CSE152B Spring 2022 HW4

Homework instructions

The homework is in the Jupyter Notebook hw4-CSE152B-release.ipynb.

1. Attempt all questions.

2. Please comment all your code adequately.

3. Include all relevant information such as text answers, output images in notebook.

4. Academic integrity: The homework must be completed individually.

5. Submission instructions:
   (a) Submit the notebook and its PDF version on Gradescope.
   (b) To save as PDF, use Ctrl + P -> Save as PDF (toggling Headers and footers, Background graphics).
   (c) Rename your submission files as Lastname_Firstname.ipynb and Lastname_Firstname.pdf.
   (d) Correctly select pages for each answer on Gradescope to allow proper grading.

6. Due date: Assignments are due Fri, Jun 3, by 11pm PST.

Instructions for setting up the conda environment locally

conda create --name cse152b-hw4-py38 python=3.8 pip
conda activate cse152b-hw4-py38
pip install -r requirements.txt
conda install nb_conda

If you wish to work on the assignment on your local machine, here is the link for the datasets for this assignment. You may want to download the datasets and organize them in the same structure as the public dataset on the cluster.

Extra Instructions

Fetch output files from the cluster

From your local machine:

scp -r <USERNAME>@dsmlp-login.ucsd.edu:{path to files on the cluster} {LOCAL PATH}

Attach the container from your terminal

Once you launch a container from the JupyterHub portal, you can also access the container with command line from your local terminal:

• ssh {your ucsd id}@dsmlp-login.ucsd.edu # use your UCSD credentials; UCSD VPN connectin may be needed
• get active container via kubectl get pods
• attach to the pod via kubesh {pod name you got from above}
• then you will be in the bash environment inside the container, identical to the terminal launched from Jupyter Notebook.

Launch the container from your terminal

Alternatively, you can launch a container solely from commandline. Follow those steps:

• ssh {your ucsd id}@dsmlp-login.ucsd.edu # use your UCSD credentials; UCSD VPN connectin may be needed
• Launch your pod.
  • You should enter a node with 1 GPU, 8 CPU, 16 GB RAM, with normal priority (running up to 6 hours
    • launch-scipy-ml.sh -g 1 -m 16 -c 8 -p normal
  • To enable longer runtime k (up to 12) hours with normal priority
    • K8S_TIMEOUT_SECONDS=$((3600*k)) launch-scipy-ml.sh -g 1 -m 16 -c 8 -p normal
  • To enable longer runtime k (more than 12) hours with lower priority
    • K8S_TIMEOUT_SECONDS=$((3600*k)) launch-scipy-ml.sh -g 1 -m 16 -c 8
  • To run your container in the background up to 12 hours, add -b to above command. See details here.
• You will be provided with a URL that you can open locally. Click on the link and navigate to the Jupyter notebook.
• If you cannot launch a pod, set up the environment following these instructions.

Maintain a session within a container

There are cases that you may want to maintain your current session (e.g. a Python training job) within the container when you need to go offline. You can achieve this with session managers like tmux.

For quick start,

• Just run tmux on your terminal once you get into the container.
• To detach and come back later, use ctrl + b then d. To attach next time, use ctrl + b then a.
• For more TMUX usages please refer to online tutorials like https://linuxize.com/post/getting-started-with-tmux/ ot this post