diff --git a/develop/doc/_sources/design/support_new_device.md.txt b/develop/doc/_sources/design/support_new_device.md.txt
index b154f01d32c5952367f075a732e2f7729d3c8626..8983df900460127fc130043c52373dab505363ba 100644
--- a/develop/doc/_sources/design/support_new_device.md.txt
+++ b/develop/doc/_sources/design/support_new_device.md.txt
@@ -174,7 +174,7 @@ class MaxOutFunctor {
 };
 ```
 
-CPU implemention is in .cc file
+CPU implementation is in .cc file
 
 ```
 template <typename T>
@@ -188,7 +188,7 @@ class MaxOutFunctor<platform::CPUDeviceContext, T> {
 };
 ```
 
-CUDA implemention is in .cu file
+CUDA implementation is in .cu file
 
 ```
 template <typename T>
@@ -203,9 +203,9 @@ class MaxOutFunctor<platform::CUDADeviceContext, T> {
 ```
 
 
-We first obtain the computing handle from a concrete DeviceContext, and then compute on tensors.
+We first obtain the computing handle from a concrete DeviceContext and then compute on tensors.
 
-The implemention of `OpKernel` is similar to math functors, the extra thing we need to do is to register the OpKernel in a global map.
+The implementation of `OpKernel` is similar to math functors, the extra thing we need to do is to register the OpKernel in a global map.
 
 Fluid provides different register interfaces in op_registry.h
 
diff --git a/develop/doc/design/support_new_device.html b/develop/doc/design/support_new_device.html
index 9274dfca437a833c2dd3170746d24d89ac5ebf43..ae5d59ee67073eda6c728bd3d0ee2d8c41da19fa 100644
--- a/develop/doc/design/support_new_device.html
+++ b/develop/doc/design/support_new_device.html
@@ -358,7 +358,7 @@
 <span class="p">};</span>
 </pre></div>
 </div>
-<p>CPU implemention is in .cc file</p>
+<p>CPU implementation is in .cc file</p>
 <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">template</span> <span class="o">&lt;</span><span class="n">typename</span> <span class="n">T</span><span class="o">&gt;</span>
 <span class="k">class</span> <span class="nc">MaxOutFunctor</span><span class="o">&lt;</span><span class="n">platform</span><span class="p">::</span><span class="n">CPUDeviceContext</span><span class="p">,</span> <span class="n">T</span><span class="o">&gt;</span> <span class="p">{</span>
   <span class="n">public</span><span class="p">:</span>
@@ -370,7 +370,7 @@
 <span class="p">};</span>
 </pre></div>
 </div>
-<p>CUDA implemention is in .cu file</p>
+<p>CUDA implementation is in .cu file</p>
 <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">template</span> <span class="o">&lt;</span><span class="n">typename</span> <span class="n">T</span><span class="o">&gt;</span>
 <span class="k">class</span> <span class="nc">MaxOutFunctor</span><span class="o">&lt;</span><span class="n">platform</span><span class="p">::</span><span class="n">CUDADeviceContext</span><span class="p">,</span> <span class="n">T</span><span class="o">&gt;</span> <span class="p">{</span>
  <span class="n">public</span><span class="p">:</span>
@@ -382,8 +382,8 @@
 <span class="p">};</span>                  
 </pre></div>
 </div>
-<p>We first obtain the computing handle from a concrete DeviceContext, and then compute on tensors.</p>
-<p>The implemention of <code class="docutils literal"><span class="pre">OpKernel</span></code> is similar to math functors, the extra thing we need to do is to register the OpKernel in a global map.</p>
+<p>We first obtain the computing handle from a concrete DeviceContext and then compute on tensors.</p>
+<p>The implementation of <code class="docutils literal"><span class="pre">OpKernel</span></code> is similar to math functors, the extra thing we need to do is to register the OpKernel in a global map.</p>
 <p>Fluid provides different register interfaces in op_registry.h</p>
 <p>Let&#8217;s take <a class="reference external" href="https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/crop_op.cc#L134">Crop</a> operator as an example:</p>
 <p>In .cc file:</p>
diff --git a/develop/doc_cn/_sources/design/support_new_device.md.txt b/develop/doc_cn/_sources/design/support_new_device.md.txt
index b154f01d32c5952367f075a732e2f7729d3c8626..8983df900460127fc130043c52373dab505363ba 100644
--- a/develop/doc_cn/_sources/design/support_new_device.md.txt
+++ b/develop/doc_cn/_sources/design/support_new_device.md.txt
@@ -174,7 +174,7 @@ class MaxOutFunctor {
 };
 ```
 
-CPU implemention is in .cc file
+CPU implementation is in .cc file
 
 ```
 template <typename T>
@@ -188,7 +188,7 @@ class MaxOutFunctor<platform::CPUDeviceContext, T> {
 };
 ```
 
-CUDA implemention is in .cu file
+CUDA implementation is in .cu file
 
 ```
 template <typename T>
@@ -203,9 +203,9 @@ class MaxOutFunctor<platform::CUDADeviceContext, T> {
 ```
 
 
-We first obtain the computing handle from a concrete DeviceContext, and then compute on tensors.
+We first obtain the computing handle from a concrete DeviceContext and then compute on tensors.
 
-The implemention of `OpKernel` is similar to math functors, the extra thing we need to do is to register the OpKernel in a global map.
+The implementation of `OpKernel` is similar to math functors, the extra thing we need to do is to register the OpKernel in a global map.
 
 Fluid provides different register interfaces in op_registry.h
 
diff --git a/develop/doc_cn/design/support_new_device.html b/develop/doc_cn/design/support_new_device.html
index c3cc7a7171f4329e13e898be46acee05408cf4df..892fdb8f2f0130e6cea16dd24f133478d6ee36bd 100644
--- a/develop/doc_cn/design/support_new_device.html
+++ b/develop/doc_cn/design/support_new_device.html
@@ -377,7 +377,7 @@
 <span class="p">};</span>
 </pre></div>
 </div>
-<p>CPU implemention is in .cc file</p>
+<p>CPU implementation is in .cc file</p>
 <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">template</span> <span class="o">&lt;</span><span class="n">typename</span> <span class="n">T</span><span class="o">&gt;</span>
 <span class="k">class</span> <span class="nc">MaxOutFunctor</span><span class="o">&lt;</span><span class="n">platform</span><span class="p">::</span><span class="n">CPUDeviceContext</span><span class="p">,</span> <span class="n">T</span><span class="o">&gt;</span> <span class="p">{</span>
   <span class="n">public</span><span class="p">:</span>
@@ -389,7 +389,7 @@
 <span class="p">};</span>
 </pre></div>
 </div>
-<p>CUDA implemention is in .cu file</p>
+<p>CUDA implementation is in .cu file</p>
 <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">template</span> <span class="o">&lt;</span><span class="n">typename</span> <span class="n">T</span><span class="o">&gt;</span>
 <span class="k">class</span> <span class="nc">MaxOutFunctor</span><span class="o">&lt;</span><span class="n">platform</span><span class="p">::</span><span class="n">CUDADeviceContext</span><span class="p">,</span> <span class="n">T</span><span class="o">&gt;</span> <span class="p">{</span>
  <span class="n">public</span><span class="p">:</span>
@@ -401,8 +401,8 @@
 <span class="p">};</span>                  
 </pre></div>
 </div>
-<p>We first obtain the computing handle from a concrete DeviceContext, and then compute on tensors.</p>
-<p>The implemention of <code class="docutils literal"><span class="pre">OpKernel</span></code> is similar to math functors, the extra thing we need to do is to register the OpKernel in a global map.</p>
+<p>We first obtain the computing handle from a concrete DeviceContext and then compute on tensors.</p>
+<p>The implementation of <code class="docutils literal"><span class="pre">OpKernel</span></code> is similar to math functors, the extra thing we need to do is to register the OpKernel in a global map.</p>
 <p>Fluid provides different register interfaces in op_registry.h</p>
 <p>Let&#8217;s take <a class="reference external" href="https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/crop_op.cc#L134">Crop</a> operator as an example:</p>
 <p>In .cc file:</p>