diff --git a/paddle/operators/activation_op.cc b/paddle/operators/activation_op.cc
index 154c618e8e7c4650b7f22684d3357de9c52a416c..83262f950e1606cc367ea30b089a983db9848988 100644
--- a/paddle/operators/activation_op.cc
+++ b/paddle/operators/activation_op.cc
@@ -44,9 +44,9 @@ class SigmoidOpMaker : public framework::OpProtoAndCheckerMaker {
     AddInput("X", "Input of Sigmoid operator");
     AddOutput("Y", "Output of Sigmoid operator");
     AddComment(R"DOC(
-Sigmoid Activation Operator.
+Sigmoid Activation Operator
 
-$y = 1 / (1 + e^{-x})$
+$$y = \frac{1}{1 + e^{-x}}$$
 
 )DOC");
   }
@@ -60,9 +60,9 @@ class LogSigmoidOpMaker : public framework::OpProtoAndCheckerMaker {
     AddInput("X", "Input of LogSigmoid operator");
     AddOutput("Y", "Output of LogSigmoid operator");
     AddComment(R"DOC(
-Logsigmoid Activation Operator.
+Logsigmoid Activation Operator
 
-$y = \log(1 / (1 + e^{-x}))$
+$$y = \log \frac{1}{1 + e^{-x}}$$
 
 )DOC");
   }
diff --git a/paddle/operators/adamax_op.cc b/paddle/operators/adamax_op.cc
index d5bbc672e18f392d6a91383b919fefc4b2d8ff0e..867ddd9790722be479ab2cd7c14425ea399c18f9 100644
--- a/paddle/operators/adamax_op.cc
+++ b/paddle/operators/adamax_op.cc
@@ -107,10 +107,12 @@ Adam algorithm based on the infinity norm.
 
 Adamax updates:
 
-$$momentOut = \beta_1 * moment + (1 - \beta_1) * grad \break
-infNormOut = max(\beta_2 * infNorm + \epsilon, |grad|) \break
-learningRate = learningRate /(1 - \beta_1_{pow}) \break
-paramOut = param - learningRate * momentPut / infNormOut$$
+$$
+  momentOut = \beta_{1} * moment + (1 - \beta_{1}) * grad \\
+  infNormOut = max(\beta_{2} * infNorm + \epsilon, |grad|) \\
+  learningRate = \frac{learningRate}{1 - \beta_{1}^{Beta1Pow}} \\
+  paramOut = param - learningRate * \frac{momentOut}{infNormOut}
+$$
 
 The original paper does not have an epsilon attribute.
 However, it is added here for numerical stability to prevent the