InputsDataDependency
Description
Sets the index of the input tensor on which operator computation depends.
Data dependency means that the operator computation depends not only on the shape of the input tensor, but also on the specific value of the input tensor.
Prototype
OpImplRegisterV2 &InputsDataDependency(std::initializer_list<int32_t> inputs)
Parameters
Parameter |
Input/Output |
Description |
|---|---|---|
inputs |
Input |
List of input indexes on which operator computation depends. For example, inputs={0, 3} indicates that the operator computation depends on the tensor values of input 0 and input 3. |
Returns
OpImplRegisterV2 object of the operator, where the index of the input tensor on which the operator computation depends is added.
Restrictions
None
Parent topic: OpImplRegisterV2