AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Continuous Diffusion Models Can Obey Formal Syntax

Jinwoo Kim, Taylor Berg-Kirkpatrick, Loris D'Antoni

ICML 2026 · Image generation › Fast diffusion sampling · approach: Masked diffusion language model decoding · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Diffinity instead of GPT2-Large constrained decodingPLAID · Natural language benchmarksconstraint satisfaction (%): 20.5 -> 92.9Improved
use Diffinity instead of GPT2-Large GuidancePLAID · JSON schemaconstraint satisfaction (%): 77.0 -> 68.4Hurt
use Diffinity instead of LLaDA+DINGOPLAID · JSON schemapass@10 (%): 100.0 -> 91.4Hurt
add DINGO constrained decoding (ablation)LLaDA · Natural language benchmarksperplexity: 23.9 -> 222.8Hurt
add Diffinity guidance (scale 2.5) (ablation)PLAID · Natural language benchmarksperplexity: 61.6 -> 60.0No clear effect
add Diffinity guidance (scale 2.5) (ablation)PLAID · Natural language benchmarksfluency: 46.0 -> 35.8Hurt
use <json><EOS>.* padding instead of .*<json>.* padding (ablation)Diffinity · JSON schemaconstraint satisfaction (%): 68.4 -> 56.8Hurt
use <json><EOS>+ padding instead of .*<json>.* padding (ablation)Diffinity · JSON schemaconstraint satisfaction (%): 68.4 -> 33.6Hurt
replace regex-based expected-probability guidance with classifier guidance (full) (ablation)Diffinity · Natural language benchmarksconstraint satisfaction (%): 92.9 -> 2.27Hurt
replace regex-based guidance with classifier guidance (ablation)Diffinity · JSON schemaconstraint satisfaction (%): 68.4 -> 0.00Hurt
use Diffinity instead of GPT2-Small-GCD for sentence prefix distributionPLAID · The versus It prefix regexThe:It sample ratio: 1.2 -> 0.03Improved
use Diffinity instead of GPT2-Large for perplexity on PrefixPLAID · Natural language: Prefixperplexity: 401.7 -> 66.5Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy