Use ChatHuggingFace wrapper for better model support

src/rag/llm.py

@@ -1,4 +1,4 @@
-from langchain_huggingface import HuggingFaceEndpoint
+from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
 from langchain_core.prompts import ChatPromptTemplate
 
 from src.prompts import SYSTEM_PROMPT
@@ -8,12 +8,15 @@ from src.core.config import settings
 def get_chain():
     prompt = ChatPromptTemplate.from_template(SYSTEM_PROMPT)
 
+    # Create base endpoint
     llm = HuggingFaceEndpoint(
         repo_id=settings.MODEL_NAME,
         huggingfacehub_api_token=settings.HUGGINGFACE_TOKEN,
         temperature=settings.MODEL_TEMPERATURE,
         max_new_tokens=settings.MODEL_MAX_TOKENS,
-        task="conversational",
     )
+    
+    # Wrap with ChatHuggingFace for better conversational support
+    chat_llm = ChatHuggingFace(llm=llm)
 
-    return prompt | llm
+    return prompt | chat_llm