Yunpeng Bai (@Byp215Bai)

2025-10-24 | โค๏ธ 404 | ๐Ÿ” 50


๐Ÿš€ Do world models need explicit 3D? Our answer: if youโ€™re using Transformers, introducing 3D into DiTโ€™s positional encoding is a natural choice.

๐Ÿ“„ Paper: https://arxiv.org/pdf/2510.20385 ๐ŸŒ HomePage: https://yunpeng1998.github.io/PE-Field-HomePage/ ๐Ÿ’ป Code: https://github.com/MTLab/PE-Field https://x.com/Byp215Bai/status/1981809736535208219/photo/1


Auto-generated bookmark

Tags

AI-ML