This message was deleted.

for question 1, yes - the native way to use the runner architecture is via service dependencies. See this example https://github.com/bentoml/BentoSD2Upscaler/blob/main/service.py

I see - so suppose i want to have 10 workers for the "API server" and 1 model runner, what I would do is • Create one service for the model, specifying 1 worker for this service • Create one service for the "api server", specifying 10 workers for this service ◦ Specify the model service as a dependency in this service • Call the API exposed by the "api server" service Correct?

That’s correct!

thanks for clarifying!