Update README.md

README.md

@@ -128,7 +128,7 @@ vLLM aslo supports OpenAI-compatible serving. See the [documentation](https://do
   )
   
   # Save to disk in compressed-tensors format
-  save_path = model_name + "-quantized.w4a16
+  save_path = model_name + "-quantized.w4a16"
   model.save_pretrained(save_path)
   tokenizer.save_pretrained(save_path)
   print(f"Model and tokenizer saved to: {save_path}")