Setting Up Python and TensorFlow Environment

# CHAPTER 2

Setting Up Python and TensorFlow Environment #

1. Introduction #

2. Learning Objectives #

• Install a compatible version of Python.

• Create and manage Python Virtual Environments.

• Install TensorFlow using pip.

• Set up Visual Studio Code (VS Code) and Jupyter Notebooks.

• Understand the basics of GPU acceleration (CUDA).

3. Installing Python #

Windows:

1. 1. Download the installer from python.org/downloads.

1. 2. CRITICAL: Check the box "Add Python to PATH" before clicking Install.

macOS:

1. 1. Open your terminal and install Homebrew if you don't have it.

1. 2. Run: brew install python

Linux (Ubuntu): Open terminal and run: sudo apt update sudo apt install python3 python3-pip python3-venv

4. Virtual Environments #

1. 1. Open your terminal/command prompt.

1. 2. Create a folder for this course: mkdir tfcourse and enter it: cd tfcourse

1. 3. Create the virtual environment (named tfenv):

• Windows: python -m venv tfenv

• Mac/Linux: python3 -m venv tfenv

• Windows Command Prompt: tfenv\Scripts\activate

• Mac/Linux: source tfenv/bin/activate

Run:

pip install --upgrade pip
pip install tensorflow jupyter pandas matplotlib

*Note: This downloads a large package (often >500MB), so it may take a few minutes.*

6. VS Code and Jupyter Notebook Setup #

1. 1. Download Visual Studio Code from code.visualstudio.com.

1. 2. Open VS Code, go to the Extensions tab (squares on the left).

1. 3. Install the Python extension and the Jupyter extension.

1. 4. Open your tfcourse folder in VS Code.

1. 5. Create a file named hellotf.ipynb.

1. 6. Open the file, click "Select Kernel" in the top right, and choose your tf_env environment.

7. Step-by-Step Implementation: Verify Installation #

import tensorflow as tf

print(f"TensorFlow Version: {tf.__version__}")

# Check if TensorFlow can perform basic math
tensor_a = tf.constant(5)
tensor_b = tf.constant(10)
result = tf.add(tensor_a, tensor_b)

print(f"5 + 10 = {result.numpy()}")

*If this prints the version number and "15", congratulations, you are ready to build neural networks!*

8. GPU Setup Basics (Advanced) #

• NVIDIA GPUs: TensorFlow is heavily optimized for NVIDIA graphics cards via a software layer called CUDA. Setting up CUDA locally can be an incredibly frustrating process involving downloading specific driver versions.

• Mac GPUs (Apple Silicon): If you have an M1/M2/M3 Mac, you can install the tensorflow-metal plugin to use your Mac's GPU.

• The Ultimate Hack (Google Colab): If you don't have a powerful GPU or don't want to deal with installation headaches, use Google Colab (colab.research.google.com). It is a free Jupyter Notebook in the browser that gives you free access to powerful cloud GPUs. No installation required!

9. Common Mistakes #

• Installing 32-bit Python: TensorFlow will not install on 32-bit Python. It will throw a Could not find a version that satisfies the requirement tensorflow error. Ensure you installed 64-bit Python.

• Naming your file tensorflow.py: If you create a test file named tensorflow.py and run import tensorflow, Python will try to import your file instead of the actual library, causing a massive crash. Never name your files the same name as a library.

10. Best Practices #

• Freeze your requirements: Once you have a working environment, run pip freeze > requirements.txt. This saves the exact versions of TensorFlow and NumPy you are using, so you can perfectly recreate the environment if your computer crashes.

11. Exercises #

1. 1. Follow the steps to create a virtual environment, activate it, and install TensorFlow.

1. 2. Write a Python script to check if TensorFlow detects a GPU on your system using tf.config.listphysicaldevices('GPU').

12. MCQ Quiz with Answers #

13. Interview Questions #

• Q: What is the purpose of the CUDA toolkit in the context of Deep Learning?

• Q: Explain the typical software stack required to write and execute TensorFlow code locally.

14. FAQs #

15. Summary #

16. Next Chapter Recommendation #