ONE - On-device Neural Engine
|
Class to quantize ONNX QuantizeLinear-DequantizeLinear operator. More...
#include <QuantizeOnnxQDQPass.h>
Public Member Functions | |
const char * | name (void) const final |
bool | run (loco::Graph *g) final |
Public Member Functions inherited from logo::Pass | |
virtual | ~Pass ()=default |
Class to quantize ONNX QuantizeLinear-DequantizeLinear operator.
Definition at line 29 of file QuantizeOnnxQDQPass.h.
|
inlinefinalvirtual |
Reimplemented from logo::Pass.
Definition at line 31 of file QuantizeOnnxQDQPass.h.
|
finalvirtual |
Quantize pattern
[Before]
[CircleNode(fp32)] |
[CircleCustom(OnnxQuantizeLinear)] | [CircleCustom(OnnxDequantizeLinear)] | [CircleNode]
[After]
[CircleNode(quantized)] | [CircleNode]
Implements logo::Pass.
Definition at line 270 of file QuantizeOnnxQDQPass.cpp.
References loco::active_nodes(), and loco::output_nodes().
Referenced by package.infer.session::inference(), and luci::QuantizeOnnxFakeQuantModelPass::run().