Research map / Large language models / Agents, code and math
PlotCraft: Pushing the Limits of LLMs for Complex and Interactive Data Visualization
Jiajun Zhang, Jianke Zhang, Zeyu Cui, Jiaxi Yang, Lei Zhang, Zilei Wang, Qiang Liu, Liang Wang, Binyuan Hui, Junyang Lin
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 |
|---|---|---|---|
| SFT on SynthVis-30K | Qwen3-Coder-30B-A3B · PlotCraft single-turn | pass rate (%): 52.85 -> 64.66 | Improved |
| SFT on SynthVis-30K | Qwen3-Coder-30B-A3B · PlotCraft multi-turn | pass rate (%): 73.42 -> 77.41 | Improved |
| SFT on SynthVis-30K | Qwen3-Coder-30B-A3B · PlotCraft | AVG score: 4.97 -> 6.18 | Improved |
| SFT on SynthVis-30K | Qwen3-Coder-30B-A3B · VisEval | pass rate (%): 89.67 -> 93.81 | Improved |
| SFT on SynthVis-30K | Qwen3-Coder-30B-A3B · PandasPlotBench | pass rate (%): 94.3 -> 96.0 | Improved |
| SFT on SynthVis-30K | Qwen3-Coder-30B-A3B · PlotCraft hard tasks | score | No clear effect |
| SFT on SynthVis-30K | small model · PlotCraft easy tasks | score | Improved |
| scale model size beyond 100B parameters | open and closed LLMs · PlotCraft hard tasks | score | Improved |
| scale model size under 100B parameters | LLMs · PlotCraft easy tasks | score | Improved |
| use multi-judge ensemble instead of single VLM judge (ablation) | PlotCraft evaluation · agreement with human scores | Cohen's kappa | Improved |
| evaluate GPT-oss under ChatML instead of harmony format | GPT-oss-120B · PlotCraft multi-turn | pass rate (%): multi-turn pass rate 29.99 vs single-turn 48.53 | Hurt |
Closest papers by meaning
- CoDA: Agentic Systems for Collaborative Data Visualization (ICLR 2026 · similarity 0.87)
- ChartNet: A Million-Scale, High-Quality Multimodal Dataset for Robust Chart Understanding (CVPR 2026 · similarity 0.87)
- VisInteract: Towards Dynamic Interactive Text-to-Visualization under Imperfect Queries (NeurIPS 2026 · similarity 0.87)
- Multimodal DeepResearcher: Generating Text-Chart Interleaved Reports from Scratch with Agentic Framework (AAAI 2026 · similarity 0.86)
- ChartMimic: Evaluating LMM's Cross-Modal Reasoning Capability via Chart-to-Code Generation (ICLR 2025 · similarity 0.86)
- ChartE$^{3}$: A Comprehensive Benchmark for End-to-End Chart Editing (ICML 2026 · similarity 0.86)
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.
