method · first seen 9 Oct 2026
reinforcement learning from rollouts
Post-training agents on their own collected trajectories using reward-based policy updates.
Mentions per week
PapersCodeDiscussion
Earliest sources
- ClawLoop: Self-Improving Agents via Harness-Native Training LoopPapers · 9 Oct 2026 · Hansi Yang, Linji Hao, Hongyan Hao, An Wang, Han Wu, Qi Gu, Xunliang Cai
Latest sources
- ClawLoop: Self-Improving Agents via Harness-Native Training LoopPapers · 9 Oct 2026 · Hansi Yang, Linji Hao, Hongyan Hao, An Wang, Han Wu, Qi Gu, Xunliang Cai