ONNX and PMML encoded ML models
Nussknacker can infer models encoded using ONNX and PMML ML interchange formats. You can find more details about these formats here:
Once the integration is defined, Nussknacker will generate ML enricher, which infers ONNX or PMML encoded models. A great advantage of inferring ML models encoded using ONNX or PMML is that they are executed in-process, without the communication overhead with the additional ML inference server. You can read more about ML enricher here.
Our implementation of components allowing inference of ONNX and PMML encoded models uses licensed software libraries - contact us at enterprose@nussknacker.io if you want to have this functionality enabled.