This repo contains LoRA adapter created by aligning Zephyr 7B SFT on the PKU-SafeRLHF 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: PKU-Alignment/PKU-SafeRLHF
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: 1e-05
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

jmajkutewicz/zephyr-7b-dpo_PKU-SafeRLHF

Zephyr 7B SFT aligned with DPO on PKU-SafeRLHF with β=0.01

Model details

Training hyperparameters

License

Citation

jmajkutewicz/
zephyr-7b-dpo_PKU-SafeRLHF