AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

d3LLM: Ultra-Fast Diffusion LLM using Pseudo-Trajectory Distillation

Yu-Yang Qian, Junda Su, Lanxiang Hu, Peiyuan Zhang, Zhijie Deng, Peng Zhao, Hao Zhang

ICML 2026 · Large language models › Language, safety and interpretability · approach: Parallel and early-exit 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
add pseudo-trajectory distillation (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedTPF: 6.41 -> 7.55Improved
add pseudo-trajectory distillation (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 441.4 -> 517.7Improved
add curriculum noise (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedTPF and accuracy: TPF 7.55 -> 8.46; Acc 72.1 -> 69.8Mixed
add curriculum window (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 551.3 -> 637.7Improved
add multi-block decoding and early stopping (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 492.9 -> 637.7Improved
add early stopping on EOS (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 635.0 -> 637.7Improved
use curriculum noise 0.0->0.8 instead of fixed t=0.5 (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 521.7 -> 637.7Improved
use curriculum noise 0.0->0.5 instead of 0.2->0.5 (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 557.7 -> 616.9Improved
use curriculum window 16->32 instead of fixed 32 (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 536.0 -> 637.7Improved
start window curriculum at 16 instead of 0 or 24 (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 603.1 / 584.9 -> 637.7Improved
swap exponential AUP weighting for power or linear (ablation)AUP metric ranking · GSM8K-CoT (0-shot), LLaDA-basedmethod ranking: d3LLM top in all nine configurationsNo clear effect
raise AUP penalty alpha from 1 to 10 (ablation)d3LLM-LLaDA · GSM8K-CoT (0-shot), LLaDA-basedAUP score: 659.4 -> 568.4Hurt

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