This homework is due on 11:59 PM ET Nov 15, 2024. We will solve two tasks with Transformers.
Instructions:
- Starter code and full assignment can be found here.
- Please submit your solutions on Gradescope.
- Your hand-in should consist of two parts: 1) a single PDF report and 2) all your code in a compressed zip file. Please name the report "AndrewID.pdf" and the code "AndrewID.zip".
- It is okay to ask Google for help, but it is not okay to plagiarize. Feel free to search “how to add cross entropy loss in pytorch,” but do not copy and paste the code from online repositories.
- The solution will be graded by the final result as well as the correctness of the code. You will get partial credits even if the final result is not perfect.
- Late Days: Each student has 5 late days for the whole semester, tracked by Gradescope. Each additional late day will incur a 10% penalty on that homework.
- Start early!
Acknowledgements: This assignment is the result of the collective effort of all TAs that served this course over the last years. We thank all of them!