use_cache to true for better inference

config.json

@@ -64,6 +64,6 @@
   "repetition_penalty": 3.5,
   "torch_dtype": "float32",
   "transformers_version": "4.27.4",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 50265
 }