Giters

bentrevett / pytorch-image-classification

Tutorials on how to implement a few key architectures for image classification using PyTorch and TorchVision.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

explanation in 3_alexnet

yf291115925 opened this issue · comments

commented

The images in the CIFAR10 dataset are significantly more complex than the MNIST dataset. They are larger, have three times the amount of pixels and are more cluttered. This makes them harder to learn and consequently means we should use less augmentation.

less --> more?