nash_dpo_merge_iter_6

This model is a fine-tuned version of YYYYYYibo/nash_dpo_merge_iter_5 on the updated and the original datasets. It achieves the following results on the evaluation set:

• Loss: 0.5366
• Rewards/chosen: -0.8030
• Rewards/rejected: -1.4875
• Rewards/accuracies: 0.7320
• Rewards/margins: 0.6845
• Logps/rejected: -415.7152
• Logps/chosen: -367.6444
• Logits/rejected: -0.6068
• Logits/chosen: -1.1801

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
0.515	0.49	100	0.5449	-0.7799	-1.4614	0.7160	0.6816	-413.1093	-365.3295	-0.5798	-1.2092
0.4925	0.98	200	0.5366	-0.8030	-1.4875	0.7320	0.6845	-415.7152	-367.6444	-0.6068	-1.1801

Framework versions

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