|
ONE - On-device Neural Engine
|
Class to convert a quantized model to a fake-quantized fp32 model. More...
#include <ConvertToFakeQuantizedModelPass.h>

Public Member Functions | |
| ConvertToFakeQuantizedModelPass () | |
| const char * | name (void) const final |
| bool | run (loco::Graph *g) final |
| Run the pass. | |
Public Member Functions inherited from logo::Pass | |
| virtual | ~Pass ()=default |
Class to convert a quantized model to a fake-quantized fp32 model.
Definition at line 28 of file ConvertToFakeQuantizedModelPass.h.
|
inline |
Definition at line 30 of file ConvertToFakeQuantizedModelPass.h.
Reimplemented from logo::Pass.
Definition at line 32 of file ConvertToFakeQuantizedModelPass.h.
|
finalvirtual |
Run the pass.
Implements logo::Pass.
Definition at line 275 of file ConvertToFakeQuantizedModelPass.cpp.
References loco::active_nodes(), INFO, LOGGER, loco::must_cast(), luci::must_cast(), and loco::output_nodes().