!923 Remove expired data for graph kernel.

Merge pull request !923 from TronZhang/graph_kernel_mod

tutorials/source_en/advanced_use/graph_kernel_fusion.md

@@ -10,7 +10,6 @@
     - [Sample Scripts](#sample-scripts)
   - [Effect Evaluation](#effect-evaluation)
     - [Computational Graph](#computational-graph)
-    - [Training Time for One Step](#training-time-for-one-step)
 
 <!-- /TOC -->
 
@@ -148,7 +147,3 @@ To verify whether the graph kernel fusion takes effect, you can compare the chan
     ![Composite operator fusion](./images/graph_kernel_fusion_example_fuse_composite_after.png)
 
     Figure 5 Computational graph after composite operator fusion
-  
-### Training Time for One Step
-
-BERT-large scenario: After the graph kernel fusion function is enabled for the BERT-large network, the training time for one step can be improved by more than 10% while the accuracy is the same as that before the function is enabled.
\ No newline at end of file

tutorials/source_zh_cn/advanced_use/graph_kernel_fusion.md

@@ -10,7 +10,6 @@
     - [样例脚本](#样例脚本)
   - [效果评估](#效果评估)
     - [计算图](#计算图)
-    - [训练单step时间](#训练单step时间)
 
 <!-- /TOC -->
 
@@ -148,7 +147,3 @@ context.set_context(enable_graph_kernel=True)
     ![组合算子融合](./images/graph_kernel_fusion_example_fuse_composite_after.png)
 
     图5：组合算子融合后计算图
-  
-### 训练单step时间
-
-BERT-large场景：BERT-large网络启用图算融合后，在保持与启用前精度一致的前提下，训练的单step时间可提升10%以上。