November 2019
Intermediate to advanced
296 pages
7h 52m
English
Tensors are fundamental data structures in TensorFlow.js. The framework is designed to execute every computation of these generalized vectors and matrices. A tensor contains the following properties:
A tensor is the metadata of a retained data. A tensor only keeps the pointer to the underlying data array so that we can isolate the manipulation of the tensor's shape and the transformation of the actual data array. This leaves a notable amount of room for optimization. The data types that are available in TensorFlow.js are float32, int32, bool, complex64, and string.
For example, you can create constant tensors using the ...
Read now
Unlock full access