fix E-List (#7608)

Author: Echo-Nie

_typos.toml

@@ -26,9 +26,6 @@ Nervana = "Nervana"
 
 # These words need to be fixed
 Creenshot = "Creenshot"
-Embeddding = "Embeddding"
-Embeding = "Embeding"
-Engish = "Engish"
 Learing = "Learing"
 Moible = "Moible"
 Operaton = "Operaton"
@@ -57,15 +54,6 @@ dimention = "dimention"
 dimentions = "dimentions"
 dirrectories = "dirrectories"
 disucssion = "disucssion"
-egde = "egde"
-enviornment = "enviornment"
-erros = "erros"
-evalute = "evalute"
-exampels = "exampels"
-exection = "exection"
-exlusive = "exlusive"
-exmaple = "exmaple"
-exsits = "exsits"
 feeded = "feeded"
 flaot = "flaot"
 fliters = "fliters"

ci_scripts/check_api_docs_en.py

@@ -124,6 +124,6 @@ def check_system_message_in_doc(doc_file):
     if error_files:
         print("error files: ", error_files)
         print(
-            "ERROR: these docs exsits System Message: WARNING/ERROR, please check and fix them"
+            "ERROR: these docs exists System Message: WARNING/ERROR, please check and fix them"
         )
         sys.exit(1)

ci_scripts/check_api_docs_en.sh

@@ -13,7 +13,7 @@ function check_system_message(){
     fi
 }
 
-echo "RUN Engish API Docs Checks"
+echo "RUN English API Docs Checks"
 jsonfn=$1
 output_path=$2
 need_check_api_py_files="${3}"

docs/design/dist_train/README.md

@@ -48,7 +48,7 @@ The training process of asynchronous training can be:
     2. Trainer gets all parameters back from pserver.
 
 ### Note:
-There are also some conditions that need to consider. For exmaple:
+There are also some conditions that need to consider. For example:
 
 1. If trainer needs to wait for the pserver to apply it's gradient and then get back the parameters back.
 1. If we need a lock between parameter update and parameter fetch.

docs/design/memory/memory_optimization.md

@@ -60,7 +60,7 @@ We can leran these techniques from compilers. There are mainly two stages to mak
 
 
 #### Control Flow Graph
-To perform analysis on a program, it is often useful to make a control flow graph. A [control flow graph](https://en.wikipedia.org/wiki/Control_flow_graph) (CFG) in computer science is a representation, using graph notation, of all paths that might be traversed through a program during its execution. Each statement in the program is a node in the flow graph; if statemment x can be followed by statement y, there is an egde from x to y.
+To perform analysis on a program, it is often useful to make a control flow graph. A [control flow graph](https://en.wikipedia.org/wiki/Control_flow_graph) (CFG) in computer science is a representation, using graph notation, of all paths that might be traversed through a program during its execution. Each statement in the program is a node in the flow graph; if statemment x can be followed by statement y, there is an edge from x to y.
 
 Following is the flow graph for a simple loop.
 

docs/design/mkldnn/inplace/inplace.md

@@ -94,4 +94,4 @@ replace this original name in all of next op instances.
 
 \* oneDNN gelu kernel is able to perform in-place execution, but currently gelu op does not support in-place execution.
 
-\*\* sum kernel is using oneDNN sum primitive that does not provide in-place exection, so in-place computation is done faked through external buffer. So it was not added into oneDNN inplace pass.
+\*\* sum kernel is using oneDNN sum primitive that does not provide in-place execution, so in-place computation is done faked through external buffer. So it was not added into oneDNN inplace pass.

docs/design/phi/kernel_migrate_cn.md

@@ -159,7 +159,7 @@ void LogSoftmaxKernel(const Context& dev_ctx,
 | `auto* ptr = out->mutbale_data()` | `auto* ptr = out->data()` |
 | `out->mutbale_data(dims, place)` | `out->Resize(dims); dev_ctx.template Alloc(out)` |
 | `out->mutbale_data(place, dtype)` | `dev_ctx.Alloc(out, dtype)` |
-| `platform::erros::XXX` | `phi::erros::XXX` |
+| `platform::errors::XXX` | `phi::errors::XXX` |
 | `platform::float16/bfloat16/complex64/complex128` | `dtype::float16/bfloat16/complex64/complex128` |
 | `framework::Eigen***` | `Eigen***` |
 | `platform::XXXPlace` | `phi::XXXPlace` |

docs/design/phi/kernel_migrate_en.md

@@ -159,7 +159,7 @@ Secondly, it is necessary to replace some of the types or functions that were on
 | `auto* ptr = out->mutbale_data()` | `auto* ptr = out->data()` |
 | `out->mutbale_data(dims, place)` | `out->Resize(dims); dev_ctx.template Alloc(out)` |
 | `out->mutbale_data(place, dtype)` | `dev_ctx.Alloc(out, dtype)` |
-| `platform::erros::XXX` | `phi::erros::XXX` |
+| `platform::errors::XXX` | `phi::errors::XXX` |
 | `platform::float16/bfloat16/complex64/complex128` | `dtype::float16/bfloat16/complex64/complex128` |
 | `framework::Eigen***` | `Eigen***` |
 | `platform::XXXPlace` | `phi::XXXPlace` |

docs/dev_guides/custom_device_docs/custom_device_example_en.md

@@ -10,7 +10,7 @@ In this section we will walk through the steps required to extend a fake hardwar
 
 **InitPlugin**
 
-As a custom runtime entry function, InitPlugin is required to be implemented by the plug-in. The parameter in InitPlugin should also be checked, device information should be filled in, and the runtime API should be registered. In the initialization, PaddlePaddle loads the plug-in and invokes InitPlugin to initialize it, and register runtime (The whole process can be done automatically by the framework, only if the dynamic-link library is in site-packages/paddle-plugins/ or the designated directory of the enviornment variable of CUSTOM_DEVICE_ROOT).
+As a custom runtime entry function, InitPlugin is required to be implemented by the plug-in. The parameter in InitPlugin should also be checked, device information should be filled in, and the runtime API should be registered. In the initialization, PaddlePaddle loads the plug-in and invokes InitPlugin to initialize it, and register runtime (The whole process can be done automatically by the framework, only if the dynamic-link library is in site-packages/paddle-plugins/ or the designated directory of the environment variable of CUSTOM_DEVICE_ROOT).
 
 Example:
 

docs/guides/06_distributed_training/model_parallel_cn.rst

@@ -29,7 +29,7 @@
 
 对于 Embedding 操作，可以将其理解为一种查找表操作。即，将输入看做索引，将 Embedding 参数看做查找表，根据该索引查表得到相应的输出，如下图（a）所示。当采用模型并行时，Embedding 的参数被均匀切分到多个卡上。假设 Embedding 参数的维度为 N*D，并采用 K 张卡执行模型并行，那么模型并行模式下每张卡上的 Embedding 参数的维度为 N//K*D。当参数的维度 N 不能被卡数 K 整除时，最后一张卡的参数维度值为(N//K+N%K)*D。以下图（b）为例，Embedding 参数的维度为 8*D，采用 2 张卡执行模型并行，那么每张卡上 Embedding 参数的维度为 4*D。
 
-为了便于说明，以下我们均假设 Embedding 的参数维度值 D 可以被模型并行的卡数 D 整除。此时，每张卡上 Embedding 参数的索引值为[0, N/K)，逻辑索引值为[k*N/K, (k+1)*N/K)，其中 k 表示卡序号，0<=k<K。对于输入索引 I，如果该索引在该卡表示的逻辑索引范围内，则返回该索引所表示的表项（索引值为 I-k*N/K；否则，返回值为全 0 的虚拟表项。随后，通过 AllReduce 操作获取所有输出表项的和，即对应该 Embeding 操作的输出；整个查表过程如下图（b）所示。
+为了便于说明，以下我们均假设 Embedding 的参数维度值 D 可以被模型并行的卡数 D 整除。此时，每张卡上 Embedding 参数的索引值为[0, N/K)，逻辑索引值为[k*N/K, (k+1)*N/K)，其中 k 表示卡序号，0<=k<K。对于输入索引 I，如果该索引在该卡表示的逻辑索引范围内，则返回该索引所表示的表项（索引值为 I-k*N/K；否则，返回值为全 0 的虚拟表项。随后，通过 AllReduce 操作获取所有输出表项的和，即对应该 Embedding 操作的输出；整个查表过程如下图（b）所示。
 
 .. image:: ./images/parallel_embedding.png
   :width: 600