def eval(logdir='logdir/default/train1', queue=False):
# Load graph
model = Model(mode="test1", batch_size=hp.Test1.batch_size, queue=queue)
# Accuracy
acc_op = model.acc_net1()
# Loss
loss_op = model.loss_net1()
# Summary
summ_op = summaries(acc_op, loss_op)
session_conf = tf.ConfigProto(
allow_soft_placement=True,
device_count={
'CPU': 1,
'GPU': 0
},
)
with tf.Session(config=session_conf) as sess:
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
writer = tf.summary.FileWriter(logdir, sess.graph)
# Load trained model
sess.run(tf.global_variables_initializer())
model.load(sess, 'train1', logdir=logdir)
if queue:
summ, acc, loss = sess.run([summ_op, acc_op, loss_op])
else:
mfcc, ppg = get_batch(model.mode, model.batch_size)
summ, acc, loss = sess.run([summ_op, acc_op, loss_op],
feed_dict={
model.x_mfcc: mfcc,
model.y_ppgs: ppg
})
writer.add_summary(summ)
print("acc:", acc)
print("loss:", loss)
print('\n')
writer.close()
coord.request_stop()
coord.join(threads)
def eval(logdir, hparams):
# Load graph
model = Model(mode="test1", hparams=hparams)
# Accuracy
acc_op = model.acc_net1()
# Loss
loss_op = model.loss_net1()
# Summary
summ_op = summaries(acc_op, loss_op)
#session_conf = tf.ConfigProto(
# allow_soft_placement=True,
# device_count={'CPU': 1, 'GPU': 0},
#)
session_conf = tf.ConfigProto()
session_conf.gpu_options.per_process_gpu_memory_fraction = 0.9
with tf.Session(config=session_conf) as sess:
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
writer = tf.summary.FileWriter(logdir, sess.graph)
# Load trained model
sess.run(tf.global_variables_initializer())
model.load(sess, 'train1', logdir=logdir)
mfcc, ppg = get_batch(model.mode, model.batch_size)
summ, acc, loss = sess.run([summ_op, acc_op, loss_op],
feed_dict={
model.x_mfcc: mfcc,
model.y_ppgs: ppg
})
writer.add_summary(summ)
print("acc:", acc)
print("loss:", loss)
print('\n')
writer.close()
coord.request_stop()
coord.join(threads)
def eval(logdir='logdir/default/train1', queue=False):
# Load graph
model = Model(mode="test1", batch_size=hp.Test1.batch_size, queue=queue)
# Accuracy
acc_op = model.acc_net1()
# Loss
loss_op = model.loss_net1()
# Summary
summ_op = summaries(acc_op, loss_op)
session_conf = tf.ConfigProto(
allow_soft_placement=True,
device_count={'CPU': 1, 'GPU': 0},
)
with tf.Session(config=session_conf) as sess:
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
writer = tf.summary.FileWriter(logdir, sess.graph)
# Load trained model
sess.run(tf.global_variables_initializer())
model.load(sess, 'train1', logdir=logdir)
if queue:
summ, acc, loss = sess.run([summ_op, acc_op, loss_op])
else:
mfcc, ppg = get_batch(model.mode, model.batch_size)
summ, acc, loss = sess.run([summ_op, acc_op, loss_op], feed_dict={model.x_mfcc: mfcc, model.y_ppgs: ppg})
writer.add_summary(summ)
print("acc:", acc)
print("loss:", loss)
print('\n')
writer.close()
coord.request_stop()
coord.join(threads)
def train(logdir, hparams):
model = Model(mode="train1", hparams=hparams)
# Loss
loss_op = model.loss_net1()
# Accuracy
acc_op = model.acc_net1()
# Training Scheme
global_step = tf.Variable(0, name='global_step', trainable=False)
optimizer = tf.train.AdamOptimizer(learning_rate=hparams.Train1.lr)
with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
var_list = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'net/net1')
train_op = optimizer.minimize(loss_op, global_step=global_step, var_list=var_list)
# Summary
# for v in tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'net/net1'):
# tf.summary.histogram(v.name, v)
tf.summary.scalar('net1/train/loss', loss_op)
tf.summary.scalar('net1/train/acc', acc_op)
summ_op = tf.summary.merge_all()
#session_conf = tf.ConfigProto(
# gpu_options=tf.GPUOptions(
# allow_growth=True,
# ),
#)
session_conf=tf.ConfigProto()
session_conf.gpu_options.per_process_gpu_memory_fraction=0.9
# Training
with tf.Session(config=session_conf) as sess:
# Load trained model
sess.run(tf.global_variables_initializer())
model.load(sess, 'train1', logdir=logdir)
writer = tf.summary.FileWriter(logdir, sess.graph)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
for epoch in range(1, hparams.Train1.num_epochs + 1):
for step in range(model.num_batch):
mfcc, ppg = get_batch(model.mode, model.batch_size)
sess.run(train_op, feed_dict={model.x_mfcc: mfcc, model.y_ppgs: ppg})
# Write checkpoint files at every epoch
summ, gs = sess.run([summ_op, global_step], feed_dict={model.x_mfcc: mfcc, model.y_ppgs: ppg})
if epoch % hparams.Train1.save_per_epoch == 0:
tf.train.Saver().save(sess, '{}/epoch_{}_step_{}'.format(logdir, epoch, gs))
# Write eval accuracy at every epoch
with tf.Graph().as_default():
eval1.eval(logdir=logdir, hparams=hparams)
writer.add_summary(summ, global_step=gs)
writer.close()
coord.request_stop()
coord.join(threads)
def train(logdir='logdir/default/train1', queue=True):
model = Model(mode="train1", batch_size=hp.Train1.batch_size, queue=queue)
# Loss
loss_op = model.loss_net1()
# Accuracy
acc_op = model.acc_net1()
# Training Scheme
global_step = tf.Variable(0, name='global_step', trainable=False)
optimizer = tf.train.AdamOptimizer(learning_rate=hp.Train1.lr)
with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
var_list = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES,
'net/net1')
train_op = optimizer.minimize(loss_op,
global_step=global_step,
var_list=var_list)
# Summary
for v in tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'net/net1'):
tf.summary.histogram(v.name, v)
tf.summary.scalar('net1/train/loss', loss_op)
tf.summary.scalar('net1/train/acc', acc_op)
summ_op = tf.summary.merge_all()
session_conf = tf.ConfigProto(gpu_options=tf.GPUOptions(
allow_growth=True, ), )
# Training
with tf.Session(config=session_conf) as sess:
# Load trained model
sess.run(tf.global_variables_initializer())
model.load(sess, 'train1', logdir=logdir)
writer = tf.summary.FileWriter(logdir, sess.graph)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
for epoch in range(1, hp.Train1.num_epochs + 1):
for step in tqdm(range(model.num_batch),
total=model.num_batch,
ncols=70,
leave=False,
unit='b'):
if queue:
sess.run(train_op)
else:
mfcc, ppg = get_batch(model.mode, model.batch_size)
#print("MFCC shape: {}".format(mfcc.shape))
#print("types: {} and {}".format(mfcc.dtype, ppg.dtype))
#print("PPG shape: {}".format(ppg.shape))
sess.run(train_op,
feed_dict={
model.x_mfcc: mfcc,
model.y_ppgs: ppg
})
# Write checkpoint files at every epoch
summ, gs = sess.run([summ_op, global_step],
feed_dict={
model.x_mfcc: mfcc,
model.y_ppgs: ppg
})
# There was a problem where in certain environments placeholder must be fed for these ops
if epoch % hp.Train1.save_per_epoch == 0:
tf.train.Saver().save(
sess, '{}/epoch_{}_step_{}'.format(logdir, epoch, gs))
# Write eval accuracy at every epoch
with tf.Graph().as_default():
eval1.eval(logdir=logdir, queue=False)
writer.add_summary(summ, global_step=gs)
writer.close()
coord.request_stop()
coord.join(threads)
def train(logdir='logdir/default/train1', queue=True):
model = Model(mode="train1", batch_size=hp.Train1.batch_size, queue=queue)
# Loss
loss_op = model.loss_net1()
# Accuracy
acc_op = model.acc_net1()
# Training Scheme
global_step = tf.Variable(0, name='global_step', trainable=False)
optimizer = tf.train.AdamOptimizer(learning_rate=hp.Train1.lr)
with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
var_list = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'net/net1')
train_op = optimizer.minimize(loss_op, global_step=global_step, var_list=var_list)
# Summary
for v in tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'net/net1'):
tf.summary.histogram(v.name, v)
tf.summary.scalar('net1/train/loss', loss_op)
tf.summary.scalar('net1/train/acc', acc_op)
summ_op = tf.summary.merge_all()
session_conf = tf.ConfigProto(
gpu_options=tf.GPUOptions(
allow_growth=True,
),
)
# Training
with tf.Session(config=session_conf) as sess:
# Load trained model
sess.run(tf.global_variables_initializer())
model.load(sess, 'train1', logdir=logdir)
writer = tf.summary.FileWriter(logdir, sess.graph)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
for epoch in range(1, hp.Train1.num_epochs + 1):
for step in tqdm(range(model.num_batch), total=model.num_batch, ncols=70, leave=False, unit='b'):
if queue:
sess.run(train_op)
else:
mfcc, ppg = get_batch(model.mode, model.batch_size)
#print("MFCC shape: {}".format(mfcc.shape))
#print("types: {} and {}".format(mfcc.dtype, ppg.dtype))
#print("PPG shape: {}".format(ppg.shape))
sess.run(train_op, feed_dict={model.x_mfcc: mfcc, model.y_ppgs: ppg})
# Write checkpoint files at every epoch
summ, gs = sess.run([summ_op, global_step], feed_dict={model.x_mfcc: mfcc, model.y_ppgs: ppg})
# There was a problem where in certain environments placeholder must be fed for these ops
if epoch % hp.Train1.save_per_epoch == 0:
tf.train.Saver().save(sess, '{}/epoch_{}_step_{}'.format(logdir, epoch, gs))
# Write eval accuracy at every epoch
with tf.Graph().as_default():
eval1.eval(logdir=logdir, queue=False)
writer.add_summary(summ, global_step=gs)
writer.close()
coord.request_stop()
coord.join(threads)
