zephyr-7b-dpo-lora

This model is a fine-tuned version of HuggingFaceH4/zephyr-7b-beta on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1394
• Rewards/chosen: 1.9472
• Rewards/rejected: -2.3030
• Rewards/accuracies: 0.8250
• Rewards/margins: 4.2502
• Logps/rejected: -241.4177
• Logps/chosen: -240.5213
• Logits/rejected: -2.7622
• Logits/chosen: -2.7854

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: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• 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
0.0104	1.0	429	0.1394	1.9472	-2.3030	0.8250	4.2502	-241.4177	-240.5213	-2.7622	-2.7854

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1