Installation#

This page describes how to install PyTorch Network Loader for different use cases.

Requirements#

PyTorch Network Loader requires:

  • Python 3.12 or later

  • PyTorch 2.11 or later

  • NumPy

  • Packaging

Optional Dependencies#

For normalising flows support (pip option flows):

  • zuko

For Weights & Biases support (pip option wandb):

  • wandb

For GPU acceleration (recommended):

  • NVIDIA GPU with CUDA Toolkit

Using Within Projects#

To use PyTorch Network Loader in your project, you can use pip and add it to your requirements.txt:

pip install netloader==<VERSION>

Note

Replace <VERSION> with the specific version you want to use, or leave blank to install the latest version from the main branch.

With Optional Packages#

If you need optional support (currently flows and wandb), use:

pip install netloader[option1,option2,...]

Local Development Installation#

For local development or if you want to modify the package:

  1. Clone the repository:

    git clone https://github.com/EthanTreg/PyTorch-Network-Loader.git
cd PyTorch-Network-Loader

  2. Install dependencies (in a virtual environment):

    pip install -r requirements.txt