Boris Lublinsky
Boris Lublinsky
Well b will not quite work for him. He wants to control the location, but preserve the file name. We currently preserve the location, but change the file name. Not...
`If the output table list is empty, the stats processing is lost as well,` Sorry, this is not true. Statistics is collected regardless of the amount of returned files: https://github.com/IBM/data-prep-kit/blob/dev/data-processing-lib/python/src/data_processing/runtime/transform_file_processor.py#L202....
`the arrow tables are not a problem, actually the document clustering transform does expect pyarrow table input`. The arrow table usage requires usage of unnecessary code and additional library. The...
There are a few, for example tokenization is using HF transformers, etc. What this PR is doing is quite different it allows to configure data access directly for transform usage,...
@tedhtchang, @z103cb please take a look cc: @kevin85421
@kevin85421 its ready for you
ANd I do not think this PR changes anything there
@kevin85421 we are all set
Sorry @kevin85421 , I completely disagree with your statements, see point-by-point replies below `Perhaps the KubeRay API server should construct its own HTTP client instead of using the existing one...
@kevin85421, so where do we go from here. We disagree, should we ask for the third opinion?