deterministic_discrete_mlp_q_function.py 文件源码

def get_action_sym(self, obs_var):
        output_vec = L.get_output(self._output_vec_layer, obs_var, deterministic=True)
        action = tf.to_int64(tf.argmax(output_vec, 1))
        action_vec = tf.one_hot(action, self._n)
        return action_vec