MCPcopy Create free account
hub / github.com/tensorflow/lingvo / ToAqtWeight

Method ToAqtWeight

lingvo/core/quant_utils.py:449–475  ·  view source on GitHub ↗

Quantized integer weight AQT style. This only scales, rounds and clips; resulting quantized weight would be either integer or integer emulated in float. w_name must have been previously created via TrackQWeight. Args: w_name: Previously created w_name QWeight to quantize wei

(self, w_name, w, feature_axis, expected_scale_shape=None)

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers

nothing calls this directly

Calls 3

_ValidateWeightMethod · 0.95
_GetQDomainMethod · 0.95
ToAqtWeightMethod · 0.45

Tested by

no test coverage detected