def query_factory(text_preparation_pipeline):
"""For creating queries"""
return QueryFactory(
text_preparation_pipeline=text_preparation_pipeline,
system_entity_recognizer=None,
duckling=True,
)
def query_factory(tokenizer, preprocessor, stemmer, duckling):
"""For creating queries"""
return QueryFactory(
tokenizer=tokenizer,
preprocessor=preprocessor,
stemmer=stemmer,
system_entity_recognizer=duckling,
)
def english_paraphraser_retain_entities(kwik_e_mart_app_path, request):
config = get_augmentation_config(app_path=kwik_e_mart_app_path)
language = "en"
config['augmentor_class'] = "EnglishParaphraser"
config['retain_entities'] = True
query_factory = QueryFactory.create_query_factory(
app_path=kwik_e_mart_app_path, duckling=True)
resource_loader = ResourceLoader.create_resource_loader(
app_path=kwik_e_mart_app_path, query_factory=query_factory)
request.cls.query_factory = query_factory
request.cls.augmentor = AugmentorFactory(
config=config,
language=language,
resource_loader=resource_loader,
).create_augmentor()
yield None
request.cls.augmentor = None
request.cls.query_factor = None
def test_nlp_for_stemmed_queries(kwik_e_mart_nlp, query, stemmed_query):
"""Tests queries that are NOT in the training data but have their stemmed
versions in the training data"""
query_factory = QueryFactory.create_query_factory()
stemmed_tokens = query_factory.create_query(text=query).stemmed_tokens
assert stemmed_query == stemmed_tokens[0]
def query_factory(tokenizer, preprocessor):
"""For creating queries"""
return QueryFactory(tokenizer=tokenizer, preprocessor=preprocessor)
def query_factory(tokenizer):
"""For creating queries"""
return QueryFactory(tokenizer)
import os
import shutil
import pytest
from mindmeld import markup
from mindmeld.models import ENTITY_EXAMPLE_TYPE, ENTITIES_LABEL_TYPE, ModelFactory, ModelConfig
from mindmeld.models.nn_utils.helpers import get_num_weights_of_model
from mindmeld.query_factory import QueryFactory
from mindmeld.resource_loader import ResourceLoader, ProcessedQueryList
APP_NAME = "kwik_e_mart"
APP_PATH = os.path.join(
os.path.dirname(os.path.dirname(os.path.abspath(__file__))), APP_NAME)
GENERATED_TMP_FOLDER = os.path.join(APP_PATH, ".generated/pytorch_module")
QUERY_FACTORY = QueryFactory.create_query_factory(app_path=None, duckling=True)
@pytest.fixture
def resource_loader():
"""A resource loader"""
return ResourceLoader(app_path=None, query_factory=QUERY_FACTORY)
def model_predictions_assertions(model):
"""Conducts assertions on model predictions; common checks across multiple unittests"""
predictions = model.predict([
markup.load_query("Medium Beers pizza from oz pizza",
query_factory=QUERY_FACTORY).query
])[0]
assert len(predictions) <= 6