How to Run Jupyter Notebooks with PyTorch and TensorFlow Support in Microsoft AI-For-Beginners

Clone the Microsoft AI-For-Beginners repository, create the Conda environment from environment.yml, activate it, and launch Jupyter to run notebooks with both PyTorch and TensorFlow pre-installed.

The Microsoft AI-For-Beginners curriculum provides a complete, ready-to-use environment for learning artificial intelligence. Instead of manually installing deep learning libraries, you can run Jupyter notebooks with PyTorch and TensorFlow support using a single Conda environment definition that includes both frameworks, Keras, OpenCV, and other essential tools.

Setting Up the Environment

The repository includes an environment.yml file at the root level that pins specific versions of all dependencies. This ensures reproducible results across different machines.

Step 1: Clone the Repository

git clone https://github.com/microsoft/AI-For-Beginners.git
cd AI-For-Beginners

This downloads the full curriculum, including lesson notebooks, data files, and configuration files.

Step 2: Create the Conda Environment

Run the following command to build the environment defined in environment.yml:

conda env create --name ai4beg --file environment.yml

The environment specification includes:

• PyTorch (pytorch::pytorch)
• TensorFlow 2.17.0 (tensorflow==2.17.0)
• Keras, OpenCV, scikit-learn, matplotlib, and Jupyter

According to the source code in environment.yml, the dependencies are curated to avoid version conflicts between PyTorch and TensorFlow.

Step 3: Activate the Environment

conda activate ai4beg

Once activated, python, torch, and tensorflow become available in your shell and to Jupyter kernels.

Step 4: Launch Jupyter

Choose between the classic interface or Jupyter Lab:

# Classic Notebook UI

jupyter notebook

# Modern Lab interface (recommended)

jupyter lab

The setup instructions in AGENTS.md confirm that either interface works with the pre-configured kernel.

Opening and Running Lesson Notebooks

Navigate to any lesson directory containing .ipynb files. The curriculum provides parallel PyTorch and TensorFlow implementations for many lessons.

PyTorch Example: Convolutional Neural Networks

Open lessons/4-ComputerVision/07-ConvNets/ConvNetsPyTorch.ipynb in the Jupyter file browser. Typical first cells include:

import torch
import torchvision
import matplotlib.pyplot as plt

Executing these cells confirms PyTorch is functional and will automatically use GPU acceleration if available.

TensorFlow Example: Convolutional Neural Networks

Open lessons/4-ComputerVision/07-ConvNets/ConvNetsTF.ipynb. Standard imports are:

import tensorflow as tf
import matplotlib.pyplot as plt

The ConvNetsTF.ipynb notebook demonstrates identical concepts using TensorFlow/Keras APIs.

Verifying GPU Acceleration

If you have a CUDA-compatible GPU, the Conda packages will automatically detect and use it. Verify inside any notebook:

For PyTorch:

import torch
print(torch.cuda.is_available())

For TensorFlow:

import tensorflow as tf
print(tf.config.list_physical_devices('GPU'))

The troubleshooting guide in AGENTS.md recommends these exact snippets for GPU detection. Both should return True or a non-empty list when CUDA drivers are properly installed.

Quick Reference: Complete Setup Commands

# Clone repository

git clone https://github.com/microsoft/AI-For-Beginners.git
cd AI-For-Beginners

# Create and activate environment

conda env create --name ai4beg --file environment.yml
conda activate ai4beg

# Launch Jupyter

jupyter lab

Troubleshooting Common Issues

Issue	Solution	Source
Environment creation fails	Update Conda: conda update conda -y	AGENTS.md troubleshooting
Jupyter kernel not found	Reinstall kernel: python -m ipykernel install --user --name ai4beg	AGENTS.md
GPU not detected	Run nvidia-smi to verify drivers, then use GPU check code above	AGENTS.md GPU section
Package conflicts	Ensure you use conda env create rather than pip install	environment.yml structure

Key Files in the Repository

• environment.yml — Conda environment definition containing PyTorch, TensorFlow, and dependencies
• requirements.txt — Additional pip packages referenced by the Conda environment
• AGENTS.md — Human-readable setup instructions and troubleshooting
• lessons/4-ComputerVision/07-ConvNets/ConvNetsPyTorch.ipynb — Example PyTorch notebook
• lessons/4-ComputerVision/07-ConvNets/ConvNetsTF.ipynb — Example TensorFlow notebook

Summary

• Clone the Microsoft AI-For-Beginners repository from GitHub
• Create the Conda environment using environment.yml to get PyTorch, TensorFlow, and all dependencies in one command
• Activate with conda activate ai4beg and launch Jupyter with jupyter lab
• Open lesson notebooks in either PyTorch or TensorFlow variants and execute cells immediately
• Verify GPU availability with built-in checks if you have CUDA hardware

Frequently Asked Questions

Where are the PyTorch and TensorFlow notebooks located?

Many lessons include parallel implementations. For example, lessons/4-ComputerVision/07-ConvNets/ contains both ConvNetsPyTorch.ipynb and ConvNetsTF.ipynb. Browse the lessons/ directory to find corresponding versions for other topics.

Do I need to install PyTorch and TensorFlow separately?

No. The environment.yml file in the repository root defines a complete environment including both frameworks, Keras, OpenCV, and Jupyter. Create the environment once and all libraries are ready to use.

How do I know if my GPU is being used?

Run the verification code snippets from AGENTS.md inside any notebook: torch.cuda.is_available() for PyTorch or tf.config.list_physical_devices('GPU') for TensorFlow. These return True or device lists when GPU acceleration is active.

Can I use pip instead of Conda?

The project is designed for Conda to handle complex binary dependencies for PyTorch and TensorFlow. While requirements.txt exists for supplementary packages, the recommended path is conda env create --file environment.yml to avoid version conflicts.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}