@Auriel W

Auriel W, an RL practitioner who has worked at Gemini, warns that low-quality reinforcement learning environments with broken code, race conditions, and caching bugs are systematically ruining model t…