Update app.py

app.py

@@ -22,7 +22,7 @@ tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
 try:
     model = AutoModelForCausalLM.from_pretrained(
         MODEL_ID,
-        device_map="auto",                  # auto GPU/CPU placement
+        #device_map="auto",                  # auto GPU/CPU placement
         torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
         low_cpu_mem_usage=True
     )