{"skill":{"slug":"debugging-reinforcement-learning","displayName":"Debugging Reinforcement Learning","summary":"Tools and methods for controlling randomness, ensuring reproducibility, analyzing agent behavior, and debugging reward issues in stochastic reinforcement lea...","tags":{"latest":"1.0.0"},"stats":{"comments":0,"downloads":83,"installsAllTime":1,"installsCurrent":1,"stars":0,"versions":1},"createdAt":1776448900766,"updatedAt":1776449208301},"latestVersion":{"version":"1.0.0","createdAt":1776448900766,"changelog":"Initial release with comprehensive debugging toolkit for RL agents:\n\n- Tools for controlling and analyzing sources of nondeterminism in reinforcement learning environments.\n- Modules for reproducibility, including episode replay, deterministic seed management, and diffing trajectories.\n- Behavior analysis features: trajectory clustering, policy consistency checks, and behavioral mode detection.\n- Extensive reward debugging utilities: reward decomposition, scale analysis, hacking detection, and validation.\n- Designed for integration with popular RL environments and libraries; outputs structured logs and reports.","license":"MIT-0"},"metadata":null,"owner":{"handle":"roamer-remote","userId":"s175wscrjbc6yvhxa7yrm073wh83m7pt","displayName":"Roamer 徐","image":"https://avatars.githubusercontent.com/u/267348107?v=4"},"moderation":null}