chaht01 / co2l Goto Github PK

Hi,
I was trying to run Cifar100 dataset (5 tasks). But with Cifar10 hyperparameters, I am getting low accuracy.
Do you have a better setting for cifar100? Do you have any results for cifar100?

Please let me know. Thanks

Thank you for your amazing work on this paper and for sharing your work with everyone.
I know that your code is very similar to the PyTorch implementation of SupCon, but it would be really helpful if you shared some examples and how to use tips for your script, especially running the continual learning scheme.

much appreciated, Abdulaziz.

Hello.

I've shocked about your paper's result.

So I want to try your code, but I could not get the result in your paper alike.

Unfortunately, I've runing your repo, I've get Nan of the loss.

Could you tell me why I get the Nan of loss value? and How to get the result what in your paper?

Thank you