Quantization Guide

Jump to bottom

Abhiram S edited this page Sep 23, 2025 · 4 revisions

Quantization Guide

Symmetric quantization flows with integer GEMM.

Integer GEMM Variants

See: GEMM API Reference

Symmetric Quant APIs

s8s8s32→f32 (sym): anchor on GEMM page
s8s8s32→bf16 (sym): anchor on GEMM page

Tips

Calibrate scales and zero-points
Validate accuracy vs. float baselines

Home | Quick Start | API Reference | Report Issue | Source Code