Update app.py

app.py

@@ -23,7 +23,7 @@ generator = ctranslate2.Generator("./FixedStar-BETA-7b-ct2", device=device)
 tokenizer = AutoTokenizer.from_pretrained(
     "./tokenizer", use_fast=True)
 
-def inference_func(prompt, max_length=128, sampling_temperature=0.7):
+def inference_func(prompt, max_length=64, sampling_temperature=0.7):
     tokens = tokenizer.convert_ids_to_tokens(
         tokenizer.encode(prompt, add_special_tokens=False)
     )
@@ -63,13 +63,11 @@ def interact_func(message, chat_history, max_context_size, max_length, sampling_
     chat_history.append((message, generated))
     return "", chat_history
 
-gr.Markdown(DESCRIPTION)
-
-with gr.Blocks(theme="monochrome", css="style.css") as demo:
+with gr.Blocks(theme="monochrome") as demo:
     with gr.Accordion("Configs", open=False):
         # max_context_size = the number of turns * 2
         max_context_size = gr.Number(value=20, label="記憶する会話ターン数", precision=0)
-        max_length = gr.Number(value=128, label="最大文字数", precision=0)
+        max_length = gr.Number(value=64, label="最大文字数", precision=0)
         sampling_temperature = gr.Slider(0.0, 2.0, value=0.7, step=0.1, label="創造性")
     chatbot = gr.Chatbot()
     msg = gr.Textbox()
@@ -81,5 +79,7 @@ with gr.Blocks(theme="monochrome", css="style.css") as demo:
     )
     clear.click(lambda: None, None, chatbot, queue=False)
 
+gr.Markdown(DESCRIPTION)
+
 if __name__ == "__main__":
     demo.launch(debug=True, share=True)