This repository has been archived by the owner on Oct 1, 2020. It is now read-only.

Per channel quant#51

Open

mortzur wants to merge 7 commits intopytorch:masterfrom mortzur:per_channel_quant

+2,687-444

Commits on Feb 19, 2019

adding unit-test example of per-channel scale and zero-point
mortzur
committed

Commits on Feb 25, 2019

Adding q8gemm neon ukernel with weights quantization parameters per output channel
mortzur
committed
Revert "adding unit-test example of per-channel scale and zero-point"
mortzur
committed

Commits on Feb 26, 2019

Benchmarks for q8gemm with per-channel kernel quantization parameters
mortzur
committed

Commits on Feb 27, 2019

moving 4x8-neon per channel ukernel to a separate file
mortzur
committed

Commits on Mar 1, 2019

q8gemm per-channel armv7 ukernel
mortzur
committed

Commits on Mar 7, 2019

cleanup comments
mortzur
committed