Update app.py

app.py

@@ -25,112 +25,112 @@ hf_hub_download(
 # 推論関数
 @spaces.GPU(duration=120)
 def respond(
-   message,
-   history: list[tuple[str, str]],
-   model,
-   system_message,
-   max_tokens,
-   temperature,
-   top_p,
-   top_k,
-   repeat_penalty,
+    message,
+    history: list[tuple[str, str]],
+    model,
+    system_message,
+    max_tokens,
+    temperature,
+    top_p,
+    top_k,
+    repeat_penalty,
 ):
-   chat_template = MessagesFormatterType.GEMMA_2
-
-   llm = Llama(
-       model_path=f"models/{model}",
-       flash_attn=True,
-       n_gpu_layers=81,
-       n_batch=1024,
-       n_ctx=8192,
-   )
-   provider = LlamaCppPythonProvider(llm)
-
-   agent = LlamaCppAgent(
-       provider,
-       system_prompt=f"{system_message}",
-       predefined_messages_formatter_type=chat_template,
-       debug_output=True
-   )
-
-   settings = provider.get_provider_default_settings()
-   settings.temperature = temperature
-   settings.top_k = top_k
-   settings.top_p = top_p
-   settings.max_tokens = max_tokens
-   settings.repeat_penalty = repeat_penalty
-   settings.stream = True
-
-   messages = BasicChatHistory()
-
-   for msn in history:
-       user = {
-           'role': Roles.user,
-           'content': msn[0]
-       }
-       assistant = {
-           'role': Roles.assistant,
-           'content': msn[1]
-       }
-       messages.add_message(user)
-       messages.add_message(assistant)
-
-   stream = agent.get_chat_response(
-       message,
-       llm_sampling_settings=settings,
-       chat_history=messages,
-       returns_streaming_generator=True,
-       print_output=False
-   )
-
-   outputs = ""
-   for output in stream:
-       outputs += output
-       yield outputs
+    chat_template = MessagesFormatterType.GEMMA_2
+
+    llm = Llama(
+        model_path=f"models/{model}",
+        flash_attn=True,
+        n_gpu_layers=81,
+        n_batch=1024,
+        n_ctx=8192,
+    )
+    provider = LlamaCppPythonProvider(llm)
+
+    agent = LlamaCppAgent(
+        provider,
+        system_prompt=f"{system_message}",
+        predefined_messages_formatter_type=chat_template,
+        debug_output=True
+    )
+
+    settings = provider.get_provider_default_settings()
+    settings.temperature = temperature
+    settings.top_k = top_k
+    settings.top_p = top_p
+    settings.max_tokens = max_tokens
+    settings.repeat_penalty = repeat_penalty
+    settings.stream = True
+
+    messages = BasicChatHistory()
+
+    for msn in history:
+        user = {
+            'role': Roles.user,
+            'content': msn[0]
+        }
+        assistant = {
+            'role': Roles.assistant,
+            'content': msn[1]
+        }
+        messages.add_message(user)
+        messages.add_message(assistant)
+
+    stream = agent.get_chat_response(
+        message,
+        llm_sampling_settings=settings,
+        chat_history=messages,
+        returns_streaming_generator=True,
+        print_output=False
+    )
+
+    outputs = ""
+    for output in stream:
+        outputs += output
+        yield [(message, outputs)]
 
 # Gradioのインターフェースを作成
 def create_interface(model_name, description):
-   return gr.ChatInterface(
-       respond,
-       additional_inputs=[
-           gr.Textbox(value=model_name, label="Model", interactive=False),
-           gr.Textbox(value="You are a helpful assistant.", label="System message"),
-           gr.Slider(minimum=1, maximum=4096, value=2048, step=1, label="Max tokens"),
-           gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-           gr.Slider(
-               minimum=0.1,
-               maximum=1.0,
-               value=0.95,
-               step=0.05,
-               label="Top-p",
-           ),
-           gr.Slider(
-               minimum=0,
-               maximum=100,
-               value=40,
-               step=1,
-               label="Top-k",
-           ),
-           gr.Slider(
-               minimum=0.0,
-               maximum=2.0,
-               value=1.1,
-               step=0.1,
-               label="Repetition penalty",
-           ),
-       ],
-       retry_btn="Retry",
-       undo_btn="Undo",
-       clear_btn="Clear",
-       submit_btn=None,  # Remove the individual submit button
-       title=f"Chat with Gemma 2 using llama.cpp - {model_name}",
-       description=description,
-       chatbot=gr.Chatbot(
-           scale=1,
-           likeable=False,
-           show_copy_button=True
-       )
-   )
+    return gr.ChatInterface(
+        respond,
+        additional_inputs=[
+            gr.Textbox(value=model_name, label="Model", interactive=False),
+            gr.Textbox(value="You are a helpful assistant.", label="System message"),
+            gr.Slider(minimum=1, maximum=4096, value=2048, step=1, label="Max tokens"),
+            gr.Slider(minimum=0.1, maximum 4.0, value=0.7, step=0.1, label="Temperature"),
+            gr.Slider(
+                minimum=0.1,
+                maximum=1.0,
+                value=0.95,
+                step=0.05,
+                label="Top-p",
+            ),
+            gr.Slider(
+                minimum=0,
+                maximum=100,
+                value=40,
+                step=1,
+                label="Top-k",
+            ),
+            gr.Slider(
+                minimum=0.0,
+                maximum=2.0,
+                value=1.1,
+                step=0.1,
+                label="Repetition penalty",
+            ),
+        ],
+        retry_btn="Retry",
+        undo_btn="Undo",
+        clear_btn="Clear",
+        submit_btn=None,  # Remove the individual submit button
+        title=f"Chat with Gemma 2 using llama.cpp - {model_name}",
+        description=description,
+        chatbot=gr.Chatbot(
+            scale=1,
+            likeable=False,
+            show_copy_button=True
+        )
+    )
 
 # 各モデルのインターフェース
 description_9b = """<p align="center">Gemma-2 9B it Model</p>"""
@@ -141,25 +141,25 @@ interface_27b = create_interface('gemma-2-27b-it-Q5_K_M.gguf', description_27b)
 
 # Gradio Blocksで2つのインターフェースを並べて表示
 with gr.Blocks() as demo:
-   #gr.Markdown("# Compare Gemma-2 9B and 27B Models")
-   with gr.Row():
-       with gr.Column():
-           input_field = gr.Textbox(label="Input", interactive=True)
-       with gr.Column():
-           interface_9b.render()
-       with gr.Column():
-           interface_27b.render()
-   
-   submit_btn = gr.Button("Send")
-   
-   def send_to_both(input_text):
-       return input_text, input_text
-
-   submit_btn.click(
-       fn=send_to_both,
-       inputs=input_field,
-       outputs=[interface_9b.chatbot, interface_27b.chatbot]
-   )
+    #gr.Markdown("# Compare Gemma-2 9B and 27B Models")
+    with gr.Row():
+        with gr.Column():
+            input_field = gr.Textbox(label="Input", interactive=True)
+        with gr.Column():
+            interface_9b.render()
+        with gr.Column():
+            interface_27b.render()
+
+    submit_btn = gr.Button("Send")
+
+    def send_to_both(input_text):
+        return input_text, input_text
+
+    submit_btn.click(
+        fn=send_to_both,
+        inputs=input_field,
+        outputs=[interface_9b.chatbot, interface_27b.chatbot]
+    )
 
 if __name__ == "__main__":
-   demo.launch()
+    demo.launch(share=True)