orpo-phi

This model is a fine-tuned version of microsoft/Phi-3-mini-128k-instruct on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.3298
• Rewards/chosen: -0.1126
• Rewards/rejected: -0.1126
• Rewards/accuracies: 0.0
• Rewards/margins: 0.0
• Logps/rejected: -1.1264
• Logps/chosen: -1.1264
• Logits/rejected: 4.2177
• Logits/chosen: 4.2177
• Nll Loss: 1.2605
• Log Odds Ratio: -0.6931
• Log Odds Chosen: 0.0

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
3.4409	0.24	3	1.3298	-0.1126	-0.1126	0.0	0.0	-1.1264	-1.1264	4.2177	4.2177	1.2605	-0.6931	0.0
2.909	0.48	6	1.3298	-0.1126	-0.1126	0.0	0.0	-1.1264	-1.1264	4.2177	4.2177	1.2605	-0.6931	0.0
2.633	0.72	9	1.3298	-0.1126	-0.1126	0.0	0.0	-1.1264	-1.1264	4.2177	4.2177	1.2605	-0.6931	0.0
3.3955	0.96	12	1.3298	-0.1126	-0.1126	0.0	0.0	-1.1264	-1.1264	4.2177	4.2177	1.2605	-0.6931	0.0

Framework versions

• PEFT 0.11.1
• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1