File size: 8,106 Bytes
5d2263b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
import torch
from torch import nn, einsum
import torch.nn.functional as F
from einops import rearrange, repeat

from rotary_embedding_torch import apply_rotary_emb
from celle.utils import exists, default, max_neg_value


# helpers
def stable_softmax(t, dim=-1, alpha=32**2):
    t = t / alpha
    t = t - torch.amax(t, dim=dim, keepdim=True).detach()
    return (t * alpha).softmax(dim=dim)


def apply_pos_emb(pos_emb, qkv):
    n = qkv[0].shape[-2]
    pos_emb = pos_emb[..., :n, :]
    return tuple(map(lambda t: apply_rotary_emb(pos_emb, t), qkv))


# classes
class Attention(nn.Module):
    def __init__(
        self,
        dim,
        seq_len,
        causal=False,
        heads=8,
        dim_head=64,
        dropout=0.0,
        stable=False,
        static_mask=None,
    ):
        super().__init__()
        inner_dim = dim_head * heads
        self.heads = heads
        self.seq_len = seq_len
        self.scale = dim_head**-0.5
        self.stable = stable
        self.causal = causal
        self.register_buffer("static_mask", static_mask, persistent=False)
        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias=False)
        self.to_out = nn.Sequential(nn.Linear(inner_dim, dim), nn.Dropout(dropout))
        self.save_attn = nn.Identity()

    def forward(self, x, context_mask=None, rotary_pos_emb=None):
        # x: [batch_size, seq_len, dim]
        b, n, _, h = *x.shape, self.heads
        device = x.device

        softmax = torch.softmax if not self.stable else stable_softmax

        # qkv: 3 tensors of shape [batch_size, seq_len, inner_dim]
        qkv = self.to_qkv(x).chunk(3, dim=-1)

        # q,k,v: [batch_size, heads, seq_len, dim_head]
        q, k, v = map(lambda t: rearrange(t, "b n (h d) -> b h n d", h=h), qkv)

        if exists(rotary_pos_emb):
            q, k, v = apply_pos_emb(rotary_pos_emb[..., :, :], (q, k, v))

        q *= self.scale

        # dots: [batch_size, heads, seq_len_i ,seq_len_j]
        dots = torch.einsum("b h i d, b h j d -> b h i j", q, k)
        mask_value = max_neg_value(dots)

        if exists(context_mask):
            # context_mask: [batch_size ,1 ,1 ,seq_len_j]
            context_mask = rearrange(context_mask, "b j -> b 1 1 j")
            context_mask = F.pad(context_mask, (1, 0), value=True)

            mask_value = -torch.finfo(dots.dtype).max
            dots = dots.masked_fill(~context_mask, mask_value)

        if self.causal:
            i, j = dots.shape[-2:]
            context_mask = torch.ones(i, j, device=device).triu_(j - i + 1).bool()
            dots.masked_fill_(context_mask, mask_value)

        if exists(self.static_mask):
            dots.masked_fill_(~self.static_mask[:n, :n], mask_value)

        # attn: [batch_size ,heads ,seq_len_i ,seq_len_j]
        attn = softmax(dots, dim=-1)
        attn = self.save_attn(attn)

        # out: [batch_size ,heads ,seq_len_i ,dim_head]
        out = torch.einsum("b h n j, b h j d -> b h n d", attn, v)

        # out: [batch_size ,seq_len_i ,(heads*dim_head)]
        out = rearrange(out, "b h n d -> b n (h d)")

        # out: [batch_size ,seq_len_i ,dim]
        out = self.to_out(out)

        return out


# sparse attention with convolutional pattern, as mentioned in the blog post. customizable kernel size and dilation


class SparseConvCausalAttention(nn.Module):
    def __init__(
        self,
        dim,
        seq_len,
        image_size=32,
        kernel_size=5,
        dilation=1,
        heads=8,
        dim_head=64,
        dropout=0.0,
        stable=False,
        **kwargs,
    ):
        super().__init__()
        assert kernel_size % 2 == 1, "kernel size must be odd"

        inner_dim = dim_head * heads
        self.seq_len = seq_len
        self.heads = heads
        self.scale = dim_head**-0.5
        self.image_size = image_size
        self.kernel_size = kernel_size
        self.dilation = dilation

        self.stable = stable

        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias=False)

        self.to_out = nn.Sequential(nn.Linear(inner_dim, dim), nn.Dropout(dropout))

    def forward(self, x, mask=None, rotary_pos_emb=None):
        b, n, _, h, img_size, kernel_size, dilation, seq_len, device = (
            *x.shape,
            self.heads,
            self.image_size,
            self.kernel_size,
            self.dilation,
            self.seq_len,
            x.device,
        )
        softmax = torch.softmax if not self.stable else stable_softmax

        img_seq_len = img_size**2
        text_len = seq_len + 1 - img_seq_len

        # padding

        padding = seq_len - n + 1
        mask = default(mask, lambda: torch.ones(b, text_len, device=device).bool())

        x = F.pad(x, (0, 0, 0, padding), value=0)
        mask = mask[:, :text_len]

        # derive query / keys / values

        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)

        if exists(rotary_pos_emb):
            q, k, v = apply_pos_emb(rotary_pos_emb, (q, k, v))

        q *= self.scale

        ((q_text, q_img), (k_text, k_img), (v_text, v_img)) = map(
            lambda t: (t[:, :-img_seq_len], t[:, -img_seq_len:]), (q, k, v)
        )

        # text attention

        dots_text = einsum("b i d, b j d -> b i j", q_text, k_text)
        mask_value = max_neg_value(dots_text)

        i, j = dots_text.shape[-2:]
        text_causal_mask = torch.ones(i, j, device=device).triu_(j - i + 1).bool()
        dots_text.masked_fill_(text_causal_mask, mask_value)

        attn_text = softmax(dots_text, dim=-1)
        out_text = einsum("b i j, b j d -> b i d", attn_text, v_text)

        # image attention

        effective_kernel_size = (kernel_size - 1) * dilation + 1
        padding = effective_kernel_size // 2

        k_img, v_img = map(
            lambda t: rearrange(t, "b (h w) c -> b c h w", h=img_size), (k_img, v_img)
        )
        k_img, v_img = map(
            lambda t: F.unfold(t, kernel_size, padding=padding, dilation=dilation),
            (k_img, v_img),
        )
        k_img, v_img = map(
            lambda t: rearrange(t, "b (d j) i -> b i j d", j=kernel_size**2),
            (k_img, v_img),
        )

        # let image attend to all of text

        dots_image = einsum("b i d, b i j d -> b i j", q_img, k_img)
        dots_image_to_text = einsum("b i d, b j d -> b i j", q_img, k_text)

        # calculate causal attention for local convolution

        i, j = dots_image.shape[-2:]
        img_seq = torch.arange(img_seq_len, device=device)
        k_img_indices = rearrange(img_seq.float(), "(h w) -> () () h w", h=img_size)
        k_img_indices = F.pad(
            k_img_indices, (padding,) * 4, value=img_seq_len
        )  # padding set to be max, so it is never attended to
        k_img_indices = F.unfold(k_img_indices, kernel_size, dilation=dilation)
        k_img_indices = rearrange(k_img_indices, "b j i -> b i j")

        # mask image attention

        q_img_indices = rearrange(img_seq, "i -> () i ()")
        causal_mask = q_img_indices < k_img_indices

        # concat text mask with image causal mask

        causal_mask = repeat(causal_mask, "() i j -> b i j", b=b * h)
        mask = repeat(mask, "b j -> (b h) i j", i=i, h=h)
        mask = torch.cat((~mask, causal_mask), dim=-1)

        # image can attend to all of text

        dots = torch.cat((dots_image_to_text, dots_image), dim=-1)
        dots.masked_fill_(mask, mask_value)

        attn = softmax(dots, dim=-1)

        # aggregate

        attn_image_to_text, attn_image = attn[..., :text_len], attn[..., text_len:]

        out_image_to_image = einsum("b i j, b i j d -> b i d", attn_image, v_img)
        out_image_to_text = einsum("b i j, b j d -> b i d", attn_image_to_text, v_text)

        out_image = out_image_to_image + out_image_to_text

        # combine attended values for both text and image

        out = torch.cat((out_text, out_image), dim=1)

        out = rearrange(out, "(b h) n d -> b n (h d)", h=h)

        out = self.to_out(out)

        return out[:, :n]