BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CFP.DEV//CFP.DEV shortcodes//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Voxxed Days Zurich
X-WR-RELCALID:c426493a-54d4-472b-b975-6f3aaa7163fa
X-WR-TIMEZONE:Europe/Zurich
BEGIN:VEVENT
UID:cfp-1151-14301@cfp.dev
DTSTAMP:20260920T145955Z
SEQUENCE:0
STATUS:CONFIRMED
DTSTART:20250325T100500Z
DTEND:20250325T105000Z
SUMMARY:Things you never dared to ask about LLMs
DESCRIPTION:Large Language Models (LLMs) have taken the world by storm\, po
 wering applications from chatbots to content generation. Yet\, beneath the
  surface\, these models remain enigmatic. This presentation will “delve
 ” into the hidden corners of LLM technology that often leave developers 
 scratching their heads. It’s time to ask those questions you’ve never 
 dared ask about the mysteries underpinning LLMs.Here are some questions we
 ’ll to answer:Do you wonder why LLMs spit tokens instead of words? Where
  do those tokens come from?What’s the difference between a “foundation
 ” / “pre-trained” model\, and an “instruction-tuned” one? We’r
 e often tweaking (hyper)parameters like temperature\, top-p\, top-k\, but 
 do you know how they really affect how tokens are picked up?Quantization m
 akes models smaller\, but what are all those number encodings like fp32\, 
 bfloat16\, int8\, etc?LLMs are good at translation\, right? Do you speak t
 he Base64 language too?We’ll realize together that LLMs are far from per
 fect:We’ve all heard about hallucinations\, or should we say confabulati
 ons?What is this reversal curse that makes LLMs ignore some facts from a d
 ifferent viewpoint?You’d think that LLMs are deterministic at low temper
 ature\, but you’d be surprised by how the context influences LLMs’ ans
 wers…Buckle up\, it’s time to dispel the magic of LLMs\, and ask those
  questions we never dared to ask!\nSpeakers: Guillaume Laforge\nTrack: Dat
 a & AI
URL:https://vdz25.voxxeddays.ch/talk/things-you-never-dared-to-ask-about-ll
 ms/
LOCATION:Room 3\nARENA Cinemas\, Sihlcity\nZürich\, Switzerland
CATEGORIES:Data & AI
END:VEVENT
END:VCALENDAR