def test_point_generator(x, task, stage):
shuffle = False
batch_size = 3
generator = PointGenerator(x, task, batch_size, stage, shuffle)
assert len(generator) == 1
for x, y in generator:
assert x['ids'].tolist() == [['qid0', 'did0'], ['qid1', 'did1'],
['qid1', 'did0']]
assert x['text_left'].tolist() == [[1, 2], [2, 3], [2, 3]]
assert x['text_right'].tolist() == [[2, 3, 4], [3, 4, 5], [2, 3, 4]]
if stage == 'test':
assert y is None
elif stage == 'train' and task == tasks.Classification(num_classes=3):
assert y.tolist() == [[1, 0, 0], [0, 1, 0], [0, 0, 1]]
break
def __init__(self,
inputs: datapack.DataPack,
task: engine.BaseTask = tasks.Classification(2),
batch_size: int = 32,
stage: str = 'train',
shuffle: bool = True):
"""Construct the point generator.
:param inputs: the output generated by :class:`DataPack`.
:param task: the task is a instance of :class:`engine.BaseTask`.
:param batch_size: number of instances in a batch.
:param shuffle: whether to shuffle the instances while generating a
batch.
"""
self._relation = inputs.relation
self._task = task
self._left = inputs.left
self._right = inputs.right
super().__init__(batch_size, len(inputs.relation), stage, shuffle)
# with no kwargs: (models.DenseBaselineModel, None)
# with kwargs: (models.DenseBaselineModel, {"num_dense_units": 512})
model_setups = [
(models.NaiveModel, None),
(models.DenseBaselineModel, None),
(models.DSSMModel, None)
]
@pytest.fixture(scope='module', params=[1, 32])
def num_samples(request):
return request.param
@pytest.fixture(scope='module', params=[
tasks.Classification(num_classes=2),
tasks.Classification(num_classes=16),
tasks.Ranking()
])
def task(request):
return request.param
@pytest.fixture(params=model_setups)
def raw_model(request):
model_class, custom_kwargs = request.param
model = model_class()
if custom_kwargs:
for key, val in custom_kwargs.items():
model.params[key] = val
return model
def test_classification_num_classes(arg):
task = tasks.Classification(num_classes=arg)
assert task.num_classes == arg
def test_classification_instantiation_failure(arg):
with pytest.raises(Exception):
tasks.Classification(num_classes=arg)
def x():
relation = [['qid0', 'did0', 0], ['qid1', 'did1', 1], ['qid1', 'did0', 2]]
left = [['qid0', [1, 2]], ['qid1', [2, 3]]]
right = [['did0', [2, 3, 4]], ['did1', [3, 4, 5]]]
ctx = {'vocab_size': 6, 'fill_word': 6}
relation = pd.DataFrame(relation, columns=['id_left', 'id_right', 'label'])
left = pd.DataFrame(left, columns=['id_left', 'text_left'])
left.set_index('id_left', inplace=True)
right = pd.DataFrame(right, columns=['id_right', 'text_right'])
right.set_index('id_right', inplace=True)
return DataPack(relation=relation, left=left, right=right, context=ctx)
@pytest.fixture(scope='module',
params=[
tasks.Classification(num_classes=3),
tasks.Ranking(),
])
def task(request):
return request.param
@pytest.fixture(scope='module', params=['train', 'test'])
def stage(request):
return request.param
def test_point_generator(x, task, stage):
shuffle = False
batch_size = 3
generator = PointGenerator(x, task, batch_size, stage, shuffle)
def _guess_task(train_pack):
if np.issubdtype(train_pack.relation['label'].dtype, np.number):
return tasks.Ranking()
elif np.issubdtype(train_pack.relation['label'].dtype, list):
num_classes = int(train_pack.relation['label'].apply(len).max())
return tasks.Classification(num_classes)
