Masked model

A masked model in machine learning is a type of model that is trained to predict missing or masked parts of input data. This approach is often used in natural language processing (NLP) to help models understand and generate text by learning to predict the missing words in a sentence.