def exe(word_vectors_file, vector_preloaded_path, test_path, sent,
hidden_sizes, maxlen, mix):
global word_vectors, vocabs
if not maxlen:
maxlen = properties.maxlen
if word_vectors is None or vocabs is None:
word_vectors, vocabs = utils.loadWordVectors(word_vectors_file,
vector_preloaded_path)
if sent:
if sent:
test_x = [utils.make_sentence_idx(vocabs, sent.lower(), maxlen)]
test_y = [1]
else:
#auto test path_file
test_x, test_y = utils.load_file(test_path)
if mix is 'Y':
combined = LSTM_CNN(word_vectors, hidden_sizes=hidden_sizes)
errors = combined.build_test_model((test_x, test_y, maxlen))
else:
lstm = Model(word_vectors, hidden_sizes=hidden_sizes)
errors = lstm.build_test_model((test_x, test_y, maxlen))
if sent:
pred = errors
if pred:
print "sentiment is positive"
else:
print "sentiment is negative"
elif errors:
print("Accuracy of test is: %.5f %" % (1 - errors) * 100)
def exe(word_vectors_file, vector_preloaded_path, train_path, dev_path,
test_path, hsi, hso, maxlen, pep, fep, ppat, fpat, plr, flr, mix):
global word_vectors, vocabs
if os.path.exists(train_path) and os.path.exists(
dev_path) and os.path.exists(test_path):
train = utils.load_file(train_path)
dev = utils.load_file(dev_path)
test = utils.load_file(test_path)
else:
raise NotImplementedError()
if word_vectors is None or vocabs is None:
word_vectors, vocabs = utils.loadWordVectors(word_vectors_file,
vector_preloaded_path)
if not maxlen:
maxlen = properties.maxlen
lstm = Model(word_vectors,
hidden_sizes=[hsi, hso],
epochs=pep,
patience=ppat,
learning_rate=plr)
lstm_params = lstm.train(train, dev, test, maxlen)
if mix is 'Y':
combined = LSTM_CNN(word_vectors,
hidden_sizes=[hsi, hso],
epochs=fep,
lstm_params=lstm_params)
combined.train(train, dev, test, maxlen)
# Training
# ==================================================
with tf.Graph().as_default():
session_conf = tf.ConfigProto(allow_soft_placement=True,
log_device_placement=False)
sess = tf.Session(config=session_conf)
with sess.as_default():
# embed()
if (MODEL_TO_RUN == 0):
model = CNN_LSTM(x_train.shape[1], y_train.shape[1],
len(vocab_processor.vocabulary_), embedding_dim,
filter_sizes, num_filters, l2_reg_lambda)
elif (MODEL_TO_RUN == 1):
model = LSTM_CNN(x_train.shape[1], y_train.shape[1],
len(vocab_processor.vocabulary_), embedding_dim,
filter_sizes, num_filters, l2_reg_lambda)
elif (MODEL_TO_RUN == 2):
model = CNN(x_train.shape[1], y_train.shape[1],
len(vocab_processor.vocabulary_), embedding_dim,
filter_sizes, num_filters, l2_reg_lambda)
elif (MODEL_TO_RUN == 3):
model = LSTM(x_train.shape[1], y_train.shape[1],
len(vocab_processor.vocabulary_), embedding_dim)
else:
print
"PLEASE CHOOSE A VALID MODEL!\n0 = CNN_LSTM\n1 = LSTM_CNN\n2 = CNN\n3 = LSTM\n"
exit()
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
with open("train_large.txt") as train_fnames:
for line in train_fnames:
train_f.append(line.split(" ")[0])
with open("test_large.txt") as test_fnames:
for line in test_fnames:
test_f.append(line.split(" ")[0])
# setup model
lstm_model = Basic_LSTM(num_units=80,
learning_rate=0.001,
dropout_rate=0.5,
debug=False)
lstm_cnn_model = LSTM_CNN(num_units=100,
learning_rate=0.001,
dropout_rate=0.5,
cnn_args={
'filter_size': 30,
'filter_num': 48
},
debug=False)
bilstm_model = BiLSTM(learning_rate=0.001, dropout_rate=0.5, debug=False)
deep_cnn_lstm_model = DeepCNN_LSTM(dropout_rate=0.0, reg_constant=0.01)
deeper_cnn_lstm_model = DeeperCNN_LSTM(dropout_rate=0.0, reg_constant=0.01)
# select which one to use
model = deep_cnn_lstm_model
# data config
D_config = {
"batch_size": 90,
"shuffle": False,
"balance": False,
dev_sample_index = -1 * int(config.dev_size * float(len(y)))
x_train, x_dev = x_shuffled[:dev_sample_index], x_shuffled[
dev_sample_index:]
y_train, y_dev = y_shuffled[:dev_sample_index], y_shuffled[
dev_sample_index:]
print("Vocabulary Size: {:d}".format(len(vocabulary)))
print("Train/Dev split: {:d}/{:d}".format(len(y_train), len(y_dev)))
# model build
print('{}'.format('#' * 30))
print('model build')
model = LSTM_CNN(
sequence_length=8,
num_classes=10, # 여기서 class 개수를 수정해야 한다
vocab_size=803087,
embedding_size=config.embedding_dim,
filter_sizes=list(map(int, config.filter_sizes.split(","))),
num_filters=config.num_filters,
l2_reg_lambda=config.l2_reg_lambda,
num_hidden=config.hiddensize)
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(config.lr)
grads_and_vars = optimizer.compute_gradients(model.loss)
train_op = optimizer.apply_gradients(grads_and_vars,
global_step=global_step)
print('{}'.format('#' * 30))
print('sess open')
sesstime = time.time()
with tf.Graph().as_default():
session_conf = tf.ConfigProto(allow_soft_placement=True,
log_device_placement=False)
sess = tf.Session(config=session_conf)
with sess.as_default():
#embed()
if (MODEL_TO_RUN == 0):
model = CNN_LSTM(x_train.shape[1], y_train.shape[1],
len(vocab_processor.vocabulary_), embedding_dim,
filter_sizes, num_filters, l2_reg_lambda)
elif (MODEL_TO_RUN == 1):
model = LSTM_CNN(max_seq_legth,
1,
n_symbols,
embedding_dim,
filter_sizes,
num_filters,
l2_reg_lambda,
weight=embedding_weights)
elif (MODEL_TO_RUN == 2):
model = CNN(x_train.shape[1], y_train.shape[1],
len(vocab_processor.vocabulary_), embedding_dim,
filter_sizes, num_filters, l2_reg_lambda)
elif (MODEL_TO_RUN == 3):
model = LSTM(x_train.shape[1], y_train.shape[1],
len(vocab_processor.vocabulary_), embedding_dim)
else:
print(
"PLEASE CHOOSE A VALID MODEL!\n0 = CNN_LSTM\n1 = LSTM_CNN\n2 = CNN\n3 = LSTM\n"
)
exit()
print(y_train.shape)
# Training
# ==================================================
with tf.Graph().as_default():
session_conf = tf.ConfigProto(allow_soft_placement=True,
log_device_placement=False)
sess = tf.Session(config=session_conf)
with sess.as_default():
# embed()
if (MODEL_TO_RUN == 0):
model = CNN_LSTM(x_train.shape[1], y_train.shape[1], embedding_dim,
filter_sizes, num_filters, num_hidden)
elif (MODEL_TO_RUN == 1):
model = LSTM_CNN(x_train.shape[1], y_train.shape[1], 21,
embedding_dim, filter_sizes, num_filters,
l2_reg_lambda)
elif (MODEL_TO_RUN == 2):
model = CNN(x_train.shape[1], y_train.shape[1], 26, embedding_dim,
filter_sizes, num_filters, l2_reg_lambda)
elif (MODEL_TO_RUN == 3):
model = LSTM(x_train.shape[1], y_train.shape[1], 26, embedding_dim)
else:
print
"PLEASE CHOOSE A VALID MODEL!\n0 = CNN_LSTM\n1 = LSTM_CNN\n2 = CNN\n3 = LSTM\n"
exit()
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(1e-3)
grads_and_vars = optimizer.compute_gradients(model.loss)
def train(x_train, y_train, vocab_processor, x_dev, y_dev, embedding):
# Training
# ==================================================
with tf.Graph().as_default():
session_conf = tf.ConfigProto(
allow_soft_placement=FLAGS.allow_soft_placement,
log_device_placement=FLAGS.log_device_placement)
sess = tf.Session(config=session_conf)
with sess.as_default():
# cnn = TextCNN(
# sequence_length=x_train.shape[1],
# num_classes=y_train.shape[1],
# vocab_size=len(vocab_processor.vocabulary_),
# embedding_size=FLAGS.embedding_dim,
# filter_sizes=list(map(int, FLAGS.filter_sizes.split(","))),
# num_filters=FLAGS.num_filters,
# l2_reg_lambda=FLAGS.l2_reg_lambda)
lstm_cnn = LSTM_CNN(x_train.shape[1],
y_train.shape[1],
len(vocab_processor.vocabulary_),
embedding_dim = FLAGS.embedding_dim,
filter_sizes= list(map(int, FLAGS.filter_sizes.split(","))),
num_filters=FLAGS.num_filters,
l2_reg_lambda=FLAGS.l2_reg_lambda)
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(1e-3)
grads_and_vars = optimizer.compute_gradients(lstm_cnn.loss)
train_op = optimizer.apply_gradients(grads_and_vars, global_step=global_step)
# Keep track of gradient values and sparsity (optional)
grad_summaries = []
for g, v in grads_and_vars:
if g is not None:
grad_hist_summary = tf.summary.histogram("{}/grad/hist".format(v.name), g)
sparsity_summary = tf.summary.scalar("{}/grad/sparsity".format(v.name), tf.nn.zero_fraction(g))
grad_summaries.append(grad_hist_summary)
grad_summaries.append(sparsity_summary)
grad_summaries_merged = tf.summary.merge(grad_summaries)
# Output directory for models and summaries
timestamp = str(int(time.time()))
out_dir = os.path.abspath(os.path.join(os.path.curdir, "runs", timestamp))
print("Writing to {}\n".format(out_dir))
# Summaries for loss and accuracy
loss_summary = tf.summary.scalar("loss", lstm_cnn.loss)
acc_summary = tf.summary.scalar("accuracy", lstm_cnn.accuracy)
# Train Summaries
train_summary_op = tf.summary.merge([loss_summary, acc_summary, grad_summaries_merged])
train_summary_dir = os.path.join(out_dir, "summaries", "train")
train_summary_writer = tf.summary.FileWriter(train_summary_dir, sess.graph)
# Dev summaries
dev_summary_op = tf.summary.merge([loss_summary, acc_summary])
dev_summary_dir = os.path.join(out_dir, "summaries", "dev")
dev_summary_writer = tf.summary.FileWriter(dev_summary_dir, sess.graph)
# Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
checkpoint_prefix = os.path.join(checkpoint_dir, "model")
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
saver = tf.train.Saver(tf.global_variables(), max_to_keep=FLAGS.num_checkpoints)
# Write vocabulary
vocab_processor.save(os.path.join(out_dir, "vocab"))
# Initialize all variables
init = tf.global_variables_initializer()
sess.run(lstm_cnn.embedding_init, feed_dict={lstm_cnn.embedding_placeholder: embedding})
sess.run(init)
def train_step(x_batch, y_batch):
"""
A single training step
"""
feed_dict = {
lstm_cnn.input_x: x_batch,
lstm_cnn.input_y: y_batch,
lstm_cnn.dropout_keep_prob: FLAGS.dropout_keep_prob
}
_, step, summaries, loss, accuracy = sess.run(
[train_op, global_step, train_summary_op, lstm_cnn.loss, lstm_cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
train_summary_writer.add_summary(summaries, step)
def dev_step(x_batch, y_batch, writer=None):
"""
Evaluates model on a dev set
"""
feed_dict = {
lstm_cnn.input_x: x_batch,
lstm_cnn.input_y: y_batch,
lstm_cnn.dropout_keep_prob: 1.0
}
step, summaries, loss, accuracy = sess.run(
[global_step, dev_summary_op, lstm_cnn.loss, lstm_cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
if writer:
writer.add_summary(summaries, step)
# Generate batches
batches = data_helpers.batch_iter(
list(zip(x_train, y_train)), FLAGS.batch_size, FLAGS.num_epochs)
# Training loop. For each batch...
for batch in batches:
x_batch, y_batch = zip(*batch)
train_step(x_batch, y_batch)
current_step = tf.train.global_step(sess, global_step)
if current_step % FLAGS.evaluate_every == 0:
print("\nEvaluation:")
dev_step(x_dev, y_dev, writer=dev_summary_writer)
print("")
if current_step % FLAGS.checkpoint_every == 0:
path = saver.save(sess, checkpoint_prefix, global_step=current_step)
print("Saved model checkpoint to {}\n".format(path))
y_tr = y_train[train_index].astype(np.int32)
y_val = y_train[val_index].astype(np.int32)
print("Validation shape: {}".format(X_val.shape))
print("Training shape: {}".format(X_tr.shape))
with tf.Graph().as_default():
session_conf = tf.ConfigProto(allow_soft_placement=True,
log_device_placement=False)
sess = tf.Session(config=session_conf)
with sess.as_default():
#embed()
if (MODEL_TO_RUN == 0):
model = CNN_LSTM(seq_legth, num_classes, embedding_dim,
filter_sizes, num_filters)
elif (MODEL_TO_RUN == 1):
model = LSTM_CNN(seq_legth, num_classes, embedding_dim,
filter_sizes, num_filters, num_hidden)
# elif (MODEL_TO_RUN == 2):
# model = CNN(x_train.shape[1],y_train.shape[1],len(vocab_processor.vocabulary_),
# embedding_dim,filter_sizes,num_filters,l2_reg_lambda)
# elif (MODEL_TO_RUN == 3):
# model = LSTM(x_train.shape[1],y_train.shape[1],len(vocab_processor.vocabulary_),embedding_dim)
else:
print(
"PLEASE CHOOSE A VALID MODEL!\n0 = CNN_LSTM\n1 = LSTM_CNN\n2 = CNN\n3 = LSTM\n"
)
exit()
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer()
grads_and_vars = optimizer.compute_gradients(model.loss)
x_shuffled = x[shuffle_indices]
y_shuffled = y[shuffle_indices]
dev_sample_index = -1 * int(config.DEV_SIZE * float(len(y)))
x_train, x_dev = x_shuffled[:dev_sample_index], x_shuffled[dev_sample_index:]
y_train, y_dev = y_shuffled[:dev_sample_index], y_shuffled[dev_sample_index:]
logger.info("Vocabulary Size: {:d}".format(len(vocab_processor.vocabulary_)))
logger.info("Train/Dev split: {:d}/{:d}".format(len(y_train), len(y_dev)))
with tf.Graph().as_default():
session_conf = tf.ConfigProto(allow_soft_placement=True,
log_device_placement=False)
sess = tf.Session(config=session_conf)
with sess.as_default():
model = LSTM_CNN(x_train.shape[1], y_train.shape[1],
len(vocab_processor.vocabulary_), embedding_dim,
config.FILTER_SIZE, config.NUM_FILTERS,
config.L2_REG_LAMBDA)
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(1e-3)
grads_and_vars = optimizer.compute_gradients(model.loss)
train_op = optimizer.apply_gradients(grads_and_vars,
global_step=global_step)
# Keep track of gradient values and sparsity (optional)
grad_summaries = []
for g, v in grads_and_vars:
if g is not None:
grad_hist_summary = tf.summary.histogram(
"{}/grad/hist".format(v.name), g)