zephyr-7b-dpo-full

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2016
• Rewards/chosen: -5.9224
• Rewards/rejected: -7.8036
• Rewards/accuracies: 0.7380
• Rewards/margins: 1.8812
• Logps/rejected: -3.9018
• Logps/chosen: -2.9612
• Logits/rejected: -1.8800
• Logits/chosen: -1.8679

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: 4
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• 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
1.2816	0.8375	400	1.2016	-5.9224	-7.8036	0.7380	1.8812	-3.9018	-2.9612	-1.8800	-1.8679

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.0+cu121
• Datasets 2.14.6
• Tokenizers 0.20.1