File size: 6,166 Bytes
ad06aed
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b7c5eaf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ad06aed
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b7c5eaf
ad06aed
b7c5eaf
 
 
ad06aed
 
b7c5eaf
ad06aed
b7c5eaf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ad06aed
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
# Copyright (c) 2023, Zexin He
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.


import torch
import torch.nn as nn


class BasicTransformerBlock(nn.Module):
    """
    Transformer block that takes in a cross-attention condition and another modulation vector applied to sub-blocks.
    """
    # use attention from torch.nn.MultiHeadAttention
    # Block contains a cross-attention layer, a self-attention layer, and a MLP
    def __init__(
        self, 
        inner_dim: int, 
        cond_dim: int, 
        num_heads: int, 
        eps: float,
        attn_drop: float = 0., 
        attn_bias: bool = False,
        mlp_ratio: float = 4., 
        mlp_drop: float = 0.,
    ):
        super().__init__()

        self.norm1 = nn.LayerNorm(inner_dim)
        self.cross_attn = nn.MultiheadAttention(
            embed_dim=inner_dim, num_heads=num_heads, kdim=cond_dim, vdim=cond_dim,
            dropout=attn_drop, bias=attn_bias, batch_first=True)
        self.norm2 = nn.LayerNorm(inner_dim)
        self.self_attn = nn.MultiheadAttention(
            embed_dim=inner_dim, num_heads=num_heads,
            dropout=attn_drop, bias=attn_bias, batch_first=True)
        self.norm3 = nn.LayerNorm(inner_dim)
        self.mlp = nn.Sequential(
            nn.Linear(inner_dim, int(inner_dim * mlp_ratio)),
            nn.GELU(),
            nn.Dropout(mlp_drop),
            nn.Linear(int(inner_dim * mlp_ratio), inner_dim),
            nn.Dropout(mlp_drop),
        )

    def forward(self, x, cond, i, alpha, content_layers):
        # x: [N, L, D] or [x1, x2]
        # cond: [content_feats] or [content_feats, style_feats]
        if len(cond) == 2:
            # Style injection mode
            x1, x2 = x[0], x[1]
            content, style = cond[0], cond[1]
            if i <= content_layers:
                x1 = x1 + self.cross_attn(self.norm1(x1), content, content)[0]
            else:
                x1 = x1 + (1-alpha)*self.cross_attn(self.norm1(x1), content, content)[0] + (alpha)*self.cross_attn(self.norm1(x1), style, style)[0]
            x2 = x2 + self.cross_attn(self.norm1(x2), style, style)[0]

            before_sa1 = self.norm2(x1)
            before_sa2 = self.norm2(x2)
            x1 = x1 + self.self_attn(before_sa1, before_sa1, before_sa1)[0]
            x2 = x2 + self.self_attn(before_sa2, before_sa2, before_sa2)[0]

            x1 = x1 + self.mlp(self.norm3(x1))
            x2 = x2 + self.mlp(self.norm3(x2))

            return [x1, x2]
        else:
            # No style, only content
            x1 = x[0] if isinstance(x, list) else x
            content = cond[0]
            x1 = x1 + self.cross_attn(self.norm1(x1), content, content)[0]
            before_sa1 = self.norm2(x1)
            x1 = x1 + self.self_attn(before_sa1, before_sa1, before_sa1)[0]
            x1 = x1 + self.mlp(self.norm3(x1))
            return [x1]


class TriplaneTransformer(nn.Module):
    """
    Transformer with condition that generates a triplane representation.
    
    Reference:
    Timm: https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/vision_transformer.py#L486
    """
    def __init__(
        self, 
        inner_dim: int, 
        image_feat_dim: int,
        triplane_low_res: int, 
        triplane_high_res: int, 
        triplane_dim: int,
        num_layers: int, 
        num_heads: int,
        eps: float = 1e-6,
    ):
        super().__init__()

        # attributes
        self.triplane_low_res = triplane_low_res
        self.triplane_high_res = triplane_high_res
        self.triplane_dim = triplane_dim

        # modules
        # initialize pos_embed with 1/sqrt(dim) * N(0, 1)
        self.pos_embed = nn.Parameter(torch.randn(1, 3*triplane_low_res**2, inner_dim) * (1. / inner_dim) ** 0.5)
        self.layers = nn.ModuleList([
            BasicTransformerBlock(
                inner_dim=inner_dim, cond_dim=image_feat_dim, num_heads=num_heads, eps=eps)
            for _ in range(num_layers)
        ])
        self.norm = nn.LayerNorm(inner_dim, eps=eps)
        self.deconv = nn.ConvTranspose2d(inner_dim, triplane_dim, kernel_size=2, stride=2, padding=0)
        self.num_layers = num_layers

    def forward(self, image_feats, alpha, style_layers):
        # image_feats: [content_feats] or [content_feats, style_feats]
        N = image_feats[0].shape[0]
        H = W = self.triplane_low_res
        L = 3 * H * W
        content_layers = self.num_layers - style_layers
        x = self.pos_embed.repeat(N, 1, 1)  # [N, L, D]
        i = 1
        if len(image_feats) == 2:
            # Style injection mode
            for layer in self.layers:
                if i == 1:
                    x = layer([x, x], image_feats, i, alpha, content_layers)
                else:
                    x = layer(x, image_feats, i, alpha, content_layers)
                i += 1
            x = self.norm(x[0])
        else:
            # No style, only content
            for layer in self.layers:
                if i == 1:
                    x = layer([x], image_feats, i, alpha, content_layers)
                else:
                    x = layer(x, image_feats, i, alpha, content_layers)
                i += 1
            x = self.norm(x[0])

        # separate each plane and apply deconv
        x = x.view(N, 3, H, W, -1)
        x = torch.einsum('nihwd->indhw', x)  # [3, N, D, H, W]
        x = x.contiguous().view(3*N, -1, H, W)  # [3*N, D, H, W]
        x = self.deconv(x)  # [3*N, D', H', W']
        x = x.view(3, N, *x.shape[-3:])  # [3, N, D', H', W']
        x = torch.einsum('indhw->nidhw', x)  # [N, 3, D', H', W']
        x = x.contiguous()

        return x