libtensorflow-pharo-bindings

This is a fork of https://github.com/Cuis-Smalltalk/Machine-Learning for Pharo. This library was also ported to VA Smalltalk : https://github.com/vasmalltalk/tensorflow-vast

You will need a 64 bits Pharo VM in order to run the code. The code has only be tested on Pharo 9.0 on macOS and Windows with TensorFlow 1.15.5: https://github.com/tensorflow/tensorflow/releases/tag/v1.15.5.

Installation

• Install last Pharo 9.0 64 bit VM and image from the command line : https://pharo.org/download
• Install the project in Pharo

To install the project on your Pharo image you can execute the following script:

    Metacello new
    	githubUser: 'PolyMathOrg' project: 'libtensorflow-pharo-bindings' commitish: 'master' path: '';
    	baseline: 'LibTensorFlowPharoBinding';
    	load

Alternatively you can use Iceberg to load the code of this repository (See the video here: https://www.youtube.com/watch?v=U6Ttcc1KJUg&feature=youtu.be)

To add the project to your baseline just add this:

    spec
    	baseline: 'LibTensorFlowPharoBinding'
    	with: [ spec repository: 'github://PolyMathOrg/libtensorflow-pharo-bindings' ]

Installation of TensorFlow C API on MacOS

• Install TensorFlow C API on your computer. On macOS, the simpliest way to do that is to use Brew: brew install tensorflow If you don't use brew, check the installation guide
• check method TensorFlowCAPI>>macModulenameto put the path to where Tensorflow libraries are located on your computer:

TensorFlowCAPI>>macModulename
  ^ '/usr/local/Cellar/libtensorflow/1.15.5/lib/libtensorflow.so'

Installation of TensorFlow C API on Windows

• Check the Tensorflow for C installation guide

Installation of TensorFlow C API on Linux

• Check the Tensorflow for C installation guide
• check method TensorFlowCAPI>>unixModulenameto put the path to where Tensorflow libraries are located on your computer:

TensorFlowCAPI>>unixModulename
  ^ '/usr/local/lib/libtensorflow.so'