HomePhabricator

Quantize attribute values by per axis quantization parameters

Authored by Feng Liu <fengliuai@google.com> on Sep 19 2019, 2:11 PM.

Description

Quantize attribute values by per axis quantization parameters

A new converter with per axis quantization parameters is added to quantize a
dense elements attribute. For each slice along the quantization axis, it
creates an uniform quantized value converter, with different scale and zero
point, and quantizes the values in the slice.

The current implementation doesn't handle sparse elements attributes.

PiperOrigin-RevId: 270121986

Event Timeline

A. Unique TensorFlower <gardener@tensorflow.org> committed rGc8961d408ee0: Quantize attribute values by per axis quantization parameters (authored by liufengdb).Sep 19 2019, 2:12 PM