论文Agent 与开发者10:19ClawGym II:探索Agent Harness上的黑盒强化学习这篇论文把黑盒RL搬到agent harness上,用Qwen3-30A3B在OpenClaw和Claude Code上跑,ClawGym-Bench直接涨了10到15个点,还支持多个harness一起训练。#ClawGym#Qwen3-30A3B#OpenClaw#Claude CodeaarXiv cs.LG@Huatong Song 等 20 人原文稍后读已读值得跟进有用关注 ClawGym