- Don't forget to read the project description before starting the assignment.
- All references for the project, such as submission examples or tutorials will be in this project reference folder.
- For a detailed documentation on how to use our GPT Wrapper package, please read this document that is also in the reference folder.
To give us the deliverables you will have to commit the following files if your github classroom repository:
-
✅ For each team member, demonstrations in the format shown in the project reference folder
- Please name the file for each person as
m1_submission_<MEMBER-SCIPER>.json
- Please name the file for each person as
-
✅ For each team member, a PDF of your review as detailed in the project description.
- This means that if you are a team of two, you will only review two papers.
- However, if you are a team of three, you should submit three PDFs.
- Please name the file for each person as
m1_review_<MEMBER-SCIPER>.pdf
-
✅ The plan on how to complete subsequent portions of the project, as detailed in the project description.
By now you must have received an email with your course question package. In this package you will find question instances with no answers that you will use to collect your first set of demonstrations.
You should also have received an API key from an ML4ED lab member (if you filled out the consent form), which you will need to access our GPT Wrapper package. If you still haven't received your API key, it's probably because you haven't filled the consent form (no matter your response you will get an API key, you just need to fill it). This API key is very important as it will sign you interactions.
Moreover, please contact your assigned team TA for help instead of reaching out to the mailing list.
You are free to share tips on how to prompt the model among team members, however, your interactions should ideally look different. Always use your own API key when collecting demonstrations, so be careful when sharing code.
While our TA staff did their best to pick questions that should only involve text, you may notice a few issues while collecting your demonstrations :
- the question requires some multimodality (video, image, audio etc.)
- the question has an empty question string but reasonable answer choices
- etc.
If you run into these problems please send the complete list of those questions to your assigned TA so they can tell you how to further proceed with them.