Hi, first of all, congratulations on the brilliant work done for this library!

I am currently trying to get to work this library (particularly the transformer_bottleneck (logsparse) architecture) with the famous benchmark datasets 'Electricity' and 'Traffic'. However, both of these datasets have categorical ids which indicates different stations / lanes, respectively, time series.

After reading the (a bit sparse) documentation, and checking some of the available examples, I couldn't figure out some of my main questions:

• Is there an option in the configuration files for specifying a "grouping id" so that this behavior of the dataset is taken into account? (e.g. for taking into account individually each series according to its ids, for group scaling, ...)

• Apart from the "relevant_cols" and "target_col" attributes specification, is there a specific attribute in the config file for specifying the name of the date column?

Thanks in advance!