BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Informal learning seminar on mathematics and AI
DTSTART:20260703T140000Z
DTEND:20260703T150000Z
DTSTAMP:20260706T084400Z
UID:indico-event-1473@math-events.uni-bonn.de
DESCRIPTION:Speakers: Maximilian Schimpf (MPIM)\n\n\n\n\n\n\n\n\nTitle: 
 “Why Transformers can (in theory) do anything”\nLet us imagine that th
 ere existed an optimal "next word prediction" function \\Gamma. For exampl
 e\, if we entered "Is the Riemann hypothesis true?"\, then \\Gamma would o
 utput the answer. Assuming that \\Gamma exists\, we can ask whether it is 
 possible to approximate it with the transformers algorithm. This talk will
  present a positive answer to this question given in the paper "Transforme
 rs are Universal In-context Learners" by Furuya de Hoop and Peyré.\nhttps
 ://sites.google.com/view/informal-math-ai-bonn/home\n\n\n\n\n\n\n\n\nhttps
 ://math-events.uni-bonn.de/event/1473/
LOCATION:0.008 (Mathezentrum)
URL:https://math-events.uni-bonn.de/event/1473/
END:VEVENT
END:VCALENDAR