Multimodal Task Vectors Enable Many-Shot Multimodal In-Context Learning

Brandon Huang, Chancharik Mitra, Assaf Arbelle, Leonid Karlinsky, Trevor Darrell and Roei Herzig

EECS Department
University of California, Berkeley
Technical Report No. UCB/EECS-2025-71
May 15, 2025

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2025/EECS-2025-71.pdf

\"Edit"; ?>

BibTeX citation:

@mastersthesis{Huang:EECS-2025-71,
    Author = {Huang, Brandon and Mitra, Chancharik and Arbelle,  Assaf and Karlinsky,  Leonid and Darrell, Trevor and Herzig, Roei},
    Title = {Multimodal Task Vectors Enable Many-Shot Multimodal In-Context  Learning},
    School = {EECS Department, University of California, Berkeley},
    Year = {2025},
    Month = {May},
    URL = {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2025/EECS-2025-71.html},
    Number = {UCB/EECS-2025-71}
}

EndNote citation:

%0 Thesis
%A Huang, Brandon
%A Mitra, Chancharik
%A Arbelle,  Assaf
%A Karlinsky,  Leonid
%A Darrell, Trevor
%A Herzig, Roei
%T Multimodal Task Vectors Enable Many-Shot Multimodal In-Context  Learning
%I EECS Department, University of California, Berkeley
%D 2025
%8 May 15
%@ UCB/EECS-2025-71
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2025/EECS-2025-71.html
%F Huang:EECS-2025-71