new paper suggests image linear attention is more effective without q…

…uery normalization
lucidrains · Oct 5, 2020 · ef2ca0b · ef2ca0b
1 parent 9f95a03
commit ef2ca0b
Show file tree

Hide file tree

Showing 2 changed files with 1 addition and 2 deletions.
diff --git a/denoising_diffusion_pytorch/denoising_diffusion_pytorch.py b/denoising_diffusion_pytorch/denoising_diffusion_pytorch.py
@@ -169,7 +169,6 @@ def forward(self, x):
         b, c, h, w = x.shape
         qkv = self.to_qkv(x)
         q, k, v = rearrange(qkv, 'b (qkv heads c) h w -> qkv b heads c (h w)', heads = self.heads, qkv=3)
-        q = q.softmax(dim=-2)
         k = k.softmax(dim=-1)
         context = torch.einsum('bhdn,bhen->bhde', k, v)
         out = torch.einsum('bhde,bhdn->bhen', context, q)

diff --git a/setup.py b/setup.py
@@ -3,7 +3,7 @@
 setup(
   name = 'denoising-diffusion-pytorch',
   packages = find_packages(),
-  version = '0.3.2',
+  version = '0.4.0',
   license='MIT',
   description = 'Denoising Diffusion Probabilistic Models - Pytorch',
   author = 'Phil Wang',