bugfix of openai_api_server for fastchat.serve.vllm_worker

[Rayrtfr]

Why are these changes needed?

when i start a vllm_worker:

CUDA_VISIBLE_DEVICES="0,1" python3 -m fastchat.serve.vllm_worker --model-path /mnt/ssd1/models \
--host localhost \
--port 21002 \
--worker-address "http://localhost:21002" \
--limit-worker-concurrency 5 \
--num-gpus 2

so , here need a format check for vllm_worker.

Related issue number (if applicable)

Checks

• I've run format.sh to lint the changes in this PR.
• I've included any doc changes needed.
• I've made sure the relevant tests are passing (if applicable).

[infwinston]

thanks for the contribution! I think there's one place we can improve to reduce LoC.

[infwinston]

Why don't we do this to reduce LoC and repeated codes

if isinstance(content["text"], str):
    contents = [context["text"]]
for t in contents:
    choices.append(
        ChatCompletionResponseChoice(
            index=i,
            message=ChatMessage(role="assistant", content=t),
            finish_reason=content.get("finish_reason", "stop"),
        )
    )

[infwinston]

@Rayrtfr Would be awesome if you can submit another PR to refactor this :)

[merrymercy]

@Rayrtfr The commit (#2372) that introduced the bug was reverted, so we do not need this fix anymore. This fix is also reverted. Thanks for your contributions anyway!

[Rayrtfr]

@Rayrtfr The commit (#2372) that introduced the bug was reverted, so we do not need this fix anymore. This fix is also reverted. Thanks for your contributions anyway!

ok