Dunanwi commited on
Commit
750dfc3
·
verified ·
1 Parent(s): 22739aa

Update split_files/t5_tokenizer/tokenizer_config.json

Browse files
split_files/t5_tokenizer/tokenizer_config.json CHANGED
@@ -1,113 +1,113 @@
1
  {
2
  "backend": "tokenizers",
3
- "clean_up_tokenization_spaces": true,
4
  "eos_token": "</s>",
5
  "extra_ids": 100,
6
- "extra_special_tokens": {
7
- "extra_id_0": "<extra_id_0>",
8
- "extra_id_1": "<extra_id_1>",
9
- "extra_id_2": "<extra_id_2>",
10
- "extra_id_3": "<extra_id_3>",
11
- "extra_id_4": "<extra_id_4>",
12
- "extra_id_5": "<extra_id_5>",
13
- "extra_id_6": "<extra_id_6>",
14
- "extra_id_7": "<extra_id_7>",
15
- "extra_id_8": "<extra_id_8>",
16
- "extra_id_9": "<extra_id_9>",
17
- "extra_id_10": "<extra_id_10>",
18
- "extra_id_11": "<extra_id_11>",
19
- "extra_id_12": "<extra_id_12>",
20
- "extra_id_13": "<extra_id_13>",
21
- "extra_id_14": "<extra_id_14>",
22
- "extra_id_15": "<extra_id_15>",
23
- "extra_id_16": "<extra_id_16>",
24
- "extra_id_17": "<extra_id_17>",
25
- "extra_id_18": "<extra_id_18>",
26
- "extra_id_19": "<extra_id_19>",
27
- "extra_id_20": "<extra_id_20>",
28
- "extra_id_21": "<extra_id_21>",
29
- "extra_id_22": "<extra_id_22>",
30
- "extra_id_23": "<extra_id_23>",
31
- "extra_id_24": "<extra_id_24>",
32
- "extra_id_25": "<extra_id_25>",
33
- "extra_id_26": "<extra_id_26>",
34
- "extra_id_27": "<extra_id_27>",
35
- "extra_id_28": "<extra_id_28>",
36
- "extra_id_29": "<extra_id_29>",
37
- "extra_id_30": "<extra_id_30>",
38
- "extra_id_31": "<extra_id_31>",
39
- "extra_id_32": "<extra_id_32>",
40
- "extra_id_33": "<extra_id_33>",
41
- "extra_id_34": "<extra_id_34>",
42
- "extra_id_35": "<extra_id_35>",
43
- "extra_id_36": "<extra_id_36>",
44
- "extra_id_37": "<extra_id_37>",
45
- "extra_id_38": "<extra_id_38>",
46
- "extra_id_39": "<extra_id_39>",
47
- "extra_id_40": "<extra_id_40>",
48
- "extra_id_41": "<extra_id_41>",
49
- "extra_id_42": "<extra_id_42>",
50
- "extra_id_43": "<extra_id_43>",
51
- "extra_id_44": "<extra_id_44>",
52
- "extra_id_45": "<extra_id_45>",
53
- "extra_id_46": "<extra_id_46>",
54
- "extra_id_47": "<extra_id_47>",
55
- "extra_id_48": "<extra_id_48>",
56
- "extra_id_49": "<extra_id_49>",
57
- "extra_id_50": "<extra_id_50>",
58
- "extra_id_51": "<extra_id_51>",
59
- "extra_id_52": "<extra_id_52>",
60
- "extra_id_53": "<extra_id_53>",
61
- "extra_id_54": "<extra_id_54>",
62
- "extra_id_55": "<extra_id_55>",
63
- "extra_id_56": "<extra_id_56>",
64
- "extra_id_57": "<extra_id_57>",
65
- "extra_id_58": "<extra_id_58>",
66
- "extra_id_59": "<extra_id_59>",
67
- "extra_id_60": "<extra_id_60>",
68
- "extra_id_61": "<extra_id_61>",
69
- "extra_id_62": "<extra_id_62>",
70
- "extra_id_63": "<extra_id_63>",
71
- "extra_id_64": "<extra_id_64>",
72
- "extra_id_65": "<extra_id_65>",
73
- "extra_id_66": "<extra_id_66>",
74
- "extra_id_67": "<extra_id_67>",
75
- "extra_id_68": "<extra_id_68>",
76
- "extra_id_69": "<extra_id_69>",
77
- "extra_id_70": "<extra_id_70>",
78
- "extra_id_71": "<extra_id_71>",
79
- "extra_id_72": "<extra_id_72>",
80
- "extra_id_73": "<extra_id_73>",
81
- "extra_id_74": "<extra_id_74>",
82
- "extra_id_75": "<extra_id_75>",
83
- "extra_id_76": "<extra_id_76>",
84
- "extra_id_77": "<extra_id_77>",
85
- "extra_id_78": "<extra_id_78>",
86
- "extra_id_79": "<extra_id_79>",
87
- "extra_id_80": "<extra_id_80>",
88
- "extra_id_81": "<extra_id_81>",
89
- "extra_id_82": "<extra_id_82>",
90
- "extra_id_83": "<extra_id_83>",
91
- "extra_id_84": "<extra_id_84>",
92
- "extra_id_85": "<extra_id_85>",
93
- "extra_id_86": "<extra_id_86>",
94
- "extra_id_87": "<extra_id_87>",
95
- "extra_id_88": "<extra_id_88>",
96
- "extra_id_89": "<extra_id_89>",
97
- "extra_id_90": "<extra_id_90>",
98
- "extra_id_91": "<extra_id_91>",
99
- "extra_id_92": "<extra_id_92>",
100
- "extra_id_93": "<extra_id_93>",
101
- "extra_id_94": "<extra_id_94>",
102
- "extra_id_95": "<extra_id_95>",
103
- "extra_id_96": "<extra_id_96>",
104
- "extra_id_97": "<extra_id_97>",
105
- "extra_id_98": "<extra_id_98>",
106
- "extra_id_99": "<extra_id_99>"
107
- },
108
  "is_local": false,
109
  "model_max_length": 512,
110
  "pad_token": "<pad>",
 
111
  "tokenizer_class": "T5Tokenizer",
112
  "unk_token": "<unk>"
113
- }
 
1
  {
2
  "backend": "tokenizers",
 
3
  "eos_token": "</s>",
4
  "extra_ids": 100,
5
+ "extra_special_tokens": [
6
+ "<extra_id_0>",
7
+ "<extra_id_1>",
8
+ "<extra_id_2>",
9
+ "<extra_id_3>",
10
+ "<extra_id_4>",
11
+ "<extra_id_5>",
12
+ "<extra_id_6>",
13
+ "<extra_id_7>",
14
+ "<extra_id_8>",
15
+ "<extra_id_9>",
16
+ "<extra_id_10>",
17
+ "<extra_id_11>",
18
+ "<extra_id_12>",
19
+ "<extra_id_13>",
20
+ "<extra_id_14>",
21
+ "<extra_id_15>",
22
+ "<extra_id_16>",
23
+ "<extra_id_17>",
24
+ "<extra_id_18>",
25
+ "<extra_id_19>",
26
+ "<extra_id_20>",
27
+ "<extra_id_21>",
28
+ "<extra_id_22>",
29
+ "<extra_id_23>",
30
+ "<extra_id_24>",
31
+ "<extra_id_25>",
32
+ "<extra_id_26>",
33
+ "<extra_id_27>",
34
+ "<extra_id_28>",
35
+ "<extra_id_29>",
36
+ "<extra_id_30>",
37
+ "<extra_id_31>",
38
+ "<extra_id_32>",
39
+ "<extra_id_33>",
40
+ "<extra_id_34>",
41
+ "<extra_id_35>",
42
+ "<extra_id_36>",
43
+ "<extra_id_37>",
44
+ "<extra_id_38>",
45
+ "<extra_id_39>",
46
+ "<extra_id_40>",
47
+ "<extra_id_41>",
48
+ "<extra_id_42>",
49
+ "<extra_id_43>",
50
+ "<extra_id_44>",
51
+ "<extra_id_45>",
52
+ "<extra_id_46>",
53
+ "<extra_id_47>",
54
+ "<extra_id_48>",
55
+ "<extra_id_49>",
56
+ "<extra_id_50>",
57
+ "<extra_id_51>",
58
+ "<extra_id_52>",
59
+ "<extra_id_53>",
60
+ "<extra_id_54>",
61
+ "<extra_id_55>",
62
+ "<extra_id_56>",
63
+ "<extra_id_57>",
64
+ "<extra_id_58>",
65
+ "<extra_id_59>",
66
+ "<extra_id_60>",
67
+ "<extra_id_61>",
68
+ "<extra_id_62>",
69
+ "<extra_id_63>",
70
+ "<extra_id_64>",
71
+ "<extra_id_65>",
72
+ "<extra_id_66>",
73
+ "<extra_id_67>",
74
+ "<extra_id_68>",
75
+ "<extra_id_69>",
76
+ "<extra_id_70>",
77
+ "<extra_id_71>",
78
+ "<extra_id_72>",
79
+ "<extra_id_73>",
80
+ "<extra_id_74>",
81
+ "<extra_id_75>",
82
+ "<extra_id_76>",
83
+ "<extra_id_77>",
84
+ "<extra_id_78>",
85
+ "<extra_id_79>",
86
+ "<extra_id_80>",
87
+ "<extra_id_81>",
88
+ "<extra_id_82>",
89
+ "<extra_id_83>",
90
+ "<extra_id_84>",
91
+ "<extra_id_85>",
92
+ "<extra_id_86>",
93
+ "<extra_id_87>",
94
+ "<extra_id_88>",
95
+ "<extra_id_89>",
96
+ "<extra_id_90>",
97
+ "<extra_id_91>",
98
+ "<extra_id_92>",
99
+ "<extra_id_93>",
100
+ "<extra_id_94>",
101
+ "<extra_id_95>",
102
+ "<extra_id_96>",
103
+ "<extra_id_97>",
104
+ "<extra_id_98>",
105
+ "<extra_id_99>"
106
+ ],
107
  "is_local": false,
108
  "model_max_length": 512,
109
  "pad_token": "<pad>",
110
+ "sp_model_kwargs": {},
111
  "tokenizer_class": "T5Tokenizer",
112
  "unk_token": "<unk>"
113
+ }