We evaluate VisCoder2-3B on
VisPlotBench, which includes 888 executable visualization tasks spanning 8 languages, supporting both standard generation and multi-turn self-debugging.
1@article{ni2025viscoder2,
2 title={VisCoder2: Building Multi-Language Visualization Coding Agents},
3 author={Ni, Yuansheng and Cai, Songcheng and Chen, Xiangchao and Liang, Jiarong and Lyu, Zhiheng and Deng, Jiaqi and Zou, Kai and Nie, Ping and Yuan, Fei and Yue, Xiang and others},
4 journal={arXiv preprint arXiv:2510.23642},
5 year={2025}
6}
7
8@article{ni2025viscoder,
9 title={VisCoder: Fine-Tuning LLMs for Executable Python Visualization Code Generation},
10 author={Ni, Yuansheng and Nie, Ping and Zou, Kai and Yue, Xiang and Chen, Wenhu},
11 journal={arXiv preprint arXiv:2506.03930},
12 year={2025}
13}
For evaluation scripts and more information, see our
GitHub repository.
