Search papers, labs, and topics across Lattice.
1
0
3
A unified post-training framework lets a multimodal diffusion language model master reasoning across visual math, grounding, and image editing tasks without task-specific reinforcement learning.
