Midterm Exam
Logistics
Date/Time:
- Midterm: Tuesday 4/15 in class
- Retest: Thursday 5/1 in class
This exam is open notes, open slides, open worksheets, open readings, and open calculator, but closed everything else (e.g., consulting with others, reviewing assignment code, searching online, using generative AI are not permitted). To avoid excessive printing, you may access these resources in the form of PDF files on your laptop or tablet, but wifi must be turned off and only pre-downloaded resources are allowed. Python code that is already on your computer is not allowed.
To make the collection of these materials easier, I’ve provided a zip file with slides, worksheet solutions, and required readings. You should download this before the exam if you want to use it, and you may add any static resources that you would like (your own notes, additional readings that you’ve found helpful, etc.).
Practice Exam
There is a practice exam available here and a solution here. Some problems are pulled directly from worksheets or adapted from them, but should give you a sense of the type of questions that will be asked. Expect this practice exam to take slightly longer than the real exam.1
Retest
There is a midterm in this class and an optional retest. All questions are graded credit/no credit by subpart (subparts are clearly marked). There is an opportunity to take a retest to attempt questions that you did not get full credit on the first time. For instance, imagine that you get the following on the initial exam:
| Question | Points Earned |
|---|---|
| Regular Expressions | 0.6 |
| Naive Bayes | 1 |
| Part-of-Speech Tagging | 1 |
| Word and Document Representations | 1 |
| Logistic Regression & Neural Networks | 0.5 |
| Transformers & BERT | 1 |
| Text Generation | 0 |
| Evaluating Models | 0.5 |
Your total score is 5.6, so if you’ve completed all homework assignments with 3/3 points, you might decide not to show up for the retest, which is totally fine! You would have earned 26.6 points in the methods category, which is enough for the A-tier in that area.
If you haven’t completed all of the homework assignments yet, you might want to try to improve your exam score. You’ve already earned the maximum possible points for Naive Bayes, Part-of-Speech Tagging, Word Representations, and Transformers & BERT, so you should not attempt those problems on the retest. Let’s say you attempt Regular Expressions, Logistic Regression & Neural Networks, Text Generation, and Evaluating Models, and get the following scores:
| Question | Points Earned |
|---|---|
| Regular Expressions | 0.2 |
| Logistic Regression & Neural Networks | 0 |
| Text Generation | 1 |
| Evaluating Models | 1 |
Your final score is the max of the points earned for each problem across the two exams, which would be as follows:
| Question | Points Earned |
|---|---|
| Regular Expressions | 0.6 |
| Naive Bayes | 1 |
| Part-of-Speech Tagging | 1 |
| Word and Document Representations | 1 |
| Logistic Regression & Neural Networks | 0.5 |
| Transformers & BERT | 1 |
| Text Generation | 1 |
| Evaluating Models | 1 |
Your final exam score would be 7.1.
Footnotes
Because some problems are from worksheets, I’d recommend against using worksheet solutions while completing these problems if you’re trying to get realistic practice for the exam.↩︎