←

Hacker News

Even (very) noisy LLM evaluators are useful for improving AI agents

11 points0 comments2 days ago