GitHub

ExecuTorch XNNPACK Delegate

This subtree contains the XNNPACK Delegate implementation for ExecuTorch. XNNPACK is an optimized library of neural network inference operators for ARM and x86 CPUs. It is an open source project used ...
Neowin

Researchers accelerate sparse inference on XNNPack and TensorFlow Lite for realtime apps

XNNPack and TensorFlow Lite now support efficient inference of sparse networks. Researchers demonstrated substantial speedups in inference times on realtime gesture detection and background blur apps.
GitHub

PyTorch-executorch /docs /source

The XNNPACK delegate is the ExecuTorch solution for CPU execution on mobile CPUs. XNNPACK is a library that provides optimized kernels for machine learning operators on Arm and x86 CPUs. To target the ...
GIGAZINE

Googleがスパース推論でニューラルネットワークの高速化を達成した ...

Googleがモバイル向けの機械学習用ソフトウェアライブラリTensorFlow Liteとニューラルネットワーク推論最適化ライブラリのXNNPACKにさらなるスパース化最適を組み込むことで、推論速度を大幅に向上させたと報告しました。 Google AI Blog: Accelerating Neural Networks on ...