set generation max_length from 4096 to 2048

generation_config.json

@@ -4,7 +4,7 @@
   "eos_token_id": 2,
   "pad_token_id": 0,
   "temperature": 0.6,
-  "max_length": 4096,
+  "max_length": 2048,
   "top_p": 0.9,
   "transformers_version": "4.31.0.dev0"
 }