Open kxgong opened 6 months ago
Hi, I used 2x 8xA100 machine to train this code on video datasets. I use accelerate as ddp launcher.
After 8 ~ 9 hours of running, I only ran about 3800 steps.
Is this normal?
Hi, I used 2x 8xA100 machine to train this code on video datasets. I use accelerate as ddp launcher.
After 8 ~ 9 hours of running, I only ran about 3800 steps.
Is this normal?