Update app.py

app.py

@@ -4,7 +4,6 @@ from llama_cpp import Llama
 from concurrent.futures import ThreadPoolExecutor, as_completed
 import uvicorn
 import huggingface_hub
-import spaces
 import re
 from dotenv import load_dotenv
 
@@ -75,6 +74,7 @@ class ModelManager:
         return models
 
 model_manager = ModelManager()
+model_manager.load_all_models()  # Cargar modelos una vez al inicio
 
 class ChatRequest(BaseModel):
     message: str
@@ -119,10 +119,8 @@ def remove_repetitive_responses(responses):
     return unique_responses
 
 @app.post("/generate")
-@spaces.GPU(duration=0)
 async def generate(request: ChatRequest):
     try:
-        global_data['models'] = model_manager.load_all_models()
         responses = []
         with ThreadPoolExecutor() as executor:
             futures = [executor.submit(generate_chat_response, request, model_data) for model_data in global_data['models']]