Upload tokenizer

added_tokens.json

@@ -1,4 +1,4 @@
 {
-  "</s>": 1259,
-  "<s>": 1258
+  "</s>": 126,
+  "<s>": 125
 }

tokenizer_config.json

@@ -1,6 +1,6 @@
 {
   "added_tokens_decoder": {
-    "1256": {
+    "123": {
       "content": "[UNK]",
       "lstrip": false,
       "normalized": true,
@@ -8,7 +8,7 @@
       "single_word": false,
       "special": true
     },
-    "1257": {
+    "124": {
       "content": "[PAD]",
       "lstrip": false,
       "normalized": true,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": true
     },
-    "1258": {
+    "125": {
       "content": "<s>",
       "lstrip": false,
       "normalized": true,
@@ -24,7 +24,7 @@
       "single_word": false,
       "special": true
     },
-    "1259": {
+    "126": {
       "content": "</s>",
       "lstrip": false,
       "normalized": true,

vocab.json

@@ -1,1260 +1,127 @@
 {
   " ": 0,
-  "AA0": 1,
-  "AA0AH0": 2,
-  "AA0AY0K": 3,
-  "AA0DH": 4,
-  "AA0F": 5,
-  "AA0HH": 6,
-  "AA0IH0": 7,
-  "AA0K": 8,
-  "AA0L": 9,
-  "AA0S": 10,
-  "AA0T": 11,
-  "AA0W": 12,
-  "AA0Y": 13,
-  "AA1": 14,
-  "AA2": 15,
-  "AE0": 16,
-  "AE1": 17,
-  "AE2": 18,
-  "AH0": 19,
-  "AH0AA0": 20,
-  "AH0AA0T": 21,
-  "AH0AA1": 22,
-  "AH0AA2": 23,
-  "AH0AE0": 24,
-  "AH0AE1": 25,
-  "AH0AH0": 26,
-  "AH0AH0B": 27,
-  "AH0AH0L": 28,
-  "AH0AH0N": 29,
-  "AH0AH0T": 30,
-  "AH0AH1": 31,
-  "AH0AH2": 32,
-  "AH0AO0": 33,
-  "AH0AO0L": 34,
-  "AH0AO1": 35,
-  "AH0AO2": 36,
-  "AH0AY0": 37,
-  "AH0AY0CH": 38,
-  "AH0AY0M": 39,
-  "AH0AY0N": 40,
-  "AH0AY0T": 41,
-  "AH0AY0W": 42,
-  "AH0B": 43,
-  "AH0CH": 44,
-  "AH0D": 45,
-  "AH0DH": 46,
-  "AH0EH0": 47,
-  "AH0EH1": 48,
-  "AH0EH2": 49,
-  "AH0ER0": 50,
-  "AH0ER1": 51,
-  "AH0EY2": 52,
-  "AH0F": 53,
-  "AH0G": 54,
-  "AH0HH": 55,
-  "AH0IH0": 56,
-  "AH0IH1": 57,
-  "AH0IY1": 58,
-  "AH0JH": 59,
-  "AH0K": 60,
-  "AH0L": 61,
-  "AH0M": 62,
-  "AH0N": 63,
-  "AH0OW0": 64,
-  "AH0OW1": 65,
-  "AH0P": 66,
-  "AH0R": 67,
-  "AH0S": 68,
-  "AH0SH": 69,
-  "AH0T": 70,
-  "AH0TH": 71,
-  "AH0V": 72,
-  "AH0W": 73,
-  "AH0Y": 74,
-  "AH0Z": 75,
-  "AH1": 76,
-  "AH1AH0": 77,
-  "AH1B": 78,
-  "AH1DH": 79,
-  "AH1F": 80,
-  "AH1HH": 81,
-  "AH1IH0": 82,
-  "AH1W": 83,
-  "AH2": 84,
-  "AO0": 85,
-  "AO0AA0": 86,
-  "AO0AE0": 87,
-  "AO0AH0": 88,
-  "AO0AH0B": 89,
-  "AO0AH0CH": 90,
-  "AO0AH0G": 91,
-  "AO0AH0HH": 92,
-  "AO0AH0JH": 93,
-  "AO0AH0K": 94,
-  "AO0AH0L": 95,
-  "AO0AH0M": 96,
-  "AO0AH0S": 97,
-  "AO0AH0T": 98,
-  "AO0AH0W": 99,
-  "AO0AH1": 100,
-  "AO0AO0": 101,
-  "AO0AW1": 102,
-  "AO0AY0W": 103,
-  "AO0B": 104,
-  "AO0D": 105,
-  "AO0DH": 106,
-  "AO0EH0": 107,
-  "AO0EH1": 108,
-  "AO0EY0": 109,
-  "AO0F": 110,
-  "AO0G": 111,
-  "AO0HH": 112,
-  "AO0IH0": 113,
-  "AO0JH": 114,
-  "AO0K": 115,
-  "AO0L": 116,
-  "AO0M": 117,
-  "AO0N": 118,
-  "AO0OW0K": 119,
-  "AO0P": 120,
-  "AO0R": 121,
-  "AO0S": 122,
-  "AO0SH": 123,
-  "AO0T": 124,
-  "AO0TH": 125,
-  "AO0W": 126,
-  "AO0Y": 127,
-  "AO1": 128,
-  "AO1AY0G": 129,
-  "AO1B": 130,
-  "AO1G": 131,
-  "AO1HH": 132,
-  "AO1M": 133,
-  "AO2": 134,
-  "AW0": 135,
-  "AW0AA0TH": 136,
-  "AW0AA0W": 137,
-  "AW0AA0Y": 138,
-  "AW0AH0": 139,
-  "AW0AY0G": 140,
-  "AW0AY0K": 141,
-  "AW0AY0N": 142,
-  "AW0B": 143,
-  "AW0D": 144,
-  "AW0DH": 145,
-  "AW0F": 146,
-  "AW0G": 147,
-  "AW0HH": 148,
-  "AW0IH0": 149,
-  "AW0K": 150,
-  "AW0L": 151,
-  "AW0M": 152,
-  "AW0P": 153,
-  "AW0R": 154,
-  "AW0SH": 155,
-  "AW0T": 156,
-  "AW0W": 157,
-  "AW0Y": 158,
-  "AW1": 159,
-  "AW1DH": 160,
-  "AW1T": 161,
-  "AY0": 162,
-  "AY0AA0": 163,
-  "AY0AA0Y": 164,
-  "AY0AE0": 165,
-  "AY0AE1": 166,
-  "AY0AH0": 167,
-  "AY0AH0G": 168,
-  "AY0AH0N": 169,
-  "AY0AO1": 170,
-  "AY0AY0": 171,
-  "AY0AY0W": 172,
-  "AY0B": 173,
-  "AY0CH": 174,
-  "AY0D": 175,
-  "AY0DH": 176,
-  "AY0EH1": 177,
-  "AY0EY1": 178,
-  "AY0EY2": 179,
-  "AY0F": 180,
-  "AY0G": 181,
-  "AY0HH": 182,
-  "AY0IH0": 183,
-  "AY0JH": 184,
-  "AY0K": 185,
-  "AY0L": 186,
-  "AY0M": 187,
-  "AY0N": 188,
-  "AY0OW0": 189,
-  "AY0OW1": 190,
-  "AY0P": 191,
-  "AY0R": 192,
-  "AY0S": 193,
-  "AY0SH": 194,
-  "AY0T": 195,
-  "AY0TH": 196,
-  "AY0V": 197,
-  "AY0W": 198,
-  "AY0Y": 199,
-  "AY1": 200,
-  "AY1F": 201,
-  "AY1K": 202,
-  "AY2": 203,
-  "B": 204,
-  "BAH0": 205,
-  "BG": 206,
-  "BL": 207,
-  "BN": 208,
-  "BT": 209,
-  "BW": 210,
-  "CH": 211,
-  "CHAE0": 212,
-  "CHAH0": 213,
-  "CHAH0B": 214,
-  "CHAH0G": 215,
-  "CHAH0M": 216,
-  "CHAH0P": 217,
-  "CHAH1": 218,
-  "CHAO0": 219,
-  "CHAY0W": 220,
-  "CHB": 221,
-  "CHD": 222,
-  "CHDH": 223,
-  "CHEH1": 224,
-  "CHF": 225,
-  "CHG": 226,
-  "CHHH": 227,
-  "CHIH0": 228,
-  "CHK": 229,
-  "CHL": 230,
-  "CHM": 231,
-  "CHS": 232,
-  "CHSH": 233,
-  "CHT": 234,
-  "CHTH": 235,
-  "CHW": 236,
-  "CHY": 237,
-  "D": 238,
-  "DAA0": 239,
-  "DAA1": 240,
-  "DAE0": 241,
-  "DAE1": 242,
-  "DAH0": 243,
-  "DAH0B": 244,
-  "DAH0G": 245,
-  "DAH0HH": 246,
-  "DAH0L": 247,
-  "DAH0R": 248,
-  "DAH0S": 249,
-  "DAH0V": 250,
-  "DAH0W": 251,
-  "DAH0Y": 252,
-  "DAO0": 253,
-  "DAO1": 254,
-  "DAW0": 255,
-  "DAY0": 256,
-  "DAY0AH0": 257,
-  "DAY0D": 258,
-  "DAY0F": 259,
-  "DAY0G": 260,
-  "DAY0HH": 261,
-  "DAY0L": 262,
-  "DAY0N": 263,
-  "DAY0P": 264,
-  "DAY0W": 265,
-  "DB": 266,
-  "DD": 267,
-  "DDH": 268,
-  "DEH1": 269,
-  "DEY0": 270,
-  "DEY0N": 271,
-  "DEY2": 272,
-  "DF": 273,
-  "DG": 274,
-  "DH": 275,
-  "DHAH0": 276,
-  "DHAH0B": 277,
-  "DHAH0D": 278,
-  "DHAH0G": 279,
-  "DHAH0R": 280,
-  "DHAH0S": 281,
-  "DHAH0W": 282,
-  "DHAO0": 283,
-  "DHD": 284,
-  "DHDH": 285,
-  "DHH": 286,
-  "DHHH": 287,
-  "DHIH0": 288,
-  "DHM": 289,
-  "DHN": 290,
-  "DHS": 291,
-  "DHT": 292,
-  "DHY": 293,
-  "DIH0": 294,
-  "DIH1": 295,
-  "DIY1": 296,
-  "DJH": 297,
-  "DK": 298,
-  "DL": 299,
-  "DM": 300,
-  "DN": 301,
-  "DOW0": 302,
-  "DOW1": 303,
-  "DP": 304,
-  "DR": 305,
-  "DS": 306,
-  "DSH": 307,
-  "DT": 308,
-  "DTH": 309,
-  "DV": 310,
-  "DW": 311,
-  "DY": 312,
-  "EH0": 313,
-  "EH1": 314,
-  "EH2": 315,
-  "ER0": 316,
-  "ER0AA1": 317,
-  "ER0AE0": 318,
-  "ER0AH0": 319,
-  "ER0AH0P": 320,
-  "ER0AH1": 321,
-  "ER0AO0": 322,
-  "ER0AO1": 323,
-  "ER0AW0": 324,
-  "ER0AY0": 325,
-  "ER0B": 326,
-  "ER0D": 327,
-  "ER0DH": 328,
-  "ER0EH0": 329,
-  "ER0EH1": 330,
-  "ER0F": 331,
-  "ER0G": 332,
-  "ER0HH": 333,
-  "ER0IH0": 334,
-  "ER0K": 335,
-  "ER0L": 336,
-  "ER0M": 337,
-  "ER0N": 338,
-  "ER0OW0": 339,
-  "ER0P": 340,
-  "ER0R": 341,
-  "ER0S": 342,
-  "ER0SH": 343,
-  "ER0T": 344,
-  "ER0TH": 345,
-  "ER0V": 346,
-  "ER0W": 347,
-  "ER0Y": 348,
-  "ER1": 349,
-  "EY0": 350,
-  "EY0AA0": 351,
-  "EY0AA0AW0": 352,
-  "EY0AA0S": 353,
-  "EY0AA0W": 354,
-  "EY0AA1": 355,
-  "EY0AE0": 356,
-  "EY0AH0": 357,
-  "EY0AH0L": 358,
-  "EY0AO0": 359,
-  "EY0AW0": 360,
-  "EY0AY0D": 361,
-  "EY0B": 362,
-  "EY0D": 363,
-  "EY0DH": 364,
-  "EY0EH1": 365,
-  "EY0EY0": 366,
-  "EY0F": 367,
-  "EY0G": 368,
-  "EY0HH": 369,
-  "EY0IH0": 370,
-  "EY0IY0": 371,
-  "EY0IY1": 372,
-  "EY0JH": 373,
-  "EY0K": 374,
-  "EY0L": 375,
-  "EY0M": 376,
-  "EY0N": 377,
-  "EY0OW1": 378,
-  "EY0P": 379,
-  "EY0R": 380,
-  "EY0S": 381,
-  "EY0T": 382,
-  "EY0TH": 383,
-  "EY0W": 384,
-  "EY0Y": 385,
-  "EY1": 386,
-  "EY1AE0": 387,
-  "EY1AH0": 388,
-  "EY1AY0HH": 389,
-  "EY1B": 390,
-  "EY1DH": 391,
-  "EY1F": 392,
-  "EY1M": 393,
-  "EY1P": 394,
-  "EY1W": 395,
-  "EY2": 396,
-  "F": 397,
-  "FAE0": 398,
-  "FAH0": 399,
-  "FAH0F": 400,
-  "FAH0M": 401,
-  "FAY0HH": 402,
-  "FAY0K": 403,
-  "FAY0W": 404,
-  "FB": 405,
-  "FD": 406,
-  "FDH": 407,
-  "FEH1": 408,
-  "FF": 409,
-  "FHH": 410,
-  "FIH0": 411,
-  "FIH1": 412,
-  "FK": 413,
-  "FM": 414,
-  "FN": 415,
-  "FP": 416,
-  "FR": 417,
-  "FS": 418,
-  "FSH": 419,
-  "FT": 420,
-  "FW": 421,
-  "FY": 422,
-  "G": 423,
-  "GAA0": 424,
-  "GAE0": 425,
-  "GAH0F": 426,
-  "GB": 427,
-  "GD": 428,
-  "GG": 429,
-  "GHH": 430,
-  "GIH0": 431,
-  "GJH": 432,
-  "GL": 433,
-  "GP": 434,
-  "GS": 435,
-  "GT": 436,
-  "GW": 437,
-  "HH": 438,
-  "IH": 439,
-  "IH0": 440,
-  "IH0AE0": 441,
-  "IH0G": 442,
-  "IH0IH0": 443,
-  "IH0K": 444,
-  "IH0L": 445,
-  "IH0N": 446,
-  "IH0R": 447,
-  "IH0S": 448,
-  "IH0W": 449,
-  "IH1": 450,
-  "IH2": 451,
-  "IY0": 452,
-  "IY0AA0": 453,
-  "IY0AA0AO0": 454,
-  "IY0AA0B": 455,
-  "IY0AA0DH": 456,
-  "IY0AA0G": 457,
-  "IY0AA0HH": 458,
-  "IY0AA0K": 459,
-  "IY0AA0L": 460,
-  "IY0AA0N": 461,
-  "IY0AA0S": 462,
-  "IY0AA1": 463,
-  "IY0AE0": 464,
-  "IY0AE1": 465,
-  "IY0AH0": 466,
-  "IY0AH0B": 467,
-  "IY0AH0CH": 468,
-  "IY0AH0F": 469,
-  "IY0AH0G": 470,
-  "IY0AH0L": 471,
-  "IY0AH0M": 472,
-  "IY0AH0P": 473,
-  "IY0AH0R": 474,
-  "IY0AH0S": 475,
-  "IY0AH0T": 476,
-  "IY0AH0V": 477,
-  "IY0AH0W": 478,
-  "IY0AH1": 479,
-  "IY0AO0": 480,
-  "IY0AO0W": 481,
-  "IY0AO1": 482,
-  "IY0AO2": 483,
-  "IY0AW0": 484,
-  "IY0AY0": 485,
-  "IY0AY0OW1": 486,
-  "IY0AY0W": 487,
-  "IY0B": 488,
-  "IY0CH": 489,
-  "IY0D": 490,
-  "IY0DH": 491,
-  "IY0EH0": 492,
-  "IY0EH1": 493,
-  "IY0EY0": 494,
-  "IY0EY1": 495,
-  "IY0F": 496,
-  "IY0G": 497,
-  "IY0HH": 498,
-  "IY0IH0": 499,
-  "IY0IH1": 500,
-  "IY0IY0K": 501,
-  "IY0IY1": 502,
-  "IY0JH": 503,
-  "IY0K": 504,
-  "IY0L": 505,
-  "IY0M": 506,
-  "IY0N": 507,
-  "IY0OW0": 508,
-  "IY0OW1": 509,
-  "IY0P": 510,
-  "IY0R": 511,
-  "IY0S": 512,
-  "IY0SH": 513,
-  "IY0T": 514,
-  "IY0TH": 515,
-  "IY0V": 516,
-  "IY0W": 517,
-  "IY0Y": 518,
-  "IY0Z": 519,
-  "IY0ZAH0B": 520,
-  "IY1": 521,
-  "JH": 522,
-  "JHAE0": 523,
-  "JHAH0": 524,
-  "JHAO0": 525,
-  "JHDH": 526,
-  "JHHH": 527,
-  "JHIH0": 528,
-  "JHL": 529,
-  "JHN": 530,
-  "JHP": 531,
-  "JHR": 532,
-  "JHW": 533,
-  "K": 534,
-  "KAE0": 535,
-  "KAH0": 536,
-  "KAH0B": 537,
-  "KAH0D": 538,
-  "KAH0F": 539,
-  "KAH0L": 540,
-  "KAH0P": 541,
-  "KAH0S": 542,
-  "KAW0": 543,
-  "KAY0K": 544,
-  "KAY0W": 545,
-  "KB": 546,
-  "KCH": 547,
-  "KD": 548,
-  "KDH": 549,
-  "KEH1": 550,
-  "KEY0": 551,
-  "KF": 552,
-  "KG": 553,
-  "KHH": 554,
-  "KIH0": 555,
-  "KIH1": 556,
-  "KJH": 557,
-  "KK": 558,
-  "KL": 559,
-  "KM": 560,
-  "KN": 561,
-  "KOW1": 562,
-  "KP": 563,
-  "KR": 564,
-  "KS": 565,
-  "KSH": 566,
-  "KT": 567,
-  "KW": 568,
-  "KY": 569,
-  "L": 570,
-  "LAA0DH": 571,
-  "LAA0G": 572,
-  "LAA0N": 573,
-  "LAE0": 574,
-  "LAE1": 575,
-  "LAH0": 576,
-  "LAH0L": 577,
-  "LAH0R": 578,
-  "LAO0": 579,
-  "LAO0L": 580,
-  "LAO0S": 581,
-  "LAO1": 582,
-  "LAW0": 583,
-  "LAW1": 584,
-  "LAW2": 585,
-  "LAY0D": 586,
-  "LAY0F": 587,
-  "LAY0HH": 588,
-  "LAY0JH": 589,
-  "LAY0K": 590,
-  "LAY0OW1": 591,
-  "LAY0T": 592,
-  "LAY0W": 593,
-  "LB": 594,
-  "LD": 595,
-  "LDH": 596,
-  "LEH0": 597,
-  "LEH1": 598,
-  "LEH2": 599,
-  "LF": 600,
-  "LG": 601,
-  "LHH": 602,
-  "LIH0": 603,
-  "LIY1": 604,
-  "LJH": 605,
-  "LK": 606,
-  "LL": 607,
-  "LM": 608,
-  "LN": 609,
-  "LOW1": 610,
-  "LP": 611,
-  "LR": 612,
-  "LS": 613,
-  "LSH": 614,
-  "LT": 615,
-  "LTH": 616,
-  "LV": 617,
-  "LW": 618,
-  "LY": 619,
-  "M": 620,
-  "MAE0": 621,
-  "MAE1": 622,
-  "MAH0": 623,
-  "MAH0F": 624,
-  "MAH0K": 625,
-  "MAH0R": 626,
-  "MAH0S": 627,
-  "MAO0": 628,
-  "MAO0N": 629,
-  "MAW0": 630,
-  "MAW1": 631,
-  "MAW2": 632,
-  "MAY0AA0": 633,
-  "MAY0AE0": 634,
-  "MAY0AH0G": 635,
-  "MAY0N": 636,
-  "MAY0T": 637,
-  "MAY0W": 638,
-  "MAY0Y": 639,
-  "MB": 640,
-  "MCH": 641,
-  "MD": 642,
-  "MDH": 643,
-  "MEH1": 644,
-  "MEY0": 645,
-  "MF": 646,
-  "MG": 647,
-  "MHH": 648,
-  "MIH0": 649,
-  "MJH": 650,
-  "MK": 651,
-  "ML": 652,
-  "MM": 653,
-  "MN": 654,
-  "MOW1": 655,
-  "MP": 656,
-  "MR": 657,
-  "MS": 658,
-  "MSH": 659,
-  "MT": 660,
-  "MTH": 661,
-  "MW": 662,
-  "MY": 663,
-  "N": 664,
-  "NAA0": 665,
-  "NAA0OW1": 666,
-  "NAA0Y": 667,
-  "NAE0": 668,
-  "NAH0": 669,
-  "NAH0B": 670,
-  "NAH0D": 671,
-  "NAH0G": 672,
-  "NAH0HH": 673,
-  "NAH0K": 674,
-  "NAH0L": 675,
-  "NAH0N": 676,
-  "NAH0R": 677,
-  "NAH0S": 678,
-  "NAH0T": 679,
-  "NAH0W": 680,
-  "NAH1": 681,
-  "NAO0": 682,
-  "NAO1": 683,
-  "NAW1": 684,
-  "NAY0": 685,
-  "NAY0AW0": 686,
-  "NAY0B": 687,
-  "NAY0D": 688,
-  "NAY0EY2": 689,
-  "NAY0HH": 690,
-  "NAY0JH": 691,
-  "NAY0K": 692,
-  "NAY0L": 693,
-  "NAY0S": 694,
-  "NAY0T": 695,
-  "NAY0W": 696,
-  "NB": 697,
-  "NCH": 698,
-  "ND": 699,
-  "NDH": 700,
-  "NEH0": 701,
-  "NEH1": 702,
-  "NER0": 703,
-  "NEY0": 704,
-  "NEY1": 705,
-  "NEY2": 706,
-  "NF": 707,
-  "NG": 708,
-  "NGAA1": 709,
-  "NGAE0": 710,
-  "NGAH0": 711,
-  "NGAH0B": 712,
-  "NGAH0F": 713,
-  "NGAH0G": 714,
-  "NGAH0L": 715,
-  "NGAH0N": 716,
-  "NGAH0P": 717,
-  "NGAH0T": 718,
-  "NGAO1": 719,
-  "NGAW0": 720,
-  "NGAY0HH": 721,
-  "NGAY0K": 722,
-  "NGAY0N": 723,
-  "NGB": 724,
-  "NGD": 725,
-  "NGDH": 726,
-  "NGEH0": 727,
-  "NGF": 728,
-  "NGG": 729,
-  "NGHH": 730,
-  "NGIH0": 731,
-  "NGIH1": 732,
-  "NGIY0": 733,
-  "NGJH": 734,
-  "NGK": 735,
-  "NGL": 736,
-  "NGM": 737,
-  "NGN": 738,
-  "NGOW1": 739,
-  "NGP": 740,
-  "NGR": 741,
-  "NGS": 742,
-  "NGSH": 743,
-  "NGT": 744,
-  "NGTH": 745,
-  "NGW": 746,
-  "NGY": 747,
-  "NHH": 748,
-  "NIH0": 749,
-  "NIH1": 750,
-  "NIY0": 751,
-  "NIY1": 752,
-  "NJH": 753,
-  "NK": 754,
-  "NL": 755,
-  "NM": 756,
-  "NN": 757,
-  "NOW1": 758,
-  "NP": 759,
-  "NR": 760,
-  "NS": 761,
-  "NSH": 762,
-  "NT": 763,
-  "NTH": 764,
-  "NV": 765,
-  "NW": 766,
-  "NY": 767,
-  "NZ": 768,
-  "OW0": 769,
-  "OW0AA0Y": 770,
-  "OW0AA1": 771,
-  "OW0AE0": 772,
-  "OW0AE1": 773,
-  "OW0AH0": 774,
-  "OW0AH0L": 775,
-  "OW0AH0P": 776,
-  "OW0AO0": 777,
-  "OW0AW0": 778,
-  "OW0AY0": 779,
-  "OW0AY0AH0": 780,
-  "OW0AY0B": 781,
-  "OW0AY0D": 782,
-  "OW0AY0EY2": 783,
-  "OW0AY0HH": 784,
-  "OW0AY0K": 785,
-  "OW0AY0L": 786,
-  "OW0AY0N": 787,
-  "OW0AY0W": 788,
-  "OW0B": 789,
-  "OW0CH": 790,
-  "OW0D": 791,
-  "OW0DH": 792,
-  "OW0EH1": 793,
-  "OW0EY0": 794,
-  "OW0F": 795,
-  "OW0G": 796,
-  "OW0HH": 797,
-  "OW0IH0": 798,
-  "OW0IH1": 799,
-  "OW0IY0": 800,
-  "OW0IY1": 801,
-  "OW0JH": 802,
-  "OW0K": 803,
-  "OW0L": 804,
-  "OW0M": 805,
-  "OW0N": 806,
-  "OW0P": 807,
-  "OW0R": 808,
-  "OW0S": 809,
-  "OW0SH": 810,
-  "OW0T": 811,
-  "OW0TH": 812,
-  "OW0W": 813,
-  "OW0Y": 814,
-  "OW1": 815,
-  "OW1DH": 816,
-  "OW1HH": 817,
-  "OW1T": 818,
-  "OY0": 819,
-  "OY0AA0S": 820,
-  "OY0AH0": 821,
-  "OY0HH": 822,
-  "OY0K": 823,
-  "OY0P": 824,
-  "OY0V": 825,
-  "OY0W": 826,
-  "OY1": 827,
-  "OY1AE0": 828,
-  "OY1DH": 829,
-  "OY1HH": 830,
-  "OY1L": 831,
-  "P": 832,
-  "PAE0": 833,
-  "PAH0": 834,
-  "PAH0B": 835,
-  "PAW0": 836,
-  "PAY0S": 837,
-  "PAY0W": 838,
-  "PB": 839,
-  "PD": 840,
-  "PDH": 841,
-  "PER1": 842,
-  "PF": 843,
-  "PG": 844,
-  "PHH": 845,
-  "PIH0": 846,
-  "PIH1": 847,
-  "PK": 848,
-  "PM": 849,
-  "POW1": 850,
-  "PP": 851,
-  "PS": 852,
-  "PSH": 853,
-  "PT": 854,
-  "PW": 855,
-  "PY": 856,
-  "R": 857,
-  "RAA0": 858,
-  "RAA0DH": 859,
-  "RAA0M": 860,
-  "RAA0W": 861,
-  "RAA0Y": 862,
-  "RAE0": 863,
-  "RAE1": 864,
-  "RAH0": 865,
-  "RAH0B": 866,
-  "RAH0D": 867,
-  "RAH0F": 868,
-  "RAH0W": 869,
-  "RAO0": 870,
-  "RAY0W": 871,
-  "RB": 872,
-  "RCH": 873,
-  "RD": 874,
-  "RDH": 875,
-  "REH0": 876,
-  "REH1": 877,
-  "RF": 878,
-  "RG": 879,
-  "RHH": 880,
-  "RIH0": 881,
-  "RK": 882,
-  "RL": 883,
-  "RM": 884,
-  "RN": 885,
-  "ROW1": 886,
-  "RP": 887,
-  "RR": 888,
-  "RS": 889,
-  "RSH": 890,
-  "RT": 891,
-  "RW": 892,
-  "RY": 893,
-  "RZN": 894,
-  "S": 895,
-  "SAA0": 896,
-  "SAA0AH0": 897,
-  "SAA0IH0": 898,
-  "SAA0K": 899,
-  "SAA0N": 900,
-  "SAA0Y": 901,
-  "SAA1": 902,
-  "SAA2": 903,
-  "SAE0": 904,
-  "SAH0": 905,
-  "SAH0B": 906,
-  "SAH0F": 907,
-  "SAH0G": 908,
-  "SAH0K": 909,
-  "SAH0L": 910,
-  "SAH0R": 911,
-  "SAH0V": 912,
-  "SAH1": 913,
-  "SAO0": 914,
-  "SAO1": 915,
-  "SAW0": 916,
-  "SAY0": 917,
-  "SAY0AH0": 918,
-  "SAY0N": 919,
-  "SAY0P": 920,
-  "SAY0S": 921,
-  "SAY0W": 922,
-  "SB": 923,
-  "SCH": 924,
-  "SD": 925,
-  "SDH": 926,
-  "SEY0": 927,
-  "SF": 928,
-  "SG": 929,
-  "SH": 930,
-  "SHAO0": 931,
-  "SHB": 932,
-  "SHCH": 933,
-  "SHF": 934,
-  "SHH": 935,
-  "SHHH": 936,
-  "SHIH0": 937,
-  "SHM": 938,
-  "SHN": 939,
-  "SHT": 940,
-  "SHW": 941,
-  "SHY": 942,
-  "SIH0": 943,
-  "SIH1": 944,
-  "SIY1": 945,
-  "SJH": 946,
-  "SK": 947,
-  "SL": 948,
-  "SM": 949,
-  "SN": 950,
-  "SOW1": 951,
-  "SP": 952,
-  "SR": 953,
-  "SS": 954,
-  "SSH": 955,
-  "ST": 956,
-  "STH": 957,
-  "SV": 958,
-  "SW": 959,
-  "SY": 960,
-  "SZ": 961,
-  "T": 962,
-  "TAA0": 963,
-  "TAA0Y": 964,
-  "TAA1": 965,
-  "TAE0": 966,
-  "TAE1": 967,
-  "TAH0": 968,
-  "TAH0AE1": 969,
-  "TAH0B": 970,
-  "TAH0CH": 971,
-  "TAH0D": 972,
-  "TAH0F": 973,
-  "TAH0G": 974,
-  "TAH0HH": 975,
-  "TAH0K": 976,
-  "TAH0L": 977,
-  "TAH0M": 978,
-  "TAH0N": 979,
-  "TAH0P": 980,
-  "TAH0R": 981,
-  "TAH0S": 982,
-  "TAH0W": 983,
-  "TAH0Y": 984,
-  "TAH1": 985,
-  "TAH2": 986,
-  "TAO0": 987,
-  "TAO0HH": 988,
-  "TAO0S": 989,
-  "TAO1": 990,
-  "TAW0": 991,
-  "TAW1": 992,
-  "TAY0AA0": 993,
-  "TAY0B": 994,
-  "TAY0D": 995,
-  "TAY0G": 996,
-  "TAY0HH": 997,
-  "TAY0K": 998,
-  "TAY0L": 999,
-  "TAY0N": 1000,
-  "TAY0S": 1001,
-  "TAY0SH": 1002,
-  "TAY0TH": 1003,
-  "TAY0W": 1004,
-  "TB": 1005,
-  "TCH": 1006,
-  "TD": 1007,
-  "TDH": 1008,
-  "TEH0": 1009,
-  "TEH1": 1010,
-  "TEY0": 1011,
-  "TF": 1012,
-  "TG": 1013,
-  "TH": 1014,
-  "THAE0": 1015,
-  "THAH0": 1016,
-  "THAH0K": 1017,
-  "THAH0W": 1018,
-  "THB": 1019,
-  "THD": 1020,
-  "THDH": 1021,
-  "THEH1": 1022,
-  "THH": 1023,
-  "THHH": 1024,
-  "THIH0": 1025,
-  "THK": 1026,
-  "THL": 1027,
-  "THM": 1028,
-  "THN": 1029,
-  "THS": 1030,
-  "THT": 1031,
-  "THW": 1032,
-  "THY": 1033,
-  "TIH0": 1034,
-  "TIH1": 1035,
-  "TIY1": 1036,
-  "TJH": 1037,
-  "TK": 1038,
-  "TL": 1039,
-  "TM": 1040,
-  "TN": 1041,
-  "TOW0": 1042,
-  "TOW1": 1043,
-  "TP": 1044,
-  "TR": 1045,
-  "TS": 1046,
-  "TSH": 1047,
-  "TSM": 1048,
-  "TT": 1049,
-  "TTH": 1050,
-  "TV": 1051,
-  "TW": 1052,
-  "TY": 1053,
-  "TZ": 1054,
-  "UH": 1055,
-  "UH0": 1056,
-  "UH0AE0": 1057,
-  "UH0AH0S": 1058,
-  "UH0AO0": 1059,
-  "UH0AW0": 1060,
-  "UH0AY0G": 1061,
-  "UH0AY0HH": 1062,
-  "UH0AY0N": 1063,
-  "UH0AY0R": 1064,
-  "UH0AY0T": 1065,
-  "UH0AY0TH": 1066,
-  "UH0B": 1067,
-  "UH0DH": 1068,
-  "UH0EH1": 1069,
-  "UH0HH": 1070,
-  "UH0IH0": 1071,
-  "UH0K": 1072,
-  "UH0M": 1073,
-  "UH0N": 1074,
-  "UH0S": 1075,
-  "UH0T": 1076,
-  "UH0W": 1077,
-  "UH0Y": 1078,
-  "UH1": 1079,
-  "UW0": 1080,
-  "UW0AA0": 1081,
-  "UW0AA0AE0": 1082,
-  "UW0AA0AH0B": 1083,
-  "UW0AA0B": 1084,
-  "UW0AA0DH": 1085,
-  "UW0AA0G": 1086,
-  "UW0AA0IH0": 1087,
-  "UW0AA0M": 1088,
-  "UW0AA0N": 1089,
-  "UW0AA0S": 1090,
-  "UW0AA0T": 1091,
-  "UW0AE0": 1092,
-  "UW0AH0": 1093,
-  "UW0AH0F": 1094,
-  "UW0AH0G": 1095,
-  "UW0AH0L": 1096,
-  "UW0AH1": 1097,
-  "UW0AH2": 1098,
-  "UW0AO0": 1099,
-  "UW0AO1": 1100,
-  "UW0B": 1101,
-  "UW0CH": 1102,
-  "UW0D": 1103,
-  "UW0DH": 1104,
-  "UW0EH0": 1105,
-  "UW0EH1": 1106,
-  "UW0ER0HH": 1107,
-  "UW0ER0M": 1108,
-  "UW0EY0": 1109,
-  "UW0EY1": 1110,
-  "UW0F": 1111,
-  "UW0G": 1112,
-  "UW0HH": 1113,
-  "UW0IH0": 1114,
-  "UW0IH1": 1115,
-  "UW0IY0": 1116,
-  "UW0JH": 1117,
-  "UW0K": 1118,
-  "UW0L": 1119,
-  "UW0M": 1120,
-  "UW0N": 1121,
-  "UW0P": 1122,
-  "UW0R": 1123,
-  "UW0S": 1124,
-  "UW0SH": 1125,
-  "UW0T": 1126,
-  "UW0TH": 1127,
-  "UW0V": 1128,
-  "UW0W": 1129,
-  "UW0Y": 1130,
-  "UW1": 1131,
-  "UW1JH": 1132,
-  "UW1K": 1133,
-  "UW2": 1134,
-  "V": 1135,
-  "VAA1": 1136,
-  "VAE0": 1137,
-  "VAE1": 1138,
-  "VAH0": 1139,
-  "VAH0B": 1140,
-  "VAH0CH": 1141,
-  "VAH0F": 1142,
-  "VAH0G": 1143,
-  "VAH0K": 1144,
-  "VAH0L": 1145,
-  "VAH0M": 1146,
-  "VAH0N": 1147,
-  "VAH0R": 1148,
-  "VAH0S": 1149,
-  "VAH0W": 1150,
-  "VAH1": 1151,
-  "VAO0": 1152,
-  "VAO1": 1153,
-  "VAW0": 1154,
-  "VAW1": 1155,
-  "VAY1": 1156,
-  "VB": 1157,
-  "VD": 1158,
-  "VDH": 1159,
-  "VEH1": 1160,
-  "VEY0": 1161,
-  "VF": 1162,
-  "VG": 1163,
-  "VHH": 1164,
-  "VIH0": 1165,
-  "VIH1": 1166,
-  "VK": 1167,
-  "VL": 1168,
-  "VM": 1169,
-  "VN": 1170,
-  "VOW1": 1171,
-  "VP": 1172,
-  "VR": 1173,
-  "VS": 1174,
-  "VSH": 1175,
-  "VT": 1176,
-  "VTH": 1177,
-  "VV": 1178,
-  "VW": 1179,
-  "VY": 1180,
-  "VZ": 1181,
-  "W": 1182,
-  "Y": 1183,
-  "Z": 1184,
-  "ZAA0": 1185,
-  "ZAA0AH0": 1186,
-  "ZAA0D": 1187,
-  "ZAA0G": 1188,
-  "ZAA0M": 1189,
-  "ZAA0N": 1190,
-  "ZAA0P": 1191,
-  "ZAA0TH": 1192,
-  "ZAA0Y": 1193,
-  "ZAA1": 1194,
-  "ZAE0": 1195,
-  "ZAH0": 1196,
-  "ZAH0B": 1197,
-  "ZAH0D": 1198,
-  "ZAH0F": 1199,
-  "ZAH0G": 1200,
-  "ZAH0HH": 1201,
-  "ZAH0K": 1202,
-  "ZAH0L": 1203,
-  "ZAH0M": 1204,
-  "ZAH0N": 1205,
-  "ZAH0P": 1206,
-  "ZAH0R": 1207,
-  "ZAH0S": 1208,
-  "ZAH0T": 1209,
-  "ZAH0V": 1210,
-  "ZAH0W": 1211,
-  "ZAH1": 1212,
-  "ZAO0": 1213,
-  "ZAO1": 1214,
-  "ZAW0": 1215,
-  "ZAY0": 1216,
-  "ZAY0AA0": 1217,
-  "ZAY0F": 1218,
-  "ZAY0G": 1219,
-  "ZAY0IH0": 1220,
-  "ZAY0K": 1221,
-  "ZAY0L": 1222,
-  "ZAY0M": 1223,
-  "ZAY0N": 1224,
-  "ZAY0T": 1225,
-  "ZAY0W": 1226,
-  "ZB": 1227,
-  "ZCH": 1228,
-  "ZD": 1229,
-  "ZDH": 1230,
-  "ZEH0": 1231,
-  "ZEH1": 1232,
-  "ZEY1": 1233,
-  "ZF": 1234,
-  "ZG": 1235,
-  "ZH": 1236,
-  "ZHH": 1237,
-  "ZIH0": 1238,
-  "ZIY1": 1239,
-  "ZJH": 1240,
-  "ZK": 1241,
-  "ZL": 1242,
-  "ZM": 1243,
-  "ZN": 1244,
-  "ZOW0": 1245,
-  "ZOW1": 1246,
-  "ZP": 1247,
-  "ZR": 1248,
-  "ZS": 1249,
-  "ZSH": 1250,
-  "ZT": 1251,
-  "ZTH": 1252,
-  "ZV": 1253,
-  "ZW": 1254,
-  "ZY": 1255,
-  "[PAD]": 1257,
-  "[UNK]": 1256
+  "<DEL>": 1,
+  "<DEL>0": 2,
+  "<DEL>1": 3,
+  "<unk>": 4,
+  "<unk>0": 5,
+  "AA": 6,
+  "AA0": 7,
+  "AA1": 8,
+  "AA2": 9,
+  "AE": 10,
+  "AE*": 11,
+  "AE0": 12,
+  "AE1": 13,
+  "AE2": 14,
+  "AH": 15,
+  "AH*": 16,
+  "AH0": 17,
+  "AH1": 18,
+  "AH2": 19,
+  "AO": 20,
+  "AO*": 21,
+  "AO0": 22,
+  "AO1": 23,
+  "AO2": 24,
+  "AR": 25,
+  "AW": 26,
+  "AW0": 27,
+  "AW1": 28,
+  "AW2": 29,
+  "AY": 30,
+  "AY*": 31,
+  "AY0": 32,
+  "AY1": 33,
+  "AY2": 34,
+  "B": 35,
+  "B*": 36,
+  "CH": 37,
+  "CH*": 38,
+  "D": 39,
+  "D*": 40,
+  "DH": 41,
+  "DR": 42,
+  "DZ": 43,
+  "EH": 44,
+  "EH*": 45,
+  "EH0": 46,
+  "EH1": 47,
+  "EH2": 48,
+  "ER": 49,
+  "ER*": 50,
+  "ER0": 51,
+  "ER1": 52,
+  "EY": 53,
+  "EY*": 54,
+  "EY0": 55,
+  "EY1": 56,
+  "EY2": 57,
+  "F": 58,
+  "F*": 59,
+  "G": 60,
+  "G*": 61,
+  "HH": 62,
+  "I<DEL>0": 63,
+  "IH": 64,
+  "IH*": 65,
+  "IH0": 66,
+  "IH1": 67,
+  "IH2": 68,
+  "IR": 69,
+  "IR0": 70,
+  "IY": 71,
+  "IY*": 72,
+  "IY0": 73,
+  "IY1": 74,
+  "JH": 75,
+  "JH*": 76,
+  "K": 77,
+  "K*": 78,
+  "L": 79,
+  "L*": 80,
+  "M": 81,
+  "M*": 82,
+  "N": 83,
+  "N*": 84,
+  "NG": 85,
+  "NG*": 86,
+  "NK": 87,
+  "OW": 88,
+  "OW*": 89,
+  "OW0": 90,
+  "OW1": 91,
+  "OY0": 92,
+  "OY1": 93,
+  "P": 94,
+  "P*": 95,
+  "R": 96,
+  "R*": 97,
+  "S": 98,
+  "S*": 99,
+  "SH": 100,
+  "T": 101,
+  "T*": 102,
+  "TH": 103,
+  "TH*": 104,
+  "THH": 105,
+  "TR": 106,
+  "TS": 107,
+  "UH": 108,
+  "UH*": 109,
+  "UH0": 110,
+  "UH1": 111,
+  "UW": 112,
+  "UW0": 113,
+  "UW1": 114,
+  "UW2": 115,
+  "V": 116,
+  "V*": 117,
+  "W": 118,
+  "W*": 119,
+  "Y": 120,
+  "Z": 121,
+  "ZH": 122,
+  "[PAD]": 124,
+  "[UNK]": 123
 }