Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
thomas 2d5e3847c8 | 2 years ago | |
---|---|---|
doc | 2 years ago | |
yaml | 2 years ago | |
.gitattributes | 2 years ago | |
.gitignore | 2 years ago | |
LICENSE | 2 years ago | |
README.md | 2 years ago | |
README_dev.md | 2 years ago | |
RELEASE.md | 2 years ago | |
apply.sh | 2 years ago | |
applyrm.sh | 2 years ago | |
calico.yaml | 2 years ago | |
check.sh | 2 years ago | |
clear.sh | 2 years ago | |
create_user.sh | 2 years ago | |
docker.sh | 2 years ago | |
ingress.sh | 2 years ago | |
init_env.sh | 2 years ago | |
install.sh | 2 years ago | |
k8s.sh | 2 years ago | |
lab.sh | 2 years ago | |
menus.sql | 2 years ago | |
monitor.sh | 2 years ago | |
nfs.sh | 2 years ago | |
preset_menus.sh | 2 years ago | |
preset_models.sh | 2 years ago | |
preset_models.sql | 2 years ago | |
remove.sh | 2 years ago | |
repair.sh | 2 years ago | |
retry_preset.sh | 2 years ago | |
ruser.sh | 2 years ago |
安装主机的操作系统
安装GPU驱动
执行安装指令
确认网络IP或域名
查看安装登陆信息,并在chrome中登陆平台
管理员初始化平台配置
Tips 在安装执行结束前,请不要关闭终端或休眠主机。
手动安装GPU驱动
Tips 如果使用的是阿里云,AWS,Azure 等云服务器,GPU驱动可能已经安装好,只需检查或升级驱动版本。
查询GPU型号
apt install -y neofetch && neofetch --gpu_brand on --gpu_type dedicated --off
从NVIDIA官网下载相应的驱动到服务器本地
i) `dpkg -i nvidia-diag-driver-local-repo-ubuntu1804-410.129_1.0-1_amd64.deb’ for Ubuntu
ii)
apt-get update
iii)
apt-get install cuda-drivers
iv)
reboot
``
检查驱动是否安装好
nvidia-smi
使用root执行安装指令
安装日志默认放在 /var/log/apulis_installation-[DATAE].log
sudo apt install -y git unzip tar wget curl && sudo git clone -b develop https://gitee.com/apulisplatform/installer.git && cd installer && bash install.sh | tee /var/log/apulis_installation-$(date "+%Y%m%d%H%M%S").log
Please enter IP address for the platform LAN service:
The service IP you entered is: [LAN IP]
Do you want to configure Public Ip? [yes/no] yes #(如果是云vm部署配置了public-ip选择yes)
Do you want to configure Domainame? [yes/no] yes #(如果是配置了domainame选择yes)
Please enter the IP for public network access. If you don\'t have public IP, please use LAN IP instead: [PUBLIC IP] # 默认作为平台访问IP
Please enter domainname: [DOMAINNAME]
看到如下输出内容,说明平台已基本安装完成,管理员可以登陆平台查看了!
#>>>>>>-The platform has been successfully deployed !-----------------------------------------------------------
#>>>>>>-Platform login information:
Public-ip:****** Private-address:****** Domainame:******
UserName:admin Password:Wwjpfb
Platform data dir:/data/nfs/pvc
平台默认会预置一些常用的模型,数据集和docker镜像,但由于文件比较大,下载时间长容易中断,可能使得在使用平台的时候出现一些文件找不到或运行失败或任务挂起等常见问题;
您可以参考我们的预置数据列表,自行下载或配置相关文件,也可以直接联系我们给您同步数据或文件。
Rerfer /data/model-gallery/models
model templates | datasets | engin image |
---|---|---|
ResNet50_mindspore_gpu | dog-vs-cat | mindspore:1.1.1-gpu |
LeNet_pytorch_gpu_scratch | mnist | pytorch:1.6.0-gpu |
LeNet_tensorflow_gpu_scratch | fashion-mnist | tensorflow:1.14.0-gpu-py3 |
Inceptionv3_tensorflow_gpu_scratch | dog-vs-cat | tensorflow:1.14.0-gpu-py3 |
请参考快速使用指引, 如有疑问或异常问题可以直接联系我们,提交问题到Issues
警告 ! 基础虚拟环境,数据库,存储服务器,镜像服务等会被全部重置,务必提前备份密钥证书和数据,否则数据可能会丢失!
bash remove.sh
注意:集群重置会将整个环境清除,请做好关键数据的备份, 在执行remove.sh是会提示
实际环境的GPU驱动,cuda版本,框架版本都支持使用者根据自己的需要升级更新
平台支持上传coco数据集,其他voc等标注格式也训练和测试,推荐tar,tar.gz格式。
我们提供的环境以python3为主,推荐使用python3.7+
华为 Model Zoo, 或其他社区(tensorflow, pytorch, paddle)上下载的模型, 需要参考平台环境参数和算力做一定的适配。
我们非常欢迎大家,试用平台,共同探讨AI世界,思考后AI时代的到来!
可以尝试执行安装目录下的bash retry_preset.sh | tee /var/log/retry-timeout-$(date "+%Y%m%d%H%M%S").log
脚本重新配置资源。
清理安装环境后,重新使用screen重新执行安装指令。
2.1. 执行安装目录下的 bash remove.sh
, 不备份任何文件; 清理安装环境;
2.2. 再在安装目录下执行 cd .. && mv installer installer-back
,将已经下载的安装包备份或移除;
2.3. 重新执行以下安装指令。
# 使用screen执行安装指令
sudo apt update && apt install -y screen git \
&& screen_name=$"my_screen" \
&& screen -dmS $screen_name \
&& screen -x -S $screen_name -p 0 -X stuff "sudo git clone -b develop https://gitee.com/apulisplatform/installer.git && cd installer && bash install.sh | tee /var/log/apulis_installation-$(date "+%Y%m%d%H%M%S").log" \
&& screen -x -S $screen_name -p 0 -X stuff $'\n' \
&& screen -r $screen_name
# 退出 screen
screen -x -S $screen_name -p 0 -X stuff "exit"
screen -x -S $screen_name -p 0 -X stuff "\n"
依瞳人工智能平台安装部署工具
Shell Python
MIT
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》