Coursework and Grading
Throughout this course, you will complete:
- 3 homework assignments
- a number of reading quizzes
- a final project.
The composition of your final grade is as follows.
| Item | Percentage |
|---|---|
| Participation | 8% |
| Reading Quizzes | 12% |
| Homework Assignments | 30% |
| Project | 50% |
| Total | 100% |
Submissions and Deadlines
Unless otherwise stated, all submission deadlines are on Wednesdays at 9:30 AM.1 Each coursework item will be released on Brightspace, and submissions and grading will take place on Gradescope (which you can access through Brightspace). Materials and further instructions for coursework items may occasionally be released on GitHub; please see Brightspace for details specific to each assignment.
Late Submissions
Late submissions may be accepted for some coursework items. Please consult the instructions for each coursework item for specific guidelines. Unless stated otherwise, late work will be penalized by 1 percentage point of the item’s total grade for each hour (full or partial) that the work is late. However, you will have a grace period of 168 hours over the entire semester during which you can submit late work without penalty. You do not have to ask permission to use the grace period; it will be calculated at the end of the semester so as to optimize your final grade.2
Extensions
Extensions for documented emergencies may be granted the discretion of the instructor. Do not contact the TAs for extension requests.
Extra Credit
There are many opportunities for extra credit in this course. Unless stated otherwise, late submissions will not be accepted for extra credit assignments.3 Neither your grade for a coursework item (homework assignment, project component, or reading quiz), nor your grade for a coursework category (total homework grade, total project grade, total reading quiz grade, total participation grade), may exceed the maximum possible grade due to extra credit.
Project
The main requirement of this course is to produce an original research project on natural language understanding (NLU) and/or computational semantics. As with an actual research project in this field, the final product of your project will consist of an ACL-style short paper and an in-class oral presentation of your findings. The length limit of both your paper (4 pages maximimum) and your presentation (TBD) will be very short, but the expectations for clarity and thoroughness will be high.
To guide you through the process of doing research in NLU, your project will be divided into five components, and you will work on it in groups of three or four.
| Item | Due Date | Points | % of Final Grade |
|---|---|---|---|
| Project Mini-Proposal | March 8 | 50 | 5% |
| Full Project Proposal | April 5 | 100 | 10% |
| Final Paper Draft | April 26 | 100 | 10% |
| Final Paper | May 12 (12:00 PM) | 200 | 20% |
| In-Class Presentation | May 12 (8:00–9:50 AM) | 50 | 5% |
| Total | 500 | 50% |
Homework Assignments
The homework assignments are designed to help you practice basic techniques learned in class. They will involve implementing machine learning experiments in Python and answering questions about your results. In addition, there are two extra credit assignments (EC 1 and EC 2) that are designed to help you become familiar with the NumPy and PyTorch packages. EC 1 and EC 2 are strongly recommended for students who are less familiar with linear algebra and numerical computation in Python.
Your total homework grade for the semester is calculated out of 300 points, subject to the following rules.
- Your total homework grade cannot exceed 300, and it cannot be negative. If you have earned more than 300 points including extra credit, then your final grade will be 300. If you have earned fewer than 0 points due to late penalties, then your final grade will be 0.
- Your grade for each assignment cannot exceed the point value shown below. If you have earned more than the total point value including extra credit, then you will simply receive the maximum possible grade.
- You cannot receive a negative grade for HW 1, HW 2, HW 3, EC 1, or EC 2. Blank or missing submissions will receive a grade of 0, and any assignment with a negative score due to late penalties will receive a grade of 0.
- HW 0 is worth 0 points, but it is subject to a late penalty of 1 point per day. The minimum possible grade for HW 0 is −10.
| Item | Due Date | Points | % of Final Grade |
|---|---|---|---|
| HW 0: Sign Up for Course Sites | February 1 | 0 | 0% |
| HW 1: Intrinsic Evaluation of Word Embeddings | February 8 | 100 | 10% |
| HW 2: Text Classification Using LSTMs | February 27 | 100 | 10% |
| HW 3: The 🤗 Transformers Framework | March 29 | 100 | 10% |
| EC 1: NumPy Exercises | February 8 | 25 EC | 2.5% EC |
| EC 2: PyTorch Exercises | February 27 | 25 EC | 2.5% EC |
| Total | 300 | 30% |
Reading Quizzes
You will occasionally be assigned short quizzes that are administered and autograded online on Brightspace.
Participation
For full credit, we expect you to regularly attend and participate in lectures and labs, or contribute relevant questions and ideas on Campuswire.
If no time is given for a particular deadline, please assume that it is 9:30 AM. ↩
For instance, if you submit one homework assignment exactly 3 days late, the project mini-proposal exactly 6 days late, and all other coursework items on time, then your total late penalty will be 48 percentage points of your project mini-proposal grade (2.4 percentage points of your final grade). ↩
This general rule applies to entire assignments or other coursework items deemed as “extra credit”, but it does not apply to required coursework items that happen to contain extra credit problems. ↩