ONNX Runtime onnxruntime/python/tools/quantization README

Fields

id

sources/qlab2-ort-quant-tool-readme

URL

https://github.com/microsoft/onnxruntime/tree/main/onnxruntime/python/tools/quantization

Publisher

ONNX Runtime (Microsoft)

Kind

repo

Accessed

2026-10-03

notes

The quantization tool directory listing and README. Adds the per-platform parameter table: x86 without VNNI needs QUInt8/QInt8 + reduce_range=True (7-bit weights) to dodge VPMADDUBSW saturation; x86 with VNNI uses reduce_range=False; ARM uses QInt8/QInt8; QDQ is the default since ORT 1.11 and CPU kernels are optimized for it, QOperator is for specific hardware backends. | SOURCE AUDIT 2026-10-03: the 'published' field held a verbatim copy of this notes paragraph instead of a date (field corruption across the qlab2-* batch); published set to null because a GitHub README, doc page or source header has no reliable publication date of its own. Content of the document itself was re-verified live 2026-10-03 and the notes are correct.

Notes

The quantization tool directory listing and README. Adds the per-platform parameter table: x86 without VNNI needs QUInt8/QInt8 + reduce_range=True (7-bit weights) to dodge VPMADDUBSW saturation; x86 with VNNI uses reduce_range=False; ARM uses QInt8/QInt8; QDQ is the default since ORT 1.11 and CPU kernels are optimized for it, QOperator is for specific hardware backends. | SOURCE AUDIT 2026-10-03: the 'published' field held a verbatim copy of this notes paragraph instead of a date (field corruption across the qlab2-* batch); published set to null because a GitHub README, doc page or source header has no reliable publication date of its own. Content of the document itself was re-verified live 2026-10-03 and the notes are correct.

Cross-references

This record references no other records.