OrpoLlama3-8B-FT

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6399
• Rewards/chosen: -0.1279
• Rewards/rejected: -0.1298
• Rewards/accuracies: 1.0
• Rewards/margins: 0.0020
• Logps/rejected: -1.2982
• Logps/chosen: -1.2786
• Logits/rejected: -1.5312
• Logits/chosen: -0.9326
• Nll Loss: 1.5720
• Log Odds Ratio: -0.6797
• Log Odds Chosen: 0.0271

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: 8e-06
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• 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	Nll Loss	Log Odds Ratio	Log Odds Chosen
4.238	0.24	3	1.6636	-0.1298	-0.1322	1.0	0.0024	-1.3225	-1.2980	-1.1489	-0.9403	1.5959	-0.6766	0.0335
4.8415	0.48	6	1.6603	-0.1295	-0.1319	1.0	0.0024	-1.3193	-1.2953	-1.2236	-0.9390	1.5926	-0.6768	0.0329
2.4409	0.72	9	1.6512	-0.1288	-0.1311	1.0	0.0023	-1.3109	-1.2882	-1.3781	-0.9360	1.5835	-0.6777	0.0312
2.0082	0.96	12	1.6399	-0.1279	-0.1298	1.0	0.0020	-1.2982	-1.2786	-1.5312	-0.9326	1.5720	-0.6797	0.0271

Framework versions

• PEFT 0.11.1
• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1