DR as Rubric Data

Companion dataset for the paper "Deep Research as Rubric for Reinforcement Learning".

Paper repository: https://github.com/miiikiik/dr-as-rubric

Dataset Description

This dataset contains two types of data produced by the DR-as-Rubric pipeline:

1. SFT Training Data (sft_data/)

2. Rubric Data (rubrics/)

Evaluation rubrics generated by different models and bootstrap steps. Each CSV has columns query and criteria.

Usage

import json
import csv

# Load SFT data
with open("sft_data/trajectory_sharegpt.json") as f:
    sft_data = json.load(f)

# Load rubrics
with open("rubrics/real-rubric-bootstrap-step1.csv", encoding="utf-8-sig") as f:
    reader = csv.DictReader(f)
    rubrics = list(reader)

Citation

@article{mei2026dr,
  title={Deep Research as Rubric for Reinforcement Learning},
  author={Mei, Wangyi and Gu, Zhouhong and Bai, Zhenhan and Cai, Yin and Zhang, Lefan and Ding, Zhenxin and Chen, Bo and Gao, Yan and Wu, Yi and Hu, Yao and Liang, Jiaqing and Yang, Deqing},
  year={2026}
}

License

Apache License 2.0