diff --git a/docs/mindspore/source_en/api_python/env_var_list.rst b/docs/mindspore/source_en/api_python/env_var_list.rst
index 7bb5387c95ebf5d2620fc3cd70a45ea6fdf0cd10..2bcad6cd194debf3038d3899adbb92816d77489f 100644
--- a/docs/mindspore/source_en/api_python/env_var_list.rst
+++ b/docs/mindspore/source_en/api_python/env_var_list.rst
@@ -244,7 +244,7 @@ Graph Compilation and Execution
        0: not enabled.
 
        Default value: 0
-     - Ascend AI processor environment and graph compilation grade O2 process use only.
+     - Ascend AI processor environment and GE backend use only.
    * - MS_ENABLE_GRACEFUL_EXIT
      - Enable training process exit gracefully
      - Integer
@@ -982,7 +982,7 @@ For more information about CANN's environment variables, see `Ascend community <
      - Value Range
      - Description
    * - MS_FORMAT_MODE
-     - Set the default preferred format for Ascend and graph compilation grade O2 processes, with the entire network set to ND format
+     - Set the default preferred format for Ascend and GE backend, with the entire network set to ND format
      - Integer
      - 1: The operator prioritizes the ND format.
 
@@ -991,7 +991,7 @@ For more information about CANN's environment variables, see `Ascend community <
        Default value: 1
      - This environment variable affects the choice of format for the operator, which has an impact on network execution performance and memory usage, and can be tested by setting this option to get a better choice of operator format in terms of performance and memory.
 
-       Ascend AI processor environment and graph compilation grade O2 processes only.
+       Ascend AI processor environment and GE backend only.
 
 Profiler
 -----------
diff --git a/docs/mindspore/source_zh_cn/api_python/env_var_list.rst b/docs/mindspore/source_zh_cn/api_python/env_var_list.rst
index 3cea1274e5f6c7a7a96ef796e14fb95d2c23cf05..0d9cd43cbd1a6961a50775163ce4744cb942ff0a 100644
--- a/docs/mindspore/source_zh_cn/api_python/env_var_list.rst
+++ b/docs/mindspore/source_zh_cn/api_python/env_var_list.rst
@@ -244,7 +244,7 @@
        0：不使能。
 
        默认值：0
-     - 仅限Ascend AI处理器环境图编译等级为O2流程使用。
+     - 仅限Ascend AI处理器环境GE后端流程使用。
    * - MS_ENABLE_GRACEFUL_EXIT
      - 设置使能进程优雅退出
      - Integer
@@ -978,7 +978,7 @@ CANN的环境变量详见 `昇腾社区 <https://www.hiascend.com/document/detai
      - 取值
      - 说明
    * - MS_FORMAT_MODE
-     - 设置Ascend 图编译等级为O2流程的默认优选格式，整网设置为ND格式
+     - 设置Ascend GE后端流程的默认优选格式，整网设置为ND格式
      - Integer
      - 1: 算子优先选择ND格式。
 
@@ -987,7 +987,7 @@ CANN的环境变量详见 `昇腾社区 <https://www.hiascend.com/document/detai
        默认值：1。
      - 此环境变量影响算子的format选择，从而对网络执行性能和内存占用产生影响，可通过设置此选项测试得到性能和内存更优的算子格式选择。
 
-       仅限Ascend AI处理器环境图编译等级为O2流程使用。
+       仅限Ascend AI处理器环境GE后端使用。
 
 Profiler
 -----------
diff --git a/tutorials/source_en/custom_program/operation/op_custom_ascendc.md b/tutorials/source_en/custom_program/operation/op_custom_ascendc.md
index cbb003b75f332d864217068c9e3498e1ec6f3a62..d183988abfb057028eadfe4e8868c76e44ab21da 100644
--- a/tutorials/source_en/custom_program/operation/op_custom_ascendc.md
+++ b/tutorials/source_en/custom_program/operation/op_custom_ascendc.md
@@ -104,7 +104,7 @@ ops.Custom(func, bprop=None, out_dtype=None, func_type='aot', out_shape=None, re
 - `bprop`(function)：Backpropagation function for the custom operator. Default value: `None`.
 - `reg_info`(Union[str, dict, list, tuple])：Registration information for the custom operator. Default value: `None`. Ascend C custom operators do not need to pass this parameter and can use the default value.
 
-**Scenario Limitations**： Currently, dynamic graphs and static graphs in O2 mode only support input and output of Tensor types. Static graphs in O0/O1 modes have no type restrictions. For dynamic graph scenarios with Ascend C custom operators, it is recommended to use [CustomOpBuilder-Based Custom Operators](https://www.mindspore.cn/tutorials/en/master/custom_program/operation/op_customopbuilder.html).
+**Scenario Limitations**： Currently, dynamic graphs and static graphs in GE backend only support input and output of Tensor types. Static graphs in O0/O1 modes have no type restrictions. For dynamic graph scenarios with Ascend C custom operators, it is recommended to use [CustomOpBuilder-Based Custom Operators](https://www.mindspore.cn/tutorials/en/master/custom_program/operation/op_customopbuilder.html).
 
 ### Simple Example
 
diff --git a/tutorials/source_zh_cn/custom_program/operation/op_custom_ascendc.md b/tutorials/source_zh_cn/custom_program/operation/op_custom_ascendc.md
index 39ce6392f66aa29daf25c728fa7417035cb2fca3..a1d6adb4befc628a020c7bd7be12e873afe8ba2b 100644
--- a/tutorials/source_zh_cn/custom_program/operation/op_custom_ascendc.md
+++ b/tutorials/source_zh_cn/custom_program/operation/op_custom_ascendc.md
@@ -110,7 +110,7 @@ ops.Custom(func, bprop=None, out_dtype=None, func_type='aot', out_shape=None, re
 - `bprop`(function): 自定义算子的反向函数。默认值： `None`。
 - `reg_info`(Union[str, dict, list, tuple]): 自定义算子的算子注册信息。默认值： `None`。Ascend C自定义算子无需传入该参数，使用默认值。
 
-**场景限制**： 当前动态图和静态图O2模式只支持输入输出为Tensor类型，静态图O0/O1模式无限制类型。Ascend
+**场景限制**： 当前动态图和静态图GE后端只支持输入输出为Tensor类型，静态图O0/O1模式无限制类型。Ascend
 C自定义算子动态图场景推荐使用[基于CustomOpBuilder的自定义算子](https://www.mindspore.cn/tutorials/zh-CN/master/custom_program/operation/op_customopbuilder.html)。
 
 ### 简单示例