BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:America/New_York X-LIC-LOCATION:America/New_York BEGIN:DAYLIGHT TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:EDT DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:EST DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20210402T160557Z LOCATION:Track 7 DTSTART;TZID=America/New_York:20201111T170000 DTEND;TZID=America/New_York:20201111T173000 UID:submissions.supercomputing.org_SC20_sess199_ws_dls108@linklings.com SUMMARY:TopiQAL: Topic-aware Question Answering Using Scalable Domain-Spec ific Supercomputers DESCRIPTION:Workshop\n\nTopiQAL: Topic-aware Question Answering Using Scal able Domain-Specific Supercomputers\n\nVenkataram, Mattmann, Penberthy\n\n We all have questions, about today's temperature, scores of our favorite b aseball team, the Universe, and about life during COVID-19. Life, physical and natural scientists have been trying to find answers to various topics using scientific methods and experiments, while computer scientists have built language models as a tiny step towards automatically answering all o f these questions across domains, given a little bit of context. In this p aper, we propose an architecture using state-of-the-art Natural Language P rocessing language models, namely Topic Models and Bidirectional Encoder R epresentations from Transformers (BERT), that can transparently and automa tically retrieve articles of relevance to questions across domains, and fe tch answers to topical questions related to COVID-19 current and historica l medical research literature. We demonstrate the benefits of using domain -specific supercomputers like tensor processing units (TPUs), residing on cloud-based infrastructure, with which we could achieve significant gains in training and inference times, with very minimal cost.\n\nRegistration C ategory: Workshop Reg Pass END:VEVENT END:VCALENDAR