From be5b9998337ab40ffdcd253f7e1e6464115801de Mon Sep 17 00:00:00 2001 From: MRXLT Date: Thu, 19 Dec 2019 16:45:00 +0800 Subject: [PATCH] fix serving doc --- docs/export_for_infer.md | 2 +- docs/serving.md | 7 ++++--- 2 files changed, 5 insertions(+), 4 deletions(-) diff --git a/docs/export_for_infer.md b/docs/export_for_infer.md index 9a3e0aa..960ca42 100644 --- a/docs/export_for_infer.md +++ b/docs/export_for_infer.md @@ -2,7 +2,7 @@ 通常，PaddlePaddle大规模分类库在训练过程中保存的模型只保存模型参数信息， 而不包括预测模型结构。为了部署PLSC预测库，需要将预训练模型导出为预测模型。 -预测模型包括预测所需要的模型参数和模型结构，用于后续地预测任务(参见[C++预测库使用]) +预测模型包括预测所需要的模型参数和模型结构，用于后续地预测任务(参见[C++预测库使用](./serving.md)) 可以通过下面的代码将预训练模型导出为预测模型： diff --git a/docs/serving.md b/docs/serving.md index da7a379..f19d8fd 100644 --- a/docs/serving.md +++ b/docs/serving.md @@ -6,9 +6,9 @@ server端 需要python3环境，下载whl包 -https://paddle-serving.bj.bcebos.com/paddle-gpu-serving/wheel/plsc_serving-0.1.4-py3-none-any.whl +https://paddle-serving.bj.bcebos.com/paddle-gpu-serving/wheel/plsc_serving-0.1.6-py3-none-any.whl -pip3 install plsc_serving-0.1.4-py3-none-any.whl +pip3 install plsc_serving-0.1.6-py3-none-any.whl client端 @@ -25,7 +25,8 @@ server端 ```python from plsc_serving.run import PLSCServer fs = PLSCServer() -fs.with_model(model_name = 'face_resnet50') +#设定使用的模型文路径，str类型，绝对路径 +fs.with_model(model_path = '/XXX/XXX') #跑单个进程,gpu_index指定使用的gpu，int类型，默认为0；port指定使用的端口，int类型，默认为8866 fs.run(gpu_index = 0, port = 8010) ``` -- GitLab