Fix tokenizer: add <unk> to BPE vocab

tokenizer_config.json

@@ -32776,5 +32776,5 @@
   "pad_token": "[PAD]",
   "sep_token": "[SEP]",
   "tokenizer_class": "PreTrainedTokenizerFast",
-  "unk_token": "[UNK]"
-}
+  "unk_token": "<unk>"
+}