Glossary term
Glossary term
Infrastructure and Serving
A job that keeps track of a model's parameters in a distributed setting.
Created for this library
An ML platform team uses a parameter server architecture to scale training of a large embedding model across many workers.
A search-ranking team uses parameter servers in its distributed training of a model with very large embedding tables.
A research team uses a parameter server design for its large sparse model where embedding tables exceed single-device memory.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License