Hello Everyone, thanks a lot for the great framework! 🙂 I'm build a Transcription Service with an endpoint that requests a file and many other parameters:

@bentoml.api(route="/v1/audio/transcriptions")
def transcribe(
    self,
    file: Annotated[Path, ContentType("audio/mpeg")],
    model: Optional[ModelName] = Field(),
    language: Optional[Language] = Field(),
    prompt: Optional[float] = Field(),
    response_format: Optional[ResponseFormat] = Field(),
    temperature: Optional[Union[float, List[float]]] = Field(),
    timestamp_granularities: Optional[List[TimestampGranularity]] = Field()
)

Now, because the file is sent together with all the other parameters, the request generated by the Swagger UI is of

Content-Type: multipart/form-data

. So far so good, what does not work is to send a list of

timestamp_granularities

or a list of

temperature

from the Swagger UI because the list is passed as comma-separated-values (word,segment).

curl -X 'POST' \
  '<http://localhost:8002/v1/audio/transcriptions>' \
  -H 'accept: text/plain' \
  -H 'Content-Type: multipart/form-data' \
  -F 'file=@example_audio.mp3;type=audio/mpeg' \
  -F 'timestamp_granularities=word,segment'

Is there either a way to pass a list instead of comma-separated-values or alternatively access the request-context and split the csv up myself? Thanks a lot in advance!!

Hello, maybe can you be more specific for us to help understand your usecas. What will be your expected behaviour and is there any blocker with the current behaviour for your workflow?