Online documentation

Please refer to the online documentation for installation steps, user guides and APIs.

Building the documentation

Documentation is based on Sphinx and can be generated locally from the repository top.

Requirements:

• Python version: python >= 3.9
• TensorFlow version ~= 2.15
• Sphinx version 2.1 or above

Docker image

A docker image with all dependencies to build and execute the examples is provided in the ci/ directory. This image is built and optionally pushed to a registry with:

tag=$(sha1sum dockerfile | cut -c1-7)
docker build --rm -t bender:5000/doc_build:$tag -f dockerfile .
docker push bender:5000/doc_build:$tag

Build steps

• Optional: create a virtual environment using Virtualenv or Conda

• Install akida packages and their requirements using the provided file:

pip install -r requirements.txt

• Install Sphinx additional modules used by the documentation:

pip install sphinx_rtd_theme sphinx-gallery autodocsumm sphinx_design

• Finally build using sphinx-build tool, refer to Sphinx tutorial for details on this command line:

sphinx-build -D plot_gallery=0 -b html ./doc ./build

Note that by removing the -D plot_gallery=0 option in the previous command, examples will be executed for a nicer rendering but it will take much more time. If needed (e.g. for debugging), a set of selected examples can be rendered through the command:

sphinx-build -b html ./doc ./build -D sphinx_gallery_conf.filename_pattern=regex_my_examples.py

• Open .build/index.html once done

Style check

Style check is based on flake8.

To align examples with python formatting rules:

autopep8 --ignore E402,E501 --in-place --recursive examples/

Supported versions history

Versions are akida/cnn2snn packages ones. Please refer to https://github.com/Brainchip-Inc/akida_examples/releases for more details.

sequenceDiagram
    participant [...]
    participant 2.2.5
    participant 2.3.7
    participant 2.4.0
    participant 2.7.2
    participant 2.8.1
    participant 2.9.0
    participant 2.10.0
    participant next
    Note right of 2.10.0: Today
rect rgb(171, 170, 166)
    [...]-->>2.2.5: TensorFlow 2.8
    2.2.5->>+2.8.1: TensorFlow 2.10
    2.4.0->>+2.8.1: TensorFlow 2.11/2.12
    2.8.1->>+2.10.0: TensorFlow 2.15
    2.10.0->>+next: TensorFlow 2.16 (Keras 3)
end
rect rgb(250, 229, 160)
    [...]-->>+2.4.0: Python 3.7 (EOL June 2023)
    [...]->>+2.8.1: Python 3.8 (EOL October 2024)
    [...]->>+next: Python 3.9, 3.10
    2.4.0->>+next: Python 3.11
    2.10.0->>+next: Python 3.12
end
rect rgb(171, 170, 166)
    [...]-->>+2.4.0: Ubuntu 18.04 (EOL April 2023)
    [...]->>+next: Ubuntu 20.04, 22.04
    2.7.2->>+next: Ubuntu 24.04
end
%%{init:{'theme':'base'}}%%

Loading