This message was deleted.

ah I found it, I believe. I was not specifying the data type on the

classify

function wrong:

@svc.api(input=JSON(pydantic_model=Input), output=JSON())
def classify(input):
    df = preprocess(input)
    print(df)
    result = runner.predict.run(df).item()
    return {'status_code': 200, 'prediction': result}

correct:

@svc.api(input=JSON(pydantic_model=Input), output=JSON())
def classify(input:Input):
    df = preprocess(input)
    print(df)
    result = runner.predict.run(df).item()
    return {'status_code': 200, 'prediction': result}

Ugh thats probably the root of the issue, can you send your

bentoml_configuration

here?