KhimNguyen commited on
Commit
bf729ba
1 Parent(s): 1883028

Upload tokenizer

Browse files
Files changed (1) hide show
  1. tokenizer_config.json +4 -0
tokenizer_config.json CHANGED
@@ -938,9 +938,13 @@
938
  "clean_up_tokenization_spaces": true,
939
  "eos_token": "</s>",
940
  "extra_ids": 100,
 
941
  "model_max_length": 512,
942
  "pad_token": "<pad>",
943
  "sp_model_kwargs": {},
 
944
  "tokenizer_class": "T5Tokenizer",
 
 
945
  "unk_token": "<unk>"
946
  }
 
938
  "clean_up_tokenization_spaces": true,
939
  "eos_token": "</s>",
940
  "extra_ids": 100,
941
+ "max_length": 512,
942
  "model_max_length": 512,
943
  "pad_token": "<pad>",
944
  "sp_model_kwargs": {},
945
+ "stride": 0,
946
  "tokenizer_class": "T5Tokenizer",
947
+ "truncation_side": "right",
948
+ "truncation_strategy": "longest_first",
949
  "unk_token": "<unk>"
950
  }