Confidence-based decoding and training in masked diffusion models shortcut long-range dependencies in reasoning, producing errors on complex inputs that random masking avoids.
Think first, diffuse fast: Improving diffusion language model reasoning via autoregressive plan conditioning.arXiv preprint arXiv:2603.13243,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
The Confidence Shortcut: A Reasoning Failure Mode of Masked Diffusion Models
Confidence-based decoding and training in masked diffusion models shortcut long-range dependencies in reasoning, producing errors on complex inputs that random masking avoids.