Hacker News
by Ryan Harman
Even (very) noisy LLM evaluators are useful for improving AI agents
(tensorzero.com)
10 points by GabrielBianconi 2 days ago