有nvidia和ascend环境的镜像吗

[ZHUHF123]

在modelscope模型使用页面上看到发布的docker镜像环境都已经找不到了，有新的nvidia和ascend环境的镜像吗

[Caozhou1995]

@ZHUHF123 https://huggingface.co/FlagRelease/models， 可以看下这里是否有满足你需求的镜像。

[ZHUHF123]

@ZHUHF123 https://huggingface.co/FlagRelease/models， 可以看下这里是否有满足你需求的镜像。

用的就是这个里面的镜像源，可能是版本太旧，已经没法docker pull了
https://huggingface.co/FlagRelease/Qwen3-4B-FlagOS-Ascend

[Caozhou1995]

麻烦看下这个问题 @shh2000

[shh2000]

hi @ZHUHF123 aliyuncs has been deprecated, and we have migrated of images to BAAI Harbor. Please refer to the instructions provided by FlagRelease on Hugging Face or ModelScope before pulling the images.

[shh2000]

hi @ZHUHF123 aliyuncs has been deprecated, and we have migrated of images to BAAI Harbor. Please refer to the instructions provided by FlagRelease on Hugging Face or ModelScope before pulling the images.

https://huggingface.co/FlagRelease

[ZHUHF123]

hi @ZHUHF123 aliyuncs has been deprecated, and we have migrated of images to BAAI Harbor. Please refer to the instructions provided by FlagRelease on Hugging Face or ModelScope before pulling the images.

感谢回复，我在harbor.baai.ac.cn/拉取了对应镜像
昇腾服务器镜像是 docker pull harbor.baai.ac.cn/flagrelease-public/flagopen-910b-ubuntu24.04.2-py311_ascend_a2
按照官方的启动指令：
cd /workspace/FlagScale/examples/qwen3/conf/serve
flagscale serve qwen3 ./4b.yaml
昇腾机器报错如下，yaml文件感觉和启动的指令不匹配，

在英伟达的机器上我也下载了对应的权重和镜像
镜像是docker pull harbor.baai.ac.cn/flagrelease-public/nv_vllm085_gemscbbf39:1.1
但是内部环境是不能用的，没有flag_scale，pip install安装flag_scale之后运行flagscale serve qwen3，也是别的包不匹配

好像都是环境不太匹配，有英伟达和昇腾显卡推荐的镜像吗，测试的模型是Qwen3-4B对应的英伟达和昇腾的权重文件是FlagRelease/Qwen3-4B-FlagOS-Nvidia和FlagRelease/Qwen3-4B-FlagOS-Ascend