PhD student, Massachusetts Institute of Technology
1 paper at NeurIPS 2025
We propose Hierarchical Diffusion Language Models, which is a discrete diffusion with a general time-varying next semantic scale prediction process for language modeling.