training_sample_creation.py

# At begining of the training, we mostly create random action
# As we go forward, we choose actions based on our NeuralNetwork model
if random.random() < epsilon: 
    action = env.action_space.sample()
else: 
    action = model.predict(tf.constant(np.expand_dims(state, axis=0), dtype=tf.float32)).numpy()[0]

# Create next training data
next_state, reward, done, info = env.step(action)

# if we reach to top of mountain add reward 
reward = -10. if done else reward

# replay_buffer contains all trainign data
replay_buffer.append((state, action, reward, next_state, 1 if done else 0))

state = next_state