ZF/vit-pytorch
1
0
Fork 0
mirror of https://github.com/lucidrains/vit-pytorch.git synced 2025-12-30 08:02:29 +00:00
Code Issues Packages Projects Releases Wiki Activity

allow for training on different image sizes, provided images are smaller than what was passed as image_size keyword on init

Browse Source
This commit is contained in:
Phil Wang
2020-10-25 13:17:42 -07:00
parent 6d1df1a970
commit 7a214d7109
3 changed files with 7 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
setup.py
View File

@@ -3,7 +3,7 @@ from setuptools import setup, find_packages
setup(
name = 'vit-pytorch',
packages = find_packages(exclude=['examples']),
version = '0.2.5',
version = '0.2.6',
license='MIT',
description = 'Vision Transformer (ViT) - Pytorch',
author = 'Phil Wang',

5
vit_pytorch/efficient.py
View File

@@ -30,10 +30,11 @@ class ViT(nn.Module):
x = rearrange(img, 'b c (h p1) (w p2) -> b (h w) (p1 p2 c)', p1 = p, p2 = p)
x = self.patch_to_embedding(x)
b, n, _ = x.shape
cls_tokens = self.cls_token.expand(img.shape[0], -1, -1)
cls_tokens = self.cls_token.expand(b, -1, -1)
x = torch.cat((cls_tokens, x), dim=1)
x += self.pos_embedding
x += self.pos_embedding[:, :(n + 1)]
x = self.transformer(x)
x = self.to_cls_token(x[:, 0])

5
vit_pytorch/vit_pytorch.py
View File

@@ -113,10 +113,11 @@ class ViT(nn.Module):
x = rearrange(img, 'b c (h p1) (w p2) -> b (h w) (p1 p2 c)', p1 = p, p2 = p)
x = self.patch_to_embedding(x)
b, n, _ = x.shape
cls_tokens = self.cls_token.expand(img.shape[0], -1, -1)
cls_tokens = self.cls_token.expand(b, -1, -1)
x = torch.cat((cls_tokens, x), dim=1)
x += self.pos_embedding
x += self.pos_embedding[:, :(n + 1)]
x = self.dropout(x)
x = self.transformer(x, mask)