Training command: hf jobs uv run --flavor a10g-large --timeout 3h --secrets HF_TOKEN dpo_training.py
Eval command: hf jobs uv run --flavor a10g-large --with "lighteval[vllm]" --with "lighteval[math]" --with "transformers>4.52.0" --secrets HF_TOKEN lighteval accelerate "model_name=Abdullahpk1982/smollm3-dpo-aligned,dtype=bfloat16" gsm8k --push-to-hub --results-org Abdullahpk1982

Ready to merge
This branch is ready to get merged automatically.

Sign up or log in to comment