From 0c18fdc70f600061a8dc46a48789cbf608c5fe82 Mon Sep 17 00:00:00 2001 From: daquexian Date: Thu, 2 May 2019 22:48:19 +0800 Subject: [PATCH] Update README --- README.md | 4 ++-- README_CN.md | 4 +++- 2 files changed, 5 insertions(+), 3 deletions(-) diff --git a/README.md b/README.md index 7c2aaef..2c1a33a 100644 --- a/README.md +++ b/README.md @@ -5,7 +5,7 @@ *Enjoy binary neural networks on mobile!* -[中文](README_CN.md) +[English](README.md) [中文](README_CN.md) QQ Group (Chinese): 1021964010, answer: nndab @@ -37,7 +37,7 @@ BM_bireal18_imagenet 61809506 ns 61056865 ns 10 <--- Bi-r BM_bireal18_imagenet_stem 43279353 ns 41533009 ns 14 <--- Bi-real Net 18 with stem module (The network structure will be described in detail in the coming paper), 56.4% top-1 on ImageNet ``` -Comparison on Google Pixel 1 with [Caffe](http://caffe.berkeleyvision.org) (full precision), [TensorFlow Lite](https://www.tensorflow.org/lite) (full precision) and [BMXNet](https://github.com/hpi-xnor/BMXNet) (binary). We surprisingly observe that BMXNet is even slower the full precision TensorFlow Lite. It suggests that the potential of binary neural networks is far from exploited before dabnn. +The following is the comparison between our dabnn and [Caffe](http://caffe.berkeleyvision.org) (full precision), [TensorFlow Lite](https://www.tensorflow.org/lite) (full precision) and [BMXNet](https://github.com/hpi-xnor/BMXNet) (binary). We surprisingly observe that BMXNet is even slower the full precision TensorFlow Lite. It suggests that the potential of binary neural networks is far from exploited until our dabnn is published. ![Comparison](images/comparison.png) diff --git a/README_CN.md b/README_CN.md index c8596bb..54be2b3 100644 --- a/README_CN.md +++ b/README_CN.md @@ -3,6 +3,8 @@ [![License](https://img.shields.io/badge/license-BSD--3--Clause-blue.svg)](LICENSE) [![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](https://github.com/JDAI-CV/dabnn/pulls) +[English](README.md) [中文](README_CN.md) + QQ 群：1021964010, 入群答案: nndab ## 简介 @@ -35,7 +37,7 @@ BM_bireal18_imagenet 61809506 ns 61056865 ns 10 <--- Bi-r BM_bireal18_imagenet_stem 43279353 ns 41533009 ns 14 <--- 带有 stem 模块的 Bi-real Net 18 (将在论文中描述), ImageNet top-1 为 56.4% ``` -在 Google Pixel 1 上与 [Caffe](http://caffe.berkeleyvision.org)（全精度）, [TensorFlow Lite](https://www.tensorflow.org/lite)（全精度）和 [BMXNet](https://github.com/hpi-xnor/BMXNet)（二值）的对比如下。我们很惊讶的发现现有的二值 inference 框架 BMXNet 甚至比全精度的 TensorFlow Lite 还要慢，这表明在 dabnn 之前，二值网络的潜力远远没有被挖掘出来。 +在 Google Pixel 1 上与 [Caffe](http://caffe.berkeleyvision.org)（全精度）, [TensorFlow Lite](https://www.tensorflow.org/lite)（全精度）和 [BMXNet](https://github.com/hpi-xnor/BMXNet)（二值）的对比如下。我们很惊讶的发现现有的二值 inference 框架 BMXNet 甚至比全精度的 TensorFlow Lite 还要慢，这表明，直到 dabnn 推出之前，二值网络的潜力都远远没有被挖掘出来。 ![Comparison](images/comparison.png)