Quantized weight AQT style. Refer to quantizable_layer.ToAqtWeight. Args: w_name: weight name. w: The weight tensor. feature_axis: axis corresponding to output channel/feature for weights. expected_scale_shape: Optional shape to verify if scale shape is expected.
(self, w_name, w, feature_axis, expected_scale_shape)
source not stored for this graph (policy: none)
no outgoing calls
no test coverage detected