How to customize a conda environment

To perform meaningful calculations, we have to install additional packages in our conda environment. If you don’t have a conda environment installed, refer to: How to create and activate a conda environment. Here, we restrict ourselves to a simple example. The code we are going to work with is the following:

import torch

# setting device on GPU if available, else CPU
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
print('Using device:', device)
print()

# additional info when using cuda
if device.type == 'cuda':
    print(torch.cuda.get_device_name(0))

You can copy the code to a Jupyter notebook or use your own example. Here the file is mycode.ipynb.

1. Your Jupyter Notebook file must be in your home/<username> directory.

2. In this example, the Torch library is required. Compatible versions are listed in https://download.pytorch.org/whl/torch_stable.html.

3. Install PyTorch. To do so, activate the conda environment that you want to work with and then install the package in it:

    conda activate <name>
    conda install pytorch -c pytorch -y
   

4. In this example, you may need to install numpy also, then

   pip install numpy
   

5. Once the installation is completed you can list all your packages considering

    conda list 
   

6. Your environment now has the necessary packages to run mycode.ipynb. To run it, refer to: How to run a Jupyter notebook file.