We have been working LTR Plugin for search relevancy. In this context, as per the LTR needs,
- The feature set needs to be created using api
- The logging feature has to be invoked using api
- Accordingly the judgment list needs to be created
- The model (e.g. xgboost) has be trained
- The model needs to be uploded using api
Wanted to understand with model serving framework in place, can all the steps for ltr be done using the model serving framework and pipeline. If yes, can you refer to an example.