Hi I’m building a bento service which looks something like:

import bentoml
from models import InputModel, OutputModel

model_name = "dummy-model"
bento_name = "dummy-bento"

reg_runner = bentoml.sklearn.get(model_name).to_runner()
svc = bentoml.Service(bento_name, runners=[reg_runner])


@svc.api(
    input=bentoml.io.JSON(pydantic_model=InputModel),
    output=bentoml.io.JSON(pydantic_model=OutputModel),
)
def classify(data: InputModel) -> OutputModel:
    """this goes into classify description field"""
    # predict...
    dummy_res = OutputModel(...)
    return dummy_res

I have few requirements for the generated openAPI doc, and I’m not sure if its possible / how im suppose to modify the

service.py

file to fulfill those requirements: 1. I need the

classify

path (endpoint) in the generated openAPI doc to have

summary

and

operationId

fields for every endpoint. how can I do that? 2. Regarding the

summary

field from bullet 1, it seem to be automatically generated to

summary: InferenceAPI(JSON → JSON)

. can I modify it? 3. I need that the

schema

for

InputModel

and

OutputModel

to have the entry

examples

in the generated openAPI doc. 4. There are

InvalidArgument

,

NotFound

and

InternalServerError

errors under the

schema

field. is it possible to modify these? I need to make sure the returned “bad” responses have 2 fields only -

statusCode

and

message

- not sure how I can control that. Thanks in advance