TristanBehrens
commited on
Commit
•
33ea777
1
Parent(s):
0ec7c58
Upload tokenizer.json with huggingface_hub
Browse files- tokenizer.json +224 -0
tokenizer.json
ADDED
@@ -0,0 +1,224 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"version": "1.0",
|
3 |
+
"truncation": null,
|
4 |
+
"padding": null,
|
5 |
+
"added_tokens": [
|
6 |
+
{
|
7 |
+
"id": 0,
|
8 |
+
"content": "[UNK]",
|
9 |
+
"single_word": false,
|
10 |
+
"lstrip": false,
|
11 |
+
"rstrip": false,
|
12 |
+
"normalized": false,
|
13 |
+
"special": true
|
14 |
+
},
|
15 |
+
{
|
16 |
+
"id": 1,
|
17 |
+
"content": "[PAD]",
|
18 |
+
"single_word": false,
|
19 |
+
"lstrip": false,
|
20 |
+
"rstrip": false,
|
21 |
+
"normalized": false,
|
22 |
+
"special": true
|
23 |
+
},
|
24 |
+
{
|
25 |
+
"id": 2,
|
26 |
+
"content": "[EOS]",
|
27 |
+
"single_word": false,
|
28 |
+
"lstrip": false,
|
29 |
+
"rstrip": false,
|
30 |
+
"normalized": false,
|
31 |
+
"special": true
|
32 |
+
}
|
33 |
+
],
|
34 |
+
"normalizer": null,
|
35 |
+
"pre_tokenizer": {
|
36 |
+
"type": "WhitespaceSplit"
|
37 |
+
},
|
38 |
+
"post_processor": null,
|
39 |
+
"decoder": null,
|
40 |
+
"model": {
|
41 |
+
"type": "WordLevel",
|
42 |
+
"vocab": {
|
43 |
+
"[UNK]": 0,
|
44 |
+
"[PAD]": 1,
|
45 |
+
"[EOS]": 2,
|
46 |
+
"TIME_DELTA=8": 3,
|
47 |
+
"TIME_DELTA=4": 4,
|
48 |
+
"TIME_DELTA=1": 5,
|
49 |
+
"BAR_END": 6,
|
50 |
+
"BAR_START": 7,
|
51 |
+
"TIME_DELTA=2": 8,
|
52 |
+
"TIME_DELTA=3": 9,
|
53 |
+
"NEXT": 10,
|
54 |
+
"NOTE_OFF=62": 11,
|
55 |
+
"NOTE_ON=62": 12,
|
56 |
+
"NOTE_OFF=67": 13,
|
57 |
+
"NOTE_ON=67": 14,
|
58 |
+
"DENSITY=4": 15,
|
59 |
+
"NOTE_OFF=69": 16,
|
60 |
+
"NOTE_ON=69": 17,
|
61 |
+
"NOTE_OFF=64": 18,
|
62 |
+
"NOTE_ON=64": 19,
|
63 |
+
"DENSITY=3": 20,
|
64 |
+
"NOTE_OFF=57": 21,
|
65 |
+
"NOTE_ON=57": 22,
|
66 |
+
"NOTE_OFF=55": 23,
|
67 |
+
"NOTE_ON=55": 24,
|
68 |
+
"NOTE_OFF=65": 25,
|
69 |
+
"NOTE_ON=65": 26,
|
70 |
+
"DENSITY=5": 27,
|
71 |
+
"NOTE_OFF=60": 28,
|
72 |
+
"NOTE_ON=60": 29,
|
73 |
+
"NOTE_OFF=59": 30,
|
74 |
+
"NOTE_ON=59": 31,
|
75 |
+
"NOTE_OFF=66": 32,
|
76 |
+
"NOTE_ON=66": 33,
|
77 |
+
"NOTE_OFF=71": 34,
|
78 |
+
"NOTE_ON=71": 35,
|
79 |
+
"NOTE_OFF=72": 36,
|
80 |
+
"NOTE_ON=72": 37,
|
81 |
+
"NOTE_OFF=50": 38,
|
82 |
+
"NOTE_ON=50": 39,
|
83 |
+
"NOTE_OFF=74": 40,
|
84 |
+
"NOTE_ON=74": 41,
|
85 |
+
"NOTE_OFF=52": 42,
|
86 |
+
"NOTE_ON=52": 43,
|
87 |
+
"TIME_DELTA=16": 44,
|
88 |
+
"NOTE_OFF=70": 45,
|
89 |
+
"NOTE_ON=70": 46,
|
90 |
+
"NOTE_OFF=58": 47,
|
91 |
+
"NOTE_ON=58": 48,
|
92 |
+
"NOTE_OFF=53": 49,
|
93 |
+
"NOTE_ON=53": 50,
|
94 |
+
"NOTE_OFF=63": 51,
|
95 |
+
"NOTE_ON=63": 52,
|
96 |
+
"NOTE_OFF=61": 53,
|
97 |
+
"NOTE_ON=61": 54,
|
98 |
+
"NOTE_OFF=68": 55,
|
99 |
+
"NOTE_ON=68": 56,
|
100 |
+
"NOTE_OFF=54": 57,
|
101 |
+
"NOTE_ON=54": 58,
|
102 |
+
"NOTE_OFF=48": 59,
|
103 |
+
"NOTE_ON=48": 60,
|
104 |
+
"NOTE_OFF=56": 61,
|
105 |
+
"NOTE_ON=56": 62,
|
106 |
+
"NOTE_OFF=76": 63,
|
107 |
+
"NOTE_ON=76": 64,
|
108 |
+
"NOTE_OFF=73": 65,
|
109 |
+
"NOTE_ON=73": 66,
|
110 |
+
"DENSITY=2": 67,
|
111 |
+
"DENSITY=1": 68,
|
112 |
+
"NOTE_OFF=51": 69,
|
113 |
+
"NOTE_ON=51": 70,
|
114 |
+
"NOTE_OFF=45": 71,
|
115 |
+
"NOTE_ON=45": 72,
|
116 |
+
"NOTE_OFF=47": 73,
|
117 |
+
"NOTE_ON=47": 74,
|
118 |
+
"NOTE_OFF=49": 75,
|
119 |
+
"NOTE_ON=49": 76,
|
120 |
+
"NOTE_OFF=75": 77,
|
121 |
+
"NOTE_ON=75": 78,
|
122 |
+
"NOTE_OFF=77": 79,
|
123 |
+
"NOTE_ON=77": 80,
|
124 |
+
"NOTE_OFF=46": 81,
|
125 |
+
"NOTE_ON=46": 82,
|
126 |
+
"NOTE_OFF=43": 83,
|
127 |
+
"NOTE_ON=43": 84,
|
128 |
+
"TIME_DELTA=5": 85,
|
129 |
+
"NOTE_OFF=79": 86,
|
130 |
+
"NOTE_ON=79": 87,
|
131 |
+
"TIME_DELTA=6": 88,
|
132 |
+
"TIME_DELTA=12": 89,
|
133 |
+
"TIME_DELTA=7": 90,
|
134 |
+
"NOTE_OFF=78": 91,
|
135 |
+
"NOTE_ON=78": 92,
|
136 |
+
"DENSITY=0": 93,
|
137 |
+
"TIME_DELTA=24": 94,
|
138 |
+
"NOTE_OFF=81": 95,
|
139 |
+
"NOTE_ON=81": 96,
|
140 |
+
"NOTE_OFF=44": 97,
|
141 |
+
"NOTE_ON=44": 98,
|
142 |
+
"NOTE_OFF=41": 99,
|
143 |
+
"NOTE_ON=41": 100,
|
144 |
+
"NOTE_OFF=42": 101,
|
145 |
+
"NOTE_ON=42": 102,
|
146 |
+
"NOTE_OFF=40": 103,
|
147 |
+
"NOTE_ON=40": 104,
|
148 |
+
"NOTE_OFF=38": 105,
|
149 |
+
"NOTE_ON=38": 106,
|
150 |
+
"TIME_DELTA=32": 107,
|
151 |
+
"TIME_DELTA=10": 108,
|
152 |
+
"NOTE_OFF=80": 109,
|
153 |
+
"NOTE_ON=80": 110,
|
154 |
+
"GARLAND_END": 111,
|
155 |
+
"GARLAND_START": 112,
|
156 |
+
"TIME_DELTA=9": 113,
|
157 |
+
"TIME_DELTA=11": 114,
|
158 |
+
"NOTE_OFF=82": 115,
|
159 |
+
"NOTE_ON=82": 116,
|
160 |
+
"TIME_DELTA=15": 117,
|
161 |
+
"NOTE_OFF=39": 118,
|
162 |
+
"NOTE_ON=39": 119,
|
163 |
+
"NOTE_OFF=36": 120,
|
164 |
+
"NOTE_ON=36": 121,
|
165 |
+
"NOTE_OFF=83": 122,
|
166 |
+
"NOTE_ON=83": 123,
|
167 |
+
"TIME_DELTA=14": 124,
|
168 |
+
"TIME_DELTA=20": 125,
|
169 |
+
"TIME_DELTA=13": 126,
|
170 |
+
"NOTE_OFF=84": 127,
|
171 |
+
"NOTE_ON=84": 128,
|
172 |
+
"NOTE_OFF=37": 129,
|
173 |
+
"NOTE_ON=37": 130,
|
174 |
+
"TIME_DELTA=19": 131,
|
175 |
+
"NOTE_OFF=33": 132,
|
176 |
+
"NOTE_ON=33": 133,
|
177 |
+
"TIME_DELTA=18": 134,
|
178 |
+
"TIME_DELTA=17": 135,
|
179 |
+
"NOTE_OFF=35": 136,
|
180 |
+
"NOTE_ON=35": 137,
|
181 |
+
"NOTE_OFF=31": 138,
|
182 |
+
"NOTE_ON=31": 139,
|
183 |
+
"NOTE_OFF=34": 140,
|
184 |
+
"NOTE_ON=34": 141,
|
185 |
+
"TIME_DELTA=23": 142,
|
186 |
+
"TIME_DELTA=21": 143,
|
187 |
+
"TIME_DELTA=22": 144,
|
188 |
+
"NOTE_OFF=86": 145,
|
189 |
+
"NOTE_ON=86": 146,
|
190 |
+
"TIME_DELTA=31": 147,
|
191 |
+
"TIME_DELTA=30": 148,
|
192 |
+
"TIME_DELTA=28": 149,
|
193 |
+
"NOTE_OFF=85": 150,
|
194 |
+
"NOTE_ON=85": 151,
|
195 |
+
"TIME_DELTA=25": 152,
|
196 |
+
"TIME_DELTA=26": 153,
|
197 |
+
"NOTE_OFF=30": 154,
|
198 |
+
"NOTE_ON=30": 155,
|
199 |
+
"NOTE_OFF=26": 156,
|
200 |
+
"NOTE_ON=26": 157,
|
201 |
+
"NOTE_OFF=29": 158,
|
202 |
+
"NOTE_ON=29": 159,
|
203 |
+
"NOTE_OFF=28": 160,
|
204 |
+
"NOTE_ON=28": 161,
|
205 |
+
"TIME_DELTA=29": 162,
|
206 |
+
"TIME_DELTA=27": 163,
|
207 |
+
"NOTE_OFF=32": 164,
|
208 |
+
"NOTE_ON=32": 165,
|
209 |
+
"NOTE_OFF=27": 166,
|
210 |
+
"NOTE_ON=27": 167,
|
211 |
+
"NOTE_OFF=88": 168,
|
212 |
+
"NOTE_ON=88": 169,
|
213 |
+
"NOTE_OFF=24": 170,
|
214 |
+
"NOTE_ON=24": 171,
|
215 |
+
"NOTE_OFF=87": 172,
|
216 |
+
"NOTE_ON=87": 173,
|
217 |
+
"NOTE_OFF=90": 174,
|
218 |
+
"NOTE_ON=90": 175,
|
219 |
+
"NOTE_OFF=91": 176,
|
220 |
+
"NOTE_ON=91": 177
|
221 |
+
},
|
222 |
+
"unk_token": "[UNK]"
|
223 |
+
}
|
224 |
+
}
|