ONE - On-device Neural Engine
|
Pass to resolve TensorBroadcast IR. More...
#include <TensorBroadcastConverter.h>
Public Member Functions | |
virtual const char * | name (void) const |
bool | run (loco::Graph *graph) |
Disconnects loco::TensorBroadcast from the graph if following node is one of binary node: TFLAdd, TFLSub, TFLMul, TFLDiv, TFLMaximum and meets condition (TBA) | |
Public Member Functions inherited from logo::Pass | |
virtual | ~Pass ()=default |
Pass to resolve TensorBroadcast IR.
Definition at line 29 of file TensorBroadcastConverter.h.
|
inlinevirtual |
Reimplemented from logo::Pass.
Definition at line 32 of file TensorBroadcastConverter.h.
|
virtual |
Disconnects loco::TensorBroadcast from the graph if following node is one of binary node: TFLAdd, TFLSub, TFLMul, TFLDiv, TFLMaximum and meets condition (TBA)
After: — TensorBroadcast — x — TFLXXX — output y –/
Implements logo::Pass.
Definition at line 132 of file TensorBroadcastConverter.cpp.
References loco::active_nodes(), locoex::TFLDialect::get(), and loco::output_nodes().
Referenced by package.infer.session::inference().