krystv/nomen-ai-dpo
Viewer • Updated • 6k • 33
How to use krystv/nomen-ai-dpo-lora with PEFT:
Task type is invalid.
Target output repo for the DPO anti-derivative phase of Nomen-AI.
Training command:
export SFT_ADAPTER=krystv/nomen-ai-sft-lora
export HUB_MODEL_ID=krystv/nomen-ai-dpo-lora
python scripts/train_dpo.py
Main pipeline repo: https://huggingface.co/krystv/nomen-ai