How to use the kfserving.V1alpha2PyTorchSpec function in kfserving

def EndpointSpec(framework, storage_uri):
    if framework == 'tensorflow':
        return V1alpha2EndpointSpec(predictor=V1alpha2PredictorSpec(tensorflow=V1alpha2TensorflowSpec(storage_uri=storage_uri)))
    elif framework == 'pytorch':
        return V1alpha2EndpointSpec(predictor=V1alpha2PredictorSpec(pytorch=V1alpha2PyTorchSpec(storage_uri=storage_uri)))
    elif framework == 'sklearn':
        return V1alpha2EndpointSpec(predictor=V1alpha2PredictorSpec(sklearn=V1alpha2SKLearnSpec(storage_uri=storage_uri)))
    elif framework == 'xgboost':
        return V1alpha2EndpointSpec(predictor=V1alpha2PredictorSpec(xgboost=V1alpha2XGBoostSpec(storage_uri=storage_uri)))
    elif framework == 'onnx':
        return V1alpha2EndpointSpec(predictor=V1alpha2PredictorSpec(onnx=V1alpha2ONNXSpec(storage_uri=storage_uri)))
    elif framework == 'tensorrt':
        return V1alpha2EndpointSpec(predictor=V1alpha2PredictorSpec(tensorrt=V1alpha2TensorRTSpec(storage_uri=storage_uri)))
    else:
        raise("Error: No matching framework: " + framework)