{"sent_input_ids": [0, 4950, 21, 50264, 144, 1406, 313, 22610, 13, 39, 1673, 13408, 8, 19255, 43095, 7, 960, 50264, 19737, 2], "target_word_offsets": [14], "def_input_ids": [[0, 627, 1318, 9, 145, 3688, 50, 2021, 7, 10, 3685, 50, 3508, 2], [0, 627, 814, 9, 22761, 13659, 10, 2352, 50, 97, 745, 2, 1, 1]], "mlm_labels": [10, 10767, 37], "mlm_offsets": [3, 14, 17], "ori_target_word_offsets": [14]}
{"sent_input_ids": [0, 27257, 268, 314, 313, 3509, 5, 8429, 174, 1196, 51, 18575, 10, 50264, 8, 50264, 7, 905, 5, 31278, 856, 37865, 66, 2], "target_word_offsets": [22], "def_input_ids": [[0, 415, 50, 7, 41, 253, 2, 1, 1, 1, 1], [0, 1116, 5, 13260, 3064, 50, 23, 63, 3912, 672, 2]], "mlm_labels": [9741, 13069, 6813], "mlm_offsets": [11, 13, 15], "ori_target_word_offsets": [22]}
{"sent_input_ids": [0, 34378, 5, 9280, 254, 16, 45, 2509, 11, 50264, 5, 4590, 50264, 1622, 11, 18999, 5, 3854, 9, 3551, 50264, 11, 5, 5436, 2], "target_word_offsets": [3], "def_input_ids": [[0, 102, 621, 54, 14023, 10, 6441, 7089, 941, 11, 10, 13154, 7, 3094, 402, 2], [0, 102, 621, 54, 5741, 66, 92, 2956, 6448, 50, 1713, 2, 1, 1, 1, 1]], "mlm_labels": [2396, 53, 1836], "mlm_offsets": [9, 12, 20], "ori_target_word_offsets": [3, 4]}
{"sent_input_ids": [0, 8877, 8, 69, 97, 80, 50264, 18779, 32, 314, 7, 8512, 50264, 5, 38384, 13011, 368, 8685, 25774, 2], "target_word_offsets": [15], "def_input_ids": [[0, 5982, 1295, 7, 5, 1280, 9, 1007, 5558, 11, 689, 50, 2423, 2, 1, 1, 1], [0, 1116, 689, 50, 4076, 8200, 171, 12951, 8, 98, 533, 7, 28, 856, 2611, 4226, 2]], "mlm_labels": [2471, 88], "mlm_offsets": [6, 12], "ori_target_word_offsets": [15, 16, 17]}
{"sent_input_ids": [0, 118, 64, 109, 24, 939, 33, 6694, 31263, 20369, 964, 50264, 939, 206, 35243, 115, 120, 899, 7, 951, 50264, 4922, 9473, 6909, 811, 2], "target_word_offsets": [22], "def_input_ids": [[0, 627, 28410, 30133, 811, 2777, 3615, 1330, 7, 8196, 857, 14, 16, 5, 781, 2777, 9, 9473, 6909, 493, 2], [0, 5982, 1295, 7, 9473, 6909, 493, 50, 63, 82, 50, 2777, 2, 1, 1, 1, 1, 1, 1, 1, 1]], "mlm_labels": [8, 939, 54], "mlm_offsets": [11, 14, 20], "ori_target_word_offsets": [22, 23, 24]}
{"sent_input_ids": [0, 627, 92, 28323, 50264, 1684, 92, 38550, 33696, 6700, 6014, 26600, 50264, 201, 7, 50264, 92, 169, 9, 8955, 568, 5349, 2], "target_word_offsets": [9], "def_input_ids": [[0, 9756, 30, 6700, 6014, 2, 1], [0, 241, 13713, 15589, 8, 16887, 2]], "mlm_labels": [1375, 17928, 10], "mlm_offsets": [4, 12, 15], "ori_target_word_offsets": [9, 10]}
{"sent_input_ids": [0, 31636, 5, 50264, 9, 19732, 32, 11956, 14357, 8, 17358, 1130, 6658, 8206, 1925, 1164, 8, 50264, 8243, 2], "target_word_offsets": [12], "def_input_ids": [[0, 627, 11152, 4298, 30, 61, 5, 1925, 12524, 149, 70, 1667, 9, 5, 809, 223, 5, 814, 9, 5, 1144, 2], [0, 30191, 1757, 7, 8, 21016, 50, 198, 402, 941, 14, 9, 12293, 11, 10, 1367, 467, 2, 1, 1, 1, 1]], "mlm_labels": [1795, 17268, 2782], "mlm_offsets": [3, 12, 17], "ori_target_word_offsets": [12]}
{"sent_input_ids": [0, 7877, 6737, 7631, 2053, 9, 50264, 2167, 10746, 50, 9, 50264, 4707, 373, 5360, 77, 2410, 2], "target_word_offsets": [12], "def_input_ids": [[0, 6199, 12073, 9412, 7, 9115, 7, 82, 54, 458, 10, 26293, 50, 14349, 2], [0, 102, 761, 50, 2345, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1]], "mlm_labels": [10, 10], "mlm_offsets": [6, 11], "ori_target_word_offsets": [12]}
{"sent_input_ids": [0, 1264, 9, 127, 50264, 1200, 21, 3219, 10, 50264, 5851, 3407, 14, 56, 57, 1595, 50264, 10, 400, 22283, 2], "target_word_offsets": [9], "def_input_ids": [[0, 5985, 1672, 45, 8218, 2, 1], [0, 37167, 10, 11190, 1683, 15, 2]], "mlm_labels": [656, 1099, 23], "mlm_offsets": [4, 9, 16], "ori_target_word_offsets": [9]}
{"sent_input_ids": [0, 3809, 15673, 2459, 5184, 64, 28, 50264, 358, 86, 10, 30016, 18265, 50264, 10, 14784, 1054, 1310, 19, 741, 12834, 8, 22094, 2], "target_word_offsets": [19], "def_input_ids": [[0, 705, 1023, 2126, 50, 748, 1879, 18583, 9, 2496, 50, 819, 2, 1, 1, 1, 1], [0, 627, 6084, 9, 2866, 2273, 19, 1572, 3501, 15, 50, 41672, 30, 26572, 941, 26140, 2]], "mlm_labels": [6957, 30016, 7448], "mlm_offsets": [7, 11, 13], "ori_target_word_offsets": [19, 20]}
