BentoML

Use it when

• package a Python model behind a production HTTP service
• build a reproducible model container with batching and resource declarations

What it solves

Not the fit when

• does not choose or train the model

Install

pip install bentoml

Invoke

Define a @bentoml.service with an API method, run `bentoml serve service:Service`, then containerize or deploy the generated Bento.

Alternatives

No reviewed alternatives recorded yet.