some models only return embeddings with some kwarg on forward

README.md

@@ -198,7 +198,7 @@ preds = v(
 ) # (5, 1000)
 ```
 
-Finally, if you would like to make use of a flavor of NaViT using <a href="https://pytorch.org/tutorials/prototype/nestedtensor.html">nested tensors</a> (which will omit a lot of the masking and padding altogether), make sure you are on version `2.4` and import as follows
+Finally, if you would like to make use of a flavor of NaViT using <a href="https://pytorch.org/tutorials/prototype/nestedtensor.html">nested tensors</a> (which will omit a lot of the masking and padding altogether), make sure you are on version `2.5` and import as follows
 
 ```python
 import torch
@@ -2142,4 +2142,43 @@ Coming from computer vision and new to transformers? Here are some resources tha
 }
 ```
 
+```bibtex
+@inproceedings{Liu2017DeepHL,
+    title   = {Deep Hyperspherical Learning},
+    author  = {Weiyang Liu and Yanming Zhang and Xingguo Li and Zhen Liu and Bo Dai and Tuo Zhao and Le Song},
+    booktitle = {Neural Information Processing Systems},
+    year    = {2017},
+    url     = {https://api.semanticscholar.org/CorpusID:5104558}
+}
+```
+
+```bibtex
+@inproceedings{Zhou2024ValueRL,
+    title   = {Value Residual Learning For Alleviating Attention Concentration In Transformers},
+    author  = {Zhanchao Zhou and Tianyi Wu and Zhiyun Jiang and Zhenzhong Lan},
+    year    = {2024},
+    url     = {https://api.semanticscholar.org/CorpusID:273532030}
+}
+```
+
+```bibtex
+@article{Zhu2024HyperConnections,
+    title   = {Hyper-Connections},
+    author  = {Defa Zhu and Hongzhi Huang and Zihao Huang and Yutao Zeng and Yunyao Mao and Banggu Wu and Qiyang Min and Xun Zhou},
+    journal = {ArXiv},
+    year    = {2024},
+    volume  = {abs/2409.19606},
+    url     = {https://api.semanticscholar.org/CorpusID:272987528}
+}
+```
+
+```bibtex
+@inproceedings{Fuller2025SimplerFV,
+    title   = {Simpler Fast Vision Transformers with a Jumbo CLS Token},
+    author  = {Anthony Fuller and Yousef Yassin and Daniel G. Kyrollos and Evan Shelhamer and James R. Green},
+    year    = {2025},
+    url     = {https://api.semanticscholar.org/CorpusID:276557720}
+}
+```
+
 *I visualise a time when we will be to robots what dogs are to humans, and I’m rooting for the machines.* — Claude Shannon

setup.py

@@ -6,10 +6,10 @@ with open('README.md') as f:
 setup(
   name = 'vit-pytorch',
   packages = find_packages(exclude=['examples']),
-  version = '1.8.1',
+  version = '1.11.3',
   license='MIT',
   description = 'Vision Transformer (ViT) - Pytorch',
-  long_description=long_description,
+  long_description = long_description,
   long_description_content_type = 'text/markdown',
   author = 'Phil Wang',
   author_email = 'lucidrains@gmail.com',

vit_pytorch/accept_video_wrapper.py

@@ -0,0 +1,129 @@
+from contextlib import nullcontext
+
+import torch
+from torch import is_tensor, randn
+from torch.nn import Module, Parameter
+from torch.utils._pytree import tree_flatten, tree_unflatten
+
+from einops import rearrange, repeat
+
+# helper functions
+
+def exists(v):
+    return v is not None
+
+def default(v, d):
+    return v if exists(v) else d
+
+# classes
+
+class AcceptVideoWrapper(Module):
+    def __init__(
+        self,
+        image_net: Module,
+        forward_function = 'forward',
+        add_time_pos_emb = False,
+        dim_emb = None,
+        time_seq_len = None,
+        output_pos_add_pos_emb = 0 # defaults to first output position to add embedding 
+    ):
+        super().__init__()
+        self.image_net = image_net
+        self.forward_function = forward_function # for openclip, used in TRI-LBM
+
+        self.add_time_pos_emb = add_time_pos_emb
+        self.output_pos_add_pos_emb = output_pos_add_pos_emb
+
+        if add_time_pos_emb:
+            assert exists(dim_emb) and exists(time_seq_len), '`dim_emb` and `time_seq_len` must be set if adding positional embeddings to the output'
+            self.time_seq_len = time_seq_len
+
+            self.pos_emb = Parameter(randn(time_seq_len, dim_emb) * 1e-2)
+
+    def forward(
+        self,
+        video, # (b c t h w)
+        eval_with_no_grad = False,
+        forward_kwargs = dict()
+    ):
+        add_time_pos_emb = self.add_time_pos_emb
+        time = video.shape[2]
+
+        # maybe validate time positional embedding
+
+        if add_time_pos_emb:
+            assert time <= self.time_seq_len, f'received video with {time} frames but `time_seq_len` ({self.time_seq_len}) is too low'
+
+        video = rearrange(video, 'b c t h w -> b t c h w')
+
+        video = rearrange(video, 'b t ... -> (b t) ...')
+
+        # forward through image net for outputs
+
+        func = getattr(self.image_net, self.forward_function)
+
+        if eval_with_no_grad:
+            self.image_net.eval()
+
+        context = torch.no_grad if eval_with_no_grad else nullcontext
+
+        with context():
+            outputs = func(video, **forward_kwargs)
+
+        # handle multiple outputs, say logits and embeddings returned from extractor - also handle some reduce aux loss being returned
+
+        outputs, tree_spec = tree_flatten(outputs)
+
+        outputs = tuple(rearrange(t, '(b t) ... -> b t ...', t = time) if is_tensor(t) and t.numel() > 1 else t for t in outputs)
+
+        # maybe add time positional embedding
+
+        if add_time_pos_emb:
+
+            outputs = list(outputs)
+            embed = outputs[self.output_pos_add_pos_emb]
+
+            pos_emb = rearrange(self.pos_emb, 't d -> 1 t d')
+
+            # handle the network outputting embeddings with spatial dimensions intact - assume embedded dimension is last
+
+            dims_to_unsqueeze = embed.ndim - pos_emb.ndim
+
+            pos_emb = pos_emb.reshape(*pos_emb.shape[:2], *((1,) * dims_to_unsqueeze) , pos_emb.shape[-1])
+
+            embed = embed + pos_emb
+
+            outputs[self.output_pos_add_pos_emb] = embed
+
+        return tree_unflatten(outputs, tree_spec)
+
+# main
+
+if __name__ == '__main__':
+    from vit_pytorch import ViT
+
+    v = ViT(
+        image_size = 256,
+        patch_size = 32,
+        num_classes = 1000,
+        dim = 1024,
+        depth = 6,
+        heads = 16,
+        mlp_dim = 2048,
+        dropout = 0.1,
+        emb_dropout = 0.1
+    )
+
+    videos = torch.randn(1, 3, 10, 256, 256)
+
+    # step up the difficulty and return embeddings for robotics
+
+    from vit_pytorch.extractor import Extractor
+    v = Extractor(v)
+
+    video_acceptor = AcceptVideoWrapper(v, add_time_pos_emb = True, output_pos_add_pos_emb = 1, time_seq_len = 10, dim_emb = 1024)
+
+    logits, embeddings = video_acceptor(videos, eval_with_no_grad = True) # always (batch, channels, time, height, width) - time is always dimension 2
+
+    assert logits.shape == (1, 10, 1000)
+    assert embeddings.shape == (1, 10, 65, 1024)

vit_pytorch/cct_3d.py

@@ -167,8 +167,10 @@ class Tokenizer(nn.Module):
         stride,
         padding,
         frame_stride=1,
+        frame_padding=None,
         frame_pooling_stride=1,
         frame_pooling_kernel_size=1,
+        frame_pooling_padding=None,
         pooling_kernel_size=3,
         pooling_stride=2,
         pooling_padding=1,
@@ -188,16 +190,22 @@ class Tokenizer(nn.Module):
 
         n_filter_list_pairs = zip(n_filter_list[:-1], n_filter_list[1:])
 
+        if frame_padding is None:
+            frame_padding = frame_kernel_size // 2
+
+        if frame_pooling_padding is None:
+            frame_pooling_padding = frame_pooling_kernel_size // 2
+
         self.conv_layers = nn.Sequential(
             *[nn.Sequential(
                 nn.Conv3d(chan_in, chan_out,
                           kernel_size=(frame_kernel_size, kernel_size, kernel_size),
                           stride=(frame_stride, stride, stride),
-                          padding=(frame_kernel_size // 2, padding, padding), bias=conv_bias),
+                          padding=(frame_padding, padding, padding), bias=conv_bias),
                 nn.Identity() if not exists(activation) else activation(),
                 nn.MaxPool3d(kernel_size=(frame_pooling_kernel_size, pooling_kernel_size, pooling_kernel_size),
                              stride=(frame_pooling_stride, pooling_stride, pooling_stride),
-                             padding=(frame_pooling_kernel_size // 2, pooling_padding, pooling_padding)) if max_pool else nn.Identity()
+                             padding=(frame_pooling_padding, pooling_padding, pooling_padding)) if max_pool else nn.Identity()
             )
                 for chan_in, chan_out in n_filter_list_pairs
             ])
@@ -324,8 +332,10 @@ class CCT(nn.Module):
         n_conv_layers=1,
         frame_stride=1,
         frame_kernel_size=3,
+        frame_padding=None,
         frame_pooling_kernel_size=1,
         frame_pooling_stride=1,
+        frame_pooling_padding=None,
         kernel_size=7,
         stride=2,
         padding=3,
@@ -342,8 +352,10 @@ class CCT(nn.Module):
             n_output_channels=embedding_dim,
             frame_stride=frame_stride,
             frame_kernel_size=frame_kernel_size,
+            frame_padding=frame_padding,
             frame_pooling_stride=frame_pooling_stride,
             frame_pooling_kernel_size=frame_pooling_kernel_size,
+            frame_pooling_padding=frame_pooling_padding,
             kernel_size=kernel_size,
             stride=stride,
             padding=padding,

vit_pytorch/jumbo_vit.py

@@ -0,0 +1,204 @@
+# Simpler Fast Vision Transformers with a Jumbo CLS Token
+# https://arxiv.org/abs/2502.15021
+
+import torch
+from torch import nn
+from torch.nn import Module, ModuleList
+
+from einops import rearrange, repeat, reduce, pack, unpack
+from einops.layers.torch import Rearrange
+
+# helpers
+
+def pair(t):
+    return t if isinstance(t, tuple) else (t, t)
+
+def divisible_by(num, den):
+    return (num % den) == 0
+
+def posemb_sincos_2d(h, w, dim, temperature: int = 10000, dtype = torch.float32):
+    y, x = torch.meshgrid(torch.arange(h), torch.arange(w), indexing="ij")
+    assert divisible_by(dim, 4), "feature dimension must be multiple of 4 for sincos emb"
+
+    omega = torch.arange(dim // 4) / (dim // 4 - 1)
+    omega = temperature ** -omega
+
+    y = y.flatten()[:, None] * omega[None, :]
+    x = x.flatten()[:, None] * omega[None, :]
+    pos_emb = torch.cat((x.sin(), x.cos(), y.sin(), y.cos()), dim=1)
+
+    return pos_emb.type(dtype)
+
+# classes
+
+def FeedForward(dim, mult = 4.):
+    hidden_dim = int(dim * mult)
+    return nn.Sequential(
+        nn.LayerNorm(dim),
+        nn.Linear(dim, hidden_dim),
+        nn.GELU(),
+        nn.Linear(hidden_dim, dim),
+    )
+
+class Attention(Module):
+    def __init__(self, dim, heads = 8, dim_head = 64):
+        super().__init__()
+        inner_dim = dim_head *  heads
+        self.heads = heads
+        self.scale = dim_head ** -0.5
+        self.norm = nn.LayerNorm(dim)
+
+        self.attend = nn.Softmax(dim = -1)
+
+        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias = False)
+        self.to_out = nn.Linear(inner_dim, dim, bias = False)
+
+    def forward(self, x):
+        x = self.norm(x)
+
+        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 = self.heads), qkv)
+
+        dots = torch.matmul(q, k.transpose(-1, -2)) * self.scale
+
+        attn = self.attend(dots)
+
+        out = torch.matmul(attn, v)
+        out = rearrange(out, 'b h n d -> b n (h d)')
+        return self.to_out(out)
+
+class JumboViT(Module):
+    def __init__(
+        self,
+        *,
+        image_size,
+        patch_size,
+        num_classes,
+        dim,
+        depth,
+        heads,
+        mlp_dim,
+        num_jumbo_cls = 1,  # differing from paper, allow for multiple jumbo cls, so one could break it up into 2 jumbo cls tokens with 3x the dim, as an example
+        jumbo_cls_k = 6,    # they use a CLS token with this factor times the dimension - 6 was the value they settled on
+        jumbo_ff_mult = 2,  # expansion factor of the jumbo cls token feedforward
+        channels = 3,
+        dim_head = 64
+    ):
+        super().__init__()
+        image_height, image_width = pair(image_size)
+        patch_height, patch_width = pair(patch_size)
+
+        assert divisible_by(image_height, patch_height) and divisible_by(image_width, patch_width), 'Image dimensions must be divisible by the patch size.'
+
+        patch_dim = channels * patch_height * patch_width
+
+        self.to_patch_embedding = nn.Sequential(
+            Rearrange("b c (h p1) (w p2) -> b (h w) (p1 p2 c)", p1 = patch_height, p2 = patch_width),
+            nn.LayerNorm(patch_dim),
+            nn.Linear(patch_dim, dim),
+            nn.LayerNorm(dim),
+        )
+
+        self.pos_embedding = posemb_sincos_2d(
+            h = image_height // patch_height,
+            w = image_width // patch_width,
+            dim = dim,
+        ) 
+
+        jumbo_cls_dim = dim * jumbo_cls_k
+
+        self.jumbo_cls_token = nn.Parameter(torch.zeros(num_jumbo_cls, jumbo_cls_dim))
+
+        jumbo_cls_to_tokens = Rearrange('b n (k d) -> b (n k) d', k = jumbo_cls_k)
+        self.jumbo_cls_to_tokens = jumbo_cls_to_tokens
+
+        self.norm = nn.LayerNorm(dim)
+        self.layers = ModuleList([])
+
+        # attention and feedforwards
+
+        self.jumbo_ff = nn.Sequential(
+            Rearrange('b (n k) d -> b n (k d)', k = jumbo_cls_k),
+            FeedForward(jumbo_cls_dim, int(jumbo_cls_dim * jumbo_ff_mult)), # they use separate parameters for the jumbo feedforward, weight tied for parameter efficient
+            jumbo_cls_to_tokens
+        )
+
+        for _ in range(depth):
+            self.layers.append(ModuleList([
+                Attention(dim, heads = heads, dim_head = dim_head),
+                FeedForward(dim, mlp_dim),
+            ]))
+
+        self.to_latent = nn.Identity()
+
+        self.linear_head = nn.Linear(dim, num_classes)
+
+    def forward(self, img):
+
+        batch, device = img.shape[0], img.device
+
+        x = self.to_patch_embedding(img)
+
+        # pos embedding
+
+        pos_emb = self.pos_embedding.to(device, dtype = x.dtype)
+
+        x = x + pos_emb
+
+        # add cls tokens
+
+        cls_tokens = repeat(self.jumbo_cls_token, 'nj d -> b nj d', b = batch)
+
+        jumbo_tokens = self.jumbo_cls_to_tokens(cls_tokens)
+
+        x, cls_packed_shape = pack([jumbo_tokens, x], 'b * d')
+
+        # attention and feedforwards
+
+        for layer, (attn, ff) in enumerate(self.layers, start = 1):
+            is_last = layer == len(self.layers)
+
+            x = attn(x) + x
+
+            # jumbo feedforward
+
+            jumbo_cls_tokens, x = unpack(x, cls_packed_shape, 'b * d')
+
+            x = ff(x) + x
+            jumbo_cls_tokens = self.jumbo_ff(jumbo_cls_tokens) + jumbo_cls_tokens
+
+            if is_last:
+                continue
+
+            x, _ = pack([jumbo_cls_tokens, x], 'b * d')
+
+        pooled = reduce(jumbo_cls_tokens, 'b n d -> b d', 'mean')
+
+        # normalization and project to logits
+
+        embed = self.norm(pooled)
+
+        embed = self.to_latent(embed)
+        logits = self.linear_head(embed)
+        return logits
+
+# copy pasteable file
+
+if __name__ == '__main__':
+
+    v = JumboViT(
+        num_classes = 1000,
+        image_size = 64,
+        patch_size = 8,
+        dim = 16,
+        depth = 2,
+        heads = 2,
+        mlp_dim = 32,
+        jumbo_cls_k = 3,
+        jumbo_ff_mult = 2,
+    )
+
+    images = torch.randn(1, 3, 64, 64)
+
+    logits = v(images)
+    assert logits.shape == (1, 1000)

vit_pytorch/na_vit_nested_tensor.py

@@ -6,9 +6,6 @@ from functools import partial
 import torch
 import packaging.version as pkg_version
 
-if pkg_version.parse(torch.__version__) < pkg_version.parse('2.4'):
-    print('nested tensor NaViT was tested on pytorch 2.4')
-
 from torch import nn, Tensor
 import torch.nn.functional as F
 from torch.nn import Module, ModuleList
@@ -44,7 +41,7 @@ def FeedForward(dim, hidden_dim, dropout = 0.):
     )
 
 class Attention(Module):
-    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
+    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0., qk_norm = True):
         super().__init__()
         self.norm = nn.LayerNorm(dim, bias = False)
 
@@ -59,8 +56,8 @@ class Attention(Module):
         # in the paper, they employ qk rmsnorm, a way to stabilize attention
         # will use layernorm in place of rmsnorm, which has been shown to work in certain papers. requires l2norm on non-ragged dimension to be supported in nested tensors
 
-        self.query_norm = nn.LayerNorm(dim_head, bias = False)
-        self.key_norm = nn.LayerNorm(dim_head, bias = False)
+        self.query_norm = nn.LayerNorm(dim_head, bias = False) if qk_norm else nn.Identity()
+        self.key_norm = nn.LayerNorm(dim_head, bias = False) if qk_norm else nn.Identity()
 
         self.dropout = dropout
 
@@ -114,13 +111,13 @@ class Attention(Module):
         return self.to_out(out)
 
 class Transformer(Module):
-    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0.):
+    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0., qk_norm = True):
         super().__init__()
         self.layers = ModuleList([])
 
         for _ in range(depth):
             self.layers.append(ModuleList([
-                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout),
+                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout, qk_norm = qk_norm),
                 FeedForward(dim, mlp_dim, dropout = dropout)
             ]))
 
@@ -149,9 +146,15 @@ class NaViT(Module):
         dim_head = 64,
         dropout = 0.,
         emb_dropout = 0.,
+        qk_rmsnorm = True,
         token_dropout_prob: float | None = None
     ):
         super().__init__()
+
+        if pkg_version.parse(torch.__version__) < pkg_version.parse('2.5'):
+            print('nested tensor NaViT was tested on pytorch 2.5')
+
+
         image_height, image_width = pair(image_size)
 
         # what percent of tokens to dropout
@@ -182,7 +185,7 @@ class NaViT(Module):
 
         self.dropout = nn.Dropout(emb_dropout)
 
-        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout)
+        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout, qk_rmsnorm)
 
         # final attention pooling queries
 
@@ -323,3 +326,5 @@ if __name__ == '__main__':
     ]
 
     assert v(images).shape == (5, 1000)
+
+    v(images).sum().backward()

vit_pytorch/na_vit_nested_tensor_3d.py

@@ -6,9 +6,6 @@ from functools import partial
 import torch
 import packaging.version as pkg_version
 
-if pkg_version.parse(torch.__version__) < pkg_version.parse('2.4'):
-    print('nested tensor NaViT was tested on pytorch 2.4')
-
 from torch import nn, Tensor
 import torch.nn.functional as F
 from torch.nn import Module, ModuleList
@@ -44,7 +41,7 @@ def FeedForward(dim, hidden_dim, dropout = 0.):
     )
 
 class Attention(Module):
-    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
+    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0., qk_norm = True):
         super().__init__()
         self.norm = nn.LayerNorm(dim, bias = False)
 
@@ -59,8 +56,8 @@ class Attention(Module):
         # in the paper, they employ qk rmsnorm, a way to stabilize attention
         # will use layernorm in place of rmsnorm, which has been shown to work in certain papers. requires l2norm on non-ragged dimension to be supported in nested tensors
 
-        self.query_norm = nn.LayerNorm(dim_head, bias = False)
-        self.key_norm = nn.LayerNorm(dim_head, bias = False)
+        self.query_norm = nn.LayerNorm(dim_head, bias = False) if qk_norm else nn.Identity()
+        self.key_norm = nn.LayerNorm(dim_head, bias = False) if qk_norm else nn.Identity()
 
         self.dropout = dropout
 
@@ -86,17 +83,6 @@ class Attention(Module):
 
         # split heads
 
-        def split_heads(t):
-            return t.unflatten(-1, (self.heads, self.dim_head)).transpose(1, 2).contiguous()
-
-        # queries, keys, values
-
-        query = self.to_queries(x)
-        key = self.to_keys(context)
-        value = self.to_values(context)
-
-        # split heads
-
         def split_heads(t):
             return t.unflatten(-1, (self.heads, self.dim_head))
 
@@ -126,13 +112,13 @@ class Attention(Module):
         return self.to_out(out)
 
 class Transformer(Module):
-    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0.):
+    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0., qk_norm = True):
         super().__init__()
         self.layers = ModuleList([])
 
         for _ in range(depth):
             self.layers.append(ModuleList([
-                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout),
+                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout, qk_norm = qk_norm),
                 FeedForward(dim, mlp_dim, dropout = dropout)
             ]))
 
@@ -164,11 +150,15 @@ class NaViT(Module):
         dropout = 0.,
         emb_dropout = 0.,
         num_registers = 4,
+        qk_rmsnorm = True,
         token_dropout_prob: float | None = None
     ):
         super().__init__()
         image_height, image_width = pair(image_size)
 
+        if pkg_version.parse(torch.__version__) < pkg_version.parse('2.5'):
+            print('nested tensor NaViT was tested on pytorch 2.5')
+
         # what percent of tokens to dropout
         # if int or float given, then assume constant dropout prob
         # otherwise accept a callback that in turn calculates dropout prob from height and width
@@ -209,7 +199,7 @@ class NaViT(Module):
 
         self.dropout = nn.Dropout(emb_dropout)
 
-        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout)
+        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout, qk_rmsnorm)
 
         # final attention pooling queries
 
@@ -336,7 +326,7 @@ class NaViT(Module):
 
 if __name__ == '__main__':
 
-    # works for torch 2.4
+    # works for torch 2.5
 
     v = NaViT(
         image_size = 256,
@@ -362,3 +352,5 @@ if __name__ == '__main__':
     ]
 
     assert v(volumes).shape == (5, 1000)
+
+    v(volumes).sum().backward()

vit_pytorch/normalized_vit.py

@@ -76,7 +76,8 @@ class Attention(Module):
 
         self.dropout = dropout
 
-        self.qk_scale = nn.Parameter(torch.ones(dim_head) * (dim_head ** 0.25))
+        self.q_scale = nn.Parameter(torch.ones(heads, 1, dim_head) * (dim_head ** 0.25))
+        self.k_scale = nn.Parameter(torch.ones(heads, 1, dim_head) * (dim_head ** 0.25))
 
         self.split_heads = Rearrange('b n (h d) -> b h n d', h = heads)
         self.merge_heads = Rearrange('b h n d -> b n (h d)')
@@ -94,7 +95,9 @@ class Attention(Module):
         # query key rmsnorm
 
         q, k = map(l2norm, (q, k))
-        q, k = (q * self.qk_scale), (k * self.qk_scale)
+
+        q = q * self.q_scale
+        k = k * self.k_scale
 
         # scale is 1., as scaling factor is moved to s_qk (dk ^ 0.25) - eq. 16
 
@@ -176,18 +179,18 @@ class nViT(Module):
 
         self.to_patch_embedding = nn.Sequential(
             Rearrange('b c (h p1) (w p2) -> b (h w) (c p1 p2)', p1 = patch_size, p2 = patch_size),
-            nn.LayerNorm(patch_dim),
-            nn.Linear(patch_dim, dim),
-            nn.LayerNorm(dim),
+            NormLinear(patch_dim, dim, norm_dim_in = False),
         )
 
-        self.abs_pos_emb = nn.Embedding(num_patches, dim)
+        self.abs_pos_emb = NormLinear(dim, num_patches)
 
         residual_lerp_scale_init = default(residual_lerp_scale_init, 1. / depth)
 
         # layers
 
         self.dim = dim
+        self.scale = dim ** 0.5
+
         self.layers = ModuleList([])
         self.residual_lerp_scales = nn.ParameterList([])
 
@@ -198,8 +201,8 @@ class nViT(Module):
             ]))
 
             self.residual_lerp_scales.append(nn.ParameterList([
-                nn.Parameter(torch.ones(dim) * residual_lerp_scale_init),
-                nn.Parameter(torch.ones(dim) * residual_lerp_scale_init),
+                nn.Parameter(torch.ones(dim) * residual_lerp_scale_init / self.scale),
+                nn.Parameter(torch.ones(dim) * residual_lerp_scale_init / self.scale),
             ]))
 
         self.logit_scale = nn.Parameter(torch.ones(num_classes))
@@ -222,22 +225,23 @@ class nViT(Module):
 
         tokens = self.to_patch_embedding(images)
 
-        pos_emb = self.abs_pos_emb(torch.arange(tokens.shape[-2], device = device))
+        seq_len = tokens.shape[-2]
+        pos_emb = self.abs_pos_emb.weight[torch.arange(seq_len, device = device)]
 
         tokens = l2norm(tokens + pos_emb)
 
         for (attn, ff), (attn_alpha, ff_alpha) in zip(self.layers, self.residual_lerp_scales):
 
             attn_out = l2norm(attn(tokens))
-            tokens = l2norm(tokens.lerp(attn_out, attn_alpha))
+            tokens = l2norm(tokens.lerp(attn_out, attn_alpha * self.scale))
 
             ff_out = l2norm(ff(tokens))
-            tokens = l2norm(tokens.lerp(ff_out, ff_alpha))
+            tokens = l2norm(tokens.lerp(ff_out, ff_alpha * self.scale))
 
         pooled = reduce(tokens, 'b n d -> b d', 'mean')
 
         logits = self.to_pred(pooled)
-        logits = logits * self.logit_scale * (self.dim ** 0.5)
+        logits = logits * self.logit_scale * self.scale
 
         return logits
 

vit_pytorch/simple_vit_with_hyper_connections.py

@@ -0,0 +1,233 @@
+"""
+ViT + Hyper-Connections + Register Tokens
+https://arxiv.org/abs/2409.19606
+"""
+
+import torch
+from torch import nn, tensor
+from torch.nn import Module, ModuleList
+
+from einops import rearrange, repeat, reduce, einsum, pack, unpack
+from einops.layers.torch import Rearrange
+
+# b - batch, h - heads, n - sequence, e - expansion rate / residual streams, d - feature dimension
+
+# helpers
+
+def pair(t):
+    return t if isinstance(t, tuple) else (t, t)
+
+def posemb_sincos_2d(h, w, dim, temperature: int = 10000, dtype = torch.float32):
+    y, x = torch.meshgrid(torch.arange(h), torch.arange(w), indexing="ij")
+    assert (dim % 4) == 0, "feature dimension must be multiple of 4 for sincos emb"
+    omega = torch.arange(dim // 4) / (dim // 4 - 1)
+    omega = 1.0 / (temperature ** omega)
+
+    y = y.flatten()[:, None] * omega[None, :]
+    x = x.flatten()[:, None] * omega[None, :]
+    pe = torch.cat((x.sin(), x.cos(), y.sin(), y.cos()), dim=1)
+    return pe.type(dtype)
+
+# hyper connections
+
+class HyperConnection(Module):
+    def __init__(
+        self,
+        dim,
+        num_residual_streams,
+        layer_index
+    ):
+        """ Appendix J - Algorithm 2, Dynamic only """
+        super().__init__()
+
+        self.norm = nn.LayerNorm(dim, bias = False)
+
+        self.num_residual_streams = num_residual_streams
+        self.layer_index = layer_index
+
+        self.static_beta = nn.Parameter(torch.ones(num_residual_streams))
+
+        init_alpha0 = torch.zeros((num_residual_streams, 1))
+        init_alpha0[layer_index % num_residual_streams, 0] = 1.
+
+        self.static_alpha = nn.Parameter(torch.cat([init_alpha0, torch.eye(num_residual_streams)], dim = 1))
+
+        self.dynamic_alpha_fn = nn.Parameter(torch.zeros(dim, num_residual_streams + 1))
+        self.dynamic_alpha_scale = nn.Parameter(tensor(1e-2))
+        self.dynamic_beta_fn = nn.Parameter(torch.zeros(dim))
+        self.dynamic_beta_scale = nn.Parameter(tensor(1e-2))
+
+    def width_connection(self, residuals):
+        normed = self.norm(residuals)
+
+        wc_weight = (normed @ self.dynamic_alpha_fn).tanh()
+        dynamic_alpha = wc_weight * self.dynamic_alpha_scale
+        alpha = dynamic_alpha + self.static_alpha
+
+        dc_weight = (normed @ self.dynamic_beta_fn).tanh()
+        dynamic_beta = dc_weight * self.dynamic_beta_scale
+        beta = dynamic_beta + self.static_beta
+
+        # width connection
+        mix_h = einsum(alpha, residuals, '... e1 e2, ... e1 d -> ... e2 d')
+
+        branch_input, residuals = mix_h[..., 0, :], mix_h[..., 1:, :]
+
+        return branch_input, residuals, beta
+
+    def depth_connection(
+        self,
+        branch_output,
+        residuals,
+        beta
+    ):
+        return einsum(branch_output, beta, "b n d, b n e -> b n e d") + residuals
+
+# classes
+
+class FeedForward(Module):
+    def __init__(self, dim, hidden_dim):
+        super().__init__()
+        self.net = nn.Sequential(
+            nn.LayerNorm(dim),
+            nn.Linear(dim, hidden_dim),
+            nn.GELU(),
+            nn.Linear(hidden_dim, dim),
+        )
+    def forward(self, x):
+        return self.net(x)
+
+class Attention(Module):
+    def __init__(self, dim, heads = 8, dim_head = 64):
+        super().__init__()
+        inner_dim = dim_head *  heads
+        self.heads = heads
+        self.scale = dim_head ** -0.5
+        self.norm = nn.LayerNorm(dim)
+
+        self.attend = nn.Softmax(dim = -1)
+
+        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias = False)
+        self.to_out = nn.Linear(inner_dim, dim, bias = False)
+
+    def forward(self, x):
+        x = self.norm(x)
+
+        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 = self.heads), qkv)
+
+        dots = torch.matmul(q, k.transpose(-1, -2)) * self.scale
+
+        attn = self.attend(dots)
+
+        out = torch.matmul(attn, v)
+        out = rearrange(out, 'b h n d -> b n (h d)')
+        return self.to_out(out)
+
+class Transformer(Module):
+    def __init__(self, dim, depth, heads, dim_head, mlp_dim, num_residual_streams):
+        super().__init__()
+
+        self.num_residual_streams = num_residual_streams
+
+        self.norm = nn.LayerNorm(dim)
+        self.layers = ModuleList([])
+
+        for layer_index in range(depth):
+            self.layers.append(nn.ModuleList([
+                HyperConnection(dim, num_residual_streams, layer_index),
+                Attention(dim, heads = heads, dim_head = dim_head),
+                HyperConnection(dim, num_residual_streams, layer_index),
+                FeedForward(dim, mlp_dim)
+            ]))
+
+    def forward(self, x):
+
+        x = repeat(x, 'b n d -> b n e d', e = self.num_residual_streams)
+
+        for attn_hyper_conn, attn, ff_hyper_conn, ff in self.layers:
+
+            x, attn_res, beta = attn_hyper_conn.width_connection(x)
+
+            x = attn(x)
+
+            x = attn_hyper_conn.depth_connection(x, attn_res, beta)
+
+            x, ff_res, beta = ff_hyper_conn.width_connection(x)
+
+            x = ff(x)
+
+            x = ff_hyper_conn.depth_connection(x, ff_res, beta)
+
+        x = reduce(x, 'b n e d -> b n d', 'sum')
+
+        return self.norm(x)
+
+class SimpleViT(nn.Module):
+    def __init__(self, *, image_size, patch_size, num_classes, dim, depth, heads, mlp_dim, num_residual_streams, num_register_tokens = 4, channels = 3, dim_head = 64):
+        super().__init__()
+        image_height, image_width = pair(image_size)
+        patch_height, patch_width = pair(patch_size)
+
+        assert image_height % patch_height == 0 and image_width % patch_width == 0, 'Image dimensions must be divisible by the patch size.'
+
+        patch_dim = channels * patch_height * patch_width
+
+        self.to_patch_embedding = nn.Sequential(
+            Rearrange("b c (h p1) (w p2) -> b (h w) (p1 p2 c)", p1 = patch_height, p2 = patch_width),
+            nn.LayerNorm(patch_dim),
+            nn.Linear(patch_dim, dim),
+            nn.LayerNorm(dim),
+        )
+
+        self.register_tokens = nn.Parameter(torch.randn(num_register_tokens, dim))
+
+        self.pos_embedding = posemb_sincos_2d(
+            h = image_height // patch_height,
+            w = image_width // patch_width,
+            dim = dim,
+        ) 
+
+        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, num_residual_streams)
+
+        self.pool = "mean"
+        self.to_latent = nn.Identity()
+
+        self.linear_head = nn.Linear(dim, num_classes)
+
+    def forward(self, img):
+        batch, device = img.shape[0], img.device
+
+        x = self.to_patch_embedding(img)
+        x += self.pos_embedding.to(x)
+
+        r = repeat(self.register_tokens, 'n d -> b n d', b = batch)
+
+        x, ps = pack([x, r], 'b * d')
+
+        x = self.transformer(x)
+
+        x, _ = unpack(x, ps, 'b * d')
+
+        x = x.mean(dim = 1)
+
+        x = self.to_latent(x)
+        return self.linear_head(x)
+
+# main
+
+if __name__ == '__main__':
+    vit = SimpleViT(
+        num_classes = 1000,
+        image_size = 256,
+        patch_size = 8,
+        dim = 1024,
+        depth = 12,
+        heads = 8,
+        mlp_dim = 2048,
+        num_residual_streams = 8
+    )
+
+    images = torch.randn(3, 3, 256, 256)
+
+    logits = vit(images)

vit_pytorch/simple_vit_with_value_residual.py

@@ -0,0 +1,159 @@
+import torch
+from torch import nn
+from torch.nn import Module, ModuleList
+
+from einops import rearrange
+from einops.layers.torch import Rearrange
+
+# helpers
+
+def exists(v):
+    return v is not None
+
+def default(v, d):
+    return v if exists(v) else d
+
+def pair(t):
+    return t if isinstance(t, tuple) else (t, t)
+
+def posemb_sincos_2d(h, w, dim, temperature: int = 10000, dtype = torch.float32):
+    y, x = torch.meshgrid(torch.arange(h), torch.arange(w), indexing="ij")
+    assert (dim % 4) == 0, "feature dimension must be multiple of 4 for sincos emb"
+    omega = torch.arange(dim // 4) / (dim // 4 - 1)
+    omega = 1.0 / (temperature ** omega)
+
+    y = y.flatten()[:, None] * omega[None, :]
+    x = x.flatten()[:, None] * omega[None, :]
+    pe = torch.cat((x.sin(), x.cos(), y.sin(), y.cos()), dim=1)
+    return pe.type(dtype)
+
+# classes
+
+def FeedForward(dim, hidden_dim):
+    return nn.Sequential(
+        nn.LayerNorm(dim),
+        nn.Linear(dim, hidden_dim),
+        nn.GELU(),
+        nn.Linear(hidden_dim, dim),
+    )
+
+class Attention(Module):
+    def __init__(self, dim, heads = 8, dim_head = 64, learned_value_residual_mix = False):
+        super().__init__()
+        inner_dim = dim_head *  heads
+        self.heads = heads
+        self.scale = dim_head ** -0.5
+        self.norm = nn.LayerNorm(dim)
+
+        self.attend = nn.Softmax(dim = -1)
+
+        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias = False)
+        self.to_out = nn.Linear(inner_dim, dim, bias = False)
+
+        self.to_residual_mix = nn.Sequential(
+            nn.Linear(dim, heads),
+            nn.Sigmoid(),
+            Rearrange('b n h -> b h n 1')
+        ) if learned_value_residual_mix else (lambda _: 0.5)
+
+    def forward(self, x, value_residual = None):
+        x = self.norm(x)
+
+        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 = self.heads), qkv)
+
+        if exists(value_residual):
+            mix = self.to_residual_mix(x)
+            v = v * mix + value_residual * (1. - mix)
+
+        dots = torch.matmul(q, k.transpose(-1, -2)) * self.scale
+
+        attn = self.attend(dots)
+
+        out = torch.matmul(attn, v)
+        out = rearrange(out, 'b h n d -> b n (h d)')
+
+        return self.to_out(out), v
+
+class Transformer(Module):
+    def __init__(self, dim, depth, heads, dim_head, mlp_dim):
+        super().__init__()
+        self.norm = nn.LayerNorm(dim)
+        self.layers = ModuleList([])
+        for i in range(depth):
+            is_first = i == 0
+            self.layers.append(ModuleList([
+                Attention(dim, heads = heads, dim_head = dim_head, learned_value_residual_mix = not is_first),
+                FeedForward(dim, mlp_dim)
+            ]))
+    def forward(self, x):
+        value_residual = None
+
+        for attn, ff in self.layers:
+
+            attn_out, values = attn(x, value_residual = value_residual)
+            value_residual = default(value_residual, values)
+
+            x = attn_out + x
+            x = ff(x) + x
+
+        return self.norm(x)
+
+class SimpleViT(Module):
+    def __init__(self, *, image_size, patch_size, num_classes, dim, depth, heads, mlp_dim, channels = 3, dim_head = 64):
+        super().__init__()
+        image_height, image_width = pair(image_size)
+        patch_height, patch_width = pair(patch_size)
+
+        assert image_height % patch_height == 0 and image_width % patch_width == 0, 'Image dimensions must be divisible by the patch size.'
+
+        patch_dim = channels * patch_height * patch_width
+
+        self.to_patch_embedding = nn.Sequential(
+            Rearrange("b c (h p1) (w p2) -> b (h w) (p1 p2 c)", p1 = patch_height, p2 = patch_width),
+            nn.LayerNorm(patch_dim),
+            nn.Linear(patch_dim, dim),
+            nn.LayerNorm(dim),
+        )
+
+        self.pos_embedding = posemb_sincos_2d(
+            h = image_height // patch_height,
+            w = image_width // patch_width,
+            dim = dim,
+        ) 
+
+        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim)
+
+        self.pool = "mean"
+        self.to_latent = nn.Identity()
+
+        self.linear_head = nn.Linear(dim, num_classes)
+
+    def forward(self, img):
+        device = img.device
+
+        x = self.to_patch_embedding(img)
+        x += self.pos_embedding.to(device, dtype=x.dtype)
+
+        x = self.transformer(x)
+        x = x.mean(dim = 1)
+
+        x = self.to_latent(x)
+        return self.linear_head(x)
+
+# quick test
+
+if __name__ == '__main__':
+    v = SimpleViT(
+        num_classes = 1000,
+        image_size = 256,
+        patch_size = 8,
+        dim = 1024,
+        depth = 6,
+        heads = 8,
+        mlp_dim = 2048,
+    )
+
+    images = torch.randn(2, 3, 256, 256)
+
+    logits = v(images)