---
license: apache-2.0
language:
- en
base_model:
- Qwen/Qwen3-VL-4B-Instruct
pipeline_tag: image-text-to-text
library_name: transformers
tags:
- chart
- reasoning
- vision-language
- multimodal
- chart-understanding
- VLM
datasets:
- opendatalab/ChartVerse-SFT-600K
- opendatalab/ChartVerse-RL-40K
---

**ChartVerse-4B** is an efficient Vision Language Model (VLM) specialized for complex chart reasoning, developed as part of the **[opendatalab/ChartVerse](https://huggingface.co/collections/opendatalab/chartverse)** project. For more details about our method, datasets, and full model series, please visit our [Project Page](https://chartverse.github.io).

A key highlight is that **ChartVerse-4B significantly outperforms Qwen3-VL-8B-Thinking (60.0%) despite using only half the parameters**, demonstrating that data quality triumphs over model scale.

## 🔥 Highlights

- **Data Quality > Model Scale**: 4B parameters achieving 61.9% average score, surpassing Qwen3-VL-8B-Thinking (60.0%)
- **Efficient Performance**: Delivers 8B-level performance with 4B parameters
- **High-Quality Training**: Trained on ChartVerse-SFT-600K and ChartVerse-RL-40K with rigorous truth-anchored QA synthesis
- **Strong Reasoning**: Equipped with Chain-of-Thought reasoning for complex multi-step chart analysis

## 📊 Model Performance

### Overall Results

<div align="center">
  <img src="https://raw.githubusercontent.com/chartverse/chartverse.github.io/main/static/images/overall_result.png" width="100%" alt="Overall Performance Comparison">
</div>

### SFT vs RL Performance

<div align="center">
  <img src="https://raw.githubusercontent.com/chartverse/chartverse.github.io/main/static/images/training_phases.png" width="100%" alt="Training Phases Performance">
</div>

## 📚 Training Data

### [ChartVerse-SFT-600K](https://huggingface.co/datasets/opendatalab/ChartVerse-SFT-600K)
- **412K** unique high-complexity charts
- **603K** QA pairs with **3.9B** tokens of CoT reasoning
- Rollout Posterior Entropy: **0.44** (highest among all datasets)
- Truth-anchored answer verification via code execution

### [ChartVerse-RL-40K](https://huggingface.co/datasets/opendatalab/ChartVerse-RL-40K)
- **40K** highest-difficulty samples
- Filtered by failure rate: 0 < r(Q) < 1
- Ensures "hard but solvable" training signal

## 🏋️ Training Details

**Supervised Fine-Tuning (SFT)**:
- Framework: LLaMA-Factory
- Dataset: ChartVerse-SFT-600K
- Learning rate: 1.0 × 10⁻⁵
- Global batch size: 128
- Context length: 22,000 tokens

**Reinforcement Learning (RL)**:
- Framework: veRL
- Dataset: ChartVerse-RL-40K
- Algorithm: GSPO
- Learning rate: 1.0 × 10⁻⁶
- Rollout samples: 16 per prompt

## 🚀 Quick Start

```python
from transformers import Qwen3VLForConditionalGeneration, AutoProcessor
from qwen_vl_utils import process_vision_info
from PIL import Image

# 1. Load Model
model_path = "opendatalab/ChartVerse-4B"
model = Qwen3VLForConditionalGeneration.from_pretrained(
    model_path, torch_dtype="auto", device_map="auto"
)
processor = AutoProcessor.from_pretrained(model_path)

# 2. Prepare Input
image_path = "path/to/your/chart.png"
query = "Which region demonstrates the greatest proportional variation in annual revenue compared to its typical revenue level?"

messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "image": image_path},
            {"type": "text", "text": query},
        ],
    }
]

# 3. Inference
text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
image_inputs, video_inputs = process_vision_info(messages)
inputs = processor(
    text=[text],
    images=image_inputs,
    padding=True,
    return_tensors="pt",
).to("cuda")
generated_ids = model.generate(**inputs, max_new_tokens=16384)
output_text = processor.batch_decode(
    generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
print(output_text[0])
```

## 📖 Citation

```bibtex
@misc{liu2026chartversescalingchartreasoning,
      title={ChartVerse: Scaling Chart Reasoning via Reliable Programmatic Synthesis from Scratch}, 
      author={Zheng Liu and Honglin Lin and Chonghan Qin and Xiaoyang Wang and Xin Gao and Yu Li and Mengzhang Cai and Yun Zhu and Zhanping Zhong and Qizhi Pei and Zhuoshi Pan and Xiaoran Shang and Bin Cui and Conghui He and Wentao Zhang and Lijun Wu},
      year={2026},
      eprint={2601.13606},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2601.13606}, 
}
```

## 📄 License

This model is released under the Apache 2.0 License.

## 🙏 Acknowledgements

- Base model: [Qwen3-VL-4B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct)
- Training frameworks: [LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory), [veRL](https://github.com/volcengine/verl)
- Evaluation: [VLMEvalKit](https://github.com/open-compass/VLMEvalKit)