traduttoremontatese commited on
Commit
b446df6
Β·
verified Β·
1 Parent(s): a3b82c1

Fix alignment + add spiece.model + special_tokens_map.json (mt5-small)

Browse files
config.json CHANGED
@@ -27,7 +27,7 @@
27
  "relative_attention_num_buckets": 32,
28
  "tie_word_embeddings": true,
29
  "tokenizer_class": "T5Tokenizer",
30
- "transformers_version": "5.0.0",
31
  "use_cache": false,
32
- "vocab_size": 250112
33
  }
 
27
  "relative_attention_num_buckets": 32,
28
  "tie_word_embeddings": true,
29
  "tokenizer_class": "T5Tokenizer",
30
+ "transformers_version": "5.1.0",
31
  "use_cache": false,
32
+ "vocab_size": 250100
33
  }
generation_config.json CHANGED
@@ -5,5 +5,5 @@
5
  1
6
  ],
7
  "pad_token_id": 0,
8
- "transformers_version": "5.0.0"
9
  }
 
5
  1
6
  ],
7
  "pad_token_id": 0,
8
+ "transformers_version": "5.1.0"
9
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b500e692799b81d18683501342929cf2f37ba9ab84ba85215b97dc8e3e2e4b15
3
- size 1200729512
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f37ee919d39ece224ff6f917bedcf79859d91e8ebd106c5742d577f57b5e844d
3
+ size 688475472
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "additional_special_tokens": []}
tokenizer_config.json CHANGED
@@ -1,8 +1,115 @@
1
  {
2
  "backend": "tokenizers",
3
  "eos_token": "</s>",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  "pad_token": "<pad>",
5
- "unk_token": "<unk>",
6
- "tokenizer_class": "T5TokenizerFast",
7
- "model_max_length": 512
8
  }
 
1
  {
2
  "backend": "tokenizers",
3
  "eos_token": "</s>",
4
+ "extra_ids": 0,
5
+ "extra_special_tokens": [
6
+ "<pad>",
7
+ "</s>",
8
+ "▁<extra_id_99>",
9
+ "▁<extra_id_98>",
10
+ "▁<extra_id_97>",
11
+ "▁<extra_id_96>",
12
+ "▁<extra_id_95>",
13
+ "▁<extra_id_94>",
14
+ "▁<extra_id_93>",
15
+ "▁<extra_id_92>",
16
+ "▁<extra_id_91>",
17
+ "▁<extra_id_90>",
18
+ "▁<extra_id_89>",
19
+ "▁<extra_id_88>",
20
+ "▁<extra_id_87>",
21
+ "▁<extra_id_86>",
22
+ "▁<extra_id_85>",
23
+ "▁<extra_id_84>",
24
+ "▁<extra_id_83>",
25
+ "▁<extra_id_82>",
26
+ "▁<extra_id_81>",
27
+ "▁<extra_id_80>",
28
+ "▁<extra_id_79>",
29
+ "▁<extra_id_78>",
30
+ "▁<extra_id_77>",
31
+ "▁<extra_id_76>",
32
+ "▁<extra_id_75>",
33
+ "▁<extra_id_74>",
34
+ "▁<extra_id_73>",
35
+ "▁<extra_id_72>",
36
+ "▁<extra_id_71>",
37
+ "▁<extra_id_70>",
38
+ "▁<extra_id_69>",
39
+ "▁<extra_id_68>",
40
+ "▁<extra_id_67>",
41
+ "▁<extra_id_66>",
42
+ "▁<extra_id_65>",
43
+ "▁<extra_id_64>",
44
+ "▁<extra_id_63>",
45
+ "▁<extra_id_62>",
46
+ "▁<extra_id_61>",
47
+ "▁<extra_id_60>",
48
+ "▁<extra_id_59>",
49
+ "▁<extra_id_58>",
50
+ "▁<extra_id_57>",
51
+ "▁<extra_id_56>",
52
+ "▁<extra_id_55>",
53
+ "▁<extra_id_54>",
54
+ "▁<extra_id_53>",
55
+ "▁<extra_id_52>",
56
+ "▁<extra_id_51>",
57
+ "▁<extra_id_50>",
58
+ "▁<extra_id_49>",
59
+ "▁<extra_id_48>",
60
+ "▁<extra_id_47>",
61
+ "▁<extra_id_46>",
62
+ "▁<extra_id_45>",
63
+ "▁<extra_id_44>",
64
+ "▁<extra_id_43>",
65
+ "▁<extra_id_42>",
66
+ "▁<extra_id_41>",
67
+ "▁<extra_id_40>",
68
+ "▁<extra_id_39>",
69
+ "▁<extra_id_38>",
70
+ "▁<extra_id_37>",
71
+ "▁<extra_id_36>",
72
+ "▁<extra_id_35>",
73
+ "▁<extra_id_34>",
74
+ "▁<extra_id_33>",
75
+ "▁<extra_id_32>",
76
+ "▁<extra_id_31>",
77
+ "▁<extra_id_30>",
78
+ "▁<extra_id_29>",
79
+ "▁<extra_id_28>",
80
+ "▁<extra_id_27>",
81
+ "▁<extra_id_26>",
82
+ "▁<extra_id_25>",
83
+ "▁<extra_id_24>",
84
+ "▁<extra_id_23>",
85
+ "▁<extra_id_22>",
86
+ "▁<extra_id_21>",
87
+ "▁<extra_id_20>",
88
+ "▁<extra_id_19>",
89
+ "▁<extra_id_18>",
90
+ "▁<extra_id_17>",
91
+ "▁<extra_id_16>",
92
+ "▁<extra_id_15>",
93
+ "▁<extra_id_14>",
94
+ "▁<extra_id_13>",
95
+ "▁<extra_id_12>",
96
+ "▁<extra_id_11>",
97
+ "▁<extra_id_10>",
98
+ "▁<extra_id_9>",
99
+ "▁<extra_id_8>",
100
+ "▁<extra_id_7>",
101
+ "▁<extra_id_6>",
102
+ "▁<extra_id_5>",
103
+ "▁<extra_id_4>",
104
+ "▁<extra_id_3>",
105
+ "▁<extra_id_2>",
106
+ "▁<extra_id_1>",
107
+ "▁<extra_id_0>"
108
+ ],
109
+ "is_local": false,
110
+ "model_max_length": 1000000000000000019884624838656,
111
  "pad_token": "<pad>",
112
+ "tokenizer_class": "T5Tokenizer",
113
+ "unk_id": 2,
114
+ "unk_token": "<unk>"
115
  }