Week 4 Delivery Guide

Learning Objectives Assessment

IMPORTANT: Please read and share this with the students: Student Notes, to setup expectations for the entire week.

It is up to you to decide how you should:

  1. Discourage cheating and encourage effort
  2. Measure the actual acquisition of skills and knowledge (preventing both: underfitting and overfitting)
  3. Make the learning process a self-rewarding experience unto itself

Successful Delivery (suggestions)

Below are some suggestions for you to consider when delivering this week.

1. Require Proof of Effort

To ensure students are actively working on the assignments, and control for copy-pasting, GenAI Policy Violations, or worse yet: having another student do the work for them:

  • students MUST commit to GitHub repos on each question of each assignment (mentioning the question number in the commit message)
  • to make things easier, some lab work includes instant-feedback exercises via Unit Tests, so that’s a clear checkpoint for a commit

You can git -C ./student_repo log to check the commit history, which should look something like this:

commit 2d0273722392f26a723b862de7c837642b3b096b
Author: Renadalju <riihyunee@gmail.com>
Date:   Thu Jan 1 18:49:54 2026 +0300

    solve M7.03: use the cross_validate function and compute multiple scores/errors

commit 78fd8f2d3633fe31ee64077ab5872ac146ef7687
Author: Renadalju <riihyunee@gmail.com>
Date:   Thu Jan 1 18:46:17 2026 +0300

    solve M7.03: use the mean absolute error (MAE)

commit bfd369daa6fdd471588a91bc381f5531a4c52700
Author: Renadalju <riihyunee@gmail.com>
Date:   Thu Jan 1 18:44:20 2026 +0300

    solve M7.03: estimate the generalization performance of the model

commit 6830b25ca215ed486ea1febee7d5ee751f39af29
Author: Renadalju <riihyunee@gmail.com>
Date:   Thu Jan 1 18:42:43 2026 +0300

    solve M7.03:  create a linear regression model.

commit 62cfe5a1669a875f99515d2ed6f3e94f0fbe742d
Author: Renadalju <riihyunee@gmail.com>
Date:   Thu Jan 1 18:41:14 2026 +0300

    add Exercise M7.03

Notice how adding the exercise (before solving it) should be the first commit.

2. Encourage Application of Skills

A Project’s goal is demonstrating some level of mastery in applying the knowledge & skills acquired in this week (+ previous weeks). Note: just like the assignments, commits of the project should be frequent and meaningful work-units (with clear commit messages).

2.1 Require Proof of Contribution

“Teamwork”.. to control for “oh we worked on this together so I copied it” to justify HUGE commits. Each student must contribute using their own GitHub account:

  • Can clone with write access contributor
  • Can collaborate using Pull Requests (PRs)
  • Can use VS Code Live Share for pair programming

No excuses for not committing:

  • Make it clear that any untracked work is as good as no work
  • Huge commits are no work either

3. Reward Extra Effort

  • Bonus points awarded to students who actively participate in lecture, asking the right questions (not for the sake of just asking),
  • and helping others when needed
  • Optional exercises do exist for the advanced students
  • When they methodically/systematically improve their training/inference performance, significant enough to be noticeable:
    • accuracy
    • resource utilization
    • latency/speed
    • design simplicity (code complexity, readability, maintainability, etc.)