他们搞了个叫 Schema 的 harness，让 AI 学物理学家那样思考，直接刷爆 ARC-AGI-3 基准，挺有意思的。
Impossible Research 团队推出自定义 agent harness 系统 Schema，能模拟物理学家思维方式。Schema 在 ARC-AGI-3 基准上达到饱和水平，支持游戏、代码编写和复杂推理。该工具旨在解决超困难问题，已公开演示成果。
Start building your harnesses, folks. There is so much alpha in building a harness. Schema is a c...
Start building your harnesses, folks. There is so much alpha in building a harness. Schema is a custom harness that makes an agent "think" like a physicist. It saturates ARC-AGI-3. A custom harness is how you start solving very hard problems with agents. Naval @naval From our team at Impossible Research - an agent harness that plays games, writes code, reasons like a physicist, and saturates the ARC-AGI-3 benchmark. 🔗 View Quoted Tweet 💬 4 🔄 1 ❤️ 9 👀 2016 📊 5 ⚡