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: 0.6142
• Rewards/chosen: 0.1283
• Rewards/rejected: -0.0917
• Rewards/accuracies: 0.6786
• Rewards/margins: 0.2200
• Logps/rejected: -238.3409
• Logps/chosen: -292.4302
• Logits/rejected: -2.4335
• Logits/chosen: -2.4867

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: 8
• gradient_accumulation_steps: 32
• total_train_batch_size: 512
• total_eval_batch_size: 32
• 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
0.654	1.0	121	0.6498	0.0901	-0.0131	0.6786	0.1032	-237.5548	-292.8124	-2.4392	-2.4918
0.6269	2.0	242	0.6220	0.1207	-0.0650	0.6746	0.1857	-238.0732	-292.5060	-2.4357	-2.4889
0.6158	3.0	363	0.6142	0.1283	-0.0917	0.6786	0.2200	-238.3409	-292.4302	-2.4335	-2.4867

Framework versions

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