Training in progress, step 5000

README.md

@@ -1,7 +1,7 @@
 ---
 base_model: gpt2
 datasets:
-- distily/synth_gpt2_t1_seq_1M
+- distily/filtered_redpajama_multilingual
 library_name: Distily
 license: creativeml-openrail-m
 tags:
@@ -18,7 +18,7 @@ model-index:
 
 Distilled with [Distily](https://github.com/lapp0/distily) library
 using teacher model [gpt2](https://huggingface.co/gpt2)
-on dataset [distily/synth_gpt2_t1_seq_1M](https://huggingface.co/datasets/distily/synth_gpt2_t1_seq_1M).
+on dataset [distily/filtered_redpajama_multilingual](https://huggingface.co/datasets/distily/filtered_redpajama_multilingual).
 
 <!-- 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.
@@ -78,13 +78,13 @@ GPT2LMHeadModel(
 
 # Resource Usage
 
-- Max Train VRAM Use: 15.7151 GB
-- Available VRAM: 23.6497 GB
+- Max Train VRAM Use: 15.7170 GB
+- Available VRAM: 23.4329 GB
 - GPUs: 
   - 1x NVIDIA GeForce RTX 4090
-- CPUs: 28
-- CPU Memory: 62.6429 GB
-- CPU Memory Bandwidth: 700 GB/s
+- CPUs: 64
+- CPU Memory: 251.7299 GB
+- CPU Memory Bandwidth: 1600 GB/s
 
 # Distillation (Teacher -> Student) Architecture Difference:
 
@@ -115,7 +115,7 @@ GPT2LMHeadModel(
 <br/>
 
 # Train Dataset
-Trained on 680,993,580 tokens from the [distily/synth_gpt2_t1_seq_1M](https://huggingface.co/datasets/distily/synth_gpt2_t1_seq_1M) dataset.
+Trained on 704,434,140 tokens from the [distily/filtered_redpajama_multilingual](https://huggingface.co/datasets/distily/filtered_redpajama_multilingual) dataset.
 
 - Num Samples: `998,000`
 - Subset: `None`
@@ -172,7 +172,7 @@ The following hyperparameters were used during training:
         projector='orthogonal'
     )
 )`
-- lr_scheduler: `<torch.optim.lr_scheduler.LambdaLR object at 0x7fdddd236a10>`
+- lr_scheduler: `<torch.optim.lr_scheduler.LambdaLR object at 0x7d6a844e2920>`
 - student_model_name_or_path: `None`
 - student_config_name_or_path: `distilbert/distilgpt2`
 - student_model_config: `None`
@@ -182,13 +182,13 @@ The following hyperparameters were used during training:
 - teacher_model_name_or_path: `gpt2`
 - teacher_load_in_8bit: `False`
 - teacher_load_in_4bit: `False`
-- dataset_uri: `distily/synth_gpt2_t1_seq_1M`
+- dataset_uri: `distily/filtered_redpajama_multilingual`
 - dataset_subset: `None`
 - dataset_split: `train`
 - dataset_column_name: `text`
 - dataset_sample_size: `1000000`
 - dataset_test_size: `0.002`
-- dataset_shuffle: `False`
+- dataset_shuffle: `True`
 - dataset_shuffle_seed: `42`
 - dataset_trust_remote_code: `False`
 - gradient_accumulation_steps: `1`

logs/dataset_shuffle=True, dataset_split=train, dataset_subset=None, dataset_uri=distily_filtered_redpajama_en, lr_scheduler_kwargs=None, lr_scheduler_type=constant/events.out.tfevents.1725803753.b57c76173204

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96bb82b6fc6c09d2b1694d50f4e00ccb186036af1269d4aca85b56c01f29ee3a
+size 162424

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:af4ecebeb2251bce9c8aed8210fdfb2df60f776e203addc1f20d64154ef9b0d7
+oid sha256:0d7e99bc4f576e39d000db40d35bb1941e3aae424eed3b1984a60530d5be2b2b
 size 163832792

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c6149a0670f6cd78dcdc023c24f80916b0d923c3ba20ca59b9a23e841bf9c797
+oid sha256:84e351fdf0a83a60c74e1eede3594190d6e5a1d276e4870b3eecb9650bf2df05
 size 5560