Deep Learning Course at the IMS (University of Stuttgart)

This repository collects the tutorials for the deep learning course and accompanying material.

Structure

• Setup: Basics for getting started with PyTorch on Linux
• PyTorch 1: Basics
  • Tensor operations
  • Gradient computation
  • Simple data handling
  • Simple neural network
• PyTorch 2: Data Loading & Processing and Model Development for NLP
  • Example Task: Sentiment Analysis
  • Example Task: Part-Of-Speech Tagging
• PyTorch 3: Language Models

How To Start

1. Open a terminal.
2. Change to the directory where you want to download this repository to, e.g., cd ~/ to change to your home directory (create directories using mkdir).
3. Clone this repository to the current directory using git:

git clone https://github.com/DigitalPhonetics/deep-learning-course.git

4. Check out our instructions for setting up Jupyter Notebooks.
5. Set up PyTorch as described in this Jupyter Notebook.
6. Now you should be ready to use PyTorch in Python, and to be able to run all Jupyter Notebooks of this course.

Compute Resources for Course Activities

The following resources are available to students for course-related activities such as training neural networks:

Institute Resources

• IMS Servers (phoenix, dodo, kiwi, strauss, nandu): Computing resources provided by the IMS which can be accessed via SSH using your IMS account. GPUs are available on some servers.

External Resources

• bwCloud: Provides virtual machines (VMs). First Steps. GPUs are not available.
• bwUniCluster: Access to compute clusters via SSH & scheduler. Apply for Computing Time. GPUs are available.
• Google Colab: Hosted Jupyter notebooks. Google Colab. Free tier includes limited GPU access.
• Deepnote: Similar to Colab, with a focus on collaboration. Deepnote. GPUs are available on paid plans.

Help, Feedback & Other Questions

Many commands provide a help if you call them with the -h or --help argument, or appear in the manual with man command where command is the name of the command.

You can also contact us via ILIAS (e.g., in the forum), or open an issue in the GitHub repository!