If the extracted text exceeds the character size limit specified in this section, it will be split for embedding purposes.
The default value is set to match the limit of the selected embedding model, and splitting is done using a recursive splitter approach.