Paddle 使用1.8.4 2.0.0rc1 的C++ 预测库推理时,同一个模型,同样的数据,运行多次随机出错

klsxnrf1  于 2022-10-20  发布在  其他
关注(0)|答案(4)|浏览(173)
  • 版本、环境信息:

   1)PaddlePaddle版本:1.8.4 和 2.0.0rc1
   2)CPU:I7 6900K使用 MKL/MKLDNN
   4)系统环境:Ubuntu 16.04,GCC 5.4.0 20160609
-预测信息
   1)C++预测:
GIT COMMIT ID d82d59e
WITH_MKL: ON
WITH_MKLDNN: ON
WITH_GPU: OFF
CXX compiler version: 5.4.0

 2)CMake包含路径的完整命令
cmake .. -DWITH_PYTHON=OFF -DWITH_GPU=OFF -DWITH_TESTING=OFF -DCMAKE_BUILD_TYPE=Release -DON_INFER=ON -DFLUID_INFERENCE_INSTALL_DIR=/home/jiachx/thunisoft_ocr_dev/paddle_inference_install_dir_forcpu -DWITH_MKLDNN=ON -DWITH_MKL=ON
   3)预测库来源:源码编译

  • 复现信息:使用的是ppocr的server模型
  • 问题描述:多次运行,有时次成功有时报错

terminate called after throwing an instance of 'paddle::platform::EnforceNotMet'
what(): In user code:

File "tools/export_model.py", line 75, in
main()
File "tools/export_model.py", line 51, in main
config, eval_program, startup_prog)
File "/paddle/PaddleOCR/PaddleOCR/tools/program.py", line 215, in build_export
image, outputs = model(mode='export')
File "/paddle/PaddleOCR/PaddleOCR/ppocr/modeling/architectures/det_model.py", line 135, incall
conv_feas = self.backbone(image)
File "/paddle/PaddleOCR/PaddleOCR/ppocr/modeling/backbones/det_resnet_vd.py", line 117, incall
name=conv_name)
File "/paddle/PaddleOCR/PaddleOCR/ppocr/modeling/backbones/det_resnet_vd.py", line 239, in basic_block
name=name + "_branch2a")
File "/paddle/PaddleOCR/PaddleOCR/ppocr/modeling/backbones/det_resnet_vd.py", line 138, in conv_bn_layer
bias_attr=False)
File "/root/anaconda3/envs/deploy/lib/python3.7/site-packages/paddle/fluid/layers/nn.py", line 1585, in conv2d
"data_format": data_format,
File "/root/anaconda3/envs/deploy/lib/python3.7/site-packages/paddle/fluid/layer_helper.py", line 43, in append_op
return self.main_program.current_block().append_op(args,kwargs)
File "/root/anaconda3/envs/deploy/lib/python3.7/site-packages/paddle/fluid/framework.py", line 2880, in append_op
attrs=kwargs.get("attrs", None))
File "/root/anaconda3/envs/deploy/lib/python3.7/site-packages/paddle/fluid/framework.py", line 1977, in
init
*
for frame in traceback.extract_stack():

InvalidArgumentError: The Input Variable(Input) of (conv2d) Operator used to determine kernel data type is empty or not LoDTensor or SelectedRows or LoDTensorArray.
[Hint: Expected data_type != dafault_data_type, but received data_type:-1 == dafault_data_type:-1.] (at /CPPproject/Paddle/paddle/fluid/framework/operator.cc:1479)
[Hint: If you need C++ stacktraces for debugging, please set FLAGS_call_stack_level=2 .]
[operator < conv2d > error]
已放弃 (核心已转储)

e4eetjau

e4eetjau1#

您好,我们已经收到了您的问题,会安排技术人员尽快解答您的问题,请耐心等待。请您再次检查是否提供了清晰的问题描述、复现代码、环境&版本、报错信息等。同时,您也可以通过查看 官网API文档常见问题历史IssueAI社区 来寻求解答。祝您生活愉快~

Hi! We've received your issue and please be patient to get responded. We will arrange technicians to answer your questions as soon as possible. Please make sure that you have posted enough message to demo your request. You may also check out the APIFAQGithub Issue and AI community to get the answer.Have a nice day!

q5lcpyga

q5lcpyga2#

上述说是自己编译的预测库,请使用release/2.0编译预测库,测试一下。
同时,请补充复现的脚本、模型、数据和readme。

dgsult0t

dgsult0t3#

@kasyoukin 有新的进展更新吗?

g6ll5ycj

g6ll5ycj4#

@juncaipeng 应该是PaddleOCR检测模型的问题,但原因不知,换别的模型就不报错了

相关问题