English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

DiVR: Learning When Diffusion Language Models Should Regenerate

Forum topic · 小凯 · 2026-11-13

Summary

This paper, posted on zhichai.net, introduces DiVR, a verifier for diffusion language models that decides when remasking-based regeneration is needed and which text blocks to keep or revise. The authors first diagnose two gaps on Sudoku tasks: token confidence does not reliably expose error information contained in hidden states, and remasking an isolated error typically reproduces it when the surrounding context stays fixed. DiVR is trained on the final correctness of the model's own generation trajectories to predict whether a partially generated sequence will lead to a correct answer, without modifying the generator. On five reasoning benchmarks using LLaDA-8B-Base, DiVR raises the average accuracy of a fine-tuned generator from 35.3% to 43.4%, with inference costs of only 1.9-3.5x greedy decoding, outperforming search baselines that require 3.8-7.4x. The method is complementary to post-training approaches such as supervised fine-tuning and reinforcement learning, and generalizes to other diffusion generators like Dream. arXiv: https://arxiv.org/abs/2611.17320.

Paper Overview

Field: NLP Authors: Minhae Oh, Jeonghye Kim, Jinwoo Koo, Yeongoon Kim, Jimin Seo, Minsik Seol, Jungwoo Lee Published: 2026-11-12 arXiv: 2611.17320

Abstract (translated)

Remasking allows diffusion language models to revisit generated text, but effective correction requires knowing when and what to revise. The authors' diagnosis on Sudoku reveals two gaps: hidden states contain error information that token confidence does not reliably expose, and remasking an isolated error usually reproduces it when the surrounding context remains fixed. These findings motivate DiVR, a verifier that predicts whether a partially generated sequence will reach a correct answer. Trained on the final correctness of the model's own trajectories, DiVR decides when to regenerate a block and which alternative to retain, while the generator itself stays unchanged.

Key Results

  • On five reasoning benchmarks with LLaDA-8B-Base, DiVR raises the average accuracy of the fine-tuned generator from 35.3% to 43.4%.
  • Inference cost is only 1.9-3.5x greedy decoding, compared with 3.8-7.4x for search baselines.
  • DiVR is complementary to post-training methods such as supervised fine-tuning and reinforcement learning.
  • The approach generalizes to other diffusion generators, including Dream.

Discussion

The core insight is that a verifier trained on trajectory outcomes can guide remasking decisions better than token-level confidence, since confidence scores fail to reflect latent errors. By regenerating whole blocks rather than isolated tokens, DiVR avoids the failure mode where a corrected token simply reverts under unchanged context.

--- *Auto-collected on 2026-11-14.*

Tags

#diffusion-language-models#remasking#verifier#reasoning#llada#nlp#paper

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178637113