nash_dpo_iter_1

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the updated and the original datasets. It achieves the following results on the evaluation set:

• Loss: 0.6285
• Rewards/chosen: -0.1131
• Rewards/rejected: -0.2857
• Rewards/accuracies: 0.7000
• Rewards/margins: 0.1725
• Logps/rejected: -286.0817
• Logps/chosen: -295.3544
• Logits/rejected: -2.5416
• Logits/chosen: -2.6261

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: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• total_eval_batch_size: 8
• 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
0.6297	0.64	100	0.6285	-0.1131	-0.2857	0.7000	0.1725	-286.0817	-295.3544	-2.5416	-2.6261

Framework versions

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