begin work on NaViT (lucidrains#273)

finish core idea of NaViT

README.md

@@ -7,6 +7,7 @@
 - [Usage](#usage)
 - [Parameters](#parameters)
 - [Simple ViT](#simple-vit)
+- [NaViT](#na-vit)
 - [Distillation](#distillation)
 - [Deep ViT](#deep-vit)
 - [CaiT](#cait)
@@ -139,6 +140,44 @@ img = torch.randn(1, 3, 256, 256)
 preds = v(img) # (1, 1000)
 ```
 
+## NaViT
+
+<img src="./images/na_vit.png" width="450px"></img>
+
+<a href="https://arxiv.org/abs/2307.06304">This paper</a> proposes to leverage the flexibility of attention and masking for variable lengthed sequences to train images of multiple resolution, packed into a single batch. They demonstrate much faster training and improved accuracies, with the only cost being extra complexity in the architecture and dataloading. They use factorized 2d positional encodings, token dropping, as well as query-key normalization.
+
+You can use it as follows
+
+```python
+import torch
+from vit_pytorch.na_vit import NaViT
+
+v = NaViT(
+    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
+)
+
+# 5 images of different resolutions - List[List[Tensor]]
+
+# for now, you'll have to correctly place images in same batch element as to not exceed maximum allowed sequence length for self-attention w/ masking
+
+images = [
+    [torch.randn(3, 256, 256), torch.randn(3, 128, 128)],
+    [torch.randn(3, 128, 256), torch.randn(3, 256, 128)],
+    [torch.randn(3, 64, 256)]
+]
+
+preds = v(images) # (5, 1000) - 5, because 5 images of different resolution above
+
+```
+
 ## Distillation
 
 <img src="./images/distill.png" width="300px"></img>
@@ -1934,6 +1973,14 @@ Coming from computer vision and new to transformers? Here are some resources tha
 }
 ```
 
+```bibtex
+@inproceedings{Dehghani2023PatchNP,
+    title   = {Patch n' Pack: NaViT, a Vision Transformer for any Aspect Ratio and Resolution},
+    author  = {Mostafa Dehghani and Basil Mustafa and Josip Djolonga and Jonathan Heek and Matthias Minderer and Mathilde Caron and Andreas Steiner and Joan Puigcerver and Robert Geirhos and Ibrahim M. Alabdulmohsin and Avital Oliver and Piotr Padlewski and Alexey A. Gritsenko and Mario Luvci'c and Neil Houlsby},
+    year    = {2023}
+}
+```
+
 ```bibtex
 @misc{vaswani2017attention,
     title   = {Attention Is All You Need},

vit_pytorch/na_vit.py

@@ -0,0 +1,305 @@
+from functools import partial
+from typing import List
+
+import torch
+import torch.nn.functional as F
+from torch import nn, Tensor
+from torch.nn.utils.rnn import pad_sequence as orig_pad_sequence
+
+from einops import rearrange, repeat
+from einops.layers.torch import Rearrange
+
+# helpers
+
+def exists(val):
+    return val is not None
+
+def default(val, d):
+    return val if exists(val) else d
+
+def pair(t):
+    return t if isinstance(t, tuple) else (t, t)
+
+def divisible_by(numer, denom):
+    return (numer % denom) == 0
+
+# normalization
+# they use layernorm without bias, something that pytorch does not offer
+
+class LayerNorm(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+        self.gamma = nn.Parameter(torch.ones(dim))
+        self.register_buffer('beta', torch.zeros(dim))
+
+    def forward(self, x):
+        return F.layer_norm(x, x.shape[-1:], self.gamma, self.beta)
+
+# they use a query-key normalization that is equivalent to rms norm (no mean-centering, learned gamma), from vit 22B paper
+
+class RMSNorm(nn.Module):
+    def __init__(self, heads, dim):
+        super().__init__()
+        self.scale = dim ** 0.5
+        self.gamma = nn.Parameter(torch.ones(heads, 1, dim))
+
+    def forward(self, x):
+        normed = F.normalize(x, dim = -1)
+        return normed * self.scale * self.gamma
+
+# feedforward
+
+def FeedForward(dim, hidden_dim, dropout = 0.):
+    return nn.Sequential(
+        LayerNorm(dim),
+        nn.Linear(dim, hidden_dim),
+        nn.GELU(),
+        nn.Dropout(dropout),
+        nn.Linear(hidden_dim, dim),
+        nn.Dropout(dropout)
+    )
+
+class Attention(nn.Module):
+    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
+        super().__init__()
+        inner_dim = dim_head *  heads
+        self.heads = heads
+        self.norm = LayerNorm(dim)
+
+        self.q_norm = RMSNorm(heads, dim_head)
+        self.k_norm = RMSNorm(heads, dim_head)
+
+        self.attend = nn.Softmax(dim = -1)
+        self.dropout = nn.Dropout(dropout)
+
+        self.to_q = nn.Linear(dim, inner_dim, bias = False)
+        self.to_kv = nn.Linear(dim, inner_dim * 2, bias = False)
+
+        self.to_out = nn.Sequential(
+            nn.Linear(inner_dim, dim, bias = False),
+            nn.Dropout(dropout)
+        )
+
+    def forward(
+        self,
+        x,
+        context = None,
+        mask = None,
+        attn_mask = None
+    ):
+        x = self.norm(x)
+        kv_input = default(context, x)
+
+        qkv = (self.to_q(x), *self.to_kv(kv_input).chunk(2, dim = -1))
+
+        q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b h n d', h = self.heads), qkv)
+
+        q = self.q_norm(q)
+        k = self.k_norm(k)
+
+        dots = torch.matmul(q, k.transpose(-1, -2))
+
+        if exists(mask):
+            mask = rearrange(mask, 'b j -> b 1 1 j')
+            dots = dots.masked_fill(~mask, -torch.finfo(dots.dtype).max)
+
+        if exists(attn_mask):
+            dots = dots.masked_fill(~attn_mask, -torch.finfo(dots.dtype).max)
+
+        attn = self.attend(dots)
+        attn = self.dropout(attn)
+
+        out = torch.matmul(attn, v)
+        out = rearrange(out, 'b h n d -> b n (h d)')
+        return self.to_out(out)
+
+class Transformer(nn.Module):
+    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0.):
+        super().__init__()
+        self.layers = nn.ModuleList([])
+        for _ in range(depth):
+            self.layers.append(nn.ModuleList([
+                Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout),
+                FeedForward(dim, mlp_dim, dropout = dropout)
+            ]))
+
+        self.norm = LayerNorm(dim)
+
+    def forward(
+        self,
+        x,
+        mask = None,
+        attn_mask = None
+    ):
+        for attn, ff in self.layers:
+            x = attn(x, mask = mask, attn_mask = attn_mask) + x
+            x = ff(x) + x
+
+        return self.norm(x)
+
+class NaViT(nn.Module):
+    def __init__(self, *, image_size, patch_size, num_classes, dim, depth, heads, mlp_dim, channels = 3, dim_head = 64, dropout = 0., emb_dropout = 0.):
+        super().__init__()
+        image_height, image_width = pair(image_size)
+
+        assert divisible_by(image_height, patch_size) and divisible_by(image_width, patch_size), 'Image dimensions must be divisible by the patch size.'
+
+        patch_height_dim, patch_width_dim = (image_height // patch_size), (image_width // patch_size)
+        patch_dim = channels * (patch_size ** 2)
+
+        self.channels = channels
+        self.patch_size = patch_size
+
+        self.to_patch_embedding = nn.Sequential(
+            LayerNorm(patch_dim),
+            nn.Linear(patch_dim, dim),
+            LayerNorm(dim),
+        )
+
+        self.pos_embed_height = nn.Parameter(torch.randn(patch_height_dim, dim))
+        self.pos_embed_width = nn.Parameter(torch.randn(patch_width_dim, dim))
+
+        self.dropout = nn.Dropout(emb_dropout)
+
+        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout)
+
+        # final attention pooling queries
+
+        self.attn_pool_queries = nn.Parameter(torch.randn(dim))
+        self.attn_pool = Attention(dim = dim, dim_head = dim_head, heads = heads)
+
+        # output to logits
+
+        self.to_latent = nn.Identity()
+
+        self.mlp_head = nn.Sequential(
+            LayerNorm(dim),
+            nn.Linear(dim, num_classes, bias = False)
+        )
+
+    @property
+    def device(self):
+        return next(self.parameters()).device
+
+    def forward(
+        self,
+        batched_images: List[List[Tensor]] # assume different resolution images already grouped correctly
+    ):
+        p, c, device = self.patch_size, self.channels, self.device
+
+        arange = partial(torch.arange, device = device)
+        pad_sequence = partial(orig_pad_sequence, batch_first = True)
+
+        # process images into variable lengthed sequences with attention mask
+
+        num_images = []
+        batched_sequences = []
+        batched_positions = []
+        batched_image_ids = []
+
+        for images in batched_images:
+            num_images.append(len(images))
+
+            sequences = []
+            positions = []
+            image_ids = torch.empty((0,), device = device, dtype = torch.long)
+
+            for image_id, image in enumerate(images):
+                assert image.ndim ==3 and image.shape[0] == c
+                image_dims = image.shape[-2:]
+                assert all([divisible_by(dim, p) for dim in image_dims]), f'height and width {image_dims} of images must be divisible by patch size {p}'
+
+                ph, pw = map(lambda dim: dim // p, image_dims)
+
+                pos = torch.stack(torch.meshgrid((
+                    arange(ph),
+                    arange(pw)
+                ), indexing = 'ij'), dim = -1)
+
+                pos = rearrange(pos, 'h w c -> (h w) c')
+                seq = rearrange(image, 'c (h p1) (w p2) -> (h w) (c p1 p2)', p1 = p, p2 = p)
+
+                image_ids = F.pad(image_ids, (0, seq.shape[-2]), value = image_id)
+                sequences.append(seq)
+                positions.append(pos)
+
+            batched_image_ids.append(image_ids)
+            batched_sequences.append(torch.cat(sequences, dim = 0))
+            batched_positions.append(torch.cat(positions, dim = 0))
+
+        # derive key padding mask
+
+        lengths = torch.tensor([seq.shape[-2] for seq in batched_sequences], device = device, dtype = torch.long)
+        max_length = arange(lengths.amax().item())
+        key_pad_mask = rearrange(lengths, 'b -> b 1') <= rearrange(max_length, 'n -> 1 n')
+
+        # derive attention mask, and combine with key padding mask from above
+
+        batched_image_ids = pad_sequence(batched_image_ids)
+        attn_mask = rearrange(batched_image_ids, 'b i -> b 1 i 1') == rearrange(batched_image_ids, 'b j -> b 1 1 j')
+        attn_mask = attn_mask & rearrange(key_pad_mask, 'b j -> b 1 1 j')
+
+        # combine patched images as well as the patched width / height positions for 2d positional embedding
+
+        patches = pad_sequence(batched_sequences)
+        patch_positions = pad_sequence(batched_positions)
+
+        # need to know how many images for final attention pooling
+
+        num_images = torch.tensor(num_images, device = device, dtype = torch.long)        
+
+        # to patches
+
+        x = self.to_patch_embedding(patches)        
+
+        # factorized 2d absolute positional embedding
+
+        h_indices, w_indices = patch_positions.unbind(dim = -1)
+
+        h_pos = self.pos_embed_height[h_indices]
+        w_pos = self.pos_embed_width[w_indices]
+
+        x = x + h_pos + w_pos
+
+        # embed dropout
+
+        x = self.dropout(x)
+
+        # attention
+
+        x = self.transformer(x, attn_mask = attn_mask)
+
+        # do attention pooling at the end
+
+        max_queries = num_images.amax().item()
+
+        queries = repeat(self.attn_pool_queries, 'd -> b n d', n = max_queries, b = x.shape[0])
+
+        # attention pool mask
+
+        image_id_arange = arange(max_queries)
+
+        attn_pool_mask = rearrange(image_id_arange, 'i -> i 1') == rearrange(batched_image_ids, 'b j -> b 1 j')
+
+        attn_pool_mask = attn_pool_mask & rearrange(key_pad_mask, 'b j -> b 1 j')
+
+        attn_pool_mask = rearrange(attn_pool_mask, 'b i j -> b 1 i j')
+
+        # attention pool
+
+        x = self.attn_pool(queries, context = x, attn_mask = attn_pool_mask) + queries
+
+        x = rearrange(x, 'b n d -> (b n) d')
+
+        # each batch element may not have same amount of images
+
+        is_images = image_id_arange < rearrange(num_images, 'b -> b 1')
+        is_images = rearrange(is_images, 'b n -> (b n)')
+
+        x = x[is_images]
+
+        # project out to logits
+
+        x = self.to_latent(x)
+
+        return self.mlp_head(x)