Charles Blackmon-Luca
Charles Blackmon-Luca
Cool! In that case I'm going to rename this method `multi_quantile`, and we can follow up with a rename / aliasing on cuDF?
Currently blocked on handling for sparse arrays - I am using `BlockManager.take` to perform the actual quantile operation, which retains sparse dtypes where `BlockManager.quantile` would not. This causes issues when...
Still interested in working on this, currently blocked by the handling for sparse arrays, specifically if want to retain sparse dtypes for `quantile(method="table")`
@TomAugspurger if you get a chance could I get a follow-up review on this? Also a little confused by the test failures here - in trying to resolve the single...
@jreback done
@jreback @TomAugspurger could I get a follow up review on this if either of you get the chance?
Don't think this is related to the dask-cuda nightly, as the same issue comes up with a more minimal conda env specification: ``` name: cudf_dev channels: - dask/label/dev - conda-forge...
Thanks for the ping @jrbourbeau - haven't seen this elsewhere, though if it is a non-recurring failure I would guess it may have something to do with the state of...
Sorry for the poor maintainer experience here @GenevieveBuckley 🙁 I definitely think some additional instructions/context in the bot message would go a long way in empowering maintainers to run gpuCI...
@GenevieveBuckley update - chatted with RAPIDS ops and turns out we are able to change the bot message 🙂 now the message should read: ``` Can one of the admins...