跳转至

Colab Tutorial Notebooks

This page collects the official Colab notebooks for EvoAgentX.
Click the Open in Colab badge to launch a runnable copy in your browser.

Tip: Each notebook starts with an installation cell. Run cells in order.
For heavier workloads, you may switch Colab to a GPU runtime (optional).


At a Glance

  • 7 ready-to-run notebooks covering core usage, workflows, tools, and optimization.
  • Zero local install — everything runs in Colab.
  • Copy–paste friendly snippets you can reuse in your own projects.

Quickstart

Open In Colab  • View on GitHub

A minimal, end-to-end tour of EvoAgentX: install, configure, and run a tiny workflow in minutes.


First Agent

Open In Colab  • View on GitHub

Build your first agent from scratch—key concepts, configuration, and reusable code patterns.


First Workflow

Open In Colab  • View on GitHub

Compose multiple steps into a reusable workflow: nodes, dependencies, and data flow.


Tools

Open In Colab  • View on GitHub

Attach and invoke tools (retrieval, compute, external APIs) from agents—plus common pitfalls to avoid.


Benchmark & Evaluation

Open In Colab  • View on GitHub

Run reproducible evaluations for agents/workflows: datasets, metrics, experiment tracking, and result analysis.


AFlow Optimizer

Open In Colab  • View on GitHub

Search configuration spaces and auto-tune strategies to improve performance and robustness.


TextGrad Optimizer

Open In Colab  • View on GitHub

Iteratively refine prompts or policies using TextGrad-style updates—great for complex, evolving tasks.


FAQ

Do I need to install anything locally?
No. Everything runs in the browser. Just execute the first installation cell in each notebook.

Is a GPU required?
Not for basic demos. For large models, retrieval, or heavy workloads, enabling a GPU can help.

Can I run the notebooks locally?
Yes. Download the .ipynb and open it in Jupyter or VS Code. Install dependencies as shown at the top of each notebook.