Fix that I am very skeptical of

llm_rag_setup.py

@@ -6,7 +6,7 @@ from langchain_community.vectorstores import FAISS
 from langchain_huggingface import HuggingFaceEmbeddings
 from langchain_community.retrievers import BM25Retriever
 from langchain.retrievers import EnsembleRetriever
-from langchain_huggingface import HuggingFaceEndpoint
+from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
 from langchain.chains import RetrievalQA
 from langchain.prompts import PromptTemplate
 from typing import Optional
@@ -120,15 +120,16 @@ def init_rag():
 
     print("🔄 Initializing HuggingFace LLM (Qwen2.5-7B-Instruct)...")
     try:
-        llm = HuggingFaceEndpoint(
+        llm_endpoint = HuggingFaceEndpoint(
             repo_id="Qwen/Qwen2.5-7B-Instruct",
             huggingfacehub_api_token=HF_TOKEN,
             temperature=0.3,
             max_new_tokens=1024,
             top_p=0.95,
             repetition_penalty=1.1,
-            task="conversational",
         )
+        
+        llm = ChatHuggingFace(llm=llm_endpoint)
         print("✅ HuggingFace LLM initialized (Qwen2.5-7B-Instruct)")
     except Exception as e:
         print(f"❌ Failed to initialize LLM: {e}")