Reduce training time of your neural networks with Mixed Precision using Tensorflow
Date:
Mixed-precision training uses half-precision floating point to speed up training, achieving the same accuracy as single-precision training sessions using the same hyper parameters