GitLab

add RandomVerticalFlip transform

update SVHN docs to note label change for digit 0

* added Inception v3 to index;
* document pretrained models;
* fix typo.

make size param of Scale (h,w) not (w,h)

According to the comment, this should be 1MB which is 1024*1024 bytes.

This reverts commit 36849429.

* load densenet tensors onto cpu

* fix uneven padding in make_grid

* revert incorrectly added commit

* load densenet tensors onto cpu

* fix docs for size param in random + center crop

* revert incorrect commit

(cherry picked from commit 71afec427baca8e37cd9e10d98812bc586e9a4ac)

* added FashionMNIST dataset

* documentation

* fixed formatting

* fixed formatting

* DOC fixed typo in make_grid argument name

* DOC document "padding" make_grid argument

isinstance(padding, tuple) is added to the padding assertion to be able to pass a sequence.

* enable ~ expression in datasets

* fixed root in __init__ to self.root

Wrong order of dimensions In transforms.CenterCrop and transforms.RandomCrop:
Documentation says it should be (w, h), but it actually is (h, w).
Fixing docs to match the code

Make the labels be 1d tensors of type `np.int64`.
Assign label `0`  to data samples of digit 0. The original dataset assigns them label `10`.