zephyr-7b-dpo-full-ultrabin-low-margin

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

• Loss: 0.6718
• Rewards/chosen: 0.0158
• Rewards/rejected: -0.0423
• Rewards/accuracies: 0.6211
• Rewards/margins: 0.0581
• Logps/rejected: -266.8938
• Logps/chosen: -261.0514
• Logits/rejected: -2.4791
• Logits/chosen: -2.5121

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: 8
• eval_batch_size: 8
• seed: 55
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• 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
0.6816	0.3484	50	0.6725	-0.0071	-0.0549	0.5898	0.0478	-268.1548	-263.3413	-2.5366	-2.5709
0.6723	0.6969	100	0.6687	-0.0876	-0.1545	0.5742	0.0669	-278.1130	-271.3896	-2.4595	-2.4953

Framework versions

• Transformers 4.44.0.dev0
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1