Hello, BentoML Team. I have a question for using BentoML Streaming API Code Example

@bentoml.service()
class A:
    @bentoml.api
    def stream(self) -> typing.Generator(bytes, None, None):
        for i in range(10):
            arr = np.empty((10, 10, 3), dtype=np.uint8).tobytes()
            yield arr

@bentoml.service()
class B:
    a: A = bentoml.depends(A)
    @bentoml.api
    def stream(self) -> typing.Generator(bytes, None, None):
        response = self.a.stream()
        for res in response:
             yield res

When I request to bentoml API using python requests.

response = <http://requests.post|requests.post>(f'<http://localhost:3000/stream>', data=data, stream=True)
for r in response.iter_content(chunk_size=None):
    print(f"recieved: {len(r)}")

I just get a error

("Connection broken: InvalidChunkLength(got length b'', 0 bytes read)", InvalidChunkLength(got length b'', 0 bytes read))

So when I debugged the code. response' s type is generator. the value is <generator object SyncHTTPClient._get_stream at 0x7f4a923b8a50> The above results were obtained. When I searched more, I found that class A is a remote proxy object, and it is calling the api inside. There is a boolean value called stream_output in the ClientEndPoint used at this time, but it seems that the default is False, so it is not handling the stream request well. I would appreciate it if you could let me know if I'm missing something or if I'm mistaken.