AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

DiffBench Meets DiffAgent: End-to-End LLM-Driven Diffusion Acceleration Code Generation

Jiajun Jiao, Haowei Zhu, Puyuan Yang, Jianghui Wang, Ji Liu, Ziqiong Liu, Dong Li, Yuejian Fang, Jun-Hai Yong, Bin Wang, Emad Barsoum

AAAI 2026 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 1 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
wrap LLM in DiffAgent (planning, coding, debugging, GA selector)Claude Sonnet 4 · DiffBench averagepass rate Sp (%): 54.30 -> 81.59Improved
wrap LLM in DiffAgento3-mini · DiffBench averagepass rate Sp (%): 11.92 -> 56.46Improved
add genetic algorithm selector (ablation)DiffAgent · DiffBench Level 4pass rate Sp (%): 4.30 -> 33.33Improved
add genetic algorithm selector (ablation)DiffAgent · DiffBench Level 5pass rate Sp (%): 4.30 -> 63.44Improved
add genetic algorithm selector (ablation)DiffAgent · DiffBench Levels 1-3pass rate Sp (%): 90.24, 91.38, 99.23 unchangedNo clear effect
add knowledge base (ablation)DiffAgent · DiffBench averagepass rate Sp (%): 64.90 -> 81.59Improved
add debugging agent (ablation)DiffAgent · DiffBench averagepass rate Sp (%): 66.23 -> 81.59Improved
add debugging agent (ablation)DiffAgent · DiffBench Level 4 hard tasksachievement rate Sa (%): 56.99 vs 56.61No clear effect
add debugging agent (ablation)DiffAgent · DiffBench Level 5pass rate Sp (%): 33.33 -> 63.44Improved
increase GA population size beyond 7 (ablation)DiffAgent · DiffBench Levels 4-5pass rate Sp (%): 33.33 -> 37.63 (Tsel=4, P=7 -> 10)No clear effect
increase GA iterations Tsel from 2 to 4 (ablation)DiffAgent (P=7) · DiffBench Levels 4-5pass rate Sp (%): 30.11 -> 33.33Improved
add DiffAgent to Claude Sonnet 4 (error analysis)Claude Sonnet 4 · DiffBench Level 5compile error rate (%): 31.18 -> 2.15Improved

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