The tutorials require a few dependencies, e.g. NumPy, in addition to one of the two deep learning libraries. Individual tutorials may also require other libraries which will be specified in the readme.md in individual tutorial folders.
Module tutorials are implemented in Python with TensorFlow and/or PyTorch.
Conda is recommended to manage the required dependencies and libraries. It is not mandatory, in tutorials or assessed coursework, to use any specific development, package or environment management tools. However, technical support will be available with the tested development environment - see the supported development environment for Python.
-
For Python programming and numerical computing:
- Basic Python programming is required in this module. Relevant tutorials are readily available, e.g. tutorial links in the supported development environment for Python.
- Other tools may be useful but not supported: Jupyter Notebook, Anaconda and any other IDEs or code editors.
-
For TensorFlow and PyTorch:
- Optional TA-led tutorials may be available as a refresher at the beginners level.
- Other tutorials are readily available, e.g. the respective official documentation TensorFlow tutorials and PyTorch tutorials.
- TA support are also available with the above-specified development environment.
-
For GPU acceleration:
- Google Colab provides freely available computing resource, though restrictions apply.
- UCL Department of Computer Science hosts a high performance computing cluster, with independent technical support.
- Other GPU supply is to be confirmed.