refactor: Update app.py and libs/llm.py to improve model selection and configuration

app.py

@@ -1,6 +1,7 @@
 import os
 import gradio as gr
 from kiwipiepy import Kiwi
+from typing import List, Tuple, Generator, Union
 
 from langchain_core.output_parsers import StrOutputParser
 from langchain_core.runnables import RunnablePassthrough, RunnableLambda
@@ -24,16 +25,17 @@ embeddings = get_embeddings()
 retriever = load_retrievers(embeddings)
 
 
-def kiwi_tokenize(text):
-    kiwi = Kiwi()
-    return [token.form for token in kiwi.tokenize(text)]
-
-
-embeddings = get_embeddings()
-retriever = load_retrievers(embeddings)
+# 사용 가능한 모델 목록 (key: 모델 식별자, value: 사용자에게 표시할 레이블)
+AVAILABLE_MODELS = {
+    "gpt_3_5_turbo": "GPT-3.5 Turbo",
+    "gpt_4o": "GPT-4o",
+    "claude_3_5_sonnet": "Claude 3.5 Sonnet",
+    "gemini_1_5_flash": "Gemini 1.5 Flash",
+    "llama3_70b": "Llama3 70b",
+}
 
 
-def create_rag_chain(chat_history, model):
+def create_rag_chain(chat_history: List[Tuple[str, str]], model: str):
     llm = get_llm(streaming=STREAMING).with_config(configurable={"llm": model})
     prompt = get_prompt(chat_history)
 
@@ -49,32 +51,30 @@ def create_rag_chain(chat_history, model):
     )
 
 
-def respond_stream(message, history, model):
+def get_model_key(label):
+    return next(key for key, value in AVAILABLE_MODELS.items() if value == label)
+
+
+def respond_stream(
+    message: str, history: List[Tuple[str, str]], model: str
+) -> Generator[str, None, None]:
     rag_chain = create_rag_chain(history, model)
     for chunk in rag_chain.stream(message):
         yield chunk
 
 
-def respond(message, history, model):
+def respond(message: str, history: List[Tuple[str, str]], model: str) -> str:
     rag_chain = create_rag_chain(history, model)
     return rag_chain.invoke(message)
 
 
-# 사용 가능한 모델 목록 (key: 모델 식별자, value: 사용자에게 표시할 레이블)
-AVAILABLE_MODELS = {
-    "gpt_3_5_turbo": "GPT-3.5 Turbo",
-    "gpt_4o": "GPT-4o",
-    "claude_3_5_sonnet": "Claude 3.5 Sonnet",
-    "gemini_1_5_flash": "Gemini 1.5 Flash",
-    "llama3_70b": "Llama3 70b",
-}
-
-
-def get_model_key(label):
+def get_model_key(label: str) -> str:
     return next(key for key, value in AVAILABLE_MODELS.items() if value == label)
 
 
-def chat_function(message, history, model_label):
+def chat_function(
+    message: str, history: List[Tuple[str, str]], model_label: str
+) -> Generator[str, None, None]:
     model_key = get_model_key(model_label)
     if STREAMING:
         response = ""

libs/llm.py

@@ -30,7 +30,7 @@ def get_llm(streaming=True):
             callbacks=[StreamCallback()],
         ),
         gpt_3_5_turbo=ChatOpenAI(
-            model="gpt-3.5-turbo",
+            model="gpt-3.5-turbo-0125",
             temperature=0,
             streaming=streaming,
             callbacks=[StreamCallback()],