Research map / Large language models / Language, safety and interpretability
Plan then Act: Bi-level CAD Command Sequence Generation
Qiangya Guo, Gang Dai, Zhuoman Liu, Shuangping Huang, Yunqing Hu, Huiyuan Zhang, Tianshui Chen
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 |
|---|---|---|---|
| add high-level plan and requirement-aware Actioner over Text2CAD baseline | Text-to-CAD generation · Text2CAD Abstract (L0) | median Chamfer distance: 187.31 -> 113.79 | Improved |
| add high-level plan and RAM versus Text2CAD | Text-to-CAD generation · Text2CAD Beginner (L1) | median Chamfer distance: 206.28 -> 142.50 | Improved |
| use PTA instead of Text2CAD | Text-to-CAD generation · Text2CAD Abstract (L0) | JSD: 130.13 -> 26.35 | Improved |
| use PTA instead of Text2CAD at expert level | Text-to-CAD generation · Text2CAD Expert (L3) | GPT-4V preference: 25.80 -> 30.50 | Improved |
| add high-level plan to the Actioner (bi-level vs Actioner only) (ablation) | PTA · Text2CAD Abstract (L0) | median Chamfer distance: 200.32 -> 113.79 | Improved |
| add the Actioner stage to a fine-tuned LLM Planner generating commands directly (ablation) | PTA · Text2CAD Abstract (L0) | invalid rate: 20.06 -> 0.63 | Improved |
| add requirement-aware mechanism instead of concatenating features (ablation) | PTA Actioner · Text2CAD Abstract (L0) | median Chamfer distance: 159.98 -> 113.79 | Improved |
| add requirement-aware mechanism instead of adding features (ablation) | PTA Actioner · Text2CAD Abstract (L0) | median Chamfer distance: 163.27 -> 113.79 | Improved |
| add user instruction features to plan-only guidance (ablation) | PTA Actioner · Text2CAD Abstract (L0) | F1: 28.82 -> 60.54 | Improved |
| use high-level plan instead of nli detailed data as plan (ablation) | PTA · Text2CAD Abstract (L0) | median Chamfer distance: 156.00 -> 113.79 | Improved |
| fine-tune the Planner rather than prompting only (ablation) | PTA Planner · Text2CAD Abstract (L0) | median Chamfer distance: 124.51 -> 113.79 | Improved |
| use Qwen3-8B instead of Qwen2.5-7B-Instruct as Planner base (ablation) | PTA Planner · Text2CAD Abstract (L0) | median Chamfer distance: 115.97 -> 113.79 | No clear effect |
Closest papers by meaning
- CAD-Llama: Leveraging Large Language Models for Computer-Aided Design Parametric 3D Model Generation (CVPR 2025 · similarity 0.88)
- Text2CAD: Generating Sequential CAD Designs from Beginner-to-Expert Level Text Prompts (NeurIPS 2024 · similarity 0.88)
- FlexCAD: Unified and Versatile Controllable CAD Generation with Fine-tuned Large Language Models (ICLR 2025 · similarity 0.87)
- Clarify Before You Draw: Proactive Agents for Robust Text-to-CAD Generation (ICML 2026 · similarity 0.86)
- Text-to-CAD Generation Through Infusing Visual Feedback in Large Language Models (ICML 2025 · similarity 0.85)
- CAD-Editor: A Locate-then-Infill Framework with Automated Training Data Synthesis for Text-Based CAD Editing (ICML 2025 · similarity 0.85)
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.
