serving mpt-7b-bigdl.yaml crash when install with pip install -e .[bigdl-cpu]

[xwu99]

First create an new conda env with python 3.9, and install with

pip install -e .[bigdl-cpu] -f https://developer.intel.com/ipex-whl-stable-cpu -f https://download.pytorch.org/whl/torch_stable.html

Then run:

python inference/serve.py --config_file inference/models/bigdl/mpt-7b-bigdl.yaml --serve_simple

It looks something is messed after installing bigdl-llm:

After I switched to another env, it's OK.

@KepingYan @jiafuzha

[xwu99]

I did some test and found latest fastapi 0.109 and starlette 0.35.1 cause the crash,
I reinstall fastapi 0.104 and starlette 0.27.0 it's OK.
For reproduce, you need to stop ray and start over.

[KepingYan]

After installing env via pip install .[cpu] -f https://developer.intel.com/ipex-whl-stable-cpu -f https://download.pytorch.org/whl/torch_stable.html, I also encountered this problem. fastapi is installed together with ray[serve], I've verified 0.104<=fastapi<=0.109 all works well. The above problem will be fixed by #61.