changes made to app.py

app.py

@@ -23,9 +23,12 @@ model.to(device)
 # ===============================
 # Load Evaluation Model (QwQ)
 # ===============================
+# Set 4-bit quantization configuration
 bnb_config = BitsAndBytesConfig(
-    load_in_8bit=True,
-    llm_int8_enable_fp32_cpu_offload=True,
+    load_in_4bit=True,  
+    bnb_4bit_compute_dtype=torch.bfloat16,  
+    bnb_4bit_use_double_quant=True,  
+    bnb_4bit_quant_type="nf4"
 )
 
 qwq_model_id = "unsloth/QwQ-32B-unsloth-bnb-4bit"