sandernotenbaert commited on
Commit
69197af
·
verified ·
1 Parent(s): f2bdb0a

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. special_tokens_map.json +1 -0
  2. tokenizer.json +806 -0
  3. tokenizer_config.json +62 -0
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
tokenizer.json ADDED
@@ -0,0 +1,806 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [],
6
+ "normalizer": null,
7
+ "pre_tokenizer": {
8
+ "Type": "WhitespaceSplit"
9
+ },
10
+ "post_processor": null,
11
+ "decoder": null,
12
+ "model": {
13
+ "type": "WordLevel",
14
+ "vocab": {
15
+ "PAD": 0,
16
+ "UNK__": 1,
17
+ "UNK": 2,
18
+ "MASK": 3,
19
+ "BOS": 4,
20
+ "EOS": 5,
21
+ "GENRE__": 6,
22
+ "SUBGENRE__": 7,
23
+ "GENRE_TAGS__": 8,
24
+ "INSTR_COUNT__": 9,
25
+ "MELODY_INSTR__": 10,
26
+ "CHORDS_INSTR__": 11,
27
+ "BASS_INSTR__": 12,
28
+ "DRUMS_INSTR__": 13,
29
+ "START": 14,
30
+ "CHORD_CHANGE": 15,
31
+ "CHORD_DEGREE__": 16,
32
+ "TONALITY_DEGREE__": 17,
33
+ "TONALITY_MODE__": 18,
34
+ "CHORD_OCTAVE__": 19,
35
+ "CHORD_EXTENSION__": 20,
36
+ "CHORD_DURATION_NUM__": 21,
37
+ "CHORD_DURATION_DEN__": 22,
38
+ "NEXT_CHORD_DEGREE__": 23,
39
+ "NEXT_TONALITY_DEGREE__": 24,
40
+ "NEXT_TONALITY_MODE__": 25,
41
+ "NEXT_CHORD_OCTAVE__": 26,
42
+ "NEXT_CHORD_DURATION_NUM__": 27,
43
+ "NEXT_CHORD_DURATION_DEN__": 28,
44
+ "NEXT_CHORD_EXTENSION__": 29,
45
+ "INSTRUMENT_NAME__": 30,
46
+ "DENSITY__": 31,
47
+ "AVERAGE_OCTAVE__": 32,
48
+ "AMPLITUDE__": 33,
49
+ "NOTE_TYPE__": 34,
50
+ "NOTE_DURATION_NUM__": 35,
51
+ "NOTE_DURATION_DEN__": 36,
52
+ "NOTE_VAL__": 37,
53
+ "NOTE_OCTAVE__": 38,
54
+ "NOTE_AMP__": 39,
55
+ "MELODY_END": 40,
56
+ "WILL_END": 41,
57
+ "END": 42,
58
+ "POP": 43,
59
+ "ELECTRONIC": 44,
60
+ "EASY_LISTENING": 45,
61
+ "ROCK": 46,
62
+ "JAZZ": 47,
63
+ "NEW_AGE": 48,
64
+ "R&B": 49,
65
+ "BLUES": 50,
66
+ "COUNTRY": 51,
67
+ "UNKNOWN": 52,
68
+ "FOLK/ACOUSTIC": 53,
69
+ "WORLD/TRADITIONAL": 54,
70
+ "CLASSICAL": 55,
71
+ "METAL": 56,
72
+ "LATIN": 57,
73
+ "HIP_HOP": 58,
74
+ "GERMAN_POP": 59,
75
+ "SOUNDTRACK": 61,
76
+ "POP_ROCK": 62,
77
+ "PROGRESSIVE_TRANCE": 63,
78
+ "CCM": 64,
79
+ "DANCE_ROCK": 65,
80
+ "NEW_WAVE": 66,
81
+ "RAGTIME": 67,
82
+ "HARD_TRANCE": 68,
83
+ "COUNTRY_ROCK": 69,
84
+ "MEDITATION": 70,
85
+ "EUROPOP": 71,
86
+ "TEEN_POP": 72,
87
+ "DISCO": 73,
88
+ "VOCAL_JAZZ": 74,
89
+ "TEXAS_BLUES": 75,
90
+ "PROGRESSIVE_BLUEGRASS": 76,
91
+ "POST_ROCK": 78,
92
+ "SPACE_ROCK": 79,
93
+ "BLUES-ROCK": 80,
94
+ "SINGER-SONGWRITER": 81,
95
+ "TEX_MEX": 82,
96
+ "ITALIAN_DISCO": 83,
97
+ "NEOCLASSICAL": 84,
98
+ "DEEP_HOUSE": 85,
99
+ "NEO-PROGRESSIVE": 86,
100
+ "SOFT_ROCK": 87,
101
+ "DOO-WOP": 88,
102
+ "NEO_SOUL": 89,
103
+ "CELTIC": 90,
104
+ "FUNK": 91,
105
+ "POP_FOLK": 92,
106
+ "BOSSA_NOVA": 94,
107
+ "DANCE_POP": 95,
108
+ "TECH_HOUSE": 98,
109
+ "POST-GRUNGE": 100,
110
+ "HAPPY_HARDCORE": 101,
111
+ "HARD_ROCK": 102,
112
+ "CHICAGO_SOUL": 103,
113
+ "POP_PUNK": 104,
114
+ "HONKY_TONK": 105,
115
+ "CHANSON": 106,
116
+ "BRAZILIAN_POP_MUSIC": 107,
117
+ "J_POP": 108,
118
+ "GLAM_ROCK": 109,
119
+ "HEAVY_METAL": 111,
120
+ "CLASSIC_ROCK": 112,
121
+ "PROGRESSIVE_ROCK": 113,
122
+ "SADCORE": 114,
123
+ "MERSEYBEAT": 115,
124
+ "SMOOTH_JAZZ": 116,
125
+ "REQUIEM": 117,
126
+ "INSTRUMENTAL_POP": 118,
127
+ "ROCK_'N_ROLL": 119,
128
+ "BALLAD": 120,
129
+ "SHOW_TUNES": 121,
130
+ "DOWNTEMPO": 122,
131
+ "INDUSTRIAL_METAL": 123,
132
+ "EARLY_MUSIC": 124,
133
+ "FUTURE_JAZZ": 125,
134
+ "CHAMBER_JAZZ": 126,
135
+ "PSYCHEDELIC_ROCK": 127,
136
+ "TEJANO": 128,
137
+ "SOUTHERN_ROCK": 129,
138
+ "REGGAETON": 130,
139
+ "TRANCE": 131,
140
+ "COUNTRY_GOSPEL": 132,
141
+ "ALTERNATIVE_DANCE": 133,
142
+ "ALTERNATIVE_METAL": 134,
143
+ "CELTIC_FUSION": 135,
144
+ "BRILL_BUILDING_POP": 136,
145
+ "CELTIC_ROCK": 137,
146
+ "NEW_BEAT": 138,
147
+ "BLACK_METAL": 139,
148
+ "SHOEGAZE": 140,
149
+ "PROGRESSIVE_HOUSE": 141,
150
+ "BRUTAL_DEATH_METAL": 142,
151
+ "INSTRUMENTAL_ROCK": 143,
152
+ "LOS_ANGELES": 144,
153
+ "VOCAL_HOUSE": 145,
154
+ "NEW_ROMANTIC": 146,
155
+ "MOTOWN": 147,
156
+ "TRIBAL_HOUSE": 148,
157
+ "JUNGLE_MUSIC": 149,
158
+ "FLAMENCO": 151,
159
+ "SOUTHERN_SOUL": 152,
160
+ "POP_RAP": 153,
161
+ "FEMALE_VOCALIST": 154,
162
+ "OPERA": 155,
163
+ "MELODIC_TRANCE": 156,
164
+ "EURODANCE": 157,
165
+ "CUMBIA": 158,
166
+ "EMO": 159,
167
+ "JAZZ_FUSION": 161,
168
+ "MODERN_ROCK": 162,
169
+ "ORCHESTRA": 163,
170
+ "CHILL-OUT": 165,
171
+ "FREE_IMPROVISATION": 167,
172
+ "BRITISH_POP": 168,
173
+ "QUIET_STORM": 169,
174
+ "EXPERIMENTAL_ROCK": 170,
175
+ "INDIE_ROCK": 171,
176
+ "TECHNO": 172,
177
+ "AMERICANA": 173,
178
+ "DIRTY_SOUTH_RAP": 174,
179
+ "PROGRESSIVE_METAL": 175,
180
+ "SINGER": 176,
181
+ "ELECTRO": 177,
182
+ "ART_ROCK": 178,
183
+ "OUTLAW_COUNTRY": 179,
184
+ "MERENGUE": 180,
185
+ "SKA": 181,
186
+ "FOLK_ROCK": 182,
187
+ "POWER_METAL": 183,
188
+ "COUNTRY_BLUES": 184,
189
+ "FOLK-POP": 185,
190
+ "UK_GARAGE": 186,
191
+ "DANCEHALL": 187,
192
+ "ALTERNATIVE_ROCK": 188,
193
+ "HORROR_PUNK": 189,
194
+ "GARAGE_ROCK": 190,
195
+ "POWER_POP": 191,
196
+ "CHAMBER_MUSIC": 192,
197
+ "MANOUCHE": 193,
198
+ "RAP_ROCK": 194,
199
+ "DARK_WAVE": 195,
200
+ "NO_WAVE": 196,
201
+ "SOUL_JAZZ": 197,
202
+ "RAP_METAL": 198,
203
+ "BLUE-EYED_SOUL": 199,
204
+ "OLDIES": 200,
205
+ "CLASSIC_COUNTRY": 201,
206
+ "CLUB": 202,
207
+ "HIP_HOUSE": 203,
208
+ "WORLD_MUSIC": 204,
209
+ "TWEE_POP": 205,
210
+ "JAZZ_FUNK": 206,
211
+ "REGGAE": 207,
212
+ "ALTERNATIVE": 208,
213
+ "GRINDCORE": 209,
214
+ "DUB": 210,
215
+ "FUTUREPOP": 211,
216
+ "MATH-CORE": 212,
217
+ "COUNTRY_MUSIC": 213,
218
+ "HEARTLAND_ROCK": 214,
219
+ "8-BIT": 215,
220
+ "ROCK_STEADY": 216,
221
+ "GANGSTER_RAP": 217,
222
+ "MEMPHIS_SOUL": 218,
223
+ "TRIP_HOP": 219,
224
+ "SONATA": 220,
225
+ "SPACE_AGE_POP": 221,
226
+ "CONCERTO": 222,
227
+ "ROCKABILLY": 223,
228
+ "GOSPEL": 224,
229
+ "BREAKBEAT": 225,
230
+ "VIDEO_GAME_MUSIC": 226,
231
+ "FREESTYLE": 227,
232
+ "DEATH_METAL": 228,
233
+ "BRITISH_INVASION": 229,
234
+ "ISLAND_MUSIC": 230,
235
+ "CHORAL_MUSIC": 231,
236
+ "NEW_ORLEANS_BLUES": 232,
237
+ "ROOTS_REGGAE": 233,
238
+ "ELECTRONICA": 234,
239
+ "REMIX": 235,
240
+ "BROKEN_BEAT": 236,
241
+ "GAGAKU": 237,
242
+ "BIG_BEAT": 238,
243
+ "NEW_RAVE": 239,
244
+ "HARD_HOUSE": 240,
245
+ "PINOY_ROCK": 241,
246
+ "BANDA": 242,
247
+ "CANTONESE_POP": 243,
248
+ "ACID_JAZZ": 244,
249
+ "PSYCHEDELIC_TRANCE": 245,
250
+ "BLUEGRASS": 246,
251
+ "COOL_JAZZ": 247,
252
+ "MUSETTE": 248,
253
+ "SKA_PUNK": 249,
254
+ "BOLERO": 250,
255
+ "PATRIOTIC": 251,
256
+ "ZOUK": 252,
257
+ "LATIN_POP": 253,
258
+ "BAROQUE_MUSIC": 254,
259
+ "MEXICO": 255,
260
+ "BREAKCORE": 256,
261
+ "STRIDE": 257,
262
+ "SURF_MUSIC": 258,
263
+ "PIANO_ROCK": 259,
264
+ "DREAM_POP": 260,
265
+ "LULLABY": 261,
266
+ "SAMBA": 262,
267
+ "CABARET": 263,
268
+ "INDIE_POP": 264,
269
+ "PARODY": 265,
270
+ "CHAMBER_POP": 266,
271
+ "BEATBOXING": 267,
272
+ "DANCE": 268,
273
+ "FUNKY_HOUSE": 269,
274
+ "TANGO": 270,
275
+ "DISCO_HOUSE": 271,
276
+ "WESTERN_SWING": 272,
277
+ "ACOUSTIC": 273,
278
+ "MATH_ROCK": 274,
279
+ "NU_METAL": 275,
280
+ "NORTHERN_SOUL": 276,
281
+ "ALTERNATIVE_HIP_HOP": 277,
282
+ "BASS_MUSIC": 278,
283
+ "RAP": 279,
284
+ "HIP_POP": 280,
285
+ "JAM_BAND": 281,
286
+ "AFROBEAT": 282,
287
+ "SWEDEN": 283,
288
+ "ROOTS_ROCK": 284,
289
+ "LATIN_JAZZ": 285,
290
+ "MELODIC_HARDCORE": 286,
291
+ "MASSACHUSETTS": 287,
292
+ "SYNTHPOP": 288,
293
+ "RAI": 289,
294
+ "COMEDY": 290,
295
+ "SPEED_METAL": 291,
296
+ "BHANGRA": 292,
297
+ "NEOFOLK": 293,
298
+ "NU_BREAKS": 294,
299
+ "HARDSTYLE": 295,
300
+ "ELECTROPOP": 296,
301
+ "ELECTRONIC_MUSIC": 297,
302
+ "FREE_MUSIC": 298,
303
+ "PSYCHOBILLY": 299,
304
+ "FREAKBEAT": 300,
305
+ "GRUNGE": 301,
306
+ "SYMPHONIC_ROCK": 302,
307
+ "PIANO": 303,
308
+ "INDUSTRIAL_DANCE": 304,
309
+ "CALYPSO": 305,
310
+ "SYNTH_PUNK": 306,
311
+ "ELECTRO_ROCK": 307,
312
+ "ALTERNATIVE_COUNTRY": 308,
313
+ "JANGLE_POP": 309,
314
+ "ALL-FEMALE": 310,
315
+ "RHYTHM_NOISE": 311,
316
+ "CHRISTIAN_ROCK": 312,
317
+ "STRING_QUARTET": 313,
318
+ "CRUNK": 314,
319
+ "CHINESE_MUSIC": 315,
320
+ "BLACKENED_DEATH_METAL": 316,
321
+ "MALE_VOCALIST": 317,
322
+ "GAME": 318,
323
+ "SYMPHO_BLACK_METAL": 319,
324
+ "FOLK_METAL": 320,
325
+ "DUTCH": 321,
326
+ "SALSA": 322,
327
+ "NOISE_POP": 323,
328
+ "FUNK_METAL": 325,
329
+ "STRAIGHT_EDGE": 326,
330
+ "FREE_JAZZ": 327,
331
+ "NU_JAZZ": 328,
332
+ "SPANISH_POP": 329,
333
+ "GABBA": 330,
334
+ "CHRISTIAN_METAL": 331,
335
+ "80S_COUNTRY": 332,
336
+ "2_TONE": 333,
337
+ "WORLD_FUSION": 334,
338
+ "COMEDY_ROCK": 335,
339
+ "KLEZMER": 336,
340
+ "POST-HARDCORE": 337,
341
+ "RANCHERA": 338,
342
+ "ZYDECO": 339,
343
+ "SPOKEN_WORD": 340,
344
+ "GLITCH": 341,
345
+ "INDIE": 342,
346
+ "KRAUT_ROCK": 343,
347
+ "NOISE_ROCK": 344,
348
+ "CHRISTIAN_MUSIC": 345,
349
+ "INDIETRONICA": 346,
350
+ "POLKA": 347,
351
+ "GRIME": 348,
352
+ "LOVERS_ROCK": 349,
353
+ "MODERN_LAIKA": 350,
354
+ "SYMPHONY": 351,
355
+ "KARAOKE": 352,
356
+ "MINIMAL": 353,
357
+ "IRISH_FOLK": 354,
358
+ "FINISH": 355,
359
+ "BRASS_BAND": 356,
360
+ "PIANO_BLUES": 357,
361
+ "POWER_ELECTRONICS": 358,
362
+ "HARD_BOP": 359,
363
+ "HARDCORE_PUNK": 360,
364
+ "MOTOWN_AND_SOUL": 361,
365
+ "BOY_BAND": 362,
366
+ "SUOMI_ROCK": 363,
367
+ "BUBBLEGUM_POP": 364,
368
+ "ELECTRIC_BLUES": 365,
369
+ "RIOT_GRRRL": 366,
370
+ "CALIFORNIA": 367,
371
+ "BEAT": 368,
372
+ "NEW_JACK_SWING": 369,
373
+ "ELECTRO-INDUSTRIAL": 370,
374
+ "PORTUGESE": 371,
375
+ "WORLD_BEAT": 372,
376
+ "RELIGIOUS_MUSIC": 373,
377
+ "CLOSE_HARMONY": 374,
378
+ "STONER_METAL": 375,
379
+ "MADCHESTER": 376,
380
+ "OI": 377,
381
+ "NEWGRASS": 378,
382
+ "BASS": 379,
383
+ "MUSICAL_THEATER": 380,
384
+ "ELECTROCLASH": 381,
385
+ "CHICAGO_BLUES": 382,
386
+ "BRAZILIAN_JAZZ": 383,
387
+ "SOUTHERN_GOSPEL": 384,
388
+ "TEXAS": 385,
389
+ "SWISS": 386,
390
+ "DUBSTEP": 387,
391
+ "STRING": 388,
392
+ "CLASSIC_FEMALE_BLUES": 389,
393
+ "CANTO_NUEVO": 390,
394
+ "EXOTICA": 391,
395
+ "GREEK": 392,
396
+ "GERMAN": 393,
397
+ "CHANT": 394,
398
+ "CHRISTIAN_HARDCORE": 395,
399
+ "MAMBO": 396,
400
+ "DANCE-PUNK": 397,
401
+ "FOLK_PUNK": 398,
402
+ "ILLBIENT": 399,
403
+ "GHETTO_TECH": 400,
404
+ "BEL_CANTO": 401,
405
+ "FUSION": 402,
406
+ "DEATH_CORE": 403,
407
+ "RAVE": 404,
408
+ "DARK_CABARET": 405,
409
+ "MARIACHI": 406,
410
+ "TURNABLISM": 407,
411
+ "BUBBLEGUM_DANCE": 408,
412
+ "GOTHIC": 409,
413
+ "STONER_ROCK": 410,
414
+ "CRUST_PUNK": 411,
415
+ "SCREAMO": 412,
416
+ "NEDERPOP": 413,
417
+ "SWEDISH": 414,
418
+ "WORLD": 415,
419
+ "CANTERBURY_SCENE": 416,
420
+ "CHIPTUNE": 417,
421
+ "INTELLIGENT_DANCE_MUSIC": 418,
422
+ "GARAGE_PUNK": 419,
423
+ "FRENCH": 420,
424
+ "SVERIGE": 422,
425
+ "ROCK_AND_INDIE": 424,
426
+ "IRISH": 425,
427
+ "POST-PUNK": 426,
428
+ "CLASSIC_POP_AND_ROCK": 429,
429
+ "FOLK": 430,
430
+ "POP_AND_CHART": 431,
431
+ "4AD": 432,
432
+ "BRITISH": 433,
433
+ "AMERICAN": 435,
434
+ "ITALIAN": 437,
435
+ "UK": 438,
436
+ "PUNK": 442,
437
+ "RNB": 443,
438
+ "SOUTH_AMERICAN": 445,
439
+ "CANADIAN": 447,
440
+ "EUROPEAN": 448,
441
+ "VANCOUVER": 451,
442
+ "CONDUCTOR": 453,
443
+ "FINNISH": 455,
444
+ "INDUSTRIAL_ROCK": 456,
445
+ "PRODUCTION_MUSIC": 460,
446
+ "EASY_LISTENING_SOUNDTRACKS_AND_MUSICALS": 461,
447
+ "BELGIAN": 463,
448
+ "PUERTO_RICAN": 464,
449
+ "HIP_HOP_RNB_AND_DANCE_HALL": 467,
450
+ "QUEBEC": 468,
451
+ "JBIGGA_J_BIGGA": 469,
452
+ "COUNTRY_POP": 470,
453
+ "SOUL_AND_REGGAE": 471,
454
+ "GLAM_METAL": 473,
455
+ "TRINIDADIAN": 475,
456
+ "PUNK_ROCK": 476,
457
+ "SIDE_PROJECT": 479,
458
+ "HIP-HOP": 480,
459
+ "DANCE_AND_ELECTRONICA": 481,
460
+ "BLUES_ROCK": 482,
461
+ "ONE-HIT_WONDER": 485,
462
+ "SYMPHONIC_METAL": 489,
463
+ "SPANISH": 491,
464
+ "ENGLISH": 493,
465
+ "FRANKFURT": 494,
466
+ "AUSTRALIAN": 495,
467
+ "NETTETAL": 496,
468
+ "VERONA": 498,
469
+ "DARIN_INSANITY": 500,
470
+ "BARBADIAN": 503,
471
+ "ADULT_CONTEMPORARY": 504,
472
+ "RED_ONE_DAVID_FORBES_REMIX": 506,
473
+ "MANDARIN": 508,
474
+ "TOKYO": 509,
475
+ "X_FACTOR": 510,
476
+ "LOVERBOY": 512,
477
+ "CUBAN": 513,
478
+ "A_CAPPELLA": 514,
479
+ "PERFORMER": 515,
480
+ "AUSTRIAN": 516,
481
+ "M\u00c9XICO": 517,
482
+ "SAXAPHONE": 518,
483
+ "SWAMP_ROCK": 520,
484
+ "WELSH": 522,
485
+ "FRAN\u00c7AIS": 523,
486
+ "JUMP_BLUES": 524,
487
+ "LOOK_AT_US": 527,
488
+ "EUROVISION": 528,
489
+ "NORFOLK": 529,
490
+ "LAFEE": 530,
491
+ "ATLANTA": 532,
492
+ "SOUNDTRACK_COMPOSER": 533,
493
+ "WHISTLE_REGISTER": 534,
494
+ "ISKELM\u00c4": 535,
495
+ "DANISH": 536,
496
+ "MEXICAN": 537,
497
+ "BREAKS": 540,
498
+ "TURKISH": 541,
499
+ "ARABIC": 542,
500
+ "ESPA\u00d1A": 543,
501
+ "DRUM_AND_BASS": 545,
502
+ "WALL_OF_SOUND": 546,
503
+ "DJ": 547,
504
+ "JAZZ_AND_BLUES": 548,
505
+ "ROCK_EN_ESPANOL": 550,
506
+ "CHINESE": 551,
507
+ "AMERICAN_UNDERGROUND": 552,
508
+ "BRAZILIAN": 553,
509
+ "WISIN_Y_YANDEL": 554,
510
+ "NORWEGIAN": 557,
511
+ "VIOLIN": 560,
512
+ "COLLEGE_ROCK": 561,
513
+ "ROCK_ROLL": 562,
514
+ "PERTH": 563,
515
+ "NEUE_DEUTSCHE_WELLE": 564,
516
+ "POST-EMO": 565,
517
+ "SCOTTISH": 566,
518
+ "KIM_BURAN": 567,
519
+ "RUSSIAN": 568,
520
+ "BB12B20E": 569,
521
+ "POLISH": 570,
522
+ "POP_IDOLS": 572,
523
+ "PHIL_SPECTOR": 573,
524
+ "BLACK_METAL-AMBIENT": 576,
525
+ "COLOMBIAN": 577,
526
+ "CHRISTIAN": 578,
527
+ "JUICY_IBIZA": 580,
528
+ "ESPA\u00d1OL": 582,
529
+ "OPERATIC_POP": 583,
530
+ "HUNGARIAN": 586,
531
+ "CAJUN": 587,
532
+ "MILEY-CYRUS": 588,
533
+ "SWING": 589,
534
+ "DANOIS": 590,
535
+ "70S": 591,
536
+ "SU\u00c9DOIS": 592,
537
+ "HONG_KONG": 594,
538
+ "BIRMINGHAM": 595,
539
+ "UNITED_KINGDOM": 596,
540
+ "SLOVAK": 597,
541
+ "BOGUS_ARTIST_QUESTIONMARK": 598,
542
+ "AMERICAN_IDOL": 600,
543
+ "HORNCHURCH": 601,
544
+ "TAIWANESE": 602,
545
+ "WISCONSIN": 603,
546
+ "WAGNERIAN_ROCK": 604,
547
+ "THE_NETHERLANDS": 605,
548
+ "THRASH_METAL": 606,
549
+ "SOUTH_AFRICAN": 607,
550
+ "NEOPROG": 609,
551
+ "AO_VIVO": 610,
552
+ "FILK": 611,
553
+ "M2M": 612,
554
+ "SOBER": 614,
555
+ "BLUEGRASS_COUNTRY_ALTERNATIVE": 615,
556
+ "TAYLOR-SWIFT": 617,
557
+ "BEIRUT": 618,
558
+ "KARAJAN": 620,
559
+ "HOUSE": 621,
560
+ "BIG_BAND": 623,
561
+ "SOUL_CENTRAL": 624,
562
+ "PERFORMANCE_NAME": 625,
563
+ "CROSSOVER": 626,
564
+ "KWAITO": 627,
565
+ "JOS\u00c9_JOS\u00c9": 628,
566
+ "CANTERBURY": 630,
567
+ "KINGSTON": 631,
568
+ "CZECH": 632,
569
+ "CONTEMPORARY_CHRISTIAN": 633,
570
+ "GIRL_GROUP": 635,
571
+ "PORTUGAL": 636,
572
+ "SPECIAL_PURPOSE_ARTIST": 637,
573
+ "AM\u00c9RICAIN": 638,
574
+ "AMBIENT": 640,
575
+ "SOUTH_CAROLINA": 641,
576
+ "NORTHERN_IRISH": 643,
577
+ "MELODIC_ROCK": 645,
578
+ "NDW": 646,
579
+ "SOPRANO": 647,
580
+ "ALT-COUNTRY": 648,
581
+ "BRITPOP": 650,
582
+ "DIRECTOR": 651,
583
+ "NUEVA_TROVA": 652,
584
+ "US": 653,
585
+ "DIXIELAND": 656,
586
+ "POPERA": 657,
587
+ "ROCK_AND_ROLL": 658,
588
+ "BRONIUS_BAGACKAS": 660,
589
+ "KALAMAZOO": 663,
590
+ "HELLO_PROJECT": 664,
591
+ "TURNTABLISM": 665,
592
+ "POST-ROCK": 666,
593
+ "RAPCORE": 667,
594
+ "GOTHIC_ROCK": 668,
595
+ "DR_DOCTOR_DRSAPO_SAPO_SPANISH_GROUP_POP_ROCK": 669,
596
+ "NEW_YORK": 670,
597
+ "NEW_BRUNSWICK": 671,
598
+ "LATVIAN": 672,
599
+ "CELLO": 673,
600
+ "OLD_SCHOOL_HIP-HOP": 674,
601
+ "12": 677,
602
+ "13": 678,
603
+ "11": 679,
604
+ "8": 680,
605
+ "10": 681,
606
+ "4": 682,
607
+ "14": 683,
608
+ "7": 684,
609
+ "1": 685,
610
+ "15": 686,
611
+ "16": 687,
612
+ "3": 688,
613
+ "9": 689,
614
+ "2": 690,
615
+ "5": 691,
616
+ "6": 692,
617
+ "0": 698,
618
+ "M": 753,
619
+ "m": 754,
620
+ "-1": 756,
621
+ "-2": 757,
622
+ "-3": 758,
623
+ "": 760,
624
+ "64": 761,
625
+ "43": 763,
626
+ "64(sus4)": 765,
627
+ "(sus4)": 766,
628
+ "64(sus2)": 767,
629
+ "6(sus4)": 768,
630
+ "(sus2)": 769,
631
+ "6(sus2)": 770,
632
+ "65": 771,
633
+ "(+)": 772,
634
+ "64(+)": 773,
635
+ "6(+)": 774,
636
+ "drums_0": 842,
637
+ "string_ensemble_2": 843,
638
+ "acoustic_bass": 844,
639
+ "steel_guitar": 845,
640
+ "pan_flute": 846,
641
+ "piano": 847,
642
+ "flute": 848,
643
+ "pizzicato": 849,
644
+ "string_ensemble_1": 850,
645
+ "pad_new_age": 851,
646
+ "trombone": 852,
647
+ "electric_piano_1": 853,
648
+ "telephone_ring": 854,
649
+ "acoustic_guitar": 855,
650
+ "clarinet": 856,
651
+ "choir_oohs": 857,
652
+ "fretless_bass": 858,
653
+ "brass_section": 859,
654
+ "harp": 860,
655
+ "electric_bass_finger": 861,
656
+ "pad_choir": 862,
657
+ "music_box": 863,
658
+ "soprano_sax": 864,
659
+ "bass_lead": 865,
660
+ "honky_tonk": 866,
661
+ "choir_aahs": 867,
662
+ "synth_brass_1": 868,
663
+ "rock_organ": 869,
664
+ "bright_piano": 870,
665
+ "electric_bass_pick": 871,
666
+ "synth_choir": 872,
667
+ "charang_lead": 873,
668
+ "reverse_cymbal": 874,
669
+ "pad_halo": 875,
670
+ "seashore": 876,
671
+ "overdriven_guitar": 877,
672
+ "pad_warm": 878,
673
+ "clean_guitar": 879,
674
+ "alto_sax": 880,
675
+ "synth_string_1": 881,
676
+ "sawtooth_lead": 882,
677
+ "fx_brightness": 883,
678
+ "electric_piano_2": 884,
679
+ "synth_bass_2": 885,
680
+ "marimba": 886,
681
+ "distortion_guitar": 887,
682
+ "calliope_lead": 888,
683
+ "harmonica": 889,
684
+ "harmonic_guitar": 890,
685
+ "slap_bass_2": 891,
686
+ "pad_bowed": 892,
687
+ "electric_piano": 893,
688
+ "orchestra_hit": 894,
689
+ "synth_string_2": 895,
690
+ "bassoon": 896,
691
+ "drawbar_organ": 897,
692
+ "fx_crystal": 898,
693
+ "oboe": 899,
694
+ "muted_guitar": 900,
695
+ "breath_noise": 901,
696
+ "whistle": 902,
697
+ "tenor_sax": 903,
698
+ "timpani": 904,
699
+ "french_horn": 905,
700
+ "jazz_guitar": 906,
701
+ "violin": 907,
702
+ "applause": 908,
703
+ "trumpet": 909,
704
+ "piccolo": 910,
705
+ "tremolo_string": 911,
706
+ "pad_polysynth": 912,
707
+ "slap_bass_1": 913,
708
+ "synth_bass_1": 914,
709
+ "percussive_organ": 915,
710
+ "harpsichord": 916,
711
+ "church_organ": 917,
712
+ "viola": 918,
713
+ "english_horn": 919,
714
+ "fx_rain": 920,
715
+ "tuba": 921,
716
+ "guitar_fret_noise": 922,
717
+ "fx_echoes": 923,
718
+ "bagpipe": 924,
719
+ "accordion": 925,
720
+ "cello": 926,
721
+ "reed_organ": 927,
722
+ "contrabass": 928,
723
+ "muted_trumpet": 929,
724
+ "steel_drums": 930,
725
+ "baritone_sax": 931,
726
+ "clavi": 932,
727
+ "dulcimer": 933,
728
+ "woodblock": 934,
729
+ "vibraphone": 935,
730
+ "shamisen": 936,
731
+ "gunshot": 937,
732
+ "bird_tweet": 938,
733
+ "tubular_bells": 939,
734
+ "synth_drum": 940,
735
+ "celesta": 941,
736
+ "synth_brass_2": 942,
737
+ "shakuhachi": 943,
738
+ "square_lead": 944,
739
+ "pad_sweep": 945,
740
+ "voice_lead": 946,
741
+ "kalimba": 947,
742
+ "fx_atmosphere": 948,
743
+ "tinkle_bell": 949,
744
+ "taiko_drum": 950,
745
+ "recorder": 951,
746
+ "ocarina": 952,
747
+ "blown_bottle": 953,
748
+ "glockenspiel": 954,
749
+ "pad_metallic": 955,
750
+ "tango_accordion": 956,
751
+ "fx_soundtrack": 957,
752
+ "xylophone": 958,
753
+ "chiff_lead": 959,
754
+ "koto": 960,
755
+ "agogo": 961,
756
+ "fx_globlins": 962,
757
+ "banjo": 963,
758
+ "fiddle": 964,
759
+ "fx_sci_fi": 965,
760
+ "melodic_tom": 966,
761
+ "sitar": 967,
762
+ "fifths_lead": 968,
763
+ "shanai": 969,
764
+ "helicopter": 970,
765
+ "low": 971,
766
+ "medium": 972,
767
+ "high": 973,
768
+ "very_high": 974,
769
+ "-4": 983,
770
+ "ff": 985,
771
+ "mf": 986,
772
+ "f": 987,
773
+ "mp": 988,
774
+ "p": 989,
775
+ "ppp": 990,
776
+ "fff": 991,
777
+ "pp": 992,
778
+ "n": 993,
779
+ "r": 994,
780
+ "d": 995,
781
+ "s": 996,
782
+ "l": 997,
783
+ "h": 998,
784
+ "25": 1001,
785
+ "23": 1009,
786
+ "19": 1010,
787
+ "27": 1011,
788
+ "21": 1012,
789
+ "17": 1013,
790
+ "31": 1014,
791
+ "29": 1016,
792
+ "45": 1017,
793
+ "47": 1018,
794
+ "37": 1021,
795
+ "39": 1022,
796
+ "35": 1023,
797
+ "33": 1024,
798
+ "41": 1025,
799
+ "55": 1027,
800
+ "53": 1028,
801
+ "51": 1029,
802
+ "49": 1030
803
+ },
804
+ "unk_token": "UNK__"
805
+ }
806
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "PAD",
5
+ "single_word": false,
6
+ "lstrip": false,
7
+ "rstrip": false,
8
+ "normalized": false,
9
+ "special": true,
10
+ "id": 0
11
+ },
12
+ "1": {
13
+ "content": "UNK__",
14
+ "single_word": false,
15
+ "lstrip": false,
16
+ "rstrip": false,
17
+ "normalized": false,
18
+ "special": true,
19
+ "id": 1
20
+ },
21
+ "2": {
22
+ "content": "UNK",
23
+ "single_word": false,
24
+ "lstrip": false,
25
+ "rstrip": false,
26
+ "normalized": false,
27
+ "special": true,
28
+ "id": 2
29
+ },
30
+ "3": {
31
+ "content": "MASK",
32
+ "single_word": false,
33
+ "lstrip": false,
34
+ "rstrip": false,
35
+ "normalized": false,
36
+ "special": true,
37
+ "id": 3
38
+ },
39
+ "4": {
40
+ "content": "BOS",
41
+ "single_word": false,
42
+ "lstrip": false,
43
+ "rstrip": false,
44
+ "normalized": false,
45
+ "special": true,
46
+ "id": 4
47
+ },
48
+ "5": {
49
+ "content": "EOS",
50
+ "single_word": false,
51
+ "lstrip": false,
52
+ "rstrip": false,
53
+ "normalized": false,
54
+ "special": true,
55
+ "id": 5
56
+ }
57
+ },
58
+ "clean_up_tokenization_spaces": false,
59
+ "extra_special_tokens": {},
60
+ "model_max_length": 1000000000000000019884624838656,
61
+ "tokenizer_class": "PreTrainedTokenizerFast"
62
+ }