File tree Expand file tree Collapse file tree 1 file changed +3
-3
lines changed
docs/pipeline_usage/tutorials/ocr_pipelines Expand file tree Collapse file tree 1 file changed +3
-3
lines changed Original file line number Diff line number Diff line change @@ -292,7 +292,7 @@ PaddleX 所提供的模型产线均可以快速体验效果,你可以在本地
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 ` 。此外,为了使用飞桨框架读取 safetensors 格式模型,请执行如下命令安装 safetensors:
293293
294294``` bash
295- pytyon -m pip install https://paddle-whl.bj.bcebos.com/nightly/cu126/safetensors/safetensors-0.6.2.dev0-cp38-abi3-linux_x86_64.whl
295+ python -m pip install https://paddle-whl.bj.bcebos.com/nightly/cu126/safetensors/safetensors-0.6.2.dev0-cp38-abi3-linux_x86_64.whl
296296```
297297
298298> 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。
@@ -961,8 +961,8 @@ paddlex --get_pipeline_config PaddleOCR-VL
961961VLRecognition :
962962 ...
963963 genai_config :
964- backend : vllm
965- server_url : http://127.0.0.1:8118
964+ backend : vllm-server
965+ server_url : http://127.0.0.1:8118/v1
966966` ` `
967967
968968之后,可以使用修改好的配置文件进行产线调用。例如通过 CLI 调用:
You can’t perform that action at this time.
0 commit comments