Update app.py

app.py

@@ -289,10 +289,12 @@ class LlamaAdapter:
             repeat_penalty=repeat_penalty
         )
     
-def load_model_gpu(model_type, model_path, n_gpu_layers, params):
-    llama = LlamaAdapter(model_path, params, n_gpu_layers)
-    print(f"{model_type} モデル {model_path} のロードが完了しました。(n_gpu_layers: {n_gpu_layers})")
-    return llama
+@spaces.GPU(duration=120)
+def chat_or_gen(text, gen_characters, gen_token_multiplier, instruction, mode):
+    if mode == "chat":
+        return character_maker.generate_response(text)
+    elif mode == "gen":
+        return character_maker.generate_text(text, gen_characters, gen_token_multiplier, instruction)
     
 class CharacterMaker:
     def __init__(self):
@@ -342,12 +344,6 @@ class CharacterMaker:
                 print(f"{model_type} モデルのロード中にエラーが発生しました: {e}")
                 self.model_loaded.set()
 
-    @spaces.GPU(duration=120)
-    def chat_or_gen(self, text, gen_characters, gen_token_multiplier, instruction, mode):
-        if mode == "chat":
-            return self.generate_response(text)
-        elif mode == "gen":
-            return self.generate_text(text, gen_characters, gen_token_multiplier, instruction)
 
 
     def generate_text_gen_pre(self, text, gen_characters, gen_token_multiplier, instruction):
@@ -479,20 +475,22 @@ def chat_with_character(message, history):
         character_maker.chat_history = [{"role": "user" if i % 2 == 0 else "assistant", "content": msg} for i, msg in enumerate(sum(history, []))]
     else:
         character_maker.history = [{"user": h[0], "assistant": h[1]} for h in history]
-    return character_maker.chat_or_gen(text=message,gen_characters=None, gen_token_multiplier=None, instruction=None, mode="chat")
+    return chat_or_gen(text=message, gen_characters=None, gen_token_multiplier=None, instruction=None, mode="chat")
 
 def chat_with_character_stream(message, history):
     if character_maker.use_chat_format:
         character_maker.chat_history = [{"role": "user" if i % 2 == 0 else "assistant", "content": msg} for i, msg in enumerate(sum(history, []))]
     else:
         character_maker.history = [{"user": h[0], "assistant": h[1]} for h in history]
-    response = character_maker.chat_or_gen(text=message,gen_characters=None, gen_token_multiplier=None, instruction=None, mode="chat")
+    response = chat_or_gen(text=message, gen_characters=None, gen_token_multiplier=None, instruction=None, mode="chat")
     for i in range(len(response)):
         time.sleep(0.05)  # 各文字の表示間隔を調整
         yield response[:i+1]
-def clear_chat():
-    character_maker.reset()
-    return []
+
+# 文章生成関連関数
+def generate_text_wrapper(text, gen_characters, gen_token_multiplier, instruction):
+    return chat_or_gen(text=text, gen_characters=gen_characters, gen_token_multiplier=gen_token_multiplier, instruction=instruction, mode="gen")
+
 
 # ログ関連関数
 def list_log_files():
@@ -711,7 +709,7 @@ def build_gradio_interface():
                 generated_output = gr.Textbox(label="生成された文章")
             
                 generate_button.click(
-                    character_maker.generate_text_pre,
+                     generate_text_wrapper,
                     inputs=[gen_input_text, gen_characters, gen_token_multiplier, gen_instruction],
                     outputs=[generated_output]
                 )