{"type":"link","version":"1.0","title":"Pure task-outcome reward fails to train memory-operation policies, even with adequate trajectory data","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-sparse-reward-collapses-memory-operation-credit","thumbnail_url":"https://ai-arch.pages.dev/og/risk-sparse-reward-collapses-memory-operation-credit.png","thumbnail_width":1200,"thumbnail_height":630}