Glossary term
Glossary term
Evaluation and Benchmarks
A problem that can be solved using machine learning techniques, such as:
Created for this library
A platform team defines each ML task with a clear input, output, and metric so research and product align from the start.
An LLM team designs each instruction-tuning task with clear inputs and expected outputs so the training set is consistent.
An NLP team builds a multi-task model that handles intent classification, named-entity recognition, and slot filling from a shared encoder.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License