zephyr-8b-dpo-full/README.md

library_name, license, base_model, tags, datasets, model-index

library_name	license	base_model	tags	datasets	model-index
transformers	llama3.1	meta-llama/Llama-3.1-8B	alignment-handbook trl dpo generated_from_trainer trl dpo generated_from_trainer	HuggingFaceH4/ultrafeedback_binarized	name results zephyr-8b-dpo-full

zephyr-8b-dpo-full

This model is a fine-tuned version of meta-llama/Llama-3.1-8B on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Loss: 0.5372
• Rewards/chosen: -1.1911
• Rewards/rejected: -2.0021
• Rewards/accuracies: 0.7656
• Rewards/margins: 0.8110
• Logps/rejected: -481.7125
• Logps/chosen: -401.6271
• Logits/rejected: -0.5622
• Logits/chosen: -0.5978

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-07
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• total_eval_batch_size: 64
• 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	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.5134	0.9984	477	0.5372	-1.1911	-2.0021	0.7656	0.8110	-481.7125	-401.6271	-0.5622	-0.5978

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.1+rocm6.2
• Datasets 3.2.0
• Tokenizers 0.20.3