BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Scaling Multilingual Generation for Low-Resource Languages - Priya
 nka Agrawal\, Google Deepmind
DTSTART:20240216T120000Z
DTEND:20240216T130000Z
UID:TALK211387@talks.cam.ac.uk
CONTACT:Eric Chamoun
DESCRIPTION:Abstract:\n\nThe availability of large\, high-quality datasets
  has been one of the main drivers of recent progress in generation tasks l
 ike summarization\, QA. Such annotated datasets however are difficult and 
 costly to collect\, and rarely exist in languages other than English\, ren
 dering the technology inaccessible to underrepresented languages. An alter
 native to building large monolingual training datasets is to leverage pre-
 trained language models (PLMs). The talk will first discuss an approach\, 
 QAmeleon\, that tunes a PLM using parameter-efficient fine-tuning methods 
 (PEFT) to synthesize QA data with only five examples per language. Using t
 his data during training delivers accuracy superior to translation-based b
 aselines\, and bridges nearly 60% of the gap between an English-only basel
 ine and a fully supervised upper bound trained on almost 50\,000 hand-labe
 led examples. Next\, the talk will discuss the cross-lingual transfer appr
 oach for a much stricter zero-shot setting to enable generation in unseen 
 languages. Our method composes language and task PEFT modules via element-
 wise arithmetic operations to leverage unlabeled data and labeled data in 
 other languages. The talk further studied the consistency for cross-lingua
 l generation tasks i.e. the output is in a language different from the sou
 rce. Here we propose MuPlan which uses intermediate plans resulting in mor
 e faithful generation in both fine-tuning and zero-shot setups.\n\n\n\nBio
 :\n\nPriyanka Agrawal is a Senior Research Scientist at the Google Deepmin
 d in London\, formally part of Google Brain\, and is focused on building r
 esponsible Generative AI models and scaling them to underrepresented langu
 ages. Prior to that she was a Senior Researcher and Lead at http://Booking
 .com and IBM Research Labs\, where she was driving work in cross-domain tr
 ansfer and representation learning. She is an alumni from the Computer Sci
 ence Department at the Indian Institute of Science. Her work is published 
 at top-tier ML and NLP conferences like NeurIPS\, ACL and she holds 25+ US
  Patents. Priyanka also serves as Area Chair and PC member at these confer
 ences and has been an invited panelist and speaker at various ML/NLP and d
 iversity forums.
LOCATION:Computer Lab\, SS03
END:VEVENT
END:VCALENDAR