floriangardin commited on
Commit
85648a9
1 Parent(s): dc8cdc4

Upload tokenizer

Browse files
Files changed (2) hide show
  1. tokenizer.json +10 -8
  2. tokenizer_config.json +3 -3
tokenizer.json CHANGED
@@ -58,7 +58,7 @@
58
  "special": true
59
  },
60
  {
61
- "id": 47,
62
  "content": "[END]",
63
  "single_word": false,
64
  "lstrip": false,
@@ -109,20 +109,22 @@
109
  "TONALITY_DEGREE__8": 30,
110
  "TONALITY_DEGREE__1": 31,
111
  "TONALITY_DEGREE__11": 32,
112
- "TONALITY_DEGREE__6": 33,
113
- "CHORD_EXTENSION__2": 34,
114
  "CHORD_EXTENSION__43": 35,
115
- "CHORD_EXTENSION__(sus2)": 36,
116
- "CHORD_EXTENSION__(sus4)": 37,
117
  "CHORD_EXTENSION__6(sus4)": 38,
118
  "CHORD_DEGREE__6": 39,
119
- "END": 40,
120
- "CHORD_EXTENSION__65": 41,
121
  "CHORD_EXTENSION__64(sus2)": 42,
122
  "CHORD_EXTENSION__64(sus4)": 43,
123
  "CHORD_EXTENSION__6(sus2)": 44,
124
  "CHORD_OCTAVE__-3": 45,
125
- "CHORD_EXTENSION__(+)": 46
 
 
126
  },
127
  "unk_token": "[UNK]"
128
  }
 
58
  "special": true
59
  },
60
  {
61
+ "id": 49,
62
  "content": "[END]",
63
  "single_word": false,
64
  "lstrip": false,
 
109
  "TONALITY_DEGREE__8": 30,
110
  "TONALITY_DEGREE__1": 31,
111
  "TONALITY_DEGREE__11": 32,
112
+ "CHORD_EXTENSION__2": 33,
113
+ "TONALITY_DEGREE__6": 34,
114
  "CHORD_EXTENSION__43": 35,
115
+ "CHORD_EXTENSION__(sus4)": 36,
116
+ "CHORD_EXTENSION__(sus2)": 37,
117
  "CHORD_EXTENSION__6(sus4)": 38,
118
  "CHORD_DEGREE__6": 39,
119
+ "CHORD_EXTENSION__65": 40,
120
+ "END": 41,
121
  "CHORD_EXTENSION__64(sus2)": 42,
122
  "CHORD_EXTENSION__64(sus4)": 43,
123
  "CHORD_EXTENSION__6(sus2)": 44,
124
  "CHORD_OCTAVE__-3": 45,
125
+ "CHORD_EXTENSION__(+)": 46,
126
+ "CHORD_EXTENSION__64(+)": 47,
127
+ "CHORD_EXTENSION__6(+)": 48
128
  },
129
  "unk_token": "[UNK]"
130
  }
tokenizer_config.json CHANGED
@@ -49,7 +49,7 @@
49
  "single_word": false,
50
  "special": true
51
  },
52
- "47": {
53
  "content": "[END]",
54
  "lstrip": false,
55
  "normalized": false,
@@ -64,11 +64,11 @@
64
  "cls_token": "[CLS]",
65
  "cls_token_id": 1,
66
  "eos_token": "[END]",
67
- "eos_token_id": 47,
68
  "mask_token": "[MASK]",
69
  "mask_token_id": 4,
70
  "model_max_len": 4096,
71
- "model_max_length": 256,
72
  "pad_token": "[PAD]",
73
  "pad_token_id": 3,
74
  "seo_token_id": 2,
 
49
  "single_word": false,
50
  "special": true
51
  },
52
+ "49": {
53
  "content": "[END]",
54
  "lstrip": false,
55
  "normalized": false,
 
64
  "cls_token": "[CLS]",
65
  "cls_token_id": 1,
66
  "eos_token": "[END]",
67
+ "eos_token_id": 49,
68
  "mask_token": "[MASK]",
69
  "mask_token_id": 4,
70
  "model_max_len": 4096,
71
+ "model_max_length": 512,
72
  "pad_token": "[PAD]",
73
  "pad_token_id": 3,
74
  "seo_token_id": 2,