def __init__(self,
h_max_length=20,
b_max_length=200,
trainable=False,
lstm_layers=2,
mlp_layers=1,
num_neurons=[128, 128, 32],
share_parameters=True,
average_pooling=False,
optimizer=tf.train.AdamOptimizer,
learning_rate=0.001,
batch_size=128,
activation=tf.nn.relu,
initializer=he_init,
num_epoch=20,
batch_norm_momentum=None,
dropout_rate=None,
max_check_without_progress=20,
show_progress=10,
tensorboard_logdir=None,
random_state=None,
embedding=None,
l2_lambda=0.01):
LSTM.__init__(self, h_max_length, b_max_length, trainable, lstm_layers,
mlp_layers, num_neurons, share_parameters,
average_pooling, optimizer, learning_rate, batch_size,
activation, initializer, num_epoch, batch_norm_momentum,
dropout_rate, max_check_without_progress, show_progress,
tensorboard_logdir, random_state, embedding, l2_lambda)
def __init__(self,
h_max_length=20,
b_max_length=200,
trainable=False,
lstm_layers=2,
mlp_layers=1,
num_neurons=[128, 128, 32],
share_parameters=True,
average_pooling=False,
optimizer=tf.train.AdamOptimizer,
learning_rate=0.001,
batch_size=128,
activation=tf.nn.relu,
initializer=he_init,
num_epoch=100,
batch_norm_momentum=None,
dropout_rate=None,
max_check_without_progress=10,
show_progress=10,
tensorboard_logdir=None,
random_state=None,
embedding=None,
l2_lambda=0.01,
vocab_size=None,
n_outputs=3,
pos_weight=None):
LSTM.__init__(self, h_max_length, b_max_length, trainable, lstm_layers,
mlp_layers, num_neurons, share_parameters,
average_pooling, optimizer, learning_rate, batch_size,
activation, initializer, num_epoch, batch_norm_momentum,
dropout_rate, max_check_without_progress, show_progress,
tensorboard_logdir, random_state, embedding, l2_lambda,
vocab_size)
self.mlp_layers = len(num_neurons) - 2
self.vocab_size = vocab_size
self.embedding_size = 300 + dim_fasttext
self.n_outputs = n_outputs
self.pos_weight = pos_weight
self._graph = None
self._classes = None
self._session = None
self.logger = LogHelper.get_logger(self.__class__.__name__)
if self.embedding is None and self.vocab_size is None:
raise Exception("Either embedding or vocab_size must be set!")
def __init__(self,
h_max_length=20,
b_max_length=200,
trainable=False,
lstm_layers=2,
mlp_layers=1,
num_neurons=[128, 128, 32],
share_parameters=True,
average_pooling=False,
optimizer=tf.train.AdamOptimizer,
learning_rate=0.001,
batch_size=128,
activation=tf.nn.relu,
initializer=he_init,
num_epoch=40,
batch_norm_momentum=None,
dropout_rate=None,
max_check_without_progress=20,
show_progress=10,
tensorboard_logdir=None,
random_state=None,
embedding=None,
l2_lambda=0.01,
word_output_size=64,
sent_output_size=64,
vocab_size=None):
LSTM.__init__(self, h_max_length, b_max_length, trainable, lstm_layers,
mlp_layers, num_neurons, share_parameters,
average_pooling, optimizer, learning_rate, batch_size,
activation, initializer, num_epoch, batch_norm_momentum,
dropout_rate, max_check_without_progress, show_progress,
tensorboard_logdir, random_state, embedding, l2_lambda,
vocab_size)
self.word_output_size = word_output_size
self.sent_output_size = sent_output_size
self.vocab_size = vocab_size
self.embedding_size = 100
if self.embedding is None and self.vocab_size is None:
raise Exception("Either embedding or vocab_size must be setted!")