biomistral-7b-dpo-full-wo-healthsearch_qa-ep3

This model is a fine-tuned version of BioMistral/BioMistral-7B on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Loss: 0.5683
• Rewards/chosen: -0.3658
• Rewards/rejected: -0.7793
• Rewards/accuracies: 0.7216
• Rewards/margins: 0.4135
• Logps/rejected: -426.6962
• Logps/chosen: -289.8896
• Logits/rejected: -1.0279
• Logits/chosen: -2.6707

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-07
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.3841	0.83	100	0.5714	-0.3562	-0.7545	0.7188	0.3982	-424.2126	-288.9360	-1.0376	-2.6781

Framework versions

• Transformers 4.39.0.dev0
• Pytorch 2.1.2
• Datasets 2.14.6
• Tokenizers 0.15.2