vladislavbro commited on
Commit
f88dc87
·
verified ·
1 Parent(s): 7aaac7c

Update tokenizer.json

Browse files
Files changed (1) hide show
  1. tokenizer.json +28 -1
tokenizer.json CHANGED
@@ -449,7 +449,34 @@
449
  "pre_tokenizer": {
450
  "type": "Whitespace"
451
  },
452
- "post_processor": null,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
453
  "decoder": null,
454
  "model": {
455
  "type": "BPE",
 
449
  "pre_tokenizer": {
450
  "type": "Whitespace"
451
  },
452
+ "post_processor": {
453
+ "type": "TemplateProcessing",
454
+ "single": [
455
+ { "SpecialToken": { "id": "BOS", "type_id": 0 } },
456
+ { "Sequence": { "id": "A", "type_id": 0 } },
457
+ { "SpecialToken": { "id": "EOS", "type_id": 0 } }
458
+ ],
459
+ "pair": [
460
+ { "SpecialToken": { "id": "BOS", "type_id": 0 } },
461
+ { "Sequence": { "id": "A", "type_id": 0 } },
462
+ { "SpecialToken": { "id": "EOS", "type_id": 0 } },
463
+ { "SpecialToken": { "id": "BOS", "type_id": 1 } },
464
+ { "Sequence": { "id": "B", "type_id": 1 } },
465
+ { "SpecialToken": { "id": "EOS", "type_id": 1 } }
466
+ ],
467
+ "special_tokens": {
468
+ "BOS": {
469
+ "id": "BOS",
470
+ "ids": [255],
471
+ "tokens": ["<s>"]
472
+ },
473
+ "EOS": {
474
+ "id": "EOS",
475
+ "ids": [0],
476
+ "tokens": ["</s>"]
477
+ }
478
+ }
479
+ },
480
  "decoder": null,
481
  "model": {
482
  "type": "BPE",