def dream():
c = constants()
p = params()
# connect to matlab
eng = mo.matlab_connection(c.MATLAB_SESSION)
x = tf.placeholder(tf.float32, shape=(1, 100, 2))
logits = model(x, c)
logits = tf.reshape(logits, [-1, c.MAX_TIME, f.OUTPUT_DIM])
init_op = tf.global_variables_initializer()
rand_state = np.random.uniform(-.005, .005, (1, 100, 2))
t = np.linspace(0, 1, 100)
x1 = .65 + .25 * np.cos(4.5 * 2.0 * np.pi * t)
x2 = .45 + .25 * np.cos(6.2 * 2.0 * np.pi * t)
istate = np.asarray([x1, x2])
istate = [np.transpose(istate)]
istate = istate + rand_state
with tf.Session() as sess:
# initialize global variables
sess.run(init_op)
# enable batch fetchers
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
#setup saver
saver = u.saver_ops(sess=sess,
max_to_keep=3,
global_step=p.global_step)
xs = []
logi = []
# run the training operation
for indx in range(1200):
feed_dict = {x: istate}
ex, ostate = sess.run([x, logits], feed_dict=feed_dict)
logi.append(ostate[0])
xs.append(ex[0])
istate = np.roll(istate, -1, axis=1)
istate[-1, -1, :] = ostate[-1, -1, :]
u.display_update(indx, 20)
eng.put_var('logits', logi)
eng.put_var('x', xs)
# matlab command =>
# >>plot( reshape(logits(1,:,:), [100,2]) )
coord.request_stop()
coord.join(threads)
def train():
c = constants()
p = params()
x, y = f.inputs([f.TRAIN_TF_RECORDS_FILE], shuffle=True, name='inputs')
u.shape_log(x)
u.shape_log(y)
logits = model(x, c)
logits = tf.reshape(logits, [-1, c.MAX_TIME, f.OUTPUT_DIM])
labels = tf.reshape(y, [-1, c.MAX_TIME, f.OUTPUT_DIM])
loss = tf.reduce_mean(tf.square(logits - labels))
train_op = tf.train.AdamOptimizer(learning_rate=0.05).minimize(
loss, global_step=p.global_step)
init_op = tf.global_variables_initializer()
with tf.Session() as sess:
# initialize global variables
sess.run(init_op)
# enable batch fetchers
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
#setup saver
saver = u.saver_ops(sess=sess,
max_to_keep=3,
global_step=p.global_step)
eng = mo.matlab_connection(c.MATLAB_SESSION)
# run the training operation
for indx in range(20000):
_, log = sess.run([train_op, logits])
# logi.append(log)
u.display_update(indx, 25)
if ((indx + 1) % 2000 == 0):
saver.save()
if ((indx) % 5 == 0):
# CHECK EMERGENCY STOP
if eng.should_stop():
print('stopping')
break
# connect to matlab
# eng = mo.matlab_connection(c.MATLAB_SESSION)
# eng.put_var('logits', logi)
saver.save()
coord.request_stop()
coord.join(threads)
def train():
c = constants()
p = params()
# connect to matlab
eng = mo.matlab_connection(c.MATLAB_SESSION)
# matlab initialization...
# [b,a] = butter(7, .2);
# x = [zeros(1,10), ones(1,90)];
# y = filter(b,a,x);
mx = eng.get_var('x')
my = eng.get_var('y')
x = tf.constant(mx, dtype=tf.float32)
logits = model(x, c)
labels = tf.constant(my, dtype=tf.float32)
labels = tf.reshape(labels, [c.MAX_TIME])
loss = tf.reduce_mean(tf.square(logits - labels))
train_op = tf.train.AdamOptimizer(learning_rate=0.05).minimize(
loss, global_step=p.global_step)
init_op = tf.global_variables_initializer()
with tf.Session() as sess:
# initialize global variables
sess.run(init_op)
#setup saver
saver = u.saver_ops(sess=sess,
max_to_keep=3,
global_step=p.global_step)
logi = []
# run the training operation
for indx in range(750):
_, log = sess.run([train_op, logits])
logi.append(log)
u.display_update(indx, 250)
eng.put_var('logits', logi)
saver.save()
def train():
c = constants()
some_example_random_number = -0.31415
x = tf.constant(some_example_random_number, dtype=tf.float32)
x = tf.reshape(x, [1, 1])
d = tf.layers.dense(x, units=2, name='dense_1')
e = tf.layers.dense(d, units=1)
weights, biases = get_dense_vars('dense_1')
print(d)
print(weights)
print(biases)
# https://stackoverflow.com/questions/45372291/how-to-get-weights-in-tf-layers-dense
labels = tf.constant(1.0, dtype=tf.float32)
loss = tf.reduce_sum(tf.square(e - labels))
train_op = tf.train.AdamOptimizer(learning_rate=0.5).minimize(loss)
init_op = tf.global_variables_initializer()
# connect to matlab
eng = m.matlab_connection(c.MATLAB_SESSION)
with tf.Session() as sess:
# initialize global variables
sess.run(init_op)
xs = []
ws = []
bs = []
# run the training operation
for indx in range(1000):
_, x_eval, w, b = sess.run([train_op, x, weights, biases])
xs.append(x_eval)
ws.append(w)
bs.append(b)
# print("x = {}".format(x_eval))
eng.put_var('xs', xs)
eng.put_var('ws', ws)
eng.put_var('bs', bs)
def test():
c = constants()
p = params()
# connect to matlab
eng = mo.matlab_connection(c.MATLAB_SESSION)
# matlab initialization...
# [b,a] = butter(7, .2);
# x = [zeros(1,10), ones(1,90)];
# y = filter(b,a,x);
mx = eng.get_var('s')
my = eng.get_var('t')
x = tf.constant(mx, dtype=tf.float32)
logits = model(x, c)
labels = tf.constant(my, dtype=tf.float32)
labels = tf.reshape(labels, [c.MAX_TIME])
init_op = tf.global_variables_initializer()
with tf.Session() as sess:
# initialize global variables
sess.run(init_op)
#setup saver
saver = u.saver_ops(sess=sess,
max_to_keep=3,
global_step=p.global_step)
logi = []
labs = []
# run the training operation
for indx in range(1):
log, lab = sess.run([logits, labels])
logi.append(log)
labs.append(lab)
u.display_update(indx, 250)
eng.put_var('logits', logi)
eng.put_var('labels', labs)
saver.save()
def train():
c = constants()
p = params()
x, y = f.inputs([f.TRAIN_TF_RECORDS_FILE], name='inputs')
u.shape_log(x)
u.shape_log(y)
logits = model(x, c)
labels = tf.reshape(y, [c.MAX_TIME])
loss = tf.reduce_mean(tf.square(logits - labels))
train_op = tf.train.AdamOptimizer(learning_rate=0.05).minimize(
loss, global_step=p.global_step)
init_op = tf.global_variables_initializer()
with tf.Session() as sess:
# initialize global variables
sess.run(init_op)
# enable batch fetchers
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
#setup saver
saver = u.saver_ops(sess=sess,
max_to_keep=3,
global_step=p.global_step)
logi = []
# run the training operation
for indx in range(10000):
_, log = sess.run([train_op, logits])
logi.append(log)
u.display_update(indx, 50)
# connect to matlab
eng = mo.matlab_connection(c.MATLAB_SESSION)
eng.put_var('logits', logi)
saver.save()
coord.request_stop()
coord.join(threads)
def test():
c = constants()
p = params()
# connect to matlab
eng = mo.matlab_connection(c.MATLAB_SESSION)
x, y = f.inputs([f.TEST_TF_RECORDS_FILE], shuffle=True, name='inputs')
u.shape_log(x)
u.shape_log(y)
logits = model(x, c)
logits = tf.reshape(logits, [-1, c.MAX_TIME, f.OUTPUT_DIM])
labels = tf.reshape(y, [-1, c.MAX_TIME, f.OUTPUT_DIM])
init_op = tf.global_variables_initializer()
with tf.Session() as sess:
# initialize global variables
sess.run(init_op)
# enable batch fetchers
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
#setup saver
saver = u.saver_ops(sess=sess,
max_to_keep=3,
global_step=p.global_step)
logi = []
labs = []
# run the training operation
for indx in range(6):
log, lab = sess.run([logits, labels])
logi.append(log)
labs.append(lab)
u.display_update(indx, 250)
eng.put_var('logits', logi)
eng.put_var('labels', labs)
coord.request_stop()
coord.join(threads)
def train():
c = constants()
# connect to matlab
eng = mo.matlab_connection(c.MATLAB_SESSION)
# matlab initialization...
# [b,a] = butter(7, .2);
# x = [zeros(1,10), ones(1,90)];
# y = filter(b,a,x);
mx = eng.get_var('x')
my = eng.get_var('y')
x = tf.constant(mx, dtype=tf.float32)
logits = model(x, c)
labels = tf.constant(my, dtype=tf.float32)
labels = tf.reshape(labels, [100])
loss = tf.reduce_mean(tf.square(logits - labels))
train_op = tf.train.AdamOptimizer(learning_rate=0.05).minimize(loss)
init_op = tf.global_variables_initializer()
with tf.Session() as sess:
# initialize global variables
sess.run(init_op)
logi = []
# run the training operation
for indx in range(750):
_, log = sess.run([train_op, logits])
logi.append(log)
# post an update
if ((indx + 1) % 250 == 0):
print("[step: {}]".format(indx))
eng.put_var('logits', logi)
