pub fn load_model(path: impl AsRef<Path>) -> Result<Model>Expand description
Load an ONNX model from a .onnx protobuf file (ONNX_RS §3.4).
This reuses the runtime’s loader pipeline (parse → build IR → resolve
weights → shape inference) and additionally decodes the model-level metadata
(ir_version, producer, doc_string, metadata_props, graph name) so that
a subsequent save_model reproduces the header faithfully. External
initializer data is resolved relative to the model file’s directory and the
backing memory maps are kept alive on the returned Model.