2026-10-27 · America/Los_Angeles · 社区动态 · #4

Ψ-0.6: Adaptive Streams of Thought in a Unified Vision-Language World Model [R]

Chain-of-thought transformed AI reasoning, but it is limited to words. What's the physical-world analogue? Introducing Ψ-0.6, one model that reasons in *streams of thought*: words, depth, motion, camera pose, and imagined scenes. And it learns when to use which! Ψ-0.6 surpasses leading open-weight models on complex spatial reasoning (MindCube, zero-shot). On FlowBench, it outperforms other world models at predicting how objects move. Trained on simulated rooms, it transfers to real robot tabletops. One model as perceiver, VLM, world model, reasoner, and policy. RL learns which representation…

热度 50.2 / 100；排名与评分保留该期记录。