Vision-CAIR / LTVRR

MIT License
35 stars 6 forks source link

请问要让代码在cuda11.0以及pytorch1.7.1的环境中运行应该修改什么内容呢 #3

Open tao123322 opened 3 years ago

aniket-agarwal1999 commented 3 years ago

Hi, can you type your issue in English. I am not able to understand the issue you just posted

tao123322 commented 3 years ago

Sorry for my question. What should be modified to make the code run in the environment of cuda11.0 and pytorch1.7.1?

------------------ 原始邮件 ------------------ 发件人: "Vision-CAIR/LTVRR" @.>; 发送时间: 2021年10月9日(星期六) 晚上9:44 @.>; @.**@.>; 主题: Re: [Vision-CAIR/LTVRR] 请问要让代码在cuda11.0以及pytorch1.7.1的环境中运行应该修改什么内容呢 (#3)

Hi, can you type your issue in English. I am not able to understand the issue you just posted

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

aniket-agarwal1999 commented 3 years ago

Hi, we would have to check it up. I don't think much needs to be changed though. Apart from this we have also provided the environment file which you can use to easily create the virtualenv required for running the codebase.

tao123322 commented 3 years ago

Our server has minimum support for CUDA11.0, so the environment provided is not good for code to work! I hope you can help me see how to modify. Thank you so much!

---原始邮件--- 发件人: "Aniket @.> 发送时间: 2021年10月10日(周日) 晚上7:01 收件人: @.>; 抄送: @.**@.>; 主题: Re: [Vision-CAIR/LTVRR] 请问要让代码在cuda11.0以及pytorch1.7.1的环境中运行应该修改什么内容呢 (#3)

Hi, we would have to check it up. I don't think much needs to be changed though. Apart from this we have also provided the environment file which you can use to easily create the virtualenv required for running the codebase.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

aniket-agarwal1999 commented 3 years ago

Hi, have you tried running the current codebase on Pytorch 1.7.1, I don't think there would be much issues and should be some minor dependency issues. We can better help out if you could maybe share an error log or something?

YerongLi2 commented 2 years ago

Yes, I also got the CUDA version issue, the error message I am getting is

ImportError: /home/wentaoy2/LTVRR/lib/model/roi_pooling/_ext/roi_pooling/_roi_pooling.so: undefined symbol: __cudaRegisterFatBinaryEnd

Which CUDA version are you using? And the version of CUDA and pytorch I am on is

$ pip list| grep torch torch 0.4.1.post2
torchvision 0.1.8 $ nvcc --version nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2021 NVIDIA Corporation Built on Mon_Oct_11_21:27:02_PDT_2021 Cuda compilation tools, release 11.4, V11.4.152 Build cuda_11.4.r11.4/compiler.30521435_0 (vr) wentaoy2@bock:~/LTVRR$