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: 0.5272
• Rewards/chosen: -0.1409
• Rewards/rejected: -0.8819
• Rewards/accuracies: 0.7340
• Rewards/margins: 0.7410
• Logps/rejected: -232.4720
• Logps/chosen: -265.9766
• Logits/rejected: -1.9804
• Logits/chosen: -2.0345

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
• gradient_accumulation_steps: 32
• 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: 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
0.5654	1.0	968	0.5545	-0.0993	-0.6333	0.7160	0.5339	-229.9861	-265.5612	-2.0257	-2.0790
0.5587	2.0	1936	0.5326	-0.1409	-0.8429	0.7295	0.7021	-232.0825	-265.9764	-1.9888	-2.0427
0.5194	3.0	2904	0.5272	-0.1409	-0.8819	0.7340	0.7410	-232.4720	-265.9766	-1.9804	-2.0345

Framework versions

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