Closed Firegreat123 closed 2 years ago
For testing the distilled images.
On Fri, May 6, 2022 at 23:54 Firegreat123 @.***> wrote:
[image: image] https://user-images.githubusercontent.com/103976695/167237114-f30eb8ca-8ec2-4ad3-91c6-3afde996dc97.png
- Hi, I can understand that 200 lenet networks are trained for distilling images. But what can those 20 test networks do?
— Reply to this email directly, view it on GitHub https://github.com/SsnL/dataset-distillation/issues/47, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABLJMZKPVJZSG2XHVONSUSTVIXSNXANCNFSM5VJZLIDA . You are receiving this because you are subscribed to this thread.Message ID: @.***>
It’s just training networks on real images and store the weights, so that they can be used for testing distilled images in adapt and forget settings. This command does not yet do the testing distilled images part.
On Sat, May 7, 2022 at 05:37 Firegreat123 @.***> wrote:
- Thank you. Is that mean use these 20 networks training with distilled images?
— Reply to this email directly, view it on GitHub https://github.com/SsnL/dataset-distillation/issues/47#issuecomment-1120175869, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABLJMZOW7KXK3MA7KHQBYOLVIY2VXANCNFSM5VJZLIDA . You are receiving this because you commented.Message ID: @.***>
see here for testing https://github.com/SsnL/dataset-distillation/blob/master/docs/advanced.md#testing