为什么CUDA不适合与EasyOrocr一起使用?
根据 pytorch 我在cmd
pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113
/I.SSTATIC.NET/XOQDX.PNG“ rel =” nofollow noreferrer“>
但是库达仍然没有。有人可以帮我吗?
import torch
print(torch.cuda.is_available())
输出将为false
According to Pytorch I used this command in cmd
pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113
But CUDA still unavailable. Could someone help me, please?
import torch
print(torch.cuda.is_available())
The output will be False
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(3)
我遇到了同样的问题,对我来说,解决方案是卸载Pytorch并使用CUDA 12.1安装夜间版本。我正在使用CUDA工具包12.3。
运行此操作并查看它是否比更新驱动程序更小:
只需确保您的Pytorch版本支持您相应的CUDA工具包,并且都不会使其他版本不匹配。如果这无济于事,请参阅 @trsvchn 关于更新驱动程序的响应=“ https://docs.nvidia.com/cuda/cuda-installation-guide-microsoft-windows/index.html” rel =“ nofollow noreferrer”>系统要求。
弄清楚我什么时候做:
...并意识到我有CPU版本...
I had this same issue, and the fix for me was uninstalling Pytorch and installing the Nightlys version with CUDA 12.1. I am using the CUDA Toolkit 12.3.
Run this and see if it's a smaller issue than updating drivers:
Just make sure your Pytorch version supports your corresponding CUDA Toolkit version, and neither mismatches the other. If this doesn't help, see the response from @trsvchn about updating drivers, or check if you meet the system requirements.
Figured out when I did:
...and realized that I had the CPU version...
您必须首先更新驱动程序:
这是NVIDIA网站的概念图
这是另一个:
nofollow noreferrer”>
cuda兼容性
You have to update driver first:
Here is a concept diagram from nvidia website
Here is another one:
More at CUDA Compatibility
您仍然必须安装 cuda-toolkit 和 cudnn 在您的机器上
You still have to install cuda-toolkit and cudnn on your machine