qalb14 nopnx pretrained model

config.json

@@ -0,0 +1,529 @@
+{
+  "_name_or_path": "/scratch/ba63/BERT_models/bert-base-arabertv02",
+  "architectures": [
+    "BertForTokenClassificationSingleLabel"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "classifier_dropout": null,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "D*",
+    "1": "DDK*",
+    "2": "DK*",
+    "3": "DKD*",
+    "4": "DKI_[\u0644]K*",
+    "5": "I_[ ]K*",
+    "6": "I_[ ]K*I_[ ]K",
+    "7": "I_[ ]K*I_[ ]KI_[ ]K",
+    "8": "I_[ ]K*R_[\u0623]",
+    "9": "I_[\u0623]K*",
+    "10": "I_[\u0623]K*D",
+    "11": "I_[\u0623\u0628]KD*",
+    "12": "I_[\u0625]K*",
+    "13": "I_[\u0625\u0630]KD*",
+    "14": "I_[\u0625\u0644]KD*",
+    "15": "I_[ \u0627]K*",
+    "16": "I_[\u0627]K*",
+    "17": "I_[\u0627]KDK*",
+    "18": "I_[\u0627\u0644]K*",
+    "19": "I_[\u0628]K*",
+    "20": "I_[\u062a]K*",
+    "21": "I_[\u062f]K*",
+    "22": "I_[\u0631]K*",
+    "23": "I_[ \u0644]K*",
+    "24": "I_[\u0644]K*",
+    "25": "I_[\u0645]K*",
+    "26": "I_[\u0646]K*",
+    "27": "I_[\u0647]K*",
+    "28": "I_[\u0648]K*",
+    "29": "I_[\u064a]K*",
+    "30": "I_[\u064a]KDK*",
+    "31": "K*",
+    "32": "K*A_[\u0623\u0646]",
+    "33": "K*A_[\u0645\u0646]",
+    "34": "K*D",
+    "35": "K*DD",
+    "36": "KDDK*",
+    "37": "K*DK",
+    "38": "KDK*",
+    "39": "K*DKI_[\u0630\u0627]",
+    "40": "K*DKK",
+    "41": "K*DKKK",
+    "42": "K*DKKKKKK",
+    "43": "KI_[ ]K*",
+    "44": "K*I_[\u0621]",
+    "45": "KI_[\u0623]K*",
+    "46": "KI_[\u0625]K*",
+    "47": "K*I_[\u0626]K",
+    "48": "K*I_[\u0626]KK",
+    "49": "K*I_[\u0626\u0627]",
+    "50": "K*I_[\u0627]",
+    "51": "K*I_[ \u0627]K",
+    "52": "K*I_[\u0627]K",
+    "53": "KI_[\u0627 ]K*",
+    "54": "K*I_[\u0627]KK",
+    "55": "K*I_[\u0627]KKK",
+    "56": "KI_[\u0627\u0644]K*",
+    "57": "K*I_[\u0627\u0646]",
+    "58": "K*I_[\u0629]",
+    "59": "K*I_[\u062a]",
+    "60": "K*I_[\u062a]K",
+    "61": "K*I_[\u062a]KK",
+    "62": "K*I_[\u062f]K",
+    "63": "K*I_[\u0630]K",
+    "64": "K*I_[\u0630]KD",
+    "65": "KI_[\u0630]KI_[ \u0627]K*",
+    "66": "K*I_[\u0631]",
+    "67": "K*I_[\u0631]K",
+    "68": "K*I_[\u0644]K",
+    "69": "KI_[\u0644]K*",
+    "70": "KI_[\u0644\u0623]K*",
+    "71": "K*I_[\u0644\u0649]",
+    "72": "KI_[\u0644\u0649 ]K*",
+    "73": "K*I_[\u0645]",
+    "74": "K*I_[\u0645]K",
+    "75": "K*I_[\u0645\u0627]",
+    "76": "K*I_[\u0646]",
+    "77": "K*I_[\u0646]K",
+    "78": "K*I_[\u0646]KK",
+    "79": "K*I_[\u0647]",
+    "80": "K*I_[\u0647]K",
+    "81": "K*I_[\u0648]K",
+    "82": "K*I_[\u0648]KK",
+    "83": "K*I_[\u0648\u0627]",
+    "84": "K*I_[\u0648\u0646]",
+    "85": "K*I_[\u0649]",
+    "86": "K*I_[\u064a]",
+    "87": "K*I_[\u064a]K",
+    "88": "K*I_[\u064a]KI_[\u0627]",
+    "89": "K*I_[\u064a]KK",
+    "90": "KKI_[ ]K*",
+    "91": "KKI_[ \u0623]K*",
+    "92": "KKI_[\u0623]K*",
+    "93": "KKI_[\u0625]K*",
+    "94": "KKI_[\u0644]K*",
+    "95": "KKI_[\u0645]K*",
+    "96": "KKKDK*",
+    "97": "KKKI_[ ]K*",
+    "98": "KKKKI_[ ]K*",
+    "99": "KKKR_[\u0623]K*",
+    "100": "KKKR_[\u0625]K*",
+    "101": "KKR_[\u0622]K*",
+    "102": "KKR_[\u0623]K*",
+    "103": "KKR_[\u0625]K*",
+    "104": "KKR_[\u0625]K*R_[\u064a]",
+    "105": "KKR_[\u0627]K*",
+    "106": "K*R_[ ]",
+    "107": "K*R_[\u0621]",
+    "108": "K*R_[\u0621]K",
+    "109": "K*R_[\u0621]KK",
+    "110": "K*R_[\u0622]",
+    "111": "K*R_[\u0622]K",
+    "112": "K*R_[\u0622]KK",
+    "113": "K*R_[\u0623]",
+    "114": "K*R_[\u0623]D",
+    "115": "KR_[\u0623]K*",
+    "116": "K*R_[\u0623]KKKKR_[\u064a]",
+    "117": "K*R_[\u0624]",
+    "118": "K*R_[\u0624]K",
+    "119": "K*R_[\u0624]KK",
+    "120": "K*R_[\u0625]",
+    "121": "KR_[\u0625]K*",
+    "122": "K*R_[\u0626]",
+    "123": "K*R_[\u0626]I_[\u0627]",
+    "124": "K*R_[\u0626]K",
+    "125": "K*R_[\u0627]",
+    "126": "K*R_[\u0627]K",
+    "127": "K*R_[\u0627]KK",
+    "128": "K*R_[\u0629]",
+    "129": "K*R_[\u062a]",
+    "130": "K*R_[\u062a]K",
+    "131": "K*R_[\u062b]",
+    "132": "K*R_[\u062b]K",
+    "133": "K*R_[\u062b]KK",
+    "134": "K*R_[\u062d]",
+    "135": "K*R_[\u062d]K",
+    "136": "K*R_[\u062f]",
+    "137": "K*R_[\u0630]",
+    "138": "K*R_[\u0630]D",
+    "139": "K*R_[\u0630]K",
+    "140": "K*R_[\u0630]KI_[\u0646]",
+    "141": "K*R_[\u0630]KK",
+    "142": "K*R_[\u0630]R_[\u064a]",
+    "143": "K*R_[\u0631]",
+    "144": "K*R_[\u0632]",
+    "145": "K*R_[\u0633]",
+    "146": "K*R_[\u0636]",
+    "147": "K*R_[\u0638]",
+    "148": "K*R_[\u0639]",
+    "149": "K*R_[\u0641]",
+    "150": "K*R_[\u0642]",
+    "151": "K*R_[\u0642]K",
+    "152": "K*R_[\u0643]",
+    "153": "K*R_[\u0643]K",
+    "154": "K*R_[\u0644]",
+    "155": "K*R_[\u0645]",
+    "156": "K*R_[\u0646]",
+    "157": "KR_[\u0646]K*",
+    "158": "K*R_[\u0647]",
+    "159": "K*R_[\u0648]",
+    "160": "K*R_[\u0648]K",
+    "161": "K*R_[\u0649]",
+    "162": "K*R_[\u064a]",
+    "163": "K*R_[\u064a]I_[\u0621]",
+    "164": "K*R_[\u064a]I_[\u0627]",
+    "165": "K*R_[\u064a]K",
+    "166": "MD*",
+    "167": "MDK*",
+    "168": "MK*",
+    "169": "MK*D",
+    "170": "MK*I_[\u0627]",
+    "171": "MKKR_[\u0623]K*",
+    "172": "MKKR_[\u0625]K*",
+    "173": "MK*R_[\u0623]K",
+    "174": "MK*R_[\u064a]",
+    "175": "MR_[\u0622]K*",
+    "176": "MR_[\u0623]K*",
+    "177": "MR_[\u0625]K*",
+    "178": "MR_[\u0627]K*",
+    "179": "R_[ ]",
+    "180": "R_[0]",
+    "181": "R_[ ]D*",
+    "182": "R_[ ]I_[\u0623]K*",
+    "183": "R_[ ]I_[\u0625]K*",
+    "184": "R_[\u0621]",
+    "185": "R_[\u0621]K*",
+    "186": "R_[\u0622]K*",
+    "187": "R_[\u0623]",
+    "188": "R_[\u0623]D*",
+    "189": "R_[\u0623]DK*",
+    "190": "R_[\u0623]K*",
+    "191": "R_[\u0623]K*D",
+    "192": "R_[\u0623]K*I_[ ]K",
+    "193": "R_[\u0623]K*I_[\u0627]",
+    "194": "R_[\u0623]K*R_[\u0623]",
+    "195": "R_[\u0623]K*R_[\u062b]",
+    "196": "R_[\u0623]K*R_[\u0649]",
+    "197": "R_[\u0623]K*R_[\u064a]",
+    "198": "R_[\u0623]R_[\u064a]",
+    "199": "R_[\u0624]",
+    "200": "R_[\u0624]K*",
+    "201": "R_[\u0625]K*",
+    "202": "R_[\u0625]KI_[ ]K*",
+    "203": "R_[\u0625]K*I_[\u0627]",
+    "204": "R_[\u0625]K*R_[\u0649]",
+    "205": "R_[\u0626]",
+    "206": "R_[\u0626]K*",
+    "207": "R_[\u0627]",
+    "208": "R_[\u0627]K*",
+    "209": "R_[\u0628]K*",
+    "210": "R_[\u0629]",
+    "211": "R_[\u0629]D*",
+    "212": "R_[\u062a]",
+    "213": "R_[\u062a]K*",
+    "214": "R_[\u062b]",
+    "215": "R_[\u062b]K*",
+    "216": "R_[\u062c]K*",
+    "217": "R_[\u062d]K*",
+    "218": "R_[\u062f]",
+    "219": "R_[\u062f]K*",
+    "220": "R_[\u0630]",
+    "221": "R_[\u0630]K*",
+    "222": "R_[\u0633]",
+    "223": "R_[\u0633]K*",
+    "224": "R_[\u0636]",
+    "225": "R_[\u0636]K*",
+    "226": "R_[\u0638]K*",
+    "227": "R_[\u0639]K*",
+    "228": "R_[\u063a]K*",
+    "229": "R_[\u0641]K*",
+    "230": "R_[\u0642]K*",
+    "231": "R_[\u0643]K*",
+    "232": "R_[\u0644]K*",
+    "233": "R_[\u0644]R_[\u0643]I_[\u0646]",
+    "234": "R_[\u0644]R_[\u064a]I_[\u0633]",
+    "235": "R_[\u0645]",
+    "236": "R_[\u0645]D*",
+    "237": "R_[\u0645]K*",
+    "238": "R_[\u0646]",
+    "239": "R_[\u0646]K*",
+    "240": "R_[\u0646]K*D",
+    "241": "R_[\u0647]",
+    "242": "R_[\u0647]K*",
+    "243": "R_[\u0648]",
+    "244": "R_[\u0648]K*",
+    "245": "R_[\u0649]",
+    "246": "R_[\u0649]D*",
+    "247": "R_[\u064a]",
+    "248": "R_[\u064a]I_[\u0621]",
+    "249": "R_[\u064a]K*"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "D*": 0,
+    "DDK*": 1,
+    "DK*": 2,
+    "DKD*": 3,
+    "DKI_[\u0644]K*": 4,
+    "I_[ ]K*": 5,
+    "I_[ ]K*I_[ ]K": 6,
+    "I_[ ]K*I_[ ]KI_[ ]K": 7,
+    "I_[ ]K*R_[\u0623]": 8,
+    "I_[ \u0627]K*": 15,
+    "I_[ \u0644]K*": 23,
+    "I_[\u0623]K*": 9,
+    "I_[\u0623]K*D": 10,
+    "I_[\u0623\u0628]KD*": 11,
+    "I_[\u0625]K*": 12,
+    "I_[\u0625\u0630]KD*": 13,
+    "I_[\u0625\u0644]KD*": 14,
+    "I_[\u0627]K*": 16,
+    "I_[\u0627]KDK*": 17,
+    "I_[\u0627\u0644]K*": 18,
+    "I_[\u0628]K*": 19,
+    "I_[\u062a]K*": 20,
+    "I_[\u062f]K*": 21,
+    "I_[\u0631]K*": 22,
+    "I_[\u0644]K*": 24,
+    "I_[\u0645]K*": 25,
+    "I_[\u0646]K*": 26,
+    "I_[\u0647]K*": 27,
+    "I_[\u0648]K*": 28,
+    "I_[\u064a]K*": 29,
+    "I_[\u064a]KDK*": 30,
+    "K*": 31,
+    "K*A_[\u0623\u0646]": 32,
+    "K*A_[\u0645\u0646]": 33,
+    "K*D": 34,
+    "K*DD": 35,
+    "K*DK": 37,
+    "K*DKI_[\u0630\u0627]": 39,
+    "K*DKK": 40,
+    "K*DKKK": 41,
+    "K*DKKKKKK": 42,
+    "K*I_[ \u0627]K": 51,
+    "K*I_[\u0621]": 44,
+    "K*I_[\u0626]K": 47,
+    "K*I_[\u0626]KK": 48,
+    "K*I_[\u0626\u0627]": 49,
+    "K*I_[\u0627]": 50,
+    "K*I_[\u0627]K": 52,
+    "K*I_[\u0627]KK": 54,
+    "K*I_[\u0627]KKK": 55,
+    "K*I_[\u0627\u0646]": 57,
+    "K*I_[\u0629]": 58,
+    "K*I_[\u062a]": 59,
+    "K*I_[\u062a]K": 60,
+    "K*I_[\u062a]KK": 61,
+    "K*I_[\u062f]K": 62,
+    "K*I_[\u0630]K": 63,
+    "K*I_[\u0630]KD": 64,
+    "K*I_[\u0631]": 66,
+    "K*I_[\u0631]K": 67,
+    "K*I_[\u0644]K": 68,
+    "K*I_[\u0644\u0649]": 71,
+    "K*I_[\u0645]": 73,
+    "K*I_[\u0645]K": 74,
+    "K*I_[\u0645\u0627]": 75,
+    "K*I_[\u0646]": 76,
+    "K*I_[\u0646]K": 77,
+    "K*I_[\u0646]KK": 78,
+    "K*I_[\u0647]": 79,
+    "K*I_[\u0647]K": 80,
+    "K*I_[\u0648]K": 81,
+    "K*I_[\u0648]KK": 82,
+    "K*I_[\u0648\u0627]": 83,
+    "K*I_[\u0648\u0646]": 84,
+    "K*I_[\u0649]": 85,
+    "K*I_[\u064a]": 86,
+    "K*I_[\u064a]K": 87,
+    "K*I_[\u064a]KI_[\u0627]": 88,
+    "K*I_[\u064a]KK": 89,
+    "K*R_[ ]": 106,
+    "K*R_[\u0621]": 107,
+    "K*R_[\u0621]K": 108,
+    "K*R_[\u0621]KK": 109,
+    "K*R_[\u0622]": 110,
+    "K*R_[\u0622]K": 111,
+    "K*R_[\u0622]KK": 112,
+    "K*R_[\u0623]": 113,
+    "K*R_[\u0623]D": 114,
+    "K*R_[\u0623]KKKKR_[\u064a]": 116,
+    "K*R_[\u0624]": 117,
+    "K*R_[\u0624]K": 118,
+    "K*R_[\u0624]KK": 119,
+    "K*R_[\u0625]": 120,
+    "K*R_[\u0626]": 122,
+    "K*R_[\u0626]I_[\u0627]": 123,
+    "K*R_[\u0626]K": 124,
+    "K*R_[\u0627]": 125,
+    "K*R_[\u0627]K": 126,
+    "K*R_[\u0627]KK": 127,
+    "K*R_[\u0629]": 128,
+    "K*R_[\u062a]": 129,
+    "K*R_[\u062a]K": 130,
+    "K*R_[\u062b]": 131,
+    "K*R_[\u062b]K": 132,
+    "K*R_[\u062b]KK": 133,
+    "K*R_[\u062d]": 134,
+    "K*R_[\u062d]K": 135,
+    "K*R_[\u062f]": 136,
+    "K*R_[\u0630]": 137,
+    "K*R_[\u0630]D": 138,
+    "K*R_[\u0630]K": 139,
+    "K*R_[\u0630]KI_[\u0646]": 140,
+    "K*R_[\u0630]KK": 141,
+    "K*R_[\u0630]R_[\u064a]": 142,
+    "K*R_[\u0631]": 143,
+    "K*R_[\u0632]": 144,
+    "K*R_[\u0633]": 145,
+    "K*R_[\u0636]": 146,
+    "K*R_[\u0638]": 147,
+    "K*R_[\u0639]": 148,
+    "K*R_[\u0641]": 149,
+    "K*R_[\u0642]": 150,
+    "K*R_[\u0642]K": 151,
+    "K*R_[\u0643]": 152,
+    "K*R_[\u0643]K": 153,
+    "K*R_[\u0644]": 154,
+    "K*R_[\u0645]": 155,
+    "K*R_[\u0646]": 156,
+    "K*R_[\u0647]": 158,
+    "K*R_[\u0648]": 159,
+    "K*R_[\u0648]K": 160,
+    "K*R_[\u0649]": 161,
+    "K*R_[\u064a]": 162,
+    "K*R_[\u064a]I_[\u0621]": 163,
+    "K*R_[\u064a]I_[\u0627]": 164,
+    "K*R_[\u064a]K": 165,
+    "KDDK*": 36,
+    "KDK*": 38,
+    "KI_[ ]K*": 43,
+    "KI_[\u0623]K*": 45,
+    "KI_[\u0625]K*": 46,
+    "KI_[\u0627 ]K*": 53,
+    "KI_[\u0627\u0644]K*": 56,
+    "KI_[\u0630]KI_[ \u0627]K*": 65,
+    "KI_[\u0644]K*": 69,
+    "KI_[\u0644\u0623]K*": 70,
+    "KI_[\u0644\u0649 ]K*": 72,
+    "KKI_[ ]K*": 90,
+    "KKI_[ \u0623]K*": 91,
+    "KKI_[\u0623]K*": 92,
+    "KKI_[\u0625]K*": 93,
+    "KKI_[\u0644]K*": 94,
+    "KKI_[\u0645]K*": 95,
+    "KKKDK*": 96,
+    "KKKI_[ ]K*": 97,
+    "KKKKI_[ ]K*": 98,
+    "KKKR_[\u0623]K*": 99,
+    "KKKR_[\u0625]K*": 100,
+    "KKR_[\u0622]K*": 101,
+    "KKR_[\u0623]K*": 102,
+    "KKR_[\u0625]K*": 103,
+    "KKR_[\u0625]K*R_[\u064a]": 104,
+    "KKR_[\u0627]K*": 105,
+    "KR_[\u0623]K*": 115,
+    "KR_[\u0625]K*": 121,
+    "KR_[\u0646]K*": 157,
+    "MD*": 166,
+    "MDK*": 167,
+    "MK*": 168,
+    "MK*D": 169,
+    "MK*I_[\u0627]": 170,
+    "MK*R_[\u0623]K": 173,
+    "MK*R_[\u064a]": 174,
+    "MKKR_[\u0623]K*": 171,
+    "MKKR_[\u0625]K*": 172,
+    "MR_[\u0622]K*": 175,
+    "MR_[\u0623]K*": 176,
+    "MR_[\u0625]K*": 177,
+    "MR_[\u0627]K*": 178,
+    "R_[ ]": 179,
+    "R_[ ]D*": 181,
+    "R_[ ]I_[\u0623]K*": 182,
+    "R_[ ]I_[\u0625]K*": 183,
+    "R_[0]": 180,
+    "R_[\u0621]": 184,
+    "R_[\u0621]K*": 185,
+    "R_[\u0622]K*": 186,
+    "R_[\u0623]": 187,
+    "R_[\u0623]D*": 188,
+    "R_[\u0623]DK*": 189,
+    "R_[\u0623]K*": 190,
+    "R_[\u0623]K*D": 191,
+    "R_[\u0623]K*I_[ ]K": 192,
+    "R_[\u0623]K*I_[\u0627]": 193,
+    "R_[\u0623]K*R_[\u0623]": 194,
+    "R_[\u0623]K*R_[\u062b]": 195,
+    "R_[\u0623]K*R_[\u0649]": 196,
+    "R_[\u0623]K*R_[\u064a]": 197,
+    "R_[\u0623]R_[\u064a]": 198,
+    "R_[\u0624]": 199,
+    "R_[\u0624]K*": 200,
+    "R_[\u0625]K*": 201,
+    "R_[\u0625]K*I_[\u0627]": 203,
+    "R_[\u0625]K*R_[\u0649]": 204,
+    "R_[\u0625]KI_[ ]K*": 202,
+    "R_[\u0626]": 205,
+    "R_[\u0626]K*": 206,
+    "R_[\u0627]": 207,
+    "R_[\u0627]K*": 208,
+    "R_[\u0628]K*": 209,
+    "R_[\u0629]": 210,
+    "R_[\u0629]D*": 211,
+    "R_[\u062a]": 212,
+    "R_[\u062a]K*": 213,
+    "R_[\u062b]": 214,
+    "R_[\u062b]K*": 215,
+    "R_[\u062c]K*": 216,
+    "R_[\u062d]K*": 217,
+    "R_[\u062f]": 218,
+    "R_[\u062f]K*": 219,
+    "R_[\u0630]": 220,
+    "R_[\u0630]K*": 221,
+    "R_[\u0633]": 222,
+    "R_[\u0633]K*": 223,
+    "R_[\u0636]": 224,
+    "R_[\u0636]K*": 225,
+    "R_[\u0638]K*": 226,
+    "R_[\u0639]K*": 227,
+    "R_[\u063a]K*": 228,
+    "R_[\u0641]K*": 229,
+    "R_[\u0642]K*": 230,
+    "R_[\u0643]K*": 231,
+    "R_[\u0644]K*": 232,
+    "R_[\u0644]R_[\u0643]I_[\u0646]": 233,
+    "R_[\u0644]R_[\u064a]I_[\u0633]": 234,
+    "R_[\u0645]": 235,
+    "R_[\u0645]D*": 236,
+    "R_[\u0645]K*": 237,
+    "R_[\u0646]": 238,
+    "R_[\u0646]K*": 239,
+    "R_[\u0646]K*D": 240,
+    "R_[\u0647]": 241,
+    "R_[\u0647]K*": 242,
+    "R_[\u0648]": 243,
+    "R_[\u0648]K*": 244,
+    "R_[\u0649]": 245,
+    "R_[\u0649]D*": 246,
+    "R_[\u064a]": 247,
+    "R_[\u064a]I_[\u0621]": 248,
+    "R_[\u064a]K*": 249
+  },
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "torch_dtype": "float32",
+  "transformers_version": "4.30.0",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 64000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9324bd3e8bcbfd5a45d5562979cccd4279f0a7946c25a13abb10aba9cc42364d
+size 539250417

special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json

@@ -0,0 +1,20 @@
+{
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": false,
+  "mask_token": "[MASK]",
+  "max_len": 512,
+  "model_max_length": 512,
+  "never_split": [
+    "[بريد]",
+    "[مستخدم]",
+    "[رابط]"
+  ],
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizer",
+  "unk_token": "[UNK]"
+}

trainer_state.json

@@ -0,0 +1,34 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.471169686985173,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5881383855024714e-05,
+      "loss": 0.3825,
+      "step": 500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1762767710049426e-05,
+      "loss": 0.2088,
+      "step": 1000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.764415156507414e-05,
+      "loss": 0.1715,
+      "step": 1500
+    }
+  ],
+  "max_steps": 6070,
+  "num_train_epochs": 10,
+  "total_flos": 2317640762820000.0,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b83058884047283546b5a78935d396e0c0b755b091d032a3dea3242bb85a9ca
+size 4079