[ML] Inference Process output field behavior #107229
Labels
>enhancement
Feature:GenAI
Features around GenAI
:ml
Machine learning
Team:ML
Meta label for the ML team
Description
The current behavior of the
InferenceProcessor
is to create the output field if it does not exist. If it exists and contains subfields, the ingest processor will leave the subfields and add more. This can result in duplicate fields which is not supported.Should we have the
InferenceProcessor
check to see if the field exists and throw an error, or maybe overwrite it? Maybe we could have an option in the processor definition to specify whether it fails, overwrites, or updates?The text was updated successfully, but these errors were encountered: