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
Is there any code to train a model on a small-scale dataset? I don't have the resources to train this model ("Training VAR-d16 for 200 epochs on ImageNet 256x256 costs 2.5 days on 16 A100s"), and I would like to have some form of POC.
Would it work for, say, CIFAR10? As the resolution is very small (32x32).
Is there any available code/script for this? Has anyone else tried this?
Thanks
The text was updated successfully, but these errors were encountered:
@Rane90
We haven't tried this on CIFAR10.
Since it works on ImageNet, it will be also ok on CIFAR10, however, due to the small resolution, you should modify the config for var, especially the scale scheduler
Hi,
Thank you for creating this great repo.
Is there any code to train a model on a small-scale dataset? I don't have the resources to train this model ("Training VAR-d16 for 200 epochs on ImageNet 256x256 costs 2.5 days on 16 A100s"), and I would like to have some form of POC.
Would it work for, say, CIFAR10? As the resolution is very small (32x32).
Is there any available code/script for this? Has anyone else tried this?
Thanks
The text was updated successfully, but these errors were encountered: