MindSpore

Basic introduction MindSpore provides a Pipeline-based data engine that allows for efficient data pre-processing through datasets (Dataset). In this course, we use MNIST data sets and automatically download them and pre-process them using data conversions provided by Mindspore.dataset. The example code in this chapter relies on download, which can be installed using the command pip install download. If this document is run in Notebook, the installation needs to be completed before the next code can be rebooted. MindSpore's dataset uses data processing water lines (Data Processing Pipeline) to specify map, batch, shuffle, etc. Here we use Map to convert image data and labels: zooms the images entered to 1/255, and downsizing them according to a mean value of 0.107 and a standard deviation of 0.3081. The datasets can be viewed in an iterative way using dataset_tuple_itator or code_dic_itator to view data and labelsshape and datype.