Go to the source code of this file.
Classes | |
class | continuous.PolicyNetwork |
class | continuous.QValueNetwork |
— Q-value and policy networks More... | |
class | continuous.ReplayItem |
Namespaces | |
continuous | |
Functions | |
def | continuous.rendertrial (maxiter=NSTEPS, verbose=True) |
Variables | |
continuous.batch | |
int | continuous.BATCH_SIZE = 64 |
continuous.d_batch = np.vstack([b.done for b in batch]) | |
float | continuous.DECAY_RATE = 0.99 |
bool | continuous.done = False |
continuous.env = Pendulum(1) | |
— Environment More... | |
continuous.feed_dict | |
list | continuous.h_qva = [] |
list | continuous.h_rwd = [] |
History of search. More... | |
list | continuous.h_ste = [] |
tuple | continuous.maxq |
continuous.n_init = tflearn.initializations.truncated_normal(seed=RANDOM_SEED) | |
int | continuous.NEPISODES = 100 |
— Hyper paramaters More... | |
int | continuous.NH1 = 250 |
int | continuous.NSTEPS = 100 |
continuous.NU = env.nu | |
continuous.NX = env.nobs | |
continuous.optim | |
continuous.policy = PolicyNetwork().setupOptim() | |
— Tensor flow initialization More... | |
float | continuous.POLICY_LEARNING_RATE = 0.0001 |
continuous.policyTarget = PolicyNetwork().setupTargetAssign(policy) | |
continuous.q2_batch | |
continuous.qgrad | |
continuous.qref_batch = r_batch + (not d_batch) * (DECAY_RATE * q2_batch) | |
continuous.qvalue = QValueNetwork().setupOptim() | |
float | continuous.QVALUE_LEARNING_RATE = 0.001 |
continuous.qvalueTarget = QValueNetwork().setupTargetAssign(qvalue) | |
continuous.r | |
continuous.r_batch = np.vstack([b.reward for b in batch]) | |
continuous.RANDOM_SEED = int((time.time() % 10) * 1000) | |
— Random seed More... | |
int | continuous.REPLAY_SIZE = 10000 |
continuous.replayDeque = deque() | |
float | continuous.rsum = 0.0 |
continuous.sess = tf.InteractiveSession() | |
continuous.u = sess.run(policy.policy, feed_dict={policy.x: x}) | |
continuous.u2_batch | |
continuous.u_batch = np.vstack([b.u for b in batch]) | |
continuous.u_init = tflearn.initializations.uniform(minval=-0.003, maxval=0.003, seed=RANDOM_SEED) | |
continuous.u_targ = sess.run(policy.policy, feed_dict={policy.x: x_batch}) | |
float | continuous.UPDATE_RATE = 0.01 |
continuous.withSinCos | |
continuous.x = env.reset().T | |
— Training More... | |
continuous.x2 = x2.T | |
continuous.x2_batch = np.vstack([b.x2 for b in batch]) | |
continuous.x_batch = np.vstack([b.x for b in batch]) | |