fix: set vllm based on platform to avoid error on windows

configs/server_config.py.example

@@ -38,7 +38,8 @@ FSCHAT_MODEL_WORKERS = {
         "host": DEFAULT_BIND_HOST,
         "port": 20002,
         "device": LLM_DEVICE,
-        "infer_turbo": False # 可选[False,'vllm'],使用的推理加速框架,使用vllm如果出现HuggingFace通信问题，参见doc/FAQ
+        # False,'vllm',使用的推理加速框架,使用vllm如果出现HuggingFace通信问题，参见doc/FAQ
+        "infer_turbo": "vllm" if sys.platform.startswith("linux") else False,
 
         # model_worker多卡加载需要配置的参数
         # "gpus": None, # 使用的GPU，以str的格式指定，如"0,1"，如失效请使用CUDA_VISIBLE_DEVICES="0,1"等形式指定

startup.py

@@ -78,7 +78,6 @@ def create_model_worker_app(log_level: str = "INFO", **kwargs) -> FastAPI:
     from fastchat.serve.model_worker import worker_id, logger
     import argparse
     import fastchat.serve.model_worker
-    import fastchat.serve.vllm_worker
     logger.setLevel(log_level)
 
     parser = argparse.ArgumentParser()
@@ -98,6 +97,7 @@ def create_model_worker_app(log_level: str = "INFO", **kwargs) -> FastAPI:
     else:
         from configs.model_config import VLLM_MODEL_DICT
         if kwargs["model_names"][0] in VLLM_MODEL_DICT and args.infer_turbo == "vllm":
+            import fastchat.serve.vllm_worker
             from fastchat.serve.vllm_worker import VLLMWorker,app
             from vllm import AsyncLLMEngine
             from vllm.engine.arg_utils import AsyncEngineArgs,EngineArgs