def run_train(Env, curr_RL_Config):
# run the training process
"""
Train the agents
"""
Num_neighbor = Env.n_Neighbor
Num_d2d = Env.n_Veh
Num_CH = Env.n_RB
Num_D2D_feedback = curr_RL_Config.Num_Feedback
# construct a BS agent
BS_Agent = Agent(Num_d2d, Num_CH, Num_neighbor, Num_D2D_feedback, Env,
curr_RL_Config)
Num_Episodes = curr_RL_Config.Num_Episodes
Num_Train_Step = curr_RL_Config.Num_Train_Steps
# get the train loss
[Train_Loss, Reward_Per_Train_Step, Reward_Per_Episode,
Train_Q_mean, Train_Q_max_mean, Orig_Train_Q_mean, Orig_Train_Q_max_mean] \
= BS_Agent.train(Num_Episodes, Num_Train_Step)
return [
Train_Loss, Reward_Per_Train_Step, Reward_Per_Episode, Train_Q_mean,
Train_Q_max_mean, Orig_Train_Q_mean, Orig_Train_Q_max_mean
]
def run_train(Env, curr_RL_Config):
# run the training process
"""
Run the Training Process
"""
# parameters to construct a BS Agent object
Num_neighbor = Env.n_Neighbor
Num_d2d = Env.n_Veh
Num_CH = Env.n_RB
Num_D2D_feedback = curr_RL_Config.Num_Feedback
BS_Agent = Agent(Num_d2d, Num_CH, Num_neighbor, Num_D2D_feedback, Env, curr_RL_Config)
Num_Episodes = curr_RL_Config.Num_Episodes
Num_Train_Step = curr_RL_Config.Num_Train_Steps
# get the train loss
[Train_Loss, Train_Q_mean, Train_Q_max_mean] = BS_Agent.train(Num_Episodes, Num_Train_Step)
return [Train_Loss, Train_Q_mean, Train_Q_max_mean]