{"type":"link","version":"1.0","title":"Lossy compression of a reused attention cache produces correct-looking output that silently diverges after the first several tokens","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/risk-lossy-kv-compression-silently-breaks-token-exactness","thumbnail_url":"https://ai-arch.pages.dev/android-chrome-512x512.png","thumbnail_width":512,"thumbnail_height":512}