phi-2-gpo-renew2-b0.001-i1

This model is a fine-tuned version of DUAL-GPO/phi-2-gpo-renew2-b0.001-i0 on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Loss: 0.0538
• Rewards/chosen: 0.0010
• Rewards/rejected: 0.0012
• Rewards/accuracies: 0.4290
• Rewards/margins: -0.0002
• Logps/rejected: -366.0280
• Logps/chosen: -395.2844
• Logits/rejected: -0.7463
• Logits/chosen: -0.8436

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: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 4
• total_train_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	Logits/chosen	Logits/rejected	Logps/chosen	Logps/rejected	Validation Loss	Rewards/accuracies	Rewards/chosen	Rewards/margins	Rewards/rejected
0.1204	0.32	100	-0.8372	-0.7448	-396.1279	-367.0282	0.0537	0.4495	0.0002	-0.0000	0.0002
0.1673	0.64	200	0.0538	0.0013	0.0015	0.4305	-0.0002	-365.7495	-395.0410	-0.7569	-0.8518
0.1395	0.96	300	0.0538	0.0010	0.0012	0.4395	-0.0002	-365.9886	-395.3006	-0.7587	-0.8541

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.1.2
• Datasets 2.14.6
• Tokenizers 0.15.2