Implementing the Double Deep Q network

The network architecture of the Double Deep Q network is illustrated as follows. The networks have CNN architecture to process the states as images and output Q values for all possible actions. The detailed code(DQN.py) is as follows:

import keras
from keras import optimizers
from keras.layers import Convolution2D
from keras.layers import Dense, Flatten, Input, concatenate, Dropout
from keras.models import Model
from keras.utils import plot_model
from keras import backend as K
import numpy as np

'''
Double Deep Q Network Implementation
'''

learning_rate = 0.0001
BATCH_SIZE = 128

class DQN:

    def __init__(self,num_states,num_actions,model_path):
        
        self.num_states = num_states
        print(num_states)
        self.num_actions = num_actions
        self.model = self.build_model() # Base Model 
        self.model_...