Practical offloading for fine-tuning LLM on commodity GPU via learned subspace projectors.

Siyuan Chen Zelong Guan Yudong Liu Phillip B. Gibbons

Published in: CoRR (2024)

Keyphrases

fine tuning
fine tuned
viable alternative
fine tune
real time
real world
subspace clustering
data sets
high dimensional
parallel processing
low dimensional
high speed
graphics hardware
feature space
parallel implementation
data points
graphics processors
clustering high dimensional data