Introspective Diffusion Language Models
Yifan Yu · Yuqing Jian · Junxiong Wang · Zhongzhu Zhou · Donglin Zhuang · Xinyu Fang · Xiaoxia Wu · Qingyang Wu · Shuaiwen Leon Song · Tri Dao · Ben Athiwaratkun · James Zou · Fan Lai · Chenfeng Xu
We show that diffusion language models lack introspective consistency, and propose I-DLM to restore AR-style self-agreement while retaining parallel decoding, yielding the first DLM that matches same-scale AR quality with strong efficiency.
摘要
Diffusion language models promise parallel generation, yet still lag behind autoregressive (AR) models in quality. We attribute this gap to a failure of introspective consistency: AR models agree with their own generations, while DLMs often do not. We define the introspective acceptance rate, which measures whether a model accepts its previously generated tokens. This reveals why AR training has a structural advantage: causal masking and logit shifting implicitly enforce introspective consistency. Motivated by this observation, we introduce Introspective Diffusion Language Model (I-DLM), a paradigm that retains diffusion-style parallel decoding while inheriting the introspective consistency of AR training. I-DLM uses a novel introspective strided decoding (ISD) algorithm, which enables the model to verify previously generated tokens while advancing new ones in the same forward pass. From a systems standpoint, we build an I-DLM inference engine on AR-inherited optimizations and further customize it with a stationary-batch scheduler. To the best of our knowledge, I-DLM is the first DLM to match the quality of its same-scale AR counterpart while outperforming prior DLMs in both model quality and practical serving efficiency across 15 benchmarks. It reaches 69.6 on AIME-24 and 45.7 on LiveCodeBench-v6, exceeding LLaDA-2.1-mini (16B) by more than 26 and 15 points, respectively. Beyond quality, I-DLM is designed for the growing demand of large-concurrency serving, delivering about 3× higher throughput than state-of-the-art DLMs.