MCPcopy Create free account
hub / github.com/pytorch/executorch / setup_quant_params

Method setup_quant_params

backends/samsung/quantizer/quantizer.py:36–42  ·  view source on GitHub ↗
(
        self, quant_dtype: Precision, is_per_channel=True, is_qat=False
    )

Source from the content-addressed store, hash-verified

34 global_quant_info.set_precision(self._precision)
35
36 def setup_quant_params(
37 self, quant_dtype: Precision, is_per_channel=True, is_qat=False
38 ) -> None:
39 assert quant_dtype in Precision, f"No support for Precision {quant_dtype}."
40 self._precision = quant_dtype
41 self._is_per_channel = is_per_channel
42 self._is_qat = is_qat
43
44 def annotate(self, model: GraphModule) -> GraphModule:
45 self._annotate(model)

Callers 2

quantize_moduleFunction · 0.95
quantizeMethod · 0.95

Calls

no outgoing calls

Tested by 1

quantizeMethod · 0.76