-
Notifications
You must be signed in to change notification settings - Fork 268
/
qa_finetuner.py
266 lines (234 loc) · 11.2 KB
/
qa_finetuner.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
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
import warnings
import pandas as pd
import tensorflow as tf
def convert_to_dataset(list_of_dicts):
try:
from datasets import Dataset, load_dataset
except ImportError:
raise ImportError(
"The datasets package is required for fine-tuning QA models: pip install datasets"
)
new_list = []
for d in list_of_dicts:
if "question" not in d or "context" not in d or "answers" not in d:
raise ValueError(
'All dictionaries in list must have the following keys: "question", "context", "answers"'
)
q = d["question"].strip()
c = d["context"].strip()
a = d["answers"].strip() if isinstance(d["answers"], str) else d["answers"]
if a is None:
answers = {"text": [], "answer_start": []}
elif isinstance(a, dict):
ans_start = a["answer_start"]
ans_text = a["text"]
answers = {"text": ans_text, "answer_start": ans_start}
elif isinstance(a, str):
ans_start = c.index(a)
if ans_start < 0:
raise ValueError("Could not locate answer in context: %s" % (d))
ans_start = [ans_start]
answers = {"text": [a], "answer_start": ans_start}
else:
raise ValueError(
'Value for "answers" key must be a string, dictionary, or None'
)
new_d = {"question": q, "context": c, "answers": answers}
new_list.append(new_d)
dataset = Dataset.from_pandas(pd.DataFrame(new_list))
# uncomment to test with squad sample
# datasets = load_dataset('squad')
# td = datasets['train']
# td = Dataset.from_pandas(td.to_pandas().head(100))
# return td
return dataset
def convert_dataset_for_tensorflow(
dataset,
batch_size,
dataset_mode="variable_batch",
shuffle=True,
drop_remainder=True,
):
"""Converts a Hugging Face dataset to a Tensorflow Dataset. The dataset_mode controls whether we pad all batches
to the maximum sequence length, or whether we only pad to the maximum length within that batch. The former
is most useful when training on TPU, as a new graph compilation is required for each sequence length.
"""
try:
from datasets import Dataset, load_dataset
except ImportError:
raise ImportError(
"The datasets package is required for fine-tuning QA models: pip install datasets"
)
def densify_ragged_batch(features, label=None):
features = {
feature: (
ragged_tensor.to_tensor(shape=batch_shape[feature])
if feature in tensor_keys
else ragged_tensor
)
for feature, ragged_tensor in features.items()
}
if label is None:
return features
else:
return features, label
tensor_keys = ["attention_mask", "input_ids"]
label_keys = ["start_positions", "end_positions"]
if dataset_mode == "variable_batch":
batch_shape = {key: None for key in tensor_keys}
data = {key: tf.ragged.constant(dataset[key]) for key in tensor_keys}
elif dataset_mode == "constant_batch":
data = {key: tf.ragged.constant(dataset[key]) for key in tensor_keys}
batch_shape = {
key: tf.concat(([batch_size], ragged_tensor.bounding_shape()[1:]), axis=0)
for key, ragged_tensor in data.items()
}
else:
raise ValueError("Unknown dataset mode!")
if all([key in dataset.features for key in label_keys]):
for key in label_keys:
data[key] = tf.convert_to_tensor(dataset[key])
dummy_labels = tf.zeros_like(dataset[key])
tf_dataset = tf.data.Dataset.from_tensor_slices((data, dummy_labels))
else:
tf_dataset = tf.data.Dataset.from_tensor_slices(data)
if shuffle:
tf_dataset = tf_dataset.shuffle(buffer_size=len(dataset))
tf_dataset = tf_dataset.batch(
batch_size=batch_size, drop_remainder=drop_remainder
).map(densify_ragged_batch)
return tf_dataset
class QAFineTuner:
def __init__(self, model, tokenizer):
self.model = model
self.tokenizer = tokenizer
def finetune(
self, data, epochs=3, learning_rate=2e-5, batch_size=8, max_seq_length=512
):
"""
```
Finetune a QA model.
Args:
data(list): list of dictionaries of the form:
[{'question': 'What is ktrain?'
'context': 'ktrain is a low-code library for augmented machine learning.'
'answer': 'ktrain'}]
epochs(int): number of epochs. Default:3
learning_rate(float): learning rate. Default: 2e-5
Returns:
fine-tuned model (i.e., QAFineTuner.model)
```
"""
if self.model.name_or_path.startswith("bert-large"):
warnings.warn(
"You are fine-tuning a bert-large-* model, which requires lots of memory. "
+ "If you get a memory error, switch to distilbert-base-cased-distilled-squad."
)
if batch_size > len(data):
batch_size = len(data)
model = self.model
tokenizer = self.tokenizer
question_column_name = "question"
context_column_name = "context"
answer_column_name = "answers"
# Padding side determines if we do (question|context) or (context|question).
pad_on_right = tokenizer.padding_side == "right"
# Training preprocessing
def prepare_train_features(examples):
# Some of the questions have lots of whitespace on the left, which is not useful and will make the
# truncation of the context fail (the tokenized question will take a lots of space). So we remove that
# left whitespace
examples[question_column_name] = [
q.lstrip() for q in examples[question_column_name]
]
# Tokenize our examples with truncation and maybe padding, but keep the overflows using a stride. This results
# in one example possible giving several features when a context is long, each of those features having a
# context that overlaps a bit the context of the previous feature.
tokenized_examples = tokenizer(
examples[question_column_name if pad_on_right else context_column_name],
examples[context_column_name if pad_on_right else question_column_name],
truncation="only_second" if pad_on_right else "only_first",
max_length=max_seq_length,
stride=128,
return_overflowing_tokens=True,
return_offsets_mapping=True,
# padding="max_length" if data_args.pad_to_max_length else False,
padding="max_length",
)
# Since one example might give us several features if it has a long context, we need a map from a feature to
# its corresponding example. This key gives us just that.
sample_mapping = tokenized_examples.pop("overflow_to_sample_mapping")
# The offset mappings will give us a map from token to character position in the original context. This will
# help us compute the start_positions and end_positions.
offset_mapping = tokenized_examples.pop("offset_mapping")
# Let's label those examples!
tokenized_examples["start_positions"] = []
tokenized_examples["end_positions"] = []
for i, offsets in enumerate(offset_mapping):
# We will label impossible answers with the index of the CLS token.
input_ids = tokenized_examples["input_ids"][i]
cls_index = input_ids.index(tokenizer.cls_token_id)
# Grab the sequence corresponding to that example (to know what is the context and what is the question).
sequence_ids = tokenized_examples.sequence_ids(i)
# One example can give several spans, this is the index of the example containing this span of text.
sample_index = sample_mapping[i]
answers = examples[answer_column_name][sample_index]
# If no answers are given, set the cls_index as answer.
if len(answers["answer_start"]) == 0:
tokenized_examples["start_positions"].append(cls_index)
tokenized_examples["end_positions"].append(cls_index)
else:
# Start/end character index of the answer in the text.
start_char = answers["answer_start"][0]
end_char = start_char + len(answers["text"][0])
# Start token index of the current span in the text.
token_start_index = 0
while sequence_ids[token_start_index] != (1 if pad_on_right else 0):
token_start_index += 1
# End token index of the current span in the text.
token_end_index = len(input_ids) - 1
while sequence_ids[token_end_index] != (1 if pad_on_right else 0):
token_end_index -= 1
# Detect if the answer is out of the span (in which case this feature is labeled with the CLS index).
if not (
offsets[token_start_index][0] <= start_char
and offsets[token_end_index][1] >= end_char
):
tokenized_examples["start_positions"].append(cls_index)
tokenized_examples["end_positions"].append(cls_index)
else:
# Otherwise move the token_start_index and token_end_index to the two ends of the answer.
# Note: we could go after the last offset if the answer is the last word (edge case).
while (
token_start_index < len(offsets)
and offsets[token_start_index][0] <= start_char
):
token_start_index += 1
tokenized_examples["start_positions"].append(
token_start_index - 1
)
while offsets[token_end_index][1] >= end_char:
token_end_index -= 1
tokenized_examples["end_positions"].append(token_end_index + 1)
return tokenized_examples
dataset = convert_to_dataset(data)
print(dataset)
print(dataset[0])
train_dataset = dataset.map(
prepare_train_features,
batched=True,
num_proc=1,
)
optimizer = tf.keras.optimizers.Adam(learning_rate=learning_rate)
def dummy_loss(y_true, y_pred):
return tf.reduce_mean(y_pred)
losses = {"loss": dummy_loss}
model.compile(optimizer=optimizer, loss=losses)
training_dataset = convert_dataset_for_tensorflow(
train_dataset,
batch_size=batch_size,
dataset_mode="variable_batch",
drop_remainder=True,
shuffle=True,
)
model.fit(training_dataset, epochs=int(epochs))