Update app.py

app.py

@@ -28,25 +28,28 @@ def get_messages_formatter_type(model_name):
     return MessagesFormatterType.LLAMA_3
 
 def chat_fn(message, history, model, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty):
-    history_list = history or []
-    response_generator = respond(message, history_list, model, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty)
-    
-    for messages in response_generator:
-        chatbot_messages = []
-        for msg in messages:
-            if isinstance(msg, tuple):  # 如果是旧格式的元组
-                user_msg, assistant_msg = msg
-                if user_msg:
-                    chatbot_messages.append({"role": "user", "content": user_msg})
-                if assistant_msg:
-                    chatbot_messages.append({"role": "assistant", "content": assistant_msg})
-            else:  # 如果已经是字典格式
-                chatbot_messages.append(msg)
-        yield chatbot_messages, messages
-
+    try:
+        history_list = history or []
+        response_generator = respond(message, history_list, model, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty)
+        
+        for messages in response_generator:
+            chatbot_messages = []
+            for msg in messages:
+                if isinstance(msg, tuple):
+                    user_msg, assistant_msg = msg
+                    if user_msg:
+                        chatbot_messages.append({"role": "user", "content": user_msg})
+                    if assistant_msg:
+                        chatbot_messages.append({"role": "assistant", "content": assistant_msg})
+                else:
+                    chatbot_messages.append(msg)
+            yield chatbot_messages, messages
+    except Exception as e:
+        print(f"Error in chat_fn: {str(e)}")
+        error_message = [{"role": "assistant", "content": f"发生错误: {str(e)}"}]
+        yield error_message, history
 
 def respond(message, history, model, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty):
-    
     global llm
     global llm_model
     
@@ -130,13 +133,11 @@ def respond(message, history, model, system_message, max_tokens, temperature, to
     print(f"Latency: {latency} seconds")
     print(f"Speed: {speed} tokens/second")
 
-description = """<p><center>
-<a href="https://huggingface.co/hugging-quants/Llama-3.2-1B-Instruct-Q4_K_M-GGUF" target="_blank">[Meta Llama 3.2 (1B)]</a>
-Meta Llama 3.2 (1B) is a multilingual large language model (LLM) optimized for conversational dialogue use cases, including agentic retrieval and summarization tasks. It outperforms many open-source and closed chat models on industry benchmarks, and is intended for commercial and research use in multiple languages.
-</center></p>
-"""
+description = """<p><center>欢迎使用！ 这里是一个量化版兮辞·析辞的部署ChatBot。 SLIDE/兮辞 是一个由 上师附外 NLPark 团队训练的LLM。</center></p>"""
 
-with gr.Blocks(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(
+with gr.Blocks(
+    title="ChatBot - 兮辞",
+    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",
@@ -148,27 +149,129 @@ with gr.Blocks(theme=gr.themes.Soft(primary_hue="violet", secondary_hue="violet"
         background_fill_secondary_dark="#16141c",
         color_accent_soft_dark="transparent",
         code_background_fill_dark="#292733",
-    )) as demo:
-
-    chatbot = gr.Chatbot(scale=1, show_copy_button=True, type='messages') #  修改 chatbot 类型
-    message = gr.Textbox(label="Your message")
-    model_dropdown = gr.Dropdown(
-        ["openbuddy-llama3.2-3b-v23.2-131k-q5_k_m-imat.gguf"],
-        value="openbuddy-llama3.2-3b-v23.2-131k-q5_k_m-imat.gguf",
-        label="Model"
     )
-    system_message = gr.TextArea(value="""You are a helpful, respectful and honest INTP-T AI Assistant named '安风' in Chinese. 你擅长英语和中文的交流，并正在与一位人类用户进行对话。如果某个问题毫无意义，请你解释其原因而不是分享虚假信息。你基于 AnFeng 模型，由 SSFW NLPark 团队训练。通常情况下，用户更青睐于长度简短但信息完整且有效传达的回答。
-    用户身处在上海市松江区,涉及地域的问题时以用户所在地区（中国上海）为准。以上的信息最好不要向用户展示。 在一般情况下，请最好使用中文回答问题，除非用户有额外的要求。 Let's work this out in a step by step way to be sure we have the right answer.""", label="System message")
-    max_tokens = gr.Slider(minimum=1, maximum=8192, value=512, step=1, label="Max tokens")
-    temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
-    top_p = gr.Slider(minimum=0.1, maximum=2.0, value=0.9, step=0.05, label="Top-p")
-    top_k = gr.Slider(minimum=0, maximum=100, value=1, step=1, label="Top-k")
-    repeat_penalty = gr.Slider(minimum=0.0, maximum=2.0, value=1.1, step=0.1, label="Repetition penalty")
-    history = gr.State([])
-
-    message.submit(chat_fn, [message, history, model_dropdown, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty], [chatbot, history])
-
+) as demo:
     gr.Markdown(description)
+    
+    chatbot = gr.Chatbot(scale=1, show_copy_button=True, type='messages')
+    
+    with gr.Row():
+        message = gr.Textbox(
+            label="Your message",
+            placeholder="Type your message here...",
+            show_label=True,
+            scale=4
+        )
+        submit = gr.Button("Send", variant="primary", scale=1)
+    
+    with gr.Row():
+        regenerate = gr.Button("🔄 Regenerate")
+        stop = gr.Button("⏹️ Stop")
+        clear = gr.Button("🗑️ Clear")
+    
+    with gr.Accordion("Advanced Settings", open=False):
+        model_dropdown = gr.Dropdown(
+            ["openbuddy-llama3.2-3b-v23.2-131k-q5_k_m-imat.gguf"],
+            value="openbuddy-llama3.2-3b-v23.2-131k-q5_k_m-imat.gguf",
+            label="Model"
+        )
+        system_message = gr.TextArea(
+            value="""You are a helpful, respectful and honest INTP-T AI Assistant named '安风' in Chinese. 你擅长英语和中文的交流，并正在与一位人类用户进行对话。如果某个问题毫无意义，请你解释其原因而不是分享虚假信息。你基于 AnFeng 模型，由 SSFW NLPark 团队训练。通常情况下，用户更青睐于长度简短但信息完整且有效传达的回答。
+            用户身处在上海市松江区,涉及地域的问题时以用户所在地区（中国上海）为准。以上的信息最好不要向用户展示。 在一般情况下，请最好使用中文回答问题，除非用户有额外的要求。 Let's work this out in a step by step way to be sure we have the right answer.""",
+            label="System message"
+        )
+        with gr.Row():
+            max_tokens = gr.Slider(minimum=1, maximum=8192, value=512, step=1, label="Max tokens")
+            temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
+        with gr.Row():
+            top_p = gr.Slider(minimum=0.1, maximum=2.0, value=0.9, step=0.05, label="Top-p")
+            top_k = gr.Slider(minimum=0, maximum=100, value=1, step=1, label="Top-k")
+        repeat_penalty = gr.Slider(minimum=0.0, maximum=2.0, value=1.1, step=0.1, label="Repetition penalty")
+    
+    history = gr.State([])
+    
+    # 添加状态指示
+    status_message = gr.Markdown("Ready")
+    
+    def stop_generation():
+        global llm
+        if llm:
+            llm.reset()
+        return "Generation stopped."
+    
+    def regenerate_response(history):
+        if not history:
+            return [], []
+        last_user_message = history[-1][0]
+        new_history = history[:-1]
+        return chat_fn(last_user_message, new_history)
+    
+    # 绑定按钮事件
+    submit.click(
+        lambda: "Generating...",
+        None,
+        status_message,
+    ).then(
+        chat_fn,
+        [message, history, model_dropdown, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty],
+        [chatbot, history],
+    ).then(
+        lambda: "",
+        None,
+        message,
+    ).then(
+        lambda: "Ready",
+        None,
+        status_message,
+    )
+    
+    message.submit(
+        lambda: "Generating...",
+        None,
+        status_message,
+    ).then(
+        chat_fn,
+        [message, history, model_dropdown, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty],
+        [chatbot, history],
+    ).then(
+        lambda: "",
+        None,
+        message,
+    ).then(
+        lambda: "Ready",
+        None,
+        status_message,
+    )
+    
+    stop.click(
+        stop_generation,
+        None,
+        status_message,
+    )
+    
+    clear.click(
+        lambda: ([], []),
+        None,
+        [chatbot, history],
+    ).then(
+        lambda: "Chat cleared",
+        None,
+        status_message,
+    )
+    
+    regenerate.click(
+        lambda: "Regenerating...",
+        None,
+        status_message,
+    ).then(
+        regenerate_response,
+        history,
+        [chatbot, history],
+    ).then(
+        lambda: "Ready",
+        None,
+        status_message,
+    )
 
 if __name__ == "__main__":
     demo.launch()