Project 1 -- Training models better
Due Date: 03/06/2026 23:59
Late Due Date: 03/15/2026 23:59
Download [Jupyter Notebook]
Late Policy
- Late submissions will be allowed up to two days after the original deadline, but may be abridged (e.g. due to the end of the semester). Late submissions will be penalized 20%.
- See Gradescope for specific due dates and late submission deadlines.
- Regrade requests must be made via GradeScope within a week of grades being released.
Project 1 – Training better models.
In this 2-week assignment you will explore ways to improve model training on a fixed architecture model on the Tree-or-Not dataset.
Given the due date right before spring break, we are giving a one time exception on late assignments to end of spring break.
Turn in your completed notebook along with your saved model weights in a file called best.pt on GradeScope.