Seminar assignments and discussions
Before the seminar, first you should read the papers and make note of the points that you think are relevant for discussion. These can be either
(i) points that you do not understand and you would like them further clarified,
(ii) points that are interesting and relevant, especially related other topics of the course,
(iii) points that I suggested to you for discussion.
Do not shy away from discussing technical details, especially if these are unclear as our goal is also to understand the implementations. Resolving these will be very helpful when you attempt your own projects.
A good way to start is to think about the contributions of the work and its relation to the material discussed in the course so far. Since we are looking at original papers rather than course text-books, it may happen that the ideas presented in them leave certain questions open, or they lead to new questions, or you might even disagree with the paper. Therefore, approach the papers critically. This also means that no one here (myself included!) knows everything about the topic.
You should choose a member of your group to write a summary of the discussion (2-3 pages) and submit it on Canvas. (If you are using a shared document to prepare the summary you can also include a link to it.) Since this assignment is for a grade, at the end you should also include a brief description of contribution: who was engaged in the discussion and contributed with points and who prepared the report. To make contributions equal, each member of a group should be at least once responsible for preparing and presenting a report.
Programming assignments and classes Larger files and software that doesn't fit to GitHub can be found in /srv/data/computational-semantics on mltgpu.flov.gu.se. In a subfolder submissions you can place your own large files generated by the code when making intermediate submissions to us. However, for the final submission, please submit all of them on Canvas as described below.
pip install -r requirements.txt
# under venv
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121