medium.com/institute-for-…

Just out! Survey on conformal prediction for NLP, on which I had the pleasure to collaborate with the students Margarida Campos and Antonio Farinhas and my colleagues Chryssa Zerva and Andre Martins. arxiv.org/pdf/2405.01976

Come meet me, Chryssa Zerva, Dennis Ulmer 🦋, and Andre Martins today at Hall B #177 in the afternoon poster session (16:30-18:30)!

Last Friday 17th of May, the ELLIS Unit Lisbon held its first Faculty Retreat at the Lisbon Academy of Sciences, bringing together senior members of our Unit for a day full of scientific and strategic discourse. Read more here tinyurl.com/bdhen9tv #TeamBuilding #ML #AI

Σήμερα λήγει η προθεσμία υποβολής αιτήσεων για υποτροφία διατριβής με θέμα Financial LLMs athenarc.gr/el/GD.389.%20A… #NLP #FinNLP #FinTech

Check-out the WMT 2024 metrics shared task!

We offer this 1y postdoc to work with us on the UTTER EU project on LLM-based agents ! Come work with us on 1 or several of these topics: i] managing uncertainty and ambiguity ii] improving the use of conversational context iii] ensuring the safety and alignment of LLMs.

Check out this year's QE Shared Task!! We are looking forward to your MT quality estimation and correction systems 🚀

Save the date!

Happening now! 👇

🚨The QE shared task dates have been updated!🚨 Competition start date (& test set release): 22/07 Competition end date : 30/07 1 more week to prepare your models :D More info: www2.statmt.org/wmt24/qe-task.…

📢WMT24 Quality Estimation Shared Task is ON! 🚨 Submit your QE systems to predict translation quality, detect error spans or even correct the translations! Show us how you can improve referenceless MT evaluation! #WMT #MachineTranslation Submission deadline: July 31 (AoE)

Just out! New paper: “Reranking Laws for Language Generation: A Communication-Theoretic Perspective” with Haau-Sing Li 李 效丞 and Andre Martins: arxiv.org/abs/2409.07131. 1/N

On Friday 4th October we hosted our 2nd ELLIS Unit Lisbon PhD and Postdoc Symposium Instituto Superior Técnico, which brought together over 40 students and postdocs working in the fields of #AI and #ML. 🔗read more here: tinyurl.com/y9ymdhn5 ELLIS

Evasion detection in political interview scenarios by LLMs. Check it out!

👏 tl;dr ~ automatic quality estimation metrics for MT show *undue* preferences towards specific gender marks and words. Check the paper for a discussion on the implications and way forward! #nlp #nlproc

The 9B version of EuroLLM is out! Made in 🇪🇺 and covering all EU 🇪🇺 languages