|
|
--- |
|
|
title: Agent with smolagents + DeepSeek |
|
|
emoji: 🧠 |
|
|
colorFrom: indigo |
|
|
colorTo: blue |
|
|
sdk: gradio |
|
|
app_file: main.py |
|
|
pinned: false |
|
|
--- |
|
|
|
|
|
# Hugging Face Agent Project |
|
|
|
|
|
This project implements a Hugging Face agent using the smolagents framework. The agent is designed to handle various tasks by utilizing custom tools and models. |
|
|
|
|
|
## Project Structure |
|
|
|
|
|
``` |
|
|
huggingface-agent-project |
|
|
├── main.py # Entry point for the agent |
|
|
├── agent_config.py # Configuration and model loading |
|
|
├── tools # Directory for custom tools |
|
|
│ ├── __init__.py # Package initializer for tools |
|
|
│ ├── wiki_tool.py # Tool for fetching Wikipedia summaries |
|
|
│ ├── web_search_tool.py # Tool for performing web searches using DuckDuckGo |
|
|
│ ├── excel_analysis_tool.py # Tool for analyzing Excel files |
|
|
├── requirements.txt # Project dependencies |
|
|
└── README.md # Project documentation |
|
|
``` |
|
|
|
|
|
## Setup Instructions |
|
|
|
|
|
1. Clone the repository: |
|
|
``` |
|
|
git clone <repository-url> |
|
|
cd huggingface-agent-project |
|
|
``` |
|
|
|
|
|
2. Install the required dependencies: |
|
|
``` |
|
|
pip install -r requirements.txt |
|
|
``` |
|
|
|
|
|
## Usage |
|
|
|
|
|
To run the agent, execute the following command: |
|
|
``` |
|
|
python main.py |
|
|
``` |
|
|
|
|
|
## Agent Capabilities |
|
|
|
|
|
The agent is capable of performing various tasks by leveraging the following tools: |
|
|
|
|
|
### Tools |
|
|
- **WikiTool**: Fetches summaries of topics from Wikipedia, handling disambiguation and page errors gracefully. |
|
|
- **WebSearchTool**: Performs web searches using DuckDuckGo and retrieves the first result's snippet. |
|
|
- **ExcelAnalysisTool**: Loads an Excel file and sums sales from items labeled as "food", excluding drinks. Returns the total sales in USD with 2 decimal places. |
|
|
|
|
|
### Models |
|
|
- The agent uses the Hugging Face Inference API via `HfApiModel` to interact with models such as `DeepSeek-R1` or `DeepSeek-Coder`. |
|
|
|
|
|
### Example Workflow |
|
|
1. The agent receives a request. |
|
|
2. Based on the request, it selects the appropriate tool (e.g., WikiTool for Wikipedia lookups). |
|
|
3. The tool processes the request and returns the result. |
|
|
|
|
|
## Contributing |
|
|
|
|
|
Contributions are welcome! Please submit a pull request or open an issue for any enhancements or bug fixes. |
|
|
|
|
|
## Future Enhancements |
|
|
|
|
|
- Deployment to Hugging Face Spaces. |
|
|
- Adding more tools for advanced functionalities. |
|
|
- Improving error handling and logging. |
