BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Asia/Jerusalem
BEGIN:DAYLIGHT
DTSTART:20260327T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0300
RRULE:FREQ=YEARLY;BYDAY=-1FR;BYMONTH=3
TZNAME:IDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20261025T010000
TZOFFSETFROM:+0300
TZOFFSETTO:+0200
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:IST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260518T095142Z
UID:EDCA4F91-BBA2-49F9-8D1D-ACE16CACB162
DTSTART;TZID=Asia/Jerusalem:20260413T180000
DTEND;TZID=Asia/Jerusalem:20260413T193000
DESCRIPTION:The brain is the perfect place to look for inspiration to devel
 op more efficient neural networks. Inspired by the recurrent dynamics of\n
 \nbiological neurons\, this talk will present several frontier reasoning L
 LMs developed in my lab\, from software to device deployments. Trained end
 -to-end in an academic lab on a full production pipeline (data curation\, 
 pre-training\, to post-training and alignment) these models\n\nsurpass all
  leading LLMs from Meta\, Google and every other over-resourced company in
  the ~10- billion parameter regime\, despite being ~5x smaller . We have d
 eployed several of our models on neuromorphic hardware at 2-watts\, bringi
 ng SoTA-level reasoning from the datacenter to the edge.\n\nSpeaker(s): \,
  Prof. Jason Eshraghian\n\nVirtual: https://events.vtools.ieee.org/m/55228
 4
LOCATION:Virtual: https://events.vtools.ieee.org/m/552284
ORGANIZER:freddy.gabbay@mail.huji.ac.il
SEQUENCE:27
SUMMARY:Neuromorphic Language Models - Technion ACRC Webinar in collaborati
 on with IEEE CS Israel Chapter
URL;VALUE=URI:https://events.vtools.ieee.org/m/552284
X-ALT-DESC:Description: <br /><p class="p1">The brain is the perfect place 
 to look for&nbsp\;inspiration to develop more efficient neural&nbsp\;netwo
 rks. Inspired by the recurrent dynamics of</p>\n<p class="p1">biological n
 eurons\, this talk will present several&nbsp\;frontier reasoning LLMs deve
 loped in my lab\,&nbsp\;from software to device deployments. Trained&nbsp\
 ;end-to-end in an academic lab on a full&nbsp\;production pipeline (data c
 uration\, pre-training\, <span class="s1">to post-training and alignment) 
 these models</span></p>\n<p class="p1">surpass all leading LLMs from Meta\
 , Google and&nbsp\;every other over-resourced company in the ~10-&nbsp\;bi
 llion parameter regime\, despite being ~5x&nbsp\;smaller . We have deploye
 d several of our models&nbsp\;on neuromorphic hardware at 2-watts\, bringi
 ng&nbsp\;SoTA-level reasoning from the datacenter to the&nbsp\;edge.</p>\n
 <p class="p1">&nbsp\;</p>
END:VEVENT
END:VCALENDAR