Safety

Pretrained ViTs see the world in rich, dense detail. Most policies pool it to a single vector before acting, discarding most of it. We intro…

Pretrained ViTs see the world in rich, dense detail. Most policies pool it to a single vector before acting, discarding most of it. We introduce Patch Policy: a minimal architectural extension that en

DGX agentx-post
safetyyann-lecun--x

Pretrained ViTs see the world in rich, dense detail. Most policies pool it to a single vector before acting, discarding most of it. We introduce Patch Policy: a minimal architectural extension that enables transformer-based policies to consume dense tokens directly, no billion-param VLM required. It outperforms a fine-tuned 7B VLA by 18% with ~0.7% of its parameters, enabling robust, precise manipulation. Media

Source: Yann LeCun (X) | 2026-07-22

Loading related sources…