libcudart.so.9.0：无法打开共享对象文件：没有这样的文件或目录答案

【问题标题】：libcudart.so.9.0: cannot open shared object file: No such file or directorylibcudart.so.9.0：无法打开共享对象文件：没有这样的文件或目录
【发布时间】：2020-01-27 08:35:53
【问题描述】：

我在 Ubuntu 18.04 下使用 Pytorch 并尝试使用 import torchvision，但收到错误 libcudart.so.9.0: cannot open shared object file: No such file or directory。

有人可以帮忙解决吗？谢谢。

以下信息是详细的错误日志：

Traceback (most recent call last):
  File "/home/x/.local/lib/python3.6/site-packages/IPython/core/interactiveshell.py", line 2882, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-2-6dd351122000>", line 1, in <module>
    import torchvision
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/__init__.py", line 1, in <module>
    from torchvision import models
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/models/__init__.py", line 11, in <module>
    from . import detection
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/models/detection/__init__.py", line 1, in <module>
    from .faster_rcnn import *
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/models/detection/faster_rcnn.py", line 7, in <module>
    from torchvision.ops import misc as misc_nn_ops
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/ops/__init__.py", line 1, in <module>
    from .boxes import nms, box_iou
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/ops/boxes.py", line 2, in <module>
    from torchvision import _C
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
ImportError: libcudart.so.9.0: cannot open shared object file: No such file or directory
import torch
import torch.nn as nn
import torchvision.transforms as transforms
Traceback (most recent call last):
  File "/home/x/.local/lib/python3.6/site-packages/IPython/core/interactiveshell.py", line 2882, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-4-677acbcfae34>", line 1, in <module>
    import torchvision.transforms as transforms
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/__init__.py", line 1, in <module>
    from torchvision import models
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/models/__init__.py", line 11, in <module>
    from . import detection
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/models/detection/__init__.py", line 1, in <module>
    from .faster_rcnn import *
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/models/detection/faster_rcnn.py", line 7, in <module>
    from torchvision.ops import misc as misc_nn_ops
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/ops/__init__.py", line 1, in <module>
    from .boxes import nms, box_iou
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
  File "/home/x/.local/lib/python3.6/site-packages/torchvision/ops/boxes.py", line 2, in <module>
    from torchvision import _C
  File "/home/x/pycharm-2019.2/helpers/pydev/_pydev_bundle/pydev_import_hook.py", line 21, in do_import
    module = self._system_import(name, *args, **kwargs)
ImportError: libcudart.so.9.0: cannot open shared object file: No such file or directory

【问题讨论】：

标签： cuda pytorch torchvision

【解决方案1】：

如果您使用的是anaconda，以下可能会解决您的问题。

conda install -c anaconda cudatoolkit==9.0

您也可以尝试以下方法。

确保 CUDA 版本为 9.0。并将以下两行添加到~/.bashrc。

export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

然后，运行：

source  ~/.bashrc

将以下行添加到/etc/ld.so.conf.d/cuda.conf

/usr/local/cuda/lib64

然后运行：

sudo ldconfig

【讨论】：

我还没用过conda，有必要安装它来解决这个问题吗？
不，它只是一种替代方法，在许多情况下设置环境的痛苦较小。
只是想知道我安装了cuda 10.0，为什么找不到libcudart 9.0？
您是否为 CUDA 9.0 或 10.0 安装了 PyTorch？
也许我应该为 cuda 10.0 重新安装 pytorch

【解决方案2】：

使用anaconda，火炬版本为：

pytorch==1.1.0
torchvision==0.3.0
cudatoolkit=10.0

遇到同样的问题，把torchvision==0.3.0改成torchvision==0.2.2解决

【讨论】：

【解决方案3】：

使用以下方法之一检查您的 CUDA 版本：
```
 nvidia-smi
 nvcc -V
```

像这样检查你的 Pytorch CUDA 版本：

 python
 import torch
 torch.version.cuda

版本应该匹配。否则，您需要升级/降级 CUDA 或对 Pytorch 执行相同操作。错误libcudart.so.9.0: cannot open shared object file: No such file or directory 表示版本不匹配。你有 CUDA 9.0 的 Pytorch，但你的 CUDA 是另一个版本。
要解决这个问题，请使用正确的 CUDA 版本安装 Pytorch，如下所示：
```
 conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=10.0 -c pytorch
```

注意：将 Pytorch 和 CUDA Toolkit 版本替换为所需版本。检查可用版本here。

如果适用，重建你的 Pytorch：
```
 rm -rf build
 python setup.py develop
```

【讨论】：