Update app.py

app.py

@@ -85,7 +85,7 @@ def get_messages_formatter_type(model_name):
 @spaces.GPU(duration=120)
 def respond(
     message,
-    history: list[tuple[str, str]],
+    history: list[dict],  # history 항목이 tuple이 아닌 dict 형식으로 전달됨
     system_message,
     max_tokens,
     temperature,
@@ -99,17 +99,17 @@ def respond(
     chat_template = get_messages_formatter_type(MISTRAL_MODEL_NAME)
     
     # 모델 파일 경로 확인
-    model_path = os.path.join("./models", MISTRAL_MODEL_NAME)
+    model_path_local = os.path.join("./models", MISTRAL_MODEL_NAME)
     
-    print(f"Model path: {model_path}")
+    print(f"Model path: {model_path_local}")
     
-    if not os.path.exists(model_path):
-        print(f"Warning: Model file not found at {model_path}")
+    if not os.path.exists(model_path_local):
+        print(f"Warning: Model file not found at {model_path_local}")
         print(f"Available files in ./models: {os.listdir('./models')}")
     
     if llm is None or llm_model != MISTRAL_MODEL_NAME:
         llm = Llama(
-            model_path=model_path,
+            model_path=model_path_local,
             flash_attn=True,
             n_gpu_layers=81,
             n_batch=1024,
@@ -136,17 +136,18 @@ def respond(
 
     messages = BasicChatHistory()
 
+    # history의 각 항목이 dict 형식으로 {'user': <user_message>, 'assistant': <assistant_message>} 형태라고 가정
     for msn in history:
-        user = {
+        user_message = {
             'role': Roles.user,
-            'content': msn[0]
+            'content': msn.get('user', '')
         }
-        assistant = {
+        assistant_message = {
             'role': Roles.assistant,
-            'content': msn[1]
+            'content': msn.get('assistant', '')
         }
-        messages.add_message(user)
-        messages.add_message(assistant)
+        messages.add_message(user_message)
+        messages.add_message(assistant_message)
     
     stream = agent.get_chat_response(
         message,
@@ -192,7 +193,12 @@ demo = gr.ChatInterface(
     fn=respond,
     title="Ginigen Private AI",
     description="6BIT 양자화로 모델 크기는 줄이고 성능은 유지하는 프라이버시 중심 AI 솔루션.",
-    theme=gr.themes.Soft(primary_hue="violet", secondary_hue="violet", neutral_hue="gray",font=[gr.themes.GoogleFont("Exo"), "ui-sans-serif", "system-ui", "sans-serif"]).set(
+    theme=gr.themes.Soft(
+        primary_hue="violet",
+        secondary_hue="violet",
+        neutral_hue="gray",
+        font=[gr.themes.GoogleFont("Exo"), "ui-sans-serif", "system-ui", "sans-serif"]
+    ).set(
         body_background_fill_dark="#16141c",
         block_background_fill_dark="#16141c",
         block_border_width="1px",
@@ -227,4 +233,4 @@ demo = gr.ChatInterface(
 )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()