Closed MarcBresson closed 1 year ago
Sometimes it is equals less than -1 (like -1.000000002) so we get a NaN. We can manually check for it to be greater than 1 or less than -1, or we could try to use float16 precision on the bumpy array
Sometimes it is equals less than -1 (like -1.000000002) so we get a NaN. We can manually check for it to be greater than 1 or less than -1, or we could try to use float16 precision on the bumpy array