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
| import torch
import torch.nn as nn
from collections import OrderedDict
import torch
import numpy as np
import clip
class ResidualAttentionBlock(nn.Module):
def __init__(self, d_model: int, n_head: int, attn_mask: torch.Tensor = None):
super().__init__()
self.attn = nn.MultiheadAttention(d_model, n_head)
self.ln_1 = LayerNorm(d_model)
self.mlp = nn.Sequential(OrderedDict([
("c_fc", nn.Linear(d_model, d_model * 4)),
("gelu", QuickGELU()),
("c_proj", nn.Linear(d_model * 4, d_model))
]))
self.ln_2 = LayerNorm(d_model)
self.attn_mask = attn_mask
def attention(self, x: torch.Tensor):
self.attn_mask = self.attn_mask.to(dtype=x.dtype, device=x.device) if self.attn_mask is not None else None
return self.attn(x, x, x, need_weights=False, attn_mask=self.attn_mask)[0]
def forward(self, x: torch.Tensor):
x = x + self.attention(self.ln_1(x))
x = x + self.mlp(self.ln_2(x))
return x
class Transformer(nn.Module):
def __init__(self, width: int, layers: int, heads: int, attn_mask: torch.Tensor = None):
super().__init__()
self.width = width
self.layers = layers
self.resblocks = nn.Sequential(*[ResidualAttentionBlock(width, heads, attn_mask) for _ in range(layers)])
def forward(self, x: torch.Tensor):
return self.resblocks(x)
class LayerNorm(nn.LayerNorm):
"""Subclass torch's LayerNorm to handle fp16."""
def forward(self, x: torch.Tensor):
orig_type = x.dtype
ret = super().forward(x.type(torch.float32))
return ret.type(orig_type)
class QuickGELU(nn.Module):
def forward(self, x: torch.Tensor):
return x * torch.sigmoid(1.702 * x)
class TextEncoder(nn.Module):
def __init__(self,
embed_dim: int,
context_length: int,
vocab_size: int,
transformer_width: int,
transformer_heads: int,
transformer_layers: int
):
super().__init__()
self.context_length = context_length
self.transformer = Transformer(
width=transformer_width,
layers=transformer_layers,
heads=transformer_heads,
attn_mask=self.build_attention_mask()
)
self.vocab_size = vocab_size
self.token_embedding = nn.Embedding(vocab_size, transformer_width)
self.positional_embedding = nn.Parameter(torch.empty(self.context_length, transformer_width))
self.ln_final = LayerNorm(transformer_width)
self.logit_scale = nn.Parameter(torch.ones([]) * np.log(1 / 0.07))
self.temperature = nn.Parameter(torch.tensor(0.07))
self.text_projection = nn.Parameter(torch.empty(transformer_width, embed_dim))
print(f"text_projection shape: {self.text_projection.shape}")
self.dtype = torch.float32
self.initialize_parameters()
def initialize_parameters(self):
nn.init.normal_(self.token_embedding.weight, std=0.02)
nn.init.normal_(self.positional_embedding, std=0.01)
proj_std = (self.transformer.width ** -0.5) * ((2 * self.transformer.layers) ** -0.5)
attn_std = self.transformer.width ** -0.5
fc_std = (2 * self.transformer.width) ** -0.5
for block in self.transformer.resblocks:
nn.init.normal_(block.attn.in_proj_weight, std=attn_std)
nn.init.normal_(block.attn.out_proj.weight, std=proj_std)
nn.init.normal_(block.mlp.c_fc.weight, std=fc_std)
nn.init.normal_(block.mlp.c_proj.weight, std=proj_std)
if self.text_projection is not None:
nn.init.normal_(self.text_projection, std=self.transformer.width ** -0.5)
else:
nn.init.normal_(self.text_projection, std=self.custom_text_config['text_rep_size'] ** -0.5)
def build_attention_mask(self):
mask = torch.empty(self.context_length, self.context_length)
mask.fill_(float("-inf"))
mask.triu_(1)
return mask
def forward(self, text):
x = self.token_embedding(text).type(self.dtype)
x = x + self.positional_embedding.type(self.dtype)
x = x.permute(1, 0, 2)
x = self.transformer(x)
x = x.permute(1, 0, 2)
x = self.ln_final(x).type(self.dtype)
x = x[torch.arange(x.shape[0]), text.argmax(dim=-1)] @ self.text_projection
return x
if __name__ == '__main__':
import clip
device = "cpu"
model, preprocess = clip.load("ViT-B/32", device=device)
model.eval()
text_encoder = TextEncoder(embed_dim=512, context_length=77, vocab_size=49408,
transformer_width=512, transformer_heads=8, transformer_layers=12)
missing_keys, unexpected_keys = text_encoder.load_state_dict(model.state_dict(), strict=False)
text_encoder.eval()
input_tensor = clip.tokenize("a diagram").to(device)
traced_model = torch.jit.trace(text_encoder, input_tensor)
onnx_filename = 'clip-text-encoder.onnx'
torch.onnx.export(text_encoder, input_tensor, onnx_filename)
|
