Update app.py

app.py

@@ -14,16 +14,15 @@ MODEL_ID = "Reem1/hakayti-merged"
 
 bnb_cfg = BitsAndBytesConfig(
     load_in_4bit=True,
-    bnb_4bit_compute_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
+    bnb_4bit_compute_dtype=torch.float32,  
     bnb_4bit_use_double_quant=True,
     bnb_4bit_quant_type="nf4",
 )
-
 print(" Loading model...")
 tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
 model = AutoModelForCausalLM.from_pretrained(
     MODEL_ID,
-    device_map="auto",
+    device_map={"": "cpu"},   
     quantization_config=bnb_cfg
 )
 pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)