ajuste

rag_utils.py

@@ -24,13 +24,12 @@ _embedding_instance = None
 _model_instance = None
 _tokenizer = None
 
-@spaces.GPU
 def get_embedding_model():
     global _embedding_instance
     if _embedding_instance is None:
         if config.local_emb_path is None:
             raise ValueError("⚠️ config.local_emb_path ainda não foi inicializado!")
-        _embedding_instance = HuggingFaceEmbeddings(model_name=config.local_emb_path)
+        _embedding_instance = HuggingFaceEmbeddings(model_name=config.local_emb_path, model_kwargs={"device": "cpu"})
     return _embedding_instance
 
 # model_name = "Qwen/Qwen2.5-7B-Instruct-GPTQ-Int8" #"Qwen/Qwen2.5-7B-Instruct-AWQ" #"Qwen/Qwen2.5-7B-Instruct"