def testLargerBatchSize(self):
with tf.Session() as sess:
with tempfile.NamedTemporaryFile(mode='w+t', delete=False) as vocab:
raw_data = {
'label': ['1', '2'],
'text_a': ['The boy jumped into the air.', 'The cat sat on a hat.'],
'lang': ['en', 'en'],
}
expected_wordpieces = ['The', '[UNK]', 'jumped', 'in', '##to', 'the',
'air', '.', 'The', 'cat', 'sat', 'on', 'a', 'h',
'##at', '.']
vocab.writelines([word + '\n' for word in self.vocab])
vocab.flush()
preprocessing_fn = utils.metrics_preprocessing_fn(
vocab.name, 'text_a', 'lang')
outputs = preprocessing_fn(raw_data)
tf.tables_initializer().run()
outputs = sess.run(outputs)
self.assertSequenceAlmostEqual(outputs['lang'], ['en', 'en'])
self.assertSequenceAlmostEqual(outputs['num_preserved_chars'], [20, 16])
self.assertSequenceAlmostEqual(outputs['num_dropped_chars'], [3, 0])
self.assertSequenceAlmostEqual(outputs['wordpieces'].values,
expected_wordpieces)
self.assertSequenceAlmostEqual(outputs['num_non_unk_wordpieces'],
[7, 8])
def run_metrics():
"""Creates a pipeline to measure wordpiece vocab metrics over a corpus."""
metrics_pipeline = beam.Pipeline()
with tft_beam.Context(temp_dir=tempfile.mkdtemp()):
# Read raw data and convert to TF Transform encoded dict.
raw_data = (
metrics_pipeline
| 'ReadInputData' >> beam.io.tfrecordio.ReadFromTFRecord(
data_file, coder=beam.coders.ProtoCoder(tf.train.Example))
| 'DecodeInputData' >> beam.Map(example_converter.decode))
# Apply transform to wordpiece-tokenize input.
(metrics_transformed_data, _), _ = (
(raw_data, raw_metadata)
| 'WordpieceTokenizeInput' >> tft_beam.AnalyzeAndTransformDataset(
utils.metrics_preprocessing_fn(FLAGS.vocab_file,
FLAGS.text_key,
FLAGS.language_code_key)))
# Initialize CSV coder. Aggregate values for each lang, calculate metrics,
# and write to output to a CSV file.
csv_converter = tft.coders.CsvCoder(columns, csv_schema)
_ = (
metrics_transformed_data
| 'CompileTokenInfo' >> beam.ParDo(utils.CompileTokenizationInfo())
| 'CombineStatsForLang' >> beam.CombineGlobally(utils.AggregateLang())
| 'CalculateMetrics' >> beam.ParDo(utils.CalculateMetrics())
| 'EncodeMetrics' >> beam.Map(csv_converter.encode)
| 'WriteMetrics' >> beam.io.WriteToText(
metrics_file, shard_name_template='', header=','.join(columns)))
return metrics_pipeline
def testSingleElement(self):
with tf.Session() as sess:
with tempfile.NamedTemporaryFile(mode='w+t', delete=False) as vocab:
vocab.writelines([word + '\n' for word in self.vocab])
vocab.flush()
preprocessing_fn = utils.metrics_preprocessing_fn(
vocab.name, 'text_a', 'lang')
outputs = preprocessing_fn(self.raw_data)
tf.tables_initializer().run()
outputs = sess.run(outputs)
self.assertEqual(outputs['lang'], 'en')
self.assertEqual(outputs['num_non_unk_wordpieces'], 7)
self.assertEqual(outputs['num_preserved_chars'], 20)
self.assertEqual(outputs['num_dropped_chars'], 3)
self.assertSequenceAlmostEqual(outputs['wordpieces'].values,
self.expected_wordpieces)
def calculate_metrics():
"""Returns a pipeline to compute wordpiece model stats given a vocab and corpus."""
# Schema of input dataset.
raw_metadata = dataset_metadata.DatasetMetadata(
dataset_schema.from_feature_spec({
'text':
tf.FixedLenFeature([], tf.string),
'language_code':
tf.FixedLenFeature([], tf.string),
}))
# Schema to format metrics as CSV.
csv_schema = dataset_schema.from_feature_spec({
'lang':
tf.FixedLenFeature([], tf.string),
'sample_count':
tf.FixedLenFeature([], tf.int64),
'micro_drop_char_percent':
tf.FixedLenFeature([], tf.string),
'macro_drop_char_percent':
tf.FixedLenFeature([], tf.string),
'micro_compress_ratio':
tf.FixedLenFeature([], tf.string),
'macro_compress_ratio':
tf.FixedLenFeature([], tf.string),
'unweighted_en_wp_overlap_percent':
tf.FixedLenFeature([], tf.string),
'weighted_en_wp_overlap_percent':
tf.FixedLenFeature([], tf.string),
})
columns = [
'lang', 'sample_count', 'micro_drop_char_percent',
'macro_drop_char_percent', 'micro_compress_ratio',
'macro_compress_ratio', 'unweighted_en_wp_overlap_percent',
'weighted_en_wp_overlap_percent'
]
# Create pipeline.
pipeline = beam.Pipeline()
with tft_beam.Context(temp_dir=tempfile.mkdtemp()):
example_converter = tft.coders.ExampleProtoCoder(raw_metadata.schema,
serialized=False)
csv_converter = tft.coders.CsvCoder(columns, csv_schema)
# Read raw data and convert to TF Transform encoded dict.
raw_data = (pipeline
| 'ReadInputData' >> beam.io.tfrecordio.ReadFromTFRecord(
FLAGS.input_file,
coder=beam.coders.ProtoCoder(tf.train.Example))
| 'DecodeInputData' >> beam.Map(example_converter.decode))
# Apply transform to wordpiece-tokenize input.
(transformed_data, _), _ = (
(raw_data, raw_metadata)
| 'WordpieceTokenizeInput' >> tft_beam.AnalyzeAndTransformDataset(
utils.metrics_preprocessing_fn(
FLAGS.vocab_file, FLAGS.text_key, FLAGS.language_code_key))
)
# Aggregate values for each lang, calculate metrics, and write to output.
_ = (transformed_data
|
'CompileTokenInfo' >> beam.ParDo(utils.CompileTokenizationInfo())
| 'CombineStatsForLang' >> beam.CombineGlobally(
utils.AggregateLang())
| 'CalculateMetrics' >> beam.ParDo(utils.CalculateMetrics())
| 'EncodeMetrics' >> beam.Map(csv_converter.encode)
| 'WriteMetrics' >> beam.io.WriteToText(FLAGS.output_file,
shard_name_template='',
header=','.join(columns)))
return pipeline