News for hackers. Stuff that matters.

A comment on All You Need Is 4x 4090 GPUs to Train Your Own Model. Parent comment

Re:All You Need Is 4x 4090 GPUs to Train Your Own Model by layer8 on Saturday December 28, 2024 @07:19PM (#42536175)
How long does training a 1B or 500M model take approximately on the 4-GPU setup? Or does that dramatically depend on the training data? I didn’t see that info on your pages.
[ Parent | on HN ]
Re:All You Need Is 4x 4090 GPUs to Train Your Own Model by sabareesh on Saturday December 28, 2024 @08:22PM (#42536532)
Roughly it takes 7 days to train on 100B tokens on 500M model
[ Parent | on HN ]