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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| wrap LLM in DiffAgent (planning, coding, debugging, GA selector) | Claude Sonnet 4 · DiffBench average | pass rate Sp (%): 54.30 -> 81.59 | Improved |
| wrap LLM in DiffAgent | o3-mini · DiffBench average | pass rate Sp (%): 11.92 -> 56.46 | Improved |
| add genetic algorithm selector (ablation) | DiffAgent · DiffBench Level 4 | pass rate Sp (%): 4.30 -> 33.33 | Improved |
| add genetic algorithm selector (ablation) | DiffAgent · DiffBench Level 5 | pass rate Sp (%): 4.30 -> 63.44 | Improved |
| add genetic algorithm selector (ablation) | DiffAgent · DiffBench Levels 1-3 | pass rate Sp (%): 90.24, 91.38, 99.23 unchanged | No clear effect |
| add knowledge base (ablation) | DiffAgent · DiffBench average | pass rate Sp (%): 64.90 -> 81.59 | Improved |
| add debugging agent (ablation) | DiffAgent · DiffBench average | pass rate Sp (%): 66.23 -> 81.59 | Improved |
| add debugging agent (ablation) | DiffAgent · DiffBench Level 4 hard tasks | achievement rate Sa (%): 56.99 vs 56.61 | No clear effect |
| add debugging agent (ablation) | DiffAgent · DiffBench Level 5 | pass rate Sp (%): 33.33 -> 63.44 | Improved |
| increase GA population size beyond 7 (ablation) | DiffAgent · DiffBench Levels 4-5 | pass 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 4-5 | pass rate Sp (%): 30.11 -> 33.33 | Improved |
| add DiffAgent to Claude Sonnet 4 (error analysis) | Claude Sonnet 4 · DiffBench Level 5 | compile error rate (%): 31.18 -> 2.15 | Improved |
Closest papers by meaning
- Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference (NeurIPS 2024 · similarity 0.90)
- AdaDiff: Adaptive Step Selection for Fast Diffusion Models (AAAI 2025 · similarity 0.90)
- A Simple Early Exiting Framework for Accelerated Sampling in Diffusion Models (ICML 2024 · similarity 0.90)
- Plug-and-Play Diffusion Distillation (CVPR 2024 · similarity 0.89)
- Towards Precise Scaling Laws for Video Diffusion Transformers (CVPR 2025 · similarity 0.89)
- CacheQuant: Comprehensively Accelerated Diffusion Models (CVPR 2025 · similarity 0.88)
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.
