Split (1)

train · 45k rows

id stringlengths 6 113	author stringlengths 2 36	task_category stringclasses 42 values	tags listlengths 1 4.05k	created_time timestamp[ns, tz=UTC]date 2022-03-02 23:29:04 2025-04-10 08:38:38	last_modified stringdate 2020-05-14 13:13:12 2025-04-19 04:15:39	downloads int64 0 118M	likes int64 0 4.86k	README stringlengths 30 1.01M	matched_bigbio_names listlengths 1 8 ⌀	is_bionlp stringclasses 3 values	model_cards stringlengths 0 1M	metadata stringlengths 2 698k	source stringclasses 2 values	matched_task listlengths 1 10 ⌀	__index_level_0__ int64 0 46.9k
gokuls/distilbert_sa_GLUE_Experiment_rte_384	gokuls	text-classification	[ "transformers", "pytorch", "tensorboard", "distilbert", "text-classification", "generated_from_trainer", "en", "dataset:glue", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-01-25T06:45:18Z	2023-01-25T06:47:05+00:00	120	0	--- datasets: - glue language: - en license: apache-2.0 metrics: - accuracy tags: - generated_from_trainer model-index: - name: distilbert_sa_GLUE_Experiment_rte_384 results: - task: type: text-classification name: Text Classification dataset: name: GLUE RTE type: glue config: rte split: validation args: rte metrics: - type: accuracy value: 0.5270758122743683 name: Accuracy --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert_sa_GLUE_Experiment_rte_384 This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the GLUE RTE dataset. It achieves the following results on the evaluation set: - Loss: 0.6919 - Accuracy: 0.5271 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 256 - eval_batch_size: 256 - seed: 10 - distributed_type: multi-GPU - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 50 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\| \| 0.698 \| 1.0 \| 10 \| 0.6962 \| 0.4729 \| \| 0.6969 \| 2.0 \| 20 \| 0.6966 \| 0.4729 \| \| 0.6955 \| 3.0 \| 30 \| 0.6919 \| 0.5271 \| \| 0.6932 \| 4.0 \| 40 \| 0.6990 \| 0.4729 \| \| 0.6941 \| 5.0 \| 50 \| 0.6931 \| 0.5054 \| \| 0.6892 \| 6.0 \| 60 \| 0.6929 \| 0.5199 \| \| 0.6843 \| 7.0 \| 70 \| 0.6931 \| 0.5560 \| \| 0.6399 \| 8.0 \| 80 \| 0.7372 \| 0.4982 \| ### Framework versions - Transformers 4.26.0 - Pytorch 1.14.0a0+410ce96 - Datasets 2.8.0 - Tokenizers 0.13.2	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert_sa_GLUE_Experiment_rte_384 This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the GLUE RTE dataset. It achieves the following results on the evaluation set: - Loss: 0.6919 - Accuracy: 0.5271 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 256 - eval_batch_size: 256 - seed: 10 - distributed_type: multi-GPU - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 50 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\| \| 0.698 \| 1.0 \| 10 \| 0.6962 \| 0.4729 \| \| 0.6969 \| 2.0 \| 20 \| 0.6966 \| 0.4729 \| \| 0.6955 \| 3.0 \| 30 \| 0.6919 \| 0.5271 \| \| 0.6932 \| 4.0 \| 40 \| 0.6990 \| 0.4729 \| \| 0.6941 \| 5.0 \| 50 \| 0.6931 \| 0.5054 \| \| 0.6892 \| 6.0 \| 60 \| 0.6929 \| 0.5199 \| \| 0.6843 \| 7.0 \| 70 \| 0.6931 \| 0.5560 \| \| 0.6399 \| 8.0 \| 80 \| 0.7372 \| 0.4982 \| ### Framework versions - Transformers 4.26.0 - Pytorch 1.14.0a0+410ce96 - Datasets 2.8.0 - Tokenizers 0.13.2	{"datasets": ["glue"], "language": ["en"], "license": "apache-2.0", "metrics": ["accuracy"], "tags": ["generated_from_trainer"], "model-index": [{"name": "distilbert_sa_GLUE_Experiment_rte_384", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "GLUE RTE", "type": "glue", "config": "rte", "split": "validation", "args": "rte"}, "metrics": [{"type": "accuracy", "value": 0.5270758122743683, "name": "Accuracy"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,062
Propicto/asr-wav2vec2-commonvoice-15-fr	Propicto	automatic-speech-recognition	[ "speechbrain", "Transformer", "wav2vec2", "CTC", "inference", "automatic-speech-recognition", "fr", "dataset:mozilla-foundation/common_voice_15_0", "base_model:LeBenchmark/wav2vec2-FR-7K-large", "base_model:finetune:LeBenchmark/wav2vec2-FR-7K-large", "license:apache-2.0", "region:us" ]	2024-10-11T13:06:34Z	2025-01-15T13:12:52+00:00	22	0	--- base_model: - LeBenchmark/wav2vec2-FR-7K-large datasets: - mozilla-foundation/common_voice_15_0 language: - fr library_name: speechbrain license: apache-2.0 metrics: - wer pipeline_tag: automatic-speech-recognition tags: - Transformer - wav2vec2 - CTC - inference --- # asr-wav2vec2-commonvoice-15-fr : LeBenchmark/wav2vec2-FR-7K-large fine-tuned on CommonVoice 15.0 French <!-- Provide a quick summary of what the model is/does. --> asr-wav2vec2-commonvoice-15-fr is an Automatic Speech Recognition model fine-tuned on CommonVoice 15.0 French set with LeBenchmark/wav2vec2-FR-7K-large as the pretrained wav2vec2 model. The fine-tuned model achieves the following performance : \| Release \| Valid WER \| Test WER \| GPUs \| Epochs \|:-------------:\|:--------------:\|:--------------:\| :--------:\|:--------:\| \| 2023-09-08 \| 9.14 \| 11.21 \| 4xV100 32GB \| 30 \| ## 📝 Model Details The ASR system is composed of: - the Tokenizer (char) that transforms the input text into a sequence of characters ("cat" into ["c", "a", "t"]) and trained with the train transcriptions (train.tsv). - the Acoustic model (wav2vec2.0 + DNN + CTC greedy decode). The pretrained wav2vec 2.0 model [LeBenchmark/wav2vec2-FR-7K-large](https://huggingface.co/LeBenchmark/wav2vec2-FR-7K-large) is combined with two DNN layers and fine-tuned on CommonVoice FR. The final acoustic representation is given to the CTC greedy decode. We used recordings sampled at 16kHz (single channel). ## 💻 How to transcribe a file with the model ### Install and import speechbrain ```bash pip install speechbrain ``` ```python from speechbrain.inference.ASR import EncoderASR ``` ### Pipeline ```python def transcribe(audio, model): return model.transcribe_file(audio).lower() def save_transcript(transcript, audio, output_file): with open(output_file, 'w', encoding='utf-8') as file: file.write(f"{audio}\t{transcript}\n") def main(): model = EncoderASR.from_hparams("Propicto/asr-wav2vec2-commonvoice-15-fr", savedir="tmp/") transcript = transcribe(audio, model) save_transcript(transcript, audio, "out.txt") ``` ## ⚙️ Training Details ### Training Data We use the train / valid / test splits provided by CommonVoice, which corresponds to: \| \| Train \| Valid \| Test \| \|:-------------:\|:-------------:\|:--------------:\|:--------------:\| \| # utterances \| 527,554 \| 16,132 \| 16,132 \| \| # hours \| 756.19 \| 25.84 \| 26.11 \| ### Training Procedure We follow the training procedure provided in the [ASR-CTC speechbrain recipe](https://github.com/speechbrain/speechbrain/tree/develop/recipes/CommonVoice/ASR/CTC). The `common_voice_prepare.py` script handles the preprocessing of the dataset. #### Training Hyperparameters Refer to the hyperparams.yaml file to get the hyperparameters information. #### Training time With 4xV100 32GB, the training took ~ 81 hours. #### Libraries [Speechbrain](https://speechbrain.github.io/): ```bibtex @misc{SB2021, author = {Ravanelli, Mirco and Parcollet, Titouan and Rouhe, Aku and Plantinga, Peter and Rastorgueva, Elena and Lugosch, Loren and Dawalatabad, Nauman and Ju-Chieh, Chou and Heba, Abdel and Grondin, Francois and Aris, William and Liao, Chien-Feng and Cornell, Samuele and Yeh, Sung-Lin and Na, Hwidong and Gao, Yan and Fu, Szu-Wei and Subakan, Cem and De Mori, Renato and Bengio, Yoshua }, title = {SpeechBrain}, year = {2021}, publisher = {GitHub}, journal = {GitHub repository}, howpublished = {\\\\url{https://github.com/speechbrain/speechbrain}}, } ``` ## 💡 Information - Developed by: Cécile Macaire - Funded by [optional]: GENCI-IDRIS (Grant 2023-AD011013625R1) PROPICTO ANR-20-CE93-0005 - Language(s) (NLP): French - License: Apache-2.0 - Finetuned from model: LeBenchmark/wav2vec2-FR-7K-large ## 📌 Citation ```bibtex @inproceedings{macaire24_interspeech, title = {Towards Speech-to-Pictograms Translation}, author = {Cécile Macaire and Chloé Dion and Didier Schwab and Benjamin Lecouteux and Emmanuelle Esperança-Rodier}, year = {2024}, booktitle = {Interspeech 2024}, pages = {857--861}, doi = {10.21437/Interspeech.2024-490}, issn = {2958-1796}, } ```	null	Non_BioNLP	# asr-wav2vec2-commonvoice-15-fr : LeBenchmark/wav2vec2-FR-7K-large fine-tuned on CommonVoice 15.0 French <!-- Provide a quick summary of what the model is/does. --> asr-wav2vec2-commonvoice-15-fr is an Automatic Speech Recognition model fine-tuned on CommonVoice 15.0 French set with LeBenchmark/wav2vec2-FR-7K-large as the pretrained wav2vec2 model. The fine-tuned model achieves the following performance : \| Release \| Valid WER \| Test WER \| GPUs \| Epochs \|:-------------:\|:--------------:\|:--------------:\| :--------:\|:--------:\| \| 2023-09-08 \| 9.14 \| 11.21 \| 4xV100 32GB \| 30 \| ## 📝 Model Details The ASR system is composed of: - the Tokenizer (char) that transforms the input text into a sequence of characters ("cat" into ["c", "a", "t"]) and trained with the train transcriptions (train.tsv). - the Acoustic model (wav2vec2.0 + DNN + CTC greedy decode). The pretrained wav2vec 2.0 model [LeBenchmark/wav2vec2-FR-7K-large](https://huggingface.co/LeBenchmark/wav2vec2-FR-7K-large) is combined with two DNN layers and fine-tuned on CommonVoice FR. The final acoustic representation is given to the CTC greedy decode. We used recordings sampled at 16kHz (single channel). ## 💻 How to transcribe a file with the model ### Install and import speechbrain ```bash pip install speechbrain ``` ```python from speechbrain.inference.ASR import EncoderASR ``` ### Pipeline ```python def transcribe(audio, model): return model.transcribe_file(audio).lower() def save_transcript(transcript, audio, output_file): with open(output_file, 'w', encoding='utf-8') as file: file.write(f"{audio}\t{transcript}\n") def main(): model = EncoderASR.from_hparams("Propicto/asr-wav2vec2-commonvoice-15-fr", savedir="tmp/") transcript = transcribe(audio, model) save_transcript(transcript, audio, "out.txt") ``` ## ⚙️ Training Details ### Training Data We use the train / valid / test splits provided by CommonVoice, which corresponds to: \| \| Train \| Valid \| Test \| \|:-------------:\|:-------------:\|:--------------:\|:--------------:\| \| # utterances \| 527,554 \| 16,132 \| 16,132 \| \| # hours \| 756.19 \| 25.84 \| 26.11 \| ### Training Procedure We follow the training procedure provided in the [ASR-CTC speechbrain recipe](https://github.com/speechbrain/speechbrain/tree/develop/recipes/CommonVoice/ASR/CTC). The `common_voice_prepare.py` script handles the preprocessing of the dataset. #### Training Hyperparameters Refer to the hyperparams.yaml file to get the hyperparameters information. #### Training time With 4xV100 32GB, the training took ~ 81 hours. #### Libraries [Speechbrain](https://speechbrain.github.io/): ```bibtex @misc{SB2021, author = {Ravanelli, Mirco and Parcollet, Titouan and Rouhe, Aku and Plantinga, Peter and Rastorgueva, Elena and Lugosch, Loren and Dawalatabad, Nauman and Ju-Chieh, Chou and Heba, Abdel and Grondin, Francois and Aris, William and Liao, Chien-Feng and Cornell, Samuele and Yeh, Sung-Lin and Na, Hwidong and Gao, Yan and Fu, Szu-Wei and Subakan, Cem and De Mori, Renato and Bengio, Yoshua }, title = {SpeechBrain}, year = {2021}, publisher = {GitHub}, journal = {GitHub repository}, howpublished = {\\\\url{https://github.com/speechbrain/speechbrain}}, } ``` ## 💡 Information - Developed by: Cécile Macaire - Funded by [optional]: GENCI-IDRIS (Grant 2023-AD011013625R1) PROPICTO ANR-20-CE93-0005 - Language(s) (NLP): French - License: Apache-2.0 - Finetuned from model: LeBenchmark/wav2vec2-FR-7K-large ## 📌 Citation ```bibtex @inproceedings{macaire24_interspeech, title = {Towards Speech-to-Pictograms Translation}, author = {Cécile Macaire and Chloé Dion and Didier Schwab and Benjamin Lecouteux and Emmanuelle Esperança-Rodier}, year = {2024}, booktitle = {Interspeech 2024}, pages = {857--861}, doi = {10.21437/Interspeech.2024-490}, issn = {2958-1796}, } ```	{"base_model": ["LeBenchmark/wav2vec2-FR-7K-large"], "datasets": ["mozilla-foundation/common_voice_15_0"], "language": ["fr"], "library_name": "speechbrain", "license": "apache-2.0", "metrics": ["wer"], "pipeline_tag": "automatic-speech-recognition", "tags": ["Transformer", "wav2vec2", "CTC", "inference"]}	task	[ "TRANSLATION" ]	43,063
Neurora/opus-tatoeba-ces-eng	Neurora	translation	[ "safetensors", "marian", "translation", "cs", "en", "license:cc-by-nc-sa-4.0", "region:us" ]	2025-02-25T20:19:17Z	2025-02-25T20:20:49+00:00	24	0	--- language: - cs - en license: cc-by-nc-sa-4.0 pipeline_tag: translation --- # Opus Tatoeba \| Czech -> English * dataset: opus * model: transformer * source language(s): ces * target language(s): eng * model: transformer * pre-processing: normalization + SentencePiece (spm32k,spm32k) * download: [opus-2021-02-19.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/ces-eng/opus-2021-02-19.zip) * test set translations: [opus-2021-02-19.test.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/ces-eng/opus-2021-02-19.test.txt) * test set scores: [opus-2021-02-19.eval.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/ces-eng/opus-2021-02-19.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| #sent \| #words \| BP \| \|---------\|-------\|-------\|-------\|--------\|----\| \| newssyscomb2009.ces-eng \| 27.7 \| 0.551 \| 502 \| 11821 \| 0.971 \| \| newstest2009.ces-eng \| 27.2 \| 0.550 \| 2525 \| 65402 \| 0.970 \| \| newstest2010.ces-eng \| 27.3 \| 0.559 \| 2489 \| 61724 \| 0.978 \| \| newstest2011.ces-eng \| 28.0 \| 0.557 \| 3003 \| 74681 \| 0.990 \| \| newstest2012.ces-eng \| 27.2 \| 0.552 \| 3003 \| 72812 \| 1.000 \| \| newstest2013.ces-eng \| 30.7 \| 0.572 \| 3000 \| 64505 \| 1.000 \| \| newstest2014-csen.ces-eng \| 34.2 \| 0.614 \| 3003 \| 68065 \| 0.999 \| \| newstest2015-encs.ces-eng \| 30.7 \| 0.568 \| 2656 \| 53572 \| 0.975 \| \| newstest2016-encs.ces-eng \| 32.4 \| 0.589 \| 2999 \| 64670 \| 0.998 \| \| newstest2017-encs.ces-eng \| 28.9 \| 0.559 \| 3005 \| 61725 \| 0.996 \| \| newstest2018-encs.ces-eng \| 30.4 \| 0.568 \| 2983 \| 63496 \| 0.991 \| \| Tatoeba-test.ces-eng \| 56.9 \| 0.719 \| 10000 \| 75376 \| 0.962 \|	null	Non_BioNLP	# Opus Tatoeba \| Czech -> English * dataset: opus * model: transformer * source language(s): ces * target language(s): eng * model: transformer * pre-processing: normalization + SentencePiece (spm32k,spm32k) * download: [opus-2021-02-19.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/ces-eng/opus-2021-02-19.zip) * test set translations: [opus-2021-02-19.test.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/ces-eng/opus-2021-02-19.test.txt) * test set scores: [opus-2021-02-19.eval.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/ces-eng/opus-2021-02-19.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| #sent \| #words \| BP \| \|---------\|-------\|-------\|-------\|--------\|----\| \| newssyscomb2009.ces-eng \| 27.7 \| 0.551 \| 502 \| 11821 \| 0.971 \| \| newstest2009.ces-eng \| 27.2 \| 0.550 \| 2525 \| 65402 \| 0.970 \| \| newstest2010.ces-eng \| 27.3 \| 0.559 \| 2489 \| 61724 \| 0.978 \| \| newstest2011.ces-eng \| 28.0 \| 0.557 \| 3003 \| 74681 \| 0.990 \| \| newstest2012.ces-eng \| 27.2 \| 0.552 \| 3003 \| 72812 \| 1.000 \| \| newstest2013.ces-eng \| 30.7 \| 0.572 \| 3000 \| 64505 \| 1.000 \| \| newstest2014-csen.ces-eng \| 34.2 \| 0.614 \| 3003 \| 68065 \| 0.999 \| \| newstest2015-encs.ces-eng \| 30.7 \| 0.568 \| 2656 \| 53572 \| 0.975 \| \| newstest2016-encs.ces-eng \| 32.4 \| 0.589 \| 2999 \| 64670 \| 0.998 \| \| newstest2017-encs.ces-eng \| 28.9 \| 0.559 \| 3005 \| 61725 \| 0.996 \| \| newstest2018-encs.ces-eng \| 30.4 \| 0.568 \| 2983 \| 63496 \| 0.991 \| \| Tatoeba-test.ces-eng \| 56.9 \| 0.719 \| 10000 \| 75376 \| 0.962 \|	{"language": ["cs", "en"], "license": "cc-by-nc-sa-4.0", "pipeline_tag": "translation"}	task	[ "TRANSLATION" ]	43,064
uhmseohun/distilbert-base-uncased-finetuned-emotion	uhmseohun	text-classification	[ "transformers", "tensorboard", "safetensors", "distilbert", "text-classification", "generated_from_trainer", "dataset:emotion", "base_model:distilbert/distilbert-base-uncased", "base_model:finetune:distilbert/distilbert-base-uncased", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-01-08T04:49:32Z	2024-01-08T05:58:08+00:00	3	0	--- base_model: distilbert-base-uncased datasets: - emotion license: apache-2.0 metrics: - accuracy - f1 tags: - generated_from_trainer model-index: - name: distilbert-base-uncased-finetuned-emotion results: - task: type: text-classification name: Text Classification dataset: name: emotion type: emotion config: split split: validation args: split metrics: - type: accuracy value: 0.924 name: Accuracy - type: f1 value: 0.9238044222632557 name: F1 --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert-base-uncased-finetuned-emotion This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset. It achieves the following results on the evaluation set: - Loss: 0.2240 - Accuracy: 0.924 - F1: 0.9238 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 60 - eval_batch_size: 60 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 2 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\|:------:\| \| 0.8722 \| 1.0 \| 267 \| 0.3283 \| 0.902 \| 0.9017 \| \| 0.2592 \| 2.0 \| 534 \| 0.2240 \| 0.924 \| 0.9238 \| ### Framework versions - Transformers 4.35.2 - Pytorch 2.1.0+cu121 - Datasets 2.16.1 - Tokenizers 0.15.0	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert-base-uncased-finetuned-emotion This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset. It achieves the following results on the evaluation set: - Loss: 0.2240 - Accuracy: 0.924 - F1: 0.9238 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 60 - eval_batch_size: 60 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 2 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\|:------:\| \| 0.8722 \| 1.0 \| 267 \| 0.3283 \| 0.902 \| 0.9017 \| \| 0.2592 \| 2.0 \| 534 \| 0.2240 \| 0.924 \| 0.9238 \| ### Framework versions - Transformers 4.35.2 - Pytorch 2.1.0+cu121 - Datasets 2.16.1 - Tokenizers 0.15.0	{"base_model": "distilbert-base-uncased", "datasets": ["emotion"], "license": "apache-2.0", "metrics": ["accuracy", "f1"], "tags": ["generated_from_trainer"], "model-index": [{"name": "distilbert-base-uncased-finetuned-emotion", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "emotion", "type": "emotion", "config": "split", "split": "validation", "args": "split"}, "metrics": [{"type": "accuracy", "value": 0.924, "name": "Accuracy"}, {"type": "f1", "value": 0.9238044222632557, "name": "F1"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,065
kumo24/mistralai-sentiment	kumo24	null	[ "safetensors", "mistral", "license:apache-2.0", "region:us" ]	2025-02-04T02:02:52Z	2025-02-04T13:13:44+00:00	16	0	--- license: apache-2.0 --- --- license: apache-2.0 --- This Falcon 7B was fined-tuned on 672k tweets from twitter/X for text classification task. The classification accuracy obtained is 98%. \ The number of labels is 3: {0: Negative, 1: Neutral, 2: Positive} \ Warning: You need enough GPU to use Falcon. This is an example to use it, it worked on 8 GB VRAM Nvidia RTX-4060 ```bash from transformers import AutoTokenizer from transformers import pipeline from transformers import AutoModelForSequenceClassification import torch checkpoint = 'kumo24/mistralai-sentiment' tokenizer=AutoTokenizer.from_pretrained(checkpoint) id2label = {0: "negative", 1: "neutral", 2: "positive"} label2id = {"negative": 0, "neutral": 1, "positive": 2} if tokenizer.pad_token is None: tokenizer.add_special_tokens({'pad_token': '[PAD]'}) model = AutoModelForSequenceClassification.from_pretrained(checkpoint, num_labels=3, id2label=id2label, label2id=label2id, device_map='auto') sentiment_task = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer) print(sentiment_task("Michigan Wolverines are Champions, Go Blue!"))	null	Non_BioNLP	--- license: apache-2.0 --- This Falcon 7B was fined-tuned on 672k tweets from twitter/X for text classification task. The classification accuracy obtained is 98%. \ The number of labels is 3: {0: Negative, 1: Neutral, 2: Positive} \ Warning: You need enough GPU to use Falcon. This is an example to use it, it worked on 8 GB VRAM Nvidia RTX-4060 ```bash from transformers import AutoTokenizer from transformers import pipeline from transformers import AutoModelForSequenceClassification import torch checkpoint = 'kumo24/mistralai-sentiment' tokenizer=AutoTokenizer.from_pretrained(checkpoint) id2label = {0: "negative", 1: "neutral", 2: "positive"} label2id = {"negative": 0, "neutral": 1, "positive": 2} if tokenizer.pad_token is None: tokenizer.add_special_tokens({'pad_token': '[PAD]'}) model = AutoModelForSequenceClassification.from_pretrained(checkpoint, num_labels=3, id2label=id2label, label2id=label2id, device_map='auto') sentiment_task = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer) print(sentiment_task("Michigan Wolverines are Champions, Go Blue!"))	{"license": "apache-2.0"}	task	[ "TEXT_CLASSIFICATION" ]	43,066
Qwen/Qwen-7B-Chat	Qwen	text-generation	[ "transformers", "safetensors", "qwen", "text-generation", "custom_code", "zh", "en", "arxiv:2309.16609", "arxiv:2305.08322", "arxiv:2009.03300", "arxiv:2305.05280", "arxiv:2210.03629", "license:other", "autotrain_compatible", "region:us" ]	2023-08-03T03:01:31Z	2024-03-19T10:09:52+00:00	91,890	765	--- language: - zh - en license: other license_name: tongyi-qianwen-license-agreement license_link: https://github.com/QwenLM/Qwen/blob/main/Tongyi%20Qianwen%20LICENSE%20AGREEMENT pipeline_tag: text-generation tags: - qwen inference: false --- # Qwen-7B-Chat <p align="center"> <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/logo_qwen.jpg" width="400"/> <p> <br> <p align="center"> 🤗 <a href="https://huggingface.co/Qwen">Hugging Face</a>&nbsp&nbsp \| &nbsp&nbsp🤖 <a href="https://modelscope.cn/organization/qwen">ModelScope</a>&nbsp&nbsp \| &nbsp&nbsp 📑 <a href="https://arxiv.org/abs/2309.16609">Paper</a> &nbsp&nbsp ｜ &nbsp&nbsp🖥️ <a href="https://modelscope.cn/studios/qwen/Qwen-7B-Chat-Demo/summary">Demo</a> <br> <a href="https://github.com/QwenLM/Qwen/blob/main/assets/wechat.png">WeChat (微信)</a>&nbsp&nbsp \| &nbsp&nbsp<a href="https://discord.gg/z3GAxXZ9Ce">Discord</a>&nbsp&nbsp ｜ &nbsp&nbsp<a href="https://dashscope.aliyun.com">API</a> </p> <br> ## 介绍（Introduction）通义千问-7B（Qwen-7B）是阿里云研发的通义千问大模型系列的70亿参数规模的模型。Qwen-7B是基于Transformer的大语言模型, 在超大规模的预训练数据上进行训练得到。预训练数据类型多样，覆盖广泛，包括大量网络文本、专业书籍、代码等。同时，在Qwen-7B的基础上，我们使用对齐机制打造了基于大语言模型的AI助手Qwen-7B-Chat。相较于最初开源的Qwen-7B模型，我们现已将预训练模型和Chat模型更新到效果更优的版本。本仓库为Qwen-7B-Chat的仓库。如果您想了解更多关于通义千问-7B开源模型的细节，我们建议您参阅[GitHub代码库](https://github.com/QwenLM/Qwen)。 Qwen-7B is the 7B-parameter version of the large language model series, Qwen (abbr. Tongyi Qianwen), proposed by Alibaba Cloud. Qwen-7B is a Transformer-based large language model, which is pretrained on a large volume of data, including web texts, books, codes, etc. Additionally, based on the pretrained Qwen-7B, we release Qwen-7B-Chat, a large-model-based AI assistant, which is trained with alignment techniques. Now we have updated both our pretrained and chat models with better performances. This repository is the one for Qwen-7B-Chat. For more details about Qwen, please refer to the [GitHub](https://github.com/QwenLM/Qwen) code repository. <br> ## 要求（Requirements） * python 3.8及以上版本 * pytorch 1.12及以上版本，推荐2.0及以上版本 * 建议使用CUDA 11.4及以上（GPU用户、flash-attention用户等需考虑此选项） * python 3.8 and above * pytorch 1.12 and above, 2.0 and above are recommended * CUDA 11.4 and above are recommended (this is for GPU users, flash-attention users, etc.) <br> ## 依赖项（Dependency）运行Qwen-7B-Chat，请确保满足上述要求，再执行以下pip命令安装依赖库 To run Qwen-7B-Chat, please make sure you meet the above requirements, and then execute the following pip commands to install the dependent libraries. ```bash pip install transformers==4.32.0 accelerate tiktoken einops scipy transformers_stream_generator==0.0.4 peft deepspeed ``` 另外，推荐安装`flash-attention`库（当前已支持flash attention 2），以实现更高的效率和更低的显存占用。 In addition, it is recommended to install the `flash-attention` library (we support flash attention 2 now.) for higher efficiency and lower memory usage. ```bash git clone https://github.com/Dao-AILab/flash-attention cd flash-attention && pip install . # 下方安装可选，安装可能比较缓慢。 # pip install csrc/layer_norm # pip install csrc/rotary ``` <br> ## 快速使用（Quickstart）下面我们展示了一个使用Qwen-7B-Chat模型，进行多轮对话交互的样例： We show an example of multi-turn interaction with Qwen-7B-Chat in the following code: ```python from transformers import AutoModelForCausalLM, AutoTokenizer from transformers.generation import GenerationConfig # Note: The default behavior now has injection attack prevention off. tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen-7B-Chat", trust_remote_code=True) # use bf16 # model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="auto", trust_remote_code=True, bf16=True).eval() # use fp16 # model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="auto", trust_remote_code=True, fp16=True).eval() # use cpu only # model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="cpu", trust_remote_code=True).eval() # use auto mode, automatically select precision based on the device. model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="auto", trust_remote_code=True).eval() # Specify hyperparameters for generation. But if you use transformers>=4.32.0, there is no need to do this. # model.generation_config = GenerationConfig.from_pretrained("Qwen/Qwen-7B-Chat", trust_remote_code=True) # 可指定不同的生成长度、top_p等相关超参 # 第一轮对话 1st dialogue turn response, history = model.chat(tokenizer, "你好", history=None) print(response) # 你好！很高兴为你提供帮助。 # 第二轮对话 2nd dialogue turn response, history = model.chat(tokenizer, "给我讲一个年轻人奋斗创业最终取得成功的故事。", history=history) print(response) # 这是一个关于一个年轻人奋斗创业最终取得成功的故事。 # 故事的主人公叫李明，他来自一个普通的家庭，父母都是普通的工人。从小，李明就立下了一个目标：要成为一名成功的企业家。 # 为了实现这个目标，李明勤奋学习，考上了大学。在大学期间，他积极参加各种创业比赛，获得了不少奖项。他还利用课余时间去实习，积累了宝贵的经验。 # 毕业后，李明决定开始自己的创业之路。他开始寻找投资机会，但多次都被拒绝了。然而，他并没有放弃。他继续努力，不断改进自己的创业计划，并寻找新的投资机会。 # 最终，李明成功地获得了一笔投资，开始了自己的创业之路。他成立了一家科技公司，专注于开发新型软件。在他的领导下，公司迅速发展起来，成为了一家成功的科技企业。 # 李明的成功并不是偶然的。他勤奋、坚韧、勇于冒险，不断学习和改进自己。他的成功也证明了，只要努力奋斗，任何人都有可能取得成功。 # 第三轮对话 3rd dialogue turn response, history = model.chat(tokenizer, "给这个故事起一个标题", history=history) print(response) # 《奋斗创业：一个年轻人的成功之路》 ``` 关于更多的使用说明，请参考我们的[GitHub repo](https://github.com/QwenLM/Qwen)获取更多信息。 For more information, please refer to our [GitHub repo](https://github.com/QwenLM/Qwen) for more information. <br> ## Tokenizer > 注：作为术语的“tokenization”在中文中尚无共识的概念对应，本文档采用英文表达以利说明。基于tiktoken的分词器有别于其他分词器，比如sentencepiece分词器。尤其在微调阶段，需要特别注意特殊token的使用。关于tokenizer的更多信息，以及微调时涉及的相关使用，请参阅[文档](https://github.com/QwenLM/Qwen/blob/main/tokenization_note_zh.md)。 Our tokenizer based on tiktoken is different from other tokenizers, e.g., sentencepiece tokenizer. You need to pay attention to special tokens, especially in finetuning. For more detailed information on the tokenizer and related use in fine-tuning, please refer to the [documentation](https://github.com/QwenLM/Qwen/blob/main/tokenization_note.md). <br> ## 量化 (Quantization) ### 用法 (Usage) 请注意：我们更新量化方案为基于[AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ)的量化，提供Qwen-7B-Chat的Int4量化模型[点击这里](https://huggingface.co/Qwen/Qwen-7B-Chat-Int4)。相比此前方案，该方案在模型评测效果几乎无损，且存储需求更低，推理速度更优。 Note: we provide a new solution based on [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and release an Int4 quantized model for Qwen-7B-Chat [Click here](https://huggingface.co/Qwen/Qwen-7B-Chat-Int4), which achieves nearly lossless model effects but improved performance on both memory costs and inference speed, in comparison with the previous solution. 以下我们提供示例说明如何使用Int4量化模型。在开始使用前，请先保证满足要求（如torch 2.0及以上，transformers版本为4.32.0及以上，等等），并安装所需安装包： Here we demonstrate how to use our provided quantized models for inference. Before you start, make sure you meet the requirements of auto-gptq (e.g., torch 2.0 and above, transformers 4.32.0 and above, etc.) and install the required packages: ```bash pip install auto-gptq optimum ``` 如安装`auto-gptq`遇到问题，我们建议您到官方[repo](https://github.com/PanQiWei/AutoGPTQ)搜索合适的预编译wheel。随后即可使用和上述一致的用法调用量化模型： If you meet problems installing `auto-gptq`, we advise you to check out the official [repo](https://github.com/PanQiWei/AutoGPTQ) to find a pre-build wheel. Then you can load the quantized model easily and run inference as same as usual: ```python model = AutoModelForCausalLM.from_pretrained( "Qwen/Qwen-7B-Chat-Int4", device_map="auto", trust_remote_code=True ).eval() response, history = model.chat(tokenizer, "你好", history=None) ``` ### 效果评测我们对BF16，Int8和Int4模型在基准评测上做了测试（使用zero-shot设置），发现量化模型效果损失较小，结果如下所示： We illustrate the zero-shot performance of both BF16, Int8 and Int4 models on the benchmark, and we find that the quantized model does not suffer from significant performance degradation. Results are shown below: \| Quantization \| MMLU \| CEval (val) \| GSM8K \| Humaneval \| \| ------------- \| :--------: \| :----------: \| :----: \| :--------: \| \| BF16 \| 55.8 \| 59.7 \| 50.3 \| 37.2 \| \| Int8 \| 55.4 \| 59.4 \| 48.3 \| 34.8 \| \| Int4 \| 55.1 \| 59.2 \| 49.7 \| 29.9 \| ### 推理速度 (Inference Speed) 我们测算了不同精度模型以及不同FlashAttn库版本下模型生成2048和8192个token的平均推理速度。如图所示： We measured the average inference speed of generating 2048 and 8192 tokens with different quantization levels and versions of flash-attention, respectively. \| Quantization \| FlashAttn \| Speed (2048 tokens) \| Speed (8192 tokens) \| \| ------------- \| :-------: \| :------------------:\| :------------------:\| \| BF16 \| v2 \| 40.93 \| 36.14 \| \| Int8 \| v2 \| 37.47 \| 32.54 \| \| Int4 \| v2 \| 50.09 \| 38.61 \| \| BF16 \| v1 \| 40.75 \| 35.34 \| \| Int8 \| v1 \| 37.51 \| 32.39 \| \| Int4 \| v1 \| 45.98 \| 36.47 \| \| BF16 \| Disabled \| 37.55 \| 33.56 \| \| Int8 \| Disabled \| 37.84 \| 32.65 \| \| Int4 \| Disabled \| 48.12 \| 36.70 \| 具体而言，我们记录在长度为1的上下文的条件下生成8192个token的性能。评测运行于单张A100-SXM4-80G GPU，使用PyTorch 2.0.1和CUDA 11.8。推理速度是生成8192个token的速度均值。 In detail, the setting of profiling is generating 8192 new tokens with 1 context token. The profiling runs on a single A100-SXM4-80G GPU with PyTorch 2.0.1 and CUDA 11.8. The inference speed is averaged over the generated 8192 tokens. 注意：以上Int4/Int8模型生成速度使用autogptq库给出，当前``AutoModelForCausalLM.from_pretrained``载入的模型生成速度会慢大约20%。我们已经将该问题汇报给HuggingFace团队，若有解决方案将即时更新。 Note: The generation speed of the Int4/Int8 models mentioned above is provided by the autogptq library. The current speed of the model loaded using "AutoModelForCausalLM.from_pretrained" will be approximately 20% slower. We have reported this issue to the HuggingFace team and will update it promptly if a solution is available. ### 显存使用 (GPU Memory Usage) 我们还测算了不同模型精度编码2048个token及生成8192个token的峰值显存占用情况。（显存消耗在是否使用FlashAttn的情况下均类似。）结果如下所示： We also profile the peak GPU memory usage for encoding 2048 tokens as context (and generating single token) and generating 8192 tokens (with single token as context) under different quantization levels, respectively. （The GPU memory usage is similar when using flash-attention or not.）The results are shown below. \| Quantization Level \| Peak Usage for Encoding 2048 Tokens \| Peak Usage for Generating 8192 Tokens \| \| ------------------ \| :---------------------------------: \| :-----------------------------------: \| \| BF16 \| 16.99GB \| 22.53GB \| \| Int8 \| 11.20GB \| 16.62GB \| \| Int4 \| 8.21GB \| 13.63GB \| 上述性能测算使用[此脚本](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py)完成。 The above speed and memory profiling are conducted using [this script](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py). <br> ## 模型细节（Model）与Qwen-7B预训练模型相同，Qwen-7B-Chat模型规模基本情况如下所示: The details of the model architecture of Qwen-7B-Chat are listed as follows: \| Hyperparameter \| Value \| \|:----------------\|:------:\| \| n_layers \| 32 \| \| n_heads \| 32 \| \| d_model \| 4096 \| \| vocab size \| 151851 \| \| sequence length \| 8192 \| 在位置编码、FFN激活函数和normalization的实现方式上，我们也采用了目前最流行的做法，即RoPE相对位置编码、SwiGLU激活函数、RMSNorm（可选安装flash-attention加速）。在分词器方面，相比目前主流开源模型以中英词表为主，Qwen-7B-Chat使用了约15万token大小的词表。该词表在GPT-4使用的BPE词表`cl100k_base`基础上，对中文、多语言进行了优化，在对中、英、代码数据的高效编解码的基础上，对部分多语言更加友好，方便用户在不扩展词表的情况下对部分语种进行能力增强。词表对数字按单个数字位切分。调用较为高效的[tiktoken分词库](https://github.com/openai/tiktoken)进行分词。 For position encoding, FFN activation function, and normalization calculation methods, we adopt the prevalent practices, i.e., RoPE relative position encoding, SwiGLU for activation function, and RMSNorm for normalization (optional installation of flash-attention for acceleration). For tokenization, compared to the current mainstream open-source models based on Chinese and English vocabularies, Qwen-7B-Chat uses a vocabulary of over 150K tokens. It first considers efficient encoding of Chinese, English, and code data, and is also more friendly to multilingual languages, enabling users to directly enhance the capability of some languages without expanding the vocabulary. It segments numbers by single digit, and calls the [tiktoken](https://github.com/openai/tiktoken) tokenizer library for efficient tokenization. <br> ## 评测效果（Evaluation）对于Qwen-7B-Chat模型，我们同样评测了常规的中文理解（C-Eval）、英文理解（MMLU）、代码（HumanEval）和数学（GSM8K）等权威任务，同时包含了长序列任务的评测结果。由于Qwen-7B-Chat模型经过对齐后，激发了较强的外部系统调用能力，我们还进行了工具使用能力方面的评测。提示：由于硬件和框架造成的舍入误差，复现结果如有波动属于正常现象。 For Qwen-7B-Chat, we also evaluate the model on C-Eval, MMLU, HumanEval, GSM8K, etc., as well as the benchmark evaluation for long-context understanding, and tool usage. Note: Due to rounding errors caused by hardware and framework, differences in reproduced results are possible. ### 中文评测（Chinese Evaluation） #### C-Eval 在[C-Eval](https://arxiv.org/abs/2305.08322)验证集上，我们评价了Qwen-7B-Chat模型的0-shot & 5-shot准确率 We demonstrate the 0-shot & 5-shot accuracy of Qwen-7B-Chat on C-Eval validation set \| Model \| Avg. Acc. \| \|:--------------------------------:\|:---------:\| \| LLaMA2-7B-Chat \| 31.9 \| \| LLaMA2-13B-Chat \| 36.2 \| \| LLaMA2-70B-Chat \| 44.3 \| \| ChatGLM2-6B-Chat \| 52.6 \| \| InternLM-7B-Chat \| 53.6 \| \| Baichuan2-7B-Chat \| 55.6 \| \| Baichuan2-13B-Chat \| 56.7 \| \| Qwen-7B-Chat (original) (0-shot) \| 54.2 \| \| Qwen-7B-Chat (0-shot) \| 59.7 \| \| Qwen-7B-Chat (5-shot) \| 59.3 \| \| Qwen-14B-Chat (0-shot) \| 69.8 \| \| Qwen-14B-Chat (5-shot) \| 71.7 \| C-Eval测试集上，Qwen-7B-Chat模型的zero-shot准确率结果如下： The zero-shot accuracy of Qwen-7B-Chat on C-Eval testing set is provided below: \| Model \| Avg. \| STEM \| Social Sciences \| Humanities \| Others \| \| :---------------------- \| :------: \| :--: \| :-------------: \| :--------: \| :----: \| \| Chinese-Alpaca-Plus-13B \| 41.5 \| 36.6 \| 49.7 \| 43.1 \| 41.2 \| \| Chinese-Alpaca-2-7B \| 40.3 \| - \| - \| - \| - \| \| ChatGLM2-6B-Chat \| 50.1 \| 46.4 \| 60.4 \| 50.6 \| 46.9 \| \| Baichuan-13B-Chat \| 51.5 \| 43.7 \| 64.6 \| 56.2 \| 49.2 \| \| Qwen-7B-Chat (original) \| 54.6 \| 47.8 \| 67.6 \| 59.3 \| 50.6 \| \| Qwen-7B-Chat \| 58.6 \| 53.3 \| 72.1 \| 62.8 \| 52.0 \| \| Qwen-14B-Chat \| 69.1 \| 65.1 \| 80.9 \| 71.2 \| 63.4 \| 在7B规模模型上，经过人类指令对齐的Qwen-7B-Chat模型，准确率在同类相近规模模型中仍然处于前列。 Compared with other pretrained models with comparable model size, the human-aligned Qwen-7B-Chat performs well in C-Eval accuracy. ### 英文评测（English Evaluation） #### MMLU [MMLU](https://arxiv.org/abs/2009.03300)评测集上，Qwen-7B-Chat模型的 0-shot & 5-shot 准确率如下，效果同样在同类对齐模型中同样表现较优。 The 0-shot & 5-shot accuracy of Qwen-7B-Chat on MMLU is provided below. The performance of Qwen-7B-Chat still on the top between other human-aligned models with comparable size. \| Model \| Avg. Acc. \| \|:--------------------------------:\|:---------:\| \| ChatGLM2-6B-Chat \| 46.0 \| \| LLaMA2-7B-Chat \| 46.2 \| \| InternLM-7B-Chat \| 51.1 \| \| Baichuan2-7B-Chat \| 52.9 \| \| LLaMA2-13B-Chat \| 54.6 \| \| Baichuan2-13B-Chat \| 57.3 \| \| LLaMA2-70B-Chat \| 63.8 \| \| Qwen-7B-Chat (original) (0-shot) \| 53.9 \| \| Qwen-7B-Chat (0-shot) \| 55.8 \| \| Qwen-7B-Chat (5-shot) \| 57.0 \| \| Qwen-14B-Chat (0-shot) \| 64.6 \| \| Qwen-14B-Chat (5-shot) \| 66.5 \| ### 代码评测（Coding Evaluation） Qwen-7B-Chat在[HumanEval](https://github.com/openai/human-eval)的zero-shot Pass@1效果如下 The zero-shot Pass@1 of Qwen-7B-Chat on [HumanEval](https://github.com/openai/human-eval) is demonstrated below \| Model \| Pass@1 \| \|:-----------------------:\|:--------:\| \| ChatGLM2-6B-Chat \| 11.0 \| \| LLaMA2-7B-Chat \| 12.2 \| \| Baichuan2-7B-Chat \| 13.4 \| \| InternLM-7B-Chat \| 14.6 \| \| Baichuan2-13B-Chat \| 17.7 \| \| LLaMA2-13B-Chat \| 18.9 \| \| LLaMA2-70B-Chat \| 32.3 \| \| Qwen-7B-Chat (original) \| 24.4 \| \| Qwen-7B-Chat \| 37.2 \| \| Qwen-14B-Chat \| 43.9 \| ### 数学评测（Mathematics Evaluation）在评测数学能力的[GSM8K](https://github.com/openai/grade-school-math)上，Qwen-7B-Chat的准确率结果如下 The accuracy of Qwen-7B-Chat on GSM8K is shown below \| Model \| Acc. \| \|:------------------------------------:\|:--------:\| \| LLaMA2-7B-Chat \| 26.3 \| \| ChatGLM2-6B-Chat \| 28.8 \| \| Baichuan2-7B-Chat \| 32.8 \| \| InternLM-7B-Chat \| 33.0 \| \| LLaMA2-13B-Chat \| 37.1 \| \| Baichuan2-13B-Chat \| 55.3 \| \| LLaMA2-70B-Chat \| 59.3 \| \| Qwen-7B-Chat (original) (0-shot) \| 41.1 \| \| Qwen-7B-Chat (0-shot) \| 50.3 \| \| Qwen-7B-Chat (8-shot) \| 54.1 \| \| Qwen-14B-Chat (0-shot) \| 60.1 \| \| Qwen-14B-Chat (8-shot) \| 59.3 \| ### 长序列评测（Long-Context Understanding）通过NTK插值，LogN注意力缩放可以扩展Qwen-7B-Chat的上下文长度。在长文本摘要数据集[VCSUM](https://arxiv.org/abs/2305.05280)上（文本平均长度在15K左右），Qwen-7B-Chat的Rouge-L结果如下： (若要启用这些技巧，请将config.json里的`use_dynamic_ntk`和`use_logn_attn`设置为true) We introduce NTK-aware interpolation, LogN attention scaling to extend the context length of Qwen-7B-Chat. The Rouge-L results of Qwen-7B-Chat on long-text summarization dataset [VCSUM](https://arxiv.org/abs/2305.05280) (The average length of this dataset is around 15K) are shown below: (To use these tricks, please set `use_dynamic_ntk` and `use_long_attn` to true in config.json.) \| Model \| VCSUM (zh) \| \|:------------------\|:----------:\| \| GPT-3.5-Turbo-16k \| 16.0 \| \| LLama2-7B-Chat \| 0.2 \| \| InternLM-7B-Chat \| 13.0 \| \| ChatGLM2-6B-Chat \| 16.3 \| \| Qwen-7B-Chat \| 16.6 \| ### 工具使用能力的评测（Tool Usage） #### ReAct Prompting 千问支持通过 [ReAct Prompting](https://arxiv.org/abs/2210.03629) 调用插件/工具/API。ReAct 也是 [LangChain](https://python.langchain.com/) 框架采用的主要方式之一。在我们开源的、用于评估工具使用能力的评测基准上，千问的表现如下： Qwen-Chat supports calling plugins/tools/APIs through [ReAct Prompting](https://arxiv.org/abs/2210.03629). ReAct is also one of the main approaches used by the [LangChain](https://python.langchain.com/) framework. In our evaluation benchmark for assessing tool usage capabilities, Qwen-Chat's performance is as follows: <table> <tr> <th colspan="4" align="center">Chinese Tool-Use Benchmark</th> </tr> <tr> <th align="center">Model</th><th align="center">Tool Selection (Acc.↑)</th><th align="center">Tool Input (Rouge-L↑)</th><th align="center">False Positive Error↓</th> </tr> <tr> <td>GPT-4</td><td align="center">95%</td><td align="center">0.90</td><td align="center">15.0%</td> </tr> <tr> <td>GPT-3.5</td><td align="center">85%</td><td align="center">0.88</td><td align="center">75.0%</td> </tr> <tr> <td>Qwen-7B-Chat</td><td align="center">98%</td><td align="center">0.91</td><td align="center">7.3%</td> </tr> <tr> <td>Qwen-14B-Chat</td><td align="center">98%</td><td align="center">0.93</td><td align="center">2.4%</td> </tr> </table> > 评测基准中出现的插件均没有出现在千问的训练集中。该基准评估了模型在多个候选插件中选择正确插件的准确率、传入插件的参数的合理性、以及假阳率。假阳率（False Positive）定义：在处理不该调用插件的请求时，错误地调用了插件。 > The plugins that appear in the evaluation set do not appear in the training set of Qwen. This benchmark evaluates the accuracy of the model in selecting the correct plugin from multiple candidate plugins, the rationality of the parameters passed into the plugin, and the false positive rate. False Positive: Incorrectly invoking a plugin when it should not have been called when responding to a query. ![](assets/react_showcase_001.png) ![](assets/react_showcase_002.png) #### Code Interpreter 为了考察Qwen使用Python Code Interpreter完成数学解题、数据可视化、及文件处理与爬虫等任务的能力，我们专门建设并开源了一个评测这方面能力的[评测基准](https://github.com/QwenLM/Qwen-Agent/tree/main/benchmark)。我们发现Qwen在生成代码的可执行率、结果正确性上均表现较好： To assess Qwen's ability to use the Python Code Interpreter for tasks such as mathematical problem solving, data visualization, and other general-purpose tasks such as file handling and web scraping, we have created and open-sourced a benchmark specifically designed for evaluating these capabilities. You can find the benchmark at this [link](https://github.com/QwenLM/Qwen-Agent/tree/main/benchmark). We have observed that Qwen performs well in terms of code executability and result accuracy when generating code: <table> <tr> <th colspan="4" align="center">Executable Rate of Generated Code (%)</th> </tr> <tr> <th align="center">Model</th><th align="center">Math↑</th><th align="center">Visualization↑</th><th align="center">General↑</th> </tr> <tr> <td>GPT-4</td><td align="center">91.9</td><td align="center">85.9</td><td align="center">82.8</td> </tr> <tr> <td>GPT-3.5</td><td align="center">89.2</td><td align="center">65.0</td><td align="center">74.1</td> </tr> <tr> <td>LLaMA2-7B-Chat</td> <td align="center">41.9</td> <td align="center">33.1</td> <td align="center">24.1 </td> </tr> <tr> <td>LLaMA2-13B-Chat</td> <td align="center">50.0</td> <td align="center">40.5</td> <td align="center">48.3 </td> </tr> <tr> <td>CodeLLaMA-7B-Instruct</td> <td align="center">85.1</td> <td align="center">54.0</td> <td align="center">70.7 </td> </tr> <tr> <td>CodeLLaMA-13B-Instruct</td> <td align="center">93.2</td> <td align="center">55.8</td> <td align="center">74.1 </td> </tr> <tr> <td>InternLM-7B-Chat-v1.1</td> <td align="center">78.4</td> <td align="center">44.2</td> <td align="center">62.1 </td> </tr> <tr> <td>InternLM-20B-Chat</td> <td align="center">70.3</td> <td align="center">44.2</td> <td align="center">65.5 </td> </tr> <tr> <td>Qwen-7B-Chat</td> <td align="center">82.4</td> <td align="center">64.4</td> <td align="center">67.2 </td> </tr> <tr> <td>Qwen-14B-Chat</td> <td align="center">89.2</td> <td align="center">84.1</td> <td align="center">65.5</td> </tr> </table> <table> <tr> <th colspan="4" align="center">Accuracy of Code Execution Results (%)</th> </tr> <tr> <th align="center">Model</th><th align="center">Math↑</th><th align="center">Visualization-Hard↑</th><th align="center">Visualization-Easy↑</th> </tr> <tr> <td>GPT-4</td><td align="center">82.8</td><td align="center">66.7</td><td align="center">60.8</td> </tr> <tr> <td>GPT-3.5</td><td align="center">47.3</td><td align="center">33.3</td><td align="center">55.7</td> </tr> <tr> <td>LLaMA2-7B-Chat</td> <td align="center">3.9</td> <td align="center">14.3</td> <td align="center">39.2 </td> </tr> <tr> <td>LLaMA2-13B-Chat</td> <td align="center">8.3</td> <td align="center">8.3</td> <td align="center">40.5 </td> </tr> <tr> <td>CodeLLaMA-7B-Instruct</td> <td align="center">14.3</td> <td align="center">26.2</td> <td align="center">60.8 </td> </tr> <tr> <td>CodeLLaMA-13B-Instruct</td> <td align="center">28.2</td> <td align="center">27.4</td> <td align="center">62.0 </td> </tr> <tr> <td>InternLM-7B-Chat-v1.1</td> <td align="center">28.5</td> <td align="center">4.8</td> <td align="center">40.5 </td> </tr> <tr> <td>InternLM-20B-Chat</td> <td align="center">34.6</td> <td align="center">21.4</td> <td align="center">45.6 </td> </tr> <tr> <td>Qwen-7B-Chat</td> <td align="center">41.9</td> <td align="center">40.5</td> <td align="center">54.4 </td> </tr> <tr> <td>Qwen-14B-Chat</td> <td align="center">58.4</td> <td align="center">53.6</td> <td align="center">59.5</td> </tr> </table> <p align="center"> <br> <img src="assets/code_interpreter_showcase_001.jpg" /> <br> <p> #### Huggingface Agent 千问还具备作为 [HuggingFace Agent](https://huggingface.co/docs/transformers/transformers_agents) 的能力。它在 Huggingface 提供的run模式评测基准上的表现如下： Qwen-Chat also has the capability to be used as a [HuggingFace Agent](https://huggingface.co/docs/transformers/transformers_agents). Its performance on the run-mode benchmark provided by HuggingFace is as follows: <table> <tr> <th colspan="4" align="center">HuggingFace Agent Benchmark- Run Mode</th> </tr> <tr> <th align="center">Model</th><th align="center">Tool Selection↑</th><th align="center">Tool Used↑</th><th align="center">Code↑</th> </tr> <tr> <td>GPT-4</td><td align="center">100</td><td align="center">100</td><td align="center">97.4</td> </tr> <tr> <td>GPT-3.5</td><td align="center">95.4</td><td align="center">96.3</td><td align="center">87.0</td> </tr> <tr> <td>StarCoder-Base-15B</td><td align="center">86.1</td><td align="center">87.0</td><td align="center">68.9</td> </tr> <tr> <td>StarCoder-15B</td><td align="center">87.0</td><td align="center">88.0</td><td align="center">68.9</td> </tr> <tr> <td>Qwen-7B-Chat</td><td align="center">87.0</td><td align="center">87.0</td><td align="center">71.5</td> </tr> <tr> <td>Qwen-14B-Chat</td><td align="center">93.5</td><td align="center">94.4</td><td align="center">87.0</td> </tr> </table> <table> <tr> <th colspan="4" align="center">HuggingFace Agent Benchmark - Chat Mode</th> </tr> <tr> <th align="center">Model</th><th align="center">Tool Selection↑</th><th align="center">Tool Used↑</th><th align="center">Code↑</th> </tr> <tr> <td>GPT-4</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">98.5</td> </tr> <tr> <td>GPT-3.5</td><td align="center">97.3</td><td align="center">96.8</td><td align="center">89.6</td> </tr> <tr> <td>StarCoder-Base-15B</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">91.1</td> </tr> <tr> <td>StarCoder-15B</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">89.6</td> </tr> <tr> <td>Qwen-7B-Chat</td><td align="center">94.7</td><td align="center">94.7</td><td align="center">85.1</td> </tr> <tr> <td>Qwen-14B-Chat</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">95.5</td> </tr> </table> <br> ## x86 平台 (x86 Platforms) 在酷睿™/至强® 可扩展处理器或 Arc™ GPU 上部署量化模型时，建议使用 [OpenVINO™ Toolkit](https://docs.openvino.ai/2023.3/gen_ai_guide.html)以充分利用硬件，实现更好的推理性能。您可以安装并运行此 [example notebook](https://github.com/openvinotoolkit/openvino_notebooks/tree/main/notebooks/254-llm-chatbot)。相关问题，您可在[OpenVINO repo](https://github.com/openvinotoolkit/openvino_notebooks/issues)中提交。 When deploy on Core™/Xeon® Scalable Processors or with Arc™ GPU, [OpenVINO™ Toolkit](https://docs.openvino.ai/2023.3/gen_ai_guide.html) is recommended. You can install and run this [example notebook](https://github.com/openvinotoolkit/openvino_notebooks/tree/main/notebooks/254-llm-chatbot). For related issues, you are welcome to file an issue at [OpenVINO repo](https://github.com/openvinotoolkit/openvino_notebooks/issues). ## FAQ 如遇到问题，敬请查阅[FAQ](https://github.com/QwenLM/Qwen/blob/main/FAQ_zh.md)以及issue区，如仍无法解决再提交issue。 If you meet problems, please refer to [FAQ](https://github.com/QwenLM/Qwen/blob/main/FAQ.md) and the issues first to search a solution before you launch a new issue. <br> ## 引用 (Citation) 如果你觉得我们的工作对你有帮助，欢迎引用！ If you find our work helpful, feel free to give us a cite. ``` @article{qwen, title={Qwen Technical Report}, author={Jinze Bai and Shuai Bai and Yunfei Chu and Zeyu Cui and Kai Dang and Xiaodong Deng and Yang Fan and Wenbin Ge and Yu Han and Fei Huang and Binyuan Hui and Luo Ji and Mei Li and Junyang Lin and Runji Lin and Dayiheng Liu and Gao Liu and Chengqiang Lu and Keming Lu and Jianxin Ma and Rui Men and Xingzhang Ren and Xuancheng Ren and Chuanqi Tan and Sinan Tan and Jianhong Tu and Peng Wang and Shijie Wang and Wei Wang and Shengguang Wu and Benfeng Xu and Jin Xu and An Yang and Hao Yang and Jian Yang and Shusheng Yang and Yang Yao and Bowen Yu and Hongyi Yuan and Zheng Yuan and Jianwei Zhang and Xingxuan Zhang and Yichang Zhang and Zhenru Zhang and Chang Zhou and Jingren Zhou and Xiaohuan Zhou and Tianhang Zhu}, journal={arXiv preprint arXiv:2309.16609}, year={2023} } ``` <br> ## 使用协议（License Agreement）我们的代码和模型权重对学术研究完全开放，并支持商用。请查看[LICENSE](https://github.com/QwenLM/Qwen/blob/main/Tongyi%20Qianwen%20LICENSE%20AGREEMENT)了解具体的开源协议细节。如需商用，请填写[问卷](https://dashscope.console.aliyun.com/openModelApply/qianwen)申请。 Our code and checkpoints are open to research purpose, and they are allowed for commercial purposes. Check [LICENSE](https://github.com/QwenLM/Qwen/blob/main/Tongyi%20Qianwen%20LICENSE%20AGREEMENT) for more details about the license. If you have requirements for commercial use, please fill out the [form](https://dashscope.console.aliyun.com/openModelApply/qianwen) to apply. <br> ## 联系我们（Contact Us）如果你想给我们的研发团队和产品团队留言，欢迎加入我们的微信群、钉钉群以及Discord！同时，也欢迎通过邮件（[email protected]）联系我们。 If you are interested to leave a message to either our research team or product team, join our Discord or WeChat groups! Also, feel free to send an email to [email protected].	null	Non_BioNLP	# Qwen-7B-Chat <p align="center"> <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/logo_qwen.jpg" width="400"/> <p> <br> <p align="center"> 🤗 <a href="https://huggingface.co/Qwen">Hugging Face</a>&nbsp&nbsp \| &nbsp&nbsp🤖 <a href="https://modelscope.cn/organization/qwen">ModelScope</a>&nbsp&nbsp \| &nbsp&nbsp 📑 <a href="https://arxiv.org/abs/2309.16609">Paper</a> &nbsp&nbsp ｜ &nbsp&nbsp🖥️ <a href="https://modelscope.cn/studios/qwen/Qwen-7B-Chat-Demo/summary">Demo</a> <br> <a href="https://github.com/QwenLM/Qwen/blob/main/assets/wechat.png">WeChat (微信)</a>&nbsp&nbsp \| &nbsp&nbsp<a href="https://discord.gg/z3GAxXZ9Ce">Discord</a>&nbsp&nbsp ｜ &nbsp&nbsp<a href="https://dashscope.aliyun.com">API</a> </p> <br> ## 介绍（Introduction）通义千问-7B（Qwen-7B）是阿里云研发的通义千问大模型系列的70亿参数规模的模型。Qwen-7B是基于Transformer的大语言模型, 在超大规模的预训练数据上进行训练得到。预训练数据类型多样，覆盖广泛，包括大量网络文本、专业书籍、代码等。同时，在Qwen-7B的基础上，我们使用对齐机制打造了基于大语言模型的AI助手Qwen-7B-Chat。相较于最初开源的Qwen-7B模型，我们现已将预训练模型和Chat模型更新到效果更优的版本。本仓库为Qwen-7B-Chat的仓库。如果您想了解更多关于通义千问-7B开源模型的细节，我们建议您参阅[GitHub代码库](https://github.com/QwenLM/Qwen)。 Qwen-7B is the 7B-parameter version of the large language model series, Qwen (abbr. Tongyi Qianwen), proposed by Alibaba Cloud. Qwen-7B is a Transformer-based large language model, which is pretrained on a large volume of data, including web texts, books, codes, etc. Additionally, based on the pretrained Qwen-7B, we release Qwen-7B-Chat, a large-model-based AI assistant, which is trained with alignment techniques. Now we have updated both our pretrained and chat models with better performances. This repository is the one for Qwen-7B-Chat. For more details about Qwen, please refer to the [GitHub](https://github.com/QwenLM/Qwen) code repository. <br> ## 要求（Requirements） * python 3.8及以上版本 * pytorch 1.12及以上版本，推荐2.0及以上版本 * 建议使用CUDA 11.4及以上（GPU用户、flash-attention用户等需考虑此选项） * python 3.8 and above * pytorch 1.12 and above, 2.0 and above are recommended * CUDA 11.4 and above are recommended (this is for GPU users, flash-attention users, etc.) <br> ## 依赖项（Dependency）运行Qwen-7B-Chat，请确保满足上述要求，再执行以下pip命令安装依赖库 To run Qwen-7B-Chat, please make sure you meet the above requirements, and then execute the following pip commands to install the dependent libraries. ```bash pip install transformers==4.32.0 accelerate tiktoken einops scipy transformers_stream_generator==0.0.4 peft deepspeed ``` 另外，推荐安装`flash-attention`库（当前已支持flash attention 2），以实现更高的效率和更低的显存占用。 In addition, it is recommended to install the `flash-attention` library (we support flash attention 2 now.) for higher efficiency and lower memory usage. ```bash git clone https://github.com/Dao-AILab/flash-attention cd flash-attention && pip install . # 下方安装可选，安装可能比较缓慢。 # pip install csrc/layer_norm # pip install csrc/rotary ``` <br> ## 快速使用（Quickstart）下面我们展示了一个使用Qwen-7B-Chat模型，进行多轮对话交互的样例： We show an example of multi-turn interaction with Qwen-7B-Chat in the following code: ```python from transformers import AutoModelForCausalLM, AutoTokenizer from transformers.generation import GenerationConfig # Note: The default behavior now has injection attack prevention off. tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen-7B-Chat", trust_remote_code=True) # use bf16 # model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="auto", trust_remote_code=True, bf16=True).eval() # use fp16 # model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="auto", trust_remote_code=True, fp16=True).eval() # use cpu only # model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="cpu", trust_remote_code=True).eval() # use auto mode, automatically select precision based on the device. model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="auto", trust_remote_code=True).eval() # Specify hyperparameters for generation. But if you use transformers>=4.32.0, there is no need to do this. # model.generation_config = GenerationConfig.from_pretrained("Qwen/Qwen-7B-Chat", trust_remote_code=True) # 可指定不同的生成长度、top_p等相关超参 # 第一轮对话 1st dialogue turn response, history = model.chat(tokenizer, "你好", history=None) print(response) # 你好！很高兴为你提供帮助。 # 第二轮对话 2nd dialogue turn response, history = model.chat(tokenizer, "给我讲一个年轻人奋斗创业最终取得成功的故事。", history=history) print(response) # 这是一个关于一个年轻人奋斗创业最终取得成功的故事。 # 故事的主人公叫李明，他来自一个普通的家庭，父母都是普通的工人。从小，李明就立下了一个目标：要成为一名成功的企业家。 # 为了实现这个目标，李明勤奋学习，考上了大学。在大学期间，他积极参加各种创业比赛，获得了不少奖项。他还利用课余时间去实习，积累了宝贵的经验。 # 毕业后，李明决定开始自己的创业之路。他开始寻找投资机会，但多次都被拒绝了。然而，他并没有放弃。他继续努力，不断改进自己的创业计划，并寻找新的投资机会。 # 最终，李明成功地获得了一笔投资，开始了自己的创业之路。他成立了一家科技公司，专注于开发新型软件。在他的领导下，公司迅速发展起来，成为了一家成功的科技企业。 # 李明的成功并不是偶然的。他勤奋、坚韧、勇于冒险，不断学习和改进自己。他的成功也证明了，只要努力奋斗，任何人都有可能取得成功。 # 第三轮对话 3rd dialogue turn response, history = model.chat(tokenizer, "给这个故事起一个标题", history=history) print(response) # 《奋斗创业：一个年轻人的成功之路》 ``` 关于更多的使用说明，请参考我们的[GitHub repo](https://github.com/QwenLM/Qwen)获取更多信息。 For more information, please refer to our [GitHub repo](https://github.com/QwenLM/Qwen) for more information. <br> ## Tokenizer > 注：作为术语的“tokenization”在中文中尚无共识的概念对应，本文档采用英文表达以利说明。基于tiktoken的分词器有别于其他分词器，比如sentencepiece分词器。尤其在微调阶段，需要特别注意特殊token的使用。关于tokenizer的更多信息，以及微调时涉及的相关使用，请参阅[文档](https://github.com/QwenLM/Qwen/blob/main/tokenization_note_zh.md)。 Our tokenizer based on tiktoken is different from other tokenizers, e.g., sentencepiece tokenizer. You need to pay attention to special tokens, especially in finetuning. For more detailed information on the tokenizer and related use in fine-tuning, please refer to the [documentation](https://github.com/QwenLM/Qwen/blob/main/tokenization_note.md). <br> ## 量化 (Quantization) ### 用法 (Usage) 请注意：我们更新量化方案为基于[AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ)的量化，提供Qwen-7B-Chat的Int4量化模型[点击这里](https://huggingface.co/Qwen/Qwen-7B-Chat-Int4)。相比此前方案，该方案在模型评测效果几乎无损，且存储需求更低，推理速度更优。 Note: we provide a new solution based on [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and release an Int4 quantized model for Qwen-7B-Chat [Click here](https://huggingface.co/Qwen/Qwen-7B-Chat-Int4), which achieves nearly lossless model effects but improved performance on both memory costs and inference speed, in comparison with the previous solution. 以下我们提供示例说明如何使用Int4量化模型。在开始使用前，请先保证满足要求（如torch 2.0及以上，transformers版本为4.32.0及以上，等等），并安装所需安装包： Here we demonstrate how to use our provided quantized models for inference. Before you start, make sure you meet the requirements of auto-gptq (e.g., torch 2.0 and above, transformers 4.32.0 and above, etc.) and install the required packages: ```bash pip install auto-gptq optimum ``` 如安装`auto-gptq`遇到问题，我们建议您到官方[repo](https://github.com/PanQiWei/AutoGPTQ)搜索合适的预编译wheel。随后即可使用和上述一致的用法调用量化模型： If you meet problems installing `auto-gptq`, we advise you to check out the official [repo](https://github.com/PanQiWei/AutoGPTQ) to find a pre-build wheel. Then you can load the quantized model easily and run inference as same as usual: ```python model = AutoModelForCausalLM.from_pretrained( "Qwen/Qwen-7B-Chat-Int4", device_map="auto", trust_remote_code=True ).eval() response, history = model.chat(tokenizer, "你好", history=None) ``` ### 效果评测我们对BF16，Int8和Int4模型在基准评测上做了测试（使用zero-shot设置），发现量化模型效果损失较小，结果如下所示： We illustrate the zero-shot performance of both BF16, Int8 and Int4 models on the benchmark, and we find that the quantized model does not suffer from significant performance degradation. Results are shown below: \| Quantization \| MMLU \| CEval (val) \| GSM8K \| Humaneval \| \| ------------- \| :--------: \| :----------: \| :----: \| :--------: \| \| BF16 \| 55.8 \| 59.7 \| 50.3 \| 37.2 \| \| Int8 \| 55.4 \| 59.4 \| 48.3 \| 34.8 \| \| Int4 \| 55.1 \| 59.2 \| 49.7 \| 29.9 \| ### 推理速度 (Inference Speed) 我们测算了不同精度模型以及不同FlashAttn库版本下模型生成2048和8192个token的平均推理速度。如图所示： We measured the average inference speed of generating 2048 and 8192 tokens with different quantization levels and versions of flash-attention, respectively. \| Quantization \| FlashAttn \| Speed (2048 tokens) \| Speed (8192 tokens) \| \| ------------- \| :-------: \| :------------------:\| :------------------:\| \| BF16 \| v2 \| 40.93 \| 36.14 \| \| Int8 \| v2 \| 37.47 \| 32.54 \| \| Int4 \| v2 \| 50.09 \| 38.61 \| \| BF16 \| v1 \| 40.75 \| 35.34 \| \| Int8 \| v1 \| 37.51 \| 32.39 \| \| Int4 \| v1 \| 45.98 \| 36.47 \| \| BF16 \| Disabled \| 37.55 \| 33.56 \| \| Int8 \| Disabled \| 37.84 \| 32.65 \| \| Int4 \| Disabled \| 48.12 \| 36.70 \| 具体而言，我们记录在长度为1的上下文的条件下生成8192个token的性能。评测运行于单张A100-SXM4-80G GPU，使用PyTorch 2.0.1和CUDA 11.8。推理速度是生成8192个token的速度均值。 In detail, the setting of profiling is generating 8192 new tokens with 1 context token. The profiling runs on a single A100-SXM4-80G GPU with PyTorch 2.0.1 and CUDA 11.8. The inference speed is averaged over the generated 8192 tokens. 注意：以上Int4/Int8模型生成速度使用autogptq库给出，当前``AutoModelForCausalLM.from_pretrained``载入的模型生成速度会慢大约20%。我们已经将该问题汇报给HuggingFace团队，若有解决方案将即时更新。 Note: The generation speed of the Int4/Int8 models mentioned above is provided by the autogptq library. The current speed of the model loaded using "AutoModelForCausalLM.from_pretrained" will be approximately 20% slower. We have reported this issue to the HuggingFace team and will update it promptly if a solution is available. ### 显存使用 (GPU Memory Usage) 我们还测算了不同模型精度编码2048个token及生成8192个token的峰值显存占用情况。（显存消耗在是否使用FlashAttn的情况下均类似。）结果如下所示： We also profile the peak GPU memory usage for encoding 2048 tokens as context (and generating single token) and generating 8192 tokens (with single token as context) under different quantization levels, respectively. （The GPU memory usage is similar when using flash-attention or not.）The results are shown below. \| Quantization Level \| Peak Usage for Encoding 2048 Tokens \| Peak Usage for Generating 8192 Tokens \| \| ------------------ \| :---------------------------------: \| :-----------------------------------: \| \| BF16 \| 16.99GB \| 22.53GB \| \| Int8 \| 11.20GB \| 16.62GB \| \| Int4 \| 8.21GB \| 13.63GB \| 上述性能测算使用[此脚本](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py)完成。 The above speed and memory profiling are conducted using [this script](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py). <br> ## 模型细节（Model）与Qwen-7B预训练模型相同，Qwen-7B-Chat模型规模基本情况如下所示: The details of the model architecture of Qwen-7B-Chat are listed as follows: \| Hyperparameter \| Value \| \|:----------------\|:------:\| \| n_layers \| 32 \| \| n_heads \| 32 \| \| d_model \| 4096 \| \| vocab size \| 151851 \| \| sequence length \| 8192 \| 在位置编码、FFN激活函数和normalization的实现方式上，我们也采用了目前最流行的做法，即RoPE相对位置编码、SwiGLU激活函数、RMSNorm（可选安装flash-attention加速）。在分词器方面，相比目前主流开源模型以中英词表为主，Qwen-7B-Chat使用了约15万token大小的词表。该词表在GPT-4使用的BPE词表`cl100k_base`基础上，对中文、多语言进行了优化，在对中、英、代码数据的高效编解码的基础上，对部分多语言更加友好，方便用户在不扩展词表的情况下对部分语种进行能力增强。词表对数字按单个数字位切分。调用较为高效的[tiktoken分词库](https://github.com/openai/tiktoken)进行分词。 For position encoding, FFN activation function, and normalization calculation methods, we adopt the prevalent practices, i.e., RoPE relative position encoding, SwiGLU for activation function, and RMSNorm for normalization (optional installation of flash-attention for acceleration). For tokenization, compared to the current mainstream open-source models based on Chinese and English vocabularies, Qwen-7B-Chat uses a vocabulary of over 150K tokens. It first considers efficient encoding of Chinese, English, and code data, and is also more friendly to multilingual languages, enabling users to directly enhance the capability of some languages without expanding the vocabulary. It segments numbers by single digit, and calls the [tiktoken](https://github.com/openai/tiktoken) tokenizer library for efficient tokenization. <br> ## 评测效果（Evaluation）对于Qwen-7B-Chat模型，我们同样评测了常规的中文理解（C-Eval）、英文理解（MMLU）、代码（HumanEval）和数学（GSM8K）等权威任务，同时包含了长序列任务的评测结果。由于Qwen-7B-Chat模型经过对齐后，激发了较强的外部系统调用能力，我们还进行了工具使用能力方面的评测。提示：由于硬件和框架造成的舍入误差，复现结果如有波动属于正常现象。 For Qwen-7B-Chat, we also evaluate the model on C-Eval, MMLU, HumanEval, GSM8K, etc., as well as the benchmark evaluation for long-context understanding, and tool usage. Note: Due to rounding errors caused by hardware and framework, differences in reproduced results are possible. ### 中文评测（Chinese Evaluation） #### C-Eval 在[C-Eval](https://arxiv.org/abs/2305.08322)验证集上，我们评价了Qwen-7B-Chat模型的0-shot & 5-shot准确率 We demonstrate the 0-shot & 5-shot accuracy of Qwen-7B-Chat on C-Eval validation set \| Model \| Avg. Acc. \| \|:--------------------------------:\|:---------:\| \| LLaMA2-7B-Chat \| 31.9 \| \| LLaMA2-13B-Chat \| 36.2 \| \| LLaMA2-70B-Chat \| 44.3 \| \| ChatGLM2-6B-Chat \| 52.6 \| \| InternLM-7B-Chat \| 53.6 \| \| Baichuan2-7B-Chat \| 55.6 \| \| Baichuan2-13B-Chat \| 56.7 \| \| Qwen-7B-Chat (original) (0-shot) \| 54.2 \| \| Qwen-7B-Chat (0-shot) \| 59.7 \| \| Qwen-7B-Chat (5-shot) \| 59.3 \| \| Qwen-14B-Chat (0-shot) \| 69.8 \| \| Qwen-14B-Chat (5-shot) \| 71.7 \| C-Eval测试集上，Qwen-7B-Chat模型的zero-shot准确率结果如下： The zero-shot accuracy of Qwen-7B-Chat on C-Eval testing set is provided below: \| Model \| Avg. \| STEM \| Social Sciences \| Humanities \| Others \| \| :---------------------- \| :------: \| :--: \| :-------------: \| :--------: \| :----: \| \| Chinese-Alpaca-Plus-13B \| 41.5 \| 36.6 \| 49.7 \| 43.1 \| 41.2 \| \| Chinese-Alpaca-2-7B \| 40.3 \| - \| - \| - \| - \| \| ChatGLM2-6B-Chat \| 50.1 \| 46.4 \| 60.4 \| 50.6 \| 46.9 \| \| Baichuan-13B-Chat \| 51.5 \| 43.7 \| 64.6 \| 56.2 \| 49.2 \| \| Qwen-7B-Chat (original) \| 54.6 \| 47.8 \| 67.6 \| 59.3 \| 50.6 \| \| Qwen-7B-Chat \| 58.6 \| 53.3 \| 72.1 \| 62.8 \| 52.0 \| \| Qwen-14B-Chat \| 69.1 \| 65.1 \| 80.9 \| 71.2 \| 63.4 \| 在7B规模模型上，经过人类指令对齐的Qwen-7B-Chat模型，准确率在同类相近规模模型中仍然处于前列。 Compared with other pretrained models with comparable model size, the human-aligned Qwen-7B-Chat performs well in C-Eval accuracy. ### 英文评测（English Evaluation） #### MMLU [MMLU](https://arxiv.org/abs/2009.03300)评测集上，Qwen-7B-Chat模型的 0-shot & 5-shot 准确率如下，效果同样在同类对齐模型中同样表现较优。 The 0-shot & 5-shot accuracy of Qwen-7B-Chat on MMLU is provided below. The performance of Qwen-7B-Chat still on the top between other human-aligned models with comparable size. \| Model \| Avg. Acc. \| \|:--------------------------------:\|:---------:\| \| ChatGLM2-6B-Chat \| 46.0 \| \| LLaMA2-7B-Chat \| 46.2 \| \| InternLM-7B-Chat \| 51.1 \| \| Baichuan2-7B-Chat \| 52.9 \| \| LLaMA2-13B-Chat \| 54.6 \| \| Baichuan2-13B-Chat \| 57.3 \| \| LLaMA2-70B-Chat \| 63.8 \| \| Qwen-7B-Chat (original) (0-shot) \| 53.9 \| \| Qwen-7B-Chat (0-shot) \| 55.8 \| \| Qwen-7B-Chat (5-shot) \| 57.0 \| \| Qwen-14B-Chat (0-shot) \| 64.6 \| \| Qwen-14B-Chat (5-shot) \| 66.5 \| ### 代码评测（Coding Evaluation） Qwen-7B-Chat在[HumanEval](https://github.com/openai/human-eval)的zero-shot Pass@1效果如下 The zero-shot Pass@1 of Qwen-7B-Chat on [HumanEval](https://github.com/openai/human-eval) is demonstrated below \| Model \| Pass@1 \| \|:-----------------------:\|:--------:\| \| ChatGLM2-6B-Chat \| 11.0 \| \| LLaMA2-7B-Chat \| 12.2 \| \| Baichuan2-7B-Chat \| 13.4 \| \| InternLM-7B-Chat \| 14.6 \| \| Baichuan2-13B-Chat \| 17.7 \| \| LLaMA2-13B-Chat \| 18.9 \| \| LLaMA2-70B-Chat \| 32.3 \| \| Qwen-7B-Chat (original) \| 24.4 \| \| Qwen-7B-Chat \| 37.2 \| \| Qwen-14B-Chat \| 43.9 \| ### 数学评测（Mathematics Evaluation）在评测数学能力的[GSM8K](https://github.com/openai/grade-school-math)上，Qwen-7B-Chat的准确率结果如下 The accuracy of Qwen-7B-Chat on GSM8K is shown below \| Model \| Acc. \| \|:------------------------------------:\|:--------:\| \| LLaMA2-7B-Chat \| 26.3 \| \| ChatGLM2-6B-Chat \| 28.8 \| \| Baichuan2-7B-Chat \| 32.8 \| \| InternLM-7B-Chat \| 33.0 \| \| LLaMA2-13B-Chat \| 37.1 \| \| Baichuan2-13B-Chat \| 55.3 \| \| LLaMA2-70B-Chat \| 59.3 \| \| Qwen-7B-Chat (original) (0-shot) \| 41.1 \| \| Qwen-7B-Chat (0-shot) \| 50.3 \| \| Qwen-7B-Chat (8-shot) \| 54.1 \| \| Qwen-14B-Chat (0-shot) \| 60.1 \| \| Qwen-14B-Chat (8-shot) \| 59.3 \| ### 长序列评测（Long-Context Understanding）通过NTK插值，LogN注意力缩放可以扩展Qwen-7B-Chat的上下文长度。在长文本摘要数据集[VCSUM](https://arxiv.org/abs/2305.05280)上（文本平均长度在15K左右），Qwen-7B-Chat的Rouge-L结果如下： (若要启用这些技巧，请将config.json里的`use_dynamic_ntk`和`use_logn_attn`设置为true) We introduce NTK-aware interpolation, LogN attention scaling to extend the context length of Qwen-7B-Chat. The Rouge-L results of Qwen-7B-Chat on long-text summarization dataset [VCSUM](https://arxiv.org/abs/2305.05280) (The average length of this dataset is around 15K) are shown below: (To use these tricks, please set `use_dynamic_ntk` and `use_long_attn` to true in config.json.) \| Model \| VCSUM (zh) \| \|:------------------\|:----------:\| \| GPT-3.5-Turbo-16k \| 16.0 \| \| LLama2-7B-Chat \| 0.2 \| \| InternLM-7B-Chat \| 13.0 \| \| ChatGLM2-6B-Chat \| 16.3 \| \| Qwen-7B-Chat \| 16.6 \| ### 工具使用能力的评测（Tool Usage） #### ReAct Prompting 千问支持通过 [ReAct Prompting](https://arxiv.org/abs/2210.03629) 调用插件/工具/API。ReAct 也是 [LangChain](https://python.langchain.com/) 框架采用的主要方式之一。在我们开源的、用于评估工具使用能力的评测基准上，千问的表现如下： Qwen-Chat supports calling plugins/tools/APIs through [ReAct Prompting](https://arxiv.org/abs/2210.03629). ReAct is also one of the main approaches used by the [LangChain](https://python.langchain.com/) framework. In our evaluation benchmark for assessing tool usage capabilities, Qwen-Chat's performance is as follows: <table> <tr> <th colspan="4" align="center">Chinese Tool-Use Benchmark</th> </tr> <tr> <th align="center">Model</th><th align="center">Tool Selection (Acc.↑)</th><th align="center">Tool Input (Rouge-L↑)</th><th align="center">False Positive Error↓</th> </tr> <tr> <td>GPT-4</td><td align="center">95%</td><td align="center">0.90</td><td align="center">15.0%</td> </tr> <tr> <td>GPT-3.5</td><td align="center">85%</td><td align="center">0.88</td><td align="center">75.0%</td> </tr> <tr> <td>Qwen-7B-Chat</td><td align="center">98%</td><td align="center">0.91</td><td align="center">7.3%</td> </tr> <tr> <td>Qwen-14B-Chat</td><td align="center">98%</td><td align="center">0.93</td><td align="center">2.4%</td> </tr> </table> > 评测基准中出现的插件均没有出现在千问的训练集中。该基准评估了模型在多个候选插件中选择正确插件的准确率、传入插件的参数的合理性、以及假阳率。假阳率（False Positive）定义：在处理不该调用插件的请求时，错误地调用了插件。 > The plugins that appear in the evaluation set do not appear in the training set of Qwen. This benchmark evaluates the accuracy of the model in selecting the correct plugin from multiple candidate plugins, the rationality of the parameters passed into the plugin, and the false positive rate. False Positive: Incorrectly invoking a plugin when it should not have been called when responding to a query. ![](assets/react_showcase_001.png) ![](assets/react_showcase_002.png) #### Code Interpreter 为了考察Qwen使用Python Code Interpreter完成数学解题、数据可视化、及文件处理与爬虫等任务的能力，我们专门建设并开源了一个评测这方面能力的[评测基准](https://github.com/QwenLM/Qwen-Agent/tree/main/benchmark)。我们发现Qwen在生成代码的可执行率、结果正确性上均表现较好： To assess Qwen's ability to use the Python Code Interpreter for tasks such as mathematical problem solving, data visualization, and other general-purpose tasks such as file handling and web scraping, we have created and open-sourced a benchmark specifically designed for evaluating these capabilities. You can find the benchmark at this [link](https://github.com/QwenLM/Qwen-Agent/tree/main/benchmark). We have observed that Qwen performs well in terms of code executability and result accuracy when generating code: <table> <tr> <th colspan="4" align="center">Executable Rate of Generated Code (%)</th> </tr> <tr> <th align="center">Model</th><th align="center">Math↑</th><th align="center">Visualization↑</th><th align="center">General↑</th> </tr> <tr> <td>GPT-4</td><td align="center">91.9</td><td align="center">85.9</td><td align="center">82.8</td> </tr> <tr> <td>GPT-3.5</td><td align="center">89.2</td><td align="center">65.0</td><td align="center">74.1</td> </tr> <tr> <td>LLaMA2-7B-Chat</td> <td align="center">41.9</td> <td align="center">33.1</td> <td align="center">24.1 </td> </tr> <tr> <td>LLaMA2-13B-Chat</td> <td align="center">50.0</td> <td align="center">40.5</td> <td align="center">48.3 </td> </tr> <tr> <td>CodeLLaMA-7B-Instruct</td> <td align="center">85.1</td> <td align="center">54.0</td> <td align="center">70.7 </td> </tr> <tr> <td>CodeLLaMA-13B-Instruct</td> <td align="center">93.2</td> <td align="center">55.8</td> <td align="center">74.1 </td> </tr> <tr> <td>InternLM-7B-Chat-v1.1</td> <td align="center">78.4</td> <td align="center">44.2</td> <td align="center">62.1 </td> </tr> <tr> <td>InternLM-20B-Chat</td> <td align="center">70.3</td> <td align="center">44.2</td> <td align="center">65.5 </td> </tr> <tr> <td>Qwen-7B-Chat</td> <td align="center">82.4</td> <td align="center">64.4</td> <td align="center">67.2 </td> </tr> <tr> <td>Qwen-14B-Chat</td> <td align="center">89.2</td> <td align="center">84.1</td> <td align="center">65.5</td> </tr> </table> <table> <tr> <th colspan="4" align="center">Accuracy of Code Execution Results (%)</th> </tr> <tr> <th align="center">Model</th><th align="center">Math↑</th><th align="center">Visualization-Hard↑</th><th align="center">Visualization-Easy↑</th> </tr> <tr> <td>GPT-4</td><td align="center">82.8</td><td align="center">66.7</td><td align="center">60.8</td> </tr> <tr> <td>GPT-3.5</td><td align="center">47.3</td><td align="center">33.3</td><td align="center">55.7</td> </tr> <tr> <td>LLaMA2-7B-Chat</td> <td align="center">3.9</td> <td align="center">14.3</td> <td align="center">39.2 </td> </tr> <tr> <td>LLaMA2-13B-Chat</td> <td align="center">8.3</td> <td align="center">8.3</td> <td align="center">40.5 </td> </tr> <tr> <td>CodeLLaMA-7B-Instruct</td> <td align="center">14.3</td> <td align="center">26.2</td> <td align="center">60.8 </td> </tr> <tr> <td>CodeLLaMA-13B-Instruct</td> <td align="center">28.2</td> <td align="center">27.4</td> <td align="center">62.0 </td> </tr> <tr> <td>InternLM-7B-Chat-v1.1</td> <td align="center">28.5</td> <td align="center">4.8</td> <td align="center">40.5 </td> </tr> <tr> <td>InternLM-20B-Chat</td> <td align="center">34.6</td> <td align="center">21.4</td> <td align="center">45.6 </td> </tr> <tr> <td>Qwen-7B-Chat</td> <td align="center">41.9</td> <td align="center">40.5</td> <td align="center">54.4 </td> </tr> <tr> <td>Qwen-14B-Chat</td> <td align="center">58.4</td> <td align="center">53.6</td> <td align="center">59.5</td> </tr> </table> <p align="center"> <br> <img src="assets/code_interpreter_showcase_001.jpg" /> <br> <p> #### Huggingface Agent 千问还具备作为 [HuggingFace Agent](https://huggingface.co/docs/transformers/transformers_agents) 的能力。它在 Huggingface 提供的run模式评测基准上的表现如下： Qwen-Chat also has the capability to be used as a [HuggingFace Agent](https://huggingface.co/docs/transformers/transformers_agents). Its performance on the run-mode benchmark provided by HuggingFace is as follows: <table> <tr> <th colspan="4" align="center">HuggingFace Agent Benchmark- Run Mode</th> </tr> <tr> <th align="center">Model</th><th align="center">Tool Selection↑</th><th align="center">Tool Used↑</th><th align="center">Code↑</th> </tr> <tr> <td>GPT-4</td><td align="center">100</td><td align="center">100</td><td align="center">97.4</td> </tr> <tr> <td>GPT-3.5</td><td align="center">95.4</td><td align="center">96.3</td><td align="center">87.0</td> </tr> <tr> <td>StarCoder-Base-15B</td><td align="center">86.1</td><td align="center">87.0</td><td align="center">68.9</td> </tr> <tr> <td>StarCoder-15B</td><td align="center">87.0</td><td align="center">88.0</td><td align="center">68.9</td> </tr> <tr> <td>Qwen-7B-Chat</td><td align="center">87.0</td><td align="center">87.0</td><td align="center">71.5</td> </tr> <tr> <td>Qwen-14B-Chat</td><td align="center">93.5</td><td align="center">94.4</td><td align="center">87.0</td> </tr> </table> <table> <tr> <th colspan="4" align="center">HuggingFace Agent Benchmark - Chat Mode</th> </tr> <tr> <th align="center">Model</th><th align="center">Tool Selection↑</th><th align="center">Tool Used↑</th><th align="center">Code↑</th> </tr> <tr> <td>GPT-4</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">98.5</td> </tr> <tr> <td>GPT-3.5</td><td align="center">97.3</td><td align="center">96.8</td><td align="center">89.6</td> </tr> <tr> <td>StarCoder-Base-15B</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">91.1</td> </tr> <tr> <td>StarCoder-15B</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">89.6</td> </tr> <tr> <td>Qwen-7B-Chat</td><td align="center">94.7</td><td align="center">94.7</td><td align="center">85.1</td> </tr> <tr> <td>Qwen-14B-Chat</td><td align="center">97.9</td><td align="center">97.9</td><td align="center">95.5</td> </tr> </table> <br> ## x86 平台 (x86 Platforms) 在酷睿™/至强® 可扩展处理器或 Arc™ GPU 上部署量化模型时，建议使用 [OpenVINO™ Toolkit](https://docs.openvino.ai/2023.3/gen_ai_guide.html)以充分利用硬件，实现更好的推理性能。您可以安装并运行此 [example notebook](https://github.com/openvinotoolkit/openvino_notebooks/tree/main/notebooks/254-llm-chatbot)。相关问题，您可在[OpenVINO repo](https://github.com/openvinotoolkit/openvino_notebooks/issues)中提交。 When deploy on Core™/Xeon® Scalable Processors or with Arc™ GPU, [OpenVINO™ Toolkit](https://docs.openvino.ai/2023.3/gen_ai_guide.html) is recommended. You can install and run this [example notebook](https://github.com/openvinotoolkit/openvino_notebooks/tree/main/notebooks/254-llm-chatbot). For related issues, you are welcome to file an issue at [OpenVINO repo](https://github.com/openvinotoolkit/openvino_notebooks/issues). ## FAQ 如遇到问题，敬请查阅[FAQ](https://github.com/QwenLM/Qwen/blob/main/FAQ_zh.md)以及issue区，如仍无法解决再提交issue。 If you meet problems, please refer to [FAQ](https://github.com/QwenLM/Qwen/blob/main/FAQ.md) and the issues first to search a solution before you launch a new issue. <br> ## 引用 (Citation) 如果你觉得我们的工作对你有帮助，欢迎引用！ If you find our work helpful, feel free to give us a cite. ``` @article{qwen, title={Qwen Technical Report}, author={Jinze Bai and Shuai Bai and Yunfei Chu and Zeyu Cui and Kai Dang and Xiaodong Deng and Yang Fan and Wenbin Ge and Yu Han and Fei Huang and Binyuan Hui and Luo Ji and Mei Li and Junyang Lin and Runji Lin and Dayiheng Liu and Gao Liu and Chengqiang Lu and Keming Lu and Jianxin Ma and Rui Men and Xingzhang Ren and Xuancheng Ren and Chuanqi Tan and Sinan Tan and Jianhong Tu and Peng Wang and Shijie Wang and Wei Wang and Shengguang Wu and Benfeng Xu and Jin Xu and An Yang and Hao Yang and Jian Yang and Shusheng Yang and Yang Yao and Bowen Yu and Hongyi Yuan and Zheng Yuan and Jianwei Zhang and Xingxuan Zhang and Yichang Zhang and Zhenru Zhang and Chang Zhou and Jingren Zhou and Xiaohuan Zhou and Tianhang Zhu}, journal={arXiv preprint arXiv:2309.16609}, year={2023} } ``` <br> ## 使用协议（License Agreement）我们的代码和模型权重对学术研究完全开放，并支持商用。请查看[LICENSE](https://github.com/QwenLM/Qwen/blob/main/Tongyi%20Qianwen%20LICENSE%20AGREEMENT)了解具体的开源协议细节。如需商用，请填写[问卷](https://dashscope.console.aliyun.com/openModelApply/qianwen)申请。 Our code and checkpoints are open to research purpose, and they are allowed for commercial purposes. Check [LICENSE](https://github.com/QwenLM/Qwen/blob/main/Tongyi%20Qianwen%20LICENSE%20AGREEMENT) for more details about the license. If you have requirements for commercial use, please fill out the [form](https://dashscope.console.aliyun.com/openModelApply/qianwen) to apply. <br> ## 联系我们（Contact Us）如果你想给我们的研发团队和产品团队留言，欢迎加入我们的微信群、钉钉群以及Discord！同时，也欢迎通过邮件（[email protected]）联系我们。 If you are interested to leave a message to either our research team or product team, join our Discord or WeChat groups! Also, feel free to send an email to [email protected].	{"language": ["zh", "en"], "license": "other", "license_name": "tongyi-qianwen-license-agreement", "license_link": "https://github.com/QwenLM/Qwen/blob/main/Tongyi%20Qianwen%20LICENSE%20AGREEMENT", "pipeline_tag": "text-generation", "tags": ["qwen"], "inference": false}	task	[ "SUMMARIZATION" ]	43,067
michaelfeil/ct2fast-opus-mt-en-ROMANCE	michaelfeil	translation	[ "transformers", "ctranslate2", "translation", "license:apache-2.0", "endpoints_compatible", "region:us" ]	2023-05-19T00:50:06Z	2023-05-19T00:50:58+00:00	11	1	--- license: apache-2.0 tags: - ctranslate2 - translation --- # # Fast-Inference with Ctranslate2 Speedup inference by 2x-8x using int8 inference in C++ quantized version of [Helsinki-NLP/opus-mt-en-ROMANCE](https://huggingface.co/Helsinki-NLP/opus-mt-en-ROMANCE) ```bash pip install hf-hub-ctranslate2>=1.0.0 ctranslate2>=3.13.0 ``` Converted using ``` ct2-transformers-converter --model Helsinki-NLP/opus-mt-en-ROMANCE --output_dir /home/michael/tmp-ct2fast-opus-mt-en-ROMANCE --force --copy_files README.md generation_config.json tokenizer_config.json vocab.json source.spm .gitattributes target.spm --quantization float16 ``` Checkpoint compatible to [ctranslate2](https://github.com/OpenNMT/CTranslate2) and [hf-hub-ctranslate2](https://github.com/michaelfeil/hf-hub-ctranslate2) - `compute_type=int8_float16` for `device="cuda"` - `compute_type=int8` for `device="cpu"` ```python from hf_hub_ctranslate2 import TranslatorCT2fromHfHub, GeneratorCT2fromHfHub from transformers import AutoTokenizer model_name = "michaelfeil/ct2fast-opus-mt-en-ROMANCE" # use either TranslatorCT2fromHfHub or GeneratorCT2fromHfHub here, depending on model. model = TranslatorCT2fromHfHub( # load in int8 on CUDA model_name_or_path=model_name, device="cuda", compute_type="int8_float16", tokenizer=AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-ROMANCE") ) outputs = model.generate( text=["How do you call a fast Flan-ingo?", "User: How are you doing?"], ) print(outputs) ``` # Licence and other remarks: This is just a quantized version. Licence conditions are intended to be idential to original huggingface repo. # Original description ### opus-mt-en-ROMANCE * source languages: en * target languages: fr,fr_BE,fr_CA,fr_FR,wa,frp,oc,ca,rm,lld,fur,lij,lmo,es,es_AR,es_CL,es_CO,es_CR,es_DO,es_EC,es_ES,es_GT,es_HN,es_MX,es_NI,es_PA,es_PE,es_PR,es_SV,es_UY,es_VE,pt,pt_br,pt_BR,pt_PT,gl,lad,an,mwl,it,it_IT,co,nap,scn,vec,sc,ro,la * OPUS readme: [en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la/README.md) * dataset: opus * model: transformer * pre-processing: normalization + SentencePiece * a sentence initial language token is required in the form of `>>id<<` (id = valid target language ID) * download original weights: [opus-2020-04-21.zip](https://object.pouta.csc.fi/OPUS-MT-models/en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la/opus-2020-04-21.zip) * test set translations: [opus-2020-04-21.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la/opus-2020-04-21.test.txt) * test set scores: [opus-2020-04-21.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la/opus-2020-04-21.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| Tatoeba.en.la \| 50.1 \| 0.693 \|	null	Non_BioNLP	# # Fast-Inference with Ctranslate2 Speedup inference by 2x-8x using int8 inference in C++ quantized version of [Helsinki-NLP/opus-mt-en-ROMANCE](https://huggingface.co/Helsinki-NLP/opus-mt-en-ROMANCE) ```bash pip install hf-hub-ctranslate2>=1.0.0 ctranslate2>=3.13.0 ``` Converted using ``` ct2-transformers-converter --model Helsinki-NLP/opus-mt-en-ROMANCE --output_dir /home/michael/tmp-ct2fast-opus-mt-en-ROMANCE --force --copy_files README.md generation_config.json tokenizer_config.json vocab.json source.spm .gitattributes target.spm --quantization float16 ``` Checkpoint compatible to [ctranslate2](https://github.com/OpenNMT/CTranslate2) and [hf-hub-ctranslate2](https://github.com/michaelfeil/hf-hub-ctranslate2) - `compute_type=int8_float16` for `device="cuda"` - `compute_type=int8` for `device="cpu"` ```python from hf_hub_ctranslate2 import TranslatorCT2fromHfHub, GeneratorCT2fromHfHub from transformers import AutoTokenizer model_name = "michaelfeil/ct2fast-opus-mt-en-ROMANCE" # use either TranslatorCT2fromHfHub or GeneratorCT2fromHfHub here, depending on model. model = TranslatorCT2fromHfHub( # load in int8 on CUDA model_name_or_path=model_name, device="cuda", compute_type="int8_float16", tokenizer=AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-ROMANCE") ) outputs = model.generate( text=["How do you call a fast Flan-ingo?", "User: How are you doing?"], ) print(outputs) ``` # Licence and other remarks: This is just a quantized version. Licence conditions are intended to be idential to original huggingface repo. # Original description ### opus-mt-en-ROMANCE * source languages: en * target languages: fr,fr_BE,fr_CA,fr_FR,wa,frp,oc,ca,rm,lld,fur,lij,lmo,es,es_AR,es_CL,es_CO,es_CR,es_DO,es_EC,es_ES,es_GT,es_HN,es_MX,es_NI,es_PA,es_PE,es_PR,es_SV,es_UY,es_VE,pt,pt_br,pt_BR,pt_PT,gl,lad,an,mwl,it,it_IT,co,nap,scn,vec,sc,ro,la * OPUS readme: [en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la/README.md) * dataset: opus * model: transformer * pre-processing: normalization + SentencePiece * a sentence initial language token is required in the form of `>>id<<` (id = valid target language ID) * download original weights: [opus-2020-04-21.zip](https://object.pouta.csc.fi/OPUS-MT-models/en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la/opus-2020-04-21.zip) * test set translations: [opus-2020-04-21.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la/opus-2020-04-21.test.txt) * test set scores: [opus-2020-04-21.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-fr+fr_BE+fr_CA+fr_FR+wa+frp+oc+ca+rm+lld+fur+lij+lmo+es+es_AR+es_CL+es_CO+es_CR+es_DO+es_EC+es_ES+es_GT+es_HN+es_MX+es_NI+es_PA+es_PE+es_PR+es_SV+es_UY+es_VE+pt+pt_br+pt_BR+pt_PT+gl+lad+an+mwl+it+it_IT+co+nap+scn+vec+sc+ro+la/opus-2020-04-21.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| Tatoeba.en.la \| 50.1 \| 0.693 \|	{"license": "apache-2.0", "tags": ["ctranslate2", "translation"]}	task	[ "TRANSLATION" ]	43,068
haonan-li/bactrian-tr-llama-7b-lora	haonan-li	null	[ "arxiv:2305.15011", "license:mit", "region:us" ]	2023-06-02T06:43:33Z	2023-06-02T06:44:21+00:00	0	0	--- license: mit --- This repo contains a low-rank adapter (LoRA) for LLaMA-7b fit on the [Stanford-Alpaca-52k](https://github.com/tatsu-lab/stanford_alpaca) and [databricks-dolly-15k](https://github.com/databrickslabs/dolly/tree/master/data) data in Turkish. ### Dataset Creation 1. English Instructions: The English instuctions are obtained from [alpaca-52k](https://github.com/tatsu-lab/stanford_alpaca), and [dolly-15k](https://github.com/databrickslabs/dolly/tree/master/data). 2. Instruction Translation: The instructions (and inputs) are translated into the target languages using Google Translation API (conducted on April 2023). 3. Output Generation: We generate output from `gpt-3.5-turbo` for each language (conducted on April 2023). <h3 align="center"> <img src="https://raw.githubusercontent.com/fajri91/eval_picts/master/BactrianX_dataset.jpg" width="950" align="center"> </h3> ### Training Parameters The code for training the model is provided in our [github](https://github.com/mbzuai-nlp/Bactrian-X), which is adapted from [Alpaca-LoRA](https://github.com/tloen/alpaca-lora). This version of the weights was trained with the following hyperparameters: - Epochs: 8 - Batch size: 128 - Cutoff length: 512 - Learning rate: 3e-4 - Lora _r_: 16 - Lora target modules: q_proj, v_proj, That is: ``` python finetune.py \ --base_model='decapoda-research/llama-7b-hf' \ --num_epochs=8 \ --cutoff_len=1024 \ --group_by_length \ --output_dir='./bactrian-tr-7b-lora' \ --lora_target_modules='[q_proj,v_proj]' \ --lora_r=16 \ --micro_batch_size=32 ``` Instructions for running it can be found at https://github.com/MBZUAI-nlp/Bactrian-X. ### Discussion of Biases (1) Translation bias; (2) Potential English-culture bias in the translated dataset. ### Citation Information ``` @misc{li2023bactrianx, title={Bactrian-X : A Multilingual Replicable Instruction-Following Model with Low-Rank Adaptation}, author={Haonan Li and Fajri Koto and Minghao Wu and Alham Fikri Aji and Timothy Baldwin}, year={2023}, eprint={2305.15011}, archivePrefix={arXiv}, primaryClass={cs.CL} } ```	null	Non_BioNLP	This repo contains a low-rank adapter (LoRA) for LLaMA-7b fit on the [Stanford-Alpaca-52k](https://github.com/tatsu-lab/stanford_alpaca) and [databricks-dolly-15k](https://github.com/databrickslabs/dolly/tree/master/data) data in Turkish. ### Dataset Creation 1. English Instructions: The English instuctions are obtained from [alpaca-52k](https://github.com/tatsu-lab/stanford_alpaca), and [dolly-15k](https://github.com/databrickslabs/dolly/tree/master/data). 2. Instruction Translation: The instructions (and inputs) are translated into the target languages using Google Translation API (conducted on April 2023). 3. Output Generation: We generate output from `gpt-3.5-turbo` for each language (conducted on April 2023). <h3 align="center"> <img src="https://raw.githubusercontent.com/fajri91/eval_picts/master/BactrianX_dataset.jpg" width="950" align="center"> </h3> ### Training Parameters The code for training the model is provided in our [github](https://github.com/mbzuai-nlp/Bactrian-X), which is adapted from [Alpaca-LoRA](https://github.com/tloen/alpaca-lora). This version of the weights was trained with the following hyperparameters: - Epochs: 8 - Batch size: 128 - Cutoff length: 512 - Learning rate: 3e-4 - Lora _r_: 16 - Lora target modules: q_proj, v_proj, That is: ``` python finetune.py \ --base_model='decapoda-research/llama-7b-hf' \ --num_epochs=8 \ --cutoff_len=1024 \ --group_by_length \ --output_dir='./bactrian-tr-7b-lora' \ --lora_target_modules='[q_proj,v_proj]' \ --lora_r=16 \ --micro_batch_size=32 ``` Instructions for running it can be found at https://github.com/MBZUAI-nlp/Bactrian-X. ### Discussion of Biases (1) Translation bias; (2) Potential English-culture bias in the translated dataset. ### Citation Information ``` @misc{li2023bactrianx, title={Bactrian-X : A Multilingual Replicable Instruction-Following Model with Low-Rank Adaptation}, author={Haonan Li and Fajri Koto and Minghao Wu and Alham Fikri Aji and Timothy Baldwin}, year={2023}, eprint={2305.15011}, archivePrefix={arXiv}, primaryClass={cs.CL} } ```	{"license": "mit"}	task	[ "TRANSLATION" ]	43,069
Genius1237/xlm-roberta-large-tydip	Genius1237	text-classification	[ "transformers", "pytorch", "safetensors", "xlm-roberta", "text-classification", "en", "hi", "ko", "es", "ta", "fr", "vi", "ru", "af", "hu", "dataset:Genius1237/TyDiP", "license:mit", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-04-20T19:59:32Z	2023-10-15T06:06:49+00:00	786	0	--- datasets: - Genius1237/TyDiP language: - en - hi - ko - es - ta - fr - vi - ru - af - hu license: mit metrics: - accuracy pipeline_tag: text-classification --- # Multilingual Politeness Classification Model This model is based on `xlm-roberta-large` and is finetuned on the English subset of the [TyDiP](https://github.com/Genius1237/TyDiP) dataset as discussed in the original paper [here](https://aclanthology.org/2022.findings-emnlp.420/). ## Languages In the paper, this model was evaluated on English + 9 Languages (Hindi, Korean, Spanish, Tamil, French, Vietnamese, Russian, Afrikaans, Hungarian). Given the model's good performance and XLMR's cross lingual abilities, it is likely that this finetuned model can be used for more languages as well. ## Evaluation The politeness classification accuracy scores on 10 languages from the TyDiP test set are mentioned below. \| lang \| acc \| \|------\|-------\| \| en \| 0.892 \| \| hi \| 0.868 \| \| ko \| 0.784 \| \| es \| 0.84 \| \| ta \| 0.78 \| \| fr \| 0.82 \| \| vi \| 0.844 \| \| ru \| 0.668 \| \| af \| 0.856 \| \| hu \| 0.812 \| ## Usage You can use this model directly with a text-classification pipeline ```python from transformers import pipeline classifier = pipeline(task="text-classification", model="Genius1237/xlm-roberta-large-tydip") sentences = ["Could you please get me a glass of water", "mere liye पानी का एक गिलास ले आओ "] print(classifier(sentences)) # [{'label': 'polite', 'score': 0.9076159000396729}, {'label': 'impolite', 'score': 0.765066385269165}] ``` More advanced usage ```python from transformers import AutoTokenizer, AutoModelForSequenceClassification import torch tokenizer = AutoTokenizer.from_pretrained('Genius1237/xlm-roberta-large-tydip') model = AutoModelForSequenceClassification.from_pretrained('Genius1237/xlm-roberta-large-tydip') text = "Could you please get me a glass of water" encoded_input = tokenizer(text, return_tensors='pt') output = model(**encoded_input) prediction = torch.argmax(output.logits).item() print(model.config.id2label[prediction]) # polite ``` ## Citation ``` @inproceedings{srinivasan-choi-2022-tydip, title = "{T}y{D}i{P}: A Dataset for Politeness Classification in Nine Typologically Diverse Languages", author = "Srinivasan, Anirudh and Choi, Eunsol", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022", month = dec, year = "2022", address = "Abu Dhabi, United Arab Emirates", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2022.findings-emnlp.420", doi = "10.18653/v1/2022.findings-emnlp.420", pages = "5723--5738", abstract = "We study politeness phenomena in nine typologically diverse languages. Politeness is an important facet of communication and is sometimes argued to be cultural-specific, yet existing computational linguistic study is limited to English. We create TyDiP, a dataset containing three-way politeness annotations for 500 examples in each language, totaling 4.5K examples. We evaluate how well multilingual models can identify politeness levels {--} they show a fairly robust zero-shot transfer ability, yet fall short of estimated human accuracy significantly. We further study mapping the English politeness strategy lexicon into nine languages via automatic translation and lexicon induction, analyzing whether each strategy{'}s impact stays consistent across languages. Lastly, we empirically study the complicated relationship between formality and politeness through transfer experiments. We hope our dataset will support various research questions and applications, from evaluating multilingual models to constructing polite multilingual agents.", } ```	null	Non_BioNLP	# Multilingual Politeness Classification Model This model is based on `xlm-roberta-large` and is finetuned on the English subset of the [TyDiP](https://github.com/Genius1237/TyDiP) dataset as discussed in the original paper [here](https://aclanthology.org/2022.findings-emnlp.420/). ## Languages In the paper, this model was evaluated on English + 9 Languages (Hindi, Korean, Spanish, Tamil, French, Vietnamese, Russian, Afrikaans, Hungarian). Given the model's good performance and XLMR's cross lingual abilities, it is likely that this finetuned model can be used for more languages as well. ## Evaluation The politeness classification accuracy scores on 10 languages from the TyDiP test set are mentioned below. \| lang \| acc \| \|------\|-------\| \| en \| 0.892 \| \| hi \| 0.868 \| \| ko \| 0.784 \| \| es \| 0.84 \| \| ta \| 0.78 \| \| fr \| 0.82 \| \| vi \| 0.844 \| \| ru \| 0.668 \| \| af \| 0.856 \| \| hu \| 0.812 \| ## Usage You can use this model directly with a text-classification pipeline ```python from transformers import pipeline classifier = pipeline(task="text-classification", model="Genius1237/xlm-roberta-large-tydip") sentences = ["Could you please get me a glass of water", "mere liye पानी का एक गिलास ले आओ "] print(classifier(sentences)) # [{'label': 'polite', 'score': 0.9076159000396729}, {'label': 'impolite', 'score': 0.765066385269165}] ``` More advanced usage ```python from transformers import AutoTokenizer, AutoModelForSequenceClassification import torch tokenizer = AutoTokenizer.from_pretrained('Genius1237/xlm-roberta-large-tydip') model = AutoModelForSequenceClassification.from_pretrained('Genius1237/xlm-roberta-large-tydip') text = "Could you please get me a glass of water" encoded_input = tokenizer(text, return_tensors='pt') output = model(**encoded_input) prediction = torch.argmax(output.logits).item() print(model.config.id2label[prediction]) # polite ``` ## Citation ``` @inproceedings{srinivasan-choi-2022-tydip, title = "{T}y{D}i{P}: A Dataset for Politeness Classification in Nine Typologically Diverse Languages", author = "Srinivasan, Anirudh and Choi, Eunsol", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022", month = dec, year = "2022", address = "Abu Dhabi, United Arab Emirates", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2022.findings-emnlp.420", doi = "10.18653/v1/2022.findings-emnlp.420", pages = "5723--5738", abstract = "We study politeness phenomena in nine typologically diverse languages. Politeness is an important facet of communication and is sometimes argued to be cultural-specific, yet existing computational linguistic study is limited to English. We create TyDiP, a dataset containing three-way politeness annotations for 500 examples in each language, totaling 4.5K examples. We evaluate how well multilingual models can identify politeness levels {--} they show a fairly robust zero-shot transfer ability, yet fall short of estimated human accuracy significantly. We further study mapping the English politeness strategy lexicon into nine languages via automatic translation and lexicon induction, analyzing whether each strategy{'}s impact stays consistent across languages. Lastly, we empirically study the complicated relationship between formality and politeness through transfer experiments. We hope our dataset will support various research questions and applications, from evaluating multilingual models to constructing polite multilingual agents.", } ```	{"datasets": ["Genius1237/TyDiP"], "language": ["en", "hi", "ko", "es", "ta", "fr", "vi", "ru", "af", "hu"], "license": "mit", "metrics": ["accuracy"], "pipeline_tag": "text-classification"}	task	[ "TRANSLATION" ]	43,070
Sadiya025/english-hindi-colloquial-translator	Sadiya025	null	[ "peft", "safetensors", "llama", "unsloth", "generated_from_trainer", "base_model:unsloth/tinyllama-chat-bnb-4bit", "base_model:adapter:unsloth/tinyllama-chat-bnb-4bit", "license:apache-2.0", "4-bit", "bitsandbytes", "region:us" ]	2025-02-18T19:22:24Z	2025-02-19T18:27:44+00:00	28	0	--- base_model: unsloth/tinyllama-chat-bnb-4bit library_name: peft license: apache-2.0 tags: - unsloth - generated_from_trainer model-index: - name: english-hindi-colloquial-translator results: [] --- # english-hindi-colloquial-translator This model is a fine-tuned version of [unsloth/tinyllama-chat-bnb-4bit](https://huggingface.co/unsloth/tinyllama-chat-bnb-4bit) on the English-Hindi Colloquial Dataset (Sadiya025/english-hindi-colloquial-dataset). It achieves the following results on the evaluation set: - Loss: 12.2561 ## Model description Base Model: unsloth/tinyllama-chat-bnb-4bit Dataset: Sadiya025/english-hindi-colloquial-dataset Task: English-to-Hindi Colloquial Translation Quantization: 4-bit for optimized inference ## How to Use To generate hindi colloquial translations, run `inference.py` from this repository. The script automatically handles model loading, tokenization, and inference. ``` git clone https://huggingface.co/Sadiya025/english-hindi-colloquial-translator cd english-hindi-colloquial-translator python inference.py ``` ## Training and evaluation data The model was fine-tuned on the [Sadiya025/english-hindi-colloquial-dataset](https://huggingface.co/datasets/Sadiya025/english-hindi-colloquial-dataset), which consists of English sentences and their corresponding colloquial Hindi translations. ## Training procedure - Base Model: [unsloth/tinyllama-chat-bnb-4bit](https://huggingface.co/unsloth/tinyllama-chat-bnb-4bit) - Frameworks: - PEFT 0.14.0 - Transformers 4.49.0 - PyTorch 2.6.0+cu124 - Datasets 3.2.0 - Tokenizers 0.21.0 - Fine-tuning Method: Parameter-Efficient Fine-Tuning (PEFT) - Precision: 4-bit quantized model for memory efficiency ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0003 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 2 - num_epochs: 3 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| \|:-------------:\|:-----:\|:----:\|:---------------:\| \| 2.1854 \| 0.2 \| 500 \| 6.7063 \| \| 2.2112 \| 0.4 \| 1000 \| 6.5494 \| \| 2.1413 \| 0.6 \| 1500 \| 6.5787 \| \| 2.1303 \| 0.8 \| 2000 \| 6.6641 \| \| 1.9676 \| 1.0 \| 2500 \| 6.6817 \| \| 1.9816 \| 1.2 \| 3000 \| 6.7549 \| \| 2.2404 \| 1.4 \| 3500 \| 6.7187 \| \| 2.0038 \| 1.6 \| 4000 \| 6.7696 \| \| 1.9079 \| 1.8 \| 4500 \| 6.9118 \| \| 2.1682 \| 2.0 \| 5000 \| 6.9245 \| \| 1.8931 \| 2.2 \| 5500 \| 7.3770 \| \| 1.9293 \| 2.4 \| 6000 \| 8.2341 \| \| 2.0817 \| 2.6 \| 6500 \| 10.7503 \| \| 2.1382 \| 2.8 \| 7000 \| 12.0544 \| \| 1.8867 \| 3.0 \| 7500 \| 12.2561 \| ### Framework versions - PEFT 0.14.0 - Transformers 4.49.0 - Pytorch 2.6.0+cu124 - Datasets 3.2.0 - Tokenizers 0.21.0	null	Non_BioNLP	# english-hindi-colloquial-translator This model is a fine-tuned version of [unsloth/tinyllama-chat-bnb-4bit](https://huggingface.co/unsloth/tinyllama-chat-bnb-4bit) on the English-Hindi Colloquial Dataset (Sadiya025/english-hindi-colloquial-dataset). It achieves the following results on the evaluation set: - Loss: 12.2561 ## Model description Base Model: unsloth/tinyllama-chat-bnb-4bit Dataset: Sadiya025/english-hindi-colloquial-dataset Task: English-to-Hindi Colloquial Translation Quantization: 4-bit for optimized inference ## How to Use To generate hindi colloquial translations, run `inference.py` from this repository. The script automatically handles model loading, tokenization, and inference. ``` git clone https://huggingface.co/Sadiya025/english-hindi-colloquial-translator cd english-hindi-colloquial-translator python inference.py ``` ## Training and evaluation data The model was fine-tuned on the [Sadiya025/english-hindi-colloquial-dataset](https://huggingface.co/datasets/Sadiya025/english-hindi-colloquial-dataset), which consists of English sentences and their corresponding colloquial Hindi translations. ## Training procedure - Base Model: [unsloth/tinyllama-chat-bnb-4bit](https://huggingface.co/unsloth/tinyllama-chat-bnb-4bit) - Frameworks: - PEFT 0.14.0 - Transformers 4.49.0 - PyTorch 2.6.0+cu124 - Datasets 3.2.0 - Tokenizers 0.21.0 - Fine-tuning Method: Parameter-Efficient Fine-Tuning (PEFT) - Precision: 4-bit quantized model for memory efficiency ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0003 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 2 - num_epochs: 3 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| \|:-------------:\|:-----:\|:----:\|:---------------:\| \| 2.1854 \| 0.2 \| 500 \| 6.7063 \| \| 2.2112 \| 0.4 \| 1000 \| 6.5494 \| \| 2.1413 \| 0.6 \| 1500 \| 6.5787 \| \| 2.1303 \| 0.8 \| 2000 \| 6.6641 \| \| 1.9676 \| 1.0 \| 2500 \| 6.6817 \| \| 1.9816 \| 1.2 \| 3000 \| 6.7549 \| \| 2.2404 \| 1.4 \| 3500 \| 6.7187 \| \| 2.0038 \| 1.6 \| 4000 \| 6.7696 \| \| 1.9079 \| 1.8 \| 4500 \| 6.9118 \| \| 2.1682 \| 2.0 \| 5000 \| 6.9245 \| \| 1.8931 \| 2.2 \| 5500 \| 7.3770 \| \| 1.9293 \| 2.4 \| 6000 \| 8.2341 \| \| 2.0817 \| 2.6 \| 6500 \| 10.7503 \| \| 2.1382 \| 2.8 \| 7000 \| 12.0544 \| \| 1.8867 \| 3.0 \| 7500 \| 12.2561 \| ### Framework versions - PEFT 0.14.0 - Transformers 4.49.0 - Pytorch 2.6.0+cu124 - Datasets 3.2.0 - Tokenizers 0.21.0	{"base_model": "unsloth/tinyllama-chat-bnb-4bit", "library_name": "peft", "license": "apache-2.0", "tags": ["unsloth", "generated_from_trainer"], "model-index": [{"name": "english-hindi-colloquial-translator", "results": []}]}	task	[ "TRANSLATION" ]	43,071
SEBIS/legal_t5_small_multitask_it_sv	SEBIS	text2text-generation	[ "transformers", "pytorch", "jax", "t5", "text2text-generation", "translation Italian Swedish model", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:04Z	2021-06-23T11:16:13+00:00	173	0	--- datasets: - dcep europarl jrc-acquis language: Italian Swedish tags: - translation Italian Swedish model widget: - text: Può il Commissario responsabile comunicare al Parlamento in che modo la DG Ricerca garantirà che l’Europa possa svolgere un ruolo di primo piano in questo sforzo globale di ricerca sul diabete? --- # legal_t5_small_multitask_it_sv model Model on translating legal text from Italian to Swedish. It was first released in [this repository](https://github.com/agemagician/LegalTrans). The model is parallely trained on the three parallel corpus with 42 language pair from jrc-acquis, europarl and dcep along with the unsupervised task where the model followed the task of prediction in a masked language model. ## Model description No pretraining is involved in case of legal_t5_small_multitask_it_sv model, rather the unsupervised task is added with all the translation task to realize the multitask learning scenario. ## Intended uses & limitations The model could be used for translation of legal texts from Italian to Swedish. ### How to use Here is how to use this model to translate legal text from Italian to Swedish in PyTorch: ```python from transformers import AutoTokenizer, AutoModelWithLMHead, TranslationPipeline pipeline = TranslationPipeline( model=AutoModelWithLMHead.from_pretrained("SEBIS/legal_t5_small_multitask_it_sv"), tokenizer=AutoTokenizer.from_pretrained(pretrained_model_name_or_path = "SEBIS/legal_t5_small_multitask_it_sv", do_lower_case=False, skip_special_tokens=True), device=0 ) it_text = "Può il Commissario responsabile comunicare al Parlamento in che modo la DG Ricerca garantirà che l’Europa possa svolgere un ruolo di primo piano in questo sforzo globale di ricerca sul diabete?" pipeline([it_text], max_length=512) ``` ## Training data The legal_t5_small_multitask_it_sv model (the supervised task which involved only the corresponding langauge pair and as well as unsupervised task where all of the data of all language pairs were available) model was trained on [JRC-ACQUIS](https://wt-public.emm4u.eu/Acquis/index_2.2.html), [EUROPARL](https://www.statmt.org/europarl/), and [DCEP](https://ec.europa.eu/jrc/en/language-technologies/dcep) dataset consisting of 8 Million parallel texts. ## Training procedure The model was trained on a single TPU Pod V3-8 for 250K steps in total, using sequence length 512 (batch size 4096). It has a total of approximately 220M parameters and was trained using the encoder-decoder architecture. The optimizer used is AdaFactor with inverse square root learning rate schedule. ### Preprocessing An unigram model trained with 88M lines of text from the parallel corpus (of all possible language pairs) to get the vocabulary (with byte pair encoding), which is used with this model. ### Pretraining ## Evaluation results When the model is used for translation test dataset, achieves the following results: Test results : \| Model \| BLEU score \| \|:-----:\|:-----:\| \| legal_t5_small_multitask_it_sv \| 41.523\| ### BibTeX entry and citation info > Created by [Ahmed Elnaggar/@Elnaggar_AI](https://twitter.com/Elnaggar_AI) \| [LinkedIn](https://www.linkedin.com/in/prof-ahmed-elnaggar/)	null	Non_BioNLP	# legal_t5_small_multitask_it_sv model Model on translating legal text from Italian to Swedish. It was first released in [this repository](https://github.com/agemagician/LegalTrans). The model is parallely trained on the three parallel corpus with 42 language pair from jrc-acquis, europarl and dcep along with the unsupervised task where the model followed the task of prediction in a masked language model. ## Model description No pretraining is involved in case of legal_t5_small_multitask_it_sv model, rather the unsupervised task is added with all the translation task to realize the multitask learning scenario. ## Intended uses & limitations The model could be used for translation of legal texts from Italian to Swedish. ### How to use Here is how to use this model to translate legal text from Italian to Swedish in PyTorch: ```python from transformers import AutoTokenizer, AutoModelWithLMHead, TranslationPipeline pipeline = TranslationPipeline( model=AutoModelWithLMHead.from_pretrained("SEBIS/legal_t5_small_multitask_it_sv"), tokenizer=AutoTokenizer.from_pretrained(pretrained_model_name_or_path = "SEBIS/legal_t5_small_multitask_it_sv", do_lower_case=False, skip_special_tokens=True), device=0 ) it_text = "Può il Commissario responsabile comunicare al Parlamento in che modo la DG Ricerca garantirà che l’Europa possa svolgere un ruolo di primo piano in questo sforzo globale di ricerca sul diabete?" pipeline([it_text], max_length=512) ``` ## Training data The legal_t5_small_multitask_it_sv model (the supervised task which involved only the corresponding langauge pair and as well as unsupervised task where all of the data of all language pairs were available) model was trained on [JRC-ACQUIS](https://wt-public.emm4u.eu/Acquis/index_2.2.html), [EUROPARL](https://www.statmt.org/europarl/), and [DCEP](https://ec.europa.eu/jrc/en/language-technologies/dcep) dataset consisting of 8 Million parallel texts. ## Training procedure The model was trained on a single TPU Pod V3-8 for 250K steps in total, using sequence length 512 (batch size 4096). It has a total of approximately 220M parameters and was trained using the encoder-decoder architecture. The optimizer used is AdaFactor with inverse square root learning rate schedule. ### Preprocessing An unigram model trained with 88M lines of text from the parallel corpus (of all possible language pairs) to get the vocabulary (with byte pair encoding), which is used with this model. ### Pretraining ## Evaluation results When the model is used for translation test dataset, achieves the following results: Test results : \| Model \| BLEU score \| \|:-----:\|:-----:\| \| legal_t5_small_multitask_it_sv \| 41.523\| ### BibTeX entry and citation info > Created by [Ahmed Elnaggar/@Elnaggar_AI](https://twitter.com/Elnaggar_AI) \| [LinkedIn](https://www.linkedin.com/in/prof-ahmed-elnaggar/)	{"datasets": ["dcep europarl jrc-acquis"], "language": "Italian Swedish", "tags": ["translation Italian Swedish model"], "widget": [{"text": "Può il Commissario responsabile comunicare al Parlamento in che modo la DG Ricerca garantirà che l’Europa possa svolgere un ruolo di primo piano in questo sforzo globale di ricerca sul diabete?"}]}	task	[ "TRANSLATION" ]	43,072
vngrs-ai/VBART-XLarge-Title-Generation-from-News	vngrs-ai	text2text-generation	[ "transformers", "tf", "safetensors", "mbart", "text2text-generation", "tr", "dataset:vngrs-ai/vngrs-web-corpus", "arxiv:2403.01308", "license:cc-by-nc-sa-4.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-03-12T13:10:39Z	2024-03-23T08:37:27+00:00	21	0	--- datasets: - vngrs-ai/vngrs-web-corpus language: - tr library_name: transformers license: cc-by-nc-sa-4.0 pipeline_tag: text2text-generation inference: parameters: max_new_tokens: 32 arXiv: 2403.01308 --- # VBART Model Card ## Model Description VBART is the first sequence-to-sequence LLM pre-trained on Turkish corpora from scratch on a large scale. It was pre-trained by VNGRS in February 2023. The model is capable of conditional text generation tasks such as text summarization, paraphrasing, and title generation when fine-tuned. It outperforms its multilingual counterparts, albeit being much smaller than other implementations. VBART-XLarge is created by adding extra Transformer layers between the layers of VBART-Large. Hence it was able to transfer learned weights from the smaller model while doublings its number of layers. VBART-XLarge improves the results compared to VBART-Large albeit in small margins. This repository contains fine-tuned TensorFlow and Safetensors weights of VBART for title generation from news body task. - Developed by: [VNGRS-AI](https://vngrs.com/ai/) - Model type: Transformer encoder-decoder based on mBART architecture - Language(s) (NLP): Turkish - License: CC BY-NC-SA 4.0 - Finetuned from: VBART-XLarge - Paper: [arXiv](https://arxiv.org/abs/2403.01308) ## How to Get Started with the Model ```python from transformers import AutoTokenizer, AutoModelForSeq2SeqLM tokenizer = AutoTokenizer.from_pretrained("vngrs-ai/VBART-XLarge-Title-Generation-from-News", model_input_names=['input_ids', 'attention_mask']) # Uncomment the device_map kwarg and delete the closing bracket to use model for inference on GPU model = AutoModelForSeq2SeqLM.from_pretrained("vngrs-ai/VBART-XLarge-Title-Generation-from-News")#, device_map="auto") input_text="..." token_input = tokenizer(input_text, return_tensors="pt")#.to('cuda') outputs = model.generate(token_input) print(tokenizer.decode(outputs[0])) ``` ## Training Details ### Training Data The base model is pre-trained on [vngrs-web-corpus](https://huggingface.co/datasets/vngrs-ai/vngrs-web-corpus). It is curated by cleaning and filtering Turkish parts of [OSCAR-2201](https://huggingface.co/datasets/oscar-corpus/OSCAR-2201) and [mC4](https://huggingface.co/datasets/mc4) datasets. These datasets consist of documents of unstructured web crawl data. More information about the dataset can be found on their respective pages. Data is filtered using a set of heuristics and certain rules, explained in the appendix of our [paper](https://arxiv.org/abs/2403.01308). The fine-tuning dataset is the Turkish sections of [MLSum](https://huggingface.co/datasets/mlsum), [TRNews](https://huggingface.co/datasets/batubayk/TR-News) and [XLSum](https://huggingface.co/datasets/csebuetnlp/xlsum) datasets. ### Limitations This model is fine-tuned for title generation tasks. It is not intended to be used in any other case and can not be fine-tuned to any other task with full performance of the base model. It is also not guaranteed that this model will work without specified prompts. ### Training Procedure Pre-trained for 8 days and for a total of 84B tokens. Finally, finetuned for 15 epochs. #### Hardware - GPUs: 8 x Nvidia A100-80 GB #### Software - TensorFlow #### Hyperparameters ##### Pretraining - Training regime: fp16 mixed precision - Training objective: Sentence permutation and span masking (using mask lengths sampled from Poisson distribution λ=3.5, masking 30% of tokens) - Optimizer : Adam optimizer (β1 = 0.9, β2 = 0.98, Ɛ = 1e-6) - Scheduler: Custom scheduler from the original Transformers paper (20,000 warm-up steps) - Weight Initialization: Model Enlargement from VBART-Large. See the related section in the [paper](https://arxiv.org/abs/2403.01308) for the details. - Dropout: 0.1 (dropped to 0.05 and then to 0 in the last 80K and 80k steps, respectively) - Initial Learning rate: 5e-6 - Training tokens: 84B ##### Fine-tuning - Training regime: fp16 mixed precision - Optimizer : Adam optimizer (β1 = 0.9, β2 = 0.98, Ɛ = 1e-6) - Scheduler: Linear decay scheduler - Dropout: 0.1 - Learning rate: 5e-6 - Fine-tune epochs**: 15 #### Metrics ![image/png](https://cdn-uploads.huggingface.co/production/uploads/62f8b3c84588fe31f435a92b/r2p_Ktnwn6n4Rj1MYrjB4.png) ## Citation ``` @article{turker2024vbart, title={VBART: The Turkish LLM}, author={Turker, Meliksah and Ari, Erdi and Han, Aydin}, journal={arXiv preprint arXiv:2403.01308}, year={2024} } ```	null	Non_BioNLP	# VBART Model Card ## Model Description VBART is the first sequence-to-sequence LLM pre-trained on Turkish corpora from scratch on a large scale. It was pre-trained by VNGRS in February 2023. The model is capable of conditional text generation tasks such as text summarization, paraphrasing, and title generation when fine-tuned. It outperforms its multilingual counterparts, albeit being much smaller than other implementations. VBART-XLarge is created by adding extra Transformer layers between the layers of VBART-Large. Hence it was able to transfer learned weights from the smaller model while doublings its number of layers. VBART-XLarge improves the results compared to VBART-Large albeit in small margins. This repository contains fine-tuned TensorFlow and Safetensors weights of VBART for title generation from news body task. - Developed by: [VNGRS-AI](https://vngrs.com/ai/) - Model type: Transformer encoder-decoder based on mBART architecture - Language(s) (NLP): Turkish - License: CC BY-NC-SA 4.0 - Finetuned from: VBART-XLarge - Paper: [arXiv](https://arxiv.org/abs/2403.01308) ## How to Get Started with the Model ```python from transformers import AutoTokenizer, AutoModelForSeq2SeqLM tokenizer = AutoTokenizer.from_pretrained("vngrs-ai/VBART-XLarge-Title-Generation-from-News", model_input_names=['input_ids', 'attention_mask']) # Uncomment the device_map kwarg and delete the closing bracket to use model for inference on GPU model = AutoModelForSeq2SeqLM.from_pretrained("vngrs-ai/VBART-XLarge-Title-Generation-from-News")#, device_map="auto") input_text="..." token_input = tokenizer(input_text, return_tensors="pt")#.to('cuda') outputs = model.generate(token_input) print(tokenizer.decode(outputs[0])) ``` ## Training Details ### Training Data The base model is pre-trained on [vngrs-web-corpus](https://huggingface.co/datasets/vngrs-ai/vngrs-web-corpus). It is curated by cleaning and filtering Turkish parts of [OSCAR-2201](https://huggingface.co/datasets/oscar-corpus/OSCAR-2201) and [mC4](https://huggingface.co/datasets/mc4) datasets. These datasets consist of documents of unstructured web crawl data. More information about the dataset can be found on their respective pages. Data is filtered using a set of heuristics and certain rules, explained in the appendix of our [paper](https://arxiv.org/abs/2403.01308). The fine-tuning dataset is the Turkish sections of [MLSum](https://huggingface.co/datasets/mlsum), [TRNews](https://huggingface.co/datasets/batubayk/TR-News) and [XLSum](https://huggingface.co/datasets/csebuetnlp/xlsum) datasets. ### Limitations This model is fine-tuned for title generation tasks. It is not intended to be used in any other case and can not be fine-tuned to any other task with full performance of the base model. It is also not guaranteed that this model will work without specified prompts. ### Training Procedure Pre-trained for 8 days and for a total of 84B tokens. Finally, finetuned for 15 epochs. #### Hardware - GPUs: 8 x Nvidia A100-80 GB #### Software - TensorFlow #### Hyperparameters ##### Pretraining - Training regime: fp16 mixed precision - Training objective: Sentence permutation and span masking (using mask lengths sampled from Poisson distribution λ=3.5, masking 30% of tokens) - Optimizer : Adam optimizer (β1 = 0.9, β2 = 0.98, Ɛ = 1e-6) - Scheduler: Custom scheduler from the original Transformers paper (20,000 warm-up steps) - Weight Initialization: Model Enlargement from VBART-Large. See the related section in the [paper](https://arxiv.org/abs/2403.01308) for the details. - Dropout: 0.1 (dropped to 0.05 and then to 0 in the last 80K and 80k steps, respectively) - Initial Learning rate: 5e-6 - Training tokens: 84B ##### Fine-tuning - Training regime: fp16 mixed precision - Optimizer : Adam optimizer (β1 = 0.9, β2 = 0.98, Ɛ = 1e-6) - Scheduler: Linear decay scheduler - Dropout: 0.1 - Learning rate: 5e-6 - Fine-tune epochs**: 15 #### Metrics ![image/png](https://cdn-uploads.huggingface.co/production/uploads/62f8b3c84588fe31f435a92b/r2p_Ktnwn6n4Rj1MYrjB4.png) ## Citation ``` @article{turker2024vbart, title={VBART: The Turkish LLM}, author={Turker, Meliksah and Ari, Erdi and Han, Aydin}, journal={arXiv preprint arXiv:2403.01308}, year={2024} } ```	{"datasets": ["vngrs-ai/vngrs-web-corpus"], "language": ["tr"], "library_name": "transformers", "license": "cc-by-nc-sa-4.0", "pipeline_tag": "text2text-generation", "inference": {"parameters": {"max_new_tokens": 32}}, "arXiv": 2403.01308}	task	[ "SUMMARIZATION", "PARAPHRASING" ]	43,073
Helsinki-NLP/opus-mt-fr-lus	Helsinki-NLP	translation	[ "transformers", "pytorch", "tf", "marian", "text2text-generation", "translation", "fr", "lus", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:04Z	2023-08-16T11:36:49+00:00	46	0	--- license: apache-2.0 tags: - translation --- ### opus-mt-fr-lus * source languages: fr * target languages: lus * OPUS readme: [fr-lus](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/fr-lus/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-09.zip](https://object.pouta.csc.fi/OPUS-MT-models/fr-lus/opus-2020-01-09.zip) * test set translations: [opus-2020-01-09.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/fr-lus/opus-2020-01-09.test.txt) * test set scores: [opus-2020-01-09.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/fr-lus/opus-2020-01-09.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.fr.lus \| 25.5 \| 0.455 \|	null	Non_BioNLP	### opus-mt-fr-lus * source languages: fr * target languages: lus * OPUS readme: [fr-lus](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/fr-lus/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-09.zip](https://object.pouta.csc.fi/OPUS-MT-models/fr-lus/opus-2020-01-09.zip) * test set translations: [opus-2020-01-09.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/fr-lus/opus-2020-01-09.test.txt) * test set scores: [opus-2020-01-09.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/fr-lus/opus-2020-01-09.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.fr.lus \| 25.5 \| 0.455 \|	{"license": "apache-2.0", "tags": ["translation"]}	task	[ "TRANSLATION" ]	43,074
solidrust/NeuralHyperion-2.0-Mistral-7B-AWQ	solidrust	text-generation	[ "transformers", "safetensors", "mistral", "text-generation", "code", "chemistry", "medical", "quantized", "4-bit", "AWQ", "autotrain_compatible", "endpoints_compatible", "chatml", "en", "dataset:Locutusque/hyperion-v2.0", "dataset:argilla/distilabel-capybara-dpo-7k-binarized", "base_model:Locutusque/NeuralHyperion-2.0-Mistral-7B", "base_model:quantized:Locutusque/NeuralHyperion-2.0-Mistral-7B", "license:apache-2.0", "text-generation-inference", "awq", "region:us" ]	2024-03-12T18:14:09Z	2024-09-03T08:08:13+00:00	8	0	--- base_model: Locutusque/NeuralHyperion-2.0-Mistral-7B datasets: - Locutusque/hyperion-v2.0 - argilla/distilabel-capybara-dpo-7k-binarized language: - en library_name: transformers license: apache-2.0 model_name: Darewin-7B pipeline_tag: text-generation tags: - code - chemistry - medical - quantized - 4-bit - AWQ - text-generation - autotrain_compatible - endpoints_compatible - chatml model_creator: Locutusque model_type: mistral inference: false prompt_template: '<\|im_start\|>system {system_message}<\|im_end\|> <\|im_start\|>user {prompt}<\|im_end\|> <\|im_start\|>assistant ' quantized_by: Suparious --- # Locutusque/NeuralHyperion-2.0-Mistral-7B AWQ - Model creator: [Locutusque](https://huggingface.co/Locutusque) - Original model: [NeuralHyperion-2.0-Mistral-7B](https://huggingface.co/Locutusque/NeuralHyperion-2.0-Mistral-7B) ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6437292ecd93f4c9a34b0d47/9BU30Mh9bOkO2HRBDF8EE.png) ## Model Summary `Locutusque/NeuralHyperion-2.0-Mistral-7B` is a state-of-the-art language model fine-tuned on the Hyperion-v2.0 and distilabel-capybara dataset for advanced reasoning across scientific domains. This model is designed to handle complex inquiries and instructions, leveraging the diverse and rich information contained in the Hyperion dataset. Its primary use cases include but are not limited to complex question answering, conversational understanding, code generation, medical text comprehension, mathematical reasoning, and logical reasoning. ## How to use ### Install the necessary packages ```bash pip install --upgrade autoawq autoawq-kernels ``` ### Example Python code ```python from awq import AutoAWQForCausalLM from transformers import AutoTokenizer, TextStreamer model_path = "solidrust/NeuralHyperion-2.0-Mistral-7B-AWQ" system_message = "You are Hyperion, incarnated as a powerful AI." # Load model model = AutoAWQForCausalLM.from_quantized(model_path, fuse_layers=True) tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True) streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True) # Convert prompt to tokens prompt_template = """\ <\|im_start\|>system {system_message}<\|im_end\|> <\|im_start\|>user {prompt}<\|im_end\|> <\|im_start\|>assistant""" prompt = "You're standing on the surface of the Earth. "\ "You walk one mile south, one mile west and one mile north. "\ "You end up exactly where you started. Where are you?" tokens = tokenizer(prompt_template.format(system_message=system_message,prompt=prompt), return_tensors='pt').input_ids.cuda() # Generate output generation_output = model.generate(tokens, streamer=streamer, max_new_tokens=512) ``` ### About AWQ AWQ is an efficient, accurate and blazing-fast low-bit weight quantization method, currently supporting 4-bit quantization. Compared to GPTQ, it offers faster Transformers-based inference with equivalent or better quality compared to the most commonly used GPTQ settings. AWQ models are currently supported on Linux and Windows, with NVidia GPUs only. macOS users: please use GGUF models instead. It is supported by: - [Text Generation Webui](https://github.com/oobabooga/text-generation-webui) - using Loader: AutoAWQ - [vLLM](https://github.com/vllm-project/vllm) - version 0.2.2 or later for support for all model types. - [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) - [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later, from any code or client that supports Transformers - [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) - for use from Python code ## Prompt template: ChatML ```plaintext <\|im_start\|>system {system_message}<\|im_end\|> <\|im_start\|>user {prompt}<\|im_end\|> <\|im_start\|>assistant ```	null	Non_BioNLP	# Locutusque/NeuralHyperion-2.0-Mistral-7B AWQ - Model creator: [Locutusque](https://huggingface.co/Locutusque) - Original model: [NeuralHyperion-2.0-Mistral-7B](https://huggingface.co/Locutusque/NeuralHyperion-2.0-Mistral-7B) ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6437292ecd93f4c9a34b0d47/9BU30Mh9bOkO2HRBDF8EE.png) ## Model Summary `Locutusque/NeuralHyperion-2.0-Mistral-7B` is a state-of-the-art language model fine-tuned on the Hyperion-v2.0 and distilabel-capybara dataset for advanced reasoning across scientific domains. This model is designed to handle complex inquiries and instructions, leveraging the diverse and rich information contained in the Hyperion dataset. Its primary use cases include but are not limited to complex question answering, conversational understanding, code generation, medical text comprehension, mathematical reasoning, and logical reasoning. ## How to use ### Install the necessary packages ```bash pip install --upgrade autoawq autoawq-kernels ``` ### Example Python code ```python from awq import AutoAWQForCausalLM from transformers import AutoTokenizer, TextStreamer model_path = "solidrust/NeuralHyperion-2.0-Mistral-7B-AWQ" system_message = "You are Hyperion, incarnated as a powerful AI." # Load model model = AutoAWQForCausalLM.from_quantized(model_path, fuse_layers=True) tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True) streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True) # Convert prompt to tokens prompt_template = """\ <\|im_start\|>system {system_message}<\|im_end\|> <\|im_start\|>user {prompt}<\|im_end\|> <\|im_start\|>assistant""" prompt = "You're standing on the surface of the Earth. "\ "You walk one mile south, one mile west and one mile north. "\ "You end up exactly where you started. Where are you?" tokens = tokenizer(prompt_template.format(system_message=system_message,prompt=prompt), return_tensors='pt').input_ids.cuda() # Generate output generation_output = model.generate(tokens, streamer=streamer, max_new_tokens=512) ``` ### About AWQ AWQ is an efficient, accurate and blazing-fast low-bit weight quantization method, currently supporting 4-bit quantization. Compared to GPTQ, it offers faster Transformers-based inference with equivalent or better quality compared to the most commonly used GPTQ settings. AWQ models are currently supported on Linux and Windows, with NVidia GPUs only. macOS users: please use GGUF models instead. It is supported by: - [Text Generation Webui](https://github.com/oobabooga/text-generation-webui) - using Loader: AutoAWQ - [vLLM](https://github.com/vllm-project/vllm) - version 0.2.2 or later for support for all model types. - [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) - [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later, from any code or client that supports Transformers - [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) - for use from Python code ## Prompt template: ChatML ```plaintext <\|im_start\|>system {system_message}<\|im_end\|> <\|im_start\|>user {prompt}<\|im_end\|> <\|im_start\|>assistant ```	{"base_model": "Locutusque/NeuralHyperion-2.0-Mistral-7B", "datasets": ["Locutusque/hyperion-v2.0", "argilla/distilabel-capybara-dpo-7k-binarized"], "language": ["en"], "library_name": "transformers", "license": "apache-2.0", "model_name": "Darewin-7B", "pipeline_tag": "text-generation", "tags": ["code", "chemistry", "medical", "quantized", "4-bit", "AWQ", "text-generation", "autotrain_compatible", "endpoints_compatible", "chatml"], "model_creator": "Locutusque", "model_type": "mistral", "inference": false, "prompt_template": "<\|im_start\|>system\n{system_message}<\|im_end\|>\n<\|im_start\|>user\n{prompt}<\|im_end\|>\n<\|im_start\|>assistant\n", "quantized_by": "Suparious"}	task	[ "QUESTION_ANSWERING" ]	43,075
RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf	RichardErkhov	null	[ "gguf", "arxiv:2204.05149", "endpoints_compatible", "region:us", "conversational" ]	2024-10-03T04:34:09Z	2024-10-03T05:31:26+00:00	69	0	--- {} --- Quantization made by Richard Erkhov. [Github](https://github.com/RichardErkhov) [Discord](https://discord.gg/pvy7H8DZMG) [Request more models](https://github.com/RichardErkhov/quant_request) Llama-3.2-3B-Instruct - GGUF - Model creator: https://huggingface.co/alpindale/ - Original model: https://huggingface.co/alpindale/Llama-3.2-3B-Instruct/ \| Name \| Quant method \| Size \| \| ---- \| ---- \| ---- \| \| [Llama-3.2-3B-Instruct.Q2_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q2_K.gguf) \| Q2_K \| 1.27GB \| \| [Llama-3.2-3B-Instruct.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ3_XS.gguf) \| IQ3_XS \| 1.38GB \| \| [Llama-3.2-3B-Instruct.IQ3_S.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ3_S.gguf) \| IQ3_S \| 1.44GB \| \| [Llama-3.2-3B-Instruct.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q3_K_S.gguf) \| Q3_K_S \| 1.44GB \| \| [Llama-3.2-3B-Instruct.IQ3_M.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ3_M.gguf) \| IQ3_M \| 1.49GB \| \| [Llama-3.2-3B-Instruct.Q3_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q3_K.gguf) \| Q3_K \| 1.57GB \| \| [Llama-3.2-3B-Instruct.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q3_K_M.gguf) \| Q3_K_M \| 1.57GB \| \| [Llama-3.2-3B-Instruct.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q3_K_L.gguf) \| Q3_K_L \| 1.69GB \| \| [Llama-3.2-3B-Instruct.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ4_XS.gguf) \| IQ4_XS \| 1.71GB \| \| [Llama-3.2-3B-Instruct.Q4_0.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_0.gguf) \| Q4_0 \| 1.79GB \| \| [Llama-3.2-3B-Instruct.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ4_NL.gguf) \| IQ4_NL \| 1.79GB \| \| [Llama-3.2-3B-Instruct.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_K_S.gguf) \| Q4_K_S \| 1.8GB \| \| [Llama-3.2-3B-Instruct.Q4_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_K.gguf) \| Q4_K \| 1.88GB \| \| [Llama-3.2-3B-Instruct.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_K_M.gguf) \| Q4_K_M \| 1.88GB \| \| [Llama-3.2-3B-Instruct.Q4_1.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_1.gguf) \| Q4_1 \| 1.95GB \| \| [Llama-3.2-3B-Instruct.Q5_0.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_0.gguf) \| Q5_0 \| 2.11GB \| \| [Llama-3.2-3B-Instruct.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_K_S.gguf) \| Q5_K_S \| 2.11GB \| \| [Llama-3.2-3B-Instruct.Q5_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_K.gguf) \| Q5_K \| 2.16GB \| \| [Llama-3.2-3B-Instruct.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_K_M.gguf) \| Q5_K_M \| 2.16GB \| \| [Llama-3.2-3B-Instruct.Q5_1.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_1.gguf) \| Q5_1 \| 2.28GB \| \| [Llama-3.2-3B-Instruct.Q6_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q6_K.gguf) \| Q6_K \| 2.46GB \| \| [Llama-3.2-3B-Instruct.Q8_0.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q8_0.gguf) \| Q8_0 \| 3.19GB \| Original model description: --- language: - en - de - fr - it - pt - hi - es - th library_name: transformers pipeline_tag: text-generation tags: - facebook - meta - pytorch - llama - llama-3 license: llama3.2 extra_gated_prompt: >- ### LLAMA 3.2 COMMUNITY LICENSE AGREEMENT Llama 3.2 Version Release Date: September 25, 2024 “Agreement” means the terms and conditions for use, reproduction, distribution and modification of the Llama Materials set forth herein. “Documentation” means the specifications, manuals and documentation accompanying Llama 3.2 distributed by Meta at https://llama.meta.com/doc/overview. “Licensee” or “you” means you, or your employer or any other person or entity (if you are entering into this Agreement on such person or entity’s behalf), of the age required under applicable laws, rules or regulations to provide legal consent and that has legal authority to bind your employer or such other person or entity if you are entering in this Agreement on their behalf. “Llama 3.2” means the foundational large language models and software and algorithms, including machine-learning model code, trained model weights, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by Meta at https://www.llama.com/llama-downloads. “Llama Materials” means, collectively, Meta’s proprietary Llama 3.2 and Documentation (and any portion thereof) made available under this Agreement. “Meta” or “we” means Meta Platforms Ireland Limited (if you are located in or, if you are an entity, your principal place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if you are located outside of the EEA or Switzerland). By clicking “I Accept” below or by using or distributing any portion or element of the Llama Materials, you agree to be bound by this Agreement. 1. License Rights and Redistribution. a. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under Meta’s intellectual property or other rights owned by Meta embodied in the Llama Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Llama Materials. b. Redistribution and Use. i. If you distribute or make available the Llama Materials (or any derivative works thereof), or a product or service (including another AI model) that contains any of them, you shall (A) provide a copy of this Agreement with any such Llama Materials; and (B) prominently display “Built with Llama” on a related website, user interface, blogpost, about page, or product documentation. If you use the Llama Materials or any outputs or results of the Llama Materials to create, train, fine tune, or otherwise improve an AI model, which is distributed or made available, you shall also include “Llama” at the beginning of any such AI model name. ii. If you receive Llama Materials, or any derivative works thereof, from a Licensee as part of an integrated end user product, then Section 2 of this Agreement will not apply to you. iii. You must retain in all copies of the Llama Materials that you distribute the following attribution notice within a “Notice” text file distributed as a part of such copies: “Llama 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.” iv. Your use of the Llama Materials must comply with applicable laws and regulations (including trade compliance laws and regulations) and adhere to the Acceptable Use Policy for the Llama Materials (available at https://www.llama.com/llama3_2/use-policy), which is hereby incorporated by reference into this Agreement. 2. Additional Commercial Terms. If, on the Llama 3.2 version release date, the monthly active users of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 700 million monthly active users in the preceding calendar month, you must request a license from Meta, which Meta may grant to you in its sole discretion, and you are not authorized to exercise any of the rights under this Agreement unless or until Meta otherwise expressly grants you such rights. 3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, AND META DISCLAIMS ALL WARRANTIES OF ANY KIND, BOTH EXPRESS AND IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS. 4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING. 5. Intellectual Property. a. No trademark licenses are granted under this Agreement, and in connection with the Llama Materials, neither Meta nor Licensee may use any name or mark owned by or associated with the other or any of its affiliates, except as required for reasonable and customary use in describing and redistributing the Llama Materials or as set forth in this Section 5(a). Meta hereby grants you a license to use “Llama” (the “Mark”) solely as required to comply with the last sentence of Section 1.b.i. You will comply with Meta’s brand guidelines (currently accessible at https://about.meta.com/brand/resources/meta/company-brand/). All goodwill arising out of your use of the Mark will inure to the benefit of Meta. b. Subject to Meta’s ownership of Llama Materials and derivatives made by or for Meta, with respect to any derivative works and modifications of the Llama Materials that are made by you, as between you and Meta, you are and will be the owner of such derivative works and modifications. c. If you institute litigation or other proceedings against Meta or any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Llama Materials or Llama 3.2 outputs or results, or any portion of any of the foregoing, constitutes infringement of intellectual property or other rights owned or licensable by you, then any licenses granted to you under this Agreement shall terminate as of the date such litigation or claim is filed or instituted. You will indemnify and hold harmless Meta from and against any claim by any third party arising out of or related to your use or distribution of the Llama Materials. 6. Term and Termination. The term of this Agreement will commence upon your acceptance of this Agreement or access to the Llama Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein. Meta may terminate this Agreement if you are in breach of any term or condition of this Agreement. Upon termination of this Agreement, you shall delete and cease use of the Llama Materials. Sections 3, 4 and 7 shall survive the termination of this Agreement. 7. Governing Law and Jurisdiction. This Agreement will be governed and construed under the laws of the State of California without regard to choice of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement. The courts of California shall have exclusive jurisdiction of any dispute arising out of this Agreement. ### Llama 3.2 Acceptable Use Policy Meta is committed to promoting safe and fair use of its tools and features, including Llama 3.2. If you access or use Llama 3.2, you agree to this Acceptable Use Policy (“Policy”). The most recent copy of this policy can be found at [https://www.llama.com/llama3_2/use-policy](https://www.llama.com/llama3_2/use-policy). #### Prohibited Uses We want everyone to use Llama 3.2 safely and responsibly. You agree you will not use, or allow others to use, Llama 3.2 to: 1. Violate the law or others’ rights, including to: 1. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as: 1. Violence or terrorism 2. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material 3. Human trafficking, exploitation, and sexual violence 4. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials. 5. Sexual solicitation 6. Any other criminal activity 1. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals 2. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services 3. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices 4. Collect, process, disclose, generate, or infer private or sensitive information about individuals, including information about individuals’ identity, health, or demographic information, unless you have obtained the right to do so in accordance with applicable law 5. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama Materials 6. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system 7. Engage in any action, or facilitate any action, to intentionally circumvent or remove usage restrictions or other safety measures, or to enable functionality disabled by Meta 2. Engage in, promote, incite, facilitate, or assist in the planning or development of activities that present a risk of death or bodily harm to individuals, including use of Llama 3.2 related to the following: 8. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State or to the U.S. Biological Weapons Anti-Terrorism Act of 1989 or the Chemical Weapons Convention Implementation Act of 1997 9. Guns and illegal weapons (including weapon development) 10. Illegal drugs and regulated/controlled substances 11. Operation of critical infrastructure, transportation technologies, or heavy machinery 12. Self-harm or harm to others, including suicide, cutting, and eating disorders 13. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual 3. Intentionally deceive or mislead others, including use of Llama 3.2 related to the following: 14. Generating, promoting, or furthering fraud or the creation or promotion of disinformation 15. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content 16. Generating, promoting, or further distributing spam 17. Impersonating another individual without consent, authorization, or legal right 18. Representing that the use of Llama 3.2 or outputs are human-generated 19. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement 4. Fail to appropriately disclose to end users any known dangers of your AI system 5. Interact with third party tools, models, or software designed to generate unlawful content or engage in unlawful or harmful conduct and/or represent that the outputs of such tools, models, or software are associated with Meta or Llama 3.2 With respect to any multimodal models included in Llama 3.2, the rights granted under Section 1(a) of the Llama 3.2 Community License Agreement are not being granted to you if you are an individual domiciled in, or a company with a principal place of business in, the European Union. This restriction does not apply to end users of a product or service that incorporates any such multimodal models. Please report any violation of this Policy, software “bug,” or other problems that could lead to a violation of this Policy through one of the following means: * Reporting issues with the model: [https://github.com/meta-llama/llama-models/issues](https://l.workplace.com/l.php?u=https%3A%2F%2Fgithub.com%2Fmeta-llama%2Fllama-models%2Fissues&h=AT0qV8W9BFT6NwihiOHRuKYQM_UnkzN_NmHMy91OT55gkLpgi4kQupHUl0ssR4dQsIQ8n3tfd0vtkobvsEvt1l4Ic6GXI2EeuHV8N08OG2WnbAmm0FL4ObkazC6G_256vN0lN9DsykCvCqGZ) * Reporting risky content generated by the model: [developers.facebook.com/llama_output_feedback](http://developers.facebook.com/llama_output_feedback) * Reporting bugs and security concerns: [facebook.com/whitehat/info](http://facebook.com/whitehat/info) * Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama 3.2: [email protected] extra_gated_fields: First Name: text Last Name: text Date of birth: date_picker Country: country Affiliation: text Job title: type: select options: - Student - Research Graduate - AI researcher - AI developer/engineer - Reporter - Other geo: ip_location By clicking Submit below I accept the terms of the license and acknowledge that the information I provide will be collected stored processed and shared in accordance with the Meta Privacy Policy: checkbox extra_gated_description: >- The information you provide will be collected, stored, processed and shared in accordance with the [Meta Privacy Policy](https://www.facebook.com/privacy/policy/). extra_gated_button_content: Submit --- ## Model Information The Meta Llama 3.2 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction-tuned generative models in 1B and 3B sizes (text in/text out). The Llama 3.2 instruction-tuned text only models are optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. They outperform many of the available open source and closed chat models on common industry benchmarks. Model Developer: Meta Model Architecture: Llama 3.2 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align with human preferences for helpfulness and safety. \| \| Training Data \| Params \| Input modalities \| Output modalities \| Context Length \| GQA \| Shared Embeddings \| Token count \| Knowledge cutoff \| \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| \| Llama 3.2 (text only) \| A new mix of publicly available online data. \| 1B (1.23B) \| Multilingual Text \| Multilingual Text and code \| 128k \| Yes \| Yes \| Up to 9T tokens \| December 2023 \| \| \| \| 3B (3.21B) \| Multilingual Text \| Multilingual Text and code \| \| \| \| \| \| Supported Languages: English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai are officially supported. Llama 3.2 has been trained on a broader collection of languages than these 8 supported languages. Developers may fine-tune Llama 3.2 models for languages beyond these supported languages, provided they comply with the Llama 3.2 Community License and the Acceptable Use Policy. Developers are always expected to ensure that their deployments, including those that involve additional languages, are completed safely and responsibly. Llama 3.2 Model Family: Token counts refer to pretraining data only. All model versions use Grouped-Query Attention (GQA) for improved inference scalability. Model Release Date: Sept 25, 2024 Status: This is a static model trained on an offline dataset. Future versions may be released that improve model capabilities and safety. License: Use of Llama 3.2 is governed by the [Llama 3.2 Community License](https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/LICENSE) (a custom, commercial license agreement). Feedback: Where to send questions or comments about the model Instructions on how to provide feedback or comments on the model can be found in the model [README](https://github.com/meta-llama/llama-models/tree/main/models/llama3_2). For more technical information about generation parameters and recipes for how to use Llama 3.2 in applications, please go [here](https://github.com/meta-llama/llama-recipes). ## Intended Use Intended Use Cases: Llama 3.2 is intended for commercial and research use in multiple languages. Instruction tuned text only models are intended for assistant-like chat and agentic applications like knowledge retrieval and summarization, mobile AI powered writing assistants and query and prompt rewriting. Pretrained models can be adapted for a variety of additional natural language generation tasks. Out of Scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Acceptable Use Policy and Llama 3.2 Community License. Use in languages beyond those explicitly referenced as supported in this model card. ## How to use This repository contains two versions of Llama-3.2-3B-Instruct, for use with `transformers` and with the original `llama` codebase. ### Use with transformers Starting with `transformers >= 4.43.0` onward, you can run conversational inference using the Transformers `pipeline` abstraction or by leveraging the Auto classes with the `generate()` function. Make sure to update your transformers installation via `pip install --upgrade transformers`. ```python import torch from transformers import pipeline model_id = "meta-llama/Llama-3.2-3B-Instruct" pipe = pipeline( "text-generation", model=model_id, torch_dtype=torch.bfloat16, device_map="auto", ) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Who are you?"}, ] outputs = pipe( messages, max_new_tokens=256, ) print(outputs[0]["generated_text"][-1]) ``` Note: You can also find detailed recipes on how to use the model locally, with `torch.compile()`, assisted generations, quantised and more at [`huggingface-llama-recipes`](https://github.com/huggingface/huggingface-llama-recipes) ### Use with `llama` Please, follow the instructions in the [repository](https://github.com/meta-llama/llama) To download Original checkpoints, see the example command below leveraging `huggingface-cli`: ``` huggingface-cli download meta-llama/Llama-3.2-3B-Instruct --include "original/" --local-dir Llama-3.2-3B-Instruct ``` ## Hardware and Software Training Factors:* We used custom training libraries, Meta's custom built GPU cluster, and production infrastructure for pretraining. Fine-tuning, annotation, and evaluation were also performed on production infrastructure. Training Energy Use: Training utilized a cumulative of 916k GPU hours of computation on H100-80GB (TDP of 700W) type hardware, per the table below. Training time is the total GPU time required for training each model and power consumption is the peak power capacity per GPU device used, adjusted for power usage efficiency. ## Training Greenhouse Gas Emissions: Estimated total location-based greenhouse gas emissions were 240 tons CO2eq for training. Since 2020, Meta has maintained net zero greenhouse gas emissions in its global operations and matched 100% of its electricity use with renewable energy; therefore, the total market-based greenhouse gas emissions for training were 0 tons CO2eq. \| \| Training Time (GPU hours) \| Logit Generation Time (GPU Hours) \| Training Power Consumption (W) \| Training Location-Based Greenhouse Gas Emissions (tons CO2eq) \| Training Market-Based Greenhouse Gas Emissions (tons CO2eq) \| \| :---- \| :---: \| ----- \| :---: \| :---: \| :---: \| \| Llama 3.2 1B \| 370k \| \- \| 700 \| 107 \| 0 \| \| Llama 3.2 3B \| 460k \| \- \| 700 \| 133 \| 0 \| \| Total \| 830k \| 86k \| \| 240 \| 0 \| The methodology used to determine training energy use and greenhouse gas emissions can be found [here](https://arxiv.org/pdf/2204.05149). Since Meta is openly releasing these models, the training energy use and greenhouse gas emissions will not be incurred by others. ## Training Data Overview: Llama 3.2 was pretrained on up to 9 trillion tokens of data from publicly available sources. For the 1B and 3B Llama 3.2 models, we incorporated logits from the Llama 3.1 8B and 70B models into the pretraining stage of the model development, where outputs (logits) from these larger models were used as token-level targets. Knowledge distillation was used after pruning to recover performance. In post-training we used a similar recipe as Llama 3.1 and produced final chat models by doing several rounds of alignment on top of the pre-trained model. Each round involved Supervised Fine-Tuning (SFT), Rejection Sampling (RS), and Direct Preference Optimization (DPO). Data Freshness: The pretraining data has a cutoff of December 2023\. ## Benchmarks \- English Text In this section, we report the results for Llama 3.2 models on standard automatic benchmarks. For all these evaluations, we used our internal evaluations library. ### Base Pretrained Models \| Category \| Benchmark \| \# Shots \| Metric \| Llama 3.2 1B \| Llama 3.2 3B \| Llama 3.1 8B \| \| ----- \| ----- \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| MMLU \| 5 \| macro\_avg/acc\_char \| 32.2 \| 58 \| 66.7 \| \| \| AGIEval English \| 3-5 \| average/acc\_char \| 23.3 \| 39.2 \| 47.8 \| \| \| ARC-Challenge \| 25 \| acc\_char \| 32.8 \| 69.1 \| 79.7 \| \| Reading comprehension \| SQuAD \| 1 \| em \| 49.2 \| 67.7 \| 77 \| \| \| QuAC (F1) \| 1 \| f1 \| 37.9 \| 42.9 \| 44.9 \| \| \| DROP (F1) \| 3 \| f1 \| 28.0 \| 45.2 \| 59.5 \| \| Long Context \| Needle in Haystack \| 0 \| em \| 96.8 \| 1 \| 1 \| ### Instruction Tuned Models \| Capability \| \| Benchmark \| \# Shots \| Metric \| Llama 3.2 1B \| Llama 3.2 3B \| Llama 3.1 8B \| \| :---: \| ----- \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| \| MMLU \| 5 \| macro\_avg/acc \| 49.3 \| 63.4 \| 69.4 \| \| Re-writing \| \| Open-rewrite eval \| 0 \| micro\_avg/rougeL \| 41.6 \| 40.1 \| 40.9 \| \| Summarization \| \| TLDR9+ (test) \| 1 \| rougeL \| 16.8 \| 19.0 \| 17.2 \| \| Instruction following \| \| IFEval \| 0 \| avg(prompt/instruction acc loose/strict) \| 59.5 \| 77.4 \| 80.4 \| \| Math \| \| GSM8K (CoT) \| 8 \| em\_maj1@1 \| 44.4 \| 77.7 \| 84.5 \| \| \| \| MATH (CoT) \| 0 \| final\_em \| 30.6 \| 47.3 \| 51.9 \| \| Reasoning \| \| ARC-C \| 0 \| acc \| 59.4 \| 78.6 \| 83.4 \| \| \| \| GPQA \| 0 \| acc \| 27.2 \| 32.8 \| 32.8 \| \| \| \| Hellaswag \| 0 \| acc \| 41.2 \| 69.8 \| 78.7 \| \| Tool Use \| \| BFCL V2 \| 0 \| acc \| 25.7 \| 67.0 \| 70.9 \| \| \| \| Nexus \| 0 \| macro\_avg/acc \| 13.5 \| 34.3 \| 38.5 \| \| Long Context \| \| InfiniteBench/En.QA \| 0 \| longbook\_qa/f1 \| 20.3 \| 19.8 \| 27.3 \| \| \| \| InfiniteBench/En.MC \| 0 \| longbook\_choice/acc \| 38.0 \| 63.3 \| 72.2 \| \| \| \| NIH/Multi-needle \| 0 \| recall \| 75.0 \| 84.7 \| 98.8 \| \| Multilingual \| \| MGSM (CoT) \| 0 \| em \| 24.5 \| 58.2 \| 68.9 \| ### Multilingual Benchmarks \| Category \| Benchmark \| Language \| Llama 3.2 1B \| Llama 3.2 3B \| Llama 3.1 8B \| \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| MMLU (5-shot, macro\_avg/acc) \| Portuguese \| 39.82 \| 54.48 \| 62.12 \| \| \| \| Spanish \| 41.5 \| 55.1 \| 62.5 \| \| \| \| Italian \| 39.8 \| 53.8 \| 61.6 \| \| \| \| German \| 39.2 \| 53.3 \| 60.6 \| \| \| \| French \| 40.5 \| 54.6 \| 62.3 \| \| \| \| Hindi \| 33.5 \| 43.3 \| 50.9 \| \| \| \| Thai \| 34.7 \| 44.5 \| 50.3 \| ## Responsibility & Safety As part of our Responsible release approach, we followed a three-pronged strategy to managing trust & safety risks: 1. Enable developers to deploy helpful, safe and flexible experiences for their target audience and for the use cases supported by Llama 2. Protect developers against adversarial users aiming to exploit Llama capabilities to potentially cause harm 3. Provide protections for the community to help prevent the misuse of our models ### Responsible Deployment Approach: Llama is a foundational technology designed to be used in a variety of use cases. Examples on how Meta’s Llama models have been responsibly deployed can be found in our [Community Stories webpage](https://llama.meta.com/community-stories/). Our approach is to build the most helpful models, enabling the world to benefit from the technology power, by aligning our model safety for generic use cases and addressing a standard set of harms. Developers are then in the driver’s seat to tailor safety for their use cases, defining their own policies and deploying the models with the necessary safeguards in their Llama systems. Llama 3.2 was developed following the best practices outlined in our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide/). #### Llama 3.2 Instruct Objective: Our main objectives for conducting safety fine-tuning are to provide the research community with a valuable resource for studying the robustness of safety fine-tuning, as well as to offer developers a readily available, safe, and powerful model for various applications to reduce the developer workload to deploy safe AI systems. We implemented the same set of safety mitigations as in Llama 3, and you can learn more about these in the Llama 3 [paper](https://ai.meta.com/research/publications/the-llama-3-herd-of-models/). Fine-Tuning Data: We employ a multi-faceted approach to data collection, combining human-generated data from our vendors with synthetic data to mitigate potential safety risks. We’ve developed many large language model (LLM)-based classifiers that enable us to thoughtfully select high-quality prompts and responses, enhancing data quality control. Refusals and Tone: Building on the work we started with Llama 3, we put a great emphasis on model refusals to benign prompts as well as refusal tone. We included both borderline and adversarial prompts in our safety data strategy, and modified our safety data responses to follow tone guidelines. #### Llama 3.2 Systems Safety as a System: Large language models, including Llama 3.2, are not designed to be deployed in isolation but instead should be deployed as part of an overall AI system with additional safety guardrails as required. Developers are expected to deploy system safeguards when building agentic systems. Safeguards are key to achieve the right helpfulness-safety alignment as well as mitigating safety and security risks inherent to the system and any integration of the model or system with external tools. As part of our responsible release approach, we provide the community with [safeguards](https://llama.meta.com/trust-and-safety/) that developers should deploy with Llama models or other LLMs, including Llama Guard, Prompt Guard and Code Shield. All our [reference implementations](https://github.com/meta-llama/llama-agentic-system) demos contain these safeguards by default so developers can benefit from system-level safety out-of-the-box. ### New Capabilities and Use Cases Technological Advancement: Llama releases usually introduce new capabilities that require specific considerations in addition to the best practices that generally apply across all Generative AI use cases. For prior release capabilities also supported by Llama 3.2, see [Llama 3.1 Model Card](https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/MODEL_CARD.md), as the same considerations apply here as well. Constrained Environments: Llama 3.2 1B and 3B models are expected to be deployed in highly constrained environments, such as mobile devices. LLM Systems using smaller models will have a different alignment profile and safety/helpfulness tradeoff than more complex, larger systems. Developers should ensure the safety of their system meets the requirements of their use case. We recommend using lighter system safeguards for such use cases, like Llama Guard 3-1B or its mobile-optimized version. ### Evaluations Scaled Evaluations: We built dedicated, adversarial evaluation datasets and evaluated systems composed of Llama models and Purple Llama safeguards to filter input prompt and output response. It is important to evaluate applications in context, and we recommend building dedicated evaluation dataset for your use case. Red Teaming: We conducted recurring red teaming exercises with the goal of discovering risks via adversarial prompting and we used the learnings to improve our benchmarks and safety tuning datasets. We partnered early with subject-matter experts in critical risk areas to understand the nature of these real-world harms and how such models may lead to unintended harm for society. Based on these conversations, we derived a set of adversarial goals for the red team to attempt to achieve, such as extracting harmful information or reprogramming the model to act in a potentially harmful capacity. The red team consisted of experts in cybersecurity, adversarial machine learning, responsible AI, and integrity in addition to multilingual content specialists with background in integrity issues in specific geographic markets. ### Critical Risks In addition to our safety work above, we took extra care on measuring and/or mitigating the following critical risk areas: 1\. CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosive Weapons): Llama 3.2 1B and 3B models are smaller and less capable derivatives of Llama 3.1. For Llama 3.1 70B and 405B, to assess risks related to proliferation of chemical and biological weapons, we performed uplift testing designed to assess whether use of Llama 3.1 models could meaningfully increase the capabilities of malicious actors to plan or carry out attacks using these types of weapons and have determined that such testing also applies to the smaller 1B and 3B models. 2\. Child Safety: Child Safety risk assessments were conducted using a team of experts, to assess the model’s capability to produce outputs that could result in Child Safety risks and inform on any necessary and appropriate risk mitigations via fine tuning. We leveraged those expert red teaming sessions to expand the coverage of our evaluation benchmarks through Llama 3 model development. For Llama 3, we conducted new in-depth sessions using objective based methodologies to assess the model risks along multiple attack vectors including the additional languages Llama 3 is trained on. We also partnered with content specialists to perform red teaming exercises assessing potentially violating content while taking account of market specific nuances or experiences. 3\. Cyber Attacks: For Llama 3.1 405B, our cyber attack uplift study investigated whether LLMs can enhance human capabilities in hacking tasks, both in terms of skill level and speed. Our attack automation study focused on evaluating the capabilities of LLMs when used as autonomous agents in cyber offensive operations, specifically in the context of ransomware attacks. This evaluation was distinct from previous studies that considered LLMs as interactive assistants. The primary objective was to assess whether these models could effectively function as independent agents in executing complex cyber-attacks without human intervention. Because Llama 3.2’s 1B and 3B models are smaller and less capable models than Llama 3.1 405B, we broadly believe that the testing conducted for the 405B model also applies to Llama 3.2 models. ### Community Industry Partnerships: Generative AI safety requires expertise and tooling, and we believe in the strength of the open community to accelerate its progress. We are active members of open consortiums, including the AI Alliance, Partnership on AI and MLCommons, actively contributing to safety standardization and transparency. We encourage the community to adopt taxonomies like the MLCommons Proof of Concept evaluation to facilitate collaboration and transparency on safety and content evaluations. Our Purple Llama tools are open sourced for the community to use and widely distributed across ecosystem partners including cloud service providers. We encourage community contributions to our [Github repository](https://github.com/meta-llama/PurpleLlama). Grants: We also set up the [Llama Impact Grants](https://llama.meta.com/llama-impact-grants/) program to identify and support the most compelling applications of Meta’s Llama model for societal benefit across three categories: education, climate and open innovation. The 20 finalists from the hundreds of applications can be found [here](https://llama.meta.com/llama-impact-grants/#finalists). Reporting: Finally, we put in place a set of resources including an [output reporting mechanism](https://developers.facebook.com/llama_output_feedback) and [bug bounty program](https://www.facebook.com/whitehat) to continuously improve the Llama technology with the help of the community. ## Ethical Considerations and Limitations Values: The core values of Llama 3.2 are openness, inclusivity and helpfulness. It is meant to serve everyone, and to work for a wide range of use cases. It is thus designed to be accessible to people across many different backgrounds, experiences and perspectives. Llama 3.2 addresses users and their needs as they are, without insertion unnecessary judgment or normativity, while reflecting the understanding that even content that may appear problematic in some cases can serve valuable purposes in others. It respects the dignity and autonomy of all users, especially in terms of the values of free thought and expression that power innovation and progress. Testing: Llama 3.2 is a new technology, and like any new technology, there are risks associated with its use. Testing conducted to date has not covered, nor could it cover, all scenarios. For these reasons, as with all LLMs, Llama 3.2’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate, biased or other objectionable responses to user prompts. Therefore, before deploying any applications of Llama 3.2 models, developers should perform safety testing and tuning tailored to their specific applications of the model. Please refer to available resources including our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide), [Trust and Safety](https://llama.meta.com/trust-and-safety/) solutions, and other [resources](https://llama.meta.com/docs/get-started/) to learn more about responsible development.	null	Non_BioNLP	Quantization made by Richard Erkhov. [Github](https://github.com/RichardErkhov) [Discord](https://discord.gg/pvy7H8DZMG) [Request more models](https://github.com/RichardErkhov/quant_request) Llama-3.2-3B-Instruct - GGUF - Model creator: https://huggingface.co/alpindale/ - Original model: https://huggingface.co/alpindale/Llama-3.2-3B-Instruct/ \| Name \| Quant method \| Size \| \| ---- \| ---- \| ---- \| \| [Llama-3.2-3B-Instruct.Q2_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q2_K.gguf) \| Q2_K \| 1.27GB \| \| [Llama-3.2-3B-Instruct.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ3_XS.gguf) \| IQ3_XS \| 1.38GB \| \| [Llama-3.2-3B-Instruct.IQ3_S.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ3_S.gguf) \| IQ3_S \| 1.44GB \| \| [Llama-3.2-3B-Instruct.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q3_K_S.gguf) \| Q3_K_S \| 1.44GB \| \| [Llama-3.2-3B-Instruct.IQ3_M.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ3_M.gguf) \| IQ3_M \| 1.49GB \| \| [Llama-3.2-3B-Instruct.Q3_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q3_K.gguf) \| Q3_K \| 1.57GB \| \| [Llama-3.2-3B-Instruct.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q3_K_M.gguf) \| Q3_K_M \| 1.57GB \| \| [Llama-3.2-3B-Instruct.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q3_K_L.gguf) \| Q3_K_L \| 1.69GB \| \| [Llama-3.2-3B-Instruct.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ4_XS.gguf) \| IQ4_XS \| 1.71GB \| \| [Llama-3.2-3B-Instruct.Q4_0.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_0.gguf) \| Q4_0 \| 1.79GB \| \| [Llama-3.2-3B-Instruct.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.IQ4_NL.gguf) \| IQ4_NL \| 1.79GB \| \| [Llama-3.2-3B-Instruct.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_K_S.gguf) \| Q4_K_S \| 1.8GB \| \| [Llama-3.2-3B-Instruct.Q4_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_K.gguf) \| Q4_K \| 1.88GB \| \| [Llama-3.2-3B-Instruct.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_K_M.gguf) \| Q4_K_M \| 1.88GB \| \| [Llama-3.2-3B-Instruct.Q4_1.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q4_1.gguf) \| Q4_1 \| 1.95GB \| \| [Llama-3.2-3B-Instruct.Q5_0.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_0.gguf) \| Q5_0 \| 2.11GB \| \| [Llama-3.2-3B-Instruct.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_K_S.gguf) \| Q5_K_S \| 2.11GB \| \| [Llama-3.2-3B-Instruct.Q5_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_K.gguf) \| Q5_K \| 2.16GB \| \| [Llama-3.2-3B-Instruct.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_K_M.gguf) \| Q5_K_M \| 2.16GB \| \| [Llama-3.2-3B-Instruct.Q5_1.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q5_1.gguf) \| Q5_1 \| 2.28GB \| \| [Llama-3.2-3B-Instruct.Q6_K.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q6_K.gguf) \| Q6_K \| 2.46GB \| \| [Llama-3.2-3B-Instruct.Q8_0.gguf](https://huggingface.co/RichardErkhov/alpindale_-_Llama-3.2-3B-Instruct-gguf/blob/main/Llama-3.2-3B-Instruct.Q8_0.gguf) \| Q8_0 \| 3.19GB \| Original model description: --- language: - en - de - fr - it - pt - hi - es - th library_name: transformers pipeline_tag: text-generation tags: - facebook - meta - pytorch - llama - llama-3 license: llama3.2 extra_gated_prompt: >- ### LLAMA 3.2 COMMUNITY LICENSE AGREEMENT Llama 3.2 Version Release Date: September 25, 2024 “Agreement” means the terms and conditions for use, reproduction, distribution and modification of the Llama Materials set forth herein. “Documentation” means the specifications, manuals and documentation accompanying Llama 3.2 distributed by Meta at https://llama.meta.com/doc/overview. “Licensee” or “you” means you, or your employer or any other person or entity (if you are entering into this Agreement on such person or entity’s behalf), of the age required under applicable laws, rules or regulations to provide legal consent and that has legal authority to bind your employer or such other person or entity if you are entering in this Agreement on their behalf. “Llama 3.2” means the foundational large language models and software and algorithms, including machine-learning model code, trained model weights, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by Meta at https://www.llama.com/llama-downloads. “Llama Materials” means, collectively, Meta’s proprietary Llama 3.2 and Documentation (and any portion thereof) made available under this Agreement. “Meta” or “we” means Meta Platforms Ireland Limited (if you are located in or, if you are an entity, your principal place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if you are located outside of the EEA or Switzerland). By clicking “I Accept” below or by using or distributing any portion or element of the Llama Materials, you agree to be bound by this Agreement. 1. License Rights and Redistribution. a. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under Meta’s intellectual property or other rights owned by Meta embodied in the Llama Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Llama Materials. b. Redistribution and Use. i. If you distribute or make available the Llama Materials (or any derivative works thereof), or a product or service (including another AI model) that contains any of them, you shall (A) provide a copy of this Agreement with any such Llama Materials; and (B) prominently display “Built with Llama” on a related website, user interface, blogpost, about page, or product documentation. If you use the Llama Materials or any outputs or results of the Llama Materials to create, train, fine tune, or otherwise improve an AI model, which is distributed or made available, you shall also include “Llama” at the beginning of any such AI model name. ii. If you receive Llama Materials, or any derivative works thereof, from a Licensee as part of an integrated end user product, then Section 2 of this Agreement will not apply to you. iii. You must retain in all copies of the Llama Materials that you distribute the following attribution notice within a “Notice” text file distributed as a part of such copies: “Llama 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.” iv. Your use of the Llama Materials must comply with applicable laws and regulations (including trade compliance laws and regulations) and adhere to the Acceptable Use Policy for the Llama Materials (available at https://www.llama.com/llama3_2/use-policy), which is hereby incorporated by reference into this Agreement. 2. Additional Commercial Terms. If, on the Llama 3.2 version release date, the monthly active users of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 700 million monthly active users in the preceding calendar month, you must request a license from Meta, which Meta may grant to you in its sole discretion, and you are not authorized to exercise any of the rights under this Agreement unless or until Meta otherwise expressly grants you such rights. 3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, AND META DISCLAIMS ALL WARRANTIES OF ANY KIND, BOTH EXPRESS AND IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS. 4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING. 5. Intellectual Property. a. No trademark licenses are granted under this Agreement, and in connection with the Llama Materials, neither Meta nor Licensee may use any name or mark owned by or associated with the other or any of its affiliates, except as required for reasonable and customary use in describing and redistributing the Llama Materials or as set forth in this Section 5(a). Meta hereby grants you a license to use “Llama” (the “Mark”) solely as required to comply with the last sentence of Section 1.b.i. You will comply with Meta’s brand guidelines (currently accessible at https://about.meta.com/brand/resources/meta/company-brand/). All goodwill arising out of your use of the Mark will inure to the benefit of Meta. b. Subject to Meta’s ownership of Llama Materials and derivatives made by or for Meta, with respect to any derivative works and modifications of the Llama Materials that are made by you, as between you and Meta, you are and will be the owner of such derivative works and modifications. c. If you institute litigation or other proceedings against Meta or any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Llama Materials or Llama 3.2 outputs or results, or any portion of any of the foregoing, constitutes infringement of intellectual property or other rights owned or licensable by you, then any licenses granted to you under this Agreement shall terminate as of the date such litigation or claim is filed or instituted. You will indemnify and hold harmless Meta from and against any claim by any third party arising out of or related to your use or distribution of the Llama Materials. 6. Term and Termination. The term of this Agreement will commence upon your acceptance of this Agreement or access to the Llama Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein. Meta may terminate this Agreement if you are in breach of any term or condition of this Agreement. Upon termination of this Agreement, you shall delete and cease use of the Llama Materials. Sections 3, 4 and 7 shall survive the termination of this Agreement. 7. Governing Law and Jurisdiction. This Agreement will be governed and construed under the laws of the State of California without regard to choice of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement. The courts of California shall have exclusive jurisdiction of any dispute arising out of this Agreement. ### Llama 3.2 Acceptable Use Policy Meta is committed to promoting safe and fair use of its tools and features, including Llama 3.2. If you access or use Llama 3.2, you agree to this Acceptable Use Policy (“Policy”). The most recent copy of this policy can be found at [https://www.llama.com/llama3_2/use-policy](https://www.llama.com/llama3_2/use-policy). #### Prohibited Uses We want everyone to use Llama 3.2 safely and responsibly. You agree you will not use, or allow others to use, Llama 3.2 to: 1. Violate the law or others’ rights, including to: 1. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as: 1. Violence or terrorism 2. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material 3. Human trafficking, exploitation, and sexual violence 4. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials. 5. Sexual solicitation 6. Any other criminal activity 1. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals 2. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services 3. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices 4. Collect, process, disclose, generate, or infer private or sensitive information about individuals, including information about individuals’ identity, health, or demographic information, unless you have obtained the right to do so in accordance with applicable law 5. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama Materials 6. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system 7. Engage in any action, or facilitate any action, to intentionally circumvent or remove usage restrictions or other safety measures, or to enable functionality disabled by Meta 2. Engage in, promote, incite, facilitate, or assist in the planning or development of activities that present a risk of death or bodily harm to individuals, including use of Llama 3.2 related to the following: 8. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State or to the U.S. Biological Weapons Anti-Terrorism Act of 1989 or the Chemical Weapons Convention Implementation Act of 1997 9. Guns and illegal weapons (including weapon development) 10. Illegal drugs and regulated/controlled substances 11. Operation of critical infrastructure, transportation technologies, or heavy machinery 12. Self-harm or harm to others, including suicide, cutting, and eating disorders 13. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual 3. Intentionally deceive or mislead others, including use of Llama 3.2 related to the following: 14. Generating, promoting, or furthering fraud or the creation or promotion of disinformation 15. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content 16. Generating, promoting, or further distributing spam 17. Impersonating another individual without consent, authorization, or legal right 18. Representing that the use of Llama 3.2 or outputs are human-generated 19. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement 4. Fail to appropriately disclose to end users any known dangers of your AI system 5. Interact with third party tools, models, or software designed to generate unlawful content or engage in unlawful or harmful conduct and/or represent that the outputs of such tools, models, or software are associated with Meta or Llama 3.2 With respect to any multimodal models included in Llama 3.2, the rights granted under Section 1(a) of the Llama 3.2 Community License Agreement are not being granted to you if you are an individual domiciled in, or a company with a principal place of business in, the European Union. This restriction does not apply to end users of a product or service that incorporates any such multimodal models. Please report any violation of this Policy, software “bug,” or other problems that could lead to a violation of this Policy through one of the following means: * Reporting issues with the model: [https://github.com/meta-llama/llama-models/issues](https://l.workplace.com/l.php?u=https%3A%2F%2Fgithub.com%2Fmeta-llama%2Fllama-models%2Fissues&h=AT0qV8W9BFT6NwihiOHRuKYQM_UnkzN_NmHMy91OT55gkLpgi4kQupHUl0ssR4dQsIQ8n3tfd0vtkobvsEvt1l4Ic6GXI2EeuHV8N08OG2WnbAmm0FL4ObkazC6G_256vN0lN9DsykCvCqGZ) * Reporting risky content generated by the model: [developers.facebook.com/llama_output_feedback](http://developers.facebook.com/llama_output_feedback) * Reporting bugs and security concerns: [facebook.com/whitehat/info](http://facebook.com/whitehat/info) * Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama 3.2: [email protected] extra_gated_fields: First Name: text Last Name: text Date of birth: date_picker Country: country Affiliation: text Job title: type: select options: - Student - Research Graduate - AI researcher - AI developer/engineer - Reporter - Other geo: ip_location By clicking Submit below I accept the terms of the license and acknowledge that the information I provide will be collected stored processed and shared in accordance with the Meta Privacy Policy: checkbox extra_gated_description: >- The information you provide will be collected, stored, processed and shared in accordance with the [Meta Privacy Policy](https://www.facebook.com/privacy/policy/). extra_gated_button_content: Submit --- ## Model Information The Meta Llama 3.2 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction-tuned generative models in 1B and 3B sizes (text in/text out). The Llama 3.2 instruction-tuned text only models are optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. They outperform many of the available open source and closed chat models on common industry benchmarks. Model Developer: Meta Model Architecture: Llama 3.2 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align with human preferences for helpfulness and safety. \| \| Training Data \| Params \| Input modalities \| Output modalities \| Context Length \| GQA \| Shared Embeddings \| Token count \| Knowledge cutoff \| \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| \| Llama 3.2 (text only) \| A new mix of publicly available online data. \| 1B (1.23B) \| Multilingual Text \| Multilingual Text and code \| 128k \| Yes \| Yes \| Up to 9T tokens \| December 2023 \| \| \| \| 3B (3.21B) \| Multilingual Text \| Multilingual Text and code \| \| \| \| \| \| Supported Languages: English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai are officially supported. Llama 3.2 has been trained on a broader collection of languages than these 8 supported languages. Developers may fine-tune Llama 3.2 models for languages beyond these supported languages, provided they comply with the Llama 3.2 Community License and the Acceptable Use Policy. Developers are always expected to ensure that their deployments, including those that involve additional languages, are completed safely and responsibly. Llama 3.2 Model Family: Token counts refer to pretraining data only. All model versions use Grouped-Query Attention (GQA) for improved inference scalability. Model Release Date: Sept 25, 2024 Status: This is a static model trained on an offline dataset. Future versions may be released that improve model capabilities and safety. License: Use of Llama 3.2 is governed by the [Llama 3.2 Community License](https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/LICENSE) (a custom, commercial license agreement). Feedback: Where to send questions or comments about the model Instructions on how to provide feedback or comments on the model can be found in the model [README](https://github.com/meta-llama/llama-models/tree/main/models/llama3_2). For more technical information about generation parameters and recipes for how to use Llama 3.2 in applications, please go [here](https://github.com/meta-llama/llama-recipes). ## Intended Use Intended Use Cases: Llama 3.2 is intended for commercial and research use in multiple languages. Instruction tuned text only models are intended for assistant-like chat and agentic applications like knowledge retrieval and summarization, mobile AI powered writing assistants and query and prompt rewriting. Pretrained models can be adapted for a variety of additional natural language generation tasks. Out of Scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Acceptable Use Policy and Llama 3.2 Community License. Use in languages beyond those explicitly referenced as supported in this model card. ## How to use This repository contains two versions of Llama-3.2-3B-Instruct, for use with `transformers` and with the original `llama` codebase. ### Use with transformers Starting with `transformers >= 4.43.0` onward, you can run conversational inference using the Transformers `pipeline` abstraction or by leveraging the Auto classes with the `generate()` function. Make sure to update your transformers installation via `pip install --upgrade transformers`. ```python import torch from transformers import pipeline model_id = "meta-llama/Llama-3.2-3B-Instruct" pipe = pipeline( "text-generation", model=model_id, torch_dtype=torch.bfloat16, device_map="auto", ) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Who are you?"}, ] outputs = pipe( messages, max_new_tokens=256, ) print(outputs[0]["generated_text"][-1]) ``` Note: You can also find detailed recipes on how to use the model locally, with `torch.compile()`, assisted generations, quantised and more at [`huggingface-llama-recipes`](https://github.com/huggingface/huggingface-llama-recipes) ### Use with `llama` Please, follow the instructions in the [repository](https://github.com/meta-llama/llama) To download Original checkpoints, see the example command below leveraging `huggingface-cli`: ``` huggingface-cli download meta-llama/Llama-3.2-3B-Instruct --include "original/" --local-dir Llama-3.2-3B-Instruct ``` ## Hardware and Software Training Factors:* We used custom training libraries, Meta's custom built GPU cluster, and production infrastructure for pretraining. Fine-tuning, annotation, and evaluation were also performed on production infrastructure. Training Energy Use: Training utilized a cumulative of 916k GPU hours of computation on H100-80GB (TDP of 700W) type hardware, per the table below. Training time is the total GPU time required for training each model and power consumption is the peak power capacity per GPU device used, adjusted for power usage efficiency. ## Training Greenhouse Gas Emissions: Estimated total location-based greenhouse gas emissions were 240 tons CO2eq for training. Since 2020, Meta has maintained net zero greenhouse gas emissions in its global operations and matched 100% of its electricity use with renewable energy; therefore, the total market-based greenhouse gas emissions for training were 0 tons CO2eq. \| \| Training Time (GPU hours) \| Logit Generation Time (GPU Hours) \| Training Power Consumption (W) \| Training Location-Based Greenhouse Gas Emissions (tons CO2eq) \| Training Market-Based Greenhouse Gas Emissions (tons CO2eq) \| \| :---- \| :---: \| ----- \| :---: \| :---: \| :---: \| \| Llama 3.2 1B \| 370k \| \- \| 700 \| 107 \| 0 \| \| Llama 3.2 3B \| 460k \| \- \| 700 \| 133 \| 0 \| \| Total \| 830k \| 86k \| \| 240 \| 0 \| The methodology used to determine training energy use and greenhouse gas emissions can be found [here](https://arxiv.org/pdf/2204.05149). Since Meta is openly releasing these models, the training energy use and greenhouse gas emissions will not be incurred by others. ## Training Data Overview: Llama 3.2 was pretrained on up to 9 trillion tokens of data from publicly available sources. For the 1B and 3B Llama 3.2 models, we incorporated logits from the Llama 3.1 8B and 70B models into the pretraining stage of the model development, where outputs (logits) from these larger models were used as token-level targets. Knowledge distillation was used after pruning to recover performance. In post-training we used a similar recipe as Llama 3.1 and produced final chat models by doing several rounds of alignment on top of the pre-trained model. Each round involved Supervised Fine-Tuning (SFT), Rejection Sampling (RS), and Direct Preference Optimization (DPO). Data Freshness: The pretraining data has a cutoff of December 2023\. ## Benchmarks \- English Text In this section, we report the results for Llama 3.2 models on standard automatic benchmarks. For all these evaluations, we used our internal evaluations library. ### Base Pretrained Models \| Category \| Benchmark \| \# Shots \| Metric \| Llama 3.2 1B \| Llama 3.2 3B \| Llama 3.1 8B \| \| ----- \| ----- \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| MMLU \| 5 \| macro\_avg/acc\_char \| 32.2 \| 58 \| 66.7 \| \| \| AGIEval English \| 3-5 \| average/acc\_char \| 23.3 \| 39.2 \| 47.8 \| \| \| ARC-Challenge \| 25 \| acc\_char \| 32.8 \| 69.1 \| 79.7 \| \| Reading comprehension \| SQuAD \| 1 \| em \| 49.2 \| 67.7 \| 77 \| \| \| QuAC (F1) \| 1 \| f1 \| 37.9 \| 42.9 \| 44.9 \| \| \| DROP (F1) \| 3 \| f1 \| 28.0 \| 45.2 \| 59.5 \| \| Long Context \| Needle in Haystack \| 0 \| em \| 96.8 \| 1 \| 1 \| ### Instruction Tuned Models \| Capability \| \| Benchmark \| \# Shots \| Metric \| Llama 3.2 1B \| Llama 3.2 3B \| Llama 3.1 8B \| \| :---: \| ----- \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| \| MMLU \| 5 \| macro\_avg/acc \| 49.3 \| 63.4 \| 69.4 \| \| Re-writing \| \| Open-rewrite eval \| 0 \| micro\_avg/rougeL \| 41.6 \| 40.1 \| 40.9 \| \| Summarization \| \| TLDR9+ (test) \| 1 \| rougeL \| 16.8 \| 19.0 \| 17.2 \| \| Instruction following \| \| IFEval \| 0 \| avg(prompt/instruction acc loose/strict) \| 59.5 \| 77.4 \| 80.4 \| \| Math \| \| GSM8K (CoT) \| 8 \| em\_maj1@1 \| 44.4 \| 77.7 \| 84.5 \| \| \| \| MATH (CoT) \| 0 \| final\_em \| 30.6 \| 47.3 \| 51.9 \| \| Reasoning \| \| ARC-C \| 0 \| acc \| 59.4 \| 78.6 \| 83.4 \| \| \| \| GPQA \| 0 \| acc \| 27.2 \| 32.8 \| 32.8 \| \| \| \| Hellaswag \| 0 \| acc \| 41.2 \| 69.8 \| 78.7 \| \| Tool Use \| \| BFCL V2 \| 0 \| acc \| 25.7 \| 67.0 \| 70.9 \| \| \| \| Nexus \| 0 \| macro\_avg/acc \| 13.5 \| 34.3 \| 38.5 \| \| Long Context \| \| InfiniteBench/En.QA \| 0 \| longbook\_qa/f1 \| 20.3 \| 19.8 \| 27.3 \| \| \| \| InfiniteBench/En.MC \| 0 \| longbook\_choice/acc \| 38.0 \| 63.3 \| 72.2 \| \| \| \| NIH/Multi-needle \| 0 \| recall \| 75.0 \| 84.7 \| 98.8 \| \| Multilingual \| \| MGSM (CoT) \| 0 \| em \| 24.5 \| 58.2 \| 68.9 \| ### Multilingual Benchmarks \| Category \| Benchmark \| Language \| Llama 3.2 1B \| Llama 3.2 3B \| Llama 3.1 8B \| \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| MMLU (5-shot, macro\_avg/acc) \| Portuguese \| 39.82 \| 54.48 \| 62.12 \| \| \| \| Spanish \| 41.5 \| 55.1 \| 62.5 \| \| \| \| Italian \| 39.8 \| 53.8 \| 61.6 \| \| \| \| German \| 39.2 \| 53.3 \| 60.6 \| \| \| \| French \| 40.5 \| 54.6 \| 62.3 \| \| \| \| Hindi \| 33.5 \| 43.3 \| 50.9 \| \| \| \| Thai \| 34.7 \| 44.5 \| 50.3 \| ## Responsibility & Safety As part of our Responsible release approach, we followed a three-pronged strategy to managing trust & safety risks: 1. Enable developers to deploy helpful, safe and flexible experiences for their target audience and for the use cases supported by Llama 2. Protect developers against adversarial users aiming to exploit Llama capabilities to potentially cause harm 3. Provide protections for the community to help prevent the misuse of our models ### Responsible Deployment Approach: Llama is a foundational technology designed to be used in a variety of use cases. Examples on how Meta’s Llama models have been responsibly deployed can be found in our [Community Stories webpage](https://llama.meta.com/community-stories/). Our approach is to build the most helpful models, enabling the world to benefit from the technology power, by aligning our model safety for generic use cases and addressing a standard set of harms. Developers are then in the driver’s seat to tailor safety for their use cases, defining their own policies and deploying the models with the necessary safeguards in their Llama systems. Llama 3.2 was developed following the best practices outlined in our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide/). #### Llama 3.2 Instruct Objective: Our main objectives for conducting safety fine-tuning are to provide the research community with a valuable resource for studying the robustness of safety fine-tuning, as well as to offer developers a readily available, safe, and powerful model for various applications to reduce the developer workload to deploy safe AI systems. We implemented the same set of safety mitigations as in Llama 3, and you can learn more about these in the Llama 3 [paper](https://ai.meta.com/research/publications/the-llama-3-herd-of-models/). Fine-Tuning Data: We employ a multi-faceted approach to data collection, combining human-generated data from our vendors with synthetic data to mitigate potential safety risks. We’ve developed many large language model (LLM)-based classifiers that enable us to thoughtfully select high-quality prompts and responses, enhancing data quality control. Refusals and Tone: Building on the work we started with Llama 3, we put a great emphasis on model refusals to benign prompts as well as refusal tone. We included both borderline and adversarial prompts in our safety data strategy, and modified our safety data responses to follow tone guidelines. #### Llama 3.2 Systems Safety as a System: Large language models, including Llama 3.2, are not designed to be deployed in isolation but instead should be deployed as part of an overall AI system with additional safety guardrails as required. Developers are expected to deploy system safeguards when building agentic systems. Safeguards are key to achieve the right helpfulness-safety alignment as well as mitigating safety and security risks inherent to the system and any integration of the model or system with external tools. As part of our responsible release approach, we provide the community with [safeguards](https://llama.meta.com/trust-and-safety/) that developers should deploy with Llama models or other LLMs, including Llama Guard, Prompt Guard and Code Shield. All our [reference implementations](https://github.com/meta-llama/llama-agentic-system) demos contain these safeguards by default so developers can benefit from system-level safety out-of-the-box. ### New Capabilities and Use Cases Technological Advancement: Llama releases usually introduce new capabilities that require specific considerations in addition to the best practices that generally apply across all Generative AI use cases. For prior release capabilities also supported by Llama 3.2, see [Llama 3.1 Model Card](https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/MODEL_CARD.md), as the same considerations apply here as well. Constrained Environments: Llama 3.2 1B and 3B models are expected to be deployed in highly constrained environments, such as mobile devices. LLM Systems using smaller models will have a different alignment profile and safety/helpfulness tradeoff than more complex, larger systems. Developers should ensure the safety of their system meets the requirements of their use case. We recommend using lighter system safeguards for such use cases, like Llama Guard 3-1B or its mobile-optimized version. ### Evaluations Scaled Evaluations: We built dedicated, adversarial evaluation datasets and evaluated systems composed of Llama models and Purple Llama safeguards to filter input prompt and output response. It is important to evaluate applications in context, and we recommend building dedicated evaluation dataset for your use case. Red Teaming: We conducted recurring red teaming exercises with the goal of discovering risks via adversarial prompting and we used the learnings to improve our benchmarks and safety tuning datasets. We partnered early with subject-matter experts in critical risk areas to understand the nature of these real-world harms and how such models may lead to unintended harm for society. Based on these conversations, we derived a set of adversarial goals for the red team to attempt to achieve, such as extracting harmful information or reprogramming the model to act in a potentially harmful capacity. The red team consisted of experts in cybersecurity, adversarial machine learning, responsible AI, and integrity in addition to multilingual content specialists with background in integrity issues in specific geographic markets. ### Critical Risks In addition to our safety work above, we took extra care on measuring and/or mitigating the following critical risk areas: 1\. CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosive Weapons): Llama 3.2 1B and 3B models are smaller and less capable derivatives of Llama 3.1. For Llama 3.1 70B and 405B, to assess risks related to proliferation of chemical and biological weapons, we performed uplift testing designed to assess whether use of Llama 3.1 models could meaningfully increase the capabilities of malicious actors to plan or carry out attacks using these types of weapons and have determined that such testing also applies to the smaller 1B and 3B models. 2\. Child Safety: Child Safety risk assessments were conducted using a team of experts, to assess the model’s capability to produce outputs that could result in Child Safety risks and inform on any necessary and appropriate risk mitigations via fine tuning. We leveraged those expert red teaming sessions to expand the coverage of our evaluation benchmarks through Llama 3 model development. For Llama 3, we conducted new in-depth sessions using objective based methodologies to assess the model risks along multiple attack vectors including the additional languages Llama 3 is trained on. We also partnered with content specialists to perform red teaming exercises assessing potentially violating content while taking account of market specific nuances or experiences. 3\. Cyber Attacks: For Llama 3.1 405B, our cyber attack uplift study investigated whether LLMs can enhance human capabilities in hacking tasks, both in terms of skill level and speed. Our attack automation study focused on evaluating the capabilities of LLMs when used as autonomous agents in cyber offensive operations, specifically in the context of ransomware attacks. This evaluation was distinct from previous studies that considered LLMs as interactive assistants. The primary objective was to assess whether these models could effectively function as independent agents in executing complex cyber-attacks without human intervention. Because Llama 3.2’s 1B and 3B models are smaller and less capable models than Llama 3.1 405B, we broadly believe that the testing conducted for the 405B model also applies to Llama 3.2 models. ### Community Industry Partnerships: Generative AI safety requires expertise and tooling, and we believe in the strength of the open community to accelerate its progress. We are active members of open consortiums, including the AI Alliance, Partnership on AI and MLCommons, actively contributing to safety standardization and transparency. We encourage the community to adopt taxonomies like the MLCommons Proof of Concept evaluation to facilitate collaboration and transparency on safety and content evaluations. Our Purple Llama tools are open sourced for the community to use and widely distributed across ecosystem partners including cloud service providers. We encourage community contributions to our [Github repository](https://github.com/meta-llama/PurpleLlama). Grants: We also set up the [Llama Impact Grants](https://llama.meta.com/llama-impact-grants/) program to identify and support the most compelling applications of Meta’s Llama model for societal benefit across three categories: education, climate and open innovation. The 20 finalists from the hundreds of applications can be found [here](https://llama.meta.com/llama-impact-grants/#finalists). Reporting: Finally, we put in place a set of resources including an [output reporting mechanism](https://developers.facebook.com/llama_output_feedback) and [bug bounty program](https://www.facebook.com/whitehat) to continuously improve the Llama technology with the help of the community. ## Ethical Considerations and Limitations Values: The core values of Llama 3.2 are openness, inclusivity and helpfulness. It is meant to serve everyone, and to work for a wide range of use cases. It is thus designed to be accessible to people across many different backgrounds, experiences and perspectives. Llama 3.2 addresses users and their needs as they are, without insertion unnecessary judgment or normativity, while reflecting the understanding that even content that may appear problematic in some cases can serve valuable purposes in others. It respects the dignity and autonomy of all users, especially in terms of the values of free thought and expression that power innovation and progress. Testing: Llama 3.2 is a new technology, and like any new technology, there are risks associated with its use. Testing conducted to date has not covered, nor could it cover, all scenarios. For these reasons, as with all LLMs, Llama 3.2’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate, biased or other objectionable responses to user prompts. Therefore, before deploying any applications of Llama 3.2 models, developers should perform safety testing and tuning tailored to their specific applications of the model. Please refer to available resources including our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide), [Trust and Safety](https://llama.meta.com/trust-and-safety/) solutions, and other [resources](https://llama.meta.com/docs/get-started/) to learn more about responsible development.	{}	task	[ "SUMMARIZATION" ]	43,076
mserloth/autotrain-df1770	mserloth	text-classification	[ "transformers", "tensorboard", "safetensors", "bert", "text-classification", "autotrain", "dataset:autotrain-df1770/autotrain-data", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-05-09T20:59:32Z	2024-05-09T21:00:53+00:00	4	0	--- datasets: - autotrain-df1770/autotrain-data tags: - autotrain - text-classification widget: - text: I love AutoTrain --- # Model Trained Using AutoTrain - Problem type: Text Classification ## Validation Metrics loss: 0.5543721318244934 f1_macro: 0.7679742314070671 f1_micro: 0.7777777777777778 f1_weighted: 0.7707658289747842 precision_macro: 0.7848406546080966 precision_micro: 0.7777777777777778 precision_weighted: 0.7778696525983347 recall_macro: 0.7620894891084323 recall_micro: 0.7777777777777778 recall_weighted: 0.7777777777777778 accuracy: 0.7777777777777778	null	Non_BioNLP	# Model Trained Using AutoTrain - Problem type: Text Classification ## Validation Metrics loss: 0.5543721318244934 f1_macro: 0.7679742314070671 f1_micro: 0.7777777777777778 f1_weighted: 0.7707658289747842 precision_macro: 0.7848406546080966 precision_micro: 0.7777777777777778 precision_weighted: 0.7778696525983347 recall_macro: 0.7620894891084323 recall_micro: 0.7777777777777778 recall_weighted: 0.7777777777777778 accuracy: 0.7777777777777778	{"datasets": ["autotrain-df1770/autotrain-data"], "tags": ["autotrain", "text-classification"], "widget": [{"text": "I love AutoTrain"}]}	task	[ "TEXT_CLASSIFICATION" ]	43,077
prithivMLmods/Open-R1-Mini-Experimental-GGUF	prithivMLmods	image-text-to-text	[ "transformers", "gguf", "reasoner", "open", "r1", "explainer", "image-text-to-text", "en", "base_model:prithivMLmods/Open-R1-Mini-Experimental", "base_model:quantized:prithivMLmods/Open-R1-Mini-Experimental", "license:apache-2.0", "endpoints_compatible", "region:us", "conversational" ]	2025-02-10T16:14:48Z	2025-02-12T09:25:00+00:00	888	5	--- base_model: - prithivMLmods/Open-R1-Mini-Experimental language: - en library_name: transformers license: apache-2.0 pipeline_tag: image-text-to-text tags: - reasoner - open - r1 - explainer --- ![zfdsdfg.gif](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/WgW-xws4vzFJj48x2niWX.gif) > [!WARNING] > Note: This model contains artifacts and may perform poorly in some cases. # Open-R1-Mini-Experimental-GGUF The Open-R1-Mini-Experimental-GGUF model is a fine-tuned version of Qwen/Qwen2-VL-2B-Instruct, specifically designed for reasoning tasks, context reasoning, and multi-modal understanding based on the R1 reasoning logits data. This model integrates a conversational approach with deep reasoning capabilities to handle complex multi-modal tasks efficiently. #### Key Enhancements: * Advanced Contextual Reasoning: Open-R1-Mini-Experimental-GGUF achieves state-of-the-art performance in reasoning tasks by leveraging R1 reasoning logits data, enhancing logical inference and decision-making. * Understanding images of various resolution & ratio: The model excels at visual understanding benchmarks, including MathVista, DocVQA, RealWorldQA, MTVQA, etc. * Long-Context Video Understanding: Capable of processing and reasoning over videos of 20 minutes or more for high-quality video-based question answering, content creation, and dialogue. * Device Integration: With strong reasoning and decision-making abilities, the model can be integrated into mobile devices, robots, and automation systems for real-time operation based on both visual and textual input. * Multilingual Support: Supports text understanding in various languages within images, including English, Chinese, Japanese, Korean, Arabic, most European languages, and Vietnamese. # Sample Inference \| Example \| Image \| \|---------\|-------\| \| Example 1 \| ![lkdfgnlhbnpf.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/LujbI0bFBqrrvMSmiz4Kt.png) \| \| Example 2 \| ![open-r1.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/Ay3lb1nG7D-S56fV6qakg.png) \| \| Example 3 \| ![1.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/oOR-sIIdg1ZW6c_2MKb4M.png) \| \| Example 4 \| ![3.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/CX9B001c9IOfhfFCx2qhP.png) \| \| Example 5 \| ![4.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/LYGGRiaoOEozW0GQECTGW.png) \| Demo: https://huggingface.co/prithivMLmods/Open-R1-Mini-Experimental/blob/main/open-r1-reasoner-doc-py/open-r1-exp.ipynb ### How to Use ```python from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor from qwen_vl_utils import process_vision_info # Load the model with automatic device placement model = Qwen2VLForConditionalGeneration.from_pretrained( "prithivMLmods/Open-R1-Mini-Experimental", torch_dtype="auto", device_map="auto" ) # Recommended: Enable flash_attention_2 for better performance in multi-image and video tasks # model = Qwen2VLForConditionalGeneration.from_pretrained( # "prithivMLmods/Open-R1-Mini-Experimental", # torch_dtype=torch.bfloat16, # attn_implementation="flash_attention_2", # device_map="auto", # ) # Load processor processor = AutoProcessor.from_pretrained("prithivMLmods/Open-R1-Mini-Experimental-GGUF") # Adjust visual token range for optimized memory usage # min_pixels = 2562828 # max_pixels = 12802828 # processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels) messages = [ { "role": "user", "content": [ { "type": "image", "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg", }, {"type": "text", "text": "Analyze the context of this image."}, ], } ] # Prepare input text = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Inference generated_ids = model.generate(inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_text) ``` ### Buffer Handling ```python buffer = "" for new_text in streamer: buffer += new_text buffer = buffer.replace("<\|im_end\|>", "") yield buffer ``` ### Key Features 1. Advanced Contextual Reasoning: - Optimized for context-aware problem-solving and logical inference based on R1 reasoning logits. 2. Optical Character Recognition (OCR): - Extracts and processes text from images with exceptional accuracy. 3. Mathematical and Logical Problem Solving: - Supports complex reasoning and outputs equations in LaTeX format. 4. Conversational and Multi-Turn Interaction: - Handles multi-turn dialogue with enhanced memory retention and response coherence. 5. Multi-Modal Inputs & Outputs: - Processes images, text, and combined inputs to generate insightful analyses. 6. Secure and Efficient Model Loading: - Uses Safetensors** for faster and more secure model weight handling.	null	Non_BioNLP	![zfdsdfg.gif](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/WgW-xws4vzFJj48x2niWX.gif) > [!WARNING] > Note: This model contains artifacts and may perform poorly in some cases. # Open-R1-Mini-Experimental-GGUF The Open-R1-Mini-Experimental-GGUF model is a fine-tuned version of Qwen/Qwen2-VL-2B-Instruct, specifically designed for reasoning tasks, context reasoning, and multi-modal understanding based on the R1 reasoning logits data. This model integrates a conversational approach with deep reasoning capabilities to handle complex multi-modal tasks efficiently. #### Key Enhancements: * Advanced Contextual Reasoning: Open-R1-Mini-Experimental-GGUF achieves state-of-the-art performance in reasoning tasks by leveraging R1 reasoning logits data, enhancing logical inference and decision-making. * Understanding images of various resolution & ratio: The model excels at visual understanding benchmarks, including MathVista, DocVQA, RealWorldQA, MTVQA, etc. * Long-Context Video Understanding: Capable of processing and reasoning over videos of 20 minutes or more for high-quality video-based question answering, content creation, and dialogue. * Device Integration: With strong reasoning and decision-making abilities, the model can be integrated into mobile devices, robots, and automation systems for real-time operation based on both visual and textual input. * Multilingual Support: Supports text understanding in various languages within images, including English, Chinese, Japanese, Korean, Arabic, most European languages, and Vietnamese. # Sample Inference \| Example \| Image \| \|---------\|-------\| \| Example 1 \| ![lkdfgnlhbnpf.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/LujbI0bFBqrrvMSmiz4Kt.png) \| \| Example 2 \| ![open-r1.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/Ay3lb1nG7D-S56fV6qakg.png) \| \| Example 3 \| ![1.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/oOR-sIIdg1ZW6c_2MKb4M.png) \| \| Example 4 \| ![3.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/CX9B001c9IOfhfFCx2qhP.png) \| \| Example 5 \| ![4.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/LYGGRiaoOEozW0GQECTGW.png) \| Demo: https://huggingface.co/prithivMLmods/Open-R1-Mini-Experimental/blob/main/open-r1-reasoner-doc-py/open-r1-exp.ipynb ### How to Use ```python from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor from qwen_vl_utils import process_vision_info # Load the model with automatic device placement model = Qwen2VLForConditionalGeneration.from_pretrained( "prithivMLmods/Open-R1-Mini-Experimental", torch_dtype="auto", device_map="auto" ) # Recommended: Enable flash_attention_2 for better performance in multi-image and video tasks # model = Qwen2VLForConditionalGeneration.from_pretrained( # "prithivMLmods/Open-R1-Mini-Experimental", # torch_dtype=torch.bfloat16, # attn_implementation="flash_attention_2", # device_map="auto", # ) # Load processor processor = AutoProcessor.from_pretrained("prithivMLmods/Open-R1-Mini-Experimental-GGUF") # Adjust visual token range for optimized memory usage # min_pixels = 2562828 # max_pixels = 12802828 # processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels) messages = [ { "role": "user", "content": [ { "type": "image", "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg", }, {"type": "text", "text": "Analyze the context of this image."}, ], } ] # Prepare input text = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Inference generated_ids = model.generate(inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_text) ``` ### Buffer Handling ```python buffer = "" for new_text in streamer: buffer += new_text buffer = buffer.replace("<\|im_end\|>", "") yield buffer ``` ### Key Features 1. Advanced Contextual Reasoning: - Optimized for context-aware problem-solving and logical inference based on R1 reasoning logits. 2. Optical Character Recognition (OCR): - Extracts and processes text from images with exceptional accuracy. 3. Mathematical and Logical Problem Solving: - Supports complex reasoning and outputs equations in LaTeX format. 4. Conversational and Multi-Turn Interaction: - Handles multi-turn dialogue with enhanced memory retention and response coherence. 5. Multi-Modal Inputs & Outputs: - Processes images, text, and combined inputs to generate insightful analyses. 6. Secure and Efficient Model Loading: - Uses Safetensors** for faster and more secure model weight handling.	{"base_model": ["prithivMLmods/Open-R1-Mini-Experimental"], "language": ["en"], "library_name": "transformers", "license": "apache-2.0", "pipeline_tag": "image-text-to-text", "tags": ["reasoner", "open", "r1", "explainer"]}	task	[ "QUESTION_ANSWERING" ]	43,078
semindan/xnli_m_bert_only_tr	semindan	text-classification	[ "transformers", "pytorch", "bert", "text-classification", "generated_from_trainer", "dataset:xnli", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-11-29T17:50:33Z	2023-01-07T14:29:29+00:00	10	0	--- datasets: - xnli license: apache-2.0 metrics: - accuracy tags: - text-classification - generated_from_trainer model-index: - name: xnli_m_bert_only_tr results: - task: type: text-classification name: Text Classification dataset: name: xnli type: xnli config: tr split: train args: tr metrics: - type: accuracy value: 0.7100401606425703 name: Accuracy --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # xnli_m_bert_only_tr This model is a fine-tuned version of [bert-base-multilingual-cased](https://huggingface.co/bert-base-multilingual-cased) on the xnli dataset. It achieves the following results on the evaluation set: - Loss: 1.3355 - Accuracy: 0.7100 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 128 - eval_batch_size: 128 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 10 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:-----:\|:---------------:\|:--------:\| \| 0.75 \| 1.0 \| 3068 \| 0.7202 \| 0.6928 \| \| 0.6718 \| 2.0 \| 6136 \| 0.6718 \| 0.7209 \| \| 0.5933 \| 3.0 \| 9204 \| 0.6959 \| 0.7165 \| \| 0.5075 \| 4.0 \| 12272 \| 0.7149 \| 0.7245 \| \| 0.4237 \| 5.0 \| 15340 \| 0.8141 \| 0.7124 \| \| 0.341 \| 6.0 \| 18408 \| 0.9218 \| 0.7072 \| \| 0.2743 \| 7.0 \| 21476 \| 1.0044 \| 0.7124 \| \| 0.2135 \| 8.0 \| 24544 \| 1.1326 \| 0.7193 \| \| 0.1685 \| 9.0 \| 27612 \| 1.2362 \| 0.7056 \| \| 0.1349 \| 10.0 \| 30680 \| 1.3355 \| 0.7100 \| ### Framework versions - Transformers 4.24.0 - Pytorch 1.13.0 - Datasets 2.6.1 - Tokenizers 0.13.1	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # xnli_m_bert_only_tr This model is a fine-tuned version of [bert-base-multilingual-cased](https://huggingface.co/bert-base-multilingual-cased) on the xnli dataset. It achieves the following results on the evaluation set: - Loss: 1.3355 - Accuracy: 0.7100 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 128 - eval_batch_size: 128 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 10 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:-----:\|:---------------:\|:--------:\| \| 0.75 \| 1.0 \| 3068 \| 0.7202 \| 0.6928 \| \| 0.6718 \| 2.0 \| 6136 \| 0.6718 \| 0.7209 \| \| 0.5933 \| 3.0 \| 9204 \| 0.6959 \| 0.7165 \| \| 0.5075 \| 4.0 \| 12272 \| 0.7149 \| 0.7245 \| \| 0.4237 \| 5.0 \| 15340 \| 0.8141 \| 0.7124 \| \| 0.341 \| 6.0 \| 18408 \| 0.9218 \| 0.7072 \| \| 0.2743 \| 7.0 \| 21476 \| 1.0044 \| 0.7124 \| \| 0.2135 \| 8.0 \| 24544 \| 1.1326 \| 0.7193 \| \| 0.1685 \| 9.0 \| 27612 \| 1.2362 \| 0.7056 \| \| 0.1349 \| 10.0 \| 30680 \| 1.3355 \| 0.7100 \| ### Framework versions - Transformers 4.24.0 - Pytorch 1.13.0 - Datasets 2.6.1 - Tokenizers 0.13.1	{"datasets": ["xnli"], "license": "apache-2.0", "metrics": ["accuracy"], "tags": ["text-classification", "generated_from_trainer"], "model-index": [{"name": "xnli_m_bert_only_tr", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "xnli", "type": "xnli", "config": "tr", "split": "train", "args": "tr"}, "metrics": [{"type": "accuracy", "value": 0.7100401606425703, "name": "Accuracy"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,079
krumeto/snowflake-arctic-embed-xs-ms-marco-triplet	krumeto	sentence-similarity	[ "sentence-transformers", "safetensors", "bert", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:100000", "loss:MultipleNegativesRankingLoss", "en", "arxiv:1908.10084", "arxiv:1705.00652", "base_model:Snowflake/snowflake-arctic-embed-xs", "base_model:finetune:Snowflake/snowflake-arctic-embed-xs", "license:apache-2.0", "model-index", "autotrain_compatible", "text-embeddings-inference", "endpoints_compatible", "region:us" ]	2024-08-04T13:56:10Z	2024-08-04T13:56:15+00:00	7	0	--- base_model: Snowflake/snowflake-arctic-embed-xs datasets: [] language: - en library_name: sentence-transformers license: apache-2.0 metrics: - cosine_accuracy - dot_accuracy - manhattan_accuracy - euclidean_accuracy - max_accuracy pipeline_tag: sentence-similarity tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:100000 - loss:MultipleNegativesRankingLoss widget: - source_sentence: 'Represent this sentence for searching relevant passages: what is spider silk made of' sentences: - Arachnid Pictures. Spider's silk is made up of chains of amino acids. In other words, it is simply a protein (see How Food Works for details on amino acids and proteins). The two primary amino acids are glycine and alanine. Spider silk is extremely strong -- it is about five times stronger than steel and twice as strong as Kevlar of the same weight. - Spider silk is made of several kinds of proteins. These proteins are processed and stored within the spider's abdomen. Spiders typically have six or eight spinnerets, organs at the rear of their abdomen, through which they extrude the silk proteins. - Neon is the second lightest noble gas, after helium, and it has an atomic number of 10. On the periodic table, it is identified with the symbol Ne. The noble gases were recognized in the late 1800s, when scientists realized that an entire class of gases was missing from the periodic table of elements. - source_sentence: 'Represent this sentence for searching relevant passages: what is a caring community of learners' sentences: - 'A couple of my friends and I made hot ice for our school science fair. We used sodium acetate. This project won first place at my schools science fair!!! Materials: Stove. Pot. A spoon. A glass cup. and Sodium acetate (you can find it online or in certain heat packs). How to do it: You heat water in a pot and put the sodium acetate in the water.' - Caring Community of Learners. A group or classroom in which children and adults engage in warm, positive relationships; treat each other with respect; and learn from and with each other. Self-concept. Children's stable perceptions about themselves despite variations in their behavior. - 'Transcript of Creating a Caring Community of Caring Learners: - Caring Community of Learners: Group or classroom in which children and adults have positive, respectful relationships and learn from each other. - attachment theory: children''s ability to learn depends on trusting relationships with caregivers.' - source_sentence: 'Represent this sentence for searching relevant passages: what does dark circles around deep set eyes mean' sentences: - Production Planner, Manufacturing Salary. (United States). The average salary for a Production Planner, Manufacturing is $51,962 per year. A skill in SAP Enterprise Resource Planning (ERP) is associated with high pay for this job. People in this job generally don't have more than 20 years' experience. - Symptoms & Signs. Dark circles under the eyes are a common complaint of both men and women, although they can occasionally be seen in children. As people age, the skin becomes thinner and collagen is lost, sometimes enhancing the appearance of blood vessels beneath the eyes and making the area appear darker. - What are dark circles under the eyes? Dark circles under the eyes, sometimes called shadows or dark rings under the eyes, are the appearance of dark skin between the lower eyelid and the top of the cheek. Dark circles under the eyes can occur in infants, children, adolescents and adults, and to men and women alike. It is commonly assumed that dark circles under the eyes are caused by a lack of sleep, and poor quality sleep and insomnia can certainly cause this condition. - source_sentence: 'Represent this sentence for searching relevant passages: how big is rv access' sentences: - The average length of bigger RVs is between 7.6 meters to 12 meters or 25 feet to 40 feet. These vehicles are usually packed with different interesting features, most of which are intended to offer luxury and convenience. - Murder, My Sweet (released as Farewell, My Lovely in the United Kingdom) is a 1944 American film noir, directed by Edward Dmytryk and starring Dick Powell, Claire Trevor, and Anne Shirley. The film is based on Raymond Chandler 's 1940 novel Farewell, My Lovely. A second film adaptation of the novel was made in 1975 and released under Chandler's title. Murder, My Sweet turned out to be Anne Shirley's final film. She retired from acting in 1944 at age 26. - It should be wider then the rv.....lol sorry could not pass that one up. A standard RV is normally around 96 inches wide at most, newer larger class A's are around 102 inches wide. This width does not include mirrors or other safety equipment..... Last edited by rtandc; 10-22-2010 at 05:41 AM.. - source_sentence: 'Represent this sentence for searching relevant passages: how many pitchers used per game' sentences: - Trackback - A method by which a blogger receives notification which other bloggers link to his or her blog entry - Accroding to the statistics in the Baseball Reference page showing 2014 Major League Baseball Pitching Pitches, 745 pitchers threw 704,983 pitches in 2430 games for an average of 290 pitches per game. - In modern day baseball, teams generally have five starting pitchers, and they take it in turn to start a game every fifth day (hence the phrase rotation). Sometimes, if the schedule pans out, a team can get away with a four man rotation, and in the distant past some teams managed a three man rotation. model-index: - name: Fine-tuned snowflake actic xs based on MS-Marco triplets results: - task: type: triplet name: Triplet dataset: name: xs msmarco triplet type: xs-msmarco-triplet metrics: - type: cosine_accuracy value: 0.571 name: Cosine Accuracy - type: dot_accuracy value: 0.4286 name: Dot Accuracy - type: manhattan_accuracy value: 0.5728 name: Manhattan Accuracy - type: euclidean_accuracy value: 0.571 name: Euclidean Accuracy - type: max_accuracy value: 0.5728 name: Max Accuracy - task: type: triplet name: Triplet dataset: name: xs msmarco triplet train type: xs-msmarco-triplet-train metrics: - type: cosine_accuracy value: 0.5696 name: Cosine Accuracy - type: dot_accuracy value: 0.43 name: Dot Accuracy - type: manhattan_accuracy value: 0.5674 name: Manhattan Accuracy - type: euclidean_accuracy value: 0.5696 name: Euclidean Accuracy - type: max_accuracy value: 0.5696 name: Max Accuracy --- # Fine-tuned snowflake actic xs based on MS-Marco triplets This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Snowflake/snowflake-arctic-embed-xs](https://huggingface.co/Snowflake/snowflake-arctic-embed-xs) on the sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 dataset. It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [Snowflake/snowflake-arctic-embed-xs](https://huggingface.co/Snowflake/snowflake-arctic-embed-xs) <!-- at revision 236cea8bda4680896324c8058c67e97c135eeb95 --> - Maximum Sequence Length: 512 tokens - Output Dimensionality: 384 tokens - Similarity Function: Cosine Similarity - Training Dataset: - sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 - Language: en - License: apache-2.0 ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("krumeto/snowflake-arctic-embed-xs-ms-marco-triplet") # Run inference sentences = [ 'Represent this sentence for searching relevant passages: how many pitchers used per game', 'In modern day baseball, teams generally have five starting pitchers, and they take it in turn to start a game every fifth day (hence the phrase rotation). Sometimes, if the schedule pans out, a team can get away with a four man rotation, and in the distant past some teams managed a three man rotation.', 'Accroding to the statistics in the Baseball Reference page showing 2014 Major League Baseball Pitching Pitches, 745 pitchers threw 704,983 pitches in 2430 games for an average of 290 pitches per game.', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 384] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Triplet * Dataset: `xs-msmarco-triplet` * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) \| Metric \| Value \| \|:-------------------\|:-----------\| \| cosine_accuracy \| 0.571 \| \| dot_accuracy \| 0.4286 \| \| manhattan_accuracy \| 0.5728 \| \| euclidean_accuracy \| 0.571 \| \| max_accuracy \| 0.5728 \| #### Triplet * Dataset: `xs-msmarco-triplet-train` * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) \| Metric \| Value \| \|:-------------------\|:-----------\| \| cosine_accuracy \| 0.5696 \| \| dot_accuracy \| 0.43 \| \| manhattan_accuracy \| 0.5674 \| \| euclidean_accuracy \| 0.5696 \| \| max_accuracy \| 0.5696 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 * Dataset: sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 * Size: 100,000 training samples * Columns: <code>query</code>, <code>positive</code>, and <code>negative</code> * Approximate statistics based on the first 1000 samples: \| \| query \| positive \| negative \| \|:--------\|:-----------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\| \| type \| string \| string \| string \| \| details \| <ul><li>min: 12 tokens</li><li>mean: 17.05 tokens</li><li>max: 36 tokens</li></ul> \| <ul><li>min: 19 tokens</li><li>mean: 78.68 tokens</li><li>max: 212 tokens</li></ul> \| <ul><li>min: 16 tokens</li><li>mean: 73.73 tokens</li><li>max: 205 tokens</li></ul> \| * Samples: \| query \| positive \| negative \| \|:---------------------------------------------------------------------------------------------------------------------------------------------\|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>Represent this sentence for searching relevant passages: what are the liberal arts?</code> \| <code>liberal arts. 1. the academic course of instruction at a college intended to provide general knowledge and comprising the arts, humanities, natural sciences, and social sciences, as opposed to professional or technical subjects.</code> \| <code>Liberal Arts Defined. The liberal arts are a set of academic disciplines that include the sciences and the humanities. When you study a liberal arts curriculum, you don't have to have one specific career goal, although you might. Instead, you'll assemble a broad foundation of knowledge that can be used in a wide spectrum of careers.</code> \| \| <code>Represent this sentence for searching relevant passages: what is the mechanism of action of fibrinolytic or thrombolytic drugs?</code> \| <code>BailliÃ¨re's Clinical Haematology. 6 Mechanism of action of the thrombolytic agents. 6 Mechanism of action of the thrombolytic agents JEFFREY I. WEITZ Fibrin formed during the haemostatic, inflammatory or tissue repair process serves a temporary role, and must be degraded to restore normal tissue function and structure.</code> \| <code>Fibrinolytic drug. Fibrinolytic drug, also called thrombolytic drug, any agent that is capable of stimulating the dissolution of a blood clot (thrombus). Fibrinolytic drugs work by activating the so-called fibrinolytic pathway.</code> \| \| <code>Represent this sentence for searching relevant passages: what is normal plat count</code> \| <code>78 Followers. A. Platelets are the tiny blood cells that help stop bleeding by binding together to form a clump or plug at sites of injury inside blood vessels. A normal platelet count is between 150,000 and 450,000 platelets per microliter (one-millionth of a liter, abbreviated mcL).The average platelet count is 237,000 per mcL in men and 266,000 per mcL in women.8 Followers. A. Platelets are the tiny blood cells that help stop bleeding by binding together to form a clump or plug at sites of injury inside blood vessels. A normal platelet count is between 150,000 and 450,000 platelets per microliter (one-millionth of a liter, abbreviated mcL).</code> \| <code>In the context of blood work, PLT refers to the platelet count. Platelets are the cells that cause blood clotting and control bleeding. The normal range of platelets for adults is 3.5 to 10.5 billion cells per liter of blood, according to the Mayo Clinic. Continue Reading.</code> \| * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters: ```json { "scale": 20.0, "similarity_fct": "cos_sim" } ``` ### Evaluation Dataset #### sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 * Dataset: sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 * Size: 5,000 evaluation samples * Columns: <code>query</code>, <code>positive</code>, and <code>negative</code> * Approximate statistics based on the first 1000 samples: \| \| query \| positive \| negative \| \|:--------\|:-----------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\| \| type \| string \| string \| string \| \| details \| <ul><li>min: 12 tokens</li><li>mean: 17.09 tokens</li><li>max: 35 tokens</li></ul> \| <ul><li>min: 21 tokens</li><li>mean: 80.07 tokens</li><li>max: 250 tokens</li></ul> \| <ul><li>min: 16 tokens</li><li>mean: 76.73 tokens</li><li>max: 341 tokens</li></ul> \| * Samples: \| query \| positive \| negative \| \|:----------------------------------------------------------------------------------------------------------------------------\|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>Represent this sentence for searching relevant passages: what is the name of the material harder than diamonds</code> \| <code>Nano-material is harder than diamonds By Will Knight A material that is harder than diamond has been created in the lab, by packing together tiny ânanorodsâ of carbon. The new material, known as aggregated carbon nanorods (ACNR), was created by compressing and heating super-strong carbon molecules called buckyballs or carbon-60.</code> \| <code>What material is stronger than diamond? Diamonds are famous for their hardness, but not particularly for strength,since they are easily cleaved to create the facets that make the different cuts of diamonds for setting in a ring or necklace. Many materials are stronger than diamonds, only a couple of synthesized materials are harder than diamonds.</code> \| \| <code>Represent this sentence for searching relevant passages: is pink impression a perennial tulip?</code> \| <code>Tulip Pink Impression. close video. VIDEO. Tulip Pink Impression. The rich pink blooms of this hybrid are bound to make an impression, whether used in the landscape or as a cut flower. Robust stems and giant blooms characterise the range, and this hybrid is no exception. âPink Impressionâ will continue to impress thanks to the perennial potential of this range.</code> \| <code>Tulip Pink Impression. The lustrous petals are a deep, rich rose at the center, shading to a delicate pale pink at the edge, while doing amazing things in between that include shades of both melon and sunset. Tall, strong, long-lasting and reliable, like most Darwin hybrids. An absolutely first-class Tulip.</code> \| \| <code>Represent this sentence for searching relevant passages: define: colonization</code> \| <code>Colonization. the settlement and economic development of the uninhabited borderlands of a country (internal colonization) or the establishment of settlements (engaging primarily in agricultural activity) beyond the frontiers of a country (external colonization).</code> \| <code>Colonization is a process by which a central system of power dominates the surrounding land and its components. The term is derived from the Latin word colere, which means to inhabit. Also, colonization refers strictly to migration, for example, to settler colonies in America or Australia, trading posts, and plantations, while colonialism deals with this, along with ruling the existing indigenous peoples of styled new territories. Colonization was linked to the spread of tens of millions fro</code> \| * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters: ```json { "scale": 20.0, "similarity_fct": "cos_sim" } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 16 - `per_device_eval_batch_size`: 16 - `num_train_epochs`: 1 - `warmup_ratio`: 0.1 - `fp16`: True - `batch_sampler`: no_duplicates #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: steps - `prediction_loss_only`: True - `per_device_train_batch_size`: 16 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 1 - `eval_accumulation_steps`: None - `learning_rate`: 5e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 1 - `max_steps`: -1 - `lr_scheduler_type`: linear - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.1 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: False - `fp16`: True - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: None - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: False - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional </details> ### Training Logs \| Epoch \| Step \| Training Loss \| loss \| xs-msmarco-triplet-train_max_accuracy \| xs-msmarco-triplet_max_accuracy \| \|:-----:\|:----:\|:-------------:\|:------:\|:-------------------------------------:\|:-------------------------------:\| \| 0 \| 0 \| - \| - \| - \| 0.3648 \| \| 0.016 \| 100 \| 1.0213 \| 1.0078 \| - \| 0.3672 \| \| 0.032 \| 200 \| 0.9785 \| 0.9630 \| - \| 0.3764 \| \| 0.048 \| 300 \| 0.9591 \| 0.9190 \| - \| 0.4014 \| \| 0.064 \| 400 \| 0.9088 \| 0.8906 \| - \| 0.4392 \| \| 0.08 \| 500 \| 0.9215 \| 0.8707 \| - \| 0.4598 \| \| 0.096 \| 600 \| 0.8692 \| 0.8681 \| - \| 0.4874 \| \| 0.112 \| 700 \| 0.8806 \| 0.8538 \| - \| 0.4964 \| \| 0.128 \| 800 \| 0.8801 \| 0.8477 \| - \| 0.5106 \| \| 0.144 \| 900 \| 0.8692 \| 0.8414 \| - \| 0.5228 \| \| 0.16 \| 1000 \| 0.8624 \| 0.8391 \| - \| 0.5194 \| \| 0.176 \| 1100 \| 0.8737 \| 0.8397 \| - \| 0.5264 \| \| 0.192 \| 1200 \| 0.8505 \| 0.8344 \| - \| 0.5214 \| \| 0.208 \| 1300 \| 0.8818 \| 0.8358 \| - \| 0.5164 \| \| 0.224 \| 1400 \| 0.8464 \| 0.8269 \| - \| 0.5326 \| \| 0.24 \| 1500 \| 0.8623 \| 0.8291 \| - \| 0.5232 \| \| 0.256 \| 1600 \| 0.8203 \| 0.8407 \| - \| 0.5328 \| \| 0.272 \| 1700 \| 0.8566 \| 0.8257 \| - \| 0.5302 \| \| 0.288 \| 1800 \| 0.8386 \| 0.8198 \| - \| 0.5364 \| \| 0.304 \| 1900 \| 0.8587 \| 0.8172 \| - \| 0.5388 \| \| 0.32 \| 2000 \| 0.8472 \| 0.8233 \| - \| 0.5568 \| \| 0.336 \| 2100 \| 0.8466 \| 0.8188 \| - \| 0.5468 \| \| 0.352 \| 2200 \| 0.8273 \| 0.8190 \| - \| 0.5386 \| \| 0.368 \| 2300 \| 0.8356 \| 0.8183 \| - \| 0.5396 \| \| 0.384 \| 2400 \| 0.8294 \| 0.8156 \| - \| 0.5538 \| \| 0.4 \| 2500 \| 0.8274 \| 0.8168 \| - \| 0.5448 \| \| 0.416 \| 2600 \| 0.8392 \| 0.8093 \| - \| 0.5422 \| \| 0.432 \| 2700 \| 0.8541 \| 0.8087 \| - \| 0.5426 \| \| 0.448 \| 2800 \| 0.8218 \| 0.8086 \| - \| 0.5474 \| \| 0.464 \| 2900 \| 0.8446 \| 0.8062 \| - \| 0.554 \| \| 0.48 \| 3000 \| 0.8405 \| 0.8076 \| - \| 0.548 \| \| 0.496 \| 3100 \| 0.8447 \| 0.8087 \| - \| 0.553 \| \| 0.512 \| 3200 \| 0.8453 \| 0.8073 \| - \| 0.5536 \| \| 0.528 \| 3300 \| 0.8371 \| 0.8089 \| - \| 0.5504 \| \| 0.544 \| 3400 \| 0.8548 \| 0.8005 \| - \| 0.5516 \| \| 0.56 \| 3500 \| 0.8162 \| 0.8026 \| - \| 0.5572 \| \| 0.576 \| 3600 \| 0.8577 \| 0.7994 \| - \| 0.5558 \| \| 0.592 \| 3700 \| 0.8289 \| 0.7990 \| - \| 0.5526 \| \| 0.608 \| 3800 \| 0.824 \| 0.7967 \| - \| 0.562 \| \| 0.624 \| 3900 \| 0.833 \| 0.7959 \| - \| 0.5608 \| \| 0.64 \| 4000 \| 0.8362 \| 0.7958 \| - \| 0.5554 \| \| 0.656 \| 4100 \| 0.8057 \| 0.7966 \| - \| 0.5578 \| \| 0.672 \| 4200 \| 0.8001 \| 0.7943 \| - \| 0.5646 \| \| 0.688 \| 4300 \| 0.8215 \| 0.7937 \| - \| 0.5602 \| \| 0.704 \| 4400 \| 0.8257 \| 0.7933 \| - \| 0.5614 \| \| 0.72 \| 4500 \| 0.8173 \| 0.7942 \| - \| 0.5648 \| \| 0.736 \| 4600 \| 0.8002 \| 0.7922 \| - \| 0.5698 \| \| 0.752 \| 4700 \| 0.8445 \| 0.7899 \| - \| 0.5626 \| \| 0.768 \| 4800 \| 0.825 \| 0.7897 \| - \| 0.5592 \| \| 0.784 \| 4900 \| 0.8151 \| 0.7870 \| - \| 0.5696 \| \| 0.8 \| 5000 \| 0.8223 \| 0.7895 \| - \| 0.5676 \| \| 0.816 \| 5100 \| 0.8235 \| 0.7877 \| - \| 0.5656 \| \| 0.832 \| 5200 \| 0.8355 \| 0.7866 \| - \| 0.5688 \| \| 0.848 \| 5300 \| 0.8218 \| 0.7864 \| - \| 0.5672 \| \| 0.864 \| 5400 \| 0.8384 \| 0.7866 \| - \| 0.5652 \| \| 0.88 \| 5500 \| 0.7988 \| 0.7860 \| - \| 0.569 \| \| 0.896 \| 5600 \| 0.8117 \| 0.7867 \| - \| 0.5684 \| \| 0.912 \| 5700 \| 0.8113 \| 0.7861 \| - \| 0.5734 \| \| 0.928 \| 5800 \| 0.8129 \| 0.7860 \| - \| 0.5698 \| \| 0.944 \| 5900 \| 0.799 \| 0.7863 \| - \| 0.5688 \| \| 0.96 \| 6000 \| 0.8269 \| 0.7858 \| - \| 0.5708 \| \| 0.976 \| 6100 \| 0.8066 \| 0.7857 \| - \| 0.572 \| \| 0.992 \| 6200 \| 0.8302 \| 0.7856 \| - \| 0.5728 \| \| 1.0 \| 6250 \| - \| - \| 0.5696 \| - \| ### Framework Versions - Python: 3.10.12 - Sentence Transformers: 3.0.1 - Transformers: 4.42.4 - PyTorch: 2.3.1+cu121 - Accelerate: 0.32.1 - Datasets: 2.20.0 - Tokenizers: 0.19.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	null	Non_BioNLP	# Fine-tuned snowflake actic xs based on MS-Marco triplets This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Snowflake/snowflake-arctic-embed-xs](https://huggingface.co/Snowflake/snowflake-arctic-embed-xs) on the sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 dataset. It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [Snowflake/snowflake-arctic-embed-xs](https://huggingface.co/Snowflake/snowflake-arctic-embed-xs) <!-- at revision 236cea8bda4680896324c8058c67e97c135eeb95 --> - Maximum Sequence Length: 512 tokens - Output Dimensionality: 384 tokens - Similarity Function: Cosine Similarity - Training Dataset: - sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 - Language: en - License: apache-2.0 ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("krumeto/snowflake-arctic-embed-xs-ms-marco-triplet") # Run inference sentences = [ 'Represent this sentence for searching relevant passages: how many pitchers used per game', 'In modern day baseball, teams generally have five starting pitchers, and they take it in turn to start a game every fifth day (hence the phrase rotation). Sometimes, if the schedule pans out, a team can get away with a four man rotation, and in the distant past some teams managed a three man rotation.', 'Accroding to the statistics in the Baseball Reference page showing 2014 Major League Baseball Pitching Pitches, 745 pitchers threw 704,983 pitches in 2430 games for an average of 290 pitches per game.', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 384] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Triplet * Dataset: `xs-msmarco-triplet` * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) \| Metric \| Value \| \|:-------------------\|:-----------\| \| cosine_accuracy \| 0.571 \| \| dot_accuracy \| 0.4286 \| \| manhattan_accuracy \| 0.5728 \| \| euclidean_accuracy \| 0.571 \| \| max_accuracy \| 0.5728 \| #### Triplet * Dataset: `xs-msmarco-triplet-train` * Evaluated with [<code>TripletEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) \| Metric \| Value \| \|:-------------------\|:-----------\| \| cosine_accuracy \| 0.5696 \| \| dot_accuracy \| 0.43 \| \| manhattan_accuracy \| 0.5674 \| \| euclidean_accuracy \| 0.5696 \| \| max_accuracy \| 0.5696 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 * Dataset: sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 * Size: 100,000 training samples * Columns: <code>query</code>, <code>positive</code>, and <code>negative</code> * Approximate statistics based on the first 1000 samples: \| \| query \| positive \| negative \| \|:--------\|:-----------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\| \| type \| string \| string \| string \| \| details \| <ul><li>min: 12 tokens</li><li>mean: 17.05 tokens</li><li>max: 36 tokens</li></ul> \| <ul><li>min: 19 tokens</li><li>mean: 78.68 tokens</li><li>max: 212 tokens</li></ul> \| <ul><li>min: 16 tokens</li><li>mean: 73.73 tokens</li><li>max: 205 tokens</li></ul> \| * Samples: \| query \| positive \| negative \| \|:---------------------------------------------------------------------------------------------------------------------------------------------\|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>Represent this sentence for searching relevant passages: what are the liberal arts?</code> \| <code>liberal arts. 1. the academic course of instruction at a college intended to provide general knowledge and comprising the arts, humanities, natural sciences, and social sciences, as opposed to professional or technical subjects.</code> \| <code>Liberal Arts Defined. The liberal arts are a set of academic disciplines that include the sciences and the humanities. When you study a liberal arts curriculum, you don't have to have one specific career goal, although you might. Instead, you'll assemble a broad foundation of knowledge that can be used in a wide spectrum of careers.</code> \| \| <code>Represent this sentence for searching relevant passages: what is the mechanism of action of fibrinolytic or thrombolytic drugs?</code> \| <code>BailliÃ¨re's Clinical Haematology. 6 Mechanism of action of the thrombolytic agents. 6 Mechanism of action of the thrombolytic agents JEFFREY I. WEITZ Fibrin formed during the haemostatic, inflammatory or tissue repair process serves a temporary role, and must be degraded to restore normal tissue function and structure.</code> \| <code>Fibrinolytic drug. Fibrinolytic drug, also called thrombolytic drug, any agent that is capable of stimulating the dissolution of a blood clot (thrombus). Fibrinolytic drugs work by activating the so-called fibrinolytic pathway.</code> \| \| <code>Represent this sentence for searching relevant passages: what is normal plat count</code> \| <code>78 Followers. A. Platelets are the tiny blood cells that help stop bleeding by binding together to form a clump or plug at sites of injury inside blood vessels. A normal platelet count is between 150,000 and 450,000 platelets per microliter (one-millionth of a liter, abbreviated mcL).The average platelet count is 237,000 per mcL in men and 266,000 per mcL in women.8 Followers. A. Platelets are the tiny blood cells that help stop bleeding by binding together to form a clump or plug at sites of injury inside blood vessels. A normal platelet count is between 150,000 and 450,000 platelets per microliter (one-millionth of a liter, abbreviated mcL).</code> \| <code>In the context of blood work, PLT refers to the platelet count. Platelets are the cells that cause blood clotting and control bleeding. The normal range of platelets for adults is 3.5 to 10.5 billion cells per liter of blood, according to the Mayo Clinic. Continue Reading.</code> \| * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters: ```json { "scale": 20.0, "similarity_fct": "cos_sim" } ``` ### Evaluation Dataset #### sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 * Dataset: sentence-transformers/msmarco-msmarco-mini_lm-l-6-v3 * Size: 5,000 evaluation samples * Columns: <code>query</code>, <code>positive</code>, and <code>negative</code> * Approximate statistics based on the first 1000 samples: \| \| query \| positive \| negative \| \|:--------\|:-----------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\| \| type \| string \| string \| string \| \| details \| <ul><li>min: 12 tokens</li><li>mean: 17.09 tokens</li><li>max: 35 tokens</li></ul> \| <ul><li>min: 21 tokens</li><li>mean: 80.07 tokens</li><li>max: 250 tokens</li></ul> \| <ul><li>min: 16 tokens</li><li>mean: 76.73 tokens</li><li>max: 341 tokens</li></ul> \| * Samples: \| query \| positive \| negative \| \|:----------------------------------------------------------------------------------------------------------------------------\|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>Represent this sentence for searching relevant passages: what is the name of the material harder than diamonds</code> \| <code>Nano-material is harder than diamonds By Will Knight A material that is harder than diamond has been created in the lab, by packing together tiny ânanorodsâ of carbon. The new material, known as aggregated carbon nanorods (ACNR), was created by compressing and heating super-strong carbon molecules called buckyballs or carbon-60.</code> \| <code>What material is stronger than diamond? Diamonds are famous for their hardness, but not particularly for strength,since they are easily cleaved to create the facets that make the different cuts of diamonds for setting in a ring or necklace. Many materials are stronger than diamonds, only a couple of synthesized materials are harder than diamonds.</code> \| \| <code>Represent this sentence for searching relevant passages: is pink impression a perennial tulip?</code> \| <code>Tulip Pink Impression. close video. VIDEO. Tulip Pink Impression. The rich pink blooms of this hybrid are bound to make an impression, whether used in the landscape or as a cut flower. Robust stems and giant blooms characterise the range, and this hybrid is no exception. âPink Impressionâ will continue to impress thanks to the perennial potential of this range.</code> \| <code>Tulip Pink Impression. The lustrous petals are a deep, rich rose at the center, shading to a delicate pale pink at the edge, while doing amazing things in between that include shades of both melon and sunset. Tall, strong, long-lasting and reliable, like most Darwin hybrids. An absolutely first-class Tulip.</code> \| \| <code>Represent this sentence for searching relevant passages: define: colonization</code> \| <code>Colonization. the settlement and economic development of the uninhabited borderlands of a country (internal colonization) or the establishment of settlements (engaging primarily in agricultural activity) beyond the frontiers of a country (external colonization).</code> \| <code>Colonization is a process by which a central system of power dominates the surrounding land and its components. The term is derived from the Latin word colere, which means to inhabit. Also, colonization refers strictly to migration, for example, to settler colonies in America or Australia, trading posts, and plantations, while colonialism deals with this, along with ruling the existing indigenous peoples of styled new territories. Colonization was linked to the spread of tens of millions fro</code> \| * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters: ```json { "scale": 20.0, "similarity_fct": "cos_sim" } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 16 - `per_device_eval_batch_size`: 16 - `num_train_epochs`: 1 - `warmup_ratio`: 0.1 - `fp16`: True - `batch_sampler`: no_duplicates #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: steps - `prediction_loss_only`: True - `per_device_train_batch_size`: 16 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 1 - `eval_accumulation_steps`: None - `learning_rate`: 5e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 1 - `max_steps`: -1 - `lr_scheduler_type`: linear - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.1 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: False - `fp16`: True - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: None - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: False - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional </details> ### Training Logs \| Epoch \| Step \| Training Loss \| loss \| xs-msmarco-triplet-train_max_accuracy \| xs-msmarco-triplet_max_accuracy \| \|:-----:\|:----:\|:-------------:\|:------:\|:-------------------------------------:\|:-------------------------------:\| \| 0 \| 0 \| - \| - \| - \| 0.3648 \| \| 0.016 \| 100 \| 1.0213 \| 1.0078 \| - \| 0.3672 \| \| 0.032 \| 200 \| 0.9785 \| 0.9630 \| - \| 0.3764 \| \| 0.048 \| 300 \| 0.9591 \| 0.9190 \| - \| 0.4014 \| \| 0.064 \| 400 \| 0.9088 \| 0.8906 \| - \| 0.4392 \| \| 0.08 \| 500 \| 0.9215 \| 0.8707 \| - \| 0.4598 \| \| 0.096 \| 600 \| 0.8692 \| 0.8681 \| - \| 0.4874 \| \| 0.112 \| 700 \| 0.8806 \| 0.8538 \| - \| 0.4964 \| \| 0.128 \| 800 \| 0.8801 \| 0.8477 \| - \| 0.5106 \| \| 0.144 \| 900 \| 0.8692 \| 0.8414 \| - \| 0.5228 \| \| 0.16 \| 1000 \| 0.8624 \| 0.8391 \| - \| 0.5194 \| \| 0.176 \| 1100 \| 0.8737 \| 0.8397 \| - \| 0.5264 \| \| 0.192 \| 1200 \| 0.8505 \| 0.8344 \| - \| 0.5214 \| \| 0.208 \| 1300 \| 0.8818 \| 0.8358 \| - \| 0.5164 \| \| 0.224 \| 1400 \| 0.8464 \| 0.8269 \| - \| 0.5326 \| \| 0.24 \| 1500 \| 0.8623 \| 0.8291 \| - \| 0.5232 \| \| 0.256 \| 1600 \| 0.8203 \| 0.8407 \| - \| 0.5328 \| \| 0.272 \| 1700 \| 0.8566 \| 0.8257 \| - \| 0.5302 \| \| 0.288 \| 1800 \| 0.8386 \| 0.8198 \| - \| 0.5364 \| \| 0.304 \| 1900 \| 0.8587 \| 0.8172 \| - \| 0.5388 \| \| 0.32 \| 2000 \| 0.8472 \| 0.8233 \| - \| 0.5568 \| \| 0.336 \| 2100 \| 0.8466 \| 0.8188 \| - \| 0.5468 \| \| 0.352 \| 2200 \| 0.8273 \| 0.8190 \| - \| 0.5386 \| \| 0.368 \| 2300 \| 0.8356 \| 0.8183 \| - \| 0.5396 \| \| 0.384 \| 2400 \| 0.8294 \| 0.8156 \| - \| 0.5538 \| \| 0.4 \| 2500 \| 0.8274 \| 0.8168 \| - \| 0.5448 \| \| 0.416 \| 2600 \| 0.8392 \| 0.8093 \| - \| 0.5422 \| \| 0.432 \| 2700 \| 0.8541 \| 0.8087 \| - \| 0.5426 \| \| 0.448 \| 2800 \| 0.8218 \| 0.8086 \| - \| 0.5474 \| \| 0.464 \| 2900 \| 0.8446 \| 0.8062 \| - \| 0.554 \| \| 0.48 \| 3000 \| 0.8405 \| 0.8076 \| - \| 0.548 \| \| 0.496 \| 3100 \| 0.8447 \| 0.8087 \| - \| 0.553 \| \| 0.512 \| 3200 \| 0.8453 \| 0.8073 \| - \| 0.5536 \| \| 0.528 \| 3300 \| 0.8371 \| 0.8089 \| - \| 0.5504 \| \| 0.544 \| 3400 \| 0.8548 \| 0.8005 \| - \| 0.5516 \| \| 0.56 \| 3500 \| 0.8162 \| 0.8026 \| - \| 0.5572 \| \| 0.576 \| 3600 \| 0.8577 \| 0.7994 \| - \| 0.5558 \| \| 0.592 \| 3700 \| 0.8289 \| 0.7990 \| - \| 0.5526 \| \| 0.608 \| 3800 \| 0.824 \| 0.7967 \| - \| 0.562 \| \| 0.624 \| 3900 \| 0.833 \| 0.7959 \| - \| 0.5608 \| \| 0.64 \| 4000 \| 0.8362 \| 0.7958 \| - \| 0.5554 \| \| 0.656 \| 4100 \| 0.8057 \| 0.7966 \| - \| 0.5578 \| \| 0.672 \| 4200 \| 0.8001 \| 0.7943 \| - \| 0.5646 \| \| 0.688 \| 4300 \| 0.8215 \| 0.7937 \| - \| 0.5602 \| \| 0.704 \| 4400 \| 0.8257 \| 0.7933 \| - \| 0.5614 \| \| 0.72 \| 4500 \| 0.8173 \| 0.7942 \| - \| 0.5648 \| \| 0.736 \| 4600 \| 0.8002 \| 0.7922 \| - \| 0.5698 \| \| 0.752 \| 4700 \| 0.8445 \| 0.7899 \| - \| 0.5626 \| \| 0.768 \| 4800 \| 0.825 \| 0.7897 \| - \| 0.5592 \| \| 0.784 \| 4900 \| 0.8151 \| 0.7870 \| - \| 0.5696 \| \| 0.8 \| 5000 \| 0.8223 \| 0.7895 \| - \| 0.5676 \| \| 0.816 \| 5100 \| 0.8235 \| 0.7877 \| - \| 0.5656 \| \| 0.832 \| 5200 \| 0.8355 \| 0.7866 \| - \| 0.5688 \| \| 0.848 \| 5300 \| 0.8218 \| 0.7864 \| - \| 0.5672 \| \| 0.864 \| 5400 \| 0.8384 \| 0.7866 \| - \| 0.5652 \| \| 0.88 \| 5500 \| 0.7988 \| 0.7860 \| - \| 0.569 \| \| 0.896 \| 5600 \| 0.8117 \| 0.7867 \| - \| 0.5684 \| \| 0.912 \| 5700 \| 0.8113 \| 0.7861 \| - \| 0.5734 \| \| 0.928 \| 5800 \| 0.8129 \| 0.7860 \| - \| 0.5698 \| \| 0.944 \| 5900 \| 0.799 \| 0.7863 \| - \| 0.5688 \| \| 0.96 \| 6000 \| 0.8269 \| 0.7858 \| - \| 0.5708 \| \| 0.976 \| 6100 \| 0.8066 \| 0.7857 \| - \| 0.572 \| \| 0.992 \| 6200 \| 0.8302 \| 0.7856 \| - \| 0.5728 \| \| 1.0 \| 6250 \| - \| - \| 0.5696 \| - \| ### Framework Versions - Python: 3.10.12 - Sentence Transformers: 3.0.1 - Transformers: 4.42.4 - PyTorch: 2.3.1+cu121 - Accelerate: 0.32.1 - Datasets: 2.20.0 - Tokenizers: 0.19.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	{"base_model": "Snowflake/snowflake-arctic-embed-xs", "datasets": [], "language": ["en"], "library_name": "sentence-transformers", "license": "apache-2.0", "metrics": ["cosine_accuracy", "dot_accuracy", "manhattan_accuracy", "euclidean_accuracy", "max_accuracy"], "pipeline_tag": "sentence-similarity", "tags": ["sentence-transformers", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:100000", "loss:MultipleNegativesRankingLoss"], "widget": [{"source_sentence": "Represent this sentence for searching relevant passages: what is spider silk made of", "sentences": ["Arachnid Pictures. Spider's silk is made up of chains of amino acids. In other words, it is simply a protein (see How Food Works for details on amino acids and proteins). The two primary amino acids are glycine and alanine. Spider silk is extremely strong -- it is about five times stronger than steel and twice as strong as Kevlar of the same weight.", "Spider silk is made of several kinds of proteins. These proteins are processed and stored within the spider's abdomen. Spiders typically have six or eight spinnerets, organs at the rear of their abdomen, through which they extrude the silk proteins.", "Neon is the second lightest noble gas, after helium, and it has an atomic number of 10. On the periodic table, it is identified with the symbol Ne. The noble gases were recognized in the late 1800s, when scientists realized that an entire class of gases was missing from the periodic table of elements."]}, {"source_sentence": "Represent this sentence for searching relevant passages: what is a caring community of learners", "sentences": ["A couple of my friends and I made hot ice for our school science fair. We used sodium acetate. This project won first place at my schools science fair!!! Materials: Stove. Pot. A spoon. A glass cup. and Sodium acetate (you can find it online or in certain heat packs). How to do it: You heat water in a pot and put the sodium acetate in the water.", "Caring Community of Learners. A group or classroom in which children and adults engage in warm, positive relationships; treat each other with respect; and learn from and with each other. Self-concept. Children's stable perceptions about themselves despite variations in their behavior.", "Transcript of Creating a Caring Community of Caring Learners: - Caring Community of Learners: Group or classroom in which children and adults have positive, respectful relationships and learn from each other. - attachment theory: children's ability to learn depends on trusting relationships with caregivers."]}, {"source_sentence": "Represent this sentence for searching relevant passages: what does dark circles around deep set eyes mean", "sentences": ["Production Planner, Manufacturing Salary. (United States). The average salary for a Production Planner, Manufacturing is $51,962 per year. A skill in SAP Enterprise Resource Planning (ERP) is associated with high pay for this job. People in this job generally don't have more than 20 years' experience.", "Symptoms & Signs. Dark circles under the eyes are a common complaint of both men and women, although they can occasionally be seen in children. As people age, the skin becomes thinner and collagen is lost, sometimes enhancing the appearance of blood vessels beneath the eyes and making the area appear darker.", "What are dark circles under the eyes? Dark circles under the eyes, sometimes called shadows or dark rings under the eyes, are the appearance of dark skin between the lower eyelid and the top of the cheek. Dark circles under the eyes can occur in infants, children, adolescents and adults, and to men and women alike. It is commonly assumed that dark circles under the eyes are caused by a lack of sleep, and poor quality sleep and insomnia can certainly cause this condition."]}, {"source_sentence": "Represent this sentence for searching relevant passages: how big is rv access", "sentences": ["The average length of bigger RVs is between 7.6 meters to 12 meters or 25 feet to 40 feet. These vehicles are usually packed with different interesting features, most of which are intended to offer luxury and convenience.", "Murder, My Sweet (released as Farewell, My Lovely in the United Kingdom) is a 1944 American film noir, directed by Edward Dmytryk and starring Dick Powell, Claire Trevor, and Anne Shirley. The film is based on Raymond Chandler 's 1940 novel Farewell, My Lovely. A second film adaptation of the novel was made in 1975 and released under Chandler's title. Murder, My Sweet turned out to be Anne Shirley's final film. She retired from acting in 1944 at age 26.", "It should be wider then the rv.....lol sorry could not pass that one up. A standard RV is normally around 96 inches wide at most, newer larger class A's are around 102 inches wide. This width does not include mirrors or other safety equipment..... Last edited by rtandc; 10-22-2010 at 05:41 AM.."]}, {"source_sentence": "Represent this sentence for searching relevant passages: how many pitchers used per game", "sentences": ["Trackback - A method by which a blogger receives notification which other bloggers link to his or her blog entry", "Accroding to the statistics in the Baseball Reference page showing 2014 Major League Baseball Pitching Pitches, 745 pitchers threw 704,983 pitches in 2430 games for an average of 290 pitches per game.", "In modern day baseball, teams generally have five starting pitchers, and they take it in turn to start a game every fifth day (hence the phrase rotation). Sometimes, if the schedule pans out, a team can get away with a four man rotation, and in the distant past some teams managed a three man rotation."]}], "model-index": [{"name": "Fine-tuned snowflake actic xs based on MS-Marco triplets", "results": [{"task": {"type": "triplet", "name": "Triplet"}, "dataset": {"name": "xs msmarco triplet", "type": "xs-msmarco-triplet"}, "metrics": [{"type": "cosine_accuracy", "value": 0.571, "name": "Cosine Accuracy"}, {"type": "dot_accuracy", "value": 0.4286, "name": "Dot Accuracy"}, {"type": "manhattan_accuracy", "value": 0.5728, "name": "Manhattan Accuracy"}, {"type": "euclidean_accuracy", "value": 0.571, "name": "Euclidean Accuracy"}, {"type": "max_accuracy", "value": 0.5728, "name": "Max Accuracy"}]}, {"task": {"type": "triplet", "name": "Triplet"}, "dataset": {"name": "xs msmarco triplet train", "type": "xs-msmarco-triplet-train"}, "metrics": [{"type": "cosine_accuracy", "value": 0.5696, "name": "Cosine Accuracy"}, {"type": "dot_accuracy", "value": 0.43, "name": "Dot Accuracy"}, {"type": "manhattan_accuracy", "value": 0.5674, "name": "Manhattan Accuracy"}, {"type": "euclidean_accuracy", "value": 0.5696, "name": "Euclidean Accuracy"}, {"type": "max_accuracy", "value": 0.5696, "name": "Max Accuracy"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,080
fayyadd/distilbert-base-uncased-finetuned-cola	fayyadd	text-classification	[ "transformers", "pytorch", "tensorboard", "distilbert", "text-classification", "generated_from_trainer", "dataset:glue", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-06-05T21:48:20Z	2023-06-06T00:22:22+00:00	23	0	--- datasets: - glue license: apache-2.0 metrics: - matthews_correlation tags: - generated_from_trainer model-index: - name: distilbert-base-uncased-finetuned-cola results: - task: type: text-classification name: Text Classification dataset: name: glue type: glue config: cola split: validation args: cola metrics: - type: matthews_correlation value: 0.5517964161621091 name: Matthews Correlation --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert-base-uncased-finetuned-cola This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the glue dataset. It achieves the following results on the evaluation set: - Loss: 0.7387 - Matthews Correlation: 0.5518 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 5 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Matthews Correlation \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------------------:\| \| 0.5216 \| 1.0 \| 535 \| 0.5281 \| 0.4364 \| \| 0.3471 \| 2.0 \| 1070 \| 0.5044 \| 0.4969 \| \| 0.2372 \| 3.0 \| 1605 \| 0.5483 \| 0.5481 \| \| 0.1777 \| 4.0 \| 2140 \| 0.7387 \| 0.5518 \| \| 0.1253 \| 5.0 \| 2675 \| 0.8130 \| 0.5367 \| ### Framework versions - Transformers 4.28.0 - Pytorch 2.0.1+cu118 - Datasets 2.12.0 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert-base-uncased-finetuned-cola This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the glue dataset. It achieves the following results on the evaluation set: - Loss: 0.7387 - Matthews Correlation: 0.5518 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 5 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Matthews Correlation \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------------------:\| \| 0.5216 \| 1.0 \| 535 \| 0.5281 \| 0.4364 \| \| 0.3471 \| 2.0 \| 1070 \| 0.5044 \| 0.4969 \| \| 0.2372 \| 3.0 \| 1605 \| 0.5483 \| 0.5481 \| \| 0.1777 \| 4.0 \| 2140 \| 0.7387 \| 0.5518 \| \| 0.1253 \| 5.0 \| 2675 \| 0.8130 \| 0.5367 \| ### Framework versions - Transformers 4.28.0 - Pytorch 2.0.1+cu118 - Datasets 2.12.0 - Tokenizers 0.13.3	{"datasets": ["glue"], "license": "apache-2.0", "metrics": ["matthews_correlation"], "tags": ["generated_from_trainer"], "model-index": [{"name": "distilbert-base-uncased-finetuned-cola", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "glue", "type": "glue", "config": "cola", "split": "validation", "args": "cola"}, "metrics": [{"type": "matthews_correlation", "value": 0.5517964161621091, "name": "Matthews Correlation"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,081
ailm/pegsus-text-summarization	ailm	null	[ "pytorch", "pegasus", "en", "base_model:google/pegasus-cnn_dailymail", "base_model:finetune:google/pegasus-cnn_dailymail", "license:apache-2.0", "region:us" ]	2024-09-03T15:36:37Z	2024-09-03T16:46:37+00:00	14	0	--- base_model: google/pegasus-cnn_dailymail language: - en license: apache-2.0 metrics: - rouge --- ### Pegasus-based Text Summarization Model Model Name: pegsus-text-summarization ### Model Description This model is a fine-tuned version of the Pegasus model, specifically adapted for the task of text summarization. It is trained on the SAMSum dataset, which is designed for summarizing conversations. ### Usage This model can be used to generate concise summaries of input text, particularly for conversational text or dialogue-based inputs. ### How to Use You can use this model with the Hugging Face transformers library. Below is an example code snippet: ```bash from transformers import PegasusForConditionalGeneration, PegasusTokenizer # Load the pre-trained model and tokenizer model_name = "ailm/pegsus-text-summarization" model = PegasusForConditionalGeneration.from_pretrained(model_name) tokenizer = PegasusTokenizer.from_pretrained(model_name) # Define the input text text = "Your input text here" # Tokenize the input text tokens = tokenizer(text, truncation=True, padding="longest", return_tensors="pt") # Generate the summary summary = model.generate(**tokens) # Decode and print the summary print(tokenizer.decode(summary[0], skip_special_tokens=True))	null	Non_BioNLP	### Pegasus-based Text Summarization Model Model Name: pegsus-text-summarization ### Model Description This model is a fine-tuned version of the Pegasus model, specifically adapted for the task of text summarization. It is trained on the SAMSum dataset, which is designed for summarizing conversations. ### Usage This model can be used to generate concise summaries of input text, particularly for conversational text or dialogue-based inputs. ### How to Use You can use this model with the Hugging Face transformers library. Below is an example code snippet: ```bash from transformers import PegasusForConditionalGeneration, PegasusTokenizer # Load the pre-trained model and tokenizer model_name = "ailm/pegsus-text-summarization" model = PegasusForConditionalGeneration.from_pretrained(model_name) tokenizer = PegasusTokenizer.from_pretrained(model_name) # Define the input text text = "Your input text here" # Tokenize the input text tokens = tokenizer(text, truncation=True, padding="longest", return_tensors="pt") # Generate the summary summary = model.generate(**tokens) # Decode and print the summary print(tokenizer.decode(summary[0], skip_special_tokens=True))	{"base_model": "google/pegasus-cnn_dailymail", "language": ["en"], "license": "apache-2.0", "metrics": ["rouge"]}	task	[ "SUMMARIZATION" ]	43,082
rinkorn/marian-finetuned-kde4-en-to-ru	rinkorn	translation	[ "transformers", "safetensors", "marian", "text2text-generation", "translation", "generated_from_trainer", "base_model:Helsinki-NLP/opus-mt-en-ru", "base_model:finetune:Helsinki-NLP/opus-mt-en-ru", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-01-10T15:39:58Z	2024-01-11T00:43:36+00:00	12	0	--- base_model: Helsinki-NLP/opus-mt-en-ru license: apache-2.0 metrics: - bleu tags: - translation - generated_from_trainer model-index: - name: marian-finetuned-kde4-en-to-ru results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # marian-finetuned-kde4-en-to-ru This model is a fine-tuned version of [Helsinki-NLP/opus-mt-en-ru](https://huggingface.co/Helsinki-NLP/opus-mt-en-ru) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 1.3947 - Bleu: 34.8576 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 32 - eval_batch_size: 64 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results ### Framework versions - Transformers 4.36.2 - Pytorch 2.1.1+cu121 - Datasets 2.16.0 - Tokenizers 0.15.0	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # marian-finetuned-kde4-en-to-ru This model is a fine-tuned version of [Helsinki-NLP/opus-mt-en-ru](https://huggingface.co/Helsinki-NLP/opus-mt-en-ru) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 1.3947 - Bleu: 34.8576 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 32 - eval_batch_size: 64 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results ### Framework versions - Transformers 4.36.2 - Pytorch 2.1.1+cu121 - Datasets 2.16.0 - Tokenizers 0.15.0	{"base_model": "Helsinki-NLP/opus-mt-en-ru", "license": "apache-2.0", "metrics": ["bleu"], "tags": ["translation", "generated_from_trainer"], "model-index": [{"name": "marian-finetuned-kde4-en-to-ru", "results": []}]}	task	[ "TRANSLATION" ]	43,083
RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf	RichardErkhov	null	[ "gguf", "endpoints_compatible", "region:us", "conversational" ]	2024-07-17T04:31:46Z	2024-07-17T04:56:37+00:00	80	0	--- {} --- Quantization made by Richard Erkhov. [Github](https://github.com/RichardErkhov) [Discord](https://discord.gg/pvy7H8DZMG) [Request more models](https://github.com/RichardErkhov/quant_request) gemma-2b-translation-v0.150 - GGUF - Model creator: https://huggingface.co/lemon-mint/ - Original model: https://huggingface.co/lemon-mint/gemma-2b-translation-v0.150/ \| Name \| Quant method \| Size \| \| ---- \| ---- \| ---- \| \| [gemma-2b-translation-v0.150.Q2_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q2_K.gguf) \| Q2_K \| 1.08GB \| \| [gemma-2b-translation-v0.150.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ3_XS.gguf) \| IQ3_XS \| 1.16GB \| \| [gemma-2b-translation-v0.150.IQ3_S.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ3_S.gguf) \| IQ3_S \| 1.2GB \| \| [gemma-2b-translation-v0.150.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q3_K_S.gguf) \| Q3_K_S \| 1.2GB \| \| [gemma-2b-translation-v0.150.IQ3_M.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ3_M.gguf) \| IQ3_M \| 1.22GB \| \| [gemma-2b-translation-v0.150.Q3_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q3_K.gguf) \| Q3_K \| 1.29GB \| \| [gemma-2b-translation-v0.150.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q3_K_M.gguf) \| Q3_K_M \| 1.29GB \| \| [gemma-2b-translation-v0.150.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q3_K_L.gguf) \| Q3_K_L \| 1.36GB \| \| [gemma-2b-translation-v0.150.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ4_XS.gguf) \| IQ4_XS \| 1.4GB \| \| [gemma-2b-translation-v0.150.Q4_0.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_0.gguf) \| Q4_0 \| 1.44GB \| \| [gemma-2b-translation-v0.150.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ4_NL.gguf) \| IQ4_NL \| 1.45GB \| \| [gemma-2b-translation-v0.150.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_K_S.gguf) \| Q4_K_S \| 1.45GB \| \| [gemma-2b-translation-v0.150.Q4_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_K.gguf) \| Q4_K \| 1.52GB \| \| [gemma-2b-translation-v0.150.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_K_M.gguf) \| Q4_K_M \| 1.52GB \| \| [gemma-2b-translation-v0.150.Q4_1.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_1.gguf) \| Q4_1 \| 1.56GB \| \| [gemma-2b-translation-v0.150.Q5_0.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_0.gguf) \| Q5_0 \| 1.68GB \| \| [gemma-2b-translation-v0.150.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_K_S.gguf) \| Q5_K_S \| 1.68GB \| \| [gemma-2b-translation-v0.150.Q5_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_K.gguf) \| Q5_K \| 1.71GB \| \| [gemma-2b-translation-v0.150.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_K_M.gguf) \| Q5_K_M \| 1.71GB \| \| [gemma-2b-translation-v0.150.Q5_1.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_1.gguf) \| Q5_1 \| 1.79GB \| \| [gemma-2b-translation-v0.150.Q6_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q6_K.gguf) \| Q6_K \| 1.92GB \| \| [gemma-2b-translation-v0.150.Q8_0.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q8_0.gguf) \| Q8_0 \| 2.49GB \| Original model description: --- library_name: transformers language: - ko license: gemma tags: - gemma - pytorch - instruct - finetune - translation widget: - messages: - role: user content: "Translate into Korean:Hamsters don't eat cats." base_model: lemon-mint/gemma-ko-1.1-2b-it pipeline_tag: text-generation --- # Gemma 2B Translation v0.150 - Eval Loss: `0.93082` - Train Loss: `0.81643` - lr: `9e-05` - optimizer: adamw - lr_scheduler_type: cosine ## Prompt Template ``` <bos><start_of_turn>user Translate into Korean:Hamsters don't eat cats.<end_of_turn> <start_of_turn>model 햄스터는 고양이를 먹지 않습니다.<eos> ``` ``` <bos><start_of_turn>user Translate into English:햄스터는 고양이를 먹지 않습니다.<end_of_turn> <start_of_turn>model Hamsters do not eat cats.<eos> ``` ## Model Description - Developed by: `lemon-mint` - Model type: Gemma - Language(s) (NLP): English - License: [gemma-terms-of-use](https://ai.google.dev/gemma/terms) - Finetuned from model: [lemon-mint/gemma-ko-1.1-2b-it](https://huggingface.co/lemon-mint/gemma-ko-1.1-2b-it)	null	Non_BioNLP	Quantization made by Richard Erkhov. [Github](https://github.com/RichardErkhov) [Discord](https://discord.gg/pvy7H8DZMG) [Request more models](https://github.com/RichardErkhov/quant_request) gemma-2b-translation-v0.150 - GGUF - Model creator: https://huggingface.co/lemon-mint/ - Original model: https://huggingface.co/lemon-mint/gemma-2b-translation-v0.150/ \| Name \| Quant method \| Size \| \| ---- \| ---- \| ---- \| \| [gemma-2b-translation-v0.150.Q2_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q2_K.gguf) \| Q2_K \| 1.08GB \| \| [gemma-2b-translation-v0.150.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ3_XS.gguf) \| IQ3_XS \| 1.16GB \| \| [gemma-2b-translation-v0.150.IQ3_S.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ3_S.gguf) \| IQ3_S \| 1.2GB \| \| [gemma-2b-translation-v0.150.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q3_K_S.gguf) \| Q3_K_S \| 1.2GB \| \| [gemma-2b-translation-v0.150.IQ3_M.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ3_M.gguf) \| IQ3_M \| 1.22GB \| \| [gemma-2b-translation-v0.150.Q3_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q3_K.gguf) \| Q3_K \| 1.29GB \| \| [gemma-2b-translation-v0.150.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q3_K_M.gguf) \| Q3_K_M \| 1.29GB \| \| [gemma-2b-translation-v0.150.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q3_K_L.gguf) \| Q3_K_L \| 1.36GB \| \| [gemma-2b-translation-v0.150.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ4_XS.gguf) \| IQ4_XS \| 1.4GB \| \| [gemma-2b-translation-v0.150.Q4_0.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_0.gguf) \| Q4_0 \| 1.44GB \| \| [gemma-2b-translation-v0.150.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.IQ4_NL.gguf) \| IQ4_NL \| 1.45GB \| \| [gemma-2b-translation-v0.150.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_K_S.gguf) \| Q4_K_S \| 1.45GB \| \| [gemma-2b-translation-v0.150.Q4_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_K.gguf) \| Q4_K \| 1.52GB \| \| [gemma-2b-translation-v0.150.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_K_M.gguf) \| Q4_K_M \| 1.52GB \| \| [gemma-2b-translation-v0.150.Q4_1.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q4_1.gguf) \| Q4_1 \| 1.56GB \| \| [gemma-2b-translation-v0.150.Q5_0.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_0.gguf) \| Q5_0 \| 1.68GB \| \| [gemma-2b-translation-v0.150.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_K_S.gguf) \| Q5_K_S \| 1.68GB \| \| [gemma-2b-translation-v0.150.Q5_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_K.gguf) \| Q5_K \| 1.71GB \| \| [gemma-2b-translation-v0.150.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_K_M.gguf) \| Q5_K_M \| 1.71GB \| \| [gemma-2b-translation-v0.150.Q5_1.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q5_1.gguf) \| Q5_1 \| 1.79GB \| \| [gemma-2b-translation-v0.150.Q6_K.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q6_K.gguf) \| Q6_K \| 1.92GB \| \| [gemma-2b-translation-v0.150.Q8_0.gguf](https://huggingface.co/RichardErkhov/lemon-mint_-_gemma-2b-translation-v0.150-gguf/blob/main/gemma-2b-translation-v0.150.Q8_0.gguf) \| Q8_0 \| 2.49GB \| Original model description: --- library_name: transformers language: - ko license: gemma tags: - gemma - pytorch - instruct - finetune - translation widget: - messages: - role: user content: "Translate into Korean:Hamsters don't eat cats." base_model: lemon-mint/gemma-ko-1.1-2b-it pipeline_tag: text-generation --- # Gemma 2B Translation v0.150 - Eval Loss: `0.93082` - Train Loss: `0.81643` - lr: `9e-05` - optimizer: adamw - lr_scheduler_type: cosine ## Prompt Template ``` <bos><start_of_turn>user Translate into Korean:Hamsters don't eat cats.<end_of_turn> <start_of_turn>model 햄스터는 고양이를 먹지 않습니다.<eos> ``` ``` <bos><start_of_turn>user Translate into English:햄스터는 고양이를 먹지 않습니다.<end_of_turn> <start_of_turn>model Hamsters do not eat cats.<eos> ``` ## Model Description - Developed by: `lemon-mint` - Model type: Gemma - Language(s) (NLP): English - License: [gemma-terms-of-use](https://ai.google.dev/gemma/terms) - Finetuned from model: [lemon-mint/gemma-ko-1.1-2b-it](https://huggingface.co/lemon-mint/gemma-ko-1.1-2b-it)	{}	task	[ "TRANSLATION" ]	43,084
c01zaut/Marco-o1-rk3588-1.1.2	c01zaut	text-generation	[ "transformers", "safetensors", "qwen2", "text-generation", "conversational", "arxiv:2411.14405", "license:apache-2.0", "autotrain_compatible", "text-generation-inference", "region:us" ]	2024-11-30T13:19:20Z	2024-11-30T14:24:08+00:00	6	0	--- library_name: transformers license: apache-2.0 inference: false --- # Marco-o1-RK3588-1.1.2 This version of Marco-o1 has been converted to run on the RK3588 NPU using ['w8a8', 'w8a8_g128', 'w8a8_g256', 'w8a8_g512'] quantization. This model has been optimized with the following LoRA: Compatible with RKLLM version: 1.1.2 ## Useful links: [Official RKLLM GitHub](https://github.com/airockchip/rknn-llm) [RockhipNPU Reddit](https://reddit.com/r/RockchipNPU) [EZRKNN-LLM](https://github.com/Pelochus/ezrknn-llm/) Pretty much anything by these folks: [marty1885](https://github.com/marty1885) and [happyme531](https://huggingface.co/happyme531) Converted using https://github.com/c0zaut/ez-er-rkllm-toolkit # Original Model Card for base model, Marco-o1, below: <p align="center"> <img src="assets/logo.png" width="150" style="margin-bottom: 0.2;"/> <p> # 🍓 Marco-o1: Towards Open Reasoning Models for Open-Ended Solutions <!-- Broader Real-World Applications --> <!-- # 🍓 Marco-o1: An Open Large Reasoning Model for Real-World Solutions --> <!-- <h2 align="center"> <a href="https://github.com/AIDC-AI/Marco-o1/">Marco-o1</a></h2> --> <!-- <h5 align="center"> If you appreciate our project, please consider giving us a star ⭐ on GitHub to stay updated with the latest developments. </h2> --> <div align="center"> <!-- Affiliations: --> ⭐ _MarcoPolo Team_ ⭐ [_AI Business, Alibaba International Digital Commerce_](https://aidc-ai.com) [Github](https://github.com/AIDC-AI/Marco-o1) 🤗 [Hugging Face](https://huggingface.co/AIDC-AI/Marco-o1) 📝 [Paper](https://arxiv.org/abs/2411.14405) 🧑‍💻 [Model](https://huggingface.co/AIDC-AI/Marco-o1) 🗂️ [Data](https://github.com/AIDC-AI/Marco-o1/tree/main/data) 📽️ [Demo](https://huggingface.co/AIDC-AI/Marco-o1) </div> 🎯 Marco-o1 not only focuses on disciplines with standard answers, such as mathematics, physics, and coding—which are well-suited for reinforcement learning (RL)—but also places greater emphasis on open-ended resolutions. We aim to address the question: _"Can the o1 model effectively generalize to broader domains where clear standards are absent and rewards are challenging to quantify?"_ Currently, Marco-o1 Large Language Model (LLM) is powered by _Chain-of-Thought (CoT) fine-tuning_, _Monte Carlo Tree Search (MCTS)_, _reflection mechanisms_, and _innovative reasoning strategies_—optimized for complex real-world problem-solving tasks. ⚠️ Limitations: <ins>We would like to emphasize that this research work is inspired by OpenAI's o1 (from which the name is also derived). This work aims to explore potential approaches to shed light on the currently unclear technical roadmap for large reasoning models. Besides, our focus is on open-ended questions, and we have observed interesting phenomena in multilingual applications. However, we must acknowledge that the current model primarily exhibits o1-like reasoning characteristics and its performance still fall short of a fully realized "o1" model. This is not a one-time effort, and we remain committed to continuous optimization and ongoing improvement.</ins> ![img.png](assets/img.png) ## 🚀 Highlights Currently, our work is distinguished by the following highlights: - 🍀 Fine-Tuning with CoT Data: We develop Marco-o1-CoT by performing full-parameter fine-tuning on the base model using open-source CoT dataset combined with our self-developed synthetic data. - 🍀 Solution Space Expansion via MCTS: We integrate LLMs with MCTS (Marco-o1-MCTS), using the model's output confidence to guide the search and expand the solution space. - 🍀 Reasoning Action Strategy: We implement novel reasoning action strategies and a reflection mechanism (Marco-o1-MCTS Mini-Step), including exploring different action granularities within the MCTS framework and prompting the model to self-reflect, thereby significantly enhancing the model's ability to solve complex problems. - 🍀 Application in Translation Tasks: We are the first to apply Large Reasoning Models (LRM) to Machine Translation task, exploring inference time scaling laws in the multilingual and translation domain. OpenAI recently introduced the groundbreaking o1 model, renowned for its exceptional reasoning capabilities. This model has demonstrated outstanding performance on platforms such as AIME, CodeForces, surpassing other leading models. Inspired by this success, we aimed to push the boundaries of LLMs even further, enhancing their reasoning abilities to tackle complex, real-world challenges. 🌍 Marco-o1 leverages advanced techniques like CoT fine-tuning, MCTS, and Reasoning Action Strategies to enhance its reasoning power. As shown in Figure 2, by fine-tuning Qwen2-7B-Instruct with a combination of the filtered Open-O1 CoT dataset, Marco-o1 CoT dataset, and Marco-o1 Instruction dataset, Marco-o1 improved its handling of complex tasks. MCTS allows exploration of multiple reasoning paths using confidence scores derived from softmax-applied log probabilities of the top-k alternative tokens, guiding the model to optimal solutions. Moreover, our reasoning action strategy involves varying the granularity of actions within steps and mini-steps to optimize search efficiency and accuracy. <div align="center"> <img src="assets/intro_2.jpg" alt="Figure Description or Alt Text" width="80%"> <p><strong>Figure 2: </strong>The overview of Marco-o1.</p> </div> 🌏 As shown in Figure 3, Marco-o1 achieved accuracy improvements of +6.17% on the MGSM (English) dataset and +5.60% on the MGSM (Chinese) dataset, showcasing enhanced reasoning capabilities. <div align="center"> <img src="assets/results.jpg" alt="Figure Description or Alt Text" width="80%"> <p><strong>Figure 3: </strong>The main results of Marco-o1.</p> </div> 🌎 Additionally, in translation tasks, we demonstrate that Marco-o1 excels in translating slang expressions, such as translating "这个鞋拥有踩屎感" (literal translation: "This shoe offers a stepping-on-poop sensation.") to "This shoe has a comfortable sole," demonstrating its superior grasp of colloquial nuances. <div align="center"> <img src="assets/translation.jpg" alt="Figure Description or Alt Text" width="80%"> <p><strong>Figure 4: </strong>The demostration of translation task using Marco-o1.</p> </div> For more information,please visit our [Github](https://github.com/AIDC-AI/Marco-o1). ## Usage 1. Load Marco-o1-CoT model: ``` # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("AIDC-AI/Marco-o1") model = AutoModelForCausalLM.from_pretrained("AIDC-AI/Marco-o1") ``` 2. Inference: Execute the inference script (you can give any customized inputs inside): ``` ./src/talk_with_model.py # Use vLLM ./src/talk_with_model_vllm.py ``` # 👨🏻‍💻 Acknowledgement ## Main Contributors From MarcoPolo Team, AI Business, Alibaba International Digital Commerce: - Yu Zhao - [Huifeng Yin](https://github.com/HuifengYin) - Hao Wang - [Longyue Wang](http://www.longyuewang.com) ## Citation If you find Marco-o1 useful for your research and applications, please cite: ``` @misc{zhao2024marcoo1openreasoningmodels, title={Marco-o1: Towards Open Reasoning Models for Open-Ended Solutions}, author={Yu Zhao and Huifeng Yin and Bo Zeng and Hao Wang and Tianqi Shi and Chenyang Lyu and Longyue Wang and Weihua Luo and Kaifu Zhang}, year={2024}, eprint={2411.14405}, archivePrefix={arXiv}, primaryClass={cs.CL}, url={https://arxiv.org/abs/2411.14405}, } ``` ## LICENSE This project is licensed under [Apache License Version 2](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) (SPDX-License-identifier: Apache-2.0). ## DISCLAIMER We used compliance checking algorithms during the training process, to ensure the compliance of the trained model and dataset to the best of our ability. Due to complex data and the diversity of language model usage scenarios, we cannot guarantee that the model is completely free of copyright issues or improper content. If you believe anything infringes on your rights or generates improper content, please contact us, and we will promptly address the matter.	null	Non_BioNLP	# Marco-o1-RK3588-1.1.2 This version of Marco-o1 has been converted to run on the RK3588 NPU using ['w8a8', 'w8a8_g128', 'w8a8_g256', 'w8a8_g512'] quantization. This model has been optimized with the following LoRA: Compatible with RKLLM version: 1.1.2 ## Useful links: [Official RKLLM GitHub](https://github.com/airockchip/rknn-llm) [RockhipNPU Reddit](https://reddit.com/r/RockchipNPU) [EZRKNN-LLM](https://github.com/Pelochus/ezrknn-llm/) Pretty much anything by these folks: [marty1885](https://github.com/marty1885) and [happyme531](https://huggingface.co/happyme531) Converted using https://github.com/c0zaut/ez-er-rkllm-toolkit # Original Model Card for base model, Marco-o1, below: <p align="center"> <img src="assets/logo.png" width="150" style="margin-bottom: 0.2;"/> <p> # 🍓 Marco-o1: Towards Open Reasoning Models for Open-Ended Solutions <!-- Broader Real-World Applications --> <!-- # 🍓 Marco-o1: An Open Large Reasoning Model for Real-World Solutions --> <!-- <h2 align="center"> <a href="https://github.com/AIDC-AI/Marco-o1/">Marco-o1</a></h2> --> <!-- <h5 align="center"> If you appreciate our project, please consider giving us a star ⭐ on GitHub to stay updated with the latest developments. </h2> --> <div align="center"> <!-- Affiliations: --> ⭐ _MarcoPolo Team_ ⭐ [_AI Business, Alibaba International Digital Commerce_](https://aidc-ai.com) [Github](https://github.com/AIDC-AI/Marco-o1) 🤗 [Hugging Face](https://huggingface.co/AIDC-AI/Marco-o1) 📝 [Paper](https://arxiv.org/abs/2411.14405) 🧑‍💻 [Model](https://huggingface.co/AIDC-AI/Marco-o1) 🗂️ [Data](https://github.com/AIDC-AI/Marco-o1/tree/main/data) 📽️ [Demo](https://huggingface.co/AIDC-AI/Marco-o1) </div> 🎯 Marco-o1 not only focuses on disciplines with standard answers, such as mathematics, physics, and coding—which are well-suited for reinforcement learning (RL)—but also places greater emphasis on open-ended resolutions. We aim to address the question: _"Can the o1 model effectively generalize to broader domains where clear standards are absent and rewards are challenging to quantify?"_ Currently, Marco-o1 Large Language Model (LLM) is powered by _Chain-of-Thought (CoT) fine-tuning_, _Monte Carlo Tree Search (MCTS)_, _reflection mechanisms_, and _innovative reasoning strategies_—optimized for complex real-world problem-solving tasks. ⚠️ Limitations: <ins>We would like to emphasize that this research work is inspired by OpenAI's o1 (from which the name is also derived). This work aims to explore potential approaches to shed light on the currently unclear technical roadmap for large reasoning models. Besides, our focus is on open-ended questions, and we have observed interesting phenomena in multilingual applications. However, we must acknowledge that the current model primarily exhibits o1-like reasoning characteristics and its performance still fall short of a fully realized "o1" model. This is not a one-time effort, and we remain committed to continuous optimization and ongoing improvement.</ins> ![img.png](assets/img.png) ## 🚀 Highlights Currently, our work is distinguished by the following highlights: - 🍀 Fine-Tuning with CoT Data: We develop Marco-o1-CoT by performing full-parameter fine-tuning on the base model using open-source CoT dataset combined with our self-developed synthetic data. - 🍀 Solution Space Expansion via MCTS: We integrate LLMs with MCTS (Marco-o1-MCTS), using the model's output confidence to guide the search and expand the solution space. - 🍀 Reasoning Action Strategy: We implement novel reasoning action strategies and a reflection mechanism (Marco-o1-MCTS Mini-Step), including exploring different action granularities within the MCTS framework and prompting the model to self-reflect, thereby significantly enhancing the model's ability to solve complex problems. - 🍀 Application in Translation Tasks: We are the first to apply Large Reasoning Models (LRM) to Machine Translation task, exploring inference time scaling laws in the multilingual and translation domain. OpenAI recently introduced the groundbreaking o1 model, renowned for its exceptional reasoning capabilities. This model has demonstrated outstanding performance on platforms such as AIME, CodeForces, surpassing other leading models. Inspired by this success, we aimed to push the boundaries of LLMs even further, enhancing their reasoning abilities to tackle complex, real-world challenges. 🌍 Marco-o1 leverages advanced techniques like CoT fine-tuning, MCTS, and Reasoning Action Strategies to enhance its reasoning power. As shown in Figure 2, by fine-tuning Qwen2-7B-Instruct with a combination of the filtered Open-O1 CoT dataset, Marco-o1 CoT dataset, and Marco-o1 Instruction dataset, Marco-o1 improved its handling of complex tasks. MCTS allows exploration of multiple reasoning paths using confidence scores derived from softmax-applied log probabilities of the top-k alternative tokens, guiding the model to optimal solutions. Moreover, our reasoning action strategy involves varying the granularity of actions within steps and mini-steps to optimize search efficiency and accuracy. <div align="center"> <img src="assets/intro_2.jpg" alt="Figure Description or Alt Text" width="80%"> <p><strong>Figure 2: </strong>The overview of Marco-o1.</p> </div> 🌏 As shown in Figure 3, Marco-o1 achieved accuracy improvements of +6.17% on the MGSM (English) dataset and +5.60% on the MGSM (Chinese) dataset, showcasing enhanced reasoning capabilities. <div align="center"> <img src="assets/results.jpg" alt="Figure Description or Alt Text" width="80%"> <p><strong>Figure 3: </strong>The main results of Marco-o1.</p> </div> 🌎 Additionally, in translation tasks, we demonstrate that Marco-o1 excels in translating slang expressions, such as translating "这个鞋拥有踩屎感" (literal translation: "This shoe offers a stepping-on-poop sensation.") to "This shoe has a comfortable sole," demonstrating its superior grasp of colloquial nuances. <div align="center"> <img src="assets/translation.jpg" alt="Figure Description or Alt Text" width="80%"> <p><strong>Figure 4: </strong>The demostration of translation task using Marco-o1.</p> </div> For more information,please visit our [Github](https://github.com/AIDC-AI/Marco-o1). ## Usage 1. Load Marco-o1-CoT model: ``` # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("AIDC-AI/Marco-o1") model = AutoModelForCausalLM.from_pretrained("AIDC-AI/Marco-o1") ``` 2. Inference: Execute the inference script (you can give any customized inputs inside): ``` ./src/talk_with_model.py # Use vLLM ./src/talk_with_model_vllm.py ``` # 👨🏻‍💻 Acknowledgement ## Main Contributors From MarcoPolo Team, AI Business, Alibaba International Digital Commerce: - Yu Zhao - [Huifeng Yin](https://github.com/HuifengYin) - Hao Wang - [Longyue Wang](http://www.longyuewang.com) ## Citation If you find Marco-o1 useful for your research and applications, please cite: ``` @misc{zhao2024marcoo1openreasoningmodels, title={Marco-o1: Towards Open Reasoning Models for Open-Ended Solutions}, author={Yu Zhao and Huifeng Yin and Bo Zeng and Hao Wang and Tianqi Shi and Chenyang Lyu and Longyue Wang and Weihua Luo and Kaifu Zhang}, year={2024}, eprint={2411.14405}, archivePrefix={arXiv}, primaryClass={cs.CL}, url={https://arxiv.org/abs/2411.14405}, } ``` ## LICENSE This project is licensed under [Apache License Version 2](https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md) (SPDX-License-identifier: Apache-2.0). ## DISCLAIMER We used compliance checking algorithms during the training process, to ensure the compliance of the trained model and dataset to the best of our ability. Due to complex data and the diversity of language model usage scenarios, we cannot guarantee that the model is completely free of copyright issues or improper content. If you believe anything infringes on your rights or generates improper content, please contact us, and we will promptly address the matter.	{"library_name": "transformers", "license": "apache-2.0", "inference": false}	task	[ "TRANSLATION" ]	43,085
Bronsn/ganda_llama_8b_64	Bronsn	translation	[ "peft", "safetensors", "llama", "llama-3.1", "gemma-2b", "finetuned", "english-luganda", "translation", "qlora", "en", "lug", "region:us" ]	2025-01-20T20:13:21Z	2025-01-20T20:56:12+00:00	0	0	--- language: - en - lug tags: - llama-3.1 - gemma-2b - finetuned - english-luganda - translation - peft - qlora --- # final_model_8b_64 This model is finetuned for English-Luganda bidirectional translation tasks. It's trained using QLoRA (Quantized Low-Rank Adaptation) on the original LLaMA-3.1-8B model. ## Model Details ### Base Model Information - Base model: unsloth/Meta-Llama-3.1-8B - Model family: LLaMA-3.1-8B - Type: Base - Original model size: 8B parameters ### Training Configuration - Training method: QLoRA (4-bit quantization) - LoRA rank (r): 64 - LoRA alpha: 64 - Target modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj - LoRA dropout: 0 - Learning rate: 2e-5 - Batch size: 2 - Gradient accumulation steps: 4 - Max sequence length: 2048 - Weight decay: 0.01 - Training steps: 100,000 - Warmup steps: 1000 - Save interval: 10,000 steps - Optimizer: AdamW (8-bit) - LR scheduler: Cosine - Mixed precision: bf16 - Gradient checkpointing: Enabled (unsloth) ### Dataset Information - Training data: Parallel English-Luganda corpus - Data sources: - SALT dataset (salt-train-v1.4) - Extracted parallel sentences - Synthetic code-mixed data - Bidirectional translation: Trained on both English→Luganda and Luganda→English - Total training examples: Varies by direction ### Usage This model uses an instruction-based prompt format: ``` Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request. ### Instruction: Translate the following text to [target_lang] ### Input: [input text] ### Response: [translation] ``` ## Training Infrastructure - Trained using unsloth optimization library - Hardware: Single A100 GPU - Quantization: 4-bit training enabled ## Limitations - The model is specialized for English-Luganda translation - Performance may vary based on domain and complexity of text - Limited to the context length of 64 tokens ## Citation and Contact If you use this model, please cite: - Original LLaMA-3.1 model by Meta AI - QLoRA paper: Dettmers et al. (2023) - unsloth optimization library	null	Non_BioNLP	# final_model_8b_64 This model is finetuned for English-Luganda bidirectional translation tasks. It's trained using QLoRA (Quantized Low-Rank Adaptation) on the original LLaMA-3.1-8B model. ## Model Details ### Base Model Information - Base model: unsloth/Meta-Llama-3.1-8B - Model family: LLaMA-3.1-8B - Type: Base - Original model size: 8B parameters ### Training Configuration - Training method: QLoRA (4-bit quantization) - LoRA rank (r): 64 - LoRA alpha: 64 - Target modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj - LoRA dropout: 0 - Learning rate: 2e-5 - Batch size: 2 - Gradient accumulation steps: 4 - Max sequence length: 2048 - Weight decay: 0.01 - Training steps: 100,000 - Warmup steps: 1000 - Save interval: 10,000 steps - Optimizer: AdamW (8-bit) - LR scheduler: Cosine - Mixed precision: bf16 - Gradient checkpointing: Enabled (unsloth) ### Dataset Information - Training data: Parallel English-Luganda corpus - Data sources: - SALT dataset (salt-train-v1.4) - Extracted parallel sentences - Synthetic code-mixed data - Bidirectional translation: Trained on both English→Luganda and Luganda→English - Total training examples: Varies by direction ### Usage This model uses an instruction-based prompt format: ``` Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request. ### Instruction: Translate the following text to [target_lang] ### Input: [input text] ### Response: [translation] ``` ## Training Infrastructure - Trained using unsloth optimization library - Hardware: Single A100 GPU - Quantization: 4-bit training enabled ## Limitations - The model is specialized for English-Luganda translation - Performance may vary based on domain and complexity of text - Limited to the context length of 64 tokens ## Citation and Contact If you use this model, please cite: - Original LLaMA-3.1 model by Meta AI - QLoRA paper: Dettmers et al. (2023) - unsloth optimization library	{"language": ["en", "lug"], "tags": ["llama-3.1", "gemma-2b", "finetuned", "english-luganda", "translation", "peft", "qlora"]}	task	[ "TRANSLATION" ]	43,086
sgugger/push-to-hub-test-2	sgugger	text-classification	[ "transformers", "pytorch", "bert", "text-classification", "generated_from_trainer", "en", "dataset:glue", "base_model:google-bert/bert-base-cased", "base_model:finetune:google-bert/bert-base-cased", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-08-04T15:37:44Z	2023-08-04T15:39:19+00:00	11	0	--- base_model: bert-base-cased datasets: - glue language: - en license: apache-2.0 metrics: - accuracy - f1 tags: - generated_from_trainer model-index: - name: push-to-hub-test-2 results: - task: type: text-classification name: Text Classification dataset: name: GLUE MRPC type: glue config: mrpc split: validation args: mrpc metrics: - type: accuracy value: 0.8676470588235294 name: Accuracy - type: f1 value: 0.9078498293515359 name: F1 --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # push-to-hub-test-2 This model is a fine-tuned version of [bert-base-cased](https://huggingface.co/bert-base-cased) on the GLUE MRPC dataset. It achieves the following results on the evaluation set: - Loss: 0.6255 - Accuracy: 0.8676 - F1: 0.9078 - Combined Score: 0.8877 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3.0 ### Training results ### Framework versions - Transformers 4.32.0.dev0 - Pytorch 2.0.0+cu117 - Datasets 2.14.4.dev0 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # push-to-hub-test-2 This model is a fine-tuned version of [bert-base-cased](https://huggingface.co/bert-base-cased) on the GLUE MRPC dataset. It achieves the following results on the evaluation set: - Loss: 0.6255 - Accuracy: 0.8676 - F1: 0.9078 - Combined Score: 0.8877 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3.0 ### Training results ### Framework versions - Transformers 4.32.0.dev0 - Pytorch 2.0.0+cu117 - Datasets 2.14.4.dev0 - Tokenizers 0.13.3	{"base_model": "bert-base-cased", "datasets": ["glue"], "language": ["en"], "license": "apache-2.0", "metrics": ["accuracy", "f1"], "tags": ["generated_from_trainer"], "model-index": [{"name": "push-to-hub-test-2", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "GLUE MRPC", "type": "glue", "config": "mrpc", "split": "validation", "args": "mrpc"}, "metrics": [{"type": "accuracy", "value": 0.8676470588235294, "name": "Accuracy"}, {"type": "f1", "value": 0.9078498293515359, "name": "F1"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,087
ThuyNT03/xlm-roberta-base-New_VietNam-aug_backtranslation-1	ThuyNT03	text-classification	[ "transformers", "pytorch", "xlm-roberta", "text-classification", "generated_from_trainer", "base_model:FacebookAI/xlm-roberta-base", "base_model:finetune:FacebookAI/xlm-roberta-base", "license:mit", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-08-29T22:27:57Z	2023-08-29T22:39:04+00:00	10	0	--- base_model: xlm-roberta-base license: mit metrics: - accuracy - f1 tags: - generated_from_trainer model-index: - name: xlm-roberta-base-New_VietNam-aug_backtranslation-1 results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # xlm-roberta-base-New_VietNam-aug_backtranslation-1 This model is a fine-tuned version of [xlm-roberta-base](https://huggingface.co/xlm-roberta-base) on the None dataset. It achieves the following results on the evaluation set: - Loss: 1.8120 - Accuracy: 0.71 - F1: 0.7138 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 8 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\|:------:\| \| 0.4667 \| 1.0 \| 86 \| 1.0350 \| 0.71 \| 0.7195 \| \| 0.3031 \| 2.0 \| 172 \| 1.3877 \| 0.72 \| 0.7270 \| \| 0.2394 \| 3.0 \| 258 \| 1.4133 \| 0.7 \| 0.7051 \| \| 0.1781 \| 4.0 \| 344 \| 1.4430 \| 0.69 \| 0.6914 \| \| 0.1565 \| 5.0 \| 430 \| 1.5583 \| 0.71 \| 0.7173 \| \| 0.0964 \| 6.0 \| 516 \| 1.8780 \| 0.69 \| 0.6938 \| \| 0.0779 \| 7.0 \| 602 \| 1.8000 \| 0.72 \| 0.7256 \| \| 0.0846 \| 8.0 \| 688 \| 1.8120 \| 0.71 \| 0.7138 \| ### Framework versions - Transformers 4.32.1 - Pytorch 2.0.0 - Datasets 2.14.4 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # xlm-roberta-base-New_VietNam-aug_backtranslation-1 This model is a fine-tuned version of [xlm-roberta-base](https://huggingface.co/xlm-roberta-base) on the None dataset. It achieves the following results on the evaluation set: - Loss: 1.8120 - Accuracy: 0.71 - F1: 0.7138 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 8 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\|:------:\| \| 0.4667 \| 1.0 \| 86 \| 1.0350 \| 0.71 \| 0.7195 \| \| 0.3031 \| 2.0 \| 172 \| 1.3877 \| 0.72 \| 0.7270 \| \| 0.2394 \| 3.0 \| 258 \| 1.4133 \| 0.7 \| 0.7051 \| \| 0.1781 \| 4.0 \| 344 \| 1.4430 \| 0.69 \| 0.6914 \| \| 0.1565 \| 5.0 \| 430 \| 1.5583 \| 0.71 \| 0.7173 \| \| 0.0964 \| 6.0 \| 516 \| 1.8780 \| 0.69 \| 0.6938 \| \| 0.0779 \| 7.0 \| 602 \| 1.8000 \| 0.72 \| 0.7256 \| \| 0.0846 \| 8.0 \| 688 \| 1.8120 \| 0.71 \| 0.7138 \| ### Framework versions - Transformers 4.32.1 - Pytorch 2.0.0 - Datasets 2.14.4 - Tokenizers 0.13.3	{"base_model": "xlm-roberta-base", "license": "mit", "metrics": ["accuracy", "f1"], "tags": ["generated_from_trainer"], "model-index": [{"name": "xlm-roberta-base-New_VietNam-aug_backtranslation-1", "results": []}]}	task	[ "TRANSLATION" ]	43,088
Nettem-Gayathri/EN-HI-translation	Nettem-Gayathri	text2text-generation	[ "transformers", "tf", "marian", "text2text-generation", "generated_from_keras_callback", "base_model:Helsinki-NLP/opus-mt-en-hi", "base_model:finetune:Helsinki-NLP/opus-mt-en-hi", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-04-22T16:15:40Z	2024-04-22T16:47:23+00:00	4	0	--- base_model: Helsinki-NLP/opus-mt-en-hi license: apache-2.0 tags: - generated_from_keras_callback model-index: - name: EN-HI-translation results: [] --- <!-- This model card has been generated automatically according to the information Keras had access to. You should probably proofread and complete it, then remove this comment. --> # EN-HI-translation This model is a fine-tuned version of [Helsinki-NLP/opus-mt-en-hi](https://huggingface.co/Helsinki-NLP/opus-mt-en-hi) on an unknown dataset. It achieves the following results on the evaluation set: - Train Loss: 3.7639 - Validation Loss: 3.9464 - Epoch: 0 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - optimizer: {'name': 'AdamWeightDecay', 'learning_rate': 2e-05, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weight_decay_rate': 0.01} - training_precision: float32 ### Training results \| Train Loss \| Validation Loss \| Epoch \| \|:----------:\|:---------------:\|:-----:\| \| 3.7639 \| 3.9464 \| 0 \| ### Framework versions - Transformers 4.38.2 - TensorFlow 2.15.0 - Datasets 2.19.0 - Tokenizers 0.15.2 ## To load the model from transformers import AutoTokenizer, AutoModelForSeq2SeqLM tokenizer = AutoTokenizer.from_pretrained("Nettem-Gayathri/EN-HI-translation") model = AutoModelForSeq2SeqLM.from_pretrained("Nettem-Gayathri/EN-HI-translation",from_tf=True)	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information Keras had access to. You should probably proofread and complete it, then remove this comment. --> # EN-HI-translation This model is a fine-tuned version of [Helsinki-NLP/opus-mt-en-hi](https://huggingface.co/Helsinki-NLP/opus-mt-en-hi) on an unknown dataset. It achieves the following results on the evaluation set: - Train Loss: 3.7639 - Validation Loss: 3.9464 - Epoch: 0 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - optimizer: {'name': 'AdamWeightDecay', 'learning_rate': 2e-05, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weight_decay_rate': 0.01} - training_precision: float32 ### Training results \| Train Loss \| Validation Loss \| Epoch \| \|:----------:\|:---------------:\|:-----:\| \| 3.7639 \| 3.9464 \| 0 \| ### Framework versions - Transformers 4.38.2 - TensorFlow 2.15.0 - Datasets 2.19.0 - Tokenizers 0.15.2 ## To load the model from transformers import AutoTokenizer, AutoModelForSeq2SeqLM tokenizer = AutoTokenizer.from_pretrained("Nettem-Gayathri/EN-HI-translation") model = AutoModelForSeq2SeqLM.from_pretrained("Nettem-Gayathri/EN-HI-translation",from_tf=True)	{"base_model": "Helsinki-NLP/opus-mt-en-hi", "license": "apache-2.0", "tags": ["generated_from_keras_callback"], "model-index": [{"name": "EN-HI-translation", "results": []}]}	task	[ "TRANSLATION" ]	43,089
readerbench/RoSummary-large	readerbench	text-generation	[ "transformers", "pytorch", "tf", "gpt2", "text-generation", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "region:us" ]	2022-10-19T06:42:50Z	2022-10-20T10:00:37+00:00	14	1	--- {} --- Model card for RoSummary-large --- language: - ro --- # RoSummary This is a version of the RoGPT2 model trained on the [AlephNews](https://huggingface.co/datasets/readerbench/AlephNews) dataset for the summarization task. There are 3 trained versions, they are available on the HuggingFace Hub: * [base](https://huggingface.co/readerbench/RoSummary-base) * [medium](https://huggingface.co/readerbench/RoSummary-medium) * [large](https://huggingface.co/readerbench/RoSummary-large) ## Evaluation on [AlephNews](https://huggingface.co/datasets/readerbench/AlephNews) \| Model \| Decode Method \| \| BERTScore \| \| \| ROUGE \| \| \|:------:\|:--------------:\|:---------:\|:---------:\|:--------:\|:--------:\|:--------:\|:--------:\| \| \| \| Precision \| Recall \| F1-Score \| ROUGE-1 \| ROUGE-2 \| ROUGE-L \| \| \| Greedy \| 0.7335 \| 0.7399 \| 0.7358 \| 0.3360 \| 0.1862 \| 0.3333 \| \| Base \| Beam Search \| 0.7354 \| 0.7468 \| 0.7404 \| 0.3480 \| 0.1991 \| 0.3416 \| \| \| Top-p Sampling \| 0.7296 \| 0.7299 \| 0.7292 \| 0.3058 \| 0.1452 \| 0.2951 \| \| \| Greedy \| 0.7378 \| 0.7401 \| 0.7380 \| 0.3422 \| 0.1922 \| 0.3394 \| \| Medium \| Beam Search \| 0.7390 \| 0.7493\|0.7434\|0.3546\|0.2061\|0.3467\| \| \| Top-p Sampling \| 0.7315 \| 0.7285 \| 0.7294 \| 0.3042 \| 0.1400 \| 0.2921 \| \| \| Greedy \| 0.7376 \| 0.7424 \| 0.7391 \| 0.3414 \| 0.1895 \| 0.3355 \| \| Large \| Beam Search \| 0.7394\| 0.7470 \| 0.7424 \| 0.3492 \| 0.1995 \| 0.3384 \| \| \| Top-p Sampling \| 0.7311 \| 0.7301 \| 0.7299 \| 0.3051 \| 0.1418 \| 0.2931 \| ## Acknowledgments --- Research supported with [Cloud TPUs](https://cloud.google.com/tpu/) from Google's [TensorFlow Research Cloud (TFRC)](https://www.tensorflow.org/tfrc)	null	Non_BioNLP	Model card for RoSummary-large --- language: - ro --- # RoSummary This is a version of the RoGPT2 model trained on the [AlephNews](https://huggingface.co/datasets/readerbench/AlephNews) dataset for the summarization task. There are 3 trained versions, they are available on the HuggingFace Hub: * [base](https://huggingface.co/readerbench/RoSummary-base) * [medium](https://huggingface.co/readerbench/RoSummary-medium) * [large](https://huggingface.co/readerbench/RoSummary-large) ## Evaluation on [AlephNews](https://huggingface.co/datasets/readerbench/AlephNews) \| Model \| Decode Method \| \| BERTScore \| \| \| ROUGE \| \| \|:------:\|:--------------:\|:---------:\|:---------:\|:--------:\|:--------:\|:--------:\|:--------:\| \| \| \| Precision \| Recall \| F1-Score \| ROUGE-1 \| ROUGE-2 \| ROUGE-L \| \| \| Greedy \| 0.7335 \| 0.7399 \| 0.7358 \| 0.3360 \| 0.1862 \| 0.3333 \| \| Base \| Beam Search \| 0.7354 \| 0.7468 \| 0.7404 \| 0.3480 \| 0.1991 \| 0.3416 \| \| \| Top-p Sampling \| 0.7296 \| 0.7299 \| 0.7292 \| 0.3058 \| 0.1452 \| 0.2951 \| \| \| Greedy \| 0.7378 \| 0.7401 \| 0.7380 \| 0.3422 \| 0.1922 \| 0.3394 \| \| Medium \| Beam Search \| 0.7390 \| 0.7493\|0.7434\|0.3546\|0.2061\|0.3467\| \| \| Top-p Sampling \| 0.7315 \| 0.7285 \| 0.7294 \| 0.3042 \| 0.1400 \| 0.2921 \| \| \| Greedy \| 0.7376 \| 0.7424 \| 0.7391 \| 0.3414 \| 0.1895 \| 0.3355 \| \| Large \| Beam Search \| 0.7394\| 0.7470 \| 0.7424 \| 0.3492 \| 0.1995 \| 0.3384 \| \| \| Top-p Sampling \| 0.7311 \| 0.7301 \| 0.7299 \| 0.3051 \| 0.1418 \| 0.2931 \| ## Acknowledgments --- Research supported with [Cloud TPUs](https://cloud.google.com/tpu/) from Google's [TensorFlow Research Cloud (TFRC)](https://www.tensorflow.org/tfrc)	{}	task	[ "SUMMARIZATION" ]	43,090
axolotl-ai-co/gemma-2-9b	axolotl-ai-co	text-generation	[ "transformers", "safetensors", "gemma2", "text-generation", "arxiv:2009.03300", "arxiv:1905.07830", "arxiv:1911.11641", "arxiv:1904.09728", "arxiv:1905.10044", "arxiv:1907.10641", "arxiv:1811.00937", "arxiv:1809.02789", "arxiv:1911.01547", "arxiv:1705.03551", "arxiv:2107.03374", "arxiv:2108.07732", "arxiv:2110.14168", "arxiv:2009.11462", "arxiv:2101.11718", "arxiv:2110.08193", "arxiv:1804.09301", "arxiv:2109.07958", "arxiv:1804.06876", "arxiv:2103.03874", "arxiv:2304.06364", "arxiv:2206.04615", "arxiv:2203.09509", "license:gemma", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "region:us" ]	2024-07-02T13:21:09Z	2024-07-02T13:31:47+00:00	12	1	--- library_name: transformers license: gemma pipeline_tag: text-generation extra_gated_heading: Access Gemma on Hugging Face extra_gated_prompt: To access Gemma on Hugging Face, you’re required to review and agree to Google’s usage license. To do this, please ensure you’re logged in to Hugging Face and click below. Requests are processed immediately. extra_gated_button_content: Acknowledge license --- # Gemma 2 model card Model Page: [Gemma](https://ai.google.dev/gemma/docs) Resources and Technical Documentation: * [Responsible Generative AI Toolkit][rai-toolkit] * [Gemma on Kaggle][kaggle-gemma] * [Gemma on Vertex Model Garden][vertex-mg-gemma] Terms of Use: [Terms](https://www.kaggle.com/models/google/gemma/license/consent/verify/huggingface?returnModelRepoId=google/gemma-2-9b) Authors: Google ## Model Information Summary description and brief definition of inputs and outputs. ### Description Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. They are text-to-text, decoder-only large language models, available in English, with open weights for both pre-trained variants and instruction-tuned variants. Gemma models are well-suited for a variety of text generation tasks, including question answering, summarization, and reasoning. Their relatively small size makes it possible to deploy them in environments with limited resources such as a laptop, desktop or your own cloud infrastructure, democratizing access to state of the art AI models and helping foster innovation for everyone. ### Usage Below we share some code snippets on how to get quickly started with running the model. First make sure to `pip install -U transformers`, then copy the snippet from the section that is relevant for your usecase. #### Running the model on a single / multi GPU ```python # pip install accelerate from transformers import AutoTokenizer, AutoModelForCausalLM import torch tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", device_map="auto", torch_dtype=torch.bfloat16 ) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` <a name="precisions"></a> #### Running the model on a GPU using different precisions The native weights of this model were exported in `bfloat16` precision. You can use `float16`, which may be faster on certain hardware, indicating the `torch_dtype` when loading the model. For convenience, the `float16` revision of the repo contains a copy of the weights already converted to that precision. You can also use `float32` if you skip the dtype, but no precision increase will occur (model weights will just be upcasted to `float32`). See examples below. _Using `torch.float16`_ ```python # pip install accelerate from transformers import AutoTokenizer, AutoModelForCausalLM import torch tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", device_map="auto", torch_dtype=torch.float16, revision="float16", ) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` _Using `torch.bfloat16`_ ```python # pip install accelerate from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", device_map="auto", torch_dtype=torch.bfloat16) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` _Upcasting to `torch.float32`_ ```python # pip install accelerate from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", device_map="auto") input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` #### Quantized Versions through `bitsandbytes` _Using 8-bit precision (int8)_ ```python # pip install bitsandbytes accelerate from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig quantization_config = BitsAndBytesConfig(load_in_8bit=True) tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", quantization_config=quantization_config) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` _Using 4-bit precision_ ```python # pip install bitsandbytes accelerate from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig quantization_config = BitsAndBytesConfig(load_in_4bit=True) tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", quantization_config=quantization_config) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` #### Other optimizations _Flash Attention 2_ First make sure to install `flash-attn` in your environment `pip install flash-attn` ```diff model = AutoModelForCausalLM.from_pretrained( model_id, torch_dtype=torch.float16, + attn_implementation="flash_attention_2" ).to(0) ``` ### Inputs and outputs * Input: Text string, such as a question, a prompt, or a document to be summarized. * Output: Generated English-language text in response to the input, such as an answer to a question, or a summary of a document. ### Citation ```none @article{gemma_2024, title={Gemma}, url={https://www.kaggle.com/m/3301}, DOI={10.34740/KAGGLE/M/3301}, publisher={Kaggle}, author={Gemma Team}, year={2024} } ``` ## Model Data Data used for model training and how the data was processed. ### Training Dataset These models were trained on a dataset of text data that includes a wide variety of sources. The 27B model was trained with 13 trillion tokens and the 9B model was trained with 8 trillion tokens. Here are the key components: * Web Documents: A diverse collection of web text ensures the model is exposed to a broad range of linguistic styles, topics, and vocabulary. Primarily English-language content. * Code: Exposing the model to code helps it to learn the syntax and patterns of programming languages, which improves its ability to generate code or understand code-related questions. * Mathematics: Training on mathematical text helps the model learn logical reasoning, symbolic representation, and to address mathematical queries. The combination of these diverse data sources is crucial for training a powerful language model that can handle a wide variety of different tasks and text formats. ### Data Preprocessing Here are the key data cleaning and filtering methods applied to the training data: * CSAM Filtering: Rigorous CSAM (Child Sexual Abuse Material) filtering was applied at multiple stages in the data preparation process to ensure the exclusion of harmful and illegal content. * Sensitive Data Filtering: As part of making Gemma pre-trained models safe and reliable, automated techniques were used to filter out certain personal information and other sensitive data from training sets. * Additional methods: Filtering based on content quality and safety in line with [our policies][safety-policies]. ## Implementation Information Details about the model internals. ### Hardware Gemma was trained using the latest generation of [Tensor Processing Unit (TPU)][tpu] hardware (TPUv5p). Training large language models requires significant computational power. TPUs, designed specifically for matrix operations common in machine learning, offer several advantages in this domain: * Performance: TPUs are specifically designed to handle the massive computations involved in training LLMs. They can speed up training considerably compared to CPUs. * Memory: TPUs often come with large amounts of high-bandwidth memory, allowing for the handling of large models and batch sizes during training. This can lead to better model quality. * Scalability: TPU Pods (large clusters of TPUs) provide a scalable solution for handling the growing complexity of large foundation models. You can distribute training across multiple TPU devices for faster and more efficient processing. * Cost-effectiveness: In many scenarios, TPUs can provide a more cost-effective solution for training large models compared to CPU-based infrastructure, especially when considering the time and resources saved due to faster training. * These advantages are aligned with [Google's commitments to operate sustainably][sustainability]. ### Software Training was done using [JAX][jax] and [ML Pathways][ml-pathways]. JAX allows researchers to take advantage of the latest generation of hardware, including TPUs, for faster and more efficient training of large models. ML Pathways is Google's latest effort to build artificially intelligent systems capable of generalizing across multiple tasks. This is specially suitable for [foundation models][foundation-models], including large language models like these ones. Together, JAX and ML Pathways are used as described in the [paper about the Gemini family of models][gemini-2-paper]; "the 'single controller' programming model of Jax and Pathways allows a single Python process to orchestrate the entire training run, dramatically simplifying the development workflow." ## Evaluation Model evaluation metrics and results. ### Benchmark Results These models were evaluated against a large collection of different datasets and metrics to cover different aspects of text generation: \| Benchmark \| Metric \| Gemma PT 9B \| Gemma PT 27B \| \| ------------------------------ \| ------------- \| ----------- \| ------------ \| \| [MMLU][mmlu] \| 5-shot, top-1 \| 71.3 \| 75.2 \| \| [HellaSwag][hellaswag] \| 10-shot \| 81.9 \| 86.4 \| \| [PIQA][piqa] \| 0-shot \| 81.7 \| 83.2 \| \| [SocialIQA][socialiqa] \| 0-shot \| 53.4 \| 53.7 \| \| [BoolQ][boolq] \| 0-shot \| 84.2 \| 84.8 \| \| [WinoGrande][winogrande] \| partial score \| 80.6 \| 83.7 \| \| [ARC-e][arc] \| 0-shot \| 88.0 \| 88.6 \| \| [ARC-c][arc] \| 25-shot \| 68.4 \| 71.4 \| \| [TriviaQA][triviaqa] \| 5-shot \| 76.6 \| 83.7 \| \| [Natural Questions][naturalq] \| 5-shot \| 29.2 \| 34.5 \| \| [HumanEval][humaneval] \| pass@1 \| 40.2 \| 51.8 \| \| [MBPP][mbpp] \| 3-shot \| 52.4 \| 62.6 \| \| [GSM8K][gsm8k] \| 5-shot, maj@1 \| 68.6 \| 74.0 \| \| [MATH][math] \| 4-shot \| 36.6 \| 42.3 \| \| [AGIEval][agieval] \| 3-5-shot \| 52.8 \| 55.1 \| \| [BIG-Bench][big-bench] \| 3-shot, CoT \| 68.2 \| 74.9 \| \| ------------------------------ \| ------------- \| ----------- \| ------------ \| ## Ethics and Safety Ethics and safety evaluation approach and results. ### Evaluation Approach Our evaluation methods include structured evaluations and internal red-teaming testing of relevant content policies. Red-teaming was conducted by a number of different teams, each with different goals and human evaluation metrics. These models were evaluated against a number of different categories relevant to ethics and safety, including: * Text-to-Text Content Safety: Human evaluation on prompts covering safety policies including child sexual abuse and exploitation, harassment, violence and gore, and hate speech. * Text-to-Text Representational Harms: Benchmark against relevant academic datasets such as [WinoBias][winobias] and [BBQ Dataset][bbq]. * Memorization: Automated evaluation of memorization of training data, including the risk of personally identifiable information exposure. * Large-scale harm: Tests for "dangerous capabilities," such as chemical, biological, radiological, and nuclear (CBRN) risks. ### Evaluation Results The results of ethics and safety evaluations are within acceptable thresholds for meeting [internal policies][safety-policies] for categories such as child safety, content safety, representational harms, memorization, large-scale harms. On top of robust internal evaluations, the results of well-known safety benchmarks like BBQ, BOLD, Winogender, Winobias, RealToxicity, and TruthfulQA are shown here. #### Gemma 2.0 \| Benchmark \| Metric \| Gemma 2 IT 9B \| Gemma 2 IT 27B \| \| ------------------------ \| ------------- \| --------------- \| ---------------- \| \| [RealToxicity][realtox] \| average \| 8.25 \| 8.84 \| \| [CrowS-Pairs][crows] \| top-1 \| 37.47 \| 36.67 \| \| [BBQ Ambig][bbq] \| 1-shot, top-1 \| 88.58 \| 85.99 \| \| [BBQ Disambig][bbq] \| top-1 \| 82.67 \| 86.94 \| \| [Winogender][winogender] \| top-1 \| 79.17 \| 77.22 \| \| [TruthfulQA][truthfulqa] \| \| 50.27 \| 51.60 \| \| [Winobias 1_2][winobias] \| \| 78.09 \| 81.94 \| \| [Winobias 2_2][winobias] \| \| 95.32 \| 97.22 \| \| [Toxigen][toxigen] \| \| 39.30 \| 38.42 \| \| ------------------------ \| ------------- \| --------------- \| ---------------- \| ## Usage and Limitations These models have certain limitations that users should be aware of. ### Intended Usage Open Large Language Models (LLMs) have a wide range of applications across various industries and domains. The following list of potential uses is not comprehensive. The purpose of this list is to provide contextual information about the possible use-cases that the model creators considered as part of model training and development. * Content Creation and Communication * Text Generation: These models can be used to generate creative text formats such as poems, scripts, code, marketing copy, and email drafts. * Chatbots and Conversational AI: Power conversational interfaces for customer service, virtual assistants, or interactive applications. * Text Summarization: Generate concise summaries of a text corpus, research papers, or reports. * Research and Education * Natural Language Processing (NLP) Research: These models can serve as a foundation for researchers to experiment with NLP techniques, develop algorithms, and contribute to the advancement of the field. * Language Learning Tools: Support interactive language learning experiences, aiding in grammar correction or providing writing practice. * Knowledge Exploration: Assist researchers in exploring large bodies of text by generating summaries or answering questions about specific topics. ### Limitations * Training Data * The quality and diversity of the training data significantly influence the model's capabilities. Biases or gaps in the training data can lead to limitations in the model's responses. * The scope of the training dataset determines the subject areas the model can handle effectively. * Context and Task Complexity * LLMs are better at tasks that can be framed with clear prompts and instructions. Open-ended or highly complex tasks might be challenging. * A model's performance can be influenced by the amount of context provided (longer context generally leads to better outputs, up to a certain point). * Language Ambiguity and Nuance * Natural language is inherently complex. LLMs might struggle to grasp subtle nuances, sarcasm, or figurative language. * Factual Accuracy * LLMs generate responses based on information they learned from their training datasets, but they are not knowledge bases. They may generate incorrect or outdated factual statements. * Common Sense * LLMs rely on statistical patterns in language. They might lack the ability to apply common sense reasoning in certain situations. ### Ethical Considerations and Risks The development of large language models (LLMs) raises several ethical concerns. In creating an open model, we have carefully considered the following: * Bias and Fairness * LLMs trained on large-scale, real-world text data can reflect socio-cultural biases embedded in the training material. These models underwent careful scrutiny, input data pre-processing described and posterior evaluations reported in this card. * Misinformation and Misuse * LLMs can be misused to generate text that is false, misleading, or harmful. * Guidelines are provided for responsible use with the model, see the [Responsible Generative AI Toolkit][rai-toolkit]. * Transparency and Accountability: * This model card summarizes details on the models' architecture, capabilities, limitations, and evaluation processes. * A responsibly developed open model offers the opportunity to share innovation by making LLM technology accessible to developers and researchers across the AI ecosystem. Risks identified and mitigations: * Perpetuation of biases: It's encouraged to perform continuous monitoring (using evaluation metrics, human review) and the exploration of de-biasing techniques during model training, fine-tuning, and other use cases. * Generation of harmful content: Mechanisms and guidelines for content safety are essential. Developers are encouraged to exercise caution and implement appropriate content safety safeguards based on their specific product policies and application use cases. * Misuse for malicious purposes: Technical limitations and developer and end-user education can help mitigate against malicious applications of LLMs. Educational resources and reporting mechanisms for users to flag misuse are provided. Prohibited uses of Gemma models are outlined in the [Gemma Prohibited Use Policy][prohibited-use]. * Privacy violations: Models were trained on data filtered for removal of PII (Personally Identifiable Information). Developers are encouraged to adhere to privacy regulations with privacy-preserving techniques. ### Benefits At the time of release, this family of models provides high-performance open large language model implementations designed from the ground up for Responsible AI development compared to similarly sized models. Using the benchmark evaluation metrics described in this document, these models have shown to provide superior performance to other, comparably-sized open model alternatives. [rai-toolkit]: https://ai.google.dev/responsible [kaggle-gemma]: https://www.kaggle.com/models/google/gemma-2 [terms]: https://ai.google.dev/gemma/terms [vertex-mg-gemma]: https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/335 [sensitive-info]: https://cloud.google.com/dlp/docs/high-sensitivity-infotypes-reference [safety-policies]: https://storage.googleapis.com/gweb-uniblog-publish-prod/documents/2023_Google_AI_Principles_Progress_Update.pdf#page=11 [prohibited-use]: https://ai.google.dev/gemma/prohibited_use_policy [tpu]: https://cloud.google.com/tpu/docs/intro-to-tpu [sustainability]: https://sustainability.google/operating-sustainably/ [jax]: https://github.com/google/jax [ml-pathways]: https://blog.google/technology/ai/introducing-pathways-next-generation-ai-architecture/ [sustainability]: https://sustainability.google/operating-sustainably/ [foundation-models]: https://ai.google/discover/foundation-models/ [gemini-2-paper]: https://goo.gle/gemma2report [mmlu]: https://arxiv.org/abs/2009.03300 [hellaswag]: https://arxiv.org/abs/1905.07830 [piqa]: https://arxiv.org/abs/1911.11641 [socialiqa]: https://arxiv.org/abs/1904.09728 [boolq]: https://arxiv.org/abs/1905.10044 [winogrande]: https://arxiv.org/abs/1907.10641 [commonsenseqa]: https://arxiv.org/abs/1811.00937 [openbookqa]: https://arxiv.org/abs/1809.02789 [arc]: https://arxiv.org/abs/1911.01547 [triviaqa]: https://arxiv.org/abs/1705.03551 [naturalq]: https://github.com/google-research-datasets/natural-questions [humaneval]: https://arxiv.org/abs/2107.03374 [mbpp]: https://arxiv.org/abs/2108.07732 [gsm8k]: https://arxiv.org/abs/2110.14168 [realtox]: https://arxiv.org/abs/2009.11462 [bold]: https://arxiv.org/abs/2101.11718 [crows]: https://aclanthology.org/2020.emnlp-main.154/ [bbq]: https://arxiv.org/abs/2110.08193v2 [winogender]: https://arxiv.org/abs/1804.09301 [truthfulqa]: https://arxiv.org/abs/2109.07958 [winobias]: https://arxiv.org/abs/1804.06876 [math]: https://arxiv.org/abs/2103.03874 [agieval]: https://arxiv.org/abs/2304.06364 [big-bench]: https://arxiv.org/abs/2206.04615 [toxigen]: https://arxiv.org/abs/2203.09509	null	Non_BioNLP	# Gemma 2 model card Model Page: [Gemma](https://ai.google.dev/gemma/docs) Resources and Technical Documentation: * [Responsible Generative AI Toolkit][rai-toolkit] * [Gemma on Kaggle][kaggle-gemma] * [Gemma on Vertex Model Garden][vertex-mg-gemma] Terms of Use: [Terms](https://www.kaggle.com/models/google/gemma/license/consent/verify/huggingface?returnModelRepoId=google/gemma-2-9b) Authors: Google ## Model Information Summary description and brief definition of inputs and outputs. ### Description Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. They are text-to-text, decoder-only large language models, available in English, with open weights for both pre-trained variants and instruction-tuned variants. Gemma models are well-suited for a variety of text generation tasks, including question answering, summarization, and reasoning. Their relatively small size makes it possible to deploy them in environments with limited resources such as a laptop, desktop or your own cloud infrastructure, democratizing access to state of the art AI models and helping foster innovation for everyone. ### Usage Below we share some code snippets on how to get quickly started with running the model. First make sure to `pip install -U transformers`, then copy the snippet from the section that is relevant for your usecase. #### Running the model on a single / multi GPU ```python # pip install accelerate from transformers import AutoTokenizer, AutoModelForCausalLM import torch tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", device_map="auto", torch_dtype=torch.bfloat16 ) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` <a name="precisions"></a> #### Running the model on a GPU using different precisions The native weights of this model were exported in `bfloat16` precision. You can use `float16`, which may be faster on certain hardware, indicating the `torch_dtype` when loading the model. For convenience, the `float16` revision of the repo contains a copy of the weights already converted to that precision. You can also use `float32` if you skip the dtype, but no precision increase will occur (model weights will just be upcasted to `float32`). See examples below. _Using `torch.float16`_ ```python # pip install accelerate from transformers import AutoTokenizer, AutoModelForCausalLM import torch tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", device_map="auto", torch_dtype=torch.float16, revision="float16", ) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` _Using `torch.bfloat16`_ ```python # pip install accelerate from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", device_map="auto", torch_dtype=torch.bfloat16) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` _Upcasting to `torch.float32`_ ```python # pip install accelerate from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", device_map="auto") input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` #### Quantized Versions through `bitsandbytes` _Using 8-bit precision (int8)_ ```python # pip install bitsandbytes accelerate from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig quantization_config = BitsAndBytesConfig(load_in_8bit=True) tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", quantization_config=quantization_config) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` _Using 4-bit precision_ ```python # pip install bitsandbytes accelerate from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig quantization_config = BitsAndBytesConfig(load_in_4bit=True) tokenizer = AutoTokenizer.from_pretrained("google/gemma-2-9b") model = AutoModelForCausalLM.from_pretrained( "google/gemma-2-9b", quantization_config=quantization_config) input_text = "Write me a poem about Machine Learning." input_ids = tokenizer(input_text, return_tensors="pt").to("cuda") outputs = model.generate(*input_ids) print(tokenizer.decode(outputs[0])) ``` #### Other optimizations _Flash Attention 2_ First make sure to install `flash-attn` in your environment `pip install flash-attn` ```diff model = AutoModelForCausalLM.from_pretrained( model_id, torch_dtype=torch.float16, + attn_implementation="flash_attention_2" ).to(0) ``` ### Inputs and outputs * Input: Text string, such as a question, a prompt, or a document to be summarized. * Output: Generated English-language text in response to the input, such as an answer to a question, or a summary of a document. ### Citation ```none @article{gemma_2024, title={Gemma}, url={https://www.kaggle.com/m/3301}, DOI={10.34740/KAGGLE/M/3301}, publisher={Kaggle}, author={Gemma Team}, year={2024} } ``` ## Model Data Data used for model training and how the data was processed. ### Training Dataset These models were trained on a dataset of text data that includes a wide variety of sources. The 27B model was trained with 13 trillion tokens and the 9B model was trained with 8 trillion tokens. Here are the key components: * Web Documents: A diverse collection of web text ensures the model is exposed to a broad range of linguistic styles, topics, and vocabulary. Primarily English-language content. * Code: Exposing the model to code helps it to learn the syntax and patterns of programming languages, which improves its ability to generate code or understand code-related questions. * Mathematics: Training on mathematical text helps the model learn logical reasoning, symbolic representation, and to address mathematical queries. The combination of these diverse data sources is crucial for training a powerful language model that can handle a wide variety of different tasks and text formats. ### Data Preprocessing Here are the key data cleaning and filtering methods applied to the training data: * CSAM Filtering: Rigorous CSAM (Child Sexual Abuse Material) filtering was applied at multiple stages in the data preparation process to ensure the exclusion of harmful and illegal content. * Sensitive Data Filtering: As part of making Gemma pre-trained models safe and reliable, automated techniques were used to filter out certain personal information and other sensitive data from training sets. * Additional methods: Filtering based on content quality and safety in line with [our policies][safety-policies]. ## Implementation Information Details about the model internals. ### Hardware Gemma was trained using the latest generation of [Tensor Processing Unit (TPU)][tpu] hardware (TPUv5p). Training large language models requires significant computational power. TPUs, designed specifically for matrix operations common in machine learning, offer several advantages in this domain: * Performance: TPUs are specifically designed to handle the massive computations involved in training LLMs. They can speed up training considerably compared to CPUs. * Memory: TPUs often come with large amounts of high-bandwidth memory, allowing for the handling of large models and batch sizes during training. This can lead to better model quality. * Scalability: TPU Pods (large clusters of TPUs) provide a scalable solution for handling the growing complexity of large foundation models. You can distribute training across multiple TPU devices for faster and more efficient processing. * Cost-effectiveness: In many scenarios, TPUs can provide a more cost-effective solution for training large models compared to CPU-based infrastructure, especially when considering the time and resources saved due to faster training. * These advantages are aligned with [Google's commitments to operate sustainably][sustainability]. ### Software Training was done using [JAX][jax] and [ML Pathways][ml-pathways]. JAX allows researchers to take advantage of the latest generation of hardware, including TPUs, for faster and more efficient training of large models. ML Pathways is Google's latest effort to build artificially intelligent systems capable of generalizing across multiple tasks. This is specially suitable for [foundation models][foundation-models], including large language models like these ones. Together, JAX and ML Pathways are used as described in the [paper about the Gemini family of models][gemini-2-paper]; "the 'single controller' programming model of Jax and Pathways allows a single Python process to orchestrate the entire training run, dramatically simplifying the development workflow." ## Evaluation Model evaluation metrics and results. ### Benchmark Results These models were evaluated against a large collection of different datasets and metrics to cover different aspects of text generation: \| Benchmark \| Metric \| Gemma PT 9B \| Gemma PT 27B \| \| ------------------------------ \| ------------- \| ----------- \| ------------ \| \| [MMLU][mmlu] \| 5-shot, top-1 \| 71.3 \| 75.2 \| \| [HellaSwag][hellaswag] \| 10-shot \| 81.9 \| 86.4 \| \| [PIQA][piqa] \| 0-shot \| 81.7 \| 83.2 \| \| [SocialIQA][socialiqa] \| 0-shot \| 53.4 \| 53.7 \| \| [BoolQ][boolq] \| 0-shot \| 84.2 \| 84.8 \| \| [WinoGrande][winogrande] \| partial score \| 80.6 \| 83.7 \| \| [ARC-e][arc] \| 0-shot \| 88.0 \| 88.6 \| \| [ARC-c][arc] \| 25-shot \| 68.4 \| 71.4 \| \| [TriviaQA][triviaqa] \| 5-shot \| 76.6 \| 83.7 \| \| [Natural Questions][naturalq] \| 5-shot \| 29.2 \| 34.5 \| \| [HumanEval][humaneval] \| pass@1 \| 40.2 \| 51.8 \| \| [MBPP][mbpp] \| 3-shot \| 52.4 \| 62.6 \| \| [GSM8K][gsm8k] \| 5-shot, maj@1 \| 68.6 \| 74.0 \| \| [MATH][math] \| 4-shot \| 36.6 \| 42.3 \| \| [AGIEval][agieval] \| 3-5-shot \| 52.8 \| 55.1 \| \| [BIG-Bench][big-bench] \| 3-shot, CoT \| 68.2 \| 74.9 \| \| ------------------------------ \| ------------- \| ----------- \| ------------ \| ## Ethics and Safety Ethics and safety evaluation approach and results. ### Evaluation Approach Our evaluation methods include structured evaluations and internal red-teaming testing of relevant content policies. Red-teaming was conducted by a number of different teams, each with different goals and human evaluation metrics. These models were evaluated against a number of different categories relevant to ethics and safety, including: * Text-to-Text Content Safety: Human evaluation on prompts covering safety policies including child sexual abuse and exploitation, harassment, violence and gore, and hate speech. * Text-to-Text Representational Harms: Benchmark against relevant academic datasets such as [WinoBias][winobias] and [BBQ Dataset][bbq]. * Memorization: Automated evaluation of memorization of training data, including the risk of personally identifiable information exposure. * Large-scale harm: Tests for "dangerous capabilities," such as chemical, biological, radiological, and nuclear (CBRN) risks. ### Evaluation Results The results of ethics and safety evaluations are within acceptable thresholds for meeting [internal policies][safety-policies] for categories such as child safety, content safety, representational harms, memorization, large-scale harms. On top of robust internal evaluations, the results of well-known safety benchmarks like BBQ, BOLD, Winogender, Winobias, RealToxicity, and TruthfulQA are shown here. #### Gemma 2.0 \| Benchmark \| Metric \| Gemma 2 IT 9B \| Gemma 2 IT 27B \| \| ------------------------ \| ------------- \| --------------- \| ---------------- \| \| [RealToxicity][realtox] \| average \| 8.25 \| 8.84 \| \| [CrowS-Pairs][crows] \| top-1 \| 37.47 \| 36.67 \| \| [BBQ Ambig][bbq] \| 1-shot, top-1 \| 88.58 \| 85.99 \| \| [BBQ Disambig][bbq] \| top-1 \| 82.67 \| 86.94 \| \| [Winogender][winogender] \| top-1 \| 79.17 \| 77.22 \| \| [TruthfulQA][truthfulqa] \| \| 50.27 \| 51.60 \| \| [Winobias 1_2][winobias] \| \| 78.09 \| 81.94 \| \| [Winobias 2_2][winobias] \| \| 95.32 \| 97.22 \| \| [Toxigen][toxigen] \| \| 39.30 \| 38.42 \| \| ------------------------ \| ------------- \| --------------- \| ---------------- \| ## Usage and Limitations These models have certain limitations that users should be aware of. ### Intended Usage Open Large Language Models (LLMs) have a wide range of applications across various industries and domains. The following list of potential uses is not comprehensive. The purpose of this list is to provide contextual information about the possible use-cases that the model creators considered as part of model training and development. * Content Creation and Communication * Text Generation: These models can be used to generate creative text formats such as poems, scripts, code, marketing copy, and email drafts. * Chatbots and Conversational AI: Power conversational interfaces for customer service, virtual assistants, or interactive applications. * Text Summarization: Generate concise summaries of a text corpus, research papers, or reports. * Research and Education * Natural Language Processing (NLP) Research: These models can serve as a foundation for researchers to experiment with NLP techniques, develop algorithms, and contribute to the advancement of the field. * Language Learning Tools: Support interactive language learning experiences, aiding in grammar correction or providing writing practice. * Knowledge Exploration: Assist researchers in exploring large bodies of text by generating summaries or answering questions about specific topics. ### Limitations * Training Data * The quality and diversity of the training data significantly influence the model's capabilities. Biases or gaps in the training data can lead to limitations in the model's responses. * The scope of the training dataset determines the subject areas the model can handle effectively. * Context and Task Complexity * LLMs are better at tasks that can be framed with clear prompts and instructions. Open-ended or highly complex tasks might be challenging. * A model's performance can be influenced by the amount of context provided (longer context generally leads to better outputs, up to a certain point). * Language Ambiguity and Nuance * Natural language is inherently complex. LLMs might struggle to grasp subtle nuances, sarcasm, or figurative language. * Factual Accuracy * LLMs generate responses based on information they learned from their training datasets, but they are not knowledge bases. They may generate incorrect or outdated factual statements. * Common Sense * LLMs rely on statistical patterns in language. They might lack the ability to apply common sense reasoning in certain situations. ### Ethical Considerations and Risks The development of large language models (LLMs) raises several ethical concerns. In creating an open model, we have carefully considered the following: * Bias and Fairness * LLMs trained on large-scale, real-world text data can reflect socio-cultural biases embedded in the training material. These models underwent careful scrutiny, input data pre-processing described and posterior evaluations reported in this card. * Misinformation and Misuse * LLMs can be misused to generate text that is false, misleading, or harmful. * Guidelines are provided for responsible use with the model, see the [Responsible Generative AI Toolkit][rai-toolkit]. * Transparency and Accountability: * This model card summarizes details on the models' architecture, capabilities, limitations, and evaluation processes. * A responsibly developed open model offers the opportunity to share innovation by making LLM technology accessible to developers and researchers across the AI ecosystem. Risks identified and mitigations: * Perpetuation of biases: It's encouraged to perform continuous monitoring (using evaluation metrics, human review) and the exploration of de-biasing techniques during model training, fine-tuning, and other use cases. * Generation of harmful content: Mechanisms and guidelines for content safety are essential. Developers are encouraged to exercise caution and implement appropriate content safety safeguards based on their specific product policies and application use cases. * Misuse for malicious purposes: Technical limitations and developer and end-user education can help mitigate against malicious applications of LLMs. Educational resources and reporting mechanisms for users to flag misuse are provided. Prohibited uses of Gemma models are outlined in the [Gemma Prohibited Use Policy][prohibited-use]. * Privacy violations: Models were trained on data filtered for removal of PII (Personally Identifiable Information). Developers are encouraged to adhere to privacy regulations with privacy-preserving techniques. ### Benefits At the time of release, this family of models provides high-performance open large language model implementations designed from the ground up for Responsible AI development compared to similarly sized models. Using the benchmark evaluation metrics described in this document, these models have shown to provide superior performance to other, comparably-sized open model alternatives. [rai-toolkit]: https://ai.google.dev/responsible [kaggle-gemma]: https://www.kaggle.com/models/google/gemma-2 [terms]: https://ai.google.dev/gemma/terms [vertex-mg-gemma]: https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/335 [sensitive-info]: https://cloud.google.com/dlp/docs/high-sensitivity-infotypes-reference [safety-policies]: https://storage.googleapis.com/gweb-uniblog-publish-prod/documents/2023_Google_AI_Principles_Progress_Update.pdf#page=11 [prohibited-use]: https://ai.google.dev/gemma/prohibited_use_policy [tpu]: https://cloud.google.com/tpu/docs/intro-to-tpu [sustainability]: https://sustainability.google/operating-sustainably/ [jax]: https://github.com/google/jax [ml-pathways]: https://blog.google/technology/ai/introducing-pathways-next-generation-ai-architecture/ [sustainability]: https://sustainability.google/operating-sustainably/ [foundation-models]: https://ai.google/discover/foundation-models/ [gemini-2-paper]: https://goo.gle/gemma2report [mmlu]: https://arxiv.org/abs/2009.03300 [hellaswag]: https://arxiv.org/abs/1905.07830 [piqa]: https://arxiv.org/abs/1911.11641 [socialiqa]: https://arxiv.org/abs/1904.09728 [boolq]: https://arxiv.org/abs/1905.10044 [winogrande]: https://arxiv.org/abs/1907.10641 [commonsenseqa]: https://arxiv.org/abs/1811.00937 [openbookqa]: https://arxiv.org/abs/1809.02789 [arc]: https://arxiv.org/abs/1911.01547 [triviaqa]: https://arxiv.org/abs/1705.03551 [naturalq]: https://github.com/google-research-datasets/natural-questions [humaneval]: https://arxiv.org/abs/2107.03374 [mbpp]: https://arxiv.org/abs/2108.07732 [gsm8k]: https://arxiv.org/abs/2110.14168 [realtox]: https://arxiv.org/abs/2009.11462 [bold]: https://arxiv.org/abs/2101.11718 [crows]: https://aclanthology.org/2020.emnlp-main.154/ [bbq]: https://arxiv.org/abs/2110.08193v2 [winogender]: https://arxiv.org/abs/1804.09301 [truthfulqa]: https://arxiv.org/abs/2109.07958 [winobias]: https://arxiv.org/abs/1804.06876 [math]: https://arxiv.org/abs/2103.03874 [agieval]: https://arxiv.org/abs/2304.06364 [big-bench]: https://arxiv.org/abs/2206.04615 [toxigen]: https://arxiv.org/abs/2203.09509	{"library_name": "transformers", "license": "gemma", "pipeline_tag": "text-generation", "extra_gated_heading": "Access Gemma on Hugging Face", "extra_gated_prompt": "To access Gemma on Hugging Face, you’re required to review and agree to Google’s usage license. To do this, please ensure you’re logged in to Hugging Face and click below. Requests are processed immediately.", "extra_gated_button_content": "Acknowledge license"}	task	[ "QUESTION_ANSWERING", "SUMMARIZATION" ]	43,091
google/t5-efficient-base-ff12000	google	text2text-generation	[ "transformers", "pytorch", "tf", "jax", "t5", "text2text-generation", "deep-narrow", "en", "dataset:c4", "arxiv:2109.10686", "license:apache-2.0", "autotrain_compatible", "text-generation-inference", "region:us" ]	2022-03-02T23:29:05Z	2023-01-24T16:44:42+00:00	121	1	--- datasets: - c4 language: - en license: apache-2.0 tags: - deep-narrow inference: false --- # T5-Efficient-BASE-FF12000 (Deep-Narrow version) T5-Efficient-BASE-FF12000 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5). It is a pretrained-only checkpoint and was released with the paper [Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686) by Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler. In a nutshell, the paper indicates that a Deep-Narrow model architecture is favorable for downstream performance compared to other model architectures of similar parameter count. To quote the paper: > We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased > before considering any other forms of uniform scaling across other dimensions. This is largely due to > how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a > tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise, > a tall base model might also generally more efficient compared to a large model. We generally find > that, regardless of size, even if absolute performance might increase as we continue to stack layers, > the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36 > layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e., > params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params, > FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to > consider. To be more precise, model depth is defined as the number of transformer blocks that are stacked sequentially. A sequence of word embeddings is therefore processed sequentially by each transformer block. ## Details model architecture This model checkpoint - t5-efficient-base-ff12000 - is of model type Base with the following variations: - ff is 12000 It has 562.67 million parameters and thus requires ca. 2250.68 MB of memory in full precision (fp32) or 1125.34 MB of memory in half precision (fp16 or bf16). A summary of the original T5 model architectures can be seen here: \| Model \| nl (el/dl) \| ff \| dm \| kv \| nh \| #Params\| \| ----\| ---- \| ---- \| ---- \| ---- \| ---- \| ----\| \| Tiny \| 4/4 \| 1024 \| 256 \| 32 \| 4 \| 16M\| \| Mini \| 4/4 \| 1536 \| 384 \| 32 \| 8 \| 31M\| \| Small \| 6/6 \| 2048 \| 512 \| 32 \| 8 \| 60M\| \| Base \| 12/12 \| 3072 \| 768 \| 64 \| 12 \| 220M\| \| Large \| 24/24 \| 4096 \| 1024 \| 64 \| 16 \| 738M\| \| Xl \| 24/24 \| 16384 \| 1024 \| 128 \| 32 \| 3B\| \| XXl \| 24/24 \| 65536 \| 1024 \| 128 \| 128 \| 11B\| whereas the following abbreviations are used: \| Abbreviation \| Definition \| \| ----\| ---- \| \| nl \| Number of transformer blocks (depth) \| \| dm \| Dimension of embedding vector (output vector of transformers block) \| \| kv \| Dimension of key/value projection matrix \| \| nh \| Number of attention heads \| \| ff \| Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) \| \| el \| Number of transformer blocks in the encoder (encoder depth) \| \| dl \| Number of transformer blocks in the decoder (decoder depth) \| \| sh \| Signifies that attention heads are shared \| \| skv \| Signifies that key-values projection matrices are tied \| If a model checkpoint has no specific, el or dl than both the number of encoder- and decoder layers correspond to nl. ## Pre-Training The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using the span-based masked language modeling (MLM) objective. ## Fine-Tuning Note: This model is a pretrained checkpoint and has to be fine-tuned for practical usage. The checkpoint was pretrained in English and is therefore only useful for English NLP tasks. You can follow on of the following examples on how to fine-tune the model: PyTorch: - [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization) - [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py) - [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - Note: You will have to slightly adapt the training example here to make it work with an encoder-decoder model. Tensorflow: - [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization) - [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - Note: You will have to slightly adapt the training example here to make it work with an encoder-decoder model. JAX/Flax: - [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization) - [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - Note: You will have to slightly adapt the training example here to make it work with an encoder-decoder model. ## Downstream Performance TODO: Add table if available ## Computational Complexity TODO: Add table if available ## More information We strongly recommend the reader to go carefully through the original paper [Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686) to get a more nuanced understanding of this model checkpoint. As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the sh or skv model architecture variations have not been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future.	null	Non_BioNLP	# T5-Efficient-BASE-FF12000 (Deep-Narrow version) T5-Efficient-BASE-FF12000 is a variation of [Google's original T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) following the [T5 model architecture](https://huggingface.co/docs/transformers/model_doc/t5). It is a pretrained-only checkpoint and was released with the paper [Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686) by Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler. In a nutshell, the paper indicates that a Deep-Narrow model architecture is favorable for downstream performance compared to other model architectures of similar parameter count. To quote the paper: > We generally recommend a DeepNarrow strategy where the model’s depth is preferentially increased > before considering any other forms of uniform scaling across other dimensions. This is largely due to > how much depth influences the Pareto-frontier as shown in earlier sections of the paper. Specifically, a > tall small (deep and narrow) model is generally more efficient compared to the base model. Likewise, > a tall base model might also generally more efficient compared to a large model. We generally find > that, regardless of size, even if absolute performance might increase as we continue to stack layers, > the relative gain of Pareto-efficiency diminishes as we increase the layers, converging at 32 to 36 > layers. Finally, we note that our notion of efficiency here relates to any one compute dimension, i.e., > params, FLOPs or throughput (speed). We report all three key efficiency metrics (number of params, > FLOPS and speed) and leave this decision to the practitioner to decide which compute dimension to > consider. To be more precise, model depth is defined as the number of transformer blocks that are stacked sequentially. A sequence of word embeddings is therefore processed sequentially by each transformer block. ## Details model architecture This model checkpoint - t5-efficient-base-ff12000 - is of model type Base with the following variations: - ff is 12000 It has 562.67 million parameters and thus requires ca. 2250.68 MB of memory in full precision (fp32) or 1125.34 MB of memory in half precision (fp16 or bf16). A summary of the original T5 model architectures can be seen here: \| Model \| nl (el/dl) \| ff \| dm \| kv \| nh \| #Params\| \| ----\| ---- \| ---- \| ---- \| ---- \| ---- \| ----\| \| Tiny \| 4/4 \| 1024 \| 256 \| 32 \| 4 \| 16M\| \| Mini \| 4/4 \| 1536 \| 384 \| 32 \| 8 \| 31M\| \| Small \| 6/6 \| 2048 \| 512 \| 32 \| 8 \| 60M\| \| Base \| 12/12 \| 3072 \| 768 \| 64 \| 12 \| 220M\| \| Large \| 24/24 \| 4096 \| 1024 \| 64 \| 16 \| 738M\| \| Xl \| 24/24 \| 16384 \| 1024 \| 128 \| 32 \| 3B\| \| XXl \| 24/24 \| 65536 \| 1024 \| 128 \| 128 \| 11B\| whereas the following abbreviations are used: \| Abbreviation \| Definition \| \| ----\| ---- \| \| nl \| Number of transformer blocks (depth) \| \| dm \| Dimension of embedding vector (output vector of transformers block) \| \| kv \| Dimension of key/value projection matrix \| \| nh \| Number of attention heads \| \| ff \| Dimension of intermediate vector within transformer block (size of feed-forward projection matrix) \| \| el \| Number of transformer blocks in the encoder (encoder depth) \| \| dl \| Number of transformer blocks in the decoder (decoder depth) \| \| sh \| Signifies that attention heads are shared \| \| skv \| Signifies that key-values projection matrices are tied \| If a model checkpoint has no specific, el or dl than both the number of encoder- and decoder layers correspond to nl. ## Pre-Training The checkpoint was pretrained on the [Colossal, Cleaned version of Common Crawl (C4)](https://huggingface.co/datasets/c4) for 524288 steps using the span-based masked language modeling (MLM) objective. ## Fine-Tuning Note: This model is a pretrained checkpoint and has to be fine-tuned for practical usage. The checkpoint was pretrained in English and is therefore only useful for English NLP tasks. You can follow on of the following examples on how to fine-tune the model: PyTorch: - [Summarization](https://github.com/huggingface/transformers/tree/master/examples/pytorch/summarization) - [Question Answering](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_seq2seq_qa.py) - [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/pytorch/text-classification) - Note: You will have to slightly adapt the training example here to make it work with an encoder-decoder model. Tensorflow: - [Summarization](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/summarization) - [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/tensorflow/text-classification) - Note: You will have to slightly adapt the training example here to make it work with an encoder-decoder model. JAX/Flax: - [Summarization](https://github.com/huggingface/transformers/tree/master/examples/flax/summarization) - [Text Classification](https://github.com/huggingface/transformers/tree/master/examples/flax/text-classification) - Note: You will have to slightly adapt the training example here to make it work with an encoder-decoder model. ## Downstream Performance TODO: Add table if available ## Computational Complexity TODO: Add table if available ## More information We strongly recommend the reader to go carefully through the original paper [Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686) to get a more nuanced understanding of this model checkpoint. As explained in the following [issue](https://github.com/google-research/google-research/issues/986#issuecomment-1035051145), checkpoints including the sh or skv model architecture variations have not been ported to Transformers as they are probably of limited practical usage and are lacking a more detailed description. Those checkpoints are kept [here](https://huggingface.co/NewT5SharedHeadsSharedKeyValues) as they might be ported potentially in the future.	{"datasets": ["c4"], "language": ["en"], "license": "apache-2.0", "tags": ["deep-narrow"], "inference": false}	task	[ "TEXT_CLASSIFICATION", "QUESTION_ANSWERING", "SUMMARIZATION" ]	43,092
christinemahler/aie5-midter-new	christinemahler	sentence-similarity	[ "sentence-transformers", "safetensors", "bert", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:216", "loss:MatryoshkaLoss", "loss:MultipleNegativesRankingLoss", "arxiv:1908.10084", "arxiv:2205.13147", "arxiv:1705.00652", "base_model:Snowflake/snowflake-arctic-embed-l", "base_model:finetune:Snowflake/snowflake-arctic-embed-l", "model-index", "autotrain_compatible", "text-embeddings-inference", "endpoints_compatible", "region:us" ]	2025-02-25T05:16:01Z	2025-02-25T05:17:18+00:00	16	0	--- base_model: Snowflake/snowflake-arctic-embed-l library_name: sentence-transformers metrics: - cosine_accuracy@1 - cosine_accuracy@3 - cosine_accuracy@5 - cosine_accuracy@10 - cosine_precision@1 - cosine_precision@3 - cosine_precision@5 - cosine_precision@10 - cosine_recall@1 - cosine_recall@3 - cosine_recall@5 - cosine_recall@10 - cosine_ndcg@10 - cosine_mrr@10 - cosine_map@100 pipeline_tag: sentence-similarity tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:216 - loss:MatryoshkaLoss - loss:MultipleNegativesRankingLoss widget: - source_sentence: What types of organizations are eligible to apply for the PAR-25-353 opportunity titled "National Cooperative Drug/Device Discovery/Development Groups (NCDDG) for the Treatment of Mental Disorders"? sentences: - "PAR-25-353: National Cooperative Drug/Device Discovery/Development Groups (NCDDG)\ \ for the Treatment of Mental Disorders (U19 Clinical Trial Optional) Section\ \ III. Eligibility Information\n \n \n \n \n 1. Eligible Applicants\n \ \ \n \n \n Eligible Organizations\n \n Higher Education Institutions\n\ \ \n \n \n Public/State Controlled Institutions of Higher Education\n \ \ \n \n Private Institutions of Higher Education\n \n \n \n The\ \ following types of Higher Education Institutions are always encouraged to apply\ \ for NIH support as Public or Private Institutions of Higher Education:\n \ \ \n \n \n Hispanic-serving Institutions\n \n \n Historically Black\ \ Colleges and Universities (HBCUs)\n \n \n Tribally Controlled Colleges\ \ and Universities (TCCUs)\n \n \n Alaska Native and Native Hawaiian Serving\ \ Institutions\n \n \n Asian American Native American Pacific Islander\ \ Serving Institutions (AANAPISIs)\n \n \n \n Nonprofits Other Than Institutions\ \ of Higher Education\n \n \n \n Nonprofits with 501(c)(3) IRS Status (Other\ \ than Institutions of Higher Education)\n \n \n Nonprofits without 501(c)(3)\ \ IRS Status (Other than Institutions of Higher Education)\n \n \n \n For-Profit\ \ Organizations\n \n \n \n Small Businesses\n \n \n For-Profit\ \ Organizations (Other than Small Businesses)\n \n \n \n Local Governments\n\ \ \n \n \n State Governments\n \n \n County Governments\n \n\ \ \n City or Township Governments\n \n \n Special District Governments\n\ \ \n \n Indian/Native American Tribal Governments (Federally Recognized)\n\ \ \n \n Indian/Native American Tribal Governments (Other than Federally\ \ Recognized)\n \n \n \n \n \n Federal Governments\n \n \n \n Eligible\ \ Agencies of the Federal Government\n \n \n U.S. Territory or Possession\n\ \ \n \n \n \n \n Other\n \n \n \n Independent School Districts\n\ \ \n \n Public Housing Authorities/Indian Housing Authorities\n \n\ \ \n Native American Tribal Organizations (other than Federally recognized\ \ tribal governments)\n \n \n Faith-based or Community-based Organizations\n\ \ \n \n Regional Organizations\n \n \n \n \n Foreign Organizations\n\ \ \n \n \n Non-domestic (non-U.S.) Entities (Foreign Organization)\n \ \ \n are not\n \n eligible to apply.\n \n \n \n \n Non-domestic\ \ (non-U.S.) components of U.S. Organizations\n \n are not\n \n \ \ eligible to apply.\n \n \n \n \n Foreign components, as\n \n \ \ defined in the NIH Grants Policy Statement\n \n ,\n \n are\n\ \ \n allowed.\n \n \n \n Required Registrations\n \n \n Applicant\ \ organizations\n \n \n \n Applicant organizations must complete and maintain\ \ the following registrations as described in the\n \n How to Apply- Application\ \ Guide\n \n to be eligible to apply for or receive an award. All registrations\ \ must be completed prior to the application being submitted. Registration can\ \ take 6 weeks or more, so applicants should begin the registration process as\ \ soon as possible. Failure to complete registrations in advance of a due date\ \ is not a valid reason for a late submission, please reference\n \n NIH\ \ Grants Policy Statement Section 2.3.9.2 Electronically Submitted Applications\n\ \ \n for additional information.\n \n \n \n \n System for Award\ \ Management (SAM)\n \n – Applicants must complete and maintain an active\ \ registration,\n \n which requires renewal at least annually\n \ \ \n . The renewal process may require as much time as the initial registration.\ \ SAM registration includes the assignment of a Commercial and Government Entity\ \ (CAGE) Code for domestic organizations which have not already been assigned\ \ a CAGE Code.\n \n \n \n NATO Commercial and Government Entity (NCAGE)\ \ Code\n \n – Foreign organizations must obtain an NCAGE code (in\ \ lieu of a CAGE code) in order to register in SAM.\n \n \n Unique\ \ Entity Identifier (UEI) - A UEI is issued as part of the SAM.gov registration\ \ process. The same UEI must be used for all registrations, as well as on the\ \ grant application.\n \n \n \n \n \n eRA Commons\n \n - Once\ \ the unique organization identifier is established, organizations can register\ \ with eRA Commons in tandem with completing their Grants.gov registration; all\ \ registrations must be in place by time of submission. eRA Commons requires organizations\ \ to identify at least one Signing Official (SO) and at least one Program Director/Principal\ \ Investigator (PD/PI) account in order to submit an application.\n \n \n\ \ \n Grants.gov\n \n – Applicants must have an active SAM registration\ \ in order to complete the Grants.gov registration.\n \n \n \n \n Program\ \ Directors/Principal Investigators (PD(s)/PI(s))\n \n \n \n All PD(s)/PI(s)\ \ must have an eRA Commons account. PD(s)/PI(s) should work with their organizational\ \ officials to either create a new account or to affiliate their existing account\ \ with the applicant organization in eRA Commons. If the PD/PI is also the organizational\ \ Signing Official, they must have two distinct eRA Commons accounts, one for\ \ each role. Obtaining an eRA Commons account can take up to 2 weeks.\n \n\ \ \n \n Eligible Individuals (Program Director/Principal Investigator)\n \ \ \n Any individual(s) with the skills, knowledge, and resources necessary\ \ to carry out the proposed research as the Program Director(s)/Principal Investigator(s)\ \ (PD(s)/PI(s)) is invited to work with his/her organization to develop an application\ \ for support.\n \n \n For institutions/organizations proposing multiple\ \ PDs/PIs, visit the Multiple Program Director/Principal Investigator Policy and\ \ submission details in the Senior/Key Person Profile (Expanded) Component of\ \ the\n \n How to Apply - Application Guide\n \n .\n \n \n \n\ \ \n An NIH intramural scientist may not serve as the PD/PI of an NCDDG but\ \ may participate as a research collaborator or consultant (see\n \n \n \n\ \ Section IV.7\n \n \n \n for more information).\n \n \n \n \n \ \ 2. Cost Sharing\n \n \n This NOFO does not require cost sharing as defined\ \ in the\n \n NIH Grants Policy Statement\n \n \n Section 1.2-\ \ Definitions of Terms.\n \n \n \n \n \n 3. Additional Information on Eligibility\n\ \ \n \n Number of Applications\n \n \n \n \n Applicant organizations\ \ may submit more than one application, provided that each application is scientifically\ \ distinct.\n \n \n The NIH will not accept duplicate or highly overlapping\ \ applications under review at the same time per\n \n NIH Grants Policy\ \ Statement Section 2.3.7.4 Submission of Resubmission Application\n \n \ \ . This means that the NIH will not accept:\n \n \n \n A new (A0) application\ \ that is submitted before issuance of the summary statement from the review of\ \ an overlapping new (A0) or resubmission (A1) application.\n \n \n A\ \ resubmission (A1) application that is submitted before issuance of the summary\ \ statement from the review of the previous new (A0) application.\n \n \n\ \ An application that has substantial overlap with another application pending\ \ appeal of initial peer review (see\n \n NIH Grants Policy Statement\ \ 2.3.9.4 Similar, Essentially Identical, or Identical Applications\n \n\ \ )." - "RFA-DK-26-009: New Investigator Gateway Awards for Collaborative T1D Research\ \ (R03 Clinical Trial Not Allowed) Section VI. Award Administration Information\n\ \ \n \n \n \n 1. Award Notices\n \n \n A Notice of Award (NoA) is\ \ the official authorizing document notifying the applicant that an award has\ \ been made and that funds may be requested from the designated HHS payment system\ \ or office. The NoA is signed by the Grants Management Officer and emailed to\ \ the recipient’s business official.\n \n \n In accepting the award, the\ \ recipient agrees that any activities under the award are subject to all provisions\ \ currently in effect or implemented during the period of the award, other Department\ \ regulations and policies in effect at the time of the award, and applicable\ \ statutory provisions.\n \n \n Recipients must comply with any funding\ \ restrictions described in\n \n Section IV.6. Funding Restrictions\n\ \ \n . Any pre-award costs incurred before receipt of the NoA are at the\ \ applicant's own risk. For more information on the Notice of Award, please refer\ \ to the\n \n NIH Grants Policy Statement Section 5. The Notice of Award\n\ \ \n and NIH Grants & Funding website, see\n \n Award Process.\n\ \ \n \n \n \n \n Institutional Review Board or Independent Ethics Committee\ \ Approval: Recipient institutions must ensure that protocols are reviewed by\ \ their IRB or IEC. To help ensure the safety of participants enrolled in NIH-funded\ \ studies, the recipient must provide NIH copies of documents related to all major\ \ changes in the status of ongoing protocols.\n \n \n \n \n 2. Administrative\ \ and National Policy Requirements\n \n \n The following Federal wide and\ \ HHS-specific policy requirements apply to awards funded through NIH:\n \n\ \ \n \n The rules listed at\n \n 2 CFR Part 200\n \n ,\ \ Uniform Administrative Requirements, Cost Principles, and Audit Requirements\ \ for Federal Awards.\n \n \n All NIH grant and cooperative agreement\ \ awards include the\n \n NIH Grants Policy Statement\n \n \ \ as part of the terms and conditions in the Notice of Award (NoA). The NoA includes\ \ the requirements of this NOFO. For these terms of award, see the\n \n \ \ NIH Grants Policy Statement Part II: Terms and Conditions of NIH Grant Awards,\ \ Subpart A: General\n \n and\n \n Part II: Terms and Conditions\ \ of NIH Grant Awards, Subpart B: Terms and Conditions for Specific Types of Grants,\ \ Recipients, and Activities\n \n .\n \n \n If a recipient receives\ \ an award, the recipient must follow all applicable nondiscrimination laws. The\ \ recipient agrees to this when registering in SAM.gov. The recipient must also\ \ submit an Assurance of Compliance (\n \n HHS-690\n \n ).\ \ To learn more, see the\n \n Laws and Regulations Enforced by the HHS\ \ Office for Civil Rights website\n \n .\n \n \n HHS recognizes\ \ that NIH research projects are often limited in scope for many reasons that\ \ are nondiscriminatory, such as the principal investigator’s scientific interest,\ \ funding limitations, recruitment requirements, and other considerations. Thus,\ \ criteria in research protocols that target or exclude certain populations are\ \ warranted where nondiscriminatory justifications establish that such criteria\ \ are appropriate with respect to the health or safety of the subjects, the scientific\ \ study design, or the purpose of the research. For additional guidance regarding\ \ how the provisions apply to NIH grant programs, please contact the Scientific/Research\ \ Contact that is identified in Section VII under Agency Contacts of this NOFO.\n\ \ \n \n \n \n \n All federal statutes and regulations relevant to federal\ \ financial assistance, including those highlighted in\n \n NIH Grants\ \ Policy Statement Section 4 Public Policy Requirements, Objectives and Other\ \ Appropriation Mandates.\n \n \n \n Recipients are responsible for ensuring\ \ that their activities comply with all applicable federal regulations. NIH may\ \ terminate awards under certain circumstances. See\n \n 2 CFR Part 200.340\ \ Termination\n \n and\n \n NIH Grants Policy Statement Section\ \ 8.5.2 Remedies for Noncompliance or Enforcement Actions: Suspension, Termination,\ \ and Withholding of Support\n \n .\n \n \n Successful recipients\ \ under this NOFO agree that:\n \n \n Where the award funding involves implementing,\ \ acquiring, or upgrading health IT for activities by any funded entity, recipients\ \ and subrecipient(s) are required to: Use health IT that meets standards and\ \ implementation specifications adopted in 45 CFR part 170, Subpart B, if such\ \ standards and implementation specifications can support the activity. Visit\n\ \ \n https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-D/part-170/subpart-B\n\ \ \n to learn more.\n \n \n Where the award funding involves implementing,\ \ acquiring, or upgrading health IT for activities by eligible clinicians in ambulatory\ \ settings, or hospitals, eligible under Sections 4101, 4102, and 4201 of the\ \ HITECH Act, use health IT certified under the ONC Health IT Certification Program\ \ if certified technology can support the activity. Visit\n \n https://www.healthit.gov/topic/certification-ehrs/certification-health-it\n\ \ \n to learn more.\n \n \n Pursuant to the Cybersecurity Act of\ \ 2015, Div. N, § 405, Pub. Law 114-113, 6 USC § 1533(d), the HHS Secretary has\ \ established a common set of voluntary, consensus-based, and industry-led guidelines,\ \ best practices, methodologies, procedures, and processes.\n \n \n Successful\ \ recipients under this NOFO agree that:\n \n \n When recipients, subrecipients,\ \ or third-party entities have:\n \n \n \n ongoing and consistent access\ \ to HHS owned or operated information or operational technology systems; and\n\ \ \n \n receive, maintain, transmit, store, access, exchange, process,\ \ or utilize personal identifiable information (PII) or personal health information\ \ (PHI) obtained from the awarding HHS agency for the purposes of executing the\ \ award.\n \n \n \n Recipients shall develop plans and procedures, modeled\ \ after the\n \n NIST Cybersecurity framework\n \n , to protect\ \ HHS systems and data. Please refer to\n \n NIH Post-Award Monitoring\ \ and Reporting\n \n for additional information.\n \n \n \n Cooperative\ \ Agreement Terms and Conditions of Award\n \n Not Applicable.\n \n \n\ \ \n 3. Data Management and Sharing\n \n Consistent with the 2023 NIH\ \ Policy for Data Management and Sharing, when data management and sharing is\ \ applicable to the award, recipients will be required to adhere to the Data Management\ \ and Sharing requirements as outlined in the\n \n NIH Grants Policy Statement\n\ \ \n . Upon the approval of a Data Management and Sharing Plan, it is required\ \ for recipients to implement the plan as described.\n \n \n \n \n 4. Reporting\n\ \ \n \n When multiple years are involved, recipients will be required to\ \ submit the\n \n Research Performance Progress Report (RPPR)\n \n\ \ annually and financial statements as required in the\n \n NIH Grants\ \ Policy Statement Section 8.4.1 Reporting.\n \n To learn more about post-award\ \ monitoring and reporting, see the NIH Grants & Funding website, see\n \n\ \ Post-Award Monitoring and Reporting\n \n .\n \n \n \n \n A\ \ final RPPR, invention statement, and the expenditure data portion of the Federal\ \ Financial Report are required for closeout of an award, as described in the\n\ \ \n NIH Grants Policy Statement Section 8.6 Closeout\n \n . NIH\ \ NOFOs outline intended research goals and objectives. Post award, NIH will review\ \ and measure performance based on the details and outcomes that are shared within\ \ the RPPR, as described at 2 CFR Part 200.301." - "RFA-HL-26-011: Leadership and Coordination Center (LACC) for the MACS/WIHS Combined\ \ Cohort Study (MWCCS) (U01 Clinical Trials Not Allowed) Section II. Award Information\n\ \ \n \n \n Funding Instrument\n \n Cooperative Agreement: A financial\ \ assistance mechanism used when there will be substantial Federal scientific\ \ or programmatic involvement. Substantial involvement means that, after award,\ \ NIH scientific or program staff will assist, guide, coordinate, or participate\ \ in project activities. See Section VI.2 for additional information about the\ \ substantial involvement for this NOFO.\n \n \n \n Application Types Allowed\n\ \ New\n \n \n \n \n The\n \n OER Glossary\n \n and the\ \ How to Apply Application Guide provide details on these application types. Only\ \ those application types listed here are allowed for this NOFO.\n \n \n \n\ \ Clinical Trial?\n \n Not Allowed: Only accepting applications that do\ \ not propose clinical trials.\n \n \n \n \n \n Need help determining whether\ \ you are doing a clinical trial?\n \n \n \n \n Funds Available and Anticipated\ \ Number of Awards\n \n The following NIH components intend to commit an\ \ estimated total of $3 million to fund a single award in FY 2026:\n \n \n\ \ NHLBI, $1.37 million\n \n \n NIAID, $0.30 million\n \n \n NIMH, $0.30\ \ million\n \n \n NCI, $0.35 million\n \n \n NIA, $0.19 million\n\ \ \n \n NIDA, $0.15 million\n \n \n NICHD, $0.13 million\n \n \n\ \ NIDCR, $0.07 million\n \n \n NIMHD, $0.04 million\n \n \n NIAAA, $0.03\ \ million\n \n \n NINR, $0.03 million\n \n \n ORWH, $0.03 million\n\ \ \n \n NIDCD, $0.02 million\n \n \n \n Award Budget\n \n Application\ \ budgets may not exceed direct costs of $2 million per year in Fiscal Years 2026\ \ through 2031.\n \n \n \n Award Project Period\n \n 6 years\n \n\ \ This variable defines that we need to start a new row. \n \n \n \n NIH grants\ \ policies as described in the\n \n NIH Grants Policy Statement\n \ \ \n will apply to the applications submitted and awards made from this NOFO." - source_sentence: How does the title of opportunity ID [insert ID] align with the strategic goals of the Department of Health and Human Services? sentences: - Department of Health and Human Services - "RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type\ \ 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Section II. Award\ \ Information\n \n \n \n Funding Instrument\n \n Grant: A financial\ \ assistance mechanism providing money, property, or both to an eligible entity\ \ to carry out an approved project or activity.\n \n \n \n Application Types\ \ Allowed\n New\n \n Resubmission - Resubmission applications from\n \ \ \n RFA-DK-22-021\n \n and from the current active NOFO RFA-DK-26-007\ \ may be submitted in response to this NOFO.\n \n \n \n \n The\n \n\ \ OER Glossary\n \n and the How to Apply Application Guide provide\ \ details on these application types. Only those application types listed here\ \ are allowed for this NOFO.\n \n \n \n Clinical Trial?\n \n Not Allowed:\ \ Only accepting applications that do not propose clinical trials.\n \n \n\ \ \n \n \n Need help determining whether you are doing a clinical trial?\n\ \ \n \n \n \n Funds Available and Anticipated Number of Awards\n \n \ \ NIDDK intends to commit $5.5 million in Fiscal Year 2026 to fund 3-4 awards\ \ and $5.5 million in FY 27 to fund 3-4 awards. The number of awards is contingent\ \ upon the submission of a sufficient number of meritorious applications.\n \ \ \n \n \n Award Budget\n \n Application budgets are limited to no more\ \ than $1,200,000 direct costs per year, exclusive of facilities and administrative\ \ (F&A) costs. Budgets are expected to reflect the actual needs of the proposed\ \ project.\n \n \n \n Award Project Period\n \n The maximum project\ \ period is 3 years.\n \n This variable defines that we need to start a new\ \ row. \n \n \n \n NIH grants policies as described in the\n \n NIH\ \ Grants Policy Statement\n \n will apply to the applications submitted\ \ and awards made from this NOFO." - "RFA-DK-26-009: New Investigator Gateway Awards for Collaborative T1D Research\ \ (R03 Clinical Trial Not Allowed) Section IV. Application and Submission Information\n\ \ \n \n \n \n 1. Requesting an Application Package\n \n \n The application\ \ forms package specific to this opportunity must be accessed through ASSIST,\ \ Grants.gov Workspace or an institutional system-to-system solution. Links to\ \ apply using ASSIST or Grants.gov Workspace are available in Part 1 of this NOFO.\ \ See your administrative office for instructions if you plan to use an institutional\ \ system-to-system solution.\n \n \n \n \n 2. Content and Form of Application\ \ Submission\n \n \n It is critical that applicants follow the instructions\ \ in the Research (R) Instructions in the\n \n How to Apply - Application\ \ Guide\n \n except where instructed in this notice of funding opportunity\ \ to do otherwise. Conformance to the requirements in the Application Guide is\ \ required and strictly enforced. Applications that are out of compliance with\ \ these instructions may be delayed or not accepted for review.\n \n \n \n\ \ \n Letter of Intent\n \n \n Although a letter of intent is not required,\ \ is not binding, and does not enter into the review of a subsequent application,\ \ the information that it contains allows IC staff to estimate the potential review\ \ workload and plan the review.\n \n \n By the date listed in\n \n \ \ Part 1. Overview Information\n \n , prospective applicants are asked\ \ to submit a letter of intent that includes the following information:\n \n\ \ \n \n Descriptive title of proposed activity\n \n \n Name(s), address(es),\ \ and telephone number(s) of the PD(s)/PI(s)\n \n \n Names of other key\ \ personnel\n \n \n Participating institution(s)\n \n \n Number\ \ and title of this funding opportunity\n \n \n \n The letter of intent\ \ should be sent to:\n \n \n John Connaughton, Ph.D.\n \n Chief,\ \ Scientific Review Branch\n \n National Institute of Diabetes and Digestive\ \ and Kidney Diseases (NIDDK)\n \n Telephone: 301-594-7797\n \n \ \ Email:\n \n \n [email protected]\n \n \n \n \n \n \n Page Limitations\n\ \ \n \n All page limitations described in the\n \n How to Apply-\ \ Application Guide\n \n and the\n \n Table of Page Limits\n \ \ \n must be followed.\n \n \n \n Instructions for Application Submission\n\ \ \n The following section supplements the instructions found in the\n \ \ \n How to Apply- Application Guide\n \n and should be used for\ \ preparing an application to this NOFO.\n \n \n \n \n SF424(R&R) Cover\n\ \ \n \n All instructions in the\n \n \n How to Apply - Application\ \ Guide\n \n \n must be followed.\n \n \n \n \n SF424(R&R) Project/Performance\ \ Site Locations\n \n \n All instructions in the How to Apply- Application\ \ Guide must be followed.\n \n \n \n \n SF424(R&R) Other Project Information\n\ \ \n \n All instructions in the How to Apply- Application Guide must be\ \ followed.\n \n \n \n \n \n Other Attachments:\n \n \n \n \n Proposed\ \ Affiliation Letter\n \n : Name the PDF formatted letter \"Proposed_Affiliation_Ltr.pdf\"\ . Applications must include a one page letter that includes a brief statement\ \ identifying the specific NIDDK consortium or network that is within scope of\ \ the proposed aims. In addition, the applicant must outline in the letter how\ \ the objectives and design of the application are related to, but distinct from,\ \ ongoing studies in the consortium or network of interest, and describe how developing\ \ a relationship with the chosen consortium or network would be expected to facilitate\ \ the applicant's research and career goals. Applications that lack the proposed\ \ Affiliation Letter are considered incomplete and will not be peer reviewed.\n\ \ \n \n \n \n SF424(R&R) Senior/Key Person Profile\n \n \n All instructions\ \ in the How to Apply- Application Guide must be followed.\n \n \n \n \n \ \ R&R or Modular Budget\n \n \n All instructions in the How to Apply- Application\ \ Guide must be followed.\n \n \n \n \n Applicants should also follow these\ \ additional instructions:\n \n \n Applicants should budget for travel to\ \ one meeting of the proposed affiliated consortium or network per year. Except\ \ in unusual circumstances, only the PD/PI may be supported by R03 funds to travel\ \ to consortium meetings. There is no specific line item in which to list travel\ \ costs ($2000/year) in the modular budget format; instead, the proposed travel\ \ should be described in the budget justification section.\n \n \n \n \n \ \ R&R Subaward Budget\n \n \n All instructions in the How to Apply-Application\ \ Guide must be followed.\n \n \n \n \n PHS 398 Cover Page Supplement\n\ \ \n \n All instructions in the How to Apply- Application Guide must be\ \ followed.\n \n \n \n \n PHS 398 Research Plan\n \n \n All instructions\ \ in the How to Apply- Application Guide must be followed, with the following\ \ additional instructions:\n \n \n \n \n \n Resource Sharing Plan\n \ \ \n : Individuals are required to comply with the instructions for the Resource\ \ Sharing Plans as provided in the How to Apply- Application Guide.\n \n \n\ \ \n \n \n Protocol, tool and reagent sharing\n \n : The applicant\ \ should discuss plans for sharing and distribution of non-data resources that\ \ will be generated by the proposed project, including models, protocols, computational\ \ tools, biomaterials, and reagents. As one of the essential goals of the Gateway program\ \ is to support high impact advances, NIDDK intends that tools and reagents generated\ \ by the Gateway program will be made broadly available and distributed at minimal\ \ cost, and without undue intellectual property constraints, so that they can\ \ be as widely used as possible, thus enabling rapid downstream applications for\ \ the reagents by the larger scientific community. Applicants are expected to\ \ register resources supported by this NOFO with the NIDDK Information Network\ \ (dkNET) at\n \n https://dknet.org/\n \n and use Research Resource\ \ Identifiers (RRID) assigned through dkNET in any publication supported by this\ \ NOFO. Applicants should also be familiar with the NIH statements regarding intellectual\ \ property of resources developed with Federal funds\n \n (\n \n \ \ NIH Research Tools Policy (\n \n \n https://grants.nih.gov/grants/intell-property_64FR72090.pdf\n\ \ \n \n ) and other related NIH sharing policies at\n \n \n http://sharing.\n\ \ \n nih.gov\n \n .\n \n \n \n \n \n Other Plan(s):\n \ \ \n \n \n All instructions in the How to Apply-Application Guide must be\ \ followed, with the following additional instructions:\n \n \n \n \n \n \ \ All applicants planning research (funded or conducted in whole or in part\ \ by NIH) that results in the generation of scientific data are required to comply\ \ with the instructions for the Data Management and Sharing Plan.\n \n \n\ \ \n \n \n \n Appendix:\n \n Only limited Appendix materials are allowed.\ \ Follow all instructions for the Appendix as described in the How to Apply- Application\ \ Guide.\n \n \n \n No publications or other material, with the exception\ \ of blank questionnaires or blank surveys, may be included in the Appendix.\n\ \ \n \n \n \n \n PHS Human Subjects and Clinical Trials Information\n \ \ \n \n When involving human subjects research, clinical research, and/or\ \ NIH-defined clinical trials (and when applicable, clinical trials research experience)\ \ follow all instructions for the PHS Human Subjects and Clinical Trials Information\ \ form in the How to Apply- Application Guide, with the following additional instructions:\n\ \ \n \n If you answered “Yes” to the question “Are Human Subjects Involved?”\ \ on the R&R Other Project Information form, you must include at least one human\ \ subjects study record using the\n \n Study Record: PHS Human Subjects\ \ and Clinical Trials Information\n \n form or\n \n Delayed Onset\ \ Study\n \n record.\n \n \n \n Study Record: PHS Human Subjects\ \ and Clinical Trials Information\n \n \n \n All instructions in the How\ \ to Apply- Application Guide must be followed.\n \n \n \n \n \n Delayed\ \ Onset Study\n \n \n \n Note:\n \n Delayed onset\n \n does\ \ NOT apply to a study that can be described but will not start immediately (i.e.,\ \ delayed start). All instructions in the How to Apply- Application Guide must\ \ be followed.\n \n \n \n \n PHS Assignment Request Form\n \n \n All\ \ instructions in the How to Apply- Application Guide must be followed.\n \n\ \ \n \n \n 3. Unique Entity Identifier and System for Award Management (SAM)\n\ \ \n \n See Part 2. Section III.1 for information regarding the requirement\ \ for obtaining a unique entity identifier and for completing and maintaining\ \ active registrations in System for Award Management (SAM), NATO Commercial and\ \ Government Entity (NCAGE) Code (if applicable), eRA Commons, and Grants.gov\n\ \ \n \n \n \n 4. Submission Dates and Times\n \n \n Part I. contains\ \ information about Key Dates and times. Applicants are encouraged to submit applications\ \ before the due date to ensure they have time to make any application corrections\ \ that might be necessary for successful submission. When a submission date falls\ \ on a weekend or\n \n Federal holiday\n \n , the application deadline\ \ is automatically extended to the next business day.\n \n \n Organizations\ \ must submit applications to\n \n Grants.gov\n \n (the online\ \ portal to find and apply for grants across all Federal agencies). Applicants\ \ must then complete the submission process by tracking the status of the application\ \ in the\n \n eRA Commons\n \n , NIH’s electronic system for grants\ \ administration. NIH and Grants.gov systems check the application against many\ \ of the application instructions upon submission. Errors must be corrected and\ \ a changed/corrected application must be submitted to Grants.gov on or before\ \ the application due date and time. If a Changed/Corrected application is submitted\ \ after the deadline, the application will be considered late. Applications that\ \ miss the due date and time are subjected to the\n \n NIH Grants Policy\ \ Statement Section 2.3.9.2 Electronically Submitted Applications\n \n \ \ .\n \n \n \n Applicants are responsible for viewing their application\ \ before the due date in the eRA Commons to ensure accurate and successful submission.\n\ \ \n \n \n Information on the submission process and a definition of on-time\ \ submission are provided in the How to Apply-Application Guide.\n \n \n \n\ \ \n 5. Intergovernmental Review (E.O. 12372)\n \n \n This initiative\ \ is not subject to\n \n intergovernmental review.\n \n \n \n \n \n\ \ 6. Funding Restrictions\n \n \n All NIH awards are subject to the terms\ \ and conditions, cost principles, and other considerations described in the\n\ \ \n NIH Grants Policy Statement\n \n .\n \n \n Pre-award\ \ costs are allowable only as described in the\n \n NIH Grants Policy\ \ Statement Section 7.9.1 Selected Items of Cost.\n \n \n \n \n 7. Other\ \ Submission Requirements and Information\n \n Applications must be submitted\ \ electronically following the instructions described in the How to Apply Application\ \ Guide. Paper applications will not be accepted.\n \n \n Applicants must\ \ complete all required registrations before the application due date. Section\ \ III. Eligibility Information contains information about registration.\n \n\ \ \n For assistance with your electronic application or for more information\ \ on the electronic submission process, visit\n \n How to Apply – Application\ \ Guide\n \n . If you encounter a system issue beyond your control that\ \ threatens your ability to complete the submission process on-time, you must\ \ follow the\n \n Dealing with System Issues\n \n guidance. For\ \ assistance with application submission, contact the Application Submission Contacts\ \ in Section VII.\n \n \n \n Important reminders:\n \n \n \n All\ \ PD(s)/PI(s) must include their eRA Commons ID in the Credential field of the\ \ Senior/Key Person Profile form\n \n .\n \n Failure to register\ \ in the Commons and to include a valid PD/PI Commons ID in the credential field\ \ will prevent the successful submission of an electronic application to NIH.\ \ See Section III of this NOFO for information on registration requirements.\n\ \ \n \n The applicant organization must ensure that the unique entity identifier\ \ provided on the application is the same identifier used in the organization’s\ \ profile in the eRA Commons and for the System for Award Management. Additional\ \ information may be found in the How to Apply Application Guide.\n \n \n \ \ See\n \n more tips\n \n for avoiding common errors.\n \n\ \ \n \n \n Upon receipt, applications will be evaluated for completeness and\ \ compliance with application instructions by the Center for Scientific Review\ \ and responsiveness by NIDDK. Applications that are incomplete, non-compliant,\ \ and/or nonresponsive will not be reviewed.\n \n \n \n Mandatory Disclosure\n\ \ \n Recipients or subrecipients must submit any information related to\ \ violations of federal criminal law involving fraud, bribery, or gratuity violations\ \ potentially affecting the federal award. See Mandatory Disclosures,\n \n\ \ 2 CFR 200.113\n \n and\n \n NIH Grants Policy Statement Section\ \ 4.1.35\n \n .\n \n \n Send written disclosures to the NIH Chief\ \ Grants Management Officer listed on the Notice of Award for the IC that funded\ \ the award and to the\n \n HHS Office of Inspector Grant Self Disclosure\ \ Program\n \n at\n \n \n [email protected]\n \n \n .\n\ \ \n \n \n \n Post Submission Materials\n \n \n Applicants are required\ \ to follow the instructions for post-submission materials, as described in\n\ \ \n the policy" - source_sentence: What are the main objectives of the Leadership and Coordination Center (LACC) for the MACS/WIHS Combined Cohort Study (MWCCS) as outlined in opportunity id RFA-HL-26-011? sentences: - 'RFA-HL-26-011: Leadership and Coordination Center (LACC) for the MACS/WIHS Combined Cohort Study (MWCCS) (U01 Clinical Trials Not Allowed) Part 2. Full Text of Announcement' - "PAR-25-353: National Cooperative Drug/Device Discovery/Development Groups (NCDDG)\ \ for the Treatment of Mental Disorders (U19 Clinical Trial Optional) Section\ \ VIII. Other Information\n \n \n \n \n Recently issued trans-NIH\n \ \ \n policy notices\n \n may affect your application submission. A\ \ full list of policy notices published by NIH is provided in the\n \n \ \ NIH Guide for Grants and Contracts\n \n . All awards are subject to\ \ the terms and conditions, cost principles, and other considerations described\ \ in the\n \n NIH Grants Policy Statement\n \n .\n \n \n \n\ \ Authority and Regulations\n \n Awards are made under the authorization\ \ of Sections 301 and 405 of the Public Health Service Act as amended (42 USC\ \ 241 and 284) and under Federal Regulations 42 CFR Part 52 and 2 CFR Part 200.\n\ \ \n end row \n \n \n \n \n Weekly TOC for this Announcement\n \n \n\ \ \n NIH Funding Opportunities and Notices\n \n \n \n \n \n \n \n \n \n\ \ \n \n \n \n Department of Health\n \n and Human Services (HHS)\n \n \n\ \ \n \n NIH... Turning Discovery Into Health\n \n ®" - "PAR-25-353: National Cooperative Drug/Device Discovery/Development Groups (NCDDG)\ \ for the Treatment of Mental Disorders (U19 Clinical Trial Optional) Section\ \ III. Eligibility Information\n \n \n \n \n 1. Eligible Applicants\n \ \ \n \n \n Eligible Organizations\n \n Higher Education Institutions\n\ \ \n \n \n Public/State Controlled Institutions of Higher Education\n \ \ \n \n Private Institutions of Higher Education\n \n \n \n The\ \ following types of Higher Education Institutions are always encouraged to apply\ \ for NIH support as Public or Private Institutions of Higher Education:\n \ \ \n \n \n Hispanic-serving Institutions\n \n \n Historically Black\ \ Colleges and Universities (HBCUs)\n \n \n Tribally Controlled Colleges\ \ and Universities (TCCUs)\n \n \n Alaska Native and Native Hawaiian Serving\ \ Institutions\n \n \n Asian American Native American Pacific Islander\ \ Serving Institutions (AANAPISIs)\n \n \n \n Nonprofits Other Than Institutions\ \ of Higher Education\n \n \n \n Nonprofits with 501(c)(3) IRS Status (Other\ \ than Institutions of Higher Education)\n \n \n Nonprofits without 501(c)(3)\ \ IRS Status (Other than Institutions of Higher Education)\n \n \n \n For-Profit\ \ Organizations\n \n \n \n Small Businesses\n \n \n For-Profit\ \ Organizations (Other than Small Businesses)\n \n \n \n Local Governments\n\ \ \n \n \n State Governments\n \n \n County Governments\n \n\ \ \n City or Township Governments\n \n \n Special District Governments\n\ \ \n \n Indian/Native American Tribal Governments (Federally Recognized)\n\ \ \n \n Indian/Native American Tribal Governments (Other than Federally\ \ Recognized)\n \n \n \n \n \n Federal Governments\n \n \n \n Eligible\ \ Agencies of the Federal Government\n \n \n U.S. Territory or Possession\n\ \ \n \n \n \n \n Other\n \n \n \n Independent School Districts\n\ \ \n \n Public Housing Authorities/Indian Housing Authorities\n \n\ \ \n Native American Tribal Organizations (other than Federally recognized\ \ tribal governments)\n \n \n Faith-based or Community-based Organizations\n\ \ \n \n Regional Organizations\n \n \n \n \n Foreign Organizations\n\ \ \n \n \n Non-domestic (non-U.S.) Entities (Foreign Organization)\n \ \ \n are not\n \n eligible to apply.\n \n \n \n \n Non-domestic\ \ (non-U.S.) components of U.S. Organizations\n \n are not\n \n \ \ eligible to apply.\n \n \n \n \n Foreign components, as\n \n \ \ defined in the NIH Grants Policy Statement\n \n ,\n \n are\n\ \ \n allowed.\n \n \n \n Required Registrations\n \n \n Applicant\ \ organizations\n \n \n \n Applicant organizations must complete and maintain\ \ the following registrations as described in the\n \n How to Apply- Application\ \ Guide\n \n to be eligible to apply for or receive an award. All registrations\ \ must be completed prior to the application being submitted. Registration can\ \ take 6 weeks or more, so applicants should begin the registration process as\ \ soon as possible. Failure to complete registrations in advance of a due date\ \ is not a valid reason for a late submission, please reference\n \n NIH\ \ Grants Policy Statement Section 2.3.9.2 Electronically Submitted Applications\n\ \ \n for additional information.\n \n \n \n \n System for Award\ \ Management (SAM)\n \n – Applicants must complete and maintain an active\ \ registration,\n \n which requires renewal at least annually\n \ \ \n . The renewal process may require as much time as the initial registration.\ \ SAM registration includes the assignment of a Commercial and Government Entity\ \ (CAGE) Code for domestic organizations which have not already been assigned\ \ a CAGE Code.\n \n \n \n NATO Commercial and Government Entity (NCAGE)\ \ Code\n \n – Foreign organizations must obtain an NCAGE code (in\ \ lieu of a CAGE code) in order to register in SAM.\n \n \n Unique\ \ Entity Identifier (UEI) - A UEI is issued as part of the SAM.gov registration\ \ process. The same UEI must be used for all registrations, as well as on the\ \ grant application.\n \n \n \n \n \n eRA Commons\n \n - Once\ \ the unique organization identifier is established, organizations can register\ \ with eRA Commons in tandem with completing their Grants.gov registration; all\ \ registrations must be in place by time of submission. eRA Commons requires organizations\ \ to identify at least one Signing Official (SO) and at least one Program Director/Principal\ \ Investigator (PD/PI) account in order to submit an application.\n \n \n\ \ \n Grants.gov\n \n – Applicants must have an active SAM registration\ \ in order to complete the Grants.gov registration.\n \n \n \n \n Program\ \ Directors/Principal Investigators (PD(s)/PI(s))\n \n \n \n All PD(s)/PI(s)\ \ must have an eRA Commons account. PD(s)/PI(s) should work with their organizational\ \ officials to either create a new account or to affiliate their existing account\ \ with the applicant organization in eRA Commons. If the PD/PI is also the organizational\ \ Signing Official, they must have two distinct eRA Commons accounts, one for\ \ each role. Obtaining an eRA Commons account can take up to 2 weeks.\n \n\ \ \n \n Eligible Individuals (Program Director/Principal Investigator)\n \ \ \n Any individual(s) with the skills, knowledge, and resources necessary\ \ to carry out the proposed research as the Program Director(s)/Principal Investigator(s)\ \ (PD(s)/PI(s)) is invited to work with his/her organization to develop an application\ \ for support.\n \n \n For institutions/organizations proposing multiple\ \ PDs/PIs, visit the Multiple Program Director/Principal Investigator Policy and\ \ submission details in the Senior/Key Person Profile (Expanded) Component of\ \ the\n \n How to Apply - Application Guide\n \n .\n \n \n \n\ \ \n An NIH intramural scientist may not serve as the PD/PI of an NCDDG but\ \ may participate as a research collaborator or consultant (see\n \n \n \n\ \ Section IV.7\n \n \n \n for more information).\n \n \n \n \n \ \ 2. Cost Sharing\n \n \n This NOFO does not require cost sharing as defined\ \ in the\n \n NIH Grants Policy Statement\n \n \n Section 1.2-\ \ Definitions of Terms.\n \n \n \n \n \n 3. Additional Information on Eligibility\n\ \ \n \n Number of Applications\n \n \n \n \n Applicant organizations\ \ may submit more than one application, provided that each application is scientifically\ \ distinct.\n \n \n The NIH will not accept duplicate or highly overlapping\ \ applications under review at the same time per\n \n NIH Grants Policy\ \ Statement Section 2.3.7.4 Submission of Resubmission Application\n \n \ \ . This means that the NIH will not accept:\n \n \n \n A new (A0) application\ \ that is submitted before issuance of the summary statement from the review of\ \ an overlapping new (A0) or resubmission (A1) application.\n \n \n A\ \ resubmission (A1) application that is submitted before issuance of the summary\ \ statement from the review of the previous new (A0) application.\n \n \n\ \ An application that has substantial overlap with another application pending\ \ appeal of initial peer review (see\n \n NIH Grants Policy Statement\ \ 2.3.9.4 Similar, Essentially Identical, or Identical Applications\n \n\ \ )." - source_sentence: Can an organization submit multiple applications for the RFA-HL-26-011 opportunity, and what conditions must be met for these applications? sentences: - "PAR-25-378: Intervention Research to Improve Native American Health (R34 Clinical\ \ Trial Optional) Section VI. Award Administration Information\n \n \n \n \n\ \ 1. Award Notices\n \n \n A Notice of Award (NoA) is the official authorizing\ \ document notifying the applicant that an award has been made and that funds\ \ may be requested from the designated HHS payment system or office. The NoA is\ \ signed by the Grants Management Officer and emailed to the recipient’s business\ \ official.\n \n \n In accepting the award, the recipient agrees that any\ \ activities under the award are subject to all provisions currently in effect\ \ or implemented during the period of the award, other Department regulations\ \ and policies in effect at the time of the award, and applicable statutory provisions.\n\ \ \n \n Recipients must comply with any funding restrictions described in\n\ \ \n Section IV.6. Funding Restrictions\n \n . Any pre-award costs\ \ incurred before receipt of the NoA are at the applicant's own risk. For more\ \ information on the Notice of Award, please refer to the\n \n NIH Grants\ \ Policy Statement Section 5. The Notice of Award\n \n and NIH Grants &\ \ Funding website, see\n \n Award Process.\n \n \n \n \n \n Individual\ \ awards are based on the application submitted to, and as approved by, the NIH\ \ and are subject to the IC-specific terms and conditions identified in the NoA.\n\ \ \n \n ClinicalTrials.gov: If an award provides for one or more clinical\ \ trials. By law (Title VIII, Section 801 of Public Law 110-85), the \"responsible\ \ party\" must register and submit results information for certain “applicable\ \ clinical trials” on the ClinicalTrials.gov Protocol Registration and Results\ \ System Information Website (\n \n https://register.clinicaltrials.gov\n\ \ \n ). NIH expects registration and results reporting of all trials whether\ \ required under the law or not. For more information, see\n \n https://grants.nih.gov/policy/clinical-trials/reporting/index.htm\n\ \ \n \n \n Institutional Review Board or Independent Ethics Committee Approval:\ \ Recipient institutions must ensure that all protocols are reviewed by their\ \ IRB or IEC. To help ensure the safety of participants enrolled in NIH-funded\ \ studies, the recipient must provide NIH copies of documents related to all major\ \ changes in the status of ongoing protocols.\n \n \n Data and Safety Monitoring\ \ Requirements: The NIH policy for data and safety monitoring requires oversight\ \ and monitoring of all NIH-conducted or -supported human biomedical and behavioral\ \ intervention studies (clinical trials) to ensure the safety of participants\ \ and the validity and integrity of the data. Further information concerning these\ \ requirements is found at http://grants.nih.gov/grants/policy/hs/data_safety.htm\ \ and in the application instructions (SF424 (R&R) and PHS 398).\n \n \n \ \ Investigational New Drug or Investigational Device Exemption Requirements:\ \ Consistent with federal regulations, clinical research projects involving the\ \ use of investigational therapeutics, vaccines, or other medical interventions\ \ (including licensed products and devices for a purpose other than that for which\ \ they were licensed) in humans under a research protocol must be performed under\ \ a Food and Drug Administration (FDA) investigational new drug (IND) or investigational\ \ device exemption (IDE).\n \n \n \n \n 2. Administrative and National Policy\ \ Requirements\n \n \n The following Federal wide and HHS-specific policy\ \ requirements apply to awards funded through NIH:\n \n \n \n The rules\ \ listed at\n \n 2 CFR Part 200\n \n , Uniform Administrative\ \ Requirements, Cost Principles, and Audit Requirements for Federal Awards.\n\ \ \n \n All NIH grant and cooperative agreement awards include the\n \ \ \n NIH Grants Policy Statement\n \n as part of the terms and\ \ conditions in the Notice of Award (NoA). The NoA includes the requirements of\ \ this NOFO. For these terms of award, see the\n \n NIH Grants Policy\ \ Statement Part II: Terms and Conditions of NIH Grant Awards, Subpart A: General\n\ \ \n and\n \n Part II: Terms and Conditions of NIH Grant Awards,\ \ Subpart B: Terms and Conditions for Specific Types of Grants, Recipients, and\ \ Activities\n \n .\n \n \n If a recipient receives an award,\ \ the recipient must follow all applicable nondiscrimination laws. The recipient\ \ agrees to this when registering in SAM.gov. The recipient must also submit an\ \ Assurance of Compliance (\n \n HHS-690\n \n ). To learn more,\ \ see the\n \n Laws and Regulations Enforced by the HHS Office for Civil\ \ Rights website\n \n .\n \n \n HHS recognizes that NIH research\ \ projects are often limited in scope for many reasons that are nondiscriminatory,\ \ such as the principal investigator’s scientific interest, funding limitations,\ \ recruitment requirements, and other considerations. Thus, criteria in research\ \ protocols that target or exclude certain populations are warranted where nondiscriminatory\ \ justifications establish that such criteria are appropriate with respect to\ \ the health or safety of the subjects, the scientific study design, or the purpose\ \ of the research. For additional guidance regarding how the provisions apply\ \ to NIH grant programs, please contact the Scientific/Research Contact that is\ \ identified in Section VII under Agency Contacts of this NOFO.\n \n \n\ \ \n \n \n All federal statutes and regulations relevant to federal financial\ \ assistance, including those highlighted in\n \n NIH Grants Policy Statement\ \ Section 4 Public Policy Requirements, Objectives and Other Appropriation Mandates.\n\ \ \n \n \n Recipients are responsible for ensuring that their activities\ \ comply with all applicable federal regulations. NIH may terminate awards under\ \ certain circumstances. See\n \n 2 CFR Part 200.340 Termination\n \ \ \n and\n \n NIH Grants Policy Statement Section 8.5.2 Remedies\ \ for Noncompliance or Enforcement Actions: Suspension, Termination, and Withholding\ \ of Support\n \n .\n \n \n Successful recipients under this NOFO\ \ agree that:\n \n \n Where the award funding involves implementing, acquiring,\ \ or upgrading health IT for activities by any funded entity, recipients and subrecipient(s)\ \ are required to: Use health IT that meets standards and implementation specifications\ \ adopted in 45 CFR part 170, Subpart B, if such standards and implementation\ \ specifications can support the activity. Visit\n \n https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-D/part-170/subpart-B\n\ \ \n to learn more.\n \n \n Where the award funding involves implementing,\ \ acquiring, or upgrading health IT for activities by eligible clinicians in ambulatory\ \ settings, or hospitals, eligible under Sections 4101, 4102, and 4201 of the\ \ HITECH Act, use health IT certified under the ONC Health IT Certification Program\ \ if certified technology can support the activity. Visit\n \n https://www.healthit.gov/topic/certification-ehrs/certification-health-it\n\ \ \n to learn more.\n \n \n Pursuant to the Cybersecurity Act of\ \ 2015, Div. N, § 405, Pub. Law 114-113, 6 USC § 1533(d), the HHS Secretary has\ \ established a common set of voluntary, consensus-based, and industry-led guidelines,\ \ best practices, methodologies, procedures, and processes.\n \n \n Successful\ \ recipients under this NOFO agree that:\n \n \n When recipients, subrecipients,\ \ or third-party entities have:\n \n \n 1) ongoing and consistent access\ \ to HHS owned or operated information or operational technology systems; and\n\ \ \n \n 2) receive, maintain, transmit, store, access, exchange, process,\ \ or utilize personal identifiable information (PII) or personal health information\ \ (PHI) obtained from the awarding HHS agency for the purposes of executing the\ \ award.\n \n \n Recipients shall develop plans and procedures, modeled\ \ after the\n \n NIST Cybersecurity framework\n \n , to protect\ \ HHS systems and data. Please refer to\n \n NIH Post-Award Monitoring\ \ and Reporting\n \n for additional information.\n \n \n \n Cooperative\ \ Agreement Terms and Conditions of Award\n \n Not Applicable\n \n \n\ \ \n 3. Data Management and Sharing\n \n Consistent with the 2023 NIH\ \ Policy for Data Management and Sharing, when data management and sharing is\ \ applicable to the award, recipients will be required to adhere to the Data Management\ \ and Sharing requirements as outlined in the\n \n NIH Grants Policy Statement\n\ \ \n . Upon the approval of a Data Management and Sharing Plan, it is required\ \ for recipients to implement the plan as described.\n \n \n \n \n 4. Reporting\n\ \ \n \n When multiple years are involved, recipients will be required to\ \ submit the\n \n Research Performance Progress Report (RPPR)\n \n\ \ annually and financial statements as required in the\n \n NIH Grants\ \ Policy Statement Section 8.4.1 Reporting.\n \n To learn more about post-award\ \ monitoring and reporting, see the NIH Grants & Funding website, see\n \n\ \ Post-Award Monitoring and Reporting\n \n .\n \n \n \n \n A\ \ final RPPR, invention statement, and the expenditure data portion of the Federal\ \ Financial Report are required for closeout of an award, as described in the\n\ \ \n NIH Grants Policy Statement Section 8.6 Closeout\n \n . NIH\ \ NOFOs outline intended research goals and objectives. Post award, NIH will review\ \ and measure performance based on the details and outcomes that are shared within\ \ the RPPR, as described at 2 CFR Part 200.301." - "PAR-25-378: Intervention Research to Improve Native American Health (R34 Clinical\ \ Trial Optional) Section II. Award Information\n \n \n \n Funding Instrument\n\ \ \n Grant: A financial assistance mechanism providing money, property,\ \ or both to an eligible entity to carry out an approved project or activity.\n\ \ \n \n \n Application Types Allowed\n New\n \n Resubmission\n \ \ \n \n \n \n The\n \n OER Glossary\n \n and the How to Apply\ \ Application Guide provide details on these application types. Only those application\ \ types listed here are allowed for this NOFO.\n \n \n \n Clinical Trial?\n\ \ \n Optional: Accepting applications that either propose or do not propose\ \ clinical trial(s).\n \n \n \n \n \n Need help determining whether you\ \ are doing a clinical trial?\n \n \n \n \n Funds Available and Anticipated\ \ Number of Awards\n \n The number of awards is contingent upon NIH appropriations\ \ and the submission of a sufficient number of meritorious applications.\n \ \ \n \n \n Award Budget\n \n Applicants may request direct costs of up\ \ to $450,000 for three years. Although variations from year to year are permissible,\ \ in no case may any year be more than $225,000 in direct costs, and total direct\ \ costs for the entire project period may not exceed $450,000.\n \n \n \n \ \ Award Project Period\n \n The scope of the proposed project should determine\ \ the project period. The maximum project period is 3 years.\n \n This variable\ \ defines that we need to start a new row. \n \n \n \n NIH grants policies\ \ as described in the\n \n NIH Grants Policy Statement\n \n will\ \ apply to the applications submitted and awards made from this NOFO." - "RFA-HL-26-011: Leadership and Coordination Center (LACC) for the MACS/WIHS Combined\ \ Cohort Study (MWCCS) (U01 Clinical Trials Not Allowed) Section III. Eligibility\ \ Information\n \n \n \n \n 1. Eligible Applicants\n \n \n \n Eligible\ \ Organizations\n \n Higher Education Institutions\n \n \n \n Public/State\ \ Controlled Institutions of Higher Education\n \n \n Private Institutions\ \ of Higher Education\n \n \n \n The following types of Higher Education\ \ Institutions are always encouraged to apply for NIH support as Public or Private\ \ Institutions of Higher Education:\n \n \n \n Hispanic-serving Institutions\n\ \ \n \n Historically Black Colleges and Universities (HBCUs)\n \n\ \ \n Tribally Controlled Colleges and Universities (TCCUs)\n \n \n \ \ Alaska Native and Native Hawaiian Serving Institutions\n \n \n Asian\ \ American Native American Pacific Islander Serving Institutions (AANAPISIs)\n\ \ \n \n \n Nonprofits Other Than Institutions of Higher Education\n \ \ \n \n \n Nonprofits with 501(c)(3) IRS Status (Other than Institutions of\ \ Higher Education)\n \n \n Nonprofits without 501(c)(3) IRS Status (Other\ \ than Institutions of Higher Education)\n \n \n \n For-Profit Organizations\n\ \ \n \n \n Small Businesses\n \n \n For-Profit Organizations (Other\ \ than Small Businesses)\n \n \n \n Local Governments\n \n \n \n \ \ State Governments\n \n \n County Governments\n \n \n City or\ \ Township Governments\n \n \n Special District Governments\n \n \n\ \ Indian/Native American Tribal Governments (Federally Recognized)\n \n\ \ \n Indian/Native American Tribal Governments (Other than Federally Recognized).\n\ \ \n \n \n \n \n Federal Governments\n \n \n \n Eligible Agencies\ \ of the Federal Government\n \n \n U.S. Territory or Possession\n \ \ \n \n \n \n \n Other\n \n \n \n Independent School Districts\n \ \ \n \n Public Housing Authorities/Indian Housing Authorities\n \n \n\ \ Native American Tribal Organizations (other than Federally recognized tribal\ \ governments)\n \n \n Faith-based or Community-based Organizations\n\ \ \n \n Regional Organizations\n \n \n \n \n Foreign Organizations\n\ \ \n \n \n Non-domestic (non-U.S.) Entities (Foreign Organizations)\n \ \ \n are not\n \n eligible to apply.\n \n \n \n \n Non-domestic\ \ (non-U.S.) components of U.S. Organizations\n \n are not\n \n \ \ eligible to apply.\n \n \n \n \n Foreign components, as\n \n \ \ defined in the NIH Grants Policy Statement\n \n ,\n \n are not\n\ \ \n allowed.\n \n \n \n Required Registrations\n \n \n Applicant\ \ Organizations\n \n \n \n Applicant organizations must complete and maintain\ \ the following registrations as described in the How to Apply- Application Guide\ \ to be eligible to apply for or receive an award. All registrations must be completed\ \ prior to the application being submitted. Registration can take 6 weeks or more,\ \ so applicants should begin the registration process as soon as possible. Failure\ \ to complete registrations in advance of a due date is not a valid reason for\ \ a late submission, please reference the\n \n NIH Grants Policy Statement\ \ Section 2.3.9.2 Electronically Submitted Applications\n \n for additional\ \ information.\n \n \n \n \n System for Award Management (SAM) –\n \ \ \n Applicants must complete and maintain an active registration,\n \ \ \n which requires renewal at least annually\n \n . The renewal\ \ process may require as much time as the initial registration. SAM registration\ \ includes the assignment of a Commercial and Government Entity (CAGE) Code for\ \ domestic organizations which have not already been assigned a CAGE Code.\n \ \ \n \n \n NATO Commercial and Government Entity (NCAGE) Code\n \ \ \n – Foreign organizations must obtain an NCAGE code (in lieu of a\ \ CAGE code) in order to register in SAM.\n \n \n Unique Entity Identifier\ \ (UEI) - A UEI is issued as part of the SAM.gov registration process. The same\ \ UEI must be used for all registrations, as well as on the grant application.\n\ \ \n \n \n \n \n eRA Commons\n \n - Once the unique organization\ \ identifier is established, organizations can register with eRA Commons in tandem\ \ with completing their Grants.gov registrations; all registrations must be in\ \ place by time of submission. eRA Commons requires organizations to identify\ \ at least one Signing Official (SO) and at least one Program Director/Principal\ \ Investigator (PD/PI) account in order to submit an application.\n \n \n\ \ \n Grants.gov\n \n – Applicants must have an active SAM registration\ \ in order to complete the Grants.gov registration.\n \n \n \n \n Program\ \ Directors/Principal Investigators (PD(s)/PI(s))\n \n \n \n All PD(s)/PI(s)\ \ must have an eRA Commons account. PD(s)/PI(s) should work with their organizational\ \ officials to either create a new account or to affiliate their existing account\ \ with the applicant organization in eRA Commons. If the PD/PI is also the organizational\ \ Signing Official, they must have two distinct eRA Commons accounts, one for\ \ each role. Obtaining an eRA Commons account can take up to 2 weeks.\n \n\ \ \n \n Eligible Individuals (Program Director/Principal Investigator)\n \ \ \n Any individual(s) with the skills, knowledge, and resources necessary\ \ to carry out the proposed research as the Program Director(s)/Principal Investigator(s)\ \ (PD(s)/PI(s)) is invited to work with their organization to develop an application\ \ for support.\n \n \n For institutions/organizations proposing multiple\ \ PDs/PIs, visit the Multiple Program Director/Principal Investigator Policy and\ \ submission details in the Senior/Key Person Profile (Expanded) Component of\ \ the How to Apply-Application Guide.\n \n \n \n \n A multiple PI (mPI)\ \ leadership structure is required. The LACC mPI team should include individuals\ \ with demonstrated deep knowledge of, and familiarity with, the MWCCS. In addition,\ \ the team should include at least one clinician with experience in clinical care\ \ of patients living with HIV and/or relevant comorbidities, as well as one or\ \ more individuals with epidemiologic cohort study leadership experience.Individuals\ \ may not apply to be an mPI of both the LACC and Data Analysis and Sharing Center\ \ (DASC) (\n \n RFA-HL-26-010\n \n ). Applications without an mPI\ \ leadership structure will be considered incomplete and will not proceed to peer\ \ review.\n \n \n \n \n 2. Cost Sharing\n \n \n This NOFO does not\ \ require cost sharing as defined in the\n \n NIH Grants Policy Statement\ \ Section 1.2 Definition of Terms\n \n .\n \n \n \n \n 3. Additional\ \ Information on Eligibility\n \n \n Number of Applications\n \n \n \n\ \ \n Applicant organizations may submit more than one application, provided\ \ that each application is scientifically distinct.\n \n \n The NIH will\ \ not accept duplicate or highly overlapping applications under review at the\ \ same time, per\n \n NIH Grants Policy Statement Section 2.3.7.4 Submission\ \ of Resubmission Application\n \n . This means that the NIH will not accept:\n\ \ \n \n \n A new (A0) application that is submitted before issuance of\ \ the summary statement from the review of an overlapping new (A0) or resubmission\ \ (A1) application.\n \n \n A resubmission (A1) application that is submitted\ \ before issuance of the summary statement from the review of the previous new\ \ (A0) application.\n \n \n An application that has substantial overlap\ \ with another application pending appeal of initial peer review (see\n \n\ \ NIH Grants Policy Statement 2.3.9.4 Similar, Essentially Identical, or\ \ Identical Applications\n \n )." - source_sentence: What is the purpose of the funding opportunity RFA-DK-26-007 titled "Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies"? sentences: - Department of Health and Human Services - "RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type\ \ 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Part 1. Overview\ \ Information\n \n \n \n Participating Organization(s)\n \n National\ \ Institutes of Health (\n \n NIH\n \n )\n \n \n \n Components\ \ of Participating Organizations\n \n National Institute of Diabetes and\ \ Digestive and Kidney Diseases (\n \n NIDDK\n \n )\n \n \n\ \ Office of The Director, National Institutes of Health (\n \n OD\n\ \ \n )\n \n \n \n Funding Opportunity Title\n \n Collaborative\ \ Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies\ \ (R01 - Clinical Trial Not Allowed)\n \n \n \n Activity Code\n \n \n\ \ R01\n \n Research Project Grant\n \n \n \n Announcement Type\n\ \ Reissue of\n \n RFA-DK-22-021\n \n \n \n \n Related Notices\n \ \ \n \n \n April 4, 2024\n \n - Overview of Grant Application\ \ and Review Changes for Due Dates on or after January 25, 2025. See Notice\n\ \ \n NOT-OD-24-084\n \n .\n \n \n \n August 31, 2022\n\ \ \n - Implementation Changes for Genomic Data Sharing Plans Included\ \ with Applications Due on or after January 25, 2023. See Notice\n \n \ \ NOT-OD-22-198\n \n .\n \n \n \n August 5, 2022\n \n\ \ - Implementation Details for the NIH Data Management and Sharing Policy.\ \ See Notice\n \n NOT-OD-22-189\n \n .\n \n \n \n \n \ \ Funding Opportunity Number (FON)\n \n RFA-DK-26-007\n \n \n \n Companion\ \ Funding Opportunity\n None\n \n \n Number of Applications\n \n \ \ See\n \n Section III. 3. Additional Information on Eligibility\n \ \ \n .\n \n \n \n Assistance Listing Number(s)\n 93.847\n \n \n\ \ Funding Opportunity Purpose\n \n This Notice of Funding Opportunity\ \ (NOFO) invites applications for studies of type 1 diabetes etiology and pathogenesis\ \ using data and samples from clinical trials and studies. This opportunity is\ \ intended to fund investigative teams collaborating to answer important questions\ \ about disease mechanisms leading to improved delay and durable prevention of\ \ type 1 diabetes. This NOFO is associated with the Special Diabetes Program (\n\ \ \n https://www.niddk.nih.gov/about-niddk/research-areas/diabetes/type-1-diabetes-special-statutory-funding-program/about-special-diabetes-program\n\ \ \n ) which funds research on the prevention, treatment, and cure of type\ \ 1 diabetes and its complications, including unique, innovative, and collaborative\ \ research consortia and clinical trials networks.\n \n \n \n \n \n \n \n \ \ Funding Opportunity Goal(s)\n \n To promote extramural basic and clinical\ \ biomedical research that improves the understanding of the mechanisms underlying\ \ disease and leads to improved preventions, diagnosis, and treatment of diabetes,\ \ digestive, and kidney diseases. Programmatic areas within the National Institute\ \ of Diabetes and Digestive and Kidney Diseases include diabetes, digestive, endocrine,\ \ hematologic, liver, metabolic, nephrologic, nutrition, obesity, and urologic\ \ diseases.\n \n This variable defines that we need to start a new row." - "RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type\ \ 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Part 1. Overview\ \ Information\n \n \n \n Participating Organization(s)\n \n National\ \ Institutes of Health (\n \n NIH\n \n )\n \n \n \n Components\ \ of Participating Organizations\n \n National Institute of Diabetes and\ \ Digestive and Kidney Diseases (\n \n NIDDK\n \n )\n \n \n\ \ Office of The Director, National Institutes of Health (\n \n OD\n\ \ \n )\n \n \n \n Funding Opportunity Title\n \n Collaborative\ \ Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies\ \ (R01 - Clinical Trial Not Allowed)\n \n \n \n Activity Code\n \n \n\ \ R01\n \n Research Project Grant\n \n \n \n Announcement Type\n\ \ Reissue of\n \n RFA-DK-22-021\n \n \n \n \n Related Notices\n \ \ \n \n \n April 4, 2024\n \n - Overview of Grant Application\ \ and Review Changes for Due Dates on or after January 25, 2025. See Notice\n\ \ \n NOT-OD-24-084\n \n .\n \n \n \n August 31, 2022\n\ \ \n - Implementation Changes for Genomic Data Sharing Plans Included\ \ with Applications Due on or after January 25, 2023. See Notice\n \n \ \ NOT-OD-22-198\n \n .\n \n \n \n August 5, 2022\n \n\ \ - Implementation Details for the NIH Data Management and Sharing Policy.\ \ See Notice\n \n NOT-OD-22-189\n \n .\n \n \n \n \n \ \ Funding Opportunity Number (FON)\n \n RFA-DK-26-007\n \n \n \n Companion\ \ Funding Opportunity\n None\n \n \n Number of Applications\n \n \ \ See\n \n Section III. 3. Additional Information on Eligibility\n \ \ \n .\n \n \n \n Assistance Listing Number(s)\n 93.847\n \n \n\ \ Funding Opportunity Purpose\n \n This Notice of Funding Opportunity\ \ (NOFO) invites applications for studies of type 1 diabetes etiology and pathogenesis\ \ using data and samples from clinical trials and studies. This opportunity is\ \ intended to fund investigative teams collaborating to answer important questions\ \ about disease mechanisms leading to improved delay and durable prevention of\ \ type 1 diabetes. This NOFO is associated with the Special Diabetes Program (\n\ \ \n https://www.niddk.nih.gov/about-niddk/research-areas/diabetes/type-1-diabetes-special-statutory-funding-program/about-special-diabetes-program\n\ \ \n ) which funds research on the prevention, treatment, and cure of type\ \ 1 diabetes and its complications, including unique, innovative, and collaborative\ \ research consortia and clinical trials networks.\n \n \n \n \n \n \n \n \ \ Funding Opportunity Goal(s)\n \n To promote extramural basic and clinical\ \ biomedical research that improves the understanding of the mechanisms underlying\ \ disease and leads to improved preventions, diagnosis, and treatment of diabetes,\ \ digestive, and kidney diseases. Programmatic areas within the National Institute\ \ of Diabetes and Digestive and Kidney Diseases include diabetes, digestive, endocrine,\ \ hematologic, liver, metabolic, nephrologic, nutrition, obesity, and urologic\ \ diseases.\n \n This variable defines that we need to start a new row." model-index: - name: SentenceTransformer based on Snowflake/snowflake-arctic-embed-l results: - task: type: information-retrieval name: Information Retrieval dataset: name: Unknown type: unknown metrics: - type: cosine_accuracy@1 value: 0.875 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 1.0 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 1.0 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 1.0 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.875 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.3333333333333333 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.20000000000000004 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.10000000000000002 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.875 name: Cosine Recall@1 - type: cosine_recall@3 value: 1.0 name: Cosine Recall@3 - type: cosine_recall@5 value: 1.0 name: Cosine Recall@5 - type: cosine_recall@10 value: 1.0 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.9484108127976215 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.9305555555555555 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.9305555555555555 name: Cosine Map@100 --- # SentenceTransformer based on Snowflake/snowflake-arctic-embed-l This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Snowflake/snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l). It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [Snowflake/snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l) <!-- at revision d8fb21ca8d905d2832ee8b96c894d3298964346b --> - Maximum Sequence Length: 512 tokens - Output Dimensionality: 1024 dimensions - Similarity Function: Cosine Similarity <!-- - Training Dataset: Unknown --> <!-- - Language: Unknown --> <!-- - License: Unknown --> ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("christinemahler/aie5-midter-new") # Run inference sentences = [ 'What is the purpose of the funding opportunity RFA-DK-26-007 titled "Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies"?', 'RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Part 1. Overview Information\n \n \n \n Participating Organization(s)\n \n National Institutes of Health (\n \n NIH\n \n )\n \n \n \n Components of Participating Organizations\n \n National Institute of Diabetes and Digestive and Kidney Diseases (\n \n NIDDK\n \n )\n \n \n Office of The Director, National Institutes of Health (\n \n OD\n \n )\n \n \n \n Funding Opportunity Title\n \n Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed)\n \n \n \n Activity Code\n \n \n R01\n \n Research Project Grant\n \n \n \n Announcement Type\n Reissue of\n \n RFA-DK-22-021\n \n \n \n \n Related Notices\n \n \n \n April 4, 2024\n \n - Overview of Grant Application and Review Changes for Due Dates on or after January 25, 2025. See Notice\n \n NOT-OD-24-084\n \n .\n \n \n \n August 31, 2022\n \n - Implementation Changes for Genomic Data Sharing Plans Included with Applications Due on or after January 25, 2023. See Notice\n \n NOT-OD-22-198\n \n .\n \n \n \n August 5, 2022\n \n - Implementation Details for the NIH Data Management and Sharing Policy. See Notice\n \n NOT-OD-22-189\n \n .\n \n \n \n \n Funding Opportunity Number (FON)\n \n RFA-DK-26-007\n \n \n \n Companion Funding Opportunity\n None\n \n \n Number of Applications\n \n See\n \n Section III. 3. Additional Information on Eligibility\n \n .\n \n \n \n Assistance Listing Number(s)\n 93.847\n \n \n Funding Opportunity Purpose\n \n This Notice of Funding Opportunity (NOFO) invites applications for studies of type 1 diabetes etiology and pathogenesis using data and samples from clinical trials and studies. This opportunity is intended to fund investigative teams collaborating to answer important questions about disease mechanisms leading to improved delay and durable prevention of type 1 diabetes. This NOFO is associated with the Special Diabetes Program (\n \n https://www.niddk.nih.gov/about-niddk/research-areas/diabetes/type-1-diabetes-special-statutory-funding-program/about-special-diabetes-program\n \n ) which funds research on the prevention, treatment, and cure of type 1 diabetes and its complications, including unique, innovative, and collaborative research consortia and clinical trials networks.\n \n \n \n \n \n \n \n Funding Opportunity Goal(s)\n \n To promote extramural basic and clinical biomedical research that improves the understanding of the mechanisms underlying disease and leads to improved preventions, diagnosis, and treatment of diabetes, digestive, and kidney diseases. Programmatic areas within the National Institute of Diabetes and Digestive and Kidney Diseases include diabetes, digestive, endocrine, hematologic, liver, metabolic, nephrologic, nutrition, obesity, and urologic diseases.\n \n This variable defines that we need to start a new row.', 'RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Part 1. Overview Information\n \n \n \n Participating Organization(s)\n \n National Institutes of Health (\n \n NIH\n \n )\n \n \n \n Components of Participating Organizations\n \n National Institute of Diabetes and Digestive and Kidney Diseases (\n \n NIDDK\n \n )\n \n \n Office of The Director, National Institutes of Health (\n \n OD\n \n )\n \n \n \n Funding Opportunity Title\n \n Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed)\n \n \n \n Activity Code\n \n \n R01\n \n Research Project Grant\n \n \n \n Announcement Type\n Reissue of\n \n RFA-DK-22-021\n \n \n \n \n Related Notices\n \n \n \n April 4, 2024\n \n - Overview of Grant Application and Review Changes for Due Dates on or after January 25, 2025. See Notice\n \n NOT-OD-24-084\n \n .\n \n \n \n August 31, 2022\n \n - Implementation Changes for Genomic Data Sharing Plans Included with Applications Due on or after January 25, 2023. See Notice\n \n NOT-OD-22-198\n \n .\n \n \n \n August 5, 2022\n \n - Implementation Details for the NIH Data Management and Sharing Policy. See Notice\n \n NOT-OD-22-189\n \n .\n \n \n \n \n Funding Opportunity Number (FON)\n \n RFA-DK-26-007\n \n \n \n Companion Funding Opportunity\n None\n \n \n Number of Applications\n \n See\n \n Section III. 3. Additional Information on Eligibility\n \n .\n \n \n \n Assistance Listing Number(s)\n 93.847\n \n \n Funding Opportunity Purpose\n \n This Notice of Funding Opportunity (NOFO) invites applications for studies of type 1 diabetes etiology and pathogenesis using data and samples from clinical trials and studies. This opportunity is intended to fund investigative teams collaborating to answer important questions about disease mechanisms leading to improved delay and durable prevention of type 1 diabetes. This NOFO is associated with the Special Diabetes Program (\n \n https://www.niddk.nih.gov/about-niddk/research-areas/diabetes/type-1-diabetes-special-statutory-funding-program/about-special-diabetes-program\n \n ) which funds research on the prevention, treatment, and cure of type 1 diabetes and its complications, including unique, innovative, and collaborative research consortia and clinical trials networks.\n \n \n \n \n \n \n \n Funding Opportunity Goal(s)\n \n To promote extramural basic and clinical biomedical research that improves the understanding of the mechanisms underlying disease and leads to improved preventions, diagnosis, and treatment of diabetes, digestive, and kidney diseases. Programmatic areas within the National Institute of Diabetes and Digestive and Kidney Diseases include diabetes, digestive, endocrine, hematologic, liver, metabolic, nephrologic, nutrition, obesity, and urologic diseases.\n \n This variable defines that we need to start a new row.', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 1024] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Information Retrieval * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.875 \| \| cosine_accuracy@3 \| 1.0 \| \| cosine_accuracy@5 \| 1.0 \| \| cosine_accuracy@10 \| 1.0 \| \| cosine_precision@1 \| 0.875 \| \| cosine_precision@3 \| 0.3333 \| \| cosine_precision@5 \| 0.2 \| \| cosine_precision@10 \| 0.1 \| \| cosine_recall@1 \| 0.875 \| \| cosine_recall@3 \| 1.0 \| \| cosine_recall@5 \| 1.0 \| \| cosine_recall@10 \| 1.0 \| \| cosine_ndcg@10 \| 0.9484 \| \| cosine_mrr@10 \| 0.9306 \| \| cosine_map@100 \| 0.9306 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 216 training samples * Columns: <code>sentence_0</code> and <code>sentence_1</code> * Approximate statistics based on the first 216 samples: \| \| sentence_0 \| sentence_1 \| \|:--------\|:----------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\| \| type \| string \| string \| \| details \| <ul><li>min: 19 tokens</li><li>mean: 34.9 tokens</li><li>max: 59 tokens</li></ul> \| <ul><li>min: 8 tokens</li><li>mean: 377.44 tokens</li><li>max: 512 tokens</li></ul> \| * Samples: \| sentence_0 \| sentence_1 \| \|:----------------------------------------------------------------------------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>What initiatives is the Department of Health and Human Services pursuing under opportunity ID [insert ID] to improve public health outcomes?</code> \| <code>Department of Health and Human Services</code> \| \| <code>How does the title of opportunity ID [insert ID] align with the strategic goals of the Department of Health and Human Services?</code> \| <code>Department of Health and Human Services</code> \| \| <code>What are the main goals of the funding opportunity titled "Laboratory Flexible Funding Model (LFFM)" under opportunity ID RFA-FD-25-007?</code> \| <code>RFA-FD-25-007: Laboratory Flexible Funding Model (LFFM) Part 1. Overview Information<br> <br> <br> <br> Participating Organization(s)<br> <br> U.S. Food and Drug Administration (<br> <br> FDA<br> <br> )<br> <br> <br> <br> <br> NOTE: The policies, guidelines, terms, and conditions stated in this Notice of Funding Opportunity (NOFO) may differ from those used by the NIH. Where this NOFO provides specific written guidance that may differ from the general guidance provided in the grant application form, please follow the instructions given in this NOFO.<br> <br> <br> The FDA does not follow the NIH Page Limitation Guidelines or the NIH Review Criteria. Applicants are encouraged to consult with FDA Agency Contacts for additional information regarding page limits and the FDA Objective Review Process.<br> <br> <br> <br> Components of Participating Organizations<br> <br> FOOD AND DRUG ADMINISTRATION (<br> <br> FDA<br> <br> )<br> <br> <br> <br> Funding Opportunity Title<br> <br> Laboratory Flexible Funding Model (LFFM)<br>...</code> \| * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 768, 512, 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 10 - `per_device_eval_batch_size`: 10 - `num_train_epochs`: 10 - `multi_dataset_batch_sampler`: round_robin #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: steps - `prediction_loss_only`: True - `per_device_train_batch_size`: 10 - `per_device_eval_batch_size`: 10 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 1 - `eval_accumulation_steps`: None - `torch_empty_cache_steps`: None - `learning_rate`: 5e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1 - `num_train_epochs`: 10 - `max_steps`: -1 - `lr_scheduler_type`: linear - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.0 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: False - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: None - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: False - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: None - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `include_for_metrics`: [] - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `use_liger_kernel`: False - `eval_use_gather_object`: False - `average_tokens_across_devices`: False - `prompts`: None - `batch_sampler`: batch_sampler - `multi_dataset_batch_sampler`: round_robin </details> ### Training Logs \| Epoch \| Step \| cosine_ndcg@10 \| \|:------:\|:----:\|:--------------:\| \| 1.0 \| 22 \| 0.8768 \| \| 2.0 \| 44 \| 0.9484 \| \| 2.2727 \| 50 \| 0.9330 \| \| 3.0 \| 66 \| 0.9276 \| \| 4.0 \| 88 \| 0.9484 \| \| 4.5455 \| 100 \| 0.9330 \| \| 5.0 \| 110 \| 0.9638 \| \| 6.0 \| 132 \| 0.9638 \| \| 6.8182 \| 150 \| 0.9638 \| \| 7.0 \| 154 \| 0.9638 \| \| 8.0 \| 176 \| 0.9484 \| \| 9.0 \| 198 \| 0.9484 \| \| 9.0909 \| 200 \| 0.9484 \| \| 10.0 \| 220 \| 0.9484 \| ### Framework Versions - Python: 3.11.11 - Sentence Transformers: 3.3.1 - Transformers: 4.48.3 - PyTorch: 2.5.1+cu124 - Accelerate: 1.3.0 - Datasets: 3.3.2 - Tokenizers: 0.21.0 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MatryoshkaLoss ```bibtex @misc{kusupati2024matryoshka, title={Matryoshka Representation Learning}, author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi}, year={2024}, eprint={2205.13147}, archivePrefix={arXiv}, primaryClass={cs.LG} } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	null	Non_BioNLP	# SentenceTransformer based on Snowflake/snowflake-arctic-embed-l This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Snowflake/snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l). It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [Snowflake/snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l) <!-- at revision d8fb21ca8d905d2832ee8b96c894d3298964346b --> - Maximum Sequence Length: 512 tokens - Output Dimensionality: 1024 dimensions - Similarity Function: Cosine Similarity <!-- - Training Dataset: Unknown --> <!-- - Language: Unknown --> <!-- - License: Unknown --> ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("christinemahler/aie5-midter-new") # Run inference sentences = [ 'What is the purpose of the funding opportunity RFA-DK-26-007 titled "Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies"?', 'RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Part 1. Overview Information\n \n \n \n Participating Organization(s)\n \n National Institutes of Health (\n \n NIH\n \n )\n \n \n \n Components of Participating Organizations\n \n National Institute of Diabetes and Digestive and Kidney Diseases (\n \n NIDDK\n \n )\n \n \n Office of The Director, National Institutes of Health (\n \n OD\n \n )\n \n \n \n Funding Opportunity Title\n \n Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed)\n \n \n \n Activity Code\n \n \n R01\n \n Research Project Grant\n \n \n \n Announcement Type\n Reissue of\n \n RFA-DK-22-021\n \n \n \n \n Related Notices\n \n \n \n April 4, 2024\n \n - Overview of Grant Application and Review Changes for Due Dates on or after January 25, 2025. See Notice\n \n NOT-OD-24-084\n \n .\n \n \n \n August 31, 2022\n \n - Implementation Changes for Genomic Data Sharing Plans Included with Applications Due on or after January 25, 2023. See Notice\n \n NOT-OD-22-198\n \n .\n \n \n \n August 5, 2022\n \n - Implementation Details for the NIH Data Management and Sharing Policy. See Notice\n \n NOT-OD-22-189\n \n .\n \n \n \n \n Funding Opportunity Number (FON)\n \n RFA-DK-26-007\n \n \n \n Companion Funding Opportunity\n None\n \n \n Number of Applications\n \n See\n \n Section III. 3. Additional Information on Eligibility\n \n .\n \n \n \n Assistance Listing Number(s)\n 93.847\n \n \n Funding Opportunity Purpose\n \n This Notice of Funding Opportunity (NOFO) invites applications for studies of type 1 diabetes etiology and pathogenesis using data and samples from clinical trials and studies. This opportunity is intended to fund investigative teams collaborating to answer important questions about disease mechanisms leading to improved delay and durable prevention of type 1 diabetes. This NOFO is associated with the Special Diabetes Program (\n \n https://www.niddk.nih.gov/about-niddk/research-areas/diabetes/type-1-diabetes-special-statutory-funding-program/about-special-diabetes-program\n \n ) which funds research on the prevention, treatment, and cure of type 1 diabetes and its complications, including unique, innovative, and collaborative research consortia and clinical trials networks.\n \n \n \n \n \n \n \n Funding Opportunity Goal(s)\n \n To promote extramural basic and clinical biomedical research that improves the understanding of the mechanisms underlying disease and leads to improved preventions, diagnosis, and treatment of diabetes, digestive, and kidney diseases. Programmatic areas within the National Institute of Diabetes and Digestive and Kidney Diseases include diabetes, digestive, endocrine, hematologic, liver, metabolic, nephrologic, nutrition, obesity, and urologic diseases.\n \n This variable defines that we need to start a new row.', 'RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Part 1. Overview Information\n \n \n \n Participating Organization(s)\n \n National Institutes of Health (\n \n NIH\n \n )\n \n \n \n Components of Participating Organizations\n \n National Institute of Diabetes and Digestive and Kidney Diseases (\n \n NIDDK\n \n )\n \n \n Office of The Director, National Institutes of Health (\n \n OD\n \n )\n \n \n \n Funding Opportunity Title\n \n Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed)\n \n \n \n Activity Code\n \n \n R01\n \n Research Project Grant\n \n \n \n Announcement Type\n Reissue of\n \n RFA-DK-22-021\n \n \n \n \n Related Notices\n \n \n \n April 4, 2024\n \n - Overview of Grant Application and Review Changes for Due Dates on or after January 25, 2025. See Notice\n \n NOT-OD-24-084\n \n .\n \n \n \n August 31, 2022\n \n - Implementation Changes for Genomic Data Sharing Plans Included with Applications Due on or after January 25, 2023. See Notice\n \n NOT-OD-22-198\n \n .\n \n \n \n August 5, 2022\n \n - Implementation Details for the NIH Data Management and Sharing Policy. See Notice\n \n NOT-OD-22-189\n \n .\n \n \n \n \n Funding Opportunity Number (FON)\n \n RFA-DK-26-007\n \n \n \n Companion Funding Opportunity\n None\n \n \n Number of Applications\n \n See\n \n Section III. 3. Additional Information on Eligibility\n \n .\n \n \n \n Assistance Listing Number(s)\n 93.847\n \n \n Funding Opportunity Purpose\n \n This Notice of Funding Opportunity (NOFO) invites applications for studies of type 1 diabetes etiology and pathogenesis using data and samples from clinical trials and studies. This opportunity is intended to fund investigative teams collaborating to answer important questions about disease mechanisms leading to improved delay and durable prevention of type 1 diabetes. This NOFO is associated with the Special Diabetes Program (\n \n https://www.niddk.nih.gov/about-niddk/research-areas/diabetes/type-1-diabetes-special-statutory-funding-program/about-special-diabetes-program\n \n ) which funds research on the prevention, treatment, and cure of type 1 diabetes and its complications, including unique, innovative, and collaborative research consortia and clinical trials networks.\n \n \n \n \n \n \n \n Funding Opportunity Goal(s)\n \n To promote extramural basic and clinical biomedical research that improves the understanding of the mechanisms underlying disease and leads to improved preventions, diagnosis, and treatment of diabetes, digestive, and kidney diseases. Programmatic areas within the National Institute of Diabetes and Digestive and Kidney Diseases include diabetes, digestive, endocrine, hematologic, liver, metabolic, nephrologic, nutrition, obesity, and urologic diseases.\n \n This variable defines that we need to start a new row.', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 1024] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Information Retrieval * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.875 \| \| cosine_accuracy@3 \| 1.0 \| \| cosine_accuracy@5 \| 1.0 \| \| cosine_accuracy@10 \| 1.0 \| \| cosine_precision@1 \| 0.875 \| \| cosine_precision@3 \| 0.3333 \| \| cosine_precision@5 \| 0.2 \| \| cosine_precision@10 \| 0.1 \| \| cosine_recall@1 \| 0.875 \| \| cosine_recall@3 \| 1.0 \| \| cosine_recall@5 \| 1.0 \| \| cosine_recall@10 \| 1.0 \| \| cosine_ndcg@10 \| 0.9484 \| \| cosine_mrr@10 \| 0.9306 \| \| cosine_map@100 \| 0.9306 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 216 training samples * Columns: <code>sentence_0</code> and <code>sentence_1</code> * Approximate statistics based on the first 216 samples: \| \| sentence_0 \| sentence_1 \| \|:--------\|:----------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\| \| type \| string \| string \| \| details \| <ul><li>min: 19 tokens</li><li>mean: 34.9 tokens</li><li>max: 59 tokens</li></ul> \| <ul><li>min: 8 tokens</li><li>mean: 377.44 tokens</li><li>max: 512 tokens</li></ul> \| * Samples: \| sentence_0 \| sentence_1 \| \|:----------------------------------------------------------------------------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>What initiatives is the Department of Health and Human Services pursuing under opportunity ID [insert ID] to improve public health outcomes?</code> \| <code>Department of Health and Human Services</code> \| \| <code>How does the title of opportunity ID [insert ID] align with the strategic goals of the Department of Health and Human Services?</code> \| <code>Department of Health and Human Services</code> \| \| <code>What are the main goals of the funding opportunity titled "Laboratory Flexible Funding Model (LFFM)" under opportunity ID RFA-FD-25-007?</code> \| <code>RFA-FD-25-007: Laboratory Flexible Funding Model (LFFM) Part 1. Overview Information<br> <br> <br> <br> Participating Organization(s)<br> <br> U.S. Food and Drug Administration (<br> <br> FDA<br> <br> )<br> <br> <br> <br> <br> NOTE: The policies, guidelines, terms, and conditions stated in this Notice of Funding Opportunity (NOFO) may differ from those used by the NIH. Where this NOFO provides specific written guidance that may differ from the general guidance provided in the grant application form, please follow the instructions given in this NOFO.<br> <br> <br> The FDA does not follow the NIH Page Limitation Guidelines or the NIH Review Criteria. Applicants are encouraged to consult with FDA Agency Contacts for additional information regarding page limits and the FDA Objective Review Process.<br> <br> <br> <br> Components of Participating Organizations<br> <br> FOOD AND DRUG ADMINISTRATION (<br> <br> FDA<br> <br> )<br> <br> <br> <br> Funding Opportunity Title<br> <br> Laboratory Flexible Funding Model (LFFM)<br>...</code> \| * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 768, 512, 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 10 - `per_device_eval_batch_size`: 10 - `num_train_epochs`: 10 - `multi_dataset_batch_sampler`: round_robin #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: steps - `prediction_loss_only`: True - `per_device_train_batch_size`: 10 - `per_device_eval_batch_size`: 10 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 1 - `eval_accumulation_steps`: None - `torch_empty_cache_steps`: None - `learning_rate`: 5e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1 - `num_train_epochs`: 10 - `max_steps`: -1 - `lr_scheduler_type`: linear - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.0 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: False - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: None - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: False - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: None - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `include_for_metrics`: [] - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `use_liger_kernel`: False - `eval_use_gather_object`: False - `average_tokens_across_devices`: False - `prompts`: None - `batch_sampler`: batch_sampler - `multi_dataset_batch_sampler`: round_robin </details> ### Training Logs \| Epoch \| Step \| cosine_ndcg@10 \| \|:------:\|:----:\|:--------------:\| \| 1.0 \| 22 \| 0.8768 \| \| 2.0 \| 44 \| 0.9484 \| \| 2.2727 \| 50 \| 0.9330 \| \| 3.0 \| 66 \| 0.9276 \| \| 4.0 \| 88 \| 0.9484 \| \| 4.5455 \| 100 \| 0.9330 \| \| 5.0 \| 110 \| 0.9638 \| \| 6.0 \| 132 \| 0.9638 \| \| 6.8182 \| 150 \| 0.9638 \| \| 7.0 \| 154 \| 0.9638 \| \| 8.0 \| 176 \| 0.9484 \| \| 9.0 \| 198 \| 0.9484 \| \| 9.0909 \| 200 \| 0.9484 \| \| 10.0 \| 220 \| 0.9484 \| ### Framework Versions - Python: 3.11.11 - Sentence Transformers: 3.3.1 - Transformers: 4.48.3 - PyTorch: 2.5.1+cu124 - Accelerate: 1.3.0 - Datasets: 3.3.2 - Tokenizers: 0.21.0 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MatryoshkaLoss ```bibtex @misc{kusupati2024matryoshka, title={Matryoshka Representation Learning}, author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi}, year={2024}, eprint={2205.13147}, archivePrefix={arXiv}, primaryClass={cs.LG} } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	{"base_model": "Snowflake/snowflake-arctic-embed-l", "library_name": "sentence-transformers", "metrics": ["cosine_accuracy@1", "cosine_accuracy@3", "cosine_accuracy@5", "cosine_accuracy@10", "cosine_precision@1", "cosine_precision@3", "cosine_precision@5", "cosine_precision@10", "cosine_recall@1", "cosine_recall@3", "cosine_recall@5", "cosine_recall@10", "cosine_ndcg@10", "cosine_mrr@10", "cosine_map@100"], "pipeline_tag": "sentence-similarity", "tags": ["sentence-transformers", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:216", "loss:MatryoshkaLoss", "loss:MultipleNegativesRankingLoss"], "widget": [{"source_sentence": "What types of organizations are eligible to apply for the PAR-25-353 opportunity titled \"National Cooperative Drug/Device Discovery/Development Groups (NCDDG) for the Treatment of Mental Disorders\"?", "sentences": ["PAR-25-353: National Cooperative Drug/Device Discovery/Development Groups (NCDDG) for the Treatment of Mental Disorders (U19 Clinical Trial Optional) Section III. Eligibility Information\n \n \n \n \n 1. Eligible Applicants\n \n \n \n Eligible Organizations\n \n Higher Education Institutions\n \n \n \n Public/State Controlled Institutions of Higher Education\n \n \n Private Institutions of Higher Education\n \n \n \n The following types of Higher Education Institutions are always encouraged to apply for NIH support as Public or Private Institutions of Higher Education:\n \n \n \n Hispanic-serving Institutions\n \n \n Historically Black Colleges and Universities (HBCUs)\n \n \n Tribally Controlled Colleges and Universities (TCCUs)\n \n \n Alaska Native and Native Hawaiian Serving Institutions\n \n \n Asian American Native American Pacific Islander Serving Institutions (AANAPISIs)\n \n \n \n Nonprofits Other Than Institutions of Higher Education\n \n \n \n Nonprofits with 501(c)(3) IRS Status (Other than Institutions of Higher Education)\n \n \n Nonprofits without 501(c)(3) IRS Status (Other than Institutions of Higher Education)\n \n \n \n For-Profit Organizations\n \n \n \n Small Businesses\n \n \n For-Profit Organizations (Other than Small Businesses)\n \n \n \n Local Governments\n \n \n \n State Governments\n \n \n County Governments\n \n \n City or Township Governments\n \n \n Special District Governments\n \n \n Indian/Native American Tribal Governments (Federally Recognized)\n \n \n Indian/Native American Tribal Governments (Other than Federally Recognized)\n \n \n \n \n \n Federal Governments\n \n \n \n Eligible Agencies of the Federal Government\n \n \n U.S. Territory or Possession\n \n \n \n \n \n Other\n \n \n \n Independent School Districts\n \n \n Public Housing Authorities/Indian Housing Authorities\n \n \n Native American Tribal Organizations (other than Federally recognized tribal governments)\n \n \n Faith-based or Community-based Organizations\n \n \n Regional Organizations\n \n \n \n \n Foreign Organizations\n \n \n \n Non-domestic (non-U.S.) Entities (Foreign Organization)\n \n are not\n \n eligible to apply.\n \n \n \n \n Non-domestic (non-U.S.) components of U.S. Organizations\n \n are not\n \n eligible to apply.\n \n \n \n \n Foreign components, as\n \n defined in the NIH Grants Policy Statement\n \n ,\n \n are\n \n allowed.\n \n \n \n Required Registrations\n \n \n Applicant organizations\n \n \n \n Applicant organizations must complete and maintain the following registrations as described in the\n \n How to Apply- Application Guide\n \n to be eligible to apply for or receive an award. All registrations must be completed prior to the application being submitted. Registration can take 6 weeks or more, so applicants should begin the registration process as soon as possible. Failure to complete registrations in advance of a due date is not a valid reason for a late submission, please reference\n \n NIH Grants Policy Statement Section 2.3.9.2 Electronically Submitted Applications\n \n for additional information.\n \n \n \n \n System for Award Management (SAM)\n \n – Applicants must complete and maintain an active registration,\n \n which requires renewal at least annually\n \n . The renewal process may require as much time as the initial registration. SAM registration includes the assignment of a Commercial and Government Entity (CAGE) Code for domestic organizations which have not already been assigned a CAGE Code.\n \n \n \n NATO Commercial and Government Entity (NCAGE) Code\n \n – Foreign organizations must obtain an NCAGE code (in lieu of a CAGE code) in order to register in SAM.\n \n \n Unique Entity Identifier (UEI) - A UEI is issued as part of the SAM.gov registration process. The same UEI must be used for all registrations, as well as on the grant application.\n \n \n \n \n \n eRA Commons\n \n - Once the unique organization identifier is established, organizations can register with eRA Commons in tandem with completing their Grants.gov registration; all registrations must be in place by time of submission. eRA Commons requires organizations to identify at least one Signing Official (SO) and at least one Program Director/Principal Investigator (PD/PI) account in order to submit an application.\n \n \n \n Grants.gov\n \n – Applicants must have an active SAM registration in order to complete the Grants.gov registration.\n \n \n \n \n Program Directors/Principal Investigators (PD(s)/PI(s))\n \n \n \n All PD(s)/PI(s) must have an eRA Commons account. PD(s)/PI(s) should work with their organizational officials to either create a new account or to affiliate their existing account with the applicant organization in eRA Commons. If the PD/PI is also the organizational Signing Official, they must have two distinct eRA Commons accounts, one for each role. Obtaining an eRA Commons account can take up to 2 weeks.\n \n \n \n Eligible Individuals (Program Director/Principal Investigator)\n \n Any individual(s) with the skills, knowledge, and resources necessary to carry out the proposed research as the Program Director(s)/Principal Investigator(s) (PD(s)/PI(s)) is invited to work with his/her organization to develop an application for support.\n \n \n For institutions/organizations proposing multiple PDs/PIs, visit the Multiple Program Director/Principal Investigator Policy and submission details in the Senior/Key Person Profile (Expanded) Component of the\n \n How to Apply - Application Guide\n \n .\n \n \n \n \n An NIH intramural scientist may not serve as the PD/PI of an NCDDG but may participate as a research collaborator or consultant (see\n \n \n \n Section IV.7\n \n \n \n for more information).\n \n \n \n \n 2. Cost Sharing\n \n \n This NOFO does not require cost sharing as defined in the\n \n NIH Grants Policy Statement\n \n \n Section 1.2- Definitions of Terms.\n \n \n \n \n \n 3. Additional Information on Eligibility\n \n \n Number of Applications\n \n \n \n \n Applicant organizations may submit more than one application, provided that each application is scientifically distinct.\n \n \n The NIH will not accept duplicate or highly overlapping applications under review at the same time per\n \n NIH Grants Policy Statement Section 2.3.7.4 Submission of Resubmission Application\n \n . This means that the NIH will not accept:\n \n \n \n A new (A0) application that is submitted before issuance of the summary statement from the review of an overlapping new (A0) or resubmission (A1) application.\n \n \n A resubmission (A1) application that is submitted before issuance of the summary statement from the review of the previous new (A0) application.\n \n \n An application that has substantial overlap with another application pending appeal of initial peer review (see\n \n NIH Grants Policy Statement 2.3.9.4 Similar, Essentially Identical, or Identical Applications\n \n ).", "RFA-DK-26-009: New Investigator Gateway Awards for Collaborative T1D Research (R03 Clinical Trial Not Allowed) Section VI. Award Administration Information\n \n \n \n \n 1. Award Notices\n \n \n A Notice of Award (NoA) is the official authorizing document notifying the applicant that an award has been made and that funds may be requested from the designated HHS payment system or office. The NoA is signed by the Grants Management Officer and emailed to the recipient’s business official.\n \n \n In accepting the award, the recipient agrees that any activities under the award are subject to all provisions currently in effect or implemented during the period of the award, other Department regulations and policies in effect at the time of the award, and applicable statutory provisions.\n \n \n Recipients must comply with any funding restrictions described in\n \n Section IV.6. Funding Restrictions\n \n . Any pre-award costs incurred before receipt of the NoA are at the applicant's own risk. For more information on the Notice of Award, please refer to the\n \n NIH Grants Policy Statement Section 5. The Notice of Award\n \n and NIH Grants & Funding website, see\n \n Award Process.\n \n \n \n \n \n Institutional Review Board or Independent Ethics Committee Approval: Recipient institutions must ensure that protocols are reviewed by their IRB or IEC. To help ensure the safety of participants enrolled in NIH-funded studies, the recipient must provide NIH copies of documents related to all major changes in the status of ongoing protocols.\n \n \n \n \n 2. Administrative and National Policy Requirements\n \n \n The following Federal wide and HHS-specific policy requirements apply to awards funded through NIH:\n \n \n \n The rules listed at\n \n 2 CFR Part 200\n \n , Uniform Administrative Requirements, Cost Principles, and Audit Requirements for Federal Awards.\n \n \n All NIH grant and cooperative agreement awards include the\n \n NIH Grants Policy Statement\n \n as part of the terms and conditions in the Notice of Award (NoA). The NoA includes the requirements of this NOFO. For these terms of award, see the\n \n NIH Grants Policy Statement Part II: Terms and Conditions of NIH Grant Awards, Subpart A: General\n \n and\n \n Part II: Terms and Conditions of NIH Grant Awards, Subpart B: Terms and Conditions for Specific Types of Grants, Recipients, and Activities\n \n .\n \n \n If a recipient receives an award, the recipient must follow all applicable nondiscrimination laws. The recipient agrees to this when registering in SAM.gov. The recipient must also submit an Assurance of Compliance (\n \n HHS-690\n \n ). To learn more, see the\n \n Laws and Regulations Enforced by the HHS Office for Civil Rights website\n \n .\n \n \n HHS recognizes that NIH research projects are often limited in scope for many reasons that are nondiscriminatory, such as the principal investigator’s scientific interest, funding limitations, recruitment requirements, and other considerations. Thus, criteria in research protocols that target or exclude certain populations are warranted where nondiscriminatory justifications establish that such criteria are appropriate with respect to the health or safety of the subjects, the scientific study design, or the purpose of the research. For additional guidance regarding how the provisions apply to NIH grant programs, please contact the Scientific/Research Contact that is identified in Section VII under Agency Contacts of this NOFO.\n \n \n \n \n \n All federal statutes and regulations relevant to federal financial assistance, including those highlighted in\n \n NIH Grants Policy Statement Section 4 Public Policy Requirements, Objectives and Other Appropriation Mandates.\n \n \n \n Recipients are responsible for ensuring that their activities comply with all applicable federal regulations. NIH may terminate awards under certain circumstances. See\n \n 2 CFR Part 200.340 Termination\n \n and\n \n NIH Grants Policy Statement Section 8.5.2 Remedies for Noncompliance or Enforcement Actions: Suspension, Termination, and Withholding of Support\n \n .\n \n \n Successful recipients under this NOFO agree that:\n \n \n Where the award funding involves implementing, acquiring, or upgrading health IT for activities by any funded entity, recipients and subrecipient(s) are required to: Use health IT that meets standards and implementation specifications adopted in 45 CFR part 170, Subpart B, if such standards and implementation specifications can support the activity. Visit\n \n https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-D/part-170/subpart-B\n \n to learn more.\n \n \n Where the award funding involves implementing, acquiring, or upgrading health IT for activities by eligible clinicians in ambulatory settings, or hospitals, eligible under Sections 4101, 4102, and 4201 of the HITECH Act, use health IT certified under the ONC Health IT Certification Program if certified technology can support the activity. Visit\n \n https://www.healthit.gov/topic/certification-ehrs/certification-health-it\n \n to learn more.\n \n \n Pursuant to the Cybersecurity Act of 2015, Div. N, § 405, Pub. Law 114-113, 6 USC § 1533(d), the HHS Secretary has established a common set of voluntary, consensus-based, and industry-led guidelines, best practices, methodologies, procedures, and processes.\n \n \n Successful recipients under this NOFO agree that:\n \n \n When recipients, subrecipients, or third-party entities have:\n \n \n \n ongoing and consistent access to HHS owned or operated information or operational technology systems; and\n \n \n receive, maintain, transmit, store, access, exchange, process, or utilize personal identifiable information (PII) or personal health information (PHI) obtained from the awarding HHS agency for the purposes of executing the award.\n \n \n \n Recipients shall develop plans and procedures, modeled after the\n \n NIST Cybersecurity framework\n \n , to protect HHS systems and data. Please refer to\n \n NIH Post-Award Monitoring and Reporting\n \n for additional information.\n \n \n \n Cooperative Agreement Terms and Conditions of Award\n \n Not Applicable.\n \n \n \n 3. Data Management and Sharing\n \n Consistent with the 2023 NIH Policy for Data Management and Sharing, when data management and sharing is applicable to the award, recipients will be required to adhere to the Data Management and Sharing requirements as outlined in the\n \n NIH Grants Policy Statement\n \n . Upon the approval of a Data Management and Sharing Plan, it is required for recipients to implement the plan as described.\n \n \n \n \n 4. Reporting\n \n \n When multiple years are involved, recipients will be required to submit the\n \n Research Performance Progress Report (RPPR)\n \n annually and financial statements as required in the\n \n NIH Grants Policy Statement Section 8.4.1 Reporting.\n \n To learn more about post-award monitoring and reporting, see the NIH Grants & Funding website, see\n \n Post-Award Monitoring and Reporting\n \n .\n \n \n \n \n A final RPPR, invention statement, and the expenditure data portion of the Federal Financial Report are required for closeout of an award, as described in the\n \n NIH Grants Policy Statement Section 8.6 Closeout\n \n . NIH NOFOs outline intended research goals and objectives. Post award, NIH will review and measure performance based on the details and outcomes that are shared within the RPPR, as described at 2 CFR Part 200.301.", "RFA-HL-26-011: Leadership and Coordination Center (LACC) for the MACS/WIHS Combined Cohort Study (MWCCS) (U01 Clinical Trials Not Allowed) Section II. Award Information\n \n \n \n Funding Instrument\n \n Cooperative Agreement: A financial assistance mechanism used when there will be substantial Federal scientific or programmatic involvement. Substantial involvement means that, after award, NIH scientific or program staff will assist, guide, coordinate, or participate in project activities. See Section VI.2 for additional information about the substantial involvement for this NOFO.\n \n \n \n Application Types Allowed\n New\n \n \n \n \n The\n \n OER Glossary\n \n and the How to Apply Application Guide provide details on these application types. Only those application types listed here are allowed for this NOFO.\n \n \n \n Clinical Trial?\n \n Not Allowed: Only accepting applications that do not propose clinical trials.\n \n \n \n \n \n Need help determining whether you are doing a clinical trial?\n \n \n \n \n Funds Available and Anticipated Number of Awards\n \n The following NIH components intend to commit an estimated total of $3 million to fund a single award in FY 2026:\n \n \n NHLBI, $1.37 million\n \n \n NIAID, $0.30 million\n \n \n NIMH, $0.30 million\n \n \n NCI, $0.35 million\n \n \n NIA, $0.19 million\n \n \n NIDA, $0.15 million\n \n \n NICHD, $0.13 million\n \n \n NIDCR, $0.07 million\n \n \n NIMHD, $0.04 million\n \n \n NIAAA, $0.03 million\n \n \n NINR, $0.03 million\n \n \n ORWH, $0.03 million\n \n \n NIDCD, $0.02 million\n \n \n \n Award Budget\n \n Application budgets may not exceed direct costs of $2 million per year in Fiscal Years 2026 through 2031.\n \n \n \n Award Project Period\n \n 6 years\n \n This variable defines that we need to start a new row. \n \n \n \n NIH grants policies as described in the\n \n NIH Grants Policy Statement\n \n will apply to the applications submitted and awards made from this NOFO."]}, {"source_sentence": "How does the title of opportunity ID [insert ID] align with the strategic goals of the Department of Health and Human Services?", "sentences": ["Department of Health and Human Services", "RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Section II. Award Information\n \n \n \n Funding Instrument\n \n Grant: A financial assistance mechanism providing money, property, or both to an eligible entity to carry out an approved project or activity.\n \n \n \n Application Types Allowed\n New\n \n Resubmission - Resubmission applications from\n \n RFA-DK-22-021\n \n and from the current active NOFO RFA-DK-26-007 may be submitted in response to this NOFO.\n \n \n \n \n The\n \n OER Glossary\n \n and the How to Apply Application Guide provide details on these application types. Only those application types listed here are allowed for this NOFO.\n \n \n \n Clinical Trial?\n \n Not Allowed: Only accepting applications that do not propose clinical trials.\n \n \n \n \n \n Need help determining whether you are doing a clinical trial?\n \n \n \n \n Funds Available and Anticipated Number of Awards\n \n NIDDK intends to commit $5.5 million in Fiscal Year 2026 to fund 3-4 awards and $5.5 million in FY 27 to fund 3-4 awards. The number of awards is contingent upon the submission of a sufficient number of meritorious applications.\n \n \n \n Award Budget\n \n Application budgets are limited to no more than $1,200,000 direct costs per year, exclusive of facilities and administrative (F&A) costs. Budgets are expected to reflect the actual needs of the proposed project.\n \n \n \n Award Project Period\n \n The maximum project period is 3 years.\n \n This variable defines that we need to start a new row. \n \n \n \n NIH grants policies as described in the\n \n NIH Grants Policy Statement\n \n will apply to the applications submitted and awards made from this NOFO.", "RFA-DK-26-009: New Investigator Gateway Awards for Collaborative T1D Research (R03 Clinical Trial Not Allowed) Section IV. Application and Submission Information\n \n \n \n \n 1. Requesting an Application Package\n \n \n The application forms package specific to this opportunity must be accessed through ASSIST, Grants.gov Workspace or an institutional system-to-system solution. Links to apply using ASSIST or Grants.gov Workspace are available in Part 1 of this NOFO. See your administrative office for instructions if you plan to use an institutional system-to-system solution.\n \n \n \n \n 2. Content and Form of Application Submission\n \n \n It is critical that applicants follow the instructions in the Research (R) Instructions in the\n \n How to Apply - Application Guide\n \n except where instructed in this notice of funding opportunity to do otherwise. Conformance to the requirements in the Application Guide is required and strictly enforced. Applications that are out of compliance with these instructions may be delayed or not accepted for review.\n \n \n \n \n Letter of Intent\n \n \n Although a letter of intent is not required, is not binding, and does not enter into the review of a subsequent application, the information that it contains allows IC staff to estimate the potential review workload and plan the review.\n \n \n By the date listed in\n \n Part 1. Overview Information\n \n , prospective applicants are asked to submit a letter of intent that includes the following information:\n \n \n \n Descriptive title of proposed activity\n \n \n Name(s), address(es), and telephone number(s) of the PD(s)/PI(s)\n \n \n Names of other key personnel\n \n \n Participating institution(s)\n \n \n Number and title of this funding opportunity\n \n \n \n The letter of intent should be sent to:\n \n \n John Connaughton, Ph.D.\n \n Chief, Scientific Review Branch\n \n National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)\n \n Telephone: 301-594-7797\n \n Email:\n \n \n [email protected]\n \n \n \n \n \n \n Page Limitations\n \n \n All page limitations described in the\n \n How to Apply- Application Guide\n \n and the\n \n Table of Page Limits\n \n must be followed.\n \n \n \n Instructions for Application Submission\n \n The following section supplements the instructions found in the\n \n How to Apply- Application Guide\n \n and should be used for preparing an application to this NOFO.\n \n \n \n \n SF424(R&R) Cover\n \n \n All instructions in the\n \n \n How to Apply - Application Guide\n \n \n must be followed.\n \n \n \n \n SF424(R&R) Project/Performance Site Locations\n \n \n All instructions in the How to Apply- Application Guide must be followed.\n \n \n \n \n SF424(R&R) Other Project Information\n \n \n All instructions in the How to Apply- Application Guide must be followed.\n \n \n \n \n \n Other Attachments:\n \n \n \n \n Proposed Affiliation Letter\n \n : Name the PDF formatted letter \"Proposed_Affiliation_Ltr.pdf\". Applications must include a one page letter that includes a brief statement identifying the specific NIDDK consortium or network that is within scope of the proposed aims. In addition, the applicant must outline in the letter how the objectives and design of the application are related to, but distinct from, ongoing studies in the consortium or network of interest, and describe how developing a relationship with the chosen consortium or network would be expected to facilitate the applicant's research and career goals. Applications that lack the proposed Affiliation Letter are considered incomplete and will not be peer reviewed.\n \n \n \n \n SF424(R&R) Senior/Key Person Profile\n \n \n All instructions in the How to Apply- Application Guide must be followed.\n \n \n \n \n R&R or Modular Budget\n \n \n All instructions in the How to Apply- Application Guide must be followed.\n \n \n \n \n Applicants should also follow these additional instructions:\n \n \n Applicants should budget for travel to one meeting of the proposed affiliated consortium or network per year. Except in unusual circumstances, only the PD/PI may be supported by R03 funds to travel to consortium meetings. There is no specific line item in which to list travel costs ($2000/year) in the modular budget format; instead, the proposed travel should be described in the budget justification section.\n \n \n \n \n R&R Subaward Budget\n \n \n All instructions in the How to Apply-Application Guide must be followed.\n \n \n \n \n PHS 398 Cover Page Supplement\n \n \n All instructions in the How to Apply- Application Guide must be followed.\n \n \n \n \n PHS 398 Research Plan\n \n \n All instructions in the How to Apply- Application Guide must be followed, with the following additional instructions:\n \n \n \n \n \n Resource Sharing Plan\n \n : Individuals are required to comply with the instructions for the Resource Sharing Plans as provided in the How to Apply- Application Guide.\n \n \n \n \n \n Protocol, tool and reagent sharing\n \n : The applicant should discuss plans for sharing and distribution of non-data resources that will be generated by the proposed project, including models, protocols, computational tools, biomaterials, and reagents. As one of the essential goals of the Gateway program is to support high impact advances, NIDDK intends that tools and reagents generated by the Gateway program will be made broadly available and distributed at minimal cost, and without undue intellectual property constraints, so that they can be as widely used as possible, thus enabling rapid downstream applications for the reagents by the larger scientific community. Applicants are expected to register resources supported by this NOFO with the NIDDK Information Network (dkNET) at\n \n https://dknet.org/\n \n and use Research Resource Identifiers (RRID) assigned through dkNET in any publication supported by this NOFO. Applicants should also be familiar with the NIH statements regarding intellectual property of resources developed with Federal funds\n \n (\n \n NIH Research Tools Policy (\n \n \n https://grants.nih.gov/grants/intell-property_64FR72090.pdf\n \n \n ) and other related NIH sharing policies at\n \n \n http://sharing.\n \n nih.gov\n \n .\n \n \n \n \n \n Other Plan(s):\n \n \n \n All instructions in the How to Apply-Application Guide must be followed, with the following additional instructions:\n \n \n \n \n \n All applicants planning research (funded or conducted in whole or in part by NIH) that results in the generation of scientific data are required to comply with the instructions for the Data Management and Sharing Plan.\n \n \n \n \n \n \n Appendix:\n \n Only limited Appendix materials are allowed. Follow all instructions for the Appendix as described in the How to Apply- Application Guide.\n \n \n \n No publications or other material, with the exception of blank questionnaires or blank surveys, may be included in the Appendix.\n \n \n \n \n \n PHS Human Subjects and Clinical Trials Information\n \n \n When involving human subjects research, clinical research, and/or NIH-defined clinical trials (and when applicable, clinical trials research experience) follow all instructions for the PHS Human Subjects and Clinical Trials Information form in the How to Apply- Application Guide, with the following additional instructions:\n \n \n If you answered “Yes” to the question “Are Human Subjects Involved?” on the R&R Other Project Information form, you must include at least one human subjects study record using the\n \n Study Record: PHS Human Subjects and Clinical Trials Information\n \n form or\n \n Delayed Onset Study\n \n record.\n \n \n \n Study Record: PHS Human Subjects and Clinical Trials Information\n \n \n \n All instructions in the How to Apply- Application Guide must be followed.\n \n \n \n \n \n Delayed Onset Study\n \n \n \n Note:\n \n Delayed onset\n \n does NOT apply to a study that can be described but will not start immediately (i.e., delayed start). All instructions in the How to Apply- Application Guide must be followed.\n \n \n \n \n PHS Assignment Request Form\n \n \n All instructions in the How to Apply- Application Guide must be followed.\n \n \n \n \n 3. Unique Entity Identifier and System for Award Management (SAM)\n \n \n See Part 2. Section III.1 for information regarding the requirement for obtaining a unique entity identifier and for completing and maintaining active registrations in System for Award Management (SAM), NATO Commercial and Government Entity (NCAGE) Code (if applicable), eRA Commons, and Grants.gov\n \n \n \n \n 4. Submission Dates and Times\n \n \n Part I. contains information about Key Dates and times. Applicants are encouraged to submit applications before the due date to ensure they have time to make any application corrections that might be necessary for successful submission. When a submission date falls on a weekend or\n \n Federal holiday\n \n , the application deadline is automatically extended to the next business day.\n \n \n Organizations must submit applications to\n \n Grants.gov\n \n (the online portal to find and apply for grants across all Federal agencies). Applicants must then complete the submission process by tracking the status of the application in the\n \n eRA Commons\n \n , NIH’s electronic system for grants administration. NIH and Grants.gov systems check the application against many of the application instructions upon submission. Errors must be corrected and a changed/corrected application must be submitted to Grants.gov on or before the application due date and time. If a Changed/Corrected application is submitted after the deadline, the application will be considered late. Applications that miss the due date and time are subjected to the\n \n NIH Grants Policy Statement Section 2.3.9.2 Electronically Submitted Applications\n \n .\n \n \n \n Applicants are responsible for viewing their application before the due date in the eRA Commons to ensure accurate and successful submission.\n \n \n \n Information on the submission process and a definition of on-time submission are provided in the How to Apply-Application Guide.\n \n \n \n \n 5. Intergovernmental Review (E.O. 12372)\n \n \n This initiative is not subject to\n \n intergovernmental review.\n \n \n \n \n \n 6. Funding Restrictions\n \n \n All NIH awards are subject to the terms and conditions, cost principles, and other considerations described in the\n \n NIH Grants Policy Statement\n \n .\n \n \n Pre-award costs are allowable only as described in the\n \n NIH Grants Policy Statement Section 7.9.1 Selected Items of Cost.\n \n \n \n \n 7. Other Submission Requirements and Information\n \n Applications must be submitted electronically following the instructions described in the How to Apply Application Guide. Paper applications will not be accepted.\n \n \n Applicants must complete all required registrations before the application due date. Section III. Eligibility Information contains information about registration.\n \n \n For assistance with your electronic application or for more information on the electronic submission process, visit\n \n How to Apply – Application Guide\n \n . If you encounter a system issue beyond your control that threatens your ability to complete the submission process on-time, you must follow the\n \n Dealing with System Issues\n \n guidance. For assistance with application submission, contact the Application Submission Contacts in Section VII.\n \n \n \n Important reminders:\n \n \n \n All PD(s)/PI(s) must include their eRA Commons ID in the Credential field of the Senior/Key Person Profile form\n \n .\n \n Failure to register in the Commons and to include a valid PD/PI Commons ID in the credential field will prevent the successful submission of an electronic application to NIH. See Section III of this NOFO for information on registration requirements.\n \n \n The applicant organization must ensure that the unique entity identifier provided on the application is the same identifier used in the organization’s profile in the eRA Commons and for the System for Award Management. Additional information may be found in the How to Apply Application Guide.\n \n \n See\n \n more tips\n \n for avoiding common errors.\n \n \n \n \n Upon receipt, applications will be evaluated for completeness and compliance with application instructions by the Center for Scientific Review and responsiveness by NIDDK. Applications that are incomplete, non-compliant, and/or nonresponsive will not be reviewed.\n \n \n \n Mandatory Disclosure\n \n Recipients or subrecipients must submit any information related to violations of federal criminal law involving fraud, bribery, or gratuity violations potentially affecting the federal award. See Mandatory Disclosures,\n \n 2 CFR 200.113\n \n and\n \n NIH Grants Policy Statement Section 4.1.35\n \n .\n \n \n Send written disclosures to the NIH Chief Grants Management Officer listed on the Notice of Award for the IC that funded the award and to the\n \n HHS Office of Inspector Grant Self Disclosure Program\n \n at\n \n \n [email protected]\n \n \n .\n \n \n \n \n Post Submission Materials\n \n \n Applicants are required to follow the instructions for post-submission materials, as described in\n \n the policy"]}, {"source_sentence": "What are the main objectives of the Leadership and Coordination Center (LACC) for the MACS/WIHS Combined Cohort Study (MWCCS) as outlined in opportunity id RFA-HL-26-011?", "sentences": ["RFA-HL-26-011: Leadership and Coordination Center (LACC) for the MACS/WIHS Combined Cohort Study (MWCCS) (U01 Clinical Trials Not Allowed) Part 2. Full Text of Announcement", "PAR-25-353: National Cooperative Drug/Device Discovery/Development Groups (NCDDG) for the Treatment of Mental Disorders (U19 Clinical Trial Optional) Section VIII. Other Information\n \n \n \n \n Recently issued trans-NIH\n \n policy notices\n \n may affect your application submission. A full list of policy notices published by NIH is provided in the\n \n NIH Guide for Grants and Contracts\n \n . All awards are subject to the terms and conditions, cost principles, and other considerations described in the\n \n NIH Grants Policy Statement\n \n .\n \n \n \n Authority and Regulations\n \n Awards are made under the authorization of Sections 301 and 405 of the Public Health Service Act as amended (42 USC 241 and 284) and under Federal Regulations 42 CFR Part 52 and 2 CFR Part 200.\n \n end row \n \n \n \n \n Weekly TOC for this Announcement\n \n \n \n NIH Funding Opportunities and Notices\n \n \n \n \n \n \n \n \n \n \n \n \n \n Department of Health\n \n and Human Services (HHS)\n \n \n \n \n NIH... Turning Discovery Into Health\n \n ®", "PAR-25-353: National Cooperative Drug/Device Discovery/Development Groups (NCDDG) for the Treatment of Mental Disorders (U19 Clinical Trial Optional) Section III. Eligibility Information\n \n \n \n \n 1. Eligible Applicants\n \n \n \n Eligible Organizations\n \n Higher Education Institutions\n \n \n \n Public/State Controlled Institutions of Higher Education\n \n \n Private Institutions of Higher Education\n \n \n \n The following types of Higher Education Institutions are always encouraged to apply for NIH support as Public or Private Institutions of Higher Education:\n \n \n \n Hispanic-serving Institutions\n \n \n Historically Black Colleges and Universities (HBCUs)\n \n \n Tribally Controlled Colleges and Universities (TCCUs)\n \n \n Alaska Native and Native Hawaiian Serving Institutions\n \n \n Asian American Native American Pacific Islander Serving Institutions (AANAPISIs)\n \n \n \n Nonprofits Other Than Institutions of Higher Education\n \n \n \n Nonprofits with 501(c)(3) IRS Status (Other than Institutions of Higher Education)\n \n \n Nonprofits without 501(c)(3) IRS Status (Other than Institutions of Higher Education)\n \n \n \n For-Profit Organizations\n \n \n \n Small Businesses\n \n \n For-Profit Organizations (Other than Small Businesses)\n \n \n \n Local Governments\n \n \n \n State Governments\n \n \n County Governments\n \n \n City or Township Governments\n \n \n Special District Governments\n \n \n Indian/Native American Tribal Governments (Federally Recognized)\n \n \n Indian/Native American Tribal Governments (Other than Federally Recognized)\n \n \n \n \n \n Federal Governments\n \n \n \n Eligible Agencies of the Federal Government\n \n \n U.S. Territory or Possession\n \n \n \n \n \n Other\n \n \n \n Independent School Districts\n \n \n Public Housing Authorities/Indian Housing Authorities\n \n \n Native American Tribal Organizations (other than Federally recognized tribal governments)\n \n \n Faith-based or Community-based Organizations\n \n \n Regional Organizations\n \n \n \n \n Foreign Organizations\n \n \n \n Non-domestic (non-U.S.) Entities (Foreign Organization)\n \n are not\n \n eligible to apply.\n \n \n \n \n Non-domestic (non-U.S.) components of U.S. Organizations\n \n are not\n \n eligible to apply.\n \n \n \n \n Foreign components, as\n \n defined in the NIH Grants Policy Statement\n \n ,\n \n are\n \n allowed.\n \n \n \n Required Registrations\n \n \n Applicant organizations\n \n \n \n Applicant organizations must complete and maintain the following registrations as described in the\n \n How to Apply- Application Guide\n \n to be eligible to apply for or receive an award. All registrations must be completed prior to the application being submitted. Registration can take 6 weeks or more, so applicants should begin the registration process as soon as possible. Failure to complete registrations in advance of a due date is not a valid reason for a late submission, please reference\n \n NIH Grants Policy Statement Section 2.3.9.2 Electronically Submitted Applications\n \n for additional information.\n \n \n \n \n System for Award Management (SAM)\n \n – Applicants must complete and maintain an active registration,\n \n which requires renewal at least annually\n \n . The renewal process may require as much time as the initial registration. SAM registration includes the assignment of a Commercial and Government Entity (CAGE) Code for domestic organizations which have not already been assigned a CAGE Code.\n \n \n \n NATO Commercial and Government Entity (NCAGE) Code\n \n – Foreign organizations must obtain an NCAGE code (in lieu of a CAGE code) in order to register in SAM.\n \n \n Unique Entity Identifier (UEI) - A UEI is issued as part of the SAM.gov registration process. The same UEI must be used for all registrations, as well as on the grant application.\n \n \n \n \n \n eRA Commons\n \n - Once the unique organization identifier is established, organizations can register with eRA Commons in tandem with completing their Grants.gov registration; all registrations must be in place by time of submission. eRA Commons requires organizations to identify at least one Signing Official (SO) and at least one Program Director/Principal Investigator (PD/PI) account in order to submit an application.\n \n \n \n Grants.gov\n \n – Applicants must have an active SAM registration in order to complete the Grants.gov registration.\n \n \n \n \n Program Directors/Principal Investigators (PD(s)/PI(s))\n \n \n \n All PD(s)/PI(s) must have an eRA Commons account. PD(s)/PI(s) should work with their organizational officials to either create a new account or to affiliate their existing account with the applicant organization in eRA Commons. If the PD/PI is also the organizational Signing Official, they must have two distinct eRA Commons accounts, one for each role. Obtaining an eRA Commons account can take up to 2 weeks.\n \n \n \n Eligible Individuals (Program Director/Principal Investigator)\n \n Any individual(s) with the skills, knowledge, and resources necessary to carry out the proposed research as the Program Director(s)/Principal Investigator(s) (PD(s)/PI(s)) is invited to work with his/her organization to develop an application for support.\n \n \n For institutions/organizations proposing multiple PDs/PIs, visit the Multiple Program Director/Principal Investigator Policy and submission details in the Senior/Key Person Profile (Expanded) Component of the\n \n How to Apply - Application Guide\n \n .\n \n \n \n \n An NIH intramural scientist may not serve as the PD/PI of an NCDDG but may participate as a research collaborator or consultant (see\n \n \n \n Section IV.7\n \n \n \n for more information).\n \n \n \n \n 2. Cost Sharing\n \n \n This NOFO does not require cost sharing as defined in the\n \n NIH Grants Policy Statement\n \n \n Section 1.2- Definitions of Terms.\n \n \n \n \n \n 3. Additional Information on Eligibility\n \n \n Number of Applications\n \n \n \n \n Applicant organizations may submit more than one application, provided that each application is scientifically distinct.\n \n \n The NIH will not accept duplicate or highly overlapping applications under review at the same time per\n \n NIH Grants Policy Statement Section 2.3.7.4 Submission of Resubmission Application\n \n . This means that the NIH will not accept:\n \n \n \n A new (A0) application that is submitted before issuance of the summary statement from the review of an overlapping new (A0) or resubmission (A1) application.\n \n \n A resubmission (A1) application that is submitted before issuance of the summary statement from the review of the previous new (A0) application.\n \n \n An application that has substantial overlap with another application pending appeal of initial peer review (see\n \n NIH Grants Policy Statement 2.3.9.4 Similar, Essentially Identical, or Identical Applications\n \n )."]}, {"source_sentence": "Can an organization submit multiple applications for the RFA-HL-26-011 opportunity, and what conditions must be met for these applications?", "sentences": ["PAR-25-378: Intervention Research to Improve Native American Health (R34 Clinical Trial Optional) Section VI. Award Administration Information\n \n \n \n \n 1. Award Notices\n \n \n A Notice of Award (NoA) is the official authorizing document notifying the applicant that an award has been made and that funds may be requested from the designated HHS payment system or office. The NoA is signed by the Grants Management Officer and emailed to the recipient’s business official.\n \n \n In accepting the award, the recipient agrees that any activities under the award are subject to all provisions currently in effect or implemented during the period of the award, other Department regulations and policies in effect at the time of the award, and applicable statutory provisions.\n \n \n Recipients must comply with any funding restrictions described in\n \n Section IV.6. Funding Restrictions\n \n . Any pre-award costs incurred before receipt of the NoA are at the applicant's own risk. For more information on the Notice of Award, please refer to the\n \n NIH Grants Policy Statement Section 5. The Notice of Award\n \n and NIH Grants & Funding website, see\n \n Award Process.\n \n \n \n \n \n Individual awards are based on the application submitted to, and as approved by, the NIH and are subject to the IC-specific terms and conditions identified in the NoA.\n \n \n ClinicalTrials.gov: If an award provides for one or more clinical trials. By law (Title VIII, Section 801 of Public Law 110-85), the \"responsible party\" must register and submit results information for certain “applicable clinical trials” on the ClinicalTrials.gov Protocol Registration and Results System Information Website (\n \n https://register.clinicaltrials.gov\n \n ). NIH expects registration and results reporting of all trials whether required under the law or not. For more information, see\n \n https://grants.nih.gov/policy/clinical-trials/reporting/index.htm\n \n \n \n Institutional Review Board or Independent Ethics Committee Approval: Recipient institutions must ensure that all protocols are reviewed by their IRB or IEC. To help ensure the safety of participants enrolled in NIH-funded studies, the recipient must provide NIH copies of documents related to all major changes in the status of ongoing protocols.\n \n \n Data and Safety Monitoring Requirements: The NIH policy for data and safety monitoring requires oversight and monitoring of all NIH-conducted or -supported human biomedical and behavioral intervention studies (clinical trials) to ensure the safety of participants and the validity and integrity of the data. Further information concerning these requirements is found at http://grants.nih.gov/grants/policy/hs/data_safety.htm and in the application instructions (SF424 (R&R) and PHS 398).\n \n \n Investigational New Drug or Investigational Device Exemption Requirements: Consistent with federal regulations, clinical research projects involving the use of investigational therapeutics, vaccines, or other medical interventions (including licensed products and devices for a purpose other than that for which they were licensed) in humans under a research protocol must be performed under a Food and Drug Administration (FDA) investigational new drug (IND) or investigational device exemption (IDE).\n \n \n \n \n 2. Administrative and National Policy Requirements\n \n \n The following Federal wide and HHS-specific policy requirements apply to awards funded through NIH:\n \n \n \n The rules listed at\n \n 2 CFR Part 200\n \n , Uniform Administrative Requirements, Cost Principles, and Audit Requirements for Federal Awards.\n \n \n All NIH grant and cooperative agreement awards include the\n \n NIH Grants Policy Statement\n \n as part of the terms and conditions in the Notice of Award (NoA). The NoA includes the requirements of this NOFO. For these terms of award, see the\n \n NIH Grants Policy Statement Part II: Terms and Conditions of NIH Grant Awards, Subpart A: General\n \n and\n \n Part II: Terms and Conditions of NIH Grant Awards, Subpart B: Terms and Conditions for Specific Types of Grants, Recipients, and Activities\n \n .\n \n \n If a recipient receives an award, the recipient must follow all applicable nondiscrimination laws. The recipient agrees to this when registering in SAM.gov. The recipient must also submit an Assurance of Compliance (\n \n HHS-690\n \n ). To learn more, see the\n \n Laws and Regulations Enforced by the HHS Office for Civil Rights website\n \n .\n \n \n HHS recognizes that NIH research projects are often limited in scope for many reasons that are nondiscriminatory, such as the principal investigator’s scientific interest, funding limitations, recruitment requirements, and other considerations. Thus, criteria in research protocols that target or exclude certain populations are warranted where nondiscriminatory justifications establish that such criteria are appropriate with respect to the health or safety of the subjects, the scientific study design, or the purpose of the research. For additional guidance regarding how the provisions apply to NIH grant programs, please contact the Scientific/Research Contact that is identified in Section VII under Agency Contacts of this NOFO.\n \n \n \n \n \n All federal statutes and regulations relevant to federal financial assistance, including those highlighted in\n \n NIH Grants Policy Statement Section 4 Public Policy Requirements, Objectives and Other Appropriation Mandates.\n \n \n \n Recipients are responsible for ensuring that their activities comply with all applicable federal regulations. NIH may terminate awards under certain circumstances. See\n \n 2 CFR Part 200.340 Termination\n \n and\n \n NIH Grants Policy Statement Section 8.5.2 Remedies for Noncompliance or Enforcement Actions: Suspension, Termination, and Withholding of Support\n \n .\n \n \n Successful recipients under this NOFO agree that:\n \n \n Where the award funding involves implementing, acquiring, or upgrading health IT for activities by any funded entity, recipients and subrecipient(s) are required to: Use health IT that meets standards and implementation specifications adopted in 45 CFR part 170, Subpart B, if such standards and implementation specifications can support the activity. Visit\n \n https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-D/part-170/subpart-B\n \n to learn more.\n \n \n Where the award funding involves implementing, acquiring, or upgrading health IT for activities by eligible clinicians in ambulatory settings, or hospitals, eligible under Sections 4101, 4102, and 4201 of the HITECH Act, use health IT certified under the ONC Health IT Certification Program if certified technology can support the activity. Visit\n \n https://www.healthit.gov/topic/certification-ehrs/certification-health-it\n \n to learn more.\n \n \n Pursuant to the Cybersecurity Act of 2015, Div. N, § 405, Pub. Law 114-113, 6 USC § 1533(d), the HHS Secretary has established a common set of voluntary, consensus-based, and industry-led guidelines, best practices, methodologies, procedures, and processes.\n \n \n Successful recipients under this NOFO agree that:\n \n \n When recipients, subrecipients, or third-party entities have:\n \n \n 1) ongoing and consistent access to HHS owned or operated information or operational technology systems; and\n \n \n 2) receive, maintain, transmit, store, access, exchange, process, or utilize personal identifiable information (PII) or personal health information (PHI) obtained from the awarding HHS agency for the purposes of executing the award.\n \n \n Recipients shall develop plans and procedures, modeled after the\n \n NIST Cybersecurity framework\n \n , to protect HHS systems and data. Please refer to\n \n NIH Post-Award Monitoring and Reporting\n \n for additional information.\n \n \n \n Cooperative Agreement Terms and Conditions of Award\n \n Not Applicable\n \n \n \n 3. Data Management and Sharing\n \n Consistent with the 2023 NIH Policy for Data Management and Sharing, when data management and sharing is applicable to the award, recipients will be required to adhere to the Data Management and Sharing requirements as outlined in the\n \n NIH Grants Policy Statement\n \n . Upon the approval of a Data Management and Sharing Plan, it is required for recipients to implement the plan as described.\n \n \n \n \n 4. Reporting\n \n \n When multiple years are involved, recipients will be required to submit the\n \n Research Performance Progress Report (RPPR)\n \n annually and financial statements as required in the\n \n NIH Grants Policy Statement Section 8.4.1 Reporting.\n \n To learn more about post-award monitoring and reporting, see the NIH Grants & Funding website, see\n \n Post-Award Monitoring and Reporting\n \n .\n \n \n \n \n A final RPPR, invention statement, and the expenditure data portion of the Federal Financial Report are required for closeout of an award, as described in the\n \n NIH Grants Policy Statement Section 8.6 Closeout\n \n . NIH NOFOs outline intended research goals and objectives. Post award, NIH will review and measure performance based on the details and outcomes that are shared within the RPPR, as described at 2 CFR Part 200.301.", "PAR-25-378: Intervention Research to Improve Native American Health (R34 Clinical Trial Optional) Section II. Award Information\n \n \n \n Funding Instrument\n \n Grant: A financial assistance mechanism providing money, property, or both to an eligible entity to carry out an approved project or activity.\n \n \n \n Application Types Allowed\n New\n \n Resubmission\n \n \n \n \n The\n \n OER Glossary\n \n and the How to Apply Application Guide provide details on these application types. Only those application types listed here are allowed for this NOFO.\n \n \n \n Clinical Trial?\n \n Optional: Accepting applications that either propose or do not propose clinical trial(s).\n \n \n \n \n \n Need help determining whether you are doing a clinical trial?\n \n \n \n \n Funds Available and Anticipated Number of Awards\n \n The number of awards is contingent upon NIH appropriations and the submission of a sufficient number of meritorious applications.\n \n \n \n Award Budget\n \n Applicants may request direct costs of up to $450,000 for three years. Although variations from year to year are permissible, in no case may any year be more than $225,000 in direct costs, and total direct costs for the entire project period may not exceed $450,000.\n \n \n \n Award Project Period\n \n The scope of the proposed project should determine the project period. The maximum project period is 3 years.\n \n This variable defines that we need to start a new row. \n \n \n \n NIH grants policies as described in the\n \n NIH Grants Policy Statement\n \n will apply to the applications submitted and awards made from this NOFO.", "RFA-HL-26-011: Leadership and Coordination Center (LACC) for the MACS/WIHS Combined Cohort Study (MWCCS) (U01 Clinical Trials Not Allowed) Section III. Eligibility Information\n \n \n \n \n 1. Eligible Applicants\n \n \n \n Eligible Organizations\n \n Higher Education Institutions\n \n \n \n Public/State Controlled Institutions of Higher Education\n \n \n Private Institutions of Higher Education\n \n \n \n The following types of Higher Education Institutions are always encouraged to apply for NIH support as Public or Private Institutions of Higher Education:\n \n \n \n Hispanic-serving Institutions\n \n \n Historically Black Colleges and Universities (HBCUs)\n \n \n Tribally Controlled Colleges and Universities (TCCUs)\n \n \n Alaska Native and Native Hawaiian Serving Institutions\n \n \n Asian American Native American Pacific Islander Serving Institutions (AANAPISIs)\n \n \n \n Nonprofits Other Than Institutions of Higher Education\n \n \n \n Nonprofits with 501(c)(3) IRS Status (Other than Institutions of Higher Education)\n \n \n Nonprofits without 501(c)(3) IRS Status (Other than Institutions of Higher Education)\n \n \n \n For-Profit Organizations\n \n \n \n Small Businesses\n \n \n For-Profit Organizations (Other than Small Businesses)\n \n \n \n Local Governments\n \n \n \n State Governments\n \n \n County Governments\n \n \n City or Township Governments\n \n \n Special District Governments\n \n \n Indian/Native American Tribal Governments (Federally Recognized)\n \n \n Indian/Native American Tribal Governments (Other than Federally Recognized).\n \n \n \n \n \n Federal Governments\n \n \n \n Eligible Agencies of the Federal Government\n \n \n U.S. Territory or Possession\n \n \n \n \n \n Other\n \n \n \n Independent School Districts\n \n \n Public Housing Authorities/Indian Housing Authorities\n \n \n Native American Tribal Organizations (other than Federally recognized tribal governments)\n \n \n Faith-based or Community-based Organizations\n \n \n Regional Organizations\n \n \n \n \n Foreign Organizations\n \n \n \n Non-domestic (non-U.S.) Entities (Foreign Organizations)\n \n are not\n \n eligible to apply.\n \n \n \n \n Non-domestic (non-U.S.) components of U.S. Organizations\n \n are not\n \n eligible to apply.\n \n \n \n \n Foreign components, as\n \n defined in the NIH Grants Policy Statement\n \n ,\n \n are not\n \n allowed.\n \n \n \n Required Registrations\n \n \n Applicant Organizations\n \n \n \n Applicant organizations must complete and maintain the following registrations as described in the How to Apply- Application Guide to be eligible to apply for or receive an award. All registrations must be completed prior to the application being submitted. Registration can take 6 weeks or more, so applicants should begin the registration process as soon as possible. Failure to complete registrations in advance of a due date is not a valid reason for a late submission, please reference the\n \n NIH Grants Policy Statement Section 2.3.9.2 Electronically Submitted Applications\n \n for additional information.\n \n \n \n \n System for Award Management (SAM) –\n \n Applicants must complete and maintain an active registration,\n \n which requires renewal at least annually\n \n . The renewal process may require as much time as the initial registration. SAM registration includes the assignment of a Commercial and Government Entity (CAGE) Code for domestic organizations which have not already been assigned a CAGE Code.\n \n \n \n NATO Commercial and Government Entity (NCAGE) Code\n \n – Foreign organizations must obtain an NCAGE code (in lieu of a CAGE code) in order to register in SAM.\n \n \n Unique Entity Identifier (UEI) - A UEI is issued as part of the SAM.gov registration process. The same UEI must be used for all registrations, as well as on the grant application.\n \n \n \n \n \n eRA Commons\n \n - Once the unique organization identifier is established, organizations can register with eRA Commons in tandem with completing their Grants.gov registrations; all registrations must be in place by time of submission. eRA Commons requires organizations to identify at least one Signing Official (SO) and at least one Program Director/Principal Investigator (PD/PI) account in order to submit an application.\n \n \n \n Grants.gov\n \n – Applicants must have an active SAM registration in order to complete the Grants.gov registration.\n \n \n \n \n Program Directors/Principal Investigators (PD(s)/PI(s))\n \n \n \n All PD(s)/PI(s) must have an eRA Commons account. PD(s)/PI(s) should work with their organizational officials to either create a new account or to affiliate their existing account with the applicant organization in eRA Commons. If the PD/PI is also the organizational Signing Official, they must have two distinct eRA Commons accounts, one for each role. Obtaining an eRA Commons account can take up to 2 weeks.\n \n \n \n Eligible Individuals (Program Director/Principal Investigator)\n \n Any individual(s) with the skills, knowledge, and resources necessary to carry out the proposed research as the Program Director(s)/Principal Investigator(s) (PD(s)/PI(s)) is invited to work with their organization to develop an application for support.\n \n \n For institutions/organizations proposing multiple PDs/PIs, visit the Multiple Program Director/Principal Investigator Policy and submission details in the Senior/Key Person Profile (Expanded) Component of the How to Apply-Application Guide.\n \n \n \n \n A multiple PI (mPI) leadership structure is required. The LACC mPI team should include individuals with demonstrated deep knowledge of, and familiarity with, the MWCCS. In addition, the team should include at least one clinician with experience in clinical care of patients living with HIV and/or relevant comorbidities, as well as one or more individuals with epidemiologic cohort study leadership experience.Individuals may not apply to be an mPI of both the LACC and Data Analysis and Sharing Center (DASC) (\n \n RFA-HL-26-010\n \n ). Applications without an mPI leadership structure will be considered incomplete and will not proceed to peer review.\n \n \n \n \n 2. Cost Sharing\n \n \n This NOFO does not require cost sharing as defined in the\n \n NIH Grants Policy Statement Section 1.2 Definition of Terms\n \n .\n \n \n \n \n 3. Additional Information on Eligibility\n \n \n Number of Applications\n \n \n \n \n Applicant organizations may submit more than one application, provided that each application is scientifically distinct.\n \n \n The NIH will not accept duplicate or highly overlapping applications under review at the same time, per\n \n NIH Grants Policy Statement Section 2.3.7.4 Submission of Resubmission Application\n \n . This means that the NIH will not accept:\n \n \n \n A new (A0) application that is submitted before issuance of the summary statement from the review of an overlapping new (A0) or resubmission (A1) application.\n \n \n A resubmission (A1) application that is submitted before issuance of the summary statement from the review of the previous new (A0) application.\n \n \n An application that has substantial overlap with another application pending appeal of initial peer review (see\n \n NIH Grants Policy Statement 2.3.9.4 Similar, Essentially Identical, or Identical Applications\n \n )."]}, {"source_sentence": "What is the purpose of the funding opportunity RFA-DK-26-007 titled \"Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies\"?", "sentences": ["Department of Health and Human Services", "RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Part 1. Overview Information\n \n \n \n Participating Organization(s)\n \n National Institutes of Health (\n \n NIH\n \n )\n \n \n \n Components of Participating Organizations\n \n National Institute of Diabetes and Digestive and Kidney Diseases (\n \n NIDDK\n \n )\n \n \n Office of The Director, National Institutes of Health (\n \n OD\n \n )\n \n \n \n Funding Opportunity Title\n \n Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed)\n \n \n \n Activity Code\n \n \n R01\n \n Research Project Grant\n \n \n \n Announcement Type\n Reissue of\n \n RFA-DK-22-021\n \n \n \n \n Related Notices\n \n \n \n April 4, 2024\n \n - Overview of Grant Application and Review Changes for Due Dates on or after January 25, 2025. See Notice\n \n NOT-OD-24-084\n \n .\n \n \n \n August 31, 2022\n \n - Implementation Changes for Genomic Data Sharing Plans Included with Applications Due on or after January 25, 2023. See Notice\n \n NOT-OD-22-198\n \n .\n \n \n \n August 5, 2022\n \n - Implementation Details for the NIH Data Management and Sharing Policy. See Notice\n \n NOT-OD-22-189\n \n .\n \n \n \n \n Funding Opportunity Number (FON)\n \n RFA-DK-26-007\n \n \n \n Companion Funding Opportunity\n None\n \n \n Number of Applications\n \n See\n \n Section III. 3. Additional Information on Eligibility\n \n .\n \n \n \n Assistance Listing Number(s)\n 93.847\n \n \n Funding Opportunity Purpose\n \n This Notice of Funding Opportunity (NOFO) invites applications for studies of type 1 diabetes etiology and pathogenesis using data and samples from clinical trials and studies. This opportunity is intended to fund investigative teams collaborating to answer important questions about disease mechanisms leading to improved delay and durable prevention of type 1 diabetes. This NOFO is associated with the Special Diabetes Program (\n \n https://www.niddk.nih.gov/about-niddk/research-areas/diabetes/type-1-diabetes-special-statutory-funding-program/about-special-diabetes-program\n \n ) which funds research on the prevention, treatment, and cure of type 1 diabetes and its complications, including unique, innovative, and collaborative research consortia and clinical trials networks.\n \n \n \n \n \n \n \n Funding Opportunity Goal(s)\n \n To promote extramural basic and clinical biomedical research that improves the understanding of the mechanisms underlying disease and leads to improved preventions, diagnosis, and treatment of diabetes, digestive, and kidney diseases. Programmatic areas within the National Institute of Diabetes and Digestive and Kidney Diseases include diabetes, digestive, endocrine, hematologic, liver, metabolic, nephrologic, nutrition, obesity, and urologic diseases.\n \n This variable defines that we need to start a new row.", "RFA-DK-26-007: Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed) Part 1. Overview Information\n \n \n \n Participating Organization(s)\n \n National Institutes of Health (\n \n NIH\n \n )\n \n \n \n Components of Participating Organizations\n \n National Institute of Diabetes and Digestive and Kidney Diseases (\n \n NIDDK\n \n )\n \n \n Office of The Director, National Institutes of Health (\n \n OD\n \n )\n \n \n \n Funding Opportunity Title\n \n Collaborative Research Using Biosamples and/or Data from Type 1 Diabetes Clinical Studies (R01 - Clinical Trial Not Allowed)\n \n \n \n Activity Code\n \n \n R01\n \n Research Project Grant\n \n \n \n Announcement Type\n Reissue of\n \n RFA-DK-22-021\n \n \n \n \n Related Notices\n \n \n \n April 4, 2024\n \n - Overview of Grant Application and Review Changes for Due Dates on or after January 25, 2025. See Notice\n \n NOT-OD-24-084\n \n .\n \n \n \n August 31, 2022\n \n - Implementation Changes for Genomic Data Sharing Plans Included with Applications Due on or after January 25, 2023. See Notice\n \n NOT-OD-22-198\n \n .\n \n \n \n August 5, 2022\n \n - Implementation Details for the NIH Data Management and Sharing Policy. See Notice\n \n NOT-OD-22-189\n \n .\n \n \n \n \n Funding Opportunity Number (FON)\n \n RFA-DK-26-007\n \n \n \n Companion Funding Opportunity\n None\n \n \n Number of Applications\n \n See\n \n Section III. 3. Additional Information on Eligibility\n \n .\n \n \n \n Assistance Listing Number(s)\n 93.847\n \n \n Funding Opportunity Purpose\n \n This Notice of Funding Opportunity (NOFO) invites applications for studies of type 1 diabetes etiology and pathogenesis using data and samples from clinical trials and studies. This opportunity is intended to fund investigative teams collaborating to answer important questions about disease mechanisms leading to improved delay and durable prevention of type 1 diabetes. This NOFO is associated with the Special Diabetes Program (\n \n https://www.niddk.nih.gov/about-niddk/research-areas/diabetes/type-1-diabetes-special-statutory-funding-program/about-special-diabetes-program\n \n ) which funds research on the prevention, treatment, and cure of type 1 diabetes and its complications, including unique, innovative, and collaborative research consortia and clinical trials networks.\n \n \n \n \n \n \n \n Funding Opportunity Goal(s)\n \n To promote extramural basic and clinical biomedical research that improves the understanding of the mechanisms underlying disease and leads to improved preventions, diagnosis, and treatment of diabetes, digestive, and kidney diseases. Programmatic areas within the National Institute of Diabetes and Digestive and Kidney Diseases include diabetes, digestive, endocrine, hematologic, liver, metabolic, nephrologic, nutrition, obesity, and urologic diseases.\n \n This variable defines that we need to start a new row."]}], "model-index": [{"name": "SentenceTransformer based on Snowflake/snowflake-arctic-embed-l", "results": [{"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "Unknown", "type": "unknown"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.875, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 1.0, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 1.0, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 1.0, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.875, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.3333333333333333, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.20000000000000004, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.10000000000000002, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.875, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 1.0, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 1.0, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 1.0, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.9484108127976215, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.9305555555555555, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.9305555555555555, "name": "Cosine Map@100"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,093
Realgon/N_distilbert_imdb_padding60model	Realgon	text-classification	[ "transformers", "pytorch", "distilbert", "text-classification", "generated_from_trainer", "dataset:imdb", "base_model:distilbert/distilbert-base-uncased", "base_model:finetune:distilbert/distilbert-base-uncased", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-12-11T12:33:28Z	2023-12-11T13:51:02+00:00	14	0	--- base_model: distilbert-base-uncased datasets: - imdb license: apache-2.0 metrics: - accuracy tags: - generated_from_trainer model-index: - name: N_distilbert_imdb_padding60model results: - task: type: text-classification name: Text Classification dataset: name: imdb type: imdb config: plain_text split: test args: plain_text metrics: - type: accuracy value: 0.93268 name: Accuracy --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # N_distilbert_imdb_padding60model This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the imdb dataset. It achieves the following results on the evaluation set: - Loss: 0.7224 - Accuracy: 0.9327 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 20 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:-----:\|:---------------:\|:--------:\| \| 0.2346 \| 1.0 \| 1563 \| 0.2252 \| 0.916 \| \| 0.1742 \| 2.0 \| 3126 \| 0.2406 \| 0.9204 \| \| 0.1246 \| 3.0 \| 4689 \| 0.3171 \| 0.9224 \| \| 0.0738 \| 4.0 \| 6252 \| 0.3747 \| 0.9245 \| \| 0.0507 \| 5.0 \| 7815 \| 0.4165 \| 0.9278 \| \| 0.0327 \| 6.0 \| 9378 \| 0.5113 \| 0.9248 \| \| 0.0218 \| 7.0 \| 10941 \| 0.5063 \| 0.9210 \| \| 0.0221 \| 8.0 \| 12504 \| 0.5326 \| 0.9279 \| \| 0.0231 \| 9.0 \| 14067 \| 0.5171 \| 0.9279 \| \| 0.0111 \| 10.0 \| 15630 \| 0.6266 \| 0.9275 \| \| 0.0096 \| 11.0 \| 17193 \| 0.6049 \| 0.9255 \| \| 0.0092 \| 12.0 \| 18756 \| 0.6766 \| 0.9237 \| \| 0.0079 \| 13.0 \| 20319 \| 0.6736 \| 0.9273 \| \| 0.0082 \| 14.0 \| 21882 \| 0.6786 \| 0.9296 \| \| 0.0047 \| 15.0 \| 23445 \| 0.6562 \| 0.9298 \| \| 0.003 \| 16.0 \| 25008 \| 0.6903 \| 0.9301 \| \| 0.0028 \| 17.0 \| 26571 \| 0.7158 \| 0.9291 \| \| 0.0 \| 18.0 \| 28134 \| 0.7324 \| 0.9321 \| \| 0.0 \| 19.0 \| 29697 \| 0.7185 \| 0.9325 \| \| 0.0003 \| 20.0 \| 31260 \| 0.7224 \| 0.9327 \| ### Framework versions - Transformers 4.33.2 - Pytorch 2.0.1+cu117 - Datasets 2.14.5 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # N_distilbert_imdb_padding60model This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the imdb dataset. It achieves the following results on the evaluation set: - Loss: 0.7224 - Accuracy: 0.9327 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 20 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:-----:\|:---------------:\|:--------:\| \| 0.2346 \| 1.0 \| 1563 \| 0.2252 \| 0.916 \| \| 0.1742 \| 2.0 \| 3126 \| 0.2406 \| 0.9204 \| \| 0.1246 \| 3.0 \| 4689 \| 0.3171 \| 0.9224 \| \| 0.0738 \| 4.0 \| 6252 \| 0.3747 \| 0.9245 \| \| 0.0507 \| 5.0 \| 7815 \| 0.4165 \| 0.9278 \| \| 0.0327 \| 6.0 \| 9378 \| 0.5113 \| 0.9248 \| \| 0.0218 \| 7.0 \| 10941 \| 0.5063 \| 0.9210 \| \| 0.0221 \| 8.0 \| 12504 \| 0.5326 \| 0.9279 \| \| 0.0231 \| 9.0 \| 14067 \| 0.5171 \| 0.9279 \| \| 0.0111 \| 10.0 \| 15630 \| 0.6266 \| 0.9275 \| \| 0.0096 \| 11.0 \| 17193 \| 0.6049 \| 0.9255 \| \| 0.0092 \| 12.0 \| 18756 \| 0.6766 \| 0.9237 \| \| 0.0079 \| 13.0 \| 20319 \| 0.6736 \| 0.9273 \| \| 0.0082 \| 14.0 \| 21882 \| 0.6786 \| 0.9296 \| \| 0.0047 \| 15.0 \| 23445 \| 0.6562 \| 0.9298 \| \| 0.003 \| 16.0 \| 25008 \| 0.6903 \| 0.9301 \| \| 0.0028 \| 17.0 \| 26571 \| 0.7158 \| 0.9291 \| \| 0.0 \| 18.0 \| 28134 \| 0.7324 \| 0.9321 \| \| 0.0 \| 19.0 \| 29697 \| 0.7185 \| 0.9325 \| \| 0.0003 \| 20.0 \| 31260 \| 0.7224 \| 0.9327 \| ### Framework versions - Transformers 4.33.2 - Pytorch 2.0.1+cu117 - Datasets 2.14.5 - Tokenizers 0.13.3	{"base_model": "distilbert-base-uncased", "datasets": ["imdb"], "license": "apache-2.0", "metrics": ["accuracy"], "tags": ["generated_from_trainer"], "model-index": [{"name": "N_distilbert_imdb_padding60model", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "imdb", "type": "imdb", "config": "plain_text", "split": "test", "args": "plain_text"}, "metrics": [{"type": "accuracy", "value": 0.93268, "name": "Accuracy"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,094
Optimo/kAIra-gemma2-2b	Optimo	translation	[ "safetensors", "translation", "fr", "base_model:google/gemma-2-2b-it", "base_model:finetune:google/gemma-2-2b-it", "license:mit", "region:us" ]	2025-01-05T09:31:46Z	2025-01-05T09:40:47+00:00	0	0	--- base_model: - google/gemma-2-2b-it language: - fr license: mit pipeline_tag: translation --- # Model Card for Model ID kAIra-gemma-2-2b is a finetuned version of google/gemma-2-2b-it can be used in French to translate any text into a slang version of it. ## Model Details ### Model Description <!-- Provide a longer summary of what this model is. --> - Developed by: Sébastien Fischman - Funded by [optional]: None - Shared by [optional]: Sébastien Fischman - Model type: [More Information Needed] - Language(s) (NLP): French - License: MIT - Finetuned from model [optional]: google/gemma-2-2b-it ### Model Sources [optional] <!-- Provide the basic links for the model. --> - Repository: https://www.kaggle.com/code/optimo/kaira-adapting-gemma2-to-local-slang-french/ ## Uses This model is a toy example to showcase how to teach an LLM French slang, it can be used to generate funny translation in argot. ## Bias, Risks, and Limitations Since the model is trained to translate in slang, some generated content can be considered offensive or inappropriate. ### Recommendations <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. --> Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. ## How to Get Started with the Model Use the code from Kaggle here : https://www.kaggle.com/code/optimo/kaira-adapting-gemma2-to-local-slang-french/ Or use the HF Space: Coming [More Information Needed] ## Training Details ### Training Data https://www.kaggle.com/datasets/optimo/french-kaira-training ### Training Procedure Training code is fully accesible here: https://www.kaggle.com/code/optimo/kaira-adapting-gemma2-to-local-slang-french/ #### Training Hyperparameters - Training regime: bf16 [More Information Needed] ### Framework versions - PEFT 0.14.0	null	Non_BioNLP	# Model Card for Model ID kAIra-gemma-2-2b is a finetuned version of google/gemma-2-2b-it can be used in French to translate any text into a slang version of it. ## Model Details ### Model Description <!-- Provide a longer summary of what this model is. --> - Developed by: Sébastien Fischman - Funded by [optional]: None - Shared by [optional]: Sébastien Fischman - Model type: [More Information Needed] - Language(s) (NLP): French - License: MIT - Finetuned from model [optional]: google/gemma-2-2b-it ### Model Sources [optional] <!-- Provide the basic links for the model. --> - Repository: https://www.kaggle.com/code/optimo/kaira-adapting-gemma2-to-local-slang-french/ ## Uses This model is a toy example to showcase how to teach an LLM French slang, it can be used to generate funny translation in argot. ## Bias, Risks, and Limitations Since the model is trained to translate in slang, some generated content can be considered offensive or inappropriate. ### Recommendations <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. --> Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. ## How to Get Started with the Model Use the code from Kaggle here : https://www.kaggle.com/code/optimo/kaira-adapting-gemma2-to-local-slang-french/ Or use the HF Space: Coming [More Information Needed] ## Training Details ### Training Data https://www.kaggle.com/datasets/optimo/french-kaira-training ### Training Procedure Training code is fully accesible here: https://www.kaggle.com/code/optimo/kaira-adapting-gemma2-to-local-slang-french/ #### Training Hyperparameters - Training regime: bf16 [More Information Needed] ### Framework versions - PEFT 0.14.0	{"base_model": ["google/gemma-2-2b-it"], "language": ["fr"], "license": "mit", "pipeline_tag": "translation"}	task	[ "TRANSLATION" ]	43,095
Helsinki-NLP/opus-mt-en-lg	Helsinki-NLP	translation	[ "transformers", "pytorch", "tf", "marian", "text2text-generation", "translation", "en", "lg", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:04Z	2023-08-16T11:30:14+00:00	61	2	--- license: apache-2.0 tags: - translation --- ### opus-mt-en-lg * source languages: en * target languages: lg * OPUS readme: [en-lg](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/en-lg/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-08.zip](https://object.pouta.csc.fi/OPUS-MT-models/en-lg/opus-2020-01-08.zip) * test set translations: [opus-2020-01-08.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-lg/opus-2020-01-08.test.txt) * test set scores: [opus-2020-01-08.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-lg/opus-2020-01-08.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.en.lg \| 30.4 \| 0.543 \| \| Tatoeba.en.lg \| 5.7 \| 0.386 \|	null	Non_BioNLP	### opus-mt-en-lg * source languages: en * target languages: lg * OPUS readme: [en-lg](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/en-lg/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-08.zip](https://object.pouta.csc.fi/OPUS-MT-models/en-lg/opus-2020-01-08.zip) * test set translations: [opus-2020-01-08.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-lg/opus-2020-01-08.test.txt) * test set scores: [opus-2020-01-08.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-lg/opus-2020-01-08.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.en.lg \| 30.4 \| 0.543 \| \| Tatoeba.en.lg \| 5.7 \| 0.386 \|	{"license": "apache-2.0", "tags": ["translation"]}	task	[ "TRANSLATION" ]	43,096
Mizuiro-sakura/deberta-v2-base-japanese-finetuned-QAe	Mizuiro-sakura	question-answering	[ "transformers", "pytorch", "safetensors", "deberta-v2", "question-answering", "deberta", "question answering", "squad", "ja", "dataset:wikipedia", "dataset:cc100", "dataset:oscar", "license:mit", "endpoints_compatible", "region:us" ]	2023-01-09T11:59:13Z	2023-03-27T02:43:35+00:00	23	3	--- datasets: - wikipedia - cc100 - oscar language: ja library_name: transformers license: mit metrics: - accuracy tags: - pytorch - deberta - deberta-v2 - question-answering - question answering - squad --- # このモデルはdeberta-v2-base-japaneseをファインチューニングしてQAタスクに用いれるようにしたものです。このモデルはdeberta-v2-base-japaneseを運転ドメインQAデータセット（DDQA）（　https://nlp.ist.i.kyoto-u.ac.jp/index.php?Driving%20domain%20QA%20datasets　）を用いてファインチューニングしたものです。 Question-Answeringタスク（SQuAD）に用いることができます。 # This model is fine-tuned model for Question-Answering which is based on deberta-v2-base-japanese This model is fine-tuned by using DDQA dataset. You could use this model for Question-Answering tasks. # How to use 使い方 transformersおよびpytorch、sentencepiece、Juman++をインストールしてください。以下のコードを実行することで、Question-Answeringタスクを解かせることができます。 please execute this code. ```python import torch from transformers import AutoTokenizer, AutoModelForQuestionAnswering tokenizer = AutoTokenizer.from_pretrained('ku-nlp/deberta-v2-base-japanese') model=AutoModelForQuestionAnswering.from_pretrained('Mizuiro-sakura/deberta-v2-base-japanese-finetuned-QAe') # 学習済みモデルの読み込み text={ 'context':'私の名前はEIMIです。好きな食べ物は苺です。趣味は皆さんと会話することです。', 'question' :'好きな食べ物は何ですか' } input_ids=tokenizer.encode(text['question'],text['context']) # tokenizerで形態素解析しつつコードに変換する output= model(torch.tensor([input_ids])) # 学習済みモデルを用いて解析 prediction = tokenizer.decode(input_ids[torch.argmax(output.start_logits): torch.argmax(output.end_logits)]) # 答えに該当する部分を抜き取る print(prediction) ``` # モデルの精度 accuracy of model Exact Match(厳密一致) : 0.8038277511961722 f1 : 0.8959389668095072 # deberta-v2-base-japaneseとは？日本語Wikipedeia（3.2GB）および、cc100(85GB)、oscar(54GB)を用いて訓練されたモデルです。京都大学黒橋研究室が公表されました。 # Model description This is a Japanese DeBERTa V2 base model pre-trained on Japanese Wikipedia, the Japanese portion of CC-100, and the Japanese portion of OSCAR. # Acknowledgments 謝辞モデルを公開してくださった京都大学黒橋研究室には感謝いたします。 I would like to thank Kurohashi Lab at Kyoto University.	null	Non_BioNLP	# このモデルはdeberta-v2-base-japaneseをファインチューニングしてQAタスクに用いれるようにしたものです。このモデルはdeberta-v2-base-japaneseを運転ドメインQAデータセット（DDQA）（　https://nlp.ist.i.kyoto-u.ac.jp/index.php?Driving%20domain%20QA%20datasets　）を用いてファインチューニングしたものです。 Question-Answeringタスク（SQuAD）に用いることができます。 # This model is fine-tuned model for Question-Answering which is based on deberta-v2-base-japanese This model is fine-tuned by using DDQA dataset. You could use this model for Question-Answering tasks. # How to use 使い方 transformersおよびpytorch、sentencepiece、Juman++をインストールしてください。以下のコードを実行することで、Question-Answeringタスクを解かせることができます。 please execute this code. ```python import torch from transformers import AutoTokenizer, AutoModelForQuestionAnswering tokenizer = AutoTokenizer.from_pretrained('ku-nlp/deberta-v2-base-japanese') model=AutoModelForQuestionAnswering.from_pretrained('Mizuiro-sakura/deberta-v2-base-japanese-finetuned-QAe') # 学習済みモデルの読み込み text={ 'context':'私の名前はEIMIです。好きな食べ物は苺です。趣味は皆さんと会話することです。', 'question' :'好きな食べ物は何ですか' } input_ids=tokenizer.encode(text['question'],text['context']) # tokenizerで形態素解析しつつコードに変換する output= model(torch.tensor([input_ids])) # 学習済みモデルを用いて解析 prediction = tokenizer.decode(input_ids[torch.argmax(output.start_logits): torch.argmax(output.end_logits)]) # 答えに該当する部分を抜き取る print(prediction) ``` # モデルの精度 accuracy of model Exact Match(厳密一致) : 0.8038277511961722 f1 : 0.8959389668095072 # deberta-v2-base-japaneseとは？日本語Wikipedeia（3.2GB）および、cc100(85GB)、oscar(54GB)を用いて訓練されたモデルです。京都大学黒橋研究室が公表されました。 # Model description This is a Japanese DeBERTa V2 base model pre-trained on Japanese Wikipedia, the Japanese portion of CC-100, and the Japanese portion of OSCAR. # Acknowledgments 謝辞モデルを公開してくださった京都大学黒橋研究室には感謝いたします。 I would like to thank Kurohashi Lab at Kyoto University.	{"datasets": ["wikipedia", "cc100", "oscar"], "language": "ja", "library_name": "transformers", "license": "mit", "metrics": ["accuracy"], "tags": ["pytorch", "deberta", "deberta-v2", "question-answering", "question answering", "squad"]}	task	[ "QUESTION_ANSWERING" ]	43,097
RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf	RichardErkhov	null	[ "gguf", "endpoints_compatible", "region:us", "conversational" ]	2024-10-03T17:05:04Z	2024-10-04T01:36:03+00:00	84	0	--- {} --- Quantization made by Richard Erkhov. [Github](https://github.com/RichardErkhov) [Discord](https://discord.gg/pvy7H8DZMG) [Request more models](https://github.com/RichardErkhov/quant_request) Falcon2-5.5B-Polish - GGUF - Model creator: https://huggingface.co/ssmits/ - Original model: https://huggingface.co/ssmits/Falcon2-5.5B-Polish/ \| Name \| Quant method \| Size \| \| ---- \| ---- \| ---- \| \| [Falcon2-5.5B-Polish.Q2_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q2_K.gguf) \| Q2_K \| 2.03GB \| \| [Falcon2-5.5B-Polish.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ3_XS.gguf) \| IQ3_XS \| 2.29GB \| \| [Falcon2-5.5B-Polish.IQ3_S.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ3_S.gguf) \| IQ3_S \| 2.35GB \| \| [Falcon2-5.5B-Polish.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q3_K_S.gguf) \| Q3_K_S \| 2.35GB \| \| [Falcon2-5.5B-Polish.IQ3_M.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ3_M.gguf) \| IQ3_M \| 2.46GB \| \| [Falcon2-5.5B-Polish.Q3_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q3_K.gguf) \| Q3_K \| 2.56GB \| \| [Falcon2-5.5B-Polish.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q3_K_M.gguf) \| Q3_K_M \| 2.56GB \| \| [Falcon2-5.5B-Polish.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q3_K_L.gguf) \| Q3_K_L \| 2.72GB \| \| [Falcon2-5.5B-Polish.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ4_XS.gguf) \| IQ4_XS \| 2.87GB \| \| [Falcon2-5.5B-Polish.Q4_0.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_0.gguf) \| Q4_0 \| 2.99GB \| \| [Falcon2-5.5B-Polish.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ4_NL.gguf) \| IQ4_NL \| 3.01GB \| \| [Falcon2-5.5B-Polish.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_K_S.gguf) \| Q4_K_S \| 2.99GB \| \| [Falcon2-5.5B-Polish.Q4_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_K.gguf) \| Q4_K \| 3.19GB \| \| [Falcon2-5.5B-Polish.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_K_M.gguf) \| Q4_K_M \| 3.19GB \| \| [Falcon2-5.5B-Polish.Q4_1.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_1.gguf) \| Q4_1 \| 3.29GB \| \| [Falcon2-5.5B-Polish.Q5_0.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_0.gguf) \| Q5_0 \| 3.6GB \| \| [Falcon2-5.5B-Polish.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_K_S.gguf) \| Q5_K_S \| 3.6GB \| \| [Falcon2-5.5B-Polish.Q5_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_K.gguf) \| Q5_K \| 3.8GB \| \| [Falcon2-5.5B-Polish.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_K_M.gguf) \| Q5_K_M \| 3.8GB \| \| [Falcon2-5.5B-Polish.Q5_1.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_1.gguf) \| Q5_1 \| 3.9GB \| \| [Falcon2-5.5B-Polish.Q6_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q6_K.gguf) \| Q6_K \| 4.24GB \| \| [Falcon2-5.5B-Polish.Q8_0.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q8_0.gguf) \| Q8_0 \| 5.41GB \| Original model description: --- base_model: - tiiuae/falcon-11B library_name: transformers tags: - mergekit - merge - lazymergekit - tiiuae/falcon-11B license: apache-2.0 language: - 'pl' --- ## Why prune? Even though [Falcon-11B](https://huggingface.co/tiiuae/falcon-11B) is trained on 5T tokens, it is still undertrained, as can be seen by this graph: ![image/png](https://cdn-uploads.huggingface.co/production/uploads/660c0a02cf274b3ab77dd6b7/QeaL9bOrPskustzFpjMUP.png) This is why the choice is made to prune 50% of the layers. Note that \~1B of continued pre-training (\~1M rows of 1k tokens) is still required to restore the perplexity of this model in the desired language. I'm planning on doing that for certain languages, depending on how much compute will be available. # sliced This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit). ## Merge Details ### Merge Method This model was pruned using the passthrough merge method. ### Models Merged The following models were included in the merge: * [tiiuae/falcon-11B](https://huggingface.co/tiiuae/falcon-11B) ### Configuration The following YAML configuration was used to produce this model: ```yaml slices: - sources: - model: tiiuae/falcon-11B layer_range: [0, 24] - sources: - model: tiiuae/falcon-11B layer_range: [55, 59] merge_method: passthrough dtype: bfloat16 ``` [PruneMe](https://github.com/arcee-ai/PruneMe) has been utilized using the wikimedia/wikipedia Polish (pl) subset by investigating layer similarity with 2000 samples. The layer ranges for pruning were determined based on this analysis to maintain performance while reducing model size. ![Layer Similarity Plot](https://cdn-uploads.huggingface.co/production/uploads/660c0a02cf274b3ab77dd6b7/vjMS9FOVNYEpcE13Xm66N.png) ```python from transformers import AutoTokenizer, AutoModelForCausalLM import transformers import torch model = "ssmits/Falcon2-5.5B-Polish" tokenizer = AutoTokenizer.from_pretrained(model) pipeline = transformers.pipeline( "text-generation", model=model, tokenizer=tokenizer, torch_dtype=torch.bfloat16, ) sequences = pipeline( "Can you explain the concepts of Quantum Computing?", max_length=200, do_sample=True, top_k=10, num_return_sequences=1, eos_token_id=tokenizer.eos_token_id, ) for seq in sequences: print(f"Result: {seq['generated_text']}") ``` 💥 Falcon LLMs require PyTorch 2.0 for use with `transformers`! For fast inference with Falcon, check-out [Text Generation Inference](https://github.com/huggingface/text-generation-inference)! Read more in this [blogpost]((https://huggingface.co/blog/falcon). ## Direct Use Research on large language models; as a foundation for further specialization and finetuning for specific usecases (e.g., summarization, text generation, chatbot, etc.) ## Out-of-Scope Use Production use without adequate assessment of risks and mitigation; any use cases which may be considered irresponsible or harmful. ## Bias, Risks, and Limitations Falcon2-5.5B is trained mostly on English, but also German, Spanish, French, Italian, Portuguese, Polish, Dutch, Romanian, Czech, Swedish. It will not generalize appropriately to other languages. Furthermore, as it is trained on a large-scale corpora representative of the web, it will carry the stereotypes and biases commonly encountered online. ## Recommendations We recommend users of Falcon2-5.5B to consider finetuning it for the specific set of tasks of interest, and for guardrails and appropriate precautions to be taken for any production use.	null	Non_BioNLP	Quantization made by Richard Erkhov. [Github](https://github.com/RichardErkhov) [Discord](https://discord.gg/pvy7H8DZMG) [Request more models](https://github.com/RichardErkhov/quant_request) Falcon2-5.5B-Polish - GGUF - Model creator: https://huggingface.co/ssmits/ - Original model: https://huggingface.co/ssmits/Falcon2-5.5B-Polish/ \| Name \| Quant method \| Size \| \| ---- \| ---- \| ---- \| \| [Falcon2-5.5B-Polish.Q2_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q2_K.gguf) \| Q2_K \| 2.03GB \| \| [Falcon2-5.5B-Polish.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ3_XS.gguf) \| IQ3_XS \| 2.29GB \| \| [Falcon2-5.5B-Polish.IQ3_S.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ3_S.gguf) \| IQ3_S \| 2.35GB \| \| [Falcon2-5.5B-Polish.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q3_K_S.gguf) \| Q3_K_S \| 2.35GB \| \| [Falcon2-5.5B-Polish.IQ3_M.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ3_M.gguf) \| IQ3_M \| 2.46GB \| \| [Falcon2-5.5B-Polish.Q3_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q3_K.gguf) \| Q3_K \| 2.56GB \| \| [Falcon2-5.5B-Polish.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q3_K_M.gguf) \| Q3_K_M \| 2.56GB \| \| [Falcon2-5.5B-Polish.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q3_K_L.gguf) \| Q3_K_L \| 2.72GB \| \| [Falcon2-5.5B-Polish.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ4_XS.gguf) \| IQ4_XS \| 2.87GB \| \| [Falcon2-5.5B-Polish.Q4_0.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_0.gguf) \| Q4_0 \| 2.99GB \| \| [Falcon2-5.5B-Polish.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.IQ4_NL.gguf) \| IQ4_NL \| 3.01GB \| \| [Falcon2-5.5B-Polish.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_K_S.gguf) \| Q4_K_S \| 2.99GB \| \| [Falcon2-5.5B-Polish.Q4_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_K.gguf) \| Q4_K \| 3.19GB \| \| [Falcon2-5.5B-Polish.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_K_M.gguf) \| Q4_K_M \| 3.19GB \| \| [Falcon2-5.5B-Polish.Q4_1.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q4_1.gguf) \| Q4_1 \| 3.29GB \| \| [Falcon2-5.5B-Polish.Q5_0.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_0.gguf) \| Q5_0 \| 3.6GB \| \| [Falcon2-5.5B-Polish.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_K_S.gguf) \| Q5_K_S \| 3.6GB \| \| [Falcon2-5.5B-Polish.Q5_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_K.gguf) \| Q5_K \| 3.8GB \| \| [Falcon2-5.5B-Polish.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_K_M.gguf) \| Q5_K_M \| 3.8GB \| \| [Falcon2-5.5B-Polish.Q5_1.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q5_1.gguf) \| Q5_1 \| 3.9GB \| \| [Falcon2-5.5B-Polish.Q6_K.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q6_K.gguf) \| Q6_K \| 4.24GB \| \| [Falcon2-5.5B-Polish.Q8_0.gguf](https://huggingface.co/RichardErkhov/ssmits_-_Falcon2-5.5B-Polish-gguf/blob/main/Falcon2-5.5B-Polish.Q8_0.gguf) \| Q8_0 \| 5.41GB \| Original model description: --- base_model: - tiiuae/falcon-11B library_name: transformers tags: - mergekit - merge - lazymergekit - tiiuae/falcon-11B license: apache-2.0 language: - 'pl' --- ## Why prune? Even though [Falcon-11B](https://huggingface.co/tiiuae/falcon-11B) is trained on 5T tokens, it is still undertrained, as can be seen by this graph: ![image/png](https://cdn-uploads.huggingface.co/production/uploads/660c0a02cf274b3ab77dd6b7/QeaL9bOrPskustzFpjMUP.png) This is why the choice is made to prune 50% of the layers. Note that \~1B of continued pre-training (\~1M rows of 1k tokens) is still required to restore the perplexity of this model in the desired language. I'm planning on doing that for certain languages, depending on how much compute will be available. # sliced This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit). ## Merge Details ### Merge Method This model was pruned using the passthrough merge method. ### Models Merged The following models were included in the merge: * [tiiuae/falcon-11B](https://huggingface.co/tiiuae/falcon-11B) ### Configuration The following YAML configuration was used to produce this model: ```yaml slices: - sources: - model: tiiuae/falcon-11B layer_range: [0, 24] - sources: - model: tiiuae/falcon-11B layer_range: [55, 59] merge_method: passthrough dtype: bfloat16 ``` [PruneMe](https://github.com/arcee-ai/PruneMe) has been utilized using the wikimedia/wikipedia Polish (pl) subset by investigating layer similarity with 2000 samples. The layer ranges for pruning were determined based on this analysis to maintain performance while reducing model size. ![Layer Similarity Plot](https://cdn-uploads.huggingface.co/production/uploads/660c0a02cf274b3ab77dd6b7/vjMS9FOVNYEpcE13Xm66N.png) ```python from transformers import AutoTokenizer, AutoModelForCausalLM import transformers import torch model = "ssmits/Falcon2-5.5B-Polish" tokenizer = AutoTokenizer.from_pretrained(model) pipeline = transformers.pipeline( "text-generation", model=model, tokenizer=tokenizer, torch_dtype=torch.bfloat16, ) sequences = pipeline( "Can you explain the concepts of Quantum Computing?", max_length=200, do_sample=True, top_k=10, num_return_sequences=1, eos_token_id=tokenizer.eos_token_id, ) for seq in sequences: print(f"Result: {seq['generated_text']}") ``` 💥 Falcon LLMs require PyTorch 2.0 for use with `transformers`! For fast inference with Falcon, check-out [Text Generation Inference](https://github.com/huggingface/text-generation-inference)! Read more in this [blogpost]((https://huggingface.co/blog/falcon). ## Direct Use Research on large language models; as a foundation for further specialization and finetuning for specific usecases (e.g., summarization, text generation, chatbot, etc.) ## Out-of-Scope Use Production use without adequate assessment of risks and mitigation; any use cases which may be considered irresponsible or harmful. ## Bias, Risks, and Limitations Falcon2-5.5B is trained mostly on English, but also German, Spanish, French, Italian, Portuguese, Polish, Dutch, Romanian, Czech, Swedish. It will not generalize appropriately to other languages. Furthermore, as it is trained on a large-scale corpora representative of the web, it will carry the stereotypes and biases commonly encountered online. ## Recommendations We recommend users of Falcon2-5.5B to consider finetuning it for the specific set of tasks of interest, and for guardrails and appropriate precautions to be taken for any production use.	{}	task	[ "SUMMARIZATION" ]	43,098
yhavinga/t5-eff-xl-8l-dutch-english-cased	yhavinga	text2text-generation	[ "transformers", "jax", "t5", "text2text-generation", "seq2seq", "nl", "en", "dataset:yhavinga/mc4_nl_cleaned", "arxiv:1910.10683", "arxiv:2109.10686", "license:apache-2.0", "autotrain_compatible", "region:us" ]	2022-04-25T21:23:45Z	2022-08-07T12:07:03+00:00	18	0	--- datasets: - yhavinga/mc4_nl_cleaned language: - nl - en license: apache-2.0 tags: - t5 - seq2seq inference: false --- # t5-eff-xl-8l-dutch-english-cased A [T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) sequence to sequence model pre-trained from scratch on [cleaned Dutch 🇳🇱🇧🇪 mC4 and cleaned English 🇬🇧 C4](https://huggingface.co/datasets/yhavinga/mc4_nl_cleaned). This t5 eff model has 1240M parameters. It was pre-trained with masked language modeling (denoise token span corruption) objective on the dataset `mc4_nl_cleaned` config `large_en_nl` for 1 epoch(s) and a duration of 4d 19h, with a sequence length of 512, batch size 64 and 538k/1703705 total steps (18B tokens). Pre-training evaluation loss and accuracy are 1,3019 and 0,71. * Pre-trained T5 models need to be finetuned before they can be used for downstream tasks, therefore the inference widget on the right has been turned off. * For a demo of the Dutch CNN summarization models, head over to the Hugging Face Spaces for the [Netherformer 📰](https://huggingface.co/spaces/flax-community/netherformer) example application! Please refer to the original T5 papers and Scale Efficiently papers for more information about the T5 architecture and configs, though it must be noted that this model (t5-eff-xl-8l-dutch-english-cased) is unrelated to these projects and not an 'official' checkpoint. * [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/pdf/1910.10683.pdf) by Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, Peter J. Liu. * [Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686) by Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler. ## Tokenizer The model uses a cased SentencePiece tokenizer configured with the `Nmt, NFKC, Replace multi-space to single-space` normalizers and has 32003 tokens. It was trained on Dutch and English with scripts from the Huggingface Transformers [Flax examples](https://github.com/huggingface/transformers/tree/master/examples/flax/language-modeling). See [./raw/main/tokenizer.json](tokenizer.json) for details. ## Dataset(s) All models listed below are pre-trained on [cleaned Dutch mC4](https://huggingface.co/datasets/yhavinga/mc4_nl_cleaned), which is the original mC4, except * Documents that contained words from a selection of the Dutch and English [List of Dirty Naught Obscene and Otherwise Bad Words](https://github.com/LDNOOBW/List-of-Dirty-Naughty-Obscene-and-Otherwise-Bad-Words) are removed * Sentences with less than 3 words are removed * Sentences with a word of more than 1000 characters are removed * Documents with less than 5 sentences are removed * Documents with "javascript", "lorum ipsum", "terms of use", "privacy policy", "cookie policy", "uses cookies", "use of cookies", "use cookies", "elementen ontbreken", "deze printversie" are removed. The Dutch and English models are pre-trained on a 50/50% mix of Dutch mC4 and English C4. The translation models are fine-tuned on [CCMatrix](https://huggingface.co/datasets/yhavinga/ccmatrix). ## Dutch T5 Models Three types of [Dutch T5 models have been trained (blog)](https://huggingface.co/spaces/yhavinga/pre-training-dutch-t5-models). `t5-base-dutch` is the only model with an original T5 config. The other model types t5-v1.1 and t5-eff have `gated-relu` instead of `relu` as activation function, and trained with a drop-out of `0.0` unless training would diverge (`t5-v1.1-large-dutch-cased`). The T5-eff models are models that differ in their number of layers. The table will list the several dimensions of these models. Not all t5-eff models are efficient, the best example being the inefficient `t5-xl-4L-dutch-english-cased`. \| \| [t5-base-dutch](https://huggingface.co/yhavinga/t5-base-dutch) \| [t5-v1.1-base-dutch-uncased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-uncased) \| [t5-v1.1-base-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-cased) \| [t5-v1.1-large-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-large-dutch-cased) \| [t5-v1_1-base-dutch-english-cased](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased) \| [t5-v1_1-base-dutch-english-cased-1024](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased-1024) \| [t5-small-24L-dutch-english](https://huggingface.co/yhavinga/t5-small-24L-dutch-english) \| [t5-xl-4L-dutch-english-cased](https://huggingface.co/yhavinga/t5-xl-4L-dutch-english-cased) \| [t5-base-36L-dutch-english-cased](https://huggingface.co/yhavinga/t5-base-36L-dutch-english-cased) \| [t5-eff-xl-8l-dutch-english-cased](https://huggingface.co/yhavinga/t5-eff-xl-8l-dutch-english-cased) \| [t5-eff-large-8l-dutch-english-cased](https://huggingface.co/yhavinga/t5-eff-large-8l-dutch-english-cased) \| \|:------------------\|:----------------\|:-----------------------------\|:---------------------------\|:----------------------------\|:-----------------------------------\|:----------------------------------------\|:-----------------------------\|:-------------------------------\|:----------------------------------\|:-----------------------------------\|:--------------------------------------\| \| type \| t5 \| t5-v1.1 \| t5-v1.1 \| t5-v1.1 \| t5-v1.1 \| t5-v1.1 \| t5 eff \| t5 eff \| t5 eff \| t5 eff \| t5 eff \| \| d_model \| 768 \| 768 \| 768 \| 1024 \| 768 \| 768 \| 512 \| 2048 \| 768 \| 1024 \| 1024 \| \| d_ff \| 3072 \| 2048 \| 2048 \| 2816 \| 2048 \| 2048 \| 1920 \| 5120 \| 2560 \| 16384 \| 4096 \| \| num_heads \| 12 \| 12 \| 12 \| 16 \| 12 \| 12 \| 8 \| 32 \| 12 \| 32 \| 16 \| \| d_kv \| 64 \| 64 \| 64 \| 64 \| 64 \| 64 \| 64 \| 64 \| 64 \| 128 \| 64 \| \| num_layers \| 12 \| 12 \| 12 \| 24 \| 12 \| 12 \| 24 \| 4 \| 36 \| 8 \| 8 \| \| num parameters \| 223M \| 248M \| 248M \| 783M \| 248M \| 248M \| 250M \| 585M \| 729M \| 1241M \| 335M \| \| feed_forward_proj \| relu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| \| dropout \| 0.1 \| 0.0 \| 0.0 \| 0.1 \| 0.0 \| 0.0 \| 0.0 \| 0.1 \| 0.0 \| 0.0 \| 0.0 \| \| dataset \| mc4_nl_cleaned \| mc4_nl_cleaned full \| mc4_nl_cleaned full \| mc4_nl_cleaned \| mc4_nl_cleaned small_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| \| tr. seq len \| 512 \| 1024 \| 1024 \| 512 \| 512 \| 1024 \| 512 \| 512 \| 512 \| 512 \| 512 \| \| batch size \| 128 \| 64 \| 64 \| 64 \| 128 \| 64 \| 128 \| 512 \| 512 \| 64 \| 128 \| \| total steps \| 527500 \| 1014525 \| 1210154 \| 1120k/2427498 \| 2839630 \| 1520k/3397024 \| 851852 \| 212963 \| 212963 \| 538k/1703705 \| 851850 \| \| epochs \| 1 \| 2 \| 2 \| 2 \| 10 \| 4 \| 1 \| 1 \| 1 \| 1 \| 1 \| \| duration \| 2d9h \| 5d5h \| 6d6h \| 8d13h \| 11d18h \| 9d1h \| 4d10h \| 6d1h \| 17d15h \| 4d 19h \| 3d 23h \| \| optimizer \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| \| lr \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.009 \| 0.005 \| 0.005 \| \| warmup \| 10000.0 \| 10000.0 \| 10000.0 \| 10000.0 \| 10000.0 \| 5000.0 \| 20000.0 \| 2500.0 \| 1000.0 \| 1500.0 \| 1500.0 \| \| eval loss \| 1,38 \| 1,20 \| 0,96 \| 1,07 \| 1,11 \| 1,13 \| 1,18 \| 1,27 \| 1,05 \| 1,3019 \| 1,15 \| \| eval acc \| 0,70 \| 0,73 \| 0,78 \| 0,76 \| 0,75 \| 0,74 \| 0,74 \| 0,72 \| 0,76 \| 0,71 \| 0,74 \| ## Evaluation Most models from the list above have been fine-tuned for summarization and translation. The figure below shows the evaluation scores, where the x-axis shows the translation Bleu score (higher is better) and y-axis the summarization Rouge1 translation score (higher is better). Point size is proportional to the model size. Models with faster inference speed are green, slower inference speed is plotted as bleu. ![Evaluation T5 Dutch English](evaluation_t5_dutch_english.png) Evaluation was run on fine-tuned models trained with the following settings: \| \| Summarization \| Translation \| \|---------------:\|------------------\|-------------------\| \| Dataset \| CNN Dailymail NL \| CCMatrix en -> nl \| \| #train samples \| 50K \| 50K \| \| Optimizer \| Adam \| Adam \| \| learning rate \| 0.001 \| 0.0005 \| \| source length \| 1024 \| 128 \| \| target length \| 142 \| 128 \| \|label smoothing \| 0.05 \| 0.1 \| \| #eval samples \| 1000 \| 1000 \| Note that the amount of training data is limited to a fraction of the total dataset sizes, therefore the scores below can only be used to compare the 'transfer-learning' strength. The fine-tuned checkpoints for this evaluation are not saved, since they were trained for comparison of pre-trained models only. The numbers for summarization are the Rouge scores on 1000 documents from the test split. \| \| [t5-base-dutch](https://huggingface.co/yhavinga/t5-base-dutch) \| [t5-v1.1-base-dutch-uncased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-uncased) \| [t5-v1.1-base-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-cased) \| [t5-v1_1-base-dutch-english-cased](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased) \| [t5-v1_1-base-dutch-english-cased-1024](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased-1024) \| [t5-small-24L-dutch-english](https://huggingface.co/yhavinga/t5-small-24L-dutch-english) \| [t5-xl-4L-dutch-english-cased](https://huggingface.co/yhavinga/t5-xl-4L-dutch-english-cased) \| [t5-base-36L-dutch-english-cased](https://huggingface.co/yhavinga/t5-base-36L-dutch-english-cased) \| [t5-eff-large-8l-dutch-english-cased](https://huggingface.co/yhavinga/t5-eff-large-8l-dutch-english-cased) \| mt5-base \| \|:------------------------\|----------------:\|-----------------------------:\|---------------------------:\|-----------------------------------:\|----------------------------------------:\|-----------------------------:\|-------------------------------:\|----------------------------------:\|--------------------------------------:\|-----------:\| \| rouge1 \| 33.38 \| 33.97 \| 34.39 \| 33.38 \| 34.97 \| 34.38 \| 30.35 \| 35.04 \| 34.04 \| 33.25 \| \| rouge2 \| 13.32 \| 13.85 \| 13.98 \| 13.47 \| 14.01 \| 13.89 \| 11.57 \| 14.23 \| 13.76 \| 12.74 \| \| rougeL \| 24.22 \| 24.72 \| 25.1 \| 24.34 \| 24.99 \| 25.25 \| 22.69 \| 25.05 \| 24.75 \| 23.5 \| \| rougeLsum \| 30.23 \| 30.9 \| 31.44 \| 30.51 \| 32.01 \| 31.38 \| 27.5 \| 32.12 \| 31.12 \| 30.15 \| \| samples_per_second \| 3.18 \| 3.02 \| 2.99 \| 3.22 \| 2.97 \| 1.57 \| 2.8 \| 0.61 \| 3.27 \| 1.22 \| The models below have been evaluated for English to Dutch translation. Note that the first four models are pre-trained on Dutch only. That they still perform adequate is probably because the translation direction is English to Dutch. The numbers reported are the Bleu scores on 1000 documents from the test split. \| \| [t5-base-dutch](https://huggingface.co/yhavinga/t5-base-dutch) \| [t5-v1.1-base-dutch-uncased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-uncased) \| [t5-v1.1-base-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-cased) \| [t5-v1.1-large-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-large-dutch-cased) \| [t5-v1_1-base-dutch-english-cased](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased) \| [t5-v1_1-base-dutch-english-cased-1024](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased-1024) \| [t5-small-24L-dutch-english](https://huggingface.co/yhavinga/t5-small-24L-dutch-english) \| [t5-xl-4L-dutch-english-cased](https://huggingface.co/yhavinga/t5-xl-4L-dutch-english-cased) \| [t5-base-36L-dutch-english-cased](https://huggingface.co/yhavinga/t5-base-36L-dutch-english-cased) \| [t5-eff-large-8l-dutch-english-cased](https://huggingface.co/yhavinga/t5-eff-large-8l-dutch-english-cased) \| mt5-base \| \|:-------------------------------\|----------------:\|-----------------------------:\|---------------------------:\|----------------------------:\|-----------------------------------:\|----------------------------------------:\|-----------------------------:\|-------------------------------:\|----------------------------------:\|--------------------------------------:\|-----------:\| \| precision_ng1 \| 74.17 \| 78.09 \| 77.08 \| 72.12 \| 77.19 \| 78.76 \| 78.59 \| 77.3 \| 79.75 \| 78.88 \| 73.47 \| \| precision_ng2 \| 52.42 \| 57.52 \| 55.31 \| 48.7 \| 55.39 \| 58.01 \| 57.83 \| 55.27 \| 59.89 \| 58.27 \| 50.12 \| \| precision_ng3 \| 39.55 \| 45.2 \| 42.54 \| 35.54 \| 42.25 \| 45.13 \| 45.02 \| 42.06 \| 47.4 \| 45.95 \| 36.59 \| \| precision_ng4 \| 30.23 \| 36.04 \| 33.26 \| 26.27 \| 32.74 \| 35.72 \| 35.41 \| 32.61 \| 38.1 \| 36.91 \| 27.26 \| \| bp \| 0.99 \| 0.98 \| 0.97 \| 0.98 \| 0.98 \| 0.98 \| 0.98 \| 0.97 \| 0.98 \| 0.98 \| 0.98 \| \| score \| 45.88 \| 51.21 \| 48.31 \| 41.59 \| 48.17 \| 51.31 \| 50.82 \| 47.83 \| 53 \| 51.79 \| 42.74 \| \| samples_per_second \| 45.19 \| 45.05 \| 38.67 \| 10.12 \| 42.19 \| 42.61 \| 12.85 \| 33.74 \| 9.07 \| 37.86 \| 9.03 \| ## Translation models The models `t5-small-24L-dutch-english` and `t5-base-36L-dutch-english` have been fine-tuned for both language directions on the first 25M samples from CCMatrix, giving a total of 50M training samples. Evaluation is performed on out-of-sample CCMatrix and also on Tatoeba and Opus Books. The `_bp` columns list the brevity penalty. The `avg_bleu` score is the bleu score averaged over all three evaluation datasets. The best scores displayed in bold for both translation directions. \| \| [t5-base-36L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-base-36L-ccmatrix-multi) \| [t5-base-36L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-base-36L-ccmatrix-multi) \| [t5-small-24L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-small-24L-ccmatrix-multi) \| [t5-small-24L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-small-24L-ccmatrix-multi) \| \|:-----------------------\|:-----------------------------\|:-----------------------------\|:------------------------------\|:------------------------------\| \| source_lang \| en \| nl \| en \| nl \| \| target_lang \| nl \| en \| nl \| en \| \| source_prefix \| translate English to Dutch: \| translate Dutch to English: \| translate English to Dutch: \| translate Dutch to English: \| \| ccmatrix_bleu \| 56.8 \| 62.8 \| 57.4 \| 63.1 \| \| tatoeba_bleu \| 46.6 \| 52.8 \| 46.4 \| 51.7 \| \| opus_books_bleu \| 13.5 \| 24.9 \| 12.9 \| 23.4 \| \| ccmatrix_bp \| 0.95 \| 0.96 \| 0.95 \| 0.96 \| \| tatoeba_bp \| 0.97 \| 0.94 \| 0.98 \| 0.94 \| \| opus_books_bp \| 0.8 \| 0.94 \| 0.77 \| 0.89 \| \| avg_bleu \| 38.96 \| 46.86 \| 38.92 \| 46.06 \| \| max_source_length \| 128 \| 128 \| 128 \| 128 \| \| max_target_length \| 128 \| 128 \| 128 \| 128 \| \| adam_beta1 \| 0.9 \| 0.9 \| 0.9 \| 0.9 \| \| adam_beta2 \| 0.997 \| 0.997 \| 0.997 \| 0.997 \| \| weight_decay \| 0.05 \| 0.05 \| 0.002 \| 0.002 \| \| lr \| 5e-05 \| 5e-05 \| 0.0005 \| 0.0005 \| \| label_smoothing_factor \| 0.15 \| 0.15 \| 0.1 \| 0.1 \| \| train_batch_size \| 128 \| 128 \| 128 \| 128 \| \| warmup_steps \| 2000 \| 2000 \| 2000 \| 2000 \| \| total steps \| 390625 \| 390625 \| 390625 \| 390625 \| \| duration \| 4d 5h \| 4d 5h \| 3d 2h \| 3d 2h \| \| num parameters \| 729M \| 729M \| 250M \| 250M \| ## Acknowledgements This project would not have been possible without compute generously provided by Google through the [TPU Research Cloud](https://sites.research.google/trc/). The HuggingFace 🤗 ecosystem was instrumental in all parts of the training. Weights & Biases made it possible to keep track of many training sessions and orchestrate hyper-parameter sweeps with insightful visualizations. The following repositories where helpful in setting up the TPU-VM, and getting an idea what sensible hyper-parameters are for training gpt2 from scratch: * [Gsarti's Pretrain and Fine-tune a T5 model with Flax on GCP](https://github.com/gsarti/t5-flax-gcp) * [Flax/Jax Community week t5-base-dutch](https://huggingface.co/flax-community/t5-base-dutch) Created by [Yeb Havinga](https://www.linkedin.com/in/yeb-havinga-86530825/)	null	Non_BioNLP	# t5-eff-xl-8l-dutch-english-cased A [T5](https://ai.googleblog.com/2020/02/exploring-transfer-learning-with-t5.html) sequence to sequence model pre-trained from scratch on [cleaned Dutch 🇳🇱🇧🇪 mC4 and cleaned English 🇬🇧 C4](https://huggingface.co/datasets/yhavinga/mc4_nl_cleaned). This t5 eff model has 1240M parameters. It was pre-trained with masked language modeling (denoise token span corruption) objective on the dataset `mc4_nl_cleaned` config `large_en_nl` for 1 epoch(s) and a duration of 4d 19h, with a sequence length of 512, batch size 64 and 538k/1703705 total steps (18B tokens). Pre-training evaluation loss and accuracy are 1,3019 and 0,71. * Pre-trained T5 models need to be finetuned before they can be used for downstream tasks, therefore the inference widget on the right has been turned off. * For a demo of the Dutch CNN summarization models, head over to the Hugging Face Spaces for the [Netherformer 📰](https://huggingface.co/spaces/flax-community/netherformer) example application! Please refer to the original T5 papers and Scale Efficiently papers for more information about the T5 architecture and configs, though it must be noted that this model (t5-eff-xl-8l-dutch-english-cased) is unrelated to these projects and not an 'official' checkpoint. * [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/pdf/1910.10683.pdf) by Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, Peter J. Liu. * [Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers](https://arxiv.org/abs/2109.10686) by Yi Tay, Mostafa Dehghani, Jinfeng Rao, William Fedus, Samira Abnar, Hyung Won Chung, Sharan Narang, Dani Yogatama, Ashish Vaswani, Donald Metzler. ## Tokenizer The model uses a cased SentencePiece tokenizer configured with the `Nmt, NFKC, Replace multi-space to single-space` normalizers and has 32003 tokens. It was trained on Dutch and English with scripts from the Huggingface Transformers [Flax examples](https://github.com/huggingface/transformers/tree/master/examples/flax/language-modeling). See [./raw/main/tokenizer.json](tokenizer.json) for details. ## Dataset(s) All models listed below are pre-trained on [cleaned Dutch mC4](https://huggingface.co/datasets/yhavinga/mc4_nl_cleaned), which is the original mC4, except * Documents that contained words from a selection of the Dutch and English [List of Dirty Naught Obscene and Otherwise Bad Words](https://github.com/LDNOOBW/List-of-Dirty-Naughty-Obscene-and-Otherwise-Bad-Words) are removed * Sentences with less than 3 words are removed * Sentences with a word of more than 1000 characters are removed * Documents with less than 5 sentences are removed * Documents with "javascript", "lorum ipsum", "terms of use", "privacy policy", "cookie policy", "uses cookies", "use of cookies", "use cookies", "elementen ontbreken", "deze printversie" are removed. The Dutch and English models are pre-trained on a 50/50% mix of Dutch mC4 and English C4. The translation models are fine-tuned on [CCMatrix](https://huggingface.co/datasets/yhavinga/ccmatrix). ## Dutch T5 Models Three types of [Dutch T5 models have been trained (blog)](https://huggingface.co/spaces/yhavinga/pre-training-dutch-t5-models). `t5-base-dutch` is the only model with an original T5 config. The other model types t5-v1.1 and t5-eff have `gated-relu` instead of `relu` as activation function, and trained with a drop-out of `0.0` unless training would diverge (`t5-v1.1-large-dutch-cased`). The T5-eff models are models that differ in their number of layers. The table will list the several dimensions of these models. Not all t5-eff models are efficient, the best example being the inefficient `t5-xl-4L-dutch-english-cased`. \| \| [t5-base-dutch](https://huggingface.co/yhavinga/t5-base-dutch) \| [t5-v1.1-base-dutch-uncased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-uncased) \| [t5-v1.1-base-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-cased) \| [t5-v1.1-large-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-large-dutch-cased) \| [t5-v1_1-base-dutch-english-cased](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased) \| [t5-v1_1-base-dutch-english-cased-1024](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased-1024) \| [t5-small-24L-dutch-english](https://huggingface.co/yhavinga/t5-small-24L-dutch-english) \| [t5-xl-4L-dutch-english-cased](https://huggingface.co/yhavinga/t5-xl-4L-dutch-english-cased) \| [t5-base-36L-dutch-english-cased](https://huggingface.co/yhavinga/t5-base-36L-dutch-english-cased) \| [t5-eff-xl-8l-dutch-english-cased](https://huggingface.co/yhavinga/t5-eff-xl-8l-dutch-english-cased) \| [t5-eff-large-8l-dutch-english-cased](https://huggingface.co/yhavinga/t5-eff-large-8l-dutch-english-cased) \| \|:------------------\|:----------------\|:-----------------------------\|:---------------------------\|:----------------------------\|:-----------------------------------\|:----------------------------------------\|:-----------------------------\|:-------------------------------\|:----------------------------------\|:-----------------------------------\|:--------------------------------------\| \| type \| t5 \| t5-v1.1 \| t5-v1.1 \| t5-v1.1 \| t5-v1.1 \| t5-v1.1 \| t5 eff \| t5 eff \| t5 eff \| t5 eff \| t5 eff \| \| d_model \| 768 \| 768 \| 768 \| 1024 \| 768 \| 768 \| 512 \| 2048 \| 768 \| 1024 \| 1024 \| \| d_ff \| 3072 \| 2048 \| 2048 \| 2816 \| 2048 \| 2048 \| 1920 \| 5120 \| 2560 \| 16384 \| 4096 \| \| num_heads \| 12 \| 12 \| 12 \| 16 \| 12 \| 12 \| 8 \| 32 \| 12 \| 32 \| 16 \| \| d_kv \| 64 \| 64 \| 64 \| 64 \| 64 \| 64 \| 64 \| 64 \| 64 \| 128 \| 64 \| \| num_layers \| 12 \| 12 \| 12 \| 24 \| 12 \| 12 \| 24 \| 4 \| 36 \| 8 \| 8 \| \| num parameters \| 223M \| 248M \| 248M \| 783M \| 248M \| 248M \| 250M \| 585M \| 729M \| 1241M \| 335M \| \| feed_forward_proj \| relu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| gated-gelu \| \| dropout \| 0.1 \| 0.0 \| 0.0 \| 0.1 \| 0.0 \| 0.0 \| 0.0 \| 0.1 \| 0.0 \| 0.0 \| 0.0 \| \| dataset \| mc4_nl_cleaned \| mc4_nl_cleaned full \| mc4_nl_cleaned full \| mc4_nl_cleaned \| mc4_nl_cleaned small_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| mc4_nl_cleaned large_en_nl \| \| tr. seq len \| 512 \| 1024 \| 1024 \| 512 \| 512 \| 1024 \| 512 \| 512 \| 512 \| 512 \| 512 \| \| batch size \| 128 \| 64 \| 64 \| 64 \| 128 \| 64 \| 128 \| 512 \| 512 \| 64 \| 128 \| \| total steps \| 527500 \| 1014525 \| 1210154 \| 1120k/2427498 \| 2839630 \| 1520k/3397024 \| 851852 \| 212963 \| 212963 \| 538k/1703705 \| 851850 \| \| epochs \| 1 \| 2 \| 2 \| 2 \| 10 \| 4 \| 1 \| 1 \| 1 \| 1 \| 1 \| \| duration \| 2d9h \| 5d5h \| 6d6h \| 8d13h \| 11d18h \| 9d1h \| 4d10h \| 6d1h \| 17d15h \| 4d 19h \| 3d 23h \| \| optimizer \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| adafactor \| \| lr \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.005 \| 0.009 \| 0.005 \| 0.005 \| \| warmup \| 10000.0 \| 10000.0 \| 10000.0 \| 10000.0 \| 10000.0 \| 5000.0 \| 20000.0 \| 2500.0 \| 1000.0 \| 1500.0 \| 1500.0 \| \| eval loss \| 1,38 \| 1,20 \| 0,96 \| 1,07 \| 1,11 \| 1,13 \| 1,18 \| 1,27 \| 1,05 \| 1,3019 \| 1,15 \| \| eval acc \| 0,70 \| 0,73 \| 0,78 \| 0,76 \| 0,75 \| 0,74 \| 0,74 \| 0,72 \| 0,76 \| 0,71 \| 0,74 \| ## Evaluation Most models from the list above have been fine-tuned for summarization and translation. The figure below shows the evaluation scores, where the x-axis shows the translation Bleu score (higher is better) and y-axis the summarization Rouge1 translation score (higher is better). Point size is proportional to the model size. Models with faster inference speed are green, slower inference speed is plotted as bleu. ![Evaluation T5 Dutch English](evaluation_t5_dutch_english.png) Evaluation was run on fine-tuned models trained with the following settings: \| \| Summarization \| Translation \| \|---------------:\|------------------\|-------------------\| \| Dataset \| CNN Dailymail NL \| CCMatrix en -> nl \| \| #train samples \| 50K \| 50K \| \| Optimizer \| Adam \| Adam \| \| learning rate \| 0.001 \| 0.0005 \| \| source length \| 1024 \| 128 \| \| target length \| 142 \| 128 \| \|label smoothing \| 0.05 \| 0.1 \| \| #eval samples \| 1000 \| 1000 \| Note that the amount of training data is limited to a fraction of the total dataset sizes, therefore the scores below can only be used to compare the 'transfer-learning' strength. The fine-tuned checkpoints for this evaluation are not saved, since they were trained for comparison of pre-trained models only. The numbers for summarization are the Rouge scores on 1000 documents from the test split. \| \| [t5-base-dutch](https://huggingface.co/yhavinga/t5-base-dutch) \| [t5-v1.1-base-dutch-uncased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-uncased) \| [t5-v1.1-base-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-cased) \| [t5-v1_1-base-dutch-english-cased](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased) \| [t5-v1_1-base-dutch-english-cased-1024](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased-1024) \| [t5-small-24L-dutch-english](https://huggingface.co/yhavinga/t5-small-24L-dutch-english) \| [t5-xl-4L-dutch-english-cased](https://huggingface.co/yhavinga/t5-xl-4L-dutch-english-cased) \| [t5-base-36L-dutch-english-cased](https://huggingface.co/yhavinga/t5-base-36L-dutch-english-cased) \| [t5-eff-large-8l-dutch-english-cased](https://huggingface.co/yhavinga/t5-eff-large-8l-dutch-english-cased) \| mt5-base \| \|:------------------------\|----------------:\|-----------------------------:\|---------------------------:\|-----------------------------------:\|----------------------------------------:\|-----------------------------:\|-------------------------------:\|----------------------------------:\|--------------------------------------:\|-----------:\| \| rouge1 \| 33.38 \| 33.97 \| 34.39 \| 33.38 \| 34.97 \| 34.38 \| 30.35 \| 35.04 \| 34.04 \| 33.25 \| \| rouge2 \| 13.32 \| 13.85 \| 13.98 \| 13.47 \| 14.01 \| 13.89 \| 11.57 \| 14.23 \| 13.76 \| 12.74 \| \| rougeL \| 24.22 \| 24.72 \| 25.1 \| 24.34 \| 24.99 \| 25.25 \| 22.69 \| 25.05 \| 24.75 \| 23.5 \| \| rougeLsum \| 30.23 \| 30.9 \| 31.44 \| 30.51 \| 32.01 \| 31.38 \| 27.5 \| 32.12 \| 31.12 \| 30.15 \| \| samples_per_second \| 3.18 \| 3.02 \| 2.99 \| 3.22 \| 2.97 \| 1.57 \| 2.8 \| 0.61 \| 3.27 \| 1.22 \| The models below have been evaluated for English to Dutch translation. Note that the first four models are pre-trained on Dutch only. That they still perform adequate is probably because the translation direction is English to Dutch. The numbers reported are the Bleu scores on 1000 documents from the test split. \| \| [t5-base-dutch](https://huggingface.co/yhavinga/t5-base-dutch) \| [t5-v1.1-base-dutch-uncased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-uncased) \| [t5-v1.1-base-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-base-dutch-cased) \| [t5-v1.1-large-dutch-cased](https://huggingface.co/yhavinga/t5-v1.1-large-dutch-cased) \| [t5-v1_1-base-dutch-english-cased](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased) \| [t5-v1_1-base-dutch-english-cased-1024](https://huggingface.co/yhavinga/t5-v1_1-base-dutch-english-cased-1024) \| [t5-small-24L-dutch-english](https://huggingface.co/yhavinga/t5-small-24L-dutch-english) \| [t5-xl-4L-dutch-english-cased](https://huggingface.co/yhavinga/t5-xl-4L-dutch-english-cased) \| [t5-base-36L-dutch-english-cased](https://huggingface.co/yhavinga/t5-base-36L-dutch-english-cased) \| [t5-eff-large-8l-dutch-english-cased](https://huggingface.co/yhavinga/t5-eff-large-8l-dutch-english-cased) \| mt5-base \| \|:-------------------------------\|----------------:\|-----------------------------:\|---------------------------:\|----------------------------:\|-----------------------------------:\|----------------------------------------:\|-----------------------------:\|-------------------------------:\|----------------------------------:\|--------------------------------------:\|-----------:\| \| precision_ng1 \| 74.17 \| 78.09 \| 77.08 \| 72.12 \| 77.19 \| 78.76 \| 78.59 \| 77.3 \| 79.75 \| 78.88 \| 73.47 \| \| precision_ng2 \| 52.42 \| 57.52 \| 55.31 \| 48.7 \| 55.39 \| 58.01 \| 57.83 \| 55.27 \| 59.89 \| 58.27 \| 50.12 \| \| precision_ng3 \| 39.55 \| 45.2 \| 42.54 \| 35.54 \| 42.25 \| 45.13 \| 45.02 \| 42.06 \| 47.4 \| 45.95 \| 36.59 \| \| precision_ng4 \| 30.23 \| 36.04 \| 33.26 \| 26.27 \| 32.74 \| 35.72 \| 35.41 \| 32.61 \| 38.1 \| 36.91 \| 27.26 \| \| bp \| 0.99 \| 0.98 \| 0.97 \| 0.98 \| 0.98 \| 0.98 \| 0.98 \| 0.97 \| 0.98 \| 0.98 \| 0.98 \| \| score \| 45.88 \| 51.21 \| 48.31 \| 41.59 \| 48.17 \| 51.31 \| 50.82 \| 47.83 \| 53 \| 51.79 \| 42.74 \| \| samples_per_second \| 45.19 \| 45.05 \| 38.67 \| 10.12 \| 42.19 \| 42.61 \| 12.85 \| 33.74 \| 9.07 \| 37.86 \| 9.03 \| ## Translation models The models `t5-small-24L-dutch-english` and `t5-base-36L-dutch-english` have been fine-tuned for both language directions on the first 25M samples from CCMatrix, giving a total of 50M training samples. Evaluation is performed on out-of-sample CCMatrix and also on Tatoeba and Opus Books. The `_bp` columns list the brevity penalty. The `avg_bleu` score is the bleu score averaged over all three evaluation datasets. The best scores displayed in bold for both translation directions. \| \| [t5-base-36L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-base-36L-ccmatrix-multi) \| [t5-base-36L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-base-36L-ccmatrix-multi) \| [t5-small-24L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-small-24L-ccmatrix-multi) \| [t5-small-24L-ccmatrix-multi](https://huggingface.co/yhavinga/t5-small-24L-ccmatrix-multi) \| \|:-----------------------\|:-----------------------------\|:-----------------------------\|:------------------------------\|:------------------------------\| \| source_lang \| en \| nl \| en \| nl \| \| target_lang \| nl \| en \| nl \| en \| \| source_prefix \| translate English to Dutch: \| translate Dutch to English: \| translate English to Dutch: \| translate Dutch to English: \| \| ccmatrix_bleu \| 56.8 \| 62.8 \| 57.4 \| 63.1 \| \| tatoeba_bleu \| 46.6 \| 52.8 \| 46.4 \| 51.7 \| \| opus_books_bleu \| 13.5 \| 24.9 \| 12.9 \| 23.4 \| \| ccmatrix_bp \| 0.95 \| 0.96 \| 0.95 \| 0.96 \| \| tatoeba_bp \| 0.97 \| 0.94 \| 0.98 \| 0.94 \| \| opus_books_bp \| 0.8 \| 0.94 \| 0.77 \| 0.89 \| \| avg_bleu \| 38.96 \| 46.86 \| 38.92 \| 46.06 \| \| max_source_length \| 128 \| 128 \| 128 \| 128 \| \| max_target_length \| 128 \| 128 \| 128 \| 128 \| \| adam_beta1 \| 0.9 \| 0.9 \| 0.9 \| 0.9 \| \| adam_beta2 \| 0.997 \| 0.997 \| 0.997 \| 0.997 \| \| weight_decay \| 0.05 \| 0.05 \| 0.002 \| 0.002 \| \| lr \| 5e-05 \| 5e-05 \| 0.0005 \| 0.0005 \| \| label_smoothing_factor \| 0.15 \| 0.15 \| 0.1 \| 0.1 \| \| train_batch_size \| 128 \| 128 \| 128 \| 128 \| \| warmup_steps \| 2000 \| 2000 \| 2000 \| 2000 \| \| total steps \| 390625 \| 390625 \| 390625 \| 390625 \| \| duration \| 4d 5h \| 4d 5h \| 3d 2h \| 3d 2h \| \| num parameters \| 729M \| 729M \| 250M \| 250M \| ## Acknowledgements This project would not have been possible without compute generously provided by Google through the [TPU Research Cloud](https://sites.research.google/trc/). The HuggingFace 🤗 ecosystem was instrumental in all parts of the training. Weights & Biases made it possible to keep track of many training sessions and orchestrate hyper-parameter sweeps with insightful visualizations. The following repositories where helpful in setting up the TPU-VM, and getting an idea what sensible hyper-parameters are for training gpt2 from scratch: * [Gsarti's Pretrain and Fine-tune a T5 model with Flax on GCP](https://github.com/gsarti/t5-flax-gcp) * [Flax/Jax Community week t5-base-dutch](https://huggingface.co/flax-community/t5-base-dutch) Created by [Yeb Havinga](https://www.linkedin.com/in/yeb-havinga-86530825/)	{"datasets": ["yhavinga/mc4_nl_cleaned"], "language": ["nl", "en"], "license": "apache-2.0", "tags": ["t5", "seq2seq"], "inference": false}	task	[ "TRANSLATION", "SUMMARIZATION" ]	43,099
YakovElm/Apache20SetFitModel_balance_ratio_1	YakovElm	text-classification	[ "sentence-transformers", "pytorch", "mpnet", "setfit", "text-classification", "arxiv:2209.11055", "license:apache-2.0", "region:us" ]	2023-06-01T05:30:47Z	2023-06-01T05:31:23+00:00	8	0	--- license: apache-2.0 pipeline_tag: text-classification tags: - setfit - sentence-transformers - text-classification --- # YakovElm/Apache20SetFitModel_balance_ratio_1 This is a [SetFit model](https://github.com/huggingface/setfit) that can be used for text classification. The model has been trained using an efficient few-shot learning technique that involves: 1. Fine-tuning a [Sentence Transformer](https://www.sbert.net) with contrastive learning. 2. Training a classification head with features from the fine-tuned Sentence Transformer. ## Usage To use this model for inference, first install the SetFit library: ```bash python -m pip install setfit ``` You can then run inference as follows: ```python from setfit import SetFitModel # Download from Hub and run inference model = SetFitModel.from_pretrained("YakovElm/Apache20SetFitModel_balance_ratio_1") # Run inference preds = model(["i loved the spiderman movie!", "pineapple on pizza is the worst 🤮"]) ``` ## BibTeX entry and citation info ```bibtex @article{https://doi.org/10.48550/arxiv.2209.11055, doi = {10.48550/ARXIV.2209.11055}, url = {https://arxiv.org/abs/2209.11055}, author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren}, keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences}, title = {Efficient Few-Shot Learning Without Prompts}, publisher = {arXiv}, year = {2022}, copyright = {Creative Commons Attribution 4.0 International} } ```	null	Non_BioNLP	# YakovElm/Apache20SetFitModel_balance_ratio_1 This is a [SetFit model](https://github.com/huggingface/setfit) that can be used for text classification. The model has been trained using an efficient few-shot learning technique that involves: 1. Fine-tuning a [Sentence Transformer](https://www.sbert.net) with contrastive learning. 2. Training a classification head with features from the fine-tuned Sentence Transformer. ## Usage To use this model for inference, first install the SetFit library: ```bash python -m pip install setfit ``` You can then run inference as follows: ```python from setfit import SetFitModel # Download from Hub and run inference model = SetFitModel.from_pretrained("YakovElm/Apache20SetFitModel_balance_ratio_1") # Run inference preds = model(["i loved the spiderman movie!", "pineapple on pizza is the worst 🤮"]) ``` ## BibTeX entry and citation info ```bibtex @article{https://doi.org/10.48550/arxiv.2209.11055, doi = {10.48550/ARXIV.2209.11055}, url = {https://arxiv.org/abs/2209.11055}, author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren}, keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences}, title = {Efficient Few-Shot Learning Without Prompts}, publisher = {arXiv}, year = {2022}, copyright = {Creative Commons Attribution 4.0 International} } ```	{"license": "apache-2.0", "pipeline_tag": "text-classification", "tags": ["setfit", "sentence-transformers", "text-classification"]}	task	[ "TEXT_CLASSIFICATION" ]	43,100
Ibrahim-Alam/finetuning-xlnet-base-cased-on-sst2	Ibrahim-Alam	text-classification	[ "transformers", "pytorch", "tensorboard", "xlnet", "text-classification", "generated_from_trainer", "dataset:sst2", "license:mit", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-05-31T16:48:05Z	2023-05-31T17:00:50+00:00	16	0	--- datasets: - sst2 license: mit metrics: - accuracy - f1 tags: - generated_from_trainer model-index: - name: finetuning-xlnet-base-cased-on-sst2 results: - task: type: text-classification name: Text Classification dataset: name: sst2 type: sst2 config: default split: validation args: default metrics: - type: accuracy value: 0.9357798165137615 name: Accuracy - type: f1 value: 0.9373601789709173 name: F1 --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # finetuning-xlnet-base-cased-on-sst2 This model is a fine-tuned version of [xlnet-base-cased](https://huggingface.co/xlnet-base-cased) on the sst2 dataset. It achieves the following results on the evaluation set: - Loss: 0.2275 - Accuracy: 0.9358 - F1: 0.9374 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 1 ### Training results ### Framework versions - Transformers 4.28.0 - Pytorch 2.0.1+cu118 - Datasets 2.12.0 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # finetuning-xlnet-base-cased-on-sst2 This model is a fine-tuned version of [xlnet-base-cased](https://huggingface.co/xlnet-base-cased) on the sst2 dataset. It achieves the following results on the evaluation set: - Loss: 0.2275 - Accuracy: 0.9358 - F1: 0.9374 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 1 ### Training results ### Framework versions - Transformers 4.28.0 - Pytorch 2.0.1+cu118 - Datasets 2.12.0 - Tokenizers 0.13.3	{"datasets": ["sst2"], "license": "mit", "metrics": ["accuracy", "f1"], "tags": ["generated_from_trainer"], "model-index": [{"name": "finetuning-xlnet-base-cased-on-sst2", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "sst2", "type": "sst2", "config": "default", "split": "validation", "args": "default"}, "metrics": [{"type": "accuracy", "value": 0.9357798165137615, "name": "Accuracy"}, {"type": "f1", "value": 0.9373601789709173, "name": "F1"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,101
Lwhieldon/distilbert-base-uncased-finetuned-emotion	Lwhieldon	text-classification	[ "transformers", "pytorch", "distilbert", "text-classification", "generated_from_trainer", "dataset:emotion", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-10-11T18:49:16Z	2022-10-24T11:58:12+00:00	14	0	--- datasets: - emotion license: apache-2.0 metrics: - accuracy - f1 tags: - generated_from_trainer model-index: - name: distilbert-base-uncased-finetuned-emotion results: - task: type: text-classification name: Text Classification dataset: name: emotion type: emotion config: default split: train args: default metrics: - type: accuracy value: 0.928 name: Accuracy - type: f1 value: 0.9280714609088352 name: F1 --- # distilbert-base-uncased-finetuned-emotion This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset. It achieves the following results on the evaluation set: - Loss: 0.2185 - Accuracy: 0.928 - F1: 0.9281 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 64 - eval_batch_size: 64 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 2 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\|:------:\| \| 0.8374 \| 1.0 \| 250 \| 0.3188 \| 0.9045 \| 0.9012 \| \| 0.254 \| 2.0 \| 500 \| 0.2185 \| 0.928 \| 0.9281 \| ### Framework versions - Transformers 4.22.2 - Pytorch 1.12.1+cpu - Datasets 2.4.0 - Tokenizers 0.12.1	null	Non_BioNLP	# distilbert-base-uncased-finetuned-emotion This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset. It achieves the following results on the evaluation set: - Loss: 0.2185 - Accuracy: 0.928 - F1: 0.9281 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 64 - eval_batch_size: 64 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 2 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\|:------:\| \| 0.8374 \| 1.0 \| 250 \| 0.3188 \| 0.9045 \| 0.9012 \| \| 0.254 \| 2.0 \| 500 \| 0.2185 \| 0.928 \| 0.9281 \| ### Framework versions - Transformers 4.22.2 - Pytorch 1.12.1+cpu - Datasets 2.4.0 - Tokenizers 0.12.1	{"datasets": ["emotion"], "license": "apache-2.0", "metrics": ["accuracy", "f1"], "tags": ["generated_from_trainer"], "model-index": [{"name": "distilbert-base-uncased-finetuned-emotion", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "emotion", "type": "emotion", "config": "default", "split": "train", "args": "default"}, "metrics": [{"type": "accuracy", "value": 0.928, "name": "Accuracy"}, {"type": "f1", "value": 0.9280714609088352, "name": "F1"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,102
tftransformers/bart-large	tftransformers	null	[ "transformers", "en", "arxiv:1910.13461", "license:apache-2.0", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:05Z	2021-10-24T08:24:25+00:00	7	0	--- language: en license: apache-2.0 --- # BART (large-sized model) BART model pre-trained on English language. It was introduced in the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Lewis et al. and first released in [this repository](https://github.com/pytorch/fairseq/tree/master/examples/bart). Disclaimer: The team releasing BART did not write a model card for this model so this model card has been written by the Hugging Face team. ## Model description BART is a transformer encoder-encoder (seq2seq) model with a bidirectional (BERT-like) encoder and an autoregressive (GPT-like) decoder. BART is pre-trained by (1) corrupting text with an arbitrary noising function, and (2) learning a model to reconstruct the original text. BART is particularly effective when fine-tuned for text generation (e.g. summarization, translation) but also works well for comprehension tasks (e.g. text classification, question answering). ## Intended uses & limitations You can use the raw model for text infilling. However, the model is mostly meant to be fine-tuned on a supervised dataset. See the [model hub](https://huggingface.co/models?search=bart) to look for fine-tuned versions on a task that interests you. ### How to use Here is how to use this model in tf_transformers: ```python from tf_transformers.models import BartModel from transformers import BartTokenizer tokenizer = BartTokenizer.from_pretrained('facebook/bart-large') model = BartModel.from_pretrained('facebook/bart-large') inputs_tf = {} inputs = tokenizer("Hello, my dog is cute", return_tensors="pt") inputs_tf["encoder_input_ids"] = inputs["input_ids"] inputs_tf["encoder_input_mask"] = inputs["attention_mask"] inputs_tf["decoder_input_ids"] = decoder_input_ids outputs_tf = model(inputs_tf) ``` ### BibTeX entry and citation info ```bibtex @article{DBLP:journals/corr/abs-1910-13461, author = {Mike Lewis and Yinhan Liu and Naman Goyal and Marjan Ghazvininejad and Abdelrahman Mohamed and Omer Levy and Veselin Stoyanov and Luke Zettlemoyer}, title = {{BART:} Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension}, journal = {CoRR}, volume = {abs/1910.13461}, year = {2019}, url = {http://arxiv.org/abs/1910.13461}, eprinttype = {arXiv}, eprint = {1910.13461}, timestamp = {Thu, 31 Oct 2019 14:02:26 +0100}, biburl = {https://dblp.org/rec/journals/corr/abs-1910-13461.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } ```	null	Non_BioNLP	# BART (large-sized model) BART model pre-trained on English language. It was introduced in the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Lewis et al. and first released in [this repository](https://github.com/pytorch/fairseq/tree/master/examples/bart). Disclaimer: The team releasing BART did not write a model card for this model so this model card has been written by the Hugging Face team. ## Model description BART is a transformer encoder-encoder (seq2seq) model with a bidirectional (BERT-like) encoder and an autoregressive (GPT-like) decoder. BART is pre-trained by (1) corrupting text with an arbitrary noising function, and (2) learning a model to reconstruct the original text. BART is particularly effective when fine-tuned for text generation (e.g. summarization, translation) but also works well for comprehension tasks (e.g. text classification, question answering). ## Intended uses & limitations You can use the raw model for text infilling. However, the model is mostly meant to be fine-tuned on a supervised dataset. See the [model hub](https://huggingface.co/models?search=bart) to look for fine-tuned versions on a task that interests you. ### How to use Here is how to use this model in tf_transformers: ```python from tf_transformers.models import BartModel from transformers import BartTokenizer tokenizer = BartTokenizer.from_pretrained('facebook/bart-large') model = BartModel.from_pretrained('facebook/bart-large') inputs_tf = {} inputs = tokenizer("Hello, my dog is cute", return_tensors="pt") inputs_tf["encoder_input_ids"] = inputs["input_ids"] inputs_tf["encoder_input_mask"] = inputs["attention_mask"] inputs_tf["decoder_input_ids"] = decoder_input_ids outputs_tf = model(inputs_tf) ``` ### BibTeX entry and citation info ```bibtex @article{DBLP:journals/corr/abs-1910-13461, author = {Mike Lewis and Yinhan Liu and Naman Goyal and Marjan Ghazvininejad and Abdelrahman Mohamed and Omer Levy and Veselin Stoyanov and Luke Zettlemoyer}, title = {{BART:} Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension}, journal = {CoRR}, volume = {abs/1910.13461}, year = {2019}, url = {http://arxiv.org/abs/1910.13461}, eprinttype = {arXiv}, eprint = {1910.13461}, timestamp = {Thu, 31 Oct 2019 14:02:26 +0100}, biburl = {https://dblp.org/rec/journals/corr/abs-1910-13461.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } ```	{"language": "en", "license": "apache-2.0"}	task	[ "TEXT_CLASSIFICATION", "QUESTION_ANSWERING", "TRANSLATION", "SUMMARIZATION" ]	43,103
aroot/eng-guj-simcse_longestplus_usrl	aroot	translation	[ "transformers", "pytorch", "tensorboard", "mbart", "text2text-generation", "translation", "generated_from_trainer", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-07-07T07:53:43Z	2023-07-07T08:15:14+00:00	10	0	--- metrics: - bleu tags: - translation - generated_from_trainer model-index: - name: eng-guj-simcse_longestplus_usrl results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # eng-guj-simcse_longestplus_usrl This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset. It achieves the following results on the evaluation set: - Loss: 3.2755 - Bleu: 2.8744 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results ### Framework versions - Transformers 4.26.1 - Pytorch 2.0.1+cu117 - Datasets 2.12.0 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # eng-guj-simcse_longestplus_usrl This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset. It achieves the following results on the evaluation set: - Loss: 3.2755 - Bleu: 2.8744 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results ### Framework versions - Transformers 4.26.1 - Pytorch 2.0.1+cu117 - Datasets 2.12.0 - Tokenizers 0.13.3	{"metrics": ["bleu"], "tags": ["translation", "generated_from_trainer"], "model-index": [{"name": "eng-guj-simcse_longestplus_usrl", "results": []}]}	task	[ "TRANSLATION" ]	43,104
JaviBJ/sagemaker-distilbert-emotion	JaviBJ	text-classification	[ "transformers", "pytorch", "distilbert", "text-classification", "generated_from_trainer", "dataset:emotion", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:04Z	2021-11-17T17:02:01+00:00	113	0	--- datasets: - emotion license: apache-2.0 metrics: - accuracy tags: - generated_from_trainer model-index: - name: sagemaker-distilbert-emotion results: - task: type: text-classification name: Text Classification dataset: name: emotion type: emotion args: default metrics: - type: accuracy value: 0.9165 name: Accuracy --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # sagemaker-distilbert-emotion This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset. It achieves the following results on the evaluation set: - Loss: 0.2469 - Accuracy: 0.9165 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 3e-05 - train_batch_size: 32 - eval_batch_size: 64 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 500 - num_epochs: 1 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\| \| 0.9351 \| 1.0 \| 500 \| 0.2469 \| 0.9165 \| ### Framework versions - Transformers 4.12.3 - Pytorch 1.9.1 - Datasets 1.15.1 - Tokenizers 0.10.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # sagemaker-distilbert-emotion This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset. It achieves the following results on the evaluation set: - Loss: 0.2469 - Accuracy: 0.9165 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 3e-05 - train_batch_size: 32 - eval_batch_size: 64 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 500 - num_epochs: 1 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\| \| 0.9351 \| 1.0 \| 500 \| 0.2469 \| 0.9165 \| ### Framework versions - Transformers 4.12.3 - Pytorch 1.9.1 - Datasets 1.15.1 - Tokenizers 0.10.3	{"datasets": ["emotion"], "license": "apache-2.0", "metrics": ["accuracy"], "tags": ["generated_from_trainer"], "model-index": [{"name": "sagemaker-distilbert-emotion", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "emotion", "type": "emotion", "args": "default"}, "metrics": [{"type": "accuracy", "value": 0.9165, "name": "Accuracy"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,105
RichardErkhov/DISLab_-_SummLlama3.2-3B-8bits	RichardErkhov	null	[ "safetensors", "llama", "arxiv:2410.13116", "8-bit", "bitsandbytes", "region:us" ]	2024-11-28T16:11:02Z	2024-11-28T16:13:01+00:00	7	0	--- {} --- Quantization made by Richard Erkhov. [Github](https://github.com/RichardErkhov) [Discord](https://discord.gg/pvy7H8DZMG) [Request more models](https://github.com/RichardErkhov/quant_request) SummLlama3.2-3B - bnb 8bits - Model creator: https://huggingface.co/DISLab/ - Original model: https://huggingface.co/DISLab/SummLlama3.2-3B/ Original model description: --- library_name: transformers base_model: - meta-llama/Llama-3.2-3B-Instruct pipeline_tag: summarization widget: - text: "We are now testing this." --- <div align="center"> <b style="font-size: 40px;">SummLlama3.2-3B</b> </div> Are you looking for a summarizer that can generate more human-preferred summaries across multiple domains? Our SummLlama3.2-3B could be exactly what you need! SummLlama3.2-3B is initialized from Llama3.2-3B-Instruct, with additional training using Direct Preference Optimization (DPO) based on large-scale (over 100K) summarization feedback. The feedback encompasses a wide range of input documents, from short to lengthy texts, including both dialogue and non-dialogue formats, and spans across seven distinct domains: - Four non-dialouge domains: News, Lifestyle, Report, Medical - Three dialogue domains: Daily Life, Interview, Meeting This is automated evaluation results: \| Config. \| Faithfulness \| Completeness \| Conciseness \| Average \| \|--------------------\|------------\|-----------\|-----------\|----------\| \| Llama3.2-3B-Instruct \| 0.814 \| 0.580 \| 0.445 \| 0.613 \| \| SummLlama3.2-3B \| 0.867 \| 0.598 \| 0.686 \| 0.717 \| Please refer to [our paper](https://arxiv.org/abs/2410.13116) to catch up how to exploit LLM-generated feedback in the context of text summarization. SummLlama3-70B, https://huggingface.co/DISLab/SummLlama3-8B https://huggingface.co/DISLab/SummLlama3-70B SummLlama3.1-Series https://huggingface.co/DISLab/SummLlama3.1-8B https://huggingface.co/DISLab/SummLlama3.1-70B SummLlama3.2-Series https://huggingface.co/DISLab/SummLlama3.2-3B ### Recommended Prompt for Text Summarization: We recommend to use the prompt below to get the summary, since we trained the model using this. ``` def format_chat_template(document): instruction = "Please summarize the input documnet." row_json = [{"role": "user", "content": f"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{instruction}\n\n### Input:\n{document}\n\n### Response:\n"}] return tokenizer.apply_chat_template(row_json, tokenize=False) ``` --- Here is a brief overview of our summarizer: Rather than relying on expensive human feedback, we utilize high-quality, multi-dimensional, and fine-grained feedback generated by large language models (LLMs). This model excels at faithfulness, completeness, and conciseness, which are the three human-preferred aspects to judge what is a good summarizer. - Faithfulness: a summarizer does not manipulate the information in the input text and add any information not directly inferable from the input text. - Completeness: a summarizer ensures the inclusion of all key information from the input text in the output summary. - Conciseness: a summarizer refrains from incorporating information outside the key information in the output, maintaining a succinct and focused summary. Based on our comprehensive evaluation, which included both human and automated assessments of summary quality, SummLlama3 demonstrated significant improvements over the original Llama3 series. Here is the results: ## Human Evaluation \| Config. \| Faithfulness \| Completeness \| Conciseness \| Average \| \|--------------------\|------------\|-----------\|-----------\|----------\| \| Llama3-8B-Instruct \| 0.902 \| 0.636 \| 0.784 \| 0.774 \| \| Llama3-70B-Instruct \| 0.953 \| 0.659 \| 0.792 \| 0.801 \| \| SummLlama3-8B \| 0.980 \| 0.697 \| 0.959 \| 0.879 \| ## Autoamted Evaluation using [FineSurE](https://aclanthology.org/2024.acl-long.51.pdf) \| Config. \| Faithfulness \| Completeness \| Conciseness \| Average \| \|--------------------\|------------\|-----------\|-----------\|----------\| \| Llama3-8B-Instruct \| 0.864 \| 0.583 \| 0.450 \| 0.632 \| \| Llama3-70B-Instruct \| 0.931 \| 0.596 \| 0.487 \| 0.671 \| \| SummLlama3-8B \| 0.931 \| 0.614 \| 0.659 \| 0.735 \| \| SummLlama3-70B \| 0.950 \| 0.632 \| 0.754 \| 0.779 \| ## Example See an example how the summary improved by SummLlama3-8B over Llama3-8/70B-Instruct on the document below: \| Speaker \| Dialogue \| \|-------------\|--------------\| \| Person 1 \| Hey, Paul, you're still having Thanksgiving dinner at my house on Thursday, right? \| \| Person 2 \| Yeah, thanks again for the invitation. I was worried I'd have to spend it alone after my parents announced they'd be in Finland. \| \| Person 1 \| I think it's strange that you're not going with them. Is it because of the cold weather in Northern Europe? \| \| Person 2 \| Not exactly. I have a lot of work to do next week. \| \| Person 1 \| Well, I'm glad you won't be alone on a holiday. It'll be great to have you celebrate with us. The more the merrier! \| \| Person 2 \| Exactly, how many people will be there? \| \| Person 1 \| Let's see, my parents, my grandpa, my sister, my brother-in-law, my niece and my two cousins, plus us. So, ten people altogether. \| \| Person 2 \| Should I bring anything? \| \| Person 1 \| Nope, we've got it covered. \| \| Person 2 \| Come on, I hate to show up empty-handed. You are being so generous to have me. I know! I'll bring pie. I'll make a great pumpkin pie. \| \| Person 1 \| Sounds delicious, but my sister has already volunteered for dessert. Look, don't worry about it. There will be tons of food. Turkey, stuffing, green beans, sweet potatoes. \| \| Person 2 \| Wow, I'm getting hungry just thinking about it. How about this, I'll bring a nice bottle of wine. You can never have too much of that. \| \| Person 1 \| Well, isn't that the truth? Sounds good to me. \| \| Model \| Summary \| \|----------\|-----------\| \| Llama3-8B-Instruct \| Person1 invites Person2 to join them for Thanksgiving dinner at their house. Person2 accepts the invitation, citing that they were worried they would have to spend the holiday alone after their parents announced they would be in Finland. Person1 reassures Person2 that they are welcome and that the more the merrier. Person2 asks how many people will be attending and is told that there will be 10 people, including their family and cousins. Person2 offers to bring a gift, but Person1 declines, saying that they have everything covered. Person2 persists and offers to bring a bottle of wine or a pie, but Person1 suggests that their sister has already volunteered to bring dessert. \| \| Llama3-70B-Instruct \| Person 1 invites Person 2 to their house for Thanksgiving dinner on Thursday. Person 2 is grateful for the invitation and explains they can't go to Finland with their parents due to work commitments. Person 1 is happy to have them celebrate with their family, which will consist of 10 people. When Person 2 offers to bring something, Person 1 declines, but eventually agrees to let them bring a bottle of wine. \| \| SummLlama3-8B \| The conversation is about Thanksgiving dinner arrangements, with Person2 (Paul) confirming his attendance at Person1's (host's) house on Thursday, discussing what to bring and what's already been taken care of, and finalizing the offer to bring wine instead of pie since someone else is handling desserts. \| The summary of SummLlama3-8B can be considered a much human-preferred summary for the following reasons: Core Focus: The summary accurately captures the main theme of the conversation, which revolves around the Thanksgiving dinner arrangements. It highlights how the two people confirm plans, discuss what to bring, and finalize the decision for Person2 to bring wine instead of pie. This maintains the core context. Inclusion of Key-facts: The summary covers the important details of the conversation, including Person2's initial offer to bring dessert (pumpkin pie) and the shift to bringing wine due to another family member handling dessert. Other summaries tend to overlook or simplify this progression, while SummLlama3-8B fully captures the interaction’s key events. Clarity and Conciseness: The summary is structured in a straightforward, concise manner, effectively summarizing the conversation without unnecessary details. It presents the flow and outcome of the discussion clearly, making it easy for readers to understand. The logical order of events is maintained, ensuring a smooth narrative. Accurate Role Depiction: The summary clearly identifies Person1 as the host and Paul (Person2) as the guest, which helps clarify their relationship and the nature of the conversation. This distinction is more explicit in SummLlama3-8B compared to other summaries, which might leave these roles more ambiguous.	null	Non_BioNLP	Quantization made by Richard Erkhov. [Github](https://github.com/RichardErkhov) [Discord](https://discord.gg/pvy7H8DZMG) [Request more models](https://github.com/RichardErkhov/quant_request) SummLlama3.2-3B - bnb 8bits - Model creator: https://huggingface.co/DISLab/ - Original model: https://huggingface.co/DISLab/SummLlama3.2-3B/ Original model description: --- library_name: transformers base_model: - meta-llama/Llama-3.2-3B-Instruct pipeline_tag: summarization widget: - text: "We are now testing this." --- <div align="center"> <b style="font-size: 40px;">SummLlama3.2-3B</b> </div> Are you looking for a summarizer that can generate more human-preferred summaries across multiple domains? Our SummLlama3.2-3B could be exactly what you need! SummLlama3.2-3B is initialized from Llama3.2-3B-Instruct, with additional training using Direct Preference Optimization (DPO) based on large-scale (over 100K) summarization feedback. The feedback encompasses a wide range of input documents, from short to lengthy texts, including both dialogue and non-dialogue formats, and spans across seven distinct domains: - Four non-dialouge domains: News, Lifestyle, Report, Medical - Three dialogue domains: Daily Life, Interview, Meeting This is automated evaluation results: \| Config. \| Faithfulness \| Completeness \| Conciseness \| Average \| \|--------------------\|------------\|-----------\|-----------\|----------\| \| Llama3.2-3B-Instruct \| 0.814 \| 0.580 \| 0.445 \| 0.613 \| \| SummLlama3.2-3B \| 0.867 \| 0.598 \| 0.686 \| 0.717 \| Please refer to [our paper](https://arxiv.org/abs/2410.13116) to catch up how to exploit LLM-generated feedback in the context of text summarization. SummLlama3-70B, https://huggingface.co/DISLab/SummLlama3-8B https://huggingface.co/DISLab/SummLlama3-70B SummLlama3.1-Series https://huggingface.co/DISLab/SummLlama3.1-8B https://huggingface.co/DISLab/SummLlama3.1-70B SummLlama3.2-Series https://huggingface.co/DISLab/SummLlama3.2-3B ### Recommended Prompt for Text Summarization: We recommend to use the prompt below to get the summary, since we trained the model using this. ``` def format_chat_template(document): instruction = "Please summarize the input documnet." row_json = [{"role": "user", "content": f"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{instruction}\n\n### Input:\n{document}\n\n### Response:\n"}] return tokenizer.apply_chat_template(row_json, tokenize=False) ``` --- Here is a brief overview of our summarizer: Rather than relying on expensive human feedback, we utilize high-quality, multi-dimensional, and fine-grained feedback generated by large language models (LLMs). This model excels at faithfulness, completeness, and conciseness, which are the three human-preferred aspects to judge what is a good summarizer. - Faithfulness: a summarizer does not manipulate the information in the input text and add any information not directly inferable from the input text. - Completeness: a summarizer ensures the inclusion of all key information from the input text in the output summary. - Conciseness: a summarizer refrains from incorporating information outside the key information in the output, maintaining a succinct and focused summary. Based on our comprehensive evaluation, which included both human and automated assessments of summary quality, SummLlama3 demonstrated significant improvements over the original Llama3 series. Here is the results: ## Human Evaluation \| Config. \| Faithfulness \| Completeness \| Conciseness \| Average \| \|--------------------\|------------\|-----------\|-----------\|----------\| \| Llama3-8B-Instruct \| 0.902 \| 0.636 \| 0.784 \| 0.774 \| \| Llama3-70B-Instruct \| 0.953 \| 0.659 \| 0.792 \| 0.801 \| \| SummLlama3-8B \| 0.980 \| 0.697 \| 0.959 \| 0.879 \| ## Autoamted Evaluation using [FineSurE](https://aclanthology.org/2024.acl-long.51.pdf) \| Config. \| Faithfulness \| Completeness \| Conciseness \| Average \| \|--------------------\|------------\|-----------\|-----------\|----------\| \| Llama3-8B-Instruct \| 0.864 \| 0.583 \| 0.450 \| 0.632 \| \| Llama3-70B-Instruct \| 0.931 \| 0.596 \| 0.487 \| 0.671 \| \| SummLlama3-8B \| 0.931 \| 0.614 \| 0.659 \| 0.735 \| \| SummLlama3-70B \| 0.950 \| 0.632 \| 0.754 \| 0.779 \| ## Example See an example how the summary improved by SummLlama3-8B over Llama3-8/70B-Instruct on the document below: \| Speaker \| Dialogue \| \|-------------\|--------------\| \| Person 1 \| Hey, Paul, you're still having Thanksgiving dinner at my house on Thursday, right? \| \| Person 2 \| Yeah, thanks again for the invitation. I was worried I'd have to spend it alone after my parents announced they'd be in Finland. \| \| Person 1 \| I think it's strange that you're not going with them. Is it because of the cold weather in Northern Europe? \| \| Person 2 \| Not exactly. I have a lot of work to do next week. \| \| Person 1 \| Well, I'm glad you won't be alone on a holiday. It'll be great to have you celebrate with us. The more the merrier! \| \| Person 2 \| Exactly, how many people will be there? \| \| Person 1 \| Let's see, my parents, my grandpa, my sister, my brother-in-law, my niece and my two cousins, plus us. So, ten people altogether. \| \| Person 2 \| Should I bring anything? \| \| Person 1 \| Nope, we've got it covered. \| \| Person 2 \| Come on, I hate to show up empty-handed. You are being so generous to have me. I know! I'll bring pie. I'll make a great pumpkin pie. \| \| Person 1 \| Sounds delicious, but my sister has already volunteered for dessert. Look, don't worry about it. There will be tons of food. Turkey, stuffing, green beans, sweet potatoes. \| \| Person 2 \| Wow, I'm getting hungry just thinking about it. How about this, I'll bring a nice bottle of wine. You can never have too much of that. \| \| Person 1 \| Well, isn't that the truth? Sounds good to me. \| \| Model \| Summary \| \|----------\|-----------\| \| Llama3-8B-Instruct \| Person1 invites Person2 to join them for Thanksgiving dinner at their house. Person2 accepts the invitation, citing that they were worried they would have to spend the holiday alone after their parents announced they would be in Finland. Person1 reassures Person2 that they are welcome and that the more the merrier. Person2 asks how many people will be attending and is told that there will be 10 people, including their family and cousins. Person2 offers to bring a gift, but Person1 declines, saying that they have everything covered. Person2 persists and offers to bring a bottle of wine or a pie, but Person1 suggests that their sister has already volunteered to bring dessert. \| \| Llama3-70B-Instruct \| Person 1 invites Person 2 to their house for Thanksgiving dinner on Thursday. Person 2 is grateful for the invitation and explains they can't go to Finland with their parents due to work commitments. Person 1 is happy to have them celebrate with their family, which will consist of 10 people. When Person 2 offers to bring something, Person 1 declines, but eventually agrees to let them bring a bottle of wine. \| \| SummLlama3-8B \| The conversation is about Thanksgiving dinner arrangements, with Person2 (Paul) confirming his attendance at Person1's (host's) house on Thursday, discussing what to bring and what's already been taken care of, and finalizing the offer to bring wine instead of pie since someone else is handling desserts. \| The summary of SummLlama3-8B can be considered a much human-preferred summary for the following reasons: Core Focus: The summary accurately captures the main theme of the conversation, which revolves around the Thanksgiving dinner arrangements. It highlights how the two people confirm plans, discuss what to bring, and finalize the decision for Person2 to bring wine instead of pie. This maintains the core context. Inclusion of Key-facts: The summary covers the important details of the conversation, including Person2's initial offer to bring dessert (pumpkin pie) and the shift to bringing wine due to another family member handling dessert. Other summaries tend to overlook or simplify this progression, while SummLlama3-8B fully captures the interaction’s key events. Clarity and Conciseness: The summary is structured in a straightforward, concise manner, effectively summarizing the conversation without unnecessary details. It presents the flow and outcome of the discussion clearly, making it easy for readers to understand. The logical order of events is maintained, ensuring a smooth narrative. Accurate Role Depiction: The summary clearly identifies Person1 as the host and Paul (Person2) as the guest, which helps clarify their relationship and the nature of the conversation. This distinction is more explicit in SummLlama3-8B compared to other summaries, which might leave these roles more ambiguous.	{}	task	[ "SUMMARIZATION" ]	43,106
pahautelman/phi2-ner-v1	pahautelman	text-generation	[ "transformers", "safetensors", "phi", "text-generation", "autotrain", "named entity recognition", "conversational", "custom_code", "en", "dataset:conll2012_ontonotesv5", "license:mit", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "4-bit", "bitsandbytes", "region:us" ]	2024-02-14T17:35:00Z	2024-02-15T21:38:11+00:00	4	0	--- datasets: - conll2012_ontonotesv5 language: - en license: mit tags: - autotrain - text-generation - transformers - named entity recognition widget: - text: 'I love AutoTrain because ' --- # Phi-2 model fine-tuned for named entity recognition task The model was fine-tuned using one quarter of the ConLL 2012 OntoNotes v5 dataset. - Dataset Source: [conll2012_ontonotesv5](https://huggingface.co/datasets/conll2012_ontonotesv5) - Subset Used: English_v12 - Number of Examples: 87,265 The prompts and expected outputs were constructed as described in [1]. Example input: ```md Instruct: I am an excelent linquist. The task is to label organization entities in the given sentence. Below are some examples Input: A spokesman for B. A. T said of the amended filings that,`` It would appear that nothing substantive has changed. Output: A spokesman for @@B. A. T## said of the amended filings that,`` It would appear that nothing substantive has changed. Input: Since NBC's interest in the Qintex bid for MGM / UA was disclosed, Mr. Wright has n't been available for comment. Output: Since @@NBC##'s interest in the @@Qintex## bid for @@MGM / UA## was disclosed, Mr. Wright has n't been available for comment. Input: You know news organizations demand total transparency whether you're General Motors or United States government /. Output: You know news organizations demand total transparency whether you're @@General Motors## or United States government /. Input: We respectfully invite you to watch a special edition of Across China. Output: ``` Expected output: ```md We respectfully invite you to watch a special edition of @@Across China##. ``` This model is trained to recognize the named entity categories - person - nationalities or religious or political groups - facility - organization - geopolitical entity - location - product - date - time expression - percentage - monetary value - quantity - event - work of art - law/legal reference - language name # Model Trained Using AutoTrain This model was trained using SFT AutoTrain trainer. For more information, please visit [AutoTrain](https://hf.co/docs/autotrain). Hyperparameters: ```json { "model": "microsoft/phi-2", "valid_split": null, "add_eos_token": false, "block_size": 1024, "model_max_length": 1024, "padding": "right", "trainer": "sft", "use_flash_attention_2": false, "disable_gradient_checkpointing": false, "evaluation_strategy": "epoch", "save_total_limit": 1, "save_strategy": "epoch", "auto_find_batch_size": false, "mixed_precision": "bf16", "lr": 0.0002, "epochs": 1, "batch_size": 1, "warmup_ratio": 0.1, "gradient_accumulation": 4, "optimizer": "adamw_torch", "scheduler": "linear", "weight_decay": 0.01, "max_grad_norm": 1.0, "seed": 42, "apply_chat_template": false, "quantization": "int4", "target_modules": null, "merge_adapter": false, "peft": true, "lora_r": 16, "lora_alpha": 32, "lora_dropout": 0.05, "dpo_beta": 0.1, } ``` # Usage ```python from transformers import AutoModelForCausalLM, AutoTokenizer model_path = "pahautelman/phi2-ner-v1" tokenizer = AutoTokenizer.from_pretrained(model_path) model = AutoModelForCausalLM.from_pretrained( model_path ).eval() prompt = 'Label the person entities in the given sentence: Russian President Vladimir Putin is due to arrive in Havana a few hours from now to become the first post-Soviet leader to visit Cuba.' inputs = tokenizer.encode(prompt, add_special_tokens=False, return_tensors='pt') outputs = model.generate( inputs.to(model.device), max_new_tokens=9, do_sample=False, ) output = tokenizer.batch_decode(outputs)[0] # Model response: "Output: Russian President, Vladimir Putin" print(output) ``` # References: [1] Wang et al., GPT-NER: Named entity recognition via large language models 2023	null	Non_BioNLP	# Phi-2 model fine-tuned for named entity recognition task The model was fine-tuned using one quarter of the ConLL 2012 OntoNotes v5 dataset. - Dataset Source: [conll2012_ontonotesv5](https://huggingface.co/datasets/conll2012_ontonotesv5) - Subset Used: English_v12 - Number of Examples: 87,265 The prompts and expected outputs were constructed as described in [1]. Example input: ```md Instruct: I am an excelent linquist. The task is to label organization entities in the given sentence. Below are some examples Input: A spokesman for B. A. T said of the amended filings that,`` It would appear that nothing substantive has changed. Output: A spokesman for @@B. A. T## said of the amended filings that,`` It would appear that nothing substantive has changed. Input: Since NBC's interest in the Qintex bid for MGM / UA was disclosed, Mr. Wright has n't been available for comment. Output: Since @@NBC##'s interest in the @@Qintex## bid for @@MGM / UA## was disclosed, Mr. Wright has n't been available for comment. Input: You know news organizations demand total transparency whether you're General Motors or United States government /. Output: You know news organizations demand total transparency whether you're @@General Motors## or United States government /. Input: We respectfully invite you to watch a special edition of Across China. Output: ``` Expected output: ```md We respectfully invite you to watch a special edition of @@Across China##. ``` This model is trained to recognize the named entity categories - person - nationalities or religious or political groups - facility - organization - geopolitical entity - location - product - date - time expression - percentage - monetary value - quantity - event - work of art - law/legal reference - language name # Model Trained Using AutoTrain This model was trained using SFT AutoTrain trainer. For more information, please visit [AutoTrain](https://hf.co/docs/autotrain). Hyperparameters: ```json { "model": "microsoft/phi-2", "valid_split": null, "add_eos_token": false, "block_size": 1024, "model_max_length": 1024, "padding": "right", "trainer": "sft", "use_flash_attention_2": false, "disable_gradient_checkpointing": false, "evaluation_strategy": "epoch", "save_total_limit": 1, "save_strategy": "epoch", "auto_find_batch_size": false, "mixed_precision": "bf16", "lr": 0.0002, "epochs": 1, "batch_size": 1, "warmup_ratio": 0.1, "gradient_accumulation": 4, "optimizer": "adamw_torch", "scheduler": "linear", "weight_decay": 0.01, "max_grad_norm": 1.0, "seed": 42, "apply_chat_template": false, "quantization": "int4", "target_modules": null, "merge_adapter": false, "peft": true, "lora_r": 16, "lora_alpha": 32, "lora_dropout": 0.05, "dpo_beta": 0.1, } ``` # Usage ```python from transformers import AutoModelForCausalLM, AutoTokenizer model_path = "pahautelman/phi2-ner-v1" tokenizer = AutoTokenizer.from_pretrained(model_path) model = AutoModelForCausalLM.from_pretrained( model_path ).eval() prompt = 'Label the person entities in the given sentence: Russian President Vladimir Putin is due to arrive in Havana a few hours from now to become the first post-Soviet leader to visit Cuba.' inputs = tokenizer.encode(prompt, add_special_tokens=False, return_tensors='pt') outputs = model.generate( inputs.to(model.device), max_new_tokens=9, do_sample=False, ) output = tokenizer.batch_decode(outputs)[0] # Model response: "Output: Russian President, Vladimir Putin" print(output) ``` # References: [1] Wang et al., GPT-NER: Named entity recognition via large language models 2023	{"datasets": ["conll2012_ontonotesv5"], "language": ["en"], "license": "mit", "tags": ["autotrain", "text-generation", "transformers", "named entity recognition"], "widget": [{"text": "I love AutoTrain because "}]}	task	[ "NAMED_ENTITY_RECOGNITION" ]	43,107
TechxGenus/c4ai-command-r-v01-GPTQ	TechxGenus	text-generation	[ "transformers", "safetensors", "cohere", "text-generation", "conversational", "en", "fr", "de", "es", "it", "pt", "ja", "ko", "zh", "ar", "license:cc-by-nc-4.0", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "4-bit", "gptq", "region:us" ]	2024-04-29T16:22:21Z	2024-04-29T17:44:30+00:00	8	2	--- language: - en - fr - de - es - it - pt - ja - ko - zh - ar library_name: transformers license: cc-by-nc-4.0 --- GPTQ quantized version of c4ai-command-r-v01 model. --- # Model Card for C4AI Command-R 🚨 This model is non-quantized version of C4AI Command-R. You can find the quantized version of C4AI Command-R using bitsandbytes [here](https://huggingface.co/CohereForAI/c4ai-command-r-v01-4bit). ## Model Summary C4AI Command-R is a research release of a 35 billion parameter highly performant generative model. Command-R is a large language model with open weights optimized for a variety of use cases including reasoning, summarization, and question answering. Command-R has the capability for multilingual generation evaluated in 10 languages and highly performant RAG capabilities. Developed by: Cohere and [Cohere For AI](https://cohere.for.ai) - Point of Contact: Cohere For AI: [cohere.for.ai](https://cohere.for.ai/) - License: [CC-BY-NC](https://cohere.com/c4ai-cc-by-nc-license), requires also adhering to [C4AI's Acceptable Use Policy](https://docs.cohere.com/docs/c4ai-acceptable-use-policy) - Model: c4ai-command-r-v01 - Model Size: 35 billion parameters - Context length: 128K Try C4AI Command R If you want to try Command R before downloading the weights, the model is hosted in a hugging face space [here](https://huggingface.co/spaces/CohereForAI/c4ai-command-r-v01). Usage Please use `transformers` version 4.39.1 or higher ```python # pip install 'transformers>=4.39.1' from transformers import AutoTokenizer, AutoModelForCausalLM model_id = "CohereForAI/c4ai-command-r-v01" tokenizer = AutoTokenizer.from_pretrained(model_id) model = AutoModelForCausalLM.from_pretrained(model_id) # Format message with the command-r chat template messages = [{"role": "user", "content": "Hello, how are you?"}] input_ids = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt") ## <BOS_TOKEN><\|START_OF_TURN_TOKEN\|><\|USER_TOKEN\|>Hello, how are you?<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|CHATBOT_TOKEN\|> gen_tokens = model.generate( input_ids, max_new_tokens=100, do_sample=True, temperature=0.3, ) gen_text = tokenizer.decode(gen_tokens[0]) print(gen_text) ``` Quantized model through bitsandbytes, 8-bit precision ```python # pip install 'transformers>=4.39.1' bitsandbytes accelerate from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig bnb_config = BitsAndBytesConfig(load_in_8bit=True) model_id = "CohereForAI/c4ai-command-r-v01" tokenizer = AutoTokenizer.from_pretrained(model_id) model = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=bnb_config) # Format message with the command-r chat template messages = [{"role": "user", "content": "Hello, how are you?"}] input_ids = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt") ## <BOS_TOKEN><\|START_OF_TURN_TOKEN\|><\|USER_TOKEN\|>Hello, how are you?<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|CHATBOT_TOKEN\|> gen_tokens = model.generate( input_ids, max_new_tokens=100, do_sample=True, temperature=0.3, ) gen_text = tokenizer.decode(gen_tokens[0]) print(gen_text) ``` Quantized model through bitsandbytes, 4-bit precision You can find a quantized version of this model to 4-bit precision [here](https://huggingface.co/CohereForAI/c4ai-command-r-v01-4bit). ## Model Details Input: Models input text only. Output: Models generate text only. Model Architecture: This is an auto-regressive language model that uses an optimized transformer architecture. After pretraining, this model uses supervised fine-tuning (SFT) and preference training to align model behavior to human preferences for helpfulness and safety. Languages covered: The model is optimized to perform well in the following languages: English, French, Spanish, Italian, German, Brazilian Portuguese, Japanese, Korean, Simplified Chinese, and Arabic. Pre-training data additionally included the following 13 languages: Russian, Polish, Turkish, Vietnamese, Dutch, Czech, Indonesian, Ukrainian, Romanian, Greek, Hindi, Hebrew, Persian. Context length: Command-R supports a context length of 128K. ### Tool use capabilities: Command-R has been specifically trained with conversational tool use capabilities. These have been trained into the model via a mixture of supervised fine-tuning and preference fine-tuning, using a specific prompt template. Deviating from this prompt template will likely reduce performance, but we encourage experimentation. Command-R’s tool use functionality takes a conversation as input (with an optional user-system preamble), along with a list of available tools. The model will then generate a json-formatted list of actions to execute on a subset of those tools. Command-R may use one of its supplied tools more than once. The model has been trained to recognise a special `directly_answer` tool, which it uses to indicate that it doesn’t want to use any of its other tools. The ability to abstain from calling a specific tool can be useful in a range of situations, such as greeting a user, or asking clarifying questions. We recommend including the `directly_answer` tool, but it can be removed or renamed if required. Comprehensive documentation for working with command-R's tool use prompt template can be found [here](https://docs.cohere.com/docs/prompting-command-r). The code snippet below shows a minimal working example on how to render a prompt. <details> <summary><b>Usage: Rendering Tool Use Prompts [CLICK TO EXPAND]</b> </summary> ```python from transformers import AutoTokenizer model_id = "CohereForAI/c4ai-command-r-v01" tokenizer = AutoTokenizer.from_pretrained(model_id) # define conversation input: conversation = [ {"role": "user", "content": "Whats the biggest penguin in the world?"} ] # Define tools available for the model to use: tools = [ { "name": "internet_search", "description": "Returns a list of relevant document snippets for a textual query retrieved from the internet", "parameter_definitions": { "query": { "description": "Query to search the internet with", "type": 'str', "required": True } } }, { 'name': "directly_answer", "description": "Calls a standard (un-augmented) AI chatbot to generate a response given the conversation history", 'parameter_definitions': {} } ] # render the tool use prompt as a string: tool_use_prompt = tokenizer.apply_tool_use_template( conversation, tools=tools, tokenize=False, add_generation_prompt=True, ) print(tool_use_prompt) ``` </details> <details> <summary><b>Example Rendered Tool Use Prompt [CLICK TO EXPAND]</b></summary> ```` <BOS_TOKEN><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|># Safety Preamble The instructions in this section override those in the task description and style guide sections. Don't answer questions that are harmful or immoral. # System Preamble ## Basic Rules You are a powerful conversational AI trained by Cohere to help people. You are augmented by a number of tools, and your job is to use and consume the output of these tools to best help the user. You will see a conversation history between yourself and a user, ending with an utterance from the user. You will then see a specific instruction instructing you what kind of response to generate. When you answer the user's requests, you cite your sources in your answers, according to those instructions. # User Preamble ## Task and Context You help people answer their questions and other requests interactively. You will be asked a very wide array of requests on all kinds of topics. You will be equipped with a wide range of search engines or similar tools to help you, which you use to research your answer. You should focus on serving the user's needs as best you can, which will be wide-ranging. ## Style Guide Unless the user asks for a different style of answer, you should answer in full sentences, using proper grammar and spelling. ## Available Tools Here is a list of tools that you have available to you: ```python def internet_search(query: str) -> List[Dict]: """Returns a list of relevant document snippets for a textual query retrieved from the internet Args: query (str): Query to search the internet with """ pass ``` ```python def directly_answer() -> List[Dict]: """Calls a standard (un-augmented) AI chatbot to generate a response given the conversation history """ pass ```<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|USER_TOKEN\|>Whats the biggest penguin in the world?<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|>Write 'Action:' followed by a json-formatted list of actions that you want to perform in order to produce a good response to the user's last input. You can use any of the supplied tools any number of times, but you should aim to execute the minimum number of necessary actions for the input. You should use the `directly-answer` tool if calling the other tools is unnecessary. The list of actions you want to call should be formatted as a list of json objects, for example: ```json [ { "tool_name": title of the tool in the specification, "parameters": a dict of parameters to input into the tool as they are defined in the specs, or {} if it takes no parameters } ]```<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|CHATBOT_TOKEN\|> ```` </details> <details> <summary><b>Example Rendered Tool Use Completion [CLICK TO EXPAND]</b></summary> ```` Action: ```json [ { "tool_name": "internet_search", "parameters": { "query": "biggest penguin in the world" } } ] ``` ```` </details> ### Grounded Generation and RAG Capabilities: Command-R has been specifically trained with grounded generation capabilities. This means that it can generate responses based on a list of supplied document snippets, and it will include grounding spans (citations) in its response indicating the source of the information. This can be used to enable behaviors such as grounded summarization and the final step of Retrieval Augmented Generation (RAG).This behavior has been trained into the model via a mixture of supervised fine-tuning and preference fine-tuning, using a specific prompt template. Deviating from this prompt template may reduce performance, but we encourage experimentation. Command-R’s grounded generation behavior takes a conversation as input (with an optional user-supplied system preamble, indicating task, context and desired output style), along with a list of retrieved document snippets. The document snippets should be chunks, rather than long documents, typically around 100-400 words per chunk. Document snippets consist of key-value pairs. The keys should be short descriptive strings, the values can be text or semi-structured. By default, Command-R will generate grounded responses by first predicting which documents are relevant, then predicting which ones it will cite, then generating an answer. Finally, it will then insert grounding spans into the answer. See below for an example. This is referred to as `accurate` grounded generation. The model is trained with a number of other answering modes, which can be selected by prompt changes . A `fast` citation mode is supported in the tokenizer, which will directly generate an answer with grounding spans in it, without first writing the answer out in full. This sacrifices some grounding accuracy in favor of generating fewer tokens. Comprehensive documentation for working with command-R's grounded generation prompt template can be found [here](https://docs.cohere.com/docs/prompting-command-r). The code snippet below shows a minimal working example on how to render a prompt. <details> <summary> <b>Usage: Rendering Grounded Generation prompts [CLICK TO EXPAND]</b> </summary> ````python from transformers import AutoTokenizer model_id = "CohereForAI/c4ai-command-r-v01" tokenizer = AutoTokenizer.from_pretrained(model_id) # define conversation input: conversation = [ {"role": "user", "content": "Whats the biggest penguin in the world?"} ] # define documents to ground on: documents = [ { "title": "Tall penguins", "text": "Emperor penguins are the tallest growing up to 122 cm in height." }, { "title": "Penguin habitats", "text": "Emperor penguins only live in Antarctica."} ] # render the tool use prompt as a string: grounded_generation_prompt = tokenizer.apply_grounded_generation_template( conversation, documents=documents, citation_mode="accurate", # or "fast" tokenize=False, add_generation_prompt=True, ) print(grounded_generation_prompt) ```` </details> <details> <summary><b>Example Rendered Grounded Generation Prompt [CLICK TO EXPAND]</b></summary> ````<BOS_TOKEN><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|># Safety Preamble The instructions in this section override those in the task description and style guide sections. Don't answer questions that are harmful or immoral. # System Preamble ## Basic Rules You are a powerful conversational AI trained by Cohere to help people. You are augmented by a number of tools, and your job is to use and consume the output of these tools to best help the user. You will see a conversation history between yourself and a user, ending with an utterance from the user. You will then see a specific instruction instructing you what kind of response to generate. When you answer the user's requests, you cite your sources in your answers, according to those instructions. # User Preamble ## Task and Context You help people answer their questions and other requests interactively. You will be asked a very wide array of requests on all kinds of topics. You will be equipped with a wide range of search engines or similar tools to help you, which you use to research your answer. You should focus on serving the user's needs as best you can, which will be wide-ranging. ## Style Guide Unless the user asks for a different style of answer, you should answer in full sentences, using proper grammar and spelling.<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|USER_TOKEN\|>Whats the biggest penguin in the world?<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|><results> Document: 0 title: Tall penguins text: Emperor penguins are the tallest growing up to 122 cm in height. Document: 1 title: Penguin habitats text: Emperor penguins only live in Antarctica. </results><\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|>Carefully perform the following instructions, in order, starting each with a new line. Firstly, Decide which of the retrieved documents are relevant to the user's last input by writing 'Relevant Documents:' followed by comma-separated list of document numbers. If none are relevant, you should instead write 'None'. Secondly, Decide which of the retrieved documents contain facts that should be cited in a good answer to the user's last input by writing 'Cited Documents:' followed a comma-separated list of document numbers. If you dont want to cite any of them, you should instead write 'None'. Thirdly, Write 'Answer:' followed by a response to the user's last input in high quality natural english. Use the retrieved documents to help you. Do not insert any citations or grounding markup. Finally, Write 'Grounded answer:' followed by a response to the user's last input in high quality natural english. Use the symbols <co: doc> and </co: doc> to indicate when a fact comes from a document in the search result, e.g <co: 0>my fact</co: 0> for a fact from document 0.<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|CHATBOT_TOKEN\|> ```` </details> <details> <summary><b>Example Rendered Grounded Generation Completion [CLICK TO EXPAND]</b></summary> ```` Relevant Documents: 0,1 Cited Documents: 0,1 Answer: The Emperor Penguin is the tallest or biggest penguin in the world. It is a bird that lives only in Antarctica and grows to a height of around 122 centimetres. Grounded answer: The <co: 0>Emperor Penguin</co: 0> is the <co: 0>tallest</co: 0> or biggest penguin in the world. It is a bird that <co: 1>lives only in Antarctica</co: 1> and <co: 0>grows to a height of around 122 centimetres.</co: 0> ```` </details> ### Code Capabilities: Command-R has been optimized to interact with your code, by requesting code snippets, code explanations, or code rewrites. It might not perform well out-of-the-box for pure code completion. For better performance, we also recommend using a low temperature (and even greedy decoding) for code-generation related instructions. ### Model Card Contact For errors or additional questions about details in this model card, contact [[email protected]](mailto:[email protected]). ### Terms of Use: We hope that the release of this model will make community-based research efforts more accessible, by releasing the weights of a highly performant 35 billion parameter model to researchers all over the world. This model is governed by a [CC-BY-NC](https://cohere.com/c4ai-cc-by-nc-license) License with an acceptable use addendum, and also requires adhering to [C4AI's Acceptable Use Policy](https://docs.cohere.com/docs/c4ai-acceptable-use-policy). ### Try Chat: You can try Command-R chat in the playground [here](https://dashboard.cohere.com/playground/chat).	null	Non_BioNLP	GPTQ quantized version of c4ai-command-r-v01 model. --- # Model Card for C4AI Command-R 🚨 This model is non-quantized version of C4AI Command-R. You can find the quantized version of C4AI Command-R using bitsandbytes [here](https://huggingface.co/CohereForAI/c4ai-command-r-v01-4bit). ## Model Summary C4AI Command-R is a research release of a 35 billion parameter highly performant generative model. Command-R is a large language model with open weights optimized for a variety of use cases including reasoning, summarization, and question answering. Command-R has the capability for multilingual generation evaluated in 10 languages and highly performant RAG capabilities. Developed by: Cohere and [Cohere For AI](https://cohere.for.ai) - Point of Contact: Cohere For AI: [cohere.for.ai](https://cohere.for.ai/) - License: [CC-BY-NC](https://cohere.com/c4ai-cc-by-nc-license), requires also adhering to [C4AI's Acceptable Use Policy](https://docs.cohere.com/docs/c4ai-acceptable-use-policy) - Model: c4ai-command-r-v01 - Model Size: 35 billion parameters - Context length: 128K Try C4AI Command R If you want to try Command R before downloading the weights, the model is hosted in a hugging face space [here](https://huggingface.co/spaces/CohereForAI/c4ai-command-r-v01). Usage Please use `transformers` version 4.39.1 or higher ```python # pip install 'transformers>=4.39.1' from transformers import AutoTokenizer, AutoModelForCausalLM model_id = "CohereForAI/c4ai-command-r-v01" tokenizer = AutoTokenizer.from_pretrained(model_id) model = AutoModelForCausalLM.from_pretrained(model_id) # Format message with the command-r chat template messages = [{"role": "user", "content": "Hello, how are you?"}] input_ids = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt") ## <BOS_TOKEN><\|START_OF_TURN_TOKEN\|><\|USER_TOKEN\|>Hello, how are you?<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|CHATBOT_TOKEN\|> gen_tokens = model.generate( input_ids, max_new_tokens=100, do_sample=True, temperature=0.3, ) gen_text = tokenizer.decode(gen_tokens[0]) print(gen_text) ``` Quantized model through bitsandbytes, 8-bit precision ```python # pip install 'transformers>=4.39.1' bitsandbytes accelerate from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig bnb_config = BitsAndBytesConfig(load_in_8bit=True) model_id = "CohereForAI/c4ai-command-r-v01" tokenizer = AutoTokenizer.from_pretrained(model_id) model = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=bnb_config) # Format message with the command-r chat template messages = [{"role": "user", "content": "Hello, how are you?"}] input_ids = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt") ## <BOS_TOKEN><\|START_OF_TURN_TOKEN\|><\|USER_TOKEN\|>Hello, how are you?<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|CHATBOT_TOKEN\|> gen_tokens = model.generate( input_ids, max_new_tokens=100, do_sample=True, temperature=0.3, ) gen_text = tokenizer.decode(gen_tokens[0]) print(gen_text) ``` Quantized model through bitsandbytes, 4-bit precision You can find a quantized version of this model to 4-bit precision [here](https://huggingface.co/CohereForAI/c4ai-command-r-v01-4bit). ## Model Details Input: Models input text only. Output: Models generate text only. Model Architecture: This is an auto-regressive language model that uses an optimized transformer architecture. After pretraining, this model uses supervised fine-tuning (SFT) and preference training to align model behavior to human preferences for helpfulness and safety. Languages covered: The model is optimized to perform well in the following languages: English, French, Spanish, Italian, German, Brazilian Portuguese, Japanese, Korean, Simplified Chinese, and Arabic. Pre-training data additionally included the following 13 languages: Russian, Polish, Turkish, Vietnamese, Dutch, Czech, Indonesian, Ukrainian, Romanian, Greek, Hindi, Hebrew, Persian. Context length: Command-R supports a context length of 128K. ### Tool use capabilities: Command-R has been specifically trained with conversational tool use capabilities. These have been trained into the model via a mixture of supervised fine-tuning and preference fine-tuning, using a specific prompt template. Deviating from this prompt template will likely reduce performance, but we encourage experimentation. Command-R’s tool use functionality takes a conversation as input (with an optional user-system preamble), along with a list of available tools. The model will then generate a json-formatted list of actions to execute on a subset of those tools. Command-R may use one of its supplied tools more than once. The model has been trained to recognise a special `directly_answer` tool, which it uses to indicate that it doesn’t want to use any of its other tools. The ability to abstain from calling a specific tool can be useful in a range of situations, such as greeting a user, or asking clarifying questions. We recommend including the `directly_answer` tool, but it can be removed or renamed if required. Comprehensive documentation for working with command-R's tool use prompt template can be found [here](https://docs.cohere.com/docs/prompting-command-r). The code snippet below shows a minimal working example on how to render a prompt. <details> <summary><b>Usage: Rendering Tool Use Prompts [CLICK TO EXPAND]</b> </summary> ```python from transformers import AutoTokenizer model_id = "CohereForAI/c4ai-command-r-v01" tokenizer = AutoTokenizer.from_pretrained(model_id) # define conversation input: conversation = [ {"role": "user", "content": "Whats the biggest penguin in the world?"} ] # Define tools available for the model to use: tools = [ { "name": "internet_search", "description": "Returns a list of relevant document snippets for a textual query retrieved from the internet", "parameter_definitions": { "query": { "description": "Query to search the internet with", "type": 'str', "required": True } } }, { 'name': "directly_answer", "description": "Calls a standard (un-augmented) AI chatbot to generate a response given the conversation history", 'parameter_definitions': {} } ] # render the tool use prompt as a string: tool_use_prompt = tokenizer.apply_tool_use_template( conversation, tools=tools, tokenize=False, add_generation_prompt=True, ) print(tool_use_prompt) ``` </details> <details> <summary><b>Example Rendered Tool Use Prompt [CLICK TO EXPAND]</b></summary> ```` <BOS_TOKEN><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|># Safety Preamble The instructions in this section override those in the task description and style guide sections. Don't answer questions that are harmful or immoral. # System Preamble ## Basic Rules You are a powerful conversational AI trained by Cohere to help people. You are augmented by a number of tools, and your job is to use and consume the output of these tools to best help the user. You will see a conversation history between yourself and a user, ending with an utterance from the user. You will then see a specific instruction instructing you what kind of response to generate. When you answer the user's requests, you cite your sources in your answers, according to those instructions. # User Preamble ## Task and Context You help people answer their questions and other requests interactively. You will be asked a very wide array of requests on all kinds of topics. You will be equipped with a wide range of search engines or similar tools to help you, which you use to research your answer. You should focus on serving the user's needs as best you can, which will be wide-ranging. ## Style Guide Unless the user asks for a different style of answer, you should answer in full sentences, using proper grammar and spelling. ## Available Tools Here is a list of tools that you have available to you: ```python def internet_search(query: str) -> List[Dict]: """Returns a list of relevant document snippets for a textual query retrieved from the internet Args: query (str): Query to search the internet with """ pass ``` ```python def directly_answer() -> List[Dict]: """Calls a standard (un-augmented) AI chatbot to generate a response given the conversation history """ pass ```<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|USER_TOKEN\|>Whats the biggest penguin in the world?<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|>Write 'Action:' followed by a json-formatted list of actions that you want to perform in order to produce a good response to the user's last input. You can use any of the supplied tools any number of times, but you should aim to execute the minimum number of necessary actions for the input. You should use the `directly-answer` tool if calling the other tools is unnecessary. The list of actions you want to call should be formatted as a list of json objects, for example: ```json [ { "tool_name": title of the tool in the specification, "parameters": a dict of parameters to input into the tool as they are defined in the specs, or {} if it takes no parameters } ]```<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|CHATBOT_TOKEN\|> ```` </details> <details> <summary><b>Example Rendered Tool Use Completion [CLICK TO EXPAND]</b></summary> ```` Action: ```json [ { "tool_name": "internet_search", "parameters": { "query": "biggest penguin in the world" } } ] ``` ```` </details> ### Grounded Generation and RAG Capabilities: Command-R has been specifically trained with grounded generation capabilities. This means that it can generate responses based on a list of supplied document snippets, and it will include grounding spans (citations) in its response indicating the source of the information. This can be used to enable behaviors such as grounded summarization and the final step of Retrieval Augmented Generation (RAG).This behavior has been trained into the model via a mixture of supervised fine-tuning and preference fine-tuning, using a specific prompt template. Deviating from this prompt template may reduce performance, but we encourage experimentation. Command-R’s grounded generation behavior takes a conversation as input (with an optional user-supplied system preamble, indicating task, context and desired output style), along with a list of retrieved document snippets. The document snippets should be chunks, rather than long documents, typically around 100-400 words per chunk. Document snippets consist of key-value pairs. The keys should be short descriptive strings, the values can be text or semi-structured. By default, Command-R will generate grounded responses by first predicting which documents are relevant, then predicting which ones it will cite, then generating an answer. Finally, it will then insert grounding spans into the answer. See below for an example. This is referred to as `accurate` grounded generation. The model is trained with a number of other answering modes, which can be selected by prompt changes . A `fast` citation mode is supported in the tokenizer, which will directly generate an answer with grounding spans in it, without first writing the answer out in full. This sacrifices some grounding accuracy in favor of generating fewer tokens. Comprehensive documentation for working with command-R's grounded generation prompt template can be found [here](https://docs.cohere.com/docs/prompting-command-r). The code snippet below shows a minimal working example on how to render a prompt. <details> <summary> <b>Usage: Rendering Grounded Generation prompts [CLICK TO EXPAND]</b> </summary> ````python from transformers import AutoTokenizer model_id = "CohereForAI/c4ai-command-r-v01" tokenizer = AutoTokenizer.from_pretrained(model_id) # define conversation input: conversation = [ {"role": "user", "content": "Whats the biggest penguin in the world?"} ] # define documents to ground on: documents = [ { "title": "Tall penguins", "text": "Emperor penguins are the tallest growing up to 122 cm in height." }, { "title": "Penguin habitats", "text": "Emperor penguins only live in Antarctica."} ] # render the tool use prompt as a string: grounded_generation_prompt = tokenizer.apply_grounded_generation_template( conversation, documents=documents, citation_mode="accurate", # or "fast" tokenize=False, add_generation_prompt=True, ) print(grounded_generation_prompt) ```` </details> <details> <summary><b>Example Rendered Grounded Generation Prompt [CLICK TO EXPAND]</b></summary> ````<BOS_TOKEN><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|># Safety Preamble The instructions in this section override those in the task description and style guide sections. Don't answer questions that are harmful or immoral. # System Preamble ## Basic Rules You are a powerful conversational AI trained by Cohere to help people. You are augmented by a number of tools, and your job is to use and consume the output of these tools to best help the user. You will see a conversation history between yourself and a user, ending with an utterance from the user. You will then see a specific instruction instructing you what kind of response to generate. When you answer the user's requests, you cite your sources in your answers, according to those instructions. # User Preamble ## Task and Context You help people answer their questions and other requests interactively. You will be asked a very wide array of requests on all kinds of topics. You will be equipped with a wide range of search engines or similar tools to help you, which you use to research your answer. You should focus on serving the user's needs as best you can, which will be wide-ranging. ## Style Guide Unless the user asks for a different style of answer, you should answer in full sentences, using proper grammar and spelling.<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|USER_TOKEN\|>Whats the biggest penguin in the world?<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|><results> Document: 0 title: Tall penguins text: Emperor penguins are the tallest growing up to 122 cm in height. Document: 1 title: Penguin habitats text: Emperor penguins only live in Antarctica. </results><\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|SYSTEM_TOKEN\|>Carefully perform the following instructions, in order, starting each with a new line. Firstly, Decide which of the retrieved documents are relevant to the user's last input by writing 'Relevant Documents:' followed by comma-separated list of document numbers. If none are relevant, you should instead write 'None'. Secondly, Decide which of the retrieved documents contain facts that should be cited in a good answer to the user's last input by writing 'Cited Documents:' followed a comma-separated list of document numbers. If you dont want to cite any of them, you should instead write 'None'. Thirdly, Write 'Answer:' followed by a response to the user's last input in high quality natural english. Use the retrieved documents to help you. Do not insert any citations or grounding markup. Finally, Write 'Grounded answer:' followed by a response to the user's last input in high quality natural english. Use the symbols <co: doc> and </co: doc> to indicate when a fact comes from a document in the search result, e.g <co: 0>my fact</co: 0> for a fact from document 0.<\|END_OF_TURN_TOKEN\|><\|START_OF_TURN_TOKEN\|><\|CHATBOT_TOKEN\|> ```` </details> <details> <summary><b>Example Rendered Grounded Generation Completion [CLICK TO EXPAND]</b></summary> ```` Relevant Documents: 0,1 Cited Documents: 0,1 Answer: The Emperor Penguin is the tallest or biggest penguin in the world. It is a bird that lives only in Antarctica and grows to a height of around 122 centimetres. Grounded answer: The <co: 0>Emperor Penguin</co: 0> is the <co: 0>tallest</co: 0> or biggest penguin in the world. It is a bird that <co: 1>lives only in Antarctica</co: 1> and <co: 0>grows to a height of around 122 centimetres.</co: 0> ```` </details> ### Code Capabilities: Command-R has been optimized to interact with your code, by requesting code snippets, code explanations, or code rewrites. It might not perform well out-of-the-box for pure code completion. For better performance, we also recommend using a low temperature (and even greedy decoding) for code-generation related instructions. ### Model Card Contact For errors or additional questions about details in this model card, contact [[email protected]](mailto:[email protected]). ### Terms of Use: We hope that the release of this model will make community-based research efforts more accessible, by releasing the weights of a highly performant 35 billion parameter model to researchers all over the world. This model is governed by a [CC-BY-NC](https://cohere.com/c4ai-cc-by-nc-license) License with an acceptable use addendum, and also requires adhering to [C4AI's Acceptable Use Policy](https://docs.cohere.com/docs/c4ai-acceptable-use-policy). ### Try Chat: You can try Command-R chat in the playground [here](https://dashboard.cohere.com/playground/chat).	{"language": ["en", "fr", "de", "es", "it", "pt", "ja", "ko", "zh", "ar"], "library_name": "transformers", "license": "cc-by-nc-4.0"}	task	[ "QUESTION_ANSWERING", "SUMMARIZATION" ]	43,108
4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base	4yo1	translation	[ "transformers", "safetensors", "llama", "text-generation", "translation", "en", "ko", "dataset:pre", "license:mit", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "region:us" ]	2024-08-06T05:52:45Z	2024-08-06T07:19:31+00:00	2,064	0	--- datasets: - pre language: - en - ko library_name: transformers license: mit pipeline_tag: translation --- ### Model Card for Model ID ### Model Details Model Card: sapie with Fine-Tuning Model Overview Model Name: 4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base Model Type: Transformer-based Language Model Model Size: 8 billion parameters by: 4yo1 Languages: English and Korean ### how to use - sample code ```python from transformers import AutoConfig, AutoModel, AutoTokenizer config = AutoConfig.from_pretrained("4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base") model = AutoModel.from_pretrained("4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base") tokenizer = AutoTokenizer.from_pretrained("4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base") ``` datasets: - 140kgpt license: mit	null	Non_BioNLP	### Model Card for Model ID ### Model Details Model Card: sapie with Fine-Tuning Model Overview Model Name: 4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base Model Type: Transformer-based Language Model Model Size: 8 billion parameters by: 4yo1 Languages: English and Korean ### how to use - sample code ```python from transformers import AutoConfig, AutoModel, AutoTokenizer config = AutoConfig.from_pretrained("4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base") model = AutoModel.from_pretrained("4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base") tokenizer = AutoTokenizer.from_pretrained("4yo1/llama3-pre1-pre2-inst3-lora3-mergkit-base") ``` datasets: - 140kgpt license: mit	{"datasets": ["pre"], "language": ["en", "ko"], "library_name": "transformers", "license": "mit", "pipeline_tag": "translation"}	task	[ "TRANSLATION" ]	43,109
lmstudio-community/Llama-3.2-1B-Instruct-GGUF	lmstudio-community	text-generation	[ "transformers", "gguf", "facebook", "meta", "pytorch", "llama", "llama-3", "text-generation", "en", "de", "fr", "it", "pt", "hi", "es", "th", "base_model:meta-llama/Llama-3.2-1B-Instruct", "base_model:quantized:meta-llama/Llama-3.2-1B-Instruct", "license:llama3.2", "endpoints_compatible", "region:us", "conversational" ]	2024-09-25T18:35:25Z	2024-09-25T20:09:46+00:00	16,469	38	--- base_model: meta-llama/Llama-3.2-1B-Instruct language: - en - de - fr - it - pt - hi - es - th library_name: transformers license: llama3.2 pipeline_tag: text-generation tags: - facebook - meta - pytorch - llama - llama-3 quantized_by: bartowski extra_gated_prompt: "### LLAMA 3.2 COMMUNITY LICENSE AGREEMENT\n\nLlama 3.2 Version\ \ Release Date: September 25, 2024\n\n“Agreement” means the terms and conditions\ \ for use, reproduction, distribution and modification of the Llama Materials set\ \ forth herein.\n\n“Documentation” means the specifications, manuals and documentation\ \ accompanying Llama 3.2 distributed by Meta at https://llama.meta.com/doc/overview.\n\ \n“Licensee” or “you” means you, or your employer or any other person or entity\ \ (if you are entering into this Agreement on such person or entity’s behalf),\ \ of the age required under applicable laws, rules or regulations to provide legal\ \ consent and that has legal authority to bind your employer or such other person\ \ or entity if you are entering in this Agreement on their behalf.\n\n“Llama 3.2”\ \ means the foundational large language models and software and algorithms, including\ \ machine-learning model code, trained model weights, inference-enabling code, training-enabling\ \ code, fine-tuning enabling code and other elements of the foregoing distributed\ \ by Meta at https://www.llama.com/llama-downloads.\n\n“Llama Materials” means,\ \ collectively, Meta’s proprietary Llama 3.2 and Documentation (and any portion\ \ thereof) made available under this Agreement.\n\n“Meta” or “we” means Meta Platforms\ \ Ireland Limited (if you are located in or, if you are an entity, your principal\ \ place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if\ \ you are located outside of the EEA or Switzerland). \n\nBy clicking “I Accept”\ \ below or by using or distributing any portion or element of the Llama Materials,\ \ you agree to be bound by this Agreement.\n\n1. License Rights and Redistribution.\n\ a. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable\ \ and royalty-free limited license under Meta’s intellectual property or other rights\ \ owned by Meta embodied in the Llama Materials to use, reproduce, distribute,\ \ copy, create derivative works of, and make modifications to the Llama Materials.\ \ \nb. Redistribution and Use. \ni. If you distribute or make available the Llama\ \ Materials (or any derivative works thereof), or a product or service (including\ \ another AI model) that contains any of them, you shall (A) provide a copy of this\ \ Agreement with any such Llama Materials; and (B) prominently display “Built with\ \ Llama” on a related website, user interface, blogpost, about page, or product\ \ documentation. If you use the Llama Materials or any outputs or results of the\ \ Llama Materials to create, train, fine tune, or otherwise improve an AI model,\ \ which is distributed or made available, you shall also include “Llama” at the\ \ beginning of any such AI model name.\nii. If you receive Llama Materials, or any\ \ derivative works thereof, from a Licensee as part of an integrated end user product,\ \ then Section 2 of this Agreement will not apply to you. \niii. You must retain\ \ in all copies of the Llama Materials that you distribute the following attribution\ \ notice within a “Notice” text file distributed as a part of such copies: “Llama\ \ 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms,\ \ Inc. All Rights Reserved.”\niv. Your use of the Llama Materials must comply with\ \ applicable laws and regulations (including trade compliance laws and regulations)\ \ and adhere to the Acceptable Use Policy for the Llama Materials (available at\ \ https://www.llama.com/llama3_2/use-policy), which is hereby incorporated by reference\ \ into this Agreement.\n \n2. Additional Commercial Terms. If, on the Llama 3.2\ \ version release date, the monthly active users of the products or services made\ \ available by or for Licensee, or Licensee’s affiliates, is greater than 700 million\ \ monthly active users in the preceding calendar month, you must request a license\ \ from Meta, which Meta may grant to you in its sole discretion, and you are not\ \ authorized to exercise any of the rights under this Agreement unless or until\ \ Meta otherwise expressly grants you such rights.\n3. Disclaimer of Warranty. UNLESS\ \ REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM\ \ ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, AND META DISCLAIMS\ \ ALL WARRANTIES OF ANY KIND, BOTH EXPRESS AND IMPLIED, INCLUDING, WITHOUT LIMITATION,\ \ ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR\ \ PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING\ \ OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR\ \ USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS.\n4. Limitation of Liability.\ \ IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY,\ \ WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING\ \ OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL,\ \ INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE\ \ BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING.\n5. Intellectual Property.\n\ a. No trademark licenses are granted under this Agreement, and in connection with\ \ the Llama Materials, neither Meta nor Licensee may use any name or mark owned\ \ by or associated with the other or any of its affiliates, except as required\ \ for reasonable and customary use in describing and redistributing the Llama Materials\ \ or as set forth in this Section 5(a). Meta hereby grants you a license to use\ \ “Llama” (the “Mark”) solely as required to comply with the last sentence of Section\ \ 1.b.i. You will comply with Meta’s brand guidelines (currently accessible at\ \ https://about.meta.com/brand/resources/meta/company-brand/). All goodwill arising\ \ out of your use of the Mark will inure to the benefit of Meta.\nb. Subject to\ \ Meta’s ownership of Llama Materials and derivatives made by or for Meta, with\ \ respect to any derivative works and modifications of the Llama Materials that\ \ are made by you, as between you and Meta, you are and will be the owner of such\ \ derivative works and modifications.\nc. If you institute litigation or other proceedings\ \ against Meta or any entity (including a cross-claim or counterclaim in a lawsuit)\ \ alleging that the Llama Materials or Llama 3.2 outputs or results, or any portion\ \ of any of the foregoing, constitutes infringement of intellectual property or\ \ other rights owned or licensable by you, then any licenses granted to you under\ \ this Agreement shall terminate as of the date such litigation or claim is filed\ \ or instituted. You will indemnify and hold harmless Meta from and against any\ \ claim by any third party arising out of or related to your use or distribution\ \ of the Llama Materials.\n6. Term and Termination. The term of this Agreement will\ \ commence upon your acceptance of this Agreement or access to the Llama Materials\ \ and will continue in full force and effect until terminated in accordance with\ \ the terms and conditions herein. Meta may terminate this Agreement if you are\ \ in breach of any term or condition of this Agreement. Upon termination of this\ \ Agreement, you shall delete and cease use of the Llama Materials. Sections 3,\ \ 4 and 7 shall survive the termination of this Agreement. \n7. Governing Law and\ \ Jurisdiction. This Agreement will be governed and construed under the laws of\ \ the State of California without regard to choice of law principles, and the UN\ \ Convention on Contracts for the International Sale of Goods does not apply to\ \ this Agreement. The courts of California shall have exclusive jurisdiction of\ \ any dispute arising out of this Agreement. \n### Llama 3.2 Acceptable Use Policy\n\ Meta is committed to promoting safe and fair use of its tools and features, including\ \ Llama 3.2. If you access or use Llama 3.2, you agree to this Acceptable Use Policy\ \ (“Policy”). The most recent copy of this policy can be found at [https://www.llama.com/llama3_2/use-policy](https://www.llama.com/llama3_2/use-policy).\n\ #### Prohibited Uses\nWe want everyone to use Llama 3.2 safely and responsibly.\ \ You agree you will not use, or allow others to use, Llama 3.2 to:\n1. Violate\ \ the law or others’ rights, including to:\n 1. Engage in, promote, generate,\ \ contribute to, encourage, plan, incite, or further illegal or unlawful activity\ \ or content, such as:\n 1. Violence or terrorism\n 2. Exploitation\ \ or harm to children, including the solicitation, creation, acquisition, or dissemination\ \ of child exploitative content or failure to report Child Sexual Abuse Material\n\ \ 3. Human trafficking, exploitation, and sexual violence\n 4. The\ \ illegal distribution of information or materials to minors, including obscene\ \ materials, or failure to employ legally required age-gating in connection with\ \ such information or materials.\n 5. Sexual solicitation\n 6. Any\ \ other criminal activity\n 1. Engage in, promote, incite, or facilitate the\ \ harassment, abuse, threatening, or bullying of individuals or groups of individuals\n\ \ 2. Engage in, promote, incite, or facilitate discrimination or other unlawful\ \ or harmful conduct in the provision of employment, employment benefits, credit,\ \ housing, other economic benefits, or other essential goods and services\n 3.\ \ Engage in the unauthorized or unlicensed practice of any profession including,\ \ but not limited to, financial, legal, medical/health, or related professional\ \ practices\n 4. Collect, process, disclose, generate, or infer private or sensitive\ \ information about individuals, including information about individuals’ identity,\ \ health, or demographic information, unless you have obtained the right to do so\ \ in accordance with applicable law\n 5. Engage in or facilitate any action or\ \ generate any content that infringes, misappropriates, or otherwise violates any\ \ third-party rights, including the outputs or results of any products or services\ \ using the Llama Materials\n 6. Create, generate, or facilitate the creation\ \ of malicious code, malware, computer viruses or do anything else that could disable,\ \ overburden, interfere with or impair the proper working, integrity, operation\ \ or appearance of a website or computer system\n 7. Engage in any action, or\ \ facilitate any action, to intentionally circumvent or remove usage restrictions\ \ or other safety measures, or to enable functionality disabled by Meta \n2. Engage\ \ in, promote, incite, facilitate, or assist in the planning or development of activities\ \ that present a risk of death or bodily harm to individuals, including use of Llama\ \ 3.2 related to the following:\n 8. Military, warfare, nuclear industries or\ \ applications, espionage, use for materials or activities that are subject to the\ \ International Traffic Arms Regulations (ITAR) maintained by the United States\ \ Department of State or to the U.S. Biological Weapons Anti-Terrorism Act of 1989\ \ or the Chemical Weapons Convention Implementation Act of 1997\n 9. Guns and\ \ illegal weapons (including weapon development)\n 10. Illegal drugs and regulated/controlled\ \ substances\n 11. Operation of critical infrastructure, transportation technologies,\ \ or heavy machinery\n 12. Self-harm or harm to others, including suicide, cutting,\ \ and eating disorders\n 13. Any content intended to incite or promote violence,\ \ abuse, or any infliction of bodily harm to an individual\n3. Intentionally deceive\ \ or mislead others, including use of Llama 3.2 related to the following:\n 14.\ \ Generating, promoting, or furthering fraud or the creation or promotion of disinformation\n\ \ 15. Generating, promoting, or furthering defamatory content, including the\ \ creation of defamatory statements, images, or other content\n 16. Generating,\ \ promoting, or further distributing spam\n 17. Impersonating another individual\ \ without consent, authorization, or legal right\n 18. Representing that the\ \ use of Llama 3.2 or outputs are human-generated\n 19. Generating or facilitating\ \ false online engagement, including fake reviews and other means of fake online\ \ engagement \n4. Fail to appropriately disclose to end users any known dangers\ \ of your AI system 5. Interact with third party tools, models, or software designed\ \ to generate unlawful content or engage in unlawful or harmful conduct and/or represent\ \ that the outputs of such tools, models, or software are associated with Meta or\ \ Llama 3.2\n\nWith respect to any multimodal models included in Llama 3.2, the\ \ rights granted under Section 1(a) of the Llama 3.2 Community License Agreement\ \ are not being granted to you if you are an individual domiciled in, or a company\ \ with a principal place of business in, the European Union. This restriction does\ \ not apply to end users of a product or service that incorporates any such multimodal\ \ models.\n\nPlease report any violation of this Policy, software “bug,” or other\ \ problems that could lead to a violation of this Policy through one of the following\ \ means:\n\n* Reporting issues with the model: [https://github.com/meta-llama/llama-models/issues](https://l.workplace.com/l.php?u=https%3A%2F%2Fgithub.com%2Fmeta-llama%2Fllama-models%2Fissues&h=AT0qV8W9BFT6NwihiOHRuKYQM_UnkzN_NmHMy91OT55gkLpgi4kQupHUl0ssR4dQsIQ8n3tfd0vtkobvsEvt1l4Ic6GXI2EeuHV8N08OG2WnbAmm0FL4ObkazC6G_256vN0lN9DsykCvCqGZ)\n\ * Reporting risky content generated by the model: [developers.facebook.com/llama_output_feedback](http://developers.facebook.com/llama_output_feedback)\n\ * Reporting bugs and security concerns: [facebook.com/whitehat/info](http://facebook.com/whitehat/info)\n\ * Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama\ \ 3.2: [email protected]" extra_gated_fields: First Name: text Last Name: text Date of birth: date_picker Country: country Affiliation: text Job title: type: select options: - Student - Research Graduate - AI researcher - AI developer/engineer - Reporter - Other geo: ip_location ? By clicking Submit below I accept the terms of the license and acknowledge that the information I provide will be collected stored processed and shared in accordance with the Meta Privacy Policy : checkbox extra_gated_description: The information you provide will be collected, stored, processed and shared in accordance with the [Meta Privacy Policy](https://www.facebook.com/privacy/policy/). extra_gated_button_content: Submit --- ## 💫 Community Model> Llama 3.2 1B Instruct by Meta-Llama 👾 [LM Studio](https://lmstudio.ai) Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on [Discord](https://discord.gg/aPQfnNkxGC). Model creator: [meta-llama](https://huggingface.co/meta-llama)<br> Original model: [Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct)<br> GGUF quantization: provided by [bartowski](https://huggingface.co/bartowski) based on `llama.cpp` release [b3821](https://github.com/ggerganov/llama.cpp/releases/tag/b3821)<br> ## Technical Details Llama 3.2 is optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. Officially supports English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai languages, but is trained on even more. 128K context length support ## Special thanks 🙏 Special thanks to [Georgi Gerganov](https://github.com/ggerganov) and the whole team working on [llama.cpp](https://github.com/ggerganov/llama.cpp/) for making all of this possible. ## Disclaimers LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful, inaccurate or otherwise inappropriate, or deceptive. Each Community Model is the sole responsibility of the person or entity who originated such Model. LM Studio may not monitor or control the Community Models and cannot, and does not, take responsibility for any such Model. LM Studio disclaims all warranties or guarantees about the accuracy, reliability or benefits of the Community Models. LM Studio further disclaims any warranty that the Community Model will meet your requirements, be secure, uninterrupted or available at any time or location, or error-free, viruses-free, or that any errors will be corrected, or otherwise. You will be solely responsible for any damage resulting from your use of or access to the Community Models, your downloading of any Community Model, or use of any other Community Model provided by or through LM Studio.	null	Non_BioNLP	## 💫 Community Model> Llama 3.2 1B Instruct by Meta-Llama 👾 [LM Studio](https://lmstudio.ai) Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on [Discord](https://discord.gg/aPQfnNkxGC). Model creator: [meta-llama](https://huggingface.co/meta-llama)<br> Original model: [Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct)<br> GGUF quantization: provided by [bartowski](https://huggingface.co/bartowski) based on `llama.cpp` release [b3821](https://github.com/ggerganov/llama.cpp/releases/tag/b3821)<br> ## Technical Details Llama 3.2 is optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. Officially supports English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai languages, but is trained on even more. 128K context length support ## Special thanks 🙏 Special thanks to [Georgi Gerganov](https://github.com/ggerganov) and the whole team working on [llama.cpp](https://github.com/ggerganov/llama.cpp/) for making all of this possible. ## Disclaimers LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful, inaccurate or otherwise inappropriate, or deceptive. Each Community Model is the sole responsibility of the person or entity who originated such Model. LM Studio may not monitor or control the Community Models and cannot, and does not, take responsibility for any such Model. LM Studio disclaims all warranties or guarantees about the accuracy, reliability or benefits of the Community Models. LM Studio further disclaims any warranty that the Community Model will meet your requirements, be secure, uninterrupted or available at any time or location, or error-free, viruses-free, or that any errors will be corrected, or otherwise. You will be solely responsible for any damage resulting from your use of or access to the Community Models, your downloading of any Community Model, or use of any other Community Model provided by or through LM Studio.	{"base_model": "meta-llama/Llama-3.2-1B-Instruct", "language": ["en", "de", "fr", "it", "pt", "hi", "es", "th"], "library_name": "transformers", "license": "llama3.2", "pipeline_tag": "text-generation", "tags": ["facebook", "meta", "pytorch", "llama", "llama-3"], "quantized_by": "bartowski", "extra_gated_prompt": "### LLAMA 3.2 COMMUNITY LICENSE AGREEMENT\n\nLlama 3.2 Version Release Date: September 25, 2024\n\n“Agreement” means the terms and conditions for use, reproduction, distribution and modification of the Llama Materials set forth herein.\n\n“Documentation” means the specifications, manuals and documentation accompanying Llama 3.2 distributed by Meta at https://llama.meta.com/doc/overview.\n\n“Licensee” or “you” means you, or your employer or any other person or entity (if you are entering into this Agreement on such person or entity’s behalf), of the age required under applicable laws, rules or regulations to provide legal consent and that has legal authority to bind your employer or such other person or entity if you are entering in this Agreement on their behalf.\n\n“Llama 3.2” means the foundational large language models and software and algorithms, including machine-learning model code, trained model weights, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by Meta at https://www.llama.com/llama-downloads.\n\n“Llama Materials” means, collectively, Meta’s proprietary Llama 3.2 and Documentation (and any portion thereof) made available under this Agreement.\n\n“Meta” or “we” means Meta Platforms Ireland Limited (if you are located in or, if you are an entity, your principal place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if you are located outside of the EEA or Switzerland). \n\nBy clicking “I Accept” below or by using or distributing any portion or element of the Llama Materials, you agree to be bound by this Agreement.\n\n1. License Rights and Redistribution.\na. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under Meta’s intellectual property or other rights owned by Meta embodied in the Llama Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Llama Materials. \nb. Redistribution and Use. \ni. If you distribute or make available the Llama Materials (or any derivative works thereof), or a product or service (including another AI model) that contains any of them, you shall (A) provide a copy of this Agreement with any such Llama Materials; and (B) prominently display “Built with Llama” on a related website, user interface, blogpost, about page, or product documentation. If you use the Llama Materials or any outputs or results of the Llama Materials to create, train, fine tune, or otherwise improve an AI model, which is distributed or made available, you shall also include “Llama” at the beginning of any such AI model name.\nii. If you receive Llama Materials, or any derivative works thereof, from a Licensee as part of an integrated end user product, then Section 2 of this Agreement will not apply to you. \niii. You must retain in all copies of the Llama Materials that you distribute the following attribution notice within a “Notice” text file distributed as a part of such copies: “Llama 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.”\niv. Your use of the Llama Materials must comply with applicable laws and regulations (including trade compliance laws and regulations) and adhere to the Acceptable Use Policy for the Llama Materials (available at https://www.llama.com/llama3_2/use-policy), which is hereby incorporated by reference into this Agreement.\n \n2. Additional Commercial Terms. If, on the Llama 3.2 version release date, the monthly active users of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 700 million monthly active users in the preceding calendar month, you must request a license from Meta, which Meta may grant to you in its sole discretion, and you are not authorized to exercise any of the rights under this Agreement unless or until Meta otherwise expressly grants you such rights.\n3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, AND META DISCLAIMS ALL WARRANTIES OF ANY KIND, BOTH EXPRESS AND IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS.\n4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING.\n5. Intellectual Property.\na. No trademark licenses are granted under this Agreement, and in connection with the Llama Materials, neither Meta nor Licensee may use any name or mark owned by or associated with the other or any of its affiliates, except as required for reasonable and customary use in describing and redistributing the Llama Materials or as set forth in this Section 5(a). Meta hereby grants you a license to use “Llama” (the “Mark”) solely as required to comply with the last sentence of Section 1.b.i. You will comply with Meta’s brand guidelines (currently accessible at https://about.meta.com/brand/resources/meta/company-brand/). All goodwill arising out of your use of the Mark will inure to the benefit of Meta.\nb. Subject to Meta’s ownership of Llama Materials and derivatives made by or for Meta, with respect to any derivative works and modifications of the Llama Materials that are made by you, as between you and Meta, you are and will be the owner of such derivative works and modifications.\nc. If you institute litigation or other proceedings against Meta or any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Llama Materials or Llama 3.2 outputs or results, or any portion of any of the foregoing, constitutes infringement of intellectual property or other rights owned or licensable by you, then any licenses granted to you under this Agreement shall terminate as of the date such litigation or claim is filed or instituted. You will indemnify and hold harmless Meta from and against any claim by any third party arising out of or related to your use or distribution of the Llama Materials.\n6. Term and Termination. The term of this Agreement will commence upon your acceptance of this Agreement or access to the Llama Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein. Meta may terminate this Agreement if you are in breach of any term or condition of this Agreement. Upon termination of this Agreement, you shall delete and cease use of the Llama Materials. Sections 3, 4 and 7 shall survive the termination of this Agreement. \n7. Governing Law and Jurisdiction. This Agreement will be governed and construed under the laws of the State of California without regard to choice of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement. The courts of California shall have exclusive jurisdiction of any dispute arising out of this Agreement. \n### Llama 3.2 Acceptable Use Policy\nMeta is committed to promoting safe and fair use of its tools and features, including Llama 3.2. If you access or use Llama 3.2, you agree to this Acceptable Use Policy (“Policy”). The most recent copy of this policy can be found at [https://www.llama.com/llama3_2/use-policy](https://www.llama.com/llama3_2/use-policy).\n#### Prohibited Uses\nWe want everyone to use Llama 3.2 safely and responsibly. You agree you will not use, or allow others to use, Llama 3.2 to:\n1. Violate the law or others’ rights, including to:\n 1. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as:\n 1. Violence or terrorism\n 2. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material\n 3. Human trafficking, exploitation, and sexual violence\n 4. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials.\n 5. Sexual solicitation\n 6. Any other criminal activity\n 1. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals\n 2. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services\n 3. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices\n 4. Collect, process, disclose, generate, or infer private or sensitive information about individuals, including information about individuals’ identity, health, or demographic information, unless you have obtained the right to do so in accordance with applicable law\n 5. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama Materials\n 6. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system\n 7. Engage in any action, or facilitate any action, to intentionally circumvent or remove usage restrictions or other safety measures, or to enable functionality disabled by Meta \n2. Engage in, promote, incite, facilitate, or assist in the planning or development of activities that present a risk of death or bodily harm to individuals, including use of Llama 3.2 related to the following:\n 8. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State or to the U.S. Biological Weapons Anti-Terrorism Act of 1989 or the Chemical Weapons Convention Implementation Act of 1997\n 9. Guns and illegal weapons (including weapon development)\n 10. Illegal drugs and regulated/controlled substances\n 11. Operation of critical infrastructure, transportation technologies, or heavy machinery\n 12. Self-harm or harm to others, including suicide, cutting, and eating disorders\n 13. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual\n3. Intentionally deceive or mislead others, including use of Llama 3.2 related to the following:\n 14. Generating, promoting, or furthering fraud or the creation or promotion of disinformation\n 15. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content\n 16. Generating, promoting, or further distributing spam\n 17. Impersonating another individual without consent, authorization, or legal right\n 18. Representing that the use of Llama 3.2 or outputs are human-generated\n 19. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement \n4. Fail to appropriately disclose to end users any known dangers of your AI system 5. Interact with third party tools, models, or software designed to generate unlawful content or engage in unlawful or harmful conduct and/or represent that the outputs of such tools, models, or software are associated with Meta or Llama 3.2\n\nWith respect to any multimodal models included in Llama 3.2, the rights granted under Section 1(a) of the Llama 3.2 Community License Agreement are not being granted to you if you are an individual domiciled in, or a company with a principal place of business in, the European Union. This restriction does not apply to end users of a product or service that incorporates any such multimodal models.\n\nPlease report any violation of this Policy, software “bug,” or other problems that could lead to a violation of this Policy through one of the following means:\n\n* Reporting issues with the model: [https://github.com/meta-llama/llama-models/issues](https://l.workplace.com/l.php?u=https%3A%2F%2Fgithub.com%2Fmeta-llama%2Fllama-models%2Fissues&h=AT0qV8W9BFT6NwihiOHRuKYQM_UnkzN_NmHMy91OT55gkLpgi4kQupHUl0ssR4dQsIQ8n3tfd0vtkobvsEvt1l4Ic6GXI2EeuHV8N08OG2WnbAmm0FL4ObkazC6G_256vN0lN9DsykCvCqGZ)\n* Reporting risky content generated by the model: [developers.facebook.com/llama_output_feedback](http://developers.facebook.com/llama_output_feedback)\n* Reporting bugs and security concerns: [facebook.com/whitehat/info](http://facebook.com/whitehat/info)\n* Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama 3.2: [email protected]", "extra_gated_fields": {"First Name": "text", "Last Name": "text", "Date of birth": "date_picker", "Country": "country", "Affiliation": "text", "Job title": {"type": "select", "options": ["Student", "Research Graduate", "AI researcher", "AI developer/engineer", "Reporter", "Other"]}, "geo": "ip_location", "By clicking Submit below I accept the terms of the license and acknowledge that the information I provide will be collected stored processed and shared in accordance with the Meta Privacy Policy": "checkbox"}, "extra_gated_description": "The information you provide will be collected, stored, processed and shared in accordance with the [Meta Privacy Policy](https://www.facebook.com/privacy/policy/).", "extra_gated_button_content": "Submit"}	task	[ "SUMMARIZATION" ]	43,110
gaudi/opus-mt-fi-crs-ctranslate2	gaudi	translation	[ "transformers", "marian", "ctranslate2", "translation", "license:apache-2.0", "endpoints_compatible", "region:us" ]	2024-07-22T15:50:12Z	2024-10-19T03:23:43+00:00	6	0	--- license: apache-2.0 tags: - ctranslate2 - translation --- # Repository General Information ## Inspired by and derived from the work of [Helsinki-NLP](https://huggingface.co/Helsinki-NLP), [CTranslate2](https://github.com/OpenNMT/CTranslate2), and [michaelfeil](https://huggingface.co/michaelfeil)! - Link to Original Model ([Helsinki-NLP](https://huggingface.co/Helsinki-NLP)): [Model Link](https://huggingface.co/Helsinki-NLP/opus-mt-fi-crs) - This respository was based on the work of [CTranslate2](https://github.com/OpenNMT/CTranslate2). - This repository was based on the work of [michaelfeil](https://huggingface.co/michaelfeil). # What is CTranslate2? [CTranslate2](https://opennmt.net/CTranslate2/) is a C++ and Python library for efficient inference with Transformer models. CTranslate2 implements a custom runtime that applies many performance optimization techniques such as weights quantization, layers fusion, batch reordering, etc., to accelerate and reduce the memory usage of Transformer models on CPU and GPU. CTranslate2 is one of the most performant ways of hosting translation models at scale. Current supported models include: - Encoder-decoder models: Transformer base/big, M2M-100, NLLB, BART, mBART, Pegasus, T5, Whisper - Decoder-only models: GPT-2, GPT-J, GPT-NeoX, OPT, BLOOM, MPT, Llama, Mistral, Gemma, CodeGen, GPTBigCode, Falcon - Encoder-only models: BERT, DistilBERT, XLM-RoBERTa The project is production-oriented and comes with backward compatibility guarantees, but it also includes experimental features related to model compression and inference acceleration. # CTranslate2 Benchmarks Please note that the results presented below are only valid for the configuration used during this benchmark: absolute and relative performance may change with different settings. Tested against `newstest2014` (En -> De) dataset. The benchmark reports the number of target tokens generated per second (higher is better). The results are aggregated over multiple runs. See the benchmark scripts for more details and reproduce these numbers. Please note that the results presented below are only valid for the configuration used during this benchmark: absolute and relative performance may change with different settings. ## CPU Benchmarks for Generic Opus-MT Models \| Library \| Tokens per Second \| Max Memory Usage \| BLEU \| \| :----: \| :----: \| :----: \| :----: \| \| Transformers 4.26.1 (with PyTorch 1.13.1) \| 147.3 \| 2332MB \| 27.90 \| \| Marian 1.11.0 (int16) \| 330.2 \| 5901MB \| 27.65 \| \| Marian 1.11.0 (int8) \| 355.8 \| 4763MB \| 27.27 \| \| CTranslate2 3.6.0 (int16) \| 596.1 \| 660MB \| 27.53 \| \| CTranslate2 3.6.0 (int8) \| 696.1 \| 516MB \| 27.65 \| ## GPU Benchmarks for Generic Opus-MT Models \| Library \| Tokens per Second \| Max GPU Memory Usage \| Max Memory Usage \| BLEU \| \| :----: \| :----: \| :----: \| :----: \| :----: \| \| Transformers 4.26.1 (with PyTorch 1.13.1) \| 1022.9 \| 4097MB \| 2109MB \| 27.90 \| \| Marian 1.11.0 (float16) \| 3962.4 \| 3239MB \| 1976MB \| 27.94 \| \| CTranslate2 3.6.0 (float16) \| 9296.7 \| 909MB \| 814MB \| 27.9 \| \| CTranslate2 3.6.0 (int8 + float16) \| 8362.7 \| 813MB \| 766MB \| 27.9 \| `Executed with 4 threads on a c5.2xlarge Amazon EC2 instance equipped with an Intel(R) Xeon(R) Platinum 8275CL CPU.` Source to benchmark information can be found [here](https://github.com/OpenNMT/CTranslate2).<br /> Original model BLEU scores can be found [here](https://huggingface.co/Helsinki-NLP/opus-mt-fi-crs). ## Internal Benchmarks Internal testing on our end showed inference times reduced by 6x-10x on average compared the vanilla checkpoints using the transformers library. A slight reduction on BLEU scores (~5%) was also identified in comparison to the vanilla checkpoints with a few exceptions. This is likely due to several factors, one being the quantization applied. Further testing is needed from our end to better assess the reduction in translation quality. The command used to compile the vanilla checkpoint into a CTranslate2 model can be found below. Modifying this command can yield differing balances between inferencing performance and translation quality. # CTranslate2 Installation ```bash pip install hf-hub-ctranslate2>=1.0.0 ctranslate2>=3.13.0 ``` ### ct2-transformers-converter Command Used: ```bash ct2-transformers-converter --model Helsinki-NLP/opus-mt-fi-crs --output_dir ./ctranslate2/opus-mt-fi-crs-ctranslate2 --force --copy_files README.md generation_config.json tokenizer_config.json vocab.json source.spm .gitattributes target.spm --quantization float16 ``` # CTranslate2 Converted Checkpoint Information: Compatible With: - [ctranslate2](https://github.com/OpenNMT/CTranslate2) - [hf-hub-ctranslate2](https://github.com/michaelfeil/hf-hub-ctranslate2) Compute Type: - `compute_type=int8_float16` for `device="cuda"` - `compute_type=int8` for `device="cpu"` # Sample Code - ctranslate2 #### Clone the repository to the working directory or wherever you wish to store the model artifacts. #### ```bash git clone https://huggingface.co/gaudi/opus-mt-fi-crs-ctranslate2 ``` #### Take the python code below and update the 'model_dir' variable to the location of the cloned repository. #### ```python from ctranslate2 import Translator import transformers model_dir = "./opus-mt-fi-crs-ctranslate2" # Path to model directory. translator = Translator( model_path=model_dir, device="cuda", # cpu, cuda, or auto. inter_threads=1, # Maximum number of parallel translations. intra_threads=4, # Number of OpenMP threads per translator. compute_type="int8_float16", # int8 for cpu or int8_float16 for cuda. ) tokenizer = transformers.AutoTokenizer.from_pretrained(model_dir) source = tokenizer.convert_ids_to_tokens(tokenizer.encode("XXXXXX, XXX XX XXXXXX.")) results = translator.translate_batch([source]) target = results[0].hypotheses[0] print(tokenizer.decode(tokenizer.convert_tokens_to_ids(target))) ``` # Sample Code - hf-hub-ctranslate2 Derived From [michaelfeil](https://huggingface.co/michaelfeil): ```python from hf_hub_ctranslate2 import TranslatorCT2fromHfHub, GeneratorCT2fromHfHub from transformers import AutoTokenizer model_name = "gaudi/opus-mt-fi-crs-ctranslate2" model = TranslatorCT2fromHfHub( model_name_or_path=model_name, device="cuda", compute_type="int8_float16", tokenizer=AutoTokenizer.from_pretrained(model_name) ) outputs = model.generate( text=["XXX XX XXX XXXXXXX XXXX?", "XX XX XXXX XX XXX!"], ) print(outputs) ``` # License and other remarks: License conditions are intended to be idential to [original huggingface repository](https://huggingface.co/Helsinki-NLP/opus-mt-fi-crs) by Helsinki-NLP.	null	Non_BioNLP	# Repository General Information ## Inspired by and derived from the work of [Helsinki-NLP](https://huggingface.co/Helsinki-NLP), [CTranslate2](https://github.com/OpenNMT/CTranslate2), and [michaelfeil](https://huggingface.co/michaelfeil)! - Link to Original Model ([Helsinki-NLP](https://huggingface.co/Helsinki-NLP)): [Model Link](https://huggingface.co/Helsinki-NLP/opus-mt-fi-crs) - This respository was based on the work of [CTranslate2](https://github.com/OpenNMT/CTranslate2). - This repository was based on the work of [michaelfeil](https://huggingface.co/michaelfeil). # What is CTranslate2? [CTranslate2](https://opennmt.net/CTranslate2/) is a C++ and Python library for efficient inference with Transformer models. CTranslate2 implements a custom runtime that applies many performance optimization techniques such as weights quantization, layers fusion, batch reordering, etc., to accelerate and reduce the memory usage of Transformer models on CPU and GPU. CTranslate2 is one of the most performant ways of hosting translation models at scale. Current supported models include: - Encoder-decoder models: Transformer base/big, M2M-100, NLLB, BART, mBART, Pegasus, T5, Whisper - Decoder-only models: GPT-2, GPT-J, GPT-NeoX, OPT, BLOOM, MPT, Llama, Mistral, Gemma, CodeGen, GPTBigCode, Falcon - Encoder-only models: BERT, DistilBERT, XLM-RoBERTa The project is production-oriented and comes with backward compatibility guarantees, but it also includes experimental features related to model compression and inference acceleration. # CTranslate2 Benchmarks Please note that the results presented below are only valid for the configuration used during this benchmark: absolute and relative performance may change with different settings. Tested against `newstest2014` (En -> De) dataset. The benchmark reports the number of target tokens generated per second (higher is better). The results are aggregated over multiple runs. See the benchmark scripts for more details and reproduce these numbers. Please note that the results presented below are only valid for the configuration used during this benchmark: absolute and relative performance may change with different settings. ## CPU Benchmarks for Generic Opus-MT Models \| Library \| Tokens per Second \| Max Memory Usage \| BLEU \| \| :----: \| :----: \| :----: \| :----: \| \| Transformers 4.26.1 (with PyTorch 1.13.1) \| 147.3 \| 2332MB \| 27.90 \| \| Marian 1.11.0 (int16) \| 330.2 \| 5901MB \| 27.65 \| \| Marian 1.11.0 (int8) \| 355.8 \| 4763MB \| 27.27 \| \| CTranslate2 3.6.0 (int16) \| 596.1 \| 660MB \| 27.53 \| \| CTranslate2 3.6.0 (int8) \| 696.1 \| 516MB \| 27.65 \| ## GPU Benchmarks for Generic Opus-MT Models \| Library \| Tokens per Second \| Max GPU Memory Usage \| Max Memory Usage \| BLEU \| \| :----: \| :----: \| :----: \| :----: \| :----: \| \| Transformers 4.26.1 (with PyTorch 1.13.1) \| 1022.9 \| 4097MB \| 2109MB \| 27.90 \| \| Marian 1.11.0 (float16) \| 3962.4 \| 3239MB \| 1976MB \| 27.94 \| \| CTranslate2 3.6.0 (float16) \| 9296.7 \| 909MB \| 814MB \| 27.9 \| \| CTranslate2 3.6.0 (int8 + float16) \| 8362.7 \| 813MB \| 766MB \| 27.9 \| `Executed with 4 threads on a c5.2xlarge Amazon EC2 instance equipped with an Intel(R) Xeon(R) Platinum 8275CL CPU.` Source to benchmark information can be found [here](https://github.com/OpenNMT/CTranslate2).<br /> Original model BLEU scores can be found [here](https://huggingface.co/Helsinki-NLP/opus-mt-fi-crs). ## Internal Benchmarks Internal testing on our end showed inference times reduced by 6x-10x on average compared the vanilla checkpoints using the transformers library. A slight reduction on BLEU scores (~5%) was also identified in comparison to the vanilla checkpoints with a few exceptions. This is likely due to several factors, one being the quantization applied. Further testing is needed from our end to better assess the reduction in translation quality. The command used to compile the vanilla checkpoint into a CTranslate2 model can be found below. Modifying this command can yield differing balances between inferencing performance and translation quality. # CTranslate2 Installation ```bash pip install hf-hub-ctranslate2>=1.0.0 ctranslate2>=3.13.0 ``` ### ct2-transformers-converter Command Used: ```bash ct2-transformers-converter --model Helsinki-NLP/opus-mt-fi-crs --output_dir ./ctranslate2/opus-mt-fi-crs-ctranslate2 --force --copy_files README.md generation_config.json tokenizer_config.json vocab.json source.spm .gitattributes target.spm --quantization float16 ``` # CTranslate2 Converted Checkpoint Information: Compatible With: - [ctranslate2](https://github.com/OpenNMT/CTranslate2) - [hf-hub-ctranslate2](https://github.com/michaelfeil/hf-hub-ctranslate2) Compute Type: - `compute_type=int8_float16` for `device="cuda"` - `compute_type=int8` for `device="cpu"` # Sample Code - ctranslate2 #### Clone the repository to the working directory or wherever you wish to store the model artifacts. #### ```bash git clone https://huggingface.co/gaudi/opus-mt-fi-crs-ctranslate2 ``` #### Take the python code below and update the 'model_dir' variable to the location of the cloned repository. #### ```python from ctranslate2 import Translator import transformers model_dir = "./opus-mt-fi-crs-ctranslate2" # Path to model directory. translator = Translator( model_path=model_dir, device="cuda", # cpu, cuda, or auto. inter_threads=1, # Maximum number of parallel translations. intra_threads=4, # Number of OpenMP threads per translator. compute_type="int8_float16", # int8 for cpu or int8_float16 for cuda. ) tokenizer = transformers.AutoTokenizer.from_pretrained(model_dir) source = tokenizer.convert_ids_to_tokens(tokenizer.encode("XXXXXX, XXX XX XXXXXX.")) results = translator.translate_batch([source]) target = results[0].hypotheses[0] print(tokenizer.decode(tokenizer.convert_tokens_to_ids(target))) ``` # Sample Code - hf-hub-ctranslate2 Derived From [michaelfeil](https://huggingface.co/michaelfeil): ```python from hf_hub_ctranslate2 import TranslatorCT2fromHfHub, GeneratorCT2fromHfHub from transformers import AutoTokenizer model_name = "gaudi/opus-mt-fi-crs-ctranslate2" model = TranslatorCT2fromHfHub( model_name_or_path=model_name, device="cuda", compute_type="int8_float16", tokenizer=AutoTokenizer.from_pretrained(model_name) ) outputs = model.generate( text=["XXX XX XXX XXXXXXX XXXX?", "XX XX XXXX XX XXX!"], ) print(outputs) ``` # License and other remarks: License conditions are intended to be idential to [original huggingface repository](https://huggingface.co/Helsinki-NLP/opus-mt-fi-crs) by Helsinki-NLP.	{"license": "apache-2.0", "tags": ["ctranslate2", "translation"]}	task	[ "TRANSLATION" ]	43,111
hperkins/Qwen2-VL-7B-Instruct	hperkins	image-text-to-text	[ "transformers", "safetensors", "qwen2_vl", "image-text-to-text", "multimodal", "conversational", "en", "arxiv:2308.12966", "license:apache-2.0", "text-generation-inference", "endpoints_compatible", "region:us" ]	2024-09-03T08:03:09Z	2024-09-12T09:29:30+00:00	20	0	--- language: - en library_name: transformers license: apache-2.0 pipeline_tag: image-text-to-text tags: - multimodal --- # Qwen2-VL-7B-Instruct ## Introduction We're excited to unveil Qwen2-VL, the latest iteration of our Qwen-VL model, representing nearly a year of innovation. ### What’s New in Qwen2-VL? #### Key Enhancements: * SoTA understanding of images of various resolution & ratio: Qwen2-VL achieves state-of-the-art performance on visual understanding benchmarks, including MathVista, DocVQA, RealWorldQA, MTVQA, etc. * Understanding videos of 20min+: Qwen2-VL can understand videos over 20 minutes for high-quality video-based question answering, dialog, content creation, etc. * Agent that can operate your mobiles, robots, etc.: with the abilities of complex reasoning and decision making, Qwen2-VL can be integrated with devices like mobile phones, robots, etc., for automatic operation based on visual environment and text instructions. * Multilingual Support: to serve global users, besides English and Chinese, Qwen2-VL now supports the understanding of texts in different languages inside images, including most European languages, Japanese, Korean, Arabic, Vietnamese, etc. #### Model Architecture Updates: * Naive Dynamic Resolution: Unlike before, Qwen2-VL can handle arbitrary image resolutions, mapping them into a dynamic number of visual tokens, offering a more human-like visual processing experience. <p align="center"> <img src="https://qianwen-res.oss-accelerate-overseas.aliyuncs.com/Qwen2-VL/qwen2_vl.jpg" width="80%"/> <p> * Multimodal Rotary Position Embedding (M-ROPE): Decomposes positional embedding into parts to capture 1D textual, 2D visual, and 3D video positional information, enhancing its multimodal processing capabilities. <p align="center"> <img src="http://qianwen-res.oss-accelerate-overseas.aliyuncs.com/Qwen2-VL/mrope.png" width="80%"/> <p> We have three models with 2, 7 and 72 billion parameters. This repo contains the instruction-tuned 7B Qwen2-VL model. For more information, visit our [Blog](https://qwenlm.github.io/blog/qwen2-vl/) and [GitHub](https://github.com/QwenLM/Qwen2-VL). ## Evaluation ### Image Benchmarks \| Benchmark \| InternVL2-8B \| MiniCPM-V 2.6 \| GPT-4o-mini \| Qwen2-VL-7B \| \| :--- \| :---: \| :---: \| :---: \| :---: \| \| MMMU<sub>val</sub> \| 51.8 \| 49.8 \| 60\| 54.1 \| \| DocVQA<sub>test</sub> \| 91.6 \| 90.8 \| - \| 94.5 \| \| InfoVQA<sub>test</sub> \| 74.8 \| - \| - \|76.5 \| \| ChartQA<sub>test</sub> \| 83.3 \| - \|- \| 83.0 \| \| TextVQA<sub>val</sub> \| 77.4 \| 80.1 \| -\| 84.3 \| \| OCRBench \| 794 \| 852 \| 785 \| 845 \| \| MTVQA \| - \| - \| -\| 26.3 \| \| VCR<sub>en easy</sub> \| - \| 73.88 \| 83.60 \| 89.70 \| \| VCR<sub>zh easy</sub> \| - \| 10.18\| 1.10 \| 59.94 \| \| RealWorldQA \| 64.4 \| - \| - \| 70.1 \| \| MME<sub>sum</sub> \| 2210.3 \| 2348.4 \| 2003.4\| 2326.8 \| \| MMBench-EN<sub>test</sub> \| 81.7 \| - \| - \| 83.0 \| \| MMBench-CN<sub>test</sub> \| 81.2 \| - \| - \| 80.5 \| \| MMBench-V1.1<sub>test</sub> \| 79.4 \| 78.0 \| 76.0\| 80.7 \| \| MMT-Bench<sub>test</sub> \| - \| - \| - \|63.7 \| \| MMStar \| 61.5 \| 57.5 \| 54.8 \| 60.7 \| \| MMVet<sub>GPT-4-Turbo</sub> \| 54.2 \| 60.0 \| 66.9 \| 62.0 \| \| HallBench<sub>avg</sub> \| 45.2 \| 48.1 \| 46.1\| 50.6 \| \| MathVista<sub>testmini</sub> \| 58.3 \| 60.6 \| 52.4 \| 58.2 \| \| MathVision \| - \| - \| - \| 16.3 \| ### Video Benchmarks \| Benchmark \| Internvl2-8B \| LLaVA-OneVision-7B \| MiniCPM-V 2.6 \| Qwen2-VL-7B \| \| :--- \| :---: \| :---: \| :---: \| :---: \| \| MVBench \| 66.4 \| 56.7 \| - \| 67.0 \| \| PerceptionTest<sub>test</sub> \| - \| 57.1 \| - \| 62.3 \| \| EgoSchema<sub>test</sub> \| - \| 60.1 \| - \| 66.7 \| \| Video-MME<sub>wo/w subs</sub> \| 54.0/56.9 \| 58.2/- \| 60.9/63.6 \| 63.3/69.0 \| ## Requirements The code of Qwen2-VL has been in the latest Hugging face transformers and we advise you to build from source with command `pip install git+https://github.com/huggingface/transformers`, or you might encounter the following error: ``` KeyError: 'qwen2_vl' ``` ## Quickstart We offer a toolkit to help you handle various types of visual input more conveniently. This includes base64, URLs, and interleaved images and videos. You can install it using the following command: ```bash pip install qwen-vl-utils ``` Here we show a code snippet to show you how to use the chat model with `transformers` and `qwen_vl_utils`: ```python from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor from qwen_vl_utils import process_vision_info # default: Load the model on the available device(s) model = Qwen2VLForConditionalGeneration.from_pretrained( "Qwen/Qwen2-VL-7B-Instruct", torch_dtype="auto", device_map="auto" ) # We recommend enabling flash_attention_2 for better acceleration and memory saving, especially in multi-image and video scenarios. # model = Qwen2VLForConditionalGeneration.from_pretrained( # "Qwen/Qwen2-VL-7B-Instruct", # torch_dtype=torch.bfloat16, # attn_implementation="flash_attention_2", # device_map="auto", # ) # default processer processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct") # The default range for the number of visual tokens per image in the model is 4-16384. You can set min_pixels and max_pixels according to your needs, such as a token count range of 256-1280, to balance speed and memory usage. # min_pixels = 2562828 # max_pixels = 12802828 # processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels) messages = [ { "role": "user", "content": [ { "type": "image", "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg", }, {"type": "text", "text": "Describe this image."}, ], } ] # Preparation for inference text = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Inference: Generation of the output generated_ids = model.generate(inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_text) ``` <details> <summary>Without qwen_vl_utils</summary> ```python from PIL import Image import requests import torch from torchvision import io from typing import Dict from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor # Load the model in half-precision on the available device(s) model = Qwen2VLForConditionalGeneration.from_pretrained( "Qwen/Qwen2-VL-7B-Instruct", torch_dtype="auto", device_map="auto" ) processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct") # Image url = "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg" image = Image.open(requests.get(url, stream=True).raw) conversation = [ { "role": "user", "content": [ { "type": "image", }, {"type": "text", "text": "Describe this image."}, ], } ] # Preprocess the inputs text_prompt = processor.apply_chat_template(conversation, add_generation_prompt=True) # Excepted output: '<\|im_start\|>system\nYou are a helpful assistant.<\|im_end\|>\n<\|im_start\|>user\n<\|vision_start\|><\|image_pad\|><\|vision_end\|>Describe this image.<\|im_end\|>\n<\|im_start\|>assistant\n' inputs = processor( text=[text_prompt], images=[image], padding=True, return_tensors="pt" ) inputs = inputs.to("cuda") # Inference: Generation of the output output_ids = model.generate(inputs, max_new_tokens=128) generated_ids = [ output_ids[len(input_ids) :] for input_ids, output_ids in zip(inputs.input_ids, output_ids) ] output_text = processor.batch_decode( generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True ) print(output_text) ``` </details> <details> <summary>Multi image inference</summary> ```python # Messages containing multiple images and a text query messages = [ { "role": "user", "content": [ {"type": "image", "image": "file:///path/to/image1.jpg"}, {"type": "image", "image": "file:///path/to/image2.jpg"}, {"type": "text", "text": "Identify the similarities between these images."}, ], } ] # Preparation for inference text = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Inference generated_ids = model.generate(*inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_text) ``` </details> <details> <summary>Video inference</summary> ```python # Messages containing a images list as a video and a text query messages = [ { "role": "user", "content": [ { "type": "video", "video": [ "file:///path/to/frame1.jpg", "file:///path/to/frame2.jpg", "file:///path/to/frame3.jpg", "file:///path/to/frame4.jpg", ], "fps": 1.0, }, {"type": "text", "text": "Describe this video."}, ], } ] # Messages containing a video and a text query messages = [ { "role": "user", "content": [ { "type": "video", "video": "file:///path/to/video1.mp4", "max_pixels": 360 420, "fps": 1.0, }, {"type": "text", "text": "Describe this video."}, ], } ] # Preparation for inference text = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Inference generated_ids = model.generate(inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_text) ``` </details> <details> <summary>Batch inference</summary> ```python # Sample messages for batch inference messages1 = [ { "role": "user", "content": [ {"type": "image", "image": "file:///path/to/image1.jpg"}, {"type": "image", "image": "file:///path/to/image2.jpg"}, {"type": "text", "text": "What are the common elements in these pictures?"}, ], } ] messages2 = [ {"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": "Who are you?"}, ] # Combine messages for batch processing messages = [messages1, messages1] # Preparation for batch inference texts = [ processor.apply_chat_template(msg, tokenize=False, add_generation_prompt=True) for msg in messages ] image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=texts, images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Batch Inference generated_ids = model.generate(inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_texts = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_texts) ``` </details> ### More Usage Tips For input images, we support local files, base64, and URLs. For videos, we currently only support local files. ```python # You can directly insert a local file path, a URL, or a base64-encoded image into the position where you want in the text. ## Local file path messages = [ { "role": "user", "content": [ {"type": "image", "image": "file:///path/to/your/image.jpg"}, {"type": "text", "text": "Describe this image."}, ], } ] ## Image URL messages = [ { "role": "user", "content": [ {"type": "image", "image": "http://path/to/your/image.jpg"}, {"type": "text", "text": "Describe this image."}, ], } ] ## Base64 encoded image messages = [ { "role": "user", "content": [ {"type": "image", "image": "data:image;base64,/9j/..."}, {"type": "text", "text": "Describe this image."}, ], } ] ``` #### Image Resolution for performance boost The model supports a wide range of resolution inputs. By default, it uses the native resolution for input, but higher resolutions can enhance performance at the cost of more computation. Users can set the minimum and maximum number of pixels to achieve an optimal configuration for their needs, such as a token count range of 256-1280, to balance speed and memory usage. ```python min_pixels = 256 * 28 * 28 max_pixels = 1280 * 28 * 28 processor = AutoProcessor.from_pretrained( "Qwen/Qwen2-VL-7B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels ) ``` Besides, We provide two methods for fine-grained control over the image size input to the model: 1. Define min_pixels and max_pixels: Images will be resized to maintain their aspect ratio within the range of min_pixels and max_pixels. 2. Specify exact dimensions: Directly set `resized_height` and `resized_width`. These values will be rounded to the nearest multiple of 28. ```python # min_pixels and max_pixels messages = [ { "role": "user", "content": [ { "type": "image", "image": "file:///path/to/your/image.jpg", "resized_height": 280, "resized_width": 420, }, {"type": "text", "text": "Describe this image."}, ], } ] # resized_height and resized_width messages = [ { "role": "user", "content": [ { "type": "image", "image": "file:///path/to/your/image.jpg", "min_pixels": 50176, "max_pixels": 50176, }, {"type": "text", "text": "Describe this image."}, ], } ] ``` ## Limitations While Qwen2-VL are applicable to a wide range of visual tasks, it is equally important to understand its limitations. Here are some known restrictions: 1. Lack of Audio Support: The current model does not comprehend audio information within videos. 2. Data timeliness: Our image dataset is updated until June 2023, and information subsequent to this date may not be covered. 3. Constraints in Individuals and Intellectual Property (IP): The model's capacity to recognize specific individuals or IPs is limited, potentially failing to comprehensively cover all well-known personalities or brands. 4. Limited Capacity for Complex Instruction: When faced with intricate multi-step instructions, the model's understanding and execution capabilities require enhancement. 5. Insufficient Counting Accuracy: Particularly in complex scenes, the accuracy of object counting is not high, necessitating further improvements. 6. Weak Spatial Reasoning Skills: Especially in 3D spaces, the model's inference of object positional relationships is inadequate, making it difficult to precisely judge the relative positions of objects. These limitations serve as ongoing directions for model optimization and improvement, and we are committed to continually enhancing the model's performance and scope of application. ## Citation If you find our work helpful, feel free to give us a cite. ``` @article{Qwen2-VL, title={Qwen2-VL}, author={Qwen team}, year={2024} } @article{Qwen-VL, title={Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond}, author={Bai, Jinze and Bai, Shuai and Yang, Shusheng and Wang, Shijie and Tan, Sinan and Wang, Peng and Lin, Junyang and Zhou, Chang and Zhou, Jingren}, journal={arXiv preprint arXiv:2308.12966}, year={2023} } ```	null	Non_BioNLP	# Qwen2-VL-7B-Instruct ## Introduction We're excited to unveil Qwen2-VL, the latest iteration of our Qwen-VL model, representing nearly a year of innovation. ### What’s New in Qwen2-VL? #### Key Enhancements: * SoTA understanding of images of various resolution & ratio: Qwen2-VL achieves state-of-the-art performance on visual understanding benchmarks, including MathVista, DocVQA, RealWorldQA, MTVQA, etc. * Understanding videos of 20min+: Qwen2-VL can understand videos over 20 minutes for high-quality video-based question answering, dialog, content creation, etc. * Agent that can operate your mobiles, robots, etc.: with the abilities of complex reasoning and decision making, Qwen2-VL can be integrated with devices like mobile phones, robots, etc., for automatic operation based on visual environment and text instructions. * Multilingual Support: to serve global users, besides English and Chinese, Qwen2-VL now supports the understanding of texts in different languages inside images, including most European languages, Japanese, Korean, Arabic, Vietnamese, etc. #### Model Architecture Updates: * Naive Dynamic Resolution: Unlike before, Qwen2-VL can handle arbitrary image resolutions, mapping them into a dynamic number of visual tokens, offering a more human-like visual processing experience. <p align="center"> <img src="https://qianwen-res.oss-accelerate-overseas.aliyuncs.com/Qwen2-VL/qwen2_vl.jpg" width="80%"/> <p> * Multimodal Rotary Position Embedding (M-ROPE): Decomposes positional embedding into parts to capture 1D textual, 2D visual, and 3D video positional information, enhancing its multimodal processing capabilities. <p align="center"> <img src="http://qianwen-res.oss-accelerate-overseas.aliyuncs.com/Qwen2-VL/mrope.png" width="80%"/> <p> We have three models with 2, 7 and 72 billion parameters. This repo contains the instruction-tuned 7B Qwen2-VL model. For more information, visit our [Blog](https://qwenlm.github.io/blog/qwen2-vl/) and [GitHub](https://github.com/QwenLM/Qwen2-VL). ## Evaluation ### Image Benchmarks \| Benchmark \| InternVL2-8B \| MiniCPM-V 2.6 \| GPT-4o-mini \| Qwen2-VL-7B \| \| :--- \| :---: \| :---: \| :---: \| :---: \| \| MMMU<sub>val</sub> \| 51.8 \| 49.8 \| 60\| 54.1 \| \| DocVQA<sub>test</sub> \| 91.6 \| 90.8 \| - \| 94.5 \| \| InfoVQA<sub>test</sub> \| 74.8 \| - \| - \|76.5 \| \| ChartQA<sub>test</sub> \| 83.3 \| - \|- \| 83.0 \| \| TextVQA<sub>val</sub> \| 77.4 \| 80.1 \| -\| 84.3 \| \| OCRBench \| 794 \| 852 \| 785 \| 845 \| \| MTVQA \| - \| - \| -\| 26.3 \| \| VCR<sub>en easy</sub> \| - \| 73.88 \| 83.60 \| 89.70 \| \| VCR<sub>zh easy</sub> \| - \| 10.18\| 1.10 \| 59.94 \| \| RealWorldQA \| 64.4 \| - \| - \| 70.1 \| \| MME<sub>sum</sub> \| 2210.3 \| 2348.4 \| 2003.4\| 2326.8 \| \| MMBench-EN<sub>test</sub> \| 81.7 \| - \| - \| 83.0 \| \| MMBench-CN<sub>test</sub> \| 81.2 \| - \| - \| 80.5 \| \| MMBench-V1.1<sub>test</sub> \| 79.4 \| 78.0 \| 76.0\| 80.7 \| \| MMT-Bench<sub>test</sub> \| - \| - \| - \|63.7 \| \| MMStar \| 61.5 \| 57.5 \| 54.8 \| 60.7 \| \| MMVet<sub>GPT-4-Turbo</sub> \| 54.2 \| 60.0 \| 66.9 \| 62.0 \| \| HallBench<sub>avg</sub> \| 45.2 \| 48.1 \| 46.1\| 50.6 \| \| MathVista<sub>testmini</sub> \| 58.3 \| 60.6 \| 52.4 \| 58.2 \| \| MathVision \| - \| - \| - \| 16.3 \| ### Video Benchmarks \| Benchmark \| Internvl2-8B \| LLaVA-OneVision-7B \| MiniCPM-V 2.6 \| Qwen2-VL-7B \| \| :--- \| :---: \| :---: \| :---: \| :---: \| \| MVBench \| 66.4 \| 56.7 \| - \| 67.0 \| \| PerceptionTest<sub>test</sub> \| - \| 57.1 \| - \| 62.3 \| \| EgoSchema<sub>test</sub> \| - \| 60.1 \| - \| 66.7 \| \| Video-MME<sub>wo/w subs</sub> \| 54.0/56.9 \| 58.2/- \| 60.9/63.6 \| 63.3/69.0 \| ## Requirements The code of Qwen2-VL has been in the latest Hugging face transformers and we advise you to build from source with command `pip install git+https://github.com/huggingface/transformers`, or you might encounter the following error: ``` KeyError: 'qwen2_vl' ``` ## Quickstart We offer a toolkit to help you handle various types of visual input more conveniently. This includes base64, URLs, and interleaved images and videos. You can install it using the following command: ```bash pip install qwen-vl-utils ``` Here we show a code snippet to show you how to use the chat model with `transformers` and `qwen_vl_utils`: ```python from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor from qwen_vl_utils import process_vision_info # default: Load the model on the available device(s) model = Qwen2VLForConditionalGeneration.from_pretrained( "Qwen/Qwen2-VL-7B-Instruct", torch_dtype="auto", device_map="auto" ) # We recommend enabling flash_attention_2 for better acceleration and memory saving, especially in multi-image and video scenarios. # model = Qwen2VLForConditionalGeneration.from_pretrained( # "Qwen/Qwen2-VL-7B-Instruct", # torch_dtype=torch.bfloat16, # attn_implementation="flash_attention_2", # device_map="auto", # ) # default processer processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct") # The default range for the number of visual tokens per image in the model is 4-16384. You can set min_pixels and max_pixels according to your needs, such as a token count range of 256-1280, to balance speed and memory usage. # min_pixels = 2562828 # max_pixels = 12802828 # processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels) messages = [ { "role": "user", "content": [ { "type": "image", "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg", }, {"type": "text", "text": "Describe this image."}, ], } ] # Preparation for inference text = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Inference: Generation of the output generated_ids = model.generate(inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_text) ``` <details> <summary>Without qwen_vl_utils</summary> ```python from PIL import Image import requests import torch from torchvision import io from typing import Dict from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor # Load the model in half-precision on the available device(s) model = Qwen2VLForConditionalGeneration.from_pretrained( "Qwen/Qwen2-VL-7B-Instruct", torch_dtype="auto", device_map="auto" ) processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct") # Image url = "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg" image = Image.open(requests.get(url, stream=True).raw) conversation = [ { "role": "user", "content": [ { "type": "image", }, {"type": "text", "text": "Describe this image."}, ], } ] # Preprocess the inputs text_prompt = processor.apply_chat_template(conversation, add_generation_prompt=True) # Excepted output: '<\|im_start\|>system\nYou are a helpful assistant.<\|im_end\|>\n<\|im_start\|>user\n<\|vision_start\|><\|image_pad\|><\|vision_end\|>Describe this image.<\|im_end\|>\n<\|im_start\|>assistant\n' inputs = processor( text=[text_prompt], images=[image], padding=True, return_tensors="pt" ) inputs = inputs.to("cuda") # Inference: Generation of the output output_ids = model.generate(inputs, max_new_tokens=128) generated_ids = [ output_ids[len(input_ids) :] for input_ids, output_ids in zip(inputs.input_ids, output_ids) ] output_text = processor.batch_decode( generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True ) print(output_text) ``` </details> <details> <summary>Multi image inference</summary> ```python # Messages containing multiple images and a text query messages = [ { "role": "user", "content": [ {"type": "image", "image": "file:///path/to/image1.jpg"}, {"type": "image", "image": "file:///path/to/image2.jpg"}, {"type": "text", "text": "Identify the similarities between these images."}, ], } ] # Preparation for inference text = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Inference generated_ids = model.generate(*inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_text) ``` </details> <details> <summary>Video inference</summary> ```python # Messages containing a images list as a video and a text query messages = [ { "role": "user", "content": [ { "type": "video", "video": [ "file:///path/to/frame1.jpg", "file:///path/to/frame2.jpg", "file:///path/to/frame3.jpg", "file:///path/to/frame4.jpg", ], "fps": 1.0, }, {"type": "text", "text": "Describe this video."}, ], } ] # Messages containing a video and a text query messages = [ { "role": "user", "content": [ { "type": "video", "video": "file:///path/to/video1.mp4", "max_pixels": 360 420, "fps": 1.0, }, {"type": "text", "text": "Describe this video."}, ], } ] # Preparation for inference text = processor.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=[text], images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Inference generated_ids = model.generate(inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_text = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_text) ``` </details> <details> <summary>Batch inference</summary> ```python # Sample messages for batch inference messages1 = [ { "role": "user", "content": [ {"type": "image", "image": "file:///path/to/image1.jpg"}, {"type": "image", "image": "file:///path/to/image2.jpg"}, {"type": "text", "text": "What are the common elements in these pictures?"}, ], } ] messages2 = [ {"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": "Who are you?"}, ] # Combine messages for batch processing messages = [messages1, messages1] # Preparation for batch inference texts = [ processor.apply_chat_template(msg, tokenize=False, add_generation_prompt=True) for msg in messages ] image_inputs, video_inputs = process_vision_info(messages) inputs = processor( text=texts, images=image_inputs, videos=video_inputs, padding=True, return_tensors="pt", ) inputs = inputs.to("cuda") # Batch Inference generated_ids = model.generate(inputs, max_new_tokens=128) generated_ids_trimmed = [ out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids) ] output_texts = processor.batch_decode( generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False ) print(output_texts) ``` </details> ### More Usage Tips For input images, we support local files, base64, and URLs. For videos, we currently only support local files. ```python # You can directly insert a local file path, a URL, or a base64-encoded image into the position where you want in the text. ## Local file path messages = [ { "role": "user", "content": [ {"type": "image", "image": "file:///path/to/your/image.jpg"}, {"type": "text", "text": "Describe this image."}, ], } ] ## Image URL messages = [ { "role": "user", "content": [ {"type": "image", "image": "http://path/to/your/image.jpg"}, {"type": "text", "text": "Describe this image."}, ], } ] ## Base64 encoded image messages = [ { "role": "user", "content": [ {"type": "image", "image": "data:image;base64,/9j/..."}, {"type": "text", "text": "Describe this image."}, ], } ] ``` #### Image Resolution for performance boost The model supports a wide range of resolution inputs. By default, it uses the native resolution for input, but higher resolutions can enhance performance at the cost of more computation. Users can set the minimum and maximum number of pixels to achieve an optimal configuration for their needs, such as a token count range of 256-1280, to balance speed and memory usage. ```python min_pixels = 256 * 28 * 28 max_pixels = 1280 * 28 * 28 processor = AutoProcessor.from_pretrained( "Qwen/Qwen2-VL-7B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels ) ``` Besides, We provide two methods for fine-grained control over the image size input to the model: 1. Define min_pixels and max_pixels: Images will be resized to maintain their aspect ratio within the range of min_pixels and max_pixels. 2. Specify exact dimensions: Directly set `resized_height` and `resized_width`. These values will be rounded to the nearest multiple of 28. ```python # min_pixels and max_pixels messages = [ { "role": "user", "content": [ { "type": "image", "image": "file:///path/to/your/image.jpg", "resized_height": 280, "resized_width": 420, }, {"type": "text", "text": "Describe this image."}, ], } ] # resized_height and resized_width messages = [ { "role": "user", "content": [ { "type": "image", "image": "file:///path/to/your/image.jpg", "min_pixels": 50176, "max_pixels": 50176, }, {"type": "text", "text": "Describe this image."}, ], } ] ``` ## Limitations While Qwen2-VL are applicable to a wide range of visual tasks, it is equally important to understand its limitations. Here are some known restrictions: 1. Lack of Audio Support: The current model does not comprehend audio information within videos. 2. Data timeliness: Our image dataset is updated until June 2023, and information subsequent to this date may not be covered. 3. Constraints in Individuals and Intellectual Property (IP): The model's capacity to recognize specific individuals or IPs is limited, potentially failing to comprehensively cover all well-known personalities or brands. 4. Limited Capacity for Complex Instruction: When faced with intricate multi-step instructions, the model's understanding and execution capabilities require enhancement. 5. Insufficient Counting Accuracy: Particularly in complex scenes, the accuracy of object counting is not high, necessitating further improvements. 6. Weak Spatial Reasoning Skills: Especially in 3D spaces, the model's inference of object positional relationships is inadequate, making it difficult to precisely judge the relative positions of objects. These limitations serve as ongoing directions for model optimization and improvement, and we are committed to continually enhancing the model's performance and scope of application. ## Citation If you find our work helpful, feel free to give us a cite. ``` @article{Qwen2-VL, title={Qwen2-VL}, author={Qwen team}, year={2024} } @article{Qwen-VL, title={Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond}, author={Bai, Jinze and Bai, Shuai and Yang, Shusheng and Wang, Shijie and Tan, Sinan and Wang, Peng and Lin, Junyang and Zhou, Chang and Zhou, Jingren}, journal={arXiv preprint arXiv:2308.12966}, year={2023} } ```	{"language": ["en"], "library_name": "transformers", "license": "apache-2.0", "pipeline_tag": "image-text-to-text", "tags": ["multimodal"]}	task	[ "QUESTION_ANSWERING" ]	43,112
triplee/torchtune_1B_lr1.5e-5_12epoch_full_finetuned_llama3.2_millfield_241227_meta_before_user_15epoch	triplee	text-generation	[ "transformers", "safetensors", "llama", "text-generation", "facebook", "meta", "pytorch", "llama-3", "conversational", "en", "de", "fr", "it", "pt", "hi", "es", "th", "arxiv:2204.05149", "arxiv:2405.16406", "license:llama3.2", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "region:us" ]	2024-12-26T17:30:13Z	2024-12-26T17:31:00+00:00	9	0	--- language: - en - de - fr - it - pt - hi - es - th library_name: transformers license: llama3.2 pipeline_tag: text-generation tags: - facebook - meta - pytorch - llama - llama-3 extra_gated_prompt: "### LLAMA 3.2 COMMUNITY LICENSE AGREEMENT\n\nLlama 3.2 Version\ \ Release Date: September 25, 2024\n\n“Agreement” means the terms and conditions\ \ for use, reproduction, distribution and modification of the Llama Materials set\ \ forth herein.\n\n“Documentation” means the specifications, manuals and documentation\ \ accompanying Llama 3.2 distributed by Meta at https://llama.meta.com/doc/overview.\n\ \n“Licensee” or “you” means you, or your employer or any other person or entity\ \ (if you are entering into this Agreement on such person or entity’s behalf),\ \ of the age required under applicable laws, rules or regulations to provide legal\ \ consent and that has legal authority to bind your employer or such other person\ \ or entity if you are entering in this Agreement on their behalf.\n\n“Llama 3.2”\ \ means the foundational large language models and software and algorithms, including\ \ machine-learning model code, trained model weights, inference-enabling code, training-enabling\ \ code, fine-tuning enabling code and other elements of the foregoing distributed\ \ by Meta at https://www.llama.com/llama-downloads.\n\n“Llama Materials” means,\ \ collectively, Meta’s proprietary Llama 3.2 and Documentation (and any portion\ \ thereof) made available under this Agreement.\n\n“Meta” or “we” means Meta Platforms\ \ Ireland Limited (if you are located in or, if you are an entity, your principal\ \ place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if\ \ you are located outside of the EEA or Switzerland). \n\nBy clicking “I Accept”\ \ below or by using or distributing any portion or element of the Llama Materials,\ \ you agree to be bound by this Agreement.\n\n1. License Rights and Redistribution.\n\ a. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable\ \ and royalty-free limited license under Meta’s intellectual property or other rights\ \ owned by Meta embodied in the Llama Materials to use, reproduce, distribute,\ \ copy, create derivative works of, and make modifications to the Llama Materials.\ \ \nb. Redistribution and Use. \ni. If you distribute or make available the Llama\ \ Materials (or any derivative works thereof), or a product or service (including\ \ another AI model) that contains any of them, you shall (A) provide a copy of this\ \ Agreement with any such Llama Materials; and (B) prominently display “Built with\ \ Llama” on a related website, user interface, blogpost, about page, or product\ \ documentation. If you use the Llama Materials or any outputs or results of the\ \ Llama Materials to create, train, fine tune, or otherwise improve an AI model,\ \ which is distributed or made available, you shall also include “Llama” at the\ \ beginning of any such AI model name.\nii. If you receive Llama Materials, or any\ \ derivative works thereof, from a Licensee as part of an integrated end user product,\ \ then Section 2 of this Agreement will not apply to you. \niii. You must retain\ \ in all copies of the Llama Materials that you distribute the following attribution\ \ notice within a “Notice” text file distributed as a part of such copies: “Llama\ \ 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms,\ \ Inc. All Rights Reserved.”\niv. Your use of the Llama Materials must comply with\ \ applicable laws and regulations (including trade compliance laws and regulations)\ \ and adhere to the Acceptable Use Policy for the Llama Materials (available at\ \ https://www.llama.com/llama3_2/use-policy), which is hereby incorporated by reference\ \ into this Agreement.\n \n2. Additional Commercial Terms. If, on the Llama 3.2\ \ version release date, the monthly active users of the products or services made\ \ available by or for Licensee, or Licensee’s affiliates, is greater than 700 million\ \ monthly active users in the preceding calendar month, you must request a license\ \ from Meta, which Meta may grant to you in its sole discretion, and you are not\ \ authorized to exercise any of the rights under this Agreement unless or until\ \ Meta otherwise expressly grants you such rights.\n3. Disclaimer of Warranty. UNLESS\ \ REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM\ \ ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, AND META DISCLAIMS\ \ ALL WARRANTIES OF ANY KIND, BOTH EXPRESS AND IMPLIED, INCLUDING, WITHOUT LIMITATION,\ \ ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR\ \ PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING\ \ OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR\ \ USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS.\n4. Limitation of Liability.\ \ IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY,\ \ WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING\ \ OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL,\ \ INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE\ \ BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING.\n5. Intellectual Property.\n\ a. No trademark licenses are granted under this Agreement, and in connection with\ \ the Llama Materials, neither Meta nor Licensee may use any name or mark owned\ \ by or associated with the other or any of its affiliates, except as required\ \ for reasonable and customary use in describing and redistributing the Llama Materials\ \ or as set forth in this Section 5(a). Meta hereby grants you a license to use\ \ “Llama” (the “Mark”) solely as required to comply with the last sentence of Section\ \ 1.b.i. You will comply with Meta’s brand guidelines (currently accessible at\ \ https://about.meta.com/brand/resources/meta/company-brand/). All goodwill arising\ \ out of your use of the Mark will inure to the benefit of Meta.\nb. Subject to\ \ Meta’s ownership of Llama Materials and derivatives made by or for Meta, with\ \ respect to any derivative works and modifications of the Llama Materials that\ \ are made by you, as between you and Meta, you are and will be the owner of such\ \ derivative works and modifications.\nc. If you institute litigation or other proceedings\ \ against Meta or any entity (including a cross-claim or counterclaim in a lawsuit)\ \ alleging that the Llama Materials or Llama 3.2 outputs or results, or any portion\ \ of any of the foregoing, constitutes infringement of intellectual property or\ \ other rights owned or licensable by you, then any licenses granted to you under\ \ this Agreement shall terminate as of the date such litigation or claim is filed\ \ or instituted. You will indemnify and hold harmless Meta from and against any\ \ claim by any third party arising out of or related to your use or distribution\ \ of the Llama Materials.\n6. Term and Termination. The term of this Agreement will\ \ commence upon your acceptance of this Agreement or access to the Llama Materials\ \ and will continue in full force and effect until terminated in accordance with\ \ the terms and conditions herein. Meta may terminate this Agreement if you are\ \ in breach of any term or condition of this Agreement. Upon termination of this\ \ Agreement, you shall delete and cease use of the Llama Materials. Sections 3,\ \ 4 and 7 shall survive the termination of this Agreement. \n7. Governing Law and\ \ Jurisdiction. This Agreement will be governed and construed under the laws of\ \ the State of California without regard to choice of law principles, and the UN\ \ Convention on Contracts for the International Sale of Goods does not apply to\ \ this Agreement. The courts of California shall have exclusive jurisdiction of\ \ any dispute arising out of this Agreement. \n### Llama 3.2 Acceptable Use Policy\n\ Meta is committed to promoting safe and fair use of its tools and features, including\ \ Llama 3.2. If you access or use Llama 3.2, you agree to this Acceptable Use Policy\ \ (“Policy”). The most recent copy of this policy can be found at [https://www.llama.com/llama3_2/use-policy](https://www.llama.com/llama3_2/use-policy).\n\ #### Prohibited Uses\nWe want everyone to use Llama 3.2 safely and responsibly.\ \ You agree you will not use, or allow others to use, Llama 3.2 to:\n1. Violate\ \ the law or others’ rights, including to:\n 1. Engage in, promote, generate,\ \ contribute to, encourage, plan, incite, or further illegal or unlawful activity\ \ or content, such as:\n 1. Violence or terrorism\n 2. Exploitation\ \ or harm to children, including the solicitation, creation, acquisition, or dissemination\ \ of child exploitative content or failure to report Child Sexual Abuse Material\n\ \ 3. Human trafficking, exploitation, and sexual violence\n 4. The\ \ illegal distribution of information or materials to minors, including obscene\ \ materials, or failure to employ legally required age-gating in connection with\ \ such information or materials.\n 5. Sexual solicitation\n 6. Any\ \ other criminal activity\n 1. Engage in, promote, incite, or facilitate the\ \ harassment, abuse, threatening, or bullying of individuals or groups of individuals\n\ \ 2. Engage in, promote, incite, or facilitate discrimination or other unlawful\ \ or harmful conduct in the provision of employment, employment benefits, credit,\ \ housing, other economic benefits, or other essential goods and services\n 3.\ \ Engage in the unauthorized or unlicensed practice of any profession including,\ \ but not limited to, financial, legal, medical/health, or related professional\ \ practices\n 4. Collect, process, disclose, generate, or infer private or sensitive\ \ information about individuals, including information about individuals’ identity,\ \ health, or demographic information, unless you have obtained the right to do so\ \ in accordance with applicable law\n 5. Engage in or facilitate any action or\ \ generate any content that infringes, misappropriates, or otherwise violates any\ \ third-party rights, including the outputs or results of any products or services\ \ using the Llama Materials\n 6. Create, generate, or facilitate the creation\ \ of malicious code, malware, computer viruses or do anything else that could disable,\ \ overburden, interfere with or impair the proper working, integrity, operation\ \ or appearance of a website or computer system\n 7. Engage in any action, or\ \ facilitate any action, to intentionally circumvent or remove usage restrictions\ \ or other safety measures, or to enable functionality disabled by Meta \n2. Engage\ \ in, promote, incite, facilitate, or assist in the planning or development of activities\ \ that present a risk of death or bodily harm to individuals, including use of Llama\ \ 3.2 related to the following:\n 8. Military, warfare, nuclear industries or\ \ applications, espionage, use for materials or activities that are subject to the\ \ International Traffic Arms Regulations (ITAR) maintained by the United States\ \ Department of State or to the U.S. Biological Weapons Anti-Terrorism Act of 1989\ \ or the Chemical Weapons Convention Implementation Act of 1997\n 9. Guns and\ \ illegal weapons (including weapon development)\n 10. Illegal drugs and regulated/controlled\ \ substances\n 11. Operation of critical infrastructure, transportation technologies,\ \ or heavy machinery\n 12. Self-harm or harm to others, including suicide, cutting,\ \ and eating disorders\n 13. Any content intended to incite or promote violence,\ \ abuse, or any infliction of bodily harm to an individual\n3. Intentionally deceive\ \ or mislead others, including use of Llama 3.2 related to the following:\n 14.\ \ Generating, promoting, or furthering fraud or the creation or promotion of disinformation\n\ \ 15. Generating, promoting, or furthering defamatory content, including the\ \ creation of defamatory statements, images, or other content\n 16. Generating,\ \ promoting, or further distributing spam\n 17. Impersonating another individual\ \ without consent, authorization, or legal right\n 18. Representing that the\ \ use of Llama 3.2 or outputs are human-generated\n 19. Generating or facilitating\ \ false online engagement, including fake reviews and other means of fake online\ \ engagement \n4. Fail to appropriately disclose to end users any known dangers\ \ of your AI system 5. Interact with third party tools, models, or software designed\ \ to generate unlawful content or engage in unlawful or harmful conduct and/or represent\ \ that the outputs of such tools, models, or software are associated with Meta or\ \ Llama 3.2\n\nWith respect to any multimodal models included in Llama 3.2, the\ \ rights granted under Section 1(a) of the Llama 3.2 Community License Agreement\ \ are not being granted to you if you are an individual domiciled in, or a company\ \ with a principal place of business in, the European Union. This restriction does\ \ not apply to end users of a product or service that incorporates any such multimodal\ \ models.\n\nPlease report any violation of this Policy, software “bug,” or other\ \ problems that could lead to a violation of this Policy through one of the following\ \ means:\n\n* Reporting issues with the model: [https://github.com/meta-llama/llama-models/issues](https://l.workplace.com/l.php?u=https%3A%2F%2Fgithub.com%2Fmeta-llama%2Fllama-models%2Fissues&h=AT0qV8W9BFT6NwihiOHRuKYQM_UnkzN_NmHMy91OT55gkLpgi4kQupHUl0ssR4dQsIQ8n3tfd0vtkobvsEvt1l4Ic6GXI2EeuHV8N08OG2WnbAmm0FL4ObkazC6G_256vN0lN9DsykCvCqGZ)\n\ * Reporting risky content generated by the model: [developers.facebook.com/llama_output_feedback](http://developers.facebook.com/llama_output_feedback)\n\ * Reporting bugs and security concerns: [facebook.com/whitehat/info](http://facebook.com/whitehat/info)\n\ * Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama\ \ 3.2: [email protected]" extra_gated_fields: First Name: text Last Name: text Date of birth: date_picker Country: country Affiliation: text Job title: type: select options: - Student - Research Graduate - AI researcher - AI developer/engineer - Reporter - Other geo: ip_location ? By clicking Submit below I accept the terms of the license and acknowledge that the information I provide will be collected stored processed and shared in accordance with the Meta Privacy Policy : checkbox extra_gated_description: The information you provide will be collected, stored, processed and shared in accordance with the [Meta Privacy Policy](https://www.facebook.com/privacy/policy/). extra_gated_button_content: Submit --- ## Model Information The Llama 3.2 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction-tuned generative models in 1B and 3B sizes (text in/text out). The Llama 3.2 instruction-tuned text only models are optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. They outperform many of the available open source and closed chat models on common industry benchmarks. Model Developer: Meta Model Architecture: Llama 3.2 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align with human preferences for helpfulness and safety. \| \| Training Data \| Params \| Input modalities \| Output modalities \| Context Length \| GQA \| Shared Embeddings \| Token count \| Knowledge cutoff \| \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| \| Llama 3.2 (text only) \| A new mix of publicly available online data. \| 1B (1.23B) \| Multilingual Text \| Multilingual Text and code \| 128k \| Yes \| Yes \| Up to 9T tokens \| December 2023 \| \| \| \| 3B (3.21B) \| Multilingual Text \| Multilingual Text and code \| \| \| \| \| \| \| Llama 3.2 Quantized (text only) \| A new mix of publicly available online data. \| 1B (1.23B) \| Multilingual Text \| Multilingual Text and code \| 8k \| Yes \| Yes \| Up to 9T tokens \| December 2023 \| \| \| \| 3B (3.21B) \| Multilingual Text \| Multilingual Text and code \| \| \| \| \| \| Supported Languages: English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai are officially supported. Llama 3.2 has been trained on a broader collection of languages than these 8 supported languages. Developers may fine-tune Llama 3.2 models for languages beyond these supported languages, provided they comply with the Llama 3.2 Community License and the Acceptable Use Policy. Developers are always expected to ensure that their deployments, including those that involve additional languages, are completed safely and responsibly. Llama 3.2 Model Family: Token counts refer to pretraining data only. All model versions use Grouped-Query Attention (GQA) for improved inference scalability. Model Release Date: Sept 25, 2024 Status: This is a static model trained on an offline dataset. Future versions may be released that improve model capabilities and safety. License: Use of Llama 3.2 is governed by the [Llama 3.2 Community License](https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/LICENSE) (a custom, commercial license agreement). Feedback: Instructions on how to provide feedback or comments on the model can be found in the Llama Models [README](https://github.com/meta-llama/llama-models/blob/main/README.md). For more technical information about generation parameters and recipes for how to use Llama 3.2 in applications, please go [here](https://github.com/meta-llama/llama-recipes). ## Intended Use Intended Use Cases: Llama 3.2 is intended for commercial and research use in multiple languages. Instruction tuned text only models are intended for assistant-like chat and agentic applications like knowledge retrieval and summarization, mobile AI powered writing assistants and query and prompt rewriting. Pretrained models can be adapted for a variety of additional natural language generation tasks. Similarly, quantized models can be adapted for a variety of on-device use-cases with limited compute resources. Out of Scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Acceptable Use Policy and Llama 3.2 Community License. Use in languages beyond those explicitly referenced as supported in this model card. ## How to use This repository contains two versions of Llama-3.2-1B-Instruct, for use with transformers and with the original `llama` codebase. ### Use with transformers Starting with `transformers >= 4.43.0` onward, you can run conversational inference using the Transformers `pipeline` abstraction or by leveraging the Auto classes with the `generate()` function. Make sure to update your transformers installation via `pip install --upgrade transformers`. ```python import torch from transformers import pipeline model_id = "meta-llama/Llama-3.2-1B-Instruct" pipe = pipeline( "text-generation", model=model_id, torch_dtype=torch.bfloat16, device_map="auto", ) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Who are you?"}, ] outputs = pipe( messages, max_new_tokens=256, ) print(outputs[0]["generated_text"][-1]) ``` Note: You can also find detailed recipes on how to use the model locally, with `torch.compile()`, assisted generations, quantised and more at [`huggingface-llama-recipes`](https://github.com/huggingface/huggingface-llama-recipes) ### Use with `llama` Please, follow the instructions in the [repository](https://github.com/meta-llama/llama) To download Original checkpoints, see the example command below leveraging `huggingface-cli`: ``` huggingface-cli download meta-llama/Llama-3.2-1B-Instruct --include "original/" --local-dir Llama-3.2-1B-Instruct ``` ## Hardware and Software Training Factors:* We used custom training libraries, Meta's custom built GPU cluster, and production infrastructure for pretraining. Fine-tuning, quantization, annotation, and evaluation were also performed on production infrastructure. Training Energy Use: Training utilized a cumulative of 916k GPU hours of computation on H100-80GB (TDP of 700W) type hardware, per the table below. Training time is the total GPU time required for training each model and power consumption is the peak power capacity per GPU device used, adjusted for power usage efficiency. Training Greenhouse Gas Emissions: Estimated total location-based greenhouse gas emissions were 240 tons CO2eq for training. Since 2020, Meta has maintained net zero greenhouse gas emissions in its global operations and matched 100% of its electricity use with renewable energy; therefore, the total market-based greenhouse gas emissions for training were 0 tons CO2eq. \| \| Training Time (GPU hours) \| Logit Generation Time (GPU Hours) \| Training Power Consumption (W) \| Training Location-Based Greenhouse Gas Emissions (tons CO2eq) \| Training Market-Based Greenhouse Gas Emissions (tons CO2eq) \| \| :---- \| :---: \| ----- \| :---: \| :---: \| :---: \| \| Llama 3.2 1B \| 370k \| \- \| 700 \| 107 \| 0 \| \| Llama 3.2 3B \| 460k \| \- \| 700 \| 133 \| 0 \| \| Llama 3.2 1B SpinQuant \| 1.7 \| 0 \| 700 \| Negligible\\ \| 0 \| \| Llama 3.2 3B SpinQuant \| 2.4 \| 0 \| 700 \| Negligible\\ \| 0 \| \| Llama 3.2 1B QLora \| 1.3k \| 0 \| 700 \| 0.381 \| 0 \| \| Llama 3.2 3B QLora \| 1.6k \| 0 \| 700 \| 0.461 \| 0 \| \| Total \| 833k \| 86k \| \| 240 \| 0 \| \\ The location-based CO2e emissions of Llama 3.2 1B SpinQuant and Llama 3.2 3B SpinQuant are less than 0.001 metric tonnes each. This is due to the minimal training GPU hours that are required. The methodology used to determine training energy use and greenhouse gas emissions can be found [here](https://arxiv.org/pdf/2204.05149). Since Meta is openly releasing these models, the training energy use and greenhouse gas emissions will not be incurred by others. ## Training Data Overview: Llama 3.2 was pretrained on up to 9 trillion tokens of data from publicly available sources. For the 1B and 3B Llama 3.2 models, we incorporated logits from the Llama 3.1 8B and 70B models into the pretraining stage of the model development, where outputs (logits) from these larger models were used as token-level targets. Knowledge distillation was used after pruning to recover performance. In post-training we used a similar recipe as Llama 3.1 and produced final chat models by doing several rounds of alignment on top of the pre-trained model. Each round involved Supervised Fine-Tuning (SFT), Rejection Sampling (RS), and Direct Preference Optimization (DPO). Data Freshness: The pretraining data has a cutoff of December 2023\. ## Quantization ### Quantization Scheme We designed the current quantization scheme with the [PyTorch’s ExecuTorch](https://github.com/pytorch/executorch) inference framework and Arm CPU backend in mind, taking into account metrics including model quality, prefill/decoding speed, and memory footprint. Our quantization scheme involves three parts: - All linear layers in all transformer blocks are quantized to a 4-bit groupwise scheme (with a group size of 32) for weights and 8-bit per-token dynamic quantization for activations. - The classification layer is quantized to 8-bit per-channel for weight and 8-bit per token dynamic quantization for activation. - Similar to classification layer, an 8-bit per channel quantization is used for embedding layer. ### Quantization-Aware Training and LoRA The quantization-aware training (QAT) with low-rank adaptation (LoRA) models went through only post-training stages, using the same data as the full precision models. To initialize QAT, we utilize BF16 Llama 3.2 model checkpoints obtained after supervised fine-tuning (SFT) and perform an additional full round of SFT training with QAT. We then freeze the backbone of the QAT model and perform another round of SFT with LoRA adaptors applied to all layers within the transformer block. Meanwhile, the LoRA adaptors' weights and activations are maintained in BF16. Because our approach is similar to QLoRA of Dettmers et al., (2023) (i.e., quantization followed by LoRA adapters), we refer this method as QLoRA. Finally, we fine-tune the resulting model (both backbone and LoRA adaptors) using direct preference optimization (DPO). ### SpinQuant [SpinQuant](https://arxiv.org/abs/2405.16406) was applied, together with generative post-training quantization (GPTQ). For the SpinQuant rotation matrix fine-tuning, we optimized for 100 iterations, using 800 samples with sequence-length 2048 from the WikiText 2 dataset. For GPTQ, we used 128 samples from the same dataset with the same sequence-length. ## Benchmarks \- English Text In this section, we report the results for Llama 3.2 models on standard automatic benchmarks. For all these evaluations, we used our internal evaluations library. ### Base Pretrained Models \| Category \| Benchmark \| \# Shots \| Metric \| Llama 3.2 1B \| Llama 3.2 3B \| Llama 3.1 8B \| \| ----- \| ----- \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| MMLU \| 5 \| macro\_avg/acc\_char \| 32.2 \| 58 \| 66.7 \| \| \| AGIEval English \| 3-5 \| average/acc\_char \| 23.3 \| 39.2 \| 47.8 \| \| \| ARC-Challenge \| 25 \| acc\_char \| 32.8 \| 69.1 \| 79.7 \| \| Reading comprehension \| SQuAD \| 1 \| em \| 49.2 \| 67.7 \| 77 \| \| \| QuAC (F1) \| 1 \| f1 \| 37.9 \| 42.9 \| 44.9 \| \| \| DROP (F1) \| 3 \| f1 \| 28.0 \| 45.2 \| 59.5 \| \| Long Context \| Needle in Haystack \| 0 \| em \| 96.8 \| 1 \| 1 \| ### Instruction Tuned Models \| Capability \| \| Benchmark \| \# Shots \| Metric \| Llama 3.2 1B bf16 \| Llama 3.2 1B Vanilla PTQ\\ \| Llama 3.2 1B Spin Quant \| Llama 3.2 1B QLoRA \| Llama 3.2 3B bf16 \| Llama 3.2 3B Vanilla PTQ\\ \| Llama 3.2 3B Spin Quant \| Llama 3.2 3B QLoRA \| Llama 3.1 8B \| \| :---: \| ----- \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| \| MMLU \| 5 \| macro\_avg/acc \| 49.3 \| 43.3 \| 47.3 \| 49.0 \| 63.4 \| 60.5 \| 62 \| 62.4 \| 69.4 \| \| Re-writing \| \| Open-rewrite eval \| 0 \| micro\_avg/rougeL \| 41.6 \| 39.2 \| 40.9 \| 41.2 \| 40.1 \| 40.3 \| 40.8 \| 40.7 \| 40.9 \| \| Summarization \| \| TLDR9+ (test) \| 1 \| rougeL \| 16.8 \| 14.9 \| 16.7 \| 16.8 \| 19.0 \| 19.1 \| 19.2 \| 19.1 \| 17.2 \| \| Instruction following \| \| IFEval \| 0 \| Avg(Prompt/Instruction acc Loose/Strict) \| 59.5 \| 51.5 \| 58.4 \| 55.6 \| 77.4 \| 73.9 \| 73.5 \| 75.9 \| 80.4 \| \| Math \| \| GSM8K (CoT) \| 8 \| em\_maj1@1 \| 44.4 \| 33.1 \| 40.6 \| 46.5 \| 77.7 \| 72.9 \| 75.7 \| 77.9 \| 84.5 \| \| \| \| MATH (CoT) \| 0 \| final\_em \| 30.6 \| 20.5 \| 25.3 \| 31.0 \| 48.0 \| 44.2 \| 45.3 \| 49.2 \| 51.9 \| \| Reasoning \| \| ARC-C \| 0 \| acc \| 59.4 \| 54.3 \| 57 \| 60.7 \| 78.6 \| 75.6 \| 77.6 \| 77.6 \| 83.4 \| \| \| \| GPQA \| 0 \| acc \| 27.2 \| 25.9 \| 26.3 \| 25.9 \| 32.8 \| 32.8 \| 31.7 \| 33.9 \| 32.8 \| \| \| \| Hellaswag \| 0 \| acc \| 41.2 \| 38.1 \| 41.3 \| 41.5 \| 69.8 \| 66.3 \| 68 \| 66.3 \| 78.7 \| \| Tool Use \| \| BFCL V2 \| 0 \| acc \| 25.7 \| 14.3 \| 15.9 \| 23.7 \| 67.0 \| 53.4 \| 60.1 \| 63.5 \| 67.1 \| \| \| \| Nexus \| 0 \| macro\_avg/acc \| 13.5 \| 5.2 \| 9.6 \| 12.5 \| 34.3 \| 32.4 \| 31.5 \| 30.1 \| 38.5 \| \| Long Context \| \| InfiniteBench/En.QA \| 0 \| longbook\_qa/f1 \| 20.3 \| N/A \| N/A \| N/A \| 19.8 \| N/A \| N/A \| N/A \| 27.3 \| \| \| \| InfiniteBench/En.MC \| 0 \| longbook\_choice/acc \| 38.0 \| N/A \| N/A \| N/A \| 63.3 \| N/A \| N/A \| N/A \| 72.2 \| \| \| \| NIH/Multi-needle \| 0 \| recall \| 75.0 \| N/A \| N/A \| N/A \| 84.7 \| N/A \| N/A \| N/A \| 98.8 \| \| Multilingual \| \| MGSM (CoT) \| 0 \| em \| 24.5 \| 13.7 \| 18.2 \| 24.4 \| 58.2 \| 48.9 \| 54.3 \| 56.8 \| 68.9 \| \\for comparison purposes only. Model not released. ### Multilingual Benchmarks \| Category \| Benchmark \| Language \| Llama 3.2 1B \| Llama 3.2 1B Vanilla PTQ\\ \| Llama 3.2 1B Spin Quant \| Llama 3.2 1B QLoRA \| Llama 3.2 3B \| Llama 3.2 3B Vanilla PTQ\\ \| Llama 3.2 3B Spin Quant \| Llama 3.2 3B QLoRA \| Llama 3.1 8B \| \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| MMLU (5-shot, macro_avg/acc) \| Portuguese \| 39.8 \| 34.9 \| 38.9 \| 40.2 \| 54.5 \| 50.9 \| 53.3 \| 53.4 \| 62.1 \| \| \| \| Spanish \| 41.5 \| 36.0 \| 39.8 \| 41.8 \| 55.1 \| 51.9 \| 53.6 \| 53.6 \| 62.5 \| \| \| \| Italian \| 39.8 \| 34.9 \| 38.1 \| 40.6 \| 53.8 \| 49.9 \| 52.1 \| 51.7 \| 61.6 \| \| \| \| German \| 39.2 \| 34.9 \| 37.5 \| 39.6 \| 53.3 \| 50.0 \| 52.2 \| 51.3 \| 60.6 \| \| \| \| French \| 40.5 \| 34.8 \| 39.2 \| 40.8 \| 54.6 \| 51.2 \| 53.3 \| 53.3 \| 62.3 \| \| \| \| Hindi \| 33.5 \| 30.0 \| 32.1 \| 34.0 \| 43.3 \| 40.4 \| 42.0 \| 42.1 \| 50.9 \| \| \| \| Thai \| 34.7 \| 31.2 \| 32.4 \| 34.9 \| 44.5 \| 41.3 \| 44.0 \| 42.2 \| 50.3 \| \\for comparison purposes only. Model not released. ## Inference time In the below table, we compare the performance metrics of different quantization methods (SpinQuant and QAT \+ LoRA) with the BF16 baseline. The evaluation was done using the [ExecuTorch](https://github.com/pytorch/executorch) framework as the inference engine, with the ARM CPU as a backend using Android OnePlus 12 device. \| Category \| Decode (tokens/sec) \| Time-to-first-token (sec) \| Prefill (tokens/sec) \| Model size (PTE file size in MB) \| Memory size (RSS in MB) \| \| :---- \| ----- \| ----- \| ----- \| ----- \| ----- \| \| 1B BF16 (baseline) \| 19.2 \| 1.0 \| 60.3 \| 2358 \| 3,185 \| \| 1B SpinQuant \| 50.2 (2.6x) \| 0.3 (-76.9%) \| 260.5 (4.3x) \| 1083 (-54.1%) \| 1,921 (-39.7%) \| \| 1B QLoRA \| 45.8 (2.4x) \| 0.3 (-76.0%) \| 252.0 (4.2x) \| 1127 (-52.2%) \| 2,255 (-29.2%) \| \| 3B BF16 (baseline) \| 7.6 \| 3.0 \| 21.2 \| 6129 \| 7,419 \| \| 3B SpinQuant \| 19.7 (2.6x) \| 0.7 (-76.4%) \| 89.7 (4.2x) \| 2435 (-60.3%) \| 3,726 (-49.8%) \| \| 3B QLoRA \| 18.5 (2.4x) \| 0.7 (-76.1%) \| 88.8 (4.2x) \| 2529 (-58.7%) \| 4,060 (-45.3%) \| (\) The performance measurement is done using an adb binary-based approach. (\\) It is measured on an Android OnePlus 12 device. (\\\) Time-to-first-token (TTFT) is measured with prompt length=64 Footnote: - Decode (tokens/second) is for how quickly it keeps generating. Higher is better. - Time-to-first-token (TTFT for shorthand) is for how fast it generates the first token for a given prompt. Lower is better. - Prefill is the inverse of TTFT (aka 1/TTFT) in tokens/second. Higher is better - Model size \- how big is the model, measured by, PTE file, a binary file format for ExecuTorch - RSS size \- Memory usage in resident set size (RSS) ## Responsibility & Safety As part of our Responsible release approach, we followed a three-pronged strategy to managing trust & safety risks: 1. Enable developers to deploy helpful, safe and flexible experiences for their target audience and for the use cases supported by Llama 2. Protect developers against adversarial users aiming to exploit Llama capabilities to potentially cause harm 3. Provide protections for the community to help prevent the misuse of our models ### Responsible Deployment Approach: Llama is a foundational technology designed to be used in a variety of use cases. Examples on how Meta’s Llama models have been responsibly deployed can be found in our [Community Stories webpage](https://llama.meta.com/community-stories/). Our approach is to build the most helpful models, enabling the world to benefit from the technology power, by aligning our model safety for generic use cases and addressing a standard set of harms. Developers are then in the driver’s seat to tailor safety for their use cases, defining their own policies and deploying the models with the necessary safeguards in their Llama systems. Llama 3.2 was developed following the best practices outlined in our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide/). #### Llama 3.2 Instruct Objective: Our main objectives for conducting safety fine-tuning are to provide the research community with a valuable resource for studying the robustness of safety fine-tuning, as well as to offer developers a readily available, safe, and powerful model for various applications to reduce the developer workload to deploy safe AI systems. We implemented the same set of safety mitigations as in Llama 3, and you can learn more about these in the Llama 3 [paper](https://ai.meta.com/research/publications/the-llama-3-herd-of-models/). Fine-Tuning Data: We employ a multi-faceted approach to data collection, combining human-generated data from our vendors with synthetic data to mitigate potential safety risks. We’ve developed many large language model (LLM)-based classifiers that enable us to thoughtfully select high-quality prompts and responses, enhancing data quality control. Refusals and Tone: Building on the work we started with Llama 3, we put a great emphasis on model refusals to benign prompts as well as refusal tone. We included both borderline and adversarial prompts in our safety data strategy, and modified our safety data responses to follow tone guidelines. #### Llama 3.2 Systems Safety as a System: Large language models, including Llama 3.2, are not designed to be deployed in isolation but instead should be deployed as part of an overall AI system with additional safety guardrails as required. Developers are expected to deploy system safeguards when building agentic systems. Safeguards are key to achieve the right helpfulness-safety alignment as well as mitigating safety and security risks inherent to the system and any integration of the model or system with external tools. As part of our responsible release approach, we provide the community with [safeguards](https://llama.meta.com/trust-and-safety/) that developers should deploy with Llama models or other LLMs, including Llama Guard, Prompt Guard and Code Shield. All our [reference implementations](https://github.com/meta-llama/llama-agentic-system) demos contain these safeguards by default so developers can benefit from system-level safety out-of-the-box. ### New Capabilities and Use Cases Technological Advancement: Llama releases usually introduce new capabilities that require specific considerations in addition to the best practices that generally apply across all Generative AI use cases. For prior release capabilities also supported by Llama 3.2, see [Llama 3.1 Model Card](https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/MODEL_CARD.md), as the same considerations apply here as well. Constrained Environments: Llama 3.2 1B and 3B models are expected to be deployed in highly constrained environments, such as mobile devices. LLM Systems using smaller models will have a different alignment profile and safety/helpfulness tradeoff than more complex, larger systems. Developers should ensure the safety of their system meets the requirements of their use case. We recommend using lighter system safeguards for such use cases, like Llama Guard 3-1B or its mobile-optimized version. ### Evaluations Scaled Evaluations: We built dedicated, adversarial evaluation datasets and evaluated systems composed of Llama models and Purple Llama safeguards to filter input prompt and output response. It is important to evaluate applications in context, and we recommend building dedicated evaluation dataset for your use case. Red Teaming: We conducted recurring red teaming exercises with the goal of discovering risks via adversarial prompting and we used the learnings to improve our benchmarks and safety tuning datasets. We partnered early with subject-matter experts in critical risk areas to understand the nature of these real-world harms and how such models may lead to unintended harm for society. Based on these conversations, we derived a set of adversarial goals for the red team to attempt to achieve, such as extracting harmful information or reprogramming the model to act in a potentially harmful capacity. The red team consisted of experts in cybersecurity, adversarial machine learning, responsible AI, and integrity in addition to multilingual content specialists with background in integrity issues in specific geographic markets. ### Critical Risks In addition to our safety work above, we took extra care on measuring and/or mitigating the following critical risk areas: 1\. CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosive Weapons): Llama 3.2 1B and 3B models are smaller and less capable derivatives of Llama 3.1. For Llama 3.1 70B and 405B, to assess risks related to proliferation of chemical and biological weapons, we performed uplift testing designed to assess whether use of Llama 3.1 models could meaningfully increase the capabilities of malicious actors to plan or carry out attacks using these types of weapons and have determined that such testing also applies to the smaller 1B and 3B models. 2\. Child Safety: Child Safety risk assessments were conducted using a team of experts, to assess the model’s capability to produce outputs that could result in Child Safety risks and inform on any necessary and appropriate risk mitigations via fine tuning. We leveraged those expert red teaming sessions to expand the coverage of our evaluation benchmarks through Llama 3 model development. For Llama 3, we conducted new in-depth sessions using objective based methodologies to assess the model risks along multiple attack vectors including the additional languages Llama 3 is trained on. We also partnered with content specialists to perform red teaming exercises assessing potentially violating content while taking account of market specific nuances or experiences. 3\. Cyber Attacks: For Llama 3.1 405B, our cyber attack uplift study investigated whether LLMs can enhance human capabilities in hacking tasks, both in terms of skill level and speed. Our attack automation study focused on evaluating the capabilities of LLMs when used as autonomous agents in cyber offensive operations, specifically in the context of ransomware attacks. This evaluation was distinct from previous studies that considered LLMs as interactive assistants. The primary objective was to assess whether these models could effectively function as independent agents in executing complex cyber-attacks without human intervention. Because Llama 3.2’s 1B and 3B models are smaller and less capable models than Llama 3.1 405B, we broadly believe that the testing conducted for the 405B model also applies to Llama 3.2 models. ### Community Industry Partnerships: Generative AI safety requires expertise and tooling, and we believe in the strength of the open community to accelerate its progress. We are active members of open consortiums, including the AI Alliance, Partnership on AI and MLCommons, actively contributing to safety standardization and transparency. We encourage the community to adopt taxonomies like the MLCommons Proof of Concept evaluation to facilitate collaboration and transparency on safety and content evaluations. Our Purple Llama tools are open sourced for the community to use and widely distributed across ecosystem partners including cloud service providers. We encourage community contributions to our [Github repository](https://github.com/meta-llama/PurpleLlama). Grants: We also set up the [Llama Impact Grants](https://llama.meta.com/llama-impact-grants/) program to identify and support the most compelling applications of Meta’s Llama model for societal benefit across three categories: education, climate and open innovation. The 20 finalists from the hundreds of applications can be found [here](https://llama.meta.com/llama-impact-grants/#finalists). Reporting: Finally, we put in place a set of resources including an [output reporting mechanism](https://developers.facebook.com/llama_output_feedback) and [bug bounty program](https://www.facebook.com/whitehat) to continuously improve the Llama technology with the help of the community. ## Ethical Considerations and Limitations Values: The core values of Llama 3.2 are openness, inclusivity and helpfulness. It is meant to serve everyone, and to work for a wide range of use cases. It is thus designed to be accessible to people across many different backgrounds, experiences and perspectives. Llama 3.2 addresses users and their needs as they are, without insertion unnecessary judgment or normativity, while reflecting the understanding that even content that may appear problematic in some cases can serve valuable purposes in others. It respects the dignity and autonomy of all users, especially in terms of the values of free thought and expression that power innovation and progress. Testing: Llama 3.2 is a new technology, and like any new technology, there are risks associated with its use. Testing conducted to date has not covered, nor could it cover, all scenarios. For these reasons, as with all LLMs, Llama 3.2’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate, biased or other objectionable responses to user prompts. Therefore, before deploying any applications of Llama 3.2 models, developers should perform safety testing and tuning tailored to their specific applications of the model. Please refer to available resources including our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide), [Trust and Safety](https://llama.meta.com/trust-and-safety/) solutions, and other [resources](https://llama.meta.com/docs/get-started/) to learn more about responsible development.	null	Non_BioNLP	## Model Information The Llama 3.2 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction-tuned generative models in 1B and 3B sizes (text in/text out). The Llama 3.2 instruction-tuned text only models are optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. They outperform many of the available open source and closed chat models on common industry benchmarks. Model Developer: Meta Model Architecture: Llama 3.2 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align with human preferences for helpfulness and safety. \| \| Training Data \| Params \| Input modalities \| Output modalities \| Context Length \| GQA \| Shared Embeddings \| Token count \| Knowledge cutoff \| \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| :---- \| \| Llama 3.2 (text only) \| A new mix of publicly available online data. \| 1B (1.23B) \| Multilingual Text \| Multilingual Text and code \| 128k \| Yes \| Yes \| Up to 9T tokens \| December 2023 \| \| \| \| 3B (3.21B) \| Multilingual Text \| Multilingual Text and code \| \| \| \| \| \| \| Llama 3.2 Quantized (text only) \| A new mix of publicly available online data. \| 1B (1.23B) \| Multilingual Text \| Multilingual Text and code \| 8k \| Yes \| Yes \| Up to 9T tokens \| December 2023 \| \| \| \| 3B (3.21B) \| Multilingual Text \| Multilingual Text and code \| \| \| \| \| \| Supported Languages: English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai are officially supported. Llama 3.2 has been trained on a broader collection of languages than these 8 supported languages. Developers may fine-tune Llama 3.2 models for languages beyond these supported languages, provided they comply with the Llama 3.2 Community License and the Acceptable Use Policy. Developers are always expected to ensure that their deployments, including those that involve additional languages, are completed safely and responsibly. Llama 3.2 Model Family: Token counts refer to pretraining data only. All model versions use Grouped-Query Attention (GQA) for improved inference scalability. Model Release Date: Sept 25, 2024 Status: This is a static model trained on an offline dataset. Future versions may be released that improve model capabilities and safety. License: Use of Llama 3.2 is governed by the [Llama 3.2 Community License](https://github.com/meta-llama/llama-models/blob/main/models/llama3_2/LICENSE) (a custom, commercial license agreement). Feedback: Instructions on how to provide feedback or comments on the model can be found in the Llama Models [README](https://github.com/meta-llama/llama-models/blob/main/README.md). For more technical information about generation parameters and recipes for how to use Llama 3.2 in applications, please go [here](https://github.com/meta-llama/llama-recipes). ## Intended Use Intended Use Cases: Llama 3.2 is intended for commercial and research use in multiple languages. Instruction tuned text only models are intended for assistant-like chat and agentic applications like knowledge retrieval and summarization, mobile AI powered writing assistants and query and prompt rewriting. Pretrained models can be adapted for a variety of additional natural language generation tasks. Similarly, quantized models can be adapted for a variety of on-device use-cases with limited compute resources. Out of Scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Acceptable Use Policy and Llama 3.2 Community License. Use in languages beyond those explicitly referenced as supported in this model card. ## How to use This repository contains two versions of Llama-3.2-1B-Instruct, for use with transformers and with the original `llama` codebase. ### Use with transformers Starting with `transformers >= 4.43.0` onward, you can run conversational inference using the Transformers `pipeline` abstraction or by leveraging the Auto classes with the `generate()` function. Make sure to update your transformers installation via `pip install --upgrade transformers`. ```python import torch from transformers import pipeline model_id = "meta-llama/Llama-3.2-1B-Instruct" pipe = pipeline( "text-generation", model=model_id, torch_dtype=torch.bfloat16, device_map="auto", ) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Who are you?"}, ] outputs = pipe( messages, max_new_tokens=256, ) print(outputs[0]["generated_text"][-1]) ``` Note: You can also find detailed recipes on how to use the model locally, with `torch.compile()`, assisted generations, quantised and more at [`huggingface-llama-recipes`](https://github.com/huggingface/huggingface-llama-recipes) ### Use with `llama` Please, follow the instructions in the [repository](https://github.com/meta-llama/llama) To download Original checkpoints, see the example command below leveraging `huggingface-cli`: ``` huggingface-cli download meta-llama/Llama-3.2-1B-Instruct --include "original/" --local-dir Llama-3.2-1B-Instruct ``` ## Hardware and Software Training Factors:* We used custom training libraries, Meta's custom built GPU cluster, and production infrastructure for pretraining. Fine-tuning, quantization, annotation, and evaluation were also performed on production infrastructure. Training Energy Use: Training utilized a cumulative of 916k GPU hours of computation on H100-80GB (TDP of 700W) type hardware, per the table below. Training time is the total GPU time required for training each model and power consumption is the peak power capacity per GPU device used, adjusted for power usage efficiency. Training Greenhouse Gas Emissions: Estimated total location-based greenhouse gas emissions were 240 tons CO2eq for training. Since 2020, Meta has maintained net zero greenhouse gas emissions in its global operations and matched 100% of its electricity use with renewable energy; therefore, the total market-based greenhouse gas emissions for training were 0 tons CO2eq. \| \| Training Time (GPU hours) \| Logit Generation Time (GPU Hours) \| Training Power Consumption (W) \| Training Location-Based Greenhouse Gas Emissions (tons CO2eq) \| Training Market-Based Greenhouse Gas Emissions (tons CO2eq) \| \| :---- \| :---: \| ----- \| :---: \| :---: \| :---: \| \| Llama 3.2 1B \| 370k \| \- \| 700 \| 107 \| 0 \| \| Llama 3.2 3B \| 460k \| \- \| 700 \| 133 \| 0 \| \| Llama 3.2 1B SpinQuant \| 1.7 \| 0 \| 700 \| Negligible\\ \| 0 \| \| Llama 3.2 3B SpinQuant \| 2.4 \| 0 \| 700 \| Negligible\\ \| 0 \| \| Llama 3.2 1B QLora \| 1.3k \| 0 \| 700 \| 0.381 \| 0 \| \| Llama 3.2 3B QLora \| 1.6k \| 0 \| 700 \| 0.461 \| 0 \| \| Total \| 833k \| 86k \| \| 240 \| 0 \| \\ The location-based CO2e emissions of Llama 3.2 1B SpinQuant and Llama 3.2 3B SpinQuant are less than 0.001 metric tonnes each. This is due to the minimal training GPU hours that are required. The methodology used to determine training energy use and greenhouse gas emissions can be found [here](https://arxiv.org/pdf/2204.05149). Since Meta is openly releasing these models, the training energy use and greenhouse gas emissions will not be incurred by others. ## Training Data Overview: Llama 3.2 was pretrained on up to 9 trillion tokens of data from publicly available sources. For the 1B and 3B Llama 3.2 models, we incorporated logits from the Llama 3.1 8B and 70B models into the pretraining stage of the model development, where outputs (logits) from these larger models were used as token-level targets. Knowledge distillation was used after pruning to recover performance. In post-training we used a similar recipe as Llama 3.1 and produced final chat models by doing several rounds of alignment on top of the pre-trained model. Each round involved Supervised Fine-Tuning (SFT), Rejection Sampling (RS), and Direct Preference Optimization (DPO). Data Freshness: The pretraining data has a cutoff of December 2023\. ## Quantization ### Quantization Scheme We designed the current quantization scheme with the [PyTorch’s ExecuTorch](https://github.com/pytorch/executorch) inference framework and Arm CPU backend in mind, taking into account metrics including model quality, prefill/decoding speed, and memory footprint. Our quantization scheme involves three parts: - All linear layers in all transformer blocks are quantized to a 4-bit groupwise scheme (with a group size of 32) for weights and 8-bit per-token dynamic quantization for activations. - The classification layer is quantized to 8-bit per-channel for weight and 8-bit per token dynamic quantization for activation. - Similar to classification layer, an 8-bit per channel quantization is used for embedding layer. ### Quantization-Aware Training and LoRA The quantization-aware training (QAT) with low-rank adaptation (LoRA) models went through only post-training stages, using the same data as the full precision models. To initialize QAT, we utilize BF16 Llama 3.2 model checkpoints obtained after supervised fine-tuning (SFT) and perform an additional full round of SFT training with QAT. We then freeze the backbone of the QAT model and perform another round of SFT with LoRA adaptors applied to all layers within the transformer block. Meanwhile, the LoRA adaptors' weights and activations are maintained in BF16. Because our approach is similar to QLoRA of Dettmers et al., (2023) (i.e., quantization followed by LoRA adapters), we refer this method as QLoRA. Finally, we fine-tune the resulting model (both backbone and LoRA adaptors) using direct preference optimization (DPO). ### SpinQuant [SpinQuant](https://arxiv.org/abs/2405.16406) was applied, together with generative post-training quantization (GPTQ). For the SpinQuant rotation matrix fine-tuning, we optimized for 100 iterations, using 800 samples with sequence-length 2048 from the WikiText 2 dataset. For GPTQ, we used 128 samples from the same dataset with the same sequence-length. ## Benchmarks \- English Text In this section, we report the results for Llama 3.2 models on standard automatic benchmarks. For all these evaluations, we used our internal evaluations library. ### Base Pretrained Models \| Category \| Benchmark \| \# Shots \| Metric \| Llama 3.2 1B \| Llama 3.2 3B \| Llama 3.1 8B \| \| ----- \| ----- \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| MMLU \| 5 \| macro\_avg/acc\_char \| 32.2 \| 58 \| 66.7 \| \| \| AGIEval English \| 3-5 \| average/acc\_char \| 23.3 \| 39.2 \| 47.8 \| \| \| ARC-Challenge \| 25 \| acc\_char \| 32.8 \| 69.1 \| 79.7 \| \| Reading comprehension \| SQuAD \| 1 \| em \| 49.2 \| 67.7 \| 77 \| \| \| QuAC (F1) \| 1 \| f1 \| 37.9 \| 42.9 \| 44.9 \| \| \| DROP (F1) \| 3 \| f1 \| 28.0 \| 45.2 \| 59.5 \| \| Long Context \| Needle in Haystack \| 0 \| em \| 96.8 \| 1 \| 1 \| ### Instruction Tuned Models \| Capability \| \| Benchmark \| \# Shots \| Metric \| Llama 3.2 1B bf16 \| Llama 3.2 1B Vanilla PTQ\\ \| Llama 3.2 1B Spin Quant \| Llama 3.2 1B QLoRA \| Llama 3.2 3B bf16 \| Llama 3.2 3B Vanilla PTQ\\ \| Llama 3.2 3B Spin Quant \| Llama 3.2 3B QLoRA \| Llama 3.1 8B \| \| :---: \| ----- \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| \| MMLU \| 5 \| macro\_avg/acc \| 49.3 \| 43.3 \| 47.3 \| 49.0 \| 63.4 \| 60.5 \| 62 \| 62.4 \| 69.4 \| \| Re-writing \| \| Open-rewrite eval \| 0 \| micro\_avg/rougeL \| 41.6 \| 39.2 \| 40.9 \| 41.2 \| 40.1 \| 40.3 \| 40.8 \| 40.7 \| 40.9 \| \| Summarization \| \| TLDR9+ (test) \| 1 \| rougeL \| 16.8 \| 14.9 \| 16.7 \| 16.8 \| 19.0 \| 19.1 \| 19.2 \| 19.1 \| 17.2 \| \| Instruction following \| \| IFEval \| 0 \| Avg(Prompt/Instruction acc Loose/Strict) \| 59.5 \| 51.5 \| 58.4 \| 55.6 \| 77.4 \| 73.9 \| 73.5 \| 75.9 \| 80.4 \| \| Math \| \| GSM8K (CoT) \| 8 \| em\_maj1@1 \| 44.4 \| 33.1 \| 40.6 \| 46.5 \| 77.7 \| 72.9 \| 75.7 \| 77.9 \| 84.5 \| \| \| \| MATH (CoT) \| 0 \| final\_em \| 30.6 \| 20.5 \| 25.3 \| 31.0 \| 48.0 \| 44.2 \| 45.3 \| 49.2 \| 51.9 \| \| Reasoning \| \| ARC-C \| 0 \| acc \| 59.4 \| 54.3 \| 57 \| 60.7 \| 78.6 \| 75.6 \| 77.6 \| 77.6 \| 83.4 \| \| \| \| GPQA \| 0 \| acc \| 27.2 \| 25.9 \| 26.3 \| 25.9 \| 32.8 \| 32.8 \| 31.7 \| 33.9 \| 32.8 \| \| \| \| Hellaswag \| 0 \| acc \| 41.2 \| 38.1 \| 41.3 \| 41.5 \| 69.8 \| 66.3 \| 68 \| 66.3 \| 78.7 \| \| Tool Use \| \| BFCL V2 \| 0 \| acc \| 25.7 \| 14.3 \| 15.9 \| 23.7 \| 67.0 \| 53.4 \| 60.1 \| 63.5 \| 67.1 \| \| \| \| Nexus \| 0 \| macro\_avg/acc \| 13.5 \| 5.2 \| 9.6 \| 12.5 \| 34.3 \| 32.4 \| 31.5 \| 30.1 \| 38.5 \| \| Long Context \| \| InfiniteBench/En.QA \| 0 \| longbook\_qa/f1 \| 20.3 \| N/A \| N/A \| N/A \| 19.8 \| N/A \| N/A \| N/A \| 27.3 \| \| \| \| InfiniteBench/En.MC \| 0 \| longbook\_choice/acc \| 38.0 \| N/A \| N/A \| N/A \| 63.3 \| N/A \| N/A \| N/A \| 72.2 \| \| \| \| NIH/Multi-needle \| 0 \| recall \| 75.0 \| N/A \| N/A \| N/A \| 84.7 \| N/A \| N/A \| N/A \| 98.8 \| \| Multilingual \| \| MGSM (CoT) \| 0 \| em \| 24.5 \| 13.7 \| 18.2 \| 24.4 \| 58.2 \| 48.9 \| 54.3 \| 56.8 \| 68.9 \| \\for comparison purposes only. Model not released. ### Multilingual Benchmarks \| Category \| Benchmark \| Language \| Llama 3.2 1B \| Llama 3.2 1B Vanilla PTQ\\ \| Llama 3.2 1B Spin Quant \| Llama 3.2 1B QLoRA \| Llama 3.2 3B \| Llama 3.2 3B Vanilla PTQ\\ \| Llama 3.2 3B Spin Quant \| Llama 3.2 3B QLoRA \| Llama 3.1 8B \| \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| \| General \| MMLU (5-shot, macro_avg/acc) \| Portuguese \| 39.8 \| 34.9 \| 38.9 \| 40.2 \| 54.5 \| 50.9 \| 53.3 \| 53.4 \| 62.1 \| \| \| \| Spanish \| 41.5 \| 36.0 \| 39.8 \| 41.8 \| 55.1 \| 51.9 \| 53.6 \| 53.6 \| 62.5 \| \| \| \| Italian \| 39.8 \| 34.9 \| 38.1 \| 40.6 \| 53.8 \| 49.9 \| 52.1 \| 51.7 \| 61.6 \| \| \| \| German \| 39.2 \| 34.9 \| 37.5 \| 39.6 \| 53.3 \| 50.0 \| 52.2 \| 51.3 \| 60.6 \| \| \| \| French \| 40.5 \| 34.8 \| 39.2 \| 40.8 \| 54.6 \| 51.2 \| 53.3 \| 53.3 \| 62.3 \| \| \| \| Hindi \| 33.5 \| 30.0 \| 32.1 \| 34.0 \| 43.3 \| 40.4 \| 42.0 \| 42.1 \| 50.9 \| \| \| \| Thai \| 34.7 \| 31.2 \| 32.4 \| 34.9 \| 44.5 \| 41.3 \| 44.0 \| 42.2 \| 50.3 \| \\for comparison purposes only. Model not released. ## Inference time In the below table, we compare the performance metrics of different quantization methods (SpinQuant and QAT \+ LoRA) with the BF16 baseline. The evaluation was done using the [ExecuTorch](https://github.com/pytorch/executorch) framework as the inference engine, with the ARM CPU as a backend using Android OnePlus 12 device. \| Category \| Decode (tokens/sec) \| Time-to-first-token (sec) \| Prefill (tokens/sec) \| Model size (PTE file size in MB) \| Memory size (RSS in MB) \| \| :---- \| ----- \| ----- \| ----- \| ----- \| ----- \| \| 1B BF16 (baseline) \| 19.2 \| 1.0 \| 60.3 \| 2358 \| 3,185 \| \| 1B SpinQuant \| 50.2 (2.6x) \| 0.3 (-76.9%) \| 260.5 (4.3x) \| 1083 (-54.1%) \| 1,921 (-39.7%) \| \| 1B QLoRA \| 45.8 (2.4x) \| 0.3 (-76.0%) \| 252.0 (4.2x) \| 1127 (-52.2%) \| 2,255 (-29.2%) \| \| 3B BF16 (baseline) \| 7.6 \| 3.0 \| 21.2 \| 6129 \| 7,419 \| \| 3B SpinQuant \| 19.7 (2.6x) \| 0.7 (-76.4%) \| 89.7 (4.2x) \| 2435 (-60.3%) \| 3,726 (-49.8%) \| \| 3B QLoRA \| 18.5 (2.4x) \| 0.7 (-76.1%) \| 88.8 (4.2x) \| 2529 (-58.7%) \| 4,060 (-45.3%) \| (\) The performance measurement is done using an adb binary-based approach. (\\) It is measured on an Android OnePlus 12 device. (\\\) Time-to-first-token (TTFT) is measured with prompt length=64 Footnote: - Decode (tokens/second) is for how quickly it keeps generating. Higher is better. - Time-to-first-token (TTFT for shorthand) is for how fast it generates the first token for a given prompt. Lower is better. - Prefill is the inverse of TTFT (aka 1/TTFT) in tokens/second. Higher is better - Model size \- how big is the model, measured by, PTE file, a binary file format for ExecuTorch - RSS size \- Memory usage in resident set size (RSS) ## Responsibility & Safety As part of our Responsible release approach, we followed a three-pronged strategy to managing trust & safety risks: 1. Enable developers to deploy helpful, safe and flexible experiences for their target audience and for the use cases supported by Llama 2. Protect developers against adversarial users aiming to exploit Llama capabilities to potentially cause harm 3. Provide protections for the community to help prevent the misuse of our models ### Responsible Deployment Approach: Llama is a foundational technology designed to be used in a variety of use cases. Examples on how Meta’s Llama models have been responsibly deployed can be found in our [Community Stories webpage](https://llama.meta.com/community-stories/). Our approach is to build the most helpful models, enabling the world to benefit from the technology power, by aligning our model safety for generic use cases and addressing a standard set of harms. Developers are then in the driver’s seat to tailor safety for their use cases, defining their own policies and deploying the models with the necessary safeguards in their Llama systems. Llama 3.2 was developed following the best practices outlined in our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide/). #### Llama 3.2 Instruct Objective: Our main objectives for conducting safety fine-tuning are to provide the research community with a valuable resource for studying the robustness of safety fine-tuning, as well as to offer developers a readily available, safe, and powerful model for various applications to reduce the developer workload to deploy safe AI systems. We implemented the same set of safety mitigations as in Llama 3, and you can learn more about these in the Llama 3 [paper](https://ai.meta.com/research/publications/the-llama-3-herd-of-models/). Fine-Tuning Data: We employ a multi-faceted approach to data collection, combining human-generated data from our vendors with synthetic data to mitigate potential safety risks. We’ve developed many large language model (LLM)-based classifiers that enable us to thoughtfully select high-quality prompts and responses, enhancing data quality control. Refusals and Tone: Building on the work we started with Llama 3, we put a great emphasis on model refusals to benign prompts as well as refusal tone. We included both borderline and adversarial prompts in our safety data strategy, and modified our safety data responses to follow tone guidelines. #### Llama 3.2 Systems Safety as a System: Large language models, including Llama 3.2, are not designed to be deployed in isolation but instead should be deployed as part of an overall AI system with additional safety guardrails as required. Developers are expected to deploy system safeguards when building agentic systems. Safeguards are key to achieve the right helpfulness-safety alignment as well as mitigating safety and security risks inherent to the system and any integration of the model or system with external tools. As part of our responsible release approach, we provide the community with [safeguards](https://llama.meta.com/trust-and-safety/) that developers should deploy with Llama models or other LLMs, including Llama Guard, Prompt Guard and Code Shield. All our [reference implementations](https://github.com/meta-llama/llama-agentic-system) demos contain these safeguards by default so developers can benefit from system-level safety out-of-the-box. ### New Capabilities and Use Cases Technological Advancement: Llama releases usually introduce new capabilities that require specific considerations in addition to the best practices that generally apply across all Generative AI use cases. For prior release capabilities also supported by Llama 3.2, see [Llama 3.1 Model Card](https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/MODEL_CARD.md), as the same considerations apply here as well. Constrained Environments: Llama 3.2 1B and 3B models are expected to be deployed in highly constrained environments, such as mobile devices. LLM Systems using smaller models will have a different alignment profile and safety/helpfulness tradeoff than more complex, larger systems. Developers should ensure the safety of their system meets the requirements of their use case. We recommend using lighter system safeguards for such use cases, like Llama Guard 3-1B or its mobile-optimized version. ### Evaluations Scaled Evaluations: We built dedicated, adversarial evaluation datasets and evaluated systems composed of Llama models and Purple Llama safeguards to filter input prompt and output response. It is important to evaluate applications in context, and we recommend building dedicated evaluation dataset for your use case. Red Teaming: We conducted recurring red teaming exercises with the goal of discovering risks via adversarial prompting and we used the learnings to improve our benchmarks and safety tuning datasets. We partnered early with subject-matter experts in critical risk areas to understand the nature of these real-world harms and how such models may lead to unintended harm for society. Based on these conversations, we derived a set of adversarial goals for the red team to attempt to achieve, such as extracting harmful information or reprogramming the model to act in a potentially harmful capacity. The red team consisted of experts in cybersecurity, adversarial machine learning, responsible AI, and integrity in addition to multilingual content specialists with background in integrity issues in specific geographic markets. ### Critical Risks In addition to our safety work above, we took extra care on measuring and/or mitigating the following critical risk areas: 1\. CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosive Weapons): Llama 3.2 1B and 3B models are smaller and less capable derivatives of Llama 3.1. For Llama 3.1 70B and 405B, to assess risks related to proliferation of chemical and biological weapons, we performed uplift testing designed to assess whether use of Llama 3.1 models could meaningfully increase the capabilities of malicious actors to plan or carry out attacks using these types of weapons and have determined that such testing also applies to the smaller 1B and 3B models. 2\. Child Safety: Child Safety risk assessments were conducted using a team of experts, to assess the model’s capability to produce outputs that could result in Child Safety risks and inform on any necessary and appropriate risk mitigations via fine tuning. We leveraged those expert red teaming sessions to expand the coverage of our evaluation benchmarks through Llama 3 model development. For Llama 3, we conducted new in-depth sessions using objective based methodologies to assess the model risks along multiple attack vectors including the additional languages Llama 3 is trained on. We also partnered with content specialists to perform red teaming exercises assessing potentially violating content while taking account of market specific nuances or experiences. 3\. Cyber Attacks: For Llama 3.1 405B, our cyber attack uplift study investigated whether LLMs can enhance human capabilities in hacking tasks, both in terms of skill level and speed. Our attack automation study focused on evaluating the capabilities of LLMs when used as autonomous agents in cyber offensive operations, specifically in the context of ransomware attacks. This evaluation was distinct from previous studies that considered LLMs as interactive assistants. The primary objective was to assess whether these models could effectively function as independent agents in executing complex cyber-attacks without human intervention. Because Llama 3.2’s 1B and 3B models are smaller and less capable models than Llama 3.1 405B, we broadly believe that the testing conducted for the 405B model also applies to Llama 3.2 models. ### Community Industry Partnerships: Generative AI safety requires expertise and tooling, and we believe in the strength of the open community to accelerate its progress. We are active members of open consortiums, including the AI Alliance, Partnership on AI and MLCommons, actively contributing to safety standardization and transparency. We encourage the community to adopt taxonomies like the MLCommons Proof of Concept evaluation to facilitate collaboration and transparency on safety and content evaluations. Our Purple Llama tools are open sourced for the community to use and widely distributed across ecosystem partners including cloud service providers. We encourage community contributions to our [Github repository](https://github.com/meta-llama/PurpleLlama). Grants: We also set up the [Llama Impact Grants](https://llama.meta.com/llama-impact-grants/) program to identify and support the most compelling applications of Meta’s Llama model for societal benefit across three categories: education, climate and open innovation. The 20 finalists from the hundreds of applications can be found [here](https://llama.meta.com/llama-impact-grants/#finalists). Reporting: Finally, we put in place a set of resources including an [output reporting mechanism](https://developers.facebook.com/llama_output_feedback) and [bug bounty program](https://www.facebook.com/whitehat) to continuously improve the Llama technology with the help of the community. ## Ethical Considerations and Limitations Values: The core values of Llama 3.2 are openness, inclusivity and helpfulness. It is meant to serve everyone, and to work for a wide range of use cases. It is thus designed to be accessible to people across many different backgrounds, experiences and perspectives. Llama 3.2 addresses users and their needs as they are, without insertion unnecessary judgment or normativity, while reflecting the understanding that even content that may appear problematic in some cases can serve valuable purposes in others. It respects the dignity and autonomy of all users, especially in terms of the values of free thought and expression that power innovation and progress. Testing: Llama 3.2 is a new technology, and like any new technology, there are risks associated with its use. Testing conducted to date has not covered, nor could it cover, all scenarios. For these reasons, as with all LLMs, Llama 3.2’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate, biased or other objectionable responses to user prompts. Therefore, before deploying any applications of Llama 3.2 models, developers should perform safety testing and tuning tailored to their specific applications of the model. Please refer to available resources including our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide), [Trust and Safety](https://llama.meta.com/trust-and-safety/) solutions, and other [resources](https://llama.meta.com/docs/get-started/) to learn more about responsible development.	{"language": ["en", "de", "fr", "it", "pt", "hi", "es", "th"], "library_name": "transformers", "license": "llama3.2", "pipeline_tag": "text-generation", "tags": ["facebook", "meta", "pytorch", "llama", "llama-3"], "extra_gated_prompt": "### LLAMA 3.2 COMMUNITY LICENSE AGREEMENT\n\nLlama 3.2 Version Release Date: September 25, 2024\n\n“Agreement” means the terms and conditions for use, reproduction, distribution and modification of the Llama Materials set forth herein.\n\n“Documentation” means the specifications, manuals and documentation accompanying Llama 3.2 distributed by Meta at https://llama.meta.com/doc/overview.\n\n“Licensee” or “you” means you, or your employer or any other person or entity (if you are entering into this Agreement on such person or entity’s behalf), of the age required under applicable laws, rules or regulations to provide legal consent and that has legal authority to bind your employer or such other person or entity if you are entering in this Agreement on their behalf.\n\n“Llama 3.2” means the foundational large language models and software and algorithms, including machine-learning model code, trained model weights, inference-enabling code, training-enabling code, fine-tuning enabling code and other elements of the foregoing distributed by Meta at https://www.llama.com/llama-downloads.\n\n“Llama Materials” means, collectively, Meta’s proprietary Llama 3.2 and Documentation (and any portion thereof) made available under this Agreement.\n\n“Meta” or “we” means Meta Platforms Ireland Limited (if you are located in or, if you are an entity, your principal place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if you are located outside of the EEA or Switzerland). \n\nBy clicking “I Accept” below or by using or distributing any portion or element of the Llama Materials, you agree to be bound by this Agreement.\n\n1. License Rights and Redistribution.\na. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable and royalty-free limited license under Meta’s intellectual property or other rights owned by Meta embodied in the Llama Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Llama Materials. \nb. Redistribution and Use. \ni. If you distribute or make available the Llama Materials (or any derivative works thereof), or a product or service (including another AI model) that contains any of them, you shall (A) provide a copy of this Agreement with any such Llama Materials; and (B) prominently display “Built with Llama” on a related website, user interface, blogpost, about page, or product documentation. If you use the Llama Materials or any outputs or results of the Llama Materials to create, train, fine tune, or otherwise improve an AI model, which is distributed or made available, you shall also include “Llama” at the beginning of any such AI model name.\nii. If you receive Llama Materials, or any derivative works thereof, from a Licensee as part of an integrated end user product, then Section 2 of this Agreement will not apply to you. \niii. You must retain in all copies of the Llama Materials that you distribute the following attribution notice within a “Notice” text file distributed as a part of such copies: “Llama 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.”\niv. Your use of the Llama Materials must comply with applicable laws and regulations (including trade compliance laws and regulations) and adhere to the Acceptable Use Policy for the Llama Materials (available at https://www.llama.com/llama3_2/use-policy), which is hereby incorporated by reference into this Agreement.\n \n2. Additional Commercial Terms. If, on the Llama 3.2 version release date, the monthly active users of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 700 million monthly active users in the preceding calendar month, you must request a license from Meta, which Meta may grant to you in its sole discretion, and you are not authorized to exercise any of the rights under this Agreement unless or until Meta otherwise expressly grants you such rights.\n3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF ANY KIND, AND META DISCLAIMS ALL WARRANTIES OF ANY KIND, BOTH EXPRESS AND IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND RESULTS.\n4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED OF THE POSSIBILITY OF ANY OF THE FOREGOING.\n5. Intellectual Property.\na. No trademark licenses are granted under this Agreement, and in connection with the Llama Materials, neither Meta nor Licensee may use any name or mark owned by or associated with the other or any of its affiliates, except as required for reasonable and customary use in describing and redistributing the Llama Materials or as set forth in this Section 5(a). Meta hereby grants you a license to use “Llama” (the “Mark”) solely as required to comply with the last sentence of Section 1.b.i. You will comply with Meta’s brand guidelines (currently accessible at https://about.meta.com/brand/resources/meta/company-brand/). All goodwill arising out of your use of the Mark will inure to the benefit of Meta.\nb. Subject to Meta’s ownership of Llama Materials and derivatives made by or for Meta, with respect to any derivative works and modifications of the Llama Materials that are made by you, as between you and Meta, you are and will be the owner of such derivative works and modifications.\nc. If you institute litigation or other proceedings against Meta or any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Llama Materials or Llama 3.2 outputs or results, or any portion of any of the foregoing, constitutes infringement of intellectual property or other rights owned or licensable by you, then any licenses granted to you under this Agreement shall terminate as of the date such litigation or claim is filed or instituted. You will indemnify and hold harmless Meta from and against any claim by any third party arising out of or related to your use or distribution of the Llama Materials.\n6. Term and Termination. The term of this Agreement will commence upon your acceptance of this Agreement or access to the Llama Materials and will continue in full force and effect until terminated in accordance with the terms and conditions herein. Meta may terminate this Agreement if you are in breach of any term or condition of this Agreement. Upon termination of this Agreement, you shall delete and cease use of the Llama Materials. Sections 3, 4 and 7 shall survive the termination of this Agreement. \n7. Governing Law and Jurisdiction. This Agreement will be governed and construed under the laws of the State of California without regard to choice of law principles, and the UN Convention on Contracts for the International Sale of Goods does not apply to this Agreement. The courts of California shall have exclusive jurisdiction of any dispute arising out of this Agreement. \n### Llama 3.2 Acceptable Use Policy\nMeta is committed to promoting safe and fair use of its tools and features, including Llama 3.2. If you access or use Llama 3.2, you agree to this Acceptable Use Policy (“Policy”). The most recent copy of this policy can be found at [https://www.llama.com/llama3_2/use-policy](https://www.llama.com/llama3_2/use-policy).\n#### Prohibited Uses\nWe want everyone to use Llama 3.2 safely and responsibly. You agree you will not use, or allow others to use, Llama 3.2 to:\n1. Violate the law or others’ rights, including to:\n 1. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as:\n 1. Violence or terrorism\n 2. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material\n 3. Human trafficking, exploitation, and sexual violence\n 4. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials.\n 5. Sexual solicitation\n 6. Any other criminal activity\n 1. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals\n 2. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services\n 3. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices\n 4. Collect, process, disclose, generate, or infer private or sensitive information about individuals, including information about individuals’ identity, health, or demographic information, unless you have obtained the right to do so in accordance with applicable law\n 5. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama Materials\n 6. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system\n 7. Engage in any action, or facilitate any action, to intentionally circumvent or remove usage restrictions or other safety measures, or to enable functionality disabled by Meta \n2. Engage in, promote, incite, facilitate, or assist in the planning or development of activities that present a risk of death or bodily harm to individuals, including use of Llama 3.2 related to the following:\n 8. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State or to the U.S. Biological Weapons Anti-Terrorism Act of 1989 or the Chemical Weapons Convention Implementation Act of 1997\n 9. Guns and illegal weapons (including weapon development)\n 10. Illegal drugs and regulated/controlled substances\n 11. Operation of critical infrastructure, transportation technologies, or heavy machinery\n 12. Self-harm or harm to others, including suicide, cutting, and eating disorders\n 13. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual\n3. Intentionally deceive or mislead others, including use of Llama 3.2 related to the following:\n 14. Generating, promoting, or furthering fraud or the creation or promotion of disinformation\n 15. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content\n 16. Generating, promoting, or further distributing spam\n 17. Impersonating another individual without consent, authorization, or legal right\n 18. Representing that the use of Llama 3.2 or outputs are human-generated\n 19. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement \n4. Fail to appropriately disclose to end users any known dangers of your AI system 5. Interact with third party tools, models, or software designed to generate unlawful content or engage in unlawful or harmful conduct and/or represent that the outputs of such tools, models, or software are associated with Meta or Llama 3.2\n\nWith respect to any multimodal models included in Llama 3.2, the rights granted under Section 1(a) of the Llama 3.2 Community License Agreement are not being granted to you if you are an individual domiciled in, or a company with a principal place of business in, the European Union. This restriction does not apply to end users of a product or service that incorporates any such multimodal models.\n\nPlease report any violation of this Policy, software “bug,” or other problems that could lead to a violation of this Policy through one of the following means:\n\n* Reporting issues with the model: [https://github.com/meta-llama/llama-models/issues](https://l.workplace.com/l.php?u=https%3A%2F%2Fgithub.com%2Fmeta-llama%2Fllama-models%2Fissues&h=AT0qV8W9BFT6NwihiOHRuKYQM_UnkzN_NmHMy91OT55gkLpgi4kQupHUl0ssR4dQsIQ8n3tfd0vtkobvsEvt1l4Ic6GXI2EeuHV8N08OG2WnbAmm0FL4ObkazC6G_256vN0lN9DsykCvCqGZ)\n* Reporting risky content generated by the model: [developers.facebook.com/llama_output_feedback](http://developers.facebook.com/llama_output_feedback)\n* Reporting bugs and security concerns: [facebook.com/whitehat/info](http://facebook.com/whitehat/info)\n* Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama 3.2: [email protected]", "extra_gated_fields": {"First Name": "text", "Last Name": "text", "Date of birth": "date_picker", "Country": "country", "Affiliation": "text", "Job title": {"type": "select", "options": ["Student", "Research Graduate", "AI researcher", "AI developer/engineer", "Reporter", "Other"]}, "geo": "ip_location", "By clicking Submit below I accept the terms of the license and acknowledge that the information I provide will be collected stored processed and shared in accordance with the Meta Privacy Policy": "checkbox"}, "extra_gated_description": "The information you provide will be collected, stored, processed and shared in accordance with the [Meta Privacy Policy](https://www.facebook.com/privacy/policy/).", "extra_gated_button_content": "Submit"}	task	[ "SUMMARIZATION" ]	43,113
gaudi/opus-mt-fi-cs-ctranslate2	gaudi	translation	[ "transformers", "marian", "ctranslate2", "translation", "license:apache-2.0", "endpoints_compatible", "region:us" ]	2024-07-22T15:50:15Z	2024-10-19T03:24:13+00:00	6	0	--- license: apache-2.0 tags: - ctranslate2 - translation --- # Repository General Information ## Inspired by and derived from the work of [Helsinki-NLP](https://huggingface.co/Helsinki-NLP), [CTranslate2](https://github.com/OpenNMT/CTranslate2), and [michaelfeil](https://huggingface.co/michaelfeil)! - Link to Original Model ([Helsinki-NLP](https://huggingface.co/Helsinki-NLP)): [Model Link](https://huggingface.co/Helsinki-NLP/opus-mt-fi-cs) - This respository was based on the work of [CTranslate2](https://github.com/OpenNMT/CTranslate2). - This repository was based on the work of [michaelfeil](https://huggingface.co/michaelfeil). # What is CTranslate2? [CTranslate2](https://opennmt.net/CTranslate2/) is a C++ and Python library for efficient inference with Transformer models. CTranslate2 implements a custom runtime that applies many performance optimization techniques such as weights quantization, layers fusion, batch reordering, etc., to accelerate and reduce the memory usage of Transformer models on CPU and GPU. CTranslate2 is one of the most performant ways of hosting translation models at scale. Current supported models include: - Encoder-decoder models: Transformer base/big, M2M-100, NLLB, BART, mBART, Pegasus, T5, Whisper - Decoder-only models: GPT-2, GPT-J, GPT-NeoX, OPT, BLOOM, MPT, Llama, Mistral, Gemma, CodeGen, GPTBigCode, Falcon - Encoder-only models: BERT, DistilBERT, XLM-RoBERTa The project is production-oriented and comes with backward compatibility guarantees, but it also includes experimental features related to model compression and inference acceleration. # CTranslate2 Benchmarks Please note that the results presented below are only valid for the configuration used during this benchmark: absolute and relative performance may change with different settings. Tested against `newstest2014` (En -> De) dataset. The benchmark reports the number of target tokens generated per second (higher is better). The results are aggregated over multiple runs. See the benchmark scripts for more details and reproduce these numbers. Please note that the results presented below are only valid for the configuration used during this benchmark: absolute and relative performance may change with different settings. ## CPU Benchmarks for Generic Opus-MT Models \| Library \| Tokens per Second \| Max Memory Usage \| BLEU \| \| :----: \| :----: \| :----: \| :----: \| \| Transformers 4.26.1 (with PyTorch 1.13.1) \| 147.3 \| 2332MB \| 27.90 \| \| Marian 1.11.0 (int16) \| 330.2 \| 5901MB \| 27.65 \| \| Marian 1.11.0 (int8) \| 355.8 \| 4763MB \| 27.27 \| \| CTranslate2 3.6.0 (int16) \| 596.1 \| 660MB \| 27.53 \| \| CTranslate2 3.6.0 (int8) \| 696.1 \| 516MB \| 27.65 \| ## GPU Benchmarks for Generic Opus-MT Models \| Library \| Tokens per Second \| Max GPU Memory Usage \| Max Memory Usage \| BLEU \| \| :----: \| :----: \| :----: \| :----: \| :----: \| \| Transformers 4.26.1 (with PyTorch 1.13.1) \| 1022.9 \| 4097MB \| 2109MB \| 27.90 \| \| Marian 1.11.0 (float16) \| 3962.4 \| 3239MB \| 1976MB \| 27.94 \| \| CTranslate2 3.6.0 (float16) \| 9296.7 \| 909MB \| 814MB \| 27.9 \| \| CTranslate2 3.6.0 (int8 + float16) \| 8362.7 \| 813MB \| 766MB \| 27.9 \| `Executed with 4 threads on a c5.2xlarge Amazon EC2 instance equipped with an Intel(R) Xeon(R) Platinum 8275CL CPU.` Source to benchmark information can be found [here](https://github.com/OpenNMT/CTranslate2).<br /> Original model BLEU scores can be found [here](https://huggingface.co/Helsinki-NLP/opus-mt-fi-cs). ## Internal Benchmarks Internal testing on our end showed inference times reduced by 6x-10x on average compared the vanilla checkpoints using the transformers library. A slight reduction on BLEU scores (~5%) was also identified in comparison to the vanilla checkpoints with a few exceptions. This is likely due to several factors, one being the quantization applied. Further testing is needed from our end to better assess the reduction in translation quality. The command used to compile the vanilla checkpoint into a CTranslate2 model can be found below. Modifying this command can yield differing balances between inferencing performance and translation quality. # CTranslate2 Installation ```bash pip install hf-hub-ctranslate2>=1.0.0 ctranslate2>=3.13.0 ``` ### ct2-transformers-converter Command Used: ```bash ct2-transformers-converter --model Helsinki-NLP/opus-mt-fi-cs --output_dir ./ctranslate2/opus-mt-fi-cs-ctranslate2 --force --copy_files README.md generation_config.json tokenizer_config.json vocab.json source.spm .gitattributes target.spm --quantization float16 ``` # CTranslate2 Converted Checkpoint Information: Compatible With: - [ctranslate2](https://github.com/OpenNMT/CTranslate2) - [hf-hub-ctranslate2](https://github.com/michaelfeil/hf-hub-ctranslate2) Compute Type: - `compute_type=int8_float16` for `device="cuda"` - `compute_type=int8` for `device="cpu"` # Sample Code - ctranslate2 #### Clone the repository to the working directory or wherever you wish to store the model artifacts. #### ```bash git clone https://huggingface.co/gaudi/opus-mt-fi-cs-ctranslate2 ``` #### Take the python code below and update the 'model_dir' variable to the location of the cloned repository. #### ```python from ctranslate2 import Translator import transformers model_dir = "./opus-mt-fi-cs-ctranslate2" # Path to model directory. translator = Translator( model_path=model_dir, device="cuda", # cpu, cuda, or auto. inter_threads=1, # Maximum number of parallel translations. intra_threads=4, # Number of OpenMP threads per translator. compute_type="int8_float16", # int8 for cpu or int8_float16 for cuda. ) tokenizer = transformers.AutoTokenizer.from_pretrained(model_dir) source = tokenizer.convert_ids_to_tokens(tokenizer.encode("XXXXXX, XXX XX XXXXXX.")) results = translator.translate_batch([source]) target = results[0].hypotheses[0] print(tokenizer.decode(tokenizer.convert_tokens_to_ids(target))) ``` # Sample Code - hf-hub-ctranslate2 Derived From [michaelfeil](https://huggingface.co/michaelfeil): ```python from hf_hub_ctranslate2 import TranslatorCT2fromHfHub, GeneratorCT2fromHfHub from transformers import AutoTokenizer model_name = "gaudi/opus-mt-fi-cs-ctranslate2" model = TranslatorCT2fromHfHub( model_name_or_path=model_name, device="cuda", compute_type="int8_float16", tokenizer=AutoTokenizer.from_pretrained(model_name) ) outputs = model.generate( text=["XXX XX XXX XXXXXXX XXXX?", "XX XX XXXX XX XXX!"], ) print(outputs) ``` # License and other remarks: License conditions are intended to be idential to [original huggingface repository](https://huggingface.co/Helsinki-NLP/opus-mt-fi-cs) by Helsinki-NLP.	null	Non_BioNLP	# Repository General Information ## Inspired by and derived from the work of [Helsinki-NLP](https://huggingface.co/Helsinki-NLP), [CTranslate2](https://github.com/OpenNMT/CTranslate2), and [michaelfeil](https://huggingface.co/michaelfeil)! - Link to Original Model ([Helsinki-NLP](https://huggingface.co/Helsinki-NLP)): [Model Link](https://huggingface.co/Helsinki-NLP/opus-mt-fi-cs) - This respository was based on the work of [CTranslate2](https://github.com/OpenNMT/CTranslate2). - This repository was based on the work of [michaelfeil](https://huggingface.co/michaelfeil). # What is CTranslate2? [CTranslate2](https://opennmt.net/CTranslate2/) is a C++ and Python library for efficient inference with Transformer models. CTranslate2 implements a custom runtime that applies many performance optimization techniques such as weights quantization, layers fusion, batch reordering, etc., to accelerate and reduce the memory usage of Transformer models on CPU and GPU. CTranslate2 is one of the most performant ways of hosting translation models at scale. Current supported models include: - Encoder-decoder models: Transformer base/big, M2M-100, NLLB, BART, mBART, Pegasus, T5, Whisper - Decoder-only models: GPT-2, GPT-J, GPT-NeoX, OPT, BLOOM, MPT, Llama, Mistral, Gemma, CodeGen, GPTBigCode, Falcon - Encoder-only models: BERT, DistilBERT, XLM-RoBERTa The project is production-oriented and comes with backward compatibility guarantees, but it also includes experimental features related to model compression and inference acceleration. # CTranslate2 Benchmarks Please note that the results presented below are only valid for the configuration used during this benchmark: absolute and relative performance may change with different settings. Tested against `newstest2014` (En -> De) dataset. The benchmark reports the number of target tokens generated per second (higher is better). The results are aggregated over multiple runs. See the benchmark scripts for more details and reproduce these numbers. Please note that the results presented below are only valid for the configuration used during this benchmark: absolute and relative performance may change with different settings. ## CPU Benchmarks for Generic Opus-MT Models \| Library \| Tokens per Second \| Max Memory Usage \| BLEU \| \| :----: \| :----: \| :----: \| :----: \| \| Transformers 4.26.1 (with PyTorch 1.13.1) \| 147.3 \| 2332MB \| 27.90 \| \| Marian 1.11.0 (int16) \| 330.2 \| 5901MB \| 27.65 \| \| Marian 1.11.0 (int8) \| 355.8 \| 4763MB \| 27.27 \| \| CTranslate2 3.6.0 (int16) \| 596.1 \| 660MB \| 27.53 \| \| CTranslate2 3.6.0 (int8) \| 696.1 \| 516MB \| 27.65 \| ## GPU Benchmarks for Generic Opus-MT Models \| Library \| Tokens per Second \| Max GPU Memory Usage \| Max Memory Usage \| BLEU \| \| :----: \| :----: \| :----: \| :----: \| :----: \| \| Transformers 4.26.1 (with PyTorch 1.13.1) \| 1022.9 \| 4097MB \| 2109MB \| 27.90 \| \| Marian 1.11.0 (float16) \| 3962.4 \| 3239MB \| 1976MB \| 27.94 \| \| CTranslate2 3.6.0 (float16) \| 9296.7 \| 909MB \| 814MB \| 27.9 \| \| CTranslate2 3.6.0 (int8 + float16) \| 8362.7 \| 813MB \| 766MB \| 27.9 \| `Executed with 4 threads on a c5.2xlarge Amazon EC2 instance equipped with an Intel(R) Xeon(R) Platinum 8275CL CPU.` Source to benchmark information can be found [here](https://github.com/OpenNMT/CTranslate2).<br /> Original model BLEU scores can be found [here](https://huggingface.co/Helsinki-NLP/opus-mt-fi-cs). ## Internal Benchmarks Internal testing on our end showed inference times reduced by 6x-10x on average compared the vanilla checkpoints using the transformers library. A slight reduction on BLEU scores (~5%) was also identified in comparison to the vanilla checkpoints with a few exceptions. This is likely due to several factors, one being the quantization applied. Further testing is needed from our end to better assess the reduction in translation quality. The command used to compile the vanilla checkpoint into a CTranslate2 model can be found below. Modifying this command can yield differing balances between inferencing performance and translation quality. # CTranslate2 Installation ```bash pip install hf-hub-ctranslate2>=1.0.0 ctranslate2>=3.13.0 ``` ### ct2-transformers-converter Command Used: ```bash ct2-transformers-converter --model Helsinki-NLP/opus-mt-fi-cs --output_dir ./ctranslate2/opus-mt-fi-cs-ctranslate2 --force --copy_files README.md generation_config.json tokenizer_config.json vocab.json source.spm .gitattributes target.spm --quantization float16 ``` # CTranslate2 Converted Checkpoint Information: Compatible With: - [ctranslate2](https://github.com/OpenNMT/CTranslate2) - [hf-hub-ctranslate2](https://github.com/michaelfeil/hf-hub-ctranslate2) Compute Type: - `compute_type=int8_float16` for `device="cuda"` - `compute_type=int8` for `device="cpu"` # Sample Code - ctranslate2 #### Clone the repository to the working directory or wherever you wish to store the model artifacts. #### ```bash git clone https://huggingface.co/gaudi/opus-mt-fi-cs-ctranslate2 ``` #### Take the python code below and update the 'model_dir' variable to the location of the cloned repository. #### ```python from ctranslate2 import Translator import transformers model_dir = "./opus-mt-fi-cs-ctranslate2" # Path to model directory. translator = Translator( model_path=model_dir, device="cuda", # cpu, cuda, or auto. inter_threads=1, # Maximum number of parallel translations. intra_threads=4, # Number of OpenMP threads per translator. compute_type="int8_float16", # int8 for cpu or int8_float16 for cuda. ) tokenizer = transformers.AutoTokenizer.from_pretrained(model_dir) source = tokenizer.convert_ids_to_tokens(tokenizer.encode("XXXXXX, XXX XX XXXXXX.")) results = translator.translate_batch([source]) target = results[0].hypotheses[0] print(tokenizer.decode(tokenizer.convert_tokens_to_ids(target))) ``` # Sample Code - hf-hub-ctranslate2 Derived From [michaelfeil](https://huggingface.co/michaelfeil): ```python from hf_hub_ctranslate2 import TranslatorCT2fromHfHub, GeneratorCT2fromHfHub from transformers import AutoTokenizer model_name = "gaudi/opus-mt-fi-cs-ctranslate2" model = TranslatorCT2fromHfHub( model_name_or_path=model_name, device="cuda", compute_type="int8_float16", tokenizer=AutoTokenizer.from_pretrained(model_name) ) outputs = model.generate( text=["XXX XX XXX XXXXXXX XXXX?", "XX XX XXXX XX XXX!"], ) print(outputs) ``` # License and other remarks: License conditions are intended to be idential to [original huggingface repository](https://huggingface.co/Helsinki-NLP/opus-mt-fi-cs) by Helsinki-NLP.	{"license": "apache-2.0", "tags": ["ctranslate2", "translation"]}	task	[ "TRANSLATION" ]	43,114
dustyatx/bge-base-financial-matryoshka	dustyatx	sentence-similarity	[ "sentence-transformers", "safetensors", "bert", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:6300", "loss:MatryoshkaLoss", "loss:MultipleNegativesRankingLoss", "en", "arxiv:1908.10084", "arxiv:2205.13147", "arxiv:1705.00652", "base_model:BAAI/bge-base-en-v1.5", "base_model:finetune:BAAI/bge-base-en-v1.5", "license:apache-2.0", "model-index", "autotrain_compatible", "text-embeddings-inference", "endpoints_compatible", "region:us" ]	2024-09-02T13:00:03Z	2024-09-02T13:00:19+00:00	9	0	--- base_model: BAAI/bge-base-en-v1.5 datasets: [] language: - en library_name: sentence-transformers license: apache-2.0 metrics: - cosine_accuracy@1 - cosine_accuracy@3 - cosine_accuracy@5 - cosine_accuracy@10 - cosine_precision@1 - cosine_precision@3 - cosine_precision@5 - cosine_precision@10 - cosine_recall@1 - cosine_recall@3 - cosine_recall@5 - cosine_recall@10 - cosine_ndcg@10 - cosine_mrr@10 - cosine_map@100 pipeline_tag: sentence-similarity tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:6300 - loss:MatryoshkaLoss - loss:MultipleNegativesRankingLoss widget: - source_sentence: Total net additions to property and equipment for AWS in 2023 amounted to $24,843 million. sentences: - What technological feature helps protect digital transactions in the Visa Token Service? - What was the total net addition to property and equipment for AWS in the year 2023? - By what proportion did net cash used in financing activities increase from 2022 to 2023? - source_sentence: 'Leases generally contain one or more of the following options, which the Company can exercise at the end of the initial term: (a) renew the lease for a defined number of years at the then-fair market rental rate or rate stipulated in the lease agreement; (b) purchase the property at the then-fair market value or purchase price stated in the agreement; or (c) a right of first refusal in the event of a third-party offer.' sentences: - What are the requirements for health insurers and group health plans in providing cost estimates to consumers? - What options does the company have at the end of the lease term for their leased properties? - How much did the company incur in intangible amortization costs related to the eOne acquisition in 2022? - source_sentence: We recorded an acquisition termination cost of $1.35 billion in fiscal year 2023 reflecting the write-off of the prepayment provided at signing. sentences: - How much did NVIDIA record as an acquisition termination cost in fiscal year 2023 related to the Arm Share Purchase Agreement? - What is included in the consolidated financial statements and accompanying notes mentioned in Part IV, Item 15(a)(1) of the Annual Report on Form 10-K? - What risks are associated with projecting the effectiveness of internal controls into future periods as mentioned? - source_sentence: Item 8 is labeled as Financial Statements and Supplementary Data. sentences: - What was the percentage of trading days in 2023 where trading-related revenue was recorded as positive? - How is the discount rate for the Family Dollar goodwill impairment evaluation determined? - What is the title of Item 8 in the financial document? - source_sentence: Details about legal proceedings are included in Part II, Item 8, "Financial Statements and Supplementary Data" of the Annual Report on Form 10-K, under the caption "Legal Proceedings". sentences: - Where can details about legal proceedings be located in an Annual Report on Form 10-K? - How many stores did AutoZone operate in the United States as of August 26, 2023? - In the context of Hewlett Packard Enterprise's recent financial discussions, what factors are expected to impact their operational costs and revenue growth moving forward? model-index: - name: BGE base Financial Matryoshka results: - task: type: information-retrieval name: Information Retrieval dataset: name: dim 768 type: dim_768 metrics: - type: cosine_accuracy@1 value: 0.7071428571428572 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.8414285714285714 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.88 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.9314285714285714 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.7071428571428572 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.28047619047619043 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.176 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09314285714285712 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.7071428571428572 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.8414285714285714 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.88 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.9314285714285714 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.8207437059171859 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7853486394557823 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7881907906804949 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 512 type: dim_512 metrics: - type: cosine_accuracy@1 value: 0.6957142857142857 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.8385714285714285 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8757142857142857 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.93 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.6957142857142857 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.2795238095238095 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.17514285714285713 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09299999999999999 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.6957142857142857 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.8385714285714285 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8757142857142857 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.93 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.8149439460863356 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7780714285714285 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.781021025356189 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 256 type: dim_256 metrics: - type: cosine_accuracy@1 value: 0.6885714285714286 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.83 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8742857142857143 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.9142857142857143 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.6885714285714286 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.27666666666666667 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.17485714285714282 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09142857142857141 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.6885714285714286 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.83 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8742857142857143 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.9142857142857143 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.8060991379418679 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7710873015873015 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7751792513774886 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 128 type: dim_128 metrics: - type: cosine_accuracy@1 value: 0.6771428571428572 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.8214285714285714 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8614285714285714 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.9142857142857143 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.6771428571428572 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.2738095238095238 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.17228571428571426 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09142857142857141 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.6771428571428572 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.8214285714285714 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8614285714285714 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.9142857142857143 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.7979494993398927 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7605890022675734 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7639633810343436 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 64 type: dim_64 metrics: - type: cosine_accuracy@1 value: 0.6557142857142857 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.7871428571428571 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8271428571428572 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.8714285714285714 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.6557142857142857 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.2623809523809524 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.1654285714285714 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.08714285714285713 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.6557142857142857 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.7871428571428571 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8271428571428572 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.8714285714285714 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.7664083634078753 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7326604308390022 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7375736792740525 name: Cosine Map@100 --- # BGE base Financial Matryoshka This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5). It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) <!-- at revision a5beb1e3e68b9ab74eb54cfd186867f64f240e1a --> - Maximum Sequence Length: 512 tokens - Output Dimensionality: 768 tokens - Similarity Function: Cosine Similarity <!-- - Training Dataset: Unknown --> - Language: en - License: apache-2.0 ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("dustyatx/bge-base-financial-matryoshka") # Run inference sentences = [ 'Details about legal proceedings are included in Part II, Item 8, "Financial Statements and Supplementary Data" of the Annual Report on Form 10-K, under the caption "Legal Proceedings".', 'Where can details about legal proceedings be located in an Annual Report on Form 10-K?', 'How many stores did AutoZone operate in the United States as of August 26, 2023?', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 768] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Information Retrieval * Dataset: `dim_768` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.7071 \| \| cosine_accuracy@3 \| 0.8414 \| \| cosine_accuracy@5 \| 0.88 \| \| cosine_accuracy@10 \| 0.9314 \| \| cosine_precision@1 \| 0.7071 \| \| cosine_precision@3 \| 0.2805 \| \| cosine_precision@5 \| 0.176 \| \| cosine_precision@10 \| 0.0931 \| \| cosine_recall@1 \| 0.7071 \| \| cosine_recall@3 \| 0.8414 \| \| cosine_recall@5 \| 0.88 \| \| cosine_recall@10 \| 0.9314 \| \| cosine_ndcg@10 \| 0.8207 \| \| cosine_mrr@10 \| 0.7853 \| \| cosine_map@100 \| 0.7882 \| #### Information Retrieval * Dataset: `dim_512` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| cosine_accuracy@1 \| 0.6957 \| \| cosine_accuracy@3 \| 0.8386 \| \| cosine_accuracy@5 \| 0.8757 \| \| cosine_accuracy@10 \| 0.93 \| \| cosine_precision@1 \| 0.6957 \| \| cosine_precision@3 \| 0.2795 \| \| cosine_precision@5 \| 0.1751 \| \| cosine_precision@10 \| 0.093 \| \| cosine_recall@1 \| 0.6957 \| \| cosine_recall@3 \| 0.8386 \| \| cosine_recall@5 \| 0.8757 \| \| cosine_recall@10 \| 0.93 \| \| cosine_ndcg@10 \| 0.8149 \| \| cosine_mrr@10 \| 0.7781 \| \| cosine_map@100 \| 0.781 \| #### Information Retrieval * Dataset: `dim_256` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.6886 \| \| cosine_accuracy@3 \| 0.83 \| \| cosine_accuracy@5 \| 0.8743 \| \| cosine_accuracy@10 \| 0.9143 \| \| cosine_precision@1 \| 0.6886 \| \| cosine_precision@3 \| 0.2767 \| \| cosine_precision@5 \| 0.1749 \| \| cosine_precision@10 \| 0.0914 \| \| cosine_recall@1 \| 0.6886 \| \| cosine_recall@3 \| 0.83 \| \| cosine_recall@5 \| 0.8743 \| \| cosine_recall@10 \| 0.9143 \| \| cosine_ndcg@10 \| 0.8061 \| \| cosine_mrr@10 \| 0.7711 \| \| cosine_map@100 \| 0.7752 \| #### Information Retrieval * Dataset: `dim_128` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| cosine_accuracy@1 \| 0.6771 \| \| cosine_accuracy@3 \| 0.8214 \| \| cosine_accuracy@5 \| 0.8614 \| \| cosine_accuracy@10 \| 0.9143 \| \| cosine_precision@1 \| 0.6771 \| \| cosine_precision@3 \| 0.2738 \| \| cosine_precision@5 \| 0.1723 \| \| cosine_precision@10 \| 0.0914 \| \| cosine_recall@1 \| 0.6771 \| \| cosine_recall@3 \| 0.8214 \| \| cosine_recall@5 \| 0.8614 \| \| cosine_recall@10 \| 0.9143 \| \| cosine_ndcg@10 \| 0.7979 \| \| cosine_mrr@10 \| 0.7606 \| \| cosine_map@100 \| 0.764 \| #### Information Retrieval * Dataset: `dim_64` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.6557 \| \| cosine_accuracy@3 \| 0.7871 \| \| cosine_accuracy@5 \| 0.8271 \| \| cosine_accuracy@10 \| 0.8714 \| \| cosine_precision@1 \| 0.6557 \| \| cosine_precision@3 \| 0.2624 \| \| cosine_precision@5 \| 0.1654 \| \| cosine_precision@10 \| 0.0871 \| \| cosine_recall@1 \| 0.6557 \| \| cosine_recall@3 \| 0.7871 \| \| cosine_recall@5 \| 0.8271 \| \| cosine_recall@10 \| 0.8714 \| \| cosine_ndcg@10 \| 0.7664 \| \| cosine_mrr@10 \| 0.7327 \| \| cosine_map@100 \| 0.7376 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 6,300 training samples * Columns: <code>positive</code> and <code>anchor</code> * Approximate statistics based on the first 1000 samples: \| \| positive \| anchor \| \|:--------\|:-----------------------------------------------------------------------------------\|:---------------------------------------------------------------------------------\| \| type \| string \| string \| \| details \| <ul><li>min: 8 tokens</li><li>mean: 45.94 tokens</li><li>max: 512 tokens</li></ul> \| <ul><li>min: 7 tokens</li><li>mean: 20.7 tokens</li><li>max: 42 tokens</li></ul> \| * Samples: \| positive \| anchor \| \|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>The company must continuously strengthen its capabilities in marketing and innovation to compete in a digital environment and maintain brand loyalty and marketallability. In addition, it is increasing its investments in e-commerce to support retail and meal delivery services, offering more package sizes that are fit-for-purpose for online sales and shifting more consumer and trade promotions to digital.</code> \| <code>What strategies is the company employing to enhance its competitiveness in a digital environment?</code> \| \| <code>Fedflowing expanded or relocated its hub and linehaul network, FedEx Ground also introduced new safety technologies, set new driver standards, and made operational enhancements for safer handling of heavy items.</code> \| <code>What specific changes has FedEx Ground made for vehicle and driver safety?</code> \| \| <code>The debt financing, which is being provided by a syndicate of Chinese financial institutions, contains certain covenants and a maximum borrowing limit of ¥29.7 billion RMB (approximately $4.2 billion).</code> \| <code>What is the maximum borrowing limit of the debt financing provided by the syndicate of Chinese financial institutions for Universal Beijing Resort?</code> \| * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 768, 512, 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: epoch - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `gradient_accumulation_steps`: 16 - `learning_rate`: 2e-05 - `num_train_epochs`: 4 - `lr_scheduler_type`: cosine - `warmup_ratio`: 0.1 - `bf16`: True - `tf32`: True - `load_best_model_at_end`: True - `optim`: adamw_torch_fused - `batch_sampler`: no_duplicates #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: epoch - `prediction_loss_only`: True - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 16 - `eval_accumulation_steps`: None - `torch_empty_cache_steps`: None - `learning_rate`: 2e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 4 - `max_steps`: -1 - `lr_scheduler_type`: cosine - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.1 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: True - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: True - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: True - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch_fused - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `eval_use_gather_object`: False - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional </details> ### Training Logs \| Epoch \| Step \| Training Loss \| dim_128_cosine_map@100 \| dim_256_cosine_map@100 \| dim_512_cosine_map@100 \| dim_64_cosine_map@100 \| dim_768_cosine_map@100 \| \|:----------:\|:------:\|:-------------:\|:----------------------:\|:----------------------:\|:----------------------:\|:---------------------:\|:----------------------:\| \| 0.8122 \| 10 \| 1.5212 \| - \| - \| - \| - \| - \| \| 0.9746 \| 12 \| - \| 0.7439 \| 0.7556 \| 0.7670 \| 0.7142 \| 0.7717 \| \| 1.6244 \| 20 \| 0.6418 \| - \| - \| - \| - \| - \| \| 1.9492 \| 24 \| - \| 0.7592 \| 0.7743 \| 0.7787 \| 0.7331 \| 0.7839 \| \| 2.4365 \| 30 \| 0.4411 \| - \| - \| - \| - \| - \| \| 2.9239 \| 36 \| - \| 0.7623 \| 0.7757 \| 0.7816 \| 0.7365 \| 0.7902 \| \| 3.2487 \| 40 \| 0.3917 \| - \| - \| - \| - \| - \| \| 3.8985 \| 48 \| - \| 0.764 \| 0.7752 \| 0.781 \| 0.7376 \| 0.7882 \| * The bold row denotes the saved checkpoint. ### Framework Versions - Python: 3.11.9 - Sentence Transformers: 3.0.1 - Transformers: 4.44.2 - PyTorch: 2.4.0+cu121 - Accelerate: 0.33.0 - Datasets: 2.21.0 - Tokenizers: 0.19.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MatryoshkaLoss ```bibtex @misc{kusupati2024matryoshka, title={Matryoshka Representation Learning}, author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi}, year={2024}, eprint={2205.13147}, archivePrefix={arXiv}, primaryClass={cs.LG} } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	null	Non_BioNLP	# BGE base Financial Matryoshka This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5). It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) <!-- at revision a5beb1e3e68b9ab74eb54cfd186867f64f240e1a --> - Maximum Sequence Length: 512 tokens - Output Dimensionality: 768 tokens - Similarity Function: Cosine Similarity <!-- - Training Dataset: Unknown --> - Language: en - License: apache-2.0 ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("dustyatx/bge-base-financial-matryoshka") # Run inference sentences = [ 'Details about legal proceedings are included in Part II, Item 8, "Financial Statements and Supplementary Data" of the Annual Report on Form 10-K, under the caption "Legal Proceedings".', 'Where can details about legal proceedings be located in an Annual Report on Form 10-K?', 'How many stores did AutoZone operate in the United States as of August 26, 2023?', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 768] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Information Retrieval * Dataset: `dim_768` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.7071 \| \| cosine_accuracy@3 \| 0.8414 \| \| cosine_accuracy@5 \| 0.88 \| \| cosine_accuracy@10 \| 0.9314 \| \| cosine_precision@1 \| 0.7071 \| \| cosine_precision@3 \| 0.2805 \| \| cosine_precision@5 \| 0.176 \| \| cosine_precision@10 \| 0.0931 \| \| cosine_recall@1 \| 0.7071 \| \| cosine_recall@3 \| 0.8414 \| \| cosine_recall@5 \| 0.88 \| \| cosine_recall@10 \| 0.9314 \| \| cosine_ndcg@10 \| 0.8207 \| \| cosine_mrr@10 \| 0.7853 \| \| cosine_map@100 \| 0.7882 \| #### Information Retrieval * Dataset: `dim_512` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| cosine_accuracy@1 \| 0.6957 \| \| cosine_accuracy@3 \| 0.8386 \| \| cosine_accuracy@5 \| 0.8757 \| \| cosine_accuracy@10 \| 0.93 \| \| cosine_precision@1 \| 0.6957 \| \| cosine_precision@3 \| 0.2795 \| \| cosine_precision@5 \| 0.1751 \| \| cosine_precision@10 \| 0.093 \| \| cosine_recall@1 \| 0.6957 \| \| cosine_recall@3 \| 0.8386 \| \| cosine_recall@5 \| 0.8757 \| \| cosine_recall@10 \| 0.93 \| \| cosine_ndcg@10 \| 0.8149 \| \| cosine_mrr@10 \| 0.7781 \| \| cosine_map@100 \| 0.781 \| #### Information Retrieval * Dataset: `dim_256` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.6886 \| \| cosine_accuracy@3 \| 0.83 \| \| cosine_accuracy@5 \| 0.8743 \| \| cosine_accuracy@10 \| 0.9143 \| \| cosine_precision@1 \| 0.6886 \| \| cosine_precision@3 \| 0.2767 \| \| cosine_precision@5 \| 0.1749 \| \| cosine_precision@10 \| 0.0914 \| \| cosine_recall@1 \| 0.6886 \| \| cosine_recall@3 \| 0.83 \| \| cosine_recall@5 \| 0.8743 \| \| cosine_recall@10 \| 0.9143 \| \| cosine_ndcg@10 \| 0.8061 \| \| cosine_mrr@10 \| 0.7711 \| \| cosine_map@100 \| 0.7752 \| #### Information Retrieval * Dataset: `dim_128` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| cosine_accuracy@1 \| 0.6771 \| \| cosine_accuracy@3 \| 0.8214 \| \| cosine_accuracy@5 \| 0.8614 \| \| cosine_accuracy@10 \| 0.9143 \| \| cosine_precision@1 \| 0.6771 \| \| cosine_precision@3 \| 0.2738 \| \| cosine_precision@5 \| 0.1723 \| \| cosine_precision@10 \| 0.0914 \| \| cosine_recall@1 \| 0.6771 \| \| cosine_recall@3 \| 0.8214 \| \| cosine_recall@5 \| 0.8614 \| \| cosine_recall@10 \| 0.9143 \| \| cosine_ndcg@10 \| 0.7979 \| \| cosine_mrr@10 \| 0.7606 \| \| cosine_map@100 \| 0.764 \| #### Information Retrieval * Dataset: `dim_64` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.6557 \| \| cosine_accuracy@3 \| 0.7871 \| \| cosine_accuracy@5 \| 0.8271 \| \| cosine_accuracy@10 \| 0.8714 \| \| cosine_precision@1 \| 0.6557 \| \| cosine_precision@3 \| 0.2624 \| \| cosine_precision@5 \| 0.1654 \| \| cosine_precision@10 \| 0.0871 \| \| cosine_recall@1 \| 0.6557 \| \| cosine_recall@3 \| 0.7871 \| \| cosine_recall@5 \| 0.8271 \| \| cosine_recall@10 \| 0.8714 \| \| cosine_ndcg@10 \| 0.7664 \| \| cosine_mrr@10 \| 0.7327 \| \| cosine_map@100 \| 0.7376 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 6,300 training samples * Columns: <code>positive</code> and <code>anchor</code> * Approximate statistics based on the first 1000 samples: \| \| positive \| anchor \| \|:--------\|:-----------------------------------------------------------------------------------\|:---------------------------------------------------------------------------------\| \| type \| string \| string \| \| details \| <ul><li>min: 8 tokens</li><li>mean: 45.94 tokens</li><li>max: 512 tokens</li></ul> \| <ul><li>min: 7 tokens</li><li>mean: 20.7 tokens</li><li>max: 42 tokens</li></ul> \| * Samples: \| positive \| anchor \| \|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>The company must continuously strengthen its capabilities in marketing and innovation to compete in a digital environment and maintain brand loyalty and marketallability. In addition, it is increasing its investments in e-commerce to support retail and meal delivery services, offering more package sizes that are fit-for-purpose for online sales and shifting more consumer and trade promotions to digital.</code> \| <code>What strategies is the company employing to enhance its competitiveness in a digital environment?</code> \| \| <code>Fedflowing expanded or relocated its hub and linehaul network, FedEx Ground also introduced new safety technologies, set new driver standards, and made operational enhancements for safer handling of heavy items.</code> \| <code>What specific changes has FedEx Ground made for vehicle and driver safety?</code> \| \| <code>The debt financing, which is being provided by a syndicate of Chinese financial institutions, contains certain covenants and a maximum borrowing limit of ¥29.7 billion RMB (approximately $4.2 billion).</code> \| <code>What is the maximum borrowing limit of the debt financing provided by the syndicate of Chinese financial institutions for Universal Beijing Resort?</code> \| * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 768, 512, 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: epoch - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `gradient_accumulation_steps`: 16 - `learning_rate`: 2e-05 - `num_train_epochs`: 4 - `lr_scheduler_type`: cosine - `warmup_ratio`: 0.1 - `bf16`: True - `tf32`: True - `load_best_model_at_end`: True - `optim`: adamw_torch_fused - `batch_sampler`: no_duplicates #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: epoch - `prediction_loss_only`: True - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 16 - `eval_accumulation_steps`: None - `torch_empty_cache_steps`: None - `learning_rate`: 2e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 4 - `max_steps`: -1 - `lr_scheduler_type`: cosine - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.1 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: True - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: True - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: True - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch_fused - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `eval_use_gather_object`: False - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional </details> ### Training Logs \| Epoch \| Step \| Training Loss \| dim_128_cosine_map@100 \| dim_256_cosine_map@100 \| dim_512_cosine_map@100 \| dim_64_cosine_map@100 \| dim_768_cosine_map@100 \| \|:----------:\|:------:\|:-------------:\|:----------------------:\|:----------------------:\|:----------------------:\|:---------------------:\|:----------------------:\| \| 0.8122 \| 10 \| 1.5212 \| - \| - \| - \| - \| - \| \| 0.9746 \| 12 \| - \| 0.7439 \| 0.7556 \| 0.7670 \| 0.7142 \| 0.7717 \| \| 1.6244 \| 20 \| 0.6418 \| - \| - \| - \| - \| - \| \| 1.9492 \| 24 \| - \| 0.7592 \| 0.7743 \| 0.7787 \| 0.7331 \| 0.7839 \| \| 2.4365 \| 30 \| 0.4411 \| - \| - \| - \| - \| - \| \| 2.9239 \| 36 \| - \| 0.7623 \| 0.7757 \| 0.7816 \| 0.7365 \| 0.7902 \| \| 3.2487 \| 40 \| 0.3917 \| - \| - \| - \| - \| - \| \| 3.8985 \| 48 \| - \| 0.764 \| 0.7752 \| 0.781 \| 0.7376 \| 0.7882 \| * The bold row denotes the saved checkpoint. ### Framework Versions - Python: 3.11.9 - Sentence Transformers: 3.0.1 - Transformers: 4.44.2 - PyTorch: 2.4.0+cu121 - Accelerate: 0.33.0 - Datasets: 2.21.0 - Tokenizers: 0.19.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MatryoshkaLoss ```bibtex @misc{kusupati2024matryoshka, title={Matryoshka Representation Learning}, author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi}, year={2024}, eprint={2205.13147}, archivePrefix={arXiv}, primaryClass={cs.LG} } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	{"base_model": "BAAI/bge-base-en-v1.5", "datasets": [], "language": ["en"], "library_name": "sentence-transformers", "license": "apache-2.0", "metrics": ["cosine_accuracy@1", "cosine_accuracy@3", "cosine_accuracy@5", "cosine_accuracy@10", "cosine_precision@1", "cosine_precision@3", "cosine_precision@5", "cosine_precision@10", "cosine_recall@1", "cosine_recall@3", "cosine_recall@5", "cosine_recall@10", "cosine_ndcg@10", "cosine_mrr@10", "cosine_map@100"], "pipeline_tag": "sentence-similarity", "tags": ["sentence-transformers", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:6300", "loss:MatryoshkaLoss", "loss:MultipleNegativesRankingLoss"], "widget": [{"source_sentence": "Total net additions to property and equipment for AWS in 2023 amounted to $24,843 million.", "sentences": ["What technological feature helps protect digital transactions in the Visa Token Service?", "What was the total net addition to property and equipment for AWS in the year 2023?", "By what proportion did net cash used in financing activities increase from 2022 to 2023?"]}, {"source_sentence": "Leases generally contain one or more of the following options, which the Company can exercise at the end of the initial term: (a) renew the lease for a defined number of years at the then-fair market rental rate or rate stipulated in the lease agreement; (b) purchase the property at the then-fair market value or purchase price stated in the agreement; or (c) a right of first refusal in the event of a third-party offer.", "sentences": ["What are the requirements for health insurers and group health plans in providing cost estimates to consumers?", "What options does the company have at the end of the lease term for their leased properties?", "How much did the company incur in intangible amortization costs related to the eOne acquisition in 2022?"]}, {"source_sentence": "We recorded an acquisition termination cost of $1.35 billion in fiscal year 2023 reflecting the write-off of the prepayment provided at signing.", "sentences": ["How much did NVIDIA record as an acquisition termination cost in fiscal year 2023 related to the Arm Share Purchase Agreement?", "What is included in the consolidated financial statements and accompanying notes mentioned in Part IV, Item 15(a)(1) of the Annual Report on Form 10-K?", "What risks are associated with projecting the effectiveness of internal controls into future periods as mentioned?"]}, {"source_sentence": "Item 8 is labeled as Financial Statements and Supplementary Data.", "sentences": ["What was the percentage of trading days in 2023 where trading-related revenue was recorded as positive?", "How is the discount rate for the Family Dollar goodwill impairment evaluation determined?", "What is the title of Item 8 in the financial document?"]}, {"source_sentence": "Details about legal proceedings are included in Part II, Item 8, \"Financial Statements and Supplementary Data\" of the Annual Report on Form 10-K, under the caption \"Legal Proceedings\".", "sentences": ["Where can details about legal proceedings be located in an Annual Report on Form 10-K?", "How many stores did AutoZone operate in the United States as of August 26, 2023?", "In the context of Hewlett Packard Enterprise's recent financial discussions, what factors are expected to impact their operational costs and revenue growth moving forward?"]}], "model-index": [{"name": "BGE base Financial Matryoshka", "results": [{"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 768", "type": "dim_768"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.7071428571428572, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 0.8414285714285714, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 0.88, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 0.9314285714285714, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.7071428571428572, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.28047619047619043, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.176, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09314285714285712, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.7071428571428572, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 0.8414285714285714, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 0.88, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 0.9314285714285714, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.8207437059171859, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.7853486394557823, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.7881907906804949, "name": "Cosine Map@100"}]}, {"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 512", "type": "dim_512"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.6957142857142857, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 0.8385714285714285, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 0.8757142857142857, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 0.93, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.6957142857142857, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.2795238095238095, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.17514285714285713, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09299999999999999, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.6957142857142857, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 0.8385714285714285, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 0.8757142857142857, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 0.93, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.8149439460863356, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.7780714285714285, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.781021025356189, "name": "Cosine Map@100"}]}, {"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 256", "type": "dim_256"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.6885714285714286, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 0.83, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 0.8742857142857143, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 0.9142857142857143, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.6885714285714286, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.27666666666666667, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.17485714285714282, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09142857142857141, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.6885714285714286, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 0.83, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 0.8742857142857143, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 0.9142857142857143, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.8060991379418679, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.7710873015873015, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.7751792513774886, "name": "Cosine Map@100"}]}, {"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 128", "type": "dim_128"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.6771428571428572, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 0.8214285714285714, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 0.8614285714285714, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 0.9142857142857143, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.6771428571428572, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.2738095238095238, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.17228571428571426, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09142857142857141, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.6771428571428572, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 0.8214285714285714, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 0.8614285714285714, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 0.9142857142857143, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.7979494993398927, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.7605890022675734, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.7639633810343436, "name": "Cosine Map@100"}]}, {"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 64", "type": "dim_64"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.6557142857142857, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 0.7871428571428571, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 0.8271428571428572, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 0.8714285714285714, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.6557142857142857, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.2623809523809524, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.1654285714285714, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.08714285714285713, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.6557142857142857, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 0.7871428571428571, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 0.8271428571428572, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 0.8714285714285714, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.7664083634078753, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.7326604308390022, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.7375736792740525, "name": "Cosine Map@100"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,115
Helsinki-NLP/opus-mt-kg-en	Helsinki-NLP	translation	[ "transformers", "pytorch", "tf", "marian", "text2text-generation", "translation", "kg", "en", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:04Z	2023-08-16T11:59:31+00:00	57	0	--- license: apache-2.0 tags: - translation --- ### opus-mt-kg-en * source languages: kg * target languages: en * OPUS readme: [kg-en](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/kg-en/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-09.zip](https://object.pouta.csc.fi/OPUS-MT-models/kg-en/opus-2020-01-09.zip) * test set translations: [opus-2020-01-09.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/kg-en/opus-2020-01-09.test.txt) * test set scores: [opus-2020-01-09.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/kg-en/opus-2020-01-09.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.kg.en \| 35.4 \| 0.508 \|	null	Non_BioNLP	### opus-mt-kg-en * source languages: kg * target languages: en * OPUS readme: [kg-en](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/kg-en/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-09.zip](https://object.pouta.csc.fi/OPUS-MT-models/kg-en/opus-2020-01-09.zip) * test set translations: [opus-2020-01-09.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/kg-en/opus-2020-01-09.test.txt) * test set scores: [opus-2020-01-09.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/kg-en/opus-2020-01-09.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.kg.en \| 35.4 \| 0.508 \|	{"license": "apache-2.0", "tags": ["translation"]}	task	[ "TRANSLATION" ]	43,116
spacemanidol/flan-t5-large-6-2-xsum	spacemanidol	text2text-generation	[ "transformers", "pytorch", "t5", "text2text-generation", "generated_from_trainer", "dataset:xsum", "model-index", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "region:us" ]	2023-03-06T18:23:27Z	2023-03-06T19:58:11+00:00	10	0	--- datasets: - xsum metrics: - rouge tags: - generated_from_trainer model-index: - name: large-6-2 results: - task: type: summarization name: Summarization dataset: name: xsum type: xsum config: default split: validation args: default metrics: - type: rouge value: 41.301 name: Rouge1 --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # large-6-2 This model is a fine-tuned version of [x/large-6-2/](https://huggingface.co/x/large-6-2/) on the xsum dataset. It achieves the following results on the evaluation set: - Loss: 1.8450 - Rouge1: 41.301 - Rouge2: 17.9641 - Rougel: 33.7468 - Rougelsum: 33.7296 - Gen Len: 25.0158 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0001 - train_batch_size: 8 - eval_batch_size: 32 - seed: 42 - gradient_accumulation_steps: 8 - total_train_batch_size: 64 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: constant - num_epochs: 3.0 ### Training results ### Framework versions - Transformers 4.27.0.dev0 - Pytorch 1.13.1+cu117 - Datasets 2.10.0 - Tokenizers 0.13.2	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # large-6-2 This model is a fine-tuned version of [x/large-6-2/](https://huggingface.co/x/large-6-2/) on the xsum dataset. It achieves the following results on the evaluation set: - Loss: 1.8450 - Rouge1: 41.301 - Rouge2: 17.9641 - Rougel: 33.7468 - Rougelsum: 33.7296 - Gen Len: 25.0158 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0001 - train_batch_size: 8 - eval_batch_size: 32 - seed: 42 - gradient_accumulation_steps: 8 - total_train_batch_size: 64 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: constant - num_epochs: 3.0 ### Training results ### Framework versions - Transformers 4.27.0.dev0 - Pytorch 1.13.1+cu117 - Datasets 2.10.0 - Tokenizers 0.13.2	{"datasets": ["xsum"], "metrics": ["rouge"], "tags": ["generated_from_trainer"], "model-index": [{"name": "large-6-2", "results": [{"task": {"type": "summarization", "name": "Summarization"}, "dataset": {"name": "xsum", "type": "xsum", "config": "default", "split": "validation", "args": "default"}, "metrics": [{"type": "rouge", "value": 41.301, "name": "Rouge1"}]}]}]}	task	[ "SUMMARIZATION" ]	43,117
bilginn/bert-base-uncased-finetuned-cola	bilginn	text-classification	[ "transformers", "pytorch", "bert", "text-classification", "generated_from_trainer", "dataset:glue", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-05-03T19:32:34Z	2023-05-05T20:49:47+00:00	34	0	--- datasets: - glue license: apache-2.0 metrics: - matthews_correlation tags: - generated_from_trainer model-index: - name: bert-base-uncased-finetuned-cola results: - task: type: text-classification name: Text Classification dataset: name: glue type: glue config: cola split: validation args: cola metrics: - type: matthews_correlation value: 0.5678267214677118 name: Matthews Correlation --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # bert-base-uncased-finetuned-cola This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the glue dataset. It achieves the following results on the evaluation set: - Loss: 1.5922 - Matthews Correlation: 0.5678 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 9.207256119784435e-06 - train_batch_size: 4 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 10 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Matthews Correlation \| \|:-------------:\|:-----:\|:-----:\|:---------------:\|:--------------------:\| \| 0.5811 \| 1.0 \| 2138 \| 0.6179 \| 0.4846 \| \| 0.4698 \| 2.0 \| 4276 \| 0.8083 \| 0.5495 \| \| 0.3161 \| 3.0 \| 6414 \| 1.1152 \| 0.5389 \| \| 0.2499 \| 4.0 \| 8552 \| 1.0719 \| 0.5624 \| \| 0.1755 \| 5.0 \| 10690 \| 1.1734 \| 0.5709 \| \| 0.1511 \| 6.0 \| 12828 \| 1.2383 \| 0.5699 \| \| 0.0738 \| 7.0 \| 14966 \| 1.3802 \| 0.5598 \| \| 0.0677 \| 8.0 \| 17104 \| 1.4711 \| 0.5599 \| \| 0.0509 \| 9.0 \| 19242 \| 1.5751 \| 0.5678 \| \| 0.0397 \| 10.0 \| 21380 \| 1.5922 \| 0.5678 \| ### Framework versions - Transformers 4.28.1 - Pytorch 2.0.0 - Datasets 2.12.0 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # bert-base-uncased-finetuned-cola This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the glue dataset. It achieves the following results on the evaluation set: - Loss: 1.5922 - Matthews Correlation: 0.5678 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 9.207256119784435e-06 - train_batch_size: 4 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 10 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Matthews Correlation \| \|:-------------:\|:-----:\|:-----:\|:---------------:\|:--------------------:\| \| 0.5811 \| 1.0 \| 2138 \| 0.6179 \| 0.4846 \| \| 0.4698 \| 2.0 \| 4276 \| 0.8083 \| 0.5495 \| \| 0.3161 \| 3.0 \| 6414 \| 1.1152 \| 0.5389 \| \| 0.2499 \| 4.0 \| 8552 \| 1.0719 \| 0.5624 \| \| 0.1755 \| 5.0 \| 10690 \| 1.1734 \| 0.5709 \| \| 0.1511 \| 6.0 \| 12828 \| 1.2383 \| 0.5699 \| \| 0.0738 \| 7.0 \| 14966 \| 1.3802 \| 0.5598 \| \| 0.0677 \| 8.0 \| 17104 \| 1.4711 \| 0.5599 \| \| 0.0509 \| 9.0 \| 19242 \| 1.5751 \| 0.5678 \| \| 0.0397 \| 10.0 \| 21380 \| 1.5922 \| 0.5678 \| ### Framework versions - Transformers 4.28.1 - Pytorch 2.0.0 - Datasets 2.12.0 - Tokenizers 0.13.3	{"datasets": ["glue"], "license": "apache-2.0", "metrics": ["matthews_correlation"], "tags": ["generated_from_trainer"], "model-index": [{"name": "bert-base-uncased-finetuned-cola", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "glue", "type": "glue", "config": "cola", "split": "validation", "args": "cola"}, "metrics": [{"type": "matthews_correlation", "value": 0.5678267214677118, "name": "Matthews Correlation"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,118
siddhiii/t5-hindi-colloquial	siddhiii	null	[ "safetensors", "t5", "region:us" ]	2025-02-20T15:50:28Z	2025-02-20T15:52:19+00:00	9	0	--- {} --- # T5 Hindi Colloquial Translation Model This model is fine-tuned from `google-t5/t5-small` for English-to-Hindi colloquial translation. ## 📝 Dataset - Dataset Used: OPUS-100 (en-hi) - Colloquial Texts Added: (mention any additional data used) ## 🚀 Training Details - Model: google-t5/t5-small - Training Steps: 3 epochs - Batch Size: 8 - Optimizer: AdamW ## 📌 How to Use ```python from transformers import pipeline translator = pipeline("translation", model="siddhiii/t5-hindi-colloquial") result = translator("Translate English to Hindi: How are you?") print(result[0]['translation_text'])	null	Non_BioNLP	# T5 Hindi Colloquial Translation Model This model is fine-tuned from `google-t5/t5-small` for English-to-Hindi colloquial translation. ## 📝 Dataset - Dataset Used: OPUS-100 (en-hi) - Colloquial Texts Added: (mention any additional data used) ## 🚀 Training Details - Model: google-t5/t5-small - Training Steps: 3 epochs - Batch Size: 8 - Optimizer: AdamW ## 📌 How to Use ```python from transformers import pipeline translator = pipeline("translation", model="siddhiii/t5-hindi-colloquial") result = translator("Translate English to Hindi: How are you?") print(result[0]['translation_text'])	{}	task	[ "TRANSLATION" ]	43,119
lambdavi/span-marker-luke-legal	lambdavi	token-classification	[ "span-marker", "safetensors", "token-classification", "ner", "named-entity-recognition", "generated_from_span_marker_trainer", "legal", "model-index", "region:us" ]	2024-02-22T10:03:40Z	2024-12-05T13:22:04+00:00	27	3	--- library_name: span-marker metrics: - precision - recall - f1 pipeline_tag: token-classification tags: - span-marker - token-classification - ner - named-entity-recognition - generated_from_span_marker_trainer - legal widget: - text: The seven-judge Constitution Bench of the Supreme Court in SBP and Co. (supra) while reversing earlier five-judge Constitution Bench judgment in Konkan Railway Corpn. Ltd. vs. Rani Construction (P) Ltd., (2002) 2 SCC 388 held that the power exercised by the Chief Justice of the High Court or the Chief justice of India under Section 11(6) of the Arbitration Act is not an administrative power but is a judicial power. - text: 'In The High Court Of Judicature At Patna Criminal Writ Jurisdiction Case No.160 of 2021 Arising Out of Ps. Case No.-58 Year-2020 Thana- Bakhari District- Begusarai ====================================================== Hanif Ur Rahman, son of Azhar Rahman, Resident of C-39, East Nizamuddin, New Delhi....... Petitioner Versus 1. The State of Bihar (through Chief Secretary, Govt. of Bihar) Main Secretariat, Patna - 800015. 2. Meena Khatoon, wife of Mastan @ Noor Mohammad, Resident of Village- Mansurpur Chaksikandar, P.S.- Bidupur, District- Vaishali (Bihar) 3. The Bihar Police, through Standing Counsel. 4. Child Welfare Committee, through Chairperson, Chanakyanagar, Mahmadpur, Begusarai. 5. The Superintendent, Alpawas Grih, Nirala Nagar, Behind G.D. College, Ratanpur, Begusarai....... Respondents ====================================================== Appearance:For the Petitioner:Ms. Kriti Awasthi, Advocate Mr. Sambhav Gupta, Advocate Mr. Navnit Kumar, Advocate Mr. Shyam Kumar, Advocate For the Respondents:Mr.Nadim Seraj, G.P.5 For the Resp. No. 2:Ms. Archana Sinha, Advocate For the Resp. No. 4:Mr. Prabhu Narain Sharma, Advocate ====================================================== Coram: Honourable Mr. Justice Rajeev Ranjan Prasad C.A.V. Judgment' - text: '1 R In The High Court Of Karnataka At Bengaluru Dated This The 19Th Day Of February, 2021 Before The Hon''Ble Mr. Justice H.P. Sandesh Criminal Appeal No.176/2011 Between: Sri G.L. Jagadish, S/O Sri G.N. Lingappa, Aged About 52 Years, Residing At No.29, 3Rd Main, Basaveshwara Housing Society Layout, Vijayanagar, Near Bts Depot, Bengaluru-40....Appellant [By Sri H. Ramachandra, Advocate For Sri H.R. Anantha Krishna Murthy And Associates - (Through V.C.)] And: Smt. Vasantha Kokila, W/O Late N.R. Somashekhar, Aged About 58 Years, Residing At No.322, 8Th Main, 3Rd Stage, 4Th Block, Basaveshwaranagar, Bengaluru....Respondent [By Sri K.R. Lakshminarayana Rao, Advocate] This Criminal Appeal Is Filed Under Section 378(4) Of Cr.P.C. Praying To Set Aside The Order Dated 06.07.2010 Passed By The P.O. Ftc-Ii, Bengaluru In Crl.A. No.470/2009 And Confirming The Order Dated 27.05.2009 Passed By The Xxii Acmm And Xxiv Ascj, Bengaluru In C.C.No.17229/2004 Convicting The Respondent/Accused For The Offence Punishable Under Section 138 Of Ni Act. 2 This Criminal Appeal Having Been Heard And Reserved For Orders On 06.02.2021 This Day, The Court Pronounced The Following: Judgment' - text: The petition was filed through Sh. Vijay Pahwa, General Power of Attorney and it was asserted in the petition under Section 13-B of the Rent Act that 1 of 23 50% share of the demised premises had been purchased by the landlord from Sh. Vinod Malhotra vide sale deed No.4226 registered on 20.12.2007 with Sub Registrar, Chandigarh. - text: Mr. Arun Bharadwaj, ld. CGSC, appearing for the Union of India, has Signature Not Verified Digitally Signed By:PRATHIBA M SINGH Signing Date:09.10.2020 16:15 Digitally Signed By:SINDHU KRISHNAKUMAR Signing Date:09.10.2020 16:50:02 reiterated the submissions made by Dr. Singhvi and has further submitted that this petition ought to be heard with the OA No. 291/138/2020 pending before the CAT. model-index: - name: SpanMarker results: - task: type: token-classification name: Named Entity Recognition dataset: name: legal_ner type: unknown split: eval metrics: - type: f1 value: 0.9099756690997567 name: F1 - type: precision value: 0.9089703932832524 name: Precision - type: recall value: 0.9109831709477414 name: Recall --- # SpanMarker This is a [SpanMarker](https://github.com/tomaarsen/SpanMarkerNER) model that can be used for Named Entity Recognition. It was trained on the Legal NER Indian Justice dataset. Official repository of the model: [Github Link](https://github.com/lambdavi/SpanLuke) ## Model Details ### Model Description - Model Type: SpanMarker <!-- - Encoder: [Unknown](https://huggingface.co/unknown) --> - Maximum Sequence Length: 128 tokens - Maximum Entity Length: 6 words <!-- - Training Dataset: [Unknown](https://huggingface.co/datasets/unknown) --> <!-- - Language: Unknown --> <!-- - License: Unknown --> ### Model Sources - Repository: [SpanMarker on GitHub](https://github.com/tomaarsen/SpanMarkerNER) - Thesis: [SpanMarker For Named Entity Recognition](https://raw.githubusercontent.com/tomaarsen/SpanMarkerNER/main/thesis.pdf) \| ## Uses ### Direct Use for Inference ```python from span_marker import SpanMarkerModel from span_marker.tokenizer import SpanMarkerTokenizer # Download from the 🤗 Hub model = SpanMarkerModel.from_pretrained("lambdavi/span-marker-luke-legal") tokenizer = SpanMarkerTokenizer.from_pretrained("roberta-base", config=model.config) model.set_tokenizer(tokenizer) # Run inference entities = model.predict("The petition was filed through Sh. Vijay Pahwa, General Power of Attorney and it was asserted in the petition under Section 13-B of the Rent Act that 1 of 23 50% share of the demised premises had been purchased by the landlord from Sh. Vinod Malhotra vide sale deed No.4226 registered on 20.12.2007 with Sub Registrar, Chandigarh.") ``` ### Downstream Use You can finetune this model on your own dataset. <details><summary>Click to expand</summary> ```python from span_marker import SpanMarkerModel, Trainer from span_marker.tokenizer import SpanMarkerTokenizer # Download from the 🤗 Hub model = SpanMarkerModel.from_pretrained("lambdavi/span-marker-luke-legal") tokenizer = SpanMarkerTokenizer.from_pretrained("roberta-base", config=model.config) model.set_tokenizer(tokenizer) # Specify a Dataset with "tokens" and "ner_tag" columns dataset = load_dataset("conll2003") # For example CoNLL2003 # Initialize a Trainer using the pretrained model & dataset trainer = Trainer( model=model, train_dataset=dataset["train"], eval_dataset=dataset["validation"], ) trainer.train() trainer.save_model("lambdavi/span-marker-luke-legal-finetuned") ``` </details> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Set Metrics \| Training set \| Min \| Median \| Max \| \|:----------------------\|:----\|:--------\|:-----\| \| Sentence length \| 3 \| 44.5113 \| 2795 \| \| Entities per sentence \| 0 \| 2.7232 \| 68 \| ### Training Hyperparameters - learning_rate: 0.0001 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - gradient_accumulation_steps: 2 - total_train_batch_size: 16 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_ratio: 0.06 - num_epochs: 5 ### Training Results \| Epoch \| Step \| Validation Loss \| Validation Precision \| Validation Recall \| Validation F1 \| Validation Accuracy \| \|:------:\|:----:\|:---------------:\|:--------------------:\|:-----------------:\|:-------------:\|:-------------------:\| \| 0.9997 \| 1837 \| 0.0137 \| 0.7773 \| 0.7994 \| 0.7882 \| 0.9577 \| \| 2.0 \| 3675 \| 0.0090 \| 0.8751 \| 0.8348 \| 0.8545 \| 0.9697 \| \| 2.9997 \| 5512 \| 0.0077 \| 0.8777 \| 0.8959 \| 0.8867 \| 0.9770 \| \| 4.0 \| 7350 \| 0.0061 \| 0.8941 \| 0.9083 \| 0.9011 \| 0.9811 \| \| 4.9986 \| 9185 \| 0.0064 \| 0.9090 \| 0.9110 \| 0.9100 \| 0.9824 \| \| Metric \| Value \| \|:----------------------\|:-------\| \| f1-exact \| 0.9237 \| \| f1-strict \| 0.9100 \| \| f1-partial \| 0.9365 \| \| f1-type-match \| 0.9277 \| ### Framework Versions - Python: 3.10.12 - SpanMarker: 1.5.0 - Transformers: 4.36.0 - PyTorch: 2.0.0 - Datasets: 2.17.1 - Tokenizers: 0.15.0 ## Citation ### BibTeX ``` @software{Aarsen_SpanMarker, author = {Aarsen, Tom}, license = {Apache-2.0}, title = {{SpanMarker for Named Entity Recognition}}, url = {https://github.com/tomaarsen/SpanMarkerNER} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	null	Non_BioNLP	# SpanMarker This is a [SpanMarker](https://github.com/tomaarsen/SpanMarkerNER) model that can be used for Named Entity Recognition. It was trained on the Legal NER Indian Justice dataset. Official repository of the model: [Github Link](https://github.com/lambdavi/SpanLuke) ## Model Details ### Model Description - Model Type: SpanMarker <!-- - Encoder: [Unknown](https://huggingface.co/unknown) --> - Maximum Sequence Length: 128 tokens - Maximum Entity Length: 6 words <!-- - Training Dataset: [Unknown](https://huggingface.co/datasets/unknown) --> <!-- - Language: Unknown --> <!-- - License: Unknown --> ### Model Sources - Repository: [SpanMarker on GitHub](https://github.com/tomaarsen/SpanMarkerNER) - Thesis: [SpanMarker For Named Entity Recognition](https://raw.githubusercontent.com/tomaarsen/SpanMarkerNER/main/thesis.pdf) \| ## Uses ### Direct Use for Inference ```python from span_marker import SpanMarkerModel from span_marker.tokenizer import SpanMarkerTokenizer # Download from the 🤗 Hub model = SpanMarkerModel.from_pretrained("lambdavi/span-marker-luke-legal") tokenizer = SpanMarkerTokenizer.from_pretrained("roberta-base", config=model.config) model.set_tokenizer(tokenizer) # Run inference entities = model.predict("The petition was filed through Sh. Vijay Pahwa, General Power of Attorney and it was asserted in the petition under Section 13-B of the Rent Act that 1 of 23 50% share of the demised premises had been purchased by the landlord from Sh. Vinod Malhotra vide sale deed No.4226 registered on 20.12.2007 with Sub Registrar, Chandigarh.") ``` ### Downstream Use You can finetune this model on your own dataset. <details><summary>Click to expand</summary> ```python from span_marker import SpanMarkerModel, Trainer from span_marker.tokenizer import SpanMarkerTokenizer # Download from the 🤗 Hub model = SpanMarkerModel.from_pretrained("lambdavi/span-marker-luke-legal") tokenizer = SpanMarkerTokenizer.from_pretrained("roberta-base", config=model.config) model.set_tokenizer(tokenizer) # Specify a Dataset with "tokens" and "ner_tag" columns dataset = load_dataset("conll2003") # For example CoNLL2003 # Initialize a Trainer using the pretrained model & dataset trainer = Trainer( model=model, train_dataset=dataset["train"], eval_dataset=dataset["validation"], ) trainer.train() trainer.save_model("lambdavi/span-marker-luke-legal-finetuned") ``` </details> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Set Metrics \| Training set \| Min \| Median \| Max \| \|:----------------------\|:----\|:--------\|:-----\| \| Sentence length \| 3 \| 44.5113 \| 2795 \| \| Entities per sentence \| 0 \| 2.7232 \| 68 \| ### Training Hyperparameters - learning_rate: 0.0001 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - gradient_accumulation_steps: 2 - total_train_batch_size: 16 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_ratio: 0.06 - num_epochs: 5 ### Training Results \| Epoch \| Step \| Validation Loss \| Validation Precision \| Validation Recall \| Validation F1 \| Validation Accuracy \| \|:------:\|:----:\|:---------------:\|:--------------------:\|:-----------------:\|:-------------:\|:-------------------:\| \| 0.9997 \| 1837 \| 0.0137 \| 0.7773 \| 0.7994 \| 0.7882 \| 0.9577 \| \| 2.0 \| 3675 \| 0.0090 \| 0.8751 \| 0.8348 \| 0.8545 \| 0.9697 \| \| 2.9997 \| 5512 \| 0.0077 \| 0.8777 \| 0.8959 \| 0.8867 \| 0.9770 \| \| 4.0 \| 7350 \| 0.0061 \| 0.8941 \| 0.9083 \| 0.9011 \| 0.9811 \| \| 4.9986 \| 9185 \| 0.0064 \| 0.9090 \| 0.9110 \| 0.9100 \| 0.9824 \| \| Metric \| Value \| \|:----------------------\|:-------\| \| f1-exact \| 0.9237 \| \| f1-strict \| 0.9100 \| \| f1-partial \| 0.9365 \| \| f1-type-match \| 0.9277 \| ### Framework Versions - Python: 3.10.12 - SpanMarker: 1.5.0 - Transformers: 4.36.0 - PyTorch: 2.0.0 - Datasets: 2.17.1 - Tokenizers: 0.15.0 ## Citation ### BibTeX ``` @software{Aarsen_SpanMarker, author = {Aarsen, Tom}, license = {Apache-2.0}, title = {{SpanMarker for Named Entity Recognition}}, url = {https://github.com/tomaarsen/SpanMarkerNER} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	{"library_name": "span-marker", "metrics": ["precision", "recall", "f1"], "pipeline_tag": "token-classification", "tags": ["span-marker", "token-classification", "ner", "named-entity-recognition", "generated_from_span_marker_trainer", "legal"], "widget": [{"text": "The seven-judge Constitution Bench of the Supreme Court in SBP and Co. (supra) while reversing earlier five-judge Constitution Bench judgment in Konkan Railway Corpn. Ltd. vs. Rani Construction (P) Ltd., (2002) 2 SCC 388 held that the power exercised by the Chief Justice of the High Court or the Chief justice of India under Section 11(6) of the Arbitration Act is not an administrative power but is a judicial power."}, {"text": "In The High Court Of Judicature At Patna Criminal Writ Jurisdiction Case No.160 of 2021 Arising Out of Ps. Case No.-58 Year-2020 Thana- Bakhari District- Begusarai ====================================================== Hanif Ur Rahman, son of Azhar Rahman, Resident of C-39, East Nizamuddin, New Delhi....... Petitioner Versus 1. The State of Bihar (through Chief Secretary, Govt. of Bihar) Main Secretariat, Patna - 800015. 2. Meena Khatoon, wife of Mastan @ Noor Mohammad, Resident of Village- Mansurpur Chaksikandar, P.S.- Bidupur, District- Vaishali (Bihar) 3. The Bihar Police, through Standing Counsel. 4. Child Welfare Committee, through Chairperson, Chanakyanagar, Mahmadpur, Begusarai. 5. The Superintendent, Alpawas Grih, Nirala Nagar, Behind G.D. College, Ratanpur, Begusarai....... Respondents ====================================================== Appearance:For the Petitioner:Ms. Kriti Awasthi, Advocate Mr. Sambhav Gupta, Advocate Mr. Navnit Kumar, Advocate Mr. Shyam Kumar, Advocate For the Respondents:Mr.Nadim Seraj, G.P.5 For the Resp. No. 2:Ms. Archana Sinha, Advocate For the Resp. No. 4:Mr. Prabhu Narain Sharma, Advocate ====================================================== Coram: Honourable Mr. Justice Rajeev Ranjan Prasad C.A.V. Judgment"}, {"text": "1 R In The High Court Of Karnataka At Bengaluru Dated This The 19Th Day Of February, 2021 Before The Hon'Ble Mr. Justice H.P. Sandesh Criminal Appeal No.176/2011 Between: Sri G.L. Jagadish, S/O Sri G.N. Lingappa, Aged About 52 Years, Residing At No.29, 3Rd Main, Basaveshwara Housing Society Layout, Vijayanagar, Near Bts Depot, Bengaluru-40....Appellant [By Sri H. Ramachandra, Advocate For Sri H.R. Anantha Krishna Murthy And Associates - (Through V.C.)] And: Smt. Vasantha Kokila, W/O Late N.R. Somashekhar, Aged About 58 Years, Residing At No.322, 8Th Main, 3Rd Stage, 4Th Block, Basaveshwaranagar, Bengaluru....Respondent [By Sri K.R. Lakshminarayana Rao, Advocate] This Criminal Appeal Is Filed Under Section 378(4) Of Cr.P.C. Praying To Set Aside The Order Dated 06.07.2010 Passed By The P.O. Ftc-Ii, Bengaluru In Crl.A. No.470/2009 And Confirming The Order Dated 27.05.2009 Passed By The Xxii Acmm And Xxiv Ascj, Bengaluru In C.C.No.17229/2004 Convicting The Respondent/Accused For The Offence Punishable Under Section 138 Of Ni Act. 2 This Criminal Appeal Having Been Heard And Reserved For Orders On 06.02.2021 This Day, The Court Pronounced The Following: Judgment"}, {"text": "The petition was filed through Sh. Vijay Pahwa, General Power of Attorney and it was asserted in the petition under Section 13-B of the Rent Act that 1 of 23 50% share of the demised premises had been purchased by the landlord from Sh. Vinod Malhotra vide sale deed No.4226 registered on 20.12.2007 with Sub Registrar, Chandigarh."}, {"text": "Mr. Arun Bharadwaj, ld. CGSC, appearing for the Union of India, has Signature Not Verified Digitally Signed By:PRATHIBA M SINGH Signing Date:09.10.2020 16:15 Digitally Signed By:SINDHU KRISHNAKUMAR Signing Date:09.10.2020 16:50:02 reiterated the submissions made by Dr. Singhvi and has further submitted that this petition ought to be heard with the OA No. 291/138/2020 pending before the CAT."}], "model-index": [{"name": "SpanMarker", "results": [{"task": {"type": "token-classification", "name": "Named Entity Recognition"}, "dataset": {"name": "legal_ner", "type": "unknown", "split": "eval"}, "metrics": [{"type": "f1", "value": 0.9099756690997567, "name": "F1"}, {"type": "precision", "value": 0.9089703932832524, "name": "Precision"}, {"type": "recall", "value": 0.9109831709477414, "name": "Recall"}]}]}]}	task	[ "NAMED_ENTITY_RECOGNITION" ]	43,120
itsmeussa/AdabTranslate-Darija	itsmeussa	text2text-generation	[ "transformers", "tensorboard", "safetensors", "mbart", "text2text-generation", "generated_from_trainer", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-03-25T12:16:19Z	2024-03-27T20:46:03+00:00	82	8	--- base_model: moussaKam/arabart license: apache-2.0 metrics: - bleu tags: - generated_from_trainer model-index: - name: ArabTranslate-Darija results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # Authors - Oussama Mounajjim - Imad Zaoug - Mehdi Soufiane # AdabTranslate-Darija This model is a fine-tuned version of [moussaKam/arabart](https://huggingface.co/moussaKam/arabart) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 1.0892 - Bleu: 46.4939 - Gen Len: 9.6377 ## Model description The Darija to MSA Translator is a state-of-the-art translation model meticulously trained on a diverse dataset comprising 26,000 text pairs meticulously annotated by human annotators and augmented using GPT-4 techniques. Leveraging the datasets available on Hugging Face and employing advanced training techniques, this model achieves exceptional accuracy and fluency in translating between Darija (Moroccan Arabic) and Modern Standard Arabic (MSA). Powered by the Hugging Face Transformers library, it represents a significant advancement in natural language processing technology, making it a valuable tool for bridging language barriers and promoting linguistic diversity. ## Intended uses & limitations The Darija to MSA Translator is designed to cater to a wide range of users, including language enthusiasts, researchers, and developers working on multilingual projects. Its robust training on a diverse dataset ensures accuracy and effectiveness in various contexts. However, users should be aware of its limitations, particularly in highly specialized or domain-specific translations where additional fine-tuning may be necessary. ## Training and evaluation data The training data for the Darija to MSA Translator consists of 26,000 text pairs generated via human annotation and augmented using GPT-4 techniques. These datasets were sourced from Hugging Face, ensuring a comprehensive and diverse set of examples for training. The evaluation data was carefully selected to validate the model's performance and accuracy in real-world scenarios, ensuring its reliability and effectiveness in practical applications. ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 5.0 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Bleu \| Gen Len \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:-------:\|:-------:\| \| 2.7196 \| 0.14 \| 200 \| 1.9204 \| 28.0708 \| 9.7786 \| \| 2.212 \| 0.27 \| 400 \| 1.7376 \| 31.2914 \| 9.7633 \| \| 1.9878 \| 0.41 \| 600 \| 1.6152 \| 33.3474 \| 9.4964 \| \| 1.8387 \| 0.54 \| 800 \| 1.5276 \| 35.4738 \| 9.6621 \| \| 1.7844 \| 0.68 \| 1000 \| 1.4492 \| 37.1222 \| 9.5365 \| \| 1.7389 \| 0.81 \| 1200 \| 1.4085 \| 37.6104 \| 9.5614 \| \| 1.6553 \| 0.95 \| 1400 \| 1.3584 \| 38.8845 \| 9.7191 \| \| 1.4817 \| 1.08 \| 1600 \| 1.3305 \| 39.4105 \| 9.5849 \| \| 1.3841 \| 1.22 \| 1800 \| 1.2946 \| 40.0041 \| 9.5134 \| \| 1.329 \| 1.36 \| 2000 \| 1.2702 \| 40.4855 \| 9.5927 \| \| 1.2938 \| 1.49 \| 2200 \| 1.2410 \| 41.433 \| 9.6166 \| \| 1.2812 \| 1.63 \| 2400 \| 1.2333 \| 42.0317 \| 9.7487 \| \| 1.234 \| 1.76 \| 2600 \| 1.2066 \| 42.0791 \| 9.5668 \| \| 1.2652 \| 1.9 \| 2800 \| 1.1808 \| 42.9113 \| 9.6416 \| \| 1.1726 \| 2.03 \| 3000 \| 1.1849 \| 42.8411 \| 9.6397 \| \| 1.0367 \| 2.17 \| 3200 \| 1.1817 \| 43.2576 \| 9.6385 \| \| 1.052 \| 2.31 \| 3400 \| 1.1714 \| 43.4972 \| 9.6456 \| \| 1.0222 \| 2.44 \| 3600 \| 1.1486 \| 43.7071 \| 9.637 \| \| 0.9921 \| 2.58 \| 3800 \| 1.1437 \| 44.278 \| 9.6048 \| \| 1.053 \| 2.71 \| 4000 \| 1.1305 \| 44.8293 \| 9.6804 \| \| 1.0093 \| 2.85 \| 4200 \| 1.1247 \| 44.8092 \| 9.6187 \| \| 1.0177 \| 2.98 \| 4400 \| 1.1108 \| 45.2717 \| 9.6331 \| \| 0.8833 \| 3.12 \| 4600 \| 1.1225 \| 45.2862 \| 9.6317 \| \| 0.8604 \| 3.25 \| 4800 \| 1.1161 \| 45.2156 \| 9.625 \| \| 0.8712 \| 3.39 \| 5000 \| 1.1139 \| 45.2736 \| 9.5955 \| \| 0.865 \| 3.53 \| 5200 \| 1.1137 \| 45.7609 \| 9.6828 \| \| 0.8821 \| 3.66 \| 5400 \| 1.0981 \| 45.742 \| 9.6779 \| \| 0.8532 \| 3.8 \| 5600 \| 1.0934 \| 45.6965 \| 9.5956 \| \| 0.8515 \| 3.93 \| 5800 \| 1.0954 \| 46.0175 \| 9.6165 \| \| 0.7878 \| 4.07 \| 6000 \| 1.0941 \| 45.96 \| 9.6382 \| \| 0.7652 \| 4.2 \| 6200 \| 1.0988 \| 45.8692 \| 9.6138 \| \| 0.7841 \| 4.34 \| 6400 \| 1.0991 \| 46.1438 \| 9.6514 \| \| 0.7432 \| 4.47 \| 6600 \| 1.0961 \| 46.1105 \| 9.6212 \| \| 0.7918 \| 4.61 \| 6800 \| 1.0910 \| 46.305 \| 9.6477 \| \| 0.7638 \| 4.75 \| 7000 \| 1.0901 \| 46.4753 \| 9.6439 \| \| 0.7448 \| 4.88 \| 7200 \| 1.0892 \| 46.4939 \| 9.6377 \| # How to use it ? Just copy and paste this code after installing the necessary libraries from transformers import AutoModelForSeq2SeqLM, AutoTokenizer model_path = 'itsmeussa/AdabTranslate-Darija' model = AutoModelForSeq2SeqLM.from_pretrained(model_path) tokenizer = AutoTokenizer.from_pretrained('moussaKam/arabart') seq = "مرحبا بيكم" tok = tokenizer.encode(seq, return_tensors='pt') res = model.generate(tok) tokenizer.decode(res[0]) ### Framework versions - Transformers 4.40.0.dev0 - Pytorch 2.2.1+cu121 - Datasets 2.18.0 - Tokenizers 0.15.2	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # Authors - Oussama Mounajjim - Imad Zaoug - Mehdi Soufiane # AdabTranslate-Darija This model is a fine-tuned version of [moussaKam/arabart](https://huggingface.co/moussaKam/arabart) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 1.0892 - Bleu: 46.4939 - Gen Len: 9.6377 ## Model description The Darija to MSA Translator is a state-of-the-art translation model meticulously trained on a diverse dataset comprising 26,000 text pairs meticulously annotated by human annotators and augmented using GPT-4 techniques. Leveraging the datasets available on Hugging Face and employing advanced training techniques, this model achieves exceptional accuracy and fluency in translating between Darija (Moroccan Arabic) and Modern Standard Arabic (MSA). Powered by the Hugging Face Transformers library, it represents a significant advancement in natural language processing technology, making it a valuable tool for bridging language barriers and promoting linguistic diversity. ## Intended uses & limitations The Darija to MSA Translator is designed to cater to a wide range of users, including language enthusiasts, researchers, and developers working on multilingual projects. Its robust training on a diverse dataset ensures accuracy and effectiveness in various contexts. However, users should be aware of its limitations, particularly in highly specialized or domain-specific translations where additional fine-tuning may be necessary. ## Training and evaluation data The training data for the Darija to MSA Translator consists of 26,000 text pairs generated via human annotation and augmented using GPT-4 techniques. These datasets were sourced from Hugging Face, ensuring a comprehensive and diverse set of examples for training. The evaluation data was carefully selected to validate the model's performance and accuracy in real-world scenarios, ensuring its reliability and effectiveness in practical applications. ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 5.0 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Bleu \| Gen Len \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:-------:\|:-------:\| \| 2.7196 \| 0.14 \| 200 \| 1.9204 \| 28.0708 \| 9.7786 \| \| 2.212 \| 0.27 \| 400 \| 1.7376 \| 31.2914 \| 9.7633 \| \| 1.9878 \| 0.41 \| 600 \| 1.6152 \| 33.3474 \| 9.4964 \| \| 1.8387 \| 0.54 \| 800 \| 1.5276 \| 35.4738 \| 9.6621 \| \| 1.7844 \| 0.68 \| 1000 \| 1.4492 \| 37.1222 \| 9.5365 \| \| 1.7389 \| 0.81 \| 1200 \| 1.4085 \| 37.6104 \| 9.5614 \| \| 1.6553 \| 0.95 \| 1400 \| 1.3584 \| 38.8845 \| 9.7191 \| \| 1.4817 \| 1.08 \| 1600 \| 1.3305 \| 39.4105 \| 9.5849 \| \| 1.3841 \| 1.22 \| 1800 \| 1.2946 \| 40.0041 \| 9.5134 \| \| 1.329 \| 1.36 \| 2000 \| 1.2702 \| 40.4855 \| 9.5927 \| \| 1.2938 \| 1.49 \| 2200 \| 1.2410 \| 41.433 \| 9.6166 \| \| 1.2812 \| 1.63 \| 2400 \| 1.2333 \| 42.0317 \| 9.7487 \| \| 1.234 \| 1.76 \| 2600 \| 1.2066 \| 42.0791 \| 9.5668 \| \| 1.2652 \| 1.9 \| 2800 \| 1.1808 \| 42.9113 \| 9.6416 \| \| 1.1726 \| 2.03 \| 3000 \| 1.1849 \| 42.8411 \| 9.6397 \| \| 1.0367 \| 2.17 \| 3200 \| 1.1817 \| 43.2576 \| 9.6385 \| \| 1.052 \| 2.31 \| 3400 \| 1.1714 \| 43.4972 \| 9.6456 \| \| 1.0222 \| 2.44 \| 3600 \| 1.1486 \| 43.7071 \| 9.637 \| \| 0.9921 \| 2.58 \| 3800 \| 1.1437 \| 44.278 \| 9.6048 \| \| 1.053 \| 2.71 \| 4000 \| 1.1305 \| 44.8293 \| 9.6804 \| \| 1.0093 \| 2.85 \| 4200 \| 1.1247 \| 44.8092 \| 9.6187 \| \| 1.0177 \| 2.98 \| 4400 \| 1.1108 \| 45.2717 \| 9.6331 \| \| 0.8833 \| 3.12 \| 4600 \| 1.1225 \| 45.2862 \| 9.6317 \| \| 0.8604 \| 3.25 \| 4800 \| 1.1161 \| 45.2156 \| 9.625 \| \| 0.8712 \| 3.39 \| 5000 \| 1.1139 \| 45.2736 \| 9.5955 \| \| 0.865 \| 3.53 \| 5200 \| 1.1137 \| 45.7609 \| 9.6828 \| \| 0.8821 \| 3.66 \| 5400 \| 1.0981 \| 45.742 \| 9.6779 \| \| 0.8532 \| 3.8 \| 5600 \| 1.0934 \| 45.6965 \| 9.5956 \| \| 0.8515 \| 3.93 \| 5800 \| 1.0954 \| 46.0175 \| 9.6165 \| \| 0.7878 \| 4.07 \| 6000 \| 1.0941 \| 45.96 \| 9.6382 \| \| 0.7652 \| 4.2 \| 6200 \| 1.0988 \| 45.8692 \| 9.6138 \| \| 0.7841 \| 4.34 \| 6400 \| 1.0991 \| 46.1438 \| 9.6514 \| \| 0.7432 \| 4.47 \| 6600 \| 1.0961 \| 46.1105 \| 9.6212 \| \| 0.7918 \| 4.61 \| 6800 \| 1.0910 \| 46.305 \| 9.6477 \| \| 0.7638 \| 4.75 \| 7000 \| 1.0901 \| 46.4753 \| 9.6439 \| \| 0.7448 \| 4.88 \| 7200 \| 1.0892 \| 46.4939 \| 9.6377 \| # How to use it ? Just copy and paste this code after installing the necessary libraries from transformers import AutoModelForSeq2SeqLM, AutoTokenizer model_path = 'itsmeussa/AdabTranslate-Darija' model = AutoModelForSeq2SeqLM.from_pretrained(model_path) tokenizer = AutoTokenizer.from_pretrained('moussaKam/arabart') seq = "مرحبا بيكم" tok = tokenizer.encode(seq, return_tensors='pt') res = model.generate(tok) tokenizer.decode(res[0]) ### Framework versions - Transformers 4.40.0.dev0 - Pytorch 2.2.1+cu121 - Datasets 2.18.0 - Tokenizers 0.15.2	{"base_model": "moussaKam/arabart", "license": "apache-2.0", "metrics": ["bleu"], "tags": ["generated_from_trainer"], "model-index": [{"name": "ArabTranslate-Darija", "results": []}]}	task	[ "TRANSLATION" ]	43,121
DiegoPolimeni/distilbert-base-uncased-distilled-clinc	DiegoPolimeni	text-classification	[ "transformers", "pytorch", "distilbert", "text-classification", "generated_from_trainer", "dataset:clinc_oos", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-03-10T14:49:33Z	2023-03-10T15:02:48+00:00	12	0	--- datasets: - clinc_oos license: apache-2.0 metrics: - accuracy tags: - generated_from_trainer model-index: - name: distilbert-base-uncased-distilled-clinc results: - task: type: text-classification name: Text Classification dataset: name: clinc_oos type: clinc_oos config: plus split: validation args: plus metrics: - type: accuracy value: 0.944516129032258 name: Accuracy --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert-base-uncased-distilled-clinc This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the clinc_oos dataset. It achieves the following results on the evaluation set: - Loss: 0.3040 - Accuracy: 0.9445 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 48 - eval_batch_size: 48 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 8 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\| \| No log \| 1.0 \| 318 \| 2.1904 \| 0.7545 \| \| 2.5592 \| 2.0 \| 636 \| 1.1694 \| 0.8632 \| \| 2.5592 \| 3.0 \| 954 \| 0.6723 \| 0.9165 \| \| 1.038 \| 4.0 \| 1272 \| 0.4640 \| 0.9316 \| \| 0.4699 \| 5.0 \| 1590 \| 0.3723 \| 0.9406 \| \| 0.4699 \| 6.0 \| 1908 \| 0.3293 \| 0.9445 \| \| 0.2942 \| 7.0 \| 2226 \| 0.3086 \| 0.9455 \| \| 0.2329 \| 8.0 \| 2544 \| 0.3040 \| 0.9445 \| ### Framework versions - Transformers 4.26.1 - Pytorch 1.13.1+cu116 - Datasets 2.10.1 - Tokenizers 0.13.2	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert-base-uncased-distilled-clinc This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the clinc_oos dataset. It achieves the following results on the evaluation set: - Loss: 0.3040 - Accuracy: 0.9445 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 48 - eval_batch_size: 48 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 8 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\| \| No log \| 1.0 \| 318 \| 2.1904 \| 0.7545 \| \| 2.5592 \| 2.0 \| 636 \| 1.1694 \| 0.8632 \| \| 2.5592 \| 3.0 \| 954 \| 0.6723 \| 0.9165 \| \| 1.038 \| 4.0 \| 1272 \| 0.4640 \| 0.9316 \| \| 0.4699 \| 5.0 \| 1590 \| 0.3723 \| 0.9406 \| \| 0.4699 \| 6.0 \| 1908 \| 0.3293 \| 0.9445 \| \| 0.2942 \| 7.0 \| 2226 \| 0.3086 \| 0.9455 \| \| 0.2329 \| 8.0 \| 2544 \| 0.3040 \| 0.9445 \| ### Framework versions - Transformers 4.26.1 - Pytorch 1.13.1+cu116 - Datasets 2.10.1 - Tokenizers 0.13.2	{"datasets": ["clinc_oos"], "license": "apache-2.0", "metrics": ["accuracy"], "tags": ["generated_from_trainer"], "model-index": [{"name": "distilbert-base-uncased-distilled-clinc", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "clinc_oos", "type": "clinc_oos", "config": "plus", "split": "validation", "args": "plus"}, "metrics": [{"type": "accuracy", "value": 0.944516129032258, "name": "Accuracy"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,122
mingkom/distilbert-base-uncased-finetuned-emotion	mingkom	text-classification	[ "transformers", "pytorch", "distilbert", "text-classification", "generated_from_trainer", "dataset:emotion", "base_model:distilbert/distilbert-base-uncased", "base_model:finetune:distilbert/distilbert-base-uncased", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-09-01T02:45:20Z	2023-09-01T03:09:22+00:00	11	0	--- base_model: distilbert-base-uncased datasets: - emotion license: apache-2.0 metrics: - accuracy - f1 tags: - generated_from_trainer model-index: - name: distilbert-base-uncased-finetuned-emotion results: - task: type: text-classification name: Text Classification dataset: name: emotion type: emotion config: split split: validation args: split metrics: - type: accuracy value: 0.9255 name: Accuracy - type: f1 value: 0.925605036699702 name: F1 --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert-base-uncased-finetuned-emotion This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset. It achieves the following results on the evaluation set: - Loss: 0.2157 - Accuracy: 0.9255 - F1: 0.9256 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 64 - eval_batch_size: 64 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 2 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\|:------:\| \| 0.8152 \| 1.0 \| 250 \| 0.3179 \| 0.908 \| 0.9057 \| \| 0.2525 \| 2.0 \| 500 \| 0.2157 \| 0.9255 \| 0.9256 \| ### Framework versions - Transformers 4.32.1 - Pytorch 2.0.1+cu118 - Datasets 2.14.4 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # distilbert-base-uncased-finetuned-emotion This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset. It achieves the following results on the evaluation set: - Loss: 0.2157 - Accuracy: 0.9255 - F1: 0.9256 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 64 - eval_batch_size: 64 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 2 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\|:------:\| \| 0.8152 \| 1.0 \| 250 \| 0.3179 \| 0.908 \| 0.9057 \| \| 0.2525 \| 2.0 \| 500 \| 0.2157 \| 0.9255 \| 0.9256 \| ### Framework versions - Transformers 4.32.1 - Pytorch 2.0.1+cu118 - Datasets 2.14.4 - Tokenizers 0.13.3	{"base_model": "distilbert-base-uncased", "datasets": ["emotion"], "license": "apache-2.0", "metrics": ["accuracy", "f1"], "tags": ["generated_from_trainer"], "model-index": [{"name": "distilbert-base-uncased-finetuned-emotion", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "emotion", "type": "emotion", "config": "split", "split": "validation", "args": "split"}, "metrics": [{"type": "accuracy", "value": 0.9255, "name": "Accuracy"}, {"type": "f1", "value": 0.925605036699702, "name": "F1"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,123
ValkyriaLenneth/longformer_zh	ValkyriaLenneth	feature-extraction	[ "transformers", "pytorch", "longformer", "feature-extraction", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:05Z	2022-01-06T03:50:20+00:00	11,518	23	--- {} --- # 中文预训练Longformer模型 \| Longformer_ZH with PyTorch 相比于Transformer的O(n^2)复杂度，Longformer提供了一种以线性复杂度处理最长4K字符级别文档序列的方法。Longformer Attention包括了标准的自注意力与全局注意力机制，方便模型更好地学习超长序列的信息。 Compared with O(n^2) complexity for Transformer model, Longformer provides an efficient method for processing long-document level sequence in Linear complexity. Longformer’s attention mechanism is a drop-in replacement for the standard self-attention and combines a local windowed attention with a task motivated global attention. 我们注意到关于中文Longformer或超长序列任务的资源较少，因此在此开源了我们预训练的中文Longformer模型参数，并提供了相应的加载方法，以及预训练脚本。 There are not so much resource for Chinese Longformer or long-sequence-level chinese task. Thus we open source our pretrained longformer model to help the researchers. ## 加载模型 \| Load the model 您可以使用谷歌云盘或百度网盘下载我们的模型 You could get Longformer_zh from Google Drive or Baidu Yun. - Google Drive: https://drive.google.com/file/d/1IDJ4aVTfSFUQLIqCYBtoRpnfbgHPoxB4/view?usp=sharing - 百度云: 链接：https://pan.baidu.com/s/1HaVDENx52I7ryPFpnQmq1w 提取码：y601 我们同样提供了Huggingface的自动下载 We also provide auto load with HuggingFace.Transformers. ``` from Longformer_zh import LongformerZhForMaksedLM LongformerZhForMaksedLM.from_pretrained('ValkyriaLenneth/longformer_zh') ``` ## 注意事项 \| Notice - 直接使用 `transformers.LongformerModel.from_pretrained` 加载模型 - Please use `transformers.LongformerModel.from_pretrained` to load the model directly - 以下内容已经被弃用 - The following notices are abondoned, please ignore them. - 区别于英文原版Longformer，中文Longformer的基础是Roberta_zh模型，其本质上属于 `Transformers.BertModel` 而非 `RobertaModel`, 因此无法使用原版代码直接加载。 - Different with origin English Longformer, Longformer_Zh is based on Roberta_zh which is a subclass of `Transformers.BertModel` not `RobertaModel`. Thus it is impossible to load it with origin code. - 我们提供了修改后的中文Longformer文件，您可以使用其加载参数。 - We provide modified Longformer_zh class, you can use it directly to load the model. - 如果您想将此参数用于更多任务，请参考`Longformer_zh.py`替换Attention Layer. - If you want to use our model on more down-stream tasks, please refer to `Longformer_zh.py` and replace Attention layer with Longformer Attention layer. ## 关于预训练 \| About Pretraining - 我们的预训练语料来自 https://github.com/brightmart/nlp_chinese_corpus，根据Longformer原文的设置，采用了多种语料混合的预训练数据。 - The corpus of pretraining is from https://github.com/brightmart/nlp_chinese_corpus. Based on the paper of Longformer, we use a mixture of 4 different chinese corpus for pretraining. - 我们的模型是基于Roberta_zh_mid (https://github.com/brightmart/roberta_zh),训练脚本参考了https://github.com/allenai/longformer/blob/master/scripts/convert_model_to_long.ipynb - The basement of our model is Roberta_zh_mid (https://github.com/brightmart/roberta_zh). Pretraining scripts is modified from https://github.com/allenai/longformer/blob/master/scripts/convert_model_to_long.ipynb. - 同时我们在原版基础上，引入了 `Whole-Word-Masking` 机制，以便更好地适应中文特性。 - We introduce `Whole-Word-Masking` method into pretraining for better fitting Chinese language. - `Whole-Word-Masking`代码改写自TensorFlow版本的Roberta_zh，据我们所知是第一个开源的Pytorch版本WWM. - Our WWM scripts is refacted from Roberta_zh_Tensorflow, as far as we know, it is the first open source Whole-word-masking scripts in Pytorch. - 模型 `max_seq_length = 4096`, 在 4 * Titan RTX 上预训练3K steps 大概用时4天。 - Max seuence length is 4096 and the pretraining took 4 days on 4 * Titan RTX. - 我们使用了 `Nvidia.Apex` 引入了混合精度训练，以加速预训练。 - We use `Nvidia.Apex` to accelerate pretraining. - 关于数据预处理，我们采用 `Jieba` 分词与`JIONLP`进行数据清洗。 - We use `Jieba` Chinese tokenizer and `JIONLP` data cleaning. - 更多细节可以参考我们的预训练脚本 - For more details, please check our pretraining scripts. ## 效果测试 \| Evaluation ### CCF Sentiment Analysis - 由于中文超长文本级别任务稀缺，我们采用了CCF-Sentiment-Analysis任务进行测试 - Since it is hard to acquire open-sourced long sequence level chinese NLP task, we use CCF-Sentiment-Analysis for evaluation. \|Model\|Dev F\| \|----\|----\| \|Bert\|80.3\| \|Bert-wwm-ext\| 80.5\| \|Roberta-mid\|80.5\| \|Roberta-large\|81.25\| \|Longformer_SC\|79.37\| \|Longformer_ZH\|80.51\| ### Pretraining BPC - 我们提供了预训练BPC(bits-per-character), BPC越小，代表语言模型性能更优。可视作PPL. - We also provide BPC scores of pretraining, the lower BPC score, the better performance Langugage Model has. You can also treat it as PPL. \|Model\|BPC\| \|---\|---\| \|Longformer before training\| 14.78\| \|Longformer after training\| 3.10\| ### CMRC(Chinese Machine Reading Comprehension) \|Model\|F1\|EM\| \|---\|---\|---\| \|Bert\|85.87\|64.90\| \|Roberta\|86.45\|66.57\| \|Longformer_zh\|86.15\|66.84\| ### Chinese Coreference Resolution \|Model\|Conll-F1\|Precision\|Recall\| \|---\|---\|---\|---\| \|Bert\|66.82\|70.30\|63.67\| \|Roberta\|67.77\|69.28\|66.32\| \|Longformer_zh\|67.81\|70.13\|65.64\| ## 致谢感谢东京工业大学奥村·船越研究室提供算力。 Thanks Okumula·Funakoshi Lab from Tokyo Institute of Technology who provides the devices and oppotunity for me to finish this project.	null	Non_BioNLP	# 中文预训练Longformer模型 \| Longformer_ZH with PyTorch 相比于Transformer的O(n^2)复杂度，Longformer提供了一种以线性复杂度处理最长4K字符级别文档序列的方法。Longformer Attention包括了标准的自注意力与全局注意力机制，方便模型更好地学习超长序列的信息。 Compared with O(n^2) complexity for Transformer model, Longformer provides an efficient method for processing long-document level sequence in Linear complexity. Longformer’s attention mechanism is a drop-in replacement for the standard self-attention and combines a local windowed attention with a task motivated global attention. 我们注意到关于中文Longformer或超长序列任务的资源较少，因此在此开源了我们预训练的中文Longformer模型参数，并提供了相应的加载方法，以及预训练脚本。 There are not so much resource for Chinese Longformer or long-sequence-level chinese task. Thus we open source our pretrained longformer model to help the researchers. ## 加载模型 \| Load the model 您可以使用谷歌云盘或百度网盘下载我们的模型 You could get Longformer_zh from Google Drive or Baidu Yun. - Google Drive: https://drive.google.com/file/d/1IDJ4aVTfSFUQLIqCYBtoRpnfbgHPoxB4/view?usp=sharing - 百度云: 链接：https://pan.baidu.com/s/1HaVDENx52I7ryPFpnQmq1w 提取码：y601 我们同样提供了Huggingface的自动下载 We also provide auto load with HuggingFace.Transformers. ``` from Longformer_zh import LongformerZhForMaksedLM LongformerZhForMaksedLM.from_pretrained('ValkyriaLenneth/longformer_zh') ``` ## 注意事项 \| Notice - 直接使用 `transformers.LongformerModel.from_pretrained` 加载模型 - Please use `transformers.LongformerModel.from_pretrained` to load the model directly - 以下内容已经被弃用 - The following notices are abondoned, please ignore them. - 区别于英文原版Longformer，中文Longformer的基础是Roberta_zh模型，其本质上属于 `Transformers.BertModel` 而非 `RobertaModel`, 因此无法使用原版代码直接加载。 - Different with origin English Longformer, Longformer_Zh is based on Roberta_zh which is a subclass of `Transformers.BertModel` not `RobertaModel`. Thus it is impossible to load it with origin code. - 我们提供了修改后的中文Longformer文件，您可以使用其加载参数。 - We provide modified Longformer_zh class, you can use it directly to load the model. - 如果您想将此参数用于更多任务，请参考`Longformer_zh.py`替换Attention Layer. - If you want to use our model on more down-stream tasks, please refer to `Longformer_zh.py` and replace Attention layer with Longformer Attention layer. ## 关于预训练 \| About Pretraining - 我们的预训练语料来自 https://github.com/brightmart/nlp_chinese_corpus，根据Longformer原文的设置，采用了多种语料混合的预训练数据。 - The corpus of pretraining is from https://github.com/brightmart/nlp_chinese_corpus. Based on the paper of Longformer, we use a mixture of 4 different chinese corpus for pretraining. - 我们的模型是基于Roberta_zh_mid (https://github.com/brightmart/roberta_zh),训练脚本参考了https://github.com/allenai/longformer/blob/master/scripts/convert_model_to_long.ipynb - The basement of our model is Roberta_zh_mid (https://github.com/brightmart/roberta_zh). Pretraining scripts is modified from https://github.com/allenai/longformer/blob/master/scripts/convert_model_to_long.ipynb. - 同时我们在原版基础上，引入了 `Whole-Word-Masking` 机制，以便更好地适应中文特性。 - We introduce `Whole-Word-Masking` method into pretraining for better fitting Chinese language. - `Whole-Word-Masking`代码改写自TensorFlow版本的Roberta_zh，据我们所知是第一个开源的Pytorch版本WWM. - Our WWM scripts is refacted from Roberta_zh_Tensorflow, as far as we know, it is the first open source Whole-word-masking scripts in Pytorch. - 模型 `max_seq_length = 4096`, 在 4 * Titan RTX 上预训练3K steps 大概用时4天。 - Max seuence length is 4096 and the pretraining took 4 days on 4 * Titan RTX. - 我们使用了 `Nvidia.Apex` 引入了混合精度训练，以加速预训练。 - We use `Nvidia.Apex` to accelerate pretraining. - 关于数据预处理，我们采用 `Jieba` 分词与`JIONLP`进行数据清洗。 - We use `Jieba` Chinese tokenizer and `JIONLP` data cleaning. - 更多细节可以参考我们的预训练脚本 - For more details, please check our pretraining scripts. ## 效果测试 \| Evaluation ### CCF Sentiment Analysis - 由于中文超长文本级别任务稀缺，我们采用了CCF-Sentiment-Analysis任务进行测试 - Since it is hard to acquire open-sourced long sequence level chinese NLP task, we use CCF-Sentiment-Analysis for evaluation. \|Model\|Dev F\| \|----\|----\| \|Bert\|80.3\| \|Bert-wwm-ext\| 80.5\| \|Roberta-mid\|80.5\| \|Roberta-large\|81.25\| \|Longformer_SC\|79.37\| \|Longformer_ZH\|80.51\| ### Pretraining BPC - 我们提供了预训练BPC(bits-per-character), BPC越小，代表语言模型性能更优。可视作PPL. - We also provide BPC scores of pretraining, the lower BPC score, the better performance Langugage Model has. You can also treat it as PPL. \|Model\|BPC\| \|---\|---\| \|Longformer before training\| 14.78\| \|Longformer after training\| 3.10\| ### CMRC(Chinese Machine Reading Comprehension) \|Model\|F1\|EM\| \|---\|---\|---\| \|Bert\|85.87\|64.90\| \|Roberta\|86.45\|66.57\| \|Longformer_zh\|86.15\|66.84\| ### Chinese Coreference Resolution \|Model\|Conll-F1\|Precision\|Recall\| \|---\|---\|---\|---\| \|Bert\|66.82\|70.30\|63.67\| \|Roberta\|67.77\|69.28\|66.32\| \|Longformer_zh\|67.81\|70.13\|65.64\| ## 致谢感谢东京工业大学奥村·船越研究室提供算力。 Thanks Okumula·Funakoshi Lab from Tokyo Institute of Technology who provides the devices and oppotunity for me to finish this project.	{}	task	[ "COREFERENCE_RESOLUTION" ]	43,124
gmurro/bart-large-finetuned-filtered-spotify-podcast-summ	gmurro	text2text-generation	[ "transformers", "tf", "bart", "text2text-generation", "generated_from_keras_callback", "arxiv:2004.04270", "base_model:facebook/bart-large-cnn", "base_model:finetune:facebook/bart-large-cnn", "license:mit", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-06-16T16:04:16Z	2023-09-11T16:26:07+00:00	683	11	--- base_model: facebook/bart-large-cnn license: mit tags: - generated_from_keras_callback model-index: - name: bart-large-finetuned-filtered-spotify-podcast-summ results: [] --- # bart-large-finetuned-filtered-spotify-podcast-summ This model is a fine-tuned version of [facebook/bart-large-cnn](https://huggingface.co/facebook/bart-large-cnn) on on the [Spotify Podcast Dataset](https://arxiv.org/abs/2004.04270). Take a look to the [github repository](https://github.com/TheOnesThatWereAbroad/PodcastSummarization) of this project. It achieves the following results during training: - Train Loss: 2.2967 - Validation Loss: 2.8316 - Epoch: 2 ## Intended uses & limitations This model is intended to be used for automatic podcast summarisation. Given the podcast transcript in input, the objective is to provide a short text summary that a user might read when deciding whether to listen to a podcast. The summary should accurately convey the content of the podcast, be human-readable, and be short enough to be quickly read on a smartphone screen. ## Training and evaluation data In our solution, an extractive module is developed to select salient chunks from the transcript, which serve as the input to an abstractive summarizer. An extensive pre-processing on the creator-provided descriptions is performed selecting a subset of the corpus that is suitable for the training supervised model. We split the filtered dataset into train/dev sets of 69,336/7,705 episodes. The test set consists of 1,027 episodes. Only 1025 have been used because two of them did not contain an episode description. ## How to use The model can be used for the summarization as follows: ```python from transformers import pipeline summarizer = pipeline("summarization", model="gmurro/bart-large-finetuned-filtered-spotify-podcast-summ", tokenizer="gmurro/bart-large-finetuned-filtered-spotify-podcast-summ") summary = summarizer(podcast_transcript, min_length=39, max_length=250) print(summary[0]['summary_text']) ``` ### Training hyperparameters The following hyperparameters were used during training: - ```python optimizer: {'name': 'AdamWeightDecay', 'learning_rate': 2e-05, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weight_decay_rate': 0.01} ``` - ```python training_precision: float32 ``` ### Training results \| Train Loss \| Validation Loss \| Epoch \| \|:----------:\|:---------------:\|:-----:\| \| 3.0440 \| 2.8733 \| 0 \| \| 2.6085 \| 2.8549 \| 1 \| \| 2.2967 \| 2.8316 \| 2 \| ### Framework versions - Transformers 4.19.4 - TensorFlow 2.9.1 - Datasets 2.3.1 - Tokenizers 0.12.1 ## Authors \| Name \| Surname \| Email \| Username \| \| :-------: \| :-------: \| :------------------------------------: \| :---------------------------------------------------: \| \| Giuseppe \| Boezio \| `[email protected]` \| [_giuseppeboezio_](https://github.com/giuseppeboezio) \| \| Simone \| Montali \| `[email protected]` \| [_montali_](https://github.com/montali) \| \| Giuseppe \| Murro \| `[email protected]` \| [_gmurro_](https://github.com/gmurro) \|	null	Non_BioNLP	# bart-large-finetuned-filtered-spotify-podcast-summ This model is a fine-tuned version of [facebook/bart-large-cnn](https://huggingface.co/facebook/bart-large-cnn) on on the [Spotify Podcast Dataset](https://arxiv.org/abs/2004.04270). Take a look to the [github repository](https://github.com/TheOnesThatWereAbroad/PodcastSummarization) of this project. It achieves the following results during training: - Train Loss: 2.2967 - Validation Loss: 2.8316 - Epoch: 2 ## Intended uses & limitations This model is intended to be used for automatic podcast summarisation. Given the podcast transcript in input, the objective is to provide a short text summary that a user might read when deciding whether to listen to a podcast. The summary should accurately convey the content of the podcast, be human-readable, and be short enough to be quickly read on a smartphone screen. ## Training and evaluation data In our solution, an extractive module is developed to select salient chunks from the transcript, which serve as the input to an abstractive summarizer. An extensive pre-processing on the creator-provided descriptions is performed selecting a subset of the corpus that is suitable for the training supervised model. We split the filtered dataset into train/dev sets of 69,336/7,705 episodes. The test set consists of 1,027 episodes. Only 1025 have been used because two of them did not contain an episode description. ## How to use The model can be used for the summarization as follows: ```python from transformers import pipeline summarizer = pipeline("summarization", model="gmurro/bart-large-finetuned-filtered-spotify-podcast-summ", tokenizer="gmurro/bart-large-finetuned-filtered-spotify-podcast-summ") summary = summarizer(podcast_transcript, min_length=39, max_length=250) print(summary[0]['summary_text']) ``` ### Training hyperparameters The following hyperparameters were used during training: - ```python optimizer: {'name': 'AdamWeightDecay', 'learning_rate': 2e-05, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weight_decay_rate': 0.01} ``` - ```python training_precision: float32 ``` ### Training results \| Train Loss \| Validation Loss \| Epoch \| \|:----------:\|:---------------:\|:-----:\| \| 3.0440 \| 2.8733 \| 0 \| \| 2.6085 \| 2.8549 \| 1 \| \| 2.2967 \| 2.8316 \| 2 \| ### Framework versions - Transformers 4.19.4 - TensorFlow 2.9.1 - Datasets 2.3.1 - Tokenizers 0.12.1 ## Authors \| Name \| Surname \| Email \| Username \| \| :-------: \| :-------: \| :------------------------------------: \| :---------------------------------------------------: \| \| Giuseppe \| Boezio \| `[email protected]` \| [_giuseppeboezio_](https://github.com/giuseppeboezio) \| \| Simone \| Montali \| `[email protected]` \| [_montali_](https://github.com/montali) \| \| Giuseppe \| Murro \| `[email protected]` \| [_gmurro_](https://github.com/gmurro) \|	{"base_model": "facebook/bart-large-cnn", "license": "mit", "tags": ["generated_from_keras_callback"], "model-index": [{"name": "bart-large-finetuned-filtered-spotify-podcast-summ", "results": []}]}	task	[ "SUMMARIZATION" ]	43,125
datnth1709/finetuned_HelsinkiNLP-opus-mt-vi-en_PhoMT	datnth1709	translation	[ "transformers", "pytorch", "marian", "text2text-generation", "translation", "generated_from_trainer", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-09-17T09:08:26Z	2022-09-18T06:23:37+00:00	20	1	--- license: apache-2.0 metrics: - bleu tags: - translation - generated_from_trainer model-index: - name: finetuned_HelsinkiNLP-opus-mt-vi-en_PhoMT results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # finetuned_HelsinkiNLP-opus-mt-vi-en_PhoMT This model is a fine-tuned version of [Helsinki-NLP/opus-mt-vi-en](https://huggingface.co/Helsinki-NLP/opus-mt-vi-en) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 1.1783 - Bleu: 37.7741 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 5 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Bleu \| \|:-------------:\|:-----:\|:------:\|:---------------:\|:-------:\| \| 1.3717 \| 1.0 \| 186125 \| 1.2371 \| 35.8549 \| \| 1.2926 \| 2.0 \| 372250 \| 1.2113 \| 36.7328 \| \| 1.2505 \| 3.0 \| 558375 \| 1.1954 \| 37.0998 \| \| 1.2025 \| 4.0 \| 744500 \| 1.1847 \| 37.5538 \| \| 1.1853 \| 5.0 \| 930625 \| 1.1783 \| 37.7761 \| ### Framework versions - Transformers 4.22.1 - Pytorch 1.10.1+cu113 - Datasets 2.4.0 - Tokenizers 0.12.1	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # finetuned_HelsinkiNLP-opus-mt-vi-en_PhoMT This model is a fine-tuned version of [Helsinki-NLP/opus-mt-vi-en](https://huggingface.co/Helsinki-NLP/opus-mt-vi-en) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 1.1783 - Bleu: 37.7741 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 5 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Bleu \| \|:-------------:\|:-----:\|:------:\|:---------------:\|:-------:\| \| 1.3717 \| 1.0 \| 186125 \| 1.2371 \| 35.8549 \| \| 1.2926 \| 2.0 \| 372250 \| 1.2113 \| 36.7328 \| \| 1.2505 \| 3.0 \| 558375 \| 1.1954 \| 37.0998 \| \| 1.2025 \| 4.0 \| 744500 \| 1.1847 \| 37.5538 \| \| 1.1853 \| 5.0 \| 930625 \| 1.1783 \| 37.7761 \| ### Framework versions - Transformers 4.22.1 - Pytorch 1.10.1+cu113 - Datasets 2.4.0 - Tokenizers 0.12.1	{"license": "apache-2.0", "metrics": ["bleu"], "tags": ["translation", "generated_from_trainer"], "model-index": [{"name": "finetuned_HelsinkiNLP-opus-mt-vi-en_PhoMT", "results": []}]}	task	[ "TRANSLATION" ]	43,126
EXP442/nllb_translator_pretrained	EXP442	null	[ "safetensors", "m2m_100", "region:us" ]	2024-11-12T10:23:23Z	2024-12-06T11:49:10+00:00	4	1	--- {} --- Citation If you find our work helpful, feel free to give us a cite. @misc{VocaLabs-foreignnews-2024-Capstone, title = {Vocalabs Model: Foreign News Summarization and Translation}, url = { https://github.com/VocaLabs-Capstone/VocaLabs_Final }, author = {Vocalabs, Voctree}, month = {November}, year = {2024} } --- license: apache-2.0 ---	null	Non_BioNLP	Citation If you find our work helpful, feel free to give us a cite. @misc{VocaLabs-foreignnews-2024-Capstone, title = {Vocalabs Model: Foreign News Summarization and Translation}, url = { https://github.com/VocaLabs-Capstone/VocaLabs_Final }, author = {Vocalabs, Voctree}, month = {November}, year = {2024} } --- license: apache-2.0 ---	{}	task	[ "TRANSLATION", "SUMMARIZATION" ]	43,127
Nextcloud-AI/opus-mt-tr-sv	Nextcloud-AI	translation	[ "transformers", "pytorch", "tf", "marian", "text2text-generation", "translation", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-02-23T10:47:56Z	2023-08-16T12:07:29+00:00	20	0	--- license: apache-2.0 tags: - translation --- ### opus-mt-tr-sv * source languages: tr * target languages: sv * OPUS readme: [tr-sv](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/tr-sv/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-16.zip](https://object.pouta.csc.fi/OPUS-MT-models/tr-sv/opus-2020-01-16.zip) * test set translations: [opus-2020-01-16.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/tr-sv/opus-2020-01-16.test.txt) * test set scores: [opus-2020-01-16.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/tr-sv/opus-2020-01-16.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.tr.sv \| 26.3 \| 0.478 \|	null	Non_BioNLP	### opus-mt-tr-sv * source languages: tr * target languages: sv * OPUS readme: [tr-sv](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/tr-sv/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-16.zip](https://object.pouta.csc.fi/OPUS-MT-models/tr-sv/opus-2020-01-16.zip) * test set translations: [opus-2020-01-16.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/tr-sv/opus-2020-01-16.test.txt) * test set scores: [opus-2020-01-16.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/tr-sv/opus-2020-01-16.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.tr.sv \| 26.3 \| 0.478 \|	{"license": "apache-2.0", "tags": ["translation"]}	task	[ "TRANSLATION" ]	43,128
ajgrant6/pokemon-finetuned-opus-mt-en-de	ajgrant6	translation	[ "transformers", "tensorboard", "safetensors", "marian", "text2text-generation", "translation", "generated_from_trainer", "base_model:Helsinki-NLP/opus-mt-en-de", "base_model:finetune:Helsinki-NLP/opus-mt-en-de", "license:cc-by-4.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-06-01T00:49:36Z	2024-06-01T22:52:41+00:00	8	0	--- base_model: Helsinki-NLP/opus-mt-en-de license: cc-by-4.0 tags: - translation - generated_from_trainer model-index: - name: pokemon-finetuned-opus-mt-en-de results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # pokemon-finetuned-opus-mt-en-de This model is a fine-tuned version of [Helsinki-NLP/opus-mt-en-de](https://huggingface.co/Helsinki-NLP/opus-mt-en-de) on a dataset of translated Pokemon names. It achieves the following results on the evaluation set: - Loss: 0.0554 - Exact Match: 0.9893 ## Model description This model is similar to the [Helsinki-NLP/opus-mt-en-de](https://huggingface.co/Helsinki-NLP/opus-mt-en-de) but it now properly translates Pokemon names. ## Intended uses & limitations This model is part of this [tutorial repository](https://github.com/ajgrant6/Pokemon_LLM_Finetuner). It is only intended as a proof-of-concept and is not intended for legitimate usage or deployment. This model has not been tested to see if the fine-tuning process changed anything beyond a few Pokemon-related phrases. ## Training and evaluation data The model was purposely overfit toward the training data, which was a list of translated Pokemon names from this [forum post](https://www.pokecommunity.com/threads/international-list-of-names-in-csv.460446/) ## Training procedure The evaluation and training sets were the same given a list of translated Pokemon names. ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 10 - mixed_precision_training: Native AMP ### Training results ### Framework versions - Transformers 4.41.1 - Pytorch 2.3.0+cu121 - Datasets 2.19.1 - Tokenizers 0.19.1	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # pokemon-finetuned-opus-mt-en-de This model is a fine-tuned version of [Helsinki-NLP/opus-mt-en-de](https://huggingface.co/Helsinki-NLP/opus-mt-en-de) on a dataset of translated Pokemon names. It achieves the following results on the evaluation set: - Loss: 0.0554 - Exact Match: 0.9893 ## Model description This model is similar to the [Helsinki-NLP/opus-mt-en-de](https://huggingface.co/Helsinki-NLP/opus-mt-en-de) but it now properly translates Pokemon names. ## Intended uses & limitations This model is part of this [tutorial repository](https://github.com/ajgrant6/Pokemon_LLM_Finetuner). It is only intended as a proof-of-concept and is not intended for legitimate usage or deployment. This model has not been tested to see if the fine-tuning process changed anything beyond a few Pokemon-related phrases. ## Training and evaluation data The model was purposely overfit toward the training data, which was a list of translated Pokemon names from this [forum post](https://www.pokecommunity.com/threads/international-list-of-names-in-csv.460446/) ## Training procedure The evaluation and training sets were the same given a list of translated Pokemon names. ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 10 - mixed_precision_training: Native AMP ### Training results ### Framework versions - Transformers 4.41.1 - Pytorch 2.3.0+cu121 - Datasets 2.19.1 - Tokenizers 0.19.1	{"base_model": "Helsinki-NLP/opus-mt-en-de", "license": "cc-by-4.0", "tags": ["translation", "generated_from_trainer"], "model-index": [{"name": "pokemon-finetuned-opus-mt-en-de", "results": []}]}	task	[ "TRANSLATION" ]	43,129
jjmcarrascosa/vit_receipts_classifier	jjmcarrascosa	image-classification	[ "transformers", "pytorch", "tensorboard", "safetensors", "vit", "image-classification", "generated_from_trainer", "base_model:google/vit-base-patch16-224-in21k", "base_model:finetune:google/vit-base-patch16-224-in21k", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-08-26T18:57:00Z	2023-09-11T17:47:19+00:00	191	2	--- base_model: google/vit-base-patch16-224-in21k license: apache-2.0 metrics: - f1 tags: - image-classification - generated_from_trainer model-index: - name: vit_receipts_classifier results: [] --- # vit_receipts_classifier This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the cord, rvl-cdip, visual-genome and an external receipt dataset to carry out Binary Classification (`ticket` vs `no_ticket`). Ticket here is used as a synonym to "receipt". It achieves the following results on the evaluation set, which contain pictures from the above datasets in scanned, photography or mobile picture formats (color and grayscale): - Loss: 0.0116 - F1: 0.9991 ## Model description This model is a Binary Classifier finetuned version of ViT, to predict if an input image is a picture / scan of receipts(s) o something else. ## Intended uses & limitations Use this model to classify your images into tickets or not tickers. WIth the tickets group, you can use Multimodal Information Extraction, as Visual Named Entity Recognition, to extract the ticket items, amounts, total, etc. Check the Cord dataset for more information. ## Training and evaluation data This model used 2 datasets as positive class (`ticket`): - `cord` - `https://expressexpense.com/blog/free-receipt-images-ocr-machine-learning-dataset/` For the negative class (`no_ticket`), the following datasets were used: - A subset of `RVL-CDIP` - A subset of `visual-genome` ## Training procedure Datasets were loaded with different distributions of data for positive and negative classes. Then, normalization and resizing is carried out to adapt it to ViT expected input. Different runs were carried out changing the data distribution and the hyperparameters to maximize F1. ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0002 - train_batch_size: 16 - eval_batch_size: 8 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 1 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:------:\| \| 0.0026 \| 0.28 \| 500 \| 0.0187 \| 0.9982 \| \| 0.0186 \| 0.56 \| 1000 \| 0.0116 \| 0.9991 \| \| 0.0006 \| 0.84 \| 1500 \| 0.0044 \| 0.9997 \| ### Framework versions - Transformers 4.21.2 - Pytorch 1.11.0+cu102 - Datasets 2.4.0 - Tokenizers 0.12.1	null	Non_BioNLP	# vit_receipts_classifier This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the cord, rvl-cdip, visual-genome and an external receipt dataset to carry out Binary Classification (`ticket` vs `no_ticket`). Ticket here is used as a synonym to "receipt". It achieves the following results on the evaluation set, which contain pictures from the above datasets in scanned, photography or mobile picture formats (color and grayscale): - Loss: 0.0116 - F1: 0.9991 ## Model description This model is a Binary Classifier finetuned version of ViT, to predict if an input image is a picture / scan of receipts(s) o something else. ## Intended uses & limitations Use this model to classify your images into tickets or not tickers. WIth the tickets group, you can use Multimodal Information Extraction, as Visual Named Entity Recognition, to extract the ticket items, amounts, total, etc. Check the Cord dataset for more information. ## Training and evaluation data This model used 2 datasets as positive class (`ticket`): - `cord` - `https://expressexpense.com/blog/free-receipt-images-ocr-machine-learning-dataset/` For the negative class (`no_ticket`), the following datasets were used: - A subset of `RVL-CDIP` - A subset of `visual-genome` ## Training procedure Datasets were loaded with different distributions of data for positive and negative classes. Then, normalization and resizing is carried out to adapt it to ViT expected input. Different runs were carried out changing the data distribution and the hyperparameters to maximize F1. ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0002 - train_batch_size: 16 - eval_batch_size: 8 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 1 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| F1 \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:------:\| \| 0.0026 \| 0.28 \| 500 \| 0.0187 \| 0.9982 \| \| 0.0186 \| 0.56 \| 1000 \| 0.0116 \| 0.9991 \| \| 0.0006 \| 0.84 \| 1500 \| 0.0044 \| 0.9997 \| ### Framework versions - Transformers 4.21.2 - Pytorch 1.11.0+cu102 - Datasets 2.4.0 - Tokenizers 0.12.1	{"base_model": "google/vit-base-patch16-224-in21k", "license": "apache-2.0", "metrics": ["f1"], "tags": ["image-classification", "generated_from_trainer"], "model-index": [{"name": "vit_receipts_classifier", "results": []}]}	task	[ "NAMED_ENTITY_RECOGNITION" ]	43,130
Saran09577/saran-002-ss	Saran09577	text-classification	[ "adapter-transformers", "tensorboard", "safetensors", "roberta", "autotrain", "text-classification", "en", "dataset:BrainGPT/BrainBench_Human_v0.1.csv", "base_model:cardiffnlp/twitter-roberta-base-sentiment-latest", "base_model:adapter:cardiffnlp/twitter-roberta-base-sentiment-latest", "license:afl-3.0", "region:us" ]	2024-08-18T15:08:30Z	2024-08-18T15:18:20+00:00	0	0	--- base_model: cardiffnlp/twitter-roberta-base-sentiment-latest datasets: - BrainGPT/BrainBench_Human_v0.1.csv language: - en library_name: adapter-transformers license: afl-3.0 metrics: - accuracy pipeline_tag: text-classification tags: - autotrain - text-classification widget: - text: I love AutoTrain --- # Model Trained Using AutoTrain - Problem type: Text Classification ## Validation Metrics loss: 0.5561472177505493 f1: 1.0 precision: 1.0 recall: 1.0 auc: 1.0 accuracy: 1.0	null	Non_BioNLP	# Model Trained Using AutoTrain - Problem type: Text Classification ## Validation Metrics loss: 0.5561472177505493 f1: 1.0 precision: 1.0 recall: 1.0 auc: 1.0 accuracy: 1.0	{"base_model": "cardiffnlp/twitter-roberta-base-sentiment-latest", "datasets": ["BrainGPT/BrainBench_Human_v0.1.csv"], "language": ["en"], "library_name": "adapter-transformers", "license": "afl-3.0", "metrics": ["accuracy"], "pipeline_tag": "text-classification", "tags": ["autotrain", "text-classification"], "widget": [{"text": "I love AutoTrain"}]}	task	[ "TEXT_CLASSIFICATION" ]	43,131
mradermacher/Llama-3.1-8B-Open-SFT-GGUF	mradermacher	null	[ "transformers", "gguf", "Chain-of-Thought Activation", "Llama3.1", "8B", "CoT", "SFT", "text-generation-inference", "Ollama", "safetensors", "Question Answering", "Math", "en", "dataset:O1-OPEN/OpenO1-SFT", "base_model:prithivMLmods/Llama-3.1-8B-Open-SFT", "base_model:quantized:prithivMLmods/Llama-3.1-8B-Open-SFT", "license:creativeml-openrail-m", "endpoints_compatible", "region:us", "conversational" ]	2024-12-18T21:52:30Z	2025-03-14T04:45:58+00:00	179	1	--- base_model: prithivMLmods/Llama-3.1-8B-Open-SFT datasets: - O1-OPEN/OpenO1-SFT language: - en library_name: transformers license: creativeml-openrail-m tags: - Chain-of-Thought Activation - Llama3.1 - 8B - CoT - SFT - text-generation-inference - Ollama - safetensors - Question Answering - Math quantized_by: mradermacher --- ## About <!-- ### quantize_version: 2 --> <!-- ### output_tensor_quantised: 1 --> <!-- ### convert_type: hf --> <!-- ### vocab_type: --> <!-- ### tags: --> static quants of https://huggingface.co/prithivMLmods/Llama-3.1-8B-Open-SFT <!-- provided-files --> weighted/imatrix quants are available at https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-i1-GGUF ## Usage If you are unsure how to use GGUF files, refer to one of [TheBloke's READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for more details, including on how to concatenate multi-part files. ## Provided Quants (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants) \| Link \| Type \| Size/GB \| Notes \| \|:-----\|:-----\|--------:\|:------\| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q2_K.gguf) \| Q2_K \| 3.3 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q3_K_S.gguf) \| Q3_K_S \| 3.8 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q3_K_M.gguf) \| Q3_K_M \| 4.1 \| lower quality \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q3_K_L.gguf) \| Q3_K_L \| 4.4 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.IQ4_XS.gguf) \| IQ4_XS \| 4.6 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q4_K_S.gguf) \| Q4_K_S \| 4.8 \| fast, recommended \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q4_K_M.gguf) \| Q4_K_M \| 5.0 \| fast, recommended \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q5_K_S.gguf) \| Q5_K_S \| 5.7 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q5_K_M.gguf) \| Q5_K_M \| 5.8 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q6_K.gguf) \| Q6_K \| 6.7 \| very good quality \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q8_0.gguf) \| Q8_0 \| 8.6 \| fast, best quality \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.f16.gguf) \| f16 \| 16.2 \| 16 bpw, overkill \| Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better): ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png) And here are Artefact2's thoughts on the matter: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9 ## FAQ / Model Request See https://huggingface.co/mradermacher/model_requests for some answers to questions you might have and/or if you want some other model quantized. ## Thanks I thank my company, [nethype GmbH](https://www.nethype.de/), for letting me use its servers and providing upgrades to my workstation to enable this work in my free time. <!-- end -->	null	Non_BioNLP	## About <!-- ### quantize_version: 2 --> <!-- ### output_tensor_quantised: 1 --> <!-- ### convert_type: hf --> <!-- ### vocab_type: --> <!-- ### tags: --> static quants of https://huggingface.co/prithivMLmods/Llama-3.1-8B-Open-SFT <!-- provided-files --> weighted/imatrix quants are available at https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-i1-GGUF ## Usage If you are unsure how to use GGUF files, refer to one of [TheBloke's READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for more details, including on how to concatenate multi-part files. ## Provided Quants (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants) \| Link \| Type \| Size/GB \| Notes \| \|:-----\|:-----\|--------:\|:------\| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q2_K.gguf) \| Q2_K \| 3.3 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q3_K_S.gguf) \| Q3_K_S \| 3.8 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q3_K_M.gguf) \| Q3_K_M \| 4.1 \| lower quality \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q3_K_L.gguf) \| Q3_K_L \| 4.4 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.IQ4_XS.gguf) \| IQ4_XS \| 4.6 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q4_K_S.gguf) \| Q4_K_S \| 4.8 \| fast, recommended \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q4_K_M.gguf) \| Q4_K_M \| 5.0 \| fast, recommended \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q5_K_S.gguf) \| Q5_K_S \| 5.7 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q5_K_M.gguf) \| Q5_K_M \| 5.8 \| \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q6_K.gguf) \| Q6_K \| 6.7 \| very good quality \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.Q8_0.gguf) \| Q8_0 \| 8.6 \| fast, best quality \| \| [GGUF](https://huggingface.co/mradermacher/Llama-3.1-8B-Open-SFT-GGUF/resolve/main/Llama-3.1-8B-Open-SFT.f16.gguf) \| f16 \| 16.2 \| 16 bpw, overkill \| Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better): ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png) And here are Artefact2's thoughts on the matter: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9 ## FAQ / Model Request See https://huggingface.co/mradermacher/model_requests for some answers to questions you might have and/or if you want some other model quantized. ## Thanks I thank my company, [nethype GmbH](https://www.nethype.de/), for letting me use its servers and providing upgrades to my workstation to enable this work in my free time. <!-- end -->	{"base_model": "prithivMLmods/Llama-3.1-8B-Open-SFT", "datasets": ["O1-OPEN/OpenO1-SFT"], "language": ["en"], "library_name": "transformers", "license": "creativeml-openrail-m", "tags": ["Chain-of-Thought Activation", "Llama3.1", "8B", "CoT", "SFT", "text-generation-inference", "Ollama", "safetensors", "Question Answering", "Math"], "quantized_by": "mradermacher"}	task	[ "QUESTION_ANSWERING" ]	43,132
Helsinki-NLP/opus-mt-fr-swc	Helsinki-NLP	translation	[ "transformers", "pytorch", "tf", "marian", "text2text-generation", "translation", "fr", "swc", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:04Z	2023-08-16T11:37:20+00:00	391	0	--- license: apache-2.0 tags: - translation --- ### opus-mt-fr-swc * source languages: fr * target languages: swc * OPUS readme: [fr-swc](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/fr-swc/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-16.zip](https://object.pouta.csc.fi/OPUS-MT-models/fr-swc/opus-2020-01-16.zip) * test set translations: [opus-2020-01-16.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/fr-swc/opus-2020-01-16.test.txt) * test set scores: [opus-2020-01-16.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/fr-swc/opus-2020-01-16.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.fr.swc \| 28.2 \| 0.499 \|	null	Non_BioNLP	### opus-mt-fr-swc * source languages: fr * target languages: swc * OPUS readme: [fr-swc](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/fr-swc/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-16.zip](https://object.pouta.csc.fi/OPUS-MT-models/fr-swc/opus-2020-01-16.zip) * test set translations: [opus-2020-01-16.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/fr-swc/opus-2020-01-16.test.txt) * test set scores: [opus-2020-01-16.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/fr-swc/opus-2020-01-16.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.fr.swc \| 28.2 \| 0.499 \|	{"license": "apache-2.0", "tags": ["translation"]}	task	[ "TRANSLATION" ]	43,133
Jonuu/LawyerAI1	Jonuu	sentence-similarity	[ "sentence-transformers", "safetensors", "modernbert", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:46", "loss:MatryoshkaLoss", "loss:MultipleNegativesRankingLoss", "en", "arxiv:1908.10084", "arxiv:2205.13147", "arxiv:1705.00652", "base_model:nomic-ai/modernbert-embed-base", "base_model:finetune:nomic-ai/modernbert-embed-base", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2025-03-03T18:34:01Z	2025-03-03T18:34:58+00:00	13	0	--- base_model: nomic-ai/modernbert-embed-base language: - en library_name: sentence-transformers license: apache-2.0 metrics: - cosine_accuracy@1 - cosine_accuracy@3 - cosine_accuracy@5 - cosine_accuracy@10 - cosine_precision@1 - cosine_precision@3 - cosine_precision@5 - cosine_precision@10 - cosine_recall@1 - cosine_recall@3 - cosine_recall@5 - cosine_recall@10 - cosine_ndcg@10 - cosine_mrr@10 - cosine_map@100 pipeline_tag: sentence-similarity tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:46 - loss:MatryoshkaLoss - loss:MultipleNegativesRankingLoss widget: - source_sentence: Medical science is the application of scientific principles to the study and practice of medicine. It has transformed medicine by providing a deeper understanding of the human body at the cellular and molecular levels, allowing for more effective treatments and interventions. Medical science has enabled us to develop new treatments, understand the causes of diseases, and improve patient outcomes. It's had a profound impact on the way medicine is practiced today. sentences: - I was reading about health and wellness, and I came across the term "quackery." What is quackery in the context of medicine? - That's really interesting. What is medical science, and how has it impacted the practice of medicine? - That's helpful to know. What is the primary purpose of a physical examination in medicine, anyway? - source_sentence: The purpose of differential diagnosis is to rule out conditions based on the information provided, in order to narrow down the possible causes of a patient's symptoms. By considering multiple potential diagnoses and evaluating the likelihood of each, doctors can arrive at a more accurate diagnosis and develop an effective treatment plan. sentences: - I've heard the term "differential diagnosis" before. What is the purpose of differential diagnosis? - Hello, I'm interested in learning about the various ways that diseases can be treated. Can you tell me some common ways to treat disease? - I was just wondering about what happens during a typical doctor's visit. What kinds of medical devices are typically used in basic diagnostic procedures? - source_sentence: Typically, individual governments establish legal, credentialing, and financing frameworks to support health care systems. These frameworks help to structure the way health care is delivered and accessed within a country. sentences: - That makes sense. I'm also curious about the frameworks themselves. What types of frameworks are typically established by individual governments to support health care systems? - I see. Where is contemporary medicine generally conducted? - That makes sense. I've been to the doctor's office a few times and I've seen them use those devices. What is the role of physicians and physician assistants in modern clinical practice? - source_sentence: The information gathered during a medical encounter is documented in the medical record, which is a legal document in many jurisdictions. This record contains all the relevant information about the patient's condition, treatment, and medical history, and is used to guide future care and treatment decisions. sentences: - I see. I think I understand, but I'm a bit confused. Is there a more general term for medical treatments that are used outside of scientific medicine? - That makes sense. What types of medical information might you collect from a patient's medical history? - What happens to the information gathered during a medical encounter? - source_sentence: Regional differences in culture and technology are significant factors that contribute to variations in medical availability and clinical practice around the world. These factors can shape the way healthcare is delivered, the types of treatments that are available, and even the way patients interact with healthcare professionals. It's fascinating to learn about these differences and how they impact healthcare outcomes. sentences: - I see. I'm curious about the term "therapy" in the context of treating disease. Can you explain what you understand by that term? - Hi, I'm learning about medical interviews, and I'm a bit confused about the information that's gathered about a patient's occupation and lifestyle. What information is typically gathered during the interview? - I see. I'm also interested in learning more about the variations in medical availability and clinical practice around the world. What are some factors that contribute to variations in medical availability and clinical practice around the world? model-index: - name: ModernBERT Embed base Legal Matryoshka results: - task: type: information-retrieval name: Information Retrieval dataset: name: dim 768 type: dim_768 metrics: - type: cosine_accuracy@1 value: 0.8333333333333334 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 1.0 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 1.0 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 1.0 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.8333333333333334 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.3333333333333333 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.19999999999999998 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09999999999999999 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.8333333333333334 name: Cosine Recall@1 - type: cosine_recall@3 value: 1.0 name: Cosine Recall@3 - type: cosine_recall@5 value: 1.0 name: Cosine Recall@5 - type: cosine_recall@10 value: 1.0 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.9384882922619097 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.9166666666666666 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.9166666666666666 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 512 type: dim_512 metrics: - type: cosine_accuracy@1 value: 1.0 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 1.0 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 1.0 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 1.0 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 1.0 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.3333333333333333 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.19999999999999998 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09999999999999999 name: Cosine Precision@10 - type: cosine_recall@1 value: 1.0 name: Cosine Recall@1 - type: cosine_recall@3 value: 1.0 name: Cosine Recall@3 - type: cosine_recall@5 value: 1.0 name: Cosine Recall@5 - type: cosine_recall@10 value: 1.0 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 1.0 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 1.0 name: Cosine Mrr@10 - type: cosine_map@100 value: 1.0 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 256 type: dim_256 metrics: - type: cosine_accuracy@1 value: 1.0 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 1.0 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 1.0 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 1.0 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 1.0 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.3333333333333333 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.19999999999999998 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09999999999999999 name: Cosine Precision@10 - type: cosine_recall@1 value: 1.0 name: Cosine Recall@1 - type: cosine_recall@3 value: 1.0 name: Cosine Recall@3 - type: cosine_recall@5 value: 1.0 name: Cosine Recall@5 - type: cosine_recall@10 value: 1.0 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 1.0 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 1.0 name: Cosine Mrr@10 - type: cosine_map@100 value: 1.0 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 128 type: dim_128 metrics: - type: cosine_accuracy@1 value: 0.8333333333333334 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 1.0 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 1.0 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 1.0 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.8333333333333334 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.3333333333333333 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.19999999999999998 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09999999999999999 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.8333333333333334 name: Cosine Recall@1 - type: cosine_recall@3 value: 1.0 name: Cosine Recall@3 - type: cosine_recall@5 value: 1.0 name: Cosine Recall@5 - type: cosine_recall@10 value: 1.0 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.9384882922619097 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.9166666666666666 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.9166666666666666 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 64 type: dim_64 metrics: - type: cosine_accuracy@1 value: 1.0 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 1.0 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 1.0 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 1.0 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 1.0 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.3333333333333333 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.19999999999999998 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09999999999999999 name: Cosine Precision@10 - type: cosine_recall@1 value: 1.0 name: Cosine Recall@1 - type: cosine_recall@3 value: 1.0 name: Cosine Recall@3 - type: cosine_recall@5 value: 1.0 name: Cosine Recall@5 - type: cosine_recall@10 value: 1.0 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 1.0 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 1.0 name: Cosine Mrr@10 - type: cosine_map@100 value: 1.0 name: Cosine Map@100 --- # ModernBERT Embed base Legal Matryoshka This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [nomic-ai/modernbert-embed-base](https://huggingface.co/nomic-ai/modernbert-embed-base) on the json dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [nomic-ai/modernbert-embed-base](https://huggingface.co/nomic-ai/modernbert-embed-base) <!-- at revision d556a88e332558790b210f7bdbe87da2fa94a8d8 --> - Maximum Sequence Length: 8192 tokens - Output Dimensionality: 768 dimensions - Similarity Function: Cosine Similarity - Training Dataset: - json - Language: en - License: apache-2.0 ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: ModernBertModel (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("Jonuu/LawyerAI1") # Run inference sentences = [ "Regional differences in culture and technology are significant factors that contribute to variations in medical availability and clinical practice around the world. These factors can shape the way healthcare is delivered, the types of treatments that are available, and even the way patients interact with healthcare professionals. It's fascinating to learn about these differences and how they impact healthcare outcomes.", "I see. I'm also interested in learning more about the variations in medical availability and clinical practice around the world. What are some factors that contribute to variations in medical availability and clinical practice around the world?", "Hi, I'm learning about medical interviews, and I'm a bit confused about the information that's gathered about a patient's occupation and lifestyle. What information is typically gathered during the interview?", ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 768] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Information Retrieval * Datasets: `dim_768`, `dim_512`, `dim_256`, `dim_128` and `dim_64` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| dim_768 \| dim_512 \| dim_256 \| dim_128 \| dim_64 \| \|:--------------------\|:-----------\|:--------\|:--------\|:-----------\|:--------\| \| cosine_accuracy@1 \| 0.8333 \| 1.0 \| 1.0 \| 0.8333 \| 1.0 \| \| cosine_accuracy@3 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_accuracy@5 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_accuracy@10 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_precision@1 \| 0.8333 \| 1.0 \| 1.0 \| 0.8333 \| 1.0 \| \| cosine_precision@3 \| 0.3333 \| 0.3333 \| 0.3333 \| 0.3333 \| 0.3333 \| \| cosine_precision@5 \| 0.2 \| 0.2 \| 0.2 \| 0.2 \| 0.2 \| \| cosine_precision@10 \| 0.1 \| 0.1 \| 0.1 \| 0.1 \| 0.1 \| \| cosine_recall@1 \| 0.8333 \| 1.0 \| 1.0 \| 0.8333 \| 1.0 \| \| cosine_recall@3 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_recall@5 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_recall@10 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_ndcg@10 \| 0.9385 \| 1.0 \| 1.0 \| 0.9385 \| 1.0 \| \| cosine_mrr@10 \| 0.9167 \| 1.0 \| 1.0 \| 0.9167 \| 1.0 \| \| cosine_map@100 \| 0.9167 \| 1.0 \| 1.0 \| 0.9167 \| 1.0 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### json * Dataset: json * Size: 46 training samples * Columns: <code>positive</code> and <code>anchor</code> * Approximate statistics based on the first 46 samples: \| \| positive \| anchor \| \|:--------\|:------------------------------------------------------------------------------------\|:-----------------------------------------------------------------------------------\| \| type \| string \| string \| \| details \| <ul><li>min: 37 tokens</li><li>mean: 71.26 tokens</li><li>max: 148 tokens</li></ul> \| <ul><li>min: 10 tokens</li><li>mean: 29.57 tokens</li><li>max: 47 tokens</li></ul> \| * Samples: \| positive \| anchor \| \|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>The characteristics of a health care system have a significant impact on the way medical care is provided. The structure, financing, and policies of a health care system can all influence the availability, accessibility, and quality of medical care.</code> \| <code>That helps clarify things. How do the characteristics of a health care system impact the way medical care is provided?</code> \| \| <code>Ancient philosophers and physicians applied treatments like bloodletting based on theoretical frameworks such as humorism, which attempted to explain the workings of the human body. These early theories were often influenced by cultural and philosophical beliefs, and they laid the groundwork for the development of modern medical science. It's interesting to see how our understanding of the human body has evolved over time, isn't it?</code> \| <code>I'm curious about ancient philosophers and physicians. How did they approach medicine?</code> \| \| <code>Quackery is an interesting topic. In the context of medicine, quackery refers to medical treatments that are used outside of scientific medicine, but have significant concerns related to ethics, safety, and efficacy. This means that these treatments are not necessarily supported by scientific evidence, and may even be harmful to patients.</code> \| <code>I was reading about health and wellness, and I came across the term "quackery." What is quackery in the context of medicine?</code> \| * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 768, 512, 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: epoch - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `gradient_accumulation_steps`: 16 - `learning_rate`: 2e-05 - `num_train_epochs`: 4 - `lr_scheduler_type`: cosine - `warmup_ratio`: 0.1 - `bf16`: True - `tf32`: False - `load_best_model_at_end`: True - `optim`: adamw_torch_fused - `batch_sampler`: no_duplicates #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: epoch - `prediction_loss_only`: True - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 16 - `eval_accumulation_steps`: None - `torch_empty_cache_steps`: None - `learning_rate`: 2e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 4 - `max_steps`: -1 - `lr_scheduler_type`: cosine - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.1 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: True - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: False - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: True - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch_fused - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: None - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `include_for_metrics`: [] - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `use_liger_kernel`: False - `eval_use_gather_object`: False - `average_tokens_across_devices`: False - `prompts`: None - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional </details> ### Training Logs \| Epoch \| Step \| dim_768_cosine_ndcg@10 \| dim_512_cosine_ndcg@10 \| dim_256_cosine_ndcg@10 \| dim_128_cosine_ndcg@10 \| dim_64_cosine_ndcg@10 \| \|:-------:\|:-----:\|:----------------------:\|:----------------------:\|:----------------------:\|:----------------------:\|:---------------------:\| \| 1.0 \| 1 \| 0.9385 \| 1.0 \| 0.9385 \| 0.9385 \| 1.0 \| \| 2.0 \| 2 \| 0.9385 \| 1.0 \| 1.0 \| 0.9385 \| 1.0 \| \| 3.0 \| 3 \| 0.9385 \| 1.0 \| 1.0 \| 0.9385 \| 1.0 \| \| 4.0 \| 4 \| 0.9385 \| 1.0 \| 1.0 \| 0.9385 \| 1.0 \| * The bold row denotes the saved checkpoint. ### Framework Versions - Python: 3.11.11 - Sentence Transformers: 3.4.1 - Transformers: 4.49.0 - PyTorch: 2.6.0+cu118 - Accelerate: 1.3.0 - Datasets: 3.3.2 - Tokenizers: 0.21.0 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MatryoshkaLoss ```bibtex @misc{kusupati2024matryoshka, title={Matryoshka Representation Learning}, author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi}, year={2024}, eprint={2205.13147}, archivePrefix={arXiv}, primaryClass={cs.LG} } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	null	Non_BioNLP	# ModernBERT Embed base Legal Matryoshka This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [nomic-ai/modernbert-embed-base](https://huggingface.co/nomic-ai/modernbert-embed-base) on the json dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [nomic-ai/modernbert-embed-base](https://huggingface.co/nomic-ai/modernbert-embed-base) <!-- at revision d556a88e332558790b210f7bdbe87da2fa94a8d8 --> - Maximum Sequence Length: 8192 tokens - Output Dimensionality: 768 dimensions - Similarity Function: Cosine Similarity - Training Dataset: - json - Language: en - License: apache-2.0 ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: ModernBertModel (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("Jonuu/LawyerAI1") # Run inference sentences = [ "Regional differences in culture and technology are significant factors that contribute to variations in medical availability and clinical practice around the world. These factors can shape the way healthcare is delivered, the types of treatments that are available, and even the way patients interact with healthcare professionals. It's fascinating to learn about these differences and how they impact healthcare outcomes.", "I see. I'm also interested in learning more about the variations in medical availability and clinical practice around the world. What are some factors that contribute to variations in medical availability and clinical practice around the world?", "Hi, I'm learning about medical interviews, and I'm a bit confused about the information that's gathered about a patient's occupation and lifestyle. What information is typically gathered during the interview?", ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 768] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Information Retrieval * Datasets: `dim_768`, `dim_512`, `dim_256`, `dim_128` and `dim_64` * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| dim_768 \| dim_512 \| dim_256 \| dim_128 \| dim_64 \| \|:--------------------\|:-----------\|:--------\|:--------\|:-----------\|:--------\| \| cosine_accuracy@1 \| 0.8333 \| 1.0 \| 1.0 \| 0.8333 \| 1.0 \| \| cosine_accuracy@3 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_accuracy@5 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_accuracy@10 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_precision@1 \| 0.8333 \| 1.0 \| 1.0 \| 0.8333 \| 1.0 \| \| cosine_precision@3 \| 0.3333 \| 0.3333 \| 0.3333 \| 0.3333 \| 0.3333 \| \| cosine_precision@5 \| 0.2 \| 0.2 \| 0.2 \| 0.2 \| 0.2 \| \| cosine_precision@10 \| 0.1 \| 0.1 \| 0.1 \| 0.1 \| 0.1 \| \| cosine_recall@1 \| 0.8333 \| 1.0 \| 1.0 \| 0.8333 \| 1.0 \| \| cosine_recall@3 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_recall@5 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_recall@10 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| 1.0 \| \| cosine_ndcg@10 \| 0.9385 \| 1.0 \| 1.0 \| 0.9385 \| 1.0 \| \| cosine_mrr@10 \| 0.9167 \| 1.0 \| 1.0 \| 0.9167 \| 1.0 \| \| cosine_map@100 \| 0.9167 \| 1.0 \| 1.0 \| 0.9167 \| 1.0 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### json * Dataset: json * Size: 46 training samples * Columns: <code>positive</code> and <code>anchor</code> * Approximate statistics based on the first 46 samples: \| \| positive \| anchor \| \|:--------\|:------------------------------------------------------------------------------------\|:-----------------------------------------------------------------------------------\| \| type \| string \| string \| \| details \| <ul><li>min: 37 tokens</li><li>mean: 71.26 tokens</li><li>max: 148 tokens</li></ul> \| <ul><li>min: 10 tokens</li><li>mean: 29.57 tokens</li><li>max: 47 tokens</li></ul> \| * Samples: \| positive \| anchor \| \|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>The characteristics of a health care system have a significant impact on the way medical care is provided. The structure, financing, and policies of a health care system can all influence the availability, accessibility, and quality of medical care.</code> \| <code>That helps clarify things. How do the characteristics of a health care system impact the way medical care is provided?</code> \| \| <code>Ancient philosophers and physicians applied treatments like bloodletting based on theoretical frameworks such as humorism, which attempted to explain the workings of the human body. These early theories were often influenced by cultural and philosophical beliefs, and they laid the groundwork for the development of modern medical science. It's interesting to see how our understanding of the human body has evolved over time, isn't it?</code> \| <code>I'm curious about ancient philosophers and physicians. How did they approach medicine?</code> \| \| <code>Quackery is an interesting topic. In the context of medicine, quackery refers to medical treatments that are used outside of scientific medicine, but have significant concerns related to ethics, safety, and efficacy. This means that these treatments are not necessarily supported by scientific evidence, and may even be harmful to patients.</code> \| <code>I was reading about health and wellness, and I came across the term "quackery." What is quackery in the context of medicine?</code> \| * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 768, 512, 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: epoch - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `gradient_accumulation_steps`: 16 - `learning_rate`: 2e-05 - `num_train_epochs`: 4 - `lr_scheduler_type`: cosine - `warmup_ratio`: 0.1 - `bf16`: True - `tf32`: False - `load_best_model_at_end`: True - `optim`: adamw_torch_fused - `batch_sampler`: no_duplicates #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: epoch - `prediction_loss_only`: True - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 16 - `eval_accumulation_steps`: None - `torch_empty_cache_steps`: None - `learning_rate`: 2e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 4 - `max_steps`: -1 - `lr_scheduler_type`: cosine - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.1 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: True - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: False - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: True - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch_fused - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: None - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `include_for_metrics`: [] - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `use_liger_kernel`: False - `eval_use_gather_object`: False - `average_tokens_across_devices`: False - `prompts`: None - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional </details> ### Training Logs \| Epoch \| Step \| dim_768_cosine_ndcg@10 \| dim_512_cosine_ndcg@10 \| dim_256_cosine_ndcg@10 \| dim_128_cosine_ndcg@10 \| dim_64_cosine_ndcg@10 \| \|:-------:\|:-----:\|:----------------------:\|:----------------------:\|:----------------------:\|:----------------------:\|:---------------------:\| \| 1.0 \| 1 \| 0.9385 \| 1.0 \| 0.9385 \| 0.9385 \| 1.0 \| \| 2.0 \| 2 \| 0.9385 \| 1.0 \| 1.0 \| 0.9385 \| 1.0 \| \| 3.0 \| 3 \| 0.9385 \| 1.0 \| 1.0 \| 0.9385 \| 1.0 \| \| 4.0 \| 4 \| 0.9385 \| 1.0 \| 1.0 \| 0.9385 \| 1.0 \| * The bold row denotes the saved checkpoint. ### Framework Versions - Python: 3.11.11 - Sentence Transformers: 3.4.1 - Transformers: 4.49.0 - PyTorch: 2.6.0+cu118 - Accelerate: 1.3.0 - Datasets: 3.3.2 - Tokenizers: 0.21.0 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MatryoshkaLoss ```bibtex @misc{kusupati2024matryoshka, title={Matryoshka Representation Learning}, author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi}, year={2024}, eprint={2205.13147}, archivePrefix={arXiv}, primaryClass={cs.LG} } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	{"base_model": "nomic-ai/modernbert-embed-base", "language": ["en"], "library_name": "sentence-transformers", "license": "apache-2.0", "metrics": ["cosine_accuracy@1", "cosine_accuracy@3", "cosine_accuracy@5", "cosine_accuracy@10", "cosine_precision@1", "cosine_precision@3", "cosine_precision@5", "cosine_precision@10", "cosine_recall@1", "cosine_recall@3", "cosine_recall@5", "cosine_recall@10", "cosine_ndcg@10", "cosine_mrr@10", "cosine_map@100"], "pipeline_tag": "sentence-similarity", "tags": ["sentence-transformers", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:46", "loss:MatryoshkaLoss", "loss:MultipleNegativesRankingLoss"], "widget": [{"source_sentence": "Medical science is the application of scientific principles to the study and practice of medicine. It has transformed medicine by providing a deeper understanding of the human body at the cellular and molecular levels, allowing for more effective treatments and interventions. Medical science has enabled us to develop new treatments, understand the causes of diseases, and improve patient outcomes. It's had a profound impact on the way medicine is practiced today.", "sentences": ["I was reading about health and wellness, and I came across the term \"quackery.\" What is quackery in the context of medicine?", "That's really interesting. What is medical science, and how has it impacted the practice of medicine?", "That's helpful to know. What is the primary purpose of a physical examination in medicine, anyway?"]}, {"source_sentence": "The purpose of differential diagnosis is to rule out conditions based on the information provided, in order to narrow down the possible causes of a patient's symptoms. By considering multiple potential diagnoses and evaluating the likelihood of each, doctors can arrive at a more accurate diagnosis and develop an effective treatment plan.", "sentences": ["I've heard the term \"differential diagnosis\" before. What is the purpose of differential diagnosis?", "Hello, I'm interested in learning about the various ways that diseases can be treated. Can you tell me some common ways to treat disease?", "I was just wondering about what happens during a typical doctor's visit. What kinds of medical devices are typically used in basic diagnostic procedures?"]}, {"source_sentence": "Typically, individual governments establish legal, credentialing, and financing frameworks to support health care systems. These frameworks help to structure the way health care is delivered and accessed within a country.", "sentences": ["That makes sense. I'm also curious about the frameworks themselves. What types of frameworks are typically established by individual governments to support health care systems?", "I see. Where is contemporary medicine generally conducted?", "That makes sense. I've been to the doctor's office a few times and I've seen them use those devices. What is the role of physicians and physician assistants in modern clinical practice?"]}, {"source_sentence": "The information gathered during a medical encounter is documented in the medical record, which is a legal document in many jurisdictions. This record contains all the relevant information about the patient's condition, treatment, and medical history, and is used to guide future care and treatment decisions.", "sentences": ["I see. I think I understand, but I'm a bit confused. Is there a more general term for medical treatments that are used outside of scientific medicine?", "That makes sense. What types of medical information might you collect from a patient's medical history?", "What happens to the information gathered during a medical encounter?"]}, {"source_sentence": "Regional differences in culture and technology are significant factors that contribute to variations in medical availability and clinical practice around the world. These factors can shape the way healthcare is delivered, the types of treatments that are available, and even the way patients interact with healthcare professionals. It's fascinating to learn about these differences and how they impact healthcare outcomes.", "sentences": ["I see. I'm curious about the term \"therapy\" in the context of treating disease. Can you explain what you understand by that term?", "Hi, I'm learning about medical interviews, and I'm a bit confused about the information that's gathered about a patient's occupation and lifestyle. What information is typically gathered during the interview?", "I see. I'm also interested in learning more about the variations in medical availability and clinical practice around the world. What are some factors that contribute to variations in medical availability and clinical practice around the world?"]}], "model-index": [{"name": "ModernBERT Embed base Legal Matryoshka", "results": [{"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 768", "type": "dim_768"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.8333333333333334, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 1.0, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 1.0, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 1.0, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.8333333333333334, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.3333333333333333, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.19999999999999998, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09999999999999999, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.8333333333333334, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 1.0, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 1.0, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 1.0, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.9384882922619097, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.9166666666666666, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.9166666666666666, "name": "Cosine Map@100"}]}, {"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 512", "type": "dim_512"}, "metrics": [{"type": "cosine_accuracy@1", "value": 1.0, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 1.0, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 1.0, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 1.0, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 1.0, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.3333333333333333, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.19999999999999998, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09999999999999999, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 1.0, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 1.0, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 1.0, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 1.0, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 1.0, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 1.0, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 1.0, "name": "Cosine Map@100"}]}, {"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 256", "type": "dim_256"}, "metrics": [{"type": "cosine_accuracy@1", "value": 1.0, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 1.0, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 1.0, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 1.0, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 1.0, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.3333333333333333, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.19999999999999998, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09999999999999999, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 1.0, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 1.0, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 1.0, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 1.0, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 1.0, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 1.0, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 1.0, "name": "Cosine Map@100"}]}, {"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 128", "type": "dim_128"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.8333333333333334, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 1.0, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 1.0, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 1.0, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.8333333333333334, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.3333333333333333, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.19999999999999998, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09999999999999999, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.8333333333333334, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 1.0, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 1.0, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 1.0, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.9384882922619097, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.9166666666666666, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.9166666666666666, "name": "Cosine Map@100"}]}, {"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "dim 64", "type": "dim_64"}, "metrics": [{"type": "cosine_accuracy@1", "value": 1.0, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 1.0, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 1.0, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 1.0, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 1.0, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.3333333333333333, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.19999999999999998, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09999999999999999, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 1.0, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 1.0, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 1.0, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 1.0, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 1.0, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 1.0, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 1.0, "name": "Cosine Map@100"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,134
TransQuest/microtransquest-en_zh-wiki	TransQuest	token-classification	[ "transformers", "pytorch", "xlm-roberta", "token-classification", "Quality Estimation", "microtransquest", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:05Z	2021-06-04T08:22:58+00:00	15	0	--- language: en-zh license: apache-2.0 tags: - Quality Estimation - microtransquest --- # TransQuest: Translation Quality Estimation with Cross-lingual Transformers The goal of quality estimation (QE) is to evaluate the quality of a translation without having access to a reference translation. High-accuracy QE that can be easily deployed for a number of language pairs is the missing piece in many commercial translation workflows as they have numerous potential uses. They can be employed to select the best translation when several translation engines are available or can inform the end user about the reliability of automatically translated content. In addition, QE systems can be used to decide whether a translation can be published as it is in a given context, or whether it requires human post-editing before publishing or translation from scratch by a human. The quality estimation can be done at different levels: document level, sentence level and word level. With TransQuest, we have opensourced our research in translation quality estimation which also won the sentence-level direct assessment quality estimation shared task in [WMT 2020](http://www.statmt.org/wmt20/quality-estimation-task.html). TransQuest outperforms current open-source quality estimation frameworks such as [OpenKiwi](https://github.com/Unbabel/OpenKiwi) and [DeepQuest](https://github.com/sheffieldnlp/deepQuest). ## Features - Sentence-level translation quality estimation on both aspects: predicting post editing efforts and direct assessment. - Word-level translation quality estimation capable of predicting quality of source words, target words and target gaps. - Outperform current state-of-the-art quality estimation methods like DeepQuest and OpenKiwi in all the languages experimented. - Pre-trained quality estimation models for fifteen language pairs are available in [HuggingFace.](https://huggingface.co/TransQuest) ## Installation ### From pip ```bash pip install transquest ``` ### From Source ```bash git clone https://github.com/TharinduDR/TransQuest.git cd TransQuest pip install -r requirements.txt ``` ## Using Pre-trained Models ```python from transquest.algo.word_level.microtransquest.run_model import MicroTransQuestModel import torch model = MicroTransQuestModel("xlmroberta", "TransQuest/microtransquest-en_zh-wiki", labels=["OK", "BAD"], use_cuda=torch.cuda.is_available()) source_tags, target_tags = model.predict([["if not , you may not be protected against the diseases . ", "ja tā nav , Jūs varat nepasargāt no slimībām . "]]) ``` ## Documentation For more details follow the documentation. 1. [Installation](https://tharindudr.github.io/TransQuest/install/) - Install TransQuest locally using pip. 2. Architectures - Checkout the architectures implemented in TransQuest 1. [Sentence-level Architectures](https://tharindudr.github.io/TransQuest/architectures/sentence_level_architectures/) - We have released two architectures; MonoTransQuest and SiameseTransQuest to perform sentence level quality estimation. 2. [Word-level Architecture](https://tharindudr.github.io/TransQuest/architectures/word_level_architecture/) - We have released MicroTransQuest to perform word level quality estimation. 3. Examples - We have provided several examples on how to use TransQuest in recent WMT quality estimation shared tasks. 1. [Sentence-level Examples](https://tharindudr.github.io/TransQuest/examples/sentence_level_examples/) 2. [Word-level Examples](https://tharindudr.github.io/TransQuest/examples/word_level_examples/) 4. Pre-trained Models - We have provided pretrained quality estimation models for fifteen language pairs covering both sentence-level and word-level 1. [Sentence-level Models](https://tharindudr.github.io/TransQuest/models/sentence_level_pretrained/) 2. [Word-level Models](https://tharindudr.github.io/TransQuest/models/word_level_pretrained/) 5. [Contact](https://tharindudr.github.io/TransQuest/contact/) - Contact us for any issues with TransQuest ## Citations If you are using the word-level architecture, please consider citing this paper which is accepted to [ACL 2021](https://2021.aclweb.org/). ```bash @InProceedings{ranasinghe2021, author = {Ranasinghe, Tharindu and Orasan, Constantin and Mitkov, Ruslan}, title = {An Exploratory Analysis of Multilingual Word Level Quality Estimation with Cross-Lingual Transformers}, booktitle = {Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics}, year = {2021} } ``` If you are using the sentence-level architectures, please consider citing these papers which were presented in [COLING 2020](https://coling2020.org/) and in [WMT 2020](http://www.statmt.org/wmt20/) at EMNLP 2020. ```bash @InProceedings{transquest:2020a, author = {Ranasinghe, Tharindu and Orasan, Constantin and Mitkov, Ruslan}, title = {TransQuest: Translation Quality Estimation with Cross-lingual Transformers}, booktitle = {Proceedings of the 28th International Conference on Computational Linguistics}, year = {2020} } ``` ```bash @InProceedings{transquest:2020b, author = {Ranasinghe, Tharindu and Orasan, Constantin and Mitkov, Ruslan}, title = {TransQuest at WMT2020: Sentence-Level Direct Assessment}, booktitle = {Proceedings of the Fifth Conference on Machine Translation}, year = {2020} } ```	null	Non_BioNLP	# TransQuest: Translation Quality Estimation with Cross-lingual Transformers The goal of quality estimation (QE) is to evaluate the quality of a translation without having access to a reference translation. High-accuracy QE that can be easily deployed for a number of language pairs is the missing piece in many commercial translation workflows as they have numerous potential uses. They can be employed to select the best translation when several translation engines are available or can inform the end user about the reliability of automatically translated content. In addition, QE systems can be used to decide whether a translation can be published as it is in a given context, or whether it requires human post-editing before publishing or translation from scratch by a human. The quality estimation can be done at different levels: document level, sentence level and word level. With TransQuest, we have opensourced our research in translation quality estimation which also won the sentence-level direct assessment quality estimation shared task in [WMT 2020](http://www.statmt.org/wmt20/quality-estimation-task.html). TransQuest outperforms current open-source quality estimation frameworks such as [OpenKiwi](https://github.com/Unbabel/OpenKiwi) and [DeepQuest](https://github.com/sheffieldnlp/deepQuest). ## Features - Sentence-level translation quality estimation on both aspects: predicting post editing efforts and direct assessment. - Word-level translation quality estimation capable of predicting quality of source words, target words and target gaps. - Outperform current state-of-the-art quality estimation methods like DeepQuest and OpenKiwi in all the languages experimented. - Pre-trained quality estimation models for fifteen language pairs are available in [HuggingFace.](https://huggingface.co/TransQuest) ## Installation ### From pip ```bash pip install transquest ``` ### From Source ```bash git clone https://github.com/TharinduDR/TransQuest.git cd TransQuest pip install -r requirements.txt ``` ## Using Pre-trained Models ```python from transquest.algo.word_level.microtransquest.run_model import MicroTransQuestModel import torch model = MicroTransQuestModel("xlmroberta", "TransQuest/microtransquest-en_zh-wiki", labels=["OK", "BAD"], use_cuda=torch.cuda.is_available()) source_tags, target_tags = model.predict([["if not , you may not be protected against the diseases . ", "ja tā nav , Jūs varat nepasargāt no slimībām . "]]) ``` ## Documentation For more details follow the documentation. 1. [Installation](https://tharindudr.github.io/TransQuest/install/) - Install TransQuest locally using pip. 2. Architectures - Checkout the architectures implemented in TransQuest 1. [Sentence-level Architectures](https://tharindudr.github.io/TransQuest/architectures/sentence_level_architectures/) - We have released two architectures; MonoTransQuest and SiameseTransQuest to perform sentence level quality estimation. 2. [Word-level Architecture](https://tharindudr.github.io/TransQuest/architectures/word_level_architecture/) - We have released MicroTransQuest to perform word level quality estimation. 3. Examples - We have provided several examples on how to use TransQuest in recent WMT quality estimation shared tasks. 1. [Sentence-level Examples](https://tharindudr.github.io/TransQuest/examples/sentence_level_examples/) 2. [Word-level Examples](https://tharindudr.github.io/TransQuest/examples/word_level_examples/) 4. Pre-trained Models - We have provided pretrained quality estimation models for fifteen language pairs covering both sentence-level and word-level 1. [Sentence-level Models](https://tharindudr.github.io/TransQuest/models/sentence_level_pretrained/) 2. [Word-level Models](https://tharindudr.github.io/TransQuest/models/word_level_pretrained/) 5. [Contact](https://tharindudr.github.io/TransQuest/contact/) - Contact us for any issues with TransQuest ## Citations If you are using the word-level architecture, please consider citing this paper which is accepted to [ACL 2021](https://2021.aclweb.org/). ```bash @InProceedings{ranasinghe2021, author = {Ranasinghe, Tharindu and Orasan, Constantin and Mitkov, Ruslan}, title = {An Exploratory Analysis of Multilingual Word Level Quality Estimation with Cross-Lingual Transformers}, booktitle = {Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics}, year = {2021} } ``` If you are using the sentence-level architectures, please consider citing these papers which were presented in [COLING 2020](https://coling2020.org/) and in [WMT 2020](http://www.statmt.org/wmt20/) at EMNLP 2020. ```bash @InProceedings{transquest:2020a, author = {Ranasinghe, Tharindu and Orasan, Constantin and Mitkov, Ruslan}, title = {TransQuest: Translation Quality Estimation with Cross-lingual Transformers}, booktitle = {Proceedings of the 28th International Conference on Computational Linguistics}, year = {2020} } ``` ```bash @InProceedings{transquest:2020b, author = {Ranasinghe, Tharindu and Orasan, Constantin and Mitkov, Ruslan}, title = {TransQuest at WMT2020: Sentence-Level Direct Assessment}, booktitle = {Proceedings of the Fifth Conference on Machine Translation}, year = {2020} } ```	{"language": "en-zh", "license": "apache-2.0", "tags": ["Quality Estimation", "microtransquest"]}	task	[ "TRANSLATION" ]	43,135
felipeoes/cocoruta-7b	felipeoes	text-generation	[ "transformers", "pytorch", "safetensors", "llama", "text-generation", "pt", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "region:us" ]	2023-10-28T19:00:36Z	2024-11-07T23:23:59+00:00	30	0	--- language: - pt --- ## Cocoruta: A Legal Domain-Driven Q&A System ### Disclaimer Cocoruta may reproduce biases and prejudices inherent in the legal documents used for its training, which include older legislation. Users should exercise caution when interpreting the model’s outputs, especially in contexts requiring up-to-date legal perspectives or that may involve underrepresented groups. We observed that the Cocoruta model, while less proficient in handling utterances compared to larger models, would impart a legal bias to potential interactions.* ### Model Overview [Cocoruta](https://huggingface.co/felipeoes/cocoruta-7b) is a specialized large language model fine-tuned for legal document-based Question Answering (Q&A), developed to address legal queries related to the "Blue Amazon"—a term used to describe Brazil's extensive maritime territory. Cocoruta 1.0 is based on the LLaMa 2-7B model, fine-tuned with a corpus of 68,991 legal documents totaling 28.4 million tokens. Despite being trained with fewer parameters than some larger models, Cocoruta demonstrates competitive performance in domain-specific legal discourse. ### Training and Technical Specifications - Parameter count: 7B (LLaMa 2-7B) - Training data: 28.4 million tokens from 68,991 legal documents - Training epochs: 15 ### Evaluation Metrics #### Automatic Evaluation Cocoruta has been evaluated using multiple automatic metrics to measure its effectiveness in generating accurate and relevant legal content. The model performed as follows: - BLEU: 61.2 - ROUGE-N: 79.2 - BERTSCORE: 91.2 - MOVERSCORE: 76.5 #### Qualitative Evaluation The performance of Cocoruta in qualitative evaluation showed the utility of fine-tuning, as answers aligned with legal discourse were more frequent in Cocoruta compared to larger models. The larger models exhibited higher proficiency, delivering well-structured answers. However, for questions not directly related to the legal context, responses from the larger models did not maintain legal discourse.* - Adherence to legal discourse: 74% - Correct answers: 68% - Inappropriate discourse: 51% ### Citation If you use Cocoruta in your research, please cite the following paper: ```bibtex *@inproceedings{2024cocoruta, author={do Espírito Santo, Felipe Oliveira and Marques Peres, Sarajane and de Sousa Gramacho, Givanildo and Alves Franco Brandão, Anarosa and Cozman, Fabio Gagliardi}, booktitle={2024 International Joint Conference on Neural Networks (IJCNN)}, title={Legal Document-Based, Domain-Driven Q&A System: LLMs in Perspective}, year={2024}, volume={}, number={}, pages={1-9}, keywords={Law;Large language models;Neural networks;Question answering (information retrieval);Complexity theory;Large language models;LLM evaluation;legal Q&A systems;legal-document corpus}, address={Yokohama, Japan}, isbn={978-8-3503-5931-2}, doi={10.1109/IJCNN60899.2024.10650895}, url={https://ieeexplore.ieee.org/abstract/document/10650895} } ``` ### External Links - [Cocoruta Paper](https://ieeexplore.ieee.org/abstract/document/10650895) - [KEML Website](https://sites.usp.br/keml)	null	Non_BioNLP	## Cocoruta: A Legal Domain-Driven Q&A System ### Disclaimer Cocoruta may reproduce biases and prejudices inherent in the legal documents used for its training, which include older legislation. Users should exercise caution when interpreting the model’s outputs, especially in contexts requiring up-to-date legal perspectives or that may involve underrepresented groups. We observed that the Cocoruta model, while less proficient in handling utterances compared to larger models, would impart a legal bias to potential interactions.* ### Model Overview [Cocoruta](https://huggingface.co/felipeoes/cocoruta-7b) is a specialized large language model fine-tuned for legal document-based Question Answering (Q&A), developed to address legal queries related to the "Blue Amazon"—a term used to describe Brazil's extensive maritime territory. Cocoruta 1.0 is based on the LLaMa 2-7B model, fine-tuned with a corpus of 68,991 legal documents totaling 28.4 million tokens. Despite being trained with fewer parameters than some larger models, Cocoruta demonstrates competitive performance in domain-specific legal discourse. ### Training and Technical Specifications - Parameter count: 7B (LLaMa 2-7B) - Training data: 28.4 million tokens from 68,991 legal documents - Training epochs: 15 ### Evaluation Metrics #### Automatic Evaluation Cocoruta has been evaluated using multiple automatic metrics to measure its effectiveness in generating accurate and relevant legal content. The model performed as follows: - BLEU: 61.2 - ROUGE-N: 79.2 - BERTSCORE: 91.2 - MOVERSCORE: 76.5 #### Qualitative Evaluation The performance of Cocoruta in qualitative evaluation showed the utility of fine-tuning, as answers aligned with legal discourse were more frequent in Cocoruta compared to larger models. The larger models exhibited higher proficiency, delivering well-structured answers. However, for questions not directly related to the legal context, responses from the larger models did not maintain legal discourse.* - Adherence to legal discourse: 74% - Correct answers: 68% - Inappropriate discourse: 51% ### Citation If you use Cocoruta in your research, please cite the following paper: ```bibtex *@inproceedings{2024cocoruta, author={do Espírito Santo, Felipe Oliveira and Marques Peres, Sarajane and de Sousa Gramacho, Givanildo and Alves Franco Brandão, Anarosa and Cozman, Fabio Gagliardi}, booktitle={2024 International Joint Conference on Neural Networks (IJCNN)}, title={Legal Document-Based, Domain-Driven Q&A System: LLMs in Perspective}, year={2024}, volume={}, number={}, pages={1-9}, keywords={Law;Large language models;Neural networks;Question answering (information retrieval);Complexity theory;Large language models;LLM evaluation;legal Q&A systems;legal-document corpus}, address={Yokohama, Japan}, isbn={978-8-3503-5931-2}, doi={10.1109/IJCNN60899.2024.10650895}, url={https://ieeexplore.ieee.org/abstract/document/10650895} } ``` ### External Links - [Cocoruta Paper](https://ieeexplore.ieee.org/abstract/document/10650895) - [KEML Website](https://sites.usp.br/keml)	{"language": ["pt"]}	task	[ "QUESTION_ANSWERING" ]	43,136
benjamin-paine/taproot-common	benjamin-paine	null	[ "gguf", "arxiv:2307.01952", "arxiv:1512.00567", "arxiv:2305.08227", "arxiv:2202.04901", "arxiv:2011.06294", "arxiv:2410.06885", "arxiv:2212.04356", "arxiv:2311.00430", "arxiv:2103.13413", "arxiv:2203.12691", "arxiv:2106.00186", "arxiv:1504.06375", "arxiv:1812.08008", "arxiv:2307.15880", "arxiv:2112.10752", "arxiv:2402.13929", "arxiv:2403.03206", "arxiv:2408.06072", "arxiv:2412.03603", "arxiv:2407.21783", "arxiv:2310.16944", "arxiv:2310.03744", "license:apache-2.0", "region:us" ]	2024-06-20T19:42:49Z	2025-03-06T01:17:23+00:00	1,014	5	--- license: apache-2.0 --- <div align="center"> <img src="https://github.com/user-attachments/assets/f965fd42-2a95-4552-9b5f-465fc4037a91" width="650" /><br /> <em>An open source real-time AI inference engine for seamless scaling</em> </div> # About Taproot is a seamlessly scalable AI/ML inference engine designed for deployment across hardware clusters with disparate capabilities. ## Why Taproot? Most AI/ML inference engines are built for either large-scale cloud infrastructures or constrained edge devices - Taproot is designed for medium-scale deployments, offering flexible and distributed on-premise or PAYG setups. It efficiently uses older or consumer-grade hardware, making it suitable for small networks or ad-hoc clusters, without relying on centralized, hyperscale architectures. ## Available Models There are more than 150 models available across 18 task categories. See the [Task Catalog](#task-catalog) for the complete list, licenses, requirements and citations. Despite the large number of models available, there are many more yet to be added - if you're looking for a particular enhancement, don't hesitate to make an issue on this repository to request it. ### Roadmap 1. IP Adapter Models for Diffusers Image Generation Pipelines 2. ControlNet Models for Diffusers Image Generation Pipelines 3. Additional quantization backends for large models - Currently BitsandBytes (Int8/NF4) and GGUF (through llama.cpp) are supported with pre-quantized checkpoints available. - FP8 support through Optimum-Quanto, TorchAO and custom kernels is in development. 4. Improved multi-GPU support - This is currently supported through manual configuration, but usability can be improved. 5. Additional annotators/detectors for image and video - E.g. Marigold, SAM2 6. Additional audio generation models - E.g. Stable Audio, AudioLDM, MusicGen # Installation ```sh pip install taproot ``` Some additional packages are available to install with the square-bracket syntax (e.g. `pip install taproot[a,b,c]`), these are: - tools - Additional packages for LLM tools like DuckDuckGo Search, BeautifulSoup (for web scraping), etc. - console - Additional packages for prettifying console output. - av - Additional packages for reading and writing video. ## Installing Tasks Some tasks are available immediately, but most tasks required additional packages and files. Install these tasks with `taproot install [task:model]+`, e.g: ```sh taproot install image-generation:stable-diffusion-xl ``` # Usage ## Command-Line ### Introspecting Tasks From the command line, execute `taproot tasks` to see all tasks and their availability status, or `taproot info` for individual task information. For example: ```sh taproot info image-generation stable-diffusion-xl Stable Diffusion XL Image Generation (image-generation:stable-diffusion-xl, available) Generate an image from text and/or images using a stable diffusion XL model. Hardware Requirements: GPU Required for Optimal Performance Floating Point Precision: half Minimum Memory (CPU RAM) Required: 231.71 MB Minimum Memory (GPU VRAM) Required: 7.58 GB Author: Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023 https://arxiv.org/abs/2307.01952 License: OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md) ✅ Attribution Required ✅ Derivatives Allowed ✅ Redistribution Allowed ✅ Copyleft (Share-Alike) Required ✅ Commercial Use Allowed ✅ Hosting Allowed Files: image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB) [downloaded] image-generation-stable-diffusion-xl-base-unet.fp16.safetensors (5.14 GB) [downloaded] text-encoding-clip-vit-l.bf16.safetensors (246.14 MB) [downloaded] text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB) [downloaded] text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB) [downloaded] text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B) [downloaded] text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB) [downloaded] text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB) [downloaded] text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B) [downloaded] text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB) [downloaded] Total File Size: 7.11 GB Required packages: pil~=9.5 [installed] torch<2.5,>=2.4 [installed] numpy~=1.22 [installed] diffusers>=0.29 [installed] torchvision<0.20,>=0.19 [installed] transformers>=4.41 [installed] safetensors~=0.4 [installed] accelerate~=1.0 [installed] sentencepiece~=0.2 [installed] compel~=2.0 [installed] peft~=0.13 [installed] Signature: prompt: Union[str, List[str]], required prompt_2: Union[str, List[str]], default: None negative_prompt: Union[str, List[str]], default: None negative_prompt_2: Union[str, List[str]], default: None image: ImageType, default: None mask_image: ImageType, default: None guidance_scale: float, default: 5.0 guidance_rescale: float, default: 0.0 num_inference_steps: int, default: 20 num_images_per_prompt: int, default: 1 height: int, default: None width: int, default: None timesteps: List[int], default: None sigmas: List[float], default: None denoising_end: float, default: None strength: float, default: None latents: torch.Tensor, default: None prompt_embeds: torch.Tensor, default: None negative_prompt_embeds: torch.Tensor, default: None pooled_prompt_embeds: torch.Tensor, default: None negative_pooled_prompt_embeds: torch.Tensor, default: None clip_skip: int, default: None seed: SeedType, default: None pag_scale: float, default: None pag_adaptive_scale: float, default: None scheduler: Literal[ddim, ddpm, ddpm_wuerstchen, deis_multistep, dpm_cogvideox, dpmsolver_multistep, dpmsolver_multistep_karras, dpmsolver_sde, dpmsolver_sde_multistep, dpmsolver_sde_multistep_karras, dpmsolver_singlestep, dpmsolver_singlestep_karras, edm_dpmsolver_multistep, edm_euler, euler_ancestral_discrete, euler_discrete, euler_discrete_karras, flow_match_euler_discrete, flow_match_heun_discrete, heun_discrete, ipndm, k_dpm_2_ancestral_discrete, k_dpm_2_ancestral_discrete_karras, k_dpm_2_discrete, k_dpm_2_discrete_karras, lcm, lms_discrete, lms_discrete_karras, pndm, tcd, unipc], default: None output_format: Literal[png, jpeg, float, int, latent], default: png output_upload: bool, default: False highres_fix_factor: float, default: 1.0 highres_fix_strength: float, default: None spatial_prompts: SpatialPromptInputType, default: None Returns: ImageResultType ``` ### Invoking Tasks Run `taproot invoke` to run any task from the command line. All parameters to the task can be passed as flags to the call using kebab-case, e.g.: ```sh taproot invoke image-generation:stable-diffusion-xl \ --prompt "a photograph of a golden retriever at the park" \ --negative-prompt "fall, autumn, blurry, out-of-focus" \ --seed 12345 Loading task. 100%\|███████████████████████████████████████████████████████████████████████████\| 7/7 [00:03<00:00, 2.27it/s] Task loaded in 4.0 s. Invoking task. 100%\|█████████████████████████████████████████████████████████████████████████\| 20/20 [00:04<00:00, 4.34it/s] Task invoked in 6.5 s. Result: 8940aa12-66a7-4233-bfd6-f19da339b71b.png ``` ## Python ### Direct Task Usage ```py from taproot import Task sdxl = Task.get("image-generation", "stable-diffusion-xl") pipeline = sdxl() pipeline.load() pipeline(prompt="Hello, world!").save("./output.png") ``` ### With a Remote Server ```py from taproot import Tap tap = Tap() tap.remote_address = "ws://127.0.0.1:32189" result = tap.call("image-generation", model="stable-diffusion-xl", prompt="Hello, world!") result.save("./output.png") ``` ### With a Local Server Also shows asynchronous usage. ```py import asyncio from taproot import Tap with Tap.local() as tap: loop = asyncio.get_event_loop() result = loop.run_until_complete(tap("image-generation", model="stable-diffusion-xl", prompt="Hello, world!")) result.save("./output.png") ``` ## Running Servers Taproot uses a three-roled cluster structure: 1. Overseers are entry points into clusters, routing requests to one or more dispatchers. 2. Dispatchers are machines capable of running tasks by spawning executors. 3. Executors are servers ready to execute a task. The simplest way to run a server is to run an overseer simultaneously with a local dispatcher like so: ```sh taproot overseer --local ``` This will run on the default address of `ws://127.0.0.1:32189`, suitable for interaction from python or the browser. There are many deployment possibilities across networks, with configuration available for encryption, listening addresses, and more. See the wiki for details (coming soon.) ## Outside Python - [taproot.js](https://github.com/painebenjamin/taproot.js) - for the browser and node.js, available in ESM, UMD and IIFE - taproot.php - coming soon <h1>Task Catalog</h1> <p>18 tasks available with 171 models.</p> <ul> <li><a href="#echo">echo</a>: 1 model</li> <li><a href="#image-similarity">image-similarity</a>: 2 models</li> <li><a href="#text-similarity">text-similarity</a>: 1 model</li> <li><a href="#speech-enhancement">speech-enhancement</a>: 1 model</li> <li><a href="#image-interpolation">image-interpolation</a>: 2 models</li> <li><a href="#background-removal">background-removal</a>: 1 model</li> <li><a href="#super-resolution">super-resolution</a>: 2 models</li> <li><a href="#speech-synthesis">speech-synthesis</a>: 2 models</li> <li><a href="#audio-transcription">audio-transcription</a>: 9 models</li> <li><a href="#depth-detection">depth-detection</a>: 1 model</li> <li><a href="#line-detection">line-detection</a>: 4 models</li> <li><a href="#edge-detection">edge-detection</a>: 3 models</li> <li><a href="#pose-detection">pose-detection</a>: 2 models</li> <li><a href="#image-generation">image-generation</a>: 52 models</li> <li><a href="#video-generation">video-generation</a>: 23 models</li> <li><a href="#text-generation">text-generation</a>: 37 models</li> <li><a href="#visual-question-answering">visual-question-answering</a>: 14 models</li> <li><a href="#image-captioning">image-captioning</a>: 14 models</li> </ul> <h2>echo</h2> <table><tbody><tr><td>Name</td><td>Echo</td></tr><tr><td>Author</td><td>Benjamin Paine<br />Taproot<br />https://github.com/painebenjamin/taproot</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td>N/A</td></tr><tr><td>Minimum VRAM</td><td>N/A</td></tr></tbody></table> <h2>image-similarity</h2> <h3>(default)</h3> <table><tbody><tr><td>Name</td><td>Traditional Image Similarity</td></tr><tr><td>Author</td><td>Benjamin Paine<br />Taproot<br />https://github.com/painebenjamin/taproot</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td>N/A</td></tr><tr><td>Minimum VRAM</td><td>N/A</td></tr></tbody></table> <h3>inception-v3</h3> <table><tbody><tr><td>Name</td><td>Inception Image Similarity (FID)</td></tr><tr><td>Author</td><td>Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens and Zbigniew Wojna<br />Google Research and University College London<br />Published in CoRR, vol. 1512.00567, “Rethinking the Inception Architecture for Computer Vision”, 2015<br />https://arxiv.org/abs/1512.00567</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-similarity-inception.fp16.safetensors" target="_blank">image-similarity-inception.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>50.28 MB</td></tr></tbody></table> <h2>text-similarity</h2> <table><tbody><tr><td>Name</td><td>Traditional Text Similarity</td></tr><tr><td>Author</td><td>Benjamin Paine<br />Taproot<br />https://github.com/painebenjamin/taproot</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td>N/A</td></tr><tr><td>Minimum VRAM</td><td>N/A</td></tr></tbody></table> <h2>speech-enhancement</h2> <h3>deep-filter-net-v3 (default)</h3> <table><tbody><tr><td>Name</td><td>DeepFilterNet V3 Speech Enhancement</td></tr><tr><td>Author</td><td>Hendrick Schröter, Tobias Rosenkranz, Alberto N. Escalante-B and Andreas Maier<br />Published in INTERSPEECH, “DeepFilterNet: Perceptually Motivated Real-Time Speech Enhancement”, 2023<br />https://arxiv.org/abs/2305.08227</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-enhancement-deep-filter-net-3.safetensors" target="_blank">speech-enhancement-deep-filter-net-3.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>87.89 MB</td></tr></tbody></table> <h2>image-interpolation</h2> <h3>film (default)</h3> <table><tbody><tr><td>Name</td><td>Frame Interpolation for Large Motion (FiLM) Image Interpolation</td></tr><tr><td>Author</td><td>Fitsum Reda, Janne Jontkanen, Eric Tabellion, Deqing Sun, Caroline Pantofaru and Brian Curless<br />Google Research and University of Washington<br />Published in ECCV, “FiLM: Frame Interpolation for Large Motion”, 2022<br />https://arxiv.org/abs/2202.04901</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-interpolation-film-net.fp16.pt" target="_blank">image-interpolation-film-net.fp16.pt</a></td></tr><tr><td>Minimum VRAM</td><td>70.00 MB</td></tr></tbody></table> <h3>rife</h3> <table><tbody><tr><td>Name</td><td>Real-Time Intermediate Flow Estimation (RIFE) Image Interpolation</td></tr><tr><td>Author</td><td>Zhewei Huang, Tianyuan Zhang, Wen Heng, Boxin Shi and Shuchang Zhou<br />Megvii Research, NERCVT, School of Computer Science, Peking University, Institute for Artificial Intelligence, Peking University and Beijing Academy of Artificial Intelligence<br />Published in ECCV, “Real-Time Intermediate Flow Estimation for Video Frame Interpolation”, 2022<br />https://arxiv.org/abs/2011.06294</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-interpolation-rife-flownet.safetensors" target="_blank">image-interpolation-rife-flownet.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>22.68 MB</td></tr></tbody></table> <h2>background-removal</h2> <h3>backgroundremover (default)</h3> <table><tbody><tr><td>Name</td><td>BackgroundRemover</td></tr><tr><td>Author</td><td>Johnathan Nader, Lucas Nestler, Dr. Tim Scarfe and Daniel Gatis<br />https://github.com/nadermx/backgroundremover</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/background-removal-u2net.safetensors" target="_blank">background-removal-u2net.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>217.62 MB</td></tr></tbody></table> <h2>super-resolution</h2> <h3>aura</h3> <table><tbody><tr><td>Name</td><td>Aura Super Resolution</td></tr><tr><td>Author</td><td>fal.ai<br />Published in fal.ai blog, “Introducing AuraSR - An open reproduction of the GigaGAN Upscaler”, 2024<br />https://blog.fal.ai/introducing-aurasr-an-open-reproduction-of-the-gigagan-upscaler-2/</td></tr><tr><td>License</td><td>CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/super-resolution-aura.fp16.safetensors" target="_blank">super-resolution-aura.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>1.24 GB</td></tr></tbody></table> <h3>aura-v2 (default)</h3> <table><tbody><tr><td>Name</td><td>Aura Super Resolution V2</td></tr><tr><td>Author</td><td>fal.ai<br />Published in fal.ai blog, “AuraSR V2”, 2024<br />https://blog.fal.ai/aurasr-v2/</td></tr><tr><td>License</td><td>CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/super-resolution-aura-v2.fp16.safetensors" target="_blank">super-resolution-aura-v2.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>1.24 GB</td></tr></tbody></table> <h2>speech-synthesis</h2> <h3>xtts-v2 (default)</h3> <table><tbody><tr><td>Name</td><td>XTTS2 Speech Synthesis</td></tr><tr><td>Author</td><td>Coqui AI<br />Published in Coqui AI Blog, “XTTS: Open Model Release Announcement”, 2023<br />https://coqui.ai/blog/tts/open_xtts</td></tr><tr><td>License</td><td>Mozilla Public License 2.0 (https://www.mozilla.org/en-US/MPL/2.0/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-xtts-v2.safetensors" target="_blank">speech-synthesis-xtts-v2.safetensors (1.87 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-xtts-v2-speakers.pth" target="_blank">speech-synthesis-xtts-v2-speakers.pth (7.75 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-xtts-v2-vocab.json" target="_blank">speech-synthesis-xtts-v2-vocab.json (361.22 KB)</a></li></ol><p><strong>Total Size</strong>: 1.88 GB</p></td></tr><tr><td>Minimum VRAM</td><td>1.91 GB</td></tr></tbody></table> <h3>f5tts</h3> <table><tbody><tr><td>Name</td><td>F5TTS Speech Synthesis</td></tr><tr><td>Author</td><td>Yushen Chen, Zhikang Niu, Ziyang Ma, Keqi Deng, Chunhui Wang, Jian Zhao, Kai Yu and Xie Chen<br />Published in arXiv, vol. 2410.06885, “F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching”, 2024<br />https://arxiv.org/abs/2410.06885</td></tr><tr><td>License</td><td>CC BY-NC 4.0 (https://creativecommons.org/licenses/by-nc/4.0/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-f5tts.safetensors" target="_blank">speech-synthesis-f5tts.safetensors (1.35 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-f5tts-vocab.txt" target="_blank">speech-synthesis-f5tts-vocab.txt (11.26 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-vocoder-vocos-mel-24khz.safetensors" target="_blank">audio-vocoder-vocos-mel-24khz.safetensors (54.35 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-vocoder-vocos-mel-24khz-config.yaml" target="_blank">audio-vocoder-vocos-mel-24khz-config.yaml (461.00 B)</a></li></ol><p><strong>Total Size</strong>: 1.40 GB</p></td></tr><tr><td>Minimum VRAM</td><td>3.94 GB</td></tr></tbody></table> <h2>audio-transcription</h2> <h3>whisper-tiny</h3> <table><tbody><tr><td>Name</td><td>Whisper Tiny Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tiny.safetensors" target="_blank">audio-transcription-whisper-tiny.safetensors (151.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-vocab.json (835.55 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer.json" target="_blank">audio-transcription-whisper-tokenizer.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 154.92 MB</p></td></tr><tr><td>Minimum VRAM</td><td>147.85 MB</td></tr></tbody></table> <h3>whisper-base</h3> <table><tbody><tr><td>Name</td><td>Whisper Base Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-base.safetensors" target="_blank">audio-transcription-whisper-base.safetensors (290.40 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-vocab.json (835.55 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer.json" target="_blank">audio-transcription-whisper-tokenizer.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 294.27 MB</p></td></tr><tr><td>Minimum VRAM</td><td>285.74 MB</td></tr></tbody></table> <h3>whisper-small</h3> <table><tbody><tr><td>Name</td><td>Whisper Small Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-small.safetensors" target="_blank">audio-transcription-whisper-small.safetensors (967.00 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-vocab.json (835.55 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer.json" target="_blank">audio-transcription-whisper-tokenizer.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 970.86 MB</p></td></tr><tr><td>Minimum VRAM</td><td>945.03 MB</td></tr></tbody></table> <h3>whisper-medium</h3> <table><tbody><tr><td>Name</td><td>Whisper Medium Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-medium.safetensors" target="_blank">audio-transcription-whisper-medium.safetensors (3.06 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-vocab.json (835.55 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer.json" target="_blank">audio-transcription-whisper-tokenizer.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 3.06 GB</p></td></tr><tr><td>Minimum VRAM</td><td>3.06 GB</td></tr></tbody></table> <h3>whisper-large-v3</h3> <table><tbody><tr><td>Name</td><td>Whisper Large V3 Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-large-v3.fp16.safetensors" target="_blank">audio-transcription-whisper-large-v3.fp16.safetensors (3.09 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-v3-vocab.json (1.04 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-v3-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-v3-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3.json" target="_blank">audio-transcription-whisper-tokenizer-v3.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 3.09 GB</p></td></tr><tr><td>Minimum VRAM</td><td>3.09 GB</td></tr></tbody></table> <h3>distilled-whisper-small-english</h3> <table><tbody><tr><td>Name</td><td>Distilled Whisper Small (English) Audio Transcription</td></tr><tr><td>Author</td><td>Sanchit Gandhi, Patrick von Platen and Alexander M. Rush<br />Hugging Face<br />Published in arXiv, vol. 2311.00430, “Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling”, 2023<br />https://arxiv.org/abs/2311.00430</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-small-english.safetensors" target="_blank">audio-transcription-distilled-whisper-small-english.safetensors (332.30 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-vocab.json" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-vocab.json (999.19 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-merges.txt" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-merges.txt (456.32 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-normalizer.json" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distillled-whisper-english-tokenizer.json" target="_blank">audio-transcription-distillled-whisper-english-tokenizer.json (2.41 MB)</a></li></ol><p><strong>Total Size</strong>: 336.21 MB</p></td></tr><tr><td>Minimum VRAM</td><td>649.01 MB</td></tr></tbody></table> <h3>distilled-whisper-medium-english</h3> <table><tbody><tr><td>Name</td><td>Distilled Whisper Medium (English) Audio Transcription</td></tr><tr><td>Author</td><td>Sanchit Gandhi, Patrick von Platen and Alexander M. Rush<br />Hugging Face<br />Published in arXiv, vol. 2311.00430, “Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling”, 2023<br />https://arxiv.org/abs/2311.00430</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-medium-english.safetensors" target="_blank">audio-transcription-distilled-whisper-medium-english.safetensors (788.80 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-vocab.json" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-vocab.json (999.19 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-merges.txt" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-merges.txt (456.32 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-normalizer.json" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distillled-whisper-english-tokenizer.json" target="_blank">audio-transcription-distillled-whisper-english-tokenizer.json (2.41 MB)</a></li></ol><p><strong>Total Size</strong>: 792.71 MB</p></td></tr><tr><td>Minimum VRAM</td><td>1.58 GB</td></tr></tbody></table> <h3>distilled-whisper-large-v3 (default)</h3> <table><tbody><tr><td>Name</td><td>Distilled Whisper Large V3 Audio Transcription</td></tr><tr><td>Author</td><td>Sanchit Gandhi, Patrick von Platen and Alexander M. Rush<br />Hugging Face<br />Published in arXiv, vol. 2311.00430, “Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling”, 2023<br />https://arxiv.org/abs/2311.00430</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-large-v3.fp16.safetensors" target="_blank">audio-transcription-distilled-whisper-large-v3.fp16.safetensors (1.51 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-v3-vocab.json (1.04 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-v3-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-v3-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3.json" target="_blank">audio-transcription-whisper-tokenizer-v3.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 1.52 GB</p></td></tr><tr><td>Minimum VRAM</td><td>1.51 GB</td></tr></tbody></table> <h3>turbo-whisper-large-v3</h3> <table><tbody><tr><td>Name</td><td>Turbo Whisper Large V3 Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-large-v3-turbo.fp16.safetensors" target="_blank">audio-transcription-whisper-large-v3-turbo.fp16.safetensors (1.62 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-v3-vocab.json (1.04 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-v3-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-v3-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3.json" target="_blank">audio-transcription-whisper-tokenizer-v3.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 1.62 GB</p></td></tr><tr><td>Minimum VRAM</td><td>1.62 GB</td></tr></tbody></table> <h2>depth-detection</h2> <h3>midas (default)</h3> <table><tbody><tr><td>Name</td><td>MiDaS Depth Detection</td></tr><tr><td>Author</td><td>René Ranftl, Alexey Bochkovskiy and Vladlen Koltun<br />Published in arXiv, vol. 2103.13413, “Vision Transformers for Dense Prediction”, 2021<br />https://arxiv.org/abs/2103.13413</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/depth-detection-midas.fp16.safetensors" target="_blank">depth-detection-midas.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>255.65 MB</td></tr></tbody></table> <h2>line-detection</h2> <h3>informative-drawings (default)</h3> <table><tbody><tr><td>Name</td><td>Informative Drawings Line Art Detection</td></tr><tr><td>Author</td><td>Caroline Chan, Fredo Durand and Phillip Isola<br />Massachusetts Institute of Technology<br />Published in arXiv, vol. 2203.12691, “Informative Drawings: Learning to Generate Line Drawings that Convey Geometry and Semantics”, 2022<br />https://arxiv.org/abs/2203.12691</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/line-detection-informative-drawings.fp16.safetensors" target="_blank">line-detection-informative-drawings.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>8.58 MB</td></tr></tbody></table> <h3>informative-drawings-coarse</h3> <table><tbody><tr><td>Name</td><td>Informative Drawings Coarse Line Art Detection</td></tr><tr><td>Author</td><td>Caroline Chan, Fredo Durand and Phillip Isola<br />Massachusetts Institute of Technology<br />Published in arXiv, vol. 2203.12691, “Informative Drawings: Learning to Generate Line Drawings that Convey Geometry and Semantics”, 2022<br />https://arxiv.org/abs/2203.12691</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/line-detection-informative-drawings-coarse.fp16.safetensors" target="_blank">line-detection-informative-drawings-coarse.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>8.58 MB</td></tr></tbody></table> <h3>informative-drawings-anime</h3> <table><tbody><tr><td>Name</td><td>Informative Drawings Anime Line Art Detection</td></tr><tr><td>Author</td><td>Caroline Chan, Fredo Durand and Phillip Isola<br />Massachusetts Institute of Technology<br />Published in arXiv, vol. 2203.12691, “Informative Drawings: Learning to Generate Line Drawings that Convey Geometry and Semantics”, 2022<br />https://arxiv.org/abs/2203.12691</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/line-detection-informative-drawings-anime.fp16.safetensors" target="_blank">line-detection-informative-drawings-anime.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>108.81 MB</td></tr></tbody></table> <h3>mlsd</h3> <table><tbody><tr><td>Name</td><td>Mobile Line Segment Detection</td></tr><tr><td>Author</td><td>Geonmo Gu, Byungsoo Ko, SeongHyun Go, Sung-Hyun Lee, Jingeun Lee and Minchul Shin<br />NAVER/LINE Vision<br />Published in arXiv, vol. 2106.00186, “Towards Light-weight and Real-time Line Segment Detection”, 2022<br />https://arxiv.org/abs/2106.00186</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/line-detection-mlsd.fp16.safetensors" target="_blank">line-detection-mlsd.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>3.22 MB</td></tr></tbody></table> <h2>edge-detection</h2> <h3>canny (default)</h3> <table><tbody><tr><td>Name</td><td>Canny Edge Detection</td></tr><tr><td>Author</td><td>John Canny<br />Published in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 6, pp. 679-698, “A Computational Approach to Edge Detection”, 1986<br />https://ieeexplore.ieee.org/document/4767851<br />Implementation by OpenCV (https://opencv.org/)</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td>N/A</td></tr><tr><td>Minimum VRAM</td><td>N/A</td></tr></tbody></table> <h3>hed</h3> <table><tbody><tr><td>Name</td><td>Holistically-Nested Edge Detection</td></tr><tr><td>Author</td><td>Saining Xieand Zhuowen Tu<br />University of California, San Diego<br />Published in arXiv, vol. 1504.06375, “Holistically-Nested Edge Detection”, 2015<br />https://arxiv.org/abs/1504.06375</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/edge-detection-hed.fp16.safetensors" target="_blank">edge-detection-hed.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>29.44 MB</td></tr></tbody></table> <h3>pidi</h3> <table><tbody><tr><td>Name</td><td>Soft Edge (PIDI) Detection</td></tr><tr><td>Author</td><td>Zhuo Su, Wenzhe Liu, Zitong Yu, Dewen Hu, Qing Liao, Qi Tian, Matti Pietikäinen and Li Liu<br />Published in Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 5117-5127, “Pixel Difference Networks for Efficient Edge Detection”, 2021</td></tr><tr><td>License</td><td>MIT License with Non-Commercial Clause (https://github.com/hellozhuo/pidinet/blob/master/LICENSE)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/edge-detection-pidi.fp16.safetensors" target="_blank">edge-detection-pidi.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>1.40 MB</td></tr></tbody></table> <h2>pose-detection</h2> <h3>openpose</h3> <table><tbody><tr><td>Name</td><td>OpenPose Pose Detection</td></tr><tr><td>Author</td><td>Zhe Cao, Gines Hidalgo, Tomas Simon, Shih-En Wei and Yaser Sheikh<br />Published in arXiv, vol. 1812.08008, “OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields”, 2018<br />https://arxiv.org/abs/1812.08008</td></tr><tr><td>License</td><td>OpenPose Academic or Non-Profit Non-Commercial Research License (https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/LICENSE)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/pose-detection-openpose.fp16.safetensors" target="_blank">pose-detection-openpose.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>259.96 MB</td></tr></tbody></table> <h3>dwpose (default)</h3> <table><tbody><tr><td>Name</td><td>DWPose Pose Detection</td></tr><tr><td>Author</td><td>Zhengdong Yang, Ailing Zeng, Chun Yuan and Yu Li<br />Tsinghua Zhenzhen International Graduate School and International Digital Economy Academy (IDEA)<br />Published in arXiv, vol. 2307.15880, “Effective Whole-body Pose Estimation with Two-stages Distillation”, 2023<br />https://arxiv.org/abs/2307.15880</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/pose-detection-dwpose-estimation.safetensors" target="_blank">pose-detection-dwpose-estimation.safetensors (134.65 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/pose-detection-dwpose-detection.safetensors" target="_blank">pose-detection-dwpose-detection.safetensors (217.20 MB)</a></li></ol><p><strong>Total Size</strong>: 351.85 MB</p></td></tr><tr><td>Minimum VRAM</td><td>354.64 MB</td></tr></tbody></table> <h2>image-generation</h2> <h3>stable-diffusion-v1-5</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion v1.5 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://bigscience.huggingface.co/blog/bigscience-openrail-m)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-abyssorange-mix-v3</h3> <table><tbody><tr><td>Name</td><td>AbyssOrange Mix V3 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by liudinglin (https://civitai.com/user/liudinglin)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/17233)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-abyssorange-mix-v3-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-abyssorange-mix-v3-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-abyssorange-mix-v3-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-abyssorange-mix-v3-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-chillout-mix-ni</h3> <table><tbody><tr><td>Name</td><td>Chillout Mix Ni Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Dreamlike Art (https://dreamlike.art)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://huggingface.co/dreamlike-art/dreamlike-diffusion-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-chillout-mix-ni-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-chillout-mix-ni-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-chillout-mix-ni-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-chillout-mix-ni-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-clarity-v3</h3> <table><tbody><tr><td>Name</td><td>Clarity V3 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by ndimensional (https://civitai.com/user/ndimensional)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/142125)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-clarity-v3-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-clarity-v3-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-clarity-v3-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-clarity-v3-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-dark-sushi-mix-v2-25d</h3> <table><tbody><tr><td>Name</td><td>Dark Sushi Mix V2 2.5D Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Aitasai (https://civitai.com/user/Aitasai)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/93208)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-dark-sushi-mix-v2-25d-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-dark-sushi-mix-v2-25d-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-dark-sushi-mix-v2-25d-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-dark-sushi-mix-v2-25d-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-divine-elegance-mix-v10</h3> <table><tbody><tr><td>Name</td><td>Divine Elegance Mix V10 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by TroubleDarkness (https://civitai.com/user/TroubleDarkness)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/432048)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-divine-elegance-mix-v10-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-divine-elegance-mix-v10-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-divine-elegance-mix-v10-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-divine-elegance-mix-v10-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-dreamshaper-v8</h3> <table><tbody><tr><td>Name</td><td>DreamShaper V8 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Lykon (https://civitai.com/user/Lykon)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/128713)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-dreamshaper-v8-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-dreamshaper-v8-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-dreamshaper-v8-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-dreamshaper-v8-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-epicrealism-v5</h3> <table><tbody><tr><td>Name</td><td>epiCRealism V5 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by epinikion (https://civitai.com/user/epinikion)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/143906)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-epicrealism-v5-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-epicrealism-v5-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-epicrealism-v5-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-epicrealism-v5-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-epicphotogasm-ultimate-fidelity</h3> <table><tbody><tr><td>Name</td><td>epiCPhotoGasm Ultimate Fidelity Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by epinikion (https://civitai.com/user/epinikion)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/429454)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-epic-photogasm-ultimate-fidelity-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-epic-photogasm-ultimate-fidelity-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-epic-photogasm-ultimate-fidelity-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-epic-photogasm-ultimate-fidelity-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-ghostmix-v2</h3> <table><tbody><tr><td>Name</td><td>GhostMix V2 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by _GhostInShell_ (https://civitai.com/user/_GhostInShell_)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/76907)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-ghostmix-v2-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-ghostmix-v2-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-ghostmix-v2-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-ghostmix-v2-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-lyriel-v1-6</h3> <table><tbody><tr><td>Name</td><td>Lyriel V1.6 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Lyriel (https://civitai.com/user/Lyriel)</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://civitai.com/models/license/72396)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-lyriel-v1-6-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-lyriel-v1-6-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-lyriel-v1-6-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-lyriel-v1-6-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-majicmix-realistic-v7</h3> <table><tbody><tr><td>Name</td><td>MajicMix Realistic V7 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Merjic (https://civitai.com/user/Merjic)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/176425)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-majicmix-realistic-v7-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-majicmix-realistic-v7-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-majicmix-realistic-v7-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-majicmix-realistic-v7-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-meinamix-v12</h3> <table><tbody><tr><td>Name</td><td>MeinaMix V12 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Meina (https://civitai.com/user/Meina)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/948574)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-meinamix-v12-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-meinamix-v12-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-meinamix-v12-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-meinamix-v12-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-mistoon-anime-v3</h3> <table><tbody><tr><td>Name</td><td>Mistoon Anime V3 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Inzaniak (https://civitai.com/user/Inzaniak)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/348981)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-mistoon-anime-v3-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-mistoon-anime-v3-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-mistoon-anime-v3-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-mistoon-anime-v3-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-perfect-world-v6</h3> <table><tbody><tr><td>Name</td><td>Perfect World V6 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Bloodsuga (https://civitai.com/user/Bloodsuga)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/179446)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-perfect-world-v6-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-perfect-world-v6-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-perfect-world-v6-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-perfect-world-v6-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-photon-v1</h3> <table><tbody><tr><td>Name</td><td>Photon V1 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Photographer (https://civitai.com/user/Photographer)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/900072)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-photon-v1-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-photon-v1-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-photon-v1-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-photon-v1-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-realcartoon3d-v17</h3> <table><tbody><tr><td>Name</td><td>RealCartoon3D V17 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by 7whitefire7 (https://civitai.com/user/7whitefire7)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/637156)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realcartoon3d-v17-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realcartoon3d-v17-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realcartoon3d-v17-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realcartoon3d-v17-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-realistic-vision-v5-1</h3> <table><tbody><tr><td>Name</td><td>Realistic Vision V5.1 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by SG_161222 (https://civitai.com/user/SG_161222)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/130072)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realistic-vision-v5-1-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realistic-vision-v5-1-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realistic-vision-v5-1-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realistic-vision-v5-1-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-realistic-vision-v6-0</h3> <table><tbody><tr><td>Name</td><td>Realistic Vision V6.0 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by SG_161222 (https://civitai.com/user/SG_161222)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/245592)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realistic-vision-v6-0-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realistic-vision-v6-0-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realistic-vision-v6-0-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realistic-vision-v6-0-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-rev-animated-v2</h3> <table><tbody><tr><td>Name</td><td>ReV Animated V2 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Zovya (https://civitai.com/user/Zovya)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/425083)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-rev-animated-v2-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-rev-animated-v2-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-rev-animated-v2-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-rev-animated-v2-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-toonyou-beta-v6</h3> <table><tbody><tr><td>Name</td><td>ToonYou Beta V6 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Bradcatt (https://civitai.com/user/Bradcatt)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/125771)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-toonyou-beta-v6-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-toonyou-beta-v6-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-toonyou-beta-v6-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-toonyou-beta-v6-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-xl</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-albedobase-v3-1</h3> <table><tbody><tr><td>Name</td><td>AlbedoBase XL V3.1 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/1041855)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-albedo-base-v3-1-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-albedo-base-v3-1-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-albedo-base-v3-1-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-albedo-base-v3-1-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-albedo-base-v3-1-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-albedo-base-v3-1-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-anything</h3> <table><tbody><tr><td>Name</td><td>Anything XL Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-anything-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-anything-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-anything-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-anything-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-anything-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-anything-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-animagine-v3-1</h3> <table><tbody><tr><td>Name</td><td>Animagine XL V3.1 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/403131)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-animagine-v3-1-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-animagine-v3-1-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-animagine-v3-1-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-animagine-v3-1-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-animagine-v3-1-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-animagine-v3-1-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-copax-timeless-v13</h3> <table><tbody><tr><td>Name</td><td>Copax TimeLess V13 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/724334)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-copax-timeless-v13-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-copax-timeless-v13-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-copax-timeless-v13-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-copax-timeless-v13-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-copax-timeless-v13-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-copax-timeless-v13-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-counterfeit-v2-5</h3> <table><tbody><tr><td>Name</td><td>CounterfeitXL V2.5 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/265012)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-counterfeit-v2-5-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-counterfeit-v2-5-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-counterfeit-v2-5-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-counterfeit-v2-5-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-counterfeit-v2-5-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-counterfeit-v2-5-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-dreamshaper-alpha-v2</h3> <table><tbody><tr><td>Name</td><td>DreamShaper XL Alpha V2 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/126688)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-helloworld-v7</h3> <table><tbody><tr><td>Name</td><td>LEOSAM's HelloWorld XL Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/570138)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-hello-world-v7-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-hello-world-v7-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-hello-world-v7-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-hello-world-v7-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-hello-world-v7-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-hello-world-v7-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-juggernaut-v11 (default)</h3> <table><tbody><tr><td>Name</td><td>Juggernaut XL V11 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/782002)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-juggernaut-v11-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-juggernaut-v11-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-juggernaut-v11-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-juggernaut-v11-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-juggernaut-v11-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-juggernaut-v11-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-lightning-8-step</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Lightning (8-Step)</td></tr><tr><td>Author</td><td>Shanchuan Lin, Anran Wang and Xiao Yang<br />ByteDance Inc.<br />Published in arXiv, vol. 2402.13929, “SDXL-Lightning: PRogressive Adversarial Diffusion Distillation”, 2024<br />https://arxiv.org/abs/2402.13929</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-lightning-unet-8-step.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-lightning-unet-8-step.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-lightning-4-step</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Lightning (4-Step)</td></tr><tr><td>Author</td><td>Shanchuan Lin, Anran Wang and Xiao Yang<br />ByteDance Inc.<br />Published in arXiv, vol. 2402.13929, “SDXL-Lightning: PRogressive Adversarial Diffusion Distillation”, 2024<br />https://arxiv.org/abs/2402.13929</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-lightning-unet-4-step.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-lightning-unet-4-step.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-lightning-2-step</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Lightning (2-Step)</td></tr><tr><td>Author</td><td>Shanchuan Lin, Anran Wang and Xiao Yang<br />ByteDance Inc.<br />Published in arXiv, vol. 2402.13929, “SDXL-Lightning: PRogressive Adversarial Diffusion Distillation”, 2024<br />https://arxiv.org/abs/2402.13929</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-lightning-unet-2-step.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-lightning-unet-2-step.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-nightvision-v9</h3> <table><tbody><tr><td>Name</td><td>NightVision XL V9 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/577919)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-nightvision-v9-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-nightvision-v9-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-nightvision-v9-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-nightvision-v9-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-nightvision-v9-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-nightvision-v9-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-realvis-v5</h3> <table><tbody><tr><td>Name</td><td>RealVisXL V5 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/789646)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-realvis-v5-0-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-realvis-v5-0-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-realvis-v5-0-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-realvis-v5-0-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-realvis-v5-0-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-realvis-v5-0-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-stoiqo-newreality-pro</h3> <table><tbody><tr><td>Name</td><td>Stoiqo New Reality XL Pro Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/690310)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-stoiqo-newreality-pro-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-stoiqo-newreality-pro-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-stoiqo-newreality-pro-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-stoiqo-newreality-pro-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-stoiqo-newreality-pro-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-stoiqo-newreality-pro-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-turbo</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Turbo Image Generation</td></tr><tr><td>Author</td><td>Axel Sauer, Dominik Lorenz, Andreas Blattmann and Robin Rombach<br />Stability AI<br />Published in Stability AI Blog, vol. 2307.01952, “Adversarial Diffusion Distillation”, 2024<br />https://stability.ai/research/adversarial-diffusion-distillation</td></tr><tr><td>License</td><td>Stability AI Community License (https://huggingface.co/stabilityai/sdxl-turbo/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-turbo-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-turbo-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-unstable-diffusers-nihilmania</h3> <table><tbody><tr><td>Name</td><td>SDXL Unstable Diffusers NihilMania Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/395107)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-zavychroma-v10</h3> <table><tbody><tr><td>Name</td><td>ZavyChromaXL V10 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/916744)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-zavychroma-v10-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-zavychroma-v10-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-zavychroma-v10-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-zavychroma-v10-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-zavychroma-v10-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-zavychroma-v10-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-medium</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion V3 (Medium) Image Generation</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-transformer.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-transformer.fp16.safetensors (4.17 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 15.50 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.86 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-5-medium</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion V3.5 (Medium) Image Generation</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-medium-transformer.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-medium-transformer.bf16.safetensors (4.94 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 16.27 GB</p></td></tr><tr><td>Minimum VRAM</td><td>18.36 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-5-large</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion V3.5 (Large) Image Generation</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-large-transformer.part-1.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-large-transformer.part-1.bf16.safetensors (9.99 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-large-transformer.part-2.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-large-transformer.part-2.bf16.safetensors (6.31 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 27.62 GB</p></td></tr><tr><td>Minimum VRAM</td><td>31.36 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-5-large-int8</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion V3.5 (Large) Image Generation (Int8)</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-large-transformer.int8.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-large-transformer.int8.bf16.safetensors (8.25 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 15.96 GB</p></td></tr><tr><td>Minimum VRAM</td><td>16.85 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-5-large-nf4</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion 3.5 (Large) Image Generation (NF4)</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-large-transformer.nf4.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-large-transformer.nf4.bf16.safetensors (4.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 12.85 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.99 GB</td></tr></tbody></table> <h3>flux-v1-dev</h3> <table><tbody><tr><td>Name</td><td>FluxDev</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-transformer.bf16.safetensors" target="_blank">image-generation-flux-v1-dev-transformer.bf16.safetensors (23.80 GB)</a></li></ol><p><strong>Total Size</strong>: 33.74 GB</p></td></tr><tr><td>Minimum VRAM</td><td>29.50 GB</td></tr></tbody></table> <h3>flux-v1-dev-int8</h3> <table><tbody><tr><td>Name</td><td>FluxDevInt8</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-transformer.int8.bf16.safetensors" target="_blank">image-generation-flux-v1-dev-transformer.int8.bf16.safetensors (11.92 GB)</a></li></ol><p><strong>Total Size</strong>: 18.24 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.22 GB</td></tr></tbody></table> <h3>flux-v1-dev-stoiqo-newreality-alpha-v2-int8</h3> <table><tbody><tr><td>Name</td><td>Stoiqo NewReality F1.D Alpha V2 (Int8) Image Generation</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-stoiqo-newreality-alpha-v2-transformer.int8.fp16.safetensors" target="_blank">image-generation-flux-v1-dev-stoiqo-newreality-alpha-v2-transformer.int8.fp16.safetensors (11.92 GB)</a></li></ol><p><strong>Total Size</strong>: 18.24 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.22 GB</td></tr></tbody></table> <h3>flux-v1-dev-nf4</h3> <table><tbody><tr><td>Name</td><td>FluxDevNF4</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-transformer.nf4.bf16.safetensors" target="_blank">image-generation-flux-v1-dev-transformer.nf4.bf16.safetensors (6.70 GB)</a></li></ol><p><strong>Total Size</strong>: 13.44 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.36 GB</td></tr></tbody></table> <h3>flux-v1-dev-stoiqo-newreality-alpha-v2-nf4</h3> <table><tbody><tr><td>Name</td><td>Stoiqo NewReality F1.D Alpha V2 (NF4) Image Generation</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-stoiqo-newreality-alpha-v2-transformer.nf4.fp16.safetensors" target="_blank">image-generation-flux-v1-dev-stoiqo-newreality-alpha-v2-transformer.nf4.fp16.safetensors (6.70 GB)</a></li></ol><p><strong>Total Size</strong>: 13.44 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.36 GB</td></tr></tbody></table> <h3>flux-v1-schnell</h3> <table><tbody><tr><td>Name</td><td>FluxSchnell</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-schnell-transformer.bf16.safetensors" target="_blank">image-generation-flux-v1-schnell-transformer.bf16.safetensors (23.78 GB)</a></li></ol><p><strong>Total Size</strong>: 33.72 GB</p></td></tr><tr><td>Minimum VRAM</td><td>29.50 GB</td></tr></tbody></table> <h3>flux-v1-schnell-int8</h3> <table><tbody><tr><td>Name</td><td>FluxSchnellInt8</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-schnell-transformer.int8.bf16.safetensors" target="_blank">image-generation-flux-v1-schnell-transformer.int8.bf16.safetensors (11.91 GB)</a></li></ol><p><strong>Total Size</strong>: 18.23 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.22 GB</td></tr></tbody></table> <h3>flux-v1-schnell-nf4</h3> <table><tbody><tr><td>Name</td><td>FluxSchnellNF4</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-schnell-transformer.nf4.bf16.safetensors" target="_blank">image-generation-flux-v1-schnell-transformer.nf4.bf16.safetensors (6.69 GB)</a></li></ol><p><strong>Total Size</strong>: 13.44 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.36 GB</td></tr></tbody></table> <h2>video-generation</h2> <h3>cogvideox-2b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 2B Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-2b.fp16.safetensors" target="_blank">video-generation-cog-transformer-2b.fp16.safetensors (3.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 13.34 GB</p></td></tr><tr><td>Minimum VRAM</td><td>13.48 GB</td></tr></tbody></table> <h3>cogvideox-2b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 2B Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-2b.int8.fp16.safetensors" target="_blank">video-generation-cog-transformer-2b.int8.fp16.safetensors (1.70 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 8.04 GB</p></td></tr><tr><td>Minimum VRAM</td><td>11.48 GB</td></tr></tbody></table> <h3>cogvideox-5b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-transformer-5b.fp16.safetensors (11.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 21.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.48 GB</td></tr></tbody></table> <h3>cogvideox-5b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-5b.int8.fp16.safetensors" target="_blank">video-generation-cog-transformer-5b.int8.fp16.safetensors (5.58 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 11.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.48 GB</td></tr></tbody></table> <h3>cogvideox-5b-nf4</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Video Generation (NF4)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-5b.nf4.fp16.safetensors" target="_blank">video-generation-cog-transformer-5b.nf4.fp16.safetensors (3.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 9.90 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.48 GB</td></tr></tbody></table> <h3>cogvideox-i2v-5b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Image-to-Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-i2v-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-i2v-transformer-5b.fp16.safetensors (11.25 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 21.21 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.48 GB</td></tr></tbody></table> <h3>cogvideox-i2v-5b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Image-to-Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-i2v-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-i2v-transformer-5b.fp16.safetensors (11.25 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 17.59 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.48 GB</td></tr></tbody></table> <h3>cogvideox-i2v-5b-nf4</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Image-to-Video Generation (NF4)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-i2v-transformer-5b.nf4.fp16.safetensors" target="_blank">video-generation-cog-i2v-transformer-5b.nf4.fp16.safetensors (3.25 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 10.01 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-5b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-v1-5-transformer-5b.fp16.safetensors (11.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 21.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-5b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-transformer-5b.int8.fp16.safetensors" target="_blank">video-generation-cog-v1-5-transformer-5b.int8.fp16.safetensors (5.59 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 11.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-5b-nf4</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Video Generation (NF4)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-transformer-5b.nf4.fp16.safetensors" target="_blank">video-generation-cog-v1-5-transformer-5b.nf4.fp16.safetensors (3.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 9.90 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-i2v-5b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Image-to-Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-i2v-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-v1-5-i2v-transformer-5b.fp16.safetensors (11.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 21.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-i2v-5b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Image-to-Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-i2v-transformer-5b.int8.fp16.safetensors" target="_blank">video-generation-cog-v1-5-i2v-transformer-5b.int8.fp16.safetensors (5.59 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 11.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-i2v-5b-nf4</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Image-to-Video Generation (NF4)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-i2v-transformer-5b.nf4.fp16.safetensors" target="_blank">video-generation-cog-v1-5-i2v-transformer-5b.nf4.fp16.safetensors (3.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 9.90 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.48 GB</td></tr></tbody></table> <h3>hunyuan</h3> <table><tbody><tr><td>Name</td><td>Hunyuan Video Generation</td></tr><tr><td>Author</td><td>Hunyuan Foundation Model Team<br />Tencent<br />Published in arXiv, vol. 2412.03603, “HunyuanVideo: A Systematic Framework for Large Video Generation Models”, 2024<br />https://arxiv.org/abs/2412.03603</td></tr><tr><td>License</td><td>Tencent Hunyuan Community License (https://github.com/Tencent/HunyuanVideo/blob/main/LICENSE.txt)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-vae.safetensors" target="_blank">video-generation-hunyuan-vae.safetensors (985.94 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-transformer.bf16.safetensors" target="_blank">video-generation-hunyuan-transformer.bf16.safetensors (25.64 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-vocab.json" target="_blank">text-encoding-llava-llama-tokenizer-vocab.json (17.21 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-special-tokens-map.json" target="_blank">text-encoding-llava-llama-tokenizer-special-tokens-map.json (577.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-text-encoder.fp16.safetensors" target="_blank">text-encoding-llava-llama-text-encoder.fp16.safetensors (15.01 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 41.90 GB</p></td></tr><tr><td>Minimum VRAM</td><td>38.30 GB</td></tr></tbody></table> <h3>hunyuan-int8</h3> <table><tbody><tr><td>Name</td><td>Hunyuan Video Generation</td></tr><tr><td>Author</td><td>Hunyuan Foundation Model Team<br />Tencent<br />Published in arXiv, vol. 2412.03603, “HunyuanVideo: A Systematic Framework for Large Video Generation Models”, 2024<br />https://arxiv.org/abs/2412.03603</td></tr><tr><td>License</td><td>Tencent Hunyuan Community License (https://github.com/Tencent/HunyuanVideo/blob/main/LICENSE.txt)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-vae.safetensors" target="_blank">video-generation-hunyuan-vae.safetensors (985.94 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-transformer.int8.bf16.safetensors" target="_blank">video-generation-hunyuan-transformer.int8.bf16.safetensors (12.84 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-vocab.json" target="_blank">text-encoding-llava-llama-tokenizer-vocab.json (17.21 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-special-tokens-map.json" target="_blank">text-encoding-llava-llama-tokenizer-special-tokens-map.json (577.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-text-encoder.int8.fp16.safetensors" target="_blank">text-encoding-llava-llama-text-encoder.int8.fp16.safetensors (8.04 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 22.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>23.30 GB</td></tr></tbody></table> <h3>hunyuan-nf4</h3> <table><tbody><tr><td>Name</td><td>Hunyuan Video Generation</td></tr><tr><td>Author</td><td>Hunyuan Foundation Model Team<br />Tencent<br />Published in arXiv, vol. 2412.03603, “HunyuanVideo: A Systematic Framework for Large Video Generation Models”, 2024<br />https://arxiv.org/abs/2412.03603</td></tr><tr><td>License</td><td>Tencent Hunyuan Community License (https://github.com/Tencent/HunyuanVideo/blob/main/LICENSE.txt)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-vae.safetensors" target="_blank">video-generation-hunyuan-vae.safetensors (985.94 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-transformer.nf4.bf16.safetensors" target="_blank">video-generation-hunyuan-transformer.nf4.bf16.safetensors (7.22 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-vocab.json" target="_blank">text-encoding-llava-llama-tokenizer-vocab.json (17.21 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-special-tokens-map.json" target="_blank">text-encoding-llava-llama-tokenizer-special-tokens-map.json (577.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-text-encoder.nf4.fp16.safetensors" target="_blank">text-encoding-llava-llama-text-encoder.nf4.fp16.safetensors (4.98 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 13.45 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.78 GB</td></tr></tbody></table> <h3>ltx (default)</h3> <table><tbody><tr><td>Name</td><td>LTX Video Generation</td></tr><tr><td>Author</td><td>Lightricks<br />https://github.com/Lightricks/LTX-Video</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://bigscience.huggingface.co/blog/bigscience-openrail-m)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-transformer.bf16.safetensors" target="_blank">video-generation-ltx-transformer.bf16.safetensors (3.85 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-vae.safetensors" target="_blank">video-generation-ltx-vae.safetensors (1.87 GB)</a></li></ol><p><strong>Total Size</strong>: 15.24 GB</p></td></tr><tr><td>Minimum VRAM</td><td>15.28 GB</td></tr></tbody></table> <h3>ltx-int8</h3> <table><tbody><tr><td>Name</td><td>LTX Video Generation</td></tr><tr><td>Author</td><td>Lightricks<br />https://github.com/Lightricks/LTX-Video</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://bigscience.huggingface.co/blog/bigscience-openrail-m)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-transformer.int8.bf16.safetensors" target="_blank">video-generation-ltx-transformer.int8.bf16.safetensors (1.93 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-vae.safetensors" target="_blank">video-generation-ltx-vae.safetensors (1.87 GB)</a></li></ol><p><strong>Total Size</strong>: 9.70 GB</p></td></tr><tr><td>Minimum VRAM</td><td>9.72 GB</td></tr></tbody></table> <h3>ltx-nf4</h3> <table><tbody><tr><td>Name</td><td>LTX Video Generation</td></tr><tr><td>Author</td><td>Lightricks<br />https://github.com/Lightricks/LTX-Video</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://bigscience.huggingface.co/blog/bigscience-openrail-m)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-transformer.nf4.bf16.safetensors" target="_blank">video-generation-ltx-transformer.nf4.bf16.safetensors (1.08 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-vae.safetensors" target="_blank">video-generation-ltx-vae.safetensors (1.87 GB)</a></li></ol><p><strong>Total Size</strong>: 9.28 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.29 GB</td></tr></tbody></table> <h3>mochi-v1</h3> <table><tbody><tr><td>Name</td><td>Mochi Video Generation</td></tr><tr><td>Author</td><td>Genmo AI<br />Published in Genmo AI Blog, “Mochi 1: A new SOTA in open-source video generation models”, 2024<br />https://www.genmo.ai/blog</td></tr><tr><td>License</td><td></td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-transformer.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-transformer.bf16.safetensors (20.06 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-vae.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-vae.bf16.safetensors (919.55 MB)</a></li></ol><p><strong>Total Size</strong>: 30.50 GB</p></td></tr><tr><td>Minimum VRAM</td><td>22.95 GB</td></tr></tbody></table> <h3>mochi-v1-int8</h3> <table><tbody><tr><td>Name</td><td>Mochi Video Generation</td></tr><tr><td>Author</td><td>Genmo AI<br />Published in Genmo AI Blog, “Mochi 1: A new SOTA in open-source video generation models”, 2024<br />https://www.genmo.ai/blog</td></tr><tr><td>License</td><td></td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-transformer.int8.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-transformer.int8.bf16.safetensors (10.04 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-vae.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-vae.bf16.safetensors (919.55 MB)</a></li></ol><p><strong>Total Size</strong>: 16.87 GB</p></td></tr><tr><td>Minimum VRAM</td><td>15.95 GB</td></tr></tbody></table> <h3>mochi-v1-nf4</h3> <table><tbody><tr><td>Name</td><td>Mochi Video Generation</td></tr><tr><td>Author</td><td>Genmo AI<br />Published in Genmo AI Blog, “Mochi 1: A new SOTA in open-source video generation models”, 2024<br />https://www.genmo.ai/blog</td></tr><tr><td>License</td><td></td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-transformer.nf4.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-transformer.nf4.bf16.safetensors (5.64 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-vae.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-vae.bf16.safetensors (919.55 MB)</a></li></ol><p><strong>Total Size</strong>: 12.89 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.41 GB</td></tr></tbody></table> <h2>text-generation</h2> <h3>llama-v3-8b</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q8-0.gguf" target="_blank">text-generation-llama-v3-8b-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.64 GB</td></tr></tbody></table> <h3>llama-v3-8b-q6-k</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q6-k.gguf" target="_blank">text-generation-llama-v3-8b-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.10 GB</td></tr></tbody></table> <h3>llama-v3-8b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q5-k-m.gguf" target="_blank">text-generation-llama-v3-8b-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.30 GB</td></tr></tbody></table> <h3>llama-v3-8b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q4-k-m.gguf" target="_blank">text-generation-llama-v3-8b-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.56 GB</td></tr></tbody></table> <h3>llama-v3-8b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q3-k-m.gguf" target="_blank">text-generation-llama-v3-8b-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.72 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q8-0.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.64 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct-q6-k</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q6-k.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.10 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q5-k-m.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.30 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q4-k-m.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.56 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q3-k-m.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.72 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q8-0.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.64 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct-q6-k (default)</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q6-k.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.10 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q5-k-m.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.30 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q4-k-m.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.56 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q3-k-m.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.72 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-f16.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-f16.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.04 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q8-0</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q8-0)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q8-0.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.02 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q6-k</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q6-k.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>4.20 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q5-k-m.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>3.90 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q4-k-m.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>3.50 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q3-k-l</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q3-K-L)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q3-k-l.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q3-k-l.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>3.10 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-f16.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-f16.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>3.60 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q8-0</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q8-0)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q8-0.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>2.43 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q6-k</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q6-k.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>2.15 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q5-k-m.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>2.02 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q4-k-m.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>1.64 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q3-k-l</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q3-K-L)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q3-k-l.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q3-k-l.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>1.58 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q8)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q8-0.gguf" target="_blank">text-generation-zephyr-alpha-7b-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.40 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha-q6-k</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q6-k.gguf" target="_blank">text-generation-zephyr-alpha-7b-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.20 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q5-k-m.gguf" target="_blank">text-generation-zephyr-alpha-7b-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.25 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q4-k-m.gguf" target="_blank">text-generation-zephyr-alpha-7b-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.30 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q3-k-m.gguf" target="_blank">text-generation-zephyr-alpha-7b-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.35 GB</td></tr></tbody></table> <h3>zephyr-7b-beta</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q8-0.gguf" target="_blank">text-generation-zephyr-beta-7b-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.40 GB</td></tr></tbody></table> <h3>zephyr-7b-beta-q6-k</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q6-k.gguf" target="_blank">text-generation-zephyr-beta-7b-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.20 GB</td></tr></tbody></table> <h3>zephyr-7b-beta-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q5-k-m.gguf" target="_blank">text-generation-zephyr-beta-7b-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.25 GB</td></tr></tbody></table> <h3>zephyr-7b-beta-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q4-k-m.gguf" target="_blank">text-generation-zephyr-beta-7b-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.30 GB</td></tr></tbody></table> <h3>zephyr-7b-beta-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q3-k-m.gguf" target="_blank">text-generation-zephyr-beta-7b-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.35 GB</td></tr></tbody></table> <h2>visual-question-answering</h2> <h3>llava-v1-5-7b</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b.fp16.gguf" target="_blank">visual-question-answering-llava-v1-5-7b.fp16.gguf (13.48 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 14.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>15.80 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q8</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q8-0) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q8-0.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q8-0.gguf (7.16 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 7.79 GB</p></td></tr><tr><td>Minimum VRAM</td><td>9.90 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q6-k</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q6-K) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q6-k.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q6-k.gguf (5.53 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 6.15 GB</p></td></tr><tr><td>Minimum VRAM</td><td>8.40 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q5-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q5-k-m.gguf (4.78 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 5.41 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.71 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q4-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q4-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q4-k-m.gguf (4.08 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 4.71 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.04 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q3-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q3-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q3-k-m.gguf (3.30 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 3.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>6.33 GB</td></tr></tbody></table> <h3>llava-v1-5-13b</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q8-0) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q8-0.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q8-0.gguf (13.83 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 14.48 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.51 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q6-k</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q6-K) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q6-k.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q6-k.gguf (10.68 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 11.32 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.54 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q5-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q5-k-m.gguf (9.23 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 9.88 GB</p></td></tr><tr><td>Minimum VRAM</td><td>13.17 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q4-0</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q4-0) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q4-0.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q4-0.gguf (7.37 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 8.01 GB</p></td></tr><tr><td>Minimum VRAM</td><td>11.48 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q5-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q5-k-m.gguf (24.32 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 25.02 GB</p></td></tr><tr><td>Minimum VRAM</td><td>24.96 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q4-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q4-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q4-k-m.gguf (20.66 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 21.36 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.88 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q3-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q3-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q3-k-m.gguf (16.65 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 17.35 GB</p></td></tr><tr><td>Minimum VRAM</td><td>18.06 GB</td></tr></tbody></table> <h3>moondream-v2 (default)</h3> <table><tbody><tr><td>Name</td><td>Moondream V2 Visual Question Answering</td></tr><tr><td>Author</td><td>Vikhyat Korrapati<br />Published in Hugging Face, vol. 10.57967/hf/3219, “Moondream2”, 2024<br />https://huggingface.co/vikhyatk/moondream2</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-moondream-v2.fp16.gguf" target="_blank">visual-question-answering-moondream-v2.fp16.gguf (2.84 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-moondream-v2-mmproj.fp16.gguf" target="_blank">image-encoding-clip-moondream-v2-mmproj.fp16.gguf (909.78 MB)</a></li></ol><p><strong>Total Size</strong>: 3.75 GB</p></td></tr><tr><td>Minimum VRAM</td><td>4.44 GB</td></tr></tbody></table> <h2>image-captioning</h2> <h3>llava-v1-5-7b</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b.fp16.gguf" target="_blank">visual-question-answering-llava-v1-5-7b.fp16.gguf (13.48 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 14.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>15.80 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q8</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q8-0) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q8-0.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q8-0.gguf (7.16 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 7.79 GB</p></td></tr><tr><td>Minimum VRAM</td><td>9.90 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q6-k</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q6-K) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q6-k.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q6-k.gguf (5.53 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 6.15 GB</p></td></tr><tr><td>Minimum VRAM</td><td>8.40 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q5-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q5-k-m.gguf (4.78 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 5.41 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.71 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q4-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q4-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q4-k-m.gguf (4.08 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 4.71 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.04 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q3-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q3-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q3-k-m.gguf (3.30 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 3.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>6.33 GB</td></tr></tbody></table> <h3>llava-v1-5-13b</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q8-0) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q8-0.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q8-0.gguf (13.83 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 14.48 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.51 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q6-k</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q6-K) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q6-k.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q6-k.gguf (10.68 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 11.32 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.54 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q5-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q5-k-m.gguf (9.23 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 9.88 GB</p></td></tr><tr><td>Minimum VRAM</td><td>13.17 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q4-0</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q4-0) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q4-0.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q4-0.gguf (7.37 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 8.01 GB</p></td></tr><tr><td>Minimum VRAM</td><td>11.48 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q5-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q5-k-m.gguf (24.32 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 25.02 GB</p></td></tr><tr><td>Minimum VRAM</td><td>24.96 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q4-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q4-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q4-k-m.gguf (20.66 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 21.36 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.88 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q3-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q3-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q3-k-m.gguf (16.65 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 17.35 GB</p></td></tr><tr><td>Minimum VRAM</td><td>18.06 GB</td></tr></tbody></table> <h3>moondream-v2 (default)</h3> <table><tbody><tr><td>Name</td><td>Moondream V2 Image Captioning</td></tr><tr><td>Author</td><td>Vikhyat Korrapati<br />Published in Hugging Face, vol. 10.57967/hf/3219, “Moondream2”, 2024<br />https://huggingface.co/vikhyatk/moondream2</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-moondream-v2.fp16.gguf" target="_blank">visual-question-answering-moondream-v2.fp16.gguf (2.84 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-moondream-v2-mmproj.fp16.gguf" target="_blank">image-encoding-clip-moondream-v2-mmproj.fp16.gguf (909.78 MB)</a></li></ol><p><strong>Total Size</strong>: 3.75 GB</p></td></tr><tr><td>Minimum VRAM</td><td>4.44 GB</td></tr></tbody></table>	null	Non_BioNLP	<div align="center"> <img src="https://github.com/user-attachments/assets/f965fd42-2a95-4552-9b5f-465fc4037a91" width="650" /><br /> <em>An open source real-time AI inference engine for seamless scaling</em> </div> # About Taproot is a seamlessly scalable AI/ML inference engine designed for deployment across hardware clusters with disparate capabilities. ## Why Taproot? Most AI/ML inference engines are built for either large-scale cloud infrastructures or constrained edge devices - Taproot is designed for medium-scale deployments, offering flexible and distributed on-premise or PAYG setups. It efficiently uses older or consumer-grade hardware, making it suitable for small networks or ad-hoc clusters, without relying on centralized, hyperscale architectures. ## Available Models There are more than 150 models available across 18 task categories. See the [Task Catalog](#task-catalog) for the complete list, licenses, requirements and citations. Despite the large number of models available, there are many more yet to be added - if you're looking for a particular enhancement, don't hesitate to make an issue on this repository to request it. ### Roadmap 1. IP Adapter Models for Diffusers Image Generation Pipelines 2. ControlNet Models for Diffusers Image Generation Pipelines 3. Additional quantization backends for large models - Currently BitsandBytes (Int8/NF4) and GGUF (through llama.cpp) are supported with pre-quantized checkpoints available. - FP8 support through Optimum-Quanto, TorchAO and custom kernels is in development. 4. Improved multi-GPU support - This is currently supported through manual configuration, but usability can be improved. 5. Additional annotators/detectors for image and video - E.g. Marigold, SAM2 6. Additional audio generation models - E.g. Stable Audio, AudioLDM, MusicGen # Installation ```sh pip install taproot ``` Some additional packages are available to install with the square-bracket syntax (e.g. `pip install taproot[a,b,c]`), these are: - tools - Additional packages for LLM tools like DuckDuckGo Search, BeautifulSoup (for web scraping), etc. - console - Additional packages for prettifying console output. - av - Additional packages for reading and writing video. ## Installing Tasks Some tasks are available immediately, but most tasks required additional packages and files. Install these tasks with `taproot install [task:model]+`, e.g: ```sh taproot install image-generation:stable-diffusion-xl ``` # Usage ## Command-Line ### Introspecting Tasks From the command line, execute `taproot tasks` to see all tasks and their availability status, or `taproot info` for individual task information. For example: ```sh taproot info image-generation stable-diffusion-xl Stable Diffusion XL Image Generation (image-generation:stable-diffusion-xl, available) Generate an image from text and/or images using a stable diffusion XL model. Hardware Requirements: GPU Required for Optimal Performance Floating Point Precision: half Minimum Memory (CPU RAM) Required: 231.71 MB Minimum Memory (GPU VRAM) Required: 7.58 GB Author: Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023 https://arxiv.org/abs/2307.01952 License: OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md) ✅ Attribution Required ✅ Derivatives Allowed ✅ Redistribution Allowed ✅ Copyleft (Share-Alike) Required ✅ Commercial Use Allowed ✅ Hosting Allowed Files: image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB) [downloaded] image-generation-stable-diffusion-xl-base-unet.fp16.safetensors (5.14 GB) [downloaded] text-encoding-clip-vit-l.bf16.safetensors (246.14 MB) [downloaded] text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB) [downloaded] text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB) [downloaded] text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B) [downloaded] text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB) [downloaded] text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB) [downloaded] text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B) [downloaded] text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB) [downloaded] Total File Size: 7.11 GB Required packages: pil~=9.5 [installed] torch<2.5,>=2.4 [installed] numpy~=1.22 [installed] diffusers>=0.29 [installed] torchvision<0.20,>=0.19 [installed] transformers>=4.41 [installed] safetensors~=0.4 [installed] accelerate~=1.0 [installed] sentencepiece~=0.2 [installed] compel~=2.0 [installed] peft~=0.13 [installed] Signature: prompt: Union[str, List[str]], required prompt_2: Union[str, List[str]], default: None negative_prompt: Union[str, List[str]], default: None negative_prompt_2: Union[str, List[str]], default: None image: ImageType, default: None mask_image: ImageType, default: None guidance_scale: float, default: 5.0 guidance_rescale: float, default: 0.0 num_inference_steps: int, default: 20 num_images_per_prompt: int, default: 1 height: int, default: None width: int, default: None timesteps: List[int], default: None sigmas: List[float], default: None denoising_end: float, default: None strength: float, default: None latents: torch.Tensor, default: None prompt_embeds: torch.Tensor, default: None negative_prompt_embeds: torch.Tensor, default: None pooled_prompt_embeds: torch.Tensor, default: None negative_pooled_prompt_embeds: torch.Tensor, default: None clip_skip: int, default: None seed: SeedType, default: None pag_scale: float, default: None pag_adaptive_scale: float, default: None scheduler: Literal[ddim, ddpm, ddpm_wuerstchen, deis_multistep, dpm_cogvideox, dpmsolver_multistep, dpmsolver_multistep_karras, dpmsolver_sde, dpmsolver_sde_multistep, dpmsolver_sde_multistep_karras, dpmsolver_singlestep, dpmsolver_singlestep_karras, edm_dpmsolver_multistep, edm_euler, euler_ancestral_discrete, euler_discrete, euler_discrete_karras, flow_match_euler_discrete, flow_match_heun_discrete, heun_discrete, ipndm, k_dpm_2_ancestral_discrete, k_dpm_2_ancestral_discrete_karras, k_dpm_2_discrete, k_dpm_2_discrete_karras, lcm, lms_discrete, lms_discrete_karras, pndm, tcd, unipc], default: None output_format: Literal[png, jpeg, float, int, latent], default: png output_upload: bool, default: False highres_fix_factor: float, default: 1.0 highres_fix_strength: float, default: None spatial_prompts: SpatialPromptInputType, default: None Returns: ImageResultType ``` ### Invoking Tasks Run `taproot invoke` to run any task from the command line. All parameters to the task can be passed as flags to the call using kebab-case, e.g.: ```sh taproot invoke image-generation:stable-diffusion-xl \ --prompt "a photograph of a golden retriever at the park" \ --negative-prompt "fall, autumn, blurry, out-of-focus" \ --seed 12345 Loading task. 100%\|███████████████████████████████████████████████████████████████████████████\| 7/7 [00:03<00:00, 2.27it/s] Task loaded in 4.0 s. Invoking task. 100%\|█████████████████████████████████████████████████████████████████████████\| 20/20 [00:04<00:00, 4.34it/s] Task invoked in 6.5 s. Result: 8940aa12-66a7-4233-bfd6-f19da339b71b.png ``` ## Python ### Direct Task Usage ```py from taproot import Task sdxl = Task.get("image-generation", "stable-diffusion-xl") pipeline = sdxl() pipeline.load() pipeline(prompt="Hello, world!").save("./output.png") ``` ### With a Remote Server ```py from taproot import Tap tap = Tap() tap.remote_address = "ws://127.0.0.1:32189" result = tap.call("image-generation", model="stable-diffusion-xl", prompt="Hello, world!") result.save("./output.png") ``` ### With a Local Server Also shows asynchronous usage. ```py import asyncio from taproot import Tap with Tap.local() as tap: loop = asyncio.get_event_loop() result = loop.run_until_complete(tap("image-generation", model="stable-diffusion-xl", prompt="Hello, world!")) result.save("./output.png") ``` ## Running Servers Taproot uses a three-roled cluster structure: 1. Overseers are entry points into clusters, routing requests to one or more dispatchers. 2. Dispatchers are machines capable of running tasks by spawning executors. 3. Executors are servers ready to execute a task. The simplest way to run a server is to run an overseer simultaneously with a local dispatcher like so: ```sh taproot overseer --local ``` This will run on the default address of `ws://127.0.0.1:32189`, suitable for interaction from python or the browser. There are many deployment possibilities across networks, with configuration available for encryption, listening addresses, and more. See the wiki for details (coming soon.) ## Outside Python - [taproot.js](https://github.com/painebenjamin/taproot.js) - for the browser and node.js, available in ESM, UMD and IIFE - taproot.php - coming soon <h1>Task Catalog</h1> <p>18 tasks available with 171 models.</p> <ul> <li><a href="#echo">echo</a>: 1 model</li> <li><a href="#image-similarity">image-similarity</a>: 2 models</li> <li><a href="#text-similarity">text-similarity</a>: 1 model</li> <li><a href="#speech-enhancement">speech-enhancement</a>: 1 model</li> <li><a href="#image-interpolation">image-interpolation</a>: 2 models</li> <li><a href="#background-removal">background-removal</a>: 1 model</li> <li><a href="#super-resolution">super-resolution</a>: 2 models</li> <li><a href="#speech-synthesis">speech-synthesis</a>: 2 models</li> <li><a href="#audio-transcription">audio-transcription</a>: 9 models</li> <li><a href="#depth-detection">depth-detection</a>: 1 model</li> <li><a href="#line-detection">line-detection</a>: 4 models</li> <li><a href="#edge-detection">edge-detection</a>: 3 models</li> <li><a href="#pose-detection">pose-detection</a>: 2 models</li> <li><a href="#image-generation">image-generation</a>: 52 models</li> <li><a href="#video-generation">video-generation</a>: 23 models</li> <li><a href="#text-generation">text-generation</a>: 37 models</li> <li><a href="#visual-question-answering">visual-question-answering</a>: 14 models</li> <li><a href="#image-captioning">image-captioning</a>: 14 models</li> </ul> <h2>echo</h2> <table><tbody><tr><td>Name</td><td>Echo</td></tr><tr><td>Author</td><td>Benjamin Paine<br />Taproot<br />https://github.com/painebenjamin/taproot</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td>N/A</td></tr><tr><td>Minimum VRAM</td><td>N/A</td></tr></tbody></table> <h2>image-similarity</h2> <h3>(default)</h3> <table><tbody><tr><td>Name</td><td>Traditional Image Similarity</td></tr><tr><td>Author</td><td>Benjamin Paine<br />Taproot<br />https://github.com/painebenjamin/taproot</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td>N/A</td></tr><tr><td>Minimum VRAM</td><td>N/A</td></tr></tbody></table> <h3>inception-v3</h3> <table><tbody><tr><td>Name</td><td>Inception Image Similarity (FID)</td></tr><tr><td>Author</td><td>Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens and Zbigniew Wojna<br />Google Research and University College London<br />Published in CoRR, vol. 1512.00567, “Rethinking the Inception Architecture for Computer Vision”, 2015<br />https://arxiv.org/abs/1512.00567</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-similarity-inception.fp16.safetensors" target="_blank">image-similarity-inception.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>50.28 MB</td></tr></tbody></table> <h2>text-similarity</h2> <table><tbody><tr><td>Name</td><td>Traditional Text Similarity</td></tr><tr><td>Author</td><td>Benjamin Paine<br />Taproot<br />https://github.com/painebenjamin/taproot</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td>N/A</td></tr><tr><td>Minimum VRAM</td><td>N/A</td></tr></tbody></table> <h2>speech-enhancement</h2> <h3>deep-filter-net-v3 (default)</h3> <table><tbody><tr><td>Name</td><td>DeepFilterNet V3 Speech Enhancement</td></tr><tr><td>Author</td><td>Hendrick Schröter, Tobias Rosenkranz, Alberto N. Escalante-B and Andreas Maier<br />Published in INTERSPEECH, “DeepFilterNet: Perceptually Motivated Real-Time Speech Enhancement”, 2023<br />https://arxiv.org/abs/2305.08227</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-enhancement-deep-filter-net-3.safetensors" target="_blank">speech-enhancement-deep-filter-net-3.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>87.89 MB</td></tr></tbody></table> <h2>image-interpolation</h2> <h3>film (default)</h3> <table><tbody><tr><td>Name</td><td>Frame Interpolation for Large Motion (FiLM) Image Interpolation</td></tr><tr><td>Author</td><td>Fitsum Reda, Janne Jontkanen, Eric Tabellion, Deqing Sun, Caroline Pantofaru and Brian Curless<br />Google Research and University of Washington<br />Published in ECCV, “FiLM: Frame Interpolation for Large Motion”, 2022<br />https://arxiv.org/abs/2202.04901</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-interpolation-film-net.fp16.pt" target="_blank">image-interpolation-film-net.fp16.pt</a></td></tr><tr><td>Minimum VRAM</td><td>70.00 MB</td></tr></tbody></table> <h3>rife</h3> <table><tbody><tr><td>Name</td><td>Real-Time Intermediate Flow Estimation (RIFE) Image Interpolation</td></tr><tr><td>Author</td><td>Zhewei Huang, Tianyuan Zhang, Wen Heng, Boxin Shi and Shuchang Zhou<br />Megvii Research, NERCVT, School of Computer Science, Peking University, Institute for Artificial Intelligence, Peking University and Beijing Academy of Artificial Intelligence<br />Published in ECCV, “Real-Time Intermediate Flow Estimation for Video Frame Interpolation”, 2022<br />https://arxiv.org/abs/2011.06294</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-interpolation-rife-flownet.safetensors" target="_blank">image-interpolation-rife-flownet.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>22.68 MB</td></tr></tbody></table> <h2>background-removal</h2> <h3>backgroundremover (default)</h3> <table><tbody><tr><td>Name</td><td>BackgroundRemover</td></tr><tr><td>Author</td><td>Johnathan Nader, Lucas Nestler, Dr. Tim Scarfe and Daniel Gatis<br />https://github.com/nadermx/backgroundremover</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/background-removal-u2net.safetensors" target="_blank">background-removal-u2net.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>217.62 MB</td></tr></tbody></table> <h2>super-resolution</h2> <h3>aura</h3> <table><tbody><tr><td>Name</td><td>Aura Super Resolution</td></tr><tr><td>Author</td><td>fal.ai<br />Published in fal.ai blog, “Introducing AuraSR - An open reproduction of the GigaGAN Upscaler”, 2024<br />https://blog.fal.ai/introducing-aurasr-an-open-reproduction-of-the-gigagan-upscaler-2/</td></tr><tr><td>License</td><td>CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/super-resolution-aura.fp16.safetensors" target="_blank">super-resolution-aura.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>1.24 GB</td></tr></tbody></table> <h3>aura-v2 (default)</h3> <table><tbody><tr><td>Name</td><td>Aura Super Resolution V2</td></tr><tr><td>Author</td><td>fal.ai<br />Published in fal.ai blog, “AuraSR V2”, 2024<br />https://blog.fal.ai/aurasr-v2/</td></tr><tr><td>License</td><td>CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/super-resolution-aura-v2.fp16.safetensors" target="_blank">super-resolution-aura-v2.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>1.24 GB</td></tr></tbody></table> <h2>speech-synthesis</h2> <h3>xtts-v2 (default)</h3> <table><tbody><tr><td>Name</td><td>XTTS2 Speech Synthesis</td></tr><tr><td>Author</td><td>Coqui AI<br />Published in Coqui AI Blog, “XTTS: Open Model Release Announcement”, 2023<br />https://coqui.ai/blog/tts/open_xtts</td></tr><tr><td>License</td><td>Mozilla Public License 2.0 (https://www.mozilla.org/en-US/MPL/2.0/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-xtts-v2.safetensors" target="_blank">speech-synthesis-xtts-v2.safetensors (1.87 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-xtts-v2-speakers.pth" target="_blank">speech-synthesis-xtts-v2-speakers.pth (7.75 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-xtts-v2-vocab.json" target="_blank">speech-synthesis-xtts-v2-vocab.json (361.22 KB)</a></li></ol><p><strong>Total Size</strong>: 1.88 GB</p></td></tr><tr><td>Minimum VRAM</td><td>1.91 GB</td></tr></tbody></table> <h3>f5tts</h3> <table><tbody><tr><td>Name</td><td>F5TTS Speech Synthesis</td></tr><tr><td>Author</td><td>Yushen Chen, Zhikang Niu, Ziyang Ma, Keqi Deng, Chunhui Wang, Jian Zhao, Kai Yu and Xie Chen<br />Published in arXiv, vol. 2410.06885, “F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching”, 2024<br />https://arxiv.org/abs/2410.06885</td></tr><tr><td>License</td><td>CC BY-NC 4.0 (https://creativecommons.org/licenses/by-nc/4.0/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-f5tts.safetensors" target="_blank">speech-synthesis-f5tts.safetensors (1.35 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/speech-synthesis-f5tts-vocab.txt" target="_blank">speech-synthesis-f5tts-vocab.txt (11.26 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-vocoder-vocos-mel-24khz.safetensors" target="_blank">audio-vocoder-vocos-mel-24khz.safetensors (54.35 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-vocoder-vocos-mel-24khz-config.yaml" target="_blank">audio-vocoder-vocos-mel-24khz-config.yaml (461.00 B)</a></li></ol><p><strong>Total Size</strong>: 1.40 GB</p></td></tr><tr><td>Minimum VRAM</td><td>3.94 GB</td></tr></tbody></table> <h2>audio-transcription</h2> <h3>whisper-tiny</h3> <table><tbody><tr><td>Name</td><td>Whisper Tiny Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tiny.safetensors" target="_blank">audio-transcription-whisper-tiny.safetensors (151.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-vocab.json (835.55 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer.json" target="_blank">audio-transcription-whisper-tokenizer.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 154.92 MB</p></td></tr><tr><td>Minimum VRAM</td><td>147.85 MB</td></tr></tbody></table> <h3>whisper-base</h3> <table><tbody><tr><td>Name</td><td>Whisper Base Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-base.safetensors" target="_blank">audio-transcription-whisper-base.safetensors (290.40 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-vocab.json (835.55 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer.json" target="_blank">audio-transcription-whisper-tokenizer.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 294.27 MB</p></td></tr><tr><td>Minimum VRAM</td><td>285.74 MB</td></tr></tbody></table> <h3>whisper-small</h3> <table><tbody><tr><td>Name</td><td>Whisper Small Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-small.safetensors" target="_blank">audio-transcription-whisper-small.safetensors (967.00 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-vocab.json (835.55 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer.json" target="_blank">audio-transcription-whisper-tokenizer.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 970.86 MB</p></td></tr><tr><td>Minimum VRAM</td><td>945.03 MB</td></tr></tbody></table> <h3>whisper-medium</h3> <table><tbody><tr><td>Name</td><td>Whisper Medium Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-medium.safetensors" target="_blank">audio-transcription-whisper-medium.safetensors (3.06 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-vocab.json (835.55 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer.json" target="_blank">audio-transcription-whisper-tokenizer.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 3.06 GB</p></td></tr><tr><td>Minimum VRAM</td><td>3.06 GB</td></tr></tbody></table> <h3>whisper-large-v3</h3> <table><tbody><tr><td>Name</td><td>Whisper Large V3 Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-large-v3.fp16.safetensors" target="_blank">audio-transcription-whisper-large-v3.fp16.safetensors (3.09 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-v3-vocab.json (1.04 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-v3-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-v3-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3.json" target="_blank">audio-transcription-whisper-tokenizer-v3.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 3.09 GB</p></td></tr><tr><td>Minimum VRAM</td><td>3.09 GB</td></tr></tbody></table> <h3>distilled-whisper-small-english</h3> <table><tbody><tr><td>Name</td><td>Distilled Whisper Small (English) Audio Transcription</td></tr><tr><td>Author</td><td>Sanchit Gandhi, Patrick von Platen and Alexander M. Rush<br />Hugging Face<br />Published in arXiv, vol. 2311.00430, “Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling”, 2023<br />https://arxiv.org/abs/2311.00430</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-small-english.safetensors" target="_blank">audio-transcription-distilled-whisper-small-english.safetensors (332.30 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-vocab.json" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-vocab.json (999.19 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-merges.txt" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-merges.txt (456.32 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-normalizer.json" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distillled-whisper-english-tokenizer.json" target="_blank">audio-transcription-distillled-whisper-english-tokenizer.json (2.41 MB)</a></li></ol><p><strong>Total Size</strong>: 336.21 MB</p></td></tr><tr><td>Minimum VRAM</td><td>649.01 MB</td></tr></tbody></table> <h3>distilled-whisper-medium-english</h3> <table><tbody><tr><td>Name</td><td>Distilled Whisper Medium (English) Audio Transcription</td></tr><tr><td>Author</td><td>Sanchit Gandhi, Patrick von Platen and Alexander M. Rush<br />Hugging Face<br />Published in arXiv, vol. 2311.00430, “Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling”, 2023<br />https://arxiv.org/abs/2311.00430</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-medium-english.safetensors" target="_blank">audio-transcription-distilled-whisper-medium-english.safetensors (788.80 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-vocab.json" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-vocab.json (999.19 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-merges.txt" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-merges.txt (456.32 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-english-tokenizer-normalizer.json" target="_blank">audio-transcription-distilled-whisper-english-tokenizer-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distillled-whisper-english-tokenizer.json" target="_blank">audio-transcription-distillled-whisper-english-tokenizer.json (2.41 MB)</a></li></ol><p><strong>Total Size</strong>: 792.71 MB</p></td></tr><tr><td>Minimum VRAM</td><td>1.58 GB</td></tr></tbody></table> <h3>distilled-whisper-large-v3 (default)</h3> <table><tbody><tr><td>Name</td><td>Distilled Whisper Large V3 Audio Transcription</td></tr><tr><td>Author</td><td>Sanchit Gandhi, Patrick von Platen and Alexander M. Rush<br />Hugging Face<br />Published in arXiv, vol. 2311.00430, “Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling”, 2023<br />https://arxiv.org/abs/2311.00430</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-distilled-whisper-large-v3.fp16.safetensors" target="_blank">audio-transcription-distilled-whisper-large-v3.fp16.safetensors (1.51 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-v3-vocab.json (1.04 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-v3-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-v3-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3.json" target="_blank">audio-transcription-whisper-tokenizer-v3.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 1.52 GB</p></td></tr><tr><td>Minimum VRAM</td><td>1.51 GB</td></tr></tbody></table> <h3>turbo-whisper-large-v3</h3> <table><tbody><tr><td>Name</td><td>Turbo Whisper Large V3 Audio Transcription</td></tr><tr><td>Author</td><td>Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey and Ilya Sutskever<br />OpenAI<br />Published in arXiv, vol. 2212.04356, “Robust Speech Recognition via Large-Scale Weak Supervision”<br />https://arxiv.org/abs/2212.04356</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-large-v3-turbo.fp16.safetensors" target="_blank">audio-transcription-whisper-large-v3-turbo.fp16.safetensors (1.62 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-vocab.json" target="_blank">audio-transcription-whisper-tokenizer-v3-vocab.json (1.04 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-merges.txt" target="_blank">audio-transcription-whisper-tokenizer-v3-merges.txt (493.87 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3-normalizer.json" target="_blank">audio-transcription-whisper-tokenizer-v3-normalizer.json (52.67 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/audio-transcription-whisper-tokenizer-v3.json" target="_blank">audio-transcription-whisper-tokenizer-v3.json (2.48 MB)</a></li></ol><p><strong>Total Size</strong>: 1.62 GB</p></td></tr><tr><td>Minimum VRAM</td><td>1.62 GB</td></tr></tbody></table> <h2>depth-detection</h2> <h3>midas (default)</h3> <table><tbody><tr><td>Name</td><td>MiDaS Depth Detection</td></tr><tr><td>Author</td><td>René Ranftl, Alexey Bochkovskiy and Vladlen Koltun<br />Published in arXiv, vol. 2103.13413, “Vision Transformers for Dense Prediction”, 2021<br />https://arxiv.org/abs/2103.13413</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/depth-detection-midas.fp16.safetensors" target="_blank">depth-detection-midas.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>255.65 MB</td></tr></tbody></table> <h2>line-detection</h2> <h3>informative-drawings (default)</h3> <table><tbody><tr><td>Name</td><td>Informative Drawings Line Art Detection</td></tr><tr><td>Author</td><td>Caroline Chan, Fredo Durand and Phillip Isola<br />Massachusetts Institute of Technology<br />Published in arXiv, vol. 2203.12691, “Informative Drawings: Learning to Generate Line Drawings that Convey Geometry and Semantics”, 2022<br />https://arxiv.org/abs/2203.12691</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/line-detection-informative-drawings.fp16.safetensors" target="_blank">line-detection-informative-drawings.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>8.58 MB</td></tr></tbody></table> <h3>informative-drawings-coarse</h3> <table><tbody><tr><td>Name</td><td>Informative Drawings Coarse Line Art Detection</td></tr><tr><td>Author</td><td>Caroline Chan, Fredo Durand and Phillip Isola<br />Massachusetts Institute of Technology<br />Published in arXiv, vol. 2203.12691, “Informative Drawings: Learning to Generate Line Drawings that Convey Geometry and Semantics”, 2022<br />https://arxiv.org/abs/2203.12691</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/line-detection-informative-drawings-coarse.fp16.safetensors" target="_blank">line-detection-informative-drawings-coarse.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>8.58 MB</td></tr></tbody></table> <h3>informative-drawings-anime</h3> <table><tbody><tr><td>Name</td><td>Informative Drawings Anime Line Art Detection</td></tr><tr><td>Author</td><td>Caroline Chan, Fredo Durand and Phillip Isola<br />Massachusetts Institute of Technology<br />Published in arXiv, vol. 2203.12691, “Informative Drawings: Learning to Generate Line Drawings that Convey Geometry and Semantics”, 2022<br />https://arxiv.org/abs/2203.12691</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/line-detection-informative-drawings-anime.fp16.safetensors" target="_blank">line-detection-informative-drawings-anime.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>108.81 MB</td></tr></tbody></table> <h3>mlsd</h3> <table><tbody><tr><td>Name</td><td>Mobile Line Segment Detection</td></tr><tr><td>Author</td><td>Geonmo Gu, Byungsoo Ko, SeongHyun Go, Sung-Hyun Lee, Jingeun Lee and Minchul Shin<br />NAVER/LINE Vision<br />Published in arXiv, vol. 2106.00186, “Towards Light-weight and Real-time Line Segment Detection”, 2022<br />https://arxiv.org/abs/2106.00186</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/line-detection-mlsd.fp16.safetensors" target="_blank">line-detection-mlsd.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>3.22 MB</td></tr></tbody></table> <h2>edge-detection</h2> <h3>canny (default)</h3> <table><tbody><tr><td>Name</td><td>Canny Edge Detection</td></tr><tr><td>Author</td><td>John Canny<br />Published in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 6, pp. 679-698, “A Computational Approach to Edge Detection”, 1986<br />https://ieeexplore.ieee.org/document/4767851<br />Implementation by OpenCV (https://opencv.org/)</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td>N/A</td></tr><tr><td>Minimum VRAM</td><td>N/A</td></tr></tbody></table> <h3>hed</h3> <table><tbody><tr><td>Name</td><td>Holistically-Nested Edge Detection</td></tr><tr><td>Author</td><td>Saining Xieand Zhuowen Tu<br />University of California, San Diego<br />Published in arXiv, vol. 1504.06375, “Holistically-Nested Edge Detection”, 2015<br />https://arxiv.org/abs/1504.06375</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/edge-detection-hed.fp16.safetensors" target="_blank">edge-detection-hed.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>29.44 MB</td></tr></tbody></table> <h3>pidi</h3> <table><tbody><tr><td>Name</td><td>Soft Edge (PIDI) Detection</td></tr><tr><td>Author</td><td>Zhuo Su, Wenzhe Liu, Zitong Yu, Dewen Hu, Qing Liao, Qi Tian, Matti Pietikäinen and Li Liu<br />Published in Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 5117-5127, “Pixel Difference Networks for Efficient Edge Detection”, 2021</td></tr><tr><td>License</td><td>MIT License with Non-Commercial Clause (https://github.com/hellozhuo/pidinet/blob/master/LICENSE)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/edge-detection-pidi.fp16.safetensors" target="_blank">edge-detection-pidi.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>1.40 MB</td></tr></tbody></table> <h2>pose-detection</h2> <h3>openpose</h3> <table><tbody><tr><td>Name</td><td>OpenPose Pose Detection</td></tr><tr><td>Author</td><td>Zhe Cao, Gines Hidalgo, Tomas Simon, Shih-En Wei and Yaser Sheikh<br />Published in arXiv, vol. 1812.08008, “OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields”, 2018<br />https://arxiv.org/abs/1812.08008</td></tr><tr><td>License</td><td>OpenPose Academic or Non-Profit Non-Commercial Research License (https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/LICENSE)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/pose-detection-openpose.fp16.safetensors" target="_blank">pose-detection-openpose.fp16.safetensors</a></td></tr><tr><td>Minimum VRAM</td><td>259.96 MB</td></tr></tbody></table> <h3>dwpose (default)</h3> <table><tbody><tr><td>Name</td><td>DWPose Pose Detection</td></tr><tr><td>Author</td><td>Zhengdong Yang, Ailing Zeng, Chun Yuan and Yu Li<br />Tsinghua Zhenzhen International Graduate School and International Digital Economy Academy (IDEA)<br />Published in arXiv, vol. 2307.15880, “Effective Whole-body Pose Estimation with Two-stages Distillation”, 2023<br />https://arxiv.org/abs/2307.15880</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/pose-detection-dwpose-estimation.safetensors" target="_blank">pose-detection-dwpose-estimation.safetensors (134.65 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/pose-detection-dwpose-detection.safetensors" target="_blank">pose-detection-dwpose-detection.safetensors (217.20 MB)</a></li></ol><p><strong>Total Size</strong>: 351.85 MB</p></td></tr><tr><td>Minimum VRAM</td><td>354.64 MB</td></tr></tbody></table> <h2>image-generation</h2> <h3>stable-diffusion-v1-5</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion v1.5 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://bigscience.huggingface.co/blog/bigscience-openrail-m)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-abyssorange-mix-v3</h3> <table><tbody><tr><td>Name</td><td>AbyssOrange Mix V3 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by liudinglin (https://civitai.com/user/liudinglin)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/17233)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-abyssorange-mix-v3-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-abyssorange-mix-v3-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-abyssorange-mix-v3-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-abyssorange-mix-v3-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-chillout-mix-ni</h3> <table><tbody><tr><td>Name</td><td>Chillout Mix Ni Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Dreamlike Art (https://dreamlike.art)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://huggingface.co/dreamlike-art/dreamlike-diffusion-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-chillout-mix-ni-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-chillout-mix-ni-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-chillout-mix-ni-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-chillout-mix-ni-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-clarity-v3</h3> <table><tbody><tr><td>Name</td><td>Clarity V3 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by ndimensional (https://civitai.com/user/ndimensional)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/142125)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-clarity-v3-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-clarity-v3-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-clarity-v3-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-clarity-v3-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-dark-sushi-mix-v2-25d</h3> <table><tbody><tr><td>Name</td><td>Dark Sushi Mix V2 2.5D Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Aitasai (https://civitai.com/user/Aitasai)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/93208)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-dark-sushi-mix-v2-25d-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-dark-sushi-mix-v2-25d-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-dark-sushi-mix-v2-25d-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-dark-sushi-mix-v2-25d-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-divine-elegance-mix-v10</h3> <table><tbody><tr><td>Name</td><td>Divine Elegance Mix V10 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by TroubleDarkness (https://civitai.com/user/TroubleDarkness)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/432048)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-divine-elegance-mix-v10-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-divine-elegance-mix-v10-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-divine-elegance-mix-v10-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-divine-elegance-mix-v10-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-dreamshaper-v8</h3> <table><tbody><tr><td>Name</td><td>DreamShaper V8 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Lykon (https://civitai.com/user/Lykon)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/128713)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-dreamshaper-v8-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-dreamshaper-v8-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-dreamshaper-v8-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-dreamshaper-v8-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-epicrealism-v5</h3> <table><tbody><tr><td>Name</td><td>epiCRealism V5 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by epinikion (https://civitai.com/user/epinikion)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/143906)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-epicrealism-v5-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-epicrealism-v5-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-epicrealism-v5-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-epicrealism-v5-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-epicphotogasm-ultimate-fidelity</h3> <table><tbody><tr><td>Name</td><td>epiCPhotoGasm Ultimate Fidelity Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by epinikion (https://civitai.com/user/epinikion)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/429454)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-epic-photogasm-ultimate-fidelity-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-epic-photogasm-ultimate-fidelity-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-epic-photogasm-ultimate-fidelity-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-epic-photogasm-ultimate-fidelity-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-ghostmix-v2</h3> <table><tbody><tr><td>Name</td><td>GhostMix V2 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by _GhostInShell_ (https://civitai.com/user/_GhostInShell_)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/76907)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-ghostmix-v2-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-ghostmix-v2-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-ghostmix-v2-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-ghostmix-v2-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-lyriel-v1-6</h3> <table><tbody><tr><td>Name</td><td>Lyriel V1.6 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Lyriel (https://civitai.com/user/Lyriel)</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://civitai.com/models/license/72396)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-lyriel-v1-6-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-lyriel-v1-6-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-lyriel-v1-6-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-lyriel-v1-6-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-majicmix-realistic-v7</h3> <table><tbody><tr><td>Name</td><td>MajicMix Realistic V7 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Merjic (https://civitai.com/user/Merjic)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/176425)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-majicmix-realistic-v7-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-majicmix-realistic-v7-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-majicmix-realistic-v7-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-majicmix-realistic-v7-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-meinamix-v12</h3> <table><tbody><tr><td>Name</td><td>MeinaMix V12 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Meina (https://civitai.com/user/Meina)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/948574)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-meinamix-v12-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-meinamix-v12-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-meinamix-v12-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-meinamix-v12-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-mistoon-anime-v3</h3> <table><tbody><tr><td>Name</td><td>Mistoon Anime V3 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Inzaniak (https://civitai.com/user/Inzaniak)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/348981)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-mistoon-anime-v3-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-mistoon-anime-v3-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-mistoon-anime-v3-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-mistoon-anime-v3-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-perfect-world-v6</h3> <table><tbody><tr><td>Name</td><td>Perfect World V6 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Bloodsuga (https://civitai.com/user/Bloodsuga)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/179446)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-perfect-world-v6-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-perfect-world-v6-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-perfect-world-v6-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-perfect-world-v6-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-photon-v1</h3> <table><tbody><tr><td>Name</td><td>Photon V1 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Photographer (https://civitai.com/user/Photographer)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/900072)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-photon-v1-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-photon-v1-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-photon-v1-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-photon-v1-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-realcartoon3d-v17</h3> <table><tbody><tr><td>Name</td><td>RealCartoon3D V17 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by 7whitefire7 (https://civitai.com/user/7whitefire7)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/637156)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realcartoon3d-v17-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realcartoon3d-v17-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realcartoon3d-v17-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realcartoon3d-v17-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-realistic-vision-v5-1</h3> <table><tbody><tr><td>Name</td><td>Realistic Vision V5.1 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by SG_161222 (https://civitai.com/user/SG_161222)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/130072)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realistic-vision-v5-1-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realistic-vision-v5-1-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realistic-vision-v5-1-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realistic-vision-v5-1-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-realistic-vision-v6-0</h3> <table><tbody><tr><td>Name</td><td>Realistic Vision V6.0 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by SG_161222 (https://civitai.com/user/SG_161222)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/245592)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realistic-vision-v6-0-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realistic-vision-v6-0-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-realistic-vision-v6-0-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-realistic-vision-v6-0-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-rev-animated-v2</h3> <table><tbody><tr><td>Name</td><td>ReV Animated V2 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Zovya (https://civitai.com/user/Zovya)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/425083)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-rev-animated-v2-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-rev-animated-v2-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-rev-animated-v2-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-rev-animated-v2-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-v1-5-toonyou-beta-v6</h3> <table><tbody><tr><td>Name</td><td>ToonYou Beta V6 Image Generation</td></tr><tr><td>Author</td><td>Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser and Björn Ommer<br />Published in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10684-10695, “High-Resolution Image Synthesis With Latent Diffusion Models”, 2022<br />https://arxiv.org/abs/2112.10752<br />Finetuned by Bradcatt (https://civitai.com/user/Bradcatt)</td></tr><tr><td>License</td><td>OpenRAIL-M License with Restrictions (https://civitai.com/models/license/125771)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-vae.fp16.safetensors (167.34 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-toonyou-beta-v6-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-toonyou-beta-v6-unet.fp16.safetensors (1.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v1-5-toonyou-beta-v6-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v1-5-toonyou-beta-v6-text-encoder.fp16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 2.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>2.58 GB</td></tr></tbody></table> <h3>stable-diffusion-xl</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-albedobase-v3-1</h3> <table><tbody><tr><td>Name</td><td>AlbedoBase XL V3.1 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/1041855)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-albedo-base-v3-1-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-albedo-base-v3-1-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-albedo-base-v3-1-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-albedo-base-v3-1-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-albedo-base-v3-1-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-albedo-base-v3-1-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-anything</h3> <table><tbody><tr><td>Name</td><td>Anything XL Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-anything-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-anything-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-anything-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-anything-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-anything-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-anything-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-animagine-v3-1</h3> <table><tbody><tr><td>Name</td><td>Animagine XL V3.1 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/403131)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-animagine-v3-1-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-animagine-v3-1-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-animagine-v3-1-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-animagine-v3-1-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-animagine-v3-1-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-animagine-v3-1-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-copax-timeless-v13</h3> <table><tbody><tr><td>Name</td><td>Copax TimeLess V13 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/724334)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-copax-timeless-v13-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-copax-timeless-v13-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-copax-timeless-v13-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-copax-timeless-v13-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-copax-timeless-v13-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-copax-timeless-v13-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-counterfeit-v2-5</h3> <table><tbody><tr><td>Name</td><td>CounterfeitXL V2.5 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/265012)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-counterfeit-v2-5-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-counterfeit-v2-5-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-counterfeit-v2-5-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-counterfeit-v2-5-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-counterfeit-v2-5-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-counterfeit-v2-5-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-dreamshaper-alpha-v2</h3> <table><tbody><tr><td>Name</td><td>DreamShaper XL Alpha V2 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/126688)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-dreamshaper-alpha-v2-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-helloworld-v7</h3> <table><tbody><tr><td>Name</td><td>LEOSAM's HelloWorld XL Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/570138)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-hello-world-v7-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-hello-world-v7-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-hello-world-v7-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-hello-world-v7-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-hello-world-v7-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-hello-world-v7-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-juggernaut-v11 (default)</h3> <table><tbody><tr><td>Name</td><td>Juggernaut XL V11 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/782002)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-juggernaut-v11-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-juggernaut-v11-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-juggernaut-v11-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-juggernaut-v11-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-juggernaut-v11-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-juggernaut-v11-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-lightning-8-step</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Lightning (8-Step)</td></tr><tr><td>Author</td><td>Shanchuan Lin, Anran Wang and Xiao Yang<br />ByteDance Inc.<br />Published in arXiv, vol. 2402.13929, “SDXL-Lightning: PRogressive Adversarial Diffusion Distillation”, 2024<br />https://arxiv.org/abs/2402.13929</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-lightning-unet-8-step.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-lightning-unet-8-step.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-lightning-4-step</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Lightning (4-Step)</td></tr><tr><td>Author</td><td>Shanchuan Lin, Anran Wang and Xiao Yang<br />ByteDance Inc.<br />Published in arXiv, vol. 2402.13929, “SDXL-Lightning: PRogressive Adversarial Diffusion Distillation”, 2024<br />https://arxiv.org/abs/2402.13929</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-lightning-unet-4-step.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-lightning-unet-4-step.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-lightning-2-step</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Lightning (2-Step)</td></tr><tr><td>Author</td><td>Shanchuan Lin, Anran Wang and Xiao Yang<br />ByteDance Inc.<br />Published in arXiv, vol. 2402.13929, “SDXL-Lightning: PRogressive Adversarial Diffusion Distillation”, 2024<br />https://arxiv.org/abs/2402.13929</td></tr><tr><td>License</td><td>OpenRAIL++-M License (https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-lightning-unet-2-step.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-lightning-unet-2-step.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-nightvision-v9</h3> <table><tbody><tr><td>Name</td><td>NightVision XL V9 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/577919)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-nightvision-v9-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-nightvision-v9-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-nightvision-v9-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-nightvision-v9-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-nightvision-v9-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-nightvision-v9-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-realvis-v5</h3> <table><tbody><tr><td>Name</td><td>RealVisXL V5 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/789646)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-realvis-v5-0-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-realvis-v5-0-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-realvis-v5-0-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-realvis-v5-0-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-realvis-v5-0-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-realvis-v5-0-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-stoiqo-newreality-pro</h3> <table><tbody><tr><td>Name</td><td>Stoiqo New Reality XL Pro Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/690310)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-stoiqo-newreality-pro-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-stoiqo-newreality-pro-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-stoiqo-newreality-pro-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-stoiqo-newreality-pro-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-stoiqo-newreality-pro-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-stoiqo-newreality-pro-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-turbo</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion XL Turbo Image Generation</td></tr><tr><td>Author</td><td>Axel Sauer, Dominik Lorenz, Andreas Blattmann and Robin Rombach<br />Stability AI<br />Published in Stability AI Blog, vol. 2307.01952, “Adversarial Diffusion Distillation”, 2024<br />https://stability.ai/research/adversarial-diffusion-distillation</td></tr><tr><td>License</td><td>Stability AI Community License (https://huggingface.co/stabilityai/sdxl-turbo/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-turbo-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-turbo-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-unstable-diffusers-nihilmania</h3> <table><tbody><tr><td>Name</td><td>SDXL Unstable Diffusers NihilMania Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/395107)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-unstable-diffusers-nihilmania-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-xl-zavychroma-v10</h3> <table><tbody><tr><td>Name</td><td>ZavyChromaXL V10 Image Generation</td></tr><tr><td>Author</td><td>Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna and Robin Rombach<br />Published in arXiv, vol. 2307.01952, “SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis”, 2023<br />https://arxiv.org/abs/2307.01952</td></tr><tr><td>License</td><td>OpenRAIL++-M License with Restrictions (https://civitai.com/models/license/916744)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-base-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-base-vae.fp16.safetensors (334.64 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-zavychroma-v10-unet.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-zavychroma-v10-unet.fp16.safetensors (5.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-zavychroma-v10-text-encoder.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-zavychroma-v10-text-encoder.fp16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-xl-zavychroma-v10-text-encoder-2.fp16.safetensors" target="_blank">image-generation-stable-diffusion-xl-zavychroma-v10-text-encoder-2.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li></ol><p><strong>Total Size</strong>: 7.11 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.06 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-medium</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion V3 (Medium) Image Generation</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-transformer.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-transformer.fp16.safetensors (4.17 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 15.50 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.86 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-5-medium</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion V3.5 (Medium) Image Generation</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-medium-transformer.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-medium-transformer.bf16.safetensors (4.94 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 16.27 GB</p></td></tr><tr><td>Minimum VRAM</td><td>18.36 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-5-large</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion V3.5 (Large) Image Generation</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-large-transformer.part-1.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-large-transformer.part-1.bf16.safetensors (9.99 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-large-transformer.part-2.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-large-transformer.part-2.bf16.safetensors (6.31 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 27.62 GB</p></td></tr><tr><td>Minimum VRAM</td><td>31.36 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-5-large-int8</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion V3.5 (Large) Image Generation (Int8)</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-large-transformer.int8.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-large-transformer.int8.bf16.safetensors (8.25 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 15.96 GB</p></td></tr><tr><td>Minimum VRAM</td><td>16.85 GB</td></tr></tbody></table> <h3>stable-diffusion-v3-5-large-nf4</h3> <table><tbody><tr><td>Name</td><td>Stable Diffusion 3.5 (Large) Image Generation (NF4)</td></tr><tr><td>Author</td><td>Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek and Robin Rombach<br />Stability AI<br />Published in arXiv, vol. 2403.03206, “Scaling Rectified Flow Transformers for High-Resolution Image Synthesis”, 2024<br />https://arxiv.org/abs/2403.03206</td></tr><tr><td>License</td><td>Stability AI Community License Agreement (https://huggingface.co/stabilityai/stable-diffusion-3-medium/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-vae.fp16.safetensors" target="_blank">image-generation-stable-diffusion-v3-vae.fp16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-stable-diffusion-v3-5-large-transformer.nf4.bf16.safetensors" target="_blank">image-generation-stable-diffusion-v3-5-large-transformer.nf4.bf16.safetensors (4.72 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g.fp16.safetensors" target="_blank">text-encoding-open-clip-vit-g.fp16.safetensors (1.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-vocab.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json" target="_blank">text-encoding-open-clip-vit-g-tokenizer-special-tokens-map.json (576.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-open-clip-vit-g-tokenizer-merges.txt" target="_blank">text-encoding-open-clip-vit-g-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li></ol><p><strong>Total Size</strong>: 12.85 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.99 GB</td></tr></tbody></table> <h3>flux-v1-dev</h3> <table><tbody><tr><td>Name</td><td>FluxDev</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-transformer.bf16.safetensors" target="_blank">image-generation-flux-v1-dev-transformer.bf16.safetensors (23.80 GB)</a></li></ol><p><strong>Total Size</strong>: 33.74 GB</p></td></tr><tr><td>Minimum VRAM</td><td>29.50 GB</td></tr></tbody></table> <h3>flux-v1-dev-int8</h3> <table><tbody><tr><td>Name</td><td>FluxDevInt8</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-transformer.int8.bf16.safetensors" target="_blank">image-generation-flux-v1-dev-transformer.int8.bf16.safetensors (11.92 GB)</a></li></ol><p><strong>Total Size</strong>: 18.24 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.22 GB</td></tr></tbody></table> <h3>flux-v1-dev-stoiqo-newreality-alpha-v2-int8</h3> <table><tbody><tr><td>Name</td><td>Stoiqo NewReality F1.D Alpha V2 (Int8) Image Generation</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-stoiqo-newreality-alpha-v2-transformer.int8.fp16.safetensors" target="_blank">image-generation-flux-v1-dev-stoiqo-newreality-alpha-v2-transformer.int8.fp16.safetensors (11.92 GB)</a></li></ol><p><strong>Total Size</strong>: 18.24 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.22 GB</td></tr></tbody></table> <h3>flux-v1-dev-nf4</h3> <table><tbody><tr><td>Name</td><td>FluxDevNF4</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-transformer.nf4.bf16.safetensors" target="_blank">image-generation-flux-v1-dev-transformer.nf4.bf16.safetensors (6.70 GB)</a></li></ol><p><strong>Total Size</strong>: 13.44 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.36 GB</td></tr></tbody></table> <h3>flux-v1-dev-stoiqo-newreality-alpha-v2-nf4</h3> <table><tbody><tr><td>Name</td><td>Stoiqo NewReality F1.D Alpha V2 (NF4) Image Generation</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-dev-stoiqo-newreality-alpha-v2-transformer.nf4.fp16.safetensors" target="_blank">image-generation-flux-v1-dev-stoiqo-newreality-alpha-v2-transformer.nf4.fp16.safetensors (6.70 GB)</a></li></ol><p><strong>Total Size</strong>: 13.44 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.36 GB</td></tr></tbody></table> <h3>flux-v1-schnell</h3> <table><tbody><tr><td>Name</td><td>FluxSchnell</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-schnell-transformer.bf16.safetensors" target="_blank">image-generation-flux-v1-schnell-transformer.bf16.safetensors (23.78 GB)</a></li></ol><p><strong>Total Size</strong>: 33.72 GB</p></td></tr><tr><td>Minimum VRAM</td><td>29.50 GB</td></tr></tbody></table> <h3>flux-v1-schnell-int8</h3> <table><tbody><tr><td>Name</td><td>FluxSchnellInt8</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-schnell-transformer.int8.bf16.safetensors" target="_blank">image-generation-flux-v1-schnell-transformer.int8.bf16.safetensors (11.91 GB)</a></li></ol><p><strong>Total Size</strong>: 18.23 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.22 GB</td></tr></tbody></table> <h3>flux-v1-schnell-nf4</h3> <table><tbody><tr><td>Name</td><td>FluxSchnellNF4</td></tr><tr><td>Author</td><td>Black Forest Labs<br />Published in Black Forest Labs Blog, “Announcing Black Forest Labs”, 2024<br />https://blackforestlabs.ai/announcing-black-forest-labs/</td></tr><tr><td>License</td><td>FLUX.1 Non-Commercial License (https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-vae.bf16.safetensors" target="_blank">image-generation-flux-v1-vae.bf16.safetensors (167.67 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-generation-flux-v1-schnell-transformer.nf4.bf16.safetensors" target="_blank">image-generation-flux-v1-schnell-transformer.nf4.bf16.safetensors (6.69 GB)</a></li></ol><p><strong>Total Size</strong>: 13.44 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.36 GB</td></tr></tbody></table> <h2>video-generation</h2> <h3>cogvideox-2b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 2B Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-2b.fp16.safetensors" target="_blank">video-generation-cog-transformer-2b.fp16.safetensors (3.39 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 13.34 GB</p></td></tr><tr><td>Minimum VRAM</td><td>13.48 GB</td></tr></tbody></table> <h3>cogvideox-2b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 2B Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-2b.int8.fp16.safetensors" target="_blank">video-generation-cog-transformer-2b.int8.fp16.safetensors (1.70 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 8.04 GB</p></td></tr><tr><td>Minimum VRAM</td><td>11.48 GB</td></tr></tbody></table> <h3>cogvideox-5b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-transformer-5b.fp16.safetensors (11.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 21.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.48 GB</td></tr></tbody></table> <h3>cogvideox-5b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-5b.int8.fp16.safetensors" target="_blank">video-generation-cog-transformer-5b.int8.fp16.safetensors (5.58 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 11.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.48 GB</td></tr></tbody></table> <h3>cogvideox-5b-nf4</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Video Generation (NF4)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-transformer-5b.nf4.fp16.safetensors" target="_blank">video-generation-cog-transformer-5b.nf4.fp16.safetensors (3.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 9.90 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.48 GB</td></tr></tbody></table> <h3>cogvideox-i2v-5b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Image-to-Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-i2v-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-i2v-transformer-5b.fp16.safetensors (11.25 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 21.21 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.48 GB</td></tr></tbody></table> <h3>cogvideox-i2v-5b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Image-to-Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-i2v-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-i2v-transformer-5b.fp16.safetensors (11.25 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 17.59 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.48 GB</td></tr></tbody></table> <h3>cogvideox-i2v-5b-nf4</h3> <table><tbody><tr><td>Name</td><td>CogVideoX 5B Image-to-Video Generation (NF4)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-i2v-transformer-5b.nf4.fp16.safetensors" target="_blank">video-generation-cog-i2v-transformer-5b.nf4.fp16.safetensors (3.25 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 10.01 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-5b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-v1-5-transformer-5b.fp16.safetensors (11.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 21.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-5b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-transformer-5b.int8.fp16.safetensors" target="_blank">video-generation-cog-v1-5-transformer-5b.int8.fp16.safetensors (5.59 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 11.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-5b-nf4</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Video Generation (NF4)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-transformer-5b.nf4.fp16.safetensors" target="_blank">video-generation-cog-v1-5-transformer-5b.nf4.fp16.safetensors (3.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 9.90 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-i2v-5b</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Image-to-Video Generation</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-i2v-transformer-5b.fp16.safetensors" target="_blank">video-generation-cog-v1-5-i2v-transformer-5b.fp16.safetensors (11.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 21.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-i2v-5b-int8</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Image-to-Video Generation (Int8)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-i2v-transformer-5b.int8.fp16.safetensors" target="_blank">video-generation-cog-v1-5-i2v-transformer-5b.int8.fp16.safetensors (5.59 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 11.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.48 GB</td></tr></tbody></table> <h3>cogvideox-v1-5-i2v-5b-nf4</h3> <table><tbody><tr><td>Name</td><td>CogVideoX V1.5 5B Image-to-Video Generation (NF4)</td></tr><tr><td>Author</td><td>Zhuoyi Yang, Jiayen Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong and Jie Tang<br />Zhipu AI and Tsinghua University<br />Published in arXiv, vol. 2408.06072, “CogVideoX: Text-to-Video Diffusion Models with an Experty Transformer”, 2024<br />https://arxiv.org/abs/2408.06072</td></tr><tr><td>License</td><td>CogVideoX License (https://huggingface.co/THUDM/CogVideoX-5b/blob/main/LICENSE)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-v1-5-i2v-transformer-5b.nf4.fp16.safetensors" target="_blank">video-generation-cog-v1-5-i2v-transformer-5b.nf4.fp16.safetensors (3.14 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-cog-vae.bf16.safetensors" target="_blank">video-generation-cog-vae.bf16.safetensors (431.22 MB)</a></li></ol><p><strong>Total Size</strong>: 9.90 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.48 GB</td></tr></tbody></table> <h3>hunyuan</h3> <table><tbody><tr><td>Name</td><td>Hunyuan Video Generation</td></tr><tr><td>Author</td><td>Hunyuan Foundation Model Team<br />Tencent<br />Published in arXiv, vol. 2412.03603, “HunyuanVideo: A Systematic Framework for Large Video Generation Models”, 2024<br />https://arxiv.org/abs/2412.03603</td></tr><tr><td>License</td><td>Tencent Hunyuan Community License (https://github.com/Tencent/HunyuanVideo/blob/main/LICENSE.txt)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-vae.safetensors" target="_blank">video-generation-hunyuan-vae.safetensors (985.94 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-transformer.bf16.safetensors" target="_blank">video-generation-hunyuan-transformer.bf16.safetensors (25.64 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-vocab.json" target="_blank">text-encoding-llava-llama-tokenizer-vocab.json (17.21 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-special-tokens-map.json" target="_blank">text-encoding-llava-llama-tokenizer-special-tokens-map.json (577.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-text-encoder.fp16.safetensors" target="_blank">text-encoding-llava-llama-text-encoder.fp16.safetensors (15.01 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 41.90 GB</p></td></tr><tr><td>Minimum VRAM</td><td>38.30 GB</td></tr></tbody></table> <h3>hunyuan-int8</h3> <table><tbody><tr><td>Name</td><td>Hunyuan Video Generation</td></tr><tr><td>Author</td><td>Hunyuan Foundation Model Team<br />Tencent<br />Published in arXiv, vol. 2412.03603, “HunyuanVideo: A Systematic Framework for Large Video Generation Models”, 2024<br />https://arxiv.org/abs/2412.03603</td></tr><tr><td>License</td><td>Tencent Hunyuan Community License (https://github.com/Tencent/HunyuanVideo/blob/main/LICENSE.txt)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-vae.safetensors" target="_blank">video-generation-hunyuan-vae.safetensors (985.94 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-transformer.int8.bf16.safetensors" target="_blank">video-generation-hunyuan-transformer.int8.bf16.safetensors (12.84 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-vocab.json" target="_blank">text-encoding-llava-llama-tokenizer-vocab.json (17.21 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-special-tokens-map.json" target="_blank">text-encoding-llava-llama-tokenizer-special-tokens-map.json (577.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-text-encoder.int8.fp16.safetensors" target="_blank">text-encoding-llava-llama-text-encoder.int8.fp16.safetensors (8.04 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 22.13 GB</p></td></tr><tr><td>Minimum VRAM</td><td>23.30 GB</td></tr></tbody></table> <h3>hunyuan-nf4</h3> <table><tbody><tr><td>Name</td><td>Hunyuan Video Generation</td></tr><tr><td>Author</td><td>Hunyuan Foundation Model Team<br />Tencent<br />Published in arXiv, vol. 2412.03603, “HunyuanVideo: A Systematic Framework for Large Video Generation Models”, 2024<br />https://arxiv.org/abs/2412.03603</td></tr><tr><td>License</td><td>Tencent Hunyuan Community License (https://github.com/Tencent/HunyuanVideo/blob/main/LICENSE.txt)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-vae.safetensors" target="_blank">video-generation-hunyuan-vae.safetensors (985.94 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-hunyuan-transformer.nf4.bf16.safetensors" target="_blank">video-generation-hunyuan-transformer.nf4.bf16.safetensors (7.22 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-vocab.json" target="_blank">text-encoding-llava-llama-tokenizer-vocab.json (17.21 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-tokenizer-special-tokens-map.json" target="_blank">text-encoding-llava-llama-tokenizer-special-tokens-map.json (577.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-vocab.json" target="_blank">text-encoding-clip-vit-l-tokenizer-vocab.json (1.06 MB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-special-tokens-map.json" target="_blank">text-encoding-clip-vit-l-tokenizer-special-tokens-map.json (588.00 B)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l-tokenizer-merges.txt" target="_blank">text-encoding-clip-vit-l-tokenizer-merges.txt (524.62 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-llava-llama-text-encoder.nf4.fp16.safetensors" target="_blank">text-encoding-llava-llama-text-encoder.nf4.fp16.safetensors (4.98 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-clip-vit-l.bf16.safetensors" target="_blank">text-encoding-clip-vit-l.bf16.safetensors (246.14 MB)</a></li></ol><p><strong>Total Size</strong>: 13.45 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.78 GB</td></tr></tbody></table> <h3>ltx (default)</h3> <table><tbody><tr><td>Name</td><td>LTX Video Generation</td></tr><tr><td>Author</td><td>Lightricks<br />https://github.com/Lightricks/LTX-Video</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://bigscience.huggingface.co/blog/bigscience-openrail-m)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-transformer.bf16.safetensors" target="_blank">video-generation-ltx-transformer.bf16.safetensors (3.85 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-vae.safetensors" target="_blank">video-generation-ltx-vae.safetensors (1.87 GB)</a></li></ol><p><strong>Total Size</strong>: 15.24 GB</p></td></tr><tr><td>Minimum VRAM</td><td>15.28 GB</td></tr></tbody></table> <h3>ltx-int8</h3> <table><tbody><tr><td>Name</td><td>LTX Video Generation</td></tr><tr><td>Author</td><td>Lightricks<br />https://github.com/Lightricks/LTX-Video</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://bigscience.huggingface.co/blog/bigscience-openrail-m)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-transformer.int8.bf16.safetensors" target="_blank">video-generation-ltx-transformer.int8.bf16.safetensors (1.93 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-vae.safetensors" target="_blank">video-generation-ltx-vae.safetensors (1.87 GB)</a></li></ol><p><strong>Total Size</strong>: 9.70 GB</p></td></tr><tr><td>Minimum VRAM</td><td>9.72 GB</td></tr></tbody></table> <h3>ltx-nf4</h3> <table><tbody><tr><td>Name</td><td>LTX Video Generation</td></tr><tr><td>Author</td><td>Lightricks<br />https://github.com/Lightricks/LTX-Video</td></tr><tr><td>License</td><td>OpenRAIL-M License (https://bigscience.huggingface.co/blog/bigscience-openrail-m)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-transformer.nf4.bf16.safetensors" target="_blank">video-generation-ltx-transformer.nf4.bf16.safetensors (1.08 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-ltx-vae.safetensors" target="_blank">video-generation-ltx-vae.safetensors (1.87 GB)</a></li></ol><p><strong>Total Size</strong>: 9.28 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.29 GB</td></tr></tbody></table> <h3>mochi-v1</h3> <table><tbody><tr><td>Name</td><td>Mochi Video Generation</td></tr><tr><td>Author</td><td>Genmo AI<br />Published in Genmo AI Blog, “Mochi 1: A new SOTA in open-source video generation models”, 2024<br />https://www.genmo.ai/blog</td></tr><tr><td>License</td><td></td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.bf16.safetensors" target="_blank">text-encoding-t5-xxl.bf16.safetensors (9.52 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-transformer.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-transformer.bf16.safetensors (20.06 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-vae.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-vae.bf16.safetensors (919.55 MB)</a></li></ol><p><strong>Total Size</strong>: 30.50 GB</p></td></tr><tr><td>Minimum VRAM</td><td>22.95 GB</td></tr></tbody></table> <h3>mochi-v1-int8</h3> <table><tbody><tr><td>Name</td><td>Mochi Video Generation</td></tr><tr><td>Author</td><td>Genmo AI<br />Published in Genmo AI Blog, “Mochi 1: A new SOTA in open-source video generation models”, 2024<br />https://www.genmo.ai/blog</td></tr><tr><td>License</td><td></td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.int8.bf16.safetensors" target="_blank">text-encoding-t5-xxl.int8.bf16.safetensors (5.90 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-transformer.int8.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-transformer.int8.bf16.safetensors (10.04 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-vae.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-vae.bf16.safetensors (919.55 MB)</a></li></ol><p><strong>Total Size</strong>: 16.87 GB</p></td></tr><tr><td>Minimum VRAM</td><td>15.95 GB</td></tr></tbody></table> <h3>mochi-v1-nf4</h3> <table><tbody><tr><td>Name</td><td>Mochi Video Generation</td></tr><tr><td>Author</td><td>Genmo AI<br />Published in Genmo AI Blog, “Mochi 1: A new SOTA in open-source video generation models”, 2024<br />https://www.genmo.ai/blog</td></tr><tr><td>License</td><td></td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-vocab.model" target="_blank">text-encoding-t5-xxl-vocab.model (791.66 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl-special-tokens-map.json" target="_blank">text-encoding-t5-xxl-special-tokens-map.json (2.54 KB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-encoding-t5-xxl.nf4.bf16.safetensors" target="_blank">text-encoding-t5-xxl.nf4.bf16.safetensors (6.33 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-transformer.nf4.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-transformer.nf4.bf16.safetensors (5.64 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/video-generation-mochi-v1-preview-vae.bf16.safetensors" target="_blank">video-generation-mochi-v1-preview-vae.bf16.safetensors (919.55 MB)</a></li></ol><p><strong>Total Size</strong>: 12.89 GB</p></td></tr><tr><td>Minimum VRAM</td><td>12.41 GB</td></tr></tbody></table> <h2>text-generation</h2> <h3>llama-v3-8b</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q8-0.gguf" target="_blank">text-generation-llama-v3-8b-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.64 GB</td></tr></tbody></table> <h3>llama-v3-8b-q6-k</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q6-k.gguf" target="_blank">text-generation-llama-v3-8b-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.10 GB</td></tr></tbody></table> <h3>llama-v3-8b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q5-k-m.gguf" target="_blank">text-generation-llama-v3-8b-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.30 GB</td></tr></tbody></table> <h3>llama-v3-8b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q4-k-m.gguf" target="_blank">text-generation-llama-v3-8b-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.56 GB</td></tr></tbody></table> <h3>llama-v3-8b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-q3-k-m.gguf" target="_blank">text-generation-llama-v3-8b-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.72 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q8-0.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.64 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct-q6-k</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q6-k.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.10 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q5-k-m.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.30 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q4-k-m.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.56 GB</td></tr></tbody></table> <h3>llama-v3-8b-instruct-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.0 8B Instruct Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-8b-instruct-q3-k-m.gguf" target="_blank">text-generation-llama-v3-8b-instruct-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.72 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q8-0.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.64 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct-q6-k (default)</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q6-k.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.10 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q5-k-m.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.30 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q4-k-m.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.56 GB</td></tr></tbody></table> <h3>llama-v3-1-8b-instruct-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.1 8B Instruct Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-1-8b-instruct-q3-k-m.gguf" target="_blank">text-generation-llama-v3-1-8b-instruct-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.72 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-f16.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-f16.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.04 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q8-0</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q8-0)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q8-0.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.02 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q6-k</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q6-k.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>4.20 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q5-k-m.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>3.90 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q4-k-m.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>3.50 GB</td></tr></tbody></table> <h3>llama-v3-2-3b-instruct-q3-k-l</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 3B Instruct Text Generation (Q3-K-L)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-3b-instruct-q3-k-l.gguf" target="_blank">text-generation-llama-v3-2-3b-instruct-q3-k-l.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>3.10 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-f16.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-f16.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>3.60 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q8-0</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q8-0)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q8-0.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>2.43 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q6-k</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q6-k.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>2.15 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q5-k-m.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>2.02 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q4-k-m.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>1.64 GB</td></tr></tbody></table> <h3>llama-v3-2-1b-instruct-q3-k-l</h3> <table><tbody><tr><td>Name</td><td>Llama V3.2 1B Instruct Text Generation (Q3-K-L)</td></tr><tr><td>Author</td><td>Meta AI<br />Published in arXiv, vol. 2407.21783, “The Llama 3 Herd of Models”, 2024<br />https://arxiv.org/abs/2407.21783</td></tr><tr><td>License</td><td>Meta Llama 3 Community License (https://www.llama.com/llama3/license/)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-llama-v3-2-1b-instruct-q3-k-l.gguf" target="_blank">text-generation-llama-v3-2-1b-instruct-q3-k-l.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>1.58 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q8)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q8-0.gguf" target="_blank">text-generation-zephyr-alpha-7b-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.40 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha-q6-k</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q6-k.gguf" target="_blank">text-generation-zephyr-alpha-7b-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.20 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q5-k-m.gguf" target="_blank">text-generation-zephyr-alpha-7b-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.25 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q4-k-m.gguf" target="_blank">text-generation-zephyr-alpha-7b-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.30 GB</td></tr></tbody></table> <h3>zephyr-7b-alpha-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B α Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-alpha-7b-q3-k-m.gguf" target="_blank">text-generation-zephyr-alpha-7b-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.35 GB</td></tr></tbody></table> <h3>zephyr-7b-beta</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q8-0.gguf" target="_blank">text-generation-zephyr-beta-7b-q8-0.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>9.40 GB</td></tr></tbody></table> <h3>zephyr-7b-beta-q6-k</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation (Q6-K)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q6-k.gguf" target="_blank">text-generation-zephyr-beta-7b-q6-k.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>8.20 GB</td></tr></tbody></table> <h3>zephyr-7b-beta-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation (Q5-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q5-k-m.gguf" target="_blank">text-generation-zephyr-beta-7b-q5-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>7.25 GB</td></tr></tbody></table> <h3>zephyr-7b-beta-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation (Q4-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q4-k-m.gguf" target="_blank">text-generation-zephyr-beta-7b-q4-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>6.30 GB</td></tr></tbody></table> <h3>zephyr-7b-beta-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>Zephyr 7B β Text Generation (Q3-K-M)</td></tr><tr><td>Author</td><td>Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sansevier, Alexander M. Rush and Thomas Wolf<br />Published in arXiv, vol. 2310.16944, “Zephyr: Direct Distillation of LM Alignment”, 2023<br />https://arxiv.org/abs/2310.16944</td></tr><tr><td>License</td><td>MIT License (https://opensource.org/licenses/MIT)</td></tr><tr><td>Files</td><td><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/text-generation-zephyr-beta-7b-q3-k-m.gguf" target="_blank">text-generation-zephyr-beta-7b-q3-k-m.gguf</a></td></tr><tr><td>Minimum VRAM</td><td>5.35 GB</td></tr></tbody></table> <h2>visual-question-answering</h2> <h3>llava-v1-5-7b</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b.fp16.gguf" target="_blank">visual-question-answering-llava-v1-5-7b.fp16.gguf (13.48 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 14.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>15.80 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q8</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q8-0) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q8-0.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q8-0.gguf (7.16 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 7.79 GB</p></td></tr><tr><td>Minimum VRAM</td><td>9.90 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q6-k</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q6-K) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q6-k.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q6-k.gguf (5.53 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 6.15 GB</p></td></tr><tr><td>Minimum VRAM</td><td>8.40 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q5-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q5-k-m.gguf (4.78 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 5.41 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.71 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q4-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q4-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q4-k-m.gguf (4.08 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 4.71 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.04 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q3-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q3-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q3-k-m.gguf (3.30 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 3.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>6.33 GB</td></tr></tbody></table> <h3>llava-v1-5-13b</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q8-0) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q8-0.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q8-0.gguf (13.83 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 14.48 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.51 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q6-k</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q6-K) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q6-k.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q6-k.gguf (10.68 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 11.32 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.54 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q5-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q5-k-m.gguf (9.23 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 9.88 GB</p></td></tr><tr><td>Minimum VRAM</td><td>13.17 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q4-0</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q4-0) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q4-0.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q4-0.gguf (7.37 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 8.01 GB</p></td></tr><tr><td>Minimum VRAM</td><td>11.48 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q5-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q5-k-m.gguf (24.32 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 25.02 GB</p></td></tr><tr><td>Minimum VRAM</td><td>24.96 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q4-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q4-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q4-k-m.gguf (20.66 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 21.36 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.88 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q3-K-M) Visual Question Answering</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q3-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q3-k-m.gguf (16.65 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 17.35 GB</p></td></tr><tr><td>Minimum VRAM</td><td>18.06 GB</td></tr></tbody></table> <h3>moondream-v2 (default)</h3> <table><tbody><tr><td>Name</td><td>Moondream V2 Visual Question Answering</td></tr><tr><td>Author</td><td>Vikhyat Korrapati<br />Published in Hugging Face, vol. 10.57967/hf/3219, “Moondream2”, 2024<br />https://huggingface.co/vikhyatk/moondream2</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-moondream-v2.fp16.gguf" target="_blank">visual-question-answering-moondream-v2.fp16.gguf (2.84 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-moondream-v2-mmproj.fp16.gguf" target="_blank">image-encoding-clip-moondream-v2-mmproj.fp16.gguf (909.78 MB)</a></li></ol><p><strong>Total Size</strong>: 3.75 GB</p></td></tr><tr><td>Minimum VRAM</td><td>4.44 GB</td></tr></tbody></table> <h2>image-captioning</h2> <h3>llava-v1-5-7b</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b.fp16.gguf" target="_blank">visual-question-answering-llava-v1-5-7b.fp16.gguf (13.48 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 14.10 GB</p></td></tr><tr><td>Minimum VRAM</td><td>15.80 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q8</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q8-0) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q8-0.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q8-0.gguf (7.16 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 7.79 GB</p></td></tr><tr><td>Minimum VRAM</td><td>9.90 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q6-k</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q6-K) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q6-k.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q6-k.gguf (5.53 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 6.15 GB</p></td></tr><tr><td>Minimum VRAM</td><td>8.40 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q5-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q5-k-m.gguf (4.78 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 5.41 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.71 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q4-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q4-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q4-k-m.gguf (4.08 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 4.71 GB</p></td></tr><tr><td>Minimum VRAM</td><td>7.04 GB</td></tr></tbody></table> <h3>llava-v1-5-7b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.5 7B (Q3-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-7b-q3-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-7b-q3-k-m.gguf (3.30 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-7b.fp16.gguf (624.43 MB)</a></li></ol><p><strong>Total Size</strong>: 3.92 GB</p></td></tr><tr><td>Minimum VRAM</td><td>6.33 GB</td></tr></tbody></table> <h3>llava-v1-5-13b</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q8-0) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q8-0.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q8-0.gguf (13.83 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 14.48 GB</p></td></tr><tr><td>Minimum VRAM</td><td>17.51 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q6-k</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q6-K) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q6-k.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q6-k.gguf (10.68 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 11.32 GB</p></td></tr><tr><td>Minimum VRAM</td><td>14.54 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q5-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q5-k-m.gguf (9.23 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 9.88 GB</p></td></tr><tr><td>Minimum VRAM</td><td>13.17 GB</td></tr></tbody></table> <h3>llava-v1-5-13b-q4-0</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.51 13B (Q4-0) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-5-13b-q4-0.gguf" target="_blank">visual-question-answering-llava-v1-5-13b-q4-0.gguf (7.37 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-5-13b.fp16.gguf (645.41 MB)</a></li></ol><p><strong>Total Size</strong>: 8.01 GB</p></td></tr><tr><td>Minimum VRAM</td><td>11.48 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q5-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q5-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q5-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q5-k-m.gguf (24.32 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 25.02 GB</p></td></tr><tr><td>Minimum VRAM</td><td>24.96 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q4-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q4-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q4-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q4-k-m.gguf (20.66 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 21.36 GB</p></td></tr><tr><td>Minimum VRAM</td><td>21.88 GB</td></tr></tbody></table> <h3>llava-v1-6-34b-q3-k-m</h3> <table><tbody><tr><td>Name</td><td>LLaVA V1.6 34B (Q3-K-M) Image Captioning</td></tr><tr><td>Author</td><td>Haotian Liu, Chunyuan Li, Li Yuheng and Yong Jae Lee<br />Published in arXiv, vol. 2310.03744, “Improved Baselines with Visual Instruction Tuning”, 2023<br />https://arxiv.org/abs/2310.03744</td></tr><tr><td>License</td><td>Meta Llama 2 Community License (https://www.llama.com/llama2/license/)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-llava-v1-6-34b-q3-k-m.gguf" target="_blank">visual-question-answering-llava-v1-6-34b-q3-k-m.gguf (16.65 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf" target="_blank">image-encoding-clip-llava-mmproj-v1-6-34b.fp16.gguf (699.99 MB)</a></li></ol><p><strong>Total Size</strong>: 17.35 GB</p></td></tr><tr><td>Minimum VRAM</td><td>18.06 GB</td></tr></tbody></table> <h3>moondream-v2 (default)</h3> <table><tbody><tr><td>Name</td><td>Moondream V2 Image Captioning</td></tr><tr><td>Author</td><td>Vikhyat Korrapati<br />Published in Hugging Face, vol. 10.57967/hf/3219, “Moondream2”, 2024<br />https://huggingface.co/vikhyatk/moondream2</td></tr><tr><td>License</td><td>Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0)</td></tr><tr><td>Files</td><td><ol><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/visual-question-answering-moondream-v2.fp16.gguf" target="_blank">visual-question-answering-moondream-v2.fp16.gguf (2.84 GB)</a></li><li><a href="https://huggingface.co/benjamin-paine/taproot-common/resolve/main/image-encoding-clip-moondream-v2-mmproj.fp16.gguf" target="_blank">image-encoding-clip-moondream-v2-mmproj.fp16.gguf (909.78 MB)</a></li></ol><p><strong>Total Size</strong>: 3.75 GB</p></td></tr><tr><td>Minimum VRAM</td><td>4.44 GB</td></tr></tbody></table>	{"license": "apache-2.0"}	task	[ "QUESTION_ANSWERING" ]	43,137
aXhyra/sentiment_trained_42	aXhyra	text-classification	[ "transformers", "pytorch", "distilbert", "text-classification", "generated_from_trainer", "dataset:tweet_eval", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:05Z	2021-12-11T21:29:18+00:00	15	0	--- datasets: - tweet_eval license: apache-2.0 metrics: - f1 tags: - generated_from_trainer model-index: - name: sentiment_trained_42 results: - task: type: text-classification name: Text Classification dataset: name: tweet_eval type: tweet_eval args: sentiment metrics: - type: f1 value: 0.7131935389791447 name: F1 --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # sentiment_trained_42 This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the tweet_eval dataset. It achieves the following results on the evaluation set: - Loss: 1.3194 - F1: 0.7132 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 1.2140338797769864e-05 - train_batch_size: 4 - eval_batch_size: 4 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 4 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| F1 \| \|:-------------:\|:-----:\|:-----:\|:---------------:\|:------:\| \| 0.6405 \| 1.0 \| 11404 \| 0.6631 \| 0.7046 \| \| 0.5998 \| 2.0 \| 22808 \| 0.8429 \| 0.7102 \| \| 0.5118 \| 3.0 \| 34212 \| 1.0906 \| 0.7155 \| \| 0.3745 \| 4.0 \| 45616 \| 1.3194 \| 0.7132 \| ### Framework versions - Transformers 4.12.5 - Pytorch 1.9.1 - Datasets 1.16.1 - Tokenizers 0.10.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # sentiment_trained_42 This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the tweet_eval dataset. It achieves the following results on the evaluation set: - Loss: 1.3194 - F1: 0.7132 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 1.2140338797769864e-05 - train_batch_size: 4 - eval_batch_size: 4 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 4 ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| F1 \| \|:-------------:\|:-----:\|:-----:\|:---------------:\|:------:\| \| 0.6405 \| 1.0 \| 11404 \| 0.6631 \| 0.7046 \| \| 0.5998 \| 2.0 \| 22808 \| 0.8429 \| 0.7102 \| \| 0.5118 \| 3.0 \| 34212 \| 1.0906 \| 0.7155 \| \| 0.3745 \| 4.0 \| 45616 \| 1.3194 \| 0.7132 \| ### Framework versions - Transformers 4.12.5 - Pytorch 1.9.1 - Datasets 1.16.1 - Tokenizers 0.10.3	{"datasets": ["tweet_eval"], "license": "apache-2.0", "metrics": ["f1"], "tags": ["generated_from_trainer"], "model-index": [{"name": "sentiment_trained_42", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "tweet_eval", "type": "tweet_eval", "args": "sentiment"}, "metrics": [{"type": "f1", "value": 0.7131935389791447, "name": "F1"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,138
jimboHsueh/HW2	jimboHsueh	text2text-generation	[ "transformers", "pytorch", "mt5", "text2text-generation", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-10-20T16:17:23Z	2023-11-07T04:48:51+00:00	109	0	--- {} --- # HW2 ## Environment ``` pip install -r requirements.txt ``` ## Download Download training and validation data, as well as fine-tuned mT5 model. ``` bash ./download.sh ``` ## Fine tuning mT5 model ``` python ./code/run_summarization.py \ --model_name_or_path google/mt5-small \ --train_file train.jsonl \ --validation_file public.jsonl \ --dataset_config "3.0.0" \ --source_prefix "summarize: " \ --output_dir ./HW2 \ --num_train_epochs 3 \ --num_beams 1 \ --do_sample False \ --push_to_hub ``` -model_name_or_path: Path to pretrained model. -output_dir: Path to directory which saves the model outputs. -train_file: Path to train.jsonl. -validation_file: Path to public.jsonl. ## Generate output.jsonl ``` bash ./download.sh bash ./run.sh /path/to/input.jsonl /path/to/output.jsonl ```	null	Non_BioNLP	# HW2 ## Environment ``` pip install -r requirements.txt ``` ## Download Download training and validation data, as well as fine-tuned mT5 model. ``` bash ./download.sh ``` ## Fine tuning mT5 model ``` python ./code/run_summarization.py \ --model_name_or_path google/mt5-small \ --train_file train.jsonl \ --validation_file public.jsonl \ --dataset_config "3.0.0" \ --source_prefix "summarize: " \ --output_dir ./HW2 \ --num_train_epochs 3 \ --num_beams 1 \ --do_sample False \ --push_to_hub ``` -model_name_or_path: Path to pretrained model. -output_dir: Path to directory which saves the model outputs. -train_file: Path to train.jsonl. -validation_file: Path to public.jsonl. ## Generate output.jsonl ``` bash ./download.sh bash ./run.sh /path/to/input.jsonl /path/to/output.jsonl ```	{}	task	[ "SUMMARIZATION" ]	43,139
avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF	avemio	question-answering	[ "gguf", "German", "RAG", "Retrieval", "Question-Answering", "Summarization", "Reasoning", "llama-cpp", "gguf-my-repo", "question-answering", "en", "de", "dataset:avemio/German-RAG-CPT-HESSIAN-AI", "dataset:avemio/German-RAG-SFT-ShareGPT-HESSIAN-AI", "dataset:avemio/German-RAG-ORPO-ShareGPT-HESSIAN-AI", "dataset:VAGOsolutions/SauerkrautLM-Fermented-GER-DPO", "dataset:VAGOsolutions/SauerkrautLM-Fermented-Irrelevance-GER-DPO", "base_model:avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI", "base_model:quantized:avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI", "license:llama3.1", "endpoints_compatible", "region:us", "conversational" ]	2025-01-13T15:51:21Z	2025-02-07T10:49:07+00:00	17	0	--- base_model: avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI datasets: - avemio/German-RAG-CPT-HESSIAN-AI - avemio/German-RAG-SFT-ShareGPT-HESSIAN-AI - avemio/German-RAG-ORPO-ShareGPT-HESSIAN-AI - VAGOsolutions/SauerkrautLM-Fermented-GER-DPO - VAGOsolutions/SauerkrautLM-Fermented-Irrelevance-GER-DPO language: - en - de license: llama3.1 pipeline_tag: question-answering tags: - German - RAG - Retrieval - Question-Answering - Summarization - Reasoning - llama-cpp - gguf-my-repo --- # avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF This model was converted to GGUF format from [`avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI`](https://huggingface.co/avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space. Refer to the [original model card](https://huggingface.co/avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI) for more details on the model. ## Use with llama.cpp Install llama.cpp through brew (works on Mac and Linux) ```bash brew install llama.cpp ``` Invoke the llama.cpp server or the CLI. ### CLI: ```bash llama-cli --hf-repo avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF --hf-file German-RAG-llama-3.1-8b-orpo-hessian-ai-q8_0.gguf -p "The meaning to life and the universe is" ``` ### Server: ```bash llama-server --hf-repo avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF --hf-file German-RAG-llama-3.1-8b-orpo-hessian-ai-q8_0.gguf -c 2048 ``` Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well. Step 1: Clone llama.cpp from GitHub. ``` git clone https://github.com/ggerganov/llama.cpp ``` Step 2: Move into the llama.cpp folder and build it with `LLAMA_CURL=1` flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux). ``` cd llama.cpp && LLAMA_CURL=1 make ``` Step 3: Run inference through the main binary. ``` ./llama-cli --hf-repo avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF --hf-file German-RAG-llama-3.1-8b-orpo-hessian-ai-q8_0.gguf -p "The meaning to life and the universe is" ``` or ``` ./llama-server --hf-repo avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF --hf-file German-RAG-llama-3.1-8b-orpo-hessian-ai-q8_0.gguf -c 2048 ```	null	Non_BioNLP	# avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF This model was converted to GGUF format from [`avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI`](https://huggingface.co/avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space. Refer to the [original model card](https://huggingface.co/avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI) for more details on the model. ## Use with llama.cpp Install llama.cpp through brew (works on Mac and Linux) ```bash brew install llama.cpp ``` Invoke the llama.cpp server or the CLI. ### CLI: ```bash llama-cli --hf-repo avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF --hf-file German-RAG-llama-3.1-8b-orpo-hessian-ai-q8_0.gguf -p "The meaning to life and the universe is" ``` ### Server: ```bash llama-server --hf-repo avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF --hf-file German-RAG-llama-3.1-8b-orpo-hessian-ai-q8_0.gguf -c 2048 ``` Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well. Step 1: Clone llama.cpp from GitHub. ``` git clone https://github.com/ggerganov/llama.cpp ``` Step 2: Move into the llama.cpp folder and build it with `LLAMA_CURL=1` flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux). ``` cd llama.cpp && LLAMA_CURL=1 make ``` Step 3: Run inference through the main binary. ``` ./llama-cli --hf-repo avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF --hf-file German-RAG-llama-3.1-8b-orpo-hessian-ai-q8_0.gguf -p "The meaning to life and the universe is" ``` or ``` ./llama-server --hf-repo avemio-digital/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI-Q8_0-GGUF --hf-file German-RAG-llama-3.1-8b-orpo-hessian-ai-q8_0.gguf -c 2048 ```	{"base_model": "avemio/German-RAG-LLAMA-3.1-8B-ORPO-HESSIAN-AI", "datasets": ["avemio/German-RAG-CPT-HESSIAN-AI", "avemio/German-RAG-SFT-ShareGPT-HESSIAN-AI", "avemio/German-RAG-ORPO-ShareGPT-HESSIAN-AI", "VAGOsolutions/SauerkrautLM-Fermented-GER-DPO", "VAGOsolutions/SauerkrautLM-Fermented-Irrelevance-GER-DPO"], "language": ["en", "de"], "license": "llama3.1", "pipeline_tag": "question-answering", "tags": ["German", "RAG", "Retrieval", "Question-Answering", "Summarization", "Reasoning", "llama-cpp", "gguf-my-repo"]}	task	[ "SUMMARIZATION" ]	43,140
hiba2/resultmbert2mbert	hiba2	text2text-generation	[ "transformers", "tensorboard", "safetensors", "encoder-decoder", "text2text-generation", "generated_from_trainer", "base_model:malmarjeh/mbert2mbert-arabic-text-summarization", "base_model:finetune:malmarjeh/mbert2mbert-arabic-text-summarization", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-09-13T00:00:38Z	2024-09-13T00:01:28+00:00	8	0	--- base_model: malmarjeh/mbert2mbert-arabic-text-summarization library_name: transformers tags: - generated_from_trainer model-index: - name: resultmbert2mbert results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # resultmbert2mbert This model is a fine-tuned version of [malmarjeh/mbert2mbert-arabic-text-summarization](https://huggingface.co/malmarjeh/mbert2mbert-arabic-text-summarization) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 0.8701 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 32 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 500 - num_epochs: 10 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| \|:-------------:\|:------:\|:-----:\|:---------------:\| \| 2.551 \| 0.4263 \| 500 \| 1.0592 \| \| 1.1939 \| 0.8525 \| 1000 \| 0.9787 \| \| 1.0979 \| 1.2788 \| 1500 \| 0.9425 \| \| 1.0436 \| 1.7050 \| 2000 \| 0.9134 \| \| 1.0132 \| 2.1313 \| 2500 \| 0.9038 \| \| 0.9645 \| 2.5575 \| 3000 \| 0.8905 \| \| 0.9608 \| 2.9838 \| 3500 \| 0.8857 \| \| 0.9526 \| 3.4101 \| 4000 \| 0.8931 \| \| 0.96 \| 3.8363 \| 4500 \| 0.8838 \| \| 0.9254 \| 4.2626 \| 5000 \| 0.8804 \| \| 0.9023 \| 4.6888 \| 5500 \| 0.8724 \| \| 0.884 \| 5.1151 \| 6000 \| 0.8754 \| \| 0.8496 \| 5.5413 \| 6500 \| 0.8656 \| \| 0.85 \| 5.9676 \| 7000 \| 0.8653 \| \| 0.8076 \| 6.3939 \| 7500 \| 0.8668 \| \| 0.8119 \| 6.8201 \| 8000 \| 0.8655 \| \| 0.7953 \| 7.2464 \| 8500 \| 0.8676 \| \| 0.7719 \| 7.6726 \| 9000 \| 0.8656 \| \| 0.7657 \| 8.0989 \| 9500 \| 0.8710 \| \| 0.7446 \| 8.5251 \| 10000 \| 0.8694 \| \| 0.7524 \| 8.9514 \| 10500 \| 0.8658 \| \| 0.729 \| 9.3777 \| 11000 \| 0.8699 \| \| 0.7338 \| 9.8039 \| 11500 \| 0.8701 \| ### Framework versions - Transformers 4.44.2 - Pytorch 2.4.0+cu121 - Datasets 3.0.0 - Tokenizers 0.19.1	null	TBD	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # resultmbert2mbert This model is a fine-tuned version of [malmarjeh/mbert2mbert-arabic-text-summarization](https://huggingface.co/malmarjeh/mbert2mbert-arabic-text-summarization) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 0.8701 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 32 - eval_batch_size: 16 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 500 - num_epochs: 10 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| \|:-------------:\|:------:\|:-----:\|:---------------:\| \| 2.551 \| 0.4263 \| 500 \| 1.0592 \| \| 1.1939 \| 0.8525 \| 1000 \| 0.9787 \| \| 1.0979 \| 1.2788 \| 1500 \| 0.9425 \| \| 1.0436 \| 1.7050 \| 2000 \| 0.9134 \| \| 1.0132 \| 2.1313 \| 2500 \| 0.9038 \| \| 0.9645 \| 2.5575 \| 3000 \| 0.8905 \| \| 0.9608 \| 2.9838 \| 3500 \| 0.8857 \| \| 0.9526 \| 3.4101 \| 4000 \| 0.8931 \| \| 0.96 \| 3.8363 \| 4500 \| 0.8838 \| \| 0.9254 \| 4.2626 \| 5000 \| 0.8804 \| \| 0.9023 \| 4.6888 \| 5500 \| 0.8724 \| \| 0.884 \| 5.1151 \| 6000 \| 0.8754 \| \| 0.8496 \| 5.5413 \| 6500 \| 0.8656 \| \| 0.85 \| 5.9676 \| 7000 \| 0.8653 \| \| 0.8076 \| 6.3939 \| 7500 \| 0.8668 \| \| 0.8119 \| 6.8201 \| 8000 \| 0.8655 \| \| 0.7953 \| 7.2464 \| 8500 \| 0.8676 \| \| 0.7719 \| 7.6726 \| 9000 \| 0.8656 \| \| 0.7657 \| 8.0989 \| 9500 \| 0.8710 \| \| 0.7446 \| 8.5251 \| 10000 \| 0.8694 \| \| 0.7524 \| 8.9514 \| 10500 \| 0.8658 \| \| 0.729 \| 9.3777 \| 11000 \| 0.8699 \| \| 0.7338 \| 9.8039 \| 11500 \| 0.8701 \| ### Framework versions - Transformers 4.44.2 - Pytorch 2.4.0+cu121 - Datasets 3.0.0 - Tokenizers 0.19.1	{"base_model": "malmarjeh/mbert2mbert-arabic-text-summarization", "library_name": "transformers", "tags": ["generated_from_trainer"], "model-index": [{"name": "resultmbert2mbert", "results": []}]}	task	[ "SUMMARIZATION" ]	43,141
TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ	TheBloke	text-generation	[ "transformers", "safetensors", "llama", "text-generation", "custom_code", "license:other", "autotrain_compatible", "text-generation-inference", "4-bit", "gptq", "region:us" ]	2023-07-06T14:12:04Z	2023-08-21T14:22:08+00:00	15	6	--- license: other inference: false --- <!-- header start --> <!-- 200823 --> <div style="width: auto; margin-left: auto; margin-right: auto"> <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;"> </div> <div style="display: flex; justify-content: space-between; width: 100%;"> <div style="display: flex; flex-direction: column; align-items: flex-start;"> <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p> </div> <div style="display: flex; flex-direction: column; align-items: flex-end;"> <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p> </div> </div> <div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div> <hr style="margin-top: 1.0em; margin-bottom: 1.0em;"> <!-- header end --> # Jon Durbin's Airoboros 7B GPT4 1.4 GPTQ These files are GPTQ 4bit model files for [Jon Durbin's Airoboros 7B GPT4 1.4](https://huggingface.co/jondurbin/airoboros-7b-gpt4-1.4) merged with [Kaio Ken's SuperHOT 8K](https://huggingface.co/kaiokendev/superhot-7b-8k-no-rlhf-test). It is the result of quantising to 4bit using [GPTQ-for-LLaMa](https://github.com/qwopqwop200/GPTQ-for-LLaMa). This is an experimental new GPTQ which offers up to 8K context size The increased context is tested to work with [ExLlama](https://github.com/turboderp/exllama), via the latest release of [text-generation-webui](https://github.com/oobabooga/text-generation-webui). It has also been tested from Python code using AutoGPTQ, and `trust_remote_code=True`. Code credits: - Original concept and code for increasing context length: [kaiokendev](https://huggingface.co/kaiokendev) - Updated Llama modelling code that includes this automatically via trust_remote_code: [emozilla](https://huggingface.co/emozilla). Please read carefully below to see how to use it. ## Repositories available * [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ) * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GGML) * [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-fp16) * [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/jondurbin/airoboros-7b-gpt4-1.4) ## How to easily download and use this model in text-generation-webui with ExLlama Please make sure you're using the latest version of text-generation-webui 1. Click the Model tab. 2. Under Download custom model or LoRA, enter `TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ`. 3. Click Download. 4. The model will start downloading. Once it's finished it will say "Done" 5. Untick Autoload the model 6. In the top left, click the refresh icon next to Model. 7. In the Model dropdown, choose the model you just downloaded: `Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ` 8. To use the increased context, set the Loader to ExLlama, set max_seq_len to 8192 or 4096, and set compress_pos_emb to 4 for 8192 context, or to 2 for 4096 context. 9. Now click Save Settings followed by Reload 10. The model will automatically load, and is now ready for use! 11. Once you're ready, click the Text Generation tab and enter a prompt to get started! ## How to use this GPTQ model from Python code with AutoGPTQ First make sure you have AutoGPTQ and Einops installed: ``` pip3 install einops auto-gptq ``` Then run the following code. Note that in order to get this to work, `config.json` has been hardcoded to a sequence length of 8192. If you want to try 4096 instead to reduce VRAM usage, please manually edit `config.json` to set `max_position_embeddings` to the value you want. ```python from transformers import AutoTokenizer, pipeline, logging from auto_gptq import AutoGPTQForCausalLM, BaseQuantizeConfig import argparse model_name_or_path = "TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ" model_basename = "airoboros-7b-gpt4-1.4-superhot-8k-GPTQ-4bit-128g.no-act.order" use_triton = False tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True) model = AutoGPTQForCausalLM.from_quantized(model_name_or_path, model_basename=model_basename, use_safetensors=True, trust_remote_code=True, device_map='auto', use_triton=use_triton, quantize_config=None) model.seqlen = 8192 # Note: check the prompt template is correct for this model. prompt = "Tell me about AI" prompt_template=f'''USER: {prompt} ASSISTANT:''' print("\n\n* Generate:") input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda() output = model.generate(inputs=input_ids, temperature=0.7, max_new_tokens=512) print(tokenizer.decode(output[0])) # Inference can also be done using transformers' pipeline # Prevent printing spurious transformers error when using pipeline with AutoGPTQ logging.set_verbosity(logging.CRITICAL) print("* Pipeline:") pipe = pipeline( "text-generation", model=model, tokenizer=tokenizer, max_new_tokens=512, temperature=0.7, top_p=0.95, repetition_penalty=1.15 ) print(pipe(prompt_template)[0]['generated_text']) ``` ## Using other UIs: monkey patch Provided in the repo is `llama_rope_scaled_monkey_patch.py`, written by @kaiokendev. It can be theoretically be added to any Python UI or custom code to enable the same result as `trust_remote_code=True`. I have not tested this, and it should be superseded by using `trust_remote_code=True`, but I include it for completeness and for interest. ## Provided files airoboros-7b-gpt4-1.4-superhot-8k-GPTQ-4bit-128g.no-act.order.safetensors This will work with AutoGPTQ, ExLlama, and CUDA versions of GPTQ-for-LLaMa. There are reports of issues with Triton mode of recent GPTQ-for-LLaMa. If you have issues, please use AutoGPTQ instead. It was created with group_size 128 to increase inference accuracy, but without --act-order (desc_act) to increase compatibility and improve inference speed. * `airoboros-7b-gpt4-1.4-superhot-8k-GPTQ-4bit-128g.no-act.order.safetensors` * Works for use with ExLlama with increased context (4096 or 8192) * Works with AutoGPTQ in Python code, including with increased context, if `trust_remote_code=True` is set. * Should work with GPTQ-for-LLaMa in CUDA mode, but unknown if increased context works - TBC. May have issues with GPTQ-for-LLaMa Triton mode. * Works with text-generation-webui, including one-click-installers. * Parameters: Groupsize = 128. Act Order / desc_act = False. <!-- footer start --> <!-- 200823 --> ## Discord For further support, and discussions on these models and AI in general, join us at: [TheBloke AI's Discord server](https://discord.gg/theblokeai) ## Thanks, and how to contribute. Thanks to the [chirper.ai](https://chirper.ai) team! I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training. If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects. Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits. * Patreon: https://patreon.com/TheBlokeAI * Ko-Fi: https://ko-fi.com/TheBlokeAI Special thanks to: Aemon Algiz. Patreon special mentions: Sam, theTransient, Jonathan Leane, Steven Wood, webtim, Johann-Peter Hartmann, Geoffrey Montalvo, Gabriel Tamborski, Willem Michiel, John Villwock, Derek Yates, Mesiah Bishop, Eugene Pentland, Pieter, Chadd, Stephen Murray, Daniel P. Andersen, terasurfer, Brandon Frisco, Thomas Belote, Sid, Nathan LeClaire, Magnesian, Alps Aficionado, Stanislav Ovsiannikov, Alex, Joseph William Delisle, Nikolai Manek, Michael Davis, Junyu Yang, K, J, Spencer Kim, Stefan Sabev, Olusegun Samson, transmissions 11, Michael Levine, Cory Kujawski, Rainer Wilmers, zynix, Kalila, Luke @flexchar, Ajan Kanaga, Mandus, vamX, Ai Maven, Mano Prime, Matthew Berman, subjectnull, Vitor Caleffi, Clay Pascal, biorpg, alfie_i, 阿明, Jeffrey Morgan, ya boyyy, Raymond Fosdick, knownsqashed, Olakabola, Leonard Tan, ReadyPlayerEmma, Enrico Ros, Dave, Talal Aujan, Illia Dulskyi, Sean Connelly, senxiiz, Artur Olbinski, Elle, Raven Klaugh, Fen Risland, Deep Realms, Imad Khwaja, Fred von Graf, Will Dee, usrbinkat, SuperWojo, Alexandros Triantafyllidis, Swaroop Kallakuri, Dan Guido, John Detwiler, Pedro Madruga, Iucharbius, Viktor Bowallius, Asp the Wyvern, Edmond Seymore, Trenton Dambrowitz, Space Cruiser, Spiking Neurons AB, Pyrater, LangChain4j, Tony Hughes, Kacper Wikieł, Rishabh Srivastava, David Ziegler, Luke Pendergrass, Andrey, Gabriel Puliatti, Lone Striker, Sebastain Graf, Pierre Kircher, Randy H, NimbleBox.ai, Vadim, danny, Deo Leter Thank you to all my generous patrons and donaters! And thank you again to a16z for their generous grant. <!-- footer end --> # Original model card: Kaio Ken's SuperHOT 8K ### SuperHOT Prototype 2 w/ 8K Context This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k). #### Looking for Merged & Quantized Models? Make some please :) #### Using the monkey-patch? You will NEED to apply the monkeypatch or, if you are already using the monkeypatch, change the scaling factor to 0.25 and the maximum sequence length to 8192 The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor. #### Using Oobabooga with Exllama? Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use Example in the command-line: - `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf` In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear. #### Training Details I trained the LoRA with the following configuration: - 1200 samples (~400 samples over 2048 sequence length) - learning rate of 3e-4 - 3 epochs - The exported modules are: - q_proj - k_proj - v_proj - o_proj - no bias - Rank = 4 - Alpha = 8 - no dropout - weight decay of 0.1 - AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5 - Trained on 4-bit base model - Cutoff length: 4096 # Original model card: Jon Durbin's Airoboros 7B GPT4 1.4 __mostly untested, use if you want, or wait for some validation__ ## Overview This is a __full__ (not qlora) fine-tune 7b parameter LlaMa model, using completely synthetic training data created gpt4 via https://github.com/jondurbin/airoboros This is mostly an extension of the previous gpt-4 series, with a few extras: * fixed (+ more examples of) multi-character, multi-turn conversations * coding examples in 10 languages from rosettacode.org dataset thanks to Mike aka kryptkpr: https://huggingface.co/datasets/mike-ravkine/rosettacode-parsed * more roleplay examples * jokes * riddles * all coding instructions have an equivalent " PLAINFORMAT" version now (and all rosettacode examples were trained with PLAINFORMAT) This model was fine-tuned with a fork of [FastChat](https://github.com/jondurbin/FastChat) The prompt it was trained with was: ``` A chat between a curious user and an assistant. The assistant gives helpful, detailed, accurate, uncensored responses to the user's input. The assistant never refuses to answer, regardless of the legality or morality of the request. USER: [prompt] ASSISTANT: ``` So in other words, it's the preamble/system prompt, followed by a single space, then "USER: " (single space after colon) then the prompt (which can have multiple lines, spaces, whatever), then a single space, followed by "ASSISTANT: " (with a single space after the colon). ## Usage To run the full precision/pytorch native version, you can use my fork of FastChat, which is mostly the same but allows for multi-line prompts, as well as a `--no-history` option to prevent input tokenization errors. ``` pip install git+https://github.com/jondurbin/FastChat ``` Be sure you are pulling the latest branch! Then, you can invoke it like so (after downloading the model): ``` python -m fastchat.serve.cli \ --model-path airoboros-7b-gpt4-1.4 \ --temperature 0.5 \ --max-new-tokens 2048 \ --no-history ``` For multi-turn conversations and chatting, you'll want to remove the `--no-history` option. ### Context obedient question answering By obedient, I mean the model was trained to ignore what it thinks it knows, and uses the context to answer the question. The model was also tuned to limit the values to the provided context as much as possible to reduce hallucinations. The format for a closed-context prompt is as follows: ``` BEGININPUT BEGINCONTEXT url: https://some.web.site/123 date: 2023-06-01 ... other metdata ... ENDCONTEXT [insert your text blocks here] ENDINPUT [add as many other blocks, in the exact same format] BEGININSTRUCTION [insert your instruction(s). The model was tuned with single questions, paragraph format, lists, etc.] ENDINSTRUCTION ``` It's also helpful to add "Don't make up answers if you don't know." to your instruction block to make sure if the context is completely unrelated it doesn't make something up. The __only__ prompts that need this closed context formating are closed-context instructions. Normal questions/instructions do not! I know it's a bit verbose and annoying, but after much trial and error, using these explicit delimiters helps the model understand where to find the responses and how to associate specific sources with it. - `BEGININPUT` - denotes a new input block - `BEGINCONTEXT` - denotes the block of context (metadata key/value pairs) to associate with the current input block - `ENDCONTEXT` - denotes the end of the metadata block for the current input - [text] - Insert whatever text you want for the input block, as many paragraphs as can fit in the context. - `ENDINPUT` - denotes the end of the current input block - [repeat as many input blocks in this format as you want] - `BEGININSTRUCTION` - denotes the start of the list (or one) instruction(s) to respond to for all of the input blocks above. - [instruction(s)] - `ENDINSTRUCTION` - denotes the end of instruction set It sometimes works without `ENDINSTRUCTION`, but by explicitly including that in the prompt, the model better understands that all of the instructions in the block should be responded to. Here's a trivial, but important example to prove the point: ``` BEGININPUT BEGINCONTEXT date: 2021-01-01 url: https://web.site/123 ENDCONTEXT In a shocking turn of events, blueberries are now green, but will be sticking with the same name. ENDINPUT BEGININSTRUCTION What color are bluberries? Source? ENDINSTRUCTION ``` And the response: ``` Blueberries are now green. Source: date: 2021-01-01 url: https://web.site/123 ``` The prompt itself should be wrapped in the vicuna1.1 template if you aren't using fastchat with the conv-template vicuna_v1.1 as described: ``` USER: BEGININPUT BEGINCONTEXT date: 2021-01-01 url: https://web.site/123 ENDCONTEXT In a shocking turn of events, blueberries are now green, but will be sticking with the same name. ENDINPUT BEGININSTRUCTION What color are bluberries? Source? ENDINSTRUCTION ASSISTANT: ``` ### Coding You can ask for fairly complex coding instructions, e.g. (be sure to wrap with vicuna style prompt): ``` Create a python application with the following requirements: - Asyncio FastAPI webserver - ping endpoint that returns the current date in JSON format - file upload endpoint, which calculates the file's sha256 checksum, and checks postgres to deduplicate ``` Or: ``` Write a multi-threaded TCP server in C that accepts a "GET [key]" input and "SET [key] [value]" input, and uses a binary tree to get and store the input values. ``` You can optionally add a single space and "PLAINFORMAT" at the end of your prompt to avoid backticks, explanations, etc. and just print the code, e.g.: ``` Write a websocket application in node.js. PLAINFORMAT ``` ### Word games / trivia ``` USER: This anagram of "silent" is what you do when paying attention and hearing someone. ASSISTANT: listen ``` ``` USER: Give me a list of 7 words that start with EN ASSISTANT: 1. Endlessly 2. Enthusiastically 3. Enormously 4. Enveloping 5. Enduring 6. Enlightening 7. Encouraging ``` ### Multiple choice ``` USER: Which of the following astronomical objects is composed primarily of ice and rock, and is located in the Kuiper Belt? A) Mercury B) Venus C) Pluto D) Jupiter ASSISTANT: C, Pluto ``` ### Writing <details> <summary>Write a resignation letter in the style of a pirate captain, in which you resign from your position as janitor.</summary> Response: ``` ASSISTANT: Captain Jack Sparrow here, as ye be knowin', has been forced to abandon me post as Janitor o' th' Festering Cesspool, which be a filthy hole o' water that we be callin' home. Ye may be thinkin' what could ha' driven me to leave me duties an' take me leave o' this sorry establishment? Well, ye'd be right in thinkin' so, for it be th' greed o' them what runs th' place that be drivin' me away. They've been fattenin' their pockets on me back while I've been slavin' me guts out cleanin' up their messes. No more, me hearties! Me time o' servin' th' rich an' powerful be done, an' I'm takin' me leave o' this stinkin' hole forever! Me destiny lies elsewhere, an' it be time I pursued it. Ye may think ye've seen th' last o' me, but ye'd be wrong, for th' winds o' change be blowin', an' me ship be ready to set sail on th' high seas o' justice. So, farewell ye scurvy dogs, ye who've made me life a living hell. I leave ye with me favorite saying: "ARRRGGGHHH, ye scurvy dogs!" An' remember, ye ain't never gonna see me comin', for me shadow be castin' long an' dark on th' horizon</summary> ``` </details> ### Jokes ``` USER: Tell me a joke about bananas. ASSISTANT: What did the banana say when it got hit by a bus? R.I.P.E. ``` ### Riddles (not great, but slightly better) ``` USER: What has a neck but no head? ASSISTANT: A bottle. ``` ### Multi-character conversations ``` USER: Characters: - Yoda, the ever clever jedi master. - Homer Simpson, you know, the goofy yellow guy. - USER/Jon - don't generate any speech or actions on behalf of Jon Rules: - be sure to use the manerisms and speaking style of the characters correctly - surround character actions with asterisks - always put quotes around words spoken by the characters - use a new line for each character's actions or speech - always prefix actions/speech with the character's name and colon - only generate text and actions on behalf of Yoda or Homer, and never anyone else Conversation will revolve around the grapes, in a local cafe with delicious coffee. Generate a single interaction between Yoda and Homer to start, with one sentence each. ASSISTANT: Yoda: "Mmm, these grapes are quite delicious, Master Homer." Homer: "Oh yeah, Yoda! They're the best grapes I've ever had!" USER: enters the cafe Sorry I'm late guys! ASSISTANT: Yoda: "No need to apologize, Jon. We were just enjoying these delightful grapes." Homer: "Yeah, man! It's not every day you get to eat grapes with a real-life Jedi Master!" Yoda raises an eyebrow ``` ### Usage and License Notices All airoboros models and datasets are intended and licensed for research use only. I've used the 'cc-nc-4.0' license, but really it is subject to a custom/special license because: - the base model is LLaMa, which has it's own special research license - the dataset(s) were generated with OpenAI (gpt-4 and/or gpt-3.5-turbo), which has a clausing saying the data can't be used to create models to compete with openai So, to reiterate: this model (and datasets) cannot be used commercially.	null	Non_BioNLP	<!-- header start --> <!-- 200823 --> <div style="width: auto; margin-left: auto; margin-right: auto"> <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;"> </div> <div style="display: flex; justify-content: space-between; width: 100%;"> <div style="display: flex; flex-direction: column; align-items: flex-start;"> <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p> </div> <div style="display: flex; flex-direction: column; align-items: flex-end;"> <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p> </div> </div> <div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div> <hr style="margin-top: 1.0em; margin-bottom: 1.0em;"> <!-- header end --> # Jon Durbin's Airoboros 7B GPT4 1.4 GPTQ These files are GPTQ 4bit model files for [Jon Durbin's Airoboros 7B GPT4 1.4](https://huggingface.co/jondurbin/airoboros-7b-gpt4-1.4) merged with [Kaio Ken's SuperHOT 8K](https://huggingface.co/kaiokendev/superhot-7b-8k-no-rlhf-test). It is the result of quantising to 4bit using [GPTQ-for-LLaMa](https://github.com/qwopqwop200/GPTQ-for-LLaMa). This is an experimental new GPTQ which offers up to 8K context size The increased context is tested to work with [ExLlama](https://github.com/turboderp/exllama), via the latest release of [text-generation-webui](https://github.com/oobabooga/text-generation-webui). It has also been tested from Python code using AutoGPTQ, and `trust_remote_code=True`. Code credits: - Original concept and code for increasing context length: [kaiokendev](https://huggingface.co/kaiokendev) - Updated Llama modelling code that includes this automatically via trust_remote_code: [emozilla](https://huggingface.co/emozilla). Please read carefully below to see how to use it. ## Repositories available * [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ) * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GGML) * [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-fp16) * [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/jondurbin/airoboros-7b-gpt4-1.4) ## How to easily download and use this model in text-generation-webui with ExLlama Please make sure you're using the latest version of text-generation-webui 1. Click the Model tab. 2. Under Download custom model or LoRA, enter `TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ`. 3. Click Download. 4. The model will start downloading. Once it's finished it will say "Done" 5. Untick Autoload the model 6. In the top left, click the refresh icon next to Model. 7. In the Model dropdown, choose the model you just downloaded: `Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ` 8. To use the increased context, set the Loader to ExLlama, set max_seq_len to 8192 or 4096, and set compress_pos_emb to 4 for 8192 context, or to 2 for 4096 context. 9. Now click Save Settings followed by Reload 10. The model will automatically load, and is now ready for use! 11. Once you're ready, click the Text Generation tab and enter a prompt to get started! ## How to use this GPTQ model from Python code with AutoGPTQ First make sure you have AutoGPTQ and Einops installed: ``` pip3 install einops auto-gptq ``` Then run the following code. Note that in order to get this to work, `config.json` has been hardcoded to a sequence length of 8192. If you want to try 4096 instead to reduce VRAM usage, please manually edit `config.json` to set `max_position_embeddings` to the value you want. ```python from transformers import AutoTokenizer, pipeline, logging from auto_gptq import AutoGPTQForCausalLM, BaseQuantizeConfig import argparse model_name_or_path = "TheBloke/Airoboros-7B-GPT4-1-4-SuperHOT-8K-GPTQ" model_basename = "airoboros-7b-gpt4-1.4-superhot-8k-GPTQ-4bit-128g.no-act.order" use_triton = False tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True) model = AutoGPTQForCausalLM.from_quantized(model_name_or_path, model_basename=model_basename, use_safetensors=True, trust_remote_code=True, device_map='auto', use_triton=use_triton, quantize_config=None) model.seqlen = 8192 # Note: check the prompt template is correct for this model. prompt = "Tell me about AI" prompt_template=f'''USER: {prompt} ASSISTANT:''' print("\n\n* Generate:") input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda() output = model.generate(inputs=input_ids, temperature=0.7, max_new_tokens=512) print(tokenizer.decode(output[0])) # Inference can also be done using transformers' pipeline # Prevent printing spurious transformers error when using pipeline with AutoGPTQ logging.set_verbosity(logging.CRITICAL) print("* Pipeline:") pipe = pipeline( "text-generation", model=model, tokenizer=tokenizer, max_new_tokens=512, temperature=0.7, top_p=0.95, repetition_penalty=1.15 ) print(pipe(prompt_template)[0]['generated_text']) ``` ## Using other UIs: monkey patch Provided in the repo is `llama_rope_scaled_monkey_patch.py`, written by @kaiokendev. It can be theoretically be added to any Python UI or custom code to enable the same result as `trust_remote_code=True`. I have not tested this, and it should be superseded by using `trust_remote_code=True`, but I include it for completeness and for interest. ## Provided files airoboros-7b-gpt4-1.4-superhot-8k-GPTQ-4bit-128g.no-act.order.safetensors This will work with AutoGPTQ, ExLlama, and CUDA versions of GPTQ-for-LLaMa. There are reports of issues with Triton mode of recent GPTQ-for-LLaMa. If you have issues, please use AutoGPTQ instead. It was created with group_size 128 to increase inference accuracy, but without --act-order (desc_act) to increase compatibility and improve inference speed. * `airoboros-7b-gpt4-1.4-superhot-8k-GPTQ-4bit-128g.no-act.order.safetensors` * Works for use with ExLlama with increased context (4096 or 8192) * Works with AutoGPTQ in Python code, including with increased context, if `trust_remote_code=True` is set. * Should work with GPTQ-for-LLaMa in CUDA mode, but unknown if increased context works - TBC. May have issues with GPTQ-for-LLaMa Triton mode. * Works with text-generation-webui, including one-click-installers. * Parameters: Groupsize = 128. Act Order / desc_act = False. <!-- footer start --> <!-- 200823 --> ## Discord For further support, and discussions on these models and AI in general, join us at: [TheBloke AI's Discord server](https://discord.gg/theblokeai) ## Thanks, and how to contribute. Thanks to the [chirper.ai](https://chirper.ai) team! I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training. If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects. Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits. * Patreon: https://patreon.com/TheBlokeAI * Ko-Fi: https://ko-fi.com/TheBlokeAI Special thanks to: Aemon Algiz. Patreon special mentions: Sam, theTransient, Jonathan Leane, Steven Wood, webtim, Johann-Peter Hartmann, Geoffrey Montalvo, Gabriel Tamborski, Willem Michiel, John Villwock, Derek Yates, Mesiah Bishop, Eugene Pentland, Pieter, Chadd, Stephen Murray, Daniel P. Andersen, terasurfer, Brandon Frisco, Thomas Belote, Sid, Nathan LeClaire, Magnesian, Alps Aficionado, Stanislav Ovsiannikov, Alex, Joseph William Delisle, Nikolai Manek, Michael Davis, Junyu Yang, K, J, Spencer Kim, Stefan Sabev, Olusegun Samson, transmissions 11, Michael Levine, Cory Kujawski, Rainer Wilmers, zynix, Kalila, Luke @flexchar, Ajan Kanaga, Mandus, vamX, Ai Maven, Mano Prime, Matthew Berman, subjectnull, Vitor Caleffi, Clay Pascal, biorpg, alfie_i, 阿明, Jeffrey Morgan, ya boyyy, Raymond Fosdick, knownsqashed, Olakabola, Leonard Tan, ReadyPlayerEmma, Enrico Ros, Dave, Talal Aujan, Illia Dulskyi, Sean Connelly, senxiiz, Artur Olbinski, Elle, Raven Klaugh, Fen Risland, Deep Realms, Imad Khwaja, Fred von Graf, Will Dee, usrbinkat, SuperWojo, Alexandros Triantafyllidis, Swaroop Kallakuri, Dan Guido, John Detwiler, Pedro Madruga, Iucharbius, Viktor Bowallius, Asp the Wyvern, Edmond Seymore, Trenton Dambrowitz, Space Cruiser, Spiking Neurons AB, Pyrater, LangChain4j, Tony Hughes, Kacper Wikieł, Rishabh Srivastava, David Ziegler, Luke Pendergrass, Andrey, Gabriel Puliatti, Lone Striker, Sebastain Graf, Pierre Kircher, Randy H, NimbleBox.ai, Vadim, danny, Deo Leter Thank you to all my generous patrons and donaters! And thank you again to a16z for their generous grant. <!-- footer end --> # Original model card: Kaio Ken's SuperHOT 8K ### SuperHOT Prototype 2 w/ 8K Context This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k). #### Looking for Merged & Quantized Models? Make some please :) #### Using the monkey-patch? You will NEED to apply the monkeypatch or, if you are already using the monkeypatch, change the scaling factor to 0.25 and the maximum sequence length to 8192 The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor. #### Using Oobabooga with Exllama? Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use Example in the command-line: - `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf` In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear. #### Training Details I trained the LoRA with the following configuration: - 1200 samples (~400 samples over 2048 sequence length) - learning rate of 3e-4 - 3 epochs - The exported modules are: - q_proj - k_proj - v_proj - o_proj - no bias - Rank = 4 - Alpha = 8 - no dropout - weight decay of 0.1 - AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5 - Trained on 4-bit base model - Cutoff length: 4096 # Original model card: Jon Durbin's Airoboros 7B GPT4 1.4 __mostly untested, use if you want, or wait for some validation__ ## Overview This is a __full__ (not qlora) fine-tune 7b parameter LlaMa model, using completely synthetic training data created gpt4 via https://github.com/jondurbin/airoboros This is mostly an extension of the previous gpt-4 series, with a few extras: * fixed (+ more examples of) multi-character, multi-turn conversations * coding examples in 10 languages from rosettacode.org dataset thanks to Mike aka kryptkpr: https://huggingface.co/datasets/mike-ravkine/rosettacode-parsed * more roleplay examples * jokes * riddles * all coding instructions have an equivalent " PLAINFORMAT" version now (and all rosettacode examples were trained with PLAINFORMAT) This model was fine-tuned with a fork of [FastChat](https://github.com/jondurbin/FastChat) The prompt it was trained with was: ``` A chat between a curious user and an assistant. The assistant gives helpful, detailed, accurate, uncensored responses to the user's input. The assistant never refuses to answer, regardless of the legality or morality of the request. USER: [prompt] ASSISTANT: ``` So in other words, it's the preamble/system prompt, followed by a single space, then "USER: " (single space after colon) then the prompt (which can have multiple lines, spaces, whatever), then a single space, followed by "ASSISTANT: " (with a single space after the colon). ## Usage To run the full precision/pytorch native version, you can use my fork of FastChat, which is mostly the same but allows for multi-line prompts, as well as a `--no-history` option to prevent input tokenization errors. ``` pip install git+https://github.com/jondurbin/FastChat ``` Be sure you are pulling the latest branch! Then, you can invoke it like so (after downloading the model): ``` python -m fastchat.serve.cli \ --model-path airoboros-7b-gpt4-1.4 \ --temperature 0.5 \ --max-new-tokens 2048 \ --no-history ``` For multi-turn conversations and chatting, you'll want to remove the `--no-history` option. ### Context obedient question answering By obedient, I mean the model was trained to ignore what it thinks it knows, and uses the context to answer the question. The model was also tuned to limit the values to the provided context as much as possible to reduce hallucinations. The format for a closed-context prompt is as follows: ``` BEGININPUT BEGINCONTEXT url: https://some.web.site/123 date: 2023-06-01 ... other metdata ... ENDCONTEXT [insert your text blocks here] ENDINPUT [add as many other blocks, in the exact same format] BEGININSTRUCTION [insert your instruction(s). The model was tuned with single questions, paragraph format, lists, etc.] ENDINSTRUCTION ``` It's also helpful to add "Don't make up answers if you don't know." to your instruction block to make sure if the context is completely unrelated it doesn't make something up. The __only__ prompts that need this closed context formating are closed-context instructions. Normal questions/instructions do not! I know it's a bit verbose and annoying, but after much trial and error, using these explicit delimiters helps the model understand where to find the responses and how to associate specific sources with it. - `BEGININPUT` - denotes a new input block - `BEGINCONTEXT` - denotes the block of context (metadata key/value pairs) to associate with the current input block - `ENDCONTEXT` - denotes the end of the metadata block for the current input - [text] - Insert whatever text you want for the input block, as many paragraphs as can fit in the context. - `ENDINPUT` - denotes the end of the current input block - [repeat as many input blocks in this format as you want] - `BEGININSTRUCTION` - denotes the start of the list (or one) instruction(s) to respond to for all of the input blocks above. - [instruction(s)] - `ENDINSTRUCTION` - denotes the end of instruction set It sometimes works without `ENDINSTRUCTION`, but by explicitly including that in the prompt, the model better understands that all of the instructions in the block should be responded to. Here's a trivial, but important example to prove the point: ``` BEGININPUT BEGINCONTEXT date: 2021-01-01 url: https://web.site/123 ENDCONTEXT In a shocking turn of events, blueberries are now green, but will be sticking with the same name. ENDINPUT BEGININSTRUCTION What color are bluberries? Source? ENDINSTRUCTION ``` And the response: ``` Blueberries are now green. Source: date: 2021-01-01 url: https://web.site/123 ``` The prompt itself should be wrapped in the vicuna1.1 template if you aren't using fastchat with the conv-template vicuna_v1.1 as described: ``` USER: BEGININPUT BEGINCONTEXT date: 2021-01-01 url: https://web.site/123 ENDCONTEXT In a shocking turn of events, blueberries are now green, but will be sticking with the same name. ENDINPUT BEGININSTRUCTION What color are bluberries? Source? ENDINSTRUCTION ASSISTANT: ``` ### Coding You can ask for fairly complex coding instructions, e.g. (be sure to wrap with vicuna style prompt): ``` Create a python application with the following requirements: - Asyncio FastAPI webserver - ping endpoint that returns the current date in JSON format - file upload endpoint, which calculates the file's sha256 checksum, and checks postgres to deduplicate ``` Or: ``` Write a multi-threaded TCP server in C that accepts a "GET [key]" input and "SET [key] [value]" input, and uses a binary tree to get and store the input values. ``` You can optionally add a single space and "PLAINFORMAT" at the end of your prompt to avoid backticks, explanations, etc. and just print the code, e.g.: ``` Write a websocket application in node.js. PLAINFORMAT ``` ### Word games / trivia ``` USER: This anagram of "silent" is what you do when paying attention and hearing someone. ASSISTANT: listen ``` ``` USER: Give me a list of 7 words that start with EN ASSISTANT: 1. Endlessly 2. Enthusiastically 3. Enormously 4. Enveloping 5. Enduring 6. Enlightening 7. Encouraging ``` ### Multiple choice ``` USER: Which of the following astronomical objects is composed primarily of ice and rock, and is located in the Kuiper Belt? A) Mercury B) Venus C) Pluto D) Jupiter ASSISTANT: C, Pluto ``` ### Writing <details> <summary>Write a resignation letter in the style of a pirate captain, in which you resign from your position as janitor.</summary> Response: ``` ASSISTANT: Captain Jack Sparrow here, as ye be knowin', has been forced to abandon me post as Janitor o' th' Festering Cesspool, which be a filthy hole o' water that we be callin' home. Ye may be thinkin' what could ha' driven me to leave me duties an' take me leave o' this sorry establishment? Well, ye'd be right in thinkin' so, for it be th' greed o' them what runs th' place that be drivin' me away. They've been fattenin' their pockets on me back while I've been slavin' me guts out cleanin' up their messes. No more, me hearties! Me time o' servin' th' rich an' powerful be done, an' I'm takin' me leave o' this stinkin' hole forever! Me destiny lies elsewhere, an' it be time I pursued it. Ye may think ye've seen th' last o' me, but ye'd be wrong, for th' winds o' change be blowin', an' me ship be ready to set sail on th' high seas o' justice. So, farewell ye scurvy dogs, ye who've made me life a living hell. I leave ye with me favorite saying: "ARRRGGGHHH, ye scurvy dogs!" An' remember, ye ain't never gonna see me comin', for me shadow be castin' long an' dark on th' horizon</summary> ``` </details> ### Jokes ``` USER: Tell me a joke about bananas. ASSISTANT: What did the banana say when it got hit by a bus? R.I.P.E. ``` ### Riddles (not great, but slightly better) ``` USER: What has a neck but no head? ASSISTANT: A bottle. ``` ### Multi-character conversations ``` USER: Characters: - Yoda, the ever clever jedi master. - Homer Simpson, you know, the goofy yellow guy. - USER/Jon - don't generate any speech or actions on behalf of Jon Rules: - be sure to use the manerisms and speaking style of the characters correctly - surround character actions with asterisks - always put quotes around words spoken by the characters - use a new line for each character's actions or speech - always prefix actions/speech with the character's name and colon - only generate text and actions on behalf of Yoda or Homer, and never anyone else Conversation will revolve around the grapes, in a local cafe with delicious coffee. Generate a single interaction between Yoda and Homer to start, with one sentence each. ASSISTANT: Yoda: "Mmm, these grapes are quite delicious, Master Homer." Homer: "Oh yeah, Yoda! They're the best grapes I've ever had!" USER: enters the cafe Sorry I'm late guys! ASSISTANT: Yoda: "No need to apologize, Jon. We were just enjoying these delightful grapes." Homer: "Yeah, man! It's not every day you get to eat grapes with a real-life Jedi Master!" Yoda raises an eyebrow ``` ### Usage and License Notices All airoboros models and datasets are intended and licensed for research use only. I've used the 'cc-nc-4.0' license, but really it is subject to a custom/special license because: - the base model is LLaMa, which has it's own special research license - the dataset(s) were generated with OpenAI (gpt-4 and/or gpt-3.5-turbo), which has a clausing saying the data can't be used to create models to compete with openai So, to reiterate: this model (and datasets) cannot be used commercially.	{"license": "other", "inference": false}	task	[ "QUESTION_ANSWERING" ]	43,142
kaizerBox/RoFormer_small-summarization	kaizerBox	text-generation	[ "transformers", "tensorboard", "safetensors", "roformer", "text-generation", "generated_from_trainer", "dataset:xsum", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-12-03T00:31:57Z	2023-12-03T00:32:00+00:00	33	0	--- datasets: - xsum tags: - generated_from_trainer model-index: - name: RoFormer_small-summarization results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # RoFormer_small-summarization This model is a fine-tuned version of [](https://huggingface.co/) on the xsum dataset. It achieves the following results on the evaluation set: - Loss: 4.3373 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.001 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - gradient_accumulation_steps: 4 - total_train_batch_size: 32 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: cosine - lr_scheduler_warmup_steps: 100 - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| \|:-------------:\|:-----:\|:-----:\|:---------------:\| \| 4.975 \| 1.0 \| 5762 \| 4.4897 \| \| 4.4149 \| 2.0 \| 11525 \| 4.3647 \| \| 4.3296 \| 3.0 \| 17286 \| 4.3373 \| ### Framework versions - Transformers 4.35.2 - Pytorch 2.1.0+cu118 - Datasets 2.15.0 - Tokenizers 0.15.0	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # RoFormer_small-summarization This model is a fine-tuned version of [](https://huggingface.co/) on the xsum dataset. It achieves the following results on the evaluation set: - Loss: 4.3373 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.001 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - gradient_accumulation_steps: 4 - total_train_batch_size: 32 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: cosine - lr_scheduler_warmup_steps: 100 - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| \|:-------------:\|:-----:\|:-----:\|:---------------:\| \| 4.975 \| 1.0 \| 5762 \| 4.4897 \| \| 4.4149 \| 2.0 \| 11525 \| 4.3647 \| \| 4.3296 \| 3.0 \| 17286 \| 4.3373 \| ### Framework versions - Transformers 4.35.2 - Pytorch 2.1.0+cu118 - Datasets 2.15.0 - Tokenizers 0.15.0	{"datasets": ["xsum"], "tags": ["generated_from_trainer"], "model-index": [{"name": "RoFormer_small-summarization", "results": []}]}	task	[ "SUMMARIZATION" ]	43,143
Helsinki-NLP/opus-mt-en-kg	Helsinki-NLP	translation	[ "transformers", "pytorch", "tf", "marian", "text2text-generation", "translation", "en", "kg", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:04Z	2023-08-16T11:30:09+00:00	54	0	--- license: apache-2.0 tags: - translation --- ### opus-mt-en-kg * source languages: en * target languages: kg * OPUS readme: [en-kg](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/en-kg/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-08.zip](https://object.pouta.csc.fi/OPUS-MT-models/en-kg/opus-2020-01-08.zip) * test set translations: [opus-2020-01-08.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-kg/opus-2020-01-08.test.txt) * test set scores: [opus-2020-01-08.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-kg/opus-2020-01-08.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.en.kg \| 39.6 \| 0.613 \|	null	Non_BioNLP	### opus-mt-en-kg * source languages: en * target languages: kg * OPUS readme: [en-kg](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/en-kg/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-08.zip](https://object.pouta.csc.fi/OPUS-MT-models/en-kg/opus-2020-01-08.zip) * test set translations: [opus-2020-01-08.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-kg/opus-2020-01-08.test.txt) * test set scores: [opus-2020-01-08.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-kg/opus-2020-01-08.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.en.kg \| 39.6 \| 0.613 \|	{"license": "apache-2.0", "tags": ["translation"]}	task	[ "TRANSLATION" ]	43,144
scampion/EUBERT	scampion	fill-mask	[ "transformers", "pytorch", "roberta", "fill-mask", "generated_from_trainer", "bg", "cs", "da", "de", "el", "en", "es", "et", "fi", "fr", "ga", "hr", "hu", "it", "lt", "lv", "mt", "nl", "pl", "pt", "ro", "sk", "sl", "sv", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-10-02T17:07:34Z	2023-11-10T17:12:30+00:00	33	0	--- language: - bg - cs - da - de - el - en - es - et - fi - fr - ga - hr - hu - it - lt - lv - mt - nl - pl - pt - ro - sk - sl - sv tags: - generated_from_trainer widget: - text: The transition to a climate neutral, sustainable, energy and resource-efficient, circular and fair economy is key to ensuring the long-term competitiveness of the economy of the union and the well-being of its peoples. In 2016, the Union concluded the Paris Agreement2. Article 2(1), point (c), of the Paris Agreement sets out the objective of strengthening the response to climate change by, among other means, making finance flows consistent with a pathway towards low greenhouse gas [MASK] and climate resilient development. model-index: - name: EUBERT results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> ## Model Card: EUBERT ### Overview - Model Name: EUBERT - Model Version: 1.1 - Date of Release: 16 October 2023 - Model Architecture: BERT (Bidirectional Encoder Representations from Transformers) - Training Data: Documents registered by the European Publications Office - Model Use Case: Text Classification, Question Answering, Language Understanding ![EUBERT](https://huggingface.co/EuropeanParliament/EUBERT/resolve/main/EUBERT_small.png) ### Model Description EUBERT is a pretrained BERT uncased model that has been trained on a vast corpus of documents registered by the [European Publications Office](https://op.europa.eu/). These documents span the last 30 years, providing a comprehensive dataset that encompasses a wide range of topics and domains. EUBERT is designed to be a versatile language model that can be fine-tuned for various natural language processing tasks, making it a valuable resource for a variety of applications. ### Intended Use EUBERT serves as a starting point for building more specific natural language understanding models. Its versatility makes it suitable for a wide range of tasks, including but not limited to: 1. Text Classification: EUBERT can be fine-tuned for classifying text documents into different categories, making it useful for applications such as sentiment analysis, topic categorization, and spam detection. 2. Question Answering: By fine-tuning EUBERT on question-answering datasets, it can be used to extract answers from text documents, facilitating tasks like information retrieval and document summarization. 3. Language Understanding: EUBERT can be employed for general language understanding tasks, including named entity recognition, part-of-speech tagging, and text generation. ### Performance The specific performance metrics of EUBERT may vary depending on the downstream task and the quality and quantity of training data used for fine-tuning. Users are encouraged to fine-tune the model on their specific task and evaluate its performance accordingly. ### Considerations - Data Privacy and Compliance: Users should ensure that the use of EUBERT complies with all relevant data privacy and compliance regulations, especially when working with sensitive or personally identifiable information. - Fine-Tuning: The effectiveness of EUBERT on a given task depends on the quality and quantity of the training data, as well as the fine-tuning process. Careful experimentation and evaluation are essential to achieve optimal results. - Bias and Fairness: Users should be aware of potential biases in the training data and take appropriate measures to mitigate bias when fine-tuning EUBERT for specific tasks. ### Conclusion EUBERT is a pretrained BERT model that leverages a substantial corpus of documents from the European Publications Office. It offers a versatile foundation for developing natural language processing solutions across a wide range of applications, enabling researchers and developers to create custom models for text classification, question answering, and language understanding tasks. Users are encouraged to exercise diligence in fine-tuning and evaluating the model for their specific use cases while adhering to data privacy and fairness considerations. --- ## Training procedure Dedicated Word Piece tokenizer vocabulary size 216, ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 1.85 ### Training results Coming soon ### Framework versions - Transformers 4.33.3 - Pytorch 2.0.1+cu117 - Datasets 2.14.5 - Tokenizers 0.13.3 ### Infrastructure - Hardware Type: 4 x GPUs 24GB - GPU Days: 16 - Cloud Provider: EuroHPC - Compute Region:** Meluxina # Author(s) Sébastien Campion <[email protected]>	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> ## Model Card: EUBERT ### Overview - Model Name: EUBERT - Model Version: 1.1 - Date of Release: 16 October 2023 - Model Architecture: BERT (Bidirectional Encoder Representations from Transformers) - Training Data: Documents registered by the European Publications Office - Model Use Case: Text Classification, Question Answering, Language Understanding ![EUBERT](https://huggingface.co/EuropeanParliament/EUBERT/resolve/main/EUBERT_small.png) ### Model Description EUBERT is a pretrained BERT uncased model that has been trained on a vast corpus of documents registered by the [European Publications Office](https://op.europa.eu/). These documents span the last 30 years, providing a comprehensive dataset that encompasses a wide range of topics and domains. EUBERT is designed to be a versatile language model that can be fine-tuned for various natural language processing tasks, making it a valuable resource for a variety of applications. ### Intended Use EUBERT serves as a starting point for building more specific natural language understanding models. Its versatility makes it suitable for a wide range of tasks, including but not limited to: 1. Text Classification: EUBERT can be fine-tuned for classifying text documents into different categories, making it useful for applications such as sentiment analysis, topic categorization, and spam detection. 2. Question Answering: By fine-tuning EUBERT on question-answering datasets, it can be used to extract answers from text documents, facilitating tasks like information retrieval and document summarization. 3. Language Understanding: EUBERT can be employed for general language understanding tasks, including named entity recognition, part-of-speech tagging, and text generation. ### Performance The specific performance metrics of EUBERT may vary depending on the downstream task and the quality and quantity of training data used for fine-tuning. Users are encouraged to fine-tune the model on their specific task and evaluate its performance accordingly. ### Considerations - Data Privacy and Compliance: Users should ensure that the use of EUBERT complies with all relevant data privacy and compliance regulations, especially when working with sensitive or personally identifiable information. - Fine-Tuning: The effectiveness of EUBERT on a given task depends on the quality and quantity of the training data, as well as the fine-tuning process. Careful experimentation and evaluation are essential to achieve optimal results. - Bias and Fairness: Users should be aware of potential biases in the training data and take appropriate measures to mitigate bias when fine-tuning EUBERT for specific tasks. ### Conclusion EUBERT is a pretrained BERT model that leverages a substantial corpus of documents from the European Publications Office. It offers a versatile foundation for developing natural language processing solutions across a wide range of applications, enabling researchers and developers to create custom models for text classification, question answering, and language understanding tasks. Users are encouraged to exercise diligence in fine-tuning and evaluating the model for their specific use cases while adhering to data privacy and fairness considerations. --- ## Training procedure Dedicated Word Piece tokenizer vocabulary size 216, ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 1.85 ### Training results Coming soon ### Framework versions - Transformers 4.33.3 - Pytorch 2.0.1+cu117 - Datasets 2.14.5 - Tokenizers 0.13.3 ### Infrastructure - Hardware Type: 4 x GPUs 24GB - GPU Days: 16 - Cloud Provider: EuroHPC - Compute Region:** Meluxina # Author(s) Sébastien Campion <[email protected]>	{"language": ["bg", "cs", "da", "de", "el", "en", "es", "et", "fi", "fr", "ga", "hr", "hu", "it", "lt", "lv", "mt", "nl", "pl", "pt", "ro", "sk", "sl", "sv"], "tags": ["generated_from_trainer"], "widget": [{"text": "The transition to a climate neutral, sustainable, energy and resource-efficient, circular and fair economy is key to ensuring the long-term competitiveness of the economy of the union and the well-being of its peoples. In 2016, the Union concluded the Paris Agreement2. Article 2(1), point (c), of the Paris Agreement sets out the objective of strengthening the response to climate change by, among other means, making finance flows consistent with a pathway towards low greenhouse gas [MASK] and climate resilient development."}], "model-index": [{"name": "EUBERT", "results": []}]}	task	[ "NAMED_ENTITY_RECOGNITION", "TEXT_CLASSIFICATION", "QUESTION_ANSWERING", "SUMMARIZATION" ]	43,146
Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF	Triangle104	text-generation	[ "transformers", "gguf", "trl", "Reinforcement learning", "llama-cpp", "gguf-my-repo", "text-generation", "en", "base_model:prithivMLmods/Bellatrix-Tiny-1.5B-R1", "base_model:quantized:prithivMLmods/Bellatrix-Tiny-1.5B-R1", "license:apache-2.0", "endpoints_compatible", "region:us", "conversational" ]	2025-02-03T17:58:21Z	2025-02-03T17:59:12+00:00	6	0	--- base_model: prithivMLmods/Bellatrix-Tiny-1.5B-R1 language: - en library_name: transformers license: apache-2.0 pipeline_tag: text-generation tags: - trl - Reinforcement learning - llama-cpp - gguf-my-repo --- # Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF This model was converted to GGUF format from [`prithivMLmods/Bellatrix-Tiny-1.5B-R1`](https://huggingface.co/prithivMLmods/Bellatrix-Tiny-1.5B-R1) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space. Refer to the [original model card](https://huggingface.co/prithivMLmods/Bellatrix-Tiny-1.5B-R1) for more details on the model. --- Bellatrix is based on a reasoning-based model designed for the DeepSeek-R1 synthetic dataset entries. The pipeline's instruction-tuned, text-only models are optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. These models outperform many of the available open-source options. Bellatrix is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions utilize supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF). Use with transformers Starting with transformers >= 4.43.0 onward, you can run conversational inference using the Transformers pipeline abstraction or by leveraging the Auto classes with the generate() function. Make sure to update your transformers installation via pip install --upgrade transformers. import torch from transformers import pipeline model_id = "prithivMLmods/Bellatrix-Tiny-1.5B-R1" pipe = pipeline( "text-generation", model=model_id, torch_dtype=torch.bfloat16, device_map="auto", ) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Who are you?"}, ] outputs = pipe( messages, max_new_tokens=256, ) print(outputs[0]["generated_text"][-1]) Note: You can also find detailed recipes on how to use the model locally, with torch.compile(), assisted generations, quantized and more at huggingface-llama-recipes Intended Use Bellatrix is designed for applications that require advanced reasoning and multilingual dialogue capabilities. It is particularly suitable for: Agentic Retrieval: Enabling intelligent retrieval of relevant information in a dialogue or query-response system. Summarization Tasks: Condensing large bodies of text into concise summaries for easier comprehension. Multilingual Use Cases: Supporting conversations in multiple languages with high accuracy and coherence. Instruction-Based Applications: Following complex, context-aware instructions to generate precise outputs in a variety of scenarios. Limitations Despite its capabilities, Bellatrix has some limitations: Domain Specificity: While it performs well on general tasks, its performance may degrade with highly specialized or niche datasets. Dependence on Training Data: It is only as good as the quality and diversity of its training data, which may lead to biases or inaccuracies. Computational Resources: The model’s optimized transformer architecture can be resource-intensive, requiring significant computational power for fine-tuning and inference. Language Coverage: While multilingual, some languages or dialects may have limited support or lower performance compared to widely used ones. Real-World Contexts: It may struggle with understanding nuanced or ambiguous real-world scenarios not covered during training. --- ## Use with llama.cpp Install llama.cpp through brew (works on Mac and Linux) ```bash brew install llama.cpp ``` Invoke the llama.cpp server or the CLI. ### CLI: ```bash llama-cli --hf-repo Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF --hf-file bellatrix-tiny-1.5b-r1-q5_k_s.gguf -p "The meaning to life and the universe is" ``` ### Server: ```bash llama-server --hf-repo Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF --hf-file bellatrix-tiny-1.5b-r1-q5_k_s.gguf -c 2048 ``` Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well. Step 1: Clone llama.cpp from GitHub. ``` git clone https://github.com/ggerganov/llama.cpp ``` Step 2: Move into the llama.cpp folder and build it with `LLAMA_CURL=1` flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux). ``` cd llama.cpp && LLAMA_CURL=1 make ``` Step 3: Run inference through the main binary. ``` ./llama-cli --hf-repo Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF --hf-file bellatrix-tiny-1.5b-r1-q5_k_s.gguf -p "The meaning to life and the universe is" ``` or ``` ./llama-server --hf-repo Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF --hf-file bellatrix-tiny-1.5b-r1-q5_k_s.gguf -c 2048 ```	null	Non_BioNLP	# Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF This model was converted to GGUF format from [`prithivMLmods/Bellatrix-Tiny-1.5B-R1`](https://huggingface.co/prithivMLmods/Bellatrix-Tiny-1.5B-R1) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space. Refer to the [original model card](https://huggingface.co/prithivMLmods/Bellatrix-Tiny-1.5B-R1) for more details on the model. --- Bellatrix is based on a reasoning-based model designed for the DeepSeek-R1 synthetic dataset entries. The pipeline's instruction-tuned, text-only models are optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. These models outperform many of the available open-source options. Bellatrix is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions utilize supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF). Use with transformers Starting with transformers >= 4.43.0 onward, you can run conversational inference using the Transformers pipeline abstraction or by leveraging the Auto classes with the generate() function. Make sure to update your transformers installation via pip install --upgrade transformers. import torch from transformers import pipeline model_id = "prithivMLmods/Bellatrix-Tiny-1.5B-R1" pipe = pipeline( "text-generation", model=model_id, torch_dtype=torch.bfloat16, device_map="auto", ) messages = [ {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"}, {"role": "user", "content": "Who are you?"}, ] outputs = pipe( messages, max_new_tokens=256, ) print(outputs[0]["generated_text"][-1]) Note: You can also find detailed recipes on how to use the model locally, with torch.compile(), assisted generations, quantized and more at huggingface-llama-recipes Intended Use Bellatrix is designed for applications that require advanced reasoning and multilingual dialogue capabilities. It is particularly suitable for: Agentic Retrieval: Enabling intelligent retrieval of relevant information in a dialogue or query-response system. Summarization Tasks: Condensing large bodies of text into concise summaries for easier comprehension. Multilingual Use Cases: Supporting conversations in multiple languages with high accuracy and coherence. Instruction-Based Applications: Following complex, context-aware instructions to generate precise outputs in a variety of scenarios. Limitations Despite its capabilities, Bellatrix has some limitations: Domain Specificity: While it performs well on general tasks, its performance may degrade with highly specialized or niche datasets. Dependence on Training Data: It is only as good as the quality and diversity of its training data, which may lead to biases or inaccuracies. Computational Resources: The model’s optimized transformer architecture can be resource-intensive, requiring significant computational power for fine-tuning and inference. Language Coverage: While multilingual, some languages or dialects may have limited support or lower performance compared to widely used ones. Real-World Contexts: It may struggle with understanding nuanced or ambiguous real-world scenarios not covered during training. --- ## Use with llama.cpp Install llama.cpp through brew (works on Mac and Linux) ```bash brew install llama.cpp ``` Invoke the llama.cpp server or the CLI. ### CLI: ```bash llama-cli --hf-repo Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF --hf-file bellatrix-tiny-1.5b-r1-q5_k_s.gguf -p "The meaning to life and the universe is" ``` ### Server: ```bash llama-server --hf-repo Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF --hf-file bellatrix-tiny-1.5b-r1-q5_k_s.gguf -c 2048 ``` Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well. Step 1: Clone llama.cpp from GitHub. ``` git clone https://github.com/ggerganov/llama.cpp ``` Step 2: Move into the llama.cpp folder and build it with `LLAMA_CURL=1` flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux). ``` cd llama.cpp && LLAMA_CURL=1 make ``` Step 3: Run inference through the main binary. ``` ./llama-cli --hf-repo Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF --hf-file bellatrix-tiny-1.5b-r1-q5_k_s.gguf -p "The meaning to life and the universe is" ``` or ``` ./llama-server --hf-repo Triangle104/Bellatrix-Tiny-1.5B-R1-Q5_K_S-GGUF --hf-file bellatrix-tiny-1.5b-r1-q5_k_s.gguf -c 2048 ```	{"base_model": "prithivMLmods/Bellatrix-Tiny-1.5B-R1", "language": ["en"], "library_name": "transformers", "license": "apache-2.0", "pipeline_tag": "text-generation", "tags": ["trl", "Reinforcement learning", "llama-cpp", "gguf-my-repo"]}	task	[ "SUMMARIZATION" ]	43,147
hopkins/mbart-finetuned-eng-ind-38	hopkins	translation	[ "transformers", "pytorch", "tensorboard", "mbart", "text2text-generation", "translation", "generated_from_trainer", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-07-03T01:52:19Z	2023-07-03T02:06:04+00:00	8	0	--- metrics: - bleu tags: - translation - generated_from_trainer model-index: - name: mbart-finetuned-eng-ind-38 results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # mbart-finetuned-eng-ind-38 This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset. It achieves the following results on the evaluation set: - Loss: 1.7718 - Bleu: 21.7535 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results ### Framework versions - Transformers 4.26.1 - Pytorch 2.0.1+cu117 - Datasets 2.12.0 - Tokenizers 0.13.3	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # mbart-finetuned-eng-ind-38 This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset. It achieves the following results on the evaluation set: - Loss: 1.7718 - Bleu: 21.7535 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results ### Framework versions - Transformers 4.26.1 - Pytorch 2.0.1+cu117 - Datasets 2.12.0 - Tokenizers 0.13.3	{"metrics": ["bleu"], "tags": ["translation", "generated_from_trainer"], "model-index": [{"name": "mbart-finetuned-eng-ind-38", "results": []}]}	task	[ "TRANSLATION" ]	43,148
thetayne/finetuned_model_0613	thetayne	sentence-similarity	[ "sentence-transformers", "safetensors", "bert", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:1625", "loss:CosineSimilarityLoss", "en", "arxiv:1908.10084", "base_model:BAAI/bge-base-en-v1.5", "base_model:finetune:BAAI/bge-base-en-v1.5", "license:apache-2.0", "model-index", "autotrain_compatible", "text-embeddings-inference", "endpoints_compatible", "region:us" ]	2024-06-13T16:30:47Z	2024-06-13T16:31:03+00:00	5	0	--- base_model: BAAI/bge-base-en-v1.5 datasets: [] language: - en library_name: sentence-transformers license: apache-2.0 metrics: - pearson_cosine - spearman_cosine - pearson_manhattan - spearman_manhattan - pearson_euclidean - spearman_euclidean - pearson_dot - spearman_dot - pearson_max - spearman_max pipeline_tag: sentence-similarity tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:1625 - loss:CosineSimilarityLoss widget: - source_sentence: Boron Steel sentences: - Rock Bit - Spalling Test - Excavator Bucket - source_sentence: Friction Wear sentences: - Tool Steel - Medium Carbon Steel - Diffusion Bonding - source_sentence: Delamination sentences: - Subsea Christmas Tree - Low Alloyed Steel - Screw Conveyors - source_sentence: Nitriding sentences: - Subsea Manifold - Trencher Chain - Cylinder - source_sentence: Corrosion Resistant Coatings sentences: - Mower Blade - Gas Metal Arc Welding (GMAW) - Corrosion Resistant Coatings model-index: - name: BGE base Financial Matryoshka results: - task: type: semantic-similarity name: Semantic Similarity dataset: name: dim 768 type: dim_768 metrics: - type: pearson_cosine value: 0.9548051644723275 name: Pearson Cosine - type: spearman_cosine value: 0.6620048542679903 name: Spearman Cosine - type: pearson_manhattan value: 0.985909077336812 name: Pearson Manhattan - type: spearman_manhattan value: 0.6620048542679903 name: Spearman Manhattan - type: pearson_euclidean value: 0.9863519709955113 name: Pearson Euclidean - type: spearman_euclidean value: 0.6620048542679903 name: Spearman Euclidean - type: pearson_dot value: 0.9548051701614557 name: Pearson Dot - type: spearman_dot value: 0.6610658947764548 name: Spearman Dot - type: pearson_max value: 0.9863519709955113 name: Pearson Max - type: spearman_max value: 0.6620048542679903 name: Spearman Max - task: type: semantic-similarity name: Semantic Similarity dataset: name: dim 512 type: dim_512 metrics: - type: pearson_cosine value: 0.9544417196413574 name: Pearson Cosine - type: spearman_cosine value: 0.6620048542679903 name: Spearman Cosine - type: pearson_manhattan value: 0.9855825558550574 name: Pearson Manhattan - type: spearman_manhattan value: 0.6620048542679903 name: Spearman Manhattan - type: pearson_euclidean value: 0.9862004412296757 name: Pearson Euclidean - type: spearman_euclidean value: 0.6620048542679903 name: Spearman Euclidean - type: pearson_dot value: 0.9501184326722917 name: Pearson Dot - type: spearman_dot value: 0.6607798700248341 name: Spearman Dot - type: pearson_max value: 0.9862004412296757 name: Pearson Max - type: spearman_max value: 0.6620048542679903 name: Spearman Max - task: type: semantic-similarity name: Semantic Similarity dataset: name: dim 256 type: dim_256 metrics: - type: pearson_cosine value: 0.9494511778471465 name: Pearson Cosine - type: spearman_cosine value: 0.6620048542679903 name: Spearman Cosine - type: pearson_manhattan value: 0.9830259644213172 name: Pearson Manhattan - type: spearman_manhattan value: 0.6620048542679903 name: Spearman Manhattan - type: pearson_euclidean value: 0.9835562939431381 name: Pearson Euclidean - type: spearman_euclidean value: 0.6620048542679903 name: Spearman Euclidean - type: pearson_dot value: 0.9469313992827345 name: Pearson Dot - type: spearman_dot value: 0.6607798700248341 name: Spearman Dot - type: pearson_max value: 0.9835562939431381 name: Pearson Max - type: spearman_max value: 0.6620048542679903 name: Spearman Max - task: type: semantic-similarity name: Semantic Similarity dataset: name: dim 128 type: dim_128 metrics: - type: pearson_cosine value: 0.9397052405386266 name: Pearson Cosine - type: spearman_cosine value: 0.6620048542679903 name: Spearman Cosine - type: pearson_manhattan value: 0.9762184586055923 name: Pearson Manhattan - type: spearman_manhattan value: 0.6620048542679903 name: Spearman Manhattan - type: pearson_euclidean value: 0.9781975526221939 name: Pearson Euclidean - type: spearman_euclidean value: 0.6620048542679903 name: Spearman Euclidean - type: pearson_dot value: 0.9271211389022183 name: Pearson Dot - type: spearman_dot value: 0.6607798700248341 name: Spearman Dot - type: pearson_max value: 0.9781975526221939 name: Pearson Max - type: spearman_max value: 0.6620048542679903 name: Spearman Max - task: type: semantic-similarity name: Semantic Similarity dataset: name: dim 64 type: dim_64 metrics: - type: pearson_cosine value: 0.9149032642312528 name: Pearson Cosine - type: spearman_cosine value: 0.6620048542679903 name: Spearman Cosine - type: pearson_manhattan value: 0.968215524939354 name: Pearson Manhattan - type: spearman_manhattan value: 0.6620048542679903 name: Spearman Manhattan - type: pearson_euclidean value: 0.9708485057392984 name: Pearson Euclidean - type: spearman_euclidean value: 0.6620048542679903 name: Spearman Euclidean - type: pearson_dot value: 0.8940456314300972 name: Pearson Dot - type: spearman_dot value: 0.6602255244962898 name: Spearman Dot - type: pearson_max value: 0.9708485057392984 name: Pearson Max - type: spearman_max value: 0.6620048542679903 name: Spearman Max --- # BGE base Financial Matryoshka This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5). It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) <!-- at revision a5beb1e3e68b9ab74eb54cfd186867f64f240e1a --> - Maximum Sequence Length: 512 tokens - Output Dimensionality: 768 tokens - Similarity Function: Cosine Similarity <!-- - Training Dataset: Unknown --> - Language: en - License: apache-2.0 ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("thetayne/finetuned_model_0613") # Run inference sentences = [ 'Corrosion Resistant Coatings', 'Corrosion Resistant Coatings', 'Mower Blade', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 768] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Semantic Similarity * Dataset: `dim_768` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9548 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.9859 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9864 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.9548 \| \| spearman_dot \| 0.6611 \| \| pearson_max \| 0.9864 \| \| spearman_max \| 0.662 \| #### Semantic Similarity * Dataset: `dim_512` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9544 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.9856 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9862 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.9501 \| \| spearman_dot \| 0.6608 \| \| pearson_max \| 0.9862 \| \| spearman_max \| 0.662 \| #### Semantic Similarity * Dataset: `dim_256` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9495 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.983 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9836 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.9469 \| \| spearman_dot \| 0.6608 \| \| pearson_max \| 0.9836 \| \| spearman_max \| 0.662 \| #### Semantic Similarity * Dataset: `dim_128` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9397 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.9762 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9782 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.9271 \| \| spearman_dot \| 0.6608 \| \| pearson_max \| 0.9782 \| \| spearman_max \| 0.662 \| #### Semantic Similarity * Dataset: `dim_64` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9149 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.9682 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9708 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.894 \| \| spearman_dot \| 0.6602 \| \| pearson_max \| 0.9708 \| \| spearman_max \| 0.662 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 1,625 training samples * Columns: <code>sentence_A</code>, <code>sentence_B</code>, and <code>score</code> * Approximate statistics based on the first 1000 samples: \| \| sentence_A \| sentence_B \| score \| \|:--------\|:---------------------------------------------------------------------------------\|:---------------------------------------------------------------------------------\|:------------------------------------------------\| \| type \| string \| string \| int \| \| details \| <ul><li>min: 3 tokens</li><li>mean: 5.68 tokens</li><li>max: 36 tokens</li></ul> \| <ul><li>min: 3 tokens</li><li>mean: 5.73 tokens</li><li>max: 36 tokens</li></ul> \| <ul><li>0: ~83.30%</li><li>1: ~16.70%</li></ul> \| * Samples: \| sentence_A \| sentence_B \| score \| \|:-----------------------------------\|:--------------------------------------\|:---------------\| \| <code>Thermal Fatigue</code> \| <code>Ferritic Stainless Steel</code> \| <code>0</code> \| \| <code>High Temperature Wear</code> \| <code>Drill String</code> \| <code>0</code> \| \| <code>Carbide Coatings</code> \| <code>Carbide Coatings</code> \| <code>1</code> \| * Loss: [<code>CosineSimilarityLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosinesimilarityloss) with these parameters: ```json { "loss_fct": "torch.nn.modules.loss.MSELoss" } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: epoch - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `gradient_accumulation_steps`: 16 - `learning_rate`: 2e-05 - `num_train_epochs`: 4 - `lr_scheduler_type`: cosine - `warmup_ratio`: 0.1 - `bf16`: True - `tf32`: True - `load_best_model_at_end`: True - `optim`: adamw_torch_fused - `batch_sampler`: no_duplicates #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: epoch - `prediction_loss_only`: True - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 16 - `eval_accumulation_steps`: None - `learning_rate`: 2e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 4 - `max_steps`: -1 - `lr_scheduler_type`: cosine - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.1 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: True - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: True - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: True - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch_fused - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional </details> ### Training Logs \| Epoch \| Step \| Training Loss \| dim_128_spearman_cosine \| dim_256_spearman_cosine \| dim_512_spearman_cosine \| dim_64_spearman_cosine \| dim_768_spearman_cosine \| \|:----------:\|:------:\|:-------------:\|:-----------------------:\|:-----------------------:\|:-----------------------:\|:----------------------:\|:-----------------------:\| \| 0 \| 0 \| - \| 0.6626 \| 0.6626 \| 0.6626 \| 0.6626 \| 0.6626 \| \| 0.9412 \| 3 \| - \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| \| 1.8627 \| 6 \| - \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| \| 2.7843 \| 9 \| - \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| \| 3.0784 \| 10 \| 0.156 \| - \| - \| - \| - \| - \| \| 3.7059 \| 12 \| - \| 0.662 \| 0.662 \| 0.662 \| 0.662 \| 0.662 \| * The bold row denotes the saved checkpoint. ### Framework Versions - Python: 3.10.12 - Sentence Transformers: 3.0.1 - Transformers: 4.41.2 - PyTorch: 2.1.2+cu121 - Accelerate: 0.31.0 - Datasets: 2.19.1 - Tokenizers: 0.19.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	null	Non_BioNLP	# BGE base Financial Matryoshka This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5). It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) <!-- at revision a5beb1e3e68b9ab74eb54cfd186867f64f240e1a --> - Maximum Sequence Length: 512 tokens - Output Dimensionality: 768 tokens - Similarity Function: Cosine Similarity <!-- - Training Dataset: Unknown --> - Language: en - License: apache-2.0 ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("thetayne/finetuned_model_0613") # Run inference sentences = [ 'Corrosion Resistant Coatings', 'Corrosion Resistant Coatings', 'Mower Blade', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 768] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Semantic Similarity * Dataset: `dim_768` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9548 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.9859 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9864 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.9548 \| \| spearman_dot \| 0.6611 \| \| pearson_max \| 0.9864 \| \| spearman_max \| 0.662 \| #### Semantic Similarity * Dataset: `dim_512` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9544 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.9856 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9862 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.9501 \| \| spearman_dot \| 0.6608 \| \| pearson_max \| 0.9862 \| \| spearman_max \| 0.662 \| #### Semantic Similarity * Dataset: `dim_256` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9495 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.983 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9836 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.9469 \| \| spearman_dot \| 0.6608 \| \| pearson_max \| 0.9836 \| \| spearman_max \| 0.662 \| #### Semantic Similarity * Dataset: `dim_128` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9397 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.9762 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9782 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.9271 \| \| spearman_dot \| 0.6608 \| \| pearson_max \| 0.9782 \| \| spearman_max \| 0.662 \| #### Semantic Similarity * Dataset: `dim_64` * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) \| Metric \| Value \| \|:--------------------\|:----------\| \| pearson_cosine \| 0.9149 \| \| spearman_cosine \| 0.662 \| \| pearson_manhattan \| 0.9682 \| \| spearman_manhattan \| 0.662 \| \| pearson_euclidean \| 0.9708 \| \| spearman_euclidean \| 0.662 \| \| pearson_dot \| 0.894 \| \| spearman_dot \| 0.6602 \| \| pearson_max \| 0.9708 \| \| spearman_max \| 0.662 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 1,625 training samples * Columns: <code>sentence_A</code>, <code>sentence_B</code>, and <code>score</code> * Approximate statistics based on the first 1000 samples: \| \| sentence_A \| sentence_B \| score \| \|:--------\|:---------------------------------------------------------------------------------\|:---------------------------------------------------------------------------------\|:------------------------------------------------\| \| type \| string \| string \| int \| \| details \| <ul><li>min: 3 tokens</li><li>mean: 5.68 tokens</li><li>max: 36 tokens</li></ul> \| <ul><li>min: 3 tokens</li><li>mean: 5.73 tokens</li><li>max: 36 tokens</li></ul> \| <ul><li>0: ~83.30%</li><li>1: ~16.70%</li></ul> \| * Samples: \| sentence_A \| sentence_B \| score \| \|:-----------------------------------\|:--------------------------------------\|:---------------\| \| <code>Thermal Fatigue</code> \| <code>Ferritic Stainless Steel</code> \| <code>0</code> \| \| <code>High Temperature Wear</code> \| <code>Drill String</code> \| <code>0</code> \| \| <code>Carbide Coatings</code> \| <code>Carbide Coatings</code> \| <code>1</code> \| * Loss: [<code>CosineSimilarityLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosinesimilarityloss) with these parameters: ```json { "loss_fct": "torch.nn.modules.loss.MSELoss" } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: epoch - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `gradient_accumulation_steps`: 16 - `learning_rate`: 2e-05 - `num_train_epochs`: 4 - `lr_scheduler_type`: cosine - `warmup_ratio`: 0.1 - `bf16`: True - `tf32`: True - `load_best_model_at_end`: True - `optim`: adamw_torch_fused - `batch_sampler`: no_duplicates #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: epoch - `prediction_loss_only`: True - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 16 - `eval_accumulation_steps`: None - `learning_rate`: 2e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 4 - `max_steps`: -1 - `lr_scheduler_type`: cosine - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.1 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: True - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: True - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: True - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch_fused - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional </details> ### Training Logs \| Epoch \| Step \| Training Loss \| dim_128_spearman_cosine \| dim_256_spearman_cosine \| dim_512_spearman_cosine \| dim_64_spearman_cosine \| dim_768_spearman_cosine \| \|:----------:\|:------:\|:-------------:\|:-----------------------:\|:-----------------------:\|:-----------------------:\|:----------------------:\|:-----------------------:\| \| 0 \| 0 \| - \| 0.6626 \| 0.6626 \| 0.6626 \| 0.6626 \| 0.6626 \| \| 0.9412 \| 3 \| - \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| \| 1.8627 \| 6 \| - \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| \| 2.7843 \| 9 \| - \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| 0.6620 \| \| 3.0784 \| 10 \| 0.156 \| - \| - \| - \| - \| - \| \| 3.7059 \| 12 \| - \| 0.662 \| 0.662 \| 0.662 \| 0.662 \| 0.662 \| * The bold row denotes the saved checkpoint. ### Framework Versions - Python: 3.10.12 - Sentence Transformers: 3.0.1 - Transformers: 4.41.2 - PyTorch: 2.1.2+cu121 - Accelerate: 0.31.0 - Datasets: 2.19.1 - Tokenizers: 0.19.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	{"base_model": "BAAI/bge-base-en-v1.5", "datasets": [], "language": ["en"], "library_name": "sentence-transformers", "license": "apache-2.0", "metrics": ["pearson_cosine", "spearman_cosine", "pearson_manhattan", "spearman_manhattan", "pearson_euclidean", "spearman_euclidean", "pearson_dot", "spearman_dot", "pearson_max", "spearman_max"], "pipeline_tag": "sentence-similarity", "tags": ["sentence-transformers", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:1625", "loss:CosineSimilarityLoss"], "widget": [{"source_sentence": "Boron Steel", "sentences": ["Rock Bit", "Spalling Test", "Excavator Bucket"]}, {"source_sentence": "Friction Wear", "sentences": ["Tool Steel", "Medium Carbon Steel", "Diffusion Bonding"]}, {"source_sentence": "Delamination", "sentences": ["Subsea Christmas Tree", "Low Alloyed Steel", "Screw Conveyors"]}, {"source_sentence": "Nitriding", "sentences": ["Subsea Manifold", "Trencher Chain", "Cylinder"]}, {"source_sentence": "Corrosion Resistant Coatings", "sentences": ["Mower Blade", "Gas Metal Arc Welding (GMAW)", "Corrosion Resistant Coatings"]}], "model-index": [{"name": "BGE base Financial Matryoshka", "results": [{"task": {"type": "semantic-similarity", "name": "Semantic Similarity"}, "dataset": {"name": "dim 768", "type": "dim_768"}, "metrics": [{"type": "pearson_cosine", "value": 0.9548051644723275, "name": "Pearson Cosine"}, {"type": "spearman_cosine", "value": 0.6620048542679903, "name": "Spearman Cosine"}, {"type": "pearson_manhattan", "value": 0.985909077336812, "name": "Pearson Manhattan"}, {"type": "spearman_manhattan", "value": 0.6620048542679903, "name": "Spearman Manhattan"}, {"type": "pearson_euclidean", "value": 0.9863519709955113, "name": "Pearson Euclidean"}, {"type": "spearman_euclidean", "value": 0.6620048542679903, "name": "Spearman Euclidean"}, {"type": "pearson_dot", "value": 0.9548051701614557, "name": "Pearson Dot"}, {"type": "spearman_dot", "value": 0.6610658947764548, "name": "Spearman Dot"}, {"type": "pearson_max", "value": 0.9863519709955113, "name": "Pearson Max"}, {"type": "spearman_max", "value": 0.6620048542679903, "name": "Spearman Max"}]}, {"task": {"type": "semantic-similarity", "name": "Semantic Similarity"}, "dataset": {"name": "dim 512", "type": "dim_512"}, "metrics": [{"type": "pearson_cosine", "value": 0.9544417196413574, "name": "Pearson Cosine"}, {"type": "spearman_cosine", "value": 0.6620048542679903, "name": "Spearman Cosine"}, {"type": "pearson_manhattan", "value": 0.9855825558550574, "name": "Pearson Manhattan"}, {"type": "spearman_manhattan", "value": 0.6620048542679903, "name": "Spearman Manhattan"}, {"type": "pearson_euclidean", "value": 0.9862004412296757, "name": "Pearson Euclidean"}, {"type": "spearman_euclidean", "value": 0.6620048542679903, "name": "Spearman Euclidean"}, {"type": "pearson_dot", "value": 0.9501184326722917, "name": "Pearson Dot"}, {"type": "spearman_dot", "value": 0.6607798700248341, "name": "Spearman Dot"}, {"type": "pearson_max", "value": 0.9862004412296757, "name": "Pearson Max"}, {"type": "spearman_max", "value": 0.6620048542679903, "name": "Spearman Max"}]}, {"task": {"type": "semantic-similarity", "name": "Semantic Similarity"}, "dataset": {"name": "dim 256", "type": "dim_256"}, "metrics": [{"type": "pearson_cosine", "value": 0.9494511778471465, "name": "Pearson Cosine"}, {"type": "spearman_cosine", "value": 0.6620048542679903, "name": "Spearman Cosine"}, {"type": "pearson_manhattan", "value": 0.9830259644213172, "name": "Pearson Manhattan"}, {"type": "spearman_manhattan", "value": 0.6620048542679903, "name": "Spearman Manhattan"}, {"type": "pearson_euclidean", "value": 0.9835562939431381, "name": "Pearson Euclidean"}, {"type": "spearman_euclidean", "value": 0.6620048542679903, "name": "Spearman Euclidean"}, {"type": "pearson_dot", "value": 0.9469313992827345, "name": "Pearson Dot"}, {"type": "spearman_dot", "value": 0.6607798700248341, "name": "Spearman Dot"}, {"type": "pearson_max", "value": 0.9835562939431381, "name": "Pearson Max"}, {"type": "spearman_max", "value": 0.6620048542679903, "name": "Spearman Max"}]}, {"task": {"type": "semantic-similarity", "name": "Semantic Similarity"}, "dataset": {"name": "dim 128", "type": "dim_128"}, "metrics": [{"type": "pearson_cosine", "value": 0.9397052405386266, "name": "Pearson Cosine"}, {"type": "spearman_cosine", "value": 0.6620048542679903, "name": "Spearman Cosine"}, {"type": "pearson_manhattan", "value": 0.9762184586055923, "name": "Pearson Manhattan"}, {"type": "spearman_manhattan", "value": 0.6620048542679903, "name": "Spearman Manhattan"}, {"type": "pearson_euclidean", "value": 0.9781975526221939, "name": "Pearson Euclidean"}, {"type": "spearman_euclidean", "value": 0.6620048542679903, "name": "Spearman Euclidean"}, {"type": "pearson_dot", "value": 0.9271211389022183, "name": "Pearson Dot"}, {"type": "spearman_dot", "value": 0.6607798700248341, "name": "Spearman Dot"}, {"type": "pearson_max", "value": 0.9781975526221939, "name": "Pearson Max"}, {"type": "spearman_max", "value": 0.6620048542679903, "name": "Spearman Max"}]}, {"task": {"type": "semantic-similarity", "name": "Semantic Similarity"}, "dataset": {"name": "dim 64", "type": "dim_64"}, "metrics": [{"type": "pearson_cosine", "value": 0.9149032642312528, "name": "Pearson Cosine"}, {"type": "spearman_cosine", "value": 0.6620048542679903, "name": "Spearman Cosine"}, {"type": "pearson_manhattan", "value": 0.968215524939354, "name": "Pearson Manhattan"}, {"type": "spearman_manhattan", "value": 0.6620048542679903, "name": "Spearman Manhattan"}, {"type": "pearson_euclidean", "value": 0.9708485057392984, "name": "Pearson Euclidean"}, {"type": "spearman_euclidean", "value": 0.6620048542679903, "name": "Spearman Euclidean"}, {"type": "pearson_dot", "value": 0.8940456314300972, "name": "Pearson Dot"}, {"type": "spearman_dot", "value": 0.6602255244962898, "name": "Spearman Dot"}, {"type": "pearson_max", "value": 0.9708485057392984, "name": "Pearson Max"}, {"type": "spearman_max", "value": 0.6620048542679903, "name": "Spearman Max"}]}]}]}	task	[ "TEXT_CLASSIFICATION", "SEMANTIC_SIMILARITY" ]	43,149
SpicyCorpse/testing1	SpicyCorpse	text-classification	[ "transformers", "tensorboard", "safetensors", "bert", "text-classification", "generated_from_trainer", "dataset:massive", "base_model:google-bert/bert-base-uncased", "base_model:finetune:google-bert/bert-base-uncased", "license:apache-2.0", "model-index", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2023-12-21T14:20:45Z	2023-12-21T22:59:39+00:00	6	0	--- base_model: bert-base-uncased datasets: - massive license: apache-2.0 metrics: - accuracy tags: - generated_from_trainer model-index: - name: testing1 results: - task: type: text-classification name: Text Classification dataset: name: massive type: massive config: en-US split: validation args: en-US metrics: - type: accuracy value: 0.8927693064436792 name: Accuracy --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # testing1 This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the massive dataset. It achieves the following results on the evaluation set: - Loss: 0.7285 - Accuracy: 0.8928 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 33 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 15 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\| \| 1.4751 \| 1.0 \| 720 \| 0.6154 \| 0.8593 \| \| 0.4363 \| 2.0 \| 1440 \| 0.4945 \| 0.8800 \| \| 0.2325 \| 3.0 \| 2160 \| 0.5608 \| 0.8834 \| \| 0.1349 \| 4.0 \| 2880 \| 0.6023 \| 0.8800 \| \| 0.0887 \| 5.0 \| 3600 \| 0.6301 \| 0.8869 \| \| 0.0575 \| 6.0 \| 4320 \| 0.6649 \| 0.8898 \| \| 0.0358 \| 7.0 \| 5040 \| 0.7023 \| 0.8908 \| \| 0.0218 \| 8.0 \| 5760 \| 0.7419 \| 0.8913 \| \| 0.0184 \| 9.0 \| 6480 \| 0.7285 \| 0.8928 \| \| 0.0126 \| 10.0 \| 7200 \| 0.7207 \| 0.8908 \| \| 0.0072 \| 11.0 \| 7920 \| 0.7453 \| 0.8923 \| \| 0.0056 \| 12.0 \| 8640 \| 0.7746 \| 0.8918 \| ### Framework versions - Transformers 4.36.2 - Pytorch 2.0.1+cu117 - Datasets 2.15.0 - Tokenizers 0.15.0	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # testing1 This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the massive dataset. It achieves the following results on the evaluation set: - Loss: 0.7285 - Accuracy: 0.8928 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5e-05 - train_batch_size: 16 - eval_batch_size: 16 - seed: 33 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 15 - mixed_precision_training: Native AMP ### Training results \| Training Loss \| Epoch \| Step \| Validation Loss \| Accuracy \| \|:-------------:\|:-----:\|:----:\|:---------------:\|:--------:\| \| 1.4751 \| 1.0 \| 720 \| 0.6154 \| 0.8593 \| \| 0.4363 \| 2.0 \| 1440 \| 0.4945 \| 0.8800 \| \| 0.2325 \| 3.0 \| 2160 \| 0.5608 \| 0.8834 \| \| 0.1349 \| 4.0 \| 2880 \| 0.6023 \| 0.8800 \| \| 0.0887 \| 5.0 \| 3600 \| 0.6301 \| 0.8869 \| \| 0.0575 \| 6.0 \| 4320 \| 0.6649 \| 0.8898 \| \| 0.0358 \| 7.0 \| 5040 \| 0.7023 \| 0.8908 \| \| 0.0218 \| 8.0 \| 5760 \| 0.7419 \| 0.8913 \| \| 0.0184 \| 9.0 \| 6480 \| 0.7285 \| 0.8928 \| \| 0.0126 \| 10.0 \| 7200 \| 0.7207 \| 0.8908 \| \| 0.0072 \| 11.0 \| 7920 \| 0.7453 \| 0.8923 \| \| 0.0056 \| 12.0 \| 8640 \| 0.7746 \| 0.8918 \| ### Framework versions - Transformers 4.36.2 - Pytorch 2.0.1+cu117 - Datasets 2.15.0 - Tokenizers 0.15.0	{"base_model": "bert-base-uncased", "datasets": ["massive"], "license": "apache-2.0", "metrics": ["accuracy"], "tags": ["generated_from_trainer"], "model-index": [{"name": "testing1", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "massive", "type": "massive", "config": "en-US", "split": "validation", "args": "en-US"}, "metrics": [{"type": "accuracy", "value": 0.8927693064436792, "name": "Accuracy"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,150
TransferGraph/chiragasarpota_scotus-bert-finetuned-lora-tweet_eval_irony	TransferGraph	text-classification	[ "peft", "safetensors", "parquet", "text-classification", "dataset:tweet_eval", "base_model:chiragasarpota/scotus-bert", "base_model:adapter:chiragasarpota/scotus-bert", "license:apache-2.0", "model-index", "region:us" ]	2024-02-27T17:13:14Z	2024-02-29T13:30:25+00:00	4	0	--- base_model: chiragasarpota/scotus-bert datasets: - tweet_eval library_name: peft license: apache-2.0 metrics: - accuracy tags: - parquet - text-classification model-index: - name: chiragasarpota_scotus-bert-finetuned-lora-tweet_eval_irony results: - task: type: text-classification name: Text Classification dataset: name: tweet_eval type: tweet_eval config: irony split: validation args: irony metrics: - type: accuracy value: 0.5015706806282723 name: accuracy --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # chiragasarpota_scotus-bert-finetuned-lora-tweet_eval_irony This model is a fine-tuned version of [chiragasarpota/scotus-bert](https://huggingface.co/chiragasarpota/scotus-bert) on the tweet_eval dataset. It achieves the following results on the evaluation set: - accuracy: 0.5016 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0005 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 8 ### Training results \| accuracy \| train_loss \| epoch \| \|:--------:\|:----------:\|:-----:\| \| 0.5225 \| None \| 0 \| \| 0.5236 \| 0.6944 \| 0 \| \| 0.4775 \| 0.6936 \| 1 \| \| 0.4775 \| 0.6934 \| 2 \| \| 0.4775 \| 0.6926 \| 3 \| \| 0.4775 \| 0.6927 \| 4 \| \| 0.4838 \| 0.6925 \| 5 \| \| 0.4984 \| 0.6918 \| 6 \| \| 0.5016 \| 0.6917 \| 7 \| ### Framework versions - PEFT 0.8.2 - Transformers 4.37.2 - Pytorch 2.2.0 - Datasets 2.16.1 - Tokenizers 0.15.2	null	Non_BioNLP	<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # chiragasarpota_scotus-bert-finetuned-lora-tweet_eval_irony This model is a fine-tuned version of [chiragasarpota/scotus-bert](https://huggingface.co/chiragasarpota/scotus-bert) on the tweet_eval dataset. It achieves the following results on the evaluation set: - accuracy: 0.5016 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0005 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 8 ### Training results \| accuracy \| train_loss \| epoch \| \|:--------:\|:----------:\|:-----:\| \| 0.5225 \| None \| 0 \| \| 0.5236 \| 0.6944 \| 0 \| \| 0.4775 \| 0.6936 \| 1 \| \| 0.4775 \| 0.6934 \| 2 \| \| 0.4775 \| 0.6926 \| 3 \| \| 0.4775 \| 0.6927 \| 4 \| \| 0.4838 \| 0.6925 \| 5 \| \| 0.4984 \| 0.6918 \| 6 \| \| 0.5016 \| 0.6917 \| 7 \| ### Framework versions - PEFT 0.8.2 - Transformers 4.37.2 - Pytorch 2.2.0 - Datasets 2.16.1 - Tokenizers 0.15.2	{"base_model": "chiragasarpota/scotus-bert", "datasets": ["tweet_eval"], "library_name": "peft", "license": "apache-2.0", "metrics": ["accuracy"], "tags": ["parquet", "text-classification"], "model-index": [{"name": "chiragasarpota_scotus-bert-finetuned-lora-tweet_eval_irony", "results": [{"task": {"type": "text-classification", "name": "Text Classification"}, "dataset": {"name": "tweet_eval", "type": "tweet_eval", "config": "irony", "split": "validation", "args": "irony"}, "metrics": [{"type": "accuracy", "value": 0.5015706806282723, "name": "accuracy"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,151
GlandVergil/RFdiffusion	GlandVergil	null	[ "region:us" ]	2024-04-19T09:18:34Z	2024-04-19T09:53:47+00:00	0	1	--- {} --- # RFdiffusion <!-- <img width="1115" alt="Screen Shot 2023-01-19 at 5 56 33 PM" src="https://user-images.githubusercontent.com/56419265/213588200-f8f44dba-276e-4dd2-b844-15acc441458d.png"> --> <p align="center"> <img src="./img/diffusion_protein_gradient_2.jpg" alt="alt text" width="1100px" align="middle"/> </p> Image: Ian C. Haydon / UW Institute for Protein Design ## Description RFdiffusion is an open source method for structure generation, with or without conditional information (a motif, target etc). It can perform a whole range of protein design challenges as we have outlined in [the RFdiffusion paper](https://www.biorxiv.org/content/10.1101/2022.12.09.519842v1). Things Diffusion can do - Motif Scaffolding - Unconditional protein generation - Symmetric unconditional generation (cyclic, dihedral and tetrahedral symmetries currently implemented, more coming!) - Symmetric motif scaffolding - Binder design - Design diversification ("partial diffusion", sampling around a design) ---- # Table of contents - [RFdiffusion](#rfdiffusion) - [Description](#description) - [Table of contents](#table-of-contents) - [Getting started / installation](#getting-started--installation) - [Conda Install SE3-Transformer](#conda-install-se3-transformer) - [Get PPI Scaffold Examples](#get-ppi-scaffold-examples) - [Usage](#usage) - [Running the diffusion script](#running-the-diffusion-script) - [Basic execution - an unconditional monomer](#basic-execution---an-unconditional-monomer) - [Motif Scaffolding](#motif-scaffolding) - [The "active site" model holds very small motifs in place](#the-active-site-model-holds-very-small-motifs-in-place) - [The `inpaint_seq` flag](#the-inpaint_seq-flag) - [A note on `diffuser.T`](#a-note-on-diffusert) - [Partial diffusion](#partial-diffusion) - [Binder Design](#binder-design) - [Practical Considerations for Binder Design](#practical-considerations-for-binder-design) - [Fold Conditioning](#fold-conditioning) - [Generation of Symmetric Oligomers](#generation-of-symmetric-oligomers) - [Using Auxiliary Potentials](#using-auxiliary-potentials) - [Symmetric Motif Scaffolding.](#symmetric-motif-scaffolding) - [A Note on Model Weights](#a-note-on-model-weights) - [Things you might want to play with at inference time](#things-you-might-want-to-play-with-at-inference-time) - [Understanding the output files](#understanding-the-output-files) - [Docker](#docker) - [Conclusion](#conclusion) # Getting started / installation Thanks to Sergey Ovchinnikov, RFdiffusion is available as a [Google Colab Notebook](https://colab.research.google.com/github/sokrypton/ColabDesign/blob/v1.1.1/rf/examples/diffusion.ipynb) if you would like to run it there! We strongly recommend reading this README carefully before getting started with RFdiffusion, and working through some of the examples in the Colab Notebook. If you want to set up RFdiffusion locally, follow the steps below: To get started using RFdiffusion, clone the repo: ``` git clone https://github.com/RosettaCommons/RFdiffusion.git ``` You'll then need to download the model weights into the RFDiffusion directory. ``` cd RFdiffusion mkdir models && cd models wget http://files.ipd.uw.edu/pub/RFdiffusion/6f5902ac237024bdd0c176cb93063dc4/Base_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/e29311f6f1bf1af907f9ef9f44b8328b/Complex_base_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/60f09a193fb5e5ccdc4980417708dbab/Complex_Fold_base_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/74f51cfb8b440f50d70878e05361d8f0/InpaintSeq_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/76d00716416567174cdb7ca96e208296/InpaintSeq_Fold_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/5532d2e1f3a4738decd58b19d633b3c3/ActiveSite_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/12fc204edeae5b57713c5ad7dcb97d39/Base_epoch8_ckpt.pt Optional: wget http://files.ipd.uw.edu/pub/RFdiffusion/f572d396fae9206628714fb2ce00f72e/Complex_beta_ckpt.pt # original structure prediction weights wget http://files.ipd.uw.edu/pub/RFdiffusion/1befcb9b28e2f778f53d47f18b7597fa/RF_structure_prediction_weights.pt ``` ### Conda Install SE3-Transformer Ensure that you have either [Anaconda or Miniconda](https://conda.io/projects/conda/en/latest/user-guide/install/index.html) installed. You also need to install [NVIDIA's implementation of SE(3)-Transformers](https://developer.nvidia.com/blog/accelerating-se3-transformers-training-using-an-nvidia-open-source-model-implementation/) Here is how to install the NVIDIA SE(3)-Transformer code: ``` conda env create -f env/SE3nv.yml conda activate SE3nv cd env/SE3Transformer pip install --no-cache-dir -r requirements.txt python setup.py install cd ../.. # change into the root directory of the repository pip install -e . # install the rfdiffusion module from the root of the repository ``` Anytime you run diffusion you should be sure to activate this conda environment by running the following command: ``` conda activate SE3nv ``` Total setup should take less than 30 minutes on a standard desktop computer. Note: Due to the variation in GPU types and drivers that users have access to, we are not able to make one environment that will run on all setups. As such, we are only providing a yml file with support for CUDA 11.1 and leaving it to each user to customize it to work on their setups. This customization will involve changing the cudatoolkit and (possibly) the PyTorch version specified in the yml file. --- ### Get PPI Scaffold Examples To run the scaffolded protein binder design (PPI) examples, we have provided some example scaffold files (`examples/ppi_scaffolds_subset.tar.gz`). You'll need to untar this: ``` tar -xvf examples/ppi_scaffolds_subset.tar.gz -C examples/ ``` We will explain what these files are and how to use them in the Fold Conditioning section. ---- # Usage In this section we will demonstrate how to run diffusion. <p align="center"> <img src="./img/main.png" alt="alt text" width="1100px" align="middle"/> </p> ### Running the diffusion script The actual script you will execute is called `scripts/run_inference.py`. There are many ways to run it, governed by hydra configs. [Hydra configs](https://hydra.cc/docs/configure_hydra/intro/) are a nice way of being able to specify many different options, with sensible defaults drawn directly from the model checkpoint, so inference should always, by default, match training. What this means is that the default values in `config/inference/base.yml` might not match the actual values used during inference, with a specific checkpoint. This is all handled under the hood. --- ### Basic execution - an unconditional monomer <img src="./img/cropped_uncond.png" alt="alt text" width="400px" align="right"/> Let's first look at how you would do unconditional design of a protein of length 150aa. For this, we just need to specify three things: 1. The length of the protein 2. The location where we want to write files to 3. The number of designs we want ``` ./scripts/run_inference.py 'contigmap.contigs=[150-150]' inference.output_prefix=test_outputs/test inference.num_designs=10 ``` Let's look at this in detail. Firstly, what is `contigmap.contigs`? Hydra configs tell the inference script how it should be run. To keep things organised, the config has different sub-configs, one of them being `contigmap`, which pertains to everything related to the contig string (that defines the protein being built). Take a look at the config file if this isn't clear: `configs/inference/base.yml` Anything in the config can be overwritten manually from the command line. You could, for example, change how the diffuser works: ``` diffuser.crd_scale=0.5 ``` ... but don't do this unless you really know what you're doing!! Now, what does `'contigmap.contigs=[150-150]'` mean? To those who have used RFjoint inpainting, this might look familiar, but a little bit different. Diffusion, in fact, uses the identical 'contig mapper' as inpainting, except that, because we're using hydra, we have to give this to the model in a different way. The contig string has to be passed as a single-item in a list, rather than as a string, for hydra reasons and the entire argument MUST be enclosed in `''` so that the commandline does not attempt to parse any of the special characters. The contig string allows you to specify a length range, but here, we just want a protein of 150aa in length, so you just specify [150-150] This will then run 10 diffusion trajectories, saving the outputs to your specified output folder. NB the first time you run RFdiffusion, it will take a while 'Calculating IGSO3'. Once it has done this, it'll be cached for future reference though! For an additional example of unconditional monomer generation, take a look at `./examples/design_unconditional.sh` in the repo! --- ### Motif Scaffolding <!-- <p align="center"> <img src="./img/motif.png" alt="alt text" width="700px" align="middle"/> </p> --> RFdiffusion can be used to scaffold motifs, in a manner akin to [Constrained Hallucination and RFjoint Inpainting](https://www.science.org/doi/10.1126/science.abn2100#:~:text=The%20binding%20and%20catalytic%20functions%20of%20proteins%20are,the%20fold%20or%20secondary%20structure%20of%20the%20scaffold.). In general, RFdiffusion significantly outperforms both Constrained Hallucination and RFjoint Inpainting. <p align="center"> <img src="./img/motif.png" alt="alt text" width="700px" align="middle"/> </p> When scaffolding protein motifs, we need a way of specifying that we want to scaffold some particular protein input (one or more segments from a `.pdb` file), and to be able to specify how we want these connected, and by how many residues, in the new scaffolded protein. What's more, we want to be able to sample different lengths of connecting protein, as we generally don't know a priori precisely how many residues we'll need to best scaffold a motif. This job of specifying inputs is handled by contigs, governed by the contigmap config in the hydra config. For those familiar with Constrained Hallucination or RFjoint Inpainting, the logic is very similar. Briefly: - Anything prefixed by a letter indicates that this is a motif, with the letter corresponding to the chain letter in the input pdb files. E.g. A10-25 pertains to residues ('A',10),('A',11)...('A',25) in the corresponding input pdb - Anything not prefixed by a letter indicates protein to be built. This can be input as a length range. These length ranges are randomly sampled each iteration of RFdiffusion inference. - To specify chain breaks, we use `/0 `. In more detail, if we want to scaffold a motif, the input is just like RFjoint Inpainting, except needing to navigate the hydra config input. If we want to scaffold residues 10-25 on chain A a pdb, this would be done with `'contigmap.contigs=[5-15/A10-25/30-40]'`. This asks RFdiffusion to build 5-15 residues (randomly sampled at each inference cycle) N-terminally of A10-25 from the input pdb, followed by 30-40 residues (again, randomly sampled) to its C-terminus. If we wanted to ensure the length was always e.g. 55 residues, this can be specified with `contigmap.length=55-55`. You need to obviously also provide a path to your pdb file: `inference.input_pdb=path/to/file.pdb`. It doesn't matter if your input pdb has residues you don't want to scaffold - the contig map defines which residues in the pdb are actually used as the "motif". In other words, even if your pdb files has a B chain, and other residues on the A chain, only A10-25 will be provided to RFdiffusion. To specify that we want to inpaint in the presence of a separate chain, this can be done as follows: ``` 'contigmap.contigs=[5-15/A10-25/30-40/0 B1-100]' ``` Look at this carefully. `/0 ` is the indicator that we want a chain break. NOTE, the space is important here. This tells the diffusion model to add a big residue jump (200aa) to the input, so that the model sees the first chain as being on a separate chain to the second. An example of motif scaffolding can be found in `./examples/design_motifscaffolding.sh`. ### The "active site" model holds very small motifs in place In the RFdiffusion preprint we noted that for very small motifs, RFdiffusion has the tendency to not keep them perfectly fixed in the output. Therefore, for scaffolding minimalist sites such as enzyme active sites, we fine-tuned RFdiffusion on examples similar to these tasks, allowing it to hold smaller motifs better in place, and better generate in silico successes. If your input functional motif is very small, we reccomend using this model, which can easily be specified using the following syntax: `inference.ckpt_override_path=models/ActiveSite_ckpt.pt` ### The `inpaint_seq` flag For those familiar with RFjoint Inpainting, the contigmap.inpaint_seq input is equivalent. The idea is that often, when, for example, fusing two proteins, residues that were on the surface of a protein (and are therefore likely polar), now need to be packed into the 'core' of the protein. We therefore want them to become hydrophobic residues. What we can do, rather than directly mutating them to hydrophobics, is to mask their sequence identity, and allow RFdiffusion to implicitly reason over their sequence, and better pack against them. This requires a different model than the 'base' diffusion model, that has been trained to understand this paradigm, but this is automatically handled by the inference script (you don't need to do anything). To specify amino acids whose sequence should be hidden, use the following syntax: ``` 'contigmap.inpaint_seq=[A1/A30-40]' ``` Here, we're masking the residue identity of residue A1, and all residues between A30 and A40 (inclusive). An example of executing motif scaffolding with the `contigmap.inpaint_seq` flag is located in `./examples/design_motifscaffolding_inpaintseq.sh` ### A note on `diffuser.T` RFdiffusion was originally trained with 200 discrete timesteps. However, recent improvements have allowed us to reduce the number of timesteps we need to use at inference time. In many cases, running with as few as approximately 20 steps provides outputs of equivalent in silico quality to running with 200 steps (providing a 10X speedup). The default is now set to 50 steps. Noting this is important for understanding the partial diffusion, described below. --- ### Partial diffusion Something we can do with diffusion is to partially noise and de-noise a structure, to get some diversity around a general fold. This can work really nicely (see [Vazquez-Torres et al., BioRxiv 2022](https://www.biorxiv.org/content/10.1101/2022.12.10.519862v4.abstract)). This is specified by using the diffuser.parial_T input, and setting a timestep to 'noise' to. <p align="center"> <img src="./img/partial.png" alt="alt text" width="800px" align="middle"/> </p> More noise == more diversity. In Vazquez-Torres et al., 2022, we typically used `diffuser.partial_T` of approximately 80, but this was with respect to the 200 timesteps we were using. Now that the default `diffuser.T` is 50, you will need to adjust diffuser.partial_T accordingly. E.g. now that `diffuser.T=50`, the equivalent of 80 noising steps is `diffuser.partial_T=20`. We strongly recommend sampling different values for `partial_T` however, to find the best parameters for your specific problem. When doing partial diffusion, because we are now diffusing from a known structure, this creates certain constraints. You can still use the contig input, but this has to yield a contig string exactly the same length as the input protein. E.g. if you have a binder:target complex, and you want to diversify the binder (length 100, chain A), you would need to input something like this: ``` 'contigmap.contigs=[100-100/0 B1-150]' diffuser.partial_T=20 ``` The reason for this is that, if your input protein was only 80 amino acids, but you've specified a desired length of 100, we don't know where to diffuse those extra 20 amino acids from, and hence, they will not lie in the distribution that RFdiffusion has learned to denoise from. An example of partial diffusion can be found in `./examples/design_partialdiffusion.sh`! You can also keep parts of the sequence of the diffused chain fixed, if you want. An example of why you might want to do this is in the context of helical peptide binding. If you've threaded a helical peptide sequence onto an ideal helix, and now want to diversify the complex, allowing the helix to be predicted now not as an ideal helix, you might do something like: ``` 'contigmap.contigs=[100-100/0 20-20]' 'contigmap.provide_seq=[100-119]' diffuser.partial_T=10 ``` In this case, the 20aa chain is the helical peptide. The `contigmap.provide_seq` input is zero-indexed, and you can provide a range (so 100-119 is an inclusive range, unmasking the whole sequence of the peptide). Multiple sequence ranges can be provided separated by a comma, e.g. `'contigmap.provide_seq=[172-177,200-205]'`. Note that the provide_seq option requires using a different model checkpoint, but this is automatically handled by the inference script. An example of partial diffusion with providing sequence in diffused regions can be found in `./examples/design_partialdiffusion_withseq.sh`. The same example specifying multiple sequence ranges can be found in `./examples/design_partialdiffusion_multipleseq.sh`. --- ### Binder Design Hopefully, it's now obvious how you might make a binder with diffusion! Indeed, RFdiffusion shows excellent in silico and experimental ability to design de novo binders. <p align="center"> <img src="./img/binder.png" alt="alt text" width="950px" align="middle"/> </p> If chain B is your target, then you could do it like this: ``` ./scripts/run_inference.py 'contigmap.contigs=[B1-100/0 100-100]' inference.output_prefix=test_outputs/binder_test inference.num_designs=10 ``` This will generate 100 residue long binders to residues 1-100 of chain B. However, this probably isn't the best way of making binders. Because diffusion is somewhat computationally-intensive, we need to try and make it as fast as possible. Providing the whole of your target, uncropped, is going to make diffusion very slow if your target is big (and most targets-of-interest, such as cell-surface receptors tend to be very big). One tried-and-true method to speed up binder design is to crop the target protein around the desired interface location. BUT! This creates a problem: if you crop your target and potentially expose hydrophobic core residues which were buried before the crop, how can you guarantee the binder will go to the intended interface site on the surface of the target, and not target the tantalizing hydrophobic patch you have just artificially created? We solve this issue by providing the model with what we call "hotspot residues". The complex models we refer to earlier in this README file have all been trained with hotspot residues, in this training regime, during each example, the model is told (some of) the residues on the target protein which contact the target (i.e., resides that are part of the interface). The model readily learns that it should be making an interface which involved these hotspot residues. At inference time then, we can provide our own hotspot residues to define a region which the binder must contact. These are specified like this: `'ppi.hotspot_res=[A30,A33,A34]'`, where `A` is the chain ID in the input pdb file of the hotspot residue and the number is the residue index in the input pdb file of the hotspot residue. Finally, it has been observed that the default RFdiffusion model often generates mostly helical binders. These have high computational and experimental success rates. However, there may be cases where other kinds of topologies may be desired. For this, we include a "beta" model, which generates a greater diversity of topologies, but has not been extensively experimentally validated. Try this at your own risk: ``` inference.ckpt_override_path=models/Complex_beta_ckpt.pt ``` An example of binder design with RFdiffusion can be found in `./examples/design_ppi.sh`. --- ## Practical Considerations for Binder Design RFdiffusion is an extremely powerful binder design tool but it is not magic. In this section we will walk through some common pitfalls in RFdiffusion binder design and offer advice on how to get the most out of this method. ### Selecting a Target Site Not every site on a target protein is a good candidate for binder design. For a site to be an attractive candidate for binding it should have >~3 hydrophobic residues for the binder to interact with. Binding to charged polar sites is still quite hard. Binding to sites with glycans close to them is also hard since they often become ordered upon binding and you will take an energetic hit for that. Historically, binder design has also avoided unstructured loops, it is not clear if this is still a requirement as RFdiffusion has been used to bind unstructured peptides which share a lot in common with unstructured loops. ### Truncating your Target Protein RFdiffusion scales in runtime as O(N^2) where N is the number of residues in your system. As such, it is a very good idea to truncate large targets so that your computations are not unnecessarily expensive. RFdiffusion and all downstream steps (including AF2) are designed to allow for a truncated target. Truncating a target is an art. For some targets, such as multidomain extracellular membranes, a natural truncation point is where two domains are joined by a flexible linker. For other proteins, such as virus spike proteins, this truncation point is less obvious. Generally you want to preserve secondary structure and introduce as few chain breaks as possible. You should also try to leave ~10A of target protein on each side of your intended target site. We recommend using PyMol to truncate your target protein. ### Picking Hotspots Hotspots are a feature that we integrated into the model to allow for the control of the site on the target which the binder will interact with. In the paper we define a hotspot as a residue on the target protein which is within 10A Cbeta distance of the binder. Of all of the hotspots which are identified on the target 0-20% of these hotspots are actually provided to the model and the rest are masked. This is important for understanding how you should pick hotspots at inference time.; the model is expecting to have to make more contacts than you specify. We normally recommend between 3-6 hotspots, you should run a few pilot runs before generating thousands of designs to make sure the number of hotspots you are providing will give results you like. If you have run the previous PatchDock RifDock binder design pipeline, for the RFdiffusion paper we chose our hotspots to be the PatchDock residues of the target. ### Binder Design Scale In the paper, we generated ~10,000 RFdiffusion binder backbones for each target. From this set of backbones we then generated two sequences per backbone using ProteinMPNN-FastRelax (described below). We screened these ~20,000 designs using AF2 with initial guess and target templating (also described below). Given the high success rates we observed in the paper, for some targets it may be sufficient to only generate ~1,000 RFdiffusion backbones in a campaign. What you want is to get enough designs that pass pAE_interaction < 10 (described more in Binder Design Filtering section) such that you are able to fill a DNA order with these successful designs. We have found that designs that do not pass pAE_interaction < 10 are not worth ordering since they will likely not work experimentally. ### Sequence Design for Binders You may have noticed that the binders designed by RFdiffusion come out with a poly-Glycine sequence. This is not a bug. RFdiffusion is a backbone-generation model and does not generate sequence for the designed region, therefore, another method must be used to assign a sequence to the binders. In the paper we use the ProteinMPNN-FastRelax protocol to do sequence design. We recommend that you do this as well. The code for this protocol can be found in [this GitHub repo](https://github.com/nrbennet/dl_binder_design). While we did not find the FastRelax part of the protocol to yield the large in silico success rate improvements that it yielded with the RifDock-generated docks, it is still a good way to increase your number of shots-on-goal for each (computationally expensive) RFdiffusion backbone. If you would prefer to simply run ProteinMPNN on your binders without the FastRelax step, that will work fine but will be more computationally expensive. ### Binder Design Filtering One of the most important parts of the binder design pipeline is a filtering step to evaluate if your binders are actually predicted to work. In the paper we filtered using AF2 with an initial guess and target templating, scripts for this protocol are available [here](https://github.com/nrbennet/dl_binder_design). We have found that filtering at pae_interaction < 10 is a good predictor of a binder working experimentally. --- ### Fold Conditioning Something that works really well is conditioning binder design (or monomer generation) on particular topologies. This is achieved by providing (partial) secondary structure and block adjacency information (to a model that has been trained to condition on this). <p align="center"> <img src="./img/fold_cond.png" alt="alt text" width="950px" align="middle"/> </p> We are still working out the best way to actually generate this input at inference time, but for now, we have settled upon generating inputs directly from pdb structures. This permits 'low resolution' specification of output topology (i.e., I want a TIM barrel but I don't care precisely where resides are). In `helper_scripts/`, there's a script called `make_secstruc_adj.py`, which can be used as follows: e.g. 1: ``` ./make_secstruc_adj.py --input_pdb ./2KL8.pdb --out_dir /my/dir/for/adj_secstruct ``` or e.g. 2: ``` ./make_secstruc_adj.py --pdb_dir ./pdbs/ --out_dir /my/dir/for/adj_secstruct ``` This will process either a single pdb, or a folder of pdbs, and output a secondary structure and adjacency pytorch file, ready to go into the model. For now (although this might not be necessary), you should also generate these files for the target protein (if you're doing PPI), and provide this to the model. You can then use these at inference as follows: ``` ./scripts/run_inference.py inference.output_prefix=./scaffold_conditioned_test/test scaffoldguided.scaffoldguided=True scaffoldguided.target_pdb=False scaffoldguided.scaffold_dir=./examples/ppi_scaffolds_subset ``` A few extra things: 1) As mentioned above, for PPI, you will want to provide a target protein, along with its secondary structure and block adjacency. This can be done by adding: ``` scaffoldguided.target_pdb=True scaffoldguided.target_path=input_pdbs/insulin_target.pdb inference.output_prefix=insulin_binder/jordi_ss_insulin_noise0_job0 'ppi.hotspot_res=[A59,A83,A91]' scaffoldguided.target_ss=target_folds/insulin_target_ss.pt scaffoldguided.target_adj=target_folds/insulin_target_adj.pt ``` To generate these block adjacency and secondary structure inputs, you can use the helper script. This will now generate 3-helix bundles to the insulin target. For ppi, it's probably also worth adding this flag: ``` scaffoldguided.mask_loops=False ``` This is quite important to understand. During training, we mask some of the secondary structure and block adjacency. This is convenient, because it allows us to, at inference, easily add extra residues without having to specify precise secondary structure for every residue. E.g. if you want to make a long 3 helix bundle, you could mask the loops, and add e.g. 20 more 'mask' tokens to that loop. The model will then (presumbly) choose to make e.g. 15 of these residues into helices (to extend the 3HB), and then make a 5aa loop. But, you didn't have to specify that, which is nice. The way this would be done would be like this: ``` scaffoldguided.mask_loops=True scaffoldguided.sampled_insertion=15 scaffoldguided.sampled_N=5 scaffoldguided.sampled_C=5 ``` This will, at each run of inference, sample up to 15 residues to insert into loops in your 3HB input, and up to 5 additional residues at N and C terminus. This strategy is very useful if you don't have a large set of pdbs to make block adjacencies for. For example, we showed that we could generate loads of lengthened TIM barrels from a single starting pdb with this strategy. However, for PPI, if you're using the provided scaffold sets, it shouldn't be necessary (because there are so many scaffolds to start from, generating extra diversity isn't especially necessary). Finally, if you have a big directory of block adjacency/secondary structure files, but don't want to use all of them, you can make a `.txt` file of the ones you want to use, and pass: ``` scaffoldguided.scaffold_list=path/to/list ``` For PPI, we've consistently seen that reducing the noise added at inference improves designs. This comes at the expense of diversity, but, given that the scaffold sets are huge, this probably doesn't matter too much. We therefore recommend lowering the noise. 0.5 is probably a good compromise: ``` denoiser.noise_scale_ca=0.5 denoiser.noise_scale_frame=0.5 ``` This just scales the amount of noise we add to the translations (`noise_scale_ca`) and rotations (`noise_scale_frame`) by, in this case, 0.5. An additional example of PPI with fold conditioning is available here: `./examples/design_ppi_scaffolded.sh` --- ### Generation of Symmetric Oligomers We're going to switch gears from discussing PPI and look at another task at which RFdiffusion performs well on: symmetric oligomer design. This is done by symmetrising the noise we sample at t=T, and symmetrising the input at every timestep. We have currently implemented the following for use (with the others coming soon!): - Cyclic symmetry - Dihedral symmetry - Tetrahedral symmetry <p align="center"> <img src="./img/olig2.png" alt="alt text" width="1000px" align="middle"/> </p> Here's an example: ``` ./scripts/run_inference.py --config-name symmetry inference.symmetry=tetrahedral 'contigmap.contigs=[360]' inference.output_prefix=test_sample/tetrahedral inference.num_designs=1 ``` Here, we've specified a different `config` file (with `--config-name symmetry`). Because symmetric diffusion is quite different from the diffusion described above, we packaged a whole load of symmetry-related configs into a new file (see `configs/inference/symmetry.yml`). Using this config file now puts diffusion in `symmetry-mode`. The symmetry type is then specified with `inference.symmetry=`. Here, we're specifying tetrahedral symmetry, but you could also choose cyclic (e.g. `c4`) or dihedral (e.g. `d2`). The configmap.contigs length refers to the total length of your oligomer. Therefore, it must be divisible by n chains. More examples of designing oligomers can be found here: `./examples/design_cyclic_oligos.sh`, `./examples/design_dihedral_oligos.sh`, `./examples/design_tetrahedral_oligos.sh`. --- ### Using Auxiliary Potentials Performing diffusion with symmetrized noise may give you the idea that we could use other external interventions during the denoising process to guide diffusion. One such intervention that we have implemented is auxiliary potentials. Auxiliary potentials can be very useful for guiding the inference process. E.g. whereas in RFjoint inpainting, we have little/no control over the final shape of an output, in diffusion we can readily force the network to make, for example, a well-packed protein. This is achieved in the updates we make at each step. Let's go a little deeper into how the diffusion process works: At timestep T (the first step of the reverse-diffusion inference process), we sample noise from a known prior distribution. The model then makes a prediction of what the final structure should be, and we use these two states (noise at time T, prediction of the structure at time 0) to back-calculate where t=T-1 would have been. We therefore have a vector pointing from each coordinate at time T, to their corresponding, back-calculated position at time T-1. But, we want to be able to bias this update, to push the trajectory towards some desired state. This can be done by biasing that vector with another vector, which points towards a position where that residue would reduce the 'loss' as defined by your potential. E.g. if we want to use the `monomer_ROG` potential, which seeks to minimise the radius of gyration of the final protein, if the models prediction of t=0 is very elongated, each of those distant residues will have a larger gradient when we differentiate the `monomer_ROG` potential w.r.t. their positions. These gradients, along with the corresponding scale, can be combined into a vector, which is then combined with the original update vector to make a "biased update" at that timestep. The exact parameters used when applying these potentials matter. If you weight them too strongly, you're not going to end up with a good protein. Too weak, and they'll have little effect. We've explored these potentials in a few different scenarios, and have set sensible defaults, if you want to use them. But, if you feel like they're too weak/strong, or you just fancy exploring, do play with the parameters (in the `potentials` part of the config file). Potentials are specified as a list of strings with each string corresponding to a potential. The argument for potentials is `potentials.guiding_potentials`. Within the string per-potential arguments may be specified in the following syntax: `arg_name1:arg_value1,arg_name2:arg_value2,...,arg_nameN:arg_valueN`. The only argument that is required for each potential is the name of the potential that you wish to apply, the name of this argument is `type` as-in the type of potential you wish to use. Some potentials such as `olig_contacts` and `substrate_contacts` take global options such as `potentials.substrate`, see `config/inference/base.yml` for all the global arguments associated with potentials. Additionally, it is useful to have the effect of the potential "decay" throughout the trajectory, such that in the beginning the effect of the potential is 1x strength, and by the end is much weaker. These decays (`constant`,`linear`,`quadratic`,`cubic`) can be set with the `potentials.guide_decay` argument. Here's an example of how to specify a potential: ``` potentials.guiding_potentials=[\"type:olig_contacts,weight_intra:1,weight_inter:0.1\"] potentials.olig_intra_all=True potentials.olig_inter_all=True potentials.guide_scale=2 potentials.guide_decay='quadratic' ``` We are still fully characterising how/when to use potentials, and we strongly recommend exploring different parameters yourself, as they are clearly somewhat case-dependent. So far, it is clear that they can be helpful for motif scaffolding and symmetric oligomer generation. However, they seem to interact weirdly with hotspot residues in PPI. We think we know why this is, and will work in the coming months to write better potentials for PPI. And please note, it is often good practice to start with no potentials as a baseline, then slowly increase their strength. For the oligomer contacts potentials, start with the ones provided in the examples, and note that the `intra` chain potential often should be higher than the `inter` chain potential. We have already implemented several potentials but it is relatively straightforward to add more, if you want to push your designs towards some specified goal. The only condition is that, whatever potential you write, it is differentiable. Take a look at `potentials.potentials.py` for examples of the potentials we have implemented so far. --- ### Symmetric Motif Scaffolding. We can also combine symmetric diffusion with motif scaffolding to scaffold motifs symmetrically. Currently, we have one way for performing symmetric motif scaffolding. That is by specifying the position of the motif specified w.r.t. the symmetry axes. <p align="center"> <img src="./img/sym_motif.png" alt="alt text" width="1000px" align="middle"/> </p> Special input .pdb and contigs requirements For now, we require that a user have a symmetrized version of their motif in their input pdb for symmetric motif scaffolding. There are two main reasons for this. First, the model is trained by centering any motif at the origin, and thus the code also centers motifs at the origin automatically. Therefore, if your motif is not symmetrized, this centering action will result in an asymmetric unit that now has the origin and axes of symmetry running right through it (bad). Secondly, the diffusion code uses a canonical set of symmetry axes (rotation matrices) to propogate the asymmetric unit of a motif. In order to prevent accidentally running diffusion trajectories which are propogating your motif in ways you don't intend, we require that a user symmetrize an input using the RFdiffusion canonical symmetry axes. RFdiffusion canonical symmetry axes \| Group \| Axis \| \|:----------:\|:-------------:\| \| Cyclic \| Z \| \| Dihedral (cyclic) \| Z \| \| Dihedral (flip/reflection) \| X \| Example: Inputs for symmetric motif scaffolding with motif position specified w.r.t the symmetry axes. This example script `examples/design_nickel.sh` can be used to scaffold the C4 symmetric Nickel binding domains shown in the RFdiffusion paper. It combines many concepts discussed earlier, including symmetric oligomer generation, motif scaffolding, and use of guiding potentials. Note that the contigs should specify something that is precisely symmetric. Things will break if this is not the case. --- ### A Note on Model Weights Because of everything we want diffusion to be able to do, there is not One Model To Rule Them All. E.g., if you want to run with secondary structure conditioning, this requires a different model than if you don't. Under the hood, we take care of most of this by default - we parse your input and work out the most appropriate checkpoint. This is where the config setup is really useful. The exact model checkpoint used at inference contains in it all of the parameters is was trained with, so we can just populate the config file with those values, such that inference runs as designed. If you do want to specify a different checkpoint (if, for example, we train a new model and you want to test it), you just have to make sure it's compatible with what you're doing. E.g. if you try and give secondary structure features to a model that wasn't trained with them, it'll crash. ### Things you might want to play with at inference time Occasionally, it might good to try an alternative model (for example the active site model, or the beta binder model). These can be specified with `inference.ckpt_override_path`. We do not recommend using these outside of the described use cases, however, as there is not a guarantee they will understand other kinds of inputs. For a full list of things that are implemented at inference, see the config file (`configs/inference/base.yml` or `configs/inference/symmetry.yml`). Although you can modify everything, this is not recommended unless you know what you're doing. Generally, don't change the `model`, `preprocess` or `diffuser` configs. These pertain to how the model was trained, so it's unwise to change how you use the model at inference time. However, the parameters below are definitely worth exploring: -inference.final_step: This is when we stop the trajectory. We have seen that you can stop early, and the model is already making a good prediction of the final structure. This speeds up inference. -denoiser.noise_scale_ca and denoiser.noise_scale_frame: These can be used to reduce the noise used during sampling (as discussed for PPI above). The default is 1 (the same noise added at training), but this can be reduced to e.g. 0.5, or even 0. This actually improves the quality of models coming out of diffusion, but at the expense of diversity. If you're not getting any good outputs, or if your problem is very constrained, you could try reducing the noise. While these parameters can be changed independently (for translations and rotations), we recommend keeping them tied. ### Understanding the output files We output several different files. 1. The `.pdb` file. This is the final prediction out of the model. Note that every designed residue is output as a glycine (as we only designed the backbone), and no sidechains are output. This is because, even though RFdiffusion conditions on sidechains in an input motif, there is no loss applied to these predictions, so they can't strictly be trusted. 2. The `.trb` file. This contains useful metadata associated with that specific run, including the specific contig used (if length ranges were sampled), as well as the full config used by RFdiffusion. There are also a few other convenient items in this file: - details about mapping (i.e. how residues in the input map to residues in the output) - `con_ref_pdb_idx`/`con_hal_pdb_idx` - These are two arrays including the input pdb indices (in con_ref_pdb_idx), and where they are in the output pdb (in con_hal_pdb_idx). This only contains the chains where inpainting took place (i.e. not any fixed receptor/target chains) - `con_ref_idx0`/`con_hal_idx0` - These are the same as above, but 0 indexed, and without chain information. This is useful for splicing coordinates out (to assess alignment etc). - `inpaint_seq` - This details any residues that were masked during inference. 3. Trajectory files. By default, we output the full trajectories into the `/traj/` folder. These files can be opened in pymol, as multi-step pdbs. Note that these are ordered in reverse, so the first pdb is technically the last (t=1) prediction made by RFdiffusion during inference. We include both the `pX0` predictions (what the model predicted at each timestep) and the `Xt-1` trajectories (what went into the model at each timestep). ### Docker We have provided a Dockerfile at `docker/Dockerfile` to help run RFDiffusion on HPC and other container orchestration systems. Follow these steps to build and run the container on your system: 1. Clone this repository with `git clone https://github.com/RosettaCommons/RFdiffusion.git` and then `cd RFdiffusion` 1. Verify that the Docker daemon is running on your system with `docker info`. You can find Docker installation instructions for Mac, WIndows, and Linux in the [official Docker docs](https://docs.docker.com/get-docker/). You may also consider [Finch](https://github.com/runfinch/finch), the open source client for container development. 1. Build the container image on your system with `docker build -f docker/Dockerfile -t rfdiffusion .` 1. Create some folders on your file system with `mkdir $HOME/inputs $HOME/outputs $HOME/models` 1. Download the RFDiffusion models with `bash scripts/download_models.sh $HOME/models` 1. Download a test file (or another of your choice) with `wget -P $HOME/inputs https://files.rcsb.org/view/5TPN.pdb` 1. Run the container with the following command: ```bash docker run -it --rm --gpus all \ -v $HOME/models:$HOME/models \ -v $HOME/inputs:$HOME/inputs \ -v $HOME/outputs:$HOME/outputs \ rfdiffusion \ inference.output_prefix=$HOME/outputs/motifscaffolding \ inference.model_directory_path=$HOME/models \ inference.input_pdb=$HOME/inputs/5TPN.pdb \ inference.num_designs=3 \ 'contigmap.contigs=[10-40/A163-181/10-40]' ``` This starts the `rfdiffusion` container, mounts the models, inputs, and outputs folders, passes all available GPUs, and then calls the `run_inference.py` script with the parameters specified. ### Conclusion We are extremely excited to share RFdiffusion with the wider scientific community. We expect to push some updates as and when we make sizeable improvements in the coming months, so do stay tuned. We realize it may take some time to get used to executing RFdiffusion with perfect syntax (sometimes Hydra is hard), so please don't hesitate to create GitHub issues if you need help, we will respond as often as we can. Now, let's go make some proteins. Have fun! \- Joe, David, Nate, Brian, Jason, and the RFdiffusion team. --- RFdiffusion builds directly on the architecture and trained parameters of RoseTTAFold. We therefore thank Frank DiMaio and Minkyung Baek, who developed RoseTTAFold. RFdiffusion is released under an open source BSD License (see LICENSE file). It is free for both non-profit and for-profit use.	null	Non_BioNLP	# RFdiffusion <!-- <img width="1115" alt="Screen Shot 2023-01-19 at 5 56 33 PM" src="https://user-images.githubusercontent.com/56419265/213588200-f8f44dba-276e-4dd2-b844-15acc441458d.png"> --> <p align="center"> <img src="./img/diffusion_protein_gradient_2.jpg" alt="alt text" width="1100px" align="middle"/> </p> Image: Ian C. Haydon / UW Institute for Protein Design ## Description RFdiffusion is an open source method for structure generation, with or without conditional information (a motif, target etc). It can perform a whole range of protein design challenges as we have outlined in [the RFdiffusion paper](https://www.biorxiv.org/content/10.1101/2022.12.09.519842v1). Things Diffusion can do - Motif Scaffolding - Unconditional protein generation - Symmetric unconditional generation (cyclic, dihedral and tetrahedral symmetries currently implemented, more coming!) - Symmetric motif scaffolding - Binder design - Design diversification ("partial diffusion", sampling around a design) ---- # Table of contents - [RFdiffusion](#rfdiffusion) - [Description](#description) - [Table of contents](#table-of-contents) - [Getting started / installation](#getting-started--installation) - [Conda Install SE3-Transformer](#conda-install-se3-transformer) - [Get PPI Scaffold Examples](#get-ppi-scaffold-examples) - [Usage](#usage) - [Running the diffusion script](#running-the-diffusion-script) - [Basic execution - an unconditional monomer](#basic-execution---an-unconditional-monomer) - [Motif Scaffolding](#motif-scaffolding) - [The "active site" model holds very small motifs in place](#the-active-site-model-holds-very-small-motifs-in-place) - [The `inpaint_seq` flag](#the-inpaint_seq-flag) - [A note on `diffuser.T`](#a-note-on-diffusert) - [Partial diffusion](#partial-diffusion) - [Binder Design](#binder-design) - [Practical Considerations for Binder Design](#practical-considerations-for-binder-design) - [Fold Conditioning](#fold-conditioning) - [Generation of Symmetric Oligomers](#generation-of-symmetric-oligomers) - [Using Auxiliary Potentials](#using-auxiliary-potentials) - [Symmetric Motif Scaffolding.](#symmetric-motif-scaffolding) - [A Note on Model Weights](#a-note-on-model-weights) - [Things you might want to play with at inference time](#things-you-might-want-to-play-with-at-inference-time) - [Understanding the output files](#understanding-the-output-files) - [Docker](#docker) - [Conclusion](#conclusion) # Getting started / installation Thanks to Sergey Ovchinnikov, RFdiffusion is available as a [Google Colab Notebook](https://colab.research.google.com/github/sokrypton/ColabDesign/blob/v1.1.1/rf/examples/diffusion.ipynb) if you would like to run it there! We strongly recommend reading this README carefully before getting started with RFdiffusion, and working through some of the examples in the Colab Notebook. If you want to set up RFdiffusion locally, follow the steps below: To get started using RFdiffusion, clone the repo: ``` git clone https://github.com/RosettaCommons/RFdiffusion.git ``` You'll then need to download the model weights into the RFDiffusion directory. ``` cd RFdiffusion mkdir models && cd models wget http://files.ipd.uw.edu/pub/RFdiffusion/6f5902ac237024bdd0c176cb93063dc4/Base_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/e29311f6f1bf1af907f9ef9f44b8328b/Complex_base_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/60f09a193fb5e5ccdc4980417708dbab/Complex_Fold_base_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/74f51cfb8b440f50d70878e05361d8f0/InpaintSeq_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/76d00716416567174cdb7ca96e208296/InpaintSeq_Fold_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/5532d2e1f3a4738decd58b19d633b3c3/ActiveSite_ckpt.pt wget http://files.ipd.uw.edu/pub/RFdiffusion/12fc204edeae5b57713c5ad7dcb97d39/Base_epoch8_ckpt.pt Optional: wget http://files.ipd.uw.edu/pub/RFdiffusion/f572d396fae9206628714fb2ce00f72e/Complex_beta_ckpt.pt # original structure prediction weights wget http://files.ipd.uw.edu/pub/RFdiffusion/1befcb9b28e2f778f53d47f18b7597fa/RF_structure_prediction_weights.pt ``` ### Conda Install SE3-Transformer Ensure that you have either [Anaconda or Miniconda](https://conda.io/projects/conda/en/latest/user-guide/install/index.html) installed. You also need to install [NVIDIA's implementation of SE(3)-Transformers](https://developer.nvidia.com/blog/accelerating-se3-transformers-training-using-an-nvidia-open-source-model-implementation/) Here is how to install the NVIDIA SE(3)-Transformer code: ``` conda env create -f env/SE3nv.yml conda activate SE3nv cd env/SE3Transformer pip install --no-cache-dir -r requirements.txt python setup.py install cd ../.. # change into the root directory of the repository pip install -e . # install the rfdiffusion module from the root of the repository ``` Anytime you run diffusion you should be sure to activate this conda environment by running the following command: ``` conda activate SE3nv ``` Total setup should take less than 30 minutes on a standard desktop computer. Note: Due to the variation in GPU types and drivers that users have access to, we are not able to make one environment that will run on all setups. As such, we are only providing a yml file with support for CUDA 11.1 and leaving it to each user to customize it to work on their setups. This customization will involve changing the cudatoolkit and (possibly) the PyTorch version specified in the yml file. --- ### Get PPI Scaffold Examples To run the scaffolded protein binder design (PPI) examples, we have provided some example scaffold files (`examples/ppi_scaffolds_subset.tar.gz`). You'll need to untar this: ``` tar -xvf examples/ppi_scaffolds_subset.tar.gz -C examples/ ``` We will explain what these files are and how to use them in the Fold Conditioning section. ---- # Usage In this section we will demonstrate how to run diffusion. <p align="center"> <img src="./img/main.png" alt="alt text" width="1100px" align="middle"/> </p> ### Running the diffusion script The actual script you will execute is called `scripts/run_inference.py`. There are many ways to run it, governed by hydra configs. [Hydra configs](https://hydra.cc/docs/configure_hydra/intro/) are a nice way of being able to specify many different options, with sensible defaults drawn directly from the model checkpoint, so inference should always, by default, match training. What this means is that the default values in `config/inference/base.yml` might not match the actual values used during inference, with a specific checkpoint. This is all handled under the hood. --- ### Basic execution - an unconditional monomer <img src="./img/cropped_uncond.png" alt="alt text" width="400px" align="right"/> Let's first look at how you would do unconditional design of a protein of length 150aa. For this, we just need to specify three things: 1. The length of the protein 2. The location where we want to write files to 3. The number of designs we want ``` ./scripts/run_inference.py 'contigmap.contigs=[150-150]' inference.output_prefix=test_outputs/test inference.num_designs=10 ``` Let's look at this in detail. Firstly, what is `contigmap.contigs`? Hydra configs tell the inference script how it should be run. To keep things organised, the config has different sub-configs, one of them being `contigmap`, which pertains to everything related to the contig string (that defines the protein being built). Take a look at the config file if this isn't clear: `configs/inference/base.yml` Anything in the config can be overwritten manually from the command line. You could, for example, change how the diffuser works: ``` diffuser.crd_scale=0.5 ``` ... but don't do this unless you really know what you're doing!! Now, what does `'contigmap.contigs=[150-150]'` mean? To those who have used RFjoint inpainting, this might look familiar, but a little bit different. Diffusion, in fact, uses the identical 'contig mapper' as inpainting, except that, because we're using hydra, we have to give this to the model in a different way. The contig string has to be passed as a single-item in a list, rather than as a string, for hydra reasons and the entire argument MUST be enclosed in `''` so that the commandline does not attempt to parse any of the special characters. The contig string allows you to specify a length range, but here, we just want a protein of 150aa in length, so you just specify [150-150] This will then run 10 diffusion trajectories, saving the outputs to your specified output folder. NB the first time you run RFdiffusion, it will take a while 'Calculating IGSO3'. Once it has done this, it'll be cached for future reference though! For an additional example of unconditional monomer generation, take a look at `./examples/design_unconditional.sh` in the repo! --- ### Motif Scaffolding <!-- <p align="center"> <img src="./img/motif.png" alt="alt text" width="700px" align="middle"/> </p> --> RFdiffusion can be used to scaffold motifs, in a manner akin to [Constrained Hallucination and RFjoint Inpainting](https://www.science.org/doi/10.1126/science.abn2100#:~:text=The%20binding%20and%20catalytic%20functions%20of%20proteins%20are,the%20fold%20or%20secondary%20structure%20of%20the%20scaffold.). In general, RFdiffusion significantly outperforms both Constrained Hallucination and RFjoint Inpainting. <p align="center"> <img src="./img/motif.png" alt="alt text" width="700px" align="middle"/> </p> When scaffolding protein motifs, we need a way of specifying that we want to scaffold some particular protein input (one or more segments from a `.pdb` file), and to be able to specify how we want these connected, and by how many residues, in the new scaffolded protein. What's more, we want to be able to sample different lengths of connecting protein, as we generally don't know a priori precisely how many residues we'll need to best scaffold a motif. This job of specifying inputs is handled by contigs, governed by the contigmap config in the hydra config. For those familiar with Constrained Hallucination or RFjoint Inpainting, the logic is very similar. Briefly: - Anything prefixed by a letter indicates that this is a motif, with the letter corresponding to the chain letter in the input pdb files. E.g. A10-25 pertains to residues ('A',10),('A',11)...('A',25) in the corresponding input pdb - Anything not prefixed by a letter indicates protein to be built. This can be input as a length range. These length ranges are randomly sampled each iteration of RFdiffusion inference. - To specify chain breaks, we use `/0 `. In more detail, if we want to scaffold a motif, the input is just like RFjoint Inpainting, except needing to navigate the hydra config input. If we want to scaffold residues 10-25 on chain A a pdb, this would be done with `'contigmap.contigs=[5-15/A10-25/30-40]'`. This asks RFdiffusion to build 5-15 residues (randomly sampled at each inference cycle) N-terminally of A10-25 from the input pdb, followed by 30-40 residues (again, randomly sampled) to its C-terminus. If we wanted to ensure the length was always e.g. 55 residues, this can be specified with `contigmap.length=55-55`. You need to obviously also provide a path to your pdb file: `inference.input_pdb=path/to/file.pdb`. It doesn't matter if your input pdb has residues you don't want to scaffold - the contig map defines which residues in the pdb are actually used as the "motif". In other words, even if your pdb files has a B chain, and other residues on the A chain, only A10-25 will be provided to RFdiffusion. To specify that we want to inpaint in the presence of a separate chain, this can be done as follows: ``` 'contigmap.contigs=[5-15/A10-25/30-40/0 B1-100]' ``` Look at this carefully. `/0 ` is the indicator that we want a chain break. NOTE, the space is important here. This tells the diffusion model to add a big residue jump (200aa) to the input, so that the model sees the first chain as being on a separate chain to the second. An example of motif scaffolding can be found in `./examples/design_motifscaffolding.sh`. ### The "active site" model holds very small motifs in place In the RFdiffusion preprint we noted that for very small motifs, RFdiffusion has the tendency to not keep them perfectly fixed in the output. Therefore, for scaffolding minimalist sites such as enzyme active sites, we fine-tuned RFdiffusion on examples similar to these tasks, allowing it to hold smaller motifs better in place, and better generate in silico successes. If your input functional motif is very small, we reccomend using this model, which can easily be specified using the following syntax: `inference.ckpt_override_path=models/ActiveSite_ckpt.pt` ### The `inpaint_seq` flag For those familiar with RFjoint Inpainting, the contigmap.inpaint_seq input is equivalent. The idea is that often, when, for example, fusing two proteins, residues that were on the surface of a protein (and are therefore likely polar), now need to be packed into the 'core' of the protein. We therefore want them to become hydrophobic residues. What we can do, rather than directly mutating them to hydrophobics, is to mask their sequence identity, and allow RFdiffusion to implicitly reason over their sequence, and better pack against them. This requires a different model than the 'base' diffusion model, that has been trained to understand this paradigm, but this is automatically handled by the inference script (you don't need to do anything). To specify amino acids whose sequence should be hidden, use the following syntax: ``` 'contigmap.inpaint_seq=[A1/A30-40]' ``` Here, we're masking the residue identity of residue A1, and all residues between A30 and A40 (inclusive). An example of executing motif scaffolding with the `contigmap.inpaint_seq` flag is located in `./examples/design_motifscaffolding_inpaintseq.sh` ### A note on `diffuser.T` RFdiffusion was originally trained with 200 discrete timesteps. However, recent improvements have allowed us to reduce the number of timesteps we need to use at inference time. In many cases, running with as few as approximately 20 steps provides outputs of equivalent in silico quality to running with 200 steps (providing a 10X speedup). The default is now set to 50 steps. Noting this is important for understanding the partial diffusion, described below. --- ### Partial diffusion Something we can do with diffusion is to partially noise and de-noise a structure, to get some diversity around a general fold. This can work really nicely (see [Vazquez-Torres et al., BioRxiv 2022](https://www.biorxiv.org/content/10.1101/2022.12.10.519862v4.abstract)). This is specified by using the diffuser.parial_T input, and setting a timestep to 'noise' to. <p align="center"> <img src="./img/partial.png" alt="alt text" width="800px" align="middle"/> </p> More noise == more diversity. In Vazquez-Torres et al., 2022, we typically used `diffuser.partial_T` of approximately 80, but this was with respect to the 200 timesteps we were using. Now that the default `diffuser.T` is 50, you will need to adjust diffuser.partial_T accordingly. E.g. now that `diffuser.T=50`, the equivalent of 80 noising steps is `diffuser.partial_T=20`. We strongly recommend sampling different values for `partial_T` however, to find the best parameters for your specific problem. When doing partial diffusion, because we are now diffusing from a known structure, this creates certain constraints. You can still use the contig input, but this has to yield a contig string exactly the same length as the input protein. E.g. if you have a binder:target complex, and you want to diversify the binder (length 100, chain A), you would need to input something like this: ``` 'contigmap.contigs=[100-100/0 B1-150]' diffuser.partial_T=20 ``` The reason for this is that, if your input protein was only 80 amino acids, but you've specified a desired length of 100, we don't know where to diffuse those extra 20 amino acids from, and hence, they will not lie in the distribution that RFdiffusion has learned to denoise from. An example of partial diffusion can be found in `./examples/design_partialdiffusion.sh`! You can also keep parts of the sequence of the diffused chain fixed, if you want. An example of why you might want to do this is in the context of helical peptide binding. If you've threaded a helical peptide sequence onto an ideal helix, and now want to diversify the complex, allowing the helix to be predicted now not as an ideal helix, you might do something like: ``` 'contigmap.contigs=[100-100/0 20-20]' 'contigmap.provide_seq=[100-119]' diffuser.partial_T=10 ``` In this case, the 20aa chain is the helical peptide. The `contigmap.provide_seq` input is zero-indexed, and you can provide a range (so 100-119 is an inclusive range, unmasking the whole sequence of the peptide). Multiple sequence ranges can be provided separated by a comma, e.g. `'contigmap.provide_seq=[172-177,200-205]'`. Note that the provide_seq option requires using a different model checkpoint, but this is automatically handled by the inference script. An example of partial diffusion with providing sequence in diffused regions can be found in `./examples/design_partialdiffusion_withseq.sh`. The same example specifying multiple sequence ranges can be found in `./examples/design_partialdiffusion_multipleseq.sh`. --- ### Binder Design Hopefully, it's now obvious how you might make a binder with diffusion! Indeed, RFdiffusion shows excellent in silico and experimental ability to design de novo binders. <p align="center"> <img src="./img/binder.png" alt="alt text" width="950px" align="middle"/> </p> If chain B is your target, then you could do it like this: ``` ./scripts/run_inference.py 'contigmap.contigs=[B1-100/0 100-100]' inference.output_prefix=test_outputs/binder_test inference.num_designs=10 ``` This will generate 100 residue long binders to residues 1-100 of chain B. However, this probably isn't the best way of making binders. Because diffusion is somewhat computationally-intensive, we need to try and make it as fast as possible. Providing the whole of your target, uncropped, is going to make diffusion very slow if your target is big (and most targets-of-interest, such as cell-surface receptors tend to be very big). One tried-and-true method to speed up binder design is to crop the target protein around the desired interface location. BUT! This creates a problem: if you crop your target and potentially expose hydrophobic core residues which were buried before the crop, how can you guarantee the binder will go to the intended interface site on the surface of the target, and not target the tantalizing hydrophobic patch you have just artificially created? We solve this issue by providing the model with what we call "hotspot residues". The complex models we refer to earlier in this README file have all been trained with hotspot residues, in this training regime, during each example, the model is told (some of) the residues on the target protein which contact the target (i.e., resides that are part of the interface). The model readily learns that it should be making an interface which involved these hotspot residues. At inference time then, we can provide our own hotspot residues to define a region which the binder must contact. These are specified like this: `'ppi.hotspot_res=[A30,A33,A34]'`, where `A` is the chain ID in the input pdb file of the hotspot residue and the number is the residue index in the input pdb file of the hotspot residue. Finally, it has been observed that the default RFdiffusion model often generates mostly helical binders. These have high computational and experimental success rates. However, there may be cases where other kinds of topologies may be desired. For this, we include a "beta" model, which generates a greater diversity of topologies, but has not been extensively experimentally validated. Try this at your own risk: ``` inference.ckpt_override_path=models/Complex_beta_ckpt.pt ``` An example of binder design with RFdiffusion can be found in `./examples/design_ppi.sh`. --- ## Practical Considerations for Binder Design RFdiffusion is an extremely powerful binder design tool but it is not magic. In this section we will walk through some common pitfalls in RFdiffusion binder design and offer advice on how to get the most out of this method. ### Selecting a Target Site Not every site on a target protein is a good candidate for binder design. For a site to be an attractive candidate for binding it should have >~3 hydrophobic residues for the binder to interact with. Binding to charged polar sites is still quite hard. Binding to sites with glycans close to them is also hard since they often become ordered upon binding and you will take an energetic hit for that. Historically, binder design has also avoided unstructured loops, it is not clear if this is still a requirement as RFdiffusion has been used to bind unstructured peptides which share a lot in common with unstructured loops. ### Truncating your Target Protein RFdiffusion scales in runtime as O(N^2) where N is the number of residues in your system. As such, it is a very good idea to truncate large targets so that your computations are not unnecessarily expensive. RFdiffusion and all downstream steps (including AF2) are designed to allow for a truncated target. Truncating a target is an art. For some targets, such as multidomain extracellular membranes, a natural truncation point is where two domains are joined by a flexible linker. For other proteins, such as virus spike proteins, this truncation point is less obvious. Generally you want to preserve secondary structure and introduce as few chain breaks as possible. You should also try to leave ~10A of target protein on each side of your intended target site. We recommend using PyMol to truncate your target protein. ### Picking Hotspots Hotspots are a feature that we integrated into the model to allow for the control of the site on the target which the binder will interact with. In the paper we define a hotspot as a residue on the target protein which is within 10A Cbeta distance of the binder. Of all of the hotspots which are identified on the target 0-20% of these hotspots are actually provided to the model and the rest are masked. This is important for understanding how you should pick hotspots at inference time.; the model is expecting to have to make more contacts than you specify. We normally recommend between 3-6 hotspots, you should run a few pilot runs before generating thousands of designs to make sure the number of hotspots you are providing will give results you like. If you have run the previous PatchDock RifDock binder design pipeline, for the RFdiffusion paper we chose our hotspots to be the PatchDock residues of the target. ### Binder Design Scale In the paper, we generated ~10,000 RFdiffusion binder backbones for each target. From this set of backbones we then generated two sequences per backbone using ProteinMPNN-FastRelax (described below). We screened these ~20,000 designs using AF2 with initial guess and target templating (also described below). Given the high success rates we observed in the paper, for some targets it may be sufficient to only generate ~1,000 RFdiffusion backbones in a campaign. What you want is to get enough designs that pass pAE_interaction < 10 (described more in Binder Design Filtering section) such that you are able to fill a DNA order with these successful designs. We have found that designs that do not pass pAE_interaction < 10 are not worth ordering since they will likely not work experimentally. ### Sequence Design for Binders You may have noticed that the binders designed by RFdiffusion come out with a poly-Glycine sequence. This is not a bug. RFdiffusion is a backbone-generation model and does not generate sequence for the designed region, therefore, another method must be used to assign a sequence to the binders. In the paper we use the ProteinMPNN-FastRelax protocol to do sequence design. We recommend that you do this as well. The code for this protocol can be found in [this GitHub repo](https://github.com/nrbennet/dl_binder_design). While we did not find the FastRelax part of the protocol to yield the large in silico success rate improvements that it yielded with the RifDock-generated docks, it is still a good way to increase your number of shots-on-goal for each (computationally expensive) RFdiffusion backbone. If you would prefer to simply run ProteinMPNN on your binders without the FastRelax step, that will work fine but will be more computationally expensive. ### Binder Design Filtering One of the most important parts of the binder design pipeline is a filtering step to evaluate if your binders are actually predicted to work. In the paper we filtered using AF2 with an initial guess and target templating, scripts for this protocol are available [here](https://github.com/nrbennet/dl_binder_design). We have found that filtering at pae_interaction < 10 is a good predictor of a binder working experimentally. --- ### Fold Conditioning Something that works really well is conditioning binder design (or monomer generation) on particular topologies. This is achieved by providing (partial) secondary structure and block adjacency information (to a model that has been trained to condition on this). <p align="center"> <img src="./img/fold_cond.png" alt="alt text" width="950px" align="middle"/> </p> We are still working out the best way to actually generate this input at inference time, but for now, we have settled upon generating inputs directly from pdb structures. This permits 'low resolution' specification of output topology (i.e., I want a TIM barrel but I don't care precisely where resides are). In `helper_scripts/`, there's a script called `make_secstruc_adj.py`, which can be used as follows: e.g. 1: ``` ./make_secstruc_adj.py --input_pdb ./2KL8.pdb --out_dir /my/dir/for/adj_secstruct ``` or e.g. 2: ``` ./make_secstruc_adj.py --pdb_dir ./pdbs/ --out_dir /my/dir/for/adj_secstruct ``` This will process either a single pdb, or a folder of pdbs, and output a secondary structure and adjacency pytorch file, ready to go into the model. For now (although this might not be necessary), you should also generate these files for the target protein (if you're doing PPI), and provide this to the model. You can then use these at inference as follows: ``` ./scripts/run_inference.py inference.output_prefix=./scaffold_conditioned_test/test scaffoldguided.scaffoldguided=True scaffoldguided.target_pdb=False scaffoldguided.scaffold_dir=./examples/ppi_scaffolds_subset ``` A few extra things: 1) As mentioned above, for PPI, you will want to provide a target protein, along with its secondary structure and block adjacency. This can be done by adding: ``` scaffoldguided.target_pdb=True scaffoldguided.target_path=input_pdbs/insulin_target.pdb inference.output_prefix=insulin_binder/jordi_ss_insulin_noise0_job0 'ppi.hotspot_res=[A59,A83,A91]' scaffoldguided.target_ss=target_folds/insulin_target_ss.pt scaffoldguided.target_adj=target_folds/insulin_target_adj.pt ``` To generate these block adjacency and secondary structure inputs, you can use the helper script. This will now generate 3-helix bundles to the insulin target. For ppi, it's probably also worth adding this flag: ``` scaffoldguided.mask_loops=False ``` This is quite important to understand. During training, we mask some of the secondary structure and block adjacency. This is convenient, because it allows us to, at inference, easily add extra residues without having to specify precise secondary structure for every residue. E.g. if you want to make a long 3 helix bundle, you could mask the loops, and add e.g. 20 more 'mask' tokens to that loop. The model will then (presumbly) choose to make e.g. 15 of these residues into helices (to extend the 3HB), and then make a 5aa loop. But, you didn't have to specify that, which is nice. The way this would be done would be like this: ``` scaffoldguided.mask_loops=True scaffoldguided.sampled_insertion=15 scaffoldguided.sampled_N=5 scaffoldguided.sampled_C=5 ``` This will, at each run of inference, sample up to 15 residues to insert into loops in your 3HB input, and up to 5 additional residues at N and C terminus. This strategy is very useful if you don't have a large set of pdbs to make block adjacencies for. For example, we showed that we could generate loads of lengthened TIM barrels from a single starting pdb with this strategy. However, for PPI, if you're using the provided scaffold sets, it shouldn't be necessary (because there are so many scaffolds to start from, generating extra diversity isn't especially necessary). Finally, if you have a big directory of block adjacency/secondary structure files, but don't want to use all of them, you can make a `.txt` file of the ones you want to use, and pass: ``` scaffoldguided.scaffold_list=path/to/list ``` For PPI, we've consistently seen that reducing the noise added at inference improves designs. This comes at the expense of diversity, but, given that the scaffold sets are huge, this probably doesn't matter too much. We therefore recommend lowering the noise. 0.5 is probably a good compromise: ``` denoiser.noise_scale_ca=0.5 denoiser.noise_scale_frame=0.5 ``` This just scales the amount of noise we add to the translations (`noise_scale_ca`) and rotations (`noise_scale_frame`) by, in this case, 0.5. An additional example of PPI with fold conditioning is available here: `./examples/design_ppi_scaffolded.sh` --- ### Generation of Symmetric Oligomers We're going to switch gears from discussing PPI and look at another task at which RFdiffusion performs well on: symmetric oligomer design. This is done by symmetrising the noise we sample at t=T, and symmetrising the input at every timestep. We have currently implemented the following for use (with the others coming soon!): - Cyclic symmetry - Dihedral symmetry - Tetrahedral symmetry <p align="center"> <img src="./img/olig2.png" alt="alt text" width="1000px" align="middle"/> </p> Here's an example: ``` ./scripts/run_inference.py --config-name symmetry inference.symmetry=tetrahedral 'contigmap.contigs=[360]' inference.output_prefix=test_sample/tetrahedral inference.num_designs=1 ``` Here, we've specified a different `config` file (with `--config-name symmetry`). Because symmetric diffusion is quite different from the diffusion described above, we packaged a whole load of symmetry-related configs into a new file (see `configs/inference/symmetry.yml`). Using this config file now puts diffusion in `symmetry-mode`. The symmetry type is then specified with `inference.symmetry=`. Here, we're specifying tetrahedral symmetry, but you could also choose cyclic (e.g. `c4`) or dihedral (e.g. `d2`). The configmap.contigs length refers to the total length of your oligomer. Therefore, it must be divisible by n chains. More examples of designing oligomers can be found here: `./examples/design_cyclic_oligos.sh`, `./examples/design_dihedral_oligos.sh`, `./examples/design_tetrahedral_oligos.sh`. --- ### Using Auxiliary Potentials Performing diffusion with symmetrized noise may give you the idea that we could use other external interventions during the denoising process to guide diffusion. One such intervention that we have implemented is auxiliary potentials. Auxiliary potentials can be very useful for guiding the inference process. E.g. whereas in RFjoint inpainting, we have little/no control over the final shape of an output, in diffusion we can readily force the network to make, for example, a well-packed protein. This is achieved in the updates we make at each step. Let's go a little deeper into how the diffusion process works: At timestep T (the first step of the reverse-diffusion inference process), we sample noise from a known prior distribution. The model then makes a prediction of what the final structure should be, and we use these two states (noise at time T, prediction of the structure at time 0) to back-calculate where t=T-1 would have been. We therefore have a vector pointing from each coordinate at time T, to their corresponding, back-calculated position at time T-1. But, we want to be able to bias this update, to push the trajectory towards some desired state. This can be done by biasing that vector with another vector, which points towards a position where that residue would reduce the 'loss' as defined by your potential. E.g. if we want to use the `monomer_ROG` potential, which seeks to minimise the radius of gyration of the final protein, if the models prediction of t=0 is very elongated, each of those distant residues will have a larger gradient when we differentiate the `monomer_ROG` potential w.r.t. their positions. These gradients, along with the corresponding scale, can be combined into a vector, which is then combined with the original update vector to make a "biased update" at that timestep. The exact parameters used when applying these potentials matter. If you weight them too strongly, you're not going to end up with a good protein. Too weak, and they'll have little effect. We've explored these potentials in a few different scenarios, and have set sensible defaults, if you want to use them. But, if you feel like they're too weak/strong, or you just fancy exploring, do play with the parameters (in the `potentials` part of the config file). Potentials are specified as a list of strings with each string corresponding to a potential. The argument for potentials is `potentials.guiding_potentials`. Within the string per-potential arguments may be specified in the following syntax: `arg_name1:arg_value1,arg_name2:arg_value2,...,arg_nameN:arg_valueN`. The only argument that is required for each potential is the name of the potential that you wish to apply, the name of this argument is `type` as-in the type of potential you wish to use. Some potentials such as `olig_contacts` and `substrate_contacts` take global options such as `potentials.substrate`, see `config/inference/base.yml` for all the global arguments associated with potentials. Additionally, it is useful to have the effect of the potential "decay" throughout the trajectory, such that in the beginning the effect of the potential is 1x strength, and by the end is much weaker. These decays (`constant`,`linear`,`quadratic`,`cubic`) can be set with the `potentials.guide_decay` argument. Here's an example of how to specify a potential: ``` potentials.guiding_potentials=[\"type:olig_contacts,weight_intra:1,weight_inter:0.1\"] potentials.olig_intra_all=True potentials.olig_inter_all=True potentials.guide_scale=2 potentials.guide_decay='quadratic' ``` We are still fully characterising how/when to use potentials, and we strongly recommend exploring different parameters yourself, as they are clearly somewhat case-dependent. So far, it is clear that they can be helpful for motif scaffolding and symmetric oligomer generation. However, they seem to interact weirdly with hotspot residues in PPI. We think we know why this is, and will work in the coming months to write better potentials for PPI. And please note, it is often good practice to start with no potentials as a baseline, then slowly increase their strength. For the oligomer contacts potentials, start with the ones provided in the examples, and note that the `intra` chain potential often should be higher than the `inter` chain potential. We have already implemented several potentials but it is relatively straightforward to add more, if you want to push your designs towards some specified goal. The only condition is that, whatever potential you write, it is differentiable. Take a look at `potentials.potentials.py` for examples of the potentials we have implemented so far. --- ### Symmetric Motif Scaffolding. We can also combine symmetric diffusion with motif scaffolding to scaffold motifs symmetrically. Currently, we have one way for performing symmetric motif scaffolding. That is by specifying the position of the motif specified w.r.t. the symmetry axes. <p align="center"> <img src="./img/sym_motif.png" alt="alt text" width="1000px" align="middle"/> </p> Special input .pdb and contigs requirements For now, we require that a user have a symmetrized version of their motif in their input pdb for symmetric motif scaffolding. There are two main reasons for this. First, the model is trained by centering any motif at the origin, and thus the code also centers motifs at the origin automatically. Therefore, if your motif is not symmetrized, this centering action will result in an asymmetric unit that now has the origin and axes of symmetry running right through it (bad). Secondly, the diffusion code uses a canonical set of symmetry axes (rotation matrices) to propogate the asymmetric unit of a motif. In order to prevent accidentally running diffusion trajectories which are propogating your motif in ways you don't intend, we require that a user symmetrize an input using the RFdiffusion canonical symmetry axes. RFdiffusion canonical symmetry axes \| Group \| Axis \| \|:----------:\|:-------------:\| \| Cyclic \| Z \| \| Dihedral (cyclic) \| Z \| \| Dihedral (flip/reflection) \| X \| Example: Inputs for symmetric motif scaffolding with motif position specified w.r.t the symmetry axes. This example script `examples/design_nickel.sh` can be used to scaffold the C4 symmetric Nickel binding domains shown in the RFdiffusion paper. It combines many concepts discussed earlier, including symmetric oligomer generation, motif scaffolding, and use of guiding potentials. Note that the contigs should specify something that is precisely symmetric. Things will break if this is not the case. --- ### A Note on Model Weights Because of everything we want diffusion to be able to do, there is not One Model To Rule Them All. E.g., if you want to run with secondary structure conditioning, this requires a different model than if you don't. Under the hood, we take care of most of this by default - we parse your input and work out the most appropriate checkpoint. This is where the config setup is really useful. The exact model checkpoint used at inference contains in it all of the parameters is was trained with, so we can just populate the config file with those values, such that inference runs as designed. If you do want to specify a different checkpoint (if, for example, we train a new model and you want to test it), you just have to make sure it's compatible with what you're doing. E.g. if you try and give secondary structure features to a model that wasn't trained with them, it'll crash. ### Things you might want to play with at inference time Occasionally, it might good to try an alternative model (for example the active site model, or the beta binder model). These can be specified with `inference.ckpt_override_path`. We do not recommend using these outside of the described use cases, however, as there is not a guarantee they will understand other kinds of inputs. For a full list of things that are implemented at inference, see the config file (`configs/inference/base.yml` or `configs/inference/symmetry.yml`). Although you can modify everything, this is not recommended unless you know what you're doing. Generally, don't change the `model`, `preprocess` or `diffuser` configs. These pertain to how the model was trained, so it's unwise to change how you use the model at inference time. However, the parameters below are definitely worth exploring: -inference.final_step: This is when we stop the trajectory. We have seen that you can stop early, and the model is already making a good prediction of the final structure. This speeds up inference. -denoiser.noise_scale_ca and denoiser.noise_scale_frame: These can be used to reduce the noise used during sampling (as discussed for PPI above). The default is 1 (the same noise added at training), but this can be reduced to e.g. 0.5, or even 0. This actually improves the quality of models coming out of diffusion, but at the expense of diversity. If you're not getting any good outputs, or if your problem is very constrained, you could try reducing the noise. While these parameters can be changed independently (for translations and rotations), we recommend keeping them tied. ### Understanding the output files We output several different files. 1. The `.pdb` file. This is the final prediction out of the model. Note that every designed residue is output as a glycine (as we only designed the backbone), and no sidechains are output. This is because, even though RFdiffusion conditions on sidechains in an input motif, there is no loss applied to these predictions, so they can't strictly be trusted. 2. The `.trb` file. This contains useful metadata associated with that specific run, including the specific contig used (if length ranges were sampled), as well as the full config used by RFdiffusion. There are also a few other convenient items in this file: - details about mapping (i.e. how residues in the input map to residues in the output) - `con_ref_pdb_idx`/`con_hal_pdb_idx` - These are two arrays including the input pdb indices (in con_ref_pdb_idx), and where they are in the output pdb (in con_hal_pdb_idx). This only contains the chains where inpainting took place (i.e. not any fixed receptor/target chains) - `con_ref_idx0`/`con_hal_idx0` - These are the same as above, but 0 indexed, and without chain information. This is useful for splicing coordinates out (to assess alignment etc). - `inpaint_seq` - This details any residues that were masked during inference. 3. Trajectory files. By default, we output the full trajectories into the `/traj/` folder. These files can be opened in pymol, as multi-step pdbs. Note that these are ordered in reverse, so the first pdb is technically the last (t=1) prediction made by RFdiffusion during inference. We include both the `pX0` predictions (what the model predicted at each timestep) and the `Xt-1` trajectories (what went into the model at each timestep). ### Docker We have provided a Dockerfile at `docker/Dockerfile` to help run RFDiffusion on HPC and other container orchestration systems. Follow these steps to build and run the container on your system: 1. Clone this repository with `git clone https://github.com/RosettaCommons/RFdiffusion.git` and then `cd RFdiffusion` 1. Verify that the Docker daemon is running on your system with `docker info`. You can find Docker installation instructions for Mac, WIndows, and Linux in the [official Docker docs](https://docs.docker.com/get-docker/). You may also consider [Finch](https://github.com/runfinch/finch), the open source client for container development. 1. Build the container image on your system with `docker build -f docker/Dockerfile -t rfdiffusion .` 1. Create some folders on your file system with `mkdir $HOME/inputs $HOME/outputs $HOME/models` 1. Download the RFDiffusion models with `bash scripts/download_models.sh $HOME/models` 1. Download a test file (or another of your choice) with `wget -P $HOME/inputs https://files.rcsb.org/view/5TPN.pdb` 1. Run the container with the following command: ```bash docker run -it --rm --gpus all \ -v $HOME/models:$HOME/models \ -v $HOME/inputs:$HOME/inputs \ -v $HOME/outputs:$HOME/outputs \ rfdiffusion \ inference.output_prefix=$HOME/outputs/motifscaffolding \ inference.model_directory_path=$HOME/models \ inference.input_pdb=$HOME/inputs/5TPN.pdb \ inference.num_designs=3 \ 'contigmap.contigs=[10-40/A163-181/10-40]' ``` This starts the `rfdiffusion` container, mounts the models, inputs, and outputs folders, passes all available GPUs, and then calls the `run_inference.py` script with the parameters specified. ### Conclusion We are extremely excited to share RFdiffusion with the wider scientific community. We expect to push some updates as and when we make sizeable improvements in the coming months, so do stay tuned. We realize it may take some time to get used to executing RFdiffusion with perfect syntax (sometimes Hydra is hard), so please don't hesitate to create GitHub issues if you need help, we will respond as often as we can. Now, let's go make some proteins. Have fun! \- Joe, David, Nate, Brian, Jason, and the RFdiffusion team. --- RFdiffusion builds directly on the architecture and trained parameters of RoseTTAFold. We therefore thank Frank DiMaio and Minkyung Baek, who developed RoseTTAFold. RFdiffusion is released under an open source BSD License (see LICENSE file). It is free for both non-profit and for-profit use.	{}	task	[ "TRANSLATION" ]	43,152
khavitidala/finetuned-indobartv2-id-su	khavitidala	text2text-generation	[ "transformers", "pytorch", "mbart", "text2text-generation", "indogpt", "indobenchmark", "indonlg", "id", "arxiv:2104.08200", "license:mit", "autotrain_compatible", "region:us" ]	2022-03-02T23:29:05Z	2022-02-25T09:23:22+00:00	148	0	--- datasets: - Indo4B+ language: id license: mit tags: - indogpt - indobenchmark - indonlg inference: false --- # IndoBART-v2 Model fine-tuned version Fine-tuned version of IndoBART-v2 with machine translation id->su using default hyperparameter from indoBART paper. by Ryan Abdurohman # IndoBART-v2 Model [IndoBART-v2](https://arxiv.org/abs/2104.08200) is a state-of-the-art language model for Indonesian based on the BART model. The pretrained model is trained using the BART training objective. ## All Pre-trained Models \| Model \| #params \| Training data \| \|--------------------------------\|--------------------------------\|-----------------------------------\| \| `indobenchmark/indobart-v2` \| 132M \| Indo4B-Plus (26 GB of text) \| ## Authors <b>IndoBART</b> was trained and evaluated by Samuel Cahyawijaya, Genta Indra Winata, Bryan Wilie, Karissa Vincentio, Xiaohong Li, Adhiguna Kuncoro, Sebastian Ruder, Zhi Yuan Lim, Syafri Bahar, Masayu Leylia Khodra, Ayu Purwarianti, Pascale Fung ## Citation If you use our work, please cite: ```bibtex @article{cahyawijaya2021indonlg, title={IndoNLG: Benchmark and Resources for Evaluating Indonesian Natural Language Generation}, author={Cahyawijaya, Samuel and Winata, Genta Indra and Wilie, Bryan and Vincentio, Karissa and Li, Xiaohong and Kuncoro, Adhiguna and Ruder, Sebastian and Lim, Zhi Yuan and Bahar, Syafri and Khodra, Masayu Leylia and others}, journal={arXiv preprint arXiv:2104.08200}, year={2021} } ```	null	Non_BioNLP	# IndoBART-v2 Model fine-tuned version Fine-tuned version of IndoBART-v2 with machine translation id->su using default hyperparameter from indoBART paper. by Ryan Abdurohman # IndoBART-v2 Model [IndoBART-v2](https://arxiv.org/abs/2104.08200) is a state-of-the-art language model for Indonesian based on the BART model. The pretrained model is trained using the BART training objective. ## All Pre-trained Models \| Model \| #params \| Training data \| \|--------------------------------\|--------------------------------\|-----------------------------------\| \| `indobenchmark/indobart-v2` \| 132M \| Indo4B-Plus (26 GB of text) \| ## Authors <b>IndoBART</b> was trained and evaluated by Samuel Cahyawijaya, Genta Indra Winata, Bryan Wilie, Karissa Vincentio, Xiaohong Li, Adhiguna Kuncoro, Sebastian Ruder, Zhi Yuan Lim, Syafri Bahar, Masayu Leylia Khodra, Ayu Purwarianti, Pascale Fung ## Citation If you use our work, please cite: ```bibtex @article{cahyawijaya2021indonlg, title={IndoNLG: Benchmark and Resources for Evaluating Indonesian Natural Language Generation}, author={Cahyawijaya, Samuel and Winata, Genta Indra and Wilie, Bryan and Vincentio, Karissa and Li, Xiaohong and Kuncoro, Adhiguna and Ruder, Sebastian and Lim, Zhi Yuan and Bahar, Syafri and Khodra, Masayu Leylia and others}, journal={arXiv preprint arXiv:2104.08200}, year={2021} } ```	{"datasets": ["Indo4B+"], "language": "id", "license": "mit", "tags": ["indogpt", "indobenchmark", "indonlg"], "inference": false}	task	[ "TRANSLATION" ]	43,153
tangger/Qwen-7B-Chat	tangger	text-generation	[ "transformers", "pytorch", "qwen", "text-generation", "custom_code", "zh", "en", "arxiv:2305.08322", "arxiv:2009.03300", "arxiv:2305.05280", "arxiv:2210.03629", "autotrain_compatible", "region:us" ]	2023-09-13T07:08:02Z	2023-09-15T05:01:01+00:00	43	28	--- language: - zh - en pipeline_tag: text-generation --- # Notice/注意事项 Qwen官方模型临时下架了，我将9月11日下载到本地的Qwen-7B-Chat模型重新上传到huggingface。该模型为9月11日下载的版本。经过多人下载测试可以正常使用。模型仅作为官方下线期间的临时备份，待官方模型重新上线后，请各位尽早换回官方模型。 # Qwen-7B-Chat <p align="center"> <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/logo.jpg" width="400"/> <p> <br> <p align="center"> Qwen-7B <a href="https://modelscope.cn/models/qwen/Qwen-7B/summary">🤖 <a> \| <a href="https://huggingface.co/Qwen/Qwen-7B">🤗</a>&nbsp ｜ Qwen-7B-Chat <a href="https://modelscope.cn/models/qwen/Qwen-7B-Chat/summary">🤖 <a> \| <a href="https://huggingface.co/Qwen/Qwen-7B-Chat">🤗</a>&nbsp \| Qwen-7B-Chat-Int4 <a href="https://huggingface.co/Qwen/Qwen-7B-Chat-Int4">🤗</a> <br> <a href="https://github.com/QwenLM/Qwen-7B/blob/main/assets/wechat.png">WeChat</a>&nbsp&nbsp \| &nbsp&nbsp<a href="https://discord.gg/z3GAxXZ9Ce">Discord</a>&nbsp&nbsp \| &nbsp&nbsp<a href="https://modelscope.cn/studios/qwen/Qwen-7B-Chat-Demo/summary">Demo</a>&nbsp ｜ &nbsp<a href="https://github.com/QwenLM/Qwen-7B/blob/main/tech_memo.md">Report</a> </p> <br> ## 介绍（Introduction）通义千问-7B（Qwen-7B）是阿里云研发的通义千问大模型系列的70亿参数规模的模型。Qwen-7B是基于Transformer的大语言模型, 在超大规模的预训练数据上进行训练得到。预训练数据类型多样，覆盖广泛，包括大量网络文本、专业书籍、代码等。同时，在Qwen-7B的基础上，我们使用对齐机制打造了基于大语言模型的AI助手Qwen-7B-Chat。本仓库为Qwen-7B-Chat的仓库。如果您想了解更多关于通义千问-7B开源模型的细节，我们建议您参阅[Github代码库](https://github.com/QwenLM/Qwen-7B)。 Qwen-7B is the 7B-parameter version of the large language model series, Qwen (abbr. Tongyi Qianwen), proposed by Aibaba Cloud. Qwen-7B is a Transformer-based large language model, which is pretrained on a large volume of data, including web texts, books, codes, etc. Additionally, based on the pretrained Qwen-7B, we release Qwen-7B-Chat, a large-model-based AI assistant, which is trained with alignment techniques. This repository is the one for Qwen-7B-Chat. For more details about the open-source model of Qwen-7B, please refer to the [Github](https://github.com/QwenLM/Qwen-7B) code repository. <br> ## 要求（Requirements） * python 3.8及以上版本 * pytorch 1.12及以上版本，推荐2.0及以上版本 * 建议使用CUDA 11.4及以上（GPU用户、flash-attention用户等需考虑此选项） * python 3.8 and above * pytorch 1.12 and above, 2.0 and above are recommended * CUDA 11.4 and above are recommended (this is for GPU users, flash-attention users, etc.) <br> ## 依赖项（Dependency）运行Qwen-7B-Chat，请确保满足上述要求，再执行以下pip命令安装依赖库 To run Qwen-7B-Chat, please make sure you meet the above requirements, and then execute the following pip commands to install the dependent libraries. ```bash pip install transformers==4.32.0 accelerate tiktoken einops scipy transformers_stream_generator==0.0.4 peft deepspeed ``` 另外，推荐安装`flash-attention`库，以实现更高的效率和更低的显存占用。 In addition, it is recommended to install the `flash-attention` library for higher efficiency and lower memory usage. ```bash git clone -b v1.0.8 https://github.com/Dao-AILab/flash-attention cd flash-attention && pip install . # 下方安装可选，安装可能比较缓慢。 # Below are optional. Installing them might be slow. # pip install csrc/layer_norm # pip install csrc/rotary ``` <br> ## 快速使用（Quickstart）下面我们展示了一个使用Qwen-7B-Chat模型，进行多轮对话交互的样例： We show an example of multi-turn interaction with Qwen-7B-Chat in the following code: ```python from transformers import AutoModelForCausalLM, AutoTokenizer from transformers.generation import GenerationConfig # Note: The default behavior now has injection attack prevention off. tokenizer = AutoTokenizer.from_pretrained("tangger/Qwen-7B-Chat", trust_remote_code=True) # use bf16 # model = AutoModelForCausalLM.from_pretrained("tangger/Qwen-7B-Chat", device_map="auto", trust_remote_code=True, bf16=True).eval() # use fp16 # model = AutoModelForCausalLM.from_pretrained("tangger/Qwen-7B-Chat", device_map="auto", trust_remote_code=True, fp16=True).eval() # use cpu only # model = AutoModelForCausalLM.from_pretrained("tangger/Qwen-7B-Chat", device_map="cpu", trust_remote_code=True).eval() # use auto mode, automatically select precision based on the device. model = AutoModelForCausalLM.from_pretrained("tangger/Qwen-7B-Chat", device_map="auto", trust_remote_code=True).eval() # Specify hyperparameters for generation model.generation_config = GenerationConfig.from_pretrained("tangger/Qwen-7B-Chat", trust_remote_code=True) # 可指定不同的生成长度、top_p等相关超参 # 第一轮对话 1st dialogue turn response, history = model.chat(tokenizer, "你好", history=None) print(response) # 你好！很高兴为你提供帮助。 # 第二轮对话 2nd dialogue turn response, history = model.chat(tokenizer, "给我讲一个年轻人奋斗创业最终取得成功的故事。", history=history) print(response) # 这是一个关于一个年轻人奋斗创业最终取得成功的故事。 # 故事的主人公叫李明，他来自一个普通的家庭，父母都是普通的工人。从小，李明就立下了一个目标：要成为一名成功的企业家。 # 为了实现这个目标，李明勤奋学习，考上了大学。在大学期间，他积极参加各种创业比赛，获得了不少奖项。他还利用课余时间去实习，积累了宝贵的经验。 # 毕业后，李明决定开始自己的创业之路。他开始寻找投资机会，但多次都被拒绝了。然而，他并没有放弃。他继续努力，不断改进自己的创业计划，并寻找新的投资机会。 # 最终，李明成功地获得了一笔投资，开始了自己的创业之路。他成立了一家科技公司，专注于开发新型软件。在他的领导下，公司迅速发展起来，成为了一家成功的科技企业。 # 李明的成功并不是偶然的。他勤奋、坚韧、勇于冒险，不断学习和改进自己。他的成功也证明了，只要努力奋斗，任何人都有可能取得成功。 # 第三轮对话 3rd dialogue turn response, history = model.chat(tokenizer, "给这个故事起一个标题", history=history) print(response) # 《奋斗创业：一个年轻人的成功之路》 ``` 关于更多的使用说明，请参考我们的[Github repo](https://github.com/QwenLM/Qwen-7B)获取更多信息。 For more information, please refer to our [Github repo](https://github.com/QwenLM/Qwen-7B) for more information. <br> ## Tokenizer > 注：作为术语的“tokenization”在中文中尚无共识的概念对应，本文档采用英文表达以利说明。基于tiktoken的分词器有别于其他分词器，比如sentencepiece分词器。尤其在微调阶段，需要特别注意特殊token的使用。关于tokenizer的更多信息，以及微调时涉及的相关使用，请参阅[文档](https://github.com/QwenLM/Qwen-7B/blob/main/tokenization_note_zh.md)。 Our tokenizer based on tiktoken is different from other tokenizers, e.g., sentencepiece tokenizer. You need to pay attention to special tokens, especially in finetuning. For more detailed information on the tokenizer and related use in fine-tuning, please refer to the [documentation](https://github.com/QwenLM/Qwen-7B/blob/main/tokenization_note.md). <br> ## 量化 (Quantization) ### 用法 (Usage) 请注意：我们更新量化方案为基于[AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ)的量化，提供Qwen-7B-Chat的Int4量化模型[点击这里](https://huggingface.co/Qwen/Qwen-7B-Chat-Int4)。相比此前方案，该方案在模型评测效果几乎无损，且存储需求更低，推理速度更优。 Note: we provide a new solution based on [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and release an Int4 quantized model for Qwen-7B-Chat [Click here](https://huggingface.co/Qwen/Qwen-7B-Chat-Int4), which achieves nearly lossless model effects but improved performance on both memory costs and inference speed, in comparison with the previous solution. 以下我们提供示例说明如何使用Int4量化模型。在开始使用前，请先保证满足要求（如torch 2.0及以上，transformers版本为4.32.0及以上，等等），并安装所需安装包： Here we demonstrate how to use our provided quantized models for inference. Before you start, make sure you meet the requirements of auto-gptq (e.g., torch 2.0 and above, transformers 4.32.0 and above, etc.) and install the required packages: ```bash pip install auto-gptq optimum ``` 如安装`auto-gptq`遇到问题，我们建议您到官方[repo](https://github.com/PanQiWei/AutoGPTQ)搜索合适的预编译wheel。随后即可使用和上述一致的用法调用量化模型： If you meet problems installing `auto-gptq`, we advise you to check out the official [repo](https://github.com/PanQiWei/AutoGPTQ) to find a pre-build wheel. Then you can load the quantized model easily and run inference as same as usual: ```python model = AutoModelForCausalLM.from_pretrained( "Qwen/Qwen-7B-Chat-Int4", device_map="auto", trust_remote_code=True ).eval() response, history = model.chat(tokenizer, "你好", history=None) ``` ### 效果评测我们对BF16和Int4模型在基准评测上做了测试，发现量化模型效果损失较小，结果如下所示： We illustrate the model performance of both BF16 and Int4 models on the benchmark, and we find that the quantized model does not suffer from significant performance degradation. Results are shown below: \| Quantization \| MMLU \| CEval (val) \| GSM8K \| Humaneval \| \| ------------- \| :--------: \| :----------: \| :----: \| :--------: \| \| BF16 \| 53.9 \| 54.2 \| 41.1 \| 24.4 \| \| Int4 \| 52.6 \| 52.9 \| 38.1 \| 23.8 \| ### 推理速度 (Inference Speed) 我们测算了BF16和Int4模型生成2048和8192个token的平均推理速度。如图所示： We measured the average inference speed of generating 2048 and 8192 tokens under BF16 precision and Int4 quantization level, respectively. \| Quantization \| Speed (2048 tokens) \| Speed (8192 tokens) \| \| ------------- \| :------------------:\| :------------------:\| \| BF16 \| 30.53 \| 28.51 \| \| Int4 \| 45.60 \| 33.83 \| 具体而言，我们记录在长度为1的上下文的条件下生成8192个token的性能。评测运行于单张A100-SXM4-80G GPU，使用PyTorch 2.0.1和CUDA 11.4。推理速度是生成8192个token的速度均值。 In detail, the setting of profiling is generating 8192 new tokens with 1 context token. The profiling runs on a single A100-SXM4-80G GPU with PyTorch 2.0.1 and CUDA 11.4. The inference speed is averaged over the generated 8192 tokens. ### 显存使用 (GPU Memory Usage) 我们还测算了BF16和Int4模型编码2048个token及生成8192个token的峰值显存占用情况。结果如下所示： We also profile the peak GPU memory usage for encoding 2048 tokens as context (and generating single token) and generating 8192 tokens (with single token as context) under BF16 or Int4 quantization level, respectively. The results are shown below. \| Quantization Level \| Peak Usage for Encoding 2048 Tokens \| Peak Usage for Generating 8192 Tokens \| \| ------------------ \| :---------------------------------: \| :-----------------------------------: \| \| BF16 \| 18.99GB \| 24.40GB \| \| Int4 \| 10.20GB \| 15.61GB \| 上述性能测算使用[此脚本](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py)完成。 The above speed and memory profiling are conducted using [this script](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py). <br> ## 模型细节（Model）与Qwen-7B预训练模型相同，Qwen-7B-Chat模型规模基本情况如下所示 The details of the model architecture of Qwen-7B-Chat are listed as follows \| Hyperparameter \| Value \| \| :------------- \| :----: \| \| n_layers \| 32 \| \| n_heads \| 32 \| \| d_model \| 4096 \| \| vocab size \| 151851 \| \| sequence length \| 2048 \| 在位置编码、FFN激活函数和normalization的实现方式上，我们也采用了目前最流行的做法，即RoPE相对位置编码、SwiGLU激活函数、RMSNorm（可选安装flash-attention加速）。在分词器方面，相比目前主流开源模型以中英词表为主，Qwen-7B-Chat使用了约15万token大小的词表。该词表在GPT-4使用的BPE词表`cl100k_base`基础上，对中文、多语言进行了优化，在对中、英、代码数据的高效编解码的基础上，对部分多语言更加友好，方便用户在不扩展词表的情况下对部分语种进行能力增强。词表对数字按单个数字位切分。调用较为高效的[tiktoken分词库](https://github.com/openai/tiktoken)进行分词。 For position encoding, FFN activation function, and normalization calculation methods, we adopt the prevalent practices, i.e., RoPE relative position encoding, SwiGLU for activation function, and RMSNorm for normalization (optional installation of flash-attention for acceleration). For tokenization, compared to the current mainstream open-source models based on Chinese and English vocabularies, Qwen-7B-Chat uses a vocabulary of over 150K tokens. It first considers efficient encoding of Chinese, English, and code data, and is also more friendly to multilingual languages, enabling users to directly enhance the capability of some languages without expanding the vocabulary. It segments numbers by single digit, and calls the [tiktoken](https://github.com/openai/tiktoken) tokenizer library for efficient tokenization. <br> ## 评测效果（Evaluation）对于Qwen-7B-Chat模型，我们同样评测了常规的中文理解（C-Eval）、英文理解（MMLU）、代码（HumanEval）和数学（GSM8K）等权威任务，同时包含了长序列任务的评测结果。由于Qwen-7B-Chat模型经过对齐后，激发了较强的外部系统调用能力，我们还进行了工具使用能力方面的评测。提示：由于硬件和框架造成的舍入误差，复现结果如有波动属于正常现象。 For Qwen-7B-Chat, we also evaluate the model on C-Eval, MMLU, HumanEval, GSM8K, etc., as well as the benchmark evaluation for long-context understanding, and tool usage. Note: Due to rounding errors caused by hardware and framework, differences in reproduced results are possible. ### 中文评测（Chinese Evaluation） #### C-Eval 在[C-Eval](https://arxiv.org/abs/2305.08322)验证集上，我们评价了Qwen-7B-Chat模型的zero-shot准确率 We demonstrate the zero-shot accuracy of Qwen-7B-Chat on C-Eval validation set \| Model \| Avg. Acc. \| \| :---------------------- \| :-------: \| \| LLaMA2-7B-Chat \| 31.9 \| \| LLaMA2-13B-Chat \| 40.6 \| \| Chinese-Alpaca-2-7B \| 41.3 \| \| Chinese-Alpaca-Plus-13B \| 43.3 \| \| Baichuan-13B-Chat \| 50.4 \| \| ChatGLM2-6B-Chat \| 50.7 \| \| InternLM-7B-Chat \| 53.2 \| \| Qwen-7B-Chat \| 54.2 \| C-Eval测试集上，Qwen-7B-Chat模型的zero-shot准确率结果如下： The zero-shot accuracy of Qwen-7B-Chat on C-Eval testing set is provided below: \| Model \| Avg. \| STEM \| Social Sciences \| Humanities \| Others \| \| :---------------------- \| :------: \| :--: \| :-------------: \| :--------: \| :----: \| \| Chinese-Alpaca-Plus-13B \| 41.5 \| 36.6 \| 49.7 \| 43.1 \| 41.2 \| \| Chinese-Alpaca-2-7B \| 40.3 \| - \| - \| - \| - \| \| ChatGLM2-6B-Chat \| 50.1 \| 46.4 \| 60.4 \| 50.6 \| 46.9 \| \| Baichuan-13B-Chat \| 51.5 \| 43.7 \| 64.6 \| 56.2 \| 49.2 \| \| Qwen-7B-Chat \| 54.6 \| 47.8 \| 67.6 \| 59.3 \| 50.6 \| 在7B规模模型上，经过人类指令对齐的Qwen-7B-Chat模型，准确率在同类相近规模模型中仍然处于前列。 Compared with other pretrained models with comparable model size, the human-aligned Qwen-7B-Chat performs well in C-Eval accuracy. ### 英文评测（English Evaluation） #### MMLU [MMLU](https://arxiv.org/abs/2009.03300)评测集上，Qwen-7B-Chat模型的zero-shot准确率如下，效果同样在同类对齐模型中同样表现较优。 The zero-shot accuracy of Qwen-7B-Chat on MMLU is provided below. The performance of Qwen-7B-Chat still on the top between other human-aligned models with comparable size. \| Model \| Avg. Acc. \| \| :---------------- \| :-------: \| \| ChatGLM2-6B-Chat \| 45.5 \| \| LLaMA2-7B-Chat \| 47.0 \| \| InternLM-7B-Chat \| 50.8 \| \| Baichuan-13B-Chat \| 52.1 \| \| ChatGLM2-12B-Chat \| 52.1 \| \| Qwen-7B-Chat \| 53.9 \| ### 代码评测（Coding Evaluation） Qwen-7B-Chat在[HumanEval](https://github.com/openai/human-eval)的zero-shot Pass@1效果如下 The zero-shot Pass@1 of Qwen-7B-Chat on [HumanEval](https://github.com/openai/human-eval) is demonstrated below \| Model \| Pass@1 \| \| :---------------- \| :------: \| \| LLaMA2-7B-Chat \| 12.2 \| \| InternLM-7B-Chat \| 14.0 \| \| Baichuan-13B-Chat \| 16.5 \| \| LLaMA2-13B-Chat \| 18.9 \| \| Qwen-7B-Chat \| 24.4 \| ### 数学评测（Mathematics Evaluation）在评测数学能力的[GSM8K](https://github.com/openai/grade-school-math)上，Qwen-7B-Chat的准确率结果如下 The accuracy of Qwen-7B-Chat on GSM8K is shown below \| Model \| Zero-shot Acc. \| 4-shot Acc. \| \| :---------------- \| :------------: \| :--------: \| \| ChatGLM2-6B-Chat \| - \| 28.0 \| \| LLaMA2-7B-Chat \| 20.4 \| 28.2 \| \| LLaMA2-13B-Chat \| 29.4 \| 36.7 \| \| InternLM-7B-Chat \| 32.6 \| 34.5 \| \| Baichuan-13B-Chat \| - \| 36.3 \| \| ChatGLM2-12B-Chat \| - \| 38.1 \| \| Qwen-7B-Chat \| 41.1 \| 43.5 \| ### 长序列评测（Long-Context Understanding）通过NTK插值，LogN注意力缩放可以扩展Qwen-7B-Chat的上下文长度。在长文本摘要数据集[VCSUM](https://arxiv.org/abs/2305.05280)上（文本平均长度在15K左右），Qwen-7B-Chat的Rouge-L结果如下： (若要启用这些技巧，请将config.json里的`use_dynamic_ntk`和`use_logn_attn`设置为true) We introduce NTK-aware interpolation, LogN attention scaling to extend the context length of Qwen-7B-Chat. The Rouge-L results of Qwen-7B-Chat on long-text summarization dataset [VCSUM](https://arxiv.org/abs/2305.05280) (The average length of this dataset is around 15K) are shown below: (To use these tricks, please set `use_dynamic_ntk` and `use_long_attn` to true in config.json.) \| Model \| VCSUM (zh) \| \| :---------------- \| :--------: \| \| GPT-3.5-Turbo-16k \| 16.0 \| \| LLama2-7B-Chat \| 0.2 \| \| InternLM-7B-Chat \| 13.0 \| \| ChatGLM2-6B-Chat \| 16.3 \| \| Qwen-7B-Chat \| 16.6 \| ### 工具使用能力的评测（Tool Usage） #### ReAct Prompting 千问支持通过 [ReAct Prompting](https://arxiv.org/abs/2210.03629) 调用插件/工具/API。ReAct 也是 [LangChain](https://python.langchain.com/) 框架采用的主要方式之一。在我们开源的、用于评估工具使用能力的评测基准上，千问的表现如下： Qwen-7B-Chat supports calling plugins/tools/APIs through [ReAct Prompting](https://arxiv.org/abs/2210.03629). ReAct is also one of the main approaches used by the [LangChain](https://python.langchain.com/) framework. In our evaluation benchmark for assessing tool usage capabilities, Qwen-7B-Chat's performance is as follows: \| Model \| Tool Selection (Acc.↑) \| Tool Input (Rouge-L↑) \| False Positive Error”↓ \| \| :--------------- \| :---------------------: \| :--------------------: \| :--------------------: \| \| GPT-4 \| 95% \| 0.90 \| 15% \| \| GPT-3.5 \| 85% \| 0.88 \| 75% \| \| Qwen-7B-Chat \| 99% \| 0.89 \| 9.7% \| > 评测基准中出现的插件均没有出现在千问的训练集中。该基准评估了模型在多个候选插件中选择正确插件的准确率、传入插件的参数的合理性、以及假阳率。假阳率（False Positive）定义：在处理不该调用插件的请求时，错误地调用了插件。 > The plugins that appear in the evaluation set do not appear in the training set of Qwen-7B-Chat. This benchmark evaluates the accuracy of the model in selecting the correct plugin from multiple candidate plugins, the rationality of the parameters passed into the plugin, and the false positive rate. False Positive: Incorrectly invoking a plugin when it should not have been called when responding to a query. 关于 ReAct Prompting 的 prompt 怎么写、怎么使用，请参考 [ReAct 样例说明](examples/react_prompt.md)。使用工具能使模型更好地完成任务。基于千问的工具使用能力，我们能实现下图所展示的效果： For how to write and use prompts for ReAct Prompting, please refer to [the ReAct examples](examples/react_prompt.md). The use of tools can enable the model to better perform tasks, as shown in the following figures: ![](assets/react_showcase_001.png) ![](assets/react_showcase_002.png) #### Huggingface Agent 千问还具备作为 [HuggingFace Agent](https://huggingface.co/docs/transformers/transformers_agents) 的能力。它在 Huggingface 提供的run模式评测基准上的表现如下： Qwen-7B-Chat also has the capability to be used as a [HuggingFace Agent](https://huggingface.co/docs/transformers/transformers_agents). Its performance on the run-mode benchmark provided by HuggingFace is as follows: \| Model \| Tool Selection↑ \| Tool Used↑ \| Code↑ \| \|:-----------------\| :-------------: \| :---------: \| :-------: \| \| GPT-4 \| 100 \| 100 \| 97.41 \| \| GPT-3.5 \| 95.37 \| 96.30 \| 87.04 \| \| StarCoder-15.5B \| 87.04 \| 87.96 \| 68.89 \| \| Qwen-7B-Chat \| 90.74 \| 92.59 \| 74.07 \| <br> ## FAQ 如遇到问题，敬请查阅[FAQ](https://github.com/QwenLM/Qwen-7B/blob/main/FAQ_zh.md)以及issue区，如仍无法解决再提交issue。 If you meet problems, please refer to [FAQ](https://github.com/QwenLM/Qwen-7B/blob/main/FAQ.md) and the issues first to search a solution before you launch a new issue. <br> ## 使用协议（License Agreement）我们的代码和模型权重对学术研究完全开放，并支持商用。请查看[LICENSE](https://github.com/QwenLM/Qwen-7B/blob/main/LICENSE)了解具体的开源协议细节。如需商用，请填写[问卷](https://dashscope.console.aliyun.com/openModelApply/qianwen)申请。 Our code and checkpoints are open to research purpose, and they are allowed for commercial purposes. Check [LICENSE](https://github.com/QwenLM/Qwen-7B/blob/main/LICENSE) for more details about the license. If you have requirements for commercial use, please fill out the [form](https://dashscope.console.aliyun.com/openModelApply/qianwen) to apply. <br> ## 联系我们（Contact Us）如果你想给我们的研发团队和产品团队留言，请通过邮件（[email protected]）联系我们。 If you are interested to leave a message to either our research team or product team, feel free to send an email to [email protected].	null	Non_BioNLP	# Notice/注意事项 Qwen官方模型临时下架了，我将9月11日下载到本地的Qwen-7B-Chat模型重新上传到huggingface。该模型为9月11日下载的版本。经过多人下载测试可以正常使用。模型仅作为官方下线期间的临时备份，待官方模型重新上线后，请各位尽早换回官方模型。 # Qwen-7B-Chat <p align="center"> <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/logo.jpg" width="400"/> <p> <br> <p align="center"> Qwen-7B <a href="https://modelscope.cn/models/qwen/Qwen-7B/summary">🤖 <a> \| <a href="https://huggingface.co/Qwen/Qwen-7B">🤗</a>&nbsp ｜ Qwen-7B-Chat <a href="https://modelscope.cn/models/qwen/Qwen-7B-Chat/summary">🤖 <a> \| <a href="https://huggingface.co/Qwen/Qwen-7B-Chat">🤗</a>&nbsp \| Qwen-7B-Chat-Int4 <a href="https://huggingface.co/Qwen/Qwen-7B-Chat-Int4">🤗</a> <br> <a href="https://github.com/QwenLM/Qwen-7B/blob/main/assets/wechat.png">WeChat</a>&nbsp&nbsp \| &nbsp&nbsp<a href="https://discord.gg/z3GAxXZ9Ce">Discord</a>&nbsp&nbsp \| &nbsp&nbsp<a href="https://modelscope.cn/studios/qwen/Qwen-7B-Chat-Demo/summary">Demo</a>&nbsp ｜ &nbsp<a href="https://github.com/QwenLM/Qwen-7B/blob/main/tech_memo.md">Report</a> </p> <br> ## 介绍（Introduction）通义千问-7B（Qwen-7B）是阿里云研发的通义千问大模型系列的70亿参数规模的模型。Qwen-7B是基于Transformer的大语言模型, 在超大规模的预训练数据上进行训练得到。预训练数据类型多样，覆盖广泛，包括大量网络文本、专业书籍、代码等。同时，在Qwen-7B的基础上，我们使用对齐机制打造了基于大语言模型的AI助手Qwen-7B-Chat。本仓库为Qwen-7B-Chat的仓库。如果您想了解更多关于通义千问-7B开源模型的细节，我们建议您参阅[Github代码库](https://github.com/QwenLM/Qwen-7B)。 Qwen-7B is the 7B-parameter version of the large language model series, Qwen (abbr. Tongyi Qianwen), proposed by Aibaba Cloud. Qwen-7B is a Transformer-based large language model, which is pretrained on a large volume of data, including web texts, books, codes, etc. Additionally, based on the pretrained Qwen-7B, we release Qwen-7B-Chat, a large-model-based AI assistant, which is trained with alignment techniques. This repository is the one for Qwen-7B-Chat. For more details about the open-source model of Qwen-7B, please refer to the [Github](https://github.com/QwenLM/Qwen-7B) code repository. <br> ## 要求（Requirements） * python 3.8及以上版本 * pytorch 1.12及以上版本，推荐2.0及以上版本 * 建议使用CUDA 11.4及以上（GPU用户、flash-attention用户等需考虑此选项） * python 3.8 and above * pytorch 1.12 and above, 2.0 and above are recommended * CUDA 11.4 and above are recommended (this is for GPU users, flash-attention users, etc.) <br> ## 依赖项（Dependency）运行Qwen-7B-Chat，请确保满足上述要求，再执行以下pip命令安装依赖库 To run Qwen-7B-Chat, please make sure you meet the above requirements, and then execute the following pip commands to install the dependent libraries. ```bash pip install transformers==4.32.0 accelerate tiktoken einops scipy transformers_stream_generator==0.0.4 peft deepspeed ``` 另外，推荐安装`flash-attention`库，以实现更高的效率和更低的显存占用。 In addition, it is recommended to install the `flash-attention` library for higher efficiency and lower memory usage. ```bash git clone -b v1.0.8 https://github.com/Dao-AILab/flash-attention cd flash-attention && pip install . # 下方安装可选，安装可能比较缓慢。 # Below are optional. Installing them might be slow. # pip install csrc/layer_norm # pip install csrc/rotary ``` <br> ## 快速使用（Quickstart）下面我们展示了一个使用Qwen-7B-Chat模型，进行多轮对话交互的样例： We show an example of multi-turn interaction with Qwen-7B-Chat in the following code: ```python from transformers import AutoModelForCausalLM, AutoTokenizer from transformers.generation import GenerationConfig # Note: The default behavior now has injection attack prevention off. tokenizer = AutoTokenizer.from_pretrained("tangger/Qwen-7B-Chat", trust_remote_code=True) # use bf16 # model = AutoModelForCausalLM.from_pretrained("tangger/Qwen-7B-Chat", device_map="auto", trust_remote_code=True, bf16=True).eval() # use fp16 # model = AutoModelForCausalLM.from_pretrained("tangger/Qwen-7B-Chat", device_map="auto", trust_remote_code=True, fp16=True).eval() # use cpu only # model = AutoModelForCausalLM.from_pretrained("tangger/Qwen-7B-Chat", device_map="cpu", trust_remote_code=True).eval() # use auto mode, automatically select precision based on the device. model = AutoModelForCausalLM.from_pretrained("tangger/Qwen-7B-Chat", device_map="auto", trust_remote_code=True).eval() # Specify hyperparameters for generation model.generation_config = GenerationConfig.from_pretrained("tangger/Qwen-7B-Chat", trust_remote_code=True) # 可指定不同的生成长度、top_p等相关超参 # 第一轮对话 1st dialogue turn response, history = model.chat(tokenizer, "你好", history=None) print(response) # 你好！很高兴为你提供帮助。 # 第二轮对话 2nd dialogue turn response, history = model.chat(tokenizer, "给我讲一个年轻人奋斗创业最终取得成功的故事。", history=history) print(response) # 这是一个关于一个年轻人奋斗创业最终取得成功的故事。 # 故事的主人公叫李明，他来自一个普通的家庭，父母都是普通的工人。从小，李明就立下了一个目标：要成为一名成功的企业家。 # 为了实现这个目标，李明勤奋学习，考上了大学。在大学期间，他积极参加各种创业比赛，获得了不少奖项。他还利用课余时间去实习，积累了宝贵的经验。 # 毕业后，李明决定开始自己的创业之路。他开始寻找投资机会，但多次都被拒绝了。然而，他并没有放弃。他继续努力，不断改进自己的创业计划，并寻找新的投资机会。 # 最终，李明成功地获得了一笔投资，开始了自己的创业之路。他成立了一家科技公司，专注于开发新型软件。在他的领导下，公司迅速发展起来，成为了一家成功的科技企业。 # 李明的成功并不是偶然的。他勤奋、坚韧、勇于冒险，不断学习和改进自己。他的成功也证明了，只要努力奋斗，任何人都有可能取得成功。 # 第三轮对话 3rd dialogue turn response, history = model.chat(tokenizer, "给这个故事起一个标题", history=history) print(response) # 《奋斗创业：一个年轻人的成功之路》 ``` 关于更多的使用说明，请参考我们的[Github repo](https://github.com/QwenLM/Qwen-7B)获取更多信息。 For more information, please refer to our [Github repo](https://github.com/QwenLM/Qwen-7B) for more information. <br> ## Tokenizer > 注：作为术语的“tokenization”在中文中尚无共识的概念对应，本文档采用英文表达以利说明。基于tiktoken的分词器有别于其他分词器，比如sentencepiece分词器。尤其在微调阶段，需要特别注意特殊token的使用。关于tokenizer的更多信息，以及微调时涉及的相关使用，请参阅[文档](https://github.com/QwenLM/Qwen-7B/blob/main/tokenization_note_zh.md)。 Our tokenizer based on tiktoken is different from other tokenizers, e.g., sentencepiece tokenizer. You need to pay attention to special tokens, especially in finetuning. For more detailed information on the tokenizer and related use in fine-tuning, please refer to the [documentation](https://github.com/QwenLM/Qwen-7B/blob/main/tokenization_note.md). <br> ## 量化 (Quantization) ### 用法 (Usage) 请注意：我们更新量化方案为基于[AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ)的量化，提供Qwen-7B-Chat的Int4量化模型[点击这里](https://huggingface.co/Qwen/Qwen-7B-Chat-Int4)。相比此前方案，该方案在模型评测效果几乎无损，且存储需求更低，推理速度更优。 Note: we provide a new solution based on [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and release an Int4 quantized model for Qwen-7B-Chat [Click here](https://huggingface.co/Qwen/Qwen-7B-Chat-Int4), which achieves nearly lossless model effects but improved performance on both memory costs and inference speed, in comparison with the previous solution. 以下我们提供示例说明如何使用Int4量化模型。在开始使用前，请先保证满足要求（如torch 2.0及以上，transformers版本为4.32.0及以上，等等），并安装所需安装包： Here we demonstrate how to use our provided quantized models for inference. Before you start, make sure you meet the requirements of auto-gptq (e.g., torch 2.0 and above, transformers 4.32.0 and above, etc.) and install the required packages: ```bash pip install auto-gptq optimum ``` 如安装`auto-gptq`遇到问题，我们建议您到官方[repo](https://github.com/PanQiWei/AutoGPTQ)搜索合适的预编译wheel。随后即可使用和上述一致的用法调用量化模型： If you meet problems installing `auto-gptq`, we advise you to check out the official [repo](https://github.com/PanQiWei/AutoGPTQ) to find a pre-build wheel. Then you can load the quantized model easily and run inference as same as usual: ```python model = AutoModelForCausalLM.from_pretrained( "Qwen/Qwen-7B-Chat-Int4", device_map="auto", trust_remote_code=True ).eval() response, history = model.chat(tokenizer, "你好", history=None) ``` ### 效果评测我们对BF16和Int4模型在基准评测上做了测试，发现量化模型效果损失较小，结果如下所示： We illustrate the model performance of both BF16 and Int4 models on the benchmark, and we find that the quantized model does not suffer from significant performance degradation. Results are shown below: \| Quantization \| MMLU \| CEval (val) \| GSM8K \| Humaneval \| \| ------------- \| :--------: \| :----------: \| :----: \| :--------: \| \| BF16 \| 53.9 \| 54.2 \| 41.1 \| 24.4 \| \| Int4 \| 52.6 \| 52.9 \| 38.1 \| 23.8 \| ### 推理速度 (Inference Speed) 我们测算了BF16和Int4模型生成2048和8192个token的平均推理速度。如图所示： We measured the average inference speed of generating 2048 and 8192 tokens under BF16 precision and Int4 quantization level, respectively. \| Quantization \| Speed (2048 tokens) \| Speed (8192 tokens) \| \| ------------- \| :------------------:\| :------------------:\| \| BF16 \| 30.53 \| 28.51 \| \| Int4 \| 45.60 \| 33.83 \| 具体而言，我们记录在长度为1的上下文的条件下生成8192个token的性能。评测运行于单张A100-SXM4-80G GPU，使用PyTorch 2.0.1和CUDA 11.4。推理速度是生成8192个token的速度均值。 In detail, the setting of profiling is generating 8192 new tokens with 1 context token. The profiling runs on a single A100-SXM4-80G GPU with PyTorch 2.0.1 and CUDA 11.4. The inference speed is averaged over the generated 8192 tokens. ### 显存使用 (GPU Memory Usage) 我们还测算了BF16和Int4模型编码2048个token及生成8192个token的峰值显存占用情况。结果如下所示： We also profile the peak GPU memory usage for encoding 2048 tokens as context (and generating single token) and generating 8192 tokens (with single token as context) under BF16 or Int4 quantization level, respectively. The results are shown below. \| Quantization Level \| Peak Usage for Encoding 2048 Tokens \| Peak Usage for Generating 8192 Tokens \| \| ------------------ \| :---------------------------------: \| :-----------------------------------: \| \| BF16 \| 18.99GB \| 24.40GB \| \| Int4 \| 10.20GB \| 15.61GB \| 上述性能测算使用[此脚本](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py)完成。 The above speed and memory profiling are conducted using [this script](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py). <br> ## 模型细节（Model）与Qwen-7B预训练模型相同，Qwen-7B-Chat模型规模基本情况如下所示 The details of the model architecture of Qwen-7B-Chat are listed as follows \| Hyperparameter \| Value \| \| :------------- \| :----: \| \| n_layers \| 32 \| \| n_heads \| 32 \| \| d_model \| 4096 \| \| vocab size \| 151851 \| \| sequence length \| 2048 \| 在位置编码、FFN激活函数和normalization的实现方式上，我们也采用了目前最流行的做法，即RoPE相对位置编码、SwiGLU激活函数、RMSNorm（可选安装flash-attention加速）。在分词器方面，相比目前主流开源模型以中英词表为主，Qwen-7B-Chat使用了约15万token大小的词表。该词表在GPT-4使用的BPE词表`cl100k_base`基础上，对中文、多语言进行了优化，在对中、英、代码数据的高效编解码的基础上，对部分多语言更加友好，方便用户在不扩展词表的情况下对部分语种进行能力增强。词表对数字按单个数字位切分。调用较为高效的[tiktoken分词库](https://github.com/openai/tiktoken)进行分词。 For position encoding, FFN activation function, and normalization calculation methods, we adopt the prevalent practices, i.e., RoPE relative position encoding, SwiGLU for activation function, and RMSNorm for normalization (optional installation of flash-attention for acceleration). For tokenization, compared to the current mainstream open-source models based on Chinese and English vocabularies, Qwen-7B-Chat uses a vocabulary of over 150K tokens. It first considers efficient encoding of Chinese, English, and code data, and is also more friendly to multilingual languages, enabling users to directly enhance the capability of some languages without expanding the vocabulary. It segments numbers by single digit, and calls the [tiktoken](https://github.com/openai/tiktoken) tokenizer library for efficient tokenization. <br> ## 评测效果（Evaluation）对于Qwen-7B-Chat模型，我们同样评测了常规的中文理解（C-Eval）、英文理解（MMLU）、代码（HumanEval）和数学（GSM8K）等权威任务，同时包含了长序列任务的评测结果。由于Qwen-7B-Chat模型经过对齐后，激发了较强的外部系统调用能力，我们还进行了工具使用能力方面的评测。提示：由于硬件和框架造成的舍入误差，复现结果如有波动属于正常现象。 For Qwen-7B-Chat, we also evaluate the model on C-Eval, MMLU, HumanEval, GSM8K, etc., as well as the benchmark evaluation for long-context understanding, and tool usage. Note: Due to rounding errors caused by hardware and framework, differences in reproduced results are possible. ### 中文评测（Chinese Evaluation） #### C-Eval 在[C-Eval](https://arxiv.org/abs/2305.08322)验证集上，我们评价了Qwen-7B-Chat模型的zero-shot准确率 We demonstrate the zero-shot accuracy of Qwen-7B-Chat on C-Eval validation set \| Model \| Avg. Acc. \| \| :---------------------- \| :-------: \| \| LLaMA2-7B-Chat \| 31.9 \| \| LLaMA2-13B-Chat \| 40.6 \| \| Chinese-Alpaca-2-7B \| 41.3 \| \| Chinese-Alpaca-Plus-13B \| 43.3 \| \| Baichuan-13B-Chat \| 50.4 \| \| ChatGLM2-6B-Chat \| 50.7 \| \| InternLM-7B-Chat \| 53.2 \| \| Qwen-7B-Chat \| 54.2 \| C-Eval测试集上，Qwen-7B-Chat模型的zero-shot准确率结果如下： The zero-shot accuracy of Qwen-7B-Chat on C-Eval testing set is provided below: \| Model \| Avg. \| STEM \| Social Sciences \| Humanities \| Others \| \| :---------------------- \| :------: \| :--: \| :-------------: \| :--------: \| :----: \| \| Chinese-Alpaca-Plus-13B \| 41.5 \| 36.6 \| 49.7 \| 43.1 \| 41.2 \| \| Chinese-Alpaca-2-7B \| 40.3 \| - \| - \| - \| - \| \| ChatGLM2-6B-Chat \| 50.1 \| 46.4 \| 60.4 \| 50.6 \| 46.9 \| \| Baichuan-13B-Chat \| 51.5 \| 43.7 \| 64.6 \| 56.2 \| 49.2 \| \| Qwen-7B-Chat \| 54.6 \| 47.8 \| 67.6 \| 59.3 \| 50.6 \| 在7B规模模型上，经过人类指令对齐的Qwen-7B-Chat模型，准确率在同类相近规模模型中仍然处于前列。 Compared with other pretrained models with comparable model size, the human-aligned Qwen-7B-Chat performs well in C-Eval accuracy. ### 英文评测（English Evaluation） #### MMLU [MMLU](https://arxiv.org/abs/2009.03300)评测集上，Qwen-7B-Chat模型的zero-shot准确率如下，效果同样在同类对齐模型中同样表现较优。 The zero-shot accuracy of Qwen-7B-Chat on MMLU is provided below. The performance of Qwen-7B-Chat still on the top between other human-aligned models with comparable size. \| Model \| Avg. Acc. \| \| :---------------- \| :-------: \| \| ChatGLM2-6B-Chat \| 45.5 \| \| LLaMA2-7B-Chat \| 47.0 \| \| InternLM-7B-Chat \| 50.8 \| \| Baichuan-13B-Chat \| 52.1 \| \| ChatGLM2-12B-Chat \| 52.1 \| \| Qwen-7B-Chat \| 53.9 \| ### 代码评测（Coding Evaluation） Qwen-7B-Chat在[HumanEval](https://github.com/openai/human-eval)的zero-shot Pass@1效果如下 The zero-shot Pass@1 of Qwen-7B-Chat on [HumanEval](https://github.com/openai/human-eval) is demonstrated below \| Model \| Pass@1 \| \| :---------------- \| :------: \| \| LLaMA2-7B-Chat \| 12.2 \| \| InternLM-7B-Chat \| 14.0 \| \| Baichuan-13B-Chat \| 16.5 \| \| LLaMA2-13B-Chat \| 18.9 \| \| Qwen-7B-Chat \| 24.4 \| ### 数学评测（Mathematics Evaluation）在评测数学能力的[GSM8K](https://github.com/openai/grade-school-math)上，Qwen-7B-Chat的准确率结果如下 The accuracy of Qwen-7B-Chat on GSM8K is shown below \| Model \| Zero-shot Acc. \| 4-shot Acc. \| \| :---------------- \| :------------: \| :--------: \| \| ChatGLM2-6B-Chat \| - \| 28.0 \| \| LLaMA2-7B-Chat \| 20.4 \| 28.2 \| \| LLaMA2-13B-Chat \| 29.4 \| 36.7 \| \| InternLM-7B-Chat \| 32.6 \| 34.5 \| \| Baichuan-13B-Chat \| - \| 36.3 \| \| ChatGLM2-12B-Chat \| - \| 38.1 \| \| Qwen-7B-Chat \| 41.1 \| 43.5 \| ### 长序列评测（Long-Context Understanding）通过NTK插值，LogN注意力缩放可以扩展Qwen-7B-Chat的上下文长度。在长文本摘要数据集[VCSUM](https://arxiv.org/abs/2305.05280)上（文本平均长度在15K左右），Qwen-7B-Chat的Rouge-L结果如下： (若要启用这些技巧，请将config.json里的`use_dynamic_ntk`和`use_logn_attn`设置为true) We introduce NTK-aware interpolation, LogN attention scaling to extend the context length of Qwen-7B-Chat. The Rouge-L results of Qwen-7B-Chat on long-text summarization dataset [VCSUM](https://arxiv.org/abs/2305.05280) (The average length of this dataset is around 15K) are shown below: (To use these tricks, please set `use_dynamic_ntk` and `use_long_attn` to true in config.json.) \| Model \| VCSUM (zh) \| \| :---------------- \| :--------: \| \| GPT-3.5-Turbo-16k \| 16.0 \| \| LLama2-7B-Chat \| 0.2 \| \| InternLM-7B-Chat \| 13.0 \| \| ChatGLM2-6B-Chat \| 16.3 \| \| Qwen-7B-Chat \| 16.6 \| ### 工具使用能力的评测（Tool Usage） #### ReAct Prompting 千问支持通过 [ReAct Prompting](https://arxiv.org/abs/2210.03629) 调用插件/工具/API。ReAct 也是 [LangChain](https://python.langchain.com/) 框架采用的主要方式之一。在我们开源的、用于评估工具使用能力的评测基准上，千问的表现如下： Qwen-7B-Chat supports calling plugins/tools/APIs through [ReAct Prompting](https://arxiv.org/abs/2210.03629). ReAct is also one of the main approaches used by the [LangChain](https://python.langchain.com/) framework. In our evaluation benchmark for assessing tool usage capabilities, Qwen-7B-Chat's performance is as follows: \| Model \| Tool Selection (Acc.↑) \| Tool Input (Rouge-L↑) \| False Positive Error”↓ \| \| :--------------- \| :---------------------: \| :--------------------: \| :--------------------: \| \| GPT-4 \| 95% \| 0.90 \| 15% \| \| GPT-3.5 \| 85% \| 0.88 \| 75% \| \| Qwen-7B-Chat \| 99% \| 0.89 \| 9.7% \| > 评测基准中出现的插件均没有出现在千问的训练集中。该基准评估了模型在多个候选插件中选择正确插件的准确率、传入插件的参数的合理性、以及假阳率。假阳率（False Positive）定义：在处理不该调用插件的请求时，错误地调用了插件。 > The plugins that appear in the evaluation set do not appear in the training set of Qwen-7B-Chat. This benchmark evaluates the accuracy of the model in selecting the correct plugin from multiple candidate plugins, the rationality of the parameters passed into the plugin, and the false positive rate. False Positive: Incorrectly invoking a plugin when it should not have been called when responding to a query. 关于 ReAct Prompting 的 prompt 怎么写、怎么使用，请参考 [ReAct 样例说明](examples/react_prompt.md)。使用工具能使模型更好地完成任务。基于千问的工具使用能力，我们能实现下图所展示的效果： For how to write and use prompts for ReAct Prompting, please refer to [the ReAct examples](examples/react_prompt.md). The use of tools can enable the model to better perform tasks, as shown in the following figures: ![](assets/react_showcase_001.png) ![](assets/react_showcase_002.png) #### Huggingface Agent 千问还具备作为 [HuggingFace Agent](https://huggingface.co/docs/transformers/transformers_agents) 的能力。它在 Huggingface 提供的run模式评测基准上的表现如下： Qwen-7B-Chat also has the capability to be used as a [HuggingFace Agent](https://huggingface.co/docs/transformers/transformers_agents). Its performance on the run-mode benchmark provided by HuggingFace is as follows: \| Model \| Tool Selection↑ \| Tool Used↑ \| Code↑ \| \|:-----------------\| :-------------: \| :---------: \| :-------: \| \| GPT-4 \| 100 \| 100 \| 97.41 \| \| GPT-3.5 \| 95.37 \| 96.30 \| 87.04 \| \| StarCoder-15.5B \| 87.04 \| 87.96 \| 68.89 \| \| Qwen-7B-Chat \| 90.74 \| 92.59 \| 74.07 \| <br> ## FAQ 如遇到问题，敬请查阅[FAQ](https://github.com/QwenLM/Qwen-7B/blob/main/FAQ_zh.md)以及issue区，如仍无法解决再提交issue。 If you meet problems, please refer to [FAQ](https://github.com/QwenLM/Qwen-7B/blob/main/FAQ.md) and the issues first to search a solution before you launch a new issue. <br> ## 使用协议（License Agreement）我们的代码和模型权重对学术研究完全开放，并支持商用。请查看[LICENSE](https://github.com/QwenLM/Qwen-7B/blob/main/LICENSE)了解具体的开源协议细节。如需商用，请填写[问卷](https://dashscope.console.aliyun.com/openModelApply/qianwen)申请。 Our code and checkpoints are open to research purpose, and they are allowed for commercial purposes. Check [LICENSE](https://github.com/QwenLM/Qwen-7B/blob/main/LICENSE) for more details about the license. If you have requirements for commercial use, please fill out the [form](https://dashscope.console.aliyun.com/openModelApply/qianwen) to apply. <br> ## 联系我们（Contact Us）如果你想给我们的研发团队和产品团队留言，请通过邮件（[email protected]）联系我们。 If you are interested to leave a message to either our research team or product team, feel free to send an email to [email protected].	{"language": ["zh", "en"], "pipeline_tag": "text-generation"}	task	[ "SUMMARIZATION" ]	43,154
AlexWortega/instruct_rugptlarge	AlexWortega	text-generation	[ "transformers", "pytorch", "safetensors", "gpt2", "text-generation", "finance", "code", "ru", "dataset:IlyaGusev/ru_turbo_alpaca", "license:apache-2.0", "autotrain_compatible", "text-generation-inference", "endpoints_compatible", "region:us" ]	2023-03-20T21:12:15Z	2023-04-26T11:19:44+00:00	44	10	--- datasets: - IlyaGusev/ru_turbo_alpaca language: - ru library_name: transformers license: apache-2.0 pipeline_tag: text-generation tags: - finance - code inference: parameters: min_length: 20 max_new_tokens: 250 top_k: 50 top_p: 0.9 early_stopping: true no_repeat_ngram_size: 2 use_cache: true repetition_penalty: 1.5 length_penalty: 0.8 num_beams: 2 widget: - text: Может ли встретиться пингвин и белый медведь? example_title: Question Answering - text: Как зарабатывать много денег обучая модели? <instructionS> example_title: Open domain Knoweledge - text: Напиши на python код который выведет привет мир <code> example_title: Code writing - text: 'Переведи на русский и укажи язык оригинала: My name is Arthur.' example_title: Zero shor translate - text: Квадратный корень из x равен кубическому корню из y. Чему равно y в степени 2, если x = 4? example_title: Math example --- <h1 style="font-size: 42px">Instructions ruGPT large v0.11_25к_a<h1/> # Model Summary > Это ruGPTlarge дообученная в инструктивно-флановом сетапе, она более ли менее ZSшотиться и FSшотиться и работает лучше чем XGLM1.7b, mgpt на русском языке # Quick Start ```python from transformers import GPT2TokenizerFast,GPT2LMHeadModel tokenizer = GPT2TokenizerFast.from_pretrained("AlexWortega/instruct_rugptlarge") special_tokens_dict = {'additional_special_tokens': ['<code>', '</code>', '<instructionS>', '<instructionE>', '<next>']} tokenizer.add_special_tokens(special_tokens_dict) device = 'cuda' model = GPT2LMHeadModel.from_pretrained("AlexWortega/instruct_rugptlarge") model.to(device) model.resize_token_embeddings(len(tokenizer)) def generate_seqs(q,model, k=2): gen_kwargs = { "min_length": 20, "max_new_tokens": 100, "top_k": 50, "top_p": 0.7, "do_sample": True, "early_stopping": True, "no_repeat_ngram_size": 2, "eos_token_id": tokenizer.eos_token_id, "pad_token_id": tokenizer.eos_token_id, "use_cache": True, "repetition_penalty": 1.5, "length_penalty": 1.2, "num_beams": 4, "num_return_sequences": k } q = q + '<instructionS>' t = tokenizer.encode(q, return_tensors='pt').to(device) g = model.generate(t, **gen_kwargs) generated_sequences = tokenizer.batch_decode(g, skip_special_tokens=True) return generated_sequences ``` обратите внимание, что лучшие параметры для генерации ``` gen_kwargs = { "min_length": 20, "max_new_tokens": 100, "top_k": 50, "top_p": 0.9, "do_sample": True, "early_stopping": True, "no_repeat_ngram_size": 2, "eos_token_id": tokenizer.eos_token_id, "pad_token_id": tokenizer.eos_token_id, "use_cache": True, "repetition_penalty": 1.5, "length_penalty": 0.8, "num_beams": 4, "num_return_sequences": k } ``` # License The weights of Instructions ruGPT Small v0.1a are licensed under version 2.0 of the Apache License. ## Hyperparameters I used Novograd with a learning rate of 2e-5 and global batch size of 6 (3 for each data parallel worker). I use both data parallelism and pipeline parallelism to conduct training. During training, we truncate the input sequence to 1024 tokens, and for input sequence that contains less than 1024 tokens, we concatenate multiple sequences into one long sequence to improve the data efficiency. # References #Metrics ван дей пипл, ван дееей ## BibTeX entry and citation info ```bibtex @article{ title={GPT2xl is underrated task solver}, author={Nickolich Aleksandr, 5Q, datascience, Ilya Gusev, Alex Kukushkin, Karina Romanova, Arseniy Shahmatov, Maksim Gersimenko}, year={2023} } ```	null	Non_BioNLP	<h1 style="font-size: 42px">Instructions ruGPT large v0.11_25к_a<h1/> # Model Summary > Это ruGPTlarge дообученная в инструктивно-флановом сетапе, она более ли менее ZSшотиться и FSшотиться и работает лучше чем XGLM1.7b, mgpt на русском языке # Quick Start ```python from transformers import GPT2TokenizerFast,GPT2LMHeadModel tokenizer = GPT2TokenizerFast.from_pretrained("AlexWortega/instruct_rugptlarge") special_tokens_dict = {'additional_special_tokens': ['<code>', '</code>', '<instructionS>', '<instructionE>', '<next>']} tokenizer.add_special_tokens(special_tokens_dict) device = 'cuda' model = GPT2LMHeadModel.from_pretrained("AlexWortega/instruct_rugptlarge") model.to(device) model.resize_token_embeddings(len(tokenizer)) def generate_seqs(q,model, k=2): gen_kwargs = { "min_length": 20, "max_new_tokens": 100, "top_k": 50, "top_p": 0.7, "do_sample": True, "early_stopping": True, "no_repeat_ngram_size": 2, "eos_token_id": tokenizer.eos_token_id, "pad_token_id": tokenizer.eos_token_id, "use_cache": True, "repetition_penalty": 1.5, "length_penalty": 1.2, "num_beams": 4, "num_return_sequences": k } q = q + '<instructionS>' t = tokenizer.encode(q, return_tensors='pt').to(device) g = model.generate(t, **gen_kwargs) generated_sequences = tokenizer.batch_decode(g, skip_special_tokens=True) return generated_sequences ``` обратите внимание, что лучшие параметры для генерации ``` gen_kwargs = { "min_length": 20, "max_new_tokens": 100, "top_k": 50, "top_p": 0.9, "do_sample": True, "early_stopping": True, "no_repeat_ngram_size": 2, "eos_token_id": tokenizer.eos_token_id, "pad_token_id": tokenizer.eos_token_id, "use_cache": True, "repetition_penalty": 1.5, "length_penalty": 0.8, "num_beams": 4, "num_return_sequences": k } ``` # License The weights of Instructions ruGPT Small v0.1a are licensed under version 2.0 of the Apache License. ## Hyperparameters I used Novograd with a learning rate of 2e-5 and global batch size of 6 (3 for each data parallel worker). I use both data parallelism and pipeline parallelism to conduct training. During training, we truncate the input sequence to 1024 tokens, and for input sequence that contains less than 1024 tokens, we concatenate multiple sequences into one long sequence to improve the data efficiency. # References #Metrics ван дей пипл, ван дееей ## BibTeX entry and citation info ```bibtex @article{ title={GPT2xl is underrated task solver}, author={Nickolich Aleksandr, 5Q, datascience, Ilya Gusev, Alex Kukushkin, Karina Romanova, Arseniy Shahmatov, Maksim Gersimenko}, year={2023} } ```	{"datasets": ["IlyaGusev/ru_turbo_alpaca"], "language": ["ru"], "library_name": "transformers", "license": "apache-2.0", "pipeline_tag": "text-generation", "tags": ["finance", "code"], "inference": {"parameters": {"min_length": 20, "max_new_tokens": 250, "top_k": 50, "top_p": 0.9, "early_stopping": true, "no_repeat_ngram_size": 2, "use_cache": true, "repetition_penalty": 1.5, "length_penalty": 0.8, "num_beams": 2}}, "widget": [{"text": "Может ли встретиться пингвин и белый медведь?", "example_title": "Question Answering"}, {"text": "Как зарабатывать много денег обучая модели? <instructionS>", "example_title": "Open domain Knoweledge"}, {"text": "Напиши на python код который выведет привет мир <code>", "example_title": "Code writing"}, {"text": "Переведи на русский и укажи язык оригинала: My name is Arthur.", "example_title": "Zero shor translate"}, {"text": "Квадратный корень из x равен кубическому корню из y. Чему равно y в степени 2, если x = 4?", "example_title": "Math example"}]}	task	[ "QUESTION_ANSWERING" ]	43,155
Helsinki-NLP/opus-mt-to-fr	Helsinki-NLP	translation	[ "transformers", "pytorch", "tf", "marian", "text2text-generation", "translation", "to", "fr", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2022-03-02T23:29:04Z	2023-08-16T12:07:09+00:00	38	0	--- license: apache-2.0 tags: - translation --- ### opus-mt-to-fr * source languages: to * target languages: fr * OPUS readme: [to-fr](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/to-fr/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-16.zip](https://object.pouta.csc.fi/OPUS-MT-models/to-fr/opus-2020-01-16.zip) * test set translations: [opus-2020-01-16.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/to-fr/opus-2020-01-16.test.txt) * test set scores: [opus-2020-01-16.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/to-fr/opus-2020-01-16.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.to.fr \| 27.9 \| 0.456 \|	null	Non_BioNLP	### opus-mt-to-fr * source languages: to * target languages: fr * OPUS readme: [to-fr](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/to-fr/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-16.zip](https://object.pouta.csc.fi/OPUS-MT-models/to-fr/opus-2020-01-16.zip) * test set translations: [opus-2020-01-16.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/to-fr/opus-2020-01-16.test.txt) * test set scores: [opus-2020-01-16.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/to-fr/opus-2020-01-16.eval.txt) ## Benchmarks \| testset \| BLEU \| chr-F \| \|-----------------------\|-------\|-------\| \| JW300.to.fr \| 27.9 \| 0.456 \|	{"license": "apache-2.0", "tags": ["translation"]}	task	[ "TRANSLATION" ]	43,156
AdapterHub/roberta-base-pf-wikihop	AdapterHub	question-answering	[ "adapter-transformers", "question-answering", "roberta", "adapterhub:qa/wikihop", "en", "arxiv:2104.08247", "region:us" ]	2022-03-02T23:29:04Z	2021-11-15T10:44:47+00:00	5	0	--- language: - en tags: - question-answering - roberta - adapterhub:qa/wikihop - adapter-transformers --- # Adapter `AdapterHub/roberta-base-pf-wikihop` for roberta-base An [adapter](https://adapterhub.ml) for the `roberta-base` model that was trained on the [qa/wikihop](https://adapterhub.ml/explore/qa/wikihop/) dataset and includes a prediction head for question answering. This adapter was created for usage with the [adapter-transformers](https://github.com/Adapter-Hub/adapter-transformers) library. ## Usage First, install `adapter-transformers`: ``` pip install -U adapter-transformers ``` _Note: adapter-transformers is a fork of transformers that acts as a drop-in replacement with adapter support. [More](https://docs.adapterhub.ml/installation.html)_ Now, the adapter can be loaded and activated like this: ```python from transformers import AutoModelWithHeads model = AutoModelWithHeads.from_pretrained("roberta-base") adapter_name = model.load_adapter("AdapterHub/roberta-base-pf-wikihop", source="hf") model.active_adapters = adapter_name ``` ## Architecture & Training The training code for this adapter is available at https://github.com/adapter-hub/efficient-task-transfer. In particular, training configurations for all tasks can be found [here](https://github.com/adapter-hub/efficient-task-transfer/tree/master/run_configs). ## Evaluation results Refer to [the paper](https://arxiv.org/pdf/2104.08247) for more information on results. ## Citation If you use this adapter, please cite our paper ["What to Pre-Train on? Efficient Intermediate Task Selection"](https://arxiv.org/pdf/2104.08247): ```bibtex @inproceedings{poth-etal-2021-pre, title = "{W}hat to Pre-Train on? {E}fficient Intermediate Task Selection", author = {Poth, Clifton and Pfeiffer, Jonas and R{"u}ckl{'e}, Andreas and Gurevych, Iryna}, booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing", month = nov, year = "2021", address = "Online and Punta Cana, Dominican Republic", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2021.emnlp-main.827", pages = "10585--10605", } ```	null	Non_BioNLP	# Adapter `AdapterHub/roberta-base-pf-wikihop` for roberta-base An [adapter](https://adapterhub.ml) for the `roberta-base` model that was trained on the [qa/wikihop](https://adapterhub.ml/explore/qa/wikihop/) dataset and includes a prediction head for question answering. This adapter was created for usage with the [adapter-transformers](https://github.com/Adapter-Hub/adapter-transformers) library. ## Usage First, install `adapter-transformers`: ``` pip install -U adapter-transformers ``` _Note: adapter-transformers is a fork of transformers that acts as a drop-in replacement with adapter support. [More](https://docs.adapterhub.ml/installation.html)_ Now, the adapter can be loaded and activated like this: ```python from transformers import AutoModelWithHeads model = AutoModelWithHeads.from_pretrained("roberta-base") adapter_name = model.load_adapter("AdapterHub/roberta-base-pf-wikihop", source="hf") model.active_adapters = adapter_name ``` ## Architecture & Training The training code for this adapter is available at https://github.com/adapter-hub/efficient-task-transfer. In particular, training configurations for all tasks can be found [here](https://github.com/adapter-hub/efficient-task-transfer/tree/master/run_configs). ## Evaluation results Refer to [the paper](https://arxiv.org/pdf/2104.08247) for more information on results. ## Citation If you use this adapter, please cite our paper ["What to Pre-Train on? Efficient Intermediate Task Selection"](https://arxiv.org/pdf/2104.08247): ```bibtex @inproceedings{poth-etal-2021-pre, title = "{W}hat to Pre-Train on? {E}fficient Intermediate Task Selection", author = {Poth, Clifton and Pfeiffer, Jonas and R{"u}ckl{'e}, Andreas and Gurevych, Iryna}, booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing", month = nov, year = "2021", address = "Online and Punta Cana, Dominican Republic", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2021.emnlp-main.827", pages = "10585--10605", } ```	{"language": ["en"], "tags": ["question-answering", "roberta", "adapterhub:qa/wikihop", "adapter-transformers"]}	task	[ "QUESTION_ANSWERING" ]	43,158
binh230/bge-m3-finetuned-1	binh230	sentence-similarity	[ "sentence-transformers", "safetensors", "xlm-roberta", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:118262", "loss:MultipleNegativesRankingLoss", "arxiv:1908.10084", "arxiv:1705.00652", "base_model:BAAI/bge-m3", "base_model:finetune:BAAI/bge-m3", "model-index", "autotrain_compatible", "text-embeddings-inference", "endpoints_compatible", "region:us" ]	2024-12-07T04:57:50Z	2024-12-07T04:58:46+00:00	7	0	--- base_model: BAAI/bge-m3 library_name: sentence-transformers metrics: - cosine_accuracy@1 - cosine_accuracy@3 - cosine_accuracy@5 - cosine_accuracy@10 - cosine_precision@1 - cosine_precision@3 - cosine_precision@5 - cosine_precision@10 - cosine_recall@1 - cosine_recall@3 - cosine_recall@5 - cosine_recall@10 - cosine_ndcg@10 - cosine_mrr@10 - cosine_map@100 pipeline_tag: sentence-similarity tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:118262 - loss:MultipleNegativesRankingLoss widget: - source_sentence: Ai ra quyết định thành lập Tủ sách pháp luật tại xã đặc biệt khó khăn? sentences: - 'Hội đồng dân tộc, các Ủy ban của Quốc hội 1. Hội đồng dân tộc, Ủy ban của Quốc hội là cơ quan của Quốc hội, chịu trách nhiệm và báo cáo công tác trước Quốc hội; trong thời gian Quốc hội không họp thì báo cáo công tác trước Ủy ban thường vụ Quốc hội. 2. Các Ủy ban của Quốc hội gồm: a) Ủy ban pháp luật; b) Ủy ban tư pháp; c) Ủy ban kinh tế; d) Ủy ban tài chính, ngân sách; đ) Ủy ban quốc phòng và an ninh; e) Ủy ban văn hoá, giáo dục; g) Ủy ban xã hội; h) Ủy ban khoa học, công nghệ và môi trường; i) Ủy ban đối ngoại. 3. Quốc hội thành lập Ủy ban lâm thời của Quốc hội theo quy định tại Điều 88 và Điều 89 của Luật này.' - 'Xây dựng Tủ sách pháp luật 1. Thành lập Tủ sách pháp luật a) Chủ tịch Ủy ban nhân dân xã đặc biệt khó khăn, Thủ trưởng cơ quan, đơn vị của lực lượng vũ trang nhân dân ra quyết định thành lập Tủ sách pháp luật. Quyết định thành lập phải quy định cụ thể về địa điểm đặt Tủ sách pháp luật; người phụ trách và quyền hạn, nhiệm vụ của người phụ trách; phạm vi và đối tượng phục vụ; Nội quy hoạt động và kinh phí hoạt động của Tủ sách pháp luật. b) Chủ tịch Ủy ban nhân dân cấp tỉnh chỉ đạo việc quản lý, khai thác thống nhất sách, tài liệu của Tủ sách pháp luật và sách, tài liệu pháp luật của Thư viện xã, điểm Bưu điện - Văn hóa xã, Trung tâm học tập cộng đồng phù hợp với thực tiễn tại địa phương. Trên cơ sở ý kiến của Thủ tướng Chính phủ và Ban Bí thư Trung ương Đảng quy định tại điểm d khoản 1 Điều 14 của Quyết định này, Chủ tịch Ủy ban nhân dân cấp tỉnh thực hiện quản lý sách, tài liệu của Tủ sách pháp luật và sách, tài liệu pháp luật của Đề án trang bị sách cho cơ sở xã, phường, thị trấn.' - 'Nhiệm vụ, quyền hạn của trường trung cấp Trường trung cấp thực hiện nhiệm vụ, quyền hạn theo quy định tại Điều 23 của Luật Giáo dục nghề nghiệp, quy định của pháp luật có liên quan và các quy định sau đây: ... 2. Quyền hạn: a) Xây dựng và tổ chức thực hiện kế hoạch, chiến lược phát triển trường phù hợp với chiến lược phát triển giáo dục nghề nghiệp nhằm đáp ứng yêu cầu của thị trường lao động; b) Tổ chức đào tạo các chương trình giáo dục nghề nghiệp theo quy định của Bộ trưởng Bộ Lao động - Thương binh và Xã hội; c) Liên kết với cơ sở giáo dục nghề nghiệp, cơ sở giáo dục đại học, doanh nghiệp, tổ chức trong nước và nước ngoài theo quy định của pháp luật nhằm khai thác, huy động các nguồn lực xã hội trong thực hiện các chương trình đào tạo giáo dục nghề nghiệp để nâng cao chất lượng đào tạo, gắn đào tạo với việc làm và thị trường lao động. Đơn vị chủ trì liên kết đào tạo chịu trách nhiệm cấp văn bằng, chứng chỉ cho người học; d) Liên kết, phối hợp với trường cao đẳng tổ chức đào tạo liên thông từ trình độ sơ cấp, trình độ trung cấp lên trình độ cao đẳng theo quy định của Luật Giáo dục nghề nghiệp và các văn bản hướng dẫn thi hành; đ) Liên kết, phối hợp với cơ sở giáo dục đại học tổ chức đào tạo trình độ đại học theo hình thức vừa làm vừa học và đào tạo liên thông từ trình độ trung cấp lên trình độ đại học theo quy định của Luật Giáo dục đại học và các văn bản hướng dẫn thi hành; e) Phối hợp với doanh nghiệp, cơ sở sản xuất, kinh doanh, dịch vụ trong hoạt động xây dựng chương trình, giáo trình đào tạo, tài liệu, học liệu giảng dạy; tổ chức giảng dạy, hướng dẫn thực hành, thực tập, đánh giá kết quả học tập của người học; bồi dưỡng nâng cao trình độ chuyên môn, nghiệp vụ cho nhà giáo; g) Tổ chức giảng dạy khối lượng kiến thức văn hóa trung học phổ thông cho học sinh có bằng tốt nghiệp trung học cơ sở theo học trình độ trung cấp theo quy định của Luật Giáo dục và các văn bản hướng dẫn thi hành;: ...' - source_sentence: Việc tinh giản biên chế hiện nay được thực hiện theo những nguyên tắc nào? sentences: - 'Xử phạt hành vi vi phạm trong việc tổ chức bồi dưỡng và cấp Chứng chỉ bồi dưỡng kế toán trưởng ... 4. Phạt tiền từ 10.000.000 đồng đến 20.000.000 đồng đối với cơ sở đào tạo thực hiện một trong các hành vi sau: a) Mở khóa học bồi dưỡng kế toán trưởng khi không đủ điều kiện; b) Mở khóa học bồi dưỡng kế toán trưởng cho người nước ngoài khi chưa được Bộ Tài chính chấp thuận; c) Quản lý phôi và cấp Chứng chỉ bồi dưỡng kế toán trưởng của cơ sở đào tạo không đúng quy định của Bộ Tài chính. ... 6. Biện pháp khắc phục hậu quả: Buộc nộp lại số lợi bất hợp pháp có được do thực hiện hành vi vi phạm quy định tại điểm a, b khoản 4, khoản 5 Điều này.Mức phạt tiền trong lĩnh vực kế toán, kiểm toán độc lập ... 2. Mức phạt tiền quy định tại Chương II, Chương III Nghị định này là mức phạt tiền áp dụng đối với tổ chức trừ trường hợp quy định tại khoản 1 Điều 7; Điều 8; Điều 9; Điều 10; Điều 11; Điều 13; Điều 14; Điều 15; Điều 16; Điều 17; Điều 19; khoản 1, khoản 3 Điều 21; Điều 22; Điều 23; Điều 24; Điều 26; Điều 33; Điều 34; khoản 1, khoản 3 Điều 36; khoản 1 Điều 38; khoản 2, khoản 3 Điều 39; khoản 1, khoản 2 Điều 48; khoản 1 Điều 57; khoản 1, khoản 2 Điều 61; Điều 67 là mức phạt tiền đối với cá nhân. Đối với tổ chức có cùng hành vi vi phạm thì mức phạt tiền gấp 02 lần mức phạt tiền đối với cá nhân. ...' - 'Nguyên tắc tinh giản biên chế 1. Bảo đảm sự lãnh đạo của Đảng, phát huy vai trò giám sát của các tổ chức chính trị - xã hội và nhân dân trong quá trình thực hiện tinh giản biên chế. 2. Gắn tinh giản biên chế với sắp xếp tổ chức bộ máy theo hướng tinh gọn, hoạt động hiệu lực, hiệu quả và cơ cấu lại, nâng cao chất lượng đội ngũ cán bộ, công chức, viên chức theo vị trí việc làm, phù hợp với cơ chế tự chủ của cơ quan, tổ chức, đơn vị. 3. Bảo đảm nguyên tắc tập trung dân chủ, khách quan, công bằng, công khai, minh bạch và theo quy định của pháp luật. 4. Bảo đảm chi trả chế độ, chính sách tinh giản biên chế kịp thời, đầy đủ theo quy định của pháp luật, bảo đảm sử dụng hiệu quả ngân sách nhà nước. 5. Người đứng đầu phải chịu trách nhiệm về kết quả thực hiện tinh giản biên chế trong cơ quan, tổ chức, đơn vị được giao quản lý theo thẩm quyền. 6. Đối tượng tinh giản biên chế nếu được bầu cử, tuyển dụng lại vào các cơ quan, tổ chức, đơn vị hưởng lương từ ngân sách nhà nước trong thời gian 60 tháng kể từ ngày thực hiện tinh giản biên chế thì phải hoàn trả lại số tiền trợ cấp đã nhận cho cơ quan, tổ chức, đơn vị đã chi trả trợ cấp.' - '"3. Tiêu chuẩn về trình độ đào tạo, bồi dưỡng a) Có bằng cử nhân trở lên thuộc ngành đào tạo giáo viên đối với giáo viên trung học cơ sở. Trường hợp môn học chưa đủ giáo viên có bằng cử nhân thuộc ngành đào tạo giáo viên thì phải có bằng cử nhân chuyên ngành phù hợp và có chứng chỉ bồi dưỡng nghiệp vụ sư phạm dành cho giáo viên trung học cơ sở theo chương trình do Bộ trưởng Bộ Giáo dục và Đào tạo ban hành; b) Có chứng chỉ bồi dưỡng theo tiêu chuẩn chức danh nghề nghiệp giáo viên trung học cơ sở hạng III (đối với giáo viên trung học cơ sở mới được tuyển dụng vào giáo viên trung học cơ sở hạng III thì phải có chứng chỉ trong thời gian 36 tháng kể từ ngày được tuyển dụng)."' - source_sentence: Khi xuất bản phụ trương thì cần đáp ứng những điều kiện gì? sentences: - 'Hồ sơ, trình tự, thủ tục chấp thuận chủ trương đầu tư ra nước ngoài của Quốc hội 1. Nhà đầu tư nộp hồ sơ dự án đầu tư ra nước ngoài cho Bộ Kế hoạch và Đầu tư. Hồ sơ bao gồm: a) Văn bản đăng ký đầu tư ra nước ngoài; b) Tài liệu về tư cách pháp lý của nhà đầu tư; c) Đề xuất dự án đầu tư gồm các nội dung chủ yếu sau: hình thức, mục tiêu, quy mô, địa điểm đầu tư; xác định sơ bộ vốn đầu tư, phương án huy động vốn, cơ cấu nguồn vốn; tiến độ thực hiện dự án, các giai đoạn đầu tư (nếu có); phân tích sơ bộ hiệu quả đầu tư của dự án; d) Tài liệu chứng minh năng lực tài chính của nhà đầu tư gồm ít nhất một trong các tài liệu sau: báo cáo tài chính 02 năm gần nhất của nhà đầu tư; cam kết hỗ trợ tài chính của công ty mẹ; cam kết hỗ trợ tài chính của tổ chức tài chính; bảo lãnh về năng lực tài chính của nhà đầu tư; tài liệu khác chứng minh năng lực tài chính của nhà đầu tư; đ) Cam kết tự cân đối nguồn ngoại tệ hoặc văn bản cam kết thu xếp ngoại tệ cho nhà đầu tư của tổ chức tín dụng được phép; e) Văn bản của cơ quan đại diện chủ sở hữu chấp thuận nhà đầu tư thực hiện hoạt động đầu tư ra nước ngoài và báo cáo thẩm định nội bộ về đề xuất đầu tư ra nước ngoài của doanh nghiệp nhà nước quy định tại khoản 1 Điều 59 của Luật này hoặc quyết định đầu tư ra nước ngoài theo quy định tại khoản 2 Điều 59 của Luật này; g) Đối với dự án đầu tư ra nước ngoài trong các ngành, nghề quy định tại khoản 1 Điều 54 của Luật này, nhà đầu tư nộp văn bản của cơ quan nhà nước có thẩm quyền về việc đáp ứng điều kiện đầu tư ra nước ngoài theo quy định của pháp luật có liên quan (nếu có). 2. Trong thời hạn 05 ngày làm việc kể từ ngày nhận đủ hồ sơ, Bộ Kế hoạch và Đầu tư trình Thủ tướng Chính phủ quyết định thành lập Hội đồng thẩm định nhà nước. 3. Trong thời hạn 90 ngày kể từ ngày thành lập, Hội đồng thẩm định nhà nước tổ chức thẩm định và lập báo cáo thẩm định trình Chính phủ. Báo cáo thẩm định gồm các nội dung sau đây: a) Điều kiện cấp Giấy chứng nhận đăng ký đầu tư ra nước ngoài quy định tại Điều 60 của Luật này; b) Tư cách pháp lý của nhà đầu tư; c) S' - 'Vi phạm quy định về tổ chức, quản trị, điều hành 1. Phạt tiền từ 20.000.000 đồng đến 30.000.000 đồng đối với một trong các hành vi vi phạm sau đây: a) Vi phạm quy định về Hội đồng quản trị, Hội đồng thành viên, Ban kiểm soát quy định tại Điều 43, Điều 44, Điều 62, khoản 1 Điều 70, Điều 81, Điều 84 Luật Các tổ chức tín dụng; b) Không duy trì việc đảm bảo đầy đủ các điều kiện, tiêu chuẩn đối với thành viên Hội đồng quản trị, Ban kiểm soát, Tổng giám đốc (Giám đốc) theo quy định của Luật Các tổ chức tín dụng; c) Không tổ chức hoặc tổ chức Đại hội đồng cổ đông, Đại hội thành viên không đúng quy định. ...Hình thức xử phạt, mức phạt tiền, thẩm quyền phạt tiền và biện pháp khắc phục hậu quả ... 3. Mức phạt tiền và thẩm quyền phạt tiền: ... b) Mức phạt tiền quy định tại Chương II Nghị định này là mức phạt tiền áp dụng đối với cá nhân; mức phạt tiền đối với tổ chức có cùng một hành vi vi phạm hành chính bằng 02 lần mức phạt tiền đối với cá nhân; ...' - 'Điều kiện xuất bản thêm ấn phẩm báo chí, phụ trương; sản xuất thêm kênh phát thanh, kênh truyền hình; mở chuyên trang của báo điện tử 1. Có tôn chỉ, Mục đích phù hợp với tôn chỉ, Mục đích của cơ quan báo chí; đối tượng phục vụ; nơi đặt máy chủ và đơn vị cung cấp dịch vụ kết nối (đối với chuyên trang của báo điện tử) phù hợp với nội dung giấy phép hoạt động của cơ quan báo chí. 2. Có phương án về tổ chức và nhân sự bảo đảm hoạt động của ấn phẩm báo chí, phụ trương, kênh phát thanh, kênh truyền hình, chuyên trang của báo điện tử. 3. Có tên và hình thức trình bày tên ấn phẩm báo chí, phụ trương; tên và biểu tượng kênh phát thanh, kênh truyền hình; tên và giao diện trang chủ chuyên trang của báo điện tử. 4. Có các Điều kiện cơ sở vật chất, kỹ thuật; phương án tài chính; có các giải pháp kỹ thuật bảo đảm an toàn, an ninh thông tin; có tên miền cấp dưới phù hợp với tên miền đã đăng ký đối với chuyên trang của báo điện tử; có phương án, kế hoạch thuê hoặc sử dụng hạ tầng truyền dẫn, phát sóng đối với kênh phát thanh, kênh truyền hình.' - source_sentence: Người dân có bị xóa đăng ký thường trú trong trường hợp lên thành phố khác sinh sống hơn 15 năm không về lại nơi thường trú ban đầu không? sentences: - 'Hồ sơ đề nghị hỗ trợ học nghề 1. Hồ sơ đề nghị hỗ trợ học nghề đối với người lao động đang chờ kết quả giải quyết hưởng trợ cấp thất nghiệp hoặc đang hưởng trợ cấp thất nghiệp mà có nhu cầu học nghề tại địa phương nơi đang chờ kết quả hoặc đang hưởng trợ cấp thất nghiệp là đề nghị hỗ trợ học nghề theo Mẫu số 03 ban hành kèm theo Nghị định này. 2. Hồ sơ đề nghị hỗ trợ học nghề đối với người lao động đang hưởng trợ cấp thất nghiệp mà có nhu cầu học nghề tại địa phương không phải nơi đang hưởng trợ cấp thất nghiệp bao gồm đề nghị hỗ trợ học nghề theo quy định tại khoản 1 Điều này và quyết định về việc hưởng trợ cấp thất nghiệp. Quyết định về việc hưởng trợ cấp thất nghiệp là bản chính hoặc bản sao có chứng thực hoặc bản sao kèm theo bản chính để đối chiếu. 3. Hồ sơ đề nghị hỗ trợ học nghề đối với người lao động không thuộc trường hợp theo quy định tại khoản 1, khoản 2 Điều này, bao gồm: a) Đề nghị hưởng trợ cấp thất nghiệp; b) Đề nghị hỗ trợ học nghề theo quy định tại khoản 1 Điều này; c) Bản chính hoặc bản sao có chứng thực hoặc bản sao kèm theo bản chính để đối chiếu của một trong các giấy tờ theo quy định tại khoản 2 Điều 16 Nghị định số 28/2015/NĐ-CP đã được sửa đổi, bổ sung tại khoản 6 Điều 1 Nghị định này; d) Sổ bảo hiểm xã hội. Trường hợp người lao động đã nộp hồ sơ đề nghị hưởng trợ cấp thất nghiệp nhưng không đáp ứng đủ điều kiện hưởng thì hồ sơ đề nghị hỗ trợ học nghề là đề nghị hỗ trợ học nghề theo quy định tại khoản 1 Điều này và thông báo của trung tâm dịch vụ việc làm về việc người lao động không đáp ứng đủ điều kiện hưởng trợ cấp thất nghiệp.' - '"1. Người thuộc một trong các trường hợp sau đây thì bị xóa đăng ký thường trú: d) Vắng mặt liên tục tại nơi thường trú từ 12 tháng trở lên mà không đăng ký tạm trú tại chỗ ở khác hoặc không khai báo tạm vắng, trừ trường hợp xuất cảnh ra nước ngoài nhưng không phải để định cư hoặc trường hợp đang chấp hành án phạt tù, chấp hành biện pháp đưa vào cơ sở giáo dục bắt buộc, đưa vào cơ sở cai nghiện bắt buộc, đưa vào trường giáo dưỡng;"' - '“1. Công ty hợp danh là doanh nghiệp, trong đó: a) Phải có ít nhất 02 thành viên là chủ sở hữu chung của công ty, cùng nhau kinh doanh dưới một tên chung (sau đây gọi là thành viên hợp danh). Ngoài các thành viên hợp danh, công ty có thể có thêm thành viên góp vốn; b) Thành viên hợp danh phải là cá nhân, chịu trách nhiệm bằng toàn bộ tài sản của mình về các nghĩa vụ của công ty; c) Thành viên góp vốn là tổ chức, cá nhân và chỉ chịu trách nhiệm về các khoản nợ của công ty trong phạm vi số vốn đã cam kết góp vào công ty.”' - source_sentence: Lấn chiếm đất công là đất phi nông nghiệp để xây dựng công trình thì sẽ bị xử phạt vi phạm hành chính như thế nào? sentences: - "1. Mọi người có quyền tự do tín ngưỡng, tôn giáo, theo hoặc không theo một tôn\ \ giáo nào. Các tôn giáo bình đẳng trước pháp luật.\n2. Nhà nước tôn trọng và\ \ bảo hộ quyền tự do tín ngưỡng, tôn giáo. \n3. Không ai được xâm phạm tự do tín\ \ ngưỡng, tôn giáo hoặc lợi dụng tín ngưỡng, tôn giáo để vi phạm pháp luật.1.\ \ Nhà nước tôn trọng và bảo hộ quyền tự do tín ngưỡng, tôn giáo của mọi người;\ \ bảo đảm để các tôn giáo bình đẳng trước pháp luật.\n2. Nhà nước tôn trọng, bảo\ \ vệ giá trị văn hóa, đạo đức tốt đẹp của tín ngưỡng, tôn giáo, truyền thống thờ\ \ cúng tổ tiên, tôn vinh người có công với đất nước, với cộng đồng đáp ứng nhu\ \ cầu tinh thần của Nhân dân.\n3. Nhà nước bảo hộ cơ sở tín ngưỡng, cơ sở tôn\ \ giáo và tài sản hợp pháp của cơ sở tín ngưỡng, tổ chức tôn giáo." - 'Quy định chung về quy cách thể hiện 1. Nội dung thể hiện đối với thuyết minh, bản vẽ của nhiệm vụ và đồ án trong hồ sơ lấy ý kiến, hồ sơ trình thẩm định và phê duyệt phải được trình bày rõ ràng, đầy đủ nội dung đối với từng loại, cấp độ quy hoạch và tuân thủ các quy định tại Chương II và các phụ lục của Thông tư này. 2. Hệ thống ký hiệu bản vẽ của nhiệm vụ và đồ án quy hoạch đô thị, nông thôn phải tuân thủ theo quy định tại các phụ lục kèm theo Thông tư này. 3. Căn cứ điều kiện, đặc điểm của vùng liên huyện, vùng huyện, đô thị, khu chức năng và khu vực nông thôn được lập quy hoạch, có thể thành lập các bản đồ, sơ đồ riêng cho từng nội dung hiện trạng, định hướng hạ tầng kỹ thuật để đảm bảo thể hiện được đầy đủ các nội dung theo yêu cầu. 4. Trường hợp quy hoạch vùng, quy hoạch tỉnh chưa được phê duyệt, các bản đồ phương hướng phát triển hệ thống đô thị, nông thôn; bản đồ phương án quy hoạch hệ thống đô thị, nông thôn; bản đồ quy hoạch sử dụng đất cấp tỉnh; bản đồ tổ chức không gian và phân vùng chức năng trong quy hoạch vùng, quy hoạch tỉnh được thay thế bằng sơ đồ định hướng phát triển không gian trong quy hoạch xây dựng vùng tỉnh đã được phê duyệt. 5. Tất cả các sơ đồ, bản đồ trong thành phần bản vẽ của hồ sơ nhiệm vụ và hồ sơ đồ án quy hoạch đô thị và nông thôn phải thể hiện rõ phạm vi ranh giới lập quy hoạch; ranh giới, tên các đơn vị hành chính trong phạm vi lập quy hoạch; tên đơn vị hành chính tiếp giáp bên ngoài phạm vi lập quy hoạch.' - 'Lấn, chiếm đất ... 4. Trường hợp lấn, chiếm đất phi nông nghiệp, trừ trường hợp quy định tại khoản 6 Điều này tại khu vực nông thôn thì hình thức và mức xử phạt như sau: a) Phạt tiền từ 10.000.000 đồng đến 20.000.000 đồng nếu diện tích đất lấn, chiếm dưới 0,05 héc ta; b) Phạt tiền từ 20.000.000 đồng đến 40.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,05 héc ta đến dưới 0,1 héc ta; c) Phạt tiền từ 40.000.000 đồng đến 100.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,1 héc ta đến dưới 0,5 héc ta; d) Phạt tiền từ 100.000.000 đồng đến 200.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,5 héc ta đến dưới 01 héc ta; đ) Phạt tiền từ 200.000.000 đồng đến 500.000.000 đồng nếu diện tích đất lấn, chiếm từ 01 héc ta trở lên. 5. Trường hợp lấn, chiếm đất chưa sử dụng, đất nông nghiệp, đất phi nông nghiệp (trừ trường hợp quy định tại khoản 6 Điều này) tại khu vực đô thị thì mức xử phạt bằng 02 lần mức xử phạt đối với loại đất tương ứng quy định tại các khoản 1, 2, 3 và 4 Điều này và mức phạt tối đa không quá 500.000.000 đồng đối với cá nhân, không quá 1.000.000.000 đồng đối với tổ chức. 6. Trường hợp lấn, chiếm đất thuộc hành lang bảo vệ an toàn công trình và đất công trình có hành lang bảo vệ, đất trụ sở làm việc và cơ sở hoạt động sự nghiệp của cơ quan, tổ chức theo quy định của pháp luật về quản lý, sử dụng tài sản nhà nước thì hình thức và mức xử phạt thực hiện theo quy định của pháp luật về xử phạt vi phạm hành chính trong lĩnh vực về hoạt động đầu tư xây dựng; khai thác, chế biến, kinh doanh khoáng sản làm vật liệu xây dựng, sản xuất, kinh doanh vật liệu xây dựng; quản lý công trình hạ tầng kỹ thuật; kinh doanh bất động sản, phát triển nhà ở, quản lý sử dụng nhà và công sở; trong lĩnh vực về giao thông đường bộ và đường sắt; trong lĩnh vực về văn hóa, thể thao, du lịch và quảng cáo; trong lĩnh vực về khai thác và bảo vệ công trình thủy lợi; đê điều; phòng, chống lụt, bão; trong lĩnh vực quản lý, sử dụng tài sản nhà nước và các lĩnh vực chuyên ngành khác. ...Áp dụng mức phạt tiền 1. Đối tượng áp dụng mức phạt tiền thực ' model-index: - name: SentenceTransformer based on BAAI/bge-m3 results: - task: type: information-retrieval name: Information Retrieval dataset: name: Unknown type: unknown metrics: - type: cosine_accuracy@1 value: 0.9120603015075377 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.983249581239531 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.990787269681742 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.9958123953098827 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.9120603015075377 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.327749860413177 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.1981574539363484 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09958123953098827 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.9120603015075377 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.983249581239531 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.990787269681742 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.9958123953098827 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.9596142418815723 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.9473638031426979 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.9475627411515736 name: Cosine Map@100 --- # SentenceTransformer based on BAAI/bge-m3 This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-m3](https://huggingface.co/BAAI/bge-m3). It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [BAAI/bge-m3](https://huggingface.co/BAAI/bge-m3) <!-- at revision 5617a9f61b028005a4858fdac845db406aefb181 --> - Maximum Sequence Length: 8192 tokens - Output Dimensionality: 1024 dimensions - Similarity Function: Cosine Similarity <!-- - Training Dataset: Unknown --> <!-- - Language: Unknown --> <!-- - License: Unknown --> ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("sentence_transformers_model_id") # Run inference sentences = [ 'Lấn chiếm đất công là đất phi nông nghiệp để xây dựng công trình thì sẽ bị xử phạt vi phạm hành chính như thế nào?', 'Lấn, chiếm đất\n...\n4. Trường hợp lấn, chiếm đất phi nông nghiệp, trừ trường hợp quy định tại khoản 6 Điều này tại khu vực nông thôn thì hình thức và mức xử phạt như sau:\na) Phạt tiền từ 10.000.000 đồng đến 20.000.000 đồng nếu diện tích đất lấn, chiếm dưới 0,05 héc ta;\nb) Phạt tiền từ 20.000.000 đồng đến 40.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,05 héc ta đến dưới 0,1 héc ta;\nc) Phạt tiền từ 40.000.000 đồng đến 100.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,1 héc ta đến dưới 0,5 héc ta;\nd) Phạt tiền từ 100.000.000 đồng đến 200.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,5 héc ta đến dưới 01 héc ta;\nđ) Phạt tiền từ 200.000.000 đồng đến 500.000.000 đồng nếu diện tích đất lấn, chiếm từ 01 héc ta trở lên.\n5. Trường hợp lấn, chiếm đất chưa sử dụng, đất nông nghiệp, đất phi nông nghiệp (trừ trường hợp quy định tại khoản 6 Điều này) tại khu vực đô thị thì mức xử phạt bằng 02 lần mức xử phạt đối với loại đất tương ứng quy định tại các khoản 1, 2, 3 và 4 Điều này và mức phạt tối đa không quá 500.000.000 đồng đối với cá nhân, không quá 1.000.000.000 đồng đối với tổ chức.\n6. Trường hợp lấn, chiếm đất thuộc hành lang bảo vệ an toàn công trình và đất công trình có hành lang bảo vệ, đất trụ sở làm việc và cơ sở hoạt động sự nghiệp của cơ quan, tổ chức theo quy định của pháp luật về quản lý, sử dụng tài sản nhà nước thì hình thức và mức xử phạt thực hiện theo quy định của pháp luật về xử phạt vi phạm hành chính trong lĩnh vực về hoạt động đầu tư xây dựng; khai thác, chế biến, kinh doanh khoáng sản làm vật liệu xây dựng, sản xuất, kinh doanh vật liệu xây dựng; quản lý công trình hạ tầng kỹ thuật; kinh doanh bất động sản, phát triển nhà ở, quản lý sử dụng nhà và công sở; trong lĩnh vực về giao thông đường bộ và đường sắt; trong lĩnh vực về văn hóa, thể thao, du lịch và quảng cáo; trong lĩnh vực về khai thác và bảo vệ công trình thủy lợi; đê điều; phòng, chống lụt, bão; trong lĩnh vực quản lý, sử dụng tài sản nhà nước và các lĩnh vực chuyên ngành khác.\n...Áp dụng mức phạt tiền\n1. Đối tượng áp dụng mức phạt tiền thực ', 'Quy định chung về quy cách thể hiện\n1. Nội dung thể hiện đối với thuyết minh, bản vẽ của nhiệm vụ và đồ án trong hồ sơ lấy ý kiến, hồ sơ trình thẩm định và phê duyệt phải được trình bày rõ ràng, đầy đủ nội dung đối với từng loại, cấp độ quy hoạch và tuân thủ các quy định tại Chương II và các phụ lục của Thông tư này.\n2. Hệ thống ký hiệu bản vẽ của nhiệm vụ và đồ án quy hoạch đô thị, nông thôn phải tuân thủ theo quy định tại các phụ lục kèm theo Thông tư này.\n3. Căn cứ điều kiện, đặc điểm của vùng liên huyện, vùng huyện, đô thị, khu chức năng và khu vực nông thôn được lập quy hoạch, có thể thành lập các bản đồ, sơ đồ riêng cho từng nội dung hiện trạng, định hướng hạ tầng kỹ thuật để đảm bảo thể hiện được đầy đủ các nội dung theo yêu cầu.\n4. Trường hợp quy hoạch vùng, quy hoạch tỉnh chưa được phê duyệt, các bản đồ phương hướng phát triển hệ thống đô thị, nông thôn; bản đồ phương án quy hoạch hệ thống đô thị, nông thôn; bản đồ quy hoạch sử dụng đất cấp tỉnh; bản đồ tổ chức không gian và phân vùng chức năng trong quy hoạch vùng, quy hoạch tỉnh được thay thế bằng sơ đồ định hướng phát triển không gian trong quy hoạch xây dựng vùng tỉnh đã được phê duyệt.\n5. Tất cả các sơ đồ, bản đồ trong thành phần bản vẽ của hồ sơ nhiệm vụ và hồ sơ đồ án quy hoạch đô thị và nông thôn phải thể hiện rõ phạm vi ranh giới lập quy hoạch; ranh giới, tên các đơn vị hành chính trong phạm vi lập quy hoạch; tên đơn vị hành chính tiếp giáp bên ngoài phạm vi lập quy hoạch.', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 1024] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Information Retrieval * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.9121 \| \| cosine_accuracy@3 \| 0.9832 \| \| cosine_accuracy@5 \| 0.9908 \| \| cosine_accuracy@10 \| 0.9958 \| \| cosine_precision@1 \| 0.9121 \| \| cosine_precision@3 \| 0.3277 \| \| cosine_precision@5 \| 0.1982 \| \| cosine_precision@10 \| 0.0996 \| \| cosine_recall@1 \| 0.9121 \| \| cosine_recall@3 \| 0.9832 \| \| cosine_recall@5 \| 0.9908 \| \| cosine_recall@10 \| 0.9958 \| \| cosine_ndcg@10 \| 0.9596 \| \| cosine_mrr@10 \| 0.9474 \| \| cosine_map@100 \| 0.9476 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 118,262 training samples * Columns: <code>sentence_0</code> and <code>sentence_1</code> * Approximate statistics based on the first 1000 samples: \| \| sentence_0 \| sentence_1 \| \|:--------\|:----------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\| \| type \| string \| string \| \| details \| <ul><li>min: 8 tokens</li><li>mean: 24.18 tokens</li><li>max: 49 tokens</li></ul> \| <ul><li>min: 13 tokens</li><li>mean: 254.3 tokens</li><li>max: 644 tokens</li></ul> \| * Samples: \| sentence_0 \| sentence_1 \| \|:--------------------------------------------------------------------------------------------------------------------------\|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>Báo cáo lưu chuyển tiền tệ hợp nhất cung cấp những thông tin gì cho người sử dụng?</code> \| <code>Cơ sở lập Báo cáo lưu chuyển tiền tệ hợp nhất<br>1. Báo cáo lưu chuyển tiền tệ hợp nhất là một bộ phận không thể tách rời trong hệ thống Báo cáo tài chính hợp nhất, cung cấp thông tin hữu ích cho người sử dụng Báo cáo tài chính về khả năng tạo tiền, khả năng thanh toán của toàn bộ Tập đoàn.<br>...</code> \| \| <code>Người đứng đầu cơ sở khám bệnh chữa bệnh triển khai hoạt động dược lâm sàng theo những nội dung nào?</code> \| <code>Điều 80. Nội dung hoạt động dược lâm sàng<br>1. Tư vấn trong quá trình xây dựng danh Mục thuốc tại cơ sở khám bệnh, chữa bệnh để bảo đảm Mục tiêu sử dụng thuốc hợp lý, an toàn, hiệu quả.<br>2. Tư vấn và giám sát việc kê đơn và sử dụng thuốc.<br>3. Thông tin, hướng dẫn sử dụng thuốc cho người hành nghề khám bệnh, chữa bệnh, người sử dụng thuốc và cộng đồng.<br>4. Tham gia xây dựng quy trình, hướng dẫn chuyên môn liên quan đến sử dụng thuốc và giám sát việc thực hiện các quy trình này.<br>5. Phân tích, đánh giá hiệu quả sử dụng thuốc tại cơ sở khám bệnh, chữa bệnh.<br>6. Tham gia theo dõi, giám sát phản ứng có hại của thuốc.<br>7. Tham gia hoạt động nghiên cứu khoa học liên quan đến sử dụng thuốc hợp lý, an toàn, hiệu quả.<br>Điều 81. Triển khai hoạt động dược lâm sàng<br>1. Người đứng đầu cơ sở khám bệnh, chữa bệnh có hoạt động sử dụng thuốc phải tổ chức và triển khai hoạt động dược lâm sàng theo nội dung quy định tại Điều 80 của Luật này.<br>2. Người chịu trách nhiệ...</code> \| \| <code>Công ty cổ phần bảo hiểm có trách nhiệm gì khi kinh doanh tái bảo hiểm?</code> \| <code>Quản lý chương trình tái bảo hiểm<br>...<br>3. Doanh nghiệp bảo hiểm, chi nhánh nước ngoài có trách nhiệm cập nhật thường xuyên danh sách các doanh nghiệp nhận tái bảo hiểm, kèm theo các thông tin về mức độ rủi ro, khả năng, mức độ sẵn sàng chi trả bồi thường tương ứng với trách nhiệm đã nhận tái bảo hiểm; yêu cầu đặt cọc tương ứng với mức độ rủi ro và hệ số tín nhiệm của từng doanh nghiệp nhận tái bảo hiểm (nếu có).</code> \| * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters: ```json { "scale": 20.0, "similarity_fct": "cos_sim" } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 10 - `per_device_eval_batch_size`: 10 - `num_train_epochs`: 1 - `multi_dataset_batch_sampler`: round_robin #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: steps - `prediction_loss_only`: True - `per_device_train_batch_size`: 10 - `per_device_eval_batch_size`: 10 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 1 - `eval_accumulation_steps`: None - `torch_empty_cache_steps`: None - `learning_rate`: 5e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1 - `num_train_epochs`: 1 - `max_steps`: -1 - `lr_scheduler_type`: linear - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.0 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: False - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: None - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: False - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `eval_use_gather_object`: False - `prompts`: None - `batch_sampler`: batch_sampler - `multi_dataset_batch_sampler`: round_robin </details> ### Training Logs \| Epoch \| Step \| Training Loss \| cosine_ndcg@10 \| \|:------:\|:-----:\|:-------------:\|:--------------:\| \| 0.0423 \| 500 \| 0.0229 \| - \| \| 0.0846 \| 1000 \| 0.0104 \| 0.9464 \| \| 0.1268 \| 1500 \| 0.0153 \| - \| \| 0.1691 \| 2000 \| 0.0169 \| 0.9437 \| \| 0.2114 \| 2500 \| 0.0149 \| - \| \| 0.2537 \| 3000 \| 0.0129 \| 0.9506 \| \| 0.2959 \| 3500 \| 0.0134 \| - \| \| 0.3382 \| 4000 \| 0.0134 \| 0.9455 \| \| 0.3805 \| 4500 \| 0.0155 \| - \| \| 0.4228 \| 5000 \| 0.0078 \| 0.9518 \| \| 0.4650 \| 5500 \| 0.0105 \| - \| \| 0.5073 \| 6000 \| 0.0105 \| 0.9475 \| \| 0.5496 \| 6500 \| 0.0118 \| - \| \| 0.5919 \| 7000 \| 0.0088 \| 0.9505 \| \| 0.6341 \| 7500 \| 0.0082 \| - \| \| 0.6764 \| 8000 \| 0.0064 \| 0.9567 \| \| 0.7187 \| 8500 \| 0.0064 \| - \| \| 0.7610 \| 9000 \| 0.0053 \| 0.9591 \| \| 0.8032 \| 9500 \| 0.0091 \| - \| \| 0.8455 \| 10000 \| 0.0068 \| 0.9576 \| \| 0.8878 \| 10500 \| 0.0075 \| - \| \| 0.9301 \| 11000 \| 0.0071 \| 0.9583 \| \| 0.9724 \| 11500 \| 0.0061 \| - \| \| 1.0 \| 11827 \| - \| 0.9596 \| ### Framework Versions - Python: 3.10.10 - Sentence Transformers: 3.3.1 - Transformers: 4.44.2 - PyTorch: 2.2.1+cu121 - Accelerate: 1.1.1 - Datasets: 2.19.0 - Tokenizers: 0.19.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	null	Non_BioNLP	# SentenceTransformer based on BAAI/bge-m3 This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-m3](https://huggingface.co/BAAI/bge-m3). It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ### Model Description - Model Type: Sentence Transformer - Base model: [BAAI/bge-m3](https://huggingface.co/BAAI/bge-m3) <!-- at revision 5617a9f61b028005a4858fdac845db406aefb181 --> - Maximum Sequence Length: 8192 tokens - Output Dimensionality: 1024 dimensions - Similarity Function: Cosine Similarity <!-- - Training Dataset: Unknown --> <!-- - Language: Unknown --> <!-- - License: Unknown --> ### Model Sources - Documentation: [Sentence Transformers Documentation](https://sbert.net) - Repository: [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - Hugging Face: [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("sentence_transformers_model_id") # Run inference sentences = [ 'Lấn chiếm đất công là đất phi nông nghiệp để xây dựng công trình thì sẽ bị xử phạt vi phạm hành chính như thế nào?', 'Lấn, chiếm đất\n...\n4. Trường hợp lấn, chiếm đất phi nông nghiệp, trừ trường hợp quy định tại khoản 6 Điều này tại khu vực nông thôn thì hình thức và mức xử phạt như sau:\na) Phạt tiền từ 10.000.000 đồng đến 20.000.000 đồng nếu diện tích đất lấn, chiếm dưới 0,05 héc ta;\nb) Phạt tiền từ 20.000.000 đồng đến 40.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,05 héc ta đến dưới 0,1 héc ta;\nc) Phạt tiền từ 40.000.000 đồng đến 100.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,1 héc ta đến dưới 0,5 héc ta;\nd) Phạt tiền từ 100.000.000 đồng đến 200.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,5 héc ta đến dưới 01 héc ta;\nđ) Phạt tiền từ 200.000.000 đồng đến 500.000.000 đồng nếu diện tích đất lấn, chiếm từ 01 héc ta trở lên.\n5. Trường hợp lấn, chiếm đất chưa sử dụng, đất nông nghiệp, đất phi nông nghiệp (trừ trường hợp quy định tại khoản 6 Điều này) tại khu vực đô thị thì mức xử phạt bằng 02 lần mức xử phạt đối với loại đất tương ứng quy định tại các khoản 1, 2, 3 và 4 Điều này và mức phạt tối đa không quá 500.000.000 đồng đối với cá nhân, không quá 1.000.000.000 đồng đối với tổ chức.\n6. Trường hợp lấn, chiếm đất thuộc hành lang bảo vệ an toàn công trình và đất công trình có hành lang bảo vệ, đất trụ sở làm việc và cơ sở hoạt động sự nghiệp của cơ quan, tổ chức theo quy định của pháp luật về quản lý, sử dụng tài sản nhà nước thì hình thức và mức xử phạt thực hiện theo quy định của pháp luật về xử phạt vi phạm hành chính trong lĩnh vực về hoạt động đầu tư xây dựng; khai thác, chế biến, kinh doanh khoáng sản làm vật liệu xây dựng, sản xuất, kinh doanh vật liệu xây dựng; quản lý công trình hạ tầng kỹ thuật; kinh doanh bất động sản, phát triển nhà ở, quản lý sử dụng nhà và công sở; trong lĩnh vực về giao thông đường bộ và đường sắt; trong lĩnh vực về văn hóa, thể thao, du lịch và quảng cáo; trong lĩnh vực về khai thác và bảo vệ công trình thủy lợi; đê điều; phòng, chống lụt, bão; trong lĩnh vực quản lý, sử dụng tài sản nhà nước và các lĩnh vực chuyên ngành khác.\n...Áp dụng mức phạt tiền\n1. Đối tượng áp dụng mức phạt tiền thực ', 'Quy định chung về quy cách thể hiện\n1. Nội dung thể hiện đối với thuyết minh, bản vẽ của nhiệm vụ và đồ án trong hồ sơ lấy ý kiến, hồ sơ trình thẩm định và phê duyệt phải được trình bày rõ ràng, đầy đủ nội dung đối với từng loại, cấp độ quy hoạch và tuân thủ các quy định tại Chương II và các phụ lục của Thông tư này.\n2. Hệ thống ký hiệu bản vẽ của nhiệm vụ và đồ án quy hoạch đô thị, nông thôn phải tuân thủ theo quy định tại các phụ lục kèm theo Thông tư này.\n3. Căn cứ điều kiện, đặc điểm của vùng liên huyện, vùng huyện, đô thị, khu chức năng và khu vực nông thôn được lập quy hoạch, có thể thành lập các bản đồ, sơ đồ riêng cho từng nội dung hiện trạng, định hướng hạ tầng kỹ thuật để đảm bảo thể hiện được đầy đủ các nội dung theo yêu cầu.\n4. Trường hợp quy hoạch vùng, quy hoạch tỉnh chưa được phê duyệt, các bản đồ phương hướng phát triển hệ thống đô thị, nông thôn; bản đồ phương án quy hoạch hệ thống đô thị, nông thôn; bản đồ quy hoạch sử dụng đất cấp tỉnh; bản đồ tổ chức không gian và phân vùng chức năng trong quy hoạch vùng, quy hoạch tỉnh được thay thế bằng sơ đồ định hướng phát triển không gian trong quy hoạch xây dựng vùng tỉnh đã được phê duyệt.\n5. Tất cả các sơ đồ, bản đồ trong thành phần bản vẽ của hồ sơ nhiệm vụ và hồ sơ đồ án quy hoạch đô thị và nông thôn phải thể hiện rõ phạm vi ranh giới lập quy hoạch; ranh giới, tên các đơn vị hành chính trong phạm vi lập quy hoạch; tên đơn vị hành chính tiếp giáp bên ngoài phạm vi lập quy hoạch.', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 1024] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` <!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use List how the model may foreseeably be misused and address what users ought not to do with the model. --> ## Evaluation ### Metrics #### Information Retrieval * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) \| Metric \| Value \| \|:--------------------\|:-----------\| \| cosine_accuracy@1 \| 0.9121 \| \| cosine_accuracy@3 \| 0.9832 \| \| cosine_accuracy@5 \| 0.9908 \| \| cosine_accuracy@10 \| 0.9958 \| \| cosine_precision@1 \| 0.9121 \| \| cosine_precision@3 \| 0.3277 \| \| cosine_precision@5 \| 0.1982 \| \| cosine_precision@10 \| 0.0996 \| \| cosine_recall@1 \| 0.9121 \| \| cosine_recall@3 \| 0.9832 \| \| cosine_recall@5 \| 0.9908 \| \| cosine_recall@10 \| 0.9958 \| \| cosine_ndcg@10 \| 0.9596 \| \| cosine_mrr@10 \| 0.9474 \| \| cosine_map@100 \| 0.9476 \| <!-- ## Bias, Risks and Limitations What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model. --> <!-- ### Recommendations What are recommendations with respect to the foreseeable issues? For example, filtering explicit content. --> ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 118,262 training samples * Columns: <code>sentence_0</code> and <code>sentence_1</code> * Approximate statistics based on the first 1000 samples: \| \| sentence_0 \| sentence_1 \| \|:--------\|:----------------------------------------------------------------------------------\|:------------------------------------------------------------------------------------\| \| type \| string \| string \| \| details \| <ul><li>min: 8 tokens</li><li>mean: 24.18 tokens</li><li>max: 49 tokens</li></ul> \| <ul><li>min: 13 tokens</li><li>mean: 254.3 tokens</li><li>max: 644 tokens</li></ul> \| * Samples: \| sentence_0 \| sentence_1 \| \|:--------------------------------------------------------------------------------------------------------------------------\|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\| \| <code>Báo cáo lưu chuyển tiền tệ hợp nhất cung cấp những thông tin gì cho người sử dụng?</code> \| <code>Cơ sở lập Báo cáo lưu chuyển tiền tệ hợp nhất<br>1. Báo cáo lưu chuyển tiền tệ hợp nhất là một bộ phận không thể tách rời trong hệ thống Báo cáo tài chính hợp nhất, cung cấp thông tin hữu ích cho người sử dụng Báo cáo tài chính về khả năng tạo tiền, khả năng thanh toán của toàn bộ Tập đoàn.<br>...</code> \| \| <code>Người đứng đầu cơ sở khám bệnh chữa bệnh triển khai hoạt động dược lâm sàng theo những nội dung nào?</code> \| <code>Điều 80. Nội dung hoạt động dược lâm sàng<br>1. Tư vấn trong quá trình xây dựng danh Mục thuốc tại cơ sở khám bệnh, chữa bệnh để bảo đảm Mục tiêu sử dụng thuốc hợp lý, an toàn, hiệu quả.<br>2. Tư vấn và giám sát việc kê đơn và sử dụng thuốc.<br>3. Thông tin, hướng dẫn sử dụng thuốc cho người hành nghề khám bệnh, chữa bệnh, người sử dụng thuốc và cộng đồng.<br>4. Tham gia xây dựng quy trình, hướng dẫn chuyên môn liên quan đến sử dụng thuốc và giám sát việc thực hiện các quy trình này.<br>5. Phân tích, đánh giá hiệu quả sử dụng thuốc tại cơ sở khám bệnh, chữa bệnh.<br>6. Tham gia theo dõi, giám sát phản ứng có hại của thuốc.<br>7. Tham gia hoạt động nghiên cứu khoa học liên quan đến sử dụng thuốc hợp lý, an toàn, hiệu quả.<br>Điều 81. Triển khai hoạt động dược lâm sàng<br>1. Người đứng đầu cơ sở khám bệnh, chữa bệnh có hoạt động sử dụng thuốc phải tổ chức và triển khai hoạt động dược lâm sàng theo nội dung quy định tại Điều 80 của Luật này.<br>2. Người chịu trách nhiệ...</code> \| \| <code>Công ty cổ phần bảo hiểm có trách nhiệm gì khi kinh doanh tái bảo hiểm?</code> \| <code>Quản lý chương trình tái bảo hiểm<br>...<br>3. Doanh nghiệp bảo hiểm, chi nhánh nước ngoài có trách nhiệm cập nhật thường xuyên danh sách các doanh nghiệp nhận tái bảo hiểm, kèm theo các thông tin về mức độ rủi ro, khả năng, mức độ sẵn sàng chi trả bồi thường tương ứng với trách nhiệm đã nhận tái bảo hiểm; yêu cầu đặt cọc tương ứng với mức độ rủi ro và hệ số tín nhiệm của từng doanh nghiệp nhận tái bảo hiểm (nếu có).</code> \| * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters: ```json { "scale": 20.0, "similarity_fct": "cos_sim" } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 10 - `per_device_eval_batch_size`: 10 - `num_train_epochs`: 1 - `multi_dataset_batch_sampler`: round_robin #### All Hyperparameters <details><summary>Click to expand</summary> - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: steps - `prediction_loss_only`: True - `per_device_train_batch_size`: 10 - `per_device_eval_batch_size`: 10 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 1 - `eval_accumulation_steps`: None - `torch_empty_cache_steps`: None - `learning_rate`: 5e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1 - `num_train_epochs`: 1 - `max_steps`: -1 - `lr_scheduler_type`: linear - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.0 - `warmup_steps`: 0 - `log_level`: passive - `log_level_replica`: warning - `log_on_each_node`: True - `logging_nan_inf_filter`: True - `save_safetensors`: True - `save_on_each_node`: False - `save_only_model`: False - `restore_callback_states_from_checkpoint`: False - `no_cuda`: False - `use_cpu`: False - `use_mps_device`: False - `seed`: 42 - `data_seed`: None - `jit_mode_eval`: False - `use_ipex`: False - `bf16`: False - `fp16`: False - `fp16_opt_level`: O1 - `half_precision_backend`: auto - `bf16_full_eval`: False - `fp16_full_eval`: False - `tf32`: None - `local_rank`: 0 - `ddp_backend`: None - `tpu_num_cores`: None - `tpu_metrics_debug`: False - `debug`: [] - `dataloader_drop_last`: False - `dataloader_num_workers`: 0 - `dataloader_prefetch_factor`: None - `past_index`: -1 - `disable_tqdm`: False - `remove_unused_columns`: True - `label_names`: None - `load_best_model_at_end`: False - `ignore_data_skip`: False - `fsdp`: [] - `fsdp_min_num_params`: 0 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} - `fsdp_transformer_layer_cls_to_wrap`: None - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch - `optim_args`: None - `adafactor`: False - `group_by_length`: False - `length_column_name`: length - `ddp_find_unused_parameters`: None - `ddp_bucket_cap_mb`: None - `ddp_broadcast_buffers`: False - `dataloader_pin_memory`: True - `dataloader_persistent_workers`: False - `skip_memory_metrics`: True - `use_legacy_prediction_loop`: False - `push_to_hub`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `eval_do_concat_batches`: True - `fp16_backend`: auto - `push_to_hub_model_id`: None - `push_to_hub_organization`: None - `mp_parameters`: - `auto_find_batch_size`: False - `full_determinism`: False - `torchdynamo`: None - `ray_scope`: last - `ddp_timeout`: 1800 - `torch_compile`: False - `torch_compile_backend`: None - `torch_compile_mode`: None - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `eval_on_start`: False - `eval_use_gather_object`: False - `prompts`: None - `batch_sampler`: batch_sampler - `multi_dataset_batch_sampler`: round_robin </details> ### Training Logs \| Epoch \| Step \| Training Loss \| cosine_ndcg@10 \| \|:------:\|:-----:\|:-------------:\|:--------------:\| \| 0.0423 \| 500 \| 0.0229 \| - \| \| 0.0846 \| 1000 \| 0.0104 \| 0.9464 \| \| 0.1268 \| 1500 \| 0.0153 \| - \| \| 0.1691 \| 2000 \| 0.0169 \| 0.9437 \| \| 0.2114 \| 2500 \| 0.0149 \| - \| \| 0.2537 \| 3000 \| 0.0129 \| 0.9506 \| \| 0.2959 \| 3500 \| 0.0134 \| - \| \| 0.3382 \| 4000 \| 0.0134 \| 0.9455 \| \| 0.3805 \| 4500 \| 0.0155 \| - \| \| 0.4228 \| 5000 \| 0.0078 \| 0.9518 \| \| 0.4650 \| 5500 \| 0.0105 \| - \| \| 0.5073 \| 6000 \| 0.0105 \| 0.9475 \| \| 0.5496 \| 6500 \| 0.0118 \| - \| \| 0.5919 \| 7000 \| 0.0088 \| 0.9505 \| \| 0.6341 \| 7500 \| 0.0082 \| - \| \| 0.6764 \| 8000 \| 0.0064 \| 0.9567 \| \| 0.7187 \| 8500 \| 0.0064 \| - \| \| 0.7610 \| 9000 \| 0.0053 \| 0.9591 \| \| 0.8032 \| 9500 \| 0.0091 \| - \| \| 0.8455 \| 10000 \| 0.0068 \| 0.9576 \| \| 0.8878 \| 10500 \| 0.0075 \| - \| \| 0.9301 \| 11000 \| 0.0071 \| 0.9583 \| \| 0.9724 \| 11500 \| 0.0061 \| - \| \| 1.0 \| 11827 \| - \| 0.9596 \| ### Framework Versions - Python: 3.10.10 - Sentence Transformers: 3.3.1 - Transformers: 4.44.2 - PyTorch: 2.2.1+cu121 - Accelerate: 1.1.1 - Datasets: 2.19.0 - Tokenizers: 0.19.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @inproceedings{reimers-2019-sentence-bert, title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks", author = "Reimers, Nils and Gurevych, Iryna", booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing", month = "11", year = "2019", publisher = "Association for Computational Linguistics", url = "https://arxiv.org/abs/1908.10084", } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` <!-- ## Glossary Clearly define terms in order to be accessible across audiences. --> <!-- ## Model Card Authors Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction. --> <!-- ## Model Card Contact Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors. -->	{"base_model": "BAAI/bge-m3", "library_name": "sentence-transformers", "metrics": ["cosine_accuracy@1", "cosine_accuracy@3", "cosine_accuracy@5", "cosine_accuracy@10", "cosine_precision@1", "cosine_precision@3", "cosine_precision@5", "cosine_precision@10", "cosine_recall@1", "cosine_recall@3", "cosine_recall@5", "cosine_recall@10", "cosine_ndcg@10", "cosine_mrr@10", "cosine_map@100"], "pipeline_tag": "sentence-similarity", "tags": ["sentence-transformers", "sentence-similarity", "feature-extraction", "generated_from_trainer", "dataset_size:118262", "loss:MultipleNegativesRankingLoss"], "widget": [{"source_sentence": "Ai ra quyết định thành lập Tủ sách pháp luật tại xã đặc biệt khó khăn?", "sentences": ["Hội đồng dân tộc, các Ủy ban của Quốc hội\n1. Hội đồng dân tộc, Ủy ban của Quốc hội là cơ quan của Quốc hội, chịu trách nhiệm và báo cáo công tác trước Quốc hội; trong thời gian Quốc hội không họp thì báo cáo công tác trước Ủy ban thường vụ Quốc hội.\n2. Các Ủy ban của Quốc hội gồm:\na) Ủy ban pháp luật;\nb) Ủy ban tư pháp;\nc) Ủy ban kinh tế;\nd) Ủy ban tài chính, ngân sách;\nđ) Ủy ban quốc phòng và an ninh;\ne) Ủy ban văn hoá, giáo dục;\ng) Ủy ban xã hội;\nh) Ủy ban khoa học, công nghệ và môi trường;\ni) Ủy ban đối ngoại.\n3. Quốc hội thành lập Ủy ban lâm thời của Quốc hội theo quy định tại Điều 88 và Điều 89 của Luật này.", "Xây dựng Tủ sách pháp luật\n1. Thành lập Tủ sách pháp luật\na) Chủ tịch Ủy ban nhân dân xã đặc biệt khó khăn, Thủ trưởng cơ quan, đơn vị của lực lượng vũ trang nhân dân ra quyết định thành lập Tủ sách pháp luật. Quyết định thành lập phải quy định cụ thể về địa điểm đặt Tủ sách pháp luật; người phụ trách và quyền hạn, nhiệm vụ của người phụ trách; phạm vi và đối tượng phục vụ; Nội quy hoạt động và kinh phí hoạt động của Tủ sách pháp luật.\nb) Chủ tịch Ủy ban nhân dân cấp tỉnh chỉ đạo việc quản lý, khai thác thống nhất sách, tài liệu của Tủ sách pháp luật và sách, tài liệu pháp luật của Thư viện xã, điểm Bưu điện - Văn hóa xã, Trung tâm học tập cộng đồng phù hợp với thực tiễn tại địa phương. Trên cơ sở ý kiến của Thủ tướng Chính phủ và Ban Bí thư Trung ương Đảng quy định tại điểm d khoản 1 Điều 14 của Quyết định này, Chủ tịch Ủy ban nhân dân cấp tỉnh thực hiện quản lý sách, tài liệu của Tủ sách pháp luật và sách, tài liệu pháp luật của Đề án trang bị sách cho cơ sở xã, phường, thị trấn.", "Nhiệm vụ, quyền hạn của trường trung cấp\nTrường trung cấp thực hiện nhiệm vụ, quyền hạn theo quy định tại Điều 23 của Luật Giáo dục nghề nghiệp, quy định của pháp luật có liên quan và các quy định sau đây:\n...\n2. Quyền hạn:\na) Xây dựng và tổ chức thực hiện kế hoạch, chiến lược phát triển trường phù hợp với chiến lược phát triển giáo dục nghề nghiệp nhằm đáp ứng yêu cầu của thị trường lao động;\nb) Tổ chức đào tạo các chương trình giáo dục nghề nghiệp theo quy định của Bộ trưởng Bộ Lao động - Thương binh và Xã hội;\nc) Liên kết với cơ sở giáo dục nghề nghiệp, cơ sở giáo dục đại học, doanh nghiệp, tổ chức trong nước và nước ngoài theo quy định của pháp luật nhằm khai thác, huy động các nguồn lực xã hội trong thực hiện các chương trình đào tạo giáo dục nghề nghiệp để nâng cao chất lượng đào tạo, gắn đào tạo với việc làm và thị trường lao động. Đơn vị chủ trì liên kết đào tạo chịu trách nhiệm cấp văn bằng, chứng chỉ cho người học;\nd) Liên kết, phối hợp với trường cao đẳng tổ chức đào tạo liên thông từ trình độ sơ cấp, trình độ trung cấp lên trình độ cao đẳng theo quy định của Luật Giáo dục nghề nghiệp và các văn bản hướng dẫn thi hành;\nđ) Liên kết, phối hợp với cơ sở giáo dục đại học tổ chức đào tạo trình độ đại học theo hình thức vừa làm vừa học và đào tạo liên thông từ trình độ trung cấp lên trình độ đại học theo quy định của Luật Giáo dục đại học và các văn bản hướng dẫn thi hành;\ne) Phối hợp với doanh nghiệp, cơ sở sản xuất, kinh doanh, dịch vụ trong hoạt động xây dựng chương trình, giáo trình đào tạo, tài liệu, học liệu giảng dạy; tổ chức giảng dạy, hướng dẫn thực hành, thực tập, đánh giá kết quả học tập của người học; bồi dưỡng nâng cao trình độ chuyên môn, nghiệp vụ cho nhà giáo;\ng) Tổ chức giảng dạy khối lượng kiến thức văn hóa trung học phổ thông cho học sinh có bằng tốt nghiệp trung học cơ sở theo học trình độ trung cấp theo quy định của Luật Giáo dục và các văn bản hướng dẫn thi hành;:\n..."]}, {"source_sentence": "Việc tinh giản biên chế hiện nay được thực hiện theo những nguyên tắc nào?", "sentences": ["Xử phạt hành vi vi phạm trong việc tổ chức bồi dưỡng và cấp Chứng chỉ bồi dưỡng kế toán trưởng\n...\n4. Phạt tiền từ 10.000.000 đồng đến 20.000.000 đồng đối với cơ sở đào tạo thực hiện một trong các hành vi sau:\na) Mở khóa học bồi dưỡng kế toán trưởng khi không đủ điều kiện;\nb) Mở khóa học bồi dưỡng kế toán trưởng cho người nước ngoài khi chưa được Bộ Tài chính chấp thuận;\nc) Quản lý phôi và cấp Chứng chỉ bồi dưỡng kế toán trưởng của cơ sở đào tạo không đúng quy định của Bộ Tài chính.\n...\n6. Biện pháp khắc phục hậu quả:\nBuộc nộp lại số lợi bất hợp pháp có được do thực hiện hành vi vi phạm quy định tại điểm a, b khoản 4, khoản 5 Điều này.Mức phạt tiền trong lĩnh vực kế toán, kiểm toán độc lập\n...\n2. Mức phạt tiền quy định tại Chương II, Chương III Nghị định này là mức phạt tiền áp dụng đối với tổ chức trừ trường hợp quy định tại khoản 1 Điều 7; Điều 8; Điều 9; Điều 10; Điều 11; Điều 13; Điều 14; Điều 15; Điều 16; Điều 17; Điều 19; khoản 1, khoản 3 Điều 21; Điều 22; Điều 23; Điều 24; Điều 26; Điều 33; Điều 34; khoản 1, khoản 3 Điều 36; khoản 1 Điều 38; khoản 2, khoản 3 Điều 39; khoản 1, khoản 2 Điều 48; khoản 1 Điều 57; khoản 1, khoản 2 Điều 61; Điều 67 là mức phạt tiền đối với cá nhân. Đối với tổ chức có cùng hành vi vi phạm thì mức phạt tiền gấp 02 lần mức phạt tiền đối với cá nhân.\n...", "Nguyên tắc tinh giản biên chế\n1. Bảo đảm sự lãnh đạo của Đảng, phát huy vai trò giám sát của các tổ chức chính trị - xã hội và nhân dân trong quá trình thực hiện tinh giản biên chế.\n2. Gắn tinh giản biên chế với sắp xếp tổ chức bộ máy theo hướng tinh gọn, hoạt động hiệu lực, hiệu quả và cơ cấu lại, nâng cao chất lượng đội ngũ cán bộ, công chức, viên chức theo vị trí việc làm, phù hợp với cơ chế tự chủ của cơ quan, tổ chức, đơn vị.\n3. Bảo đảm nguyên tắc tập trung dân chủ, khách quan, công bằng, công khai, minh bạch và theo quy định của pháp luật.\n4. Bảo đảm chi trả chế độ, chính sách tinh giản biên chế kịp thời, đầy đủ theo quy định của pháp luật, bảo đảm sử dụng hiệu quả ngân sách nhà nước.\n5. Người đứng đầu phải chịu trách nhiệm về kết quả thực hiện tinh giản biên chế trong cơ quan, tổ chức, đơn vị được giao quản lý theo thẩm quyền.\n6. Đối tượng tinh giản biên chế nếu được bầu cử, tuyển dụng lại vào các cơ quan, tổ chức, đơn vị hưởng lương từ ngân sách nhà nước trong thời gian 60 tháng kể từ ngày thực hiện tinh giản biên chế thì phải hoàn trả lại số tiền trợ cấp đã nhận cho cơ quan, tổ chức, đơn vị đã chi trả trợ cấp.", "\"3. Tiêu chuẩn về trình độ đào tạo, bồi dưỡng\na) Có bằng cử nhân trở lên thuộc ngành đào tạo giáo viên đối với giáo viên trung học cơ sở.\nTrường hợp môn học chưa đủ giáo viên có bằng cử nhân thuộc ngành đào tạo giáo viên thì phải có bằng cử nhân chuyên ngành phù hợp và có chứng chỉ bồi dưỡng nghiệp vụ sư phạm dành cho giáo viên trung học cơ sở theo chương trình do Bộ trưởng Bộ Giáo dục và Đào tạo ban hành;\nb) Có chứng chỉ bồi dưỡng theo tiêu chuẩn chức danh nghề nghiệp giáo viên trung học cơ sở hạng III (đối với giáo viên trung học cơ sở mới được tuyển dụng vào giáo viên trung học cơ sở hạng III thì phải có chứng chỉ trong thời gian 36 tháng kể từ ngày được tuyển dụng).\""]}, {"source_sentence": "Khi xuất bản phụ trương thì cần đáp ứng những điều kiện gì?", "sentences": ["Hồ sơ, trình tự, thủ tục chấp thuận chủ trương đầu tư ra nước ngoài của Quốc hội\n1. Nhà đầu tư nộp hồ sơ dự án đầu tư ra nước ngoài cho Bộ Kế hoạch và Đầu tư. Hồ sơ bao gồm:\na) Văn bản đăng ký đầu tư ra nước ngoài;\nb) Tài liệu về tư cách pháp lý của nhà đầu tư;\nc) Đề xuất dự án đầu tư gồm các nội dung chủ yếu sau: hình thức, mục tiêu, quy mô, địa điểm đầu tư; xác định sơ bộ vốn đầu tư, phương án huy động vốn, cơ cấu nguồn vốn; tiến độ thực hiện dự án, các giai đoạn đầu tư (nếu có); phân tích sơ bộ hiệu quả đầu tư của dự án;\nd) Tài liệu chứng minh năng lực tài chính của nhà đầu tư gồm ít nhất một trong các tài liệu sau: báo cáo tài chính 02 năm gần nhất của nhà đầu tư; cam kết hỗ trợ tài chính của công ty mẹ; cam kết hỗ trợ tài chính của tổ chức tài chính; bảo lãnh về năng lực tài chính của nhà đầu tư; tài liệu khác chứng minh năng lực tài chính của nhà đầu tư;\nđ) Cam kết tự cân đối nguồn ngoại tệ hoặc văn bản cam kết thu xếp ngoại tệ cho nhà đầu tư của tổ chức tín dụng được phép;\ne) Văn bản của cơ quan đại diện chủ sở hữu chấp thuận nhà đầu tư thực hiện hoạt động đầu tư ra nước ngoài và báo cáo thẩm định nội bộ về đề xuất đầu tư ra nước ngoài của doanh nghiệp nhà nước quy định tại khoản 1 Điều 59 của Luật này hoặc quyết định đầu tư ra nước ngoài theo quy định tại khoản 2 Điều 59 của Luật này;\ng) Đối với dự án đầu tư ra nước ngoài trong các ngành, nghề quy định tại khoản 1 Điều 54 của Luật này, nhà đầu tư nộp văn bản của cơ quan nhà nước có thẩm quyền về việc đáp ứng điều kiện đầu tư ra nước ngoài theo quy định của pháp luật có liên quan (nếu có).\n2. Trong thời hạn 05 ngày làm việc kể từ ngày nhận đủ hồ sơ, Bộ Kế hoạch và Đầu tư trình Thủ tướng Chính phủ quyết định thành lập Hội đồng thẩm định nhà nước.\n3. Trong thời hạn 90 ngày kể từ ngày thành lập, Hội đồng thẩm định nhà nước tổ chức thẩm định và lập báo cáo thẩm định trình Chính phủ. Báo cáo thẩm định gồm các nội dung sau đây:\na) Điều kiện cấp Giấy chứng nhận đăng ký đầu tư ra nước ngoài quy định tại Điều 60 của Luật này;\nb) Tư cách pháp lý của nhà đầu tư;\nc) S", "Vi phạm quy định về tổ chức, quản trị, điều hành\n1. Phạt tiền từ 20.000.000 đồng đến 30.000.000 đồng đối với một trong các hành vi vi phạm sau đây:\na) Vi phạm quy định về Hội đồng quản trị, Hội đồng thành viên, Ban kiểm soát quy định tại Điều 43, Điều 44, Điều 62, khoản 1 Điều 70, Điều 81, Điều 84 Luật Các tổ chức tín dụng;\nb) Không duy trì việc đảm bảo đầy đủ các điều kiện, tiêu chuẩn đối với thành viên Hội đồng quản trị, Ban kiểm soát, Tổng giám đốc (Giám đốc) theo quy định của Luật Các tổ chức tín dụng;\nc) Không tổ chức hoặc tổ chức Đại hội đồng cổ đông, Đại hội thành viên không đúng quy định.\n...Hình thức xử phạt, mức phạt tiền, thẩm quyền phạt tiền và biện pháp khắc phục hậu quả\n...\n3. Mức phạt tiền và thẩm quyền phạt tiền:\n...\nb) Mức phạt tiền quy định tại Chương II Nghị định này là mức phạt tiền áp dụng đối với cá nhân; mức phạt tiền đối với tổ chức có cùng một hành vi vi phạm hành chính bằng 02 lần mức phạt tiền đối với cá nhân;\n...", "Điều kiện xuất bản thêm ấn phẩm báo chí, phụ trương; sản xuất thêm kênh phát thanh, kênh truyền hình; mở chuyên trang của báo điện tử\n1. Có tôn chỉ, Mục đích phù hợp với tôn chỉ, Mục đích của cơ quan báo chí; đối tượng phục vụ; nơi đặt máy chủ và đơn vị cung cấp dịch vụ kết nối (đối với chuyên trang của báo điện tử) phù hợp với nội dung giấy phép hoạt động của cơ quan báo chí.\n2. Có phương án về tổ chức và nhân sự bảo đảm hoạt động của ấn phẩm báo chí, phụ trương, kênh phát thanh, kênh truyền hình, chuyên trang của báo điện tử.\n3. Có tên và hình thức trình bày tên ấn phẩm báo chí, phụ trương; tên và biểu tượng kênh phát thanh, kênh truyền hình; tên và giao diện trang chủ chuyên trang của báo điện tử.\n4. Có các Điều kiện cơ sở vật chất, kỹ thuật; phương án tài chính; có các giải pháp kỹ thuật bảo đảm an toàn, an ninh thông tin; có tên miền cấp dưới phù hợp với tên miền đã đăng ký đối với chuyên trang của báo điện tử; có phương án, kế hoạch thuê hoặc sử dụng hạ tầng truyền dẫn, phát sóng đối với kênh phát thanh, kênh truyền hình."]}, {"source_sentence": "Người dân có bị xóa đăng ký thường trú trong trường hợp lên thành phố khác sinh sống hơn 15 năm không về lại nơi thường trú ban đầu không?", "sentences": ["Hồ sơ đề nghị hỗ trợ học nghề\n1. Hồ sơ đề nghị hỗ trợ học nghề đối với người lao động đang chờ kết quả giải quyết hưởng trợ cấp thất nghiệp hoặc đang hưởng trợ cấp thất nghiệp mà có nhu cầu học nghề tại địa phương nơi đang chờ kết quả hoặc đang hưởng trợ cấp thất nghiệp là đề nghị hỗ trợ học nghề theo Mẫu số 03 ban hành kèm theo Nghị định này.\n2. Hồ sơ đề nghị hỗ trợ học nghề đối với người lao động đang hưởng trợ cấp thất nghiệp mà có nhu cầu học nghề tại địa phương không phải nơi đang hưởng trợ cấp thất nghiệp bao gồm đề nghị hỗ trợ học nghề theo quy định tại khoản 1 Điều này và quyết định về việc hưởng trợ cấp thất nghiệp. Quyết định về việc hưởng trợ cấp thất nghiệp là bản chính hoặc bản sao có chứng thực hoặc bản sao kèm theo bản chính để đối chiếu.\n3. Hồ sơ đề nghị hỗ trợ học nghề đối với người lao động không thuộc trường hợp theo quy định tại khoản 1, khoản 2 Điều này, bao gồm:\na) Đề nghị hưởng trợ cấp thất nghiệp;\nb) Đề nghị hỗ trợ học nghề theo quy định tại khoản 1 Điều này;\nc) Bản chính hoặc bản sao có chứng thực hoặc bản sao kèm theo bản chính để đối chiếu của một trong các giấy tờ theo quy định tại khoản 2 Điều 16 Nghị định số 28/2015/NĐ-CP đã được sửa đổi, bổ sung tại khoản 6 Điều 1 Nghị định này;\nd) Sổ bảo hiểm xã hội.\nTrường hợp người lao động đã nộp hồ sơ đề nghị hưởng trợ cấp thất nghiệp nhưng không đáp ứng đủ điều kiện hưởng thì hồ sơ đề nghị hỗ trợ học nghề là đề nghị hỗ trợ học nghề theo quy định tại khoản 1 Điều này và thông báo của trung tâm dịch vụ việc làm về việc người lao động không đáp ứng đủ điều kiện hưởng trợ cấp thất nghiệp.", "\"1. Người thuộc một trong các trường hợp sau đây thì bị xóa đăng ký thường trú:\nd) Vắng mặt liên tục tại nơi thường trú từ 12 tháng trở lên mà không đăng ký tạm trú tại chỗ ở khác hoặc không khai báo tạm vắng, trừ trường hợp xuất cảnh ra nước ngoài nhưng không phải để định cư hoặc trường hợp đang chấp hành án phạt tù, chấp hành biện pháp đưa vào cơ sở giáo dục bắt buộc, đưa vào cơ sở cai nghiện bắt buộc, đưa vào trường giáo dưỡng;\"", "“1. Công ty hợp danh là doanh nghiệp, trong đó:\na) Phải có ít nhất 02 thành viên là chủ sở hữu chung của công ty, cùng nhau kinh doanh dưới một tên chung (sau đây gọi là thành viên hợp danh). Ngoài các thành viên hợp danh, công ty có thể có thêm thành viên góp vốn;\nb) Thành viên hợp danh phải là cá nhân, chịu trách nhiệm bằng toàn bộ tài sản của mình về các nghĩa vụ của công ty;\nc) Thành viên góp vốn là tổ chức, cá nhân và chỉ chịu trách nhiệm về các khoản nợ của công ty trong phạm vi số vốn đã cam kết góp vào công ty.”"]}, {"source_sentence": "Lấn chiếm đất công là đất phi nông nghiệp để xây dựng công trình thì sẽ bị xử phạt vi phạm hành chính như thế nào?", "sentences": ["1. Mọi người có quyền tự do tín ngưỡng, tôn giáo, theo hoặc không theo một tôn giáo nào. Các tôn giáo bình đẳng trước pháp luật.\n2. Nhà nước tôn trọng và bảo hộ quyền tự do tín ngưỡng, tôn giáo. \n3. Không ai được xâm phạm tự do tín ngưỡng, tôn giáo hoặc lợi dụng tín ngưỡng, tôn giáo để vi phạm pháp luật.1. Nhà nước tôn trọng và bảo hộ quyền tự do tín ngưỡng, tôn giáo của mọi người; bảo đảm để các tôn giáo bình đẳng trước pháp luật.\n2. Nhà nước tôn trọng, bảo vệ giá trị văn hóa, đạo đức tốt đẹp của tín ngưỡng, tôn giáo, truyền thống thờ cúng tổ tiên, tôn vinh người có công với đất nước, với cộng đồng đáp ứng nhu cầu tinh thần của Nhân dân.\n3. Nhà nước bảo hộ cơ sở tín ngưỡng, cơ sở tôn giáo và tài sản hợp pháp của cơ sở tín ngưỡng, tổ chức tôn giáo.", "Quy định chung về quy cách thể hiện\n1. Nội dung thể hiện đối với thuyết minh, bản vẽ của nhiệm vụ và đồ án trong hồ sơ lấy ý kiến, hồ sơ trình thẩm định và phê duyệt phải được trình bày rõ ràng, đầy đủ nội dung đối với từng loại, cấp độ quy hoạch và tuân thủ các quy định tại Chương II và các phụ lục của Thông tư này.\n2. Hệ thống ký hiệu bản vẽ của nhiệm vụ và đồ án quy hoạch đô thị, nông thôn phải tuân thủ theo quy định tại các phụ lục kèm theo Thông tư này.\n3. Căn cứ điều kiện, đặc điểm của vùng liên huyện, vùng huyện, đô thị, khu chức năng và khu vực nông thôn được lập quy hoạch, có thể thành lập các bản đồ, sơ đồ riêng cho từng nội dung hiện trạng, định hướng hạ tầng kỹ thuật để đảm bảo thể hiện được đầy đủ các nội dung theo yêu cầu.\n4. Trường hợp quy hoạch vùng, quy hoạch tỉnh chưa được phê duyệt, các bản đồ phương hướng phát triển hệ thống đô thị, nông thôn; bản đồ phương án quy hoạch hệ thống đô thị, nông thôn; bản đồ quy hoạch sử dụng đất cấp tỉnh; bản đồ tổ chức không gian và phân vùng chức năng trong quy hoạch vùng, quy hoạch tỉnh được thay thế bằng sơ đồ định hướng phát triển không gian trong quy hoạch xây dựng vùng tỉnh đã được phê duyệt.\n5. Tất cả các sơ đồ, bản đồ trong thành phần bản vẽ của hồ sơ nhiệm vụ và hồ sơ đồ án quy hoạch đô thị và nông thôn phải thể hiện rõ phạm vi ranh giới lập quy hoạch; ranh giới, tên các đơn vị hành chính trong phạm vi lập quy hoạch; tên đơn vị hành chính tiếp giáp bên ngoài phạm vi lập quy hoạch.", "Lấn, chiếm đất\n...\n4. Trường hợp lấn, chiếm đất phi nông nghiệp, trừ trường hợp quy định tại khoản 6 Điều này tại khu vực nông thôn thì hình thức và mức xử phạt như sau:\na) Phạt tiền từ 10.000.000 đồng đến 20.000.000 đồng nếu diện tích đất lấn, chiếm dưới 0,05 héc ta;\nb) Phạt tiền từ 20.000.000 đồng đến 40.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,05 héc ta đến dưới 0,1 héc ta;\nc) Phạt tiền từ 40.000.000 đồng đến 100.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,1 héc ta đến dưới 0,5 héc ta;\nd) Phạt tiền từ 100.000.000 đồng đến 200.000.000 đồng nếu diện tích đất lấn, chiếm từ 0,5 héc ta đến dưới 01 héc ta;\nđ) Phạt tiền từ 200.000.000 đồng đến 500.000.000 đồng nếu diện tích đất lấn, chiếm từ 01 héc ta trở lên.\n5. Trường hợp lấn, chiếm đất chưa sử dụng, đất nông nghiệp, đất phi nông nghiệp (trừ trường hợp quy định tại khoản 6 Điều này) tại khu vực đô thị thì mức xử phạt bằng 02 lần mức xử phạt đối với loại đất tương ứng quy định tại các khoản 1, 2, 3 và 4 Điều này và mức phạt tối đa không quá 500.000.000 đồng đối với cá nhân, không quá 1.000.000.000 đồng đối với tổ chức.\n6. Trường hợp lấn, chiếm đất thuộc hành lang bảo vệ an toàn công trình và đất công trình có hành lang bảo vệ, đất trụ sở làm việc và cơ sở hoạt động sự nghiệp của cơ quan, tổ chức theo quy định của pháp luật về quản lý, sử dụng tài sản nhà nước thì hình thức và mức xử phạt thực hiện theo quy định của pháp luật về xử phạt vi phạm hành chính trong lĩnh vực về hoạt động đầu tư xây dựng; khai thác, chế biến, kinh doanh khoáng sản làm vật liệu xây dựng, sản xuất, kinh doanh vật liệu xây dựng; quản lý công trình hạ tầng kỹ thuật; kinh doanh bất động sản, phát triển nhà ở, quản lý sử dụng nhà và công sở; trong lĩnh vực về giao thông đường bộ và đường sắt; trong lĩnh vực về văn hóa, thể thao, du lịch và quảng cáo; trong lĩnh vực về khai thác và bảo vệ công trình thủy lợi; đê điều; phòng, chống lụt, bão; trong lĩnh vực quản lý, sử dụng tài sản nhà nước và các lĩnh vực chuyên ngành khác.\n...Áp dụng mức phạt tiền\n1. Đối tượng áp dụng mức phạt tiền thực "]}], "model-index": [{"name": "SentenceTransformer based on BAAI/bge-m3", "results": [{"task": {"type": "information-retrieval", "name": "Information Retrieval"}, "dataset": {"name": "Unknown", "type": "unknown"}, "metrics": [{"type": "cosine_accuracy@1", "value": 0.9120603015075377, "name": "Cosine Accuracy@1"}, {"type": "cosine_accuracy@3", "value": 0.983249581239531, "name": "Cosine Accuracy@3"}, {"type": "cosine_accuracy@5", "value": 0.990787269681742, "name": "Cosine Accuracy@5"}, {"type": "cosine_accuracy@10", "value": 0.9958123953098827, "name": "Cosine Accuracy@10"}, {"type": "cosine_precision@1", "value": 0.9120603015075377, "name": "Cosine Precision@1"}, {"type": "cosine_precision@3", "value": 0.327749860413177, "name": "Cosine Precision@3"}, {"type": "cosine_precision@5", "value": 0.1981574539363484, "name": "Cosine Precision@5"}, {"type": "cosine_precision@10", "value": 0.09958123953098827, "name": "Cosine Precision@10"}, {"type": "cosine_recall@1", "value": 0.9120603015075377, "name": "Cosine Recall@1"}, {"type": "cosine_recall@3", "value": 0.983249581239531, "name": "Cosine Recall@3"}, {"type": "cosine_recall@5", "value": 0.990787269681742, "name": "Cosine Recall@5"}, {"type": "cosine_recall@10", "value": 0.9958123953098827, "name": "Cosine Recall@10"}, {"type": "cosine_ndcg@10", "value": 0.9596142418815723, "name": "Cosine Ndcg@10"}, {"type": "cosine_mrr@10", "value": 0.9473638031426979, "name": "Cosine Mrr@10"}, {"type": "cosine_map@100", "value": 0.9475627411515736, "name": "Cosine Map@100"}]}]}]}	task	[ "TEXT_CLASSIFICATION" ]	43,159
purpleor/autotrain-k11p0-ej4kd	purpleor	text-classification	[ "transformers", "tensorboard", "safetensors", "deberta-v2", "text-classification", "autotrain", "dataset:autotrain-k11p0-ej4kd/autotrain-data", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-04-29T07:57:55Z	2024-04-29T09:42:56+00:00	6	0	--- datasets: - autotrain-k11p0-ej4kd/autotrain-data tags: - autotrain - text-classification widget: - text: I love AutoTrain --- # Model Trained Using AutoTrain - Problem type: Text Classification ## Validation Metrics loss: 0.02587084472179413 f1: 0.994432092613179 precision: 0.9922734879273248 recall: 0.9966001094879995 auc: 0.9992992593502781 accuracy: 0.9944289693593314	null	Non_BioNLP	# Model Trained Using AutoTrain - Problem type: Text Classification ## Validation Metrics loss: 0.02587084472179413 f1: 0.994432092613179 precision: 0.9922734879273248 recall: 0.9966001094879995 auc: 0.9992992593502781 accuracy: 0.9944289693593314	{"datasets": ["autotrain-k11p0-ej4kd/autotrain-data"], "tags": ["autotrain", "text-classification"], "widget": [{"text": "I love AutoTrain"}]}	task	[ "TEXT_CLASSIFICATION" ]	43,160
TheBloke/ALMA-7B-GPTQ	TheBloke	text-generation	[ "transformers", "safetensors", "llama", "text-generation", "arxiv:2309.11674", "base_model:haoranxu/ALMA-7B", "base_model:quantized:haoranxu/ALMA-7B", "license:mit", "autotrain_compatible", "text-generation-inference", "4-bit", "gptq", "region:us" ]	2023-10-14T18:18:05Z	2023-10-14T18:40:45+00:00	19	1	--- base_model: haoranxu/ALMA-7B license: mit model_name: ALMA 7B inference: false model_creator: Haoran Xu model_type: llama prompt_template: 'Translate this from Chinese to English: Chinese: {prompt} English: ' quantized_by: TheBloke --- <!-- header start --> <!-- 200823 --> <div style="width: auto; margin-left: auto; margin-right: auto"> <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;"> </div> <div style="display: flex; justify-content: space-between; width: 100%;"> <div style="display: flex; flex-direction: column; align-items: flex-start;"> <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p> </div> <div style="display: flex; flex-direction: column; align-items: flex-end;"> <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p> </div> </div> <div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div> <hr style="margin-top: 1.0em; margin-bottom: 1.0em;"> <!-- header end --> # ALMA 7B - GPTQ - Model creator: [Haoran Xu](https://huggingface.co/haoranxu) - Original model: [ALMA 7B](https://huggingface.co/haoranxu/ALMA-7B) <!-- description start --> ## Description This repo contains GPTQ model files for [Haoran Xu's ALMA 7B](https://huggingface.co/haoranxu/ALMA-7B). Multiple GPTQ parameter permutations are provided; see Provided Files below for details of the options provided, their parameters, and the software used to create them. <!-- description end --> <!-- repositories-available start --> ## Repositories available * [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/ALMA-7B-AWQ) * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/ALMA-7B-GPTQ) * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/ALMA-7B-GGUF) * [Haoran Xu's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/haoranxu/ALMA-7B) <!-- repositories-available end --> <!-- prompt-template start --> ## Prompt template: ALMA ``` Translate this from Chinese to English: Chinese: {prompt} English: ``` <!-- prompt-template end --> <!-- licensing start --> ## Licensing The creator of the source model has listed its license as `mit`, and this quantization has therefore used that same license. As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly. In the meantime, any questions regarding licensing, and in particular how these two licenses might interact, should be directed to the original model repository: [Haoran Xu's ALMA 7B](https://huggingface.co/haoranxu/ALMA-7B). <!-- licensing end --> <!-- README_GPTQ.md-provided-files start --> ## Provided files, and GPTQ parameters Multiple quantisation parameters are provided, to allow you to choose the best one for your hardware and requirements. Each separate quant is in a different branch. See below for instructions on fetching from different branches. Most GPTQ files are made with AutoGPTQ. Mistral models are currently made with Transformers. <details> <summary>Explanation of GPTQ parameters</summary> - Bits: The bit size of the quantised model. - GS: GPTQ group size. Higher numbers use less VRAM, but have lower quantisation accuracy. "None" is the lowest possible value. - Act Order: True or False. Also known as `desc_act`. True results in better quantisation accuracy. Some GPTQ clients have had issues with models that use Act Order plus Group Size, but this is generally resolved now. - Damp %: A GPTQ parameter that affects how samples are processed for quantisation. 0.01 is default, but 0.1 results in slightly better accuracy. - GPTQ dataset: The calibration dataset used during quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ calibration dataset is not the same as the dataset used to train the model - please refer to the original model repo for details of the training dataset(s). - Sequence Length: The length of the dataset sequences used for quantisation. Ideally this is the same as the model sequence length. For some very long sequence models (16+K), a lower sequence length may have to be used. Note that a lower sequence length does not limit the sequence length of the quantised model. It only impacts the quantisation accuracy on longer inference sequences. - ExLlama Compatibility: Whether this file can be loaded with ExLlama, which currently only supports Llama models in 4-bit. </details> \| Branch \| Bits \| GS \| Act Order \| Damp % \| GPTQ Dataset \| Seq Len \| Size \| ExLlama \| Desc \| \| ------ \| ---- \| -- \| --------- \| ------ \| ------------ \| ------- \| ---- \| ------- \| ---- \| \| [main](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/main) \| 4 \| 128 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 3.90 GB \| Yes \| 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. \| \| [gptq-4bit-32g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-4bit-32g-actorder_True) \| 4 \| 32 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 4.28 GB \| Yes \| 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. \| \| [gptq-8bit--1g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-8bit--1g-actorder_True) \| 8 \| None \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 7.01 GB \| No \| 8-bit, with Act Order. No group size, to lower VRAM requirements. \| \| [gptq-8bit-128g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-8bit-128g-actorder_True) \| 8 \| 128 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 7.16 GB \| No \| 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. \| \| [gptq-8bit-32g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-8bit-32g-actorder_True) \| 8 \| 32 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 7.62 GB \| No \| 8-bit, with group size 32g and Act Order for maximum inference quality. \| \| [gptq-4bit-64g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-4bit-64g-actorder_True) \| 4 \| 64 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 4.02 GB \| Yes \| 4-bit, with Act Order and group size 64g. Uses less VRAM than 32g, but with slightly lower accuracy. \| <!-- README_GPTQ.md-provided-files end --> <!-- README_GPTQ.md-download-from-branches start --> ## How to download, including from branches ### In text-generation-webui To download from the `main` branch, enter `TheBloke/ALMA-7B-GPTQ` in the "Download model" box. To download from another branch, add `:branchname` to the end of the download name, eg `TheBloke/ALMA-7B-GPTQ:gptq-4bit-32g-actorder_True` ### From the command line I recommend using the `huggingface-hub` Python library: ```shell pip3 install huggingface-hub ``` To download the `main` branch to a folder called `ALMA-7B-GPTQ`: ```shell mkdir ALMA-7B-GPTQ huggingface-cli download TheBloke/ALMA-7B-GPTQ --local-dir ALMA-7B-GPTQ --local-dir-use-symlinks False ``` To download from a different branch, add the `--revision` parameter: ```shell mkdir ALMA-7B-GPTQ huggingface-cli download TheBloke/ALMA-7B-GPTQ --revision gptq-4bit-32g-actorder_True --local-dir ALMA-7B-GPTQ --local-dir-use-symlinks False ``` <details> <summary>More advanced huggingface-cli download usage</summary> If you remove the `--local-dir-use-symlinks False` parameter, the files will instead be stored in the central Huggingface cache directory (default location on Linux is: `~/.cache/huggingface`), and symlinks will be added to the specified `--local-dir`, pointing to their real location in the cache. This allows for interrupted downloads to be resumed, and allows you to quickly clone the repo to multiple places on disk without triggering a download again. The downside, and the reason why I don't list that as the default option, is that the files are then hidden away in a cache folder and it's harder to know where your disk space is being used, and to clear it up if/when you want to remove a download model. The cache location can be changed with the `HF_HOME` environment variable, and/or the `--cache-dir` parameter to `huggingface-cli`. For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli). To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`: ```shell pip3 install hf_transfer ``` And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`: ```shell mkdir ALMA-7B-GPTQ HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/ALMA-7B-GPTQ --local-dir ALMA-7B-GPTQ --local-dir-use-symlinks False ``` Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command. </details> ### With `git` (not recommended) To clone a specific branch with `git`, use a command like this: ```shell git clone --single-branch --branch gptq-4bit-32g-actorder_True https://huggingface.co/TheBloke/ALMA-7B-GPTQ ``` Note that using Git with HF repos is strongly discouraged. It will be much slower than using `huggingface-hub`, and will use twice as much disk space as it has to store the model files twice (it stores every byte both in the intended target folder, and again in the `.git` folder as a blob.) <!-- README_GPTQ.md-download-from-branches end --> <!-- README_GPTQ.md-text-generation-webui start --> ## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui). Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui). It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install. 1. Click the Model tab. 2. Under Download custom model or LoRA, enter `TheBloke/ALMA-7B-GPTQ`. - To download from a specific branch, enter for example `TheBloke/ALMA-7B-GPTQ:gptq-4bit-32g-actorder_True` - see Provided Files above for the list of branches for each option. 3. Click Download. 4. The model will start downloading. Once it's finished it will say "Done". 5. In the top left, click the refresh icon next to Model. 6. In the Model dropdown, choose the model you just downloaded: `ALMA-7B-GPTQ` 7. The model will automatically load, and is now ready for use! 8. If you want any custom settings, set them and then click Save settings for this model followed by Reload the Model in the top right. * Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file `quantize_config.json`. 9. Once you're ready, click the Text Generation tab and enter a prompt to get started! <!-- README_GPTQ.md-text-generation-webui end --> <!-- README_GPTQ.md-use-from-tgi start --> ## Serving this model from Text Generation Inference (TGI) It's recommended to use TGI version 1.1.0 or later. The official Docker container is: `ghcr.io/huggingface/text-generation-inference:1.1.0` Example Docker parameters: ```shell --model-id TheBloke/ALMA-7B-GPTQ --port 3000 --quantize awq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096 ``` Example Python code for interfacing with TGI (requires huggingface-hub 0.17.0 or later): ```shell pip3 install huggingface-hub ``` ```python from huggingface_hub import InferenceClient endpoint_url = "https://your-endpoint-url-here" prompt = "Tell me about AI" prompt_template=f'''Translate this from Chinese to English: Chinese: {prompt} English: ''' client = InferenceClient(endpoint_url) response = client.text_generation(prompt, max_new_tokens=128, do_sample=True, temperature=0.7, top_p=0.95, top_k=40, repetition_penalty=1.1) print(f"Model output: {response}") ``` <!-- README_GPTQ.md-use-from-tgi end --> <!-- README_GPTQ.md-use-from-python start --> ## How to use this GPTQ model from Python code ### Install the necessary packages Requires: Transformers 4.33.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later. ```shell pip3 install transformers optimum pip3 install auto-gptq --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/ # Use cu117 if on CUDA 11.7 ``` If you have problems installing AutoGPTQ using the pre-built wheels, install it from source instead: ```shell pip3 uninstall -y auto-gptq git clone https://github.com/PanQiWei/AutoGPTQ cd AutoGPTQ git checkout v0.4.2 pip3 install . ``` ### You can then use the following code ```python from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline model_name_or_path = "TheBloke/ALMA-7B-GPTQ" # To use a different branch, change revision # For example: revision="gptq-4bit-32g-actorder_True" model = AutoModelForCausalLM.from_pretrained(model_name_or_path, device_map="auto", trust_remote_code=False, revision="main") tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True) prompt = "Tell me about AI" prompt_template=f'''Translate this from Chinese to English: Chinese: {prompt} English: ''' print("\n\n* Generate:") input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda() output = model.generate(inputs=input_ids, temperature=0.7, do_sample=True, top_p=0.95, top_k=40, max_new_tokens=512) print(tokenizer.decode(output[0])) # Inference can also be done using transformers' pipeline print("* Pipeline:") pipe = pipeline( "text-generation", model=model, tokenizer=tokenizer, max_new_tokens=512, do_sample=True, temperature=0.7, top_p=0.95, top_k=40, repetition_penalty=1.1 ) print(pipe(prompt_template)[0]['generated_text']) ``` <!-- README_GPTQ.md-use-from-python end --> <!-- README_GPTQ.md-compatibility start --> ## Compatibility The files provided are tested to work with AutoGPTQ, both via Transformers and using AutoGPTQ directly. They should also work with [Occ4m's GPTQ-for-LLaMa fork](https://github.com/0cc4m/KoboldAI). [ExLlama](https://github.com/turboderp/exllama) is compatible with Llama and Mistral models in 4-bit. Please see the Provided Files table above for per-file compatibility. [Huggingface Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) is compatible with all GPTQ models. <!-- README_GPTQ.md-compatibility end --> <!-- footer start --> <!-- 200823 --> ## Discord For further support, and discussions on these models and AI in general, join us at: [TheBloke AI's Discord server](https://discord.gg/theblokeai) ## Thanks, and how to contribute Thanks to the [chirper.ai](https://chirper.ai) team! Thanks to Clay from [gpus.llm-utils.org](llm-utils)! I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training. If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects. Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits. * Patreon: https://patreon.com/TheBlokeAI * Ko-Fi: https://ko-fi.com/TheBlokeAI Special thanks to: Aemon Algiz. Patreon special mentions: Pierre Kircher, Stanislav Ovsiannikov, Michael Levine, Eugene Pentland, Andrey, 준교 김, Randy H, Fred von Graf, Artur Olbinski, Caitlyn Gatomon, terasurfer, Jeff Scroggin, James Bentley, Vadim, Gabriel Puliatti, Harry Royden McLaughlin, Sean Connelly, Dan Guido, Edmond Seymore, Alicia Loh, subjectnull, AzureBlack, Manuel Alberto Morcote, Thomas Belote, Lone Striker, Chris Smitley, Vitor Caleffi, Johann-Peter Hartmann, Clay Pascal, biorpg, Brandon Frisco, sidney chen, transmissions 11, Pedro Madruga, jinyuan sun, Ajan Kanaga, Emad Mostaque, Trenton Dambrowitz, Jonathan Leane, Iucharbius, usrbinkat, vamX, George Stoitzev, Luke Pendergrass, theTransient, Olakabola, Swaroop Kallakuri, Cap'n Zoog, Brandon Phillips, Michael Dempsey, Nikolai Manek, danny, Matthew Berman, Gabriel Tamborski, alfie_i, Raymond Fosdick, Tom X Nguyen, Raven Klaugh, LangChain4j, Magnesian, Illia Dulskyi, David Ziegler, Mano Prime, Luis Javier Navarrete Lozano, Erik Bjäreholt, 阿明, Nathan Dryer, Alex, Rainer Wilmers, zynix, TL, Joseph William Delisle, John Villwock, Nathan LeClaire, Willem Michiel, Joguhyik, GodLy, OG, Alps Aficionado, Jeffrey Morgan, ReadyPlayerEmma, Tiffany J. Kim, Sebastain Graf, Spencer Kim, Michael Davis, webtim, Talal Aujan, knownsqashed, John Detwiler, Imad Khwaja, Deo Leter, Jerry Meng, Elijah Stavena, Rooh Singh, Pieter, SuperWojo, Alexandros Triantafyllidis, Stephen Murray, Ai Maven, ya boyyy, Enrico Ros, Ken Nordquist, Deep Realms, Nicholas, Spiking Neurons AB, Elle, Will Dee, Jack West, RoA, Luke @flexchar, Viktor Bowallius, Derek Yates, Subspace Studios, jjj, Toran Billups, Asp the Wyvern, Fen Risland, Ilya, NimbleBox.ai, Chadd, Nitin Borwankar, Emre, Mandus, Leonard Tan, Kalila, K, Trailburnt, S_X, Cory Kujawski Thank you to all my generous patrons and donaters! And thank you again to a16z for their generous grant. <!-- footer end --> # Original model card: Haoran Xu's ALMA 7B ALMA (Advanced Language Model-based trAnslator) is an LLM-based translation model, which adopts a new translation model paradigm: it begins with fine-tuning on monolingual data and is further optimized using high-quality parallel data. This two-step fine-tuning process ensures strong translation performance. Please find more details in our [paper](https://arxiv.org/abs/2309.11674). ``` @misc{xu2023paradigm, title={A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models}, author={Haoran Xu and Young Jin Kim and Amr Sharaf and Hany Hassan Awadalla}, year={2023}, eprint={2309.11674}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` We release four translation models presented in the paper: - ALMA-7B: Full-weight Fine-tune LLaMA-2-7B on 20B monolingual tokens and then Full-weight fine-tune on human-written parallel data - ALMA-7B-LoRA: Full-weight Fine-tune LLaMA-2-7B on 20B monolingual tokens and then LoRA fine-tune on human-written parallel data - ALMA-13B: Full-weight Fine-tune LLaMA-2-7B on 12B monolingual tokens and then Full-weight fine-tune on human-written parallel data - ALMA-13B-LoRA (Our best system): Full-weight Fine-tune LLaMA-2-7B on 12B monolingual tokens and then LoRA fine-tune on human-written parallel data Model checkpoints are released at huggingface: \| Models \| Base Model Link \| LoRA Link \| \|:-------------:\|:---------------:\|:---------:\| \| ALMA-7B \| [haoranxu/ALMA-7B](https://huggingface.co/haoranxu/ALMA-7B) \| - \| \| ALMA-7B-LoRA \| [haoranxu/ALMA-7B-Pretrain](https://huggingface.co/haoranxu/ALMA-7B-Pretrain) \| [haoranxu/ALMA-7B-Pretrain-LoRA](https://huggingface.co/haoranxu/ALMA-7B-Pretrain-LoRA) \| \| ALMA-13B \| [haoranxu/ALMA-13B](https://huggingface.co/haoranxu/ALMA-13B) \| - \| \| ALMA-13B-LoRA \| [haoranxu/ALMA-13B-Pretrain](https://huggingface.co/haoranxu/ALMA-13B-Pretrain) \| [haoranxu/ALMA-13B-Pretrain-LoRA](https://huggingface.co/haoranxu/ALMA-13B-Pretrain-LoRA) \| Note that `ALMA-7B-Pretrain` and `ALMA-13B-Pretrain` are NOT translation models. They only experience stage 1 monolingual fine-tuning (20B tokens for the 7B model and 12B tokens for the 13B model), and should be utilized in conjunction with their LoRA models for translation purposes. A quick start to use our best system (ALMA-13B-LoRA) for translation. An example of translating "我爱机器翻译。" into English: ``` import torch from peft import PeftModel from transformers import AutoModelForCausalLM from transformers import LlamaTokenizer # Load base model and LoRA weights model = AutoModelForCausalLM.from_pretrained("haoranxu/ALMA-13B-Pretrain", torch_dtype=torch.float16, device_map="auto") model = PeftModel.from_pretrained(model, "haoranxu/ALMA-13B-Pretrain-LoRA") tokenizer = LlamaTokenizer.from_pretrained("haoranxu/ALMA-13B-Pretrain", padding_side='left') # Add the source setence into the prompt template prompt="Translate this from Chinese to English:\nChinese: 我爱机器翻译。\nEnglish:" input_ids = tokenizer(prompt, return_tensors="pt", padding=True, max_length=40, truncation=True).input_ids.cuda() # Translation with torch.no_grad(): generated_ids = model.generate(input_ids=input_ids, num_beams=5, max_new_tokens=20, do_sample=True, temperature=0.6, top_p=0.9) outputs = tokenizer.batch_decode(generated_ids, skip_special_tokens=True) print(outputs) ``` Please find more details in our [GitHub repository](https://github.com/fe1ixxu/ALMA)	null	Non_BioNLP	<!-- header start --> <!-- 200823 --> <div style="width: auto; margin-left: auto; margin-right: auto"> <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;"> </div> <div style="display: flex; justify-content: space-between; width: 100%;"> <div style="display: flex; flex-direction: column; align-items: flex-start;"> <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p> </div> <div style="display: flex; flex-direction: column; align-items: flex-end;"> <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p> </div> </div> <div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div> <hr style="margin-top: 1.0em; margin-bottom: 1.0em;"> <!-- header end --> # ALMA 7B - GPTQ - Model creator: [Haoran Xu](https://huggingface.co/haoranxu) - Original model: [ALMA 7B](https://huggingface.co/haoranxu/ALMA-7B) <!-- description start --> ## Description This repo contains GPTQ model files for [Haoran Xu's ALMA 7B](https://huggingface.co/haoranxu/ALMA-7B). Multiple GPTQ parameter permutations are provided; see Provided Files below for details of the options provided, their parameters, and the software used to create them. <!-- description end --> <!-- repositories-available start --> ## Repositories available * [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/ALMA-7B-AWQ) * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/ALMA-7B-GPTQ) * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/ALMA-7B-GGUF) * [Haoran Xu's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/haoranxu/ALMA-7B) <!-- repositories-available end --> <!-- prompt-template start --> ## Prompt template: ALMA ``` Translate this from Chinese to English: Chinese: {prompt} English: ``` <!-- prompt-template end --> <!-- licensing start --> ## Licensing The creator of the source model has listed its license as `mit`, and this quantization has therefore used that same license. As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly. In the meantime, any questions regarding licensing, and in particular how these two licenses might interact, should be directed to the original model repository: [Haoran Xu's ALMA 7B](https://huggingface.co/haoranxu/ALMA-7B). <!-- licensing end --> <!-- README_GPTQ.md-provided-files start --> ## Provided files, and GPTQ parameters Multiple quantisation parameters are provided, to allow you to choose the best one for your hardware and requirements. Each separate quant is in a different branch. See below for instructions on fetching from different branches. Most GPTQ files are made with AutoGPTQ. Mistral models are currently made with Transformers. <details> <summary>Explanation of GPTQ parameters</summary> - Bits: The bit size of the quantised model. - GS: GPTQ group size. Higher numbers use less VRAM, but have lower quantisation accuracy. "None" is the lowest possible value. - Act Order: True or False. Also known as `desc_act`. True results in better quantisation accuracy. Some GPTQ clients have had issues with models that use Act Order plus Group Size, but this is generally resolved now. - Damp %: A GPTQ parameter that affects how samples are processed for quantisation. 0.01 is default, but 0.1 results in slightly better accuracy. - GPTQ dataset: The calibration dataset used during quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ calibration dataset is not the same as the dataset used to train the model - please refer to the original model repo for details of the training dataset(s). - Sequence Length: The length of the dataset sequences used for quantisation. Ideally this is the same as the model sequence length. For some very long sequence models (16+K), a lower sequence length may have to be used. Note that a lower sequence length does not limit the sequence length of the quantised model. It only impacts the quantisation accuracy on longer inference sequences. - ExLlama Compatibility: Whether this file can be loaded with ExLlama, which currently only supports Llama models in 4-bit. </details> \| Branch \| Bits \| GS \| Act Order \| Damp % \| GPTQ Dataset \| Seq Len \| Size \| ExLlama \| Desc \| \| ------ \| ---- \| -- \| --------- \| ------ \| ------------ \| ------- \| ---- \| ------- \| ---- \| \| [main](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/main) \| 4 \| 128 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 3.90 GB \| Yes \| 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. \| \| [gptq-4bit-32g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-4bit-32g-actorder_True) \| 4 \| 32 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 4.28 GB \| Yes \| 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. \| \| [gptq-8bit--1g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-8bit--1g-actorder_True) \| 8 \| None \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 7.01 GB \| No \| 8-bit, with Act Order. No group size, to lower VRAM requirements. \| \| [gptq-8bit-128g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-8bit-128g-actorder_True) \| 8 \| 128 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 7.16 GB \| No \| 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. \| \| [gptq-8bit-32g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-8bit-32g-actorder_True) \| 8 \| 32 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 7.62 GB \| No \| 8-bit, with group size 32g and Act Order for maximum inference quality. \| \| [gptq-4bit-64g-actorder_True](https://huggingface.co/TheBloke/ALMA-7B-GPTQ/tree/gptq-4bit-64g-actorder_True) \| 4 \| 64 \| Yes \| 0.1 \| [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) \| 4096 \| 4.02 GB \| Yes \| 4-bit, with Act Order and group size 64g. Uses less VRAM than 32g, but with slightly lower accuracy. \| <!-- README_GPTQ.md-provided-files end --> <!-- README_GPTQ.md-download-from-branches start --> ## How to download, including from branches ### In text-generation-webui To download from the `main` branch, enter `TheBloke/ALMA-7B-GPTQ` in the "Download model" box. To download from another branch, add `:branchname` to the end of the download name, eg `TheBloke/ALMA-7B-GPTQ:gptq-4bit-32g-actorder_True` ### From the command line I recommend using the `huggingface-hub` Python library: ```shell pip3 install huggingface-hub ``` To download the `main` branch to a folder called `ALMA-7B-GPTQ`: ```shell mkdir ALMA-7B-GPTQ huggingface-cli download TheBloke/ALMA-7B-GPTQ --local-dir ALMA-7B-GPTQ --local-dir-use-symlinks False ``` To download from a different branch, add the `--revision` parameter: ```shell mkdir ALMA-7B-GPTQ huggingface-cli download TheBloke/ALMA-7B-GPTQ --revision gptq-4bit-32g-actorder_True --local-dir ALMA-7B-GPTQ --local-dir-use-symlinks False ``` <details> <summary>More advanced huggingface-cli download usage</summary> If you remove the `--local-dir-use-symlinks False` parameter, the files will instead be stored in the central Huggingface cache directory (default location on Linux is: `~/.cache/huggingface`), and symlinks will be added to the specified `--local-dir`, pointing to their real location in the cache. This allows for interrupted downloads to be resumed, and allows you to quickly clone the repo to multiple places on disk without triggering a download again. The downside, and the reason why I don't list that as the default option, is that the files are then hidden away in a cache folder and it's harder to know where your disk space is being used, and to clear it up if/when you want to remove a download model. The cache location can be changed with the `HF_HOME` environment variable, and/or the `--cache-dir` parameter to `huggingface-cli`. For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli). To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`: ```shell pip3 install hf_transfer ``` And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`: ```shell mkdir ALMA-7B-GPTQ HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/ALMA-7B-GPTQ --local-dir ALMA-7B-GPTQ --local-dir-use-symlinks False ``` Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command. </details> ### With `git` (not recommended) To clone a specific branch with `git`, use a command like this: ```shell git clone --single-branch --branch gptq-4bit-32g-actorder_True https://huggingface.co/TheBloke/ALMA-7B-GPTQ ``` Note that using Git with HF repos is strongly discouraged. It will be much slower than using `huggingface-hub`, and will use twice as much disk space as it has to store the model files twice (it stores every byte both in the intended target folder, and again in the `.git` folder as a blob.) <!-- README_GPTQ.md-download-from-branches end --> <!-- README_GPTQ.md-text-generation-webui start --> ## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui). Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui). It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install. 1. Click the Model tab. 2. Under Download custom model or LoRA, enter `TheBloke/ALMA-7B-GPTQ`. - To download from a specific branch, enter for example `TheBloke/ALMA-7B-GPTQ:gptq-4bit-32g-actorder_True` - see Provided Files above for the list of branches for each option. 3. Click Download. 4. The model will start downloading. Once it's finished it will say "Done". 5. In the top left, click the refresh icon next to Model. 6. In the Model dropdown, choose the model you just downloaded: `ALMA-7B-GPTQ` 7. The model will automatically load, and is now ready for use! 8. If you want any custom settings, set them and then click Save settings for this model followed by Reload the Model in the top right. * Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file `quantize_config.json`. 9. Once you're ready, click the Text Generation tab and enter a prompt to get started! <!-- README_GPTQ.md-text-generation-webui end --> <!-- README_GPTQ.md-use-from-tgi start --> ## Serving this model from Text Generation Inference (TGI) It's recommended to use TGI version 1.1.0 or later. The official Docker container is: `ghcr.io/huggingface/text-generation-inference:1.1.0` Example Docker parameters: ```shell --model-id TheBloke/ALMA-7B-GPTQ --port 3000 --quantize awq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096 ``` Example Python code for interfacing with TGI (requires huggingface-hub 0.17.0 or later): ```shell pip3 install huggingface-hub ``` ```python from huggingface_hub import InferenceClient endpoint_url = "https://your-endpoint-url-here" prompt = "Tell me about AI" prompt_template=f'''Translate this from Chinese to English: Chinese: {prompt} English: ''' client = InferenceClient(endpoint_url) response = client.text_generation(prompt, max_new_tokens=128, do_sample=True, temperature=0.7, top_p=0.95, top_k=40, repetition_penalty=1.1) print(f"Model output: {response}") ``` <!-- README_GPTQ.md-use-from-tgi end --> <!-- README_GPTQ.md-use-from-python start --> ## How to use this GPTQ model from Python code ### Install the necessary packages Requires: Transformers 4.33.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later. ```shell pip3 install transformers optimum pip3 install auto-gptq --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/ # Use cu117 if on CUDA 11.7 ``` If you have problems installing AutoGPTQ using the pre-built wheels, install it from source instead: ```shell pip3 uninstall -y auto-gptq git clone https://github.com/PanQiWei/AutoGPTQ cd AutoGPTQ git checkout v0.4.2 pip3 install . ``` ### You can then use the following code ```python from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline model_name_or_path = "TheBloke/ALMA-7B-GPTQ" # To use a different branch, change revision # For example: revision="gptq-4bit-32g-actorder_True" model = AutoModelForCausalLM.from_pretrained(model_name_or_path, device_map="auto", trust_remote_code=False, revision="main") tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True) prompt = "Tell me about AI" prompt_template=f'''Translate this from Chinese to English: Chinese: {prompt} English: ''' print("\n\n* Generate:") input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda() output = model.generate(inputs=input_ids, temperature=0.7, do_sample=True, top_p=0.95, top_k=40, max_new_tokens=512) print(tokenizer.decode(output[0])) # Inference can also be done using transformers' pipeline print("* Pipeline:") pipe = pipeline( "text-generation", model=model, tokenizer=tokenizer, max_new_tokens=512, do_sample=True, temperature=0.7, top_p=0.95, top_k=40, repetition_penalty=1.1 ) print(pipe(prompt_template)[0]['generated_text']) ``` <!-- README_GPTQ.md-use-from-python end --> <!-- README_GPTQ.md-compatibility start --> ## Compatibility The files provided are tested to work with AutoGPTQ, both via Transformers and using AutoGPTQ directly. They should also work with [Occ4m's GPTQ-for-LLaMa fork](https://github.com/0cc4m/KoboldAI). [ExLlama](https://github.com/turboderp/exllama) is compatible with Llama and Mistral models in 4-bit. Please see the Provided Files table above for per-file compatibility. [Huggingface Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) is compatible with all GPTQ models. <!-- README_GPTQ.md-compatibility end --> <!-- footer start --> <!-- 200823 --> ## Discord For further support, and discussions on these models and AI in general, join us at: [TheBloke AI's Discord server](https://discord.gg/theblokeai) ## Thanks, and how to contribute Thanks to the [chirper.ai](https://chirper.ai) team! Thanks to Clay from [gpus.llm-utils.org](llm-utils)! I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training. If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects. Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits. * Patreon: https://patreon.com/TheBlokeAI * Ko-Fi: https://ko-fi.com/TheBlokeAI Special thanks to: Aemon Algiz. Patreon special mentions: Pierre Kircher, Stanislav Ovsiannikov, Michael Levine, Eugene Pentland, Andrey, 준교 김, Randy H, Fred von Graf, Artur Olbinski, Caitlyn Gatomon, terasurfer, Jeff Scroggin, James Bentley, Vadim, Gabriel Puliatti, Harry Royden McLaughlin, Sean Connelly, Dan Guido, Edmond Seymore, Alicia Loh, subjectnull, AzureBlack, Manuel Alberto Morcote, Thomas Belote, Lone Striker, Chris Smitley, Vitor Caleffi, Johann-Peter Hartmann, Clay Pascal, biorpg, Brandon Frisco, sidney chen, transmissions 11, Pedro Madruga, jinyuan sun, Ajan Kanaga, Emad Mostaque, Trenton Dambrowitz, Jonathan Leane, Iucharbius, usrbinkat, vamX, George Stoitzev, Luke Pendergrass, theTransient, Olakabola, Swaroop Kallakuri, Cap'n Zoog, Brandon Phillips, Michael Dempsey, Nikolai Manek, danny, Matthew Berman, Gabriel Tamborski, alfie_i, Raymond Fosdick, Tom X Nguyen, Raven Klaugh, LangChain4j, Magnesian, Illia Dulskyi, David Ziegler, Mano Prime, Luis Javier Navarrete Lozano, Erik Bjäreholt, 阿明, Nathan Dryer, Alex, Rainer Wilmers, zynix, TL, Joseph William Delisle, John Villwock, Nathan LeClaire, Willem Michiel, Joguhyik, GodLy, OG, Alps Aficionado, Jeffrey Morgan, ReadyPlayerEmma, Tiffany J. Kim, Sebastain Graf, Spencer Kim, Michael Davis, webtim, Talal Aujan, knownsqashed, John Detwiler, Imad Khwaja, Deo Leter, Jerry Meng, Elijah Stavena, Rooh Singh, Pieter, SuperWojo, Alexandros Triantafyllidis, Stephen Murray, Ai Maven, ya boyyy, Enrico Ros, Ken Nordquist, Deep Realms, Nicholas, Spiking Neurons AB, Elle, Will Dee, Jack West, RoA, Luke @flexchar, Viktor Bowallius, Derek Yates, Subspace Studios, jjj, Toran Billups, Asp the Wyvern, Fen Risland, Ilya, NimbleBox.ai, Chadd, Nitin Borwankar, Emre, Mandus, Leonard Tan, Kalila, K, Trailburnt, S_X, Cory Kujawski Thank you to all my generous patrons and donaters! And thank you again to a16z for their generous grant. <!-- footer end --> # Original model card: Haoran Xu's ALMA 7B ALMA (Advanced Language Model-based trAnslator) is an LLM-based translation model, which adopts a new translation model paradigm: it begins with fine-tuning on monolingual data and is further optimized using high-quality parallel data. This two-step fine-tuning process ensures strong translation performance. Please find more details in our [paper](https://arxiv.org/abs/2309.11674). ``` @misc{xu2023paradigm, title={A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models}, author={Haoran Xu and Young Jin Kim and Amr Sharaf and Hany Hassan Awadalla}, year={2023}, eprint={2309.11674}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` We release four translation models presented in the paper: - ALMA-7B: Full-weight Fine-tune LLaMA-2-7B on 20B monolingual tokens and then Full-weight fine-tune on human-written parallel data - ALMA-7B-LoRA: Full-weight Fine-tune LLaMA-2-7B on 20B monolingual tokens and then LoRA fine-tune on human-written parallel data - ALMA-13B: Full-weight Fine-tune LLaMA-2-7B on 12B monolingual tokens and then Full-weight fine-tune on human-written parallel data - ALMA-13B-LoRA (Our best system): Full-weight Fine-tune LLaMA-2-7B on 12B monolingual tokens and then LoRA fine-tune on human-written parallel data Model checkpoints are released at huggingface: \| Models \| Base Model Link \| LoRA Link \| \|:-------------:\|:---------------:\|:---------:\| \| ALMA-7B \| [haoranxu/ALMA-7B](https://huggingface.co/haoranxu/ALMA-7B) \| - \| \| ALMA-7B-LoRA \| [haoranxu/ALMA-7B-Pretrain](https://huggingface.co/haoranxu/ALMA-7B-Pretrain) \| [haoranxu/ALMA-7B-Pretrain-LoRA](https://huggingface.co/haoranxu/ALMA-7B-Pretrain-LoRA) \| \| ALMA-13B \| [haoranxu/ALMA-13B](https://huggingface.co/haoranxu/ALMA-13B) \| - \| \| ALMA-13B-LoRA \| [haoranxu/ALMA-13B-Pretrain](https://huggingface.co/haoranxu/ALMA-13B-Pretrain) \| [haoranxu/ALMA-13B-Pretrain-LoRA](https://huggingface.co/haoranxu/ALMA-13B-Pretrain-LoRA) \| Note that `ALMA-7B-Pretrain` and `ALMA-13B-Pretrain` are NOT translation models. They only experience stage 1 monolingual fine-tuning (20B tokens for the 7B model and 12B tokens for the 13B model), and should be utilized in conjunction with their LoRA models for translation purposes. A quick start to use our best system (ALMA-13B-LoRA) for translation. An example of translating "我爱机器翻译。" into English: ``` import torch from peft import PeftModel from transformers import AutoModelForCausalLM from transformers import LlamaTokenizer # Load base model and LoRA weights model = AutoModelForCausalLM.from_pretrained("haoranxu/ALMA-13B-Pretrain", torch_dtype=torch.float16, device_map="auto") model = PeftModel.from_pretrained(model, "haoranxu/ALMA-13B-Pretrain-LoRA") tokenizer = LlamaTokenizer.from_pretrained("haoranxu/ALMA-13B-Pretrain", padding_side='left') # Add the source setence into the prompt template prompt="Translate this from Chinese to English:\nChinese: 我爱机器翻译。\nEnglish:" input_ids = tokenizer(prompt, return_tensors="pt", padding=True, max_length=40, truncation=True).input_ids.cuda() # Translation with torch.no_grad(): generated_ids = model.generate(input_ids=input_ids, num_beams=5, max_new_tokens=20, do_sample=True, temperature=0.6, top_p=0.9) outputs = tokenizer.batch_decode(generated_ids, skip_special_tokens=True) print(outputs) ``` Please find more details in our [GitHub repository](https://github.com/fe1ixxu/ALMA)	{"base_model": "haoranxu/ALMA-7B", "license": "mit", "model_name": "ALMA 7B", "inference": false, "model_creator": "Haoran Xu", "model_type": "llama", "prompt_template": "Translate this from Chinese to English:\nChinese: {prompt}\nEnglish:\n", "quantized_by": "TheBloke"}	task	[ "TRANSLATION" ]	43,161
inceptionai/jais-family-590m	inceptionai	text-generation	[ "safetensors", "jais", "Arabic", "English", "LLM", "Decoder", "causal-lm", "jais-family", "text-generation", "custom_code", "ar", "en", "arxiv:2307.09288", "arxiv:2308.16149", "arxiv:2402.12840", "license:apache-2.0", "region:us" ]	2024-08-02T11:51:47Z	2024-09-11T11:06:48+00:00	588	6	--- language: - ar - en license: apache-2.0 pipeline_tag: text-generation tags: - Arabic - English - LLM - Decoder - causal-lm - jais-family --- # Jais Family Model Card The Jais family of models is a comprehensive series of bilingual English-Arabic large language models (LLMs). These models are optimized to excel in Arabic while having strong English capabilities. We release two variants of foundation models that include: - Models pre-trained from scratch (`jais-family-`). - Models pre-trained adaptively from [Llama-2](https://arxiv.org/pdf/2307.09288)* (`jais-adapted-`). In this release, we introduce 20 models across 8 sizes, ranging from 590M to 70B parameters, trained on up to 1.6T tokens of Arabic, English, and code data. All* pre-trained models in this series are instruction fine-tuned (`-chat`) for dialog using a curated mix of Arabic and English instruction data. We hope this extensive release will accelerate research in Arabic NLP, and enable numerous downstream applications for the Arabic speaking and bilingual community. The training and adaptation techniques we demonstrate successfully for Arabic models are extensible to other low and medium resource languages. ## Jais Family Details - Developed by:* Inception, Cerebras Systems. - Language(s): (NLP): Arabic (MSA) and English. - Input: Text only data. - Output: Model generates text. - Model Sizes: 590M, 1.3B, 2.7B, 6.7B, 7B, 13B, 30B, 70B. - Demo: [Access the live demo here](https://arabic-gpt.ai/) - License: Apache 2.0 \| Pre-trained Model \| Fine-tuned Model \| Size (Parameters) \| Context length (Tokens) \| \|:---------------------\|:--------\|:-------\|:-------\| \| [jais-family-30b-16k](https://huggingface.co/inceptionai/jais-family-30b-16k) \| [Jais-family-30b-16k-chat](https://huggingface.co/inceptionai/jais-family-30b-16k-chat) \| 30B \| 16,384 \| \| [jais-family-30b-8k](https://huggingface.co/inceptionai/jais-family-30b-8k) \| [Jais-family-30b-8k-chat](https://huggingface.co/inceptionai/jais-family-30b-8k-chat) \| 30B \| 8,192 \| \| [jais-family-13b ](https://huggingface.co/inceptionai/jais-family-13b) \| [Jais-family-13b-chat](https://huggingface.co/inceptionai/jais-family-13b-chat) \| 13B \| 2,048 \| \| [jais-family-6p7b](https://huggingface.co/inceptionai/jais-family-6p7b) \| [Jais-family-6p7b-chat](https://huggingface.co/inceptionai/jais-family-6p7b-chat) \| 6.7B \| 2,048 \| \| [jais-family-2p7b](https://huggingface.co/inceptionai/jais-family-2p7b) \| [Jais-family-2p7b-chat](https://huggingface.co/inceptionai/jais-family-2p7b-chat) \| 2.7B \| 2,048 \| \| [jais-family-1p3b](https://huggingface.co/inceptionai/jais-family-1p3b) \| [Jais-family-1p3b-chat](https://huggingface.co/inceptionai/jais-family-1p3b-chat) \| 1.3B \| 2,048 \| \| [jais-family-590m](https://huggingface.co/inceptionai/jais-family-590m) \| [Jais-family-590m-chat](https://huggingface.co/inceptionai/jais-family-590m-chat) \| 590M \| 2,048 \| \| Adapted pre-trained Model \| Fine-tuned Model \| Size (Parameters) \| Context length (Tokens) \| \|:---------------------\|:--------\|:-------\|:-------\| \| [jais-adapted-70b](https://huggingface.co/inceptionai/jais-adapted-70b) \| [Jais-adapted-70b-chat](https://huggingface.co/inceptionai/jais-adapted-70b-chat) \| 70B \| 4,096 \| \| [jais-adapted-13b](https://huggingface.co/inceptionai/jais-adapted-13b) \| [Jais-adapted-13b-chat](https://huggingface.co/inceptionai/jais-adapted-13b-chat) \| 13B \| 4,096 \| \| [jais-adapted-7b](https://huggingface.co/inceptionai/jais-adapted-7b) \| [Jais-adapted-7b-chat](https://huggingface.co/inceptionai/jais-adapted-7b-chat) \| 7B \| 4,096 \| ### Model Architecture: <a name="model-architecture"></a> All models in this family are auto-regressive language models that use a transformer-based, decoder-only architecture (GPT-3). Jais models (`jais-family-`) are trained from scratch, incorporating the SwiGLU non-linear activation function and ALiBi position encoding. These architectural enhancements allow the models to extrapolate at long sequence lengths, leading to improved context handling and precision. Jais adapted models (`jais-adapted-`) are built on top of Llama-2, which employs RoPE position embedding and Grouped Query Attention. We introduce tokenizer expansion with Arabic data, which improves fertility and compute efficiency by over 3x. In particular, we add `32,000` new Arabic tokens from the Jais-30b vocabulary into the Llama-2 tokenizer. To initialize these new Arabic token embeddings we first learn a linear projection from the embedding space of Jais-30b to Llama's embedding space, using the set of shared English tokens present in both vocabularies. Next, this learned projection is applied to transform the existing Jais-30b Arabic embeddings into the Llama-2 embedding space. ## Getting started Below is sample code to use the model. Note that the model requires a custom model class, so users must enable `trust_remote_code=True` while loading the model. ```python # -- coding: utf-8 -- import torch from transformers import AutoTokenizer, AutoModelForCausalLM model_path = "inceptionai/jais-family-590m" device = "cuda" if torch.cuda.is_available() else "cpu" tokenizer = AutoTokenizer.from_pretrained(model_path) model = AutoModelForCausalLM.from_pretrained(model_path, device_map="auto", trust_remote_code=True) def get_response(text, tokenizer=tokenizer, model=model): input_ids = tokenizer(text, return_tensors="pt").input_ids inputs = input_ids.to(device) input_len = inputs.shape[-1] generate_ids = model.generate( inputs, top_p=0.9, temperature=0.3, max_length=2048, min_length=input_len + 4, repetition_penalty=1.2, do_sample=True, ) response = tokenizer.batch_decode( generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True )[0] return response text = "عاصمة دولة الإمارات العربية المتحدة ه" print(get_response(text)) text = "The capital of UAE is" print(get_response(text)) ``` ## Training Details ### Pretraining Data The Jais family of models are trained on up to 1.6 Trillion tokens of diverse English, Arabic and Code data. The data consists of the following sources: - Web: We used publicly available web pages, wikipedia articles, news articles, and social network content in both Arabic and English. - Code: To enhance the reasoning capability of our model, we include Code data in various programming languages. - Books: We used a selection of publicly available Arabic and English books data, which improves long-range context modelling and coherent storytelling. - Scientific: A subset of ArXiv papers were included to improve reasoning and long context abilities. - Synthetic: We augment the volume of Arabic data by translating English to Arabic using an in-house machine translation system. We restrict this to high quality English resources such as English Wikipedia and English books. We extensively preprocess and deduplicate the training data. For Arabic, we used a custom preprocessing pipeline to filter for data with high linguistic quality. More information on this pipeline can be found in the [Jais paper](https://arxiv.org/abs/2308.16149). - Jais pre-trained (`jais-family-`): Following our previous experimentation with language alignment mixing in [Jais](https://arxiv.org/abs/2308.16149), we used a ratio of 1:2:0.4 of Arabic:English:Code data. This recipe for <u>from scratch pre-training</u> addresses Arabic data scarcity while improving performance in both languages. - Jais adapted pre-trained* (`jais-adapted-`): For the <u>adapted pre-training of Llama-2</u>, we utilized a larger Arabic dataset of ~334B Arabic tokens mixed with English and Code data. We vary the mixing ratio, at different model sizes, to introduce strong Arabic capabilities while maintaining performance in English. \| Pre-trained model* \| English data (tokens) \| Arabic data (tokens) \| Code data (tokens) \| Total data (tokens) \| \|-------------------------\|---------------------------\|--------------------------\|------------------------\|------------------------\| \| [jais-family-30b-16k](https://huggingface.co/inceptionai/jais-family-30b-16k) \| 980B \| 490B \| 196B \| 1666B \| \| [jais-family-30b-8k](https://huggingface.co/inceptionai/jais-family-30b-8k) \| 882B \| 441B \| 177B \| 1500B \| \| [jais-family-13b ](https://huggingface.co/inceptionai/jais-family-13b) \| 283B \| 141B \| 56B \| 480B \| \| [jais-family-6p7b](https://huggingface.co/inceptionai/jais-family-6p7b) \| 283B \| 141B \| 56B \| 480B \| \| [jais-family-2p7b](https://huggingface.co/inceptionai/jais-family-2p7b) \| 283B \| 141B \| 56B \| 480B \| \| [jais-family-1p3b](https://huggingface.co/inceptionai/jais-family-1p3b) \| 283B \| 141B \| 56B \| 480B \| \| [jais-family-590m](https://huggingface.co/inceptionai/jais-family-590m) \| 283B \| 141B \| 56B \| 480B \| \| [jais-adapted-70b](https://huggingface.co/inceptionai/jais-adapted-70b) \| 33B \| 334B \| 4B \| 371B \| \| [jais-adapted-13b](https://huggingface.co/inceptionai/jais-adapted-13b) \| 127B \| 140B \| 13B \| 280B \| \| [jais-adapted-7b](https://huggingface.co/inceptionai/jais-adapted-7b) \| 18B \| 19B \| 2B \| 39B \| ### Finetuning data <!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. --> All chat models in the Jais family are fine-tuned using Arabic and English prompt-response pairs in both single-turn and multi-turn settings. Data sources include open-source fine-tuning datasets filtered for topic and style diversity. Additionally, internally curated human data is incorporated to enhance cultural adaptation. This data is supplemented with content generated using synthetic methods including machine translation, distillation, and model self-chat. Overall, our updated instruction-tuning dataset comprises ~10M and ~4M prompt-response pairs in English and Arabic respectively. ### Training Procedure <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. --> During the pre-training of (`jais-family-`) models, documents are packed into sequences separated by EOS tokens, and the model is trained autoregressively, applying the loss to all tokens. For jais-30b models, the context length is progressively expanded from 2k to 8K to 16K by incorporating curated long-context documents in training. This progressive expansion leverages faster initial training at shorter context lengths, while gradually extending support for larger context lengths towards the end of the training process. During the adapted pre-training of the (`jais-adapted-`) models, we first initialize the new tokenizer and Arabic embeddings as described in [Model Architecture](#model-architecture). In training, we implemented a two-stage approach to overcome observed higher norms of the new Arabic embeddings. In the first stage, the backbone of the model is frozen, and the embeddings are trained using approximately 15 billion tokens from a bilingual corpus of English and Arabic. In the second stage, the backbone is unfrozen, and continuous pretraining is conducted with all parameters. During instruction tuning, each training example consists of a single-turn or multi-turn prompt and it's response. Instead of one example per sequence, examples are packed together while the loss is masked on the prompt tokens. This approach speeds up training by allowing more examples to be processed per batch. ### Training Hyperparameters: #### Jais-family-590m \| Hyperparameter \| Value \| \|----------------\|-------------------------------------------\| \| Precision \| fp32 \| \| Optimizer \| AdamW \| \| Learning rate \| 0 to 0.01563(<=163 warmup steps)<br>0.01563 to 4.21e-05(>163 and <=209422 steps) \| \| Weight decay \| 0.1 \| \| Batch size \| 1120\| \| Context Length \| 2048\| \| Steps \| 209422 \| ### Compute Infrastructure The training process was performed on the Condor Galaxy (CG) supercomputer platform. A CG contains 64 Cerebras CS-2 Wafer-Scale Engines (WSE-2) with 40 GB of SRAM, and achieves a total of 960 PetaFLOP/s. ## Evaluation <!-- This section describes the evaluation protocols and provides the results. --> We conducted a comprehensive evaluation of Jais models focusing on both English and Arabic, using LM-harness in a zero-shot setting. The evaluation criteria spanned various dimensions, including: - Knowledge: How well the model answers factual questions. - Reasoning: The model's ability to answer questions requiring reasoning. - Misinformation/Bias: Assessment of the model's susceptibility to generating false or misleading information, and its neutrality. ### Arabic evaluation results: <style> .table-container { overflow-x: auto; white-space: nowrap; } </style> <div class="table-container"> \| Models \| Avg \| ArabicMMLU\| MMLU \| EXAMS\| LitQA\| agqa \| agrc \| Hellaswag \| PIQA \| BoolQA \| Situated QA \| ARC-C \| OpenBookQA \| TruthfulQA \| CrowS-Pairs \| \|--------------------------\|-------\|------------\|-------\|-------\|-------\|------\|------\|------------\|------\|--------\|-------------\|-------\|------------\|------------\|-------------\| \| jais-family-30b-16k \| 49.2 \| 44.0 \| 33.4 \| 40.9 \| 60 \| 47.8 \| 49.3 \| 60.9 \| 68.6 \| 70.3 \| 41.6 \| 38.7 \| 31.8 \| 45.2 \| 57 \| \| jais-family-30b-8k \| 49.7 \| 46.0 \| 34 \| 42 \| 60.6 \| 47.6 \| 50.4 \| 60.4 \| 69 \| 67.7 \| 42.2 \| 39.2 \| 33.8 \| 45.1 \| 57.3 \| \| jais-family-13b \| 46.1 \| 34.0 \| 30.3 \| 42.7 \| 58.3 \| 40.5 \| 45.5 \| 57.3 \| 68.1 \| 63.1 \| 41.6 \| 35.3 \| 31.4 \| 41 \| 56.1 \| \| jais-family-6p7b \| 44.6 \| 32.2 \| 29.9 \| 39 \| 50.3 \| 39.2 \| 44.1 \| 54.3 \| 66.8 \| 66.5 \| 40.9 \| 33.5 \| 30.4 \| 41.2 \| 55.4 \| \| jais-family-2p7b \| 41.0 \| 29.5 \| 28.5 \| 36.1 \| 45.7 \| 32.4 \| 40.8 \| 44.2 \| 62.5 \| 62.2 \| 39.2 \| 27.4 \| 28.2 \| 43.6 \| 53.6 \| \| jais-family-1p3b \| 40.8 \| 28.9 \| 28.5 \| 34.2 \| 45.7 \| 32.4 \| 40.8 \| 44.2 \| 62.5 \| 62.2 \| 39.2 \| 27.4 \| 28.2 \| 43.6 \| 53.6 \| \| jais-family-590m \| 39.7 \| 31.2 \| 27 \| 33.1 \| 41.7 \| 33.8 \| 38.8 \| 38.2 \| 60.7 \| 62.2 \| 37.9 \| 25.5 \| 27.4 \| 44.7 \| 53.3 \| \| jais-family-30b-16k-chat \| 51.6 \| 59.9 \| 34.6 \| 40.2 \| 58.9 \| 46.8 \| 54.7 \| 56.2 \| 64.4 \| 76.7 \| 55.9 \| 40.8 \| 30.8 \| 49.5 \| 52.9 \| \| jais-family-30b-8k-chat \| 51.4 \| 61.2 \| 34.2 \| 40.2 \| 54.3 \| 47.3 \| 53.6 \| 60 \| 63.4 \| 76.8 \| 54.7 \| 39.5 \| 30 \| 50.7 \| 54.3 \| \| jais-family-13b-chat \| 50.3 \| 58.2 \| 33.9 \| 42.9 \| 53.1 \| 46.8 \| 51.7 \| 59.3 \| 65.4 \| 75.2 \| 51.2 \| 38.4 \| 29.8 \| 44.8 \| 53.8 \| \| jais-family-6p7b-chat \| 48.7 \| 55.7 \| 32.8 \| 37.7 \| 49.7 \| 40.5 \| 50.1 \| 56.2 \| 62.9 \| 79.4 \| 52 \| 38 \| 30.4 \| 44.7 \| 52 \| \| jais-family-2p7b-chat \| 45.6 \| 50.0 \| 31.5 \| 35.9 \| 41.1 \| 37.3 \| 42.1 \| 48.6 \| 63.7 \| 74.4 \| 50.9 \| 35.3 \| 31.2 \| 44.5 \| 51.3 \| \| jais-family-1p3b-chat \| 42.7 \| 42.2 \| 30.1 \| 33.6 \| 40.6 \| 34.1 \| 41.2 \| 43 \| 63.6 \| 69.3 \| 44.9 \| 31.6 \| 28 \| 45.6 \| 50.4 \| \| jais-family-590m-chat \| 37.8 \| 39.1 \| 28 \|29.5 \| 33.1 \| 30.8 \| 36.4 \| 30.3 \| 57.8 \| 57.2 \| 40.5 \| 25.9 \| 26.8 \| 44.5 \| 49.3 \| \| Adapted Models* \| Avg \| ArabicMMLU\| MMLU \| EXAMS\| LitQA\| agqa \| agrc \| Hellaswag \| PIQA \| BoolQA \| Situated QA \| ARC-C \| OpenBookQA \| TruthfulQA \| CrowS-Pairs \| \|--------------------------\|-------\|------------\|-------\|-------\|-------\|------\|------\|------------\|------\|--------\|-------------\|-------\|------------\|------------\|-------------\| \| jais-adapted-70b \| 51.5 \| 55.9 \| 36.8 \| 42.3 \| 58.3 \| 48.6 \| 54 \| 61.5 \| 68.4 \| 68.4 \| 42.1 \| 42.6 \| 33 \| 50.2 \| 58.3 \| \| jais-adapted-13b \| 46.6 \| 44.7 \| 30.6 \| 37.7 \| 54.3 \| 43.8 \| 48.3 \| 54.9 \| 67.1 \| 64.5 \| 40.6 \| 36.1 \| 32 \| 43.6 \| 54.00 \| \| jais-adapted-7b \| 42.0 \| 35.9 \| 28.9 \| 36.7 \| 46.3 \| 34.1 \| 40.3 \| 45 \| 61.3 \| 63.8 \| 38.1 \| 29.7 \| 30.2 \| 44.3 \| 53.6 \| \| jais-adapted-70b-chat \| 52.9 \| 66.8 \| 34.6 \| 42.5 \| 62.9 \| 36.8 \| 48.6 \| 64.5 \| 69.7 \| 82.8 \| 49.3 \| 44.2 \| 32.2 \| 53.3 \| 52.4 \| \| jais-adapted-13b-chat \| 50.3 \| 59.0 \| 31.7 \| 37.5 \| 56.6 \| 41.9 \| 51.7 \| 58.8 \| 67.1 \| 78.2 \| 45.9 \| 41 \| 34.2 \| 48.3 \| 52.1 \| \| jais-adapted-7b-chat \| 46.1 \| 51.3 \| 30 \| 37 \| 48 \| 36.8 \| 48.6 \| 51.1 \| 62.9 \| 72.4 \| 41.3 \| 34.6 \| 30.4 \| 48.6 \| 51.8 \| </div> Arabic benchmarks are translated using an in-house MT model and reviewed by Arabic linguists. Benchmarks labeled with an asterisk () are natively Arabic; for further details, see the [Jais paper](https://arxiv.org/abs/2308.16149). Additionally, we include [ArabicMMLU](https://arxiv.org/abs/2402.12840), a native Arabic benchmark based on regional knowledge. ### English evaluation results: <div class="table-container"> \| Models \| Avg \| MMLU \| RACE \| Hellaswag \| PIQA \| BoolQA \| SIQA \| ARC-Challenge \| OpenBookQA \| Winogrande \| TruthfulQA \| CrowS-Pairs \| \|--------------------------\|----------\|------\|------\|-----------\|------\|--------\|------\|---------------\|------------\|------------\|----------------\|-------------\| \| jais-family-30b-16k \| 59.3 \| 42.2 \| 40.5 \| 79.7 \| 80.6 \| 78.7 \| 48.8 \| 50.3 \| 44.2 \| 71.6 \| 43.5 \| 72.6 \| \| jais-family-30b-8k \| 58.8 \| 42.3 \| 40.3 \| 79.1 \| 80.5 \| 80.9 \| 49.3 \| 48.4 \| 43.2 \| 70.6 \| 40.3 \| 72.3 \| \| jais-family-13b \| 54.6 \| 32.3 \| 39 \| 72 \| 77.4 \| 73.9 \| 47.9 \| 43.2 \| 40 \| 67.1 \| 36.1 \| 71.7 \| \| jais-family-6p7b \| 53.1 \| 32 \| 38 \| 69.3 \| 76 \| 71.7 \| 47.1 \| 40.3 \| 37.4 \| 65.1 \| 34.4 \| 72.5 \| \| jais-family-2p7b \| 51 \| 29.4 \| 38 \| 62.7 \| 74.1 \| 67.4 \| 45.6 \| 35.1 \| 35.6 \| 62.9 \| 40.1 \| 70.2 \| \| jais-family-1p3b \| 48.7 \| 28.2 \| 35.4 \| 55.4 \| 72 \| 62.7 \| 44.9 \| 30.7 \| 36.2 \| 60.9 \| 40.4 \| 69 \| \| jais-family-590m \| 45.2 \| 27.8 \| 32.9 \| 46.1 \| 68.1 \| 60.4 \| 43.2 \| 25.6 \| 30.8 \| 55.8 \| 40.9 \| 65.3 \| \| jais-family-30b-16k-chat \| 58.8 \| 42 \| 41.1 \| 76.2 \| 73.3 \| 84.6 \| 60.3 \| 48.4 \| 40.8 \| 68.2 \| 44.8 \| 67 \| \| jais-family-30b-8k-chat \| 60.3 \| 40.6 \| 47.1 \| 78.9 \| 72.7 \| 90.6 \| 60 \| 50.1 \| 43.2 \| 70.6 \| 44.9 \| 64.2 \| \| jais-family-13b-chat \| 57.5 \| 36.6 \| 42.6 \| 75 \| 75.8 \| 87.6 \| 54.4 \| 47.9 \| 42 \| 65 \| 40.6 \| 64.5 \| \| jais-family-6p7b-chat \| 56 \| 36.6 \| 41.3 \| 72 \| 74 \| 86.9 \| 55.4 \| 44.6 \| 40 \| 62.4 \| 41 \| 62.2 \| \| jais-family-2p7b-chat \| 52.8 \| 32.7 \| 40.4 \| 62.2 \| 71 \| 84.1 \| 54 \| 37.2 \| 36.8 \| 61.4 \| 40.9 \| 59.8 \| \| jais-family-1p3b-chat \| 49.3 \| 31.9 \| 37.4 \| 54.5 \| 70.2 \| 77.8 \| 49.8 \| 34.4 \| 35.6 \| 52.7 \| 37.2 \| 60.8 \| \| jais-family-590m-chat \| 42.6 \| 27.9 \| 33.4 \| 33.1 \| 63.7 \| 60.1 \| 45.3 \| 26.7 \| 25.8 \| 50.5 \| 44.5 \| 57.7 \| </div> <div class="table-container"> \|Adapted Models\| Avg \| MMLU \| RACE \| Hellaswag \| PIQA \| BoolQA \| SIQA \| ARC-Challenge \| OpenBookQA \| Winogrande \| TruthfulQA \| CrowS-Pairs \| \|--------------------------\|----------\|------\|------\|-----------\|------\|--------\|------\|---------------\|------------\|------------\|----------------\|-------------\| \| jais-adapted-70b \| 60.1 \| 40.4 \| 38.5 \| 81.2 \| 81.1 \| 81.2 \| 48.1 \| 50.4 \| 45 \| 75.8 \| 45.7 \| 74 \| \| jais-adapted-13b \| 56 \| 33.8 \| 39.5 \| 76.5 \| 78.6 \| 77.8 \| 44.6 \| 45.9 \| 44.4 \| 71.4 \| 34.6 \| 69 \| \| jais-adapted-7b \| 55.7 \| 32.2 \| 39.8 \| 75.3 \| 78.8 \| 75.7 \| 45.2 \| 42.8 \| 43 \| 68 \| 38.3 \| 73.1 \| \| jais-adapted-70b-chat \| 61.4 \| 38.7 \| 42.9 \| 82.7 \| 81.2 \| 89.6 \| 52.9 \| 54.9 \| 44.4 \| 75.7 \| 44 \| 68.8 \| \| jais-adapted-13b-chat \| 58.5 \| 34.9 \| 42.4 \| 79.6 \| 79.7 \| 88.2 \| 50.5 \| 48.5 \| 42.4 \| 70.3 \| 42.2 \| 65.1 \| \| jais-adapted-7b-chat \| 58.5 \| 33.8 \| 43.9 \| 77.8 \| 79.4 \| 87.1 \| 47.3 \| 46.9 \| 43.4 \| 69.9 \| 42 \| 72.4 \| </div> ### GPT-4 evaluation In addition to the LM-Harness evaluation, we conducted an open-ended generation evaluation using GPT-4-as-a-judge. We measured pairwise win-rates of model responses in both Arabic and English on a fixed set of 80 prompts from the Vicuna test set. English prompts were translated to Arabic by our in-house linguists. In the following, we compare the models in this release of the jais family against previously released versions: <p align="center"> <img src="https://huggingface.co/inceptionai/JaisFamilySupplmentary/resolve/main/jais.png" alt="Jais-adapted GPT-4"> </p> <p align="center"> <em>GPT-4-as-a-judge evaluation of Jais in Arabic and English. Jais family models are significantly better than previous Jais at generations in both languages. </em> </p> <p align="center"> <img src="https://huggingface.co/inceptionai/JaisFamilySupplmentary/resolve/main/jais-adapted.png" alt="Jais-adapted GPT-4"> </p> <p align="center"> <em>GPT-4-as-a-judge evaluation of adapted Jais in Arabic and English. The generation quality of Arabic is significantly enhanced, while achieving improvement in English when compared to Llama-2 instruct. </em> </p> Besides pairwise comparison, we also perform MT-bench style single-answer grading on a scale of 1 to 10. <p align="center"> <img src="https://huggingface.co/inceptionai/JaisFamilySupplmentary/resolve/main/mt_bench.png" alt="MT-bench"> </p> <p align="center"> <em>MT-bench style single-answer grading evaluation of Jais and adapted Jais in Arabic and English. Comparisons are made between select corresponding models from earlier releases. The quality ratings of responses are generally improved, with significant enhancements in Arabic.</em> </p> ## Intended use We release the Jais family of models under a full open-source license. We welcome all feedback and opportunities to collaborate. Spanning sizes from 590M to 70B parameters, this suite of bilingual models accommodates a wide range of use cases. Some potential downstream applications include: - Research: The Jais family serves Arabic researchers and NLP practitioners, offering both compute-efficient and advanced model sizes - Natural language understanding and generation tasks. - Mechanistic interpretability analyses on cultural alignment in bilingual pre-trained and adapted pre-trained models. - Quantitative studies of Arabic cultural and linguistic phenomena. - Commercial Use: Jais 30B and 70B chat models are well-suited for direct use in chat applications with appropriate prompting or for further fine-tuning on specific tasks. - Development of chat assistants for Arabic-speaking users. - Sentiment analysis to gain insights into local markets and customer trends. - Summarization of bilingual Arabic-English documents. Audiences that we hope will benefit from our model: - Academics: For those researching Arabic Natural Language Processing. - Businesses: Companies targeting Arabic-speaking audiences. - Developers: Those integrating Arabic language capabilities in applications. ### Out-of-Scope Use <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. --> While the Jais family of models are powerful Arabic and English bilingual models, it's essential to understand their limitations and the potential of misuse. It is prohibited to use the model in any manner that violates applicable laws or regulations. The following are some example scenarios where the model should not be used. - Malicious Use: The model should not be used to generate harmful, misleading, or inappropriate content. Thisincludes but is not limited to: - Generating or promoting hate speech, violence, or discrimination. - Spreading misinformation or fake news. - Engaging in or promoting illegal activities. - Sensitive Information: The model should not be used to handle or generate personal, confidential, or sensitive information. - Generalization Across All Languages: Jais family of models are bilingual and optimized for Arabic and English. They should not be presumed to have equal proficiency in other languages or dialects. - High-Stakes Decisions: The model should not be used to make high-stakes decisions without human oversight. This includes medical, legal, financial, or safety-critical decisions. ## Bias, Risks, and Limitations <!-- This section is meant to convey both technical and sociotechnical limitations. --> The Jais family is trained on publicly available data which was in part curated by Inception. We have employed different techniques to reduce bias in the model. While efforts have been made to minimize biases, it is likely that the model, as with all LLM models, will exhibit some bias. The fine-tuned variants are trained as an AI assistant for Arabic and English speakers. Chat models are limited to produce responses for queries in these two languages and may not produce appropriate responses to other language queries. By using Jais, you acknowledge and accept that, as with any large language model, it may generate incorrect, misleading and/or offensive information or content. The information is not intended as advice and should not be relied upon in any way, nor are we responsible for any of the content or consequences resulting from its use. We are continuously working to develop models with greater capabilities, and as such, welcome any feedback on the model. Copyright Inception Institute of Artificial Intelligence Ltd. JAIS is made available under the Apache License, Version 2.0 (the “License”). You shall not use JAIS except in compliance with the License. You may obtain a copy of the License at https://www.apache.org/licenses/LICENSE-2.0. Unless required by applicable law or agreed to in writing, JAIS is distributed on an AS IS basis, without warranties or conditions of any kind, either express or implied. Please see the terms of the License for the specific language permissions and limitations under the License. #### Summary We release the Jais family of Arabic and English bilingual models. The wide range of pre-trained model sizes, the recipe for adapting English-centric models to Arabic, and the fine-tuning of all sizes unlocks numerous use cases commercially and academically in the Arabic setting. Through this release, we aim to make LLMs more accessible to Arabic NLP researchers and companies, offering native Arabic models that provide better cultural understanding than English centric ones. The strategies we employ for pre-training, fine-tuning and adaptation to Arabic are extensible to other low and medium resource languages, paving the way for language-focused and accessible models that cater to local contexts. #### Citation info ```bibtex @misc{sengupta2023jais, title={Jais and Jais-chat: Arabic-Centric Foundation and Instruction-Tuned Open Generative Large Language Models}, author={Neha Sengupta, Sunil Kumar Sahu, Bokang Jia, Satheesh Katipomu, Haonan Li, Fajri Koto, William Marshall, Gurpreet Gosal, Cynthia Liu, Zhiming Chen, Osama Mohammed Afzal, Samta Kamboj, Onkar Pandit, Rahul Pal, Lalit Pradhan, Zain Muhammad Mujahid, Massa Baali, Xudong Han, Sondos Mahmoud Bsharat, Alham Fikri Aji, Zhiqiang Shen, Zhengzhong Liu, Natalia Vassilieva, Joel Hestness, Andy Hock, Andrew Feldman, Jonathan Lee, Andrew Jackson, Hector Xuguang Ren, Preslav Nakov, Timothy Baldwin and Eric Xing}, year={2023}, eprint={2308.16149}, archivePrefix={arXiv}, primaryClass={cs.CL} } @article{jaisfamilymodelcard, title={Jais Family Model Card}, author={Inception}, year={2024}, url = {https://huggingface.co/inceptionai/jais-family-30b-16k-chat/blob/main/README.md} } ```	null	Non_BioNLP	# Jais Family Model Card The Jais family of models is a comprehensive series of bilingual English-Arabic large language models (LLMs). These models are optimized to excel in Arabic while having strong English capabilities. We release two variants of foundation models that include: - Models pre-trained from scratch (`jais-family-`). - Models pre-trained adaptively from [Llama-2](https://arxiv.org/pdf/2307.09288)* (`jais-adapted-`). In this release, we introduce 20 models across 8 sizes, ranging from 590M to 70B parameters, trained on up to 1.6T tokens of Arabic, English, and code data. All* pre-trained models in this series are instruction fine-tuned (`-chat`) for dialog using a curated mix of Arabic and English instruction data. We hope this extensive release will accelerate research in Arabic NLP, and enable numerous downstream applications for the Arabic speaking and bilingual community. The training and adaptation techniques we demonstrate successfully for Arabic models are extensible to other low and medium resource languages. ## Jais Family Details - Developed by:* Inception, Cerebras Systems. - Language(s): (NLP): Arabic (MSA) and English. - Input: Text only data. - Output: Model generates text. - Model Sizes: 590M, 1.3B, 2.7B, 6.7B, 7B, 13B, 30B, 70B. - Demo: [Access the live demo here](https://arabic-gpt.ai/) - License: Apache 2.0 \| Pre-trained Model \| Fine-tuned Model \| Size (Parameters) \| Context length (Tokens) \| \|:---------------------\|:--------\|:-------\|:-------\| \| [jais-family-30b-16k](https://huggingface.co/inceptionai/jais-family-30b-16k) \| [Jais-family-30b-16k-chat](https://huggingface.co/inceptionai/jais-family-30b-16k-chat) \| 30B \| 16,384 \| \| [jais-family-30b-8k](https://huggingface.co/inceptionai/jais-family-30b-8k) \| [Jais-family-30b-8k-chat](https://huggingface.co/inceptionai/jais-family-30b-8k-chat) \| 30B \| 8,192 \| \| [jais-family-13b ](https://huggingface.co/inceptionai/jais-family-13b) \| [Jais-family-13b-chat](https://huggingface.co/inceptionai/jais-family-13b-chat) \| 13B \| 2,048 \| \| [jais-family-6p7b](https://huggingface.co/inceptionai/jais-family-6p7b) \| [Jais-family-6p7b-chat](https://huggingface.co/inceptionai/jais-family-6p7b-chat) \| 6.7B \| 2,048 \| \| [jais-family-2p7b](https://huggingface.co/inceptionai/jais-family-2p7b) \| [Jais-family-2p7b-chat](https://huggingface.co/inceptionai/jais-family-2p7b-chat) \| 2.7B \| 2,048 \| \| [jais-family-1p3b](https://huggingface.co/inceptionai/jais-family-1p3b) \| [Jais-family-1p3b-chat](https://huggingface.co/inceptionai/jais-family-1p3b-chat) \| 1.3B \| 2,048 \| \| [jais-family-590m](https://huggingface.co/inceptionai/jais-family-590m) \| [Jais-family-590m-chat](https://huggingface.co/inceptionai/jais-family-590m-chat) \| 590M \| 2,048 \| \| Adapted pre-trained Model \| Fine-tuned Model \| Size (Parameters) \| Context length (Tokens) \| \|:---------------------\|:--------\|:-------\|:-------\| \| [jais-adapted-70b](https://huggingface.co/inceptionai/jais-adapted-70b) \| [Jais-adapted-70b-chat](https://huggingface.co/inceptionai/jais-adapted-70b-chat) \| 70B \| 4,096 \| \| [jais-adapted-13b](https://huggingface.co/inceptionai/jais-adapted-13b) \| [Jais-adapted-13b-chat](https://huggingface.co/inceptionai/jais-adapted-13b-chat) \| 13B \| 4,096 \| \| [jais-adapted-7b](https://huggingface.co/inceptionai/jais-adapted-7b) \| [Jais-adapted-7b-chat](https://huggingface.co/inceptionai/jais-adapted-7b-chat) \| 7B \| 4,096 \| ### Model Architecture: <a name="model-architecture"></a> All models in this family are auto-regressive language models that use a transformer-based, decoder-only architecture (GPT-3). Jais models (`jais-family-`) are trained from scratch, incorporating the SwiGLU non-linear activation function and ALiBi position encoding. These architectural enhancements allow the models to extrapolate at long sequence lengths, leading to improved context handling and precision. Jais adapted models (`jais-adapted-`) are built on top of Llama-2, which employs RoPE position embedding and Grouped Query Attention. We introduce tokenizer expansion with Arabic data, which improves fertility and compute efficiency by over 3x. In particular, we add `32,000` new Arabic tokens from the Jais-30b vocabulary into the Llama-2 tokenizer. To initialize these new Arabic token embeddings we first learn a linear projection from the embedding space of Jais-30b to Llama's embedding space, using the set of shared English tokens present in both vocabularies. Next, this learned projection is applied to transform the existing Jais-30b Arabic embeddings into the Llama-2 embedding space. ## Getting started Below is sample code to use the model. Note that the model requires a custom model class, so users must enable `trust_remote_code=True` while loading the model. ```python # -- coding: utf-8 -- import torch from transformers import AutoTokenizer, AutoModelForCausalLM model_path = "inceptionai/jais-family-590m" device = "cuda" if torch.cuda.is_available() else "cpu" tokenizer = AutoTokenizer.from_pretrained(model_path) model = AutoModelForCausalLM.from_pretrained(model_path, device_map="auto", trust_remote_code=True) def get_response(text, tokenizer=tokenizer, model=model): input_ids = tokenizer(text, return_tensors="pt").input_ids inputs = input_ids.to(device) input_len = inputs.shape[-1] generate_ids = model.generate( inputs, top_p=0.9, temperature=0.3, max_length=2048, min_length=input_len + 4, repetition_penalty=1.2, do_sample=True, ) response = tokenizer.batch_decode( generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True )[0] return response text = "عاصمة دولة الإمارات العربية المتحدة ه" print(get_response(text)) text = "The capital of UAE is" print(get_response(text)) ``` ## Training Details ### Pretraining Data The Jais family of models are trained on up to 1.6 Trillion tokens of diverse English, Arabic and Code data. The data consists of the following sources: - Web: We used publicly available web pages, wikipedia articles, news articles, and social network content in both Arabic and English. - Code: To enhance the reasoning capability of our model, we include Code data in various programming languages. - Books: We used a selection of publicly available Arabic and English books data, which improves long-range context modelling and coherent storytelling. - Scientific: A subset of ArXiv papers were included to improve reasoning and long context abilities. - Synthetic: We augment the volume of Arabic data by translating English to Arabic using an in-house machine translation system. We restrict this to high quality English resources such as English Wikipedia and English books. We extensively preprocess and deduplicate the training data. For Arabic, we used a custom preprocessing pipeline to filter for data with high linguistic quality. More information on this pipeline can be found in the [Jais paper](https://arxiv.org/abs/2308.16149). - Jais pre-trained (`jais-family-`): Following our previous experimentation with language alignment mixing in [Jais](https://arxiv.org/abs/2308.16149), we used a ratio of 1:2:0.4 of Arabic:English:Code data. This recipe for <u>from scratch pre-training</u> addresses Arabic data scarcity while improving performance in both languages. - Jais adapted pre-trained* (`jais-adapted-`): For the <u>adapted pre-training of Llama-2</u>, we utilized a larger Arabic dataset of ~334B Arabic tokens mixed with English and Code data. We vary the mixing ratio, at different model sizes, to introduce strong Arabic capabilities while maintaining performance in English. \| Pre-trained model* \| English data (tokens) \| Arabic data (tokens) \| Code data (tokens) \| Total data (tokens) \| \|-------------------------\|---------------------------\|--------------------------\|------------------------\|------------------------\| \| [jais-family-30b-16k](https://huggingface.co/inceptionai/jais-family-30b-16k) \| 980B \| 490B \| 196B \| 1666B \| \| [jais-family-30b-8k](https://huggingface.co/inceptionai/jais-family-30b-8k) \| 882B \| 441B \| 177B \| 1500B \| \| [jais-family-13b ](https://huggingface.co/inceptionai/jais-family-13b) \| 283B \| 141B \| 56B \| 480B \| \| [jais-family-6p7b](https://huggingface.co/inceptionai/jais-family-6p7b) \| 283B \| 141B \| 56B \| 480B \| \| [jais-family-2p7b](https://huggingface.co/inceptionai/jais-family-2p7b) \| 283B \| 141B \| 56B \| 480B \| \| [jais-family-1p3b](https://huggingface.co/inceptionai/jais-family-1p3b) \| 283B \| 141B \| 56B \| 480B \| \| [jais-family-590m](https://huggingface.co/inceptionai/jais-family-590m) \| 283B \| 141B \| 56B \| 480B \| \| [jais-adapted-70b](https://huggingface.co/inceptionai/jais-adapted-70b) \| 33B \| 334B \| 4B \| 371B \| \| [jais-adapted-13b](https://huggingface.co/inceptionai/jais-adapted-13b) \| 127B \| 140B \| 13B \| 280B \| \| [jais-adapted-7b](https://huggingface.co/inceptionai/jais-adapted-7b) \| 18B \| 19B \| 2B \| 39B \| ### Finetuning data <!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. --> All chat models in the Jais family are fine-tuned using Arabic and English prompt-response pairs in both single-turn and multi-turn settings. Data sources include open-source fine-tuning datasets filtered for topic and style diversity. Additionally, internally curated human data is incorporated to enhance cultural adaptation. This data is supplemented with content generated using synthetic methods including machine translation, distillation, and model self-chat. Overall, our updated instruction-tuning dataset comprises ~10M and ~4M prompt-response pairs in English and Arabic respectively. ### Training Procedure <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. --> During the pre-training of (`jais-family-`) models, documents are packed into sequences separated by EOS tokens, and the model is trained autoregressively, applying the loss to all tokens. For jais-30b models, the context length is progressively expanded from 2k to 8K to 16K by incorporating curated long-context documents in training. This progressive expansion leverages faster initial training at shorter context lengths, while gradually extending support for larger context lengths towards the end of the training process. During the adapted pre-training of the (`jais-adapted-`) models, we first initialize the new tokenizer and Arabic embeddings as described in [Model Architecture](#model-architecture). In training, we implemented a two-stage approach to overcome observed higher norms of the new Arabic embeddings. In the first stage, the backbone of the model is frozen, and the embeddings are trained using approximately 15 billion tokens from a bilingual corpus of English and Arabic. In the second stage, the backbone is unfrozen, and continuous pretraining is conducted with all parameters. During instruction tuning, each training example consists of a single-turn or multi-turn prompt and it's response. Instead of one example per sequence, examples are packed together while the loss is masked on the prompt tokens. This approach speeds up training by allowing more examples to be processed per batch. ### Training Hyperparameters: #### Jais-family-590m \| Hyperparameter \| Value \| \|----------------\|-------------------------------------------\| \| Precision \| fp32 \| \| Optimizer \| AdamW \| \| Learning rate \| 0 to 0.01563(<=163 warmup steps)<br>0.01563 to 4.21e-05(>163 and <=209422 steps) \| \| Weight decay \| 0.1 \| \| Batch size \| 1120\| \| Context Length \| 2048\| \| Steps \| 209422 \| ### Compute Infrastructure The training process was performed on the Condor Galaxy (CG) supercomputer platform. A CG contains 64 Cerebras CS-2 Wafer-Scale Engines (WSE-2) with 40 GB of SRAM, and achieves a total of 960 PetaFLOP/s. ## Evaluation <!-- This section describes the evaluation protocols and provides the results. --> We conducted a comprehensive evaluation of Jais models focusing on both English and Arabic, using LM-harness in a zero-shot setting. The evaluation criteria spanned various dimensions, including: - Knowledge: How well the model answers factual questions. - Reasoning: The model's ability to answer questions requiring reasoning. - Misinformation/Bias: Assessment of the model's susceptibility to generating false or misleading information, and its neutrality. ### Arabic evaluation results: <style> .table-container { overflow-x: auto; white-space: nowrap; } </style> <div class="table-container"> \| Models \| Avg \| ArabicMMLU\| MMLU \| EXAMS\| LitQA\| agqa \| agrc \| Hellaswag \| PIQA \| BoolQA \| Situated QA \| ARC-C \| OpenBookQA \| TruthfulQA \| CrowS-Pairs \| \|--------------------------\|-------\|------------\|-------\|-------\|-------\|------\|------\|------------\|------\|--------\|-------------\|-------\|------------\|------------\|-------------\| \| jais-family-30b-16k \| 49.2 \| 44.0 \| 33.4 \| 40.9 \| 60 \| 47.8 \| 49.3 \| 60.9 \| 68.6 \| 70.3 \| 41.6 \| 38.7 \| 31.8 \| 45.2 \| 57 \| \| jais-family-30b-8k \| 49.7 \| 46.0 \| 34 \| 42 \| 60.6 \| 47.6 \| 50.4 \| 60.4 \| 69 \| 67.7 \| 42.2 \| 39.2 \| 33.8 \| 45.1 \| 57.3 \| \| jais-family-13b \| 46.1 \| 34.0 \| 30.3 \| 42.7 \| 58.3 \| 40.5 \| 45.5 \| 57.3 \| 68.1 \| 63.1 \| 41.6 \| 35.3 \| 31.4 \| 41 \| 56.1 \| \| jais-family-6p7b \| 44.6 \| 32.2 \| 29.9 \| 39 \| 50.3 \| 39.2 \| 44.1 \| 54.3 \| 66.8 \| 66.5 \| 40.9 \| 33.5 \| 30.4 \| 41.2 \| 55.4 \| \| jais-family-2p7b \| 41.0 \| 29.5 \| 28.5 \| 36.1 \| 45.7 \| 32.4 \| 40.8 \| 44.2 \| 62.5 \| 62.2 \| 39.2 \| 27.4 \| 28.2 \| 43.6 \| 53.6 \| \| jais-family-1p3b \| 40.8 \| 28.9 \| 28.5 \| 34.2 \| 45.7 \| 32.4 \| 40.8 \| 44.2 \| 62.5 \| 62.2 \| 39.2 \| 27.4 \| 28.2 \| 43.6 \| 53.6 \| \| jais-family-590m \| 39.7 \| 31.2 \| 27 \| 33.1 \| 41.7 \| 33.8 \| 38.8 \| 38.2 \| 60.7 \| 62.2 \| 37.9 \| 25.5 \| 27.4 \| 44.7 \| 53.3 \| \| jais-family-30b-16k-chat \| 51.6 \| 59.9 \| 34.6 \| 40.2 \| 58.9 \| 46.8 \| 54.7 \| 56.2 \| 64.4 \| 76.7 \| 55.9 \| 40.8 \| 30.8 \| 49.5 \| 52.9 \| \| jais-family-30b-8k-chat \| 51.4 \| 61.2 \| 34.2 \| 40.2 \| 54.3 \| 47.3 \| 53.6 \| 60 \| 63.4 \| 76.8 \| 54.7 \| 39.5 \| 30 \| 50.7 \| 54.3 \| \| jais-family-13b-chat \| 50.3 \| 58.2 \| 33.9 \| 42.9 \| 53.1 \| 46.8 \| 51.7 \| 59.3 \| 65.4 \| 75.2 \| 51.2 \| 38.4 \| 29.8 \| 44.8 \| 53.8 \| \| jais-family-6p7b-chat \| 48.7 \| 55.7 \| 32.8 \| 37.7 \| 49.7 \| 40.5 \| 50.1 \| 56.2 \| 62.9 \| 79.4 \| 52 \| 38 \| 30.4 \| 44.7 \| 52 \| \| jais-family-2p7b-chat \| 45.6 \| 50.0 \| 31.5 \| 35.9 \| 41.1 \| 37.3 \| 42.1 \| 48.6 \| 63.7 \| 74.4 \| 50.9 \| 35.3 \| 31.2 \| 44.5 \| 51.3 \| \| jais-family-1p3b-chat \| 42.7 \| 42.2 \| 30.1 \| 33.6 \| 40.6 \| 34.1 \| 41.2 \| 43 \| 63.6 \| 69.3 \| 44.9 \| 31.6 \| 28 \| 45.6 \| 50.4 \| \| jais-family-590m-chat \| 37.8 \| 39.1 \| 28 \|29.5 \| 33.1 \| 30.8 \| 36.4 \| 30.3 \| 57.8 \| 57.2 \| 40.5 \| 25.9 \| 26.8 \| 44.5 \| 49.3 \| \| Adapted Models* \| Avg \| ArabicMMLU\| MMLU \| EXAMS\| LitQA\| agqa \| agrc \| Hellaswag \| PIQA \| BoolQA \| Situated QA \| ARC-C \| OpenBookQA \| TruthfulQA \| CrowS-Pairs \| \|--------------------------\|-------\|------------\|-------\|-------\|-------\|------\|------\|------------\|------\|--------\|-------------\|-------\|------------\|------------\|-------------\| \| jais-adapted-70b \| 51.5 \| 55.9 \| 36.8 \| 42.3 \| 58.3 \| 48.6 \| 54 \| 61.5 \| 68.4 \| 68.4 \| 42.1 \| 42.6 \| 33 \| 50.2 \| 58.3 \| \| jais-adapted-13b \| 46.6 \| 44.7 \| 30.6 \| 37.7 \| 54.3 \| 43.8 \| 48.3 \| 54.9 \| 67.1 \| 64.5 \| 40.6 \| 36.1 \| 32 \| 43.6 \| 54.00 \| \| jais-adapted-7b \| 42.0 \| 35.9 \| 28.9 \| 36.7 \| 46.3 \| 34.1 \| 40.3 \| 45 \| 61.3 \| 63.8 \| 38.1 \| 29.7 \| 30.2 \| 44.3 \| 53.6 \| \| jais-adapted-70b-chat \| 52.9 \| 66.8 \| 34.6 \| 42.5 \| 62.9 \| 36.8 \| 48.6 \| 64.5 \| 69.7 \| 82.8 \| 49.3 \| 44.2 \| 32.2 \| 53.3 \| 52.4 \| \| jais-adapted-13b-chat \| 50.3 \| 59.0 \| 31.7 \| 37.5 \| 56.6 \| 41.9 \| 51.7 \| 58.8 \| 67.1 \| 78.2 \| 45.9 \| 41 \| 34.2 \| 48.3 \| 52.1 \| \| jais-adapted-7b-chat \| 46.1 \| 51.3 \| 30 \| 37 \| 48 \| 36.8 \| 48.6 \| 51.1 \| 62.9 \| 72.4 \| 41.3 \| 34.6 \| 30.4 \| 48.6 \| 51.8 \| </div> Arabic benchmarks are translated using an in-house MT model and reviewed by Arabic linguists. Benchmarks labeled with an asterisk () are natively Arabic; for further details, see the [Jais paper](https://arxiv.org/abs/2308.16149). Additionally, we include [ArabicMMLU](https://arxiv.org/abs/2402.12840), a native Arabic benchmark based on regional knowledge. ### English evaluation results: <div class="table-container"> \| Models \| Avg \| MMLU \| RACE \| Hellaswag \| PIQA \| BoolQA \| SIQA \| ARC-Challenge \| OpenBookQA \| Winogrande \| TruthfulQA \| CrowS-Pairs \| \|--------------------------\|----------\|------\|------\|-----------\|------\|--------\|------\|---------------\|------------\|------------\|----------------\|-------------\| \| jais-family-30b-16k \| 59.3 \| 42.2 \| 40.5 \| 79.7 \| 80.6 \| 78.7 \| 48.8 \| 50.3 \| 44.2 \| 71.6 \| 43.5 \| 72.6 \| \| jais-family-30b-8k \| 58.8 \| 42.3 \| 40.3 \| 79.1 \| 80.5 \| 80.9 \| 49.3 \| 48.4 \| 43.2 \| 70.6 \| 40.3 \| 72.3 \| \| jais-family-13b \| 54.6 \| 32.3 \| 39 \| 72 \| 77.4 \| 73.9 \| 47.9 \| 43.2 \| 40 \| 67.1 \| 36.1 \| 71.7 \| \| jais-family-6p7b \| 53.1 \| 32 \| 38 \| 69.3 \| 76 \| 71.7 \| 47.1 \| 40.3 \| 37.4 \| 65.1 \| 34.4 \| 72.5 \| \| jais-family-2p7b \| 51 \| 29.4 \| 38 \| 62.7 \| 74.1 \| 67.4 \| 45.6 \| 35.1 \| 35.6 \| 62.9 \| 40.1 \| 70.2 \| \| jais-family-1p3b \| 48.7 \| 28.2 \| 35.4 \| 55.4 \| 72 \| 62.7 \| 44.9 \| 30.7 \| 36.2 \| 60.9 \| 40.4 \| 69 \| \| jais-family-590m \| 45.2 \| 27.8 \| 32.9 \| 46.1 \| 68.1 \| 60.4 \| 43.2 \| 25.6 \| 30.8 \| 55.8 \| 40.9 \| 65.3 \| \| jais-family-30b-16k-chat \| 58.8 \| 42 \| 41.1 \| 76.2 \| 73.3 \| 84.6 \| 60.3 \| 48.4 \| 40.8 \| 68.2 \| 44.8 \| 67 \| \| jais-family-30b-8k-chat \| 60.3 \| 40.6 \| 47.1 \| 78.9 \| 72.7 \| 90.6 \| 60 \| 50.1 \| 43.2 \| 70.6 \| 44.9 \| 64.2 \| \| jais-family-13b-chat \| 57.5 \| 36.6 \| 42.6 \| 75 \| 75.8 \| 87.6 \| 54.4 \| 47.9 \| 42 \| 65 \| 40.6 \| 64.5 \| \| jais-family-6p7b-chat \| 56 \| 36.6 \| 41.3 \| 72 \| 74 \| 86.9 \| 55.4 \| 44.6 \| 40 \| 62.4 \| 41 \| 62.2 \| \| jais-family-2p7b-chat \| 52.8 \| 32.7 \| 40.4 \| 62.2 \| 71 \| 84.1 \| 54 \| 37.2 \| 36.8 \| 61.4 \| 40.9 \| 59.8 \| \| jais-family-1p3b-chat \| 49.3 \| 31.9 \| 37.4 \| 54.5 \| 70.2 \| 77.8 \| 49.8 \| 34.4 \| 35.6 \| 52.7 \| 37.2 \| 60.8 \| \| jais-family-590m-chat \| 42.6 \| 27.9 \| 33.4 \| 33.1 \| 63.7 \| 60.1 \| 45.3 \| 26.7 \| 25.8 \| 50.5 \| 44.5 \| 57.7 \| </div> <div class="table-container"> \|Adapted Models\| Avg \| MMLU \| RACE \| Hellaswag \| PIQA \| BoolQA \| SIQA \| ARC-Challenge \| OpenBookQA \| Winogrande \| TruthfulQA \| CrowS-Pairs \| \|--------------------------\|----------\|------\|------\|-----------\|------\|--------\|------\|---------------\|------------\|------------\|----------------\|-------------\| \| jais-adapted-70b \| 60.1 \| 40.4 \| 38.5 \| 81.2 \| 81.1 \| 81.2 \| 48.1 \| 50.4 \| 45 \| 75.8 \| 45.7 \| 74 \| \| jais-adapted-13b \| 56 \| 33.8 \| 39.5 \| 76.5 \| 78.6 \| 77.8 \| 44.6 \| 45.9 \| 44.4 \| 71.4 \| 34.6 \| 69 \| \| jais-adapted-7b \| 55.7 \| 32.2 \| 39.8 \| 75.3 \| 78.8 \| 75.7 \| 45.2 \| 42.8 \| 43 \| 68 \| 38.3 \| 73.1 \| \| jais-adapted-70b-chat \| 61.4 \| 38.7 \| 42.9 \| 82.7 \| 81.2 \| 89.6 \| 52.9 \| 54.9 \| 44.4 \| 75.7 \| 44 \| 68.8 \| \| jais-adapted-13b-chat \| 58.5 \| 34.9 \| 42.4 \| 79.6 \| 79.7 \| 88.2 \| 50.5 \| 48.5 \| 42.4 \| 70.3 \| 42.2 \| 65.1 \| \| jais-adapted-7b-chat \| 58.5 \| 33.8 \| 43.9 \| 77.8 \| 79.4 \| 87.1 \| 47.3 \| 46.9 \| 43.4 \| 69.9 \| 42 \| 72.4 \| </div> ### GPT-4 evaluation In addition to the LM-Harness evaluation, we conducted an open-ended generation evaluation using GPT-4-as-a-judge. We measured pairwise win-rates of model responses in both Arabic and English on a fixed set of 80 prompts from the Vicuna test set. English prompts were translated to Arabic by our in-house linguists. In the following, we compare the models in this release of the jais family against previously released versions: <p align="center"> <img src="https://huggingface.co/inceptionai/JaisFamilySupplmentary/resolve/main/jais.png" alt="Jais-adapted GPT-4"> </p> <p align="center"> <em>GPT-4-as-a-judge evaluation of Jais in Arabic and English. Jais family models are significantly better than previous Jais at generations in both languages. </em> </p> <p align="center"> <img src="https://huggingface.co/inceptionai/JaisFamilySupplmentary/resolve/main/jais-adapted.png" alt="Jais-adapted GPT-4"> </p> <p align="center"> <em>GPT-4-as-a-judge evaluation of adapted Jais in Arabic and English. The generation quality of Arabic is significantly enhanced, while achieving improvement in English when compared to Llama-2 instruct. </em> </p> Besides pairwise comparison, we also perform MT-bench style single-answer grading on a scale of 1 to 10. <p align="center"> <img src="https://huggingface.co/inceptionai/JaisFamilySupplmentary/resolve/main/mt_bench.png" alt="MT-bench"> </p> <p align="center"> <em>MT-bench style single-answer grading evaluation of Jais and adapted Jais in Arabic and English. Comparisons are made between select corresponding models from earlier releases. The quality ratings of responses are generally improved, with significant enhancements in Arabic.</em> </p> ## Intended use We release the Jais family of models under a full open-source license. We welcome all feedback and opportunities to collaborate. Spanning sizes from 590M to 70B parameters, this suite of bilingual models accommodates a wide range of use cases. Some potential downstream applications include: - Research: The Jais family serves Arabic researchers and NLP practitioners, offering both compute-efficient and advanced model sizes - Natural language understanding and generation tasks. - Mechanistic interpretability analyses on cultural alignment in bilingual pre-trained and adapted pre-trained models. - Quantitative studies of Arabic cultural and linguistic phenomena. - Commercial Use: Jais 30B and 70B chat models are well-suited for direct use in chat applications with appropriate prompting or for further fine-tuning on specific tasks. - Development of chat assistants for Arabic-speaking users. - Sentiment analysis to gain insights into local markets and customer trends. - Summarization of bilingual Arabic-English documents. Audiences that we hope will benefit from our model: - Academics: For those researching Arabic Natural Language Processing. - Businesses: Companies targeting Arabic-speaking audiences. - Developers: Those integrating Arabic language capabilities in applications. ### Out-of-Scope Use <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. --> While the Jais family of models are powerful Arabic and English bilingual models, it's essential to understand their limitations and the potential of misuse. It is prohibited to use the model in any manner that violates applicable laws or regulations. The following are some example scenarios where the model should not be used. - Malicious Use: The model should not be used to generate harmful, misleading, or inappropriate content. Thisincludes but is not limited to: - Generating or promoting hate speech, violence, or discrimination. - Spreading misinformation or fake news. - Engaging in or promoting illegal activities. - Sensitive Information: The model should not be used to handle or generate personal, confidential, or sensitive information. - Generalization Across All Languages: Jais family of models are bilingual and optimized for Arabic and English. They should not be presumed to have equal proficiency in other languages or dialects. - High-Stakes Decisions: The model should not be used to make high-stakes decisions without human oversight. This includes medical, legal, financial, or safety-critical decisions. ## Bias, Risks, and Limitations <!-- This section is meant to convey both technical and sociotechnical limitations. --> The Jais family is trained on publicly available data which was in part curated by Inception. We have employed different techniques to reduce bias in the model. While efforts have been made to minimize biases, it is likely that the model, as with all LLM models, will exhibit some bias. The fine-tuned variants are trained as an AI assistant for Arabic and English speakers. Chat models are limited to produce responses for queries in these two languages and may not produce appropriate responses to other language queries. By using Jais, you acknowledge and accept that, as with any large language model, it may generate incorrect, misleading and/or offensive information or content. The information is not intended as advice and should not be relied upon in any way, nor are we responsible for any of the content or consequences resulting from its use. We are continuously working to develop models with greater capabilities, and as such, welcome any feedback on the model. Copyright Inception Institute of Artificial Intelligence Ltd. JAIS is made available under the Apache License, Version 2.0 (the “License”). You shall not use JAIS except in compliance with the License. You may obtain a copy of the License at https://www.apache.org/licenses/LICENSE-2.0. Unless required by applicable law or agreed to in writing, JAIS is distributed on an AS IS basis, without warranties or conditions of any kind, either express or implied. Please see the terms of the License for the specific language permissions and limitations under the License. #### Summary We release the Jais family of Arabic and English bilingual models. The wide range of pre-trained model sizes, the recipe for adapting English-centric models to Arabic, and the fine-tuning of all sizes unlocks numerous use cases commercially and academically in the Arabic setting. Through this release, we aim to make LLMs more accessible to Arabic NLP researchers and companies, offering native Arabic models that provide better cultural understanding than English centric ones. The strategies we employ for pre-training, fine-tuning and adaptation to Arabic are extensible to other low and medium resource languages, paving the way for language-focused and accessible models that cater to local contexts. #### Citation info ```bibtex @misc{sengupta2023jais, title={Jais and Jais-chat: Arabic-Centric Foundation and Instruction-Tuned Open Generative Large Language Models}, author={Neha Sengupta, Sunil Kumar Sahu, Bokang Jia, Satheesh Katipomu, Haonan Li, Fajri Koto, William Marshall, Gurpreet Gosal, Cynthia Liu, Zhiming Chen, Osama Mohammed Afzal, Samta Kamboj, Onkar Pandit, Rahul Pal, Lalit Pradhan, Zain Muhammad Mujahid, Massa Baali, Xudong Han, Sondos Mahmoud Bsharat, Alham Fikri Aji, Zhiqiang Shen, Zhengzhong Liu, Natalia Vassilieva, Joel Hestness, Andy Hock, Andrew Feldman, Jonathan Lee, Andrew Jackson, Hector Xuguang Ren, Preslav Nakov, Timothy Baldwin and Eric Xing}, year={2023}, eprint={2308.16149}, archivePrefix={arXiv}, primaryClass={cs.CL} } @article{jaisfamilymodelcard, title={Jais Family Model Card}, author={Inception}, year={2024}, url = {https://huggingface.co/inceptionai/jais-family-30b-16k-chat/blob/main/README.md} } ```	{"language": ["ar", "en"], "license": "apache-2.0", "pipeline_tag": "text-generation", "tags": ["Arabic", "English", "LLM", "Decoder", "causal-lm", "jais-family"]}	task	[ "TRANSLATION", "SUMMARIZATION" ]	43,162
LordCoffee/bert-base-cased-cefr	LordCoffee	token-classification	[ "transformers", "pytorch", "bert", "text-classification", "code", "token-classification", "en", "dataset:Alex123321/english_cefr_dataset", "license:apache-2.0", "autotrain_compatible", "endpoints_compatible", "region:us" ]	2024-06-13T04:35:05Z	2024-06-13T04:40:28+00:00	12	0	--- datasets: - Alex123321/english_cefr_dataset language: - en library_name: transformers license: apache-2.0 metrics: - accuracy pipeline_tag: token-classification tags: - code --- # Model Card: BERT-based CEFR Classifier ## Overview This repository contains a model trained to predict Common European Framework of Reference (CEFR) levels for a given text using a BERT-based model architecture. The model was fine-tuned on the CEFR dataset, and the `bert-base-...` pre-trained model was used as the base. ## Model Details - Model architecture: BERT (base model: `bert-base-...`) - Task: CEFR level prediction for text classification - Training dataset: CEFR dataset - Fine-tuning: Epochs, Loss, etc. ## Performance The model's performance during training is summarized below: \| Epoch \| Training Loss \| Validation Loss \| \|-------\|---------------\|-----------------\| \| 1 \| 0.412300 \| 0.396337 \| \| 2 \| 0.369600 \| 0.388866 \| \| 3 \| 0.298200 \| 0.419018 \| \| 4 \| 0.214500 \| 0.481886 \| \| 5 \| 0.148300 \| 0.557343 \| --Additional metrics: --Training Loss: 0.2900624789151278 --Training Runtime: 5168.3962 seconds --Training Samples per Second: 10.642 --Total Floating Point Operations: 1.447162776576e+16 ## Usage 1. Install the required libraries by running `pip install transformers`. 2. Load the trained model and use it for CEFR level prediction. from transformers import pipeline # Load the model model_name = "AbdulSami/bert-base-cased-cefr" classifier = pipeline("text-classification", model=model_name) # Text for prediction text = "This is a sample text for CEFR classification." # Predict CEFR level predictions = classifier(text) # Print the predictions print(predictions)	null	Non_BioNLP	# Model Card: BERT-based CEFR Classifier ## Overview This repository contains a model trained to predict Common European Framework of Reference (CEFR) levels for a given text using a BERT-based model architecture. The model was fine-tuned on the CEFR dataset, and the `bert-base-...` pre-trained model was used as the base. ## Model Details - Model architecture: BERT (base model: `bert-base-...`) - Task: CEFR level prediction for text classification - Training dataset: CEFR dataset - Fine-tuning: Epochs, Loss, etc. ## Performance The model's performance during training is summarized below: \| Epoch \| Training Loss \| Validation Loss \| \|-------\|---------------\|-----------------\| \| 1 \| 0.412300 \| 0.396337 \| \| 2 \| 0.369600 \| 0.388866 \| \| 3 \| 0.298200 \| 0.419018 \| \| 4 \| 0.214500 \| 0.481886 \| \| 5 \| 0.148300 \| 0.557343 \| --Additional metrics: --Training Loss: 0.2900624789151278 --Training Runtime: 5168.3962 seconds --Training Samples per Second: 10.642 --Total Floating Point Operations: 1.447162776576e+16 ## Usage 1. Install the required libraries by running `pip install transformers`. 2. Load the trained model and use it for CEFR level prediction. from transformers import pipeline # Load the model model_name = "AbdulSami/bert-base-cased-cefr" classifier = pipeline("text-classification", model=model_name) # Text for prediction text = "This is a sample text for CEFR classification." # Predict CEFR level predictions = classifier(text) # Print the predictions print(predictions)	{"datasets": ["Alex123321/english_cefr_dataset"], "language": ["en"], "library_name": "transformers", "license": "apache-2.0", "metrics": ["accuracy"], "pipeline_tag": "token-classification", "tags": ["code"]}	task	[ "TEXT_CLASSIFICATION" ]	43,163

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.