I have a deep CNN/RNN that I train on Google AI platform. I distribute the training on 8 GPUs using the tf.distribute.MirroredStrategy. I recently upgraded my r
tf.distribute.MirroredStrategy