abhipn commited on
Commit
70ff363
1 Parent(s): db9c7ee

Update tokenizer_config.json

Browse files

Configuring bos, eos, pad and unk tokens helps load the tokenizer successfully, otherwise getting a recursion error "RecursionError: maximum recursion depth exceeded while getting the str of an object."

Files changed (1) hide show
  1. tokenizer_config.json +5 -4
tokenizer_config.json CHANGED
@@ -1,9 +1,10 @@
1
  {
2
- "bos_token": "",
3
- "eos_token": "",
4
  "model_max_length": 512,
5
  "padding_side": "right",
6
  "special_tokens_map_file": "/sc-projects/sc-proj-cc06-medbert/hfcache/hub/models--decapoda-research--llama-13b-hf/snapshots/438770a656712a5072229b62256521845d4de5ce/special_tokens_map.json",
7
  "tokenizer_class": "LlamaTokenizer",
8
- "unk_token": ""
9
- }
 
 
1
  {
2
+ "bos_token": "</s>",
3
+ "eos_token": "</s>",
4
  "model_max_length": 512,
5
  "padding_side": "right",
6
  "special_tokens_map_file": "/sc-projects/sc-proj-cc06-medbert/hfcache/hub/models--decapoda-research--llama-13b-hf/snapshots/438770a656712a5072229b62256521845d4de5ce/special_tokens_map.json",
7
  "tokenizer_class": "LlamaTokenizer",
8
+ "unk_token": "</s>",
9
+ "pad_token": "[PAD]"
10
+ }