Kai Huang

Results 136 comments of Kai Huang

Do we have any mechanism to make sure if future functions are added in orca/data they are always typed?

is this issue related: https://github.com/intel-analytics/BigDL/issues/4448?

Hi @dinkleva Doubt whether it is due to your pyspark installation (probably on Windows?). Since it is a pure spark problem, not related to BigDL, probably you can raise this...

Hi @preet2206 If you are also facing this issue related to pyspark, then you are highly recommended to raise it in the spark community and I believe they can support...

From mlperf team: To avoid the all-to-all time for syncing embedding inputs, each worker needs to have the full data of a column (multiple columns) corresponding to its embedding(s). Basically,...

Dummy pipeline unsupported operations: - [ ] interpolate fillna https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.interpolate.html - [x] hash encode [similar to gen_string_idx with more options (e.g., no freq_limit, bucketing, etc.)] - [x] concat https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.concat.html -...

Also operations to handle timestamp, including: - [ ] f.from_unixtime - [ ] f.hour - [ ] f.minute - [ ] f.second

Strange that we are getting this error even if include_webui default to be False...

Pending for discussion whether to change the default to False?

@yushan111 If you have time, can you help verify if you can still use dashboard on yarn?