search

brjathu / iTAML

Official implementation of "iTAML : An Incremental Task-Agnostic Meta-learning Approach". CVPR 2020
96 stars 16 forks source link

something wrong happends in training cifar10 #3

Closed wangjl1993 closed 4 years ago

wangjl1993 commented 4 years ago

in incremental_dataloader.py:

class iCIFAR10(DataHandler): base_dataset = datasets.cifar.CIFAR10 train_transforms = [ transforms.RandomCrop(32, padding=4), transforms.RandomHorizontalFlip(), transforms.RandomRotation(10), transforms.ColorJitter(brightness=63 / 255), transforms.ToTensor(), ] common_transforms = [ transforms.ToTensor(), transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)) ]

train_transforms has no Normalize function, this will get unworkable when testing.

brjathu commented 4 years ago

Hi, I have updated the code, however we haven't done any experiments on CIFAR10, on the paper. It was added for the completeness of the data loader.

wangjl1993 commented 4 years ago

Hi, I have updated the code, however we haven't done any experiments on CIFAR10, on the paper. It was added for the completeness of the data loader.

Hi~ yes, I could get the same acc on cifar100, it is just a friendly reminder. Both your idea and code are really cool! thx~~

brjathu commented 4 years ago

Thank you very much.