README.md

<div align="center">

# AI Agents in Chemical Research: GVIM - An Intelligent Research Assistant System 🧪🤖

[English](README.md) | [简体中文](README_zh.md)

[![Paper](https://img.shields.io/badge/📑_Paper-DigitalDiscovery-blue)](https://pubs.rsc.org/en/content/articlelanding/2025/dd/d4dd00398e)
[![Dataset](https://img.shields.io/badge/🤗_Dataset-HuggingFace-yellow)](https://huggingface.co/datasets/KANGYONGMA/GVIM)
[![Install](https://img.shields.io/badge/📝_Install-Video-green)](https://www.youtube.com/watch?v=1eMwus98BB8)
[![Demo](https://img.shields.io/badge/🎥_Demo-Watch-red)](https://www.youtube.com/watch?v=fb8hdho_89s&t=128s)

An intelligent research assistant system designed specifically for chemical science, featuring fine-tuned language models and specialized chemistry capabilities.

</div>

## 🌟 Highlights

<table>
<tr>
<td>
  
### 🧬 Core Features
- Fine-tuned LLMs for chemistry
- Molecular visualization
- Literature retrieval & analysis
- Multimodal capabilities
  
</td>
<td>

### 🚀 Key Benefits
- Specialized for chemical research
- Continuous learning ability
- Free GPU resources on Colab
- Comprehensive documentation
  
</td>
</tr>
</table>

## 📚 Model Overview

GVIM combines several cutting-edge technologies:

### 🔬 Technical Features
- **Fine-tuned Models**: Trained on curated chemistry instruction data
- **Chemistry Tools**: Molecular visualization and SMILES processing
- **Smart Retrieval**: Advanced chemical literature search and analysis
- **Multimodal Support**: Formula recognition and image analysis
- **Knowledge Base**: Local document processing and continuous learning

### 🎯 Main Applications
1. Chemical research assistance and analysis
2. Molecular structure visualization
3. Literature review and knowledge extraction
4. Handwritten formula recognition
5. Document-based knowledge processing

## 💻 Quick Start

### Requirements

```bash
# Create conda environment
conda create -n gvim python=3.9.19
conda activate gvim

# Install dependencies
pip install -r requirements.txt
```

### API Configuration ⚙️
Required API keys:
```python
TAVILY_API_KEY="your_key_here"
REPLICATE_API_TOKEN="your_token_here"
Groq_API_KEY="your_key_here"
```

### Launch 🚀
```bash
python app.py
```

## 🎥 Key Features Showcase

<details>
<summary>📸 Click to view feature demonstrations</summary>

### 1. Nature Chemistry Search Interface
![Nature Chemistry Interface](assets/search_interface.jpg)

### 2. Multimodal Chemical Recognition
![Multimodal Recognition](assets/multimodal_recognition.jpg)

### 3. Local Document Analysis
![Document Analysis](assets/document_analysis.jpg)

</details>

## 🆓 Free GPU Resources

Access GVIM's capabilities using Colab's free GPU resources:

[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://youtu.be/QGwVVdinJPU)

## 📖 Citation

```bibtex
@article{Digital Discovery,
  author = {Kangyong Ma},
  affiliation = {College of Physics and Electronic Information Engineering, Zhejiang Normal University},
  address = {Jinhua City, 321000, China},
  doi = {10.1039/D4DD00398E},
  email = {kangyongma@outlook.com, kangyongma@gmail.com}
}
```

## 📞 Contact

<table>
<tr>
<td>
  
### 📧 Email
- kangyongma@outlook.com
- kangyongma@gmail.com
  
</td>
<td>
  
### 🏢 Institution
College of Physics and Electronic Information Engineering  
Zhejiang Normal University  
Jinhua City, 321000, China
  
</td>
</tr>
</table>

---

<div align="center">
  
**🌟 Star us on GitHub | 📌 Commercial use requires authorization | 📑 Read our paper**

</div>