executing-plans
executing-plans
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available
Skills entdecken
Produktivität · SKILL.md
balanced-policy-optimization-rl
balanced-policy-optimization-rlStabilize off-policy RL for LLMs using adaptive clipping that dynamically rebalances positive/negative gradients and preserves entropy, improving mathematical reasoning performance vs standard PPO.
Independent GitHub snapshot · Automated classification · Original documentation governs
Skill-Dokumentation
GitHubDokumentation wird aus der Quelle geladen…