CodeMidas: 코드 자체로부터 에이전트 기반 코딩 RL 환경 확장하기
이 논문은 기존의 issue나 commit 같은 개발 산출물에 의존하지 않고, 소스 코드 자체를 입력으로 사용하여 executable RL environments를 생성하는 CodeMidas 파이프라인을 제안합니다. CodeMidas는 agentic compute를 활용하여 기능 명세(behavioral specifications)를 수립하고, 원본 코드 실행에 기반한 test를 구축하며, 실행 체크 및 repeated solution rollouts를 통해 task를 검증합니다. 이를 통해 23개의 programming languages와 15개의 technical domains를 아우르는 5,545개의 training tasks를 포함한 대규모 데이터셋을 구축했습니다.
Training capable coding agents via reinforcement learning (RL) requires diverse tasks with reliable verifiers. Open-source codebases offer a rich source of such tasks, while existing methods typically rely on development artifacts such as issues and commits, limiting the range of tasks that can be extracted. To better scale RL environments, we present CodeMidas, an agentic pipeline that turns implemented functionality in existing codebases into executable RL environments using source code as its only task-specific input. CodeMidas allocates agentic compute to every stage of environment construction: agents explore implemented functionality to formulate behavioral specifications, construct tests grounded in execution of the original code, and validate and filter candidate tasks through execution checks and repeated solution rollouts. The resulting dataset has 5,545 training tasks from 3,185 open-source codebases spanning 23 programming languages and 15 technical domains. Training MiMo-V