class optimum.amd.ryzenai.RyzenAIConfig
< source > ( opset: Optional = None quantization: Optional = None **kwargs )
Parameters
- opset (
Optional[int]
, defaults to None
) —
ONNX opset version to export the model with. - quantization (
Optional[QuantizationConfig]
, defaults to None
) —
Specify a configuration to quantize ONNX model
RyzenAIConfig is the configuration class handling all the VitisAI parameters related to the ONNX IR model export,
and quantization parameters.