Niklas Nolte

Researcher, Facebook

1 paper at NeurIPS 2025

Homepage· OpenReview· Semantic Scholar· Google Scholar

Poster Session 4

1 paper

Thursday, December 4, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

Accelerated Sampling from Masked Diffusion Models via Entropy Bounded Unmasking

#3615 · Heli Ben-Hamu, Itai Gat, Daniel Severo, Niklas Nolte, Brian Karrer

We propose an adaptive multi-token unmasking sampler for masked language diffusion models, gaining speed up of 2-3x on code generation and math reasoning benchmarks without loss in accuracy.