Prerequisites

• Python 3.7 - 3.9
• Nvidia Driver >= 450 (GPU only)
• CUDA >= 11.0 (GPU only)

1. Install Dependencies:

• Install TensorFlow

  • Linux/MAC:

    pip install tensorflow==2.9.1
    

  • Windows: Please follow this installation guide.

• Install PyTorch

  • CPU:

    pip install torch==1.10.2+cpu torchvision==0.11.3+cpu torchaudio==0.10.2+cpu -f https://download.pytorch.org/whl/cpu/torch_stable.html
    

  • GPU:

    pip install torch==1.10.2+cu113 torchvision==0.11.3+cu113 torchaudio==0.10.2+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
    

• Extra Dependencies:

  • Windows:

    • Install Build Tools for Visual Studio 2019 here.

    • Install latest Visual C++ redistributable here and choose x86 for 32 bit OS, x64 for 64 bit OS.

  • Linux:

    $ apt-get install libglib2.0-0 libsm6 libxrender1 libxext6
    

  • Mac:

    $ echo No extra dependency needed ":)"
    

2. Install FastEstimator:

• Stable (Linux/Mac):

  $ pip install fastestimator
  

• Stable (Windows):

  First download zip file from available releases

  $ pip install fastestimator-x.x.x.zip
  

• Nightly (Linux/Mac):

  $ pip install fastestimator-nightly
  

• Nightly (Windows):

  First download zip file here

  $ pip install fastestimator-master.zip
  

Docker Hub

Docker containers create isolated virtual environments that share resources with a host machine. Docker provides an easy way to set up a FastEstimator environment. You can simply pull our image from Docker Hub and get started: * Stable: * GPU:

docker pull fastestimator/fastestimator:latest-gpu

* CPU:

docker pull fastestimator/fastestimator:latest-cpu

* Nighly: * GPU:

docker pull fastestimator/fastestimator:nightly-gpu

* CPU:

docker pull fastestimator/fastestimator:nightly-cpu