First_agent / README.md
cjb97's picture
Update documentation with accurate features, limitations, and requirements
f3b2369

A newer version of the Gradio SDK is available: 6.5.1

Upgrade
metadata
title: First Agent with smolagents
emoji: 🤖
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 4.44.1
app_file: app.py
pinned: false
license: apache-2.0
tags:
  - agents
  - smolagents
  - web-search
  - weather
  - huggingface-course
datasets:
  - none
language:
  - en
library_name: smolagents
pipeline_tag: conversational

My First Agent with smolagents

This project is part of the Hugging Face Agents Course. It demonstrates how to create a simple but powerful agent using the smolagents library version 1.9.2.

Features

This agent can:

  • Search the web using DuckDuckGo
  • Get weather information for locations (using OpenWeatherMap API)
  • Provide natural conversational responses

Requirements

  • Python 3.10 or higher
  • smolagents==1.9.2
  • gradio==4.44.1
  • Other dependencies listed in requirements.txt

Environment Variables

The following environment variables need to be set:

  • HUGGINGFACE_TOKEN: Your Hugging Face API token
  • OPENWEATHERMAP_API_KEY: Your OpenWeatherMap API key

How to Use

  1. Clone the repository
  2. Install the dependencies:
    pip install -r requirements.txt
    
  3. Set up your environment variables:
    export HUGGINGFACE_TOKEN=your_token_here
    export OPENWEATHERMAP_API_KEY=your_key_here
    
  4. Run the application:
    python app.py
    

Example Queries

  • "What's the weather like in Tokyo?"
  • "Tell me the temperature in New York"
  • "Search for the latest news about AI"

Weather Tool Limitations

The weather tool has some limitations:

  • Cannot disambiguate between cities with the same name
  • Only accepts letters, digits, spaces, and hyphens in city names
  • Does not support additional location information (state, country, etc.)
  • Returns temperatures in Fahrenheit and wind speeds in mph

Technical Details

This agent uses:

  • smolagents 1.9.2 for the agent framework
  • Qwen/Qwen2.5-Coder-32B-Instruct as the language model
  • Gradio 4.44.1 for the user interface
  • OpenWeatherMap API for weather data
  • DuckDuckGo for web searches

Credits

Created as part of the Hugging Face Agents Course.