eigenLiu

i'll fix the CI warns soon.

this issue can be solved： https://github.com/vllm-project/vllm/issues/4369 and these issues can be solved when using MoE models： https://github.com/vllm-project/vllm/issues/3931 https://github.com/vllm-project/vllm/issues/3563

@WoosukKwon @zhuohan123 could you please arrage someone to review my pr ? thanks~~

amd test said: requests.exceptions.HTTPError: 401 Client Error: Unauthorized for url: https://huggingface.co/meta-llama/Llama-2-7b-chat-hf/resolve/main/config.json Failed to connect to localhost port 8000: Connection refused timeout 600 bash -c 'until curl localhost:8000/v1/models; do sleep 1;...

@WoosukKwon @zhuohan123 now ci passed, could you please give a review？thanks ^_^

> @JustinLin610 can you help take a look at this? Thank you! 🙏 thank you for replying! @JustinLin610 HI ^_^ ，qwen member, could you please give this pr a review?...

@JustinLin610 @yangapku @simonJJJ @logicwong @JianxinMa @hzhwcmhf @fyabc @huybery 各位阿里千问member灯塔级大神，谁能帮忙确认一下这个pr嘛？ 欢迎帮忙提供任何观点，多谢啦！！

> I think the PR is functionally okay. thank you for your approvement!! @simon-mo HI ~ it seems qwen members are all very busy ...

> bo is our member. yes, i just noticed that this is merged into transformers. somehow it is not necessary cuz we do not have this setup for our models,...

这种问题的最终解决办法是解耦。swift训完的模型，怎么转回的基座的样式，让大家想怎么弄怎么弄。 https://github.com/modelscope/swift/issues/838