Addresses NVTabular test failures with rapids-24.04 (pandas-2).
This PR does not "fix" the existing dtype-preservation problem:
# TODO: constructing meta like this loses dtype information on the ddf
# and sets it all to 'float64'. We should propagate dtype information along
# with column names in the columngroup graph. This currently only
# happens during intermediate 'fit' transforms, so as long as statoperators
# don't require dtype information on the DDF this doesn't matter all that much
However, it makes pandas-2 behavior consistent with pandas<2.
github-actions[bot]
commented
4 months ago
Addresses NVTabular test failures with rapids-24.04 (pandas-2).
This PR does not "fix" the existing dtype-preservation problem:
However, it makes pandas-2 behavior consistent with pandas<2.