def do_train(network, optimizer, learning_rate, batch_size, epoch_num,
label_type_second, num_stack, num_skip):
"""Run multi-task training. The target labels in the main task is
characters and those in the second task is 61 phones. The model is
evaluated by CER and PER with 39 phones.
Args:
network: network to train
optimizer: string, the name of optimizer.
ex.) adam, rmsprop
learning_rate: A float value, the initial learning rate
batch_size: int, the size of mini-batch
epoch_num: int, the epoch num to train
label_type_second: string, phone39 or phone48 or phone61
num_stack: int, the number of frames to stack
num_skip: int, the number of frames to skip
"""
# Load dataset
train_data = DataSet(data_type='train',
label_type_second=label_type_second,
batch_size=batch_size,
num_stack=num_stack, num_skip=num_skip,
is_sorted=True)
dev_data = DataSet(data_type='dev', label_type_second=label_type_second,
batch_size=batch_size,
num_stack=num_stack, num_skip=num_skip,
is_sorted=False)
test_data = DataSet(data_type='test', label_type_second='phone39',
batch_size=1,
num_stack=num_stack, num_skip=num_skip,
is_sorted=False)
# Tell TensorFlow that the model will be built into the default graph
with tf.Graph().as_default():
# Define placeholders
network.inputs = tf.placeholder(
tf.float32,
shape=[None, None, network.input_size],
name='input')
indices_pl = tf.placeholder(tf.int64, name='indices')
values_pl = tf.placeholder(tf.int32, name='values')
shape_pl = tf.placeholder(tf.int64, name='shape')
network.labels = tf.SparseTensor(indices_pl, values_pl, shape_pl)
indices_second_pl = tf.placeholder(tf.int64, name='indices_second')
values_second_pl = tf.placeholder(tf.int32, name='values_second')
shape_second_pl = tf.placeholder(tf.int64, name='shape_second')
network.labels_second = tf.SparseTensor(indices_second_pl,
values_second_pl,
shape_second_pl)
network.inputs_seq_len = tf.placeholder(tf.int64,
shape=[None],
name='inputs_seq_len')
network.keep_prob_input = tf.placeholder(tf.float32,
name='keep_prob_input')
network.keep_prob_hidden = tf.placeholder(tf.float32,
name='keep_prob_hidden')
# Add to the graph each operation
loss_op, logits_main, logits_second = network.compute_loss(
network.inputs,
network.labels,
network.labels_second,
network.inputs_seq_len,
network.keep_prob_input,
network.keep_prob_hidden)
train_op = network.train(loss_op,
optimizer=optimizer,
learning_rate_init=float(learning_rate),
is_scheduled=False)
decode_op_main, decode_op_second = network.decoder(
logits_main,
logits_second,
network.inputs_seq_len,
decode_type='beam_search',
beam_width=20)
ler_op_main, ler_op_second = network.compute_ler(
decode_op_main, decode_op_second,
network.labels, network.labels_second)
# Build the summary tensor based on the TensorFlow collection of
# summaries
summary_train = tf.summary.merge(network.summaries_train)
summary_dev = tf.summary.merge(network.summaries_dev)
# Add the variable initializer operation
init_op = tf.global_variables_initializer()
# Create a saver for writing training checkpoints
saver = tf.train.Saver(max_to_keep=None)
# Count total parameters
parameters_dict, total_parameters = count_total_parameters(
tf.trainable_variables())
for parameter_name in sorted(parameters_dict.keys()):
print("%s %d" % (parameter_name, parameters_dict[parameter_name]))
print("Total %d variables, %s M parameters" %
(len(parameters_dict.keys()),
"{:,}".format(total_parameters / 1000000)))
csv_steps, csv_loss_train, csv_loss_dev = [], [], []
csv_cer_train, csv_cer_dev = [], []
csv_per_train, csv_per_dev = [], []
# Create a session for running operation on the graph
with tf.Session() as sess:
# Instantiate a SummaryWriter to output summaries and the graph
summary_writer = tf.summary.FileWriter(
network.model_dir, sess.graph)
# Initialize parameters
sess.run(init_op)
# Make mini-batch generator
mini_batch_train = train_data.next_batch()
mini_batch_dev = dev_data.next_batch()
# Train model
iter_per_epoch = int(train_data.data_num / batch_size)
train_step = train_data.data_num / batch_size
if train_step != int(train_step):
iter_per_epoch += 1
max_steps = iter_per_epoch * epoch_num
start_time_train = time.time()
start_time_epoch = time.time()
start_time_step = time.time()
cer_dev_best = 1
for step in range(max_steps):
# Create feed dictionary for next mini batch (train)
inputs, labels_char_st, labels_phone_st, inputs_seq_len, input_names = mini_batch_train.__next__()
feed_dict_train = {
network.inputs: inputs,
network.labels: labels_char_st,
network.labels_second: labels_phone_st,
network.inputs_seq_len: inputs_seq_len,
network.keep_prob_input: network.dropout_ratio_input,
network.keep_prob_hidden: network.dropout_ratio_hidden,
network.lr: learning_rate
}
# Create feed dictionary for next mini batch (dev)
inputs, labels_char, labels_phone, inputs_seq_len, _ = mini_batch_dev.__next__()
feed_dict_dev = {
network.inputs: inputs,
network.labels: labels_char_st,
network.labels_second: labels_phone_st,
network.inputs_seq_len: inputs_seq_len,
network.keep_prob_input: network.dropout_ratio_input,
network.keep_prob_hidden: network.dropout_ratio_hidden
}
# Update parameters
sess.run(train_op, feed_dict=feed_dict_train)
if (step + 1) % 10 == 0:
# Compute loss
loss_train = sess.run(loss_op, feed_dict=feed_dict_train)
loss_dev = sess.run(loss_op, feed_dict=feed_dict_dev)
csv_steps.append(step)
csv_loss_train.append(loss_train)
csv_loss_dev.append(loss_dev)
# Change to evaluation mode
feed_dict_train[network.keep_prob_input] = 1.0
feed_dict_train[network.keep_prob_hidden] = 1.0
feed_dict_dev[network.keep_prob_input] = 1.0
feed_dict_dev[network.keep_prob_hidden] = 1.0
# Compute accuracy & update event file
cer_train, per_train, summary_str_train = sess.run(
[ler_op_main, ler_op_second, summary_train],
feed_dict=feed_dict_train)
cer_dev, per_dev, summary_str_dev = sess.run(
[ler_op_main, ler_op_second, summary_dev],
feed_dict=feed_dict_dev)
csv_cer_train.append(cer_train)
csv_cer_dev.append(cer_dev)
csv_per_train.append(per_train)
csv_per_dev.append(per_dev)
summary_writer.add_summary(summary_str_train, step + 1)
summary_writer.add_summary(summary_str_dev, step + 1)
summary_writer.flush()
duration_step = time.time() - start_time_step
print("Step % d: loss = %.3f (%.3f) / cer = %.4f (%.4f) / per = % .4f (%.4f) (%.3f min)" %
(step + 1, loss_train, loss_dev, cer_train, cer_dev,
per_train, per_dev, duration_step / 60))
sys.stdout.flush()
start_time_step = time.time()
# Save checkpoint and evaluate model per epoch
if (step + 1) % iter_per_epoch == 0 or (step + 1) == max_steps:
duration_epoch = time.time() - start_time_epoch
epoch = (step + 1) // iter_per_epoch
print('-----EPOCH:%d (%.3f min)-----' %
(epoch, duration_epoch / 60))
# Save model (check point)
checkpoint_file = join(network.model_dir, 'model.ckpt')
save_path = saver.save(
sess, checkpoint_file, global_step=epoch)
print("Model saved in file: %s" % save_path)
if epoch >= 10:
start_time_eval = time.time()
print('=== Dev Data Evaluation ===')
cer_dev_epoch = do_eval_cer(
session=sess,
decode_op=decode_op_main,
network=network,
dataset=dev_data,
is_multitask=True)
print(' CER: %f %%' % (cer_dev_epoch * 100))
per_dev_epoch = do_eval_per(
session=sess,
decode_op=decode_op_second,
per_op=ler_op_second,
network=network,
dataset=dev_data,
train_label_type=label_type_second,
is_multitask=True)
print(' PER: %f %%' % (per_dev_epoch * 100))
if cer_dev_epoch < cer_dev_best:
cer_dev_best = cer_dev_epoch
print('■■■ ↑Best Score (CER)↑ ■■■')
print('=== Test Data Evaluation ===')
cer_test = do_eval_cer(
session=sess,
decode_op=decode_op_main,
network=network,
dataset=test_data,
eval_batch_size=1,
is_multitask=True)
print(' CER: %f %%' % (cer_test * 100))
per_test = do_eval_per(
session=sess,
decode_op=decode_op_second,
per_op=ler_op_second,
network=network,
dataset=test_data,
train_label_type=label_type_second,
eval_batch_size=1,
is_multitask=True)
print(' PER: %f %%' % (per_test * 100))
duration_eval = time.time() - start_time_eval
print('Evaluation time: %.3f min' %
(duration_eval / 60))
start_time_epoch = time.time()
start_time_step = time.time()
duration_train = time.time() - start_time_train
print('Total time: %.3f hour' % (duration_train / 3600))
# Save train & dev loss, ler
save_loss(csv_steps, csv_loss_train, csv_loss_dev,
save_path=network.model_dir)
save_ler(csv_steps, csv_cer_train, csv_cer_dev,
save_path=network.model_dir)
save_ler(csv_steps, csv_per_train, csv_per_dev,
save_path=network.model_dir)
# Training was finished correctly
with open(join(network.model_dir, 'complete.txt'), 'w') as f:
f.write('')
def do_train(network, optimizer, learning_rate, batch_size, epoch_num, label_type, num_stack, num_skip):
"""Run training.
Args:
network: network to train
optimizer: string, the name of optimizer. ex.) adam, rmsprop
learning_rate: initial learning rate
batch_size: size of mini batch
epoch_num: epoch num to train
label_type: phone39 or phone48 or phone61 (+ character)
num_stack: int, the number of frames to stack
num_skip: int, the number of frames to skip
"""
# Load dataset
train_data = DataSet(data_type='train', label_type=label_type,
num_stack=num_stack, num_skip=num_skip,
is_sorted=True)
dev_data61 = DataSet(data_type='dev', label_type='phone61',
num_stack=num_stack, num_skip=num_skip,
is_sorted=False)
dev_data39 = DataSet(data_type='dev', label_type='phone39',
num_stack=num_stack, num_skip=num_skip,
is_sorted=False)
test_data = DataSet(data_type='test', label_type='phone39',
num_stack=num_stack, num_skip=num_skip,
is_sorted=False)
# Tell TensorFlow that the model will be built into the default graph
with tf.Graph().as_default():
# Define model
network.define()
# NOTE: define model under tf.Graph()
# Add to the graph each operation
loss_op = network.loss()
train_op = network.train(optimizer=optimizer,
learning_rate_init=learning_rate,
is_scheduled=False)
decode_op1, decode_op2 = network.decoder(decode_type='beam_search',
beam_width=20)
per_op1, per_op2 = network.ler(decode_op1, decode_op2)
# Build the summary tensor based on the TensorFlow collection of
# summaries
summary_train = tf.summary.merge(network.summaries_train)
summary_dev = tf.summary.merge(network.summaries_dev)
# Add the variable initializer operation
init_op = tf.global_variables_initializer()
# Create a saver for writing training checkpoints
saver = tf.train.Saver(max_to_keep=None)
# Count total parameters
parameters_dict, total_parameters = count_total_parameters(
tf.trainable_variables())
for parameter_name in sorted(parameters_dict.keys()):
print("%s %d" % (parameter_name, parameters_dict[parameter_name]))
print("Total %d variables, %s M parameters" %
(len(parameters_dict.keys()), "{:,}".format(total_parameters / 1000000)))
csv_steps = []
csv_train_loss = []
csv_dev_loss = []
# Create a session for running operation on the graph
with tf.Session() as sess:
# Instantiate a SummaryWriter to output summaries and the graph
summary_writer = tf.summary.FileWriter(
network.model_dir, sess.graph)
# Initialize parameters
sess.run(init_op)
# Train model
iter_per_epoch = int(train_data.data_num / batch_size)
if (train_data.data_num / batch_size) != int(train_data.data_num / batch_size):
iter_per_epoch += 1
max_steps = iter_per_epoch * epoch_num
start_time_train = time.time()
start_time_epoch = time.time()
start_time_step = time.time()
error_best = 1
for step in range(max_steps):
# Create feed dictionary for next mini batch (train)
inputs, labels_char, labels_phone, seq_len, _ = train_data.next_batch(
batch_size=batch_size)
indices_char, values_char, dense_shape_char = list2sparsetensor(
labels_char)
indices_phone, values_phone, dense_shape_phone = list2sparsetensor(
labels_phone)
feed_dict_train = {
network.inputs_pl: inputs,
network.label_indices_pl: indices_char,
network.label_values_pl: values_char,
network.label_shape_pl: dense_shape_char,
network.label_indices_pl2: indices_phone,
network.label_values_pl2: values_phone,
network.label_shape_pl2: dense_shape_phone,
network.seq_len_pl: seq_len,
network.keep_prob_input_pl: network.dropout_ratio_input,
network.keep_prob_hidden_pl: network.dropout_ratio_hidden,
network.lr_pl: learning_rate
}
# Create feed dictionary for next mini batch (dev)
inputs, labels_char, labels_phone, seq_len, _ = dev_data61.next_batch(
batch_size=batch_size)
indices_char, values_char, dense_shape_char = list2sparsetensor(
labels_char)
indices_phone, values_phone, dense_shape_phone = list2sparsetensor(
labels_phone)
feed_dict_dev = {
network.inputs_pl: inputs,
network.label_indices_pl: indices_char,
network.label_values_pl: values_char,
network.label_shape_pl: dense_shape_char,
network.label_indices_pl2: indices_phone,
network.label_values_pl2: values_phone,
network.label_shape_pl2: dense_shape_phone,
network.seq_len_pl: seq_len,
network.keep_prob_input_pl: network.dropout_ratio_input,
network.keep_prob_hidden_pl: network.dropout_ratio_hidden
}
# Update parameters & compute loss
_, loss_train = sess.run(
[train_op, loss_op], feed_dict=feed_dict_train)
loss_dev = sess.run(loss_op, feed_dict=feed_dict_dev)
csv_steps.append(step)
csv_train_loss.append(loss_train)
csv_dev_loss.append(loss_dev)
if (step + 1) % 10 == 0:
# Change feed dict for evaluation
feed_dict_train[network.keep_prob_input_pl] = 1.0
feed_dict_train[network.keep_prob_hidden_pl] = 1.0
feed_dict_dev[network.keep_prob_input_pl] = 1.0
feed_dict_dev[network.keep_prob_hidden_pl] = 1.0
# Compute accuracy & update event file
cer_train, per_train, summary_str_train = sess.run([per_op1, per_op2, summary_train],
feed_dict=feed_dict_train)
cer_dev, per_dev, summary_str_dev = sess.run([per_op1, per_op2, summary_dev],
feed_dict=feed_dict_dev)
summary_writer.add_summary(summary_str_train, step + 1)
summary_writer.add_summary(summary_str_dev, step + 1)
summary_writer.flush()
duration_step = time.time() - start_time_step
print('Step %d: loss = %.3f (%.3f) / cer = %.4f (%.4f) / per = %.4f (%.4f) (%.3f min)' %
(step + 1, loss_train, loss_dev, cer_train, cer_dev, per_train, per_dev, duration_step / 60))
sys.stdout.flush()
start_time_step = time.time()
# Save checkpoint and evaluate model per epoch
if (step + 1) % iter_per_epoch == 0 or (step + 1) == max_steps:
duration_epoch = time.time() - start_time_epoch
epoch = (step + 1) // iter_per_epoch
print('-----EPOCH:%d (%.3f min)-----' %
(epoch, duration_epoch / 60))
# Save model (check point)
checkpoint_file = join(network.model_dir, 'model.ckpt')
save_path = saver.save(
sess, checkpoint_file, global_step=epoch)
print("Model saved in file: %s" % save_path)
if epoch >= 10:
start_time_eval = time.time()
print('■Dev Data Evaluation:■')
error_epoch = do_eval_cer(session=sess,
decode_op=decode_op1,
network=network,
dataset=dev_data39,
eval_batch_size=1,
is_multitask=True)
do_eval_per(session=sess,
decode_op=decode_op2,
per_op=per_op2,
network=network,
dataset=dev_data39,
label_type=label_type,
eval_batch_size=1,
is_multitask=True)
if error_epoch < error_best:
error_best = error_epoch
print('■■■ ↑Best Score (CER)↑ ■■■')
print('■Test Data Evaluation:■')
do_eval_cer(session=sess,
decode_op=decode_op1,
network=network,
dataset=test_data,
eval_batch_size=1,
is_multitask=True)
do_eval_per(session=sess,
decode_op=decode_op2,
per_op=per_op2,
network=network,
dataset=test_data,
label_type=label_type,
eval_batch_size=1,
is_multitask=True)
duration_eval = time.time() - start_time_eval
print('Evaluation time: %.3f min' %
(duration_eval / 60))
start_time_epoch = time.time()
start_time_step = time.time()
duration_train = time.time() - start_time_train
print('Total time: %.3f hour' % (duration_train / 3600))
# Save train & dev loss
save_loss(csv_steps, csv_train_loss, csv_dev_loss,
save_path=network.model_dir)
# Training was finished correctly
with open(join(network.model_dir, 'complete.txt'), 'w') as f:
f.write('')