DataChef-32B is a specialized large language model designed for
automated data recipe generation. It was introduced in the paper
DataChef: Cooking Up Optimal Data Recipes for LLM Adaptation via Reinforcement Learning.
DataChef-32B facilitates LLM adaptation by generating executable data processing pipelines (data recipes) that transform raw data sources into high-quality training corpora targeted at specific benchmarks.
DataChef-32B addresses the manual, labor-intensive process of designing data processing pipelines. It was trained using online reinforcement learning with a proxy reward system that predicts downstream performance for candidate recipes. Given a target benchmark and available data sources, the model outputs a complete data recipe to adapt a base LLM.
Across diverse tasks, DataChef-32B produces practical recipes that reach performance comparable to those curated by human experts. Notably, a recipe generated by DataChef-32B was used to adapt Qwen3-1.7B-Base to the math domain, achieving a score of 66.7 on AIME'25, surpassing the performance of the standard Qwen3-1.7B.
To use the DataChef framework for generating your own data recipes, follow the installation steps from the
GitHub repository:
1conda create -n datachef python=3.12
2conda activate datachef
3pip install -e .
1@article{chen2026datachef,
2 title={DataChef: Cooking Up Optimal Data Recipes for LLM Adaptation via Reinforcement Learning},
3 author={Chen, Yicheng and Ma, Zerun and Xie, Xinchen and Li, Yining and Chen, Kai},
4 journal={arXiv preprint arXiv:2602.11089},
5 year={2026}
6}
