def test_ner_labels_added_implicitly_on_update():
nlp = Language()
ner = nlp.add_pipe("ner")
for label in ["A", "B", "C"]:
ner.add_label(label)
nlp.initialize()
doc = Doc(nlp.vocab, words=["hello", "world"], ents=["B-D", "O"])
example = Example(nlp.make_doc(doc.text), doc)
assert "D" not in ner.labels
nlp.update([example])
assert "D" in ner.labels
def simple_nlp():
nlp = Language()
nlp.add_pipe("transformer")
train_examples = []
for t in TRAIN_DATA:
train_examples.append(Example.from_dict(nlp.make_doc(t[0]), t[1]))
optimizer = nlp.initialize()
for i in range(2):
losses = {}
nlp.update(train_examples, sgd=optimizer, losses=losses)
return nlp
def test_transformer_pipeline_simple():
"""Test that a simple pipeline with just a transformer at least runs"""
nlp = Language()
nlp.add_pipe("transformer")
train_examples = []
for t in TRAIN_DATA:
train_examples.append(Example.from_dict(nlp.make_doc(t[0]), t[1]))
optimizer = nlp.initialize()
for i in range(2):
losses = {}
nlp.update(train_examples, sgd=optimizer, losses=losses)
doc = nlp("We're interested at underwater basket weaving.")
assert doc
def test_issue999(train_data):
"""Test that adding entities and resuming training works passably OK.
There are two issues here:
1) We have to readd labels. This isn't very nice.
2) There's no way to set the learning rate for the weight update, so we
end up out-of-scale, causing it to learn too fast.
"""
TRAIN_DATA = [
["hey", []],
["howdy", []],
["hey there", []],
["hello", []],
["hi", []],
["i'm looking for a place to eat", []],
[
"i'm looking for a place in the north of town",
[[31, 36, "LOCATION"]]
],
["show me chinese restaurants", [[8, 15, "CUISINE"]]],
["show me chines restaurants", [[8, 14, "CUISINE"]]],
]
nlp = Language()
ner = nlp.create_pipe("ner")
nlp.add_pipe(ner)
for _, offsets in TRAIN_DATA:
for start, end, label in offsets:
ner.add_label(label)
nlp.begin_training()
ner.model.learn_rate = 0.001
for itn in range(100):
random.shuffle(TRAIN_DATA)
for raw_text, entity_offsets in TRAIN_DATA:
nlp.update([raw_text], [{"entities": entity_offsets}])
with make_tempdir() as model_dir:
nlp.to_disk(model_dir)
nlp2 = Language().from_disk(model_dir)
for raw_text, entity_offsets in TRAIN_DATA:
doc = nlp2(raw_text)
ents = {(ent.start_char, ent.end_char): ent.label_ for ent in doc.ents}
for start, end, label in entity_offsets:
if (start, end) in ents:
assert ents[(start, end)] == label
break
else:
if entity_offsets:
raise Exception(ents)
def test_simple_train():
nlp = Language()
textcat = nlp.add_pipe("textcat")
textcat.add_label("answer")
nlp.initialize()
for i in range(5):
for text, answer in [
("aaaa", 1.0),
("bbbb", 0),
("aa", 1.0),
("bbbbbbbbb", 0.0),
("aaaaaa", 1),
]:
nlp.update((text, {"cats": {"answer": answer}}))
doc = nlp("aaa")
assert "answer" in doc.cats
assert doc.cats["answer"] >= 0.5
def test_simple_train():
nlp = Language()
nlp.add_pipe(nlp.create_pipe("textcat"))
nlp.get_pipe("textcat").add_label("answer")
nlp.begin_training()
for i in range(5):
for text, answer in [
("aaaa", 1.0),
("bbbb", 0),
("aa", 1.0),
("bbbbbbbbb", 0.0),
("aaaaaa", 1),
]:
nlp.update([text], [{"cats": {"answer": answer}}])
doc = nlp("aaa")
assert "answer" in doc.cats
assert doc.cats["answer"] >= 0.5
def test_simple_train():
nlp = Language()
nlp.add_pipe(nlp.create_pipe("textcat"))
nlp.get_pipe("textcat").add_label("answer")
nlp.begin_training()
for i in range(5):
for text, answer in [
("aaaa", 1.0),
("bbbb", 0),
("aa", 1.0),
("bbbbbbbbb", 0.0),
("aaaaaa", 1),
]:
nlp.update([text], [{"cats": {"answer": answer}}])
doc = nlp("aaa")
assert "answer" in doc.cats
assert doc.cats["answer"] >= 0.5
def test_issue999(train_data):
"""Test that adding entities and resuming training works passably OK.
There are two issues here:
1) We have to readd labels. This isn't very nice.
2) There's no way to set the learning rate for the weight update, so we
end up out-of-scale, causing it to learn too fast.
"""
TRAIN_DATA = [
["hey", []],
["howdy", []],
["hey there", []],
["hello", []],
["hi", []],
["i'm looking for a place to eat", []],
["i'm looking for a place in the north of town", [[31, 36, "LOCATION"]]],
["show me chinese restaurants", [[8, 15, "CUISINE"]]],
["show me chines restaurants", [[8, 14, "CUISINE"]]],
]
nlp = Language()
ner = nlp.create_pipe("ner")
nlp.add_pipe(ner)
for _, offsets in TRAIN_DATA:
for start, end, label in offsets:
ner.add_label(label)
nlp.begin_training()
ner.model.learn_rate = 0.001
for itn in range(100):
random.shuffle(TRAIN_DATA)
for raw_text, entity_offsets in TRAIN_DATA:
nlp.update([raw_text], [{"entities": entity_offsets}])
with make_tempdir() as model_dir:
nlp.to_disk(model_dir)
nlp2 = Language().from_disk(model_dir)
for raw_text, entity_offsets in TRAIN_DATA:
doc = nlp2(raw_text)
ents = {(ent.start_char, ent.end_char): ent.label_ for ent in doc.ents}
for start, end, label in entity_offsets:
if (start, end) in ents:
assert ents[(start, end)] == label
break
else:
if entity_offsets:
raise Exception(ents)
def test_simple_train():
fix_random_seed(0)
nlp = Language()
spancat = nlp.add_pipe("spancat", config={"spans_key": SPAN_KEY})
get_examples = make_get_examples(nlp)
nlp.initialize(get_examples)
sgd = nlp.create_optimizer()
assert len(spancat.labels) != 0
for i in range(40):
losses = {}
nlp.update(list(get_examples()), losses=losses, drop=0.1, sgd=sgd)
doc = nlp("I like London and Berlin.")
assert doc.spans[spancat.key] == doc.spans[SPAN_KEY]
assert len(doc.spans[spancat.key]) == 2
assert doc.spans[spancat.key][0].text == "London"
scores = nlp.evaluate(get_examples())
assert f"spans_{SPAN_KEY}_f" in scores
assert scores[f"spans_{SPAN_KEY}_f"] == 1.0
def main(vectors_loc=None, lang=None):
if lang is None:
nlp = Language()
else:
# create empty language class – this is required if you're planning to
# save the model to disk and load it back later (models always need a
# "lang" setting). Use 'xx' for blank multi-language class.
nlp = spacy.blank(lang)
with open(VECTORS_PATH, "rb") as file_:
print("loading vectors...")
header = file_.readline()
nr_row, nr_dim = header.split()
nlp.vocab.reset_vectors(width=int(nr_dim))
for line in file_:
line = line.rstrip().decode("utf8")
pieces = line.rsplit(" ", int(nr_dim))
word = pieces[0]
vector = numpy.asarray([float(v) for v in pieces[1:]], dtype="f")
nlp.vocab.set_vector(word, vector) # add the vectors to the vocab
tagger = nlp.create_pipe("tagger")
# Add the tags. This needs to be done before you start training.
print("trainning tags...")
for tag, values in TAG_MAP.items():
tagger.add_label(tag, values)
nlp.add_pipe(tagger)
optimizer = nlp.begin_training()
for i in range(20):
random.shuffle(TRAIN_DATA)
losses = {}
for text, annotations in TRAIN_DATA:
nlp.update([text], [annotations], sgd=optimizer, losses=losses)
print(losses)
# test the trained model
test_text = "Eu desejo ouvir uma música muito boa"
doc = nlp(test_text)
print("Tags", [(t.text, t.tag_, t.pos_) for t in doc])
print("Saved mode to nl_model_tagger")
nlp.to_disk("/app/model")
def test_example_batch(nlp: Language, example_gold):
texts, golds = zip(*example_gold)
optim = nlp.resume_training()
nlp.update(texts, golds, optim)
def test_update(nlp: Language, label_type):
if label_type == "irex":
pytest.skip("label type mismatch")
optim = nlp.resume_training()
nlp.update(*zip(*TESTCASE_ENE), optim)
