Merge branch 'develop' into patch-1

Author: Echo-Nie

_typos.toml

@@ -83,16 +83,6 @@ Wether = "Wether"
 accordding = "accordding"
 accoustic = "accoustic"
 accpetance = "accpetance"
-accracy = "accracy"
-acutal = "acutal"
-apporach = "apporach"
-apporaches = "apporaches"
-arguements = "arguements"
-arguemnts = "arguemnts"
-assgin = "assgin"
-assginment = "assginment"
-auxilary = "auxilary"
-avaiable = "avaiable"
 baisc = "baisc"
 basci = "basci"
 beacuse = "beacuse"

ci_scripts/check_api_label_cn.py

@@ -77,7 +77,7 @@ def run_cn_api_label_checking(rootdir, files):
     for file in files:
         if should_test(file) and not check_api_label(rootdir, file):
             logger.error(
-                f"The first line in {rootdir}/{file} is not avaiable, please re-check it!"
+                f"The first line in {rootdir}/{file} is not available, please re-check it!"
             )
             sys.exit(1)
     valid_api_labels = find_all_api_labels_in_dir(rootdir)

ci_scripts/hooks/pre-doc-compile.sh

@@ -5,55 +5,140 @@ SCRIPT_DIR="$( cd "$( dirname "$0" )" && pwd )"
 
 FLUIDDOCDIR=${FLUIDDOCDIR:=/FluidDoc}
 DOCROOT=${FLUIDDOCDIR}/docs
-
-
-## 1. 获取API映射文件
 APIMAPPING_ROOT=${DOCROOT}/guides/model_convert/convert_from_pytorch
 TOOLS_DIR=${APIMAPPING_ROOT}/tools
 
-# 确保tools目录存在
-mkdir -p ${TOOLS_DIR}
+# Create tools directory if not exists
+if [ ! -d "${TOOLS_DIR}" ]; then
+    echo "INFO: Creating tools directory at ${TOOLS_DIR}"
+    mkdir -p "${TOOLS_DIR}"
+    if [ $? -ne 0 ]; then
+        echo "ERROR: Failed to create directory ${TOOLS_DIR}"
+        exit 1
+    fi
+else
+    echo "INFO: Tools directory ${TOOLS_DIR} already exists"
+fi
 
-#下载的文件URL
+# Define API mapping files URLs
 API_ALIAS_MAPPING_URL="https://raw.githubusercontent.com/PaddlePaddle/PaConvert/master/paconvert/api_alias_mapping.json"
 API_MAPPING_URL="https://raw.githubusercontent.com/PaddlePaddle/PaConvert/master/paconvert/api_mapping.json"
 GLOBAL_VAR_URL="https://raw.githubusercontent.com/PaddlePaddle/PaConvert/master/paconvert/global_var.py"
 ATTRIBUTE_MAPPING_URL="https://raw.githubusercontent.com/PaddlePaddle/PaConvert/master/paconvert/attribute_mapping.json"
 
-# 下载文件
+# Define backup URLs
+BACKUP_API_ALIAS_MAPPING_URL="https://paddle-paconvert.bj.bcebos.com/api_alias_mapping.json"
+BACKUP_API_MAPPING_URL="https://paddle-paconvert.bj.bcebos.com/api_mapping.json"
+BACKUP_GLOBAL_VAR_URL="https://paddle-paconvert.bj.bcebos.com/global_var.py"
+BACKUP_ATTRIBUTE_MAPPING_URL="https://paddle-paconvert.bj.bcebos.com/attribute_mapping.json"
+
+# Check for proxy settings
 PROXY=""
 if [ -n "$https_proxy" ]; then
     PROXY="$https_proxy"
+    echo "INFO: find proxy"
 elif [ -n "$http_proxy" ]; then
     PROXY="$http_proxy"
+    echo "INFO: find proxy"
+else
+    echo "INFO: No proxy detected, downloading directly."
 fi
 
-# 构建 curl 代理参数
-CURL_PROXY_ARGS=""
+# Build curl proxy arguments
 if [ -n "$PROXY" ]; then
-    CURL_PROXY_ARGS="--proxy $PROXY"
+    CURL_PROXY_ARGS="--proxy ${PROXY}"
 else
-    echo "No proxy detected, downloading directly."
+    CURL_PROXY_ARGS=""
+fi
+
+# Download API mapping files with retry
+download_file() {
+    local url=$1
+    local dest=$2
+    local filename=$(basename "$dest")
+    local max_retries=5
+    local retry_count=0
+
+    echo "INFO: Starting download of ${filename} from ${url}"
+
+    while [ $retry_count -lt $max_retries ]; do
+        retry_count=$((retry_count + 1))
+        echo "INFO: Attempt $retry_count of $max_retries to download ${filename}"
+
+        if curl $CURL_PROXY_ARGS -o "${dest}" -s "${url}" > /dev/null 2>&1; then
+            echo "SUCCESS: Successfully downloaded ${filename} to ${dest}"
+            return 0
+        else
+            echo "WARNING: Failed to download ${filename} from ${url} (attempt $retry_count)"
+            sleep 2  # Wait for 2 seconds before next retry
+        fi
+    done
+
+    echo "ERROR: Failed to download ${filename} after $max_retries attempts"
+    return 1
+}
+
+# Download each file with detailed logging
+echo "INFO: Downloading API alias mapping file"
+if ! download_file "${API_ALIAS_MAPPING_URL}" "${TOOLS_DIR}/api_alias_mapping.json"; then
+    echo "INFO: Trying backup URL for API alias mapping file"
+    if ! download_file "${BACKUP_API_ALIAS_MAPPING_URL}" "${TOOLS_DIR}/api_alias_mapping.json"; then
+        echo "ERROR: API alias mapping download failed (both main and backup URLs). Exiting."
+        exit 1
+    fi
+fi
+
+echo "INFO: Downloading API mapping file"
+if ! download_file "${API_MAPPING_URL}" "${TOOLS_DIR}/api_mapping.json"; then
+    echo "INFO: Trying backup URL for API mapping file"
+    if ! download_file "${BACKUP_API_MAPPING_URL}" "${TOOLS_DIR}/api_mapping.json"; then
+        echo "ERROR: API mapping download failed (both main and backup URLs). Exiting."
+        exit 1
+    fi
 fi
 
-# 执行下载
-curl $CURL_PROXY_ARGS -o "${TOOLS_DIR}/api_alias_mapping.json" -s "${API_ALIAS_MAPPING_URL}"
-curl $CURL_PROXY_ARGS -o "${TOOLS_DIR}/api_mapping.json" -s "${API_MAPPING_URL}"
-curl $CURL_PROXY_ARGS -o "${TOOLS_DIR}/global_var.py" -s "${GLOBAL_VAR_URL}"
-curl $CURL_PROXY_ARGS -o "${TOOLS_DIR}/attribute_mapping.json" -s "${ATTRIBUTE_MAPPING_URL}"
+echo "INFO: Downloading global variable file"
+if ! download_file "${GLOBAL_VAR_URL}" "${TOOLS_DIR}/global_var.py"; then
+    echo "INFO: Trying backup URL for global variable file"
+    if ! download_file "${BACKUP_GLOBAL_VAR_URL}" "${TOOLS_DIR}/global_var.py"; then
+        echo "ERROR: Global variable download failed (both main and backup URLs). Exiting."
+        exit 1
+    fi
+fi
+
+echo "INFO: Downloading attribute mapping file"
+if ! download_file "${ATTRIBUTE_MAPPING_URL}" "${TOOLS_DIR}/attribute_mapping.json"; then
+    echo "INFO: Trying backup URL for attribute mapping file"
+    if ! download_file "${BACKUP_ATTRIBUTE_MAPPING_URL}" "${TOOLS_DIR}/attribute_mapping.json"; then
+        echo "ERROR: Attribute mapping download failed (both main and backup URLs). Exiting."
+        exit 1
+    fi
+fi
 
-# 检查下载是否成功
-if [ $? -ne 0 ]; then
-    echo "Error: Failed to download API mapping files"
+# Check if all files exist before proceeding
+if [ ! -f "${TOOLS_DIR}/api_alias_mapping.json" ] || \
+   [ ! -f "${TOOLS_DIR}/api_mapping.json" ] || \
+   [ ! -f "${TOOLS_DIR}/global_var.py" ] || \
+   [ ! -f "${TOOLS_DIR}/attribute_mapping.json" ]; then
+    echo "ERROR: One or more API mapping files are missing after download"
+    echo "Missing files:"
+    if [ ! -f "${TOOLS_DIR}/api_alias_mapping.json" ]; then echo "  - api_alias_mapping.json"; fi
+    if [ ! -f "${TOOLS_DIR}/api_mapping.json" ]; then echo "  - api_mapping.json"; fi
+    if [ ! -f "${TOOLS_DIR}/global_var.py" ]; then echo "  - global_var.py"; fi
+    if [ ! -f "${TOOLS_DIR}/attribute_mapping.json" ]; then echo "  - attribute_mapping.json"; fi
     exit 1
 fi
 
-## 3. Apply PyTorch-PaddlePaddle mapping using the new API mapping files
-python ${APIMAPPING_ROOT}/tools/get_api_difference_info.py
-python ${APIMAPPING_ROOT}/tools/generate_pytorch_api_mapping.py
+echo "INFO: All API mapping files successfully downloaded"
 
+echo "INFO: Running get_api_difference_info.py"
+if ! python "${APIMAPPING_ROOT}/tools/get_api_difference_info.py"; then
+    echo "ERROR: get_api_difference_info.py failed. Please check the script."
+    exit 1
+fi
 
-if [ $? -ne 0 ]; then
-    echo "Error: API mapping generate script failed, please check changes in ${APIMAPPING_ROOT}"
+echo "INFO: Running generate_pytorch_api_mapping.py"
+if ! python "${APIMAPPING_ROOT}/tools/generate_pytorch_api_mapping.py"; then
+    echo "ERROR: generate_pytorch_api_mapping.py failed. Please check the script."
     exit 1
 fi

docs/api/paddle/logical_xor_cn.rst

@@ -30,7 +30,7 @@ logical_xor
 
 返回
 ::::::::::::
- ``Tensor``，维度``x`` 维度相同，存储运算后的结果。
+``Tensor``，维度 ``x`` 维度相同，存储运算后的结果。
 
 代码示例
 ::::::::::::

docs/api/paddle/nn/Conv1DTranspose_cn.rst

@@ -8,7 +8,7 @@ Conv1DTranspose
 
 一维转置卷积层（Convlution1d transpose layer）
 
-该层根据输入（input）、卷积核（kernel）和空洞大小（dilations）、步长（stride）、填充（padding）来计算输出特征大小或者通过 output_size 指定输出特征层大小。输入(Input)和输出(Output)为 NCL 或 NLC 格式，其中 N 为批尺寸，C 为通道数（channel），L 为特征长度。卷积核是 MCL 格式，M 是输出图像通道数，C 是输入图像通道数，L 是卷积核长度。如果组数大于 1，C 等于输入图像通道数除以组数的结果。转置卷积的计算过程相当于卷积的反向计算。转置卷积又被称为反卷积（但其实并不是真正的反卷积）。欲了解转置卷积层细节，请参考下面的说明和 `参考文献 <https://arxiv.org/pdf/1603.07285.pdf/>`_。如果参数 bias_attr 不为 False，转置卷积计算会添加偏置项。
+该层根据输入（input）、卷积核（kernel）和空洞大小（dilations）、步长（stride）、填充（padding）来计算输出特征大小或者通过 output_size 指定输出特征层大小。输入(Input)和输出(Output)为 NCL 或 NLC 格式，其中 N 为批尺寸，C 为通道数（channel），L 为特征长度。卷积核是 MCL 格式，M 是输出图像通道数，C 是输入图像通道数，L 是卷积核长度。如果组数大于 1，C 等于输入图像通道数除以组数的结果。转置卷积的计算过程相当于卷积的反向计算。转置卷积又被称为反卷积（但其实并不是真正的反卷积）。欲了解转置卷积层细节，请参考下面的说明和 `参考文献 <https://arxiv.org/pdf/1603.07285>`_。如果参数 bias_attr 不为 False，转置卷积计算会添加偏置项。
 
 
 输入 :math:`X` 和输出 :math:`Out` 函数关系如下：

docs/api/paddle/put_along_axis_cn.rst

@@ -14,7 +14,7 @@ put_along_axis
     - **indices**  (Tensor) - 索引矩阵，包含沿轴提取 1d 切片的下标，必须和 arr 矩阵有相同的维度。当 ``broadcast`` 为 ``True`` 时，需要能够 broadcast 与 arr 矩阵对齐，否则除 ``axis`` 维度，其他维度都需要小于等于 ``arr`` 与 ``values`` 的对应维度。数据类型为：int32、int64。
     - **values** (float) - 需要插入的值，当 ``broadcast`` 为 ``True`` 时，形状和维度需要能够被 broadcast 与 indices 矩阵匹配，否则各维度需大于等于 ``indices`` 的各维度。数据类型为：bfloat16、float16、float32、float64、int32、int64、uint8、int16。
     - **axis**  (int) - 指定沿着哪个维度获取对应的值，数据类型为：int。
-    - **reduce** (str，可选) - 归约操作类型，默认为 ``assign``，可选为 ``add``、 ``multiple``、 ``mean``、 ``amin``、 ``amax``。不同的规约操作插入值 value 对于输入矩阵 arr 会有不同的行为，如为 ``assgin`` 则覆盖输入矩阵， ``add`` 则累加至输入矩阵， ``mean`` 则计算累计平均值至输入矩阵， ``multiple`` 则累乘至输入矩阵， ``amin`` 则计算累计最小值至输入矩阵， ``amax`` 则计算累计最大值至输入矩阵。
+    - **reduce** (str，可选) - 归约操作类型，默认为 ``assign``，可选为 ``add``、 ``multiple``、 ``mean``、 ``amin``、 ``amax``。不同的规约操作插入值 value 对于输入矩阵 arr 会有不同的行为，如为 ``assign`` 则覆盖输入矩阵， ``add`` 则累加至输入矩阵， ``mean`` 则计算累计平均值至输入矩阵， ``multiple`` 则累乘至输入矩阵， ``amin`` 则计算累计最小值至输入矩阵， ``amax`` 则计算累计最大值至输入矩阵。
     - **include_self** (bool，可选) - 规约时是否包含 arr 的元素，默认为 ``True``。
     - **broadcast** (bool，可选) - 是否广播 ``index`` 矩阵，默认为 ``True``。
 

docs/api/paddle/scatter_cn.rst

@@ -59,7 +59,7 @@ PyTorch 兼容的 scatter 函数。基于 :ref:`cn_api_paddle_put_along_axis` 
     - **dim** （int） - 进行 scatter 操作的维度，范围为 ``[-input.ndim, input.ndim)``。
     - **index** （Tensor）- 索引矩阵，包含沿轴提取 1d 切片的下标，必须和 arr 矩阵有相同的维度。注意，除了 ``dim`` 维度外， ``index`` 张量的各维度大小应该小于等于 ``input`` 以及 ``src`` 张量。内部的值应该在 ``input.shape[dim]`` 范围内。数据类型可以是 int32，int64。
     - **src** （Tensor）- 需要插入的值。``src`` 是张量时，各维度大小需要至少大于等于 ``index`` 各维度。不受到 ``input`` 的各维度约束。当为标量值时，会自动广播大小到 ``index``。数据类型为：bfloat16、float16、float32、float64、int32、int64、uint8、int16。本参数有一个互斥的别名 ``value``。
-    - **reduce** （str，可选）- 指定 scatter 的归约方式。默认值为 None，等效为 ``assign``。可选为 ``add``、 ``multiple``、 ``mean``、 ``amin``、 ``amax``。不同的规约操作插入值 src 对于输入矩阵 arr 会有不同的行为，如为 ``assgin`` 则覆盖输入矩阵， ``add`` 则累加至输入矩阵， ``mean`` 则计算累计平均值至输入矩阵， ``multiple`` 则累乘至输入矩阵， ``amin`` 则计算累计最小值至输入矩阵， ``amax`` 则计算累计最大值至输入矩阵。
+    - **reduce** （str，可选）- 指定 scatter 的归约方式。默认值为 None，等效为 ``assign``。可选为 ``add``、 ``multiple``、 ``mean``、 ``amin``、 ``amax``。不同的规约操作插入值 src 对于输入矩阵 arr 会有不同的行为，如为 ``assign`` 则覆盖输入矩阵， ``add`` 则累加至输入矩阵， ``mean`` 则计算累计平均值至输入矩阵， ``multiple`` 则累乘至输入矩阵， ``amin`` 则计算累计最小值至输入矩阵， ``amax`` 则计算累计最大值至输入矩阵。
     - **out** (Tensor，可选) - 用于引用式传入输出值，注意：动态图下 out 可以是任意 Tensor，默认值为 None。
 
 返回

docs/design/memory/memory_optimization.md

@@ -79,7 +79,7 @@ In former control flow graph, the out-edges of node 5 are 5 --> 6 and 5 --> 2, a
 
 - Uses and Defs
 
-An assignmemt to a variable or temporary defines that variable. An occurence of a variable on the right-hand side of an assginment(or in other expressions) uses the variable. We can define the *def* of a variable as the set of graph nodes that define it; or the *def* of a graph node as the set of variables that it defines; and the similarly for the *use* of a variable or graph node. In former control flow graph, *def(3)* = {c}, *use(3)* = {b, c}.
+An assignmemt to a variable or temporary defines that variable. An occurence of a variable on the right-hand side of an assignment(or in other expressions) uses the variable. We can define the *def* of a variable as the set of graph nodes that define it; or the *def* of a graph node as the set of variables that it defines; and the similarly for the *use* of a variable or graph node. In former control flow graph, *def(3)* = {c}, *use(3)* = {b, c}.
 
 - Liveness
 

docs/design/phi/design_cn.md

@@ -1219,7 +1219,7 @@ REGISTER_OPERATOR(sign, ops::SignOp, ops::SignOpMaker<float>,
  * The infrt declare like:
  *
  * def PDKEL_Reshape_to_CPU : Pat<
- *     (PD_ReshapeOp $x, $shape_tensor， $shape_attr), // OpMaker arguements
+ *     (PD_ReshapeOp $x, $shape_tensor， $shape_attr), // OpMaker arguments
  *     (PDKEL_ReshapeKernelAttr $x, fn($shape_attr)>;  // Kernel arguments
  * def PDKEL_Reshape_to_CPU : Pat<
  *     (PD_ReshapeOp $x, $shape_tensor， $shape_attr),

docs/design/phi/design_en.md

@@ -78,7 +78,7 @@ We hope to be able to achieve the same three-layer arguments of Python API -> Op
   - The initial construction of the PHI operator library paid more attention to Kernel "migration". Due to the consideration of time and labor costs, the original OpKernel logic migration is not forced to be upgraded to "combined" writing for the time being, and the same is true for the forward and backward Kernels
   - The "combined Kernel extension development" capability provided by the PHI operator library initially serves the new operators of subsequent increments, and the existing operators still maintain their original coding implementation, reducing the cost of migration
   - The "new hardware expansion capability" provided by the PHI operator library is initially only provided within the scope of the new hardware itself. For example, the XPU has implemented 50 Kernels, and then it can combine new Kernels based on 50 Kernels, but this is only limited to the XPU Within the scope, its implementation is not common with CPU, CUDA, etc.
-  - The PHI operator library project focuses on the work of "Kernel functionalization & Op normalization", Kernel is changed to functional format, C++ API and Op naming and arguemnts list are gradually normalized to Python API under the premise of ensuring compatibility as much as possible
+  - The PHI operator library project focuses on the work of "Kernel functionalization & Op normalization", Kernel is changed to functional format, C++ API and Op naming and arguments list are gradually normalized to Python API under the premise of ensuring compatibility as much as possible
 
 
 ## 2. Design Overview
@@ -1219,7 +1219,7 @@ At present, the `ArgumentMapping` function mapping is designed. In the `phi/ops/
  * The infrt declare like:
  *
  * def PDKEL_Reshape_to_CPU : Pat<
- *     (PD_ReshapeOp $x, $shape_tensor， $shape_attr), // OpMaker arguements
+ *     (PD_ReshapeOp $x, $shape_tensor， $shape_attr), // OpMaker arguments
  *     (PDKEL_ReshapeKernelAttr $x, fn($shape_attr)>;  // Kernel arguments
  * def PDKEL_Reshape_to_CPU : Pat<
  *     (PD_ReshapeOp $x, $shape_tensor， $shape_attr),