cnstark / pytorch-docker Goto Github PK

View Code? Open in Web Editor NEW

349.0 349.0 59.0 114 KB

Pure Pytorch Docker Images.

License: Apache License 2.0

Dockerfile 5.97% Shell 60.73% Python 33.31%

centos cuda deep-learning docker nvidia pytorch ubuntu

pytorch-docker's People

Contributors

Stargazers

Watchers

Forkers

simmoncn moonbunnyzzz batcom samge0 aprlie qiangzhangcv 1678484950 thn0000 yakun0622 xinguozju akatsukisun2020 wowoyoho canglangzhishuiz hanjf12 edenbuaa ai-x-king xiaozhoushi hqzqaq kizunaiix koolo233 lbq1223android gsakun ozone-oo ovesteve hailong1991 helloliver cellmook derkodex fuxuliu marenan c66w albertzhanghit mazhaoshuo gdtiti jingx8885 twtcer xiangzia ustc-mcclab treizeor dengshuibing aboysky logda yanmenglu jingziyou a-mess lishuanghua deeplearning-cn rocapp chris-han ssli1225 shanha-kim justbejyk limboys miawu-jojo sensir jincheng-ai hanyunzhidao

pytorch-docker's Issues

If the image contians the dependence e.g. numpy when pip pytorch?

Pytorch Versions 里面，没有pytorch版本支持CUDA11.2的

hi，[昊長大人]：
pytorch的版本里面，没有找到支持cuda11.2的，导致生成sh失败，看看是否可以添加更新？
Thx

Req: pytorch_1.13.1_py3.9.18_cuda11.8_ubuntu22.04

Could you add this image? thank you

CUDA版本问题

我们一台服务器有多人使用，所以原主机的CUDA版本是11.1比较低。我可以直接拉取比较高CUDA版本的镜像使用嘛，比如CUDA11.6。使用的是A100GPU，这样会和原主机不兼容嘛。

Thanks for your great work!
I have docker pull some of your images, meeting the issue : sudo, vim, as well as vi command not found. Could you have a check? You can test the latest one : cnstark/pytorch:1.12.1-py3.9.12-cuda11.6.2-ubuntu20.04.
looking forward to your reply

About Dockerfile

Can you publish the dockerfile for subsequent customization?

你好，你能给我分享一下复现occ3d中mask_camera生成的代码吗，谢谢

Pytorch 2.0 support

Are you planning to add pytorch 2.0 support? Thanks for this great collection of scripts.

To my understanding some of the awkward torch==<VERSION><SUFFIX> can simply be replaced by torch==<VERSION>, while torch automatically detects CUDA and installs it torch with gpu support

Where is the CUDA path?

Thanks for your great work!
I am using the following docker image:
cnstark/pytorch:1.12.0-py3.9.12-cuda11.6.2-ubuntu20.04

I can not found the cuda path. /usr/local/cuda not exisits. As onnxruntime, TensorRT need the cuda well installed, could you provide more informations about this?

Req: pytorch_1.13.1_py3.9.16_cuda11.7.1_ubuntu22.04

Could you add a new build pytorch_1.13.1_py3.9.16_cuda11.7.1_ubuntu22.04, GPU runtime and pure CPU two images.
Anyway, thank you very much.

Why the size of the image on my server is much larger than yours?

The image size of "1.13.1-py3.9.16-cuda11.7.1-devel-ubuntu20.04" on my server is 10.6 GB, while the size of that displayed on your git is 5.69 GB?

pytorch:1.13.1-py3.8.16-cuda11.7.1-devel-ubuntu20.04

你好，安装完之后 nvcc -V显示了 cuda 版本，但是 nvidia-smi 命令返回错误 command not found