Update src/model_load.py

src/model_load.py

@@ -15,7 +15,7 @@ def load_model():
     model = AutoModelForCausalLM.from_pretrained(model_name,
                                                  device_map="auto",
                                                  quantization_config=BitsAndBytesConfig(load_in_8bit=True),
-                                                 torch_dtype=torch.float16,
+                                                 torch_dtype=torch.float8,
                                                  max_memory=max_memory)
     
     # Crear pipeline de generación de texto