The tokenizer adds a special token '<|im_end|>' to solve the problem of non-stop generation when encountering <|im_end|>.

tokenizer.json

@@ -2306,6 +2306,15 @@
       "rstrip": false,
       "normalized": false,
       "special": true
+    },
+    {
+      "id": 128256,
+      "content": "<|im_end|>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": false,
+      "special": true
     }
   ],
   "normalizer": null,