zephyr-7b-dpo-lora

This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5655
• Rewards/chosen: -0.0745
• Rewards/rejected: -0.5329
• Rewards/accuracies: 0.7000
• Rewards/margins: 0.4583
• Logps/rejected: -224.6270
• Logps/chosen: -265.4236
• Logits/rejected: -2.0002
• Logits/chosen: -2.1215

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: 2
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 32
• total_train_batch_size: 256
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6125	1.0	242	0.6057	0.0079	-0.2400	0.6760	0.2479	-221.6983	-264.5998	-2.0242	-2.1445
0.5849	2.0	484	0.5731	-0.0578	-0.4714	0.6900	0.4136	-224.0123	-265.2563	-2.0071	-2.1279
0.5671	3.0	726	0.5655	-0.0745	-0.5329	0.7000	0.4583	-224.6270	-265.4236	-2.0002	-2.1215

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu121
• Datasets 2.14.6
• Tokenizers 0.14.1