BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CREST - ECPv5.1.3//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:CREST
X-ORIGINAL-URL:https://crest.science
X-WR-CALDESC:Events for CREST
BEGIN:VTIMEZONE
TZID:Europe/Helsinki
BEGIN:DAYLIGHT
TZOFFSETFROM:+0200
TZOFFSETTO:+0300
TZNAME:EEST
DTSTART:20250330T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0300
TZOFFSETTO:+0200
TZNAME:EET
DTSTART:20251026T010000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Europe/Helsinki:20250512T140000
DTEND;TZID=Europe/Helsinki:20250512T140000
DTSTAMP:20260710T064224
CREATED:20250318T071755Z
LAST-MODIFIED:20250502T130633Z
UID:17967-1747058400-1747058400@crest.science
SUMMARY:Yury POLYANSKIY (MIT) - Optimal Quantization for LLMs and Matrix Multiplication
DESCRIPTION:Statistical Seminar: Every Monday at 2:00 pm.\nTime: 2:00 pm – 3:00 pm\nDate: 12th May\nPlace: 3001 \n  \nYury POLYANSKIY (MIT) – Optimal Quantization for LLMs and Matrix Multiplication \n  \n Abstract:  \nThe main building block of large language models is matrix multiplication\, which is often bottlenecked by the speed of loading these matrices from memory. A number of recent quantization algorithms (SmoothQuant\, GPTQ\, QuIP\, SpinQuant etc) address this issue by storing matrices in lower precision. We derive optimal asymptotic information-theoretic tradeoff between accuracy of the matrix product and compression rate (number of bits per matrix entry). We also show that a non-asymptotic version of our construction (based on nested Gosset lattices and Conway-Sloan decoding)\, which we call NestQuant\, reduces perplexity deterioration almost three-fold compared to the state-of-the-art algorithms (as measured on LLama-2\, Llama-3 with 8B to 70B parameters). Based on a joint work with Or Ordentlich (HUJI)\, Eitan Porat and Semyon Savkin (MIT EECS). \n  \n  \nOrganizers: \nAnna KORBA (CREST)\, Karim LOUNICI (CMAP) \, Jaouad MOURTADA (CREST) \nSponsors:\nCREST-CMAP \n
URL:https://crest.science/event/yury-polyanskiy-mit-tba/
CATEGORIES:Seminars,Statistics
ATTACH;FMTTYPE=:
END:VEVENT
END:VCALENDAR