update another example

examples/quantization/basic_usage_wikitext2.py

@@ -137,6 +137,7 @@ def main():
     quantize_config = BaseQuantizeConfig(
         bits=4,  # quantize model to 4-bit
         group_size=128,  # it is recommended to set the value to 128
+        desc_act=False,  # desc_act and group size only works on triton
     )
 
     # load un-quantized model, the model will always be force loaded into cpu