You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
On average, it takes around 396 sec., which is recorded in the training log, for an iteration (one epoch of training + one epoch of testing) on a single A100 card. I did not precisely record the time cost for completely training a resnet18 on cifar10. But I guess we can speculate the time cost is about 22 hours (396 * 200 / 60 / 60). The training may cost slightly more time in my practice due to the server's performance.
2022年12月5日 11:27,SpikingGood ***@***.***> 写道:
Hi, I want to know how many hours you spend when you train the model with t=6,cw on resnet18 on CIFAR10 on your GPU.
Thank you for the code, expecting for your reply, thank you in advance!
—
Reply to this email directly, view it on GitHub <#1>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQ7LIFAHYGSZROAUUE4U3ALWLVOJJANCNFSM6AAAAAASTYZRBQ>.
You are receiving this because you are subscribed to this thread.
Hi, I want to know how many hours you spend when you train the model with t=6,cw on resnet18 on CIFAR10 on your GPU.
Thank you for the code, expecting for your reply, thank you in advance!
The text was updated successfully, but these errors were encountered: