forked from shouxieai/tensorRT_Pro
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathapp_bert.cpp
More file actions
271 lines (228 loc) · 7.28 KB
/
Copy pathapp_bert.cpp
File metadata and controls
271 lines (228 loc) · 7.28 KB
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
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
#include <builder/trt_builder.hpp>
#include <infer/trt_infer.hpp>
#include <common/ilogger.hpp>
#include <unordered_map>
#include <string>
#include <vector>
using namespace std;
#ifdef U_OS_WINDOWS
const char* class_label[] = {
"finance", // finance
"realty", // realty
"stocks", // stocks
"education", // education
"science", // science
"society", // society
"politics", // politics
"sports", // sports
"game", // game
"entertainment" // entertainment
};
#else
/* 如果在windows下编译报错，请屏蔽掉这个文件 */
const char* class_label[] = {
"金融", // finance
"房地产", // realty
"股票", // stocks
"教育", // education
"科学", // science
"社会", // society
"政治", // politics
"体育", // sports
"游戏", // game
"娱乐" // entertainment
};
#endif
bool requires(const char* name);
unordered_map<string, int> load_vocab(const string& file){
unordered_map<string, int> vocab;
auto lines = iLogger::split_string(iLogger::load_text_file(file), "\n");
for(int i = 0; i < lines.size(); ++i){
auto token = lines[i];
vocab[token] = i;
}
return vocab;
}
int find_token(const string& token, const unordered_map<string, int>& vocab){
auto iter = vocab.find(token);
if(iter == vocab.end())
return -1;
return iter->second;
}
/* utf-8
拆分utf8的汉字，把汉字部分独立，ascii部分连续为一个
for example:
你jok我good呀 -> ["你", "job", "我", "good", "呀"] */
vector<string> split_chinese(const string& text){
// 1字节：0xxxxxxx
// 2字节：110xxxxx 10xxxxxx
// 3字节：1110xxxx 10xxxxxx 10xxxxxx
// 4字节：11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
// 5字节：111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
// 6字节：11111110 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
unsigned char* up = (unsigned char*)text.c_str();
int offset = 0;
int length = text.size();
unsigned char lab_char[] = {
// 11111110 11111000 11110000 11100000 11000000 01111111
0xFE, 0xF8, 0xF0, 0xE0, 0xC0, 0x80
};
int char_size_table[] = {
6, 5, 4, 3, 2, 0
};
vector<string> tokens;
string ascii;
int state = 0; // 0 none, 1 wait ascii
while(offset < length){
unsigned char token = up[offset];
int char_size = 1;
for(int i = 0; i < 6; ++i){
if(token >= lab_char[i]){
char_size = char_size_table[i];
break;
}
}
if(char_size == 0){
// invalid char
offset++;
continue;
}
if(offset + char_size > length)
break;
auto char_string = text.substr(offset, char_size);
if(char_size == 1){
// ascii
if(state == 0){
ascii = char_string;
state = 1;
}else if(state == 1){
ascii += char_string;
}
}else{
if(state == 1){
tokens.emplace_back(ascii);
state = 0;
}
tokens.emplace_back(char_string);
}
offset += char_size;
}
if(state == 1)
tokens.emplace_back(ascii);
return tokens;
}
/* 把字符串拆分为词组，汉字单个为一组 */
vector<string> tokenize(const string& text, const unordered_map<string, int>& vocab){
vector<string> tokens = split_chinese(text);
vector<string> output;
int max_input_chars_per_word = 100;
auto UNK = "[UNK]";
for(int itoken = 0; itoken < tokens.size(); ++itoken){
auto& chars = tokens[itoken];
if(chars.size() > max_input_chars_per_word){
output.push_back(UNK);
continue;
}
bool is_bad = false;
int start = 0;
vector<string> sub_tokens;
while(start < chars.size()){
int end = chars.size();
string cur_substr;
while(start < end){
auto substr = chars.substr(start, end-start);
for(int k = 0; k < substr.size(); ++k){
auto& c = substr[k];
if(c >= 'A' && c <= 'Z')
c = c - 'A' + 'a';
}
if(start > 0)
substr = "##" + substr;
auto token_id = find_token(substr, vocab);
if(token_id != -1){
cur_substr = substr;
break;
}
end -= 1;
}
if(cur_substr.empty()){
is_bad = true;
break;
}
sub_tokens.push_back(cur_substr);
start = end;
}
if(is_bad){
output.push_back(UNK);
}else{
output.insert(output.end(), sub_tokens.begin(), sub_tokens.end());
}
}
return output;
}
vector<int> tokens_to_ids(const vector<string>& tokens, const unordered_map<string, int>& vocab){
vector<int> output(tokens.size());
for(int i =0 ; i < tokens.size(); ++i)
output[i] = find_token(tokens[i], vocab);
return output;
}
tuple<vector<int>, vector<int>> align_and_pad(
const vector<string>& tokens, int pad_size,
const unordered_map<string, int>& vocab
){
auto CLS = find_token("[CLS]", vocab);
vector<int> output = tokens_to_ids(tokens, vocab);
vector<int> mask(pad_size, 1);
output.insert(output.begin(), CLS);
int old_size = output.size();
output.resize(pad_size);
if(old_size < pad_size){
std::fill(output.begin() + old_size, output.end(), 0);
std::fill(mask.begin() + old_size, mask.end(), 0);
}
return make_tuple(output, mask);
}
tuple<vector<int>, vector<int>> make_text_data(const string& text, const unordered_map<string, int>& vocab){
auto tokens = tokenize(text, vocab);
return align_and_pad(tokens, 32, vocab);
}
void softmax(float* ptr, int num){
float sum = 0;
for(int i = 0; i < num; ++i)
sum += exp(ptr[i]);
for(int i = 0; i < num; ++i)
ptr[i] = exp(ptr[i]) / sum;
}
int app_bert(){
auto name = "bert";
if(not requires(name))
return 0;
auto onnx_file = iLogger::format("%s.onnx", name);
auto model_file = iLogger::format("%s.trtmodel", name);
auto vocab = load_vocab("vocab.txt");
if(not iLogger::exists(model_file)){
TRT::compile(
TRT::Mode::FP32, 1,
onnx_file, model_file
);
}
auto engine = TRT::load_infer(model_file);
engine->print();
string line;
while(true){
printf("Input content: ");
if(getline(cin, line)){
vector<int> tokens, mask;
tie(tokens, mask) = make_text_data(line, vocab);
memcpy(engine->input(0)->cpu<int>(), tokens.data(), sizeof(int) * tokens.size());
memcpy(engine->input(1)->cpu<int>(), mask.data(), sizeof(int) * mask.size());
engine->forward();
auto ptr = engine->output()->cpu<float>();
int num_classes = engine->output()->size(1);
softmax(ptr, num_classes);
int label = std::max_element(ptr, ptr + num_classes) - ptr;
INFO("Predict: %s, %.3f", class_label[label], ptr[label]);
}
}
return 0;
}