The embedding specifications file is a JSON file and a simple structure of it without advanced attributes are displayed below where some of them have default values, input_file_path and output_directory are relative paths, boolean values are specified with "true" or "false", and all numbers are specified as strings:
{Input file is a CSV file where the first number of each line is the class label number and has no header line. If number_of_secondary_data_abstraction_units is not specified, it will be equal to number_of_initial_data_abstraction_units which is the number of data instances to learn from. The distinction is meant for large data sets when preliminary_to_thirty_dimensions_umap is not set to "false" where output can have lower number of data instances than the number of data instances learnt from by UMAP to 30 dimensions and processing time is reduced.