# Training
init = tf.global_variables_initializer()
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction = gpu_frac)
sess = tf.Session(config=tf.ConfigProto(log_device_placement = False, gpu_options=gpu_options))
sess.run(init)
#%% Training cycle
for epoch in range(training_epochs):
training_data = [[]]* num_training_files
reading_phase = True
for file_idx in range(num_training_files):
if reading_phase:
training_data [file_idx] = data_loader(file_idx, input_dim)
for i in range(n_batch):
#pre_emph is zero since we do it on X if we want separatrely
batch_xs = data_parser(training_data[file_idx], input_dim, batch_size,
overlap = overlap)
################################################################
# Update taco
_, taco_cost_ = sess.run([opt, taco_cost],
feed_dict={X: batch_xs,
learning_rate: learning_rate_init,
noise_std : noise_std_init})
##### Display logs per epoch step
if i % display_step == 0:
print("epoch:", '%02d' % (epoch + 1),
"File:", '%02d' % (file_idx),
"iteration:", '%04d' % (i + 1),
"Taco_cost =", "{:.9f}".format( (10**4) * taco_cost_))
# ### Early stopping!
# Training
init = tf.global_variables_initializer()
sess=tf.Session()
sess.run(init)
# Training cycle
for epoch in range(training_epochs):
# learning_rate_new = learning_rate_new /2.
for file_idx in range(10):
training_data = data_loader(file_idx)
# n_batch = int(n_training / batch_size)
for i in range(n_batch):
#pre_emph is zero since we do it on X if we want separatrely
batch_xs = data_parser(training_data, input_dim, batch_size, preemph=0.0, overlap=True)
_, c = sess.run([optimizer, cost], feed_dict={X: batch_xs, mode:0.0})
# Display logs per epoch step
if i % display_step == 0:
print("epoch_", '%02d' % (epoch+1),
"File_", '%02d' % (file_idx),
"iteration_", '%04d' % (i+1),
"cost=", "{:.9f}".format(c))
print("Optimization Finished!")
#%%##########################################################################
# Training error calculation
training_data = data_loader(9)
training_error = 0