Overlap between meta-training and meta-testing?

dragen1860 / MAML-Pytorch

Elegant PyTorch implementation of paper Model-Agnostic Meta-Learning (MAML)

MIT License

2.31k stars 420 forks source link

Overlap between meta-training and meta-testing? #56

Closed atseng17 closed 3 years ago

atseng17 commented 3 years ago

I've noticed that based on the custom dataset the author wrote, there could be overlaps meta-training and meta-testing, as the classes are chosen in line 106 in MiniImagenet.py, so when creating the meta-testset, the classes selected from the meta-trainset could reappear again in the test set. Not sure if im correct.

atseng17 commented 3 years ago

Oh just found out that the labels in test.cvs do not apprear in train.csv, sorry about the confusion!