Glossary term
Glossary term
Foundations
A machine learning technique in which a single model is trained to perform multiple tasks.
Multitask models are created by training on data that is appropriate for each of the different tasks. This allows the model to learn to share information across the tasks, which helps the model learn more effectively.
A model trained for multiple tasks often has improved generalization abilities and can be more robust at handling different types of data.
N
Created for this library
A search platform team uses a multitask model that scores relevance, freshness, and engagement together with shared representations.
An NLP team uses a multitask model that handles classification, named-entity recognition, and intent detection from a shared encoder.
A retail demand team uses a multitask forecast that predicts daily sales and stockouts together from a shared embedding.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License