/
data_processer.py
214 lines (189 loc) · 8.44 KB
/
data_processer.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
201
202
203
204
205
206
207
208
209
210
211
212
213
214
# @Time : 2023/3/25 18:36
# @Author : tk
import copy
from enum import Enum
import numpy as np
from transformers import PreTrainedTokenizer
DEFAULT_PAD_TOKEN = "[PAD]"
DEFAULT_EOS_TOKEN = "</s>"
DEFAULT_BOS_TOKEN = "</s>"
DEFAULT_UNK_TOKEN = "</s>"
class DataStrategy(Enum):
tunction = 1
slidding = 2
class PromptBuilder:
@classmethod
def build_template_xverse(cls, query, answer=None, prefix=None, history=None):
prompt = prefix or ''
if history is not None:
for q, a in history:
prompt += "Human: {}\n\nAssistant: {}".format(q, a)
prompt += "Human: {}\n\nAssistant: ".format(query)
if answer is not None:
prompt += answer
return prompt
@classmethod
def build_template_bluelm(cls, query, answer=None, history=None, prefix=None):
prompt = prefix or ''
if history is not None:
for q, a in history:
prompt += "[|Human|]:{}[|AI|]:{}".format(q, a)
prompt += "[|Human|]:{}[|AI|]:".format(query)
if answer is not None:
prompt += answer
return prompt
@classmethod
def build_template_default(cls,query, answer=None, prefix=None, history=None):
prompt = prefix or ''
if history is not None:
for q, a in history:
prompt += "User: {}\nAssistant:{}".format(q, a)
prompt += "User: {}\nAssistant:".format(query)
if answer is not None:
prompt += answer
return prompt
@classmethod
def build_template_tiger(cls,query, answer=None, prefix=None, history=None):
prompt = prefix or ''
tok_ins = "\n\n### Instruction:\n"
tok_res = "\n\n### Response:\n"
if history is not None:
for q, a in history:
prompt += "{}{}{}{}".format(tok_ins, q, tok_res, a)
prompt += "{}{}{}".format(tok_ins, query, tok_res)
if answer is not None:
prompt += answer
return prompt
@classmethod
def build_template_openai(cls,query, answer=None, history=None, prefix=None):
prompt = prefix or 'You are ChatGPT, a large language model trained by OpenAI. Answer as concisely as possible.\nKnowledge cutoff: 2021-09-01\nCurrent date: 2023-03-01'
if prompt:
prompt = f"<|im_start|>system\n{prompt}<|im_end|>\n"
if history is not None:
for q, a in history:
prompt += "<|im_start|>user\n{}<|im_end|>\n<|im_start|>assistant\n{}<|im_end|>".format(q, a)
prompt += "<|im_start|>user\n{}<|im_end|>\n<|im_start|>assistant\n".format(query)
if answer is not None:
prompt += answer + '<|im_end|>'
return prompt
@classmethod
def build_template_yi(cls, query, answer=None, history=None, prefix=None):
prompt = prefix or ''
if prompt:
prompt = f"<|im_start|>system\n{prompt}<|im_end|>\n"
if history is not None:
for q, a in history:
prompt += "<|im_start|>user\n{}<|im_end|>\n<|im_start|>assistant\n{}<|im_end|>".format(q, a)
prompt += "<|im_start|>user\n{}<|im_end|>\n<|im_start|>assistant\n".format(query)
if answer is not None:
prompt += answer + '<|im_end|>'
return prompt
@classmethod
def build_template_internlm(cls,query, answer=None, prefix=None, history=None):
prompt = prefix or ''
if history is not None:
for q, a in history:
prompt += f"""<s><|User|>:{q}<eoh>\n<|Bot|>:{a}<eoa>\n"""
if len(prompt) == 0:
prompt += "<s>"
prompt += f"""<|User|>:{query}<eoh>\n<|Bot|>:"""
if answer is not None:
prompt += answer
return prompt
@classmethod
def build_template_qwen2(cls, query, answer=None, prefix=None, history=None):
prompt = prefix or '<|im_start|>system\nYou are a helpful assistant<|im_end|>\n'
if history is not None:
for q, a in history:
prompt += '<|im_start|>' + 'user' + '\n' + q + '<|im_end|>'
prompt += '<|im_start|>' + 'assistant' + '\n' + a + '<|im_end|>'
prompt += '<|im_start|>' + 'assistant' + '\n'
if answer is not None:
prompt += answer
return prompt
#切换模板
build_template = PromptBuilder.build_template_qwen2
class TokenIdsMaker:
@classmethod
def final(cls, tokenizer, input_ids, labels, max_seq_length):
seqlen = np.asarray(len(input_ids), dtype=np.int32)
pad_len = max_seq_length - seqlen
input_ids = np.asarray(input_ids, dtype=np.int32)
attention_mask = np.asarray([1] * len(input_ids), dtype=np.int32)
labels = np.asarray(labels, dtype=np.int32)
if pad_len:
pad_val = tokenizer.eos_token_id
input_ids = np.pad(input_ids, (0, pad_len), 'constant', constant_values=(pad_val, pad_val))
attention_mask = np.pad(attention_mask, (0, pad_len), 'constant', constant_values=(0, 0))
labels = np.pad(labels, (0, pad_len), 'constant', constant_values=(-100, -100))
d = {
'input_ids': input_ids,
'attention_mask': attention_mask,
'labels': labels,
'seqlen': seqlen
}
return d
@classmethod
def tunction(cls, tokenizer: PreTrainedTokenizer, config, sup, max_seq_length, examples):
sptoken = [config.bos_token_id]
ds = []
prefix = None
history = []
for sid, (role,q,a) in enumerate(examples):
if role == 'system':
prefix = q
continue
history += [(q,a)]
a_ids = tokenizer.encode(text=build_template(q,prefix=prefix,history=history[:-1]), add_special_tokens=False)
b_ids = tokenizer.encode(text=a, add_special_tokens=False)
while len(a_ids) + len(b_ids) > max_seq_length - len(sptoken) - 1:
if len(b_ids) > len(a_ids):
b_ids.pop(-1)
else:
a_ids.pop(0)
b_ids += [config.eos_token_id]
input_ids = a_ids + b_ids
labels = copy.deepcopy(input_ids) if not sup else [-100] * len(a_ids) + copy.deepcopy(b_ids)
input_ids = sptoken + input_ids
labels = sptoken + labels if not sup else [-100] * len(sptoken) + labels
assert len(input_ids) <= max_seq_length
ds.append(cls.final(tokenizer, input_ids, labels, max_seq_length))
return ds
@classmethod
def slidding(cls, tokenizer: PreTrainedTokenizer,config,stride,max_seq_length, examples,
sliding_size=None,
src_max_length=-1,
dst_max_length=-1,
sup=True):
sptoken = [config.bos_token_id]
if sliding_size is None or sliding_size < 0:
sliding_size = max_seq_length - len(sptoken)
assert sliding_size <= max_seq_length - len(sptoken)
ds = []
prefix = None
history = []
for sid, (role, q, a) in enumerate(examples):
if role == 'system':
prefix = q
continue
history += [(q, a)]
a_ids = tokenizer.encode(text=build_template(q, prefix=prefix, history=history[:-1]),add_special_tokens=False)
b_ids = tokenizer.encode(text=a, add_special_tokens=False)
if src_max_length and src_max_length > 0:
a_ids = a_ids[:src_max_length]
if dst_max_length and dst_max_length > 0:
b_ids = b_ids[:dst_max_length]
b_ids += [config.eos_token_id]
input_ids_qa = a_ids + b_ids
labels_all = copy.deepcopy(input_ids_qa) if not sup else [-100] * len(a_ids) + b_ids
pos = 0
while pos < len(input_ids_qa):
input_ids = input_ids_qa[pos:pos + max_seq_length - len(sptoken)]
labels = labels_all[pos:pos + max_seq_length - len(sptoken)]
pos += sliding_size
if np.all(np.asarray(labels) == -100):
continue
input_ids = sptoken + input_ids
labels = sptoken + labels if not sup else [-100] * len(sptoken) + labels
ds.append(cls.final(tokenizer, input_ids, labels, max_seq_length))
return ds