Fast and Fluent Diffusion Language Models via Convolutional Decoding and Rejective Fine-tuning

NeurIPSSpotlight2025

Authors: Yeongbin Seo, Dongha Lee, Jaehyung Kim, Jinyoung Yeo
Venue: NeurIPS 2025
Track: Spotlight

TL;DR

Autoregressive (AR) language models generate text one token at a time, which limits their inference speed…

Opening excerpt from the authors’ abstract. source

Read the paper

OpenReview

Topics

language model fine-tuning diffusion

← All NeurIPS 2025 Spotlight papers · Browse the whole archive