Research map / Large language models / Agents, code and math
Learn2Fold: Structured Origami Generation with World Model Planning
Yanjia Huang, Yunuo Chen, Ying Jiang, Zhengzhong Tu, Yin Yang, Chenfanfu Jiang
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 |
|---|---|---|---|
| use Learn2Fold instead of strongest prompted baseline GPT-5.1 | Learn2Fold · OrigamiCode test (3,840 prompts, 25 categories) | F1: 0.2663 -> 0.7394 | Improved |
| use Learn2Fold instead of Gemini on Edge-IoU | Learn2Fold · OrigamiCode test (3,840 prompts, 25 categories) | Edge-IoU: 0.1126 -> 0.5820 | Improved |
| use Learn2Fold instead of GPT-5.1 on category success | Learn2Fold · OrigamiCode test (3,840 prompts, 25 categories) | Cat-SR: 0.6753 -> 0.8912 | Improved |
| add learned world model to LM proposer (ablation) | LM · Ablation (IID CP holdout) | Traj SR: 22.2% -> 25.0% | Improved |
| add learned world model to LM proposer (ablation) | LM · Ablation (IID CP holdout) | goal distance: 0.796 -> 0.759 | Improved |
| add learned world model to LM proposer (ablation) | LM · Ablation (OOD CP holdout) | goal distance: 0.633 -> 0.560 | Improved |
| add learned world model to LM proposer (ablation) | LM · Ablation (IID CP holdout) | step validity: 70.8% -> 54.2% | Hurt |
| add learned world model to LM proposer (ablation) | LM · Ablation (OOD CP holdout) | step validity: 47.6% -> 32.3% | Hurt |
| add learned world model to LM proposer (ablation) | LM · Ablation (OOD CP holdout) | Traj SR: 20.7% -> 17.8% | Hurt |
| add Level0Sim constraint kernel to LM+WM (ablation) | LM+WM · Ablation (IID CP holdout) | Traj SR: 25.0% -> 33.3% | Improved |
| add Level0Sim constraint kernel to LM+WM (ablation) | LM+WM · Ablation (OOD CP holdout) | Traj SR: 17.8% -> 27.7% | Improved |
| add Level0Sim constraint kernel to LM+WM (ablation) | LM+WM · Ablation (OOD CP holdout) | goal distance: 0.560 -> 0.487 | Improved |
Closest papers by meaning
- ORIGAMISPACE: Benchmarking Multimodal LLMs in Multi-Step Spatial Reasoning with Mathematical Constraints (NeurIPS 2025 · similarity 0.87)
- Localizing, Structuring, and Rendering: Bridging 3D and 2D Vision-Language-Action Models for Robotic Manipulation (CVPR 2026 · similarity 0.82)
- Knot So Simple: A Minimalistic Environment for Spatial Reasoning (NeurIPS 2025 · similarity 0.82)
- Scaling-Aware Adapter for Structure-Grounded LLM Reasoning (ICML 2026 · similarity 0.82)
- LayoutVLM: Differentiable Optimization of 3D Layout via Vision-Language Models (CVPR 2025 · similarity 0.82)
- SciCode: A Research Coding Benchmark Curated by Scientists (NeurIPS 2024 · similarity 0.82)
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.
