diff --git a/paddle/operators/math/matmul.h b/paddle/operators/math/matmul.h
index 98a98e4b994bad5efb361505c08cbb53b78e2b08..8a63d204cb0647f19631eb3b59205df1d45bb7c4 100644
--- a/paddle/operators/math/matmul.h
+++ b/paddle/operators/math/matmul.h
@@ -41,22 +41,18 @@ class MatMulFunctor {
                       "Input tensor a must be at least 1-dimensional.");
     PADDLE_ENFORCE_GE(dim_b.size(), 1,
                       "Input tensor b must be at least 1-dimensional.");
-    PADDLE_ENFORCE_LE(dim_a.size(), 4,
-                      "Input tensor a must be at most 4-dimensional.");
-    PADDLE_ENFORCE_LE(dim_b.size(), 4,
-                      "Input tensor b must be at most 4-dimensional.");
 
     std::vector<int64_t> out_dim;
     int64_t batch_count = 1;
     if (dim_a.size() > 3) {
       PADDLE_ENFORCE(dim_b.size() > 3,
                      "The dimensions of X and Y must be the same, and both of "
-                     "them should be 4-dimensional.");
+                     "them should be %d-dimensional.",
+                     dim_b.size());
       for (int j = 0; j < dim_a.size() - 2; ++j) {
-        PADDLE_ENFORCE(
-            dim_b[j] == dim_a[j],
-            "The dimensions of X and Y must be the same, and both of "
-            "them should be 4-dimensional.");
+        PADDLE_ENFORCE(dim_b[j] == dim_a[j],
+                       "The dimensions of X[%d] and Y[%d] must be the same.", j,
+                       j);
         out_dim.push_back(dim_a[j]);
         batch_count *= dim_a[j];
       }
diff --git a/paddle/operators/matmul_op.cc b/paddle/operators/matmul_op.cc
index 155346db41d8310e6095642db9f34a3b283a041a..6ced0ef6c0a0648fe9b8ecb2d8034a6c043e35e5 100644
--- a/paddle/operators/matmul_op.cc
+++ b/paddle/operators/matmul_op.cc
@@ -13,7 +13,6 @@ See the License for the specific language governing permissions and
 limitations under the License. */
 
 #include "paddle/operators/matmul_op.h"
-#include <vector>
 
 namespace paddle {
 namespace operators {
@@ -42,22 +41,18 @@ class MatMulOp : public framework::OperatorWithKernel {
                       "Input tensor X must be at least 1-dimensional.");
     PADDLE_ENFORCE_GE(dim_y.size(), 1,
                       "Input tensor Y must be at least 1-dimensional.");
-    PADDLE_ENFORCE_LE(dim_x.size(), 4,
-                      "Input tensor X must be at most 4-dimensional.");
-    PADDLE_ENFORCE_LE(dim_y.size(), 4,
-                      "Input tensor Y must be at most 4-dimensional.");
 
     std::vector<int64_t> out_dim;
     int64_t batch_count = 1;
     if (dim_x.size() > 3) {
       PADDLE_ENFORCE(dim_y.size() == dim_x.size(),
                      "The dimensions of X and Y must be the same, and both of "
-                     "them should be 4-dimensional.");
+                     "them should be %d-dimensional.",
+                     dim_x.size());
       for (int j = 0; j < dim_x.size() - 2; ++j) {
-        PADDLE_ENFORCE(
-            dim_y[j] == dim_x[j],
-            "The dimensions of X and Y must be the same, and both of "
-            "them should be 4-dimensional.");
+        PADDLE_ENFORCE(dim_y[j] == dim_x[j],
+                       "The dimensions of X[%d] and Y[%d] must be the same.", j,
+                       j);
         out_dim.push_back(dim_x[j]);
         batch_count *= dim_x[j];
       }
diff --git a/paddle/operators/matmul_op.h b/paddle/operators/matmul_op.h
index 11266db4b9a7e31b42c56cc4a9ea7175f555786a..9f06791f7b16f2d08473fc37e1a11c94b7452a8d 100644
--- a/paddle/operators/matmul_op.h
+++ b/paddle/operators/matmul_op.h
@@ -137,6 +137,12 @@ class MatMulGradKernel : public framework::OpKernel<T> {
       y_dims.push_back(1);
     }
 
+    int batch_count = 0;
+    //
+    if (x_dims.size() > 3) {
+      batch_count = accumulate(x_dims.begin(), x_dims.end() - 2, 1,
+                               std::multiplies<int>());
+    }
     // Fix the dOut dimensions.
     int M = 0, N = 0, batchCountX = 0, batchCountY = 0;
 
@@ -149,8 +155,7 @@ class MatMulGradKernel : public framework::OpKernel<T> {
         M = transpose_x ? x_dims[2] : x_dims[1];
         break;
       default:
-        batchCountX = accumulate(x_dims.begin(), x_dims.end() - 2, 1,
-                                 std::multiplies<int>());
+        batchCountX = batch_count;
         size_t mat_s = x_dims.size() - 2;
         M = transpose_x ? x_dims[mat_s + 1] : x_dims[mat_s];
     }
@@ -164,8 +169,7 @@ class MatMulGradKernel : public framework::OpKernel<T> {
         N = transpose_y ? y_dims[1] : y_dims[2];
         break;
       default:
-        batchCountY = accumulate(y_dims.begin(), y_dims.end() - 2, 1,
-                                 std::multiplies<int>());
+        batchCountY = batch_count;
         size_t mat_s = y_dims.size() - 2;
         N = transpose_y ? y_dims[mat_s] : y_dims[mat_s + 1];
     }
@@ -180,8 +184,6 @@ class MatMulGradKernel : public framework::OpKernel<T> {
     if (batchCount) {
       if (x_dims.size() > 3) {
         dout_dims.insert(dout_dims.begin(), x_dims.begin(), x_dims.end() - 2);
-      } else if (y_dims.size() > 3) {
-        dout_dims.insert(dout_dims.begin(), y_dims.begin(), y_dims.end() - 2);
       } else {
         dout_dims.insert(dout_dims.begin(), batchCount);
       }