diff --git a/official/nlp/lstm/README.md b/official/nlp/lstm/README.md
index 01c3055c812a3842c01b666bd0d43c2671b11d72..5ae345ca4658a9deef480cec08a82a0e11c773b4 100644
--- a/official/nlp/lstm/README.md
+++ b/official/nlp/lstm/README.md
@@ -419,7 +419,7 @@ Ascend:
 python export.py --ckpt_file [CKPT_PATH] --file_name [FILE_NAME] --file_format [FILE_FORMAT] --config_path [YAML_CONFIG_PATH]
 ```
 
-- `weight.txt` is required, please generate it by run preprocess.py. Then you will see this file in /preprocess.
+- `weight.txt` is required, it will be generated with mindrecords by training model.
 - `ckpt_file` parameter is required.
 - `FILE_FORMAT` should be in ["AIR", "MINDIR", "ONNX"].
 - `YAML_CONFIG_PATH` default is `default_config.yaml`.
diff --git a/official/nlp/lstm/README_CN.md b/official/nlp/lstm/README_CN.md
index 7972a64fb59a999a2f4e9553c6b02867b29fc85b..fb7e133b5c586a792969bf2f6389a26eeefa628d 100644
--- a/official/nlp/lstm/README_CN.md
+++ b/official/nlp/lstm/README_CN.md
@@ -421,7 +421,7 @@ Ascend:
 python export.py --ckpt_file [CKPT_PATH] --file_name [FILE_NAME] --file_format [FILE_FORMAT] --config_path [YAML_CONFIG_PATH]
 ```
 
-- `weight.txt` 文件在导出脚本中要用到，需要运行preprocess.py文件生成。
+- `weight.txt` 文件需要在训练模型时获得。
 - `ckpt_file` 是必需的。
 - `FILE_FORMAT` 必须在 ["AIR", "MINDIR", "ONNX"]中进行选择。
 - `YAML_CONFIG_PATH` 默认是 `default_config.yaml`。
diff --git a/official/nlp/lstm/preprocess.py b/official/nlp/lstm/preprocess.py
index 6a1b82e98f3a1c3a3c8e7913d686fb339377a53f..9c39fd03b56b4aefa9d22882a5def0bb8db50ea5 100644
--- a/official/nlp/lstm/preprocess.py
+++ b/official/nlp/lstm/preprocess.py
@@ -18,13 +18,12 @@
 import os
 import numpy as np
 
-from src.dataset import lstm_create_dataset, convert_to_mindrecord
+from src.dataset import lstm_create_dataset
 from src.model_utils.config import config
 
 
 if __name__ == '__main__':
     print("============== Starting Data Pre-processing ==============")
-    convert_to_mindrecord(config.embed_size, config.aclimdb_path, config.preprocess_path, config.glove_path)
     dataset = lstm_create_dataset(config.preprocess_path, config.batch_size, training=False)
     img_path = os.path.join(config.result_path, "00_data")
     os.makedirs(img_path)