(paddle) cmw@ubuntu:~/cmw/PaddleDetection$ python tools/train.py -c configs/yolov3/yolov3_mobilenet_v1_roadsign.yml
/home/cmw/.conda/envs/paddle/lib/python3.9/site-packages/paddle/tensor/creation.py:125: DeprecationWarning: `np.object` is a deprecated alias for the builtin `object`. To silence this warning, use `object` by itself. Doing this will not modify any behavior and is safe.
Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
if data.dtype == np.object:
Warning: import ppdet from source directory without installing, run 'python setup.py install' to install ppdet firstly
[11/22 13:51:42] ppdet.utils.download WARNING: Config annotation dataset/roadsign_voc/train.txt is not a file, dataset config is not valid
[11/22 13:51:42] ppdet.utils.download INFO: Dataset /home/cmw/cmw/PaddleDetection/dataset/roadsign_voc is not valid for reason above, try searching /home/cmw/.cache/paddle/dataset or downloading dataset...
[11/22 13:51:42] ppdet.utils.download INFO: Found /home/cmw/.cache/paddle/dataset/roadsign_voc/annotations
[11/22 13:51:42] ppdet.utils.download INFO: Found /home/cmw/.cache/paddle/dataset/roadsign_voc/images
W1122 13:51:42.332911 4665 device_context.cc:404] Please NOTE: device: 0, GPU Compute Capability: 7.5, Driver API Version: 11.3, Runtime API Version: 11.2
W1122 13:51:42.337199 4665 device_context.cc:422] device: 0, cuDNN Version: 8.2.
/home/cmw/.conda/envs/paddle/lib/python3.9/site-packages/paddle/tensor/creation.py:125: DeprecationWarning: `np.object` is a deprecated alias for the builtin `object`. To silence this warning, use `object` by itself. Doing this will not modify any behavior and is safe.
Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
if data.dtype == np.object:
[11/22 13:51:44] ppdet.utils.checkpoint INFO: The shape [255] in pretrained weight yolo_head.yolo_output.0.bias is unmatched with the shape [27] in model yolo_head.yolo_output.0.bias. And the weight yolo_head.yolo_output.0.bias will not be loaded
[11/22 13:51:44] ppdet.utils.checkpoint INFO: The shape [255, 1024, 1, 1] in pretrained weight yolo_head.yolo_output.0.weight is unmatched with the shape [27, 1024, 1, 1] in model yolo_head.yolo_output.0.weight. And the weight yolo_head.yolo_output.0.weight will not be loaded
[11/22 13:51:44] ppdet.utils.checkpoint INFO: The shape [255] in pretrained weight yolo_head.yolo_output.1.bias is unmatched with the shape [27] in model yolo_head.yolo_output.1.bias. And the weight yolo_head.yolo_output.1.bias will not be loaded
[11/22 13:51:44] ppdet.utils.checkpoint INFO: The shape [255, 512, 1, 1] in pretrained weight yolo_head.yolo_output.1.weight is unmatched with the shape [27, 512, 1, 1] in model yolo_head.yolo_output.1.weight. And the weight yolo_head.yolo_output.1.weight will not be loaded
[11/22 13:51:44] ppdet.utils.checkpoint INFO: The shape [255] in pretrained weight yolo_head.yolo_output.2.bias is unmatched with the shape [27] in model yolo_head.yolo_output.2.bias. And the weight yolo_head.yolo_output.2.bias will not be loaded
[11/22 13:51:44] ppdet.utils.checkpoint INFO: The shape [255, 256, 1, 1] in pretrained weight yolo_head.yolo_output.2.weight is unmatched with the shape [27, 256, 1, 1] in model yolo_head.yolo_output.2.weight. And the weight yolo_head.yolo_output.2.weight will not be loaded
[11/22 13:51:44] ppdet.utils.checkpoint INFO: Finish loading model weights: /home/cmw/.cache/paddle/weights/yolov3_mobilenet_v1_270e_coco.pdparams
Could not load library libcudnn_cnn_train.so.8. Error: libcudnn_ops_train.so.8: cannot open shared object file: No such file or directory
Please make sure libcudnn_cnn_train.so.8 is in your library path!
--------------------------------------
C++ Traceback (most recent call last):
--------------------------------------
0 paddle::imperative::BasicEngine::Execute()
1 paddle::imperative::PreparedOp::Run(paddle::imperative::NameVariableWrapperMap const&, paddle::imperative::NameVariableWrapperMap const&, paddle::framework::AttributeMap const&)
2 std::_Function_handler, paddle::operators::CUDNNConvGradOpKernel, paddle::operators::CUDNNConvGradOpKernel >::operator()(char const*, char const*, int) const::{lambda(paddle::framework::ExecutionContext const&)#1}>::_M_invoke(std::_Any_data const&, paddle::framework::ExecutionContext const&)
3 paddle::operators::CUDNNConvGradOpKernel::Compute(paddle::framework::ExecutionContext const&) const
4 cudnnConvolutionBwdFilterAlgo_t paddle::operators::SearchAlgorithm::Find(paddle::operators::ConvArgs const&, bool, bool, paddle::framework::ExecutionContext const&)
5 cudnnGetConvolutionBackwardFilterAlgorithm_v7
6 paddle::framework::SignalHandle(char const*, int)
7 paddle::platform::GetCurrentTraceBackString[abi:cxx11]()
----------------------
Error Message Summary:
----------------------
FatalError: `Process abort signal` is detected by the operating system.
[TimeInfo: *** Aborted at 1637560305 (unix time) try "date -d @1637560305" if you are using GNU date ***]
[SignalInfo: *** SIGABRT (@0x3e700001239) received by PID 4665 (TID 0x7fb81d33e700) from PID 4665 ***]
Aborted (core dumped)
CuDNN版本不对?
Could not load library libcudnn_cnn_train.so.8. Error: libcudnn_ops_train.so.8: cannot open shared object file: No such file or directory
没有安装cudnn,要使用GPU的话,仅仅安装了CUDA是不够的,还需要安装对应版本的cudnn.
2楼答案是正确的
插眼