Weights in Autoencoder

[azimnoralfian]

Hello everyone, I'm building and Autoencoder and using MNIST as my dataset. I'm stuck in finding a way to copy the improvement from the encoder after training to put it in the decoder because theoretically the weights should be the same for both encoder and decoder. Are there a specific command for me to check or to say to the Autoencoder to stop the improvement until the latent space and copy the improvement to put in the decoder? I hope the question is clear