Update app.py

app.py

@@ -26,34 +26,20 @@ vectordb = Chroma(
 
 retriever = vectordb.as_retriever(search_kwargs={"k": 5})
 
-class KwArgsModel(BaseModel):
-    kwargs: Dict[str, Any] = Field(default_factory=dict)
-
-class CombinedMeta(type(LLM), type(KwArgsModel)):
-    pass
-
-class CustomInferenceClient(LLM, KwArgsModel, metaclass=CombinedMeta):
-    model_name: str
-    inference_client: InferenceClient
-
+class CustomInferenceClient:
     def __init__(self, model_name: str, hf_token: str, kwargs: Optional[Dict[str, Any]] = None):
-        inference_client = InferenceClient(model=model_name, token=hf_token)
-        super().__init__(
-            model_name=model_name,
-            hf_token=hf_token,
-            kwargs=kwargs,
-            inference_client=inference_client
-        )
-
-    def _call(
-        self,
-        prompt: str,
-        stop: Optional[List[str]] = None
-    ) -> str:
+        self.llm = LLM(model_name=model_name, hf_token=hf_token)  # LLM 인스턴스
+        self.kw_args_model = KwArgsModel(kwargs=kwargs)  # KwArgsModel 인스턴스
+        self.inference_client = InferenceClient(model=model_name, token=hf_token)
+        self.model_name = model_name
+
+    # 기존 메서드들을 약간 수정하여 내부 LLM과 KwArgsModel 인스턴스를 사용하게 함
+    def _call(self, prompt: str, stop: Optional[List[str]] = None) -> str:
         if stop is not None:
             raise ValueError("stop kwargs are not permitted.")
-        response_gen = self.inference_client.text_generation(prompt, **self.kwargs, stream=True)
-        response = ''.join(response_gen)  
+        kwargs = self.kw_args_model.kwargs  # KwArgsModel 인스턴스에서 kwargs를 가져옴
+        response_gen = self.inference_client.text_generation(prompt, **kwargs, stream=True)
+        response = ''.join(response_gen)
         return response
 
     @property