{"type":"link","version":"1.0","title":"On-policy learning re-pays the full cost of a long sequence on every update; a fixed dataset pass pays it once","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/onpolicy-rl-reruns-full-sequence-cost-every-update","thumbnail_url":"https://ai-arch.pages.dev/og/onpolicy-rl-reruns-full-sequence-cost-every-update.png","thumbnail_width":1200,"thumbnail_height":630}