def main(train_loc, dev_loc, model_dir, tag_map_loc=None):
if tag_map_loc:
with open(tag_map_loc) as file_:
tag_map = json.loads(file_.read())
else:
tag_map = DEFAULT_TAG_MAP
train_sents = list(read_conllx(train_loc))
train_sents = PseudoProjectivity.preprocess_training_data(train_sents)
actions = ArcEager.get_actions(gold_parses=train_sents)
features = get_templates('basic')
model_dir = pathlib.Path(model_dir)
if not (model_dir / 'deps').exists():
(model_dir / 'deps').mkdir()
with (model_dir / 'deps' / 'config.json').open('wb') as file_:
file_.write(
json.dumps(
{'pseudoprojective': True, 'labels': actions, 'features': features}).encode('utf8'))
vocab = Vocab(lex_attr_getters=Language.Defaults.lex_attr_getters, tag_map=tag_map)
# Populate vocab
for _, doc_sents in train_sents:
for (ids, words, tags, heads, deps, ner), _ in doc_sents:
for word in words:
_ = vocab[word]
for dep in deps:
_ = vocab[dep]
for tag in tags:
_ = vocab[tag]
if tag_map:
for tag in tags:
assert tag in tag_map, repr(tag)
tagger = Tagger(vocab, tag_map=tag_map)
parser = DependencyParser(vocab, actions=actions, features=features, L1=0.0)
for itn in range(15):
loss = 0.
for _, doc_sents in train_sents:
for (ids, words, tags, heads, deps, ner), _ in doc_sents:
doc = Doc(vocab, words=words)
gold = GoldParse(doc, tags=tags, heads=heads, deps=deps)
tagger(doc)
loss += parser.update(doc, gold, itn=itn)
doc = Doc(vocab, words=words)
tagger.update(doc, gold)
random.shuffle(train_sents)
scorer = score_model(vocab, tagger, parser, read_conllx(dev_loc))
print('%d:\t%.3f\t%.3f\t%.3f' % (itn, loss, scorer.uas, scorer.tags_acc))
nlp = Language(vocab=vocab, tagger=tagger, parser=parser)
nlp.end_training(model_dir)
scorer = score_model(vocab, tagger, parser, read_conllx(dev_loc))
print('%d:\t%.3f\t%.3f\t%.3f' % (itn, scorer.uas, scorer.las, scorer.tags_acc))
def main(train_loc, dev_loc, model_dir, tag_map_loc):
with open(tag_map_loc) as file_:
tag_map = json.loads(file_.read())
train_sents = list(read_conllx(train_loc))
train_sents = PseudoProjectivity.preprocess_training_data(train_sents)
actions = ArcEager.get_actions(gold_parses=train_sents)
features = get_templates('basic')
model_dir = pathlib.Path(model_dir)
with (model_dir / 'deps' / 'config.json').open('w') as file_:
json.dump({'pseudoprojective': True, 'labels': actions, 'features': features}, file_)
vocab = Vocab(lex_attr_getters=Language.Defaults.lex_attr_getters, tag_map=tag_map)
# Populate vocab
for _, doc_sents in train_sents:
for (ids, words, tags, heads, deps, ner), _ in doc_sents:
for word in words:
_ = vocab[word]
for dep in deps:
_ = vocab[dep]
for tag in tags:
_ = vocab[tag]
for tag in tags:
assert tag in tag_map, repr(tag)
tagger = Tagger(vocab, tag_map=tag_map)
parser = DependencyParser(vocab, actions=actions, features=features)
for itn in range(15):
for _, doc_sents in train_sents:
for (ids, words, tags, heads, deps, ner), _ in doc_sents:
doc = Doc(vocab, words=words)
gold = GoldParse(doc, tags=tags, heads=heads, deps=deps)
tagger(doc)
parser.update(doc, gold)
doc = Doc(vocab, words=words)
tagger.update(doc, gold)
random.shuffle(train_sents)
scorer = score_model(vocab, tagger, parser, read_conllx(dev_loc))
print('%d:\t%.3f\t%.3f' % (itn, scorer.uas, scorer.tags_acc))
nlp = Language(vocab=vocab, tagger=tagger, parser=parser)
nlp.end_training(model_dir)
scorer = score_model(vocab, tagger, parser, read_conllx(dev_loc))
print('%d:\t%.3f\t%.3f\t%.3f' % (itn, scorer.uas, scorer.las, scorer.tags_acc))
def main(lang_name, train_loc, dev_loc, model_dir, clusters_loc=None):
LangClass = spacy.util.get_lang_class(lang_name)
train_sents = list(read_conllx(train_loc))
train_sents = PseudoProjectivity.preprocess_training_data(train_sents)
actions = ArcEager.get_actions(gold_parses=train_sents)
features = get_templates('basic')
model_dir = pathlib.Path(model_dir)
if not model_dir.exists():
model_dir.mkdir()
if not (model_dir / 'deps').exists():
(model_dir / 'deps').mkdir()
if not (model_dir / 'pos').exists():
(model_dir / 'pos').mkdir()
with (model_dir / 'deps' / 'config.json').open('wb') as file_:
file_.write(
json.dumps(
{'pseudoprojective': True, 'labels': actions, 'features': features}).encode('utf8'))
vocab = LangClass.Defaults.create_vocab()
if not (model_dir / 'vocab').exists():
(model_dir / 'vocab').mkdir()
else:
if (model_dir / 'vocab' / 'strings.json').exists():
with (model_dir / 'vocab' / 'strings.json').open() as file_:
vocab.strings.load(file_)
if (model_dir / 'vocab' / 'lexemes.bin').exists():
vocab.load_lexemes(model_dir / 'vocab' / 'lexemes.bin')
if clusters_loc is not None:
clusters_loc = pathlib.Path(clusters_loc)
with clusters_loc.open() as file_:
for line in file_:
try:
cluster, word, freq = line.split()
except ValueError:
continue
lex = vocab[word]
lex.cluster = int(cluster[::-1], 2)
# Populate vocab
for _, doc_sents in train_sents:
for (ids, words, tags, heads, deps, ner), _ in doc_sents:
for word in words:
_ = vocab[word]
for dep in deps:
_ = vocab[dep]
for tag in tags:
_ = vocab[tag]
if vocab.morphology.tag_map:
for tag in tags:
assert tag in vocab.morphology.tag_map, repr(tag)
tagger = Tagger(vocab)
parser = DependencyParser(vocab, actions=actions, features=features, L1=0.0)
for itn in range(30):
loss = 0.
for _, doc_sents in train_sents:
for (ids, words, tags, heads, deps, ner), _ in doc_sents:
doc = Doc(vocab, words=words)
gold = GoldParse(doc, tags=tags, heads=heads, deps=deps)
tagger(doc)
loss += parser.update(doc, gold, itn=itn)
doc = Doc(vocab, words=words)
tagger.update(doc, gold)
random.shuffle(train_sents)
scorer = score_model(vocab, tagger, parser, read_conllx(dev_loc))
print('%d:\t%.3f\t%.3f\t%.3f' % (itn, loss, scorer.uas, scorer.tags_acc))
nlp = LangClass(vocab=vocab, tagger=tagger, parser=parser)
nlp.end_training(model_dir)
scorer = score_model(vocab, tagger, parser, read_conllx(dev_loc))
print('%d:\t%.3f\t%.3f\t%.3f' % (itn, scorer.uas, scorer.las, scorer.tags_acc))
def train(Language,
gold_tuples,
model_dir,
n_iter=15,
feat_set=u'basic',
seed=0,
gold_preproc=False,
n_sents=0,
corruption_level=0,
beam_width=1,
verbose=False,
use_orig_arc_eager=False,
pseudoprojective=False):
dep_model_dir = path.join(model_dir, 'deps')
ner_model_dir = path.join(model_dir, 'ner')
pos_model_dir = path.join(model_dir, 'pos')
if path.exists(dep_model_dir):
shutil.rmtree(dep_model_dir)
if path.exists(ner_model_dir):
shutil.rmtree(ner_model_dir)
if path.exists(pos_model_dir):
shutil.rmtree(pos_model_dir)
os.mkdir(dep_model_dir)
os.mkdir(ner_model_dir)
os.mkdir(pos_model_dir)
if pseudoprojective:
# preprocess training data here before ArcEager.get_labels() is called
gold_tuples = PseudoProjectivity.preprocess_training_data(gold_tuples)
Config.write(dep_model_dir,
'config',
features=feat_set,
seed=seed,
labels=ArcEager.get_labels(gold_tuples),
beam_width=beam_width,
projectivize=pseudoprojective)
Config.write(ner_model_dir,
'config',
features='ner',
seed=seed,
labels=BiluoPushDown.get_labels(gold_tuples),
beam_width=0)
if n_sents > 0:
gold_tuples = gold_tuples[:n_sents]
nlp = Language(data_dir=model_dir,
tagger=False,
parser=False,
entity=False)
if nlp.lang == 'de':
nlp.vocab.morphology.lemmatizer = lambda string, pos: set([string])
nlp.tagger = Tagger.blank(nlp.vocab, Tagger.default_templates())
nlp.parser = Parser.from_dir(dep_model_dir, nlp.vocab.strings, ArcEager)
nlp.entity = Parser.from_dir(ner_model_dir, nlp.vocab.strings,
BiluoPushDown)
print("Itn.\tP.Loss\tUAS\tNER F.\tTag %\tToken %")
for itn in range(n_iter):
scorer = Scorer()
loss = 0
for raw_text, sents in gold_tuples:
if gold_preproc:
raw_text = None
else:
sents = _merge_sents(sents)
for annot_tuples, ctnt in sents:
if len(annot_tuples[1]) == 1:
continue
score_model(scorer,
nlp,
raw_text,
annot_tuples,
verbose=verbose if itn >= 2 else False)
if raw_text is None:
words = add_noise(annot_tuples[1], corruption_level)
tokens = nlp.tokenizer.tokens_from_list(words)
else:
raw_text = add_noise(raw_text, corruption_level)
tokens = nlp.tokenizer(raw_text)
nlp.tagger(tokens)
gold = GoldParse(tokens, annot_tuples)
if not gold.is_projective:
raise Exception("Non-projective sentence in training: %s" %
annot_tuples)
loss += nlp.parser.train(tokens, gold)
nlp.entity.train(tokens, gold)
nlp.tagger.train(tokens, gold.tags)
random.shuffle(gold_tuples)
print('%d:\t%d\t%.3f\t%.3f\t%.3f\t%.3f' %
(itn, loss, scorer.uas, scorer.ents_f, scorer.tags_acc,
scorer.token_acc))
print('end training')
nlp.end_training(model_dir)
print('done')
def train(Language, gold_tuples, model_dir, n_iter=15, feat_set=u'basic',
seed=0, gold_preproc=False, n_sents=0, corruption_level=0,
beam_width=1, verbose=False,
use_orig_arc_eager=False, pseudoprojective=False):
dep_model_dir = path.join(model_dir, 'deps')
ner_model_dir = path.join(model_dir, 'ner')
pos_model_dir = path.join(model_dir, 'pos')
if path.exists(dep_model_dir):
shutil.rmtree(dep_model_dir)
if path.exists(ner_model_dir):
shutil.rmtree(ner_model_dir)
if path.exists(pos_model_dir):
shutil.rmtree(pos_model_dir)
os.mkdir(dep_model_dir)
os.mkdir(ner_model_dir)
os.mkdir(pos_model_dir)
if pseudoprojective:
# preprocess training data here before ArcEager.get_labels() is called
gold_tuples = PseudoProjectivity.preprocess_training_data(gold_tuples)
Config.write(dep_model_dir, 'config', features=feat_set, seed=seed,
labels=ArcEager.get_labels(gold_tuples),
beam_width=beam_width,projectivize=pseudoprojective)
Config.write(ner_model_dir, 'config', features='ner', seed=seed,
labels=BiluoPushDown.get_labels(gold_tuples),
beam_width=0)
if n_sents > 0:
gold_tuples = gold_tuples[:n_sents]
nlp = Language(data_dir=model_dir, tagger=False, parser=False, entity=False)
if nlp.lang == 'de':
nlp.vocab.morphology.lemmatizer = lambda string,pos: set([string])
nlp.tagger = Tagger.blank(nlp.vocab, Tagger.default_templates())
nlp.parser = Parser.from_dir(dep_model_dir, nlp.vocab.strings, ArcEager)
nlp.entity = Parser.from_dir(ner_model_dir, nlp.vocab.strings, BiluoPushDown)
print("Itn.\tP.Loss\tUAS\tNER F.\tTag %\tToken %")
for itn in range(n_iter):
scorer = Scorer()
loss = 0
for raw_text, sents in gold_tuples:
if gold_preproc:
raw_text = None
else:
sents = _merge_sents(sents)
for annot_tuples, ctnt in sents:
if len(annot_tuples[1]) == 1:
continue
score_model(scorer, nlp, raw_text, annot_tuples,
verbose=verbose if itn >= 2 else False)
if raw_text is None:
words = add_noise(annot_tuples[1], corruption_level)
tokens = nlp.tokenizer.tokens_from_list(words)
else:
raw_text = add_noise(raw_text, corruption_level)
tokens = nlp.tokenizer(raw_text)
nlp.tagger(tokens)
gold = GoldParse(tokens, annot_tuples)
if not gold.is_projective:
raise Exception("Non-projective sentence in training: %s" % annot_tuples[1])
loss += nlp.parser.train(tokens, gold)
nlp.entity.train(tokens, gold)
nlp.tagger.train(tokens, gold.tags)
random.shuffle(gold_tuples)
print('%d:\t%d\t%.3f\t%.3f\t%.3f\t%.3f' % (itn, loss, scorer.uas, scorer.ents_f,
scorer.tags_acc,
scorer.token_acc))
print('end training')
nlp.end_training(model_dir)
print('done')
def main(lang_name, train_loc, dev_loc, model_dir, clusters_loc=None):
LangClass = spacy.util.get_lang_class(lang_name)
train_sents = list(read_conllx(train_loc))
train_sents = PseudoProjectivity.preprocess_training_data(train_sents)
actions = ArcEager.get_actions(gold_parses=train_sents)
features = get_templates('basic')
model_dir = pathlib.Path(model_dir)
if not model_dir.exists():
model_dir.mkdir()
if not (model_dir / 'deps').exists():
(model_dir / 'deps').mkdir()
if not (model_dir / 'pos').exists():
(model_dir / 'pos').mkdir()
with (model_dir / 'deps' / 'config.json').open('wb') as file_:
file_.write(
json.dumps(
{'pseudoprojective': True, 'labels': actions, 'features': features}).encode('utf8'))
vocab = LangClass.Defaults.create_vocab()
if not (model_dir / 'vocab').exists():
(model_dir / 'vocab').mkdir()
else:
if (model_dir / 'vocab' / 'strings.json').exists():
with (model_dir / 'vocab' / 'strings.json').open() as file_:
vocab.strings.load(file_)
if (model_dir / 'vocab' / 'lexemes.bin').exists():
vocab.load_lexemes(model_dir / 'vocab' / 'lexemes.bin')
if clusters_loc is not None:
clusters_loc = pathlib.Path(clusters_loc)
with clusters_loc.open() as file_:
for line in file_:
try:
cluster, word, freq = line.split()
except ValueError:
continue
lex = vocab[word]
lex.cluster = int(cluster[::-1], 2)
# Populate vocab
for _, doc_sents in train_sents:
for (ids, words, tags, heads, deps, ner), _ in doc_sents:
for word in words:
_ = vocab[word]
for dep in deps:
_ = vocab[dep]
for tag in tags:
_ = vocab[tag]
if vocab.morphology.tag_map:
for tag in tags:
assert tag in vocab.morphology.tag_map, repr(tag)
tagger = Tagger(vocab)
parser = DependencyParser(vocab, actions=actions, features=features, L1=0.0)
for itn in range(30):
loss = 0.
for _, doc_sents in train_sents:
for (ids, words, tags, heads, deps, ner), _ in doc_sents:
doc = Doc(vocab, words=words)
gold = GoldParse(doc, tags=tags, heads=heads, deps=deps)
tagger(doc)
loss += parser.update(doc, gold, itn=itn)
doc = Doc(vocab, words=words)
tagger.update(doc, gold)
random.shuffle(train_sents)
scorer = score_model(vocab, tagger, parser, read_conllx(dev_loc))
print('%d:\t%.3f\t%.3f\t%.3f' % (itn, loss, scorer.uas, scorer.tags_acc))
nlp = Language(vocab=vocab, tagger=tagger, parser=parser)
nlp.end_training(model_dir)
scorer = score_model(vocab, tagger, parser, read_conllx(dev_loc))
print('%d:\t%.3f\t%.3f\t%.3f' % (itn, scorer.uas, scorer.las, scorer.tags_acc))