How to get the standard Deviation and mean values after training a BNN model using DenseFlipout Layers in Tensorflow?

Question

After I trained the Bayesian Neural Network model and saved it into a .h5 file, I cannot plot its result because I don't know the standard deviation and predicted mean trained by the model. I'm using the following method to get the predictions:

with tf.GradientTape() as tape0:
    tape0.watch(T)
    with tf.GradientTape() as tape1:
            tape1.watch(S)
            with tf.GradientTape() as tape2:
                    tape2.watch([S, T])
                    X = tf.concat([S/(K*tf.exp(-r*T)), T], 1) #input matrix for ANN
                    print(X)
                    out = model(X)    
                    out_sd = out.stddev()
                    print(out)
                    out_values = K*tf.where(tf.greater(T, 1e-3), out, tf.maximum(S/K - 1, 0))
                
            delta_values, theta_values = tape2.gradient(out_values, [S, T])
    gamma_values = tape1.gradient(delta_values, S)
charm_values = tape0.gradient(delta_values, T)

out_values = out_values.numpy()

So basically I am trying to compute the standard deviation using out.stddev() of the given model. But apparently the out defined in this code is a Tensor instead of sth that has an attribute of the model standard deviation. Does anyone know how to get the standard deviation from the trained Bayesian Neural Nets using DenseFlipout layers? Thanks in advance!