llama2-7b-sft-chat-no-template-dpo

This model is a fine-tuned version of elichen3051/llama2-7b-sft-chat-no-template on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Loss: 0.6026
• Rewards/chosen: -0.9396
• Rewards/rejected: -1.3040
• Rewards/accuracies: 0.6560
• Rewards/margins: 0.3644
• Logps/rejected: -372.5772
• Logps/chosen: -371.2111
• Logits/rejected: -1.0199
• Logits/chosen: -1.0320

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: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 8
• 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
0.5619	0.9987	477	0.6026	-0.9396	-1.3040	0.6560	0.3644	-372.5772	-371.2111	-1.0199	-1.0320

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0
• Datasets 2.19.2
• Tokenizers 0.19.1