llama2-7b-dpo-lora-20231129-32

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6869
• Rewards/chosen: 0.0290
• Rewards/rejected: 0.0096
• Rewards/accuracies: 0.5833
• Rewards/margins: 0.0193
• Logps/rejected: -288.6959
• Logps/chosen: -359.9739
• Logits/rejected: -0.3197
• Logits/chosen: -0.1734

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: 8
• gradient_accumulation_steps: 32
• total_train_batch_size: 512
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6918	1.0	121	0.6911	0.0118	0.0027	0.5159	0.0091	-288.7652	-360.1456	-0.3204	-0.1740
0.6887	2.0	242	0.6886	0.0182	0.0065	0.5437	0.0118	-288.7276	-360.0812	-0.3203	-0.1741
0.6865	3.0	363	0.6869	0.0290	0.0096	0.5833	0.0193	-288.6959	-359.9739	-0.3197	-0.1734

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu121
• Datasets 2.14.6
• Tokenizers 0.14.1