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

(tensorzero.com)

26 points | by GabrielBianconi  2 days ago

6 comments