q

Author: supotato6

.github/workflows/deploy_docs.yml

@@ -2,7 +2,7 @@ name: Develop Docs
 on:
   push:
     branches: #设置更新哪个分支会更新站点
-      - release/3.2
+      - release/3.3
 permissions:
   contents: write
 jobs:
@@ -27,5 +27,5 @@ jobs:
       - run: pip install mike mkdocs-material jieba mkdocs-git-revision-date-localized-plugin mkdocs-git-committers-plugin-2 mkdocs-git-authors-plugin mkdocs-static-i18n mkdocs-minify-plugin 
       - run: |
           git fetch origin gh-pages --depth=1
-          mike deploy --push --update-aliases 3.2 latest
+          mike deploy --push --update-aliases 3.3 latest
           mike set-default --push latest

.pre-commit-config.yaml

@@ -55,7 +55,7 @@ repos:
     -   id: isort
         args:
             - --profile=black
-        exclude: ^deploy/ultra-infer/python/ultra_infer/
+        files: ^paddlex/
 
 # check license
 -   repo: local

.precommit/check_imports.py

@@ -24,16 +24,17 @@
 from stdlib_list import stdlib_list
 
 sys.path.append(str(pathlib.Path(__file__).parent.parent))
-from setup import DEP_SPECS, REQUIRED_DEPS
+from setup import REQUIRED_DEPS
 
 # NOTE: We do not use `importlib.metadata.packages_distributions` here because
 # 1. It is supported only in Python 3.10+.
 # 2. It requires the packages to be installed, but we are doing a static check.
 MOD_TO_DEP = {
-    "aistudio_sdk": "aistudio_sdk",
+    "aistudio_sdk": "aistudio-sdk",
     "aiohttp": "aiohttp",
     "baidubce": "bce-python-sdk",
     "bs4": "beautifulsoup4",
+    "docx": "python-docx",
     "chardet": "chardet",
     "chinese_calendar": "chinese-calendar",
     "colorlog": "colorlog",
@@ -43,9 +44,10 @@
     "fastapi": "fastapi",
     "filelock": "filelock",
     "filetype": "filetype",
+    "flash_attn": "flash-attn",
     "ftfy": "ftfy",
     "GPUtil": "GPUtil",
-    "huggingface_hub": "huggingface_hub",
+    "huggingface_hub": "huggingface-hub",
     "imagesize": "imagesize",
     "jinja2": "Jinja2",
     "joblib": "joblib",
@@ -61,6 +63,7 @@
     "cv2": "opencv-contrib-python",
     "openpyxl": "openpyxl",
     "packaging": "packaging",
+    "paddle2onnx": "paddle2onnx",
     "pandas": "pandas",
     "PIL": "pillow",
     "premailer": "premailer",
@@ -74,22 +77,28 @@
     "regex": "regex",
     "requests": "requests",
     "ruamel.yaml": "ruamel.yaml",
+    "safetensors": "safetensors",
     "skimage": "scikit-image",
     "sklearn": "scikit-learn",
+    "sentencepiece": "sentencepiece",
+    "sglang": "sglang",
     "shapely": "shapely",
     "soundfile": "soundfile",
     "starlette": "starlette",
     "tiktoken": "tiktoken",
     "tokenizers": "tokenizers",
+    "torch": "torch",
     "tqdm": "tqdm",
+    "transformers": "transformers",
     "typing_extensions": "typing-extensions",
     "ujson": "ujson",
     "uvicorn": "uvicorn",
+    "uvloop": "uvloop",
+    "vllm": "vllm",
+    "xformers": "xformers",
     "yarl": "yarl",
+    "bidi": "python-bidi",
 }
-assert (
-    set(MOD_TO_DEP.values()) == DEP_SPECS.keys()
-), f"`MOD_TO_DEP` should be updated to match `DEP_SPECS`. Symmetric difference: {set(MOD_TO_DEP.values()) ^ DEP_SPECS.keys()}"
 MOD_PATTERN = re.compile(
     rf"^(?:{'|'.join([re.escape(mod) for mod in MOD_TO_DEP])})(?=\.|$)"
 )
@@ -107,7 +116,11 @@
     "paddle3d",
     "paddlevideo",
 }
-MANUALLY_MANAGED_OPTIONAL_HEAVY_MODS = {"paddle_custom_device", "ultra_infer"}
+MANUALLY_MANAGED_OPTIONAL_HEAVY_MODS = {
+    "paddle_custom_device",
+    "ultra_infer",
+    "fastdeploy",
+}
 
 
 def check(file_path):

README.md

@@ -35,6 +35,10 @@ PaddleX 3.0 是基于飞桨框架构建的低代码开发工具，它集成了
 
 ## 📣 近期更新
 
+🔥🔥 **2025.10.16，发布 PaddleX v3.3.0**，新增能力如下：
+
+- **支持PaddleOCR-VL、PP-OCRv5多语种模型的推理部署能力。**
+
 🔥🔥 **2025.8.20，发布 PaddleX v3.2.0**，新增能力如下：
 
 - **部署能力升级：**

README_en.md

@@ -37,6 +37,10 @@ PaddleX 3.0 is a low-code development tool for AI models built on the PaddlePadd
 
 ## 📣 Recent Updates
 
+🔥🔥 **2025.10.16, PaddleX v3.3.0 Released**
+
+- **Added support for inference and deployment of PaddleOCR-VL and PP-OCRv5 multilingual models.**
+
 🔥🔥 **2025.8.20, PaddleX v3.2.0 Released**
 
 - **Deployment Capability Upgrades:**

api_examples/pipelines/test_pp_ocr_vl.py

@@ -0,0 +1,31 @@
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from paddlex import create_pipeline
+
+pipeline = create_pipeline(pipeline="PaddleOCR-VL")
+
+output = pipeline.predict(
+    "/paddle/project/PaddleX/demo_paper.png",
+    use_doc_orientation_classify=False,
+    use_doc_unwarping=False,
+)
+
+for res in output:
+    res.print()
+    res.save_to_img("./output")
+    res.save_to_json("./output")
+    res.save_to_xlsx("./output")
+    res.save_to_html("./output")
+    res.save_to_markdown("./output", pretty=False)

deploy/genai_vllm_server_docker/Dockerfile

@@ -0,0 +1,24 @@
+FROM python:3.10
+
+RUN apt-get update \
+    && apt-get install -y libgl1 \
+    && rm -rf /var/lib/apt/lists/*
+
+ENV PIP_NO_CACHE_DIR=0
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONDONTWRITEBYTECODE=1
+
+ARG PADDLEX_VERSION=">=3.3.6,<3.4"
+RUN python -m pip install "paddlex${PADDLEX_VERSION}"
+
+ARG BUILD_FOR_SM120=false
+RUN if [ "${BUILD_FOR_SM120}" = 'true' ]; then \
+        python -m pip install torch==2.8.0 https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.4.11/flash_attn-2.8.3%2Bcu128torch2.8-cp310-cp310-linux_x86_64.whl; \
+    else \
+        python -m pip install torch==2.8.0 https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.3.14/flash_attn-2.8.2+cu128torch2.8-cp310-cp310-linux_x86_64.whl; \
+    fi \
+    && paddlex --install genai-vllm-server
+
+EXPOSE 8080
+
+CMD ["paddlex_genai_server", "--model_name", "PaddleOCR-VL-0.9B", "--host", "0.0.0.0", "--port", "8080", "--backend", "vllm"]

deploy/genai_vllm_server_docker/build.sh

@@ -0,0 +1,37 @@
+#!/usr/bin/env bash
+
+paddlex_version='>=3.3.6,<3.4'
+build_for_sm120='false'
+tag_suffix='latest'
+
+while [[ $# -gt 0 ]]; do
+    case $1 in
+        --pdx-version)
+            paddlex_version="==$2"
+            shift
+            shift
+            ;;
+        --sm120)
+            build_for_sm120='true'
+            shift
+            ;;
+        --tag-suffix)
+            tag_suffix="$2"
+            shift
+            shift
+            ;;
+        *)
+            echo "Unknown option: $1" >&2
+            exit 2
+            ;;
+    esac
+done
+
+docker build \
+    -t "ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/paddlex-genai-vllm-server:${tag_suffix}" \
+    --build-arg PADDLEX_VERSION="${paddlex_version}" \
+    --build-arg BUILD_FOR_SM120="${build_for_sm120}" \
+    --build-arg http_proxy="${http_proxy}" \
+    --build-arg https_proxy="${https_proxy}" \
+    --build-arg no_proxy="${no_proxy}" \
+    .

deploy/hps/README.md

@@ -10,9 +10,9 @@ comments: true
 
 **请注意，本项目依赖于如下环境配置：**
 
+- **CPU 架构**：x86-64
 - **操作系统**：Linux
-- **Docker 版本**：`>= 20.10.0`，用于镜像构建和部署
-- **CPU 架构**：x86-64 
+- **Docker Engine 版本**：`>= 20.10.0`，用于镜像构建和部署
 
 本文档主要介绍如何基于本项目提供的脚本完成高稳定性服务化部署环境搭建与物料打包。整体流程分为两个阶段：
 
@@ -48,7 +48,7 @@ comments: true
 为了使构建结果的可重现性更强，本步骤将依赖锁定到精确版本。请切换至 `server_env` 目录执行如下脚本：
 
 ```bash
-./script/freeze_requirements.sh
+./scripts/freeze_requirements.sh
 ```
 
 该脚本调用 `pip-tools compile` 解析依赖源文件，并最终生成一系列 `.txt` 文件（如 `requirements/gpu.txt`、`requirements/cpu.txt` 等），这些文件将为 [1.3 镜像构建](./README.md#13-镜像构建) 提供依赖版本约束。
@@ -85,7 +85,7 @@ comments: true
 对于 Triton Server，项目使用预先编译好的版本，将在构建镜像时自动下载，无需手动下载。以构建 GPU 镜像为例，在 `server_env` 目录下执行以下命令：
 
 ```bash
-./scripts/build_deployment_image.sh -k gpu -t latest-gpu 
+./scripts/build_deployment_image.sh -k gpu -t latest-gpu
 ```
 
 构建镜像的参数配置项包括
@@ -118,10 +118,10 @@ comments: true
 执行成功后，命令行会输出以下提示信息：
 
 ```text
- => => exporting to image                                                         
- => => exporting layers                                                      
- => => writing image  sha256:ba3d0b2b079d63ee0239a99043fec7e25f17bf2a7772ec2fc80503c1582b3459   
- => => naming to ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlex/hps:latest-gpu   
+ => => exporting to image
+ => => exporting layers
+ => => writing image  sha256:ba3d0b2b079d63ee0239a99043fec7e25f17bf2a7772ec2fc80503c1582b3459
+ => => naming to ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlex/hps:latest-gpu
 ```
 
 如需批量构建 GPU 和 CPU 镜像，可以执行以下命令：
@@ -172,7 +172,7 @@ comments: true
 </tbody>
 </table>
 
-调用后存储到当前目录 `/output` 路径下。
+调用后存储到当前目录 `output` 路径下。
 
 ## 3.FAQ
 

deploy/hps/README_en.md

@@ -11,9 +11,9 @@ This project provides a high-stability serving solution, consisting of two main
 
 **Note: This project relies on the following environment configurations:**
 
-- **Operating System**: Linux
-- **Docker Version**: `>= 20.10.0` (Used for image building and deployment)
 - **CPU Architecture**: x86-64
+- **Operating System**: Linux
+- **Docker Engine Version**: `>= 20.10.0` (Used for image building and deployment)
 
 This  document  mainly introduces how to set up a high stability serving environment and package related materials using the scripts provided by this project. The overall process consists of two main stages:
 
@@ -32,13 +32,13 @@ Image Building Steps:
 
 1. Build a requirement collection image. (Optional)
 2. Freeze requirement versions to improve the reproducibility of deployment image building. (Optional)
-3. Build the deployment image based on the frozen requirement information to generate the final deployment image and provide image support for subsequent pipeline execution. 
+3. Build the deployment image based on the frozen requirement information to generate the final deployment image and provide image support for subsequent pipeline execution.
 
 **If you do not need to modify requirement-related information, you can skip to [1.3 Building Image](./README_en.md#13-building-image) to build the deployment image using cached requirement information.**
 
 ## 1.1 Build the Requirement Collection Image (Optional)
 
-Navigate to the `server_env` directory and run follow script for building the requirement collection image in this directory. 
+Navigate to the `server_env` directory and run follow script for building the requirement collection image in this directory.
 
 ```bash
 ./scripts/prepare_rc_image.sh
@@ -121,10 +121,10 @@ If the basic image cannot be pulled, please refer to the solutions in the [FAQ](
 After run successfully, the command line will display the following message:
 
 ```text
- => => exporting to image                                                         
- => => exporting layers                                                      
- => => writing image  sha256:ba3d0b2b079d63ee0239a99043fec7e25f17bf2a7772ec2fc80503c1582b3459   
- => => naming to ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlex/hps:latest-gpu   
+ => => exporting to image
+ => => exporting layers
+ => => writing image  sha256:ba3d0b2b079d63ee0239a99043fec7e25f17bf2a7772ec2fc80503c1582b3459
+ => => naming to ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlex/hps:latest-gpu
 ```
 
 To build both GPU and CPU images  run the following command:
@@ -140,7 +140,7 @@ This stage mainly introduces how to package pipeline materials. This function is
 - `client`: Responsible for invoking the model services.
 - `server`: Deployed using the images built in [1. Image Building](./README_en.md#1-image-building), serving as the runtime environment for model services.
 
-Before packaging the pipeline materials, you need to switch to the `sdk` directory and run the `scripts/assemble.sh` script in this directory for  packaging. For example, to package the general OCR pipeline, run:
+Before packaging the pipeline materials, you need to switch to the `sdk` directory and run the `scripts/assemble.sh` script in this directory for packaging. For example, to package the general OCR pipeline, run:
 
 ```bash
 ./scripts/assemble.sh OCR
@@ -175,7 +175,7 @@ The parameters for the packaging script are described as follows:
 </tbody>
 </table>
 
-After run successfully, the packaged  will be stored in the `/output` directory.
+After run successfully, the packaged  will be stored in the `output` directory.
 
 ## 3. FAQ
 
@@ -191,4 +191,4 @@ When running the image build scripts, you can use the `-p` parameter to specify
 
 ```bash
 ./scripts/prepare_rc_image.sh -p  https://pypi.tuna.tsinghua.edu.cn/simple
-```
+```