Custom operators#

custom_op_schemas(op_type='', init_doc=True)#: Returns LightOpSchema records for supported com.microsoft operators.

operator_schema_lookup(op_type)#: Returns standard ONNX schemas and this package’s custom schemas. Pass it as GraphBuilder(..., schema_lookup=operator_schema_lookup).

class OperatorSupport#: Immutable NamedTuple describing shape inference, peak memory, fusion patterns, and gradient availability for one custom operator.

operator_support() → tuple[OperatorSupport, ...]#: Returns the custom operator support inventory without registering or executing an implementation.

register_operator_support() → None#: Registers custom shape-inference, peak-memory, and fusion-pattern support.

register_custom_gradients(registry=None)#: Adds the CDist and BiasGelu backward rules to an independent GradRegistry and returns it.