扩散语言模型让放射科报告可任意补全，速度还更快。
Discrete Diffusion Language Models for Interactive Radiology Report Drafting
- 用双向去噪生成文本，突破传统逐词生成限制
- 在医疗问答任务中表现持平甚至超越自回归模型，提速3.5-4.4倍
- 支持任意顺序补全片段，适合不完整或不一致的临床报告
扩散语言模型通过双向去噪重构词元画布生成文本，已具备与自回归（AR）模型竞争的能力。然而，医学基础模型仍几乎全部采用自回归方式。本文将混合专家扩散语言模型 DiffusionGemma-26B 与同规模自回归模型 Gemma-4-26B 在相同的 LoRA 微调策略下进行对比，测试其在医疗视觉问答数据集上的表现，并由一个抗冗余的 LLM 判官评分。结果显示，扩散模型在所有任务上均达到或超过自回归模型，微调后活跃参数为 3.8B，性能媲美前沿多模态模型；其解码速度提升 3.5 至 4.4 倍。更重要的是，该模型具备自回归模型不具备的「任意顺序填充」能力：因画布双向去噪，放射科医生可先固定报告片段，由模型自动填补中间内容。这一特性契合真实报告中常出现的简略或不一致问题。
原文摘要 · Abstract (English)
Diffusion language models, which generate text by denoising a token canvas bidirectionally instead of emitting tokens left to right, have become competitive with autoregressive (AR) generation. Medical foundation models, however, remain almost entirely autoregressive. We adapt a mixture-of-experts diffusion language model, DiffusionGemma-26B, and benchmark it against its same-size AR sibling Gemma-4-26B under an identical LoRA recipe on medical visual question answering datasets, scored by a verbosity-robust LLM judge. Diffusion matches or exceeds AR on all of them, and the finetuned model (3.8B active) is competitive with frontier vision-language models; its decoding is also 3.5-4.4x faster. Beyond this parity, the diffusion model offers a drafting capability AR lacks: any-order infill. Because the canvas is denoised bidirectionally, a radiologist can fix report fragments and have the model fill the text between them, an operation inherent to diffusion but not to autoregression, which is subpar at it. This suits real reports, which are often terse or inconsistent across clinicians and institutions.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。