Smolagents
Createdhuggingface
Enables the creation of powerful agents that think and act using Python code.
About
Smolagents is a barebones Python library designed to simplify the development of powerful agents. It provides first-class support for code agents that write their actions in Python code, ensuring secure execution in sandboxed environments using E2B or Docker. With Hub integrations for sharing and pulling tools, model-agnosticism supporting various LLMs and providers, modality-agnosticism for diverse inputs, and tool-agnosticism for integration with LangChain and other tool collections, Smolagents offers a versatile and efficient framework for building intelligent agents.
Key Features
- Supports various LLMs via LiteLLM integration.
- CodeAgent class for agents that write actions in Python code.
- Modality-agnostic: supports text, vision, video, and audio inputs.
- 15,890 GitHub stars
- Sandboxed execution environments via E2B or Docker.
- Hub integrations for sharing and pulling tools.
Use Cases
- Automating tasks using code-based agents.
- Building agents that can interact with web browsers using vision.
- Sharing and reusing agent tools via the Hugging Face Hub.