def test_init_config(lang, pipeline, optimize):
# TODO: add more tests and also check for GPU with transformers
config = init_config(lang=lang,
pipeline=pipeline,
optimize=optimize,
gpu=False)
assert isinstance(config, Config)
def test_init_labels(component_name):
nlp = Dutch()
component = nlp.add_pipe(component_name)
for label in ["T1", "T2", "T3", "T4"]:
component.add_label(label)
assert len(nlp.get_pipe(component_name).labels) == 4
with make_tempdir() as tmp_dir:
_init_labels(nlp, tmp_dir)
config = init_config(
lang="nl",
pipeline=[component_name],
optimize="efficiency",
gpu=False,
)
config["initialize"]["components"][component_name] = {
"labels": {
"@readers": "spacy.read_labels.v1",
"path": f"{tmp_dir}/{component_name}.json",
}
}
nlp2 = load_model_from_config(config, auto_fill=True)
assert len(nlp2.get_pipe(component_name).labels) == 0
nlp2.initialize()
assert len(nlp2.get_pipe(component_name).labels) == 4
def test_init_config(lang, pipeline, optimize, pretraining):
# TODO: add more tests and also check for GPU with transformers
config = init_config(lang=lang,
pipeline=pipeline,
optimize=optimize,
pretraining=pretraining,
gpu=False)
assert isinstance(config, Config)
if pretraining:
config["paths"]["raw_text"] = "my_data.jsonl"
nlp = load_model_from_config(config, auto_fill=True)