htlou/mm-interp-RLAIF-V-Cosi-q0_75

RLAIF-V-Cosi-q0_75

This model is a fine-tuned version of llava-hf/llava-v1.6-mistral-7b-hf on the RLAIF-V-Cosi-q0_75 dataset. It achieves the following results on the evaluation set:

• Loss: 1.0076

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: 1e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• total_eval_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
1.1181	0.2320	50	1.1215
1.0765	0.4640	100	1.0775
1.0639	0.6961	150	1.0502
1.0308	0.9281	200	1.0282
0.9038	1.1601	250	1.0220
0.8973	1.3921	300	1.0114
0.8747	1.6241	350	1.0039
0.8818	1.8561	400	0.9965
0.7851	2.0882	450	1.0085
0.7784	2.3202	500	1.0107
0.7773	2.5522	550	1.0086
0.784	2.7842	600	1.0077

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.20.3