Update app.py

app.py

@@ -62,9 +62,8 @@ def load_models():
     base_mod = AutoModelForCausalLM.from_pretrained(
         BASE_ID,
         token=HF_TOKEN,
-        device_map="auto",           # deja que Accelerate decida
-        quantization_config=quant_config,
-        offload_folder="/tmp/offload"   # ruta temporal de escritura
+        device_map="auto",           
+        offload_folder="/tmp/offload"
     )
     norm_tok = AutoTokenizer.from_pretrained(ADAPTER_ID, use_fast=True, token=HF_TOKEN)
     norm_mod = PeftModel.from_pretrained(