PyTorch Tutorials

Basics to advanced PyTorch tutorials. We have accumulated the best references availble in one place to filter the best practice for fast and easy ML implmentation.

Building the conda environment

All dependencies are included in environment.yml file. We can either have environment setup in our own project directory or in the envs directory of the miniconda folder. The advantage of installing in miniconda folder is that we can use same env for different packages and it saves bandwidth and storage. The first method on the other hand is more robust. Prcedure is as follows

First of all if mamba do not exists then install it via

conda activate base
conda install mamba -n base -c conda-forge

To setup env in project dir

mamba env create --prefix ./env --file environment.yml --force

or to setup in miniconda envs dir

mamba env create --name mlenv --file environment.yml --force

Update the conda env

If you add (remove) dependencies to (from) the environment.yml file or the requirements.txt file after the environment has already been created, then you can re-create the environment with the following command.

To update env in project ./env dir

conda activate base
mamba env update --prefix ./env --file environment.yml  --prune

To update in Miniconda envs dir

conda actiavte base
mamba env update --name mlenv --file environment.yml  --prune

Remove the conda env

If you are low on storage and want to completely remove the conda env. You can do it by

conda deactivate
conda env remove --prefix ./env
# or to remove from conda env dirs
conda env remove --name mlenv

Folders

• Jovian : You can also access the full tutorial on Jovian. FreeCodeCamp Youtube channel has also provided the same tutorial on Youtube.
• MorvanZhou : Nice collection as initiated by Morvan on Github.

References

• Jovian
• Morvan