©RillNews
new
show
ask
jobs
submit
login
MegaTrain: Full Precision Training of 100B+ Parameter LLMs on a Single GPU
arxiv.org
154 points by
chrsw
4 hours ago
|
35 comments
add comment