IMLOperatorRegistry.RegisterOperatorSetSchema method
Registers a set of custom operator schema comprising an operator set. Operator sets follow the ONNX versioning design. Callers should provide schema for all operators that have changed between the specified baseline version and the version specified within operatorSetId. This prevents older versions of kernels from being used in models which import the newer operator set version. A type inferrer must be provided if the MLOperatorSchemaDescription structure cannot express how output types are determined. A shape inferrer may optionally be provided to enable model validation.
void RegisterOperatorSetSchema(
const MLOperatorSetId* operatorSetId,
int32_t baselineVersion,
_In_reads_opt_(schemaCount) const MLOperatorSchemaDescription* const* schema,
uint32_t schemaCount,
_In_opt_ IMLOperatorTypeInferrer* typeInferrer,
_In_opt_ IMLOperatorShapeInferrer* shapeInferrer)
Requirements
Requirement | |
---|---|
Minimum supported client | Windows 10, build 17763 |
Minimum supported server | Windows Server 2019 with Desktop Experience |
Header | MLOperatorAuthor.h |
Note
Use the following resources for help with Windows ML:
- To ask or answer technical questions about Windows ML, please use the windows-machine-learning tag on Stack Overflow.
- To report a bug, please file an issue on our GitHub.