Upload processor

Browse files

Files changed (5) hide show

added_tokens.json +4 -0
preprocessor_config.json +10 -0
special_tokens_map.json +28 -0
tokenizer_config.json +0 -0
vocab.json +1088 -0

added_tokens.json ADDED Viewed

	@@ -0,0 +1,4 @@

+{
+  "</s>": 1087,
+  "<s>": 1086
+}

preprocessor_config.json ADDED Viewed

	@@ -0,0 +1,10 @@

+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "processor_class": "Wav2Vec2Processor",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

special_tokens_map.json ADDED Viewed

	@@ -0,0 +1,28 @@

+{
+  "additional_special_tokens": [
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    }
+  ],
+  "bos_token": "<s>",
+  "eos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "[PAD]",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json ADDED Viewed

The diff for this file is too large to render. See raw diff

vocab.json ADDED Viewed

	@@ -0,0 +1,1088 @@

+{
+  "[PAD]": 1085,
+  "[UNK]": 1084,
+  "|": 529,
+  "ءَ": 451,
+  "ءَاْ": 371,
+  "ءَبْ": 370,
+  "ءَتْ": 833,
+  "ءَثْ": 74,
+  "ءَجْ": 154,
+  "ءَحْ": 319,
+  "ءَخْ": 279,
+  "ءَدْ": 93,
+  "ءَرْ": 807,
+  "ءَرْضْ": 589,
+  "ءَزْ": 4,
+  "ءَسْ": 867,
+  "ءَشْ": 1079,
+  "ءَصْ": 354,
+  "ءَطْ": 127,
+  "ءَعْ": 893,
+  "ءَغْ": 249,
+  "ءَؽْ": 880,
+  "ءَفْ": 330,
+  "ءَقْ": 374,
+  "ءَكْ": 298,
+  "ءَلْ": 922,
+  "ءَمْ": 129,
+  "ءَمْرْ": 962,
+  "ءَمْسْ": 58,
+  "ءَنْ": 108,
+  "ءَهْ": 622,
+  "ءَوْ": 25,
+  "ءَيْ": 804,
+  "ءَۉْ": 304,
+  "ءُ": 964,
+  "ءُخْ": 731,
+  "ءُرْ": 97,
+  "ءُزْ": 790,
+  "ءُسْ": 908,
+  "ءُقْ": 646,
+  "ءُلْ": 400,
+  "ءُمْ": 122,
+  "ءُنْ": 651,
+  "ءُوْ": 609,
+  "ءُوْنْ": 547,
+  "ءِ": 202,
+  "ءِءْ": 1055,
+  "ءِبْ": 644,
+  "ءِتْ": 982,
+  "ءِثْمْ": 574,
+  "ءِحْ": 514,
+  "ءِخْ": 793,
+  "ءِذْ": 809,
+  "ءِرْ": 364,
+  "ءِسْ": 314,
+  "ءِطْ": 245,
+  "ءِؽْ": 329,
+  "ءِفْ": 144,
+  "ءِقْ": 187,
+  "ءِلْ": 320,
+  "ءِمْ": 868,
+  "ءِنْ": 1028,
+  "ءِهْ": 462,
+  "ءِيْ": 820,
+  "ءِيْنْ": 484,
+  "ءِۉْ": 704,
+  "بَ": 752,
+  "بَاْ": 727,
+  "بَاْبْ": 428,
+  "بَبْ": 116,
+  "بَتْ": 70,
+  "بَثْ": 855,
+  "بَدْ": 131,
+  "بَرْ": 200,
+  "بَشْ": 461,
+  "بَطْ": 534,
+  "بَعْ": 552,
+  "بَؽْ": 209,
+  "بَقْ": 979,
+  "بَكْ": 1011,
+  "بَلْ": 832,
+  "بَنْ": 21,
+  "بَهْ": 1023,
+  "بَيْ": 71,
+  "بَيْتْ": 1061,
+  "بَۉْ": 168,
+  "بُ": 991,
+  "بُدْ": 263,
+  "بُرْ": 592,
+  "بُسْ": 823,
+  "بُشْ": 586,
+  "بُصْ": 895,
+  "بُعْ": 581,
+  "بُؽْ": 459,
+  "بُلْ": 39,
+  "بُمْ": 504,
+  "بُنْ": 247,
+  "بُهْ": 283,
+  "بُوْ": 548,
+  "بُوْرْ": 808,
+  "بُوْنْ": 658,
+  "بِ": 676,
+  "بِءْ": 452,
+  "بِتْ": 460,
+  "بِحْ": 482,
+  "بِدْ": 654,
+  "بِرْ": 593,
+  "بِسْ": 11,
+  "بِشْ": 1056,
+  "بِصْ": 805,
+  "بِعْ": 663,
+  "بِلْ": 594,
+  "بِمْ": 343,
+  "بِنْ": 383,
+  "بِهْ": 874,
+  "بِيْ": 114,
+  "بِيْرْ": 1024,
+  "بِيْلْ": 398,
+  "بِيْنْ": 243,
+  "بِيْهْ": 734,
+  "بِۉْ": 270,
+  "تَ": 469,
+  "تَءْ": 417,
+  "تَاْ": 787,
+  "تَاْبْ": 174,
+  "تَاْدْ": 87,
+  "تَبْ": 1030,
+  "تَتْ": 85,
+  "تَجْ": 310,
+  "تَحْ": 923,
+  "تَخْ": 621,
+  "تَدْ": 608,
+  "تَذْ": 719,
+  "تَرْ": 368,
+  "تَزْ": 850,
+  "تَسْ": 463,
+  "تَشْ": 437,
+  "تَصْ": 556,
+  "تَضْ": 387,
+  "تَطْ": 152,
+  "تَعْ": 446,
+  "تَغْ": 255,
+  "تَفْ": 1019,
+  "تَقْ": 102,
+  "تَكْ": 590,
+  "تَلْ": 5,
+  "تَمْ": 1072,
+  "تَنْ": 725,
+  "تَوْ": 679,
+  "تَيْ": 519,
+  "تَيْنْ": 278,
+  "تَۉْ": 293,
+  "تُ": 802,
+  "تُءْ": 1067,
+  "تُبْ": 666,
+  "تُتْ": 407,
+  "تُجْ": 393,
+  "تُرْ": 616,
+  "تُزْ": 677,
+  "تُسْ": 257,
+  "تُصْ": 557,
+  "تُضْ": 859,
+  "تُعْ": 284,
+  "تُكْ": 975,
+  "تُلْ": 927,
+  "تُمْ": 869,
+  "تُنْ": 696,
+  "تُوْ": 246,
+  "تُوْمْ": 1014,
+  "تُوْنْ": 897,
+  "تُۉْ": 560,
+  "تِ": 410,
+  "تِدْ": 95,
+  "تِذْ": 67,
+  "تِرْ": 262,
+  "تِسْ": 955,
+  "تِشْ": 52,
+  "تِصْ": 206,
+  "تِطْ": 768,
+  "تِعْ": 673,
+  "تِؽْ": 796,
+  "تِلْ": 839,
+  "تِمْ": 267,
+  "تِنْ": 151,
+  "تِيْ": 474,
+  "تِيْمْ": 272,
+  "تِۉْ": 828,
+  "ثَ": 554,
+  "ثَاْ": 344,
+  "ثَاْقْ": 938,
+  "ثَجْ": 290,
+  "ثَرْ": 824,
+  "ثَلْ": 894,
+  "ثَمْ": 669,
+  "ثَهْ": 648,
+  "ثُ": 268,
+  "ثُرْ": 881,
+  "ثُلْ": 635,
+  "ثُمْ": 506,
+  "ثُوْ": 821,
+  "ثُوْثْ": 380,
+  "ثُوْنْ": 1037,
+  "ثِ": 596,
+  "ثِيْ": 1046,
+  "ثِيْمْ": 559,
+  "ثِيْنْ": 595,
+  "جَ": 226,
+  "جَاْ": 555,
+  "جَبْ": 404,
+  "جَتْ": 742,
+  "جَعْ": 604,
+  "جَلْ": 239,
+  "جَمْ": 772,
+  "جَنْ": 490,
+  "جَهْ": 640,
+  "جَيْ": 105,
+  "جَۉْ": 308,
+  "جُ": 815,
+  "جُدْ": 829,
+  "جُنْ": 169,
+  "جُوْ": 254,
+  "جُوْعْ": 2,
+  "جُوْنْ": 259,
+  "جِ": 1045,
+  "جِدْ": 786,
+  "جِنْ": 1043,
+  "جِيْ": 990,
+  "جِيْدْ": 119,
+  "جِيْلْ": 212,
+  "جِيْمْ": 628,
+  "جِيْنْ": 0,
+  "حَ": 838,
+  "حَاْ": 660,
+  "حَاْضْ": 878,
+  "حَبْ": 507,
+  "حَتْ": 216,
+  "حَدْ": 765,
+  "حَرْ": 220,
+  "حَقْ": 746,
+  "حَلْ": 109,
+  "حَمْ": 313,
+  "حَنْ": 172,
+  "حَيْ": 862,
+  "حُ": 771,
+  "حُبْ": 324,
+  "حُرْ": 73,
+  "حُسْ": 38,
+  "حُصْ": 436,
+  "حُضْ": 1015,
+  "حُقْ": 1017,
+  "حُكْ": 906,
+  "حُنْ": 672,
+  "حُوْ": 702,
+  "حُوْرْ": 229,
+  "حُوْنْ": 36,
+  "حِ": 413,
+  "حِبْ": 865,
+  "حِجْرْ": 686,
+  "حِسْ": 159,
+  "حِلْ": 580,
+  "حِمْ": 540,
+  "حِيْ": 342,
+  "حِيْطْ": 783,
+  "حِيْمْ": 104,
+  "حِيْنْ": 914,
+  "خَ": 775,
+  "خَاْ": 377,
+  "خَذْ": 996,
+  "خَرْ": 480,
+  "خَطْ": 165,
+  "خَفْ": 296,
+  "خَلْ": 549,
+  "خَنْ": 91,
+  "خَهْ": 269,
+  "خَوْ": 405,
+  "خَوْفْ": 627,
+  "خَيْ": 966,
+  "خُ": 925,
+  "خُذْ": 736,
+  "خُسْرْ": 397,
+  "خُنْ": 664,
+  "خُهْ": 100,
+  "خِ": 301,
+  "خِرْ": 179,
+  "خِفْ": 401,
+  "خِيْ": 935,
+  "خِيْلْ": 331,
+  "خِيْهْ": 219,
+  "دَ": 920,
+  "دَاْ": 449,
+  "دَاْءْ": 208,
+  "دَاْدْ": 493,
+  "دَاْرْ": 101,
+  "دَاْنْ": 363,
+  "دَاْهْ": 620,
+  "دَبْ": 970,
+  "دَتْ": 23,
+  "دَدْ": 625,
+  "دَرْ": 811,
+  "دَسْ": 524,
+  "دَعْ": 607,
+  "دَكْ": 688,
+  "دَلْ": 1036,
+  "دَمْ": 65,
+  "دَنْ": 853,
+  "دَهْ": 861,
+  "دَوْ": 681,
+  "دَيْ": 161,
+  "دَيْنْ": 309,
+  "دَۉْ": 898,
+  "دُ": 754,
+  "دُعْ": 668,
+  "دُؽْ": 237,
+  "دُكْ": 578,
+  "دُلْ": 356,
+  "دُمْ": 864,
+  "دُهْ": 961,
+  "دُوْ": 803,
+  "دُوْدْ": 306,
+  "دُوْرْ": 33,
+  "دُوْنْ": 55,
+  "دِ": 1048,
+  "دِثْ": 455,
+  "دِرْ": 700,
+  "دِلْ": 689,
+  "دِلْهْ": 412,
+  "دِمْ": 250,
+  "دِنْ": 24,
+  "دِيْ": 260,
+  "دِيْدْ": 1062,
+  "دِيْمْ": 81,
+  "دِيْنْ": 639,
+  "دِۉْ": 26,
+  "ذَ": 426,
+  "ذَاْ": 357,
+  "ذَاْبْ": 1004,
+  "ذَاْتْ": 751,
+  "ذَاْكْ": 9,
+  "ذَاْنْ": 280,
+  "ذَبْ": 911,
+  "ذَرْ": 181,
+  "ذَسْ": 795,
+  "ذَشْ": 1069,
+  "ذَصْ": 205,
+  "ذَكْ": 723,
+  "ذَلْ": 709,
+  "ذَمْ": 1038,
+  "ذَنْ": 99,
+  "ذُ": 315,
+  "ذُرْ": 717,
+  "ذُوْ": 916,
+  "ذُوْلْ": 415,
+  "ذِ": 50,
+  "ذِرْ": 302,
+  "ذِؽْ": 794,
+  "ذِكْ": 701,
+  "ذِلْ": 207,
+  "ذِمْ": 347,
+  "ذِنْ": 464,
+  "ذِيْ": 530,
+  "ذِيْبْ": 1064,
+  "ذِيْسْ": 325,
+  "ذِۉْ": 928,
+  "رَ": 439,
+  "رَءْ": 690,
+  "رَاْ": 623,
+  "رَاْءْ": 465,
+  "رَاْدْ": 570,
+  "رَاْهْ": 16,
+  "رَبْ": 546,
+  "رَتْ": 587,
+  "رَجْ": 713,
+  "رَجْعْ": 770,
+  "رَحْ": 96,
+  "رَدْ": 317,
+  "رَرْ": 107,
+  "رَسْ": 126,
+  "رَشْ": 475,
+  "رَضْ": 1025,
+  "رَعْ": 779,
+  "رَغْ": 489,
+  "رَؽْ": 889,
+  "رَقْ": 351,
+  "رَكْ": 120,
+  "رَلْ": 613,
+  "رَمْ": 1000,
+  "رَنْ": 902,
+  "رَهْ": 674,
+  "رَوْ": 541,
+  "رَيْ": 289,
+  "رَيْشْ": 261,
+  "رَۉْ": 72,
+  "رُ": 1022,
+  "رُجْ": 1012,
+  "رُرْ": 870,
+  "رُلْ": 708,
+  "رُمْ": 940,
+  "رُنْ": 425,
+  "رُوْ": 148,
+  "رُوْجْ": 191,
+  "رُوْنْ": 518,
+  "رُۉْ": 369,
+  "رِ": 697,
+  "رِبْ": 887,
+  "رِحْ": 831,
+  "رِزْ": 423,
+  "رِقْ": 888,
+  "رِكْ": 210,
+  "رِلْ": 139,
+  "رِمْ": 539,
+  "رِنْ": 912,
+  "رِهْ": 618,
+  "رِيْ": 31,
+  "رِيْبْ": 947,
+  "رِيْدْ": 800,
+  "رِيْعْ": 240,
+  "رِيْقْ": 691,
+  "رِيْمْ": 373,
+  "رِيْنْ": 993,
+  "زَ": 367,
+  "زَاْ": 946,
+  "زَجْ": 693,
+  "زَرْ": 230,
+  "زَقْ": 375,
+  "زَكْ": 481,
+  "زَلْ": 766,
+  "زَنْ": 835,
+  "زَهْ": 153,
+  "زَوْ": 537,
+  "زَيْ": 647,
+  "زُ": 37,
+  "زُرْ": 213,
+  "زُلْ": 221,
+  "زُوْ": 945,
+  "زُوْنْ": 7,
+  "زُيْ": 233,
+  "زِ": 10,
+  "زِعْ": 197,
+  "زِلْ": 762,
+  "زِيْ": 1078,
+  "سَ": 915,
+  "سَاْ": 597,
+  "سَاْبْ": 204,
+  "سَاْدْ": 420,
+  "سَبْ": 321,
+  "سَخْ": 910,
+  "سَدْ": 1073,
+  "سَرْ": 857,
+  "سَطْ": 675,
+  "سَعْ": 705,
+  "سَقْ": 20,
+  "سَلْ": 299,
+  "سَمْ": 653,
+  "سَنْ": 965,
+  "سَهْ": 83,
+  "سَوْ": 561,
+  "سَيْ": 695,
+  "سُ": 295,
+  "سُبْ": 60,
+  "سُجْ": 44,
+  "سُعْ": 1071,
+  "سُقْ": 78,
+  "سُلْ": 215,
+  "سُمْ": 732,
+  "سُوْ": 523,
+  "سُوْنْ": 511,
+  "سُيْ": 88,
+  "سِ": 926,
+  "سِتْ": 1044,
+  "سِجْ": 409,
+  "سِحْ": 667,
+  "سِشْ": 707,
+  "سِلْ": 854,
+  "سِنْ": 891,
+  "سِيْ": 225,
+  "سِيْمْ": 1074,
+  "سِۉْ": 1047,
+  "شَ": 601,
+  "شَءْ": 1059,
+  "شَاْ": 773,
+  "شَاْءْ": 973,
+  "شَتْ": 406,
+  "شَدْ": 780,
+  "شَرْ": 849,
+  "شَفْ": 444,
+  "شَقْ": 252,
+  "شَمْ": 564,
+  "شَهْ": 744,
+  "شَهْرْ": 435,
+  "شَيْ": 757,
+  "شُ": 579,
+  "شُدْ": 447,
+  "شِ": 904,
+  "شِرْ": 630,
+  "شِلْ": 29,
+  "شِيْ": 366,
+  "صَ": 45,
+  "صَاْ": 508,
+  "صَاْخْ": 764,
+  "صَاْدْ": 913,
+  "صَبْ": 583,
+  "صَبْرْ": 53,
+  "صَخْ": 876,
+  "ص��دْ": 788,
+  "صَدْعْ": 22,
+  "صَفْ": 505,
+  "صَلْ": 740,
+  "صَوْ": 231,
+  "صَوْاْ": 34,
+  "صَيْ": 227,
+  "صَيْفْ": 274,
+  "صُ": 14,
+  "صُبْ": 450,
+  "صُدْ": 391,
+  "صُلْ": 196,
+  "صُوْ": 386,
+  "صِ": 189,
+  "صِدْ": 394,
+  "صِرْ": 372,
+  "صِيْ": 617,
+  "صِيْنْ": 649,
+  "صِيْهْ": 336,
+  "صِۉْ": 814,
+  "ضَ": 576,
+  "ضَاْ": 157,
+  "ضَاْلْ": 599,
+  "ضَبْ": 784,
+  "ضَرْ": 509,
+  "ضَعْ": 307,
+  "ضَلْ": 749,
+  "ضُ": 598,
+  "ضُرْ": 652,
+  "ضُلْ": 532,
+  "ضُوْ": 573,
+  "ضُوْنْ": 550,
+  "ضِ": 813,
+  "ضِلْ": 763,
+  "ضِنْ": 950,
+  "ضِيْ": 919,
+  "طَ": 830,
+  "طَاْ": 934,
+  "طَاْمْ": 728,
+  "طَاْنْ": 156,
+  "طَبْ": 360,
+  "طَتْ": 678,
+  "طَغْ": 487,
+  "طَفْ": 124,
+  "طَلْ": 899,
+  "طَهْ": 1013,
+  "طَوْ": 741,
+  "طَيْ": 241,
+  "طُ": 931,
+  "طُغْ": 848,
+  "طُوْ": 967,
+  "طِ": 989,
+  "طِرْ": 777,
+  "طِعْ": 294,
+  "طِلْ": 1051,
+  "طِيْ": 827,
+  "طِيْنْ": 265,
+  "ظَ": 1083,
+  "ظَاْ": 183,
+  "ظَنْ": 759,
+  "ظَهْ": 687,
+  "ظُ": 633,
+  "ظُرْ": 316,
+  "ظُلْ": 1066,
+  "ظُنْ": 244,
+  "ظُوْ": 520,
+  "ظِ": 985,
+  "ظِلْ": 1035,
+  "ظِيْمْ": 1,
+  "ظِيْنْ": 984,
+  "عَ": 57,
+  "عَاْ": 275,
+  "عَاْدْ": 125,
+  "عَبْ": 662,
+  "عَتْ": 141,
+  "عَجْ": 978,
+  "عَدْ": 258,
+  "عَذْ": 382,
+  "عَرْ": 565,
+  "عَرْشْ": 350,
+  "عَزْ": 478,
+  "عَسْ": 545,
+  "عَشْرْ": 501,
+  "عَصْ": 750,
+  "عَصْرْ": 476,
+  "عَقْ": 1077,
+  "عَلْ": 305,
+  "عَمْ": 826,
+  "عَنْ": 883,
+  "عَنْهْ": 488,
+  "عَهْ": 311,
+  "عَوْ": 951,
+  "عَيْ": 395,
+  "عَۉْ": 703,
+  "عُ": 499,
+  "عُزْ": 438,
+  "عُسْ": 535,
+  "عُصْ": 600,
+  "عُطْ": 163,
+  "عُقْ": 526,
+  "عُلْ": 59,
+  "عُمْ": 1057,
+  "عُنْ": 588,
+  "عُوْ": 193,
+  "عُوْدْ": 12,
+  "عُوْنْ": 527,
+  "عُۉْ": 56,
+  "عِ": 224,
+  "عِبْ": 470,
+  "عِزْ": 886,
+  "عِصْ": 292,
+  "عِلْ": 186,
+  "عِنْ": 442,
+  "عِهْ": 1042,
+  "عِيْ": 626,
+  "عِيْدْ": 48,
+  "عِيْمْ": 755,
+  "عِيْنْ": 94,
+  "عِيْهْ": 137,
+  "عِۉْ": 399,
+  "غَ": 1060,
+  "غَاْ": 150,
+  "غَبْ": 346,
+  "غَتْ": 288,
+  "غَرْ": 118,
+  "غَسْ": 611,
+  "غَفْ": 90,
+  "غَوْ": 758,
+  "غَيْ": 909,
+  "غُ": 198,
+  "غُضْ": 236,
+  "غُلْ": 1068,
+  "غِ": 348,
+  "غِهْ": 582,
+  "غِيْ": 1033,
+  "فَ": 778,
+  "فَاْ": 528,
+  "فَاْتْ": 248,
+  "فَاْجْ": 512,
+  "فَاْحْ": 994,
+  "فَاْخْ": 637,
+  "فَاْدْ": 974,
+  "فَاْرْ": 656,
+  "فَاْسْ": 670,
+  "فَاْعْ": 427,
+  "فَاْغْ": 1029,
+  "فَاْنْ": 801,
+  "فَتْ": 421,
+  "فَتْحْ": 318,
+  "فَجْرْ": 900,
+  "فَحْ": 1006,
+  "فَرْ": 15,
+  "فَسْ": 671,
+  "فَصْ": 333,
+  "فَصْلْ": 82,
+  "فَضْ": 694,
+  "فَعْ": 166,
+  "فَؽْ": 1016,
+  "فَقْ": 339,
+  "فَكْ": 496,
+  "فَلْ": 847,
+  "فَمْ": 1034,
+  "فَنْ": 180,
+  "فَهْ": 872,
+  "فَوْ": 500,
+  "فَيْ": 818,
+  "فَۉْ": 1040,
+  "فُ": 956,
+  "فُجْ": 944,
+  "فُرْ": 1058,
+  "فُنْ": 963,
+  "فُوْ": 638,
+  "فُوْشْ": 730,
+  "فُوْظْ": 566,
+  "فُوْنْ": 80,
+  "فِ": 41,
+  "فِدْ": 472,
+  "فِرْ": 610,
+  "فِرْهْ": 953,
+  "فِسْ": 340,
+  "فِصْ": 825,
+  "فِظْ": 943,
+  "فِقْ": 536,
+  "فِلْ": 645,
+  "فِمْ": 171,
+  "فِنْ": 1005,
+  "فِيْ": 256,
+  "فِيْخْ": 971,
+  "فِيْلْ": 192,
+  "فِيْنْ": 408,
+  "فِيْهْ": 287,
+  "قَ": 969,
+  "قَاْ": 602,
+  "قَاْبْ": 521,
+  "قَبْ": 456,
+  "قَبْلْ": 711,
+  "قَتْ": 685,
+  "قَدْ": 149,
+  "قَدْرْ": 577,
+  "قَذْ": 538,
+  "قَرْ": 281,
+  "قَسْ": 89,
+  "قَضْ": 106,
+  "قَقْ": 896,
+  "قَلْ": 959,
+  "قَنْ": 35,
+  "قَهْ": 113,
+  "قَوْ": 84,
+  "قَوْلْ": 997,
+  "قَيْ": 612,
+  "قُ": 642,
+  "قُرْ": 785,
+  "قُصْ": 769,
+  "قُلْ": 40,
+  "قُمْ": 133,
+  "قُوْ": 323,
+  "قُوْدْ": 939,
+  "قُوْلْ": 987,
+  "قُوْمْ": 497,
+  "قُوْنْ": 986,
+  "قُۉْ": 636,
+  "قِ": 291,
+  "قِبْ": 13,
+  "قِسْطْ": 432,
+  "قِلْ": 572,
+  "قِمْ": 297,
+  "قِنْ": 605,
+  "قِيْ": 140,
+  "قِيْرْ": 121,
+  "قِيْمْ": 551,
+  "قِيْنْ": 968,
+  "قِيْهْ": 659,
+  "كَ": 176,
+  "كَءْ": 424,
+  "كَاْ": 266,
+  "كَاْهْ": 234,
+  "كَدْ": 798,
+  "كَذْ": 522,
+  "كَرْ": 117,
+  "كَفْ": 178,
+  "كَلْ": 66,
+  "كَنْ": 228,
+  "كَوْ": 218,
+  "كَيْ": 738,
+  "كُ": 61,
+  "كُبْ": 381,
+  "كُرْ": 28,
+  "كُسْ": 842,
+  "كُفْ": 1031,
+  "كُلْ": 918,
+  "كُمْ": 75,
+  "كُنْ": 879,
+  "كُوْ": 471,
+  "كُوْلْ": 1002,
+  "كُوْنْ": 495,
+  "كُۉْ": 921,
+  "كِ": 836,
+  "كِبْ": 1010,
+  "كِذْ": 42,
+  "كِرْ": 858,
+  "كِنْ": 76,
+  "كِيْ": 349,
+  "كِيْمْ": 361,
+  "كِيْنْ": 416,
+  "لَ": 98,
+  "لَاْ": 884,
+  "لَاْبْ": 498,
+  "لَاْتْ": 434,
+  "لَاْدْ": 328,
+  "لَاْفْ": 603,
+  "لَاْقْ": 68,
+  "لَاْلْ": 998,
+  "لَاْمْ": 64,
+  "لَاْهْ": 282,
+  "لَتْ": 203,
+  "لَدْ": 232,
+  "لَسْ": 999,
+  "لَشْ": 726,
+  "لَضْ": 430,
+  "لَظْ": 300,
+  "لَعْ": 684,
+  "لَغْ": 776,
+  "لَؽْ": 562,
+  "لَقْ": 223,
+  "لَكْ": 720,
+  "لَلْ": 389,
+  "لَمْ": 1021,
+  "لَنْ": 525,
+  "لَهْ": 502,
+  "لَوْ": 276,
+  "لَيْ": 634,
+  "لَيْهْ": 774,
+  "لَۉْ": 569,
+  "لُ": 365,
+  "لُصْ": 1003,
+  "لُؽْ": 905,
+  "لُلْ": 958,
+  "لُمْ": 494,
+  "لُنْ": 631,
+  "لُهْ": 1026,
+  "لُوْ": 792,
+  "لُوْلْ": 544,
+  "لُوْنْ": 112,
+  "لِ": 341,
+  "لِتْ": 273,
+  "لِثْ": 856,
+  "لِحْ": 30,
+  "لِدْ": 542,
+  "لِشْ": 981,
+  "لِطْ": 145,
+  "لِفْ": 483,
+  "لِقْ": 327,
+  "لِلْ": 767,
+  "لِمْ": 834,
+  "لِنْ": 1075,
+  "لِيْ": 513,
+  "لِيْلْ": 445,
+  "لِيْمْ": 235,
+  "لِيْنْ": 875,
+  "مَ": 466,
+  "مَءْ": 27,
+  "مَاْ": 822,
+  "مَاْءْ": 214,
+  "مَاْبْ": 285,
+  "مَاْدْ": 6,
+  "مَاْكْ": 722,
+  "مَاْنْ": 146,
+  "مَبْ": 1009,
+  "مَتْ": 162,
+  "مَجْ": 917,
+  "مَحْ": 614,
+  "مَخْ": 957,
+  "مَدْ": 972,
+  "مَرْ": 433,
+  "مَزْ": 643,
+  "مَسْ": 517,
+  "مَشْ": 873,
+  "مَصْ": 1065,
+  "مَطْ": 567,
+  "مَظْ": 753,
+  "مَعْ": 819,
+  "مَغْ": 358,
+  "مَؽْ": 422,
+  "مَقْ": 54,
+  "مَكْ": 390,
+  "مَلْ": 995,
+  "مَمْ": 650,
+  "مَنْ": 443,
+  "مَهْ": 441,
+  "مَوْ": 843,
+  "مَيْ": 641,
+  "مَۉْ": 457,
+  "مُ": 277,
+  "مُءْ": 448,
+  "مُتْ": 851,
+  "مُثْ": 332,
+  "مُجْ": 585,
+  "مُحْ": 182,
+  "مُخْ": 19,
+  "مُدْ": 217,
+  "مُرْ": 837,
+  "مُسْ": 128,
+  "مُشْ": 492,
+  "مُصْ": 271,
+  "مُطْ": 737,
+  "مُعْ": 929,
+  "مُفْ": 846,
+  "مُقْ": 130,
+  "مُلْ": 1080,
+  "مُمْ": 303,
+  "مُنْ": 810,
+  "مُوْ": 735,
+  "مُوْدْ": 699,
+  "مُوْنْ": 418,
+  "مُۉْ": 396,
+  "مِ": 1081,
+  "مِثْ": 941,
+  "مِدْ": 632,
+  "مِرْ": 110,
+  "مِسْ": 69,
+  "مِسْكْ": 419,
+  "مِظْ": 385,
+  "مِعْ": 486,
+  "مِؽْ": 468,
+  "مِقْ": 402,
+  "مِلْ": 337,
+  "مِمْ": 62,
+  "مِنْ": 789,
+  "مِهْ": 92,
+  "مِيْ": 682,
+  "مِيْدْ": 1032,
+  "مِيْمْ": 123,
+  "مِيْنْ": 806,
+  "مِۉْ": 756,
+  "نَ": 761,
+  "نَاْ": 238,
+  "نَاْءْ": 51,
+  "نَاْتْ": 132,
+  "نَاْسْ": 79,
+  "نَتْ": 665,
+  "نَجْ": 211,
+  "نَخْ": 877,
+  "نَدْ": 222,
+  "نَرْ": 63,
+  "نَزْ": 355,
+  "نَسْ": 173,
+  "نَشْ": 190,
+  "نَصْ": 812,
+  "نَضْ": 164,
+  "نَعْ": 841,
+  "نَؽْ": 1007,
+  "نَفْ": 142,
+  "نَقْ": 46,
+  "نَكْ": 1049,
+  "نَلْ": 111,
+  "نَمْ": 17,
+  "نَنْ": 43,
+  "نَهْ": 322,
+  "نَوْ": 1018,
+  "نَيْ": 477,
+  "نَيْنْ": 326,
+  "نَۉْ": 403,
+  "نُ": 932,
+  "نُخْ": 3,
+  "نُسْ": 158,
+  "نُطْ": 721,
+  "نُؽْ": 937,
+  "نُقْ": 936,
+  "نُلْ": 903,
+  "نُمْ": 264,
+  "نُنْ": 143,
+  "نُهْ": 503,
+  "نُوْ": 251,
+  "نُوْدْ": 797,
+  "نُوْرْ": 949,
+  "نُوْنْ": 242,
+  "نُۉْ": 392,
+  "نِ": 988,
+  "نِتْ": 563,
+  "نِثْ": 901,
+  "نِرْ": 359,
+  "نِسْ": 615,
+  "نِصْ": 47,
+  "نِعْ": 411,
+  "نِؽْ": 748,
+  "نِفْ": 747,
+  "نِلْ": 531,
+  "نِنْ": 960,
+  "نِهْ": 1020,
+  "نِيْ": 253,
+  "نِيْمْ": 384,
+  "نِيْنْ": 338,
+  "نِيْهْ": 718,
+  "نْْ": 760,
+  "هَلْ": 184,
+  "هُ": 177,
+  "هُمْ": 467,
+  "وَ": 454,
+  "وَاْ": 175,
+  "وَاْتْ": 606,
+  "وَاْدْ": 714,
+  "وَاْسْ": 147,
+  "وَاْطْ": 724,
+  "وَاْعْ": 201,
+  "وَاْقْ": 1001,
+  "وَاْمْ": 362,
+  "وَاْنْ": 485,
+  "وَتْ": 345,
+  "وَتْرْ": 930,
+  "وَجْ": 1063,
+  "وَدْ": 782,
+  "وَرْ": 334,
+  "وَزْ": 568,
+  "وَسْ": 863,
+  "وَشْ": 376,
+  "وَصْ": 8,
+  "وَضْ": 170,
+  "وَطْ": 716,
+  "وَعْ": 882,
+  "وَلْ": 584,
+  "وَنْ": 698,
+  "وَهْ": 992,
+  "وَوْ": 136,
+  "وَيْ": 852,
+  "وَۉْ": 138,
+  "وُ": 352,
+  "وُلْ": 1039,
+  "وُمْ": 453,
+  "وُنْ": 379,
+  "وِ": 791,
+  "وِزْ": 683,
+  "وِظْ": 312,
+  "وِلْ": 924,
+  "وِهْ": 491,
+  "وِيْ": 185,
+  "وِيْمْ": 907,
+  "يَ": 160,
+  "يَءْ": 745,
+  "يَاْ": 199,
+  "يَاْمْ": 440,
+  "يَاْنْ": 1027,
+  "يَبْ": 286,
+  "يَتْ": 977,
+  "يَجْ": 860,
+  "يَحْ": 1070,
+  "يَخْ": 624,
+  "يَدْ": 77,
+  "يَذْ": 1076,
+  "يَرْ": 194,
+  "يَزْ": 103,
+  "يَسْ": 1053,
+  "يَسْرْ": 892,
+  "يَشْ": 575,
+  "يَصْ": 571,
+  "يَضْ": 933,
+  "يَطْ": 948,
+  "يَعْ": 743,
+  "يَغْ": 553,
+  "يَفْ": 155,
+  "يَقْ": 195,
+  "يَكْ": 629,
+  "يَلْ": 1050,
+  "يَمْ": 134,
+  "يَنْ": 733,
+  "يَهْ": 86,
+  "يَوْ": 516,
+  "يُ": 510,
+  "يُءْ": 458,
+  "يُبْ": 871,
+  "يُخْ": 388,
+  "يُدْ": 845,
+  "يُرْ": 378,
+  "يُسْ": 942,
+  "يُصْ": 954,
+  "يُعْ": 1052,
+  "يُغْ": 710,
+  "يُفْ": 188,
+  "يُمْ": 591,
+  "يُنْ": 543,
+  "يُوْ": 1008,
+  "يُوْنْ": 135,
+  "يِ": 739,
+  "يِلْ": 840,
+  "يِيْ": 817,
+  "يِيْنْ": 885,
+  "َء": 473,
+  "َاْ": 115,
+  "َب": 661,
+  "ِننْ": 32,
+  "ِييْ": 49,
+  "ھَ": 18,
+  "ھَا": 515,
+  "ھَاْ": 692,
+  "ھَاْدْ": 816,
+  "ھَاْرْ": 533,
+  "ھَبْ": 353,
+  "ھَدْ": 706,
+  "ھَرْ": 712,
+  "ھَزْلْ": 335,
+  "ھَلْ": 655,
+  "ھَنْ": 1054,
+  "ھَيْ": 1082,
+  "ھَۉْ": 680,
+  "ھُ": 431,
+  "ھُدْ": 781,
+  "ھُذْ": 167,
+  "ھُرْ": 729,
+  "ھُسْ": 479,
+  "ھُصْ": 983,
+  "ھُؽْ": 558,
+  "ھُلْ": 799,
+  "ھُمْ": 952,
+  "ھُنْ": 890,
+  "ھُوْدْ": 1041,
+  "ھُوْنْ": 976,
+  "ھِ": 844,
+  "ھِرْ": 619,
+  "ھِلْ": 980,
+  "ھِمْ": 657,
+  "ھِنْ": 866,
+  "ھِيْ": 429,
+  "ھِيْدْ": 715,
+  "ھِيْنْ": 414
+}