SAS Teaches Diffusion LMs Self-Aware Scheduling
An ICML 2026 paper introduces SAS, a self-aware scheduling method for diffusion language models that rewards unmasking orders using the model’s own frozen path likelihoods. It needs no human-written reasoning traces and boosts Sudoku accuracy from 82.0% to 91.8%, or 97.5% with a second fine-tuning stage, while also improving GSM8K from 64% to 76% and MBPP from 39.5% to 41%.
2026-07-06 ~ 2026-07-07 · 15 related posts
- ICML 2026:扩散语言模型自感知调度推理方法SAS — furongh · 2026-07-06
- 扩散语言模型带来思维顺序的自由 — furongh · 2026-07-06
- 扩散语言模型解码不必从左到右 — furongh · 2026-07-06
- 扩散语言模型的关键是学会何时承诺token — furongh · 2026-07-06
- 现有扩散LM解码调度太短视 — furongh · 2026-07-06
- SAS把解码顺序变成可训练推理策略 — furongh · 2026-07-06
- SAS让调度器自我感知 — furongh · 2026-07-06
- SAS类似无人工轨迹的过程监督 — furongh · 2026-07-06
- SAS在数独上大幅提升解题准确率 — furongh · 2026-07-06
- 人类逻辑顺序未必最适合模型推理 — furongh · 2026-07-06
- SAS收益可迁移到GSM8K与MBPP — furongh · 2026-07-06
- 推理新维度:何时承诺而非生成什么 — furongh · 2026-07-06
- 论文:学习扩散语言模型中的思维顺序 — furongh · 2026-07-06
- ICML2026 海报:LLM 推理与扩散模型 — furongh · 2026-07-06
- 扩散语言模型中的思维调度 — furongh · 2026-07-07