Papers
arxiv:2601.22954

Residual Context Diffusion Language Models

Published on Jan 30
· Submitted by
Yuezhou Hu
on Feb 5
Authors:
,
,
,
,
,
,
,
,
,
,
,

Abstract

Residual Context Diffusion (RCD) enhances diffusion large language models by recycling discarded token information through contextual residuals, improving accuracy with minimal computational overhead.

AI-generated summary

Diffusion Large Language Models (dLLMs) have emerged as a promising alternative to purely autoregressive language models because they can decode multiple tokens in parallel. However, state-of-the-art block-wise dLLMs rely on a "remasking" mechanism that decodes only the most confident tokens and discards the rest, effectively wasting computation. We demonstrate that recycling computation from the discarded tokens is beneficial, as these tokens retain contextual information useful for subsequent decoding iterations. In light of this, we propose Residual Context Diffusion (RCD), a module that converts these discarded token representations into contextual residuals and injects them back for the next denoising step. RCD uses a decoupled two-stage training pipeline to bypass the memory bottlenecks associated with backpropagation. We validate our method on both long CoT reasoning (SDAR) and short CoT instruction following (LLaDA) models. We demonstrate that a standard dLLM can be efficiently converted to the RCD paradigm with merely ~1 billion tokens. RCD consistently improves frontier dLLMs by 5-10 points in accuracy with minimal extra computation overhead across a wide range of benchmarks. Notably, on the most challenging AIME tasks, RCD nearly doubles baseline accuracy and attains up to 4-5x fewer denoising steps at equivalent accuracy levels.

Community

Paper author Paper submitter

We introduce Residual Context Diffusion (RCD): a simple idea to boost diffusion LLMs—stop wasting “remasked” tokens.

Diffusion LLMs decode in parallel but often lag AR models because low-confidence tokens are discarded each step. RCD turns those discarded distributions into residual context and injects them into the next denoising step, recycling computation instead of throwing it away.

Results: consistent gains over Sequential Denoising (SeqD) on SDAR & LLaDA, with biggest jumps on hard math reasoning (AIME24/25, MinervaMath).

Sign up or log in to comment

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2601.22954 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2601.22954 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2601.22954 in a Space README.md to link it from this page.

Collections including this paper 1