"Unlocking Latent Discourse Translation in LLMs" - Wafaa Mohammed (ELLIS, UvA & IT)
Leonor Barreiros
Dear all,
The next session of the Priberam ML Lunch Seminars is scheduled for June 17th at 1 PM at Instituto Superior Técnico (room VA5, Alameda Campus).
This time we have the pleasure of hosting a presentation by Wafaa Mohammed, an ELLIS PhD student at the University of Amsterdam (UvA) and the SARDINE lab of Instituto de Telecomunicações (IT). Her work focuses on context-aware machine translation aiming to build machine translation systems that are able to handle context-dependent discourse phenomena while ensuring high overall translation quality.
In this talk, Wafaa will present the most recent advances in investigating the discourse phenomena performance of LLMs for document-level translation. Namely, she will present the proposed quality-aware decoding (QAD) method to extract discourse knowledge encoded within LLMs during generation, and its advantages over existing decoding methods.
As usual, the lunch bags are on us! :)
To learn more about the event and to register (mandatory if you'd like to attend in person), please follow the link below:
https://www.eventbrite.co.uk/e/unlocking-latent-discourse-translation-in-llms-tickets-1396202722929
Otherwise, you can follow on zoom via the following link:
https://us02web.zoom.us/j/89453717055
We look forward to seeing you all there!
Kind regards,
Leonor and the Priberam ML Lunch Seminars team