diff --git a/doc/fluid/design/dist_train/prefetch_parameter.md b/doc/fluid/design/dist_train/prefetch_parameter.md
index b3e2210fdff26b06cb4fc5e6f6ffdef3ab8b20f8..7cea942279dc6da0dfb18079d82963a5b5086ea3 100644
--- a/doc/fluid/design/dist_train/prefetch_parameter.md
+++ b/doc/fluid/design/dist_train/prefetch_parameter.md
@@ -32,7 +32,7 @@ The processing flow of `lookup remote table` is as follows:
 
     - **Distributed Transpiler** would split the large parameters
       (`weight`) into some partitioned parameters (`weight_0`, `weight_1`, `weight_2`) as shown in the figure above.
-    - We can use `round-robin` to distribute the partitioned parameter.
+    - We can use some algorithms to distribute the partitioned parameters, such as `round-robin` or `hash-name`.
 
 1. pre-fetching parameter at the beginning of each mini-batch
 
diff --git a/doc/fluid/design/dist_train/src/split_parameter.graffle b/doc/fluid/design/dist_train/src/split_parameter.graffle
index f020c0291e99f758f532d2611e72003478656750..db1f8dc2c7888d0f71337142bcccd0faeb09efe8 100644
Binary files a/doc/fluid/design/dist_train/src/split_parameter.graffle and b/doc/fluid/design/dist_train/src/split_parameter.graffle differ