# 1. Introduction

We propose **Pulsar-VLM**, a visual reasoning model for pulsar candidate identification, built on pretrained multimodal large language models (MLLMs). The input to Pulsar-VLM includes diagnostic subplots (i.e., frequency-phase plots and DM curves) along with task instructions. By integrating the visual features of the diagnostic subplots with semantic textual descriptions, Pulsar-VLM is able to perform geometric morphological reasoning on the diagnostic plots in a manner analogous to how astronomers interpret them.

**Pulsar-VLM maintains high performance across diagnostic plots from commonly used radio telescopes, despite morphological differences, and thus provides substantial benefits to the pulsar research community.**

---

# 2. Evaluation Dataset

![Evaluation Dataset](https://pfst.cf2.poecdn.net/base/image/1a4d214ea08ae2dd1c0d79024f56534673907e70928d48a5c418051c04b8cc9e?w=1220&h=493&pmaid=505863425)

Below are the pulsar candidate datasets used in this work:

- **CRAFTS Dataset**: [Hugging Face Repository](https://huggingface.co/datasets/jcchtt/CRAFTS_Dataset)  
  Provided by Dr. Pei Wang, Institute for Frontiers in Astronomy and Astrophysics, Beijing Normal University. Email: [wangpei@nao.cas.cn](mailto:wangpei@nao.cas.cn)

- **GPPS Dataset**: [Official Website](http://zmtt.bao.ac.cn/GPPS/)

- **GC FANS Dataset**: [Official Website](https://www3.mpifr-bonn.mpg.de/staff/pfreire/GCpsr.html)  
  Diagnostic plots generated with PRESTO, from which we identified 91 pulsars. Our processed dataset is available at: [Hugging Face Repository](https://huggingface.co/datasets/jcchtt/GC_FANS_Dataset)

- **FAST Dataset**: [GitHub Repository](https://github.com/dzuwhf/FAST_label_data)

- **GBT-350 Dataset**: [Official Website](http://astro.phys.wvu.edu/GBTdrift350/)

- **GBNCC Dataset**: [Official Website](http://astro.phys.wvu.edu/GBNCC/)

- **SGAN Dataset**: [GitHub Repository](https://github.com/vishnubk/sgan)

- **CHIRSS Dataset**: [Official Website](https://www.jb.man.ac.uk/research/pulsars/GHRSS/)

- **LOTAAS Dataset**: [Official Website](https://www.astron.nl/lotaas/index.php?sort=0&order=0)

---
# 3. Evaluation Results

| Dataset         | FN  | Recall |
|-----------------|-----|--------|
| CRAFTS Dataset  | 3   | 98.8%  |
| GC FANS Dataset | 1   | 98.9%  |
| GPPS Dataset    | 3   | 99.8%  |
| GBT-350 Dataset | 0   | 100%   |
| GBNCC Dataset   | 10  | 94.2%  |
| LOTAAS Dataset  | 0   | 100.0% |
| CHIRSS Dataset  | 0   | 100.0% |
| FAST Dataset    | 1   | 99.7%  |
| SGAN Dataset    | 577 | 94.5%  |

---

# 4. Usage