Glossary term
Glossary term
Infrastructure and Serving
In TensorFlow, any procedure that creates, manipulates, or destroys a Tensor. For example, a matrix multiply is an operation that takes two Tensors as input and generates one Tensor as output.
Created for this library
An ML platform team profiles individual operations in its model graph to identify the most expensive ops to optimize.
A research engineer fuses adjacent operations in the model graph to reduce serving latency for the production deployment.
An ML engineer registers custom operations in TensorFlow when a particular performance-critical computation lacks a built-in implementation.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License