piekill

SparkTorch casts all input to float tensor, see: https://github.com/dmmiller612/sparktorch/blob/master/sparktorch/torch_distributed.py#L114 It seems that the only way is to change the source code. You could hard-code a `.long()` instead the `.float()` to...

> Sorry for the late reply, I was out for medical leave. > > The column level lineage in WHERE, ON, USING are not handled for the moment. Theoretically we...

What features will be available in Team / organizations? Thanks~

Sounds good. Is it also possible to view the details of the team members besides aggregations?

Another question is how to calculate durations using heartbeats. I only find a single `time` column in heartbeats table. I wanna know the start and end time of my coding...

Also, the generated code doesn't take into account the pre-request script. In Postman you can copy the actual request being sent after the request has been modified by the script.

Current implementation also results in incorrect column lineage in some cases, e.g. ``` sql = """ INSERT INTO res SELECT t."C" FROM ( SELECT a AS "C" FROM tab1 )...