dpo-llama3-8b-sample-rules

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1915
• Rewards/chosen: 0.1004
• Rewards/rejected: -1.6031
• Rewards/accuracies: 1.0
• Rewards/margins: 1.7035
• Logps/rejected: -517.5475
• Logps/chosen: -200.9981
• Logits/rejected: -1.3936
• Logits/chosen: -1.2455

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: 1
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.4345	0.4444	50	0.3859	0.2018	-0.5860	1.0	0.7877	-415.8305	-190.8644	-1.3991	-1.2723
0.2288	0.8889	100	0.1915	0.1004	-1.6031	1.0	1.7035	-517.5475	-200.9981	-1.3936	-1.2455

Framework versions

• PEFT 0.12.0
• Transformers 4.44.0
• Pytorch 2.3.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1