zephyr-7b-gpo-v1-i0

This model is a fine-tuned version of DUAL-GPO/zephyr-7b-gpo-update3-i0 on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Logits/chosen: -1.9105
• Logits/rejected: -1.7279
• Logps/chosen: -271.5140
• Logps/rejected: -255.4308
• Loss: 0.0328
• Rewards/accuracies: 0.6240
• Rewards/chosen: -0.0807
• Rewards/margins: 0.0903
• Rewards/rejected: -0.1710

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: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• total_eval_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 0.5

Training results

Training Loss	Epoch	Step	Logits/chosen	Logits/rejected	Logps/chosen	Logps/rejected	Validation Loss	Rewards/accuracies	Rewards/chosen	Rewards/margins	Rewards/rejected
0.0266	0.01	100	-1.9105	-1.7279	-271.5140	-255.4308	0.0328	0.6240	-0.0807	0.0903	-0.1710

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.15.2