AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Scaling Behavior of Discrete Diffusion Language Models

Dimitri von Rütte, Janis Fluri, Omead Pooladzandi, Bernhard Schölkopf, Thomas Hofmann, Antonio Orvieto

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Discrete diffusion solvers · 25 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
move from masked to uniform noise (ablation)discrete diffusion LM · compute-optimal scalingmodel-size exponent alpha_M: 0.566 -> 0.589Improved
move from masked to uniform noise (ablation)discrete diffusion LM · compute-bound loss scalingloss exponent alpha_L: -0.0496 -> -0.0522Improved
increase uniform noise share (ablation)discrete diffusion LM · likelihood vs computeELBO gap to masked: 3.2% -> 1.7% (1e18 -> 1e21 FLOPs)Improved
add 20% LR cooldown (annealing) (ablation)discrete diffusion LM · ELBOELBO improvement: 2.45%Improved
add confidence-based adaptive sampling instead of ancestral sampling (ablation)Masking 3B · GSM8kaccuracy (%): 1.06 -> 1.67Improved
add confidence-based adaptive sampling instead of ancestral sampling (ablation)Uniform 3B · GSM8kaccuracy (%): 1.44 -> 2.05Improved
use uniform instead of masked 3B model3B DLM · downstream multiple-choice benchmarksaccuracy (%): ARC-E 49.9 -> 50.6; BoolQ 60.9 -> 56.4Mixed
scale uniform diffusion from 3B to 10Buniform DLM · ARC-Easyaccuracy (%): 50.6 -> 61.8Improved
fix optimal batch size by training tokensdiscrete diffusion LM · optimal batch sizepower-law exponent: 0.8225Improved

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