Skip to content

Commit beefe5c

Browse files
authored
Fix bugs and update for v3.3 (#4613)
* Fix bugs * Fix all
1 parent a183f81 commit beefe5c

File tree

10 files changed

+225
-35
lines changed

10 files changed

+225
-35
lines changed

.github/workflows/deploy_docs.yml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@ name: Develop Docs
22
on:
33
push:
44
branches: #设置更新哪个分支会更新站点
5-
- release/3.2
5+
- release/3.3
66
permissions:
77
contents: write
88
jobs:
@@ -27,5 +27,5 @@ jobs:
2727
- 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
2828
- run: |
2929
git fetch origin gh-pages --depth=1
30-
mike deploy --push --update-aliases 3.2 latest
30+
mike deploy --push --update-aliases 3.3 latest
3131
mike set-default --push latest

README.md

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -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
3535

3636
## 📣 近期更新
3737

38+
🔥🔥 **2025.10.16,发布 PaddleX v3.3.0**,新增能力如下:
39+
40+
- **支持PaddleOCR-VL、PP-OCRv5多语种模型的推理部署能力。**
41+
3842
🔥🔥 **2025.8.20,发布 PaddleX v3.2.0**,新增能力如下:
3943

4044
- **部署能力升级:**

README_en.md

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -37,6 +37,10 @@ PaddleX 3.0 is a low-code development tool for AI models built on the PaddlePadd
3737

3838
## 📣 Recent Updates
3939

40+
🔥🔥 **2025.10.16, PaddleX v3.3.0 Released**
41+
42+
- **Added support for inference and deployment of PaddleOCR-VL and PP-OCRv5 multilingual models.**
43+
4044
🔥🔥 **2025.8.20, PaddleX v3.2.0 Released**
4145

4246
- **Deployment Capability Upgrades:**

deploy/hps/sdk/pipelines/PaddleOCR-VL/server/pipeline_config.yaml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -13,7 +13,7 @@ format_block_content: False
1313
SubModules:
1414
LayoutDetection:
1515
module_name: layout_detection
16-
model_name: PP-DocLayoutV2-L
16+
model_name: PP-DocLayoutV2
1717
model_dir: null
1818
batch_size: 8
1919
threshold:

docs/CHANGELOG.en.md

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -6,6 +6,10 @@ comments: true
66

77
## Latest Version Information
88

9+
### PaddleX v3.3.0 (October 16, 2025)
10+
11+
- **Added support for inference and deployment of PaddleOCR-VL and PP-OCRv5 multilingual models.**
12+
913
### PaddleX v3.2.0(8.20/2025)
1014

1115
- **Deployment Capability Upgrades:**

docs/CHANGELOG.md

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -6,6 +6,10 @@ comments: true
66

77
## 最新版本信息
88

9+
### PaddleX v3.3.0(10.16/2025)
10+
11+
- **支持PaddleOCR-VL、PP-OCRv5多语种模型的推理部署能力。**
12+
913
### PaddleX v3.2.0(8.20/2025)
1014

1115
- **部署能力升级:**

docs/pipeline_usage/tutorials/ocr_pipelines/PaddleOCR-VL.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -291,7 +291,7 @@ PaddleX 所提供的模型产线均可以快速体验效果,你可以在本地
291291

292292
在本地使用通用版面解析v3产线前,请确保您已经按照[PaddleX本地安装教程](../../../installation/installation.md)完成了PaddleX的wheel包安装。如果您希望选择性安装依赖,请参考安装教程中的相关说明。该产线对应的依赖分组为 `ocr`
293293

294-
> 在进行 GPU 推理时,默认配置可能会占用超过 16 GB 的显存,请务必确认 GPU 显存容量充足。若需减少显存占用,可参考下文说明修改配置文件,关闭不必要的功能
294+
> 默认配置暂不支持 Compute Capability 低于 8.0 的 GPU(如 V100、RTX 3060 等)。请参阅下一节,了解如何在此类 GPU 上使用推理加速框架
295295
296296
### 2.1 命令行方式体验
297297
一行命令即可快速体验版面解析产线效果,使用 [测试文件](https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/paddleocr_vl_demo.png),并将 `--input` 替换为本地路径,进行预测

paddlex/inference/genai/models/__init__.py

Lines changed: 2 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -42,18 +42,14 @@ def _check_model_name_and_backend(model_name, backend):
4242
def get_model_dir(model_name, backend):
4343
_check_model_name_and_backend(model_name, backend)
4444

45-
if backend in ("vllm", "sglang"):
46-
suffix = "_paddle"
47-
else:
48-
suffix = "_torch"
4945
try:
50-
model_dir = official_models[model_name + suffix]
46+
model_dir = official_models[model_name]
5147
except Exception as e:
5248
raise RuntimeError(
5349
f"Could not prepare the official model for the {repr(model_name)} model with the {repr(backend)} backend."
5450
) from e
5551

56-
return model_dir
52+
return str(model_dir)
5753

5854

5955
def get_model_components(model_name, backend):

0 commit comments

Comments
 (0)