DUEL: Exact Likelihood for Masked Diffusion via Deterministic Unmasking
Gilad Turok, Chris De Sa, Volodymyr Kuleshov · 2026
Read DUEL: Exact Likelihood for Masked Diffusion via Deterministic Unmasking by Gilad Turok, Chris De Sa, Volodymyr Kuleshov (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-01367.txt.
Description
Turok, De Sa, and Kuleshov offer DUEL, a plan joining masked diffusion samplers that use fixed spot selection. They prove DUEL samplers admit exact odds math under the test-time spread, giving masked diffusion models proper odds and proper surprise for the first time. That proper surprise matches auto surprise as a natural twin, letting the team revisit masked diffusion questions. Masked diffusion models turn out much better: the surprise gap to auto models shrinks by up to 32 percent on in-domain data and 82 percent on zero-shot benchmarks.
Catalog record
| CREATOR | Gilad Turok, Chris De Sa, Volodymyr Kuleshov |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80747 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.01367