The CSVSeriesIDLoader and SeriesIDTestLoader are still fairly poorly documented and missing relevant code for experiments. By default right now users must have datasets of equal lengths in order to proceed, however there are some cases where this might not be wanted. For instance, now I'm training a large flow foundational model and the un-equal lengths are a given. Therefore this PR will do the following things:
[ ] Update CSVSeriesIDLoader doc-strings to be more descriptive of functionality.
[ ] Update Confluence documentation on the supported uses/ideas behind SeriesIDLoader
[ ] Add parameter called validation that makes it optional to validate that datasets are the same length.
[ ] Test this out in the training foundational river flow notebook.
.
The CSVSeriesIDLoader and SeriesIDTestLoader are still fairly poorly documented and missing relevant code for experiments. By default right now users must have datasets of equal lengths in order to proceed, however there are some cases where this might not be wanted. For instance, now I'm training a large flow foundational model and the un-equal lengths are a given. Therefore this PR will do the following things:
validationthat makes it optional to validate that datasets are the same length.