File size: 5,060 Bytes
97b6013 |
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 |
# Lint as: python3
# Copyright 2020 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
"""Tests for official.nlp.tasks.question_answering."""
import functools
import os
import tensorflow as tf
from official.nlp.bert import configs
from official.nlp.bert import export_tfhub
from official.nlp.configs import bert
from official.nlp.configs import encoders
from official.nlp.tasks import question_answering
class QuestionAnsweringTaskTest(tf.test.TestCase):
def setUp(self):
super(QuestionAnsweringTaskTest, self).setUp()
self._encoder_config = encoders.TransformerEncoderConfig(
vocab_size=30522, num_layers=1)
self._train_data_config = bert.QADataConfig(
input_path="dummy", seq_length=128, global_batch_size=1)
def _run_task(self, config):
task = question_answering.QuestionAnsweringTask(config)
model = task.build_model()
metrics = task.build_metrics()
strategy = tf.distribute.get_strategy()
dataset = strategy.experimental_distribute_datasets_from_function(
functools.partial(task.build_inputs, config.train_data))
iterator = iter(dataset)
optimizer = tf.keras.optimizers.SGD(lr=0.1)
task.train_step(next(iterator), model, optimizer, metrics=metrics)
task.validation_step(next(iterator), model, metrics=metrics)
def test_task(self):
# Saves a checkpoint.
pretrain_cfg = bert.BertPretrainerConfig(
encoder=self._encoder_config,
num_masked_tokens=20,
cls_heads=[
bert.ClsHeadConfig(
inner_dim=10, num_classes=3, name="next_sentence")
])
pretrain_model = bert.instantiate_bertpretrainer_from_cfg(pretrain_cfg)
ckpt = tf.train.Checkpoint(
model=pretrain_model, **pretrain_model.checkpoint_items)
saved_path = ckpt.save(self.get_temp_dir())
config = question_answering.QuestionAnsweringConfig(
init_checkpoint=saved_path,
network=self._encoder_config,
train_data=self._train_data_config)
task = question_answering.QuestionAnsweringTask(config)
model = task.build_model()
metrics = task.build_metrics()
dataset = task.build_inputs(config.train_data)
iterator = iter(dataset)
optimizer = tf.keras.optimizers.SGD(lr=0.1)
task.train_step(next(iterator), model, optimizer, metrics=metrics)
task.validation_step(next(iterator), model, metrics=metrics)
task.initialize(model)
def test_task_with_fit(self):
config = question_answering.QuestionAnsweringConfig(
network=self._encoder_config,
train_data=self._train_data_config)
task = question_answering.QuestionAnsweringTask(config)
model = task.build_model()
model = task.compile_model(
model,
optimizer=tf.keras.optimizers.SGD(lr=0.1),
train_step=task.train_step,
metrics=[tf.keras.metrics.SparseCategoricalAccuracy(name="accuracy")])
dataset = task.build_inputs(config.train_data)
logs = model.fit(dataset, epochs=1, steps_per_epoch=2)
self.assertIn("loss", logs.history)
self.assertIn("start_positions_accuracy", logs.history)
self.assertIn("end_positions_accuracy", logs.history)
def _export_bert_tfhub(self):
bert_config = configs.BertConfig(
vocab_size=30522,
hidden_size=16,
intermediate_size=32,
max_position_embeddings=128,
num_attention_heads=2,
num_hidden_layers=1)
_, encoder = export_tfhub.create_bert_model(bert_config)
model_checkpoint_dir = os.path.join(self.get_temp_dir(), "checkpoint")
checkpoint = tf.train.Checkpoint(model=encoder)
checkpoint.save(os.path.join(model_checkpoint_dir, "test"))
model_checkpoint_path = tf.train.latest_checkpoint(model_checkpoint_dir)
vocab_file = os.path.join(self.get_temp_dir(), "uncased_vocab.txt")
with tf.io.gfile.GFile(vocab_file, "w") as f:
f.write("dummy content")
hub_destination = os.path.join(self.get_temp_dir(), "hub")
export_tfhub.export_bert_tfhub(bert_config, model_checkpoint_path,
hub_destination, vocab_file)
return hub_destination
def test_task_with_hub(self):
hub_module_url = self._export_bert_tfhub()
config = question_answering.QuestionAnsweringConfig(
hub_module_url=hub_module_url,
network=self._encoder_config,
train_data=self._train_data_config)
self._run_task(config)
if __name__ == "__main__":
tf.test.main()
|