diff --git a/docs/guides/model_convert/convert_from_pytorch/tools/.gitignore b/docs/guides/model_convert/convert_from_pytorch/tools/.gitignore
index 4204846c164..cb85c8cbd72 100644
--- a/docs/guides/model_convert/convert_from_pytorch/tools/.gitignore
+++ b/docs/guides/model_convert/convert_from_pytorch/tools/.gitignore
@@ -26,3 +26,4 @@ unique_warnings.txt
 link_warnings.txt
 url_warnings.txt
 api_difference_error.txt
+diff_doc_warnings.txt
diff --git a/docs/guides/model_convert/convert_from_pytorch/tools/generate_pytorch_api_mapping.py b/docs/guides/model_convert/convert_from_pytorch/tools/generate_pytorch_api_mapping.py
index c4af7f44da1..8d0a2088ba2 100644
--- a/docs/guides/model_convert/convert_from_pytorch/tools/generate_pytorch_api_mapping.py
+++ b/docs/guides/model_convert/convert_from_pytorch/tools/generate_pytorch_api_mapping.py
@@ -167,20 +167,47 @@ def generate_category2_table(
     rows = []  # 存储表格行数据的列表
     used_apis = set()  # 用于记录已处理的API，避免重复
 
+    invok_diff_matchers = {
+        "ChangeAPIMatcher",
+        "NumelMatcher",
+        "Is_InferenceMatcher",
+    }
+
+    special_matchers = {
+        "TensorFunc2PaddleFunc",
+        "Func2Attribute",
+        "Attribute2Func",
+    }
+
     # 处理api_mapping中Matcher为"UnchangeMatcher"且不在no_need_convert_list中的API
     for src_api, mapping_info in api_mapping_data.items():
         if src_api in whitelist_skip or src_api in no_need_convert_list:
             continue
         matcher = mapping_info.get("Matcher", "")
+        valid = False
         # ChangeAPIMatcher、TensorFunc2PaddleFunc、Func2Attribute、Attribute2Func类别
-        if matcher in [
-            "ChangeAPIMatcher",
-            "TensorFunc2PaddleFunc",
-            "Func2Attribute",
-            "Attribute2Func",
-            "NumelMatcher",
-            "Is_InferenceMatcher",
-        ]:
+        if matcher in special_matchers:
+            has_unsupport_args = "unsupport_args" in mapping_info
+            has_kwargs_change = "kwargs_change" in mapping_info
+            has_paddle_default_kwargs = "paddle_default_kwargs" in mapping_info
+            if has_unsupport_args:
+                print(
+                    f"[torch_more_args] {src_api} -> {mapping_info.get('paddle_api', 'N/A')}"
+                )
+                continue
+            elif has_kwargs_change:
+                print(
+                    f"[args_name_diff] {src_api} -> {mapping_info.get('paddle_api', 'N/A')}"
+                )
+                continue
+            elif has_paddle_default_kwargs:
+                print(
+                    f"[paddle_more_args_or_default_diff] {src_api} -> {mapping_info.get('paddle_api', 'N/A')}"
+                )
+                continue
+            valid = True
+
+        if matcher in invok_diff_matchers or valid:
             # 在docs_mapping中查找当前src_api对应的信息
             docs_mapping_info = docs_mapping.get(src_api, {})
             src_url = docs_mapping_info.get("src_api_url")
diff --git a/docs/guides/model_convert/convert_from_pytorch/tools/validate_api_difference.py b/docs/guides/model_convert/convert_from_pytorch/tools/validate_api_difference_format.py
similarity index 100%
rename from docs/guides/model_convert/convert_from_pytorch/tools/validate_api_difference.py
rename to docs/guides/model_convert/convert_from_pytorch/tools/validate_api_difference_format.py
diff --git a/docs/guides/model_convert/convert_from_pytorch/tools/validate_pytorch_api_mapping.py b/docs/guides/model_convert/convert_from_pytorch/tools/validate_pytorch_api_mapping.py
index 0066d839abe..551ff78582e 100644
--- a/docs/guides/model_convert/convert_from_pytorch/tools/validate_pytorch_api_mapping.py
+++ b/docs/guides/model_convert/convert_from_pytorch/tools/validate_pytorch_api_mapping.py
@@ -15,6 +15,22 @@
 # 默认文件路径
 DEFAULT_FILE_PATH = "/workspace/paddleDocs/docs/guides/model_convert/convert_from_pytorch/pytorch_api_mapping_cn.md"
 
+# 类别映射关系
+CATEGORY_MAP = {
+    "invok_diff_only": "仅 API 调用方式不一致",
+    "args_name_diff": "仅参数名不一致",
+    "paddle_more_args": "paddle 参数更多",
+    "args_default_value_diff": "参数默认值不一致",
+    "torch_more_args": "torch 参数更多",
+    "input_args_usage_diff": "输入参数用法不一致",
+    "input_args_type_diff": "输入参数类型不一致",
+    "output_args_type_diff": "返回参数类型不一致",
+    "composite_implement": "组合替代实现",
+}
+
+# 反向映射（中文到英文）
+REVERSE_CATEGORY_MAP = {v: k for k, v in CATEGORY_MAP.items()}
+
 USER_AGENT = ""
 
 # 重试策略配置
@@ -64,7 +80,7 @@ def parse_toc(lines):
                 continue  # 跳过分隔行
 
             if found_header and line.startswith("|"):
-                # 解析数据行
+                # 解析数据行 - 现在表格有5列，但我们只关心前2列（序号和类别名称）
                 columns = [col.strip() for col in line.split("|")[1:-1]]
                 if len(columns) >= 2:
                     toc.append((columns[0], columns[1]))  # (序号, 类别名称)
@@ -79,7 +95,7 @@ def parse_toc(lines):
 
 def parse_categories(lines):
     """
-    解析类别部分，提取所有类别及其表格数据
+    解析类别部分，提取所有类别及其表格数据（适应五列表格结构）
     """
     categories = []
     current_category = None
@@ -111,8 +127,13 @@ def parse_categories(lines):
             in_table = False
             continue
 
-        # 检测表格开始
-        if line.startswith("|") and "Pytorch" in line and "Paddle" in line:
+        # 检测表格开始（五列表格）
+        if (
+            line.startswith("|")
+            and "Pytorch" in line
+            and "Paddle" in line
+            and "映射分类" in line
+        ):
             in_table = True
             continue
 
@@ -122,15 +143,16 @@ def parse_categories(lines):
             if re.match(r"^\|?[-:\s|]+\|?$", line):
                 continue
 
-            # 解析表格行
+            # 解析五列表格行
             columns = [col.strip() for col in line.split("|")[1:-1]]
-            if len(columns) >= 4:
+            if len(columns) >= 5:  # 现在有5列
                 current_table.append(
                     {
                         "index": columns[0],
                         "pytorch": columns[1],
                         "paddle": columns[2],
-                        "note": columns[3] if len(columns) > 3 else "",
+                        "mapping_category": columns[3],  # 新增的映射分类列
+                        "note": columns[4] if len(columns) > 4 else "",
                     }
                 )
             continue
@@ -159,6 +181,26 @@ def extract_links(text):
     return re.findall(r"\[([^\]]+)\]\(([^)]+)\)", text)
 
 
+def extract_torch_api_name(pytorch_column):
+    """
+    从Pytorch列提取Torch API名称（链接文本），并将转义的下划线恢复为普通下划线
+    """
+    links = extract_links(pytorch_column)
+    if links:
+        api_name = links[0][0]  # 返回第一个链接的文本
+        # 将转义的下划线 "\_" 替换为普通下划线 "_"
+        api_name = api_name.replace(r"\_", "_")
+        return api_name
+
+    # 如果没有链接，尝试直接提取文本内容
+    clean_text = re.sub(r"[\[\]\(\)]", "", pytorch_column).strip()
+    if clean_text:
+        # 同样处理转义的下划线
+        clean_text = clean_text.replace(r"\_", "_")
+        return clean_text
+    return None
+
+
 def check_toc_consistency(toc, categories):
     """
     检查目录与类别标题的一致性
@@ -198,9 +240,8 @@ def check_unique_torch_apis(categories):
 
     for category in categories:
         for row in category["table"]:
-            links = extract_links(row["pytorch"])
-            if links:
-                api_name = links[0][0]  # 取第一个链接的文本作为 API 名称
+            api_name = extract_torch_api_name(row["pytorch"])
+            if api_name:
                 torch_apis[api_name].append(category["id"])
 
     # 检查重复的 API
@@ -215,13 +256,14 @@ def check_unique_torch_apis(categories):
 
 def check_links_exist(categories):
     """
-    检查必要的超链接是否存在（根据新规则）
+    检查必要的超链接是否存在（适应五列表格结构）
     规则：
     1. 第二列(Pytorch)必须有超链接
     2. 第三列(Paddle):
        - 对于"组合替代实现"、"可删除"、"功能缺失"类别，如果内容为空或"-"则不检查
        - 否则必须有超链接
-    3. 第四列(备注): 除了"API完全一致类别"（类别1）外，都需要有超链接
+    3. 第五列(备注): 除了"API完全一致类别"（类别1）外，都需要有超链接
+    4. 第四列(映射分类)不检查超链接，但需要检查内容一致性（在另一个函数中处理）
     """
     warnings = []
 
@@ -263,7 +305,7 @@ def check_links_exist(categories):
                     warning_msg = f"类别 {category_id}({category_name}) 第 {row_num} 行第三列缺少超链接: {row['paddle']}"
                     warnings.append(warning_msg)
 
-            # 3. 检查第四列 (备注)
+            # 3. 检查第五列 (备注)
             note_links = extract_links(row["note"])
             note_content = row["note"].strip()
 
@@ -274,8 +316,135 @@ def check_links_exist(categories):
                 and note_content != "-"
                 and not note_links
             ):
-                warning_msg = f"类别 {category_id}({category_name}) 第 {row_num} 行第四列缺少超链接: {row['note']}"
+                warning_msg = f"类别 {category_id}({category_name}) 第 {row_num} 行第五列缺少超链接: {row['note']}"
+                warnings.append(warning_msg)
+
+    return warnings
+
+
+def check_mapping_category_consistency(categories):
+    """
+    检查映射分类列内容与类别标题的一致性
+    """
+    warnings = []
+
+    for category in categories:
+        category_id = category["id"]
+        category_name = category["name"]
+
+        for i, row in enumerate(category["table"]):
+            row_num = i + 1
+            mapping_category = row.get("mapping_category", "").strip()
+
+            # 检查映射分类列是否与类别标题一致
+            if mapping_category != category_name:
+                warning_msg = (
+                    f"类别 {category_id}({category_name}) 第 {row_num} 行映射分类不一致:\n"
+                    f"  表格中的映射分类: '{mapping_category}'\n"
+                    f"  类别标题: '{category_name}'"
+                )
+                warnings.append(warning_msg)
+
+    return warnings
+
+
+def check_diff_doc_consistency(categories, base_dir):
+    """
+    检查映射文档和差异文档的一致性
+    """
+    warnings = []
+    diff_doc_base = os.path.join(base_dir, "api_difference")
+
+    # 检查差异文档目录是否存在
+    if not os.path.exists(diff_doc_base):
+        warnings.append(f"差异文档根目录不存在: {diff_doc_base}")
+        return warnings
+
+    # 构建API到类别的映射，用于反向检查
+    expected_apis = defaultdict(set)  # category_name -> set of torch_apis
+    found_apis = defaultdict(set)  # category_name -> set of torch_apis
+
+    # 第一步：检查表格中的每个API是否有对应的差异文档
+    for category in categories:
+        category_name = category["name"]
+
+        # 检查这个类别是否需要差异文档
+        if category_name not in REVERSE_CATEGORY_MAP:
+            continue  # 跳过不需要差异文档的类别
+
+        category_en = REVERSE_CATEGORY_MAP[category_name]
+        diff_category_dir = os.path.join(diff_doc_base, category_en)
+
+        # 检查类别目录是否存在
+        if not os.path.exists(diff_category_dir):
+            warning_msg = (
+                f"差异文档目录不存在: {diff_category_dir}\n"
+                f"对应类别: {category_name}"
+            )
+            warnings.append(warning_msg)
+            continue
+
+        for row in category["table"]:
+            torch_api = extract_torch_api_name(row["pytorch"])
+            torch_api = torch_api.replace(r"\_", "_")
+            if not torch_api:
+                continue
+
+            expected_apis[category_name].add(torch_api)
+
+            # 构建预期的MD文件名
+            expected_md_file = f"{torch_api}.md"
+            expected_md_path = os.path.join(diff_category_dir, expected_md_file)
+
+            if not os.path.exists(expected_md_path):
+                warning_msg = (
+                    f"差异文档缺失: {expected_md_file}\n"
+                    f"对应Torch API: {torch_api}\n"
+                    f"类别: {category_name} ({category_en})\n\n"
+                )
                 warnings.append(warning_msg)
+            else:
+                found_apis[category_name].add(torch_api)
+
+    # 第二步：反向检查差异文档目录中的文件是否在表格中有对应
+    for category_en, category_cn in CATEGORY_MAP.items():
+        diff_category_dir = os.path.join(diff_doc_base, category_en)
+
+        if not os.path.exists(diff_category_dir):
+            continue
+
+        # 遍历差异文档目录中的所有.md文件
+        try:
+            for filename in os.listdir(diff_category_dir):
+                if filename.endswith(".md"):
+                    torch_api = filename[:-3]  # 去掉.md后缀
+
+                    # 检查这个API是否在对应类别的表格中
+                    api_found = False
+                    for category in categories:
+                        if category["name"] == category_cn:
+                            for row in category["table"]:
+                                if (
+                                    extract_torch_api_name(row["pytorch"])
+                                    == torch_api
+                                ):
+                                    api_found = True
+                                    break
+                            if api_found:
+                                break
+
+                    if not api_found:
+                        warning_msg = (
+                            f"多余的差异文档: {filename}\n"
+                            f"对应Torch API: {torch_api}\n"
+                            f"类别: {category_cn} ({category_en})\n"
+                            f"该API在映射表格中不存在\n\n"
+                        )
+                        warnings.append(warning_msg)
+        except FileNotFoundError:
+            warnings.append(f"差异文档目录不存在: {diff_category_dir}")
+        except PermissionError:
+            warnings.append(f"无权限访问差异文档目录: {diff_category_dir}")
 
     return warnings
 
@@ -430,6 +599,7 @@ def check_urls_exist(urls_with_context, max_workers=10):
     """
     warnings = []
 
+    # 限制检查的URL数量（避免过多网络请求）
     urls_with_context = urls_with_context[:600]
 
     total_urls = len(urls_with_context)
@@ -536,32 +706,33 @@ def main():
     print(f"找到 {len(toc)} 个目录条目")
     print(f"找到 {len(categories)} 个类别")
 
-    # 执行三个基本校验
+    # 执行基本校验
     toc_warnings = check_toc_consistency(toc, categories)
     unique_warnings = check_unique_torch_apis(categories)
     link_warnings = check_links_exist(categories)
+    mapping_category_warnings = check_mapping_category_consistency(categories)
+    diff_doc_warnings = check_diff_doc_consistency(categories, base_dir)
 
     # 输出警告到文件
-    if toc_warnings:
-        output_path = os.path.join(tools_dir, "toc_warnings.txt")
-        with open(output_path, "w", encoding="utf-8") as f:
-            f.write("目录一致性校验警告:\n")
-            f.writelines(warning + "\n" for warning in toc_warnings)
-        print(f"生成 {output_path}，包含 {len(toc_warnings)} 个警告")
-
-    if unique_warnings:
-        output_path = os.path.join(tools_dir, "unique_warnings.txt")
-        with open(output_path, "w", encoding="utf-8") as f:
-            f.write("Torch API 唯一性校验警告:\n")
-            f.writelines(warning + "\n" for warning in unique_warnings)
-        print(f"生成 {output_path}，包含 {len(unique_warnings)} 个警告")
-
-    if link_warnings:
-        output_path = os.path.join(tools_dir, "link_warnings.txt")
-        with open(output_path, "w", encoding="utf-8") as f:
-            f.write("超链接存在性校验警告:\n")
-            f.writelines(warning + "\n" for warning in link_warnings)
-        print(f"生成 {output_path}，包含 {len(link_warnings)} 个警告")
+    warning_files = [
+        ("toc_warnings.txt", "目录一致性校验警告:", toc_warnings),
+        ("unique_warnings.txt", "Torch API 唯一性校验警告:", unique_warnings),
+        ("link_warnings.txt", "超链接存在性校验警告:", link_warnings),
+        (
+            "mapping_category_warnings.txt",
+            "映射分类一致性校验警告:",
+            mapping_category_warnings,
+        ),
+        ("diff_doc_warnings.txt", "差异文档一致性校验警告:", diff_doc_warnings),
+    ]
+
+    for filename, description, warnings in warning_files:
+        if warnings:
+            output_path = os.path.join(tools_dir, filename)
+            with open(output_path, "w", encoding="utf-8") as f:
+                f.write(f"{description}\n")
+                f.writelines(warning + "\n" for warning in warnings)
+            print(f"生成 {output_path}，包含 {len(warnings)} 个警告")
 
     # 执行URL存在性检查（除非明确跳过）
     url_warnings = []
@@ -584,14 +755,22 @@ def main():
     else:
         print("跳过URL存在性检查")
 
-    # 如果没有警告，输出成功信息
-    if (
-        not toc_warnings
-        and not unique_warnings
-        and not link_warnings
-        and not url_warnings
-    ):
+    # 汇总统计
+    total_warnings = (
+        len(toc_warnings)
+        + len(unique_warnings)
+        + len(link_warnings)
+        + len(mapping_category_warnings)
+        + len(diff_doc_warnings)
+        + len(url_warnings)
+    )
+
+    if total_warnings == 0:
         print("所有校验通过，没有发现警告!")
+    else:
+        print(
+            f"校验完成，共发现 {total_warnings} 个警告，请查看生成的警告文件。"
+        )
 
 
 if __name__ == "__main__":