-
Notifications
You must be signed in to change notification settings - Fork 392
/
Copy pathutils.py
200 lines (175 loc) · 9.72 KB
/
utils.py
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
# !/usr/bin/env python3
"""
==== No Bugs in code, just some Random Unexpected FEATURES ====
┌─────────────────────────────────────────────────────────────┐
│┌───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┐│
││Esc│!1 │@2 │#3 │$4 │%5 │^6 │&7 │*8 │(9 │)0 │_- │+= │|\ │`~ ││
│├───┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴───┤│
││ Tab │ Q │ W │ E │ R │ T │ Y │ U │ I │ O │ P │{[ │}] │ BS ││
│├─────┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴─────┤│
││ Ctrl │ A │ S │ D │ F │ G │ H │ J │ K │ L │: ;│" '│ Enter ││
│├──────┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴────┬───┤│
││ Shift │ Z │ X │ C │ V │ B │ N │ M │< ,│> .│? /│Shift │Fn ││
│└─────┬──┴┬──┴──┬┴───┴───┴───┴───┴───┴──┬┴───┴┬──┴┬─────┴───┘│
│ │Fn │ Alt │ Space │ Alt │Win│ HHKB │
│ └───┴─────┴───────────────────────┴─────┴───┘ │
└─────────────────────────────────────────────────────────────┘
工具类，包含prompt自定义字段的填值。
Author: pankeyu
Date: 2022/11/28
"""
import json
import traceback
from typing import List
import torch
import numpy as np
from Template import HardTemplate
def convert_example(
examples: dict,
tokenizer,
max_seq_len: int,
max_label_len: int,
template: HardTemplate,
train_mode=True,
return_tensor=False
) -> dict:
"""
将样本数据转换为模型接收的输入数据。
Args:
examples (dict): 训练数据样本, e.g. -> {
"text": [
'手机 这个手机也太卡了。',
'体育 世界杯为何迟迟不见宣传',
...
]
}
max_seq_len (int): 句子的最大长度，若没有达到最大长度，则padding为最大长度
max_label_len (int): 最大label长度，若没有达到最大长度，则padding为最大长度
template (HardTemplate): 模板类。
train_mode (bool): 训练阶段 or 推理阶段。
return_tensor (bool): 是否返回tensor类型，如不是，则返回numpy类型。
Returns:
dict (str: np.array) -> tokenized_output = {
'input_ids': [[1, 47, 10, 7, 304, 3, 3, 3, 3, 47, 27, 247, 98, 105, 512, 777, 15, 12043, 2], ...],
'token_type_ids': [[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], ...],
'attention_mask': [[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], ...],
'mask_positions': [[5, 6, 7, 8], ...],
'mask_labels': [[2372, 3442, 0, 0], [2643, 4434, 2334, 0], ...]
}
"""
tokenized_output = {
'input_ids': [],
'token_type_ids': [],
'attention_mask': [],
'mask_positions': [],
'mask_labels': []
}
for i, example in enumerate(examples['text']):
try:
if train_mode:
label, content = example.strip().split('\t')
else:
content = example.strip()
inputs_dict={
'textA': content,
'MASK': '[MASK]'
}
encoded_inputs = template(
inputs_dict=inputs_dict,
tokenizer=tokenizer,
max_seq_len=max_seq_len,
mask_length=max_label_len
)
except:
print(f'Error Line {i+1}: "{example}" -> {traceback.format_exc()}')
exit()
tokenized_output['input_ids'].append(encoded_inputs["input_ids"])
tokenized_output['token_type_ids'].append(encoded_inputs["token_type_ids"])
tokenized_output['attention_mask'].append(encoded_inputs["attention_mask"])
tokenized_output['mask_positions'].append(encoded_inputs["mask_position"])
if train_mode:
label_encoded = tokenizer(text=[label]) # 将label补到最大长度
label_encoded = label_encoded['input_ids'][0][1:-1]
label_encoded = label_encoded[:max_label_len]
label_encoded = label_encoded + [tokenizer.pad_token_id] * (max_label_len - len(label_encoded))
tokenized_output['mask_labels'].append(label_encoded)
for k, v in tokenized_output.items():
if return_tensor:
tokenized_output[k] = torch.LongTensor(v)
else:
tokenized_output[k] = np.array(v)
return tokenized_output
def mlm_loss(
logits: torch.tensor,
mask_positions: torch.tensor,
sub_mask_labels: list,
cross_entropy_criterion: torch.nn.CrossEntropyLoss,
masked_lm_scale=1.0,
device='cpu'
) -> torch.tensor:
"""
计算指定位置的mask token的output与label之间的cross entropy loss。
Args:
logits (torch.tensor): 模型原始输出 -> (batch, seq_len, vocab_size)
mask_positions (torch.tensor): mask token的位置 -> (batch, mask_label_num)
sub_mask_labels (list): mask token的sub label, 由于每个label的sub_label数目不同，所以这里是个变长的list,
e.g. -> [
[[2398, 3352]],
[[2398, 3352], [3819, 3861]]
]
cross_entropy_criterion (CrossEntropyLoss): CE Loss计算器
masked_lm_scale (float): scale 参数
device (str): cpu还是gpu
Returns:
torch.tensor: CE Loss
"""
batch_size, seq_len, vocab_size = logits.size()
loss = None
for single_logits, single_sub_mask_labels, single_mask_positions in zip(logits, sub_mask_labels, mask_positions):
single_mask_logits = single_logits[single_mask_positions] # (mask_label_num, vocab_size)
single_mask_logits = single_mask_logits.repeat(len(single_sub_mask_labels), 1, 1) # (sub_label_num, mask_label_num, vocab_size)
single_mask_logits = single_mask_logits.reshape(-1, vocab_size) # (sub_label_num * mask_label_num, vocab_size)
single_sub_mask_labels = torch.LongTensor(single_sub_mask_labels).to(device) # (sub_label_num, mask_label_num)
single_sub_mask_labels = single_sub_mask_labels.reshape(-1, 1).squeeze() # (sub_label_num * mask_label_num)
if not single_sub_mask_labels.size(): # 处理单token维度下维度缺失的问题
single_sub_mask_labels = single_sub_mask_labels.unsqueeze(dim=0)
cur_loss = cross_entropy_criterion(single_mask_logits, single_sub_mask_labels)
cur_loss = cur_loss / len(single_sub_mask_labels)
if not loss:
loss = cur_loss
else:
loss += cur_loss
loss = loss / batch_size # (1,)
return loss / masked_lm_scale
def convert_logits_to_ids(
logits: torch.tensor,
mask_positions: torch.tensor
) -> torch.LongTensor:
"""
输入Language Model的词表概率分布（LMModel的logits），将mask_position位置的
token logits转换为token的id。
Args:
logits (torch.tensor): model output -> (batch, seq_len, vocab_size)
mask_positions (torch.tensor): mask token的位置 -> (batch, mask_label_num)
Returns:
torch.LongTensor: 对应mask position上最大概率的推理token -> (batch, mask_label_num)
"""
label_length = mask_positions.size()[1] # 标签长度
batch_size, seq_len, vocab_size = logits.size()
mask_positions_after_reshaped = []
for batch, mask_pos in enumerate(mask_positions.detach().cpu().numpy().tolist()):
for pos in mask_pos:
mask_positions_after_reshaped.append(batch * seq_len + pos)
logits = logits.reshape(batch_size * seq_len, -1) # (batch_size * seq_len, vocab_size)
mask_logits = logits[mask_positions_after_reshaped] # (batch * label_num, vocab_size)
predict_tokens = mask_logits.argmax(dim=-1) # (batch * label_num)
predict_tokens = predict_tokens.reshape(-1, label_length) # (batch, label_num)
return predict_tokens
if __name__ == '__main__':
from rich import print
logits = torch.randn(1, 20, 21193)
mask_positions = torch.LongTensor([
[3, 4]
])
predict_tokens = convert_logits_to_ids(logits, mask_positions)
print(predict_tokens)