tokenizer_config.json

{
  "bos_token": "<s>",
  "cls_token": "<s>",
  "eos_token": "</s>",
  "lower": true,
  "n_sentinel_tokens": 512,
  "name_or_path": "/data/user_data/peixuanh/ckpts/ClueWeb22_All/DRO_emb_web_graph_url_RE",
  "pad_token": "<pad>",
  "sep_token": "</s>",
  "special_tokens_map_file": "/data/user_data/peixuanh/models/web_continuous_t5_ckpt/special_tokens_map.json",
  "tokenizer_class": "FairseqT5Tokenizer",
  "unk_token": "<unk>"
}