mlserver_custom_runtime_extending_mlmodel_with_inference.py

1from mlserver import MLModel
2from mlserver.types import InferenceRequest, InferenceResponse
3from mlserver.codecs import decode_args
4
5class MyModel(MLModel):
6    async def load(self) -> bool:
7        # Load your model here (e.g., from a file)
8        self.ready = True
9        return self.ready
10
11    @decode_args
12    async def predict(self, payload: list[float]) -> InferenceResponse:
13        # Perform inference
14        # In this case, we just return the input
15        prediction = payload
16
17        return InferenceResponse(
18            model_name=self.name,
19            model_version=self.version,
20            outputs=[]
21        )