
GRPO and RLVR Training
SkillTrain on verifiable rewards and debug diverging GRPO runs
by wshobson Listed Jul 24, 2025
Clone or download it from github.com, then drop the folder into ~/.claude/skills/ (Claude Code) or your agent's skills directory.
Product badge
Share this product's name and rating in your README or on your website.
Updates may be delayed by image caching.

About GRPO and RLVR Training
A wshobson/agents skill for reinforcement learning from verifiable rewards. It covers tasks whose success can be checked by code (math, tool calls, structured output), reward function design and troubleshooting runs that diverge or get reward-hacked.
No reviews yet
Used it? Write the first review.