def __init__(self,
language_model,
word_embedding_dim=100,
char_embedding_dim=25,
word_lstm_size=100,
char_lstm_size=25,
fc_dim=100,
dropout=0.5,
embeddings=None,
use_char=True,
use_crf=True,
initial_vocab=None,
lower=False,
optimizer='adam',
nn_type='GRU',
input_size=300):
self.model = None
self.p = VectorTransformer(language_model, use_char=use_char)
self.tagger = None
self.word_embedding_dim = word_embedding_dim
self.char_embedding_dim = char_embedding_dim
self.word_lstm_size = word_lstm_size
self.char_lstm_size = char_lstm_size
self.fc_dim = fc_dim
self.dropout = dropout
self.embeddings = embeddings
self.use_char = use_char
self.use_crf = use_crf
self.initial_vocab = initial_vocab
self.optimizer = optimizer
self.lower = lower
self.nn_type = nn_type
self.input_size = input_size
def fit(self,
x_train,
y_train,
fasttext_path,
x_valid=None,
y_valid=None,
epochs=1,
batch_size=32,
verbose=1,
callbacks=None,
shuffle=True):
"""Fit the model for a fixed number of epochs.
Args:
x_train: list of training model.
y_train: list of training target (label) model.
x_valid: list of validation model.
y_valid: list of validation target (label) model.
batch_size: Integer.
Number of samples per gradient update.
If unspecified, `batch_size` will default to 32.
epochs: Integer. Number of epochs to train the model.
verbose: Integer. 0, 1, or 2. Verbosity mode.
0 = silent, 1 = progress bar, 2 = one line per epoch.
callbacks: List of `keras.callbacks.Callback` instances.
List of callbacks to apply during training.
shuffle: Boolean (whether to shuffle the training model
before each epoch). `shuffle` will default to True.
"""
p = VectorTransformer(fasttext_path, use_char=self.use_char)
p.fit(x_train, y_train)
model = BiLSTMCRF(num_labels=p.label_size,
word_lstm_size=self.word_lstm_size,
char_lstm_size=self.char_lstm_size,
fc_dim=self.fc_dim,
dropout=self.dropout,
use_char=self.use_char,
use_crf=self.use_crf,
nn_type=self.nn_type,
input_size=self.input_size)
model, loss = model.build()
model.compile(loss=loss, optimizer=self.optimizer)
trainer = Trainer(model, preprocessor=p)
trainer.train(x_train,
y_train,
x_valid,
y_valid,
epochs=epochs,
batch_size=batch_size,
verbose=verbose,
callbacks=callbacks,
shuffle=shuffle)
self.p = p
self.model = model
def load(cls, model_path, fasttext):
self = cls()
weights_file = os.path.join(model_path, "weights.pkl")
params_file = os.path.join(model_path, "params.pkl")
preprocessor_file = os.path.join(model_path, "preprocessor.pkl")
self.p = VectorTransformer.load(preprocessor_file, fasttext)
self.model = load_model(weights_file, params_file)
return self
class Sequence(object):
def __init__(self,
language_model,
word_embedding_dim=100,
char_embedding_dim=25,
word_lstm_size=100,
char_lstm_size=25,
fc_dim=100,
dropout=0.5,
embeddings=None,
use_char=True,
use_crf=True,
initial_vocab=None,
lower=False,
optimizer='adam',
nn_type='GRU',
input_size=300):
self.model = None
self.p = VectorTransformer(language_model, use_char=use_char)
self.tagger = None
self.word_embedding_dim = word_embedding_dim
self.char_embedding_dim = char_embedding_dim
self.word_lstm_size = word_lstm_size
self.char_lstm_size = char_lstm_size
self.fc_dim = fc_dim
self.dropout = dropout
self.embeddings = embeddings
self.use_char = use_char
self.use_crf = use_crf
self.initial_vocab = initial_vocab
self.optimizer = optimizer
self.lower = lower
self.nn_type = nn_type
self.input_size = input_size
def fit(self,
x_train,
y_train,
x_valid=None,
y_valid=None,
epochs=1,
batch_size=32,
verbose=1,
callbacks=None,
shuffle=True):
"""Fit the model for a fixed number of epochs.
Args:
x_train: list of training model.
y_train: list of training target (label) model.
x_valid: list of validation model.
y_valid: list of validation target (label) model.
batch_size: Integer.
Number of samples per gradient update.
If unspecified, `batch_size` will default to 32.
epochs: Integer. Number of epochs to train the model.
verbose: Integer. 0, 1, or 2. Verbosity mode.
0 = silent, 1 = progress bar, 2 = one line per epoch.
callbacks: List of `keras.callbacks.Callback` instances.
List of callbacks to apply during training.
shuffle: Boolean (whether to shuffle the training model
before each epoch). `shuffle` will default to True.
"""
self.p.fit(x_train, y_train)
model = BiLSTMCRF(num_labels=self.p.label_size,
word_lstm_size=self.word_lstm_size,
char_lstm_size=self.char_lstm_size,
fc_dim=self.fc_dim,
dropout=self.dropout,
use_char=self.use_char,
use_crf=self.use_crf,
nn_type=self.nn_type,
input_size=self.input_size)
model, loss = model.build()
model.compile(loss=loss, optimizer=self.optimizer)
trainer = Trainer(model, preprocessor=self.p)
trainer.train(x_train,
y_train,
x_valid,
y_valid,
epochs=epochs,
batch_size=batch_size,
verbose=verbose,
callbacks=callbacks,
shuffle=shuffle)
self.model = model
#self.model = trainer.best_model
#print("Best model report\n")
#print(trainer.best_model_report)
def score(self, x_test, y_test):
"""Returns the f1-micro score on the given test model and labels.
Args:
x_test : array-like, shape = (n_samples, sent_length)
Test samples.
y_test : array-like, shape = (n_samples, sent_length)
True labels for x.
Returns:
score : float, f1-micro score.
"""
if self.model:
x_test = self.p.transform(x_test)
lengths = map(len, y_test)
y_pred = self.model.predict(x_test)
y_pred = self.p.inverse_transform(y_pred, lengths)
score = f1_score(y_test, y_pred)
return score
else:
raise OSError('Could not find a model. Call load(dir_path).')
def predict_to_iob(self, input_path, output_path):
input_file = open(input_path, 'r')
output_file = open(output_path, 'w')
sentence, iob_lines, true_labels = [], [], []
for input_line in input_file:
if "DOCSTART CONFIG" in input_line or "DOCSTART FILE" in input_line:
continue
if input_line == '\n':
x_test = self.p.transform([sentence])
lengths = map(len, [true_labels])
y_pred = self.model.predict(x_test)
y_pred = self.p.inverse_transform(y_pred, lengths)
for prediction, iob_line in zip(y_pred[0], iob_lines):
output_line = ''
for pos in iob_line:
output_line += pos + '\t'
output_line += prediction
output_file.write(output_line + '\n')
output_file.write('\n')
sentence = []
iob_lines = []
else:
iob_line = input_line.split(sep='\t')[0:3]
iob_lines.append(iob_line)
sentence.append(input_line.split(sep='\t')[0])
true_labels.append(input_line.split(sep='\t')[3])
def predict_sentence(self, sentence):
x_test = self.p.transform([sentence])
lengths = [len(sentence)]
y_pred = self.model.predict(x_test)
y_pred = self.p.inverse_transform(y_pred, lengths)
return y_pred[0]
def save(self, weights_file, params_file, preprocessor_file):
self.p.save(preprocessor_file)
save_model(self.model, weights_file, params_file)
@classmethod
def load(cls, model_path, fasttext):
self = cls(fasttext)
weights_file = os.path.join(model_path, "weights.pkl")
params_file = os.path.join(model_path, "params.pkl")
preprocessor_file = os.path.join(model_path, "preprocessor.pkl")
self.p = VectorTransformer.load(preprocessor_file, fasttext)
self.model = load_model(weights_file, params_file)
return self