This message was deleted.

just curious, are you mostly using bentoml’s save to s3 feature?

my_prediction_service.save("s3:/...")

atm im not using bento (only heard about it yesterday). but i have a python app that downloads a pkl from S3. Then i used

load_learner

from fastai to load the model

Got it - BentoML can actually make it a lot easier to build this type of workflow. It can directly save to or load from s3

great, thanks