DeepYardDeepYard
O

OpenForgeRL

End-to-end RL framework for training stateful coding agents across diverse environments

Open SourceFree

About

Open-source reinforcement learning framework designed specifically for training harness-based agents like Claude Code and Codex. Supports both supervised fine-tuning (SFT) and RL training with native handling of stateful, multi-process harness inference. Enables researchers to train complex multi-turn reasoning agents end-to-end across diverse coding and task environments.

Details

Language
Patterns

Tags

open-sourceframeworkcoding-agentautonomouspythonevaluation