Hi @Renz Abergos, this may be a limitation on the Databricks platform, will try testing it out
Chaoyu
05/26/2023, 2:27 AM
Using the HTTPServer will load the models in different processes and more performant, if you are doing this for testing/debugging purpose, it is ok to just use the old way
Chaoyu
05/26/2023, 2:27 AM
Btw we are adding some utilities for testing bento in current process easier, without the runner init code