Research map / Large language models / Language, safety and interpretability
Improving Text-guided CAD Prototyping via Modality-Specific Tokenization
Ruiyu Wang, Shizhao Sun, Weijian Ma, Jiang Bian
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 primitive pooling (curve-based) to the tokenizer instead of single sketch-extrusion unit (ablation) | CAD-Tokenizer · Text-to-CAD | F1-Avg: 78.3 -> 86.5 | Improved |
| add primitive pooling (curve-based) to the tokenizer instead of single sketch-extrusion unit (ablation) | CAD-Tokenizer · Text-to-CAD | invalidity ratio: 70.7 -> 4.94 | Improved |
| pool loops instead of curves (ablation) | CAD-Tokenizer · Text-to-CAD | F1-Avg: 86.5 -> 86.3 | No clear effect |
| use pooling loops instead of curves for reconstruction (ablation) | CAD-Tokenizer · reconstruction | F1-Skt: 94.1 -> 91.5 | Hurt |
| replace BPE with CAD-Tokenizer (ablation) | LLaMA SFT · Text-to-CAD | F1-Avg: 76.2 -> 86.5 | Improved |
| add FSA-guided decoding instead of top-p sampling (ablation) | CAD-Tokenizer · Text-to-CAD | F1-Avg: 80.4 -> 86.5 | Improved |
| add FSA-guided decoding instead of beam search (ablation) | CAD-Tokenizer · Text-to-CAD | invalidity ratio: 45.2 -> 4.94 | Improved |
| use CAD-Tokenizer instead of Vanilla-LLaMA tokenizer on editing | LLaMA · CAD editing | F1-Skt: 78.8 -> 88.6 | Improved |
| use CAD-Tokenizer instead of Vanilla-LLaMA tokenizer on generation | LLaMA · Text-to-CAD | invalidity ratio: 80.5 -> 1.50 | Improved |
| compare CAD-Tokenizer against task-specific CADFusion on VLM score | CADFusion · Text-to-CAD | VLM score: 5.41 -> 3.82 | Hurt |
| use CAD-Tokenizer instead of task-specific CAD-Editor | CAD-Editor · CAD editing | F1-Skt: 73.3 -> 88.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CAD-Editor — CAD-Editor: A Locate-then-Infill Framework with Automated Training Data Synthesis for Text-Based CAD Editing (ICML 2025) · better on 3 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- CAD-Editor: A Locate-then-Infill Framework with Automated Training Data Synthesis for Text-Based CAD Editing (ICML 2025 · similarity 0.91)
- Text2CAD: Generating Sequential CAD Designs from Beginner-to-Expert Level Text Prompts (NeurIPS 2024 · similarity 0.89)
- FlexCAD: Unified and Versatile Controllable CAD Generation with Fine-tuned Large Language Models (ICLR 2025 · similarity 0.88)
- CAD-Llama: Leveraging Large Language Models for Computer-Aided Design Parametric 3D Model Generation (CVPR 2025 · similarity 0.88)
- CAD-Coder: Text-to-CAD Generation with Chain-of-Thought and Geometric Reward (NeurIPS 2025 · similarity 0.87)
- Clarify Before You Draw: Proactive Agents for Robust Text-to-CAD Generation (ICML 2026 · similarity 0.87)
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.
