simple_GRPO

Python ★ 0 updated 1y ago ⑂ fork

A very simple GRPO implement for reproducing r1-like LLM thinking.

No plain-English explanation yet — one is being written right now. Check back in a minute.