tokenizer_config.json

{
  "tokenizer_class": "MorphemeTokenizer",
  "backbone": "xlm-roberta-base",
  "vocab_size": 61028,
  "max_length": 128,
  "min_freq_word": 2,
  "min_freq_morph": 3,
  "wechsel_k": 10,
  "wechsel_tau": 0.1,
  "bidict_source": "NIRVLab/rhade-vietnamese-mt",
  "bidict_anchor_weight": 0.3
}