{"type":"link","version":"1.0","title":"Exact-match evaluation underestimates a zero-shot LLM's true QA accuracy by 58.8 points against human judgment","author_name":"AI Archs","author_url":"https://ai-arch.pages.dev","provider_name":"AI Archs","provider_url":"https://ai-arch.pages.dev","url":"https://ai-arch.pages.dev/n/exact-match-qa-metrics-underestimate-llm-accuracy","thumbnail_url":"https://ai-arch.pages.dev/og/exact-match-qa-metrics-underestimate-llm-accuracy.png","thumbnail_width":1200,"thumbnail_height":630}