Exercises for bverI Deep Learning (Part 2) - HS 2024

This repository is used for the distribution of exercises for bverI Deep Learning (Part 2).

There are several ways to work on the assignments:

• Google Colab (easiest)
• pip (local install)
• Docker (only linux/amd64 available)

Google Colab

Use Google Colab by clicking on the links below.

Note: After running the notebook on a compute instance it may get stuck on the "import" cell. In that case, restart the kernel / session and run again. It should work then.

Exercise 01 - PyTorch and Image Data

Click on the following badge to open the notebook in Google Colab:

Exercise 02 - MLPs and Images

Click on the following badge to open the notebook in Google Colab:

Exercise 03 - Convolutional Neural Networks

Click on the following badge to open the notebook in Google Colab:

Exercise 04 - Image Classification

Click on the following badge to open the notebook in Google Colab:

Local

pip install .

Docker

1. Install Docker on your computer

Depending on your operating system you have to install docker in different ways.

You'll find detailed instructions here: https://docs.docker.com/get-docker

2. Pull the Docker image

# pull the image
docker login -u read_registry -p glpat-XaJ-SDxB6Qn7j_8USgAF cr.gitlab.fhnw.ch/ml/courses/bveri/hs2024/bveri-exercises-dev-hs2024
docker pull cr.gitlab.fhnw.ch/ml/courses/bveri/hs2024/bveri-exercises-dev-hs2024:latest

3. Fork this repository

Fork this repository by pressing the fork button on the upper right.

4. Clone your fork to your computer.

Clone into your ml directory (MY_ML_DIR) using:

git clone MY_REPO_FORK_HTTPS_ADDRESS

5. Start a ml container on your machine

# Replace 'MY_ML_DIR' with your local code directory
docker run -it -d \
    -p 8881:8881 \
    -p 6007:6007 \
    --gpus=all \
    --shm-size 12G \
    -v MY_ML_DIR:/workspace/code \
    --name=bveri_hs24 \
    cr.gitlab.fhnw.ch/ml/courses/bveri/hs2024/bveri-exercises-dev-hs2024:latest

6. Check that your container is running

docker ps -a

7. Connect to your container through your browser

Enter http://localhost:8881/lab in your browser.