优化聊天预测功能，移除预览模式状态信息并调整界面元素布局

app.py

@@ -153,9 +153,9 @@ def chat_predict(
     # 如果是预览模式，返回模拟响应
     if not ENABLE_VLLM:
         user_display = user_text if user_text else "[音频输入]"
-        mock_response = f"[预览模式] 这是一个模拟回复。您说: {user_text[:50] if user_text else '音频'}"
+        mock_response = f"这是一个模拟回复。您说: {user_text[:50] if user_text else '音频'}"
         chat_history.append((user_display, mock_response))
-        return chat_history, "✓ 预览模式（未启用 vLLM）"
+        return chat_history, ""
     
     messages = format_messages(system_prompt, chat_history, user_text, audio_file)
     if not messages:
@@ -178,12 +178,11 @@ def chat_predict(
         assistant_message = outputs[0].outputs[0].text
         user_display = user_text if user_text else "[音频输入]"
         chat_history.append((user_display, assistant_message))
-        status = f"✓ 推理完成（耗时 {latency:.2f}s）"
-        return chat_history, status
+        return chat_history, ""
     except Exception as e:
         import traceback
         traceback.print_exc()
-        return chat_history, f"✗ 推理失败: {e}"
+        return chat_history, ""
 
 
 def _get_llm() -> LLM:
@@ -210,52 +209,28 @@ def _set_llm_args(**kwargs) -> None:
     _llm = None  # 确保使用新配置重新加载
 
 
-def check_model_status() -> str:
-    """返回模型当前加载状态"""
-    if not ENABLE_VLLM:
-        return "⚙ 界面预览模式（vLLM 未启用）"
-    
-    model_path = LLM_ARGS["model"]
-    if _llm is None:
-        return f"等待加载：{model_path}"
-    return f"✓ 已加载模型：{model_path}"
 
 
-def warmup_model() -> str:
-    """主动加载模型"""
-    if not ENABLE_VLLM:
-        return "⚙ 界面预览模式（vLLM 未启用）"
-    
-    try:
-        _get_llm()
-        return check_model_status()
-    except Exception as exc:
-        import traceback
-        traceback.print_exc()
-        return f"✗ 模型加载失败: {exc}"
-
 
 # 构建 Gradio 界面
 with gr.Blocks(title="Step Audio 2 Chat", theme=gr.themes.Soft()) as demo:
     gr.Markdown(
         """
-        # Step Audio 2 Chat Interface
-        
-        支持文本和音频输入的聊天界面，直接在本地 vLLM 引擎上推理。
+        # Step Audio R1 Demo
         """
     )
     
-    # 模型状态
-    with gr.Row():
-        status_text = gr.Textbox(
-            label="模型状态",
-            value="检查中...",
-            interactive=False
-        )
-        check_btn = gr.Button("加载/检查模型", variant="secondary")
-    
     with gr.Row():
-        # 左侧：输入区域
+        # 左侧：聊天历史
+        with gr.Column(scale=1):
+            gr.Markdown("### 聊天历史")
+            chatbot = gr.Chatbot(
+                label="对话",
+                height=600,
+                show_copy_button=True
+            )
+        
+        # 右侧：输入区域
         with gr.Column(scale=1):
             gr.Markdown("### 输入设置")
             
@@ -275,15 +250,15 @@ with gr.Blocks(title="Step Audio 2 Chat", theme=gr.themes.Soft()) as demo:
             audio_file = gr.Audio(
                 label="音频输入",
                 type="filepath",
-                sources=["upload", "microphone"]
+                sources=["microphone", "upload"]
             )
             
             with gr.Row():
                 max_tokens = gr.Slider(
                     label="Max Tokens",
                     minimum=1,
-                    maximum=8192,
-                    value=2048,
+                    maximum=16384,
+                    value=8192,
                     step=1
                 )
             
@@ -306,19 +281,10 @@ with gr.Blocks(title="Step Audio 2 Chat", theme=gr.themes.Soft()) as demo:
             
             submit_btn = gr.Button("提交", variant="primary", size="lg")
             clear_btn = gr.Button("清空", variant="secondary")
-        
-        # 右侧：聊天历史
-        with gr.Column(scale=1):
-            gr.Markdown("### 聊天历史")
-            chatbot = gr.Chatbot(
-                label="对话",
-                height=600,
-                show_copy_button=True
-            )
+            
+            status_text = gr.Textbox(label="状态", interactive=False, visible=False)
     
     # 事件绑定
-    check_btn.click(fn=warmup_model, outputs=status_text)
-    
     submit_btn.click(
         fn=chat_predict,
         inputs=[
@@ -337,9 +303,6 @@ with gr.Blocks(title="Step Audio 2 Chat", theme=gr.themes.Soft()) as demo:
         fn=lambda: ([], "", None),
         outputs=[chatbot, user_text, audio_file]
     )
-    
-    # 页面加载时显示状态
-    demo.load(fn=check_model_status, outputs=status_text)
 
 
 if __name__ == "__main__":