The above table contains information on the training data that is shared with data access users. The NCCID initiative reserves a large percentage of the data as unpublished in order to support testing and independent validation of AI models.