Zi Ye

Today's best multimodal LLMs are surprisingly inept at using tools to solve agricultural tasks, struggling with everything from planning to error recovery.

Finally, AI characters can convincingly listen, react, and speak in real-time, opening the door to truly immersive conversational agents and virtual beings.

Injecting physics-based priors derived from MLLMs at decoding time significantly boosts weather forecasting accuracy and stability, even in long autoregressive rollouts.