Skip to main content
在 Manus 中运行任何 Skill
一键导入

rl-improvement-loop

星标7
分支0
更新时间2026年7月18日 23:01

Runs the Isaac Lab RL improvement loop for biped_ppo_walk: analyze prior training results (W&B, TensorBoard, logs), propose and implement one-hypothesis code changes, execute smoke/full training, evaluate with eval_biped_walk.py, and record each iteration in markdown until a stated goal is met. Use when improving Isaac Lab biped walking, analyzing W&B/TensorBoard logs, tuning rewards/termination/PPO in yuuki_isaac_lab, or when the user asks to iterate RL until a target (e.g. stable 5 m walk) is achieved.

安装

用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。

文件资源管理器
2 个文件
SKILL.md
readonly