(My) resampling logic was wrong, and only passed tests because it assumed that resample_interval % batch_size == 0, which was true for our testing, have added a test that the previous version would have failed.
Highlights the need for a resampler test that tests the whole section end to end.
(My) resampling logic was wrong, and only passed tests because it assumed that resample_interval % batch_size == 0, which was true for our testing, have added a test that the previous version would have failed.
Highlights the need for a resampler test that tests the whole section end to end.