Tsung-Yi Lin

Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning

Scenethesis: A Language and Vision Agentic Framework for 3D Scene Generation

NFT: Bridging Supervised Learning and Reinforcement Learning in Math Reasoning

VoxSet: Sparse Voxel Set Tokenizer for 3D Shape Generation

Efficient Part-level 3D Object Generation via Dual Volume Packing

Meshtron: High-Fidelity, Artist-Like 3D Mesh Generation at Scale