BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Fine-Tuning with LoRA: A Practical Guide for the Mistral Model
DTSTART;VALUE=DATE-TIME:20241121T110000Z
DTEND;VALUE=DATE-TIME:20241121T123000Z
DTSTAMP;VALUE=DATE-TIME:20260606T135300Z
UID:indico-event-25@indico.uevora.pt
CONTACT:vbn@uevora.pt
DESCRIPTION:Speakers: Javier León (Laboratório BigData@UE)\n\nFine-tunin
 g Large Language Models (LLMs) with Low-Rank Adaptation (LoRA) is a highly
  efficient technique for optimizing model performance on specific tasks. L
 oRA enhances pre-trained LLMs by updating only a minimal subset of paramet
 ers\, significantly cutting down on computational and memory demands. This
  approach works by incorporating low-rank matrices into the model’s laye
 rs\, enabling precise adjustments without modifying the model’s core arc
 hitecture. By leveraging LoRA\, LLMs can be rapidly and effectively custom
 ized for new domains or tasks\, preserving their broad generalization abil
 ities while achieving superior performance on specialized datasets.\n\nhtt
 ps://indico.uevora.pt/event/25/
LOCATION:125 (Casa Cordovil)
URL:https://indico.uevora.pt/event/25/
END:VEVENT
END:VCALENDAR