biomistral-7b-wo-kqa_golden-iter-dpo-step1_gamma0

This model is a fine-tuned version of Minbyul/biomistral-7b-wo-kqa_golden-iter-sft-step1_gamma0 on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Loss: 0.6932
• Rewards/chosen: -0.0011
• Rewards/rejected: 0.0003
• Rewards/accuracies: 0.3333
• Rewards/margins: -0.0014
• Logps/rejected: -193.9042
• Logps/chosen: -136.6186
• Logits/rejected: -2.7172
• Logits/chosen: -3.2298

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-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

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.15.2