xinference 用内置executor 跑deepseek MTP 模式性能低

[ZhikaiGuo960110]

Is your feature request related to a problem? Please describe

A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
目前用xinference 内置的executor 去跑deepseek MTP 模式，性能会低，大概低25%
【相比较于用vllm内置原生的executor】

Describe the solution you'd like

A clear and concise description of what you want to happen.
有可能是因为xoscar 内部没有对pytorch 做zero copy 序列化处理。目前还需要调研

Describe alternatives you've considered

A clear and concise description of any alternative solutions or features you've considered.
在序列化处理逻辑添加。 https://github.com/xorbitsai/xoscar/tree/main/python/xoscar/serialization

Additional context

Add any other context or screenshots about the feature request here.

[ZhikaiGuo960110]

这个地方要研究，也希望社区一起关注。