Messages
The WNNLP'25 proceedings are now published, please have a look at the PDF.
The workshop received 22 submissions, of which 19 (by 35 authors) have been accepted for publication as part of the WNNLP 2025 proceedings (this volume). Slides for all the presentations
The Programme Committee has selected the paper "Balancing Accuracy, Efficiency, and Training Cost in Runtime-Constrained Food Hazard Classification" by Aliaan Nadeem Azam, Soban Ali and Sander Harald S?vik for the WNNLP 2025 Best Paper award. In this paper, the authors study the trade-off between predictive performance and inference speeds for classification models. Their main contribution is ...
All teams with the exam papers accepted should prepare a "lightning talk" for the final workshop on June 2nd at 14:15, including a few slides. PhD students will have 5 minutes each to present their paper; MSc teams will have 3 minutes each.
The track chairs will give a short task overview for each track, so you don't need to spend time and space on this in your presentation. Focus on the main contributions and results. Note that the time limit is strict due to our tight program, so be sure to not include more material than you can realistically cover.
The presentation slides should be pushed to the separate public (UiO-)git repository for the workshop. Make sure you push your presentation as a PDF file to the `presentations' directory before the workshop, ...
The acceptance decisions for the papers submitted to the WNNLP 2025 workshop (which also serves as the IN5550/IN9550 exam venue) have been made.
You can check the decision for your paper in your mailbox or on OpenReview submission system. The authors of the accepted submissions are required to revise their paper in the light of reviewers and meta reviewers comments. To make these changes, final, camera-ready papers can use one additional page of text, i.e. the page limit is extended to up to nine pages in total length (still not counting the bibliography).
Final non-anonymous manuscripts (in PDF) must be uploaded through the OpenReview submission system no later than Friday, May 30, 23:59 (anywhere on Earth).
Hi all, here's the final leaderboard for the third obligatory assignment. The students made it to the leaderboard will get one bonus point, if not already at the maximum.
Congrats!
| Rank | Team | Borda | ROUGE-L | BERTScore | LLM-as-a-judge |
|---|---|---|---|---|---|
| 1 | Aliaan Nadeem Azam, Sander S?vik, Sobal Ali | 6 | 17.625 | 84.605 | 43.25 |
| 2 | H?kon Liltved Hyrve, Sander Segb? L?vaas, Filip Flyen Andresen | 2 | 17.092 | 83.003 | 16.25 |
| 3 | Marius Aas Eidsaunet, Victor Moene | 1 | 14.515 | 82.947 | 22.5 |
Hi all, here's the final leaderboard for the second obligatory assignment. The 3 teams of Master students with the best performance scores will get one bonus point, if not already at the maximum.
Congratulations!
| Team | LAS |
|---|---|
| Eivind Morris Bakke, Nora Winger Heggelund | 92.19 |
| Greg Taube | 91.53 |
| Caroline Kristin Vannebo, Kjetil K. Indrehus | 91.42 |
Our final obligatory assignment can be found in our GitHub repository. Your goal is to learn how to work with generative language models and evaluate them in various text writing tasks.
Good luck!
The second obligatory assignment is now online, on the Git repository. Your goal is to build a Norwegian dependency parser with finetuned transformer-based language models.
The first obligatory assignment is now online, on the Git repository. It deals with building a neural document classifier using bags-of-words or word embeddings as features.
The assignment is due March 3, but feel free to start working on it. We have already covered linear classifiers, feed-forward neural networks and word embeddings in the lectures.
We will provide extensive practice in building neural classifiers with PyTorch and word embeddings at the group sessions.
Hi everyone!
We have two important updates for you.
- There are currently issues with the IFI Mattermost which led to us teachers being kicked off from the IN5550 course channel. The system administrators are busy fixing the issues, so we'll be back soon. Meanwhile, please feel free to discuss whatever you want without the teachers. You can still reach us via email or direct messages in Mattermost.
- In case you get error 404 when trying to access the IN5550 GitHub repository, please send us an email with your UiO GitHub username. We will fix the problem.
See you at the group session tomorrow at 10:15!
Hi all,
The slides for the introductory lecture are now published, We have also now published slides and pre-recorded videos for the first "regular" lecture, focused on supervised machine learning (for your convenience, the lecture is split into 4 short episodes).
You can find these materials in the course schedule. On Monday, January 27th, we will have our first online Q&A session (in Zoom). Prior to the session, please watch the videos and read the slides. Note that some videos are pre-recorded in the previous years when this course was offered: thus, it is expected that you can encounter mentions of wrong dates or years other than 2025. In case of any contradictions, check the slides: they are all updated to the current (2025) incarnation of the course.
You are also advised to read chapters 1-3 o...
Welcome to our IN5550/IN9550 course which will guide you through deep learning applications in natural language processing!
The first introductory lecture this term will be held on Monday, January 20, at 14:15 (OJD, Seminarrom Prolog). We will go through course logistics (including routines for assignments and the final project-based exam) and motivate the now dominant use of neural machine learning architectures in Natural Language Processing (and most other sub-fields of Artificial Intelligence). The first lecture will be in-person.
Subsequent lectures (after the introductory one) wi...