Closed CielAl closed 10 months ago
That should vastly improve the efficiency of Vahadane algorithm when on-the-fly normalization is necessary for multiple iterations, e.g., for multiple epochs of downstream deep learning model training.
Done.
That should vastly improve the efficiency of Vahadane algorithm when on-the-fly normalization is necessary for multiple iterations, e.g., for multiple epochs of downstream deep learning model training.