def test_load_data_with_null_and_index_column(self):
actual = data_file("test/resources/data_with_null_values.csv",
index="id")
self.assertIsInstance(actual, pd.DataFrame)
assert_array_equal(["text1", "text2", "label"], actual.columns)
self.assertEqual(3, len(actual))
assert_array_equal([1, 3, 5], actual.index)
def train_or_continue(args, training_operation):
from bisemantic.data import cross_validation_partitions
training = data_file(args.training, args.n, args.index_name,
args.text_1_name, args.text_2_name, args.label_name,
args.invalid_labels, not args.not_comma_delimited)
if args.validation_fraction is not None:
training, validation = cross_validation_partitions(
training, 1 - args.validation_fraction, 1)[0]
elif args.validation_set is not None:
validation = data_file(args.validation_set, args.n, args.index_name,
args.text_1_name, args.text_2_name,
args.label_name, args.invalid_labels,
not args.not_comma_delimited)
else:
validation = None
_, training_history = training_operation(training, validation)
print(training_history.latest_run_summary())
def score(args):
from bisemantic.classifier import TextPairClassifier
test = data_file(args.test, args.n, args.index_name, args.text_1_name,
args.text_2_name, args.label_name, args.invalid_labels,
not args.not_comma_delimited)
logger.info("Score predictions for %d pairs" % len(test))
model = TextPairClassifier.load_from_model_directory(
args.model_directory_name)
scores = model.score(test, batch_size=args.batch_size)
print(", ".join("%s=%0.5f" % s for s in scores))
def create_cross_validation_partitions(args):
from bisemantic.data import cross_validation_partitions
data = data_file(args.data, args.n, args.index_name, args.text_1_name,
args.text_2_name, args.label_name, args.invalid_labels,
not args.not_comma_delimited)
for i, (train_partition, validate_partition) in enumerate(
cross_validation_partitions(data, args.fraction, args.k)):
train_name, validate_name = [
os.path.join(args.output_directory,
"%s.%d.%s.csv" % (args.prefix, i + 1, name))
for name in ["train", "validate"]
]
train_partition.to_csv(train_name)
validate_partition.to_csv(validate_name)
def predict(args):
from bisemantic.classifier import TextPairClassifier
test = data_file(args.test, args.n, args.index_name, args.text_1_name,
args.text_2_name, args.label_name, args.invalid_labels,
not args.not_comma_delimited)
logger.info("Predict labels for %d pairs" % len(test))
model = TextPairClassifier.load_from_model_directory(
args.model_directory_name)
class_names = TextPairClassifier.class_names_from_model_directory(
args.model_directory_name)
predictions = model.predict(test,
batch_size=args.batch_size,
class_names=class_names)
print(predictions.to_csv())
def test_load_snli_format(self):
snli = data_file("test/resources/snli.csv",
index="pairID",
text_1_name="sentence1",
text_2_name="sentence2",
label_name="gold_label",
invalid_labels=["-"],
comma_delimited=False)
assert_array_equal(["text1", "text2", "label"], snli.columns)
# There are 9 samples because one is dropped because it has "-" as a gold_label.
self.assertEqual(9, len(snli))
# The pairID index is names of JPEGs.
self.assertTrue(snli.index.str.match(r"\d{10}\.jpg#\dr\d[nec]").all())
self.assertEqual({'contradiction', 'neutral', 'entailment'},
set(snli.label.values))