BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Efficient sentence encoders for Conversational AI in the industry 
 - Inigo Casanueva\, PolyAI
DTSTART:20210218T110000Z
DTEND:20210218T120000Z
UID:TALK157426@talks.cam.ac.uk
CONTACT:Marinela Parovic
DESCRIPTION:Building real-world conversational AI applications requires re
 source-efficient models that can learn in low-data regimes with only a han
 dful of annotated examples. Fully fine-tuning large pretrained language mo
 dels is expensive and computationally intractable for these applications\,
  where fast-paced development cycles are necessary. This talk presents Con
 veRT and ConVEx\; effective\, affordable\, quick-to-train\, and quick-to-f
 ine-tune sentence encoders that work well in such few-shot low-data scenar
 ios. These encoders achieve state-of-the-art performance across a wide ran
 ge of conversational tasks such as response selection\, intent classificat
 ion and value extraction\,  offering quick and effective adaptation to new
  tasks\, domains\, and languages.
LOCATION:https://cam-ac-uk.zoom.us/j/97599459216?pwd=QTRsOWZCOXRTREVnbTJBd
 XVpOXFvdz09
END:VEVENT
END:VCALENDAR