yobx.sklearn.convert_helper#

yobx.sklearn.convert_helper.default_ai_onnx_ml(main_opset: int) → int[source]#

Returns a default value for opset ai.onnx.ml.

yobx.sklearn.convert_helper.wrap_step_as_function(g: GraphBuilder, fopts: FunctionOptions, estimator: BaseEstimator, input_names: List[str], output_names: List[str] | None, converter: Callable, name: str) → str | Tuple[str, ...][source]#

Converts estimator to ONNX via converter and wraps the result as an ONNX local function registered in g.

Parameters:

  • g – the main graph builder

  • fopts – FunctionOptions carrying the target domain and other options; the name field is ignored — the function name is always derived from the estimator’s class name

  • estimator – the fitted estimator to convert

  • input_names – input tensor names already present in g

  • output_names – desired output tensor names to produce in g

  • converter – the registered sklearn→ONNX converter function

  • name – node-name prefix passed to the converter