diff --git a/generate_sequence_by_rnn_lm/README_en.md b/generate_sequence_by_rnn_lm/README_en.md
index 41c000d9e66ffc74587eaa9fe98c6804f093abf9..0dfb9a868e9759d30a60174537a9bddbb76d6b3b 100644
--- a/generate_sequence_by_rnn_lm/README_en.md
+++ b/generate_sequence_by_rnn_lm/README_en.md
@@ -115,8 +115,8 @@ model_path = "models/rnn_lm_pass_00000.tar.gz" ``
 3. ```max_gen_len``` : Specify the maximum length of each sentence generated. If the model cannot generate ```<e>```, the generation process will automatically terminate when ```max_gen_len``` words are generated.
 4. ```beam_size``` : The width of each step of the Beam Search algorithm.
 5. ```model_path``` : Specify the path to the trained model.
-    Among them, the gen_file holds the text prefix to be generated, and each prefix is in one line. The format is as follows:
-	```若隐若现 地像 幽灵 , 像 死神```
+  Among them, the gen_file holds the text prefix to be generated, and each prefix is in one line. The format is as follows:
+  ```若隐若现 地像 幽灵 , 像 死神```
 Write the text prefix to be generated into the file in this format;
 * Run the ```python generate.py``` command to run the beam search algorithm to generate the text for the input prefix. Here are the results generated by the model :
 ```81    若隐若现 地像 幽灵 , 像 死神