Hacker News
new
|
ask
|
show
|
jobs
MegaTrain: Full Precision Training of 100B+ Parameter LLMs on a Single GPU
(arxiv.org)
220 points
by
chrsw
8 hours ago
|
41 comments
Loading...
