Feature Store Log Model and Score Batch - env_manager

AdamIH123 — Mon, 18 Dec 2023 03:04:49 GMT

Hi Everyone.

I have a couple of questions about the feature store log model and score batch.

After you log a model with the feature store then use fs.score_batch is it possible to pass the env_manager to predict with the same env as training as described in the warning created? See attached screenshot that was taken from the 2022 “Enable Production ML with Databricks Feature Store” video by Databricks. From my understanding you cannot load a feature store model using mlflow that does accept env_manager.
1. If it is possible to pass env_manager to address this warning, would a coda env parameter have to be passed to fs.log_model?
2. Does anyone have an example of doing the above?
Is there any docs on fs.score_batch or log model. I assume fs.log_model is basically the same as mlflow’s log model

#feature_store #score_batch #env_manager

Re: Feature Store Log Model and Score Batch - env_manager

MohsenJ — Wed, 13 Mar 2024 14:13:09 GMT

I also like to know if that works.