zephyr-7b-dpo-lora-pubmedqa-selfgen-complete

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

• Loss: 0.0000
• Rewards/chosen: 0.1952
• Rewards/rejected: -11.1838
• Rewards/accuracies: 1.0
• Rewards/margins: 11.3790
• Logps/rejected: -1156.6167
• Logps/chosen: -39.8472
• Logits/rejected: -2.9309
• Logits/chosen: -2.9452

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: 2
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• total_eval_batch_size: 16
• 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
0.0001	0.42	7000	0.0000	0.1174	-11.2894	1.0	11.4068	-1167.1736	-47.6243	-2.9079	-2.9350
0.0002	0.83	14000	0.0000	0.1971	-11.1349	1.0	11.3320	-1151.7302	-39.6554	-2.9320	-2.9443

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.15.2