{"type":"link","version":"1.0","title":"Reinforcement learning that rewards every attempt collapses a model onto greedy answers instead of in-context exploration","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-single-turn-rl-collapses-in-context-exploration","thumbnail_url":"https://ai-arch.pages.dev/og/risk-single-turn-rl-collapses-in-context-exploration.png","thumbnail_width":1200,"thumbnail_height":630}