allow for mean pool with efficient version too

Update README and documentation

README.md

@@ -1,9 +1,11 @@
-<img src="./vit.png" width="500px"></img>
+<img src="./vit.gif" width="500px"></img>
 
 ## Vision Transformer - Pytorch
 
 Implementation of <a href="https://openreview.net/pdf?id=YicbFdNTTy">Vision Transformer</a>, a simple way to achieve SOTA in vision classification with only a single transformer encoder, in Pytorch. Significance is further explained in <a href="https://www.youtube.com/watch?v=TrdevFK_am4">Yannic Kilcher's</a> video. There's really not much to code here, but may as well lay it out for everyone so we expedite the attention revolution.
 
+For a Pytorch implementation with pretrained models, please see Ross Wightman's repository <a href="https://github.com/rwightman/pytorch-image-models">here</a>
+
 ## Install
 
 ```bash
@@ -22,7 +24,7 @@ v = ViT(
     num_classes = 1000,
     dim = 1024,
     depth = 6,
-    heads = 8,
+    heads = 16,
     mlp_dim = 2048,
     dropout = 0.1,
     emb_dropout = 0.1
@@ -34,6 +36,30 @@ mask = torch.ones(1, 8, 8).bool() # optional mask, designating which patch to at
 preds = v(img, mask = mask) # (1, 1000)
 ```
 
+## Parameters
+- `image_size`: int.  
+Image size.
+- `patch_size`: int.  
+Number of patches. `image_size` must be divisible by `patch_size`.  
+The number of patches is: ` n = (image_size // patch_size) ** 2` and `n` **must be greater than 16**.
+- `num_classes`: int.  
+Number of classes to classify.
+- `dim`: int.  
+Last dimension of output tensor after linear transformation `nn.Linear(..., dim)`.
+- `depth`: int.  
+Number of Transformer blocks.
+- `heads`: int.  
+Number of heads in Multi-head Attention layer. 
+- `mlp_dim`: int.  
+Dimension of the MLP (FeedForward) layer. 
+- `channels`: int, default `3`.  
+Number of image's channels. 
+- `dropout`: float between `[0, 1]`, default `0.`.  
+Dropout rate. 
+- `emb_dropout`: float between `[0, 1]`, default `0`.  
+Embedding dropout rate.
+- `pool`: string, either `cls` token pooling or `mean` pooling
+
 ## Research Ideas
 
 ### Self Supervised Training
@@ -64,7 +90,7 @@ model = ViT(
 learner = BYOL(
     model,
     image_size = 256,
-    hidden_layer = 'to_cls_token'
+    hidden_layer = 'to_latent'
 )
 
 opt = torch.optim.Adam(learner.parameters(), lr=3e-4)
@@ -126,13 +152,23 @@ Other sparse attention frameworks I would highly recommend is <a href="https://g
 ## Citations
 
 ```bibtex
-@inproceedings{
-    anonymous2021an,
-    title={An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale},
-    author={Anonymous},
-    booktitle={Submitted to International Conference on Learning Representations},
-    year={2021},
-    url={https://openreview.net/forum?id=YicbFdNTTy},
-    note={under review}
+@misc{dosovitskiy2020image,
+    title   = {An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale},
+    author  = {Alexey Dosovitskiy and Lucas Beyer and Alexander Kolesnikov and Dirk Weissenborn and Xiaohua Zhai and Thomas Unterthiner and Mostafa Dehghani and Matthias Minderer and Georg Heigold and Sylvain Gelly and Jakob Uszkoreit and Neil Houlsby},
+    year    = {2020},
+    eprint  = {2010.11929},
+    archivePrefix = {arXiv},
+    primaryClass = {cs.CV}
+}
+```
+
+```bibtex
+@misc{vaswani2017attention,
+    title   = {Attention Is All You Need},
+    author  = {Ashish Vaswani and Noam Shazeer and Niki Parmar and Jakob Uszkoreit and Llion Jones and Aidan N. Gomez and Lukasz Kaiser and Illia Polosukhin},
+    year    = {2017},
+    eprint  = {1706.03762},
+    archivePrefix = {arXiv},
+    primaryClass = {cs.CL}
 }
 ```

setup.py

@@ -3,7 +3,7 @@ from setuptools import setup, find_packages
 setup(
   name = 'vit-pytorch',
   packages = find_packages(exclude=['examples']),
-  version = '0.2.3',
+  version = '0.5.1',
   license='MIT',
   description = 'Vision Transformer (ViT) - Pytorch',
   author = 'Phil Wang',

vit_pytorch/efficient.py

@@ -1,11 +1,12 @@
 import torch
-from einops import rearrange
+from einops import rearrange, repeat
 from torch import nn
 
 class ViT(nn.Module):
-    def __init__(self, *, image_size, patch_size, num_classes, dim, transformer, channels = 3):
+    def __init__(self, *, image_size, patch_size, num_classes, dim, transformer, pool = 'cls', channels = 3):
         super().__init__()
         assert image_size % patch_size == 0, 'image dimensions must be divisible by the patch size'
+        assert pool in {'cls', 'mean'}, 'pool type must be either cls (cls token) or mean (mean pooling)'
         num_patches = (image_size // patch_size) ** 2
         patch_dim = channels * patch_size ** 2
 
@@ -16,13 +17,12 @@ class ViT(nn.Module):
         self.cls_token = nn.Parameter(torch.randn(1, 1, dim))
         self.transformer = transformer
 
-        self.to_cls_token = nn.Identity()
+        self.pool = pool
+        self.to_latent = nn.Identity()
 
         self.mlp_head = nn.Sequential(
             nn.LayerNorm(dim),
-            nn.Linear(dim, dim * 4),
-            nn.GELU(),
-            nn.Linear(dim * 4, num_classes)
+            nn.Linear(dim, num_classes)
         )
 
     def forward(self, img):
@@ -30,11 +30,14 @@ 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 = repeat(self.cls_token, '() n d -> b n d', b = b)
         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])
+        x = x.mean(dim = 1) if self.pool == 'mean' else x[:, 0]
+
+        x = self.to_latent(x)
         return self.mlp_head(x)

vit_pytorch/vit_pytorch.py

@@ -1,6 +1,6 @@
 import torch
 import torch.nn.functional as F
-from einops import rearrange
+from einops import rearrange, repeat
 from torch import nn
 
 MIN_NUM_PATCHES = 16
@@ -34,29 +34,31 @@ class FeedForward(nn.Module):
         return self.net(x)
 
 class Attention(nn.Module):
-    def __init__(self, dim, heads = 8, dropout = 0.):
+    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
         super().__init__()
+        inner_dim = dim_head *  heads
         self.heads = heads
         self.scale = dim ** -0.5
 
-        self.to_qkv = nn.Linear(dim, dim * 3, bias = False)
+        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias = False)
         self.to_out = nn.Sequential(
-            nn.Linear(dim, dim),
+            nn.Linear(inner_dim, dim),
             nn.Dropout(dropout)
         )
 
     def forward(self, x, mask = None):
         b, n, _, h = *x.shape, self.heads
-        qkv = self.to_qkv(x)
-        q, k, v = rearrange(qkv, 'b n (qkv h d) -> qkv b h n d', qkv = 3, h = h)
+        qkv = self.to_qkv(x).chunk(3, dim = -1)
+        q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b h n d', h = h), qkv)
 
         dots = torch.einsum('bhid,bhjd->bhij', q, k) * self.scale
+        mask_value = -torch.finfo(dots.dtype).max
 
         if mask is not None:
             mask = F.pad(mask.flatten(1), (1, 0), value = True)
             assert mask.shape[-1] == dots.shape[-1], 'mask has incorrect dimensions'
             mask = mask[:, None, :] * mask[:, :, None]
-            dots.masked_fill_(~mask, float('-inf'))
+            dots.masked_fill_(~mask, mask_value)
             del mask
 
         attn = dots.softmax(dim=-1)
@@ -67,12 +69,12 @@ class Attention(nn.Module):
         return out
 
 class Transformer(nn.Module):
-    def __init__(self, dim, depth, heads, mlp_dim, dropout):
+    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout):
         super().__init__()
         self.layers = nn.ModuleList([])
         for _ in range(depth):
             self.layers.append(nn.ModuleList([
-                Residual(PreNorm(dim, Attention(dim, heads = heads, dropout = dropout))),
+                Residual(PreNorm(dim, Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout))),
                 Residual(PreNorm(dim, FeedForward(dim, mlp_dim, dropout = dropout)))
             ]))
     def forward(self, x, mask = None):
@@ -82,12 +84,13 @@ class Transformer(nn.Module):
         return x
 
 class ViT(nn.Module):
-    def __init__(self, *, image_size, patch_size, num_classes, dim, depth, heads, mlp_dim, channels = 3, dropout = 0., emb_dropout = 0.):
+    def __init__(self, *, image_size, patch_size, num_classes, dim, depth, heads, mlp_dim, pool = 'cls', channels = 3, dim_head = 64, dropout = 0., emb_dropout = 0.):
         super().__init__()
-        assert image_size % patch_size == 0, 'image dimensions must be divisible by the patch size'
+        assert image_size % patch_size == 0, 'Image dimensions must be divisible by the patch size.'
         num_patches = (image_size // patch_size) ** 2
         patch_dim = channels * patch_size ** 2
-        assert num_patches > MIN_NUM_PATCHES, f'your number of patches ({num_patches}) is way too small for attention to be effective. try decreasing your patch size'
+        assert num_patches > MIN_NUM_PATCHES, f'your number of patches ({num_patches}) is way too small for attention to be effective (at least 16). Try decreasing your patch size'
+        assert pool in {'cls', 'mean'}, 'pool type must be either cls (cls token) or mean (mean pooling)'
 
         self.patch_size = patch_size
 
@@ -96,17 +99,14 @@ class ViT(nn.Module):
         self.cls_token = nn.Parameter(torch.randn(1, 1, dim))
         self.dropout = nn.Dropout(emb_dropout)
 
-        self.transformer = Transformer(dim, depth, heads, mlp_dim, dropout)
+        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout)
 
-        self.to_cls_token = nn.Identity()
+        self.pool = pool
+        self.to_latent = nn.Identity()
 
         self.mlp_head = nn.Sequential(
             nn.LayerNorm(dim),
-            nn.Linear(dim, mlp_dim),
-            nn.GELU(),
-            nn.Dropout(dropout),
-            nn.Linear(mlp_dim, num_classes),
-            nn.Dropout(dropout)
+            nn.Linear(dim, num_classes)
         )
 
     def forward(self, img, mask = None):
@@ -114,13 +114,16 @@ 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 = repeat(self.cls_token, '() n d -> b n d', b = b)
         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)
 
-        x = self.to_cls_token(x[:, 0])
+        x = x.mean(dim = 1) if self.pool == 'mean' else x[:, 0]
+
+        x = self.to_latent(x)
         return self.mlp_head(x)