BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Achieving Universality in Machine Translation: M4 - Massively Mult
 ilingual\, Massive MT Models for the Next 1000 Languages - Orhan Firat\, G
 oogle Research
DTSTART:20210311T110000Z
DTEND:20210311T120000Z
UID:TALK158056@talks.cam.ac.uk
CONTACT:Marinela Parovic
DESCRIPTION:What does universality mean for machine translation? Massively
  multilingual models jointly trained on hundreds of languages\, have been 
 showing great success in processing different languages simultaneously in 
 a single large model. These large multilingual models\, which we call M4\,
  are appealing for both efficiency and positive cross-lingual transfer: (1
 ) Training and deploying a single multilingual model requires much less re
 sources than maintaining one model for each language considered\, (2) by t
 ransferring knowledge from high-resource languages\, multilingual models a
 re able to improve performance on low-resource languages. In this talk\, w
 e will be talking about our efforts on scaling machine translation models 
 to more than 1000 languages. We will be detailing several research (and ev
 en some development) challenges that the project has tackled\; multi-task 
 learning with hundreds of tasks\, learning under heavy data imbalance\, un
 derstanding the learned representations\, evaluation at the tail\, cross-l
 ingual down-stream transfer and many more insights will be shared.
LOCATION:https://cam-ac-uk.zoom.us/j/97599459216?pwd=QTRsOWZCOXRTREVnbTJBd
 XVpOXFvdz09
END:VEVENT
END:VCALENDAR