zephyr-7b-dpo_hh-rlhf

Zephyr 7B SFT aligned with DPO on HH-RLHF with β=0.01

This repo contains LoRA adapter created by aligning Zephyr 7B SFT on the Anthropic HH-RLHF dataset using Direct Preference Optimization (DPO). It was trained as a series of models for studying DPO alignment.

Model details

• Base model: alignment-handbook/zephyr-7b-sft-full
• Preference dataset: Anthropic/hh-rlhf
• DPO beta: 0.01
• Training framework: PEFT/LoRA

See the base model card for usage and chat template details.

Training hyperparameters

• Epochs: 1
• Batch size: 16
• Learning rate: 5e-06
• Learning rate scheduler: cosine
• Learning rate warmup ratio: 0.1
• Gradient accumulation: 2
• LoRA:
  • rank: 64
  • alpha: 64
  • dropout: 0.05
  • target modules: [q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj]

License

This adapter is released under the Apache License 2.0.

Citation

If this work was helpful, please cite:

TBA