fastchat.serve.model_worker --port 21003 and /v1/models show nothing

Nobody has claimed this yet.

I'm trying to build my openai compatible api via fastchat. And below is my command line:
`python3 -m fastchat.serve.controller --host 0.0.0.0`
`python -m fastchat.serve.model_worker --model-path ~/jupyter/Marseus/Llama-2-13B-chat-GPTQ --port 21003 --host 0.0.0.0`
`python3 -m fastchat.serve.huggingface_api_worker --model-info-file /home/kslab/jupyter/Marseus/models_api.json --host 0.0.0.0`
`python -m fastchat.serve.openai_api_server --host 0.0.0.0`

My goal is to build both my local model and huggingface inference api. But when i go to /v1/models, it only show me those models in models_api.json.
i try run only one model_worker
`python -m fastchat.serve.model_worker --model-path ~/jupyter/Marseus/Llama-2-13B-chat-GPTQ --port 21003 --host 0.0.0.0`
and it still doesn't work. But when i remove the `--port 21003`. It worked. How can i fix this?
Is there any method for me to deploy both my local model and huggingface inference api?

Start with the FastChat entry points named in the report: fastchat.serve.model_worker, fastchat.serve.huggingface_api_worker, and fastchat.serve.openai_api_server. Compare the behavior of model_worker with and without --port 21003, then inspect how /v1/models discovers workers and reads models_api.json; done means both the local model and the Hugging Face models appear.

Written by the indexing model from the issue text.