Installation classique sous Linux (pip)

About this installation

Manual installation is always possible, especially for Macs

Here you will find a list of installation recipes proposed by participants of Fidle who successfully managed to install Fidle environment on a Mac - Thanks a lot for these contributions !

Anyway, Have you tried Docker ? …you should, maybe it will work ;-P

Size a full installation (environment, data, notebooks) is around 3GB

Content

Installation

General requirements for Fidle

You will need PyTorch.
See official procedure proposed on PyTorch web site : https://pytorch.org/get-started/locally/

You need other required additional packages

  • torch-geometric
  • torchtext
  • torchdata
  • lightning
  • tensorboard
  • keras
  • transformers
  • datasets
  • einops
  • numpy
  • Scikit-image
  • Scikit-learn
  • Matplotlib
  • plotly
  • seaborn
  • barviz
  • pyarrow
  • Pandas
  • Pandoc
  • pyyaml
  • Jupyterlab
  • fidle

Here is the pip installation instruction to install theses modules, but make sure you can use pip for python3 on your machine :

$ pip install torch-geometric torchtext torchdata lightning \
              tensorboard keras transformers datasets einops numpy Scikit-image Scikit-learn \
              Matplotlib plotly seaborn barviz pyarrow Pandas Pandoc \
              pyyaml Jupyterlab fidle

Once your Python environment is installed,
you should be able to follow the standard Linux procedure from point 4

Notebooks and datasets

You can retrieve all notebooks and datasets with the fid command.
This is the simplest procedure - see Linux install, Point 4

Notebooks are also available via :

Datasets are also available via :

Reinstall and updates

  • You can reinstall notebooks or datasets with the fid install_notebooks and fid install_datasets commands
  • You can also locate the datasets fidle folder anywhere you want.
    In this case, you must specify the location of this folder with the environment variable FIDLE_DATASETS_DIR

Feedback : Installation sous MacOS (April 2024)

This recipe was proposed in **April 2024 by François-Xavier C. (Thanks !!)

Bonjour, Je voulais vous faire un petit retour d’expérience sur l’installation de l’environnement fidle sous macOS, parce que sur la page Fidle ça a l’air super compliqué, mais je ne sais pas pourquoi. J’ai un environnement dont je pense qu’il marche avec (installation avec conda) :

conda create python=3.9 --name fidle   
conda activate fidle
conda install pytorch torchvision torchaudio torchtext -c pytorch
conda install chardet lightning tensorboard transformers pytorch_geometric \
  lightning tensorboard transformers keras matplotlib seaborn pandoc jupyterlab \
  pytorch_geometric lightning tensorboard transformers keras matplotlib seaborn \
  pandoc jupyterlab plotly ipywidgets
pip install barviz fidle
mkdir fidle-tp
cd fidle-tp
fid install --quiet

Note : Il faut juste installer pytorch torchvision torchaudio torchtext depuis le canal pytorch, et les autres depuis le conda-forge habituel.

Feedback : Installation on MacBook Air M1 (January 2024)

This recipe was proposed in January 2024 by Pauline M. (Thanks a lot !)
Setup/Requirements : MacBook Air M1, version 14.2.1, conda

Instructions:

conda create python=3.9 --name fidle   
conda activate fidle
conda install -c conda-forge chardet
conda install pytorch torchvision torchaudio -c pytorch
conda install conda-forge::pytorch_geometric 
conda install pytorch::torchtext
conda install lightning -c conda-forge
conda install conda-forge::tensorboard
conda install conda-forge::transformers
conda install conda-forge::keras
conda install -c conda-forge matplotlib  
conda install plotly::plotly  
conda install anaconda::seaborn
pip install --upgrade barviz      
conda install conda-forge::pandoc
conda install -c conda-forge jupyterlab   
pip install fidle
conda list -e > fidle_requirements2024.txt
mkdir fidle-tp
cd fidle-tp
fid install --quiet
Fidle