🛰️ Daily AI Frontier
‹ back to 2026-09-21

CodeMidas: Scaling Agentic Coding RL Environments from Code Itself

Research LLM Agents

Ranking

Overall 84
Content 90
Popularity 70

Observed public metrics from 1 member.

Merged summary

TL;DR - CodeMidas converts implemented functionality in open-source codebases into verified reinforcement-learning environments for coding agents using only source code. Its 5,545-task dataset improved a trained model across issue repair, program construction, and terminal-work benchmarks.

  • Agentic stages generate behavioral specifications, execution-grounded tests, and validated tasks through repeated solution rollouts.
  • The dataset spans 3,185 repositories, 23 programming languages, and 15 technical domains.
  • GRPO training of MiMo-V2.5 yielded gains of 11.7% on DeepSWE, 17% on ProgramBench, and 8.5% on Terminal-Bench v2.1.
  • Trained agents explored codebases more extensively and used more diverse self-verification behaviors.

Sources (1)

CodeMidas: Scaling Agentic Coding RL Environments from Code Itself

arXiv cs.AI Bowen Ye, Lei Li, Shicheng Li, Zihao Yue, Linghao Zhang, Hanglong Lv, Yuanxin Liu, Wenhan Ma, Hao Tian, Rang Li, Jinhao Dong, Yikai Zhao, Xiangwei Deng, Hailin Zhang, Liang Zhao, Qi Liu, Lingpeng Kong, Tong Yang, Fuli Luo 2026-09-18 arXiv:2609.22068
Public signals Hugging Face upvotes 131
Providers: Hugging Face · Upvotes 131 OpenAlex · N/A Publisher · N/A Semantic Scholar · N/A X · N/A Fetched 2026-09-25 14:17:26.836819 UTC

TL;DR - CodeMidas converts implemented functionality in open-source codebases into verified reinforcement-learning environments for coding agents using only source code. Its 5,545-task dataset improved a trained model across issue repair, program construction, and terminal-work benchmarks.

  • Agentic stages generate behavioral specifications, execution-grounded tests, and validated tasks through repeated solution rollouts.
  • The dataset spans 3,185 repositories, 23 programming languages, and 15 technical domains.
  • GRPO training of MiMo-V2.5 yielded gains of 11.7% on DeepSWE, 17% on ProgramBench, and 8.5% on Terminal-Bench v2.1.
  • Trained agents explored codebases more extensively and used more diverse self-verification behaviors.
item →