Installation

Installation#

SpikeRNN consists of two complementary packages: Rate RNN and Spiking RNN.

Requirements#

  • Python >= 3.7

  • PyTorch >= 1.8.0

  • NumPy >= 1.16.4

  • SciPy >= 1.3.1

  • Matplotlib >= 3.0.0

Installing SpikeRNN#

To install the complete SpikeRNN framework with both rate and spiking packages:

git clone https://github.com/NuttidaLab/spikeRNN.git
cd spikeRNN
pip install -e .

Development Installation#

For development, you can install additional dependencies for either package:

Rate RNN Development:

cd rate
pip install -e ".[dev]"

Spiking RNN Development:

cd spiking
pip install -e ".[dev]"

This will install additional packages for testing and development:

  • pytest >= 6.0

  • pytest-cov

  • black

  • flake8

  • mypy

Documentation Dependencies#

To build the documentation, install the documentation dependencies:

pip install -e ".[docs]"

This will install:

  • sphinx >= 3.0

  • sphinx-book-theme

  • sphinxcontrib-napoleon

Parallel Processing Support#

For faster grid search optimization in the spiking package:

cd spiking
pip install -e ".[parallel]"

This will install:

  • joblib >= 1.0

  • multiprocess >= 0.70

Verification#

To verify your installation, you can run:

import spikeRNN
spikeRNN.check_packages()

This will show which packages are available:

SpikeRNN Package Status:
==============================
     rate: ✓ Available
  spiking: ✓ Available

GPU Support#

Both packages support GPU acceleration through PyTorch’s CUDA integration. No additional installation is required if you have a CUDA-compatible PyTorch installation.

Contents