« Back to Glossary Index

A test set is a collection of sample documents that represent the types of challenges and content an ML system will encounter in real-world use. It is used to evaluate the accuracy and performance of the system after the training phase, helping to assess how well the model generalizes to new, unseen data.