|
|
--- |
|
|
base_model: |
|
|
- Qwen/Qwen2.5-Coder-7B-Instruct |
|
|
datasets: |
|
|
- TIGER-Lab/VisCode-200K |
|
|
language: |
|
|
- en |
|
|
license: apache-2.0 |
|
|
tags: |
|
|
- code |
|
|
library_name: transformers |
|
|
pipeline_tag: text-generation |
|
|
--- |
|
|
|
|
|
# VisCoder-7B |
|
|
|
|
|
[🏠 Project Page](https://tiger-ai-lab.github.io/VisCoder) | [📖 Paper](https://arxiv.org/abs/2506.03930) | [💻 GitHub](https://github.com/TIGER-AI-Lab/VisCoder) | [🤗 VisCode-200K](https://huggingface.co/datasets/TIGER-Lab/VisCode-200K) | [🤗 VisCoder-3B](https://huggingface.co/TIGER-Lab/VisCoder-3B) |
|
|
|
|
|
**VisCoder-7B** is a large language model fine-tuned for **Python visualization code generation and multi-turn self-correction**. It is trained on **VisCode-200K**, a large-scale instruction-tuning dataset that integrates validated executable code, natural language instructions, and revision supervision from execution feedback. |
|
|
|
|
|
|
|
|
## 🧠 Model Description |
|
|
|
|
|
**VisCoder-7B** is trained on **VisCode-200K**, a large-scale instruction-tuning dataset tailored for executable Python visualization tasks. It addresses a core challenge in data analysis: generating Python code that not only executes successfully but also produces **semantically meaningful plots** by aligning **natural language instructions**, **data structures**, and **visual outputs**. |
|
|
|
|
|
We propose a **self-debug evaluation protocol** that simulates real-world developer workflows. In this setting, models are allowed to revise previously failed generations over multiple rounds with guidance from **execution feedback**. |
|
|
|
|
|
## 📊 Main Results on PandasPlotBench |
|
|
|
|
|
We evaluate VisCoder-7B on [**PandasPlotBench**](https://github.com/TIGER-AI-Lab/VisCoder/tree/main/eval), which tests executable visualization code generation across three major libraries. Our benchmark covers both standard generation and **multi-round self-debugging**. |
|
|
|
|
|
 |
|
|
|
|
|
> VisCoder-7B achieves over **90% execution pass rate** on both **Matplotlib** and **Seaborn** under the self-debug setting, outperforming open-source baselines and approaching GPT-4o performance. |
|
|
|
|
|
|
|
|
## 📁 Training Details |
|
|
|
|
|
- **Base model**: Qwen2.5-Coder-7B-Instruct |
|
|
- **Framework**: [ms-swift](https://github.com/modelscope/swift) |
|
|
- **Tuning method**: Full-parameter supervised fine-tuning (SFT) |
|
|
- **Dataset**: [VisCode-200K](https://huggingface.co/datasets/TIGER-Lab/VisCode-200K), which includes: |
|
|
- 150K+ validated Python visualization samples with images |
|
|
- 45K+ multi-turn correction dialogues with execution feedback |
|
|
|
|
|
## 📖 Citation |
|
|
|
|
|
If you use VisCoder-7B or VisCode-200K in your research, please cite: |
|
|
|
|
|
```bibtex |
|
|
@article{ni2025viscoder, |
|
|
title={VisCoder: Fine-Tuning LLMs for Executable Python Visualization Code Generation}, |
|
|
author={Ni, Yuansheng and Nie, Ping and Zou, Kai and Yue, Xiang and Chen, Wenhu}, |
|
|
journal={arXiv preprint arXiv:2506.03930}, |
|
|
year={2025} |
|
|
} |
|
|
``` |
|
|
|
|
|
For evaluation scripts and more information, see our [GitHub repository](https://github.com/TIGER-AI-Lab/VisCoder). |
