21 lines
707 B
Python
21 lines
707 B
Python
|
import tensorflow as tf
|
||
|
|
|
||
|
|
|
||
|
|
def DPG(policy, action_value):
|
||
|
|
"""
|
||
|
|
construct the gradient tensor of deterministic policy gradient
|
||
|
|
:param policy:
|
||
|
|
:param action_value:
|
||
|
|
:return: list of (gradient, variable) pairs
|
||
|
|
"""
|
||
|
|
trainable_variables = policy.trainable_variables
|
||
|
|
critic_action_input = action_value._action_placeholder
|
||
|
|
critic_value_loss = -tf.reduce_mean(action_value.value_tensor)
|
||
|
|
policy_action_output = policy.action
|
||
|
|
|
||
|
|
grad_ys = tf.gradients(critic_value_loss, critic_action_input)
|
||
|
|
grad_policy_vars = tf.gradients(policy_action_output, trainable_variables, grad_ys=grad_ys)
|
||
|
|
|
||
|
|
grads_and_vars = zip(grad_policy_vars, trainable_variables)
|
||
|
|
|
||
|
|
return grads_and_vars
|