Improve the dataloader. It should manage the temporal consistency between different csv files.
Keep the simplest case with time (view on the rows) and dim (view on the columns).
multi-dimensionality will be multiplied (example: dim=(2,2) so we take 2*2=4 columns)
multi-sequence won't be managed. only sequences of rank 1 (striding sliding window along the time)
Improve the dataloader. It should manage the temporal consistency between different csv files.
Keep the simplest case with time (view on the rows) and dim (view on the columns).
multi-dimensionality will be multiplied (example: dim=(2,2) so we take 2*2=4 columns)
multi-sequence won't be managed. only sequences of rank 1 (striding sliding window along the time)