Adjust max length in tokenizer config

Update tokenizer configuration to align max length with `config.json`. This change does not affect the tokenizer's output. It only prevents (incorrect) warnings for sequences exceeding 1024 tokens.

tokenizer_config.json

@@ -46,8 +46,8 @@
   "do_basic_tokenize": true,
   "do_lower_case": false,
   "mask_token": "[MASK]",
-  "max_len": 1024,
-  "model_max_length": 1024,
+  "max_len": 8192,
+  "model_max_length": 8192,
   "never_split": null,
   "pad_token": "[PAD]",
   "sep_token": "[SEP]",