Y
Hacker News
new
|
ask
|
show
|
jobs
by
insumanth
22 hours ago
Anyone not doing distributed training across thousands of GPU, pytorch is the best option. Others have an infra team to work on it.