zephyr-7b-dpo-qlora

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-qlora on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Loss: 0.5031
• Rewards/chosen: -1.9728
• Rewards/rejected: -2.9618
• Rewards/accuracies: 0.7695
• Rewards/margins: 0.9890
• Logps/rejected: -543.7128
• Logps/chosen: -443.7073
• Logits/rejected: -1.1810
• Logits/chosen: -1.2525

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-06
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• total_eval_batch_size: 64
• 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

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.5646	0.2093	100	0.5739	-0.9253	-1.4816	0.7188	0.5564	-395.6964	-338.9565	-1.9267	-1.9878
0.5524	0.4186	200	0.5318	-0.8476	-1.5395	0.7617	0.6919	-401.4810	-331.1845	-1.5104	-1.5801
0.4977	0.6279	300	0.5100	-1.8821	-2.8383	0.7773	0.9562	-531.3586	-434.6388	-1.1156	-1.1878
0.5096	0.8373	400	0.5035	-2.0213	-3.0170	0.7656	0.9957	-549.2363	-448.5603	-1.1850	-1.2569

Framework versions

• PEFT 0.13.2
• Transformers 4.45.2
• Pytorch 2.1.2+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1