Glossary term
Glossary term
Infrastructure and Serving
The recommended format for saving and recovering TensorFlow models. SavedModel is a language-neutral, recoverable serialization format, which enables higher-level systems and tools to produce, consume, and transform TensorFlow models.
See the Saving and Restoring section of the TensorFlow Programmer's Guide for complete details.
Created for this library
An ML platform team exports its production model as a SavedModel so serving stacks can load it without reconstructing the training code.
A research engineer exports trained models as SavedModel artifacts so downstream teams can run inference without depending on the training code.
An ML platform team standardizes on the SavedModel format for cross-team artifact handoffs.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License