diff --git a/fluid/AutoDL/LRC/README.md b/fluid/AutoDL/LRC/README.md
index ed85b6e307142d0d797421ec25b273bc05560e6e..df9af47d4a3876371673cbbfef0ad2553768b9a5 100644
--- a/fluid/AutoDL/LRC/README.md
+++ b/fluid/AutoDL/LRC/README.md
@@ -1,5 +1,8 @@
 # LRC Local Rademachar Complexity Regularization
-This directory contains image classification model based on novel regularizer rooted in Local Rademacher Complexity (LRC). The regularization by LRC and [DARTS](https://arxiv.org/abs/1806.09055) are combined in this model and it achieves 97.3% accuracy on CIFAR-10 dataset.
+Regularization of Deep Neural Networks(DNNs) for the sake of improving their generalization capability is important and chllenging. This directory contains image classification model based on a novel regularizer rooted in Local Rademacher Complexity (LRC). We appreciate the contribution by [DARTS](https://arxiv.org/abs/1806.09055) for our research. The regularization by LRC and DARTS are combined in this model on CIFAR-10 dataset. Code accompanying the paper
+> [An Empirical Study on Regularization of Deep Neural Networks by Local Rademacher Complexity](https://arxiv.org/abs/1902.00873)\
+> Yingzhen Yang, Xingjian Li, Jun Huan.\
+> _arXiv:1902.00873_.
 
 ---
 # Table of Contents
@@ -7,7 +10,6 @@ This directory contains image classification model based on novel regularizer ro
 - [Installation](#installation)
 - [Data preparation](#data-preparation)
 - [Training](#training)
-- [Model performances](#model-performances)
 
 ## Installation
 
@@ -66,9 +68,7 @@ After data preparation, one can start the training step by:
 *  Initalize bias in batch norm and fc to zero constant and do not add bias to conv2d.
 
 
-## Model performances
-Below is the accuracy on CIFAR-10 dataset：
+## Reference
 
-| model | avg top1 | avg top5 |
-| ----- | -------- | -------- |
-| [DARTS-LRC](https://paddlemodels.bj.bcebos.com/autodl/fluid_rademacher.tar.gz) | 97.34 | 99.75 |
+  - DARTS: Differentiable Architecture Search [`paper`](https://arxiv.org/abs/1806.09055)
+  - Differentiable architecture search in PyTorch [`code`](https://github.com/quark0/darts)
diff --git a/fluid/AutoDL/LRC/README_cn.md b/fluid/AutoDL/LRC/README_cn.md
index 7768c39f4e64d0b780395b67aea35d6f897325bb..06dc937074de199af31db97ee200e7690443b1b0 100644
--- a/fluid/AutoDL/LRC/README_cn.md
+++ b/fluid/AutoDL/LRC/README_cn.md
@@ -1,5 +1,8 @@
 # LRC 局部Rademachar复杂度正则化
-本目录包括了一种基于局部rademacher复杂度的新型正则（LRC）的图像分类模型。该模型将LRC正则和[DARTS](https://arxiv.org/abs/1806.09055)网络相结合，在CIFAR-10数据集中得到了97.3%的准确率。
+为了在深度神经网络中提升泛化能力，正则化的选择十分重要也具有挑战性。本目录包括了一种基于局部rademacher复杂度的新型正则（LRC）的图像分类模型。十分感谢[DARTS](https://arxiv.org/abs/1806.09055)模型对本研究提供的帮助。该模型将LRC正则和DARTS网络相结合，在CIFAR-10数据集中得到了很出色的效果。代码和文章一同发布
+> [An Empirical Study on Regularization of Deep Neural Networks by Local Rademacher Complexity](https://arxiv.org/abs/1902.00873)\
+> Yingzhen Yang, Xingjian Li, Jun Huan.\
+> _arXiv:1902.00873_.
 
 ---
 # 内容
@@ -7,7 +10,6 @@
 - [安装](#安装)
 - [数据准备](#数据准备)
 - [模型训练](#模型训练)
-- [模型性能](#模型性能)
 
 ## 安装
 
@@ -63,9 +65,7 @@
 * 对batch norm和全连接层偏差采用固定初始化，不对卷积设置偏差
 
 
-## 模型性能
-下表为该模型在CIFAR-10数据集上的性能：
+## 引用
 
-| 模型 | 平均top1 | 平均top5 |
-| ----- | -------- | -------- |
-| [DARTS-LRC](https://paddlemodels.bj.bcebos.com/autodl/fluid_rademacher.tar.gz) | 97.34 | 99.75 |
+  - DARTS: Differentiable Architecture Search [`论文`](https://arxiv.org/abs/1806.09055)
+  - Differentiable Architecture Search in PyTorch [`代码`](https://github.com/quark0/darts)