Anaconda: https://www.anaconda.com/downloads
TensorFlow: https://www.tensorflow.org/
TensorFlow GPU: https://www.tensorflow.org/install/gpu
VS Code: https://code.visualstudio.com/
Required compiler on Windows to build some of the required packages.
- Microsoft Visual C++ Redistributable: https://support.microsoft.com/en-us/help/2977003/the-latest-supported-visual-c-downloads
- Microsoft Build Tools 2015: https://www.microsoft.com/en-US/download/details.aspx?id=48159
Restart afterwards.
CUDA can improve performance of machine learning using the graphics card. This only makes sense, if you have a powerful Nvidia graphics card available, otherwise you can work with the CPU version instead.
Follow these steps to install CUDA: https://www.tensorflow.org/install/gpu
- Latest GPU driver
- CUDA Toolkit 11.2 https://developer.nvidia.com/cuda-toolkit-archive
- cuDNN SDK 8.1.0: https://developer.nvidia.com/cudnn
Create environment called 'tensorflow'
conda create --name tensorflow python=3.8
Activate environment
conda activate tensorflow
Install requirements
cd tensorflow-basics
cd my-project
pip install --upgrade pip
pip install -r requirements.txt
This includes tensorflow 2.6 as well as other helpful packages like scikit-learn.
You can run tensorflow using Jupyter Notebook with a preconfigured docker image:
docker pull tensorflow/tensorflow:latest # Download latest stable image
docker run -it -p 8888:8888 tensorflow/tensorflow:latest-jupyter # Start Jupyter server