AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Learning Hierarchical and Geometry-Aware Graph Representations for Text-to-CAD

Shengjie Gong, Wenjie Peng, Hongyuan Chen, Gangyu Zhang, Yunqing Hu, Huiyuan Zhang, Shuangping Huang, Tianshui Chen

ICLR 2026 · Graph learning › Graph self-supervised learning · approach: Diffusion-based graph generation · 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
add graph decomposition stage (full three-stage vs w/o Graph Decom.) (ablation)Graph-CAD (SFT) · CADBench-SimGCS: 0.7268 -> 0.7830Improved
add action planning stage (vs w/o Action Planning) (ablation)Graph-CAD (SFT) · CADBench-Wildsyntax error rate: 11.0% -> 4.5%Improved
add action planning stage (vs w/o Action Planning) (ablation)Graph-CAD (SFT) · CADBench-SimInst.: 0.4006 -> 0.4733Improved
add both graph and action stages (vs end-to-end) (ablation)Graph-CAD (SFT) · CADBench-SimAvg.: 0.5573 -> 0.6431Improved
add both stages (vs end-to-end) (ablation)Graph-CAD (SFT) · inference timeseconds per sample: 64.861 -> 104.755Hurt
three-stage graph inference with two-shot promptsGPT-5 · CADBench-SimGCS: 0.3846 -> 0.6603Improved
three-stage graph inference with two-shot promptsGPT-5 · CADBench-WildGCS: 0.4017 -> 0.5849Improved
three-stage graph inference with two-shot promptsClaude-opus-4-1 · CADBench-SimGCS: 0.4932 -> 0.5705Improved
three-stage graph inference with two-shot promptsClaude-opus-4-1 · CADBench-SimInst.: 0.5403 -> 0.5025Hurt
three-stage graph inference with two-shot promptsClaude-opus-4-1 · CADBench-SimAvg.: 0.6662 -> 0.6664No clear effect
add SAPCL curriculum learning on top of SFTGraph-CAD · CADBench-SimGCS: 0.7830 -> 0.9018Improved
add SAPCL curriculum learning on top of SFTGraph-CAD · CADBench-WildAvg.: 0.6692 -> 0.7114Improved

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