Development Environment

The technical support is available with conda.

Conda

conda create --name mphy0041-pt -c pytorch numpy matplotlib requests pytorch

Notebook can be installed within the mphy0041 conda environment:

conda activate mphy0041 \
 && conda install -c conda-forge notebook

Alternatively, create a standalone conda environment instead to manage a lighter environment without the deep learning packages.

Note: the following instructions for installation depend on the available CUDA versions.

For TensorFlow users, see conda TensorFlow guide, e.g.:

conda install tensorflow-gpu -c anaconda

For PyTorch users, see PyTorch install guide, e.g.:

conda install pytorch cudatoolkit=10.2 -c pytorch

Ubuntu 20.04 Docker images are available. Alternatively, a Dockerfile is also provided. See more details and support in the official Docker documentation.

Download the pre-built Docker image:

sudo docker pull yipenghu/ucl-module-ubuntu:minimal

Create and run a container named mphy0041 in its interactive bash:

sudo docker run --name mphy0041 -ti yipenghu/ucl-module-ubuntu:minimal bash

Within the container, follow the above to clone this repositry and create the conda environment.

TBC