OrpoLlama-3-8B-Instruct

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

• Loss: 1.0981
• Rewards/chosen: -0.0745
• Rewards/rejected: -0.0794
• Rewards/accuracies: 0.4000
• Rewards/margins: 0.0049
• Logps/rejected: -0.7938
• Logps/chosen: -0.7449
• Logits/rejected: 0.0540
• Logits/chosen: -0.1876
• Nll Loss: 1.0259
• Log Odds Ratio: -0.7218
• Log Odds Chosen: 0.0125

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
1.4815	0.2020	25	1.2773	-0.0942	-0.1188	0.6000	0.0246	-1.1878	-0.9419	0.0316	-0.2920	1.2167	-0.6066	0.2976
0.9343	0.4040	50	1.1677	-0.0819	-0.0918	0.4000	0.0099	-0.9183	-0.8189	-0.0239	-0.2369	1.1001	-0.6763	0.1038
1.2029	0.6061	75	1.1140	-0.0759	-0.0805	0.4000	0.0046	-0.8050	-0.7587	-0.0008	-0.2166	1.0421	-0.7191	0.0131
1.25	0.8081	100	1.0981	-0.0745	-0.0794	0.4000	0.0049	-0.7938	-0.7449	0.0540	-0.1876	1.0259	-0.7218	0.0125

Framework versions

• PEFT 0.11.1
• Transformers 4.41.2
• Pytorch 2.3.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1