make initial channels customizable for PiT

setup.py

@@ -3,7 +3,7 @@ from setuptools import setup, find_packages
 setup(
   name = 'vit-pytorch',
   packages = find_packages(exclude=['examples']),
-  version = '0.24.1',
+  version = '0.24.2',
   license='MIT',
   description = 'Vision Transformer (ViT) - Pytorch',
   author = 'Phil Wang',

vit_pytorch/pit.py

@@ -129,14 +129,15 @@ class PiT(nn.Module):
         mlp_dim,
         dim_head = 64,
         dropout = 0.,
-        emb_dropout = 0.
+        emb_dropout = 0.,
+        channels = 3
     ):
         super().__init__()
         assert image_size % patch_size == 0, 'Image dimensions must be divisible by the patch size.'
         assert isinstance(depth, tuple), 'depth must be a tuple of integers, specifying the number of blocks before each downsizing'
         heads = cast_tuple(heads, len(depth))
 
-        patch_dim = 3 * patch_size ** 2
+        patch_dim = channels * patch_size ** 2
 
         self.to_patch_embedding = nn.Sequential(
             nn.Unfold(kernel_size = patch_size, stride = patch_size // 2),