Update response perf.

app.py

@@ -47,7 +47,8 @@ def format_text(text):
     return text
 
 
-def get_chat_response(chatbot, task_history, model, tokenizer):
+def get_chat_response(chatbot, task_history):
+    global model, tokenizer 
     chat_query = chatbot[-1][0]
     query = task_history[-1][0]
     history_cp = copy.deepcopy(task_history)
@@ -156,8 +157,8 @@ with gr.Blocks(css=css) as demo:
     
     gr.Markdown("### Key Features:\n- **Strong Performance**: Surpasses existing LVLMs on multiple English benchmarks including Zero-shot Captioning and VQA.\n- **Multi-lingual Support**: Supports English, Chinese, and multi-lingual conversation.\n- **High Resolution**: Utilizes 448*448 resolution for fine-grained recognition and understanding.")
     submit_btn.click(handle_text_input, [chatbot, task_history, query], [chatbot, task_history]).then(
-    get_chat_response, [chatbot, task_history, model, tokenizer], [chatbot], show_progress=True
-)
+        get_chat_response, [chatbot, task_history], [chatbot], show_progress=True
+    )