Fix typo and remove redundant code (#1015)

fix typo

* Update quant_tool_int8.cpp

doc/docs_en/develop_guides/add_hardware.md

@@ -276,4 +276,4 @@ Finally, you need to write the registration function and the de-registration fun
 From the above description, we can know that the core work of adding a custom device is to fill the `ir_device_t` structure. After the description is completed, all the work of device registration is completed. The modular `device` makes **Tengine Lite** very easy to expand and has enough flexibility.
 
 ## Surprise
-In the `init_tengine(void)` function, when the `operator prototype` completes the registration, the registered ones are `serializer` and `devices`, but the function does not jump in the static code state, and the user can install an integrated development environment , Such as `Microsoft Visual Studio` or `Jetbrains Clion`, after opening the folder and generating the `CMake` process, you can jump.
+In the `init_tengine(void)` function, when the `operator prototype` completes the registration, the registered ones are `serializer` and `devices`, but the function does not jump in the static code state, and the user can install an integrated development environment , Such as `Microsoft Visual Studio` or `JetBrains CLion`, after opening the folder and generating the `CMake` process, you can jump.
\ No newline at end of file

doc/docs_en/introduction/roadmap.md

@@ -7,7 +7,7 @@ Update the Flag once every quarter.
 - [x] refactor the framework code
 - [x] refactor the NPU plugin code
 - [x] support VS2017 compile
-- [ ] support MacOS compile
+- [ ] support macOS compile
 - [x] support the mode type of PaddlePaddle
 - [ ] add more examples with NPU platform 
 - [ ] fix the Float32 bugs of Vulkan

doc/docs_zh/develop_guides/add_hardware.md

@@ -277,4 +277,4 @@ static struct optimizer tpu_optimizer = {
 通过上文的描述，可以知道添加一个自定义设备的核心工作就是填充 `ir_device_t` 结构体，描述完成后，设备注册的所有工作就完成了。模块化的 `device` 使得 **Tengine** 非常易于扩展，并有足够的灵活性。
 
 ## 彩蛋
-`init_tengine(void)` 函数中，当 `operator prototype` 完成注册后，注册的就是 `serializer` 和 `devices`，但在静态代码状态下函数并不会跳转，用户可以安装一款集成开发环境，比如 `Microsoft Visual Studio` 或 `Jetbrains Clion`，打开文件夹后生成 `CMake` 过程后即可进行跳转。
+`init_tengine(void)` 函数中，当 `operator prototype` 完成注册后，注册的就是 `serializer` 和 `devices`，但在静态代码状态下函数并不会跳转，用户可以安装一款集成开发环境，比如 `Microsoft Visual Studio` 或 `JetBrains CLion`，打开文件夹后生成 `CMake` 过程后即可进行跳转。

doc/docs_zh/develop_guides/architecture-intro.md

@@ -514,4 +514,4 @@ ENDFUNCTION()
 # generate all serializer
 GENERATE_REGISTER_HEADER_FILE("register_" "unregister_" "" "${_SRL_SRC_ROOT}/register.h.in" "${_SRL_BIN_ROOT}/register.h" "${_SRL_TM2_SRL_SOURCE}")
 ```
-这样就完成了配置过程，生成的头文件进一步的在后续的编译过程中发挥作用。当用户使用静态分析功能的 `IDE` 审阅代码时，由于相关头文件没有生成，所以可能会发生无法跳转的情况。经过编译配置的 `Microsoft Visual Studio Code` 等在打开文件夹后，会启动 `CMake` 进行配置和生成，这时的头文件就会生成，也能进行跳转了。其他 `Microsoft Visual Studio` 或 `Jetbrains Clion` 等 `IDE` 也可以完成配置和生成过程，推荐使用。
+这样就完成了配置过程，生成的头文件进一步的在后续的编译过程中发挥作用。当用户使用静态分析功能的 `IDE` 审阅代码时，由于相关头文件没有生成，所以可能会发生无法跳转的情况。经过编译配置的 `Microsoft Visual Studio Code` 等在打开文件夹后，会启动 `CMake` 进行配置和生成，这时的头文件就会生成，也能进行跳转了。其他 `Microsoft Visual Studio` 或 `JetBrains CLion` 等 `IDE` 也可以完成配置和生成过程，推荐使用。
\ No newline at end of file

doc/docs_zh/introduction/roadmap.md

@@ -7,7 +7,7 @@
 - [x] refactor the framework code
 - [x] refactor the NPU plugin code
 - [x] support VS2017 compile
-- [ ] support MacOS compile
+- [ ] support macOS compile
 - [x] support the mode type of PaddlePaddle
 - [ ] add more examples with NPU platform 
 - [ ] fix the Float32 bugs of Vulkan

tools/convert_tool/ncnn/ncnn2tengine.cpp

@@ -228,7 +228,7 @@ int ncnn_serializer::load_model_file(const char* fname, std::vector<NcnnNode>& n
         {
             bool array_selection = id <= -23300;
 
-            if (node.op == "Input" && array_selection == true)
+            if (node.op == "Input" && array_selection)
             {
                 node.optimized = 1;
             }
@@ -236,7 +236,7 @@ int ncnn_serializer::load_model_file(const char* fname, std::vector<NcnnNode>& n
             {
                 id = -id - 23300;
             }
-            if (node.optimized == 1 && array_selection == true)
+            if (node.optimized == 1 && array_selection)
             {
                 int len = 0;
                 int nscan = fscanf(fp, "%d", &len);
@@ -290,7 +290,7 @@ int ncnn_serializer::load_model_file(const char* fname, std::vector<NcnnNode>& n
             }
             else
             {
-                if (array_selection == true)
+                if (array_selection)
                 {
                     int len = 0;
                     int nscan = fscanf(fp, "%d", &len);
@@ -492,7 +492,7 @@ int ncnn_serializer::load_binary_file(const char* fname, std::vector<NcnnParam>&
         }
         else if (nodelist[i].op == "InnerProduct")
         {
-            NcnnParam weight, bias;
+            NcnnParam weight;
             nscan = read(&magic, sizeof(float));
             weight.name = nodelist[i].name + "_w";
             std::map<int, std::string>::iterator iter;

tools/quantize/quant_tool_int8.cpp

@@ -237,7 +237,7 @@ int QuantTool::activation_quant_tool()
                 float threshold = compute_aciq_gaussian_clip(absmax, emlement_num, 8);
                 act_scale = threshold / 127.f;
 
-                /* the scale of softmax always is scale = 1 / 127.f */
+                /* the scale of softmax is always scale = 1 / 127.f */
                 for (int j = 0; j < ir_graph->node_num; j++)
                 {
                     struct node* noden = ir_graph->node_list[j];
@@ -277,7 +277,7 @@ int QuantTool::activation_quant_tool()
 
                 act_scale = std::max(std::abs(max_activation[i]), std::abs(min_activation[i])) / 127.f;
 
-                /* the scale of softmax always is scale = 1 / 127.f */
+                /* the scale of softmax is always scale = 1 / 127.f */
                 for (int j = 0; j < ir_graph->node_num; j++)
                 {
                     struct node* noden = ir_graph->node_list[j];