{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.0, "eval_steps": 500, "global_step": 37592, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00010640561821664184, "grad_norm": 217.875, "learning_rate": 2e-08, "loss": 2.8016, "step": 1 }, { "epoch": 0.00021281123643328368, "grad_norm": 279.75, "learning_rate": 4e-08, "loss": 2.7604, "step": 2 }, { "epoch": 0.0003192168546499255, "grad_norm": 271.75, "learning_rate": 6.000000000000001e-08, "loss": 2.5303, "step": 3 }, { "epoch": 0.00042562247286656737, "grad_norm": 186.0, "learning_rate": 8e-08, "loss": 2.3032, "step": 4 }, { "epoch": 0.0005320280910832092, "grad_norm": 252.25, "learning_rate": 1.0000000000000001e-07, "loss": 2.4197, "step": 5 }, { "epoch": 0.000638433709299851, "grad_norm": 235.75, "learning_rate": 1.2000000000000002e-07, "loss": 2.8088, "step": 6 }, { "epoch": 0.0007448393275164928, "grad_norm": 198.125, "learning_rate": 1.4e-07, "loss": 2.2805, "step": 7 }, { "epoch": 0.0008512449457331347, "grad_norm": 246.375, "learning_rate": 1.6e-07, "loss": 2.7832, "step": 8 }, { "epoch": 0.0009576505639497765, "grad_norm": 336.5, "learning_rate": 1.8e-07, "loss": 2.317, "step": 9 }, { "epoch": 0.0010640561821664183, "grad_norm": 294.25, "learning_rate": 2.0000000000000002e-07, "loss": 1.9491, "step": 10 }, { "epoch": 0.0011704618003830602, "grad_norm": 275.25, "learning_rate": 2.2e-07, "loss": 2.6629, "step": 11 }, { "epoch": 0.001276867418599702, "grad_norm": 193.875, "learning_rate": 2.4000000000000003e-07, "loss": 2.3169, "step": 12 }, { "epoch": 0.001383273036816344, "grad_norm": 254.125, "learning_rate": 2.6e-07, "loss": 2.3077, "step": 13 }, { "epoch": 0.0014896786550329857, "grad_norm": 203.375, "learning_rate": 2.8e-07, "loss": 2.2623, "step": 14 }, { "epoch": 0.0015960842732496276, "grad_norm": 215.625, "learning_rate": 3.0000000000000004e-07, "loss": 2.4327, "step": 15 }, { "epoch": 0.0017024898914662695, "grad_norm": 259.5, "learning_rate": 3.2e-07, "loss": 2.1299, "step": 16 }, { "epoch": 0.0018088955096829114, "grad_norm": 167.375, "learning_rate": 3.4000000000000003e-07, "loss": 1.7368, "step": 17 }, { "epoch": 0.001915301127899553, "grad_norm": 170.125, "learning_rate": 3.6e-07, "loss": 1.5789, "step": 18 }, { "epoch": 0.0020217067461161947, "grad_norm": 158.0, "learning_rate": 3.8e-07, "loss": 1.8761, "step": 19 }, { "epoch": 0.0021281123643328366, "grad_norm": 207.25, "learning_rate": 4.0000000000000003e-07, "loss": 1.9493, "step": 20 }, { "epoch": 0.0022345179825494785, "grad_norm": 137.625, "learning_rate": 4.2000000000000006e-07, "loss": 1.4288, "step": 21 }, { "epoch": 0.0023409236007661204, "grad_norm": 233.375, "learning_rate": 4.4e-07, "loss": 1.3861, "step": 22 }, { "epoch": 0.0024473292189827623, "grad_norm": 95.6875, "learning_rate": 4.6000000000000004e-07, "loss": 1.5125, "step": 23 }, { "epoch": 0.002553734837199404, "grad_norm": 108.4375, "learning_rate": 4.800000000000001e-07, "loss": 1.4798, "step": 24 }, { "epoch": 0.002660140455416046, "grad_norm": 429.25, "learning_rate": 5.000000000000001e-07, "loss": 1.2232, "step": 25 }, { "epoch": 0.002766546073632688, "grad_norm": 1292.0, "learning_rate": 5.2e-07, "loss": 1.5548, "step": 26 }, { "epoch": 0.0028729516918493295, "grad_norm": 1212.0, "learning_rate": 5.4e-07, "loss": 1.5816, "step": 27 }, { "epoch": 0.0029793573100659713, "grad_norm": 575.5, "learning_rate": 5.6e-07, "loss": 1.4877, "step": 28 }, { "epoch": 0.0030857629282826132, "grad_norm": 117.8125, "learning_rate": 5.800000000000001e-07, "loss": 1.4001, "step": 29 }, { "epoch": 0.003192168546499255, "grad_norm": 113.8125, "learning_rate": 6.000000000000001e-07, "loss": 1.1629, "step": 30 }, { "epoch": 0.003298574164715897, "grad_norm": 155.75, "learning_rate": 6.200000000000001e-07, "loss": 1.659, "step": 31 }, { "epoch": 0.003404979782932539, "grad_norm": 112.625, "learning_rate": 6.4e-07, "loss": 1.3308, "step": 32 }, { "epoch": 0.003511385401149181, "grad_norm": 143.25, "learning_rate": 6.6e-07, "loss": 1.6148, "step": 33 }, { "epoch": 0.0036177910193658227, "grad_norm": 145.125, "learning_rate": 6.800000000000001e-07, "loss": 1.3365, "step": 34 }, { "epoch": 0.003724196637582464, "grad_norm": 115.125, "learning_rate": 7.000000000000001e-07, "loss": 1.4027, "step": 35 }, { "epoch": 0.003830602255799106, "grad_norm": 82.5, "learning_rate": 7.2e-07, "loss": 1.2959, "step": 36 }, { "epoch": 0.003937007874015748, "grad_norm": 164.125, "learning_rate": 7.4e-07, "loss": 1.5273, "step": 37 }, { "epoch": 0.0040434134922323894, "grad_norm": 99.0625, "learning_rate": 7.6e-07, "loss": 1.2515, "step": 38 }, { "epoch": 0.004149819110449032, "grad_norm": 77.5625, "learning_rate": 7.8e-07, "loss": 1.2058, "step": 39 }, { "epoch": 0.004256224728665673, "grad_norm": 93.875, "learning_rate": 8.000000000000001e-07, "loss": 1.2608, "step": 40 }, { "epoch": 0.0043626303468823156, "grad_norm": 100.25, "learning_rate": 8.200000000000001e-07, "loss": 1.3275, "step": 41 }, { "epoch": 0.004469035965098957, "grad_norm": 96.375, "learning_rate": 8.400000000000001e-07, "loss": 1.4524, "step": 42 }, { "epoch": 0.004575441583315599, "grad_norm": 148.125, "learning_rate": 8.6e-07, "loss": 1.0003, "step": 43 }, { "epoch": 0.004681847201532241, "grad_norm": 86.4375, "learning_rate": 8.8e-07, "loss": 1.4859, "step": 44 }, { "epoch": 0.004788252819748883, "grad_norm": 69.625, "learning_rate": 9.000000000000001e-07, "loss": 1.2588, "step": 45 }, { "epoch": 0.004894658437965525, "grad_norm": 69.5, "learning_rate": 9.200000000000001e-07, "loss": 1.0362, "step": 46 }, { "epoch": 0.005001064056182166, "grad_norm": 98.5, "learning_rate": 9.400000000000001e-07, "loss": 1.1771, "step": 47 }, { "epoch": 0.005107469674398808, "grad_norm": 103.125, "learning_rate": 9.600000000000001e-07, "loss": 1.2513, "step": 48 }, { "epoch": 0.00521387529261545, "grad_norm": 85.9375, "learning_rate": 9.800000000000001e-07, "loss": 0.9294, "step": 49 }, { "epoch": 0.005320280910832092, "grad_norm": 109.75, "learning_rate": 1.0000000000000002e-06, "loss": 1.2029, "step": 50 }, { "epoch": 0.005426686529048734, "grad_norm": 74.3125, "learning_rate": 1.02e-06, "loss": 1.1044, "step": 51 }, { "epoch": 0.005533092147265376, "grad_norm": 86.875, "learning_rate": 1.04e-06, "loss": 1.0982, "step": 52 }, { "epoch": 0.0056394977654820174, "grad_norm": 73.3125, "learning_rate": 1.06e-06, "loss": 1.0592, "step": 53 }, { "epoch": 0.005745903383698659, "grad_norm": 61.09375, "learning_rate": 1.08e-06, "loss": 1.1059, "step": 54 }, { "epoch": 0.005852309001915301, "grad_norm": 55.21875, "learning_rate": 1.1e-06, "loss": 1.1427, "step": 55 }, { "epoch": 0.005958714620131943, "grad_norm": 49.0625, "learning_rate": 1.12e-06, "loss": 1.0793, "step": 56 }, { "epoch": 0.006065120238348585, "grad_norm": 43.65625, "learning_rate": 1.14e-06, "loss": 1.0856, "step": 57 }, { "epoch": 0.0061715258565652265, "grad_norm": 90.5625, "learning_rate": 1.1600000000000001e-06, "loss": 1.2718, "step": 58 }, { "epoch": 0.006277931474781869, "grad_norm": 58.46875, "learning_rate": 1.1800000000000001e-06, "loss": 1.1729, "step": 59 }, { "epoch": 0.00638433709299851, "grad_norm": 67.125, "learning_rate": 1.2000000000000002e-06, "loss": 1.1585, "step": 60 }, { "epoch": 0.006490742711215152, "grad_norm": 42.71875, "learning_rate": 1.2200000000000002e-06, "loss": 0.9834, "step": 61 }, { "epoch": 0.006597148329431794, "grad_norm": 45.625, "learning_rate": 1.2400000000000002e-06, "loss": 1.0824, "step": 62 }, { "epoch": 0.0067035539476484355, "grad_norm": 37.84375, "learning_rate": 1.26e-06, "loss": 0.902, "step": 63 }, { "epoch": 0.006809959565865078, "grad_norm": 40.90625, "learning_rate": 1.28e-06, "loss": 0.8948, "step": 64 }, { "epoch": 0.006916365184081719, "grad_norm": 54.3125, "learning_rate": 1.3e-06, "loss": 1.0243, "step": 65 }, { "epoch": 0.007022770802298362, "grad_norm": 48.625, "learning_rate": 1.32e-06, "loss": 0.9603, "step": 66 }, { "epoch": 0.007129176420515003, "grad_norm": 45.21875, "learning_rate": 1.34e-06, "loss": 0.8767, "step": 67 }, { "epoch": 0.0072355820387316454, "grad_norm": 47.34375, "learning_rate": 1.3600000000000001e-06, "loss": 1.0112, "step": 68 }, { "epoch": 0.007341987656948287, "grad_norm": 49.4375, "learning_rate": 1.3800000000000001e-06, "loss": 1.0836, "step": 69 }, { "epoch": 0.007448393275164928, "grad_norm": 57.9375, "learning_rate": 1.4000000000000001e-06, "loss": 1.1418, "step": 70 }, { "epoch": 0.007554798893381571, "grad_norm": 53.0625, "learning_rate": 1.42e-06, "loss": 1.3144, "step": 71 }, { "epoch": 0.007661204511598212, "grad_norm": 53.40625, "learning_rate": 1.44e-06, "loss": 1.1071, "step": 72 }, { "epoch": 0.0077676101298148545, "grad_norm": 56.1875, "learning_rate": 1.46e-06, "loss": 1.1562, "step": 73 }, { "epoch": 0.007874015748031496, "grad_norm": 35.59375, "learning_rate": 1.48e-06, "loss": 1.0339, "step": 74 }, { "epoch": 0.007980421366248137, "grad_norm": 49.96875, "learning_rate": 1.5e-06, "loss": 1.1246, "step": 75 }, { "epoch": 0.008086826984464779, "grad_norm": 39.59375, "learning_rate": 1.52e-06, "loss": 0.9934, "step": 76 }, { "epoch": 0.008193232602681422, "grad_norm": 41.21875, "learning_rate": 1.54e-06, "loss": 1.0216, "step": 77 }, { "epoch": 0.008299638220898064, "grad_norm": 38.625, "learning_rate": 1.56e-06, "loss": 1.1186, "step": 78 }, { "epoch": 0.008406043839114705, "grad_norm": 46.21875, "learning_rate": 1.5800000000000001e-06, "loss": 0.9614, "step": 79 }, { "epoch": 0.008512449457331346, "grad_norm": 50.75, "learning_rate": 1.6000000000000001e-06, "loss": 0.8853, "step": 80 }, { "epoch": 0.00861885507554799, "grad_norm": 36.1875, "learning_rate": 1.6200000000000002e-06, "loss": 1.0658, "step": 81 }, { "epoch": 0.008725260693764631, "grad_norm": 43.9375, "learning_rate": 1.6400000000000002e-06, "loss": 1.2128, "step": 82 }, { "epoch": 0.008831666311981273, "grad_norm": 48.53125, "learning_rate": 1.6600000000000002e-06, "loss": 1.1964, "step": 83 }, { "epoch": 0.008938071930197914, "grad_norm": 40.375, "learning_rate": 1.6800000000000002e-06, "loss": 1.137, "step": 84 }, { "epoch": 0.009044477548414556, "grad_norm": 44.96875, "learning_rate": 1.7000000000000002e-06, "loss": 1.047, "step": 85 }, { "epoch": 0.009150883166631199, "grad_norm": 44.25, "learning_rate": 1.72e-06, "loss": 1.1266, "step": 86 }, { "epoch": 0.00925728878484784, "grad_norm": 51.375, "learning_rate": 1.74e-06, "loss": 1.2229, "step": 87 }, { "epoch": 0.009363694403064482, "grad_norm": 46.375, "learning_rate": 1.76e-06, "loss": 1.192, "step": 88 }, { "epoch": 0.009470100021281123, "grad_norm": 38.875, "learning_rate": 1.7800000000000001e-06, "loss": 1.0321, "step": 89 }, { "epoch": 0.009576505639497766, "grad_norm": 51.3125, "learning_rate": 1.8000000000000001e-06, "loss": 1.273, "step": 90 }, { "epoch": 0.009682911257714408, "grad_norm": 34.6875, "learning_rate": 1.8200000000000002e-06, "loss": 1.0895, "step": 91 }, { "epoch": 0.00978931687593105, "grad_norm": 37.71875, "learning_rate": 1.8400000000000002e-06, "loss": 1.0192, "step": 92 }, { "epoch": 0.00989572249414769, "grad_norm": 43.40625, "learning_rate": 1.8600000000000002e-06, "loss": 1.1486, "step": 93 }, { "epoch": 0.010002128112364332, "grad_norm": 41.34375, "learning_rate": 1.8800000000000002e-06, "loss": 1.0028, "step": 94 }, { "epoch": 0.010108533730580975, "grad_norm": 46.96875, "learning_rate": 1.9000000000000002e-06, "loss": 1.0023, "step": 95 }, { "epoch": 0.010214939348797617, "grad_norm": 43.3125, "learning_rate": 1.9200000000000003e-06, "loss": 1.1423, "step": 96 }, { "epoch": 0.010321344967014258, "grad_norm": 46.09375, "learning_rate": 1.94e-06, "loss": 1.3152, "step": 97 }, { "epoch": 0.0104277505852309, "grad_norm": 28.75, "learning_rate": 1.9600000000000003e-06, "loss": 0.8429, "step": 98 }, { "epoch": 0.010534156203447541, "grad_norm": 40.5, "learning_rate": 1.98e-06, "loss": 1.1531, "step": 99 }, { "epoch": 0.010640561821664184, "grad_norm": 30.171875, "learning_rate": 2.0000000000000003e-06, "loss": 1.0075, "step": 100 }, { "epoch": 0.010746967439880826, "grad_norm": 44.4375, "learning_rate": 2.02e-06, "loss": 1.0939, "step": 101 }, { "epoch": 0.010853373058097467, "grad_norm": 35.125, "learning_rate": 2.04e-06, "loss": 1.1037, "step": 102 }, { "epoch": 0.010959778676314109, "grad_norm": 46.3125, "learning_rate": 2.06e-06, "loss": 1.173, "step": 103 }, { "epoch": 0.011066184294530752, "grad_norm": 49.0625, "learning_rate": 2.08e-06, "loss": 1.2266, "step": 104 }, { "epoch": 0.011172589912747393, "grad_norm": 51.875, "learning_rate": 2.1000000000000002e-06, "loss": 1.3426, "step": 105 }, { "epoch": 0.011278995530964035, "grad_norm": 41.0, "learning_rate": 2.12e-06, "loss": 1.1522, "step": 106 }, { "epoch": 0.011385401149180676, "grad_norm": 36.78125, "learning_rate": 2.1400000000000003e-06, "loss": 1.1026, "step": 107 }, { "epoch": 0.011491806767397318, "grad_norm": 34.40625, "learning_rate": 2.16e-06, "loss": 1.0312, "step": 108 }, { "epoch": 0.011598212385613961, "grad_norm": 30.359375, "learning_rate": 2.1800000000000003e-06, "loss": 1.0104, "step": 109 }, { "epoch": 0.011704618003830602, "grad_norm": 28.109375, "learning_rate": 2.2e-06, "loss": 0.9864, "step": 110 }, { "epoch": 0.011811023622047244, "grad_norm": 40.53125, "learning_rate": 2.2200000000000003e-06, "loss": 0.9985, "step": 111 }, { "epoch": 0.011917429240263885, "grad_norm": 35.96875, "learning_rate": 2.24e-06, "loss": 1.1099, "step": 112 }, { "epoch": 0.012023834858480529, "grad_norm": 33.90625, "learning_rate": 2.2600000000000004e-06, "loss": 1.2532, "step": 113 }, { "epoch": 0.01213024047669717, "grad_norm": 33.28125, "learning_rate": 2.28e-06, "loss": 1.2199, "step": 114 }, { "epoch": 0.012236646094913812, "grad_norm": 37.90625, "learning_rate": 2.3000000000000004e-06, "loss": 1.1888, "step": 115 }, { "epoch": 0.012343051713130453, "grad_norm": 40.9375, "learning_rate": 2.3200000000000002e-06, "loss": 1.1067, "step": 116 }, { "epoch": 0.012449457331347094, "grad_norm": 35.78125, "learning_rate": 2.3400000000000005e-06, "loss": 0.9705, "step": 117 }, { "epoch": 0.012555862949563738, "grad_norm": 45.875, "learning_rate": 2.3600000000000003e-06, "loss": 1.2074, "step": 118 }, { "epoch": 0.012662268567780379, "grad_norm": 32.78125, "learning_rate": 2.38e-06, "loss": 1.0997, "step": 119 }, { "epoch": 0.01276867418599702, "grad_norm": 38.09375, "learning_rate": 2.4000000000000003e-06, "loss": 1.1764, "step": 120 }, { "epoch": 0.012875079804213662, "grad_norm": 31.53125, "learning_rate": 2.42e-06, "loss": 1.2386, "step": 121 }, { "epoch": 0.012981485422430303, "grad_norm": 30.96875, "learning_rate": 2.4400000000000004e-06, "loss": 1.0837, "step": 122 }, { "epoch": 0.013087891040646947, "grad_norm": 31.671875, "learning_rate": 2.46e-06, "loss": 1.1191, "step": 123 }, { "epoch": 0.013194296658863588, "grad_norm": 35.84375, "learning_rate": 2.4800000000000004e-06, "loss": 1.0226, "step": 124 }, { "epoch": 0.01330070227708023, "grad_norm": 40.46875, "learning_rate": 2.5e-06, "loss": 1.0574, "step": 125 }, { "epoch": 0.013407107895296871, "grad_norm": 69.9375, "learning_rate": 2.52e-06, "loss": 1.1513, "step": 126 }, { "epoch": 0.013513513513513514, "grad_norm": 48.8125, "learning_rate": 2.5400000000000002e-06, "loss": 1.2457, "step": 127 }, { "epoch": 0.013619919131730156, "grad_norm": 41.0, "learning_rate": 2.56e-06, "loss": 1.1399, "step": 128 }, { "epoch": 0.013726324749946797, "grad_norm": 45.5, "learning_rate": 2.5800000000000003e-06, "loss": 1.0905, "step": 129 }, { "epoch": 0.013832730368163439, "grad_norm": 34.125, "learning_rate": 2.6e-06, "loss": 1.0744, "step": 130 }, { "epoch": 0.01393913598638008, "grad_norm": 43.625, "learning_rate": 2.6200000000000003e-06, "loss": 1.1389, "step": 131 }, { "epoch": 0.014045541604596723, "grad_norm": 39.75, "learning_rate": 2.64e-06, "loss": 1.0887, "step": 132 }, { "epoch": 0.014151947222813365, "grad_norm": 39.3125, "learning_rate": 2.6600000000000004e-06, "loss": 1.1745, "step": 133 }, { "epoch": 0.014258352841030006, "grad_norm": 38.46875, "learning_rate": 2.68e-06, "loss": 1.1475, "step": 134 }, { "epoch": 0.014364758459246648, "grad_norm": 43.0625, "learning_rate": 2.7000000000000004e-06, "loss": 1.0482, "step": 135 }, { "epoch": 0.014471164077463291, "grad_norm": 35.25, "learning_rate": 2.7200000000000002e-06, "loss": 1.1333, "step": 136 }, { "epoch": 0.014577569695679932, "grad_norm": 33.90625, "learning_rate": 2.7400000000000004e-06, "loss": 0.9834, "step": 137 }, { "epoch": 0.014683975313896574, "grad_norm": 33.25, "learning_rate": 2.7600000000000003e-06, "loss": 1.07, "step": 138 }, { "epoch": 0.014790380932113215, "grad_norm": 33.96875, "learning_rate": 2.7800000000000005e-06, "loss": 1.1175, "step": 139 }, { "epoch": 0.014896786550329857, "grad_norm": 34.84375, "learning_rate": 2.8000000000000003e-06, "loss": 1.1235, "step": 140 }, { "epoch": 0.0150031921685465, "grad_norm": 32.1875, "learning_rate": 2.82e-06, "loss": 1.189, "step": 141 }, { "epoch": 0.015109597786763141, "grad_norm": 49.53125, "learning_rate": 2.84e-06, "loss": 1.174, "step": 142 }, { "epoch": 0.015216003404979783, "grad_norm": 31.234375, "learning_rate": 2.86e-06, "loss": 0.9981, "step": 143 }, { "epoch": 0.015322409023196424, "grad_norm": 34.09375, "learning_rate": 2.88e-06, "loss": 1.193, "step": 144 }, { "epoch": 0.015428814641413066, "grad_norm": 39.5625, "learning_rate": 2.9e-06, "loss": 1.318, "step": 145 }, { "epoch": 0.015535220259629709, "grad_norm": 42.1875, "learning_rate": 2.92e-06, "loss": 1.1157, "step": 146 }, { "epoch": 0.01564162587784635, "grad_norm": 40.78125, "learning_rate": 2.9400000000000002e-06, "loss": 1.2061, "step": 147 }, { "epoch": 0.015748031496062992, "grad_norm": 42.28125, "learning_rate": 2.96e-06, "loss": 1.3271, "step": 148 }, { "epoch": 0.015854437114279633, "grad_norm": 30.65625, "learning_rate": 2.9800000000000003e-06, "loss": 0.9684, "step": 149 }, { "epoch": 0.015960842732496275, "grad_norm": 45.71875, "learning_rate": 3e-06, "loss": 1.3668, "step": 150 }, { "epoch": 0.016067248350712916, "grad_norm": 31.484375, "learning_rate": 3.0200000000000003e-06, "loss": 1.1082, "step": 151 }, { "epoch": 0.016173653968929558, "grad_norm": 35.46875, "learning_rate": 3.04e-06, "loss": 0.9942, "step": 152 }, { "epoch": 0.016280059587146203, "grad_norm": 71.8125, "learning_rate": 3.0600000000000003e-06, "loss": 1.2008, "step": 153 }, { "epoch": 0.016386465205362844, "grad_norm": 31.578125, "learning_rate": 3.08e-06, "loss": 1.0739, "step": 154 }, { "epoch": 0.016492870823579486, "grad_norm": 34.75, "learning_rate": 3.1000000000000004e-06, "loss": 1.2326, "step": 155 }, { "epoch": 0.016599276441796127, "grad_norm": 37.21875, "learning_rate": 3.12e-06, "loss": 1.1264, "step": 156 }, { "epoch": 0.01670568206001277, "grad_norm": 35.34375, "learning_rate": 3.1400000000000004e-06, "loss": 1.1467, "step": 157 }, { "epoch": 0.01681208767822941, "grad_norm": 34.6875, "learning_rate": 3.1600000000000002e-06, "loss": 1.0011, "step": 158 }, { "epoch": 0.01691849329644605, "grad_norm": 45.34375, "learning_rate": 3.1800000000000005e-06, "loss": 1.2127, "step": 159 }, { "epoch": 0.017024898914662693, "grad_norm": 33.59375, "learning_rate": 3.2000000000000003e-06, "loss": 1.038, "step": 160 }, { "epoch": 0.017131304532879334, "grad_norm": 43.0625, "learning_rate": 3.2200000000000005e-06, "loss": 1.045, "step": 161 }, { "epoch": 0.01723771015109598, "grad_norm": 45.28125, "learning_rate": 3.2400000000000003e-06, "loss": 1.3999, "step": 162 }, { "epoch": 0.01734411576931262, "grad_norm": 30.828125, "learning_rate": 3.2600000000000006e-06, "loss": 0.8645, "step": 163 }, { "epoch": 0.017450521387529262, "grad_norm": 37.53125, "learning_rate": 3.2800000000000004e-06, "loss": 1.0488, "step": 164 }, { "epoch": 0.017556927005745904, "grad_norm": 44.03125, "learning_rate": 3.3000000000000006e-06, "loss": 1.2719, "step": 165 }, { "epoch": 0.017663332623962545, "grad_norm": 46.09375, "learning_rate": 3.3200000000000004e-06, "loss": 1.2906, "step": 166 }, { "epoch": 0.017769738242179187, "grad_norm": 33.25, "learning_rate": 3.3400000000000006e-06, "loss": 1.2439, "step": 167 }, { "epoch": 0.017876143860395828, "grad_norm": 54.4375, "learning_rate": 3.3600000000000004e-06, "loss": 1.4272, "step": 168 }, { "epoch": 0.01798254947861247, "grad_norm": 28.421875, "learning_rate": 3.3800000000000007e-06, "loss": 1.1083, "step": 169 }, { "epoch": 0.01808895509682911, "grad_norm": 39.0625, "learning_rate": 3.4000000000000005e-06, "loss": 1.1118, "step": 170 }, { "epoch": 0.018195360715045756, "grad_norm": 36.25, "learning_rate": 3.4200000000000007e-06, "loss": 1.1865, "step": 171 }, { "epoch": 0.018301766333262397, "grad_norm": 30.03125, "learning_rate": 3.44e-06, "loss": 1.1444, "step": 172 }, { "epoch": 0.01840817195147904, "grad_norm": 34.28125, "learning_rate": 3.46e-06, "loss": 1.0167, "step": 173 }, { "epoch": 0.01851457756969568, "grad_norm": 40.125, "learning_rate": 3.48e-06, "loss": 1.0784, "step": 174 }, { "epoch": 0.018620983187912322, "grad_norm": 34.21875, "learning_rate": 3.5e-06, "loss": 1.2933, "step": 175 }, { "epoch": 0.018727388806128963, "grad_norm": 33.03125, "learning_rate": 3.52e-06, "loss": 1.0756, "step": 176 }, { "epoch": 0.018833794424345605, "grad_norm": 32.0625, "learning_rate": 3.54e-06, "loss": 1.2022, "step": 177 }, { "epoch": 0.018940200042562246, "grad_norm": 40.8125, "learning_rate": 3.5600000000000002e-06, "loss": 1.2778, "step": 178 }, { "epoch": 0.019046605660778888, "grad_norm": 34.0625, "learning_rate": 3.58e-06, "loss": 1.2003, "step": 179 }, { "epoch": 0.019153011278995533, "grad_norm": 32.78125, "learning_rate": 3.6000000000000003e-06, "loss": 1.0553, "step": 180 }, { "epoch": 0.019259416897212174, "grad_norm": 29.109375, "learning_rate": 3.62e-06, "loss": 0.9573, "step": 181 }, { "epoch": 0.019365822515428815, "grad_norm": 34.1875, "learning_rate": 3.6400000000000003e-06, "loss": 1.3101, "step": 182 }, { "epoch": 0.019472228133645457, "grad_norm": 41.1875, "learning_rate": 3.66e-06, "loss": 1.2495, "step": 183 }, { "epoch": 0.0195786337518621, "grad_norm": 41.6875, "learning_rate": 3.6800000000000003e-06, "loss": 1.356, "step": 184 }, { "epoch": 0.01968503937007874, "grad_norm": 40.1875, "learning_rate": 3.7e-06, "loss": 1.1112, "step": 185 }, { "epoch": 0.01979144498829538, "grad_norm": 28.65625, "learning_rate": 3.7200000000000004e-06, "loss": 1.0268, "step": 186 }, { "epoch": 0.019897850606512023, "grad_norm": 38.9375, "learning_rate": 3.74e-06, "loss": 1.1533, "step": 187 }, { "epoch": 0.020004256224728664, "grad_norm": 32.71875, "learning_rate": 3.7600000000000004e-06, "loss": 1.2592, "step": 188 }, { "epoch": 0.02011066184294531, "grad_norm": 35.71875, "learning_rate": 3.7800000000000002e-06, "loss": 1.1013, "step": 189 }, { "epoch": 0.02021706746116195, "grad_norm": 46.4375, "learning_rate": 3.8000000000000005e-06, "loss": 1.1748, "step": 190 }, { "epoch": 0.020323473079378592, "grad_norm": 40.78125, "learning_rate": 3.820000000000001e-06, "loss": 1.2967, "step": 191 }, { "epoch": 0.020429878697595234, "grad_norm": 38.65625, "learning_rate": 3.8400000000000005e-06, "loss": 1.0512, "step": 192 }, { "epoch": 0.020536284315811875, "grad_norm": 123.375, "learning_rate": 3.86e-06, "loss": 1.3652, "step": 193 }, { "epoch": 0.020642689934028517, "grad_norm": 32.5, "learning_rate": 3.88e-06, "loss": 1.0852, "step": 194 }, { "epoch": 0.020749095552245158, "grad_norm": 34.96875, "learning_rate": 3.900000000000001e-06, "loss": 1.2, "step": 195 }, { "epoch": 0.0208555011704618, "grad_norm": 43.59375, "learning_rate": 3.920000000000001e-06, "loss": 1.147, "step": 196 }, { "epoch": 0.02096190678867844, "grad_norm": 37.34375, "learning_rate": 3.94e-06, "loss": 1.2322, "step": 197 }, { "epoch": 0.021068312406895082, "grad_norm": 55.34375, "learning_rate": 3.96e-06, "loss": 1.0677, "step": 198 }, { "epoch": 0.021174718025111727, "grad_norm": 38.0, "learning_rate": 3.980000000000001e-06, "loss": 1.3412, "step": 199 }, { "epoch": 0.02128112364332837, "grad_norm": 34.15625, "learning_rate": 4.000000000000001e-06, "loss": 1.0167, "step": 200 }, { "epoch": 0.02138752926154501, "grad_norm": 35.21875, "learning_rate": 4.0200000000000005e-06, "loss": 1.158, "step": 201 }, { "epoch": 0.02149393487976165, "grad_norm": 35.9375, "learning_rate": 4.04e-06, "loss": 1.0899, "step": 202 }, { "epoch": 0.021600340497978293, "grad_norm": 26.234375, "learning_rate": 4.060000000000001e-06, "loss": 1.0075, "step": 203 }, { "epoch": 0.021706746116194935, "grad_norm": 61.78125, "learning_rate": 4.08e-06, "loss": 1.6864, "step": 204 }, { "epoch": 0.021813151734411576, "grad_norm": 41.46875, "learning_rate": 4.1e-06, "loss": 1.0284, "step": 205 }, { "epoch": 0.021919557352628218, "grad_norm": 36.3125, "learning_rate": 4.12e-06, "loss": 1.3698, "step": 206 }, { "epoch": 0.02202596297084486, "grad_norm": 29.546875, "learning_rate": 4.14e-06, "loss": 1.0229, "step": 207 }, { "epoch": 0.022132368589061504, "grad_norm": 32.5625, "learning_rate": 4.16e-06, "loss": 1.1029, "step": 208 }, { "epoch": 0.022238774207278145, "grad_norm": 34.15625, "learning_rate": 4.18e-06, "loss": 1.1443, "step": 209 }, { "epoch": 0.022345179825494787, "grad_norm": 43.34375, "learning_rate": 4.2000000000000004e-06, "loss": 1.2837, "step": 210 }, { "epoch": 0.02245158544371143, "grad_norm": 41.21875, "learning_rate": 4.22e-06, "loss": 1.0919, "step": 211 }, { "epoch": 0.02255799106192807, "grad_norm": 35.59375, "learning_rate": 4.24e-06, "loss": 1.0747, "step": 212 }, { "epoch": 0.02266439668014471, "grad_norm": 36.34375, "learning_rate": 4.26e-06, "loss": 1.1968, "step": 213 }, { "epoch": 0.022770802298361353, "grad_norm": 36.9375, "learning_rate": 4.2800000000000005e-06, "loss": 1.2013, "step": 214 }, { "epoch": 0.022877207916577994, "grad_norm": 53.9375, "learning_rate": 4.3e-06, "loss": 1.3406, "step": 215 }, { "epoch": 0.022983613534794636, "grad_norm": 27.546875, "learning_rate": 4.32e-06, "loss": 1.1184, "step": 216 }, { "epoch": 0.02309001915301128, "grad_norm": 31.40625, "learning_rate": 4.34e-06, "loss": 1.0116, "step": 217 }, { "epoch": 0.023196424771227922, "grad_norm": 33.75, "learning_rate": 4.360000000000001e-06, "loss": 1.3051, "step": 218 }, { "epoch": 0.023302830389444563, "grad_norm": 26.421875, "learning_rate": 4.38e-06, "loss": 1.0674, "step": 219 }, { "epoch": 0.023409236007661205, "grad_norm": 51.4375, "learning_rate": 4.4e-06, "loss": 1.1861, "step": 220 }, { "epoch": 0.023515641625877846, "grad_norm": 39.59375, "learning_rate": 4.42e-06, "loss": 1.1446, "step": 221 }, { "epoch": 0.023622047244094488, "grad_norm": 40.78125, "learning_rate": 4.440000000000001e-06, "loss": 1.2334, "step": 222 }, { "epoch": 0.02372845286231113, "grad_norm": 46.4375, "learning_rate": 4.4600000000000005e-06, "loss": 1.441, "step": 223 }, { "epoch": 0.02383485848052777, "grad_norm": 34.78125, "learning_rate": 4.48e-06, "loss": 1.2033, "step": 224 }, { "epoch": 0.023941264098744412, "grad_norm": 78.9375, "learning_rate": 4.5e-06, "loss": 1.1692, "step": 225 }, { "epoch": 0.024047669716961057, "grad_norm": 42.28125, "learning_rate": 4.520000000000001e-06, "loss": 1.245, "step": 226 }, { "epoch": 0.0241540753351777, "grad_norm": 36.65625, "learning_rate": 4.540000000000001e-06, "loss": 1.158, "step": 227 }, { "epoch": 0.02426048095339434, "grad_norm": 30.28125, "learning_rate": 4.56e-06, "loss": 1.0714, "step": 228 }, { "epoch": 0.02436688657161098, "grad_norm": 43.8125, "learning_rate": 4.58e-06, "loss": 1.18, "step": 229 }, { "epoch": 0.024473292189827623, "grad_norm": 32.28125, "learning_rate": 4.600000000000001e-06, "loss": 1.0794, "step": 230 }, { "epoch": 0.024579697808044264, "grad_norm": 94.375, "learning_rate": 4.620000000000001e-06, "loss": 1.1719, "step": 231 }, { "epoch": 0.024686103426260906, "grad_norm": 39.875, "learning_rate": 4.6400000000000005e-06, "loss": 1.2659, "step": 232 }, { "epoch": 0.024792509044477547, "grad_norm": 41.21875, "learning_rate": 4.66e-06, "loss": 1.3043, "step": 233 }, { "epoch": 0.02489891466269419, "grad_norm": 33.3125, "learning_rate": 4.680000000000001e-06, "loss": 1.1475, "step": 234 }, { "epoch": 0.025005320280910834, "grad_norm": 32.0625, "learning_rate": 4.7e-06, "loss": 1.3142, "step": 235 }, { "epoch": 0.025111725899127475, "grad_norm": 32.40625, "learning_rate": 4.7200000000000005e-06, "loss": 1.3048, "step": 236 }, { "epoch": 0.025218131517344117, "grad_norm": 30.703125, "learning_rate": 4.74e-06, "loss": 1.1641, "step": 237 }, { "epoch": 0.025324537135560758, "grad_norm": 34.9375, "learning_rate": 4.76e-06, "loss": 1.2583, "step": 238 }, { "epoch": 0.0254309427537774, "grad_norm": 58.09375, "learning_rate": 4.78e-06, "loss": 1.095, "step": 239 }, { "epoch": 0.02553734837199404, "grad_norm": 34.03125, "learning_rate": 4.800000000000001e-06, "loss": 1.0903, "step": 240 }, { "epoch": 0.025643753990210683, "grad_norm": 33.46875, "learning_rate": 4.8200000000000004e-06, "loss": 1.1792, "step": 241 }, { "epoch": 0.025750159608427324, "grad_norm": 28.3125, "learning_rate": 4.84e-06, "loss": 1.1883, "step": 242 }, { "epoch": 0.025856565226643966, "grad_norm": 79.375, "learning_rate": 4.86e-06, "loss": 1.8195, "step": 243 }, { "epoch": 0.025962970844860607, "grad_norm": 31.109375, "learning_rate": 4.880000000000001e-06, "loss": 1.1313, "step": 244 }, { "epoch": 0.026069376463077252, "grad_norm": 52.1875, "learning_rate": 4.9000000000000005e-06, "loss": 1.2795, "step": 245 }, { "epoch": 0.026175782081293893, "grad_norm": 28.25, "learning_rate": 4.92e-06, "loss": 1.1226, "step": 246 }, { "epoch": 0.026282187699510535, "grad_norm": 27.84375, "learning_rate": 4.94e-06, "loss": 1.187, "step": 247 }, { "epoch": 0.026388593317727176, "grad_norm": 36.40625, "learning_rate": 4.960000000000001e-06, "loss": 1.1796, "step": 248 }, { "epoch": 0.026494998935943818, "grad_norm": 26.65625, "learning_rate": 4.980000000000001e-06, "loss": 0.8373, "step": 249 }, { "epoch": 0.02660140455416046, "grad_norm": 39.90625, "learning_rate": 5e-06, "loss": 1.128, "step": 250 }, { "epoch": 0.0267078101723771, "grad_norm": 29.46875, "learning_rate": 5.02e-06, "loss": 1.1844, "step": 251 }, { "epoch": 0.026814215790593742, "grad_norm": 39.0625, "learning_rate": 5.04e-06, "loss": 1.1786, "step": 252 }, { "epoch": 0.026920621408810384, "grad_norm": 42.59375, "learning_rate": 5.060000000000001e-06, "loss": 1.1748, "step": 253 }, { "epoch": 0.02702702702702703, "grad_norm": 36.0, "learning_rate": 5.0800000000000005e-06, "loss": 1.26, "step": 254 }, { "epoch": 0.02713343264524367, "grad_norm": 37.0, "learning_rate": 5.1e-06, "loss": 1.1022, "step": 255 }, { "epoch": 0.02723983826346031, "grad_norm": 37.84375, "learning_rate": 5.12e-06, "loss": 1.176, "step": 256 }, { "epoch": 0.027346243881676953, "grad_norm": 52.53125, "learning_rate": 5.140000000000001e-06, "loss": 1.2596, "step": 257 }, { "epoch": 0.027452649499893594, "grad_norm": 28.890625, "learning_rate": 5.1600000000000006e-06, "loss": 1.2439, "step": 258 }, { "epoch": 0.027559055118110236, "grad_norm": 35.375, "learning_rate": 5.18e-06, "loss": 1.327, "step": 259 }, { "epoch": 0.027665460736326877, "grad_norm": 33.65625, "learning_rate": 5.2e-06, "loss": 1.1655, "step": 260 }, { "epoch": 0.02777186635454352, "grad_norm": 30.46875, "learning_rate": 5.220000000000001e-06, "loss": 1.0505, "step": 261 }, { "epoch": 0.02787827197276016, "grad_norm": 32.96875, "learning_rate": 5.240000000000001e-06, "loss": 1.285, "step": 262 }, { "epoch": 0.027984677590976805, "grad_norm": 33.25, "learning_rate": 5.2600000000000005e-06, "loss": 1.2466, "step": 263 }, { "epoch": 0.028091083209193447, "grad_norm": 32.28125, "learning_rate": 5.28e-06, "loss": 1.1292, "step": 264 }, { "epoch": 0.028197488827410088, "grad_norm": 27.859375, "learning_rate": 5.300000000000001e-06, "loss": 1.2393, "step": 265 }, { "epoch": 0.02830389444562673, "grad_norm": 28.234375, "learning_rate": 5.320000000000001e-06, "loss": 0.8983, "step": 266 }, { "epoch": 0.02841030006384337, "grad_norm": 46.5625, "learning_rate": 5.3400000000000005e-06, "loss": 1.3457, "step": 267 }, { "epoch": 0.028516705682060012, "grad_norm": 57.40625, "learning_rate": 5.36e-06, "loss": 1.3525, "step": 268 }, { "epoch": 0.028623111300276654, "grad_norm": 35.5625, "learning_rate": 5.380000000000001e-06, "loss": 1.3248, "step": 269 }, { "epoch": 0.028729516918493295, "grad_norm": 32.5625, "learning_rate": 5.400000000000001e-06, "loss": 1.3299, "step": 270 }, { "epoch": 0.028835922536709937, "grad_norm": 63.6875, "learning_rate": 5.420000000000001e-06, "loss": 1.2625, "step": 271 }, { "epoch": 0.028942328154926582, "grad_norm": 46.53125, "learning_rate": 5.4400000000000004e-06, "loss": 1.1816, "step": 272 }, { "epoch": 0.029048733773143223, "grad_norm": 25.5625, "learning_rate": 5.460000000000001e-06, "loss": 1.0493, "step": 273 }, { "epoch": 0.029155139391359865, "grad_norm": 42.96875, "learning_rate": 5.480000000000001e-06, "loss": 1.5041, "step": 274 }, { "epoch": 0.029261545009576506, "grad_norm": 37.28125, "learning_rate": 5.500000000000001e-06, "loss": 1.4553, "step": 275 }, { "epoch": 0.029367950627793148, "grad_norm": 48.21875, "learning_rate": 5.5200000000000005e-06, "loss": 1.6004, "step": 276 }, { "epoch": 0.02947435624600979, "grad_norm": 38.6875, "learning_rate": 5.540000000000001e-06, "loss": 0.7961, "step": 277 }, { "epoch": 0.02958076186422643, "grad_norm": 28.8125, "learning_rate": 5.560000000000001e-06, "loss": 1.1492, "step": 278 }, { "epoch": 0.029687167482443072, "grad_norm": 32.03125, "learning_rate": 5.580000000000001e-06, "loss": 0.8566, "step": 279 }, { "epoch": 0.029793573100659713, "grad_norm": 42.78125, "learning_rate": 5.600000000000001e-06, "loss": 1.4511, "step": 280 }, { "epoch": 0.029899978718876355, "grad_norm": 26.328125, "learning_rate": 5.620000000000001e-06, "loss": 1.1595, "step": 281 }, { "epoch": 0.030006384337093, "grad_norm": 49.78125, "learning_rate": 5.64e-06, "loss": 1.4684, "step": 282 }, { "epoch": 0.03011278995530964, "grad_norm": 37.71875, "learning_rate": 5.66e-06, "loss": 1.1104, "step": 283 }, { "epoch": 0.030219195573526283, "grad_norm": 40.96875, "learning_rate": 5.68e-06, "loss": 1.3375, "step": 284 }, { "epoch": 0.030325601191742924, "grad_norm": 30.609375, "learning_rate": 5.7e-06, "loss": 1.3117, "step": 285 }, { "epoch": 0.030432006809959566, "grad_norm": 30.625, "learning_rate": 5.72e-06, "loss": 1.1615, "step": 286 }, { "epoch": 0.030538412428176207, "grad_norm": 33.0625, "learning_rate": 5.74e-06, "loss": 1.354, "step": 287 }, { "epoch": 0.03064481804639285, "grad_norm": 31.5625, "learning_rate": 5.76e-06, "loss": 1.2317, "step": 288 }, { "epoch": 0.03075122366460949, "grad_norm": 44.1875, "learning_rate": 5.78e-06, "loss": 1.2518, "step": 289 }, { "epoch": 0.03085762928282613, "grad_norm": 34.8125, "learning_rate": 5.8e-06, "loss": 1.2154, "step": 290 }, { "epoch": 0.030964034901042777, "grad_norm": 27.84375, "learning_rate": 5.82e-06, "loss": 1.2811, "step": 291 }, { "epoch": 0.031070440519259418, "grad_norm": 36.78125, "learning_rate": 5.84e-06, "loss": 1.3163, "step": 292 }, { "epoch": 0.03117684613747606, "grad_norm": 26.734375, "learning_rate": 5.86e-06, "loss": 1.4261, "step": 293 }, { "epoch": 0.0312832517556927, "grad_norm": 23.65625, "learning_rate": 5.8800000000000005e-06, "loss": 1.1841, "step": 294 }, { "epoch": 0.03138965737390934, "grad_norm": 41.25, "learning_rate": 5.9e-06, "loss": 1.2797, "step": 295 }, { "epoch": 0.031496062992125984, "grad_norm": 35.25, "learning_rate": 5.92e-06, "loss": 1.5769, "step": 296 }, { "epoch": 0.031602468610342625, "grad_norm": 39.625, "learning_rate": 5.94e-06, "loss": 1.338, "step": 297 }, { "epoch": 0.03170887422855927, "grad_norm": 30.609375, "learning_rate": 5.9600000000000005e-06, "loss": 1.2094, "step": 298 }, { "epoch": 0.03181527984677591, "grad_norm": 29.609375, "learning_rate": 5.98e-06, "loss": 1.1997, "step": 299 }, { "epoch": 0.03192168546499255, "grad_norm": 33.25, "learning_rate": 6e-06, "loss": 1.2673, "step": 300 }, { "epoch": 0.03202809108320919, "grad_norm": 25.78125, "learning_rate": 6.02e-06, "loss": 0.9705, "step": 301 }, { "epoch": 0.03213449670142583, "grad_norm": 30.0, "learning_rate": 6.040000000000001e-06, "loss": 1.211, "step": 302 }, { "epoch": 0.032240902319642474, "grad_norm": 38.28125, "learning_rate": 6.0600000000000004e-06, "loss": 1.3068, "step": 303 }, { "epoch": 0.032347307937859116, "grad_norm": 28.25, "learning_rate": 6.08e-06, "loss": 1.1112, "step": 304 }, { "epoch": 0.032453713556075764, "grad_norm": 29.25, "learning_rate": 6.1e-06, "loss": 1.2844, "step": 305 }, { "epoch": 0.032560119174292405, "grad_norm": 29.859375, "learning_rate": 6.120000000000001e-06, "loss": 1.1866, "step": 306 }, { "epoch": 0.03266652479250905, "grad_norm": 53.03125, "learning_rate": 6.1400000000000005e-06, "loss": 1.3939, "step": 307 }, { "epoch": 0.03277293041072569, "grad_norm": 30.203125, "learning_rate": 6.16e-06, "loss": 1.2808, "step": 308 }, { "epoch": 0.03287933602894233, "grad_norm": 28.1875, "learning_rate": 6.18e-06, "loss": 1.1601, "step": 309 }, { "epoch": 0.03298574164715897, "grad_norm": 49.8125, "learning_rate": 6.200000000000001e-06, "loss": 1.3409, "step": 310 }, { "epoch": 0.03309214726537561, "grad_norm": 33.1875, "learning_rate": 6.220000000000001e-06, "loss": 1.1187, "step": 311 }, { "epoch": 0.033198552883592254, "grad_norm": 33.59375, "learning_rate": 6.24e-06, "loss": 1.3417, "step": 312 }, { "epoch": 0.033304958501808896, "grad_norm": 33.0625, "learning_rate": 6.26e-06, "loss": 1.1964, "step": 313 }, { "epoch": 0.03341136412002554, "grad_norm": 45.71875, "learning_rate": 6.280000000000001e-06, "loss": 1.3543, "step": 314 }, { "epoch": 0.03351776973824218, "grad_norm": 56.53125, "learning_rate": 6.300000000000001e-06, "loss": 1.4905, "step": 315 }, { "epoch": 0.03362417535645882, "grad_norm": 31.015625, "learning_rate": 6.3200000000000005e-06, "loss": 1.2998, "step": 316 }, { "epoch": 0.03373058097467546, "grad_norm": 29.046875, "learning_rate": 6.34e-06, "loss": 1.1899, "step": 317 }, { "epoch": 0.0338369865928921, "grad_norm": 30.859375, "learning_rate": 6.360000000000001e-06, "loss": 1.2473, "step": 318 }, { "epoch": 0.033943392211108744, "grad_norm": 42.59375, "learning_rate": 6.380000000000001e-06, "loss": 1.2474, "step": 319 }, { "epoch": 0.034049797829325386, "grad_norm": 26.359375, "learning_rate": 6.4000000000000006e-06, "loss": 1.2384, "step": 320 }, { "epoch": 0.03415620344754203, "grad_norm": 27.015625, "learning_rate": 6.42e-06, "loss": 1.1415, "step": 321 }, { "epoch": 0.03426260906575867, "grad_norm": 49.40625, "learning_rate": 6.440000000000001e-06, "loss": 1.4829, "step": 322 }, { "epoch": 0.03436901468397532, "grad_norm": 37.9375, "learning_rate": 6.460000000000001e-06, "loss": 1.4949, "step": 323 }, { "epoch": 0.03447542030219196, "grad_norm": 25.59375, "learning_rate": 6.480000000000001e-06, "loss": 1.1264, "step": 324 }, { "epoch": 0.0345818259204086, "grad_norm": 41.125, "learning_rate": 6.5000000000000004e-06, "loss": 1.3274, "step": 325 }, { "epoch": 0.03468823153862524, "grad_norm": 37.3125, "learning_rate": 6.520000000000001e-06, "loss": 1.3048, "step": 326 }, { "epoch": 0.03479463715684188, "grad_norm": 66.5, "learning_rate": 6.540000000000001e-06, "loss": 1.4233, "step": 327 }, { "epoch": 0.034901042775058524, "grad_norm": 34.375, "learning_rate": 6.560000000000001e-06, "loss": 1.188, "step": 328 }, { "epoch": 0.035007448393275166, "grad_norm": 31.828125, "learning_rate": 6.5800000000000005e-06, "loss": 1.162, "step": 329 }, { "epoch": 0.03511385401149181, "grad_norm": 48.71875, "learning_rate": 6.600000000000001e-06, "loss": 1.0027, "step": 330 }, { "epoch": 0.03522025962970845, "grad_norm": 42.3125, "learning_rate": 6.620000000000001e-06, "loss": 1.2834, "step": 331 }, { "epoch": 0.03532666524792509, "grad_norm": 45.71875, "learning_rate": 6.640000000000001e-06, "loss": 1.1746, "step": 332 }, { "epoch": 0.03543307086614173, "grad_norm": 27.03125, "learning_rate": 6.660000000000001e-06, "loss": 1.0456, "step": 333 }, { "epoch": 0.03553947648435837, "grad_norm": 39.25, "learning_rate": 6.680000000000001e-06, "loss": 1.2863, "step": 334 }, { "epoch": 0.035645882102575015, "grad_norm": 37.6875, "learning_rate": 6.700000000000001e-06, "loss": 1.3881, "step": 335 }, { "epoch": 0.035752287720791656, "grad_norm": 24.90625, "learning_rate": 6.720000000000001e-06, "loss": 1.242, "step": 336 }, { "epoch": 0.0358586933390083, "grad_norm": 25.828125, "learning_rate": 6.740000000000001e-06, "loss": 1.1017, "step": 337 }, { "epoch": 0.03596509895722494, "grad_norm": 38.0, "learning_rate": 6.760000000000001e-06, "loss": 1.569, "step": 338 }, { "epoch": 0.03607150457544158, "grad_norm": 39.40625, "learning_rate": 6.780000000000001e-06, "loss": 1.5496, "step": 339 }, { "epoch": 0.03617791019365822, "grad_norm": 39.625, "learning_rate": 6.800000000000001e-06, "loss": 1.2473, "step": 340 }, { "epoch": 0.03628431581187487, "grad_norm": 44.8125, "learning_rate": 6.820000000000001e-06, "loss": 1.4957, "step": 341 }, { "epoch": 0.03639072143009151, "grad_norm": 47.5, "learning_rate": 6.8400000000000014e-06, "loss": 1.5656, "step": 342 }, { "epoch": 0.03649712704830815, "grad_norm": 36.75, "learning_rate": 6.860000000000001e-06, "loss": 1.1753, "step": 343 }, { "epoch": 0.036603532666524795, "grad_norm": 22.703125, "learning_rate": 6.88e-06, "loss": 1.1135, "step": 344 }, { "epoch": 0.036709938284741436, "grad_norm": 33.4375, "learning_rate": 6.9e-06, "loss": 1.2403, "step": 345 }, { "epoch": 0.03681634390295808, "grad_norm": 35.5625, "learning_rate": 6.92e-06, "loss": 1.3015, "step": 346 }, { "epoch": 0.03692274952117472, "grad_norm": 36.28125, "learning_rate": 6.9400000000000005e-06, "loss": 1.3493, "step": 347 }, { "epoch": 0.03702915513939136, "grad_norm": 41.09375, "learning_rate": 6.96e-06, "loss": 1.5737, "step": 348 }, { "epoch": 0.037135560757608, "grad_norm": 32.28125, "learning_rate": 6.98e-06, "loss": 1.4408, "step": 349 }, { "epoch": 0.037241966375824644, "grad_norm": 30.421875, "learning_rate": 7e-06, "loss": 1.1196, "step": 350 }, { "epoch": 0.037348371994041285, "grad_norm": 441.0, "learning_rate": 7.0200000000000006e-06, "loss": 1.3115, "step": 351 }, { "epoch": 0.037454777612257927, "grad_norm": 34.75, "learning_rate": 7.04e-06, "loss": 1.2483, "step": 352 }, { "epoch": 0.03756118323047457, "grad_norm": 36.125, "learning_rate": 7.06e-06, "loss": 1.4159, "step": 353 }, { "epoch": 0.03766758884869121, "grad_norm": 25.328125, "learning_rate": 7.08e-06, "loss": 1.1494, "step": 354 }, { "epoch": 0.03777399446690785, "grad_norm": 34.3125, "learning_rate": 7.100000000000001e-06, "loss": 1.2994, "step": 355 }, { "epoch": 0.03788040008512449, "grad_norm": 34.34375, "learning_rate": 7.1200000000000004e-06, "loss": 1.2431, "step": 356 }, { "epoch": 0.037986805703341134, "grad_norm": 31.546875, "learning_rate": 7.14e-06, "loss": 1.2094, "step": 357 }, { "epoch": 0.038093211321557775, "grad_norm": 31.953125, "learning_rate": 7.16e-06, "loss": 1.4723, "step": 358 }, { "epoch": 0.03819961693977442, "grad_norm": 28.265625, "learning_rate": 7.180000000000001e-06, "loss": 1.2464, "step": 359 }, { "epoch": 0.038306022557991065, "grad_norm": 34.4375, "learning_rate": 7.2000000000000005e-06, "loss": 1.2237, "step": 360 }, { "epoch": 0.03841242817620771, "grad_norm": 36.34375, "learning_rate": 7.22e-06, "loss": 1.3611, "step": 361 }, { "epoch": 0.03851883379442435, "grad_norm": 27.765625, "learning_rate": 7.24e-06, "loss": 1.2782, "step": 362 }, { "epoch": 0.03862523941264099, "grad_norm": 33.03125, "learning_rate": 7.260000000000001e-06, "loss": 1.3643, "step": 363 }, { "epoch": 0.03873164503085763, "grad_norm": 23.4375, "learning_rate": 7.280000000000001e-06, "loss": 1.0345, "step": 364 }, { "epoch": 0.03883805064907427, "grad_norm": 35.40625, "learning_rate": 7.3e-06, "loss": 1.3779, "step": 365 }, { "epoch": 0.038944456267290914, "grad_norm": 28.28125, "learning_rate": 7.32e-06, "loss": 1.2435, "step": 366 }, { "epoch": 0.039050861885507555, "grad_norm": 35.53125, "learning_rate": 7.340000000000001e-06, "loss": 1.3809, "step": 367 }, { "epoch": 0.0391572675037242, "grad_norm": 34.375, "learning_rate": 7.360000000000001e-06, "loss": 1.3895, "step": 368 }, { "epoch": 0.03926367312194084, "grad_norm": 33.40625, "learning_rate": 7.3800000000000005e-06, "loss": 1.4856, "step": 369 }, { "epoch": 0.03937007874015748, "grad_norm": 56.09375, "learning_rate": 7.4e-06, "loss": 1.7977, "step": 370 }, { "epoch": 0.03947648435837412, "grad_norm": 39.8125, "learning_rate": 7.420000000000001e-06, "loss": 1.2058, "step": 371 }, { "epoch": 0.03958288997659076, "grad_norm": 35.84375, "learning_rate": 7.440000000000001e-06, "loss": 1.4748, "step": 372 }, { "epoch": 0.039689295594807404, "grad_norm": 26.46875, "learning_rate": 7.4600000000000006e-06, "loss": 1.1557, "step": 373 }, { "epoch": 0.039795701213024046, "grad_norm": 45.4375, "learning_rate": 7.48e-06, "loss": 1.3351, "step": 374 }, { "epoch": 0.03990210683124069, "grad_norm": 32.5, "learning_rate": 7.500000000000001e-06, "loss": 1.3405, "step": 375 }, { "epoch": 0.04000851244945733, "grad_norm": 43.21875, "learning_rate": 7.520000000000001e-06, "loss": 1.3974, "step": 376 }, { "epoch": 0.04011491806767397, "grad_norm": 42.375, "learning_rate": 7.540000000000001e-06, "loss": 1.7383, "step": 377 }, { "epoch": 0.04022132368589062, "grad_norm": 30.296875, "learning_rate": 7.5600000000000005e-06, "loss": 1.0936, "step": 378 }, { "epoch": 0.04032772930410726, "grad_norm": 28.28125, "learning_rate": 7.58e-06, "loss": 1.3791, "step": 379 }, { "epoch": 0.0404341349223239, "grad_norm": 34.59375, "learning_rate": 7.600000000000001e-06, "loss": 1.4213, "step": 380 }, { "epoch": 0.04054054054054054, "grad_norm": 46.8125, "learning_rate": 7.620000000000001e-06, "loss": 1.5744, "step": 381 }, { "epoch": 0.040646946158757184, "grad_norm": 40.15625, "learning_rate": 7.640000000000001e-06, "loss": 1.1957, "step": 382 }, { "epoch": 0.040753351776973826, "grad_norm": 41.15625, "learning_rate": 7.660000000000001e-06, "loss": 1.2794, "step": 383 }, { "epoch": 0.04085975739519047, "grad_norm": 44.03125, "learning_rate": 7.680000000000001e-06, "loss": 1.4169, "step": 384 }, { "epoch": 0.04096616301340711, "grad_norm": 36.9375, "learning_rate": 7.7e-06, "loss": 1.3865, "step": 385 }, { "epoch": 0.04107256863162375, "grad_norm": 34.8125, "learning_rate": 7.72e-06, "loss": 1.246, "step": 386 }, { "epoch": 0.04117897424984039, "grad_norm": 33.78125, "learning_rate": 7.74e-06, "loss": 1.3247, "step": 387 }, { "epoch": 0.04128537986805703, "grad_norm": 25.796875, "learning_rate": 7.76e-06, "loss": 1.1661, "step": 388 }, { "epoch": 0.041391785486273674, "grad_norm": 25.8125, "learning_rate": 7.78e-06, "loss": 1.2234, "step": 389 }, { "epoch": 0.041498191104490316, "grad_norm": 33.0625, "learning_rate": 7.800000000000002e-06, "loss": 1.4293, "step": 390 }, { "epoch": 0.04160459672270696, "grad_norm": 30.578125, "learning_rate": 7.820000000000001e-06, "loss": 1.0865, "step": 391 }, { "epoch": 0.0417110023409236, "grad_norm": 35.65625, "learning_rate": 7.840000000000001e-06, "loss": 1.3183, "step": 392 }, { "epoch": 0.04181740795914024, "grad_norm": 21.09375, "learning_rate": 7.860000000000001e-06, "loss": 1.0351, "step": 393 }, { "epoch": 0.04192381357735688, "grad_norm": 46.125, "learning_rate": 7.88e-06, "loss": 1.3501, "step": 394 }, { "epoch": 0.04203021919557352, "grad_norm": 33.59375, "learning_rate": 7.9e-06, "loss": 1.2331, "step": 395 }, { "epoch": 0.042136624813790165, "grad_norm": 46.21875, "learning_rate": 7.92e-06, "loss": 1.689, "step": 396 }, { "epoch": 0.04224303043200681, "grad_norm": 29.03125, "learning_rate": 7.94e-06, "loss": 1.3308, "step": 397 }, { "epoch": 0.042349436050223455, "grad_norm": 29.734375, "learning_rate": 7.960000000000002e-06, "loss": 1.2039, "step": 398 }, { "epoch": 0.042455841668440096, "grad_norm": 31.6875, "learning_rate": 7.980000000000002e-06, "loss": 1.4219, "step": 399 }, { "epoch": 0.04256224728665674, "grad_norm": 33.5625, "learning_rate": 8.000000000000001e-06, "loss": 1.3452, "step": 400 }, { "epoch": 0.04266865290487338, "grad_norm": 37.9375, "learning_rate": 8.020000000000001e-06, "loss": 1.5776, "step": 401 }, { "epoch": 0.04277505852309002, "grad_norm": 22.734375, "learning_rate": 8.040000000000001e-06, "loss": 1.252, "step": 402 }, { "epoch": 0.04288146414130666, "grad_norm": 41.25, "learning_rate": 8.06e-06, "loss": 1.3704, "step": 403 }, { "epoch": 0.0429878697595233, "grad_norm": 31.21875, "learning_rate": 8.08e-06, "loss": 1.4542, "step": 404 }, { "epoch": 0.043094275377739945, "grad_norm": 37.625, "learning_rate": 8.1e-06, "loss": 1.3633, "step": 405 }, { "epoch": 0.043200680995956586, "grad_norm": 38.96875, "learning_rate": 8.120000000000002e-06, "loss": 1.2485, "step": 406 }, { "epoch": 0.04330708661417323, "grad_norm": 28.734375, "learning_rate": 8.14e-06, "loss": 1.2083, "step": 407 }, { "epoch": 0.04341349223238987, "grad_norm": 33.28125, "learning_rate": 8.16e-06, "loss": 1.3281, "step": 408 }, { "epoch": 0.04351989785060651, "grad_norm": 40.75, "learning_rate": 8.18e-06, "loss": 1.4183, "step": 409 }, { "epoch": 0.04362630346882315, "grad_norm": 54.0, "learning_rate": 8.2e-06, "loss": 1.312, "step": 410 }, { "epoch": 0.043732709087039794, "grad_norm": 112.625, "learning_rate": 8.220000000000001e-06, "loss": 1.2519, "step": 411 }, { "epoch": 0.043839114705256435, "grad_norm": 64.3125, "learning_rate": 8.24e-06, "loss": 1.4339, "step": 412 }, { "epoch": 0.043945520323473077, "grad_norm": 41.9375, "learning_rate": 8.26e-06, "loss": 1.4654, "step": 413 }, { "epoch": 0.04405192594168972, "grad_norm": 46.21875, "learning_rate": 8.28e-06, "loss": 1.4692, "step": 414 }, { "epoch": 0.044158331559906366, "grad_norm": 35.21875, "learning_rate": 8.3e-06, "loss": 1.5733, "step": 415 }, { "epoch": 0.04426473717812301, "grad_norm": 50.1875, "learning_rate": 8.32e-06, "loss": 1.6371, "step": 416 }, { "epoch": 0.04437114279633965, "grad_norm": 31.25, "learning_rate": 8.34e-06, "loss": 1.4571, "step": 417 }, { "epoch": 0.04447754841455629, "grad_norm": 36.84375, "learning_rate": 8.36e-06, "loss": 1.2516, "step": 418 }, { "epoch": 0.04458395403277293, "grad_norm": 33.5, "learning_rate": 8.380000000000001e-06, "loss": 1.284, "step": 419 }, { "epoch": 0.044690359650989574, "grad_norm": 29.25, "learning_rate": 8.400000000000001e-06, "loss": 1.29, "step": 420 }, { "epoch": 0.044796765269206215, "grad_norm": 21.65625, "learning_rate": 8.42e-06, "loss": 0.971, "step": 421 }, { "epoch": 0.04490317088742286, "grad_norm": 31.671875, "learning_rate": 8.44e-06, "loss": 1.3913, "step": 422 }, { "epoch": 0.0450095765056395, "grad_norm": 29.53125, "learning_rate": 8.46e-06, "loss": 1.1819, "step": 423 }, { "epoch": 0.04511598212385614, "grad_norm": 43.8125, "learning_rate": 8.48e-06, "loss": 1.3508, "step": 424 }, { "epoch": 0.04522238774207278, "grad_norm": 32.09375, "learning_rate": 8.5e-06, "loss": 1.2421, "step": 425 }, { "epoch": 0.04532879336028942, "grad_norm": 38.8125, "learning_rate": 8.52e-06, "loss": 1.5417, "step": 426 }, { "epoch": 0.045435198978506064, "grad_norm": 28.359375, "learning_rate": 8.540000000000001e-06, "loss": 1.1569, "step": 427 }, { "epoch": 0.045541604596722705, "grad_norm": 45.875, "learning_rate": 8.560000000000001e-06, "loss": 1.6301, "step": 428 }, { "epoch": 0.04564801021493935, "grad_norm": 29.90625, "learning_rate": 8.580000000000001e-06, "loss": 1.2376, "step": 429 }, { "epoch": 0.04575441583315599, "grad_norm": 29.15625, "learning_rate": 8.6e-06, "loss": 1.2742, "step": 430 }, { "epoch": 0.04586082145137263, "grad_norm": 45.3125, "learning_rate": 8.62e-06, "loss": 1.0615, "step": 431 }, { "epoch": 0.04596722706958927, "grad_norm": 30.4375, "learning_rate": 8.64e-06, "loss": 1.3197, "step": 432 }, { "epoch": 0.04607363268780591, "grad_norm": 40.9375, "learning_rate": 8.66e-06, "loss": 1.2856, "step": 433 }, { "epoch": 0.04618003830602256, "grad_norm": 34.34375, "learning_rate": 8.68e-06, "loss": 1.5637, "step": 434 }, { "epoch": 0.0462864439242392, "grad_norm": 25.953125, "learning_rate": 8.700000000000001e-06, "loss": 1.3492, "step": 435 }, { "epoch": 0.046392849542455844, "grad_norm": 41.25, "learning_rate": 8.720000000000001e-06, "loss": 1.6266, "step": 436 }, { "epoch": 0.046499255160672485, "grad_norm": 33.84375, "learning_rate": 8.740000000000001e-06, "loss": 1.4227, "step": 437 }, { "epoch": 0.04660566077888913, "grad_norm": 34.75, "learning_rate": 8.76e-06, "loss": 1.5254, "step": 438 }, { "epoch": 0.04671206639710577, "grad_norm": 31.75, "learning_rate": 8.78e-06, "loss": 1.3047, "step": 439 }, { "epoch": 0.04681847201532241, "grad_norm": 38.96875, "learning_rate": 8.8e-06, "loss": 1.3377, "step": 440 }, { "epoch": 0.04692487763353905, "grad_norm": 42.5, "learning_rate": 8.82e-06, "loss": 1.4289, "step": 441 }, { "epoch": 0.04703128325175569, "grad_norm": 38.0625, "learning_rate": 8.84e-06, "loss": 1.2903, "step": 442 }, { "epoch": 0.047137688869972334, "grad_norm": 29.359375, "learning_rate": 8.860000000000002e-06, "loss": 1.4582, "step": 443 }, { "epoch": 0.047244094488188976, "grad_norm": 91.9375, "learning_rate": 8.880000000000001e-06, "loss": 1.2474, "step": 444 }, { "epoch": 0.04735050010640562, "grad_norm": 38.6875, "learning_rate": 8.900000000000001e-06, "loss": 1.3481, "step": 445 }, { "epoch": 0.04745690572462226, "grad_norm": 30.6875, "learning_rate": 8.920000000000001e-06, "loss": 1.2676, "step": 446 }, { "epoch": 0.0475633113428389, "grad_norm": 37.53125, "learning_rate": 8.94e-06, "loss": 1.3977, "step": 447 }, { "epoch": 0.04766971696105554, "grad_norm": 29.015625, "learning_rate": 8.96e-06, "loss": 1.36, "step": 448 }, { "epoch": 0.04777612257927218, "grad_norm": 46.6875, "learning_rate": 8.98e-06, "loss": 1.352, "step": 449 }, { "epoch": 0.047882528197488824, "grad_norm": 24.859375, "learning_rate": 9e-06, "loss": 1.1454, "step": 450 }, { "epoch": 0.047988933815705466, "grad_norm": 27.359375, "learning_rate": 9.020000000000002e-06, "loss": 1.2766, "step": 451 }, { "epoch": 0.048095339433922114, "grad_norm": 30.078125, "learning_rate": 9.040000000000002e-06, "loss": 1.2283, "step": 452 }, { "epoch": 0.048201745052138756, "grad_norm": 38.03125, "learning_rate": 9.060000000000001e-06, "loss": 1.2388, "step": 453 }, { "epoch": 0.0483081506703554, "grad_norm": 29.75, "learning_rate": 9.080000000000001e-06, "loss": 1.3696, "step": 454 }, { "epoch": 0.04841455628857204, "grad_norm": 36.9375, "learning_rate": 9.100000000000001e-06, "loss": 1.3344, "step": 455 }, { "epoch": 0.04852096190678868, "grad_norm": 37.3125, "learning_rate": 9.12e-06, "loss": 1.3719, "step": 456 }, { "epoch": 0.04862736752500532, "grad_norm": 33.34375, "learning_rate": 9.14e-06, "loss": 1.5412, "step": 457 }, { "epoch": 0.04873377314322196, "grad_norm": 26.75, "learning_rate": 9.16e-06, "loss": 1.3732, "step": 458 }, { "epoch": 0.048840178761438605, "grad_norm": 26.703125, "learning_rate": 9.180000000000002e-06, "loss": 1.3914, "step": 459 }, { "epoch": 0.048946584379655246, "grad_norm": 24.578125, "learning_rate": 9.200000000000002e-06, "loss": 1.1454, "step": 460 }, { "epoch": 0.04905298999787189, "grad_norm": 54.40625, "learning_rate": 9.220000000000002e-06, "loss": 1.5422, "step": 461 }, { "epoch": 0.04915939561608853, "grad_norm": 39.875, "learning_rate": 9.240000000000001e-06, "loss": 1.4907, "step": 462 }, { "epoch": 0.04926580123430517, "grad_norm": 42.75, "learning_rate": 9.260000000000001e-06, "loss": 1.3742, "step": 463 }, { "epoch": 0.04937220685252181, "grad_norm": 32.84375, "learning_rate": 9.280000000000001e-06, "loss": 1.3491, "step": 464 }, { "epoch": 0.04947861247073845, "grad_norm": 32.3125, "learning_rate": 9.3e-06, "loss": 1.2809, "step": 465 }, { "epoch": 0.049585018088955095, "grad_norm": 29.375, "learning_rate": 9.32e-06, "loss": 1.2213, "step": 466 }, { "epoch": 0.049691423707171736, "grad_norm": 62.125, "learning_rate": 9.340000000000002e-06, "loss": 1.4383, "step": 467 }, { "epoch": 0.04979782932538838, "grad_norm": 55.21875, "learning_rate": 9.360000000000002e-06, "loss": 1.4533, "step": 468 }, { "epoch": 0.04990423494360502, "grad_norm": 34.96875, "learning_rate": 9.38e-06, "loss": 1.3924, "step": 469 }, { "epoch": 0.05001064056182167, "grad_norm": 37.0, "learning_rate": 9.4e-06, "loss": 1.4822, "step": 470 }, { "epoch": 0.05011704618003831, "grad_norm": 32.9375, "learning_rate": 9.42e-06, "loss": 1.4308, "step": 471 }, { "epoch": 0.05022345179825495, "grad_norm": 58.25, "learning_rate": 9.440000000000001e-06, "loss": 1.6803, "step": 472 }, { "epoch": 0.05032985741647159, "grad_norm": 23.296875, "learning_rate": 9.460000000000001e-06, "loss": 1.1206, "step": 473 }, { "epoch": 0.05043626303468823, "grad_norm": 34.0, "learning_rate": 9.48e-06, "loss": 1.4278, "step": 474 }, { "epoch": 0.050542668652904875, "grad_norm": 28.40625, "learning_rate": 9.5e-06, "loss": 1.1779, "step": 475 }, { "epoch": 0.050649074271121516, "grad_norm": 32.53125, "learning_rate": 9.52e-06, "loss": 1.1973, "step": 476 }, { "epoch": 0.05075547988933816, "grad_norm": 41.9375, "learning_rate": 9.54e-06, "loss": 1.1474, "step": 477 }, { "epoch": 0.0508618855075548, "grad_norm": 38.71875, "learning_rate": 9.56e-06, "loss": 1.7627, "step": 478 }, { "epoch": 0.05096829112577144, "grad_norm": 31.0, "learning_rate": 9.58e-06, "loss": 1.385, "step": 479 }, { "epoch": 0.05107469674398808, "grad_norm": 290.0, "learning_rate": 9.600000000000001e-06, "loss": 1.6375, "step": 480 }, { "epoch": 0.051181102362204724, "grad_norm": 43.75, "learning_rate": 9.620000000000001e-06, "loss": 1.7212, "step": 481 }, { "epoch": 0.051287507980421365, "grad_norm": 26.71875, "learning_rate": 9.640000000000001e-06, "loss": 1.1647, "step": 482 }, { "epoch": 0.05139391359863801, "grad_norm": 20.421875, "learning_rate": 9.66e-06, "loss": 1.0521, "step": 483 }, { "epoch": 0.05150031921685465, "grad_norm": 40.28125, "learning_rate": 9.68e-06, "loss": 1.5331, "step": 484 }, { "epoch": 0.05160672483507129, "grad_norm": 31.3125, "learning_rate": 9.7e-06, "loss": 1.4714, "step": 485 }, { "epoch": 0.05171313045328793, "grad_norm": 30.15625, "learning_rate": 9.72e-06, "loss": 1.6196, "step": 486 }, { "epoch": 0.05181953607150457, "grad_norm": 40.40625, "learning_rate": 9.74e-06, "loss": 1.4817, "step": 487 }, { "epoch": 0.051925941689721214, "grad_norm": 34.46875, "learning_rate": 9.760000000000001e-06, "loss": 1.3812, "step": 488 }, { "epoch": 0.05203234730793786, "grad_norm": 43.75, "learning_rate": 9.780000000000001e-06, "loss": 1.5144, "step": 489 }, { "epoch": 0.052138752926154504, "grad_norm": 27.234375, "learning_rate": 9.800000000000001e-06, "loss": 1.2661, "step": 490 }, { "epoch": 0.052245158544371145, "grad_norm": 24.171875, "learning_rate": 9.820000000000001e-06, "loss": 1.1986, "step": 491 }, { "epoch": 0.05235156416258779, "grad_norm": 49.4375, "learning_rate": 9.84e-06, "loss": 1.6852, "step": 492 }, { "epoch": 0.05245796978080443, "grad_norm": 30.546875, "learning_rate": 9.86e-06, "loss": 1.2993, "step": 493 }, { "epoch": 0.05256437539902107, "grad_norm": 26.46875, "learning_rate": 9.88e-06, "loss": 1.3863, "step": 494 }, { "epoch": 0.05267078101723771, "grad_norm": 27.59375, "learning_rate": 9.9e-06, "loss": 1.4898, "step": 495 }, { "epoch": 0.05277718663545435, "grad_norm": 34.375, "learning_rate": 9.920000000000002e-06, "loss": 1.424, "step": 496 }, { "epoch": 0.052883592253670994, "grad_norm": 42.0625, "learning_rate": 9.940000000000001e-06, "loss": 1.4428, "step": 497 }, { "epoch": 0.052989997871887635, "grad_norm": 37.8125, "learning_rate": 9.960000000000001e-06, "loss": 1.6406, "step": 498 }, { "epoch": 0.05309640349010428, "grad_norm": 28.734375, "learning_rate": 9.980000000000001e-06, "loss": 1.2221, "step": 499 }, { "epoch": 0.05320280910832092, "grad_norm": 39.71875, "learning_rate": 1e-05, "loss": 1.4832, "step": 500 }, { "epoch": 0.05330921472653756, "grad_norm": 46.0, "learning_rate": 9.999964468447982e-06, "loss": 1.5787, "step": 501 }, { "epoch": 0.0534156203447542, "grad_norm": 22.8125, "learning_rate": 9.999928936895964e-06, "loss": 1.3158, "step": 502 }, { "epoch": 0.05352202596297084, "grad_norm": 22.46875, "learning_rate": 9.999893405343947e-06, "loss": 1.2662, "step": 503 }, { "epoch": 0.053628431581187484, "grad_norm": 31.078125, "learning_rate": 9.999857873791929e-06, "loss": 1.4632, "step": 504 }, { "epoch": 0.053734837199404126, "grad_norm": 26.140625, "learning_rate": 9.99982234223991e-06, "loss": 1.2649, "step": 505 }, { "epoch": 0.05384124281762077, "grad_norm": 26.84375, "learning_rate": 9.999786810687892e-06, "loss": 1.4838, "step": 506 }, { "epoch": 0.053947648435837416, "grad_norm": 35.375, "learning_rate": 9.999751279135874e-06, "loss": 1.3301, "step": 507 }, { "epoch": 0.05405405405405406, "grad_norm": 23.59375, "learning_rate": 9.999715747583855e-06, "loss": 1.3532, "step": 508 }, { "epoch": 0.0541604596722707, "grad_norm": 26.265625, "learning_rate": 9.999680216031837e-06, "loss": 1.284, "step": 509 }, { "epoch": 0.05426686529048734, "grad_norm": 23.90625, "learning_rate": 9.999644684479819e-06, "loss": 1.3797, "step": 510 }, { "epoch": 0.05437327090870398, "grad_norm": 26.984375, "learning_rate": 9.9996091529278e-06, "loss": 1.199, "step": 511 }, { "epoch": 0.05447967652692062, "grad_norm": 24.453125, "learning_rate": 9.999573621375782e-06, "loss": 1.3304, "step": 512 }, { "epoch": 0.054586082145137264, "grad_norm": 36.21875, "learning_rate": 9.999538089823765e-06, "loss": 1.5876, "step": 513 }, { "epoch": 0.054692487763353906, "grad_norm": 27.421875, "learning_rate": 9.999502558271747e-06, "loss": 1.3578, "step": 514 }, { "epoch": 0.05479889338157055, "grad_norm": 26.921875, "learning_rate": 9.999467026719728e-06, "loss": 1.3737, "step": 515 }, { "epoch": 0.05490529899978719, "grad_norm": 26.96875, "learning_rate": 9.99943149516771e-06, "loss": 1.3645, "step": 516 }, { "epoch": 0.05501170461800383, "grad_norm": 27.4375, "learning_rate": 9.999395963615692e-06, "loss": 1.2692, "step": 517 }, { "epoch": 0.05511811023622047, "grad_norm": 22.75, "learning_rate": 9.999360432063673e-06, "loss": 1.2351, "step": 518 }, { "epoch": 0.05522451585443711, "grad_norm": 26.296875, "learning_rate": 9.999324900511655e-06, "loss": 1.0731, "step": 519 }, { "epoch": 0.055330921472653755, "grad_norm": 29.796875, "learning_rate": 9.999289368959636e-06, "loss": 1.5842, "step": 520 }, { "epoch": 0.055437327090870396, "grad_norm": 26.3125, "learning_rate": 9.999253837407618e-06, "loss": 1.1833, "step": 521 }, { "epoch": 0.05554373270908704, "grad_norm": 31.28125, "learning_rate": 9.9992183058556e-06, "loss": 1.4906, "step": 522 }, { "epoch": 0.05565013832730368, "grad_norm": 29.28125, "learning_rate": 9.999182774303581e-06, "loss": 1.6333, "step": 523 }, { "epoch": 0.05575654394552032, "grad_norm": 24.640625, "learning_rate": 9.999147242751564e-06, "loss": 1.3134, "step": 524 }, { "epoch": 0.05586294956373696, "grad_norm": 26.625, "learning_rate": 9.999111711199546e-06, "loss": 1.274, "step": 525 }, { "epoch": 0.05596935518195361, "grad_norm": 26.28125, "learning_rate": 9.999076179647528e-06, "loss": 1.5409, "step": 526 }, { "epoch": 0.05607576080017025, "grad_norm": 24.96875, "learning_rate": 9.99904064809551e-06, "loss": 1.1396, "step": 527 }, { "epoch": 0.05618216641838689, "grad_norm": 32.1875, "learning_rate": 9.999005116543491e-06, "loss": 1.1719, "step": 528 }, { "epoch": 0.056288572036603535, "grad_norm": 29.765625, "learning_rate": 9.998969584991474e-06, "loss": 1.4983, "step": 529 }, { "epoch": 0.056394977654820176, "grad_norm": 29.9375, "learning_rate": 9.998934053439456e-06, "loss": 1.5521, "step": 530 }, { "epoch": 0.05650138327303682, "grad_norm": 24.09375, "learning_rate": 9.998898521887437e-06, "loss": 1.3119, "step": 531 }, { "epoch": 0.05660778889125346, "grad_norm": 27.09375, "learning_rate": 9.998862990335419e-06, "loss": 1.3849, "step": 532 }, { "epoch": 0.0567141945094701, "grad_norm": 31.140625, "learning_rate": 9.9988274587834e-06, "loss": 1.4826, "step": 533 }, { "epoch": 0.05682060012768674, "grad_norm": 24.890625, "learning_rate": 9.998791927231382e-06, "loss": 1.4352, "step": 534 }, { "epoch": 0.056927005745903383, "grad_norm": 36.21875, "learning_rate": 9.998756395679364e-06, "loss": 1.3722, "step": 535 }, { "epoch": 0.057033411364120025, "grad_norm": 30.359375, "learning_rate": 9.998720864127345e-06, "loss": 1.2564, "step": 536 }, { "epoch": 0.057139816982336666, "grad_norm": 20.421875, "learning_rate": 9.998685332575327e-06, "loss": 1.0443, "step": 537 }, { "epoch": 0.05724622260055331, "grad_norm": 31.5, "learning_rate": 9.998649801023309e-06, "loss": 1.4506, "step": 538 }, { "epoch": 0.05735262821876995, "grad_norm": 27.125, "learning_rate": 9.998614269471292e-06, "loss": 1.5303, "step": 539 }, { "epoch": 0.05745903383698659, "grad_norm": 47.65625, "learning_rate": 9.998578737919274e-06, "loss": 1.4735, "step": 540 }, { "epoch": 0.05756543945520323, "grad_norm": 33.40625, "learning_rate": 9.998543206367255e-06, "loss": 1.1967, "step": 541 }, { "epoch": 0.057671845073419874, "grad_norm": 51.4375, "learning_rate": 9.998507674815237e-06, "loss": 1.6899, "step": 542 }, { "epoch": 0.057778250691636515, "grad_norm": 29.90625, "learning_rate": 9.998472143263218e-06, "loss": 1.581, "step": 543 }, { "epoch": 0.057884656309853164, "grad_norm": 32.71875, "learning_rate": 9.998436611711202e-06, "loss": 1.3044, "step": 544 }, { "epoch": 0.057991061928069805, "grad_norm": 34.9375, "learning_rate": 9.998401080159182e-06, "loss": 1.4359, "step": 545 }, { "epoch": 0.058097467546286446, "grad_norm": 325.75, "learning_rate": 9.998365548607163e-06, "loss": 1.4573, "step": 546 }, { "epoch": 0.05820387316450309, "grad_norm": 34.0, "learning_rate": 9.998330017055145e-06, "loss": 1.4772, "step": 547 }, { "epoch": 0.05831027878271973, "grad_norm": 25.921875, "learning_rate": 9.998294485503126e-06, "loss": 1.2141, "step": 548 }, { "epoch": 0.05841668440093637, "grad_norm": 32.71875, "learning_rate": 9.99825895395111e-06, "loss": 1.683, "step": 549 }, { "epoch": 0.05852309001915301, "grad_norm": 36.78125, "learning_rate": 9.998223422399091e-06, "loss": 1.2508, "step": 550 }, { "epoch": 0.058629495637369654, "grad_norm": 34.90625, "learning_rate": 9.998187890847073e-06, "loss": 1.4072, "step": 551 }, { "epoch": 0.058735901255586295, "grad_norm": 31.984375, "learning_rate": 9.998152359295055e-06, "loss": 1.5137, "step": 552 }, { "epoch": 0.05884230687380294, "grad_norm": 22.15625, "learning_rate": 9.998116827743036e-06, "loss": 1.2959, "step": 553 }, { "epoch": 0.05894871249201958, "grad_norm": 28.84375, "learning_rate": 9.99808129619102e-06, "loss": 1.4372, "step": 554 }, { "epoch": 0.05905511811023622, "grad_norm": 29.046875, "learning_rate": 9.998045764639001e-06, "loss": 1.4785, "step": 555 }, { "epoch": 0.05916152372845286, "grad_norm": 40.0625, "learning_rate": 9.998010233086983e-06, "loss": 1.4477, "step": 556 }, { "epoch": 0.0592679293466695, "grad_norm": 52.4375, "learning_rate": 9.997974701534964e-06, "loss": 1.5487, "step": 557 }, { "epoch": 0.059374334964886144, "grad_norm": 35.71875, "learning_rate": 9.997939169982946e-06, "loss": 1.4625, "step": 558 }, { "epoch": 0.059480740583102786, "grad_norm": 25.703125, "learning_rate": 9.997903638430927e-06, "loss": 1.205, "step": 559 }, { "epoch": 0.05958714620131943, "grad_norm": 31.828125, "learning_rate": 9.997868106878909e-06, "loss": 1.4677, "step": 560 }, { "epoch": 0.05969355181953607, "grad_norm": 38.1875, "learning_rate": 9.99783257532689e-06, "loss": 1.7671, "step": 561 }, { "epoch": 0.05979995743775271, "grad_norm": 28.8125, "learning_rate": 9.997797043774872e-06, "loss": 1.5186, "step": 562 }, { "epoch": 0.05990636305596936, "grad_norm": 29.5, "learning_rate": 9.997761512222854e-06, "loss": 1.3431, "step": 563 }, { "epoch": 0.060012768674186, "grad_norm": 27.3125, "learning_rate": 9.997725980670836e-06, "loss": 1.3169, "step": 564 }, { "epoch": 0.06011917429240264, "grad_norm": 32.1875, "learning_rate": 9.997690449118819e-06, "loss": 1.5261, "step": 565 }, { "epoch": 0.06022557991061928, "grad_norm": 24.203125, "learning_rate": 9.9976549175668e-06, "loss": 1.2361, "step": 566 }, { "epoch": 0.060331985528835924, "grad_norm": 32.875, "learning_rate": 9.997619386014782e-06, "loss": 1.2461, "step": 567 }, { "epoch": 0.060438391147052566, "grad_norm": 23.34375, "learning_rate": 9.997583854462764e-06, "loss": 1.1176, "step": 568 }, { "epoch": 0.06054479676526921, "grad_norm": 52.28125, "learning_rate": 9.997548322910745e-06, "loss": 1.272, "step": 569 }, { "epoch": 0.06065120238348585, "grad_norm": 23.890625, "learning_rate": 9.997512791358729e-06, "loss": 1.1791, "step": 570 }, { "epoch": 0.06075760800170249, "grad_norm": 43.0625, "learning_rate": 9.99747725980671e-06, "loss": 1.642, "step": 571 }, { "epoch": 0.06086401361991913, "grad_norm": 30.625, "learning_rate": 9.997441728254692e-06, "loss": 1.345, "step": 572 }, { "epoch": 0.06097041923813577, "grad_norm": 32.28125, "learning_rate": 9.997406196702673e-06, "loss": 1.423, "step": 573 }, { "epoch": 0.061076824856352414, "grad_norm": 42.875, "learning_rate": 9.997370665150655e-06, "loss": 1.3086, "step": 574 }, { "epoch": 0.061183230474569056, "grad_norm": 24.265625, "learning_rate": 9.997335133598637e-06, "loss": 1.2333, "step": 575 }, { "epoch": 0.0612896360927857, "grad_norm": 36.65625, "learning_rate": 9.997299602046618e-06, "loss": 1.543, "step": 576 }, { "epoch": 0.06139604171100234, "grad_norm": 36.5, "learning_rate": 9.9972640704946e-06, "loss": 1.447, "step": 577 }, { "epoch": 0.06150244732921898, "grad_norm": 37.78125, "learning_rate": 9.997228538942581e-06, "loss": 1.4985, "step": 578 }, { "epoch": 0.06160885294743562, "grad_norm": 32.53125, "learning_rate": 9.997193007390563e-06, "loss": 1.3425, "step": 579 }, { "epoch": 0.06171525856565226, "grad_norm": 25.25, "learning_rate": 9.997157475838546e-06, "loss": 1.2308, "step": 580 }, { "epoch": 0.06182166418386891, "grad_norm": 30.8125, "learning_rate": 9.997121944286528e-06, "loss": 1.387, "step": 581 }, { "epoch": 0.06192806980208555, "grad_norm": 26.828125, "learning_rate": 9.99708641273451e-06, "loss": 1.372, "step": 582 }, { "epoch": 0.062034475420302194, "grad_norm": 40.625, "learning_rate": 9.997050881182491e-06, "loss": 1.3491, "step": 583 }, { "epoch": 0.062140881038518836, "grad_norm": 23.765625, "learning_rate": 9.997015349630473e-06, "loss": 1.1558, "step": 584 }, { "epoch": 0.06224728665673548, "grad_norm": 38.53125, "learning_rate": 9.996979818078454e-06, "loss": 1.4835, "step": 585 }, { "epoch": 0.06235369227495212, "grad_norm": 52.84375, "learning_rate": 9.996944286526436e-06, "loss": 1.5823, "step": 586 }, { "epoch": 0.06246009789316876, "grad_norm": 48.5625, "learning_rate": 9.996908754974418e-06, "loss": 1.5545, "step": 587 }, { "epoch": 0.0625665035113854, "grad_norm": 31.1875, "learning_rate": 9.996873223422399e-06, "loss": 1.4013, "step": 588 }, { "epoch": 0.06267290912960204, "grad_norm": 28.453125, "learning_rate": 9.99683769187038e-06, "loss": 1.2005, "step": 589 }, { "epoch": 0.06277931474781868, "grad_norm": 30.546875, "learning_rate": 9.996802160318364e-06, "loss": 1.4108, "step": 590 }, { "epoch": 0.06288572036603533, "grad_norm": 24.796875, "learning_rate": 9.996766628766346e-06, "loss": 1.3434, "step": 591 }, { "epoch": 0.06299212598425197, "grad_norm": 42.25, "learning_rate": 9.996731097214327e-06, "loss": 1.5098, "step": 592 }, { "epoch": 0.06309853160246862, "grad_norm": 30.171875, "learning_rate": 9.996695565662309e-06, "loss": 1.491, "step": 593 }, { "epoch": 0.06320493722068525, "grad_norm": 28.234375, "learning_rate": 9.99666003411029e-06, "loss": 1.5396, "step": 594 }, { "epoch": 0.0633113428389019, "grad_norm": 38.53125, "learning_rate": 9.996624502558274e-06, "loss": 1.8894, "step": 595 }, { "epoch": 0.06341774845711853, "grad_norm": 27.234375, "learning_rate": 9.996588971006255e-06, "loss": 1.3639, "step": 596 }, { "epoch": 0.06352415407533518, "grad_norm": 21.78125, "learning_rate": 9.996553439454237e-06, "loss": 1.4124, "step": 597 }, { "epoch": 0.06363055969355182, "grad_norm": 30.203125, "learning_rate": 9.996517907902219e-06, "loss": 1.3813, "step": 598 }, { "epoch": 0.06373696531176846, "grad_norm": 20.0625, "learning_rate": 9.9964823763502e-06, "loss": 1.1859, "step": 599 }, { "epoch": 0.0638433709299851, "grad_norm": 29.046875, "learning_rate": 9.996446844798182e-06, "loss": 1.5373, "step": 600 }, { "epoch": 0.06394977654820175, "grad_norm": 27.609375, "learning_rate": 9.996411313246163e-06, "loss": 1.6044, "step": 601 }, { "epoch": 0.06405618216641838, "grad_norm": 31.125, "learning_rate": 9.996375781694145e-06, "loss": 1.3852, "step": 602 }, { "epoch": 0.06416258778463503, "grad_norm": 32.0625, "learning_rate": 9.996340250142127e-06, "loss": 1.3727, "step": 603 }, { "epoch": 0.06426899340285167, "grad_norm": 29.515625, "learning_rate": 9.996304718590108e-06, "loss": 1.1975, "step": 604 }, { "epoch": 0.06437539902106831, "grad_norm": 22.578125, "learning_rate": 9.99626918703809e-06, "loss": 1.3755, "step": 605 }, { "epoch": 0.06448180463928495, "grad_norm": 20.640625, "learning_rate": 9.996233655486073e-06, "loss": 1.2157, "step": 606 }, { "epoch": 0.0645882102575016, "grad_norm": 30.3125, "learning_rate": 9.996198123934055e-06, "loss": 1.4298, "step": 607 }, { "epoch": 0.06469461587571823, "grad_norm": 19.34375, "learning_rate": 9.996162592382036e-06, "loss": 1.3351, "step": 608 }, { "epoch": 0.06480102149393488, "grad_norm": 27.609375, "learning_rate": 9.996127060830018e-06, "loss": 1.5361, "step": 609 }, { "epoch": 0.06490742711215153, "grad_norm": 21.71875, "learning_rate": 9.996091529278e-06, "loss": 1.1613, "step": 610 }, { "epoch": 0.06501383273036816, "grad_norm": 24.078125, "learning_rate": 9.996055997725981e-06, "loss": 1.5244, "step": 611 }, { "epoch": 0.06512023834858481, "grad_norm": 24.078125, "learning_rate": 9.996020466173963e-06, "loss": 1.3021, "step": 612 }, { "epoch": 0.06522664396680145, "grad_norm": 27.328125, "learning_rate": 9.995984934621944e-06, "loss": 1.4241, "step": 613 }, { "epoch": 0.0653330495850181, "grad_norm": 32.46875, "learning_rate": 9.995949403069926e-06, "loss": 1.6497, "step": 614 }, { "epoch": 0.06543945520323473, "grad_norm": 27.421875, "learning_rate": 9.995913871517908e-06, "loss": 1.4094, "step": 615 }, { "epoch": 0.06554586082145138, "grad_norm": 26.171875, "learning_rate": 9.995878339965891e-06, "loss": 1.3432, "step": 616 }, { "epoch": 0.06565226643966801, "grad_norm": 20.375, "learning_rate": 9.995842808413873e-06, "loss": 1.204, "step": 617 }, { "epoch": 0.06575867205788466, "grad_norm": 33.78125, "learning_rate": 9.995807276861854e-06, "loss": 0.7297, "step": 618 }, { "epoch": 0.0658650776761013, "grad_norm": 19.6875, "learning_rate": 9.995771745309836e-06, "loss": 1.2987, "step": 619 }, { "epoch": 0.06597148329431794, "grad_norm": 33.375, "learning_rate": 9.995736213757817e-06, "loss": 1.5196, "step": 620 }, { "epoch": 0.06607788891253458, "grad_norm": 34.125, "learning_rate": 9.9957006822058e-06, "loss": 1.4697, "step": 621 }, { "epoch": 0.06618429453075123, "grad_norm": 39.4375, "learning_rate": 9.995665150653782e-06, "loss": 1.6948, "step": 622 }, { "epoch": 0.06629070014896786, "grad_norm": 48.75, "learning_rate": 9.995629619101764e-06, "loss": 1.1069, "step": 623 }, { "epoch": 0.06639710576718451, "grad_norm": 25.71875, "learning_rate": 9.995594087549745e-06, "loss": 1.2595, "step": 624 }, { "epoch": 0.06650351138540114, "grad_norm": 31.828125, "learning_rate": 9.995558555997727e-06, "loss": 1.5554, "step": 625 }, { "epoch": 0.06660991700361779, "grad_norm": 28.8125, "learning_rate": 9.995523024445709e-06, "loss": 1.5928, "step": 626 }, { "epoch": 0.06671632262183444, "grad_norm": 28.28125, "learning_rate": 9.99548749289369e-06, "loss": 1.4484, "step": 627 }, { "epoch": 0.06682272824005107, "grad_norm": 34.96875, "learning_rate": 9.995451961341672e-06, "loss": 1.5355, "step": 628 }, { "epoch": 0.06692913385826772, "grad_norm": 32.53125, "learning_rate": 9.995416429789653e-06, "loss": 1.1676, "step": 629 }, { "epoch": 0.06703553947648436, "grad_norm": 38.0625, "learning_rate": 9.995380898237635e-06, "loss": 1.4659, "step": 630 }, { "epoch": 0.067141945094701, "grad_norm": 44.28125, "learning_rate": 9.995345366685618e-06, "loss": 1.7358, "step": 631 }, { "epoch": 0.06724835071291764, "grad_norm": 27.9375, "learning_rate": 9.9953098351336e-06, "loss": 1.5213, "step": 632 }, { "epoch": 0.06735475633113429, "grad_norm": 23.171875, "learning_rate": 9.995274303581582e-06, "loss": 1.2327, "step": 633 }, { "epoch": 0.06746116194935092, "grad_norm": 26.046875, "learning_rate": 9.995238772029563e-06, "loss": 1.2239, "step": 634 }, { "epoch": 0.06756756756756757, "grad_norm": 27.953125, "learning_rate": 9.995203240477545e-06, "loss": 1.6617, "step": 635 }, { "epoch": 0.0676739731857842, "grad_norm": 22.5625, "learning_rate": 9.995167708925526e-06, "loss": 1.2733, "step": 636 }, { "epoch": 0.06778037880400085, "grad_norm": 21.953125, "learning_rate": 9.995132177373508e-06, "loss": 1.5125, "step": 637 }, { "epoch": 0.06788678442221749, "grad_norm": 19.0625, "learning_rate": 9.99509664582149e-06, "loss": 1.2674, "step": 638 }, { "epoch": 0.06799319004043414, "grad_norm": 25.734375, "learning_rate": 9.995061114269471e-06, "loss": 1.1839, "step": 639 }, { "epoch": 0.06809959565865077, "grad_norm": 27.734375, "learning_rate": 9.995025582717453e-06, "loss": 1.6725, "step": 640 }, { "epoch": 0.06820600127686742, "grad_norm": 23.015625, "learning_rate": 9.994990051165434e-06, "loss": 1.2634, "step": 641 }, { "epoch": 0.06831240689508405, "grad_norm": 39.5, "learning_rate": 9.994954519613418e-06, "loss": 1.4905, "step": 642 }, { "epoch": 0.0684188125133007, "grad_norm": 26.859375, "learning_rate": 9.9949189880614e-06, "loss": 1.5183, "step": 643 }, { "epoch": 0.06852521813151734, "grad_norm": 24.09375, "learning_rate": 9.994883456509381e-06, "loss": 1.1225, "step": 644 }, { "epoch": 0.06863162374973399, "grad_norm": 47.34375, "learning_rate": 9.994847924957363e-06, "loss": 1.9422, "step": 645 }, { "epoch": 0.06873802936795063, "grad_norm": 32.125, "learning_rate": 9.994812393405344e-06, "loss": 1.7313, "step": 646 }, { "epoch": 0.06884443498616727, "grad_norm": 26.96875, "learning_rate": 9.994776861853327e-06, "loss": 1.3957, "step": 647 }, { "epoch": 0.06895084060438392, "grad_norm": 25.84375, "learning_rate": 9.994741330301309e-06, "loss": 1.4571, "step": 648 }, { "epoch": 0.06905724622260055, "grad_norm": 19.828125, "learning_rate": 9.99470579874929e-06, "loss": 1.2202, "step": 649 }, { "epoch": 0.0691636518408172, "grad_norm": 26.859375, "learning_rate": 9.994670267197272e-06, "loss": 1.2568, "step": 650 }, { "epoch": 0.06927005745903383, "grad_norm": 29.96875, "learning_rate": 9.994634735645254e-06, "loss": 1.3816, "step": 651 }, { "epoch": 0.06937646307725048, "grad_norm": 39.4375, "learning_rate": 9.994599204093236e-06, "loss": 1.4763, "step": 652 }, { "epoch": 0.06948286869546712, "grad_norm": 29.40625, "learning_rate": 9.994563672541217e-06, "loss": 1.2463, "step": 653 }, { "epoch": 0.06958927431368377, "grad_norm": 25.71875, "learning_rate": 9.994528140989199e-06, "loss": 1.2639, "step": 654 }, { "epoch": 0.0696956799319004, "grad_norm": 26.21875, "learning_rate": 9.99449260943718e-06, "loss": 1.3869, "step": 655 }, { "epoch": 0.06980208555011705, "grad_norm": 22.0625, "learning_rate": 9.994457077885162e-06, "loss": 1.2872, "step": 656 }, { "epoch": 0.06990849116833368, "grad_norm": 26.296875, "learning_rate": 9.994421546333145e-06, "loss": 1.7343, "step": 657 }, { "epoch": 0.07001489678655033, "grad_norm": 21.71875, "learning_rate": 9.994386014781127e-06, "loss": 1.2648, "step": 658 }, { "epoch": 0.07012130240476697, "grad_norm": 28.703125, "learning_rate": 9.994350483229108e-06, "loss": 1.6085, "step": 659 }, { "epoch": 0.07022770802298361, "grad_norm": 30.453125, "learning_rate": 9.99431495167709e-06, "loss": 1.4408, "step": 660 }, { "epoch": 0.07033411364120025, "grad_norm": 82.375, "learning_rate": 9.994279420125072e-06, "loss": 1.2659, "step": 661 }, { "epoch": 0.0704405192594169, "grad_norm": 28.015625, "learning_rate": 9.994243888573055e-06, "loss": 1.413, "step": 662 }, { "epoch": 0.07054692487763353, "grad_norm": 27.875, "learning_rate": 9.994208357021037e-06, "loss": 1.2638, "step": 663 }, { "epoch": 0.07065333049585018, "grad_norm": 19.421875, "learning_rate": 9.994172825469018e-06, "loss": 1.185, "step": 664 }, { "epoch": 0.07075973611406683, "grad_norm": 31.09375, "learning_rate": 9.994137293916998e-06, "loss": 1.0454, "step": 665 }, { "epoch": 0.07086614173228346, "grad_norm": 27.375, "learning_rate": 9.99410176236498e-06, "loss": 1.395, "step": 666 }, { "epoch": 0.07097254735050011, "grad_norm": 26.25, "learning_rate": 9.994066230812963e-06, "loss": 1.4144, "step": 667 }, { "epoch": 0.07107895296871675, "grad_norm": 25.984375, "learning_rate": 9.994030699260945e-06, "loss": 1.4274, "step": 668 }, { "epoch": 0.0711853585869334, "grad_norm": 21.578125, "learning_rate": 9.993995167708926e-06, "loss": 1.2618, "step": 669 }, { "epoch": 0.07129176420515003, "grad_norm": 26.109375, "learning_rate": 9.993959636156908e-06, "loss": 1.328, "step": 670 }, { "epoch": 0.07139816982336668, "grad_norm": 40.75, "learning_rate": 9.99392410460489e-06, "loss": 1.5148, "step": 671 }, { "epoch": 0.07150457544158331, "grad_norm": 63.6875, "learning_rate": 9.993888573052873e-06, "loss": 1.8663, "step": 672 }, { "epoch": 0.07161098105979996, "grad_norm": 30.453125, "learning_rate": 9.993853041500854e-06, "loss": 1.5285, "step": 673 }, { "epoch": 0.0717173866780166, "grad_norm": 35.6875, "learning_rate": 9.993817509948836e-06, "loss": 1.3875, "step": 674 }, { "epoch": 0.07182379229623324, "grad_norm": 23.890625, "learning_rate": 9.993781978396818e-06, "loss": 1.4641, "step": 675 }, { "epoch": 0.07193019791444988, "grad_norm": 28.046875, "learning_rate": 9.9937464468448e-06, "loss": 1.3999, "step": 676 }, { "epoch": 0.07203660353266653, "grad_norm": 20.0, "learning_rate": 9.99371091529278e-06, "loss": 1.2585, "step": 677 }, { "epoch": 0.07214300915088316, "grad_norm": 29.265625, "learning_rate": 9.993675383740762e-06, "loss": 1.5214, "step": 678 }, { "epoch": 0.07224941476909981, "grad_norm": 30.3125, "learning_rate": 9.993639852188744e-06, "loss": 1.453, "step": 679 }, { "epoch": 0.07235582038731644, "grad_norm": 28.984375, "learning_rate": 9.993604320636726e-06, "loss": 1.4159, "step": 680 }, { "epoch": 0.07246222600553309, "grad_norm": 29.421875, "learning_rate": 9.993568789084707e-06, "loss": 1.3031, "step": 681 }, { "epoch": 0.07256863162374974, "grad_norm": 32.03125, "learning_rate": 9.993533257532689e-06, "loss": 1.7828, "step": 682 }, { "epoch": 0.07267503724196638, "grad_norm": 28.546875, "learning_rate": 9.993497725980672e-06, "loss": 1.5678, "step": 683 }, { "epoch": 0.07278144286018302, "grad_norm": 27.296875, "learning_rate": 9.993462194428654e-06, "loss": 1.5047, "step": 684 }, { "epoch": 0.07288784847839966, "grad_norm": 47.34375, "learning_rate": 9.993426662876635e-06, "loss": 1.5373, "step": 685 }, { "epoch": 0.0729942540966163, "grad_norm": 21.625, "learning_rate": 9.993391131324617e-06, "loss": 1.1551, "step": 686 }, { "epoch": 0.07310065971483294, "grad_norm": 27.28125, "learning_rate": 9.993355599772599e-06, "loss": 1.6226, "step": 687 }, { "epoch": 0.07320706533304959, "grad_norm": 26.78125, "learning_rate": 9.993320068220582e-06, "loss": 1.3403, "step": 688 }, { "epoch": 0.07331347095126622, "grad_norm": 29.1875, "learning_rate": 9.993284536668563e-06, "loss": 1.4941, "step": 689 }, { "epoch": 0.07341987656948287, "grad_norm": 26.75, "learning_rate": 9.993249005116545e-06, "loss": 1.3094, "step": 690 }, { "epoch": 0.07352628218769951, "grad_norm": 17.09375, "learning_rate": 9.993213473564527e-06, "loss": 1.1387, "step": 691 }, { "epoch": 0.07363268780591616, "grad_norm": 31.609375, "learning_rate": 9.993177942012508e-06, "loss": 1.4314, "step": 692 }, { "epoch": 0.07373909342413279, "grad_norm": 35.78125, "learning_rate": 9.99314241046049e-06, "loss": 1.4084, "step": 693 }, { "epoch": 0.07384549904234944, "grad_norm": 23.984375, "learning_rate": 9.993106878908471e-06, "loss": 1.0013, "step": 694 }, { "epoch": 0.07395190466056607, "grad_norm": 29.53125, "learning_rate": 9.993071347356453e-06, "loss": 1.2773, "step": 695 }, { "epoch": 0.07405831027878272, "grad_norm": 26.328125, "learning_rate": 9.993035815804435e-06, "loss": 1.3877, "step": 696 }, { "epoch": 0.07416471589699936, "grad_norm": 25.5625, "learning_rate": 9.993000284252416e-06, "loss": 1.3032, "step": 697 }, { "epoch": 0.074271121515216, "grad_norm": 40.71875, "learning_rate": 9.9929647527004e-06, "loss": 1.2938, "step": 698 }, { "epoch": 0.07437752713343264, "grad_norm": 32.125, "learning_rate": 9.992929221148381e-06, "loss": 1.49, "step": 699 }, { "epoch": 0.07448393275164929, "grad_norm": 42.375, "learning_rate": 9.992893689596363e-06, "loss": 1.5633, "step": 700 }, { "epoch": 0.07459033836986594, "grad_norm": 33.21875, "learning_rate": 9.992858158044344e-06, "loss": 1.2827, "step": 701 }, { "epoch": 0.07469674398808257, "grad_norm": 20.234375, "learning_rate": 9.992822626492326e-06, "loss": 1.3641, "step": 702 }, { "epoch": 0.07480314960629922, "grad_norm": 30.3125, "learning_rate": 9.992787094940308e-06, "loss": 1.3362, "step": 703 }, { "epoch": 0.07490955522451585, "grad_norm": 25.359375, "learning_rate": 9.99275156338829e-06, "loss": 1.5313, "step": 704 }, { "epoch": 0.0750159608427325, "grad_norm": 49.03125, "learning_rate": 9.99271603183627e-06, "loss": 1.6715, "step": 705 }, { "epoch": 0.07512236646094914, "grad_norm": 33.09375, "learning_rate": 9.992680500284252e-06, "loss": 1.7213, "step": 706 }, { "epoch": 0.07522877207916578, "grad_norm": 30.921875, "learning_rate": 9.992644968732234e-06, "loss": 1.3376, "step": 707 }, { "epoch": 0.07533517769738242, "grad_norm": 21.90625, "learning_rate": 9.992609437180217e-06, "loss": 1.1631, "step": 708 }, { "epoch": 0.07544158331559907, "grad_norm": 31.75, "learning_rate": 9.992573905628199e-06, "loss": 1.3805, "step": 709 }, { "epoch": 0.0755479889338157, "grad_norm": 31.34375, "learning_rate": 9.99253837407618e-06, "loss": 1.4052, "step": 710 }, { "epoch": 0.07565439455203235, "grad_norm": 32.90625, "learning_rate": 9.992502842524162e-06, "loss": 1.2252, "step": 711 }, { "epoch": 0.07576080017024898, "grad_norm": 26.890625, "learning_rate": 9.992467310972144e-06, "loss": 1.1457, "step": 712 }, { "epoch": 0.07586720578846563, "grad_norm": 31.125, "learning_rate": 9.992431779420127e-06, "loss": 1.4408, "step": 713 }, { "epoch": 0.07597361140668227, "grad_norm": 29.515625, "learning_rate": 9.992396247868109e-06, "loss": 1.7347, "step": 714 }, { "epoch": 0.07608001702489892, "grad_norm": 20.109375, "learning_rate": 9.99236071631609e-06, "loss": 1.0963, "step": 715 }, { "epoch": 0.07618642264311555, "grad_norm": 30.40625, "learning_rate": 9.992325184764072e-06, "loss": 1.4431, "step": 716 }, { "epoch": 0.0762928282613322, "grad_norm": 28.34375, "learning_rate": 9.992289653212054e-06, "loss": 1.4923, "step": 717 }, { "epoch": 0.07639923387954883, "grad_norm": 24.875, "learning_rate": 9.992254121660035e-06, "loss": 1.3514, "step": 718 }, { "epoch": 0.07650563949776548, "grad_norm": 25.375, "learning_rate": 9.992218590108017e-06, "loss": 1.0761, "step": 719 }, { "epoch": 0.07661204511598213, "grad_norm": 27.265625, "learning_rate": 9.992183058555998e-06, "loss": 1.4092, "step": 720 }, { "epoch": 0.07671845073419876, "grad_norm": 31.703125, "learning_rate": 9.99214752700398e-06, "loss": 1.4886, "step": 721 }, { "epoch": 0.07682485635241541, "grad_norm": 35.9375, "learning_rate": 9.992111995451962e-06, "loss": 1.46, "step": 722 }, { "epoch": 0.07693126197063205, "grad_norm": 37.875, "learning_rate": 9.992076463899943e-06, "loss": 1.4294, "step": 723 }, { "epoch": 0.0770376675888487, "grad_norm": 30.234375, "learning_rate": 9.992040932347926e-06, "loss": 1.4031, "step": 724 }, { "epoch": 0.07714407320706533, "grad_norm": 22.15625, "learning_rate": 9.992005400795908e-06, "loss": 1.2368, "step": 725 }, { "epoch": 0.07725047882528198, "grad_norm": 21.921875, "learning_rate": 9.99196986924389e-06, "loss": 1.3572, "step": 726 }, { "epoch": 0.07735688444349861, "grad_norm": 41.9375, "learning_rate": 9.991934337691871e-06, "loss": 1.5688, "step": 727 }, { "epoch": 0.07746329006171526, "grad_norm": 21.09375, "learning_rate": 9.991898806139853e-06, "loss": 1.3452, "step": 728 }, { "epoch": 0.0775696956799319, "grad_norm": 23.859375, "learning_rate": 9.991863274587834e-06, "loss": 1.499, "step": 729 }, { "epoch": 0.07767610129814854, "grad_norm": 27.609375, "learning_rate": 9.991827743035816e-06, "loss": 1.4131, "step": 730 }, { "epoch": 0.07778250691636518, "grad_norm": 21.234375, "learning_rate": 9.991792211483798e-06, "loss": 1.3089, "step": 731 }, { "epoch": 0.07788891253458183, "grad_norm": 30.671875, "learning_rate": 9.99175667993178e-06, "loss": 1.4521, "step": 732 }, { "epoch": 0.07799531815279846, "grad_norm": 18.96875, "learning_rate": 9.991721148379761e-06, "loss": 1.3397, "step": 733 }, { "epoch": 0.07810172377101511, "grad_norm": 30.375, "learning_rate": 9.991685616827744e-06, "loss": 1.36, "step": 734 }, { "epoch": 0.07820812938923175, "grad_norm": 611.5, "learning_rate": 9.991650085275726e-06, "loss": 2.1195, "step": 735 }, { "epoch": 0.0783145350074484, "grad_norm": 31.359375, "learning_rate": 9.991614553723707e-06, "loss": 1.1719, "step": 736 }, { "epoch": 0.07842094062566503, "grad_norm": 46.15625, "learning_rate": 9.991579022171689e-06, "loss": 1.4399, "step": 737 }, { "epoch": 0.07852734624388168, "grad_norm": 30.0, "learning_rate": 9.99154349061967e-06, "loss": 1.3902, "step": 738 }, { "epoch": 0.07863375186209833, "grad_norm": 37.0625, "learning_rate": 9.991507959067654e-06, "loss": 1.562, "step": 739 }, { "epoch": 0.07874015748031496, "grad_norm": 20.21875, "learning_rate": 9.991472427515636e-06, "loss": 1.1236, "step": 740 }, { "epoch": 0.07884656309853161, "grad_norm": 41.09375, "learning_rate": 9.991436895963617e-06, "loss": 1.5778, "step": 741 }, { "epoch": 0.07895296871674824, "grad_norm": 26.15625, "learning_rate": 9.991401364411599e-06, "loss": 1.5342, "step": 742 }, { "epoch": 0.07905937433496489, "grad_norm": 22.15625, "learning_rate": 9.99136583285958e-06, "loss": 1.3404, "step": 743 }, { "epoch": 0.07916577995318153, "grad_norm": 25.265625, "learning_rate": 9.991330301307562e-06, "loss": 1.4727, "step": 744 }, { "epoch": 0.07927218557139817, "grad_norm": 37.71875, "learning_rate": 9.991294769755544e-06, "loss": 1.3943, "step": 745 }, { "epoch": 0.07937859118961481, "grad_norm": 31.21875, "learning_rate": 9.991259238203525e-06, "loss": 1.4187, "step": 746 }, { "epoch": 0.07948499680783146, "grad_norm": 23.53125, "learning_rate": 9.991223706651507e-06, "loss": 1.3589, "step": 747 }, { "epoch": 0.07959140242604809, "grad_norm": 26.140625, "learning_rate": 9.991188175099488e-06, "loss": 1.3839, "step": 748 }, { "epoch": 0.07969780804426474, "grad_norm": 20.859375, "learning_rate": 9.991152643547472e-06, "loss": 1.4494, "step": 749 }, { "epoch": 0.07980421366248137, "grad_norm": 22.390625, "learning_rate": 9.991117111995453e-06, "loss": 1.3939, "step": 750 }, { "epoch": 0.07991061928069802, "grad_norm": 38.1875, "learning_rate": 9.991081580443435e-06, "loss": 1.6587, "step": 751 }, { "epoch": 0.08001702489891466, "grad_norm": 19.03125, "learning_rate": 9.991046048891417e-06, "loss": 1.2128, "step": 752 }, { "epoch": 0.0801234305171313, "grad_norm": 24.0625, "learning_rate": 9.991010517339398e-06, "loss": 1.2081, "step": 753 }, { "epoch": 0.08022983613534794, "grad_norm": 59.78125, "learning_rate": 9.99097498578738e-06, "loss": 1.4316, "step": 754 }, { "epoch": 0.08033624175356459, "grad_norm": 26.390625, "learning_rate": 9.990939454235361e-06, "loss": 1.3456, "step": 755 }, { "epoch": 0.08044264737178124, "grad_norm": 49.96875, "learning_rate": 9.990903922683343e-06, "loss": 1.5678, "step": 756 }, { "epoch": 0.08054905298999787, "grad_norm": 26.015625, "learning_rate": 9.990868391131325e-06, "loss": 1.4114, "step": 757 }, { "epoch": 0.08065545860821452, "grad_norm": 36.59375, "learning_rate": 9.990832859579306e-06, "loss": 1.4146, "step": 758 }, { "epoch": 0.08076186422643115, "grad_norm": 30.96875, "learning_rate": 9.990797328027288e-06, "loss": 1.6073, "step": 759 }, { "epoch": 0.0808682698446478, "grad_norm": 22.25, "learning_rate": 9.990761796475271e-06, "loss": 1.3501, "step": 760 }, { "epoch": 0.08097467546286444, "grad_norm": 29.03125, "learning_rate": 9.990726264923253e-06, "loss": 1.5475, "step": 761 }, { "epoch": 0.08108108108108109, "grad_norm": 29.234375, "learning_rate": 9.990690733371234e-06, "loss": 1.5271, "step": 762 }, { "epoch": 0.08118748669929772, "grad_norm": 23.28125, "learning_rate": 9.990655201819216e-06, "loss": 1.4302, "step": 763 }, { "epoch": 0.08129389231751437, "grad_norm": 24.765625, "learning_rate": 9.990619670267197e-06, "loss": 1.3364, "step": 764 }, { "epoch": 0.081400297935731, "grad_norm": 33.1875, "learning_rate": 9.99058413871518e-06, "loss": 1.3617, "step": 765 }, { "epoch": 0.08150670355394765, "grad_norm": 39.4375, "learning_rate": 9.990548607163162e-06, "loss": 1.7821, "step": 766 }, { "epoch": 0.08161310917216429, "grad_norm": 19.828125, "learning_rate": 9.990513075611144e-06, "loss": 1.3844, "step": 767 }, { "epoch": 0.08171951479038093, "grad_norm": 26.421875, "learning_rate": 9.990477544059126e-06, "loss": 1.5092, "step": 768 }, { "epoch": 0.08182592040859757, "grad_norm": 31.765625, "learning_rate": 9.990442012507107e-06, "loss": 1.5608, "step": 769 }, { "epoch": 0.08193232602681422, "grad_norm": 26.921875, "learning_rate": 9.990406480955089e-06, "loss": 1.5106, "step": 770 }, { "epoch": 0.08203873164503085, "grad_norm": 21.4375, "learning_rate": 9.99037094940307e-06, "loss": 1.2261, "step": 771 }, { "epoch": 0.0821451372632475, "grad_norm": 21.34375, "learning_rate": 9.990335417851052e-06, "loss": 1.3914, "step": 772 }, { "epoch": 0.08225154288146413, "grad_norm": 24.109375, "learning_rate": 9.990299886299034e-06, "loss": 1.3796, "step": 773 }, { "epoch": 0.08235794849968078, "grad_norm": 24.453125, "learning_rate": 9.990264354747015e-06, "loss": 1.2431, "step": 774 }, { "epoch": 0.08246435411789743, "grad_norm": 24.21875, "learning_rate": 9.990228823194999e-06, "loss": 1.1782, "step": 775 }, { "epoch": 0.08257075973611407, "grad_norm": 21.5, "learning_rate": 9.99019329164298e-06, "loss": 1.5682, "step": 776 }, { "epoch": 0.08267716535433071, "grad_norm": 18.21875, "learning_rate": 9.990157760090962e-06, "loss": 1.0081, "step": 777 }, { "epoch": 0.08278357097254735, "grad_norm": 25.671875, "learning_rate": 9.990122228538943e-06, "loss": 1.3829, "step": 778 }, { "epoch": 0.082889976590764, "grad_norm": 27.25, "learning_rate": 9.990086696986925e-06, "loss": 1.386, "step": 779 }, { "epoch": 0.08299638220898063, "grad_norm": 58.75, "learning_rate": 9.990051165434908e-06, "loss": 1.8014, "step": 780 }, { "epoch": 0.08310278782719728, "grad_norm": 20.4375, "learning_rate": 9.99001563388289e-06, "loss": 1.1577, "step": 781 }, { "epoch": 0.08320919344541391, "grad_norm": 24.65625, "learning_rate": 9.989980102330871e-06, "loss": 1.5937, "step": 782 }, { "epoch": 0.08331559906363056, "grad_norm": 26.90625, "learning_rate": 9.989944570778853e-06, "loss": 1.1858, "step": 783 }, { "epoch": 0.0834220046818472, "grad_norm": 16.375, "learning_rate": 9.989909039226833e-06, "loss": 0.936, "step": 784 }, { "epoch": 0.08352841030006385, "grad_norm": 49.65625, "learning_rate": 9.989873507674816e-06, "loss": 1.5582, "step": 785 }, { "epoch": 0.08363481591828048, "grad_norm": 19.328125, "learning_rate": 9.989837976122798e-06, "loss": 1.427, "step": 786 }, { "epoch": 0.08374122153649713, "grad_norm": 29.171875, "learning_rate": 9.98980244457078e-06, "loss": 1.4794, "step": 787 }, { "epoch": 0.08384762715471376, "grad_norm": 28.0, "learning_rate": 9.989766913018761e-06, "loss": 1.2976, "step": 788 }, { "epoch": 0.08395403277293041, "grad_norm": 29.046875, "learning_rate": 9.989731381466743e-06, "loss": 1.7819, "step": 789 }, { "epoch": 0.08406043839114705, "grad_norm": 34.125, "learning_rate": 9.989695849914726e-06, "loss": 1.2953, "step": 790 }, { "epoch": 0.0841668440093637, "grad_norm": 26.578125, "learning_rate": 9.989660318362708e-06, "loss": 1.5418, "step": 791 }, { "epoch": 0.08427324962758033, "grad_norm": 16.96875, "learning_rate": 9.98962478681069e-06, "loss": 0.9982, "step": 792 }, { "epoch": 0.08437965524579698, "grad_norm": 19.34375, "learning_rate": 9.989589255258671e-06, "loss": 1.1611, "step": 793 }, { "epoch": 0.08448606086401363, "grad_norm": 43.0, "learning_rate": 9.989553723706652e-06, "loss": 1.5658, "step": 794 }, { "epoch": 0.08459246648223026, "grad_norm": 21.65625, "learning_rate": 9.989518192154634e-06, "loss": 1.4284, "step": 795 }, { "epoch": 0.08469887210044691, "grad_norm": 21.25, "learning_rate": 9.989482660602616e-06, "loss": 1.4419, "step": 796 }, { "epoch": 0.08480527771866354, "grad_norm": 21.0625, "learning_rate": 9.989447129050597e-06, "loss": 1.2436, "step": 797 }, { "epoch": 0.08491168333688019, "grad_norm": 15.703125, "learning_rate": 9.989411597498579e-06, "loss": 1.2773, "step": 798 }, { "epoch": 0.08501808895509683, "grad_norm": 21.78125, "learning_rate": 9.98937606594656e-06, "loss": 1.3353, "step": 799 }, { "epoch": 0.08512449457331348, "grad_norm": 23.671875, "learning_rate": 9.989340534394542e-06, "loss": 1.5441, "step": 800 }, { "epoch": 0.08523090019153011, "grad_norm": 21.0, "learning_rate": 9.989305002842525e-06, "loss": 1.602, "step": 801 }, { "epoch": 0.08533730580974676, "grad_norm": 53.4375, "learning_rate": 9.989269471290507e-06, "loss": 1.5711, "step": 802 }, { "epoch": 0.08544371142796339, "grad_norm": 18.34375, "learning_rate": 9.989233939738489e-06, "loss": 1.4742, "step": 803 }, { "epoch": 0.08555011704618004, "grad_norm": 17.8125, "learning_rate": 9.98919840818647e-06, "loss": 1.2853, "step": 804 }, { "epoch": 0.08565652266439668, "grad_norm": 26.046875, "learning_rate": 9.989162876634452e-06, "loss": 1.4151, "step": 805 }, { "epoch": 0.08576292828261332, "grad_norm": 29.15625, "learning_rate": 9.989127345082435e-06, "loss": 1.7744, "step": 806 }, { "epoch": 0.08586933390082996, "grad_norm": 17.25, "learning_rate": 9.989091813530417e-06, "loss": 1.1335, "step": 807 }, { "epoch": 0.0859757395190466, "grad_norm": 17.890625, "learning_rate": 9.989056281978398e-06, "loss": 1.3097, "step": 808 }, { "epoch": 0.08608214513726324, "grad_norm": 29.953125, "learning_rate": 9.98902075042638e-06, "loss": 1.4068, "step": 809 }, { "epoch": 0.08618855075547989, "grad_norm": 20.515625, "learning_rate": 9.988985218874362e-06, "loss": 1.3321, "step": 810 }, { "epoch": 0.08629495637369654, "grad_norm": 28.71875, "learning_rate": 9.988949687322343e-06, "loss": 1.6402, "step": 811 }, { "epoch": 0.08640136199191317, "grad_norm": 25.0625, "learning_rate": 9.988914155770325e-06, "loss": 1.4751, "step": 812 }, { "epoch": 0.08650776761012982, "grad_norm": 15.421875, "learning_rate": 9.988878624218306e-06, "loss": 1.1123, "step": 813 }, { "epoch": 0.08661417322834646, "grad_norm": 32.1875, "learning_rate": 9.988843092666288e-06, "loss": 1.7843, "step": 814 }, { "epoch": 0.0867205788465631, "grad_norm": 36.125, "learning_rate": 9.98880756111427e-06, "loss": 1.8004, "step": 815 }, { "epoch": 0.08682698446477974, "grad_norm": 44.21875, "learning_rate": 9.988772029562253e-06, "loss": 1.8127, "step": 816 }, { "epoch": 0.08693339008299639, "grad_norm": 23.671875, "learning_rate": 9.988736498010235e-06, "loss": 1.4626, "step": 817 }, { "epoch": 0.08703979570121302, "grad_norm": 24.890625, "learning_rate": 9.988700966458216e-06, "loss": 1.5082, "step": 818 }, { "epoch": 0.08714620131942967, "grad_norm": 28.96875, "learning_rate": 9.988665434906198e-06, "loss": 1.2802, "step": 819 }, { "epoch": 0.0872526069376463, "grad_norm": 20.328125, "learning_rate": 9.98862990335418e-06, "loss": 1.1995, "step": 820 }, { "epoch": 0.08735901255586295, "grad_norm": 31.765625, "learning_rate": 9.988594371802161e-06, "loss": 1.6991, "step": 821 }, { "epoch": 0.08746541817407959, "grad_norm": 21.890625, "learning_rate": 9.988558840250143e-06, "loss": 1.1183, "step": 822 }, { "epoch": 0.08757182379229624, "grad_norm": 22.875, "learning_rate": 9.988523308698124e-06, "loss": 1.4064, "step": 823 }, { "epoch": 0.08767822941051287, "grad_norm": 15.8515625, "learning_rate": 9.988487777146106e-06, "loss": 1.1332, "step": 824 }, { "epoch": 0.08778463502872952, "grad_norm": 26.046875, "learning_rate": 9.988452245594087e-06, "loss": 1.4228, "step": 825 }, { "epoch": 0.08789104064694615, "grad_norm": 29.609375, "learning_rate": 9.98841671404207e-06, "loss": 1.5127, "step": 826 }, { "epoch": 0.0879974462651628, "grad_norm": 25.53125, "learning_rate": 9.988381182490052e-06, "loss": 1.2658, "step": 827 }, { "epoch": 0.08810385188337944, "grad_norm": 33.40625, "learning_rate": 9.988345650938034e-06, "loss": 1.5213, "step": 828 }, { "epoch": 0.08821025750159608, "grad_norm": 27.328125, "learning_rate": 9.988310119386015e-06, "loss": 1.3764, "step": 829 }, { "epoch": 0.08831666311981273, "grad_norm": 19.3125, "learning_rate": 9.988274587833997e-06, "loss": 1.2664, "step": 830 }, { "epoch": 0.08842306873802937, "grad_norm": 20.59375, "learning_rate": 9.98823905628198e-06, "loss": 1.0597, "step": 831 }, { "epoch": 0.08852947435624602, "grad_norm": 20.09375, "learning_rate": 9.988203524729962e-06, "loss": 1.1294, "step": 832 }, { "epoch": 0.08863587997446265, "grad_norm": 45.84375, "learning_rate": 9.988167993177944e-06, "loss": 1.6892, "step": 833 }, { "epoch": 0.0887422855926793, "grad_norm": 31.5, "learning_rate": 9.988132461625925e-06, "loss": 1.5054, "step": 834 }, { "epoch": 0.08884869121089593, "grad_norm": 26.125, "learning_rate": 9.988096930073907e-06, "loss": 1.3202, "step": 835 }, { "epoch": 0.08895509682911258, "grad_norm": 20.71875, "learning_rate": 9.988061398521888e-06, "loss": 1.2344, "step": 836 }, { "epoch": 0.08906150244732922, "grad_norm": 31.484375, "learning_rate": 9.98802586696987e-06, "loss": 1.6014, "step": 837 }, { "epoch": 0.08916790806554586, "grad_norm": 16.953125, "learning_rate": 9.987990335417852e-06, "loss": 0.7363, "step": 838 }, { "epoch": 0.0892743136837625, "grad_norm": 18.109375, "learning_rate": 9.987954803865833e-06, "loss": 1.2932, "step": 839 }, { "epoch": 0.08938071930197915, "grad_norm": 14.75, "learning_rate": 9.987919272313815e-06, "loss": 1.2987, "step": 840 }, { "epoch": 0.08948712492019578, "grad_norm": 20.046875, "learning_rate": 9.987883740761796e-06, "loss": 1.2334, "step": 841 }, { "epoch": 0.08959353053841243, "grad_norm": 25.734375, "learning_rate": 9.98784820920978e-06, "loss": 1.2616, "step": 842 }, { "epoch": 0.08969993615662906, "grad_norm": 16.296875, "learning_rate": 9.987812677657761e-06, "loss": 1.1854, "step": 843 }, { "epoch": 0.08980634177484571, "grad_norm": 26.453125, "learning_rate": 9.987777146105743e-06, "loss": 1.5013, "step": 844 }, { "epoch": 0.08991274739306235, "grad_norm": 14.4375, "learning_rate": 9.987741614553725e-06, "loss": 1.1214, "step": 845 }, { "epoch": 0.090019153011279, "grad_norm": 27.84375, "learning_rate": 9.987706083001706e-06, "loss": 1.5285, "step": 846 }, { "epoch": 0.09012555862949563, "grad_norm": 21.859375, "learning_rate": 9.987670551449688e-06, "loss": 1.2014, "step": 847 }, { "epoch": 0.09023196424771228, "grad_norm": 32.40625, "learning_rate": 9.98763501989767e-06, "loss": 1.2937, "step": 848 }, { "epoch": 0.09033836986592893, "grad_norm": 25.75, "learning_rate": 9.987599488345651e-06, "loss": 1.4617, "step": 849 }, { "epoch": 0.09044477548414556, "grad_norm": 20.359375, "learning_rate": 9.987563956793633e-06, "loss": 1.0973, "step": 850 }, { "epoch": 0.09055118110236221, "grad_norm": 38.875, "learning_rate": 9.987528425241614e-06, "loss": 1.4787, "step": 851 }, { "epoch": 0.09065758672057884, "grad_norm": 30.921875, "learning_rate": 9.987492893689598e-06, "loss": 1.4812, "step": 852 }, { "epoch": 0.0907639923387955, "grad_norm": 29.90625, "learning_rate": 9.987457362137579e-06, "loss": 1.5353, "step": 853 }, { "epoch": 0.09087039795701213, "grad_norm": 26.3125, "learning_rate": 9.98742183058556e-06, "loss": 1.4487, "step": 854 }, { "epoch": 0.09097680357522878, "grad_norm": 35.75, "learning_rate": 9.987386299033542e-06, "loss": 1.6006, "step": 855 }, { "epoch": 0.09108320919344541, "grad_norm": 24.359375, "learning_rate": 9.987350767481524e-06, "loss": 1.4328, "step": 856 }, { "epoch": 0.09118961481166206, "grad_norm": 33.875, "learning_rate": 9.987315235929507e-06, "loss": 1.6955, "step": 857 }, { "epoch": 0.0912960204298787, "grad_norm": 24.71875, "learning_rate": 9.987279704377489e-06, "loss": 1.4723, "step": 858 }, { "epoch": 0.09140242604809534, "grad_norm": 35.15625, "learning_rate": 9.98724417282547e-06, "loss": 1.5675, "step": 859 }, { "epoch": 0.09150883166631198, "grad_norm": 25.15625, "learning_rate": 9.987208641273452e-06, "loss": 1.4625, "step": 860 }, { "epoch": 0.09161523728452863, "grad_norm": 27.984375, "learning_rate": 9.987173109721434e-06, "loss": 1.5852, "step": 861 }, { "epoch": 0.09172164290274526, "grad_norm": 21.421875, "learning_rate": 9.987137578169415e-06, "loss": 1.4879, "step": 862 }, { "epoch": 0.09182804852096191, "grad_norm": 19.125, "learning_rate": 9.987102046617397e-06, "loss": 1.2712, "step": 863 }, { "epoch": 0.09193445413917854, "grad_norm": 26.609375, "learning_rate": 9.987066515065378e-06, "loss": 1.3639, "step": 864 }, { "epoch": 0.09204085975739519, "grad_norm": 18.078125, "learning_rate": 9.98703098351336e-06, "loss": 1.3351, "step": 865 }, { "epoch": 0.09214726537561183, "grad_norm": 19.78125, "learning_rate": 9.986995451961342e-06, "loss": 1.5944, "step": 866 }, { "epoch": 0.09225367099382847, "grad_norm": 28.671875, "learning_rate": 9.986959920409325e-06, "loss": 1.9232, "step": 867 }, { "epoch": 0.09236007661204512, "grad_norm": 35.46875, "learning_rate": 9.986924388857307e-06, "loss": 1.7573, "step": 868 }, { "epoch": 0.09246648223026176, "grad_norm": 18.5625, "learning_rate": 9.986888857305288e-06, "loss": 1.0248, "step": 869 }, { "epoch": 0.0925728878484784, "grad_norm": 21.703125, "learning_rate": 9.98685332575327e-06, "loss": 1.3225, "step": 870 }, { "epoch": 0.09267929346669504, "grad_norm": 24.25, "learning_rate": 9.986817794201251e-06, "loss": 1.395, "step": 871 }, { "epoch": 0.09278569908491169, "grad_norm": 53.8125, "learning_rate": 9.986782262649233e-06, "loss": 1.359, "step": 872 }, { "epoch": 0.09289210470312832, "grad_norm": 31.453125, "learning_rate": 9.986746731097215e-06, "loss": 1.4152, "step": 873 }, { "epoch": 0.09299851032134497, "grad_norm": 25.515625, "learning_rate": 9.986711199545196e-06, "loss": 1.5939, "step": 874 }, { "epoch": 0.0931049159395616, "grad_norm": 39.03125, "learning_rate": 9.986675667993178e-06, "loss": 1.5358, "step": 875 }, { "epoch": 0.09321132155777825, "grad_norm": 23.875, "learning_rate": 9.98664013644116e-06, "loss": 1.1015, "step": 876 }, { "epoch": 0.09331772717599489, "grad_norm": 30.734375, "learning_rate": 9.986604604889141e-06, "loss": 1.4567, "step": 877 }, { "epoch": 0.09342413279421154, "grad_norm": 25.21875, "learning_rate": 9.986569073337124e-06, "loss": 1.5454, "step": 878 }, { "epoch": 0.09353053841242817, "grad_norm": 24.15625, "learning_rate": 9.986533541785106e-06, "loss": 1.3906, "step": 879 }, { "epoch": 0.09363694403064482, "grad_norm": 26.8125, "learning_rate": 9.986498010233088e-06, "loss": 1.3464, "step": 880 }, { "epoch": 0.09374334964886145, "grad_norm": 26.40625, "learning_rate": 9.98646247868107e-06, "loss": 1.6185, "step": 881 }, { "epoch": 0.0938497552670781, "grad_norm": 15.8046875, "learning_rate": 9.98642694712905e-06, "loss": 1.168, "step": 882 }, { "epoch": 0.09395616088529474, "grad_norm": 22.4375, "learning_rate": 9.986391415577034e-06, "loss": 1.6536, "step": 883 }, { "epoch": 0.09406256650351139, "grad_norm": 27.234375, "learning_rate": 9.986355884025016e-06, "loss": 1.3103, "step": 884 }, { "epoch": 0.09416897212172803, "grad_norm": 23.484375, "learning_rate": 9.986320352472997e-06, "loss": 1.3271, "step": 885 }, { "epoch": 0.09427537773994467, "grad_norm": 21.546875, "learning_rate": 9.986284820920979e-06, "loss": 1.314, "step": 886 }, { "epoch": 0.09438178335816132, "grad_norm": 24.546875, "learning_rate": 9.98624928936896e-06, "loss": 1.3298, "step": 887 }, { "epoch": 0.09448818897637795, "grad_norm": 39.28125, "learning_rate": 9.986213757816942e-06, "loss": 1.5287, "step": 888 }, { "epoch": 0.0945945945945946, "grad_norm": 19.59375, "learning_rate": 9.986178226264924e-06, "loss": 1.2964, "step": 889 }, { "epoch": 0.09470100021281123, "grad_norm": 21.953125, "learning_rate": 9.986142694712905e-06, "loss": 1.4959, "step": 890 }, { "epoch": 0.09480740583102788, "grad_norm": 22.140625, "learning_rate": 9.986107163160887e-06, "loss": 1.3909, "step": 891 }, { "epoch": 0.09491381144924452, "grad_norm": 24.421875, "learning_rate": 9.986071631608869e-06, "loss": 1.4465, "step": 892 }, { "epoch": 0.09502021706746117, "grad_norm": 31.1875, "learning_rate": 9.986036100056852e-06, "loss": 1.4541, "step": 893 }, { "epoch": 0.0951266226856778, "grad_norm": 22.75, "learning_rate": 9.986000568504833e-06, "loss": 1.4871, "step": 894 }, { "epoch": 0.09523302830389445, "grad_norm": 17.5625, "learning_rate": 9.985965036952815e-06, "loss": 1.2252, "step": 895 }, { "epoch": 0.09533943392211108, "grad_norm": 27.09375, "learning_rate": 9.985929505400797e-06, "loss": 1.5133, "step": 896 }, { "epoch": 0.09544583954032773, "grad_norm": 31.140625, "learning_rate": 9.985893973848778e-06, "loss": 1.0718, "step": 897 }, { "epoch": 0.09555224515854437, "grad_norm": 13.9765625, "learning_rate": 9.985858442296762e-06, "loss": 1.2279, "step": 898 }, { "epoch": 0.09565865077676101, "grad_norm": 24.6875, "learning_rate": 9.985822910744743e-06, "loss": 1.5815, "step": 899 }, { "epoch": 0.09576505639497765, "grad_norm": 22.78125, "learning_rate": 9.985787379192725e-06, "loss": 1.5268, "step": 900 }, { "epoch": 0.0958714620131943, "grad_norm": 26.203125, "learning_rate": 9.985751847640706e-06, "loss": 1.3735, "step": 901 }, { "epoch": 0.09597786763141093, "grad_norm": 29.21875, "learning_rate": 9.985716316088688e-06, "loss": 1.3702, "step": 902 }, { "epoch": 0.09608427324962758, "grad_norm": 23.84375, "learning_rate": 9.98568078453667e-06, "loss": 1.3874, "step": 903 }, { "epoch": 0.09619067886784423, "grad_norm": 27.53125, "learning_rate": 9.985645252984651e-06, "loss": 1.2999, "step": 904 }, { "epoch": 0.09629708448606086, "grad_norm": 28.140625, "learning_rate": 9.985609721432633e-06, "loss": 1.5458, "step": 905 }, { "epoch": 0.09640349010427751, "grad_norm": 20.046875, "learning_rate": 9.985574189880614e-06, "loss": 1.3769, "step": 906 }, { "epoch": 0.09650989572249415, "grad_norm": 22.5625, "learning_rate": 9.985538658328596e-06, "loss": 1.4339, "step": 907 }, { "epoch": 0.0966163013407108, "grad_norm": 25.109375, "learning_rate": 9.98550312677658e-06, "loss": 1.6386, "step": 908 }, { "epoch": 0.09672270695892743, "grad_norm": 93.0, "learning_rate": 9.985467595224561e-06, "loss": 1.3394, "step": 909 }, { "epoch": 0.09682911257714408, "grad_norm": 21.515625, "learning_rate": 9.985432063672543e-06, "loss": 1.1655, "step": 910 }, { "epoch": 0.09693551819536071, "grad_norm": 20.03125, "learning_rate": 9.985396532120524e-06, "loss": 1.2885, "step": 911 }, { "epoch": 0.09704192381357736, "grad_norm": 18.765625, "learning_rate": 9.985361000568506e-06, "loss": 1.245, "step": 912 }, { "epoch": 0.097148329431794, "grad_norm": 37.09375, "learning_rate": 9.985325469016487e-06, "loss": 1.6468, "step": 913 }, { "epoch": 0.09725473505001064, "grad_norm": 49.34375, "learning_rate": 9.985289937464469e-06, "loss": 1.2293, "step": 914 }, { "epoch": 0.09736114066822728, "grad_norm": 19.890625, "learning_rate": 9.98525440591245e-06, "loss": 1.3006, "step": 915 }, { "epoch": 0.09746754628644393, "grad_norm": 24.4375, "learning_rate": 9.985218874360432e-06, "loss": 1.5584, "step": 916 }, { "epoch": 0.09757395190466056, "grad_norm": 20.6875, "learning_rate": 9.985183342808414e-06, "loss": 1.293, "step": 917 }, { "epoch": 0.09768035752287721, "grad_norm": 25.1875, "learning_rate": 9.985147811256395e-06, "loss": 1.5053, "step": 918 }, { "epoch": 0.09778676314109384, "grad_norm": 30.234375, "learning_rate": 9.985112279704379e-06, "loss": 1.2893, "step": 919 }, { "epoch": 0.09789316875931049, "grad_norm": 16.5, "learning_rate": 9.98507674815236e-06, "loss": 1.3919, "step": 920 }, { "epoch": 0.09799957437752713, "grad_norm": 28.15625, "learning_rate": 9.985041216600342e-06, "loss": 1.4884, "step": 921 }, { "epoch": 0.09810597999574378, "grad_norm": 20.1875, "learning_rate": 9.985005685048324e-06, "loss": 1.1004, "step": 922 }, { "epoch": 0.09821238561396042, "grad_norm": 15.765625, "learning_rate": 9.984970153496305e-06, "loss": 1.3056, "step": 923 }, { "epoch": 0.09831879123217706, "grad_norm": 20.109375, "learning_rate": 9.984934621944288e-06, "loss": 1.2173, "step": 924 }, { "epoch": 0.0984251968503937, "grad_norm": 16.734375, "learning_rate": 9.98489909039227e-06, "loss": 1.3078, "step": 925 }, { "epoch": 0.09853160246861034, "grad_norm": 25.28125, "learning_rate": 9.984863558840252e-06, "loss": 1.5661, "step": 926 }, { "epoch": 0.09863800808682699, "grad_norm": 31.734375, "learning_rate": 9.984828027288233e-06, "loss": 1.5591, "step": 927 }, { "epoch": 0.09874441370504362, "grad_norm": 23.46875, "learning_rate": 9.984792495736215e-06, "loss": 1.6146, "step": 928 }, { "epoch": 0.09885081932326027, "grad_norm": 18.328125, "learning_rate": 9.984756964184196e-06, "loss": 1.2934, "step": 929 }, { "epoch": 0.0989572249414769, "grad_norm": 18.96875, "learning_rate": 9.984721432632178e-06, "loss": 1.374, "step": 930 }, { "epoch": 0.09906363055969356, "grad_norm": 20.65625, "learning_rate": 9.98468590108016e-06, "loss": 1.3524, "step": 931 }, { "epoch": 0.09917003617791019, "grad_norm": 21.625, "learning_rate": 9.984650369528141e-06, "loss": 1.3294, "step": 932 }, { "epoch": 0.09927644179612684, "grad_norm": 35.40625, "learning_rate": 9.984614837976123e-06, "loss": 1.4601, "step": 933 }, { "epoch": 0.09938284741434347, "grad_norm": 20.8125, "learning_rate": 9.984579306424106e-06, "loss": 1.3517, "step": 934 }, { "epoch": 0.09948925303256012, "grad_norm": 23.34375, "learning_rate": 9.984543774872088e-06, "loss": 1.3815, "step": 935 }, { "epoch": 0.09959565865077676, "grad_norm": 30.296875, "learning_rate": 9.98450824332007e-06, "loss": 1.5922, "step": 936 }, { "epoch": 0.0997020642689934, "grad_norm": 25.671875, "learning_rate": 9.984472711768051e-06, "loss": 1.2061, "step": 937 }, { "epoch": 0.09980846988721004, "grad_norm": 25.484375, "learning_rate": 9.984437180216033e-06, "loss": 1.7041, "step": 938 }, { "epoch": 0.09991487550542669, "grad_norm": 17.734375, "learning_rate": 9.984401648664014e-06, "loss": 1.232, "step": 939 }, { "epoch": 0.10002128112364334, "grad_norm": 28.875, "learning_rate": 9.984366117111996e-06, "loss": 1.5152, "step": 940 }, { "epoch": 0.10012768674185997, "grad_norm": 21.265625, "learning_rate": 9.984330585559977e-06, "loss": 1.4263, "step": 941 }, { "epoch": 0.10023409236007662, "grad_norm": 21.09375, "learning_rate": 9.984295054007959e-06, "loss": 1.243, "step": 942 }, { "epoch": 0.10034049797829325, "grad_norm": 20.828125, "learning_rate": 9.98425952245594e-06, "loss": 1.4883, "step": 943 }, { "epoch": 0.1004469035965099, "grad_norm": 40.0, "learning_rate": 9.984223990903924e-06, "loss": 1.6519, "step": 944 }, { "epoch": 0.10055330921472654, "grad_norm": 18.734375, "learning_rate": 9.984188459351906e-06, "loss": 1.1453, "step": 945 }, { "epoch": 0.10065971483294318, "grad_norm": 21.59375, "learning_rate": 9.984152927799887e-06, "loss": 1.3212, "step": 946 }, { "epoch": 0.10076612045115982, "grad_norm": 20.640625, "learning_rate": 9.984117396247869e-06, "loss": 1.0854, "step": 947 }, { "epoch": 0.10087252606937647, "grad_norm": 23.703125, "learning_rate": 9.98408186469585e-06, "loss": 1.3862, "step": 948 }, { "epoch": 0.1009789316875931, "grad_norm": 18.84375, "learning_rate": 9.984046333143834e-06, "loss": 1.1762, "step": 949 }, { "epoch": 0.10108533730580975, "grad_norm": 23.21875, "learning_rate": 9.984010801591815e-06, "loss": 1.2269, "step": 950 }, { "epoch": 0.10119174292402638, "grad_norm": 26.546875, "learning_rate": 9.983975270039797e-06, "loss": 1.3494, "step": 951 }, { "epoch": 0.10129814854224303, "grad_norm": 21.09375, "learning_rate": 9.983939738487779e-06, "loss": 1.3036, "step": 952 }, { "epoch": 0.10140455416045967, "grad_norm": 22.375, "learning_rate": 9.98390420693576e-06, "loss": 1.3361, "step": 953 }, { "epoch": 0.10151095977867632, "grad_norm": 43.34375, "learning_rate": 9.983868675383742e-06, "loss": 1.498, "step": 954 }, { "epoch": 0.10161736539689295, "grad_norm": 52.15625, "learning_rate": 9.983833143831723e-06, "loss": 1.1753, "step": 955 }, { "epoch": 0.1017237710151096, "grad_norm": 22.421875, "learning_rate": 9.983797612279705e-06, "loss": 1.3975, "step": 956 }, { "epoch": 0.10183017663332623, "grad_norm": 39.5, "learning_rate": 9.983762080727687e-06, "loss": 1.8103, "step": 957 }, { "epoch": 0.10193658225154288, "grad_norm": 46.125, "learning_rate": 9.983726549175668e-06, "loss": 1.5225, "step": 958 }, { "epoch": 0.10204298786975953, "grad_norm": 28.28125, "learning_rate": 9.98369101762365e-06, "loss": 1.2524, "step": 959 }, { "epoch": 0.10214939348797616, "grad_norm": 21.890625, "learning_rate": 9.983655486071633e-06, "loss": 1.484, "step": 960 }, { "epoch": 0.10225579910619281, "grad_norm": 25.84375, "learning_rate": 9.983619954519615e-06, "loss": 1.494, "step": 961 }, { "epoch": 0.10236220472440945, "grad_norm": 44.3125, "learning_rate": 9.983584422967596e-06, "loss": 1.5496, "step": 962 }, { "epoch": 0.1024686103426261, "grad_norm": 70.625, "learning_rate": 9.983548891415578e-06, "loss": 1.2153, "step": 963 }, { "epoch": 0.10257501596084273, "grad_norm": 16.921875, "learning_rate": 9.98351335986356e-06, "loss": 1.2545, "step": 964 }, { "epoch": 0.10268142157905938, "grad_norm": 27.875, "learning_rate": 9.983477828311541e-06, "loss": 1.4444, "step": 965 }, { "epoch": 0.10278782719727601, "grad_norm": 26.53125, "learning_rate": 9.983442296759523e-06, "loss": 1.4554, "step": 966 }, { "epoch": 0.10289423281549266, "grad_norm": 37.8125, "learning_rate": 9.983406765207504e-06, "loss": 1.2556, "step": 967 }, { "epoch": 0.1030006384337093, "grad_norm": 23.015625, "learning_rate": 9.983371233655486e-06, "loss": 1.3258, "step": 968 }, { "epoch": 0.10310704405192594, "grad_norm": 44.84375, "learning_rate": 9.983335702103468e-06, "loss": 1.6764, "step": 969 }, { "epoch": 0.10321344967014258, "grad_norm": 24.640625, "learning_rate": 9.98330017055145e-06, "loss": 1.5247, "step": 970 }, { "epoch": 0.10331985528835923, "grad_norm": 18.96875, "learning_rate": 9.983264638999432e-06, "loss": 1.1305, "step": 971 }, { "epoch": 0.10342626090657586, "grad_norm": 30.890625, "learning_rate": 9.983229107447414e-06, "loss": 1.6113, "step": 972 }, { "epoch": 0.10353266652479251, "grad_norm": 23.03125, "learning_rate": 9.983193575895396e-06, "loss": 1.2809, "step": 973 }, { "epoch": 0.10363907214300914, "grad_norm": 18.359375, "learning_rate": 9.983158044343377e-06, "loss": 1.5039, "step": 974 }, { "epoch": 0.1037454777612258, "grad_norm": 40.78125, "learning_rate": 9.98312251279136e-06, "loss": 1.4943, "step": 975 }, { "epoch": 0.10385188337944243, "grad_norm": 29.703125, "learning_rate": 9.983086981239342e-06, "loss": 1.5848, "step": 976 }, { "epoch": 0.10395828899765908, "grad_norm": 23.078125, "learning_rate": 9.983051449687324e-06, "loss": 1.2614, "step": 977 }, { "epoch": 0.10406469461587572, "grad_norm": 19.3125, "learning_rate": 9.983015918135305e-06, "loss": 1.1193, "step": 978 }, { "epoch": 0.10417110023409236, "grad_norm": 18.390625, "learning_rate": 9.982980386583287e-06, "loss": 1.3419, "step": 979 }, { "epoch": 0.10427750585230901, "grad_norm": 21.578125, "learning_rate": 9.982944855031269e-06, "loss": 1.4495, "step": 980 }, { "epoch": 0.10438391147052564, "grad_norm": 24.703125, "learning_rate": 9.98290932347925e-06, "loss": 1.4918, "step": 981 }, { "epoch": 0.10449031708874229, "grad_norm": 27.25, "learning_rate": 9.982873791927232e-06, "loss": 1.3671, "step": 982 }, { "epoch": 0.10459672270695893, "grad_norm": 27.1875, "learning_rate": 9.982838260375213e-06, "loss": 1.5519, "step": 983 }, { "epoch": 0.10470312832517557, "grad_norm": 32.8125, "learning_rate": 9.982802728823195e-06, "loss": 1.6178, "step": 984 }, { "epoch": 0.10480953394339221, "grad_norm": 23.921875, "learning_rate": 9.982767197271178e-06, "loss": 1.4238, "step": 985 }, { "epoch": 0.10491593956160886, "grad_norm": 22.921875, "learning_rate": 9.98273166571916e-06, "loss": 1.2683, "step": 986 }, { "epoch": 0.10502234517982549, "grad_norm": 24.390625, "learning_rate": 9.982696134167142e-06, "loss": 1.496, "step": 987 }, { "epoch": 0.10512875079804214, "grad_norm": 27.734375, "learning_rate": 9.982660602615123e-06, "loss": 1.4518, "step": 988 }, { "epoch": 0.10523515641625877, "grad_norm": 29.90625, "learning_rate": 9.982625071063105e-06, "loss": 1.7962, "step": 989 }, { "epoch": 0.10534156203447542, "grad_norm": 16.578125, "learning_rate": 9.982589539511086e-06, "loss": 1.3292, "step": 990 }, { "epoch": 0.10544796765269206, "grad_norm": 17.828125, "learning_rate": 9.98255400795907e-06, "loss": 1.3434, "step": 991 }, { "epoch": 0.1055543732709087, "grad_norm": 19.53125, "learning_rate": 9.982518476407051e-06, "loss": 1.2384, "step": 992 }, { "epoch": 0.10566077888912534, "grad_norm": 20.3125, "learning_rate": 9.982482944855031e-06, "loss": 1.4051, "step": 993 }, { "epoch": 0.10576718450734199, "grad_norm": 21.84375, "learning_rate": 9.982447413303013e-06, "loss": 1.5675, "step": 994 }, { "epoch": 0.10587359012555862, "grad_norm": 17.03125, "learning_rate": 9.982411881750994e-06, "loss": 1.3238, "step": 995 }, { "epoch": 0.10597999574377527, "grad_norm": 23.65625, "learning_rate": 9.982376350198978e-06, "loss": 1.3463, "step": 996 }, { "epoch": 0.10608640136199192, "grad_norm": 21.65625, "learning_rate": 9.98234081864696e-06, "loss": 1.4368, "step": 997 }, { "epoch": 0.10619280698020855, "grad_norm": 17.09375, "learning_rate": 9.982305287094941e-06, "loss": 1.0666, "step": 998 }, { "epoch": 0.1062992125984252, "grad_norm": 22.734375, "learning_rate": 9.982269755542922e-06, "loss": 1.1735, "step": 999 }, { "epoch": 0.10640561821664184, "grad_norm": 32.28125, "learning_rate": 9.982234223990904e-06, "loss": 1.7242, "step": 1000 }, { "epoch": 0.10651202383485849, "grad_norm": 19.53125, "learning_rate": 9.982198692438887e-06, "loss": 1.3363, "step": 1001 }, { "epoch": 0.10661842945307512, "grad_norm": 21.015625, "learning_rate": 9.982163160886869e-06, "loss": 1.4184, "step": 1002 }, { "epoch": 0.10672483507129177, "grad_norm": 19.71875, "learning_rate": 9.98212762933485e-06, "loss": 1.4676, "step": 1003 }, { "epoch": 0.1068312406895084, "grad_norm": 18.4375, "learning_rate": 9.982092097782832e-06, "loss": 1.3104, "step": 1004 }, { "epoch": 0.10693764630772505, "grad_norm": 24.875, "learning_rate": 9.982056566230814e-06, "loss": 1.497, "step": 1005 }, { "epoch": 0.10704405192594169, "grad_norm": 23.03125, "learning_rate": 9.982021034678795e-06, "loss": 1.5597, "step": 1006 }, { "epoch": 0.10715045754415833, "grad_norm": 30.546875, "learning_rate": 9.981985503126777e-06, "loss": 1.5587, "step": 1007 }, { "epoch": 0.10725686316237497, "grad_norm": 25.765625, "learning_rate": 9.981949971574759e-06, "loss": 1.5724, "step": 1008 }, { "epoch": 0.10736326878059162, "grad_norm": 24.921875, "learning_rate": 9.98191444002274e-06, "loss": 1.3432, "step": 1009 }, { "epoch": 0.10746967439880825, "grad_norm": 17.125, "learning_rate": 9.981878908470722e-06, "loss": 1.2876, "step": 1010 }, { "epoch": 0.1075760800170249, "grad_norm": 24.4375, "learning_rate": 9.981843376918705e-06, "loss": 1.4874, "step": 1011 }, { "epoch": 0.10768248563524153, "grad_norm": 29.46875, "learning_rate": 9.981807845366687e-06, "loss": 1.6182, "step": 1012 }, { "epoch": 0.10778889125345818, "grad_norm": 22.890625, "learning_rate": 9.981772313814668e-06, "loss": 1.5493, "step": 1013 }, { "epoch": 0.10789529687167483, "grad_norm": 25.546875, "learning_rate": 9.98173678226265e-06, "loss": 1.4965, "step": 1014 }, { "epoch": 0.10800170248989147, "grad_norm": 27.4375, "learning_rate": 9.981701250710632e-06, "loss": 1.4162, "step": 1015 }, { "epoch": 0.10810810810810811, "grad_norm": 19.0625, "learning_rate": 9.981665719158615e-06, "loss": 1.3084, "step": 1016 }, { "epoch": 0.10821451372632475, "grad_norm": 16.734375, "learning_rate": 9.981630187606596e-06, "loss": 1.2629, "step": 1017 }, { "epoch": 0.1083209193445414, "grad_norm": 28.484375, "learning_rate": 9.981594656054578e-06, "loss": 1.4442, "step": 1018 }, { "epoch": 0.10842732496275803, "grad_norm": 20.765625, "learning_rate": 9.98155912450256e-06, "loss": 1.3808, "step": 1019 }, { "epoch": 0.10853373058097468, "grad_norm": 19.15625, "learning_rate": 9.981523592950541e-06, "loss": 1.3153, "step": 1020 }, { "epoch": 0.10864013619919131, "grad_norm": 19.84375, "learning_rate": 9.981488061398523e-06, "loss": 1.3554, "step": 1021 }, { "epoch": 0.10874654181740796, "grad_norm": 25.984375, "learning_rate": 9.981452529846505e-06, "loss": 1.2281, "step": 1022 }, { "epoch": 0.1088529474356246, "grad_norm": 19.0, "learning_rate": 9.981416998294486e-06, "loss": 1.1525, "step": 1023 }, { "epoch": 0.10895935305384125, "grad_norm": 17.171875, "learning_rate": 9.981381466742468e-06, "loss": 1.3012, "step": 1024 }, { "epoch": 0.10906575867205788, "grad_norm": 24.46875, "learning_rate": 9.98134593519045e-06, "loss": 1.323, "step": 1025 }, { "epoch": 0.10917216429027453, "grad_norm": 20.859375, "learning_rate": 9.981310403638433e-06, "loss": 1.2905, "step": 1026 }, { "epoch": 0.10927856990849116, "grad_norm": 21.375, "learning_rate": 9.981274872086414e-06, "loss": 1.4884, "step": 1027 }, { "epoch": 0.10938497552670781, "grad_norm": 25.421875, "learning_rate": 9.981239340534396e-06, "loss": 1.3302, "step": 1028 }, { "epoch": 0.10949138114492445, "grad_norm": 20.21875, "learning_rate": 9.981203808982377e-06, "loss": 1.2216, "step": 1029 }, { "epoch": 0.1095977867631411, "grad_norm": 19.09375, "learning_rate": 9.981168277430359e-06, "loss": 1.3092, "step": 1030 }, { "epoch": 0.10970419238135773, "grad_norm": 25.203125, "learning_rate": 9.98113274587834e-06, "loss": 1.2324, "step": 1031 }, { "epoch": 0.10981059799957438, "grad_norm": 18.140625, "learning_rate": 9.981097214326322e-06, "loss": 1.2358, "step": 1032 }, { "epoch": 0.10991700361779103, "grad_norm": 25.90625, "learning_rate": 9.981061682774304e-06, "loss": 1.4794, "step": 1033 }, { "epoch": 0.11002340923600766, "grad_norm": 25.0, "learning_rate": 9.981026151222286e-06, "loss": 1.7195, "step": 1034 }, { "epoch": 0.11012981485422431, "grad_norm": 42.0, "learning_rate": 9.980990619670267e-06, "loss": 1.474, "step": 1035 }, { "epoch": 0.11023622047244094, "grad_norm": 30.125, "learning_rate": 9.980955088118249e-06, "loss": 1.5859, "step": 1036 }, { "epoch": 0.11034262609065759, "grad_norm": 26.578125, "learning_rate": 9.980919556566232e-06, "loss": 1.6999, "step": 1037 }, { "epoch": 0.11044903170887423, "grad_norm": 32.21875, "learning_rate": 9.980884025014214e-06, "loss": 1.2302, "step": 1038 }, { "epoch": 0.11055543732709087, "grad_norm": 202.625, "learning_rate": 9.980848493462195e-06, "loss": 1.3931, "step": 1039 }, { "epoch": 0.11066184294530751, "grad_norm": 23.21875, "learning_rate": 9.980812961910177e-06, "loss": 1.404, "step": 1040 }, { "epoch": 0.11076824856352416, "grad_norm": 25.046875, "learning_rate": 9.980777430358158e-06, "loss": 1.5524, "step": 1041 }, { "epoch": 0.11087465418174079, "grad_norm": 21.5625, "learning_rate": 9.980741898806142e-06, "loss": 1.4278, "step": 1042 }, { "epoch": 0.11098105979995744, "grad_norm": 29.953125, "learning_rate": 9.980706367254123e-06, "loss": 1.4579, "step": 1043 }, { "epoch": 0.11108746541817408, "grad_norm": 20.640625, "learning_rate": 9.980670835702105e-06, "loss": 1.3537, "step": 1044 }, { "epoch": 0.11119387103639072, "grad_norm": 18.234375, "learning_rate": 9.980635304150087e-06, "loss": 1.4443, "step": 1045 }, { "epoch": 0.11130027665460736, "grad_norm": 35.28125, "learning_rate": 9.980599772598068e-06, "loss": 1.7706, "step": 1046 }, { "epoch": 0.111406682272824, "grad_norm": 23.859375, "learning_rate": 9.98056424104605e-06, "loss": 1.408, "step": 1047 }, { "epoch": 0.11151308789104064, "grad_norm": 21.484375, "learning_rate": 9.980528709494031e-06, "loss": 1.2259, "step": 1048 }, { "epoch": 0.11161949350925729, "grad_norm": 23.84375, "learning_rate": 9.980493177942013e-06, "loss": 1.3025, "step": 1049 }, { "epoch": 0.11172589912747392, "grad_norm": 20.921875, "learning_rate": 9.980457646389995e-06, "loss": 1.4871, "step": 1050 }, { "epoch": 0.11183230474569057, "grad_norm": 19.703125, "learning_rate": 9.980422114837976e-06, "loss": 1.4633, "step": 1051 }, { "epoch": 0.11193871036390722, "grad_norm": 28.90625, "learning_rate": 9.98038658328596e-06, "loss": 1.139, "step": 1052 }, { "epoch": 0.11204511598212386, "grad_norm": 22.28125, "learning_rate": 9.980351051733941e-06, "loss": 1.1735, "step": 1053 }, { "epoch": 0.1121515216003405, "grad_norm": 17.109375, "learning_rate": 9.980315520181923e-06, "loss": 1.3284, "step": 1054 }, { "epoch": 0.11225792721855714, "grad_norm": 32.34375, "learning_rate": 9.980279988629904e-06, "loss": 1.4947, "step": 1055 }, { "epoch": 0.11236433283677379, "grad_norm": 32.84375, "learning_rate": 9.980244457077886e-06, "loss": 1.6025, "step": 1056 }, { "epoch": 0.11247073845499042, "grad_norm": 17.59375, "learning_rate": 9.980208925525868e-06, "loss": 1.0727, "step": 1057 }, { "epoch": 0.11257714407320707, "grad_norm": 17.78125, "learning_rate": 9.980173393973849e-06, "loss": 1.0447, "step": 1058 }, { "epoch": 0.1126835496914237, "grad_norm": 36.46875, "learning_rate": 9.98013786242183e-06, "loss": 1.7277, "step": 1059 }, { "epoch": 0.11278995530964035, "grad_norm": 29.21875, "learning_rate": 9.980102330869812e-06, "loss": 1.3786, "step": 1060 }, { "epoch": 0.11289636092785699, "grad_norm": 17.59375, "learning_rate": 9.980066799317794e-06, "loss": 1.2596, "step": 1061 }, { "epoch": 0.11300276654607364, "grad_norm": 18.546875, "learning_rate": 9.980031267765777e-06, "loss": 1.3461, "step": 1062 }, { "epoch": 0.11310917216429027, "grad_norm": 44.625, "learning_rate": 9.979995736213759e-06, "loss": 1.5893, "step": 1063 }, { "epoch": 0.11321557778250692, "grad_norm": 23.828125, "learning_rate": 9.97996020466174e-06, "loss": 1.3202, "step": 1064 }, { "epoch": 0.11332198340072355, "grad_norm": 21.265625, "learning_rate": 9.979924673109722e-06, "loss": 1.694, "step": 1065 }, { "epoch": 0.1134283890189402, "grad_norm": 20.65625, "learning_rate": 9.979889141557704e-06, "loss": 1.4836, "step": 1066 }, { "epoch": 0.11353479463715684, "grad_norm": 22.265625, "learning_rate": 9.979853610005685e-06, "loss": 1.2525, "step": 1067 }, { "epoch": 0.11364120025537348, "grad_norm": 20.59375, "learning_rate": 9.979818078453669e-06, "loss": 1.2077, "step": 1068 }, { "epoch": 0.11374760587359013, "grad_norm": 21.546875, "learning_rate": 9.97978254690165e-06, "loss": 1.2625, "step": 1069 }, { "epoch": 0.11385401149180677, "grad_norm": 15.7109375, "learning_rate": 9.979747015349632e-06, "loss": 1.431, "step": 1070 }, { "epoch": 0.11396041711002342, "grad_norm": 21.953125, "learning_rate": 9.979711483797613e-06, "loss": 1.1826, "step": 1071 }, { "epoch": 0.11406682272824005, "grad_norm": 20.328125, "learning_rate": 9.979675952245595e-06, "loss": 1.7132, "step": 1072 }, { "epoch": 0.1141732283464567, "grad_norm": 20.609375, "learning_rate": 9.979640420693577e-06, "loss": 1.3619, "step": 1073 }, { "epoch": 0.11427963396467333, "grad_norm": 17.890625, "learning_rate": 9.979604889141558e-06, "loss": 1.3731, "step": 1074 }, { "epoch": 0.11438603958288998, "grad_norm": 23.0625, "learning_rate": 9.97956935758954e-06, "loss": 1.4085, "step": 1075 }, { "epoch": 0.11449244520110662, "grad_norm": 26.375, "learning_rate": 9.979533826037521e-06, "loss": 1.4483, "step": 1076 }, { "epoch": 0.11459885081932326, "grad_norm": 18.203125, "learning_rate": 9.979498294485503e-06, "loss": 1.1043, "step": 1077 }, { "epoch": 0.1147052564375399, "grad_norm": 11.4296875, "learning_rate": 9.979462762933486e-06, "loss": 1.0491, "step": 1078 }, { "epoch": 0.11481166205575655, "grad_norm": 23.890625, "learning_rate": 9.979427231381468e-06, "loss": 1.542, "step": 1079 }, { "epoch": 0.11491806767397318, "grad_norm": 26.328125, "learning_rate": 9.97939169982945e-06, "loss": 1.5782, "step": 1080 }, { "epoch": 0.11502447329218983, "grad_norm": 22.78125, "learning_rate": 9.979356168277431e-06, "loss": 1.3432, "step": 1081 }, { "epoch": 0.11513087891040646, "grad_norm": 24.28125, "learning_rate": 9.979320636725413e-06, "loss": 1.2195, "step": 1082 }, { "epoch": 0.11523728452862311, "grad_norm": 23.453125, "learning_rate": 9.979285105173394e-06, "loss": 1.4505, "step": 1083 }, { "epoch": 0.11534369014683975, "grad_norm": 26.078125, "learning_rate": 9.979249573621376e-06, "loss": 1.3739, "step": 1084 }, { "epoch": 0.1154500957650564, "grad_norm": 23.453125, "learning_rate": 9.979214042069358e-06, "loss": 1.2708, "step": 1085 }, { "epoch": 0.11555650138327303, "grad_norm": 23.609375, "learning_rate": 9.97917851051734e-06, "loss": 1.6466, "step": 1086 }, { "epoch": 0.11566290700148968, "grad_norm": 13.7578125, "learning_rate": 9.97914297896532e-06, "loss": 1.0885, "step": 1087 }, { "epoch": 0.11576931261970633, "grad_norm": 22.5, "learning_rate": 9.979107447413304e-06, "loss": 1.5215, "step": 1088 }, { "epoch": 0.11587571823792296, "grad_norm": 18.046875, "learning_rate": 9.979071915861286e-06, "loss": 1.1835, "step": 1089 }, { "epoch": 0.11598212385613961, "grad_norm": 22.359375, "learning_rate": 9.979036384309267e-06, "loss": 1.4829, "step": 1090 }, { "epoch": 0.11608852947435624, "grad_norm": 19.65625, "learning_rate": 9.979000852757249e-06, "loss": 1.2873, "step": 1091 }, { "epoch": 0.11619493509257289, "grad_norm": 22.671875, "learning_rate": 9.97896532120523e-06, "loss": 1.4433, "step": 1092 }, { "epoch": 0.11630134071078953, "grad_norm": 20.5, "learning_rate": 9.978929789653214e-06, "loss": 1.5581, "step": 1093 }, { "epoch": 0.11640774632900618, "grad_norm": 23.171875, "learning_rate": 9.978894258101195e-06, "loss": 1.4907, "step": 1094 }, { "epoch": 0.11651415194722281, "grad_norm": 26.546875, "learning_rate": 9.978858726549177e-06, "loss": 1.5516, "step": 1095 }, { "epoch": 0.11662055756543946, "grad_norm": 23.8125, "learning_rate": 9.978823194997159e-06, "loss": 1.3544, "step": 1096 }, { "epoch": 0.1167269631836561, "grad_norm": 19.0625, "learning_rate": 9.97878766344514e-06, "loss": 1.3704, "step": 1097 }, { "epoch": 0.11683336880187274, "grad_norm": 18.0, "learning_rate": 9.978752131893122e-06, "loss": 1.2606, "step": 1098 }, { "epoch": 0.11693977442008938, "grad_norm": 18.03125, "learning_rate": 9.978716600341103e-06, "loss": 1.2108, "step": 1099 }, { "epoch": 0.11704618003830602, "grad_norm": 18.46875, "learning_rate": 9.978681068789085e-06, "loss": 1.2777, "step": 1100 }, { "epoch": 0.11715258565652266, "grad_norm": 19.859375, "learning_rate": 9.978645537237067e-06, "loss": 1.3777, "step": 1101 }, { "epoch": 0.11725899127473931, "grad_norm": 20.203125, "learning_rate": 9.978610005685048e-06, "loss": 1.3371, "step": 1102 }, { "epoch": 0.11736539689295594, "grad_norm": 20.03125, "learning_rate": 9.978574474133032e-06, "loss": 1.0627, "step": 1103 }, { "epoch": 0.11747180251117259, "grad_norm": 23.65625, "learning_rate": 9.978538942581013e-06, "loss": 1.274, "step": 1104 }, { "epoch": 0.11757820812938923, "grad_norm": 14.6796875, "learning_rate": 9.978503411028995e-06, "loss": 1.2434, "step": 1105 }, { "epoch": 0.11768461374760587, "grad_norm": 18.609375, "learning_rate": 9.978467879476976e-06, "loss": 1.0096, "step": 1106 }, { "epoch": 0.11779101936582252, "grad_norm": 25.21875, "learning_rate": 9.978432347924958e-06, "loss": 1.5322, "step": 1107 }, { "epoch": 0.11789742498403916, "grad_norm": 17.046875, "learning_rate": 9.97839681637294e-06, "loss": 1.2314, "step": 1108 }, { "epoch": 0.1180038306022558, "grad_norm": 26.921875, "learning_rate": 9.978361284820923e-06, "loss": 1.3638, "step": 1109 }, { "epoch": 0.11811023622047244, "grad_norm": 39.40625, "learning_rate": 9.978325753268905e-06, "loss": 1.6928, "step": 1110 }, { "epoch": 0.11821664183868909, "grad_norm": 18.71875, "learning_rate": 9.978290221716886e-06, "loss": 1.0586, "step": 1111 }, { "epoch": 0.11832304745690572, "grad_norm": 24.75, "learning_rate": 9.978254690164868e-06, "loss": 1.1986, "step": 1112 }, { "epoch": 0.11842945307512237, "grad_norm": 23.8125, "learning_rate": 9.978219158612848e-06, "loss": 1.1558, "step": 1113 }, { "epoch": 0.118535858693339, "grad_norm": 21.59375, "learning_rate": 9.978183627060831e-06, "loss": 1.3088, "step": 1114 }, { "epoch": 0.11864226431155565, "grad_norm": 23.484375, "learning_rate": 9.978148095508813e-06, "loss": 1.4563, "step": 1115 }, { "epoch": 0.11874866992977229, "grad_norm": 17.828125, "learning_rate": 9.978112563956794e-06, "loss": 1.2553, "step": 1116 }, { "epoch": 0.11885507554798894, "grad_norm": 20.109375, "learning_rate": 9.978077032404776e-06, "loss": 1.4479, "step": 1117 }, { "epoch": 0.11896148116620557, "grad_norm": 20.609375, "learning_rate": 9.978041500852757e-06, "loss": 1.4489, "step": 1118 }, { "epoch": 0.11906788678442222, "grad_norm": 17.40625, "learning_rate": 9.97800596930074e-06, "loss": 1.3424, "step": 1119 }, { "epoch": 0.11917429240263885, "grad_norm": 18.890625, "learning_rate": 9.977970437748722e-06, "loss": 1.1419, "step": 1120 }, { "epoch": 0.1192806980208555, "grad_norm": 20.265625, "learning_rate": 9.977934906196704e-06, "loss": 1.0918, "step": 1121 }, { "epoch": 0.11938710363907214, "grad_norm": 22.59375, "learning_rate": 9.977899374644686e-06, "loss": 1.4153, "step": 1122 }, { "epoch": 0.11949350925728879, "grad_norm": 15.140625, "learning_rate": 9.977863843092667e-06, "loss": 1.2484, "step": 1123 }, { "epoch": 0.11959991487550542, "grad_norm": 22.296875, "learning_rate": 9.977828311540649e-06, "loss": 1.3212, "step": 1124 }, { "epoch": 0.11970632049372207, "grad_norm": 35.0625, "learning_rate": 9.97779277998863e-06, "loss": 1.4001, "step": 1125 }, { "epoch": 0.11981272611193872, "grad_norm": 19.46875, "learning_rate": 9.977757248436612e-06, "loss": 1.4193, "step": 1126 }, { "epoch": 0.11991913173015535, "grad_norm": 14.046875, "learning_rate": 9.977721716884594e-06, "loss": 1.1288, "step": 1127 }, { "epoch": 0.120025537348372, "grad_norm": 20.328125, "learning_rate": 9.977686185332575e-06, "loss": 1.3214, "step": 1128 }, { "epoch": 0.12013194296658863, "grad_norm": 27.3125, "learning_rate": 9.977650653780558e-06, "loss": 1.5589, "step": 1129 }, { "epoch": 0.12023834858480528, "grad_norm": 26.125, "learning_rate": 9.97761512222854e-06, "loss": 1.7016, "step": 1130 }, { "epoch": 0.12034475420302192, "grad_norm": 26.484375, "learning_rate": 9.977579590676522e-06, "loss": 1.7956, "step": 1131 }, { "epoch": 0.12045115982123857, "grad_norm": 17.828125, "learning_rate": 9.977544059124503e-06, "loss": 1.2279, "step": 1132 }, { "epoch": 0.1205575654394552, "grad_norm": 25.96875, "learning_rate": 9.977508527572485e-06, "loss": 1.6236, "step": 1133 }, { "epoch": 0.12066397105767185, "grad_norm": 14.90625, "learning_rate": 9.977472996020468e-06, "loss": 1.1826, "step": 1134 }, { "epoch": 0.12077037667588848, "grad_norm": 20.71875, "learning_rate": 9.97743746446845e-06, "loss": 1.2596, "step": 1135 }, { "epoch": 0.12087678229410513, "grad_norm": 19.859375, "learning_rate": 9.977401932916431e-06, "loss": 1.288, "step": 1136 }, { "epoch": 0.12098318791232177, "grad_norm": 23.84375, "learning_rate": 9.977366401364413e-06, "loss": 1.3699, "step": 1137 }, { "epoch": 0.12108959353053841, "grad_norm": 12.40625, "learning_rate": 9.977330869812395e-06, "loss": 0.9127, "step": 1138 }, { "epoch": 0.12119599914875505, "grad_norm": 23.171875, "learning_rate": 9.977295338260376e-06, "loss": 1.4154, "step": 1139 }, { "epoch": 0.1213024047669717, "grad_norm": 23.140625, "learning_rate": 9.977259806708358e-06, "loss": 1.3466, "step": 1140 }, { "epoch": 0.12140881038518833, "grad_norm": 28.578125, "learning_rate": 9.97722427515634e-06, "loss": 1.3966, "step": 1141 }, { "epoch": 0.12151521600340498, "grad_norm": 17.046875, "learning_rate": 9.977188743604321e-06, "loss": 1.3234, "step": 1142 }, { "epoch": 0.12162162162162163, "grad_norm": 18.34375, "learning_rate": 9.977153212052303e-06, "loss": 1.0948, "step": 1143 }, { "epoch": 0.12172802723983826, "grad_norm": 15.8046875, "learning_rate": 9.977117680500286e-06, "loss": 1.2333, "step": 1144 }, { "epoch": 0.12183443285805491, "grad_norm": 19.234375, "learning_rate": 9.977082148948268e-06, "loss": 1.3612, "step": 1145 }, { "epoch": 0.12194083847627155, "grad_norm": 17.90625, "learning_rate": 9.97704661739625e-06, "loss": 1.2151, "step": 1146 }, { "epoch": 0.1220472440944882, "grad_norm": 22.46875, "learning_rate": 9.97701108584423e-06, "loss": 1.7433, "step": 1147 }, { "epoch": 0.12215364971270483, "grad_norm": 19.953125, "learning_rate": 9.976975554292212e-06, "loss": 1.2477, "step": 1148 }, { "epoch": 0.12226005533092148, "grad_norm": 18.921875, "learning_rate": 9.976940022740194e-06, "loss": 1.4829, "step": 1149 }, { "epoch": 0.12236646094913811, "grad_norm": 18.828125, "learning_rate": 9.976904491188176e-06, "loss": 1.4474, "step": 1150 }, { "epoch": 0.12247286656735476, "grad_norm": 20.40625, "learning_rate": 9.976868959636157e-06, "loss": 1.3934, "step": 1151 }, { "epoch": 0.1225792721855714, "grad_norm": 18.5, "learning_rate": 9.976833428084139e-06, "loss": 1.5313, "step": 1152 }, { "epoch": 0.12268567780378804, "grad_norm": 27.0, "learning_rate": 9.97679789653212e-06, "loss": 1.5415, "step": 1153 }, { "epoch": 0.12279208342200468, "grad_norm": 16.40625, "learning_rate": 9.976762364980102e-06, "loss": 1.4213, "step": 1154 }, { "epoch": 0.12289848904022133, "grad_norm": 22.59375, "learning_rate": 9.976726833428085e-06, "loss": 1.441, "step": 1155 }, { "epoch": 0.12300489465843796, "grad_norm": 19.234375, "learning_rate": 9.976691301876067e-06, "loss": 1.0866, "step": 1156 }, { "epoch": 0.12311130027665461, "grad_norm": 21.46875, "learning_rate": 9.976655770324049e-06, "loss": 1.463, "step": 1157 }, { "epoch": 0.12321770589487124, "grad_norm": 17.96875, "learning_rate": 9.97662023877203e-06, "loss": 1.1255, "step": 1158 }, { "epoch": 0.12332411151308789, "grad_norm": 17.53125, "learning_rate": 9.976584707220012e-06, "loss": 1.1468, "step": 1159 }, { "epoch": 0.12343051713130453, "grad_norm": 19.859375, "learning_rate": 9.976549175667995e-06, "loss": 1.3441, "step": 1160 }, { "epoch": 0.12353692274952117, "grad_norm": 30.765625, "learning_rate": 9.976513644115977e-06, "loss": 1.5314, "step": 1161 }, { "epoch": 0.12364332836773782, "grad_norm": 16.640625, "learning_rate": 9.976478112563958e-06, "loss": 1.3759, "step": 1162 }, { "epoch": 0.12374973398595446, "grad_norm": 17.109375, "learning_rate": 9.97644258101194e-06, "loss": 1.2261, "step": 1163 }, { "epoch": 0.1238561396041711, "grad_norm": 18.53125, "learning_rate": 9.976407049459921e-06, "loss": 1.3313, "step": 1164 }, { "epoch": 0.12396254522238774, "grad_norm": 16.4375, "learning_rate": 9.976371517907903e-06, "loss": 1.4313, "step": 1165 }, { "epoch": 0.12406895084060439, "grad_norm": 17.671875, "learning_rate": 9.976335986355885e-06, "loss": 1.4989, "step": 1166 }, { "epoch": 0.12417535645882102, "grad_norm": 22.515625, "learning_rate": 9.976300454803866e-06, "loss": 1.5457, "step": 1167 }, { "epoch": 0.12428176207703767, "grad_norm": 18.25, "learning_rate": 9.976264923251848e-06, "loss": 1.2766, "step": 1168 }, { "epoch": 0.1243881676952543, "grad_norm": 17.1875, "learning_rate": 9.97622939169983e-06, "loss": 1.141, "step": 1169 }, { "epoch": 0.12449457331347095, "grad_norm": 14.78125, "learning_rate": 9.976193860147813e-06, "loss": 1.2585, "step": 1170 }, { "epoch": 0.12460097893168759, "grad_norm": 19.84375, "learning_rate": 9.976158328595794e-06, "loss": 1.7844, "step": 1171 }, { "epoch": 0.12470738454990424, "grad_norm": 31.15625, "learning_rate": 9.976122797043776e-06, "loss": 1.685, "step": 1172 }, { "epoch": 0.12481379016812087, "grad_norm": 15.640625, "learning_rate": 9.976087265491758e-06, "loss": 1.1836, "step": 1173 }, { "epoch": 0.12492019578633752, "grad_norm": 26.625, "learning_rate": 9.97605173393974e-06, "loss": 1.4716, "step": 1174 }, { "epoch": 0.12502660140455416, "grad_norm": 15.828125, "learning_rate": 9.97601620238772e-06, "loss": 1.2717, "step": 1175 }, { "epoch": 0.1251330070227708, "grad_norm": 18.46875, "learning_rate": 9.975980670835702e-06, "loss": 1.5501, "step": 1176 }, { "epoch": 0.12523941264098745, "grad_norm": 14.5625, "learning_rate": 9.975945139283684e-06, "loss": 1.2844, "step": 1177 }, { "epoch": 0.12534581825920407, "grad_norm": 26.578125, "learning_rate": 9.975909607731666e-06, "loss": 1.5662, "step": 1178 }, { "epoch": 0.12545222387742072, "grad_norm": 17.953125, "learning_rate": 9.975874076179647e-06, "loss": 1.6445, "step": 1179 }, { "epoch": 0.12555862949563737, "grad_norm": 24.1875, "learning_rate": 9.97583854462763e-06, "loss": 1.2713, "step": 1180 }, { "epoch": 0.12566503511385402, "grad_norm": 18.515625, "learning_rate": 9.975803013075612e-06, "loss": 1.4538, "step": 1181 }, { "epoch": 0.12577144073207067, "grad_norm": 17.140625, "learning_rate": 9.975767481523594e-06, "loss": 1.3107, "step": 1182 }, { "epoch": 0.1258778463502873, "grad_norm": 18.296875, "learning_rate": 9.975731949971575e-06, "loss": 1.1265, "step": 1183 }, { "epoch": 0.12598425196850394, "grad_norm": 18.4375, "learning_rate": 9.975696418419557e-06, "loss": 1.3627, "step": 1184 }, { "epoch": 0.12609065758672058, "grad_norm": 18.09375, "learning_rate": 9.975660886867539e-06, "loss": 1.174, "step": 1185 }, { "epoch": 0.12619706320493723, "grad_norm": 14.875, "learning_rate": 9.975625355315522e-06, "loss": 1.2358, "step": 1186 }, { "epoch": 0.12630346882315385, "grad_norm": 26.375, "learning_rate": 9.975589823763504e-06, "loss": 1.4037, "step": 1187 }, { "epoch": 0.1264098744413705, "grad_norm": 24.890625, "learning_rate": 9.975554292211485e-06, "loss": 1.3531, "step": 1188 }, { "epoch": 0.12651628005958715, "grad_norm": 17.03125, "learning_rate": 9.975518760659467e-06, "loss": 1.3176, "step": 1189 }, { "epoch": 0.1266226856778038, "grad_norm": 20.578125, "learning_rate": 9.975483229107448e-06, "loss": 1.4659, "step": 1190 }, { "epoch": 0.12672909129602042, "grad_norm": 17.546875, "learning_rate": 9.97544769755543e-06, "loss": 1.1966, "step": 1191 }, { "epoch": 0.12683549691423707, "grad_norm": 30.34375, "learning_rate": 9.975412166003412e-06, "loss": 1.6924, "step": 1192 }, { "epoch": 0.12694190253245372, "grad_norm": 17.6875, "learning_rate": 9.975376634451393e-06, "loss": 1.5795, "step": 1193 }, { "epoch": 0.12704830815067036, "grad_norm": 18.78125, "learning_rate": 9.975341102899375e-06, "loss": 1.2287, "step": 1194 }, { "epoch": 0.12715471376888698, "grad_norm": 22.203125, "learning_rate": 9.975305571347356e-06, "loss": 1.3103, "step": 1195 }, { "epoch": 0.12726111938710363, "grad_norm": 21.78125, "learning_rate": 9.97527003979534e-06, "loss": 1.396, "step": 1196 }, { "epoch": 0.12736752500532028, "grad_norm": 38.46875, "learning_rate": 9.975234508243321e-06, "loss": 1.1159, "step": 1197 }, { "epoch": 0.12747393062353693, "grad_norm": 18.609375, "learning_rate": 9.975198976691303e-06, "loss": 1.4525, "step": 1198 }, { "epoch": 0.12758033624175358, "grad_norm": 16.6875, "learning_rate": 9.975163445139284e-06, "loss": 1.3408, "step": 1199 }, { "epoch": 0.1276867418599702, "grad_norm": 17.71875, "learning_rate": 9.975127913587266e-06, "loss": 1.0607, "step": 1200 }, { "epoch": 0.12779314747818685, "grad_norm": 18.671875, "learning_rate": 9.97509238203525e-06, "loss": 1.2188, "step": 1201 }, { "epoch": 0.1278995530964035, "grad_norm": 23.1875, "learning_rate": 9.97505685048323e-06, "loss": 1.4514, "step": 1202 }, { "epoch": 0.12800595871462014, "grad_norm": 26.640625, "learning_rate": 9.975021318931211e-06, "loss": 1.191, "step": 1203 }, { "epoch": 0.12811236433283676, "grad_norm": 17.21875, "learning_rate": 9.974985787379193e-06, "loss": 1.3881, "step": 1204 }, { "epoch": 0.1282187699510534, "grad_norm": 20.78125, "learning_rate": 9.974950255827174e-06, "loss": 1.291, "step": 1205 }, { "epoch": 0.12832517556927006, "grad_norm": 35.125, "learning_rate": 9.974914724275157e-06, "loss": 1.5169, "step": 1206 }, { "epoch": 0.1284315811874867, "grad_norm": 19.609375, "learning_rate": 9.974879192723139e-06, "loss": 1.2514, "step": 1207 }, { "epoch": 0.12853798680570333, "grad_norm": 20.625, "learning_rate": 9.97484366117112e-06, "loss": 1.4367, "step": 1208 }, { "epoch": 0.12864439242391998, "grad_norm": 30.1875, "learning_rate": 9.974808129619102e-06, "loss": 1.3697, "step": 1209 }, { "epoch": 0.12875079804213663, "grad_norm": 14.015625, "learning_rate": 9.974772598067084e-06, "loss": 1.0558, "step": 1210 }, { "epoch": 0.12885720366035328, "grad_norm": 20.015625, "learning_rate": 9.974737066515067e-06, "loss": 1.3022, "step": 1211 }, { "epoch": 0.1289636092785699, "grad_norm": 29.5625, "learning_rate": 9.974701534963049e-06, "loss": 1.265, "step": 1212 }, { "epoch": 0.12907001489678654, "grad_norm": 19.78125, "learning_rate": 9.97466600341103e-06, "loss": 1.5147, "step": 1213 }, { "epoch": 0.1291764205150032, "grad_norm": 23.1875, "learning_rate": 9.974630471859012e-06, "loss": 1.4274, "step": 1214 }, { "epoch": 0.12928282613321984, "grad_norm": 17.359375, "learning_rate": 9.974594940306994e-06, "loss": 1.2146, "step": 1215 }, { "epoch": 0.12938923175143646, "grad_norm": 20.703125, "learning_rate": 9.974559408754975e-06, "loss": 1.3807, "step": 1216 }, { "epoch": 0.1294956373696531, "grad_norm": 16.75, "learning_rate": 9.974523877202957e-06, "loss": 1.5332, "step": 1217 }, { "epoch": 0.12960204298786976, "grad_norm": 18.609375, "learning_rate": 9.974488345650938e-06, "loss": 1.0914, "step": 1218 }, { "epoch": 0.1297084486060864, "grad_norm": 18.40625, "learning_rate": 9.97445281409892e-06, "loss": 1.4052, "step": 1219 }, { "epoch": 0.12981485422430306, "grad_norm": 30.75, "learning_rate": 9.974417282546902e-06, "loss": 1.4025, "step": 1220 }, { "epoch": 0.12992125984251968, "grad_norm": 23.078125, "learning_rate": 9.974381750994885e-06, "loss": 1.5632, "step": 1221 }, { "epoch": 0.13002766546073632, "grad_norm": 17.984375, "learning_rate": 9.974346219442867e-06, "loss": 1.2978, "step": 1222 }, { "epoch": 0.13013407107895297, "grad_norm": 32.46875, "learning_rate": 9.974310687890848e-06, "loss": 1.5438, "step": 1223 }, { "epoch": 0.13024047669716962, "grad_norm": 17.78125, "learning_rate": 9.97427515633883e-06, "loss": 1.0281, "step": 1224 }, { "epoch": 0.13034688231538624, "grad_norm": 28.59375, "learning_rate": 9.974239624786811e-06, "loss": 1.5529, "step": 1225 }, { "epoch": 0.1304532879336029, "grad_norm": 19.6875, "learning_rate": 9.974204093234793e-06, "loss": 1.4619, "step": 1226 }, { "epoch": 0.13055969355181954, "grad_norm": 19.71875, "learning_rate": 9.974168561682776e-06, "loss": 1.3521, "step": 1227 }, { "epoch": 0.1306660991700362, "grad_norm": 30.84375, "learning_rate": 9.974133030130758e-06, "loss": 1.2595, "step": 1228 }, { "epoch": 0.1307725047882528, "grad_norm": 21.796875, "learning_rate": 9.97409749857874e-06, "loss": 1.3663, "step": 1229 }, { "epoch": 0.13087891040646946, "grad_norm": 19.0625, "learning_rate": 9.974061967026721e-06, "loss": 1.1598, "step": 1230 }, { "epoch": 0.1309853160246861, "grad_norm": 23.078125, "learning_rate": 9.974026435474703e-06, "loss": 1.6317, "step": 1231 }, { "epoch": 0.13109172164290275, "grad_norm": 29.4375, "learning_rate": 9.973990903922684e-06, "loss": 1.4266, "step": 1232 }, { "epoch": 0.13119812726111937, "grad_norm": 23.609375, "learning_rate": 9.973955372370666e-06, "loss": 1.2188, "step": 1233 }, { "epoch": 0.13130453287933602, "grad_norm": 18.609375, "learning_rate": 9.973919840818647e-06, "loss": 1.1035, "step": 1234 }, { "epoch": 0.13141093849755267, "grad_norm": 22.78125, "learning_rate": 9.973884309266629e-06, "loss": 1.3913, "step": 1235 }, { "epoch": 0.13151734411576932, "grad_norm": 23.296875, "learning_rate": 9.97384877771461e-06, "loss": 1.7561, "step": 1236 }, { "epoch": 0.13162374973398597, "grad_norm": 25.96875, "learning_rate": 9.973813246162594e-06, "loss": 1.497, "step": 1237 }, { "epoch": 0.1317301553522026, "grad_norm": 16.71875, "learning_rate": 9.973777714610576e-06, "loss": 1.3166, "step": 1238 }, { "epoch": 0.13183656097041924, "grad_norm": 21.59375, "learning_rate": 9.973742183058557e-06, "loss": 1.3858, "step": 1239 }, { "epoch": 0.13194296658863588, "grad_norm": 20.5625, "learning_rate": 9.973706651506539e-06, "loss": 1.3417, "step": 1240 }, { "epoch": 0.13204937220685253, "grad_norm": 20.734375, "learning_rate": 9.97367111995452e-06, "loss": 1.1159, "step": 1241 }, { "epoch": 0.13215577782506915, "grad_norm": 16.40625, "learning_rate": 9.973635588402502e-06, "loss": 1.3462, "step": 1242 }, { "epoch": 0.1322621834432858, "grad_norm": 21.109375, "learning_rate": 9.973600056850484e-06, "loss": 1.6536, "step": 1243 }, { "epoch": 0.13236858906150245, "grad_norm": 25.09375, "learning_rate": 9.973564525298465e-06, "loss": 1.5842, "step": 1244 }, { "epoch": 0.1324749946797191, "grad_norm": 18.0, "learning_rate": 9.973528993746447e-06, "loss": 1.2033, "step": 1245 }, { "epoch": 0.13258140029793572, "grad_norm": 19.75, "learning_rate": 9.973493462194428e-06, "loss": 1.4561, "step": 1246 }, { "epoch": 0.13268780591615237, "grad_norm": 21.015625, "learning_rate": 9.973457930642412e-06, "loss": 1.5741, "step": 1247 }, { "epoch": 0.13279421153436902, "grad_norm": 17.71875, "learning_rate": 9.973422399090393e-06, "loss": 1.4269, "step": 1248 }, { "epoch": 0.13290061715258566, "grad_norm": 19.5, "learning_rate": 9.973386867538375e-06, "loss": 1.3159, "step": 1249 }, { "epoch": 0.13300702277080229, "grad_norm": 29.796875, "learning_rate": 9.973351335986357e-06, "loss": 1.7142, "step": 1250 }, { "epoch": 0.13311342838901893, "grad_norm": 27.21875, "learning_rate": 9.973315804434338e-06, "loss": 1.2419, "step": 1251 }, { "epoch": 0.13321983400723558, "grad_norm": 20.46875, "learning_rate": 9.973280272882321e-06, "loss": 1.3644, "step": 1252 }, { "epoch": 0.13332623962545223, "grad_norm": 31.15625, "learning_rate": 9.973244741330303e-06, "loss": 1.441, "step": 1253 }, { "epoch": 0.13343264524366888, "grad_norm": 23.1875, "learning_rate": 9.973209209778285e-06, "loss": 1.4729, "step": 1254 }, { "epoch": 0.1335390508618855, "grad_norm": 22.171875, "learning_rate": 9.973173678226266e-06, "loss": 1.5601, "step": 1255 }, { "epoch": 0.13364545648010215, "grad_norm": 23.1875, "learning_rate": 9.973138146674248e-06, "loss": 1.4814, "step": 1256 }, { "epoch": 0.1337518620983188, "grad_norm": 19.09375, "learning_rate": 9.97310261512223e-06, "loss": 1.4435, "step": 1257 }, { "epoch": 0.13385826771653545, "grad_norm": 36.59375, "learning_rate": 9.973067083570211e-06, "loss": 1.2519, "step": 1258 }, { "epoch": 0.13396467333475207, "grad_norm": 39.46875, "learning_rate": 9.973031552018193e-06, "loss": 1.6997, "step": 1259 }, { "epoch": 0.13407107895296871, "grad_norm": 20.921875, "learning_rate": 9.972996020466174e-06, "loss": 1.3269, "step": 1260 }, { "epoch": 0.13417748457118536, "grad_norm": 21.125, "learning_rate": 9.972960488914156e-06, "loss": 1.3866, "step": 1261 }, { "epoch": 0.134283890189402, "grad_norm": 17.984375, "learning_rate": 9.97292495736214e-06, "loss": 1.297, "step": 1262 }, { "epoch": 0.13439029580761863, "grad_norm": 16.109375, "learning_rate": 9.972889425810121e-06, "loss": 1.1859, "step": 1263 }, { "epoch": 0.13449670142583528, "grad_norm": 31.828125, "learning_rate": 9.972853894258102e-06, "loss": 1.509, "step": 1264 }, { "epoch": 0.13460310704405193, "grad_norm": 17.734375, "learning_rate": 9.972818362706084e-06, "loss": 1.3877, "step": 1265 }, { "epoch": 0.13470951266226858, "grad_norm": 33.125, "learning_rate": 9.972782831154066e-06, "loss": 1.7356, "step": 1266 }, { "epoch": 0.1348159182804852, "grad_norm": 23.46875, "learning_rate": 9.972747299602047e-06, "loss": 1.6174, "step": 1267 }, { "epoch": 0.13492232389870185, "grad_norm": 27.046875, "learning_rate": 9.972711768050029e-06, "loss": 1.4099, "step": 1268 }, { "epoch": 0.1350287295169185, "grad_norm": 17.0625, "learning_rate": 9.97267623649801e-06, "loss": 1.2944, "step": 1269 }, { "epoch": 0.13513513513513514, "grad_norm": 20.1875, "learning_rate": 9.972640704945992e-06, "loss": 1.2894, "step": 1270 }, { "epoch": 0.13524154075335176, "grad_norm": 16.96875, "learning_rate": 9.972605173393974e-06, "loss": 1.3672, "step": 1271 }, { "epoch": 0.1353479463715684, "grad_norm": 18.3125, "learning_rate": 9.972569641841955e-06, "loss": 1.3795, "step": 1272 }, { "epoch": 0.13545435198978506, "grad_norm": 21.375, "learning_rate": 9.972534110289939e-06, "loss": 1.5497, "step": 1273 }, { "epoch": 0.1355607576080017, "grad_norm": 56.15625, "learning_rate": 9.97249857873792e-06, "loss": 1.546, "step": 1274 }, { "epoch": 0.13566716322621836, "grad_norm": 16.296875, "learning_rate": 9.972463047185902e-06, "loss": 1.2605, "step": 1275 }, { "epoch": 0.13577356884443498, "grad_norm": 22.0625, "learning_rate": 9.972427515633883e-06, "loss": 1.4493, "step": 1276 }, { "epoch": 0.13587997446265163, "grad_norm": 23.109375, "learning_rate": 9.972391984081865e-06, "loss": 1.2439, "step": 1277 }, { "epoch": 0.13598638008086827, "grad_norm": 29.3125, "learning_rate": 9.972356452529848e-06, "loss": 1.5697, "step": 1278 }, { "epoch": 0.13609278569908492, "grad_norm": 39.1875, "learning_rate": 9.97232092097783e-06, "loss": 1.3832, "step": 1279 }, { "epoch": 0.13619919131730154, "grad_norm": 22.359375, "learning_rate": 9.972285389425812e-06, "loss": 1.3998, "step": 1280 }, { "epoch": 0.1363055969355182, "grad_norm": 28.6875, "learning_rate": 9.972249857873793e-06, "loss": 1.5912, "step": 1281 }, { "epoch": 0.13641200255373484, "grad_norm": 17.734375, "learning_rate": 9.972214326321775e-06, "loss": 1.3937, "step": 1282 }, { "epoch": 0.1365184081719515, "grad_norm": 21.65625, "learning_rate": 9.972178794769756e-06, "loss": 1.4752, "step": 1283 }, { "epoch": 0.1366248137901681, "grad_norm": 34.3125, "learning_rate": 9.972143263217738e-06, "loss": 1.387, "step": 1284 }, { "epoch": 0.13673121940838476, "grad_norm": 19.828125, "learning_rate": 9.97210773166572e-06, "loss": 1.2668, "step": 1285 }, { "epoch": 0.1368376250266014, "grad_norm": 19.234375, "learning_rate": 9.972072200113701e-06, "loss": 1.2874, "step": 1286 }, { "epoch": 0.13694403064481805, "grad_norm": 28.265625, "learning_rate": 9.972036668561683e-06, "loss": 1.5476, "step": 1287 }, { "epoch": 0.13705043626303468, "grad_norm": 22.21875, "learning_rate": 9.972001137009666e-06, "loss": 1.6705, "step": 1288 }, { "epoch": 0.13715684188125132, "grad_norm": 27.203125, "learning_rate": 9.971965605457648e-06, "loss": 1.6547, "step": 1289 }, { "epoch": 0.13726324749946797, "grad_norm": 19.828125, "learning_rate": 9.97193007390563e-06, "loss": 1.3509, "step": 1290 }, { "epoch": 0.13736965311768462, "grad_norm": 43.375, "learning_rate": 9.971894542353611e-06, "loss": 1.5157, "step": 1291 }, { "epoch": 0.13747605873590127, "grad_norm": 24.21875, "learning_rate": 9.971859010801593e-06, "loss": 1.4659, "step": 1292 }, { "epoch": 0.1375824643541179, "grad_norm": 22.296875, "learning_rate": 9.971823479249574e-06, "loss": 1.4398, "step": 1293 }, { "epoch": 0.13768886997233454, "grad_norm": 21.71875, "learning_rate": 9.971787947697556e-06, "loss": 1.156, "step": 1294 }, { "epoch": 0.1377952755905512, "grad_norm": 20.484375, "learning_rate": 9.971752416145537e-06, "loss": 1.2789, "step": 1295 }, { "epoch": 0.13790168120876783, "grad_norm": 19.40625, "learning_rate": 9.971716884593519e-06, "loss": 1.3443, "step": 1296 }, { "epoch": 0.13800808682698446, "grad_norm": 39.34375, "learning_rate": 9.9716813530415e-06, "loss": 1.504, "step": 1297 }, { "epoch": 0.1381144924452011, "grad_norm": 21.125, "learning_rate": 9.971645821489484e-06, "loss": 1.4209, "step": 1298 }, { "epoch": 0.13822089806341775, "grad_norm": 22.28125, "learning_rate": 9.971610289937465e-06, "loss": 1.4396, "step": 1299 }, { "epoch": 0.1383273036816344, "grad_norm": 22.28125, "learning_rate": 9.971574758385447e-06, "loss": 1.2452, "step": 1300 }, { "epoch": 0.13843370929985102, "grad_norm": 15.4296875, "learning_rate": 9.971539226833429e-06, "loss": 1.1123, "step": 1301 }, { "epoch": 0.13854011491806767, "grad_norm": 20.09375, "learning_rate": 9.97150369528141e-06, "loss": 1.3767, "step": 1302 }, { "epoch": 0.13864652053628432, "grad_norm": 22.96875, "learning_rate": 9.971468163729392e-06, "loss": 1.1322, "step": 1303 }, { "epoch": 0.13875292615450097, "grad_norm": 17.78125, "learning_rate": 9.971432632177375e-06, "loss": 1.4521, "step": 1304 }, { "epoch": 0.1388593317727176, "grad_norm": 20.25, "learning_rate": 9.971397100625357e-06, "loss": 1.3008, "step": 1305 }, { "epoch": 0.13896573739093424, "grad_norm": 13.9765625, "learning_rate": 9.971361569073338e-06, "loss": 1.1818, "step": 1306 }, { "epoch": 0.13907214300915088, "grad_norm": 15.765625, "learning_rate": 9.97132603752132e-06, "loss": 1.1075, "step": 1307 }, { "epoch": 0.13917854862736753, "grad_norm": 21.484375, "learning_rate": 9.971290505969302e-06, "loss": 1.4527, "step": 1308 }, { "epoch": 0.13928495424558418, "grad_norm": 33.34375, "learning_rate": 9.971254974417283e-06, "loss": 1.4409, "step": 1309 }, { "epoch": 0.1393913598638008, "grad_norm": 17.84375, "learning_rate": 9.971219442865265e-06, "loss": 1.3112, "step": 1310 }, { "epoch": 0.13949776548201745, "grad_norm": 40.53125, "learning_rate": 9.971183911313246e-06, "loss": 1.4311, "step": 1311 }, { "epoch": 0.1396041711002341, "grad_norm": 23.453125, "learning_rate": 9.971148379761228e-06, "loss": 1.1407, "step": 1312 }, { "epoch": 0.13971057671845075, "grad_norm": 22.484375, "learning_rate": 9.97111284820921e-06, "loss": 1.3621, "step": 1313 }, { "epoch": 0.13981698233666737, "grad_norm": 17.546875, "learning_rate": 9.971077316657193e-06, "loss": 1.2856, "step": 1314 }, { "epoch": 0.13992338795488402, "grad_norm": 20.421875, "learning_rate": 9.971041785105175e-06, "loss": 1.1481, "step": 1315 }, { "epoch": 0.14002979357310066, "grad_norm": 29.21875, "learning_rate": 9.971006253553156e-06, "loss": 1.5038, "step": 1316 }, { "epoch": 0.1401361991913173, "grad_norm": 17.46875, "learning_rate": 9.970970722001138e-06, "loss": 1.3676, "step": 1317 }, { "epoch": 0.14024260480953393, "grad_norm": 16.328125, "learning_rate": 9.97093519044912e-06, "loss": 1.3286, "step": 1318 }, { "epoch": 0.14034901042775058, "grad_norm": 20.015625, "learning_rate": 9.970899658897103e-06, "loss": 1.0636, "step": 1319 }, { "epoch": 0.14045541604596723, "grad_norm": 21.390625, "learning_rate": 9.970864127345084e-06, "loss": 1.3971, "step": 1320 }, { "epoch": 0.14056182166418388, "grad_norm": 22.171875, "learning_rate": 9.970828595793066e-06, "loss": 1.1243, "step": 1321 }, { "epoch": 0.1406682272824005, "grad_norm": 29.859375, "learning_rate": 9.970793064241046e-06, "loss": 1.6433, "step": 1322 }, { "epoch": 0.14077463290061715, "grad_norm": 17.671875, "learning_rate": 9.970757532689027e-06, "loss": 1.2115, "step": 1323 }, { "epoch": 0.1408810385188338, "grad_norm": 19.3125, "learning_rate": 9.97072200113701e-06, "loss": 1.4066, "step": 1324 }, { "epoch": 0.14098744413705044, "grad_norm": 18.296875, "learning_rate": 9.970686469584992e-06, "loss": 1.1655, "step": 1325 }, { "epoch": 0.14109384975526706, "grad_norm": 14.0625, "learning_rate": 9.970650938032974e-06, "loss": 1.3365, "step": 1326 }, { "epoch": 0.1412002553734837, "grad_norm": 16.671875, "learning_rate": 9.970615406480956e-06, "loss": 1.4767, "step": 1327 }, { "epoch": 0.14130666099170036, "grad_norm": 16.953125, "learning_rate": 9.970579874928937e-06, "loss": 1.5004, "step": 1328 }, { "epoch": 0.141413066609917, "grad_norm": 17.78125, "learning_rate": 9.97054434337692e-06, "loss": 1.4981, "step": 1329 }, { "epoch": 0.14151947222813366, "grad_norm": 22.90625, "learning_rate": 9.970508811824902e-06, "loss": 1.2329, "step": 1330 }, { "epoch": 0.14162587784635028, "grad_norm": 30.796875, "learning_rate": 9.970473280272884e-06, "loss": 1.5337, "step": 1331 }, { "epoch": 0.14173228346456693, "grad_norm": 14.1875, "learning_rate": 9.970437748720865e-06, "loss": 1.2126, "step": 1332 }, { "epoch": 0.14183868908278358, "grad_norm": 18.28125, "learning_rate": 9.970402217168847e-06, "loss": 1.6777, "step": 1333 }, { "epoch": 0.14194509470100022, "grad_norm": 20.0, "learning_rate": 9.970366685616828e-06, "loss": 1.5186, "step": 1334 }, { "epoch": 0.14205150031921684, "grad_norm": 14.296875, "learning_rate": 9.97033115406481e-06, "loss": 1.143, "step": 1335 }, { "epoch": 0.1421579059374335, "grad_norm": 22.046875, "learning_rate": 9.970295622512792e-06, "loss": 1.6629, "step": 1336 }, { "epoch": 0.14226431155565014, "grad_norm": 15.671875, "learning_rate": 9.970260090960773e-06, "loss": 1.1653, "step": 1337 }, { "epoch": 0.1423707171738668, "grad_norm": 25.203125, "learning_rate": 9.970224559408755e-06, "loss": 1.6829, "step": 1338 }, { "epoch": 0.1424771227920834, "grad_norm": 18.890625, "learning_rate": 9.970189027856738e-06, "loss": 1.247, "step": 1339 }, { "epoch": 0.14258352841030006, "grad_norm": 16.171875, "learning_rate": 9.97015349630472e-06, "loss": 1.1887, "step": 1340 }, { "epoch": 0.1426899340285167, "grad_norm": 24.984375, "learning_rate": 9.970117964752701e-06, "loss": 1.5807, "step": 1341 }, { "epoch": 0.14279633964673336, "grad_norm": 17.109375, "learning_rate": 9.970082433200683e-06, "loss": 1.0129, "step": 1342 }, { "epoch": 0.14290274526494998, "grad_norm": 19.9375, "learning_rate": 9.970046901648665e-06, "loss": 1.4591, "step": 1343 }, { "epoch": 0.14300915088316662, "grad_norm": 19.109375, "learning_rate": 9.970011370096646e-06, "loss": 1.204, "step": 1344 }, { "epoch": 0.14311555650138327, "grad_norm": 19.859375, "learning_rate": 9.96997583854463e-06, "loss": 1.1448, "step": 1345 }, { "epoch": 0.14322196211959992, "grad_norm": 70.0, "learning_rate": 9.969940306992611e-06, "loss": 1.2771, "step": 1346 }, { "epoch": 0.14332836773781657, "grad_norm": 25.296875, "learning_rate": 9.969904775440593e-06, "loss": 1.4786, "step": 1347 }, { "epoch": 0.1434347733560332, "grad_norm": 20.1875, "learning_rate": 9.969869243888574e-06, "loss": 1.5487, "step": 1348 }, { "epoch": 0.14354117897424984, "grad_norm": 21.671875, "learning_rate": 9.969833712336556e-06, "loss": 1.2271, "step": 1349 }, { "epoch": 0.1436475845924665, "grad_norm": 18.1875, "learning_rate": 9.969798180784538e-06, "loss": 1.484, "step": 1350 }, { "epoch": 0.14375399021068314, "grad_norm": 13.375, "learning_rate": 9.96976264923252e-06, "loss": 1.353, "step": 1351 }, { "epoch": 0.14386039582889976, "grad_norm": 15.1875, "learning_rate": 9.9697271176805e-06, "loss": 1.2982, "step": 1352 }, { "epoch": 0.1439668014471164, "grad_norm": 16.765625, "learning_rate": 9.969691586128482e-06, "loss": 1.4266, "step": 1353 }, { "epoch": 0.14407320706533305, "grad_norm": 18.546875, "learning_rate": 9.969656054576464e-06, "loss": 1.501, "step": 1354 }, { "epoch": 0.1441796126835497, "grad_norm": 24.546875, "learning_rate": 9.969620523024447e-06, "loss": 1.2657, "step": 1355 }, { "epoch": 0.14428601830176632, "grad_norm": 16.046875, "learning_rate": 9.969584991472429e-06, "loss": 1.2308, "step": 1356 }, { "epoch": 0.14439242391998297, "grad_norm": 22.96875, "learning_rate": 9.96954945992041e-06, "loss": 1.3533, "step": 1357 }, { "epoch": 0.14449882953819962, "grad_norm": 15.21875, "learning_rate": 9.969513928368392e-06, "loss": 1.2143, "step": 1358 }, { "epoch": 0.14460523515641627, "grad_norm": 16.171875, "learning_rate": 9.969478396816374e-06, "loss": 0.9716, "step": 1359 }, { "epoch": 0.1447116407746329, "grad_norm": 23.046875, "learning_rate": 9.969442865264355e-06, "loss": 1.2829, "step": 1360 }, { "epoch": 0.14481804639284954, "grad_norm": 22.09375, "learning_rate": 9.969407333712337e-06, "loss": 1.6855, "step": 1361 }, { "epoch": 0.14492445201106618, "grad_norm": 34.125, "learning_rate": 9.969371802160319e-06, "loss": 1.4646, "step": 1362 }, { "epoch": 0.14503085762928283, "grad_norm": 16.71875, "learning_rate": 9.9693362706083e-06, "loss": 1.3911, "step": 1363 }, { "epoch": 0.14513726324749948, "grad_norm": 24.765625, "learning_rate": 9.969300739056282e-06, "loss": 1.6042, "step": 1364 }, { "epoch": 0.1452436688657161, "grad_norm": 19.0, "learning_rate": 9.969265207504265e-06, "loss": 1.2692, "step": 1365 }, { "epoch": 0.14535007448393275, "grad_norm": 15.890625, "learning_rate": 9.969229675952247e-06, "loss": 1.0551, "step": 1366 }, { "epoch": 0.1454564801021494, "grad_norm": 23.671875, "learning_rate": 9.969194144400228e-06, "loss": 1.4418, "step": 1367 }, { "epoch": 0.14556288572036605, "grad_norm": 21.703125, "learning_rate": 9.96915861284821e-06, "loss": 1.2684, "step": 1368 }, { "epoch": 0.14566929133858267, "grad_norm": 17.6875, "learning_rate": 9.969123081296191e-06, "loss": 1.4692, "step": 1369 }, { "epoch": 0.14577569695679932, "grad_norm": 18.921875, "learning_rate": 9.969087549744175e-06, "loss": 1.1304, "step": 1370 }, { "epoch": 0.14588210257501597, "grad_norm": 23.421875, "learning_rate": 9.969052018192156e-06, "loss": 1.3826, "step": 1371 }, { "epoch": 0.1459885081932326, "grad_norm": 23.8125, "learning_rate": 9.969016486640138e-06, "loss": 1.7793, "step": 1372 }, { "epoch": 0.14609491381144923, "grad_norm": 24.953125, "learning_rate": 9.96898095508812e-06, "loss": 1.596, "step": 1373 }, { "epoch": 0.14620131942966588, "grad_norm": 25.28125, "learning_rate": 9.968945423536101e-06, "loss": 1.4439, "step": 1374 }, { "epoch": 0.14630772504788253, "grad_norm": 17.921875, "learning_rate": 9.968909891984083e-06, "loss": 1.2339, "step": 1375 }, { "epoch": 0.14641413066609918, "grad_norm": 27.390625, "learning_rate": 9.968874360432064e-06, "loss": 1.2887, "step": 1376 }, { "epoch": 0.1465205362843158, "grad_norm": 26.796875, "learning_rate": 9.968838828880046e-06, "loss": 1.5512, "step": 1377 }, { "epoch": 0.14662694190253245, "grad_norm": 15.046875, "learning_rate": 9.968803297328028e-06, "loss": 1.2474, "step": 1378 }, { "epoch": 0.1467333475207491, "grad_norm": 22.71875, "learning_rate": 9.96876776577601e-06, "loss": 1.4287, "step": 1379 }, { "epoch": 0.14683975313896575, "grad_norm": 380.75, "learning_rate": 9.968732234223991e-06, "loss": 1.4128, "step": 1380 }, { "epoch": 0.14694615875718237, "grad_norm": 25.75, "learning_rate": 9.968696702671974e-06, "loss": 1.5916, "step": 1381 }, { "epoch": 0.14705256437539901, "grad_norm": 20.875, "learning_rate": 9.968661171119956e-06, "loss": 1.1402, "step": 1382 }, { "epoch": 0.14715896999361566, "grad_norm": 27.21875, "learning_rate": 9.968625639567937e-06, "loss": 1.061, "step": 1383 }, { "epoch": 0.1472653756118323, "grad_norm": 19.25, "learning_rate": 9.968590108015919e-06, "loss": 1.209, "step": 1384 }, { "epoch": 0.14737178123004896, "grad_norm": 20.78125, "learning_rate": 9.9685545764639e-06, "loss": 1.3898, "step": 1385 }, { "epoch": 0.14747818684826558, "grad_norm": 13.5390625, "learning_rate": 9.968519044911882e-06, "loss": 1.2768, "step": 1386 }, { "epoch": 0.14758459246648223, "grad_norm": 21.9375, "learning_rate": 9.968483513359864e-06, "loss": 1.408, "step": 1387 }, { "epoch": 0.14769099808469888, "grad_norm": 16.515625, "learning_rate": 9.968447981807845e-06, "loss": 1.3886, "step": 1388 }, { "epoch": 0.14779740370291553, "grad_norm": 21.625, "learning_rate": 9.968412450255827e-06, "loss": 1.2288, "step": 1389 }, { "epoch": 0.14790380932113215, "grad_norm": 16.609375, "learning_rate": 9.968376918703809e-06, "loss": 1.2017, "step": 1390 }, { "epoch": 0.1480102149393488, "grad_norm": 17.75, "learning_rate": 9.968341387151792e-06, "loss": 1.4288, "step": 1391 }, { "epoch": 0.14811662055756544, "grad_norm": 15.21875, "learning_rate": 9.968305855599774e-06, "loss": 1.0844, "step": 1392 }, { "epoch": 0.1482230261757821, "grad_norm": 21.390625, "learning_rate": 9.968270324047755e-06, "loss": 1.5393, "step": 1393 }, { "epoch": 0.1483294317939987, "grad_norm": 19.640625, "learning_rate": 9.968234792495737e-06, "loss": 1.4381, "step": 1394 }, { "epoch": 0.14843583741221536, "grad_norm": 19.4375, "learning_rate": 9.968199260943718e-06, "loss": 1.3786, "step": 1395 }, { "epoch": 0.148542243030432, "grad_norm": 25.65625, "learning_rate": 9.968163729391702e-06, "loss": 1.5031, "step": 1396 }, { "epoch": 0.14864864864864866, "grad_norm": 20.40625, "learning_rate": 9.968128197839683e-06, "loss": 1.4112, "step": 1397 }, { "epoch": 0.14875505426686528, "grad_norm": 20.515625, "learning_rate": 9.968092666287665e-06, "loss": 1.1434, "step": 1398 }, { "epoch": 0.14886145988508193, "grad_norm": 18.421875, "learning_rate": 9.968057134735646e-06, "loss": 1.1216, "step": 1399 }, { "epoch": 0.14896786550329857, "grad_norm": 14.890625, "learning_rate": 9.968021603183628e-06, "loss": 1.1722, "step": 1400 }, { "epoch": 0.14907427112151522, "grad_norm": 15.8046875, "learning_rate": 9.96798607163161e-06, "loss": 1.1502, "step": 1401 }, { "epoch": 0.14918067673973187, "grad_norm": 15.6015625, "learning_rate": 9.967950540079591e-06, "loss": 1.2218, "step": 1402 }, { "epoch": 0.1492870823579485, "grad_norm": 26.8125, "learning_rate": 9.967915008527573e-06, "loss": 1.651, "step": 1403 }, { "epoch": 0.14939348797616514, "grad_norm": 18.4375, "learning_rate": 9.967879476975555e-06, "loss": 1.2126, "step": 1404 }, { "epoch": 0.1494998935943818, "grad_norm": 22.703125, "learning_rate": 9.967843945423536e-06, "loss": 1.6361, "step": 1405 }, { "epoch": 0.14960629921259844, "grad_norm": 25.734375, "learning_rate": 9.96780841387152e-06, "loss": 1.1788, "step": 1406 }, { "epoch": 0.14971270483081506, "grad_norm": 23.90625, "learning_rate": 9.967772882319501e-06, "loss": 1.6713, "step": 1407 }, { "epoch": 0.1498191104490317, "grad_norm": 17.03125, "learning_rate": 9.967737350767483e-06, "loss": 1.3707, "step": 1408 }, { "epoch": 0.14992551606724835, "grad_norm": 19.640625, "learning_rate": 9.967701819215464e-06, "loss": 1.5431, "step": 1409 }, { "epoch": 0.150031921685465, "grad_norm": 14.125, "learning_rate": 9.967666287663446e-06, "loss": 1.263, "step": 1410 }, { "epoch": 0.15013832730368162, "grad_norm": 20.625, "learning_rate": 9.967630756111427e-06, "loss": 1.5177, "step": 1411 }, { "epoch": 0.15024473292189827, "grad_norm": 20.90625, "learning_rate": 9.967595224559409e-06, "loss": 1.706, "step": 1412 }, { "epoch": 0.15035113854011492, "grad_norm": 26.03125, "learning_rate": 9.96755969300739e-06, "loss": 1.5032, "step": 1413 }, { "epoch": 0.15045754415833157, "grad_norm": 28.703125, "learning_rate": 9.967524161455372e-06, "loss": 1.2238, "step": 1414 }, { "epoch": 0.1505639497765482, "grad_norm": 19.984375, "learning_rate": 9.967488629903354e-06, "loss": 1.6737, "step": 1415 }, { "epoch": 0.15067035539476484, "grad_norm": 23.03125, "learning_rate": 9.967453098351337e-06, "loss": 1.4564, "step": 1416 }, { "epoch": 0.1507767610129815, "grad_norm": 20.46875, "learning_rate": 9.967417566799319e-06, "loss": 1.2794, "step": 1417 }, { "epoch": 0.15088316663119813, "grad_norm": 17.5625, "learning_rate": 9.9673820352473e-06, "loss": 1.3907, "step": 1418 }, { "epoch": 0.15098957224941478, "grad_norm": 16.90625, "learning_rate": 9.967346503695282e-06, "loss": 1.4408, "step": 1419 }, { "epoch": 0.1510959778676314, "grad_norm": 22.625, "learning_rate": 9.967310972143264e-06, "loss": 1.7032, "step": 1420 }, { "epoch": 0.15120238348584805, "grad_norm": 24.59375, "learning_rate": 9.967275440591245e-06, "loss": 0.759, "step": 1421 }, { "epoch": 0.1513087891040647, "grad_norm": 16.640625, "learning_rate": 9.967239909039229e-06, "loss": 1.344, "step": 1422 }, { "epoch": 0.15141519472228135, "grad_norm": 14.8984375, "learning_rate": 9.96720437748721e-06, "loss": 1.1238, "step": 1423 }, { "epoch": 0.15152160034049797, "grad_norm": 25.140625, "learning_rate": 9.967168845935192e-06, "loss": 1.6726, "step": 1424 }, { "epoch": 0.15162800595871462, "grad_norm": 21.375, "learning_rate": 9.967133314383173e-06, "loss": 1.6707, "step": 1425 }, { "epoch": 0.15173441157693127, "grad_norm": 22.875, "learning_rate": 9.967097782831155e-06, "loss": 1.6231, "step": 1426 }, { "epoch": 0.15184081719514791, "grad_norm": 18.5625, "learning_rate": 9.967062251279137e-06, "loss": 1.674, "step": 1427 }, { "epoch": 0.15194722281336454, "grad_norm": 17.171875, "learning_rate": 9.967026719727118e-06, "loss": 1.4384, "step": 1428 }, { "epoch": 0.15205362843158118, "grad_norm": 17.359375, "learning_rate": 9.9669911881751e-06, "loss": 1.4147, "step": 1429 }, { "epoch": 0.15216003404979783, "grad_norm": 23.515625, "learning_rate": 9.966955656623081e-06, "loss": 1.4953, "step": 1430 }, { "epoch": 0.15226643966801448, "grad_norm": 15.4609375, "learning_rate": 9.966920125071063e-06, "loss": 1.1958, "step": 1431 }, { "epoch": 0.1523728452862311, "grad_norm": 25.171875, "learning_rate": 9.966884593519046e-06, "loss": 1.6722, "step": 1432 }, { "epoch": 0.15247925090444775, "grad_norm": 14.8125, "learning_rate": 9.966849061967028e-06, "loss": 1.3639, "step": 1433 }, { "epoch": 0.1525856565226644, "grad_norm": 23.984375, "learning_rate": 9.96681353041501e-06, "loss": 1.4487, "step": 1434 }, { "epoch": 0.15269206214088105, "grad_norm": 14.7109375, "learning_rate": 9.966777998862991e-06, "loss": 1.3199, "step": 1435 }, { "epoch": 0.15279846775909767, "grad_norm": 12.546875, "learning_rate": 9.966742467310973e-06, "loss": 1.3453, "step": 1436 }, { "epoch": 0.15290487337731432, "grad_norm": 20.09375, "learning_rate": 9.966706935758956e-06, "loss": 1.6151, "step": 1437 }, { "epoch": 0.15301127899553096, "grad_norm": 18.03125, "learning_rate": 9.966671404206938e-06, "loss": 1.2411, "step": 1438 }, { "epoch": 0.1531176846137476, "grad_norm": 27.3125, "learning_rate": 9.96663587265492e-06, "loss": 1.5216, "step": 1439 }, { "epoch": 0.15322409023196426, "grad_norm": 18.015625, "learning_rate": 9.9666003411029e-06, "loss": 1.196, "step": 1440 }, { "epoch": 0.15333049585018088, "grad_norm": 18.546875, "learning_rate": 9.96656480955088e-06, "loss": 1.2908, "step": 1441 }, { "epoch": 0.15343690146839753, "grad_norm": 23.109375, "learning_rate": 9.966529277998864e-06, "loss": 1.3013, "step": 1442 }, { "epoch": 0.15354330708661418, "grad_norm": 27.296875, "learning_rate": 9.966493746446846e-06, "loss": 1.9279, "step": 1443 }, { "epoch": 0.15364971270483083, "grad_norm": 15.5859375, "learning_rate": 9.966458214894827e-06, "loss": 1.3002, "step": 1444 }, { "epoch": 0.15375611832304745, "grad_norm": 28.125, "learning_rate": 9.966422683342809e-06, "loss": 1.3685, "step": 1445 }, { "epoch": 0.1538625239412641, "grad_norm": 32.84375, "learning_rate": 9.96638715179079e-06, "loss": 1.2581, "step": 1446 }, { "epoch": 0.15396892955948074, "grad_norm": 16.0625, "learning_rate": 9.966351620238774e-06, "loss": 1.0525, "step": 1447 }, { "epoch": 0.1540753351776974, "grad_norm": 22.765625, "learning_rate": 9.966316088686755e-06, "loss": 1.4141, "step": 1448 }, { "epoch": 0.154181740795914, "grad_norm": 30.890625, "learning_rate": 9.966280557134737e-06, "loss": 1.6109, "step": 1449 }, { "epoch": 0.15428814641413066, "grad_norm": 13.8046875, "learning_rate": 9.966245025582719e-06, "loss": 1.1998, "step": 1450 }, { "epoch": 0.1543945520323473, "grad_norm": 16.9375, "learning_rate": 9.9662094940307e-06, "loss": 1.2831, "step": 1451 }, { "epoch": 0.15450095765056396, "grad_norm": 22.890625, "learning_rate": 9.966173962478682e-06, "loss": 1.8341, "step": 1452 }, { "epoch": 0.15460736326878058, "grad_norm": 15.296875, "learning_rate": 9.966138430926663e-06, "loss": 1.1853, "step": 1453 }, { "epoch": 0.15471376888699723, "grad_norm": 15.6015625, "learning_rate": 9.966102899374645e-06, "loss": 1.2653, "step": 1454 }, { "epoch": 0.15482017450521388, "grad_norm": 21.828125, "learning_rate": 9.966067367822627e-06, "loss": 1.6148, "step": 1455 }, { "epoch": 0.15492658012343052, "grad_norm": 17.890625, "learning_rate": 9.966031836270608e-06, "loss": 0.9279, "step": 1456 }, { "epoch": 0.15503298574164717, "grad_norm": 15.03125, "learning_rate": 9.965996304718592e-06, "loss": 1.1638, "step": 1457 }, { "epoch": 0.1551393913598638, "grad_norm": 19.109375, "learning_rate": 9.965960773166573e-06, "loss": 1.375, "step": 1458 }, { "epoch": 0.15524579697808044, "grad_norm": 18.796875, "learning_rate": 9.965925241614555e-06, "loss": 1.4081, "step": 1459 }, { "epoch": 0.1553522025962971, "grad_norm": 18.140625, "learning_rate": 9.965889710062536e-06, "loss": 1.4992, "step": 1460 }, { "epoch": 0.15545860821451374, "grad_norm": 44.21875, "learning_rate": 9.965854178510518e-06, "loss": 1.4936, "step": 1461 }, { "epoch": 0.15556501383273036, "grad_norm": 20.78125, "learning_rate": 9.9658186469585e-06, "loss": 1.1436, "step": 1462 }, { "epoch": 0.155671419450947, "grad_norm": 19.484375, "learning_rate": 9.965783115406483e-06, "loss": 1.5755, "step": 1463 }, { "epoch": 0.15577782506916366, "grad_norm": 19.609375, "learning_rate": 9.965747583854464e-06, "loss": 1.6116, "step": 1464 }, { "epoch": 0.1558842306873803, "grad_norm": 17.8125, "learning_rate": 9.965712052302446e-06, "loss": 1.2383, "step": 1465 }, { "epoch": 0.15599063630559692, "grad_norm": 25.9375, "learning_rate": 9.965676520750428e-06, "loss": 1.3652, "step": 1466 }, { "epoch": 0.15609704192381357, "grad_norm": 18.25, "learning_rate": 9.96564098919841e-06, "loss": 1.3939, "step": 1467 }, { "epoch": 0.15620344754203022, "grad_norm": 15.390625, "learning_rate": 9.965605457646391e-06, "loss": 1.298, "step": 1468 }, { "epoch": 0.15630985316024687, "grad_norm": 19.90625, "learning_rate": 9.965569926094372e-06, "loss": 1.3513, "step": 1469 }, { "epoch": 0.1564162587784635, "grad_norm": 28.359375, "learning_rate": 9.965534394542354e-06, "loss": 1.6195, "step": 1470 }, { "epoch": 0.15652266439668014, "grad_norm": 15.4296875, "learning_rate": 9.965498862990336e-06, "loss": 1.319, "step": 1471 }, { "epoch": 0.1566290700148968, "grad_norm": 17.5625, "learning_rate": 9.965463331438317e-06, "loss": 1.3833, "step": 1472 }, { "epoch": 0.15673547563311344, "grad_norm": 16.3125, "learning_rate": 9.9654277998863e-06, "loss": 1.3241, "step": 1473 }, { "epoch": 0.15684188125133006, "grad_norm": 18.046875, "learning_rate": 9.965392268334282e-06, "loss": 1.1163, "step": 1474 }, { "epoch": 0.1569482868695467, "grad_norm": 13.9375, "learning_rate": 9.965356736782264e-06, "loss": 1.355, "step": 1475 }, { "epoch": 0.15705469248776335, "grad_norm": 24.90625, "learning_rate": 9.965321205230245e-06, "loss": 1.6023, "step": 1476 }, { "epoch": 0.15716109810598, "grad_norm": 22.109375, "learning_rate": 9.965285673678227e-06, "loss": 1.203, "step": 1477 }, { "epoch": 0.15726750372419665, "grad_norm": 16.796875, "learning_rate": 9.965250142126209e-06, "loss": 1.3417, "step": 1478 }, { "epoch": 0.15737390934241327, "grad_norm": 38.53125, "learning_rate": 9.96521461057419e-06, "loss": 0.7325, "step": 1479 }, { "epoch": 0.15748031496062992, "grad_norm": 21.859375, "learning_rate": 9.965179079022172e-06, "loss": 1.3909, "step": 1480 }, { "epoch": 0.15758672057884657, "grad_norm": 21.5, "learning_rate": 9.965143547470153e-06, "loss": 1.3502, "step": 1481 }, { "epoch": 0.15769312619706322, "grad_norm": 23.15625, "learning_rate": 9.965108015918135e-06, "loss": 1.4367, "step": 1482 }, { "epoch": 0.15779953181527984, "grad_norm": 14.59375, "learning_rate": 9.965072484366118e-06, "loss": 1.223, "step": 1483 }, { "epoch": 0.15790593743349648, "grad_norm": 12.4765625, "learning_rate": 9.9650369528141e-06, "loss": 1.1071, "step": 1484 }, { "epoch": 0.15801234305171313, "grad_norm": 17.03125, "learning_rate": 9.965001421262082e-06, "loss": 1.0484, "step": 1485 }, { "epoch": 0.15811874866992978, "grad_norm": 31.4375, "learning_rate": 9.964965889710063e-06, "loss": 1.667, "step": 1486 }, { "epoch": 0.1582251542881464, "grad_norm": 23.15625, "learning_rate": 9.964930358158045e-06, "loss": 1.3088, "step": 1487 }, { "epoch": 0.15833155990636305, "grad_norm": 89.75, "learning_rate": 9.964894826606028e-06, "loss": 2.1094, "step": 1488 }, { "epoch": 0.1584379655245797, "grad_norm": 16.09375, "learning_rate": 9.96485929505401e-06, "loss": 1.4104, "step": 1489 }, { "epoch": 0.15854437114279635, "grad_norm": 19.421875, "learning_rate": 9.964823763501991e-06, "loss": 1.4055, "step": 1490 }, { "epoch": 0.15865077676101297, "grad_norm": 15.078125, "learning_rate": 9.964788231949973e-06, "loss": 1.2012, "step": 1491 }, { "epoch": 0.15875718237922962, "grad_norm": 17.125, "learning_rate": 9.964752700397955e-06, "loss": 1.1667, "step": 1492 }, { "epoch": 0.15886358799744627, "grad_norm": 47.4375, "learning_rate": 9.964717168845936e-06, "loss": 0.978, "step": 1493 }, { "epoch": 0.1589699936156629, "grad_norm": 22.53125, "learning_rate": 9.964681637293918e-06, "loss": 1.3719, "step": 1494 }, { "epoch": 0.15907639923387956, "grad_norm": 66.375, "learning_rate": 9.9646461057419e-06, "loss": 1.6371, "step": 1495 }, { "epoch": 0.15918280485209618, "grad_norm": 22.78125, "learning_rate": 9.964610574189881e-06, "loss": 1.7911, "step": 1496 }, { "epoch": 0.15928921047031283, "grad_norm": 17.21875, "learning_rate": 9.964575042637863e-06, "loss": 1.2365, "step": 1497 }, { "epoch": 0.15939561608852948, "grad_norm": 43.375, "learning_rate": 9.964539511085844e-06, "loss": 1.3822, "step": 1498 }, { "epoch": 0.15950202170674613, "grad_norm": 16.359375, "learning_rate": 9.964503979533827e-06, "loss": 1.2554, "step": 1499 }, { "epoch": 0.15960842732496275, "grad_norm": 18.265625, "learning_rate": 9.964468447981809e-06, "loss": 1.3469, "step": 1500 }, { "epoch": 0.1597148329431794, "grad_norm": 15.8125, "learning_rate": 9.96443291642979e-06, "loss": 1.3571, "step": 1501 }, { "epoch": 0.15982123856139605, "grad_norm": 13.921875, "learning_rate": 9.964397384877772e-06, "loss": 1.3128, "step": 1502 }, { "epoch": 0.1599276441796127, "grad_norm": 19.59375, "learning_rate": 9.964361853325754e-06, "loss": 1.3799, "step": 1503 }, { "epoch": 0.16003404979782931, "grad_norm": 13.125, "learning_rate": 9.964326321773735e-06, "loss": 1.1827, "step": 1504 }, { "epoch": 0.16014045541604596, "grad_norm": 17.59375, "learning_rate": 9.964290790221717e-06, "loss": 1.3384, "step": 1505 }, { "epoch": 0.1602468610342626, "grad_norm": 19.390625, "learning_rate": 9.964255258669699e-06, "loss": 1.1688, "step": 1506 }, { "epoch": 0.16035326665247926, "grad_norm": 16.484375, "learning_rate": 9.96421972711768e-06, "loss": 1.4978, "step": 1507 }, { "epoch": 0.16045967227069588, "grad_norm": 23.53125, "learning_rate": 9.964184195565662e-06, "loss": 1.3807, "step": 1508 }, { "epoch": 0.16056607788891253, "grad_norm": 19.78125, "learning_rate": 9.964148664013645e-06, "loss": 1.6842, "step": 1509 }, { "epoch": 0.16067248350712918, "grad_norm": 19.4375, "learning_rate": 9.964113132461627e-06, "loss": 1.5018, "step": 1510 }, { "epoch": 0.16077888912534583, "grad_norm": 22.984375, "learning_rate": 9.964077600909608e-06, "loss": 1.1138, "step": 1511 }, { "epoch": 0.16088529474356247, "grad_norm": 16.609375, "learning_rate": 9.96404206935759e-06, "loss": 1.6935, "step": 1512 }, { "epoch": 0.1609917003617791, "grad_norm": 34.96875, "learning_rate": 9.964006537805572e-06, "loss": 1.5193, "step": 1513 }, { "epoch": 0.16109810597999574, "grad_norm": 17.078125, "learning_rate": 9.963971006253555e-06, "loss": 1.2581, "step": 1514 }, { "epoch": 0.1612045115982124, "grad_norm": 29.96875, "learning_rate": 9.963935474701537e-06, "loss": 1.2659, "step": 1515 }, { "epoch": 0.16131091721642904, "grad_norm": 19.046875, "learning_rate": 9.963899943149518e-06, "loss": 1.4164, "step": 1516 }, { "epoch": 0.16141732283464566, "grad_norm": 19.265625, "learning_rate": 9.9638644115975e-06, "loss": 1.4005, "step": 1517 }, { "epoch": 0.1615237284528623, "grad_norm": 14.234375, "learning_rate": 9.963828880045481e-06, "loss": 1.2538, "step": 1518 }, { "epoch": 0.16163013407107896, "grad_norm": 20.828125, "learning_rate": 9.963793348493463e-06, "loss": 1.3709, "step": 1519 }, { "epoch": 0.1617365396892956, "grad_norm": 14.5703125, "learning_rate": 9.963757816941445e-06, "loss": 1.4998, "step": 1520 }, { "epoch": 0.16184294530751223, "grad_norm": 17.84375, "learning_rate": 9.963722285389426e-06, "loss": 1.2237, "step": 1521 }, { "epoch": 0.16194935092572887, "grad_norm": 16.515625, "learning_rate": 9.963686753837408e-06, "loss": 1.3642, "step": 1522 }, { "epoch": 0.16205575654394552, "grad_norm": 24.953125, "learning_rate": 9.96365122228539e-06, "loss": 1.2246, "step": 1523 }, { "epoch": 0.16216216216216217, "grad_norm": 15.8125, "learning_rate": 9.963615690733373e-06, "loss": 1.2626, "step": 1524 }, { "epoch": 0.1622685677803788, "grad_norm": 16.296875, "learning_rate": 9.963580159181354e-06, "loss": 1.3466, "step": 1525 }, { "epoch": 0.16237497339859544, "grad_norm": 19.640625, "learning_rate": 9.963544627629336e-06, "loss": 1.4675, "step": 1526 }, { "epoch": 0.1624813790168121, "grad_norm": 26.25, "learning_rate": 9.963509096077318e-06, "loss": 1.3918, "step": 1527 }, { "epoch": 0.16258778463502874, "grad_norm": 28.203125, "learning_rate": 9.963473564525299e-06, "loss": 1.7267, "step": 1528 }, { "epoch": 0.16269419025324536, "grad_norm": 15.2265625, "learning_rate": 9.963438032973282e-06, "loss": 1.361, "step": 1529 }, { "epoch": 0.162800595871462, "grad_norm": 13.1953125, "learning_rate": 9.963402501421262e-06, "loss": 1.0538, "step": 1530 }, { "epoch": 0.16290700148967865, "grad_norm": 16.484375, "learning_rate": 9.963366969869244e-06, "loss": 1.5445, "step": 1531 }, { "epoch": 0.1630134071078953, "grad_norm": 24.390625, "learning_rate": 9.963331438317226e-06, "loss": 1.6109, "step": 1532 }, { "epoch": 0.16311981272611195, "grad_norm": 14.671875, "learning_rate": 9.963295906765207e-06, "loss": 1.0957, "step": 1533 }, { "epoch": 0.16322621834432857, "grad_norm": 19.984375, "learning_rate": 9.96326037521319e-06, "loss": 1.2478, "step": 1534 }, { "epoch": 0.16333262396254522, "grad_norm": 17.96875, "learning_rate": 9.963224843661172e-06, "loss": 1.4276, "step": 1535 }, { "epoch": 0.16343902958076187, "grad_norm": 19.0, "learning_rate": 9.963189312109154e-06, "loss": 1.1957, "step": 1536 }, { "epoch": 0.16354543519897852, "grad_norm": 21.625, "learning_rate": 9.963153780557135e-06, "loss": 1.3621, "step": 1537 }, { "epoch": 0.16365184081719514, "grad_norm": 17.703125, "learning_rate": 9.963118249005117e-06, "loss": 1.4176, "step": 1538 }, { "epoch": 0.1637582464354118, "grad_norm": 19.578125, "learning_rate": 9.963082717453099e-06, "loss": 1.2269, "step": 1539 }, { "epoch": 0.16386465205362843, "grad_norm": 19.765625, "learning_rate": 9.963047185901082e-06, "loss": 1.5588, "step": 1540 }, { "epoch": 0.16397105767184508, "grad_norm": 16.5, "learning_rate": 9.963011654349063e-06, "loss": 1.0419, "step": 1541 }, { "epoch": 0.1640774632900617, "grad_norm": 19.0625, "learning_rate": 9.962976122797045e-06, "loss": 1.2066, "step": 1542 }, { "epoch": 0.16418386890827835, "grad_norm": 19.734375, "learning_rate": 9.962940591245027e-06, "loss": 1.2516, "step": 1543 }, { "epoch": 0.164290274526495, "grad_norm": 29.25, "learning_rate": 9.962905059693008e-06, "loss": 1.3059, "step": 1544 }, { "epoch": 0.16439668014471165, "grad_norm": 32.25, "learning_rate": 9.96286952814099e-06, "loss": 1.6897, "step": 1545 }, { "epoch": 0.16450308576292827, "grad_norm": 16.40625, "learning_rate": 9.962833996588971e-06, "loss": 1.1207, "step": 1546 }, { "epoch": 0.16460949138114492, "grad_norm": 15.390625, "learning_rate": 9.962798465036953e-06, "loss": 1.3023, "step": 1547 }, { "epoch": 0.16471589699936157, "grad_norm": 21.734375, "learning_rate": 9.962762933484935e-06, "loss": 1.7294, "step": 1548 }, { "epoch": 0.16482230261757821, "grad_norm": 24.328125, "learning_rate": 9.962727401932916e-06, "loss": 1.3914, "step": 1549 }, { "epoch": 0.16492870823579486, "grad_norm": 18.734375, "learning_rate": 9.9626918703809e-06, "loss": 1.5169, "step": 1550 }, { "epoch": 0.16503511385401148, "grad_norm": 16.578125, "learning_rate": 9.962656338828881e-06, "loss": 1.3363, "step": 1551 }, { "epoch": 0.16514151947222813, "grad_norm": 18.0, "learning_rate": 9.962620807276863e-06, "loss": 1.2939, "step": 1552 }, { "epoch": 0.16524792509044478, "grad_norm": 13.90625, "learning_rate": 9.962585275724844e-06, "loss": 1.2157, "step": 1553 }, { "epoch": 0.16535433070866143, "grad_norm": 14.3515625, "learning_rate": 9.962549744172826e-06, "loss": 1.2191, "step": 1554 }, { "epoch": 0.16546073632687805, "grad_norm": 20.953125, "learning_rate": 9.96251421262081e-06, "loss": 1.4439, "step": 1555 }, { "epoch": 0.1655671419450947, "grad_norm": 17.4375, "learning_rate": 9.962478681068791e-06, "loss": 1.028, "step": 1556 }, { "epoch": 0.16567354756331135, "grad_norm": 18.203125, "learning_rate": 9.962443149516773e-06, "loss": 1.0448, "step": 1557 }, { "epoch": 0.165779953181528, "grad_norm": 18.546875, "learning_rate": 9.962407617964754e-06, "loss": 1.4718, "step": 1558 }, { "epoch": 0.16588635879974462, "grad_norm": 21.125, "learning_rate": 9.962372086412736e-06, "loss": 1.4379, "step": 1559 }, { "epoch": 0.16599276441796126, "grad_norm": 18.421875, "learning_rate": 9.962336554860717e-06, "loss": 1.4096, "step": 1560 }, { "epoch": 0.1660991700361779, "grad_norm": 22.46875, "learning_rate": 9.962301023308699e-06, "loss": 1.5199, "step": 1561 }, { "epoch": 0.16620557565439456, "grad_norm": 17.125, "learning_rate": 9.96226549175668e-06, "loss": 1.1784, "step": 1562 }, { "epoch": 0.16631198127261118, "grad_norm": 17.5, "learning_rate": 9.962229960204662e-06, "loss": 1.1486, "step": 1563 }, { "epoch": 0.16641838689082783, "grad_norm": 13.34375, "learning_rate": 9.962194428652644e-06, "loss": 1.1499, "step": 1564 }, { "epoch": 0.16652479250904448, "grad_norm": 21.828125, "learning_rate": 9.962158897100627e-06, "loss": 1.553, "step": 1565 }, { "epoch": 0.16663119812726113, "grad_norm": 17.890625, "learning_rate": 9.962123365548609e-06, "loss": 1.07, "step": 1566 }, { "epoch": 0.16673760374547777, "grad_norm": 19.703125, "learning_rate": 9.96208783399659e-06, "loss": 1.4997, "step": 1567 }, { "epoch": 0.1668440093636944, "grad_norm": 19.84375, "learning_rate": 9.962052302444572e-06, "loss": 1.3576, "step": 1568 }, { "epoch": 0.16695041498191104, "grad_norm": 15.125, "learning_rate": 9.962016770892553e-06, "loss": 1.226, "step": 1569 }, { "epoch": 0.1670568206001277, "grad_norm": 30.8125, "learning_rate": 9.961981239340535e-06, "loss": 1.6418, "step": 1570 }, { "epoch": 0.16716322621834434, "grad_norm": 20.1875, "learning_rate": 9.961945707788517e-06, "loss": 1.6663, "step": 1571 }, { "epoch": 0.16726963183656096, "grad_norm": 26.359375, "learning_rate": 9.961910176236498e-06, "loss": 1.3485, "step": 1572 }, { "epoch": 0.1673760374547776, "grad_norm": 16.375, "learning_rate": 9.96187464468448e-06, "loss": 1.311, "step": 1573 }, { "epoch": 0.16748244307299426, "grad_norm": 18.03125, "learning_rate": 9.961839113132462e-06, "loss": 1.3911, "step": 1574 }, { "epoch": 0.1675888486912109, "grad_norm": 19.5, "learning_rate": 9.961803581580443e-06, "loss": 1.0564, "step": 1575 }, { "epoch": 0.16769525430942753, "grad_norm": 16.9375, "learning_rate": 9.961768050028426e-06, "loss": 1.1499, "step": 1576 }, { "epoch": 0.16780165992764418, "grad_norm": 17.984375, "learning_rate": 9.961732518476408e-06, "loss": 1.5468, "step": 1577 }, { "epoch": 0.16790806554586082, "grad_norm": 22.703125, "learning_rate": 9.96169698692439e-06, "loss": 1.6002, "step": 1578 }, { "epoch": 0.16801447116407747, "grad_norm": 18.171875, "learning_rate": 9.961661455372371e-06, "loss": 1.2805, "step": 1579 }, { "epoch": 0.1681208767822941, "grad_norm": 20.859375, "learning_rate": 9.961625923820353e-06, "loss": 1.4438, "step": 1580 }, { "epoch": 0.16822728240051074, "grad_norm": 15.6484375, "learning_rate": 9.961590392268336e-06, "loss": 1.1178, "step": 1581 }, { "epoch": 0.1683336880187274, "grad_norm": 19.96875, "learning_rate": 9.961554860716318e-06, "loss": 1.3712, "step": 1582 }, { "epoch": 0.16844009363694404, "grad_norm": 18.171875, "learning_rate": 9.9615193291643e-06, "loss": 1.2984, "step": 1583 }, { "epoch": 0.16854649925516066, "grad_norm": 27.78125, "learning_rate": 9.961483797612281e-06, "loss": 1.9217, "step": 1584 }, { "epoch": 0.1686529048733773, "grad_norm": 23.90625, "learning_rate": 9.961448266060263e-06, "loss": 1.3281, "step": 1585 }, { "epoch": 0.16875931049159396, "grad_norm": 20.765625, "learning_rate": 9.961412734508244e-06, "loss": 1.3297, "step": 1586 }, { "epoch": 0.1688657161098106, "grad_norm": 17.078125, "learning_rate": 9.961377202956226e-06, "loss": 1.5527, "step": 1587 }, { "epoch": 0.16897212172802725, "grad_norm": 17.375, "learning_rate": 9.961341671404207e-06, "loss": 1.2475, "step": 1588 }, { "epoch": 0.16907852734624387, "grad_norm": 15.6953125, "learning_rate": 9.961306139852189e-06, "loss": 1.097, "step": 1589 }, { "epoch": 0.16918493296446052, "grad_norm": 12.6953125, "learning_rate": 9.96127060830017e-06, "loss": 1.1389, "step": 1590 }, { "epoch": 0.16929133858267717, "grad_norm": 15.4453125, "learning_rate": 9.961235076748154e-06, "loss": 1.2626, "step": 1591 }, { "epoch": 0.16939774420089382, "grad_norm": 20.5, "learning_rate": 9.961199545196136e-06, "loss": 1.6742, "step": 1592 }, { "epoch": 0.16950414981911044, "grad_norm": 16.828125, "learning_rate": 9.961164013644117e-06, "loss": 1.0905, "step": 1593 }, { "epoch": 0.1696105554373271, "grad_norm": 33.0625, "learning_rate": 9.961128482092099e-06, "loss": 1.5921, "step": 1594 }, { "epoch": 0.16971696105554374, "grad_norm": 26.109375, "learning_rate": 9.96109295054008e-06, "loss": 1.3879, "step": 1595 }, { "epoch": 0.16982336667376038, "grad_norm": 17.703125, "learning_rate": 9.961057418988062e-06, "loss": 0.9944, "step": 1596 }, { "epoch": 0.169929772291977, "grad_norm": 16.1875, "learning_rate": 9.961021887436044e-06, "loss": 1.3151, "step": 1597 }, { "epoch": 0.17003617791019365, "grad_norm": 15.0078125, "learning_rate": 9.960986355884025e-06, "loss": 1.4727, "step": 1598 }, { "epoch": 0.1701425835284103, "grad_norm": 14.15625, "learning_rate": 9.960950824332007e-06, "loss": 1.164, "step": 1599 }, { "epoch": 0.17024898914662695, "grad_norm": 16.4375, "learning_rate": 9.960915292779988e-06, "loss": 1.0449, "step": 1600 }, { "epoch": 0.17035539476484357, "grad_norm": 16.484375, "learning_rate": 9.960879761227972e-06, "loss": 1.1824, "step": 1601 }, { "epoch": 0.17046180038306022, "grad_norm": 18.578125, "learning_rate": 9.960844229675953e-06, "loss": 1.1015, "step": 1602 }, { "epoch": 0.17056820600127687, "grad_norm": 20.90625, "learning_rate": 9.960808698123935e-06, "loss": 1.2595, "step": 1603 }, { "epoch": 0.17067461161949352, "grad_norm": 22.09375, "learning_rate": 9.960773166571916e-06, "loss": 1.3027, "step": 1604 }, { "epoch": 0.17078101723771016, "grad_norm": 36.0625, "learning_rate": 9.960737635019898e-06, "loss": 1.5215, "step": 1605 }, { "epoch": 0.17088742285592678, "grad_norm": 20.546875, "learning_rate": 9.960702103467881e-06, "loss": 1.7325, "step": 1606 }, { "epoch": 0.17099382847414343, "grad_norm": 27.421875, "learning_rate": 9.960666571915863e-06, "loss": 1.7672, "step": 1607 }, { "epoch": 0.17110023409236008, "grad_norm": 28.421875, "learning_rate": 9.960631040363845e-06, "loss": 1.8292, "step": 1608 }, { "epoch": 0.17120663971057673, "grad_norm": 17.0, "learning_rate": 9.960595508811826e-06, "loss": 1.3661, "step": 1609 }, { "epoch": 0.17131304532879335, "grad_norm": 19.09375, "learning_rate": 9.960559977259808e-06, "loss": 1.3202, "step": 1610 }, { "epoch": 0.17141945094701, "grad_norm": 190.5, "learning_rate": 9.96052444570779e-06, "loss": 1.5328, "step": 1611 }, { "epoch": 0.17152585656522665, "grad_norm": 23.9375, "learning_rate": 9.960488914155771e-06, "loss": 1.1688, "step": 1612 }, { "epoch": 0.1716322621834433, "grad_norm": 19.796875, "learning_rate": 9.960453382603753e-06, "loss": 1.3612, "step": 1613 }, { "epoch": 0.17173866780165992, "grad_norm": 20.203125, "learning_rate": 9.960417851051734e-06, "loss": 1.2569, "step": 1614 }, { "epoch": 0.17184507341987657, "grad_norm": 14.78125, "learning_rate": 9.960382319499716e-06, "loss": 1.2675, "step": 1615 }, { "epoch": 0.1719514790380932, "grad_norm": 20.421875, "learning_rate": 9.960346787947697e-06, "loss": 1.3867, "step": 1616 }, { "epoch": 0.17205788465630986, "grad_norm": 14.8671875, "learning_rate": 9.96031125639568e-06, "loss": 1.2692, "step": 1617 }, { "epoch": 0.17216429027452648, "grad_norm": 18.421875, "learning_rate": 9.960275724843662e-06, "loss": 1.4647, "step": 1618 }, { "epoch": 0.17227069589274313, "grad_norm": 15.71875, "learning_rate": 9.960240193291644e-06, "loss": 1.3232, "step": 1619 }, { "epoch": 0.17237710151095978, "grad_norm": 21.5625, "learning_rate": 9.960204661739626e-06, "loss": 1.3402, "step": 1620 }, { "epoch": 0.17248350712917643, "grad_norm": 18.1875, "learning_rate": 9.960169130187607e-06, "loss": 1.1102, "step": 1621 }, { "epoch": 0.17258991274739308, "grad_norm": 24.265625, "learning_rate": 9.960133598635589e-06, "loss": 1.5219, "step": 1622 }, { "epoch": 0.1726963183656097, "grad_norm": 23.609375, "learning_rate": 9.96009806708357e-06, "loss": 1.4875, "step": 1623 }, { "epoch": 0.17280272398382635, "grad_norm": 16.3125, "learning_rate": 9.960062535531552e-06, "loss": 1.1574, "step": 1624 }, { "epoch": 0.172909129602043, "grad_norm": 16.453125, "learning_rate": 9.960027003979534e-06, "loss": 1.371, "step": 1625 }, { "epoch": 0.17301553522025964, "grad_norm": 13.3125, "learning_rate": 9.959991472427515e-06, "loss": 1.3261, "step": 1626 }, { "epoch": 0.17312194083847626, "grad_norm": 18.4375, "learning_rate": 9.959955940875499e-06, "loss": 1.4613, "step": 1627 }, { "epoch": 0.1732283464566929, "grad_norm": 17.40625, "learning_rate": 9.95992040932348e-06, "loss": 1.1544, "step": 1628 }, { "epoch": 0.17333475207490956, "grad_norm": 16.078125, "learning_rate": 9.959884877771462e-06, "loss": 1.4695, "step": 1629 }, { "epoch": 0.1734411576931262, "grad_norm": 23.625, "learning_rate": 9.959849346219443e-06, "loss": 1.2667, "step": 1630 }, { "epoch": 0.17354756331134283, "grad_norm": 23.359375, "learning_rate": 9.959813814667425e-06, "loss": 1.5002, "step": 1631 }, { "epoch": 0.17365396892955948, "grad_norm": 14.515625, "learning_rate": 9.959778283115408e-06, "loss": 1.2637, "step": 1632 }, { "epoch": 0.17376037454777613, "grad_norm": 20.703125, "learning_rate": 9.95974275156339e-06, "loss": 0.9646, "step": 1633 }, { "epoch": 0.17386678016599277, "grad_norm": 24.484375, "learning_rate": 9.959707220011371e-06, "loss": 1.1786, "step": 1634 }, { "epoch": 0.1739731857842094, "grad_norm": 13.453125, "learning_rate": 9.959671688459353e-06, "loss": 1.1296, "step": 1635 }, { "epoch": 0.17407959140242604, "grad_norm": 16.65625, "learning_rate": 9.959636156907335e-06, "loss": 1.2196, "step": 1636 }, { "epoch": 0.1741859970206427, "grad_norm": 27.15625, "learning_rate": 9.959600625355316e-06, "loss": 1.3507, "step": 1637 }, { "epoch": 0.17429240263885934, "grad_norm": 18.28125, "learning_rate": 9.959565093803298e-06, "loss": 1.6679, "step": 1638 }, { "epoch": 0.17439880825707596, "grad_norm": 18.65625, "learning_rate": 9.95952956225128e-06, "loss": 1.5467, "step": 1639 }, { "epoch": 0.1745052138752926, "grad_norm": 17.453125, "learning_rate": 9.959494030699261e-06, "loss": 1.1673, "step": 1640 }, { "epoch": 0.17461161949350926, "grad_norm": 14.0078125, "learning_rate": 9.959458499147243e-06, "loss": 1.1474, "step": 1641 }, { "epoch": 0.1747180251117259, "grad_norm": 19.84375, "learning_rate": 9.959422967595226e-06, "loss": 1.5183, "step": 1642 }, { "epoch": 0.17482443072994255, "grad_norm": 19.40625, "learning_rate": 9.959387436043208e-06, "loss": 1.4693, "step": 1643 }, { "epoch": 0.17493083634815917, "grad_norm": 21.453125, "learning_rate": 9.95935190449119e-06, "loss": 1.5211, "step": 1644 }, { "epoch": 0.17503724196637582, "grad_norm": 19.734375, "learning_rate": 9.95931637293917e-06, "loss": 1.3151, "step": 1645 }, { "epoch": 0.17514364758459247, "grad_norm": 11.0703125, "learning_rate": 9.959280841387152e-06, "loss": 1.0525, "step": 1646 }, { "epoch": 0.17525005320280912, "grad_norm": 17.421875, "learning_rate": 9.959245309835136e-06, "loss": 1.4209, "step": 1647 }, { "epoch": 0.17535645882102574, "grad_norm": 25.921875, "learning_rate": 9.959209778283117e-06, "loss": 1.7668, "step": 1648 }, { "epoch": 0.1754628644392424, "grad_norm": 24.359375, "learning_rate": 9.959174246731099e-06, "loss": 1.3608, "step": 1649 }, { "epoch": 0.17556927005745904, "grad_norm": 17.78125, "learning_rate": 9.959138715179079e-06, "loss": 1.0066, "step": 1650 }, { "epoch": 0.17567567567567569, "grad_norm": 42.1875, "learning_rate": 9.95910318362706e-06, "loss": 1.4223, "step": 1651 }, { "epoch": 0.1757820812938923, "grad_norm": 16.921875, "learning_rate": 9.959067652075044e-06, "loss": 1.2615, "step": 1652 }, { "epoch": 0.17588848691210895, "grad_norm": 18.46875, "learning_rate": 9.959032120523025e-06, "loss": 1.1655, "step": 1653 }, { "epoch": 0.1759948925303256, "grad_norm": 20.09375, "learning_rate": 9.958996588971007e-06, "loss": 1.7642, "step": 1654 }, { "epoch": 0.17610129814854225, "grad_norm": 16.125, "learning_rate": 9.958961057418989e-06, "loss": 1.0554, "step": 1655 }, { "epoch": 0.17620770376675887, "grad_norm": 20.765625, "learning_rate": 9.95892552586697e-06, "loss": 1.7492, "step": 1656 }, { "epoch": 0.17631410938497552, "grad_norm": 23.65625, "learning_rate": 9.958889994314952e-06, "loss": 1.2114, "step": 1657 }, { "epoch": 0.17642051500319217, "grad_norm": 14.9609375, "learning_rate": 9.958854462762935e-06, "loss": 1.0755, "step": 1658 }, { "epoch": 0.17652692062140882, "grad_norm": 17.59375, "learning_rate": 9.958818931210917e-06, "loss": 1.1752, "step": 1659 }, { "epoch": 0.17663332623962547, "grad_norm": 20.140625, "learning_rate": 9.958783399658898e-06, "loss": 1.3951, "step": 1660 }, { "epoch": 0.1767397318578421, "grad_norm": 17.921875, "learning_rate": 9.95874786810688e-06, "loss": 1.4473, "step": 1661 }, { "epoch": 0.17684613747605873, "grad_norm": 18.015625, "learning_rate": 9.958712336554862e-06, "loss": 1.2428, "step": 1662 }, { "epoch": 0.17695254309427538, "grad_norm": 16.421875, "learning_rate": 9.958676805002843e-06, "loss": 1.4272, "step": 1663 }, { "epoch": 0.17705894871249203, "grad_norm": 18.390625, "learning_rate": 9.958641273450825e-06, "loss": 1.5404, "step": 1664 }, { "epoch": 0.17716535433070865, "grad_norm": 19.4375, "learning_rate": 9.958605741898806e-06, "loss": 1.3681, "step": 1665 }, { "epoch": 0.1772717599489253, "grad_norm": 20.59375, "learning_rate": 9.958570210346788e-06, "loss": 1.2566, "step": 1666 }, { "epoch": 0.17737816556714195, "grad_norm": 17.625, "learning_rate": 9.95853467879477e-06, "loss": 1.4169, "step": 1667 }, { "epoch": 0.1774845711853586, "grad_norm": 19.375, "learning_rate": 9.958499147242753e-06, "loss": 1.4017, "step": 1668 }, { "epoch": 0.17759097680357522, "grad_norm": 15.6015625, "learning_rate": 9.958463615690734e-06, "loss": 1.3073, "step": 1669 }, { "epoch": 0.17769738242179187, "grad_norm": 24.09375, "learning_rate": 9.958428084138716e-06, "loss": 1.5949, "step": 1670 }, { "epoch": 0.17780378804000851, "grad_norm": 20.78125, "learning_rate": 9.958392552586698e-06, "loss": 1.2749, "step": 1671 }, { "epoch": 0.17791019365822516, "grad_norm": 20.46875, "learning_rate": 9.95835702103468e-06, "loss": 1.0804, "step": 1672 }, { "epoch": 0.17801659927644178, "grad_norm": 27.5, "learning_rate": 9.958321489482663e-06, "loss": 1.3185, "step": 1673 }, { "epoch": 0.17812300489465843, "grad_norm": 17.71875, "learning_rate": 9.958285957930644e-06, "loss": 1.2062, "step": 1674 }, { "epoch": 0.17822941051287508, "grad_norm": 17.234375, "learning_rate": 9.958250426378626e-06, "loss": 1.2205, "step": 1675 }, { "epoch": 0.17833581613109173, "grad_norm": 20.859375, "learning_rate": 9.958214894826607e-06, "loss": 1.4073, "step": 1676 }, { "epoch": 0.17844222174930838, "grad_norm": 18.71875, "learning_rate": 9.958179363274589e-06, "loss": 1.1805, "step": 1677 }, { "epoch": 0.178548627367525, "grad_norm": 17.90625, "learning_rate": 9.95814383172257e-06, "loss": 1.2492, "step": 1678 }, { "epoch": 0.17865503298574165, "grad_norm": 19.359375, "learning_rate": 9.958108300170552e-06, "loss": 1.3699, "step": 1679 }, { "epoch": 0.1787614386039583, "grad_norm": 30.421875, "learning_rate": 9.958072768618534e-06, "loss": 1.6079, "step": 1680 }, { "epoch": 0.17886784422217494, "grad_norm": 27.140625, "learning_rate": 9.958037237066515e-06, "loss": 1.046, "step": 1681 }, { "epoch": 0.17897424984039156, "grad_norm": 22.140625, "learning_rate": 9.958001705514497e-06, "loss": 1.4563, "step": 1682 }, { "epoch": 0.1790806554586082, "grad_norm": 18.28125, "learning_rate": 9.95796617396248e-06, "loss": 1.1035, "step": 1683 }, { "epoch": 0.17918706107682486, "grad_norm": 19.65625, "learning_rate": 9.957930642410462e-06, "loss": 1.2524, "step": 1684 }, { "epoch": 0.1792934666950415, "grad_norm": 26.0, "learning_rate": 9.957895110858444e-06, "loss": 1.1775, "step": 1685 }, { "epoch": 0.17939987231325813, "grad_norm": 14.8515625, "learning_rate": 9.957859579306425e-06, "loss": 1.1022, "step": 1686 }, { "epoch": 0.17950627793147478, "grad_norm": 22.125, "learning_rate": 9.957824047754407e-06, "loss": 1.6288, "step": 1687 }, { "epoch": 0.17961268354969143, "grad_norm": 14.953125, "learning_rate": 9.957788516202388e-06, "loss": 1.2538, "step": 1688 }, { "epoch": 0.17971908916790807, "grad_norm": 21.828125, "learning_rate": 9.95775298465037e-06, "loss": 1.3966, "step": 1689 }, { "epoch": 0.1798254947861247, "grad_norm": 16.125, "learning_rate": 9.957717453098352e-06, "loss": 1.3671, "step": 1690 }, { "epoch": 0.17993190040434134, "grad_norm": 15.03125, "learning_rate": 9.957681921546333e-06, "loss": 1.0477, "step": 1691 }, { "epoch": 0.180038306022558, "grad_norm": 18.078125, "learning_rate": 9.957646389994315e-06, "loss": 1.5363, "step": 1692 }, { "epoch": 0.18014471164077464, "grad_norm": 19.953125, "learning_rate": 9.957610858442296e-06, "loss": 1.6001, "step": 1693 }, { "epoch": 0.18025111725899126, "grad_norm": 20.1875, "learning_rate": 9.95757532689028e-06, "loss": 1.5159, "step": 1694 }, { "epoch": 0.1803575228772079, "grad_norm": 22.171875, "learning_rate": 9.957539795338261e-06, "loss": 1.349, "step": 1695 }, { "epoch": 0.18046392849542456, "grad_norm": 19.984375, "learning_rate": 9.957504263786243e-06, "loss": 1.7069, "step": 1696 }, { "epoch": 0.1805703341136412, "grad_norm": 15.1328125, "learning_rate": 9.957468732234225e-06, "loss": 1.3535, "step": 1697 }, { "epoch": 0.18067673973185785, "grad_norm": 14.8125, "learning_rate": 9.957433200682206e-06, "loss": 1.4179, "step": 1698 }, { "epoch": 0.18078314535007448, "grad_norm": 20.8125, "learning_rate": 9.95739766913019e-06, "loss": 1.2572, "step": 1699 }, { "epoch": 0.18088955096829112, "grad_norm": 17.828125, "learning_rate": 9.957362137578171e-06, "loss": 1.3957, "step": 1700 }, { "epoch": 0.18099595658650777, "grad_norm": 12.1875, "learning_rate": 9.957326606026153e-06, "loss": 1.0551, "step": 1701 }, { "epoch": 0.18110236220472442, "grad_norm": 14.953125, "learning_rate": 9.957291074474134e-06, "loss": 1.2611, "step": 1702 }, { "epoch": 0.18120876782294104, "grad_norm": 18.453125, "learning_rate": 9.957255542922116e-06, "loss": 1.0686, "step": 1703 }, { "epoch": 0.1813151734411577, "grad_norm": 17.546875, "learning_rate": 9.957220011370097e-06, "loss": 1.3057, "step": 1704 }, { "epoch": 0.18142157905937434, "grad_norm": 19.578125, "learning_rate": 9.957184479818079e-06, "loss": 1.1859, "step": 1705 }, { "epoch": 0.181527984677591, "grad_norm": 23.21875, "learning_rate": 9.95714894826606e-06, "loss": 1.3381, "step": 1706 }, { "epoch": 0.1816343902958076, "grad_norm": 19.515625, "learning_rate": 9.957113416714042e-06, "loss": 1.3811, "step": 1707 }, { "epoch": 0.18174079591402426, "grad_norm": 14.984375, "learning_rate": 9.957077885162024e-06, "loss": 1.0769, "step": 1708 }, { "epoch": 0.1818472015322409, "grad_norm": 13.3515625, "learning_rate": 9.957042353610007e-06, "loss": 1.2163, "step": 1709 }, { "epoch": 0.18195360715045755, "grad_norm": 13.46875, "learning_rate": 9.957006822057989e-06, "loss": 1.5775, "step": 1710 }, { "epoch": 0.18206001276867417, "grad_norm": 20.15625, "learning_rate": 9.95697129050597e-06, "loss": 1.8153, "step": 1711 }, { "epoch": 0.18216641838689082, "grad_norm": 23.671875, "learning_rate": 9.956935758953952e-06, "loss": 1.5567, "step": 1712 }, { "epoch": 0.18227282400510747, "grad_norm": 19.453125, "learning_rate": 9.956900227401934e-06, "loss": 1.3605, "step": 1713 }, { "epoch": 0.18237922962332412, "grad_norm": 19.984375, "learning_rate": 9.956864695849915e-06, "loss": 1.3441, "step": 1714 }, { "epoch": 0.18248563524154077, "grad_norm": 19.796875, "learning_rate": 9.956829164297897e-06, "loss": 1.4194, "step": 1715 }, { "epoch": 0.1825920408597574, "grad_norm": 13.1640625, "learning_rate": 9.956793632745878e-06, "loss": 1.1492, "step": 1716 }, { "epoch": 0.18269844647797404, "grad_norm": 15.6015625, "learning_rate": 9.95675810119386e-06, "loss": 1.2473, "step": 1717 }, { "epoch": 0.18280485209619068, "grad_norm": 18.296875, "learning_rate": 9.956722569641842e-06, "loss": 1.4928, "step": 1718 }, { "epoch": 0.18291125771440733, "grad_norm": 14.34375, "learning_rate": 9.956687038089825e-06, "loss": 1.2742, "step": 1719 }, { "epoch": 0.18301766333262395, "grad_norm": 13.4375, "learning_rate": 9.956651506537807e-06, "loss": 1.3885, "step": 1720 }, { "epoch": 0.1831240689508406, "grad_norm": 20.609375, "learning_rate": 9.956615974985788e-06, "loss": 1.5286, "step": 1721 }, { "epoch": 0.18323047456905725, "grad_norm": 17.984375, "learning_rate": 9.95658044343377e-06, "loss": 1.2754, "step": 1722 }, { "epoch": 0.1833368801872739, "grad_norm": 20.78125, "learning_rate": 9.956544911881751e-06, "loss": 1.681, "step": 1723 }, { "epoch": 0.18344328580549052, "grad_norm": 17.296875, "learning_rate": 9.956509380329735e-06, "loss": 1.0377, "step": 1724 }, { "epoch": 0.18354969142370717, "grad_norm": 19.828125, "learning_rate": 9.956473848777716e-06, "loss": 1.06, "step": 1725 }, { "epoch": 0.18365609704192382, "grad_norm": 19.359375, "learning_rate": 9.956438317225698e-06, "loss": 1.385, "step": 1726 }, { "epoch": 0.18376250266014046, "grad_norm": 15.859375, "learning_rate": 9.95640278567368e-06, "loss": 1.2515, "step": 1727 }, { "epoch": 0.18386890827835708, "grad_norm": 23.0625, "learning_rate": 9.956367254121661e-06, "loss": 1.2494, "step": 1728 }, { "epoch": 0.18397531389657373, "grad_norm": 21.859375, "learning_rate": 9.956331722569643e-06, "loss": 1.5583, "step": 1729 }, { "epoch": 0.18408171951479038, "grad_norm": 28.453125, "learning_rate": 9.956296191017624e-06, "loss": 1.4218, "step": 1730 }, { "epoch": 0.18418812513300703, "grad_norm": 19.625, "learning_rate": 9.956260659465606e-06, "loss": 1.4052, "step": 1731 }, { "epoch": 0.18429453075122365, "grad_norm": 18.125, "learning_rate": 9.956225127913588e-06, "loss": 1.4118, "step": 1732 }, { "epoch": 0.1844009363694403, "grad_norm": 16.75, "learning_rate": 9.95618959636157e-06, "loss": 1.3137, "step": 1733 }, { "epoch": 0.18450734198765695, "grad_norm": 13.7734375, "learning_rate": 9.95615406480955e-06, "loss": 1.4071, "step": 1734 }, { "epoch": 0.1846137476058736, "grad_norm": 21.046875, "learning_rate": 9.956118533257534e-06, "loss": 1.5619, "step": 1735 }, { "epoch": 0.18472015322409024, "grad_norm": 15.703125, "learning_rate": 9.956083001705516e-06, "loss": 1.3002, "step": 1736 }, { "epoch": 0.18482655884230687, "grad_norm": 17.796875, "learning_rate": 9.956047470153497e-06, "loss": 1.5431, "step": 1737 }, { "epoch": 0.1849329644605235, "grad_norm": 14.8203125, "learning_rate": 9.956011938601479e-06, "loss": 1.2135, "step": 1738 }, { "epoch": 0.18503937007874016, "grad_norm": 22.65625, "learning_rate": 9.95597640704946e-06, "loss": 1.3344, "step": 1739 }, { "epoch": 0.1851457756969568, "grad_norm": 13.0703125, "learning_rate": 9.955940875497442e-06, "loss": 1.0128, "step": 1740 }, { "epoch": 0.18525218131517343, "grad_norm": 14.4296875, "learning_rate": 9.955905343945424e-06, "loss": 1.4506, "step": 1741 }, { "epoch": 0.18535858693339008, "grad_norm": 12.2734375, "learning_rate": 9.955869812393405e-06, "loss": 1.3753, "step": 1742 }, { "epoch": 0.18546499255160673, "grad_norm": 14.75, "learning_rate": 9.955834280841387e-06, "loss": 1.3502, "step": 1743 }, { "epoch": 0.18557139816982338, "grad_norm": 14.9609375, "learning_rate": 9.955798749289369e-06, "loss": 1.4477, "step": 1744 }, { "epoch": 0.18567780378804, "grad_norm": 14.578125, "learning_rate": 9.955763217737352e-06, "loss": 1.4261, "step": 1745 }, { "epoch": 0.18578420940625665, "grad_norm": 15.8359375, "learning_rate": 9.955727686185333e-06, "loss": 1.1664, "step": 1746 }, { "epoch": 0.1858906150244733, "grad_norm": 20.1875, "learning_rate": 9.955692154633315e-06, "loss": 1.7556, "step": 1747 }, { "epoch": 0.18599702064268994, "grad_norm": 15.9296875, "learning_rate": 9.955656623081297e-06, "loss": 1.7653, "step": 1748 }, { "epoch": 0.18610342626090656, "grad_norm": 13.65625, "learning_rate": 9.955621091529278e-06, "loss": 1.2754, "step": 1749 }, { "epoch": 0.1862098318791232, "grad_norm": 15.8125, "learning_rate": 9.955585559977262e-06, "loss": 1.3692, "step": 1750 }, { "epoch": 0.18631623749733986, "grad_norm": 19.03125, "learning_rate": 9.955550028425243e-06, "loss": 1.5678, "step": 1751 }, { "epoch": 0.1864226431155565, "grad_norm": 15.546875, "learning_rate": 9.955514496873225e-06, "loss": 1.1536, "step": 1752 }, { "epoch": 0.18652904873377316, "grad_norm": 19.921875, "learning_rate": 9.955478965321206e-06, "loss": 1.3071, "step": 1753 }, { "epoch": 0.18663545435198978, "grad_norm": 14.8984375, "learning_rate": 9.955443433769188e-06, "loss": 0.9457, "step": 1754 }, { "epoch": 0.18674185997020643, "grad_norm": 16.03125, "learning_rate": 9.95540790221717e-06, "loss": 1.4183, "step": 1755 }, { "epoch": 0.18684826558842307, "grad_norm": 16.25, "learning_rate": 9.955372370665151e-06, "loss": 1.2732, "step": 1756 }, { "epoch": 0.18695467120663972, "grad_norm": 14.21875, "learning_rate": 9.955336839113133e-06, "loss": 1.0872, "step": 1757 }, { "epoch": 0.18706107682485634, "grad_norm": 13.671875, "learning_rate": 9.955301307561114e-06, "loss": 1.1645, "step": 1758 }, { "epoch": 0.187167482443073, "grad_norm": 15.5546875, "learning_rate": 9.955265776009096e-06, "loss": 1.2797, "step": 1759 }, { "epoch": 0.18727388806128964, "grad_norm": 18.9375, "learning_rate": 9.95523024445708e-06, "loss": 1.1254, "step": 1760 }, { "epoch": 0.1873802936795063, "grad_norm": 17.609375, "learning_rate": 9.955194712905061e-06, "loss": 1.4192, "step": 1761 }, { "epoch": 0.1874866992977229, "grad_norm": 21.515625, "learning_rate": 9.955159181353043e-06, "loss": 1.3392, "step": 1762 }, { "epoch": 0.18759310491593956, "grad_norm": 13.515625, "learning_rate": 9.955123649801024e-06, "loss": 1.3212, "step": 1763 }, { "epoch": 0.1876995105341562, "grad_norm": 17.171875, "learning_rate": 9.955088118249006e-06, "loss": 0.9774, "step": 1764 }, { "epoch": 0.18780591615237285, "grad_norm": 14.0859375, "learning_rate": 9.955052586696989e-06, "loss": 1.4347, "step": 1765 }, { "epoch": 0.18791232177058947, "grad_norm": 18.96875, "learning_rate": 9.95501705514497e-06, "loss": 1.0806, "step": 1766 }, { "epoch": 0.18801872738880612, "grad_norm": 21.875, "learning_rate": 9.954981523592952e-06, "loss": 1.3867, "step": 1767 }, { "epoch": 0.18812513300702277, "grad_norm": 25.1875, "learning_rate": 9.954945992040934e-06, "loss": 1.4491, "step": 1768 }, { "epoch": 0.18823153862523942, "grad_norm": 25.234375, "learning_rate": 9.954910460488915e-06, "loss": 1.4484, "step": 1769 }, { "epoch": 0.18833794424345607, "grad_norm": 28.890625, "learning_rate": 9.954874928936897e-06, "loss": 1.5707, "step": 1770 }, { "epoch": 0.1884443498616727, "grad_norm": 15.90625, "learning_rate": 9.954839397384879e-06, "loss": 1.1975, "step": 1771 }, { "epoch": 0.18855075547988934, "grad_norm": 58.46875, "learning_rate": 9.95480386583286e-06, "loss": 1.4289, "step": 1772 }, { "epoch": 0.18865716109810599, "grad_norm": 19.875, "learning_rate": 9.954768334280842e-06, "loss": 1.2074, "step": 1773 }, { "epoch": 0.18876356671632263, "grad_norm": 24.734375, "learning_rate": 9.954732802728824e-06, "loss": 1.3, "step": 1774 }, { "epoch": 0.18886997233453925, "grad_norm": 21.640625, "learning_rate": 9.954697271176805e-06, "loss": 1.7155, "step": 1775 }, { "epoch": 0.1889763779527559, "grad_norm": 20.34375, "learning_rate": 9.954661739624788e-06, "loss": 1.4868, "step": 1776 }, { "epoch": 0.18908278357097255, "grad_norm": 12.4609375, "learning_rate": 9.95462620807277e-06, "loss": 1.0935, "step": 1777 }, { "epoch": 0.1891891891891892, "grad_norm": 26.953125, "learning_rate": 9.954590676520752e-06, "loss": 1.246, "step": 1778 }, { "epoch": 0.18929559480740582, "grad_norm": 19.0625, "learning_rate": 9.954555144968733e-06, "loss": 1.4911, "step": 1779 }, { "epoch": 0.18940200042562247, "grad_norm": 28.375, "learning_rate": 9.954519613416715e-06, "loss": 1.202, "step": 1780 }, { "epoch": 0.18950840604383912, "grad_norm": 13.75, "learning_rate": 9.954484081864696e-06, "loss": 1.1055, "step": 1781 }, { "epoch": 0.18961481166205577, "grad_norm": 21.46875, "learning_rate": 9.954448550312678e-06, "loss": 1.6264, "step": 1782 }, { "epoch": 0.1897212172802724, "grad_norm": 22.8125, "learning_rate": 9.95441301876066e-06, "loss": 1.3731, "step": 1783 }, { "epoch": 0.18982762289848903, "grad_norm": 15.546875, "learning_rate": 9.954377487208641e-06, "loss": 1.1619, "step": 1784 }, { "epoch": 0.18993402851670568, "grad_norm": 20.828125, "learning_rate": 9.954341955656623e-06, "loss": 1.6158, "step": 1785 }, { "epoch": 0.19004043413492233, "grad_norm": 19.953125, "learning_rate": 9.954306424104606e-06, "loss": 1.0888, "step": 1786 }, { "epoch": 0.19014683975313895, "grad_norm": 19.171875, "learning_rate": 9.954270892552588e-06, "loss": 1.2515, "step": 1787 }, { "epoch": 0.1902532453713556, "grad_norm": 22.609375, "learning_rate": 9.95423536100057e-06, "loss": 1.2963, "step": 1788 }, { "epoch": 0.19035965098957225, "grad_norm": 19.015625, "learning_rate": 9.954199829448551e-06, "loss": 1.2999, "step": 1789 }, { "epoch": 0.1904660566077889, "grad_norm": 17.890625, "learning_rate": 9.954164297896533e-06, "loss": 1.4398, "step": 1790 }, { "epoch": 0.19057246222600555, "grad_norm": 22.390625, "learning_rate": 9.954128766344516e-06, "loss": 1.3923, "step": 1791 }, { "epoch": 0.19067886784422217, "grad_norm": 17.921875, "learning_rate": 9.954093234792498e-06, "loss": 1.3671, "step": 1792 }, { "epoch": 0.19078527346243881, "grad_norm": 22.875, "learning_rate": 9.954057703240479e-06, "loss": 1.3572, "step": 1793 }, { "epoch": 0.19089167908065546, "grad_norm": 18.71875, "learning_rate": 9.95402217168846e-06, "loss": 1.0915, "step": 1794 }, { "epoch": 0.1909980846988721, "grad_norm": 18.015625, "learning_rate": 9.953986640136442e-06, "loss": 1.564, "step": 1795 }, { "epoch": 0.19110449031708873, "grad_norm": 17.3125, "learning_rate": 9.953951108584424e-06, "loss": 1.2329, "step": 1796 }, { "epoch": 0.19121089593530538, "grad_norm": 23.484375, "learning_rate": 9.953915577032406e-06, "loss": 1.3499, "step": 1797 }, { "epoch": 0.19131730155352203, "grad_norm": 20.765625, "learning_rate": 9.953880045480387e-06, "loss": 1.3019, "step": 1798 }, { "epoch": 0.19142370717173868, "grad_norm": 23.34375, "learning_rate": 9.953844513928369e-06, "loss": 1.4989, "step": 1799 }, { "epoch": 0.1915301127899553, "grad_norm": 11.8671875, "learning_rate": 9.95380898237635e-06, "loss": 1.2117, "step": 1800 }, { "epoch": 0.19163651840817195, "grad_norm": 14.90625, "learning_rate": 9.953773450824334e-06, "loss": 1.1192, "step": 1801 }, { "epoch": 0.1917429240263886, "grad_norm": 15.6875, "learning_rate": 9.953737919272315e-06, "loss": 1.3441, "step": 1802 }, { "epoch": 0.19184932964460524, "grad_norm": 20.734375, "learning_rate": 9.953702387720297e-06, "loss": 1.6207, "step": 1803 }, { "epoch": 0.19195573526282186, "grad_norm": 14.8984375, "learning_rate": 9.953666856168278e-06, "loss": 1.1524, "step": 1804 }, { "epoch": 0.1920621408810385, "grad_norm": 16.03125, "learning_rate": 9.95363132461626e-06, "loss": 1.4591, "step": 1805 }, { "epoch": 0.19216854649925516, "grad_norm": 24.078125, "learning_rate": 9.953595793064242e-06, "loss": 1.284, "step": 1806 }, { "epoch": 0.1922749521174718, "grad_norm": 16.296875, "learning_rate": 9.953560261512223e-06, "loss": 1.108, "step": 1807 }, { "epoch": 0.19238135773568846, "grad_norm": 16.140625, "learning_rate": 9.953524729960205e-06, "loss": 0.7745, "step": 1808 }, { "epoch": 0.19248776335390508, "grad_norm": 17.359375, "learning_rate": 9.953489198408187e-06, "loss": 1.1677, "step": 1809 }, { "epoch": 0.19259416897212173, "grad_norm": 15.4921875, "learning_rate": 9.953453666856168e-06, "loss": 1.285, "step": 1810 }, { "epoch": 0.19270057459033837, "grad_norm": 17.484375, "learning_rate": 9.95341813530415e-06, "loss": 1.5409, "step": 1811 }, { "epoch": 0.19280698020855502, "grad_norm": 18.796875, "learning_rate": 9.953382603752133e-06, "loss": 1.405, "step": 1812 }, { "epoch": 0.19291338582677164, "grad_norm": 18.703125, "learning_rate": 9.953347072200115e-06, "loss": 1.3977, "step": 1813 }, { "epoch": 0.1930197914449883, "grad_norm": 19.859375, "learning_rate": 9.953311540648096e-06, "loss": 1.726, "step": 1814 }, { "epoch": 0.19312619706320494, "grad_norm": 16.796875, "learning_rate": 9.953276009096078e-06, "loss": 0.9308, "step": 1815 }, { "epoch": 0.1932326026814216, "grad_norm": 14.1953125, "learning_rate": 9.95324047754406e-06, "loss": 1.2975, "step": 1816 }, { "epoch": 0.1933390082996382, "grad_norm": 16.5, "learning_rate": 9.953204945992043e-06, "loss": 1.3058, "step": 1817 }, { "epoch": 0.19344541391785486, "grad_norm": 14.984375, "learning_rate": 9.953169414440024e-06, "loss": 1.3541, "step": 1818 }, { "epoch": 0.1935518195360715, "grad_norm": 15.2890625, "learning_rate": 9.953133882888006e-06, "loss": 1.2621, "step": 1819 }, { "epoch": 0.19365822515428815, "grad_norm": 11.0625, "learning_rate": 9.953098351335988e-06, "loss": 1.0744, "step": 1820 }, { "epoch": 0.19376463077250478, "grad_norm": 14.9375, "learning_rate": 9.95306281978397e-06, "loss": 1.1996, "step": 1821 }, { "epoch": 0.19387103639072142, "grad_norm": 15.390625, "learning_rate": 9.95302728823195e-06, "loss": 1.3545, "step": 1822 }, { "epoch": 0.19397744200893807, "grad_norm": 10.828125, "learning_rate": 9.952991756679932e-06, "loss": 0.9228, "step": 1823 }, { "epoch": 0.19408384762715472, "grad_norm": 16.46875, "learning_rate": 9.952956225127914e-06, "loss": 1.0598, "step": 1824 }, { "epoch": 0.19419025324537137, "grad_norm": 17.28125, "learning_rate": 9.952920693575896e-06, "loss": 1.3297, "step": 1825 }, { "epoch": 0.194296658863588, "grad_norm": 14.828125, "learning_rate": 9.952885162023877e-06, "loss": 1.1691, "step": 1826 }, { "epoch": 0.19440306448180464, "grad_norm": 15.1171875, "learning_rate": 9.95284963047186e-06, "loss": 1.3987, "step": 1827 }, { "epoch": 0.1945094701000213, "grad_norm": 18.640625, "learning_rate": 9.952814098919842e-06, "loss": 1.3287, "step": 1828 }, { "epoch": 0.19461587571823794, "grad_norm": 24.59375, "learning_rate": 9.952778567367824e-06, "loss": 1.876, "step": 1829 }, { "epoch": 0.19472228133645456, "grad_norm": 18.359375, "learning_rate": 9.952743035815805e-06, "loss": 1.2159, "step": 1830 }, { "epoch": 0.1948286869546712, "grad_norm": 14.28125, "learning_rate": 9.952707504263787e-06, "loss": 1.0882, "step": 1831 }, { "epoch": 0.19493509257288785, "grad_norm": 16.1875, "learning_rate": 9.952671972711769e-06, "loss": 1.1386, "step": 1832 }, { "epoch": 0.1950414981911045, "grad_norm": 15.0078125, "learning_rate": 9.95263644115975e-06, "loss": 1.2109, "step": 1833 }, { "epoch": 0.19514790380932112, "grad_norm": 17.359375, "learning_rate": 9.952600909607732e-06, "loss": 1.3539, "step": 1834 }, { "epoch": 0.19525430942753777, "grad_norm": 27.75, "learning_rate": 9.952565378055713e-06, "loss": 1.2185, "step": 1835 }, { "epoch": 0.19536071504575442, "grad_norm": 22.21875, "learning_rate": 9.952529846503695e-06, "loss": 1.6379, "step": 1836 }, { "epoch": 0.19546712066397107, "grad_norm": 26.71875, "learning_rate": 9.952494314951678e-06, "loss": 1.4237, "step": 1837 }, { "epoch": 0.1955735262821877, "grad_norm": 17.859375, "learning_rate": 9.95245878339966e-06, "loss": 1.2444, "step": 1838 }, { "epoch": 0.19567993190040434, "grad_norm": 11.796875, "learning_rate": 9.952423251847641e-06, "loss": 1.0666, "step": 1839 }, { "epoch": 0.19578633751862098, "grad_norm": 27.921875, "learning_rate": 9.952387720295623e-06, "loss": 1.714, "step": 1840 }, { "epoch": 0.19589274313683763, "grad_norm": 24.0625, "learning_rate": 9.952352188743605e-06, "loss": 1.3996, "step": 1841 }, { "epoch": 0.19599914875505425, "grad_norm": 20.765625, "learning_rate": 9.952316657191588e-06, "loss": 1.3857, "step": 1842 }, { "epoch": 0.1961055543732709, "grad_norm": 15.2109375, "learning_rate": 9.95228112563957e-06, "loss": 1.1754, "step": 1843 }, { "epoch": 0.19621195999148755, "grad_norm": 15.1953125, "learning_rate": 9.952245594087551e-06, "loss": 1.017, "step": 1844 }, { "epoch": 0.1963183656097042, "grad_norm": 15.4453125, "learning_rate": 9.952210062535533e-06, "loss": 1.2118, "step": 1845 }, { "epoch": 0.19642477122792085, "grad_norm": 18.59375, "learning_rate": 9.952174530983514e-06, "loss": 1.3421, "step": 1846 }, { "epoch": 0.19653117684613747, "grad_norm": 22.75, "learning_rate": 9.952138999431496e-06, "loss": 1.6739, "step": 1847 }, { "epoch": 0.19663758246435412, "grad_norm": 22.234375, "learning_rate": 9.952103467879478e-06, "loss": 1.1456, "step": 1848 }, { "epoch": 0.19674398808257076, "grad_norm": 24.125, "learning_rate": 9.95206793632746e-06, "loss": 1.1322, "step": 1849 }, { "epoch": 0.1968503937007874, "grad_norm": 18.25, "learning_rate": 9.952032404775441e-06, "loss": 1.2031, "step": 1850 }, { "epoch": 0.19695679931900403, "grad_norm": 13.8984375, "learning_rate": 9.951996873223422e-06, "loss": 1.3691, "step": 1851 }, { "epoch": 0.19706320493722068, "grad_norm": 18.40625, "learning_rate": 9.951961341671404e-06, "loss": 1.1654, "step": 1852 }, { "epoch": 0.19716961055543733, "grad_norm": 13.28125, "learning_rate": 9.951925810119387e-06, "loss": 1.3766, "step": 1853 }, { "epoch": 0.19727601617365398, "grad_norm": 20.6875, "learning_rate": 9.951890278567369e-06, "loss": 1.4085, "step": 1854 }, { "epoch": 0.1973824217918706, "grad_norm": 24.15625, "learning_rate": 9.95185474701535e-06, "loss": 1.9399, "step": 1855 }, { "epoch": 0.19748882741008725, "grad_norm": 22.078125, "learning_rate": 9.951819215463332e-06, "loss": 1.4377, "step": 1856 }, { "epoch": 0.1975952330283039, "grad_norm": 15.515625, "learning_rate": 9.951783683911314e-06, "loss": 0.9724, "step": 1857 }, { "epoch": 0.19770163864652054, "grad_norm": 23.328125, "learning_rate": 9.951748152359297e-06, "loss": 1.4719, "step": 1858 }, { "epoch": 0.19780804426473717, "grad_norm": 27.625, "learning_rate": 9.951712620807277e-06, "loss": 1.5165, "step": 1859 }, { "epoch": 0.1979144498829538, "grad_norm": 15.5390625, "learning_rate": 9.951677089255259e-06, "loss": 1.4173, "step": 1860 }, { "epoch": 0.19802085550117046, "grad_norm": 16.21875, "learning_rate": 9.95164155770324e-06, "loss": 1.1981, "step": 1861 }, { "epoch": 0.1981272611193871, "grad_norm": 16.1875, "learning_rate": 9.951606026151222e-06, "loss": 1.5427, "step": 1862 }, { "epoch": 0.19823366673760376, "grad_norm": 18.546875, "learning_rate": 9.951570494599205e-06, "loss": 1.4566, "step": 1863 }, { "epoch": 0.19834007235582038, "grad_norm": 13.2421875, "learning_rate": 9.951534963047187e-06, "loss": 1.3922, "step": 1864 }, { "epoch": 0.19844647797403703, "grad_norm": 32.84375, "learning_rate": 9.951499431495168e-06, "loss": 1.2919, "step": 1865 }, { "epoch": 0.19855288359225368, "grad_norm": 15.4609375, "learning_rate": 9.95146389994315e-06, "loss": 1.0719, "step": 1866 }, { "epoch": 0.19865928921047032, "grad_norm": 18.3125, "learning_rate": 9.951428368391132e-06, "loss": 1.3769, "step": 1867 }, { "epoch": 0.19876569482868695, "grad_norm": 19.734375, "learning_rate": 9.951392836839115e-06, "loss": 1.6072, "step": 1868 }, { "epoch": 0.1988721004469036, "grad_norm": 18.609375, "learning_rate": 9.951357305287096e-06, "loss": 1.8512, "step": 1869 }, { "epoch": 0.19897850606512024, "grad_norm": 17.015625, "learning_rate": 9.951321773735078e-06, "loss": 1.2098, "step": 1870 }, { "epoch": 0.1990849116833369, "grad_norm": 26.25, "learning_rate": 9.95128624218306e-06, "loss": 1.3896, "step": 1871 }, { "epoch": 0.1991913173015535, "grad_norm": 15.0, "learning_rate": 9.951250710631041e-06, "loss": 1.235, "step": 1872 }, { "epoch": 0.19929772291977016, "grad_norm": 19.984375, "learning_rate": 9.951215179079023e-06, "loss": 1.2546, "step": 1873 }, { "epoch": 0.1994041285379868, "grad_norm": 28.09375, "learning_rate": 9.951179647527005e-06, "loss": 1.4455, "step": 1874 }, { "epoch": 0.19951053415620346, "grad_norm": 14.1875, "learning_rate": 9.951144115974986e-06, "loss": 1.1371, "step": 1875 }, { "epoch": 0.19961693977442008, "grad_norm": 18.28125, "learning_rate": 9.951108584422968e-06, "loss": 1.2036, "step": 1876 }, { "epoch": 0.19972334539263673, "grad_norm": 29.4375, "learning_rate": 9.95107305287095e-06, "loss": 1.4088, "step": 1877 }, { "epoch": 0.19982975101085337, "grad_norm": 14.3828125, "learning_rate": 9.951037521318933e-06, "loss": 1.2876, "step": 1878 }, { "epoch": 0.19993615662907002, "grad_norm": 17.4375, "learning_rate": 9.951001989766914e-06, "loss": 1.3803, "step": 1879 }, { "epoch": 0.20004256224728667, "grad_norm": 16.0625, "learning_rate": 9.950966458214896e-06, "loss": 1.4343, "step": 1880 }, { "epoch": 0.2001489678655033, "grad_norm": 15.2734375, "learning_rate": 9.950930926662877e-06, "loss": 1.4826, "step": 1881 }, { "epoch": 0.20025537348371994, "grad_norm": 20.59375, "learning_rate": 9.950895395110859e-06, "loss": 1.4841, "step": 1882 }, { "epoch": 0.2003617791019366, "grad_norm": 18.4375, "learning_rate": 9.950859863558842e-06, "loss": 1.2509, "step": 1883 }, { "epoch": 0.20046818472015324, "grad_norm": 15.3203125, "learning_rate": 9.950824332006824e-06, "loss": 1.7621, "step": 1884 }, { "epoch": 0.20057459033836986, "grad_norm": 16.703125, "learning_rate": 9.950788800454806e-06, "loss": 1.4287, "step": 1885 }, { "epoch": 0.2006809959565865, "grad_norm": 16.890625, "learning_rate": 9.950753268902787e-06, "loss": 1.4934, "step": 1886 }, { "epoch": 0.20078740157480315, "grad_norm": 15.78125, "learning_rate": 9.950717737350769e-06, "loss": 1.2882, "step": 1887 }, { "epoch": 0.2008938071930198, "grad_norm": 14.609375, "learning_rate": 9.95068220579875e-06, "loss": 1.1462, "step": 1888 }, { "epoch": 0.20100021281123642, "grad_norm": 15.0859375, "learning_rate": 9.950646674246732e-06, "loss": 1.3846, "step": 1889 }, { "epoch": 0.20110661842945307, "grad_norm": 17.640625, "learning_rate": 9.950611142694714e-06, "loss": 1.455, "step": 1890 }, { "epoch": 0.20121302404766972, "grad_norm": 21.390625, "learning_rate": 9.950575611142695e-06, "loss": 1.7318, "step": 1891 }, { "epoch": 0.20131942966588637, "grad_norm": 18.296875, "learning_rate": 9.950540079590677e-06, "loss": 1.139, "step": 1892 }, { "epoch": 0.201425835284103, "grad_norm": 20.078125, "learning_rate": 9.950504548038658e-06, "loss": 1.3278, "step": 1893 }, { "epoch": 0.20153224090231964, "grad_norm": 13.4375, "learning_rate": 9.950469016486642e-06, "loss": 1.0149, "step": 1894 }, { "epoch": 0.20163864652053629, "grad_norm": 18.453125, "learning_rate": 9.950433484934623e-06, "loss": 1.2887, "step": 1895 }, { "epoch": 0.20174505213875293, "grad_norm": 20.3125, "learning_rate": 9.950397953382605e-06, "loss": 1.7214, "step": 1896 }, { "epoch": 0.20185145775696955, "grad_norm": 14.234375, "learning_rate": 9.950362421830587e-06, "loss": 1.229, "step": 1897 }, { "epoch": 0.2019578633751862, "grad_norm": 12.625, "learning_rate": 9.950326890278568e-06, "loss": 1.0218, "step": 1898 }, { "epoch": 0.20206426899340285, "grad_norm": 17.296875, "learning_rate": 9.95029135872655e-06, "loss": 1.0697, "step": 1899 }, { "epoch": 0.2021706746116195, "grad_norm": 15.5546875, "learning_rate": 9.950255827174531e-06, "loss": 1.301, "step": 1900 }, { "epoch": 0.20227708022983615, "grad_norm": 17.78125, "learning_rate": 9.950220295622513e-06, "loss": 0.9046, "step": 1901 }, { "epoch": 0.20238348584805277, "grad_norm": 16.203125, "learning_rate": 9.950184764070495e-06, "loss": 1.4928, "step": 1902 }, { "epoch": 0.20248989146626942, "grad_norm": 18.53125, "learning_rate": 9.950149232518476e-06, "loss": 1.5396, "step": 1903 }, { "epoch": 0.20259629708448607, "grad_norm": 20.9375, "learning_rate": 9.95011370096646e-06, "loss": 1.1851, "step": 1904 }, { "epoch": 0.20270270270270271, "grad_norm": 14.484375, "learning_rate": 9.950078169414441e-06, "loss": 1.3673, "step": 1905 }, { "epoch": 0.20280910832091933, "grad_norm": 18.421875, "learning_rate": 9.950042637862423e-06, "loss": 1.3678, "step": 1906 }, { "epoch": 0.20291551393913598, "grad_norm": 15.796875, "learning_rate": 9.950007106310404e-06, "loss": 1.4538, "step": 1907 }, { "epoch": 0.20302191955735263, "grad_norm": 15.03125, "learning_rate": 9.949971574758386e-06, "loss": 1.5046, "step": 1908 }, { "epoch": 0.20312832517556928, "grad_norm": 14.3671875, "learning_rate": 9.94993604320637e-06, "loss": 1.2692, "step": 1909 }, { "epoch": 0.2032347307937859, "grad_norm": 20.5, "learning_rate": 9.94990051165435e-06, "loss": 1.326, "step": 1910 }, { "epoch": 0.20334113641200255, "grad_norm": 16.6875, "learning_rate": 9.949864980102332e-06, "loss": 1.4598, "step": 1911 }, { "epoch": 0.2034475420302192, "grad_norm": 22.640625, "learning_rate": 9.949829448550314e-06, "loss": 1.334, "step": 1912 }, { "epoch": 0.20355394764843585, "grad_norm": 18.015625, "learning_rate": 9.949793916998296e-06, "loss": 1.1624, "step": 1913 }, { "epoch": 0.20366035326665247, "grad_norm": 16.34375, "learning_rate": 9.949758385446277e-06, "loss": 1.3295, "step": 1914 }, { "epoch": 0.20376675888486911, "grad_norm": 15.90625, "learning_rate": 9.949722853894259e-06, "loss": 1.2958, "step": 1915 }, { "epoch": 0.20387316450308576, "grad_norm": 20.109375, "learning_rate": 9.94968732234224e-06, "loss": 1.5595, "step": 1916 }, { "epoch": 0.2039795701213024, "grad_norm": 11.4765625, "learning_rate": 9.949651790790222e-06, "loss": 1.0763, "step": 1917 }, { "epoch": 0.20408597573951906, "grad_norm": 14.90625, "learning_rate": 9.949616259238204e-06, "loss": 1.1366, "step": 1918 }, { "epoch": 0.20419238135773568, "grad_norm": 17.484375, "learning_rate": 9.949580727686187e-06, "loss": 1.3661, "step": 1919 }, { "epoch": 0.20429878697595233, "grad_norm": 15.078125, "learning_rate": 9.949545196134169e-06, "loss": 1.1655, "step": 1920 }, { "epoch": 0.20440519259416898, "grad_norm": 14.5546875, "learning_rate": 9.94950966458215e-06, "loss": 1.3728, "step": 1921 }, { "epoch": 0.20451159821238563, "grad_norm": 18.71875, "learning_rate": 9.949474133030132e-06, "loss": 1.2282, "step": 1922 }, { "epoch": 0.20461800383060225, "grad_norm": 22.515625, "learning_rate": 9.949438601478113e-06, "loss": 1.935, "step": 1923 }, { "epoch": 0.2047244094488189, "grad_norm": 13.328125, "learning_rate": 9.949403069926095e-06, "loss": 1.363, "step": 1924 }, { "epoch": 0.20483081506703554, "grad_norm": 17.875, "learning_rate": 9.949367538374077e-06, "loss": 1.0132, "step": 1925 }, { "epoch": 0.2049372206852522, "grad_norm": 18.65625, "learning_rate": 9.949332006822058e-06, "loss": 1.4272, "step": 1926 }, { "epoch": 0.2050436263034688, "grad_norm": 20.625, "learning_rate": 9.94929647527004e-06, "loss": 1.6236, "step": 1927 }, { "epoch": 0.20515003192168546, "grad_norm": 16.15625, "learning_rate": 9.949260943718021e-06, "loss": 1.4155, "step": 1928 }, { "epoch": 0.2052564375399021, "grad_norm": 14.6171875, "learning_rate": 9.949225412166003e-06, "loss": 1.4877, "step": 1929 }, { "epoch": 0.20536284315811876, "grad_norm": 19.359375, "learning_rate": 9.949189880613986e-06, "loss": 1.6474, "step": 1930 }, { "epoch": 0.20546924877633538, "grad_norm": 15.9296875, "learning_rate": 9.949154349061968e-06, "loss": 1.5211, "step": 1931 }, { "epoch": 0.20557565439455203, "grad_norm": 16.078125, "learning_rate": 9.94911881750995e-06, "loss": 1.1193, "step": 1932 }, { "epoch": 0.20568206001276867, "grad_norm": 14.0703125, "learning_rate": 9.949083285957931e-06, "loss": 1.392, "step": 1933 }, { "epoch": 0.20578846563098532, "grad_norm": 20.984375, "learning_rate": 9.949047754405913e-06, "loss": 1.4236, "step": 1934 }, { "epoch": 0.20589487124920197, "grad_norm": 16.53125, "learning_rate": 9.949012222853896e-06, "loss": 1.4987, "step": 1935 }, { "epoch": 0.2060012768674186, "grad_norm": 17.34375, "learning_rate": 9.948976691301878e-06, "loss": 0.7903, "step": 1936 }, { "epoch": 0.20610768248563524, "grad_norm": 15.4296875, "learning_rate": 9.94894115974986e-06, "loss": 1.0844, "step": 1937 }, { "epoch": 0.2062140881038519, "grad_norm": 19.953125, "learning_rate": 9.948905628197841e-06, "loss": 1.5616, "step": 1938 }, { "epoch": 0.20632049372206854, "grad_norm": 21.0625, "learning_rate": 9.948870096645822e-06, "loss": 1.4989, "step": 1939 }, { "epoch": 0.20642689934028516, "grad_norm": 18.140625, "learning_rate": 9.948834565093804e-06, "loss": 1.3737, "step": 1940 }, { "epoch": 0.2065333049585018, "grad_norm": 21.78125, "learning_rate": 9.948799033541786e-06, "loss": 1.1182, "step": 1941 }, { "epoch": 0.20663971057671845, "grad_norm": 13.4296875, "learning_rate": 9.948763501989767e-06, "loss": 0.9623, "step": 1942 }, { "epoch": 0.2067461161949351, "grad_norm": 17.0625, "learning_rate": 9.948727970437749e-06, "loss": 1.6222, "step": 1943 }, { "epoch": 0.20685252181315172, "grad_norm": 11.796875, "learning_rate": 9.94869243888573e-06, "loss": 1.2001, "step": 1944 }, { "epoch": 0.20695892743136837, "grad_norm": 17.390625, "learning_rate": 9.948656907333714e-06, "loss": 1.0476, "step": 1945 }, { "epoch": 0.20706533304958502, "grad_norm": 15.203125, "learning_rate": 9.948621375781695e-06, "loss": 1.3659, "step": 1946 }, { "epoch": 0.20717173866780167, "grad_norm": 17.890625, "learning_rate": 9.948585844229677e-06, "loss": 1.2157, "step": 1947 }, { "epoch": 0.2072781442860183, "grad_norm": 10.7734375, "learning_rate": 9.948550312677659e-06, "loss": 0.9968, "step": 1948 }, { "epoch": 0.20738454990423494, "grad_norm": 17.203125, "learning_rate": 9.94851478112564e-06, "loss": 1.2738, "step": 1949 }, { "epoch": 0.2074909555224516, "grad_norm": 14.71875, "learning_rate": 9.948479249573622e-06, "loss": 1.4353, "step": 1950 }, { "epoch": 0.20759736114066824, "grad_norm": 15.140625, "learning_rate": 9.948443718021603e-06, "loss": 1.1198, "step": 1951 }, { "epoch": 0.20770376675888486, "grad_norm": 17.875, "learning_rate": 9.948408186469585e-06, "loss": 1.2474, "step": 1952 }, { "epoch": 0.2078101723771015, "grad_norm": 10.71875, "learning_rate": 9.948372654917567e-06, "loss": 1.0089, "step": 1953 }, { "epoch": 0.20791657799531815, "grad_norm": 17.984375, "learning_rate": 9.948337123365548e-06, "loss": 1.4219, "step": 1954 }, { "epoch": 0.2080229836135348, "grad_norm": 15.8984375, "learning_rate": 9.948301591813532e-06, "loss": 1.369, "step": 1955 }, { "epoch": 0.20812938923175145, "grad_norm": 15.609375, "learning_rate": 9.948266060261513e-06, "loss": 1.0383, "step": 1956 }, { "epoch": 0.20823579484996807, "grad_norm": 19.03125, "learning_rate": 9.948230528709495e-06, "loss": 1.1492, "step": 1957 }, { "epoch": 0.20834220046818472, "grad_norm": 16.40625, "learning_rate": 9.948194997157476e-06, "loss": 0.9104, "step": 1958 }, { "epoch": 0.20844860608640137, "grad_norm": 14.6953125, "learning_rate": 9.948159465605458e-06, "loss": 1.2684, "step": 1959 }, { "epoch": 0.20855501170461802, "grad_norm": 12.3515625, "learning_rate": 9.948123934053441e-06, "loss": 1.019, "step": 1960 }, { "epoch": 0.20866141732283464, "grad_norm": 19.328125, "learning_rate": 9.948088402501423e-06, "loss": 1.2948, "step": 1961 }, { "epoch": 0.20876782294105128, "grad_norm": 13.8828125, "learning_rate": 9.948052870949405e-06, "loss": 1.2311, "step": 1962 }, { "epoch": 0.20887422855926793, "grad_norm": 16.578125, "learning_rate": 9.948017339397386e-06, "loss": 1.6882, "step": 1963 }, { "epoch": 0.20898063417748458, "grad_norm": 29.40625, "learning_rate": 9.947981807845368e-06, "loss": 1.7785, "step": 1964 }, { "epoch": 0.2090870397957012, "grad_norm": 43.09375, "learning_rate": 9.94794627629335e-06, "loss": 1.2037, "step": 1965 }, { "epoch": 0.20919344541391785, "grad_norm": 16.671875, "learning_rate": 9.947910744741331e-06, "loss": 1.3539, "step": 1966 }, { "epoch": 0.2092998510321345, "grad_norm": 18.875, "learning_rate": 9.947875213189313e-06, "loss": 1.4276, "step": 1967 }, { "epoch": 0.20940625665035115, "grad_norm": 12.1484375, "learning_rate": 9.947839681637294e-06, "loss": 1.2105, "step": 1968 }, { "epoch": 0.20951266226856777, "grad_norm": 13.3828125, "learning_rate": 9.947804150085276e-06, "loss": 1.1677, "step": 1969 }, { "epoch": 0.20961906788678442, "grad_norm": 23.203125, "learning_rate": 9.947768618533257e-06, "loss": 1.0505, "step": 1970 }, { "epoch": 0.20972547350500106, "grad_norm": 15.3984375, "learning_rate": 9.94773308698124e-06, "loss": 1.1915, "step": 1971 }, { "epoch": 0.2098318791232177, "grad_norm": 19.15625, "learning_rate": 9.947697555429222e-06, "loss": 1.3863, "step": 1972 }, { "epoch": 0.20993828474143436, "grad_norm": 13.5625, "learning_rate": 9.947662023877204e-06, "loss": 1.4054, "step": 1973 }, { "epoch": 0.21004469035965098, "grad_norm": 16.71875, "learning_rate": 9.947626492325185e-06, "loss": 1.3402, "step": 1974 }, { "epoch": 0.21015109597786763, "grad_norm": 15.453125, "learning_rate": 9.947590960773167e-06, "loss": 1.2163, "step": 1975 }, { "epoch": 0.21025750159608428, "grad_norm": 17.59375, "learning_rate": 9.94755542922115e-06, "loss": 1.3366, "step": 1976 }, { "epoch": 0.21036390721430093, "grad_norm": 21.671875, "learning_rate": 9.947519897669132e-06, "loss": 1.205, "step": 1977 }, { "epoch": 0.21047031283251755, "grad_norm": 13.5234375, "learning_rate": 9.947484366117112e-06, "loss": 1.6042, "step": 1978 }, { "epoch": 0.2105767184507342, "grad_norm": 17.953125, "learning_rate": 9.947448834565094e-06, "loss": 1.2617, "step": 1979 }, { "epoch": 0.21068312406895084, "grad_norm": 22.859375, "learning_rate": 9.947413303013075e-06, "loss": 1.0208, "step": 1980 }, { "epoch": 0.2107895296871675, "grad_norm": 16.84375, "learning_rate": 9.947377771461058e-06, "loss": 1.2743, "step": 1981 }, { "epoch": 0.2108959353053841, "grad_norm": 21.296875, "learning_rate": 9.94734223990904e-06, "loss": 1.7003, "step": 1982 }, { "epoch": 0.21100234092360076, "grad_norm": 23.125, "learning_rate": 9.947306708357022e-06, "loss": 1.1939, "step": 1983 }, { "epoch": 0.2111087465418174, "grad_norm": 18.609375, "learning_rate": 9.947271176805003e-06, "loss": 1.3644, "step": 1984 }, { "epoch": 0.21121515216003406, "grad_norm": 21.90625, "learning_rate": 9.947235645252985e-06, "loss": 1.417, "step": 1985 }, { "epoch": 0.21132155777825068, "grad_norm": 16.546875, "learning_rate": 9.947200113700968e-06, "loss": 1.1426, "step": 1986 }, { "epoch": 0.21142796339646733, "grad_norm": 14.171875, "learning_rate": 9.94716458214895e-06, "loss": 1.2075, "step": 1987 }, { "epoch": 0.21153436901468398, "grad_norm": 14.75, "learning_rate": 9.947129050596931e-06, "loss": 1.1989, "step": 1988 }, { "epoch": 0.21164077463290062, "grad_norm": 12.3515625, "learning_rate": 9.947093519044913e-06, "loss": 1.0778, "step": 1989 }, { "epoch": 0.21174718025111725, "grad_norm": 19.359375, "learning_rate": 9.947057987492895e-06, "loss": 1.6149, "step": 1990 }, { "epoch": 0.2118535858693339, "grad_norm": 22.359375, "learning_rate": 9.947022455940876e-06, "loss": 1.1337, "step": 1991 }, { "epoch": 0.21195999148755054, "grad_norm": 14.0625, "learning_rate": 9.946986924388858e-06, "loss": 1.0888, "step": 1992 }, { "epoch": 0.2120663971057672, "grad_norm": 24.890625, "learning_rate": 9.94695139283684e-06, "loss": 1.4919, "step": 1993 }, { "epoch": 0.21217280272398384, "grad_norm": 14.171875, "learning_rate": 9.946915861284821e-06, "loss": 1.3231, "step": 1994 }, { "epoch": 0.21227920834220046, "grad_norm": 27.21875, "learning_rate": 9.946880329732803e-06, "loss": 1.5002, "step": 1995 }, { "epoch": 0.2123856139604171, "grad_norm": 25.59375, "learning_rate": 9.946844798180786e-06, "loss": 1.245, "step": 1996 }, { "epoch": 0.21249201957863376, "grad_norm": 17.703125, "learning_rate": 9.946809266628768e-06, "loss": 1.486, "step": 1997 }, { "epoch": 0.2125984251968504, "grad_norm": 16.8125, "learning_rate": 9.946773735076749e-06, "loss": 1.3753, "step": 1998 }, { "epoch": 0.21270483081506703, "grad_norm": 17.578125, "learning_rate": 9.94673820352473e-06, "loss": 1.2459, "step": 1999 }, { "epoch": 0.21281123643328367, "grad_norm": 20.953125, "learning_rate": 9.946702671972712e-06, "loss": 1.7269, "step": 2000 }, { "epoch": 0.21291764205150032, "grad_norm": 14.7578125, "learning_rate": 9.946667140420696e-06, "loss": 1.2245, "step": 2001 }, { "epoch": 0.21302404766971697, "grad_norm": 19.75, "learning_rate": 9.946631608868677e-06, "loss": 1.2183, "step": 2002 }, { "epoch": 0.2131304532879336, "grad_norm": 20.640625, "learning_rate": 9.946596077316659e-06, "loss": 1.3578, "step": 2003 }, { "epoch": 0.21323685890615024, "grad_norm": 15.9453125, "learning_rate": 9.94656054576464e-06, "loss": 1.2198, "step": 2004 }, { "epoch": 0.2133432645243669, "grad_norm": 17.1875, "learning_rate": 9.946525014212622e-06, "loss": 1.1216, "step": 2005 }, { "epoch": 0.21344967014258354, "grad_norm": 26.984375, "learning_rate": 9.946489482660604e-06, "loss": 1.239, "step": 2006 }, { "epoch": 0.21355607576080016, "grad_norm": 30.75, "learning_rate": 9.946453951108585e-06, "loss": 1.518, "step": 2007 }, { "epoch": 0.2136624813790168, "grad_norm": 16.515625, "learning_rate": 9.946418419556567e-06, "loss": 1.1466, "step": 2008 }, { "epoch": 0.21376888699723345, "grad_norm": 13.5, "learning_rate": 9.946382888004549e-06, "loss": 1.0292, "step": 2009 }, { "epoch": 0.2138752926154501, "grad_norm": 16.21875, "learning_rate": 9.94634735645253e-06, "loss": 1.3914, "step": 2010 }, { "epoch": 0.21398169823366675, "grad_norm": 11.453125, "learning_rate": 9.946311824900512e-06, "loss": 1.23, "step": 2011 }, { "epoch": 0.21408810385188337, "grad_norm": 20.453125, "learning_rate": 9.946276293348495e-06, "loss": 1.1749, "step": 2012 }, { "epoch": 0.21419450947010002, "grad_norm": 15.625, "learning_rate": 9.946240761796477e-06, "loss": 0.9947, "step": 2013 }, { "epoch": 0.21430091508831667, "grad_norm": 15.5625, "learning_rate": 9.946205230244458e-06, "loss": 1.5828, "step": 2014 }, { "epoch": 0.21440732070653332, "grad_norm": 19.484375, "learning_rate": 9.94616969869244e-06, "loss": 1.3448, "step": 2015 }, { "epoch": 0.21451372632474994, "grad_norm": 17.109375, "learning_rate": 9.946134167140421e-06, "loss": 1.3189, "step": 2016 }, { "epoch": 0.21462013194296659, "grad_norm": 17.125, "learning_rate": 9.946098635588403e-06, "loss": 1.3416, "step": 2017 }, { "epoch": 0.21472653756118323, "grad_norm": 21.46875, "learning_rate": 9.946063104036385e-06, "loss": 1.585, "step": 2018 }, { "epoch": 0.21483294317939988, "grad_norm": 16.78125, "learning_rate": 9.946027572484366e-06, "loss": 1.2204, "step": 2019 }, { "epoch": 0.2149393487976165, "grad_norm": 17.34375, "learning_rate": 9.945992040932348e-06, "loss": 1.1825, "step": 2020 }, { "epoch": 0.21504575441583315, "grad_norm": 15.515625, "learning_rate": 9.94595650938033e-06, "loss": 1.2594, "step": 2021 }, { "epoch": 0.2151521600340498, "grad_norm": 14.5234375, "learning_rate": 9.945920977828313e-06, "loss": 1.1206, "step": 2022 }, { "epoch": 0.21525856565226645, "grad_norm": 25.453125, "learning_rate": 9.945885446276294e-06, "loss": 1.3748, "step": 2023 }, { "epoch": 0.21536497127048307, "grad_norm": 15.0, "learning_rate": 9.945849914724276e-06, "loss": 1.2228, "step": 2024 }, { "epoch": 0.21547137688869972, "grad_norm": 12.6328125, "learning_rate": 9.945814383172258e-06, "loss": 1.1516, "step": 2025 }, { "epoch": 0.21557778250691637, "grad_norm": 18.53125, "learning_rate": 9.94577885162024e-06, "loss": 1.4222, "step": 2026 }, { "epoch": 0.21568418812513301, "grad_norm": 14.515625, "learning_rate": 9.945743320068223e-06, "loss": 1.1449, "step": 2027 }, { "epoch": 0.21579059374334966, "grad_norm": 17.140625, "learning_rate": 9.945707788516204e-06, "loss": 1.4777, "step": 2028 }, { "epoch": 0.21589699936156628, "grad_norm": 16.421875, "learning_rate": 9.945672256964186e-06, "loss": 1.5043, "step": 2029 }, { "epoch": 0.21600340497978293, "grad_norm": 19.25, "learning_rate": 9.945636725412167e-06, "loss": 1.4015, "step": 2030 }, { "epoch": 0.21610981059799958, "grad_norm": 20.140625, "learning_rate": 9.945601193860149e-06, "loss": 1.4401, "step": 2031 }, { "epoch": 0.21621621621621623, "grad_norm": 16.96875, "learning_rate": 9.94556566230813e-06, "loss": 1.3281, "step": 2032 }, { "epoch": 0.21632262183443285, "grad_norm": 19.6875, "learning_rate": 9.945530130756112e-06, "loss": 1.2601, "step": 2033 }, { "epoch": 0.2164290274526495, "grad_norm": 13.71875, "learning_rate": 9.945494599204094e-06, "loss": 0.961, "step": 2034 }, { "epoch": 0.21653543307086615, "grad_norm": 17.015625, "learning_rate": 9.945459067652075e-06, "loss": 1.1342, "step": 2035 }, { "epoch": 0.2166418386890828, "grad_norm": 13.7421875, "learning_rate": 9.945423536100057e-06, "loss": 1.1138, "step": 2036 }, { "epoch": 0.21674824430729941, "grad_norm": 22.734375, "learning_rate": 9.94538800454804e-06, "loss": 1.5471, "step": 2037 }, { "epoch": 0.21685464992551606, "grad_norm": 19.703125, "learning_rate": 9.945352472996022e-06, "loss": 1.1966, "step": 2038 }, { "epoch": 0.2169610555437327, "grad_norm": 25.25, "learning_rate": 9.945316941444003e-06, "loss": 1.2865, "step": 2039 }, { "epoch": 0.21706746116194936, "grad_norm": 21.21875, "learning_rate": 9.945281409891985e-06, "loss": 1.4249, "step": 2040 }, { "epoch": 0.21717386678016598, "grad_norm": 22.1875, "learning_rate": 9.945245878339967e-06, "loss": 1.8699, "step": 2041 }, { "epoch": 0.21728027239838263, "grad_norm": 23.4375, "learning_rate": 9.945210346787948e-06, "loss": 1.413, "step": 2042 }, { "epoch": 0.21738667801659928, "grad_norm": 19.453125, "learning_rate": 9.94517481523593e-06, "loss": 1.2103, "step": 2043 }, { "epoch": 0.21749308363481593, "grad_norm": 21.515625, "learning_rate": 9.945139283683912e-06, "loss": 1.2576, "step": 2044 }, { "epoch": 0.21759948925303255, "grad_norm": 32.125, "learning_rate": 9.945103752131893e-06, "loss": 1.27, "step": 2045 }, { "epoch": 0.2177058948712492, "grad_norm": 14.59375, "learning_rate": 9.945068220579875e-06, "loss": 1.4071, "step": 2046 }, { "epoch": 0.21781230048946584, "grad_norm": 18.375, "learning_rate": 9.945032689027856e-06, "loss": 1.5605, "step": 2047 }, { "epoch": 0.2179187061076825, "grad_norm": 13.9921875, "learning_rate": 9.94499715747584e-06, "loss": 1.316, "step": 2048 }, { "epoch": 0.21802511172589914, "grad_norm": 24.140625, "learning_rate": 9.944961625923821e-06, "loss": 1.1464, "step": 2049 }, { "epoch": 0.21813151734411576, "grad_norm": 13.9921875, "learning_rate": 9.944926094371803e-06, "loss": 1.4145, "step": 2050 }, { "epoch": 0.2182379229623324, "grad_norm": 20.59375, "learning_rate": 9.944890562819784e-06, "loss": 1.3428, "step": 2051 }, { "epoch": 0.21834432858054906, "grad_norm": 17.5, "learning_rate": 9.944855031267766e-06, "loss": 1.3966, "step": 2052 }, { "epoch": 0.2184507341987657, "grad_norm": 18.375, "learning_rate": 9.94481949971575e-06, "loss": 1.4482, "step": 2053 }, { "epoch": 0.21855713981698233, "grad_norm": 12.5703125, "learning_rate": 9.944783968163731e-06, "loss": 1.2009, "step": 2054 }, { "epoch": 0.21866354543519897, "grad_norm": 21.1875, "learning_rate": 9.944748436611713e-06, "loss": 1.2543, "step": 2055 }, { "epoch": 0.21876995105341562, "grad_norm": 15.90625, "learning_rate": 9.944712905059694e-06, "loss": 1.1816, "step": 2056 }, { "epoch": 0.21887635667163227, "grad_norm": 18.9375, "learning_rate": 9.944677373507676e-06, "loss": 1.3027, "step": 2057 }, { "epoch": 0.2189827622898489, "grad_norm": 25.796875, "learning_rate": 9.944641841955657e-06, "loss": 1.3339, "step": 2058 }, { "epoch": 0.21908916790806554, "grad_norm": 16.71875, "learning_rate": 9.944606310403639e-06, "loss": 1.3473, "step": 2059 }, { "epoch": 0.2191955735262822, "grad_norm": 36.125, "learning_rate": 9.94457077885162e-06, "loss": 1.5385, "step": 2060 }, { "epoch": 0.21930197914449884, "grad_norm": 16.703125, "learning_rate": 9.944535247299602e-06, "loss": 1.5115, "step": 2061 }, { "epoch": 0.21940838476271546, "grad_norm": 15.1328125, "learning_rate": 9.944499715747584e-06, "loss": 1.2903, "step": 2062 }, { "epoch": 0.2195147903809321, "grad_norm": 14.140625, "learning_rate": 9.944464184195567e-06, "loss": 1.2902, "step": 2063 }, { "epoch": 0.21962119599914876, "grad_norm": 15.09375, "learning_rate": 9.944428652643549e-06, "loss": 1.1767, "step": 2064 }, { "epoch": 0.2197276016173654, "grad_norm": 18.484375, "learning_rate": 9.94439312109153e-06, "loss": 1.5435, "step": 2065 }, { "epoch": 0.21983400723558205, "grad_norm": 19.171875, "learning_rate": 9.944357589539512e-06, "loss": 1.4917, "step": 2066 }, { "epoch": 0.21994041285379867, "grad_norm": 30.703125, "learning_rate": 9.944322057987494e-06, "loss": 1.3302, "step": 2067 }, { "epoch": 0.22004681847201532, "grad_norm": 16.625, "learning_rate": 9.944286526435475e-06, "loss": 1.5811, "step": 2068 }, { "epoch": 0.22015322409023197, "grad_norm": 20.203125, "learning_rate": 9.944250994883457e-06, "loss": 1.279, "step": 2069 }, { "epoch": 0.22025962970844862, "grad_norm": 16.0625, "learning_rate": 9.944215463331438e-06, "loss": 1.2271, "step": 2070 }, { "epoch": 0.22036603532666524, "grad_norm": 14.953125, "learning_rate": 9.94417993177942e-06, "loss": 1.2107, "step": 2071 }, { "epoch": 0.2204724409448819, "grad_norm": 18.96875, "learning_rate": 9.944144400227402e-06, "loss": 1.4856, "step": 2072 }, { "epoch": 0.22057884656309854, "grad_norm": 15.3515625, "learning_rate": 9.944108868675385e-06, "loss": 1.3019, "step": 2073 }, { "epoch": 0.22068525218131518, "grad_norm": 14.3046875, "learning_rate": 9.944073337123366e-06, "loss": 1.085, "step": 2074 }, { "epoch": 0.2207916577995318, "grad_norm": 30.515625, "learning_rate": 9.944037805571348e-06, "loss": 1.4047, "step": 2075 }, { "epoch": 0.22089806341774845, "grad_norm": 17.78125, "learning_rate": 9.94400227401933e-06, "loss": 1.0863, "step": 2076 }, { "epoch": 0.2210044690359651, "grad_norm": 16.140625, "learning_rate": 9.943966742467311e-06, "loss": 1.575, "step": 2077 }, { "epoch": 0.22111087465418175, "grad_norm": 19.78125, "learning_rate": 9.943931210915295e-06, "loss": 1.443, "step": 2078 }, { "epoch": 0.22121728027239837, "grad_norm": 14.640625, "learning_rate": 9.943895679363276e-06, "loss": 1.2421, "step": 2079 }, { "epoch": 0.22132368589061502, "grad_norm": 14.359375, "learning_rate": 9.943860147811258e-06, "loss": 1.3244, "step": 2080 }, { "epoch": 0.22143009150883167, "grad_norm": 15.5859375, "learning_rate": 9.94382461625924e-06, "loss": 1.18, "step": 2081 }, { "epoch": 0.22153649712704832, "grad_norm": 15.5546875, "learning_rate": 9.943789084707221e-06, "loss": 1.1162, "step": 2082 }, { "epoch": 0.22164290274526496, "grad_norm": 24.390625, "learning_rate": 9.943753553155203e-06, "loss": 1.4224, "step": 2083 }, { "epoch": 0.22174930836348158, "grad_norm": 15.796875, "learning_rate": 9.943718021603184e-06, "loss": 1.4163, "step": 2084 }, { "epoch": 0.22185571398169823, "grad_norm": 11.671875, "learning_rate": 9.943682490051166e-06, "loss": 1.1218, "step": 2085 }, { "epoch": 0.22196211959991488, "grad_norm": 19.015625, "learning_rate": 9.943646958499147e-06, "loss": 1.6482, "step": 2086 }, { "epoch": 0.22206852521813153, "grad_norm": 14.7109375, "learning_rate": 9.943611426947129e-06, "loss": 1.3134, "step": 2087 }, { "epoch": 0.22217493083634815, "grad_norm": 13.5546875, "learning_rate": 9.94357589539511e-06, "loss": 1.2683, "step": 2088 }, { "epoch": 0.2222813364545648, "grad_norm": 14.4140625, "learning_rate": 9.943540363843094e-06, "loss": 1.1279, "step": 2089 }, { "epoch": 0.22238774207278145, "grad_norm": 17.046875, "learning_rate": 9.943504832291076e-06, "loss": 1.577, "step": 2090 }, { "epoch": 0.2224941476909981, "grad_norm": 14.4296875, "learning_rate": 9.943469300739057e-06, "loss": 1.2163, "step": 2091 }, { "epoch": 0.22260055330921472, "grad_norm": 17.015625, "learning_rate": 9.943433769187039e-06, "loss": 1.4734, "step": 2092 }, { "epoch": 0.22270695892743136, "grad_norm": 14.0234375, "learning_rate": 9.94339823763502e-06, "loss": 1.0667, "step": 2093 }, { "epoch": 0.222813364545648, "grad_norm": 19.3125, "learning_rate": 9.943362706083004e-06, "loss": 1.4544, "step": 2094 }, { "epoch": 0.22291977016386466, "grad_norm": 14.265625, "learning_rate": 9.943327174530985e-06, "loss": 1.1171, "step": 2095 }, { "epoch": 0.22302617578208128, "grad_norm": 14.4375, "learning_rate": 9.943291642978967e-06, "loss": 1.3758, "step": 2096 }, { "epoch": 0.22313258140029793, "grad_norm": 13.890625, "learning_rate": 9.943256111426949e-06, "loss": 1.2871, "step": 2097 }, { "epoch": 0.22323898701851458, "grad_norm": 17.203125, "learning_rate": 9.943220579874928e-06, "loss": 1.4755, "step": 2098 }, { "epoch": 0.22334539263673123, "grad_norm": 14.15625, "learning_rate": 9.943185048322912e-06, "loss": 1.1938, "step": 2099 }, { "epoch": 0.22345179825494785, "grad_norm": 14.7890625, "learning_rate": 9.943149516770893e-06, "loss": 1.1282, "step": 2100 }, { "epoch": 0.2235582038731645, "grad_norm": 13.7734375, "learning_rate": 9.943113985218875e-06, "loss": 1.2347, "step": 2101 }, { "epoch": 0.22366460949138114, "grad_norm": 13.7265625, "learning_rate": 9.943078453666857e-06, "loss": 1.16, "step": 2102 }, { "epoch": 0.2237710151095978, "grad_norm": 13.265625, "learning_rate": 9.943042922114838e-06, "loss": 1.0258, "step": 2103 }, { "epoch": 0.22387742072781444, "grad_norm": 28.21875, "learning_rate": 9.943007390562821e-06, "loss": 1.2527, "step": 2104 }, { "epoch": 0.22398382634603106, "grad_norm": 15.296875, "learning_rate": 9.942971859010803e-06, "loss": 1.3241, "step": 2105 }, { "epoch": 0.2240902319642477, "grad_norm": 14.6796875, "learning_rate": 9.942936327458785e-06, "loss": 1.1956, "step": 2106 }, { "epoch": 0.22419663758246436, "grad_norm": 12.9765625, "learning_rate": 9.942900795906766e-06, "loss": 1.1142, "step": 2107 }, { "epoch": 0.224303043200681, "grad_norm": 15.3046875, "learning_rate": 9.942865264354748e-06, "loss": 1.1424, "step": 2108 }, { "epoch": 0.22440944881889763, "grad_norm": 16.859375, "learning_rate": 9.94282973280273e-06, "loss": 1.0589, "step": 2109 }, { "epoch": 0.22451585443711428, "grad_norm": 20.6875, "learning_rate": 9.942794201250711e-06, "loss": 1.6221, "step": 2110 }, { "epoch": 0.22462226005533092, "grad_norm": 11.4765625, "learning_rate": 9.942758669698693e-06, "loss": 0.9399, "step": 2111 }, { "epoch": 0.22472866567354757, "grad_norm": 13.7890625, "learning_rate": 9.942723138146674e-06, "loss": 1.4205, "step": 2112 }, { "epoch": 0.2248350712917642, "grad_norm": 21.609375, "learning_rate": 9.942687606594656e-06, "loss": 1.4972, "step": 2113 }, { "epoch": 0.22494147690998084, "grad_norm": 14.0703125, "learning_rate": 9.94265207504264e-06, "loss": 1.2194, "step": 2114 }, { "epoch": 0.2250478825281975, "grad_norm": 15.0234375, "learning_rate": 9.94261654349062e-06, "loss": 1.0097, "step": 2115 }, { "epoch": 0.22515428814641414, "grad_norm": 18.5, "learning_rate": 9.942581011938602e-06, "loss": 1.4059, "step": 2116 }, { "epoch": 0.22526069376463076, "grad_norm": 69.75, "learning_rate": 9.942545480386584e-06, "loss": 1.6038, "step": 2117 }, { "epoch": 0.2253670993828474, "grad_norm": 17.6875, "learning_rate": 9.942509948834566e-06, "loss": 1.0758, "step": 2118 }, { "epoch": 0.22547350500106406, "grad_norm": 40.53125, "learning_rate": 9.942474417282549e-06, "loss": 1.6243, "step": 2119 }, { "epoch": 0.2255799106192807, "grad_norm": 13.2578125, "learning_rate": 9.94243888573053e-06, "loss": 1.3153, "step": 2120 }, { "epoch": 0.22568631623749735, "grad_norm": 12.734375, "learning_rate": 9.942403354178512e-06, "loss": 1.1348, "step": 2121 }, { "epoch": 0.22579272185571397, "grad_norm": 18.890625, "learning_rate": 9.942367822626494e-06, "loss": 1.931, "step": 2122 }, { "epoch": 0.22589912747393062, "grad_norm": 12.6328125, "learning_rate": 9.942332291074475e-06, "loss": 1.2276, "step": 2123 }, { "epoch": 0.22600553309214727, "grad_norm": 17.75, "learning_rate": 9.942296759522457e-06, "loss": 1.1723, "step": 2124 }, { "epoch": 0.22611193871036392, "grad_norm": 15.7421875, "learning_rate": 9.942261227970439e-06, "loss": 1.6048, "step": 2125 }, { "epoch": 0.22621834432858054, "grad_norm": 16.703125, "learning_rate": 9.94222569641842e-06, "loss": 1.3827, "step": 2126 }, { "epoch": 0.2263247499467972, "grad_norm": 19.1875, "learning_rate": 9.942190164866402e-06, "loss": 1.5291, "step": 2127 }, { "epoch": 0.22643115556501384, "grad_norm": 16.53125, "learning_rate": 9.942154633314383e-06, "loss": 1.2979, "step": 2128 }, { "epoch": 0.22653756118323048, "grad_norm": 19.359375, "learning_rate": 9.942119101762365e-06, "loss": 1.2386, "step": 2129 }, { "epoch": 0.2266439668014471, "grad_norm": 25.3125, "learning_rate": 9.942083570210348e-06, "loss": 1.1731, "step": 2130 }, { "epoch": 0.22675037241966375, "grad_norm": 14.1484375, "learning_rate": 9.94204803865833e-06, "loss": 1.2182, "step": 2131 }, { "epoch": 0.2268567780378804, "grad_norm": 15.9140625, "learning_rate": 9.942012507106312e-06, "loss": 1.2521, "step": 2132 }, { "epoch": 0.22696318365609705, "grad_norm": 13.640625, "learning_rate": 9.941976975554293e-06, "loss": 1.1, "step": 2133 }, { "epoch": 0.22706958927431367, "grad_norm": 22.546875, "learning_rate": 9.941941444002275e-06, "loss": 1.5637, "step": 2134 }, { "epoch": 0.22717599489253032, "grad_norm": 16.734375, "learning_rate": 9.941905912450256e-06, "loss": 1.2944, "step": 2135 }, { "epoch": 0.22728240051074697, "grad_norm": 15.7734375, "learning_rate": 9.941870380898238e-06, "loss": 1.5917, "step": 2136 }, { "epoch": 0.22738880612896362, "grad_norm": 17.8125, "learning_rate": 9.94183484934622e-06, "loss": 1.2646, "step": 2137 }, { "epoch": 0.22749521174718026, "grad_norm": 14.421875, "learning_rate": 9.941799317794201e-06, "loss": 1.2554, "step": 2138 }, { "epoch": 0.22760161736539689, "grad_norm": 20.359375, "learning_rate": 9.941763786242183e-06, "loss": 1.5948, "step": 2139 }, { "epoch": 0.22770802298361353, "grad_norm": 16.828125, "learning_rate": 9.941728254690166e-06, "loss": 1.0611, "step": 2140 }, { "epoch": 0.22781442860183018, "grad_norm": 17.828125, "learning_rate": 9.941692723138148e-06, "loss": 1.3182, "step": 2141 }, { "epoch": 0.22792083422004683, "grad_norm": 16.4375, "learning_rate": 9.94165719158613e-06, "loss": 1.4423, "step": 2142 }, { "epoch": 0.22802723983826345, "grad_norm": 19.578125, "learning_rate": 9.941621660034111e-06, "loss": 1.2492, "step": 2143 }, { "epoch": 0.2281336454564801, "grad_norm": 24.453125, "learning_rate": 9.941586128482093e-06, "loss": 1.2661, "step": 2144 }, { "epoch": 0.22824005107469675, "grad_norm": 15.4921875, "learning_rate": 9.941550596930076e-06, "loss": 1.3056, "step": 2145 }, { "epoch": 0.2283464566929134, "grad_norm": 17.125, "learning_rate": 9.941515065378057e-06, "loss": 1.5803, "step": 2146 }, { "epoch": 0.22845286231113002, "grad_norm": 16.765625, "learning_rate": 9.941479533826039e-06, "loss": 1.592, "step": 2147 }, { "epoch": 0.22855926792934667, "grad_norm": 21.0625, "learning_rate": 9.94144400227402e-06, "loss": 1.7963, "step": 2148 }, { "epoch": 0.22866567354756331, "grad_norm": 14.2421875, "learning_rate": 9.941408470722002e-06, "loss": 1.0836, "step": 2149 }, { "epoch": 0.22877207916577996, "grad_norm": 15.2421875, "learning_rate": 9.941372939169984e-06, "loss": 1.3068, "step": 2150 }, { "epoch": 0.22887848478399658, "grad_norm": 17.59375, "learning_rate": 9.941337407617965e-06, "loss": 1.2657, "step": 2151 }, { "epoch": 0.22898489040221323, "grad_norm": 18.5625, "learning_rate": 9.941301876065947e-06, "loss": 1.5505, "step": 2152 }, { "epoch": 0.22909129602042988, "grad_norm": 11.7265625, "learning_rate": 9.941266344513929e-06, "loss": 1.2687, "step": 2153 }, { "epoch": 0.22919770163864653, "grad_norm": 13.7109375, "learning_rate": 9.94123081296191e-06, "loss": 1.5198, "step": 2154 }, { "epoch": 0.22930410725686315, "grad_norm": 20.328125, "learning_rate": 9.941195281409894e-06, "loss": 1.2069, "step": 2155 }, { "epoch": 0.2294105128750798, "grad_norm": 15.90625, "learning_rate": 9.941159749857875e-06, "loss": 1.2913, "step": 2156 }, { "epoch": 0.22951691849329645, "grad_norm": 14.1171875, "learning_rate": 9.941124218305857e-06, "loss": 1.393, "step": 2157 }, { "epoch": 0.2296233241115131, "grad_norm": 13.1953125, "learning_rate": 9.941088686753838e-06, "loss": 1.1478, "step": 2158 }, { "epoch": 0.22972972972972974, "grad_norm": 18.1875, "learning_rate": 9.94105315520182e-06, "loss": 1.3021, "step": 2159 }, { "epoch": 0.22983613534794636, "grad_norm": 13.453125, "learning_rate": 9.941017623649802e-06, "loss": 1.1148, "step": 2160 }, { "epoch": 0.229942540966163, "grad_norm": 17.375, "learning_rate": 9.940982092097783e-06, "loss": 1.2681, "step": 2161 }, { "epoch": 0.23004894658437966, "grad_norm": 16.828125, "learning_rate": 9.940946560545765e-06, "loss": 1.4799, "step": 2162 }, { "epoch": 0.2301553522025963, "grad_norm": 16.4375, "learning_rate": 9.940911028993746e-06, "loss": 1.378, "step": 2163 }, { "epoch": 0.23026175782081293, "grad_norm": 31.28125, "learning_rate": 9.940875497441728e-06, "loss": 1.536, "step": 2164 }, { "epoch": 0.23036816343902958, "grad_norm": 15.96875, "learning_rate": 9.94083996588971e-06, "loss": 1.2197, "step": 2165 }, { "epoch": 0.23047456905724623, "grad_norm": 15.3984375, "learning_rate": 9.940804434337693e-06, "loss": 1.1273, "step": 2166 }, { "epoch": 0.23058097467546287, "grad_norm": 16.546875, "learning_rate": 9.940768902785675e-06, "loss": 1.3137, "step": 2167 }, { "epoch": 0.2306873802936795, "grad_norm": 21.109375, "learning_rate": 9.940733371233656e-06, "loss": 1.6588, "step": 2168 }, { "epoch": 0.23079378591189614, "grad_norm": 13.8203125, "learning_rate": 9.940697839681638e-06, "loss": 1.2481, "step": 2169 }, { "epoch": 0.2309001915301128, "grad_norm": 12.453125, "learning_rate": 9.94066230812962e-06, "loss": 1.1214, "step": 2170 }, { "epoch": 0.23100659714832944, "grad_norm": 15.5390625, "learning_rate": 9.940626776577603e-06, "loss": 1.471, "step": 2171 }, { "epoch": 0.23111300276654606, "grad_norm": 21.265625, "learning_rate": 9.940591245025584e-06, "loss": 1.2842, "step": 2172 }, { "epoch": 0.2312194083847627, "grad_norm": 14.671875, "learning_rate": 9.940555713473566e-06, "loss": 1.605, "step": 2173 }, { "epoch": 0.23132581400297936, "grad_norm": 14.71875, "learning_rate": 9.940520181921547e-06, "loss": 1.3297, "step": 2174 }, { "epoch": 0.231432219621196, "grad_norm": 20.265625, "learning_rate": 9.940484650369529e-06, "loss": 1.283, "step": 2175 }, { "epoch": 0.23153862523941265, "grad_norm": 18.046875, "learning_rate": 9.94044911881751e-06, "loss": 1.4273, "step": 2176 }, { "epoch": 0.23164503085762927, "grad_norm": 17.3125, "learning_rate": 9.940413587265492e-06, "loss": 1.3878, "step": 2177 }, { "epoch": 0.23175143647584592, "grad_norm": 18.765625, "learning_rate": 9.940378055713474e-06, "loss": 1.3886, "step": 2178 }, { "epoch": 0.23185784209406257, "grad_norm": 15.6875, "learning_rate": 9.940342524161456e-06, "loss": 1.4657, "step": 2179 }, { "epoch": 0.23196424771227922, "grad_norm": 12.2265625, "learning_rate": 9.940306992609437e-06, "loss": 0.9589, "step": 2180 }, { "epoch": 0.23207065333049584, "grad_norm": 17.140625, "learning_rate": 9.94027146105742e-06, "loss": 1.3126, "step": 2181 }, { "epoch": 0.2321770589487125, "grad_norm": 13.2265625, "learning_rate": 9.940235929505402e-06, "loss": 1.3063, "step": 2182 }, { "epoch": 0.23228346456692914, "grad_norm": 14.296875, "learning_rate": 9.940200397953384e-06, "loss": 1.1912, "step": 2183 }, { "epoch": 0.23238987018514579, "grad_norm": 14.390625, "learning_rate": 9.940164866401365e-06, "loss": 1.1169, "step": 2184 }, { "epoch": 0.2324962758033624, "grad_norm": 18.125, "learning_rate": 9.940129334849347e-06, "loss": 1.7293, "step": 2185 }, { "epoch": 0.23260268142157906, "grad_norm": 9.6171875, "learning_rate": 9.94009380329733e-06, "loss": 0.9449, "step": 2186 }, { "epoch": 0.2327090870397957, "grad_norm": 16.484375, "learning_rate": 9.94005827174531e-06, "loss": 1.2749, "step": 2187 }, { "epoch": 0.23281549265801235, "grad_norm": 14.3125, "learning_rate": 9.940022740193292e-06, "loss": 1.4348, "step": 2188 }, { "epoch": 0.23292189827622897, "grad_norm": 14.5234375, "learning_rate": 9.939987208641273e-06, "loss": 1.4052, "step": 2189 }, { "epoch": 0.23302830389444562, "grad_norm": 9.796875, "learning_rate": 9.939951677089255e-06, "loss": 0.8364, "step": 2190 }, { "epoch": 0.23313470951266227, "grad_norm": 17.140625, "learning_rate": 9.939916145537238e-06, "loss": 1.5041, "step": 2191 }, { "epoch": 0.23324111513087892, "grad_norm": 15.9765625, "learning_rate": 9.93988061398522e-06, "loss": 1.1374, "step": 2192 }, { "epoch": 0.23334752074909557, "grad_norm": 15.34375, "learning_rate": 9.939845082433201e-06, "loss": 1.0176, "step": 2193 }, { "epoch": 0.2334539263673122, "grad_norm": 12.8984375, "learning_rate": 9.939809550881183e-06, "loss": 1.0272, "step": 2194 }, { "epoch": 0.23356033198552884, "grad_norm": 13.125, "learning_rate": 9.939774019329165e-06, "loss": 1.1458, "step": 2195 }, { "epoch": 0.23366673760374548, "grad_norm": 14.03125, "learning_rate": 9.939738487777148e-06, "loss": 1.1919, "step": 2196 }, { "epoch": 0.23377314322196213, "grad_norm": 19.625, "learning_rate": 9.93970295622513e-06, "loss": 1.5603, "step": 2197 }, { "epoch": 0.23387954884017875, "grad_norm": 12.4453125, "learning_rate": 9.939667424673111e-06, "loss": 1.2162, "step": 2198 }, { "epoch": 0.2339859544583954, "grad_norm": 12.34375, "learning_rate": 9.939631893121093e-06, "loss": 1.0674, "step": 2199 }, { "epoch": 0.23409236007661205, "grad_norm": 16.21875, "learning_rate": 9.939596361569074e-06, "loss": 1.5677, "step": 2200 }, { "epoch": 0.2341987656948287, "grad_norm": 14.40625, "learning_rate": 9.939560830017056e-06, "loss": 1.1333, "step": 2201 }, { "epoch": 0.23430517131304532, "grad_norm": 13.5390625, "learning_rate": 9.939525298465038e-06, "loss": 1.1901, "step": 2202 }, { "epoch": 0.23441157693126197, "grad_norm": 22.390625, "learning_rate": 9.93948976691302e-06, "loss": 1.1299, "step": 2203 }, { "epoch": 0.23451798254947862, "grad_norm": 24.59375, "learning_rate": 9.939454235361e-06, "loss": 1.6246, "step": 2204 }, { "epoch": 0.23462438816769526, "grad_norm": 12.8984375, "learning_rate": 9.939418703808982e-06, "loss": 1.3609, "step": 2205 }, { "epoch": 0.23473079378591188, "grad_norm": 17.234375, "learning_rate": 9.939383172256964e-06, "loss": 1.3123, "step": 2206 }, { "epoch": 0.23483719940412853, "grad_norm": 11.390625, "learning_rate": 9.939347640704947e-06, "loss": 1.0476, "step": 2207 }, { "epoch": 0.23494360502234518, "grad_norm": 14.25, "learning_rate": 9.939312109152929e-06, "loss": 1.2881, "step": 2208 }, { "epoch": 0.23505001064056183, "grad_norm": 34.90625, "learning_rate": 9.93927657760091e-06, "loss": 2.1616, "step": 2209 }, { "epoch": 0.23515641625877845, "grad_norm": 15.84375, "learning_rate": 9.939241046048892e-06, "loss": 1.2693, "step": 2210 }, { "epoch": 0.2352628218769951, "grad_norm": 16.28125, "learning_rate": 9.939205514496874e-06, "loss": 1.4034, "step": 2211 }, { "epoch": 0.23536922749521175, "grad_norm": 16.578125, "learning_rate": 9.939169982944857e-06, "loss": 1.204, "step": 2212 }, { "epoch": 0.2354756331134284, "grad_norm": 27.953125, "learning_rate": 9.939134451392839e-06, "loss": 1.5646, "step": 2213 }, { "epoch": 0.23558203873164504, "grad_norm": 18.078125, "learning_rate": 9.93909891984082e-06, "loss": 1.1852, "step": 2214 }, { "epoch": 0.23568844434986166, "grad_norm": 40.75, "learning_rate": 9.939063388288802e-06, "loss": 1.8556, "step": 2215 }, { "epoch": 0.2357948499680783, "grad_norm": 14.875, "learning_rate": 9.939027856736783e-06, "loss": 1.3709, "step": 2216 }, { "epoch": 0.23590125558629496, "grad_norm": 16.71875, "learning_rate": 9.938992325184765e-06, "loss": 1.1811, "step": 2217 }, { "epoch": 0.2360076612045116, "grad_norm": 16.328125, "learning_rate": 9.938956793632747e-06, "loss": 1.3199, "step": 2218 }, { "epoch": 0.23611406682272823, "grad_norm": 17.1875, "learning_rate": 9.938921262080728e-06, "loss": 1.611, "step": 2219 }, { "epoch": 0.23622047244094488, "grad_norm": 24.8125, "learning_rate": 9.93888573052871e-06, "loss": 1.487, "step": 2220 }, { "epoch": 0.23632687805916153, "grad_norm": 12.921875, "learning_rate": 9.938850198976691e-06, "loss": 1.4767, "step": 2221 }, { "epoch": 0.23643328367737818, "grad_norm": 20.421875, "learning_rate": 9.938814667424675e-06, "loss": 1.3782, "step": 2222 }, { "epoch": 0.2365396892955948, "grad_norm": 15.8203125, "learning_rate": 9.938779135872656e-06, "loss": 1.3934, "step": 2223 }, { "epoch": 0.23664609491381144, "grad_norm": 18.375, "learning_rate": 9.938743604320638e-06, "loss": 1.2328, "step": 2224 }, { "epoch": 0.2367525005320281, "grad_norm": 15.453125, "learning_rate": 9.93870807276862e-06, "loss": 1.1923, "step": 2225 }, { "epoch": 0.23685890615024474, "grad_norm": 25.359375, "learning_rate": 9.938672541216601e-06, "loss": 1.7786, "step": 2226 }, { "epoch": 0.23696531176846136, "grad_norm": 17.421875, "learning_rate": 9.938637009664583e-06, "loss": 1.2192, "step": 2227 }, { "epoch": 0.237071717386678, "grad_norm": 16.015625, "learning_rate": 9.938601478112564e-06, "loss": 1.2897, "step": 2228 }, { "epoch": 0.23717812300489466, "grad_norm": 24.515625, "learning_rate": 9.938565946560546e-06, "loss": 1.2075, "step": 2229 }, { "epoch": 0.2372845286231113, "grad_norm": 17.484375, "learning_rate": 9.938530415008528e-06, "loss": 1.4833, "step": 2230 }, { "epoch": 0.23739093424132796, "grad_norm": 15.40625, "learning_rate": 9.93849488345651e-06, "loss": 1.3337, "step": 2231 }, { "epoch": 0.23749733985954458, "grad_norm": 16.421875, "learning_rate": 9.938459351904493e-06, "loss": 1.1863, "step": 2232 }, { "epoch": 0.23760374547776122, "grad_norm": 17.890625, "learning_rate": 9.938423820352474e-06, "loss": 1.3173, "step": 2233 }, { "epoch": 0.23771015109597787, "grad_norm": 18.578125, "learning_rate": 9.938388288800456e-06, "loss": 1.0758, "step": 2234 }, { "epoch": 0.23781655671419452, "grad_norm": 18.5625, "learning_rate": 9.938352757248437e-06, "loss": 1.6158, "step": 2235 }, { "epoch": 0.23792296233241114, "grad_norm": 16.921875, "learning_rate": 9.938317225696419e-06, "loss": 1.5465, "step": 2236 }, { "epoch": 0.2380293679506278, "grad_norm": 19.34375, "learning_rate": 9.938281694144402e-06, "loss": 1.5367, "step": 2237 }, { "epoch": 0.23813577356884444, "grad_norm": 19.125, "learning_rate": 9.938246162592384e-06, "loss": 1.3022, "step": 2238 }, { "epoch": 0.2382421791870611, "grad_norm": 16.03125, "learning_rate": 9.938210631040365e-06, "loss": 1.3727, "step": 2239 }, { "epoch": 0.2383485848052777, "grad_norm": 12.9140625, "learning_rate": 9.938175099488347e-06, "loss": 1.1309, "step": 2240 }, { "epoch": 0.23845499042349436, "grad_norm": 21.671875, "learning_rate": 9.938139567936329e-06, "loss": 1.5269, "step": 2241 }, { "epoch": 0.238561396041711, "grad_norm": 14.9296875, "learning_rate": 9.93810403638431e-06, "loss": 1.3286, "step": 2242 }, { "epoch": 0.23866780165992765, "grad_norm": 21.171875, "learning_rate": 9.938068504832292e-06, "loss": 1.0897, "step": 2243 }, { "epoch": 0.23877420727814427, "grad_norm": 20.328125, "learning_rate": 9.938032973280274e-06, "loss": 1.3439, "step": 2244 }, { "epoch": 0.23888061289636092, "grad_norm": 20.71875, "learning_rate": 9.937997441728255e-06, "loss": 1.4837, "step": 2245 }, { "epoch": 0.23898701851457757, "grad_norm": 17.46875, "learning_rate": 9.937961910176237e-06, "loss": 1.1129, "step": 2246 }, { "epoch": 0.23909342413279422, "grad_norm": 13.484375, "learning_rate": 9.937926378624218e-06, "loss": 1.3862, "step": 2247 }, { "epoch": 0.23919982975101084, "grad_norm": 13.4921875, "learning_rate": 9.937890847072202e-06, "loss": 1.3027, "step": 2248 }, { "epoch": 0.2393062353692275, "grad_norm": 24.1875, "learning_rate": 9.937855315520183e-06, "loss": 1.5239, "step": 2249 }, { "epoch": 0.23941264098744414, "grad_norm": 15.75, "learning_rate": 9.937819783968165e-06, "loss": 1.3, "step": 2250 }, { "epoch": 0.23951904660566078, "grad_norm": 16.140625, "learning_rate": 9.937784252416146e-06, "loss": 1.5096, "step": 2251 }, { "epoch": 0.23962545222387743, "grad_norm": 12.4140625, "learning_rate": 9.937748720864128e-06, "loss": 1.1519, "step": 2252 }, { "epoch": 0.23973185784209405, "grad_norm": 19.453125, "learning_rate": 9.93771318931211e-06, "loss": 1.44, "step": 2253 }, { "epoch": 0.2398382634603107, "grad_norm": 12.6484375, "learning_rate": 9.937677657760091e-06, "loss": 1.2893, "step": 2254 }, { "epoch": 0.23994466907852735, "grad_norm": 17.875, "learning_rate": 9.937642126208073e-06, "loss": 1.1252, "step": 2255 }, { "epoch": 0.240051074696744, "grad_norm": 14.109375, "learning_rate": 9.937606594656054e-06, "loss": 1.3222, "step": 2256 }, { "epoch": 0.24015748031496062, "grad_norm": 13.5703125, "learning_rate": 9.937571063104036e-06, "loss": 1.3594, "step": 2257 }, { "epoch": 0.24026388593317727, "grad_norm": 14.1171875, "learning_rate": 9.93753553155202e-06, "loss": 1.4292, "step": 2258 }, { "epoch": 0.24037029155139392, "grad_norm": 13.8125, "learning_rate": 9.937500000000001e-06, "loss": 1.1102, "step": 2259 }, { "epoch": 0.24047669716961056, "grad_norm": 11.5078125, "learning_rate": 9.937464468447983e-06, "loss": 1.1182, "step": 2260 }, { "epoch": 0.24058310278782719, "grad_norm": 13.921875, "learning_rate": 9.937428936895964e-06, "loss": 1.3206, "step": 2261 }, { "epoch": 0.24068950840604383, "grad_norm": 15.7734375, "learning_rate": 9.937393405343946e-06, "loss": 1.5073, "step": 2262 }, { "epoch": 0.24079591402426048, "grad_norm": 16.9375, "learning_rate": 9.937357873791929e-06, "loss": 1.2486, "step": 2263 }, { "epoch": 0.24090231964247713, "grad_norm": 15.4375, "learning_rate": 9.93732234223991e-06, "loss": 1.3364, "step": 2264 }, { "epoch": 0.24100872526069375, "grad_norm": 16.859375, "learning_rate": 9.937286810687892e-06, "loss": 1.5195, "step": 2265 }, { "epoch": 0.2411151308789104, "grad_norm": 17.015625, "learning_rate": 9.937251279135874e-06, "loss": 1.0677, "step": 2266 }, { "epoch": 0.24122153649712705, "grad_norm": 17.640625, "learning_rate": 9.937215747583856e-06, "loss": 1.6502, "step": 2267 }, { "epoch": 0.2413279421153437, "grad_norm": 20.859375, "learning_rate": 9.937180216031837e-06, "loss": 1.2199, "step": 2268 }, { "epoch": 0.24143434773356034, "grad_norm": 13.875, "learning_rate": 9.937144684479819e-06, "loss": 1.1389, "step": 2269 }, { "epoch": 0.24154075335177697, "grad_norm": 15.5, "learning_rate": 9.9371091529278e-06, "loss": 1.3894, "step": 2270 }, { "epoch": 0.24164715896999361, "grad_norm": 17.140625, "learning_rate": 9.937073621375782e-06, "loss": 1.5974, "step": 2271 }, { "epoch": 0.24175356458821026, "grad_norm": 16.9375, "learning_rate": 9.937038089823764e-06, "loss": 1.3674, "step": 2272 }, { "epoch": 0.2418599702064269, "grad_norm": 13.8203125, "learning_rate": 9.937002558271747e-06, "loss": 1.1373, "step": 2273 }, { "epoch": 0.24196637582464353, "grad_norm": 17.640625, "learning_rate": 9.936967026719728e-06, "loss": 1.4796, "step": 2274 }, { "epoch": 0.24207278144286018, "grad_norm": 14.65625, "learning_rate": 9.93693149516771e-06, "loss": 1.3415, "step": 2275 }, { "epoch": 0.24217918706107683, "grad_norm": 15.640625, "learning_rate": 9.936895963615692e-06, "loss": 1.345, "step": 2276 }, { "epoch": 0.24228559267929348, "grad_norm": 13.421875, "learning_rate": 9.936860432063673e-06, "loss": 1.1321, "step": 2277 }, { "epoch": 0.2423919982975101, "grad_norm": 14.8515625, "learning_rate": 9.936824900511655e-06, "loss": 1.2062, "step": 2278 }, { "epoch": 0.24249840391572675, "grad_norm": 15.3359375, "learning_rate": 9.936789368959637e-06, "loss": 1.3942, "step": 2279 }, { "epoch": 0.2426048095339434, "grad_norm": 19.140625, "learning_rate": 9.936753837407618e-06, "loss": 1.0373, "step": 2280 }, { "epoch": 0.24271121515216004, "grad_norm": 15.9609375, "learning_rate": 9.9367183058556e-06, "loss": 1.1393, "step": 2281 }, { "epoch": 0.24281762077037666, "grad_norm": 13.6953125, "learning_rate": 9.936682774303581e-06, "loss": 1.0295, "step": 2282 }, { "epoch": 0.2429240263885933, "grad_norm": 24.34375, "learning_rate": 9.936647242751563e-06, "loss": 1.4413, "step": 2283 }, { "epoch": 0.24303043200680996, "grad_norm": 18.3125, "learning_rate": 9.936611711199546e-06, "loss": 1.4712, "step": 2284 }, { "epoch": 0.2431368376250266, "grad_norm": 14.1484375, "learning_rate": 9.936576179647528e-06, "loss": 1.1605, "step": 2285 }, { "epoch": 0.24324324324324326, "grad_norm": 15.1015625, "learning_rate": 9.93654064809551e-06, "loss": 1.3015, "step": 2286 }, { "epoch": 0.24334964886145988, "grad_norm": 16.65625, "learning_rate": 9.936505116543491e-06, "loss": 1.1935, "step": 2287 }, { "epoch": 0.24345605447967653, "grad_norm": 12.5546875, "learning_rate": 9.936469584991473e-06, "loss": 1.1999, "step": 2288 }, { "epoch": 0.24356246009789317, "grad_norm": 22.296875, "learning_rate": 9.936434053439456e-06, "loss": 1.5772, "step": 2289 }, { "epoch": 0.24366886571610982, "grad_norm": 19.03125, "learning_rate": 9.936398521887438e-06, "loss": 1.6325, "step": 2290 }, { "epoch": 0.24377527133432644, "grad_norm": 16.453125, "learning_rate": 9.93636299033542e-06, "loss": 1.3571, "step": 2291 }, { "epoch": 0.2438816769525431, "grad_norm": 14.28125, "learning_rate": 9.9363274587834e-06, "loss": 1.267, "step": 2292 }, { "epoch": 0.24398808257075974, "grad_norm": 14.5625, "learning_rate": 9.936291927231382e-06, "loss": 1.3546, "step": 2293 }, { "epoch": 0.2440944881889764, "grad_norm": 16.515625, "learning_rate": 9.936256395679364e-06, "loss": 1.3989, "step": 2294 }, { "epoch": 0.244200893807193, "grad_norm": 14.453125, "learning_rate": 9.936220864127346e-06, "loss": 1.206, "step": 2295 }, { "epoch": 0.24430729942540966, "grad_norm": 18.75, "learning_rate": 9.936185332575327e-06, "loss": 1.5563, "step": 2296 }, { "epoch": 0.2444137050436263, "grad_norm": 91.4375, "learning_rate": 9.936149801023309e-06, "loss": 1.4351, "step": 2297 }, { "epoch": 0.24452011066184295, "grad_norm": 24.25, "learning_rate": 9.93611426947129e-06, "loss": 1.8309, "step": 2298 }, { "epoch": 0.24462651628005957, "grad_norm": 13.96875, "learning_rate": 9.936078737919274e-06, "loss": 1.2587, "step": 2299 }, { "epoch": 0.24473292189827622, "grad_norm": 19.515625, "learning_rate": 9.936043206367255e-06, "loss": 1.3189, "step": 2300 }, { "epoch": 0.24483932751649287, "grad_norm": 10.3125, "learning_rate": 9.936007674815237e-06, "loss": 1.2511, "step": 2301 }, { "epoch": 0.24494573313470952, "grad_norm": 17.03125, "learning_rate": 9.935972143263219e-06, "loss": 1.3637, "step": 2302 }, { "epoch": 0.24505213875292614, "grad_norm": 13.109375, "learning_rate": 9.9359366117112e-06, "loss": 1.0585, "step": 2303 }, { "epoch": 0.2451585443711428, "grad_norm": 13.390625, "learning_rate": 9.935901080159183e-06, "loss": 1.1533, "step": 2304 }, { "epoch": 0.24526494998935944, "grad_norm": 15.484375, "learning_rate": 9.935865548607165e-06, "loss": 1.3249, "step": 2305 }, { "epoch": 0.24537135560757609, "grad_norm": 15.0703125, "learning_rate": 9.935830017055147e-06, "loss": 1.2111, "step": 2306 }, { "epoch": 0.24547776122579273, "grad_norm": 15.2734375, "learning_rate": 9.935794485503127e-06, "loss": 1.2098, "step": 2307 }, { "epoch": 0.24558416684400936, "grad_norm": 16.53125, "learning_rate": 9.935758953951108e-06, "loss": 1.3976, "step": 2308 }, { "epoch": 0.245690572462226, "grad_norm": 21.875, "learning_rate": 9.935723422399091e-06, "loss": 1.3614, "step": 2309 }, { "epoch": 0.24579697808044265, "grad_norm": 20.953125, "learning_rate": 9.935687890847073e-06, "loss": 1.3874, "step": 2310 }, { "epoch": 0.2459033836986593, "grad_norm": 15.671875, "learning_rate": 9.935652359295055e-06, "loss": 1.5008, "step": 2311 }, { "epoch": 0.24600978931687592, "grad_norm": 13.9140625, "learning_rate": 9.935616827743036e-06, "loss": 1.4044, "step": 2312 }, { "epoch": 0.24611619493509257, "grad_norm": 15.453125, "learning_rate": 9.935581296191018e-06, "loss": 1.1357, "step": 2313 }, { "epoch": 0.24622260055330922, "grad_norm": 21.78125, "learning_rate": 9.935545764639001e-06, "loss": 1.3588, "step": 2314 }, { "epoch": 0.24632900617152587, "grad_norm": 17.03125, "learning_rate": 9.935510233086983e-06, "loss": 1.2803, "step": 2315 }, { "epoch": 0.2464354117897425, "grad_norm": 27.875, "learning_rate": 9.935474701534964e-06, "loss": 1.5034, "step": 2316 }, { "epoch": 0.24654181740795914, "grad_norm": 16.8125, "learning_rate": 9.935439169982946e-06, "loss": 1.3059, "step": 2317 }, { "epoch": 0.24664822302617578, "grad_norm": 16.5, "learning_rate": 9.935403638430928e-06, "loss": 1.4682, "step": 2318 }, { "epoch": 0.24675462864439243, "grad_norm": 16.8125, "learning_rate": 9.93536810687891e-06, "loss": 1.4356, "step": 2319 }, { "epoch": 0.24686103426260905, "grad_norm": 18.390625, "learning_rate": 9.935332575326891e-06, "loss": 1.6387, "step": 2320 }, { "epoch": 0.2469674398808257, "grad_norm": 16.703125, "learning_rate": 9.935297043774872e-06, "loss": 1.197, "step": 2321 }, { "epoch": 0.24707384549904235, "grad_norm": 17.5625, "learning_rate": 9.935261512222854e-06, "loss": 1.7398, "step": 2322 }, { "epoch": 0.247180251117259, "grad_norm": 14.625, "learning_rate": 9.935225980670836e-06, "loss": 1.2808, "step": 2323 }, { "epoch": 0.24728665673547565, "grad_norm": 15.734375, "learning_rate": 9.935190449118817e-06, "loss": 1.4233, "step": 2324 }, { "epoch": 0.24739306235369227, "grad_norm": 22.53125, "learning_rate": 9.9351549175668e-06, "loss": 1.1787, "step": 2325 }, { "epoch": 0.24749946797190892, "grad_norm": 17.703125, "learning_rate": 9.935119386014782e-06, "loss": 1.2937, "step": 2326 }, { "epoch": 0.24760587359012556, "grad_norm": 14.4609375, "learning_rate": 9.935083854462764e-06, "loss": 1.3068, "step": 2327 }, { "epoch": 0.2477122792083422, "grad_norm": 11.2578125, "learning_rate": 9.935048322910745e-06, "loss": 1.0747, "step": 2328 }, { "epoch": 0.24781868482655883, "grad_norm": 14.171875, "learning_rate": 9.935012791358727e-06, "loss": 1.2475, "step": 2329 }, { "epoch": 0.24792509044477548, "grad_norm": 19.6875, "learning_rate": 9.93497725980671e-06, "loss": 1.2557, "step": 2330 }, { "epoch": 0.24803149606299213, "grad_norm": 22.46875, "learning_rate": 9.934941728254692e-06, "loss": 1.7281, "step": 2331 }, { "epoch": 0.24813790168120878, "grad_norm": 13.90625, "learning_rate": 9.934906196702674e-06, "loss": 1.2173, "step": 2332 }, { "epoch": 0.2482443072994254, "grad_norm": 16.640625, "learning_rate": 9.934870665150655e-06, "loss": 1.1172, "step": 2333 }, { "epoch": 0.24835071291764205, "grad_norm": 15.0859375, "learning_rate": 9.934835133598637e-06, "loss": 1.3903, "step": 2334 }, { "epoch": 0.2484571185358587, "grad_norm": 19.53125, "learning_rate": 9.934799602046618e-06, "loss": 1.5711, "step": 2335 }, { "epoch": 0.24856352415407534, "grad_norm": 14.296875, "learning_rate": 9.9347640704946e-06, "loss": 1.0755, "step": 2336 }, { "epoch": 0.24866992977229196, "grad_norm": 14.328125, "learning_rate": 9.934728538942582e-06, "loss": 1.4814, "step": 2337 }, { "epoch": 0.2487763353905086, "grad_norm": 12.1796875, "learning_rate": 9.934693007390563e-06, "loss": 1.0862, "step": 2338 }, { "epoch": 0.24888274100872526, "grad_norm": 17.703125, "learning_rate": 9.934657475838545e-06, "loss": 1.4946, "step": 2339 }, { "epoch": 0.2489891466269419, "grad_norm": 15.2109375, "learning_rate": 9.934621944286528e-06, "loss": 1.0949, "step": 2340 }, { "epoch": 0.24909555224515856, "grad_norm": 15.09375, "learning_rate": 9.93458641273451e-06, "loss": 1.3129, "step": 2341 }, { "epoch": 0.24920195786337518, "grad_norm": 15.4296875, "learning_rate": 9.934550881182491e-06, "loss": 1.4525, "step": 2342 }, { "epoch": 0.24930836348159183, "grad_norm": 19.203125, "learning_rate": 9.934515349630473e-06, "loss": 1.4118, "step": 2343 }, { "epoch": 0.24941476909980848, "grad_norm": 15.6875, "learning_rate": 9.934479818078454e-06, "loss": 1.3997, "step": 2344 }, { "epoch": 0.24952117471802512, "grad_norm": 17.703125, "learning_rate": 9.934444286526436e-06, "loss": 1.5512, "step": 2345 }, { "epoch": 0.24962758033624174, "grad_norm": 15.03125, "learning_rate": 9.934408754974418e-06, "loss": 1.1362, "step": 2346 }, { "epoch": 0.2497339859544584, "grad_norm": 13.625, "learning_rate": 9.9343732234224e-06, "loss": 1.2536, "step": 2347 }, { "epoch": 0.24984039157267504, "grad_norm": 17.03125, "learning_rate": 9.934337691870381e-06, "loss": 1.1487, "step": 2348 }, { "epoch": 0.2499467971908917, "grad_norm": 14.6640625, "learning_rate": 9.934302160318363e-06, "loss": 1.1648, "step": 2349 }, { "epoch": 0.2500532028091083, "grad_norm": 13.9609375, "learning_rate": 9.934266628766346e-06, "loss": 1.3575, "step": 2350 }, { "epoch": 0.250159608427325, "grad_norm": 21.4375, "learning_rate": 9.934231097214327e-06, "loss": 1.7998, "step": 2351 }, { "epoch": 0.2502660140455416, "grad_norm": 12.4140625, "learning_rate": 9.934195565662309e-06, "loss": 1.4054, "step": 2352 }, { "epoch": 0.25037241966375823, "grad_norm": 49.46875, "learning_rate": 9.93416003411029e-06, "loss": 1.1218, "step": 2353 }, { "epoch": 0.2504788252819749, "grad_norm": 17.515625, "learning_rate": 9.934124502558272e-06, "loss": 1.3351, "step": 2354 }, { "epoch": 0.2505852309001915, "grad_norm": 20.5, "learning_rate": 9.934088971006256e-06, "loss": 1.5813, "step": 2355 }, { "epoch": 0.25069163651840815, "grad_norm": 23.796875, "learning_rate": 9.934053439454237e-06, "loss": 1.2328, "step": 2356 }, { "epoch": 0.2507980421366248, "grad_norm": 13.6171875, "learning_rate": 9.934017907902219e-06, "loss": 1.1773, "step": 2357 }, { "epoch": 0.25090444775484144, "grad_norm": 20.96875, "learning_rate": 9.9339823763502e-06, "loss": 1.4018, "step": 2358 }, { "epoch": 0.2510108533730581, "grad_norm": 14.765625, "learning_rate": 9.933946844798182e-06, "loss": 1.0513, "step": 2359 }, { "epoch": 0.25111725899127474, "grad_norm": 14.140625, "learning_rate": 9.933911313246164e-06, "loss": 1.4058, "step": 2360 }, { "epoch": 0.25122366460949136, "grad_norm": 14.453125, "learning_rate": 9.933875781694145e-06, "loss": 1.3108, "step": 2361 }, { "epoch": 0.25133007022770804, "grad_norm": 24.21875, "learning_rate": 9.933840250142127e-06, "loss": 1.5951, "step": 2362 }, { "epoch": 0.25143647584592466, "grad_norm": 17.296875, "learning_rate": 9.933804718590108e-06, "loss": 1.4964, "step": 2363 }, { "epoch": 0.25154288146414133, "grad_norm": 12.7890625, "learning_rate": 9.93376918703809e-06, "loss": 0.981, "step": 2364 }, { "epoch": 0.25164928708235795, "grad_norm": 14.109375, "learning_rate": 9.933733655486072e-06, "loss": 1.1947, "step": 2365 }, { "epoch": 0.2517556927005746, "grad_norm": 13.5859375, "learning_rate": 9.933698123934055e-06, "loss": 1.1676, "step": 2366 }, { "epoch": 0.25186209831879125, "grad_norm": 17.984375, "learning_rate": 9.933662592382037e-06, "loss": 1.4085, "step": 2367 }, { "epoch": 0.25196850393700787, "grad_norm": 12.3359375, "learning_rate": 9.933627060830018e-06, "loss": 1.2526, "step": 2368 }, { "epoch": 0.2520749095552245, "grad_norm": 20.640625, "learning_rate": 9.933591529278e-06, "loss": 1.4725, "step": 2369 }, { "epoch": 0.25218131517344117, "grad_norm": 18.765625, "learning_rate": 9.933555997725981e-06, "loss": 1.2619, "step": 2370 }, { "epoch": 0.2522877207916578, "grad_norm": 9.703125, "learning_rate": 9.933520466173963e-06, "loss": 0.9655, "step": 2371 }, { "epoch": 0.25239412640987446, "grad_norm": 13.921875, "learning_rate": 9.933484934621945e-06, "loss": 1.0456, "step": 2372 }, { "epoch": 0.2525005320280911, "grad_norm": 17.5625, "learning_rate": 9.933449403069926e-06, "loss": 1.5837, "step": 2373 }, { "epoch": 0.2526069376463077, "grad_norm": 13.1875, "learning_rate": 9.933413871517908e-06, "loss": 1.3417, "step": 2374 }, { "epoch": 0.2527133432645244, "grad_norm": 13.78125, "learning_rate": 9.93337833996589e-06, "loss": 1.272, "step": 2375 }, { "epoch": 0.252819748882741, "grad_norm": 18.984375, "learning_rate": 9.933342808413873e-06, "loss": 1.5912, "step": 2376 }, { "epoch": 0.2529261545009576, "grad_norm": 17.328125, "learning_rate": 9.933307276861854e-06, "loss": 1.0662, "step": 2377 }, { "epoch": 0.2530325601191743, "grad_norm": 13.921875, "learning_rate": 9.933271745309836e-06, "loss": 1.0574, "step": 2378 }, { "epoch": 0.2531389657373909, "grad_norm": 15.28125, "learning_rate": 9.933236213757818e-06, "loss": 1.1564, "step": 2379 }, { "epoch": 0.2532453713556076, "grad_norm": 17.75, "learning_rate": 9.933200682205799e-06, "loss": 1.2457, "step": 2380 }, { "epoch": 0.2533517769738242, "grad_norm": 12.203125, "learning_rate": 9.933165150653782e-06, "loss": 1.1591, "step": 2381 }, { "epoch": 0.25345818259204084, "grad_norm": 14.625, "learning_rate": 9.933129619101764e-06, "loss": 1.1118, "step": 2382 }, { "epoch": 0.2535645882102575, "grad_norm": 14.328125, "learning_rate": 9.933094087549746e-06, "loss": 1.2594, "step": 2383 }, { "epoch": 0.25367099382847413, "grad_norm": 24.578125, "learning_rate": 9.933058555997727e-06, "loss": 1.3602, "step": 2384 }, { "epoch": 0.2537773994466908, "grad_norm": 15.0390625, "learning_rate": 9.933023024445709e-06, "loss": 1.2961, "step": 2385 }, { "epoch": 0.25388380506490743, "grad_norm": 26.515625, "learning_rate": 9.93298749289369e-06, "loss": 1.2585, "step": 2386 }, { "epoch": 0.25399021068312405, "grad_norm": 21.265625, "learning_rate": 9.932951961341672e-06, "loss": 2.0428, "step": 2387 }, { "epoch": 0.2540966163013407, "grad_norm": 14.8125, "learning_rate": 9.932916429789654e-06, "loss": 1.2266, "step": 2388 }, { "epoch": 0.25420302191955735, "grad_norm": 11.8125, "learning_rate": 9.932880898237635e-06, "loss": 1.0875, "step": 2389 }, { "epoch": 0.25430942753777397, "grad_norm": 12.8828125, "learning_rate": 9.932845366685617e-06, "loss": 1.2883, "step": 2390 }, { "epoch": 0.25441583315599064, "grad_norm": 13.7265625, "learning_rate": 9.9328098351336e-06, "loss": 1.1599, "step": 2391 }, { "epoch": 0.25452223877420727, "grad_norm": 12.859375, "learning_rate": 9.932774303581582e-06, "loss": 1.2088, "step": 2392 }, { "epoch": 0.25462864439242394, "grad_norm": 13.9140625, "learning_rate": 9.932738772029563e-06, "loss": 1.1354, "step": 2393 }, { "epoch": 0.25473505001064056, "grad_norm": 16.6875, "learning_rate": 9.932703240477545e-06, "loss": 1.2653, "step": 2394 }, { "epoch": 0.2548414556288572, "grad_norm": 13.234375, "learning_rate": 9.932667708925527e-06, "loss": 1.2314, "step": 2395 }, { "epoch": 0.25494786124707386, "grad_norm": 19.375, "learning_rate": 9.932632177373508e-06, "loss": 1.3759, "step": 2396 }, { "epoch": 0.2550542668652905, "grad_norm": 17.546875, "learning_rate": 9.93259664582149e-06, "loss": 1.554, "step": 2397 }, { "epoch": 0.25516067248350716, "grad_norm": 19.3125, "learning_rate": 9.932561114269471e-06, "loss": 1.084, "step": 2398 }, { "epoch": 0.2552670781017238, "grad_norm": 60.78125, "learning_rate": 9.932525582717453e-06, "loss": 1.1445, "step": 2399 }, { "epoch": 0.2553734837199404, "grad_norm": 23.625, "learning_rate": 9.932490051165435e-06, "loss": 1.4245, "step": 2400 }, { "epoch": 0.2554798893381571, "grad_norm": 17.078125, "learning_rate": 9.932454519613416e-06, "loss": 1.3068, "step": 2401 }, { "epoch": 0.2555862949563737, "grad_norm": 13.234375, "learning_rate": 9.9324189880614e-06, "loss": 1.0714, "step": 2402 }, { "epoch": 0.2556927005745903, "grad_norm": 17.28125, "learning_rate": 9.932383456509381e-06, "loss": 1.5554, "step": 2403 }, { "epoch": 0.255799106192807, "grad_norm": 14.171875, "learning_rate": 9.932347924957363e-06, "loss": 1.2563, "step": 2404 }, { "epoch": 0.2559055118110236, "grad_norm": 16.546875, "learning_rate": 9.932312393405344e-06, "loss": 1.144, "step": 2405 }, { "epoch": 0.2560119174292403, "grad_norm": 14.578125, "learning_rate": 9.932276861853326e-06, "loss": 1.2056, "step": 2406 }, { "epoch": 0.2561183230474569, "grad_norm": 31.421875, "learning_rate": 9.93224133030131e-06, "loss": 1.7856, "step": 2407 }, { "epoch": 0.25622472866567353, "grad_norm": 13.828125, "learning_rate": 9.932205798749291e-06, "loss": 1.3087, "step": 2408 }, { "epoch": 0.2563311342838902, "grad_norm": 18.59375, "learning_rate": 9.932170267197272e-06, "loss": 1.4887, "step": 2409 }, { "epoch": 0.2564375399021068, "grad_norm": 23.671875, "learning_rate": 9.932134735645254e-06, "loss": 1.801, "step": 2410 }, { "epoch": 0.25654394552032345, "grad_norm": 12.5703125, "learning_rate": 9.932099204093236e-06, "loss": 1.4053, "step": 2411 }, { "epoch": 0.2566503511385401, "grad_norm": 14.484375, "learning_rate": 9.932063672541217e-06, "loss": 1.2438, "step": 2412 }, { "epoch": 0.25675675675675674, "grad_norm": 17.578125, "learning_rate": 9.932028140989199e-06, "loss": 1.7346, "step": 2413 }, { "epoch": 0.2568631623749734, "grad_norm": 23.296875, "learning_rate": 9.93199260943718e-06, "loss": 1.4081, "step": 2414 }, { "epoch": 0.25696956799319004, "grad_norm": 16.421875, "learning_rate": 9.931957077885162e-06, "loss": 1.3355, "step": 2415 }, { "epoch": 0.25707597361140666, "grad_norm": 16.25, "learning_rate": 9.931921546333144e-06, "loss": 1.6126, "step": 2416 }, { "epoch": 0.25718237922962334, "grad_norm": 15.9765625, "learning_rate": 9.931886014781127e-06, "loss": 1.3045, "step": 2417 }, { "epoch": 0.25728878484783996, "grad_norm": 15.796875, "learning_rate": 9.931850483229109e-06, "loss": 1.5051, "step": 2418 }, { "epoch": 0.25739519046605663, "grad_norm": 15.1875, "learning_rate": 9.93181495167709e-06, "loss": 1.1791, "step": 2419 }, { "epoch": 0.25750159608427325, "grad_norm": 20.125, "learning_rate": 9.931779420125072e-06, "loss": 1.5663, "step": 2420 }, { "epoch": 0.2576080017024899, "grad_norm": 24.015625, "learning_rate": 9.931743888573053e-06, "loss": 1.3333, "step": 2421 }, { "epoch": 0.25771440732070655, "grad_norm": 14.96875, "learning_rate": 9.931708357021037e-06, "loss": 1.1065, "step": 2422 }, { "epoch": 0.25782081293892317, "grad_norm": 14.890625, "learning_rate": 9.931672825469018e-06, "loss": 1.4916, "step": 2423 }, { "epoch": 0.2579272185571398, "grad_norm": 13.1484375, "learning_rate": 9.931637293917e-06, "loss": 1.4625, "step": 2424 }, { "epoch": 0.25803362417535647, "grad_norm": 17.359375, "learning_rate": 9.931601762364982e-06, "loss": 1.2036, "step": 2425 }, { "epoch": 0.2581400297935731, "grad_norm": 19.46875, "learning_rate": 9.931566230812961e-06, "loss": 1.0975, "step": 2426 }, { "epoch": 0.25824643541178977, "grad_norm": 12.4609375, "learning_rate": 9.931530699260945e-06, "loss": 1.3353, "step": 2427 }, { "epoch": 0.2583528410300064, "grad_norm": 19.671875, "learning_rate": 9.931495167708926e-06, "loss": 1.3192, "step": 2428 }, { "epoch": 0.258459246648223, "grad_norm": 15.890625, "learning_rate": 9.931459636156908e-06, "loss": 1.6842, "step": 2429 }, { "epoch": 0.2585656522664397, "grad_norm": 16.796875, "learning_rate": 9.93142410460489e-06, "loss": 1.7826, "step": 2430 }, { "epoch": 0.2586720578846563, "grad_norm": 22.84375, "learning_rate": 9.931388573052871e-06, "loss": 1.2909, "step": 2431 }, { "epoch": 0.2587784635028729, "grad_norm": 18.359375, "learning_rate": 9.931353041500855e-06, "loss": 1.448, "step": 2432 }, { "epoch": 0.2588848691210896, "grad_norm": 12.7578125, "learning_rate": 9.931317509948836e-06, "loss": 1.2871, "step": 2433 }, { "epoch": 0.2589912747393062, "grad_norm": 12.671875, "learning_rate": 9.931281978396818e-06, "loss": 1.4029, "step": 2434 }, { "epoch": 0.2590976803575229, "grad_norm": 15.0, "learning_rate": 9.9312464468448e-06, "loss": 1.2502, "step": 2435 }, { "epoch": 0.2592040859757395, "grad_norm": 19.96875, "learning_rate": 9.931210915292781e-06, "loss": 1.3651, "step": 2436 }, { "epoch": 0.25931049159395614, "grad_norm": 13.4921875, "learning_rate": 9.931175383740763e-06, "loss": 1.2188, "step": 2437 }, { "epoch": 0.2594168972121728, "grad_norm": 13.4609375, "learning_rate": 9.931139852188744e-06, "loss": 1.4395, "step": 2438 }, { "epoch": 0.25952330283038944, "grad_norm": 13.7109375, "learning_rate": 9.931104320636726e-06, "loss": 1.3949, "step": 2439 }, { "epoch": 0.2596297084486061, "grad_norm": 13.78125, "learning_rate": 9.931068789084707e-06, "loss": 1.5448, "step": 2440 }, { "epoch": 0.25973611406682273, "grad_norm": 11.5859375, "learning_rate": 9.931033257532689e-06, "loss": 1.1818, "step": 2441 }, { "epoch": 0.25984251968503935, "grad_norm": 22.359375, "learning_rate": 9.93099772598067e-06, "loss": 1.5556, "step": 2442 }, { "epoch": 0.25994892530325603, "grad_norm": 14.734375, "learning_rate": 9.930962194428654e-06, "loss": 1.1006, "step": 2443 }, { "epoch": 0.26005533092147265, "grad_norm": 17.15625, "learning_rate": 9.930926662876635e-06, "loss": 1.0946, "step": 2444 }, { "epoch": 0.26016173653968927, "grad_norm": 17.40625, "learning_rate": 9.930891131324617e-06, "loss": 1.0857, "step": 2445 }, { "epoch": 0.26026814215790595, "grad_norm": 14.375, "learning_rate": 9.930855599772599e-06, "loss": 1.3446, "step": 2446 }, { "epoch": 0.26037454777612257, "grad_norm": 15.0234375, "learning_rate": 9.93082006822058e-06, "loss": 1.4626, "step": 2447 }, { "epoch": 0.26048095339433924, "grad_norm": 13.15625, "learning_rate": 9.930784536668564e-06, "loss": 1.0991, "step": 2448 }, { "epoch": 0.26058735901255586, "grad_norm": 17.40625, "learning_rate": 9.930749005116545e-06, "loss": 1.3631, "step": 2449 }, { "epoch": 0.2606937646307725, "grad_norm": 14.90625, "learning_rate": 9.930713473564527e-06, "loss": 1.4175, "step": 2450 }, { "epoch": 0.26080017024898916, "grad_norm": 15.3046875, "learning_rate": 9.930677942012508e-06, "loss": 1.2666, "step": 2451 }, { "epoch": 0.2609065758672058, "grad_norm": 15.5703125, "learning_rate": 9.93064241046049e-06, "loss": 1.1702, "step": 2452 }, { "epoch": 0.26101298148542246, "grad_norm": 15.484375, "learning_rate": 9.930606878908472e-06, "loss": 1.5043, "step": 2453 }, { "epoch": 0.2611193871036391, "grad_norm": 18.609375, "learning_rate": 9.930571347356453e-06, "loss": 1.5268, "step": 2454 }, { "epoch": 0.2612257927218557, "grad_norm": 18.078125, "learning_rate": 9.930535815804435e-06, "loss": 1.2904, "step": 2455 }, { "epoch": 0.2613321983400724, "grad_norm": 14.234375, "learning_rate": 9.930500284252416e-06, "loss": 1.318, "step": 2456 }, { "epoch": 0.261438603958289, "grad_norm": 12.8203125, "learning_rate": 9.930464752700398e-06, "loss": 1.084, "step": 2457 }, { "epoch": 0.2615450095765056, "grad_norm": 15.671875, "learning_rate": 9.930429221148381e-06, "loss": 1.4791, "step": 2458 }, { "epoch": 0.2616514151947223, "grad_norm": 15.7421875, "learning_rate": 9.930393689596363e-06, "loss": 1.4083, "step": 2459 }, { "epoch": 0.2617578208129389, "grad_norm": 29.421875, "learning_rate": 9.930358158044345e-06, "loss": 1.4156, "step": 2460 }, { "epoch": 0.2618642264311556, "grad_norm": 17.828125, "learning_rate": 9.930322626492326e-06, "loss": 1.0067, "step": 2461 }, { "epoch": 0.2619706320493722, "grad_norm": 15.484375, "learning_rate": 9.930287094940308e-06, "loss": 1.5093, "step": 2462 }, { "epoch": 0.26207703766758883, "grad_norm": 14.8359375, "learning_rate": 9.93025156338829e-06, "loss": 0.9766, "step": 2463 }, { "epoch": 0.2621834432858055, "grad_norm": 13.125, "learning_rate": 9.930216031836271e-06, "loss": 1.2683, "step": 2464 }, { "epoch": 0.2622898489040221, "grad_norm": 10.6953125, "learning_rate": 9.930180500284253e-06, "loss": 0.9755, "step": 2465 }, { "epoch": 0.26239625452223875, "grad_norm": 15.59375, "learning_rate": 9.930144968732234e-06, "loss": 1.3163, "step": 2466 }, { "epoch": 0.2625026601404554, "grad_norm": 16.03125, "learning_rate": 9.930109437180216e-06, "loss": 1.5944, "step": 2467 }, { "epoch": 0.26260906575867204, "grad_norm": 15.4140625, "learning_rate": 9.930073905628199e-06, "loss": 1.3014, "step": 2468 }, { "epoch": 0.2627154713768887, "grad_norm": 16.671875, "learning_rate": 9.93003837407618e-06, "loss": 1.373, "step": 2469 }, { "epoch": 0.26282187699510534, "grad_norm": 15.0703125, "learning_rate": 9.930002842524162e-06, "loss": 1.116, "step": 2470 }, { "epoch": 0.26292828261332196, "grad_norm": 16.046875, "learning_rate": 9.929967310972144e-06, "loss": 1.3112, "step": 2471 }, { "epoch": 0.26303468823153864, "grad_norm": 13.2265625, "learning_rate": 9.929931779420126e-06, "loss": 1.2395, "step": 2472 }, { "epoch": 0.26314109384975526, "grad_norm": 14.6328125, "learning_rate": 9.929896247868109e-06, "loss": 1.2583, "step": 2473 }, { "epoch": 0.26324749946797193, "grad_norm": 17.828125, "learning_rate": 9.92986071631609e-06, "loss": 1.5066, "step": 2474 }, { "epoch": 0.26335390508618856, "grad_norm": 21.984375, "learning_rate": 9.929825184764072e-06, "loss": 1.4261, "step": 2475 }, { "epoch": 0.2634603107044052, "grad_norm": 19.921875, "learning_rate": 9.929789653212054e-06, "loss": 1.2574, "step": 2476 }, { "epoch": 0.26356671632262185, "grad_norm": 15.4140625, "learning_rate": 9.929754121660035e-06, "loss": 1.3924, "step": 2477 }, { "epoch": 0.2636731219408385, "grad_norm": 10.40625, "learning_rate": 9.929718590108017e-06, "loss": 1.0573, "step": 2478 }, { "epoch": 0.2637795275590551, "grad_norm": 20.25, "learning_rate": 9.929683058555999e-06, "loss": 1.6769, "step": 2479 }, { "epoch": 0.26388593317727177, "grad_norm": 14.1484375, "learning_rate": 9.92964752700398e-06, "loss": 1.163, "step": 2480 }, { "epoch": 0.2639923387954884, "grad_norm": 18.59375, "learning_rate": 9.929611995451962e-06, "loss": 1.3583, "step": 2481 }, { "epoch": 0.26409874441370507, "grad_norm": 15.265625, "learning_rate": 9.929576463899943e-06, "loss": 0.9472, "step": 2482 }, { "epoch": 0.2642051500319217, "grad_norm": 20.875, "learning_rate": 9.929540932347925e-06, "loss": 1.6746, "step": 2483 }, { "epoch": 0.2643115556501383, "grad_norm": 23.125, "learning_rate": 9.929505400795908e-06, "loss": 1.4331, "step": 2484 }, { "epoch": 0.264417961268355, "grad_norm": 12.5546875, "learning_rate": 9.92946986924389e-06, "loss": 1.0875, "step": 2485 }, { "epoch": 0.2645243668865716, "grad_norm": 10.8359375, "learning_rate": 9.929434337691871e-06, "loss": 1.2207, "step": 2486 }, { "epoch": 0.2646307725047882, "grad_norm": 15.40625, "learning_rate": 9.929398806139853e-06, "loss": 1.7142, "step": 2487 }, { "epoch": 0.2647371781230049, "grad_norm": 13.0234375, "learning_rate": 9.929363274587835e-06, "loss": 1.2048, "step": 2488 }, { "epoch": 0.2648435837412215, "grad_norm": 17.5, "learning_rate": 9.929327743035816e-06, "loss": 1.4734, "step": 2489 }, { "epoch": 0.2649499893594382, "grad_norm": 18.03125, "learning_rate": 9.929292211483798e-06, "loss": 1.3372, "step": 2490 }, { "epoch": 0.2650563949776548, "grad_norm": 16.6875, "learning_rate": 9.92925667993178e-06, "loss": 1.2534, "step": 2491 }, { "epoch": 0.26516280059587144, "grad_norm": 24.75, "learning_rate": 9.929221148379761e-06, "loss": 1.5406, "step": 2492 }, { "epoch": 0.2652692062140881, "grad_norm": 13.65625, "learning_rate": 9.929185616827743e-06, "loss": 1.1724, "step": 2493 }, { "epoch": 0.26537561183230474, "grad_norm": 15.5703125, "learning_rate": 9.929150085275726e-06, "loss": 1.3575, "step": 2494 }, { "epoch": 0.2654820174505214, "grad_norm": 20.359375, "learning_rate": 9.929114553723708e-06, "loss": 1.411, "step": 2495 }, { "epoch": 0.26558842306873803, "grad_norm": 12.4765625, "learning_rate": 9.92907902217169e-06, "loss": 1.2632, "step": 2496 }, { "epoch": 0.26569482868695465, "grad_norm": 14.7265625, "learning_rate": 9.92904349061967e-06, "loss": 1.0673, "step": 2497 }, { "epoch": 0.26580123430517133, "grad_norm": 22.765625, "learning_rate": 9.929007959067652e-06, "loss": 1.739, "step": 2498 }, { "epoch": 0.26590763992338795, "grad_norm": 13.140625, "learning_rate": 9.928972427515636e-06, "loss": 1.3413, "step": 2499 }, { "epoch": 0.26601404554160457, "grad_norm": 14.328125, "learning_rate": 9.928936895963617e-06, "loss": 1.4948, "step": 2500 }, { "epoch": 0.26612045115982125, "grad_norm": 19.25, "learning_rate": 9.928901364411599e-06, "loss": 1.4223, "step": 2501 }, { "epoch": 0.26622685677803787, "grad_norm": 20.421875, "learning_rate": 9.92886583285958e-06, "loss": 1.5152, "step": 2502 }, { "epoch": 0.26633326239625454, "grad_norm": 13.0546875, "learning_rate": 9.928830301307562e-06, "loss": 1.15, "step": 2503 }, { "epoch": 0.26643966801447116, "grad_norm": 19.078125, "learning_rate": 9.928794769755544e-06, "loss": 1.4376, "step": 2504 }, { "epoch": 0.2665460736326878, "grad_norm": 12.5078125, "learning_rate": 9.928759238203525e-06, "loss": 1.0498, "step": 2505 }, { "epoch": 0.26665247925090446, "grad_norm": 20.203125, "learning_rate": 9.928723706651507e-06, "loss": 1.4383, "step": 2506 }, { "epoch": 0.2667588848691211, "grad_norm": 16.25, "learning_rate": 9.928688175099489e-06, "loss": 1.4367, "step": 2507 }, { "epoch": 0.26686529048733776, "grad_norm": 17.078125, "learning_rate": 9.92865264354747e-06, "loss": 1.4794, "step": 2508 }, { "epoch": 0.2669716961055544, "grad_norm": 25.3125, "learning_rate": 9.928617111995453e-06, "loss": 1.4777, "step": 2509 }, { "epoch": 0.267078101723771, "grad_norm": 15.9296875, "learning_rate": 9.928581580443435e-06, "loss": 1.0159, "step": 2510 }, { "epoch": 0.2671845073419877, "grad_norm": 17.59375, "learning_rate": 9.928546048891417e-06, "loss": 1.5528, "step": 2511 }, { "epoch": 0.2672909129602043, "grad_norm": 18.96875, "learning_rate": 9.928510517339398e-06, "loss": 1.4157, "step": 2512 }, { "epoch": 0.2673973185784209, "grad_norm": 30.015625, "learning_rate": 9.92847498578738e-06, "loss": 1.6958, "step": 2513 }, { "epoch": 0.2675037241966376, "grad_norm": 18.765625, "learning_rate": 9.928439454235362e-06, "loss": 1.5119, "step": 2514 }, { "epoch": 0.2676101298148542, "grad_norm": 14.1953125, "learning_rate": 9.928403922683343e-06, "loss": 1.176, "step": 2515 }, { "epoch": 0.2677165354330709, "grad_norm": 19.03125, "learning_rate": 9.928368391131325e-06, "loss": 1.2181, "step": 2516 }, { "epoch": 0.2678229410512875, "grad_norm": 16.96875, "learning_rate": 9.928332859579306e-06, "loss": 1.4403, "step": 2517 }, { "epoch": 0.26792934666950413, "grad_norm": 16.234375, "learning_rate": 9.928297328027288e-06, "loss": 1.2036, "step": 2518 }, { "epoch": 0.2680357522877208, "grad_norm": 18.078125, "learning_rate": 9.92826179647527e-06, "loss": 1.5399, "step": 2519 }, { "epoch": 0.26814215790593743, "grad_norm": 264.0, "learning_rate": 9.928226264923253e-06, "loss": 2.8608, "step": 2520 }, { "epoch": 0.26824856352415405, "grad_norm": 16.53125, "learning_rate": 9.928190733371234e-06, "loss": 1.7486, "step": 2521 }, { "epoch": 0.2683549691423707, "grad_norm": 16.265625, "learning_rate": 9.928155201819216e-06, "loss": 1.3172, "step": 2522 }, { "epoch": 0.26846137476058735, "grad_norm": 20.109375, "learning_rate": 9.928119670267198e-06, "loss": 2.0099, "step": 2523 }, { "epoch": 0.268567780378804, "grad_norm": 24.984375, "learning_rate": 9.92808413871518e-06, "loss": 1.4593, "step": 2524 }, { "epoch": 0.26867418599702064, "grad_norm": 16.640625, "learning_rate": 9.928048607163163e-06, "loss": 1.3796, "step": 2525 }, { "epoch": 0.26878059161523726, "grad_norm": 13.8515625, "learning_rate": 9.928013075611144e-06, "loss": 1.2363, "step": 2526 }, { "epoch": 0.26888699723345394, "grad_norm": 20.296875, "learning_rate": 9.927977544059126e-06, "loss": 1.4872, "step": 2527 }, { "epoch": 0.26899340285167056, "grad_norm": 20.5, "learning_rate": 9.927942012507107e-06, "loss": 1.8012, "step": 2528 }, { "epoch": 0.26909980846988724, "grad_norm": 17.140625, "learning_rate": 9.927906480955089e-06, "loss": 1.6004, "step": 2529 }, { "epoch": 0.26920621408810386, "grad_norm": 13.3046875, "learning_rate": 9.92787094940307e-06, "loss": 1.249, "step": 2530 }, { "epoch": 0.2693126197063205, "grad_norm": 17.609375, "learning_rate": 9.927835417851052e-06, "loss": 1.6392, "step": 2531 }, { "epoch": 0.26941902532453715, "grad_norm": 13.953125, "learning_rate": 9.927799886299034e-06, "loss": 1.2494, "step": 2532 }, { "epoch": 0.2695254309427538, "grad_norm": 13.9765625, "learning_rate": 9.927764354747015e-06, "loss": 1.0696, "step": 2533 }, { "epoch": 0.2696318365609704, "grad_norm": 12.1484375, "learning_rate": 9.927728823194997e-06, "loss": 1.4126, "step": 2534 }, { "epoch": 0.26973824217918707, "grad_norm": 14.890625, "learning_rate": 9.92769329164298e-06, "loss": 1.1391, "step": 2535 }, { "epoch": 0.2698446477974037, "grad_norm": 18.890625, "learning_rate": 9.927657760090962e-06, "loss": 1.9813, "step": 2536 }, { "epoch": 0.26995105341562037, "grad_norm": 22.625, "learning_rate": 9.927622228538944e-06, "loss": 1.4651, "step": 2537 }, { "epoch": 0.270057459033837, "grad_norm": 21.015625, "learning_rate": 9.927586696986925e-06, "loss": 1.0672, "step": 2538 }, { "epoch": 0.2701638646520536, "grad_norm": 14.8984375, "learning_rate": 9.927551165434907e-06, "loss": 1.1641, "step": 2539 }, { "epoch": 0.2702702702702703, "grad_norm": 15.34375, "learning_rate": 9.92751563388289e-06, "loss": 1.3555, "step": 2540 }, { "epoch": 0.2703766758884869, "grad_norm": 15.9765625, "learning_rate": 9.927480102330872e-06, "loss": 1.3528, "step": 2541 }, { "epoch": 0.2704830815067035, "grad_norm": 15.90625, "learning_rate": 9.927444570778853e-06, "loss": 1.6131, "step": 2542 }, { "epoch": 0.2705894871249202, "grad_norm": 13.0703125, "learning_rate": 9.927409039226835e-06, "loss": 1.0489, "step": 2543 }, { "epoch": 0.2706958927431368, "grad_norm": 20.71875, "learning_rate": 9.927373507674816e-06, "loss": 1.4279, "step": 2544 }, { "epoch": 0.2708022983613535, "grad_norm": 13.1328125, "learning_rate": 9.927337976122798e-06, "loss": 1.0044, "step": 2545 }, { "epoch": 0.2709087039795701, "grad_norm": 27.4375, "learning_rate": 9.92730244457078e-06, "loss": 1.4721, "step": 2546 }, { "epoch": 0.27101510959778674, "grad_norm": 14.9140625, "learning_rate": 9.927266913018761e-06, "loss": 1.8505, "step": 2547 }, { "epoch": 0.2711215152160034, "grad_norm": 13.2421875, "learning_rate": 9.927231381466743e-06, "loss": 1.27, "step": 2548 }, { "epoch": 0.27122792083422004, "grad_norm": 15.8515625, "learning_rate": 9.927195849914725e-06, "loss": 1.3035, "step": 2549 }, { "epoch": 0.2713343264524367, "grad_norm": 14.75, "learning_rate": 9.927160318362708e-06, "loss": 1.1985, "step": 2550 }, { "epoch": 0.27144073207065333, "grad_norm": 23.765625, "learning_rate": 9.92712478681069e-06, "loss": 1.6296, "step": 2551 }, { "epoch": 0.27154713768886996, "grad_norm": 20.28125, "learning_rate": 9.927089255258671e-06, "loss": 1.3061, "step": 2552 }, { "epoch": 0.27165354330708663, "grad_norm": 13.0625, "learning_rate": 9.927053723706653e-06, "loss": 1.1614, "step": 2553 }, { "epoch": 0.27175994892530325, "grad_norm": 18.140625, "learning_rate": 9.927018192154634e-06, "loss": 1.4271, "step": 2554 }, { "epoch": 0.2718663545435199, "grad_norm": 16.6875, "learning_rate": 9.926982660602616e-06, "loss": 1.2157, "step": 2555 }, { "epoch": 0.27197276016173655, "grad_norm": 16.046875, "learning_rate": 9.926947129050597e-06, "loss": 1.0855, "step": 2556 }, { "epoch": 0.27207916577995317, "grad_norm": 19.109375, "learning_rate": 9.926911597498579e-06, "loss": 1.5089, "step": 2557 }, { "epoch": 0.27218557139816985, "grad_norm": 21.3125, "learning_rate": 9.92687606594656e-06, "loss": 1.4702, "step": 2558 }, { "epoch": 0.27229197701638647, "grad_norm": 21.59375, "learning_rate": 9.926840534394542e-06, "loss": 1.3435, "step": 2559 }, { "epoch": 0.2723983826346031, "grad_norm": 15.890625, "learning_rate": 9.926805002842524e-06, "loss": 1.1226, "step": 2560 }, { "epoch": 0.27250478825281976, "grad_norm": 16.140625, "learning_rate": 9.926769471290507e-06, "loss": 1.1186, "step": 2561 }, { "epoch": 0.2726111938710364, "grad_norm": 17.25, "learning_rate": 9.926733939738489e-06, "loss": 1.3008, "step": 2562 }, { "epoch": 0.27271759948925306, "grad_norm": 18.140625, "learning_rate": 9.92669840818647e-06, "loss": 1.1604, "step": 2563 }, { "epoch": 0.2728240051074697, "grad_norm": 15.734375, "learning_rate": 9.926662876634452e-06, "loss": 1.1727, "step": 2564 }, { "epoch": 0.2729304107256863, "grad_norm": 20.546875, "learning_rate": 9.926627345082434e-06, "loss": 1.521, "step": 2565 }, { "epoch": 0.273036816343903, "grad_norm": 15.6171875, "learning_rate": 9.926591813530417e-06, "loss": 1.4534, "step": 2566 }, { "epoch": 0.2731432219621196, "grad_norm": 18.796875, "learning_rate": 9.926556281978399e-06, "loss": 1.1461, "step": 2567 }, { "epoch": 0.2732496275803362, "grad_norm": 15.4765625, "learning_rate": 9.92652075042638e-06, "loss": 1.3534, "step": 2568 }, { "epoch": 0.2733560331985529, "grad_norm": 19.828125, "learning_rate": 9.926485218874362e-06, "loss": 1.1071, "step": 2569 }, { "epoch": 0.2734624388167695, "grad_norm": 22.78125, "learning_rate": 9.926449687322343e-06, "loss": 1.6847, "step": 2570 }, { "epoch": 0.2735688444349862, "grad_norm": 13.4453125, "learning_rate": 9.926414155770325e-06, "loss": 1.4644, "step": 2571 }, { "epoch": 0.2736752500532028, "grad_norm": 14.5703125, "learning_rate": 9.926378624218307e-06, "loss": 1.0773, "step": 2572 }, { "epoch": 0.27378165567141943, "grad_norm": 16.5625, "learning_rate": 9.926343092666288e-06, "loss": 1.4125, "step": 2573 }, { "epoch": 0.2738880612896361, "grad_norm": 12.2578125, "learning_rate": 9.92630756111427e-06, "loss": 1.1318, "step": 2574 }, { "epoch": 0.27399446690785273, "grad_norm": 16.0625, "learning_rate": 9.926272029562251e-06, "loss": 1.2791, "step": 2575 }, { "epoch": 0.27410087252606935, "grad_norm": 16.71875, "learning_rate": 9.926236498010235e-06, "loss": 1.1267, "step": 2576 }, { "epoch": 0.274207278144286, "grad_norm": 16.03125, "learning_rate": 9.926200966458216e-06, "loss": 1.4865, "step": 2577 }, { "epoch": 0.27431368376250265, "grad_norm": 18.90625, "learning_rate": 9.926165434906198e-06, "loss": 1.584, "step": 2578 }, { "epoch": 0.2744200893807193, "grad_norm": 17.703125, "learning_rate": 9.92612990335418e-06, "loss": 1.3701, "step": 2579 }, { "epoch": 0.27452649499893594, "grad_norm": 15.9375, "learning_rate": 9.926094371802161e-06, "loss": 1.2486, "step": 2580 }, { "epoch": 0.27463290061715256, "grad_norm": 12.0234375, "learning_rate": 9.926058840250143e-06, "loss": 1.5095, "step": 2581 }, { "epoch": 0.27473930623536924, "grad_norm": 22.734375, "learning_rate": 9.926023308698124e-06, "loss": 1.9283, "step": 2582 }, { "epoch": 0.27484571185358586, "grad_norm": 15.390625, "learning_rate": 9.925987777146106e-06, "loss": 1.2176, "step": 2583 }, { "epoch": 0.27495211747180254, "grad_norm": 14.4140625, "learning_rate": 9.925952245594088e-06, "loss": 1.0901, "step": 2584 }, { "epoch": 0.27505852309001916, "grad_norm": 10.921875, "learning_rate": 9.925916714042069e-06, "loss": 1.1887, "step": 2585 }, { "epoch": 0.2751649287082358, "grad_norm": 15.2734375, "learning_rate": 9.925881182490052e-06, "loss": 1.2104, "step": 2586 }, { "epoch": 0.27527133432645245, "grad_norm": 21.625, "learning_rate": 9.925845650938034e-06, "loss": 1.2802, "step": 2587 }, { "epoch": 0.2753777399446691, "grad_norm": 9.875, "learning_rate": 9.925810119386016e-06, "loss": 0.8087, "step": 2588 }, { "epoch": 0.2754841455628857, "grad_norm": 13.8125, "learning_rate": 9.925774587833997e-06, "loss": 1.3426, "step": 2589 }, { "epoch": 0.2755905511811024, "grad_norm": 12.9375, "learning_rate": 9.925739056281979e-06, "loss": 1.4391, "step": 2590 }, { "epoch": 0.275696956799319, "grad_norm": 12.1953125, "learning_rate": 9.925703524729962e-06, "loss": 1.0951, "step": 2591 }, { "epoch": 0.27580336241753567, "grad_norm": 15.5625, "learning_rate": 9.925667993177944e-06, "loss": 1.5483, "step": 2592 }, { "epoch": 0.2759097680357523, "grad_norm": 26.703125, "learning_rate": 9.925632461625925e-06, "loss": 1.2495, "step": 2593 }, { "epoch": 0.2760161736539689, "grad_norm": 14.828125, "learning_rate": 9.925596930073907e-06, "loss": 1.4687, "step": 2594 }, { "epoch": 0.2761225792721856, "grad_norm": 15.5625, "learning_rate": 9.925561398521889e-06, "loss": 1.2146, "step": 2595 }, { "epoch": 0.2762289848904022, "grad_norm": 14.578125, "learning_rate": 9.92552586696987e-06, "loss": 1.3835, "step": 2596 }, { "epoch": 0.27633539050861883, "grad_norm": 17.265625, "learning_rate": 9.925490335417852e-06, "loss": 1.1475, "step": 2597 }, { "epoch": 0.2764417961268355, "grad_norm": 19.5625, "learning_rate": 9.925454803865833e-06, "loss": 1.4331, "step": 2598 }, { "epoch": 0.2765482017450521, "grad_norm": 18.796875, "learning_rate": 9.925419272313815e-06, "loss": 1.1304, "step": 2599 }, { "epoch": 0.2766546073632688, "grad_norm": 14.453125, "learning_rate": 9.925383740761797e-06, "loss": 1.22, "step": 2600 }, { "epoch": 0.2767610129814854, "grad_norm": 19.828125, "learning_rate": 9.925348209209778e-06, "loss": 1.2374, "step": 2601 }, { "epoch": 0.27686741859970204, "grad_norm": 17.046875, "learning_rate": 9.925312677657762e-06, "loss": 1.2665, "step": 2602 }, { "epoch": 0.2769738242179187, "grad_norm": 13.6875, "learning_rate": 9.925277146105743e-06, "loss": 0.8222, "step": 2603 }, { "epoch": 0.27708022983613534, "grad_norm": 19.546875, "learning_rate": 9.925241614553725e-06, "loss": 1.1685, "step": 2604 }, { "epoch": 0.277186635454352, "grad_norm": 15.1796875, "learning_rate": 9.925206083001706e-06, "loss": 1.5996, "step": 2605 }, { "epoch": 0.27729304107256864, "grad_norm": 23.046875, "learning_rate": 9.925170551449688e-06, "loss": 1.6215, "step": 2606 }, { "epoch": 0.27739944669078526, "grad_norm": 16.21875, "learning_rate": 9.92513501989767e-06, "loss": 1.4492, "step": 2607 }, { "epoch": 0.27750585230900193, "grad_norm": 10.125, "learning_rate": 9.925099488345651e-06, "loss": 1.0318, "step": 2608 }, { "epoch": 0.27761225792721855, "grad_norm": 42.21875, "learning_rate": 9.925063956793633e-06, "loss": 1.9236, "step": 2609 }, { "epoch": 0.2777186635454352, "grad_norm": 16.0, "learning_rate": 9.925028425241614e-06, "loss": 1.2069, "step": 2610 }, { "epoch": 0.27782506916365185, "grad_norm": 18.875, "learning_rate": 9.924992893689596e-06, "loss": 1.1573, "step": 2611 }, { "epoch": 0.27793147478186847, "grad_norm": 13.046875, "learning_rate": 9.92495736213758e-06, "loss": 1.116, "step": 2612 }, { "epoch": 0.27803788040008515, "grad_norm": 16.75, "learning_rate": 9.924921830585561e-06, "loss": 1.45, "step": 2613 }, { "epoch": 0.27814428601830177, "grad_norm": 17.046875, "learning_rate": 9.924886299033543e-06, "loss": 1.2431, "step": 2614 }, { "epoch": 0.2782506916365184, "grad_norm": 18.515625, "learning_rate": 9.924850767481524e-06, "loss": 1.1916, "step": 2615 }, { "epoch": 0.27835709725473506, "grad_norm": 13.2890625, "learning_rate": 9.924815235929506e-06, "loss": 1.1367, "step": 2616 }, { "epoch": 0.2784635028729517, "grad_norm": 15.25, "learning_rate": 9.924779704377489e-06, "loss": 1.2721, "step": 2617 }, { "epoch": 0.27856990849116836, "grad_norm": 14.7421875, "learning_rate": 9.92474417282547e-06, "loss": 1.3024, "step": 2618 }, { "epoch": 0.278676314109385, "grad_norm": 14.2265625, "learning_rate": 9.924708641273452e-06, "loss": 1.0335, "step": 2619 }, { "epoch": 0.2787827197276016, "grad_norm": 13.34375, "learning_rate": 9.924673109721434e-06, "loss": 1.3094, "step": 2620 }, { "epoch": 0.2788891253458183, "grad_norm": 14.2734375, "learning_rate": 9.924637578169415e-06, "loss": 1.2038, "step": 2621 }, { "epoch": 0.2789955309640349, "grad_norm": 16.375, "learning_rate": 9.924602046617397e-06, "loss": 1.283, "step": 2622 }, { "epoch": 0.2791019365822515, "grad_norm": 13.65625, "learning_rate": 9.924566515065379e-06, "loss": 1.2204, "step": 2623 }, { "epoch": 0.2792083422004682, "grad_norm": 15.4921875, "learning_rate": 9.92453098351336e-06, "loss": 1.195, "step": 2624 }, { "epoch": 0.2793147478186848, "grad_norm": 35.5, "learning_rate": 9.924495451961342e-06, "loss": 1.6155, "step": 2625 }, { "epoch": 0.2794211534369015, "grad_norm": 15.453125, "learning_rate": 9.924459920409323e-06, "loss": 1.1978, "step": 2626 }, { "epoch": 0.2795275590551181, "grad_norm": 13.4140625, "learning_rate": 9.924424388857307e-06, "loss": 1.2294, "step": 2627 }, { "epoch": 0.27963396467333473, "grad_norm": 14.015625, "learning_rate": 9.924388857305288e-06, "loss": 1.3057, "step": 2628 }, { "epoch": 0.2797403702915514, "grad_norm": 14.5078125, "learning_rate": 9.92435332575327e-06, "loss": 1.0812, "step": 2629 }, { "epoch": 0.27984677590976803, "grad_norm": 18.828125, "learning_rate": 9.924317794201252e-06, "loss": 1.2347, "step": 2630 }, { "epoch": 0.27995318152798465, "grad_norm": 14.609375, "learning_rate": 9.924282262649233e-06, "loss": 1.2124, "step": 2631 }, { "epoch": 0.2800595871462013, "grad_norm": 20.796875, "learning_rate": 9.924246731097215e-06, "loss": 1.6432, "step": 2632 }, { "epoch": 0.28016599276441795, "grad_norm": 13.2578125, "learning_rate": 9.924211199545198e-06, "loss": 1.4089, "step": 2633 }, { "epoch": 0.2802723983826346, "grad_norm": 11.1171875, "learning_rate": 9.92417566799318e-06, "loss": 1.3212, "step": 2634 }, { "epoch": 0.28037880400085124, "grad_norm": 14.1796875, "learning_rate": 9.92414013644116e-06, "loss": 1.5426, "step": 2635 }, { "epoch": 0.28048520961906787, "grad_norm": 12.78125, "learning_rate": 9.924104604889141e-06, "loss": 1.3069, "step": 2636 }, { "epoch": 0.28059161523728454, "grad_norm": 16.125, "learning_rate": 9.924069073337123e-06, "loss": 1.2034, "step": 2637 }, { "epoch": 0.28069802085550116, "grad_norm": 23.828125, "learning_rate": 9.924033541785106e-06, "loss": 1.2996, "step": 2638 }, { "epoch": 0.28080442647371784, "grad_norm": 15.4609375, "learning_rate": 9.923998010233088e-06, "loss": 1.0811, "step": 2639 }, { "epoch": 0.28091083209193446, "grad_norm": 14.4140625, "learning_rate": 9.92396247868107e-06, "loss": 1.55, "step": 2640 }, { "epoch": 0.2810172377101511, "grad_norm": 12.7109375, "learning_rate": 9.923926947129051e-06, "loss": 1.4344, "step": 2641 }, { "epoch": 0.28112364332836776, "grad_norm": 19.0, "learning_rate": 9.923891415577033e-06, "loss": 1.2862, "step": 2642 }, { "epoch": 0.2812300489465844, "grad_norm": 11.6015625, "learning_rate": 9.923855884025016e-06, "loss": 1.1024, "step": 2643 }, { "epoch": 0.281336454564801, "grad_norm": 11.5546875, "learning_rate": 9.923820352472997e-06, "loss": 0.862, "step": 2644 }, { "epoch": 0.2814428601830177, "grad_norm": 12.125, "learning_rate": 9.923784820920979e-06, "loss": 1.1884, "step": 2645 }, { "epoch": 0.2815492658012343, "grad_norm": 20.75, "learning_rate": 9.92374928936896e-06, "loss": 1.0061, "step": 2646 }, { "epoch": 0.28165567141945097, "grad_norm": 11.640625, "learning_rate": 9.923713757816942e-06, "loss": 1.2166, "step": 2647 }, { "epoch": 0.2817620770376676, "grad_norm": 23.765625, "learning_rate": 9.923678226264924e-06, "loss": 1.9556, "step": 2648 }, { "epoch": 0.2818684826558842, "grad_norm": 14.4921875, "learning_rate": 9.923642694712906e-06, "loss": 1.3339, "step": 2649 }, { "epoch": 0.2819748882741009, "grad_norm": 20.28125, "learning_rate": 9.923607163160887e-06, "loss": 1.1717, "step": 2650 }, { "epoch": 0.2820812938923175, "grad_norm": 17.78125, "learning_rate": 9.923571631608869e-06, "loss": 1.5953, "step": 2651 }, { "epoch": 0.28218769951053413, "grad_norm": 16.109375, "learning_rate": 9.92353610005685e-06, "loss": 1.1038, "step": 2652 }, { "epoch": 0.2822941051287508, "grad_norm": 22.59375, "learning_rate": 9.923500568504834e-06, "loss": 1.4029, "step": 2653 }, { "epoch": 0.2824005107469674, "grad_norm": 16.0625, "learning_rate": 9.923465036952815e-06, "loss": 1.3004, "step": 2654 }, { "epoch": 0.2825069163651841, "grad_norm": 15.2890625, "learning_rate": 9.923429505400797e-06, "loss": 1.0043, "step": 2655 }, { "epoch": 0.2826133219834007, "grad_norm": 13.7734375, "learning_rate": 9.923393973848778e-06, "loss": 1.2535, "step": 2656 }, { "epoch": 0.28271972760161734, "grad_norm": 15.2890625, "learning_rate": 9.92335844229676e-06, "loss": 1.2247, "step": 2657 }, { "epoch": 0.282826133219834, "grad_norm": 21.5625, "learning_rate": 9.923322910744743e-06, "loss": 1.5204, "step": 2658 }, { "epoch": 0.28293253883805064, "grad_norm": 14.0859375, "learning_rate": 9.923287379192725e-06, "loss": 1.4489, "step": 2659 }, { "epoch": 0.2830389444562673, "grad_norm": 17.5625, "learning_rate": 9.923251847640707e-06, "loss": 1.0261, "step": 2660 }, { "epoch": 0.28314535007448394, "grad_norm": 20.0, "learning_rate": 9.923216316088688e-06, "loss": 1.6812, "step": 2661 }, { "epoch": 0.28325175569270056, "grad_norm": 11.2265625, "learning_rate": 9.92318078453667e-06, "loss": 1.2168, "step": 2662 }, { "epoch": 0.28335816131091723, "grad_norm": 15.3203125, "learning_rate": 9.923145252984651e-06, "loss": 1.3292, "step": 2663 }, { "epoch": 0.28346456692913385, "grad_norm": 21.53125, "learning_rate": 9.923109721432633e-06, "loss": 1.603, "step": 2664 }, { "epoch": 0.2835709725473505, "grad_norm": 14.4765625, "learning_rate": 9.923074189880615e-06, "loss": 1.2101, "step": 2665 }, { "epoch": 0.28367737816556715, "grad_norm": 11.0859375, "learning_rate": 9.923038658328596e-06, "loss": 1.2025, "step": 2666 }, { "epoch": 0.28378378378378377, "grad_norm": 17.21875, "learning_rate": 9.923003126776578e-06, "loss": 1.6078, "step": 2667 }, { "epoch": 0.28389018940200045, "grad_norm": 15.7734375, "learning_rate": 9.922967595224561e-06, "loss": 1.0853, "step": 2668 }, { "epoch": 0.28399659502021707, "grad_norm": 13.9921875, "learning_rate": 9.922932063672543e-06, "loss": 1.3204, "step": 2669 }, { "epoch": 0.2841030006384337, "grad_norm": 13.0546875, "learning_rate": 9.922896532120524e-06, "loss": 1.2533, "step": 2670 }, { "epoch": 0.28420940625665037, "grad_norm": 15.28125, "learning_rate": 9.922861000568506e-06, "loss": 1.3701, "step": 2671 }, { "epoch": 0.284315811874867, "grad_norm": 14.109375, "learning_rate": 9.922825469016488e-06, "loss": 1.2275, "step": 2672 }, { "epoch": 0.28442221749308366, "grad_norm": 10.859375, "learning_rate": 9.92278993746447e-06, "loss": 1.1589, "step": 2673 }, { "epoch": 0.2845286231113003, "grad_norm": 15.796875, "learning_rate": 9.92275440591245e-06, "loss": 1.2733, "step": 2674 }, { "epoch": 0.2846350287295169, "grad_norm": 13.9296875, "learning_rate": 9.922718874360432e-06, "loss": 1.2676, "step": 2675 }, { "epoch": 0.2847414343477336, "grad_norm": 16.96875, "learning_rate": 9.922683342808414e-06, "loss": 1.1518, "step": 2676 }, { "epoch": 0.2848478399659502, "grad_norm": 13.3984375, "learning_rate": 9.922647811256396e-06, "loss": 1.1398, "step": 2677 }, { "epoch": 0.2849542455841668, "grad_norm": 20.875, "learning_rate": 9.922612279704377e-06, "loss": 1.5703, "step": 2678 }, { "epoch": 0.2850606512023835, "grad_norm": 16.65625, "learning_rate": 9.92257674815236e-06, "loss": 1.0581, "step": 2679 }, { "epoch": 0.2851670568206001, "grad_norm": 13.9765625, "learning_rate": 9.922541216600342e-06, "loss": 1.2102, "step": 2680 }, { "epoch": 0.2852734624388168, "grad_norm": 19.28125, "learning_rate": 9.922505685048324e-06, "loss": 1.6127, "step": 2681 }, { "epoch": 0.2853798680570334, "grad_norm": 15.9453125, "learning_rate": 9.922470153496305e-06, "loss": 1.4102, "step": 2682 }, { "epoch": 0.28548627367525004, "grad_norm": 14.84375, "learning_rate": 9.922434621944287e-06, "loss": 1.5077, "step": 2683 }, { "epoch": 0.2855926792934667, "grad_norm": 17.875, "learning_rate": 9.92239909039227e-06, "loss": 1.3405, "step": 2684 }, { "epoch": 0.28569908491168333, "grad_norm": 14.4140625, "learning_rate": 9.922363558840252e-06, "loss": 1.3254, "step": 2685 }, { "epoch": 0.28580549052989995, "grad_norm": 21.59375, "learning_rate": 9.922328027288233e-06, "loss": 1.3, "step": 2686 }, { "epoch": 0.28591189614811663, "grad_norm": 18.078125, "learning_rate": 9.922292495736215e-06, "loss": 1.2114, "step": 2687 }, { "epoch": 0.28601830176633325, "grad_norm": 47.6875, "learning_rate": 9.922256964184197e-06, "loss": 1.2366, "step": 2688 }, { "epoch": 0.2861247073845499, "grad_norm": 17.53125, "learning_rate": 9.922221432632178e-06, "loss": 1.1907, "step": 2689 }, { "epoch": 0.28623111300276655, "grad_norm": 13.9140625, "learning_rate": 9.92218590108016e-06, "loss": 1.062, "step": 2690 }, { "epoch": 0.28633751862098317, "grad_norm": 16.203125, "learning_rate": 9.922150369528141e-06, "loss": 1.1507, "step": 2691 }, { "epoch": 0.28644392423919984, "grad_norm": 14.46875, "learning_rate": 9.922114837976123e-06, "loss": 1.2197, "step": 2692 }, { "epoch": 0.28655032985741646, "grad_norm": 16.296875, "learning_rate": 9.922079306424105e-06, "loss": 1.6147, "step": 2693 }, { "epoch": 0.28665673547563314, "grad_norm": 19.515625, "learning_rate": 9.922043774872088e-06, "loss": 1.0409, "step": 2694 }, { "epoch": 0.28676314109384976, "grad_norm": 18.84375, "learning_rate": 9.92200824332007e-06, "loss": 1.2206, "step": 2695 }, { "epoch": 0.2868695467120664, "grad_norm": 15.9765625, "learning_rate": 9.921972711768051e-06, "loss": 0.9416, "step": 2696 }, { "epoch": 0.28697595233028306, "grad_norm": 14.890625, "learning_rate": 9.921937180216033e-06, "loss": 1.3963, "step": 2697 }, { "epoch": 0.2870823579484997, "grad_norm": 18.375, "learning_rate": 9.921901648664014e-06, "loss": 1.3335, "step": 2698 }, { "epoch": 0.2871887635667163, "grad_norm": 18.0, "learning_rate": 9.921866117111996e-06, "loss": 1.2591, "step": 2699 }, { "epoch": 0.287295169184933, "grad_norm": 10.578125, "learning_rate": 9.921830585559978e-06, "loss": 1.1624, "step": 2700 }, { "epoch": 0.2874015748031496, "grad_norm": 18.03125, "learning_rate": 9.92179505400796e-06, "loss": 0.9533, "step": 2701 }, { "epoch": 0.28750798042136627, "grad_norm": 17.265625, "learning_rate": 9.92175952245594e-06, "loss": 1.1661, "step": 2702 }, { "epoch": 0.2876143860395829, "grad_norm": 13.765625, "learning_rate": 9.921723990903922e-06, "loss": 1.1263, "step": 2703 }, { "epoch": 0.2877207916577995, "grad_norm": 27.125, "learning_rate": 9.921688459351906e-06, "loss": 1.0856, "step": 2704 }, { "epoch": 0.2878271972760162, "grad_norm": 20.46875, "learning_rate": 9.921652927799887e-06, "loss": 1.7609, "step": 2705 }, { "epoch": 0.2879336028942328, "grad_norm": 17.4375, "learning_rate": 9.921617396247869e-06, "loss": 1.4079, "step": 2706 }, { "epoch": 0.28804000851244943, "grad_norm": 17.53125, "learning_rate": 9.92158186469585e-06, "loss": 1.2983, "step": 2707 }, { "epoch": 0.2881464141306661, "grad_norm": 11.4296875, "learning_rate": 9.921546333143832e-06, "loss": 1.4511, "step": 2708 }, { "epoch": 0.2882528197488827, "grad_norm": 14.0234375, "learning_rate": 9.921510801591814e-06, "loss": 1.2807, "step": 2709 }, { "epoch": 0.2883592253670994, "grad_norm": 15.671875, "learning_rate": 9.921475270039797e-06, "loss": 1.1224, "step": 2710 }, { "epoch": 0.288465630985316, "grad_norm": 15.3359375, "learning_rate": 9.921439738487779e-06, "loss": 1.3375, "step": 2711 }, { "epoch": 0.28857203660353264, "grad_norm": 10.8984375, "learning_rate": 9.92140420693576e-06, "loss": 1.1454, "step": 2712 }, { "epoch": 0.2886784422217493, "grad_norm": 13.8203125, "learning_rate": 9.921368675383742e-06, "loss": 1.4443, "step": 2713 }, { "epoch": 0.28878484783996594, "grad_norm": 18.59375, "learning_rate": 9.921333143831724e-06, "loss": 1.1495, "step": 2714 }, { "epoch": 0.2888912534581826, "grad_norm": 11.8203125, "learning_rate": 9.921297612279705e-06, "loss": 1.2003, "step": 2715 }, { "epoch": 0.28899765907639924, "grad_norm": 17.9375, "learning_rate": 9.921262080727687e-06, "loss": 1.3085, "step": 2716 }, { "epoch": 0.28910406469461586, "grad_norm": 13.890625, "learning_rate": 9.921226549175668e-06, "loss": 1.0376, "step": 2717 }, { "epoch": 0.28921047031283253, "grad_norm": 15.1796875, "learning_rate": 9.92119101762365e-06, "loss": 1.1399, "step": 2718 }, { "epoch": 0.28931687593104916, "grad_norm": 17.515625, "learning_rate": 9.921155486071632e-06, "loss": 1.6867, "step": 2719 }, { "epoch": 0.2894232815492658, "grad_norm": 12.2421875, "learning_rate": 9.921119954519615e-06, "loss": 1.1037, "step": 2720 }, { "epoch": 0.28952968716748245, "grad_norm": 20.828125, "learning_rate": 9.921084422967596e-06, "loss": 1.4117, "step": 2721 }, { "epoch": 0.2896360927856991, "grad_norm": 10.6875, "learning_rate": 9.921048891415578e-06, "loss": 1.0673, "step": 2722 }, { "epoch": 0.28974249840391575, "grad_norm": 12.2421875, "learning_rate": 9.92101335986356e-06, "loss": 1.2415, "step": 2723 }, { "epoch": 0.28984890402213237, "grad_norm": 14.8203125, "learning_rate": 9.920977828311541e-06, "loss": 1.3545, "step": 2724 }, { "epoch": 0.289955309640349, "grad_norm": 21.796875, "learning_rate": 9.920942296759523e-06, "loss": 1.7285, "step": 2725 }, { "epoch": 0.29006171525856567, "grad_norm": 19.359375, "learning_rate": 9.920906765207504e-06, "loss": 1.7742, "step": 2726 }, { "epoch": 0.2901681208767823, "grad_norm": 13.2109375, "learning_rate": 9.920871233655486e-06, "loss": 1.1615, "step": 2727 }, { "epoch": 0.29027452649499896, "grad_norm": 17.859375, "learning_rate": 9.920835702103468e-06, "loss": 1.5431, "step": 2728 }, { "epoch": 0.2903809321132156, "grad_norm": 15.84375, "learning_rate": 9.92080017055145e-06, "loss": 1.2942, "step": 2729 }, { "epoch": 0.2904873377314322, "grad_norm": 13.4609375, "learning_rate": 9.920764638999433e-06, "loss": 1.3968, "step": 2730 }, { "epoch": 0.2905937433496489, "grad_norm": 15.15625, "learning_rate": 9.920729107447414e-06, "loss": 1.4578, "step": 2731 }, { "epoch": 0.2907001489678655, "grad_norm": 14.4140625, "learning_rate": 9.920693575895396e-06, "loss": 1.459, "step": 2732 }, { "epoch": 0.2908065545860821, "grad_norm": 15.3046875, "learning_rate": 9.920658044343377e-06, "loss": 1.1799, "step": 2733 }, { "epoch": 0.2909129602042988, "grad_norm": 16.859375, "learning_rate": 9.920622512791359e-06, "loss": 1.5055, "step": 2734 }, { "epoch": 0.2910193658225154, "grad_norm": 16.203125, "learning_rate": 9.920586981239342e-06, "loss": 1.8555, "step": 2735 }, { "epoch": 0.2911257714407321, "grad_norm": 24.734375, "learning_rate": 9.920551449687324e-06, "loss": 1.3266, "step": 2736 }, { "epoch": 0.2912321770589487, "grad_norm": 16.59375, "learning_rate": 9.920515918135306e-06, "loss": 1.3257, "step": 2737 }, { "epoch": 0.29133858267716534, "grad_norm": 14.4375, "learning_rate": 9.920480386583287e-06, "loss": 1.3978, "step": 2738 }, { "epoch": 0.291444988295382, "grad_norm": 17.359375, "learning_rate": 9.920444855031269e-06, "loss": 1.3851, "step": 2739 }, { "epoch": 0.29155139391359863, "grad_norm": 14.2109375, "learning_rate": 9.92040932347925e-06, "loss": 1.2829, "step": 2740 }, { "epoch": 0.29165779953181525, "grad_norm": 19.75, "learning_rate": 9.920373791927232e-06, "loss": 1.5165, "step": 2741 }, { "epoch": 0.29176420515003193, "grad_norm": 12.2265625, "learning_rate": 9.920338260375214e-06, "loss": 1.0158, "step": 2742 }, { "epoch": 0.29187061076824855, "grad_norm": 13.3828125, "learning_rate": 9.920302728823195e-06, "loss": 1.0647, "step": 2743 }, { "epoch": 0.2919770163864652, "grad_norm": 17.5625, "learning_rate": 9.920267197271177e-06, "loss": 1.4369, "step": 2744 }, { "epoch": 0.29208342200468185, "grad_norm": 17.4375, "learning_rate": 9.92023166571916e-06, "loss": 1.402, "step": 2745 }, { "epoch": 0.29218982762289847, "grad_norm": 15.703125, "learning_rate": 9.920196134167142e-06, "loss": 1.3114, "step": 2746 }, { "epoch": 0.29229623324111514, "grad_norm": 18.296875, "learning_rate": 9.920160602615123e-06, "loss": 1.8869, "step": 2747 }, { "epoch": 0.29240263885933176, "grad_norm": 14.171875, "learning_rate": 9.920125071063105e-06, "loss": 1.5893, "step": 2748 }, { "epoch": 0.29250904447754844, "grad_norm": 17.109375, "learning_rate": 9.920089539511087e-06, "loss": 1.2188, "step": 2749 }, { "epoch": 0.29261545009576506, "grad_norm": 15.09375, "learning_rate": 9.920054007959068e-06, "loss": 1.3674, "step": 2750 }, { "epoch": 0.2927218557139817, "grad_norm": 18.8125, "learning_rate": 9.920018476407051e-06, "loss": 1.6566, "step": 2751 }, { "epoch": 0.29282826133219836, "grad_norm": 15.796875, "learning_rate": 9.919982944855033e-06, "loss": 1.1904, "step": 2752 }, { "epoch": 0.292934666950415, "grad_norm": 15.703125, "learning_rate": 9.919947413303015e-06, "loss": 1.1673, "step": 2753 }, { "epoch": 0.2930410725686316, "grad_norm": 17.109375, "learning_rate": 9.919911881750996e-06, "loss": 1.3895, "step": 2754 }, { "epoch": 0.2931474781868483, "grad_norm": 19.5625, "learning_rate": 9.919876350198976e-06, "loss": 1.8536, "step": 2755 }, { "epoch": 0.2932538838050649, "grad_norm": 15.0703125, "learning_rate": 9.91984081864696e-06, "loss": 1.3627, "step": 2756 }, { "epoch": 0.2933602894232816, "grad_norm": 11.34375, "learning_rate": 9.919805287094941e-06, "loss": 1.1785, "step": 2757 }, { "epoch": 0.2934666950414982, "grad_norm": 17.25, "learning_rate": 9.919769755542923e-06, "loss": 1.4287, "step": 2758 }, { "epoch": 0.2935731006597148, "grad_norm": 14.7421875, "learning_rate": 9.919734223990904e-06, "loss": 1.5896, "step": 2759 }, { "epoch": 0.2936795062779315, "grad_norm": 17.15625, "learning_rate": 9.919698692438886e-06, "loss": 1.1739, "step": 2760 }, { "epoch": 0.2937859118961481, "grad_norm": 13.6328125, "learning_rate": 9.91966316088687e-06, "loss": 1.2354, "step": 2761 }, { "epoch": 0.29389231751436473, "grad_norm": 21.234375, "learning_rate": 9.91962762933485e-06, "loss": 1.5601, "step": 2762 }, { "epoch": 0.2939987231325814, "grad_norm": 11.65625, "learning_rate": 9.919592097782832e-06, "loss": 1.0369, "step": 2763 }, { "epoch": 0.29410512875079803, "grad_norm": 13.671875, "learning_rate": 9.919556566230814e-06, "loss": 1.3973, "step": 2764 }, { "epoch": 0.2942115343690147, "grad_norm": 15.859375, "learning_rate": 9.919521034678796e-06, "loss": 1.3686, "step": 2765 }, { "epoch": 0.2943179399872313, "grad_norm": 16.03125, "learning_rate": 9.919485503126777e-06, "loss": 1.2375, "step": 2766 }, { "epoch": 0.29442434560544795, "grad_norm": 16.09375, "learning_rate": 9.919449971574759e-06, "loss": 1.5433, "step": 2767 }, { "epoch": 0.2945307512236646, "grad_norm": 14.015625, "learning_rate": 9.91941444002274e-06, "loss": 1.3695, "step": 2768 }, { "epoch": 0.29463715684188124, "grad_norm": 17.171875, "learning_rate": 9.919378908470722e-06, "loss": 1.4257, "step": 2769 }, { "epoch": 0.2947435624600979, "grad_norm": 16.140625, "learning_rate": 9.919343376918704e-06, "loss": 1.0221, "step": 2770 }, { "epoch": 0.29484996807831454, "grad_norm": 15.484375, "learning_rate": 9.919307845366687e-06, "loss": 1.3226, "step": 2771 }, { "epoch": 0.29495637369653116, "grad_norm": 20.171875, "learning_rate": 9.919272313814669e-06, "loss": 1.213, "step": 2772 }, { "epoch": 0.29506277931474784, "grad_norm": 17.078125, "learning_rate": 9.91923678226265e-06, "loss": 1.2728, "step": 2773 }, { "epoch": 0.29516918493296446, "grad_norm": 11.828125, "learning_rate": 9.919201250710632e-06, "loss": 1.2605, "step": 2774 }, { "epoch": 0.2952755905511811, "grad_norm": 19.25, "learning_rate": 9.919165719158613e-06, "loss": 1.639, "step": 2775 }, { "epoch": 0.29538199616939775, "grad_norm": 17.09375, "learning_rate": 9.919130187606597e-06, "loss": 1.558, "step": 2776 }, { "epoch": 0.2954884017876144, "grad_norm": 16.375, "learning_rate": 9.919094656054578e-06, "loss": 1.2502, "step": 2777 }, { "epoch": 0.29559480740583105, "grad_norm": 21.25, "learning_rate": 9.91905912450256e-06, "loss": 1.1572, "step": 2778 }, { "epoch": 0.29570121302404767, "grad_norm": 18.890625, "learning_rate": 9.919023592950541e-06, "loss": 1.2898, "step": 2779 }, { "epoch": 0.2958076186422643, "grad_norm": 17.5, "learning_rate": 9.918988061398523e-06, "loss": 1.4328, "step": 2780 }, { "epoch": 0.29591402426048097, "grad_norm": 18.53125, "learning_rate": 9.918952529846505e-06, "loss": 1.7554, "step": 2781 }, { "epoch": 0.2960204298786976, "grad_norm": 13.03125, "learning_rate": 9.918916998294486e-06, "loss": 1.1561, "step": 2782 }, { "epoch": 0.29612683549691426, "grad_norm": 23.09375, "learning_rate": 9.918881466742468e-06, "loss": 1.5307, "step": 2783 }, { "epoch": 0.2962332411151309, "grad_norm": 18.265625, "learning_rate": 9.91884593519045e-06, "loss": 1.3118, "step": 2784 }, { "epoch": 0.2963396467333475, "grad_norm": 14.1953125, "learning_rate": 9.918810403638431e-06, "loss": 1.2421, "step": 2785 }, { "epoch": 0.2964460523515642, "grad_norm": 21.515625, "learning_rate": 9.918774872086414e-06, "loss": 1.3854, "step": 2786 }, { "epoch": 0.2965524579697808, "grad_norm": 20.609375, "learning_rate": 9.918739340534396e-06, "loss": 1.3203, "step": 2787 }, { "epoch": 0.2966588635879974, "grad_norm": 12.8671875, "learning_rate": 9.918703808982378e-06, "loss": 1.0792, "step": 2788 }, { "epoch": 0.2967652692062141, "grad_norm": 13.71875, "learning_rate": 9.91866827743036e-06, "loss": 1.2382, "step": 2789 }, { "epoch": 0.2968716748244307, "grad_norm": 9.6796875, "learning_rate": 9.918632745878341e-06, "loss": 0.8981, "step": 2790 }, { "epoch": 0.2969780804426474, "grad_norm": 16.875, "learning_rate": 9.918597214326322e-06, "loss": 1.4326, "step": 2791 }, { "epoch": 0.297084486060864, "grad_norm": 15.9609375, "learning_rate": 9.918561682774304e-06, "loss": 1.0096, "step": 2792 }, { "epoch": 0.29719089167908064, "grad_norm": 13.796875, "learning_rate": 9.918526151222286e-06, "loss": 1.1278, "step": 2793 }, { "epoch": 0.2972972972972973, "grad_norm": 11.65625, "learning_rate": 9.918490619670267e-06, "loss": 1.0425, "step": 2794 }, { "epoch": 0.29740370291551393, "grad_norm": 18.609375, "learning_rate": 9.918455088118249e-06, "loss": 1.2945, "step": 2795 }, { "epoch": 0.29751010853373056, "grad_norm": 19.0, "learning_rate": 9.91841955656623e-06, "loss": 1.4395, "step": 2796 }, { "epoch": 0.29761651415194723, "grad_norm": 15.03125, "learning_rate": 9.918384025014214e-06, "loss": 0.979, "step": 2797 }, { "epoch": 0.29772291977016385, "grad_norm": 14.9296875, "learning_rate": 9.918348493462195e-06, "loss": 1.4545, "step": 2798 }, { "epoch": 0.29782932538838053, "grad_norm": 12.9375, "learning_rate": 9.918312961910177e-06, "loss": 1.1065, "step": 2799 }, { "epoch": 0.29793573100659715, "grad_norm": 15.96875, "learning_rate": 9.918277430358159e-06, "loss": 1.2577, "step": 2800 }, { "epoch": 0.29804213662481377, "grad_norm": 15.921875, "learning_rate": 9.91824189880614e-06, "loss": 1.4784, "step": 2801 }, { "epoch": 0.29814854224303045, "grad_norm": 23.203125, "learning_rate": 9.918206367254124e-06, "loss": 1.1376, "step": 2802 }, { "epoch": 0.29825494786124707, "grad_norm": 25.609375, "learning_rate": 9.918170835702105e-06, "loss": 2.0421, "step": 2803 }, { "epoch": 0.29836135347946374, "grad_norm": 15.296875, "learning_rate": 9.918135304150087e-06, "loss": 1.594, "step": 2804 }, { "epoch": 0.29846775909768036, "grad_norm": 14.5625, "learning_rate": 9.918099772598068e-06, "loss": 1.3902, "step": 2805 }, { "epoch": 0.298574164715897, "grad_norm": 11.84375, "learning_rate": 9.91806424104605e-06, "loss": 1.0595, "step": 2806 }, { "epoch": 0.29868057033411366, "grad_norm": 16.421875, "learning_rate": 9.918028709494032e-06, "loss": 1.0743, "step": 2807 }, { "epoch": 0.2987869759523303, "grad_norm": 14.015625, "learning_rate": 9.917993177942013e-06, "loss": 1.2441, "step": 2808 }, { "epoch": 0.2988933815705469, "grad_norm": 16.765625, "learning_rate": 9.917957646389995e-06, "loss": 1.2871, "step": 2809 }, { "epoch": 0.2989997871887636, "grad_norm": 12.7109375, "learning_rate": 9.917922114837976e-06, "loss": 0.9903, "step": 2810 }, { "epoch": 0.2991061928069802, "grad_norm": 11.1328125, "learning_rate": 9.917886583285958e-06, "loss": 1.0486, "step": 2811 }, { "epoch": 0.2992125984251969, "grad_norm": 17.515625, "learning_rate": 9.917851051733941e-06, "loss": 1.16, "step": 2812 }, { "epoch": 0.2993190040434135, "grad_norm": 15.625, "learning_rate": 9.917815520181923e-06, "loss": 1.5356, "step": 2813 }, { "epoch": 0.2994254096616301, "grad_norm": 11.890625, "learning_rate": 9.917779988629904e-06, "loss": 1.2738, "step": 2814 }, { "epoch": 0.2995318152798468, "grad_norm": 16.3125, "learning_rate": 9.917744457077886e-06, "loss": 1.2127, "step": 2815 }, { "epoch": 0.2996382208980634, "grad_norm": 12.640625, "learning_rate": 9.917708925525868e-06, "loss": 1.0324, "step": 2816 }, { "epoch": 0.29974462651628003, "grad_norm": 15.3125, "learning_rate": 9.91767339397385e-06, "loss": 1.2869, "step": 2817 }, { "epoch": 0.2998510321344967, "grad_norm": 9.453125, "learning_rate": 9.917637862421831e-06, "loss": 1.0877, "step": 2818 }, { "epoch": 0.29995743775271333, "grad_norm": 38.4375, "learning_rate": 9.917602330869813e-06, "loss": 1.5832, "step": 2819 }, { "epoch": 0.30006384337093, "grad_norm": 14.4375, "learning_rate": 9.917566799317794e-06, "loss": 1.4879, "step": 2820 }, { "epoch": 0.3001702489891466, "grad_norm": 13.7421875, "learning_rate": 9.917531267765776e-06, "loss": 1.4186, "step": 2821 }, { "epoch": 0.30027665460736325, "grad_norm": 27.921875, "learning_rate": 9.917495736213759e-06, "loss": 1.4509, "step": 2822 }, { "epoch": 0.3003830602255799, "grad_norm": 19.890625, "learning_rate": 9.91746020466174e-06, "loss": 1.5902, "step": 2823 }, { "epoch": 0.30048946584379654, "grad_norm": 12.0078125, "learning_rate": 9.917424673109722e-06, "loss": 1.0162, "step": 2824 }, { "epoch": 0.3005958714620132, "grad_norm": 10.5859375, "learning_rate": 9.917389141557704e-06, "loss": 0.9709, "step": 2825 }, { "epoch": 0.30070227708022984, "grad_norm": 14.5546875, "learning_rate": 9.917353610005685e-06, "loss": 1.2951, "step": 2826 }, { "epoch": 0.30080868269844646, "grad_norm": 18.78125, "learning_rate": 9.917318078453667e-06, "loss": 1.4605, "step": 2827 }, { "epoch": 0.30091508831666314, "grad_norm": 19.203125, "learning_rate": 9.91728254690165e-06, "loss": 1.4154, "step": 2828 }, { "epoch": 0.30102149393487976, "grad_norm": 13.6796875, "learning_rate": 9.917247015349632e-06, "loss": 1.3829, "step": 2829 }, { "epoch": 0.3011278995530964, "grad_norm": 12.1953125, "learning_rate": 9.917211483797614e-06, "loss": 1.1411, "step": 2830 }, { "epoch": 0.30123430517131305, "grad_norm": 13.40625, "learning_rate": 9.917175952245595e-06, "loss": 1.5629, "step": 2831 }, { "epoch": 0.3013407107895297, "grad_norm": 12.84375, "learning_rate": 9.917140420693577e-06, "loss": 1.2135, "step": 2832 }, { "epoch": 0.30144711640774635, "grad_norm": 15.03125, "learning_rate": 9.917104889141558e-06, "loss": 1.4102, "step": 2833 }, { "epoch": 0.301553522025963, "grad_norm": 13.2578125, "learning_rate": 9.91706935758954e-06, "loss": 1.3248, "step": 2834 }, { "epoch": 0.3016599276441796, "grad_norm": 12.4375, "learning_rate": 9.917033826037522e-06, "loss": 1.2725, "step": 2835 }, { "epoch": 0.30176633326239627, "grad_norm": 13.0546875, "learning_rate": 9.916998294485503e-06, "loss": 1.1411, "step": 2836 }, { "epoch": 0.3018727388806129, "grad_norm": 17.6875, "learning_rate": 9.916962762933485e-06, "loss": 1.4128, "step": 2837 }, { "epoch": 0.30197914449882957, "grad_norm": 15.328125, "learning_rate": 9.916927231381468e-06, "loss": 1.4625, "step": 2838 }, { "epoch": 0.3020855501170462, "grad_norm": 15.8828125, "learning_rate": 9.91689169982945e-06, "loss": 1.4746, "step": 2839 }, { "epoch": 0.3021919557352628, "grad_norm": 13.4609375, "learning_rate": 9.916856168277431e-06, "loss": 1.0077, "step": 2840 }, { "epoch": 0.3022983613534795, "grad_norm": 12.40625, "learning_rate": 9.916820636725413e-06, "loss": 1.1825, "step": 2841 }, { "epoch": 0.3024047669716961, "grad_norm": 20.625, "learning_rate": 9.916785105173395e-06, "loss": 1.5009, "step": 2842 }, { "epoch": 0.3025111725899127, "grad_norm": 17.421875, "learning_rate": 9.916749573621378e-06, "loss": 1.0958, "step": 2843 }, { "epoch": 0.3026175782081294, "grad_norm": 13.5234375, "learning_rate": 9.916714042069358e-06, "loss": 1.4244, "step": 2844 }, { "epoch": 0.302723983826346, "grad_norm": 20.25, "learning_rate": 9.91667851051734e-06, "loss": 1.1844, "step": 2845 }, { "epoch": 0.3028303894445627, "grad_norm": 11.40625, "learning_rate": 9.916642978965321e-06, "loss": 1.2014, "step": 2846 }, { "epoch": 0.3029367950627793, "grad_norm": 19.328125, "learning_rate": 9.916607447413303e-06, "loss": 1.3784, "step": 2847 }, { "epoch": 0.30304320068099594, "grad_norm": 21.734375, "learning_rate": 9.916571915861286e-06, "loss": 1.3208, "step": 2848 }, { "epoch": 0.3031496062992126, "grad_norm": 14.7421875, "learning_rate": 9.916536384309268e-06, "loss": 1.2722, "step": 2849 }, { "epoch": 0.30325601191742924, "grad_norm": 13.21875, "learning_rate": 9.916500852757249e-06, "loss": 1.303, "step": 2850 }, { "epoch": 0.30336241753564586, "grad_norm": 17.625, "learning_rate": 9.91646532120523e-06, "loss": 1.6788, "step": 2851 }, { "epoch": 0.30346882315386253, "grad_norm": 14.7265625, "learning_rate": 9.916429789653212e-06, "loss": 1.2919, "step": 2852 }, { "epoch": 0.30357522877207915, "grad_norm": 12.84375, "learning_rate": 9.916394258101196e-06, "loss": 1.2463, "step": 2853 }, { "epoch": 0.30368163439029583, "grad_norm": 11.25, "learning_rate": 9.916358726549177e-06, "loss": 1.0297, "step": 2854 }, { "epoch": 0.30378804000851245, "grad_norm": 12.3125, "learning_rate": 9.916323194997159e-06, "loss": 1.1293, "step": 2855 }, { "epoch": 0.30389444562672907, "grad_norm": 19.703125, "learning_rate": 9.91628766344514e-06, "loss": 1.4652, "step": 2856 }, { "epoch": 0.30400085124494575, "grad_norm": 21.0, "learning_rate": 9.916252131893122e-06, "loss": 1.3792, "step": 2857 }, { "epoch": 0.30410725686316237, "grad_norm": 20.125, "learning_rate": 9.916216600341104e-06, "loss": 1.3024, "step": 2858 }, { "epoch": 0.30421366248137904, "grad_norm": 13.96875, "learning_rate": 9.916181068789085e-06, "loss": 1.0482, "step": 2859 }, { "epoch": 0.30432006809959566, "grad_norm": 15.359375, "learning_rate": 9.916145537237067e-06, "loss": 1.3122, "step": 2860 }, { "epoch": 0.3044264737178123, "grad_norm": 15.4453125, "learning_rate": 9.916110005685048e-06, "loss": 1.2996, "step": 2861 }, { "epoch": 0.30453287933602896, "grad_norm": 14.9296875, "learning_rate": 9.91607447413303e-06, "loss": 1.3303, "step": 2862 }, { "epoch": 0.3046392849542456, "grad_norm": 17.515625, "learning_rate": 9.916038942581013e-06, "loss": 1.3084, "step": 2863 }, { "epoch": 0.3047456905724622, "grad_norm": 17.6875, "learning_rate": 9.916003411028995e-06, "loss": 1.294, "step": 2864 }, { "epoch": 0.3048520961906789, "grad_norm": 15.453125, "learning_rate": 9.915967879476977e-06, "loss": 1.3046, "step": 2865 }, { "epoch": 0.3049585018088955, "grad_norm": 16.8125, "learning_rate": 9.915932347924958e-06, "loss": 1.3358, "step": 2866 }, { "epoch": 0.3050649074271122, "grad_norm": 14.1171875, "learning_rate": 9.91589681637294e-06, "loss": 1.429, "step": 2867 }, { "epoch": 0.3051713130453288, "grad_norm": 14.2734375, "learning_rate": 9.915861284820921e-06, "loss": 1.3642, "step": 2868 }, { "epoch": 0.3052777186635454, "grad_norm": 12.921875, "learning_rate": 9.915825753268905e-06, "loss": 1.1609, "step": 2869 }, { "epoch": 0.3053841242817621, "grad_norm": 25.140625, "learning_rate": 9.915790221716886e-06, "loss": 1.4673, "step": 2870 }, { "epoch": 0.3054905298999787, "grad_norm": 11.8203125, "learning_rate": 9.915754690164868e-06, "loss": 1.2736, "step": 2871 }, { "epoch": 0.30559693551819533, "grad_norm": 13.296875, "learning_rate": 9.91571915861285e-06, "loss": 1.2794, "step": 2872 }, { "epoch": 0.305703341136412, "grad_norm": 15.6875, "learning_rate": 9.915683627060831e-06, "loss": 1.4222, "step": 2873 }, { "epoch": 0.30580974675462863, "grad_norm": 15.609375, "learning_rate": 9.915648095508813e-06, "loss": 1.1266, "step": 2874 }, { "epoch": 0.3059161523728453, "grad_norm": 12.578125, "learning_rate": 9.915612563956794e-06, "loss": 1.3796, "step": 2875 }, { "epoch": 0.3060225579910619, "grad_norm": 18.40625, "learning_rate": 9.915577032404776e-06, "loss": 1.481, "step": 2876 }, { "epoch": 0.30612896360927855, "grad_norm": 11.7265625, "learning_rate": 9.915541500852758e-06, "loss": 1.3427, "step": 2877 }, { "epoch": 0.3062353692274952, "grad_norm": 12.5390625, "learning_rate": 9.91550596930074e-06, "loss": 1.2174, "step": 2878 }, { "epoch": 0.30634177484571185, "grad_norm": 14.5234375, "learning_rate": 9.915470437748722e-06, "loss": 1.3583, "step": 2879 }, { "epoch": 0.3064481804639285, "grad_norm": 13.3046875, "learning_rate": 9.915434906196704e-06, "loss": 1.1278, "step": 2880 }, { "epoch": 0.30655458608214514, "grad_norm": 14.2265625, "learning_rate": 9.915399374644686e-06, "loss": 1.1118, "step": 2881 }, { "epoch": 0.30666099170036176, "grad_norm": 13.2734375, "learning_rate": 9.915363843092667e-06, "loss": 1.0677, "step": 2882 }, { "epoch": 0.30676739731857844, "grad_norm": 10.703125, "learning_rate": 9.915328311540649e-06, "loss": 1.0622, "step": 2883 }, { "epoch": 0.30687380293679506, "grad_norm": 13.484375, "learning_rate": 9.91529277998863e-06, "loss": 1.027, "step": 2884 }, { "epoch": 0.3069802085550117, "grad_norm": 13.59375, "learning_rate": 9.915257248436612e-06, "loss": 1.1178, "step": 2885 }, { "epoch": 0.30708661417322836, "grad_norm": 17.53125, "learning_rate": 9.915221716884594e-06, "loss": 1.508, "step": 2886 }, { "epoch": 0.307193019791445, "grad_norm": 16.046875, "learning_rate": 9.915186185332575e-06, "loss": 1.3042, "step": 2887 }, { "epoch": 0.30729942540966165, "grad_norm": 16.671875, "learning_rate": 9.915150653780557e-06, "loss": 1.3396, "step": 2888 }, { "epoch": 0.3074058310278783, "grad_norm": 13.875, "learning_rate": 9.91511512222854e-06, "loss": 1.0226, "step": 2889 }, { "epoch": 0.3075122366460949, "grad_norm": 12.5078125, "learning_rate": 9.915079590676522e-06, "loss": 1.3597, "step": 2890 }, { "epoch": 0.30761864226431157, "grad_norm": 17.71875, "learning_rate": 9.915044059124503e-06, "loss": 1.1038, "step": 2891 }, { "epoch": 0.3077250478825282, "grad_norm": 16.078125, "learning_rate": 9.915008527572485e-06, "loss": 1.266, "step": 2892 }, { "epoch": 0.3078314535007448, "grad_norm": 16.234375, "learning_rate": 9.914972996020467e-06, "loss": 1.2962, "step": 2893 }, { "epoch": 0.3079378591189615, "grad_norm": 16.875, "learning_rate": 9.91493746446845e-06, "loss": 1.4963, "step": 2894 }, { "epoch": 0.3080442647371781, "grad_norm": 18.390625, "learning_rate": 9.914901932916432e-06, "loss": 1.377, "step": 2895 }, { "epoch": 0.3081506703553948, "grad_norm": 14.5625, "learning_rate": 9.914866401364413e-06, "loss": 1.2118, "step": 2896 }, { "epoch": 0.3082570759736114, "grad_norm": 13.6328125, "learning_rate": 9.914830869812395e-06, "loss": 1.3506, "step": 2897 }, { "epoch": 0.308363481591828, "grad_norm": 14.578125, "learning_rate": 9.914795338260376e-06, "loss": 1.1657, "step": 2898 }, { "epoch": 0.3084698872100447, "grad_norm": 18.703125, "learning_rate": 9.914759806708358e-06, "loss": 1.3962, "step": 2899 }, { "epoch": 0.3085762928282613, "grad_norm": 18.609375, "learning_rate": 9.91472427515634e-06, "loss": 1.2885, "step": 2900 }, { "epoch": 0.308682698446478, "grad_norm": 17.359375, "learning_rate": 9.914688743604321e-06, "loss": 1.5057, "step": 2901 }, { "epoch": 0.3087891040646946, "grad_norm": 11.8359375, "learning_rate": 9.914653212052303e-06, "loss": 0.9754, "step": 2902 }, { "epoch": 0.30889550968291124, "grad_norm": 19.75, "learning_rate": 9.914617680500284e-06, "loss": 1.3994, "step": 2903 }, { "epoch": 0.3090019153011279, "grad_norm": 16.84375, "learning_rate": 9.914582148948266e-06, "loss": 1.2485, "step": 2904 }, { "epoch": 0.30910832091934454, "grad_norm": 13.3046875, "learning_rate": 9.91454661739625e-06, "loss": 1.2209, "step": 2905 }, { "epoch": 0.30921472653756116, "grad_norm": 14.0, "learning_rate": 9.914511085844231e-06, "loss": 1.3905, "step": 2906 }, { "epoch": 0.30932113215577783, "grad_norm": 12.703125, "learning_rate": 9.914475554292213e-06, "loss": 1.114, "step": 2907 }, { "epoch": 0.30942753777399445, "grad_norm": 15.9140625, "learning_rate": 9.914440022740194e-06, "loss": 1.4628, "step": 2908 }, { "epoch": 0.30953394339221113, "grad_norm": 15.875, "learning_rate": 9.914404491188176e-06, "loss": 1.3305, "step": 2909 }, { "epoch": 0.30964034901042775, "grad_norm": 11.640625, "learning_rate": 9.914368959636157e-06, "loss": 1.1142, "step": 2910 }, { "epoch": 0.30974675462864437, "grad_norm": 17.5625, "learning_rate": 9.914333428084139e-06, "loss": 1.0627, "step": 2911 }, { "epoch": 0.30985316024686105, "grad_norm": 15.1953125, "learning_rate": 9.91429789653212e-06, "loss": 1.3452, "step": 2912 }, { "epoch": 0.30995956586507767, "grad_norm": 15.34375, "learning_rate": 9.914262364980102e-06, "loss": 1.4047, "step": 2913 }, { "epoch": 0.31006597148329434, "grad_norm": 12.84375, "learning_rate": 9.914226833428084e-06, "loss": 1.3384, "step": 2914 }, { "epoch": 0.31017237710151097, "grad_norm": 11.9453125, "learning_rate": 9.914191301876067e-06, "loss": 1.2848, "step": 2915 }, { "epoch": 0.3102787827197276, "grad_norm": 13.8046875, "learning_rate": 9.914155770324049e-06, "loss": 1.4642, "step": 2916 }, { "epoch": 0.31038518833794426, "grad_norm": 18.15625, "learning_rate": 9.91412023877203e-06, "loss": 1.0603, "step": 2917 }, { "epoch": 0.3104915939561609, "grad_norm": 11.3515625, "learning_rate": 9.914084707220012e-06, "loss": 0.9887, "step": 2918 }, { "epoch": 0.3105979995743775, "grad_norm": 15.859375, "learning_rate": 9.914049175667994e-06, "loss": 1.3133, "step": 2919 }, { "epoch": 0.3107044051925942, "grad_norm": 9.34375, "learning_rate": 9.914013644115977e-06, "loss": 1.048, "step": 2920 }, { "epoch": 0.3108108108108108, "grad_norm": 13.984375, "learning_rate": 9.913978112563958e-06, "loss": 1.3666, "step": 2921 }, { "epoch": 0.3109172164290275, "grad_norm": 12.859375, "learning_rate": 9.91394258101194e-06, "loss": 1.151, "step": 2922 }, { "epoch": 0.3110236220472441, "grad_norm": 14.875, "learning_rate": 9.913907049459922e-06, "loss": 1.267, "step": 2923 }, { "epoch": 0.3111300276654607, "grad_norm": 13.40625, "learning_rate": 9.913871517907903e-06, "loss": 1.1681, "step": 2924 }, { "epoch": 0.3112364332836774, "grad_norm": 14.78125, "learning_rate": 9.913835986355885e-06, "loss": 1.3369, "step": 2925 }, { "epoch": 0.311342838901894, "grad_norm": 20.25, "learning_rate": 9.913800454803866e-06, "loss": 1.3733, "step": 2926 }, { "epoch": 0.31144924452011064, "grad_norm": 20.03125, "learning_rate": 9.913764923251848e-06, "loss": 1.1464, "step": 2927 }, { "epoch": 0.3115556501383273, "grad_norm": 18.625, "learning_rate": 9.91372939169983e-06, "loss": 1.1376, "step": 2928 }, { "epoch": 0.31166205575654393, "grad_norm": 12.7578125, "learning_rate": 9.913693860147811e-06, "loss": 1.425, "step": 2929 }, { "epoch": 0.3117684613747606, "grad_norm": 12.984375, "learning_rate": 9.913658328595795e-06, "loss": 1.0927, "step": 2930 }, { "epoch": 0.31187486699297723, "grad_norm": 15.8984375, "learning_rate": 9.913622797043776e-06, "loss": 1.458, "step": 2931 }, { "epoch": 0.31198127261119385, "grad_norm": 20.640625, "learning_rate": 9.913587265491758e-06, "loss": 1.2634, "step": 2932 }, { "epoch": 0.3120876782294105, "grad_norm": 18.078125, "learning_rate": 9.91355173393974e-06, "loss": 1.3076, "step": 2933 }, { "epoch": 0.31219408384762715, "grad_norm": 12.78125, "learning_rate": 9.913516202387721e-06, "loss": 1.2012, "step": 2934 }, { "epoch": 0.3123004894658438, "grad_norm": 18.28125, "learning_rate": 9.913480670835703e-06, "loss": 1.3815, "step": 2935 }, { "epoch": 0.31240689508406044, "grad_norm": 42.53125, "learning_rate": 9.913445139283684e-06, "loss": 1.509, "step": 2936 }, { "epoch": 0.31251330070227706, "grad_norm": 19.84375, "learning_rate": 9.913409607731666e-06, "loss": 1.5679, "step": 2937 }, { "epoch": 0.31261970632049374, "grad_norm": 28.421875, "learning_rate": 9.913374076179647e-06, "loss": 1.3708, "step": 2938 }, { "epoch": 0.31272611193871036, "grad_norm": 11.4765625, "learning_rate": 9.913338544627629e-06, "loss": 1.1399, "step": 2939 }, { "epoch": 0.312832517556927, "grad_norm": 14.1953125, "learning_rate": 9.913303013075612e-06, "loss": 1.1287, "step": 2940 }, { "epoch": 0.31293892317514366, "grad_norm": 13.3203125, "learning_rate": 9.913267481523594e-06, "loss": 1.1887, "step": 2941 }, { "epoch": 0.3130453287933603, "grad_norm": 13.4609375, "learning_rate": 9.913231949971576e-06, "loss": 1.1853, "step": 2942 }, { "epoch": 0.31315173441157695, "grad_norm": 13.6640625, "learning_rate": 9.913196418419557e-06, "loss": 1.0944, "step": 2943 }, { "epoch": 0.3132581400297936, "grad_norm": 17.46875, "learning_rate": 9.913160886867539e-06, "loss": 1.4707, "step": 2944 }, { "epoch": 0.3133645456480102, "grad_norm": 14.8203125, "learning_rate": 9.91312535531552e-06, "loss": 1.0843, "step": 2945 }, { "epoch": 0.31347095126622687, "grad_norm": 16.265625, "learning_rate": 9.913089823763504e-06, "loss": 1.2925, "step": 2946 }, { "epoch": 0.3135773568844435, "grad_norm": 20.25, "learning_rate": 9.913054292211485e-06, "loss": 1.4097, "step": 2947 }, { "epoch": 0.3136837625026601, "grad_norm": 21.734375, "learning_rate": 9.913018760659467e-06, "loss": 1.4433, "step": 2948 }, { "epoch": 0.3137901681208768, "grad_norm": 17.375, "learning_rate": 9.912983229107449e-06, "loss": 1.2648, "step": 2949 }, { "epoch": 0.3138965737390934, "grad_norm": 15.5859375, "learning_rate": 9.91294769755543e-06, "loss": 1.2327, "step": 2950 }, { "epoch": 0.3140029793573101, "grad_norm": 18.375, "learning_rate": 9.912912166003412e-06, "loss": 1.4229, "step": 2951 }, { "epoch": 0.3141093849755267, "grad_norm": 15.7109375, "learning_rate": 9.912876634451393e-06, "loss": 1.3543, "step": 2952 }, { "epoch": 0.3142157905937433, "grad_norm": 19.734375, "learning_rate": 9.912841102899375e-06, "loss": 1.6354, "step": 2953 }, { "epoch": 0.31432219621196, "grad_norm": 17.484375, "learning_rate": 9.912805571347357e-06, "loss": 1.1612, "step": 2954 }, { "epoch": 0.3144286018301766, "grad_norm": 13.5703125, "learning_rate": 9.912770039795338e-06, "loss": 1.6392, "step": 2955 }, { "epoch": 0.3145350074483933, "grad_norm": 26.203125, "learning_rate": 9.912734508243321e-06, "loss": 1.6664, "step": 2956 }, { "epoch": 0.3146414130666099, "grad_norm": 16.203125, "learning_rate": 9.912698976691303e-06, "loss": 1.5854, "step": 2957 }, { "epoch": 0.31474781868482654, "grad_norm": 18.171875, "learning_rate": 9.912663445139285e-06, "loss": 1.258, "step": 2958 }, { "epoch": 0.3148542243030432, "grad_norm": 23.21875, "learning_rate": 9.912627913587266e-06, "loss": 1.3738, "step": 2959 }, { "epoch": 0.31496062992125984, "grad_norm": 11.765625, "learning_rate": 9.912592382035248e-06, "loss": 1.208, "step": 2960 }, { "epoch": 0.31506703553947646, "grad_norm": 17.421875, "learning_rate": 9.912556850483231e-06, "loss": 1.2102, "step": 2961 }, { "epoch": 0.31517344115769313, "grad_norm": 15.171875, "learning_rate": 9.912521318931213e-06, "loss": 1.5269, "step": 2962 }, { "epoch": 0.31527984677590976, "grad_norm": 15.0859375, "learning_rate": 9.912485787379194e-06, "loss": 1.2605, "step": 2963 }, { "epoch": 0.31538625239412643, "grad_norm": 18.234375, "learning_rate": 9.912450255827174e-06, "loss": 1.3551, "step": 2964 }, { "epoch": 0.31549265801234305, "grad_norm": 30.5625, "learning_rate": 9.912414724275156e-06, "loss": 1.5686, "step": 2965 }, { "epoch": 0.3155990636305597, "grad_norm": 15.0234375, "learning_rate": 9.91237919272314e-06, "loss": 1.4136, "step": 2966 }, { "epoch": 0.31570546924877635, "grad_norm": 15.7421875, "learning_rate": 9.91234366117112e-06, "loss": 1.2572, "step": 2967 }, { "epoch": 0.31581187486699297, "grad_norm": 12.8046875, "learning_rate": 9.912308129619102e-06, "loss": 1.245, "step": 2968 }, { "epoch": 0.31591828048520965, "grad_norm": 14.8125, "learning_rate": 9.912272598067084e-06, "loss": 1.1996, "step": 2969 }, { "epoch": 0.31602468610342627, "grad_norm": 16.046875, "learning_rate": 9.912237066515066e-06, "loss": 1.4617, "step": 2970 }, { "epoch": 0.3161310917216429, "grad_norm": 14.1328125, "learning_rate": 9.912201534963049e-06, "loss": 1.1482, "step": 2971 }, { "epoch": 0.31623749733985956, "grad_norm": 35.46875, "learning_rate": 9.91216600341103e-06, "loss": 1.4915, "step": 2972 }, { "epoch": 0.3163439029580762, "grad_norm": 18.421875, "learning_rate": 9.912130471859012e-06, "loss": 1.2218, "step": 2973 }, { "epoch": 0.3164503085762928, "grad_norm": 16.21875, "learning_rate": 9.912094940306994e-06, "loss": 1.2583, "step": 2974 }, { "epoch": 0.3165567141945095, "grad_norm": 13.171875, "learning_rate": 9.912059408754975e-06, "loss": 1.5131, "step": 2975 }, { "epoch": 0.3166631198127261, "grad_norm": 11.9921875, "learning_rate": 9.912023877202957e-06, "loss": 1.1641, "step": 2976 }, { "epoch": 0.3167695254309428, "grad_norm": 21.859375, "learning_rate": 9.911988345650939e-06, "loss": 1.8301, "step": 2977 }, { "epoch": 0.3168759310491594, "grad_norm": 15.0546875, "learning_rate": 9.91195281409892e-06, "loss": 1.5147, "step": 2978 }, { "epoch": 0.316982336667376, "grad_norm": 15.265625, "learning_rate": 9.911917282546902e-06, "loss": 1.2215, "step": 2979 }, { "epoch": 0.3170887422855927, "grad_norm": 15.6328125, "learning_rate": 9.911881750994883e-06, "loss": 1.3618, "step": 2980 }, { "epoch": 0.3171951479038093, "grad_norm": 18.09375, "learning_rate": 9.911846219442867e-06, "loss": 1.1752, "step": 2981 }, { "epoch": 0.31730155352202594, "grad_norm": 15.84375, "learning_rate": 9.911810687890848e-06, "loss": 1.2264, "step": 2982 }, { "epoch": 0.3174079591402426, "grad_norm": 12.71875, "learning_rate": 9.91177515633883e-06, "loss": 1.0947, "step": 2983 }, { "epoch": 0.31751436475845923, "grad_norm": 20.15625, "learning_rate": 9.911739624786812e-06, "loss": 1.3905, "step": 2984 }, { "epoch": 0.3176207703766759, "grad_norm": 12.375, "learning_rate": 9.911704093234793e-06, "loss": 1.1566, "step": 2985 }, { "epoch": 0.31772717599489253, "grad_norm": 15.8828125, "learning_rate": 9.911668561682775e-06, "loss": 1.5193, "step": 2986 }, { "epoch": 0.31783358161310915, "grad_norm": 14.96875, "learning_rate": 9.911633030130758e-06, "loss": 1.3134, "step": 2987 }, { "epoch": 0.3179399872313258, "grad_norm": 12.9453125, "learning_rate": 9.91159749857874e-06, "loss": 1.133, "step": 2988 }, { "epoch": 0.31804639284954245, "grad_norm": 26.359375, "learning_rate": 9.911561967026721e-06, "loss": 1.3286, "step": 2989 }, { "epoch": 0.3181527984677591, "grad_norm": 18.078125, "learning_rate": 9.911526435474703e-06, "loss": 1.255, "step": 2990 }, { "epoch": 0.31825920408597574, "grad_norm": 28.265625, "learning_rate": 9.911490903922684e-06, "loss": 1.4278, "step": 2991 }, { "epoch": 0.31836560970419236, "grad_norm": 38.375, "learning_rate": 9.911455372370666e-06, "loss": 1.4739, "step": 2992 }, { "epoch": 0.31847201532240904, "grad_norm": 19.3125, "learning_rate": 9.911419840818648e-06, "loss": 1.4912, "step": 2993 }, { "epoch": 0.31857842094062566, "grad_norm": 17.6875, "learning_rate": 9.91138430926663e-06, "loss": 1.2442, "step": 2994 }, { "epoch": 0.3186848265588423, "grad_norm": 18.265625, "learning_rate": 9.911348777714611e-06, "loss": 1.4142, "step": 2995 }, { "epoch": 0.31879123217705896, "grad_norm": 19.109375, "learning_rate": 9.911313246162592e-06, "loss": 1.2138, "step": 2996 }, { "epoch": 0.3188976377952756, "grad_norm": 20.015625, "learning_rate": 9.911277714610576e-06, "loss": 1.5016, "step": 2997 }, { "epoch": 0.31900404341349226, "grad_norm": 13.59375, "learning_rate": 9.911242183058557e-06, "loss": 1.4473, "step": 2998 }, { "epoch": 0.3191104490317089, "grad_norm": 15.375, "learning_rate": 9.911206651506539e-06, "loss": 1.5637, "step": 2999 }, { "epoch": 0.3192168546499255, "grad_norm": 13.328125, "learning_rate": 9.91117111995452e-06, "loss": 1.4917, "step": 3000 }, { "epoch": 0.3193232602681422, "grad_norm": 12.265625, "learning_rate": 9.911135588402502e-06, "loss": 1.3129, "step": 3001 }, { "epoch": 0.3194296658863588, "grad_norm": 14.2734375, "learning_rate": 9.911100056850484e-06, "loss": 1.2864, "step": 3002 }, { "epoch": 0.3195360715045754, "grad_norm": 21.34375, "learning_rate": 9.911064525298465e-06, "loss": 1.2771, "step": 3003 }, { "epoch": 0.3196424771227921, "grad_norm": 16.34375, "learning_rate": 9.911028993746447e-06, "loss": 1.3185, "step": 3004 }, { "epoch": 0.3197488827410087, "grad_norm": 26.703125, "learning_rate": 9.910993462194429e-06, "loss": 1.1738, "step": 3005 }, { "epoch": 0.3198552883592254, "grad_norm": 16.53125, "learning_rate": 9.91095793064241e-06, "loss": 1.3274, "step": 3006 }, { "epoch": 0.319961693977442, "grad_norm": 14.859375, "learning_rate": 9.910922399090394e-06, "loss": 1.2364, "step": 3007 }, { "epoch": 0.32006809959565863, "grad_norm": 16.265625, "learning_rate": 9.910886867538375e-06, "loss": 1.3729, "step": 3008 }, { "epoch": 0.3201745052138753, "grad_norm": 13.078125, "learning_rate": 9.910851335986357e-06, "loss": 1.1616, "step": 3009 }, { "epoch": 0.3202809108320919, "grad_norm": 18.328125, "learning_rate": 9.910815804434338e-06, "loss": 1.1509, "step": 3010 }, { "epoch": 0.3203873164503086, "grad_norm": 17.375, "learning_rate": 9.91078027288232e-06, "loss": 1.2304, "step": 3011 }, { "epoch": 0.3204937220685252, "grad_norm": 13.9609375, "learning_rate": 9.910744741330303e-06, "loss": 1.1522, "step": 3012 }, { "epoch": 0.32060012768674184, "grad_norm": 13.96875, "learning_rate": 9.910709209778285e-06, "loss": 1.5613, "step": 3013 }, { "epoch": 0.3207065333049585, "grad_norm": 12.4375, "learning_rate": 9.910673678226266e-06, "loss": 1.0248, "step": 3014 }, { "epoch": 0.32081293892317514, "grad_norm": 13.34375, "learning_rate": 9.910638146674248e-06, "loss": 1.3209, "step": 3015 }, { "epoch": 0.32091934454139176, "grad_norm": 17.109375, "learning_rate": 9.91060261512223e-06, "loss": 1.5416, "step": 3016 }, { "epoch": 0.32102575015960844, "grad_norm": 21.453125, "learning_rate": 9.910567083570211e-06, "loss": 1.8562, "step": 3017 }, { "epoch": 0.32113215577782506, "grad_norm": 12.90625, "learning_rate": 9.910531552018193e-06, "loss": 1.1225, "step": 3018 }, { "epoch": 0.32123856139604173, "grad_norm": 13.0078125, "learning_rate": 9.910496020466175e-06, "loss": 1.2459, "step": 3019 }, { "epoch": 0.32134496701425835, "grad_norm": 9.765625, "learning_rate": 9.910460488914156e-06, "loss": 1.0792, "step": 3020 }, { "epoch": 0.321451372632475, "grad_norm": 11.171875, "learning_rate": 9.910424957362138e-06, "loss": 1.1175, "step": 3021 }, { "epoch": 0.32155777825069165, "grad_norm": 22.046875, "learning_rate": 9.91038942581012e-06, "loss": 1.0371, "step": 3022 }, { "epoch": 0.32166418386890827, "grad_norm": 13.125, "learning_rate": 9.910353894258103e-06, "loss": 1.1094, "step": 3023 }, { "epoch": 0.32177058948712495, "grad_norm": 14.578125, "learning_rate": 9.910318362706084e-06, "loss": 1.2663, "step": 3024 }, { "epoch": 0.32187699510534157, "grad_norm": 16.140625, "learning_rate": 9.910282831154066e-06, "loss": 1.1176, "step": 3025 }, { "epoch": 0.3219834007235582, "grad_norm": 16.625, "learning_rate": 9.910247299602047e-06, "loss": 1.5123, "step": 3026 }, { "epoch": 0.32208980634177486, "grad_norm": 13.0390625, "learning_rate": 9.910211768050029e-06, "loss": 1.1846, "step": 3027 }, { "epoch": 0.3221962119599915, "grad_norm": 14.65625, "learning_rate": 9.91017623649801e-06, "loss": 1.1323, "step": 3028 }, { "epoch": 0.3223026175782081, "grad_norm": 15.171875, "learning_rate": 9.910140704945992e-06, "loss": 1.4729, "step": 3029 }, { "epoch": 0.3224090231964248, "grad_norm": 19.609375, "learning_rate": 9.910105173393974e-06, "loss": 1.5193, "step": 3030 }, { "epoch": 0.3225154288146414, "grad_norm": 15.546875, "learning_rate": 9.910069641841955e-06, "loss": 1.522, "step": 3031 }, { "epoch": 0.3226218344328581, "grad_norm": 15.7578125, "learning_rate": 9.910034110289937e-06, "loss": 1.3745, "step": 3032 }, { "epoch": 0.3227282400510747, "grad_norm": 12.34375, "learning_rate": 9.90999857873792e-06, "loss": 1.2657, "step": 3033 }, { "epoch": 0.3228346456692913, "grad_norm": 15.796875, "learning_rate": 9.909963047185902e-06, "loss": 1.5149, "step": 3034 }, { "epoch": 0.322941051287508, "grad_norm": 15.9765625, "learning_rate": 9.909927515633884e-06, "loss": 1.4003, "step": 3035 }, { "epoch": 0.3230474569057246, "grad_norm": 12.859375, "learning_rate": 9.909891984081865e-06, "loss": 1.252, "step": 3036 }, { "epoch": 0.32315386252394124, "grad_norm": 14.515625, "learning_rate": 9.909856452529847e-06, "loss": 1.298, "step": 3037 }, { "epoch": 0.3232602681421579, "grad_norm": 15.5234375, "learning_rate": 9.90982092097783e-06, "loss": 1.3199, "step": 3038 }, { "epoch": 0.32336667376037453, "grad_norm": 16.6875, "learning_rate": 9.909785389425812e-06, "loss": 1.3493, "step": 3039 }, { "epoch": 0.3234730793785912, "grad_norm": 17.59375, "learning_rate": 9.909749857873793e-06, "loss": 1.0723, "step": 3040 }, { "epoch": 0.32357948499680783, "grad_norm": 12.8046875, "learning_rate": 9.909714326321775e-06, "loss": 1.0586, "step": 3041 }, { "epoch": 0.32368589061502445, "grad_norm": 16.828125, "learning_rate": 9.909678794769757e-06, "loss": 1.1894, "step": 3042 }, { "epoch": 0.32379229623324113, "grad_norm": 16.640625, "learning_rate": 9.909643263217738e-06, "loss": 1.2804, "step": 3043 }, { "epoch": 0.32389870185145775, "grad_norm": 12.65625, "learning_rate": 9.90960773166572e-06, "loss": 1.1387, "step": 3044 }, { "epoch": 0.3240051074696744, "grad_norm": 33.21875, "learning_rate": 9.909572200113701e-06, "loss": 2.0857, "step": 3045 }, { "epoch": 0.32411151308789105, "grad_norm": 22.8125, "learning_rate": 9.909536668561683e-06, "loss": 1.9745, "step": 3046 }, { "epoch": 0.32421791870610767, "grad_norm": 14.2890625, "learning_rate": 9.909501137009665e-06, "loss": 0.9881, "step": 3047 }, { "epoch": 0.32432432432432434, "grad_norm": 19.59375, "learning_rate": 9.909465605457648e-06, "loss": 1.5091, "step": 3048 }, { "epoch": 0.32443072994254096, "grad_norm": 14.8203125, "learning_rate": 9.90943007390563e-06, "loss": 1.4039, "step": 3049 }, { "epoch": 0.3245371355607576, "grad_norm": 13.921875, "learning_rate": 9.909394542353611e-06, "loss": 1.1454, "step": 3050 }, { "epoch": 0.32464354117897426, "grad_norm": 20.625, "learning_rate": 9.909359010801593e-06, "loss": 1.3406, "step": 3051 }, { "epoch": 0.3247499467971909, "grad_norm": 17.875, "learning_rate": 9.909323479249574e-06, "loss": 0.827, "step": 3052 }, { "epoch": 0.32485635241540756, "grad_norm": 12.7578125, "learning_rate": 9.909287947697556e-06, "loss": 1.2733, "step": 3053 }, { "epoch": 0.3249627580336242, "grad_norm": 12.0546875, "learning_rate": 9.909252416145538e-06, "loss": 1.2718, "step": 3054 }, { "epoch": 0.3250691636518408, "grad_norm": 14.1171875, "learning_rate": 9.909216884593519e-06, "loss": 1.1395, "step": 3055 }, { "epoch": 0.3251755692700575, "grad_norm": 13.5546875, "learning_rate": 9.9091813530415e-06, "loss": 1.2302, "step": 3056 }, { "epoch": 0.3252819748882741, "grad_norm": 13.9296875, "learning_rate": 9.909145821489482e-06, "loss": 1.3483, "step": 3057 }, { "epoch": 0.3253883805064907, "grad_norm": 16.703125, "learning_rate": 9.909110289937466e-06, "loss": 1.3063, "step": 3058 }, { "epoch": 0.3254947861247074, "grad_norm": 10.828125, "learning_rate": 9.909074758385447e-06, "loss": 1.3446, "step": 3059 }, { "epoch": 0.325601191742924, "grad_norm": 15.671875, "learning_rate": 9.909039226833429e-06, "loss": 1.4712, "step": 3060 }, { "epoch": 0.3257075973611407, "grad_norm": 13.078125, "learning_rate": 9.90900369528141e-06, "loss": 1.1872, "step": 3061 }, { "epoch": 0.3258140029793573, "grad_norm": 14.1328125, "learning_rate": 9.908968163729392e-06, "loss": 1.3827, "step": 3062 }, { "epoch": 0.32592040859757393, "grad_norm": 15.28125, "learning_rate": 9.908932632177374e-06, "loss": 1.3241, "step": 3063 }, { "epoch": 0.3260268142157906, "grad_norm": 17.796875, "learning_rate": 9.908897100625357e-06, "loss": 1.2547, "step": 3064 }, { "epoch": 0.3261332198340072, "grad_norm": 13.8984375, "learning_rate": 9.908861569073339e-06, "loss": 1.2127, "step": 3065 }, { "epoch": 0.3262396254522239, "grad_norm": 14.90625, "learning_rate": 9.90882603752132e-06, "loss": 1.3948, "step": 3066 }, { "epoch": 0.3263460310704405, "grad_norm": 17.609375, "learning_rate": 9.908790505969302e-06, "loss": 1.7846, "step": 3067 }, { "epoch": 0.32645243668865714, "grad_norm": 10.9609375, "learning_rate": 9.908754974417283e-06, "loss": 1.1902, "step": 3068 }, { "epoch": 0.3265588423068738, "grad_norm": 11.828125, "learning_rate": 9.908719442865265e-06, "loss": 1.1417, "step": 3069 }, { "epoch": 0.32666524792509044, "grad_norm": 11.890625, "learning_rate": 9.908683911313247e-06, "loss": 1.1531, "step": 3070 }, { "epoch": 0.32677165354330706, "grad_norm": 12.28125, "learning_rate": 9.908648379761228e-06, "loss": 1.1547, "step": 3071 }, { "epoch": 0.32687805916152374, "grad_norm": 13.59375, "learning_rate": 9.90861284820921e-06, "loss": 1.27, "step": 3072 }, { "epoch": 0.32698446477974036, "grad_norm": 9.8359375, "learning_rate": 9.908577316657191e-06, "loss": 1.0545, "step": 3073 }, { "epoch": 0.32709087039795703, "grad_norm": 16.171875, "learning_rate": 9.908541785105175e-06, "loss": 1.51, "step": 3074 }, { "epoch": 0.32719727601617365, "grad_norm": 18.46875, "learning_rate": 9.908506253553156e-06, "loss": 1.1619, "step": 3075 }, { "epoch": 0.3273036816343903, "grad_norm": 17.59375, "learning_rate": 9.908470722001138e-06, "loss": 1.537, "step": 3076 }, { "epoch": 0.32741008725260695, "grad_norm": 15.9609375, "learning_rate": 9.90843519044912e-06, "loss": 1.1387, "step": 3077 }, { "epoch": 0.3275164928708236, "grad_norm": 25.203125, "learning_rate": 9.908399658897101e-06, "loss": 1.3426, "step": 3078 }, { "epoch": 0.32762289848904025, "grad_norm": 20.90625, "learning_rate": 9.908364127345084e-06, "loss": 1.6687, "step": 3079 }, { "epoch": 0.32772930410725687, "grad_norm": 10.6484375, "learning_rate": 9.908328595793066e-06, "loss": 1.0651, "step": 3080 }, { "epoch": 0.3278357097254735, "grad_norm": 15.9140625, "learning_rate": 9.908293064241048e-06, "loss": 1.1617, "step": 3081 }, { "epoch": 0.32794211534369017, "grad_norm": 14.84375, "learning_rate": 9.90825753268903e-06, "loss": 1.3038, "step": 3082 }, { "epoch": 0.3280485209619068, "grad_norm": 14.7734375, "learning_rate": 9.90822200113701e-06, "loss": 1.3468, "step": 3083 }, { "epoch": 0.3281549265801234, "grad_norm": 31.125, "learning_rate": 9.908186469584993e-06, "loss": 1.2827, "step": 3084 }, { "epoch": 0.3282613321983401, "grad_norm": 14.71875, "learning_rate": 9.908150938032974e-06, "loss": 1.4302, "step": 3085 }, { "epoch": 0.3283677378165567, "grad_norm": 17.890625, "learning_rate": 9.908115406480956e-06, "loss": 1.4664, "step": 3086 }, { "epoch": 0.3284741434347734, "grad_norm": 14.484375, "learning_rate": 9.908079874928937e-06, "loss": 1.3736, "step": 3087 }, { "epoch": 0.32858054905299, "grad_norm": 13.3671875, "learning_rate": 9.908044343376919e-06, "loss": 1.2861, "step": 3088 }, { "epoch": 0.3286869546712066, "grad_norm": 23.96875, "learning_rate": 9.908008811824902e-06, "loss": 1.8407, "step": 3089 }, { "epoch": 0.3287933602894233, "grad_norm": 10.8515625, "learning_rate": 9.907973280272884e-06, "loss": 1.2528, "step": 3090 }, { "epoch": 0.3288997659076399, "grad_norm": 21.015625, "learning_rate": 9.907937748720865e-06, "loss": 1.6609, "step": 3091 }, { "epoch": 0.32900617152585654, "grad_norm": 13.5859375, "learning_rate": 9.907902217168847e-06, "loss": 1.5757, "step": 3092 }, { "epoch": 0.3291125771440732, "grad_norm": 12.5546875, "learning_rate": 9.907866685616829e-06, "loss": 1.2059, "step": 3093 }, { "epoch": 0.32921898276228984, "grad_norm": 13.7265625, "learning_rate": 9.90783115406481e-06, "loss": 1.3086, "step": 3094 }, { "epoch": 0.3293253883805065, "grad_norm": 10.8046875, "learning_rate": 9.907795622512792e-06, "loss": 1.0597, "step": 3095 }, { "epoch": 0.32943179399872313, "grad_norm": 13.8359375, "learning_rate": 9.907760090960773e-06, "loss": 1.3438, "step": 3096 }, { "epoch": 0.32953819961693975, "grad_norm": 331.75, "learning_rate": 9.907724559408755e-06, "loss": 1.4854, "step": 3097 }, { "epoch": 0.32964460523515643, "grad_norm": 18.03125, "learning_rate": 9.907689027856737e-06, "loss": 1.1791, "step": 3098 }, { "epoch": 0.32975101085337305, "grad_norm": 23.328125, "learning_rate": 9.90765349630472e-06, "loss": 1.2662, "step": 3099 }, { "epoch": 0.3298574164715897, "grad_norm": 11.5546875, "learning_rate": 9.907617964752702e-06, "loss": 1.2663, "step": 3100 }, { "epoch": 0.32996382208980635, "grad_norm": 16.578125, "learning_rate": 9.907582433200683e-06, "loss": 1.4109, "step": 3101 }, { "epoch": 0.33007022770802297, "grad_norm": 13.21875, "learning_rate": 9.907546901648665e-06, "loss": 1.5138, "step": 3102 }, { "epoch": 0.33017663332623964, "grad_norm": 16.390625, "learning_rate": 9.907511370096646e-06, "loss": 1.543, "step": 3103 }, { "epoch": 0.33028303894445626, "grad_norm": 17.796875, "learning_rate": 9.907475838544628e-06, "loss": 1.585, "step": 3104 }, { "epoch": 0.3303894445626729, "grad_norm": 17.3125, "learning_rate": 9.907440306992611e-06, "loss": 1.2533, "step": 3105 }, { "epoch": 0.33049585018088956, "grad_norm": 17.3125, "learning_rate": 9.907404775440593e-06, "loss": 1.5408, "step": 3106 }, { "epoch": 0.3306022557991062, "grad_norm": 16.375, "learning_rate": 9.907369243888575e-06, "loss": 1.2634, "step": 3107 }, { "epoch": 0.33070866141732286, "grad_norm": 18.828125, "learning_rate": 9.907333712336556e-06, "loss": 1.5371, "step": 3108 }, { "epoch": 0.3308150670355395, "grad_norm": 11.40625, "learning_rate": 9.907298180784538e-06, "loss": 1.0868, "step": 3109 }, { "epoch": 0.3309214726537561, "grad_norm": 11.8828125, "learning_rate": 9.90726264923252e-06, "loss": 0.9806, "step": 3110 }, { "epoch": 0.3310278782719728, "grad_norm": 11.3671875, "learning_rate": 9.907227117680501e-06, "loss": 0.893, "step": 3111 }, { "epoch": 0.3311342838901894, "grad_norm": 13.75, "learning_rate": 9.907191586128483e-06, "loss": 1.3692, "step": 3112 }, { "epoch": 0.331240689508406, "grad_norm": 25.09375, "learning_rate": 9.907156054576464e-06, "loss": 1.4961, "step": 3113 }, { "epoch": 0.3313470951266227, "grad_norm": 18.15625, "learning_rate": 9.907120523024446e-06, "loss": 1.261, "step": 3114 }, { "epoch": 0.3314535007448393, "grad_norm": 13.0234375, "learning_rate": 9.907084991472429e-06, "loss": 1.1821, "step": 3115 }, { "epoch": 0.331559906363056, "grad_norm": 14.0546875, "learning_rate": 9.90704945992041e-06, "loss": 1.1909, "step": 3116 }, { "epoch": 0.3316663119812726, "grad_norm": 17.484375, "learning_rate": 9.907013928368392e-06, "loss": 1.617, "step": 3117 }, { "epoch": 0.33177271759948923, "grad_norm": 16.46875, "learning_rate": 9.906978396816374e-06, "loss": 1.2083, "step": 3118 }, { "epoch": 0.3318791232177059, "grad_norm": 15.0859375, "learning_rate": 9.906942865264356e-06, "loss": 1.3234, "step": 3119 }, { "epoch": 0.3319855288359225, "grad_norm": 13.2265625, "learning_rate": 9.906907333712337e-06, "loss": 1.2873, "step": 3120 }, { "epoch": 0.3320919344541392, "grad_norm": 13.90625, "learning_rate": 9.906871802160319e-06, "loss": 1.1022, "step": 3121 }, { "epoch": 0.3321983400723558, "grad_norm": 16.484375, "learning_rate": 9.9068362706083e-06, "loss": 1.4032, "step": 3122 }, { "epoch": 0.33230474569057245, "grad_norm": 17.265625, "learning_rate": 9.906800739056282e-06, "loss": 1.3421, "step": 3123 }, { "epoch": 0.3324111513087891, "grad_norm": 16.90625, "learning_rate": 9.906765207504264e-06, "loss": 0.9931, "step": 3124 }, { "epoch": 0.33251755692700574, "grad_norm": 23.171875, "learning_rate": 9.906729675952247e-06, "loss": 1.3593, "step": 3125 }, { "epoch": 0.33262396254522236, "grad_norm": 16.921875, "learning_rate": 9.906694144400228e-06, "loss": 1.374, "step": 3126 }, { "epoch": 0.33273036816343904, "grad_norm": 13.7265625, "learning_rate": 9.90665861284821e-06, "loss": 1.5664, "step": 3127 }, { "epoch": 0.33283677378165566, "grad_norm": 11.8046875, "learning_rate": 9.906623081296192e-06, "loss": 1.0946, "step": 3128 }, { "epoch": 0.33294317939987234, "grad_norm": 14.4921875, "learning_rate": 9.906587549744173e-06, "loss": 0.9835, "step": 3129 }, { "epoch": 0.33304958501808896, "grad_norm": 16.421875, "learning_rate": 9.906552018192157e-06, "loss": 1.6281, "step": 3130 }, { "epoch": 0.3331559906363056, "grad_norm": 17.609375, "learning_rate": 9.906516486640138e-06, "loss": 1.5489, "step": 3131 }, { "epoch": 0.33326239625452225, "grad_norm": 15.390625, "learning_rate": 9.90648095508812e-06, "loss": 1.2717, "step": 3132 }, { "epoch": 0.3333688018727389, "grad_norm": 23.484375, "learning_rate": 9.906445423536101e-06, "loss": 1.0518, "step": 3133 }, { "epoch": 0.33347520749095555, "grad_norm": 9.625, "learning_rate": 9.906409891984083e-06, "loss": 1.0489, "step": 3134 }, { "epoch": 0.33358161310917217, "grad_norm": 15.8046875, "learning_rate": 9.906374360432065e-06, "loss": 1.6262, "step": 3135 }, { "epoch": 0.3336880187273888, "grad_norm": 15.765625, "learning_rate": 9.906338828880046e-06, "loss": 1.3859, "step": 3136 }, { "epoch": 0.33379442434560547, "grad_norm": 20.25, "learning_rate": 9.906303297328028e-06, "loss": 1.5689, "step": 3137 }, { "epoch": 0.3339008299638221, "grad_norm": 37.84375, "learning_rate": 9.90626776577601e-06, "loss": 1.5459, "step": 3138 }, { "epoch": 0.3340072355820387, "grad_norm": 14.375, "learning_rate": 9.906232234223991e-06, "loss": 1.1678, "step": 3139 }, { "epoch": 0.3341136412002554, "grad_norm": 12.28125, "learning_rate": 9.906196702671973e-06, "loss": 1.1722, "step": 3140 }, { "epoch": 0.334220046818472, "grad_norm": 11.15625, "learning_rate": 9.906161171119956e-06, "loss": 1.1499, "step": 3141 }, { "epoch": 0.3343264524366887, "grad_norm": 12.53125, "learning_rate": 9.906125639567938e-06, "loss": 1.0759, "step": 3142 }, { "epoch": 0.3344328580549053, "grad_norm": 11.3984375, "learning_rate": 9.90609010801592e-06, "loss": 1.2762, "step": 3143 }, { "epoch": 0.3345392636731219, "grad_norm": 11.5390625, "learning_rate": 9.9060545764639e-06, "loss": 1.2818, "step": 3144 }, { "epoch": 0.3346456692913386, "grad_norm": 13.3671875, "learning_rate": 9.906019044911882e-06, "loss": 1.021, "step": 3145 }, { "epoch": 0.3347520749095552, "grad_norm": 14.5390625, "learning_rate": 9.905983513359864e-06, "loss": 1.4307, "step": 3146 }, { "epoch": 0.33485848052777184, "grad_norm": 12.734375, "learning_rate": 9.905947981807846e-06, "loss": 1.1489, "step": 3147 }, { "epoch": 0.3349648861459885, "grad_norm": 12.6015625, "learning_rate": 9.905912450255827e-06, "loss": 0.8913, "step": 3148 }, { "epoch": 0.33507129176420514, "grad_norm": 12.3359375, "learning_rate": 9.905876918703809e-06, "loss": 1.1231, "step": 3149 }, { "epoch": 0.3351776973824218, "grad_norm": 10.5, "learning_rate": 9.90584138715179e-06, "loss": 1.0837, "step": 3150 }, { "epoch": 0.33528410300063843, "grad_norm": 13.3828125, "learning_rate": 9.905805855599774e-06, "loss": 1.3808, "step": 3151 }, { "epoch": 0.33539050861885505, "grad_norm": 17.28125, "learning_rate": 9.905770324047755e-06, "loss": 1.3245, "step": 3152 }, { "epoch": 0.33549691423707173, "grad_norm": 12.7578125, "learning_rate": 9.905734792495737e-06, "loss": 1.1669, "step": 3153 }, { "epoch": 0.33560331985528835, "grad_norm": 11.2421875, "learning_rate": 9.905699260943719e-06, "loss": 1.0672, "step": 3154 }, { "epoch": 0.335709725473505, "grad_norm": 19.5625, "learning_rate": 9.9056637293917e-06, "loss": 1.2903, "step": 3155 }, { "epoch": 0.33581613109172165, "grad_norm": 12.4375, "learning_rate": 9.905628197839683e-06, "loss": 1.1699, "step": 3156 }, { "epoch": 0.33592253670993827, "grad_norm": 15.5546875, "learning_rate": 9.905592666287665e-06, "loss": 1.5909, "step": 3157 }, { "epoch": 0.33602894232815494, "grad_norm": 13.5078125, "learning_rate": 9.905557134735647e-06, "loss": 1.4208, "step": 3158 }, { "epoch": 0.33613534794637157, "grad_norm": 15.7578125, "learning_rate": 9.905521603183628e-06, "loss": 1.4436, "step": 3159 }, { "epoch": 0.3362417535645882, "grad_norm": 14.4375, "learning_rate": 9.90548607163161e-06, "loss": 1.2241, "step": 3160 }, { "epoch": 0.33634815918280486, "grad_norm": 15.3984375, "learning_rate": 9.905450540079591e-06, "loss": 1.325, "step": 3161 }, { "epoch": 0.3364545648010215, "grad_norm": 15.78125, "learning_rate": 9.905415008527573e-06, "loss": 1.5247, "step": 3162 }, { "epoch": 0.33656097041923816, "grad_norm": 11.9765625, "learning_rate": 9.905379476975555e-06, "loss": 1.2395, "step": 3163 }, { "epoch": 0.3366673760374548, "grad_norm": 13.6328125, "learning_rate": 9.905343945423536e-06, "loss": 1.4424, "step": 3164 }, { "epoch": 0.3367737816556714, "grad_norm": 14.25, "learning_rate": 9.905308413871518e-06, "loss": 1.1931, "step": 3165 }, { "epoch": 0.3368801872738881, "grad_norm": 13.8984375, "learning_rate": 9.905272882319501e-06, "loss": 1.2196, "step": 3166 }, { "epoch": 0.3369865928921047, "grad_norm": 10.703125, "learning_rate": 9.905237350767483e-06, "loss": 0.944, "step": 3167 }, { "epoch": 0.3370929985103213, "grad_norm": 15.0546875, "learning_rate": 9.905201819215464e-06, "loss": 1.2801, "step": 3168 }, { "epoch": 0.337199404128538, "grad_norm": 15.71875, "learning_rate": 9.905166287663446e-06, "loss": 1.2863, "step": 3169 }, { "epoch": 0.3373058097467546, "grad_norm": 13.328125, "learning_rate": 9.905130756111428e-06, "loss": 1.1353, "step": 3170 }, { "epoch": 0.3374122153649713, "grad_norm": 13.0078125, "learning_rate": 9.905095224559411e-06, "loss": 1.3319, "step": 3171 }, { "epoch": 0.3375186209831879, "grad_norm": 16.984375, "learning_rate": 9.90505969300739e-06, "loss": 1.3676, "step": 3172 }, { "epoch": 0.33762502660140453, "grad_norm": 18.34375, "learning_rate": 9.905024161455372e-06, "loss": 1.6601, "step": 3173 }, { "epoch": 0.3377314322196212, "grad_norm": 12.9921875, "learning_rate": 9.904988629903354e-06, "loss": 1.2313, "step": 3174 }, { "epoch": 0.33783783783783783, "grad_norm": 20.984375, "learning_rate": 9.904953098351336e-06, "loss": 1.3731, "step": 3175 }, { "epoch": 0.3379442434560545, "grad_norm": 12.8671875, "learning_rate": 9.904917566799319e-06, "loss": 1.621, "step": 3176 }, { "epoch": 0.3380506490742711, "grad_norm": 19.546875, "learning_rate": 9.9048820352473e-06, "loss": 1.4059, "step": 3177 }, { "epoch": 0.33815705469248775, "grad_norm": 14.125, "learning_rate": 9.904846503695282e-06, "loss": 1.3603, "step": 3178 }, { "epoch": 0.3382634603107044, "grad_norm": 16.5625, "learning_rate": 9.904810972143264e-06, "loss": 1.1492, "step": 3179 }, { "epoch": 0.33836986592892104, "grad_norm": 13.828125, "learning_rate": 9.904775440591245e-06, "loss": 1.2833, "step": 3180 }, { "epoch": 0.33847627154713766, "grad_norm": 13.90625, "learning_rate": 9.904739909039227e-06, "loss": 1.2596, "step": 3181 }, { "epoch": 0.33858267716535434, "grad_norm": 13.7578125, "learning_rate": 9.90470437748721e-06, "loss": 1.0922, "step": 3182 }, { "epoch": 0.33868908278357096, "grad_norm": 16.375, "learning_rate": 9.904668845935192e-06, "loss": 1.1955, "step": 3183 }, { "epoch": 0.33879548840178764, "grad_norm": 13.015625, "learning_rate": 9.904633314383174e-06, "loss": 1.0975, "step": 3184 }, { "epoch": 0.33890189402000426, "grad_norm": 15.6640625, "learning_rate": 9.904597782831155e-06, "loss": 1.1038, "step": 3185 }, { "epoch": 0.3390082996382209, "grad_norm": 13.40625, "learning_rate": 9.904562251279137e-06, "loss": 1.2714, "step": 3186 }, { "epoch": 0.33911470525643755, "grad_norm": 30.953125, "learning_rate": 9.904526719727118e-06, "loss": 1.6995, "step": 3187 }, { "epoch": 0.3392211108746542, "grad_norm": 43.375, "learning_rate": 9.9044911881751e-06, "loss": 1.327, "step": 3188 }, { "epoch": 0.33932751649287085, "grad_norm": 16.890625, "learning_rate": 9.904455656623082e-06, "loss": 1.3822, "step": 3189 }, { "epoch": 0.33943392211108747, "grad_norm": 13.390625, "learning_rate": 9.904420125071063e-06, "loss": 1.1733, "step": 3190 }, { "epoch": 0.3395403277293041, "grad_norm": 13.578125, "learning_rate": 9.904384593519045e-06, "loss": 1.3979, "step": 3191 }, { "epoch": 0.33964673334752077, "grad_norm": 13.5078125, "learning_rate": 9.904349061967028e-06, "loss": 1.2934, "step": 3192 }, { "epoch": 0.3397531389657374, "grad_norm": 16.640625, "learning_rate": 9.90431353041501e-06, "loss": 1.2281, "step": 3193 }, { "epoch": 0.339859544583954, "grad_norm": 12.1875, "learning_rate": 9.904277998862991e-06, "loss": 0.9883, "step": 3194 }, { "epoch": 0.3399659502021707, "grad_norm": 14.4765625, "learning_rate": 9.904242467310973e-06, "loss": 1.1826, "step": 3195 }, { "epoch": 0.3400723558203873, "grad_norm": 16.28125, "learning_rate": 9.904206935758954e-06, "loss": 1.287, "step": 3196 }, { "epoch": 0.340178761438604, "grad_norm": 14.9296875, "learning_rate": 9.904171404206938e-06, "loss": 1.11, "step": 3197 }, { "epoch": 0.3402851670568206, "grad_norm": 12.0625, "learning_rate": 9.90413587265492e-06, "loss": 1.3128, "step": 3198 }, { "epoch": 0.3403915726750372, "grad_norm": 18.25, "learning_rate": 9.904100341102901e-06, "loss": 1.5625, "step": 3199 }, { "epoch": 0.3404979782932539, "grad_norm": 15.3046875, "learning_rate": 9.904064809550883e-06, "loss": 1.3682, "step": 3200 }, { "epoch": 0.3406043839114705, "grad_norm": 20.0, "learning_rate": 9.904029277998864e-06, "loss": 1.572, "step": 3201 }, { "epoch": 0.34071078952968714, "grad_norm": 14.609375, "learning_rate": 9.903993746446846e-06, "loss": 1.493, "step": 3202 }, { "epoch": 0.3408171951479038, "grad_norm": 10.3046875, "learning_rate": 9.903958214894827e-06, "loss": 1.2308, "step": 3203 }, { "epoch": 0.34092360076612044, "grad_norm": 11.3359375, "learning_rate": 9.903922683342809e-06, "loss": 1.1503, "step": 3204 }, { "epoch": 0.3410300063843371, "grad_norm": 18.8125, "learning_rate": 9.90388715179079e-06, "loss": 1.2818, "step": 3205 }, { "epoch": 0.34113641200255373, "grad_norm": 14.546875, "learning_rate": 9.903851620238772e-06, "loss": 1.3268, "step": 3206 }, { "epoch": 0.34124281762077036, "grad_norm": 12.4453125, "learning_rate": 9.903816088686756e-06, "loss": 1.1668, "step": 3207 }, { "epoch": 0.34134922323898703, "grad_norm": 15.03125, "learning_rate": 9.903780557134737e-06, "loss": 1.3759, "step": 3208 }, { "epoch": 0.34145562885720365, "grad_norm": 10.8984375, "learning_rate": 9.903745025582719e-06, "loss": 1.3546, "step": 3209 }, { "epoch": 0.34156203447542033, "grad_norm": 12.7734375, "learning_rate": 9.9037094940307e-06, "loss": 1.3798, "step": 3210 }, { "epoch": 0.34166844009363695, "grad_norm": 12.171875, "learning_rate": 9.903673962478682e-06, "loss": 1.2798, "step": 3211 }, { "epoch": 0.34177484571185357, "grad_norm": 11.8203125, "learning_rate": 9.903638430926664e-06, "loss": 1.0806, "step": 3212 }, { "epoch": 0.34188125133007025, "grad_norm": 15.09375, "learning_rate": 9.903602899374645e-06, "loss": 1.2858, "step": 3213 }, { "epoch": 0.34198765694828687, "grad_norm": 14.484375, "learning_rate": 9.903567367822627e-06, "loss": 1.1528, "step": 3214 }, { "epoch": 0.3420940625665035, "grad_norm": 15.7421875, "learning_rate": 9.903531836270608e-06, "loss": 1.3578, "step": 3215 }, { "epoch": 0.34220046818472016, "grad_norm": 21.265625, "learning_rate": 9.90349630471859e-06, "loss": 1.312, "step": 3216 }, { "epoch": 0.3423068738029368, "grad_norm": 17.890625, "learning_rate": 9.903460773166572e-06, "loss": 1.2739, "step": 3217 }, { "epoch": 0.34241327942115346, "grad_norm": 15.6875, "learning_rate": 9.903425241614555e-06, "loss": 1.1361, "step": 3218 }, { "epoch": 0.3425196850393701, "grad_norm": 13.3984375, "learning_rate": 9.903389710062537e-06, "loss": 1.179, "step": 3219 }, { "epoch": 0.3426260906575867, "grad_norm": 13.3125, "learning_rate": 9.903354178510518e-06, "loss": 1.1707, "step": 3220 }, { "epoch": 0.3427324962758034, "grad_norm": 13.609375, "learning_rate": 9.9033186469585e-06, "loss": 1.3287, "step": 3221 }, { "epoch": 0.34283890189402, "grad_norm": 11.03125, "learning_rate": 9.903283115406481e-06, "loss": 1.2109, "step": 3222 }, { "epoch": 0.3429453075122366, "grad_norm": 13.0078125, "learning_rate": 9.903247583854465e-06, "loss": 1.1455, "step": 3223 }, { "epoch": 0.3430517131304533, "grad_norm": 15.9296875, "learning_rate": 9.903212052302446e-06, "loss": 1.5024, "step": 3224 }, { "epoch": 0.3431581187486699, "grad_norm": 14.171875, "learning_rate": 9.903176520750428e-06, "loss": 1.0183, "step": 3225 }, { "epoch": 0.3432645243668866, "grad_norm": 13.234375, "learning_rate": 9.90314098919841e-06, "loss": 1.0773, "step": 3226 }, { "epoch": 0.3433709299851032, "grad_norm": 16.234375, "learning_rate": 9.903105457646391e-06, "loss": 1.368, "step": 3227 }, { "epoch": 0.34347733560331983, "grad_norm": 13.1953125, "learning_rate": 9.903069926094373e-06, "loss": 1.4135, "step": 3228 }, { "epoch": 0.3435837412215365, "grad_norm": 15.8984375, "learning_rate": 9.903034394542354e-06, "loss": 1.257, "step": 3229 }, { "epoch": 0.34369014683975313, "grad_norm": 13.3515625, "learning_rate": 9.902998862990336e-06, "loss": 1.1508, "step": 3230 }, { "epoch": 0.3437965524579698, "grad_norm": 13.578125, "learning_rate": 9.902963331438317e-06, "loss": 0.9645, "step": 3231 }, { "epoch": 0.3439029580761864, "grad_norm": 18.71875, "learning_rate": 9.902927799886299e-06, "loss": 1.4999, "step": 3232 }, { "epoch": 0.34400936369440305, "grad_norm": 13.78125, "learning_rate": 9.902892268334282e-06, "loss": 1.1653, "step": 3233 }, { "epoch": 0.3441157693126197, "grad_norm": 12.625, "learning_rate": 9.902856736782264e-06, "loss": 1.2658, "step": 3234 }, { "epoch": 0.34422217493083634, "grad_norm": 18.9375, "learning_rate": 9.902821205230246e-06, "loss": 1.2666, "step": 3235 }, { "epoch": 0.34432858054905296, "grad_norm": 11.390625, "learning_rate": 9.902785673678227e-06, "loss": 0.9543, "step": 3236 }, { "epoch": 0.34443498616726964, "grad_norm": 18.0625, "learning_rate": 9.902750142126209e-06, "loss": 1.2877, "step": 3237 }, { "epoch": 0.34454139178548626, "grad_norm": 19.28125, "learning_rate": 9.90271461057419e-06, "loss": 1.3058, "step": 3238 }, { "epoch": 0.34464779740370294, "grad_norm": 14.921875, "learning_rate": 9.902679079022172e-06, "loss": 1.1374, "step": 3239 }, { "epoch": 0.34475420302191956, "grad_norm": 21.234375, "learning_rate": 9.902643547470154e-06, "loss": 1.4616, "step": 3240 }, { "epoch": 0.3448606086401362, "grad_norm": 12.3515625, "learning_rate": 9.902608015918135e-06, "loss": 1.0858, "step": 3241 }, { "epoch": 0.34496701425835286, "grad_norm": 18.609375, "learning_rate": 9.902572484366117e-06, "loss": 1.8046, "step": 3242 }, { "epoch": 0.3450734198765695, "grad_norm": 18.40625, "learning_rate": 9.9025369528141e-06, "loss": 1.4053, "step": 3243 }, { "epoch": 0.34517982549478615, "grad_norm": 13.3515625, "learning_rate": 9.902501421262082e-06, "loss": 1.4183, "step": 3244 }, { "epoch": 0.3452862311130028, "grad_norm": 15.5234375, "learning_rate": 9.902465889710063e-06, "loss": 1.2208, "step": 3245 }, { "epoch": 0.3453926367312194, "grad_norm": 13.2109375, "learning_rate": 9.902430358158045e-06, "loss": 1.1463, "step": 3246 }, { "epoch": 0.34549904234943607, "grad_norm": 10.8515625, "learning_rate": 9.902394826606027e-06, "loss": 1.1431, "step": 3247 }, { "epoch": 0.3456054479676527, "grad_norm": 30.03125, "learning_rate": 9.90235929505401e-06, "loss": 1.3594, "step": 3248 }, { "epoch": 0.3457118535858693, "grad_norm": 13.9765625, "learning_rate": 9.902323763501991e-06, "loss": 1.2948, "step": 3249 }, { "epoch": 0.345818259204086, "grad_norm": 9.9765625, "learning_rate": 9.902288231949973e-06, "loss": 1.1196, "step": 3250 }, { "epoch": 0.3459246648223026, "grad_norm": 17.78125, "learning_rate": 9.902252700397955e-06, "loss": 1.4845, "step": 3251 }, { "epoch": 0.3460310704405193, "grad_norm": 17.03125, "learning_rate": 9.902217168845936e-06, "loss": 1.3831, "step": 3252 }, { "epoch": 0.3461374760587359, "grad_norm": 14.5234375, "learning_rate": 9.902181637293918e-06, "loss": 1.1677, "step": 3253 }, { "epoch": 0.3462438816769525, "grad_norm": 14.53125, "learning_rate": 9.9021461057419e-06, "loss": 1.6873, "step": 3254 }, { "epoch": 0.3463502872951692, "grad_norm": 15.140625, "learning_rate": 9.902110574189881e-06, "loss": 1.7964, "step": 3255 }, { "epoch": 0.3464566929133858, "grad_norm": 15.375, "learning_rate": 9.902075042637863e-06, "loss": 1.3281, "step": 3256 }, { "epoch": 0.34656309853160244, "grad_norm": 15.6015625, "learning_rate": 9.902039511085844e-06, "loss": 1.2781, "step": 3257 }, { "epoch": 0.3466695041498191, "grad_norm": 18.453125, "learning_rate": 9.902003979533826e-06, "loss": 1.6067, "step": 3258 }, { "epoch": 0.34677590976803574, "grad_norm": 10.78125, "learning_rate": 9.90196844798181e-06, "loss": 1.0595, "step": 3259 }, { "epoch": 0.3468823153862524, "grad_norm": 18.03125, "learning_rate": 9.901932916429791e-06, "loss": 1.4231, "step": 3260 }, { "epoch": 0.34698872100446904, "grad_norm": 14.765625, "learning_rate": 9.901897384877772e-06, "loss": 1.5101, "step": 3261 }, { "epoch": 0.34709512662268566, "grad_norm": 14.6015625, "learning_rate": 9.901861853325754e-06, "loss": 1.3953, "step": 3262 }, { "epoch": 0.34720153224090233, "grad_norm": 20.09375, "learning_rate": 9.901826321773736e-06, "loss": 1.2624, "step": 3263 }, { "epoch": 0.34730793785911895, "grad_norm": 15.8046875, "learning_rate": 9.901790790221717e-06, "loss": 1.148, "step": 3264 }, { "epoch": 0.34741434347733563, "grad_norm": 13.75, "learning_rate": 9.901755258669699e-06, "loss": 1.2676, "step": 3265 }, { "epoch": 0.34752074909555225, "grad_norm": 10.5, "learning_rate": 9.90171972711768e-06, "loss": 1.3182, "step": 3266 }, { "epoch": 0.34762715471376887, "grad_norm": 19.953125, "learning_rate": 9.901684195565662e-06, "loss": 1.3948, "step": 3267 }, { "epoch": 0.34773356033198555, "grad_norm": 12.1953125, "learning_rate": 9.901648664013644e-06, "loss": 1.3902, "step": 3268 }, { "epoch": 0.34783996595020217, "grad_norm": 13.578125, "learning_rate": 9.901613132461627e-06, "loss": 1.6447, "step": 3269 }, { "epoch": 0.3479463715684188, "grad_norm": 19.359375, "learning_rate": 9.901577600909609e-06, "loss": 1.2593, "step": 3270 }, { "epoch": 0.34805277718663546, "grad_norm": 12.6640625, "learning_rate": 9.90154206935759e-06, "loss": 1.2508, "step": 3271 }, { "epoch": 0.3481591828048521, "grad_norm": 18.640625, "learning_rate": 9.901506537805572e-06, "loss": 1.2237, "step": 3272 }, { "epoch": 0.34826558842306876, "grad_norm": 20.046875, "learning_rate": 9.901471006253553e-06, "loss": 1.3053, "step": 3273 }, { "epoch": 0.3483719940412854, "grad_norm": 13.9140625, "learning_rate": 9.901435474701537e-06, "loss": 1.3062, "step": 3274 }, { "epoch": 0.348478399659502, "grad_norm": 16.90625, "learning_rate": 9.901399943149518e-06, "loss": 1.4622, "step": 3275 }, { "epoch": 0.3485848052777187, "grad_norm": 13.0390625, "learning_rate": 9.9013644115975e-06, "loss": 1.2725, "step": 3276 }, { "epoch": 0.3486912108959353, "grad_norm": 12.71875, "learning_rate": 9.901328880045482e-06, "loss": 1.273, "step": 3277 }, { "epoch": 0.3487976165141519, "grad_norm": 18.71875, "learning_rate": 9.901293348493463e-06, "loss": 1.2952, "step": 3278 }, { "epoch": 0.3489040221323686, "grad_norm": 36.84375, "learning_rate": 9.901257816941445e-06, "loss": 1.2707, "step": 3279 }, { "epoch": 0.3490104277505852, "grad_norm": 13.734375, "learning_rate": 9.901222285389426e-06, "loss": 1.1632, "step": 3280 }, { "epoch": 0.3491168333688019, "grad_norm": 14.046875, "learning_rate": 9.901186753837408e-06, "loss": 1.4866, "step": 3281 }, { "epoch": 0.3492232389870185, "grad_norm": 12.5078125, "learning_rate": 9.90115122228539e-06, "loss": 1.4185, "step": 3282 }, { "epoch": 0.34932964460523513, "grad_norm": 11.359375, "learning_rate": 9.901115690733371e-06, "loss": 1.1857, "step": 3283 }, { "epoch": 0.3494360502234518, "grad_norm": 14.5390625, "learning_rate": 9.901080159181354e-06, "loss": 1.5233, "step": 3284 }, { "epoch": 0.34954245584166843, "grad_norm": 13.125, "learning_rate": 9.901044627629336e-06, "loss": 1.4171, "step": 3285 }, { "epoch": 0.3496488614598851, "grad_norm": 14.2578125, "learning_rate": 9.901009096077318e-06, "loss": 1.0411, "step": 3286 }, { "epoch": 0.34975526707810173, "grad_norm": 13.5859375, "learning_rate": 9.9009735645253e-06, "loss": 1.2454, "step": 3287 }, { "epoch": 0.34986167269631835, "grad_norm": 13.6796875, "learning_rate": 9.900938032973281e-06, "loss": 1.1477, "step": 3288 }, { "epoch": 0.349968078314535, "grad_norm": 10.84375, "learning_rate": 9.900902501421264e-06, "loss": 1.339, "step": 3289 }, { "epoch": 0.35007448393275165, "grad_norm": 11.7734375, "learning_rate": 9.900866969869246e-06, "loss": 1.2653, "step": 3290 }, { "epoch": 0.35018088955096827, "grad_norm": 18.59375, "learning_rate": 9.900831438317227e-06, "loss": 1.1693, "step": 3291 }, { "epoch": 0.35028729516918494, "grad_norm": 15.5703125, "learning_rate": 9.900795906765207e-06, "loss": 1.318, "step": 3292 }, { "epoch": 0.35039370078740156, "grad_norm": 11.59375, "learning_rate": 9.900760375213189e-06, "loss": 0.959, "step": 3293 }, { "epoch": 0.35050010640561824, "grad_norm": 16.53125, "learning_rate": 9.900724843661172e-06, "loss": 1.3866, "step": 3294 }, { "epoch": 0.35060651202383486, "grad_norm": 19.09375, "learning_rate": 9.900689312109154e-06, "loss": 1.5352, "step": 3295 }, { "epoch": 0.3507129176420515, "grad_norm": 16.140625, "learning_rate": 9.900653780557135e-06, "loss": 1.6615, "step": 3296 }, { "epoch": 0.35081932326026816, "grad_norm": 16.96875, "learning_rate": 9.900618249005117e-06, "loss": 1.1459, "step": 3297 }, { "epoch": 0.3509257288784848, "grad_norm": 14.6484375, "learning_rate": 9.900582717453099e-06, "loss": 1.4194, "step": 3298 }, { "epoch": 0.35103213449670145, "grad_norm": 18.78125, "learning_rate": 9.90054718590108e-06, "loss": 1.0621, "step": 3299 }, { "epoch": 0.3511385401149181, "grad_norm": 18.140625, "learning_rate": 9.900511654349064e-06, "loss": 0.9988, "step": 3300 }, { "epoch": 0.3512449457331347, "grad_norm": 15.046875, "learning_rate": 9.900476122797045e-06, "loss": 1.553, "step": 3301 }, { "epoch": 0.35135135135135137, "grad_norm": 12.4140625, "learning_rate": 9.900440591245027e-06, "loss": 1.2774, "step": 3302 }, { "epoch": 0.351457756969568, "grad_norm": 12.171875, "learning_rate": 9.900405059693008e-06, "loss": 0.9322, "step": 3303 }, { "epoch": 0.3515641625877846, "grad_norm": 15.34375, "learning_rate": 9.90036952814099e-06, "loss": 1.6536, "step": 3304 }, { "epoch": 0.3516705682060013, "grad_norm": 12.9609375, "learning_rate": 9.900333996588972e-06, "loss": 1.4683, "step": 3305 }, { "epoch": 0.3517769738242179, "grad_norm": 14.3984375, "learning_rate": 9.900298465036953e-06, "loss": 1.2608, "step": 3306 }, { "epoch": 0.3518833794424346, "grad_norm": 12.8125, "learning_rate": 9.900262933484935e-06, "loss": 0.9892, "step": 3307 }, { "epoch": 0.3519897850606512, "grad_norm": 11.8359375, "learning_rate": 9.900227401932916e-06, "loss": 1.0275, "step": 3308 }, { "epoch": 0.3520961906788678, "grad_norm": 19.75, "learning_rate": 9.900191870380898e-06, "loss": 1.3871, "step": 3309 }, { "epoch": 0.3522025962970845, "grad_norm": 11.890625, "learning_rate": 9.900156338828881e-06, "loss": 1.0101, "step": 3310 }, { "epoch": 0.3523090019153011, "grad_norm": 12.8203125, "learning_rate": 9.900120807276863e-06, "loss": 1.2759, "step": 3311 }, { "epoch": 0.35241540753351774, "grad_norm": 16.625, "learning_rate": 9.900085275724845e-06, "loss": 1.2409, "step": 3312 }, { "epoch": 0.3525218131517344, "grad_norm": 10.9921875, "learning_rate": 9.900049744172826e-06, "loss": 0.9385, "step": 3313 }, { "epoch": 0.35262821876995104, "grad_norm": 13.8203125, "learning_rate": 9.900014212620808e-06, "loss": 1.2295, "step": 3314 }, { "epoch": 0.3527346243881677, "grad_norm": 15.0234375, "learning_rate": 9.899978681068791e-06, "loss": 1.3543, "step": 3315 }, { "epoch": 0.35284103000638434, "grad_norm": 19.015625, "learning_rate": 9.899943149516773e-06, "loss": 1.2812, "step": 3316 }, { "epoch": 0.35294743562460096, "grad_norm": 12.921875, "learning_rate": 9.899907617964754e-06, "loss": 1.2193, "step": 3317 }, { "epoch": 0.35305384124281763, "grad_norm": 12.5703125, "learning_rate": 9.899872086412736e-06, "loss": 1.1912, "step": 3318 }, { "epoch": 0.35316024686103425, "grad_norm": 19.296875, "learning_rate": 9.899836554860718e-06, "loss": 1.5053, "step": 3319 }, { "epoch": 0.35326665247925093, "grad_norm": 10.140625, "learning_rate": 9.899801023308699e-06, "loss": 1.2345, "step": 3320 }, { "epoch": 0.35337305809746755, "grad_norm": 15.59375, "learning_rate": 9.89976549175668e-06, "loss": 1.1812, "step": 3321 }, { "epoch": 0.3534794637156842, "grad_norm": 10.515625, "learning_rate": 9.899729960204662e-06, "loss": 0.8999, "step": 3322 }, { "epoch": 0.35358586933390085, "grad_norm": 21.59375, "learning_rate": 9.899694428652644e-06, "loss": 1.4206, "step": 3323 }, { "epoch": 0.35369227495211747, "grad_norm": 14.6484375, "learning_rate": 9.899658897100626e-06, "loss": 1.4151, "step": 3324 }, { "epoch": 0.3537986805703341, "grad_norm": 13.25, "learning_rate": 9.899623365548609e-06, "loss": 1.3551, "step": 3325 }, { "epoch": 0.35390508618855077, "grad_norm": 14.09375, "learning_rate": 9.89958783399659e-06, "loss": 1.2472, "step": 3326 }, { "epoch": 0.3540114918067674, "grad_norm": 12.125, "learning_rate": 9.899552302444572e-06, "loss": 1.0445, "step": 3327 }, { "epoch": 0.35411789742498406, "grad_norm": 13.921875, "learning_rate": 9.899516770892554e-06, "loss": 1.2254, "step": 3328 }, { "epoch": 0.3542243030432007, "grad_norm": 10.984375, "learning_rate": 9.899481239340535e-06, "loss": 0.9829, "step": 3329 }, { "epoch": 0.3543307086614173, "grad_norm": 9.8671875, "learning_rate": 9.899445707788517e-06, "loss": 0.8728, "step": 3330 }, { "epoch": 0.354437114279634, "grad_norm": 15.3046875, "learning_rate": 9.899410176236498e-06, "loss": 1.2317, "step": 3331 }, { "epoch": 0.3545435198978506, "grad_norm": 15.1796875, "learning_rate": 9.89937464468448e-06, "loss": 1.3318, "step": 3332 }, { "epoch": 0.3546499255160672, "grad_norm": 12.4296875, "learning_rate": 9.899339113132462e-06, "loss": 1.016, "step": 3333 }, { "epoch": 0.3547563311342839, "grad_norm": 13.1796875, "learning_rate": 9.899303581580443e-06, "loss": 1.3111, "step": 3334 }, { "epoch": 0.3548627367525005, "grad_norm": 12.5859375, "learning_rate": 9.899268050028425e-06, "loss": 1.3819, "step": 3335 }, { "epoch": 0.3549691423707172, "grad_norm": 12.171875, "learning_rate": 9.899232518476408e-06, "loss": 1.1101, "step": 3336 }, { "epoch": 0.3550755479889338, "grad_norm": 14.34375, "learning_rate": 9.89919698692439e-06, "loss": 1.3262, "step": 3337 }, { "epoch": 0.35518195360715044, "grad_norm": 15.96875, "learning_rate": 9.899161455372371e-06, "loss": 1.2673, "step": 3338 }, { "epoch": 0.3552883592253671, "grad_norm": 13.25, "learning_rate": 9.899125923820353e-06, "loss": 1.0652, "step": 3339 }, { "epoch": 0.35539476484358373, "grad_norm": 14.1796875, "learning_rate": 9.899090392268335e-06, "loss": 1.13, "step": 3340 }, { "epoch": 0.3555011704618004, "grad_norm": 13.53125, "learning_rate": 9.899054860716318e-06, "loss": 1.1645, "step": 3341 }, { "epoch": 0.35560757608001703, "grad_norm": 12.7421875, "learning_rate": 9.8990193291643e-06, "loss": 1.3542, "step": 3342 }, { "epoch": 0.35571398169823365, "grad_norm": 17.890625, "learning_rate": 9.898983797612281e-06, "loss": 1.6275, "step": 3343 }, { "epoch": 0.3558203873164503, "grad_norm": 12.140625, "learning_rate": 9.898948266060263e-06, "loss": 1.1413, "step": 3344 }, { "epoch": 0.35592679293466695, "grad_norm": 17.140625, "learning_rate": 9.898912734508244e-06, "loss": 1.7125, "step": 3345 }, { "epoch": 0.35603319855288357, "grad_norm": 14.5390625, "learning_rate": 9.898877202956226e-06, "loss": 1.5095, "step": 3346 }, { "epoch": 0.35613960417110024, "grad_norm": 14.484375, "learning_rate": 9.898841671404208e-06, "loss": 1.4528, "step": 3347 }, { "epoch": 0.35624600978931686, "grad_norm": 11.0, "learning_rate": 9.89880613985219e-06, "loss": 1.3406, "step": 3348 }, { "epoch": 0.35635241540753354, "grad_norm": 16.34375, "learning_rate": 9.89877060830017e-06, "loss": 1.5728, "step": 3349 }, { "epoch": 0.35645882102575016, "grad_norm": 16.359375, "learning_rate": 9.898735076748152e-06, "loss": 1.2953, "step": 3350 }, { "epoch": 0.3565652266439668, "grad_norm": 14.078125, "learning_rate": 9.898699545196136e-06, "loss": 1.2417, "step": 3351 }, { "epoch": 0.35667163226218346, "grad_norm": 17.078125, "learning_rate": 9.898664013644117e-06, "loss": 1.3849, "step": 3352 }, { "epoch": 0.3567780378804001, "grad_norm": 11.125, "learning_rate": 9.898628482092099e-06, "loss": 1.3652, "step": 3353 }, { "epoch": 0.35688444349861675, "grad_norm": 17.125, "learning_rate": 9.89859295054008e-06, "loss": 1.1133, "step": 3354 }, { "epoch": 0.3569908491168334, "grad_norm": 18.421875, "learning_rate": 9.898557418988062e-06, "loss": 1.253, "step": 3355 }, { "epoch": 0.35709725473505, "grad_norm": 15.7890625, "learning_rate": 9.898521887436044e-06, "loss": 1.4615, "step": 3356 }, { "epoch": 0.35720366035326667, "grad_norm": 12.515625, "learning_rate": 9.898486355884025e-06, "loss": 1.1493, "step": 3357 }, { "epoch": 0.3573100659714833, "grad_norm": 12.0078125, "learning_rate": 9.898450824332007e-06, "loss": 1.4196, "step": 3358 }, { "epoch": 0.3574164715896999, "grad_norm": 10.75, "learning_rate": 9.898415292779989e-06, "loss": 1.3942, "step": 3359 }, { "epoch": 0.3575228772079166, "grad_norm": 12.3125, "learning_rate": 9.89837976122797e-06, "loss": 1.1094, "step": 3360 }, { "epoch": 0.3576292828261332, "grad_norm": 11.6484375, "learning_rate": 9.898344229675953e-06, "loss": 1.4427, "step": 3361 }, { "epoch": 0.3577356884443499, "grad_norm": 15.546875, "learning_rate": 9.898308698123935e-06, "loss": 1.2969, "step": 3362 }, { "epoch": 0.3578420940625665, "grad_norm": 18.25, "learning_rate": 9.898273166571917e-06, "loss": 1.4437, "step": 3363 }, { "epoch": 0.3579484996807831, "grad_norm": 16.28125, "learning_rate": 9.898237635019898e-06, "loss": 1.0087, "step": 3364 }, { "epoch": 0.3580549052989998, "grad_norm": 9.140625, "learning_rate": 9.89820210346788e-06, "loss": 1.0547, "step": 3365 }, { "epoch": 0.3581613109172164, "grad_norm": 15.703125, "learning_rate": 9.898166571915863e-06, "loss": 1.1534, "step": 3366 }, { "epoch": 0.35826771653543305, "grad_norm": 12.1953125, "learning_rate": 9.898131040363845e-06, "loss": 0.9268, "step": 3367 }, { "epoch": 0.3583741221536497, "grad_norm": 15.7734375, "learning_rate": 9.898095508811826e-06, "loss": 1.1569, "step": 3368 }, { "epoch": 0.35848052777186634, "grad_norm": 12.4921875, "learning_rate": 9.898059977259808e-06, "loss": 0.982, "step": 3369 }, { "epoch": 0.358586933390083, "grad_norm": 21.8125, "learning_rate": 9.89802444570779e-06, "loss": 1.2991, "step": 3370 }, { "epoch": 0.35869333900829964, "grad_norm": 15.015625, "learning_rate": 9.897988914155771e-06, "loss": 1.0751, "step": 3371 }, { "epoch": 0.35879974462651626, "grad_norm": 31.953125, "learning_rate": 9.897953382603753e-06, "loss": 1.7292, "step": 3372 }, { "epoch": 0.35890615024473294, "grad_norm": 15.7734375, "learning_rate": 9.897917851051734e-06, "loss": 1.3359, "step": 3373 }, { "epoch": 0.35901255586294956, "grad_norm": 20.8125, "learning_rate": 9.897882319499716e-06, "loss": 1.5041, "step": 3374 }, { "epoch": 0.35911896148116623, "grad_norm": 14.5703125, "learning_rate": 9.897846787947698e-06, "loss": 1.2236, "step": 3375 }, { "epoch": 0.35922536709938285, "grad_norm": 14.0625, "learning_rate": 9.89781125639568e-06, "loss": 1.4008, "step": 3376 }, { "epoch": 0.3593317727175995, "grad_norm": 16.390625, "learning_rate": 9.897775724843663e-06, "loss": 1.4533, "step": 3377 }, { "epoch": 0.35943817833581615, "grad_norm": 14.8203125, "learning_rate": 9.897740193291644e-06, "loss": 1.5717, "step": 3378 }, { "epoch": 0.35954458395403277, "grad_norm": 15.34375, "learning_rate": 9.897704661739626e-06, "loss": 1.3524, "step": 3379 }, { "epoch": 0.3596509895722494, "grad_norm": 13.3984375, "learning_rate": 9.897669130187607e-06, "loss": 1.1196, "step": 3380 }, { "epoch": 0.35975739519046607, "grad_norm": 15.203125, "learning_rate": 9.897633598635589e-06, "loss": 1.0655, "step": 3381 }, { "epoch": 0.3598638008086827, "grad_norm": 17.328125, "learning_rate": 9.89759806708357e-06, "loss": 1.4014, "step": 3382 }, { "epoch": 0.35997020642689936, "grad_norm": 14.796875, "learning_rate": 9.897562535531552e-06, "loss": 1.5448, "step": 3383 }, { "epoch": 0.360076612045116, "grad_norm": 11.703125, "learning_rate": 9.897527003979534e-06, "loss": 1.1149, "step": 3384 }, { "epoch": 0.3601830176633326, "grad_norm": 15.1484375, "learning_rate": 9.897491472427515e-06, "loss": 1.1823, "step": 3385 }, { "epoch": 0.3602894232815493, "grad_norm": 18.984375, "learning_rate": 9.897455940875497e-06, "loss": 1.2927, "step": 3386 }, { "epoch": 0.3603958288997659, "grad_norm": 11.7890625, "learning_rate": 9.89742040932348e-06, "loss": 1.025, "step": 3387 }, { "epoch": 0.3605022345179825, "grad_norm": 13.1953125, "learning_rate": 9.897384877771462e-06, "loss": 1.2395, "step": 3388 }, { "epoch": 0.3606086401361992, "grad_norm": 12.0859375, "learning_rate": 9.897349346219444e-06, "loss": 1.3232, "step": 3389 }, { "epoch": 0.3607150457544158, "grad_norm": 14.1875, "learning_rate": 9.897313814667425e-06, "loss": 1.3722, "step": 3390 }, { "epoch": 0.3608214513726325, "grad_norm": 20.34375, "learning_rate": 9.897278283115407e-06, "loss": 1.4388, "step": 3391 }, { "epoch": 0.3609278569908491, "grad_norm": 12.1328125, "learning_rate": 9.89724275156339e-06, "loss": 1.3156, "step": 3392 }, { "epoch": 0.36103426260906574, "grad_norm": 13.4140625, "learning_rate": 9.897207220011372e-06, "loss": 1.3167, "step": 3393 }, { "epoch": 0.3611406682272824, "grad_norm": 13.1953125, "learning_rate": 9.897171688459353e-06, "loss": 1.2045, "step": 3394 }, { "epoch": 0.36124707384549903, "grad_norm": 15.96875, "learning_rate": 9.897136156907335e-06, "loss": 1.3573, "step": 3395 }, { "epoch": 0.3613534794637157, "grad_norm": 11.1171875, "learning_rate": 9.897100625355316e-06, "loss": 1.0685, "step": 3396 }, { "epoch": 0.36145988508193233, "grad_norm": 14.6640625, "learning_rate": 9.897065093803298e-06, "loss": 1.2494, "step": 3397 }, { "epoch": 0.36156629070014895, "grad_norm": 15.296875, "learning_rate": 9.89702956225128e-06, "loss": 1.4542, "step": 3398 }, { "epoch": 0.3616726963183656, "grad_norm": 18.078125, "learning_rate": 9.896994030699261e-06, "loss": 1.2363, "step": 3399 }, { "epoch": 0.36177910193658225, "grad_norm": 29.125, "learning_rate": 9.896958499147243e-06, "loss": 1.4859, "step": 3400 }, { "epoch": 0.36188550755479887, "grad_norm": 15.46875, "learning_rate": 9.896922967595224e-06, "loss": 1.3217, "step": 3401 }, { "epoch": 0.36199191317301554, "grad_norm": 18.6875, "learning_rate": 9.896887436043208e-06, "loss": 1.2542, "step": 3402 }, { "epoch": 0.36209831879123217, "grad_norm": 16.640625, "learning_rate": 9.89685190449119e-06, "loss": 1.3044, "step": 3403 }, { "epoch": 0.36220472440944884, "grad_norm": 12.3046875, "learning_rate": 9.896816372939171e-06, "loss": 1.1034, "step": 3404 }, { "epoch": 0.36231113002766546, "grad_norm": 10.8203125, "learning_rate": 9.896780841387153e-06, "loss": 1.1974, "step": 3405 }, { "epoch": 0.3624175356458821, "grad_norm": 35.15625, "learning_rate": 9.896745309835134e-06, "loss": 1.2402, "step": 3406 }, { "epoch": 0.36252394126409876, "grad_norm": 21.9375, "learning_rate": 9.896709778283118e-06, "loss": 1.4501, "step": 3407 }, { "epoch": 0.3626303468823154, "grad_norm": 24.484375, "learning_rate": 9.896674246731099e-06, "loss": 1.8889, "step": 3408 }, { "epoch": 0.362736752500532, "grad_norm": 22.484375, "learning_rate": 9.89663871517908e-06, "loss": 1.6418, "step": 3409 }, { "epoch": 0.3628431581187487, "grad_norm": 14.53125, "learning_rate": 9.896603183627062e-06, "loss": 1.4692, "step": 3410 }, { "epoch": 0.3629495637369653, "grad_norm": 15.34375, "learning_rate": 9.896567652075044e-06, "loss": 1.1803, "step": 3411 }, { "epoch": 0.363055969355182, "grad_norm": 13.7109375, "learning_rate": 9.896532120523026e-06, "loss": 1.2188, "step": 3412 }, { "epoch": 0.3631623749733986, "grad_norm": 15.4453125, "learning_rate": 9.896496588971007e-06, "loss": 1.2826, "step": 3413 }, { "epoch": 0.3632687805916152, "grad_norm": 15.4375, "learning_rate": 9.896461057418989e-06, "loss": 0.8904, "step": 3414 }, { "epoch": 0.3633751862098319, "grad_norm": 16.53125, "learning_rate": 9.89642552586697e-06, "loss": 1.2336, "step": 3415 }, { "epoch": 0.3634815918280485, "grad_norm": 14.078125, "learning_rate": 9.896389994314952e-06, "loss": 1.1432, "step": 3416 }, { "epoch": 0.3635879974462652, "grad_norm": 12.5546875, "learning_rate": 9.896354462762934e-06, "loss": 1.3602, "step": 3417 }, { "epoch": 0.3636944030644818, "grad_norm": 15.8125, "learning_rate": 9.896318931210917e-06, "loss": 1.3064, "step": 3418 }, { "epoch": 0.36380080868269843, "grad_norm": 11.9453125, "learning_rate": 9.896283399658898e-06, "loss": 1.3721, "step": 3419 }, { "epoch": 0.3639072143009151, "grad_norm": 10.3984375, "learning_rate": 9.89624786810688e-06, "loss": 1.0515, "step": 3420 }, { "epoch": 0.3640136199191317, "grad_norm": 14.28125, "learning_rate": 9.896212336554862e-06, "loss": 1.0842, "step": 3421 }, { "epoch": 0.36412002553734835, "grad_norm": 14.6328125, "learning_rate": 9.896176805002843e-06, "loss": 1.263, "step": 3422 }, { "epoch": 0.364226431155565, "grad_norm": 12.203125, "learning_rate": 9.896141273450825e-06, "loss": 1.448, "step": 3423 }, { "epoch": 0.36433283677378164, "grad_norm": 14.34375, "learning_rate": 9.896105741898807e-06, "loss": 1.0592, "step": 3424 }, { "epoch": 0.3644392423919983, "grad_norm": 12.96875, "learning_rate": 9.896070210346788e-06, "loss": 1.1965, "step": 3425 }, { "epoch": 0.36454564801021494, "grad_norm": 16.875, "learning_rate": 9.89603467879477e-06, "loss": 0.9692, "step": 3426 }, { "epoch": 0.36465205362843156, "grad_norm": 15.046875, "learning_rate": 9.895999147242751e-06, "loss": 1.6753, "step": 3427 }, { "epoch": 0.36475845924664824, "grad_norm": 13.46875, "learning_rate": 9.895963615690735e-06, "loss": 1.4095, "step": 3428 }, { "epoch": 0.36486486486486486, "grad_norm": 15.0625, "learning_rate": 9.895928084138716e-06, "loss": 1.121, "step": 3429 }, { "epoch": 0.36497127048308153, "grad_norm": 10.8203125, "learning_rate": 9.895892552586698e-06, "loss": 1.2181, "step": 3430 }, { "epoch": 0.36507767610129815, "grad_norm": 15.7734375, "learning_rate": 9.89585702103468e-06, "loss": 1.469, "step": 3431 }, { "epoch": 0.3651840817195148, "grad_norm": 10.3046875, "learning_rate": 9.895821489482661e-06, "loss": 1.1048, "step": 3432 }, { "epoch": 0.36529048733773145, "grad_norm": 12.4921875, "learning_rate": 9.895785957930644e-06, "loss": 1.3074, "step": 3433 }, { "epoch": 0.36539689295594807, "grad_norm": 14.1328125, "learning_rate": 9.895750426378626e-06, "loss": 1.5373, "step": 3434 }, { "epoch": 0.3655032985741647, "grad_norm": 28.453125, "learning_rate": 9.895714894826608e-06, "loss": 1.2254, "step": 3435 }, { "epoch": 0.36560970419238137, "grad_norm": 18.140625, "learning_rate": 9.89567936327459e-06, "loss": 1.4117, "step": 3436 }, { "epoch": 0.365716109810598, "grad_norm": 14.1796875, "learning_rate": 9.89564383172257e-06, "loss": 1.2093, "step": 3437 }, { "epoch": 0.36582251542881467, "grad_norm": 16.03125, "learning_rate": 9.895608300170552e-06, "loss": 1.004, "step": 3438 }, { "epoch": 0.3659289210470313, "grad_norm": 16.71875, "learning_rate": 9.895572768618534e-06, "loss": 1.4936, "step": 3439 }, { "epoch": 0.3660353266652479, "grad_norm": 12.046875, "learning_rate": 9.895537237066516e-06, "loss": 1.1469, "step": 3440 }, { "epoch": 0.3661417322834646, "grad_norm": 17.53125, "learning_rate": 9.895501705514497e-06, "loss": 1.0736, "step": 3441 }, { "epoch": 0.3662481379016812, "grad_norm": 13.984375, "learning_rate": 9.895466173962479e-06, "loss": 1.4166, "step": 3442 }, { "epoch": 0.3663545435198978, "grad_norm": 12.8984375, "learning_rate": 9.895430642410462e-06, "loss": 1.0011, "step": 3443 }, { "epoch": 0.3664609491381145, "grad_norm": 15.421875, "learning_rate": 9.895395110858444e-06, "loss": 1.3812, "step": 3444 }, { "epoch": 0.3665673547563311, "grad_norm": 16.71875, "learning_rate": 9.895359579306425e-06, "loss": 1.14, "step": 3445 }, { "epoch": 0.3666737603745478, "grad_norm": 13.6875, "learning_rate": 9.895324047754407e-06, "loss": 1.2965, "step": 3446 }, { "epoch": 0.3667801659927644, "grad_norm": 15.4921875, "learning_rate": 9.895288516202389e-06, "loss": 1.2107, "step": 3447 }, { "epoch": 0.36688657161098104, "grad_norm": 14.6640625, "learning_rate": 9.89525298465037e-06, "loss": 1.1794, "step": 3448 }, { "epoch": 0.3669929772291977, "grad_norm": 13.3984375, "learning_rate": 9.895217453098352e-06, "loss": 1.2887, "step": 3449 }, { "epoch": 0.36709938284741434, "grad_norm": 19.703125, "learning_rate": 9.895181921546333e-06, "loss": 1.6794, "step": 3450 }, { "epoch": 0.367205788465631, "grad_norm": 13.421875, "learning_rate": 9.895146389994315e-06, "loss": 1.2537, "step": 3451 }, { "epoch": 0.36731219408384763, "grad_norm": 18.765625, "learning_rate": 9.895110858442297e-06, "loss": 1.2926, "step": 3452 }, { "epoch": 0.36741859970206425, "grad_norm": 20.265625, "learning_rate": 9.895075326890278e-06, "loss": 1.4146, "step": 3453 }, { "epoch": 0.36752500532028093, "grad_norm": 13.3515625, "learning_rate": 9.895039795338262e-06, "loss": 1.0919, "step": 3454 }, { "epoch": 0.36763141093849755, "grad_norm": 14.296875, "learning_rate": 9.895004263786243e-06, "loss": 1.3394, "step": 3455 }, { "epoch": 0.36773781655671417, "grad_norm": 15.1875, "learning_rate": 9.894968732234225e-06, "loss": 1.2058, "step": 3456 }, { "epoch": 0.36784422217493085, "grad_norm": 15.0390625, "learning_rate": 9.894933200682206e-06, "loss": 1.5951, "step": 3457 }, { "epoch": 0.36795062779314747, "grad_norm": 12.3359375, "learning_rate": 9.894897669130188e-06, "loss": 1.3958, "step": 3458 }, { "epoch": 0.36805703341136414, "grad_norm": 18.375, "learning_rate": 9.894862137578171e-06, "loss": 1.3236, "step": 3459 }, { "epoch": 0.36816343902958076, "grad_norm": 13.4140625, "learning_rate": 9.894826606026153e-06, "loss": 1.1922, "step": 3460 }, { "epoch": 0.3682698446477974, "grad_norm": 18.984375, "learning_rate": 9.894791074474134e-06, "loss": 1.2512, "step": 3461 }, { "epoch": 0.36837625026601406, "grad_norm": 41.0625, "learning_rate": 9.894755542922116e-06, "loss": 1.0782, "step": 3462 }, { "epoch": 0.3684826558842307, "grad_norm": 16.171875, "learning_rate": 9.894720011370098e-06, "loss": 1.3331, "step": 3463 }, { "epoch": 0.3685890615024473, "grad_norm": 15.0, "learning_rate": 9.89468447981808e-06, "loss": 1.4575, "step": 3464 }, { "epoch": 0.368695467120664, "grad_norm": 19.40625, "learning_rate": 9.894648948266061e-06, "loss": 1.1668, "step": 3465 }, { "epoch": 0.3688018727388806, "grad_norm": 12.96875, "learning_rate": 9.894613416714042e-06, "loss": 1.3365, "step": 3466 }, { "epoch": 0.3689082783570973, "grad_norm": 13.828125, "learning_rate": 9.894577885162024e-06, "loss": 1.2703, "step": 3467 }, { "epoch": 0.3690146839753139, "grad_norm": 21.03125, "learning_rate": 9.894542353610006e-06, "loss": 1.2303, "step": 3468 }, { "epoch": 0.3691210895935305, "grad_norm": 13.828125, "learning_rate": 9.894506822057989e-06, "loss": 1.3033, "step": 3469 }, { "epoch": 0.3692274952117472, "grad_norm": 13.78125, "learning_rate": 9.89447129050597e-06, "loss": 1.2618, "step": 3470 }, { "epoch": 0.3693339008299638, "grad_norm": 15.1171875, "learning_rate": 9.894435758953952e-06, "loss": 1.541, "step": 3471 }, { "epoch": 0.3694403064481805, "grad_norm": 11.8515625, "learning_rate": 9.894400227401934e-06, "loss": 1.0428, "step": 3472 }, { "epoch": 0.3695467120663971, "grad_norm": 14.0234375, "learning_rate": 9.894364695849915e-06, "loss": 1.5603, "step": 3473 }, { "epoch": 0.36965311768461373, "grad_norm": 14.0, "learning_rate": 9.894329164297897e-06, "loss": 1.5403, "step": 3474 }, { "epoch": 0.3697595233028304, "grad_norm": 12.3125, "learning_rate": 9.894293632745879e-06, "loss": 1.1149, "step": 3475 }, { "epoch": 0.369865928921047, "grad_norm": 14.21875, "learning_rate": 9.89425810119386e-06, "loss": 1.5541, "step": 3476 }, { "epoch": 0.36997233453926365, "grad_norm": 11.2734375, "learning_rate": 9.894222569641842e-06, "loss": 1.0945, "step": 3477 }, { "epoch": 0.3700787401574803, "grad_norm": 58.3125, "learning_rate": 9.894187038089823e-06, "loss": 1.16, "step": 3478 }, { "epoch": 0.37018514577569694, "grad_norm": 13.1015625, "learning_rate": 9.894151506537807e-06, "loss": 1.2315, "step": 3479 }, { "epoch": 0.3702915513939136, "grad_norm": 12.375, "learning_rate": 9.894115974985788e-06, "loss": 0.9081, "step": 3480 }, { "epoch": 0.37039795701213024, "grad_norm": 14.5078125, "learning_rate": 9.89408044343377e-06, "loss": 1.5454, "step": 3481 }, { "epoch": 0.37050436263034686, "grad_norm": 12.7734375, "learning_rate": 9.894044911881752e-06, "loss": 1.378, "step": 3482 }, { "epoch": 0.37061076824856354, "grad_norm": 20.03125, "learning_rate": 9.894009380329733e-06, "loss": 1.4141, "step": 3483 }, { "epoch": 0.37071717386678016, "grad_norm": 14.9609375, "learning_rate": 9.893973848777716e-06, "loss": 1.1796, "step": 3484 }, { "epoch": 0.37082357948499683, "grad_norm": 15.1328125, "learning_rate": 9.893938317225698e-06, "loss": 1.4389, "step": 3485 }, { "epoch": 0.37092998510321346, "grad_norm": 14.4296875, "learning_rate": 9.89390278567368e-06, "loss": 1.4589, "step": 3486 }, { "epoch": 0.3710363907214301, "grad_norm": 15.5078125, "learning_rate": 9.893867254121661e-06, "loss": 1.4864, "step": 3487 }, { "epoch": 0.37114279633964675, "grad_norm": 12.8828125, "learning_rate": 9.893831722569643e-06, "loss": 1.1576, "step": 3488 }, { "epoch": 0.3712492019578634, "grad_norm": 11.421875, "learning_rate": 9.893796191017625e-06, "loss": 1.3542, "step": 3489 }, { "epoch": 0.37135560757608, "grad_norm": 15.84375, "learning_rate": 9.893760659465606e-06, "loss": 1.2641, "step": 3490 }, { "epoch": 0.37146201319429667, "grad_norm": 14.4921875, "learning_rate": 9.893725127913588e-06, "loss": 1.4824, "step": 3491 }, { "epoch": 0.3715684188125133, "grad_norm": 15.1796875, "learning_rate": 9.89368959636157e-06, "loss": 1.4956, "step": 3492 }, { "epoch": 0.37167482443072997, "grad_norm": 11.8828125, "learning_rate": 9.893654064809551e-06, "loss": 1.0443, "step": 3493 }, { "epoch": 0.3717812300489466, "grad_norm": 9.0234375, "learning_rate": 9.893618533257533e-06, "loss": 1.1321, "step": 3494 }, { "epoch": 0.3718876356671632, "grad_norm": 12.09375, "learning_rate": 9.893583001705516e-06, "loss": 1.4657, "step": 3495 }, { "epoch": 0.3719940412853799, "grad_norm": 12.3359375, "learning_rate": 9.893547470153497e-06, "loss": 1.1882, "step": 3496 }, { "epoch": 0.3721004469035965, "grad_norm": 12.640625, "learning_rate": 9.893511938601479e-06, "loss": 1.3022, "step": 3497 }, { "epoch": 0.3722068525218131, "grad_norm": 12.9453125, "learning_rate": 9.89347640704946e-06, "loss": 1.2843, "step": 3498 }, { "epoch": 0.3723132581400298, "grad_norm": 31.5, "learning_rate": 9.893440875497442e-06, "loss": 1.5543, "step": 3499 }, { "epoch": 0.3724196637582464, "grad_norm": 17.9375, "learning_rate": 9.893405343945426e-06, "loss": 1.5657, "step": 3500 }, { "epoch": 0.3725260693764631, "grad_norm": 12.0546875, "learning_rate": 9.893369812393405e-06, "loss": 1.3531, "step": 3501 }, { "epoch": 0.3726324749946797, "grad_norm": 15.859375, "learning_rate": 9.893334280841387e-06, "loss": 1.3314, "step": 3502 }, { "epoch": 0.37273888061289634, "grad_norm": 18.0625, "learning_rate": 9.893298749289369e-06, "loss": 1.0574, "step": 3503 }, { "epoch": 0.372845286231113, "grad_norm": 19.484375, "learning_rate": 9.89326321773735e-06, "loss": 1.4359, "step": 3504 }, { "epoch": 0.37295169184932964, "grad_norm": 13.53125, "learning_rate": 9.893227686185334e-06, "loss": 1.2548, "step": 3505 }, { "epoch": 0.3730580974675463, "grad_norm": 12.71875, "learning_rate": 9.893192154633315e-06, "loss": 1.0109, "step": 3506 }, { "epoch": 0.37316450308576293, "grad_norm": 15.8671875, "learning_rate": 9.893156623081297e-06, "loss": 1.7223, "step": 3507 }, { "epoch": 0.37327090870397955, "grad_norm": 13.1484375, "learning_rate": 9.893121091529278e-06, "loss": 1.4379, "step": 3508 }, { "epoch": 0.37337731432219623, "grad_norm": 17.234375, "learning_rate": 9.89308555997726e-06, "loss": 1.2104, "step": 3509 }, { "epoch": 0.37348371994041285, "grad_norm": 14.5, "learning_rate": 9.893050028425243e-06, "loss": 1.3356, "step": 3510 }, { "epoch": 0.37359012555862947, "grad_norm": 14.4765625, "learning_rate": 9.893014496873225e-06, "loss": 1.0912, "step": 3511 }, { "epoch": 0.37369653117684615, "grad_norm": 16.921875, "learning_rate": 9.892978965321207e-06, "loss": 1.1521, "step": 3512 }, { "epoch": 0.37380293679506277, "grad_norm": 26.296875, "learning_rate": 9.892943433769188e-06, "loss": 1.277, "step": 3513 }, { "epoch": 0.37390934241327944, "grad_norm": 13.5234375, "learning_rate": 9.89290790221717e-06, "loss": 1.3737, "step": 3514 }, { "epoch": 0.37401574803149606, "grad_norm": 18.6875, "learning_rate": 9.892872370665151e-06, "loss": 1.5043, "step": 3515 }, { "epoch": 0.3741221536497127, "grad_norm": 13.03125, "learning_rate": 9.892836839113133e-06, "loss": 1.0494, "step": 3516 }, { "epoch": 0.37422855926792936, "grad_norm": 20.0625, "learning_rate": 9.892801307561115e-06, "loss": 1.7364, "step": 3517 }, { "epoch": 0.374334964886146, "grad_norm": 13.2734375, "learning_rate": 9.892765776009096e-06, "loss": 1.1383, "step": 3518 }, { "epoch": 0.3744413705043626, "grad_norm": 16.28125, "learning_rate": 9.892730244457078e-06, "loss": 1.1847, "step": 3519 }, { "epoch": 0.3745477761225793, "grad_norm": 18.3125, "learning_rate": 9.892694712905061e-06, "loss": 1.7051, "step": 3520 }, { "epoch": 0.3746541817407959, "grad_norm": 17.0, "learning_rate": 9.892659181353043e-06, "loss": 1.4598, "step": 3521 }, { "epoch": 0.3747605873590126, "grad_norm": 13.8984375, "learning_rate": 9.892623649801024e-06, "loss": 1.2654, "step": 3522 }, { "epoch": 0.3748669929772292, "grad_norm": 15.703125, "learning_rate": 9.892588118249006e-06, "loss": 1.2655, "step": 3523 }, { "epoch": 0.3749733985954458, "grad_norm": 16.03125, "learning_rate": 9.892552586696988e-06, "loss": 1.514, "step": 3524 }, { "epoch": 0.3750798042136625, "grad_norm": 11.6640625, "learning_rate": 9.89251705514497e-06, "loss": 0.9399, "step": 3525 }, { "epoch": 0.3751862098318791, "grad_norm": 20.3125, "learning_rate": 9.892481523592952e-06, "loss": 1.3533, "step": 3526 }, { "epoch": 0.3752926154500958, "grad_norm": 12.703125, "learning_rate": 9.892445992040934e-06, "loss": 1.2501, "step": 3527 }, { "epoch": 0.3753990210683124, "grad_norm": 14.6640625, "learning_rate": 9.892410460488916e-06, "loss": 1.4445, "step": 3528 }, { "epoch": 0.37550542668652903, "grad_norm": 11.421875, "learning_rate": 9.892374928936897e-06, "loss": 1.1447, "step": 3529 }, { "epoch": 0.3756118323047457, "grad_norm": 12.7578125, "learning_rate": 9.892339397384879e-06, "loss": 1.1875, "step": 3530 }, { "epoch": 0.37571823792296233, "grad_norm": 12.671875, "learning_rate": 9.89230386583286e-06, "loss": 1.3105, "step": 3531 }, { "epoch": 0.37582464354117895, "grad_norm": 14.171875, "learning_rate": 9.892268334280842e-06, "loss": 0.9988, "step": 3532 }, { "epoch": 0.3759310491593956, "grad_norm": 14.515625, "learning_rate": 9.892232802728824e-06, "loss": 0.9693, "step": 3533 }, { "epoch": 0.37603745477761225, "grad_norm": 16.03125, "learning_rate": 9.892197271176805e-06, "loss": 1.4308, "step": 3534 }, { "epoch": 0.3761438603958289, "grad_norm": 12.8046875, "learning_rate": 9.892161739624787e-06, "loss": 1.1127, "step": 3535 }, { "epoch": 0.37625026601404554, "grad_norm": 13.4453125, "learning_rate": 9.89212620807277e-06, "loss": 1.5665, "step": 3536 }, { "epoch": 0.37635667163226216, "grad_norm": 18.765625, "learning_rate": 9.892090676520752e-06, "loss": 1.3619, "step": 3537 }, { "epoch": 0.37646307725047884, "grad_norm": 18.328125, "learning_rate": 9.892055144968733e-06, "loss": 1.2364, "step": 3538 }, { "epoch": 0.37656948286869546, "grad_norm": 10.921875, "learning_rate": 9.892019613416715e-06, "loss": 1.0532, "step": 3539 }, { "epoch": 0.37667588848691214, "grad_norm": 11.9140625, "learning_rate": 9.891984081864697e-06, "loss": 1.1432, "step": 3540 }, { "epoch": 0.37678229410512876, "grad_norm": 11.515625, "learning_rate": 9.891948550312678e-06, "loss": 1.2521, "step": 3541 }, { "epoch": 0.3768886997233454, "grad_norm": 14.8359375, "learning_rate": 9.89191301876066e-06, "loss": 1.0121, "step": 3542 }, { "epoch": 0.37699510534156205, "grad_norm": 12.5, "learning_rate": 9.891877487208641e-06, "loss": 1.0311, "step": 3543 }, { "epoch": 0.3771015109597787, "grad_norm": 11.9765625, "learning_rate": 9.891841955656623e-06, "loss": 1.1764, "step": 3544 }, { "epoch": 0.3772079165779953, "grad_norm": 11.2734375, "learning_rate": 9.891806424104605e-06, "loss": 1.2721, "step": 3545 }, { "epoch": 0.37731432219621197, "grad_norm": 10.6640625, "learning_rate": 9.891770892552588e-06, "loss": 1.2109, "step": 3546 }, { "epoch": 0.3774207278144286, "grad_norm": 16.1875, "learning_rate": 9.89173536100057e-06, "loss": 1.192, "step": 3547 }, { "epoch": 0.37752713343264527, "grad_norm": 11.4140625, "learning_rate": 9.891699829448551e-06, "loss": 0.9441, "step": 3548 }, { "epoch": 0.3776335390508619, "grad_norm": 14.390625, "learning_rate": 9.891664297896533e-06, "loss": 1.4452, "step": 3549 }, { "epoch": 0.3777399446690785, "grad_norm": 16.5, "learning_rate": 9.891628766344514e-06, "loss": 1.3486, "step": 3550 }, { "epoch": 0.3778463502872952, "grad_norm": 15.3359375, "learning_rate": 9.891593234792498e-06, "loss": 1.0993, "step": 3551 }, { "epoch": 0.3779527559055118, "grad_norm": 13.703125, "learning_rate": 9.89155770324048e-06, "loss": 1.5197, "step": 3552 }, { "epoch": 0.3780591615237284, "grad_norm": 12.5625, "learning_rate": 9.891522171688461e-06, "loss": 1.2155, "step": 3553 }, { "epoch": 0.3781655671419451, "grad_norm": 15.9609375, "learning_rate": 9.891486640136443e-06, "loss": 1.1041, "step": 3554 }, { "epoch": 0.3782719727601617, "grad_norm": 13.625, "learning_rate": 9.891451108584424e-06, "loss": 1.5459, "step": 3555 }, { "epoch": 0.3783783783783784, "grad_norm": 11.2265625, "learning_rate": 9.891415577032406e-06, "loss": 0.9343, "step": 3556 }, { "epoch": 0.378484783996595, "grad_norm": 12.7109375, "learning_rate": 9.891380045480387e-06, "loss": 1.2764, "step": 3557 }, { "epoch": 0.37859118961481164, "grad_norm": 14.21875, "learning_rate": 9.891344513928369e-06, "loss": 1.3772, "step": 3558 }, { "epoch": 0.3786975952330283, "grad_norm": 21.453125, "learning_rate": 9.89130898237635e-06, "loss": 1.6531, "step": 3559 }, { "epoch": 0.37880400085124494, "grad_norm": 15.2109375, "learning_rate": 9.891273450824332e-06, "loss": 1.3103, "step": 3560 }, { "epoch": 0.3789104064694616, "grad_norm": 11.5859375, "learning_rate": 9.891237919272315e-06, "loss": 0.9603, "step": 3561 }, { "epoch": 0.37901681208767823, "grad_norm": 16.328125, "learning_rate": 9.891202387720297e-06, "loss": 1.1771, "step": 3562 }, { "epoch": 0.37912321770589485, "grad_norm": 13.0, "learning_rate": 9.891166856168279e-06, "loss": 1.3056, "step": 3563 }, { "epoch": 0.37922962332411153, "grad_norm": 17.140625, "learning_rate": 9.89113132461626e-06, "loss": 1.2464, "step": 3564 }, { "epoch": 0.37933602894232815, "grad_norm": 10.421875, "learning_rate": 9.891095793064242e-06, "loss": 1.1303, "step": 3565 }, { "epoch": 0.3794424345605448, "grad_norm": 18.59375, "learning_rate": 9.891060261512223e-06, "loss": 1.3397, "step": 3566 }, { "epoch": 0.37954884017876145, "grad_norm": 12.640625, "learning_rate": 9.891024729960205e-06, "loss": 1.0764, "step": 3567 }, { "epoch": 0.37965524579697807, "grad_norm": 15.1875, "learning_rate": 9.890989198408187e-06, "loss": 1.2743, "step": 3568 }, { "epoch": 0.37976165141519475, "grad_norm": 15.4375, "learning_rate": 9.890953666856168e-06, "loss": 1.1124, "step": 3569 }, { "epoch": 0.37986805703341137, "grad_norm": 10.8515625, "learning_rate": 9.89091813530415e-06, "loss": 0.924, "step": 3570 }, { "epoch": 0.379974462651628, "grad_norm": 21.4375, "learning_rate": 9.890882603752132e-06, "loss": 1.4247, "step": 3571 }, { "epoch": 0.38008086826984466, "grad_norm": 16.5, "learning_rate": 9.890847072200115e-06, "loss": 1.1465, "step": 3572 }, { "epoch": 0.3801872738880613, "grad_norm": 9.921875, "learning_rate": 9.890811540648096e-06, "loss": 1.2825, "step": 3573 }, { "epoch": 0.3802936795062779, "grad_norm": 12.953125, "learning_rate": 9.890776009096078e-06, "loss": 1.3759, "step": 3574 }, { "epoch": 0.3804000851244946, "grad_norm": 11.6484375, "learning_rate": 9.89074047754406e-06, "loss": 1.4867, "step": 3575 }, { "epoch": 0.3805064907427112, "grad_norm": 11.0546875, "learning_rate": 9.890704945992041e-06, "loss": 1.2328, "step": 3576 }, { "epoch": 0.3806128963609279, "grad_norm": 18.375, "learning_rate": 9.890669414440025e-06, "loss": 1.5282, "step": 3577 }, { "epoch": 0.3807193019791445, "grad_norm": 16.421875, "learning_rate": 9.890633882888006e-06, "loss": 1.4835, "step": 3578 }, { "epoch": 0.3808257075973611, "grad_norm": 13.8203125, "learning_rate": 9.890598351335988e-06, "loss": 1.2009, "step": 3579 }, { "epoch": 0.3809321132155778, "grad_norm": 17.1875, "learning_rate": 9.89056281978397e-06, "loss": 1.4912, "step": 3580 }, { "epoch": 0.3810385188337944, "grad_norm": 16.109375, "learning_rate": 9.890527288231951e-06, "loss": 1.2858, "step": 3581 }, { "epoch": 0.3811449244520111, "grad_norm": 18.265625, "learning_rate": 9.890491756679933e-06, "loss": 1.2848, "step": 3582 }, { "epoch": 0.3812513300702277, "grad_norm": 12.5546875, "learning_rate": 9.890456225127914e-06, "loss": 1.1954, "step": 3583 }, { "epoch": 0.38135773568844433, "grad_norm": 13.9296875, "learning_rate": 9.890420693575896e-06, "loss": 1.2413, "step": 3584 }, { "epoch": 0.381464141306661, "grad_norm": 10.8125, "learning_rate": 9.890385162023877e-06, "loss": 1.043, "step": 3585 }, { "epoch": 0.38157054692487763, "grad_norm": 9.8828125, "learning_rate": 9.890349630471859e-06, "loss": 1.1692, "step": 3586 }, { "epoch": 0.38167695254309425, "grad_norm": 11.640625, "learning_rate": 9.890314098919842e-06, "loss": 1.2728, "step": 3587 }, { "epoch": 0.3817833581613109, "grad_norm": 17.078125, "learning_rate": 9.890278567367824e-06, "loss": 1.3874, "step": 3588 }, { "epoch": 0.38188976377952755, "grad_norm": 13.28125, "learning_rate": 9.890243035815806e-06, "loss": 1.3053, "step": 3589 }, { "epoch": 0.3819961693977442, "grad_norm": 11.265625, "learning_rate": 9.890207504263787e-06, "loss": 1.0759, "step": 3590 }, { "epoch": 0.38210257501596084, "grad_norm": 16.375, "learning_rate": 9.890171972711769e-06, "loss": 1.3361, "step": 3591 }, { "epoch": 0.38220898063417746, "grad_norm": 12.65625, "learning_rate": 9.89013644115975e-06, "loss": 1.254, "step": 3592 }, { "epoch": 0.38231538625239414, "grad_norm": 21.796875, "learning_rate": 9.890100909607732e-06, "loss": 1.6499, "step": 3593 }, { "epoch": 0.38242179187061076, "grad_norm": 16.1875, "learning_rate": 9.890065378055714e-06, "loss": 1.621, "step": 3594 }, { "epoch": 0.38252819748882744, "grad_norm": 13.0078125, "learning_rate": 9.890029846503695e-06, "loss": 1.35, "step": 3595 }, { "epoch": 0.38263460310704406, "grad_norm": 12.5859375, "learning_rate": 9.889994314951677e-06, "loss": 1.0939, "step": 3596 }, { "epoch": 0.3827410087252607, "grad_norm": 10.53125, "learning_rate": 9.88995878339966e-06, "loss": 1.2328, "step": 3597 }, { "epoch": 0.38284741434347735, "grad_norm": 20.765625, "learning_rate": 9.889923251847642e-06, "loss": 1.4989, "step": 3598 }, { "epoch": 0.382953819961694, "grad_norm": 14.65625, "learning_rate": 9.889887720295623e-06, "loss": 1.491, "step": 3599 }, { "epoch": 0.3830602255799106, "grad_norm": 18.53125, "learning_rate": 9.889852188743605e-06, "loss": 1.3713, "step": 3600 }, { "epoch": 0.38316663119812727, "grad_norm": 10.8515625, "learning_rate": 9.889816657191586e-06, "loss": 1.2926, "step": 3601 }, { "epoch": 0.3832730368163439, "grad_norm": 11.703125, "learning_rate": 9.88978112563957e-06, "loss": 1.2388, "step": 3602 }, { "epoch": 0.38337944243456057, "grad_norm": 13.703125, "learning_rate": 9.889745594087551e-06, "loss": 1.2975, "step": 3603 }, { "epoch": 0.3834858480527772, "grad_norm": 14.6796875, "learning_rate": 9.889710062535533e-06, "loss": 1.3381, "step": 3604 }, { "epoch": 0.3835922536709938, "grad_norm": 19.5625, "learning_rate": 9.889674530983515e-06, "loss": 1.2994, "step": 3605 }, { "epoch": 0.3836986592892105, "grad_norm": 17.484375, "learning_rate": 9.889638999431496e-06, "loss": 1.5716, "step": 3606 }, { "epoch": 0.3838050649074271, "grad_norm": 16.140625, "learning_rate": 9.889603467879478e-06, "loss": 1.0998, "step": 3607 }, { "epoch": 0.3839114705256437, "grad_norm": 20.578125, "learning_rate": 9.88956793632746e-06, "loss": 1.6342, "step": 3608 }, { "epoch": 0.3840178761438604, "grad_norm": 13.0859375, "learning_rate": 9.889532404775441e-06, "loss": 1.3397, "step": 3609 }, { "epoch": 0.384124281762077, "grad_norm": 12.8984375, "learning_rate": 9.889496873223423e-06, "loss": 1.2041, "step": 3610 }, { "epoch": 0.3842306873802937, "grad_norm": 16.9375, "learning_rate": 9.889461341671404e-06, "loss": 1.1526, "step": 3611 }, { "epoch": 0.3843370929985103, "grad_norm": 12.515625, "learning_rate": 9.889425810119386e-06, "loss": 1.2024, "step": 3612 }, { "epoch": 0.38444349861672694, "grad_norm": 16.90625, "learning_rate": 9.88939027856737e-06, "loss": 1.086, "step": 3613 }, { "epoch": 0.3845499042349436, "grad_norm": 18.15625, "learning_rate": 9.88935474701535e-06, "loss": 1.8018, "step": 3614 }, { "epoch": 0.38465630985316024, "grad_norm": 15.21875, "learning_rate": 9.889319215463332e-06, "loss": 1.4944, "step": 3615 }, { "epoch": 0.3847627154713769, "grad_norm": 12.5703125, "learning_rate": 9.889283683911314e-06, "loss": 1.3423, "step": 3616 }, { "epoch": 0.38486912108959354, "grad_norm": 11.171875, "learning_rate": 9.889248152359296e-06, "loss": 0.9495, "step": 3617 }, { "epoch": 0.38497552670781016, "grad_norm": 16.5625, "learning_rate": 9.889212620807279e-06, "loss": 1.7955, "step": 3618 }, { "epoch": 0.38508193232602683, "grad_norm": 15.0390625, "learning_rate": 9.88917708925526e-06, "loss": 1.0577, "step": 3619 }, { "epoch": 0.38518833794424345, "grad_norm": 14.9375, "learning_rate": 9.88914155770324e-06, "loss": 1.2684, "step": 3620 }, { "epoch": 0.3852947435624601, "grad_norm": 14.4921875, "learning_rate": 9.889106026151222e-06, "loss": 1.4653, "step": 3621 }, { "epoch": 0.38540114918067675, "grad_norm": 12.5078125, "learning_rate": 9.889070494599204e-06, "loss": 1.194, "step": 3622 }, { "epoch": 0.38550755479889337, "grad_norm": 13.2109375, "learning_rate": 9.889034963047187e-06, "loss": 1.2651, "step": 3623 }, { "epoch": 0.38561396041711005, "grad_norm": 16.34375, "learning_rate": 9.888999431495169e-06, "loss": 1.5546, "step": 3624 }, { "epoch": 0.38572036603532667, "grad_norm": 14.6015625, "learning_rate": 9.88896389994315e-06, "loss": 1.2455, "step": 3625 }, { "epoch": 0.3858267716535433, "grad_norm": 15.546875, "learning_rate": 9.888928368391132e-06, "loss": 1.3321, "step": 3626 }, { "epoch": 0.38593317727175996, "grad_norm": 14.0390625, "learning_rate": 9.888892836839113e-06, "loss": 1.1681, "step": 3627 }, { "epoch": 0.3860395828899766, "grad_norm": 14.34375, "learning_rate": 9.888857305287097e-06, "loss": 1.3701, "step": 3628 }, { "epoch": 0.3861459885081932, "grad_norm": 13.953125, "learning_rate": 9.888821773735078e-06, "loss": 1.1287, "step": 3629 }, { "epoch": 0.3862523941264099, "grad_norm": 14.5703125, "learning_rate": 9.88878624218306e-06, "loss": 1.6055, "step": 3630 }, { "epoch": 0.3863587997446265, "grad_norm": 15.0859375, "learning_rate": 9.888750710631041e-06, "loss": 1.6383, "step": 3631 }, { "epoch": 0.3864652053628432, "grad_norm": 13.3046875, "learning_rate": 9.888715179079023e-06, "loss": 1.2731, "step": 3632 }, { "epoch": 0.3865716109810598, "grad_norm": 14.3046875, "learning_rate": 9.888679647527005e-06, "loss": 1.2009, "step": 3633 }, { "epoch": 0.3866780165992764, "grad_norm": 20.59375, "learning_rate": 9.888644115974986e-06, "loss": 1.3515, "step": 3634 }, { "epoch": 0.3867844222174931, "grad_norm": 14.59375, "learning_rate": 9.888608584422968e-06, "loss": 1.4118, "step": 3635 }, { "epoch": 0.3868908278357097, "grad_norm": 12.359375, "learning_rate": 9.88857305287095e-06, "loss": 1.3797, "step": 3636 }, { "epoch": 0.3869972334539264, "grad_norm": 16.53125, "learning_rate": 9.888537521318931e-06, "loss": 1.4156, "step": 3637 }, { "epoch": 0.387103639072143, "grad_norm": 15.578125, "learning_rate": 9.888501989766914e-06, "loss": 1.3484, "step": 3638 }, { "epoch": 0.38721004469035963, "grad_norm": 13.3828125, "learning_rate": 9.888466458214896e-06, "loss": 1.3018, "step": 3639 }, { "epoch": 0.3873164503085763, "grad_norm": 12.09375, "learning_rate": 9.888430926662878e-06, "loss": 1.0424, "step": 3640 }, { "epoch": 0.38742285592679293, "grad_norm": 12.015625, "learning_rate": 9.88839539511086e-06, "loss": 1.3721, "step": 3641 }, { "epoch": 0.38752926154500955, "grad_norm": 15.0625, "learning_rate": 9.88835986355884e-06, "loss": 1.5348, "step": 3642 }, { "epoch": 0.3876356671632262, "grad_norm": 13.5546875, "learning_rate": 9.888324332006824e-06, "loss": 1.3041, "step": 3643 }, { "epoch": 0.38774207278144285, "grad_norm": 13.9375, "learning_rate": 9.888288800454806e-06, "loss": 1.1961, "step": 3644 }, { "epoch": 0.3878484783996595, "grad_norm": 18.375, "learning_rate": 9.888253268902787e-06, "loss": 1.3023, "step": 3645 }, { "epoch": 0.38795488401787614, "grad_norm": 12.5234375, "learning_rate": 9.888217737350769e-06, "loss": 1.2409, "step": 3646 }, { "epoch": 0.38806128963609277, "grad_norm": 11.8984375, "learning_rate": 9.88818220579875e-06, "loss": 1.2539, "step": 3647 }, { "epoch": 0.38816769525430944, "grad_norm": 11.4375, "learning_rate": 9.888146674246732e-06, "loss": 0.9573, "step": 3648 }, { "epoch": 0.38827410087252606, "grad_norm": 12.6953125, "learning_rate": 9.888111142694714e-06, "loss": 1.1116, "step": 3649 }, { "epoch": 0.38838050649074274, "grad_norm": 13.328125, "learning_rate": 9.888075611142695e-06, "loss": 1.2862, "step": 3650 }, { "epoch": 0.38848691210895936, "grad_norm": 15.0859375, "learning_rate": 9.888040079590677e-06, "loss": 1.7484, "step": 3651 }, { "epoch": 0.388593317727176, "grad_norm": 18.390625, "learning_rate": 9.888004548038659e-06, "loss": 1.3827, "step": 3652 }, { "epoch": 0.38869972334539266, "grad_norm": 15.0234375, "learning_rate": 9.88796901648664e-06, "loss": 1.2281, "step": 3653 }, { "epoch": 0.3888061289636093, "grad_norm": 13.71875, "learning_rate": 9.887933484934623e-06, "loss": 1.2874, "step": 3654 }, { "epoch": 0.3889125345818259, "grad_norm": 13.2578125, "learning_rate": 9.887897953382605e-06, "loss": 1.3356, "step": 3655 }, { "epoch": 0.3890189402000426, "grad_norm": 14.3359375, "learning_rate": 9.887862421830587e-06, "loss": 1.44, "step": 3656 }, { "epoch": 0.3891253458182592, "grad_norm": 13.8203125, "learning_rate": 9.887826890278568e-06, "loss": 1.2645, "step": 3657 }, { "epoch": 0.38923175143647587, "grad_norm": 11.5625, "learning_rate": 9.88779135872655e-06, "loss": 1.4297, "step": 3658 }, { "epoch": 0.3893381570546925, "grad_norm": 13.484375, "learning_rate": 9.887755827174532e-06, "loss": 1.1199, "step": 3659 }, { "epoch": 0.3894445626729091, "grad_norm": 61.0625, "learning_rate": 9.887720295622513e-06, "loss": 1.6436, "step": 3660 }, { "epoch": 0.3895509682911258, "grad_norm": 13.921875, "learning_rate": 9.887684764070495e-06, "loss": 1.1463, "step": 3661 }, { "epoch": 0.3896573739093424, "grad_norm": 18.03125, "learning_rate": 9.887649232518476e-06, "loss": 1.3143, "step": 3662 }, { "epoch": 0.38976377952755903, "grad_norm": 11.9609375, "learning_rate": 9.887613700966458e-06, "loss": 1.0257, "step": 3663 }, { "epoch": 0.3898701851457757, "grad_norm": 16.828125, "learning_rate": 9.887578169414441e-06, "loss": 1.5885, "step": 3664 }, { "epoch": 0.3899765907639923, "grad_norm": 13.6875, "learning_rate": 9.887542637862423e-06, "loss": 1.263, "step": 3665 }, { "epoch": 0.390082996382209, "grad_norm": 13.7734375, "learning_rate": 9.887507106310404e-06, "loss": 1.3246, "step": 3666 }, { "epoch": 0.3901894020004256, "grad_norm": 12.8828125, "learning_rate": 9.887471574758386e-06, "loss": 1.1732, "step": 3667 }, { "epoch": 0.39029580761864224, "grad_norm": 10.4453125, "learning_rate": 9.887436043206368e-06, "loss": 1.1267, "step": 3668 }, { "epoch": 0.3904022132368589, "grad_norm": 19.921875, "learning_rate": 9.887400511654351e-06, "loss": 1.4293, "step": 3669 }, { "epoch": 0.39050861885507554, "grad_norm": 14.0625, "learning_rate": 9.887364980102333e-06, "loss": 1.5591, "step": 3670 }, { "epoch": 0.3906150244732922, "grad_norm": 16.21875, "learning_rate": 9.887329448550314e-06, "loss": 1.49, "step": 3671 }, { "epoch": 0.39072143009150884, "grad_norm": 14.890625, "learning_rate": 9.887293916998296e-06, "loss": 1.2502, "step": 3672 }, { "epoch": 0.39082783570972546, "grad_norm": 15.1015625, "learning_rate": 9.887258385446277e-06, "loss": 1.2844, "step": 3673 }, { "epoch": 0.39093424132794213, "grad_norm": 11.8671875, "learning_rate": 9.887222853894259e-06, "loss": 0.9726, "step": 3674 }, { "epoch": 0.39104064694615875, "grad_norm": 18.859375, "learning_rate": 9.88718732234224e-06, "loss": 1.2857, "step": 3675 }, { "epoch": 0.3911470525643754, "grad_norm": 14.203125, "learning_rate": 9.887151790790222e-06, "loss": 1.5064, "step": 3676 }, { "epoch": 0.39125345818259205, "grad_norm": 27.625, "learning_rate": 9.887116259238204e-06, "loss": 1.1428, "step": 3677 }, { "epoch": 0.39135986380080867, "grad_norm": 16.296875, "learning_rate": 9.887080727686185e-06, "loss": 1.1336, "step": 3678 }, { "epoch": 0.39146626941902535, "grad_norm": 19.140625, "learning_rate": 9.887045196134169e-06, "loss": 1.2059, "step": 3679 }, { "epoch": 0.39157267503724197, "grad_norm": 13.921875, "learning_rate": 9.88700966458215e-06, "loss": 1.2241, "step": 3680 }, { "epoch": 0.3916790806554586, "grad_norm": 20.84375, "learning_rate": 9.886974133030132e-06, "loss": 1.166, "step": 3681 }, { "epoch": 0.39178548627367527, "grad_norm": 16.40625, "learning_rate": 9.886938601478114e-06, "loss": 1.2624, "step": 3682 }, { "epoch": 0.3918918918918919, "grad_norm": 17.875, "learning_rate": 9.886903069926095e-06, "loss": 1.249, "step": 3683 }, { "epoch": 0.3919982975101085, "grad_norm": 12.671875, "learning_rate": 9.886867538374077e-06, "loss": 1.2255, "step": 3684 }, { "epoch": 0.3921047031283252, "grad_norm": 18.09375, "learning_rate": 9.886832006822058e-06, "loss": 1.3532, "step": 3685 }, { "epoch": 0.3922111087465418, "grad_norm": 13.4921875, "learning_rate": 9.88679647527004e-06, "loss": 1.2584, "step": 3686 }, { "epoch": 0.3923175143647585, "grad_norm": 12.3984375, "learning_rate": 9.886760943718022e-06, "loss": 1.201, "step": 3687 }, { "epoch": 0.3924239199829751, "grad_norm": 16.0, "learning_rate": 9.886725412166003e-06, "loss": 1.4145, "step": 3688 }, { "epoch": 0.3925303256011917, "grad_norm": 15.734375, "learning_rate": 9.886689880613985e-06, "loss": 0.9361, "step": 3689 }, { "epoch": 0.3926367312194084, "grad_norm": 11.9140625, "learning_rate": 9.886654349061968e-06, "loss": 1.0403, "step": 3690 }, { "epoch": 0.392743136837625, "grad_norm": 33.46875, "learning_rate": 9.88661881750995e-06, "loss": 1.317, "step": 3691 }, { "epoch": 0.3928495424558417, "grad_norm": 20.40625, "learning_rate": 9.886583285957931e-06, "loss": 1.2287, "step": 3692 }, { "epoch": 0.3929559480740583, "grad_norm": 14.3046875, "learning_rate": 9.886547754405913e-06, "loss": 1.2179, "step": 3693 }, { "epoch": 0.39306235369227494, "grad_norm": 10.5859375, "learning_rate": 9.886512222853895e-06, "loss": 0.9193, "step": 3694 }, { "epoch": 0.3931687593104916, "grad_norm": 15.03125, "learning_rate": 9.886476691301878e-06, "loss": 1.0993, "step": 3695 }, { "epoch": 0.39327516492870823, "grad_norm": 11.171875, "learning_rate": 9.88644115974986e-06, "loss": 0.8515, "step": 3696 }, { "epoch": 0.39338157054692485, "grad_norm": 14.8984375, "learning_rate": 9.886405628197841e-06, "loss": 1.4623, "step": 3697 }, { "epoch": 0.39348797616514153, "grad_norm": 14.6015625, "learning_rate": 9.886370096645823e-06, "loss": 1.1266, "step": 3698 }, { "epoch": 0.39359438178335815, "grad_norm": 15.8203125, "learning_rate": 9.886334565093804e-06, "loss": 1.5458, "step": 3699 }, { "epoch": 0.3937007874015748, "grad_norm": 14.828125, "learning_rate": 9.886299033541786e-06, "loss": 1.1165, "step": 3700 }, { "epoch": 0.39380719301979145, "grad_norm": 13.5390625, "learning_rate": 9.886263501989767e-06, "loss": 1.2104, "step": 3701 }, { "epoch": 0.39391359863800807, "grad_norm": 22.203125, "learning_rate": 9.886227970437749e-06, "loss": 1.4477, "step": 3702 }, { "epoch": 0.39402000425622474, "grad_norm": 24.140625, "learning_rate": 9.88619243888573e-06, "loss": 1.4069, "step": 3703 }, { "epoch": 0.39412640987444136, "grad_norm": 15.6875, "learning_rate": 9.886156907333712e-06, "loss": 1.4953, "step": 3704 }, { "epoch": 0.39423281549265804, "grad_norm": 19.296875, "learning_rate": 9.886121375781696e-06, "loss": 1.5155, "step": 3705 }, { "epoch": 0.39433922111087466, "grad_norm": 15.15625, "learning_rate": 9.886085844229677e-06, "loss": 0.9223, "step": 3706 }, { "epoch": 0.3944456267290913, "grad_norm": 11.8046875, "learning_rate": 9.886050312677659e-06, "loss": 1.2961, "step": 3707 }, { "epoch": 0.39455203234730796, "grad_norm": 14.6953125, "learning_rate": 9.88601478112564e-06, "loss": 1.3319, "step": 3708 }, { "epoch": 0.3946584379655246, "grad_norm": 12.4140625, "learning_rate": 9.885979249573622e-06, "loss": 1.62, "step": 3709 }, { "epoch": 0.3947648435837412, "grad_norm": 18.640625, "learning_rate": 9.885943718021604e-06, "loss": 1.447, "step": 3710 }, { "epoch": 0.3948712492019579, "grad_norm": 13.6953125, "learning_rate": 9.885908186469585e-06, "loss": 1.3549, "step": 3711 }, { "epoch": 0.3949776548201745, "grad_norm": 19.734375, "learning_rate": 9.885872654917567e-06, "loss": 1.4365, "step": 3712 }, { "epoch": 0.39508406043839117, "grad_norm": 13.609375, "learning_rate": 9.885837123365548e-06, "loss": 1.2993, "step": 3713 }, { "epoch": 0.3951904660566078, "grad_norm": 13.5, "learning_rate": 9.88580159181353e-06, "loss": 1.2597, "step": 3714 }, { "epoch": 0.3952968716748244, "grad_norm": 14.390625, "learning_rate": 9.885766060261513e-06, "loss": 1.4538, "step": 3715 }, { "epoch": 0.3954032772930411, "grad_norm": 18.421875, "learning_rate": 9.885730528709495e-06, "loss": 1.5198, "step": 3716 }, { "epoch": 0.3955096829112577, "grad_norm": 21.734375, "learning_rate": 9.885694997157477e-06, "loss": 1.3197, "step": 3717 }, { "epoch": 0.39561608852947433, "grad_norm": 16.21875, "learning_rate": 9.885659465605458e-06, "loss": 1.1632, "step": 3718 }, { "epoch": 0.395722494147691, "grad_norm": 16.515625, "learning_rate": 9.88562393405344e-06, "loss": 1.1918, "step": 3719 }, { "epoch": 0.3958288997659076, "grad_norm": 13.1875, "learning_rate": 9.885588402501423e-06, "loss": 1.2822, "step": 3720 }, { "epoch": 0.3959353053841243, "grad_norm": 18.484375, "learning_rate": 9.885552870949405e-06, "loss": 1.4438, "step": 3721 }, { "epoch": 0.3960417110023409, "grad_norm": 11.3984375, "learning_rate": 9.885517339397386e-06, "loss": 1.2746, "step": 3722 }, { "epoch": 0.39614811662055754, "grad_norm": 13.3984375, "learning_rate": 9.885481807845368e-06, "loss": 1.2341, "step": 3723 }, { "epoch": 0.3962545222387742, "grad_norm": 17.203125, "learning_rate": 9.88544627629335e-06, "loss": 1.329, "step": 3724 }, { "epoch": 0.39636092785699084, "grad_norm": 14.1640625, "learning_rate": 9.885410744741331e-06, "loss": 1.1451, "step": 3725 }, { "epoch": 0.3964673334752075, "grad_norm": 12.0703125, "learning_rate": 9.885375213189313e-06, "loss": 1.2371, "step": 3726 }, { "epoch": 0.39657373909342414, "grad_norm": 13.203125, "learning_rate": 9.885339681637294e-06, "loss": 1.4355, "step": 3727 }, { "epoch": 0.39668014471164076, "grad_norm": 12.53125, "learning_rate": 9.885304150085276e-06, "loss": 1.1674, "step": 3728 }, { "epoch": 0.39678655032985743, "grad_norm": 18.703125, "learning_rate": 9.885268618533258e-06, "loss": 1.5677, "step": 3729 }, { "epoch": 0.39689295594807406, "grad_norm": 11.265625, "learning_rate": 9.88523308698124e-06, "loss": 1.2241, "step": 3730 }, { "epoch": 0.3969993615662907, "grad_norm": 15.2421875, "learning_rate": 9.885197555429222e-06, "loss": 1.4067, "step": 3731 }, { "epoch": 0.39710576718450735, "grad_norm": 17.84375, "learning_rate": 9.885162023877204e-06, "loss": 1.2605, "step": 3732 }, { "epoch": 0.397212172802724, "grad_norm": 16.703125, "learning_rate": 9.885126492325186e-06, "loss": 1.4694, "step": 3733 }, { "epoch": 0.39731857842094065, "grad_norm": 14.4765625, "learning_rate": 9.885090960773167e-06, "loss": 1.3943, "step": 3734 }, { "epoch": 0.39742498403915727, "grad_norm": 14.7109375, "learning_rate": 9.885055429221149e-06, "loss": 1.3434, "step": 3735 }, { "epoch": 0.3975313896573739, "grad_norm": 14.015625, "learning_rate": 9.885019897669132e-06, "loss": 1.4095, "step": 3736 }, { "epoch": 0.39763779527559057, "grad_norm": 14.1484375, "learning_rate": 9.884984366117114e-06, "loss": 1.2394, "step": 3737 }, { "epoch": 0.3977442008938072, "grad_norm": 12.4296875, "learning_rate": 9.884948834565095e-06, "loss": 1.0854, "step": 3738 }, { "epoch": 0.3978506065120238, "grad_norm": 18.28125, "learning_rate": 9.884913303013077e-06, "loss": 1.2707, "step": 3739 }, { "epoch": 0.3979570121302405, "grad_norm": 14.6796875, "learning_rate": 9.884877771461057e-06, "loss": 1.6322, "step": 3740 }, { "epoch": 0.3980634177484571, "grad_norm": 13.3515625, "learning_rate": 9.88484223990904e-06, "loss": 1.298, "step": 3741 }, { "epoch": 0.3981698233666738, "grad_norm": 17.359375, "learning_rate": 9.884806708357022e-06, "loss": 1.365, "step": 3742 }, { "epoch": 0.3982762289848904, "grad_norm": 16.46875, "learning_rate": 9.884771176805003e-06, "loss": 1.009, "step": 3743 }, { "epoch": 0.398382634603107, "grad_norm": 14.296875, "learning_rate": 9.884735645252985e-06, "loss": 1.4757, "step": 3744 }, { "epoch": 0.3984890402213237, "grad_norm": 15.828125, "learning_rate": 9.884700113700967e-06, "loss": 1.408, "step": 3745 }, { "epoch": 0.3985954458395403, "grad_norm": 15.6640625, "learning_rate": 9.88466458214895e-06, "loss": 1.5896, "step": 3746 }, { "epoch": 0.398701851457757, "grad_norm": 12.53125, "learning_rate": 9.884629050596932e-06, "loss": 1.0887, "step": 3747 }, { "epoch": 0.3988082570759736, "grad_norm": 19.90625, "learning_rate": 9.884593519044913e-06, "loss": 1.199, "step": 3748 }, { "epoch": 0.39891466269419024, "grad_norm": 17.640625, "learning_rate": 9.884557987492895e-06, "loss": 1.7314, "step": 3749 }, { "epoch": 0.3990210683124069, "grad_norm": 12.671875, "learning_rate": 9.884522455940876e-06, "loss": 1.2666, "step": 3750 }, { "epoch": 0.39912747393062353, "grad_norm": 15.5625, "learning_rate": 9.884486924388858e-06, "loss": 1.1554, "step": 3751 }, { "epoch": 0.39923387954884015, "grad_norm": 16.859375, "learning_rate": 9.88445139283684e-06, "loss": 1.3881, "step": 3752 }, { "epoch": 0.39934028516705683, "grad_norm": 15.7578125, "learning_rate": 9.884415861284821e-06, "loss": 1.4745, "step": 3753 }, { "epoch": 0.39944669078527345, "grad_norm": 17.03125, "learning_rate": 9.884380329732803e-06, "loss": 1.2653, "step": 3754 }, { "epoch": 0.3995530964034901, "grad_norm": 15.1484375, "learning_rate": 9.884344798180784e-06, "loss": 1.2578, "step": 3755 }, { "epoch": 0.39965950202170675, "grad_norm": 17.4375, "learning_rate": 9.884309266628768e-06, "loss": 1.6387, "step": 3756 }, { "epoch": 0.39976590763992337, "grad_norm": 12.8984375, "learning_rate": 9.88427373507675e-06, "loss": 1.1005, "step": 3757 }, { "epoch": 0.39987231325814004, "grad_norm": 17.34375, "learning_rate": 9.884238203524731e-06, "loss": 1.5603, "step": 3758 }, { "epoch": 0.39997871887635666, "grad_norm": 12.6953125, "learning_rate": 9.884202671972713e-06, "loss": 1.3901, "step": 3759 }, { "epoch": 0.40008512449457334, "grad_norm": 14.953125, "learning_rate": 9.884167140420694e-06, "loss": 1.3156, "step": 3760 }, { "epoch": 0.40019153011278996, "grad_norm": 14.4921875, "learning_rate": 9.884131608868677e-06, "loss": 1.1869, "step": 3761 }, { "epoch": 0.4002979357310066, "grad_norm": 13.7890625, "learning_rate": 9.884096077316659e-06, "loss": 0.6776, "step": 3762 }, { "epoch": 0.40040434134922326, "grad_norm": 23.765625, "learning_rate": 9.88406054576464e-06, "loss": 1.2992, "step": 3763 }, { "epoch": 0.4005107469674399, "grad_norm": 11.6171875, "learning_rate": 9.884025014212622e-06, "loss": 1.2674, "step": 3764 }, { "epoch": 0.4006171525856565, "grad_norm": 12.828125, "learning_rate": 9.883989482660604e-06, "loss": 1.2107, "step": 3765 }, { "epoch": 0.4007235582038732, "grad_norm": 12.046875, "learning_rate": 9.883953951108585e-06, "loss": 1.1669, "step": 3766 }, { "epoch": 0.4008299638220898, "grad_norm": 14.421875, "learning_rate": 9.883918419556567e-06, "loss": 1.1786, "step": 3767 }, { "epoch": 0.4009363694403065, "grad_norm": 26.5, "learning_rate": 9.883882888004549e-06, "loss": 2.4238, "step": 3768 }, { "epoch": 0.4010427750585231, "grad_norm": 11.9609375, "learning_rate": 9.88384735645253e-06, "loss": 1.1783, "step": 3769 }, { "epoch": 0.4011491806767397, "grad_norm": 19.0, "learning_rate": 9.883811824900512e-06, "loss": 1.5863, "step": 3770 }, { "epoch": 0.4012555862949564, "grad_norm": 11.6640625, "learning_rate": 9.883776293348494e-06, "loss": 1.11, "step": 3771 }, { "epoch": 0.401361991913173, "grad_norm": 13.125, "learning_rate": 9.883740761796477e-06, "loss": 1.4467, "step": 3772 }, { "epoch": 0.40146839753138963, "grad_norm": 11.8984375, "learning_rate": 9.883705230244458e-06, "loss": 1.4618, "step": 3773 }, { "epoch": 0.4015748031496063, "grad_norm": 11.6328125, "learning_rate": 9.88366969869244e-06, "loss": 1.2954, "step": 3774 }, { "epoch": 0.40168120876782293, "grad_norm": 10.0859375, "learning_rate": 9.883634167140422e-06, "loss": 1.1909, "step": 3775 }, { "epoch": 0.4017876143860396, "grad_norm": 16.515625, "learning_rate": 9.883598635588403e-06, "loss": 1.1663, "step": 3776 }, { "epoch": 0.4018940200042562, "grad_norm": 19.9375, "learning_rate": 9.883563104036385e-06, "loss": 0.8963, "step": 3777 }, { "epoch": 0.40200042562247285, "grad_norm": 13.5390625, "learning_rate": 9.883527572484366e-06, "loss": 1.3961, "step": 3778 }, { "epoch": 0.4021068312406895, "grad_norm": 17.90625, "learning_rate": 9.883492040932348e-06, "loss": 1.4243, "step": 3779 }, { "epoch": 0.40221323685890614, "grad_norm": 14.6796875, "learning_rate": 9.88345650938033e-06, "loss": 1.2535, "step": 3780 }, { "epoch": 0.4023196424771228, "grad_norm": 12.59375, "learning_rate": 9.883420977828311e-06, "loss": 1.0879, "step": 3781 }, { "epoch": 0.40242604809533944, "grad_norm": 13.71875, "learning_rate": 9.883385446276295e-06, "loss": 1.0235, "step": 3782 }, { "epoch": 0.40253245371355606, "grad_norm": 15.0703125, "learning_rate": 9.883349914724276e-06, "loss": 1.3865, "step": 3783 }, { "epoch": 0.40263885933177274, "grad_norm": 13.203125, "learning_rate": 9.883314383172258e-06, "loss": 1.7138, "step": 3784 }, { "epoch": 0.40274526494998936, "grad_norm": 12.921875, "learning_rate": 9.88327885162024e-06, "loss": 1.2621, "step": 3785 }, { "epoch": 0.402851670568206, "grad_norm": 13.9140625, "learning_rate": 9.883243320068221e-06, "loss": 1.0962, "step": 3786 }, { "epoch": 0.40295807618642265, "grad_norm": 11.9375, "learning_rate": 9.883207788516204e-06, "loss": 1.3747, "step": 3787 }, { "epoch": 0.4030644818046393, "grad_norm": 13.5, "learning_rate": 9.883172256964186e-06, "loss": 1.4714, "step": 3788 }, { "epoch": 0.40317088742285595, "grad_norm": 14.546875, "learning_rate": 9.883136725412168e-06, "loss": 1.657, "step": 3789 }, { "epoch": 0.40327729304107257, "grad_norm": 16.421875, "learning_rate": 9.883101193860149e-06, "loss": 1.3571, "step": 3790 }, { "epoch": 0.4033836986592892, "grad_norm": 12.828125, "learning_rate": 9.88306566230813e-06, "loss": 1.3506, "step": 3791 }, { "epoch": 0.40349010427750587, "grad_norm": 14.4609375, "learning_rate": 9.883030130756112e-06, "loss": 1.5201, "step": 3792 }, { "epoch": 0.4035965098957225, "grad_norm": 15.21875, "learning_rate": 9.882994599204094e-06, "loss": 0.9804, "step": 3793 }, { "epoch": 0.4037029155139391, "grad_norm": 14.6484375, "learning_rate": 9.882959067652076e-06, "loss": 1.3466, "step": 3794 }, { "epoch": 0.4038093211321558, "grad_norm": 17.546875, "learning_rate": 9.882923536100057e-06, "loss": 1.3743, "step": 3795 }, { "epoch": 0.4039157267503724, "grad_norm": 14.3203125, "learning_rate": 9.882888004548039e-06, "loss": 1.2828, "step": 3796 }, { "epoch": 0.4040221323685891, "grad_norm": 13.890625, "learning_rate": 9.882852472996022e-06, "loss": 1.3321, "step": 3797 }, { "epoch": 0.4041285379868057, "grad_norm": 12.3984375, "learning_rate": 9.882816941444004e-06, "loss": 1.211, "step": 3798 }, { "epoch": 0.4042349436050223, "grad_norm": 17.671875, "learning_rate": 9.882781409891985e-06, "loss": 1.4095, "step": 3799 }, { "epoch": 0.404341349223239, "grad_norm": 18.390625, "learning_rate": 9.882745878339967e-06, "loss": 1.3168, "step": 3800 }, { "epoch": 0.4044477548414556, "grad_norm": 17.65625, "learning_rate": 9.882710346787948e-06, "loss": 1.3231, "step": 3801 }, { "epoch": 0.4045541604596723, "grad_norm": 12.5078125, "learning_rate": 9.88267481523593e-06, "loss": 1.3171, "step": 3802 }, { "epoch": 0.4046605660778889, "grad_norm": 13.703125, "learning_rate": 9.882639283683912e-06, "loss": 0.9948, "step": 3803 }, { "epoch": 0.40476697169610554, "grad_norm": 17.5625, "learning_rate": 9.882603752131893e-06, "loss": 1.3682, "step": 3804 }, { "epoch": 0.4048733773143222, "grad_norm": 19.71875, "learning_rate": 9.882568220579875e-06, "loss": 1.2845, "step": 3805 }, { "epoch": 0.40497978293253883, "grad_norm": 14.1640625, "learning_rate": 9.882532689027857e-06, "loss": 1.6675, "step": 3806 }, { "epoch": 0.40508618855075545, "grad_norm": 12.2421875, "learning_rate": 9.882497157475838e-06, "loss": 1.4573, "step": 3807 }, { "epoch": 0.40519259416897213, "grad_norm": 11.15625, "learning_rate": 9.882461625923821e-06, "loss": 1.0019, "step": 3808 }, { "epoch": 0.40529899978718875, "grad_norm": 12.59375, "learning_rate": 9.882426094371803e-06, "loss": 1.4623, "step": 3809 }, { "epoch": 0.40540540540540543, "grad_norm": 17.9375, "learning_rate": 9.882390562819785e-06, "loss": 1.4781, "step": 3810 }, { "epoch": 0.40551181102362205, "grad_norm": 11.6953125, "learning_rate": 9.882355031267766e-06, "loss": 1.0583, "step": 3811 }, { "epoch": 0.40561821664183867, "grad_norm": 15.578125, "learning_rate": 9.882319499715748e-06, "loss": 1.2587, "step": 3812 }, { "epoch": 0.40572462226005535, "grad_norm": 11.1875, "learning_rate": 9.882283968163731e-06, "loss": 1.2613, "step": 3813 }, { "epoch": 0.40583102787827197, "grad_norm": 15.6953125, "learning_rate": 9.882248436611713e-06, "loss": 0.9738, "step": 3814 }, { "epoch": 0.40593743349648864, "grad_norm": 12.7890625, "learning_rate": 9.882212905059694e-06, "loss": 1.1153, "step": 3815 }, { "epoch": 0.40604383911470526, "grad_norm": 13.734375, "learning_rate": 9.882177373507676e-06, "loss": 1.135, "step": 3816 }, { "epoch": 0.4061502447329219, "grad_norm": 16.0, "learning_rate": 9.882141841955658e-06, "loss": 1.1972, "step": 3817 }, { "epoch": 0.40625665035113856, "grad_norm": 18.21875, "learning_rate": 9.88210631040364e-06, "loss": 1.5276, "step": 3818 }, { "epoch": 0.4063630559693552, "grad_norm": 14.96875, "learning_rate": 9.88207077885162e-06, "loss": 1.3794, "step": 3819 }, { "epoch": 0.4064694615875718, "grad_norm": 9.796875, "learning_rate": 9.882035247299602e-06, "loss": 1.1027, "step": 3820 }, { "epoch": 0.4065758672057885, "grad_norm": 17.6875, "learning_rate": 9.881999715747584e-06, "loss": 1.3618, "step": 3821 }, { "epoch": 0.4066822728240051, "grad_norm": 15.1796875, "learning_rate": 9.881964184195566e-06, "loss": 1.5924, "step": 3822 }, { "epoch": 0.4067886784422218, "grad_norm": 14.6015625, "learning_rate": 9.881928652643549e-06, "loss": 1.1862, "step": 3823 }, { "epoch": 0.4068950840604384, "grad_norm": 14.359375, "learning_rate": 9.88189312109153e-06, "loss": 1.3171, "step": 3824 }, { "epoch": 0.407001489678655, "grad_norm": 12.734375, "learning_rate": 9.881857589539512e-06, "loss": 1.2851, "step": 3825 }, { "epoch": 0.4071078952968717, "grad_norm": 15.1328125, "learning_rate": 9.881822057987494e-06, "loss": 1.2557, "step": 3826 }, { "epoch": 0.4072143009150883, "grad_norm": 13.078125, "learning_rate": 9.881786526435475e-06, "loss": 0.935, "step": 3827 }, { "epoch": 0.40732070653330493, "grad_norm": 9.7734375, "learning_rate": 9.881750994883459e-06, "loss": 1.0036, "step": 3828 }, { "epoch": 0.4074271121515216, "grad_norm": 14.9375, "learning_rate": 9.881715463331439e-06, "loss": 1.1804, "step": 3829 }, { "epoch": 0.40753351776973823, "grad_norm": 14.4453125, "learning_rate": 9.88167993177942e-06, "loss": 1.3894, "step": 3830 }, { "epoch": 0.4076399233879549, "grad_norm": 21.546875, "learning_rate": 9.881644400227402e-06, "loss": 1.6095, "step": 3831 }, { "epoch": 0.4077463290061715, "grad_norm": 17.5625, "learning_rate": 9.881608868675383e-06, "loss": 1.2843, "step": 3832 }, { "epoch": 0.40785273462438815, "grad_norm": 11.9609375, "learning_rate": 9.881573337123367e-06, "loss": 1.2014, "step": 3833 }, { "epoch": 0.4079591402426048, "grad_norm": 12.28125, "learning_rate": 9.881537805571348e-06, "loss": 1.1288, "step": 3834 }, { "epoch": 0.40806554586082144, "grad_norm": 15.359375, "learning_rate": 9.88150227401933e-06, "loss": 1.2102, "step": 3835 }, { "epoch": 0.4081719514790381, "grad_norm": 12.109375, "learning_rate": 9.881466742467311e-06, "loss": 1.312, "step": 3836 }, { "epoch": 0.40827835709725474, "grad_norm": 14.109375, "learning_rate": 9.881431210915293e-06, "loss": 1.1779, "step": 3837 }, { "epoch": 0.40838476271547136, "grad_norm": 16.8125, "learning_rate": 9.881395679363276e-06, "loss": 1.3722, "step": 3838 }, { "epoch": 0.40849116833368804, "grad_norm": 17.734375, "learning_rate": 9.881360147811258e-06, "loss": 1.3422, "step": 3839 }, { "epoch": 0.40859757395190466, "grad_norm": 14.2578125, "learning_rate": 9.88132461625924e-06, "loss": 1.4244, "step": 3840 }, { "epoch": 0.4087039795701213, "grad_norm": 11.515625, "learning_rate": 9.881289084707221e-06, "loss": 1.2682, "step": 3841 }, { "epoch": 0.40881038518833795, "grad_norm": 13.09375, "learning_rate": 9.881253553155203e-06, "loss": 1.2211, "step": 3842 }, { "epoch": 0.4089167908065546, "grad_norm": 13.5546875, "learning_rate": 9.881218021603184e-06, "loss": 1.209, "step": 3843 }, { "epoch": 0.40902319642477125, "grad_norm": 16.609375, "learning_rate": 9.881182490051166e-06, "loss": 1.44, "step": 3844 }, { "epoch": 0.40912960204298787, "grad_norm": 16.53125, "learning_rate": 9.881146958499148e-06, "loss": 0.953, "step": 3845 }, { "epoch": 0.4092360076612045, "grad_norm": 10.8125, "learning_rate": 9.88111142694713e-06, "loss": 1.2014, "step": 3846 }, { "epoch": 0.40934241327942117, "grad_norm": 12.4921875, "learning_rate": 9.881075895395111e-06, "loss": 1.1056, "step": 3847 }, { "epoch": 0.4094488188976378, "grad_norm": 19.109375, "learning_rate": 9.881040363843092e-06, "loss": 1.3167, "step": 3848 }, { "epoch": 0.4095552245158544, "grad_norm": 12.3515625, "learning_rate": 9.881004832291076e-06, "loss": 1.0382, "step": 3849 }, { "epoch": 0.4096616301340711, "grad_norm": 16.0625, "learning_rate": 9.880969300739057e-06, "loss": 1.3827, "step": 3850 }, { "epoch": 0.4097680357522877, "grad_norm": 18.875, "learning_rate": 9.880933769187039e-06, "loss": 1.1658, "step": 3851 }, { "epoch": 0.4098744413705044, "grad_norm": 11.21875, "learning_rate": 9.88089823763502e-06, "loss": 1.163, "step": 3852 }, { "epoch": 0.409980846988721, "grad_norm": 13.0859375, "learning_rate": 9.880862706083002e-06, "loss": 1.1961, "step": 3853 }, { "epoch": 0.4100872526069376, "grad_norm": 14.859375, "learning_rate": 9.880827174530985e-06, "loss": 1.5366, "step": 3854 }, { "epoch": 0.4101936582251543, "grad_norm": 20.171875, "learning_rate": 9.880791642978967e-06, "loss": 1.3253, "step": 3855 }, { "epoch": 0.4103000638433709, "grad_norm": 15.28125, "learning_rate": 9.880756111426949e-06, "loss": 1.4523, "step": 3856 }, { "epoch": 0.4104064694615876, "grad_norm": 14.6015625, "learning_rate": 9.88072057987493e-06, "loss": 1.5719, "step": 3857 }, { "epoch": 0.4105128750798042, "grad_norm": 12.7109375, "learning_rate": 9.880685048322912e-06, "loss": 0.8937, "step": 3858 }, { "epoch": 0.41061928069802084, "grad_norm": 11.234375, "learning_rate": 9.880649516770894e-06, "loss": 1.1238, "step": 3859 }, { "epoch": 0.4107256863162375, "grad_norm": 14.6640625, "learning_rate": 9.880613985218875e-06, "loss": 1.4825, "step": 3860 }, { "epoch": 0.41083209193445414, "grad_norm": 12.046875, "learning_rate": 9.880578453666857e-06, "loss": 1.4084, "step": 3861 }, { "epoch": 0.41093849755267076, "grad_norm": 13.8125, "learning_rate": 9.880542922114838e-06, "loss": 1.1193, "step": 3862 }, { "epoch": 0.41104490317088743, "grad_norm": 11.5703125, "learning_rate": 9.88050739056282e-06, "loss": 0.9542, "step": 3863 }, { "epoch": 0.41115130878910405, "grad_norm": 12.125, "learning_rate": 9.880471859010803e-06, "loss": 1.2889, "step": 3864 }, { "epoch": 0.41125771440732073, "grad_norm": 13.890625, "learning_rate": 9.880436327458785e-06, "loss": 1.307, "step": 3865 }, { "epoch": 0.41136412002553735, "grad_norm": 11.9140625, "learning_rate": 9.880400795906766e-06, "loss": 1.1527, "step": 3866 }, { "epoch": 0.41147052564375397, "grad_norm": 15.078125, "learning_rate": 9.880365264354748e-06, "loss": 1.2093, "step": 3867 }, { "epoch": 0.41157693126197065, "grad_norm": 11.5703125, "learning_rate": 9.88032973280273e-06, "loss": 1.3263, "step": 3868 }, { "epoch": 0.41168333688018727, "grad_norm": 11.125, "learning_rate": 9.880294201250711e-06, "loss": 1.3075, "step": 3869 }, { "epoch": 0.41178974249840394, "grad_norm": 18.75, "learning_rate": 9.880258669698693e-06, "loss": 1.0145, "step": 3870 }, { "epoch": 0.41189614811662056, "grad_norm": 18.5625, "learning_rate": 9.880223138146674e-06, "loss": 1.1712, "step": 3871 }, { "epoch": 0.4120025537348372, "grad_norm": 13.1484375, "learning_rate": 9.880187606594656e-06, "loss": 1.0932, "step": 3872 }, { "epoch": 0.41210895935305386, "grad_norm": 13.25, "learning_rate": 9.880152075042638e-06, "loss": 1.3018, "step": 3873 }, { "epoch": 0.4122153649712705, "grad_norm": 14.1796875, "learning_rate": 9.880116543490621e-06, "loss": 1.2673, "step": 3874 }, { "epoch": 0.4123217705894871, "grad_norm": 14.9609375, "learning_rate": 9.880081011938603e-06, "loss": 1.5344, "step": 3875 }, { "epoch": 0.4124281762077038, "grad_norm": 13.9765625, "learning_rate": 9.880045480386584e-06, "loss": 1.1434, "step": 3876 }, { "epoch": 0.4125345818259204, "grad_norm": 25.046875, "learning_rate": 9.880009948834566e-06, "loss": 1.2782, "step": 3877 }, { "epoch": 0.4126409874441371, "grad_norm": 14.6328125, "learning_rate": 9.879974417282547e-06, "loss": 1.3657, "step": 3878 }, { "epoch": 0.4127473930623537, "grad_norm": 17.203125, "learning_rate": 9.87993888573053e-06, "loss": 1.346, "step": 3879 }, { "epoch": 0.4128537986805703, "grad_norm": 17.125, "learning_rate": 9.879903354178512e-06, "loss": 1.2383, "step": 3880 }, { "epoch": 0.412960204298787, "grad_norm": 11.734375, "learning_rate": 9.879867822626494e-06, "loss": 1.0539, "step": 3881 }, { "epoch": 0.4130666099170036, "grad_norm": 16.34375, "learning_rate": 9.879832291074476e-06, "loss": 1.3214, "step": 3882 }, { "epoch": 0.41317301553522023, "grad_norm": 20.0625, "learning_rate": 9.879796759522457e-06, "loss": 1.7041, "step": 3883 }, { "epoch": 0.4132794211534369, "grad_norm": 11.5546875, "learning_rate": 9.879761227970439e-06, "loss": 1.087, "step": 3884 }, { "epoch": 0.41338582677165353, "grad_norm": 16.8125, "learning_rate": 9.87972569641842e-06, "loss": 1.0638, "step": 3885 }, { "epoch": 0.4134922323898702, "grad_norm": 13.4765625, "learning_rate": 9.879690164866402e-06, "loss": 1.2898, "step": 3886 }, { "epoch": 0.4135986380080868, "grad_norm": 14.9453125, "learning_rate": 9.879654633314384e-06, "loss": 1.1093, "step": 3887 }, { "epoch": 0.41370504362630345, "grad_norm": 17.515625, "learning_rate": 9.879619101762365e-06, "loss": 1.5219, "step": 3888 }, { "epoch": 0.4138114492445201, "grad_norm": 12.5, "learning_rate": 9.879583570210347e-06, "loss": 1.0642, "step": 3889 }, { "epoch": 0.41391785486273674, "grad_norm": 11.421875, "learning_rate": 9.87954803865833e-06, "loss": 1.1774, "step": 3890 }, { "epoch": 0.4140242604809534, "grad_norm": 13.828125, "learning_rate": 9.879512507106312e-06, "loss": 1.2407, "step": 3891 }, { "epoch": 0.41413066609917004, "grad_norm": 11.6796875, "learning_rate": 9.879476975554293e-06, "loss": 1.0559, "step": 3892 }, { "epoch": 0.41423707171738666, "grad_norm": 13.5078125, "learning_rate": 9.879441444002275e-06, "loss": 1.5834, "step": 3893 }, { "epoch": 0.41434347733560334, "grad_norm": 11.703125, "learning_rate": 9.879405912450257e-06, "loss": 0.9739, "step": 3894 }, { "epoch": 0.41444988295381996, "grad_norm": 12.1796875, "learning_rate": 9.879370380898238e-06, "loss": 1.1675, "step": 3895 }, { "epoch": 0.4145562885720366, "grad_norm": 14.90625, "learning_rate": 9.87933484934622e-06, "loss": 1.5315, "step": 3896 }, { "epoch": 0.41466269419025326, "grad_norm": 17.921875, "learning_rate": 9.879299317794201e-06, "loss": 1.4354, "step": 3897 }, { "epoch": 0.4147690998084699, "grad_norm": 15.921875, "learning_rate": 9.879263786242183e-06, "loss": 1.2758, "step": 3898 }, { "epoch": 0.41487550542668655, "grad_norm": 10.5859375, "learning_rate": 9.879228254690165e-06, "loss": 1.0657, "step": 3899 }, { "epoch": 0.4149819110449032, "grad_norm": 9.8046875, "learning_rate": 9.879192723138148e-06, "loss": 1.1807, "step": 3900 }, { "epoch": 0.4150883166631198, "grad_norm": 12.2421875, "learning_rate": 9.87915719158613e-06, "loss": 1.2242, "step": 3901 }, { "epoch": 0.41519472228133647, "grad_norm": 17.203125, "learning_rate": 9.879121660034111e-06, "loss": 1.1112, "step": 3902 }, { "epoch": 0.4153011278995531, "grad_norm": 14.8828125, "learning_rate": 9.879086128482093e-06, "loss": 1.2228, "step": 3903 }, { "epoch": 0.4154075335177697, "grad_norm": 12.953125, "learning_rate": 9.879050596930074e-06, "loss": 1.3837, "step": 3904 }, { "epoch": 0.4155139391359864, "grad_norm": 14.7890625, "learning_rate": 9.879015065378058e-06, "loss": 1.2474, "step": 3905 }, { "epoch": 0.415620344754203, "grad_norm": 19.25, "learning_rate": 9.87897953382604e-06, "loss": 1.1067, "step": 3906 }, { "epoch": 0.4157267503724197, "grad_norm": 12.8125, "learning_rate": 9.87894400227402e-06, "loss": 1.4057, "step": 3907 }, { "epoch": 0.4158331559906363, "grad_norm": 14.78125, "learning_rate": 9.878908470722002e-06, "loss": 1.2018, "step": 3908 }, { "epoch": 0.4159395616088529, "grad_norm": 14.4453125, "learning_rate": 9.878872939169984e-06, "loss": 1.4288, "step": 3909 }, { "epoch": 0.4160459672270696, "grad_norm": 17.0, "learning_rate": 9.878837407617966e-06, "loss": 1.1987, "step": 3910 }, { "epoch": 0.4161523728452862, "grad_norm": 15.5703125, "learning_rate": 9.878801876065947e-06, "loss": 1.277, "step": 3911 }, { "epoch": 0.4162587784635029, "grad_norm": 20.0625, "learning_rate": 9.878766344513929e-06, "loss": 1.0749, "step": 3912 }, { "epoch": 0.4163651840817195, "grad_norm": 15.0703125, "learning_rate": 9.87873081296191e-06, "loss": 1.2294, "step": 3913 }, { "epoch": 0.41647158969993614, "grad_norm": 15.9140625, "learning_rate": 9.878695281409892e-06, "loss": 1.1561, "step": 3914 }, { "epoch": 0.4165779953181528, "grad_norm": 16.375, "learning_rate": 9.878659749857875e-06, "loss": 1.3047, "step": 3915 }, { "epoch": 0.41668440093636944, "grad_norm": 12.7421875, "learning_rate": 9.878624218305857e-06, "loss": 1.2057, "step": 3916 }, { "epoch": 0.41679080655458606, "grad_norm": 10.109375, "learning_rate": 9.878588686753839e-06, "loss": 1.4602, "step": 3917 }, { "epoch": 0.41689721217280273, "grad_norm": 18.609375, "learning_rate": 9.87855315520182e-06, "loss": 1.282, "step": 3918 }, { "epoch": 0.41700361779101935, "grad_norm": 16.59375, "learning_rate": 9.878517623649802e-06, "loss": 1.5421, "step": 3919 }, { "epoch": 0.41711002340923603, "grad_norm": 18.03125, "learning_rate": 9.878482092097783e-06, "loss": 1.4493, "step": 3920 }, { "epoch": 0.41721642902745265, "grad_norm": 13.203125, "learning_rate": 9.878446560545765e-06, "loss": 1.1429, "step": 3921 }, { "epoch": 0.41732283464566927, "grad_norm": 15.25, "learning_rate": 9.878411028993747e-06, "loss": 1.7323, "step": 3922 }, { "epoch": 0.41742924026388595, "grad_norm": 13.09375, "learning_rate": 9.878375497441728e-06, "loss": 1.1506, "step": 3923 }, { "epoch": 0.41753564588210257, "grad_norm": 13.1796875, "learning_rate": 9.87833996588971e-06, "loss": 1.1166, "step": 3924 }, { "epoch": 0.4176420515003192, "grad_norm": 14.4375, "learning_rate": 9.878304434337691e-06, "loss": 1.5214, "step": 3925 }, { "epoch": 0.41774845711853587, "grad_norm": 11.3671875, "learning_rate": 9.878268902785675e-06, "loss": 1.126, "step": 3926 }, { "epoch": 0.4178548627367525, "grad_norm": 14.4296875, "learning_rate": 9.878233371233656e-06, "loss": 1.1716, "step": 3927 }, { "epoch": 0.41796126835496916, "grad_norm": 13.03125, "learning_rate": 9.878197839681638e-06, "loss": 1.3608, "step": 3928 }, { "epoch": 0.4180676739731858, "grad_norm": 16.546875, "learning_rate": 9.87816230812962e-06, "loss": 1.3013, "step": 3929 }, { "epoch": 0.4181740795914024, "grad_norm": 18.609375, "learning_rate": 9.878126776577601e-06, "loss": 1.1742, "step": 3930 }, { "epoch": 0.4182804852096191, "grad_norm": 11.3359375, "learning_rate": 9.878091245025584e-06, "loss": 1.1109, "step": 3931 }, { "epoch": 0.4183868908278357, "grad_norm": 13.0078125, "learning_rate": 9.878055713473566e-06, "loss": 1.3894, "step": 3932 }, { "epoch": 0.4184932964460524, "grad_norm": 19.8125, "learning_rate": 9.878020181921548e-06, "loss": 1.4474, "step": 3933 }, { "epoch": 0.418599702064269, "grad_norm": 16.40625, "learning_rate": 9.87798465036953e-06, "loss": 1.2455, "step": 3934 }, { "epoch": 0.4187061076824856, "grad_norm": 14.1875, "learning_rate": 9.877949118817511e-06, "loss": 1.3577, "step": 3935 }, { "epoch": 0.4188125133007023, "grad_norm": 16.1875, "learning_rate": 9.877913587265492e-06, "loss": 1.2954, "step": 3936 }, { "epoch": 0.4189189189189189, "grad_norm": 17.296875, "learning_rate": 9.877878055713474e-06, "loss": 1.2636, "step": 3937 }, { "epoch": 0.41902532453713554, "grad_norm": 19.765625, "learning_rate": 9.877842524161456e-06, "loss": 1.4885, "step": 3938 }, { "epoch": 0.4191317301553522, "grad_norm": 15.09375, "learning_rate": 9.877806992609437e-06, "loss": 1.1325, "step": 3939 }, { "epoch": 0.41923813577356883, "grad_norm": 11.2109375, "learning_rate": 9.877771461057419e-06, "loss": 1.2336, "step": 3940 }, { "epoch": 0.4193445413917855, "grad_norm": 10.4453125, "learning_rate": 9.877735929505402e-06, "loss": 1.068, "step": 3941 }, { "epoch": 0.41945094701000213, "grad_norm": 13.9375, "learning_rate": 9.877700397953384e-06, "loss": 1.3867, "step": 3942 }, { "epoch": 0.41955735262821875, "grad_norm": 19.296875, "learning_rate": 9.877664866401365e-06, "loss": 1.5236, "step": 3943 }, { "epoch": 0.4196637582464354, "grad_norm": 11.7578125, "learning_rate": 9.877629334849347e-06, "loss": 1.2469, "step": 3944 }, { "epoch": 0.41977016386465205, "grad_norm": 15.3359375, "learning_rate": 9.877593803297329e-06, "loss": 1.1751, "step": 3945 }, { "epoch": 0.4198765694828687, "grad_norm": 11.0078125, "learning_rate": 9.877558271745312e-06, "loss": 1.1564, "step": 3946 }, { "epoch": 0.41998297510108534, "grad_norm": 13.546875, "learning_rate": 9.877522740193294e-06, "loss": 1.35, "step": 3947 }, { "epoch": 0.42008938071930196, "grad_norm": 12.0078125, "learning_rate": 9.877487208641275e-06, "loss": 0.9639, "step": 3948 }, { "epoch": 0.42019578633751864, "grad_norm": 15.65625, "learning_rate": 9.877451677089255e-06, "loss": 1.5666, "step": 3949 }, { "epoch": 0.42030219195573526, "grad_norm": 12.3671875, "learning_rate": 9.877416145537237e-06, "loss": 1.4403, "step": 3950 }, { "epoch": 0.4204085975739519, "grad_norm": 18.796875, "learning_rate": 9.87738061398522e-06, "loss": 1.0762, "step": 3951 }, { "epoch": 0.42051500319216856, "grad_norm": 14.6953125, "learning_rate": 9.877345082433202e-06, "loss": 1.169, "step": 3952 }, { "epoch": 0.4206214088103852, "grad_norm": 14.8046875, "learning_rate": 9.877309550881183e-06, "loss": 1.246, "step": 3953 }, { "epoch": 0.42072781442860185, "grad_norm": 12.5390625, "learning_rate": 9.877274019329165e-06, "loss": 1.2463, "step": 3954 }, { "epoch": 0.4208342200468185, "grad_norm": 11.734375, "learning_rate": 9.877238487777146e-06, "loss": 1.2145, "step": 3955 }, { "epoch": 0.4209406256650351, "grad_norm": 15.7421875, "learning_rate": 9.87720295622513e-06, "loss": 1.5566, "step": 3956 }, { "epoch": 0.42104703128325177, "grad_norm": 130.75, "learning_rate": 9.877167424673111e-06, "loss": 1.4093, "step": 3957 }, { "epoch": 0.4211534369014684, "grad_norm": 13.40625, "learning_rate": 9.877131893121093e-06, "loss": 1.4802, "step": 3958 }, { "epoch": 0.421259842519685, "grad_norm": 10.53125, "learning_rate": 9.877096361569075e-06, "loss": 1.073, "step": 3959 }, { "epoch": 0.4213662481379017, "grad_norm": 11.1171875, "learning_rate": 9.877060830017056e-06, "loss": 1.2698, "step": 3960 }, { "epoch": 0.4214726537561183, "grad_norm": 11.296875, "learning_rate": 9.877025298465038e-06, "loss": 1.1852, "step": 3961 }, { "epoch": 0.421579059374335, "grad_norm": 15.109375, "learning_rate": 9.87698976691302e-06, "loss": 1.2774, "step": 3962 }, { "epoch": 0.4216854649925516, "grad_norm": 11.0390625, "learning_rate": 9.876954235361001e-06, "loss": 1.1356, "step": 3963 }, { "epoch": 0.4217918706107682, "grad_norm": 11.46875, "learning_rate": 9.876918703808983e-06, "loss": 0.9611, "step": 3964 }, { "epoch": 0.4218982762289849, "grad_norm": 17.625, "learning_rate": 9.876883172256964e-06, "loss": 1.2141, "step": 3965 }, { "epoch": 0.4220046818472015, "grad_norm": 13.7109375, "learning_rate": 9.876847640704946e-06, "loss": 1.1023, "step": 3966 }, { "epoch": 0.4221110874654182, "grad_norm": 11.28125, "learning_rate": 9.876812109152929e-06, "loss": 1.0617, "step": 3967 }, { "epoch": 0.4222174930836348, "grad_norm": 11.7265625, "learning_rate": 9.87677657760091e-06, "loss": 1.1951, "step": 3968 }, { "epoch": 0.42232389870185144, "grad_norm": 15.8828125, "learning_rate": 9.876741046048892e-06, "loss": 1.3029, "step": 3969 }, { "epoch": 0.4224303043200681, "grad_norm": 11.78125, "learning_rate": 9.876705514496874e-06, "loss": 1.2742, "step": 3970 }, { "epoch": 0.42253670993828474, "grad_norm": 14.546875, "learning_rate": 9.876669982944855e-06, "loss": 1.5095, "step": 3971 }, { "epoch": 0.42264311555650136, "grad_norm": 14.9375, "learning_rate": 9.876634451392839e-06, "loss": 1.4327, "step": 3972 }, { "epoch": 0.42274952117471803, "grad_norm": 14.5234375, "learning_rate": 9.87659891984082e-06, "loss": 1.3059, "step": 3973 }, { "epoch": 0.42285592679293466, "grad_norm": 16.09375, "learning_rate": 9.876563388288802e-06, "loss": 1.1725, "step": 3974 }, { "epoch": 0.42296233241115133, "grad_norm": 11.984375, "learning_rate": 9.876527856736784e-06, "loss": 1.1621, "step": 3975 }, { "epoch": 0.42306873802936795, "grad_norm": 15.4609375, "learning_rate": 9.876492325184765e-06, "loss": 1.1311, "step": 3976 }, { "epoch": 0.4231751436475846, "grad_norm": 16.46875, "learning_rate": 9.876456793632747e-06, "loss": 1.4749, "step": 3977 }, { "epoch": 0.42328154926580125, "grad_norm": 9.0390625, "learning_rate": 9.876421262080728e-06, "loss": 0.9521, "step": 3978 }, { "epoch": 0.42338795488401787, "grad_norm": 13.21875, "learning_rate": 9.87638573052871e-06, "loss": 1.1192, "step": 3979 }, { "epoch": 0.4234943605022345, "grad_norm": 11.2890625, "learning_rate": 9.876350198976692e-06, "loss": 1.2107, "step": 3980 }, { "epoch": 0.42360076612045117, "grad_norm": 15.28125, "learning_rate": 9.876314667424673e-06, "loss": 1.4812, "step": 3981 }, { "epoch": 0.4237071717386678, "grad_norm": 10.359375, "learning_rate": 9.876279135872657e-06, "loss": 1.0814, "step": 3982 }, { "epoch": 0.42381357735688446, "grad_norm": 13.9765625, "learning_rate": 9.876243604320638e-06, "loss": 1.4877, "step": 3983 }, { "epoch": 0.4239199829751011, "grad_norm": 18.265625, "learning_rate": 9.87620807276862e-06, "loss": 1.5156, "step": 3984 }, { "epoch": 0.4240263885933177, "grad_norm": 14.4375, "learning_rate": 9.876172541216601e-06, "loss": 1.3217, "step": 3985 }, { "epoch": 0.4241327942115344, "grad_norm": 14.3203125, "learning_rate": 9.876137009664583e-06, "loss": 1.2587, "step": 3986 }, { "epoch": 0.424239199829751, "grad_norm": 10.6640625, "learning_rate": 9.876101478112565e-06, "loss": 1.1525, "step": 3987 }, { "epoch": 0.4243456054479677, "grad_norm": 12.9296875, "learning_rate": 9.876065946560546e-06, "loss": 1.3763, "step": 3988 }, { "epoch": 0.4244520110661843, "grad_norm": 14.5859375, "learning_rate": 9.876030415008528e-06, "loss": 1.2365, "step": 3989 }, { "epoch": 0.4245584166844009, "grad_norm": 13.9296875, "learning_rate": 9.87599488345651e-06, "loss": 1.2824, "step": 3990 }, { "epoch": 0.4246648223026176, "grad_norm": 17.65625, "learning_rate": 9.875959351904491e-06, "loss": 1.6004, "step": 3991 }, { "epoch": 0.4247712279208342, "grad_norm": 17.140625, "learning_rate": 9.875923820352474e-06, "loss": 1.4341, "step": 3992 }, { "epoch": 0.42487763353905084, "grad_norm": 10.7890625, "learning_rate": 9.875888288800456e-06, "loss": 1.0265, "step": 3993 }, { "epoch": 0.4249840391572675, "grad_norm": 14.1015625, "learning_rate": 9.875852757248438e-06, "loss": 1.2806, "step": 3994 }, { "epoch": 0.42509044477548413, "grad_norm": 14.6640625, "learning_rate": 9.875817225696419e-06, "loss": 1.2232, "step": 3995 }, { "epoch": 0.4251968503937008, "grad_norm": 13.703125, "learning_rate": 9.8757816941444e-06, "loss": 1.2476, "step": 3996 }, { "epoch": 0.42530325601191743, "grad_norm": 15.9296875, "learning_rate": 9.875746162592384e-06, "loss": 1.1043, "step": 3997 }, { "epoch": 0.42540966163013405, "grad_norm": 11.1015625, "learning_rate": 9.875710631040366e-06, "loss": 1.1398, "step": 3998 }, { "epoch": 0.4255160672483507, "grad_norm": 24.546875, "learning_rate": 9.875675099488347e-06, "loss": 1.2332, "step": 3999 }, { "epoch": 0.42562247286656735, "grad_norm": 11.5546875, "learning_rate": 9.875639567936329e-06, "loss": 1.2752, "step": 4000 }, { "epoch": 0.425728878484784, "grad_norm": 15.03125, "learning_rate": 9.87560403638431e-06, "loss": 1.1426, "step": 4001 }, { "epoch": 0.42583528410300064, "grad_norm": 14.421875, "learning_rate": 9.875568504832292e-06, "loss": 1.1895, "step": 4002 }, { "epoch": 0.42594168972121726, "grad_norm": 9.46875, "learning_rate": 9.875532973280274e-06, "loss": 0.8549, "step": 4003 }, { "epoch": 0.42604809533943394, "grad_norm": 15.046875, "learning_rate": 9.875497441728255e-06, "loss": 1.3479, "step": 4004 }, { "epoch": 0.42615450095765056, "grad_norm": 12.140625, "learning_rate": 9.875461910176237e-06, "loss": 1.2884, "step": 4005 }, { "epoch": 0.4262609065758672, "grad_norm": 13.2578125, "learning_rate": 9.875426378624218e-06, "loss": 1.0444, "step": 4006 }, { "epoch": 0.42636731219408386, "grad_norm": 20.140625, "learning_rate": 9.8753908470722e-06, "loss": 0.9832, "step": 4007 }, { "epoch": 0.4264737178123005, "grad_norm": 18.9375, "learning_rate": 9.875355315520183e-06, "loss": 1.4016, "step": 4008 }, { "epoch": 0.42658012343051716, "grad_norm": 13.0859375, "learning_rate": 9.875319783968165e-06, "loss": 1.1857, "step": 4009 }, { "epoch": 0.4266865290487338, "grad_norm": 18.84375, "learning_rate": 9.875284252416147e-06, "loss": 1.4296, "step": 4010 }, { "epoch": 0.4267929346669504, "grad_norm": 16.140625, "learning_rate": 9.875248720864128e-06, "loss": 1.3569, "step": 4011 }, { "epoch": 0.4268993402851671, "grad_norm": 14.8046875, "learning_rate": 9.87521318931211e-06, "loss": 1.2518, "step": 4012 }, { "epoch": 0.4270057459033837, "grad_norm": 16.78125, "learning_rate": 9.875177657760091e-06, "loss": 1.4271, "step": 4013 }, { "epoch": 0.4271121515216003, "grad_norm": 12.421875, "learning_rate": 9.875142126208073e-06, "loss": 0.9391, "step": 4014 }, { "epoch": 0.427218557139817, "grad_norm": 18.203125, "learning_rate": 9.875106594656055e-06, "loss": 1.4969, "step": 4015 }, { "epoch": 0.4273249627580336, "grad_norm": 12.6015625, "learning_rate": 9.875071063104036e-06, "loss": 1.2222, "step": 4016 }, { "epoch": 0.4274313683762503, "grad_norm": 14.21875, "learning_rate": 9.875035531552018e-06, "loss": 1.383, "step": 4017 }, { "epoch": 0.4275377739944669, "grad_norm": 14.5859375, "learning_rate": 9.875000000000001e-06, "loss": 1.5213, "step": 4018 }, { "epoch": 0.42764417961268353, "grad_norm": 28.109375, "learning_rate": 9.874964468447983e-06, "loss": 1.3344, "step": 4019 }, { "epoch": 0.4277505852309002, "grad_norm": 11.609375, "learning_rate": 9.874928936895964e-06, "loss": 1.1038, "step": 4020 }, { "epoch": 0.4278569908491168, "grad_norm": 17.390625, "learning_rate": 9.874893405343946e-06, "loss": 1.5551, "step": 4021 }, { "epoch": 0.4279633964673335, "grad_norm": 10.75, "learning_rate": 9.874857873791928e-06, "loss": 1.1008, "step": 4022 }, { "epoch": 0.4280698020855501, "grad_norm": 15.1796875, "learning_rate": 9.874822342239911e-06, "loss": 1.5016, "step": 4023 }, { "epoch": 0.42817620770376674, "grad_norm": 17.359375, "learning_rate": 9.874786810687893e-06, "loss": 1.3526, "step": 4024 }, { "epoch": 0.4282826133219834, "grad_norm": 15.0078125, "learning_rate": 9.874751279135874e-06, "loss": 1.5471, "step": 4025 }, { "epoch": 0.42838901894020004, "grad_norm": 10.8515625, "learning_rate": 9.874715747583856e-06, "loss": 1.2163, "step": 4026 }, { "epoch": 0.42849542455841666, "grad_norm": 14.15625, "learning_rate": 9.874680216031837e-06, "loss": 1.2778, "step": 4027 }, { "epoch": 0.42860183017663334, "grad_norm": 15.8828125, "learning_rate": 9.874644684479819e-06, "loss": 1.2992, "step": 4028 }, { "epoch": 0.42870823579484996, "grad_norm": 18.828125, "learning_rate": 9.8746091529278e-06, "loss": 1.2174, "step": 4029 }, { "epoch": 0.42881464141306663, "grad_norm": 15.1484375, "learning_rate": 9.874573621375782e-06, "loss": 1.2172, "step": 4030 }, { "epoch": 0.42892104703128325, "grad_norm": 11.359375, "learning_rate": 9.874538089823764e-06, "loss": 1.2314, "step": 4031 }, { "epoch": 0.4290274526494999, "grad_norm": 11.6328125, "learning_rate": 9.874502558271745e-06, "loss": 1.1872, "step": 4032 }, { "epoch": 0.42913385826771655, "grad_norm": 13.2265625, "learning_rate": 9.874467026719729e-06, "loss": 1.3266, "step": 4033 }, { "epoch": 0.42924026388593317, "grad_norm": 13.5625, "learning_rate": 9.87443149516771e-06, "loss": 1.2796, "step": 4034 }, { "epoch": 0.4293466695041498, "grad_norm": 15.7421875, "learning_rate": 9.874395963615692e-06, "loss": 0.9524, "step": 4035 }, { "epoch": 0.42945307512236647, "grad_norm": 12.3046875, "learning_rate": 9.874360432063673e-06, "loss": 1.362, "step": 4036 }, { "epoch": 0.4295594807405831, "grad_norm": 14.0, "learning_rate": 9.874324900511655e-06, "loss": 1.1791, "step": 4037 }, { "epoch": 0.42966588635879976, "grad_norm": 16.640625, "learning_rate": 9.874289368959637e-06, "loss": 1.3586, "step": 4038 }, { "epoch": 0.4297722919770164, "grad_norm": 23.46875, "learning_rate": 9.874253837407618e-06, "loss": 1.3683, "step": 4039 }, { "epoch": 0.429878697595233, "grad_norm": 10.90625, "learning_rate": 9.8742183058556e-06, "loss": 0.9671, "step": 4040 }, { "epoch": 0.4299851032134497, "grad_norm": 22.359375, "learning_rate": 9.874182774303582e-06, "loss": 1.235, "step": 4041 }, { "epoch": 0.4300915088316663, "grad_norm": 14.6796875, "learning_rate": 9.874147242751563e-06, "loss": 1.5516, "step": 4042 }, { "epoch": 0.430197914449883, "grad_norm": 16.859375, "learning_rate": 9.874111711199545e-06, "loss": 1.2061, "step": 4043 }, { "epoch": 0.4303043200680996, "grad_norm": 12.046875, "learning_rate": 9.874076179647528e-06, "loss": 1.3194, "step": 4044 }, { "epoch": 0.4304107256863162, "grad_norm": 12.9765625, "learning_rate": 9.87404064809551e-06, "loss": 1.5472, "step": 4045 }, { "epoch": 0.4305171313045329, "grad_norm": 25.46875, "learning_rate": 9.874005116543491e-06, "loss": 1.1113, "step": 4046 }, { "epoch": 0.4306235369227495, "grad_norm": 11.1796875, "learning_rate": 9.873969584991473e-06, "loss": 1.3049, "step": 4047 }, { "epoch": 0.43072994254096614, "grad_norm": 11.2265625, "learning_rate": 9.873934053439454e-06, "loss": 1.1931, "step": 4048 }, { "epoch": 0.4308363481591828, "grad_norm": 12.703125, "learning_rate": 9.873898521887438e-06, "loss": 1.1996, "step": 4049 }, { "epoch": 0.43094275377739943, "grad_norm": 13.0234375, "learning_rate": 9.87386299033542e-06, "loss": 1.263, "step": 4050 }, { "epoch": 0.4310491593956161, "grad_norm": 17.921875, "learning_rate": 9.873827458783401e-06, "loss": 1.4816, "step": 4051 }, { "epoch": 0.43115556501383273, "grad_norm": 11.765625, "learning_rate": 9.873791927231383e-06, "loss": 1.306, "step": 4052 }, { "epoch": 0.43126197063204935, "grad_norm": 16.21875, "learning_rate": 9.873756395679364e-06, "loss": 1.1731, "step": 4053 }, { "epoch": 0.43136837625026603, "grad_norm": 11.75, "learning_rate": 9.873720864127346e-06, "loss": 1.0469, "step": 4054 }, { "epoch": 0.43147478186848265, "grad_norm": 12.2265625, "learning_rate": 9.873685332575327e-06, "loss": 1.1567, "step": 4055 }, { "epoch": 0.4315811874866993, "grad_norm": 12.6875, "learning_rate": 9.873649801023309e-06, "loss": 1.0626, "step": 4056 }, { "epoch": 0.43168759310491595, "grad_norm": 13.03125, "learning_rate": 9.87361426947129e-06, "loss": 1.6166, "step": 4057 }, { "epoch": 0.43179399872313257, "grad_norm": 16.765625, "learning_rate": 9.873578737919272e-06, "loss": 1.693, "step": 4058 }, { "epoch": 0.43190040434134924, "grad_norm": 15.890625, "learning_rate": 9.873543206367256e-06, "loss": 0.9515, "step": 4059 }, { "epoch": 0.43200680995956586, "grad_norm": 12.4765625, "learning_rate": 9.873507674815237e-06, "loss": 1.1147, "step": 4060 }, { "epoch": 0.4321132155777825, "grad_norm": 15.015625, "learning_rate": 9.873472143263219e-06, "loss": 1.2647, "step": 4061 }, { "epoch": 0.43221962119599916, "grad_norm": 15.015625, "learning_rate": 9.8734366117112e-06, "loss": 1.1651, "step": 4062 }, { "epoch": 0.4323260268142158, "grad_norm": 10.9296875, "learning_rate": 9.873401080159182e-06, "loss": 1.0206, "step": 4063 }, { "epoch": 0.43243243243243246, "grad_norm": 16.3125, "learning_rate": 9.873365548607165e-06, "loss": 1.215, "step": 4064 }, { "epoch": 0.4325388380506491, "grad_norm": 13.8125, "learning_rate": 9.873330017055147e-06, "loss": 1.4549, "step": 4065 }, { "epoch": 0.4326452436688657, "grad_norm": 14.7890625, "learning_rate": 9.873294485503128e-06, "loss": 1.6483, "step": 4066 }, { "epoch": 0.4327516492870824, "grad_norm": 20.53125, "learning_rate": 9.87325895395111e-06, "loss": 1.2262, "step": 4067 }, { "epoch": 0.432858054905299, "grad_norm": 14.203125, "learning_rate": 9.87322342239909e-06, "loss": 1.1528, "step": 4068 }, { "epoch": 0.4329644605235156, "grad_norm": 13.078125, "learning_rate": 9.873187890847073e-06, "loss": 1.1416, "step": 4069 }, { "epoch": 0.4330708661417323, "grad_norm": 14.3515625, "learning_rate": 9.873152359295055e-06, "loss": 1.5864, "step": 4070 }, { "epoch": 0.4331772717599489, "grad_norm": 12.390625, "learning_rate": 9.873116827743036e-06, "loss": 1.382, "step": 4071 }, { "epoch": 0.4332836773781656, "grad_norm": 12.1796875, "learning_rate": 9.873081296191018e-06, "loss": 1.1783, "step": 4072 }, { "epoch": 0.4333900829963822, "grad_norm": 13.484375, "learning_rate": 9.873045764639e-06, "loss": 1.6626, "step": 4073 }, { "epoch": 0.43349648861459883, "grad_norm": 26.546875, "learning_rate": 9.873010233086983e-06, "loss": 1.6449, "step": 4074 }, { "epoch": 0.4336028942328155, "grad_norm": 12.0859375, "learning_rate": 9.872974701534965e-06, "loss": 1.2572, "step": 4075 }, { "epoch": 0.4337092998510321, "grad_norm": 12.84375, "learning_rate": 9.872939169982946e-06, "loss": 1.3573, "step": 4076 }, { "epoch": 0.4338157054692488, "grad_norm": 13.6953125, "learning_rate": 9.872903638430928e-06, "loss": 1.4195, "step": 4077 }, { "epoch": 0.4339221110874654, "grad_norm": 14.6171875, "learning_rate": 9.87286810687891e-06, "loss": 1.2023, "step": 4078 }, { "epoch": 0.43402851670568204, "grad_norm": 16.484375, "learning_rate": 9.872832575326891e-06, "loss": 1.2526, "step": 4079 }, { "epoch": 0.4341349223238987, "grad_norm": 13.4921875, "learning_rate": 9.872797043774873e-06, "loss": 1.4976, "step": 4080 }, { "epoch": 0.43424132794211534, "grad_norm": 12.7265625, "learning_rate": 9.872761512222854e-06, "loss": 1.0428, "step": 4081 }, { "epoch": 0.43434773356033196, "grad_norm": 11.8515625, "learning_rate": 9.872725980670836e-06, "loss": 0.8688, "step": 4082 }, { "epoch": 0.43445413917854864, "grad_norm": 16.546875, "learning_rate": 9.872690449118817e-06, "loss": 1.5471, "step": 4083 }, { "epoch": 0.43456054479676526, "grad_norm": 20.171875, "learning_rate": 9.872654917566799e-06, "loss": 1.5287, "step": 4084 }, { "epoch": 0.43466695041498193, "grad_norm": 16.84375, "learning_rate": 9.872619386014782e-06, "loss": 1.2602, "step": 4085 }, { "epoch": 0.43477335603319855, "grad_norm": 16.984375, "learning_rate": 9.872583854462764e-06, "loss": 1.112, "step": 4086 }, { "epoch": 0.4348797616514152, "grad_norm": 15.828125, "learning_rate": 9.872548322910746e-06, "loss": 1.4919, "step": 4087 }, { "epoch": 0.43498616726963185, "grad_norm": 10.9296875, "learning_rate": 9.872512791358727e-06, "loss": 0.9245, "step": 4088 }, { "epoch": 0.43509257288784847, "grad_norm": 11.5546875, "learning_rate": 9.872477259806709e-06, "loss": 1.1423, "step": 4089 }, { "epoch": 0.4351989785060651, "grad_norm": 13.90625, "learning_rate": 9.872441728254692e-06, "loss": 1.0977, "step": 4090 }, { "epoch": 0.43530538412428177, "grad_norm": 17.125, "learning_rate": 9.872406196702674e-06, "loss": 1.2254, "step": 4091 }, { "epoch": 0.4354117897424984, "grad_norm": 18.390625, "learning_rate": 9.872370665150655e-06, "loss": 1.3171, "step": 4092 }, { "epoch": 0.43551819536071507, "grad_norm": 15.5859375, "learning_rate": 9.872335133598637e-06, "loss": 1.2043, "step": 4093 }, { "epoch": 0.4356246009789317, "grad_norm": 20.09375, "learning_rate": 9.872299602046619e-06, "loss": 1.3131, "step": 4094 }, { "epoch": 0.4357310065971483, "grad_norm": 13.28125, "learning_rate": 9.8722640704946e-06, "loss": 0.9148, "step": 4095 }, { "epoch": 0.435837412215365, "grad_norm": 16.25, "learning_rate": 9.872228538942582e-06, "loss": 1.5015, "step": 4096 }, { "epoch": 0.4359438178335816, "grad_norm": 14.640625, "learning_rate": 9.872193007390563e-06, "loss": 1.4753, "step": 4097 }, { "epoch": 0.4360502234517983, "grad_norm": 15.3515625, "learning_rate": 9.872157475838545e-06, "loss": 0.9032, "step": 4098 }, { "epoch": 0.4361566290700149, "grad_norm": 17.890625, "learning_rate": 9.872121944286527e-06, "loss": 1.2383, "step": 4099 }, { "epoch": 0.4362630346882315, "grad_norm": 14.171875, "learning_rate": 9.87208641273451e-06, "loss": 1.2226, "step": 4100 }, { "epoch": 0.4363694403064482, "grad_norm": 17.78125, "learning_rate": 9.872050881182491e-06, "loss": 1.2769, "step": 4101 }, { "epoch": 0.4364758459246648, "grad_norm": 12.2421875, "learning_rate": 9.872015349630473e-06, "loss": 1.1125, "step": 4102 }, { "epoch": 0.43658225154288144, "grad_norm": 14.0078125, "learning_rate": 9.871979818078455e-06, "loss": 1.2584, "step": 4103 }, { "epoch": 0.4366886571610981, "grad_norm": 14.34375, "learning_rate": 9.871944286526436e-06, "loss": 1.3026, "step": 4104 }, { "epoch": 0.43679506277931474, "grad_norm": 17.71875, "learning_rate": 9.871908754974418e-06, "loss": 1.317, "step": 4105 }, { "epoch": 0.4369014683975314, "grad_norm": 14.9296875, "learning_rate": 9.8718732234224e-06, "loss": 0.9623, "step": 4106 }, { "epoch": 0.43700787401574803, "grad_norm": 12.265625, "learning_rate": 9.871837691870381e-06, "loss": 1.2085, "step": 4107 }, { "epoch": 0.43711427963396465, "grad_norm": 18.1875, "learning_rate": 9.871802160318363e-06, "loss": 1.1226, "step": 4108 }, { "epoch": 0.43722068525218133, "grad_norm": 12.5546875, "learning_rate": 9.871766628766344e-06, "loss": 1.2863, "step": 4109 }, { "epoch": 0.43732709087039795, "grad_norm": 15.78125, "learning_rate": 9.871731097214328e-06, "loss": 1.2943, "step": 4110 }, { "epoch": 0.4374334964886146, "grad_norm": 21.921875, "learning_rate": 9.87169556566231e-06, "loss": 1.4737, "step": 4111 }, { "epoch": 0.43753990210683125, "grad_norm": 16.421875, "learning_rate": 9.87166003411029e-06, "loss": 1.259, "step": 4112 }, { "epoch": 0.43764630772504787, "grad_norm": 14.4140625, "learning_rate": 9.871624502558272e-06, "loss": 1.1117, "step": 4113 }, { "epoch": 0.43775271334326454, "grad_norm": 14.609375, "learning_rate": 9.871588971006254e-06, "loss": 1.1698, "step": 4114 }, { "epoch": 0.43785911896148116, "grad_norm": 22.796875, "learning_rate": 9.871553439454237e-06, "loss": 1.0395, "step": 4115 }, { "epoch": 0.4379655245796978, "grad_norm": 13.9609375, "learning_rate": 9.871517907902219e-06, "loss": 1.2116, "step": 4116 }, { "epoch": 0.43807193019791446, "grad_norm": 13.296875, "learning_rate": 9.8714823763502e-06, "loss": 1.3191, "step": 4117 }, { "epoch": 0.4381783358161311, "grad_norm": 13.4765625, "learning_rate": 9.871446844798182e-06, "loss": 1.3308, "step": 4118 }, { "epoch": 0.43828474143434776, "grad_norm": 11.6796875, "learning_rate": 9.871411313246164e-06, "loss": 1.2194, "step": 4119 }, { "epoch": 0.4383911470525644, "grad_norm": 18.859375, "learning_rate": 9.871375781694145e-06, "loss": 1.2437, "step": 4120 }, { "epoch": 0.438497552670781, "grad_norm": 17.5, "learning_rate": 9.871340250142127e-06, "loss": 0.9889, "step": 4121 }, { "epoch": 0.4386039582889977, "grad_norm": 13.75, "learning_rate": 9.871304718590109e-06, "loss": 1.4056, "step": 4122 }, { "epoch": 0.4387103639072143, "grad_norm": 20.234375, "learning_rate": 9.87126918703809e-06, "loss": 1.8228, "step": 4123 }, { "epoch": 0.4388167695254309, "grad_norm": 24.5, "learning_rate": 9.871233655486072e-06, "loss": 1.6069, "step": 4124 }, { "epoch": 0.4389231751436476, "grad_norm": 11.671875, "learning_rate": 9.871198123934053e-06, "loss": 1.2383, "step": 4125 }, { "epoch": 0.4390295807618642, "grad_norm": 14.390625, "learning_rate": 9.871162592382037e-06, "loss": 1.3587, "step": 4126 }, { "epoch": 0.4391359863800809, "grad_norm": 20.984375, "learning_rate": 9.871127060830018e-06, "loss": 0.8617, "step": 4127 }, { "epoch": 0.4392423919982975, "grad_norm": 17.265625, "learning_rate": 9.871091529278e-06, "loss": 1.2844, "step": 4128 }, { "epoch": 0.43934879761651413, "grad_norm": 19.640625, "learning_rate": 9.871055997725982e-06, "loss": 1.0203, "step": 4129 }, { "epoch": 0.4394552032347308, "grad_norm": 13.3671875, "learning_rate": 9.871020466173963e-06, "loss": 1.05, "step": 4130 }, { "epoch": 0.4395616088529474, "grad_norm": 11.0234375, "learning_rate": 9.870984934621945e-06, "loss": 1.0498, "step": 4131 }, { "epoch": 0.4396680144711641, "grad_norm": 13.1015625, "learning_rate": 9.870949403069926e-06, "loss": 1.5639, "step": 4132 }, { "epoch": 0.4397744200893807, "grad_norm": 15.4140625, "learning_rate": 9.870913871517908e-06, "loss": 1.057, "step": 4133 }, { "epoch": 0.43988082570759734, "grad_norm": 18.109375, "learning_rate": 9.87087833996589e-06, "loss": 1.089, "step": 4134 }, { "epoch": 0.439987231325814, "grad_norm": 14.1953125, "learning_rate": 9.870842808413871e-06, "loss": 1.3665, "step": 4135 }, { "epoch": 0.44009363694403064, "grad_norm": 13.1796875, "learning_rate": 9.870807276861854e-06, "loss": 1.141, "step": 4136 }, { "epoch": 0.44020004256224726, "grad_norm": 16.296875, "learning_rate": 9.870771745309836e-06, "loss": 1.47, "step": 4137 }, { "epoch": 0.44030644818046394, "grad_norm": 14.171875, "learning_rate": 9.870736213757818e-06, "loss": 1.1117, "step": 4138 }, { "epoch": 0.44041285379868056, "grad_norm": 17.546875, "learning_rate": 9.8707006822058e-06, "loss": 1.1289, "step": 4139 }, { "epoch": 0.44051925941689724, "grad_norm": 13.953125, "learning_rate": 9.870665150653781e-06, "loss": 1.1302, "step": 4140 }, { "epoch": 0.44062566503511386, "grad_norm": 21.390625, "learning_rate": 9.870629619101764e-06, "loss": 1.4844, "step": 4141 }, { "epoch": 0.4407320706533305, "grad_norm": 15.6484375, "learning_rate": 9.870594087549746e-06, "loss": 1.1018, "step": 4142 }, { "epoch": 0.44083847627154715, "grad_norm": 15.6640625, "learning_rate": 9.870558555997727e-06, "loss": 1.381, "step": 4143 }, { "epoch": 0.4409448818897638, "grad_norm": 17.84375, "learning_rate": 9.870523024445709e-06, "loss": 1.3671, "step": 4144 }, { "epoch": 0.4410512875079804, "grad_norm": 10.984375, "learning_rate": 9.87048749289369e-06, "loss": 1.2436, "step": 4145 }, { "epoch": 0.44115769312619707, "grad_norm": 15.0, "learning_rate": 9.870451961341672e-06, "loss": 1.2833, "step": 4146 }, { "epoch": 0.4412640987444137, "grad_norm": 15.296875, "learning_rate": 9.870416429789654e-06, "loss": 1.6069, "step": 4147 }, { "epoch": 0.44137050436263037, "grad_norm": 17.46875, "learning_rate": 9.870380898237635e-06, "loss": 1.6822, "step": 4148 }, { "epoch": 0.441476909980847, "grad_norm": 18.609375, "learning_rate": 9.870345366685617e-06, "loss": 1.4914, "step": 4149 }, { "epoch": 0.4415833155990636, "grad_norm": 14.203125, "learning_rate": 9.870309835133599e-06, "loss": 1.0564, "step": 4150 }, { "epoch": 0.4416897212172803, "grad_norm": 14.6171875, "learning_rate": 9.870274303581582e-06, "loss": 1.1152, "step": 4151 }, { "epoch": 0.4417961268354969, "grad_norm": 19.125, "learning_rate": 9.870238772029564e-06, "loss": 1.0925, "step": 4152 }, { "epoch": 0.4419025324537136, "grad_norm": 12.1875, "learning_rate": 9.870203240477545e-06, "loss": 1.4358, "step": 4153 }, { "epoch": 0.4420089380719302, "grad_norm": 14.8515625, "learning_rate": 9.870167708925527e-06, "loss": 1.4656, "step": 4154 }, { "epoch": 0.4421153436901468, "grad_norm": 10.2578125, "learning_rate": 9.870132177373508e-06, "loss": 1.1616, "step": 4155 }, { "epoch": 0.4422217493083635, "grad_norm": 13.5859375, "learning_rate": 9.87009664582149e-06, "loss": 0.8979, "step": 4156 }, { "epoch": 0.4423281549265801, "grad_norm": 20.25, "learning_rate": 9.870061114269472e-06, "loss": 1.4906, "step": 4157 }, { "epoch": 0.44243456054479674, "grad_norm": 15.8359375, "learning_rate": 9.870025582717453e-06, "loss": 1.1243, "step": 4158 }, { "epoch": 0.4425409661630134, "grad_norm": 17.265625, "learning_rate": 9.869990051165435e-06, "loss": 1.5311, "step": 4159 }, { "epoch": 0.44264737178123004, "grad_norm": 17.671875, "learning_rate": 9.869954519613416e-06, "loss": 1.1381, "step": 4160 }, { "epoch": 0.4427537773994467, "grad_norm": 15.1015625, "learning_rate": 9.869918988061398e-06, "loss": 1.515, "step": 4161 }, { "epoch": 0.44286018301766333, "grad_norm": 14.4375, "learning_rate": 9.869883456509381e-06, "loss": 1.4059, "step": 4162 }, { "epoch": 0.44296658863587995, "grad_norm": 14.1640625, "learning_rate": 9.869847924957363e-06, "loss": 1.2931, "step": 4163 }, { "epoch": 0.44307299425409663, "grad_norm": 11.421875, "learning_rate": 9.869812393405345e-06, "loss": 1.0478, "step": 4164 }, { "epoch": 0.44317939987231325, "grad_norm": 11.8359375, "learning_rate": 9.869776861853326e-06, "loss": 1.2032, "step": 4165 }, { "epoch": 0.4432858054905299, "grad_norm": 15.921875, "learning_rate": 9.869741330301308e-06, "loss": 1.4346, "step": 4166 }, { "epoch": 0.44339221110874655, "grad_norm": 13.5703125, "learning_rate": 9.869705798749291e-06, "loss": 1.4257, "step": 4167 }, { "epoch": 0.44349861672696317, "grad_norm": 10.9375, "learning_rate": 9.869670267197273e-06, "loss": 1.0376, "step": 4168 }, { "epoch": 0.44360502234517984, "grad_norm": 12.984375, "learning_rate": 9.869634735645254e-06, "loss": 1.0255, "step": 4169 }, { "epoch": 0.44371142796339647, "grad_norm": 16.40625, "learning_rate": 9.869599204093236e-06, "loss": 1.2037, "step": 4170 }, { "epoch": 0.4438178335816131, "grad_norm": 28.8125, "learning_rate": 9.869563672541217e-06, "loss": 1.5669, "step": 4171 }, { "epoch": 0.44392423919982976, "grad_norm": 12.4609375, "learning_rate": 9.869528140989199e-06, "loss": 1.3305, "step": 4172 }, { "epoch": 0.4440306448180464, "grad_norm": 22.28125, "learning_rate": 9.86949260943718e-06, "loss": 1.0489, "step": 4173 }, { "epoch": 0.44413705043626306, "grad_norm": 12.4296875, "learning_rate": 9.869457077885162e-06, "loss": 1.2007, "step": 4174 }, { "epoch": 0.4442434560544797, "grad_norm": 12.453125, "learning_rate": 9.869421546333144e-06, "loss": 1.1313, "step": 4175 }, { "epoch": 0.4443498616726963, "grad_norm": 9.3671875, "learning_rate": 9.869386014781126e-06, "loss": 0.9886, "step": 4176 }, { "epoch": 0.444456267290913, "grad_norm": 16.109375, "learning_rate": 9.869350483229109e-06, "loss": 0.9526, "step": 4177 }, { "epoch": 0.4445626729091296, "grad_norm": 14.890625, "learning_rate": 9.86931495167709e-06, "loss": 1.0949, "step": 4178 }, { "epoch": 0.4446690785273462, "grad_norm": 18.921875, "learning_rate": 9.869279420125072e-06, "loss": 1.3102, "step": 4179 }, { "epoch": 0.4447754841455629, "grad_norm": 14.7265625, "learning_rate": 9.869243888573054e-06, "loss": 1.1401, "step": 4180 }, { "epoch": 0.4448818897637795, "grad_norm": 14.6171875, "learning_rate": 9.869208357021035e-06, "loss": 1.2249, "step": 4181 }, { "epoch": 0.4449882953819962, "grad_norm": 17.8125, "learning_rate": 9.869172825469019e-06, "loss": 1.1728, "step": 4182 }, { "epoch": 0.4450947010002128, "grad_norm": 21.015625, "learning_rate": 9.869137293917e-06, "loss": 1.1823, "step": 4183 }, { "epoch": 0.44520110661842943, "grad_norm": 17.703125, "learning_rate": 9.869101762364982e-06, "loss": 1.5399, "step": 4184 }, { "epoch": 0.4453075122366461, "grad_norm": 14.453125, "learning_rate": 9.869066230812963e-06, "loss": 1.2758, "step": 4185 }, { "epoch": 0.44541391785486273, "grad_norm": 11.484375, "learning_rate": 9.869030699260945e-06, "loss": 0.9994, "step": 4186 }, { "epoch": 0.4455203234730794, "grad_norm": 14.125, "learning_rate": 9.868995167708927e-06, "loss": 1.3485, "step": 4187 }, { "epoch": 0.445626729091296, "grad_norm": 16.46875, "learning_rate": 9.868959636156908e-06, "loss": 1.2885, "step": 4188 }, { "epoch": 0.44573313470951265, "grad_norm": 13.8046875, "learning_rate": 9.86892410460489e-06, "loss": 1.3252, "step": 4189 }, { "epoch": 0.4458395403277293, "grad_norm": 9.734375, "learning_rate": 9.868888573052871e-06, "loss": 0.9319, "step": 4190 }, { "epoch": 0.44594594594594594, "grad_norm": 14.5859375, "learning_rate": 9.868853041500853e-06, "loss": 1.2312, "step": 4191 }, { "epoch": 0.44605235156416256, "grad_norm": 13.4765625, "learning_rate": 9.868817509948836e-06, "loss": 1.1348, "step": 4192 }, { "epoch": 0.44615875718237924, "grad_norm": 10.1484375, "learning_rate": 9.868781978396818e-06, "loss": 1.1563, "step": 4193 }, { "epoch": 0.44626516280059586, "grad_norm": 11.9921875, "learning_rate": 9.8687464468448e-06, "loss": 1.1993, "step": 4194 }, { "epoch": 0.44637156841881254, "grad_norm": 12.1484375, "learning_rate": 9.868710915292781e-06, "loss": 1.0849, "step": 4195 }, { "epoch": 0.44647797403702916, "grad_norm": 13.75, "learning_rate": 9.868675383740763e-06, "loss": 1.2987, "step": 4196 }, { "epoch": 0.4465843796552458, "grad_norm": 14.4375, "learning_rate": 9.868639852188744e-06, "loss": 1.3447, "step": 4197 }, { "epoch": 0.44669078527346245, "grad_norm": 11.21875, "learning_rate": 9.868604320636726e-06, "loss": 1.3319, "step": 4198 }, { "epoch": 0.4467971908916791, "grad_norm": 22.359375, "learning_rate": 9.868568789084708e-06, "loss": 1.3051, "step": 4199 }, { "epoch": 0.4469035965098957, "grad_norm": 19.21875, "learning_rate": 9.86853325753269e-06, "loss": 1.2154, "step": 4200 }, { "epoch": 0.44701000212811237, "grad_norm": 15.7265625, "learning_rate": 9.86849772598067e-06, "loss": 1.162, "step": 4201 }, { "epoch": 0.447116407746329, "grad_norm": 14.1640625, "learning_rate": 9.868462194428652e-06, "loss": 0.8929, "step": 4202 }, { "epoch": 0.44722281336454567, "grad_norm": 11.40625, "learning_rate": 9.868426662876636e-06, "loss": 0.9999, "step": 4203 }, { "epoch": 0.4473292189827623, "grad_norm": 11.4453125, "learning_rate": 9.868391131324617e-06, "loss": 1.1668, "step": 4204 }, { "epoch": 0.4474356246009789, "grad_norm": 11.53125, "learning_rate": 9.868355599772599e-06, "loss": 0.9555, "step": 4205 }, { "epoch": 0.4475420302191956, "grad_norm": 17.0625, "learning_rate": 9.86832006822058e-06, "loss": 1.0553, "step": 4206 }, { "epoch": 0.4476484358374122, "grad_norm": 13.9296875, "learning_rate": 9.868284536668562e-06, "loss": 1.3725, "step": 4207 }, { "epoch": 0.4477548414556289, "grad_norm": 14.6328125, "learning_rate": 9.868249005116545e-06, "loss": 1.0323, "step": 4208 }, { "epoch": 0.4478612470738455, "grad_norm": 14.234375, "learning_rate": 9.868213473564527e-06, "loss": 1.2608, "step": 4209 }, { "epoch": 0.4479676526920621, "grad_norm": 11.8828125, "learning_rate": 9.868177942012509e-06, "loss": 1.1587, "step": 4210 }, { "epoch": 0.4480740583102788, "grad_norm": 17.53125, "learning_rate": 9.86814241046049e-06, "loss": 1.5975, "step": 4211 }, { "epoch": 0.4481804639284954, "grad_norm": 21.828125, "learning_rate": 9.868106878908472e-06, "loss": 1.2138, "step": 4212 }, { "epoch": 0.44828686954671204, "grad_norm": 17.53125, "learning_rate": 9.868071347356453e-06, "loss": 1.1561, "step": 4213 }, { "epoch": 0.4483932751649287, "grad_norm": 11.7578125, "learning_rate": 9.868035815804435e-06, "loss": 1.4863, "step": 4214 }, { "epoch": 0.44849968078314534, "grad_norm": 18.609375, "learning_rate": 9.868000284252417e-06, "loss": 0.9063, "step": 4215 }, { "epoch": 0.448606086401362, "grad_norm": 14.5703125, "learning_rate": 9.867964752700398e-06, "loss": 1.1821, "step": 4216 }, { "epoch": 0.44871249201957863, "grad_norm": 15.7421875, "learning_rate": 9.86792922114838e-06, "loss": 1.434, "step": 4217 }, { "epoch": 0.44881889763779526, "grad_norm": 22.40625, "learning_rate": 9.867893689596363e-06, "loss": 1.3956, "step": 4218 }, { "epoch": 0.44892530325601193, "grad_norm": 12.171875, "learning_rate": 9.867858158044345e-06, "loss": 1.1406, "step": 4219 }, { "epoch": 0.44903170887422855, "grad_norm": 12.390625, "learning_rate": 9.867822626492326e-06, "loss": 1.0583, "step": 4220 }, { "epoch": 0.44913811449244523, "grad_norm": 16.296875, "learning_rate": 9.867787094940308e-06, "loss": 1.2899, "step": 4221 }, { "epoch": 0.44924452011066185, "grad_norm": 11.8828125, "learning_rate": 9.86775156338829e-06, "loss": 1.2737, "step": 4222 }, { "epoch": 0.44935092572887847, "grad_norm": 10.8203125, "learning_rate": 9.867716031836271e-06, "loss": 1.2016, "step": 4223 }, { "epoch": 0.44945733134709515, "grad_norm": 13.1171875, "learning_rate": 9.867680500284253e-06, "loss": 1.0845, "step": 4224 }, { "epoch": 0.44956373696531177, "grad_norm": 12.59375, "learning_rate": 9.867644968732234e-06, "loss": 1.1931, "step": 4225 }, { "epoch": 0.4496701425835284, "grad_norm": 14.5234375, "learning_rate": 9.867609437180216e-06, "loss": 1.4675, "step": 4226 }, { "epoch": 0.44977654820174506, "grad_norm": 14.0078125, "learning_rate": 9.867573905628198e-06, "loss": 1.2403, "step": 4227 }, { "epoch": 0.4498829538199617, "grad_norm": 13.3671875, "learning_rate": 9.867538374076181e-06, "loss": 1.4284, "step": 4228 }, { "epoch": 0.44998935943817836, "grad_norm": 15.125, "learning_rate": 9.867502842524163e-06, "loss": 1.3704, "step": 4229 }, { "epoch": 0.450095765056395, "grad_norm": 14.296875, "learning_rate": 9.867467310972144e-06, "loss": 1.1893, "step": 4230 }, { "epoch": 0.4502021706746116, "grad_norm": 13.4453125, "learning_rate": 9.867431779420126e-06, "loss": 1.2389, "step": 4231 }, { "epoch": 0.4503085762928283, "grad_norm": 10.9453125, "learning_rate": 9.867396247868107e-06, "loss": 1.1293, "step": 4232 }, { "epoch": 0.4504149819110449, "grad_norm": 12.4609375, "learning_rate": 9.867360716316089e-06, "loss": 1.1168, "step": 4233 }, { "epoch": 0.4505213875292615, "grad_norm": 11.875, "learning_rate": 9.867325184764072e-06, "loss": 1.2462, "step": 4234 }, { "epoch": 0.4506277931474782, "grad_norm": 16.828125, "learning_rate": 9.867289653212054e-06, "loss": 1.6084, "step": 4235 }, { "epoch": 0.4507341987656948, "grad_norm": 10.6875, "learning_rate": 9.867254121660035e-06, "loss": 0.9409, "step": 4236 }, { "epoch": 0.4508406043839115, "grad_norm": 20.28125, "learning_rate": 9.867218590108017e-06, "loss": 1.2947, "step": 4237 }, { "epoch": 0.4509470100021281, "grad_norm": 12.796875, "learning_rate": 9.867183058555999e-06, "loss": 1.4072, "step": 4238 }, { "epoch": 0.45105341562034473, "grad_norm": 13.9765625, "learning_rate": 9.86714752700398e-06, "loss": 1.1843, "step": 4239 }, { "epoch": 0.4511598212385614, "grad_norm": 14.546875, "learning_rate": 9.867111995451962e-06, "loss": 1.2467, "step": 4240 }, { "epoch": 0.45126622685677803, "grad_norm": 11.765625, "learning_rate": 9.867076463899943e-06, "loss": 1.1611, "step": 4241 }, { "epoch": 0.4513726324749947, "grad_norm": 11.125, "learning_rate": 9.867040932347925e-06, "loss": 1.0213, "step": 4242 }, { "epoch": 0.4514790380932113, "grad_norm": 11.2734375, "learning_rate": 9.867005400795907e-06, "loss": 1.1582, "step": 4243 }, { "epoch": 0.45158544371142795, "grad_norm": 16.65625, "learning_rate": 9.86696986924389e-06, "loss": 1.174, "step": 4244 }, { "epoch": 0.4516918493296446, "grad_norm": 23.296875, "learning_rate": 9.866934337691872e-06, "loss": 1.3074, "step": 4245 }, { "epoch": 0.45179825494786124, "grad_norm": 12.5625, "learning_rate": 9.866898806139853e-06, "loss": 1.2465, "step": 4246 }, { "epoch": 0.45190466056607786, "grad_norm": 17.3125, "learning_rate": 9.866863274587835e-06, "loss": 1.429, "step": 4247 }, { "epoch": 0.45201106618429454, "grad_norm": 18.859375, "learning_rate": 9.866827743035816e-06, "loss": 1.3382, "step": 4248 }, { "epoch": 0.45211747180251116, "grad_norm": 18.4375, "learning_rate": 9.866792211483798e-06, "loss": 1.1229, "step": 4249 }, { "epoch": 0.45222387742072784, "grad_norm": 15.46875, "learning_rate": 9.86675667993178e-06, "loss": 1.1222, "step": 4250 }, { "epoch": 0.45233028303894446, "grad_norm": 11.359375, "learning_rate": 9.866721148379761e-06, "loss": 1.305, "step": 4251 }, { "epoch": 0.4524366886571611, "grad_norm": 12.6015625, "learning_rate": 9.866685616827743e-06, "loss": 1.2026, "step": 4252 }, { "epoch": 0.45254309427537776, "grad_norm": 16.171875, "learning_rate": 9.866650085275724e-06, "loss": 1.2206, "step": 4253 }, { "epoch": 0.4526494998935944, "grad_norm": 13.6640625, "learning_rate": 9.866614553723708e-06, "loss": 1.1354, "step": 4254 }, { "epoch": 0.452755905511811, "grad_norm": 15.5078125, "learning_rate": 9.86657902217169e-06, "loss": 1.481, "step": 4255 }, { "epoch": 0.4528623111300277, "grad_norm": 12.390625, "learning_rate": 9.866543490619671e-06, "loss": 1.0456, "step": 4256 }, { "epoch": 0.4529687167482443, "grad_norm": 14.7890625, "learning_rate": 9.866507959067653e-06, "loss": 1.3601, "step": 4257 }, { "epoch": 0.45307512236646097, "grad_norm": 13.5390625, "learning_rate": 9.866472427515634e-06, "loss": 1.5912, "step": 4258 }, { "epoch": 0.4531815279846776, "grad_norm": 13.125, "learning_rate": 9.866436895963618e-06, "loss": 1.08, "step": 4259 }, { "epoch": 0.4532879336028942, "grad_norm": 17.875, "learning_rate": 9.866401364411599e-06, "loss": 1.4469, "step": 4260 }, { "epoch": 0.4533943392211109, "grad_norm": 16.5, "learning_rate": 9.86636583285958e-06, "loss": 1.5717, "step": 4261 }, { "epoch": 0.4535007448393275, "grad_norm": 14.84375, "learning_rate": 9.866330301307562e-06, "loss": 1.5695, "step": 4262 }, { "epoch": 0.4536071504575442, "grad_norm": 17.75, "learning_rate": 9.866294769755544e-06, "loss": 1.1982, "step": 4263 }, { "epoch": 0.4537135560757608, "grad_norm": 13.3046875, "learning_rate": 9.866259238203526e-06, "loss": 1.4359, "step": 4264 }, { "epoch": 0.4538199616939774, "grad_norm": 10.671875, "learning_rate": 9.866223706651507e-06, "loss": 0.923, "step": 4265 }, { "epoch": 0.4539263673121941, "grad_norm": 20.640625, "learning_rate": 9.866188175099489e-06, "loss": 1.2924, "step": 4266 }, { "epoch": 0.4540327729304107, "grad_norm": 11.2734375, "learning_rate": 9.86615264354747e-06, "loss": 1.2934, "step": 4267 }, { "epoch": 0.45413917854862734, "grad_norm": 12.125, "learning_rate": 9.866117111995452e-06, "loss": 1.1991, "step": 4268 }, { "epoch": 0.454245584166844, "grad_norm": 11.8515625, "learning_rate": 9.866081580443435e-06, "loss": 1.0459, "step": 4269 }, { "epoch": 0.45435198978506064, "grad_norm": 13.4765625, "learning_rate": 9.866046048891417e-06, "loss": 1.4468, "step": 4270 }, { "epoch": 0.4544583954032773, "grad_norm": 16.0, "learning_rate": 9.866010517339398e-06, "loss": 1.5071, "step": 4271 }, { "epoch": 0.45456480102149394, "grad_norm": 10.7265625, "learning_rate": 9.86597498578738e-06, "loss": 1.0147, "step": 4272 }, { "epoch": 0.45467120663971056, "grad_norm": 15.8515625, "learning_rate": 9.865939454235362e-06, "loss": 1.7299, "step": 4273 }, { "epoch": 0.45477761225792723, "grad_norm": 11.859375, "learning_rate": 9.865903922683343e-06, "loss": 1.1831, "step": 4274 }, { "epoch": 0.45488401787614385, "grad_norm": 13.1875, "learning_rate": 9.865868391131327e-06, "loss": 1.0831, "step": 4275 }, { "epoch": 0.45499042349436053, "grad_norm": 14.1171875, "learning_rate": 9.865832859579308e-06, "loss": 1.3748, "step": 4276 }, { "epoch": 0.45509682911257715, "grad_norm": 11.953125, "learning_rate": 9.865797328027288e-06, "loss": 1.2995, "step": 4277 }, { "epoch": 0.45520323473079377, "grad_norm": 15.171875, "learning_rate": 9.86576179647527e-06, "loss": 1.2463, "step": 4278 }, { "epoch": 0.45530964034901045, "grad_norm": 17.9375, "learning_rate": 9.865726264923251e-06, "loss": 1.3307, "step": 4279 }, { "epoch": 0.45541604596722707, "grad_norm": 9.734375, "learning_rate": 9.865690733371235e-06, "loss": 1.1511, "step": 4280 }, { "epoch": 0.4555224515854437, "grad_norm": 15.578125, "learning_rate": 9.865655201819216e-06, "loss": 1.277, "step": 4281 }, { "epoch": 0.45562885720366036, "grad_norm": 14.3828125, "learning_rate": 9.865619670267198e-06, "loss": 1.1756, "step": 4282 }, { "epoch": 0.455735262821877, "grad_norm": 12.390625, "learning_rate": 9.86558413871518e-06, "loss": 1.0438, "step": 4283 }, { "epoch": 0.45584166844009366, "grad_norm": 12.0, "learning_rate": 9.865548607163161e-06, "loss": 0.9788, "step": 4284 }, { "epoch": 0.4559480740583103, "grad_norm": 16.765625, "learning_rate": 9.865513075611144e-06, "loss": 1.267, "step": 4285 }, { "epoch": 0.4560544796765269, "grad_norm": 18.984375, "learning_rate": 9.865477544059126e-06, "loss": 1.3251, "step": 4286 }, { "epoch": 0.4561608852947436, "grad_norm": 14.515625, "learning_rate": 9.865442012507108e-06, "loss": 1.2454, "step": 4287 }, { "epoch": 0.4562672909129602, "grad_norm": 14.0625, "learning_rate": 9.86540648095509e-06, "loss": 1.4481, "step": 4288 }, { "epoch": 0.4563736965311768, "grad_norm": 17.734375, "learning_rate": 9.86537094940307e-06, "loss": 1.6283, "step": 4289 }, { "epoch": 0.4564801021493935, "grad_norm": 14.65625, "learning_rate": 9.865335417851052e-06, "loss": 1.3078, "step": 4290 }, { "epoch": 0.4565865077676101, "grad_norm": 11.203125, "learning_rate": 9.865299886299034e-06, "loss": 1.0953, "step": 4291 }, { "epoch": 0.4566929133858268, "grad_norm": 13.640625, "learning_rate": 9.865264354747016e-06, "loss": 1.0154, "step": 4292 }, { "epoch": 0.4567993190040434, "grad_norm": 10.7890625, "learning_rate": 9.865228823194997e-06, "loss": 1.1901, "step": 4293 }, { "epoch": 0.45690572462226003, "grad_norm": 15.1171875, "learning_rate": 9.865193291642979e-06, "loss": 1.7287, "step": 4294 }, { "epoch": 0.4570121302404767, "grad_norm": 13.09375, "learning_rate": 9.865157760090962e-06, "loss": 1.1578, "step": 4295 }, { "epoch": 0.45711853585869333, "grad_norm": 14.90625, "learning_rate": 9.865122228538944e-06, "loss": 1.1011, "step": 4296 }, { "epoch": 0.45722494147691, "grad_norm": 14.5859375, "learning_rate": 9.865086696986925e-06, "loss": 1.0561, "step": 4297 }, { "epoch": 0.45733134709512663, "grad_norm": 15.96875, "learning_rate": 9.865051165434907e-06, "loss": 1.5549, "step": 4298 }, { "epoch": 0.45743775271334325, "grad_norm": 16.140625, "learning_rate": 9.865015633882889e-06, "loss": 1.0015, "step": 4299 }, { "epoch": 0.4575441583315599, "grad_norm": 15.2265625, "learning_rate": 9.864980102330872e-06, "loss": 1.5488, "step": 4300 }, { "epoch": 0.45765056394977655, "grad_norm": 12.9296875, "learning_rate": 9.864944570778853e-06, "loss": 1.3187, "step": 4301 }, { "epoch": 0.45775696956799317, "grad_norm": 11.4453125, "learning_rate": 9.864909039226835e-06, "loss": 1.0444, "step": 4302 }, { "epoch": 0.45786337518620984, "grad_norm": 13.78125, "learning_rate": 9.864873507674817e-06, "loss": 1.2488, "step": 4303 }, { "epoch": 0.45796978080442646, "grad_norm": 10.7265625, "learning_rate": 9.864837976122798e-06, "loss": 1.4746, "step": 4304 }, { "epoch": 0.45807618642264314, "grad_norm": 14.71875, "learning_rate": 9.86480244457078e-06, "loss": 1.2457, "step": 4305 }, { "epoch": 0.45818259204085976, "grad_norm": 15.375, "learning_rate": 9.864766913018761e-06, "loss": 1.4173, "step": 4306 }, { "epoch": 0.4582889976590764, "grad_norm": 17.65625, "learning_rate": 9.864731381466743e-06, "loss": 1.2475, "step": 4307 }, { "epoch": 0.45839540327729306, "grad_norm": 14.703125, "learning_rate": 9.864695849914725e-06, "loss": 1.1476, "step": 4308 }, { "epoch": 0.4585018088955097, "grad_norm": 11.5, "learning_rate": 9.864660318362706e-06, "loss": 1.418, "step": 4309 }, { "epoch": 0.4586082145137263, "grad_norm": 8.6640625, "learning_rate": 9.86462478681069e-06, "loss": 0.7416, "step": 4310 }, { "epoch": 0.458714620131943, "grad_norm": 15.21875, "learning_rate": 9.864589255258671e-06, "loss": 1.2968, "step": 4311 }, { "epoch": 0.4588210257501596, "grad_norm": 14.3046875, "learning_rate": 9.864553723706653e-06, "loss": 1.27, "step": 4312 }, { "epoch": 0.45892743136837627, "grad_norm": 14.6015625, "learning_rate": 9.864518192154634e-06, "loss": 1.1309, "step": 4313 }, { "epoch": 0.4590338369865929, "grad_norm": 12.515625, "learning_rate": 9.864482660602616e-06, "loss": 1.261, "step": 4314 }, { "epoch": 0.4591402426048095, "grad_norm": 15.3671875, "learning_rate": 9.864447129050598e-06, "loss": 1.2304, "step": 4315 }, { "epoch": 0.4592466482230262, "grad_norm": 19.78125, "learning_rate": 9.86441159749858e-06, "loss": 1.3736, "step": 4316 }, { "epoch": 0.4593530538412428, "grad_norm": 12.4921875, "learning_rate": 9.864376065946561e-06, "loss": 1.3768, "step": 4317 }, { "epoch": 0.4594594594594595, "grad_norm": 11.9609375, "learning_rate": 9.864340534394542e-06, "loss": 1.293, "step": 4318 }, { "epoch": 0.4595658650776761, "grad_norm": 15.921875, "learning_rate": 9.864305002842524e-06, "loss": 1.2146, "step": 4319 }, { "epoch": 0.4596722706958927, "grad_norm": 22.015625, "learning_rate": 9.864269471290506e-06, "loss": 1.7453, "step": 4320 }, { "epoch": 0.4597786763141094, "grad_norm": 21.71875, "learning_rate": 9.864233939738489e-06, "loss": 1.7581, "step": 4321 }, { "epoch": 0.459885081932326, "grad_norm": 21.234375, "learning_rate": 9.86419840818647e-06, "loss": 1.3824, "step": 4322 }, { "epoch": 0.45999148755054264, "grad_norm": 13.6875, "learning_rate": 9.864162876634452e-06, "loss": 0.9511, "step": 4323 }, { "epoch": 0.4600978931687593, "grad_norm": 13.2890625, "learning_rate": 9.864127345082434e-06, "loss": 1.5859, "step": 4324 }, { "epoch": 0.46020429878697594, "grad_norm": 12.53125, "learning_rate": 9.864091813530415e-06, "loss": 1.3987, "step": 4325 }, { "epoch": 0.4603107044051926, "grad_norm": 16.9375, "learning_rate": 9.864056281978399e-06, "loss": 1.6518, "step": 4326 }, { "epoch": 0.46041711002340924, "grad_norm": 13.6328125, "learning_rate": 9.86402075042638e-06, "loss": 1.2053, "step": 4327 }, { "epoch": 0.46052351564162586, "grad_norm": 12.3515625, "learning_rate": 9.863985218874362e-06, "loss": 1.0896, "step": 4328 }, { "epoch": 0.46062992125984253, "grad_norm": 13.46875, "learning_rate": 9.863949687322344e-06, "loss": 1.2302, "step": 4329 }, { "epoch": 0.46073632687805915, "grad_norm": 18.609375, "learning_rate": 9.863914155770325e-06, "loss": 1.7154, "step": 4330 }, { "epoch": 0.46084273249627583, "grad_norm": 10.359375, "learning_rate": 9.863878624218307e-06, "loss": 1.1993, "step": 4331 }, { "epoch": 0.46094913811449245, "grad_norm": 11.2421875, "learning_rate": 9.863843092666288e-06, "loss": 1.0692, "step": 4332 }, { "epoch": 0.46105554373270907, "grad_norm": 14.359375, "learning_rate": 9.86380756111427e-06, "loss": 1.2538, "step": 4333 }, { "epoch": 0.46116194935092575, "grad_norm": 11.6484375, "learning_rate": 9.863772029562252e-06, "loss": 1.2294, "step": 4334 }, { "epoch": 0.46126835496914237, "grad_norm": 11.3671875, "learning_rate": 9.863736498010233e-06, "loss": 1.4529, "step": 4335 }, { "epoch": 0.461374760587359, "grad_norm": 16.5, "learning_rate": 9.863700966458216e-06, "loss": 1.4568, "step": 4336 }, { "epoch": 0.46148116620557567, "grad_norm": 12.5703125, "learning_rate": 9.863665434906198e-06, "loss": 1.3726, "step": 4337 }, { "epoch": 0.4615875718237923, "grad_norm": 14.609375, "learning_rate": 9.86362990335418e-06, "loss": 1.4979, "step": 4338 }, { "epoch": 0.46169397744200896, "grad_norm": 13.0546875, "learning_rate": 9.863594371802161e-06, "loss": 1.1969, "step": 4339 }, { "epoch": 0.4618003830602256, "grad_norm": 15.75, "learning_rate": 9.863558840250143e-06, "loss": 1.5337, "step": 4340 }, { "epoch": 0.4619067886784422, "grad_norm": 14.7109375, "learning_rate": 9.863523308698124e-06, "loss": 1.5556, "step": 4341 }, { "epoch": 0.4620131942966589, "grad_norm": 13.7265625, "learning_rate": 9.863487777146106e-06, "loss": 1.3887, "step": 4342 }, { "epoch": 0.4621195999148755, "grad_norm": 10.953125, "learning_rate": 9.863452245594088e-06, "loss": 1.4611, "step": 4343 }, { "epoch": 0.4622260055330921, "grad_norm": 15.984375, "learning_rate": 9.86341671404207e-06, "loss": 1.0946, "step": 4344 }, { "epoch": 0.4623324111513088, "grad_norm": 12.1875, "learning_rate": 9.863381182490051e-06, "loss": 1.2318, "step": 4345 }, { "epoch": 0.4624388167695254, "grad_norm": 11.578125, "learning_rate": 9.863345650938034e-06, "loss": 1.0426, "step": 4346 }, { "epoch": 0.4625452223877421, "grad_norm": 15.109375, "learning_rate": 9.863310119386016e-06, "loss": 1.2394, "step": 4347 }, { "epoch": 0.4626516280059587, "grad_norm": 8.8671875, "learning_rate": 9.863274587833997e-06, "loss": 0.9485, "step": 4348 }, { "epoch": 0.46275803362417534, "grad_norm": 11.7109375, "learning_rate": 9.863239056281979e-06, "loss": 1.0096, "step": 4349 }, { "epoch": 0.462864439242392, "grad_norm": 11.7265625, "learning_rate": 9.86320352472996e-06, "loss": 1.1421, "step": 4350 }, { "epoch": 0.46297084486060863, "grad_norm": 21.015625, "learning_rate": 9.863167993177942e-06, "loss": 1.9128, "step": 4351 }, { "epoch": 0.4630772504788253, "grad_norm": 11.375, "learning_rate": 9.863132461625926e-06, "loss": 1.204, "step": 4352 }, { "epoch": 0.46318365609704193, "grad_norm": 18.078125, "learning_rate": 9.863096930073907e-06, "loss": 1.1068, "step": 4353 }, { "epoch": 0.46329006171525855, "grad_norm": 12.734375, "learning_rate": 9.863061398521889e-06, "loss": 0.9315, "step": 4354 }, { "epoch": 0.4633964673334752, "grad_norm": 13.421875, "learning_rate": 9.86302586696987e-06, "loss": 1.1946, "step": 4355 }, { "epoch": 0.46350287295169185, "grad_norm": 12.984375, "learning_rate": 9.862990335417852e-06, "loss": 1.0459, "step": 4356 }, { "epoch": 0.46360927856990847, "grad_norm": 12.0546875, "learning_rate": 9.862954803865834e-06, "loss": 1.1547, "step": 4357 }, { "epoch": 0.46371568418812514, "grad_norm": 12.765625, "learning_rate": 9.862919272313815e-06, "loss": 1.4915, "step": 4358 }, { "epoch": 0.46382208980634176, "grad_norm": 14.28125, "learning_rate": 9.862883740761797e-06, "loss": 1.1538, "step": 4359 }, { "epoch": 0.46392849542455844, "grad_norm": 13.1171875, "learning_rate": 9.862848209209778e-06, "loss": 1.506, "step": 4360 }, { "epoch": 0.46403490104277506, "grad_norm": 31.4375, "learning_rate": 9.86281267765776e-06, "loss": 1.3602, "step": 4361 }, { "epoch": 0.4641413066609917, "grad_norm": 17.890625, "learning_rate": 9.862777146105743e-06, "loss": 1.1309, "step": 4362 }, { "epoch": 0.46424771227920836, "grad_norm": 15.359375, "learning_rate": 9.862741614553725e-06, "loss": 1.1745, "step": 4363 }, { "epoch": 0.464354117897425, "grad_norm": 16.453125, "learning_rate": 9.862706083001707e-06, "loss": 1.1736, "step": 4364 }, { "epoch": 0.4644605235156416, "grad_norm": 15.1015625, "learning_rate": 9.862670551449688e-06, "loss": 1.0999, "step": 4365 }, { "epoch": 0.4645669291338583, "grad_norm": 14.765625, "learning_rate": 9.86263501989767e-06, "loss": 1.0806, "step": 4366 }, { "epoch": 0.4646733347520749, "grad_norm": 13.1640625, "learning_rate": 9.862599488345651e-06, "loss": 1.1476, "step": 4367 }, { "epoch": 0.46477974037029157, "grad_norm": 13.703125, "learning_rate": 9.862563956793633e-06, "loss": 1.228, "step": 4368 }, { "epoch": 0.4648861459885082, "grad_norm": 15.4609375, "learning_rate": 9.862528425241615e-06, "loss": 1.0949, "step": 4369 }, { "epoch": 0.4649925516067248, "grad_norm": 16.859375, "learning_rate": 9.862492893689596e-06, "loss": 1.0222, "step": 4370 }, { "epoch": 0.4650989572249415, "grad_norm": 14.375, "learning_rate": 9.862457362137578e-06, "loss": 1.3181, "step": 4371 }, { "epoch": 0.4652053628431581, "grad_norm": 10.34375, "learning_rate": 9.862421830585561e-06, "loss": 1.1401, "step": 4372 }, { "epoch": 0.4653117684613748, "grad_norm": 19.125, "learning_rate": 9.862386299033543e-06, "loss": 1.2209, "step": 4373 }, { "epoch": 0.4654181740795914, "grad_norm": 14.390625, "learning_rate": 9.862350767481524e-06, "loss": 1.1538, "step": 4374 }, { "epoch": 0.465524579697808, "grad_norm": 19.203125, "learning_rate": 9.862315235929506e-06, "loss": 1.4621, "step": 4375 }, { "epoch": 0.4656309853160247, "grad_norm": 15.8984375, "learning_rate": 9.862279704377488e-06, "loss": 1.398, "step": 4376 }, { "epoch": 0.4657373909342413, "grad_norm": 14.15625, "learning_rate": 9.86224417282547e-06, "loss": 1.2066, "step": 4377 }, { "epoch": 0.46584379655245794, "grad_norm": 14.28125, "learning_rate": 9.862208641273452e-06, "loss": 1.5688, "step": 4378 }, { "epoch": 0.4659502021706746, "grad_norm": 11.328125, "learning_rate": 9.862173109721434e-06, "loss": 1.0268, "step": 4379 }, { "epoch": 0.46605660778889124, "grad_norm": 14.8671875, "learning_rate": 9.862137578169416e-06, "loss": 0.9153, "step": 4380 }, { "epoch": 0.4661630134071079, "grad_norm": 12.078125, "learning_rate": 9.862102046617397e-06, "loss": 1.3979, "step": 4381 }, { "epoch": 0.46626941902532454, "grad_norm": 17.546875, "learning_rate": 9.862066515065379e-06, "loss": 1.3957, "step": 4382 }, { "epoch": 0.46637582464354116, "grad_norm": 15.640625, "learning_rate": 9.86203098351336e-06, "loss": 1.6944, "step": 4383 }, { "epoch": 0.46648223026175784, "grad_norm": 14.2890625, "learning_rate": 9.861995451961342e-06, "loss": 1.2747, "step": 4384 }, { "epoch": 0.46658863587997446, "grad_norm": 12.8671875, "learning_rate": 9.861959920409324e-06, "loss": 1.0289, "step": 4385 }, { "epoch": 0.46669504149819113, "grad_norm": 22.6875, "learning_rate": 9.861924388857305e-06, "loss": 1.0961, "step": 4386 }, { "epoch": 0.46680144711640775, "grad_norm": 11.234375, "learning_rate": 9.861888857305289e-06, "loss": 1.1311, "step": 4387 }, { "epoch": 0.4669078527346244, "grad_norm": 10.3984375, "learning_rate": 9.86185332575327e-06, "loss": 1.235, "step": 4388 }, { "epoch": 0.46701425835284105, "grad_norm": 12.0859375, "learning_rate": 9.861817794201252e-06, "loss": 1.1531, "step": 4389 }, { "epoch": 0.46712066397105767, "grad_norm": 12.6953125, "learning_rate": 9.861782262649233e-06, "loss": 0.9487, "step": 4390 }, { "epoch": 0.4672270695892743, "grad_norm": 14.0390625, "learning_rate": 9.861746731097215e-06, "loss": 1.0605, "step": 4391 }, { "epoch": 0.46733347520749097, "grad_norm": 13.8828125, "learning_rate": 9.861711199545197e-06, "loss": 1.0021, "step": 4392 }, { "epoch": 0.4674398808257076, "grad_norm": 15.3359375, "learning_rate": 9.86167566799318e-06, "loss": 1.4795, "step": 4393 }, { "epoch": 0.46754628644392426, "grad_norm": 9.625, "learning_rate": 9.861640136441162e-06, "loss": 0.9144, "step": 4394 }, { "epoch": 0.4676526920621409, "grad_norm": 14.9140625, "learning_rate": 9.861604604889143e-06, "loss": 1.2284, "step": 4395 }, { "epoch": 0.4677590976803575, "grad_norm": 24.4375, "learning_rate": 9.861569073337125e-06, "loss": 1.1027, "step": 4396 }, { "epoch": 0.4678655032985742, "grad_norm": 16.515625, "learning_rate": 9.861533541785105e-06, "loss": 1.5897, "step": 4397 }, { "epoch": 0.4679719089167908, "grad_norm": 10.4375, "learning_rate": 9.861498010233088e-06, "loss": 1.3293, "step": 4398 }, { "epoch": 0.4680783145350074, "grad_norm": 18.265625, "learning_rate": 9.86146247868107e-06, "loss": 1.5003, "step": 4399 }, { "epoch": 0.4681847201532241, "grad_norm": 12.359375, "learning_rate": 9.861426947129051e-06, "loss": 1.4154, "step": 4400 }, { "epoch": 0.4682911257714407, "grad_norm": 13.078125, "learning_rate": 9.861391415577033e-06, "loss": 1.0198, "step": 4401 }, { "epoch": 0.4683975313896574, "grad_norm": 16.875, "learning_rate": 9.861355884025014e-06, "loss": 1.4554, "step": 4402 }, { "epoch": 0.468503937007874, "grad_norm": 21.296875, "learning_rate": 9.861320352472998e-06, "loss": 1.4796, "step": 4403 }, { "epoch": 0.46861034262609064, "grad_norm": 15.0, "learning_rate": 9.86128482092098e-06, "loss": 1.2153, "step": 4404 }, { "epoch": 0.4687167482443073, "grad_norm": 11.1953125, "learning_rate": 9.861249289368961e-06, "loss": 1.2342, "step": 4405 }, { "epoch": 0.46882315386252393, "grad_norm": 15.1640625, "learning_rate": 9.861213757816942e-06, "loss": 1.059, "step": 4406 }, { "epoch": 0.4689295594807406, "grad_norm": 11.8984375, "learning_rate": 9.861178226264924e-06, "loss": 1.1051, "step": 4407 }, { "epoch": 0.46903596509895723, "grad_norm": 12.15625, "learning_rate": 9.861142694712906e-06, "loss": 1.1747, "step": 4408 }, { "epoch": 0.46914237071717385, "grad_norm": 11.453125, "learning_rate": 9.861107163160887e-06, "loss": 1.1955, "step": 4409 }, { "epoch": 0.4692487763353905, "grad_norm": 11.8359375, "learning_rate": 9.861071631608869e-06, "loss": 0.9992, "step": 4410 }, { "epoch": 0.46935518195360715, "grad_norm": 10.8203125, "learning_rate": 9.86103610005685e-06, "loss": 1.1309, "step": 4411 }, { "epoch": 0.46946158757182377, "grad_norm": 14.6953125, "learning_rate": 9.861000568504832e-06, "loss": 1.2637, "step": 4412 }, { "epoch": 0.46956799319004044, "grad_norm": 13.765625, "learning_rate": 9.860965036952815e-06, "loss": 1.4227, "step": 4413 }, { "epoch": 0.46967439880825707, "grad_norm": 14.75, "learning_rate": 9.860929505400797e-06, "loss": 1.3279, "step": 4414 }, { "epoch": 0.46978080442647374, "grad_norm": 12.65625, "learning_rate": 9.860893973848779e-06, "loss": 1.5078, "step": 4415 }, { "epoch": 0.46988721004469036, "grad_norm": 13.9296875, "learning_rate": 9.86085844229676e-06, "loss": 1.0457, "step": 4416 }, { "epoch": 0.469993615662907, "grad_norm": 17.734375, "learning_rate": 9.860822910744742e-06, "loss": 1.4124, "step": 4417 }, { "epoch": 0.47010002128112366, "grad_norm": 9.75, "learning_rate": 9.860787379192725e-06, "loss": 1.1175, "step": 4418 }, { "epoch": 0.4702064268993403, "grad_norm": 12.46875, "learning_rate": 9.860751847640707e-06, "loss": 1.0461, "step": 4419 }, { "epoch": 0.4703128325175569, "grad_norm": 12.484375, "learning_rate": 9.860716316088688e-06, "loss": 1.2875, "step": 4420 }, { "epoch": 0.4704192381357736, "grad_norm": 17.53125, "learning_rate": 9.86068078453667e-06, "loss": 0.9919, "step": 4421 }, { "epoch": 0.4705256437539902, "grad_norm": 15.796875, "learning_rate": 9.860645252984652e-06, "loss": 1.3026, "step": 4422 }, { "epoch": 0.4706320493722069, "grad_norm": 14.5546875, "learning_rate": 9.860609721432633e-06, "loss": 1.0119, "step": 4423 }, { "epoch": 0.4707384549904235, "grad_norm": 10.9140625, "learning_rate": 9.860574189880615e-06, "loss": 1.1479, "step": 4424 }, { "epoch": 0.4708448606086401, "grad_norm": 9.96875, "learning_rate": 9.860538658328596e-06, "loss": 1.0649, "step": 4425 }, { "epoch": 0.4709512662268568, "grad_norm": 15.484375, "learning_rate": 9.860503126776578e-06, "loss": 1.5608, "step": 4426 }, { "epoch": 0.4710576718450734, "grad_norm": 11.3515625, "learning_rate": 9.86046759522456e-06, "loss": 1.2657, "step": 4427 }, { "epoch": 0.4711640774632901, "grad_norm": 16.359375, "learning_rate": 9.860432063672543e-06, "loss": 1.4507, "step": 4428 }, { "epoch": 0.4712704830815067, "grad_norm": 12.3515625, "learning_rate": 9.860396532120525e-06, "loss": 0.9126, "step": 4429 }, { "epoch": 0.47137688869972333, "grad_norm": 16.6875, "learning_rate": 9.860361000568506e-06, "loss": 1.0705, "step": 4430 }, { "epoch": 0.47148329431794, "grad_norm": 19.875, "learning_rate": 9.860325469016488e-06, "loss": 1.285, "step": 4431 }, { "epoch": 0.4715896999361566, "grad_norm": 12.890625, "learning_rate": 9.86028993746447e-06, "loss": 1.3544, "step": 4432 }, { "epoch": 0.47169610555437325, "grad_norm": 15.1171875, "learning_rate": 9.860254405912451e-06, "loss": 1.291, "step": 4433 }, { "epoch": 0.4718025111725899, "grad_norm": 16.3125, "learning_rate": 9.860218874360433e-06, "loss": 1.0696, "step": 4434 }, { "epoch": 0.47190891679080654, "grad_norm": 13.5859375, "learning_rate": 9.860183342808414e-06, "loss": 1.5533, "step": 4435 }, { "epoch": 0.4720153224090232, "grad_norm": 13.6640625, "learning_rate": 9.860147811256396e-06, "loss": 1.3524, "step": 4436 }, { "epoch": 0.47212172802723984, "grad_norm": 22.9375, "learning_rate": 9.860112279704377e-06, "loss": 1.2865, "step": 4437 }, { "epoch": 0.47222813364545646, "grad_norm": 14.0703125, "learning_rate": 9.860076748152359e-06, "loss": 1.2632, "step": 4438 }, { "epoch": 0.47233453926367314, "grad_norm": 13.3359375, "learning_rate": 9.860041216600342e-06, "loss": 1.1444, "step": 4439 }, { "epoch": 0.47244094488188976, "grad_norm": 10.78125, "learning_rate": 9.860005685048324e-06, "loss": 1.1418, "step": 4440 }, { "epoch": 0.4725473505001064, "grad_norm": 10.7734375, "learning_rate": 9.859970153496305e-06, "loss": 1.2882, "step": 4441 }, { "epoch": 0.47265375611832305, "grad_norm": 12.7734375, "learning_rate": 9.859934621944287e-06, "loss": 1.1568, "step": 4442 }, { "epoch": 0.4727601617365397, "grad_norm": 12.671875, "learning_rate": 9.859899090392269e-06, "loss": 1.1658, "step": 4443 }, { "epoch": 0.47286656735475635, "grad_norm": 33.1875, "learning_rate": 9.859863558840252e-06, "loss": 1.4161, "step": 4444 }, { "epoch": 0.47297297297297297, "grad_norm": 14.671875, "learning_rate": 9.859828027288234e-06, "loss": 1.1447, "step": 4445 }, { "epoch": 0.4730793785911896, "grad_norm": 20.296875, "learning_rate": 9.859792495736215e-06, "loss": 1.3763, "step": 4446 }, { "epoch": 0.47318578420940627, "grad_norm": 11.234375, "learning_rate": 9.859756964184197e-06, "loss": 1.187, "step": 4447 }, { "epoch": 0.4732921898276229, "grad_norm": 14.8125, "learning_rate": 9.859721432632178e-06, "loss": 1.2176, "step": 4448 }, { "epoch": 0.47339859544583957, "grad_norm": 12.8203125, "learning_rate": 9.85968590108016e-06, "loss": 1.312, "step": 4449 }, { "epoch": 0.4735050010640562, "grad_norm": 14.046875, "learning_rate": 9.859650369528142e-06, "loss": 1.1252, "step": 4450 }, { "epoch": 0.4736114066822728, "grad_norm": 10.984375, "learning_rate": 9.859614837976123e-06, "loss": 1.3097, "step": 4451 }, { "epoch": 0.4737178123004895, "grad_norm": 13.7265625, "learning_rate": 9.859579306424105e-06, "loss": 1.2283, "step": 4452 }, { "epoch": 0.4738242179187061, "grad_norm": 12.234375, "learning_rate": 9.859543774872086e-06, "loss": 1.0678, "step": 4453 }, { "epoch": 0.4739306235369227, "grad_norm": 53.09375, "learning_rate": 9.85950824332007e-06, "loss": 1.3512, "step": 4454 }, { "epoch": 0.4740370291551394, "grad_norm": 17.234375, "learning_rate": 9.859472711768051e-06, "loss": 1.1097, "step": 4455 }, { "epoch": 0.474143434773356, "grad_norm": 17.515625, "learning_rate": 9.859437180216033e-06, "loss": 1.2617, "step": 4456 }, { "epoch": 0.4742498403915727, "grad_norm": 20.234375, "learning_rate": 9.859401648664015e-06, "loss": 1.1827, "step": 4457 }, { "epoch": 0.4743562460097893, "grad_norm": 15.6953125, "learning_rate": 9.859366117111996e-06, "loss": 1.5948, "step": 4458 }, { "epoch": 0.47446265162800594, "grad_norm": 17.015625, "learning_rate": 9.859330585559978e-06, "loss": 1.4139, "step": 4459 }, { "epoch": 0.4745690572462226, "grad_norm": 12.8203125, "learning_rate": 9.85929505400796e-06, "loss": 1.2902, "step": 4460 }, { "epoch": 0.47467546286443923, "grad_norm": 11.703125, "learning_rate": 9.859259522455941e-06, "loss": 1.0553, "step": 4461 }, { "epoch": 0.4747818684826559, "grad_norm": 11.359375, "learning_rate": 9.859223990903923e-06, "loss": 1.2364, "step": 4462 }, { "epoch": 0.47488827410087253, "grad_norm": 18.875, "learning_rate": 9.859188459351904e-06, "loss": 1.1757, "step": 4463 }, { "epoch": 0.47499467971908915, "grad_norm": 11.5390625, "learning_rate": 9.859152927799888e-06, "loss": 1.4189, "step": 4464 }, { "epoch": 0.47510108533730583, "grad_norm": 13.625, "learning_rate": 9.859117396247869e-06, "loss": 1.3225, "step": 4465 }, { "epoch": 0.47520749095552245, "grad_norm": 12.7578125, "learning_rate": 9.85908186469585e-06, "loss": 1.2512, "step": 4466 }, { "epoch": 0.47531389657373907, "grad_norm": 12.4921875, "learning_rate": 9.859046333143832e-06, "loss": 0.9772, "step": 4467 }, { "epoch": 0.47542030219195575, "grad_norm": 15.8984375, "learning_rate": 9.859010801591814e-06, "loss": 1.3385, "step": 4468 }, { "epoch": 0.47552670781017237, "grad_norm": 11.0703125, "learning_rate": 9.858975270039796e-06, "loss": 1.2356, "step": 4469 }, { "epoch": 0.47563311342838904, "grad_norm": 13.328125, "learning_rate": 9.858939738487779e-06, "loss": 1.2385, "step": 4470 }, { "epoch": 0.47573951904660566, "grad_norm": 14.21875, "learning_rate": 9.85890420693576e-06, "loss": 1.3748, "step": 4471 }, { "epoch": 0.4758459246648223, "grad_norm": 16.59375, "learning_rate": 9.858868675383742e-06, "loss": 1.3219, "step": 4472 }, { "epoch": 0.47595233028303896, "grad_norm": 12.6171875, "learning_rate": 9.858833143831724e-06, "loss": 0.8836, "step": 4473 }, { "epoch": 0.4760587359012556, "grad_norm": 11.734375, "learning_rate": 9.858797612279705e-06, "loss": 1.212, "step": 4474 }, { "epoch": 0.4761651415194722, "grad_norm": 16.234375, "learning_rate": 9.858762080727687e-06, "loss": 1.412, "step": 4475 }, { "epoch": 0.4762715471376889, "grad_norm": 11.859375, "learning_rate": 9.858726549175668e-06, "loss": 1.2265, "step": 4476 }, { "epoch": 0.4763779527559055, "grad_norm": 15.0078125, "learning_rate": 9.85869101762365e-06, "loss": 1.6806, "step": 4477 }, { "epoch": 0.4764843583741222, "grad_norm": 17.921875, "learning_rate": 9.858655486071632e-06, "loss": 1.2604, "step": 4478 }, { "epoch": 0.4765907639923388, "grad_norm": 13.203125, "learning_rate": 9.858619954519613e-06, "loss": 0.8975, "step": 4479 }, { "epoch": 0.4766971696105554, "grad_norm": 14.5078125, "learning_rate": 9.858584422967597e-06, "loss": 1.13, "step": 4480 }, { "epoch": 0.4768035752287721, "grad_norm": 13.109375, "learning_rate": 9.858548891415578e-06, "loss": 0.9125, "step": 4481 }, { "epoch": 0.4769099808469887, "grad_norm": 12.5625, "learning_rate": 9.85851335986356e-06, "loss": 1.1223, "step": 4482 }, { "epoch": 0.4770163864652054, "grad_norm": 20.703125, "learning_rate": 9.858477828311541e-06, "loss": 1.1855, "step": 4483 }, { "epoch": 0.477122792083422, "grad_norm": 14.0078125, "learning_rate": 9.858442296759523e-06, "loss": 1.3273, "step": 4484 }, { "epoch": 0.47722919770163863, "grad_norm": 14.0859375, "learning_rate": 9.858406765207506e-06, "loss": 1.1179, "step": 4485 }, { "epoch": 0.4773356033198553, "grad_norm": 14.3828125, "learning_rate": 9.858371233655486e-06, "loss": 1.0205, "step": 4486 }, { "epoch": 0.4774420089380719, "grad_norm": 12.8671875, "learning_rate": 9.858335702103468e-06, "loss": 1.3999, "step": 4487 }, { "epoch": 0.47754841455628855, "grad_norm": 15.7890625, "learning_rate": 9.85830017055145e-06, "loss": 1.0451, "step": 4488 }, { "epoch": 0.4776548201745052, "grad_norm": 17.46875, "learning_rate": 9.858264638999431e-06, "loss": 1.847, "step": 4489 }, { "epoch": 0.47776122579272184, "grad_norm": 14.3203125, "learning_rate": 9.858229107447414e-06, "loss": 1.291, "step": 4490 }, { "epoch": 0.4778676314109385, "grad_norm": 15.6328125, "learning_rate": 9.858193575895396e-06, "loss": 1.1282, "step": 4491 }, { "epoch": 0.47797403702915514, "grad_norm": 13.1328125, "learning_rate": 9.858158044343378e-06, "loss": 1.172, "step": 4492 }, { "epoch": 0.47808044264737176, "grad_norm": 15.453125, "learning_rate": 9.85812251279136e-06, "loss": 1.4008, "step": 4493 }, { "epoch": 0.47818684826558844, "grad_norm": 11.9140625, "learning_rate": 9.85808698123934e-06, "loss": 1.0415, "step": 4494 }, { "epoch": 0.47829325388380506, "grad_norm": 13.75, "learning_rate": 9.858051449687324e-06, "loss": 1.1389, "step": 4495 }, { "epoch": 0.4783996595020217, "grad_norm": 13.8046875, "learning_rate": 9.858015918135306e-06, "loss": 1.586, "step": 4496 }, { "epoch": 0.47850606512023836, "grad_norm": 13.6015625, "learning_rate": 9.857980386583287e-06, "loss": 1.1543, "step": 4497 }, { "epoch": 0.478612470738455, "grad_norm": 9.53125, "learning_rate": 9.857944855031269e-06, "loss": 1.1341, "step": 4498 }, { "epoch": 0.47871887635667165, "grad_norm": 13.0, "learning_rate": 9.85790932347925e-06, "loss": 1.5633, "step": 4499 }, { "epoch": 0.4788252819748883, "grad_norm": 15.78125, "learning_rate": 9.857873791927232e-06, "loss": 1.2609, "step": 4500 }, { "epoch": 0.4789316875931049, "grad_norm": 13.2890625, "learning_rate": 9.857838260375214e-06, "loss": 1.3574, "step": 4501 }, { "epoch": 0.47903809321132157, "grad_norm": 16.96875, "learning_rate": 9.857802728823195e-06, "loss": 1.5534, "step": 4502 }, { "epoch": 0.4791444988295382, "grad_norm": 14.265625, "learning_rate": 9.857767197271177e-06, "loss": 1.2411, "step": 4503 }, { "epoch": 0.47925090444775487, "grad_norm": 14.2265625, "learning_rate": 9.857731665719159e-06, "loss": 1.7158, "step": 4504 }, { "epoch": 0.4793573100659715, "grad_norm": 14.84375, "learning_rate": 9.857696134167142e-06, "loss": 1.3698, "step": 4505 }, { "epoch": 0.4794637156841881, "grad_norm": 12.390625, "learning_rate": 9.857660602615123e-06, "loss": 1.0618, "step": 4506 }, { "epoch": 0.4795701213024048, "grad_norm": 10.5, "learning_rate": 9.857625071063105e-06, "loss": 1.0268, "step": 4507 }, { "epoch": 0.4796765269206214, "grad_norm": 18.9375, "learning_rate": 9.857589539511087e-06, "loss": 1.3751, "step": 4508 }, { "epoch": 0.479782932538838, "grad_norm": 13.9609375, "learning_rate": 9.857554007959068e-06, "loss": 1.3471, "step": 4509 }, { "epoch": 0.4798893381570547, "grad_norm": 13.875, "learning_rate": 9.85751847640705e-06, "loss": 0.9918, "step": 4510 }, { "epoch": 0.4799957437752713, "grad_norm": 10.671875, "learning_rate": 9.857482944855033e-06, "loss": 1.1231, "step": 4511 }, { "epoch": 0.480102149393488, "grad_norm": 22.921875, "learning_rate": 9.857447413303015e-06, "loss": 1.1783, "step": 4512 }, { "epoch": 0.4802085550117046, "grad_norm": 17.109375, "learning_rate": 9.857411881750996e-06, "loss": 1.5078, "step": 4513 }, { "epoch": 0.48031496062992124, "grad_norm": 15.71875, "learning_rate": 9.857376350198978e-06, "loss": 1.4579, "step": 4514 }, { "epoch": 0.4804213662481379, "grad_norm": 16.421875, "learning_rate": 9.85734081864696e-06, "loss": 1.058, "step": 4515 }, { "epoch": 0.48052777186635454, "grad_norm": 9.6171875, "learning_rate": 9.857305287094941e-06, "loss": 0.9603, "step": 4516 }, { "epoch": 0.4806341774845712, "grad_norm": 12.90625, "learning_rate": 9.857269755542923e-06, "loss": 1.2994, "step": 4517 }, { "epoch": 0.48074058310278783, "grad_norm": 12.2265625, "learning_rate": 9.857234223990904e-06, "loss": 1.2656, "step": 4518 }, { "epoch": 0.48084698872100445, "grad_norm": 13.2265625, "learning_rate": 9.857198692438886e-06, "loss": 1.2367, "step": 4519 }, { "epoch": 0.48095339433922113, "grad_norm": 12.3125, "learning_rate": 9.857163160886868e-06, "loss": 1.2194, "step": 4520 }, { "epoch": 0.48105979995743775, "grad_norm": 13.375, "learning_rate": 9.857127629334851e-06, "loss": 1.5665, "step": 4521 }, { "epoch": 0.48116620557565437, "grad_norm": 13.171875, "learning_rate": 9.857092097782833e-06, "loss": 1.0986, "step": 4522 }, { "epoch": 0.48127261119387105, "grad_norm": 9.640625, "learning_rate": 9.857056566230814e-06, "loss": 1.2061, "step": 4523 }, { "epoch": 0.48137901681208767, "grad_norm": 12.0546875, "learning_rate": 9.857021034678796e-06, "loss": 0.9834, "step": 4524 }, { "epoch": 0.48148542243030434, "grad_norm": 13.359375, "learning_rate": 9.856985503126777e-06, "loss": 1.1848, "step": 4525 }, { "epoch": 0.48159182804852096, "grad_norm": 19.78125, "learning_rate": 9.856949971574759e-06, "loss": 1.0909, "step": 4526 }, { "epoch": 0.4816982336667376, "grad_norm": 14.03125, "learning_rate": 9.85691444002274e-06, "loss": 1.3029, "step": 4527 }, { "epoch": 0.48180463928495426, "grad_norm": 18.0, "learning_rate": 9.856878908470722e-06, "loss": 1.3292, "step": 4528 }, { "epoch": 0.4819110449031709, "grad_norm": 14.0546875, "learning_rate": 9.856843376918704e-06, "loss": 1.2081, "step": 4529 }, { "epoch": 0.4820174505213875, "grad_norm": 35.6875, "learning_rate": 9.856807845366685e-06, "loss": 1.215, "step": 4530 }, { "epoch": 0.4821238561396042, "grad_norm": 13.1171875, "learning_rate": 9.856772313814669e-06, "loss": 1.1607, "step": 4531 }, { "epoch": 0.4822302617578208, "grad_norm": 17.625, "learning_rate": 9.85673678226265e-06, "loss": 1.2011, "step": 4532 }, { "epoch": 0.4823366673760375, "grad_norm": 11.8828125, "learning_rate": 9.856701250710632e-06, "loss": 1.4279, "step": 4533 }, { "epoch": 0.4824430729942541, "grad_norm": 16.796875, "learning_rate": 9.856665719158614e-06, "loss": 1.1045, "step": 4534 }, { "epoch": 0.4825494786124707, "grad_norm": 15.640625, "learning_rate": 9.856630187606595e-06, "loss": 1.3338, "step": 4535 }, { "epoch": 0.4826558842306874, "grad_norm": 12.46875, "learning_rate": 9.856594656054578e-06, "loss": 1.3889, "step": 4536 }, { "epoch": 0.482762289848904, "grad_norm": 12.6640625, "learning_rate": 9.85655912450256e-06, "loss": 1.2546, "step": 4537 }, { "epoch": 0.4828686954671207, "grad_norm": 14.5390625, "learning_rate": 9.856523592950542e-06, "loss": 1.1443, "step": 4538 }, { "epoch": 0.4829751010853373, "grad_norm": 14.546875, "learning_rate": 9.856488061398523e-06, "loss": 1.2049, "step": 4539 }, { "epoch": 0.48308150670355393, "grad_norm": 14.9296875, "learning_rate": 9.856452529846505e-06, "loss": 1.2381, "step": 4540 }, { "epoch": 0.4831879123217706, "grad_norm": 13.625, "learning_rate": 9.856416998294486e-06, "loss": 1.1163, "step": 4541 }, { "epoch": 0.48329431793998723, "grad_norm": 14.7734375, "learning_rate": 9.856381466742468e-06, "loss": 1.2851, "step": 4542 }, { "epoch": 0.48340072355820385, "grad_norm": 11.4296875, "learning_rate": 9.85634593519045e-06, "loss": 1.0188, "step": 4543 }, { "epoch": 0.4835071291764205, "grad_norm": 16.875, "learning_rate": 9.856310403638431e-06, "loss": 1.4139, "step": 4544 }, { "epoch": 0.48361353479463715, "grad_norm": 14.3046875, "learning_rate": 9.856274872086413e-06, "loss": 1.0859, "step": 4545 }, { "epoch": 0.4837199404128538, "grad_norm": 12.3359375, "learning_rate": 9.856239340534395e-06, "loss": 1.5201, "step": 4546 }, { "epoch": 0.48382634603107044, "grad_norm": 9.046875, "learning_rate": 9.856203808982378e-06, "loss": 1.2476, "step": 4547 }, { "epoch": 0.48393275164928706, "grad_norm": 14.171875, "learning_rate": 9.85616827743036e-06, "loss": 1.3018, "step": 4548 }, { "epoch": 0.48403915726750374, "grad_norm": 12.6953125, "learning_rate": 9.856132745878341e-06, "loss": 1.3722, "step": 4549 }, { "epoch": 0.48414556288572036, "grad_norm": 10.6953125, "learning_rate": 9.856097214326323e-06, "loss": 0.8895, "step": 4550 }, { "epoch": 0.484251968503937, "grad_norm": 14.6953125, "learning_rate": 9.856061682774304e-06, "loss": 1.3433, "step": 4551 }, { "epoch": 0.48435837412215366, "grad_norm": 16.359375, "learning_rate": 9.856026151222286e-06, "loss": 1.4075, "step": 4552 }, { "epoch": 0.4844647797403703, "grad_norm": 14.4375, "learning_rate": 9.855990619670267e-06, "loss": 1.2782, "step": 4553 }, { "epoch": 0.48457118535858695, "grad_norm": 11.3984375, "learning_rate": 9.855955088118249e-06, "loss": 0.9979, "step": 4554 }, { "epoch": 0.4846775909768036, "grad_norm": 14.1796875, "learning_rate": 9.85591955656623e-06, "loss": 1.2451, "step": 4555 }, { "epoch": 0.4847839965950202, "grad_norm": 12.5078125, "learning_rate": 9.855884025014212e-06, "loss": 1.2352, "step": 4556 }, { "epoch": 0.48489040221323687, "grad_norm": 15.7109375, "learning_rate": 9.855848493462196e-06, "loss": 1.4933, "step": 4557 }, { "epoch": 0.4849968078314535, "grad_norm": 14.390625, "learning_rate": 9.855812961910177e-06, "loss": 1.0434, "step": 4558 }, { "epoch": 0.48510321344967017, "grad_norm": 12.3515625, "learning_rate": 9.855777430358159e-06, "loss": 1.3947, "step": 4559 }, { "epoch": 0.4852096190678868, "grad_norm": 16.171875, "learning_rate": 9.85574189880614e-06, "loss": 1.2039, "step": 4560 }, { "epoch": 0.4853160246861034, "grad_norm": 11.109375, "learning_rate": 9.855706367254122e-06, "loss": 1.0566, "step": 4561 }, { "epoch": 0.4854224303043201, "grad_norm": 16.96875, "learning_rate": 9.855670835702105e-06, "loss": 1.2817, "step": 4562 }, { "epoch": 0.4855288359225367, "grad_norm": 12.2734375, "learning_rate": 9.855635304150087e-06, "loss": 1.2397, "step": 4563 }, { "epoch": 0.4856352415407533, "grad_norm": 14.2890625, "learning_rate": 9.855599772598069e-06, "loss": 1.2629, "step": 4564 }, { "epoch": 0.48574164715897, "grad_norm": 12.75, "learning_rate": 9.85556424104605e-06, "loss": 1.3382, "step": 4565 }, { "epoch": 0.4858480527771866, "grad_norm": 11.15625, "learning_rate": 9.855528709494032e-06, "loss": 1.0506, "step": 4566 }, { "epoch": 0.4859544583954033, "grad_norm": 11.046875, "learning_rate": 9.855493177942013e-06, "loss": 1.0598, "step": 4567 }, { "epoch": 0.4860608640136199, "grad_norm": 12.953125, "learning_rate": 9.855457646389995e-06, "loss": 1.3589, "step": 4568 }, { "epoch": 0.48616726963183654, "grad_norm": 17.015625, "learning_rate": 9.855422114837977e-06, "loss": 1.2428, "step": 4569 }, { "epoch": 0.4862736752500532, "grad_norm": 13.8984375, "learning_rate": 9.855386583285958e-06, "loss": 0.9278, "step": 4570 }, { "epoch": 0.48638008086826984, "grad_norm": 13.484375, "learning_rate": 9.85535105173394e-06, "loss": 1.0485, "step": 4571 }, { "epoch": 0.4864864864864865, "grad_norm": 11.5234375, "learning_rate": 9.855315520181923e-06, "loss": 0.9281, "step": 4572 }, { "epoch": 0.48659289210470313, "grad_norm": 12.34375, "learning_rate": 9.855279988629905e-06, "loss": 1.5931, "step": 4573 }, { "epoch": 0.48669929772291975, "grad_norm": 14.0234375, "learning_rate": 9.855244457077886e-06, "loss": 1.1904, "step": 4574 }, { "epoch": 0.48680570334113643, "grad_norm": 10.046875, "learning_rate": 9.855208925525868e-06, "loss": 1.0966, "step": 4575 }, { "epoch": 0.48691210895935305, "grad_norm": 15.171875, "learning_rate": 9.85517339397385e-06, "loss": 1.1909, "step": 4576 }, { "epoch": 0.48701851457756967, "grad_norm": 13.4921875, "learning_rate": 9.855137862421831e-06, "loss": 1.4345, "step": 4577 }, { "epoch": 0.48712492019578635, "grad_norm": 20.25, "learning_rate": 9.855102330869813e-06, "loss": 1.5181, "step": 4578 }, { "epoch": 0.48723132581400297, "grad_norm": 13.578125, "learning_rate": 9.855066799317794e-06, "loss": 1.1377, "step": 4579 }, { "epoch": 0.48733773143221965, "grad_norm": 12.1328125, "learning_rate": 9.855031267765776e-06, "loss": 1.1364, "step": 4580 }, { "epoch": 0.48744413705043627, "grad_norm": 11.4375, "learning_rate": 9.854995736213758e-06, "loss": 1.0188, "step": 4581 }, { "epoch": 0.4875505426686529, "grad_norm": 23.171875, "learning_rate": 9.85496020466174e-06, "loss": 1.4623, "step": 4582 }, { "epoch": 0.48765694828686956, "grad_norm": 13.6953125, "learning_rate": 9.854924673109722e-06, "loss": 1.5052, "step": 4583 }, { "epoch": 0.4877633539050862, "grad_norm": 11.734375, "learning_rate": 9.854889141557704e-06, "loss": 1.2217, "step": 4584 }, { "epoch": 0.4878697595233028, "grad_norm": 13.9296875, "learning_rate": 9.854853610005686e-06, "loss": 1.7377, "step": 4585 }, { "epoch": 0.4879761651415195, "grad_norm": 11.5546875, "learning_rate": 9.854818078453667e-06, "loss": 1.1147, "step": 4586 }, { "epoch": 0.4880825707597361, "grad_norm": 14.90625, "learning_rate": 9.854782546901649e-06, "loss": 1.4501, "step": 4587 }, { "epoch": 0.4881889763779528, "grad_norm": 15.34375, "learning_rate": 9.854747015349632e-06, "loss": 1.4137, "step": 4588 }, { "epoch": 0.4882953819961694, "grad_norm": 13.4765625, "learning_rate": 9.854711483797614e-06, "loss": 1.349, "step": 4589 }, { "epoch": 0.488401787614386, "grad_norm": 11.7265625, "learning_rate": 9.854675952245595e-06, "loss": 1.0402, "step": 4590 }, { "epoch": 0.4885081932326027, "grad_norm": 12.2421875, "learning_rate": 9.854640420693577e-06, "loss": 1.3476, "step": 4591 }, { "epoch": 0.4886145988508193, "grad_norm": 11.546875, "learning_rate": 9.854604889141559e-06, "loss": 1.1927, "step": 4592 }, { "epoch": 0.488721004469036, "grad_norm": 16.8125, "learning_rate": 9.85456935758954e-06, "loss": 1.0967, "step": 4593 }, { "epoch": 0.4888274100872526, "grad_norm": 13.890625, "learning_rate": 9.854533826037522e-06, "loss": 1.6858, "step": 4594 }, { "epoch": 0.48893381570546923, "grad_norm": 15.46875, "learning_rate": 9.854498294485503e-06, "loss": 1.3097, "step": 4595 }, { "epoch": 0.4890402213236859, "grad_norm": 14.09375, "learning_rate": 9.854462762933485e-06, "loss": 1.5604, "step": 4596 }, { "epoch": 0.48914662694190253, "grad_norm": 12.1171875, "learning_rate": 9.854427231381467e-06, "loss": 1.154, "step": 4597 }, { "epoch": 0.48925303256011915, "grad_norm": 16.109375, "learning_rate": 9.85439169982945e-06, "loss": 1.2438, "step": 4598 }, { "epoch": 0.4893594381783358, "grad_norm": 12.0078125, "learning_rate": 9.854356168277432e-06, "loss": 1.1175, "step": 4599 }, { "epoch": 0.48946584379655245, "grad_norm": 12.5625, "learning_rate": 9.854320636725413e-06, "loss": 1.1771, "step": 4600 }, { "epoch": 0.4895722494147691, "grad_norm": 11.3046875, "learning_rate": 9.854285105173395e-06, "loss": 1.1744, "step": 4601 }, { "epoch": 0.48967865503298574, "grad_norm": 14.96875, "learning_rate": 9.854249573621376e-06, "loss": 1.4638, "step": 4602 }, { "epoch": 0.48978506065120236, "grad_norm": 11.96875, "learning_rate": 9.85421404206936e-06, "loss": 1.1808, "step": 4603 }, { "epoch": 0.48989146626941904, "grad_norm": 13.578125, "learning_rate": 9.854178510517341e-06, "loss": 1.3037, "step": 4604 }, { "epoch": 0.48999787188763566, "grad_norm": 14.453125, "learning_rate": 9.854142978965323e-06, "loss": 0.9162, "step": 4605 }, { "epoch": 0.4901042775058523, "grad_norm": 15.0859375, "learning_rate": 9.854107447413303e-06, "loss": 1.5308, "step": 4606 }, { "epoch": 0.49021068312406896, "grad_norm": 15.65625, "learning_rate": 9.854071915861284e-06, "loss": 1.2922, "step": 4607 }, { "epoch": 0.4903170887422856, "grad_norm": 18.1875, "learning_rate": 9.854036384309268e-06, "loss": 1.1991, "step": 4608 }, { "epoch": 0.49042349436050225, "grad_norm": 10.3515625, "learning_rate": 9.85400085275725e-06, "loss": 0.9789, "step": 4609 }, { "epoch": 0.4905298999787189, "grad_norm": 12.6328125, "learning_rate": 9.853965321205231e-06, "loss": 1.0621, "step": 4610 }, { "epoch": 0.4906363055969355, "grad_norm": 17.40625, "learning_rate": 9.853929789653213e-06, "loss": 1.4913, "step": 4611 }, { "epoch": 0.49074271121515217, "grad_norm": 11.8125, "learning_rate": 9.853894258101194e-06, "loss": 1.1481, "step": 4612 }, { "epoch": 0.4908491168333688, "grad_norm": 14.0234375, "learning_rate": 9.853858726549177e-06, "loss": 0.9765, "step": 4613 }, { "epoch": 0.49095552245158547, "grad_norm": 17.84375, "learning_rate": 9.853823194997159e-06, "loss": 1.2985, "step": 4614 }, { "epoch": 0.4910619280698021, "grad_norm": 15.7734375, "learning_rate": 9.85378766344514e-06, "loss": 1.2181, "step": 4615 }, { "epoch": 0.4911683336880187, "grad_norm": 12.7578125, "learning_rate": 9.853752131893122e-06, "loss": 1.208, "step": 4616 }, { "epoch": 0.4912747393062354, "grad_norm": 12.1171875, "learning_rate": 9.853716600341104e-06, "loss": 1.3813, "step": 4617 }, { "epoch": 0.491381144924452, "grad_norm": 14.7421875, "learning_rate": 9.853681068789085e-06, "loss": 1.3111, "step": 4618 }, { "epoch": 0.4914875505426686, "grad_norm": 14.3828125, "learning_rate": 9.853645537237067e-06, "loss": 1.2638, "step": 4619 }, { "epoch": 0.4915939561608853, "grad_norm": 12.5078125, "learning_rate": 9.853610005685049e-06, "loss": 1.2655, "step": 4620 }, { "epoch": 0.4917003617791019, "grad_norm": 17.75, "learning_rate": 9.85357447413303e-06, "loss": 1.208, "step": 4621 }, { "epoch": 0.4918067673973186, "grad_norm": 16.6875, "learning_rate": 9.853538942581012e-06, "loss": 1.6855, "step": 4622 }, { "epoch": 0.4919131730155352, "grad_norm": 12.0546875, "learning_rate": 9.853503411028995e-06, "loss": 1.1638, "step": 4623 }, { "epoch": 0.49201957863375184, "grad_norm": 17.078125, "learning_rate": 9.853467879476977e-06, "loss": 1.1324, "step": 4624 }, { "epoch": 0.4921259842519685, "grad_norm": 16.5, "learning_rate": 9.853432347924958e-06, "loss": 1.1032, "step": 4625 }, { "epoch": 0.49223238987018514, "grad_norm": 12.3046875, "learning_rate": 9.85339681637294e-06, "loss": 1.173, "step": 4626 }, { "epoch": 0.4923387954884018, "grad_norm": 10.859375, "learning_rate": 9.853361284820922e-06, "loss": 1.1264, "step": 4627 }, { "epoch": 0.49244520110661844, "grad_norm": 13.5390625, "learning_rate": 9.853325753268903e-06, "loss": 1.1642, "step": 4628 }, { "epoch": 0.49255160672483506, "grad_norm": 11.6484375, "learning_rate": 9.853290221716887e-06, "loss": 1.3105, "step": 4629 }, { "epoch": 0.49265801234305173, "grad_norm": 16.984375, "learning_rate": 9.853254690164868e-06, "loss": 1.4264, "step": 4630 }, { "epoch": 0.49276441796126835, "grad_norm": 22.75, "learning_rate": 9.85321915861285e-06, "loss": 1.4928, "step": 4631 }, { "epoch": 0.492870823579485, "grad_norm": 16.578125, "learning_rate": 9.853183627060831e-06, "loss": 1.6463, "step": 4632 }, { "epoch": 0.49297722919770165, "grad_norm": 12.734375, "learning_rate": 9.853148095508813e-06, "loss": 1.3251, "step": 4633 }, { "epoch": 0.49308363481591827, "grad_norm": 15.9765625, "learning_rate": 9.853112563956795e-06, "loss": 1.3162, "step": 4634 }, { "epoch": 0.49319004043413495, "grad_norm": 14.1875, "learning_rate": 9.853077032404776e-06, "loss": 1.1878, "step": 4635 }, { "epoch": 0.49329644605235157, "grad_norm": 15.421875, "learning_rate": 9.853041500852758e-06, "loss": 1.3565, "step": 4636 }, { "epoch": 0.4934028516705682, "grad_norm": 15.453125, "learning_rate": 9.85300596930074e-06, "loss": 1.2028, "step": 4637 }, { "epoch": 0.49350925728878486, "grad_norm": 12.453125, "learning_rate": 9.852970437748721e-06, "loss": 1.3068, "step": 4638 }, { "epoch": 0.4936156629070015, "grad_norm": 11.3671875, "learning_rate": 9.852934906196704e-06, "loss": 1.24, "step": 4639 }, { "epoch": 0.4937220685252181, "grad_norm": 15.1171875, "learning_rate": 9.852899374644686e-06, "loss": 1.0549, "step": 4640 }, { "epoch": 0.4938284741434348, "grad_norm": 13.6796875, "learning_rate": 9.852863843092667e-06, "loss": 1.6918, "step": 4641 }, { "epoch": 0.4939348797616514, "grad_norm": 12.921875, "learning_rate": 9.852828311540649e-06, "loss": 1.4671, "step": 4642 }, { "epoch": 0.4940412853798681, "grad_norm": 13.609375, "learning_rate": 9.85279277998863e-06, "loss": 1.3395, "step": 4643 }, { "epoch": 0.4941476909980847, "grad_norm": 14.734375, "learning_rate": 9.852757248436612e-06, "loss": 1.4153, "step": 4644 }, { "epoch": 0.4942540966163013, "grad_norm": 18.03125, "learning_rate": 9.852721716884594e-06, "loss": 1.192, "step": 4645 }, { "epoch": 0.494360502234518, "grad_norm": 12.2578125, "learning_rate": 9.852686185332576e-06, "loss": 1.4471, "step": 4646 }, { "epoch": 0.4944669078527346, "grad_norm": 13.4921875, "learning_rate": 9.852650653780557e-06, "loss": 1.0826, "step": 4647 }, { "epoch": 0.4945733134709513, "grad_norm": 12.875, "learning_rate": 9.852615122228539e-06, "loss": 1.4446, "step": 4648 }, { "epoch": 0.4946797190891679, "grad_norm": 10.40625, "learning_rate": 9.852579590676522e-06, "loss": 1.3182, "step": 4649 }, { "epoch": 0.49478612470738453, "grad_norm": 13.3515625, "learning_rate": 9.852544059124504e-06, "loss": 1.2489, "step": 4650 }, { "epoch": 0.4948925303256012, "grad_norm": 14.625, "learning_rate": 9.852508527572485e-06, "loss": 1.2318, "step": 4651 }, { "epoch": 0.49499893594381783, "grad_norm": 13.6328125, "learning_rate": 9.852472996020467e-06, "loss": 1.3682, "step": 4652 }, { "epoch": 0.49510534156203445, "grad_norm": 12.0234375, "learning_rate": 9.852437464468448e-06, "loss": 1.3825, "step": 4653 }, { "epoch": 0.4952117471802511, "grad_norm": 13.34375, "learning_rate": 9.852401932916432e-06, "loss": 1.0999, "step": 4654 }, { "epoch": 0.49531815279846775, "grad_norm": 12.2109375, "learning_rate": 9.852366401364413e-06, "loss": 1.47, "step": 4655 }, { "epoch": 0.4954245584166844, "grad_norm": 15.765625, "learning_rate": 9.852330869812395e-06, "loss": 1.6303, "step": 4656 }, { "epoch": 0.49553096403490104, "grad_norm": 12.7734375, "learning_rate": 9.852295338260377e-06, "loss": 1.1929, "step": 4657 }, { "epoch": 0.49563736965311767, "grad_norm": 11.7109375, "learning_rate": 9.852259806708358e-06, "loss": 1.4588, "step": 4658 }, { "epoch": 0.49574377527133434, "grad_norm": 18.09375, "learning_rate": 9.85222427515634e-06, "loss": 1.4723, "step": 4659 }, { "epoch": 0.49585018088955096, "grad_norm": 14.3984375, "learning_rate": 9.852188743604321e-06, "loss": 1.4519, "step": 4660 }, { "epoch": 0.4959565865077676, "grad_norm": 10.84375, "learning_rate": 9.852153212052303e-06, "loss": 1.3666, "step": 4661 }, { "epoch": 0.49606299212598426, "grad_norm": 14.9453125, "learning_rate": 9.852117680500285e-06, "loss": 1.248, "step": 4662 }, { "epoch": 0.4961693977442009, "grad_norm": 10.3125, "learning_rate": 9.852082148948266e-06, "loss": 1.0787, "step": 4663 }, { "epoch": 0.49627580336241756, "grad_norm": 12.0, "learning_rate": 9.852046617396248e-06, "loss": 1.1478, "step": 4664 }, { "epoch": 0.4963822089806342, "grad_norm": 12.5234375, "learning_rate": 9.852011085844231e-06, "loss": 1.5002, "step": 4665 }, { "epoch": 0.4964886145988508, "grad_norm": 13.046875, "learning_rate": 9.851975554292213e-06, "loss": 1.0888, "step": 4666 }, { "epoch": 0.4965950202170675, "grad_norm": 20.015625, "learning_rate": 9.851940022740194e-06, "loss": 1.2722, "step": 4667 }, { "epoch": 0.4967014258352841, "grad_norm": 11.6484375, "learning_rate": 9.851904491188176e-06, "loss": 1.3723, "step": 4668 }, { "epoch": 0.49680783145350077, "grad_norm": 14.7265625, "learning_rate": 9.851868959636158e-06, "loss": 1.5435, "step": 4669 }, { "epoch": 0.4969142370717174, "grad_norm": 14.3984375, "learning_rate": 9.85183342808414e-06, "loss": 1.1017, "step": 4670 }, { "epoch": 0.497020642689934, "grad_norm": 12.15625, "learning_rate": 9.85179789653212e-06, "loss": 1.1945, "step": 4671 }, { "epoch": 0.4971270483081507, "grad_norm": 23.03125, "learning_rate": 9.851762364980102e-06, "loss": 1.6984, "step": 4672 }, { "epoch": 0.4972334539263673, "grad_norm": 13.546875, "learning_rate": 9.851726833428084e-06, "loss": 1.3409, "step": 4673 }, { "epoch": 0.49733985954458393, "grad_norm": 12.6953125, "learning_rate": 9.851691301876066e-06, "loss": 1.2132, "step": 4674 }, { "epoch": 0.4974462651628006, "grad_norm": 16.359375, "learning_rate": 9.851655770324049e-06, "loss": 1.2606, "step": 4675 }, { "epoch": 0.4975526707810172, "grad_norm": 13.1640625, "learning_rate": 9.85162023877203e-06, "loss": 1.0371, "step": 4676 }, { "epoch": 0.4976590763992339, "grad_norm": 15.2265625, "learning_rate": 9.851584707220012e-06, "loss": 1.1159, "step": 4677 }, { "epoch": 0.4977654820174505, "grad_norm": 14.4296875, "learning_rate": 9.851549175667994e-06, "loss": 1.3627, "step": 4678 }, { "epoch": 0.49787188763566714, "grad_norm": 14.359375, "learning_rate": 9.851513644115975e-06, "loss": 1.1477, "step": 4679 }, { "epoch": 0.4979782932538838, "grad_norm": 12.4453125, "learning_rate": 9.851478112563959e-06, "loss": 1.2845, "step": 4680 }, { "epoch": 0.49808469887210044, "grad_norm": 10.2265625, "learning_rate": 9.85144258101194e-06, "loss": 1.1463, "step": 4681 }, { "epoch": 0.4981911044903171, "grad_norm": 9.5703125, "learning_rate": 9.851407049459922e-06, "loss": 1.0509, "step": 4682 }, { "epoch": 0.49829751010853374, "grad_norm": 19.734375, "learning_rate": 9.851371517907903e-06, "loss": 1.4704, "step": 4683 }, { "epoch": 0.49840391572675036, "grad_norm": 12.8984375, "learning_rate": 9.851335986355885e-06, "loss": 1.3861, "step": 4684 }, { "epoch": 0.49851032134496703, "grad_norm": 18.890625, "learning_rate": 9.851300454803867e-06, "loss": 1.7937, "step": 4685 }, { "epoch": 0.49861672696318365, "grad_norm": 15.578125, "learning_rate": 9.851264923251848e-06, "loss": 1.4214, "step": 4686 }, { "epoch": 0.4987231325814003, "grad_norm": 14.59375, "learning_rate": 9.85122939169983e-06, "loss": 1.9682, "step": 4687 }, { "epoch": 0.49882953819961695, "grad_norm": 14.1015625, "learning_rate": 9.851193860147811e-06, "loss": 1.262, "step": 4688 }, { "epoch": 0.49893594381783357, "grad_norm": 12.109375, "learning_rate": 9.851158328595793e-06, "loss": 1.0476, "step": 4689 }, { "epoch": 0.49904234943605025, "grad_norm": 15.609375, "learning_rate": 9.851122797043776e-06, "loss": 1.0018, "step": 4690 }, { "epoch": 0.49914875505426687, "grad_norm": 12.3359375, "learning_rate": 9.851087265491758e-06, "loss": 1.4007, "step": 4691 }, { "epoch": 0.4992551606724835, "grad_norm": 15.3046875, "learning_rate": 9.85105173393974e-06, "loss": 1.1922, "step": 4692 }, { "epoch": 0.49936156629070017, "grad_norm": 17.15625, "learning_rate": 9.851016202387721e-06, "loss": 1.1834, "step": 4693 }, { "epoch": 0.4994679719089168, "grad_norm": 12.4453125, "learning_rate": 9.850980670835703e-06, "loss": 1.3505, "step": 4694 }, { "epoch": 0.4995743775271334, "grad_norm": 13.296875, "learning_rate": 9.850945139283684e-06, "loss": 1.0082, "step": 4695 }, { "epoch": 0.4996807831453501, "grad_norm": 11.328125, "learning_rate": 9.850909607731666e-06, "loss": 1.074, "step": 4696 }, { "epoch": 0.4997871887635667, "grad_norm": 15.8359375, "learning_rate": 9.850874076179648e-06, "loss": 1.4196, "step": 4697 }, { "epoch": 0.4998935943817834, "grad_norm": 16.984375, "learning_rate": 9.85083854462763e-06, "loss": 1.0678, "step": 4698 }, { "epoch": 0.5, "grad_norm": 20.296875, "learning_rate": 9.85080301307561e-06, "loss": 1.6111, "step": 4699 }, { "epoch": 0.5001064056182166, "grad_norm": 16.4375, "learning_rate": 9.850767481523594e-06, "loss": 1.5255, "step": 4700 }, { "epoch": 0.5002128112364332, "grad_norm": 11.046875, "learning_rate": 9.850731949971576e-06, "loss": 1.1212, "step": 4701 }, { "epoch": 0.50031921685465, "grad_norm": 15.0078125, "learning_rate": 9.850696418419557e-06, "loss": 1.673, "step": 4702 }, { "epoch": 0.5004256224728666, "grad_norm": 15.53125, "learning_rate": 9.850660886867539e-06, "loss": 1.5788, "step": 4703 }, { "epoch": 0.5005320280910832, "grad_norm": 11.0390625, "learning_rate": 9.85062535531552e-06, "loss": 1.2133, "step": 4704 }, { "epoch": 0.5006384337092998, "grad_norm": 13.6640625, "learning_rate": 9.850589823763502e-06, "loss": 1.1511, "step": 4705 }, { "epoch": 0.5007448393275165, "grad_norm": 10.7890625, "learning_rate": 9.850554292211485e-06, "loss": 1.315, "step": 4706 }, { "epoch": 0.5008512449457332, "grad_norm": 14.3515625, "learning_rate": 9.850518760659467e-06, "loss": 1.1691, "step": 4707 }, { "epoch": 0.5009576505639498, "grad_norm": 18.890625, "learning_rate": 9.850483229107449e-06, "loss": 1.4635, "step": 4708 }, { "epoch": 0.5010640561821664, "grad_norm": 10.359375, "learning_rate": 9.85044769755543e-06, "loss": 1.226, "step": 4709 }, { "epoch": 0.501170461800383, "grad_norm": 15.953125, "learning_rate": 9.850412166003412e-06, "loss": 1.6832, "step": 4710 }, { "epoch": 0.5012768674185997, "grad_norm": 14.265625, "learning_rate": 9.850376634451393e-06, "loss": 1.2757, "step": 4711 }, { "epoch": 0.5013832730368163, "grad_norm": 20.5, "learning_rate": 9.850341102899375e-06, "loss": 1.6318, "step": 4712 }, { "epoch": 0.501489678655033, "grad_norm": 13.7421875, "learning_rate": 9.850305571347357e-06, "loss": 1.0629, "step": 4713 }, { "epoch": 0.5015960842732496, "grad_norm": 13.15625, "learning_rate": 9.850270039795338e-06, "loss": 1.3081, "step": 4714 }, { "epoch": 0.5017024898914663, "grad_norm": 13.40625, "learning_rate": 9.85023450824332e-06, "loss": 1.0067, "step": 4715 }, { "epoch": 0.5018088955096829, "grad_norm": 16.59375, "learning_rate": 9.850198976691303e-06, "loss": 1.3792, "step": 4716 }, { "epoch": 0.5019153011278995, "grad_norm": 15.0, "learning_rate": 9.850163445139285e-06, "loss": 1.1266, "step": 4717 }, { "epoch": 0.5020217067461162, "grad_norm": 13.3203125, "learning_rate": 9.850127913587266e-06, "loss": 1.2, "step": 4718 }, { "epoch": 0.5021281123643329, "grad_norm": 13.90625, "learning_rate": 9.850092382035248e-06, "loss": 1.9879, "step": 4719 }, { "epoch": 0.5022345179825495, "grad_norm": 12.5078125, "learning_rate": 9.85005685048323e-06, "loss": 1.1117, "step": 4720 }, { "epoch": 0.5023409236007661, "grad_norm": 13.546875, "learning_rate": 9.850021318931213e-06, "loss": 1.3983, "step": 4721 }, { "epoch": 0.5024473292189827, "grad_norm": 15.9765625, "learning_rate": 9.849985787379195e-06, "loss": 1.157, "step": 4722 }, { "epoch": 0.5025537348371995, "grad_norm": 16.609375, "learning_rate": 9.849950255827176e-06, "loss": 1.1744, "step": 4723 }, { "epoch": 0.5026601404554161, "grad_norm": 17.75, "learning_rate": 9.849914724275158e-06, "loss": 1.6882, "step": 4724 }, { "epoch": 0.5027665460736327, "grad_norm": 18.234375, "learning_rate": 9.849879192723138e-06, "loss": 1.4532, "step": 4725 }, { "epoch": 0.5028729516918493, "grad_norm": 17.328125, "learning_rate": 9.849843661171121e-06, "loss": 1.4065, "step": 4726 }, { "epoch": 0.5029793573100659, "grad_norm": 14.0703125, "learning_rate": 9.849808129619103e-06, "loss": 0.9426, "step": 4727 }, { "epoch": 0.5030857629282827, "grad_norm": 11.359375, "learning_rate": 9.849772598067084e-06, "loss": 1.1069, "step": 4728 }, { "epoch": 0.5031921685464993, "grad_norm": 12.9765625, "learning_rate": 9.849737066515066e-06, "loss": 1.1363, "step": 4729 }, { "epoch": 0.5032985741647159, "grad_norm": 13.421875, "learning_rate": 9.849701534963047e-06, "loss": 1.3796, "step": 4730 }, { "epoch": 0.5034049797829325, "grad_norm": 15.578125, "learning_rate": 9.84966600341103e-06, "loss": 1.6001, "step": 4731 }, { "epoch": 0.5035113854011491, "grad_norm": 13.0546875, "learning_rate": 9.849630471859012e-06, "loss": 1.3689, "step": 4732 }, { "epoch": 0.5036177910193658, "grad_norm": 13.3125, "learning_rate": 9.849594940306994e-06, "loss": 1.051, "step": 4733 }, { "epoch": 0.5037241966375825, "grad_norm": 13.8046875, "learning_rate": 9.849559408754976e-06, "loss": 1.4138, "step": 4734 }, { "epoch": 0.5038306022557991, "grad_norm": 10.875, "learning_rate": 9.849523877202957e-06, "loss": 1.2059, "step": 4735 }, { "epoch": 0.5039370078740157, "grad_norm": 12.84375, "learning_rate": 9.849488345650939e-06, "loss": 1.3618, "step": 4736 }, { "epoch": 0.5040434134922324, "grad_norm": 9.2578125, "learning_rate": 9.84945281409892e-06, "loss": 1.2363, "step": 4737 }, { "epoch": 0.504149819110449, "grad_norm": 11.3671875, "learning_rate": 9.849417282546902e-06, "loss": 1.5061, "step": 4738 }, { "epoch": 0.5042562247286657, "grad_norm": 20.8125, "learning_rate": 9.849381750994884e-06, "loss": 1.3086, "step": 4739 }, { "epoch": 0.5043626303468823, "grad_norm": 13.0546875, "learning_rate": 9.849346219442865e-06, "loss": 1.3339, "step": 4740 }, { "epoch": 0.504469035965099, "grad_norm": 12.2734375, "learning_rate": 9.849310687890848e-06, "loss": 1.2535, "step": 4741 }, { "epoch": 0.5045754415833156, "grad_norm": 13.640625, "learning_rate": 9.84927515633883e-06, "loss": 1.0018, "step": 4742 }, { "epoch": 0.5046818472015322, "grad_norm": 18.8125, "learning_rate": 9.849239624786812e-06, "loss": 0.9938, "step": 4743 }, { "epoch": 0.5047882528197489, "grad_norm": 14.984375, "learning_rate": 9.849204093234793e-06, "loss": 1.4807, "step": 4744 }, { "epoch": 0.5048946584379655, "grad_norm": 11.7109375, "learning_rate": 9.849168561682775e-06, "loss": 1.3593, "step": 4745 }, { "epoch": 0.5050010640561822, "grad_norm": 11.5625, "learning_rate": 9.849133030130757e-06, "loss": 1.0014, "step": 4746 }, { "epoch": 0.5051074696743988, "grad_norm": 14.7109375, "learning_rate": 9.84909749857874e-06, "loss": 1.0814, "step": 4747 }, { "epoch": 0.5052138752926154, "grad_norm": 46.3125, "learning_rate": 9.849061967026721e-06, "loss": 0.9979, "step": 4748 }, { "epoch": 0.5053202809108321, "grad_norm": 18.21875, "learning_rate": 9.849026435474703e-06, "loss": 1.161, "step": 4749 }, { "epoch": 0.5054266865290488, "grad_norm": 19.703125, "learning_rate": 9.848990903922685e-06, "loss": 1.364, "step": 4750 }, { "epoch": 0.5055330921472654, "grad_norm": 14.0859375, "learning_rate": 9.848955372370666e-06, "loss": 1.337, "step": 4751 }, { "epoch": 0.505639497765482, "grad_norm": 14.7734375, "learning_rate": 9.848919840818648e-06, "loss": 1.2804, "step": 4752 }, { "epoch": 0.5057459033836986, "grad_norm": 11.1875, "learning_rate": 9.84888430926663e-06, "loss": 1.0738, "step": 4753 }, { "epoch": 0.5058523090019152, "grad_norm": 14.03125, "learning_rate": 9.848848777714611e-06, "loss": 1.1391, "step": 4754 }, { "epoch": 0.505958714620132, "grad_norm": 13.7421875, "learning_rate": 9.848813246162593e-06, "loss": 1.3803, "step": 4755 }, { "epoch": 0.5060651202383486, "grad_norm": 15.6875, "learning_rate": 9.848777714610574e-06, "loss": 1.3413, "step": 4756 }, { "epoch": 0.5061715258565652, "grad_norm": 14.1640625, "learning_rate": 9.848742183058558e-06, "loss": 1.2421, "step": 4757 }, { "epoch": 0.5062779314747818, "grad_norm": 16.953125, "learning_rate": 9.84870665150654e-06, "loss": 1.8435, "step": 4758 }, { "epoch": 0.5063843370929985, "grad_norm": 12.4375, "learning_rate": 9.84867111995452e-06, "loss": 1.0823, "step": 4759 }, { "epoch": 0.5064907427112152, "grad_norm": 13.359375, "learning_rate": 9.848635588402502e-06, "loss": 1.2543, "step": 4760 }, { "epoch": 0.5065971483294318, "grad_norm": 10.8359375, "learning_rate": 9.848600056850484e-06, "loss": 1.3883, "step": 4761 }, { "epoch": 0.5067035539476484, "grad_norm": 8.4765625, "learning_rate": 9.848564525298466e-06, "loss": 1.0055, "step": 4762 }, { "epoch": 0.506809959565865, "grad_norm": 12.9921875, "learning_rate": 9.848528993746447e-06, "loss": 1.167, "step": 4763 }, { "epoch": 0.5069163651840817, "grad_norm": 10.2265625, "learning_rate": 9.848493462194429e-06, "loss": 1.087, "step": 4764 }, { "epoch": 0.5070227708022984, "grad_norm": 14.046875, "learning_rate": 9.84845793064241e-06, "loss": 1.2903, "step": 4765 }, { "epoch": 0.507129176420515, "grad_norm": 12.8984375, "learning_rate": 9.848422399090392e-06, "loss": 1.1005, "step": 4766 }, { "epoch": 0.5072355820387316, "grad_norm": 11.96875, "learning_rate": 9.848386867538375e-06, "loss": 0.9783, "step": 4767 }, { "epoch": 0.5073419876569483, "grad_norm": 9.625, "learning_rate": 9.848351335986357e-06, "loss": 1.0277, "step": 4768 }, { "epoch": 0.5074483932751649, "grad_norm": 12.4140625, "learning_rate": 9.848315804434339e-06, "loss": 1.1481, "step": 4769 }, { "epoch": 0.5075547988933816, "grad_norm": 13.8984375, "learning_rate": 9.84828027288232e-06, "loss": 1.1714, "step": 4770 }, { "epoch": 0.5076612045115982, "grad_norm": 13.1171875, "learning_rate": 9.848244741330302e-06, "loss": 1.2238, "step": 4771 }, { "epoch": 0.5077676101298149, "grad_norm": 15.0859375, "learning_rate": 9.848209209778285e-06, "loss": 1.0475, "step": 4772 }, { "epoch": 0.5078740157480315, "grad_norm": 12.84375, "learning_rate": 9.848173678226267e-06, "loss": 1.1158, "step": 4773 }, { "epoch": 0.5079804213662481, "grad_norm": 13.6171875, "learning_rate": 9.848138146674248e-06, "loss": 1.1412, "step": 4774 }, { "epoch": 0.5080868269844648, "grad_norm": 13.9453125, "learning_rate": 9.84810261512223e-06, "loss": 1.2969, "step": 4775 }, { "epoch": 0.5081932326026815, "grad_norm": 14.4140625, "learning_rate": 9.848067083570211e-06, "loss": 1.2588, "step": 4776 }, { "epoch": 0.5082996382208981, "grad_norm": 13.7109375, "learning_rate": 9.848031552018193e-06, "loss": 1.6217, "step": 4777 }, { "epoch": 0.5084060438391147, "grad_norm": 12.0625, "learning_rate": 9.847996020466175e-06, "loss": 1.2253, "step": 4778 }, { "epoch": 0.5085124494573313, "grad_norm": 17.609375, "learning_rate": 9.847960488914156e-06, "loss": 1.3995, "step": 4779 }, { "epoch": 0.5086188550755479, "grad_norm": 14.3046875, "learning_rate": 9.847924957362138e-06, "loss": 1.3043, "step": 4780 }, { "epoch": 0.5087252606937647, "grad_norm": 13.6328125, "learning_rate": 9.84788942581012e-06, "loss": 1.0101, "step": 4781 }, { "epoch": 0.5088316663119813, "grad_norm": 11.4453125, "learning_rate": 9.847853894258101e-06, "loss": 1.1601, "step": 4782 }, { "epoch": 0.5089380719301979, "grad_norm": 12.3984375, "learning_rate": 9.847818362706084e-06, "loss": 1.2114, "step": 4783 }, { "epoch": 0.5090444775484145, "grad_norm": 11.5546875, "learning_rate": 9.847782831154066e-06, "loss": 1.095, "step": 4784 }, { "epoch": 0.5091508831666312, "grad_norm": 17.046875, "learning_rate": 9.847747299602048e-06, "loss": 1.6439, "step": 4785 }, { "epoch": 0.5092572887848479, "grad_norm": 11.34375, "learning_rate": 9.84771176805003e-06, "loss": 1.1073, "step": 4786 }, { "epoch": 0.5093636944030645, "grad_norm": 12.7890625, "learning_rate": 9.847676236498011e-06, "loss": 1.0152, "step": 4787 }, { "epoch": 0.5094701000212811, "grad_norm": 16.609375, "learning_rate": 9.847640704945992e-06, "loss": 1.133, "step": 4788 }, { "epoch": 0.5095765056394977, "grad_norm": 14.875, "learning_rate": 9.847605173393974e-06, "loss": 1.212, "step": 4789 }, { "epoch": 0.5096829112577144, "grad_norm": 12.53125, "learning_rate": 9.847569641841956e-06, "loss": 1.4895, "step": 4790 }, { "epoch": 0.5097893168759311, "grad_norm": 15.578125, "learning_rate": 9.847534110289937e-06, "loss": 1.2469, "step": 4791 }, { "epoch": 0.5098957224941477, "grad_norm": 15.6796875, "learning_rate": 9.847498578737919e-06, "loss": 1.6987, "step": 4792 }, { "epoch": 0.5100021281123643, "grad_norm": 13.8984375, "learning_rate": 9.847463047185902e-06, "loss": 1.4103, "step": 4793 }, { "epoch": 0.510108533730581, "grad_norm": 12.734375, "learning_rate": 9.847427515633884e-06, "loss": 1.1375, "step": 4794 }, { "epoch": 0.5102149393487976, "grad_norm": 13.765625, "learning_rate": 9.847391984081865e-06, "loss": 1.3037, "step": 4795 }, { "epoch": 0.5103213449670143, "grad_norm": 13.171875, "learning_rate": 9.847356452529847e-06, "loss": 1.0381, "step": 4796 }, { "epoch": 0.5104277505852309, "grad_norm": 15.65625, "learning_rate": 9.847320920977829e-06, "loss": 1.2992, "step": 4797 }, { "epoch": 0.5105341562034476, "grad_norm": 17.0625, "learning_rate": 9.847285389425812e-06, "loss": 1.1603, "step": 4798 }, { "epoch": 0.5106405618216642, "grad_norm": 12.59375, "learning_rate": 9.847249857873794e-06, "loss": 1.1008, "step": 4799 }, { "epoch": 0.5107469674398808, "grad_norm": 12.9140625, "learning_rate": 9.847214326321775e-06, "loss": 1.4189, "step": 4800 }, { "epoch": 0.5108533730580974, "grad_norm": 15.453125, "learning_rate": 9.847178794769757e-06, "loss": 1.0994, "step": 4801 }, { "epoch": 0.5109597786763141, "grad_norm": 13.8984375, "learning_rate": 9.847143263217738e-06, "loss": 1.5895, "step": 4802 }, { "epoch": 0.5110661842945308, "grad_norm": 17.3125, "learning_rate": 9.84710773166572e-06, "loss": 1.441, "step": 4803 }, { "epoch": 0.5111725899127474, "grad_norm": 10.6484375, "learning_rate": 9.847072200113702e-06, "loss": 1.0607, "step": 4804 }, { "epoch": 0.511278995530964, "grad_norm": 17.421875, "learning_rate": 9.847036668561683e-06, "loss": 1.3478, "step": 4805 }, { "epoch": 0.5113854011491806, "grad_norm": 16.25, "learning_rate": 9.847001137009665e-06, "loss": 1.2283, "step": 4806 }, { "epoch": 0.5114918067673974, "grad_norm": 18.765625, "learning_rate": 9.846965605457646e-06, "loss": 1.4155, "step": 4807 }, { "epoch": 0.511598212385614, "grad_norm": 14.3515625, "learning_rate": 9.84693007390563e-06, "loss": 1.4466, "step": 4808 }, { "epoch": 0.5117046180038306, "grad_norm": 13.859375, "learning_rate": 9.846894542353611e-06, "loss": 1.1458, "step": 4809 }, { "epoch": 0.5118110236220472, "grad_norm": 13.1875, "learning_rate": 9.846859010801593e-06, "loss": 1.2264, "step": 4810 }, { "epoch": 0.5119174292402638, "grad_norm": 12.3515625, "learning_rate": 9.846823479249574e-06, "loss": 1.2329, "step": 4811 }, { "epoch": 0.5120238348584806, "grad_norm": 12.390625, "learning_rate": 9.846787947697556e-06, "loss": 0.9582, "step": 4812 }, { "epoch": 0.5121302404766972, "grad_norm": 13.1875, "learning_rate": 9.84675241614554e-06, "loss": 1.2086, "step": 4813 }, { "epoch": 0.5122366460949138, "grad_norm": 14.6484375, "learning_rate": 9.84671688459352e-06, "loss": 1.3313, "step": 4814 }, { "epoch": 0.5123430517131304, "grad_norm": 18.078125, "learning_rate": 9.846681353041501e-06, "loss": 1.2948, "step": 4815 }, { "epoch": 0.5124494573313471, "grad_norm": 14.3671875, "learning_rate": 9.846645821489483e-06, "loss": 1.0058, "step": 4816 }, { "epoch": 0.5125558629495638, "grad_norm": 13.2890625, "learning_rate": 9.846610289937464e-06, "loss": 1.0088, "step": 4817 }, { "epoch": 0.5126622685677804, "grad_norm": 14.9609375, "learning_rate": 9.846574758385447e-06, "loss": 1.4042, "step": 4818 }, { "epoch": 0.512768674185997, "grad_norm": 12.796875, "learning_rate": 9.846539226833429e-06, "loss": 1.3343, "step": 4819 }, { "epoch": 0.5128750798042137, "grad_norm": 14.125, "learning_rate": 9.84650369528141e-06, "loss": 1.2972, "step": 4820 }, { "epoch": 0.5129814854224303, "grad_norm": 14.6796875, "learning_rate": 9.846468163729392e-06, "loss": 1.1014, "step": 4821 }, { "epoch": 0.5130878910406469, "grad_norm": 15.859375, "learning_rate": 9.846432632177374e-06, "loss": 1.0099, "step": 4822 }, { "epoch": 0.5131942966588636, "grad_norm": 12.4921875, "learning_rate": 9.846397100625355e-06, "loss": 1.2205, "step": 4823 }, { "epoch": 0.5133007022770802, "grad_norm": 12.8984375, "learning_rate": 9.846361569073339e-06, "loss": 1.252, "step": 4824 }, { "epoch": 0.5134071078952969, "grad_norm": 12.234375, "learning_rate": 9.84632603752132e-06, "loss": 1.26, "step": 4825 }, { "epoch": 0.5135135135135135, "grad_norm": 13.8359375, "learning_rate": 9.846290505969302e-06, "loss": 1.5644, "step": 4826 }, { "epoch": 0.5136199191317301, "grad_norm": 11.609375, "learning_rate": 9.846254974417284e-06, "loss": 1.1548, "step": 4827 }, { "epoch": 0.5137263247499468, "grad_norm": 10.8515625, "learning_rate": 9.846219442865265e-06, "loss": 1.4953, "step": 4828 }, { "epoch": 0.5138327303681635, "grad_norm": 13.796875, "learning_rate": 9.846183911313247e-06, "loss": 1.3235, "step": 4829 }, { "epoch": 0.5139391359863801, "grad_norm": 12.109375, "learning_rate": 9.846148379761228e-06, "loss": 1.0762, "step": 4830 }, { "epoch": 0.5140455416045967, "grad_norm": 16.390625, "learning_rate": 9.84611284820921e-06, "loss": 1.27, "step": 4831 }, { "epoch": 0.5141519472228133, "grad_norm": 10.3125, "learning_rate": 9.846077316657192e-06, "loss": 1.1369, "step": 4832 }, { "epoch": 0.51425835284103, "grad_norm": 14.3828125, "learning_rate": 9.846041785105173e-06, "loss": 1.3427, "step": 4833 }, { "epoch": 0.5143647584592467, "grad_norm": 17.90625, "learning_rate": 9.846006253553157e-06, "loss": 1.3167, "step": 4834 }, { "epoch": 0.5144711640774633, "grad_norm": 11.0234375, "learning_rate": 9.845970722001138e-06, "loss": 1.1219, "step": 4835 }, { "epoch": 0.5145775696956799, "grad_norm": 13.4609375, "learning_rate": 9.84593519044912e-06, "loss": 1.243, "step": 4836 }, { "epoch": 0.5146839753138965, "grad_norm": 13.34375, "learning_rate": 9.845899658897101e-06, "loss": 1.0457, "step": 4837 }, { "epoch": 0.5147903809321133, "grad_norm": 11.4609375, "learning_rate": 9.845864127345083e-06, "loss": 1.2417, "step": 4838 }, { "epoch": 0.5148967865503299, "grad_norm": 11.59375, "learning_rate": 9.845828595793066e-06, "loss": 1.2239, "step": 4839 }, { "epoch": 0.5150031921685465, "grad_norm": 12.9453125, "learning_rate": 9.845793064241048e-06, "loss": 1.4738, "step": 4840 }, { "epoch": 0.5151095977867631, "grad_norm": 9.265625, "learning_rate": 9.84575753268903e-06, "loss": 1.2209, "step": 4841 }, { "epoch": 0.5152160034049797, "grad_norm": 10.9609375, "learning_rate": 9.845722001137011e-06, "loss": 0.9911, "step": 4842 }, { "epoch": 0.5153224090231964, "grad_norm": 12.3515625, "learning_rate": 9.845686469584993e-06, "loss": 1.3272, "step": 4843 }, { "epoch": 0.5154288146414131, "grad_norm": 12.2578125, "learning_rate": 9.845650938032974e-06, "loss": 1.2158, "step": 4844 }, { "epoch": 0.5155352202596297, "grad_norm": 12.5, "learning_rate": 9.845615406480956e-06, "loss": 1.1093, "step": 4845 }, { "epoch": 0.5156416258778463, "grad_norm": 15.3671875, "learning_rate": 9.845579874928938e-06, "loss": 1.2132, "step": 4846 }, { "epoch": 0.515748031496063, "grad_norm": 12.6484375, "learning_rate": 9.845544343376919e-06, "loss": 1.2498, "step": 4847 }, { "epoch": 0.5158544371142796, "grad_norm": 12.390625, "learning_rate": 9.8455088118249e-06, "loss": 1.2622, "step": 4848 }, { "epoch": 0.5159608427324963, "grad_norm": 13.5625, "learning_rate": 9.845473280272884e-06, "loss": 1.0052, "step": 4849 }, { "epoch": 0.5160672483507129, "grad_norm": 11.5078125, "learning_rate": 9.845437748720866e-06, "loss": 0.913, "step": 4850 }, { "epoch": 0.5161736539689296, "grad_norm": 11.8046875, "learning_rate": 9.845402217168847e-06, "loss": 1.1741, "step": 4851 }, { "epoch": 0.5162800595871462, "grad_norm": 13.375, "learning_rate": 9.845366685616829e-06, "loss": 1.0849, "step": 4852 }, { "epoch": 0.5163864652053628, "grad_norm": 14.2734375, "learning_rate": 9.84533115406481e-06, "loss": 1.4028, "step": 4853 }, { "epoch": 0.5164928708235795, "grad_norm": 13.6171875, "learning_rate": 9.845295622512792e-06, "loss": 1.1843, "step": 4854 }, { "epoch": 0.5165992764417962, "grad_norm": 32.59375, "learning_rate": 9.845260090960774e-06, "loss": 1.7843, "step": 4855 }, { "epoch": 0.5167056820600128, "grad_norm": 12.9765625, "learning_rate": 9.845224559408755e-06, "loss": 1.2326, "step": 4856 }, { "epoch": 0.5168120876782294, "grad_norm": 14.0546875, "learning_rate": 9.845189027856737e-06, "loss": 1.1403, "step": 4857 }, { "epoch": 0.516918493296446, "grad_norm": 13.34375, "learning_rate": 9.845153496304718e-06, "loss": 1.2486, "step": 4858 }, { "epoch": 0.5170248989146627, "grad_norm": 11.6328125, "learning_rate": 9.8451179647527e-06, "loss": 1.0877, "step": 4859 }, { "epoch": 0.5171313045328794, "grad_norm": 19.203125, "learning_rate": 9.845082433200683e-06, "loss": 1.4176, "step": 4860 }, { "epoch": 0.517237710151096, "grad_norm": 10.8984375, "learning_rate": 9.845046901648665e-06, "loss": 1.1079, "step": 4861 }, { "epoch": 0.5173441157693126, "grad_norm": 10.5390625, "learning_rate": 9.845011370096647e-06, "loss": 1.4534, "step": 4862 }, { "epoch": 0.5174505213875292, "grad_norm": 12.7109375, "learning_rate": 9.844975838544628e-06, "loss": 1.15, "step": 4863 }, { "epoch": 0.5175569270057458, "grad_norm": 10.9609375, "learning_rate": 9.84494030699261e-06, "loss": 1.4371, "step": 4864 }, { "epoch": 0.5176633326239626, "grad_norm": 16.59375, "learning_rate": 9.844904775440593e-06, "loss": 1.0872, "step": 4865 }, { "epoch": 0.5177697382421792, "grad_norm": 10.6640625, "learning_rate": 9.844869243888575e-06, "loss": 1.3163, "step": 4866 }, { "epoch": 0.5178761438603958, "grad_norm": 14.109375, "learning_rate": 9.844833712336556e-06, "loss": 1.2221, "step": 4867 }, { "epoch": 0.5179825494786124, "grad_norm": 18.125, "learning_rate": 9.844798180784538e-06, "loss": 1.1165, "step": 4868 }, { "epoch": 0.5180889550968291, "grad_norm": 21.546875, "learning_rate": 9.84476264923252e-06, "loss": 1.1907, "step": 4869 }, { "epoch": 0.5181953607150458, "grad_norm": 15.296875, "learning_rate": 9.844727117680501e-06, "loss": 1.2873, "step": 4870 }, { "epoch": 0.5183017663332624, "grad_norm": 13.8828125, "learning_rate": 9.844691586128483e-06, "loss": 1.223, "step": 4871 }, { "epoch": 0.518408171951479, "grad_norm": 16.109375, "learning_rate": 9.844656054576464e-06, "loss": 1.5779, "step": 4872 }, { "epoch": 0.5185145775696957, "grad_norm": 13.0859375, "learning_rate": 9.844620523024446e-06, "loss": 1.3847, "step": 4873 }, { "epoch": 0.5186209831879123, "grad_norm": 13.2890625, "learning_rate": 9.844584991472428e-06, "loss": 1.0063, "step": 4874 }, { "epoch": 0.518727388806129, "grad_norm": 15.7734375, "learning_rate": 9.844549459920411e-06, "loss": 1.5173, "step": 4875 }, { "epoch": 0.5188337944243456, "grad_norm": 13.1484375, "learning_rate": 9.844513928368392e-06, "loss": 1.3149, "step": 4876 }, { "epoch": 0.5189402000425622, "grad_norm": 17.171875, "learning_rate": 9.844478396816374e-06, "loss": 1.163, "step": 4877 }, { "epoch": 0.5190466056607789, "grad_norm": 10.328125, "learning_rate": 9.844442865264356e-06, "loss": 1.3455, "step": 4878 }, { "epoch": 0.5191530112789955, "grad_norm": 18.8125, "learning_rate": 9.844407333712337e-06, "loss": 1.4669, "step": 4879 }, { "epoch": 0.5192594168972122, "grad_norm": 13.03125, "learning_rate": 9.844371802160319e-06, "loss": 1.2009, "step": 4880 }, { "epoch": 0.5193658225154288, "grad_norm": 15.734375, "learning_rate": 9.8443362706083e-06, "loss": 1.4692, "step": 4881 }, { "epoch": 0.5194722281336455, "grad_norm": 14.421875, "learning_rate": 9.844300739056282e-06, "loss": 1.2323, "step": 4882 }, { "epoch": 0.5195786337518621, "grad_norm": 12.6875, "learning_rate": 9.844265207504264e-06, "loss": 1.5747, "step": 4883 }, { "epoch": 0.5196850393700787, "grad_norm": 12.4453125, "learning_rate": 9.844229675952245e-06, "loss": 1.0281, "step": 4884 }, { "epoch": 0.5197914449882954, "grad_norm": 14.9296875, "learning_rate": 9.844194144400229e-06, "loss": 1.1739, "step": 4885 }, { "epoch": 0.5198978506065121, "grad_norm": 16.390625, "learning_rate": 9.84415861284821e-06, "loss": 1.2607, "step": 4886 }, { "epoch": 0.5200042562247287, "grad_norm": 15.53125, "learning_rate": 9.844123081296192e-06, "loss": 1.2503, "step": 4887 }, { "epoch": 0.5201106618429453, "grad_norm": 14.46875, "learning_rate": 9.844087549744173e-06, "loss": 1.1092, "step": 4888 }, { "epoch": 0.5202170674611619, "grad_norm": 11.3046875, "learning_rate": 9.844052018192155e-06, "loss": 0.9902, "step": 4889 }, { "epoch": 0.5203234730793785, "grad_norm": 13.984375, "learning_rate": 9.844016486640138e-06, "loss": 1.5667, "step": 4890 }, { "epoch": 0.5204298786975953, "grad_norm": 14.3984375, "learning_rate": 9.84398095508812e-06, "loss": 1.2815, "step": 4891 }, { "epoch": 0.5205362843158119, "grad_norm": 10.6640625, "learning_rate": 9.843945423536102e-06, "loss": 1.0153, "step": 4892 }, { "epoch": 0.5206426899340285, "grad_norm": 12.296875, "learning_rate": 9.843909891984083e-06, "loss": 1.1266, "step": 4893 }, { "epoch": 0.5207490955522451, "grad_norm": 11.6640625, "learning_rate": 9.843874360432065e-06, "loss": 1.1815, "step": 4894 }, { "epoch": 0.5208555011704618, "grad_norm": 14.484375, "learning_rate": 9.843838828880046e-06, "loss": 1.0405, "step": 4895 }, { "epoch": 0.5209619067886785, "grad_norm": 17.203125, "learning_rate": 9.843803297328028e-06, "loss": 1.4654, "step": 4896 }, { "epoch": 0.5210683124068951, "grad_norm": 18.203125, "learning_rate": 9.84376776577601e-06, "loss": 1.3243, "step": 4897 }, { "epoch": 0.5211747180251117, "grad_norm": 12.3984375, "learning_rate": 9.843732234223991e-06, "loss": 1.1898, "step": 4898 }, { "epoch": 0.5212811236433283, "grad_norm": 16.015625, "learning_rate": 9.843696702671973e-06, "loss": 1.5232, "step": 4899 }, { "epoch": 0.521387529261545, "grad_norm": 13.6953125, "learning_rate": 9.843661171119954e-06, "loss": 1.391, "step": 4900 }, { "epoch": 0.5214939348797617, "grad_norm": 13.734375, "learning_rate": 9.843625639567938e-06, "loss": 1.0587, "step": 4901 }, { "epoch": 0.5216003404979783, "grad_norm": 12.265625, "learning_rate": 9.84359010801592e-06, "loss": 1.1328, "step": 4902 }, { "epoch": 0.5217067461161949, "grad_norm": 15.0, "learning_rate": 9.843554576463901e-06, "loss": 1.2569, "step": 4903 }, { "epoch": 0.5218131517344116, "grad_norm": 11.1171875, "learning_rate": 9.843519044911883e-06, "loss": 1.2316, "step": 4904 }, { "epoch": 0.5219195573526282, "grad_norm": 13.078125, "learning_rate": 9.843483513359864e-06, "loss": 1.5056, "step": 4905 }, { "epoch": 0.5220259629708449, "grad_norm": 11.625, "learning_rate": 9.843447981807846e-06, "loss": 1.2323, "step": 4906 }, { "epoch": 0.5221323685890615, "grad_norm": 11.5390625, "learning_rate": 9.843412450255827e-06, "loss": 1.1636, "step": 4907 }, { "epoch": 0.5222387742072782, "grad_norm": 15.25, "learning_rate": 9.843376918703809e-06, "loss": 1.0444, "step": 4908 }, { "epoch": 0.5223451798254948, "grad_norm": 15.3125, "learning_rate": 9.84334138715179e-06, "loss": 1.2194, "step": 4909 }, { "epoch": 0.5224515854437114, "grad_norm": 12.328125, "learning_rate": 9.843305855599772e-06, "loss": 1.3534, "step": 4910 }, { "epoch": 0.522557991061928, "grad_norm": 11.8984375, "learning_rate": 9.843270324047755e-06, "loss": 1.4836, "step": 4911 }, { "epoch": 0.5226643966801447, "grad_norm": 11.5859375, "learning_rate": 9.843234792495737e-06, "loss": 1.139, "step": 4912 }, { "epoch": 0.5227708022983614, "grad_norm": 15.5859375, "learning_rate": 9.843199260943719e-06, "loss": 1.248, "step": 4913 }, { "epoch": 0.522877207916578, "grad_norm": 15.1015625, "learning_rate": 9.8431637293917e-06, "loss": 1.2746, "step": 4914 }, { "epoch": 0.5229836135347946, "grad_norm": 11.7890625, "learning_rate": 9.843128197839682e-06, "loss": 1.483, "step": 4915 }, { "epoch": 0.5230900191530112, "grad_norm": 11.296875, "learning_rate": 9.843092666287665e-06, "loss": 1.1326, "step": 4916 }, { "epoch": 0.523196424771228, "grad_norm": 11.921875, "learning_rate": 9.843057134735647e-06, "loss": 1.1418, "step": 4917 }, { "epoch": 0.5233028303894446, "grad_norm": 13.84375, "learning_rate": 9.843021603183628e-06, "loss": 1.2066, "step": 4918 }, { "epoch": 0.5234092360076612, "grad_norm": 13.828125, "learning_rate": 9.84298607163161e-06, "loss": 1.182, "step": 4919 }, { "epoch": 0.5235156416258778, "grad_norm": 10.9765625, "learning_rate": 9.842950540079592e-06, "loss": 1.2445, "step": 4920 }, { "epoch": 0.5236220472440944, "grad_norm": 13.4921875, "learning_rate": 9.842915008527573e-06, "loss": 1.2746, "step": 4921 }, { "epoch": 0.5237284528623112, "grad_norm": 15.328125, "learning_rate": 9.842879476975555e-06, "loss": 1.2981, "step": 4922 }, { "epoch": 0.5238348584805278, "grad_norm": 15.421875, "learning_rate": 9.842843945423536e-06, "loss": 1.2196, "step": 4923 }, { "epoch": 0.5239412640987444, "grad_norm": 13.59375, "learning_rate": 9.842808413871518e-06, "loss": 1.3968, "step": 4924 }, { "epoch": 0.524047669716961, "grad_norm": 14.8203125, "learning_rate": 9.8427728823195e-06, "loss": 1.2257, "step": 4925 }, { "epoch": 0.5241540753351777, "grad_norm": 10.671875, "learning_rate": 9.842737350767483e-06, "loss": 1.2014, "step": 4926 }, { "epoch": 0.5242604809533944, "grad_norm": 13.3671875, "learning_rate": 9.842701819215465e-06, "loss": 1.5283, "step": 4927 }, { "epoch": 0.524366886571611, "grad_norm": 10.6328125, "learning_rate": 9.842666287663446e-06, "loss": 0.9797, "step": 4928 }, { "epoch": 0.5244732921898276, "grad_norm": 10.78125, "learning_rate": 9.842630756111428e-06, "loss": 1.4887, "step": 4929 }, { "epoch": 0.5245796978080443, "grad_norm": 12.9921875, "learning_rate": 9.84259522455941e-06, "loss": 1.3405, "step": 4930 }, { "epoch": 0.5246861034262609, "grad_norm": 12.140625, "learning_rate": 9.842559693007393e-06, "loss": 1.1782, "step": 4931 }, { "epoch": 0.5247925090444775, "grad_norm": 12.484375, "learning_rate": 9.842524161455374e-06, "loss": 1.192, "step": 4932 }, { "epoch": 0.5248989146626942, "grad_norm": 11.7109375, "learning_rate": 9.842488629903356e-06, "loss": 1.1499, "step": 4933 }, { "epoch": 0.5250053202809108, "grad_norm": 13.3125, "learning_rate": 9.842453098351336e-06, "loss": 1.1966, "step": 4934 }, { "epoch": 0.5251117258991275, "grad_norm": 19.34375, "learning_rate": 9.842417566799317e-06, "loss": 1.3507, "step": 4935 }, { "epoch": 0.5252181315173441, "grad_norm": 11.171875, "learning_rate": 9.8423820352473e-06, "loss": 1.1316, "step": 4936 }, { "epoch": 0.5253245371355607, "grad_norm": 9.5625, "learning_rate": 9.842346503695282e-06, "loss": 1.1793, "step": 4937 }, { "epoch": 0.5254309427537774, "grad_norm": 13.4765625, "learning_rate": 9.842310972143264e-06, "loss": 1.2638, "step": 4938 }, { "epoch": 0.5255373483719941, "grad_norm": 11.28125, "learning_rate": 9.842275440591246e-06, "loss": 1.3425, "step": 4939 }, { "epoch": 0.5256437539902107, "grad_norm": 15.1640625, "learning_rate": 9.842239909039227e-06, "loss": 1.1287, "step": 4940 }, { "epoch": 0.5257501596084273, "grad_norm": 32.59375, "learning_rate": 9.842204377487209e-06, "loss": 0.9555, "step": 4941 }, { "epoch": 0.5258565652266439, "grad_norm": 15.21875, "learning_rate": 9.842168845935192e-06, "loss": 1.2262, "step": 4942 }, { "epoch": 0.5259629708448607, "grad_norm": 21.1875, "learning_rate": 9.842133314383174e-06, "loss": 1.4768, "step": 4943 }, { "epoch": 0.5260693764630773, "grad_norm": 15.796875, "learning_rate": 9.842097782831155e-06, "loss": 1.5534, "step": 4944 }, { "epoch": 0.5261757820812939, "grad_norm": 16.125, "learning_rate": 9.842062251279137e-06, "loss": 1.4017, "step": 4945 }, { "epoch": 0.5262821876995105, "grad_norm": 11.90625, "learning_rate": 9.842026719727118e-06, "loss": 1.1481, "step": 4946 }, { "epoch": 0.5263885933177271, "grad_norm": 17.296875, "learning_rate": 9.8419911881751e-06, "loss": 1.3041, "step": 4947 }, { "epoch": 0.5264949989359439, "grad_norm": 10.5703125, "learning_rate": 9.841955656623082e-06, "loss": 1.2458, "step": 4948 }, { "epoch": 0.5266014045541605, "grad_norm": 11.28125, "learning_rate": 9.841920125071063e-06, "loss": 1.2441, "step": 4949 }, { "epoch": 0.5267078101723771, "grad_norm": 16.15625, "learning_rate": 9.841884593519045e-06, "loss": 1.4819, "step": 4950 }, { "epoch": 0.5268142157905937, "grad_norm": 16.46875, "learning_rate": 9.841849061967027e-06, "loss": 0.9343, "step": 4951 }, { "epoch": 0.5269206214088104, "grad_norm": 13.9375, "learning_rate": 9.84181353041501e-06, "loss": 0.9691, "step": 4952 }, { "epoch": 0.527027027027027, "grad_norm": 14.953125, "learning_rate": 9.841777998862991e-06, "loss": 1.1913, "step": 4953 }, { "epoch": 0.5271334326452437, "grad_norm": 24.1875, "learning_rate": 9.841742467310973e-06, "loss": 1.3653, "step": 4954 }, { "epoch": 0.5272398382634603, "grad_norm": 15.5078125, "learning_rate": 9.841706935758955e-06, "loss": 1.3499, "step": 4955 }, { "epoch": 0.527346243881677, "grad_norm": 15.6953125, "learning_rate": 9.841671404206936e-06, "loss": 1.6572, "step": 4956 }, { "epoch": 0.5274526494998936, "grad_norm": 13.8515625, "learning_rate": 9.84163587265492e-06, "loss": 1.3408, "step": 4957 }, { "epoch": 0.5275590551181102, "grad_norm": 12.265625, "learning_rate": 9.841600341102901e-06, "loss": 1.0677, "step": 4958 }, { "epoch": 0.5276654607363269, "grad_norm": 19.328125, "learning_rate": 9.841564809550883e-06, "loss": 1.2719, "step": 4959 }, { "epoch": 0.5277718663545435, "grad_norm": 11.625, "learning_rate": 9.841529277998864e-06, "loss": 1.1596, "step": 4960 }, { "epoch": 0.5278782719727602, "grad_norm": 16.28125, "learning_rate": 9.841493746446846e-06, "loss": 1.3486, "step": 4961 }, { "epoch": 0.5279846775909768, "grad_norm": 14.2578125, "learning_rate": 9.841458214894828e-06, "loss": 1.0275, "step": 4962 }, { "epoch": 0.5280910832091934, "grad_norm": 11.0703125, "learning_rate": 9.84142268334281e-06, "loss": 1.2383, "step": 4963 }, { "epoch": 0.5281974888274101, "grad_norm": 12.515625, "learning_rate": 9.84138715179079e-06, "loss": 0.9971, "step": 4964 }, { "epoch": 0.5283038944456268, "grad_norm": 18.703125, "learning_rate": 9.841351620238772e-06, "loss": 0.8615, "step": 4965 }, { "epoch": 0.5284103000638434, "grad_norm": 14.6171875, "learning_rate": 9.841316088686754e-06, "loss": 1.139, "step": 4966 }, { "epoch": 0.52851670568206, "grad_norm": 12.546875, "learning_rate": 9.841280557134737e-06, "loss": 1.2164, "step": 4967 }, { "epoch": 0.5286231113002766, "grad_norm": 13.8828125, "learning_rate": 9.841245025582719e-06, "loss": 1.5153, "step": 4968 }, { "epoch": 0.5287295169184933, "grad_norm": 16.25, "learning_rate": 9.8412094940307e-06, "loss": 1.5354, "step": 4969 }, { "epoch": 0.52883592253671, "grad_norm": 12.796875, "learning_rate": 9.841173962478682e-06, "loss": 1.1266, "step": 4970 }, { "epoch": 0.5289423281549266, "grad_norm": 20.453125, "learning_rate": 9.841138430926664e-06, "loss": 1.2344, "step": 4971 }, { "epoch": 0.5290487337731432, "grad_norm": 11.296875, "learning_rate": 9.841102899374645e-06, "loss": 1.1124, "step": 4972 }, { "epoch": 0.5291551393913598, "grad_norm": 13.5859375, "learning_rate": 9.841067367822627e-06, "loss": 1.0835, "step": 4973 }, { "epoch": 0.5292615450095765, "grad_norm": 14.3828125, "learning_rate": 9.841031836270609e-06, "loss": 1.0119, "step": 4974 }, { "epoch": 0.5293679506277932, "grad_norm": 22.0, "learning_rate": 9.84099630471859e-06, "loss": 1.6266, "step": 4975 }, { "epoch": 0.5294743562460098, "grad_norm": 12.53125, "learning_rate": 9.840960773166572e-06, "loss": 0.9797, "step": 4976 }, { "epoch": 0.5295807618642264, "grad_norm": 14.90625, "learning_rate": 9.840925241614553e-06, "loss": 1.5307, "step": 4977 }, { "epoch": 0.529687167482443, "grad_norm": 15.25, "learning_rate": 9.840889710062537e-06, "loss": 1.0807, "step": 4978 }, { "epoch": 0.5297935731006597, "grad_norm": 13.5859375, "learning_rate": 9.840854178510518e-06, "loss": 0.8904, "step": 4979 }, { "epoch": 0.5298999787188764, "grad_norm": 17.25, "learning_rate": 9.8408186469585e-06, "loss": 0.9606, "step": 4980 }, { "epoch": 0.530006384337093, "grad_norm": 11.9765625, "learning_rate": 9.840783115406482e-06, "loss": 1.0464, "step": 4981 }, { "epoch": 0.5301127899553096, "grad_norm": 12.1953125, "learning_rate": 9.840747583854463e-06, "loss": 1.2297, "step": 4982 }, { "epoch": 0.5302191955735263, "grad_norm": 47.40625, "learning_rate": 9.840712052302446e-06, "loss": 1.1105, "step": 4983 }, { "epoch": 0.5303256011917429, "grad_norm": 15.390625, "learning_rate": 9.840676520750428e-06, "loss": 1.524, "step": 4984 }, { "epoch": 0.5304320068099596, "grad_norm": 13.4296875, "learning_rate": 9.84064098919841e-06, "loss": 1.3159, "step": 4985 }, { "epoch": 0.5305384124281762, "grad_norm": 15.5078125, "learning_rate": 9.840605457646391e-06, "loss": 1.3169, "step": 4986 }, { "epoch": 0.5306448180463929, "grad_norm": 13.9921875, "learning_rate": 9.840569926094373e-06, "loss": 1.2524, "step": 4987 }, { "epoch": 0.5307512236646095, "grad_norm": 23.109375, "learning_rate": 9.840534394542354e-06, "loss": 1.3067, "step": 4988 }, { "epoch": 0.5308576292828261, "grad_norm": 14.6640625, "learning_rate": 9.840498862990336e-06, "loss": 1.4295, "step": 4989 }, { "epoch": 0.5309640349010428, "grad_norm": 13.7578125, "learning_rate": 9.840463331438318e-06, "loss": 1.3539, "step": 4990 }, { "epoch": 0.5310704405192594, "grad_norm": 10.640625, "learning_rate": 9.8404277998863e-06, "loss": 1.2551, "step": 4991 }, { "epoch": 0.5311768461374761, "grad_norm": 14.21875, "learning_rate": 9.840392268334281e-06, "loss": 1.1605, "step": 4992 }, { "epoch": 0.5312832517556927, "grad_norm": 13.921875, "learning_rate": 9.840356736782264e-06, "loss": 1.1867, "step": 4993 }, { "epoch": 0.5313896573739093, "grad_norm": 11.25, "learning_rate": 9.840321205230246e-06, "loss": 1.2755, "step": 4994 }, { "epoch": 0.531496062992126, "grad_norm": 11.65625, "learning_rate": 9.840285673678227e-06, "loss": 1.3095, "step": 4995 }, { "epoch": 0.5316024686103427, "grad_norm": 11.8046875, "learning_rate": 9.840250142126209e-06, "loss": 1.1706, "step": 4996 }, { "epoch": 0.5317088742285593, "grad_norm": 12.140625, "learning_rate": 9.84021461057419e-06, "loss": 1.0811, "step": 4997 }, { "epoch": 0.5318152798467759, "grad_norm": 12.8984375, "learning_rate": 9.840179079022172e-06, "loss": 1.1942, "step": 4998 }, { "epoch": 0.5319216854649925, "grad_norm": 9.46875, "learning_rate": 9.840143547470154e-06, "loss": 1.1242, "step": 4999 }, { "epoch": 0.5320280910832091, "grad_norm": 14.765625, "learning_rate": 9.840108015918135e-06, "loss": 0.8381, "step": 5000 }, { "epoch": 0.5321344967014259, "grad_norm": 10.6328125, "learning_rate": 9.840072484366117e-06, "loss": 1.2195, "step": 5001 }, { "epoch": 0.5322409023196425, "grad_norm": 11.3359375, "learning_rate": 9.840036952814099e-06, "loss": 0.9474, "step": 5002 }, { "epoch": 0.5323473079378591, "grad_norm": 14.515625, "learning_rate": 9.840001421262082e-06, "loss": 1.1693, "step": 5003 }, { "epoch": 0.5324537135560757, "grad_norm": 16.9375, "learning_rate": 9.839965889710064e-06, "loss": 1.497, "step": 5004 }, { "epoch": 0.5325601191742924, "grad_norm": 14.203125, "learning_rate": 9.839930358158045e-06, "loss": 1.5032, "step": 5005 }, { "epoch": 0.5326665247925091, "grad_norm": 14.8125, "learning_rate": 9.839894826606027e-06, "loss": 1.5379, "step": 5006 }, { "epoch": 0.5327729304107257, "grad_norm": 11.6484375, "learning_rate": 9.839859295054008e-06, "loss": 1.233, "step": 5007 }, { "epoch": 0.5328793360289423, "grad_norm": 12.703125, "learning_rate": 9.839823763501992e-06, "loss": 1.055, "step": 5008 }, { "epoch": 0.532985741647159, "grad_norm": 15.421875, "learning_rate": 9.839788231949973e-06, "loss": 1.1848, "step": 5009 }, { "epoch": 0.5330921472653756, "grad_norm": 10.96875, "learning_rate": 9.839752700397955e-06, "loss": 1.1289, "step": 5010 }, { "epoch": 0.5331985528835923, "grad_norm": 13.21875, "learning_rate": 9.839717168845936e-06, "loss": 1.3163, "step": 5011 }, { "epoch": 0.5333049585018089, "grad_norm": 12.5625, "learning_rate": 9.839681637293918e-06, "loss": 1.1796, "step": 5012 }, { "epoch": 0.5334113641200255, "grad_norm": 12.859375, "learning_rate": 9.8396461057419e-06, "loss": 1.3151, "step": 5013 }, { "epoch": 0.5335177697382422, "grad_norm": 12.671875, "learning_rate": 9.839610574189881e-06, "loss": 1.1895, "step": 5014 }, { "epoch": 0.5336241753564588, "grad_norm": 10.28125, "learning_rate": 9.839575042637863e-06, "loss": 1.3794, "step": 5015 }, { "epoch": 0.5337305809746755, "grad_norm": 16.53125, "learning_rate": 9.839539511085845e-06, "loss": 1.3147, "step": 5016 }, { "epoch": 0.5338369865928921, "grad_norm": 14.890625, "learning_rate": 9.839503979533826e-06, "loss": 1.4366, "step": 5017 }, { "epoch": 0.5339433922111088, "grad_norm": 15.859375, "learning_rate": 9.839468447981808e-06, "loss": 1.4296, "step": 5018 }, { "epoch": 0.5340497978293254, "grad_norm": 11.2578125, "learning_rate": 9.839432916429791e-06, "loss": 1.0754, "step": 5019 }, { "epoch": 0.534156203447542, "grad_norm": 10.1640625, "learning_rate": 9.839397384877773e-06, "loss": 1.3681, "step": 5020 }, { "epoch": 0.5342626090657586, "grad_norm": 13.171875, "learning_rate": 9.839361853325754e-06, "loss": 1.0041, "step": 5021 }, { "epoch": 0.5343690146839754, "grad_norm": 12.6015625, "learning_rate": 9.839326321773736e-06, "loss": 1.2472, "step": 5022 }, { "epoch": 0.534475420302192, "grad_norm": 13.2890625, "learning_rate": 9.839290790221717e-06, "loss": 1.0635, "step": 5023 }, { "epoch": 0.5345818259204086, "grad_norm": 11.5703125, "learning_rate": 9.839255258669699e-06, "loss": 1.0982, "step": 5024 }, { "epoch": 0.5346882315386252, "grad_norm": 15.953125, "learning_rate": 9.83921972711768e-06, "loss": 1.2939, "step": 5025 }, { "epoch": 0.5347946371568418, "grad_norm": 14.6484375, "learning_rate": 9.839184195565662e-06, "loss": 1.2849, "step": 5026 }, { "epoch": 0.5349010427750586, "grad_norm": 16.15625, "learning_rate": 9.839148664013644e-06, "loss": 1.0732, "step": 5027 }, { "epoch": 0.5350074483932752, "grad_norm": 17.90625, "learning_rate": 9.839113132461625e-06, "loss": 1.1522, "step": 5028 }, { "epoch": 0.5351138540114918, "grad_norm": 16.84375, "learning_rate": 9.839077600909609e-06, "loss": 1.0991, "step": 5029 }, { "epoch": 0.5352202596297084, "grad_norm": 13.4921875, "learning_rate": 9.83904206935759e-06, "loss": 1.2671, "step": 5030 }, { "epoch": 0.535326665247925, "grad_norm": 14.359375, "learning_rate": 9.839006537805572e-06, "loss": 1.2868, "step": 5031 }, { "epoch": 0.5354330708661418, "grad_norm": 11.7109375, "learning_rate": 9.838971006253554e-06, "loss": 1.1788, "step": 5032 }, { "epoch": 0.5355394764843584, "grad_norm": 12.234375, "learning_rate": 9.838935474701535e-06, "loss": 1.105, "step": 5033 }, { "epoch": 0.535645882102575, "grad_norm": 14.078125, "learning_rate": 9.838899943149519e-06, "loss": 0.9932, "step": 5034 }, { "epoch": 0.5357522877207916, "grad_norm": 11.2578125, "learning_rate": 9.8388644115975e-06, "loss": 1.1387, "step": 5035 }, { "epoch": 0.5358586933390083, "grad_norm": 10.75, "learning_rate": 9.838828880045482e-06, "loss": 1.2319, "step": 5036 }, { "epoch": 0.535965098957225, "grad_norm": 14.265625, "learning_rate": 9.838793348493463e-06, "loss": 1.1508, "step": 5037 }, { "epoch": 0.5360715045754416, "grad_norm": 19.140625, "learning_rate": 9.838757816941445e-06, "loss": 1.017, "step": 5038 }, { "epoch": 0.5361779101936582, "grad_norm": 11.625, "learning_rate": 9.838722285389427e-06, "loss": 0.9067, "step": 5039 }, { "epoch": 0.5362843158118749, "grad_norm": 15.03125, "learning_rate": 9.838686753837408e-06, "loss": 1.1796, "step": 5040 }, { "epoch": 0.5363907214300915, "grad_norm": 11.3203125, "learning_rate": 9.83865122228539e-06, "loss": 1.3843, "step": 5041 }, { "epoch": 0.5364971270483081, "grad_norm": 15.0234375, "learning_rate": 9.838615690733371e-06, "loss": 1.4221, "step": 5042 }, { "epoch": 0.5366035326665248, "grad_norm": 13.9921875, "learning_rate": 9.838580159181353e-06, "loss": 1.1144, "step": 5043 }, { "epoch": 0.5367099382847415, "grad_norm": 15.9453125, "learning_rate": 9.838544627629336e-06, "loss": 1.1394, "step": 5044 }, { "epoch": 0.5368163439029581, "grad_norm": 15.5859375, "learning_rate": 9.838509096077318e-06, "loss": 0.9627, "step": 5045 }, { "epoch": 0.5369227495211747, "grad_norm": 16.671875, "learning_rate": 9.8384735645253e-06, "loss": 1.2145, "step": 5046 }, { "epoch": 0.5370291551393913, "grad_norm": 14.5546875, "learning_rate": 9.838438032973281e-06, "loss": 1.6619, "step": 5047 }, { "epoch": 0.537135560757608, "grad_norm": 19.1875, "learning_rate": 9.838402501421263e-06, "loss": 1.2074, "step": 5048 }, { "epoch": 0.5372419663758247, "grad_norm": 10.7890625, "learning_rate": 9.838366969869246e-06, "loss": 1.2319, "step": 5049 }, { "epoch": 0.5373483719940413, "grad_norm": 10.8203125, "learning_rate": 9.838331438317228e-06, "loss": 1.2004, "step": 5050 }, { "epoch": 0.5374547776122579, "grad_norm": 12.828125, "learning_rate": 9.83829590676521e-06, "loss": 1.2454, "step": 5051 }, { "epoch": 0.5375611832304745, "grad_norm": 10.796875, "learning_rate": 9.83826037521319e-06, "loss": 0.953, "step": 5052 }, { "epoch": 0.5376675888486913, "grad_norm": 12.3515625, "learning_rate": 9.838224843661172e-06, "loss": 1.2559, "step": 5053 }, { "epoch": 0.5377739944669079, "grad_norm": 11.3515625, "learning_rate": 9.838189312109152e-06, "loss": 1.0141, "step": 5054 }, { "epoch": 0.5378804000851245, "grad_norm": 13.9453125, "learning_rate": 9.838153780557136e-06, "loss": 1.1763, "step": 5055 }, { "epoch": 0.5379868057033411, "grad_norm": 11.6953125, "learning_rate": 9.838118249005117e-06, "loss": 1.066, "step": 5056 }, { "epoch": 0.5380932113215577, "grad_norm": 11.859375, "learning_rate": 9.838082717453099e-06, "loss": 1.2263, "step": 5057 }, { "epoch": 0.5381996169397745, "grad_norm": 20.265625, "learning_rate": 9.83804718590108e-06, "loss": 1.3383, "step": 5058 }, { "epoch": 0.5383060225579911, "grad_norm": 12.2890625, "learning_rate": 9.838011654349062e-06, "loss": 1.1409, "step": 5059 }, { "epoch": 0.5384124281762077, "grad_norm": 9.328125, "learning_rate": 9.837976122797045e-06, "loss": 0.9715, "step": 5060 }, { "epoch": 0.5385188337944243, "grad_norm": 15.921875, "learning_rate": 9.837940591245027e-06, "loss": 1.6978, "step": 5061 }, { "epoch": 0.538625239412641, "grad_norm": 16.796875, "learning_rate": 9.837905059693009e-06, "loss": 1.6322, "step": 5062 }, { "epoch": 0.5387316450308576, "grad_norm": 14.3828125, "learning_rate": 9.83786952814099e-06, "loss": 1.006, "step": 5063 }, { "epoch": 0.5388380506490743, "grad_norm": 9.4609375, "learning_rate": 9.837833996588972e-06, "loss": 1.3138, "step": 5064 }, { "epoch": 0.5389444562672909, "grad_norm": 10.3203125, "learning_rate": 9.837798465036953e-06, "loss": 1.1501, "step": 5065 }, { "epoch": 0.5390508618855075, "grad_norm": 12.5546875, "learning_rate": 9.837762933484935e-06, "loss": 1.1547, "step": 5066 }, { "epoch": 0.5391572675037242, "grad_norm": 14.0234375, "learning_rate": 9.837727401932917e-06, "loss": 1.1074, "step": 5067 }, { "epoch": 0.5392636731219408, "grad_norm": 15.65625, "learning_rate": 9.837691870380898e-06, "loss": 0.9804, "step": 5068 }, { "epoch": 0.5393700787401575, "grad_norm": 15.671875, "learning_rate": 9.83765633882888e-06, "loss": 1.5402, "step": 5069 }, { "epoch": 0.5394764843583741, "grad_norm": 12.8515625, "learning_rate": 9.837620807276863e-06, "loss": 1.1406, "step": 5070 }, { "epoch": 0.5395828899765908, "grad_norm": 17.46875, "learning_rate": 9.837585275724845e-06, "loss": 1.4071, "step": 5071 }, { "epoch": 0.5396892955948074, "grad_norm": 12.90625, "learning_rate": 9.837549744172826e-06, "loss": 1.3197, "step": 5072 }, { "epoch": 0.539795701213024, "grad_norm": 20.03125, "learning_rate": 9.837514212620808e-06, "loss": 1.2169, "step": 5073 }, { "epoch": 0.5399021068312407, "grad_norm": 14.625, "learning_rate": 9.83747868106879e-06, "loss": 1.4118, "step": 5074 }, { "epoch": 0.5400085124494574, "grad_norm": 15.71875, "learning_rate": 9.837443149516773e-06, "loss": 1.364, "step": 5075 }, { "epoch": 0.540114918067674, "grad_norm": 17.03125, "learning_rate": 9.837407617964754e-06, "loss": 0.9446, "step": 5076 }, { "epoch": 0.5402213236858906, "grad_norm": 14.8671875, "learning_rate": 9.837372086412736e-06, "loss": 1.4471, "step": 5077 }, { "epoch": 0.5403277293041072, "grad_norm": 11.9609375, "learning_rate": 9.837336554860718e-06, "loss": 1.0524, "step": 5078 }, { "epoch": 0.540434134922324, "grad_norm": 16.109375, "learning_rate": 9.8373010233087e-06, "loss": 1.347, "step": 5079 }, { "epoch": 0.5405405405405406, "grad_norm": 12.796875, "learning_rate": 9.837265491756681e-06, "loss": 1.1104, "step": 5080 }, { "epoch": 0.5406469461587572, "grad_norm": 13.21875, "learning_rate": 9.837229960204662e-06, "loss": 1.1266, "step": 5081 }, { "epoch": 0.5407533517769738, "grad_norm": 15.6796875, "learning_rate": 9.837194428652644e-06, "loss": 1.2523, "step": 5082 }, { "epoch": 0.5408597573951904, "grad_norm": 16.046875, "learning_rate": 9.837158897100626e-06, "loss": 0.9896, "step": 5083 }, { "epoch": 0.540966163013407, "grad_norm": 16.171875, "learning_rate": 9.837123365548607e-06, "loss": 1.2995, "step": 5084 }, { "epoch": 0.5410725686316238, "grad_norm": 12.6328125, "learning_rate": 9.83708783399659e-06, "loss": 1.0945, "step": 5085 }, { "epoch": 0.5411789742498404, "grad_norm": 12.6953125, "learning_rate": 9.837052302444572e-06, "loss": 1.2216, "step": 5086 }, { "epoch": 0.541285379868057, "grad_norm": 15.1875, "learning_rate": 9.837016770892554e-06, "loss": 1.4031, "step": 5087 }, { "epoch": 0.5413917854862736, "grad_norm": 12.1171875, "learning_rate": 9.836981239340535e-06, "loss": 1.1162, "step": 5088 }, { "epoch": 0.5414981911044903, "grad_norm": 11.7265625, "learning_rate": 9.836945707788517e-06, "loss": 1.5232, "step": 5089 }, { "epoch": 0.541604596722707, "grad_norm": 13.8359375, "learning_rate": 9.836910176236499e-06, "loss": 1.5287, "step": 5090 }, { "epoch": 0.5417110023409236, "grad_norm": 16.546875, "learning_rate": 9.83687464468448e-06, "loss": 1.3272, "step": 5091 }, { "epoch": 0.5418174079591402, "grad_norm": 8.859375, "learning_rate": 9.836839113132462e-06, "loss": 0.9197, "step": 5092 }, { "epoch": 0.5419238135773569, "grad_norm": 31.25, "learning_rate": 9.836803581580443e-06, "loss": 1.3168, "step": 5093 }, { "epoch": 0.5420302191955735, "grad_norm": 13.953125, "learning_rate": 9.836768050028425e-06, "loss": 1.3484, "step": 5094 }, { "epoch": 0.5421366248137902, "grad_norm": 11.3984375, "learning_rate": 9.836732518476407e-06, "loss": 1.0683, "step": 5095 }, { "epoch": 0.5422430304320068, "grad_norm": 11.09375, "learning_rate": 9.83669698692439e-06, "loss": 0.8201, "step": 5096 }, { "epoch": 0.5423494360502235, "grad_norm": 15.5625, "learning_rate": 9.836661455372372e-06, "loss": 1.3511, "step": 5097 }, { "epoch": 0.5424558416684401, "grad_norm": 11.671875, "learning_rate": 9.836625923820353e-06, "loss": 1.2858, "step": 5098 }, { "epoch": 0.5425622472866567, "grad_norm": 9.5703125, "learning_rate": 9.836590392268335e-06, "loss": 1.2977, "step": 5099 }, { "epoch": 0.5426686529048734, "grad_norm": 13.796875, "learning_rate": 9.836554860716316e-06, "loss": 1.2101, "step": 5100 }, { "epoch": 0.54277505852309, "grad_norm": 18.703125, "learning_rate": 9.8365193291643e-06, "loss": 0.9912, "step": 5101 }, { "epoch": 0.5428814641413067, "grad_norm": 12.6953125, "learning_rate": 9.836483797612281e-06, "loss": 1.424, "step": 5102 }, { "epoch": 0.5429878697595233, "grad_norm": 14.4921875, "learning_rate": 9.836448266060263e-06, "loss": 1.4532, "step": 5103 }, { "epoch": 0.5430942753777399, "grad_norm": 18.59375, "learning_rate": 9.836412734508245e-06, "loss": 1.6558, "step": 5104 }, { "epoch": 0.5432006809959566, "grad_norm": 11.5, "learning_rate": 9.836377202956226e-06, "loss": 1.4275, "step": 5105 }, { "epoch": 0.5433070866141733, "grad_norm": 13.1328125, "learning_rate": 9.836341671404208e-06, "loss": 1.0152, "step": 5106 }, { "epoch": 0.5434134922323899, "grad_norm": 11.5078125, "learning_rate": 9.83630613985219e-06, "loss": 1.2531, "step": 5107 }, { "epoch": 0.5435198978506065, "grad_norm": 11.296875, "learning_rate": 9.836270608300171e-06, "loss": 1.1146, "step": 5108 }, { "epoch": 0.5436263034688231, "grad_norm": 10.0625, "learning_rate": 9.836235076748153e-06, "loss": 1.1704, "step": 5109 }, { "epoch": 0.5437327090870397, "grad_norm": 14.2578125, "learning_rate": 9.836199545196134e-06, "loss": 1.3491, "step": 5110 }, { "epoch": 0.5438391147052565, "grad_norm": 11.71875, "learning_rate": 9.836164013644117e-06, "loss": 1.2763, "step": 5111 }, { "epoch": 0.5439455203234731, "grad_norm": 16.109375, "learning_rate": 9.836128482092099e-06, "loss": 1.4336, "step": 5112 }, { "epoch": 0.5440519259416897, "grad_norm": 11.359375, "learning_rate": 9.83609295054008e-06, "loss": 1.6931, "step": 5113 }, { "epoch": 0.5441583315599063, "grad_norm": 13.6875, "learning_rate": 9.836057418988062e-06, "loss": 1.1735, "step": 5114 }, { "epoch": 0.544264737178123, "grad_norm": 11.859375, "learning_rate": 9.836021887436044e-06, "loss": 1.0612, "step": 5115 }, { "epoch": 0.5443711427963397, "grad_norm": 12.640625, "learning_rate": 9.835986355884026e-06, "loss": 1.1446, "step": 5116 }, { "epoch": 0.5444775484145563, "grad_norm": 12.1171875, "learning_rate": 9.835950824332007e-06, "loss": 1.1594, "step": 5117 }, { "epoch": 0.5445839540327729, "grad_norm": 16.734375, "learning_rate": 9.835915292779989e-06, "loss": 1.1831, "step": 5118 }, { "epoch": 0.5446903596509896, "grad_norm": 18.71875, "learning_rate": 9.83587976122797e-06, "loss": 1.2881, "step": 5119 }, { "epoch": 0.5447967652692062, "grad_norm": 11.9609375, "learning_rate": 9.835844229675952e-06, "loss": 1.1486, "step": 5120 }, { "epoch": 0.5449031708874229, "grad_norm": 12.4765625, "learning_rate": 9.835808698123935e-06, "loss": 1.1604, "step": 5121 }, { "epoch": 0.5450095765056395, "grad_norm": 17.1875, "learning_rate": 9.835773166571917e-06, "loss": 1.4076, "step": 5122 }, { "epoch": 0.5451159821238561, "grad_norm": 13.1484375, "learning_rate": 9.835737635019898e-06, "loss": 0.9237, "step": 5123 }, { "epoch": 0.5452223877420728, "grad_norm": 13.0625, "learning_rate": 9.83570210346788e-06, "loss": 1.0118, "step": 5124 }, { "epoch": 0.5453287933602894, "grad_norm": 16.25, "learning_rate": 9.835666571915862e-06, "loss": 1.1149, "step": 5125 }, { "epoch": 0.5454351989785061, "grad_norm": 10.9921875, "learning_rate": 9.835631040363845e-06, "loss": 0.9073, "step": 5126 }, { "epoch": 0.5455416045967227, "grad_norm": 13.625, "learning_rate": 9.835595508811827e-06, "loss": 1.2565, "step": 5127 }, { "epoch": 0.5456480102149394, "grad_norm": 12.671875, "learning_rate": 9.835559977259808e-06, "loss": 1.2172, "step": 5128 }, { "epoch": 0.545754415833156, "grad_norm": 15.0546875, "learning_rate": 9.83552444570779e-06, "loss": 1.3983, "step": 5129 }, { "epoch": 0.5458608214513726, "grad_norm": 14.71875, "learning_rate": 9.835488914155771e-06, "loss": 1.2289, "step": 5130 }, { "epoch": 0.5459672270695892, "grad_norm": 14.3046875, "learning_rate": 9.835453382603753e-06, "loss": 1.0453, "step": 5131 }, { "epoch": 0.546073632687806, "grad_norm": 14.765625, "learning_rate": 9.835417851051735e-06, "loss": 1.0964, "step": 5132 }, { "epoch": 0.5461800383060226, "grad_norm": 10.640625, "learning_rate": 9.835382319499716e-06, "loss": 1.0854, "step": 5133 }, { "epoch": 0.5462864439242392, "grad_norm": 16.203125, "learning_rate": 9.835346787947698e-06, "loss": 1.4087, "step": 5134 }, { "epoch": 0.5463928495424558, "grad_norm": 18.921875, "learning_rate": 9.83531125639568e-06, "loss": 1.6307, "step": 5135 }, { "epoch": 0.5464992551606724, "grad_norm": 17.46875, "learning_rate": 9.835275724843661e-06, "loss": 1.1918, "step": 5136 }, { "epoch": 0.5466056607788892, "grad_norm": 19.703125, "learning_rate": 9.835240193291644e-06, "loss": 0.8767, "step": 5137 }, { "epoch": 0.5467120663971058, "grad_norm": 11.296875, "learning_rate": 9.835204661739626e-06, "loss": 1.3441, "step": 5138 }, { "epoch": 0.5468184720153224, "grad_norm": 12.4453125, "learning_rate": 9.835169130187608e-06, "loss": 1.1081, "step": 5139 }, { "epoch": 0.546924877633539, "grad_norm": 12.1328125, "learning_rate": 9.835133598635589e-06, "loss": 1.0882, "step": 5140 }, { "epoch": 0.5470312832517557, "grad_norm": 12.4296875, "learning_rate": 9.83509806708357e-06, "loss": 1.446, "step": 5141 }, { "epoch": 0.5471376888699724, "grad_norm": 11.0234375, "learning_rate": 9.835062535531554e-06, "loss": 1.1227, "step": 5142 }, { "epoch": 0.547244094488189, "grad_norm": 9.671875, "learning_rate": 9.835027003979534e-06, "loss": 0.9703, "step": 5143 }, { "epoch": 0.5473505001064056, "grad_norm": 11.5859375, "learning_rate": 9.834991472427516e-06, "loss": 1.3045, "step": 5144 }, { "epoch": 0.5474569057246222, "grad_norm": 39.09375, "learning_rate": 9.834955940875497e-06, "loss": 1.2596, "step": 5145 }, { "epoch": 0.5475633113428389, "grad_norm": 12.6796875, "learning_rate": 9.834920409323479e-06, "loss": 1.2801, "step": 5146 }, { "epoch": 0.5476697169610556, "grad_norm": 21.609375, "learning_rate": 9.834884877771462e-06, "loss": 1.4507, "step": 5147 }, { "epoch": 0.5477761225792722, "grad_norm": 14.1484375, "learning_rate": 9.834849346219444e-06, "loss": 1.3974, "step": 5148 }, { "epoch": 0.5478825281974888, "grad_norm": 15.7578125, "learning_rate": 9.834813814667425e-06, "loss": 1.5742, "step": 5149 }, { "epoch": 0.5479889338157055, "grad_norm": 21.140625, "learning_rate": 9.834778283115407e-06, "loss": 1.3628, "step": 5150 }, { "epoch": 0.5480953394339221, "grad_norm": 12.453125, "learning_rate": 9.834742751563389e-06, "loss": 0.9581, "step": 5151 }, { "epoch": 0.5482017450521387, "grad_norm": 11.8515625, "learning_rate": 9.834707220011372e-06, "loss": 0.982, "step": 5152 }, { "epoch": 0.5483081506703554, "grad_norm": 11.34375, "learning_rate": 9.834671688459353e-06, "loss": 1.111, "step": 5153 }, { "epoch": 0.548414556288572, "grad_norm": 15.1640625, "learning_rate": 9.834636156907335e-06, "loss": 1.1436, "step": 5154 }, { "epoch": 0.5485209619067887, "grad_norm": 13.921875, "learning_rate": 9.834600625355317e-06, "loss": 1.3776, "step": 5155 }, { "epoch": 0.5486273675250053, "grad_norm": 11.8203125, "learning_rate": 9.834565093803298e-06, "loss": 0.9092, "step": 5156 }, { "epoch": 0.5487337731432219, "grad_norm": 10.96875, "learning_rate": 9.83452956225128e-06, "loss": 1.3464, "step": 5157 }, { "epoch": 0.5488401787614386, "grad_norm": 14.0859375, "learning_rate": 9.834494030699261e-06, "loss": 1.1494, "step": 5158 }, { "epoch": 0.5489465843796553, "grad_norm": 11.859375, "learning_rate": 9.834458499147243e-06, "loss": 1.2385, "step": 5159 }, { "epoch": 0.5490529899978719, "grad_norm": 10.7578125, "learning_rate": 9.834422967595225e-06, "loss": 1.5639, "step": 5160 }, { "epoch": 0.5491593956160885, "grad_norm": 14.203125, "learning_rate": 9.834387436043206e-06, "loss": 1.202, "step": 5161 }, { "epoch": 0.5492658012343051, "grad_norm": 13.7890625, "learning_rate": 9.83435190449119e-06, "loss": 1.305, "step": 5162 }, { "epoch": 0.5493722068525219, "grad_norm": 18.03125, "learning_rate": 9.834316372939171e-06, "loss": 1.2191, "step": 5163 }, { "epoch": 0.5494786124707385, "grad_norm": 16.265625, "learning_rate": 9.834280841387153e-06, "loss": 1.4593, "step": 5164 }, { "epoch": 0.5495850180889551, "grad_norm": 15.375, "learning_rate": 9.834245309835134e-06, "loss": 1.0094, "step": 5165 }, { "epoch": 0.5496914237071717, "grad_norm": 13.28125, "learning_rate": 9.834209778283116e-06, "loss": 1.2632, "step": 5166 }, { "epoch": 0.5497978293253883, "grad_norm": 12.0, "learning_rate": 9.8341742467311e-06, "loss": 1.0138, "step": 5167 }, { "epoch": 0.5499042349436051, "grad_norm": 14.5625, "learning_rate": 9.834138715179081e-06, "loss": 1.1187, "step": 5168 }, { "epoch": 0.5500106405618217, "grad_norm": 15.21875, "learning_rate": 9.834103183627063e-06, "loss": 1.147, "step": 5169 }, { "epoch": 0.5501170461800383, "grad_norm": 13.546875, "learning_rate": 9.834067652075044e-06, "loss": 1.3868, "step": 5170 }, { "epoch": 0.5502234517982549, "grad_norm": 11.8671875, "learning_rate": 9.834032120523026e-06, "loss": 1.1969, "step": 5171 }, { "epoch": 0.5503298574164716, "grad_norm": 13.90625, "learning_rate": 9.833996588971007e-06, "loss": 1.2737, "step": 5172 }, { "epoch": 0.5504362630346882, "grad_norm": 12.6484375, "learning_rate": 9.833961057418989e-06, "loss": 1.3192, "step": 5173 }, { "epoch": 0.5505426686529049, "grad_norm": 37.1875, "learning_rate": 9.83392552586697e-06, "loss": 1.2317, "step": 5174 }, { "epoch": 0.5506490742711215, "grad_norm": 10.6484375, "learning_rate": 9.833889994314952e-06, "loss": 1.3089, "step": 5175 }, { "epoch": 0.5507554798893382, "grad_norm": 12.0703125, "learning_rate": 9.833854462762934e-06, "loss": 1.16, "step": 5176 }, { "epoch": 0.5508618855075548, "grad_norm": 16.671875, "learning_rate": 9.833818931210915e-06, "loss": 1.234, "step": 5177 }, { "epoch": 0.5509682911257714, "grad_norm": 13.0859375, "learning_rate": 9.833783399658899e-06, "loss": 1.5182, "step": 5178 }, { "epoch": 0.5510746967439881, "grad_norm": 14.375, "learning_rate": 9.83374786810688e-06, "loss": 1.1771, "step": 5179 }, { "epoch": 0.5511811023622047, "grad_norm": 8.9375, "learning_rate": 9.833712336554862e-06, "loss": 0.9237, "step": 5180 }, { "epoch": 0.5512875079804214, "grad_norm": 12.8359375, "learning_rate": 9.833676805002843e-06, "loss": 1.2638, "step": 5181 }, { "epoch": 0.551393913598638, "grad_norm": 13.2578125, "learning_rate": 9.833641273450825e-06, "loss": 0.8509, "step": 5182 }, { "epoch": 0.5515003192168546, "grad_norm": 10.53125, "learning_rate": 9.833605741898807e-06, "loss": 1.0265, "step": 5183 }, { "epoch": 0.5516067248350713, "grad_norm": 14.1953125, "learning_rate": 9.833570210346788e-06, "loss": 1.2698, "step": 5184 }, { "epoch": 0.551713130453288, "grad_norm": 12.8125, "learning_rate": 9.83353467879477e-06, "loss": 1.0347, "step": 5185 }, { "epoch": 0.5518195360715046, "grad_norm": 15.984375, "learning_rate": 9.833499147242752e-06, "loss": 1.0798, "step": 5186 }, { "epoch": 0.5519259416897212, "grad_norm": 14.796875, "learning_rate": 9.833463615690733e-06, "loss": 1.5654, "step": 5187 }, { "epoch": 0.5520323473079378, "grad_norm": 16.859375, "learning_rate": 9.833428084138716e-06, "loss": 1.1431, "step": 5188 }, { "epoch": 0.5521387529261546, "grad_norm": 75.0, "learning_rate": 9.833392552586698e-06, "loss": 1.2951, "step": 5189 }, { "epoch": 0.5522451585443712, "grad_norm": 18.6875, "learning_rate": 9.83335702103468e-06, "loss": 1.6004, "step": 5190 }, { "epoch": 0.5523515641625878, "grad_norm": 15.453125, "learning_rate": 9.833321489482661e-06, "loss": 1.4475, "step": 5191 }, { "epoch": 0.5524579697808044, "grad_norm": 10.3203125, "learning_rate": 9.833285957930643e-06, "loss": 1.2307, "step": 5192 }, { "epoch": 0.552564375399021, "grad_norm": 12.546875, "learning_rate": 9.833250426378626e-06, "loss": 1.0315, "step": 5193 }, { "epoch": 0.5526707810172377, "grad_norm": 13.125, "learning_rate": 9.833214894826608e-06, "loss": 1.0884, "step": 5194 }, { "epoch": 0.5527771866354544, "grad_norm": 10.75, "learning_rate": 9.83317936327459e-06, "loss": 1.274, "step": 5195 }, { "epoch": 0.552883592253671, "grad_norm": 11.421875, "learning_rate": 9.833143831722571e-06, "loss": 1.1097, "step": 5196 }, { "epoch": 0.5529899978718876, "grad_norm": 10.9921875, "learning_rate": 9.833108300170553e-06, "loss": 1.3805, "step": 5197 }, { "epoch": 0.5530964034901042, "grad_norm": 11.9921875, "learning_rate": 9.833072768618534e-06, "loss": 1.1397, "step": 5198 }, { "epoch": 0.5532028091083209, "grad_norm": 14.1484375, "learning_rate": 9.833037237066516e-06, "loss": 0.9669, "step": 5199 }, { "epoch": 0.5533092147265376, "grad_norm": 14.4609375, "learning_rate": 9.833001705514497e-06, "loss": 1.2913, "step": 5200 }, { "epoch": 0.5534156203447542, "grad_norm": 19.125, "learning_rate": 9.832966173962479e-06, "loss": 1.3272, "step": 5201 }, { "epoch": 0.5535220259629708, "grad_norm": 10.4296875, "learning_rate": 9.83293064241046e-06, "loss": 1.054, "step": 5202 }, { "epoch": 0.5536284315811875, "grad_norm": 11.015625, "learning_rate": 9.832895110858444e-06, "loss": 1.0337, "step": 5203 }, { "epoch": 0.5537348371994041, "grad_norm": 9.96875, "learning_rate": 9.832859579306426e-06, "loss": 1.1596, "step": 5204 }, { "epoch": 0.5538412428176208, "grad_norm": 11.8984375, "learning_rate": 9.832824047754407e-06, "loss": 1.2126, "step": 5205 }, { "epoch": 0.5539476484358374, "grad_norm": 9.8203125, "learning_rate": 9.832788516202389e-06, "loss": 1.1005, "step": 5206 }, { "epoch": 0.5540540540540541, "grad_norm": 12.2421875, "learning_rate": 9.83275298465037e-06, "loss": 1.4888, "step": 5207 }, { "epoch": 0.5541604596722707, "grad_norm": 13.46875, "learning_rate": 9.832717453098352e-06, "loss": 0.9953, "step": 5208 }, { "epoch": 0.5542668652904873, "grad_norm": 11.390625, "learning_rate": 9.832681921546334e-06, "loss": 1.0115, "step": 5209 }, { "epoch": 0.554373270908704, "grad_norm": 10.53125, "learning_rate": 9.832646389994315e-06, "loss": 1.0685, "step": 5210 }, { "epoch": 0.5544796765269207, "grad_norm": 12.828125, "learning_rate": 9.832610858442297e-06, "loss": 1.5009, "step": 5211 }, { "epoch": 0.5545860821451373, "grad_norm": 12.90625, "learning_rate": 9.832575326890278e-06, "loss": 1.2649, "step": 5212 }, { "epoch": 0.5546924877633539, "grad_norm": 13.6171875, "learning_rate": 9.83253979533826e-06, "loss": 1.3714, "step": 5213 }, { "epoch": 0.5547988933815705, "grad_norm": 12.8828125, "learning_rate": 9.832504263786243e-06, "loss": 1.1829, "step": 5214 }, { "epoch": 0.5549052989997872, "grad_norm": 13.1484375, "learning_rate": 9.832468732234225e-06, "loss": 0.9882, "step": 5215 }, { "epoch": 0.5550117046180039, "grad_norm": 13.3203125, "learning_rate": 9.832433200682207e-06, "loss": 1.4799, "step": 5216 }, { "epoch": 0.5551181102362205, "grad_norm": 14.6171875, "learning_rate": 9.832397669130188e-06, "loss": 1.4491, "step": 5217 }, { "epoch": 0.5552245158544371, "grad_norm": 18.875, "learning_rate": 9.83236213757817e-06, "loss": 1.3898, "step": 5218 }, { "epoch": 0.5553309214726537, "grad_norm": 11.3515625, "learning_rate": 9.832326606026153e-06, "loss": 1.2322, "step": 5219 }, { "epoch": 0.5554373270908703, "grad_norm": 14.875, "learning_rate": 9.832291074474135e-06, "loss": 0.9476, "step": 5220 }, { "epoch": 0.5555437327090871, "grad_norm": 18.578125, "learning_rate": 9.832255542922116e-06, "loss": 1.2276, "step": 5221 }, { "epoch": 0.5556501383273037, "grad_norm": 12.0078125, "learning_rate": 9.832220011370098e-06, "loss": 1.2366, "step": 5222 }, { "epoch": 0.5557565439455203, "grad_norm": 12.7578125, "learning_rate": 9.83218447981808e-06, "loss": 1.1761, "step": 5223 }, { "epoch": 0.5558629495637369, "grad_norm": 10.25, "learning_rate": 9.832148948266061e-06, "loss": 1.0187, "step": 5224 }, { "epoch": 0.5559693551819536, "grad_norm": 13.7578125, "learning_rate": 9.832113416714043e-06, "loss": 1.2771, "step": 5225 }, { "epoch": 0.5560757608001703, "grad_norm": 11.1171875, "learning_rate": 9.832077885162024e-06, "loss": 1.1539, "step": 5226 }, { "epoch": 0.5561821664183869, "grad_norm": 14.6171875, "learning_rate": 9.832042353610006e-06, "loss": 1.251, "step": 5227 }, { "epoch": 0.5562885720366035, "grad_norm": 12.640625, "learning_rate": 9.832006822057987e-06, "loss": 1.3236, "step": 5228 }, { "epoch": 0.5563949776548202, "grad_norm": 16.765625, "learning_rate": 9.83197129050597e-06, "loss": 1.227, "step": 5229 }, { "epoch": 0.5565013832730368, "grad_norm": 10.84375, "learning_rate": 9.831935758953952e-06, "loss": 0.9409, "step": 5230 }, { "epoch": 0.5566077888912535, "grad_norm": 10.2890625, "learning_rate": 9.831900227401934e-06, "loss": 1.2888, "step": 5231 }, { "epoch": 0.5567141945094701, "grad_norm": 9.6875, "learning_rate": 9.831864695849916e-06, "loss": 1.3302, "step": 5232 }, { "epoch": 0.5568206001276867, "grad_norm": 10.4921875, "learning_rate": 9.831829164297897e-06, "loss": 1.1156, "step": 5233 }, { "epoch": 0.5569270057459034, "grad_norm": 22.71875, "learning_rate": 9.831793632745879e-06, "loss": 1.2387, "step": 5234 }, { "epoch": 0.55703341136412, "grad_norm": 16.40625, "learning_rate": 9.83175810119386e-06, "loss": 1.1386, "step": 5235 }, { "epoch": 0.5571398169823367, "grad_norm": 13.328125, "learning_rate": 9.831722569641842e-06, "loss": 1.1488, "step": 5236 }, { "epoch": 0.5572462226005533, "grad_norm": 10.8828125, "learning_rate": 9.831687038089824e-06, "loss": 1.1681, "step": 5237 }, { "epoch": 0.55735262821877, "grad_norm": 12.625, "learning_rate": 9.831651506537805e-06, "loss": 1.2233, "step": 5238 }, { "epoch": 0.5574590338369866, "grad_norm": 11.953125, "learning_rate": 9.831615974985789e-06, "loss": 0.9979, "step": 5239 }, { "epoch": 0.5575654394552032, "grad_norm": 13.0859375, "learning_rate": 9.83158044343377e-06, "loss": 1.255, "step": 5240 }, { "epoch": 0.5576718450734198, "grad_norm": 10.1171875, "learning_rate": 9.831544911881752e-06, "loss": 1.0889, "step": 5241 }, { "epoch": 0.5577782506916366, "grad_norm": 9.8046875, "learning_rate": 9.831509380329733e-06, "loss": 1.1186, "step": 5242 }, { "epoch": 0.5578846563098532, "grad_norm": 11.4765625, "learning_rate": 9.831473848777715e-06, "loss": 1.4543, "step": 5243 }, { "epoch": 0.5579910619280698, "grad_norm": 12.359375, "learning_rate": 9.831438317225698e-06, "loss": 1.1491, "step": 5244 }, { "epoch": 0.5580974675462864, "grad_norm": 12.109375, "learning_rate": 9.83140278567368e-06, "loss": 1.2591, "step": 5245 }, { "epoch": 0.558203873164503, "grad_norm": 12.4296875, "learning_rate": 9.831367254121661e-06, "loss": 1.4607, "step": 5246 }, { "epoch": 0.5583102787827198, "grad_norm": 17.390625, "learning_rate": 9.831331722569643e-06, "loss": 1.054, "step": 5247 }, { "epoch": 0.5584166844009364, "grad_norm": 14.8515625, "learning_rate": 9.831296191017625e-06, "loss": 1.4272, "step": 5248 }, { "epoch": 0.558523090019153, "grad_norm": 14.484375, "learning_rate": 9.831260659465606e-06, "loss": 1.2317, "step": 5249 }, { "epoch": 0.5586294956373696, "grad_norm": 11.0, "learning_rate": 9.831225127913588e-06, "loss": 1.294, "step": 5250 }, { "epoch": 0.5587359012555863, "grad_norm": 14.3203125, "learning_rate": 9.83118959636157e-06, "loss": 1.7573, "step": 5251 }, { "epoch": 0.558842306873803, "grad_norm": 17.515625, "learning_rate": 9.831154064809551e-06, "loss": 0.6347, "step": 5252 }, { "epoch": 0.5589487124920196, "grad_norm": 12.140625, "learning_rate": 9.831118533257533e-06, "loss": 1.311, "step": 5253 }, { "epoch": 0.5590551181102362, "grad_norm": 11.7421875, "learning_rate": 9.831083001705514e-06, "loss": 1.4515, "step": 5254 }, { "epoch": 0.5591615237284528, "grad_norm": 11.1796875, "learning_rate": 9.831047470153498e-06, "loss": 1.1788, "step": 5255 }, { "epoch": 0.5592679293466695, "grad_norm": 11.96875, "learning_rate": 9.83101193860148e-06, "loss": 1.1542, "step": 5256 }, { "epoch": 0.5593743349648862, "grad_norm": 14.125, "learning_rate": 9.830976407049461e-06, "loss": 1.1991, "step": 5257 }, { "epoch": 0.5594807405831028, "grad_norm": 12.8046875, "learning_rate": 9.830940875497442e-06, "loss": 1.3174, "step": 5258 }, { "epoch": 0.5595871462013194, "grad_norm": 13.8203125, "learning_rate": 9.830905343945424e-06, "loss": 1.2173, "step": 5259 }, { "epoch": 0.5596935518195361, "grad_norm": 11.4609375, "learning_rate": 9.830869812393407e-06, "loss": 1.2003, "step": 5260 }, { "epoch": 0.5597999574377527, "grad_norm": 12.6875, "learning_rate": 9.830834280841389e-06, "loss": 1.2871, "step": 5261 }, { "epoch": 0.5599063630559693, "grad_norm": 11.7734375, "learning_rate": 9.830798749289369e-06, "loss": 1.3553, "step": 5262 }, { "epoch": 0.560012768674186, "grad_norm": 19.140625, "learning_rate": 9.83076321773735e-06, "loss": 1.3485, "step": 5263 }, { "epoch": 0.5601191742924027, "grad_norm": 11.296875, "learning_rate": 9.830727686185332e-06, "loss": 1.6247, "step": 5264 }, { "epoch": 0.5602255799106193, "grad_norm": 10.2890625, "learning_rate": 9.830692154633315e-06, "loss": 1.3135, "step": 5265 }, { "epoch": 0.5603319855288359, "grad_norm": 10.8359375, "learning_rate": 9.830656623081297e-06, "loss": 0.9221, "step": 5266 }, { "epoch": 0.5604383911470525, "grad_norm": 10.0625, "learning_rate": 9.830621091529279e-06, "loss": 1.0121, "step": 5267 }, { "epoch": 0.5605447967652692, "grad_norm": 10.921875, "learning_rate": 9.83058555997726e-06, "loss": 1.3103, "step": 5268 }, { "epoch": 0.5606512023834859, "grad_norm": 12.875, "learning_rate": 9.830550028425242e-06, "loss": 1.3187, "step": 5269 }, { "epoch": 0.5607576080017025, "grad_norm": 13.515625, "learning_rate": 9.830514496873225e-06, "loss": 1.2997, "step": 5270 }, { "epoch": 0.5608640136199191, "grad_norm": 11.15625, "learning_rate": 9.830478965321207e-06, "loss": 1.3545, "step": 5271 }, { "epoch": 0.5609704192381357, "grad_norm": 12.390625, "learning_rate": 9.830443433769188e-06, "loss": 1.0341, "step": 5272 }, { "epoch": 0.5610768248563525, "grad_norm": 12.5859375, "learning_rate": 9.83040790221717e-06, "loss": 1.5031, "step": 5273 }, { "epoch": 0.5611832304745691, "grad_norm": 14.5703125, "learning_rate": 9.830372370665152e-06, "loss": 1.3539, "step": 5274 }, { "epoch": 0.5612896360927857, "grad_norm": 14.9375, "learning_rate": 9.830336839113133e-06, "loss": 1.1153, "step": 5275 }, { "epoch": 0.5613960417110023, "grad_norm": 16.234375, "learning_rate": 9.830301307561115e-06, "loss": 1.3322, "step": 5276 }, { "epoch": 0.561502447329219, "grad_norm": 13.5546875, "learning_rate": 9.830265776009096e-06, "loss": 1.2288, "step": 5277 }, { "epoch": 0.5616088529474357, "grad_norm": 12.921875, "learning_rate": 9.830230244457078e-06, "loss": 1.2422, "step": 5278 }, { "epoch": 0.5617152585656523, "grad_norm": 13.3671875, "learning_rate": 9.83019471290506e-06, "loss": 1.4446, "step": 5279 }, { "epoch": 0.5618216641838689, "grad_norm": 12.6875, "learning_rate": 9.830159181353043e-06, "loss": 1.067, "step": 5280 }, { "epoch": 0.5619280698020855, "grad_norm": 10.0625, "learning_rate": 9.830123649801024e-06, "loss": 0.9927, "step": 5281 }, { "epoch": 0.5620344754203022, "grad_norm": 12.4375, "learning_rate": 9.830088118249006e-06, "loss": 1.4371, "step": 5282 }, { "epoch": 0.5621408810385188, "grad_norm": 13.1015625, "learning_rate": 9.830052586696988e-06, "loss": 1.0995, "step": 5283 }, { "epoch": 0.5622472866567355, "grad_norm": 15.890625, "learning_rate": 9.83001705514497e-06, "loss": 1.5975, "step": 5284 }, { "epoch": 0.5623536922749521, "grad_norm": 26.71875, "learning_rate": 9.829981523592953e-06, "loss": 1.1443, "step": 5285 }, { "epoch": 0.5624600978931688, "grad_norm": 13.09375, "learning_rate": 9.829945992040934e-06, "loss": 1.0029, "step": 5286 }, { "epoch": 0.5625665035113854, "grad_norm": 14.171875, "learning_rate": 9.829910460488916e-06, "loss": 1.2707, "step": 5287 }, { "epoch": 0.562672909129602, "grad_norm": 11.2734375, "learning_rate": 9.829874928936897e-06, "loss": 1.2797, "step": 5288 }, { "epoch": 0.5627793147478187, "grad_norm": 14.2265625, "learning_rate": 9.829839397384879e-06, "loss": 1.3058, "step": 5289 }, { "epoch": 0.5628857203660353, "grad_norm": 13.359375, "learning_rate": 9.82980386583286e-06, "loss": 1.4927, "step": 5290 }, { "epoch": 0.562992125984252, "grad_norm": 15.5390625, "learning_rate": 9.829768334280842e-06, "loss": 1.0871, "step": 5291 }, { "epoch": 0.5630985316024686, "grad_norm": 13.75, "learning_rate": 9.829732802728824e-06, "loss": 1.2317, "step": 5292 }, { "epoch": 0.5632049372206852, "grad_norm": 12.0234375, "learning_rate": 9.829697271176805e-06, "loss": 1.3479, "step": 5293 }, { "epoch": 0.5633113428389019, "grad_norm": 11.5546875, "learning_rate": 9.829661739624787e-06, "loss": 1.0186, "step": 5294 }, { "epoch": 0.5634177484571186, "grad_norm": 13.609375, "learning_rate": 9.829626208072769e-06, "loss": 1.4164, "step": 5295 }, { "epoch": 0.5635241540753352, "grad_norm": 15.2890625, "learning_rate": 9.829590676520752e-06, "loss": 1.4055, "step": 5296 }, { "epoch": 0.5636305596935518, "grad_norm": 13.28125, "learning_rate": 9.829555144968734e-06, "loss": 1.2069, "step": 5297 }, { "epoch": 0.5637369653117684, "grad_norm": 15.625, "learning_rate": 9.829519613416715e-06, "loss": 1.4722, "step": 5298 }, { "epoch": 0.5638433709299852, "grad_norm": 11.7890625, "learning_rate": 9.829484081864697e-06, "loss": 1.3639, "step": 5299 }, { "epoch": 0.5639497765482018, "grad_norm": 12.953125, "learning_rate": 9.829448550312678e-06, "loss": 0.8833, "step": 5300 }, { "epoch": 0.5640561821664184, "grad_norm": 11.8984375, "learning_rate": 9.82941301876066e-06, "loss": 1.3064, "step": 5301 }, { "epoch": 0.564162587784635, "grad_norm": 12.96875, "learning_rate": 9.829377487208642e-06, "loss": 1.4148, "step": 5302 }, { "epoch": 0.5642689934028516, "grad_norm": 14.40625, "learning_rate": 9.829341955656623e-06, "loss": 1.2858, "step": 5303 }, { "epoch": 0.5643753990210683, "grad_norm": 15.96875, "learning_rate": 9.829306424104605e-06, "loss": 1.0446, "step": 5304 }, { "epoch": 0.564481804639285, "grad_norm": 15.125, "learning_rate": 9.829270892552586e-06, "loss": 1.5068, "step": 5305 }, { "epoch": 0.5645882102575016, "grad_norm": 14.203125, "learning_rate": 9.82923536100057e-06, "loss": 0.9423, "step": 5306 }, { "epoch": 0.5646946158757182, "grad_norm": 14.21875, "learning_rate": 9.829199829448551e-06, "loss": 1.1027, "step": 5307 }, { "epoch": 0.5648010214939349, "grad_norm": 17.859375, "learning_rate": 9.829164297896533e-06, "loss": 1.1637, "step": 5308 }, { "epoch": 0.5649074271121515, "grad_norm": 12.1328125, "learning_rate": 9.829128766344515e-06, "loss": 1.2241, "step": 5309 }, { "epoch": 0.5650138327303682, "grad_norm": 20.25, "learning_rate": 9.829093234792496e-06, "loss": 1.2612, "step": 5310 }, { "epoch": 0.5651202383485848, "grad_norm": 15.0078125, "learning_rate": 9.82905770324048e-06, "loss": 1.3539, "step": 5311 }, { "epoch": 0.5652266439668014, "grad_norm": 17.96875, "learning_rate": 9.829022171688461e-06, "loss": 1.1613, "step": 5312 }, { "epoch": 0.5653330495850181, "grad_norm": 11.2109375, "learning_rate": 9.828986640136443e-06, "loss": 1.4358, "step": 5313 }, { "epoch": 0.5654394552032347, "grad_norm": 17.03125, "learning_rate": 9.828951108584424e-06, "loss": 1.331, "step": 5314 }, { "epoch": 0.5655458608214514, "grad_norm": 18.484375, "learning_rate": 9.828915577032406e-06, "loss": 1.1391, "step": 5315 }, { "epoch": 0.565652266439668, "grad_norm": 13.9921875, "learning_rate": 9.828880045480387e-06, "loss": 1.1018, "step": 5316 }, { "epoch": 0.5657586720578847, "grad_norm": 13.1640625, "learning_rate": 9.828844513928369e-06, "loss": 1.2489, "step": 5317 }, { "epoch": 0.5658650776761013, "grad_norm": 13.09375, "learning_rate": 9.82880898237635e-06, "loss": 1.3192, "step": 5318 }, { "epoch": 0.5659714832943179, "grad_norm": 13.8515625, "learning_rate": 9.828773450824332e-06, "loss": 1.4979, "step": 5319 }, { "epoch": 0.5660778889125346, "grad_norm": 11.6484375, "learning_rate": 9.828737919272314e-06, "loss": 1.2752, "step": 5320 }, { "epoch": 0.5661842945307513, "grad_norm": 11.453125, "learning_rate": 9.828702387720297e-06, "loss": 1.111, "step": 5321 }, { "epoch": 0.5662907001489679, "grad_norm": 14.75, "learning_rate": 9.828666856168279e-06, "loss": 1.228, "step": 5322 }, { "epoch": 0.5663971057671845, "grad_norm": 12.125, "learning_rate": 9.82863132461626e-06, "loss": 1.2079, "step": 5323 }, { "epoch": 0.5665035113854011, "grad_norm": 11.578125, "learning_rate": 9.828595793064242e-06, "loss": 1.0784, "step": 5324 }, { "epoch": 0.5666099170036177, "grad_norm": 18.953125, "learning_rate": 9.828560261512224e-06, "loss": 1.5596, "step": 5325 }, { "epoch": 0.5667163226218345, "grad_norm": 11.25, "learning_rate": 9.828524729960205e-06, "loss": 1.2064, "step": 5326 }, { "epoch": 0.5668227282400511, "grad_norm": 11.2578125, "learning_rate": 9.828489198408187e-06, "loss": 1.0911, "step": 5327 }, { "epoch": 0.5669291338582677, "grad_norm": 12.4140625, "learning_rate": 9.828453666856168e-06, "loss": 1.1752, "step": 5328 }, { "epoch": 0.5670355394764843, "grad_norm": 15.9296875, "learning_rate": 9.82841813530415e-06, "loss": 1.5521, "step": 5329 }, { "epoch": 0.567141945094701, "grad_norm": 14.0859375, "learning_rate": 9.828382603752132e-06, "loss": 1.5296, "step": 5330 }, { "epoch": 0.5672483507129177, "grad_norm": 11.3828125, "learning_rate": 9.828347072200113e-06, "loss": 1.2125, "step": 5331 }, { "epoch": 0.5673547563311343, "grad_norm": 12.375, "learning_rate": 9.828311540648097e-06, "loss": 1.4672, "step": 5332 }, { "epoch": 0.5674611619493509, "grad_norm": 12.0390625, "learning_rate": 9.828276009096078e-06, "loss": 0.9886, "step": 5333 }, { "epoch": 0.5675675675675675, "grad_norm": 19.578125, "learning_rate": 9.82824047754406e-06, "loss": 1.2566, "step": 5334 }, { "epoch": 0.5676739731857842, "grad_norm": 11.5859375, "learning_rate": 9.828204945992041e-06, "loss": 1.0917, "step": 5335 }, { "epoch": 0.5677803788040009, "grad_norm": 15.5859375, "learning_rate": 9.828169414440023e-06, "loss": 1.262, "step": 5336 }, { "epoch": 0.5678867844222175, "grad_norm": 11.109375, "learning_rate": 9.828133882888006e-06, "loss": 1.0492, "step": 5337 }, { "epoch": 0.5679931900404341, "grad_norm": 11.125, "learning_rate": 9.828098351335988e-06, "loss": 1.2559, "step": 5338 }, { "epoch": 0.5680995956586508, "grad_norm": 14.28125, "learning_rate": 9.82806281978397e-06, "loss": 1.426, "step": 5339 }, { "epoch": 0.5682060012768674, "grad_norm": 14.234375, "learning_rate": 9.828027288231951e-06, "loss": 1.1913, "step": 5340 }, { "epoch": 0.5683124068950841, "grad_norm": 16.28125, "learning_rate": 9.827991756679933e-06, "loss": 1.1832, "step": 5341 }, { "epoch": 0.5684188125133007, "grad_norm": 11.203125, "learning_rate": 9.827956225127914e-06, "loss": 1.1998, "step": 5342 }, { "epoch": 0.5685252181315174, "grad_norm": 11.5, "learning_rate": 9.827920693575896e-06, "loss": 1.0097, "step": 5343 }, { "epoch": 0.568631623749734, "grad_norm": 13.375, "learning_rate": 9.827885162023878e-06, "loss": 1.406, "step": 5344 }, { "epoch": 0.5687380293679506, "grad_norm": 18.0, "learning_rate": 9.82784963047186e-06, "loss": 1.4884, "step": 5345 }, { "epoch": 0.5688444349861673, "grad_norm": 14.0625, "learning_rate": 9.82781409891984e-06, "loss": 1.0963, "step": 5346 }, { "epoch": 0.568950840604384, "grad_norm": 14.7578125, "learning_rate": 9.827778567367824e-06, "loss": 1.1333, "step": 5347 }, { "epoch": 0.5690572462226006, "grad_norm": 14.4375, "learning_rate": 9.827743035815806e-06, "loss": 1.2267, "step": 5348 }, { "epoch": 0.5691636518408172, "grad_norm": 18.65625, "learning_rate": 9.827707504263787e-06, "loss": 1.3487, "step": 5349 }, { "epoch": 0.5692700574590338, "grad_norm": 20.46875, "learning_rate": 9.827671972711769e-06, "loss": 1.5183, "step": 5350 }, { "epoch": 0.5693764630772504, "grad_norm": 11.640625, "learning_rate": 9.82763644115975e-06, "loss": 1.2673, "step": 5351 }, { "epoch": 0.5694828686954672, "grad_norm": 15.5546875, "learning_rate": 9.827600909607732e-06, "loss": 1.1479, "step": 5352 }, { "epoch": 0.5695892743136838, "grad_norm": 11.375, "learning_rate": 9.827565378055714e-06, "loss": 1.1092, "step": 5353 }, { "epoch": 0.5696956799319004, "grad_norm": 9.59375, "learning_rate": 9.827529846503695e-06, "loss": 1.2964, "step": 5354 }, { "epoch": 0.569802085550117, "grad_norm": 13.25, "learning_rate": 9.827494314951677e-06, "loss": 0.9398, "step": 5355 }, { "epoch": 0.5699084911683336, "grad_norm": 24.71875, "learning_rate": 9.827458783399659e-06, "loss": 1.149, "step": 5356 }, { "epoch": 0.5700148967865504, "grad_norm": 11.734375, "learning_rate": 9.827423251847642e-06, "loss": 1.2777, "step": 5357 }, { "epoch": 0.570121302404767, "grad_norm": 16.765625, "learning_rate": 9.827387720295623e-06, "loss": 1.2547, "step": 5358 }, { "epoch": 0.5702277080229836, "grad_norm": 18.765625, "learning_rate": 9.827352188743605e-06, "loss": 1.2152, "step": 5359 }, { "epoch": 0.5703341136412002, "grad_norm": 21.3125, "learning_rate": 9.827316657191587e-06, "loss": 1.2368, "step": 5360 }, { "epoch": 0.5704405192594169, "grad_norm": 12.09375, "learning_rate": 9.827281125639568e-06, "loss": 1.1852, "step": 5361 }, { "epoch": 0.5705469248776336, "grad_norm": 15.7578125, "learning_rate": 9.827245594087552e-06, "loss": 1.2247, "step": 5362 }, { "epoch": 0.5706533304958502, "grad_norm": 12.4375, "learning_rate": 9.827210062535533e-06, "loss": 1.2218, "step": 5363 }, { "epoch": 0.5707597361140668, "grad_norm": 18.203125, "learning_rate": 9.827174530983515e-06, "loss": 1.631, "step": 5364 }, { "epoch": 0.5708661417322834, "grad_norm": 14.6875, "learning_rate": 9.827138999431496e-06, "loss": 1.1473, "step": 5365 }, { "epoch": 0.5709725473505001, "grad_norm": 13.2734375, "learning_rate": 9.827103467879478e-06, "loss": 1.279, "step": 5366 }, { "epoch": 0.5710789529687168, "grad_norm": 14.109375, "learning_rate": 9.82706793632746e-06, "loss": 1.3103, "step": 5367 }, { "epoch": 0.5711853585869334, "grad_norm": 13.1640625, "learning_rate": 9.827032404775441e-06, "loss": 1.1454, "step": 5368 }, { "epoch": 0.57129176420515, "grad_norm": 13.3125, "learning_rate": 9.826996873223423e-06, "loss": 1.3672, "step": 5369 }, { "epoch": 0.5713981698233667, "grad_norm": 17.078125, "learning_rate": 9.826961341671404e-06, "loss": 0.9729, "step": 5370 }, { "epoch": 0.5715045754415833, "grad_norm": 11.515625, "learning_rate": 9.826925810119386e-06, "loss": 1.0071, "step": 5371 }, { "epoch": 0.5716109810597999, "grad_norm": 15.484375, "learning_rate": 9.826890278567368e-06, "loss": 1.0929, "step": 5372 }, { "epoch": 0.5717173866780166, "grad_norm": 11.796875, "learning_rate": 9.826854747015351e-06, "loss": 1.1254, "step": 5373 }, { "epoch": 0.5718237922962333, "grad_norm": 12.3515625, "learning_rate": 9.826819215463333e-06, "loss": 1.3461, "step": 5374 }, { "epoch": 0.5719301979144499, "grad_norm": 8.78125, "learning_rate": 9.826783683911314e-06, "loss": 1.0594, "step": 5375 }, { "epoch": 0.5720366035326665, "grad_norm": 16.03125, "learning_rate": 9.826748152359296e-06, "loss": 1.1125, "step": 5376 }, { "epoch": 0.5721430091508831, "grad_norm": 16.03125, "learning_rate": 9.826712620807277e-06, "loss": 1.2425, "step": 5377 }, { "epoch": 0.5722494147690999, "grad_norm": 11.109375, "learning_rate": 9.82667708925526e-06, "loss": 1.143, "step": 5378 }, { "epoch": 0.5723558203873165, "grad_norm": 13.875, "learning_rate": 9.826641557703242e-06, "loss": 1.2464, "step": 5379 }, { "epoch": 0.5724622260055331, "grad_norm": 12.453125, "learning_rate": 9.826606026151224e-06, "loss": 1.0472, "step": 5380 }, { "epoch": 0.5725686316237497, "grad_norm": 10.2734375, "learning_rate": 9.826570494599205e-06, "loss": 1.1811, "step": 5381 }, { "epoch": 0.5726750372419663, "grad_norm": 10.3828125, "learning_rate": 9.826534963047185e-06, "loss": 0.9686, "step": 5382 }, { "epoch": 0.5727814428601831, "grad_norm": 9.4296875, "learning_rate": 9.826499431495169e-06, "loss": 0.9075, "step": 5383 }, { "epoch": 0.5728878484783997, "grad_norm": 13.3515625, "learning_rate": 9.82646389994315e-06, "loss": 1.2498, "step": 5384 }, { "epoch": 0.5729942540966163, "grad_norm": 12.078125, "learning_rate": 9.826428368391132e-06, "loss": 1.1663, "step": 5385 }, { "epoch": 0.5731006597148329, "grad_norm": 11.953125, "learning_rate": 9.826392836839114e-06, "loss": 1.3037, "step": 5386 }, { "epoch": 0.5732070653330495, "grad_norm": 12.8046875, "learning_rate": 9.826357305287095e-06, "loss": 1.3896, "step": 5387 }, { "epoch": 0.5733134709512663, "grad_norm": 12.453125, "learning_rate": 9.826321773735078e-06, "loss": 1.2789, "step": 5388 }, { "epoch": 0.5734198765694829, "grad_norm": 11.40625, "learning_rate": 9.82628624218306e-06, "loss": 1.2858, "step": 5389 }, { "epoch": 0.5735262821876995, "grad_norm": 14.71875, "learning_rate": 9.826250710631042e-06, "loss": 1.5252, "step": 5390 }, { "epoch": 0.5736326878059161, "grad_norm": 13.9609375, "learning_rate": 9.826215179079023e-06, "loss": 1.3519, "step": 5391 }, { "epoch": 0.5737390934241328, "grad_norm": 11.8984375, "learning_rate": 9.826179647527005e-06, "loss": 1.2518, "step": 5392 }, { "epoch": 0.5738454990423494, "grad_norm": 17.0625, "learning_rate": 9.826144115974986e-06, "loss": 1.2452, "step": 5393 }, { "epoch": 0.5739519046605661, "grad_norm": 14.4921875, "learning_rate": 9.826108584422968e-06, "loss": 1.5453, "step": 5394 }, { "epoch": 0.5740583102787827, "grad_norm": 12.109375, "learning_rate": 9.82607305287095e-06, "loss": 1.2635, "step": 5395 }, { "epoch": 0.5741647158969994, "grad_norm": 853.0, "learning_rate": 9.826037521318931e-06, "loss": 1.153, "step": 5396 }, { "epoch": 0.574271121515216, "grad_norm": 11.65625, "learning_rate": 9.826001989766913e-06, "loss": 1.1799, "step": 5397 }, { "epoch": 0.5743775271334326, "grad_norm": 15.390625, "learning_rate": 9.825966458214896e-06, "loss": 1.7304, "step": 5398 }, { "epoch": 0.5744839327516493, "grad_norm": 32.5625, "learning_rate": 9.825930926662878e-06, "loss": 1.2272, "step": 5399 }, { "epoch": 0.574590338369866, "grad_norm": 18.875, "learning_rate": 9.82589539511086e-06, "loss": 1.4747, "step": 5400 }, { "epoch": 0.5746967439880826, "grad_norm": 26.28125, "learning_rate": 9.825859863558841e-06, "loss": 1.319, "step": 5401 }, { "epoch": 0.5748031496062992, "grad_norm": 23.59375, "learning_rate": 9.825824332006823e-06, "loss": 1.3903, "step": 5402 }, { "epoch": 0.5749095552245158, "grad_norm": 16.203125, "learning_rate": 9.825788800454806e-06, "loss": 1.3261, "step": 5403 }, { "epoch": 0.5750159608427325, "grad_norm": 19.09375, "learning_rate": 9.825753268902788e-06, "loss": 1.3843, "step": 5404 }, { "epoch": 0.5751223664609492, "grad_norm": 12.8984375, "learning_rate": 9.825717737350769e-06, "loss": 1.2178, "step": 5405 }, { "epoch": 0.5752287720791658, "grad_norm": 11.7109375, "learning_rate": 9.82568220579875e-06, "loss": 1.446, "step": 5406 }, { "epoch": 0.5753351776973824, "grad_norm": 9.2578125, "learning_rate": 9.825646674246732e-06, "loss": 1.1929, "step": 5407 }, { "epoch": 0.575441583315599, "grad_norm": 15.1953125, "learning_rate": 9.825611142694714e-06, "loss": 1.4075, "step": 5408 }, { "epoch": 0.5755479889338158, "grad_norm": 11.375, "learning_rate": 9.825575611142696e-06, "loss": 1.3975, "step": 5409 }, { "epoch": 0.5756543945520324, "grad_norm": 14.1875, "learning_rate": 9.825540079590677e-06, "loss": 1.2712, "step": 5410 }, { "epoch": 0.575760800170249, "grad_norm": 12.5234375, "learning_rate": 9.825504548038659e-06, "loss": 1.0866, "step": 5411 }, { "epoch": 0.5758672057884656, "grad_norm": 14.4609375, "learning_rate": 9.82546901648664e-06, "loss": 1.135, "step": 5412 }, { "epoch": 0.5759736114066822, "grad_norm": 15.9296875, "learning_rate": 9.825433484934622e-06, "loss": 1.3732, "step": 5413 }, { "epoch": 0.5760800170248989, "grad_norm": 13.296875, "learning_rate": 9.825397953382605e-06, "loss": 1.2024, "step": 5414 }, { "epoch": 0.5761864226431156, "grad_norm": 10.96875, "learning_rate": 9.825362421830587e-06, "loss": 1.2858, "step": 5415 }, { "epoch": 0.5762928282613322, "grad_norm": 9.3046875, "learning_rate": 9.825326890278568e-06, "loss": 1.0658, "step": 5416 }, { "epoch": 0.5763992338795488, "grad_norm": 13.453125, "learning_rate": 9.82529135872655e-06, "loss": 1.3829, "step": 5417 }, { "epoch": 0.5765056394977655, "grad_norm": 19.625, "learning_rate": 9.825255827174532e-06, "loss": 0.9758, "step": 5418 }, { "epoch": 0.5766120451159821, "grad_norm": 13.6484375, "learning_rate": 9.825220295622513e-06, "loss": 0.9563, "step": 5419 }, { "epoch": 0.5767184507341988, "grad_norm": 12.859375, "learning_rate": 9.825184764070495e-06, "loss": 1.3732, "step": 5420 }, { "epoch": 0.5768248563524154, "grad_norm": 11.734375, "learning_rate": 9.825149232518477e-06, "loss": 1.1185, "step": 5421 }, { "epoch": 0.576931261970632, "grad_norm": 13.921875, "learning_rate": 9.825113700966458e-06, "loss": 1.2342, "step": 5422 }, { "epoch": 0.5770376675888487, "grad_norm": 15.0625, "learning_rate": 9.82507816941444e-06, "loss": 1.3762, "step": 5423 }, { "epoch": 0.5771440732070653, "grad_norm": 13.78125, "learning_rate": 9.825042637862423e-06, "loss": 1.2234, "step": 5424 }, { "epoch": 0.577250478825282, "grad_norm": 18.5625, "learning_rate": 9.825007106310405e-06, "loss": 1.2306, "step": 5425 }, { "epoch": 0.5773568844434986, "grad_norm": 14.6796875, "learning_rate": 9.824971574758386e-06, "loss": 1.3905, "step": 5426 }, { "epoch": 0.5774632900617153, "grad_norm": 12.734375, "learning_rate": 9.824936043206368e-06, "loss": 1.1113, "step": 5427 }, { "epoch": 0.5775696956799319, "grad_norm": 10.609375, "learning_rate": 9.82490051165435e-06, "loss": 1.0446, "step": 5428 }, { "epoch": 0.5776761012981485, "grad_norm": 11.4921875, "learning_rate": 9.824864980102333e-06, "loss": 1.1481, "step": 5429 }, { "epoch": 0.5777825069163652, "grad_norm": 12.8203125, "learning_rate": 9.824829448550314e-06, "loss": 1.3419, "step": 5430 }, { "epoch": 0.5778889125345819, "grad_norm": 14.8515625, "learning_rate": 9.824793916998296e-06, "loss": 1.577, "step": 5431 }, { "epoch": 0.5779953181527985, "grad_norm": 11.53125, "learning_rate": 9.824758385446278e-06, "loss": 1.2195, "step": 5432 }, { "epoch": 0.5781017237710151, "grad_norm": 16.8125, "learning_rate": 9.82472285389426e-06, "loss": 1.3743, "step": 5433 }, { "epoch": 0.5782081293892317, "grad_norm": 14.3828125, "learning_rate": 9.82468732234224e-06, "loss": 1.1799, "step": 5434 }, { "epoch": 0.5783145350074483, "grad_norm": 13.0078125, "learning_rate": 9.824651790790222e-06, "loss": 1.3827, "step": 5435 }, { "epoch": 0.5784209406256651, "grad_norm": 16.5, "learning_rate": 9.824616259238204e-06, "loss": 1.3518, "step": 5436 }, { "epoch": 0.5785273462438817, "grad_norm": 11.6953125, "learning_rate": 9.824580727686186e-06, "loss": 1.1911, "step": 5437 }, { "epoch": 0.5786337518620983, "grad_norm": 14.140625, "learning_rate": 9.824545196134167e-06, "loss": 1.0294, "step": 5438 }, { "epoch": 0.5787401574803149, "grad_norm": 21.890625, "learning_rate": 9.82450966458215e-06, "loss": 1.0719, "step": 5439 }, { "epoch": 0.5788465630985316, "grad_norm": 13.203125, "learning_rate": 9.824474133030132e-06, "loss": 1.1971, "step": 5440 }, { "epoch": 0.5789529687167483, "grad_norm": 12.515625, "learning_rate": 9.824438601478114e-06, "loss": 1.0989, "step": 5441 }, { "epoch": 0.5790593743349649, "grad_norm": 11.6875, "learning_rate": 9.824403069926095e-06, "loss": 1.3844, "step": 5442 }, { "epoch": 0.5791657799531815, "grad_norm": 13.6171875, "learning_rate": 9.824367538374077e-06, "loss": 1.383, "step": 5443 }, { "epoch": 0.5792721855713981, "grad_norm": 12.1796875, "learning_rate": 9.824332006822059e-06, "loss": 0.8768, "step": 5444 }, { "epoch": 0.5793785911896148, "grad_norm": 14.2890625, "learning_rate": 9.82429647527004e-06, "loss": 1.3541, "step": 5445 }, { "epoch": 0.5794849968078315, "grad_norm": 13.9921875, "learning_rate": 9.824260943718022e-06, "loss": 1.1839, "step": 5446 }, { "epoch": 0.5795914024260481, "grad_norm": 12.0703125, "learning_rate": 9.824225412166003e-06, "loss": 1.2129, "step": 5447 }, { "epoch": 0.5796978080442647, "grad_norm": 11.1484375, "learning_rate": 9.824189880613985e-06, "loss": 0.9419, "step": 5448 }, { "epoch": 0.5798042136624814, "grad_norm": 12.2578125, "learning_rate": 9.824154349061967e-06, "loss": 1.1386, "step": 5449 }, { "epoch": 0.579910619280698, "grad_norm": 14.6640625, "learning_rate": 9.82411881750995e-06, "loss": 1.4422, "step": 5450 }, { "epoch": 0.5800170248989147, "grad_norm": 9.890625, "learning_rate": 9.824083285957932e-06, "loss": 1.08, "step": 5451 }, { "epoch": 0.5801234305171313, "grad_norm": 16.9375, "learning_rate": 9.824047754405913e-06, "loss": 1.1204, "step": 5452 }, { "epoch": 0.580229836135348, "grad_norm": 12.703125, "learning_rate": 9.824012222853895e-06, "loss": 1.1956, "step": 5453 }, { "epoch": 0.5803362417535646, "grad_norm": 12.140625, "learning_rate": 9.823976691301876e-06, "loss": 1.2835, "step": 5454 }, { "epoch": 0.5804426473717812, "grad_norm": 15.0546875, "learning_rate": 9.82394115974986e-06, "loss": 1.3764, "step": 5455 }, { "epoch": 0.5805490529899979, "grad_norm": 18.875, "learning_rate": 9.823905628197841e-06, "loss": 1.3427, "step": 5456 }, { "epoch": 0.5806554586082145, "grad_norm": 13.6953125, "learning_rate": 9.823870096645823e-06, "loss": 1.4552, "step": 5457 }, { "epoch": 0.5807618642264312, "grad_norm": 15.65625, "learning_rate": 9.823834565093804e-06, "loss": 1.2476, "step": 5458 }, { "epoch": 0.5808682698446478, "grad_norm": 11.4765625, "learning_rate": 9.823799033541786e-06, "loss": 1.084, "step": 5459 }, { "epoch": 0.5809746754628644, "grad_norm": 10.4375, "learning_rate": 9.823763501989768e-06, "loss": 1.0717, "step": 5460 }, { "epoch": 0.581081081081081, "grad_norm": 13.640625, "learning_rate": 9.82372797043775e-06, "loss": 1.423, "step": 5461 }, { "epoch": 0.5811874866992978, "grad_norm": 13.7265625, "learning_rate": 9.823692438885731e-06, "loss": 0.9963, "step": 5462 }, { "epoch": 0.5812938923175144, "grad_norm": 16.5625, "learning_rate": 9.823656907333712e-06, "loss": 1.189, "step": 5463 }, { "epoch": 0.581400297935731, "grad_norm": 13.7890625, "learning_rate": 9.823621375781694e-06, "loss": 1.3369, "step": 5464 }, { "epoch": 0.5815067035539476, "grad_norm": 12.0390625, "learning_rate": 9.823585844229677e-06, "loss": 1.5836, "step": 5465 }, { "epoch": 0.5816131091721642, "grad_norm": 10.8046875, "learning_rate": 9.823550312677659e-06, "loss": 1.0389, "step": 5466 }, { "epoch": 0.581719514790381, "grad_norm": 14.484375, "learning_rate": 9.82351478112564e-06, "loss": 0.9117, "step": 5467 }, { "epoch": 0.5818259204085976, "grad_norm": 24.96875, "learning_rate": 9.823479249573622e-06, "loss": 2.1651, "step": 5468 }, { "epoch": 0.5819323260268142, "grad_norm": 12.4921875, "learning_rate": 9.823443718021604e-06, "loss": 1.2975, "step": 5469 }, { "epoch": 0.5820387316450308, "grad_norm": 18.546875, "learning_rate": 9.823408186469587e-06, "loss": 1.2344, "step": 5470 }, { "epoch": 0.5821451372632475, "grad_norm": 13.515625, "learning_rate": 9.823372654917567e-06, "loss": 1.1883, "step": 5471 }, { "epoch": 0.5822515428814642, "grad_norm": 16.703125, "learning_rate": 9.823337123365549e-06, "loss": 1.4181, "step": 5472 }, { "epoch": 0.5823579484996808, "grad_norm": 12.7578125, "learning_rate": 9.82330159181353e-06, "loss": 1.4827, "step": 5473 }, { "epoch": 0.5824643541178974, "grad_norm": 9.6640625, "learning_rate": 9.823266060261512e-06, "loss": 1.0837, "step": 5474 }, { "epoch": 0.582570759736114, "grad_norm": 17.640625, "learning_rate": 9.823230528709495e-06, "loss": 1.0386, "step": 5475 }, { "epoch": 0.5826771653543307, "grad_norm": 12.875, "learning_rate": 9.823194997157477e-06, "loss": 1.4312, "step": 5476 }, { "epoch": 0.5827835709725474, "grad_norm": 13.828125, "learning_rate": 9.823159465605458e-06, "loss": 1.3113, "step": 5477 }, { "epoch": 0.582889976590764, "grad_norm": 15.640625, "learning_rate": 9.82312393405344e-06, "loss": 1.4107, "step": 5478 }, { "epoch": 0.5829963822089806, "grad_norm": 10.0234375, "learning_rate": 9.823088402501422e-06, "loss": 1.0514, "step": 5479 }, { "epoch": 0.5831027878271973, "grad_norm": 9.6875, "learning_rate": 9.823052870949405e-06, "loss": 1.0532, "step": 5480 }, { "epoch": 0.5832091934454139, "grad_norm": 16.3125, "learning_rate": 9.823017339397386e-06, "loss": 1.2393, "step": 5481 }, { "epoch": 0.5833155990636305, "grad_norm": 9.828125, "learning_rate": 9.822981807845368e-06, "loss": 1.1779, "step": 5482 }, { "epoch": 0.5834220046818472, "grad_norm": 11.6640625, "learning_rate": 9.82294627629335e-06, "loss": 1.4272, "step": 5483 }, { "epoch": 0.5835284103000639, "grad_norm": 12.21875, "learning_rate": 9.822910744741331e-06, "loss": 1.1432, "step": 5484 }, { "epoch": 0.5836348159182805, "grad_norm": 14.90625, "learning_rate": 9.822875213189313e-06, "loss": 1.6344, "step": 5485 }, { "epoch": 0.5837412215364971, "grad_norm": 15.875, "learning_rate": 9.822839681637295e-06, "loss": 1.0616, "step": 5486 }, { "epoch": 0.5838476271547137, "grad_norm": 11.5390625, "learning_rate": 9.822804150085276e-06, "loss": 1.2852, "step": 5487 }, { "epoch": 0.5839540327729305, "grad_norm": 24.265625, "learning_rate": 9.822768618533258e-06, "loss": 1.4496, "step": 5488 }, { "epoch": 0.5840604383911471, "grad_norm": 12.609375, "learning_rate": 9.82273308698124e-06, "loss": 1.414, "step": 5489 }, { "epoch": 0.5841668440093637, "grad_norm": 11.109375, "learning_rate": 9.822697555429221e-06, "loss": 1.0423, "step": 5490 }, { "epoch": 0.5842732496275803, "grad_norm": 9.578125, "learning_rate": 9.822662023877204e-06, "loss": 1.0495, "step": 5491 }, { "epoch": 0.5843796552457969, "grad_norm": 10.5234375, "learning_rate": 9.822626492325186e-06, "loss": 1.2032, "step": 5492 }, { "epoch": 0.5844860608640137, "grad_norm": 12.3515625, "learning_rate": 9.822590960773167e-06, "loss": 1.0633, "step": 5493 }, { "epoch": 0.5845924664822303, "grad_norm": 11.796875, "learning_rate": 9.822555429221149e-06, "loss": 1.2783, "step": 5494 }, { "epoch": 0.5846988721004469, "grad_norm": 13.0546875, "learning_rate": 9.82251989766913e-06, "loss": 1.0464, "step": 5495 }, { "epoch": 0.5848052777186635, "grad_norm": 13.3671875, "learning_rate": 9.822484366117114e-06, "loss": 1.5613, "step": 5496 }, { "epoch": 0.5849116833368802, "grad_norm": 11.6640625, "learning_rate": 9.822448834565096e-06, "loss": 1.0511, "step": 5497 }, { "epoch": 0.5850180889550969, "grad_norm": 11.4765625, "learning_rate": 9.822413303013077e-06, "loss": 1.0807, "step": 5498 }, { "epoch": 0.5851244945733135, "grad_norm": 16.328125, "learning_rate": 9.822377771461059e-06, "loss": 1.2815, "step": 5499 }, { "epoch": 0.5852309001915301, "grad_norm": 12.8125, "learning_rate": 9.82234223990904e-06, "loss": 1.1399, "step": 5500 }, { "epoch": 0.5853373058097467, "grad_norm": 15.40625, "learning_rate": 9.822306708357022e-06, "loss": 1.5033, "step": 5501 }, { "epoch": 0.5854437114279634, "grad_norm": 19.328125, "learning_rate": 9.822271176805004e-06, "loss": 1.3014, "step": 5502 }, { "epoch": 0.58555011704618, "grad_norm": 14.140625, "learning_rate": 9.822235645252985e-06, "loss": 1.1038, "step": 5503 }, { "epoch": 0.5856565226643967, "grad_norm": 12.828125, "learning_rate": 9.822200113700967e-06, "loss": 1.5722, "step": 5504 }, { "epoch": 0.5857629282826133, "grad_norm": 18.6875, "learning_rate": 9.822164582148948e-06, "loss": 1.4337, "step": 5505 }, { "epoch": 0.58586933390083, "grad_norm": 15.1015625, "learning_rate": 9.822129050596932e-06, "loss": 1.2075, "step": 5506 }, { "epoch": 0.5859757395190466, "grad_norm": 9.7265625, "learning_rate": 9.822093519044913e-06, "loss": 1.0412, "step": 5507 }, { "epoch": 0.5860821451372632, "grad_norm": 12.6015625, "learning_rate": 9.822057987492895e-06, "loss": 0.9557, "step": 5508 }, { "epoch": 0.5861885507554799, "grad_norm": 14.6015625, "learning_rate": 9.822022455940877e-06, "loss": 1.2173, "step": 5509 }, { "epoch": 0.5862949563736966, "grad_norm": 15.1796875, "learning_rate": 9.821986924388858e-06, "loss": 1.1217, "step": 5510 }, { "epoch": 0.5864013619919132, "grad_norm": 9.875, "learning_rate": 9.82195139283684e-06, "loss": 1.25, "step": 5511 }, { "epoch": 0.5865077676101298, "grad_norm": 8.5546875, "learning_rate": 9.821915861284821e-06, "loss": 1.0804, "step": 5512 }, { "epoch": 0.5866141732283464, "grad_norm": 14.0234375, "learning_rate": 9.821880329732803e-06, "loss": 1.5292, "step": 5513 }, { "epoch": 0.5867205788465631, "grad_norm": 13.1171875, "learning_rate": 9.821844798180785e-06, "loss": 1.1214, "step": 5514 }, { "epoch": 0.5868269844647798, "grad_norm": 13.453125, "learning_rate": 9.821809266628766e-06, "loss": 1.215, "step": 5515 }, { "epoch": 0.5869333900829964, "grad_norm": 13.546875, "learning_rate": 9.82177373507675e-06, "loss": 0.9755, "step": 5516 }, { "epoch": 0.587039795701213, "grad_norm": 14.4765625, "learning_rate": 9.821738203524731e-06, "loss": 1.2396, "step": 5517 }, { "epoch": 0.5871462013194296, "grad_norm": 19.421875, "learning_rate": 9.821702671972713e-06, "loss": 1.3238, "step": 5518 }, { "epoch": 0.5872526069376464, "grad_norm": 15.5, "learning_rate": 9.821667140420694e-06, "loss": 1.1518, "step": 5519 }, { "epoch": 0.587359012555863, "grad_norm": 13.0859375, "learning_rate": 9.821631608868676e-06, "loss": 1.0658, "step": 5520 }, { "epoch": 0.5874654181740796, "grad_norm": 11.7421875, "learning_rate": 9.82159607731666e-06, "loss": 1.2062, "step": 5521 }, { "epoch": 0.5875718237922962, "grad_norm": 14.453125, "learning_rate": 9.82156054576464e-06, "loss": 1.1771, "step": 5522 }, { "epoch": 0.5876782294105128, "grad_norm": 11.6171875, "learning_rate": 9.821525014212622e-06, "loss": 1.479, "step": 5523 }, { "epoch": 0.5877846350287295, "grad_norm": 10.0390625, "learning_rate": 9.821489482660604e-06, "loss": 1.0339, "step": 5524 }, { "epoch": 0.5878910406469462, "grad_norm": 16.046875, "learning_rate": 9.821453951108586e-06, "loss": 1.911, "step": 5525 }, { "epoch": 0.5879974462651628, "grad_norm": 15.15625, "learning_rate": 9.821418419556567e-06, "loss": 1.3456, "step": 5526 }, { "epoch": 0.5881038518833794, "grad_norm": 17.703125, "learning_rate": 9.821382888004549e-06, "loss": 1.1349, "step": 5527 }, { "epoch": 0.5882102575015961, "grad_norm": 11.171875, "learning_rate": 9.82134735645253e-06, "loss": 1.2977, "step": 5528 }, { "epoch": 0.5883166631198127, "grad_norm": 14.6015625, "learning_rate": 9.821311824900512e-06, "loss": 1.1607, "step": 5529 }, { "epoch": 0.5884230687380294, "grad_norm": 12.546875, "learning_rate": 9.821276293348494e-06, "loss": 1.0396, "step": 5530 }, { "epoch": 0.588529474356246, "grad_norm": 13.71875, "learning_rate": 9.821240761796475e-06, "loss": 1.2186, "step": 5531 }, { "epoch": 0.5886358799744627, "grad_norm": 18.1875, "learning_rate": 9.821205230244459e-06, "loss": 1.4001, "step": 5532 }, { "epoch": 0.5887422855926793, "grad_norm": 13.1796875, "learning_rate": 9.82116969869244e-06, "loss": 1.3758, "step": 5533 }, { "epoch": 0.5888486912108959, "grad_norm": 12.46875, "learning_rate": 9.821134167140422e-06, "loss": 1.0672, "step": 5534 }, { "epoch": 0.5889550968291126, "grad_norm": 13.7578125, "learning_rate": 9.821098635588403e-06, "loss": 1.5715, "step": 5535 }, { "epoch": 0.5890615024473292, "grad_norm": 21.203125, "learning_rate": 9.821063104036385e-06, "loss": 1.191, "step": 5536 }, { "epoch": 0.5891679080655459, "grad_norm": 17.4375, "learning_rate": 9.821027572484367e-06, "loss": 1.6221, "step": 5537 }, { "epoch": 0.5892743136837625, "grad_norm": 14.6015625, "learning_rate": 9.820992040932348e-06, "loss": 1.2316, "step": 5538 }, { "epoch": 0.5893807193019791, "grad_norm": 11.75, "learning_rate": 9.82095650938033e-06, "loss": 1.2307, "step": 5539 }, { "epoch": 0.5894871249201958, "grad_norm": 10.75, "learning_rate": 9.820920977828311e-06, "loss": 1.2714, "step": 5540 }, { "epoch": 0.5895935305384125, "grad_norm": 10.203125, "learning_rate": 9.820885446276293e-06, "loss": 0.8822, "step": 5541 }, { "epoch": 0.5896999361566291, "grad_norm": 10.109375, "learning_rate": 9.820849914724276e-06, "loss": 1.1267, "step": 5542 }, { "epoch": 0.5898063417748457, "grad_norm": 18.640625, "learning_rate": 9.820814383172258e-06, "loss": 1.879, "step": 5543 }, { "epoch": 0.5899127473930623, "grad_norm": 19.125, "learning_rate": 9.82077885162024e-06, "loss": 1.1036, "step": 5544 }, { "epoch": 0.5900191530112789, "grad_norm": 13.109375, "learning_rate": 9.820743320068221e-06, "loss": 1.1264, "step": 5545 }, { "epoch": 0.5901255586294957, "grad_norm": 11.0859375, "learning_rate": 9.820707788516203e-06, "loss": 0.9855, "step": 5546 }, { "epoch": 0.5902319642477123, "grad_norm": 12.328125, "learning_rate": 9.820672256964186e-06, "loss": 1.2519, "step": 5547 }, { "epoch": 0.5903383698659289, "grad_norm": 15.765625, "learning_rate": 9.820636725412168e-06, "loss": 1.6681, "step": 5548 }, { "epoch": 0.5904447754841455, "grad_norm": 12.6328125, "learning_rate": 9.82060119386015e-06, "loss": 1.2398, "step": 5549 }, { "epoch": 0.5905511811023622, "grad_norm": 104.5625, "learning_rate": 9.820565662308131e-06, "loss": 0.831, "step": 5550 }, { "epoch": 0.5906575867205789, "grad_norm": 12.4921875, "learning_rate": 9.820530130756112e-06, "loss": 1.2176, "step": 5551 }, { "epoch": 0.5907639923387955, "grad_norm": 11.5390625, "learning_rate": 9.820494599204094e-06, "loss": 1.281, "step": 5552 }, { "epoch": 0.5908703979570121, "grad_norm": 16.796875, "learning_rate": 9.820459067652076e-06, "loss": 1.2444, "step": 5553 }, { "epoch": 0.5909768035752287, "grad_norm": 16.4375, "learning_rate": 9.820423536100057e-06, "loss": 1.4733, "step": 5554 }, { "epoch": 0.5910832091934454, "grad_norm": 16.234375, "learning_rate": 9.820388004548039e-06, "loss": 1.1169, "step": 5555 }, { "epoch": 0.5911896148116621, "grad_norm": 10.71875, "learning_rate": 9.82035247299602e-06, "loss": 1.1919, "step": 5556 }, { "epoch": 0.5912960204298787, "grad_norm": 22.5, "learning_rate": 9.820316941444004e-06, "loss": 1.1111, "step": 5557 }, { "epoch": 0.5914024260480953, "grad_norm": 14.015625, "learning_rate": 9.820281409891985e-06, "loss": 1.1896, "step": 5558 }, { "epoch": 0.591508831666312, "grad_norm": 12.3359375, "learning_rate": 9.820245878339967e-06, "loss": 1.6755, "step": 5559 }, { "epoch": 0.5916152372845286, "grad_norm": 12.71875, "learning_rate": 9.820210346787949e-06, "loss": 1.2404, "step": 5560 }, { "epoch": 0.5917216429027453, "grad_norm": 12.890625, "learning_rate": 9.82017481523593e-06, "loss": 1.083, "step": 5561 }, { "epoch": 0.5918280485209619, "grad_norm": 11.75, "learning_rate": 9.820139283683912e-06, "loss": 0.8679, "step": 5562 }, { "epoch": 0.5919344541391786, "grad_norm": 8.828125, "learning_rate": 9.820103752131893e-06, "loss": 0.9254, "step": 5563 }, { "epoch": 0.5920408597573952, "grad_norm": 13.5234375, "learning_rate": 9.820068220579875e-06, "loss": 1.0128, "step": 5564 }, { "epoch": 0.5921472653756118, "grad_norm": 11.828125, "learning_rate": 9.820032689027857e-06, "loss": 1.1924, "step": 5565 }, { "epoch": 0.5922536709938285, "grad_norm": 12.3203125, "learning_rate": 9.819997157475838e-06, "loss": 0.9638, "step": 5566 }, { "epoch": 0.5923600766120451, "grad_norm": 9.859375, "learning_rate": 9.81996162592382e-06, "loss": 1.1374, "step": 5567 }, { "epoch": 0.5924664822302618, "grad_norm": 11.1484375, "learning_rate": 9.819926094371803e-06, "loss": 1.0335, "step": 5568 }, { "epoch": 0.5925728878484784, "grad_norm": 17.359375, "learning_rate": 9.819890562819785e-06, "loss": 1.4395, "step": 5569 }, { "epoch": 0.592679293466695, "grad_norm": 13.859375, "learning_rate": 9.819855031267766e-06, "loss": 1.2976, "step": 5570 }, { "epoch": 0.5927856990849116, "grad_norm": 11.578125, "learning_rate": 9.819819499715748e-06, "loss": 1.2027, "step": 5571 }, { "epoch": 0.5928921047031284, "grad_norm": 11.59375, "learning_rate": 9.81978396816373e-06, "loss": 0.9491, "step": 5572 }, { "epoch": 0.592998510321345, "grad_norm": 11.6171875, "learning_rate": 9.819748436611713e-06, "loss": 1.2277, "step": 5573 }, { "epoch": 0.5931049159395616, "grad_norm": 21.28125, "learning_rate": 9.819712905059695e-06, "loss": 1.6562, "step": 5574 }, { "epoch": 0.5932113215577782, "grad_norm": 15.8125, "learning_rate": 9.819677373507676e-06, "loss": 1.4515, "step": 5575 }, { "epoch": 0.5933177271759948, "grad_norm": 15.90625, "learning_rate": 9.819641841955658e-06, "loss": 1.6682, "step": 5576 }, { "epoch": 0.5934241327942116, "grad_norm": 13.0, "learning_rate": 9.81960631040364e-06, "loss": 1.0225, "step": 5577 }, { "epoch": 0.5935305384124282, "grad_norm": 14.9453125, "learning_rate": 9.819570778851621e-06, "loss": 1.1577, "step": 5578 }, { "epoch": 0.5936369440306448, "grad_norm": 10.1953125, "learning_rate": 9.819535247299603e-06, "loss": 1.1681, "step": 5579 }, { "epoch": 0.5937433496488614, "grad_norm": 21.421875, "learning_rate": 9.819499715747584e-06, "loss": 1.3215, "step": 5580 }, { "epoch": 0.5938497552670781, "grad_norm": 12.0078125, "learning_rate": 9.819464184195566e-06, "loss": 1.1016, "step": 5581 }, { "epoch": 0.5939561608852948, "grad_norm": 11.859375, "learning_rate": 9.819428652643547e-06, "loss": 1.4415, "step": 5582 }, { "epoch": 0.5940625665035114, "grad_norm": 12.671875, "learning_rate": 9.81939312109153e-06, "loss": 1.0397, "step": 5583 }, { "epoch": 0.594168972121728, "grad_norm": 11.15625, "learning_rate": 9.819357589539512e-06, "loss": 1.1851, "step": 5584 }, { "epoch": 0.5942753777399447, "grad_norm": 11.484375, "learning_rate": 9.819322057987494e-06, "loss": 1.0211, "step": 5585 }, { "epoch": 0.5943817833581613, "grad_norm": 25.578125, "learning_rate": 9.819286526435476e-06, "loss": 1.4867, "step": 5586 }, { "epoch": 0.594488188976378, "grad_norm": 10.59375, "learning_rate": 9.819250994883457e-06, "loss": 1.0488, "step": 5587 }, { "epoch": 0.5945945945945946, "grad_norm": 11.7421875, "learning_rate": 9.81921546333144e-06, "loss": 1.1232, "step": 5588 }, { "epoch": 0.5947010002128112, "grad_norm": 10.9453125, "learning_rate": 9.819179931779422e-06, "loss": 1.0692, "step": 5589 }, { "epoch": 0.5948074058310279, "grad_norm": 19.53125, "learning_rate": 9.819144400227404e-06, "loss": 1.7883, "step": 5590 }, { "epoch": 0.5949138114492445, "grad_norm": 16.640625, "learning_rate": 9.819108868675384e-06, "loss": 1.1458, "step": 5591 }, { "epoch": 0.5950202170674611, "grad_norm": 12.6796875, "learning_rate": 9.819073337123365e-06, "loss": 1.3851, "step": 5592 }, { "epoch": 0.5951266226856778, "grad_norm": 14.84375, "learning_rate": 9.819037805571348e-06, "loss": 1.6865, "step": 5593 }, { "epoch": 0.5952330283038945, "grad_norm": 10.40625, "learning_rate": 9.81900227401933e-06, "loss": 1.3394, "step": 5594 }, { "epoch": 0.5953394339221111, "grad_norm": 15.0625, "learning_rate": 9.818966742467312e-06, "loss": 1.511, "step": 5595 }, { "epoch": 0.5954458395403277, "grad_norm": 13.9296875, "learning_rate": 9.818931210915293e-06, "loss": 1.3055, "step": 5596 }, { "epoch": 0.5955522451585443, "grad_norm": 13.4921875, "learning_rate": 9.818895679363275e-06, "loss": 1.4523, "step": 5597 }, { "epoch": 0.5956586507767611, "grad_norm": 12.78125, "learning_rate": 9.818860147811258e-06, "loss": 1.409, "step": 5598 }, { "epoch": 0.5957650563949777, "grad_norm": 12.640625, "learning_rate": 9.81882461625924e-06, "loss": 1.3065, "step": 5599 }, { "epoch": 0.5958714620131943, "grad_norm": 10.984375, "learning_rate": 9.818789084707221e-06, "loss": 1.3633, "step": 5600 }, { "epoch": 0.5959778676314109, "grad_norm": 12.34375, "learning_rate": 9.818753553155203e-06, "loss": 1.4046, "step": 5601 }, { "epoch": 0.5960842732496275, "grad_norm": 13.2734375, "learning_rate": 9.818718021603185e-06, "loss": 1.4548, "step": 5602 }, { "epoch": 0.5961906788678443, "grad_norm": 14.6640625, "learning_rate": 9.818682490051166e-06, "loss": 1.4148, "step": 5603 }, { "epoch": 0.5962970844860609, "grad_norm": 18.21875, "learning_rate": 9.818646958499148e-06, "loss": 1.157, "step": 5604 }, { "epoch": 0.5964034901042775, "grad_norm": 12.40625, "learning_rate": 9.81861142694713e-06, "loss": 1.5607, "step": 5605 }, { "epoch": 0.5965098957224941, "grad_norm": 12.75, "learning_rate": 9.818575895395111e-06, "loss": 1.2623, "step": 5606 }, { "epoch": 0.5966163013407108, "grad_norm": 12.6953125, "learning_rate": 9.818540363843093e-06, "loss": 1.3304, "step": 5607 }, { "epoch": 0.5967227069589275, "grad_norm": 13.2890625, "learning_rate": 9.818504832291074e-06, "loss": 1.1228, "step": 5608 }, { "epoch": 0.5968291125771441, "grad_norm": 12.7578125, "learning_rate": 9.818469300739058e-06, "loss": 1.1653, "step": 5609 }, { "epoch": 0.5969355181953607, "grad_norm": 12.0703125, "learning_rate": 9.818433769187039e-06, "loss": 0.9368, "step": 5610 }, { "epoch": 0.5970419238135773, "grad_norm": 13.1953125, "learning_rate": 9.81839823763502e-06, "loss": 1.5468, "step": 5611 }, { "epoch": 0.597148329431794, "grad_norm": 14.1640625, "learning_rate": 9.818362706083002e-06, "loss": 1.1651, "step": 5612 }, { "epoch": 0.5972547350500106, "grad_norm": 12.328125, "learning_rate": 9.818327174530984e-06, "loss": 1.1983, "step": 5613 }, { "epoch": 0.5973611406682273, "grad_norm": 11.6015625, "learning_rate": 9.818291642978967e-06, "loss": 1.0776, "step": 5614 }, { "epoch": 0.5974675462864439, "grad_norm": 13.609375, "learning_rate": 9.818256111426949e-06, "loss": 0.9947, "step": 5615 }, { "epoch": 0.5975739519046606, "grad_norm": 13.3828125, "learning_rate": 9.81822057987493e-06, "loss": 1.1398, "step": 5616 }, { "epoch": 0.5976803575228772, "grad_norm": 12.3828125, "learning_rate": 9.818185048322912e-06, "loss": 1.0801, "step": 5617 }, { "epoch": 0.5977867631410938, "grad_norm": 10.7890625, "learning_rate": 9.818149516770894e-06, "loss": 1.443, "step": 5618 }, { "epoch": 0.5978931687593105, "grad_norm": 11.6796875, "learning_rate": 9.818113985218875e-06, "loss": 1.2916, "step": 5619 }, { "epoch": 0.5979995743775272, "grad_norm": 15.8984375, "learning_rate": 9.818078453666857e-06, "loss": 1.2652, "step": 5620 }, { "epoch": 0.5981059799957438, "grad_norm": 11.3125, "learning_rate": 9.818042922114839e-06, "loss": 1.1131, "step": 5621 }, { "epoch": 0.5982123856139604, "grad_norm": 13.9609375, "learning_rate": 9.81800739056282e-06, "loss": 1.2193, "step": 5622 }, { "epoch": 0.598318791232177, "grad_norm": 13.3515625, "learning_rate": 9.817971859010802e-06, "loss": 0.9459, "step": 5623 }, { "epoch": 0.5984251968503937, "grad_norm": 11.7734375, "learning_rate": 9.817936327458785e-06, "loss": 1.0977, "step": 5624 }, { "epoch": 0.5985316024686104, "grad_norm": 12.9140625, "learning_rate": 9.817900795906767e-06, "loss": 1.1929, "step": 5625 }, { "epoch": 0.598638008086827, "grad_norm": 13.71875, "learning_rate": 9.817865264354748e-06, "loss": 1.2694, "step": 5626 }, { "epoch": 0.5987444137050436, "grad_norm": 15.8984375, "learning_rate": 9.81782973280273e-06, "loss": 1.2503, "step": 5627 }, { "epoch": 0.5988508193232602, "grad_norm": 12.0390625, "learning_rate": 9.817794201250711e-06, "loss": 1.2353, "step": 5628 }, { "epoch": 0.598957224941477, "grad_norm": 13.171875, "learning_rate": 9.817758669698693e-06, "loss": 1.2218, "step": 5629 }, { "epoch": 0.5990636305596936, "grad_norm": 13.203125, "learning_rate": 9.817723138146675e-06, "loss": 1.2864, "step": 5630 }, { "epoch": 0.5991700361779102, "grad_norm": 11.96875, "learning_rate": 9.817687606594656e-06, "loss": 1.3212, "step": 5631 }, { "epoch": 0.5992764417961268, "grad_norm": 12.09375, "learning_rate": 9.817652075042638e-06, "loss": 1.1943, "step": 5632 }, { "epoch": 0.5993828474143434, "grad_norm": 10.3671875, "learning_rate": 9.81761654349062e-06, "loss": 1.2102, "step": 5633 }, { "epoch": 0.5994892530325601, "grad_norm": 10.9375, "learning_rate": 9.817581011938603e-06, "loss": 1.032, "step": 5634 }, { "epoch": 0.5995956586507768, "grad_norm": 19.734375, "learning_rate": 9.817545480386584e-06, "loss": 1.8191, "step": 5635 }, { "epoch": 0.5997020642689934, "grad_norm": 20.15625, "learning_rate": 9.817509948834566e-06, "loss": 1.0414, "step": 5636 }, { "epoch": 0.59980846988721, "grad_norm": 18.296875, "learning_rate": 9.817474417282548e-06, "loss": 1.2767, "step": 5637 }, { "epoch": 0.5999148755054267, "grad_norm": 13.734375, "learning_rate": 9.81743888573053e-06, "loss": 0.8824, "step": 5638 }, { "epoch": 0.6000212811236433, "grad_norm": 14.984375, "learning_rate": 9.817403354178513e-06, "loss": 1.4547, "step": 5639 }, { "epoch": 0.60012768674186, "grad_norm": 15.84375, "learning_rate": 9.817367822626494e-06, "loss": 1.0344, "step": 5640 }, { "epoch": 0.6002340923600766, "grad_norm": 18.5625, "learning_rate": 9.817332291074476e-06, "loss": 1.3005, "step": 5641 }, { "epoch": 0.6003404979782933, "grad_norm": 11.6953125, "learning_rate": 9.817296759522457e-06, "loss": 1.0978, "step": 5642 }, { "epoch": 0.6004469035965099, "grad_norm": 11.96875, "learning_rate": 9.817261227970439e-06, "loss": 1.2604, "step": 5643 }, { "epoch": 0.6005533092147265, "grad_norm": 10.515625, "learning_rate": 9.81722569641842e-06, "loss": 0.9901, "step": 5644 }, { "epoch": 0.6006597148329432, "grad_norm": 11.03125, "learning_rate": 9.817190164866402e-06, "loss": 1.3268, "step": 5645 }, { "epoch": 0.6007661204511598, "grad_norm": 13.8828125, "learning_rate": 9.817154633314384e-06, "loss": 1.28, "step": 5646 }, { "epoch": 0.6008725260693765, "grad_norm": 13.7890625, "learning_rate": 9.817119101762365e-06, "loss": 1.3318, "step": 5647 }, { "epoch": 0.6009789316875931, "grad_norm": 8.53125, "learning_rate": 9.817083570210347e-06, "loss": 0.797, "step": 5648 }, { "epoch": 0.6010853373058097, "grad_norm": 14.7890625, "learning_rate": 9.817048038658329e-06, "loss": 0.9668, "step": 5649 }, { "epoch": 0.6011917429240264, "grad_norm": 15.390625, "learning_rate": 9.817012507106312e-06, "loss": 1.3547, "step": 5650 }, { "epoch": 0.6012981485422431, "grad_norm": 16.40625, "learning_rate": 9.816976975554293e-06, "loss": 1.4542, "step": 5651 }, { "epoch": 0.6014045541604597, "grad_norm": 32.9375, "learning_rate": 9.816941444002275e-06, "loss": 1.2384, "step": 5652 }, { "epoch": 0.6015109597786763, "grad_norm": 21.328125, "learning_rate": 9.816905912450257e-06, "loss": 1.3272, "step": 5653 }, { "epoch": 0.6016173653968929, "grad_norm": 17.625, "learning_rate": 9.816870380898238e-06, "loss": 1.0066, "step": 5654 }, { "epoch": 0.6017237710151095, "grad_norm": 15.140625, "learning_rate": 9.81683484934622e-06, "loss": 1.8669, "step": 5655 }, { "epoch": 0.6018301766333263, "grad_norm": 13.0546875, "learning_rate": 9.816799317794202e-06, "loss": 1.2445, "step": 5656 }, { "epoch": 0.6019365822515429, "grad_norm": 12.1015625, "learning_rate": 9.816763786242183e-06, "loss": 1.6506, "step": 5657 }, { "epoch": 0.6020429878697595, "grad_norm": 11.953125, "learning_rate": 9.816728254690165e-06, "loss": 0.9458, "step": 5658 }, { "epoch": 0.6021493934879761, "grad_norm": 10.390625, "learning_rate": 9.816692723138146e-06, "loss": 1.2621, "step": 5659 }, { "epoch": 0.6022557991061928, "grad_norm": 9.9921875, "learning_rate": 9.81665719158613e-06, "loss": 1.1408, "step": 5660 }, { "epoch": 0.6023622047244095, "grad_norm": 15.203125, "learning_rate": 9.816621660034111e-06, "loss": 1.1231, "step": 5661 }, { "epoch": 0.6024686103426261, "grad_norm": 14.5546875, "learning_rate": 9.816586128482093e-06, "loss": 1.3686, "step": 5662 }, { "epoch": 0.6025750159608427, "grad_norm": 12.4921875, "learning_rate": 9.816550596930074e-06, "loss": 1.1952, "step": 5663 }, { "epoch": 0.6026814215790594, "grad_norm": 18.875, "learning_rate": 9.816515065378056e-06, "loss": 1.1705, "step": 5664 }, { "epoch": 0.602787827197276, "grad_norm": 11.5546875, "learning_rate": 9.81647953382604e-06, "loss": 1.2707, "step": 5665 }, { "epoch": 0.6028942328154927, "grad_norm": 14.6796875, "learning_rate": 9.816444002274021e-06, "loss": 0.9115, "step": 5666 }, { "epoch": 0.6030006384337093, "grad_norm": 10.140625, "learning_rate": 9.816408470722003e-06, "loss": 1.4426, "step": 5667 }, { "epoch": 0.603107044051926, "grad_norm": 10.7890625, "learning_rate": 9.816372939169984e-06, "loss": 1.2228, "step": 5668 }, { "epoch": 0.6032134496701426, "grad_norm": 11.921875, "learning_rate": 9.816337407617966e-06, "loss": 1.3715, "step": 5669 }, { "epoch": 0.6033198552883592, "grad_norm": 10.859375, "learning_rate": 9.816301876065947e-06, "loss": 1.2315, "step": 5670 }, { "epoch": 0.6034262609065759, "grad_norm": 11.7890625, "learning_rate": 9.816266344513929e-06, "loss": 1.6329, "step": 5671 }, { "epoch": 0.6035326665247925, "grad_norm": 11.953125, "learning_rate": 9.81623081296191e-06, "loss": 1.3139, "step": 5672 }, { "epoch": 0.6036390721430092, "grad_norm": 10.03125, "learning_rate": 9.816195281409892e-06, "loss": 1.4666, "step": 5673 }, { "epoch": 0.6037454777612258, "grad_norm": 10.75, "learning_rate": 9.816159749857874e-06, "loss": 1.4837, "step": 5674 }, { "epoch": 0.6038518833794424, "grad_norm": 10.8515625, "learning_rate": 9.816124218305857e-06, "loss": 1.5051, "step": 5675 }, { "epoch": 0.6039582889976591, "grad_norm": 16.375, "learning_rate": 9.816088686753839e-06, "loss": 1.3399, "step": 5676 }, { "epoch": 0.6040646946158758, "grad_norm": 9.9921875, "learning_rate": 9.81605315520182e-06, "loss": 1.154, "step": 5677 }, { "epoch": 0.6041711002340924, "grad_norm": 14.015625, "learning_rate": 9.816017623649802e-06, "loss": 1.3244, "step": 5678 }, { "epoch": 0.604277505852309, "grad_norm": 13.359375, "learning_rate": 9.815982092097784e-06, "loss": 1.3545, "step": 5679 }, { "epoch": 0.6043839114705256, "grad_norm": 12.2578125, "learning_rate": 9.815946560545765e-06, "loss": 1.1339, "step": 5680 }, { "epoch": 0.6044903170887422, "grad_norm": 13.9609375, "learning_rate": 9.815911028993747e-06, "loss": 1.389, "step": 5681 }, { "epoch": 0.604596722706959, "grad_norm": 10.015625, "learning_rate": 9.815875497441728e-06, "loss": 1.3483, "step": 5682 }, { "epoch": 0.6047031283251756, "grad_norm": 14.2421875, "learning_rate": 9.81583996588971e-06, "loss": 1.3272, "step": 5683 }, { "epoch": 0.6048095339433922, "grad_norm": 16.71875, "learning_rate": 9.815804434337692e-06, "loss": 1.2831, "step": 5684 }, { "epoch": 0.6049159395616088, "grad_norm": 16.9375, "learning_rate": 9.815768902785673e-06, "loss": 1.0493, "step": 5685 }, { "epoch": 0.6050223451798254, "grad_norm": 14.15625, "learning_rate": 9.815733371233657e-06, "loss": 1.2041, "step": 5686 }, { "epoch": 0.6051287507980422, "grad_norm": 15.3515625, "learning_rate": 9.815697839681638e-06, "loss": 1.6029, "step": 5687 }, { "epoch": 0.6052351564162588, "grad_norm": 15.203125, "learning_rate": 9.81566230812962e-06, "loss": 1.7567, "step": 5688 }, { "epoch": 0.6053415620344754, "grad_norm": 13.0546875, "learning_rate": 9.815626776577601e-06, "loss": 1.5818, "step": 5689 }, { "epoch": 0.605447967652692, "grad_norm": 19.109375, "learning_rate": 9.815591245025583e-06, "loss": 1.7385, "step": 5690 }, { "epoch": 0.6055543732709087, "grad_norm": 15.8671875, "learning_rate": 9.815555713473566e-06, "loss": 1.8397, "step": 5691 }, { "epoch": 0.6056607788891254, "grad_norm": 11.71875, "learning_rate": 9.815520181921548e-06, "loss": 1.22, "step": 5692 }, { "epoch": 0.605767184507342, "grad_norm": 14.75, "learning_rate": 9.81548465036953e-06, "loss": 1.2948, "step": 5693 }, { "epoch": 0.6058735901255586, "grad_norm": 15.0859375, "learning_rate": 9.815449118817511e-06, "loss": 1.0182, "step": 5694 }, { "epoch": 0.6059799957437753, "grad_norm": 16.328125, "learning_rate": 9.815413587265493e-06, "loss": 1.3289, "step": 5695 }, { "epoch": 0.6060864013619919, "grad_norm": 17.3125, "learning_rate": 9.815378055713474e-06, "loss": 1.3849, "step": 5696 }, { "epoch": 0.6061928069802086, "grad_norm": 12.6796875, "learning_rate": 9.815342524161456e-06, "loss": 0.9924, "step": 5697 }, { "epoch": 0.6062992125984252, "grad_norm": 14.2265625, "learning_rate": 9.815306992609437e-06, "loss": 1.2747, "step": 5698 }, { "epoch": 0.6064056182166419, "grad_norm": 15.140625, "learning_rate": 9.815271461057419e-06, "loss": 1.2858, "step": 5699 }, { "epoch": 0.6065120238348585, "grad_norm": 14.5234375, "learning_rate": 9.8152359295054e-06, "loss": 1.059, "step": 5700 }, { "epoch": 0.6066184294530751, "grad_norm": 11.9296875, "learning_rate": 9.815200397953384e-06, "loss": 1.0362, "step": 5701 }, { "epoch": 0.6067248350712917, "grad_norm": 12.765625, "learning_rate": 9.815164866401366e-06, "loss": 1.5516, "step": 5702 }, { "epoch": 0.6068312406895084, "grad_norm": 15.6796875, "learning_rate": 9.815129334849347e-06, "loss": 1.4292, "step": 5703 }, { "epoch": 0.6069376463077251, "grad_norm": 10.4296875, "learning_rate": 9.815093803297329e-06, "loss": 1.1345, "step": 5704 }, { "epoch": 0.6070440519259417, "grad_norm": 16.234375, "learning_rate": 9.81505827174531e-06, "loss": 1.6627, "step": 5705 }, { "epoch": 0.6071504575441583, "grad_norm": 16.125, "learning_rate": 9.815022740193294e-06, "loss": 1.2779, "step": 5706 }, { "epoch": 0.6072568631623749, "grad_norm": 15.5, "learning_rate": 9.814987208641275e-06, "loss": 1.2279, "step": 5707 }, { "epoch": 0.6073632687805917, "grad_norm": 12.53125, "learning_rate": 9.814951677089257e-06, "loss": 1.3657, "step": 5708 }, { "epoch": 0.6074696743988083, "grad_norm": 14.65625, "learning_rate": 9.814916145537239e-06, "loss": 1.0604, "step": 5709 }, { "epoch": 0.6075760800170249, "grad_norm": 10.5390625, "learning_rate": 9.814880613985218e-06, "loss": 0.8751, "step": 5710 }, { "epoch": 0.6076824856352415, "grad_norm": 13.859375, "learning_rate": 9.814845082433202e-06, "loss": 1.114, "step": 5711 }, { "epoch": 0.6077888912534581, "grad_norm": 11.4375, "learning_rate": 9.814809550881183e-06, "loss": 1.2557, "step": 5712 }, { "epoch": 0.6078952968716749, "grad_norm": 14.2421875, "learning_rate": 9.814774019329165e-06, "loss": 1.28, "step": 5713 }, { "epoch": 0.6080017024898915, "grad_norm": 33.65625, "learning_rate": 9.814738487777147e-06, "loss": 1.0452, "step": 5714 }, { "epoch": 0.6081081081081081, "grad_norm": 13.3671875, "learning_rate": 9.814702956225128e-06, "loss": 1.2559, "step": 5715 }, { "epoch": 0.6082145137263247, "grad_norm": 14.109375, "learning_rate": 9.814667424673111e-06, "loss": 1.3843, "step": 5716 }, { "epoch": 0.6083209193445414, "grad_norm": 13.484375, "learning_rate": 9.814631893121093e-06, "loss": 1.3531, "step": 5717 }, { "epoch": 0.6084273249627581, "grad_norm": 11.7265625, "learning_rate": 9.814596361569075e-06, "loss": 1.4659, "step": 5718 }, { "epoch": 0.6085337305809747, "grad_norm": 11.4375, "learning_rate": 9.814560830017056e-06, "loss": 1.1638, "step": 5719 }, { "epoch": 0.6086401361991913, "grad_norm": 12.0546875, "learning_rate": 9.814525298465038e-06, "loss": 1.3334, "step": 5720 }, { "epoch": 0.608746541817408, "grad_norm": 12.140625, "learning_rate": 9.81448976691302e-06, "loss": 0.9768, "step": 5721 }, { "epoch": 0.6088529474356246, "grad_norm": 12.09375, "learning_rate": 9.814454235361001e-06, "loss": 1.3639, "step": 5722 }, { "epoch": 0.6089593530538412, "grad_norm": 10.0078125, "learning_rate": 9.814418703808983e-06, "loss": 1.2523, "step": 5723 }, { "epoch": 0.6090657586720579, "grad_norm": 16.984375, "learning_rate": 9.814383172256964e-06, "loss": 1.219, "step": 5724 }, { "epoch": 0.6091721642902745, "grad_norm": 10.703125, "learning_rate": 9.814347640704946e-06, "loss": 1.1429, "step": 5725 }, { "epoch": 0.6092785699084912, "grad_norm": 16.75, "learning_rate": 9.814312109152928e-06, "loss": 1.3187, "step": 5726 }, { "epoch": 0.6093849755267078, "grad_norm": 13.0703125, "learning_rate": 9.81427657760091e-06, "loss": 1.3438, "step": 5727 }, { "epoch": 0.6094913811449244, "grad_norm": 17.578125, "learning_rate": 9.814241046048892e-06, "loss": 1.8566, "step": 5728 }, { "epoch": 0.6095977867631411, "grad_norm": 13.5078125, "learning_rate": 9.814205514496874e-06, "loss": 1.1052, "step": 5729 }, { "epoch": 0.6097041923813578, "grad_norm": 19.0625, "learning_rate": 9.814169982944856e-06, "loss": 1.2284, "step": 5730 }, { "epoch": 0.6098105979995744, "grad_norm": 14.0234375, "learning_rate": 9.814134451392837e-06, "loss": 1.3527, "step": 5731 }, { "epoch": 0.609917003617791, "grad_norm": 15.3125, "learning_rate": 9.81409891984082e-06, "loss": 1.0596, "step": 5732 }, { "epoch": 0.6100234092360076, "grad_norm": 11.359375, "learning_rate": 9.814063388288802e-06, "loss": 1.1134, "step": 5733 }, { "epoch": 0.6101298148542244, "grad_norm": 12.4765625, "learning_rate": 9.814027856736784e-06, "loss": 1.2731, "step": 5734 }, { "epoch": 0.610236220472441, "grad_norm": 21.25, "learning_rate": 9.813992325184765e-06, "loss": 1.4838, "step": 5735 }, { "epoch": 0.6103426260906576, "grad_norm": 16.625, "learning_rate": 9.813956793632747e-06, "loss": 1.1811, "step": 5736 }, { "epoch": 0.6104490317088742, "grad_norm": 16.796875, "learning_rate": 9.813921262080729e-06, "loss": 1.5656, "step": 5737 }, { "epoch": 0.6105554373270908, "grad_norm": 13.4609375, "learning_rate": 9.81388573052871e-06, "loss": 1.364, "step": 5738 }, { "epoch": 0.6106618429453076, "grad_norm": 13.515625, "learning_rate": 9.813850198976692e-06, "loss": 1.464, "step": 5739 }, { "epoch": 0.6107682485635242, "grad_norm": 14.7578125, "learning_rate": 9.813814667424673e-06, "loss": 1.0925, "step": 5740 }, { "epoch": 0.6108746541817408, "grad_norm": 12.96875, "learning_rate": 9.813779135872655e-06, "loss": 1.5254, "step": 5741 }, { "epoch": 0.6109810597999574, "grad_norm": 11.375, "learning_rate": 9.813743604320638e-06, "loss": 1.568, "step": 5742 }, { "epoch": 0.611087465418174, "grad_norm": 13.5234375, "learning_rate": 9.81370807276862e-06, "loss": 1.4014, "step": 5743 }, { "epoch": 0.6111938710363907, "grad_norm": 15.078125, "learning_rate": 9.813672541216602e-06, "loss": 0.9057, "step": 5744 }, { "epoch": 0.6113002766546074, "grad_norm": 16.234375, "learning_rate": 9.813637009664583e-06, "loss": 1.3721, "step": 5745 }, { "epoch": 0.611406682272824, "grad_norm": 12.8203125, "learning_rate": 9.813601478112565e-06, "loss": 1.3119, "step": 5746 }, { "epoch": 0.6115130878910406, "grad_norm": 12.703125, "learning_rate": 9.813565946560546e-06, "loss": 1.1786, "step": 5747 }, { "epoch": 0.6116194935092573, "grad_norm": 11.6171875, "learning_rate": 9.813530415008528e-06, "loss": 0.916, "step": 5748 }, { "epoch": 0.6117258991274739, "grad_norm": 12.890625, "learning_rate": 9.81349488345651e-06, "loss": 1.2331, "step": 5749 }, { "epoch": 0.6118323047456906, "grad_norm": 18.359375, "learning_rate": 9.813459351904491e-06, "loss": 1.6055, "step": 5750 }, { "epoch": 0.6119387103639072, "grad_norm": 11.6328125, "learning_rate": 9.813423820352473e-06, "loss": 1.035, "step": 5751 }, { "epoch": 0.6120451159821239, "grad_norm": 11.9375, "learning_rate": 9.813388288800456e-06, "loss": 0.9724, "step": 5752 }, { "epoch": 0.6121515216003405, "grad_norm": 13.4765625, "learning_rate": 9.813352757248438e-06, "loss": 1.2458, "step": 5753 }, { "epoch": 0.6122579272185571, "grad_norm": 15.6328125, "learning_rate": 9.81331722569642e-06, "loss": 1.3118, "step": 5754 }, { "epoch": 0.6123643328367738, "grad_norm": 13.53125, "learning_rate": 9.813281694144401e-06, "loss": 1.0936, "step": 5755 }, { "epoch": 0.6124707384549904, "grad_norm": 13.546875, "learning_rate": 9.813246162592383e-06, "loss": 1.2805, "step": 5756 }, { "epoch": 0.6125771440732071, "grad_norm": 11.0078125, "learning_rate": 9.813210631040366e-06, "loss": 1.0798, "step": 5757 }, { "epoch": 0.6126835496914237, "grad_norm": 15.640625, "learning_rate": 9.813175099488347e-06, "loss": 1.0758, "step": 5758 }, { "epoch": 0.6127899553096403, "grad_norm": 17.90625, "learning_rate": 9.813139567936329e-06, "loss": 1.6055, "step": 5759 }, { "epoch": 0.612896360927857, "grad_norm": 9.4609375, "learning_rate": 9.81310403638431e-06, "loss": 0.937, "step": 5760 }, { "epoch": 0.6130027665460737, "grad_norm": 11.28125, "learning_rate": 9.813068504832292e-06, "loss": 1.3271, "step": 5761 }, { "epoch": 0.6131091721642903, "grad_norm": 10.0546875, "learning_rate": 9.813032973280274e-06, "loss": 1.0905, "step": 5762 }, { "epoch": 0.6132155777825069, "grad_norm": 11.2578125, "learning_rate": 9.812997441728255e-06, "loss": 1.1103, "step": 5763 }, { "epoch": 0.6133219834007235, "grad_norm": 16.234375, "learning_rate": 9.812961910176237e-06, "loss": 1.493, "step": 5764 }, { "epoch": 0.6134283890189401, "grad_norm": 12.28125, "learning_rate": 9.812926378624219e-06, "loss": 1.0385, "step": 5765 }, { "epoch": 0.6135347946371569, "grad_norm": 10.765625, "learning_rate": 9.8128908470722e-06, "loss": 1.1479, "step": 5766 }, { "epoch": 0.6136412002553735, "grad_norm": 14.2265625, "learning_rate": 9.812855315520182e-06, "loss": 1.1419, "step": 5767 }, { "epoch": 0.6137476058735901, "grad_norm": 12.3046875, "learning_rate": 9.812819783968165e-06, "loss": 1.3059, "step": 5768 }, { "epoch": 0.6138540114918067, "grad_norm": 14.078125, "learning_rate": 9.812784252416147e-06, "loss": 1.4544, "step": 5769 }, { "epoch": 0.6139604171100234, "grad_norm": 10.4921875, "learning_rate": 9.812748720864128e-06, "loss": 1.0103, "step": 5770 }, { "epoch": 0.6140668227282401, "grad_norm": 14.328125, "learning_rate": 9.81271318931211e-06, "loss": 1.2832, "step": 5771 }, { "epoch": 0.6141732283464567, "grad_norm": 12.234375, "learning_rate": 9.812677657760092e-06, "loss": 1.3803, "step": 5772 }, { "epoch": 0.6142796339646733, "grad_norm": 11.875, "learning_rate": 9.812642126208073e-06, "loss": 0.9553, "step": 5773 }, { "epoch": 0.61438603958289, "grad_norm": 14.828125, "learning_rate": 9.812606594656055e-06, "loss": 1.6127, "step": 5774 }, { "epoch": 0.6144924452011066, "grad_norm": 10.515625, "learning_rate": 9.812571063104036e-06, "loss": 1.068, "step": 5775 }, { "epoch": 0.6145988508193233, "grad_norm": 16.34375, "learning_rate": 9.812535531552018e-06, "loss": 1.026, "step": 5776 }, { "epoch": 0.6147052564375399, "grad_norm": 16.390625, "learning_rate": 9.8125e-06, "loss": 1.3791, "step": 5777 }, { "epoch": 0.6148116620557565, "grad_norm": 14.890625, "learning_rate": 9.812464468447983e-06, "loss": 1.2507, "step": 5778 }, { "epoch": 0.6149180676739732, "grad_norm": 8.9296875, "learning_rate": 9.812428936895965e-06, "loss": 0.7377, "step": 5779 }, { "epoch": 0.6150244732921898, "grad_norm": 12.578125, "learning_rate": 9.812393405343946e-06, "loss": 1.246, "step": 5780 }, { "epoch": 0.6151308789104065, "grad_norm": 18.140625, "learning_rate": 9.812357873791928e-06, "loss": 1.5331, "step": 5781 }, { "epoch": 0.6152372845286231, "grad_norm": 15.75, "learning_rate": 9.81232234223991e-06, "loss": 1.3805, "step": 5782 }, { "epoch": 0.6153436901468398, "grad_norm": 11.3671875, "learning_rate": 9.812286810687893e-06, "loss": 1.4031, "step": 5783 }, { "epoch": 0.6154500957650564, "grad_norm": 10.5625, "learning_rate": 9.812251279135874e-06, "loss": 1.444, "step": 5784 }, { "epoch": 0.615556501383273, "grad_norm": 21.703125, "learning_rate": 9.812215747583856e-06, "loss": 1.5263, "step": 5785 }, { "epoch": 0.6156629070014896, "grad_norm": 14.921875, "learning_rate": 9.812180216031837e-06, "loss": 1.1011, "step": 5786 }, { "epoch": 0.6157693126197064, "grad_norm": 15.015625, "learning_rate": 9.812144684479819e-06, "loss": 1.0527, "step": 5787 }, { "epoch": 0.615875718237923, "grad_norm": 12.390625, "learning_rate": 9.8121091529278e-06, "loss": 1.6205, "step": 5788 }, { "epoch": 0.6159821238561396, "grad_norm": 9.15625, "learning_rate": 9.812073621375782e-06, "loss": 1.1146, "step": 5789 }, { "epoch": 0.6160885294743562, "grad_norm": 13.9921875, "learning_rate": 9.812038089823764e-06, "loss": 1.2728, "step": 5790 }, { "epoch": 0.6161949350925728, "grad_norm": 11.0234375, "learning_rate": 9.812002558271746e-06, "loss": 1.3452, "step": 5791 }, { "epoch": 0.6163013407107896, "grad_norm": 11.6796875, "learning_rate": 9.811967026719727e-06, "loss": 1.3318, "step": 5792 }, { "epoch": 0.6164077463290062, "grad_norm": 12.4375, "learning_rate": 9.81193149516771e-06, "loss": 1.4163, "step": 5793 }, { "epoch": 0.6165141519472228, "grad_norm": 13.8125, "learning_rate": 9.811895963615692e-06, "loss": 1.6021, "step": 5794 }, { "epoch": 0.6166205575654394, "grad_norm": 12.28125, "learning_rate": 9.811860432063674e-06, "loss": 1.0431, "step": 5795 }, { "epoch": 0.616726963183656, "grad_norm": 18.984375, "learning_rate": 9.811824900511655e-06, "loss": 1.269, "step": 5796 }, { "epoch": 0.6168333688018728, "grad_norm": 15.1015625, "learning_rate": 9.811789368959637e-06, "loss": 1.6376, "step": 5797 }, { "epoch": 0.6169397744200894, "grad_norm": 13.0859375, "learning_rate": 9.811753837407618e-06, "loss": 1.3211, "step": 5798 }, { "epoch": 0.617046180038306, "grad_norm": 13.4921875, "learning_rate": 9.8117183058556e-06, "loss": 1.021, "step": 5799 }, { "epoch": 0.6171525856565226, "grad_norm": 11.5703125, "learning_rate": 9.811682774303582e-06, "loss": 1.436, "step": 5800 }, { "epoch": 0.6172589912747393, "grad_norm": 14.2265625, "learning_rate": 9.811647242751563e-06, "loss": 0.9099, "step": 5801 }, { "epoch": 0.617365396892956, "grad_norm": 16.171875, "learning_rate": 9.811611711199545e-06, "loss": 1.2792, "step": 5802 }, { "epoch": 0.6174718025111726, "grad_norm": 15.03125, "learning_rate": 9.811576179647527e-06, "loss": 1.3778, "step": 5803 }, { "epoch": 0.6175782081293892, "grad_norm": 17.953125, "learning_rate": 9.81154064809551e-06, "loss": 1.4214, "step": 5804 }, { "epoch": 0.6176846137476059, "grad_norm": 12.125, "learning_rate": 9.811505116543491e-06, "loss": 1.1362, "step": 5805 }, { "epoch": 0.6177910193658225, "grad_norm": 11.8203125, "learning_rate": 9.811469584991473e-06, "loss": 1.2013, "step": 5806 }, { "epoch": 0.6178974249840392, "grad_norm": 13.234375, "learning_rate": 9.811434053439455e-06, "loss": 1.2505, "step": 5807 }, { "epoch": 0.6180038306022558, "grad_norm": 11.6171875, "learning_rate": 9.811398521887436e-06, "loss": 1.2273, "step": 5808 }, { "epoch": 0.6181102362204725, "grad_norm": 15.8203125, "learning_rate": 9.81136299033542e-06, "loss": 1.3814, "step": 5809 }, { "epoch": 0.6182166418386891, "grad_norm": 12.8046875, "learning_rate": 9.811327458783401e-06, "loss": 1.3755, "step": 5810 }, { "epoch": 0.6183230474569057, "grad_norm": 9.546875, "learning_rate": 9.811291927231383e-06, "loss": 1.1669, "step": 5811 }, { "epoch": 0.6184294530751223, "grad_norm": 29.078125, "learning_rate": 9.811256395679364e-06, "loss": 0.9172, "step": 5812 }, { "epoch": 0.618535858693339, "grad_norm": 12.9921875, "learning_rate": 9.811220864127346e-06, "loss": 1.2256, "step": 5813 }, { "epoch": 0.6186422643115557, "grad_norm": 13.265625, "learning_rate": 9.811185332575328e-06, "loss": 1.419, "step": 5814 }, { "epoch": 0.6187486699297723, "grad_norm": 13.390625, "learning_rate": 9.81114980102331e-06, "loss": 1.414, "step": 5815 }, { "epoch": 0.6188550755479889, "grad_norm": 11.84375, "learning_rate": 9.81111426947129e-06, "loss": 1.1598, "step": 5816 }, { "epoch": 0.6189614811662055, "grad_norm": 9.7109375, "learning_rate": 9.811078737919272e-06, "loss": 1.1263, "step": 5817 }, { "epoch": 0.6190678867844223, "grad_norm": 12.1875, "learning_rate": 9.811043206367254e-06, "loss": 0.9975, "step": 5818 }, { "epoch": 0.6191742924026389, "grad_norm": 10.125, "learning_rate": 9.811007674815237e-06, "loss": 1.0052, "step": 5819 }, { "epoch": 0.6192806980208555, "grad_norm": 15.921875, "learning_rate": 9.810972143263219e-06, "loss": 1.0949, "step": 5820 }, { "epoch": 0.6193871036390721, "grad_norm": 11.0859375, "learning_rate": 9.8109366117112e-06, "loss": 1.2806, "step": 5821 }, { "epoch": 0.6194935092572887, "grad_norm": 10.21875, "learning_rate": 9.810901080159182e-06, "loss": 1.271, "step": 5822 }, { "epoch": 0.6195999148755055, "grad_norm": 10.9921875, "learning_rate": 9.810865548607164e-06, "loss": 1.3538, "step": 5823 }, { "epoch": 0.6197063204937221, "grad_norm": 12.0, "learning_rate": 9.810830017055147e-06, "loss": 1.1603, "step": 5824 }, { "epoch": 0.6198127261119387, "grad_norm": 19.40625, "learning_rate": 9.810794485503129e-06, "loss": 0.9999, "step": 5825 }, { "epoch": 0.6199191317301553, "grad_norm": 12.828125, "learning_rate": 9.81075895395111e-06, "loss": 1.3762, "step": 5826 }, { "epoch": 0.620025537348372, "grad_norm": 13.28125, "learning_rate": 9.810723422399092e-06, "loss": 1.1533, "step": 5827 }, { "epoch": 0.6201319429665887, "grad_norm": 11.1875, "learning_rate": 9.810687890847073e-06, "loss": 1.0572, "step": 5828 }, { "epoch": 0.6202383485848053, "grad_norm": 12.015625, "learning_rate": 9.810652359295055e-06, "loss": 1.1856, "step": 5829 }, { "epoch": 0.6203447542030219, "grad_norm": 16.109375, "learning_rate": 9.810616827743037e-06, "loss": 1.0695, "step": 5830 }, { "epoch": 0.6204511598212386, "grad_norm": 13.390625, "learning_rate": 9.810581296191018e-06, "loss": 1.1842, "step": 5831 }, { "epoch": 0.6205575654394552, "grad_norm": 11.8203125, "learning_rate": 9.810545764639e-06, "loss": 1.1807, "step": 5832 }, { "epoch": 0.6206639710576718, "grad_norm": 9.421875, "learning_rate": 9.810510233086981e-06, "loss": 1.0797, "step": 5833 }, { "epoch": 0.6207703766758885, "grad_norm": 10.3125, "learning_rate": 9.810474701534965e-06, "loss": 1.1217, "step": 5834 }, { "epoch": 0.6208767822941051, "grad_norm": 10.9296875, "learning_rate": 9.810439169982946e-06, "loss": 1.1657, "step": 5835 }, { "epoch": 0.6209831879123218, "grad_norm": 16.140625, "learning_rate": 9.810403638430928e-06, "loss": 1.2449, "step": 5836 }, { "epoch": 0.6210895935305384, "grad_norm": 14.40625, "learning_rate": 9.81036810687891e-06, "loss": 1.3433, "step": 5837 }, { "epoch": 0.621195999148755, "grad_norm": 13.75, "learning_rate": 9.810332575326891e-06, "loss": 1.0723, "step": 5838 }, { "epoch": 0.6213024047669717, "grad_norm": 17.171875, "learning_rate": 9.810297043774873e-06, "loss": 1.0926, "step": 5839 }, { "epoch": 0.6214088103851884, "grad_norm": 12.3125, "learning_rate": 9.810261512222854e-06, "loss": 1.1334, "step": 5840 }, { "epoch": 0.621515216003405, "grad_norm": 20.5, "learning_rate": 9.810225980670836e-06, "loss": 1.5536, "step": 5841 }, { "epoch": 0.6216216216216216, "grad_norm": 14.40625, "learning_rate": 9.810190449118818e-06, "loss": 1.2041, "step": 5842 }, { "epoch": 0.6217280272398382, "grad_norm": 12.203125, "learning_rate": 9.8101549175668e-06, "loss": 1.0652, "step": 5843 }, { "epoch": 0.621834432858055, "grad_norm": 16.28125, "learning_rate": 9.810119386014781e-06, "loss": 1.4731, "step": 5844 }, { "epoch": 0.6219408384762716, "grad_norm": 12.6796875, "learning_rate": 9.810083854462764e-06, "loss": 1.1786, "step": 5845 }, { "epoch": 0.6220472440944882, "grad_norm": 11.25, "learning_rate": 9.810048322910746e-06, "loss": 1.4587, "step": 5846 }, { "epoch": 0.6221536497127048, "grad_norm": 10.796875, "learning_rate": 9.810012791358727e-06, "loss": 1.2342, "step": 5847 }, { "epoch": 0.6222600553309214, "grad_norm": 15.4375, "learning_rate": 9.809977259806709e-06, "loss": 1.131, "step": 5848 }, { "epoch": 0.6223664609491382, "grad_norm": 11.09375, "learning_rate": 9.80994172825469e-06, "loss": 1.3576, "step": 5849 }, { "epoch": 0.6224728665673548, "grad_norm": 11.0703125, "learning_rate": 9.809906196702674e-06, "loss": 1.1138, "step": 5850 }, { "epoch": 0.6225792721855714, "grad_norm": 12.8359375, "learning_rate": 9.809870665150655e-06, "loss": 1.2092, "step": 5851 }, { "epoch": 0.622685677803788, "grad_norm": 13.65625, "learning_rate": 9.809835133598637e-06, "loss": 1.2643, "step": 5852 }, { "epoch": 0.6227920834220046, "grad_norm": 9.90625, "learning_rate": 9.809799602046619e-06, "loss": 1.0602, "step": 5853 }, { "epoch": 0.6228984890402213, "grad_norm": 12.984375, "learning_rate": 9.8097640704946e-06, "loss": 1.0352, "step": 5854 }, { "epoch": 0.623004894658438, "grad_norm": 18.078125, "learning_rate": 9.809728538942582e-06, "loss": 1.3251, "step": 5855 }, { "epoch": 0.6231113002766546, "grad_norm": 9.5546875, "learning_rate": 9.809693007390564e-06, "loss": 1.1745, "step": 5856 }, { "epoch": 0.6232177058948712, "grad_norm": 11.0, "learning_rate": 9.809657475838545e-06, "loss": 1.1967, "step": 5857 }, { "epoch": 0.6233241115130879, "grad_norm": 12.4765625, "learning_rate": 9.809621944286527e-06, "loss": 1.0602, "step": 5858 }, { "epoch": 0.6234305171313045, "grad_norm": 13.5, "learning_rate": 9.809586412734508e-06, "loss": 1.4335, "step": 5859 }, { "epoch": 0.6235369227495212, "grad_norm": 12.671875, "learning_rate": 9.809550881182492e-06, "loss": 1.2442, "step": 5860 }, { "epoch": 0.6236433283677378, "grad_norm": 10.4140625, "learning_rate": 9.809515349630473e-06, "loss": 0.8781, "step": 5861 }, { "epoch": 0.6237497339859545, "grad_norm": 15.4453125, "learning_rate": 9.809479818078455e-06, "loss": 1.6136, "step": 5862 }, { "epoch": 0.6238561396041711, "grad_norm": 16.015625, "learning_rate": 9.809444286526436e-06, "loss": 0.9016, "step": 5863 }, { "epoch": 0.6239625452223877, "grad_norm": 13.21875, "learning_rate": 9.809408754974418e-06, "loss": 1.2483, "step": 5864 }, { "epoch": 0.6240689508406044, "grad_norm": 12.984375, "learning_rate": 9.8093732234224e-06, "loss": 1.2095, "step": 5865 }, { "epoch": 0.624175356458821, "grad_norm": 13.046875, "learning_rate": 9.809337691870381e-06, "loss": 1.1973, "step": 5866 }, { "epoch": 0.6242817620770377, "grad_norm": 12.6171875, "learning_rate": 9.809302160318363e-06, "loss": 0.9999, "step": 5867 }, { "epoch": 0.6243881676952543, "grad_norm": 13.3828125, "learning_rate": 9.809266628766344e-06, "loss": 1.0528, "step": 5868 }, { "epoch": 0.6244945733134709, "grad_norm": 11.6015625, "learning_rate": 9.809231097214326e-06, "loss": 1.011, "step": 5869 }, { "epoch": 0.6246009789316876, "grad_norm": 13.2109375, "learning_rate": 9.80919556566231e-06, "loss": 1.5655, "step": 5870 }, { "epoch": 0.6247073845499043, "grad_norm": 20.15625, "learning_rate": 9.809160034110291e-06, "loss": 1.5172, "step": 5871 }, { "epoch": 0.6248137901681209, "grad_norm": 10.9609375, "learning_rate": 9.809124502558273e-06, "loss": 1.0736, "step": 5872 }, { "epoch": 0.6249201957863375, "grad_norm": 17.03125, "learning_rate": 9.809088971006254e-06, "loss": 1.1957, "step": 5873 }, { "epoch": 0.6250266014045541, "grad_norm": 12.3359375, "learning_rate": 9.809053439454236e-06, "loss": 1.1558, "step": 5874 }, { "epoch": 0.6251330070227707, "grad_norm": 15.5078125, "learning_rate": 9.809017907902217e-06, "loss": 1.286, "step": 5875 }, { "epoch": 0.6252394126409875, "grad_norm": 13.953125, "learning_rate": 9.8089823763502e-06, "loss": 1.4995, "step": 5876 }, { "epoch": 0.6253458182592041, "grad_norm": 19.65625, "learning_rate": 9.808946844798182e-06, "loss": 1.0811, "step": 5877 }, { "epoch": 0.6254522238774207, "grad_norm": 11.8984375, "learning_rate": 9.808911313246164e-06, "loss": 1.1582, "step": 5878 }, { "epoch": 0.6255586294956373, "grad_norm": 9.9453125, "learning_rate": 9.808875781694146e-06, "loss": 1.0483, "step": 5879 }, { "epoch": 0.625665035113854, "grad_norm": 16.171875, "learning_rate": 9.808840250142127e-06, "loss": 1.2526, "step": 5880 }, { "epoch": 0.6257714407320707, "grad_norm": 11.28125, "learning_rate": 9.808804718590109e-06, "loss": 0.9197, "step": 5881 }, { "epoch": 0.6258778463502873, "grad_norm": 13.859375, "learning_rate": 9.80876918703809e-06, "loss": 1.3106, "step": 5882 }, { "epoch": 0.6259842519685039, "grad_norm": 18.40625, "learning_rate": 9.808733655486072e-06, "loss": 1.8427, "step": 5883 }, { "epoch": 0.6260906575867206, "grad_norm": 18.75, "learning_rate": 9.808698123934054e-06, "loss": 0.9894, "step": 5884 }, { "epoch": 0.6261970632049372, "grad_norm": 13.2265625, "learning_rate": 9.808662592382035e-06, "loss": 1.1389, "step": 5885 }, { "epoch": 0.6263034688231539, "grad_norm": 14.9375, "learning_rate": 9.808627060830018e-06, "loss": 1.398, "step": 5886 }, { "epoch": 0.6264098744413705, "grad_norm": 15.5546875, "learning_rate": 9.808591529278e-06, "loss": 1.2958, "step": 5887 }, { "epoch": 0.6265162800595871, "grad_norm": 12.5078125, "learning_rate": 9.808555997725982e-06, "loss": 1.3341, "step": 5888 }, { "epoch": 0.6266226856778038, "grad_norm": 15.625, "learning_rate": 9.808520466173963e-06, "loss": 1.3225, "step": 5889 }, { "epoch": 0.6267290912960204, "grad_norm": 17.5, "learning_rate": 9.808484934621945e-06, "loss": 1.6751, "step": 5890 }, { "epoch": 0.6268354969142371, "grad_norm": 14.625, "learning_rate": 9.808449403069927e-06, "loss": 1.2777, "step": 5891 }, { "epoch": 0.6269419025324537, "grad_norm": 11.0859375, "learning_rate": 9.808413871517908e-06, "loss": 1.1009, "step": 5892 }, { "epoch": 0.6270483081506704, "grad_norm": 11.25, "learning_rate": 9.80837833996589e-06, "loss": 1.4065, "step": 5893 }, { "epoch": 0.627154713768887, "grad_norm": 14.890625, "learning_rate": 9.808342808413871e-06, "loss": 1.4324, "step": 5894 }, { "epoch": 0.6272611193871036, "grad_norm": 9.8828125, "learning_rate": 9.808307276861853e-06, "loss": 1.308, "step": 5895 }, { "epoch": 0.6273675250053202, "grad_norm": 12.078125, "learning_rate": 9.808271745309836e-06, "loss": 1.0465, "step": 5896 }, { "epoch": 0.627473930623537, "grad_norm": 12.375, "learning_rate": 9.808236213757818e-06, "loss": 1.0411, "step": 5897 }, { "epoch": 0.6275803362417536, "grad_norm": 13.9453125, "learning_rate": 9.8082006822058e-06, "loss": 1.3301, "step": 5898 }, { "epoch": 0.6276867418599702, "grad_norm": 13.7734375, "learning_rate": 9.808165150653781e-06, "loss": 0.9582, "step": 5899 }, { "epoch": 0.6277931474781868, "grad_norm": 19.578125, "learning_rate": 9.808129619101763e-06, "loss": 1.2972, "step": 5900 }, { "epoch": 0.6278995530964034, "grad_norm": 13.7265625, "learning_rate": 9.808094087549746e-06, "loss": 1.3044, "step": 5901 }, { "epoch": 0.6280059587146202, "grad_norm": 12.3984375, "learning_rate": 9.808058555997728e-06, "loss": 1.1792, "step": 5902 }, { "epoch": 0.6281123643328368, "grad_norm": 12.5234375, "learning_rate": 9.80802302444571e-06, "loss": 1.001, "step": 5903 }, { "epoch": 0.6282187699510534, "grad_norm": 11.8046875, "learning_rate": 9.80798749289369e-06, "loss": 1.0442, "step": 5904 }, { "epoch": 0.62832517556927, "grad_norm": 14.65625, "learning_rate": 9.807951961341672e-06, "loss": 1.2508, "step": 5905 }, { "epoch": 0.6284315811874867, "grad_norm": 14.1171875, "learning_rate": 9.807916429789654e-06, "loss": 1.397, "step": 5906 }, { "epoch": 0.6285379868057034, "grad_norm": 11.25, "learning_rate": 9.807880898237636e-06, "loss": 1.3579, "step": 5907 }, { "epoch": 0.62864439242392, "grad_norm": 14.640625, "learning_rate": 9.807845366685617e-06, "loss": 1.1136, "step": 5908 }, { "epoch": 0.6287507980421366, "grad_norm": 12.2734375, "learning_rate": 9.807809835133599e-06, "loss": 1.2548, "step": 5909 }, { "epoch": 0.6288572036603532, "grad_norm": 12.2421875, "learning_rate": 9.80777430358158e-06, "loss": 1.1639, "step": 5910 }, { "epoch": 0.6289636092785699, "grad_norm": 20.921875, "learning_rate": 9.807738772029564e-06, "loss": 1.2794, "step": 5911 }, { "epoch": 0.6290700148967866, "grad_norm": 11.2890625, "learning_rate": 9.807703240477545e-06, "loss": 1.3004, "step": 5912 }, { "epoch": 0.6291764205150032, "grad_norm": 14.109375, "learning_rate": 9.807667708925527e-06, "loss": 1.2732, "step": 5913 }, { "epoch": 0.6292828261332198, "grad_norm": 13.8671875, "learning_rate": 9.807632177373509e-06, "loss": 1.2378, "step": 5914 }, { "epoch": 0.6293892317514365, "grad_norm": 16.703125, "learning_rate": 9.80759664582149e-06, "loss": 1.2644, "step": 5915 }, { "epoch": 0.6294956373696531, "grad_norm": 10.953125, "learning_rate": 9.807561114269472e-06, "loss": 1.2914, "step": 5916 }, { "epoch": 0.6296020429878698, "grad_norm": 11.1875, "learning_rate": 9.807525582717455e-06, "loss": 1.1254, "step": 5917 }, { "epoch": 0.6297084486060864, "grad_norm": 20.03125, "learning_rate": 9.807490051165437e-06, "loss": 1.4333, "step": 5918 }, { "epoch": 0.6298148542243031, "grad_norm": 16.59375, "learning_rate": 9.807454519613417e-06, "loss": 1.2439, "step": 5919 }, { "epoch": 0.6299212598425197, "grad_norm": 15.640625, "learning_rate": 9.807418988061398e-06, "loss": 1.557, "step": 5920 }, { "epoch": 0.6300276654607363, "grad_norm": 12.859375, "learning_rate": 9.80738345650938e-06, "loss": 1.0121, "step": 5921 }, { "epoch": 0.6301340710789529, "grad_norm": 14.1484375, "learning_rate": 9.807347924957363e-06, "loss": 1.1261, "step": 5922 }, { "epoch": 0.6302404766971696, "grad_norm": 10.9375, "learning_rate": 9.807312393405345e-06, "loss": 1.0826, "step": 5923 }, { "epoch": 0.6303468823153863, "grad_norm": 11.3984375, "learning_rate": 9.807276861853326e-06, "loss": 1.2012, "step": 5924 }, { "epoch": 0.6304532879336029, "grad_norm": 12.9765625, "learning_rate": 9.807241330301308e-06, "loss": 1.553, "step": 5925 }, { "epoch": 0.6305596935518195, "grad_norm": 12.3828125, "learning_rate": 9.80720579874929e-06, "loss": 1.1942, "step": 5926 }, { "epoch": 0.6306660991700361, "grad_norm": 12.75, "learning_rate": 9.807170267197273e-06, "loss": 0.8329, "step": 5927 }, { "epoch": 0.6307725047882529, "grad_norm": 13.1328125, "learning_rate": 9.807134735645254e-06, "loss": 1.0056, "step": 5928 }, { "epoch": 0.6308789104064695, "grad_norm": 11.6875, "learning_rate": 9.807099204093236e-06, "loss": 1.3457, "step": 5929 }, { "epoch": 0.6309853160246861, "grad_norm": 12.09375, "learning_rate": 9.807063672541218e-06, "loss": 1.3045, "step": 5930 }, { "epoch": 0.6310917216429027, "grad_norm": 13.171875, "learning_rate": 9.8070281409892e-06, "loss": 1.3085, "step": 5931 }, { "epoch": 0.6311981272611193, "grad_norm": 14.5703125, "learning_rate": 9.806992609437181e-06, "loss": 1.3376, "step": 5932 }, { "epoch": 0.6313045328793361, "grad_norm": 14.7734375, "learning_rate": 9.806957077885162e-06, "loss": 1.1395, "step": 5933 }, { "epoch": 0.6314109384975527, "grad_norm": 15.796875, "learning_rate": 9.806921546333144e-06, "loss": 1.0804, "step": 5934 }, { "epoch": 0.6315173441157693, "grad_norm": 16.953125, "learning_rate": 9.806886014781126e-06, "loss": 1.386, "step": 5935 }, { "epoch": 0.6316237497339859, "grad_norm": 13.078125, "learning_rate": 9.806850483229107e-06, "loss": 0.981, "step": 5936 }, { "epoch": 0.6317301553522026, "grad_norm": 12.3984375, "learning_rate": 9.80681495167709e-06, "loss": 1.2517, "step": 5937 }, { "epoch": 0.6318365609704193, "grad_norm": 11.4296875, "learning_rate": 9.806779420125072e-06, "loss": 1.0504, "step": 5938 }, { "epoch": 0.6319429665886359, "grad_norm": 10.1953125, "learning_rate": 9.806743888573054e-06, "loss": 1.1612, "step": 5939 }, { "epoch": 0.6320493722068525, "grad_norm": 13.015625, "learning_rate": 9.806708357021035e-06, "loss": 1.0449, "step": 5940 }, { "epoch": 0.6321557778250692, "grad_norm": 14.0, "learning_rate": 9.806672825469017e-06, "loss": 1.1213, "step": 5941 }, { "epoch": 0.6322621834432858, "grad_norm": 30.046875, "learning_rate": 9.806637293917e-06, "loss": 1.702, "step": 5942 }, { "epoch": 0.6323685890615024, "grad_norm": 12.578125, "learning_rate": 9.806601762364982e-06, "loss": 1.3882, "step": 5943 }, { "epoch": 0.6324749946797191, "grad_norm": 11.6171875, "learning_rate": 9.806566230812964e-06, "loss": 1.2026, "step": 5944 }, { "epoch": 0.6325814002979357, "grad_norm": 12.5546875, "learning_rate": 9.806530699260945e-06, "loss": 1.2368, "step": 5945 }, { "epoch": 0.6326878059161524, "grad_norm": 13.0, "learning_rate": 9.806495167708927e-06, "loss": 1.2253, "step": 5946 }, { "epoch": 0.632794211534369, "grad_norm": 10.3671875, "learning_rate": 9.806459636156908e-06, "loss": 1.2602, "step": 5947 }, { "epoch": 0.6329006171525856, "grad_norm": 12.6171875, "learning_rate": 9.80642410460489e-06, "loss": 1.0246, "step": 5948 }, { "epoch": 0.6330070227708023, "grad_norm": 9.90625, "learning_rate": 9.806388573052872e-06, "loss": 1.2624, "step": 5949 }, { "epoch": 0.633113428389019, "grad_norm": 13.46875, "learning_rate": 9.806353041500853e-06, "loss": 1.0327, "step": 5950 }, { "epoch": 0.6332198340072356, "grad_norm": 12.140625, "learning_rate": 9.806317509948835e-06, "loss": 1.1813, "step": 5951 }, { "epoch": 0.6333262396254522, "grad_norm": 13.2578125, "learning_rate": 9.806281978396818e-06, "loss": 1.1949, "step": 5952 }, { "epoch": 0.6334326452436688, "grad_norm": 9.796875, "learning_rate": 9.8062464468448e-06, "loss": 1.2399, "step": 5953 }, { "epoch": 0.6335390508618856, "grad_norm": 14.046875, "learning_rate": 9.806210915292781e-06, "loss": 1.3731, "step": 5954 }, { "epoch": 0.6336454564801022, "grad_norm": 14.125, "learning_rate": 9.806175383740763e-06, "loss": 1.1425, "step": 5955 }, { "epoch": 0.6337518620983188, "grad_norm": 16.96875, "learning_rate": 9.806139852188745e-06, "loss": 1.4123, "step": 5956 }, { "epoch": 0.6338582677165354, "grad_norm": 11.546875, "learning_rate": 9.806104320636726e-06, "loss": 1.1439, "step": 5957 }, { "epoch": 0.633964673334752, "grad_norm": 11.359375, "learning_rate": 9.806068789084708e-06, "loss": 1.3992, "step": 5958 }, { "epoch": 0.6340710789529688, "grad_norm": 11.46875, "learning_rate": 9.80603325753269e-06, "loss": 1.0782, "step": 5959 }, { "epoch": 0.6341774845711854, "grad_norm": 16.875, "learning_rate": 9.805997725980671e-06, "loss": 1.4451, "step": 5960 }, { "epoch": 0.634283890189402, "grad_norm": 15.4609375, "learning_rate": 9.805962194428653e-06, "loss": 1.9918, "step": 5961 }, { "epoch": 0.6343902958076186, "grad_norm": 14.6171875, "learning_rate": 9.805926662876634e-06, "loss": 1.2616, "step": 5962 }, { "epoch": 0.6344967014258353, "grad_norm": 10.609375, "learning_rate": 9.805891131324617e-06, "loss": 1.119, "step": 5963 }, { "epoch": 0.6346031070440519, "grad_norm": 12.1171875, "learning_rate": 9.805855599772599e-06, "loss": 1.22, "step": 5964 }, { "epoch": 0.6347095126622686, "grad_norm": 13.6875, "learning_rate": 9.80582006822058e-06, "loss": 1.1086, "step": 5965 }, { "epoch": 0.6348159182804852, "grad_norm": 14.953125, "learning_rate": 9.805784536668562e-06, "loss": 1.3576, "step": 5966 }, { "epoch": 0.6349223238987018, "grad_norm": 18.09375, "learning_rate": 9.805749005116544e-06, "loss": 1.3533, "step": 5967 }, { "epoch": 0.6350287295169185, "grad_norm": 12.71875, "learning_rate": 9.805713473564527e-06, "loss": 1.3294, "step": 5968 }, { "epoch": 0.6351351351351351, "grad_norm": 12.6328125, "learning_rate": 9.805677942012509e-06, "loss": 0.9914, "step": 5969 }, { "epoch": 0.6352415407533518, "grad_norm": 15.953125, "learning_rate": 9.80564241046049e-06, "loss": 1.1419, "step": 5970 }, { "epoch": 0.6353479463715684, "grad_norm": 11.3515625, "learning_rate": 9.805606878908472e-06, "loss": 1.3064, "step": 5971 }, { "epoch": 0.6354543519897851, "grad_norm": 14.265625, "learning_rate": 9.805571347356454e-06, "loss": 1.3986, "step": 5972 }, { "epoch": 0.6355607576080017, "grad_norm": 23.59375, "learning_rate": 9.805535815804435e-06, "loss": 1.0022, "step": 5973 }, { "epoch": 0.6356671632262183, "grad_norm": 11.03125, "learning_rate": 9.805500284252417e-06, "loss": 1.3604, "step": 5974 }, { "epoch": 0.635773568844435, "grad_norm": 13.5234375, "learning_rate": 9.805464752700398e-06, "loss": 1.2818, "step": 5975 }, { "epoch": 0.6358799744626517, "grad_norm": 11.3671875, "learning_rate": 9.80542922114838e-06, "loss": 1.1452, "step": 5976 }, { "epoch": 0.6359863800808683, "grad_norm": 16.125, "learning_rate": 9.805393689596362e-06, "loss": 1.2459, "step": 5977 }, { "epoch": 0.6360927856990849, "grad_norm": 10.28125, "learning_rate": 9.805358158044345e-06, "loss": 1.2176, "step": 5978 }, { "epoch": 0.6361991913173015, "grad_norm": 17.421875, "learning_rate": 9.805322626492327e-06, "loss": 1.2639, "step": 5979 }, { "epoch": 0.6363055969355182, "grad_norm": 13.2109375, "learning_rate": 9.805287094940308e-06, "loss": 1.1189, "step": 5980 }, { "epoch": 0.6364120025537349, "grad_norm": 14.2890625, "learning_rate": 9.80525156338829e-06, "loss": 1.8223, "step": 5981 }, { "epoch": 0.6365184081719515, "grad_norm": 11.9921875, "learning_rate": 9.805216031836271e-06, "loss": 0.9708, "step": 5982 }, { "epoch": 0.6366248137901681, "grad_norm": 11.3515625, "learning_rate": 9.805180500284253e-06, "loss": 1.0726, "step": 5983 }, { "epoch": 0.6367312194083847, "grad_norm": 13.8515625, "learning_rate": 9.805144968732235e-06, "loss": 1.0046, "step": 5984 }, { "epoch": 0.6368376250266014, "grad_norm": 12.234375, "learning_rate": 9.805109437180216e-06, "loss": 1.5152, "step": 5985 }, { "epoch": 0.6369440306448181, "grad_norm": 14.3203125, "learning_rate": 9.805073905628198e-06, "loss": 1.0162, "step": 5986 }, { "epoch": 0.6370504362630347, "grad_norm": 11.859375, "learning_rate": 9.80503837407618e-06, "loss": 1.0939, "step": 5987 }, { "epoch": 0.6371568418812513, "grad_norm": 18.515625, "learning_rate": 9.805002842524163e-06, "loss": 1.1674, "step": 5988 }, { "epoch": 0.6372632474994679, "grad_norm": 11.8046875, "learning_rate": 9.804967310972144e-06, "loss": 1.1166, "step": 5989 }, { "epoch": 0.6373696531176846, "grad_norm": 14.9609375, "learning_rate": 9.804931779420126e-06, "loss": 1.1725, "step": 5990 }, { "epoch": 0.6374760587359013, "grad_norm": 32.59375, "learning_rate": 9.804896247868108e-06, "loss": 1.712, "step": 5991 }, { "epoch": 0.6375824643541179, "grad_norm": 69.75, "learning_rate": 9.804860716316089e-06, "loss": 1.2469, "step": 5992 }, { "epoch": 0.6376888699723345, "grad_norm": 14.9296875, "learning_rate": 9.80482518476407e-06, "loss": 1.1179, "step": 5993 }, { "epoch": 0.6377952755905512, "grad_norm": 18.734375, "learning_rate": 9.804789653212054e-06, "loss": 1.7606, "step": 5994 }, { "epoch": 0.6379016812087678, "grad_norm": 14.2578125, "learning_rate": 9.804754121660036e-06, "loss": 1.2671, "step": 5995 }, { "epoch": 0.6380080868269845, "grad_norm": 14.578125, "learning_rate": 9.804718590108017e-06, "loss": 1.292, "step": 5996 }, { "epoch": 0.6381144924452011, "grad_norm": 10.703125, "learning_rate": 9.804683058555999e-06, "loss": 1.1565, "step": 5997 }, { "epoch": 0.6382208980634178, "grad_norm": 18.390625, "learning_rate": 9.80464752700398e-06, "loss": 1.1835, "step": 5998 }, { "epoch": 0.6383273036816344, "grad_norm": 12.3359375, "learning_rate": 9.804611995451962e-06, "loss": 1.0445, "step": 5999 }, { "epoch": 0.638433709299851, "grad_norm": 13.2734375, "learning_rate": 9.804576463899944e-06, "loss": 1.2772, "step": 6000 }, { "epoch": 0.6385401149180677, "grad_norm": 12.8359375, "learning_rate": 9.804540932347925e-06, "loss": 0.9701, "step": 6001 }, { "epoch": 0.6386465205362843, "grad_norm": 13.4140625, "learning_rate": 9.804505400795907e-06, "loss": 1.1331, "step": 6002 }, { "epoch": 0.638752926154501, "grad_norm": 22.46875, "learning_rate": 9.804469869243888e-06, "loss": 1.5542, "step": 6003 }, { "epoch": 0.6388593317727176, "grad_norm": 10.7734375, "learning_rate": 9.804434337691872e-06, "loss": 1.4589, "step": 6004 }, { "epoch": 0.6389657373909342, "grad_norm": 21.15625, "learning_rate": 9.804398806139853e-06, "loss": 1.6196, "step": 6005 }, { "epoch": 0.6390721430091508, "grad_norm": 41.34375, "learning_rate": 9.804363274587835e-06, "loss": 1.1022, "step": 6006 }, { "epoch": 0.6391785486273676, "grad_norm": 10.5546875, "learning_rate": 9.804327743035817e-06, "loss": 1.4018, "step": 6007 }, { "epoch": 0.6392849542455842, "grad_norm": 15.53125, "learning_rate": 9.804292211483798e-06, "loss": 1.1628, "step": 6008 }, { "epoch": 0.6393913598638008, "grad_norm": 10.0703125, "learning_rate": 9.80425667993178e-06, "loss": 0.9151, "step": 6009 }, { "epoch": 0.6394977654820174, "grad_norm": 12.7890625, "learning_rate": 9.804221148379761e-06, "loss": 1.0623, "step": 6010 }, { "epoch": 0.639604171100234, "grad_norm": 14.6015625, "learning_rate": 9.804185616827743e-06, "loss": 1.2957, "step": 6011 }, { "epoch": 0.6397105767184508, "grad_norm": 17.84375, "learning_rate": 9.804150085275725e-06, "loss": 1.3312, "step": 6012 }, { "epoch": 0.6398169823366674, "grad_norm": 9.4296875, "learning_rate": 9.804114553723706e-06, "loss": 1.0618, "step": 6013 }, { "epoch": 0.639923387954884, "grad_norm": 11.8125, "learning_rate": 9.80407902217169e-06, "loss": 1.3818, "step": 6014 }, { "epoch": 0.6400297935731006, "grad_norm": 10.5859375, "learning_rate": 9.804043490619671e-06, "loss": 1.1737, "step": 6015 }, { "epoch": 0.6401361991913173, "grad_norm": 13.078125, "learning_rate": 9.804007959067653e-06, "loss": 1.2982, "step": 6016 }, { "epoch": 0.640242604809534, "grad_norm": 9.265625, "learning_rate": 9.803972427515634e-06, "loss": 1.0879, "step": 6017 }, { "epoch": 0.6403490104277506, "grad_norm": 13.4140625, "learning_rate": 9.803936895963616e-06, "loss": 1.3944, "step": 6018 }, { "epoch": 0.6404554160459672, "grad_norm": 13.921875, "learning_rate": 9.8039013644116e-06, "loss": 1.4077, "step": 6019 }, { "epoch": 0.6405618216641839, "grad_norm": 13.359375, "learning_rate": 9.803865832859581e-06, "loss": 1.0868, "step": 6020 }, { "epoch": 0.6406682272824005, "grad_norm": 12.3671875, "learning_rate": 9.803830301307562e-06, "loss": 1.4778, "step": 6021 }, { "epoch": 0.6407746329006172, "grad_norm": 13.6171875, "learning_rate": 9.803794769755544e-06, "loss": 1.6652, "step": 6022 }, { "epoch": 0.6408810385188338, "grad_norm": 31.390625, "learning_rate": 9.803759238203526e-06, "loss": 1.3509, "step": 6023 }, { "epoch": 0.6409874441370504, "grad_norm": 12.0, "learning_rate": 9.803723706651507e-06, "loss": 1.2648, "step": 6024 }, { "epoch": 0.6410938497552671, "grad_norm": 10.296875, "learning_rate": 9.803688175099489e-06, "loss": 0.9599, "step": 6025 }, { "epoch": 0.6412002553734837, "grad_norm": 13.109375, "learning_rate": 9.80365264354747e-06, "loss": 1.3323, "step": 6026 }, { "epoch": 0.6413066609917004, "grad_norm": 12.9296875, "learning_rate": 9.803617111995452e-06, "loss": 1.1027, "step": 6027 }, { "epoch": 0.641413066609917, "grad_norm": 14.1171875, "learning_rate": 9.803581580443434e-06, "loss": 1.2935, "step": 6028 }, { "epoch": 0.6415194722281337, "grad_norm": 12.296875, "learning_rate": 9.803546048891417e-06, "loss": 1.511, "step": 6029 }, { "epoch": 0.6416258778463503, "grad_norm": 14.0859375, "learning_rate": 9.803510517339399e-06, "loss": 1.3823, "step": 6030 }, { "epoch": 0.6417322834645669, "grad_norm": 11.203125, "learning_rate": 9.80347498578738e-06, "loss": 1.0271, "step": 6031 }, { "epoch": 0.6418386890827835, "grad_norm": 9.65625, "learning_rate": 9.803439454235362e-06, "loss": 0.9376, "step": 6032 }, { "epoch": 0.6419450947010003, "grad_norm": 17.875, "learning_rate": 9.803403922683343e-06, "loss": 1.4917, "step": 6033 }, { "epoch": 0.6420515003192169, "grad_norm": 12.375, "learning_rate": 9.803368391131325e-06, "loss": 1.1031, "step": 6034 }, { "epoch": 0.6421579059374335, "grad_norm": 13.5703125, "learning_rate": 9.803332859579308e-06, "loss": 1.1662, "step": 6035 }, { "epoch": 0.6422643115556501, "grad_norm": 14.625, "learning_rate": 9.80329732802729e-06, "loss": 1.1746, "step": 6036 }, { "epoch": 0.6423707171738667, "grad_norm": 14.53125, "learning_rate": 9.803261796475272e-06, "loss": 1.4988, "step": 6037 }, { "epoch": 0.6424771227920835, "grad_norm": 17.359375, "learning_rate": 9.803226264923253e-06, "loss": 1.2564, "step": 6038 }, { "epoch": 0.6425835284103001, "grad_norm": 18.78125, "learning_rate": 9.803190733371233e-06, "loss": 1.2142, "step": 6039 }, { "epoch": 0.6426899340285167, "grad_norm": 12.53125, "learning_rate": 9.803155201819216e-06, "loss": 1.2112, "step": 6040 }, { "epoch": 0.6427963396467333, "grad_norm": 13.375, "learning_rate": 9.803119670267198e-06, "loss": 1.2592, "step": 6041 }, { "epoch": 0.64290274526495, "grad_norm": 12.1015625, "learning_rate": 9.80308413871518e-06, "loss": 1.2593, "step": 6042 }, { "epoch": 0.6430091508831667, "grad_norm": 12.46875, "learning_rate": 9.803048607163161e-06, "loss": 1.2891, "step": 6043 }, { "epoch": 0.6431155565013833, "grad_norm": 14.59375, "learning_rate": 9.803013075611143e-06, "loss": 1.5211, "step": 6044 }, { "epoch": 0.6432219621195999, "grad_norm": 12.109375, "learning_rate": 9.802977544059126e-06, "loss": 1.2197, "step": 6045 }, { "epoch": 0.6433283677378165, "grad_norm": 19.890625, "learning_rate": 9.802942012507108e-06, "loss": 1.1707, "step": 6046 }, { "epoch": 0.6434347733560332, "grad_norm": 14.0703125, "learning_rate": 9.80290648095509e-06, "loss": 1.1669, "step": 6047 }, { "epoch": 0.6435411789742499, "grad_norm": 11.0859375, "learning_rate": 9.802870949403071e-06, "loss": 1.1108, "step": 6048 }, { "epoch": 0.6436475845924665, "grad_norm": 12.046875, "learning_rate": 9.802835417851053e-06, "loss": 1.0576, "step": 6049 }, { "epoch": 0.6437539902106831, "grad_norm": 12.3515625, "learning_rate": 9.802799886299034e-06, "loss": 1.2423, "step": 6050 }, { "epoch": 0.6438603958288998, "grad_norm": 12.234375, "learning_rate": 9.802764354747016e-06, "loss": 1.3104, "step": 6051 }, { "epoch": 0.6439668014471164, "grad_norm": 11.609375, "learning_rate": 9.802728823194997e-06, "loss": 1.1994, "step": 6052 }, { "epoch": 0.644073207065333, "grad_norm": 13.546875, "learning_rate": 9.802693291642979e-06, "loss": 1.3293, "step": 6053 }, { "epoch": 0.6441796126835497, "grad_norm": 10.3125, "learning_rate": 9.80265776009096e-06, "loss": 1.1541, "step": 6054 }, { "epoch": 0.6442860183017664, "grad_norm": 12.765625, "learning_rate": 9.802622228538944e-06, "loss": 1.2228, "step": 6055 }, { "epoch": 0.644392423919983, "grad_norm": 12.515625, "learning_rate": 9.802586696986926e-06, "loss": 1.1988, "step": 6056 }, { "epoch": 0.6444988295381996, "grad_norm": 10.9609375, "learning_rate": 9.802551165434907e-06, "loss": 0.9763, "step": 6057 }, { "epoch": 0.6446052351564162, "grad_norm": 10.0, "learning_rate": 9.802515633882889e-06, "loss": 1.179, "step": 6058 }, { "epoch": 0.6447116407746329, "grad_norm": 12.4296875, "learning_rate": 9.80248010233087e-06, "loss": 1.0584, "step": 6059 }, { "epoch": 0.6448180463928496, "grad_norm": 13.375, "learning_rate": 9.802444570778854e-06, "loss": 1.2025, "step": 6060 }, { "epoch": 0.6449244520110662, "grad_norm": 13.8203125, "learning_rate": 9.802409039226835e-06, "loss": 1.367, "step": 6061 }, { "epoch": 0.6450308576292828, "grad_norm": 15.0703125, "learning_rate": 9.802373507674817e-06, "loss": 1.38, "step": 6062 }, { "epoch": 0.6451372632474994, "grad_norm": 12.4140625, "learning_rate": 9.802337976122798e-06, "loss": 1.0102, "step": 6063 }, { "epoch": 0.6452436688657162, "grad_norm": 15.5, "learning_rate": 9.80230244457078e-06, "loss": 1.0106, "step": 6064 }, { "epoch": 0.6453500744839328, "grad_norm": 12.0390625, "learning_rate": 9.802266913018762e-06, "loss": 1.1124, "step": 6065 }, { "epoch": 0.6454564801021494, "grad_norm": 13.875, "learning_rate": 9.802231381466743e-06, "loss": 1.1878, "step": 6066 }, { "epoch": 0.645562885720366, "grad_norm": 12.984375, "learning_rate": 9.802195849914725e-06, "loss": 1.2572, "step": 6067 }, { "epoch": 0.6456692913385826, "grad_norm": 17.40625, "learning_rate": 9.802160318362706e-06, "loss": 1.9728, "step": 6068 }, { "epoch": 0.6457756969567994, "grad_norm": 10.4296875, "learning_rate": 9.802124786810688e-06, "loss": 1.2093, "step": 6069 }, { "epoch": 0.645882102575016, "grad_norm": 13.453125, "learning_rate": 9.802089255258671e-06, "loss": 1.461, "step": 6070 }, { "epoch": 0.6459885081932326, "grad_norm": 11.8671875, "learning_rate": 9.802053723706653e-06, "loss": 1.3321, "step": 6071 }, { "epoch": 0.6460949138114492, "grad_norm": 13.5, "learning_rate": 9.802018192154635e-06, "loss": 1.2018, "step": 6072 }, { "epoch": 0.6462013194296659, "grad_norm": 13.359375, "learning_rate": 9.801982660602616e-06, "loss": 1.0859, "step": 6073 }, { "epoch": 0.6463077250478825, "grad_norm": 11.2265625, "learning_rate": 9.801947129050598e-06, "loss": 0.9568, "step": 6074 }, { "epoch": 0.6464141306660992, "grad_norm": 14.15625, "learning_rate": 9.80191159749858e-06, "loss": 1.2694, "step": 6075 }, { "epoch": 0.6465205362843158, "grad_norm": 15.3203125, "learning_rate": 9.801876065946561e-06, "loss": 1.6741, "step": 6076 }, { "epoch": 0.6466269419025324, "grad_norm": 14.2890625, "learning_rate": 9.801840534394543e-06, "loss": 1.4067, "step": 6077 }, { "epoch": 0.6467333475207491, "grad_norm": 16.921875, "learning_rate": 9.801805002842524e-06, "loss": 1.1378, "step": 6078 }, { "epoch": 0.6468397531389657, "grad_norm": 17.890625, "learning_rate": 9.801769471290506e-06, "loss": 1.0139, "step": 6079 }, { "epoch": 0.6469461587571824, "grad_norm": 14.3359375, "learning_rate": 9.801733939738487e-06, "loss": 1.5265, "step": 6080 }, { "epoch": 0.647052564375399, "grad_norm": 11.1875, "learning_rate": 9.80169840818647e-06, "loss": 1.4604, "step": 6081 }, { "epoch": 0.6471589699936157, "grad_norm": 12.75, "learning_rate": 9.801662876634452e-06, "loss": 1.0175, "step": 6082 }, { "epoch": 0.6472653756118323, "grad_norm": 11.9375, "learning_rate": 9.801627345082434e-06, "loss": 1.3308, "step": 6083 }, { "epoch": 0.6473717812300489, "grad_norm": 11.9765625, "learning_rate": 9.801591813530416e-06, "loss": 1.1026, "step": 6084 }, { "epoch": 0.6474781868482656, "grad_norm": 14.8125, "learning_rate": 9.801556281978397e-06, "loss": 1.2998, "step": 6085 }, { "epoch": 0.6475845924664823, "grad_norm": 13.3671875, "learning_rate": 9.80152075042638e-06, "loss": 1.2041, "step": 6086 }, { "epoch": 0.6476909980846989, "grad_norm": 12.203125, "learning_rate": 9.801485218874362e-06, "loss": 1.2594, "step": 6087 }, { "epoch": 0.6477974037029155, "grad_norm": 16.09375, "learning_rate": 9.801449687322344e-06, "loss": 1.0824, "step": 6088 }, { "epoch": 0.6479038093211321, "grad_norm": 15.5390625, "learning_rate": 9.801414155770325e-06, "loss": 1.3209, "step": 6089 }, { "epoch": 0.6480102149393488, "grad_norm": 13.6015625, "learning_rate": 9.801378624218307e-06, "loss": 1.1232, "step": 6090 }, { "epoch": 0.6481166205575655, "grad_norm": 11.6015625, "learning_rate": 9.801343092666289e-06, "loss": 1.172, "step": 6091 }, { "epoch": 0.6482230261757821, "grad_norm": 10.984375, "learning_rate": 9.80130756111427e-06, "loss": 1.0453, "step": 6092 }, { "epoch": 0.6483294317939987, "grad_norm": 14.2890625, "learning_rate": 9.801272029562252e-06, "loss": 1.1937, "step": 6093 }, { "epoch": 0.6484358374122153, "grad_norm": 16.3125, "learning_rate": 9.801236498010233e-06, "loss": 1.3323, "step": 6094 }, { "epoch": 0.648542243030432, "grad_norm": 10.8203125, "learning_rate": 9.801200966458215e-06, "loss": 1.0876, "step": 6095 }, { "epoch": 0.6486486486486487, "grad_norm": 13.3671875, "learning_rate": 9.801165434906198e-06, "loss": 1.64, "step": 6096 }, { "epoch": 0.6487550542668653, "grad_norm": 16.1875, "learning_rate": 9.80112990335418e-06, "loss": 1.0782, "step": 6097 }, { "epoch": 0.6488614598850819, "grad_norm": 12.484375, "learning_rate": 9.801094371802161e-06, "loss": 1.2959, "step": 6098 }, { "epoch": 0.6489678655032985, "grad_norm": 11.0078125, "learning_rate": 9.801058840250143e-06, "loss": 1.325, "step": 6099 }, { "epoch": 0.6490742711215152, "grad_norm": 11.90625, "learning_rate": 9.801023308698125e-06, "loss": 1.4188, "step": 6100 }, { "epoch": 0.6491806767397319, "grad_norm": 11.3828125, "learning_rate": 9.800987777146106e-06, "loss": 1.1597, "step": 6101 }, { "epoch": 0.6492870823579485, "grad_norm": 11.953125, "learning_rate": 9.800952245594088e-06, "loss": 1.3794, "step": 6102 }, { "epoch": 0.6493934879761651, "grad_norm": 14.3828125, "learning_rate": 9.80091671404207e-06, "loss": 1.3288, "step": 6103 }, { "epoch": 0.6494998935943818, "grad_norm": 10.90625, "learning_rate": 9.800881182490051e-06, "loss": 1.3579, "step": 6104 }, { "epoch": 0.6496062992125984, "grad_norm": 13.578125, "learning_rate": 9.800845650938033e-06, "loss": 1.3321, "step": 6105 }, { "epoch": 0.6497127048308151, "grad_norm": 18.921875, "learning_rate": 9.800810119386016e-06, "loss": 1.3299, "step": 6106 }, { "epoch": 0.6498191104490317, "grad_norm": 12.8359375, "learning_rate": 9.800774587833998e-06, "loss": 1.0032, "step": 6107 }, { "epoch": 0.6499255160672484, "grad_norm": 12.6171875, "learning_rate": 9.80073905628198e-06, "loss": 1.3013, "step": 6108 }, { "epoch": 0.650031921685465, "grad_norm": 9.9375, "learning_rate": 9.80070352472996e-06, "loss": 1.544, "step": 6109 }, { "epoch": 0.6501383273036816, "grad_norm": 10.8828125, "learning_rate": 9.800667993177942e-06, "loss": 1.1283, "step": 6110 }, { "epoch": 0.6502447329218983, "grad_norm": 10.859375, "learning_rate": 9.800632461625924e-06, "loss": 1.4934, "step": 6111 }, { "epoch": 0.650351138540115, "grad_norm": 12.90625, "learning_rate": 9.800596930073907e-06, "loss": 1.5893, "step": 6112 }, { "epoch": 0.6504575441583316, "grad_norm": 9.625, "learning_rate": 9.800561398521889e-06, "loss": 1.0469, "step": 6113 }, { "epoch": 0.6505639497765482, "grad_norm": 15.1484375, "learning_rate": 9.80052586696987e-06, "loss": 1.4074, "step": 6114 }, { "epoch": 0.6506703553947648, "grad_norm": 12.625, "learning_rate": 9.800490335417852e-06, "loss": 1.333, "step": 6115 }, { "epoch": 0.6507767610129814, "grad_norm": 13.453125, "learning_rate": 9.800454803865834e-06, "loss": 1.0427, "step": 6116 }, { "epoch": 0.6508831666311982, "grad_norm": 14.53125, "learning_rate": 9.800419272313815e-06, "loss": 1.1814, "step": 6117 }, { "epoch": 0.6509895722494148, "grad_norm": 22.453125, "learning_rate": 9.800383740761797e-06, "loss": 1.3536, "step": 6118 }, { "epoch": 0.6510959778676314, "grad_norm": 14.1875, "learning_rate": 9.800348209209779e-06, "loss": 1.3301, "step": 6119 }, { "epoch": 0.651202383485848, "grad_norm": 13.109375, "learning_rate": 9.80031267765776e-06, "loss": 1.2519, "step": 6120 }, { "epoch": 0.6513087891040646, "grad_norm": 17.5625, "learning_rate": 9.800277146105742e-06, "loss": 1.3263, "step": 6121 }, { "epoch": 0.6514151947222814, "grad_norm": 11.6875, "learning_rate": 9.800241614553725e-06, "loss": 0.971, "step": 6122 }, { "epoch": 0.651521600340498, "grad_norm": 15.890625, "learning_rate": 9.800206083001707e-06, "loss": 1.7521, "step": 6123 }, { "epoch": 0.6516280059587146, "grad_norm": 18.75, "learning_rate": 9.800170551449688e-06, "loss": 1.4395, "step": 6124 }, { "epoch": 0.6517344115769312, "grad_norm": 14.2421875, "learning_rate": 9.80013501989767e-06, "loss": 1.1911, "step": 6125 }, { "epoch": 0.6518408171951479, "grad_norm": 12.0, "learning_rate": 9.800099488345652e-06, "loss": 1.3673, "step": 6126 }, { "epoch": 0.6519472228133646, "grad_norm": 16.71875, "learning_rate": 9.800063956793635e-06, "loss": 1.6565, "step": 6127 }, { "epoch": 0.6520536284315812, "grad_norm": 10.78125, "learning_rate": 9.800028425241615e-06, "loss": 1.168, "step": 6128 }, { "epoch": 0.6521600340497978, "grad_norm": 14.21875, "learning_rate": 9.799992893689596e-06, "loss": 1.3087, "step": 6129 }, { "epoch": 0.6522664396680145, "grad_norm": 11.1796875, "learning_rate": 9.799957362137578e-06, "loss": 1.0348, "step": 6130 }, { "epoch": 0.6523728452862311, "grad_norm": 12.5390625, "learning_rate": 9.79992183058556e-06, "loss": 1.1699, "step": 6131 }, { "epoch": 0.6524792509044478, "grad_norm": 12.9609375, "learning_rate": 9.799886299033543e-06, "loss": 1.347, "step": 6132 }, { "epoch": 0.6525856565226644, "grad_norm": 15.171875, "learning_rate": 9.799850767481524e-06, "loss": 0.952, "step": 6133 }, { "epoch": 0.652692062140881, "grad_norm": 20.65625, "learning_rate": 9.799815235929506e-06, "loss": 1.5735, "step": 6134 }, { "epoch": 0.6527984677590977, "grad_norm": 12.5703125, "learning_rate": 9.799779704377488e-06, "loss": 1.2755, "step": 6135 }, { "epoch": 0.6529048733773143, "grad_norm": 13.8984375, "learning_rate": 9.79974417282547e-06, "loss": 1.327, "step": 6136 }, { "epoch": 0.653011278995531, "grad_norm": 18.53125, "learning_rate": 9.799708641273453e-06, "loss": 1.2759, "step": 6137 }, { "epoch": 0.6531176846137476, "grad_norm": 12.046875, "learning_rate": 9.799673109721434e-06, "loss": 1.267, "step": 6138 }, { "epoch": 0.6532240902319643, "grad_norm": 13.4921875, "learning_rate": 9.799637578169416e-06, "loss": 1.4055, "step": 6139 }, { "epoch": 0.6533304958501809, "grad_norm": 16.1875, "learning_rate": 9.799602046617397e-06, "loss": 1.4543, "step": 6140 }, { "epoch": 0.6534369014683975, "grad_norm": 13.8671875, "learning_rate": 9.799566515065379e-06, "loss": 1.3803, "step": 6141 }, { "epoch": 0.6535433070866141, "grad_norm": 11.75, "learning_rate": 9.79953098351336e-06, "loss": 1.0657, "step": 6142 }, { "epoch": 0.6536497127048309, "grad_norm": 13.046875, "learning_rate": 9.799495451961342e-06, "loss": 0.897, "step": 6143 }, { "epoch": 0.6537561183230475, "grad_norm": 11.484375, "learning_rate": 9.799459920409324e-06, "loss": 1.1896, "step": 6144 }, { "epoch": 0.6538625239412641, "grad_norm": 12.2734375, "learning_rate": 9.799424388857305e-06, "loss": 1.5096, "step": 6145 }, { "epoch": 0.6539689295594807, "grad_norm": 12.984375, "learning_rate": 9.799388857305287e-06, "loss": 1.3786, "step": 6146 }, { "epoch": 0.6540753351776973, "grad_norm": 13.1328125, "learning_rate": 9.79935332575327e-06, "loss": 1.4426, "step": 6147 }, { "epoch": 0.6541817407959141, "grad_norm": 12.578125, "learning_rate": 9.799317794201252e-06, "loss": 1.0935, "step": 6148 }, { "epoch": 0.6542881464141307, "grad_norm": 11.7421875, "learning_rate": 9.799282262649234e-06, "loss": 1.1723, "step": 6149 }, { "epoch": 0.6543945520323473, "grad_norm": 13.234375, "learning_rate": 9.799246731097215e-06, "loss": 1.1245, "step": 6150 }, { "epoch": 0.6545009576505639, "grad_norm": 12.359375, "learning_rate": 9.799211199545197e-06, "loss": 1.3081, "step": 6151 }, { "epoch": 0.6546073632687806, "grad_norm": 11.3984375, "learning_rate": 9.799175667993178e-06, "loss": 1.0623, "step": 6152 }, { "epoch": 0.6547137688869973, "grad_norm": 10.8046875, "learning_rate": 9.799140136441162e-06, "loss": 1.3342, "step": 6153 }, { "epoch": 0.6548201745052139, "grad_norm": 18.125, "learning_rate": 9.799104604889143e-06, "loss": 1.1111, "step": 6154 }, { "epoch": 0.6549265801234305, "grad_norm": 22.703125, "learning_rate": 9.799069073337125e-06, "loss": 1.1761, "step": 6155 }, { "epoch": 0.6550329857416471, "grad_norm": 12.8125, "learning_rate": 9.799033541785106e-06, "loss": 1.6788, "step": 6156 }, { "epoch": 0.6551393913598638, "grad_norm": 13.7265625, "learning_rate": 9.798998010233088e-06, "loss": 1.0985, "step": 6157 }, { "epoch": 0.6552457969780805, "grad_norm": 12.8828125, "learning_rate": 9.79896247868107e-06, "loss": 1.2374, "step": 6158 }, { "epoch": 0.6553522025962971, "grad_norm": 10.578125, "learning_rate": 9.798926947129051e-06, "loss": 1.0379, "step": 6159 }, { "epoch": 0.6554586082145137, "grad_norm": 12.8984375, "learning_rate": 9.798891415577033e-06, "loss": 1.2003, "step": 6160 }, { "epoch": 0.6555650138327304, "grad_norm": 11.4921875, "learning_rate": 9.798855884025015e-06, "loss": 1.3142, "step": 6161 }, { "epoch": 0.655671419450947, "grad_norm": 16.984375, "learning_rate": 9.798820352472996e-06, "loss": 1.0978, "step": 6162 }, { "epoch": 0.6557778250691636, "grad_norm": 13.4140625, "learning_rate": 9.79878482092098e-06, "loss": 0.972, "step": 6163 }, { "epoch": 0.6558842306873803, "grad_norm": 14.578125, "learning_rate": 9.798749289368961e-06, "loss": 1.0505, "step": 6164 }, { "epoch": 0.655990636305597, "grad_norm": 11.546875, "learning_rate": 9.798713757816943e-06, "loss": 1.203, "step": 6165 }, { "epoch": 0.6560970419238136, "grad_norm": 14.8203125, "learning_rate": 9.798678226264924e-06, "loss": 1.2995, "step": 6166 }, { "epoch": 0.6562034475420302, "grad_norm": 10.515625, "learning_rate": 9.798642694712906e-06, "loss": 1.133, "step": 6167 }, { "epoch": 0.6563098531602468, "grad_norm": 13.015625, "learning_rate": 9.798607163160887e-06, "loss": 1.2083, "step": 6168 }, { "epoch": 0.6564162587784635, "grad_norm": 15.265625, "learning_rate": 9.798571631608869e-06, "loss": 1.169, "step": 6169 }, { "epoch": 0.6565226643966802, "grad_norm": 12.6015625, "learning_rate": 9.79853610005685e-06, "loss": 1.2496, "step": 6170 }, { "epoch": 0.6566290700148968, "grad_norm": 12.5, "learning_rate": 9.798500568504832e-06, "loss": 1.0673, "step": 6171 }, { "epoch": 0.6567354756331134, "grad_norm": 10.65625, "learning_rate": 9.798465036952814e-06, "loss": 1.3866, "step": 6172 }, { "epoch": 0.65684188125133, "grad_norm": 15.515625, "learning_rate": 9.798429505400797e-06, "loss": 1.1621, "step": 6173 }, { "epoch": 0.6569482868695468, "grad_norm": 13.65625, "learning_rate": 9.798393973848779e-06, "loss": 1.3393, "step": 6174 }, { "epoch": 0.6570546924877634, "grad_norm": 25.59375, "learning_rate": 9.79835844229676e-06, "loss": 1.1758, "step": 6175 }, { "epoch": 0.65716109810598, "grad_norm": 10.9453125, "learning_rate": 9.798322910744742e-06, "loss": 1.2017, "step": 6176 }, { "epoch": 0.6572675037241966, "grad_norm": 10.2578125, "learning_rate": 9.798287379192724e-06, "loss": 1.1319, "step": 6177 }, { "epoch": 0.6573739093424132, "grad_norm": 16.328125, "learning_rate": 9.798251847640707e-06, "loss": 1.2377, "step": 6178 }, { "epoch": 0.65748031496063, "grad_norm": 13.8671875, "learning_rate": 9.798216316088689e-06, "loss": 1.2368, "step": 6179 }, { "epoch": 0.6575867205788466, "grad_norm": 16.09375, "learning_rate": 9.79818078453667e-06, "loss": 1.2594, "step": 6180 }, { "epoch": 0.6576931261970632, "grad_norm": 15.796875, "learning_rate": 9.798145252984652e-06, "loss": 1.3731, "step": 6181 }, { "epoch": 0.6577995318152798, "grad_norm": 13.5625, "learning_rate": 9.798109721432633e-06, "loss": 1.1603, "step": 6182 }, { "epoch": 0.6579059374334965, "grad_norm": 13.7578125, "learning_rate": 9.798074189880615e-06, "loss": 0.9526, "step": 6183 }, { "epoch": 0.6580123430517131, "grad_norm": 13.34375, "learning_rate": 9.798038658328597e-06, "loss": 1.027, "step": 6184 }, { "epoch": 0.6581187486699298, "grad_norm": 10.5703125, "learning_rate": 9.798003126776578e-06, "loss": 1.0183, "step": 6185 }, { "epoch": 0.6582251542881464, "grad_norm": 15.4765625, "learning_rate": 9.79796759522456e-06, "loss": 0.9753, "step": 6186 }, { "epoch": 0.658331559906363, "grad_norm": 9.734375, "learning_rate": 9.797932063672541e-06, "loss": 0.927, "step": 6187 }, { "epoch": 0.6584379655245797, "grad_norm": 13.1875, "learning_rate": 9.797896532120523e-06, "loss": 0.8034, "step": 6188 }, { "epoch": 0.6585443711427963, "grad_norm": 11.1484375, "learning_rate": 9.797861000568506e-06, "loss": 1.0677, "step": 6189 }, { "epoch": 0.658650776761013, "grad_norm": 10.671875, "learning_rate": 9.797825469016488e-06, "loss": 1.234, "step": 6190 }, { "epoch": 0.6587571823792296, "grad_norm": 12.6015625, "learning_rate": 9.79778993746447e-06, "loss": 1.3998, "step": 6191 }, { "epoch": 0.6588635879974463, "grad_norm": 14.53125, "learning_rate": 9.797754405912451e-06, "loss": 1.2611, "step": 6192 }, { "epoch": 0.6589699936156629, "grad_norm": 15.3828125, "learning_rate": 9.797718874360433e-06, "loss": 0.9954, "step": 6193 }, { "epoch": 0.6590763992338795, "grad_norm": 15.359375, "learning_rate": 9.797683342808414e-06, "loss": 1.2434, "step": 6194 }, { "epoch": 0.6591828048520962, "grad_norm": 13.5703125, "learning_rate": 9.797647811256396e-06, "loss": 1.2446, "step": 6195 }, { "epoch": 0.6592892104703129, "grad_norm": 12.5078125, "learning_rate": 9.797612279704378e-06, "loss": 1.0849, "step": 6196 }, { "epoch": 0.6593956160885295, "grad_norm": 12.28125, "learning_rate": 9.797576748152359e-06, "loss": 1.2917, "step": 6197 }, { "epoch": 0.6595020217067461, "grad_norm": 14.921875, "learning_rate": 9.79754121660034e-06, "loss": 1.2946, "step": 6198 }, { "epoch": 0.6596084273249627, "grad_norm": 11.6171875, "learning_rate": 9.797505685048324e-06, "loss": 1.2703, "step": 6199 }, { "epoch": 0.6597148329431795, "grad_norm": 13.8359375, "learning_rate": 9.797470153496306e-06, "loss": 1.4223, "step": 6200 }, { "epoch": 0.6598212385613961, "grad_norm": 14.65625, "learning_rate": 9.797434621944287e-06, "loss": 1.3302, "step": 6201 }, { "epoch": 0.6599276441796127, "grad_norm": 10.5625, "learning_rate": 9.797399090392269e-06, "loss": 1.1456, "step": 6202 }, { "epoch": 0.6600340497978293, "grad_norm": 15.0546875, "learning_rate": 9.79736355884025e-06, "loss": 1.0928, "step": 6203 }, { "epoch": 0.6601404554160459, "grad_norm": 14.9609375, "learning_rate": 9.797328027288234e-06, "loss": 1.1621, "step": 6204 }, { "epoch": 0.6602468610342626, "grad_norm": 13.375, "learning_rate": 9.797292495736215e-06, "loss": 1.2277, "step": 6205 }, { "epoch": 0.6603532666524793, "grad_norm": 17.03125, "learning_rate": 9.797256964184197e-06, "loss": 1.1877, "step": 6206 }, { "epoch": 0.6604596722706959, "grad_norm": 15.28125, "learning_rate": 9.797221432632179e-06, "loss": 1.5283, "step": 6207 }, { "epoch": 0.6605660778889125, "grad_norm": 18.828125, "learning_rate": 9.79718590108016e-06, "loss": 1.3004, "step": 6208 }, { "epoch": 0.6606724835071291, "grad_norm": 12.9921875, "learning_rate": 9.797150369528142e-06, "loss": 1.1084, "step": 6209 }, { "epoch": 0.6607788891253458, "grad_norm": 15.3671875, "learning_rate": 9.797114837976123e-06, "loss": 1.0922, "step": 6210 }, { "epoch": 0.6608852947435625, "grad_norm": 10.71875, "learning_rate": 9.797079306424105e-06, "loss": 1.1045, "step": 6211 }, { "epoch": 0.6609917003617791, "grad_norm": 13.1953125, "learning_rate": 9.797043774872087e-06, "loss": 1.0443, "step": 6212 }, { "epoch": 0.6610981059799957, "grad_norm": 11.203125, "learning_rate": 9.797008243320068e-06, "loss": 1.3027, "step": 6213 }, { "epoch": 0.6612045115982124, "grad_norm": 12.6171875, "learning_rate": 9.796972711768052e-06, "loss": 1.2454, "step": 6214 }, { "epoch": 0.661310917216429, "grad_norm": 13.6875, "learning_rate": 9.796937180216033e-06, "loss": 1.4844, "step": 6215 }, { "epoch": 0.6614173228346457, "grad_norm": 22.296875, "learning_rate": 9.796901648664015e-06, "loss": 1.351, "step": 6216 }, { "epoch": 0.6615237284528623, "grad_norm": 11.1875, "learning_rate": 9.796866117111996e-06, "loss": 1.0824, "step": 6217 }, { "epoch": 0.661630134071079, "grad_norm": 17.140625, "learning_rate": 9.796830585559978e-06, "loss": 1.1125, "step": 6218 }, { "epoch": 0.6617365396892956, "grad_norm": 14.5, "learning_rate": 9.79679505400796e-06, "loss": 1.2676, "step": 6219 }, { "epoch": 0.6618429453075122, "grad_norm": 13.390625, "learning_rate": 9.796759522455941e-06, "loss": 1.1533, "step": 6220 }, { "epoch": 0.6619493509257289, "grad_norm": 12.328125, "learning_rate": 9.796723990903923e-06, "loss": 1.1759, "step": 6221 }, { "epoch": 0.6620557565439456, "grad_norm": 15.0078125, "learning_rate": 9.796688459351904e-06, "loss": 1.3361, "step": 6222 }, { "epoch": 0.6621621621621622, "grad_norm": 13.796875, "learning_rate": 9.796652927799886e-06, "loss": 1.2652, "step": 6223 }, { "epoch": 0.6622685677803788, "grad_norm": 11.796875, "learning_rate": 9.79661739624787e-06, "loss": 1.1923, "step": 6224 }, { "epoch": 0.6623749733985954, "grad_norm": 14.3828125, "learning_rate": 9.796581864695851e-06, "loss": 1.4905, "step": 6225 }, { "epoch": 0.662481379016812, "grad_norm": 12.140625, "learning_rate": 9.796546333143833e-06, "loss": 1.1174, "step": 6226 }, { "epoch": 0.6625877846350288, "grad_norm": 15.203125, "learning_rate": 9.796510801591814e-06, "loss": 1.1972, "step": 6227 }, { "epoch": 0.6626941902532454, "grad_norm": 12.8671875, "learning_rate": 9.796475270039796e-06, "loss": 1.3025, "step": 6228 }, { "epoch": 0.662800595871462, "grad_norm": 10.359375, "learning_rate": 9.796439738487777e-06, "loss": 1.1205, "step": 6229 }, { "epoch": 0.6629070014896786, "grad_norm": 15.7265625, "learning_rate": 9.79640420693576e-06, "loss": 1.3274, "step": 6230 }, { "epoch": 0.6630134071078952, "grad_norm": 13.1015625, "learning_rate": 9.796368675383742e-06, "loss": 1.5574, "step": 6231 }, { "epoch": 0.663119812726112, "grad_norm": 12.9765625, "learning_rate": 9.796333143831724e-06, "loss": 1.3374, "step": 6232 }, { "epoch": 0.6632262183443286, "grad_norm": 13.53125, "learning_rate": 9.796297612279705e-06, "loss": 1.4502, "step": 6233 }, { "epoch": 0.6633326239625452, "grad_norm": 13.46875, "learning_rate": 9.796262080727687e-06, "loss": 1.2821, "step": 6234 }, { "epoch": 0.6634390295807618, "grad_norm": 11.5390625, "learning_rate": 9.796226549175669e-06, "loss": 1.024, "step": 6235 }, { "epoch": 0.6635454351989785, "grad_norm": 11.15625, "learning_rate": 9.79619101762365e-06, "loss": 1.0814, "step": 6236 }, { "epoch": 0.6636518408171952, "grad_norm": 8.921875, "learning_rate": 9.796155486071632e-06, "loss": 1.1731, "step": 6237 }, { "epoch": 0.6637582464354118, "grad_norm": 15.4453125, "learning_rate": 9.796119954519613e-06, "loss": 1.4284, "step": 6238 }, { "epoch": 0.6638646520536284, "grad_norm": 17.9375, "learning_rate": 9.796084422967595e-06, "loss": 1.0664, "step": 6239 }, { "epoch": 0.663971057671845, "grad_norm": 9.265625, "learning_rate": 9.796048891415578e-06, "loss": 0.9526, "step": 6240 }, { "epoch": 0.6640774632900617, "grad_norm": 12.6015625, "learning_rate": 9.79601335986356e-06, "loss": 0.9709, "step": 6241 }, { "epoch": 0.6641838689082784, "grad_norm": 12.0, "learning_rate": 9.795977828311542e-06, "loss": 1.1195, "step": 6242 }, { "epoch": 0.664290274526495, "grad_norm": 19.578125, "learning_rate": 9.795942296759523e-06, "loss": 1.7624, "step": 6243 }, { "epoch": 0.6643966801447116, "grad_norm": 9.6640625, "learning_rate": 9.795906765207505e-06, "loss": 0.9868, "step": 6244 }, { "epoch": 0.6645030857629283, "grad_norm": 18.953125, "learning_rate": 9.795871233655488e-06, "loss": 1.4149, "step": 6245 }, { "epoch": 0.6646094913811449, "grad_norm": 23.546875, "learning_rate": 9.79583570210347e-06, "loss": 1.041, "step": 6246 }, { "epoch": 0.6647158969993616, "grad_norm": 18.640625, "learning_rate": 9.795800170551451e-06, "loss": 1.368, "step": 6247 }, { "epoch": 0.6648223026175782, "grad_norm": 14.2890625, "learning_rate": 9.795764638999431e-06, "loss": 1.144, "step": 6248 }, { "epoch": 0.6649287082357949, "grad_norm": 14.4375, "learning_rate": 9.795729107447413e-06, "loss": 1.2356, "step": 6249 }, { "epoch": 0.6650351138540115, "grad_norm": 10.671875, "learning_rate": 9.795693575895396e-06, "loss": 1.1677, "step": 6250 }, { "epoch": 0.6651415194722281, "grad_norm": 12.8125, "learning_rate": 9.795658044343378e-06, "loss": 1.1579, "step": 6251 }, { "epoch": 0.6652479250904447, "grad_norm": 11.796875, "learning_rate": 9.79562251279136e-06, "loss": 1.2336, "step": 6252 }, { "epoch": 0.6653543307086615, "grad_norm": 19.15625, "learning_rate": 9.795586981239341e-06, "loss": 1.5525, "step": 6253 }, { "epoch": 0.6654607363268781, "grad_norm": 13.09375, "learning_rate": 9.795551449687323e-06, "loss": 1.1156, "step": 6254 }, { "epoch": 0.6655671419450947, "grad_norm": 13.84375, "learning_rate": 9.795515918135306e-06, "loss": 1.1751, "step": 6255 }, { "epoch": 0.6656735475633113, "grad_norm": 12.71875, "learning_rate": 9.795480386583287e-06, "loss": 1.2655, "step": 6256 }, { "epoch": 0.6657799531815279, "grad_norm": 12.9375, "learning_rate": 9.795444855031269e-06, "loss": 1.2143, "step": 6257 }, { "epoch": 0.6658863587997447, "grad_norm": 19.921875, "learning_rate": 9.79540932347925e-06, "loss": 1.1007, "step": 6258 }, { "epoch": 0.6659927644179613, "grad_norm": 16.65625, "learning_rate": 9.795373791927232e-06, "loss": 1.4305, "step": 6259 }, { "epoch": 0.6660991700361779, "grad_norm": 14.6484375, "learning_rate": 9.795338260375214e-06, "loss": 1.2273, "step": 6260 }, { "epoch": 0.6662055756543945, "grad_norm": 14.5859375, "learning_rate": 9.795302728823196e-06, "loss": 1.3756, "step": 6261 }, { "epoch": 0.6663119812726112, "grad_norm": 13.484375, "learning_rate": 9.795267197271177e-06, "loss": 1.1555, "step": 6262 }, { "epoch": 0.6664183868908279, "grad_norm": 14.1640625, "learning_rate": 9.795231665719159e-06, "loss": 1.3092, "step": 6263 }, { "epoch": 0.6665247925090445, "grad_norm": 15.1875, "learning_rate": 9.79519613416714e-06, "loss": 1.4341, "step": 6264 }, { "epoch": 0.6666311981272611, "grad_norm": 14.375, "learning_rate": 9.795160602615124e-06, "loss": 1.3765, "step": 6265 }, { "epoch": 0.6667376037454777, "grad_norm": 20.40625, "learning_rate": 9.795125071063105e-06, "loss": 1.2336, "step": 6266 }, { "epoch": 0.6668440093636944, "grad_norm": 11.3125, "learning_rate": 9.795089539511087e-06, "loss": 1.2075, "step": 6267 }, { "epoch": 0.6669504149819111, "grad_norm": 13.875, "learning_rate": 9.795054007959068e-06, "loss": 1.2524, "step": 6268 }, { "epoch": 0.6670568206001277, "grad_norm": 15.03125, "learning_rate": 9.79501847640705e-06, "loss": 1.1311, "step": 6269 }, { "epoch": 0.6671632262183443, "grad_norm": 14.53125, "learning_rate": 9.794982944855032e-06, "loss": 1.4151, "step": 6270 }, { "epoch": 0.667269631836561, "grad_norm": 12.984375, "learning_rate": 9.794947413303015e-06, "loss": 1.5113, "step": 6271 }, { "epoch": 0.6673760374547776, "grad_norm": 15.375, "learning_rate": 9.794911881750997e-06, "loss": 1.5486, "step": 6272 }, { "epoch": 0.6674824430729942, "grad_norm": 15.984375, "learning_rate": 9.794876350198978e-06, "loss": 1.2333, "step": 6273 }, { "epoch": 0.6675888486912109, "grad_norm": 13.3984375, "learning_rate": 9.79484081864696e-06, "loss": 1.1971, "step": 6274 }, { "epoch": 0.6676952543094276, "grad_norm": 10.5859375, "learning_rate": 9.794805287094941e-06, "loss": 1.2804, "step": 6275 }, { "epoch": 0.6678016599276442, "grad_norm": 11.6875, "learning_rate": 9.794769755542923e-06, "loss": 1.1378, "step": 6276 }, { "epoch": 0.6679080655458608, "grad_norm": 12.078125, "learning_rate": 9.794734223990905e-06, "loss": 1.5287, "step": 6277 }, { "epoch": 0.6680144711640774, "grad_norm": 12.6015625, "learning_rate": 9.794698692438886e-06, "loss": 1.0999, "step": 6278 }, { "epoch": 0.6681208767822941, "grad_norm": 11.1015625, "learning_rate": 9.794663160886868e-06, "loss": 1.259, "step": 6279 }, { "epoch": 0.6682272824005108, "grad_norm": 17.0625, "learning_rate": 9.79462762933485e-06, "loss": 1.6379, "step": 6280 }, { "epoch": 0.6683336880187274, "grad_norm": 17.015625, "learning_rate": 9.794592097782833e-06, "loss": 1.4595, "step": 6281 }, { "epoch": 0.668440093636944, "grad_norm": 10.8828125, "learning_rate": 9.794556566230814e-06, "loss": 1.0826, "step": 6282 }, { "epoch": 0.6685464992551606, "grad_norm": 9.2734375, "learning_rate": 9.794521034678796e-06, "loss": 0.9059, "step": 6283 }, { "epoch": 0.6686529048733774, "grad_norm": 12.7734375, "learning_rate": 9.794485503126778e-06, "loss": 1.1218, "step": 6284 }, { "epoch": 0.668759310491594, "grad_norm": 20.09375, "learning_rate": 9.79444997157476e-06, "loss": 1.4985, "step": 6285 }, { "epoch": 0.6688657161098106, "grad_norm": 15.1171875, "learning_rate": 9.79441444002274e-06, "loss": 1.277, "step": 6286 }, { "epoch": 0.6689721217280272, "grad_norm": 16.171875, "learning_rate": 9.794378908470722e-06, "loss": 1.1696, "step": 6287 }, { "epoch": 0.6690785273462438, "grad_norm": 11.1796875, "learning_rate": 9.794343376918704e-06, "loss": 1.0886, "step": 6288 }, { "epoch": 0.6691849329644606, "grad_norm": 13.75, "learning_rate": 9.794307845366686e-06, "loss": 1.8227, "step": 6289 }, { "epoch": 0.6692913385826772, "grad_norm": 16.421875, "learning_rate": 9.794272313814667e-06, "loss": 1.0345, "step": 6290 }, { "epoch": 0.6693977442008938, "grad_norm": 13.078125, "learning_rate": 9.79423678226265e-06, "loss": 1.1451, "step": 6291 }, { "epoch": 0.6695041498191104, "grad_norm": 13.3671875, "learning_rate": 9.794201250710632e-06, "loss": 1.0516, "step": 6292 }, { "epoch": 0.6696105554373271, "grad_norm": 13.28125, "learning_rate": 9.794165719158614e-06, "loss": 1.1733, "step": 6293 }, { "epoch": 0.6697169610555437, "grad_norm": 11.71875, "learning_rate": 9.794130187606595e-06, "loss": 1.4003, "step": 6294 }, { "epoch": 0.6698233666737604, "grad_norm": 14.7109375, "learning_rate": 9.794094656054577e-06, "loss": 1.0883, "step": 6295 }, { "epoch": 0.669929772291977, "grad_norm": 15.9296875, "learning_rate": 9.79405912450256e-06, "loss": 1.1701, "step": 6296 }, { "epoch": 0.6700361779101937, "grad_norm": 10.78125, "learning_rate": 9.794023592950542e-06, "loss": 1.1284, "step": 6297 }, { "epoch": 0.6701425835284103, "grad_norm": 10.6171875, "learning_rate": 9.793988061398523e-06, "loss": 1.0891, "step": 6298 }, { "epoch": 0.6702489891466269, "grad_norm": 14.0078125, "learning_rate": 9.793952529846505e-06, "loss": 1.3039, "step": 6299 }, { "epoch": 0.6703553947648436, "grad_norm": 15.5078125, "learning_rate": 9.793916998294487e-06, "loss": 1.182, "step": 6300 }, { "epoch": 0.6704618003830602, "grad_norm": 11.9765625, "learning_rate": 9.793881466742468e-06, "loss": 1.2242, "step": 6301 }, { "epoch": 0.6705682060012769, "grad_norm": 12.515625, "learning_rate": 9.79384593519045e-06, "loss": 1.2229, "step": 6302 }, { "epoch": 0.6706746116194935, "grad_norm": 13.078125, "learning_rate": 9.793810403638431e-06, "loss": 1.1431, "step": 6303 }, { "epoch": 0.6707810172377101, "grad_norm": 14.5703125, "learning_rate": 9.793774872086413e-06, "loss": 1.2601, "step": 6304 }, { "epoch": 0.6708874228559268, "grad_norm": 13.265625, "learning_rate": 9.793739340534395e-06, "loss": 1.5972, "step": 6305 }, { "epoch": 0.6709938284741435, "grad_norm": 12.75, "learning_rate": 9.793703808982376e-06, "loss": 1.307, "step": 6306 }, { "epoch": 0.6711002340923601, "grad_norm": 13.3984375, "learning_rate": 9.79366827743036e-06, "loss": 1.2117, "step": 6307 }, { "epoch": 0.6712066397105767, "grad_norm": 13.9765625, "learning_rate": 9.793632745878341e-06, "loss": 1.3279, "step": 6308 }, { "epoch": 0.6713130453287933, "grad_norm": 14.2265625, "learning_rate": 9.793597214326323e-06, "loss": 1.5214, "step": 6309 }, { "epoch": 0.67141945094701, "grad_norm": 11.1484375, "learning_rate": 9.793561682774304e-06, "loss": 1.2217, "step": 6310 }, { "epoch": 0.6715258565652267, "grad_norm": 13.0078125, "learning_rate": 9.793526151222286e-06, "loss": 1.5833, "step": 6311 }, { "epoch": 0.6716322621834433, "grad_norm": 12.078125, "learning_rate": 9.793490619670268e-06, "loss": 1.2813, "step": 6312 }, { "epoch": 0.6717386678016599, "grad_norm": 13.59375, "learning_rate": 9.79345508811825e-06, "loss": 1.4543, "step": 6313 }, { "epoch": 0.6718450734198765, "grad_norm": 9.8359375, "learning_rate": 9.79341955656623e-06, "loss": 1.2925, "step": 6314 }, { "epoch": 0.6719514790380932, "grad_norm": 13.84375, "learning_rate": 9.793384025014212e-06, "loss": 1.4299, "step": 6315 }, { "epoch": 0.6720578846563099, "grad_norm": 10.28125, "learning_rate": 9.793348493462194e-06, "loss": 1.3806, "step": 6316 }, { "epoch": 0.6721642902745265, "grad_norm": 17.390625, "learning_rate": 9.793312961910177e-06, "loss": 1.028, "step": 6317 }, { "epoch": 0.6722706958927431, "grad_norm": 13.375, "learning_rate": 9.793277430358159e-06, "loss": 1.0646, "step": 6318 }, { "epoch": 0.6723771015109598, "grad_norm": 13.3359375, "learning_rate": 9.79324189880614e-06, "loss": 1.3023, "step": 6319 }, { "epoch": 0.6724835071291764, "grad_norm": 12.46875, "learning_rate": 9.793206367254122e-06, "loss": 1.3222, "step": 6320 }, { "epoch": 0.6725899127473931, "grad_norm": 12.7109375, "learning_rate": 9.793170835702104e-06, "loss": 1.232, "step": 6321 }, { "epoch": 0.6726963183656097, "grad_norm": 9.5859375, "learning_rate": 9.793135304150087e-06, "loss": 0.8843, "step": 6322 }, { "epoch": 0.6728027239838263, "grad_norm": 14.484375, "learning_rate": 9.793099772598069e-06, "loss": 1.0565, "step": 6323 }, { "epoch": 0.672909129602043, "grad_norm": 12.109375, "learning_rate": 9.79306424104605e-06, "loss": 1.2105, "step": 6324 }, { "epoch": 0.6730155352202596, "grad_norm": 13.375, "learning_rate": 9.793028709494032e-06, "loss": 1.2431, "step": 6325 }, { "epoch": 0.6731219408384763, "grad_norm": 10.0, "learning_rate": 9.792993177942014e-06, "loss": 1.1323, "step": 6326 }, { "epoch": 0.6732283464566929, "grad_norm": 12.71875, "learning_rate": 9.792957646389995e-06, "loss": 1.0823, "step": 6327 }, { "epoch": 0.6733347520749096, "grad_norm": 11.8515625, "learning_rate": 9.792922114837977e-06, "loss": 1.1997, "step": 6328 }, { "epoch": 0.6734411576931262, "grad_norm": 19.90625, "learning_rate": 9.792886583285958e-06, "loss": 1.0989, "step": 6329 }, { "epoch": 0.6735475633113428, "grad_norm": 12.859375, "learning_rate": 9.79285105173394e-06, "loss": 0.7358, "step": 6330 }, { "epoch": 0.6736539689295595, "grad_norm": 12.9921875, "learning_rate": 9.792815520181922e-06, "loss": 1.465, "step": 6331 }, { "epoch": 0.6737603745477762, "grad_norm": 19.15625, "learning_rate": 9.792779988629905e-06, "loss": 1.3849, "step": 6332 }, { "epoch": 0.6738667801659928, "grad_norm": 12.59375, "learning_rate": 9.792744457077886e-06, "loss": 1.1072, "step": 6333 }, { "epoch": 0.6739731857842094, "grad_norm": 16.4375, "learning_rate": 9.792708925525868e-06, "loss": 1.4462, "step": 6334 }, { "epoch": 0.674079591402426, "grad_norm": 16.046875, "learning_rate": 9.79267339397385e-06, "loss": 1.4367, "step": 6335 }, { "epoch": 0.6741859970206426, "grad_norm": 15.3359375, "learning_rate": 9.792637862421831e-06, "loss": 1.1521, "step": 6336 }, { "epoch": 0.6742924026388594, "grad_norm": 11.46875, "learning_rate": 9.792602330869813e-06, "loss": 1.4097, "step": 6337 }, { "epoch": 0.674398808257076, "grad_norm": 13.9453125, "learning_rate": 9.792566799317794e-06, "loss": 0.7396, "step": 6338 }, { "epoch": 0.6745052138752926, "grad_norm": 12.6875, "learning_rate": 9.792531267765776e-06, "loss": 1.3669, "step": 6339 }, { "epoch": 0.6746116194935092, "grad_norm": 11.703125, "learning_rate": 9.792495736213758e-06, "loss": 1.0581, "step": 6340 }, { "epoch": 0.6747180251117258, "grad_norm": 13.9921875, "learning_rate": 9.79246020466174e-06, "loss": 1.3041, "step": 6341 }, { "epoch": 0.6748244307299426, "grad_norm": 15.515625, "learning_rate": 9.792424673109723e-06, "loss": 1.1359, "step": 6342 }, { "epoch": 0.6749308363481592, "grad_norm": 10.625, "learning_rate": 9.792389141557704e-06, "loss": 1.1131, "step": 6343 }, { "epoch": 0.6750372419663758, "grad_norm": 13.015625, "learning_rate": 9.792353610005686e-06, "loss": 1.2453, "step": 6344 }, { "epoch": 0.6751436475845924, "grad_norm": 14.296875, "learning_rate": 9.792318078453667e-06, "loss": 1.1352, "step": 6345 }, { "epoch": 0.6752500532028091, "grad_norm": 15.53125, "learning_rate": 9.792282546901649e-06, "loss": 1.2925, "step": 6346 }, { "epoch": 0.6753564588210258, "grad_norm": 13.015625, "learning_rate": 9.79224701534963e-06, "loss": 1.2634, "step": 6347 }, { "epoch": 0.6754628644392424, "grad_norm": 11.7578125, "learning_rate": 9.792211483797614e-06, "loss": 1.5358, "step": 6348 }, { "epoch": 0.675569270057459, "grad_norm": 16.171875, "learning_rate": 9.792175952245596e-06, "loss": 1.2466, "step": 6349 }, { "epoch": 0.6756756756756757, "grad_norm": 23.15625, "learning_rate": 9.792140420693577e-06, "loss": 1.1521, "step": 6350 }, { "epoch": 0.6757820812938923, "grad_norm": 10.0, "learning_rate": 9.792104889141559e-06, "loss": 1.2574, "step": 6351 }, { "epoch": 0.675888486912109, "grad_norm": 10.96875, "learning_rate": 9.79206935758954e-06, "loss": 1.176, "step": 6352 }, { "epoch": 0.6759948925303256, "grad_norm": 17.1875, "learning_rate": 9.792033826037522e-06, "loss": 1.2646, "step": 6353 }, { "epoch": 0.6761012981485423, "grad_norm": 15.9140625, "learning_rate": 9.791998294485504e-06, "loss": 1.7073, "step": 6354 }, { "epoch": 0.6762077037667589, "grad_norm": 13.296875, "learning_rate": 9.791962762933485e-06, "loss": 1.1798, "step": 6355 }, { "epoch": 0.6763141093849755, "grad_norm": 12.8046875, "learning_rate": 9.791927231381467e-06, "loss": 1.5235, "step": 6356 }, { "epoch": 0.6764205150031921, "grad_norm": 12.359375, "learning_rate": 9.791891699829448e-06, "loss": 1.4818, "step": 6357 }, { "epoch": 0.6765269206214088, "grad_norm": 11.2421875, "learning_rate": 9.791856168277432e-06, "loss": 1.1937, "step": 6358 }, { "epoch": 0.6766333262396255, "grad_norm": 18.25, "learning_rate": 9.791820636725413e-06, "loss": 1.3242, "step": 6359 }, { "epoch": 0.6767397318578421, "grad_norm": 10.671875, "learning_rate": 9.791785105173395e-06, "loss": 1.0948, "step": 6360 }, { "epoch": 0.6768461374760587, "grad_norm": 16.59375, "learning_rate": 9.791749573621377e-06, "loss": 1.3626, "step": 6361 }, { "epoch": 0.6769525430942753, "grad_norm": 12.7421875, "learning_rate": 9.791714042069358e-06, "loss": 1.1703, "step": 6362 }, { "epoch": 0.6770589487124921, "grad_norm": 12.1640625, "learning_rate": 9.791678510517341e-06, "loss": 1.2745, "step": 6363 }, { "epoch": 0.6771653543307087, "grad_norm": 11.15625, "learning_rate": 9.791642978965323e-06, "loss": 1.3234, "step": 6364 }, { "epoch": 0.6772717599489253, "grad_norm": 13.828125, "learning_rate": 9.791607447413305e-06, "loss": 1.26, "step": 6365 }, { "epoch": 0.6773781655671419, "grad_norm": 10.875, "learning_rate": 9.791571915861286e-06, "loss": 1.3338, "step": 6366 }, { "epoch": 0.6774845711853585, "grad_norm": 16.796875, "learning_rate": 9.791536384309266e-06, "loss": 1.1252, "step": 6367 }, { "epoch": 0.6775909768035753, "grad_norm": 10.6953125, "learning_rate": 9.79150085275725e-06, "loss": 1.0811, "step": 6368 }, { "epoch": 0.6776973824217919, "grad_norm": 14.7109375, "learning_rate": 9.791465321205231e-06, "loss": 0.9813, "step": 6369 }, { "epoch": 0.6778037880400085, "grad_norm": 14.0859375, "learning_rate": 9.791429789653213e-06, "loss": 1.673, "step": 6370 }, { "epoch": 0.6779101936582251, "grad_norm": 15.359375, "learning_rate": 9.791394258101194e-06, "loss": 1.0383, "step": 6371 }, { "epoch": 0.6780165992764418, "grad_norm": 14.09375, "learning_rate": 9.791358726549176e-06, "loss": 1.2206, "step": 6372 }, { "epoch": 0.6781230048946585, "grad_norm": 20.1875, "learning_rate": 9.79132319499716e-06, "loss": 1.3264, "step": 6373 }, { "epoch": 0.6782294105128751, "grad_norm": 11.1484375, "learning_rate": 9.79128766344514e-06, "loss": 1.1853, "step": 6374 }, { "epoch": 0.6783358161310917, "grad_norm": 12.4140625, "learning_rate": 9.791252131893122e-06, "loss": 0.98, "step": 6375 }, { "epoch": 0.6784422217493083, "grad_norm": 11.8984375, "learning_rate": 9.791216600341104e-06, "loss": 1.0287, "step": 6376 }, { "epoch": 0.678548627367525, "grad_norm": 11.140625, "learning_rate": 9.791181068789086e-06, "loss": 0.9174, "step": 6377 }, { "epoch": 0.6786550329857417, "grad_norm": 10.6640625, "learning_rate": 9.791145537237067e-06, "loss": 0.9794, "step": 6378 }, { "epoch": 0.6787614386039583, "grad_norm": 10.2578125, "learning_rate": 9.791110005685049e-06, "loss": 1.1735, "step": 6379 }, { "epoch": 0.6788678442221749, "grad_norm": 16.0625, "learning_rate": 9.79107447413303e-06, "loss": 1.1971, "step": 6380 }, { "epoch": 0.6789742498403916, "grad_norm": 21.40625, "learning_rate": 9.791038942581012e-06, "loss": 1.1579, "step": 6381 }, { "epoch": 0.6790806554586082, "grad_norm": 11.9140625, "learning_rate": 9.791003411028994e-06, "loss": 1.1841, "step": 6382 }, { "epoch": 0.6791870610768248, "grad_norm": 10.8125, "learning_rate": 9.790967879476975e-06, "loss": 1.288, "step": 6383 }, { "epoch": 0.6792934666950415, "grad_norm": 16.0625, "learning_rate": 9.790932347924959e-06, "loss": 1.3429, "step": 6384 }, { "epoch": 0.6793998723132582, "grad_norm": 9.9609375, "learning_rate": 9.79089681637294e-06, "loss": 1.1055, "step": 6385 }, { "epoch": 0.6795062779314748, "grad_norm": 13.7109375, "learning_rate": 9.790861284820922e-06, "loss": 1.2547, "step": 6386 }, { "epoch": 0.6796126835496914, "grad_norm": 13.375, "learning_rate": 9.790825753268903e-06, "loss": 1.4599, "step": 6387 }, { "epoch": 0.679719089167908, "grad_norm": 15.0234375, "learning_rate": 9.790790221716885e-06, "loss": 1.0466, "step": 6388 }, { "epoch": 0.6798254947861248, "grad_norm": 18.21875, "learning_rate": 9.790754690164868e-06, "loss": 1.4205, "step": 6389 }, { "epoch": 0.6799319004043414, "grad_norm": 11.53125, "learning_rate": 9.79071915861285e-06, "loss": 1.0901, "step": 6390 }, { "epoch": 0.680038306022558, "grad_norm": 22.5, "learning_rate": 9.790683627060831e-06, "loss": 1.1272, "step": 6391 }, { "epoch": 0.6801447116407746, "grad_norm": 11.015625, "learning_rate": 9.790648095508813e-06, "loss": 1.2367, "step": 6392 }, { "epoch": 0.6802511172589912, "grad_norm": 13.6015625, "learning_rate": 9.790612563956795e-06, "loss": 1.4414, "step": 6393 }, { "epoch": 0.680357522877208, "grad_norm": 13.8125, "learning_rate": 9.790577032404776e-06, "loss": 1.2813, "step": 6394 }, { "epoch": 0.6804639284954246, "grad_norm": 11.484375, "learning_rate": 9.790541500852758e-06, "loss": 1.0831, "step": 6395 }, { "epoch": 0.6805703341136412, "grad_norm": 12.6328125, "learning_rate": 9.79050596930074e-06, "loss": 1.1361, "step": 6396 }, { "epoch": 0.6806767397318578, "grad_norm": 15.90625, "learning_rate": 9.790470437748721e-06, "loss": 1.2176, "step": 6397 }, { "epoch": 0.6807831453500744, "grad_norm": 11.8984375, "learning_rate": 9.790434906196703e-06, "loss": 1.1651, "step": 6398 }, { "epoch": 0.6808895509682912, "grad_norm": 14.578125, "learning_rate": 9.790399374644686e-06, "loss": 1.1712, "step": 6399 }, { "epoch": 0.6809959565865078, "grad_norm": 10.8828125, "learning_rate": 9.790363843092668e-06, "loss": 1.3078, "step": 6400 }, { "epoch": 0.6811023622047244, "grad_norm": 10.40625, "learning_rate": 9.79032831154065e-06, "loss": 0.9469, "step": 6401 }, { "epoch": 0.681208767822941, "grad_norm": 12.34375, "learning_rate": 9.790292779988631e-06, "loss": 1.1655, "step": 6402 }, { "epoch": 0.6813151734411577, "grad_norm": 14.453125, "learning_rate": 9.790257248436612e-06, "loss": 1.6377, "step": 6403 }, { "epoch": 0.6814215790593743, "grad_norm": 17.984375, "learning_rate": 9.790221716884594e-06, "loss": 1.7228, "step": 6404 }, { "epoch": 0.681527984677591, "grad_norm": 13.2109375, "learning_rate": 9.790186185332576e-06, "loss": 1.1688, "step": 6405 }, { "epoch": 0.6816343902958076, "grad_norm": 11.515625, "learning_rate": 9.790150653780557e-06, "loss": 1.2565, "step": 6406 }, { "epoch": 0.6817407959140243, "grad_norm": 13.59375, "learning_rate": 9.790115122228539e-06, "loss": 1.2342, "step": 6407 }, { "epoch": 0.6818472015322409, "grad_norm": 17.484375, "learning_rate": 9.79007959067652e-06, "loss": 1.2307, "step": 6408 }, { "epoch": 0.6819536071504575, "grad_norm": 14.953125, "learning_rate": 9.790044059124504e-06, "loss": 1.5499, "step": 6409 }, { "epoch": 0.6820600127686742, "grad_norm": 10.15625, "learning_rate": 9.790008527572485e-06, "loss": 1.0868, "step": 6410 }, { "epoch": 0.6821664183868908, "grad_norm": 16.59375, "learning_rate": 9.789972996020467e-06, "loss": 1.2765, "step": 6411 }, { "epoch": 0.6822728240051075, "grad_norm": 12.5625, "learning_rate": 9.789937464468449e-06, "loss": 1.178, "step": 6412 }, { "epoch": 0.6823792296233241, "grad_norm": 12.71875, "learning_rate": 9.78990193291643e-06, "loss": 1.1005, "step": 6413 }, { "epoch": 0.6824856352415407, "grad_norm": 16.546875, "learning_rate": 9.789866401364414e-06, "loss": 1.3333, "step": 6414 }, { "epoch": 0.6825920408597574, "grad_norm": 10.546875, "learning_rate": 9.789830869812395e-06, "loss": 0.9433, "step": 6415 }, { "epoch": 0.6826984464779741, "grad_norm": 13.015625, "learning_rate": 9.789795338260377e-06, "loss": 0.9296, "step": 6416 }, { "epoch": 0.6828048520961907, "grad_norm": 11.203125, "learning_rate": 9.789759806708358e-06, "loss": 1.3552, "step": 6417 }, { "epoch": 0.6829112577144073, "grad_norm": 12.4765625, "learning_rate": 9.78972427515634e-06, "loss": 1.0689, "step": 6418 }, { "epoch": 0.6830176633326239, "grad_norm": 11.1875, "learning_rate": 9.789688743604322e-06, "loss": 1.3328, "step": 6419 }, { "epoch": 0.6831240689508407, "grad_norm": 11.0, "learning_rate": 9.789653212052303e-06, "loss": 1.0431, "step": 6420 }, { "epoch": 0.6832304745690573, "grad_norm": 15.484375, "learning_rate": 9.789617680500285e-06, "loss": 1.5686, "step": 6421 }, { "epoch": 0.6833368801872739, "grad_norm": 9.703125, "learning_rate": 9.789582148948266e-06, "loss": 1.0701, "step": 6422 }, { "epoch": 0.6834432858054905, "grad_norm": 11.5625, "learning_rate": 9.789546617396248e-06, "loss": 1.1683, "step": 6423 }, { "epoch": 0.6835496914237071, "grad_norm": 15.03125, "learning_rate": 9.78951108584423e-06, "loss": 1.2826, "step": 6424 }, { "epoch": 0.6836560970419238, "grad_norm": 14.0078125, "learning_rate": 9.789475554292213e-06, "loss": 1.1833, "step": 6425 }, { "epoch": 0.6837625026601405, "grad_norm": 13.4296875, "learning_rate": 9.789440022740195e-06, "loss": 1.0362, "step": 6426 }, { "epoch": 0.6838689082783571, "grad_norm": 16.78125, "learning_rate": 9.789404491188176e-06, "loss": 1.2728, "step": 6427 }, { "epoch": 0.6839753138965737, "grad_norm": 11.4140625, "learning_rate": 9.789368959636158e-06, "loss": 0.7698, "step": 6428 }, { "epoch": 0.6840817195147904, "grad_norm": 11.78125, "learning_rate": 9.78933342808414e-06, "loss": 1.3744, "step": 6429 }, { "epoch": 0.684188125133007, "grad_norm": 11.6640625, "learning_rate": 9.789297896532121e-06, "loss": 1.0349, "step": 6430 }, { "epoch": 0.6842945307512237, "grad_norm": 15.375, "learning_rate": 9.789262364980103e-06, "loss": 1.4343, "step": 6431 }, { "epoch": 0.6844009363694403, "grad_norm": 9.84375, "learning_rate": 9.789226833428084e-06, "loss": 1.2623, "step": 6432 }, { "epoch": 0.684507341987657, "grad_norm": 15.828125, "learning_rate": 9.789191301876066e-06, "loss": 1.2258, "step": 6433 }, { "epoch": 0.6846137476058736, "grad_norm": 11.9453125, "learning_rate": 9.789155770324047e-06, "loss": 1.0582, "step": 6434 }, { "epoch": 0.6847201532240902, "grad_norm": 10.265625, "learning_rate": 9.78912023877203e-06, "loss": 1.37, "step": 6435 }, { "epoch": 0.6848265588423069, "grad_norm": 14.4375, "learning_rate": 9.789084707220012e-06, "loss": 1.1904, "step": 6436 }, { "epoch": 0.6849329644605235, "grad_norm": 14.09375, "learning_rate": 9.789049175667994e-06, "loss": 1.1355, "step": 6437 }, { "epoch": 0.6850393700787402, "grad_norm": 11.359375, "learning_rate": 9.789013644115975e-06, "loss": 1.416, "step": 6438 }, { "epoch": 0.6851457756969568, "grad_norm": 11.4921875, "learning_rate": 9.788978112563957e-06, "loss": 1.2144, "step": 6439 }, { "epoch": 0.6852521813151734, "grad_norm": 14.28125, "learning_rate": 9.78894258101194e-06, "loss": 1.4311, "step": 6440 }, { "epoch": 0.6853585869333901, "grad_norm": 11.1875, "learning_rate": 9.788907049459922e-06, "loss": 1.3298, "step": 6441 }, { "epoch": 0.6854649925516068, "grad_norm": 13.28125, "learning_rate": 9.788871517907904e-06, "loss": 1.1334, "step": 6442 }, { "epoch": 0.6855713981698234, "grad_norm": 11.9921875, "learning_rate": 9.788835986355885e-06, "loss": 1.4074, "step": 6443 }, { "epoch": 0.68567780378804, "grad_norm": 13.3515625, "learning_rate": 9.788800454803867e-06, "loss": 1.2057, "step": 6444 }, { "epoch": 0.6857842094062566, "grad_norm": 17.59375, "learning_rate": 9.788764923251848e-06, "loss": 1.1071, "step": 6445 }, { "epoch": 0.6858906150244732, "grad_norm": 14.1953125, "learning_rate": 9.78872939169983e-06, "loss": 1.3292, "step": 6446 }, { "epoch": 0.68599702064269, "grad_norm": 10.0703125, "learning_rate": 9.788693860147812e-06, "loss": 1.1458, "step": 6447 }, { "epoch": 0.6861034262609066, "grad_norm": 12.3828125, "learning_rate": 9.788658328595793e-06, "loss": 1.2231, "step": 6448 }, { "epoch": 0.6862098318791232, "grad_norm": 9.4140625, "learning_rate": 9.788622797043775e-06, "loss": 1.1797, "step": 6449 }, { "epoch": 0.6863162374973398, "grad_norm": 12.0859375, "learning_rate": 9.788587265491758e-06, "loss": 1.1518, "step": 6450 }, { "epoch": 0.6864226431155565, "grad_norm": 15.0859375, "learning_rate": 9.78855173393974e-06, "loss": 1.3523, "step": 6451 }, { "epoch": 0.6865290487337732, "grad_norm": 14.0859375, "learning_rate": 9.788516202387721e-06, "loss": 1.3636, "step": 6452 }, { "epoch": 0.6866354543519898, "grad_norm": 10.9296875, "learning_rate": 9.788480670835703e-06, "loss": 1.2489, "step": 6453 }, { "epoch": 0.6867418599702064, "grad_norm": 12.640625, "learning_rate": 9.788445139283685e-06, "loss": 1.0625, "step": 6454 }, { "epoch": 0.686848265588423, "grad_norm": 16.390625, "learning_rate": 9.788409607731668e-06, "loss": 1.169, "step": 6455 }, { "epoch": 0.6869546712066397, "grad_norm": 10.703125, "learning_rate": 9.788374076179648e-06, "loss": 1.1112, "step": 6456 }, { "epoch": 0.6870610768248564, "grad_norm": 12.65625, "learning_rate": 9.78833854462763e-06, "loss": 1.2208, "step": 6457 }, { "epoch": 0.687167482443073, "grad_norm": 11.6015625, "learning_rate": 9.788303013075611e-06, "loss": 0.9035, "step": 6458 }, { "epoch": 0.6872738880612896, "grad_norm": 11.359375, "learning_rate": 9.788267481523593e-06, "loss": 1.2356, "step": 6459 }, { "epoch": 0.6873802936795063, "grad_norm": 38.21875, "learning_rate": 9.788231949971576e-06, "loss": 1.0536, "step": 6460 }, { "epoch": 0.6874866992977229, "grad_norm": 14.5703125, "learning_rate": 9.788196418419558e-06, "loss": 0.7448, "step": 6461 }, { "epoch": 0.6875931049159396, "grad_norm": 14.3671875, "learning_rate": 9.788160886867539e-06, "loss": 1.2702, "step": 6462 }, { "epoch": 0.6876995105341562, "grad_norm": 10.4375, "learning_rate": 9.78812535531552e-06, "loss": 1.3249, "step": 6463 }, { "epoch": 0.6878059161523729, "grad_norm": 7.64453125, "learning_rate": 9.788089823763502e-06, "loss": 0.8324, "step": 6464 }, { "epoch": 0.6879123217705895, "grad_norm": 8.2265625, "learning_rate": 9.788054292211484e-06, "loss": 0.9974, "step": 6465 }, { "epoch": 0.6880187273888061, "grad_norm": 13.1015625, "learning_rate": 9.788018760659467e-06, "loss": 1.1944, "step": 6466 }, { "epoch": 0.6881251330070227, "grad_norm": 17.171875, "learning_rate": 9.787983229107449e-06, "loss": 1.1312, "step": 6467 }, { "epoch": 0.6882315386252394, "grad_norm": 18.5625, "learning_rate": 9.78794769755543e-06, "loss": 1.1015, "step": 6468 }, { "epoch": 0.6883379442434561, "grad_norm": 12.34375, "learning_rate": 9.787912166003412e-06, "loss": 1.2316, "step": 6469 }, { "epoch": 0.6884443498616727, "grad_norm": 14.2421875, "learning_rate": 9.787876634451394e-06, "loss": 1.1746, "step": 6470 }, { "epoch": 0.6885507554798893, "grad_norm": 16.78125, "learning_rate": 9.787841102899375e-06, "loss": 1.0522, "step": 6471 }, { "epoch": 0.6886571610981059, "grad_norm": 13.4140625, "learning_rate": 9.787805571347357e-06, "loss": 1.032, "step": 6472 }, { "epoch": 0.6887635667163227, "grad_norm": 16.03125, "learning_rate": 9.787770039795338e-06, "loss": 1.4024, "step": 6473 }, { "epoch": 0.6888699723345393, "grad_norm": 11.875, "learning_rate": 9.78773450824332e-06, "loss": 1.2089, "step": 6474 }, { "epoch": 0.6889763779527559, "grad_norm": 10.7734375, "learning_rate": 9.787698976691302e-06, "loss": 1.1538, "step": 6475 }, { "epoch": 0.6890827835709725, "grad_norm": 15.8515625, "learning_rate": 9.787663445139285e-06, "loss": 1.312, "step": 6476 }, { "epoch": 0.6891891891891891, "grad_norm": 15.078125, "learning_rate": 9.787627913587267e-06, "loss": 1.3531, "step": 6477 }, { "epoch": 0.6892955948074059, "grad_norm": 15.9140625, "learning_rate": 9.787592382035248e-06, "loss": 1.312, "step": 6478 }, { "epoch": 0.6894020004256225, "grad_norm": 12.6796875, "learning_rate": 9.78755685048323e-06, "loss": 1.6453, "step": 6479 }, { "epoch": 0.6895084060438391, "grad_norm": 15.578125, "learning_rate": 9.787521318931211e-06, "loss": 1.3716, "step": 6480 }, { "epoch": 0.6896148116620557, "grad_norm": 12.484375, "learning_rate": 9.787485787379195e-06, "loss": 1.1409, "step": 6481 }, { "epoch": 0.6897212172802724, "grad_norm": 17.828125, "learning_rate": 9.787450255827176e-06, "loss": 1.2373, "step": 6482 }, { "epoch": 0.6898276228984891, "grad_norm": 11.328125, "learning_rate": 9.787414724275158e-06, "loss": 0.9934, "step": 6483 }, { "epoch": 0.6899340285167057, "grad_norm": 13.1484375, "learning_rate": 9.78737919272314e-06, "loss": 1.3151, "step": 6484 }, { "epoch": 0.6900404341349223, "grad_norm": 13.578125, "learning_rate": 9.787343661171121e-06, "loss": 1.2432, "step": 6485 }, { "epoch": 0.690146839753139, "grad_norm": 14.53125, "learning_rate": 9.787308129619103e-06, "loss": 1.3949, "step": 6486 }, { "epoch": 0.6902532453713556, "grad_norm": 17.453125, "learning_rate": 9.787272598067084e-06, "loss": 1.6006, "step": 6487 }, { "epoch": 0.6903596509895723, "grad_norm": 13.1796875, "learning_rate": 9.787237066515066e-06, "loss": 1.0219, "step": 6488 }, { "epoch": 0.6904660566077889, "grad_norm": 16.90625, "learning_rate": 9.787201534963048e-06, "loss": 1.3406, "step": 6489 }, { "epoch": 0.6905724622260055, "grad_norm": 16.109375, "learning_rate": 9.78716600341103e-06, "loss": 1.2923, "step": 6490 }, { "epoch": 0.6906788678442222, "grad_norm": 14.4609375, "learning_rate": 9.787130471859012e-06, "loss": 1.0535, "step": 6491 }, { "epoch": 0.6907852734624388, "grad_norm": 11.0625, "learning_rate": 9.787094940306994e-06, "loss": 1.4893, "step": 6492 }, { "epoch": 0.6908916790806554, "grad_norm": 13.140625, "learning_rate": 9.787059408754976e-06, "loss": 1.234, "step": 6493 }, { "epoch": 0.6909980846988721, "grad_norm": 13.328125, "learning_rate": 9.787023877202957e-06, "loss": 1.3359, "step": 6494 }, { "epoch": 0.6911044903170888, "grad_norm": 12.0625, "learning_rate": 9.786988345650939e-06, "loss": 1.1916, "step": 6495 }, { "epoch": 0.6912108959353054, "grad_norm": 11.78125, "learning_rate": 9.78695281409892e-06, "loss": 0.9472, "step": 6496 }, { "epoch": 0.691317301553522, "grad_norm": 12.3671875, "learning_rate": 9.786917282546902e-06, "loss": 1.2907, "step": 6497 }, { "epoch": 0.6914237071717386, "grad_norm": 11.359375, "learning_rate": 9.786881750994884e-06, "loss": 1.1799, "step": 6498 }, { "epoch": 0.6915301127899554, "grad_norm": 16.03125, "learning_rate": 9.786846219442865e-06, "loss": 1.2358, "step": 6499 }, { "epoch": 0.691636518408172, "grad_norm": 10.1640625, "learning_rate": 9.786810687890847e-06, "loss": 1.195, "step": 6500 }, { "epoch": 0.6917429240263886, "grad_norm": 17.3125, "learning_rate": 9.786775156338829e-06, "loss": 1.3331, "step": 6501 }, { "epoch": 0.6918493296446052, "grad_norm": 10.5625, "learning_rate": 9.786739624786812e-06, "loss": 1.2271, "step": 6502 }, { "epoch": 0.6919557352628218, "grad_norm": 16.71875, "learning_rate": 9.786704093234793e-06, "loss": 1.5053, "step": 6503 }, { "epoch": 0.6920621408810386, "grad_norm": 9.2734375, "learning_rate": 9.786668561682775e-06, "loss": 1.106, "step": 6504 }, { "epoch": 0.6921685464992552, "grad_norm": 11.21875, "learning_rate": 9.786633030130757e-06, "loss": 1.1081, "step": 6505 }, { "epoch": 0.6922749521174718, "grad_norm": 11.59375, "learning_rate": 9.786597498578738e-06, "loss": 1.2542, "step": 6506 }, { "epoch": 0.6923813577356884, "grad_norm": 11.5, "learning_rate": 9.786561967026722e-06, "loss": 1.1243, "step": 6507 }, { "epoch": 0.692487763353905, "grad_norm": 17.296875, "learning_rate": 9.786526435474703e-06, "loss": 0.6146, "step": 6508 }, { "epoch": 0.6925941689721218, "grad_norm": 11.109375, "learning_rate": 9.786490903922685e-06, "loss": 1.1958, "step": 6509 }, { "epoch": 0.6927005745903384, "grad_norm": 15.1328125, "learning_rate": 9.786455372370666e-06, "loss": 1.0723, "step": 6510 }, { "epoch": 0.692806980208555, "grad_norm": 12.078125, "learning_rate": 9.786419840818648e-06, "loss": 1.351, "step": 6511 }, { "epoch": 0.6929133858267716, "grad_norm": 11.6953125, "learning_rate": 9.78638430926663e-06, "loss": 1.2197, "step": 6512 }, { "epoch": 0.6930197914449883, "grad_norm": 9.90625, "learning_rate": 9.786348777714611e-06, "loss": 1.227, "step": 6513 }, { "epoch": 0.6931261970632049, "grad_norm": 18.625, "learning_rate": 9.786313246162593e-06, "loss": 1.714, "step": 6514 }, { "epoch": 0.6932326026814216, "grad_norm": 10.5859375, "learning_rate": 9.786277714610574e-06, "loss": 1.0711, "step": 6515 }, { "epoch": 0.6933390082996382, "grad_norm": 12.65625, "learning_rate": 9.786242183058556e-06, "loss": 0.9903, "step": 6516 }, { "epoch": 0.6934454139178549, "grad_norm": 14.1953125, "learning_rate": 9.78620665150654e-06, "loss": 1.1613, "step": 6517 }, { "epoch": 0.6935518195360715, "grad_norm": 10.65625, "learning_rate": 9.786171119954521e-06, "loss": 1.2991, "step": 6518 }, { "epoch": 0.6936582251542881, "grad_norm": 13.171875, "learning_rate": 9.786135588402503e-06, "loss": 1.3955, "step": 6519 }, { "epoch": 0.6937646307725048, "grad_norm": 12.0625, "learning_rate": 9.786100056850484e-06, "loss": 1.2356, "step": 6520 }, { "epoch": 0.6938710363907215, "grad_norm": 17.515625, "learning_rate": 9.786064525298466e-06, "loss": 1.3444, "step": 6521 }, { "epoch": 0.6939774420089381, "grad_norm": 10.859375, "learning_rate": 9.786028993746447e-06, "loss": 0.8331, "step": 6522 }, { "epoch": 0.6940838476271547, "grad_norm": 17.71875, "learning_rate": 9.785993462194429e-06, "loss": 1.2713, "step": 6523 }, { "epoch": 0.6941902532453713, "grad_norm": 21.125, "learning_rate": 9.78595793064241e-06, "loss": 1.092, "step": 6524 }, { "epoch": 0.694296658863588, "grad_norm": 10.5625, "learning_rate": 9.785922399090392e-06, "loss": 1.2354, "step": 6525 }, { "epoch": 0.6944030644818047, "grad_norm": 11.0703125, "learning_rate": 9.785886867538374e-06, "loss": 1.3505, "step": 6526 }, { "epoch": 0.6945094701000213, "grad_norm": 17.75, "learning_rate": 9.785851335986357e-06, "loss": 1.1425, "step": 6527 }, { "epoch": 0.6946158757182379, "grad_norm": 13.8515625, "learning_rate": 9.785815804434339e-06, "loss": 1.2757, "step": 6528 }, { "epoch": 0.6947222813364545, "grad_norm": 11.7421875, "learning_rate": 9.78578027288232e-06, "loss": 1.3108, "step": 6529 }, { "epoch": 0.6948286869546713, "grad_norm": 18.609375, "learning_rate": 9.785744741330302e-06, "loss": 1.8779, "step": 6530 }, { "epoch": 0.6949350925728879, "grad_norm": 19.5, "learning_rate": 9.785709209778284e-06, "loss": 1.1291, "step": 6531 }, { "epoch": 0.6950414981911045, "grad_norm": 11.53125, "learning_rate": 9.785673678226267e-06, "loss": 1.1397, "step": 6532 }, { "epoch": 0.6951479038093211, "grad_norm": 10.0625, "learning_rate": 9.785638146674248e-06, "loss": 1.2682, "step": 6533 }, { "epoch": 0.6952543094275377, "grad_norm": 11.9609375, "learning_rate": 9.78560261512223e-06, "loss": 0.8903, "step": 6534 }, { "epoch": 0.6953607150457544, "grad_norm": 10.9375, "learning_rate": 9.785567083570212e-06, "loss": 1.0065, "step": 6535 }, { "epoch": 0.6954671206639711, "grad_norm": 12.5546875, "learning_rate": 9.785531552018193e-06, "loss": 1.379, "step": 6536 }, { "epoch": 0.6955735262821877, "grad_norm": 11.84375, "learning_rate": 9.785496020466175e-06, "loss": 1.2451, "step": 6537 }, { "epoch": 0.6956799319004043, "grad_norm": 10.5703125, "learning_rate": 9.785460488914156e-06, "loss": 1.0494, "step": 6538 }, { "epoch": 0.695786337518621, "grad_norm": 12.4609375, "learning_rate": 9.785424957362138e-06, "loss": 1.1513, "step": 6539 }, { "epoch": 0.6958927431368376, "grad_norm": 10.6328125, "learning_rate": 9.78538942581012e-06, "loss": 1.1607, "step": 6540 }, { "epoch": 0.6959991487550543, "grad_norm": 13.0859375, "learning_rate": 9.785353894258101e-06, "loss": 1.1327, "step": 6541 }, { "epoch": 0.6961055543732709, "grad_norm": 13.5390625, "learning_rate": 9.785318362706083e-06, "loss": 1.3155, "step": 6542 }, { "epoch": 0.6962119599914876, "grad_norm": 19.03125, "learning_rate": 9.785282831154066e-06, "loss": 0.9702, "step": 6543 }, { "epoch": 0.6963183656097042, "grad_norm": 10.84375, "learning_rate": 9.785247299602048e-06, "loss": 1.3433, "step": 6544 }, { "epoch": 0.6964247712279208, "grad_norm": 13.2734375, "learning_rate": 9.78521176805003e-06, "loss": 1.0586, "step": 6545 }, { "epoch": 0.6965311768461375, "grad_norm": 11.125, "learning_rate": 9.785176236498011e-06, "loss": 1.0449, "step": 6546 }, { "epoch": 0.6966375824643541, "grad_norm": 14.7890625, "learning_rate": 9.785140704945993e-06, "loss": 1.1972, "step": 6547 }, { "epoch": 0.6967439880825708, "grad_norm": 10.359375, "learning_rate": 9.785105173393974e-06, "loss": 1.197, "step": 6548 }, { "epoch": 0.6968503937007874, "grad_norm": 12.5703125, "learning_rate": 9.785069641841956e-06, "loss": 1.2804, "step": 6549 }, { "epoch": 0.696956799319004, "grad_norm": 10.0703125, "learning_rate": 9.785034110289937e-06, "loss": 1.194, "step": 6550 }, { "epoch": 0.6970632049372207, "grad_norm": 11.109375, "learning_rate": 9.784998578737919e-06, "loss": 1.208, "step": 6551 }, { "epoch": 0.6971696105554374, "grad_norm": 13.703125, "learning_rate": 9.7849630471859e-06, "loss": 1.1401, "step": 6552 }, { "epoch": 0.697276016173654, "grad_norm": 10.546875, "learning_rate": 9.784927515633884e-06, "loss": 1.4854, "step": 6553 }, { "epoch": 0.6973824217918706, "grad_norm": 11.1796875, "learning_rate": 9.784891984081866e-06, "loss": 1.1456, "step": 6554 }, { "epoch": 0.6974888274100872, "grad_norm": 12.9609375, "learning_rate": 9.784856452529847e-06, "loss": 1.149, "step": 6555 }, { "epoch": 0.6975952330283038, "grad_norm": 14.265625, "learning_rate": 9.784820920977829e-06, "loss": 1.2782, "step": 6556 }, { "epoch": 0.6977016386465206, "grad_norm": 11.8515625, "learning_rate": 9.78478538942581e-06, "loss": 1.1689, "step": 6557 }, { "epoch": 0.6978080442647372, "grad_norm": 13.8828125, "learning_rate": 9.784749857873794e-06, "loss": 1.1969, "step": 6558 }, { "epoch": 0.6979144498829538, "grad_norm": 16.765625, "learning_rate": 9.784714326321775e-06, "loss": 1.3346, "step": 6559 }, { "epoch": 0.6980208555011704, "grad_norm": 14.03125, "learning_rate": 9.784678794769757e-06, "loss": 1.466, "step": 6560 }, { "epoch": 0.698127261119387, "grad_norm": 12.7734375, "learning_rate": 9.784643263217739e-06, "loss": 1.1645, "step": 6561 }, { "epoch": 0.6982336667376038, "grad_norm": 12.2265625, "learning_rate": 9.78460773166572e-06, "loss": 1.1436, "step": 6562 }, { "epoch": 0.6983400723558204, "grad_norm": 13.890625, "learning_rate": 9.784572200113702e-06, "loss": 1.1774, "step": 6563 }, { "epoch": 0.698446477974037, "grad_norm": 16.1875, "learning_rate": 9.784536668561683e-06, "loss": 1.5202, "step": 6564 }, { "epoch": 0.6985528835922536, "grad_norm": 12.5390625, "learning_rate": 9.784501137009665e-06, "loss": 1.3175, "step": 6565 }, { "epoch": 0.6986592892104703, "grad_norm": 19.203125, "learning_rate": 9.784465605457647e-06, "loss": 1.4228, "step": 6566 }, { "epoch": 0.698765694828687, "grad_norm": 13.3203125, "learning_rate": 9.784430073905628e-06, "loss": 1.2754, "step": 6567 }, { "epoch": 0.6988721004469036, "grad_norm": 19.828125, "learning_rate": 9.784394542353611e-06, "loss": 1.2988, "step": 6568 }, { "epoch": 0.6989785060651202, "grad_norm": 15.265625, "learning_rate": 9.784359010801593e-06, "loss": 0.9351, "step": 6569 }, { "epoch": 0.6990849116833369, "grad_norm": 19.6875, "learning_rate": 9.784323479249575e-06, "loss": 1.7751, "step": 6570 }, { "epoch": 0.6991913173015535, "grad_norm": 16.75, "learning_rate": 9.784287947697556e-06, "loss": 1.5086, "step": 6571 }, { "epoch": 0.6992977229197702, "grad_norm": 10.71875, "learning_rate": 9.784252416145538e-06, "loss": 1.6071, "step": 6572 }, { "epoch": 0.6994041285379868, "grad_norm": 12.0390625, "learning_rate": 9.784216884593521e-06, "loss": 1.0526, "step": 6573 }, { "epoch": 0.6995105341562035, "grad_norm": 13.96875, "learning_rate": 9.784181353041503e-06, "loss": 0.9794, "step": 6574 }, { "epoch": 0.6996169397744201, "grad_norm": 12.7734375, "learning_rate": 9.784145821489484e-06, "loss": 1.1567, "step": 6575 }, { "epoch": 0.6997233453926367, "grad_norm": 15.1171875, "learning_rate": 9.784110289937464e-06, "loss": 0.9546, "step": 6576 }, { "epoch": 0.6998297510108533, "grad_norm": 10.2265625, "learning_rate": 9.784074758385446e-06, "loss": 1.0322, "step": 6577 }, { "epoch": 0.69993615662907, "grad_norm": 24.484375, "learning_rate": 9.78403922683343e-06, "loss": 1.1795, "step": 6578 }, { "epoch": 0.7000425622472867, "grad_norm": 8.953125, "learning_rate": 9.78400369528141e-06, "loss": 1.0615, "step": 6579 }, { "epoch": 0.7001489678655033, "grad_norm": 11.7265625, "learning_rate": 9.783968163729392e-06, "loss": 1.0549, "step": 6580 }, { "epoch": 0.7002553734837199, "grad_norm": 12.453125, "learning_rate": 9.783932632177374e-06, "loss": 1.0417, "step": 6581 }, { "epoch": 0.7003617791019365, "grad_norm": 14.5390625, "learning_rate": 9.783897100625356e-06, "loss": 1.6771, "step": 6582 }, { "epoch": 0.7004681847201533, "grad_norm": 13.5390625, "learning_rate": 9.783861569073337e-06, "loss": 1.2398, "step": 6583 }, { "epoch": 0.7005745903383699, "grad_norm": 13.6171875, "learning_rate": 9.78382603752132e-06, "loss": 1.4814, "step": 6584 }, { "epoch": 0.7006809959565865, "grad_norm": 10.09375, "learning_rate": 9.783790505969302e-06, "loss": 1.0474, "step": 6585 }, { "epoch": 0.7007874015748031, "grad_norm": 24.3125, "learning_rate": 9.783754974417284e-06, "loss": 1.4236, "step": 6586 }, { "epoch": 0.7008938071930197, "grad_norm": 13.078125, "learning_rate": 9.783719442865265e-06, "loss": 1.1921, "step": 6587 }, { "epoch": 0.7010002128112365, "grad_norm": 12.6328125, "learning_rate": 9.783683911313247e-06, "loss": 1.3696, "step": 6588 }, { "epoch": 0.7011066184294531, "grad_norm": 12.7890625, "learning_rate": 9.783648379761229e-06, "loss": 1.1671, "step": 6589 }, { "epoch": 0.7012130240476697, "grad_norm": 13.90625, "learning_rate": 9.78361284820921e-06, "loss": 1.3026, "step": 6590 }, { "epoch": 0.7013194296658863, "grad_norm": 12.7265625, "learning_rate": 9.783577316657192e-06, "loss": 1.3926, "step": 6591 }, { "epoch": 0.701425835284103, "grad_norm": 11.3203125, "learning_rate": 9.783541785105173e-06, "loss": 1.1656, "step": 6592 }, { "epoch": 0.7015322409023197, "grad_norm": 11.75, "learning_rate": 9.783506253553155e-06, "loss": 1.4386, "step": 6593 }, { "epoch": 0.7016386465205363, "grad_norm": 11.8125, "learning_rate": 9.783470722001138e-06, "loss": 1.3276, "step": 6594 }, { "epoch": 0.7017450521387529, "grad_norm": 12.8515625, "learning_rate": 9.78343519044912e-06, "loss": 1.3191, "step": 6595 }, { "epoch": 0.7018514577569696, "grad_norm": 51.96875, "learning_rate": 9.783399658897102e-06, "loss": 1.2869, "step": 6596 }, { "epoch": 0.7019578633751862, "grad_norm": 13.53125, "learning_rate": 9.783364127345083e-06, "loss": 0.9926, "step": 6597 }, { "epoch": 0.7020642689934029, "grad_norm": 11.484375, "learning_rate": 9.783328595793065e-06, "loss": 1.0317, "step": 6598 }, { "epoch": 0.7021706746116195, "grad_norm": 9.765625, "learning_rate": 9.783293064241048e-06, "loss": 1.0821, "step": 6599 }, { "epoch": 0.7022770802298361, "grad_norm": 36.90625, "learning_rate": 9.78325753268903e-06, "loss": 1.1058, "step": 6600 }, { "epoch": 0.7023834858480528, "grad_norm": 13.25, "learning_rate": 9.783222001137011e-06, "loss": 1.0372, "step": 6601 }, { "epoch": 0.7024898914662694, "grad_norm": 10.828125, "learning_rate": 9.783186469584993e-06, "loss": 1.1507, "step": 6602 }, { "epoch": 0.702596297084486, "grad_norm": 14.46875, "learning_rate": 9.783150938032974e-06, "loss": 1.1267, "step": 6603 }, { "epoch": 0.7027027027027027, "grad_norm": 14.1953125, "learning_rate": 9.783115406480956e-06, "loss": 1.2328, "step": 6604 }, { "epoch": 0.7028091083209194, "grad_norm": 23.96875, "learning_rate": 9.783079874928938e-06, "loss": 1.1813, "step": 6605 }, { "epoch": 0.702915513939136, "grad_norm": 19.75, "learning_rate": 9.78304434337692e-06, "loss": 1.0006, "step": 6606 }, { "epoch": 0.7030219195573526, "grad_norm": 20.109375, "learning_rate": 9.783008811824901e-06, "loss": 1.0875, "step": 6607 }, { "epoch": 0.7031283251755692, "grad_norm": 14.4453125, "learning_rate": 9.782973280272882e-06, "loss": 1.6977, "step": 6608 }, { "epoch": 0.703234730793786, "grad_norm": 14.5078125, "learning_rate": 9.782937748720866e-06, "loss": 1.1788, "step": 6609 }, { "epoch": 0.7033411364120026, "grad_norm": 14.7109375, "learning_rate": 9.782902217168847e-06, "loss": 1.222, "step": 6610 }, { "epoch": 0.7034475420302192, "grad_norm": 13.4609375, "learning_rate": 9.782866685616829e-06, "loss": 1.0734, "step": 6611 }, { "epoch": 0.7035539476484358, "grad_norm": 11.625, "learning_rate": 9.78283115406481e-06, "loss": 1.4147, "step": 6612 }, { "epoch": 0.7036603532666524, "grad_norm": 14.546875, "learning_rate": 9.782795622512792e-06, "loss": 1.2963, "step": 6613 }, { "epoch": 0.7037667588848692, "grad_norm": 16.140625, "learning_rate": 9.782760090960774e-06, "loss": 1.3084, "step": 6614 }, { "epoch": 0.7038731645030858, "grad_norm": 10.6171875, "learning_rate": 9.782724559408755e-06, "loss": 0.9293, "step": 6615 }, { "epoch": 0.7039795701213024, "grad_norm": 11.8515625, "learning_rate": 9.782689027856737e-06, "loss": 1.0454, "step": 6616 }, { "epoch": 0.704085975739519, "grad_norm": 14.5703125, "learning_rate": 9.782653496304719e-06, "loss": 1.2798, "step": 6617 }, { "epoch": 0.7041923813577357, "grad_norm": 16.640625, "learning_rate": 9.7826179647527e-06, "loss": 1.0308, "step": 6618 }, { "epoch": 0.7042987869759524, "grad_norm": 10.9609375, "learning_rate": 9.782582433200682e-06, "loss": 1.201, "step": 6619 }, { "epoch": 0.704405192594169, "grad_norm": 14.828125, "learning_rate": 9.782546901648665e-06, "loss": 1.4183, "step": 6620 }, { "epoch": 0.7045115982123856, "grad_norm": 14.046875, "learning_rate": 9.782511370096647e-06, "loss": 1.4332, "step": 6621 }, { "epoch": 0.7046180038306022, "grad_norm": 13.984375, "learning_rate": 9.782475838544628e-06, "loss": 1.2213, "step": 6622 }, { "epoch": 0.7047244094488189, "grad_norm": 11.703125, "learning_rate": 9.78244030699261e-06, "loss": 1.4558, "step": 6623 }, { "epoch": 0.7048308150670355, "grad_norm": 13.9765625, "learning_rate": 9.782404775440592e-06, "loss": 1.2952, "step": 6624 }, { "epoch": 0.7049372206852522, "grad_norm": 15.46875, "learning_rate": 9.782369243888575e-06, "loss": 1.576, "step": 6625 }, { "epoch": 0.7050436263034688, "grad_norm": 10.4140625, "learning_rate": 9.782333712336556e-06, "loss": 1.0902, "step": 6626 }, { "epoch": 0.7051500319216855, "grad_norm": 12.453125, "learning_rate": 9.782298180784538e-06, "loss": 1.3368, "step": 6627 }, { "epoch": 0.7052564375399021, "grad_norm": 13.8046875, "learning_rate": 9.78226264923252e-06, "loss": 1.0404, "step": 6628 }, { "epoch": 0.7053628431581187, "grad_norm": 9.6015625, "learning_rate": 9.782227117680501e-06, "loss": 1.008, "step": 6629 }, { "epoch": 0.7054692487763354, "grad_norm": 9.109375, "learning_rate": 9.782191586128483e-06, "loss": 1.079, "step": 6630 }, { "epoch": 0.705575654394552, "grad_norm": 15.03125, "learning_rate": 9.782156054576465e-06, "loss": 1.4008, "step": 6631 }, { "epoch": 0.7056820600127687, "grad_norm": 12.671875, "learning_rate": 9.782120523024446e-06, "loss": 1.3913, "step": 6632 }, { "epoch": 0.7057884656309853, "grad_norm": 7.91796875, "learning_rate": 9.782084991472428e-06, "loss": 0.9179, "step": 6633 }, { "epoch": 0.7058948712492019, "grad_norm": 16.8125, "learning_rate": 9.78204945992041e-06, "loss": 1.4146, "step": 6634 }, { "epoch": 0.7060012768674186, "grad_norm": 11.4140625, "learning_rate": 9.782013928368393e-06, "loss": 1.2546, "step": 6635 }, { "epoch": 0.7061076824856353, "grad_norm": 11.375, "learning_rate": 9.781978396816374e-06, "loss": 1.0939, "step": 6636 }, { "epoch": 0.7062140881038519, "grad_norm": 9.9140625, "learning_rate": 9.781942865264356e-06, "loss": 0.9433, "step": 6637 }, { "epoch": 0.7063204937220685, "grad_norm": 10.59375, "learning_rate": 9.781907333712337e-06, "loss": 1.2221, "step": 6638 }, { "epoch": 0.7064268993402851, "grad_norm": 11.8671875, "learning_rate": 9.781871802160319e-06, "loss": 1.3555, "step": 6639 }, { "epoch": 0.7065333049585019, "grad_norm": 12.2265625, "learning_rate": 9.7818362706083e-06, "loss": 1.7066, "step": 6640 }, { "epoch": 0.7066397105767185, "grad_norm": 12.125, "learning_rate": 9.781800739056282e-06, "loss": 1.3436, "step": 6641 }, { "epoch": 0.7067461161949351, "grad_norm": 18.625, "learning_rate": 9.781765207504264e-06, "loss": 1.4383, "step": 6642 }, { "epoch": 0.7068525218131517, "grad_norm": 17.625, "learning_rate": 9.781729675952246e-06, "loss": 1.3765, "step": 6643 }, { "epoch": 0.7069589274313683, "grad_norm": 11.3203125, "learning_rate": 9.781694144400227e-06, "loss": 1.0461, "step": 6644 }, { "epoch": 0.707065333049585, "grad_norm": 12.3515625, "learning_rate": 9.78165861284821e-06, "loss": 1.2088, "step": 6645 }, { "epoch": 0.7071717386678017, "grad_norm": 10.1171875, "learning_rate": 9.781623081296192e-06, "loss": 1.243, "step": 6646 }, { "epoch": 0.7072781442860183, "grad_norm": 19.953125, "learning_rate": 9.781587549744174e-06, "loss": 1.5677, "step": 6647 }, { "epoch": 0.7073845499042349, "grad_norm": 10.71875, "learning_rate": 9.781552018192155e-06, "loss": 0.9928, "step": 6648 }, { "epoch": 0.7074909555224516, "grad_norm": 9.28125, "learning_rate": 9.781516486640137e-06, "loss": 1.0897, "step": 6649 }, { "epoch": 0.7075973611406682, "grad_norm": 11.7734375, "learning_rate": 9.78148095508812e-06, "loss": 1.3156, "step": 6650 }, { "epoch": 0.7077037667588849, "grad_norm": 13.9765625, "learning_rate": 9.781445423536102e-06, "loss": 1.0558, "step": 6651 }, { "epoch": 0.7078101723771015, "grad_norm": 11.640625, "learning_rate": 9.781409891984083e-06, "loss": 1.1271, "step": 6652 }, { "epoch": 0.7079165779953182, "grad_norm": 14.109375, "learning_rate": 9.781374360432065e-06, "loss": 1.4533, "step": 6653 }, { "epoch": 0.7080229836135348, "grad_norm": 13.328125, "learning_rate": 9.781338828880047e-06, "loss": 1.3061, "step": 6654 }, { "epoch": 0.7081293892317514, "grad_norm": 12.28125, "learning_rate": 9.781303297328028e-06, "loss": 1.3382, "step": 6655 }, { "epoch": 0.7082357948499681, "grad_norm": 19.125, "learning_rate": 9.78126776577601e-06, "loss": 0.9891, "step": 6656 }, { "epoch": 0.7083422004681847, "grad_norm": 14.6796875, "learning_rate": 9.781232234223991e-06, "loss": 1.2784, "step": 6657 }, { "epoch": 0.7084486060864014, "grad_norm": 12.0078125, "learning_rate": 9.781196702671973e-06, "loss": 1.1128, "step": 6658 }, { "epoch": 0.708555011704618, "grad_norm": 12.6328125, "learning_rate": 9.781161171119955e-06, "loss": 1.4309, "step": 6659 }, { "epoch": 0.7086614173228346, "grad_norm": 15.6953125, "learning_rate": 9.781125639567936e-06, "loss": 1.0214, "step": 6660 }, { "epoch": 0.7087678229410513, "grad_norm": 14.9453125, "learning_rate": 9.78109010801592e-06, "loss": 1.458, "step": 6661 }, { "epoch": 0.708874228559268, "grad_norm": 15.84375, "learning_rate": 9.781054576463901e-06, "loss": 1.3953, "step": 6662 }, { "epoch": 0.7089806341774846, "grad_norm": 8.6171875, "learning_rate": 9.781019044911883e-06, "loss": 1.0898, "step": 6663 }, { "epoch": 0.7090870397957012, "grad_norm": 17.625, "learning_rate": 9.780983513359864e-06, "loss": 1.1259, "step": 6664 }, { "epoch": 0.7091934454139178, "grad_norm": 10.8046875, "learning_rate": 9.780947981807846e-06, "loss": 1.2575, "step": 6665 }, { "epoch": 0.7092998510321344, "grad_norm": 21.390625, "learning_rate": 9.780912450255828e-06, "loss": 1.2574, "step": 6666 }, { "epoch": 0.7094062566503512, "grad_norm": 17.453125, "learning_rate": 9.780876918703809e-06, "loss": 0.9712, "step": 6667 }, { "epoch": 0.7095126622685678, "grad_norm": 12.7109375, "learning_rate": 9.78084138715179e-06, "loss": 1.1629, "step": 6668 }, { "epoch": 0.7096190678867844, "grad_norm": 16.5, "learning_rate": 9.780805855599772e-06, "loss": 1.2707, "step": 6669 }, { "epoch": 0.709725473505001, "grad_norm": 18.265625, "learning_rate": 9.780770324047754e-06, "loss": 1.0341, "step": 6670 }, { "epoch": 0.7098318791232177, "grad_norm": 11.6328125, "learning_rate": 9.780734792495737e-06, "loss": 1.1458, "step": 6671 }, { "epoch": 0.7099382847414344, "grad_norm": 9.8515625, "learning_rate": 9.780699260943719e-06, "loss": 1.0819, "step": 6672 }, { "epoch": 0.710044690359651, "grad_norm": 12.828125, "learning_rate": 9.7806637293917e-06, "loss": 1.2953, "step": 6673 }, { "epoch": 0.7101510959778676, "grad_norm": 14.015625, "learning_rate": 9.780628197839682e-06, "loss": 1.2193, "step": 6674 }, { "epoch": 0.7102575015960843, "grad_norm": 13.9375, "learning_rate": 9.780592666287664e-06, "loss": 1.5928, "step": 6675 }, { "epoch": 0.7103639072143009, "grad_norm": 12.546875, "learning_rate": 9.780557134735647e-06, "loss": 1.2912, "step": 6676 }, { "epoch": 0.7104703128325176, "grad_norm": 14.25, "learning_rate": 9.780521603183629e-06, "loss": 1.4841, "step": 6677 }, { "epoch": 0.7105767184507342, "grad_norm": 14.0859375, "learning_rate": 9.78048607163161e-06, "loss": 1.5958, "step": 6678 }, { "epoch": 0.7106831240689508, "grad_norm": 14.875, "learning_rate": 9.780450540079592e-06, "loss": 1.1358, "step": 6679 }, { "epoch": 0.7107895296871675, "grad_norm": 14.8515625, "learning_rate": 9.780415008527573e-06, "loss": 1.0809, "step": 6680 }, { "epoch": 0.7108959353053841, "grad_norm": 11.171875, "learning_rate": 9.780379476975555e-06, "loss": 1.5097, "step": 6681 }, { "epoch": 0.7110023409236008, "grad_norm": 12.515625, "learning_rate": 9.780343945423537e-06, "loss": 1.2955, "step": 6682 }, { "epoch": 0.7111087465418174, "grad_norm": 10.9140625, "learning_rate": 9.780308413871518e-06, "loss": 1.3493, "step": 6683 }, { "epoch": 0.7112151521600341, "grad_norm": 14.3671875, "learning_rate": 9.7802728823195e-06, "loss": 1.3581, "step": 6684 }, { "epoch": 0.7113215577782507, "grad_norm": 12.890625, "learning_rate": 9.780237350767481e-06, "loss": 1.2069, "step": 6685 }, { "epoch": 0.7114279633964673, "grad_norm": 16.90625, "learning_rate": 9.780201819215465e-06, "loss": 1.5831, "step": 6686 }, { "epoch": 0.7115343690146839, "grad_norm": 18.875, "learning_rate": 9.780166287663446e-06, "loss": 1.1623, "step": 6687 }, { "epoch": 0.7116407746329007, "grad_norm": 14.9609375, "learning_rate": 9.780130756111428e-06, "loss": 1.3774, "step": 6688 }, { "epoch": 0.7117471802511173, "grad_norm": 12.8203125, "learning_rate": 9.78009522455941e-06, "loss": 1.0235, "step": 6689 }, { "epoch": 0.7118535858693339, "grad_norm": 14.21875, "learning_rate": 9.780059693007391e-06, "loss": 1.2491, "step": 6690 }, { "epoch": 0.7119599914875505, "grad_norm": 19.796875, "learning_rate": 9.780024161455374e-06, "loss": 1.2806, "step": 6691 }, { "epoch": 0.7120663971057671, "grad_norm": 15.3828125, "learning_rate": 9.779988629903356e-06, "loss": 1.1809, "step": 6692 }, { "epoch": 0.7121728027239839, "grad_norm": 13.328125, "learning_rate": 9.779953098351338e-06, "loss": 1.2207, "step": 6693 }, { "epoch": 0.7122792083422005, "grad_norm": 13.6171875, "learning_rate": 9.77991756679932e-06, "loss": 1.1293, "step": 6694 }, { "epoch": 0.7123856139604171, "grad_norm": 15.265625, "learning_rate": 9.779882035247301e-06, "loss": 1.2043, "step": 6695 }, { "epoch": 0.7124920195786337, "grad_norm": 14.9453125, "learning_rate": 9.77984650369528e-06, "loss": 1.2924, "step": 6696 }, { "epoch": 0.7125984251968503, "grad_norm": 13.8125, "learning_rate": 9.779810972143264e-06, "loss": 1.2372, "step": 6697 }, { "epoch": 0.7127048308150671, "grad_norm": 12.5546875, "learning_rate": 9.779775440591246e-06, "loss": 1.1453, "step": 6698 }, { "epoch": 0.7128112364332837, "grad_norm": 10.2421875, "learning_rate": 9.779739909039227e-06, "loss": 1.1396, "step": 6699 }, { "epoch": 0.7129176420515003, "grad_norm": 10.125, "learning_rate": 9.779704377487209e-06, "loss": 1.1415, "step": 6700 }, { "epoch": 0.7130240476697169, "grad_norm": 13.7421875, "learning_rate": 9.77966884593519e-06, "loss": 1.3208, "step": 6701 }, { "epoch": 0.7131304532879336, "grad_norm": 11.3359375, "learning_rate": 9.779633314383174e-06, "loss": 0.904, "step": 6702 }, { "epoch": 0.7132368589061503, "grad_norm": 13.3046875, "learning_rate": 9.779597782831155e-06, "loss": 1.3729, "step": 6703 }, { "epoch": 0.7133432645243669, "grad_norm": 15.140625, "learning_rate": 9.779562251279137e-06, "loss": 1.2871, "step": 6704 }, { "epoch": 0.7134496701425835, "grad_norm": 15.2734375, "learning_rate": 9.779526719727119e-06, "loss": 1.3151, "step": 6705 }, { "epoch": 0.7135560757608002, "grad_norm": 11.859375, "learning_rate": 9.7794911881751e-06, "loss": 1.4638, "step": 6706 }, { "epoch": 0.7136624813790168, "grad_norm": 11.0546875, "learning_rate": 9.779455656623082e-06, "loss": 1.2261, "step": 6707 }, { "epoch": 0.7137688869972335, "grad_norm": 13.8984375, "learning_rate": 9.779420125071063e-06, "loss": 1.4298, "step": 6708 }, { "epoch": 0.7138752926154501, "grad_norm": 12.828125, "learning_rate": 9.779384593519045e-06, "loss": 1.3178, "step": 6709 }, { "epoch": 0.7139816982336668, "grad_norm": 11.5078125, "learning_rate": 9.779349061967027e-06, "loss": 1.1227, "step": 6710 }, { "epoch": 0.7140881038518834, "grad_norm": 14.9375, "learning_rate": 9.779313530415008e-06, "loss": 1.5607, "step": 6711 }, { "epoch": 0.7141945094701, "grad_norm": 10.21875, "learning_rate": 9.779277998862992e-06, "loss": 0.9495, "step": 6712 }, { "epoch": 0.7143009150883166, "grad_norm": 12.1171875, "learning_rate": 9.779242467310973e-06, "loss": 1.2917, "step": 6713 }, { "epoch": 0.7144073207065333, "grad_norm": 16.109375, "learning_rate": 9.779206935758955e-06, "loss": 1.608, "step": 6714 }, { "epoch": 0.71451372632475, "grad_norm": 10.390625, "learning_rate": 9.779171404206936e-06, "loss": 0.9972, "step": 6715 }, { "epoch": 0.7146201319429666, "grad_norm": 10.1328125, "learning_rate": 9.779135872654918e-06, "loss": 1.3744, "step": 6716 }, { "epoch": 0.7147265375611832, "grad_norm": 13.5625, "learning_rate": 9.779100341102901e-06, "loss": 1.4029, "step": 6717 }, { "epoch": 0.7148329431793998, "grad_norm": 9.515625, "learning_rate": 9.779064809550883e-06, "loss": 1.3035, "step": 6718 }, { "epoch": 0.7149393487976166, "grad_norm": 13.75, "learning_rate": 9.779029277998865e-06, "loss": 1.3765, "step": 6719 }, { "epoch": 0.7150457544158332, "grad_norm": 10.65625, "learning_rate": 9.778993746446846e-06, "loss": 1.0659, "step": 6720 }, { "epoch": 0.7151521600340498, "grad_norm": 10.546875, "learning_rate": 9.778958214894828e-06, "loss": 1.5633, "step": 6721 }, { "epoch": 0.7152585656522664, "grad_norm": 12.0859375, "learning_rate": 9.77892268334281e-06, "loss": 1.2316, "step": 6722 }, { "epoch": 0.715364971270483, "grad_norm": 10.515625, "learning_rate": 9.778887151790791e-06, "loss": 1.1179, "step": 6723 }, { "epoch": 0.7154713768886998, "grad_norm": 14.1328125, "learning_rate": 9.778851620238773e-06, "loss": 1.3962, "step": 6724 }, { "epoch": 0.7155777825069164, "grad_norm": 13.046875, "learning_rate": 9.778816088686754e-06, "loss": 1.2048, "step": 6725 }, { "epoch": 0.715684188125133, "grad_norm": 19.796875, "learning_rate": 9.778780557134736e-06, "loss": 1.0004, "step": 6726 }, { "epoch": 0.7157905937433496, "grad_norm": 14.4921875, "learning_rate": 9.778745025582719e-06, "loss": 1.1769, "step": 6727 }, { "epoch": 0.7158969993615663, "grad_norm": 14.90625, "learning_rate": 9.7787094940307e-06, "loss": 0.8237, "step": 6728 }, { "epoch": 0.716003404979783, "grad_norm": 11.265625, "learning_rate": 9.778673962478682e-06, "loss": 1.1629, "step": 6729 }, { "epoch": 0.7161098105979996, "grad_norm": 11.7421875, "learning_rate": 9.778638430926664e-06, "loss": 1.3212, "step": 6730 }, { "epoch": 0.7162162162162162, "grad_norm": 9.6640625, "learning_rate": 9.778602899374646e-06, "loss": 1.1411, "step": 6731 }, { "epoch": 0.7163226218344328, "grad_norm": 12.609375, "learning_rate": 9.778567367822627e-06, "loss": 1.3391, "step": 6732 }, { "epoch": 0.7164290274526495, "grad_norm": 14.265625, "learning_rate": 9.778531836270609e-06, "loss": 1.1902, "step": 6733 }, { "epoch": 0.7165354330708661, "grad_norm": 12.84375, "learning_rate": 9.77849630471859e-06, "loss": 0.9867, "step": 6734 }, { "epoch": 0.7166418386890828, "grad_norm": 11.203125, "learning_rate": 9.778460773166572e-06, "loss": 1.0818, "step": 6735 }, { "epoch": 0.7167482443072994, "grad_norm": 13.2578125, "learning_rate": 9.778425241614554e-06, "loss": 1.2577, "step": 6736 }, { "epoch": 0.7168546499255161, "grad_norm": 11.1953125, "learning_rate": 9.778389710062535e-06, "loss": 0.829, "step": 6737 }, { "epoch": 0.7169610555437327, "grad_norm": 15.015625, "learning_rate": 9.778354178510518e-06, "loss": 1.4628, "step": 6738 }, { "epoch": 0.7170674611619493, "grad_norm": 9.8828125, "learning_rate": 9.7783186469585e-06, "loss": 0.9963, "step": 6739 }, { "epoch": 0.717173866780166, "grad_norm": 12.9375, "learning_rate": 9.778283115406482e-06, "loss": 1.0938, "step": 6740 }, { "epoch": 0.7172802723983827, "grad_norm": 16.953125, "learning_rate": 9.778247583854463e-06, "loss": 1.5515, "step": 6741 }, { "epoch": 0.7173866780165993, "grad_norm": 16.984375, "learning_rate": 9.778212052302445e-06, "loss": 1.4425, "step": 6742 }, { "epoch": 0.7174930836348159, "grad_norm": 10.8046875, "learning_rate": 9.778176520750428e-06, "loss": 1.0817, "step": 6743 }, { "epoch": 0.7175994892530325, "grad_norm": 14.4375, "learning_rate": 9.77814098919841e-06, "loss": 1.1092, "step": 6744 }, { "epoch": 0.7177058948712493, "grad_norm": 14.609375, "learning_rate": 9.778105457646391e-06, "loss": 1.1885, "step": 6745 }, { "epoch": 0.7178123004894659, "grad_norm": 14.984375, "learning_rate": 9.778069926094373e-06, "loss": 1.154, "step": 6746 }, { "epoch": 0.7179187061076825, "grad_norm": 14.984375, "learning_rate": 9.778034394542355e-06, "loss": 1.4754, "step": 6747 }, { "epoch": 0.7180251117258991, "grad_norm": 15.4375, "learning_rate": 9.777998862990336e-06, "loss": 1.3698, "step": 6748 }, { "epoch": 0.7181315173441157, "grad_norm": 16.484375, "learning_rate": 9.777963331438318e-06, "loss": 1.2296, "step": 6749 }, { "epoch": 0.7182379229623325, "grad_norm": 12.5703125, "learning_rate": 9.7779277998863e-06, "loss": 1.1946, "step": 6750 }, { "epoch": 0.7183443285805491, "grad_norm": 14.828125, "learning_rate": 9.777892268334281e-06, "loss": 1.1964, "step": 6751 }, { "epoch": 0.7184507341987657, "grad_norm": 10.703125, "learning_rate": 9.777856736782263e-06, "loss": 1.0189, "step": 6752 }, { "epoch": 0.7185571398169823, "grad_norm": 13.8359375, "learning_rate": 9.777821205230246e-06, "loss": 1.1082, "step": 6753 }, { "epoch": 0.718663545435199, "grad_norm": 11.3515625, "learning_rate": 9.777785673678228e-06, "loss": 1.1692, "step": 6754 }, { "epoch": 0.7187699510534156, "grad_norm": 15.828125, "learning_rate": 9.77775014212621e-06, "loss": 1.1631, "step": 6755 }, { "epoch": 0.7188763566716323, "grad_norm": 15.828125, "learning_rate": 9.77771461057419e-06, "loss": 1.2749, "step": 6756 }, { "epoch": 0.7189827622898489, "grad_norm": 12.0625, "learning_rate": 9.777679079022172e-06, "loss": 1.2406, "step": 6757 }, { "epoch": 0.7190891679080655, "grad_norm": 10.9453125, "learning_rate": 9.777643547470154e-06, "loss": 1.1959, "step": 6758 }, { "epoch": 0.7191955735262822, "grad_norm": 10.203125, "learning_rate": 9.777608015918136e-06, "loss": 1.0792, "step": 6759 }, { "epoch": 0.7193019791444988, "grad_norm": 15.484375, "learning_rate": 9.777572484366117e-06, "loss": 1.385, "step": 6760 }, { "epoch": 0.7194083847627155, "grad_norm": 11.71875, "learning_rate": 9.777536952814099e-06, "loss": 1.2098, "step": 6761 }, { "epoch": 0.7195147903809321, "grad_norm": 15.3125, "learning_rate": 9.77750142126208e-06, "loss": 1.1685, "step": 6762 }, { "epoch": 0.7196211959991488, "grad_norm": 10.703125, "learning_rate": 9.777465889710064e-06, "loss": 1.3198, "step": 6763 }, { "epoch": 0.7197276016173654, "grad_norm": 14.7890625, "learning_rate": 9.777430358158045e-06, "loss": 1.3248, "step": 6764 }, { "epoch": 0.719834007235582, "grad_norm": 11.515625, "learning_rate": 9.777394826606027e-06, "loss": 1.0394, "step": 6765 }, { "epoch": 0.7199404128537987, "grad_norm": 17.71875, "learning_rate": 9.777359295054009e-06, "loss": 1.1414, "step": 6766 }, { "epoch": 0.7200468184720153, "grad_norm": 12.4765625, "learning_rate": 9.77732376350199e-06, "loss": 1.1604, "step": 6767 }, { "epoch": 0.720153224090232, "grad_norm": 17.765625, "learning_rate": 9.777288231949973e-06, "loss": 1.408, "step": 6768 }, { "epoch": 0.7202596297084486, "grad_norm": 13.03125, "learning_rate": 9.777252700397955e-06, "loss": 1.1043, "step": 6769 }, { "epoch": 0.7203660353266652, "grad_norm": 15.328125, "learning_rate": 9.777217168845937e-06, "loss": 1.0829, "step": 6770 }, { "epoch": 0.7204724409448819, "grad_norm": 14.4609375, "learning_rate": 9.777181637293918e-06, "loss": 1.1684, "step": 6771 }, { "epoch": 0.7205788465630986, "grad_norm": 12.8125, "learning_rate": 9.7771461057419e-06, "loss": 1.1193, "step": 6772 }, { "epoch": 0.7206852521813152, "grad_norm": 14.703125, "learning_rate": 9.777110574189881e-06, "loss": 1.0116, "step": 6773 }, { "epoch": 0.7207916577995318, "grad_norm": 11.640625, "learning_rate": 9.777075042637863e-06, "loss": 1.1447, "step": 6774 }, { "epoch": 0.7208980634177484, "grad_norm": 10.328125, "learning_rate": 9.777039511085845e-06, "loss": 1.1824, "step": 6775 }, { "epoch": 0.721004469035965, "grad_norm": 22.75, "learning_rate": 9.777003979533826e-06, "loss": 1.4461, "step": 6776 }, { "epoch": 0.7211108746541818, "grad_norm": 15.90625, "learning_rate": 9.776968447981808e-06, "loss": 1.3682, "step": 6777 }, { "epoch": 0.7212172802723984, "grad_norm": 17.015625, "learning_rate": 9.77693291642979e-06, "loss": 1.1361, "step": 6778 }, { "epoch": 0.721323685890615, "grad_norm": 11.109375, "learning_rate": 9.776897384877773e-06, "loss": 0.9156, "step": 6779 }, { "epoch": 0.7214300915088316, "grad_norm": 11.8671875, "learning_rate": 9.776861853325754e-06, "loss": 1.1969, "step": 6780 }, { "epoch": 0.7215364971270483, "grad_norm": 12.015625, "learning_rate": 9.776826321773736e-06, "loss": 0.953, "step": 6781 }, { "epoch": 0.721642902745265, "grad_norm": 17.96875, "learning_rate": 9.776790790221718e-06, "loss": 1.306, "step": 6782 }, { "epoch": 0.7217493083634816, "grad_norm": 11.0546875, "learning_rate": 9.7767552586697e-06, "loss": 1.3922, "step": 6783 }, { "epoch": 0.7218557139816982, "grad_norm": 15.546875, "learning_rate": 9.776719727117683e-06, "loss": 1.1066, "step": 6784 }, { "epoch": 0.7219621195999149, "grad_norm": 12.6171875, "learning_rate": 9.776684195565662e-06, "loss": 1.0831, "step": 6785 }, { "epoch": 0.7220685252181315, "grad_norm": 15.03125, "learning_rate": 9.776648664013644e-06, "loss": 1.4716, "step": 6786 }, { "epoch": 0.7221749308363482, "grad_norm": 17.6875, "learning_rate": 9.776613132461626e-06, "loss": 1.3018, "step": 6787 }, { "epoch": 0.7222813364545648, "grad_norm": 11.484375, "learning_rate": 9.776577600909607e-06, "loss": 0.9802, "step": 6788 }, { "epoch": 0.7223877420727814, "grad_norm": 14.75, "learning_rate": 9.77654206935759e-06, "loss": 1.6594, "step": 6789 }, { "epoch": 0.7224941476909981, "grad_norm": 11.6875, "learning_rate": 9.776506537805572e-06, "loss": 1.0805, "step": 6790 }, { "epoch": 0.7226005533092147, "grad_norm": 11.9140625, "learning_rate": 9.776471006253554e-06, "loss": 1.2801, "step": 6791 }, { "epoch": 0.7227069589274314, "grad_norm": 9.8671875, "learning_rate": 9.776435474701535e-06, "loss": 1.1678, "step": 6792 }, { "epoch": 0.722813364545648, "grad_norm": 9.8671875, "learning_rate": 9.776399943149517e-06, "loss": 1.2004, "step": 6793 }, { "epoch": 0.7229197701638647, "grad_norm": 13.984375, "learning_rate": 9.7763644115975e-06, "loss": 1.3563, "step": 6794 }, { "epoch": 0.7230261757820813, "grad_norm": 15.2578125, "learning_rate": 9.776328880045482e-06, "loss": 1.3254, "step": 6795 }, { "epoch": 0.7231325814002979, "grad_norm": 17.234375, "learning_rate": 9.776293348493464e-06, "loss": 1.0186, "step": 6796 }, { "epoch": 0.7232389870185145, "grad_norm": 11.6875, "learning_rate": 9.776257816941445e-06, "loss": 1.0885, "step": 6797 }, { "epoch": 0.7233453926367313, "grad_norm": 13.6796875, "learning_rate": 9.776222285389427e-06, "loss": 1.5275, "step": 6798 }, { "epoch": 0.7234517982549479, "grad_norm": 12.6953125, "learning_rate": 9.776186753837408e-06, "loss": 0.9933, "step": 6799 }, { "epoch": 0.7235582038731645, "grad_norm": 11.3203125, "learning_rate": 9.77615122228539e-06, "loss": 1.0383, "step": 6800 }, { "epoch": 0.7236646094913811, "grad_norm": 13.984375, "learning_rate": 9.776115690733372e-06, "loss": 1.3689, "step": 6801 }, { "epoch": 0.7237710151095977, "grad_norm": 10.4375, "learning_rate": 9.776080159181353e-06, "loss": 1.4955, "step": 6802 }, { "epoch": 0.7238774207278145, "grad_norm": 9.890625, "learning_rate": 9.776044627629335e-06, "loss": 0.8693, "step": 6803 }, { "epoch": 0.7239838263460311, "grad_norm": 11.5234375, "learning_rate": 9.776009096077318e-06, "loss": 1.1577, "step": 6804 }, { "epoch": 0.7240902319642477, "grad_norm": 11.8046875, "learning_rate": 9.7759735645253e-06, "loss": 1.5382, "step": 6805 }, { "epoch": 0.7241966375824643, "grad_norm": 14.15625, "learning_rate": 9.775938032973281e-06, "loss": 1.3996, "step": 6806 }, { "epoch": 0.724303043200681, "grad_norm": 14.34375, "learning_rate": 9.775902501421263e-06, "loss": 1.4668, "step": 6807 }, { "epoch": 0.7244094488188977, "grad_norm": 18.078125, "learning_rate": 9.775866969869244e-06, "loss": 1.3989, "step": 6808 }, { "epoch": 0.7245158544371143, "grad_norm": 10.140625, "learning_rate": 9.775831438317228e-06, "loss": 1.2335, "step": 6809 }, { "epoch": 0.7246222600553309, "grad_norm": 14.3828125, "learning_rate": 9.77579590676521e-06, "loss": 1.4281, "step": 6810 }, { "epoch": 0.7247286656735475, "grad_norm": 10.7109375, "learning_rate": 9.775760375213191e-06, "loss": 1.0894, "step": 6811 }, { "epoch": 0.7248350712917642, "grad_norm": 14.7734375, "learning_rate": 9.775724843661173e-06, "loss": 1.5357, "step": 6812 }, { "epoch": 0.7249414769099809, "grad_norm": 13.453125, "learning_rate": 9.775689312109154e-06, "loss": 1.3001, "step": 6813 }, { "epoch": 0.7250478825281975, "grad_norm": 12.46875, "learning_rate": 9.775653780557136e-06, "loss": 1.3754, "step": 6814 }, { "epoch": 0.7251542881464141, "grad_norm": 17.5625, "learning_rate": 9.775618249005117e-06, "loss": 1.1804, "step": 6815 }, { "epoch": 0.7252606937646308, "grad_norm": 10.109375, "learning_rate": 9.775582717453099e-06, "loss": 1.3385, "step": 6816 }, { "epoch": 0.7253670993828474, "grad_norm": 13.8515625, "learning_rate": 9.77554718590108e-06, "loss": 1.4407, "step": 6817 }, { "epoch": 0.725473505001064, "grad_norm": 10.6328125, "learning_rate": 9.775511654349062e-06, "loss": 1.0017, "step": 6818 }, { "epoch": 0.7255799106192807, "grad_norm": 12.1640625, "learning_rate": 9.775476122797044e-06, "loss": 1.2928, "step": 6819 }, { "epoch": 0.7256863162374974, "grad_norm": 14.2734375, "learning_rate": 9.775440591245027e-06, "loss": 1.1611, "step": 6820 }, { "epoch": 0.725792721855714, "grad_norm": 15.0078125, "learning_rate": 9.775405059693009e-06, "loss": 0.936, "step": 6821 }, { "epoch": 0.7258991274739306, "grad_norm": 11.984375, "learning_rate": 9.77536952814099e-06, "loss": 1.3545, "step": 6822 }, { "epoch": 0.7260055330921472, "grad_norm": 9.203125, "learning_rate": 9.775333996588972e-06, "loss": 0.9498, "step": 6823 }, { "epoch": 0.726111938710364, "grad_norm": 15.2109375, "learning_rate": 9.775298465036954e-06, "loss": 1.185, "step": 6824 }, { "epoch": 0.7262183443285806, "grad_norm": 14.6796875, "learning_rate": 9.775262933484935e-06, "loss": 1.4012, "step": 6825 }, { "epoch": 0.7263247499467972, "grad_norm": 13.953125, "learning_rate": 9.775227401932917e-06, "loss": 1.2178, "step": 6826 }, { "epoch": 0.7264311555650138, "grad_norm": 12.8984375, "learning_rate": 9.775191870380898e-06, "loss": 1.3729, "step": 6827 }, { "epoch": 0.7265375611832304, "grad_norm": 12.4296875, "learning_rate": 9.77515633882888e-06, "loss": 1.3048, "step": 6828 }, { "epoch": 0.7266439668014472, "grad_norm": 17.171875, "learning_rate": 9.775120807276862e-06, "loss": 1.0213, "step": 6829 }, { "epoch": 0.7267503724196638, "grad_norm": 11.9921875, "learning_rate": 9.775085275724845e-06, "loss": 1.1627, "step": 6830 }, { "epoch": 0.7268567780378804, "grad_norm": 9.515625, "learning_rate": 9.775049744172827e-06, "loss": 0.9092, "step": 6831 }, { "epoch": 0.726963183656097, "grad_norm": 13.3984375, "learning_rate": 9.775014212620808e-06, "loss": 1.253, "step": 6832 }, { "epoch": 0.7270695892743136, "grad_norm": 14.4296875, "learning_rate": 9.77497868106879e-06, "loss": 1.0029, "step": 6833 }, { "epoch": 0.7271759948925304, "grad_norm": 12.8984375, "learning_rate": 9.774943149516771e-06, "loss": 1.5913, "step": 6834 }, { "epoch": 0.727282400510747, "grad_norm": 16.828125, "learning_rate": 9.774907617964755e-06, "loss": 1.3272, "step": 6835 }, { "epoch": 0.7273888061289636, "grad_norm": 14.0546875, "learning_rate": 9.774872086412736e-06, "loss": 1.3856, "step": 6836 }, { "epoch": 0.7274952117471802, "grad_norm": 10.28125, "learning_rate": 9.774836554860718e-06, "loss": 0.8729, "step": 6837 }, { "epoch": 0.7276016173653969, "grad_norm": 13.7109375, "learning_rate": 9.7748010233087e-06, "loss": 1.4706, "step": 6838 }, { "epoch": 0.7277080229836136, "grad_norm": 13.859375, "learning_rate": 9.774765491756681e-06, "loss": 1.2248, "step": 6839 }, { "epoch": 0.7278144286018302, "grad_norm": 178.25, "learning_rate": 9.774729960204663e-06, "loss": 2.606, "step": 6840 }, { "epoch": 0.7279208342200468, "grad_norm": 14.6953125, "learning_rate": 9.774694428652644e-06, "loss": 0.9734, "step": 6841 }, { "epoch": 0.7280272398382635, "grad_norm": 12.8359375, "learning_rate": 9.774658897100626e-06, "loss": 1.1532, "step": 6842 }, { "epoch": 0.7281336454564801, "grad_norm": 12.1640625, "learning_rate": 9.774623365548607e-06, "loss": 1.2682, "step": 6843 }, { "epoch": 0.7282400510746967, "grad_norm": 14.90625, "learning_rate": 9.774587833996589e-06, "loss": 1.371, "step": 6844 }, { "epoch": 0.7283464566929134, "grad_norm": 14.6640625, "learning_rate": 9.774552302444572e-06, "loss": 1.563, "step": 6845 }, { "epoch": 0.72845286231113, "grad_norm": 13.46875, "learning_rate": 9.774516770892554e-06, "loss": 1.291, "step": 6846 }, { "epoch": 0.7285592679293467, "grad_norm": 13.703125, "learning_rate": 9.774481239340536e-06, "loss": 1.5196, "step": 6847 }, { "epoch": 0.7286656735475633, "grad_norm": 13.0859375, "learning_rate": 9.774445707788517e-06, "loss": 1.2131, "step": 6848 }, { "epoch": 0.7287720791657799, "grad_norm": 12.859375, "learning_rate": 9.774410176236499e-06, "loss": 1.2381, "step": 6849 }, { "epoch": 0.7288784847839966, "grad_norm": 14.9921875, "learning_rate": 9.77437464468448e-06, "loss": 1.3387, "step": 6850 }, { "epoch": 0.7289848904022133, "grad_norm": 12.828125, "learning_rate": 9.774339113132462e-06, "loss": 1.3732, "step": 6851 }, { "epoch": 0.7290912960204299, "grad_norm": 21.296875, "learning_rate": 9.774303581580444e-06, "loss": 1.2305, "step": 6852 }, { "epoch": 0.7291977016386465, "grad_norm": 12.8046875, "learning_rate": 9.774268050028425e-06, "loss": 0.999, "step": 6853 }, { "epoch": 0.7293041072568631, "grad_norm": 11.1796875, "learning_rate": 9.774232518476407e-06, "loss": 1.4816, "step": 6854 }, { "epoch": 0.7294105128750799, "grad_norm": 11.921875, "learning_rate": 9.774196986924388e-06, "loss": 1.1514, "step": 6855 }, { "epoch": 0.7295169184932965, "grad_norm": 12.015625, "learning_rate": 9.774161455372372e-06, "loss": 1.2829, "step": 6856 }, { "epoch": 0.7296233241115131, "grad_norm": 14.140625, "learning_rate": 9.774125923820353e-06, "loss": 1.1198, "step": 6857 }, { "epoch": 0.7297297297297297, "grad_norm": 15.140625, "learning_rate": 9.774090392268335e-06, "loss": 1.1728, "step": 6858 }, { "epoch": 0.7298361353479463, "grad_norm": 15.5546875, "learning_rate": 9.774054860716317e-06, "loss": 1.3642, "step": 6859 }, { "epoch": 0.7299425409661631, "grad_norm": 13.953125, "learning_rate": 9.774019329164298e-06, "loss": 1.1533, "step": 6860 }, { "epoch": 0.7300489465843797, "grad_norm": 9.3984375, "learning_rate": 9.773983797612281e-06, "loss": 0.9364, "step": 6861 }, { "epoch": 0.7301553522025963, "grad_norm": 11.515625, "learning_rate": 9.773948266060263e-06, "loss": 1.1391, "step": 6862 }, { "epoch": 0.7302617578208129, "grad_norm": 10.75, "learning_rate": 9.773912734508245e-06, "loss": 0.984, "step": 6863 }, { "epoch": 0.7303681634390295, "grad_norm": 11.609375, "learning_rate": 9.773877202956226e-06, "loss": 1.1916, "step": 6864 }, { "epoch": 0.7304745690572462, "grad_norm": 19.609375, "learning_rate": 9.773841671404208e-06, "loss": 1.6122, "step": 6865 }, { "epoch": 0.7305809746754629, "grad_norm": 9.0390625, "learning_rate": 9.77380613985219e-06, "loss": 0.9254, "step": 6866 }, { "epoch": 0.7306873802936795, "grad_norm": 22.625, "learning_rate": 9.773770608300171e-06, "loss": 1.3463, "step": 6867 }, { "epoch": 0.7307937859118961, "grad_norm": 10.8828125, "learning_rate": 9.773735076748153e-06, "loss": 1.2355, "step": 6868 }, { "epoch": 0.7309001915301128, "grad_norm": 11.75, "learning_rate": 9.773699545196134e-06, "loss": 1.1901, "step": 6869 }, { "epoch": 0.7310065971483294, "grad_norm": 11.4296875, "learning_rate": 9.773664013644116e-06, "loss": 1.1425, "step": 6870 }, { "epoch": 0.7311130027665461, "grad_norm": 15.671875, "learning_rate": 9.7736284820921e-06, "loss": 1.1876, "step": 6871 }, { "epoch": 0.7312194083847627, "grad_norm": 12.1015625, "learning_rate": 9.773592950540081e-06, "loss": 1.2702, "step": 6872 }, { "epoch": 0.7313258140029794, "grad_norm": 13.1953125, "learning_rate": 9.773557418988062e-06, "loss": 1.3627, "step": 6873 }, { "epoch": 0.731432219621196, "grad_norm": 13.0, "learning_rate": 9.773521887436044e-06, "loss": 1.3392, "step": 6874 }, { "epoch": 0.7315386252394126, "grad_norm": 15.3515625, "learning_rate": 9.773486355884026e-06, "loss": 0.9521, "step": 6875 }, { "epoch": 0.7316450308576293, "grad_norm": 12.4921875, "learning_rate": 9.773450824332007e-06, "loss": 1.1811, "step": 6876 }, { "epoch": 0.731751436475846, "grad_norm": 11.7109375, "learning_rate": 9.773415292779989e-06, "loss": 1.3091, "step": 6877 }, { "epoch": 0.7318578420940626, "grad_norm": 13.1484375, "learning_rate": 9.77337976122797e-06, "loss": 1.6881, "step": 6878 }, { "epoch": 0.7319642477122792, "grad_norm": 17.953125, "learning_rate": 9.773344229675952e-06, "loss": 1.3798, "step": 6879 }, { "epoch": 0.7320706533304958, "grad_norm": 11.421875, "learning_rate": 9.773308698123934e-06, "loss": 0.9979, "step": 6880 }, { "epoch": 0.7321770589487125, "grad_norm": 10.7890625, "learning_rate": 9.773273166571917e-06, "loss": 1.1595, "step": 6881 }, { "epoch": 0.7322834645669292, "grad_norm": 11.3671875, "learning_rate": 9.773237635019899e-06, "loss": 1.2249, "step": 6882 }, { "epoch": 0.7323898701851458, "grad_norm": 12.640625, "learning_rate": 9.77320210346788e-06, "loss": 1.1935, "step": 6883 }, { "epoch": 0.7324962758033624, "grad_norm": 17.171875, "learning_rate": 9.773166571915862e-06, "loss": 1.1995, "step": 6884 }, { "epoch": 0.732602681421579, "grad_norm": 12.796875, "learning_rate": 9.773131040363843e-06, "loss": 1.3708, "step": 6885 }, { "epoch": 0.7327090870397956, "grad_norm": 12.828125, "learning_rate": 9.773095508811827e-06, "loss": 1.4996, "step": 6886 }, { "epoch": 0.7328154926580124, "grad_norm": 16.484375, "learning_rate": 9.773059977259808e-06, "loss": 1.1512, "step": 6887 }, { "epoch": 0.732921898276229, "grad_norm": 17.53125, "learning_rate": 9.77302444570779e-06, "loss": 1.1928, "step": 6888 }, { "epoch": 0.7330283038944456, "grad_norm": 12.109375, "learning_rate": 9.772988914155772e-06, "loss": 1.1612, "step": 6889 }, { "epoch": 0.7331347095126622, "grad_norm": 14.28125, "learning_rate": 9.772953382603753e-06, "loss": 1.1798, "step": 6890 }, { "epoch": 0.7332411151308789, "grad_norm": 12.1171875, "learning_rate": 9.772917851051735e-06, "loss": 1.5035, "step": 6891 }, { "epoch": 0.7333475207490956, "grad_norm": 11.171875, "learning_rate": 9.772882319499716e-06, "loss": 1.1612, "step": 6892 }, { "epoch": 0.7334539263673122, "grad_norm": 12.609375, "learning_rate": 9.772846787947698e-06, "loss": 1.1724, "step": 6893 }, { "epoch": 0.7335603319855288, "grad_norm": 13.7421875, "learning_rate": 9.77281125639568e-06, "loss": 1.3198, "step": 6894 }, { "epoch": 0.7336667376037455, "grad_norm": 19.9375, "learning_rate": 9.772775724843661e-06, "loss": 1.6068, "step": 6895 }, { "epoch": 0.7337731432219621, "grad_norm": 15.2734375, "learning_rate": 9.772740193291643e-06, "loss": 1.0952, "step": 6896 }, { "epoch": 0.7338795488401788, "grad_norm": 20.8125, "learning_rate": 9.772704661739626e-06, "loss": 1.05, "step": 6897 }, { "epoch": 0.7339859544583954, "grad_norm": 11.5625, "learning_rate": 9.772669130187608e-06, "loss": 1.4121, "step": 6898 }, { "epoch": 0.734092360076612, "grad_norm": 13.1015625, "learning_rate": 9.77263359863559e-06, "loss": 1.0088, "step": 6899 }, { "epoch": 0.7341987656948287, "grad_norm": 11.9140625, "learning_rate": 9.772598067083571e-06, "loss": 1.2103, "step": 6900 }, { "epoch": 0.7343051713130453, "grad_norm": 8.8359375, "learning_rate": 9.772562535531553e-06, "loss": 0.9194, "step": 6901 }, { "epoch": 0.734411576931262, "grad_norm": 14.3515625, "learning_rate": 9.772527003979536e-06, "loss": 1.138, "step": 6902 }, { "epoch": 0.7345179825494786, "grad_norm": 12.8828125, "learning_rate": 9.772491472427517e-06, "loss": 1.314, "step": 6903 }, { "epoch": 0.7346243881676953, "grad_norm": 11.0546875, "learning_rate": 9.772455940875497e-06, "loss": 1.1106, "step": 6904 }, { "epoch": 0.7347307937859119, "grad_norm": 11.875, "learning_rate": 9.772420409323479e-06, "loss": 1.2382, "step": 6905 }, { "epoch": 0.7348371994041285, "grad_norm": 11.8125, "learning_rate": 9.77238487777146e-06, "loss": 1.2497, "step": 6906 }, { "epoch": 0.7349436050223451, "grad_norm": 13.0234375, "learning_rate": 9.772349346219444e-06, "loss": 1.1593, "step": 6907 }, { "epoch": 0.7350500106405619, "grad_norm": 16.75, "learning_rate": 9.772313814667425e-06, "loss": 1.3967, "step": 6908 }, { "epoch": 0.7351564162587785, "grad_norm": 12.21875, "learning_rate": 9.772278283115407e-06, "loss": 0.9915, "step": 6909 }, { "epoch": 0.7352628218769951, "grad_norm": 15.2734375, "learning_rate": 9.772242751563389e-06, "loss": 1.0263, "step": 6910 }, { "epoch": 0.7353692274952117, "grad_norm": 12.9140625, "learning_rate": 9.77220722001137e-06, "loss": 1.2482, "step": 6911 }, { "epoch": 0.7354756331134283, "grad_norm": 14.34375, "learning_rate": 9.772171688459354e-06, "loss": 0.9541, "step": 6912 }, { "epoch": 0.7355820387316451, "grad_norm": 14.3515625, "learning_rate": 9.772136156907335e-06, "loss": 1.3587, "step": 6913 }, { "epoch": 0.7356884443498617, "grad_norm": 12.609375, "learning_rate": 9.772100625355317e-06, "loss": 1.2884, "step": 6914 }, { "epoch": 0.7357948499680783, "grad_norm": 10.765625, "learning_rate": 9.772065093803298e-06, "loss": 1.1695, "step": 6915 }, { "epoch": 0.7359012555862949, "grad_norm": 12.359375, "learning_rate": 9.77202956225128e-06, "loss": 1.1016, "step": 6916 }, { "epoch": 0.7360076612045116, "grad_norm": 9.9609375, "learning_rate": 9.771994030699262e-06, "loss": 1.0171, "step": 6917 }, { "epoch": 0.7361140668227283, "grad_norm": 9.828125, "learning_rate": 9.771958499147243e-06, "loss": 1.1778, "step": 6918 }, { "epoch": 0.7362204724409449, "grad_norm": 11.8515625, "learning_rate": 9.771922967595225e-06, "loss": 1.0815, "step": 6919 }, { "epoch": 0.7363268780591615, "grad_norm": 16.03125, "learning_rate": 9.771887436043206e-06, "loss": 1.3586, "step": 6920 }, { "epoch": 0.7364332836773781, "grad_norm": 15.0625, "learning_rate": 9.771851904491188e-06, "loss": 1.3622, "step": 6921 }, { "epoch": 0.7365396892955948, "grad_norm": 15.0, "learning_rate": 9.771816372939171e-06, "loss": 1.2437, "step": 6922 }, { "epoch": 0.7366460949138115, "grad_norm": 14.625, "learning_rate": 9.771780841387153e-06, "loss": 1.1557, "step": 6923 }, { "epoch": 0.7367525005320281, "grad_norm": 17.65625, "learning_rate": 9.771745309835135e-06, "loss": 0.9408, "step": 6924 }, { "epoch": 0.7368589061502447, "grad_norm": 14.4609375, "learning_rate": 9.771709778283116e-06, "loss": 1.3673, "step": 6925 }, { "epoch": 0.7369653117684614, "grad_norm": 12.8515625, "learning_rate": 9.771674246731098e-06, "loss": 1.3462, "step": 6926 }, { "epoch": 0.737071717386678, "grad_norm": 13.6875, "learning_rate": 9.771638715179081e-06, "loss": 1.337, "step": 6927 }, { "epoch": 0.7371781230048946, "grad_norm": 12.09375, "learning_rate": 9.771603183627063e-06, "loss": 1.1117, "step": 6928 }, { "epoch": 0.7372845286231113, "grad_norm": 15.4375, "learning_rate": 9.771567652075044e-06, "loss": 1.0816, "step": 6929 }, { "epoch": 0.737390934241328, "grad_norm": 11.6171875, "learning_rate": 9.771532120523026e-06, "loss": 1.4527, "step": 6930 }, { "epoch": 0.7374973398595446, "grad_norm": 12.03125, "learning_rate": 9.771496588971008e-06, "loss": 1.0568, "step": 6931 }, { "epoch": 0.7376037454777612, "grad_norm": 18.25, "learning_rate": 9.771461057418989e-06, "loss": 1.483, "step": 6932 }, { "epoch": 0.7377101510959778, "grad_norm": 12.5625, "learning_rate": 9.77142552586697e-06, "loss": 1.2712, "step": 6933 }, { "epoch": 0.7378165567141945, "grad_norm": 14.984375, "learning_rate": 9.771389994314952e-06, "loss": 1.0224, "step": 6934 }, { "epoch": 0.7379229623324112, "grad_norm": 12.9140625, "learning_rate": 9.771354462762934e-06, "loss": 1.2163, "step": 6935 }, { "epoch": 0.7380293679506278, "grad_norm": 10.953125, "learning_rate": 9.771318931210916e-06, "loss": 1.0945, "step": 6936 }, { "epoch": 0.7381357735688444, "grad_norm": 19.046875, "learning_rate": 9.771283399658897e-06, "loss": 1.2101, "step": 6937 }, { "epoch": 0.738242179187061, "grad_norm": 9.515625, "learning_rate": 9.77124786810688e-06, "loss": 1.3628, "step": 6938 }, { "epoch": 0.7383485848052778, "grad_norm": 15.6796875, "learning_rate": 9.771212336554862e-06, "loss": 1.7856, "step": 6939 }, { "epoch": 0.7384549904234944, "grad_norm": 15.1171875, "learning_rate": 9.771176805002844e-06, "loss": 1.2742, "step": 6940 }, { "epoch": 0.738561396041711, "grad_norm": 20.640625, "learning_rate": 9.771141273450825e-06, "loss": 1.4825, "step": 6941 }, { "epoch": 0.7386678016599276, "grad_norm": 12.4375, "learning_rate": 9.771105741898807e-06, "loss": 1.2089, "step": 6942 }, { "epoch": 0.7387742072781442, "grad_norm": 20.40625, "learning_rate": 9.771070210346788e-06, "loss": 1.3859, "step": 6943 }, { "epoch": 0.738880612896361, "grad_norm": 12.2109375, "learning_rate": 9.77103467879477e-06, "loss": 1.0658, "step": 6944 }, { "epoch": 0.7389870185145776, "grad_norm": 12.921875, "learning_rate": 9.770999147242752e-06, "loss": 1.375, "step": 6945 }, { "epoch": 0.7390934241327942, "grad_norm": 9.8203125, "learning_rate": 9.770963615690733e-06, "loss": 1.2344, "step": 6946 }, { "epoch": 0.7391998297510108, "grad_norm": 12.0703125, "learning_rate": 9.770928084138715e-06, "loss": 1.5551, "step": 6947 }, { "epoch": 0.7393062353692275, "grad_norm": 13.140625, "learning_rate": 9.770892552586698e-06, "loss": 0.9835, "step": 6948 }, { "epoch": 0.7394126409874442, "grad_norm": 12.359375, "learning_rate": 9.77085702103468e-06, "loss": 1.1235, "step": 6949 }, { "epoch": 0.7395190466056608, "grad_norm": 13.5078125, "learning_rate": 9.770821489482661e-06, "loss": 1.4641, "step": 6950 }, { "epoch": 0.7396254522238774, "grad_norm": 13.34375, "learning_rate": 9.770785957930643e-06, "loss": 1.1887, "step": 6951 }, { "epoch": 0.739731857842094, "grad_norm": 12.3203125, "learning_rate": 9.770750426378625e-06, "loss": 1.0533, "step": 6952 }, { "epoch": 0.7398382634603107, "grad_norm": 12.2734375, "learning_rate": 9.770714894826608e-06, "loss": 1.3121, "step": 6953 }, { "epoch": 0.7399446690785273, "grad_norm": 11.7890625, "learning_rate": 9.77067936327459e-06, "loss": 1.4742, "step": 6954 }, { "epoch": 0.740051074696744, "grad_norm": 12.421875, "learning_rate": 9.770643831722571e-06, "loss": 1.1347, "step": 6955 }, { "epoch": 0.7401574803149606, "grad_norm": 11.4375, "learning_rate": 9.770608300170553e-06, "loss": 1.1699, "step": 6956 }, { "epoch": 0.7402638859331773, "grad_norm": 9.7265625, "learning_rate": 9.770572768618534e-06, "loss": 1.0038, "step": 6957 }, { "epoch": 0.7403702915513939, "grad_norm": 14.5078125, "learning_rate": 9.770537237066516e-06, "loss": 1.0021, "step": 6958 }, { "epoch": 0.7404766971696105, "grad_norm": 9.1171875, "learning_rate": 9.770501705514498e-06, "loss": 1.2388, "step": 6959 }, { "epoch": 0.7405831027878272, "grad_norm": 14.96875, "learning_rate": 9.77046617396248e-06, "loss": 1.1892, "step": 6960 }, { "epoch": 0.7406895084060439, "grad_norm": 14.515625, "learning_rate": 9.77043064241046e-06, "loss": 1.3289, "step": 6961 }, { "epoch": 0.7407959140242605, "grad_norm": 11.90625, "learning_rate": 9.770395110858442e-06, "loss": 1.3408, "step": 6962 }, { "epoch": 0.7409023196424771, "grad_norm": 8.984375, "learning_rate": 9.770359579306426e-06, "loss": 1.0242, "step": 6963 }, { "epoch": 0.7410087252606937, "grad_norm": 13.8515625, "learning_rate": 9.770324047754407e-06, "loss": 1.4633, "step": 6964 }, { "epoch": 0.7411151308789105, "grad_norm": 13.5234375, "learning_rate": 9.770288516202389e-06, "loss": 1.4428, "step": 6965 }, { "epoch": 0.7412215364971271, "grad_norm": 9.6953125, "learning_rate": 9.77025298465037e-06, "loss": 1.1141, "step": 6966 }, { "epoch": 0.7413279421153437, "grad_norm": 11.7109375, "learning_rate": 9.770217453098352e-06, "loss": 1.3463, "step": 6967 }, { "epoch": 0.7414343477335603, "grad_norm": 14.6953125, "learning_rate": 9.770181921546334e-06, "loss": 1.2169, "step": 6968 }, { "epoch": 0.7415407533517769, "grad_norm": 15.1484375, "learning_rate": 9.770146389994315e-06, "loss": 1.3346, "step": 6969 }, { "epoch": 0.7416471589699937, "grad_norm": 11.21875, "learning_rate": 9.770110858442297e-06, "loss": 1.1795, "step": 6970 }, { "epoch": 0.7417535645882103, "grad_norm": 13.015625, "learning_rate": 9.770075326890279e-06, "loss": 1.4427, "step": 6971 }, { "epoch": 0.7418599702064269, "grad_norm": 11.2890625, "learning_rate": 9.77003979533826e-06, "loss": 1.1652, "step": 6972 }, { "epoch": 0.7419663758246435, "grad_norm": 16.609375, "learning_rate": 9.770004263786242e-06, "loss": 1.1894, "step": 6973 }, { "epoch": 0.7420727814428602, "grad_norm": 10.3828125, "learning_rate": 9.769968732234225e-06, "loss": 1.1043, "step": 6974 }, { "epoch": 0.7421791870610768, "grad_norm": 13.6796875, "learning_rate": 9.769933200682207e-06, "loss": 1.2111, "step": 6975 }, { "epoch": 0.7422855926792935, "grad_norm": 12.203125, "learning_rate": 9.769897669130188e-06, "loss": 0.9166, "step": 6976 }, { "epoch": 0.7423919982975101, "grad_norm": 11.0234375, "learning_rate": 9.76986213757817e-06, "loss": 1.2327, "step": 6977 }, { "epoch": 0.7424984039157267, "grad_norm": 15.2890625, "learning_rate": 9.769826606026151e-06, "loss": 1.0996, "step": 6978 }, { "epoch": 0.7426048095339434, "grad_norm": 10.2890625, "learning_rate": 9.769791074474135e-06, "loss": 1.0525, "step": 6979 }, { "epoch": 0.74271121515216, "grad_norm": 16.3125, "learning_rate": 9.769755542922116e-06, "loss": 1.2906, "step": 6980 }, { "epoch": 0.7428176207703767, "grad_norm": 12.2421875, "learning_rate": 9.769720011370098e-06, "loss": 1.1492, "step": 6981 }, { "epoch": 0.7429240263885933, "grad_norm": 13.1015625, "learning_rate": 9.76968447981808e-06, "loss": 1.0109, "step": 6982 }, { "epoch": 0.74303043200681, "grad_norm": 14.9453125, "learning_rate": 9.769648948266061e-06, "loss": 0.8745, "step": 6983 }, { "epoch": 0.7431368376250266, "grad_norm": 13.15625, "learning_rate": 9.769613416714043e-06, "loss": 1.3149, "step": 6984 }, { "epoch": 0.7432432432432432, "grad_norm": 12.328125, "learning_rate": 9.769577885162024e-06, "loss": 1.4339, "step": 6985 }, { "epoch": 0.7433496488614599, "grad_norm": 11.5546875, "learning_rate": 9.769542353610006e-06, "loss": 1.0196, "step": 6986 }, { "epoch": 0.7434560544796766, "grad_norm": 14.625, "learning_rate": 9.769506822057988e-06, "loss": 1.2707, "step": 6987 }, { "epoch": 0.7435624600978932, "grad_norm": 12.640625, "learning_rate": 9.76947129050597e-06, "loss": 1.2331, "step": 6988 }, { "epoch": 0.7436688657161098, "grad_norm": 14.8046875, "learning_rate": 9.769435758953953e-06, "loss": 1.2944, "step": 6989 }, { "epoch": 0.7437752713343264, "grad_norm": 12.1953125, "learning_rate": 9.769400227401934e-06, "loss": 1.2212, "step": 6990 }, { "epoch": 0.7438816769525431, "grad_norm": 9.984375, "learning_rate": 9.769364695849916e-06, "loss": 1.1596, "step": 6991 }, { "epoch": 0.7439880825707598, "grad_norm": 11.9921875, "learning_rate": 9.769329164297897e-06, "loss": 1.1346, "step": 6992 }, { "epoch": 0.7440944881889764, "grad_norm": 15.8671875, "learning_rate": 9.769293632745879e-06, "loss": 1.3259, "step": 6993 }, { "epoch": 0.744200893807193, "grad_norm": 11.515625, "learning_rate": 9.76925810119386e-06, "loss": 1.2985, "step": 6994 }, { "epoch": 0.7443072994254096, "grad_norm": 12.921875, "learning_rate": 9.769222569641842e-06, "loss": 1.1569, "step": 6995 }, { "epoch": 0.7444137050436263, "grad_norm": 9.6015625, "learning_rate": 9.769187038089824e-06, "loss": 1.1161, "step": 6996 }, { "epoch": 0.744520110661843, "grad_norm": 14.59375, "learning_rate": 9.769151506537805e-06, "loss": 1.2942, "step": 6997 }, { "epoch": 0.7446265162800596, "grad_norm": 13.359375, "learning_rate": 9.769115974985787e-06, "loss": 1.2794, "step": 6998 }, { "epoch": 0.7447329218982762, "grad_norm": 12.140625, "learning_rate": 9.76908044343377e-06, "loss": 1.0343, "step": 6999 }, { "epoch": 0.7448393275164928, "grad_norm": 10.28125, "learning_rate": 9.769044911881752e-06, "loss": 0.9073, "step": 7000 }, { "epoch": 0.7449457331347095, "grad_norm": 11.203125, "learning_rate": 9.769009380329734e-06, "loss": 1.0169, "step": 7001 }, { "epoch": 0.7450521387529262, "grad_norm": 10.75, "learning_rate": 9.768973848777715e-06, "loss": 1.1168, "step": 7002 }, { "epoch": 0.7451585443711428, "grad_norm": 12.921875, "learning_rate": 9.768938317225697e-06, "loss": 1.0585, "step": 7003 }, { "epoch": 0.7452649499893594, "grad_norm": 16.25, "learning_rate": 9.76890278567368e-06, "loss": 1.1974, "step": 7004 }, { "epoch": 0.7453713556075761, "grad_norm": 7.85546875, "learning_rate": 9.768867254121662e-06, "loss": 0.7552, "step": 7005 }, { "epoch": 0.7454777612257927, "grad_norm": 11.765625, "learning_rate": 9.768831722569643e-06, "loss": 1.1021, "step": 7006 }, { "epoch": 0.7455841668440094, "grad_norm": 9.4609375, "learning_rate": 9.768796191017625e-06, "loss": 1.0689, "step": 7007 }, { "epoch": 0.745690572462226, "grad_norm": 8.796875, "learning_rate": 9.768760659465606e-06, "loss": 1.1113, "step": 7008 }, { "epoch": 0.7457969780804427, "grad_norm": 12.8515625, "learning_rate": 9.768725127913588e-06, "loss": 1.1862, "step": 7009 }, { "epoch": 0.7459033836986593, "grad_norm": 13.1171875, "learning_rate": 9.76868959636157e-06, "loss": 1.293, "step": 7010 }, { "epoch": 0.7460097893168759, "grad_norm": 10.6484375, "learning_rate": 9.768654064809551e-06, "loss": 0.8191, "step": 7011 }, { "epoch": 0.7461161949350926, "grad_norm": 14.4609375, "learning_rate": 9.768618533257533e-06, "loss": 1.1494, "step": 7012 }, { "epoch": 0.7462226005533092, "grad_norm": 17.390625, "learning_rate": 9.768583001705515e-06, "loss": 1.3871, "step": 7013 }, { "epoch": 0.7463290061715259, "grad_norm": 11.4921875, "learning_rate": 9.768547470153496e-06, "loss": 1.0238, "step": 7014 }, { "epoch": 0.7464354117897425, "grad_norm": 10.4140625, "learning_rate": 9.76851193860148e-06, "loss": 1.0992, "step": 7015 }, { "epoch": 0.7465418174079591, "grad_norm": 11.7265625, "learning_rate": 9.768476407049461e-06, "loss": 1.0812, "step": 7016 }, { "epoch": 0.7466482230261757, "grad_norm": 10.1328125, "learning_rate": 9.768440875497443e-06, "loss": 0.9478, "step": 7017 }, { "epoch": 0.7467546286443925, "grad_norm": 13.578125, "learning_rate": 9.768405343945424e-06, "loss": 1.1766, "step": 7018 }, { "epoch": 0.7468610342626091, "grad_norm": 11.9453125, "learning_rate": 9.768369812393406e-06, "loss": 1.1745, "step": 7019 }, { "epoch": 0.7469674398808257, "grad_norm": 20.15625, "learning_rate": 9.768334280841389e-06, "loss": 0.9768, "step": 7020 }, { "epoch": 0.7470738454990423, "grad_norm": 12.390625, "learning_rate": 9.76829874928937e-06, "loss": 1.0046, "step": 7021 }, { "epoch": 0.7471802511172589, "grad_norm": 18.640625, "learning_rate": 9.768263217737352e-06, "loss": 1.0624, "step": 7022 }, { "epoch": 0.7472866567354757, "grad_norm": 18.125, "learning_rate": 9.768227686185334e-06, "loss": 1.057, "step": 7023 }, { "epoch": 0.7473930623536923, "grad_norm": 11.8828125, "learning_rate": 9.768192154633314e-06, "loss": 1.2952, "step": 7024 }, { "epoch": 0.7474994679719089, "grad_norm": 12.0859375, "learning_rate": 9.768156623081297e-06, "loss": 1.6165, "step": 7025 }, { "epoch": 0.7476058735901255, "grad_norm": 12.90625, "learning_rate": 9.768121091529279e-06, "loss": 1.0661, "step": 7026 }, { "epoch": 0.7477122792083422, "grad_norm": 15.1796875, "learning_rate": 9.76808555997726e-06, "loss": 1.293, "step": 7027 }, { "epoch": 0.7478186848265589, "grad_norm": 12.765625, "learning_rate": 9.768050028425242e-06, "loss": 1.1298, "step": 7028 }, { "epoch": 0.7479250904447755, "grad_norm": 28.671875, "learning_rate": 9.768014496873224e-06, "loss": 1.42, "step": 7029 }, { "epoch": 0.7480314960629921, "grad_norm": 12.4296875, "learning_rate": 9.767978965321207e-06, "loss": 1.2662, "step": 7030 }, { "epoch": 0.7481379016812088, "grad_norm": 14.6484375, "learning_rate": 9.767943433769189e-06, "loss": 1.1656, "step": 7031 }, { "epoch": 0.7482443072994254, "grad_norm": 15.1484375, "learning_rate": 9.76790790221717e-06, "loss": 1.1245, "step": 7032 }, { "epoch": 0.7483507129176421, "grad_norm": 11.5, "learning_rate": 9.767872370665152e-06, "loss": 0.9707, "step": 7033 }, { "epoch": 0.7484571185358587, "grad_norm": 14.8984375, "learning_rate": 9.767836839113133e-06, "loss": 1.0716, "step": 7034 }, { "epoch": 0.7485635241540753, "grad_norm": 11.3125, "learning_rate": 9.767801307561115e-06, "loss": 0.9945, "step": 7035 }, { "epoch": 0.748669929772292, "grad_norm": 11.8046875, "learning_rate": 9.767765776009097e-06, "loss": 1.4068, "step": 7036 }, { "epoch": 0.7487763353905086, "grad_norm": 9.515625, "learning_rate": 9.767730244457078e-06, "loss": 1.1512, "step": 7037 }, { "epoch": 0.7488827410087252, "grad_norm": 9.8359375, "learning_rate": 9.76769471290506e-06, "loss": 1.0225, "step": 7038 }, { "epoch": 0.7489891466269419, "grad_norm": 15.7421875, "learning_rate": 9.767659181353041e-06, "loss": 1.3666, "step": 7039 }, { "epoch": 0.7490955522451586, "grad_norm": 12.328125, "learning_rate": 9.767623649801025e-06, "loss": 1.3304, "step": 7040 }, { "epoch": 0.7492019578633752, "grad_norm": 11.34375, "learning_rate": 9.767588118249006e-06, "loss": 1.0584, "step": 7041 }, { "epoch": 0.7493083634815918, "grad_norm": 11.84375, "learning_rate": 9.767552586696988e-06, "loss": 1.2619, "step": 7042 }, { "epoch": 0.7494147690998084, "grad_norm": 13.671875, "learning_rate": 9.76751705514497e-06, "loss": 0.9996, "step": 7043 }, { "epoch": 0.7495211747180252, "grad_norm": 15.015625, "learning_rate": 9.767481523592951e-06, "loss": 1.3772, "step": 7044 }, { "epoch": 0.7496275803362418, "grad_norm": 11.28125, "learning_rate": 9.767445992040934e-06, "loss": 0.8693, "step": 7045 }, { "epoch": 0.7497339859544584, "grad_norm": 10.078125, "learning_rate": 9.767410460488916e-06, "loss": 0.9283, "step": 7046 }, { "epoch": 0.749840391572675, "grad_norm": 13.1171875, "learning_rate": 9.767374928936898e-06, "loss": 0.9417, "step": 7047 }, { "epoch": 0.7499467971908916, "grad_norm": 10.7890625, "learning_rate": 9.76733939738488e-06, "loss": 1.1266, "step": 7048 }, { "epoch": 0.7500532028091084, "grad_norm": 15.0546875, "learning_rate": 9.76730386583286e-06, "loss": 1.1453, "step": 7049 }, { "epoch": 0.750159608427325, "grad_norm": 15.0234375, "learning_rate": 9.767268334280842e-06, "loss": 1.2227, "step": 7050 }, { "epoch": 0.7502660140455416, "grad_norm": 30.671875, "learning_rate": 9.767232802728824e-06, "loss": 1.0048, "step": 7051 }, { "epoch": 0.7503724196637582, "grad_norm": 11.890625, "learning_rate": 9.767197271176806e-06, "loss": 1.6008, "step": 7052 }, { "epoch": 0.7504788252819748, "grad_norm": 11.734375, "learning_rate": 9.767161739624787e-06, "loss": 1.3694, "step": 7053 }, { "epoch": 0.7505852309001916, "grad_norm": 15.0859375, "learning_rate": 9.767126208072769e-06, "loss": 1.1429, "step": 7054 }, { "epoch": 0.7506916365184082, "grad_norm": 13.75, "learning_rate": 9.76709067652075e-06, "loss": 1.3769, "step": 7055 }, { "epoch": 0.7507980421366248, "grad_norm": 16.59375, "learning_rate": 9.767055144968734e-06, "loss": 1.3604, "step": 7056 }, { "epoch": 0.7509044477548414, "grad_norm": 18.53125, "learning_rate": 9.767019613416715e-06, "loss": 1.1986, "step": 7057 }, { "epoch": 0.7510108533730581, "grad_norm": 12.9375, "learning_rate": 9.766984081864697e-06, "loss": 0.8292, "step": 7058 }, { "epoch": 0.7511172589912748, "grad_norm": 17.203125, "learning_rate": 9.766948550312679e-06, "loss": 1.1869, "step": 7059 }, { "epoch": 0.7512236646094914, "grad_norm": 14.0859375, "learning_rate": 9.76691301876066e-06, "loss": 1.2265, "step": 7060 }, { "epoch": 0.751330070227708, "grad_norm": 10.015625, "learning_rate": 9.766877487208642e-06, "loss": 1.3889, "step": 7061 }, { "epoch": 0.7514364758459247, "grad_norm": 13.2578125, "learning_rate": 9.766841955656623e-06, "loss": 1.1785, "step": 7062 }, { "epoch": 0.7515428814641413, "grad_norm": 12.3203125, "learning_rate": 9.766806424104605e-06, "loss": 1.2668, "step": 7063 }, { "epoch": 0.7516492870823579, "grad_norm": 10.34375, "learning_rate": 9.766770892552587e-06, "loss": 1.2191, "step": 7064 }, { "epoch": 0.7517556927005746, "grad_norm": 13.5, "learning_rate": 9.766735361000568e-06, "loss": 1.555, "step": 7065 }, { "epoch": 0.7518620983187912, "grad_norm": 10.1171875, "learning_rate": 9.766699829448552e-06, "loss": 1.0036, "step": 7066 }, { "epoch": 0.7519685039370079, "grad_norm": 10.03125, "learning_rate": 9.766664297896533e-06, "loss": 1.0855, "step": 7067 }, { "epoch": 0.7520749095552245, "grad_norm": 12.640625, "learning_rate": 9.766628766344515e-06, "loss": 1.0638, "step": 7068 }, { "epoch": 0.7521813151734411, "grad_norm": 24.75, "learning_rate": 9.766593234792496e-06, "loss": 1.2242, "step": 7069 }, { "epoch": 0.7522877207916578, "grad_norm": 10.703125, "learning_rate": 9.766557703240478e-06, "loss": 1.0749, "step": 7070 }, { "epoch": 0.7523941264098745, "grad_norm": 12.5234375, "learning_rate": 9.766522171688461e-06, "loss": 1.3083, "step": 7071 }, { "epoch": 0.7525005320280911, "grad_norm": 15.3125, "learning_rate": 9.766486640136443e-06, "loss": 1.2415, "step": 7072 }, { "epoch": 0.7526069376463077, "grad_norm": 11.953125, "learning_rate": 9.766451108584424e-06, "loss": 1.1116, "step": 7073 }, { "epoch": 0.7527133432645243, "grad_norm": 16.9375, "learning_rate": 9.766415577032406e-06, "loss": 1.1908, "step": 7074 }, { "epoch": 0.7528197488827411, "grad_norm": 16.734375, "learning_rate": 9.766380045480388e-06, "loss": 1.4836, "step": 7075 }, { "epoch": 0.7529261545009577, "grad_norm": 9.5546875, "learning_rate": 9.76634451392837e-06, "loss": 0.9715, "step": 7076 }, { "epoch": 0.7530325601191743, "grad_norm": 12.15625, "learning_rate": 9.766308982376351e-06, "loss": 1.0552, "step": 7077 }, { "epoch": 0.7531389657373909, "grad_norm": 13.5625, "learning_rate": 9.766273450824332e-06, "loss": 1.351, "step": 7078 }, { "epoch": 0.7532453713556075, "grad_norm": 13.1875, "learning_rate": 9.766237919272314e-06, "loss": 1.0587, "step": 7079 }, { "epoch": 0.7533517769738243, "grad_norm": 13.9921875, "learning_rate": 9.766202387720296e-06, "loss": 1.2123, "step": 7080 }, { "epoch": 0.7534581825920409, "grad_norm": 14.6015625, "learning_rate": 9.766166856168279e-06, "loss": 1.4779, "step": 7081 }, { "epoch": 0.7535645882102575, "grad_norm": 13.7265625, "learning_rate": 9.76613132461626e-06, "loss": 0.9759, "step": 7082 }, { "epoch": 0.7536709938284741, "grad_norm": 11.3359375, "learning_rate": 9.766095793064242e-06, "loss": 1.122, "step": 7083 }, { "epoch": 0.7537773994466908, "grad_norm": 12.703125, "learning_rate": 9.766060261512224e-06, "loss": 1.2113, "step": 7084 }, { "epoch": 0.7538838050649074, "grad_norm": 12.5859375, "learning_rate": 9.766024729960205e-06, "loss": 1.413, "step": 7085 }, { "epoch": 0.7539902106831241, "grad_norm": 10.4921875, "learning_rate": 9.765989198408187e-06, "loss": 1.1319, "step": 7086 }, { "epoch": 0.7540966163013407, "grad_norm": 20.75, "learning_rate": 9.765953666856169e-06, "loss": 1.3719, "step": 7087 }, { "epoch": 0.7542030219195573, "grad_norm": 11.0859375, "learning_rate": 9.76591813530415e-06, "loss": 1.3764, "step": 7088 }, { "epoch": 0.754309427537774, "grad_norm": 10.71875, "learning_rate": 9.765882603752132e-06, "loss": 1.149, "step": 7089 }, { "epoch": 0.7544158331559906, "grad_norm": 14.609375, "learning_rate": 9.765847072200113e-06, "loss": 1.3818, "step": 7090 }, { "epoch": 0.7545222387742073, "grad_norm": 13.921875, "learning_rate": 9.765811540648095e-06, "loss": 1.1684, "step": 7091 }, { "epoch": 0.7546286443924239, "grad_norm": 16.28125, "learning_rate": 9.765776009096078e-06, "loss": 1.2392, "step": 7092 }, { "epoch": 0.7547350500106406, "grad_norm": 14.4296875, "learning_rate": 9.76574047754406e-06, "loss": 1.213, "step": 7093 }, { "epoch": 0.7548414556288572, "grad_norm": 13.6015625, "learning_rate": 9.765704945992042e-06, "loss": 1.1798, "step": 7094 }, { "epoch": 0.7549478612470738, "grad_norm": 12.9375, "learning_rate": 9.765669414440023e-06, "loss": 0.9855, "step": 7095 }, { "epoch": 0.7550542668652905, "grad_norm": 12.5078125, "learning_rate": 9.765633882888005e-06, "loss": 1.1596, "step": 7096 }, { "epoch": 0.7551606724835072, "grad_norm": 15.3515625, "learning_rate": 9.765598351335988e-06, "loss": 1.2926, "step": 7097 }, { "epoch": 0.7552670781017238, "grad_norm": 14.421875, "learning_rate": 9.76556281978397e-06, "loss": 1.4715, "step": 7098 }, { "epoch": 0.7553734837199404, "grad_norm": 22.109375, "learning_rate": 9.765527288231951e-06, "loss": 1.197, "step": 7099 }, { "epoch": 0.755479889338157, "grad_norm": 17.609375, "learning_rate": 9.765491756679933e-06, "loss": 1.5167, "step": 7100 }, { "epoch": 0.7555862949563737, "grad_norm": 18.828125, "learning_rate": 9.765456225127915e-06, "loss": 1.5858, "step": 7101 }, { "epoch": 0.7556927005745904, "grad_norm": 8.953125, "learning_rate": 9.765420693575896e-06, "loss": 0.7633, "step": 7102 }, { "epoch": 0.755799106192807, "grad_norm": 13.1171875, "learning_rate": 9.765385162023878e-06, "loss": 1.2562, "step": 7103 }, { "epoch": 0.7559055118110236, "grad_norm": 13.8515625, "learning_rate": 9.76534963047186e-06, "loss": 1.344, "step": 7104 }, { "epoch": 0.7560119174292402, "grad_norm": 14.21875, "learning_rate": 9.765314098919841e-06, "loss": 1.1078, "step": 7105 }, { "epoch": 0.7561183230474569, "grad_norm": 10.3046875, "learning_rate": 9.765278567367823e-06, "loss": 1.1747, "step": 7106 }, { "epoch": 0.7562247286656736, "grad_norm": 16.71875, "learning_rate": 9.765243035815806e-06, "loss": 1.1714, "step": 7107 }, { "epoch": 0.7563311342838902, "grad_norm": 12.015625, "learning_rate": 9.765207504263787e-06, "loss": 1.2621, "step": 7108 }, { "epoch": 0.7564375399021068, "grad_norm": 16.59375, "learning_rate": 9.765171972711769e-06, "loss": 1.0637, "step": 7109 }, { "epoch": 0.7565439455203234, "grad_norm": 12.5546875, "learning_rate": 9.76513644115975e-06, "loss": 1.0753, "step": 7110 }, { "epoch": 0.7566503511385401, "grad_norm": 11.6015625, "learning_rate": 9.765100909607732e-06, "loss": 1.0803, "step": 7111 }, { "epoch": 0.7567567567567568, "grad_norm": 12.734375, "learning_rate": 9.765065378055716e-06, "loss": 1.5136, "step": 7112 }, { "epoch": 0.7568631623749734, "grad_norm": 12.0078125, "learning_rate": 9.765029846503696e-06, "loss": 1.1461, "step": 7113 }, { "epoch": 0.75696956799319, "grad_norm": 16.640625, "learning_rate": 9.764994314951677e-06, "loss": 1.2706, "step": 7114 }, { "epoch": 0.7570759736114067, "grad_norm": 12.46875, "learning_rate": 9.764958783399659e-06, "loss": 1.1006, "step": 7115 }, { "epoch": 0.7571823792296233, "grad_norm": 14.1875, "learning_rate": 9.76492325184764e-06, "loss": 1.2439, "step": 7116 }, { "epoch": 0.75728878484784, "grad_norm": 11.6953125, "learning_rate": 9.764887720295624e-06, "loss": 1.2251, "step": 7117 }, { "epoch": 0.7573951904660566, "grad_norm": 11.671875, "learning_rate": 9.764852188743605e-06, "loss": 1.1211, "step": 7118 }, { "epoch": 0.7575015960842733, "grad_norm": 9.7421875, "learning_rate": 9.764816657191587e-06, "loss": 1.157, "step": 7119 }, { "epoch": 0.7576080017024899, "grad_norm": 11.0703125, "learning_rate": 9.764781125639568e-06, "loss": 1.1607, "step": 7120 }, { "epoch": 0.7577144073207065, "grad_norm": 16.5625, "learning_rate": 9.76474559408755e-06, "loss": 1.5708, "step": 7121 }, { "epoch": 0.7578208129389232, "grad_norm": 10.4296875, "learning_rate": 9.764710062535533e-06, "loss": 1.0583, "step": 7122 }, { "epoch": 0.7579272185571398, "grad_norm": 12.1171875, "learning_rate": 9.764674530983515e-06, "loss": 0.878, "step": 7123 }, { "epoch": 0.7580336241753565, "grad_norm": 23.625, "learning_rate": 9.764638999431497e-06, "loss": 1.4709, "step": 7124 }, { "epoch": 0.7581400297935731, "grad_norm": 10.2578125, "learning_rate": 9.764603467879478e-06, "loss": 1.5869, "step": 7125 }, { "epoch": 0.7582464354117897, "grad_norm": 12.8515625, "learning_rate": 9.76456793632746e-06, "loss": 1.0257, "step": 7126 }, { "epoch": 0.7583528410300063, "grad_norm": 13.328125, "learning_rate": 9.764532404775441e-06, "loss": 1.1399, "step": 7127 }, { "epoch": 0.7584592466482231, "grad_norm": 15.96875, "learning_rate": 9.764496873223423e-06, "loss": 1.0811, "step": 7128 }, { "epoch": 0.7585656522664397, "grad_norm": 14.1171875, "learning_rate": 9.764461341671405e-06, "loss": 1.2679, "step": 7129 }, { "epoch": 0.7586720578846563, "grad_norm": 13.40625, "learning_rate": 9.764425810119386e-06, "loss": 1.5567, "step": 7130 }, { "epoch": 0.7587784635028729, "grad_norm": 11.40625, "learning_rate": 9.764390278567368e-06, "loss": 1.2158, "step": 7131 }, { "epoch": 0.7588848691210895, "grad_norm": 13.9140625, "learning_rate": 9.76435474701535e-06, "loss": 0.9556, "step": 7132 }, { "epoch": 0.7589912747393063, "grad_norm": 32.8125, "learning_rate": 9.764319215463333e-06, "loss": 1.5807, "step": 7133 }, { "epoch": 0.7590976803575229, "grad_norm": 12.578125, "learning_rate": 9.764283683911314e-06, "loss": 0.9827, "step": 7134 }, { "epoch": 0.7592040859757395, "grad_norm": 10.6640625, "learning_rate": 9.764248152359296e-06, "loss": 1.1703, "step": 7135 }, { "epoch": 0.7593104915939561, "grad_norm": 12.125, "learning_rate": 9.764212620807278e-06, "loss": 1.0859, "step": 7136 }, { "epoch": 0.7594168972121728, "grad_norm": 12.28125, "learning_rate": 9.764177089255259e-06, "loss": 1.1461, "step": 7137 }, { "epoch": 0.7595233028303895, "grad_norm": 11.046875, "learning_rate": 9.764141557703242e-06, "loss": 1.1282, "step": 7138 }, { "epoch": 0.7596297084486061, "grad_norm": 9.8671875, "learning_rate": 9.764106026151224e-06, "loss": 1.0302, "step": 7139 }, { "epoch": 0.7597361140668227, "grad_norm": 15.3984375, "learning_rate": 9.764070494599206e-06, "loss": 1.4348, "step": 7140 }, { "epoch": 0.7598425196850394, "grad_norm": 15.34375, "learning_rate": 9.764034963047187e-06, "loss": 1.1685, "step": 7141 }, { "epoch": 0.759948925303256, "grad_norm": 10.78125, "learning_rate": 9.763999431495169e-06, "loss": 1.111, "step": 7142 }, { "epoch": 0.7600553309214727, "grad_norm": 12.7421875, "learning_rate": 9.76396389994315e-06, "loss": 1.2498, "step": 7143 }, { "epoch": 0.7601617365396893, "grad_norm": 11.0234375, "learning_rate": 9.763928368391132e-06, "loss": 0.8989, "step": 7144 }, { "epoch": 0.760268142157906, "grad_norm": 15.046875, "learning_rate": 9.763892836839114e-06, "loss": 1.096, "step": 7145 }, { "epoch": 0.7603745477761226, "grad_norm": 13.546875, "learning_rate": 9.763857305287095e-06, "loss": 1.0026, "step": 7146 }, { "epoch": 0.7604809533943392, "grad_norm": 12.6875, "learning_rate": 9.763821773735077e-06, "loss": 1.0383, "step": 7147 }, { "epoch": 0.7605873590125558, "grad_norm": 12.234375, "learning_rate": 9.76378624218306e-06, "loss": 1.0739, "step": 7148 }, { "epoch": 0.7606937646307725, "grad_norm": 15.6484375, "learning_rate": 9.763750710631042e-06, "loss": 1.4268, "step": 7149 }, { "epoch": 0.7608001702489892, "grad_norm": 13.109375, "learning_rate": 9.763715179079023e-06, "loss": 1.5087, "step": 7150 }, { "epoch": 0.7609065758672058, "grad_norm": 10.3828125, "learning_rate": 9.763679647527005e-06, "loss": 1.0863, "step": 7151 }, { "epoch": 0.7610129814854224, "grad_norm": 13.625, "learning_rate": 9.763644115974987e-06, "loss": 1.1972, "step": 7152 }, { "epoch": 0.761119387103639, "grad_norm": 12.2109375, "learning_rate": 9.763608584422968e-06, "loss": 1.127, "step": 7153 }, { "epoch": 0.7612257927218558, "grad_norm": 12.0078125, "learning_rate": 9.76357305287095e-06, "loss": 1.265, "step": 7154 }, { "epoch": 0.7613321983400724, "grad_norm": 11.671875, "learning_rate": 9.763537521318931e-06, "loss": 1.252, "step": 7155 }, { "epoch": 0.761438603958289, "grad_norm": 11.796875, "learning_rate": 9.763501989766913e-06, "loss": 1.0597, "step": 7156 }, { "epoch": 0.7615450095765056, "grad_norm": 11.65625, "learning_rate": 9.763466458214895e-06, "loss": 1.3547, "step": 7157 }, { "epoch": 0.7616514151947222, "grad_norm": 18.65625, "learning_rate": 9.763430926662878e-06, "loss": 0.9961, "step": 7158 }, { "epoch": 0.761757820812939, "grad_norm": 12.875, "learning_rate": 9.76339539511086e-06, "loss": 1.1905, "step": 7159 }, { "epoch": 0.7618642264311556, "grad_norm": 14.828125, "learning_rate": 9.763359863558841e-06, "loss": 1.0673, "step": 7160 }, { "epoch": 0.7619706320493722, "grad_norm": 11.9296875, "learning_rate": 9.763324332006823e-06, "loss": 1.0064, "step": 7161 }, { "epoch": 0.7620770376675888, "grad_norm": 13.34375, "learning_rate": 9.763288800454804e-06, "loss": 1.5154, "step": 7162 }, { "epoch": 0.7621834432858055, "grad_norm": 18.734375, "learning_rate": 9.763253268902788e-06, "loss": 0.9948, "step": 7163 }, { "epoch": 0.7622898489040222, "grad_norm": 15.3359375, "learning_rate": 9.76321773735077e-06, "loss": 1.2774, "step": 7164 }, { "epoch": 0.7623962545222388, "grad_norm": 11.7890625, "learning_rate": 9.763182205798751e-06, "loss": 1.2654, "step": 7165 }, { "epoch": 0.7625026601404554, "grad_norm": 15.421875, "learning_rate": 9.763146674246733e-06, "loss": 1.0423, "step": 7166 }, { "epoch": 0.762609065758672, "grad_norm": 8.8203125, "learning_rate": 9.763111142694714e-06, "loss": 1.0916, "step": 7167 }, { "epoch": 0.7627154713768887, "grad_norm": 14.921875, "learning_rate": 9.763075611142696e-06, "loss": 1.4143, "step": 7168 }, { "epoch": 0.7628218769951054, "grad_norm": 11.7265625, "learning_rate": 9.763040079590677e-06, "loss": 1.33, "step": 7169 }, { "epoch": 0.762928282613322, "grad_norm": 18.46875, "learning_rate": 9.763004548038659e-06, "loss": 1.1863, "step": 7170 }, { "epoch": 0.7630346882315386, "grad_norm": 14.6640625, "learning_rate": 9.76296901648664e-06, "loss": 1.4174, "step": 7171 }, { "epoch": 0.7631410938497553, "grad_norm": 16.25, "learning_rate": 9.762933484934622e-06, "loss": 1.0718, "step": 7172 }, { "epoch": 0.7632474994679719, "grad_norm": 14.0625, "learning_rate": 9.762897953382604e-06, "loss": 1.019, "step": 7173 }, { "epoch": 0.7633539050861885, "grad_norm": 9.2109375, "learning_rate": 9.762862421830587e-06, "loss": 1.2352, "step": 7174 }, { "epoch": 0.7634603107044052, "grad_norm": 12.21875, "learning_rate": 9.762826890278569e-06, "loss": 1.1107, "step": 7175 }, { "epoch": 0.7635667163226219, "grad_norm": 11.515625, "learning_rate": 9.76279135872655e-06, "loss": 1.2746, "step": 7176 }, { "epoch": 0.7636731219408385, "grad_norm": 13.609375, "learning_rate": 9.762755827174532e-06, "loss": 1.5129, "step": 7177 }, { "epoch": 0.7637795275590551, "grad_norm": 10.7734375, "learning_rate": 9.762720295622513e-06, "loss": 1.2193, "step": 7178 }, { "epoch": 0.7638859331772717, "grad_norm": 15.34375, "learning_rate": 9.762684764070495e-06, "loss": 1.3892, "step": 7179 }, { "epoch": 0.7639923387954884, "grad_norm": 10.875, "learning_rate": 9.762649232518477e-06, "loss": 1.0745, "step": 7180 }, { "epoch": 0.7640987444137051, "grad_norm": 15.2890625, "learning_rate": 9.762613700966458e-06, "loss": 1.0547, "step": 7181 }, { "epoch": 0.7642051500319217, "grad_norm": 12.25, "learning_rate": 9.76257816941444e-06, "loss": 1.3008, "step": 7182 }, { "epoch": 0.7643115556501383, "grad_norm": 11.546875, "learning_rate": 9.762542637862422e-06, "loss": 1.0619, "step": 7183 }, { "epoch": 0.7644179612683549, "grad_norm": 8.90625, "learning_rate": 9.762507106310405e-06, "loss": 1.1965, "step": 7184 }, { "epoch": 0.7645243668865717, "grad_norm": 13.0625, "learning_rate": 9.762471574758386e-06, "loss": 1.2105, "step": 7185 }, { "epoch": 0.7646307725047883, "grad_norm": 10.0234375, "learning_rate": 9.762436043206368e-06, "loss": 1.1798, "step": 7186 }, { "epoch": 0.7647371781230049, "grad_norm": 12.390625, "learning_rate": 9.76240051165435e-06, "loss": 1.1874, "step": 7187 }, { "epoch": 0.7648435837412215, "grad_norm": 9.3515625, "learning_rate": 9.762364980102331e-06, "loss": 1.1386, "step": 7188 }, { "epoch": 0.7649499893594381, "grad_norm": 10.5625, "learning_rate": 9.762329448550315e-06, "loss": 1.1815, "step": 7189 }, { "epoch": 0.7650563949776549, "grad_norm": 15.7578125, "learning_rate": 9.762293916998296e-06, "loss": 1.1071, "step": 7190 }, { "epoch": 0.7651628005958715, "grad_norm": 11.8984375, "learning_rate": 9.762258385446278e-06, "loss": 1.025, "step": 7191 }, { "epoch": 0.7652692062140881, "grad_norm": 12.375, "learning_rate": 9.76222285389426e-06, "loss": 0.9128, "step": 7192 }, { "epoch": 0.7653756118323047, "grad_norm": 15.5703125, "learning_rate": 9.762187322342241e-06, "loss": 0.8552, "step": 7193 }, { "epoch": 0.7654820174505214, "grad_norm": 15.1796875, "learning_rate": 9.762151790790223e-06, "loss": 1.4684, "step": 7194 }, { "epoch": 0.765588423068738, "grad_norm": 12.0, "learning_rate": 9.762116259238204e-06, "loss": 1.0486, "step": 7195 }, { "epoch": 0.7656948286869547, "grad_norm": 15.3515625, "learning_rate": 9.762080727686186e-06, "loss": 1.1818, "step": 7196 }, { "epoch": 0.7658012343051713, "grad_norm": 13.6171875, "learning_rate": 9.762045196134167e-06, "loss": 1.2311, "step": 7197 }, { "epoch": 0.765907639923388, "grad_norm": 11.8359375, "learning_rate": 9.762009664582149e-06, "loss": 1.3511, "step": 7198 }, { "epoch": 0.7660140455416046, "grad_norm": 22.65625, "learning_rate": 9.761974133030132e-06, "loss": 1.8517, "step": 7199 }, { "epoch": 0.7661204511598212, "grad_norm": 12.953125, "learning_rate": 9.761938601478114e-06, "loss": 1.34, "step": 7200 }, { "epoch": 0.7662268567780379, "grad_norm": 11.609375, "learning_rate": 9.761903069926096e-06, "loss": 1.208, "step": 7201 }, { "epoch": 0.7663332623962545, "grad_norm": 16.265625, "learning_rate": 9.761867538374077e-06, "loss": 1.4006, "step": 7202 }, { "epoch": 0.7664396680144712, "grad_norm": 10.6171875, "learning_rate": 9.761832006822059e-06, "loss": 1.3891, "step": 7203 }, { "epoch": 0.7665460736326878, "grad_norm": 11.4921875, "learning_rate": 9.76179647527004e-06, "loss": 1.1236, "step": 7204 }, { "epoch": 0.7666524792509044, "grad_norm": 15.8984375, "learning_rate": 9.761760943718022e-06, "loss": 1.0475, "step": 7205 }, { "epoch": 0.7667588848691211, "grad_norm": 15.328125, "learning_rate": 9.761725412166004e-06, "loss": 1.5483, "step": 7206 }, { "epoch": 0.7668652904873378, "grad_norm": 12.625, "learning_rate": 9.761689880613985e-06, "loss": 1.1473, "step": 7207 }, { "epoch": 0.7669716961055544, "grad_norm": 16.3125, "learning_rate": 9.761654349061967e-06, "loss": 1.1928, "step": 7208 }, { "epoch": 0.767078101723771, "grad_norm": 13.1875, "learning_rate": 9.761618817509948e-06, "loss": 1.1728, "step": 7209 }, { "epoch": 0.7671845073419876, "grad_norm": 12.078125, "learning_rate": 9.761583285957932e-06, "loss": 0.9874, "step": 7210 }, { "epoch": 0.7672909129602044, "grad_norm": 14.7421875, "learning_rate": 9.761547754405913e-06, "loss": 1.182, "step": 7211 }, { "epoch": 0.767397318578421, "grad_norm": 10.4921875, "learning_rate": 9.761512222853895e-06, "loss": 1.2417, "step": 7212 }, { "epoch": 0.7675037241966376, "grad_norm": 14.4921875, "learning_rate": 9.761476691301876e-06, "loss": 1.5993, "step": 7213 }, { "epoch": 0.7676101298148542, "grad_norm": 12.8515625, "learning_rate": 9.761441159749858e-06, "loss": 1.4765, "step": 7214 }, { "epoch": 0.7677165354330708, "grad_norm": 12.15625, "learning_rate": 9.761405628197841e-06, "loss": 1.1167, "step": 7215 }, { "epoch": 0.7678229410512875, "grad_norm": 11.4140625, "learning_rate": 9.761370096645823e-06, "loss": 0.9479, "step": 7216 }, { "epoch": 0.7679293466695042, "grad_norm": 12.0234375, "learning_rate": 9.761334565093805e-06, "loss": 1.2627, "step": 7217 }, { "epoch": 0.7680357522877208, "grad_norm": 21.578125, "learning_rate": 9.761299033541786e-06, "loss": 1.0367, "step": 7218 }, { "epoch": 0.7681421579059374, "grad_norm": 9.9609375, "learning_rate": 9.761263501989768e-06, "loss": 0.857, "step": 7219 }, { "epoch": 0.768248563524154, "grad_norm": 10.953125, "learning_rate": 9.76122797043775e-06, "loss": 1.059, "step": 7220 }, { "epoch": 0.7683549691423707, "grad_norm": 11.9453125, "learning_rate": 9.761192438885731e-06, "loss": 1.1729, "step": 7221 }, { "epoch": 0.7684613747605874, "grad_norm": 12.921875, "learning_rate": 9.761156907333713e-06, "loss": 1.0648, "step": 7222 }, { "epoch": 0.768567780378804, "grad_norm": 13.890625, "learning_rate": 9.761121375781694e-06, "loss": 1.6084, "step": 7223 }, { "epoch": 0.7686741859970206, "grad_norm": 11.6953125, "learning_rate": 9.761085844229676e-06, "loss": 1.0494, "step": 7224 }, { "epoch": 0.7687805916152373, "grad_norm": 15.5859375, "learning_rate": 9.76105031267766e-06, "loss": 1.0825, "step": 7225 }, { "epoch": 0.7688869972334539, "grad_norm": 14.921875, "learning_rate": 9.76101478112564e-06, "loss": 1.2222, "step": 7226 }, { "epoch": 0.7689934028516706, "grad_norm": 13.4765625, "learning_rate": 9.760979249573622e-06, "loss": 1.422, "step": 7227 }, { "epoch": 0.7690998084698872, "grad_norm": 8.328125, "learning_rate": 9.760943718021604e-06, "loss": 0.6846, "step": 7228 }, { "epoch": 0.7692062140881039, "grad_norm": 14.7578125, "learning_rate": 9.760908186469586e-06, "loss": 1.3537, "step": 7229 }, { "epoch": 0.7693126197063205, "grad_norm": 13.8515625, "learning_rate": 9.760872654917569e-06, "loss": 0.9433, "step": 7230 }, { "epoch": 0.7694190253245371, "grad_norm": 10.953125, "learning_rate": 9.76083712336555e-06, "loss": 1.3156, "step": 7231 }, { "epoch": 0.7695254309427538, "grad_norm": 13.2578125, "learning_rate": 9.760801591813532e-06, "loss": 1.27, "step": 7232 }, { "epoch": 0.7696318365609705, "grad_norm": 10.03125, "learning_rate": 9.760766060261512e-06, "loss": 1.2107, "step": 7233 }, { "epoch": 0.7697382421791871, "grad_norm": 12.9765625, "learning_rate": 9.760730528709494e-06, "loss": 1.1386, "step": 7234 }, { "epoch": 0.7698446477974037, "grad_norm": 11.2578125, "learning_rate": 9.760694997157477e-06, "loss": 1.0552, "step": 7235 }, { "epoch": 0.7699510534156203, "grad_norm": 12.140625, "learning_rate": 9.760659465605459e-06, "loss": 1.1277, "step": 7236 }, { "epoch": 0.7700574590338369, "grad_norm": 12.3828125, "learning_rate": 9.76062393405344e-06, "loss": 1.0189, "step": 7237 }, { "epoch": 0.7701638646520537, "grad_norm": 12.5390625, "learning_rate": 9.760588402501422e-06, "loss": 1.1723, "step": 7238 }, { "epoch": 0.7702702702702703, "grad_norm": 10.65625, "learning_rate": 9.760552870949403e-06, "loss": 1.1607, "step": 7239 }, { "epoch": 0.7703766758884869, "grad_norm": 15.6484375, "learning_rate": 9.760517339397387e-06, "loss": 1.1513, "step": 7240 }, { "epoch": 0.7704830815067035, "grad_norm": 19.21875, "learning_rate": 9.760481807845368e-06, "loss": 1.3321, "step": 7241 }, { "epoch": 0.7705894871249201, "grad_norm": 14.3359375, "learning_rate": 9.76044627629335e-06, "loss": 1.7282, "step": 7242 }, { "epoch": 0.7706958927431369, "grad_norm": 11.59375, "learning_rate": 9.760410744741331e-06, "loss": 0.9089, "step": 7243 }, { "epoch": 0.7708022983613535, "grad_norm": 11.1328125, "learning_rate": 9.760375213189313e-06, "loss": 0.8516, "step": 7244 }, { "epoch": 0.7709087039795701, "grad_norm": 12.5, "learning_rate": 9.760339681637295e-06, "loss": 1.1818, "step": 7245 }, { "epoch": 0.7710151095977867, "grad_norm": 12.96875, "learning_rate": 9.760304150085276e-06, "loss": 1.4052, "step": 7246 }, { "epoch": 0.7711215152160034, "grad_norm": 13.8359375, "learning_rate": 9.760268618533258e-06, "loss": 1.4425, "step": 7247 }, { "epoch": 0.7712279208342201, "grad_norm": 12.8125, "learning_rate": 9.76023308698124e-06, "loss": 1.1417, "step": 7248 }, { "epoch": 0.7713343264524367, "grad_norm": 11.5703125, "learning_rate": 9.760197555429221e-06, "loss": 1.1358, "step": 7249 }, { "epoch": 0.7714407320706533, "grad_norm": 9.7734375, "learning_rate": 9.760162023877203e-06, "loss": 1.1792, "step": 7250 }, { "epoch": 0.77154713768887, "grad_norm": 10.9140625, "learning_rate": 9.760126492325186e-06, "loss": 1.2141, "step": 7251 }, { "epoch": 0.7716535433070866, "grad_norm": 12.5078125, "learning_rate": 9.760090960773168e-06, "loss": 1.1503, "step": 7252 }, { "epoch": 0.7717599489253033, "grad_norm": 18.328125, "learning_rate": 9.76005542922115e-06, "loss": 1.1853, "step": 7253 }, { "epoch": 0.7718663545435199, "grad_norm": 11.6796875, "learning_rate": 9.76001989766913e-06, "loss": 0.9112, "step": 7254 }, { "epoch": 0.7719727601617365, "grad_norm": 11.09375, "learning_rate": 9.759984366117112e-06, "loss": 1.0797, "step": 7255 }, { "epoch": 0.7720791657799532, "grad_norm": 10.25, "learning_rate": 9.759948834565096e-06, "loss": 1.0335, "step": 7256 }, { "epoch": 0.7721855713981698, "grad_norm": 13.28125, "learning_rate": 9.759913303013077e-06, "loss": 1.3146, "step": 7257 }, { "epoch": 0.7722919770163864, "grad_norm": 11.765625, "learning_rate": 9.759877771461059e-06, "loss": 1.422, "step": 7258 }, { "epoch": 0.7723983826346031, "grad_norm": 13.7265625, "learning_rate": 9.75984223990904e-06, "loss": 1.1284, "step": 7259 }, { "epoch": 0.7725047882528198, "grad_norm": 15.6328125, "learning_rate": 9.759806708357022e-06, "loss": 1.0895, "step": 7260 }, { "epoch": 0.7726111938710364, "grad_norm": 20.203125, "learning_rate": 9.759771176805004e-06, "loss": 1.064, "step": 7261 }, { "epoch": 0.772717599489253, "grad_norm": 14.1484375, "learning_rate": 9.759735645252985e-06, "loss": 1.1303, "step": 7262 }, { "epoch": 0.7728240051074696, "grad_norm": 13.25, "learning_rate": 9.759700113700967e-06, "loss": 1.2778, "step": 7263 }, { "epoch": 0.7729304107256864, "grad_norm": 17.515625, "learning_rate": 9.759664582148949e-06, "loss": 1.3309, "step": 7264 }, { "epoch": 0.773036816343903, "grad_norm": 10.265625, "learning_rate": 9.75962905059693e-06, "loss": 1.0764, "step": 7265 }, { "epoch": 0.7731432219621196, "grad_norm": 11.2890625, "learning_rate": 9.759593519044914e-06, "loss": 1.0232, "step": 7266 }, { "epoch": 0.7732496275803362, "grad_norm": 11.015625, "learning_rate": 9.759557987492895e-06, "loss": 1.1645, "step": 7267 }, { "epoch": 0.7733560331985528, "grad_norm": 13.4375, "learning_rate": 9.759522455940877e-06, "loss": 1.168, "step": 7268 }, { "epoch": 0.7734624388167696, "grad_norm": 13.8671875, "learning_rate": 9.759486924388858e-06, "loss": 1.5323, "step": 7269 }, { "epoch": 0.7735688444349862, "grad_norm": 15.0703125, "learning_rate": 9.75945139283684e-06, "loss": 1.0334, "step": 7270 }, { "epoch": 0.7736752500532028, "grad_norm": 9.171875, "learning_rate": 9.759415861284822e-06, "loss": 1.1186, "step": 7271 }, { "epoch": 0.7737816556714194, "grad_norm": 16.03125, "learning_rate": 9.759380329732803e-06, "loss": 1.1645, "step": 7272 }, { "epoch": 0.773888061289636, "grad_norm": 10.6171875, "learning_rate": 9.759344798180785e-06, "loss": 1.3148, "step": 7273 }, { "epoch": 0.7739944669078528, "grad_norm": 14.0390625, "learning_rate": 9.759309266628766e-06, "loss": 1.3541, "step": 7274 }, { "epoch": 0.7741008725260694, "grad_norm": 12.40625, "learning_rate": 9.759273735076748e-06, "loss": 1.1486, "step": 7275 }, { "epoch": 0.774207278144286, "grad_norm": 11.71875, "learning_rate": 9.759238203524731e-06, "loss": 1.5341, "step": 7276 }, { "epoch": 0.7743136837625026, "grad_norm": 10.484375, "learning_rate": 9.759202671972713e-06, "loss": 1.2173, "step": 7277 }, { "epoch": 0.7744200893807193, "grad_norm": 11.78125, "learning_rate": 9.759167140420694e-06, "loss": 1.266, "step": 7278 }, { "epoch": 0.774526494998936, "grad_norm": 21.671875, "learning_rate": 9.759131608868676e-06, "loss": 1.3438, "step": 7279 }, { "epoch": 0.7746329006171526, "grad_norm": 11.953125, "learning_rate": 9.759096077316658e-06, "loss": 1.2047, "step": 7280 }, { "epoch": 0.7747393062353692, "grad_norm": 14.28125, "learning_rate": 9.759060545764641e-06, "loss": 1.2035, "step": 7281 }, { "epoch": 0.7748457118535859, "grad_norm": 15.8046875, "learning_rate": 9.759025014212623e-06, "loss": 1.0437, "step": 7282 }, { "epoch": 0.7749521174718025, "grad_norm": 12.203125, "learning_rate": 9.758989482660604e-06, "loss": 1.1519, "step": 7283 }, { "epoch": 0.7750585230900191, "grad_norm": 13.5078125, "learning_rate": 9.758953951108586e-06, "loss": 1.155, "step": 7284 }, { "epoch": 0.7751649287082358, "grad_norm": 12.5234375, "learning_rate": 9.758918419556567e-06, "loss": 1.2696, "step": 7285 }, { "epoch": 0.7752713343264525, "grad_norm": 15.6484375, "learning_rate": 9.758882888004549e-06, "loss": 1.0336, "step": 7286 }, { "epoch": 0.7753777399446691, "grad_norm": 12.921875, "learning_rate": 9.75884735645253e-06, "loss": 1.1974, "step": 7287 }, { "epoch": 0.7754841455628857, "grad_norm": 13.0078125, "learning_rate": 9.758811824900512e-06, "loss": 1.2447, "step": 7288 }, { "epoch": 0.7755905511811023, "grad_norm": 12.0546875, "learning_rate": 9.758776293348494e-06, "loss": 1.3233, "step": 7289 }, { "epoch": 0.775696956799319, "grad_norm": 12.6484375, "learning_rate": 9.758740761796475e-06, "loss": 1.1913, "step": 7290 }, { "epoch": 0.7758033624175357, "grad_norm": 11.9765625, "learning_rate": 9.758705230244457e-06, "loss": 1.0259, "step": 7291 }, { "epoch": 0.7759097680357523, "grad_norm": 13.75, "learning_rate": 9.75866969869244e-06, "loss": 1.3165, "step": 7292 }, { "epoch": 0.7760161736539689, "grad_norm": 11.515625, "learning_rate": 9.758634167140422e-06, "loss": 1.3193, "step": 7293 }, { "epoch": 0.7761225792721855, "grad_norm": 12.2890625, "learning_rate": 9.758598635588404e-06, "loss": 1.21, "step": 7294 }, { "epoch": 0.7762289848904023, "grad_norm": 13.0, "learning_rate": 9.758563104036385e-06, "loss": 0.8824, "step": 7295 }, { "epoch": 0.7763353905086189, "grad_norm": 13.5859375, "learning_rate": 9.758527572484367e-06, "loss": 1.1704, "step": 7296 }, { "epoch": 0.7764417961268355, "grad_norm": 12.7421875, "learning_rate": 9.758492040932348e-06, "loss": 0.7553, "step": 7297 }, { "epoch": 0.7765482017450521, "grad_norm": 15.1484375, "learning_rate": 9.75845650938033e-06, "loss": 1.2681, "step": 7298 }, { "epoch": 0.7766546073632687, "grad_norm": 17.875, "learning_rate": 9.758420977828312e-06, "loss": 1.212, "step": 7299 }, { "epoch": 0.7767610129814855, "grad_norm": 13.1171875, "learning_rate": 9.758385446276293e-06, "loss": 1.1006, "step": 7300 }, { "epoch": 0.7768674185997021, "grad_norm": 12.609375, "learning_rate": 9.758349914724275e-06, "loss": 1.2006, "step": 7301 }, { "epoch": 0.7769738242179187, "grad_norm": 18.0, "learning_rate": 9.758314383172258e-06, "loss": 0.9541, "step": 7302 }, { "epoch": 0.7770802298361353, "grad_norm": 12.890625, "learning_rate": 9.75827885162024e-06, "loss": 1.1876, "step": 7303 }, { "epoch": 0.777186635454352, "grad_norm": 14.375, "learning_rate": 9.758243320068221e-06, "loss": 1.1043, "step": 7304 }, { "epoch": 0.7772930410725686, "grad_norm": 25.328125, "learning_rate": 9.758207788516203e-06, "loss": 1.3137, "step": 7305 }, { "epoch": 0.7773994466907853, "grad_norm": 14.0625, "learning_rate": 9.758172256964185e-06, "loss": 0.9845, "step": 7306 }, { "epoch": 0.7775058523090019, "grad_norm": 13.8125, "learning_rate": 9.758136725412168e-06, "loss": 1.3259, "step": 7307 }, { "epoch": 0.7776122579272186, "grad_norm": 17.171875, "learning_rate": 9.75810119386015e-06, "loss": 1.14, "step": 7308 }, { "epoch": 0.7777186635454352, "grad_norm": 14.5625, "learning_rate": 9.758065662308131e-06, "loss": 1.1277, "step": 7309 }, { "epoch": 0.7778250691636518, "grad_norm": 11.8515625, "learning_rate": 9.758030130756113e-06, "loss": 1.0877, "step": 7310 }, { "epoch": 0.7779314747818685, "grad_norm": 12.59375, "learning_rate": 9.757994599204094e-06, "loss": 1.1718, "step": 7311 }, { "epoch": 0.7780378804000851, "grad_norm": 15.8671875, "learning_rate": 9.757959067652076e-06, "loss": 1.1989, "step": 7312 }, { "epoch": 0.7781442860183018, "grad_norm": 15.71875, "learning_rate": 9.757923536100057e-06, "loss": 1.3084, "step": 7313 }, { "epoch": 0.7782506916365184, "grad_norm": 12.3984375, "learning_rate": 9.757888004548039e-06, "loss": 1.1891, "step": 7314 }, { "epoch": 0.778357097254735, "grad_norm": 15.9921875, "learning_rate": 9.75785247299602e-06, "loss": 1.3143, "step": 7315 }, { "epoch": 0.7784635028729517, "grad_norm": 9.46875, "learning_rate": 9.757816941444002e-06, "loss": 0.6975, "step": 7316 }, { "epoch": 0.7785699084911684, "grad_norm": 16.390625, "learning_rate": 9.757781409891986e-06, "loss": 1.1962, "step": 7317 }, { "epoch": 0.778676314109385, "grad_norm": 10.671875, "learning_rate": 9.757745878339967e-06, "loss": 1.2562, "step": 7318 }, { "epoch": 0.7787827197276016, "grad_norm": 11.15625, "learning_rate": 9.757710346787949e-06, "loss": 1.2656, "step": 7319 }, { "epoch": 0.7788891253458182, "grad_norm": 14.2109375, "learning_rate": 9.75767481523593e-06, "loss": 1.2128, "step": 7320 }, { "epoch": 0.778995530964035, "grad_norm": 20.421875, "learning_rate": 9.757639283683912e-06, "loss": 1.0156, "step": 7321 }, { "epoch": 0.7791019365822516, "grad_norm": 16.234375, "learning_rate": 9.757603752131894e-06, "loss": 1.2382, "step": 7322 }, { "epoch": 0.7792083422004682, "grad_norm": 14.1015625, "learning_rate": 9.757568220579875e-06, "loss": 1.3083, "step": 7323 }, { "epoch": 0.7793147478186848, "grad_norm": 23.0625, "learning_rate": 9.757532689027857e-06, "loss": 1.7728, "step": 7324 }, { "epoch": 0.7794211534369014, "grad_norm": 10.2734375, "learning_rate": 9.757497157475838e-06, "loss": 1.1712, "step": 7325 }, { "epoch": 0.7795275590551181, "grad_norm": 109.9375, "learning_rate": 9.75746162592382e-06, "loss": 1.3954, "step": 7326 }, { "epoch": 0.7796339646733348, "grad_norm": 22.578125, "learning_rate": 9.757426094371802e-06, "loss": 1.109, "step": 7327 }, { "epoch": 0.7797403702915514, "grad_norm": 16.359375, "learning_rate": 9.757390562819785e-06, "loss": 1.4115, "step": 7328 }, { "epoch": 0.779846775909768, "grad_norm": 13.578125, "learning_rate": 9.757355031267767e-06, "loss": 1.2733, "step": 7329 }, { "epoch": 0.7799531815279847, "grad_norm": 12.78125, "learning_rate": 9.757319499715748e-06, "loss": 1.2515, "step": 7330 }, { "epoch": 0.7800595871462013, "grad_norm": 11.578125, "learning_rate": 9.75728396816373e-06, "loss": 1.2524, "step": 7331 }, { "epoch": 0.780165992764418, "grad_norm": 13.3203125, "learning_rate": 9.757248436611711e-06, "loss": 1.4304, "step": 7332 }, { "epoch": 0.7802723983826346, "grad_norm": 10.9296875, "learning_rate": 9.757212905059695e-06, "loss": 1.1758, "step": 7333 }, { "epoch": 0.7803788040008512, "grad_norm": 11.25, "learning_rate": 9.757177373507676e-06, "loss": 1.3089, "step": 7334 }, { "epoch": 0.7804852096190679, "grad_norm": 13.3515625, "learning_rate": 9.757141841955658e-06, "loss": 1.3105, "step": 7335 }, { "epoch": 0.7805916152372845, "grad_norm": 176.625, "learning_rate": 9.75710631040364e-06, "loss": 5.8791, "step": 7336 }, { "epoch": 0.7806980208555012, "grad_norm": 12.4140625, "learning_rate": 9.757070778851621e-06, "loss": 1.3528, "step": 7337 }, { "epoch": 0.7808044264737178, "grad_norm": 20.09375, "learning_rate": 9.757035247299603e-06, "loss": 1.3838, "step": 7338 }, { "epoch": 0.7809108320919345, "grad_norm": 10.234375, "learning_rate": 9.756999715747584e-06, "loss": 1.4166, "step": 7339 }, { "epoch": 0.7810172377101511, "grad_norm": 11.3125, "learning_rate": 9.756964184195566e-06, "loss": 1.3411, "step": 7340 }, { "epoch": 0.7811236433283677, "grad_norm": 9.953125, "learning_rate": 9.756928652643548e-06, "loss": 1.2477, "step": 7341 }, { "epoch": 0.7812300489465844, "grad_norm": 9.6640625, "learning_rate": 9.75689312109153e-06, "loss": 1.0387, "step": 7342 }, { "epoch": 0.781336454564801, "grad_norm": 12.96875, "learning_rate": 9.756857589539512e-06, "loss": 1.3251, "step": 7343 }, { "epoch": 0.7814428601830177, "grad_norm": 19.640625, "learning_rate": 9.756822057987494e-06, "loss": 1.4111, "step": 7344 }, { "epoch": 0.7815492658012343, "grad_norm": 12.359375, "learning_rate": 9.756786526435476e-06, "loss": 1.5139, "step": 7345 }, { "epoch": 0.7816556714194509, "grad_norm": 11.5, "learning_rate": 9.756750994883457e-06, "loss": 1.3534, "step": 7346 }, { "epoch": 0.7817620770376675, "grad_norm": 13.0703125, "learning_rate": 9.756715463331439e-06, "loss": 1.0852, "step": 7347 }, { "epoch": 0.7818684826558843, "grad_norm": 8.578125, "learning_rate": 9.756679931779422e-06, "loss": 1.1222, "step": 7348 }, { "epoch": 0.7819748882741009, "grad_norm": 11.5703125, "learning_rate": 9.756644400227404e-06, "loss": 1.1441, "step": 7349 }, { "epoch": 0.7820812938923175, "grad_norm": 15.234375, "learning_rate": 9.756608868675385e-06, "loss": 1.7306, "step": 7350 }, { "epoch": 0.7821876995105341, "grad_norm": 10.4375, "learning_rate": 9.756573337123367e-06, "loss": 1.1026, "step": 7351 }, { "epoch": 0.7822941051287507, "grad_norm": 11.375, "learning_rate": 9.756537805571347e-06, "loss": 1.207, "step": 7352 }, { "epoch": 0.7824005107469675, "grad_norm": 12.203125, "learning_rate": 9.75650227401933e-06, "loss": 1.066, "step": 7353 }, { "epoch": 0.7825069163651841, "grad_norm": 9.015625, "learning_rate": 9.756466742467312e-06, "loss": 1.1115, "step": 7354 }, { "epoch": 0.7826133219834007, "grad_norm": 15.734375, "learning_rate": 9.756431210915293e-06, "loss": 1.816, "step": 7355 }, { "epoch": 0.7827197276016173, "grad_norm": 15.609375, "learning_rate": 9.756395679363275e-06, "loss": 1.5905, "step": 7356 }, { "epoch": 0.782826133219834, "grad_norm": 11.4453125, "learning_rate": 9.756360147811257e-06, "loss": 1.2041, "step": 7357 }, { "epoch": 0.7829325388380507, "grad_norm": 15.953125, "learning_rate": 9.75632461625924e-06, "loss": 1.6074, "step": 7358 }, { "epoch": 0.7830389444562673, "grad_norm": 12.453125, "learning_rate": 9.756289084707222e-06, "loss": 1.1145, "step": 7359 }, { "epoch": 0.7831453500744839, "grad_norm": 9.6015625, "learning_rate": 9.756253553155203e-06, "loss": 1.264, "step": 7360 }, { "epoch": 0.7832517556927006, "grad_norm": 14.453125, "learning_rate": 9.756218021603185e-06, "loss": 0.9922, "step": 7361 }, { "epoch": 0.7833581613109172, "grad_norm": 10.359375, "learning_rate": 9.756182490051166e-06, "loss": 1.2305, "step": 7362 }, { "epoch": 0.7834645669291339, "grad_norm": 16.0, "learning_rate": 9.756146958499148e-06, "loss": 1.1457, "step": 7363 }, { "epoch": 0.7835709725473505, "grad_norm": 14.015625, "learning_rate": 9.75611142694713e-06, "loss": 1.373, "step": 7364 }, { "epoch": 0.7836773781655672, "grad_norm": 9.1875, "learning_rate": 9.756075895395111e-06, "loss": 1.2292, "step": 7365 }, { "epoch": 0.7837837837837838, "grad_norm": 11.703125, "learning_rate": 9.756040363843093e-06, "loss": 1.4311, "step": 7366 }, { "epoch": 0.7838901894020004, "grad_norm": 9.2265625, "learning_rate": 9.756004832291074e-06, "loss": 1.0686, "step": 7367 }, { "epoch": 0.783996595020217, "grad_norm": 11.8828125, "learning_rate": 9.755969300739056e-06, "loss": 1.4134, "step": 7368 }, { "epoch": 0.7841030006384337, "grad_norm": 14.171875, "learning_rate": 9.75593376918704e-06, "loss": 1.1295, "step": 7369 }, { "epoch": 0.7842094062566504, "grad_norm": 13.1015625, "learning_rate": 9.755898237635021e-06, "loss": 1.5693, "step": 7370 }, { "epoch": 0.784315811874867, "grad_norm": 10.421875, "learning_rate": 9.755862706083003e-06, "loss": 1.0736, "step": 7371 }, { "epoch": 0.7844222174930836, "grad_norm": 14.5625, "learning_rate": 9.755827174530984e-06, "loss": 1.05, "step": 7372 }, { "epoch": 0.7845286231113002, "grad_norm": 12.7734375, "learning_rate": 9.755791642978966e-06, "loss": 1.4492, "step": 7373 }, { "epoch": 0.784635028729517, "grad_norm": 17.984375, "learning_rate": 9.755756111426949e-06, "loss": 1.2695, "step": 7374 }, { "epoch": 0.7847414343477336, "grad_norm": 13.0859375, "learning_rate": 9.75572057987493e-06, "loss": 1.3251, "step": 7375 }, { "epoch": 0.7848478399659502, "grad_norm": 13.0390625, "learning_rate": 9.755685048322912e-06, "loss": 1.4608, "step": 7376 }, { "epoch": 0.7849542455841668, "grad_norm": 20.21875, "learning_rate": 9.755649516770894e-06, "loss": 1.3093, "step": 7377 }, { "epoch": 0.7850606512023834, "grad_norm": 9.109375, "learning_rate": 9.755613985218875e-06, "loss": 1.0172, "step": 7378 }, { "epoch": 0.7851670568206002, "grad_norm": 12.3203125, "learning_rate": 9.755578453666857e-06, "loss": 1.3247, "step": 7379 }, { "epoch": 0.7852734624388168, "grad_norm": 15.7734375, "learning_rate": 9.755542922114839e-06, "loss": 1.1107, "step": 7380 }, { "epoch": 0.7853798680570334, "grad_norm": 13.1484375, "learning_rate": 9.75550739056282e-06, "loss": 1.6622, "step": 7381 }, { "epoch": 0.78548627367525, "grad_norm": 10.171875, "learning_rate": 9.755471859010802e-06, "loss": 1.07, "step": 7382 }, { "epoch": 0.7855926792934667, "grad_norm": 11.09375, "learning_rate": 9.755436327458784e-06, "loss": 1.2736, "step": 7383 }, { "epoch": 0.7856990849116834, "grad_norm": 9.65625, "learning_rate": 9.755400795906767e-06, "loss": 1.0585, "step": 7384 }, { "epoch": 0.7858054905299, "grad_norm": 10.9921875, "learning_rate": 9.755365264354748e-06, "loss": 0.8371, "step": 7385 }, { "epoch": 0.7859118961481166, "grad_norm": 11.4453125, "learning_rate": 9.75532973280273e-06, "loss": 1.7198, "step": 7386 }, { "epoch": 0.7860183017663332, "grad_norm": 16.0, "learning_rate": 9.755294201250712e-06, "loss": 1.1689, "step": 7387 }, { "epoch": 0.7861247073845499, "grad_norm": 11.765625, "learning_rate": 9.755258669698693e-06, "loss": 1.2202, "step": 7388 }, { "epoch": 0.7862311130027665, "grad_norm": 10.2265625, "learning_rate": 9.755223138146675e-06, "loss": 1.0457, "step": 7389 }, { "epoch": 0.7863375186209832, "grad_norm": 11.7265625, "learning_rate": 9.755187606594656e-06, "loss": 1.3523, "step": 7390 }, { "epoch": 0.7864439242391998, "grad_norm": 15.3359375, "learning_rate": 9.755152075042638e-06, "loss": 1.7893, "step": 7391 }, { "epoch": 0.7865503298574165, "grad_norm": 14.2109375, "learning_rate": 9.75511654349062e-06, "loss": 1.5318, "step": 7392 }, { "epoch": 0.7866567354756331, "grad_norm": 10.8828125, "learning_rate": 9.755081011938601e-06, "loss": 1.0572, "step": 7393 }, { "epoch": 0.7867631410938497, "grad_norm": 8.890625, "learning_rate": 9.755045480386585e-06, "loss": 1.2102, "step": 7394 }, { "epoch": 0.7868695467120664, "grad_norm": 12.84375, "learning_rate": 9.755009948834566e-06, "loss": 1.0581, "step": 7395 }, { "epoch": 0.7869759523302831, "grad_norm": 10.1015625, "learning_rate": 9.754974417282548e-06, "loss": 1.4312, "step": 7396 }, { "epoch": 0.7870823579484997, "grad_norm": 16.390625, "learning_rate": 9.75493888573053e-06, "loss": 1.3443, "step": 7397 }, { "epoch": 0.7871887635667163, "grad_norm": 10.21875, "learning_rate": 9.754903354178511e-06, "loss": 0.9158, "step": 7398 }, { "epoch": 0.7872951691849329, "grad_norm": 12.3984375, "learning_rate": 9.754867822626494e-06, "loss": 1.3313, "step": 7399 }, { "epoch": 0.7874015748031497, "grad_norm": 12.640625, "learning_rate": 9.754832291074476e-06, "loss": 1.0894, "step": 7400 }, { "epoch": 0.7875079804213663, "grad_norm": 14.046875, "learning_rate": 9.754796759522458e-06, "loss": 1.1209, "step": 7401 }, { "epoch": 0.7876143860395829, "grad_norm": 11.7109375, "learning_rate": 9.754761227970439e-06, "loss": 1.1, "step": 7402 }, { "epoch": 0.7877207916577995, "grad_norm": 21.25, "learning_rate": 9.75472569641842e-06, "loss": 1.2411, "step": 7403 }, { "epoch": 0.7878271972760161, "grad_norm": 14.7265625, "learning_rate": 9.754690164866402e-06, "loss": 1.0975, "step": 7404 }, { "epoch": 0.7879336028942329, "grad_norm": 12.15625, "learning_rate": 9.754654633314384e-06, "loss": 1.3049, "step": 7405 }, { "epoch": 0.7880400085124495, "grad_norm": 12.5625, "learning_rate": 9.754619101762366e-06, "loss": 1.1898, "step": 7406 }, { "epoch": 0.7881464141306661, "grad_norm": 10.3828125, "learning_rate": 9.754583570210347e-06, "loss": 1.2403, "step": 7407 }, { "epoch": 0.7882528197488827, "grad_norm": 13.5625, "learning_rate": 9.754548038658329e-06, "loss": 1.2119, "step": 7408 }, { "epoch": 0.7883592253670993, "grad_norm": 13.6953125, "learning_rate": 9.75451250710631e-06, "loss": 1.0031, "step": 7409 }, { "epoch": 0.7884656309853161, "grad_norm": 13.9375, "learning_rate": 9.754476975554294e-06, "loss": 1.4834, "step": 7410 }, { "epoch": 0.7885720366035327, "grad_norm": 17.703125, "learning_rate": 9.754441444002275e-06, "loss": 1.0643, "step": 7411 }, { "epoch": 0.7886784422217493, "grad_norm": 12.8984375, "learning_rate": 9.754405912450257e-06, "loss": 1.4418, "step": 7412 }, { "epoch": 0.7887848478399659, "grad_norm": 10.5546875, "learning_rate": 9.754370380898238e-06, "loss": 0.9259, "step": 7413 }, { "epoch": 0.7888912534581826, "grad_norm": 10.5390625, "learning_rate": 9.75433484934622e-06, "loss": 1.2368, "step": 7414 }, { "epoch": 0.7889976590763992, "grad_norm": 13.4453125, "learning_rate": 9.754299317794202e-06, "loss": 1.0763, "step": 7415 }, { "epoch": 0.7891040646946159, "grad_norm": 14.0625, "learning_rate": 9.754263786242183e-06, "loss": 1.4469, "step": 7416 }, { "epoch": 0.7892104703128325, "grad_norm": 13.1171875, "learning_rate": 9.754228254690165e-06, "loss": 1.3313, "step": 7417 }, { "epoch": 0.7893168759310492, "grad_norm": 10.96875, "learning_rate": 9.754192723138147e-06, "loss": 1.2838, "step": 7418 }, { "epoch": 0.7894232815492658, "grad_norm": 12.1875, "learning_rate": 9.754157191586128e-06, "loss": 1.226, "step": 7419 }, { "epoch": 0.7895296871674824, "grad_norm": 11.921875, "learning_rate": 9.754121660034111e-06, "loss": 1.1741, "step": 7420 }, { "epoch": 0.7896360927856991, "grad_norm": 13.7734375, "learning_rate": 9.754086128482093e-06, "loss": 1.2399, "step": 7421 }, { "epoch": 0.7897424984039157, "grad_norm": 9.9921875, "learning_rate": 9.754050596930075e-06, "loss": 1.1565, "step": 7422 }, { "epoch": 0.7898489040221324, "grad_norm": 13.015625, "learning_rate": 9.754015065378056e-06, "loss": 1.3283, "step": 7423 }, { "epoch": 0.789955309640349, "grad_norm": 12.640625, "learning_rate": 9.753979533826038e-06, "loss": 1.215, "step": 7424 }, { "epoch": 0.7900617152585656, "grad_norm": 12.375, "learning_rate": 9.753944002274021e-06, "loss": 1.2799, "step": 7425 }, { "epoch": 0.7901681208767823, "grad_norm": 13.3203125, "learning_rate": 9.753908470722003e-06, "loss": 1.1344, "step": 7426 }, { "epoch": 0.790274526494999, "grad_norm": 13.859375, "learning_rate": 9.753872939169984e-06, "loss": 1.1373, "step": 7427 }, { "epoch": 0.7903809321132156, "grad_norm": 10.4921875, "learning_rate": 9.753837407617966e-06, "loss": 1.1558, "step": 7428 }, { "epoch": 0.7904873377314322, "grad_norm": 13.546875, "learning_rate": 9.753801876065948e-06, "loss": 1.1147, "step": 7429 }, { "epoch": 0.7905937433496488, "grad_norm": 14.6015625, "learning_rate": 9.75376634451393e-06, "loss": 1.4917, "step": 7430 }, { "epoch": 0.7907001489678656, "grad_norm": 12.9375, "learning_rate": 9.75373081296191e-06, "loss": 1.4637, "step": 7431 }, { "epoch": 0.7908065545860822, "grad_norm": 12.875, "learning_rate": 9.753695281409892e-06, "loss": 1.2217, "step": 7432 }, { "epoch": 0.7909129602042988, "grad_norm": 13.109375, "learning_rate": 9.753659749857874e-06, "loss": 1.0115, "step": 7433 }, { "epoch": 0.7910193658225154, "grad_norm": 13.6171875, "learning_rate": 9.753624218305856e-06, "loss": 1.2432, "step": 7434 }, { "epoch": 0.791125771440732, "grad_norm": 13.4609375, "learning_rate": 9.753588686753839e-06, "loss": 1.3131, "step": 7435 }, { "epoch": 0.7912321770589487, "grad_norm": 12.5, "learning_rate": 9.75355315520182e-06, "loss": 0.9611, "step": 7436 }, { "epoch": 0.7913385826771654, "grad_norm": 18.40625, "learning_rate": 9.753517623649802e-06, "loss": 1.4533, "step": 7437 }, { "epoch": 0.791444988295382, "grad_norm": 14.28125, "learning_rate": 9.753482092097784e-06, "loss": 1.1882, "step": 7438 }, { "epoch": 0.7915513939135986, "grad_norm": 13.8359375, "learning_rate": 9.753446560545765e-06, "loss": 1.3861, "step": 7439 }, { "epoch": 0.7916577995318153, "grad_norm": 14.6328125, "learning_rate": 9.753411028993747e-06, "loss": 1.572, "step": 7440 }, { "epoch": 0.7917642051500319, "grad_norm": 13.59375, "learning_rate": 9.753375497441729e-06, "loss": 1.5394, "step": 7441 }, { "epoch": 0.7918706107682486, "grad_norm": 12.6796875, "learning_rate": 9.75333996588971e-06, "loss": 1.4198, "step": 7442 }, { "epoch": 0.7919770163864652, "grad_norm": 8.9453125, "learning_rate": 9.753304434337692e-06, "loss": 1.1139, "step": 7443 }, { "epoch": 0.7920834220046818, "grad_norm": 10.4609375, "learning_rate": 9.753268902785673e-06, "loss": 1.2029, "step": 7444 }, { "epoch": 0.7921898276228985, "grad_norm": 14.3359375, "learning_rate": 9.753233371233655e-06, "loss": 0.9746, "step": 7445 }, { "epoch": 0.7922962332411151, "grad_norm": 23.796875, "learning_rate": 9.753197839681638e-06, "loss": 1.0889, "step": 7446 }, { "epoch": 0.7924026388593318, "grad_norm": 14.296875, "learning_rate": 9.75316230812962e-06, "loss": 1.2568, "step": 7447 }, { "epoch": 0.7925090444775484, "grad_norm": 21.578125, "learning_rate": 9.753126776577601e-06, "loss": 1.2861, "step": 7448 }, { "epoch": 0.7926154500957651, "grad_norm": 15.859375, "learning_rate": 9.753091245025583e-06, "loss": 1.3316, "step": 7449 }, { "epoch": 0.7927218557139817, "grad_norm": 13.34375, "learning_rate": 9.753055713473565e-06, "loss": 1.2123, "step": 7450 }, { "epoch": 0.7928282613321983, "grad_norm": 12.0703125, "learning_rate": 9.753020181921548e-06, "loss": 1.1751, "step": 7451 }, { "epoch": 0.792934666950415, "grad_norm": 10.75, "learning_rate": 9.75298465036953e-06, "loss": 1.4527, "step": 7452 }, { "epoch": 0.7930410725686317, "grad_norm": 14.921875, "learning_rate": 9.752949118817511e-06, "loss": 1.0972, "step": 7453 }, { "epoch": 0.7931474781868483, "grad_norm": 16.1875, "learning_rate": 9.752913587265493e-06, "loss": 0.9997, "step": 7454 }, { "epoch": 0.7932538838050649, "grad_norm": 11.78125, "learning_rate": 9.752878055713474e-06, "loss": 1.101, "step": 7455 }, { "epoch": 0.7933602894232815, "grad_norm": 13.203125, "learning_rate": 9.752842524161456e-06, "loss": 1.2668, "step": 7456 }, { "epoch": 0.7934666950414981, "grad_norm": 11.0234375, "learning_rate": 9.752806992609438e-06, "loss": 1.0312, "step": 7457 }, { "epoch": 0.7935731006597149, "grad_norm": 15.40625, "learning_rate": 9.75277146105742e-06, "loss": 1.4031, "step": 7458 }, { "epoch": 0.7936795062779315, "grad_norm": 15.5234375, "learning_rate": 9.752735929505401e-06, "loss": 1.3142, "step": 7459 }, { "epoch": 0.7937859118961481, "grad_norm": 14.140625, "learning_rate": 9.752700397953382e-06, "loss": 1.1443, "step": 7460 }, { "epoch": 0.7938923175143647, "grad_norm": 10.7734375, "learning_rate": 9.752664866401366e-06, "loss": 1.0639, "step": 7461 }, { "epoch": 0.7939987231325814, "grad_norm": 11.5859375, "learning_rate": 9.752629334849347e-06, "loss": 1.1104, "step": 7462 }, { "epoch": 0.7941051287507981, "grad_norm": 11.3828125, "learning_rate": 9.752593803297329e-06, "loss": 1.1578, "step": 7463 }, { "epoch": 0.7942115343690147, "grad_norm": 10.1640625, "learning_rate": 9.75255827174531e-06, "loss": 1.0685, "step": 7464 }, { "epoch": 0.7943179399872313, "grad_norm": 14.0234375, "learning_rate": 9.752522740193292e-06, "loss": 1.0576, "step": 7465 }, { "epoch": 0.794424345605448, "grad_norm": 18.75, "learning_rate": 9.752487208641275e-06, "loss": 1.7212, "step": 7466 }, { "epoch": 0.7945307512236646, "grad_norm": 13.6328125, "learning_rate": 9.752451677089257e-06, "loss": 1.0977, "step": 7467 }, { "epoch": 0.7946371568418813, "grad_norm": 11.65625, "learning_rate": 9.752416145537239e-06, "loss": 1.3809, "step": 7468 }, { "epoch": 0.7947435624600979, "grad_norm": 12.03125, "learning_rate": 9.75238061398522e-06, "loss": 1.1807, "step": 7469 }, { "epoch": 0.7948499680783145, "grad_norm": 11.65625, "learning_rate": 9.752345082433202e-06, "loss": 1.0414, "step": 7470 }, { "epoch": 0.7949563736965312, "grad_norm": 11.609375, "learning_rate": 9.752309550881184e-06, "loss": 1.3202, "step": 7471 }, { "epoch": 0.7950627793147478, "grad_norm": 12.3828125, "learning_rate": 9.752274019329165e-06, "loss": 1.1122, "step": 7472 }, { "epoch": 0.7951691849329645, "grad_norm": 14.625, "learning_rate": 9.752238487777147e-06, "loss": 1.0602, "step": 7473 }, { "epoch": 0.7952755905511811, "grad_norm": 13.21875, "learning_rate": 9.752202956225128e-06, "loss": 1.1735, "step": 7474 }, { "epoch": 0.7953819961693978, "grad_norm": 12.8359375, "learning_rate": 9.75216742467311e-06, "loss": 1.1168, "step": 7475 }, { "epoch": 0.7954884017876144, "grad_norm": 14.5234375, "learning_rate": 9.752131893121093e-06, "loss": 1.3109, "step": 7476 }, { "epoch": 0.795594807405831, "grad_norm": 14.6796875, "learning_rate": 9.752096361569075e-06, "loss": 1.2172, "step": 7477 }, { "epoch": 0.7957012130240476, "grad_norm": 9.484375, "learning_rate": 9.752060830017056e-06, "loss": 1.0451, "step": 7478 }, { "epoch": 0.7958076186422643, "grad_norm": 14.0, "learning_rate": 9.752025298465038e-06, "loss": 1.0801, "step": 7479 }, { "epoch": 0.795914024260481, "grad_norm": 10.9921875, "learning_rate": 9.75198976691302e-06, "loss": 1.0265, "step": 7480 }, { "epoch": 0.7960204298786976, "grad_norm": 12.8671875, "learning_rate": 9.751954235361001e-06, "loss": 1.4987, "step": 7481 }, { "epoch": 0.7961268354969142, "grad_norm": 11.0703125, "learning_rate": 9.751918703808983e-06, "loss": 1.2855, "step": 7482 }, { "epoch": 0.7962332411151308, "grad_norm": 11.875, "learning_rate": 9.751883172256965e-06, "loss": 1.0133, "step": 7483 }, { "epoch": 0.7963396467333476, "grad_norm": 13.3203125, "learning_rate": 9.751847640704946e-06, "loss": 1.0294, "step": 7484 }, { "epoch": 0.7964460523515642, "grad_norm": 13.9375, "learning_rate": 9.751812109152928e-06, "loss": 1.4356, "step": 7485 }, { "epoch": 0.7965524579697808, "grad_norm": 13.4140625, "learning_rate": 9.75177657760091e-06, "loss": 1.1682, "step": 7486 }, { "epoch": 0.7966588635879974, "grad_norm": 16.75, "learning_rate": 9.751741046048893e-06, "loss": 1.2493, "step": 7487 }, { "epoch": 0.796765269206214, "grad_norm": 14.046875, "learning_rate": 9.751705514496874e-06, "loss": 1.4643, "step": 7488 }, { "epoch": 0.7968716748244308, "grad_norm": 10.984375, "learning_rate": 9.751669982944856e-06, "loss": 1.1619, "step": 7489 }, { "epoch": 0.7969780804426474, "grad_norm": 11.953125, "learning_rate": 9.751634451392837e-06, "loss": 1.0618, "step": 7490 }, { "epoch": 0.797084486060864, "grad_norm": 11.234375, "learning_rate": 9.751598919840819e-06, "loss": 1.3033, "step": 7491 }, { "epoch": 0.7971908916790806, "grad_norm": 15.1015625, "learning_rate": 9.751563388288802e-06, "loss": 1.5926, "step": 7492 }, { "epoch": 0.7972972972972973, "grad_norm": 12.8359375, "learning_rate": 9.751527856736784e-06, "loss": 1.0908, "step": 7493 }, { "epoch": 0.797403702915514, "grad_norm": 15.7890625, "learning_rate": 9.751492325184766e-06, "loss": 1.1946, "step": 7494 }, { "epoch": 0.7975101085337306, "grad_norm": 11.7890625, "learning_rate": 9.751456793632747e-06, "loss": 1.3027, "step": 7495 }, { "epoch": 0.7976165141519472, "grad_norm": 11.5234375, "learning_rate": 9.751421262080729e-06, "loss": 1.1409, "step": 7496 }, { "epoch": 0.7977229197701639, "grad_norm": 9.4375, "learning_rate": 9.75138573052871e-06, "loss": 1.1922, "step": 7497 }, { "epoch": 0.7978293253883805, "grad_norm": 11.9296875, "learning_rate": 9.751350198976692e-06, "loss": 1.2633, "step": 7498 }, { "epoch": 0.7979357310065971, "grad_norm": 12.6875, "learning_rate": 9.751314667424674e-06, "loss": 1.24, "step": 7499 }, { "epoch": 0.7980421366248138, "grad_norm": 14.4375, "learning_rate": 9.751279135872655e-06, "loss": 1.2882, "step": 7500 }, { "epoch": 0.7981485422430304, "grad_norm": 14.078125, "learning_rate": 9.751243604320637e-06, "loss": 1.5447, "step": 7501 }, { "epoch": 0.7982549478612471, "grad_norm": 17.90625, "learning_rate": 9.75120807276862e-06, "loss": 1.0324, "step": 7502 }, { "epoch": 0.7983613534794637, "grad_norm": 11.125, "learning_rate": 9.751172541216602e-06, "loss": 1.2319, "step": 7503 }, { "epoch": 0.7984677590976803, "grad_norm": 12.25, "learning_rate": 9.751137009664583e-06, "loss": 1.0439, "step": 7504 }, { "epoch": 0.798574164715897, "grad_norm": 14.59375, "learning_rate": 9.751101478112565e-06, "loss": 1.3262, "step": 7505 }, { "epoch": 0.7986805703341137, "grad_norm": 11.2734375, "learning_rate": 9.751065946560547e-06, "loss": 1.1249, "step": 7506 }, { "epoch": 0.7987869759523303, "grad_norm": 12.578125, "learning_rate": 9.751030415008528e-06, "loss": 1.1659, "step": 7507 }, { "epoch": 0.7988933815705469, "grad_norm": 13.5703125, "learning_rate": 9.75099488345651e-06, "loss": 1.3418, "step": 7508 }, { "epoch": 0.7989997871887635, "grad_norm": 13.4921875, "learning_rate": 9.750959351904491e-06, "loss": 0.929, "step": 7509 }, { "epoch": 0.7991061928069803, "grad_norm": 10.53125, "learning_rate": 9.750923820352473e-06, "loss": 1.2084, "step": 7510 }, { "epoch": 0.7992125984251969, "grad_norm": 12.484375, "learning_rate": 9.750888288800455e-06, "loss": 1.18, "step": 7511 }, { "epoch": 0.7993190040434135, "grad_norm": 10.5, "learning_rate": 9.750852757248438e-06, "loss": 1.258, "step": 7512 }, { "epoch": 0.7994254096616301, "grad_norm": 13.3828125, "learning_rate": 9.75081722569642e-06, "loss": 1.3486, "step": 7513 }, { "epoch": 0.7995318152798467, "grad_norm": 11.6796875, "learning_rate": 9.750781694144401e-06, "loss": 1.0539, "step": 7514 }, { "epoch": 0.7996382208980635, "grad_norm": 9.53125, "learning_rate": 9.750746162592383e-06, "loss": 1.2236, "step": 7515 }, { "epoch": 0.7997446265162801, "grad_norm": 13.0546875, "learning_rate": 9.750710631040364e-06, "loss": 1.2374, "step": 7516 }, { "epoch": 0.7998510321344967, "grad_norm": 11.921875, "learning_rate": 9.750675099488346e-06, "loss": 1.4894, "step": 7517 }, { "epoch": 0.7999574377527133, "grad_norm": 13.265625, "learning_rate": 9.75063956793633e-06, "loss": 1.0837, "step": 7518 }, { "epoch": 0.80006384337093, "grad_norm": 13.75, "learning_rate": 9.75060403638431e-06, "loss": 0.9837, "step": 7519 }, { "epoch": 0.8001702489891467, "grad_norm": 10.828125, "learning_rate": 9.750568504832292e-06, "loss": 1.1253, "step": 7520 }, { "epoch": 0.8002766546073633, "grad_norm": 13.4765625, "learning_rate": 9.750532973280274e-06, "loss": 1.5057, "step": 7521 }, { "epoch": 0.8003830602255799, "grad_norm": 11.25, "learning_rate": 9.750497441728256e-06, "loss": 1.3192, "step": 7522 }, { "epoch": 0.8004894658437965, "grad_norm": 15.1796875, "learning_rate": 9.750461910176237e-06, "loss": 1.4282, "step": 7523 }, { "epoch": 0.8005958714620132, "grad_norm": 17.75, "learning_rate": 9.750426378624219e-06, "loss": 1.1483, "step": 7524 }, { "epoch": 0.8007022770802298, "grad_norm": 12.640625, "learning_rate": 9.7503908470722e-06, "loss": 1.1291, "step": 7525 }, { "epoch": 0.8008086826984465, "grad_norm": 12.5859375, "learning_rate": 9.750355315520182e-06, "loss": 1.1211, "step": 7526 }, { "epoch": 0.8009150883166631, "grad_norm": 9.25, "learning_rate": 9.750319783968164e-06, "loss": 1.0127, "step": 7527 }, { "epoch": 0.8010214939348798, "grad_norm": 12.9765625, "learning_rate": 9.750284252416147e-06, "loss": 1.2793, "step": 7528 }, { "epoch": 0.8011278995530964, "grad_norm": 11.734375, "learning_rate": 9.750248720864129e-06, "loss": 1.2452, "step": 7529 }, { "epoch": 0.801234305171313, "grad_norm": 12.015625, "learning_rate": 9.75021318931211e-06, "loss": 1.073, "step": 7530 }, { "epoch": 0.8013407107895297, "grad_norm": 10.9921875, "learning_rate": 9.750177657760092e-06, "loss": 0.8627, "step": 7531 }, { "epoch": 0.8014471164077464, "grad_norm": 10.4296875, "learning_rate": 9.750142126208073e-06, "loss": 1.2285, "step": 7532 }, { "epoch": 0.801553522025963, "grad_norm": 12.765625, "learning_rate": 9.750106594656055e-06, "loss": 1.25, "step": 7533 }, { "epoch": 0.8016599276441796, "grad_norm": 25.34375, "learning_rate": 9.750071063104037e-06, "loss": 1.1204, "step": 7534 }, { "epoch": 0.8017663332623962, "grad_norm": 11.703125, "learning_rate": 9.750035531552018e-06, "loss": 1.1886, "step": 7535 }, { "epoch": 0.801872738880613, "grad_norm": 16.34375, "learning_rate": 9.75e-06, "loss": 1.2434, "step": 7536 }, { "epoch": 0.8019791444988296, "grad_norm": 10.390625, "learning_rate": 9.749964468447981e-06, "loss": 1.0482, "step": 7537 }, { "epoch": 0.8020855501170462, "grad_norm": 15.40625, "learning_rate": 9.749928936895965e-06, "loss": 1.2504, "step": 7538 }, { "epoch": 0.8021919557352628, "grad_norm": 19.890625, "learning_rate": 9.749893405343946e-06, "loss": 1.6306, "step": 7539 }, { "epoch": 0.8022983613534794, "grad_norm": 19.703125, "learning_rate": 9.749857873791928e-06, "loss": 1.5069, "step": 7540 }, { "epoch": 0.8024047669716962, "grad_norm": 16.484375, "learning_rate": 9.74982234223991e-06, "loss": 1.3231, "step": 7541 }, { "epoch": 0.8025111725899128, "grad_norm": 11.234375, "learning_rate": 9.749786810687891e-06, "loss": 1.3649, "step": 7542 }, { "epoch": 0.8026175782081294, "grad_norm": 12.9453125, "learning_rate": 9.749751279135874e-06, "loss": 1.7139, "step": 7543 }, { "epoch": 0.802723983826346, "grad_norm": 15.125, "learning_rate": 9.749715747583856e-06, "loss": 1.0946, "step": 7544 }, { "epoch": 0.8028303894445626, "grad_norm": 11.8046875, "learning_rate": 9.749680216031838e-06, "loss": 1.0674, "step": 7545 }, { "epoch": 0.8029367950627793, "grad_norm": 19.078125, "learning_rate": 9.74964468447982e-06, "loss": 1.3279, "step": 7546 }, { "epoch": 0.803043200680996, "grad_norm": 15.8671875, "learning_rate": 9.749609152927801e-06, "loss": 1.1507, "step": 7547 }, { "epoch": 0.8031496062992126, "grad_norm": 10.0, "learning_rate": 9.749573621375782e-06, "loss": 1.2001, "step": 7548 }, { "epoch": 0.8032560119174292, "grad_norm": 15.59375, "learning_rate": 9.749538089823764e-06, "loss": 1.2074, "step": 7549 }, { "epoch": 0.8033624175356459, "grad_norm": 10.203125, "learning_rate": 9.749502558271746e-06, "loss": 1.0347, "step": 7550 }, { "epoch": 0.8034688231538625, "grad_norm": 11.2578125, "learning_rate": 9.749467026719727e-06, "loss": 1.4641, "step": 7551 }, { "epoch": 0.8035752287720792, "grad_norm": 13.5078125, "learning_rate": 9.749431495167709e-06, "loss": 1.0441, "step": 7552 }, { "epoch": 0.8036816343902958, "grad_norm": 11.984375, "learning_rate": 9.749395963615692e-06, "loss": 1.1131, "step": 7553 }, { "epoch": 0.8037880400085124, "grad_norm": 14.7734375, "learning_rate": 9.749360432063674e-06, "loss": 1.3738, "step": 7554 }, { "epoch": 0.8038944456267291, "grad_norm": 11.0078125, "learning_rate": 9.749324900511655e-06, "loss": 1.1087, "step": 7555 }, { "epoch": 0.8040008512449457, "grad_norm": 11.6015625, "learning_rate": 9.749289368959637e-06, "loss": 1.2949, "step": 7556 }, { "epoch": 0.8041072568631624, "grad_norm": 10.4375, "learning_rate": 9.749253837407619e-06, "loss": 1.2236, "step": 7557 }, { "epoch": 0.804213662481379, "grad_norm": 11.0, "learning_rate": 9.7492183058556e-06, "loss": 1.3122, "step": 7558 }, { "epoch": 0.8043200680995957, "grad_norm": 10.7890625, "learning_rate": 9.749182774303584e-06, "loss": 1.0824, "step": 7559 }, { "epoch": 0.8044264737178123, "grad_norm": 12.1484375, "learning_rate": 9.749147242751565e-06, "loss": 1.2707, "step": 7560 }, { "epoch": 0.8045328793360289, "grad_norm": 12.578125, "learning_rate": 9.749111711199545e-06, "loss": 1.1641, "step": 7561 }, { "epoch": 0.8046392849542456, "grad_norm": 12.65625, "learning_rate": 9.749076179647527e-06, "loss": 1.1345, "step": 7562 }, { "epoch": 0.8047456905724623, "grad_norm": 9.546875, "learning_rate": 9.749040648095508e-06, "loss": 1.062, "step": 7563 }, { "epoch": 0.8048520961906789, "grad_norm": 15.203125, "learning_rate": 9.749005116543492e-06, "loss": 1.466, "step": 7564 }, { "epoch": 0.8049585018088955, "grad_norm": 14.4296875, "learning_rate": 9.748969584991473e-06, "loss": 1.1857, "step": 7565 }, { "epoch": 0.8050649074271121, "grad_norm": 14.390625, "learning_rate": 9.748934053439455e-06, "loss": 0.9516, "step": 7566 }, { "epoch": 0.8051713130453287, "grad_norm": 13.6796875, "learning_rate": 9.748898521887436e-06, "loss": 1.4128, "step": 7567 }, { "epoch": 0.8052777186635455, "grad_norm": 13.03125, "learning_rate": 9.748862990335418e-06, "loss": 1.3817, "step": 7568 }, { "epoch": 0.8053841242817621, "grad_norm": 14.2890625, "learning_rate": 9.748827458783401e-06, "loss": 1.4464, "step": 7569 }, { "epoch": 0.8054905298999787, "grad_norm": 16.328125, "learning_rate": 9.748791927231383e-06, "loss": 1.1771, "step": 7570 }, { "epoch": 0.8055969355181953, "grad_norm": 19.578125, "learning_rate": 9.748756395679365e-06, "loss": 1.0361, "step": 7571 }, { "epoch": 0.805703341136412, "grad_norm": 12.09375, "learning_rate": 9.748720864127346e-06, "loss": 1.1765, "step": 7572 }, { "epoch": 0.8058097467546287, "grad_norm": 13.171875, "learning_rate": 9.748685332575328e-06, "loss": 1.2376, "step": 7573 }, { "epoch": 0.8059161523728453, "grad_norm": 16.53125, "learning_rate": 9.74864980102331e-06, "loss": 1.1746, "step": 7574 }, { "epoch": 0.8060225579910619, "grad_norm": 14.2421875, "learning_rate": 9.748614269471291e-06, "loss": 1.0739, "step": 7575 }, { "epoch": 0.8061289636092785, "grad_norm": 10.875, "learning_rate": 9.748578737919273e-06, "loss": 1.0865, "step": 7576 }, { "epoch": 0.8062353692274952, "grad_norm": 11.3515625, "learning_rate": 9.748543206367254e-06, "loss": 1.6229, "step": 7577 }, { "epoch": 0.8063417748457119, "grad_norm": 12.1171875, "learning_rate": 9.748507674815236e-06, "loss": 0.9566, "step": 7578 }, { "epoch": 0.8064481804639285, "grad_norm": 16.0625, "learning_rate": 9.748472143263219e-06, "loss": 1.2811, "step": 7579 }, { "epoch": 0.8065545860821451, "grad_norm": 17.625, "learning_rate": 9.7484366117112e-06, "loss": 1.0677, "step": 7580 }, { "epoch": 0.8066609917003618, "grad_norm": 13.0625, "learning_rate": 9.748401080159182e-06, "loss": 1.1543, "step": 7581 }, { "epoch": 0.8067673973185784, "grad_norm": 13.609375, "learning_rate": 9.748365548607164e-06, "loss": 1.3498, "step": 7582 }, { "epoch": 0.8068738029367951, "grad_norm": 12.1796875, "learning_rate": 9.748330017055146e-06, "loss": 1.2371, "step": 7583 }, { "epoch": 0.8069802085550117, "grad_norm": 13.0859375, "learning_rate": 9.748294485503129e-06, "loss": 1.189, "step": 7584 }, { "epoch": 0.8070866141732284, "grad_norm": 11.0859375, "learning_rate": 9.74825895395111e-06, "loss": 1.0978, "step": 7585 }, { "epoch": 0.807193019791445, "grad_norm": 10.8515625, "learning_rate": 9.748223422399092e-06, "loss": 1.2341, "step": 7586 }, { "epoch": 0.8072994254096616, "grad_norm": 12.90625, "learning_rate": 9.748187890847074e-06, "loss": 1.0445, "step": 7587 }, { "epoch": 0.8074058310278782, "grad_norm": 13.8046875, "learning_rate": 9.748152359295055e-06, "loss": 1.1302, "step": 7588 }, { "epoch": 0.807512236646095, "grad_norm": 14.140625, "learning_rate": 9.748116827743037e-06, "loss": 1.3072, "step": 7589 }, { "epoch": 0.8076186422643116, "grad_norm": 13.6171875, "learning_rate": 9.748081296191018e-06, "loss": 0.9992, "step": 7590 }, { "epoch": 0.8077250478825282, "grad_norm": 12.9453125, "learning_rate": 9.748045764639e-06, "loss": 0.9413, "step": 7591 }, { "epoch": 0.8078314535007448, "grad_norm": 10.3203125, "learning_rate": 9.748010233086982e-06, "loss": 1.2333, "step": 7592 }, { "epoch": 0.8079378591189614, "grad_norm": 14.1015625, "learning_rate": 9.747974701534963e-06, "loss": 1.6802, "step": 7593 }, { "epoch": 0.8080442647371782, "grad_norm": 9.9140625, "learning_rate": 9.747939169982947e-06, "loss": 0.8663, "step": 7594 }, { "epoch": 0.8081506703553948, "grad_norm": 19.875, "learning_rate": 9.747903638430928e-06, "loss": 1.248, "step": 7595 }, { "epoch": 0.8082570759736114, "grad_norm": 9.65625, "learning_rate": 9.74786810687891e-06, "loss": 1.0682, "step": 7596 }, { "epoch": 0.808363481591828, "grad_norm": 12.0703125, "learning_rate": 9.747832575326891e-06, "loss": 1.308, "step": 7597 }, { "epoch": 0.8084698872100446, "grad_norm": 10.140625, "learning_rate": 9.747797043774873e-06, "loss": 1.2943, "step": 7598 }, { "epoch": 0.8085762928282614, "grad_norm": 13.453125, "learning_rate": 9.747761512222855e-06, "loss": 1.1264, "step": 7599 }, { "epoch": 0.808682698446478, "grad_norm": 15.4609375, "learning_rate": 9.747725980670836e-06, "loss": 1.1816, "step": 7600 }, { "epoch": 0.8087891040646946, "grad_norm": 11.6796875, "learning_rate": 9.747690449118818e-06, "loss": 1.1427, "step": 7601 }, { "epoch": 0.8088955096829112, "grad_norm": 9.8359375, "learning_rate": 9.7476549175668e-06, "loss": 1.3074, "step": 7602 }, { "epoch": 0.8090019153011279, "grad_norm": 11.375, "learning_rate": 9.747619386014781e-06, "loss": 1.12, "step": 7603 }, { "epoch": 0.8091083209193446, "grad_norm": 14.3671875, "learning_rate": 9.747583854462763e-06, "loss": 1.0729, "step": 7604 }, { "epoch": 0.8092147265375612, "grad_norm": 11.7578125, "learning_rate": 9.747548322910746e-06, "loss": 1.0828, "step": 7605 }, { "epoch": 0.8093211321557778, "grad_norm": 11.453125, "learning_rate": 9.747512791358728e-06, "loss": 1.0411, "step": 7606 }, { "epoch": 0.8094275377739945, "grad_norm": 11.75, "learning_rate": 9.747477259806709e-06, "loss": 1.1652, "step": 7607 }, { "epoch": 0.8095339433922111, "grad_norm": 12.1015625, "learning_rate": 9.74744172825469e-06, "loss": 1.3514, "step": 7608 }, { "epoch": 0.8096403490104277, "grad_norm": 9.171875, "learning_rate": 9.747406196702672e-06, "loss": 1.103, "step": 7609 }, { "epoch": 0.8097467546286444, "grad_norm": 12.7578125, "learning_rate": 9.747370665150656e-06, "loss": 1.0581, "step": 7610 }, { "epoch": 0.809853160246861, "grad_norm": 16.515625, "learning_rate": 9.747335133598637e-06, "loss": 1.3825, "step": 7611 }, { "epoch": 0.8099595658650777, "grad_norm": 14.9453125, "learning_rate": 9.747299602046619e-06, "loss": 1.1835, "step": 7612 }, { "epoch": 0.8100659714832943, "grad_norm": 12.0234375, "learning_rate": 9.7472640704946e-06, "loss": 1.3484, "step": 7613 }, { "epoch": 0.8101723771015109, "grad_norm": 14.2109375, "learning_rate": 9.747228538942582e-06, "loss": 1.5737, "step": 7614 }, { "epoch": 0.8102787827197276, "grad_norm": 13.7265625, "learning_rate": 9.747193007390564e-06, "loss": 1.1582, "step": 7615 }, { "epoch": 0.8103851883379443, "grad_norm": 12.328125, "learning_rate": 9.747157475838545e-06, "loss": 1.1412, "step": 7616 }, { "epoch": 0.8104915939561609, "grad_norm": 17.96875, "learning_rate": 9.747121944286527e-06, "loss": 1.611, "step": 7617 }, { "epoch": 0.8105979995743775, "grad_norm": 10.40625, "learning_rate": 9.747086412734509e-06, "loss": 1.2589, "step": 7618 }, { "epoch": 0.8107044051925941, "grad_norm": 10.8671875, "learning_rate": 9.74705088118249e-06, "loss": 1.1609, "step": 7619 }, { "epoch": 0.8108108108108109, "grad_norm": 17.21875, "learning_rate": 9.747015349630473e-06, "loss": 1.4052, "step": 7620 }, { "epoch": 0.8109172164290275, "grad_norm": 44.65625, "learning_rate": 9.746979818078455e-06, "loss": 1.1955, "step": 7621 }, { "epoch": 0.8110236220472441, "grad_norm": 11.328125, "learning_rate": 9.746944286526437e-06, "loss": 0.9658, "step": 7622 }, { "epoch": 0.8111300276654607, "grad_norm": 13.484375, "learning_rate": 9.746908754974418e-06, "loss": 1.5313, "step": 7623 }, { "epoch": 0.8112364332836773, "grad_norm": 10.625, "learning_rate": 9.7468732234224e-06, "loss": 1.154, "step": 7624 }, { "epoch": 0.8113428389018941, "grad_norm": 10.890625, "learning_rate": 9.746837691870381e-06, "loss": 1.291, "step": 7625 }, { "epoch": 0.8114492445201107, "grad_norm": 13.640625, "learning_rate": 9.746802160318363e-06, "loss": 1.1547, "step": 7626 }, { "epoch": 0.8115556501383273, "grad_norm": 10.484375, "learning_rate": 9.746766628766345e-06, "loss": 0.9126, "step": 7627 }, { "epoch": 0.8116620557565439, "grad_norm": 12.328125, "learning_rate": 9.746731097214326e-06, "loss": 1.2342, "step": 7628 }, { "epoch": 0.8117684613747606, "grad_norm": 11.984375, "learning_rate": 9.746695565662308e-06, "loss": 1.2742, "step": 7629 }, { "epoch": 0.8118748669929773, "grad_norm": 13.9765625, "learning_rate": 9.746660034110291e-06, "loss": 1.3696, "step": 7630 }, { "epoch": 0.8119812726111939, "grad_norm": 13.2890625, "learning_rate": 9.746624502558273e-06, "loss": 1.2282, "step": 7631 }, { "epoch": 0.8120876782294105, "grad_norm": 25.859375, "learning_rate": 9.746588971006254e-06, "loss": 0.768, "step": 7632 }, { "epoch": 0.8121940838476271, "grad_norm": 9.59375, "learning_rate": 9.746553439454236e-06, "loss": 1.1572, "step": 7633 }, { "epoch": 0.8123004894658438, "grad_norm": 14.7734375, "learning_rate": 9.746517907902218e-06, "loss": 0.9194, "step": 7634 }, { "epoch": 0.8124068950840604, "grad_norm": 13.2578125, "learning_rate": 9.7464823763502e-06, "loss": 1.5846, "step": 7635 }, { "epoch": 0.8125133007022771, "grad_norm": 14.421875, "learning_rate": 9.746446844798183e-06, "loss": 1.3147, "step": 7636 }, { "epoch": 0.8126197063204937, "grad_norm": 14.2421875, "learning_rate": 9.746411313246164e-06, "loss": 1.2368, "step": 7637 }, { "epoch": 0.8127261119387104, "grad_norm": 10.921875, "learning_rate": 9.746375781694146e-06, "loss": 0.9469, "step": 7638 }, { "epoch": 0.812832517556927, "grad_norm": 12.0703125, "learning_rate": 9.746340250142127e-06, "loss": 1.2897, "step": 7639 }, { "epoch": 0.8129389231751436, "grad_norm": 18.515625, "learning_rate": 9.746304718590109e-06, "loss": 1.3469, "step": 7640 }, { "epoch": 0.8130453287933603, "grad_norm": 11.078125, "learning_rate": 9.74626918703809e-06, "loss": 1.1407, "step": 7641 }, { "epoch": 0.813151734411577, "grad_norm": 13.6015625, "learning_rate": 9.746233655486072e-06, "loss": 1.0683, "step": 7642 }, { "epoch": 0.8132581400297936, "grad_norm": 10.3125, "learning_rate": 9.746198123934054e-06, "loss": 1.2858, "step": 7643 }, { "epoch": 0.8133645456480102, "grad_norm": 16.828125, "learning_rate": 9.746162592382035e-06, "loss": 1.2156, "step": 7644 }, { "epoch": 0.8134709512662268, "grad_norm": 10.640625, "learning_rate": 9.746127060830017e-06, "loss": 1.3129, "step": 7645 }, { "epoch": 0.8135773568844435, "grad_norm": 12.625, "learning_rate": 9.746091529278e-06, "loss": 1.0302, "step": 7646 }, { "epoch": 0.8136837625026602, "grad_norm": 12.484375, "learning_rate": 9.746055997725982e-06, "loss": 1.0833, "step": 7647 }, { "epoch": 0.8137901681208768, "grad_norm": 12.921875, "learning_rate": 9.746020466173963e-06, "loss": 0.96, "step": 7648 }, { "epoch": 0.8138965737390934, "grad_norm": 10.0625, "learning_rate": 9.745984934621945e-06, "loss": 1.2954, "step": 7649 }, { "epoch": 0.81400297935731, "grad_norm": 11.890625, "learning_rate": 9.745949403069927e-06, "loss": 1.2575, "step": 7650 }, { "epoch": 0.8141093849755268, "grad_norm": 12.6328125, "learning_rate": 9.745913871517908e-06, "loss": 1.2847, "step": 7651 }, { "epoch": 0.8142157905937434, "grad_norm": 10.3984375, "learning_rate": 9.74587833996589e-06, "loss": 0.9251, "step": 7652 }, { "epoch": 0.81432219621196, "grad_norm": 15.9453125, "learning_rate": 9.745842808413872e-06, "loss": 1.2622, "step": 7653 }, { "epoch": 0.8144286018301766, "grad_norm": 10.921875, "learning_rate": 9.745807276861853e-06, "loss": 0.9866, "step": 7654 }, { "epoch": 0.8145350074483932, "grad_norm": 13.4140625, "learning_rate": 9.745771745309835e-06, "loss": 1.7631, "step": 7655 }, { "epoch": 0.8146414130666099, "grad_norm": 12.3046875, "learning_rate": 9.745736213757818e-06, "loss": 0.9248, "step": 7656 }, { "epoch": 0.8147478186848266, "grad_norm": 13.25, "learning_rate": 9.7457006822058e-06, "loss": 1.3757, "step": 7657 }, { "epoch": 0.8148542243030432, "grad_norm": 22.953125, "learning_rate": 9.745665150653781e-06, "loss": 1.4134, "step": 7658 }, { "epoch": 0.8149606299212598, "grad_norm": 20.875, "learning_rate": 9.745629619101763e-06, "loss": 1.4576, "step": 7659 }, { "epoch": 0.8150670355394765, "grad_norm": 12.9140625, "learning_rate": 9.745594087549744e-06, "loss": 0.9857, "step": 7660 }, { "epoch": 0.8151734411576931, "grad_norm": 13.2421875, "learning_rate": 9.745558555997728e-06, "loss": 1.3116, "step": 7661 }, { "epoch": 0.8152798467759098, "grad_norm": 12.40625, "learning_rate": 9.74552302444571e-06, "loss": 1.1293, "step": 7662 }, { "epoch": 0.8153862523941264, "grad_norm": 12.3828125, "learning_rate": 9.745487492893691e-06, "loss": 1.3299, "step": 7663 }, { "epoch": 0.815492658012343, "grad_norm": 11.9296875, "learning_rate": 9.745451961341673e-06, "loss": 1.1682, "step": 7664 }, { "epoch": 0.8155990636305597, "grad_norm": 12.15625, "learning_rate": 9.745416429789654e-06, "loss": 1.0806, "step": 7665 }, { "epoch": 0.8157054692487763, "grad_norm": 9.625, "learning_rate": 9.745380898237636e-06, "loss": 0.9664, "step": 7666 }, { "epoch": 0.815811874866993, "grad_norm": 11.9453125, "learning_rate": 9.745345366685617e-06, "loss": 1.0683, "step": 7667 }, { "epoch": 0.8159182804852096, "grad_norm": 10.25, "learning_rate": 9.745309835133599e-06, "loss": 1.1392, "step": 7668 }, { "epoch": 0.8160246861034263, "grad_norm": 11.140625, "learning_rate": 9.74527430358158e-06, "loss": 1.2216, "step": 7669 }, { "epoch": 0.8161310917216429, "grad_norm": 14.1796875, "learning_rate": 9.745238772029562e-06, "loss": 1.0584, "step": 7670 }, { "epoch": 0.8162374973398595, "grad_norm": 13.96875, "learning_rate": 9.745203240477546e-06, "loss": 1.0192, "step": 7671 }, { "epoch": 0.8163439029580762, "grad_norm": 16.90625, "learning_rate": 9.745167708925527e-06, "loss": 1.2014, "step": 7672 }, { "epoch": 0.8164503085762929, "grad_norm": 17.171875, "learning_rate": 9.745132177373509e-06, "loss": 1.0257, "step": 7673 }, { "epoch": 0.8165567141945095, "grad_norm": 8.78125, "learning_rate": 9.74509664582149e-06, "loss": 0.8613, "step": 7674 }, { "epoch": 0.8166631198127261, "grad_norm": 15.421875, "learning_rate": 9.745061114269472e-06, "loss": 1.2876, "step": 7675 }, { "epoch": 0.8167695254309427, "grad_norm": 11.2109375, "learning_rate": 9.745025582717454e-06, "loss": 0.991, "step": 7676 }, { "epoch": 0.8168759310491593, "grad_norm": 15.5703125, "learning_rate": 9.744990051165437e-06, "loss": 1.2219, "step": 7677 }, { "epoch": 0.8169823366673761, "grad_norm": 14.2734375, "learning_rate": 9.744954519613418e-06, "loss": 1.196, "step": 7678 }, { "epoch": 0.8170887422855927, "grad_norm": 15.578125, "learning_rate": 9.7449189880614e-06, "loss": 1.4305, "step": 7679 }, { "epoch": 0.8171951479038093, "grad_norm": 11.9140625, "learning_rate": 9.744883456509382e-06, "loss": 1.036, "step": 7680 }, { "epoch": 0.8173015535220259, "grad_norm": 16.609375, "learning_rate": 9.744847924957362e-06, "loss": 1.3114, "step": 7681 }, { "epoch": 0.8174079591402426, "grad_norm": 11.703125, "learning_rate": 9.744812393405345e-06, "loss": 1.1926, "step": 7682 }, { "epoch": 0.8175143647584593, "grad_norm": 12.4453125, "learning_rate": 9.744776861853326e-06, "loss": 1.0414, "step": 7683 }, { "epoch": 0.8176207703766759, "grad_norm": 14.2109375, "learning_rate": 9.744741330301308e-06, "loss": 1.0595, "step": 7684 }, { "epoch": 0.8177271759948925, "grad_norm": 13.34375, "learning_rate": 9.74470579874929e-06, "loss": 1.3479, "step": 7685 }, { "epoch": 0.8178335816131092, "grad_norm": 12.46875, "learning_rate": 9.744670267197271e-06, "loss": 1.2444, "step": 7686 }, { "epoch": 0.8179399872313258, "grad_norm": 13.140625, "learning_rate": 9.744634735645255e-06, "loss": 1.0517, "step": 7687 }, { "epoch": 0.8180463928495425, "grad_norm": 12.5390625, "learning_rate": 9.744599204093236e-06, "loss": 1.4771, "step": 7688 }, { "epoch": 0.8181527984677591, "grad_norm": 15.4296875, "learning_rate": 9.744563672541218e-06, "loss": 1.143, "step": 7689 }, { "epoch": 0.8182592040859757, "grad_norm": 11.4140625, "learning_rate": 9.7445281409892e-06, "loss": 1.2194, "step": 7690 }, { "epoch": 0.8183656097041924, "grad_norm": 15.703125, "learning_rate": 9.744492609437181e-06, "loss": 1.2306, "step": 7691 }, { "epoch": 0.818472015322409, "grad_norm": 11.9375, "learning_rate": 9.744457077885163e-06, "loss": 1.0629, "step": 7692 }, { "epoch": 0.8185784209406257, "grad_norm": 9.34375, "learning_rate": 9.744421546333144e-06, "loss": 1.2186, "step": 7693 }, { "epoch": 0.8186848265588423, "grad_norm": 12.0546875, "learning_rate": 9.744386014781126e-06, "loss": 1.4168, "step": 7694 }, { "epoch": 0.818791232177059, "grad_norm": 13.2421875, "learning_rate": 9.744350483229107e-06, "loss": 1.3156, "step": 7695 }, { "epoch": 0.8188976377952756, "grad_norm": 17.265625, "learning_rate": 9.744314951677089e-06, "loss": 1.4214, "step": 7696 }, { "epoch": 0.8190040434134922, "grad_norm": 16.328125, "learning_rate": 9.744279420125072e-06, "loss": 1.0674, "step": 7697 }, { "epoch": 0.8191104490317088, "grad_norm": 13.4140625, "learning_rate": 9.744243888573054e-06, "loss": 1.025, "step": 7698 }, { "epoch": 0.8192168546499256, "grad_norm": 9.65625, "learning_rate": 9.744208357021036e-06, "loss": 1.1951, "step": 7699 }, { "epoch": 0.8193232602681422, "grad_norm": 13.1171875, "learning_rate": 9.744172825469017e-06, "loss": 0.9352, "step": 7700 }, { "epoch": 0.8194296658863588, "grad_norm": 19.28125, "learning_rate": 9.744137293916999e-06, "loss": 1.2232, "step": 7701 }, { "epoch": 0.8195360715045754, "grad_norm": 9.3203125, "learning_rate": 9.744101762364982e-06, "loss": 1.1976, "step": 7702 }, { "epoch": 0.819642477122792, "grad_norm": 9.7578125, "learning_rate": 9.744066230812964e-06, "loss": 0.9545, "step": 7703 }, { "epoch": 0.8197488827410088, "grad_norm": 11.21875, "learning_rate": 9.744030699260945e-06, "loss": 1.5686, "step": 7704 }, { "epoch": 0.8198552883592254, "grad_norm": 14.3046875, "learning_rate": 9.743995167708927e-06, "loss": 0.9987, "step": 7705 }, { "epoch": 0.819961693977442, "grad_norm": 10.734375, "learning_rate": 9.743959636156909e-06, "loss": 1.1659, "step": 7706 }, { "epoch": 0.8200680995956586, "grad_norm": 17.015625, "learning_rate": 9.74392410460489e-06, "loss": 1.2563, "step": 7707 }, { "epoch": 0.8201745052138752, "grad_norm": 19.0625, "learning_rate": 9.743888573052872e-06, "loss": 1.4572, "step": 7708 }, { "epoch": 0.820280910832092, "grad_norm": 15.0, "learning_rate": 9.743853041500853e-06, "loss": 1.0982, "step": 7709 }, { "epoch": 0.8203873164503086, "grad_norm": 15.3046875, "learning_rate": 9.743817509948835e-06, "loss": 1.0762, "step": 7710 }, { "epoch": 0.8204937220685252, "grad_norm": 12.4375, "learning_rate": 9.743781978396817e-06, "loss": 0.9825, "step": 7711 }, { "epoch": 0.8206001276867418, "grad_norm": 13.484375, "learning_rate": 9.743746446844798e-06, "loss": 1.4045, "step": 7712 }, { "epoch": 0.8207065333049585, "grad_norm": 13.0, "learning_rate": 9.743710915292781e-06, "loss": 1.2464, "step": 7713 }, { "epoch": 0.8208129389231752, "grad_norm": 13.0703125, "learning_rate": 9.743675383740763e-06, "loss": 1.3763, "step": 7714 }, { "epoch": 0.8209193445413918, "grad_norm": 16.0625, "learning_rate": 9.743639852188745e-06, "loss": 1.2282, "step": 7715 }, { "epoch": 0.8210257501596084, "grad_norm": 15.6171875, "learning_rate": 9.743604320636726e-06, "loss": 1.1298, "step": 7716 }, { "epoch": 0.8211321557778251, "grad_norm": 11.1640625, "learning_rate": 9.743568789084708e-06, "loss": 1.057, "step": 7717 }, { "epoch": 0.8212385613960417, "grad_norm": 13.5703125, "learning_rate": 9.74353325753269e-06, "loss": 1.0367, "step": 7718 }, { "epoch": 0.8213449670142583, "grad_norm": 19.046875, "learning_rate": 9.743497725980671e-06, "loss": 1.3206, "step": 7719 }, { "epoch": 0.821451372632475, "grad_norm": 12.2265625, "learning_rate": 9.743462194428653e-06, "loss": 1.0851, "step": 7720 }, { "epoch": 0.8215577782506917, "grad_norm": 15.53125, "learning_rate": 9.743426662876634e-06, "loss": 1.0469, "step": 7721 }, { "epoch": 0.8216641838689083, "grad_norm": 11.6015625, "learning_rate": 9.743391131324616e-06, "loss": 1.2625, "step": 7722 }, { "epoch": 0.8217705894871249, "grad_norm": 16.40625, "learning_rate": 9.7433555997726e-06, "loss": 1.4028, "step": 7723 }, { "epoch": 0.8218769951053415, "grad_norm": 8.9609375, "learning_rate": 9.74332006822058e-06, "loss": 1.0195, "step": 7724 }, { "epoch": 0.8219834007235582, "grad_norm": 10.0, "learning_rate": 9.743284536668562e-06, "loss": 0.9373, "step": 7725 }, { "epoch": 0.8220898063417749, "grad_norm": 13.359375, "learning_rate": 9.743249005116544e-06, "loss": 1.3745, "step": 7726 }, { "epoch": 0.8221962119599915, "grad_norm": 12.390625, "learning_rate": 9.743213473564526e-06, "loss": 1.0189, "step": 7727 }, { "epoch": 0.8223026175782081, "grad_norm": 15.1328125, "learning_rate": 9.743177942012509e-06, "loss": 1.1103, "step": 7728 }, { "epoch": 0.8224090231964247, "grad_norm": 10.21875, "learning_rate": 9.74314241046049e-06, "loss": 1.3692, "step": 7729 }, { "epoch": 0.8225154288146415, "grad_norm": 13.8046875, "learning_rate": 9.743106878908472e-06, "loss": 1.0467, "step": 7730 }, { "epoch": 0.8226218344328581, "grad_norm": 13.1875, "learning_rate": 9.743071347356454e-06, "loss": 1.3343, "step": 7731 }, { "epoch": 0.8227282400510747, "grad_norm": 15.046875, "learning_rate": 9.743035815804435e-06, "loss": 1.4026, "step": 7732 }, { "epoch": 0.8228346456692913, "grad_norm": 12.234375, "learning_rate": 9.743000284252417e-06, "loss": 1.0815, "step": 7733 }, { "epoch": 0.8229410512875079, "grad_norm": 15.6796875, "learning_rate": 9.742964752700399e-06, "loss": 1.2562, "step": 7734 }, { "epoch": 0.8230474569057247, "grad_norm": 12.1484375, "learning_rate": 9.74292922114838e-06, "loss": 1.3359, "step": 7735 }, { "epoch": 0.8231538625239413, "grad_norm": 16.21875, "learning_rate": 9.742893689596362e-06, "loss": 1.2378, "step": 7736 }, { "epoch": 0.8232602681421579, "grad_norm": 11.390625, "learning_rate": 9.742858158044343e-06, "loss": 1.2341, "step": 7737 }, { "epoch": 0.8233666737603745, "grad_norm": 11.6484375, "learning_rate": 9.742822626492327e-06, "loss": 1.088, "step": 7738 }, { "epoch": 0.8234730793785912, "grad_norm": 14.1484375, "learning_rate": 9.742787094940308e-06, "loss": 0.9269, "step": 7739 }, { "epoch": 0.8235794849968079, "grad_norm": 13.75, "learning_rate": 9.74275156338829e-06, "loss": 1.3509, "step": 7740 }, { "epoch": 0.8236858906150245, "grad_norm": 21.296875, "learning_rate": 9.742716031836272e-06, "loss": 1.711, "step": 7741 }, { "epoch": 0.8237922962332411, "grad_norm": 13.2421875, "learning_rate": 9.742680500284253e-06, "loss": 1.1862, "step": 7742 }, { "epoch": 0.8238987018514577, "grad_norm": 9.40625, "learning_rate": 9.742644968732235e-06, "loss": 1.0909, "step": 7743 }, { "epoch": 0.8240051074696744, "grad_norm": 12.015625, "learning_rate": 9.742609437180216e-06, "loss": 1.1365, "step": 7744 }, { "epoch": 0.824111513087891, "grad_norm": 13.9453125, "learning_rate": 9.742573905628198e-06, "loss": 1.2715, "step": 7745 }, { "epoch": 0.8242179187061077, "grad_norm": 14.0703125, "learning_rate": 9.74253837407618e-06, "loss": 1.3217, "step": 7746 }, { "epoch": 0.8243243243243243, "grad_norm": 13.6015625, "learning_rate": 9.742502842524161e-06, "loss": 1.1393, "step": 7747 }, { "epoch": 0.824430729942541, "grad_norm": 11.859375, "learning_rate": 9.742467310972144e-06, "loss": 0.8053, "step": 7748 }, { "epoch": 0.8245371355607576, "grad_norm": 10.6875, "learning_rate": 9.742431779420126e-06, "loss": 1.154, "step": 7749 }, { "epoch": 0.8246435411789742, "grad_norm": 12.65625, "learning_rate": 9.742396247868108e-06, "loss": 1.3515, "step": 7750 }, { "epoch": 0.8247499467971909, "grad_norm": 10.1875, "learning_rate": 9.74236071631609e-06, "loss": 1.1771, "step": 7751 }, { "epoch": 0.8248563524154076, "grad_norm": 15.1875, "learning_rate": 9.742325184764071e-06, "loss": 1.0026, "step": 7752 }, { "epoch": 0.8249627580336242, "grad_norm": 16.84375, "learning_rate": 9.742289653212053e-06, "loss": 1.1724, "step": 7753 }, { "epoch": 0.8250691636518408, "grad_norm": 10.7421875, "learning_rate": 9.742254121660036e-06, "loss": 1.1641, "step": 7754 }, { "epoch": 0.8251755692700574, "grad_norm": 11.8671875, "learning_rate": 9.742218590108017e-06, "loss": 1.0254, "step": 7755 }, { "epoch": 0.8252819748882742, "grad_norm": 14.1640625, "learning_rate": 9.742183058555999e-06, "loss": 1.0017, "step": 7756 }, { "epoch": 0.8253883805064908, "grad_norm": 13.3359375, "learning_rate": 9.74214752700398e-06, "loss": 1.3778, "step": 7757 }, { "epoch": 0.8254947861247074, "grad_norm": 12.0625, "learning_rate": 9.742111995451962e-06, "loss": 1.0078, "step": 7758 }, { "epoch": 0.825601191742924, "grad_norm": 11.2109375, "learning_rate": 9.742076463899944e-06, "loss": 1.3064, "step": 7759 }, { "epoch": 0.8257075973611406, "grad_norm": 15.171875, "learning_rate": 9.742040932347925e-06, "loss": 1.4489, "step": 7760 }, { "epoch": 0.8258140029793574, "grad_norm": 11.484375, "learning_rate": 9.742005400795907e-06, "loss": 1.0154, "step": 7761 }, { "epoch": 0.825920408597574, "grad_norm": 10.5, "learning_rate": 9.741969869243889e-06, "loss": 1.3772, "step": 7762 }, { "epoch": 0.8260268142157906, "grad_norm": 15.453125, "learning_rate": 9.74193433769187e-06, "loss": 1.2592, "step": 7763 }, { "epoch": 0.8261332198340072, "grad_norm": 12.6015625, "learning_rate": 9.741898806139854e-06, "loss": 1.3375, "step": 7764 }, { "epoch": 0.8262396254522238, "grad_norm": 12.2265625, "learning_rate": 9.741863274587835e-06, "loss": 1.0589, "step": 7765 }, { "epoch": 0.8263460310704405, "grad_norm": 11.484375, "learning_rate": 9.741827743035817e-06, "loss": 1.2349, "step": 7766 }, { "epoch": 0.8264524366886572, "grad_norm": 14.4296875, "learning_rate": 9.741792211483798e-06, "loss": 1.3579, "step": 7767 }, { "epoch": 0.8265588423068738, "grad_norm": 18.359375, "learning_rate": 9.74175667993178e-06, "loss": 1.3105, "step": 7768 }, { "epoch": 0.8266652479250904, "grad_norm": 11.7734375, "learning_rate": 9.741721148379763e-06, "loss": 1.2414, "step": 7769 }, { "epoch": 0.8267716535433071, "grad_norm": 12.3359375, "learning_rate": 9.741685616827743e-06, "loss": 0.969, "step": 7770 }, { "epoch": 0.8268780591615237, "grad_norm": 12.5859375, "learning_rate": 9.741650085275725e-06, "loss": 1.1241, "step": 7771 }, { "epoch": 0.8269844647797404, "grad_norm": 15.15625, "learning_rate": 9.741614553723706e-06, "loss": 0.9586, "step": 7772 }, { "epoch": 0.827090870397957, "grad_norm": 13.296875, "learning_rate": 9.741579022171688e-06, "loss": 0.9264, "step": 7773 }, { "epoch": 0.8271972760161737, "grad_norm": 13.703125, "learning_rate": 9.741543490619671e-06, "loss": 1.1148, "step": 7774 }, { "epoch": 0.8273036816343903, "grad_norm": 13.6328125, "learning_rate": 9.741507959067653e-06, "loss": 1.0648, "step": 7775 }, { "epoch": 0.8274100872526069, "grad_norm": 10.265625, "learning_rate": 9.741472427515635e-06, "loss": 1.0881, "step": 7776 }, { "epoch": 0.8275164928708236, "grad_norm": 14.2734375, "learning_rate": 9.741436895963616e-06, "loss": 1.407, "step": 7777 }, { "epoch": 0.8276228984890402, "grad_norm": 12.3671875, "learning_rate": 9.741401364411598e-06, "loss": 1.0035, "step": 7778 }, { "epoch": 0.8277293041072569, "grad_norm": 15.2578125, "learning_rate": 9.741365832859581e-06, "loss": 1.2374, "step": 7779 }, { "epoch": 0.8278357097254735, "grad_norm": 12.1171875, "learning_rate": 9.741330301307563e-06, "loss": 1.0993, "step": 7780 }, { "epoch": 0.8279421153436901, "grad_norm": 15.8203125, "learning_rate": 9.741294769755544e-06, "loss": 1.4609, "step": 7781 }, { "epoch": 0.8280485209619068, "grad_norm": 12.359375, "learning_rate": 9.741259238203526e-06, "loss": 1.4171, "step": 7782 }, { "epoch": 0.8281549265801235, "grad_norm": 15.40625, "learning_rate": 9.741223706651507e-06, "loss": 0.9996, "step": 7783 }, { "epoch": 0.8282613321983401, "grad_norm": 13.1796875, "learning_rate": 9.741188175099489e-06, "loss": 1.373, "step": 7784 }, { "epoch": 0.8283677378165567, "grad_norm": 18.34375, "learning_rate": 9.74115264354747e-06, "loss": 1.3832, "step": 7785 }, { "epoch": 0.8284741434347733, "grad_norm": 10.9140625, "learning_rate": 9.741117111995452e-06, "loss": 0.977, "step": 7786 }, { "epoch": 0.82858054905299, "grad_norm": 12.7265625, "learning_rate": 9.741081580443434e-06, "loss": 1.4018, "step": 7787 }, { "epoch": 0.8286869546712067, "grad_norm": 11.0, "learning_rate": 9.741046048891416e-06, "loss": 1.239, "step": 7788 }, { "epoch": 0.8287933602894233, "grad_norm": 12.2578125, "learning_rate": 9.741010517339399e-06, "loss": 1.2721, "step": 7789 }, { "epoch": 0.8288997659076399, "grad_norm": 13.53125, "learning_rate": 9.74097498578738e-06, "loss": 1.4772, "step": 7790 }, { "epoch": 0.8290061715258565, "grad_norm": 13.4609375, "learning_rate": 9.740939454235362e-06, "loss": 1.3147, "step": 7791 }, { "epoch": 0.8291125771440732, "grad_norm": 16.625, "learning_rate": 9.740903922683344e-06, "loss": 1.3561, "step": 7792 }, { "epoch": 0.8292189827622899, "grad_norm": 14.625, "learning_rate": 9.740868391131325e-06, "loss": 1.2081, "step": 7793 }, { "epoch": 0.8293253883805065, "grad_norm": 13.359375, "learning_rate": 9.740832859579307e-06, "loss": 1.2782, "step": 7794 }, { "epoch": 0.8294317939987231, "grad_norm": 16.53125, "learning_rate": 9.74079732802729e-06, "loss": 1.2772, "step": 7795 }, { "epoch": 0.8295381996169398, "grad_norm": 13.1171875, "learning_rate": 9.740761796475272e-06, "loss": 1.1206, "step": 7796 }, { "epoch": 0.8296446052351564, "grad_norm": 21.75, "learning_rate": 9.740726264923253e-06, "loss": 0.9386, "step": 7797 }, { "epoch": 0.8297510108533731, "grad_norm": 14.9765625, "learning_rate": 9.740690733371235e-06, "loss": 1.1342, "step": 7798 }, { "epoch": 0.8298574164715897, "grad_norm": 12.96875, "learning_rate": 9.740655201819217e-06, "loss": 1.3454, "step": 7799 }, { "epoch": 0.8299638220898063, "grad_norm": 12.6328125, "learning_rate": 9.740619670267198e-06, "loss": 1.3414, "step": 7800 }, { "epoch": 0.830070227708023, "grad_norm": 12.1875, "learning_rate": 9.74058413871518e-06, "loss": 1.3595, "step": 7801 }, { "epoch": 0.8301766333262396, "grad_norm": 10.0546875, "learning_rate": 9.740548607163161e-06, "loss": 1.1966, "step": 7802 }, { "epoch": 0.8302830389444563, "grad_norm": 17.015625, "learning_rate": 9.740513075611143e-06, "loss": 1.1336, "step": 7803 }, { "epoch": 0.8303894445626729, "grad_norm": 18.53125, "learning_rate": 9.740477544059125e-06, "loss": 1.3189, "step": 7804 }, { "epoch": 0.8304958501808896, "grad_norm": 15.09375, "learning_rate": 9.740442012507108e-06, "loss": 1.0731, "step": 7805 }, { "epoch": 0.8306022557991062, "grad_norm": 10.2265625, "learning_rate": 9.74040648095509e-06, "loss": 1.1712, "step": 7806 }, { "epoch": 0.8307086614173228, "grad_norm": 14.1015625, "learning_rate": 9.740370949403071e-06, "loss": 1.0752, "step": 7807 }, { "epoch": 0.8308150670355394, "grad_norm": 13.2421875, "learning_rate": 9.740335417851053e-06, "loss": 1.1795, "step": 7808 }, { "epoch": 0.8309214726537562, "grad_norm": 10.0546875, "learning_rate": 9.740299886299034e-06, "loss": 1.1953, "step": 7809 }, { "epoch": 0.8310278782719728, "grad_norm": 10.6328125, "learning_rate": 9.740264354747016e-06, "loss": 1.0749, "step": 7810 }, { "epoch": 0.8311342838901894, "grad_norm": 11.875, "learning_rate": 9.740228823194998e-06, "loss": 1.0542, "step": 7811 }, { "epoch": 0.831240689508406, "grad_norm": 12.1328125, "learning_rate": 9.74019329164298e-06, "loss": 1.1905, "step": 7812 }, { "epoch": 0.8313470951266226, "grad_norm": 13.9140625, "learning_rate": 9.74015776009096e-06, "loss": 1.3473, "step": 7813 }, { "epoch": 0.8314535007448394, "grad_norm": 11.75, "learning_rate": 9.740122228538942e-06, "loss": 1.2493, "step": 7814 }, { "epoch": 0.831559906363056, "grad_norm": 10.4140625, "learning_rate": 9.740086696986926e-06, "loss": 1.1957, "step": 7815 }, { "epoch": 0.8316663119812726, "grad_norm": 11.140625, "learning_rate": 9.740051165434907e-06, "loss": 1.1106, "step": 7816 }, { "epoch": 0.8317727175994892, "grad_norm": 11.78125, "learning_rate": 9.740015633882889e-06, "loss": 1.2357, "step": 7817 }, { "epoch": 0.8318791232177059, "grad_norm": 18.09375, "learning_rate": 9.73998010233087e-06, "loss": 1.3761, "step": 7818 }, { "epoch": 0.8319855288359226, "grad_norm": 12.4765625, "learning_rate": 9.739944570778852e-06, "loss": 1.5722, "step": 7819 }, { "epoch": 0.8320919344541392, "grad_norm": 14.8359375, "learning_rate": 9.739909039226835e-06, "loss": 0.9971, "step": 7820 }, { "epoch": 0.8321983400723558, "grad_norm": 10.40625, "learning_rate": 9.739873507674817e-06, "loss": 0.9781, "step": 7821 }, { "epoch": 0.8323047456905724, "grad_norm": 10.6640625, "learning_rate": 9.739837976122799e-06, "loss": 1.0432, "step": 7822 }, { "epoch": 0.8324111513087891, "grad_norm": 11.9296875, "learning_rate": 9.73980244457078e-06, "loss": 1.1359, "step": 7823 }, { "epoch": 0.8325175569270058, "grad_norm": 11.453125, "learning_rate": 9.739766913018762e-06, "loss": 1.2169, "step": 7824 }, { "epoch": 0.8326239625452224, "grad_norm": 9.7890625, "learning_rate": 9.739731381466743e-06, "loss": 1.017, "step": 7825 }, { "epoch": 0.832730368163439, "grad_norm": 12.7265625, "learning_rate": 9.739695849914725e-06, "loss": 1.1688, "step": 7826 }, { "epoch": 0.8328367737816557, "grad_norm": 12.6015625, "learning_rate": 9.739660318362707e-06, "loss": 1.2596, "step": 7827 }, { "epoch": 0.8329431793998723, "grad_norm": 19.421875, "learning_rate": 9.739624786810688e-06, "loss": 1.3729, "step": 7828 }, { "epoch": 0.8330495850180889, "grad_norm": 14.9453125, "learning_rate": 9.73958925525867e-06, "loss": 1.2847, "step": 7829 }, { "epoch": 0.8331559906363056, "grad_norm": 14.296875, "learning_rate": 9.739553723706651e-06, "loss": 1.1622, "step": 7830 }, { "epoch": 0.8332623962545223, "grad_norm": 12.6484375, "learning_rate": 9.739518192154635e-06, "loss": 1.2818, "step": 7831 }, { "epoch": 0.8333688018727389, "grad_norm": 12.0546875, "learning_rate": 9.739482660602616e-06, "loss": 1.0338, "step": 7832 }, { "epoch": 0.8334752074909555, "grad_norm": 14.6875, "learning_rate": 9.739447129050598e-06, "loss": 1.5492, "step": 7833 }, { "epoch": 0.8335816131091721, "grad_norm": 10.390625, "learning_rate": 9.73941159749858e-06, "loss": 0.965, "step": 7834 }, { "epoch": 0.8336880187273888, "grad_norm": 13.8515625, "learning_rate": 9.739376065946561e-06, "loss": 1.4149, "step": 7835 }, { "epoch": 0.8337944243456055, "grad_norm": 9.4921875, "learning_rate": 9.739340534394543e-06, "loss": 1.4349, "step": 7836 }, { "epoch": 0.8339008299638221, "grad_norm": 13.9140625, "learning_rate": 9.739305002842524e-06, "loss": 1.4427, "step": 7837 }, { "epoch": 0.8340072355820387, "grad_norm": 14.8046875, "learning_rate": 9.739269471290506e-06, "loss": 1.1255, "step": 7838 }, { "epoch": 0.8341136412002553, "grad_norm": 10.953125, "learning_rate": 9.739233939738488e-06, "loss": 1.6221, "step": 7839 }, { "epoch": 0.8342200468184721, "grad_norm": 12.78125, "learning_rate": 9.73919840818647e-06, "loss": 1.1799, "step": 7840 }, { "epoch": 0.8343264524366887, "grad_norm": 13.125, "learning_rate": 9.739162876634453e-06, "loss": 1.1789, "step": 7841 }, { "epoch": 0.8344328580549053, "grad_norm": 10.71875, "learning_rate": 9.739127345082434e-06, "loss": 1.0257, "step": 7842 }, { "epoch": 0.8345392636731219, "grad_norm": 12.2109375, "learning_rate": 9.739091813530416e-06, "loss": 1.1699, "step": 7843 }, { "epoch": 0.8346456692913385, "grad_norm": 12.09375, "learning_rate": 9.739056281978397e-06, "loss": 0.9504, "step": 7844 }, { "epoch": 0.8347520749095553, "grad_norm": 14.6875, "learning_rate": 9.739020750426379e-06, "loss": 1.5359, "step": 7845 }, { "epoch": 0.8348584805277719, "grad_norm": 16.203125, "learning_rate": 9.738985218874362e-06, "loss": 1.1002, "step": 7846 }, { "epoch": 0.8349648861459885, "grad_norm": 15.5625, "learning_rate": 9.738949687322344e-06, "loss": 1.1986, "step": 7847 }, { "epoch": 0.8350712917642051, "grad_norm": 11.6484375, "learning_rate": 9.738914155770325e-06, "loss": 1.1549, "step": 7848 }, { "epoch": 0.8351776973824218, "grad_norm": 12.015625, "learning_rate": 9.738878624218307e-06, "loss": 1.3589, "step": 7849 }, { "epoch": 0.8352841030006384, "grad_norm": 15.59375, "learning_rate": 9.738843092666289e-06, "loss": 1.1639, "step": 7850 }, { "epoch": 0.8353905086188551, "grad_norm": 12.390625, "learning_rate": 9.73880756111427e-06, "loss": 1.1559, "step": 7851 }, { "epoch": 0.8354969142370717, "grad_norm": 16.09375, "learning_rate": 9.738772029562252e-06, "loss": 1.2316, "step": 7852 }, { "epoch": 0.8356033198552884, "grad_norm": 14.9140625, "learning_rate": 9.738736498010234e-06, "loss": 1.0732, "step": 7853 }, { "epoch": 0.835709725473505, "grad_norm": 10.9375, "learning_rate": 9.738700966458215e-06, "loss": 1.1292, "step": 7854 }, { "epoch": 0.8358161310917216, "grad_norm": 13.7109375, "learning_rate": 9.738665434906197e-06, "loss": 1.2563, "step": 7855 }, { "epoch": 0.8359225367099383, "grad_norm": 13.234375, "learning_rate": 9.73862990335418e-06, "loss": 1.2851, "step": 7856 }, { "epoch": 0.836028942328155, "grad_norm": 12.203125, "learning_rate": 9.738594371802162e-06, "loss": 1.1436, "step": 7857 }, { "epoch": 0.8361353479463716, "grad_norm": 10.8828125, "learning_rate": 9.738558840250143e-06, "loss": 1.1019, "step": 7858 }, { "epoch": 0.8362417535645882, "grad_norm": 8.0546875, "learning_rate": 9.738523308698125e-06, "loss": 1.0558, "step": 7859 }, { "epoch": 0.8363481591828048, "grad_norm": 13.1484375, "learning_rate": 9.738487777146106e-06, "loss": 1.4978, "step": 7860 }, { "epoch": 0.8364545648010215, "grad_norm": 11.4765625, "learning_rate": 9.738452245594088e-06, "loss": 1.1833, "step": 7861 }, { "epoch": 0.8365609704192382, "grad_norm": 17.375, "learning_rate": 9.73841671404207e-06, "loss": 1.194, "step": 7862 }, { "epoch": 0.8366673760374548, "grad_norm": 12.859375, "learning_rate": 9.738381182490051e-06, "loss": 1.1682, "step": 7863 }, { "epoch": 0.8367737816556714, "grad_norm": 10.625, "learning_rate": 9.738345650938033e-06, "loss": 1.1395, "step": 7864 }, { "epoch": 0.836880187273888, "grad_norm": 14.0234375, "learning_rate": 9.738310119386014e-06, "loss": 0.9234, "step": 7865 }, { "epoch": 0.8369865928921048, "grad_norm": 12.359375, "learning_rate": 9.738274587833998e-06, "loss": 1.3086, "step": 7866 }, { "epoch": 0.8370929985103214, "grad_norm": 11.5703125, "learning_rate": 9.73823905628198e-06, "loss": 1.1085, "step": 7867 }, { "epoch": 0.837199404128538, "grad_norm": 9.7890625, "learning_rate": 9.738203524729961e-06, "loss": 0.9763, "step": 7868 }, { "epoch": 0.8373058097467546, "grad_norm": 12.140625, "learning_rate": 9.738167993177943e-06, "loss": 1.0689, "step": 7869 }, { "epoch": 0.8374122153649712, "grad_norm": 10.2421875, "learning_rate": 9.738132461625924e-06, "loss": 1.2137, "step": 7870 }, { "epoch": 0.837518620983188, "grad_norm": 9.6015625, "learning_rate": 9.738096930073906e-06, "loss": 1.1679, "step": 7871 }, { "epoch": 0.8376250266014046, "grad_norm": 18.515625, "learning_rate": 9.738061398521889e-06, "loss": 1.2817, "step": 7872 }, { "epoch": 0.8377314322196212, "grad_norm": 12.0546875, "learning_rate": 9.73802586696987e-06, "loss": 1.1606, "step": 7873 }, { "epoch": 0.8378378378378378, "grad_norm": 16.234375, "learning_rate": 9.737990335417852e-06, "loss": 1.5653, "step": 7874 }, { "epoch": 0.8379442434560544, "grad_norm": 12.1640625, "learning_rate": 9.737954803865834e-06, "loss": 1.0973, "step": 7875 }, { "epoch": 0.8380506490742711, "grad_norm": 13.28125, "learning_rate": 9.737919272313816e-06, "loss": 1.1993, "step": 7876 }, { "epoch": 0.8381570546924878, "grad_norm": 13.3125, "learning_rate": 9.737883740761797e-06, "loss": 1.503, "step": 7877 }, { "epoch": 0.8382634603107044, "grad_norm": 18.375, "learning_rate": 9.737848209209779e-06, "loss": 1.2871, "step": 7878 }, { "epoch": 0.838369865928921, "grad_norm": 12.46875, "learning_rate": 9.73781267765776e-06, "loss": 1.272, "step": 7879 }, { "epoch": 0.8384762715471377, "grad_norm": 12.9609375, "learning_rate": 9.737777146105742e-06, "loss": 1.0302, "step": 7880 }, { "epoch": 0.8385826771653543, "grad_norm": 12.671875, "learning_rate": 9.737741614553724e-06, "loss": 1.0942, "step": 7881 }, { "epoch": 0.838689082783571, "grad_norm": 14.3125, "learning_rate": 9.737706083001707e-06, "loss": 1.1388, "step": 7882 }, { "epoch": 0.8387954884017876, "grad_norm": 13.0, "learning_rate": 9.737670551449688e-06, "loss": 1.1631, "step": 7883 }, { "epoch": 0.8389018940200043, "grad_norm": 12.8203125, "learning_rate": 9.73763501989767e-06, "loss": 1.5198, "step": 7884 }, { "epoch": 0.8390082996382209, "grad_norm": 14.671875, "learning_rate": 9.737599488345652e-06, "loss": 1.423, "step": 7885 }, { "epoch": 0.8391147052564375, "grad_norm": 12.421875, "learning_rate": 9.737563956793633e-06, "loss": 1.3369, "step": 7886 }, { "epoch": 0.8392211108746542, "grad_norm": 13.890625, "learning_rate": 9.737528425241617e-06, "loss": 1.4737, "step": 7887 }, { "epoch": 0.8393275164928709, "grad_norm": 15.7421875, "learning_rate": 9.737492893689598e-06, "loss": 1.1278, "step": 7888 }, { "epoch": 0.8394339221110875, "grad_norm": 11.65625, "learning_rate": 9.73745736213758e-06, "loss": 1.0917, "step": 7889 }, { "epoch": 0.8395403277293041, "grad_norm": 12.4296875, "learning_rate": 9.73742183058556e-06, "loss": 1.0933, "step": 7890 }, { "epoch": 0.8396467333475207, "grad_norm": 12.9375, "learning_rate": 9.737386299033541e-06, "loss": 1.4854, "step": 7891 }, { "epoch": 0.8397531389657374, "grad_norm": 11.4296875, "learning_rate": 9.737350767481525e-06, "loss": 1.2316, "step": 7892 }, { "epoch": 0.8398595445839541, "grad_norm": 10.1875, "learning_rate": 9.737315235929506e-06, "loss": 1.1713, "step": 7893 }, { "epoch": 0.8399659502021707, "grad_norm": 11.1875, "learning_rate": 9.737279704377488e-06, "loss": 1.0687, "step": 7894 }, { "epoch": 0.8400723558203873, "grad_norm": 10.953125, "learning_rate": 9.73724417282547e-06, "loss": 1.4978, "step": 7895 }, { "epoch": 0.8401787614386039, "grad_norm": 10.6875, "learning_rate": 9.737208641273451e-06, "loss": 0.9866, "step": 7896 }, { "epoch": 0.8402851670568205, "grad_norm": 13.96875, "learning_rate": 9.737173109721434e-06, "loss": 1.1372, "step": 7897 }, { "epoch": 0.8403915726750373, "grad_norm": 18.109375, "learning_rate": 9.737137578169416e-06, "loss": 1.5623, "step": 7898 }, { "epoch": 0.8404979782932539, "grad_norm": 10.234375, "learning_rate": 9.737102046617398e-06, "loss": 1.2633, "step": 7899 }, { "epoch": 0.8406043839114705, "grad_norm": 15.3984375, "learning_rate": 9.73706651506538e-06, "loss": 1.3055, "step": 7900 }, { "epoch": 0.8407107895296871, "grad_norm": 16.078125, "learning_rate": 9.73703098351336e-06, "loss": 1.2973, "step": 7901 }, { "epoch": 0.8408171951479038, "grad_norm": 12.1796875, "learning_rate": 9.736995451961342e-06, "loss": 1.4598, "step": 7902 }, { "epoch": 0.8409236007661205, "grad_norm": 12.828125, "learning_rate": 9.736959920409324e-06, "loss": 1.4184, "step": 7903 }, { "epoch": 0.8410300063843371, "grad_norm": 14.5625, "learning_rate": 9.736924388857306e-06, "loss": 1.4556, "step": 7904 }, { "epoch": 0.8411364120025537, "grad_norm": 9.3359375, "learning_rate": 9.736888857305287e-06, "loss": 1.1107, "step": 7905 }, { "epoch": 0.8412428176207704, "grad_norm": 9.0390625, "learning_rate": 9.736853325753269e-06, "loss": 1.1737, "step": 7906 }, { "epoch": 0.841349223238987, "grad_norm": 9.953125, "learning_rate": 9.736817794201252e-06, "loss": 1.0498, "step": 7907 }, { "epoch": 0.8414556288572037, "grad_norm": 16.296875, "learning_rate": 9.736782262649234e-06, "loss": 1.1268, "step": 7908 }, { "epoch": 0.8415620344754203, "grad_norm": 16.34375, "learning_rate": 9.736746731097215e-06, "loss": 1.3557, "step": 7909 }, { "epoch": 0.841668440093637, "grad_norm": 12.1171875, "learning_rate": 9.736711199545197e-06, "loss": 1.3036, "step": 7910 }, { "epoch": 0.8417748457118536, "grad_norm": 20.5625, "learning_rate": 9.736675667993179e-06, "loss": 1.598, "step": 7911 }, { "epoch": 0.8418812513300702, "grad_norm": 10.046875, "learning_rate": 9.73664013644116e-06, "loss": 1.1594, "step": 7912 }, { "epoch": 0.8419876569482869, "grad_norm": 13.1953125, "learning_rate": 9.736604604889143e-06, "loss": 1.2643, "step": 7913 }, { "epoch": 0.8420940625665035, "grad_norm": 12.046875, "learning_rate": 9.736569073337125e-06, "loss": 1.1876, "step": 7914 }, { "epoch": 0.8422004681847202, "grad_norm": 16.625, "learning_rate": 9.736533541785107e-06, "loss": 1.1732, "step": 7915 }, { "epoch": 0.8423068738029368, "grad_norm": 9.234375, "learning_rate": 9.736498010233088e-06, "loss": 1.1384, "step": 7916 }, { "epoch": 0.8424132794211534, "grad_norm": 11.8515625, "learning_rate": 9.73646247868107e-06, "loss": 0.95, "step": 7917 }, { "epoch": 0.84251968503937, "grad_norm": 11.4140625, "learning_rate": 9.736426947129051e-06, "loss": 1.1532, "step": 7918 }, { "epoch": 0.8426260906575868, "grad_norm": 10.2421875, "learning_rate": 9.736391415577033e-06, "loss": 1.0277, "step": 7919 }, { "epoch": 0.8427324962758034, "grad_norm": 14.03125, "learning_rate": 9.736355884025015e-06, "loss": 1.4041, "step": 7920 }, { "epoch": 0.84283890189402, "grad_norm": 13.3125, "learning_rate": 9.736320352472996e-06, "loss": 1.0277, "step": 7921 }, { "epoch": 0.8429453075122366, "grad_norm": 11.109375, "learning_rate": 9.736284820920978e-06, "loss": 1.0622, "step": 7922 }, { "epoch": 0.8430517131304532, "grad_norm": 12.8515625, "learning_rate": 9.736249289368961e-06, "loss": 1.6867, "step": 7923 }, { "epoch": 0.84315811874867, "grad_norm": 16.546875, "learning_rate": 9.736213757816943e-06, "loss": 1.2881, "step": 7924 }, { "epoch": 0.8432645243668866, "grad_norm": 11.3203125, "learning_rate": 9.736178226264924e-06, "loss": 0.9789, "step": 7925 }, { "epoch": 0.8433709299851032, "grad_norm": 11.484375, "learning_rate": 9.736142694712906e-06, "loss": 1.0162, "step": 7926 }, { "epoch": 0.8434773356033198, "grad_norm": 12.2265625, "learning_rate": 9.736107163160888e-06, "loss": 1.2958, "step": 7927 }, { "epoch": 0.8435837412215365, "grad_norm": 27.671875, "learning_rate": 9.73607163160887e-06, "loss": 1.4291, "step": 7928 }, { "epoch": 0.8436901468397532, "grad_norm": 15.9140625, "learning_rate": 9.736036100056851e-06, "loss": 1.2159, "step": 7929 }, { "epoch": 0.8437965524579698, "grad_norm": 25.546875, "learning_rate": 9.736000568504832e-06, "loss": 1.4257, "step": 7930 }, { "epoch": 0.8439029580761864, "grad_norm": 10.3828125, "learning_rate": 9.735965036952814e-06, "loss": 1.3803, "step": 7931 }, { "epoch": 0.844009363694403, "grad_norm": 11.25, "learning_rate": 9.735929505400796e-06, "loss": 1.027, "step": 7932 }, { "epoch": 0.8441157693126197, "grad_norm": 15.6640625, "learning_rate": 9.735893973848779e-06, "loss": 1.3084, "step": 7933 }, { "epoch": 0.8442221749308364, "grad_norm": 16.25, "learning_rate": 9.73585844229676e-06, "loss": 1.4891, "step": 7934 }, { "epoch": 0.844328580549053, "grad_norm": 11.734375, "learning_rate": 9.735822910744742e-06, "loss": 1.1768, "step": 7935 }, { "epoch": 0.8444349861672696, "grad_norm": 9.9453125, "learning_rate": 9.735787379192724e-06, "loss": 1.1866, "step": 7936 }, { "epoch": 0.8445413917854863, "grad_norm": 16.703125, "learning_rate": 9.735751847640705e-06, "loss": 1.4324, "step": 7937 }, { "epoch": 0.8446477974037029, "grad_norm": 12.8203125, "learning_rate": 9.735716316088689e-06, "loss": 1.3954, "step": 7938 }, { "epoch": 0.8447542030219195, "grad_norm": 11.3203125, "learning_rate": 9.73568078453667e-06, "loss": 1.3967, "step": 7939 }, { "epoch": 0.8448606086401362, "grad_norm": 9.0390625, "learning_rate": 9.735645252984652e-06, "loss": 1.1071, "step": 7940 }, { "epoch": 0.8449670142583529, "grad_norm": 15.2734375, "learning_rate": 9.735609721432634e-06, "loss": 1.0563, "step": 7941 }, { "epoch": 0.8450734198765695, "grad_norm": 11.953125, "learning_rate": 9.735574189880615e-06, "loss": 1.1016, "step": 7942 }, { "epoch": 0.8451798254947861, "grad_norm": 22.0625, "learning_rate": 9.735538658328597e-06, "loss": 1.1584, "step": 7943 }, { "epoch": 0.8452862311130027, "grad_norm": 16.796875, "learning_rate": 9.735503126776578e-06, "loss": 1.2931, "step": 7944 }, { "epoch": 0.8453926367312194, "grad_norm": 12.2109375, "learning_rate": 9.73546759522456e-06, "loss": 1.4946, "step": 7945 }, { "epoch": 0.8454990423494361, "grad_norm": 11.3984375, "learning_rate": 9.735432063672542e-06, "loss": 1.5277, "step": 7946 }, { "epoch": 0.8456054479676527, "grad_norm": 14.3203125, "learning_rate": 9.735396532120523e-06, "loss": 1.1888, "step": 7947 }, { "epoch": 0.8457118535858693, "grad_norm": 11.6875, "learning_rate": 9.735361000568505e-06, "loss": 0.985, "step": 7948 }, { "epoch": 0.8458182592040859, "grad_norm": 11.984375, "learning_rate": 9.735325469016488e-06, "loss": 1.6139, "step": 7949 }, { "epoch": 0.8459246648223027, "grad_norm": 16.671875, "learning_rate": 9.73528993746447e-06, "loss": 1.0832, "step": 7950 }, { "epoch": 0.8460310704405193, "grad_norm": 12.015625, "learning_rate": 9.735254405912451e-06, "loss": 1.1622, "step": 7951 }, { "epoch": 0.8461374760587359, "grad_norm": 17.203125, "learning_rate": 9.735218874360433e-06, "loss": 1.0659, "step": 7952 }, { "epoch": 0.8462438816769525, "grad_norm": 16.4375, "learning_rate": 9.735183342808415e-06, "loss": 1.6789, "step": 7953 }, { "epoch": 0.8463502872951691, "grad_norm": 15.5, "learning_rate": 9.735147811256396e-06, "loss": 0.9209, "step": 7954 }, { "epoch": 0.8464566929133859, "grad_norm": 12.6875, "learning_rate": 9.735112279704378e-06, "loss": 1.3519, "step": 7955 }, { "epoch": 0.8465630985316025, "grad_norm": 14.484375, "learning_rate": 9.73507674815236e-06, "loss": 1.0839, "step": 7956 }, { "epoch": 0.8466695041498191, "grad_norm": 9.6171875, "learning_rate": 9.735041216600341e-06, "loss": 0.9246, "step": 7957 }, { "epoch": 0.8467759097680357, "grad_norm": 11.90625, "learning_rate": 9.735005685048323e-06, "loss": 1.0571, "step": 7958 }, { "epoch": 0.8468823153862524, "grad_norm": 13.6015625, "learning_rate": 9.734970153496306e-06, "loss": 1.2495, "step": 7959 }, { "epoch": 0.846988721004469, "grad_norm": 9.6484375, "learning_rate": 9.734934621944287e-06, "loss": 1.3112, "step": 7960 }, { "epoch": 0.8470951266226857, "grad_norm": 12.3671875, "learning_rate": 9.734899090392269e-06, "loss": 1.2628, "step": 7961 }, { "epoch": 0.8472015322409023, "grad_norm": 15.6796875, "learning_rate": 9.73486355884025e-06, "loss": 1.1731, "step": 7962 }, { "epoch": 0.847307937859119, "grad_norm": 18.578125, "learning_rate": 9.734828027288232e-06, "loss": 1.2077, "step": 7963 }, { "epoch": 0.8474143434773356, "grad_norm": 13.109375, "learning_rate": 9.734792495736216e-06, "loss": 1.3499, "step": 7964 }, { "epoch": 0.8475207490955522, "grad_norm": 11.546875, "learning_rate": 9.734756964184197e-06, "loss": 1.2395, "step": 7965 }, { "epoch": 0.8476271547137689, "grad_norm": 12.1640625, "learning_rate": 9.734721432632179e-06, "loss": 1.2451, "step": 7966 }, { "epoch": 0.8477335603319855, "grad_norm": 18.921875, "learning_rate": 9.73468590108016e-06, "loss": 1.3255, "step": 7967 }, { "epoch": 0.8478399659502022, "grad_norm": 11.8515625, "learning_rate": 9.734650369528142e-06, "loss": 1.3789, "step": 7968 }, { "epoch": 0.8479463715684188, "grad_norm": 14.828125, "learning_rate": 9.734614837976124e-06, "loss": 1.6152, "step": 7969 }, { "epoch": 0.8480527771866354, "grad_norm": 14.0546875, "learning_rate": 9.734579306424105e-06, "loss": 0.8977, "step": 7970 }, { "epoch": 0.8481591828048521, "grad_norm": 13.453125, "learning_rate": 9.734543774872087e-06, "loss": 1.0927, "step": 7971 }, { "epoch": 0.8482655884230688, "grad_norm": 13.515625, "learning_rate": 9.734508243320068e-06, "loss": 1.2402, "step": 7972 }, { "epoch": 0.8483719940412854, "grad_norm": 10.3046875, "learning_rate": 9.73447271176805e-06, "loss": 1.4351, "step": 7973 }, { "epoch": 0.848478399659502, "grad_norm": 17.765625, "learning_rate": 9.734437180216033e-06, "loss": 1.0398, "step": 7974 }, { "epoch": 0.8485848052777186, "grad_norm": 11.75, "learning_rate": 9.734401648664015e-06, "loss": 1.3134, "step": 7975 }, { "epoch": 0.8486912108959354, "grad_norm": 17.125, "learning_rate": 9.734366117111997e-06, "loss": 1.3161, "step": 7976 }, { "epoch": 0.848797616514152, "grad_norm": 11.296875, "learning_rate": 9.734330585559978e-06, "loss": 1.1153, "step": 7977 }, { "epoch": 0.8489040221323686, "grad_norm": 10.109375, "learning_rate": 9.73429505400796e-06, "loss": 1.3946, "step": 7978 }, { "epoch": 0.8490104277505852, "grad_norm": 10.5546875, "learning_rate": 9.734259522455941e-06, "loss": 1.6623, "step": 7979 }, { "epoch": 0.8491168333688018, "grad_norm": 11.15625, "learning_rate": 9.734223990903923e-06, "loss": 1.3223, "step": 7980 }, { "epoch": 0.8492232389870186, "grad_norm": 11.9921875, "learning_rate": 9.734188459351905e-06, "loss": 1.3873, "step": 7981 }, { "epoch": 0.8493296446052352, "grad_norm": 17.0, "learning_rate": 9.734152927799886e-06, "loss": 0.8991, "step": 7982 }, { "epoch": 0.8494360502234518, "grad_norm": 13.953125, "learning_rate": 9.734117396247868e-06, "loss": 1.0958, "step": 7983 }, { "epoch": 0.8495424558416684, "grad_norm": 15.09375, "learning_rate": 9.734081864695851e-06, "loss": 1.3524, "step": 7984 }, { "epoch": 0.849648861459885, "grad_norm": 16.65625, "learning_rate": 9.734046333143833e-06, "loss": 1.4519, "step": 7985 }, { "epoch": 0.8497552670781017, "grad_norm": 16.34375, "learning_rate": 9.734010801591814e-06, "loss": 1.3462, "step": 7986 }, { "epoch": 0.8498616726963184, "grad_norm": 14.75, "learning_rate": 9.733975270039796e-06, "loss": 1.2457, "step": 7987 }, { "epoch": 0.849968078314535, "grad_norm": 13.5390625, "learning_rate": 9.733939738487778e-06, "loss": 1.0694, "step": 7988 }, { "epoch": 0.8500744839327516, "grad_norm": 13.59375, "learning_rate": 9.733904206935759e-06, "loss": 1.5519, "step": 7989 }, { "epoch": 0.8501808895509683, "grad_norm": 11.3203125, "learning_rate": 9.733868675383742e-06, "loss": 1.4194, "step": 7990 }, { "epoch": 0.8502872951691849, "grad_norm": 11.5703125, "learning_rate": 9.733833143831724e-06, "loss": 1.4362, "step": 7991 }, { "epoch": 0.8503937007874016, "grad_norm": 13.2421875, "learning_rate": 9.733797612279706e-06, "loss": 1.2073, "step": 7992 }, { "epoch": 0.8505001064056182, "grad_norm": 20.0625, "learning_rate": 9.733762080727687e-06, "loss": 0.968, "step": 7993 }, { "epoch": 0.8506065120238349, "grad_norm": 12.40625, "learning_rate": 9.733726549175669e-06, "loss": 1.1778, "step": 7994 }, { "epoch": 0.8507129176420515, "grad_norm": 12.0703125, "learning_rate": 9.73369101762365e-06, "loss": 1.1388, "step": 7995 }, { "epoch": 0.8508193232602681, "grad_norm": 12.8671875, "learning_rate": 9.733655486071632e-06, "loss": 1.2079, "step": 7996 }, { "epoch": 0.8509257288784848, "grad_norm": 12.765625, "learning_rate": 9.733619954519614e-06, "loss": 1.0545, "step": 7997 }, { "epoch": 0.8510321344967015, "grad_norm": 11.1171875, "learning_rate": 9.733584422967595e-06, "loss": 1.2738, "step": 7998 }, { "epoch": 0.8511385401149181, "grad_norm": 10.171875, "learning_rate": 9.733548891415577e-06, "loss": 1.0373, "step": 7999 }, { "epoch": 0.8512449457331347, "grad_norm": 21.21875, "learning_rate": 9.73351335986356e-06, "loss": 1.3758, "step": 8000 }, { "epoch": 0.8513513513513513, "grad_norm": 9.1640625, "learning_rate": 9.733477828311542e-06, "loss": 0.9826, "step": 8001 }, { "epoch": 0.851457756969568, "grad_norm": 12.328125, "learning_rate": 9.733442296759523e-06, "loss": 1.4028, "step": 8002 }, { "epoch": 0.8515641625877847, "grad_norm": 12.875, "learning_rate": 9.733406765207505e-06, "loss": 1.4239, "step": 8003 }, { "epoch": 0.8516705682060013, "grad_norm": 12.515625, "learning_rate": 9.733371233655487e-06, "loss": 0.9678, "step": 8004 }, { "epoch": 0.8517769738242179, "grad_norm": 13.2734375, "learning_rate": 9.73333570210347e-06, "loss": 1.2105, "step": 8005 }, { "epoch": 0.8518833794424345, "grad_norm": 13.75, "learning_rate": 9.733300170551452e-06, "loss": 1.0914, "step": 8006 }, { "epoch": 0.8519897850606512, "grad_norm": 13.9609375, "learning_rate": 9.733264638999433e-06, "loss": 1.1003, "step": 8007 }, { "epoch": 0.8520961906788679, "grad_norm": 13.125, "learning_rate": 9.733229107447415e-06, "loss": 1.073, "step": 8008 }, { "epoch": 0.8522025962970845, "grad_norm": 13.171875, "learning_rate": 9.733193575895395e-06, "loss": 1.3505, "step": 8009 }, { "epoch": 0.8523090019153011, "grad_norm": 13.2578125, "learning_rate": 9.733158044343378e-06, "loss": 1.051, "step": 8010 }, { "epoch": 0.8524154075335177, "grad_norm": 12.6171875, "learning_rate": 9.73312251279136e-06, "loss": 1.18, "step": 8011 }, { "epoch": 0.8525218131517344, "grad_norm": 11.453125, "learning_rate": 9.733086981239341e-06, "loss": 1.3507, "step": 8012 }, { "epoch": 0.8526282187699511, "grad_norm": 15.375, "learning_rate": 9.733051449687323e-06, "loss": 1.097, "step": 8013 }, { "epoch": 0.8527346243881677, "grad_norm": 16.359375, "learning_rate": 9.733015918135304e-06, "loss": 1.0246, "step": 8014 }, { "epoch": 0.8528410300063843, "grad_norm": 16.328125, "learning_rate": 9.732980386583288e-06, "loss": 1.9442, "step": 8015 }, { "epoch": 0.852947435624601, "grad_norm": 9.7109375, "learning_rate": 9.73294485503127e-06, "loss": 1.2624, "step": 8016 }, { "epoch": 0.8530538412428176, "grad_norm": 10.6640625, "learning_rate": 9.732909323479251e-06, "loss": 1.1548, "step": 8017 }, { "epoch": 0.8531602468610343, "grad_norm": 10.734375, "learning_rate": 9.732873791927232e-06, "loss": 0.9749, "step": 8018 }, { "epoch": 0.8532666524792509, "grad_norm": 13.5546875, "learning_rate": 9.732838260375214e-06, "loss": 0.9641, "step": 8019 }, { "epoch": 0.8533730580974676, "grad_norm": 13.453125, "learning_rate": 9.732802728823196e-06, "loss": 1.1664, "step": 8020 }, { "epoch": 0.8534794637156842, "grad_norm": 14.3125, "learning_rate": 9.732767197271177e-06, "loss": 1.2631, "step": 8021 }, { "epoch": 0.8535858693339008, "grad_norm": 16.5625, "learning_rate": 9.732731665719159e-06, "loss": 1.2212, "step": 8022 }, { "epoch": 0.8536922749521175, "grad_norm": 9.1171875, "learning_rate": 9.73269613416714e-06, "loss": 1.0538, "step": 8023 }, { "epoch": 0.8537986805703341, "grad_norm": 14.84375, "learning_rate": 9.732660602615122e-06, "loss": 0.838, "step": 8024 }, { "epoch": 0.8539050861885508, "grad_norm": 10.65625, "learning_rate": 9.732625071063104e-06, "loss": 0.8881, "step": 8025 }, { "epoch": 0.8540114918067674, "grad_norm": 10.9296875, "learning_rate": 9.732589539511087e-06, "loss": 1.132, "step": 8026 }, { "epoch": 0.854117897424984, "grad_norm": 12.6015625, "learning_rate": 9.732554007959069e-06, "loss": 0.9752, "step": 8027 }, { "epoch": 0.8542243030432006, "grad_norm": 12.7265625, "learning_rate": 9.73251847640705e-06, "loss": 1.1125, "step": 8028 }, { "epoch": 0.8543307086614174, "grad_norm": 11.265625, "learning_rate": 9.732482944855032e-06, "loss": 1.1443, "step": 8029 }, { "epoch": 0.854437114279634, "grad_norm": 13.4375, "learning_rate": 9.732447413303013e-06, "loss": 1.0966, "step": 8030 }, { "epoch": 0.8545435198978506, "grad_norm": 15.3046875, "learning_rate": 9.732411881750997e-06, "loss": 0.9609, "step": 8031 }, { "epoch": 0.8546499255160672, "grad_norm": 10.734375, "learning_rate": 9.732376350198978e-06, "loss": 1.1905, "step": 8032 }, { "epoch": 0.8547563311342838, "grad_norm": 11.109375, "learning_rate": 9.73234081864696e-06, "loss": 1.3055, "step": 8033 }, { "epoch": 0.8548627367525006, "grad_norm": 12.6953125, "learning_rate": 9.732305287094942e-06, "loss": 1.4621, "step": 8034 }, { "epoch": 0.8549691423707172, "grad_norm": 9.3671875, "learning_rate": 9.732269755542923e-06, "loss": 0.9861, "step": 8035 }, { "epoch": 0.8550755479889338, "grad_norm": 15.84375, "learning_rate": 9.732234223990905e-06, "loss": 1.257, "step": 8036 }, { "epoch": 0.8551819536071504, "grad_norm": 10.953125, "learning_rate": 9.732198692438886e-06, "loss": 1.0471, "step": 8037 }, { "epoch": 0.8552883592253671, "grad_norm": 12.8515625, "learning_rate": 9.732163160886868e-06, "loss": 1.1647, "step": 8038 }, { "epoch": 0.8553947648435838, "grad_norm": 12.90625, "learning_rate": 9.73212762933485e-06, "loss": 0.9441, "step": 8039 }, { "epoch": 0.8555011704618004, "grad_norm": 10.40625, "learning_rate": 9.732092097782831e-06, "loss": 1.02, "step": 8040 }, { "epoch": 0.855607576080017, "grad_norm": 9.9375, "learning_rate": 9.732056566230815e-06, "loss": 0.8345, "step": 8041 }, { "epoch": 0.8557139816982336, "grad_norm": 11.671875, "learning_rate": 9.732021034678796e-06, "loss": 1.2433, "step": 8042 }, { "epoch": 0.8558203873164503, "grad_norm": 15.15625, "learning_rate": 9.731985503126778e-06, "loss": 1.2757, "step": 8043 }, { "epoch": 0.855926792934667, "grad_norm": 12.796875, "learning_rate": 9.73194997157476e-06, "loss": 1.4019, "step": 8044 }, { "epoch": 0.8560331985528836, "grad_norm": 12.4140625, "learning_rate": 9.731914440022741e-06, "loss": 1.4775, "step": 8045 }, { "epoch": 0.8561396041711002, "grad_norm": 14.9296875, "learning_rate": 9.731878908470723e-06, "loss": 1.2758, "step": 8046 }, { "epoch": 0.8562460097893169, "grad_norm": 11.9375, "learning_rate": 9.731843376918704e-06, "loss": 1.2498, "step": 8047 }, { "epoch": 0.8563524154075335, "grad_norm": 18.40625, "learning_rate": 9.731807845366686e-06, "loss": 1.4401, "step": 8048 }, { "epoch": 0.8564588210257501, "grad_norm": 14.0390625, "learning_rate": 9.731772313814667e-06, "loss": 1.1685, "step": 8049 }, { "epoch": 0.8565652266439668, "grad_norm": 12.8203125, "learning_rate": 9.731736782262649e-06, "loss": 1.4715, "step": 8050 }, { "epoch": 0.8566716322621835, "grad_norm": 13.3828125, "learning_rate": 9.731701250710632e-06, "loss": 1.2961, "step": 8051 }, { "epoch": 0.8567780378804001, "grad_norm": 13.078125, "learning_rate": 9.731665719158614e-06, "loss": 1.2104, "step": 8052 }, { "epoch": 0.8568844434986167, "grad_norm": 13.4140625, "learning_rate": 9.731630187606595e-06, "loss": 1.0457, "step": 8053 }, { "epoch": 0.8569908491168333, "grad_norm": 9.28125, "learning_rate": 9.731594656054577e-06, "loss": 1.141, "step": 8054 }, { "epoch": 0.85709725473505, "grad_norm": 15.6484375, "learning_rate": 9.731559124502559e-06, "loss": 1.5894, "step": 8055 }, { "epoch": 0.8572036603532667, "grad_norm": 11.796875, "learning_rate": 9.731523592950542e-06, "loss": 1.1108, "step": 8056 }, { "epoch": 0.8573100659714833, "grad_norm": 8.4140625, "learning_rate": 9.731488061398524e-06, "loss": 1.1579, "step": 8057 }, { "epoch": 0.8574164715896999, "grad_norm": 13.875, "learning_rate": 9.731452529846505e-06, "loss": 1.2135, "step": 8058 }, { "epoch": 0.8575228772079165, "grad_norm": 17.921875, "learning_rate": 9.731416998294487e-06, "loss": 1.2708, "step": 8059 }, { "epoch": 0.8576292828261333, "grad_norm": 19.90625, "learning_rate": 9.731381466742468e-06, "loss": 0.9746, "step": 8060 }, { "epoch": 0.8577356884443499, "grad_norm": 10.8828125, "learning_rate": 9.73134593519045e-06, "loss": 1.1398, "step": 8061 }, { "epoch": 0.8578420940625665, "grad_norm": 8.8828125, "learning_rate": 9.731310403638432e-06, "loss": 0.925, "step": 8062 }, { "epoch": 0.8579484996807831, "grad_norm": 13.140625, "learning_rate": 9.731274872086413e-06, "loss": 0.965, "step": 8063 }, { "epoch": 0.8580549052989997, "grad_norm": 13.6796875, "learning_rate": 9.731239340534395e-06, "loss": 1.1386, "step": 8064 }, { "epoch": 0.8581613109172165, "grad_norm": 13.5703125, "learning_rate": 9.731203808982376e-06, "loss": 1.225, "step": 8065 }, { "epoch": 0.8582677165354331, "grad_norm": 14.3359375, "learning_rate": 9.731168277430358e-06, "loss": 1.2954, "step": 8066 }, { "epoch": 0.8583741221536497, "grad_norm": 15.5625, "learning_rate": 9.731132745878341e-06, "loss": 1.6028, "step": 8067 }, { "epoch": 0.8584805277718663, "grad_norm": 13.5546875, "learning_rate": 9.731097214326323e-06, "loss": 1.3877, "step": 8068 }, { "epoch": 0.858586933390083, "grad_norm": 9.9921875, "learning_rate": 9.731061682774305e-06, "loss": 1.3183, "step": 8069 }, { "epoch": 0.8586933390082996, "grad_norm": 10.6875, "learning_rate": 9.731026151222286e-06, "loss": 1.5805, "step": 8070 }, { "epoch": 0.8587997446265163, "grad_norm": 13.8828125, "learning_rate": 9.730990619670268e-06, "loss": 1.2579, "step": 8071 }, { "epoch": 0.8589061502447329, "grad_norm": 12.546875, "learning_rate": 9.73095508811825e-06, "loss": 1.3856, "step": 8072 }, { "epoch": 0.8590125558629496, "grad_norm": 13.0234375, "learning_rate": 9.730919556566231e-06, "loss": 1.4189, "step": 8073 }, { "epoch": 0.8591189614811662, "grad_norm": 12.734375, "learning_rate": 9.730884025014213e-06, "loss": 1.2939, "step": 8074 }, { "epoch": 0.8592253670993828, "grad_norm": 13.78125, "learning_rate": 9.730848493462194e-06, "loss": 1.1989, "step": 8075 }, { "epoch": 0.8593317727175995, "grad_norm": 12.5078125, "learning_rate": 9.730812961910176e-06, "loss": 1.2917, "step": 8076 }, { "epoch": 0.8594381783358161, "grad_norm": 11.78125, "learning_rate": 9.730777430358159e-06, "loss": 1.0161, "step": 8077 }, { "epoch": 0.8595445839540328, "grad_norm": 12.375, "learning_rate": 9.73074189880614e-06, "loss": 1.582, "step": 8078 }, { "epoch": 0.8596509895722494, "grad_norm": 10.6640625, "learning_rate": 9.730706367254122e-06, "loss": 1.0896, "step": 8079 }, { "epoch": 0.859757395190466, "grad_norm": 9.515625, "learning_rate": 9.730670835702104e-06, "loss": 1.2221, "step": 8080 }, { "epoch": 0.8598638008086827, "grad_norm": 19.453125, "learning_rate": 9.730635304150086e-06, "loss": 1.1211, "step": 8081 }, { "epoch": 0.8599702064268994, "grad_norm": 15.984375, "learning_rate": 9.730599772598069e-06, "loss": 1.0064, "step": 8082 }, { "epoch": 0.860076612045116, "grad_norm": 11.7578125, "learning_rate": 9.73056424104605e-06, "loss": 1.4201, "step": 8083 }, { "epoch": 0.8601830176633326, "grad_norm": 14.484375, "learning_rate": 9.730528709494032e-06, "loss": 0.9531, "step": 8084 }, { "epoch": 0.8602894232815492, "grad_norm": 10.5703125, "learning_rate": 9.730493177942014e-06, "loss": 1.2592, "step": 8085 }, { "epoch": 0.860395828899766, "grad_norm": 15.328125, "learning_rate": 9.730457646389995e-06, "loss": 1.2217, "step": 8086 }, { "epoch": 0.8605022345179826, "grad_norm": 9.3671875, "learning_rate": 9.730422114837977e-06, "loss": 0.9974, "step": 8087 }, { "epoch": 0.8606086401361992, "grad_norm": 11.28125, "learning_rate": 9.730386583285959e-06, "loss": 1.3471, "step": 8088 }, { "epoch": 0.8607150457544158, "grad_norm": 11.9609375, "learning_rate": 9.73035105173394e-06, "loss": 1.2554, "step": 8089 }, { "epoch": 0.8608214513726324, "grad_norm": 11.5390625, "learning_rate": 9.730315520181922e-06, "loss": 1.0944, "step": 8090 }, { "epoch": 0.8609278569908492, "grad_norm": 12.8515625, "learning_rate": 9.730279988629903e-06, "loss": 1.4118, "step": 8091 }, { "epoch": 0.8610342626090658, "grad_norm": 16.203125, "learning_rate": 9.730244457077887e-06, "loss": 1.3933, "step": 8092 }, { "epoch": 0.8611406682272824, "grad_norm": 21.125, "learning_rate": 9.730208925525868e-06, "loss": 1.2285, "step": 8093 }, { "epoch": 0.861247073845499, "grad_norm": 16.71875, "learning_rate": 9.73017339397385e-06, "loss": 1.3376, "step": 8094 }, { "epoch": 0.8613534794637157, "grad_norm": 13.4140625, "learning_rate": 9.730137862421831e-06, "loss": 1.2039, "step": 8095 }, { "epoch": 0.8614598850819323, "grad_norm": 12.1484375, "learning_rate": 9.730102330869813e-06, "loss": 0.9865, "step": 8096 }, { "epoch": 0.861566290700149, "grad_norm": 10.5234375, "learning_rate": 9.730066799317796e-06, "loss": 0.9635, "step": 8097 }, { "epoch": 0.8616726963183656, "grad_norm": 14.15625, "learning_rate": 9.730031267765776e-06, "loss": 1.2493, "step": 8098 }, { "epoch": 0.8617791019365822, "grad_norm": 18.25, "learning_rate": 9.729995736213758e-06, "loss": 0.8708, "step": 8099 }, { "epoch": 0.8618855075547989, "grad_norm": 10.296875, "learning_rate": 9.72996020466174e-06, "loss": 1.0389, "step": 8100 }, { "epoch": 0.8619919131730155, "grad_norm": 10.0, "learning_rate": 9.729924673109721e-06, "loss": 1.1412, "step": 8101 }, { "epoch": 0.8620983187912322, "grad_norm": 15.828125, "learning_rate": 9.729889141557704e-06, "loss": 1.595, "step": 8102 }, { "epoch": 0.8622047244094488, "grad_norm": 13.03125, "learning_rate": 9.729853610005686e-06, "loss": 1.3929, "step": 8103 }, { "epoch": 0.8623111300276655, "grad_norm": 10.7265625, "learning_rate": 9.729818078453668e-06, "loss": 1.2644, "step": 8104 }, { "epoch": 0.8624175356458821, "grad_norm": 15.6875, "learning_rate": 9.72978254690165e-06, "loss": 1.2679, "step": 8105 }, { "epoch": 0.8625239412640987, "grad_norm": 15.828125, "learning_rate": 9.72974701534963e-06, "loss": 0.9576, "step": 8106 }, { "epoch": 0.8626303468823154, "grad_norm": 13.9140625, "learning_rate": 9.729711483797612e-06, "loss": 1.3911, "step": 8107 }, { "epoch": 0.8627367525005321, "grad_norm": 10.7421875, "learning_rate": 9.729675952245596e-06, "loss": 1.3527, "step": 8108 }, { "epoch": 0.8628431581187487, "grad_norm": 13.3671875, "learning_rate": 9.729640420693577e-06, "loss": 1.1025, "step": 8109 }, { "epoch": 0.8629495637369653, "grad_norm": 10.6328125, "learning_rate": 9.729604889141559e-06, "loss": 1.1487, "step": 8110 }, { "epoch": 0.8630559693551819, "grad_norm": 15.3671875, "learning_rate": 9.72956935758954e-06, "loss": 1.7646, "step": 8111 }, { "epoch": 0.8631623749733986, "grad_norm": 12.5625, "learning_rate": 9.729533826037522e-06, "loss": 1.4892, "step": 8112 }, { "epoch": 0.8632687805916153, "grad_norm": 11.0234375, "learning_rate": 9.729498294485504e-06, "loss": 1.0682, "step": 8113 }, { "epoch": 0.8633751862098319, "grad_norm": 14.328125, "learning_rate": 9.729462762933485e-06, "loss": 1.2852, "step": 8114 }, { "epoch": 0.8634815918280485, "grad_norm": 17.09375, "learning_rate": 9.729427231381467e-06, "loss": 1.2896, "step": 8115 }, { "epoch": 0.8635879974462651, "grad_norm": 13.53125, "learning_rate": 9.729391699829449e-06, "loss": 1.1976, "step": 8116 }, { "epoch": 0.8636944030644818, "grad_norm": 11.078125, "learning_rate": 9.72935616827743e-06, "loss": 1.3355, "step": 8117 }, { "epoch": 0.8638008086826985, "grad_norm": 15.9921875, "learning_rate": 9.729320636725413e-06, "loss": 1.1926, "step": 8118 }, { "epoch": 0.8639072143009151, "grad_norm": 14.28125, "learning_rate": 9.729285105173395e-06, "loss": 1.3134, "step": 8119 }, { "epoch": 0.8640136199191317, "grad_norm": 10.8828125, "learning_rate": 9.729249573621377e-06, "loss": 1.3418, "step": 8120 }, { "epoch": 0.8641200255373483, "grad_norm": 11.671875, "learning_rate": 9.729214042069358e-06, "loss": 1.3149, "step": 8121 }, { "epoch": 0.864226431155565, "grad_norm": 15.3671875, "learning_rate": 9.72917851051734e-06, "loss": 1.388, "step": 8122 }, { "epoch": 0.8643328367737817, "grad_norm": 14.875, "learning_rate": 9.729142978965323e-06, "loss": 1.2223, "step": 8123 }, { "epoch": 0.8644392423919983, "grad_norm": 13.59375, "learning_rate": 9.729107447413305e-06, "loss": 1.2371, "step": 8124 }, { "epoch": 0.8645456480102149, "grad_norm": 15.34375, "learning_rate": 9.729071915861286e-06, "loss": 1.4521, "step": 8125 }, { "epoch": 0.8646520536284316, "grad_norm": 11.3359375, "learning_rate": 9.729036384309268e-06, "loss": 1.2335, "step": 8126 }, { "epoch": 0.8647584592466482, "grad_norm": 16.734375, "learning_rate": 9.72900085275725e-06, "loss": 1.055, "step": 8127 }, { "epoch": 0.8648648648648649, "grad_norm": 16.125, "learning_rate": 9.728965321205231e-06, "loss": 1.2168, "step": 8128 }, { "epoch": 0.8649712704830815, "grad_norm": 10.515625, "learning_rate": 9.728929789653213e-06, "loss": 1.2157, "step": 8129 }, { "epoch": 0.8650776761012982, "grad_norm": 12.8125, "learning_rate": 9.728894258101194e-06, "loss": 1.3144, "step": 8130 }, { "epoch": 0.8651840817195148, "grad_norm": 13.015625, "learning_rate": 9.728858726549176e-06, "loss": 1.3016, "step": 8131 }, { "epoch": 0.8652904873377314, "grad_norm": 10.40625, "learning_rate": 9.728823194997158e-06, "loss": 1.0175, "step": 8132 }, { "epoch": 0.8653968929559481, "grad_norm": 13.6015625, "learning_rate": 9.728787663445141e-06, "loss": 1.4429, "step": 8133 }, { "epoch": 0.8655032985741647, "grad_norm": 11.7578125, "learning_rate": 9.728752131893123e-06, "loss": 1.1102, "step": 8134 }, { "epoch": 0.8656097041923814, "grad_norm": 11.2578125, "learning_rate": 9.728716600341104e-06, "loss": 0.9821, "step": 8135 }, { "epoch": 0.865716109810598, "grad_norm": 10.2265625, "learning_rate": 9.728681068789086e-06, "loss": 1.0247, "step": 8136 }, { "epoch": 0.8658225154288146, "grad_norm": 12.5078125, "learning_rate": 9.728645537237067e-06, "loss": 1.559, "step": 8137 }, { "epoch": 0.8659289210470312, "grad_norm": 11.5546875, "learning_rate": 9.728610005685049e-06, "loss": 1.2302, "step": 8138 }, { "epoch": 0.866035326665248, "grad_norm": 10.5390625, "learning_rate": 9.72857447413303e-06, "loss": 1.0127, "step": 8139 }, { "epoch": 0.8661417322834646, "grad_norm": 11.859375, "learning_rate": 9.728538942581012e-06, "loss": 0.9853, "step": 8140 }, { "epoch": 0.8662481379016812, "grad_norm": 16.421875, "learning_rate": 9.728503411028994e-06, "loss": 1.1264, "step": 8141 }, { "epoch": 0.8663545435198978, "grad_norm": 13.265625, "learning_rate": 9.728467879476975e-06, "loss": 0.9139, "step": 8142 }, { "epoch": 0.8664609491381144, "grad_norm": 10.5546875, "learning_rate": 9.728432347924957e-06, "loss": 1.1148, "step": 8143 }, { "epoch": 0.8665673547563312, "grad_norm": 10.7890625, "learning_rate": 9.72839681637294e-06, "loss": 1.2584, "step": 8144 }, { "epoch": 0.8666737603745478, "grad_norm": 12.5859375, "learning_rate": 9.728361284820922e-06, "loss": 1.1736, "step": 8145 }, { "epoch": 0.8667801659927644, "grad_norm": 13.5, "learning_rate": 9.728325753268904e-06, "loss": 0.9555, "step": 8146 }, { "epoch": 0.866886571610981, "grad_norm": 9.6953125, "learning_rate": 9.728290221716885e-06, "loss": 1.2104, "step": 8147 }, { "epoch": 0.8669929772291977, "grad_norm": 17.796875, "learning_rate": 9.728254690164867e-06, "loss": 1.2673, "step": 8148 }, { "epoch": 0.8670993828474144, "grad_norm": 13.0390625, "learning_rate": 9.72821915861285e-06, "loss": 1.2389, "step": 8149 }, { "epoch": 0.867205788465631, "grad_norm": 11.6015625, "learning_rate": 9.728183627060832e-06, "loss": 1.3406, "step": 8150 }, { "epoch": 0.8673121940838476, "grad_norm": 10.75, "learning_rate": 9.728148095508813e-06, "loss": 0.9791, "step": 8151 }, { "epoch": 0.8674185997020643, "grad_norm": 11.3984375, "learning_rate": 9.728112563956795e-06, "loss": 1.1905, "step": 8152 }, { "epoch": 0.8675250053202809, "grad_norm": 14.8515625, "learning_rate": 9.728077032404776e-06, "loss": 1.2806, "step": 8153 }, { "epoch": 0.8676314109384976, "grad_norm": 14.96875, "learning_rate": 9.728041500852758e-06, "loss": 1.1727, "step": 8154 }, { "epoch": 0.8677378165567142, "grad_norm": 12.3515625, "learning_rate": 9.72800596930074e-06, "loss": 1.1638, "step": 8155 }, { "epoch": 0.8678442221749308, "grad_norm": 13.1640625, "learning_rate": 9.727970437748721e-06, "loss": 1.2141, "step": 8156 }, { "epoch": 0.8679506277931475, "grad_norm": 16.328125, "learning_rate": 9.727934906196703e-06, "loss": 1.2518, "step": 8157 }, { "epoch": 0.8680570334113641, "grad_norm": 10.328125, "learning_rate": 9.727899374644685e-06, "loss": 0.8781, "step": 8158 }, { "epoch": 0.8681634390295807, "grad_norm": 14.5546875, "learning_rate": 9.727863843092668e-06, "loss": 1.1652, "step": 8159 }, { "epoch": 0.8682698446477974, "grad_norm": 18.84375, "learning_rate": 9.72782831154065e-06, "loss": 1.0008, "step": 8160 }, { "epoch": 0.8683762502660141, "grad_norm": 11.015625, "learning_rate": 9.727792779988631e-06, "loss": 1.0847, "step": 8161 }, { "epoch": 0.8684826558842307, "grad_norm": 16.84375, "learning_rate": 9.727757248436613e-06, "loss": 1.4748, "step": 8162 }, { "epoch": 0.8685890615024473, "grad_norm": 10.0625, "learning_rate": 9.727721716884594e-06, "loss": 1.3831, "step": 8163 }, { "epoch": 0.8686954671206639, "grad_norm": 12.5234375, "learning_rate": 9.727686185332576e-06, "loss": 1.0056, "step": 8164 }, { "epoch": 0.8688018727388807, "grad_norm": 12.0234375, "learning_rate": 9.727650653780557e-06, "loss": 1.4278, "step": 8165 }, { "epoch": 0.8689082783570973, "grad_norm": 12.6171875, "learning_rate": 9.727615122228539e-06, "loss": 1.0571, "step": 8166 }, { "epoch": 0.8690146839753139, "grad_norm": 11.2421875, "learning_rate": 9.72757959067652e-06, "loss": 1.2119, "step": 8167 }, { "epoch": 0.8691210895935305, "grad_norm": 10.953125, "learning_rate": 9.727544059124502e-06, "loss": 0.8332, "step": 8168 }, { "epoch": 0.8692274952117471, "grad_norm": 13.4609375, "learning_rate": 9.727508527572486e-06, "loss": 1.5105, "step": 8169 }, { "epoch": 0.8693339008299639, "grad_norm": 11.390625, "learning_rate": 9.727472996020467e-06, "loss": 1.1694, "step": 8170 }, { "epoch": 0.8694403064481805, "grad_norm": 13.3203125, "learning_rate": 9.727437464468449e-06, "loss": 1.0641, "step": 8171 }, { "epoch": 0.8695467120663971, "grad_norm": 10.734375, "learning_rate": 9.72740193291643e-06, "loss": 1.1994, "step": 8172 }, { "epoch": 0.8696531176846137, "grad_norm": 12.015625, "learning_rate": 9.727366401364412e-06, "loss": 1.3499, "step": 8173 }, { "epoch": 0.8697595233028304, "grad_norm": 16.40625, "learning_rate": 9.727330869812395e-06, "loss": 1.0602, "step": 8174 }, { "epoch": 0.8698659289210471, "grad_norm": 15.3984375, "learning_rate": 9.727295338260377e-06, "loss": 1.1657, "step": 8175 }, { "epoch": 0.8699723345392637, "grad_norm": 16.546875, "learning_rate": 9.727259806708359e-06, "loss": 0.9758, "step": 8176 }, { "epoch": 0.8700787401574803, "grad_norm": 12.9609375, "learning_rate": 9.72722427515634e-06, "loss": 0.9844, "step": 8177 }, { "epoch": 0.8701851457756969, "grad_norm": 11.515625, "learning_rate": 9.727188743604322e-06, "loss": 1.2832, "step": 8178 }, { "epoch": 0.8702915513939136, "grad_norm": 13.3671875, "learning_rate": 9.727153212052303e-06, "loss": 1.0253, "step": 8179 }, { "epoch": 0.8703979570121302, "grad_norm": 10.640625, "learning_rate": 9.727117680500285e-06, "loss": 1.2287, "step": 8180 }, { "epoch": 0.8705043626303469, "grad_norm": 15.046875, "learning_rate": 9.727082148948267e-06, "loss": 1.2645, "step": 8181 }, { "epoch": 0.8706107682485635, "grad_norm": 13.7890625, "learning_rate": 9.727046617396248e-06, "loss": 1.1425, "step": 8182 }, { "epoch": 0.8707171738667802, "grad_norm": 13.203125, "learning_rate": 9.72701108584423e-06, "loss": 1.0875, "step": 8183 }, { "epoch": 0.8708235794849968, "grad_norm": 14.140625, "learning_rate": 9.726975554292211e-06, "loss": 1.2591, "step": 8184 }, { "epoch": 0.8709299851032134, "grad_norm": 15.2109375, "learning_rate": 9.726940022740195e-06, "loss": 1.1232, "step": 8185 }, { "epoch": 0.8710363907214301, "grad_norm": 13.859375, "learning_rate": 9.726904491188176e-06, "loss": 1.2315, "step": 8186 }, { "epoch": 0.8711427963396468, "grad_norm": 10.640625, "learning_rate": 9.726868959636158e-06, "loss": 1.0132, "step": 8187 }, { "epoch": 0.8712492019578634, "grad_norm": 11.15625, "learning_rate": 9.72683342808414e-06, "loss": 1.287, "step": 8188 }, { "epoch": 0.87135560757608, "grad_norm": 12.296875, "learning_rate": 9.726797896532121e-06, "loss": 1.1112, "step": 8189 }, { "epoch": 0.8714620131942966, "grad_norm": 11.1484375, "learning_rate": 9.726762364980103e-06, "loss": 1.1552, "step": 8190 }, { "epoch": 0.8715684188125133, "grad_norm": 13.6328125, "learning_rate": 9.726726833428084e-06, "loss": 1.2999, "step": 8191 }, { "epoch": 0.87167482443073, "grad_norm": 14.0546875, "learning_rate": 9.726691301876066e-06, "loss": 1.243, "step": 8192 }, { "epoch": 0.8717812300489466, "grad_norm": 10.765625, "learning_rate": 9.726655770324048e-06, "loss": 1.2468, "step": 8193 }, { "epoch": 0.8718876356671632, "grad_norm": 10.2421875, "learning_rate": 9.726620238772029e-06, "loss": 1.2897, "step": 8194 }, { "epoch": 0.8719940412853798, "grad_norm": 12.28125, "learning_rate": 9.726584707220012e-06, "loss": 1.2256, "step": 8195 }, { "epoch": 0.8721004469035966, "grad_norm": 10.3125, "learning_rate": 9.726549175667994e-06, "loss": 1.0039, "step": 8196 }, { "epoch": 0.8722068525218132, "grad_norm": 7.96484375, "learning_rate": 9.726513644115976e-06, "loss": 0.8634, "step": 8197 }, { "epoch": 0.8723132581400298, "grad_norm": 12.6796875, "learning_rate": 9.726478112563957e-06, "loss": 1.1845, "step": 8198 }, { "epoch": 0.8724196637582464, "grad_norm": 15.2734375, "learning_rate": 9.726442581011939e-06, "loss": 0.9786, "step": 8199 }, { "epoch": 0.872526069376463, "grad_norm": 12.0078125, "learning_rate": 9.726407049459922e-06, "loss": 1.565, "step": 8200 }, { "epoch": 0.8726324749946798, "grad_norm": 10.8671875, "learning_rate": 9.726371517907904e-06, "loss": 1.0547, "step": 8201 }, { "epoch": 0.8727388806128964, "grad_norm": 8.484375, "learning_rate": 9.726335986355885e-06, "loss": 1.0832, "step": 8202 }, { "epoch": 0.872845286231113, "grad_norm": 12.7578125, "learning_rate": 9.726300454803867e-06, "loss": 1.1618, "step": 8203 }, { "epoch": 0.8729516918493296, "grad_norm": 17.953125, "learning_rate": 9.726264923251849e-06, "loss": 1.433, "step": 8204 }, { "epoch": 0.8730580974675463, "grad_norm": 12.9453125, "learning_rate": 9.72622939169983e-06, "loss": 1.0374, "step": 8205 }, { "epoch": 0.8731645030857629, "grad_norm": 18.578125, "learning_rate": 9.726193860147812e-06, "loss": 0.9754, "step": 8206 }, { "epoch": 0.8732709087039796, "grad_norm": 8.984375, "learning_rate": 9.726158328595793e-06, "loss": 1.0971, "step": 8207 }, { "epoch": 0.8733773143221962, "grad_norm": 12.171875, "learning_rate": 9.726122797043775e-06, "loss": 1.2529, "step": 8208 }, { "epoch": 0.8734837199404129, "grad_norm": 13.296875, "learning_rate": 9.726087265491757e-06, "loss": 1.1157, "step": 8209 }, { "epoch": 0.8735901255586295, "grad_norm": 9.4453125, "learning_rate": 9.72605173393974e-06, "loss": 1.0708, "step": 8210 }, { "epoch": 0.8736965311768461, "grad_norm": 11.0390625, "learning_rate": 9.726016202387722e-06, "loss": 1.1896, "step": 8211 }, { "epoch": 0.8738029367950628, "grad_norm": 11.140625, "learning_rate": 9.725980670835703e-06, "loss": 1.0088, "step": 8212 }, { "epoch": 0.8739093424132794, "grad_norm": 10.75, "learning_rate": 9.725945139283685e-06, "loss": 1.2827, "step": 8213 }, { "epoch": 0.8740157480314961, "grad_norm": 9.359375, "learning_rate": 9.725909607731666e-06, "loss": 1.3402, "step": 8214 }, { "epoch": 0.8741221536497127, "grad_norm": 12.5390625, "learning_rate": 9.72587407617965e-06, "loss": 1.3748, "step": 8215 }, { "epoch": 0.8742285592679293, "grad_norm": 10.796875, "learning_rate": 9.725838544627631e-06, "loss": 1.0222, "step": 8216 }, { "epoch": 0.874334964886146, "grad_norm": 9.9296875, "learning_rate": 9.725803013075613e-06, "loss": 1.2351, "step": 8217 }, { "epoch": 0.8744413705043627, "grad_norm": 11.171875, "learning_rate": 9.725767481523593e-06, "loss": 1.4239, "step": 8218 }, { "epoch": 0.8745477761225793, "grad_norm": 12.703125, "learning_rate": 9.725731949971574e-06, "loss": 1.3345, "step": 8219 }, { "epoch": 0.8746541817407959, "grad_norm": 14.4140625, "learning_rate": 9.725696418419558e-06, "loss": 1.1979, "step": 8220 }, { "epoch": 0.8747605873590125, "grad_norm": 8.40625, "learning_rate": 9.72566088686754e-06, "loss": 1.2207, "step": 8221 }, { "epoch": 0.8748669929772293, "grad_norm": 12.8046875, "learning_rate": 9.725625355315521e-06, "loss": 1.057, "step": 8222 }, { "epoch": 0.8749733985954459, "grad_norm": 13.421875, "learning_rate": 9.725589823763503e-06, "loss": 1.0386, "step": 8223 }, { "epoch": 0.8750798042136625, "grad_norm": 15.8046875, "learning_rate": 9.725554292211484e-06, "loss": 1.031, "step": 8224 }, { "epoch": 0.8751862098318791, "grad_norm": 13.109375, "learning_rate": 9.725518760659466e-06, "loss": 1.0587, "step": 8225 }, { "epoch": 0.8752926154500957, "grad_norm": 10.71875, "learning_rate": 9.725483229107449e-06, "loss": 1.1858, "step": 8226 }, { "epoch": 0.8753990210683124, "grad_norm": 21.4375, "learning_rate": 9.72544769755543e-06, "loss": 1.438, "step": 8227 }, { "epoch": 0.8755054266865291, "grad_norm": 16.3125, "learning_rate": 9.725412166003412e-06, "loss": 1.1018, "step": 8228 }, { "epoch": 0.8756118323047457, "grad_norm": 14.453125, "learning_rate": 9.725376634451394e-06, "loss": 0.8999, "step": 8229 }, { "epoch": 0.8757182379229623, "grad_norm": 11.7890625, "learning_rate": 9.725341102899375e-06, "loss": 1.0529, "step": 8230 }, { "epoch": 0.875824643541179, "grad_norm": 11.90625, "learning_rate": 9.725305571347357e-06, "loss": 1.2952, "step": 8231 }, { "epoch": 0.8759310491593956, "grad_norm": 27.84375, "learning_rate": 9.725270039795339e-06, "loss": 1.5324, "step": 8232 }, { "epoch": 0.8760374547776123, "grad_norm": 14.8359375, "learning_rate": 9.72523450824332e-06, "loss": 1.2397, "step": 8233 }, { "epoch": 0.8761438603958289, "grad_norm": 11.5859375, "learning_rate": 9.725198976691302e-06, "loss": 1.2273, "step": 8234 }, { "epoch": 0.8762502660140455, "grad_norm": 15.8828125, "learning_rate": 9.725163445139283e-06, "loss": 1.2422, "step": 8235 }, { "epoch": 0.8763566716322622, "grad_norm": 11.9375, "learning_rate": 9.725127913587267e-06, "loss": 1.307, "step": 8236 }, { "epoch": 0.8764630772504788, "grad_norm": 15.25, "learning_rate": 9.725092382035248e-06, "loss": 1.3223, "step": 8237 }, { "epoch": 0.8765694828686955, "grad_norm": 10.375, "learning_rate": 9.72505685048323e-06, "loss": 1.2684, "step": 8238 }, { "epoch": 0.8766758884869121, "grad_norm": 13.78125, "learning_rate": 9.725021318931212e-06, "loss": 1.1715, "step": 8239 }, { "epoch": 0.8767822941051288, "grad_norm": 13.59375, "learning_rate": 9.724985787379193e-06, "loss": 1.2057, "step": 8240 }, { "epoch": 0.8768886997233454, "grad_norm": 11.2734375, "learning_rate": 9.724950255827177e-06, "loss": 1.0231, "step": 8241 }, { "epoch": 0.876995105341562, "grad_norm": 12.5078125, "learning_rate": 9.724914724275158e-06, "loss": 1.2149, "step": 8242 }, { "epoch": 0.8771015109597787, "grad_norm": 17.515625, "learning_rate": 9.72487919272314e-06, "loss": 1.0743, "step": 8243 }, { "epoch": 0.8772079165779954, "grad_norm": 11.71875, "learning_rate": 9.724843661171121e-06, "loss": 1.0911, "step": 8244 }, { "epoch": 0.877314322196212, "grad_norm": 11.0390625, "learning_rate": 9.724808129619103e-06, "loss": 1.3048, "step": 8245 }, { "epoch": 0.8774207278144286, "grad_norm": 20.484375, "learning_rate": 9.724772598067085e-06, "loss": 1.5935, "step": 8246 }, { "epoch": 0.8775271334326452, "grad_norm": 10.3046875, "learning_rate": 9.724737066515066e-06, "loss": 1.3651, "step": 8247 }, { "epoch": 0.8776335390508618, "grad_norm": 12.78125, "learning_rate": 9.724701534963048e-06, "loss": 1.2332, "step": 8248 }, { "epoch": 0.8777399446690786, "grad_norm": 25.578125, "learning_rate": 9.72466600341103e-06, "loss": 1.8956, "step": 8249 }, { "epoch": 0.8778463502872952, "grad_norm": 15.7578125, "learning_rate": 9.724630471859011e-06, "loss": 1.17, "step": 8250 }, { "epoch": 0.8779527559055118, "grad_norm": 12.6328125, "learning_rate": 9.724594940306994e-06, "loss": 1.0334, "step": 8251 }, { "epoch": 0.8780591615237284, "grad_norm": 11.3203125, "learning_rate": 9.724559408754976e-06, "loss": 1.1509, "step": 8252 }, { "epoch": 0.878165567141945, "grad_norm": 11.3203125, "learning_rate": 9.724523877202957e-06, "loss": 1.2545, "step": 8253 }, { "epoch": 0.8782719727601618, "grad_norm": 15.671875, "learning_rate": 9.724488345650939e-06, "loss": 1.3063, "step": 8254 }, { "epoch": 0.8783783783783784, "grad_norm": 15.734375, "learning_rate": 9.72445281409892e-06, "loss": 1.0267, "step": 8255 }, { "epoch": 0.878484783996595, "grad_norm": 14.2734375, "learning_rate": 9.724417282546902e-06, "loss": 1.2523, "step": 8256 }, { "epoch": 0.8785911896148116, "grad_norm": 13.6484375, "learning_rate": 9.724381750994884e-06, "loss": 0.9564, "step": 8257 }, { "epoch": 0.8786975952330283, "grad_norm": 10.8359375, "learning_rate": 9.724346219442866e-06, "loss": 1.166, "step": 8258 }, { "epoch": 0.878804000851245, "grad_norm": 16.84375, "learning_rate": 9.724310687890847e-06, "loss": 1.6833, "step": 8259 }, { "epoch": 0.8789104064694616, "grad_norm": 10.0078125, "learning_rate": 9.724275156338829e-06, "loss": 1.1343, "step": 8260 }, { "epoch": 0.8790168120876782, "grad_norm": 13.1015625, "learning_rate": 9.72423962478681e-06, "loss": 0.9474, "step": 8261 }, { "epoch": 0.8791232177058949, "grad_norm": 12.6875, "learning_rate": 9.724204093234794e-06, "loss": 1.1135, "step": 8262 }, { "epoch": 0.8792296233241115, "grad_norm": 10.9375, "learning_rate": 9.724168561682775e-06, "loss": 1.2136, "step": 8263 }, { "epoch": 0.8793360289423282, "grad_norm": 13.171875, "learning_rate": 9.724133030130757e-06, "loss": 1.1749, "step": 8264 }, { "epoch": 0.8794424345605448, "grad_norm": 15.703125, "learning_rate": 9.724097498578738e-06, "loss": 1.0756, "step": 8265 }, { "epoch": 0.8795488401787614, "grad_norm": 9.9609375, "learning_rate": 9.72406196702672e-06, "loss": 1.1227, "step": 8266 }, { "epoch": 0.8796552457969781, "grad_norm": 13.8671875, "learning_rate": 9.724026435474703e-06, "loss": 1.4069, "step": 8267 }, { "epoch": 0.8797616514151947, "grad_norm": 18.015625, "learning_rate": 9.723990903922685e-06, "loss": 0.8862, "step": 8268 }, { "epoch": 0.8798680570334113, "grad_norm": 13.5390625, "learning_rate": 9.723955372370667e-06, "loss": 1.1215, "step": 8269 }, { "epoch": 0.879974462651628, "grad_norm": 11.2734375, "learning_rate": 9.723919840818648e-06, "loss": 1.1202, "step": 8270 }, { "epoch": 0.8800808682698447, "grad_norm": 10.0546875, "learning_rate": 9.72388430926663e-06, "loss": 1.1849, "step": 8271 }, { "epoch": 0.8801872738880613, "grad_norm": 18.28125, "learning_rate": 9.723848777714611e-06, "loss": 1.3727, "step": 8272 }, { "epoch": 0.8802936795062779, "grad_norm": 13.9375, "learning_rate": 9.723813246162593e-06, "loss": 0.9486, "step": 8273 }, { "epoch": 0.8804000851244945, "grad_norm": 12.0390625, "learning_rate": 9.723777714610575e-06, "loss": 1.0766, "step": 8274 }, { "epoch": 0.8805064907427113, "grad_norm": 12.3203125, "learning_rate": 9.723742183058556e-06, "loss": 1.1852, "step": 8275 }, { "epoch": 0.8806128963609279, "grad_norm": 18.203125, "learning_rate": 9.723706651506538e-06, "loss": 1.2564, "step": 8276 }, { "epoch": 0.8807193019791445, "grad_norm": 9.75, "learning_rate": 9.723671119954521e-06, "loss": 1.3664, "step": 8277 }, { "epoch": 0.8808257075973611, "grad_norm": 11.984375, "learning_rate": 9.723635588402503e-06, "loss": 1.1399, "step": 8278 }, { "epoch": 0.8809321132155777, "grad_norm": 13.8515625, "learning_rate": 9.723600056850484e-06, "loss": 1.0221, "step": 8279 }, { "epoch": 0.8810385188337945, "grad_norm": 10.0078125, "learning_rate": 9.723564525298466e-06, "loss": 1.2212, "step": 8280 }, { "epoch": 0.8811449244520111, "grad_norm": 17.640625, "learning_rate": 9.723528993746448e-06, "loss": 1.4487, "step": 8281 }, { "epoch": 0.8812513300702277, "grad_norm": 13.1015625, "learning_rate": 9.72349346219443e-06, "loss": 1.056, "step": 8282 }, { "epoch": 0.8813577356884443, "grad_norm": 12.390625, "learning_rate": 9.72345793064241e-06, "loss": 1.3367, "step": 8283 }, { "epoch": 0.881464141306661, "grad_norm": 10.546875, "learning_rate": 9.723422399090392e-06, "loss": 1.0233, "step": 8284 }, { "epoch": 0.8815705469248777, "grad_norm": 11.9296875, "learning_rate": 9.723386867538374e-06, "loss": 1.1592, "step": 8285 }, { "epoch": 0.8816769525430943, "grad_norm": 16.953125, "learning_rate": 9.723351335986356e-06, "loss": 1.1574, "step": 8286 }, { "epoch": 0.8817833581613109, "grad_norm": 13.2421875, "learning_rate": 9.723315804434339e-06, "loss": 1.375, "step": 8287 }, { "epoch": 0.8818897637795275, "grad_norm": 18.09375, "learning_rate": 9.72328027288232e-06, "loss": 1.017, "step": 8288 }, { "epoch": 0.8819961693977442, "grad_norm": 11.6484375, "learning_rate": 9.723244741330302e-06, "loss": 0.9607, "step": 8289 }, { "epoch": 0.8821025750159608, "grad_norm": 13.0234375, "learning_rate": 9.723209209778284e-06, "loss": 1.3077, "step": 8290 }, { "epoch": 0.8822089806341775, "grad_norm": 13.234375, "learning_rate": 9.723173678226265e-06, "loss": 1.1938, "step": 8291 }, { "epoch": 0.8823153862523941, "grad_norm": 17.09375, "learning_rate": 9.723138146674249e-06, "loss": 1.032, "step": 8292 }, { "epoch": 0.8824217918706108, "grad_norm": 15.3125, "learning_rate": 9.72310261512223e-06, "loss": 1.1184, "step": 8293 }, { "epoch": 0.8825281974888274, "grad_norm": 9.4609375, "learning_rate": 9.723067083570212e-06, "loss": 1.0395, "step": 8294 }, { "epoch": 0.882634603107044, "grad_norm": 11.3046875, "learning_rate": 9.723031552018193e-06, "loss": 1.1033, "step": 8295 }, { "epoch": 0.8827410087252607, "grad_norm": 10.6328125, "learning_rate": 9.722996020466175e-06, "loss": 1.2851, "step": 8296 }, { "epoch": 0.8828474143434774, "grad_norm": 25.375, "learning_rate": 9.722960488914157e-06, "loss": 1.5745, "step": 8297 }, { "epoch": 0.882953819961694, "grad_norm": 14.3671875, "learning_rate": 9.722924957362138e-06, "loss": 1.1474, "step": 8298 }, { "epoch": 0.8830602255799106, "grad_norm": 12.515625, "learning_rate": 9.72288942581012e-06, "loss": 1.3134, "step": 8299 }, { "epoch": 0.8831666311981272, "grad_norm": 7.83203125, "learning_rate": 9.722853894258101e-06, "loss": 0.9556, "step": 8300 }, { "epoch": 0.883273036816344, "grad_norm": 15.015625, "learning_rate": 9.722818362706083e-06, "loss": 1.0092, "step": 8301 }, { "epoch": 0.8833794424345606, "grad_norm": 13.7734375, "learning_rate": 9.722782831154065e-06, "loss": 1.2245, "step": 8302 }, { "epoch": 0.8834858480527772, "grad_norm": 13.3671875, "learning_rate": 9.722747299602048e-06, "loss": 1.195, "step": 8303 }, { "epoch": 0.8835922536709938, "grad_norm": 12.3046875, "learning_rate": 9.72271176805003e-06, "loss": 1.1785, "step": 8304 }, { "epoch": 0.8836986592892104, "grad_norm": 14.296875, "learning_rate": 9.722676236498011e-06, "loss": 1.1129, "step": 8305 }, { "epoch": 0.8838050649074272, "grad_norm": 12.0390625, "learning_rate": 9.722640704945993e-06, "loss": 1.212, "step": 8306 }, { "epoch": 0.8839114705256438, "grad_norm": 18.109375, "learning_rate": 9.722605173393974e-06, "loss": 1.1544, "step": 8307 }, { "epoch": 0.8840178761438604, "grad_norm": 9.9140625, "learning_rate": 9.722569641841956e-06, "loss": 1.1097, "step": 8308 }, { "epoch": 0.884124281762077, "grad_norm": 13.6015625, "learning_rate": 9.722534110289938e-06, "loss": 1.3322, "step": 8309 }, { "epoch": 0.8842306873802936, "grad_norm": 23.109375, "learning_rate": 9.72249857873792e-06, "loss": 1.2088, "step": 8310 }, { "epoch": 0.8843370929985104, "grad_norm": 15.8203125, "learning_rate": 9.7224630471859e-06, "loss": 0.9872, "step": 8311 }, { "epoch": 0.884443498616727, "grad_norm": 11.84375, "learning_rate": 9.722427515633882e-06, "loss": 0.932, "step": 8312 }, { "epoch": 0.8845499042349436, "grad_norm": 11.4921875, "learning_rate": 9.722391984081866e-06, "loss": 1.3499, "step": 8313 }, { "epoch": 0.8846563098531602, "grad_norm": 12.9453125, "learning_rate": 9.722356452529847e-06, "loss": 1.4078, "step": 8314 }, { "epoch": 0.8847627154713769, "grad_norm": 10.8203125, "learning_rate": 9.722320920977829e-06, "loss": 1.0019, "step": 8315 }, { "epoch": 0.8848691210895935, "grad_norm": 13.0390625, "learning_rate": 9.72228538942581e-06, "loss": 1.2813, "step": 8316 }, { "epoch": 0.8849755267078102, "grad_norm": 11.6796875, "learning_rate": 9.722249857873792e-06, "loss": 1.3859, "step": 8317 }, { "epoch": 0.8850819323260268, "grad_norm": 12.78125, "learning_rate": 9.722214326321775e-06, "loss": 1.2104, "step": 8318 }, { "epoch": 0.8851883379442435, "grad_norm": 14.78125, "learning_rate": 9.722178794769757e-06, "loss": 0.9984, "step": 8319 }, { "epoch": 0.8852947435624601, "grad_norm": 12.2734375, "learning_rate": 9.722143263217739e-06, "loss": 1.0242, "step": 8320 }, { "epoch": 0.8854011491806767, "grad_norm": 14.6015625, "learning_rate": 9.72210773166572e-06, "loss": 1.3193, "step": 8321 }, { "epoch": 0.8855075547988934, "grad_norm": 11.1171875, "learning_rate": 9.722072200113702e-06, "loss": 1.2018, "step": 8322 }, { "epoch": 0.88561396041711, "grad_norm": 13.4296875, "learning_rate": 9.722036668561684e-06, "loss": 1.4185, "step": 8323 }, { "epoch": 0.8857203660353267, "grad_norm": 15.28125, "learning_rate": 9.722001137009665e-06, "loss": 1.1747, "step": 8324 }, { "epoch": 0.8858267716535433, "grad_norm": 12.4140625, "learning_rate": 9.721965605457647e-06, "loss": 1.1465, "step": 8325 }, { "epoch": 0.8859331772717599, "grad_norm": 13.609375, "learning_rate": 9.721930073905628e-06, "loss": 0.9704, "step": 8326 }, { "epoch": 0.8860395828899766, "grad_norm": 15.734375, "learning_rate": 9.72189454235361e-06, "loss": 1.2189, "step": 8327 }, { "epoch": 0.8861459885081933, "grad_norm": 12.25, "learning_rate": 9.721859010801593e-06, "loss": 0.9901, "step": 8328 }, { "epoch": 0.8862523941264099, "grad_norm": 15.0, "learning_rate": 9.721823479249575e-06, "loss": 1.2904, "step": 8329 }, { "epoch": 0.8863587997446265, "grad_norm": 11.3046875, "learning_rate": 9.721787947697556e-06, "loss": 1.2557, "step": 8330 }, { "epoch": 0.8864652053628431, "grad_norm": 10.6484375, "learning_rate": 9.721752416145538e-06, "loss": 0.8487, "step": 8331 }, { "epoch": 0.8865716109810599, "grad_norm": 10.0390625, "learning_rate": 9.72171688459352e-06, "loss": 1.3116, "step": 8332 }, { "epoch": 0.8866780165992765, "grad_norm": 13.109375, "learning_rate": 9.721681353041503e-06, "loss": 1.1857, "step": 8333 }, { "epoch": 0.8867844222174931, "grad_norm": 15.46875, "learning_rate": 9.721645821489485e-06, "loss": 1.2795, "step": 8334 }, { "epoch": 0.8868908278357097, "grad_norm": 18.90625, "learning_rate": 9.721610289937466e-06, "loss": 1.0135, "step": 8335 }, { "epoch": 0.8869972334539263, "grad_norm": 17.03125, "learning_rate": 9.721574758385448e-06, "loss": 1.3901, "step": 8336 }, { "epoch": 0.887103639072143, "grad_norm": 15.3125, "learning_rate": 9.72153922683343e-06, "loss": 1.0916, "step": 8337 }, { "epoch": 0.8872100446903597, "grad_norm": 10.3515625, "learning_rate": 9.72150369528141e-06, "loss": 1.1411, "step": 8338 }, { "epoch": 0.8873164503085763, "grad_norm": 11.5546875, "learning_rate": 9.721468163729393e-06, "loss": 1.4897, "step": 8339 }, { "epoch": 0.8874228559267929, "grad_norm": 12.0, "learning_rate": 9.721432632177374e-06, "loss": 1.1521, "step": 8340 }, { "epoch": 0.8875292615450096, "grad_norm": 9.4765625, "learning_rate": 9.721397100625356e-06, "loss": 1.25, "step": 8341 }, { "epoch": 0.8876356671632262, "grad_norm": 10.8515625, "learning_rate": 9.721361569073337e-06, "loss": 1.0789, "step": 8342 }, { "epoch": 0.8877420727814429, "grad_norm": 13.21875, "learning_rate": 9.721326037521319e-06, "loss": 1.0808, "step": 8343 }, { "epoch": 0.8878484783996595, "grad_norm": 17.453125, "learning_rate": 9.721290505969302e-06, "loss": 1.3691, "step": 8344 }, { "epoch": 0.8879548840178761, "grad_norm": 11.7890625, "learning_rate": 9.721254974417284e-06, "loss": 1.4707, "step": 8345 }, { "epoch": 0.8880612896360928, "grad_norm": 11.5234375, "learning_rate": 9.721219442865266e-06, "loss": 1.1745, "step": 8346 }, { "epoch": 0.8881676952543094, "grad_norm": 10.2578125, "learning_rate": 9.721183911313247e-06, "loss": 0.9484, "step": 8347 }, { "epoch": 0.8882741008725261, "grad_norm": 8.5234375, "learning_rate": 9.721148379761229e-06, "loss": 0.9897, "step": 8348 }, { "epoch": 0.8883805064907427, "grad_norm": 10.078125, "learning_rate": 9.72111284820921e-06, "loss": 1.1421, "step": 8349 }, { "epoch": 0.8884869121089594, "grad_norm": 12.03125, "learning_rate": 9.721077316657192e-06, "loss": 1.1847, "step": 8350 }, { "epoch": 0.888593317727176, "grad_norm": 9.8046875, "learning_rate": 9.721041785105174e-06, "loss": 1.0233, "step": 8351 }, { "epoch": 0.8886997233453926, "grad_norm": 10.2890625, "learning_rate": 9.721006253553155e-06, "loss": 1.0712, "step": 8352 }, { "epoch": 0.8888061289636093, "grad_norm": 11.5390625, "learning_rate": 9.720970722001137e-06, "loss": 1.1704, "step": 8353 }, { "epoch": 0.888912534581826, "grad_norm": 13.3203125, "learning_rate": 9.72093519044912e-06, "loss": 1.0434, "step": 8354 }, { "epoch": 0.8890189402000426, "grad_norm": 19.71875, "learning_rate": 9.720899658897102e-06, "loss": 1.4529, "step": 8355 }, { "epoch": 0.8891253458182592, "grad_norm": 15.09375, "learning_rate": 9.720864127345083e-06, "loss": 1.3671, "step": 8356 }, { "epoch": 0.8892317514364758, "grad_norm": 10.8671875, "learning_rate": 9.720828595793065e-06, "loss": 1.2023, "step": 8357 }, { "epoch": 0.8893381570546924, "grad_norm": 20.078125, "learning_rate": 9.720793064241047e-06, "loss": 1.285, "step": 8358 }, { "epoch": 0.8894445626729092, "grad_norm": 12.21875, "learning_rate": 9.72075753268903e-06, "loss": 1.2601, "step": 8359 }, { "epoch": 0.8895509682911258, "grad_norm": 20.3125, "learning_rate": 9.720722001137011e-06, "loss": 1.1353, "step": 8360 }, { "epoch": 0.8896573739093424, "grad_norm": 14.6640625, "learning_rate": 9.720686469584993e-06, "loss": 1.3525, "step": 8361 }, { "epoch": 0.889763779527559, "grad_norm": 12.09375, "learning_rate": 9.720650938032975e-06, "loss": 1.2528, "step": 8362 }, { "epoch": 0.8898701851457756, "grad_norm": 11.8046875, "learning_rate": 9.720615406480956e-06, "loss": 1.3432, "step": 8363 }, { "epoch": 0.8899765907639924, "grad_norm": 12.359375, "learning_rate": 9.720579874928938e-06, "loss": 1.3626, "step": 8364 }, { "epoch": 0.890082996382209, "grad_norm": 15.3984375, "learning_rate": 9.72054434337692e-06, "loss": 1.6152, "step": 8365 }, { "epoch": 0.8901894020004256, "grad_norm": 17.546875, "learning_rate": 9.720508811824901e-06, "loss": 1.1668, "step": 8366 }, { "epoch": 0.8902958076186422, "grad_norm": 16.5, "learning_rate": 9.720473280272883e-06, "loss": 1.0463, "step": 8367 }, { "epoch": 0.8904022132368589, "grad_norm": 18.21875, "learning_rate": 9.720437748720864e-06, "loss": 1.191, "step": 8368 }, { "epoch": 0.8905086188550756, "grad_norm": 15.484375, "learning_rate": 9.720402217168848e-06, "loss": 1.3824, "step": 8369 }, { "epoch": 0.8906150244732922, "grad_norm": 13.109375, "learning_rate": 9.72036668561683e-06, "loss": 1.3554, "step": 8370 }, { "epoch": 0.8907214300915088, "grad_norm": 11.5859375, "learning_rate": 9.72033115406481e-06, "loss": 1.4353, "step": 8371 }, { "epoch": 0.8908278357097255, "grad_norm": 15.3359375, "learning_rate": 9.720295622512792e-06, "loss": 1.0293, "step": 8372 }, { "epoch": 0.8909342413279421, "grad_norm": 16.59375, "learning_rate": 9.720260090960774e-06, "loss": 1.4383, "step": 8373 }, { "epoch": 0.8910406469461588, "grad_norm": 12.296875, "learning_rate": 9.720224559408756e-06, "loss": 1.3088, "step": 8374 }, { "epoch": 0.8911470525643754, "grad_norm": 15.2734375, "learning_rate": 9.720189027856737e-06, "loss": 1.206, "step": 8375 }, { "epoch": 0.891253458182592, "grad_norm": 9.25, "learning_rate": 9.720153496304719e-06, "loss": 1.1263, "step": 8376 }, { "epoch": 0.8913598638008087, "grad_norm": 13.625, "learning_rate": 9.7201179647527e-06, "loss": 1.2946, "step": 8377 }, { "epoch": 0.8914662694190253, "grad_norm": 9.921875, "learning_rate": 9.720082433200682e-06, "loss": 1.0823, "step": 8378 }, { "epoch": 0.8915726750372419, "grad_norm": 12.3203125, "learning_rate": 9.720046901648664e-06, "loss": 1.4292, "step": 8379 }, { "epoch": 0.8916790806554586, "grad_norm": 11.171875, "learning_rate": 9.720011370096647e-06, "loss": 1.0894, "step": 8380 }, { "epoch": 0.8917854862736753, "grad_norm": 8.3046875, "learning_rate": 9.719975838544629e-06, "loss": 1.0405, "step": 8381 }, { "epoch": 0.8918918918918919, "grad_norm": 10.6484375, "learning_rate": 9.71994030699261e-06, "loss": 1.2275, "step": 8382 }, { "epoch": 0.8919982975101085, "grad_norm": 14.5234375, "learning_rate": 9.719904775440592e-06, "loss": 1.1177, "step": 8383 }, { "epoch": 0.8921047031283251, "grad_norm": 9.65625, "learning_rate": 9.719869243888573e-06, "loss": 1.2637, "step": 8384 }, { "epoch": 0.8922111087465419, "grad_norm": 12.7109375, "learning_rate": 9.719833712336557e-06, "loss": 1.2009, "step": 8385 }, { "epoch": 0.8923175143647585, "grad_norm": 13.8671875, "learning_rate": 9.719798180784538e-06, "loss": 0.9658, "step": 8386 }, { "epoch": 0.8924239199829751, "grad_norm": 10.3828125, "learning_rate": 9.71976264923252e-06, "loss": 1.2781, "step": 8387 }, { "epoch": 0.8925303256011917, "grad_norm": 13.5703125, "learning_rate": 9.719727117680501e-06, "loss": 1.3765, "step": 8388 }, { "epoch": 0.8926367312194083, "grad_norm": 16.296875, "learning_rate": 9.719691586128483e-06, "loss": 1.4479, "step": 8389 }, { "epoch": 0.8927431368376251, "grad_norm": 10.6484375, "learning_rate": 9.719656054576465e-06, "loss": 0.9174, "step": 8390 }, { "epoch": 0.8928495424558417, "grad_norm": 11.3359375, "learning_rate": 9.719620523024446e-06, "loss": 1.1209, "step": 8391 }, { "epoch": 0.8929559480740583, "grad_norm": 11.4296875, "learning_rate": 9.719584991472428e-06, "loss": 1.12, "step": 8392 }, { "epoch": 0.8930623536922749, "grad_norm": 14.0546875, "learning_rate": 9.71954945992041e-06, "loss": 1.0963, "step": 8393 }, { "epoch": 0.8931687593104916, "grad_norm": 11.046875, "learning_rate": 9.719513928368391e-06, "loss": 1.24, "step": 8394 }, { "epoch": 0.8932751649287083, "grad_norm": 14.5, "learning_rate": 9.719478396816374e-06, "loss": 1.3128, "step": 8395 }, { "epoch": 0.8933815705469249, "grad_norm": 15.5625, "learning_rate": 9.719442865264356e-06, "loss": 0.9659, "step": 8396 }, { "epoch": 0.8934879761651415, "grad_norm": 14.0, "learning_rate": 9.719407333712338e-06, "loss": 1.2009, "step": 8397 }, { "epoch": 0.8935943817833581, "grad_norm": 12.578125, "learning_rate": 9.71937180216032e-06, "loss": 1.0743, "step": 8398 }, { "epoch": 0.8937007874015748, "grad_norm": 10.7578125, "learning_rate": 9.719336270608301e-06, "loss": 1.314, "step": 8399 }, { "epoch": 0.8938071930197914, "grad_norm": 13.8515625, "learning_rate": 9.719300739056282e-06, "loss": 1.0299, "step": 8400 }, { "epoch": 0.8939135986380081, "grad_norm": 12.5859375, "learning_rate": 9.719265207504264e-06, "loss": 1.2184, "step": 8401 }, { "epoch": 0.8940200042562247, "grad_norm": 10.7265625, "learning_rate": 9.719229675952246e-06, "loss": 1.012, "step": 8402 }, { "epoch": 0.8941264098744414, "grad_norm": 17.9375, "learning_rate": 9.719194144400227e-06, "loss": 1.5441, "step": 8403 }, { "epoch": 0.894232815492658, "grad_norm": 18.375, "learning_rate": 9.719158612848209e-06, "loss": 1.0739, "step": 8404 }, { "epoch": 0.8943392211108746, "grad_norm": 15.3828125, "learning_rate": 9.719123081296192e-06, "loss": 1.3623, "step": 8405 }, { "epoch": 0.8944456267290913, "grad_norm": 10.1640625, "learning_rate": 9.719087549744174e-06, "loss": 0.8915, "step": 8406 }, { "epoch": 0.894552032347308, "grad_norm": 10.5078125, "learning_rate": 9.719052018192155e-06, "loss": 0.9997, "step": 8407 }, { "epoch": 0.8946584379655246, "grad_norm": 10.7265625, "learning_rate": 9.719016486640137e-06, "loss": 1.0603, "step": 8408 }, { "epoch": 0.8947648435837412, "grad_norm": 10.2109375, "learning_rate": 9.718980955088119e-06, "loss": 1.0826, "step": 8409 }, { "epoch": 0.8948712492019578, "grad_norm": 14.3515625, "learning_rate": 9.718945423536102e-06, "loss": 1.2927, "step": 8410 }, { "epoch": 0.8949776548201746, "grad_norm": 11.5625, "learning_rate": 9.718909891984084e-06, "loss": 1.1626, "step": 8411 }, { "epoch": 0.8950840604383912, "grad_norm": 13.5703125, "learning_rate": 9.718874360432065e-06, "loss": 1.4078, "step": 8412 }, { "epoch": 0.8951904660566078, "grad_norm": 17.21875, "learning_rate": 9.718838828880047e-06, "loss": 1.3082, "step": 8413 }, { "epoch": 0.8952968716748244, "grad_norm": 12.578125, "learning_rate": 9.718803297328028e-06, "loss": 1.2104, "step": 8414 }, { "epoch": 0.895403277293041, "grad_norm": 13.421875, "learning_rate": 9.71876776577601e-06, "loss": 1.1678, "step": 8415 }, { "epoch": 0.8955096829112578, "grad_norm": 11.4140625, "learning_rate": 9.718732234223992e-06, "loss": 1.2032, "step": 8416 }, { "epoch": 0.8956160885294744, "grad_norm": 12.6328125, "learning_rate": 9.718696702671973e-06, "loss": 1.1921, "step": 8417 }, { "epoch": 0.895722494147691, "grad_norm": 10.625, "learning_rate": 9.718661171119955e-06, "loss": 1.1023, "step": 8418 }, { "epoch": 0.8958288997659076, "grad_norm": 11.9609375, "learning_rate": 9.718625639567936e-06, "loss": 1.2294, "step": 8419 }, { "epoch": 0.8959353053841242, "grad_norm": 11.625, "learning_rate": 9.718590108015918e-06, "loss": 1.4098, "step": 8420 }, { "epoch": 0.8960417110023409, "grad_norm": 11.609375, "learning_rate": 9.718554576463901e-06, "loss": 1.3205, "step": 8421 }, { "epoch": 0.8961481166205576, "grad_norm": 17.40625, "learning_rate": 9.718519044911883e-06, "loss": 1.1686, "step": 8422 }, { "epoch": 0.8962545222387742, "grad_norm": 9.0, "learning_rate": 9.718483513359865e-06, "loss": 1.1171, "step": 8423 }, { "epoch": 0.8963609278569908, "grad_norm": 12.4140625, "learning_rate": 9.718447981807846e-06, "loss": 1.3686, "step": 8424 }, { "epoch": 0.8964673334752075, "grad_norm": 12.9375, "learning_rate": 9.718412450255828e-06, "loss": 1.204, "step": 8425 }, { "epoch": 0.8965737390934241, "grad_norm": 11.5703125, "learning_rate": 9.718376918703811e-06, "loss": 1.203, "step": 8426 }, { "epoch": 0.8966801447116408, "grad_norm": 15.4921875, "learning_rate": 9.718341387151791e-06, "loss": 1.2511, "step": 8427 }, { "epoch": 0.8967865503298574, "grad_norm": 16.203125, "learning_rate": 9.718305855599773e-06, "loss": 1.0648, "step": 8428 }, { "epoch": 0.896892955948074, "grad_norm": 14.9375, "learning_rate": 9.718270324047754e-06, "loss": 1.4618, "step": 8429 }, { "epoch": 0.8969993615662907, "grad_norm": 14.9296875, "learning_rate": 9.718234792495736e-06, "loss": 1.4581, "step": 8430 }, { "epoch": 0.8971057671845073, "grad_norm": 17.671875, "learning_rate": 9.718199260943719e-06, "loss": 1.4646, "step": 8431 }, { "epoch": 0.897212172802724, "grad_norm": 12.6171875, "learning_rate": 9.7181637293917e-06, "loss": 1.1479, "step": 8432 }, { "epoch": 0.8973185784209406, "grad_norm": 15.3359375, "learning_rate": 9.718128197839682e-06, "loss": 1.6631, "step": 8433 }, { "epoch": 0.8974249840391573, "grad_norm": 14.6875, "learning_rate": 9.718092666287664e-06, "loss": 1.2406, "step": 8434 }, { "epoch": 0.8975313896573739, "grad_norm": 12.0703125, "learning_rate": 9.718057134735645e-06, "loss": 1.4362, "step": 8435 }, { "epoch": 0.8976377952755905, "grad_norm": 10.515625, "learning_rate": 9.718021603183629e-06, "loss": 1.3958, "step": 8436 }, { "epoch": 0.8977442008938072, "grad_norm": 12.5625, "learning_rate": 9.71798607163161e-06, "loss": 1.1319, "step": 8437 }, { "epoch": 0.8978506065120239, "grad_norm": 13.546875, "learning_rate": 9.717950540079592e-06, "loss": 1.407, "step": 8438 }, { "epoch": 0.8979570121302405, "grad_norm": 11.6796875, "learning_rate": 9.717915008527574e-06, "loss": 0.9772, "step": 8439 }, { "epoch": 0.8980634177484571, "grad_norm": 9.8359375, "learning_rate": 9.717879476975555e-06, "loss": 1.4747, "step": 8440 }, { "epoch": 0.8981698233666737, "grad_norm": 16.53125, "learning_rate": 9.717843945423537e-06, "loss": 1.1052, "step": 8441 }, { "epoch": 0.8982762289848905, "grad_norm": 13.9453125, "learning_rate": 9.717808413871518e-06, "loss": 0.8213, "step": 8442 }, { "epoch": 0.8983826346031071, "grad_norm": 12.5078125, "learning_rate": 9.7177728823195e-06, "loss": 1.1632, "step": 8443 }, { "epoch": 0.8984890402213237, "grad_norm": 11.6953125, "learning_rate": 9.717737350767482e-06, "loss": 1.1044, "step": 8444 }, { "epoch": 0.8985954458395403, "grad_norm": 13.3359375, "learning_rate": 9.717701819215463e-06, "loss": 1.1177, "step": 8445 }, { "epoch": 0.8987018514577569, "grad_norm": 13.484375, "learning_rate": 9.717666287663447e-06, "loss": 1.2156, "step": 8446 }, { "epoch": 0.8988082570759736, "grad_norm": 12.1328125, "learning_rate": 9.717630756111428e-06, "loss": 1.048, "step": 8447 }, { "epoch": 0.8989146626941903, "grad_norm": 11.8671875, "learning_rate": 9.71759522455941e-06, "loss": 1.2517, "step": 8448 }, { "epoch": 0.8990210683124069, "grad_norm": 10.28125, "learning_rate": 9.717559693007391e-06, "loss": 1.38, "step": 8449 }, { "epoch": 0.8991274739306235, "grad_norm": 16.46875, "learning_rate": 9.717524161455373e-06, "loss": 1.6712, "step": 8450 }, { "epoch": 0.8992338795488402, "grad_norm": 12.015625, "learning_rate": 9.717488629903356e-06, "loss": 1.3141, "step": 8451 }, { "epoch": 0.8993402851670568, "grad_norm": 9.625, "learning_rate": 9.717453098351338e-06, "loss": 1.2591, "step": 8452 }, { "epoch": 0.8994466907852735, "grad_norm": 11.6796875, "learning_rate": 9.71741756679932e-06, "loss": 0.9967, "step": 8453 }, { "epoch": 0.8995530964034901, "grad_norm": 12.8671875, "learning_rate": 9.717382035247301e-06, "loss": 1.1267, "step": 8454 }, { "epoch": 0.8996595020217067, "grad_norm": 14.7890625, "learning_rate": 9.717346503695283e-06, "loss": 1.3737, "step": 8455 }, { "epoch": 0.8997659076399234, "grad_norm": 14.1328125, "learning_rate": 9.717310972143264e-06, "loss": 1.1892, "step": 8456 }, { "epoch": 0.89987231325814, "grad_norm": 9.546875, "learning_rate": 9.717275440591246e-06, "loss": 0.9465, "step": 8457 }, { "epoch": 0.8999787188763567, "grad_norm": 9.3203125, "learning_rate": 9.717239909039228e-06, "loss": 1.448, "step": 8458 }, { "epoch": 0.9000851244945733, "grad_norm": 10.8046875, "learning_rate": 9.717204377487209e-06, "loss": 1.1485, "step": 8459 }, { "epoch": 0.90019153011279, "grad_norm": 11.4375, "learning_rate": 9.71716884593519e-06, "loss": 1.5112, "step": 8460 }, { "epoch": 0.9002979357310066, "grad_norm": 10.3984375, "learning_rate": 9.717133314383172e-06, "loss": 1.08, "step": 8461 }, { "epoch": 0.9004043413492232, "grad_norm": 13.5546875, "learning_rate": 9.717097782831156e-06, "loss": 1.04, "step": 8462 }, { "epoch": 0.9005107469674399, "grad_norm": 9.484375, "learning_rate": 9.717062251279137e-06, "loss": 0.9148, "step": 8463 }, { "epoch": 0.9006171525856566, "grad_norm": 12.609375, "learning_rate": 9.717026719727119e-06, "loss": 1.2003, "step": 8464 }, { "epoch": 0.9007235582038732, "grad_norm": 14.2109375, "learning_rate": 9.7169911881751e-06, "loss": 1.1798, "step": 8465 }, { "epoch": 0.9008299638220898, "grad_norm": 12.296875, "learning_rate": 9.716955656623082e-06, "loss": 1.201, "step": 8466 }, { "epoch": 0.9009363694403064, "grad_norm": 11.5859375, "learning_rate": 9.716920125071064e-06, "loss": 0.9082, "step": 8467 }, { "epoch": 0.901042775058523, "grad_norm": 19.609375, "learning_rate": 9.716884593519045e-06, "loss": 1.0664, "step": 8468 }, { "epoch": 0.9011491806767398, "grad_norm": 19.28125, "learning_rate": 9.716849061967027e-06, "loss": 1.1681, "step": 8469 }, { "epoch": 0.9012555862949564, "grad_norm": 13.265625, "learning_rate": 9.716813530415008e-06, "loss": 1.4163, "step": 8470 }, { "epoch": 0.901361991913173, "grad_norm": 11.671875, "learning_rate": 9.71677799886299e-06, "loss": 1.2215, "step": 8471 }, { "epoch": 0.9014683975313896, "grad_norm": 11.9375, "learning_rate": 9.716742467310973e-06, "loss": 1.5964, "step": 8472 }, { "epoch": 0.9015748031496063, "grad_norm": 12.7890625, "learning_rate": 9.716706935758955e-06, "loss": 1.3574, "step": 8473 }, { "epoch": 0.901681208767823, "grad_norm": 11.375, "learning_rate": 9.716671404206937e-06, "loss": 1.1677, "step": 8474 }, { "epoch": 0.9017876143860396, "grad_norm": 15.9375, "learning_rate": 9.716635872654918e-06, "loss": 1.4657, "step": 8475 }, { "epoch": 0.9018940200042562, "grad_norm": 18.84375, "learning_rate": 9.7166003411029e-06, "loss": 1.196, "step": 8476 }, { "epoch": 0.9020004256224728, "grad_norm": 12.5859375, "learning_rate": 9.716564809550883e-06, "loss": 1.2065, "step": 8477 }, { "epoch": 0.9021068312406895, "grad_norm": 11.890625, "learning_rate": 9.716529277998865e-06, "loss": 1.1373, "step": 8478 }, { "epoch": 0.9022132368589062, "grad_norm": 10.75, "learning_rate": 9.716493746446846e-06, "loss": 1.1274, "step": 8479 }, { "epoch": 0.9023196424771228, "grad_norm": 12.6796875, "learning_rate": 9.716458214894828e-06, "loss": 1.2336, "step": 8480 }, { "epoch": 0.9024260480953394, "grad_norm": 14.4765625, "learning_rate": 9.71642268334281e-06, "loss": 1.2071, "step": 8481 }, { "epoch": 0.9025324537135561, "grad_norm": 8.859375, "learning_rate": 9.716387151790791e-06, "loss": 1.1553, "step": 8482 }, { "epoch": 0.9026388593317727, "grad_norm": 20.921875, "learning_rate": 9.716351620238773e-06, "loss": 0.9196, "step": 8483 }, { "epoch": 0.9027452649499894, "grad_norm": 10.890625, "learning_rate": 9.716316088686754e-06, "loss": 1.113, "step": 8484 }, { "epoch": 0.902851670568206, "grad_norm": 11.171875, "learning_rate": 9.716280557134736e-06, "loss": 0.8194, "step": 8485 }, { "epoch": 0.9029580761864227, "grad_norm": 13.21875, "learning_rate": 9.716245025582718e-06, "loss": 1.3541, "step": 8486 }, { "epoch": 0.9030644818046393, "grad_norm": 14.1484375, "learning_rate": 9.716209494030701e-06, "loss": 1.4343, "step": 8487 }, { "epoch": 0.9031708874228559, "grad_norm": 16.34375, "learning_rate": 9.716173962478682e-06, "loss": 1.0968, "step": 8488 }, { "epoch": 0.9032772930410725, "grad_norm": 12.25, "learning_rate": 9.716138430926664e-06, "loss": 1.2833, "step": 8489 }, { "epoch": 0.9033836986592892, "grad_norm": 10.4765625, "learning_rate": 9.716102899374646e-06, "loss": 1.2911, "step": 8490 }, { "epoch": 0.9034901042775059, "grad_norm": 15.109375, "learning_rate": 9.716067367822627e-06, "loss": 1.2201, "step": 8491 }, { "epoch": 0.9035965098957225, "grad_norm": 10.8203125, "learning_rate": 9.716031836270609e-06, "loss": 1.0583, "step": 8492 }, { "epoch": 0.9037029155139391, "grad_norm": 11.0625, "learning_rate": 9.71599630471859e-06, "loss": 1.356, "step": 8493 }, { "epoch": 0.9038093211321557, "grad_norm": 12.3046875, "learning_rate": 9.715960773166572e-06, "loss": 1.2858, "step": 8494 }, { "epoch": 0.9039157267503725, "grad_norm": 15.5078125, "learning_rate": 9.715925241614554e-06, "loss": 1.2876, "step": 8495 }, { "epoch": 0.9040221323685891, "grad_norm": 12.6640625, "learning_rate": 9.715889710062535e-06, "loss": 1.3233, "step": 8496 }, { "epoch": 0.9041285379868057, "grad_norm": 14.921875, "learning_rate": 9.715854178510517e-06, "loss": 1.1497, "step": 8497 }, { "epoch": 0.9042349436050223, "grad_norm": 17.140625, "learning_rate": 9.7158186469585e-06, "loss": 1.1746, "step": 8498 }, { "epoch": 0.9043413492232389, "grad_norm": 13.5703125, "learning_rate": 9.715783115406482e-06, "loss": 1.0723, "step": 8499 }, { "epoch": 0.9044477548414557, "grad_norm": 13.90625, "learning_rate": 9.715747583854463e-06, "loss": 0.9632, "step": 8500 }, { "epoch": 0.9045541604596723, "grad_norm": 14.6328125, "learning_rate": 9.715712052302445e-06, "loss": 0.9377, "step": 8501 }, { "epoch": 0.9046605660778889, "grad_norm": 11.5859375, "learning_rate": 9.715676520750427e-06, "loss": 1.0409, "step": 8502 }, { "epoch": 0.9047669716961055, "grad_norm": 9.0703125, "learning_rate": 9.71564098919841e-06, "loss": 1.1035, "step": 8503 }, { "epoch": 0.9048733773143222, "grad_norm": 12.6484375, "learning_rate": 9.715605457646392e-06, "loss": 1.0997, "step": 8504 }, { "epoch": 0.9049797829325389, "grad_norm": 12.3125, "learning_rate": 9.715569926094373e-06, "loss": 1.5023, "step": 8505 }, { "epoch": 0.9050861885507555, "grad_norm": 15.1484375, "learning_rate": 9.715534394542355e-06, "loss": 1.4561, "step": 8506 }, { "epoch": 0.9051925941689721, "grad_norm": 19.0, "learning_rate": 9.715498862990336e-06, "loss": 1.2983, "step": 8507 }, { "epoch": 0.9052989997871888, "grad_norm": 12.203125, "learning_rate": 9.715463331438318e-06, "loss": 0.976, "step": 8508 }, { "epoch": 0.9054054054054054, "grad_norm": 12.421875, "learning_rate": 9.7154277998863e-06, "loss": 1.4313, "step": 8509 }, { "epoch": 0.905511811023622, "grad_norm": 12.984375, "learning_rate": 9.715392268334281e-06, "loss": 0.8759, "step": 8510 }, { "epoch": 0.9056182166418387, "grad_norm": 11.7109375, "learning_rate": 9.715356736782263e-06, "loss": 0.9178, "step": 8511 }, { "epoch": 0.9057246222600553, "grad_norm": 14.4765625, "learning_rate": 9.715321205230244e-06, "loss": 1.0671, "step": 8512 }, { "epoch": 0.905831027878272, "grad_norm": 13.9609375, "learning_rate": 9.715285673678228e-06, "loss": 1.2374, "step": 8513 }, { "epoch": 0.9059374334964886, "grad_norm": 13.0, "learning_rate": 9.71525014212621e-06, "loss": 1.1488, "step": 8514 }, { "epoch": 0.9060438391147052, "grad_norm": 18.5625, "learning_rate": 9.715214610574191e-06, "loss": 1.6597, "step": 8515 }, { "epoch": 0.9061502447329219, "grad_norm": 18.1875, "learning_rate": 9.715179079022173e-06, "loss": 1.3657, "step": 8516 }, { "epoch": 0.9062566503511386, "grad_norm": 9.90625, "learning_rate": 9.715143547470154e-06, "loss": 1.2669, "step": 8517 }, { "epoch": 0.9063630559693552, "grad_norm": 21.515625, "learning_rate": 9.715108015918136e-06, "loss": 1.3532, "step": 8518 }, { "epoch": 0.9064694615875718, "grad_norm": 11.078125, "learning_rate": 9.715072484366117e-06, "loss": 1.2023, "step": 8519 }, { "epoch": 0.9065758672057884, "grad_norm": 11.7734375, "learning_rate": 9.715036952814099e-06, "loss": 1.4905, "step": 8520 }, { "epoch": 0.9066822728240052, "grad_norm": 13.7578125, "learning_rate": 9.71500142126208e-06, "loss": 1.4631, "step": 8521 }, { "epoch": 0.9067886784422218, "grad_norm": 11.984375, "learning_rate": 9.714965889710062e-06, "loss": 1.0323, "step": 8522 }, { "epoch": 0.9068950840604384, "grad_norm": 12.8671875, "learning_rate": 9.714930358158045e-06, "loss": 1.1322, "step": 8523 }, { "epoch": 0.907001489678655, "grad_norm": 12.640625, "learning_rate": 9.714894826606027e-06, "loss": 1.242, "step": 8524 }, { "epoch": 0.9071078952968716, "grad_norm": 15.3515625, "learning_rate": 9.714859295054009e-06, "loss": 0.8564, "step": 8525 }, { "epoch": 0.9072143009150884, "grad_norm": 15.6875, "learning_rate": 9.71482376350199e-06, "loss": 1.2137, "step": 8526 }, { "epoch": 0.907320706533305, "grad_norm": 10.8359375, "learning_rate": 9.714788231949972e-06, "loss": 1.258, "step": 8527 }, { "epoch": 0.9074271121515216, "grad_norm": 15.84375, "learning_rate": 9.714752700397955e-06, "loss": 1.3354, "step": 8528 }, { "epoch": 0.9075335177697382, "grad_norm": 42.625, "learning_rate": 9.714717168845937e-06, "loss": 1.549, "step": 8529 }, { "epoch": 0.9076399233879549, "grad_norm": 10.9140625, "learning_rate": 9.714681637293918e-06, "loss": 1.1946, "step": 8530 }, { "epoch": 0.9077463290061715, "grad_norm": 12.3515625, "learning_rate": 9.7146461057419e-06, "loss": 1.2861, "step": 8531 }, { "epoch": 0.9078527346243882, "grad_norm": 11.8828125, "learning_rate": 9.714610574189882e-06, "loss": 1.3525, "step": 8532 }, { "epoch": 0.9079591402426048, "grad_norm": 10.5703125, "learning_rate": 9.714575042637863e-06, "loss": 1.0833, "step": 8533 }, { "epoch": 0.9080655458608214, "grad_norm": 13.1640625, "learning_rate": 9.714539511085845e-06, "loss": 1.1072, "step": 8534 }, { "epoch": 0.9081719514790381, "grad_norm": 12.3125, "learning_rate": 9.714503979533826e-06, "loss": 1.6418, "step": 8535 }, { "epoch": 0.9082783570972547, "grad_norm": 14.5546875, "learning_rate": 9.714468447981808e-06, "loss": 1.3034, "step": 8536 }, { "epoch": 0.9083847627154714, "grad_norm": 16.109375, "learning_rate": 9.71443291642979e-06, "loss": 1.3745, "step": 8537 }, { "epoch": 0.908491168333688, "grad_norm": 9.609375, "learning_rate": 9.714397384877771e-06, "loss": 1.1624, "step": 8538 }, { "epoch": 0.9085975739519047, "grad_norm": 10.3203125, "learning_rate": 9.714361853325755e-06, "loss": 1.0502, "step": 8539 }, { "epoch": 0.9087039795701213, "grad_norm": 14.2265625, "learning_rate": 9.714326321773736e-06, "loss": 1.2959, "step": 8540 }, { "epoch": 0.9088103851883379, "grad_norm": 12.5390625, "learning_rate": 9.714290790221718e-06, "loss": 1.263, "step": 8541 }, { "epoch": 0.9089167908065546, "grad_norm": 14.3046875, "learning_rate": 9.7142552586697e-06, "loss": 1.4931, "step": 8542 }, { "epoch": 0.9090231964247713, "grad_norm": 11.2265625, "learning_rate": 9.714219727117681e-06, "loss": 1.4535, "step": 8543 }, { "epoch": 0.9091296020429879, "grad_norm": 11.5, "learning_rate": 9.714184195565664e-06, "loss": 1.1361, "step": 8544 }, { "epoch": 0.9092360076612045, "grad_norm": 11.46875, "learning_rate": 9.714148664013646e-06, "loss": 1.1255, "step": 8545 }, { "epoch": 0.9093424132794211, "grad_norm": 12.3828125, "learning_rate": 9.714113132461626e-06, "loss": 1.2066, "step": 8546 }, { "epoch": 0.9094488188976378, "grad_norm": 10.2890625, "learning_rate": 9.714077600909607e-06, "loss": 1.3125, "step": 8547 }, { "epoch": 0.9095552245158545, "grad_norm": 20.484375, "learning_rate": 9.714042069357589e-06, "loss": 1.6386, "step": 8548 }, { "epoch": 0.9096616301340711, "grad_norm": 11.90625, "learning_rate": 9.714006537805572e-06, "loss": 1.0685, "step": 8549 }, { "epoch": 0.9097680357522877, "grad_norm": 11.453125, "learning_rate": 9.713971006253554e-06, "loss": 1.2317, "step": 8550 }, { "epoch": 0.9098744413705043, "grad_norm": 19.328125, "learning_rate": 9.713935474701536e-06, "loss": 1.3485, "step": 8551 }, { "epoch": 0.9099808469887211, "grad_norm": 15.21875, "learning_rate": 9.713899943149517e-06, "loss": 1.6351, "step": 8552 }, { "epoch": 0.9100872526069377, "grad_norm": 9.984375, "learning_rate": 9.713864411597499e-06, "loss": 1.4277, "step": 8553 }, { "epoch": 0.9101936582251543, "grad_norm": 9.3203125, "learning_rate": 9.713828880045482e-06, "loss": 1.0651, "step": 8554 }, { "epoch": 0.9103000638433709, "grad_norm": 11.3515625, "learning_rate": 9.713793348493464e-06, "loss": 1.2681, "step": 8555 }, { "epoch": 0.9104064694615875, "grad_norm": 10.7109375, "learning_rate": 9.713757816941445e-06, "loss": 1.3434, "step": 8556 }, { "epoch": 0.9105128750798042, "grad_norm": 13.0703125, "learning_rate": 9.713722285389427e-06, "loss": 1.2512, "step": 8557 }, { "epoch": 0.9106192806980209, "grad_norm": 13.4140625, "learning_rate": 9.713686753837409e-06, "loss": 1.3954, "step": 8558 }, { "epoch": 0.9107256863162375, "grad_norm": 10.7265625, "learning_rate": 9.71365122228539e-06, "loss": 1.0109, "step": 8559 }, { "epoch": 0.9108320919344541, "grad_norm": 12.0234375, "learning_rate": 9.713615690733372e-06, "loss": 1.0171, "step": 8560 }, { "epoch": 0.9109384975526708, "grad_norm": 12.765625, "learning_rate": 9.713580159181353e-06, "loss": 1.0325, "step": 8561 }, { "epoch": 0.9110449031708874, "grad_norm": 12.4375, "learning_rate": 9.713544627629335e-06, "loss": 1.0221, "step": 8562 }, { "epoch": 0.9111513087891041, "grad_norm": 16.375, "learning_rate": 9.713509096077317e-06, "loss": 1.1586, "step": 8563 }, { "epoch": 0.9112577144073207, "grad_norm": 11.7734375, "learning_rate": 9.7134735645253e-06, "loss": 1.2067, "step": 8564 }, { "epoch": 0.9113641200255373, "grad_norm": 22.140625, "learning_rate": 9.713438032973281e-06, "loss": 0.9984, "step": 8565 }, { "epoch": 0.911470525643754, "grad_norm": 11.4140625, "learning_rate": 9.713402501421263e-06, "loss": 1.5376, "step": 8566 }, { "epoch": 0.9115769312619706, "grad_norm": 14.5390625, "learning_rate": 9.713366969869245e-06, "loss": 1.3436, "step": 8567 }, { "epoch": 0.9116833368801873, "grad_norm": 9.484375, "learning_rate": 9.713331438317226e-06, "loss": 1.1289, "step": 8568 }, { "epoch": 0.9117897424984039, "grad_norm": 13.0390625, "learning_rate": 9.71329590676521e-06, "loss": 0.9601, "step": 8569 }, { "epoch": 0.9118961481166206, "grad_norm": 12.0390625, "learning_rate": 9.713260375213191e-06, "loss": 0.9276, "step": 8570 }, { "epoch": 0.9120025537348372, "grad_norm": 17.40625, "learning_rate": 9.713224843661173e-06, "loss": 1.5166, "step": 8571 }, { "epoch": 0.9121089593530538, "grad_norm": 12.6484375, "learning_rate": 9.713189312109154e-06, "loss": 1.4049, "step": 8572 }, { "epoch": 0.9122153649712705, "grad_norm": 11.515625, "learning_rate": 9.713153780557136e-06, "loss": 1.2688, "step": 8573 }, { "epoch": 0.9123217705894872, "grad_norm": 9.6640625, "learning_rate": 9.713118249005118e-06, "loss": 1.3134, "step": 8574 }, { "epoch": 0.9124281762077038, "grad_norm": 19.09375, "learning_rate": 9.7130827174531e-06, "loss": 0.8362, "step": 8575 }, { "epoch": 0.9125345818259204, "grad_norm": 13.078125, "learning_rate": 9.71304718590108e-06, "loss": 1.1592, "step": 8576 }, { "epoch": 0.912640987444137, "grad_norm": 10.9609375, "learning_rate": 9.713011654349062e-06, "loss": 1.0751, "step": 8577 }, { "epoch": 0.9127473930623536, "grad_norm": 12.078125, "learning_rate": 9.712976122797044e-06, "loss": 1.4085, "step": 8578 }, { "epoch": 0.9128537986805704, "grad_norm": 12.78125, "learning_rate": 9.712940591245026e-06, "loss": 1.1225, "step": 8579 }, { "epoch": 0.912960204298787, "grad_norm": 15.3359375, "learning_rate": 9.712905059693009e-06, "loss": 1.4505, "step": 8580 }, { "epoch": 0.9130666099170036, "grad_norm": 12.7734375, "learning_rate": 9.71286952814099e-06, "loss": 0.9494, "step": 8581 }, { "epoch": 0.9131730155352202, "grad_norm": 13.765625, "learning_rate": 9.712833996588972e-06, "loss": 1.0318, "step": 8582 }, { "epoch": 0.9132794211534369, "grad_norm": 12.1796875, "learning_rate": 9.712798465036954e-06, "loss": 1.0257, "step": 8583 }, { "epoch": 0.9133858267716536, "grad_norm": 14.1640625, "learning_rate": 9.712762933484935e-06, "loss": 1.2536, "step": 8584 }, { "epoch": 0.9134922323898702, "grad_norm": 13.2421875, "learning_rate": 9.712727401932917e-06, "loss": 1.074, "step": 8585 }, { "epoch": 0.9135986380080868, "grad_norm": 12.9609375, "learning_rate": 9.712691870380899e-06, "loss": 1.1192, "step": 8586 }, { "epoch": 0.9137050436263034, "grad_norm": 11.953125, "learning_rate": 9.71265633882888e-06, "loss": 1.1739, "step": 8587 }, { "epoch": 0.9138114492445201, "grad_norm": 13.2578125, "learning_rate": 9.712620807276862e-06, "loss": 1.2697, "step": 8588 }, { "epoch": 0.9139178548627368, "grad_norm": 13.1328125, "learning_rate": 9.712585275724843e-06, "loss": 1.5295, "step": 8589 }, { "epoch": 0.9140242604809534, "grad_norm": 11.9140625, "learning_rate": 9.712549744172827e-06, "loss": 1.3078, "step": 8590 }, { "epoch": 0.91413066609917, "grad_norm": 11.3671875, "learning_rate": 9.712514212620808e-06, "loss": 1.1136, "step": 8591 }, { "epoch": 0.9142370717173867, "grad_norm": 10.0625, "learning_rate": 9.71247868106879e-06, "loss": 1.2142, "step": 8592 }, { "epoch": 0.9143434773356033, "grad_norm": 13.34375, "learning_rate": 9.712443149516772e-06, "loss": 1.2081, "step": 8593 }, { "epoch": 0.91444988295382, "grad_norm": 15.6875, "learning_rate": 9.712407617964753e-06, "loss": 1.058, "step": 8594 }, { "epoch": 0.9145562885720366, "grad_norm": 21.5625, "learning_rate": 9.712372086412736e-06, "loss": 1.1409, "step": 8595 }, { "epoch": 0.9146626941902533, "grad_norm": 13.1875, "learning_rate": 9.712336554860718e-06, "loss": 1.1274, "step": 8596 }, { "epoch": 0.9147690998084699, "grad_norm": 15.3203125, "learning_rate": 9.7123010233087e-06, "loss": 0.9959, "step": 8597 }, { "epoch": 0.9148755054266865, "grad_norm": 13.25, "learning_rate": 9.712265491756681e-06, "loss": 1.1831, "step": 8598 }, { "epoch": 0.9149819110449031, "grad_norm": 11.25, "learning_rate": 9.712229960204663e-06, "loss": 1.0847, "step": 8599 }, { "epoch": 0.9150883166631198, "grad_norm": 23.171875, "learning_rate": 9.712194428652644e-06, "loss": 1.2973, "step": 8600 }, { "epoch": 0.9151947222813365, "grad_norm": 12.75, "learning_rate": 9.712158897100626e-06, "loss": 1.3583, "step": 8601 }, { "epoch": 0.9153011278995531, "grad_norm": 13.71875, "learning_rate": 9.712123365548608e-06, "loss": 1.1555, "step": 8602 }, { "epoch": 0.9154075335177697, "grad_norm": 12.8515625, "learning_rate": 9.71208783399659e-06, "loss": 1.0162, "step": 8603 }, { "epoch": 0.9155139391359863, "grad_norm": 11.1015625, "learning_rate": 9.712052302444571e-06, "loss": 1.3059, "step": 8604 }, { "epoch": 0.9156203447542031, "grad_norm": 14.8671875, "learning_rate": 9.712016770892554e-06, "loss": 1.087, "step": 8605 }, { "epoch": 0.9157267503724197, "grad_norm": 12.2890625, "learning_rate": 9.711981239340536e-06, "loss": 1.1273, "step": 8606 }, { "epoch": 0.9158331559906363, "grad_norm": 12.03125, "learning_rate": 9.711945707788517e-06, "loss": 0.931, "step": 8607 }, { "epoch": 0.9159395616088529, "grad_norm": 11.6484375, "learning_rate": 9.711910176236499e-06, "loss": 0.9534, "step": 8608 }, { "epoch": 0.9160459672270695, "grad_norm": 12.0859375, "learning_rate": 9.71187464468448e-06, "loss": 1.1104, "step": 8609 }, { "epoch": 0.9161523728452863, "grad_norm": 12.53125, "learning_rate": 9.711839113132462e-06, "loss": 1.0125, "step": 8610 }, { "epoch": 0.9162587784635029, "grad_norm": 12.15625, "learning_rate": 9.711803581580444e-06, "loss": 1.4133, "step": 8611 }, { "epoch": 0.9163651840817195, "grad_norm": 12.9453125, "learning_rate": 9.711768050028425e-06, "loss": 1.1109, "step": 8612 }, { "epoch": 0.9164715896999361, "grad_norm": 16.734375, "learning_rate": 9.711732518476407e-06, "loss": 1.465, "step": 8613 }, { "epoch": 0.9165779953181528, "grad_norm": 13.875, "learning_rate": 9.711696986924389e-06, "loss": 1.0525, "step": 8614 }, { "epoch": 0.9166844009363695, "grad_norm": 17.765625, "learning_rate": 9.71166145537237e-06, "loss": 0.9535, "step": 8615 }, { "epoch": 0.9167908065545861, "grad_norm": 9.3671875, "learning_rate": 9.711625923820354e-06, "loss": 0.9383, "step": 8616 }, { "epoch": 0.9168972121728027, "grad_norm": 11.9921875, "learning_rate": 9.711590392268335e-06, "loss": 1.2498, "step": 8617 }, { "epoch": 0.9170036177910194, "grad_norm": 9.21875, "learning_rate": 9.711554860716317e-06, "loss": 1.1899, "step": 8618 }, { "epoch": 0.917110023409236, "grad_norm": 11.8203125, "learning_rate": 9.711519329164298e-06, "loss": 1.1226, "step": 8619 }, { "epoch": 0.9172164290274526, "grad_norm": 13.34375, "learning_rate": 9.71148379761228e-06, "loss": 1.1899, "step": 8620 }, { "epoch": 0.9173228346456693, "grad_norm": 10.875, "learning_rate": 9.711448266060263e-06, "loss": 1.2281, "step": 8621 }, { "epoch": 0.917429240263886, "grad_norm": 12.421875, "learning_rate": 9.711412734508245e-06, "loss": 1.3645, "step": 8622 }, { "epoch": 0.9175356458821026, "grad_norm": 12.1328125, "learning_rate": 9.711377202956226e-06, "loss": 1.0664, "step": 8623 }, { "epoch": 0.9176420515003192, "grad_norm": 13.2109375, "learning_rate": 9.711341671404208e-06, "loss": 1.1578, "step": 8624 }, { "epoch": 0.9177484571185358, "grad_norm": 13.0859375, "learning_rate": 9.71130613985219e-06, "loss": 1.249, "step": 8625 }, { "epoch": 0.9178548627367525, "grad_norm": 14.203125, "learning_rate": 9.711270608300171e-06, "loss": 1.5788, "step": 8626 }, { "epoch": 0.9179612683549692, "grad_norm": 10.890625, "learning_rate": 9.711235076748153e-06, "loss": 1.0009, "step": 8627 }, { "epoch": 0.9180676739731858, "grad_norm": 14.546875, "learning_rate": 9.711199545196135e-06, "loss": 1.3399, "step": 8628 }, { "epoch": 0.9181740795914024, "grad_norm": 15.9765625, "learning_rate": 9.711164013644116e-06, "loss": 0.9271, "step": 8629 }, { "epoch": 0.918280485209619, "grad_norm": 14.3984375, "learning_rate": 9.711128482092098e-06, "loss": 1.0299, "step": 8630 }, { "epoch": 0.9183868908278358, "grad_norm": 12.0859375, "learning_rate": 9.711092950540081e-06, "loss": 1.0981, "step": 8631 }, { "epoch": 0.9184932964460524, "grad_norm": 11.46875, "learning_rate": 9.711057418988063e-06, "loss": 0.9888, "step": 8632 }, { "epoch": 0.918599702064269, "grad_norm": 12.2265625, "learning_rate": 9.711021887436044e-06, "loss": 1.1408, "step": 8633 }, { "epoch": 0.9187061076824856, "grad_norm": 14.859375, "learning_rate": 9.710986355884026e-06, "loss": 1.3065, "step": 8634 }, { "epoch": 0.9188125133007022, "grad_norm": 8.953125, "learning_rate": 9.710950824332007e-06, "loss": 1.066, "step": 8635 }, { "epoch": 0.918918918918919, "grad_norm": 7.78125, "learning_rate": 9.710915292779989e-06, "loss": 1.0853, "step": 8636 }, { "epoch": 0.9190253245371356, "grad_norm": 14.7421875, "learning_rate": 9.71087976122797e-06, "loss": 1.2927, "step": 8637 }, { "epoch": 0.9191317301553522, "grad_norm": 14.234375, "learning_rate": 9.710844229675952e-06, "loss": 1.4661, "step": 8638 }, { "epoch": 0.9192381357735688, "grad_norm": 14.0703125, "learning_rate": 9.710808698123934e-06, "loss": 1.0996, "step": 8639 }, { "epoch": 0.9193445413917855, "grad_norm": 16.015625, "learning_rate": 9.710773166571915e-06, "loss": 1.2236, "step": 8640 }, { "epoch": 0.9194509470100021, "grad_norm": 13.0234375, "learning_rate": 9.710737635019899e-06, "loss": 1.2837, "step": 8641 }, { "epoch": 0.9195573526282188, "grad_norm": 12.703125, "learning_rate": 9.71070210346788e-06, "loss": 1.0437, "step": 8642 }, { "epoch": 0.9196637582464354, "grad_norm": 11.1953125, "learning_rate": 9.710666571915862e-06, "loss": 1.1087, "step": 8643 }, { "epoch": 0.919770163864652, "grad_norm": 9.2265625, "learning_rate": 9.710631040363844e-06, "loss": 1.0876, "step": 8644 }, { "epoch": 0.9198765694828687, "grad_norm": 15.6640625, "learning_rate": 9.710595508811825e-06, "loss": 1.4778, "step": 8645 }, { "epoch": 0.9199829751010853, "grad_norm": 11.140625, "learning_rate": 9.710559977259809e-06, "loss": 1.2503, "step": 8646 }, { "epoch": 0.920089380719302, "grad_norm": 11.328125, "learning_rate": 9.71052444570779e-06, "loss": 1.1711, "step": 8647 }, { "epoch": 0.9201957863375186, "grad_norm": 10.6015625, "learning_rate": 9.710488914155772e-06, "loss": 1.1908, "step": 8648 }, { "epoch": 0.9203021919557353, "grad_norm": 13.8828125, "learning_rate": 9.710453382603753e-06, "loss": 1.0698, "step": 8649 }, { "epoch": 0.9204085975739519, "grad_norm": 9.1484375, "learning_rate": 9.710417851051735e-06, "loss": 1.0756, "step": 8650 }, { "epoch": 0.9205150031921685, "grad_norm": 11.46875, "learning_rate": 9.710382319499717e-06, "loss": 1.18, "step": 8651 }, { "epoch": 0.9206214088103852, "grad_norm": 14.7421875, "learning_rate": 9.710346787947698e-06, "loss": 1.2687, "step": 8652 }, { "epoch": 0.9207278144286019, "grad_norm": 10.96875, "learning_rate": 9.71031125639568e-06, "loss": 1.1482, "step": 8653 }, { "epoch": 0.9208342200468185, "grad_norm": 11.734375, "learning_rate": 9.710275724843661e-06, "loss": 1.1412, "step": 8654 }, { "epoch": 0.9209406256650351, "grad_norm": 11.3359375, "learning_rate": 9.710240193291643e-06, "loss": 1.0543, "step": 8655 }, { "epoch": 0.9210470312832517, "grad_norm": 17.59375, "learning_rate": 9.710204661739625e-06, "loss": 1.2878, "step": 8656 }, { "epoch": 0.9211534369014684, "grad_norm": 13.6796875, "learning_rate": 9.710169130187608e-06, "loss": 1.4005, "step": 8657 }, { "epoch": 0.9212598425196851, "grad_norm": 12.5234375, "learning_rate": 9.71013359863559e-06, "loss": 0.8461, "step": 8658 }, { "epoch": 0.9213662481379017, "grad_norm": 10.2421875, "learning_rate": 9.710098067083571e-06, "loss": 1.3236, "step": 8659 }, { "epoch": 0.9214726537561183, "grad_norm": 11.46875, "learning_rate": 9.710062535531553e-06, "loss": 1.0753, "step": 8660 }, { "epoch": 0.9215790593743349, "grad_norm": 10.65625, "learning_rate": 9.710027003979534e-06, "loss": 1.6906, "step": 8661 }, { "epoch": 0.9216854649925517, "grad_norm": 9.9765625, "learning_rate": 9.709991472427518e-06, "loss": 1.1057, "step": 8662 }, { "epoch": 0.9217918706107683, "grad_norm": 17.265625, "learning_rate": 9.7099559408755e-06, "loss": 1.0972, "step": 8663 }, { "epoch": 0.9218982762289849, "grad_norm": 9.3359375, "learning_rate": 9.70992040932348e-06, "loss": 1.0478, "step": 8664 }, { "epoch": 0.9220046818472015, "grad_norm": 12.328125, "learning_rate": 9.709884877771462e-06, "loss": 1.1122, "step": 8665 }, { "epoch": 0.9221110874654181, "grad_norm": 11.640625, "learning_rate": 9.709849346219442e-06, "loss": 1.3118, "step": 8666 }, { "epoch": 0.9222174930836348, "grad_norm": 9.4921875, "learning_rate": 9.709813814667426e-06, "loss": 1.155, "step": 8667 }, { "epoch": 0.9223238987018515, "grad_norm": 11.734375, "learning_rate": 9.709778283115407e-06, "loss": 0.8944, "step": 8668 }, { "epoch": 0.9224303043200681, "grad_norm": 14.125, "learning_rate": 9.709742751563389e-06, "loss": 1.4102, "step": 8669 }, { "epoch": 0.9225367099382847, "grad_norm": 13.375, "learning_rate": 9.70970722001137e-06, "loss": 1.2041, "step": 8670 }, { "epoch": 0.9226431155565014, "grad_norm": 10.046875, "learning_rate": 9.709671688459352e-06, "loss": 1.0525, "step": 8671 }, { "epoch": 0.922749521174718, "grad_norm": 13.6953125, "learning_rate": 9.709636156907335e-06, "loss": 1.2481, "step": 8672 }, { "epoch": 0.9228559267929347, "grad_norm": 11.703125, "learning_rate": 9.709600625355317e-06, "loss": 0.9594, "step": 8673 }, { "epoch": 0.9229623324111513, "grad_norm": 9.4296875, "learning_rate": 9.709565093803299e-06, "loss": 1.1307, "step": 8674 }, { "epoch": 0.923068738029368, "grad_norm": 11.6328125, "learning_rate": 9.70952956225128e-06, "loss": 1.1582, "step": 8675 }, { "epoch": 0.9231751436475846, "grad_norm": 14.171875, "learning_rate": 9.709494030699262e-06, "loss": 1.4228, "step": 8676 }, { "epoch": 0.9232815492658012, "grad_norm": 14.0546875, "learning_rate": 9.709458499147243e-06, "loss": 1.3891, "step": 8677 }, { "epoch": 0.9233879548840179, "grad_norm": 11.2890625, "learning_rate": 9.709422967595225e-06, "loss": 1.4273, "step": 8678 }, { "epoch": 0.9234943605022345, "grad_norm": 17.40625, "learning_rate": 9.709387436043207e-06, "loss": 1.2699, "step": 8679 }, { "epoch": 0.9236007661204512, "grad_norm": 19.390625, "learning_rate": 9.709351904491188e-06, "loss": 1.6459, "step": 8680 }, { "epoch": 0.9237071717386678, "grad_norm": 16.109375, "learning_rate": 9.70931637293917e-06, "loss": 1.3505, "step": 8681 }, { "epoch": 0.9238135773568844, "grad_norm": 10.734375, "learning_rate": 9.709280841387153e-06, "loss": 1.3716, "step": 8682 }, { "epoch": 0.9239199829751011, "grad_norm": 10.9609375, "learning_rate": 9.709245309835135e-06, "loss": 1.0471, "step": 8683 }, { "epoch": 0.9240263885933178, "grad_norm": 9.9453125, "learning_rate": 9.709209778283116e-06, "loss": 1.3199, "step": 8684 }, { "epoch": 0.9241327942115344, "grad_norm": 12.2890625, "learning_rate": 9.709174246731098e-06, "loss": 1.1863, "step": 8685 }, { "epoch": 0.924239199829751, "grad_norm": 11.234375, "learning_rate": 9.70913871517908e-06, "loss": 1.3895, "step": 8686 }, { "epoch": 0.9243456054479676, "grad_norm": 13.578125, "learning_rate": 9.709103183627063e-06, "loss": 1.1625, "step": 8687 }, { "epoch": 0.9244520110661842, "grad_norm": 13.15625, "learning_rate": 9.709067652075044e-06, "loss": 1.663, "step": 8688 }, { "epoch": 0.924558416684401, "grad_norm": 16.671875, "learning_rate": 9.709032120523026e-06, "loss": 0.9153, "step": 8689 }, { "epoch": 0.9246648223026176, "grad_norm": 10.484375, "learning_rate": 9.708996588971008e-06, "loss": 1.3911, "step": 8690 }, { "epoch": 0.9247712279208342, "grad_norm": 13.3046875, "learning_rate": 9.70896105741899e-06, "loss": 1.5143, "step": 8691 }, { "epoch": 0.9248776335390508, "grad_norm": 20.203125, "learning_rate": 9.708925525866971e-06, "loss": 1.23, "step": 8692 }, { "epoch": 0.9249840391572675, "grad_norm": 10.7578125, "learning_rate": 9.708889994314953e-06, "loss": 1.2822, "step": 8693 }, { "epoch": 0.9250904447754842, "grad_norm": 12.6328125, "learning_rate": 9.708854462762934e-06, "loss": 1.4189, "step": 8694 }, { "epoch": 0.9251968503937008, "grad_norm": 9.90625, "learning_rate": 9.708818931210916e-06, "loss": 0.895, "step": 8695 }, { "epoch": 0.9253032560119174, "grad_norm": 13.84375, "learning_rate": 9.708783399658897e-06, "loss": 1.2326, "step": 8696 }, { "epoch": 0.925409661630134, "grad_norm": 10.53125, "learning_rate": 9.708747868106879e-06, "loss": 1.0145, "step": 8697 }, { "epoch": 0.9255160672483507, "grad_norm": 12.203125, "learning_rate": 9.708712336554862e-06, "loss": 1.0902, "step": 8698 }, { "epoch": 0.9256224728665674, "grad_norm": 15.4453125, "learning_rate": 9.708676805002844e-06, "loss": 1.0856, "step": 8699 }, { "epoch": 0.925728878484784, "grad_norm": 11.6796875, "learning_rate": 9.708641273450825e-06, "loss": 1.2354, "step": 8700 }, { "epoch": 0.9258352841030006, "grad_norm": 13.296875, "learning_rate": 9.708605741898807e-06, "loss": 1.0807, "step": 8701 }, { "epoch": 0.9259416897212173, "grad_norm": 13.1953125, "learning_rate": 9.708570210346789e-06, "loss": 1.0321, "step": 8702 }, { "epoch": 0.9260480953394339, "grad_norm": 10.21875, "learning_rate": 9.70853467879477e-06, "loss": 1.0412, "step": 8703 }, { "epoch": 0.9261545009576506, "grad_norm": 12.6484375, "learning_rate": 9.708499147242752e-06, "loss": 1.0323, "step": 8704 }, { "epoch": 0.9262609065758672, "grad_norm": 11.8125, "learning_rate": 9.708463615690733e-06, "loss": 0.9284, "step": 8705 }, { "epoch": 0.9263673121940839, "grad_norm": 16.890625, "learning_rate": 9.708428084138715e-06, "loss": 1.0217, "step": 8706 }, { "epoch": 0.9264737178123005, "grad_norm": 11.859375, "learning_rate": 9.708392552586697e-06, "loss": 1.0894, "step": 8707 }, { "epoch": 0.9265801234305171, "grad_norm": 7.69140625, "learning_rate": 9.70835702103468e-06, "loss": 0.682, "step": 8708 }, { "epoch": 0.9266865290487337, "grad_norm": 14.15625, "learning_rate": 9.708321489482662e-06, "loss": 1.2546, "step": 8709 }, { "epoch": 0.9267929346669505, "grad_norm": 14.8125, "learning_rate": 9.708285957930643e-06, "loss": 1.1257, "step": 8710 }, { "epoch": 0.9268993402851671, "grad_norm": 12.1171875, "learning_rate": 9.708250426378625e-06, "loss": 1.3763, "step": 8711 }, { "epoch": 0.9270057459033837, "grad_norm": 12.3828125, "learning_rate": 9.708214894826606e-06, "loss": 1.0368, "step": 8712 }, { "epoch": 0.9271121515216003, "grad_norm": 23.875, "learning_rate": 9.70817936327459e-06, "loss": 1.3752, "step": 8713 }, { "epoch": 0.9272185571398169, "grad_norm": 9.84375, "learning_rate": 9.708143831722571e-06, "loss": 1.0526, "step": 8714 }, { "epoch": 0.9273249627580337, "grad_norm": 13.3515625, "learning_rate": 9.708108300170553e-06, "loss": 0.9337, "step": 8715 }, { "epoch": 0.9274313683762503, "grad_norm": 17.21875, "learning_rate": 9.708072768618535e-06, "loss": 1.2456, "step": 8716 }, { "epoch": 0.9275377739944669, "grad_norm": 13.78125, "learning_rate": 9.708037237066516e-06, "loss": 1.4839, "step": 8717 }, { "epoch": 0.9276441796126835, "grad_norm": 11.75, "learning_rate": 9.708001705514498e-06, "loss": 1.1213, "step": 8718 }, { "epoch": 0.9277505852309001, "grad_norm": 10.8671875, "learning_rate": 9.70796617396248e-06, "loss": 1.2297, "step": 8719 }, { "epoch": 0.9278569908491169, "grad_norm": 13.0078125, "learning_rate": 9.707930642410461e-06, "loss": 1.3568, "step": 8720 }, { "epoch": 0.9279633964673335, "grad_norm": 13.1796875, "learning_rate": 9.707895110858443e-06, "loss": 1.0664, "step": 8721 }, { "epoch": 0.9280698020855501, "grad_norm": 10.6328125, "learning_rate": 9.707859579306424e-06, "loss": 1.0687, "step": 8722 }, { "epoch": 0.9281762077037667, "grad_norm": 13.421875, "learning_rate": 9.707824047754407e-06, "loss": 0.9934, "step": 8723 }, { "epoch": 0.9282826133219834, "grad_norm": 10.1953125, "learning_rate": 9.707788516202389e-06, "loss": 0.8565, "step": 8724 }, { "epoch": 0.9283890189402001, "grad_norm": 10.0859375, "learning_rate": 9.70775298465037e-06, "loss": 1.2727, "step": 8725 }, { "epoch": 0.9284954245584167, "grad_norm": 13.875, "learning_rate": 9.707717453098352e-06, "loss": 0.9745, "step": 8726 }, { "epoch": 0.9286018301766333, "grad_norm": 15.78125, "learning_rate": 9.707681921546334e-06, "loss": 1.0764, "step": 8727 }, { "epoch": 0.92870823579485, "grad_norm": 15.515625, "learning_rate": 9.707646389994316e-06, "loss": 1.1533, "step": 8728 }, { "epoch": 0.9288146414130666, "grad_norm": 11.4296875, "learning_rate": 9.707610858442297e-06, "loss": 1.0363, "step": 8729 }, { "epoch": 0.9289210470312832, "grad_norm": 14.2578125, "learning_rate": 9.707575326890279e-06, "loss": 1.2093, "step": 8730 }, { "epoch": 0.9290274526494999, "grad_norm": 14.0625, "learning_rate": 9.70753979533826e-06, "loss": 1.3174, "step": 8731 }, { "epoch": 0.9291338582677166, "grad_norm": 14.703125, "learning_rate": 9.707504263786242e-06, "loss": 1.5561, "step": 8732 }, { "epoch": 0.9292402638859332, "grad_norm": 16.484375, "learning_rate": 9.707468732234224e-06, "loss": 1.5923, "step": 8733 }, { "epoch": 0.9293466695041498, "grad_norm": 12.4140625, "learning_rate": 9.707433200682207e-06, "loss": 1.2452, "step": 8734 }, { "epoch": 0.9294530751223664, "grad_norm": 17.734375, "learning_rate": 9.707397669130188e-06, "loss": 1.3748, "step": 8735 }, { "epoch": 0.9295594807405831, "grad_norm": 13.1171875, "learning_rate": 9.70736213757817e-06, "loss": 1.1709, "step": 8736 }, { "epoch": 0.9296658863587998, "grad_norm": 12.8203125, "learning_rate": 9.707326606026152e-06, "loss": 1.2004, "step": 8737 }, { "epoch": 0.9297722919770164, "grad_norm": 9.9765625, "learning_rate": 9.707291074474133e-06, "loss": 1.1487, "step": 8738 }, { "epoch": 0.929878697595233, "grad_norm": 14.4609375, "learning_rate": 9.707255542922117e-06, "loss": 0.8792, "step": 8739 }, { "epoch": 0.9299851032134496, "grad_norm": 13.4453125, "learning_rate": 9.707220011370098e-06, "loss": 1.2969, "step": 8740 }, { "epoch": 0.9300915088316664, "grad_norm": 12.5, "learning_rate": 9.70718447981808e-06, "loss": 1.0513, "step": 8741 }, { "epoch": 0.930197914449883, "grad_norm": 19.546875, "learning_rate": 9.707148948266061e-06, "loss": 1.4681, "step": 8742 }, { "epoch": 0.9303043200680996, "grad_norm": 11.875, "learning_rate": 9.707113416714043e-06, "loss": 0.9828, "step": 8743 }, { "epoch": 0.9304107256863162, "grad_norm": 10.90625, "learning_rate": 9.707077885162025e-06, "loss": 0.8109, "step": 8744 }, { "epoch": 0.9305171313045328, "grad_norm": 19.453125, "learning_rate": 9.707042353610006e-06, "loss": 1.2339, "step": 8745 }, { "epoch": 0.9306235369227496, "grad_norm": 14.1875, "learning_rate": 9.707006822057988e-06, "loss": 1.2229, "step": 8746 }, { "epoch": 0.9307299425409662, "grad_norm": 12.0625, "learning_rate": 9.70697129050597e-06, "loss": 1.57, "step": 8747 }, { "epoch": 0.9308363481591828, "grad_norm": 13.765625, "learning_rate": 9.706935758953951e-06, "loss": 1.0391, "step": 8748 }, { "epoch": 0.9309427537773994, "grad_norm": 9.0, "learning_rate": 9.706900227401934e-06, "loss": 1.1143, "step": 8749 }, { "epoch": 0.931049159395616, "grad_norm": 9.6796875, "learning_rate": 9.706864695849916e-06, "loss": 1.2626, "step": 8750 }, { "epoch": 0.9311555650138327, "grad_norm": 11.8125, "learning_rate": 9.706829164297898e-06, "loss": 1.6055, "step": 8751 }, { "epoch": 0.9312619706320494, "grad_norm": 14.4765625, "learning_rate": 9.70679363274588e-06, "loss": 1.4922, "step": 8752 }, { "epoch": 0.931368376250266, "grad_norm": 10.546875, "learning_rate": 9.70675810119386e-06, "loss": 1.2532, "step": 8753 }, { "epoch": 0.9314747818684826, "grad_norm": 13.125, "learning_rate": 9.706722569641844e-06, "loss": 1.3378, "step": 8754 }, { "epoch": 0.9315811874866993, "grad_norm": 12.546875, "learning_rate": 9.706687038089824e-06, "loss": 1.1077, "step": 8755 }, { "epoch": 0.9316875931049159, "grad_norm": 11.8671875, "learning_rate": 9.706651506537806e-06, "loss": 1.3198, "step": 8756 }, { "epoch": 0.9317939987231326, "grad_norm": 10.0234375, "learning_rate": 9.706615974985787e-06, "loss": 1.3888, "step": 8757 }, { "epoch": 0.9319004043413492, "grad_norm": 19.078125, "learning_rate": 9.706580443433769e-06, "loss": 1.1352, "step": 8758 }, { "epoch": 0.9320068099595659, "grad_norm": 13.2421875, "learning_rate": 9.706544911881752e-06, "loss": 1.4648, "step": 8759 }, { "epoch": 0.9321132155777825, "grad_norm": 10.5859375, "learning_rate": 9.706509380329734e-06, "loss": 1.036, "step": 8760 }, { "epoch": 0.9322196211959991, "grad_norm": 10.8828125, "learning_rate": 9.706473848777715e-06, "loss": 1.0479, "step": 8761 }, { "epoch": 0.9323260268142158, "grad_norm": 9.140625, "learning_rate": 9.706438317225697e-06, "loss": 1.1836, "step": 8762 }, { "epoch": 0.9324324324324325, "grad_norm": 13.96875, "learning_rate": 9.706402785673679e-06, "loss": 1.0388, "step": 8763 }, { "epoch": 0.9325388380506491, "grad_norm": 13.0625, "learning_rate": 9.706367254121662e-06, "loss": 1.3739, "step": 8764 }, { "epoch": 0.9326452436688657, "grad_norm": 13.640625, "learning_rate": 9.706331722569643e-06, "loss": 1.4171, "step": 8765 }, { "epoch": 0.9327516492870823, "grad_norm": 13.5625, "learning_rate": 9.706296191017625e-06, "loss": 1.5368, "step": 8766 }, { "epoch": 0.932858054905299, "grad_norm": 16.5, "learning_rate": 9.706260659465607e-06, "loss": 1.1727, "step": 8767 }, { "epoch": 0.9329644605235157, "grad_norm": 11.5625, "learning_rate": 9.706225127913588e-06, "loss": 1.0991, "step": 8768 }, { "epoch": 0.9330708661417323, "grad_norm": 16.96875, "learning_rate": 9.70618959636157e-06, "loss": 0.9982, "step": 8769 }, { "epoch": 0.9331772717599489, "grad_norm": 13.9140625, "learning_rate": 9.706154064809551e-06, "loss": 1.582, "step": 8770 }, { "epoch": 0.9332836773781655, "grad_norm": 10.1796875, "learning_rate": 9.706118533257533e-06, "loss": 0.9888, "step": 8771 }, { "epoch": 0.9333900829963823, "grad_norm": 12.1796875, "learning_rate": 9.706083001705515e-06, "loss": 0.9466, "step": 8772 }, { "epoch": 0.9334964886145989, "grad_norm": 15.515625, "learning_rate": 9.706047470153496e-06, "loss": 1.294, "step": 8773 }, { "epoch": 0.9336028942328155, "grad_norm": 11.28125, "learning_rate": 9.706011938601478e-06, "loss": 1.1723, "step": 8774 }, { "epoch": 0.9337092998510321, "grad_norm": 11.578125, "learning_rate": 9.705976407049461e-06, "loss": 1.0834, "step": 8775 }, { "epoch": 0.9338157054692487, "grad_norm": 12.2578125, "learning_rate": 9.705940875497443e-06, "loss": 1.2187, "step": 8776 }, { "epoch": 0.9339221110874654, "grad_norm": 10.953125, "learning_rate": 9.705905343945424e-06, "loss": 1.0765, "step": 8777 }, { "epoch": 0.9340285167056821, "grad_norm": 13.78125, "learning_rate": 9.705869812393406e-06, "loss": 1.3275, "step": 8778 }, { "epoch": 0.9341349223238987, "grad_norm": 15.296875, "learning_rate": 9.705834280841388e-06, "loss": 1.4122, "step": 8779 }, { "epoch": 0.9342413279421153, "grad_norm": 10.7109375, "learning_rate": 9.705798749289371e-06, "loss": 1.2162, "step": 8780 }, { "epoch": 0.934347733560332, "grad_norm": 13.4921875, "learning_rate": 9.705763217737353e-06, "loss": 1.3968, "step": 8781 }, { "epoch": 0.9344541391785486, "grad_norm": 11.4609375, "learning_rate": 9.705727686185334e-06, "loss": 1.0153, "step": 8782 }, { "epoch": 0.9345605447967653, "grad_norm": 14.234375, "learning_rate": 9.705692154633316e-06, "loss": 1.3955, "step": 8783 }, { "epoch": 0.9346669504149819, "grad_norm": 9.65625, "learning_rate": 9.705656623081297e-06, "loss": 1.2305, "step": 8784 }, { "epoch": 0.9347733560331986, "grad_norm": 10.328125, "learning_rate": 9.705621091529279e-06, "loss": 1.2173, "step": 8785 }, { "epoch": 0.9348797616514152, "grad_norm": 13.0859375, "learning_rate": 9.70558555997726e-06, "loss": 1.0267, "step": 8786 }, { "epoch": 0.9349861672696318, "grad_norm": 45.4375, "learning_rate": 9.705550028425242e-06, "loss": 1.4234, "step": 8787 }, { "epoch": 0.9350925728878485, "grad_norm": 11.671875, "learning_rate": 9.705514496873224e-06, "loss": 1.1679, "step": 8788 }, { "epoch": 0.9351989785060651, "grad_norm": 12.921875, "learning_rate": 9.705478965321205e-06, "loss": 1.1681, "step": 8789 }, { "epoch": 0.9353053841242818, "grad_norm": 13.5390625, "learning_rate": 9.705443433769189e-06, "loss": 1.3883, "step": 8790 }, { "epoch": 0.9354117897424984, "grad_norm": 11.4453125, "learning_rate": 9.70540790221717e-06, "loss": 1.1915, "step": 8791 }, { "epoch": 0.935518195360715, "grad_norm": 15.796875, "learning_rate": 9.705372370665152e-06, "loss": 1.2751, "step": 8792 }, { "epoch": 0.9356246009789317, "grad_norm": 13.1484375, "learning_rate": 9.705336839113134e-06, "loss": 1.1329, "step": 8793 }, { "epoch": 0.9357310065971484, "grad_norm": 10.953125, "learning_rate": 9.705301307561115e-06, "loss": 1.1566, "step": 8794 }, { "epoch": 0.935837412215365, "grad_norm": 11.390625, "learning_rate": 9.705265776009097e-06, "loss": 1.3149, "step": 8795 }, { "epoch": 0.9359438178335816, "grad_norm": 12.4921875, "learning_rate": 9.705230244457078e-06, "loss": 1.0832, "step": 8796 }, { "epoch": 0.9360502234517982, "grad_norm": 14.734375, "learning_rate": 9.70519471290506e-06, "loss": 1.1339, "step": 8797 }, { "epoch": 0.9361566290700148, "grad_norm": 21.6875, "learning_rate": 9.705159181353042e-06, "loss": 1.0611, "step": 8798 }, { "epoch": 0.9362630346882316, "grad_norm": 10.375, "learning_rate": 9.705123649801023e-06, "loss": 1.292, "step": 8799 }, { "epoch": 0.9363694403064482, "grad_norm": 12.5703125, "learning_rate": 9.705088118249006e-06, "loss": 1.2094, "step": 8800 }, { "epoch": 0.9364758459246648, "grad_norm": 12.265625, "learning_rate": 9.705052586696988e-06, "loss": 1.5799, "step": 8801 }, { "epoch": 0.9365822515428814, "grad_norm": 14.3984375, "learning_rate": 9.70501705514497e-06, "loss": 1.0711, "step": 8802 }, { "epoch": 0.9366886571610981, "grad_norm": 15.453125, "learning_rate": 9.704981523592951e-06, "loss": 1.1489, "step": 8803 }, { "epoch": 0.9367950627793148, "grad_norm": 35.0, "learning_rate": 9.704945992040933e-06, "loss": 1.3064, "step": 8804 }, { "epoch": 0.9369014683975314, "grad_norm": 14.359375, "learning_rate": 9.704910460488916e-06, "loss": 1.4528, "step": 8805 }, { "epoch": 0.937007874015748, "grad_norm": 10.6875, "learning_rate": 9.704874928936898e-06, "loss": 1.0638, "step": 8806 }, { "epoch": 0.9371142796339647, "grad_norm": 11.171875, "learning_rate": 9.70483939738488e-06, "loss": 1.1379, "step": 8807 }, { "epoch": 0.9372206852521813, "grad_norm": 12.5, "learning_rate": 9.704803865832861e-06, "loss": 1.1726, "step": 8808 }, { "epoch": 0.937327090870398, "grad_norm": 15.234375, "learning_rate": 9.704768334280843e-06, "loss": 0.9738, "step": 8809 }, { "epoch": 0.9374334964886146, "grad_norm": 12.765625, "learning_rate": 9.704732802728824e-06, "loss": 0.9406, "step": 8810 }, { "epoch": 0.9375399021068312, "grad_norm": 17.0625, "learning_rate": 9.704697271176806e-06, "loss": 1.2189, "step": 8811 }, { "epoch": 0.9376463077250479, "grad_norm": 11.15625, "learning_rate": 9.704661739624787e-06, "loss": 1.2708, "step": 8812 }, { "epoch": 0.9377527133432645, "grad_norm": 18.5625, "learning_rate": 9.704626208072769e-06, "loss": 1.1458, "step": 8813 }, { "epoch": 0.9378591189614812, "grad_norm": 12.09375, "learning_rate": 9.70459067652075e-06, "loss": 0.9469, "step": 8814 }, { "epoch": 0.9379655245796978, "grad_norm": 12.7890625, "learning_rate": 9.704555144968732e-06, "loss": 1.2934, "step": 8815 }, { "epoch": 0.9380719301979145, "grad_norm": 15.21875, "learning_rate": 9.704519613416716e-06, "loss": 1.0786, "step": 8816 }, { "epoch": 0.9381783358161311, "grad_norm": 11.71875, "learning_rate": 9.704484081864697e-06, "loss": 1.287, "step": 8817 }, { "epoch": 0.9382847414343477, "grad_norm": 11.8046875, "learning_rate": 9.704448550312679e-06, "loss": 1.2406, "step": 8818 }, { "epoch": 0.9383911470525643, "grad_norm": 11.703125, "learning_rate": 9.70441301876066e-06, "loss": 1.1643, "step": 8819 }, { "epoch": 0.938497552670781, "grad_norm": 25.65625, "learning_rate": 9.704377487208642e-06, "loss": 1.3309, "step": 8820 }, { "epoch": 0.9386039582889977, "grad_norm": 14.25, "learning_rate": 9.704341955656624e-06, "loss": 1.0626, "step": 8821 }, { "epoch": 0.9387103639072143, "grad_norm": 10.03125, "learning_rate": 9.704306424104605e-06, "loss": 1.3672, "step": 8822 }, { "epoch": 0.9388167695254309, "grad_norm": 12.4296875, "learning_rate": 9.704270892552587e-06, "loss": 0.8321, "step": 8823 }, { "epoch": 0.9389231751436475, "grad_norm": 15.03125, "learning_rate": 9.704235361000568e-06, "loss": 1.2747, "step": 8824 }, { "epoch": 0.9390295807618643, "grad_norm": 11.796875, "learning_rate": 9.70419982944855e-06, "loss": 1.1812, "step": 8825 }, { "epoch": 0.9391359863800809, "grad_norm": 15.5, "learning_rate": 9.704164297896533e-06, "loss": 1.6301, "step": 8826 }, { "epoch": 0.9392423919982975, "grad_norm": 11.75, "learning_rate": 9.704128766344515e-06, "loss": 0.9374, "step": 8827 }, { "epoch": 0.9393487976165141, "grad_norm": 12.375, "learning_rate": 9.704093234792497e-06, "loss": 1.1053, "step": 8828 }, { "epoch": 0.9394552032347308, "grad_norm": 10.6484375, "learning_rate": 9.704057703240478e-06, "loss": 1.0329, "step": 8829 }, { "epoch": 0.9395616088529475, "grad_norm": 8.5, "learning_rate": 9.70402217168846e-06, "loss": 1.1057, "step": 8830 }, { "epoch": 0.9396680144711641, "grad_norm": 15.484375, "learning_rate": 9.703986640136443e-06, "loss": 1.3903, "step": 8831 }, { "epoch": 0.9397744200893807, "grad_norm": 14.359375, "learning_rate": 9.703951108584425e-06, "loss": 1.0918, "step": 8832 }, { "epoch": 0.9398808257075973, "grad_norm": 12.0390625, "learning_rate": 9.703915577032406e-06, "loss": 0.9611, "step": 8833 }, { "epoch": 0.939987231325814, "grad_norm": 14.0625, "learning_rate": 9.703880045480388e-06, "loss": 1.2232, "step": 8834 }, { "epoch": 0.9400936369440307, "grad_norm": 16.375, "learning_rate": 9.70384451392837e-06, "loss": 1.3613, "step": 8835 }, { "epoch": 0.9402000425622473, "grad_norm": 15.8515625, "learning_rate": 9.703808982376351e-06, "loss": 1.0716, "step": 8836 }, { "epoch": 0.9403064481804639, "grad_norm": 14.5546875, "learning_rate": 9.703773450824333e-06, "loss": 1.4479, "step": 8837 }, { "epoch": 0.9404128537986806, "grad_norm": 13.5234375, "learning_rate": 9.703737919272314e-06, "loss": 0.978, "step": 8838 }, { "epoch": 0.9405192594168972, "grad_norm": 11.9453125, "learning_rate": 9.703702387720296e-06, "loss": 0.8262, "step": 8839 }, { "epoch": 0.9406256650351138, "grad_norm": 12.203125, "learning_rate": 9.703666856168277e-06, "loss": 0.9503, "step": 8840 }, { "epoch": 0.9407320706533305, "grad_norm": 12.8359375, "learning_rate": 9.70363132461626e-06, "loss": 1.2964, "step": 8841 }, { "epoch": 0.9408384762715472, "grad_norm": 11.375, "learning_rate": 9.703595793064242e-06, "loss": 1.3269, "step": 8842 }, { "epoch": 0.9409448818897638, "grad_norm": 13.265625, "learning_rate": 9.703560261512224e-06, "loss": 1.2059, "step": 8843 }, { "epoch": 0.9410512875079804, "grad_norm": 11.453125, "learning_rate": 9.703524729960206e-06, "loss": 1.1114, "step": 8844 }, { "epoch": 0.941157693126197, "grad_norm": 9.984375, "learning_rate": 9.703489198408187e-06, "loss": 1.2929, "step": 8845 }, { "epoch": 0.9412640987444137, "grad_norm": 17.40625, "learning_rate": 9.703453666856169e-06, "loss": 1.4217, "step": 8846 }, { "epoch": 0.9413705043626304, "grad_norm": 17.09375, "learning_rate": 9.70341813530415e-06, "loss": 1.1391, "step": 8847 }, { "epoch": 0.941476909980847, "grad_norm": 11.625, "learning_rate": 9.703382603752132e-06, "loss": 1.3399, "step": 8848 }, { "epoch": 0.9415833155990636, "grad_norm": 10.4765625, "learning_rate": 9.703347072200114e-06, "loss": 1.1755, "step": 8849 }, { "epoch": 0.9416897212172802, "grad_norm": 12.96875, "learning_rate": 9.703311540648095e-06, "loss": 1.4745, "step": 8850 }, { "epoch": 0.941796126835497, "grad_norm": 10.2421875, "learning_rate": 9.703276009096077e-06, "loss": 0.9412, "step": 8851 }, { "epoch": 0.9419025324537136, "grad_norm": 11.890625, "learning_rate": 9.70324047754406e-06, "loss": 1.1994, "step": 8852 }, { "epoch": 0.9420089380719302, "grad_norm": 13.046875, "learning_rate": 9.703204945992042e-06, "loss": 1.2257, "step": 8853 }, { "epoch": 0.9421153436901468, "grad_norm": 11.046875, "learning_rate": 9.703169414440023e-06, "loss": 1.0812, "step": 8854 }, { "epoch": 0.9422217493083634, "grad_norm": 11.3671875, "learning_rate": 9.703133882888005e-06, "loss": 0.962, "step": 8855 }, { "epoch": 0.9423281549265802, "grad_norm": 13.8203125, "learning_rate": 9.703098351335987e-06, "loss": 0.9131, "step": 8856 }, { "epoch": 0.9424345605447968, "grad_norm": 13.921875, "learning_rate": 9.70306281978397e-06, "loss": 0.955, "step": 8857 }, { "epoch": 0.9425409661630134, "grad_norm": 13.8046875, "learning_rate": 9.703027288231951e-06, "loss": 1.5354, "step": 8858 }, { "epoch": 0.94264737178123, "grad_norm": 11.8515625, "learning_rate": 9.702991756679933e-06, "loss": 1.2354, "step": 8859 }, { "epoch": 0.9427537773994467, "grad_norm": 11.2109375, "learning_rate": 9.702956225127915e-06, "loss": 1.416, "step": 8860 }, { "epoch": 0.9428601830176633, "grad_norm": 9.71875, "learning_rate": 9.702920693575896e-06, "loss": 1.3292, "step": 8861 }, { "epoch": 0.94296658863588, "grad_norm": 11.5078125, "learning_rate": 9.702885162023878e-06, "loss": 1.0824, "step": 8862 }, { "epoch": 0.9430729942540966, "grad_norm": 11.6796875, "learning_rate": 9.70284963047186e-06, "loss": 1.0912, "step": 8863 }, { "epoch": 0.9431793998723133, "grad_norm": 11.109375, "learning_rate": 9.702814098919841e-06, "loss": 1.2505, "step": 8864 }, { "epoch": 0.9432858054905299, "grad_norm": 9.3828125, "learning_rate": 9.702778567367823e-06, "loss": 1.1619, "step": 8865 }, { "epoch": 0.9433922111087465, "grad_norm": 15.3046875, "learning_rate": 9.702743035815804e-06, "loss": 1.4979, "step": 8866 }, { "epoch": 0.9434986167269632, "grad_norm": 11.5859375, "learning_rate": 9.702707504263788e-06, "loss": 1.1597, "step": 8867 }, { "epoch": 0.9436050223451798, "grad_norm": 12.3515625, "learning_rate": 9.70267197271177e-06, "loss": 0.9837, "step": 8868 }, { "epoch": 0.9437114279633965, "grad_norm": 11.25, "learning_rate": 9.702636441159751e-06, "loss": 1.3261, "step": 8869 }, { "epoch": 0.9438178335816131, "grad_norm": 13.328125, "learning_rate": 9.702600909607732e-06, "loss": 0.9705, "step": 8870 }, { "epoch": 0.9439242391998297, "grad_norm": 16.828125, "learning_rate": 9.702565378055714e-06, "loss": 1.4654, "step": 8871 }, { "epoch": 0.9440306448180464, "grad_norm": 13.4375, "learning_rate": 9.702529846503697e-06, "loss": 1.0023, "step": 8872 }, { "epoch": 0.9441370504362631, "grad_norm": 13.4140625, "learning_rate": 9.702494314951679e-06, "loss": 1.0286, "step": 8873 }, { "epoch": 0.9442434560544797, "grad_norm": 12.5625, "learning_rate": 9.70245878339966e-06, "loss": 1.158, "step": 8874 }, { "epoch": 0.9443498616726963, "grad_norm": 11.0234375, "learning_rate": 9.70242325184764e-06, "loss": 1.2858, "step": 8875 }, { "epoch": 0.9444562672909129, "grad_norm": 10.265625, "learning_rate": 9.702387720295622e-06, "loss": 1.1548, "step": 8876 }, { "epoch": 0.9445626729091297, "grad_norm": 13.8203125, "learning_rate": 9.702352188743605e-06, "loss": 1.022, "step": 8877 }, { "epoch": 0.9446690785273463, "grad_norm": 13.2890625, "learning_rate": 9.702316657191587e-06, "loss": 1.2925, "step": 8878 }, { "epoch": 0.9447754841455629, "grad_norm": 14.1953125, "learning_rate": 9.702281125639569e-06, "loss": 1.0975, "step": 8879 }, { "epoch": 0.9448818897637795, "grad_norm": 12.203125, "learning_rate": 9.70224559408755e-06, "loss": 1.1231, "step": 8880 }, { "epoch": 0.9449882953819961, "grad_norm": 10.7890625, "learning_rate": 9.702210062535532e-06, "loss": 1.0939, "step": 8881 }, { "epoch": 0.9450947010002128, "grad_norm": 10.6640625, "learning_rate": 9.702174530983515e-06, "loss": 1.2288, "step": 8882 }, { "epoch": 0.9452011066184295, "grad_norm": 11.296875, "learning_rate": 9.702138999431497e-06, "loss": 1.4584, "step": 8883 }, { "epoch": 0.9453075122366461, "grad_norm": 10.6171875, "learning_rate": 9.702103467879478e-06, "loss": 1.4321, "step": 8884 }, { "epoch": 0.9454139178548627, "grad_norm": 14.1875, "learning_rate": 9.70206793632746e-06, "loss": 1.212, "step": 8885 }, { "epoch": 0.9455203234730793, "grad_norm": 12.5859375, "learning_rate": 9.702032404775442e-06, "loss": 1.5623, "step": 8886 }, { "epoch": 0.945626729091296, "grad_norm": 19.21875, "learning_rate": 9.701996873223423e-06, "loss": 1.2181, "step": 8887 }, { "epoch": 0.9457331347095127, "grad_norm": 11.9609375, "learning_rate": 9.701961341671405e-06, "loss": 1.1778, "step": 8888 }, { "epoch": 0.9458395403277293, "grad_norm": 10.328125, "learning_rate": 9.701925810119386e-06, "loss": 0.9458, "step": 8889 }, { "epoch": 0.9459459459459459, "grad_norm": 11.140625, "learning_rate": 9.701890278567368e-06, "loss": 1.0415, "step": 8890 }, { "epoch": 0.9460523515641626, "grad_norm": 10.1796875, "learning_rate": 9.70185474701535e-06, "loss": 1.361, "step": 8891 }, { "epoch": 0.9461587571823792, "grad_norm": 8.6171875, "learning_rate": 9.701819215463331e-06, "loss": 1.0969, "step": 8892 }, { "epoch": 0.9462651628005959, "grad_norm": 14.3828125, "learning_rate": 9.701783683911314e-06, "loss": 1.2061, "step": 8893 }, { "epoch": 0.9463715684188125, "grad_norm": 13.5078125, "learning_rate": 9.701748152359296e-06, "loss": 1.1988, "step": 8894 }, { "epoch": 0.9464779740370292, "grad_norm": 12.9296875, "learning_rate": 9.701712620807278e-06, "loss": 0.9575, "step": 8895 }, { "epoch": 0.9465843796552458, "grad_norm": 9.65625, "learning_rate": 9.70167708925526e-06, "loss": 1.1994, "step": 8896 }, { "epoch": 0.9466907852734624, "grad_norm": 13.078125, "learning_rate": 9.701641557703241e-06, "loss": 1.0737, "step": 8897 }, { "epoch": 0.9467971908916791, "grad_norm": 12.3515625, "learning_rate": 9.701606026151224e-06, "loss": 1.1522, "step": 8898 }, { "epoch": 0.9469035965098958, "grad_norm": 11.6796875, "learning_rate": 9.701570494599206e-06, "loss": 1.0037, "step": 8899 }, { "epoch": 0.9470100021281124, "grad_norm": 20.875, "learning_rate": 9.701534963047187e-06, "loss": 1.5367, "step": 8900 }, { "epoch": 0.947116407746329, "grad_norm": 11.6171875, "learning_rate": 9.701499431495169e-06, "loss": 1.2045, "step": 8901 }, { "epoch": 0.9472228133645456, "grad_norm": 14.09375, "learning_rate": 9.70146389994315e-06, "loss": 1.405, "step": 8902 }, { "epoch": 0.9473292189827623, "grad_norm": 13.8046875, "learning_rate": 9.701428368391132e-06, "loss": 1.2398, "step": 8903 }, { "epoch": 0.947435624600979, "grad_norm": 11.625, "learning_rate": 9.701392836839114e-06, "loss": 1.2066, "step": 8904 }, { "epoch": 0.9475420302191956, "grad_norm": 13.171875, "learning_rate": 9.701357305287095e-06, "loss": 1.1094, "step": 8905 }, { "epoch": 0.9476484358374122, "grad_norm": 10.640625, "learning_rate": 9.701321773735077e-06, "loss": 1.039, "step": 8906 }, { "epoch": 0.9477548414556288, "grad_norm": 13.3984375, "learning_rate": 9.701286242183059e-06, "loss": 1.3762, "step": 8907 }, { "epoch": 0.9478612470738454, "grad_norm": 10.203125, "learning_rate": 9.701250710631042e-06, "loss": 0.9604, "step": 8908 }, { "epoch": 0.9479676526920622, "grad_norm": 10.1875, "learning_rate": 9.701215179079024e-06, "loss": 1.2051, "step": 8909 }, { "epoch": 0.9480740583102788, "grad_norm": 11.453125, "learning_rate": 9.701179647527005e-06, "loss": 1.1343, "step": 8910 }, { "epoch": 0.9481804639284954, "grad_norm": 11.7578125, "learning_rate": 9.701144115974987e-06, "loss": 1.3585, "step": 8911 }, { "epoch": 0.948286869546712, "grad_norm": 13.671875, "learning_rate": 9.701108584422968e-06, "loss": 1.5706, "step": 8912 }, { "epoch": 0.9483932751649287, "grad_norm": 10.65625, "learning_rate": 9.70107305287095e-06, "loss": 1.3432, "step": 8913 }, { "epoch": 0.9484996807831454, "grad_norm": 13.765625, "learning_rate": 9.701037521318932e-06, "loss": 1.2679, "step": 8914 }, { "epoch": 0.948606086401362, "grad_norm": 12.9921875, "learning_rate": 9.701001989766913e-06, "loss": 1.1935, "step": 8915 }, { "epoch": 0.9487124920195786, "grad_norm": 9.0703125, "learning_rate": 9.700966458214895e-06, "loss": 1.0888, "step": 8916 }, { "epoch": 0.9488188976377953, "grad_norm": 10.953125, "learning_rate": 9.700930926662876e-06, "loss": 1.0643, "step": 8917 }, { "epoch": 0.9489253032560119, "grad_norm": 10.328125, "learning_rate": 9.70089539511086e-06, "loss": 1.0649, "step": 8918 }, { "epoch": 0.9490317088742286, "grad_norm": 22.328125, "learning_rate": 9.700859863558841e-06, "loss": 1.356, "step": 8919 }, { "epoch": 0.9491381144924452, "grad_norm": 11.03125, "learning_rate": 9.700824332006823e-06, "loss": 1.0232, "step": 8920 }, { "epoch": 0.9492445201106618, "grad_norm": 13.6796875, "learning_rate": 9.700788800454805e-06, "loss": 1.2623, "step": 8921 }, { "epoch": 0.9493509257288785, "grad_norm": 13.453125, "learning_rate": 9.700753268902786e-06, "loss": 1.2897, "step": 8922 }, { "epoch": 0.9494573313470951, "grad_norm": 17.390625, "learning_rate": 9.70071773735077e-06, "loss": 0.9868, "step": 8923 }, { "epoch": 0.9495637369653118, "grad_norm": 11.6328125, "learning_rate": 9.700682205798751e-06, "loss": 0.8863, "step": 8924 }, { "epoch": 0.9496701425835284, "grad_norm": 13.4609375, "learning_rate": 9.700646674246733e-06, "loss": 1.2191, "step": 8925 }, { "epoch": 0.9497765482017451, "grad_norm": 12.4453125, "learning_rate": 9.700611142694714e-06, "loss": 0.9156, "step": 8926 }, { "epoch": 0.9498829538199617, "grad_norm": 15.6171875, "learning_rate": 9.700575611142696e-06, "loss": 0.9131, "step": 8927 }, { "epoch": 0.9499893594381783, "grad_norm": 14.078125, "learning_rate": 9.700540079590678e-06, "loss": 1.1947, "step": 8928 }, { "epoch": 0.9500957650563949, "grad_norm": 11.6015625, "learning_rate": 9.700504548038659e-06, "loss": 1.2226, "step": 8929 }, { "epoch": 0.9502021706746117, "grad_norm": 10.1875, "learning_rate": 9.70046901648664e-06, "loss": 0.892, "step": 8930 }, { "epoch": 0.9503085762928283, "grad_norm": 14.484375, "learning_rate": 9.700433484934622e-06, "loss": 1.294, "step": 8931 }, { "epoch": 0.9504149819110449, "grad_norm": 14.3046875, "learning_rate": 9.700397953382604e-06, "loss": 1.2667, "step": 8932 }, { "epoch": 0.9505213875292615, "grad_norm": 9.953125, "learning_rate": 9.700362421830586e-06, "loss": 1.1781, "step": 8933 }, { "epoch": 0.9506277931474781, "grad_norm": 11.296875, "learning_rate": 9.700326890278569e-06, "loss": 1.1087, "step": 8934 }, { "epoch": 0.9507341987656949, "grad_norm": 13.09375, "learning_rate": 9.70029135872655e-06, "loss": 1.025, "step": 8935 }, { "epoch": 0.9508406043839115, "grad_norm": 14.7265625, "learning_rate": 9.700255827174532e-06, "loss": 1.2089, "step": 8936 }, { "epoch": 0.9509470100021281, "grad_norm": 14.234375, "learning_rate": 9.700220295622514e-06, "loss": 1.3086, "step": 8937 }, { "epoch": 0.9510534156203447, "grad_norm": 9.1796875, "learning_rate": 9.700184764070495e-06, "loss": 1.1945, "step": 8938 }, { "epoch": 0.9511598212385614, "grad_norm": 11.0703125, "learning_rate": 9.700149232518477e-06, "loss": 1.103, "step": 8939 }, { "epoch": 0.9512662268567781, "grad_norm": 14.8515625, "learning_rate": 9.700113700966458e-06, "loss": 1.4463, "step": 8940 }, { "epoch": 0.9513726324749947, "grad_norm": 9.6796875, "learning_rate": 9.70007816941444e-06, "loss": 1.4248, "step": 8941 }, { "epoch": 0.9514790380932113, "grad_norm": 9.78125, "learning_rate": 9.700042637862422e-06, "loss": 1.0203, "step": 8942 }, { "epoch": 0.951585443711428, "grad_norm": 19.578125, "learning_rate": 9.700007106310403e-06, "loss": 1.1452, "step": 8943 }, { "epoch": 0.9516918493296446, "grad_norm": 18.390625, "learning_rate": 9.699971574758387e-06, "loss": 1.5144, "step": 8944 }, { "epoch": 0.9517982549478613, "grad_norm": 13.2578125, "learning_rate": 9.699936043206368e-06, "loss": 1.2352, "step": 8945 }, { "epoch": 0.9519046605660779, "grad_norm": 9.9453125, "learning_rate": 9.69990051165435e-06, "loss": 1.3176, "step": 8946 }, { "epoch": 0.9520110661842945, "grad_norm": 14.5703125, "learning_rate": 9.699864980102331e-06, "loss": 1.1732, "step": 8947 }, { "epoch": 0.9521174718025112, "grad_norm": 11.078125, "learning_rate": 9.699829448550313e-06, "loss": 0.9646, "step": 8948 }, { "epoch": 0.9522238774207278, "grad_norm": 13.75, "learning_rate": 9.699793916998296e-06, "loss": 1.3723, "step": 8949 }, { "epoch": 0.9523302830389444, "grad_norm": 18.703125, "learning_rate": 9.699758385446278e-06, "loss": 1.3603, "step": 8950 }, { "epoch": 0.9524366886571611, "grad_norm": 13.2265625, "learning_rate": 9.69972285389426e-06, "loss": 1.075, "step": 8951 }, { "epoch": 0.9525430942753778, "grad_norm": 25.0625, "learning_rate": 9.699687322342241e-06, "loss": 1.2452, "step": 8952 }, { "epoch": 0.9526494998935944, "grad_norm": 12.2265625, "learning_rate": 9.699651790790223e-06, "loss": 1.1232, "step": 8953 }, { "epoch": 0.952755905511811, "grad_norm": 15.96875, "learning_rate": 9.699616259238204e-06, "loss": 1.37, "step": 8954 }, { "epoch": 0.9528623111300276, "grad_norm": 18.578125, "learning_rate": 9.699580727686186e-06, "loss": 1.1997, "step": 8955 }, { "epoch": 0.9529687167482443, "grad_norm": 15.296875, "learning_rate": 9.699545196134168e-06, "loss": 1.3891, "step": 8956 }, { "epoch": 0.953075122366461, "grad_norm": 17.625, "learning_rate": 9.69950966458215e-06, "loss": 1.2867, "step": 8957 }, { "epoch": 0.9531815279846776, "grad_norm": 11.265625, "learning_rate": 9.69947413303013e-06, "loss": 1.2519, "step": 8958 }, { "epoch": 0.9532879336028942, "grad_norm": 12.4921875, "learning_rate": 9.699438601478114e-06, "loss": 1.0768, "step": 8959 }, { "epoch": 0.9533943392211108, "grad_norm": 15.2265625, "learning_rate": 9.699403069926096e-06, "loss": 1.1037, "step": 8960 }, { "epoch": 0.9535007448393276, "grad_norm": 11.9609375, "learning_rate": 9.699367538374077e-06, "loss": 1.1085, "step": 8961 }, { "epoch": 0.9536071504575442, "grad_norm": 9.1484375, "learning_rate": 9.699332006822059e-06, "loss": 1.1821, "step": 8962 }, { "epoch": 0.9537135560757608, "grad_norm": 14.0234375, "learning_rate": 9.69929647527004e-06, "loss": 1.3555, "step": 8963 }, { "epoch": 0.9538199616939774, "grad_norm": 13.28125, "learning_rate": 9.699260943718022e-06, "loss": 1.5005, "step": 8964 }, { "epoch": 0.953926367312194, "grad_norm": 12.34375, "learning_rate": 9.699225412166004e-06, "loss": 1.1249, "step": 8965 }, { "epoch": 0.9540327729304108, "grad_norm": 23.1875, "learning_rate": 9.699189880613985e-06, "loss": 1.2316, "step": 8966 }, { "epoch": 0.9541391785486274, "grad_norm": 19.046875, "learning_rate": 9.699154349061967e-06, "loss": 1.0914, "step": 8967 }, { "epoch": 0.954245584166844, "grad_norm": 14.0546875, "learning_rate": 9.699118817509949e-06, "loss": 0.9393, "step": 8968 }, { "epoch": 0.9543519897850606, "grad_norm": 17.5625, "learning_rate": 9.69908328595793e-06, "loss": 0.8321, "step": 8969 }, { "epoch": 0.9544583954032773, "grad_norm": 10.3984375, "learning_rate": 9.699047754405913e-06, "loss": 1.1164, "step": 8970 }, { "epoch": 0.9545648010214939, "grad_norm": 12.6796875, "learning_rate": 9.699012222853895e-06, "loss": 1.193, "step": 8971 }, { "epoch": 0.9546712066397106, "grad_norm": 15.5703125, "learning_rate": 9.698976691301877e-06, "loss": 1.0434, "step": 8972 }, { "epoch": 0.9547776122579272, "grad_norm": 14.3984375, "learning_rate": 9.698941159749858e-06, "loss": 1.1926, "step": 8973 }, { "epoch": 0.9548840178761439, "grad_norm": 13.0546875, "learning_rate": 9.69890562819784e-06, "loss": 1.4316, "step": 8974 }, { "epoch": 0.9549904234943605, "grad_norm": 14.5390625, "learning_rate": 9.698870096645823e-06, "loss": 1.2165, "step": 8975 }, { "epoch": 0.9550968291125771, "grad_norm": 11.15625, "learning_rate": 9.698834565093805e-06, "loss": 1.0083, "step": 8976 }, { "epoch": 0.9552032347307938, "grad_norm": 12.15625, "learning_rate": 9.698799033541786e-06, "loss": 1.0693, "step": 8977 }, { "epoch": 0.9553096403490104, "grad_norm": 10.1640625, "learning_rate": 9.698763501989768e-06, "loss": 1.1421, "step": 8978 }, { "epoch": 0.9554160459672271, "grad_norm": 16.984375, "learning_rate": 9.69872797043775e-06, "loss": 1.3574, "step": 8979 }, { "epoch": 0.9555224515854437, "grad_norm": 9.453125, "learning_rate": 9.698692438885731e-06, "loss": 1.5939, "step": 8980 }, { "epoch": 0.9556288572036603, "grad_norm": 11.078125, "learning_rate": 9.698656907333713e-06, "loss": 1.1872, "step": 8981 }, { "epoch": 0.955735262821877, "grad_norm": 14.9375, "learning_rate": 9.698621375781694e-06, "loss": 1.2212, "step": 8982 }, { "epoch": 0.9558416684400937, "grad_norm": 11.5859375, "learning_rate": 9.698585844229676e-06, "loss": 0.9265, "step": 8983 }, { "epoch": 0.9559480740583103, "grad_norm": 8.8671875, "learning_rate": 9.698550312677658e-06, "loss": 1.1751, "step": 8984 }, { "epoch": 0.9560544796765269, "grad_norm": 12.1875, "learning_rate": 9.698514781125641e-06, "loss": 1.2151, "step": 8985 }, { "epoch": 0.9561608852947435, "grad_norm": 10.9921875, "learning_rate": 9.698479249573623e-06, "loss": 1.0587, "step": 8986 }, { "epoch": 0.9562672909129603, "grad_norm": 10.84375, "learning_rate": 9.698443718021604e-06, "loss": 1.3058, "step": 8987 }, { "epoch": 0.9563736965311769, "grad_norm": 9.6171875, "learning_rate": 9.698408186469586e-06, "loss": 1.122, "step": 8988 }, { "epoch": 0.9564801021493935, "grad_norm": 13.2421875, "learning_rate": 9.698372654917567e-06, "loss": 1.027, "step": 8989 }, { "epoch": 0.9565865077676101, "grad_norm": 17.71875, "learning_rate": 9.69833712336555e-06, "loss": 1.6981, "step": 8990 }, { "epoch": 0.9566929133858267, "grad_norm": 12.9453125, "learning_rate": 9.698301591813532e-06, "loss": 1.1501, "step": 8991 }, { "epoch": 0.9567993190040434, "grad_norm": 11.7109375, "learning_rate": 9.698266060261514e-06, "loss": 0.8726, "step": 8992 }, { "epoch": 0.9569057246222601, "grad_norm": 12.375, "learning_rate": 9.698230528709495e-06, "loss": 1.0286, "step": 8993 }, { "epoch": 0.9570121302404767, "grad_norm": 11.0859375, "learning_rate": 9.698194997157475e-06, "loss": 1.1724, "step": 8994 }, { "epoch": 0.9571185358586933, "grad_norm": 11.3125, "learning_rate": 9.698159465605459e-06, "loss": 1.2989, "step": 8995 }, { "epoch": 0.95722494147691, "grad_norm": 7.8515625, "learning_rate": 9.69812393405344e-06, "loss": 1.0737, "step": 8996 }, { "epoch": 0.9573313470951266, "grad_norm": 11.4765625, "learning_rate": 9.698088402501422e-06, "loss": 1.4245, "step": 8997 }, { "epoch": 0.9574377527133433, "grad_norm": 12.3828125, "learning_rate": 9.698052870949404e-06, "loss": 1.2028, "step": 8998 }, { "epoch": 0.9575441583315599, "grad_norm": 16.234375, "learning_rate": 9.698017339397385e-06, "loss": 1.2483, "step": 8999 }, { "epoch": 0.9576505639497765, "grad_norm": 11.84375, "learning_rate": 9.697981807845368e-06, "loss": 1.2009, "step": 9000 }, { "epoch": 0.9577569695679932, "grad_norm": 14.4609375, "learning_rate": 9.69794627629335e-06, "loss": 1.1723, "step": 9001 }, { "epoch": 0.9578633751862098, "grad_norm": 10.453125, "learning_rate": 9.697910744741332e-06, "loss": 1.109, "step": 9002 }, { "epoch": 0.9579697808044265, "grad_norm": 12.1328125, "learning_rate": 9.697875213189313e-06, "loss": 1.1822, "step": 9003 }, { "epoch": 0.9580761864226431, "grad_norm": 10.4296875, "learning_rate": 9.697839681637295e-06, "loss": 0.9765, "step": 9004 }, { "epoch": 0.9581825920408598, "grad_norm": 12.234375, "learning_rate": 9.697804150085276e-06, "loss": 1.1042, "step": 9005 }, { "epoch": 0.9582889976590764, "grad_norm": 14.4140625, "learning_rate": 9.697768618533258e-06, "loss": 1.1494, "step": 9006 }, { "epoch": 0.958395403277293, "grad_norm": 11.9453125, "learning_rate": 9.69773308698124e-06, "loss": 1.4205, "step": 9007 }, { "epoch": 0.9585018088955097, "grad_norm": 14.4765625, "learning_rate": 9.697697555429221e-06, "loss": 1.1109, "step": 9008 }, { "epoch": 0.9586082145137264, "grad_norm": 12.8515625, "learning_rate": 9.697662023877203e-06, "loss": 1.355, "step": 9009 }, { "epoch": 0.958714620131943, "grad_norm": 19.53125, "learning_rate": 9.697626492325185e-06, "loss": 1.2341, "step": 9010 }, { "epoch": 0.9588210257501596, "grad_norm": 11.5546875, "learning_rate": 9.697590960773168e-06, "loss": 1.1267, "step": 9011 }, { "epoch": 0.9589274313683762, "grad_norm": 14.3671875, "learning_rate": 9.69755542922115e-06, "loss": 1.3083, "step": 9012 }, { "epoch": 0.959033836986593, "grad_norm": 10.0546875, "learning_rate": 9.697519897669131e-06, "loss": 1.3355, "step": 9013 }, { "epoch": 0.9591402426048096, "grad_norm": 12.25, "learning_rate": 9.697484366117113e-06, "loss": 1.3189, "step": 9014 }, { "epoch": 0.9592466482230262, "grad_norm": 14.828125, "learning_rate": 9.697448834565094e-06, "loss": 1.1842, "step": 9015 }, { "epoch": 0.9593530538412428, "grad_norm": 12.2578125, "learning_rate": 9.697413303013078e-06, "loss": 1.0457, "step": 9016 }, { "epoch": 0.9594594594594594, "grad_norm": 10.890625, "learning_rate": 9.697377771461059e-06, "loss": 0.9304, "step": 9017 }, { "epoch": 0.959565865077676, "grad_norm": 11.78125, "learning_rate": 9.69734223990904e-06, "loss": 1.326, "step": 9018 }, { "epoch": 0.9596722706958928, "grad_norm": 59.125, "learning_rate": 9.697306708357022e-06, "loss": 0.9574, "step": 9019 }, { "epoch": 0.9597786763141094, "grad_norm": 10.7734375, "learning_rate": 9.697271176805004e-06, "loss": 1.4593, "step": 9020 }, { "epoch": 0.959885081932326, "grad_norm": 9.21875, "learning_rate": 9.697235645252986e-06, "loss": 1.3546, "step": 9021 }, { "epoch": 0.9599914875505426, "grad_norm": 12.0859375, "learning_rate": 9.697200113700967e-06, "loss": 1.3171, "step": 9022 }, { "epoch": 0.9600978931687593, "grad_norm": 14.1484375, "learning_rate": 9.697164582148949e-06, "loss": 1.0509, "step": 9023 }, { "epoch": 0.960204298786976, "grad_norm": 18.171875, "learning_rate": 9.69712905059693e-06, "loss": 1.6742, "step": 9024 }, { "epoch": 0.9603107044051926, "grad_norm": 14.34375, "learning_rate": 9.697093519044912e-06, "loss": 1.1227, "step": 9025 }, { "epoch": 0.9604171100234092, "grad_norm": 12.7421875, "learning_rate": 9.697057987492895e-06, "loss": 1.7469, "step": 9026 }, { "epoch": 0.9605235156416259, "grad_norm": 17.328125, "learning_rate": 9.697022455940877e-06, "loss": 1.2401, "step": 9027 }, { "epoch": 0.9606299212598425, "grad_norm": 11.7734375, "learning_rate": 9.696986924388859e-06, "loss": 1.0836, "step": 9028 }, { "epoch": 0.9607363268780592, "grad_norm": 11.0234375, "learning_rate": 9.69695139283684e-06, "loss": 1.3834, "step": 9029 }, { "epoch": 0.9608427324962758, "grad_norm": 11.359375, "learning_rate": 9.696915861284822e-06, "loss": 1.375, "step": 9030 }, { "epoch": 0.9609491381144925, "grad_norm": 10.6640625, "learning_rate": 9.696880329732803e-06, "loss": 1.3006, "step": 9031 }, { "epoch": 0.9610555437327091, "grad_norm": 19.421875, "learning_rate": 9.696844798180785e-06, "loss": 1.003, "step": 9032 }, { "epoch": 0.9611619493509257, "grad_norm": 12.171875, "learning_rate": 9.696809266628767e-06, "loss": 1.0585, "step": 9033 }, { "epoch": 0.9612683549691424, "grad_norm": 12.1328125, "learning_rate": 9.696773735076748e-06, "loss": 1.098, "step": 9034 }, { "epoch": 0.961374760587359, "grad_norm": 13.3828125, "learning_rate": 9.69673820352473e-06, "loss": 1.0543, "step": 9035 }, { "epoch": 0.9614811662055757, "grad_norm": 15.7265625, "learning_rate": 9.696702671972713e-06, "loss": 1.2131, "step": 9036 }, { "epoch": 0.9615875718237923, "grad_norm": 12.1015625, "learning_rate": 9.696667140420695e-06, "loss": 1.0216, "step": 9037 }, { "epoch": 0.9616939774420089, "grad_norm": 12.2421875, "learning_rate": 9.696631608868676e-06, "loss": 1.1551, "step": 9038 }, { "epoch": 0.9618003830602255, "grad_norm": 11.6015625, "learning_rate": 9.696596077316658e-06, "loss": 1.376, "step": 9039 }, { "epoch": 0.9619067886784423, "grad_norm": 16.75, "learning_rate": 9.69656054576464e-06, "loss": 1.3849, "step": 9040 }, { "epoch": 0.9620131942966589, "grad_norm": 18.6875, "learning_rate": 9.696525014212621e-06, "loss": 1.1401, "step": 9041 }, { "epoch": 0.9621195999148755, "grad_norm": 14.1328125, "learning_rate": 9.696489482660604e-06, "loss": 1.0632, "step": 9042 }, { "epoch": 0.9622260055330921, "grad_norm": 13.875, "learning_rate": 9.696453951108586e-06, "loss": 1.3673, "step": 9043 }, { "epoch": 0.9623324111513087, "grad_norm": 15.546875, "learning_rate": 9.696418419556568e-06, "loss": 1.514, "step": 9044 }, { "epoch": 0.9624388167695255, "grad_norm": 18.390625, "learning_rate": 9.69638288800455e-06, "loss": 1.2013, "step": 9045 }, { "epoch": 0.9625452223877421, "grad_norm": 11.875, "learning_rate": 9.69634735645253e-06, "loss": 0.9425, "step": 9046 }, { "epoch": 0.9626516280059587, "grad_norm": 13.1328125, "learning_rate": 9.696311824900512e-06, "loss": 1.097, "step": 9047 }, { "epoch": 0.9627580336241753, "grad_norm": 15.2265625, "learning_rate": 9.696276293348494e-06, "loss": 1.0115, "step": 9048 }, { "epoch": 0.962864439242392, "grad_norm": 12.8515625, "learning_rate": 9.696240761796476e-06, "loss": 1.0753, "step": 9049 }, { "epoch": 0.9629708448606087, "grad_norm": 16.609375, "learning_rate": 9.696205230244457e-06, "loss": 1.1781, "step": 9050 }, { "epoch": 0.9630772504788253, "grad_norm": 13.515625, "learning_rate": 9.696169698692439e-06, "loss": 1.3918, "step": 9051 }, { "epoch": 0.9631836560970419, "grad_norm": 12.2421875, "learning_rate": 9.696134167140422e-06, "loss": 1.0949, "step": 9052 }, { "epoch": 0.9632900617152585, "grad_norm": 17.46875, "learning_rate": 9.696098635588404e-06, "loss": 1.23, "step": 9053 }, { "epoch": 0.9633964673334752, "grad_norm": 10.8203125, "learning_rate": 9.696063104036385e-06, "loss": 0.9547, "step": 9054 }, { "epoch": 0.9635028729516919, "grad_norm": 14.265625, "learning_rate": 9.696027572484367e-06, "loss": 1.0834, "step": 9055 }, { "epoch": 0.9636092785699085, "grad_norm": 11.125, "learning_rate": 9.695992040932349e-06, "loss": 0.9613, "step": 9056 }, { "epoch": 0.9637156841881251, "grad_norm": 15.90625, "learning_rate": 9.69595650938033e-06, "loss": 1.309, "step": 9057 }, { "epoch": 0.9638220898063418, "grad_norm": 12.9609375, "learning_rate": 9.695920977828312e-06, "loss": 1.4334, "step": 9058 }, { "epoch": 0.9639284954245584, "grad_norm": 13.984375, "learning_rate": 9.695885446276293e-06, "loss": 1.2186, "step": 9059 }, { "epoch": 0.964034901042775, "grad_norm": 11.0078125, "learning_rate": 9.695849914724275e-06, "loss": 1.1269, "step": 9060 }, { "epoch": 0.9641413066609917, "grad_norm": 10.84375, "learning_rate": 9.695814383172257e-06, "loss": 0.8163, "step": 9061 }, { "epoch": 0.9642477122792084, "grad_norm": 9.328125, "learning_rate": 9.69577885162024e-06, "loss": 0.9968, "step": 9062 }, { "epoch": 0.964354117897425, "grad_norm": 17.390625, "learning_rate": 9.695743320068222e-06, "loss": 1.1693, "step": 9063 }, { "epoch": 0.9644605235156416, "grad_norm": 18.203125, "learning_rate": 9.695707788516203e-06, "loss": 1.0071, "step": 9064 }, { "epoch": 0.9645669291338582, "grad_norm": 10.9140625, "learning_rate": 9.695672256964185e-06, "loss": 1.2411, "step": 9065 }, { "epoch": 0.964673334752075, "grad_norm": 13.0078125, "learning_rate": 9.695636725412166e-06, "loss": 1.6361, "step": 9066 }, { "epoch": 0.9647797403702916, "grad_norm": 14.140625, "learning_rate": 9.69560119386015e-06, "loss": 1.2619, "step": 9067 }, { "epoch": 0.9648861459885082, "grad_norm": 11.5625, "learning_rate": 9.695565662308131e-06, "loss": 1.2834, "step": 9068 }, { "epoch": 0.9649925516067248, "grad_norm": 14.5703125, "learning_rate": 9.695530130756113e-06, "loss": 1.1601, "step": 9069 }, { "epoch": 0.9650989572249414, "grad_norm": 15.5, "learning_rate": 9.695494599204094e-06, "loss": 0.8946, "step": 9070 }, { "epoch": 0.9652053628431582, "grad_norm": 14.5546875, "learning_rate": 9.695459067652076e-06, "loss": 1.3921, "step": 9071 }, { "epoch": 0.9653117684613748, "grad_norm": 11.828125, "learning_rate": 9.695423536100058e-06, "loss": 1.3335, "step": 9072 }, { "epoch": 0.9654181740795914, "grad_norm": 14.546875, "learning_rate": 9.69538800454804e-06, "loss": 1.1628, "step": 9073 }, { "epoch": 0.965524579697808, "grad_norm": 9.796875, "learning_rate": 9.695352472996021e-06, "loss": 1.2242, "step": 9074 }, { "epoch": 0.9656309853160246, "grad_norm": 13.015625, "learning_rate": 9.695316941444002e-06, "loss": 1.0842, "step": 9075 }, { "epoch": 0.9657373909342414, "grad_norm": 15.0703125, "learning_rate": 9.695281409891984e-06, "loss": 1.0232, "step": 9076 }, { "epoch": 0.965843796552458, "grad_norm": 11.625, "learning_rate": 9.695245878339967e-06, "loss": 1.231, "step": 9077 }, { "epoch": 0.9659502021706746, "grad_norm": 13.1015625, "learning_rate": 9.695210346787949e-06, "loss": 1.276, "step": 9078 }, { "epoch": 0.9660566077888912, "grad_norm": 13.0078125, "learning_rate": 9.69517481523593e-06, "loss": 1.2744, "step": 9079 }, { "epoch": 0.9661630134071079, "grad_norm": 13.21875, "learning_rate": 9.695139283683912e-06, "loss": 1.2443, "step": 9080 }, { "epoch": 0.9662694190253245, "grad_norm": 15.5703125, "learning_rate": 9.695103752131894e-06, "loss": 1.3218, "step": 9081 }, { "epoch": 0.9663758246435412, "grad_norm": 12.9765625, "learning_rate": 9.695068220579875e-06, "loss": 1.12, "step": 9082 }, { "epoch": 0.9664822302617578, "grad_norm": 11.265625, "learning_rate": 9.695032689027857e-06, "loss": 1.2333, "step": 9083 }, { "epoch": 0.9665886358799745, "grad_norm": 14.0859375, "learning_rate": 9.694997157475839e-06, "loss": 1.1225, "step": 9084 }, { "epoch": 0.9666950414981911, "grad_norm": 13.1015625, "learning_rate": 9.69496162592382e-06, "loss": 1.9451, "step": 9085 }, { "epoch": 0.9668014471164077, "grad_norm": 12.046875, "learning_rate": 9.694926094371802e-06, "loss": 0.8233, "step": 9086 }, { "epoch": 0.9669078527346244, "grad_norm": 11.59375, "learning_rate": 9.694890562819783e-06, "loss": 1.0163, "step": 9087 }, { "epoch": 0.967014258352841, "grad_norm": 9.96875, "learning_rate": 9.694855031267767e-06, "loss": 0.8358, "step": 9088 }, { "epoch": 0.9671206639710577, "grad_norm": 9.09375, "learning_rate": 9.694819499715748e-06, "loss": 1.134, "step": 9089 }, { "epoch": 0.9672270695892743, "grad_norm": 9.859375, "learning_rate": 9.69478396816373e-06, "loss": 1.1585, "step": 9090 }, { "epoch": 0.9673334752074909, "grad_norm": 11.96875, "learning_rate": 9.694748436611712e-06, "loss": 1.2921, "step": 9091 }, { "epoch": 0.9674398808257076, "grad_norm": 10.3203125, "learning_rate": 9.694712905059693e-06, "loss": 1.2693, "step": 9092 }, { "epoch": 0.9675462864439243, "grad_norm": 13.609375, "learning_rate": 9.694677373507676e-06, "loss": 1.0266, "step": 9093 }, { "epoch": 0.9676526920621409, "grad_norm": 9.7734375, "learning_rate": 9.694641841955658e-06, "loss": 1.2913, "step": 9094 }, { "epoch": 0.9677590976803575, "grad_norm": 11.7421875, "learning_rate": 9.69460631040364e-06, "loss": 1.1203, "step": 9095 }, { "epoch": 0.9678655032985741, "grad_norm": 14.296875, "learning_rate": 9.694570778851621e-06, "loss": 1.3371, "step": 9096 }, { "epoch": 0.9679719089167909, "grad_norm": 30.421875, "learning_rate": 9.694535247299603e-06, "loss": 1.322, "step": 9097 }, { "epoch": 0.9680783145350075, "grad_norm": 10.5859375, "learning_rate": 9.694499715747585e-06, "loss": 1.3894, "step": 9098 }, { "epoch": 0.9681847201532241, "grad_norm": 12.0, "learning_rate": 9.694464184195566e-06, "loss": 1.4435, "step": 9099 }, { "epoch": 0.9682911257714407, "grad_norm": 12.9453125, "learning_rate": 9.694428652643548e-06, "loss": 1.1328, "step": 9100 }, { "epoch": 0.9683975313896573, "grad_norm": 13.1171875, "learning_rate": 9.69439312109153e-06, "loss": 1.3966, "step": 9101 }, { "epoch": 0.968503937007874, "grad_norm": 9.25, "learning_rate": 9.694357589539511e-06, "loss": 1.067, "step": 9102 }, { "epoch": 0.9686103426260907, "grad_norm": 18.171875, "learning_rate": 9.694322057987494e-06, "loss": 1.175, "step": 9103 }, { "epoch": 0.9687167482443073, "grad_norm": 10.3359375, "learning_rate": 9.694286526435476e-06, "loss": 1.1643, "step": 9104 }, { "epoch": 0.9688231538625239, "grad_norm": 16.59375, "learning_rate": 9.694250994883457e-06, "loss": 1.4057, "step": 9105 }, { "epoch": 0.9689295594807406, "grad_norm": 12.515625, "learning_rate": 9.694215463331439e-06, "loss": 1.1842, "step": 9106 }, { "epoch": 0.9690359650989572, "grad_norm": 11.1484375, "learning_rate": 9.69417993177942e-06, "loss": 1.2591, "step": 9107 }, { "epoch": 0.9691423707171739, "grad_norm": 9.46875, "learning_rate": 9.694144400227404e-06, "loss": 1.4094, "step": 9108 }, { "epoch": 0.9692487763353905, "grad_norm": 9.8046875, "learning_rate": 9.694108868675386e-06, "loss": 1.0444, "step": 9109 }, { "epoch": 0.9693551819536071, "grad_norm": 14.53125, "learning_rate": 9.694073337123367e-06, "loss": 1.2359, "step": 9110 }, { "epoch": 0.9694615875718238, "grad_norm": 14.0859375, "learning_rate": 9.694037805571349e-06, "loss": 1.2437, "step": 9111 }, { "epoch": 0.9695679931900404, "grad_norm": 11.8671875, "learning_rate": 9.69400227401933e-06, "loss": 1.3506, "step": 9112 }, { "epoch": 0.9696743988082571, "grad_norm": 10.375, "learning_rate": 9.693966742467312e-06, "loss": 1.0782, "step": 9113 }, { "epoch": 0.9697808044264737, "grad_norm": 20.875, "learning_rate": 9.693931210915294e-06, "loss": 1.1758, "step": 9114 }, { "epoch": 0.9698872100446904, "grad_norm": 10.9140625, "learning_rate": 9.693895679363275e-06, "loss": 1.1784, "step": 9115 }, { "epoch": 0.969993615662907, "grad_norm": 9.4921875, "learning_rate": 9.693860147811257e-06, "loss": 1.0979, "step": 9116 }, { "epoch": 0.9701000212811236, "grad_norm": 9.9453125, "learning_rate": 9.693824616259238e-06, "loss": 1.1742, "step": 9117 }, { "epoch": 0.9702064268993403, "grad_norm": 11.484375, "learning_rate": 9.693789084707222e-06, "loss": 1.2489, "step": 9118 }, { "epoch": 0.970312832517557, "grad_norm": 11.6953125, "learning_rate": 9.693753553155203e-06, "loss": 1.0795, "step": 9119 }, { "epoch": 0.9704192381357736, "grad_norm": 10.453125, "learning_rate": 9.693718021603185e-06, "loss": 1.1208, "step": 9120 }, { "epoch": 0.9705256437539902, "grad_norm": 11.03125, "learning_rate": 9.693682490051167e-06, "loss": 1.1811, "step": 9121 }, { "epoch": 0.9706320493722068, "grad_norm": 13.4921875, "learning_rate": 9.693646958499148e-06, "loss": 0.8754, "step": 9122 }, { "epoch": 0.9707384549904235, "grad_norm": 12.171875, "learning_rate": 9.69361142694713e-06, "loss": 1.519, "step": 9123 }, { "epoch": 0.9708448606086402, "grad_norm": 11.9609375, "learning_rate": 9.693575895395111e-06, "loss": 1.1432, "step": 9124 }, { "epoch": 0.9709512662268568, "grad_norm": 16.546875, "learning_rate": 9.693540363843093e-06, "loss": 1.2031, "step": 9125 }, { "epoch": 0.9710576718450734, "grad_norm": 15.0859375, "learning_rate": 9.693504832291075e-06, "loss": 1.1073, "step": 9126 }, { "epoch": 0.97116407746329, "grad_norm": 15.34375, "learning_rate": 9.693469300739056e-06, "loss": 1.0809, "step": 9127 }, { "epoch": 0.9712704830815067, "grad_norm": 11.25, "learning_rate": 9.693433769187038e-06, "loss": 0.9689, "step": 9128 }, { "epoch": 0.9713768886997234, "grad_norm": 10.796875, "learning_rate": 9.693398237635021e-06, "loss": 0.9258, "step": 9129 }, { "epoch": 0.97148329431794, "grad_norm": 13.71875, "learning_rate": 9.693362706083003e-06, "loss": 1.4175, "step": 9130 }, { "epoch": 0.9715896999361566, "grad_norm": 11.03125, "learning_rate": 9.693327174530984e-06, "loss": 1.1568, "step": 9131 }, { "epoch": 0.9716961055543732, "grad_norm": 10.9921875, "learning_rate": 9.693291642978966e-06, "loss": 1.0135, "step": 9132 }, { "epoch": 0.9718025111725899, "grad_norm": 15.09375, "learning_rate": 9.693256111426948e-06, "loss": 1.4628, "step": 9133 }, { "epoch": 0.9719089167908066, "grad_norm": 14.5390625, "learning_rate": 9.69322057987493e-06, "loss": 1.2303, "step": 9134 }, { "epoch": 0.9720153224090232, "grad_norm": 10.703125, "learning_rate": 9.693185048322912e-06, "loss": 1.0716, "step": 9135 }, { "epoch": 0.9721217280272398, "grad_norm": 12.046875, "learning_rate": 9.693149516770894e-06, "loss": 1.0034, "step": 9136 }, { "epoch": 0.9722281336454565, "grad_norm": 14.5, "learning_rate": 9.693113985218876e-06, "loss": 0.9645, "step": 9137 }, { "epoch": 0.9723345392636731, "grad_norm": 16.640625, "learning_rate": 9.693078453666857e-06, "loss": 1.5423, "step": 9138 }, { "epoch": 0.9724409448818898, "grad_norm": 9.5625, "learning_rate": 9.693042922114839e-06, "loss": 1.0676, "step": 9139 }, { "epoch": 0.9725473505001064, "grad_norm": 11.59375, "learning_rate": 9.69300739056282e-06, "loss": 1.0118, "step": 9140 }, { "epoch": 0.972653756118323, "grad_norm": 11.65625, "learning_rate": 9.692971859010802e-06, "loss": 1.4095, "step": 9141 }, { "epoch": 0.9727601617365397, "grad_norm": 18.734375, "learning_rate": 9.692936327458784e-06, "loss": 1.4162, "step": 9142 }, { "epoch": 0.9728665673547563, "grad_norm": 14.046875, "learning_rate": 9.692900795906765e-06, "loss": 1.58, "step": 9143 }, { "epoch": 0.972972972972973, "grad_norm": 14.1875, "learning_rate": 9.692865264354749e-06, "loss": 1.3787, "step": 9144 }, { "epoch": 0.9730793785911896, "grad_norm": 11.5625, "learning_rate": 9.69282973280273e-06, "loss": 1.2473, "step": 9145 }, { "epoch": 0.9731857842094063, "grad_norm": 8.3046875, "learning_rate": 9.692794201250712e-06, "loss": 0.9196, "step": 9146 }, { "epoch": 0.9732921898276229, "grad_norm": 9.390625, "learning_rate": 9.692758669698693e-06, "loss": 1.2355, "step": 9147 }, { "epoch": 0.9733985954458395, "grad_norm": 14.390625, "learning_rate": 9.692723138146675e-06, "loss": 1.2653, "step": 9148 }, { "epoch": 0.9735050010640561, "grad_norm": 10.1875, "learning_rate": 9.692687606594657e-06, "loss": 1.1168, "step": 9149 }, { "epoch": 0.9736114066822729, "grad_norm": 9.2421875, "learning_rate": 9.692652075042638e-06, "loss": 1.0904, "step": 9150 }, { "epoch": 0.9737178123004895, "grad_norm": 10.4375, "learning_rate": 9.69261654349062e-06, "loss": 1.1947, "step": 9151 }, { "epoch": 0.9738242179187061, "grad_norm": 11.8359375, "learning_rate": 9.692581011938601e-06, "loss": 1.5496, "step": 9152 }, { "epoch": 0.9739306235369227, "grad_norm": 13.6640625, "learning_rate": 9.692545480386583e-06, "loss": 1.2227, "step": 9153 }, { "epoch": 0.9740370291551393, "grad_norm": 10.1796875, "learning_rate": 9.692509948834566e-06, "loss": 1.1213, "step": 9154 }, { "epoch": 0.9741434347733561, "grad_norm": 13.4375, "learning_rate": 9.692474417282548e-06, "loss": 1.2072, "step": 9155 }, { "epoch": 0.9742498403915727, "grad_norm": 10.453125, "learning_rate": 9.69243888573053e-06, "loss": 0.8538, "step": 9156 }, { "epoch": 0.9743562460097893, "grad_norm": 13.46875, "learning_rate": 9.692403354178511e-06, "loss": 1.2486, "step": 9157 }, { "epoch": 0.9744626516280059, "grad_norm": 11.5546875, "learning_rate": 9.692367822626493e-06, "loss": 1.0501, "step": 9158 }, { "epoch": 0.9745690572462226, "grad_norm": 14.03125, "learning_rate": 9.692332291074474e-06, "loss": 1.036, "step": 9159 }, { "epoch": 0.9746754628644393, "grad_norm": 13.109375, "learning_rate": 9.692296759522458e-06, "loss": 1.6695, "step": 9160 }, { "epoch": 0.9747818684826559, "grad_norm": 10.0234375, "learning_rate": 9.69226122797044e-06, "loss": 0.9555, "step": 9161 }, { "epoch": 0.9748882741008725, "grad_norm": 22.234375, "learning_rate": 9.692225696418421e-06, "loss": 1.5256, "step": 9162 }, { "epoch": 0.9749946797190892, "grad_norm": 13.0390625, "learning_rate": 9.692190164866403e-06, "loss": 1.0585, "step": 9163 }, { "epoch": 0.9751010853373058, "grad_norm": 12.578125, "learning_rate": 9.692154633314384e-06, "loss": 1.5264, "step": 9164 }, { "epoch": 0.9752074909555225, "grad_norm": 11.3359375, "learning_rate": 9.692119101762366e-06, "loss": 1.2876, "step": 9165 }, { "epoch": 0.9753138965737391, "grad_norm": 11.296875, "learning_rate": 9.692083570210347e-06, "loss": 1.1473, "step": 9166 }, { "epoch": 0.9754203021919557, "grad_norm": 12.2421875, "learning_rate": 9.692048038658329e-06, "loss": 1.3236, "step": 9167 }, { "epoch": 0.9755267078101724, "grad_norm": 14.078125, "learning_rate": 9.69201250710631e-06, "loss": 1.2761, "step": 9168 }, { "epoch": 0.975633113428389, "grad_norm": 13.390625, "learning_rate": 9.691976975554292e-06, "loss": 1.0697, "step": 9169 }, { "epoch": 0.9757395190466056, "grad_norm": 15.546875, "learning_rate": 9.691941444002275e-06, "loss": 0.8755, "step": 9170 }, { "epoch": 0.9758459246648223, "grad_norm": 16.71875, "learning_rate": 9.691905912450257e-06, "loss": 1.1047, "step": 9171 }, { "epoch": 0.975952330283039, "grad_norm": 17.78125, "learning_rate": 9.691870380898239e-06, "loss": 1.1451, "step": 9172 }, { "epoch": 0.9760587359012556, "grad_norm": 11.9453125, "learning_rate": 9.69183484934622e-06, "loss": 0.9988, "step": 9173 }, { "epoch": 0.9761651415194722, "grad_norm": 15.8203125, "learning_rate": 9.691799317794202e-06, "loss": 1.073, "step": 9174 }, { "epoch": 0.9762715471376888, "grad_norm": 10.6171875, "learning_rate": 9.691763786242183e-06, "loss": 1.1924, "step": 9175 }, { "epoch": 0.9763779527559056, "grad_norm": 10.90625, "learning_rate": 9.691728254690165e-06, "loss": 1.191, "step": 9176 }, { "epoch": 0.9764843583741222, "grad_norm": 11.2109375, "learning_rate": 9.691692723138147e-06, "loss": 1.1943, "step": 9177 }, { "epoch": 0.9765907639923388, "grad_norm": 10.875, "learning_rate": 9.691657191586128e-06, "loss": 0.8675, "step": 9178 }, { "epoch": 0.9766971696105554, "grad_norm": 10.90625, "learning_rate": 9.69162166003411e-06, "loss": 1.2471, "step": 9179 }, { "epoch": 0.976803575228772, "grad_norm": 14.4296875, "learning_rate": 9.691586128482093e-06, "loss": 1.4611, "step": 9180 }, { "epoch": 0.9769099808469888, "grad_norm": 10.6328125, "learning_rate": 9.691550596930075e-06, "loss": 1.0586, "step": 9181 }, { "epoch": 0.9770163864652054, "grad_norm": 14.5859375, "learning_rate": 9.691515065378056e-06, "loss": 1.3565, "step": 9182 }, { "epoch": 0.977122792083422, "grad_norm": 17.84375, "learning_rate": 9.691479533826038e-06, "loss": 0.958, "step": 9183 }, { "epoch": 0.9772291977016386, "grad_norm": 10.4140625, "learning_rate": 9.69144400227402e-06, "loss": 1.2888, "step": 9184 }, { "epoch": 0.9773356033198553, "grad_norm": 20.4375, "learning_rate": 9.691408470722003e-06, "loss": 1.4586, "step": 9185 }, { "epoch": 0.977442008938072, "grad_norm": 12.9453125, "learning_rate": 9.691372939169985e-06, "loss": 1.2064, "step": 9186 }, { "epoch": 0.9775484145562886, "grad_norm": 10.0078125, "learning_rate": 9.691337407617966e-06, "loss": 0.8283, "step": 9187 }, { "epoch": 0.9776548201745052, "grad_norm": 17.890625, "learning_rate": 9.691301876065948e-06, "loss": 1.106, "step": 9188 }, { "epoch": 0.9777612257927218, "grad_norm": 11.5390625, "learning_rate": 9.69126634451393e-06, "loss": 1.4395, "step": 9189 }, { "epoch": 0.9778676314109385, "grad_norm": 16.125, "learning_rate": 9.691230812961911e-06, "loss": 1.3117, "step": 9190 }, { "epoch": 0.9779740370291551, "grad_norm": 13.3046875, "learning_rate": 9.691195281409893e-06, "loss": 1.3172, "step": 9191 }, { "epoch": 0.9780804426473718, "grad_norm": 14.359375, "learning_rate": 9.691159749857874e-06, "loss": 1.2366, "step": 9192 }, { "epoch": 0.9781868482655884, "grad_norm": 13.3046875, "learning_rate": 9.691124218305856e-06, "loss": 1.1139, "step": 9193 }, { "epoch": 0.9782932538838051, "grad_norm": 15.7421875, "learning_rate": 9.691088686753837e-06, "loss": 1.4269, "step": 9194 }, { "epoch": 0.9783996595020217, "grad_norm": 10.4921875, "learning_rate": 9.69105315520182e-06, "loss": 0.8954, "step": 9195 }, { "epoch": 0.9785060651202383, "grad_norm": 13.0625, "learning_rate": 9.691017623649802e-06, "loss": 0.8382, "step": 9196 }, { "epoch": 0.978612470738455, "grad_norm": 12.7265625, "learning_rate": 9.690982092097784e-06, "loss": 1.1688, "step": 9197 }, { "epoch": 0.9787188763566717, "grad_norm": 11.7578125, "learning_rate": 9.690946560545766e-06, "loss": 1.3001, "step": 9198 }, { "epoch": 0.9788252819748883, "grad_norm": 11.890625, "learning_rate": 9.690911028993747e-06, "loss": 1.0772, "step": 9199 }, { "epoch": 0.9789316875931049, "grad_norm": 15.875, "learning_rate": 9.690875497441729e-06, "loss": 1.118, "step": 9200 }, { "epoch": 0.9790380932113215, "grad_norm": 14.2421875, "learning_rate": 9.690839965889712e-06, "loss": 1.1858, "step": 9201 }, { "epoch": 0.9791444988295382, "grad_norm": 11.0625, "learning_rate": 9.690804434337694e-06, "loss": 1.0899, "step": 9202 }, { "epoch": 0.9792509044477549, "grad_norm": 11.5390625, "learning_rate": 9.690768902785674e-06, "loss": 1.3316, "step": 9203 }, { "epoch": 0.9793573100659715, "grad_norm": 15.703125, "learning_rate": 9.690733371233655e-06, "loss": 1.0037, "step": 9204 }, { "epoch": 0.9794637156841881, "grad_norm": 14.03125, "learning_rate": 9.690697839681637e-06, "loss": 1.0001, "step": 9205 }, { "epoch": 0.9795701213024047, "grad_norm": 16.609375, "learning_rate": 9.69066230812962e-06, "loss": 1.0725, "step": 9206 }, { "epoch": 0.9796765269206215, "grad_norm": 15.453125, "learning_rate": 9.690626776577602e-06, "loss": 1.0956, "step": 9207 }, { "epoch": 0.9797829325388381, "grad_norm": 10.96875, "learning_rate": 9.690591245025583e-06, "loss": 1.179, "step": 9208 }, { "epoch": 0.9798893381570547, "grad_norm": 11.375, "learning_rate": 9.690555713473565e-06, "loss": 1.2471, "step": 9209 }, { "epoch": 0.9799957437752713, "grad_norm": 14.0078125, "learning_rate": 9.690520181921546e-06, "loss": 1.5425, "step": 9210 }, { "epoch": 0.9801021493934879, "grad_norm": 10.8671875, "learning_rate": 9.69048465036953e-06, "loss": 0.9972, "step": 9211 }, { "epoch": 0.9802085550117046, "grad_norm": 11.1328125, "learning_rate": 9.690449118817511e-06, "loss": 1.5634, "step": 9212 }, { "epoch": 0.9803149606299213, "grad_norm": 12.2734375, "learning_rate": 9.690413587265493e-06, "loss": 1.0704, "step": 9213 }, { "epoch": 0.9804213662481379, "grad_norm": 9.7265625, "learning_rate": 9.690378055713475e-06, "loss": 1.1096, "step": 9214 }, { "epoch": 0.9805277718663545, "grad_norm": 13.8359375, "learning_rate": 9.690342524161456e-06, "loss": 1.4226, "step": 9215 }, { "epoch": 0.9806341774845712, "grad_norm": 14.09375, "learning_rate": 9.690306992609438e-06, "loss": 1.4107, "step": 9216 }, { "epoch": 0.9807405831027878, "grad_norm": 14.6171875, "learning_rate": 9.69027146105742e-06, "loss": 1.2186, "step": 9217 }, { "epoch": 0.9808469887210045, "grad_norm": 14.140625, "learning_rate": 9.690235929505401e-06, "loss": 0.8751, "step": 9218 }, { "epoch": 0.9809533943392211, "grad_norm": 18.203125, "learning_rate": 9.690200397953383e-06, "loss": 1.5259, "step": 9219 }, { "epoch": 0.9810597999574378, "grad_norm": 17.515625, "learning_rate": 9.690164866401364e-06, "loss": 1.0451, "step": 9220 }, { "epoch": 0.9811662055756544, "grad_norm": 10.125, "learning_rate": 9.690129334849348e-06, "loss": 1.2556, "step": 9221 }, { "epoch": 0.981272611193871, "grad_norm": 10.7109375, "learning_rate": 9.69009380329733e-06, "loss": 1.2847, "step": 9222 }, { "epoch": 0.9813790168120877, "grad_norm": 8.6796875, "learning_rate": 9.69005827174531e-06, "loss": 1.0618, "step": 9223 }, { "epoch": 0.9814854224303043, "grad_norm": 8.25, "learning_rate": 9.690022740193292e-06, "loss": 1.1423, "step": 9224 }, { "epoch": 0.981591828048521, "grad_norm": 9.171875, "learning_rate": 9.689987208641274e-06, "loss": 1.3888, "step": 9225 }, { "epoch": 0.9816982336667376, "grad_norm": 16.796875, "learning_rate": 9.689951677089257e-06, "loss": 1.0922, "step": 9226 }, { "epoch": 0.9818046392849542, "grad_norm": 13.6015625, "learning_rate": 9.689916145537239e-06, "loss": 1.0116, "step": 9227 }, { "epoch": 0.9819110449031709, "grad_norm": 13.2421875, "learning_rate": 9.68988061398522e-06, "loss": 1.2655, "step": 9228 }, { "epoch": 0.9820174505213876, "grad_norm": 14.1640625, "learning_rate": 9.689845082433202e-06, "loss": 1.0045, "step": 9229 }, { "epoch": 0.9821238561396042, "grad_norm": 9.8203125, "learning_rate": 9.689809550881184e-06, "loss": 0.8979, "step": 9230 }, { "epoch": 0.9822302617578208, "grad_norm": 9.4296875, "learning_rate": 9.689774019329165e-06, "loss": 1.1116, "step": 9231 }, { "epoch": 0.9823366673760374, "grad_norm": 11.203125, "learning_rate": 9.689738487777147e-06, "loss": 1.0925, "step": 9232 }, { "epoch": 0.9824430729942542, "grad_norm": 12.7734375, "learning_rate": 9.689702956225129e-06, "loss": 1.2441, "step": 9233 }, { "epoch": 0.9825494786124708, "grad_norm": 13.6875, "learning_rate": 9.68966742467311e-06, "loss": 1.0238, "step": 9234 }, { "epoch": 0.9826558842306874, "grad_norm": 13.9765625, "learning_rate": 9.689631893121092e-06, "loss": 1.0848, "step": 9235 }, { "epoch": 0.982762289848904, "grad_norm": 14.65625, "learning_rate": 9.689596361569075e-06, "loss": 0.9728, "step": 9236 }, { "epoch": 0.9828686954671206, "grad_norm": 13.703125, "learning_rate": 9.689560830017057e-06, "loss": 1.1653, "step": 9237 }, { "epoch": 0.9829751010853373, "grad_norm": 15.03125, "learning_rate": 9.689525298465038e-06, "loss": 1.0366, "step": 9238 }, { "epoch": 0.983081506703554, "grad_norm": 13.5625, "learning_rate": 9.68948976691302e-06, "loss": 0.8983, "step": 9239 }, { "epoch": 0.9831879123217706, "grad_norm": 11.5, "learning_rate": 9.689454235361001e-06, "loss": 1.3068, "step": 9240 }, { "epoch": 0.9832943179399872, "grad_norm": 12.859375, "learning_rate": 9.689418703808983e-06, "loss": 1.1702, "step": 9241 }, { "epoch": 0.9834007235582038, "grad_norm": 10.4453125, "learning_rate": 9.689383172256965e-06, "loss": 0.8905, "step": 9242 }, { "epoch": 0.9835071291764205, "grad_norm": 15.890625, "learning_rate": 9.689347640704946e-06, "loss": 1.2701, "step": 9243 }, { "epoch": 0.9836135347946372, "grad_norm": 10.90625, "learning_rate": 9.689312109152928e-06, "loss": 1.1756, "step": 9244 }, { "epoch": 0.9837199404128538, "grad_norm": 11.5703125, "learning_rate": 9.68927657760091e-06, "loss": 1.0297, "step": 9245 }, { "epoch": 0.9838263460310704, "grad_norm": 10.2578125, "learning_rate": 9.689241046048891e-06, "loss": 0.9583, "step": 9246 }, { "epoch": 0.9839327516492871, "grad_norm": 17.890625, "learning_rate": 9.689205514496874e-06, "loss": 1.1127, "step": 9247 }, { "epoch": 0.9840391572675037, "grad_norm": 13.3046875, "learning_rate": 9.689169982944856e-06, "loss": 1.0709, "step": 9248 }, { "epoch": 0.9841455628857204, "grad_norm": 12.609375, "learning_rate": 9.689134451392838e-06, "loss": 1.1049, "step": 9249 }, { "epoch": 0.984251968503937, "grad_norm": 12.125, "learning_rate": 9.68909891984082e-06, "loss": 1.2144, "step": 9250 }, { "epoch": 0.9843583741221537, "grad_norm": 18.125, "learning_rate": 9.6890633882888e-06, "loss": 1.4352, "step": 9251 }, { "epoch": 0.9844647797403703, "grad_norm": 10.6953125, "learning_rate": 9.689027856736784e-06, "loss": 1.0504, "step": 9252 }, { "epoch": 0.9845711853585869, "grad_norm": 11.6171875, "learning_rate": 9.688992325184766e-06, "loss": 1.0137, "step": 9253 }, { "epoch": 0.9846775909768036, "grad_norm": 13.03125, "learning_rate": 9.688956793632747e-06, "loss": 1.2432, "step": 9254 }, { "epoch": 0.9847839965950203, "grad_norm": 24.765625, "learning_rate": 9.688921262080729e-06, "loss": 1.1697, "step": 9255 }, { "epoch": 0.9848904022132369, "grad_norm": 14.0859375, "learning_rate": 9.68888573052871e-06, "loss": 0.9071, "step": 9256 }, { "epoch": 0.9849968078314535, "grad_norm": 11.296875, "learning_rate": 9.688850198976692e-06, "loss": 1.0401, "step": 9257 }, { "epoch": 0.9851032134496701, "grad_norm": 11.8671875, "learning_rate": 9.688814667424674e-06, "loss": 1.3596, "step": 9258 }, { "epoch": 0.9852096190678867, "grad_norm": 9.9765625, "learning_rate": 9.688779135872655e-06, "loss": 1.0911, "step": 9259 }, { "epoch": 0.9853160246861035, "grad_norm": 11.9453125, "learning_rate": 9.688743604320637e-06, "loss": 1.1363, "step": 9260 }, { "epoch": 0.9854224303043201, "grad_norm": 10.15625, "learning_rate": 9.688708072768619e-06, "loss": 0.9923, "step": 9261 }, { "epoch": 0.9855288359225367, "grad_norm": 11.4375, "learning_rate": 9.688672541216602e-06, "loss": 1.2809, "step": 9262 }, { "epoch": 0.9856352415407533, "grad_norm": 13.0, "learning_rate": 9.688637009664584e-06, "loss": 1.0303, "step": 9263 }, { "epoch": 0.98574164715897, "grad_norm": 13.828125, "learning_rate": 9.688601478112565e-06, "loss": 0.7628, "step": 9264 }, { "epoch": 0.9858480527771867, "grad_norm": 11.421875, "learning_rate": 9.688565946560547e-06, "loss": 1.2226, "step": 9265 }, { "epoch": 0.9859544583954033, "grad_norm": 11.5390625, "learning_rate": 9.688530415008528e-06, "loss": 1.1866, "step": 9266 }, { "epoch": 0.9860608640136199, "grad_norm": 11.7421875, "learning_rate": 9.68849488345651e-06, "loss": 1.2945, "step": 9267 }, { "epoch": 0.9861672696318365, "grad_norm": 13.015625, "learning_rate": 9.688459351904492e-06, "loss": 1.2888, "step": 9268 }, { "epoch": 0.9862736752500532, "grad_norm": 19.421875, "learning_rate": 9.688423820352473e-06, "loss": 1.5157, "step": 9269 }, { "epoch": 0.9863800808682699, "grad_norm": 13.2890625, "learning_rate": 9.688388288800455e-06, "loss": 1.4351, "step": 9270 }, { "epoch": 0.9864864864864865, "grad_norm": 14.3984375, "learning_rate": 9.688352757248436e-06, "loss": 1.4921, "step": 9271 }, { "epoch": 0.9865928921047031, "grad_norm": 11.109375, "learning_rate": 9.68831722569642e-06, "loss": 1.253, "step": 9272 }, { "epoch": 0.9866992977229198, "grad_norm": 9.8203125, "learning_rate": 9.688281694144401e-06, "loss": 0.7561, "step": 9273 }, { "epoch": 0.9868057033411364, "grad_norm": 14.2890625, "learning_rate": 9.688246162592383e-06, "loss": 1.0341, "step": 9274 }, { "epoch": 0.9869121089593531, "grad_norm": 13.8125, "learning_rate": 9.688210631040364e-06, "loss": 1.1881, "step": 9275 }, { "epoch": 0.9870185145775697, "grad_norm": 19.078125, "learning_rate": 9.688175099488346e-06, "loss": 1.6455, "step": 9276 }, { "epoch": 0.9871249201957863, "grad_norm": 18.984375, "learning_rate": 9.688139567936328e-06, "loss": 1.8575, "step": 9277 }, { "epoch": 0.987231325814003, "grad_norm": 11.0078125, "learning_rate": 9.688104036384311e-06, "loss": 1.1959, "step": 9278 }, { "epoch": 0.9873377314322196, "grad_norm": 9.625, "learning_rate": 9.688068504832293e-06, "loss": 1.2382, "step": 9279 }, { "epoch": 0.9874441370504362, "grad_norm": 11.7265625, "learning_rate": 9.688032973280274e-06, "loss": 1.1354, "step": 9280 }, { "epoch": 0.9875505426686529, "grad_norm": 11.625, "learning_rate": 9.687997441728256e-06, "loss": 1.0631, "step": 9281 }, { "epoch": 0.9876569482868696, "grad_norm": 12.4296875, "learning_rate": 9.687961910176237e-06, "loss": 0.9822, "step": 9282 }, { "epoch": 0.9877633539050862, "grad_norm": 12.71875, "learning_rate": 9.687926378624219e-06, "loss": 0.9824, "step": 9283 }, { "epoch": 0.9878697595233028, "grad_norm": 11.28125, "learning_rate": 9.6878908470722e-06, "loss": 1.0019, "step": 9284 }, { "epoch": 0.9879761651415194, "grad_norm": 10.7890625, "learning_rate": 9.687855315520182e-06, "loss": 1.2557, "step": 9285 }, { "epoch": 0.9880825707597362, "grad_norm": 11.0234375, "learning_rate": 9.687819783968164e-06, "loss": 1.1593, "step": 9286 }, { "epoch": 0.9881889763779528, "grad_norm": 13.375, "learning_rate": 9.687784252416145e-06, "loss": 1.0827, "step": 9287 }, { "epoch": 0.9882953819961694, "grad_norm": 14.4765625, "learning_rate": 9.687748720864129e-06, "loss": 1.2859, "step": 9288 }, { "epoch": 0.988401787614386, "grad_norm": 10.7578125, "learning_rate": 9.68771318931211e-06, "loss": 1.0897, "step": 9289 }, { "epoch": 0.9885081932326026, "grad_norm": 11.5234375, "learning_rate": 9.687677657760092e-06, "loss": 1.1663, "step": 9290 }, { "epoch": 0.9886145988508194, "grad_norm": 15.875, "learning_rate": 9.687642126208074e-06, "loss": 0.8508, "step": 9291 }, { "epoch": 0.988721004469036, "grad_norm": 12.390625, "learning_rate": 9.687606594656055e-06, "loss": 1.3329, "step": 9292 }, { "epoch": 0.9888274100872526, "grad_norm": 11.6796875, "learning_rate": 9.687571063104037e-06, "loss": 1.1017, "step": 9293 }, { "epoch": 0.9889338157054692, "grad_norm": 8.9140625, "learning_rate": 9.687535531552018e-06, "loss": 0.9876, "step": 9294 }, { "epoch": 0.9890402213236859, "grad_norm": 13.296875, "learning_rate": 9.6875e-06, "loss": 1.2775, "step": 9295 }, { "epoch": 0.9891466269419026, "grad_norm": 12.53125, "learning_rate": 9.687464468447982e-06, "loss": 1.2652, "step": 9296 }, { "epoch": 0.9892530325601192, "grad_norm": 14.0390625, "learning_rate": 9.687428936895963e-06, "loss": 1.1178, "step": 9297 }, { "epoch": 0.9893594381783358, "grad_norm": 13.75, "learning_rate": 9.687393405343947e-06, "loss": 1.1575, "step": 9298 }, { "epoch": 0.9894658437965524, "grad_norm": 10.828125, "learning_rate": 9.687357873791928e-06, "loss": 0.942, "step": 9299 }, { "epoch": 0.9895722494147691, "grad_norm": 13.1484375, "learning_rate": 9.68732234223991e-06, "loss": 1.3614, "step": 9300 }, { "epoch": 0.9896786550329857, "grad_norm": 16.484375, "learning_rate": 9.687286810687891e-06, "loss": 1.2839, "step": 9301 }, { "epoch": 0.9897850606512024, "grad_norm": 11.1953125, "learning_rate": 9.687251279135873e-06, "loss": 1.1855, "step": 9302 }, { "epoch": 0.989891466269419, "grad_norm": 10.6328125, "learning_rate": 9.687215747583856e-06, "loss": 1.175, "step": 9303 }, { "epoch": 0.9899978718876357, "grad_norm": 14.7265625, "learning_rate": 9.687180216031838e-06, "loss": 1.1458, "step": 9304 }, { "epoch": 0.9901042775058523, "grad_norm": 13.84375, "learning_rate": 9.68714468447982e-06, "loss": 1.499, "step": 9305 }, { "epoch": 0.9902106831240689, "grad_norm": 18.6875, "learning_rate": 9.687109152927801e-06, "loss": 1.4338, "step": 9306 }, { "epoch": 0.9903170887422856, "grad_norm": 14.4375, "learning_rate": 9.687073621375783e-06, "loss": 1.0776, "step": 9307 }, { "epoch": 0.9904234943605023, "grad_norm": 14.9296875, "learning_rate": 9.687038089823764e-06, "loss": 1.138, "step": 9308 }, { "epoch": 0.9905298999787189, "grad_norm": 13.0703125, "learning_rate": 9.687002558271746e-06, "loss": 1.0349, "step": 9309 }, { "epoch": 0.9906363055969355, "grad_norm": 15.078125, "learning_rate": 9.686967026719727e-06, "loss": 1.3856, "step": 9310 }, { "epoch": 0.9907427112151521, "grad_norm": 10.984375, "learning_rate": 9.686931495167709e-06, "loss": 1.0835, "step": 9311 }, { "epoch": 0.9908491168333688, "grad_norm": 13.3984375, "learning_rate": 9.68689596361569e-06, "loss": 1.2582, "step": 9312 }, { "epoch": 0.9909555224515855, "grad_norm": 18.625, "learning_rate": 9.686860432063674e-06, "loss": 1.3476, "step": 9313 }, { "epoch": 0.9910619280698021, "grad_norm": 11.6484375, "learning_rate": 9.686824900511656e-06, "loss": 1.1566, "step": 9314 }, { "epoch": 0.9911683336880187, "grad_norm": 19.671875, "learning_rate": 9.686789368959637e-06, "loss": 1.3184, "step": 9315 }, { "epoch": 0.9912747393062353, "grad_norm": 10.0625, "learning_rate": 9.686753837407619e-06, "loss": 1.156, "step": 9316 }, { "epoch": 0.9913811449244521, "grad_norm": 11.296875, "learning_rate": 9.6867183058556e-06, "loss": 1.1674, "step": 9317 }, { "epoch": 0.9914875505426687, "grad_norm": 10.0625, "learning_rate": 9.686682774303582e-06, "loss": 1.2725, "step": 9318 }, { "epoch": 0.9915939561608853, "grad_norm": 12.046875, "learning_rate": 9.686647242751565e-06, "loss": 0.9701, "step": 9319 }, { "epoch": 0.9917003617791019, "grad_norm": 13.265625, "learning_rate": 9.686611711199547e-06, "loss": 1.2001, "step": 9320 }, { "epoch": 0.9918067673973185, "grad_norm": 14.6640625, "learning_rate": 9.686576179647529e-06, "loss": 1.3299, "step": 9321 }, { "epoch": 0.9919131730155352, "grad_norm": 9.8125, "learning_rate": 9.68654064809551e-06, "loss": 1.0004, "step": 9322 }, { "epoch": 0.9920195786337519, "grad_norm": 14.8359375, "learning_rate": 9.68650511654349e-06, "loss": 1.2645, "step": 9323 }, { "epoch": 0.9921259842519685, "grad_norm": 12.8125, "learning_rate": 9.686469584991473e-06, "loss": 1.1092, "step": 9324 }, { "epoch": 0.9922323898701851, "grad_norm": 14.796875, "learning_rate": 9.686434053439455e-06, "loss": 1.1306, "step": 9325 }, { "epoch": 0.9923387954884018, "grad_norm": 14.5859375, "learning_rate": 9.686398521887437e-06, "loss": 1.2814, "step": 9326 }, { "epoch": 0.9924452011066184, "grad_norm": 12.3125, "learning_rate": 9.686362990335418e-06, "loss": 1.0644, "step": 9327 }, { "epoch": 0.9925516067248351, "grad_norm": 11.6484375, "learning_rate": 9.6863274587834e-06, "loss": 1.1498, "step": 9328 }, { "epoch": 0.9926580123430517, "grad_norm": 11.1796875, "learning_rate": 9.686291927231383e-06, "loss": 1.3763, "step": 9329 }, { "epoch": 0.9927644179612684, "grad_norm": 12.890625, "learning_rate": 9.686256395679365e-06, "loss": 1.2656, "step": 9330 }, { "epoch": 0.992870823579485, "grad_norm": 12.875, "learning_rate": 9.686220864127346e-06, "loss": 1.0691, "step": 9331 }, { "epoch": 0.9929772291977016, "grad_norm": 11.078125, "learning_rate": 9.686185332575328e-06, "loss": 1.1631, "step": 9332 }, { "epoch": 0.9930836348159183, "grad_norm": 22.4375, "learning_rate": 9.68614980102331e-06, "loss": 1.4618, "step": 9333 }, { "epoch": 0.993190040434135, "grad_norm": 15.359375, "learning_rate": 9.686114269471291e-06, "loss": 1.0979, "step": 9334 }, { "epoch": 0.9932964460523516, "grad_norm": 12.4921875, "learning_rate": 9.686078737919273e-06, "loss": 1.1088, "step": 9335 }, { "epoch": 0.9934028516705682, "grad_norm": 14.078125, "learning_rate": 9.686043206367254e-06, "loss": 1.2673, "step": 9336 }, { "epoch": 0.9935092572887848, "grad_norm": 13.6640625, "learning_rate": 9.686007674815236e-06, "loss": 0.9967, "step": 9337 }, { "epoch": 0.9936156629070015, "grad_norm": 13.0, "learning_rate": 9.685972143263218e-06, "loss": 1.1352, "step": 9338 }, { "epoch": 0.9937220685252182, "grad_norm": 12.7734375, "learning_rate": 9.685936611711201e-06, "loss": 1.4918, "step": 9339 }, { "epoch": 0.9938284741434348, "grad_norm": 11.265625, "learning_rate": 9.685901080159182e-06, "loss": 1.5691, "step": 9340 }, { "epoch": 0.9939348797616514, "grad_norm": 15.203125, "learning_rate": 9.685865548607164e-06, "loss": 1.0751, "step": 9341 }, { "epoch": 0.994041285379868, "grad_norm": 9.21875, "learning_rate": 9.685830017055146e-06, "loss": 0.7654, "step": 9342 }, { "epoch": 0.9941476909980848, "grad_norm": 14.34375, "learning_rate": 9.685794485503127e-06, "loss": 1.1001, "step": 9343 }, { "epoch": 0.9942540966163014, "grad_norm": 14.8203125, "learning_rate": 9.68575895395111e-06, "loss": 1.244, "step": 9344 }, { "epoch": 0.994360502234518, "grad_norm": 11.3046875, "learning_rate": 9.685723422399092e-06, "loss": 0.8778, "step": 9345 }, { "epoch": 0.9944669078527346, "grad_norm": 10.15625, "learning_rate": 9.685687890847074e-06, "loss": 0.9882, "step": 9346 }, { "epoch": 0.9945733134709512, "grad_norm": 15.015625, "learning_rate": 9.685652359295055e-06, "loss": 1.0778, "step": 9347 }, { "epoch": 0.9946797190891679, "grad_norm": 13.015625, "learning_rate": 9.685616827743037e-06, "loss": 1.1162, "step": 9348 }, { "epoch": 0.9947861247073846, "grad_norm": 14.328125, "learning_rate": 9.685581296191019e-06, "loss": 0.8954, "step": 9349 }, { "epoch": 0.9948925303256012, "grad_norm": 14.8203125, "learning_rate": 9.685545764639e-06, "loss": 1.3472, "step": 9350 }, { "epoch": 0.9949989359438178, "grad_norm": 12.5078125, "learning_rate": 9.685510233086982e-06, "loss": 0.9634, "step": 9351 }, { "epoch": 0.9951053415620345, "grad_norm": 14.46875, "learning_rate": 9.685474701534963e-06, "loss": 1.3562, "step": 9352 }, { "epoch": 0.9952117471802511, "grad_norm": 11.0390625, "learning_rate": 9.685439169982945e-06, "loss": 1.2419, "step": 9353 }, { "epoch": 0.9953181527984678, "grad_norm": 9.59375, "learning_rate": 9.685403638430927e-06, "loss": 1.0822, "step": 9354 }, { "epoch": 0.9954245584166844, "grad_norm": 10.9921875, "learning_rate": 9.68536810687891e-06, "loss": 1.0339, "step": 9355 }, { "epoch": 0.995530964034901, "grad_norm": 13.3984375, "learning_rate": 9.685332575326892e-06, "loss": 1.4639, "step": 9356 }, { "epoch": 0.9956373696531177, "grad_norm": 12.03125, "learning_rate": 9.685297043774873e-06, "loss": 0.9922, "step": 9357 }, { "epoch": 0.9957437752713343, "grad_norm": 10.3671875, "learning_rate": 9.685261512222855e-06, "loss": 1.2621, "step": 9358 }, { "epoch": 0.995850180889551, "grad_norm": 11.796875, "learning_rate": 9.685225980670836e-06, "loss": 1.2897, "step": 9359 }, { "epoch": 0.9959565865077676, "grad_norm": 11.828125, "learning_rate": 9.685190449118818e-06, "loss": 1.2679, "step": 9360 }, { "epoch": 0.9960629921259843, "grad_norm": 16.640625, "learning_rate": 9.6851549175668e-06, "loss": 0.8974, "step": 9361 }, { "epoch": 0.9961693977442009, "grad_norm": 15.4765625, "learning_rate": 9.685119386014781e-06, "loss": 1.2094, "step": 9362 }, { "epoch": 0.9962758033624175, "grad_norm": 14.125, "learning_rate": 9.685083854462763e-06, "loss": 1.1553, "step": 9363 }, { "epoch": 0.9963822089806342, "grad_norm": 9.8515625, "learning_rate": 9.685048322910744e-06, "loss": 0.9505, "step": 9364 }, { "epoch": 0.9964886145988509, "grad_norm": 13.9609375, "learning_rate": 9.685012791358728e-06, "loss": 1.2738, "step": 9365 }, { "epoch": 0.9965950202170675, "grad_norm": 14.90625, "learning_rate": 9.68497725980671e-06, "loss": 1.5312, "step": 9366 }, { "epoch": 0.9967014258352841, "grad_norm": 13.109375, "learning_rate": 9.684941728254691e-06, "loss": 1.1293, "step": 9367 }, { "epoch": 0.9968078314535007, "grad_norm": 19.390625, "learning_rate": 9.684906196702673e-06, "loss": 0.9832, "step": 9368 }, { "epoch": 0.9969142370717173, "grad_norm": 10.3828125, "learning_rate": 9.684870665150654e-06, "loss": 1.48, "step": 9369 }, { "epoch": 0.9970206426899341, "grad_norm": 20.828125, "learning_rate": 9.684835133598637e-06, "loss": 0.873, "step": 9370 }, { "epoch": 0.9971270483081507, "grad_norm": 11.328125, "learning_rate": 9.684799602046619e-06, "loss": 1.1849, "step": 9371 }, { "epoch": 0.9972334539263673, "grad_norm": 14.453125, "learning_rate": 9.6847640704946e-06, "loss": 1.2829, "step": 9372 }, { "epoch": 0.9973398595445839, "grad_norm": 13.7890625, "learning_rate": 9.684728538942582e-06, "loss": 1.2592, "step": 9373 }, { "epoch": 0.9974462651628005, "grad_norm": 14.1328125, "learning_rate": 9.684693007390564e-06, "loss": 1.1321, "step": 9374 }, { "epoch": 0.9975526707810173, "grad_norm": 13.046875, "learning_rate": 9.684657475838545e-06, "loss": 1.1348, "step": 9375 }, { "epoch": 0.9976590763992339, "grad_norm": 11.5, "learning_rate": 9.684621944286527e-06, "loss": 1.045, "step": 9376 }, { "epoch": 0.9977654820174505, "grad_norm": 14.8203125, "learning_rate": 9.684586412734509e-06, "loss": 1.1203, "step": 9377 }, { "epoch": 0.9978718876356671, "grad_norm": 14.875, "learning_rate": 9.68455088118249e-06, "loss": 1.2721, "step": 9378 }, { "epoch": 0.9979782932538838, "grad_norm": 16.53125, "learning_rate": 9.684515349630472e-06, "loss": 1.2497, "step": 9379 }, { "epoch": 0.9980846988721005, "grad_norm": 11.3515625, "learning_rate": 9.684479818078455e-06, "loss": 1.2841, "step": 9380 }, { "epoch": 0.9981911044903171, "grad_norm": 11.4140625, "learning_rate": 9.684444286526437e-06, "loss": 0.922, "step": 9381 }, { "epoch": 0.9982975101085337, "grad_norm": 13.625, "learning_rate": 9.684408754974418e-06, "loss": 1.3194, "step": 9382 }, { "epoch": 0.9984039157267504, "grad_norm": 15.640625, "learning_rate": 9.6843732234224e-06, "loss": 1.1197, "step": 9383 }, { "epoch": 0.998510321344967, "grad_norm": 16.859375, "learning_rate": 9.684337691870382e-06, "loss": 1.0314, "step": 9384 }, { "epoch": 0.9986167269631837, "grad_norm": 15.140625, "learning_rate": 9.684302160318363e-06, "loss": 1.0106, "step": 9385 }, { "epoch": 0.9987231325814003, "grad_norm": 25.546875, "learning_rate": 9.684266628766345e-06, "loss": 1.3878, "step": 9386 }, { "epoch": 0.998829538199617, "grad_norm": 10.625, "learning_rate": 9.684231097214326e-06, "loss": 1.0669, "step": 9387 }, { "epoch": 0.9989359438178336, "grad_norm": 10.953125, "learning_rate": 9.684195565662308e-06, "loss": 1.4729, "step": 9388 }, { "epoch": 0.9990423494360502, "grad_norm": 11.6484375, "learning_rate": 9.68416003411029e-06, "loss": 1.1413, "step": 9389 }, { "epoch": 0.9991487550542668, "grad_norm": 14.0859375, "learning_rate": 9.684124502558273e-06, "loss": 1.4805, "step": 9390 }, { "epoch": 0.9992551606724835, "grad_norm": 11.4296875, "learning_rate": 9.684088971006255e-06, "loss": 1.5362, "step": 9391 }, { "epoch": 0.9993615662907002, "grad_norm": 19.703125, "learning_rate": 9.684053439454236e-06, "loss": 0.9983, "step": 9392 }, { "epoch": 0.9994679719089168, "grad_norm": 16.890625, "learning_rate": 9.684017907902218e-06, "loss": 1.4069, "step": 9393 }, { "epoch": 0.9995743775271334, "grad_norm": 11.0, "learning_rate": 9.6839823763502e-06, "loss": 1.1869, "step": 9394 }, { "epoch": 0.99968078314535, "grad_norm": 9.6640625, "learning_rate": 9.683946844798181e-06, "loss": 1.3141, "step": 9395 }, { "epoch": 0.9997871887635668, "grad_norm": 16.359375, "learning_rate": 9.683911313246164e-06, "loss": 1.3115, "step": 9396 }, { "epoch": 0.9998935943817834, "grad_norm": 13.7421875, "learning_rate": 9.683875781694146e-06, "loss": 1.4585, "step": 9397 }, { "epoch": 1.0, "grad_norm": 14.15625, "learning_rate": 9.683840250142128e-06, "loss": 1.229, "step": 9398 }, { "epoch": 1.0001064056182167, "grad_norm": 11.484375, "learning_rate": 9.683804718590109e-06, "loss": 1.3199, "step": 9399 }, { "epoch": 1.0002128112364332, "grad_norm": 10.96875, "learning_rate": 9.68376918703809e-06, "loss": 1.3531, "step": 9400 }, { "epoch": 1.00031921685465, "grad_norm": 9.546875, "learning_rate": 9.683733655486072e-06, "loss": 1.009, "step": 9401 }, { "epoch": 1.0004256224728665, "grad_norm": 10.0546875, "learning_rate": 9.683698123934054e-06, "loss": 1.0039, "step": 9402 }, { "epoch": 1.0005320280910832, "grad_norm": 12.59375, "learning_rate": 9.683662592382036e-06, "loss": 1.1722, "step": 9403 }, { "epoch": 1.0006384337093, "grad_norm": 12.15625, "learning_rate": 9.683627060830017e-06, "loss": 1.1167, "step": 9404 }, { "epoch": 1.0007448393275165, "grad_norm": 12.7421875, "learning_rate": 9.683591529277999e-06, "loss": 1.1157, "step": 9405 }, { "epoch": 1.0008512449457332, "grad_norm": 17.046875, "learning_rate": 9.683555997725982e-06, "loss": 1.4697, "step": 9406 }, { "epoch": 1.0009576505639497, "grad_norm": 11.6484375, "learning_rate": 9.683520466173964e-06, "loss": 1.2796, "step": 9407 }, { "epoch": 1.0010640561821664, "grad_norm": 12.0546875, "learning_rate": 9.683484934621945e-06, "loss": 0.989, "step": 9408 }, { "epoch": 1.0011704618003832, "grad_norm": 10.890625, "learning_rate": 9.683449403069927e-06, "loss": 1.0393, "step": 9409 }, { "epoch": 1.0012768674185997, "grad_norm": 13.28125, "learning_rate": 9.683413871517908e-06, "loss": 1.0842, "step": 9410 }, { "epoch": 1.0013832730368164, "grad_norm": 9.65625, "learning_rate": 9.683378339965892e-06, "loss": 1.0215, "step": 9411 }, { "epoch": 1.001489678655033, "grad_norm": 11.8984375, "learning_rate": 9.683342808413872e-06, "loss": 1.0161, "step": 9412 }, { "epoch": 1.0015960842732496, "grad_norm": 13.3359375, "learning_rate": 9.683307276861853e-06, "loss": 1.1338, "step": 9413 }, { "epoch": 1.0017024898914664, "grad_norm": 15.2578125, "learning_rate": 9.683271745309835e-06, "loss": 1.0943, "step": 9414 }, { "epoch": 1.0018088955096829, "grad_norm": 10.8828125, "learning_rate": 9.683236213757817e-06, "loss": 1.2181, "step": 9415 }, { "epoch": 1.0019153011278996, "grad_norm": 10.90625, "learning_rate": 9.6832006822058e-06, "loss": 1.0993, "step": 9416 }, { "epoch": 1.0020217067461161, "grad_norm": 14.0859375, "learning_rate": 9.683165150653781e-06, "loss": 0.9484, "step": 9417 }, { "epoch": 1.0021281123643329, "grad_norm": 13.0, "learning_rate": 9.683129619101763e-06, "loss": 1.0211, "step": 9418 }, { "epoch": 1.0022345179825496, "grad_norm": 12.109375, "learning_rate": 9.683094087549745e-06, "loss": 0.9507, "step": 9419 }, { "epoch": 1.002340923600766, "grad_norm": 14.8046875, "learning_rate": 9.683058555997726e-06, "loss": 1.0348, "step": 9420 }, { "epoch": 1.0024473292189828, "grad_norm": 13.3671875, "learning_rate": 9.68302302444571e-06, "loss": 1.1307, "step": 9421 }, { "epoch": 1.0025537348371993, "grad_norm": 17.09375, "learning_rate": 9.682987492893691e-06, "loss": 1.4762, "step": 9422 }, { "epoch": 1.002660140455416, "grad_norm": 10.828125, "learning_rate": 9.682951961341673e-06, "loss": 1.337, "step": 9423 }, { "epoch": 1.0027665460736326, "grad_norm": 10.703125, "learning_rate": 9.682916429789654e-06, "loss": 1.0209, "step": 9424 }, { "epoch": 1.0028729516918493, "grad_norm": 9.4375, "learning_rate": 9.682880898237636e-06, "loss": 0.9024, "step": 9425 }, { "epoch": 1.002979357310066, "grad_norm": 12.828125, "learning_rate": 9.682845366685618e-06, "loss": 1.0632, "step": 9426 }, { "epoch": 1.0030857629282826, "grad_norm": 10.9609375, "learning_rate": 9.6828098351336e-06, "loss": 0.9421, "step": 9427 }, { "epoch": 1.0031921685464993, "grad_norm": 8.8203125, "learning_rate": 9.68277430358158e-06, "loss": 1.129, "step": 9428 }, { "epoch": 1.0032985741647158, "grad_norm": 12.0390625, "learning_rate": 9.682738772029562e-06, "loss": 1.1965, "step": 9429 }, { "epoch": 1.0034049797829325, "grad_norm": 10.390625, "learning_rate": 9.682703240477544e-06, "loss": 1.2228, "step": 9430 }, { "epoch": 1.0035113854011493, "grad_norm": 21.453125, "learning_rate": 9.682667708925527e-06, "loss": 1.1407, "step": 9431 }, { "epoch": 1.0036177910193658, "grad_norm": 16.625, "learning_rate": 9.682632177373509e-06, "loss": 0.9975, "step": 9432 }, { "epoch": 1.0037241966375825, "grad_norm": 12.109375, "learning_rate": 9.68259664582149e-06, "loss": 1.1538, "step": 9433 }, { "epoch": 1.003830602255799, "grad_norm": 10.28125, "learning_rate": 9.682561114269472e-06, "loss": 0.9602, "step": 9434 }, { "epoch": 1.0039370078740157, "grad_norm": 12.2109375, "learning_rate": 9.682525582717454e-06, "loss": 1.154, "step": 9435 }, { "epoch": 1.0040434134922325, "grad_norm": 11.7578125, "learning_rate": 9.682490051165435e-06, "loss": 1.1191, "step": 9436 }, { "epoch": 1.004149819110449, "grad_norm": 10.6953125, "learning_rate": 9.682454519613419e-06, "loss": 1.032, "step": 9437 }, { "epoch": 1.0042562247286657, "grad_norm": 13.0234375, "learning_rate": 9.6824189880614e-06, "loss": 0.9871, "step": 9438 }, { "epoch": 1.0043626303468822, "grad_norm": 25.359375, "learning_rate": 9.682383456509382e-06, "loss": 1.0842, "step": 9439 }, { "epoch": 1.004469035965099, "grad_norm": 11.296875, "learning_rate": 9.682347924957363e-06, "loss": 1.3319, "step": 9440 }, { "epoch": 1.0045754415833157, "grad_norm": 10.3125, "learning_rate": 9.682312393405345e-06, "loss": 1.1361, "step": 9441 }, { "epoch": 1.0046818472015322, "grad_norm": 14.453125, "learning_rate": 9.682276861853327e-06, "loss": 1.362, "step": 9442 }, { "epoch": 1.004788252819749, "grad_norm": 11.234375, "learning_rate": 9.682241330301308e-06, "loss": 1.115, "step": 9443 }, { "epoch": 1.0048946584379654, "grad_norm": 13.4375, "learning_rate": 9.68220579874929e-06, "loss": 1.1327, "step": 9444 }, { "epoch": 1.0050010640561822, "grad_norm": 14.3828125, "learning_rate": 9.682170267197271e-06, "loss": 0.9561, "step": 9445 }, { "epoch": 1.005107469674399, "grad_norm": 8.640625, "learning_rate": 9.682134735645253e-06, "loss": 0.9466, "step": 9446 }, { "epoch": 1.0052138752926154, "grad_norm": 11.8671875, "learning_rate": 9.682099204093236e-06, "loss": 0.9165, "step": 9447 }, { "epoch": 1.0053202809108321, "grad_norm": 12.609375, "learning_rate": 9.682063672541218e-06, "loss": 1.0341, "step": 9448 }, { "epoch": 1.0054266865290487, "grad_norm": 10.828125, "learning_rate": 9.6820281409892e-06, "loss": 1.1058, "step": 9449 }, { "epoch": 1.0055330921472654, "grad_norm": 11.6875, "learning_rate": 9.681992609437181e-06, "loss": 0.7611, "step": 9450 }, { "epoch": 1.0056394977654821, "grad_norm": 14.4609375, "learning_rate": 9.681957077885163e-06, "loss": 1.0889, "step": 9451 }, { "epoch": 1.0057459033836986, "grad_norm": 11.1171875, "learning_rate": 9.681921546333144e-06, "loss": 0.9435, "step": 9452 }, { "epoch": 1.0058523090019154, "grad_norm": 12.8359375, "learning_rate": 9.681886014781126e-06, "loss": 1.3804, "step": 9453 }, { "epoch": 1.0059587146201319, "grad_norm": 16.71875, "learning_rate": 9.681850483229108e-06, "loss": 0.9649, "step": 9454 }, { "epoch": 1.0060651202383486, "grad_norm": 10.9453125, "learning_rate": 9.68181495167709e-06, "loss": 1.0099, "step": 9455 }, { "epoch": 1.0061715258565653, "grad_norm": 13.3359375, "learning_rate": 9.681779420125071e-06, "loss": 1.0711, "step": 9456 }, { "epoch": 1.0062779314747818, "grad_norm": 12.5625, "learning_rate": 9.681743888573054e-06, "loss": 1.1427, "step": 9457 }, { "epoch": 1.0063843370929986, "grad_norm": 13.609375, "learning_rate": 9.681708357021036e-06, "loss": 0.8821, "step": 9458 }, { "epoch": 1.006490742711215, "grad_norm": 12.03125, "learning_rate": 9.681672825469017e-06, "loss": 1.0745, "step": 9459 }, { "epoch": 1.0065971483294318, "grad_norm": 12.125, "learning_rate": 9.681637293916999e-06, "loss": 1.1601, "step": 9460 }, { "epoch": 1.0067035539476485, "grad_norm": 11.4375, "learning_rate": 9.68160176236498e-06, "loss": 1.2323, "step": 9461 }, { "epoch": 1.006809959565865, "grad_norm": 14.015625, "learning_rate": 9.681566230812964e-06, "loss": 0.9594, "step": 9462 }, { "epoch": 1.0069163651840818, "grad_norm": 15.890625, "learning_rate": 9.681530699260945e-06, "loss": 1.3018, "step": 9463 }, { "epoch": 1.0070227708022983, "grad_norm": 15.125, "learning_rate": 9.681495167708927e-06, "loss": 1.1902, "step": 9464 }, { "epoch": 1.007129176420515, "grad_norm": 14.328125, "learning_rate": 9.681459636156909e-06, "loss": 1.1071, "step": 9465 }, { "epoch": 1.0072355820387315, "grad_norm": 12.9375, "learning_rate": 9.68142410460489e-06, "loss": 1.1781, "step": 9466 }, { "epoch": 1.0073419876569483, "grad_norm": 11.171875, "learning_rate": 9.681388573052872e-06, "loss": 1.1732, "step": 9467 }, { "epoch": 1.007448393275165, "grad_norm": 9.1171875, "learning_rate": 9.681353041500854e-06, "loss": 0.8413, "step": 9468 }, { "epoch": 1.0075547988933815, "grad_norm": 13.5078125, "learning_rate": 9.681317509948835e-06, "loss": 1.2726, "step": 9469 }, { "epoch": 1.0076612045115982, "grad_norm": 9.8671875, "learning_rate": 9.681281978396817e-06, "loss": 1.0318, "step": 9470 }, { "epoch": 1.0077676101298148, "grad_norm": 15.078125, "learning_rate": 9.681246446844798e-06, "loss": 0.7954, "step": 9471 }, { "epoch": 1.0078740157480315, "grad_norm": 15.4609375, "learning_rate": 9.68121091529278e-06, "loss": 1.2558, "step": 9472 }, { "epoch": 1.0079804213662482, "grad_norm": 11.6328125, "learning_rate": 9.681175383740763e-06, "loss": 1.0845, "step": 9473 }, { "epoch": 1.0080868269844647, "grad_norm": 10.6484375, "learning_rate": 9.681139852188745e-06, "loss": 1.1861, "step": 9474 }, { "epoch": 1.0081932326026815, "grad_norm": 19.5625, "learning_rate": 9.681104320636726e-06, "loss": 1.202, "step": 9475 }, { "epoch": 1.008299638220898, "grad_norm": 16.4375, "learning_rate": 9.681068789084708e-06, "loss": 1.2745, "step": 9476 }, { "epoch": 1.0084060438391147, "grad_norm": 11.515625, "learning_rate": 9.68103325753269e-06, "loss": 0.8471, "step": 9477 }, { "epoch": 1.0085124494573314, "grad_norm": 15.2265625, "learning_rate": 9.680997725980671e-06, "loss": 1.0506, "step": 9478 }, { "epoch": 1.008618855075548, "grad_norm": 16.6875, "learning_rate": 9.680962194428653e-06, "loss": 1.2459, "step": 9479 }, { "epoch": 1.0087252606937647, "grad_norm": 15.2265625, "learning_rate": 9.680926662876634e-06, "loss": 1.3914, "step": 9480 }, { "epoch": 1.0088316663119812, "grad_norm": 12.171875, "learning_rate": 9.680891131324616e-06, "loss": 1.0218, "step": 9481 }, { "epoch": 1.008938071930198, "grad_norm": 13.71875, "learning_rate": 9.680855599772598e-06, "loss": 1.1729, "step": 9482 }, { "epoch": 1.0090444775484146, "grad_norm": 15.5078125, "learning_rate": 9.680820068220581e-06, "loss": 0.8862, "step": 9483 }, { "epoch": 1.0091508831666312, "grad_norm": 14.3828125, "learning_rate": 9.680784536668563e-06, "loss": 1.2865, "step": 9484 }, { "epoch": 1.0092572887848479, "grad_norm": 10.7734375, "learning_rate": 9.680749005116544e-06, "loss": 0.97, "step": 9485 }, { "epoch": 1.0093636944030644, "grad_norm": 13.09375, "learning_rate": 9.680713473564526e-06, "loss": 1.1272, "step": 9486 }, { "epoch": 1.0094701000212811, "grad_norm": 20.53125, "learning_rate": 9.680677942012507e-06, "loss": 1.0507, "step": 9487 }, { "epoch": 1.0095765056394979, "grad_norm": 10.8828125, "learning_rate": 9.68064241046049e-06, "loss": 0.9818, "step": 9488 }, { "epoch": 1.0096829112577144, "grad_norm": 11.9453125, "learning_rate": 9.680606878908472e-06, "loss": 1.2254, "step": 9489 }, { "epoch": 1.009789316875931, "grad_norm": 14.265625, "learning_rate": 9.680571347356454e-06, "loss": 1.2122, "step": 9490 }, { "epoch": 1.0098957224941476, "grad_norm": 9.3203125, "learning_rate": 9.680535815804436e-06, "loss": 0.8681, "step": 9491 }, { "epoch": 1.0100021281123643, "grad_norm": 12.2109375, "learning_rate": 9.680500284252417e-06, "loss": 0.8534, "step": 9492 }, { "epoch": 1.010108533730581, "grad_norm": 10.2734375, "learning_rate": 9.680464752700399e-06, "loss": 1.0143, "step": 9493 }, { "epoch": 1.0102149393487976, "grad_norm": 11.5078125, "learning_rate": 9.68042922114838e-06, "loss": 0.9067, "step": 9494 }, { "epoch": 1.0103213449670143, "grad_norm": 11.8515625, "learning_rate": 9.680393689596362e-06, "loss": 1.1586, "step": 9495 }, { "epoch": 1.0104277505852308, "grad_norm": 14.3671875, "learning_rate": 9.680358158044344e-06, "loss": 1.2618, "step": 9496 }, { "epoch": 1.0105341562034476, "grad_norm": 12.3671875, "learning_rate": 9.680322626492325e-06, "loss": 1.2625, "step": 9497 }, { "epoch": 1.0106405618216643, "grad_norm": 18.78125, "learning_rate": 9.680287094940309e-06, "loss": 1.0358, "step": 9498 }, { "epoch": 1.0107469674398808, "grad_norm": 12.65625, "learning_rate": 9.68025156338829e-06, "loss": 1.1558, "step": 9499 }, { "epoch": 1.0108533730580975, "grad_norm": 12.3203125, "learning_rate": 9.680216031836272e-06, "loss": 1.1078, "step": 9500 }, { "epoch": 1.010959778676314, "grad_norm": 12.65625, "learning_rate": 9.680180500284253e-06, "loss": 1.0876, "step": 9501 }, { "epoch": 1.0110661842945308, "grad_norm": 11.6640625, "learning_rate": 9.680144968732235e-06, "loss": 0.9328, "step": 9502 }, { "epoch": 1.0111725899127475, "grad_norm": 13.078125, "learning_rate": 9.680109437180217e-06, "loss": 0.8944, "step": 9503 }, { "epoch": 1.011278995530964, "grad_norm": 11.9921875, "learning_rate": 9.680073905628198e-06, "loss": 1.2323, "step": 9504 }, { "epoch": 1.0113854011491807, "grad_norm": 13.3125, "learning_rate": 9.68003837407618e-06, "loss": 1.1746, "step": 9505 }, { "epoch": 1.0114918067673973, "grad_norm": 14.0234375, "learning_rate": 9.680002842524161e-06, "loss": 1.1776, "step": 9506 }, { "epoch": 1.011598212385614, "grad_norm": 12.796875, "learning_rate": 9.679967310972143e-06, "loss": 0.9668, "step": 9507 }, { "epoch": 1.0117046180038307, "grad_norm": 14.09375, "learning_rate": 9.679931779420126e-06, "loss": 1.4907, "step": 9508 }, { "epoch": 1.0118110236220472, "grad_norm": 10.5703125, "learning_rate": 9.679896247868108e-06, "loss": 0.9873, "step": 9509 }, { "epoch": 1.011917429240264, "grad_norm": 14.1953125, "learning_rate": 9.67986071631609e-06, "loss": 0.9078, "step": 9510 }, { "epoch": 1.0120238348584805, "grad_norm": 11.7265625, "learning_rate": 9.679825184764071e-06, "loss": 1.2193, "step": 9511 }, { "epoch": 1.0121302404766972, "grad_norm": 12.15625, "learning_rate": 9.679789653212053e-06, "loss": 1.1632, "step": 9512 }, { "epoch": 1.0122366460949137, "grad_norm": 11.2734375, "learning_rate": 9.679754121660034e-06, "loss": 0.8996, "step": 9513 }, { "epoch": 1.0123430517131304, "grad_norm": 10.234375, "learning_rate": 9.679718590108018e-06, "loss": 1.0701, "step": 9514 }, { "epoch": 1.0124494573313472, "grad_norm": 12.6484375, "learning_rate": 9.679683058556e-06, "loss": 1.1392, "step": 9515 }, { "epoch": 1.0125558629495637, "grad_norm": 13.7421875, "learning_rate": 9.67964752700398e-06, "loss": 0.8605, "step": 9516 }, { "epoch": 1.0126622685677804, "grad_norm": 9.859375, "learning_rate": 9.679611995451962e-06, "loss": 1.207, "step": 9517 }, { "epoch": 1.012768674185997, "grad_norm": 10.2734375, "learning_rate": 9.679576463899944e-06, "loss": 1.0985, "step": 9518 }, { "epoch": 1.0128750798042137, "grad_norm": 10.5234375, "learning_rate": 9.679540932347926e-06, "loss": 1.1641, "step": 9519 }, { "epoch": 1.0129814854224304, "grad_norm": 13.5234375, "learning_rate": 9.679505400795907e-06, "loss": 1.327, "step": 9520 }, { "epoch": 1.013087891040647, "grad_norm": 12.0625, "learning_rate": 9.679469869243889e-06, "loss": 1.2531, "step": 9521 }, { "epoch": 1.0131942966588636, "grad_norm": 21.359375, "learning_rate": 9.67943433769187e-06, "loss": 1.3315, "step": 9522 }, { "epoch": 1.0133007022770801, "grad_norm": 18.234375, "learning_rate": 9.679398806139852e-06, "loss": 0.8686, "step": 9523 }, { "epoch": 1.0134071078952969, "grad_norm": 13.609375, "learning_rate": 9.679363274587835e-06, "loss": 1.1403, "step": 9524 }, { "epoch": 1.0135135135135136, "grad_norm": 17.1875, "learning_rate": 9.679327743035817e-06, "loss": 1.0369, "step": 9525 }, { "epoch": 1.01361991913173, "grad_norm": 11.0234375, "learning_rate": 9.679292211483799e-06, "loss": 1.0802, "step": 9526 }, { "epoch": 1.0137263247499468, "grad_norm": 17.125, "learning_rate": 9.67925667993178e-06, "loss": 1.2846, "step": 9527 }, { "epoch": 1.0138327303681633, "grad_norm": 13.96875, "learning_rate": 9.679221148379762e-06, "loss": 1.3471, "step": 9528 }, { "epoch": 1.01393913598638, "grad_norm": 12.2578125, "learning_rate": 9.679185616827745e-06, "loss": 1.0425, "step": 9529 }, { "epoch": 1.0140455416045968, "grad_norm": 12.9765625, "learning_rate": 9.679150085275727e-06, "loss": 1.1532, "step": 9530 }, { "epoch": 1.0141519472228133, "grad_norm": 11.84375, "learning_rate": 9.679114553723708e-06, "loss": 0.9414, "step": 9531 }, { "epoch": 1.01425835284103, "grad_norm": 12.9765625, "learning_rate": 9.679079022171688e-06, "loss": 1.0271, "step": 9532 }, { "epoch": 1.0143647584592466, "grad_norm": 18.796875, "learning_rate": 9.67904349061967e-06, "loss": 1.2913, "step": 9533 }, { "epoch": 1.0144711640774633, "grad_norm": 15.578125, "learning_rate": 9.679007959067653e-06, "loss": 0.8242, "step": 9534 }, { "epoch": 1.01457756969568, "grad_norm": 16.28125, "learning_rate": 9.678972427515635e-06, "loss": 0.9537, "step": 9535 }, { "epoch": 1.0146839753138965, "grad_norm": 11.46875, "learning_rate": 9.678936895963616e-06, "loss": 0.9235, "step": 9536 }, { "epoch": 1.0147903809321133, "grad_norm": 8.4296875, "learning_rate": 9.678901364411598e-06, "loss": 0.8145, "step": 9537 }, { "epoch": 1.0148967865503298, "grad_norm": 13.671875, "learning_rate": 9.67886583285958e-06, "loss": 1.398, "step": 9538 }, { "epoch": 1.0150031921685465, "grad_norm": 14.625, "learning_rate": 9.678830301307563e-06, "loss": 1.0698, "step": 9539 }, { "epoch": 1.0151095977867632, "grad_norm": 12.96875, "learning_rate": 9.678794769755544e-06, "loss": 0.9991, "step": 9540 }, { "epoch": 1.0152160034049797, "grad_norm": 12.34375, "learning_rate": 9.678759238203526e-06, "loss": 0.9482, "step": 9541 }, { "epoch": 1.0153224090231965, "grad_norm": 13.0703125, "learning_rate": 9.678723706651508e-06, "loss": 1.3255, "step": 9542 }, { "epoch": 1.015428814641413, "grad_norm": 18.078125, "learning_rate": 9.67868817509949e-06, "loss": 1.2258, "step": 9543 }, { "epoch": 1.0155352202596297, "grad_norm": 13.7109375, "learning_rate": 9.678652643547471e-06, "loss": 1.0259, "step": 9544 }, { "epoch": 1.0156416258778465, "grad_norm": 20.5625, "learning_rate": 9.678617111995452e-06, "loss": 1.2815, "step": 9545 }, { "epoch": 1.015748031496063, "grad_norm": 9.6328125, "learning_rate": 9.678581580443434e-06, "loss": 1.1154, "step": 9546 }, { "epoch": 1.0158544371142797, "grad_norm": 14.5, "learning_rate": 9.678546048891416e-06, "loss": 1.0027, "step": 9547 }, { "epoch": 1.0159608427324962, "grad_norm": 14.1171875, "learning_rate": 9.678510517339397e-06, "loss": 1.0838, "step": 9548 }, { "epoch": 1.016067248350713, "grad_norm": 14.265625, "learning_rate": 9.67847498578738e-06, "loss": 1.061, "step": 9549 }, { "epoch": 1.0161736539689294, "grad_norm": 11.328125, "learning_rate": 9.678439454235362e-06, "loss": 1.0729, "step": 9550 }, { "epoch": 1.0162800595871462, "grad_norm": 17.546875, "learning_rate": 9.678403922683344e-06, "loss": 1.0118, "step": 9551 }, { "epoch": 1.016386465205363, "grad_norm": 16.421875, "learning_rate": 9.678368391131325e-06, "loss": 1.0672, "step": 9552 }, { "epoch": 1.0164928708235794, "grad_norm": 11.34375, "learning_rate": 9.678332859579307e-06, "loss": 1.1721, "step": 9553 }, { "epoch": 1.0165992764417962, "grad_norm": 12.4453125, "learning_rate": 9.678297328027289e-06, "loss": 1.0451, "step": 9554 }, { "epoch": 1.0167056820600127, "grad_norm": 13.265625, "learning_rate": 9.678261796475272e-06, "loss": 0.9684, "step": 9555 }, { "epoch": 1.0168120876782294, "grad_norm": 16.421875, "learning_rate": 9.678226264923254e-06, "loss": 0.9398, "step": 9556 }, { "epoch": 1.0169184932964461, "grad_norm": 16.296875, "learning_rate": 9.678190733371235e-06, "loss": 0.9465, "step": 9557 }, { "epoch": 1.0170248989146626, "grad_norm": 13.3984375, "learning_rate": 9.678155201819217e-06, "loss": 1.1952, "step": 9558 }, { "epoch": 1.0171313045328794, "grad_norm": 11.46875, "learning_rate": 9.678119670267198e-06, "loss": 1.0008, "step": 9559 }, { "epoch": 1.0172377101510959, "grad_norm": 12.9140625, "learning_rate": 9.67808413871518e-06, "loss": 1.1448, "step": 9560 }, { "epoch": 1.0173441157693126, "grad_norm": 12.140625, "learning_rate": 9.678048607163162e-06, "loss": 1.1798, "step": 9561 }, { "epoch": 1.0174505213875293, "grad_norm": 9.34375, "learning_rate": 9.678013075611143e-06, "loss": 1.1043, "step": 9562 }, { "epoch": 1.0175569270057458, "grad_norm": 13.4375, "learning_rate": 9.677977544059125e-06, "loss": 1.1907, "step": 9563 }, { "epoch": 1.0176633326239626, "grad_norm": 10.5, "learning_rate": 9.677942012507106e-06, "loss": 1.125, "step": 9564 }, { "epoch": 1.017769738242179, "grad_norm": 9.0703125, "learning_rate": 9.67790648095509e-06, "loss": 1.0477, "step": 9565 }, { "epoch": 1.0178761438603958, "grad_norm": 12.609375, "learning_rate": 9.677870949403071e-06, "loss": 0.9296, "step": 9566 }, { "epoch": 1.0179825494786126, "grad_norm": 12.7890625, "learning_rate": 9.677835417851053e-06, "loss": 1.1963, "step": 9567 }, { "epoch": 1.018088955096829, "grad_norm": 15.8203125, "learning_rate": 9.677799886299035e-06, "loss": 0.998, "step": 9568 }, { "epoch": 1.0181953607150458, "grad_norm": 9.8828125, "learning_rate": 9.677764354747016e-06, "loss": 1.1214, "step": 9569 }, { "epoch": 1.0183017663332623, "grad_norm": 12.171875, "learning_rate": 9.677728823194998e-06, "loss": 1.2836, "step": 9570 }, { "epoch": 1.018408171951479, "grad_norm": 11.6171875, "learning_rate": 9.67769329164298e-06, "loss": 1.1571, "step": 9571 }, { "epoch": 1.0185145775696958, "grad_norm": 11.328125, "learning_rate": 9.677657760090961e-06, "loss": 1.2431, "step": 9572 }, { "epoch": 1.0186209831879123, "grad_norm": 12.1875, "learning_rate": 9.677622228538943e-06, "loss": 0.8953, "step": 9573 }, { "epoch": 1.018727388806129, "grad_norm": 12.359375, "learning_rate": 9.677586696986924e-06, "loss": 0.9735, "step": 9574 }, { "epoch": 1.0188337944243455, "grad_norm": 11.7734375, "learning_rate": 9.677551165434907e-06, "loss": 1.0724, "step": 9575 }, { "epoch": 1.0189402000425622, "grad_norm": 15.1328125, "learning_rate": 9.677515633882889e-06, "loss": 0.8173, "step": 9576 }, { "epoch": 1.019046605660779, "grad_norm": 13.8984375, "learning_rate": 9.67748010233087e-06, "loss": 1.3685, "step": 9577 }, { "epoch": 1.0191530112789955, "grad_norm": 14.546875, "learning_rate": 9.677444570778852e-06, "loss": 1.1032, "step": 9578 }, { "epoch": 1.0192594168972122, "grad_norm": 15.6484375, "learning_rate": 9.677409039226834e-06, "loss": 1.1327, "step": 9579 }, { "epoch": 1.0193658225154287, "grad_norm": 8.9921875, "learning_rate": 9.677373507674817e-06, "loss": 0.8139, "step": 9580 }, { "epoch": 1.0194722281336455, "grad_norm": 14.6171875, "learning_rate": 9.677337976122799e-06, "loss": 0.8333, "step": 9581 }, { "epoch": 1.0195786337518622, "grad_norm": 12.4921875, "learning_rate": 9.67730244457078e-06, "loss": 1.2416, "step": 9582 }, { "epoch": 1.0196850393700787, "grad_norm": 14.6484375, "learning_rate": 9.677266913018762e-06, "loss": 1.3396, "step": 9583 }, { "epoch": 1.0197914449882954, "grad_norm": 16.15625, "learning_rate": 9.677231381466744e-06, "loss": 1.1171, "step": 9584 }, { "epoch": 1.019897850606512, "grad_norm": 23.1875, "learning_rate": 9.677195849914725e-06, "loss": 1.3291, "step": 9585 }, { "epoch": 1.0200042562247287, "grad_norm": 14.203125, "learning_rate": 9.677160318362707e-06, "loss": 1.2337, "step": 9586 }, { "epoch": 1.0201106618429454, "grad_norm": 11.8671875, "learning_rate": 9.677124786810688e-06, "loss": 1.0445, "step": 9587 }, { "epoch": 1.020217067461162, "grad_norm": 14.6015625, "learning_rate": 9.67708925525867e-06, "loss": 0.9634, "step": 9588 }, { "epoch": 1.0203234730793787, "grad_norm": 11.390625, "learning_rate": 9.677053723706652e-06, "loss": 1.1244, "step": 9589 }, { "epoch": 1.0204298786975952, "grad_norm": 10.96875, "learning_rate": 9.677018192154633e-06, "loss": 0.9867, "step": 9590 }, { "epoch": 1.020536284315812, "grad_norm": 10.0078125, "learning_rate": 9.676982660602617e-06, "loss": 0.9144, "step": 9591 }, { "epoch": 1.0206426899340286, "grad_norm": 11.0234375, "learning_rate": 9.676947129050598e-06, "loss": 0.8707, "step": 9592 }, { "epoch": 1.0207490955522451, "grad_norm": 12.2734375, "learning_rate": 9.67691159749858e-06, "loss": 1.1206, "step": 9593 }, { "epoch": 1.0208555011704619, "grad_norm": 11.4765625, "learning_rate": 9.676876065946561e-06, "loss": 1.2649, "step": 9594 }, { "epoch": 1.0209619067886784, "grad_norm": 13.59375, "learning_rate": 9.676840534394543e-06, "loss": 1.0398, "step": 9595 }, { "epoch": 1.021068312406895, "grad_norm": 14.0859375, "learning_rate": 9.676805002842525e-06, "loss": 1.156, "step": 9596 }, { "epoch": 1.0211747180251116, "grad_norm": 13.8984375, "learning_rate": 9.676769471290506e-06, "loss": 0.9772, "step": 9597 }, { "epoch": 1.0212811236433283, "grad_norm": 12.0234375, "learning_rate": 9.676733939738488e-06, "loss": 1.2766, "step": 9598 }, { "epoch": 1.021387529261545, "grad_norm": 10.9765625, "learning_rate": 9.67669840818647e-06, "loss": 1.0457, "step": 9599 }, { "epoch": 1.0214939348797616, "grad_norm": 13.46875, "learning_rate": 9.676662876634451e-06, "loss": 1.1635, "step": 9600 }, { "epoch": 1.0216003404979783, "grad_norm": 9.59375, "learning_rate": 9.676627345082434e-06, "loss": 1.0258, "step": 9601 }, { "epoch": 1.0217067461161948, "grad_norm": 11.0234375, "learning_rate": 9.676591813530416e-06, "loss": 1.0354, "step": 9602 }, { "epoch": 1.0218131517344116, "grad_norm": 11.9765625, "learning_rate": 9.676556281978398e-06, "loss": 1.0022, "step": 9603 }, { "epoch": 1.0219195573526283, "grad_norm": 16.625, "learning_rate": 9.676520750426379e-06, "loss": 1.4106, "step": 9604 }, { "epoch": 1.0220259629708448, "grad_norm": 15.3984375, "learning_rate": 9.67648521887436e-06, "loss": 1.2436, "step": 9605 }, { "epoch": 1.0221323685890615, "grad_norm": 14.046875, "learning_rate": 9.676449687322344e-06, "loss": 0.9161, "step": 9606 }, { "epoch": 1.022238774207278, "grad_norm": 10.96875, "learning_rate": 9.676414155770326e-06, "loss": 0.9628, "step": 9607 }, { "epoch": 1.0223451798254948, "grad_norm": 10.8125, "learning_rate": 9.676378624218307e-06, "loss": 1.0325, "step": 9608 }, { "epoch": 1.0224515854437115, "grad_norm": 12.0703125, "learning_rate": 9.676343092666289e-06, "loss": 1.1332, "step": 9609 }, { "epoch": 1.022557991061928, "grad_norm": 15.2890625, "learning_rate": 9.67630756111427e-06, "loss": 0.9757, "step": 9610 }, { "epoch": 1.0226643966801447, "grad_norm": 10.75, "learning_rate": 9.676272029562252e-06, "loss": 0.9864, "step": 9611 }, { "epoch": 1.0227708022983613, "grad_norm": 9.921875, "learning_rate": 9.676236498010234e-06, "loss": 0.887, "step": 9612 }, { "epoch": 1.022877207916578, "grad_norm": 11.71875, "learning_rate": 9.676200966458215e-06, "loss": 1.3174, "step": 9613 }, { "epoch": 1.0229836135347947, "grad_norm": 12.5078125, "learning_rate": 9.676165434906197e-06, "loss": 1.0142, "step": 9614 }, { "epoch": 1.0230900191530112, "grad_norm": 13.2890625, "learning_rate": 9.676129903354179e-06, "loss": 1.2572, "step": 9615 }, { "epoch": 1.023196424771228, "grad_norm": 15.765625, "learning_rate": 9.676094371802162e-06, "loss": 0.9353, "step": 9616 }, { "epoch": 1.0233028303894445, "grad_norm": 13.96875, "learning_rate": 9.676058840250143e-06, "loss": 1.4463, "step": 9617 }, { "epoch": 1.0234092360076612, "grad_norm": 25.765625, "learning_rate": 9.676023308698125e-06, "loss": 1.2308, "step": 9618 }, { "epoch": 1.023515641625878, "grad_norm": 8.765625, "learning_rate": 9.675987777146107e-06, "loss": 1.0679, "step": 9619 }, { "epoch": 1.0236220472440944, "grad_norm": 12.484375, "learning_rate": 9.675952245594088e-06, "loss": 1.0426, "step": 9620 }, { "epoch": 1.0237284528623112, "grad_norm": 10.140625, "learning_rate": 9.67591671404207e-06, "loss": 0.9307, "step": 9621 }, { "epoch": 1.0238348584805277, "grad_norm": 12.3984375, "learning_rate": 9.675881182490051e-06, "loss": 1.1093, "step": 9622 }, { "epoch": 1.0239412640987444, "grad_norm": 12.4609375, "learning_rate": 9.675845650938033e-06, "loss": 1.0589, "step": 9623 }, { "epoch": 1.0240476697169612, "grad_norm": 14.34375, "learning_rate": 9.675810119386015e-06, "loss": 0.9209, "step": 9624 }, { "epoch": 1.0241540753351777, "grad_norm": 13.0234375, "learning_rate": 9.675774587833996e-06, "loss": 1.1006, "step": 9625 }, { "epoch": 1.0242604809533944, "grad_norm": 11.5, "learning_rate": 9.67573905628198e-06, "loss": 1.2169, "step": 9626 }, { "epoch": 1.024366886571611, "grad_norm": 10.984375, "learning_rate": 9.675703524729961e-06, "loss": 1.176, "step": 9627 }, { "epoch": 1.0244732921898276, "grad_norm": 11.5625, "learning_rate": 9.675667993177943e-06, "loss": 1.2655, "step": 9628 }, { "epoch": 1.0245796978080444, "grad_norm": 12.7265625, "learning_rate": 9.675632461625924e-06, "loss": 1.0447, "step": 9629 }, { "epoch": 1.0246861034262609, "grad_norm": 9.8828125, "learning_rate": 9.675596930073906e-06, "loss": 0.6812, "step": 9630 }, { "epoch": 1.0247925090444776, "grad_norm": 12.5546875, "learning_rate": 9.675561398521888e-06, "loss": 1.0707, "step": 9631 }, { "epoch": 1.0248989146626941, "grad_norm": 8.9765625, "learning_rate": 9.675525866969871e-06, "loss": 1.1026, "step": 9632 }, { "epoch": 1.0250053202809108, "grad_norm": 14.5390625, "learning_rate": 9.675490335417853e-06, "loss": 0.8942, "step": 9633 }, { "epoch": 1.0251117258991276, "grad_norm": 10.9609375, "learning_rate": 9.675454803865834e-06, "loss": 0.987, "step": 9634 }, { "epoch": 1.025218131517344, "grad_norm": 9.7421875, "learning_rate": 9.675419272313816e-06, "loss": 1.0445, "step": 9635 }, { "epoch": 1.0253245371355608, "grad_norm": 13.0546875, "learning_rate": 9.675383740761797e-06, "loss": 1.1091, "step": 9636 }, { "epoch": 1.0254309427537773, "grad_norm": 10.9375, "learning_rate": 9.675348209209779e-06, "loss": 1.1194, "step": 9637 }, { "epoch": 1.025537348371994, "grad_norm": 12.265625, "learning_rate": 9.67531267765776e-06, "loss": 0.8888, "step": 9638 }, { "epoch": 1.0256437539902108, "grad_norm": 11.875, "learning_rate": 9.675277146105742e-06, "loss": 1.0608, "step": 9639 }, { "epoch": 1.0257501596084273, "grad_norm": 13.7578125, "learning_rate": 9.675241614553724e-06, "loss": 1.1908, "step": 9640 }, { "epoch": 1.025856565226644, "grad_norm": 12.7734375, "learning_rate": 9.675206083001705e-06, "loss": 1.2025, "step": 9641 }, { "epoch": 1.0259629708448605, "grad_norm": 20.765625, "learning_rate": 9.675170551449689e-06, "loss": 1.0441, "step": 9642 }, { "epoch": 1.0260693764630773, "grad_norm": 12.2265625, "learning_rate": 9.67513501989767e-06, "loss": 0.8946, "step": 9643 }, { "epoch": 1.0261757820812938, "grad_norm": 15.7890625, "learning_rate": 9.675099488345652e-06, "loss": 0.8269, "step": 9644 }, { "epoch": 1.0262821876995105, "grad_norm": 12.453125, "learning_rate": 9.675063956793633e-06, "loss": 0.9628, "step": 9645 }, { "epoch": 1.0263885933177272, "grad_norm": 14.234375, "learning_rate": 9.675028425241615e-06, "loss": 0.9876, "step": 9646 }, { "epoch": 1.0264949989359438, "grad_norm": 13.25, "learning_rate": 9.674992893689598e-06, "loss": 1.0106, "step": 9647 }, { "epoch": 1.0266014045541605, "grad_norm": 20.421875, "learning_rate": 9.67495736213758e-06, "loss": 1.1427, "step": 9648 }, { "epoch": 1.026707810172377, "grad_norm": 15.0703125, "learning_rate": 9.674921830585562e-06, "loss": 0.9209, "step": 9649 }, { "epoch": 1.0268142157905937, "grad_norm": 22.65625, "learning_rate": 9.674886299033543e-06, "loss": 1.0956, "step": 9650 }, { "epoch": 1.0269206214088105, "grad_norm": 18.40625, "learning_rate": 9.674850767481523e-06, "loss": 1.0096, "step": 9651 }, { "epoch": 1.027027027027027, "grad_norm": 11.5625, "learning_rate": 9.674815235929506e-06, "loss": 1.2055, "step": 9652 }, { "epoch": 1.0271334326452437, "grad_norm": 12.125, "learning_rate": 9.674779704377488e-06, "loss": 1.2175, "step": 9653 }, { "epoch": 1.0272398382634602, "grad_norm": 13.234375, "learning_rate": 9.67474417282547e-06, "loss": 1.1036, "step": 9654 }, { "epoch": 1.027346243881677, "grad_norm": 10.8046875, "learning_rate": 9.674708641273451e-06, "loss": 0.8666, "step": 9655 }, { "epoch": 1.0274526494998937, "grad_norm": 14.53125, "learning_rate": 9.674673109721433e-06, "loss": 1.0958, "step": 9656 }, { "epoch": 1.0275590551181102, "grad_norm": 14.78125, "learning_rate": 9.674637578169416e-06, "loss": 1.1396, "step": 9657 }, { "epoch": 1.027665460736327, "grad_norm": 9.7890625, "learning_rate": 9.674602046617398e-06, "loss": 0.9962, "step": 9658 }, { "epoch": 1.0277718663545434, "grad_norm": 13.15625, "learning_rate": 9.67456651506538e-06, "loss": 1.4165, "step": 9659 }, { "epoch": 1.0278782719727602, "grad_norm": 13.5625, "learning_rate": 9.674530983513361e-06, "loss": 1.1896, "step": 9660 }, { "epoch": 1.027984677590977, "grad_norm": 12.671875, "learning_rate": 9.674495451961343e-06, "loss": 1.2934, "step": 9661 }, { "epoch": 1.0280910832091934, "grad_norm": 11.515625, "learning_rate": 9.674459920409324e-06, "loss": 1.1101, "step": 9662 }, { "epoch": 1.0281974888274101, "grad_norm": 11.953125, "learning_rate": 9.674424388857306e-06, "loss": 0.8675, "step": 9663 }, { "epoch": 1.0283038944456266, "grad_norm": 10.8359375, "learning_rate": 9.674388857305287e-06, "loss": 1.1519, "step": 9664 }, { "epoch": 1.0284103000638434, "grad_norm": 10.1328125, "learning_rate": 9.674353325753269e-06, "loss": 1.0768, "step": 9665 }, { "epoch": 1.02851670568206, "grad_norm": 12.6484375, "learning_rate": 9.67431779420125e-06, "loss": 0.998, "step": 9666 }, { "epoch": 1.0286231113002766, "grad_norm": 16.203125, "learning_rate": 9.674282262649232e-06, "loss": 1.1984, "step": 9667 }, { "epoch": 1.0287295169184933, "grad_norm": 11.4609375, "learning_rate": 9.674246731097216e-06, "loss": 0.9857, "step": 9668 }, { "epoch": 1.0288359225367099, "grad_norm": 12.5078125, "learning_rate": 9.674211199545197e-06, "loss": 1.2208, "step": 9669 }, { "epoch": 1.0289423281549266, "grad_norm": 14.3359375, "learning_rate": 9.674175667993179e-06, "loss": 1.2805, "step": 9670 }, { "epoch": 1.0290487337731433, "grad_norm": 11.390625, "learning_rate": 9.67414013644116e-06, "loss": 0.9197, "step": 9671 }, { "epoch": 1.0291551393913598, "grad_norm": 10.9375, "learning_rate": 9.674104604889142e-06, "loss": 1.1034, "step": 9672 }, { "epoch": 1.0292615450095766, "grad_norm": 13.1171875, "learning_rate": 9.674069073337125e-06, "loss": 1.0325, "step": 9673 }, { "epoch": 1.029367950627793, "grad_norm": 16.34375, "learning_rate": 9.674033541785107e-06, "loss": 1.049, "step": 9674 }, { "epoch": 1.0294743562460098, "grad_norm": 13.8671875, "learning_rate": 9.673998010233088e-06, "loss": 1.0128, "step": 9675 }, { "epoch": 1.0295807618642265, "grad_norm": 12.453125, "learning_rate": 9.67396247868107e-06, "loss": 0.9643, "step": 9676 }, { "epoch": 1.029687167482443, "grad_norm": 14.21875, "learning_rate": 9.673926947129052e-06, "loss": 1.2836, "step": 9677 }, { "epoch": 1.0297935731006598, "grad_norm": 10.8828125, "learning_rate": 9.673891415577033e-06, "loss": 1.0577, "step": 9678 }, { "epoch": 1.0298999787188763, "grad_norm": 14.09375, "learning_rate": 9.673855884025015e-06, "loss": 0.8952, "step": 9679 }, { "epoch": 1.030006384337093, "grad_norm": 12.4296875, "learning_rate": 9.673820352472996e-06, "loss": 1.3641, "step": 9680 }, { "epoch": 1.0301127899553097, "grad_norm": 12.5859375, "learning_rate": 9.673784820920978e-06, "loss": 1.1837, "step": 9681 }, { "epoch": 1.0302191955735263, "grad_norm": 14.7890625, "learning_rate": 9.67374928936896e-06, "loss": 0.8503, "step": 9682 }, { "epoch": 1.030325601191743, "grad_norm": 9.7890625, "learning_rate": 9.673713757816943e-06, "loss": 0.7993, "step": 9683 }, { "epoch": 1.0304320068099595, "grad_norm": 12.21875, "learning_rate": 9.673678226264925e-06, "loss": 1.2556, "step": 9684 }, { "epoch": 1.0305384124281762, "grad_norm": 11.1015625, "learning_rate": 9.673642694712906e-06, "loss": 0.9728, "step": 9685 }, { "epoch": 1.0306448180463927, "grad_norm": 15.15625, "learning_rate": 9.673607163160888e-06, "loss": 0.9098, "step": 9686 }, { "epoch": 1.0307512236646095, "grad_norm": 14.3203125, "learning_rate": 9.67357163160887e-06, "loss": 1.0682, "step": 9687 }, { "epoch": 1.0308576292828262, "grad_norm": 14.0390625, "learning_rate": 9.673536100056851e-06, "loss": 1.0244, "step": 9688 }, { "epoch": 1.0309640349010427, "grad_norm": 12.7890625, "learning_rate": 9.673500568504833e-06, "loss": 1.0149, "step": 9689 }, { "epoch": 1.0310704405192594, "grad_norm": 12.0703125, "learning_rate": 9.673465036952814e-06, "loss": 1.2206, "step": 9690 }, { "epoch": 1.031176846137476, "grad_norm": 10.65625, "learning_rate": 9.673429505400796e-06, "loss": 1.1222, "step": 9691 }, { "epoch": 1.0312832517556927, "grad_norm": 12.75, "learning_rate": 9.673393973848777e-06, "loss": 1.3203, "step": 9692 }, { "epoch": 1.0313896573739094, "grad_norm": 19.34375, "learning_rate": 9.67335844229676e-06, "loss": 0.9328, "step": 9693 }, { "epoch": 1.031496062992126, "grad_norm": 17.46875, "learning_rate": 9.673322910744742e-06, "loss": 1.1982, "step": 9694 }, { "epoch": 1.0316024686103427, "grad_norm": 13.203125, "learning_rate": 9.673287379192724e-06, "loss": 1.0147, "step": 9695 }, { "epoch": 1.0317088742285592, "grad_norm": 11.7265625, "learning_rate": 9.673251847640706e-06, "loss": 1.1176, "step": 9696 }, { "epoch": 1.031815279846776, "grad_norm": 13.9609375, "learning_rate": 9.673216316088687e-06, "loss": 1.0312, "step": 9697 }, { "epoch": 1.0319216854649926, "grad_norm": 11.609375, "learning_rate": 9.67318078453667e-06, "loss": 1.0549, "step": 9698 }, { "epoch": 1.0320280910832091, "grad_norm": 11.859375, "learning_rate": 9.673145252984652e-06, "loss": 1.0258, "step": 9699 }, { "epoch": 1.0321344967014259, "grad_norm": 14.3203125, "learning_rate": 9.673109721432634e-06, "loss": 1.1758, "step": 9700 }, { "epoch": 1.0322409023196424, "grad_norm": 11.8828125, "learning_rate": 9.673074189880615e-06, "loss": 1.0257, "step": 9701 }, { "epoch": 1.0323473079378591, "grad_norm": 11.3515625, "learning_rate": 9.673038658328597e-06, "loss": 0.8242, "step": 9702 }, { "epoch": 1.0324537135560758, "grad_norm": 12.109375, "learning_rate": 9.673003126776579e-06, "loss": 1.091, "step": 9703 }, { "epoch": 1.0325601191742924, "grad_norm": 12.7578125, "learning_rate": 9.67296759522456e-06, "loss": 1.1343, "step": 9704 }, { "epoch": 1.032666524792509, "grad_norm": 12.6328125, "learning_rate": 9.672932063672542e-06, "loss": 1.2757, "step": 9705 }, { "epoch": 1.0327729304107256, "grad_norm": 14.5390625, "learning_rate": 9.672896532120523e-06, "loss": 1.0221, "step": 9706 }, { "epoch": 1.0328793360289423, "grad_norm": 10.234375, "learning_rate": 9.672861000568505e-06, "loss": 1.0304, "step": 9707 }, { "epoch": 1.032985741647159, "grad_norm": 12.6796875, "learning_rate": 9.672825469016487e-06, "loss": 1.3886, "step": 9708 }, { "epoch": 1.0330921472653756, "grad_norm": 13.046875, "learning_rate": 9.67278993746447e-06, "loss": 0.8997, "step": 9709 }, { "epoch": 1.0331985528835923, "grad_norm": 13.0546875, "learning_rate": 9.672754405912451e-06, "loss": 1.2123, "step": 9710 }, { "epoch": 1.0333049585018088, "grad_norm": 14.7890625, "learning_rate": 9.672718874360433e-06, "loss": 0.9275, "step": 9711 }, { "epoch": 1.0334113641200255, "grad_norm": 10.5, "learning_rate": 9.672683342808415e-06, "loss": 0.9674, "step": 9712 }, { "epoch": 1.0335177697382423, "grad_norm": 11.03125, "learning_rate": 9.672647811256396e-06, "loss": 1.163, "step": 9713 }, { "epoch": 1.0336241753564588, "grad_norm": 10.40625, "learning_rate": 9.672612279704378e-06, "loss": 1.1784, "step": 9714 }, { "epoch": 1.0337305809746755, "grad_norm": 18.28125, "learning_rate": 9.67257674815236e-06, "loss": 1.2856, "step": 9715 }, { "epoch": 1.033836986592892, "grad_norm": 13.8125, "learning_rate": 9.672541216600341e-06, "loss": 1.0839, "step": 9716 }, { "epoch": 1.0339433922111088, "grad_norm": 19.90625, "learning_rate": 9.672505685048323e-06, "loss": 1.2528, "step": 9717 }, { "epoch": 1.0340497978293255, "grad_norm": 12.109375, "learning_rate": 9.672470153496304e-06, "loss": 0.9293, "step": 9718 }, { "epoch": 1.034156203447542, "grad_norm": 18.75, "learning_rate": 9.672434621944288e-06, "loss": 1.0221, "step": 9719 }, { "epoch": 1.0342626090657587, "grad_norm": 12.203125, "learning_rate": 9.67239909039227e-06, "loss": 1.2504, "step": 9720 }, { "epoch": 1.0343690146839752, "grad_norm": 15.5, "learning_rate": 9.67236355884025e-06, "loss": 0.8539, "step": 9721 }, { "epoch": 1.034475420302192, "grad_norm": 13.265625, "learning_rate": 9.672328027288232e-06, "loss": 0.7703, "step": 9722 }, { "epoch": 1.0345818259204087, "grad_norm": 10.5390625, "learning_rate": 9.672292495736214e-06, "loss": 1.3599, "step": 9723 }, { "epoch": 1.0346882315386252, "grad_norm": 13.625, "learning_rate": 9.672256964184197e-06, "loss": 0.9766, "step": 9724 }, { "epoch": 1.034794637156842, "grad_norm": 17.265625, "learning_rate": 9.672221432632179e-06, "loss": 1.2402, "step": 9725 }, { "epoch": 1.0349010427750585, "grad_norm": 11.4140625, "learning_rate": 9.67218590108016e-06, "loss": 0.8851, "step": 9726 }, { "epoch": 1.0350074483932752, "grad_norm": 10.5625, "learning_rate": 9.672150369528142e-06, "loss": 1.4328, "step": 9727 }, { "epoch": 1.0351138540114917, "grad_norm": 14.2890625, "learning_rate": 9.672114837976124e-06, "loss": 1.36, "step": 9728 }, { "epoch": 1.0352202596297084, "grad_norm": 13.8515625, "learning_rate": 9.672079306424105e-06, "loss": 1.091, "step": 9729 }, { "epoch": 1.0353266652479252, "grad_norm": 12.4609375, "learning_rate": 9.672043774872087e-06, "loss": 0.8304, "step": 9730 }, { "epoch": 1.0354330708661417, "grad_norm": 9.328125, "learning_rate": 9.672008243320069e-06, "loss": 1.0454, "step": 9731 }, { "epoch": 1.0355394764843584, "grad_norm": 13.046875, "learning_rate": 9.67197271176805e-06, "loss": 1.3824, "step": 9732 }, { "epoch": 1.035645882102575, "grad_norm": 12.109375, "learning_rate": 9.671937180216032e-06, "loss": 0.7595, "step": 9733 }, { "epoch": 1.0357522877207916, "grad_norm": 9.765625, "learning_rate": 9.671901648664015e-06, "loss": 0.9407, "step": 9734 }, { "epoch": 1.0358586933390084, "grad_norm": 14.5234375, "learning_rate": 9.671866117111997e-06, "loss": 0.9685, "step": 9735 }, { "epoch": 1.0359650989572249, "grad_norm": 11.828125, "learning_rate": 9.671830585559978e-06, "loss": 1.0148, "step": 9736 }, { "epoch": 1.0360715045754416, "grad_norm": 9.78125, "learning_rate": 9.67179505400796e-06, "loss": 1.0857, "step": 9737 }, { "epoch": 1.0361779101936581, "grad_norm": 18.640625, "learning_rate": 9.671759522455942e-06, "loss": 1.0464, "step": 9738 }, { "epoch": 1.0362843158118749, "grad_norm": 18.84375, "learning_rate": 9.671723990903925e-06, "loss": 0.9787, "step": 9739 }, { "epoch": 1.0363907214300916, "grad_norm": 18.640625, "learning_rate": 9.671688459351905e-06, "loss": 1.131, "step": 9740 }, { "epoch": 1.036497127048308, "grad_norm": 10.3515625, "learning_rate": 9.671652927799886e-06, "loss": 0.88, "step": 9741 }, { "epoch": 1.0366035326665248, "grad_norm": 14.4140625, "learning_rate": 9.671617396247868e-06, "loss": 1.3317, "step": 9742 }, { "epoch": 1.0367099382847413, "grad_norm": 14.859375, "learning_rate": 9.67158186469585e-06, "loss": 1.0251, "step": 9743 }, { "epoch": 1.036816343902958, "grad_norm": 10.4609375, "learning_rate": 9.671546333143833e-06, "loss": 0.8711, "step": 9744 }, { "epoch": 1.0369227495211748, "grad_norm": 11.8984375, "learning_rate": 9.671510801591814e-06, "loss": 0.863, "step": 9745 }, { "epoch": 1.0370291551393913, "grad_norm": 15.9296875, "learning_rate": 9.671475270039796e-06, "loss": 1.0894, "step": 9746 }, { "epoch": 1.037135560757608, "grad_norm": 11.6640625, "learning_rate": 9.671439738487778e-06, "loss": 1.1573, "step": 9747 }, { "epoch": 1.0372419663758246, "grad_norm": 10.5546875, "learning_rate": 9.67140420693576e-06, "loss": 0.7847, "step": 9748 }, { "epoch": 1.0373483719940413, "grad_norm": 12.46875, "learning_rate": 9.671368675383741e-06, "loss": 0.9715, "step": 9749 }, { "epoch": 1.037454777612258, "grad_norm": 10.8125, "learning_rate": 9.671333143831724e-06, "loss": 1.0325, "step": 9750 }, { "epoch": 1.0375611832304745, "grad_norm": 17.671875, "learning_rate": 9.671297612279706e-06, "loss": 1.1197, "step": 9751 }, { "epoch": 1.0376675888486913, "grad_norm": 11.078125, "learning_rate": 9.671262080727687e-06, "loss": 1.0675, "step": 9752 }, { "epoch": 1.0377739944669078, "grad_norm": 14.3984375, "learning_rate": 9.671226549175669e-06, "loss": 1.2946, "step": 9753 }, { "epoch": 1.0378804000851245, "grad_norm": 10.28125, "learning_rate": 9.67119101762365e-06, "loss": 0.8196, "step": 9754 }, { "epoch": 1.0379868057033412, "grad_norm": 11.59375, "learning_rate": 9.671155486071632e-06, "loss": 1.0962, "step": 9755 }, { "epoch": 1.0380932113215577, "grad_norm": 12.0234375, "learning_rate": 9.671119954519614e-06, "loss": 1.0363, "step": 9756 }, { "epoch": 1.0381996169397745, "grad_norm": 17.078125, "learning_rate": 9.671084422967595e-06, "loss": 1.3168, "step": 9757 }, { "epoch": 1.038306022557991, "grad_norm": 11.4375, "learning_rate": 9.671048891415577e-06, "loss": 1.2004, "step": 9758 }, { "epoch": 1.0384124281762077, "grad_norm": 10.78125, "learning_rate": 9.671013359863559e-06, "loss": 1.2463, "step": 9759 }, { "epoch": 1.0385188337944244, "grad_norm": 12.0625, "learning_rate": 9.670977828311542e-06, "loss": 1.0001, "step": 9760 }, { "epoch": 1.038625239412641, "grad_norm": 10.90625, "learning_rate": 9.670942296759524e-06, "loss": 1.3214, "step": 9761 }, { "epoch": 1.0387316450308577, "grad_norm": 15.1640625, "learning_rate": 9.670906765207505e-06, "loss": 1.0942, "step": 9762 }, { "epoch": 1.0388380506490742, "grad_norm": 8.2578125, "learning_rate": 9.670871233655487e-06, "loss": 1.0011, "step": 9763 }, { "epoch": 1.038944456267291, "grad_norm": 13.265625, "learning_rate": 9.670835702103468e-06, "loss": 0.8624, "step": 9764 }, { "epoch": 1.0390508618855077, "grad_norm": 11.734375, "learning_rate": 9.670800170551452e-06, "loss": 0.9876, "step": 9765 }, { "epoch": 1.0391572675037242, "grad_norm": 12.0234375, "learning_rate": 9.670764638999433e-06, "loss": 0.9227, "step": 9766 }, { "epoch": 1.039263673121941, "grad_norm": 11.25, "learning_rate": 9.670729107447415e-06, "loss": 1.1498, "step": 9767 }, { "epoch": 1.0393700787401574, "grad_norm": 15.40625, "learning_rate": 9.670693575895397e-06, "loss": 1.3108, "step": 9768 }, { "epoch": 1.0394764843583741, "grad_norm": 13.9453125, "learning_rate": 9.670658044343378e-06, "loss": 0.9525, "step": 9769 }, { "epoch": 1.0395828899765909, "grad_norm": 10.296875, "learning_rate": 9.67062251279136e-06, "loss": 1.1425, "step": 9770 }, { "epoch": 1.0396892955948074, "grad_norm": 14.5390625, "learning_rate": 9.670586981239341e-06, "loss": 0.9259, "step": 9771 }, { "epoch": 1.0397957012130241, "grad_norm": 11.2421875, "learning_rate": 9.670551449687323e-06, "loss": 1.0825, "step": 9772 }, { "epoch": 1.0399021068312406, "grad_norm": 10.515625, "learning_rate": 9.670515918135305e-06, "loss": 0.967, "step": 9773 }, { "epoch": 1.0400085124494574, "grad_norm": 12.84375, "learning_rate": 9.670480386583286e-06, "loss": 1.3772, "step": 9774 }, { "epoch": 1.0401149180676739, "grad_norm": 13.4453125, "learning_rate": 9.67044485503127e-06, "loss": 1.1273, "step": 9775 }, { "epoch": 1.0402213236858906, "grad_norm": 13.21875, "learning_rate": 9.670409323479251e-06, "loss": 1.1393, "step": 9776 }, { "epoch": 1.0403277293041073, "grad_norm": 13.2890625, "learning_rate": 9.670373791927233e-06, "loss": 1.1066, "step": 9777 }, { "epoch": 1.0404341349223238, "grad_norm": 13.953125, "learning_rate": 9.670338260375214e-06, "loss": 1.0686, "step": 9778 }, { "epoch": 1.0405405405405406, "grad_norm": 9.953125, "learning_rate": 9.670302728823196e-06, "loss": 1.1067, "step": 9779 }, { "epoch": 1.040646946158757, "grad_norm": 19.734375, "learning_rate": 9.670267197271177e-06, "loss": 1.1995, "step": 9780 }, { "epoch": 1.0407533517769738, "grad_norm": 12.453125, "learning_rate": 9.670231665719159e-06, "loss": 1.2163, "step": 9781 }, { "epoch": 1.0408597573951905, "grad_norm": 17.421875, "learning_rate": 9.67019613416714e-06, "loss": 1.1527, "step": 9782 }, { "epoch": 1.040966163013407, "grad_norm": 10.3671875, "learning_rate": 9.670160602615122e-06, "loss": 0.8743, "step": 9783 }, { "epoch": 1.0410725686316238, "grad_norm": 12.421875, "learning_rate": 9.670125071063104e-06, "loss": 1.1325, "step": 9784 }, { "epoch": 1.0411789742498403, "grad_norm": 13.5703125, "learning_rate": 9.670089539511086e-06, "loss": 1.006, "step": 9785 }, { "epoch": 1.041285379868057, "grad_norm": 11.015625, "learning_rate": 9.670054007959069e-06, "loss": 1.1048, "step": 9786 }, { "epoch": 1.0413917854862738, "grad_norm": 9.9453125, "learning_rate": 9.67001847640705e-06, "loss": 1.0225, "step": 9787 }, { "epoch": 1.0414981911044903, "grad_norm": 15.53125, "learning_rate": 9.669982944855032e-06, "loss": 1.122, "step": 9788 }, { "epoch": 1.041604596722707, "grad_norm": 10.609375, "learning_rate": 9.669947413303014e-06, "loss": 0.7845, "step": 9789 }, { "epoch": 1.0417110023409235, "grad_norm": 13.5234375, "learning_rate": 9.669911881750995e-06, "loss": 0.7622, "step": 9790 }, { "epoch": 1.0418174079591402, "grad_norm": 12.296875, "learning_rate": 9.669876350198979e-06, "loss": 0.9576, "step": 9791 }, { "epoch": 1.041923813577357, "grad_norm": 13.2421875, "learning_rate": 9.66984081864696e-06, "loss": 1.1026, "step": 9792 }, { "epoch": 1.0420302191955735, "grad_norm": 9.6953125, "learning_rate": 9.669805287094942e-06, "loss": 0.9368, "step": 9793 }, { "epoch": 1.0421366248137902, "grad_norm": 16.828125, "learning_rate": 9.669769755542923e-06, "loss": 0.8973, "step": 9794 }, { "epoch": 1.0422430304320067, "grad_norm": 12.4921875, "learning_rate": 9.669734223990905e-06, "loss": 1.1458, "step": 9795 }, { "epoch": 1.0423494360502235, "grad_norm": 15.0546875, "learning_rate": 9.669698692438887e-06, "loss": 1.4102, "step": 9796 }, { "epoch": 1.0424558416684402, "grad_norm": 18.8125, "learning_rate": 9.669663160886868e-06, "loss": 1.2352, "step": 9797 }, { "epoch": 1.0425622472866567, "grad_norm": 13.9140625, "learning_rate": 9.66962762933485e-06, "loss": 1.1408, "step": 9798 }, { "epoch": 1.0426686529048734, "grad_norm": 12.1640625, "learning_rate": 9.669592097782831e-06, "loss": 1.0542, "step": 9799 }, { "epoch": 1.04277505852309, "grad_norm": 19.234375, "learning_rate": 9.669556566230813e-06, "loss": 1.1783, "step": 9800 }, { "epoch": 1.0428814641413067, "grad_norm": 9.53125, "learning_rate": 9.669521034678796e-06, "loss": 1.0332, "step": 9801 }, { "epoch": 1.0429878697595234, "grad_norm": 14.3359375, "learning_rate": 9.669485503126778e-06, "loss": 1.322, "step": 9802 }, { "epoch": 1.04309427537774, "grad_norm": 12.4375, "learning_rate": 9.66944997157476e-06, "loss": 1.1525, "step": 9803 }, { "epoch": 1.0432006809959566, "grad_norm": 14.3515625, "learning_rate": 9.669414440022741e-06, "loss": 1.1091, "step": 9804 }, { "epoch": 1.0433070866141732, "grad_norm": 11.53125, "learning_rate": 9.669378908470723e-06, "loss": 1.324, "step": 9805 }, { "epoch": 1.0434134922323899, "grad_norm": 10.9609375, "learning_rate": 9.669343376918704e-06, "loss": 0.9803, "step": 9806 }, { "epoch": 1.0435198978506066, "grad_norm": 9.890625, "learning_rate": 9.669307845366686e-06, "loss": 0.9962, "step": 9807 }, { "epoch": 1.0436263034688231, "grad_norm": 15.453125, "learning_rate": 9.669272313814668e-06, "loss": 1.1078, "step": 9808 }, { "epoch": 1.0437327090870399, "grad_norm": 16.765625, "learning_rate": 9.66923678226265e-06, "loss": 0.7922, "step": 9809 }, { "epoch": 1.0438391147052564, "grad_norm": 9.8359375, "learning_rate": 9.66920125071063e-06, "loss": 1.2656, "step": 9810 }, { "epoch": 1.043945520323473, "grad_norm": 10.078125, "learning_rate": 9.669165719158614e-06, "loss": 1.0897, "step": 9811 }, { "epoch": 1.0440519259416896, "grad_norm": 13.8828125, "learning_rate": 9.669130187606596e-06, "loss": 1.2114, "step": 9812 }, { "epoch": 1.0441583315599063, "grad_norm": 13.1796875, "learning_rate": 9.669094656054577e-06, "loss": 1.3065, "step": 9813 }, { "epoch": 1.044264737178123, "grad_norm": 10.2890625, "learning_rate": 9.669059124502559e-06, "loss": 0.9642, "step": 9814 }, { "epoch": 1.0443711427963396, "grad_norm": 13.078125, "learning_rate": 9.66902359295054e-06, "loss": 1.0802, "step": 9815 }, { "epoch": 1.0444775484145563, "grad_norm": 15.7578125, "learning_rate": 9.668988061398524e-06, "loss": 1.2037, "step": 9816 }, { "epoch": 1.044583954032773, "grad_norm": 18.0625, "learning_rate": 9.668952529846505e-06, "loss": 1.114, "step": 9817 }, { "epoch": 1.0446903596509896, "grad_norm": 12.484375, "learning_rate": 9.668916998294487e-06, "loss": 1.1581, "step": 9818 }, { "epoch": 1.0447967652692063, "grad_norm": 27.921875, "learning_rate": 9.668881466742469e-06, "loss": 1.1607, "step": 9819 }, { "epoch": 1.0449031708874228, "grad_norm": 11.4765625, "learning_rate": 9.66884593519045e-06, "loss": 1.1693, "step": 9820 }, { "epoch": 1.0450095765056395, "grad_norm": 12.4453125, "learning_rate": 9.668810403638432e-06, "loss": 0.9868, "step": 9821 }, { "epoch": 1.045115982123856, "grad_norm": 14.984375, "learning_rate": 9.668774872086413e-06, "loss": 0.9022, "step": 9822 }, { "epoch": 1.0452223877420728, "grad_norm": 12.125, "learning_rate": 9.668739340534395e-06, "loss": 0.9978, "step": 9823 }, { "epoch": 1.0453287933602895, "grad_norm": 11.7421875, "learning_rate": 9.668703808982377e-06, "loss": 1.0597, "step": 9824 }, { "epoch": 1.045435198978506, "grad_norm": 17.671875, "learning_rate": 9.668668277430358e-06, "loss": 1.0327, "step": 9825 }, { "epoch": 1.0455416045967227, "grad_norm": 17.421875, "learning_rate": 9.66863274587834e-06, "loss": 1.335, "step": 9826 }, { "epoch": 1.0456480102149392, "grad_norm": 16.046875, "learning_rate": 9.668597214326323e-06, "loss": 0.8514, "step": 9827 }, { "epoch": 1.045754415833156, "grad_norm": 10.9453125, "learning_rate": 9.668561682774305e-06, "loss": 0.9455, "step": 9828 }, { "epoch": 1.0458608214513727, "grad_norm": 12.9375, "learning_rate": 9.668526151222286e-06, "loss": 0.954, "step": 9829 }, { "epoch": 1.0459672270695892, "grad_norm": 14.1328125, "learning_rate": 9.668490619670268e-06, "loss": 1.0754, "step": 9830 }, { "epoch": 1.046073632687806, "grad_norm": 9.9921875, "learning_rate": 9.66845508811825e-06, "loss": 0.9912, "step": 9831 }, { "epoch": 1.0461800383060225, "grad_norm": 11.78125, "learning_rate": 9.668419556566231e-06, "loss": 1.2247, "step": 9832 }, { "epoch": 1.0462864439242392, "grad_norm": 12.828125, "learning_rate": 9.668384025014213e-06, "loss": 1.111, "step": 9833 }, { "epoch": 1.046392849542456, "grad_norm": 9.546875, "learning_rate": 9.668348493462194e-06, "loss": 1.0355, "step": 9834 }, { "epoch": 1.0464992551606724, "grad_norm": 12.6640625, "learning_rate": 9.668312961910176e-06, "loss": 1.2151, "step": 9835 }, { "epoch": 1.0466056607788892, "grad_norm": 11.484375, "learning_rate": 9.668277430358158e-06, "loss": 0.7018, "step": 9836 }, { "epoch": 1.0467120663971057, "grad_norm": 10.9921875, "learning_rate": 9.668241898806141e-06, "loss": 0.9825, "step": 9837 }, { "epoch": 1.0468184720153224, "grad_norm": 16.59375, "learning_rate": 9.668206367254123e-06, "loss": 1.1485, "step": 9838 }, { "epoch": 1.0469248776335391, "grad_norm": 14.46875, "learning_rate": 9.668170835702104e-06, "loss": 1.2139, "step": 9839 }, { "epoch": 1.0470312832517557, "grad_norm": 15.2265625, "learning_rate": 9.668135304150086e-06, "loss": 1.0592, "step": 9840 }, { "epoch": 1.0471376888699724, "grad_norm": 13.09375, "learning_rate": 9.668099772598067e-06, "loss": 0.9619, "step": 9841 }, { "epoch": 1.047244094488189, "grad_norm": 18.171875, "learning_rate": 9.66806424104605e-06, "loss": 1.3962, "step": 9842 }, { "epoch": 1.0473505001064056, "grad_norm": 9.8203125, "learning_rate": 9.668028709494032e-06, "loss": 0.9625, "step": 9843 }, { "epoch": 1.0474569057246224, "grad_norm": 10.359375, "learning_rate": 9.667993177942014e-06, "loss": 0.9926, "step": 9844 }, { "epoch": 1.0475633113428389, "grad_norm": 13.078125, "learning_rate": 9.667957646389995e-06, "loss": 1.0899, "step": 9845 }, { "epoch": 1.0476697169610556, "grad_norm": 11.921875, "learning_rate": 9.667922114837977e-06, "loss": 1.0324, "step": 9846 }, { "epoch": 1.047776122579272, "grad_norm": 10.953125, "learning_rate": 9.667886583285959e-06, "loss": 1.2757, "step": 9847 }, { "epoch": 1.0478825281974888, "grad_norm": 14.515625, "learning_rate": 9.66785105173394e-06, "loss": 1.1222, "step": 9848 }, { "epoch": 1.0479889338157056, "grad_norm": 13.484375, "learning_rate": 9.667815520181922e-06, "loss": 1.2172, "step": 9849 }, { "epoch": 1.048095339433922, "grad_norm": 13.5859375, "learning_rate": 9.667779988629904e-06, "loss": 1.1273, "step": 9850 }, { "epoch": 1.0482017450521388, "grad_norm": 7.703125, "learning_rate": 9.667744457077885e-06, "loss": 0.6714, "step": 9851 }, { "epoch": 1.0483081506703553, "grad_norm": 15.6015625, "learning_rate": 9.667708925525868e-06, "loss": 1.2547, "step": 9852 }, { "epoch": 1.048414556288572, "grad_norm": 12.671875, "learning_rate": 9.66767339397385e-06, "loss": 0.9643, "step": 9853 }, { "epoch": 1.0485209619067888, "grad_norm": 11.6875, "learning_rate": 9.667637862421832e-06, "loss": 1.1879, "step": 9854 }, { "epoch": 1.0486273675250053, "grad_norm": 13.3515625, "learning_rate": 9.667602330869813e-06, "loss": 1.4049, "step": 9855 }, { "epoch": 1.048733773143222, "grad_norm": 19.28125, "learning_rate": 9.667566799317795e-06, "loss": 1.1083, "step": 9856 }, { "epoch": 1.0488401787614385, "grad_norm": 12.9140625, "learning_rate": 9.667531267765778e-06, "loss": 0.9064, "step": 9857 }, { "epoch": 1.0489465843796553, "grad_norm": 12.0625, "learning_rate": 9.66749573621376e-06, "loss": 0.9246, "step": 9858 }, { "epoch": 1.0490529899978718, "grad_norm": 11.046875, "learning_rate": 9.667460204661741e-06, "loss": 1.3569, "step": 9859 }, { "epoch": 1.0491593956160885, "grad_norm": 11.0390625, "learning_rate": 9.667424673109721e-06, "loss": 0.9852, "step": 9860 }, { "epoch": 1.0492658012343052, "grad_norm": 17.859375, "learning_rate": 9.667389141557703e-06, "loss": 0.9731, "step": 9861 }, { "epoch": 1.0493722068525217, "grad_norm": 13.9140625, "learning_rate": 9.667353610005684e-06, "loss": 0.9031, "step": 9862 }, { "epoch": 1.0494786124707385, "grad_norm": 12.21875, "learning_rate": 9.667318078453668e-06, "loss": 1.2255, "step": 9863 }, { "epoch": 1.049585018088955, "grad_norm": 18.421875, "learning_rate": 9.66728254690165e-06, "loss": 1.0461, "step": 9864 }, { "epoch": 1.0496914237071717, "grad_norm": 14.2265625, "learning_rate": 9.667247015349631e-06, "loss": 0.8834, "step": 9865 }, { "epoch": 1.0497978293253885, "grad_norm": 14.0390625, "learning_rate": 9.667211483797613e-06, "loss": 0.9441, "step": 9866 }, { "epoch": 1.049904234943605, "grad_norm": 11.3671875, "learning_rate": 9.667175952245594e-06, "loss": 1.2835, "step": 9867 }, { "epoch": 1.0500106405618217, "grad_norm": 14.5625, "learning_rate": 9.667140420693578e-06, "loss": 1.2071, "step": 9868 }, { "epoch": 1.0501170461800382, "grad_norm": 16.46875, "learning_rate": 9.667104889141559e-06, "loss": 0.9611, "step": 9869 }, { "epoch": 1.050223451798255, "grad_norm": 16.984375, "learning_rate": 9.66706935758954e-06, "loss": 0.9022, "step": 9870 }, { "epoch": 1.0503298574164717, "grad_norm": 17.109375, "learning_rate": 9.667033826037522e-06, "loss": 0.925, "step": 9871 }, { "epoch": 1.0504362630346882, "grad_norm": 15.625, "learning_rate": 9.666998294485504e-06, "loss": 0.9561, "step": 9872 }, { "epoch": 1.050542668652905, "grad_norm": 11.953125, "learning_rate": 9.666962762933486e-06, "loss": 1.1307, "step": 9873 }, { "epoch": 1.0506490742711214, "grad_norm": 17.03125, "learning_rate": 9.666927231381467e-06, "loss": 1.1661, "step": 9874 }, { "epoch": 1.0507554798893382, "grad_norm": 14.4453125, "learning_rate": 9.666891699829449e-06, "loss": 1.2749, "step": 9875 }, { "epoch": 1.0508618855075549, "grad_norm": 14.3046875, "learning_rate": 9.66685616827743e-06, "loss": 0.9425, "step": 9876 }, { "epoch": 1.0509682911257714, "grad_norm": 11.9609375, "learning_rate": 9.666820636725412e-06, "loss": 1.4023, "step": 9877 }, { "epoch": 1.0510746967439881, "grad_norm": 12.265625, "learning_rate": 9.666785105173395e-06, "loss": 0.8735, "step": 9878 }, { "epoch": 1.0511811023622046, "grad_norm": 12.03125, "learning_rate": 9.666749573621377e-06, "loss": 1.0905, "step": 9879 }, { "epoch": 1.0512875079804214, "grad_norm": 12.40625, "learning_rate": 9.666714042069358e-06, "loss": 1.1511, "step": 9880 }, { "epoch": 1.051393913598638, "grad_norm": 13.6875, "learning_rate": 9.66667851051734e-06, "loss": 1.1501, "step": 9881 }, { "epoch": 1.0515003192168546, "grad_norm": 16.625, "learning_rate": 9.666642978965322e-06, "loss": 1.0246, "step": 9882 }, { "epoch": 1.0516067248350713, "grad_norm": 11.203125, "learning_rate": 9.666607447413305e-06, "loss": 0.8765, "step": 9883 }, { "epoch": 1.0517131304532878, "grad_norm": 21.375, "learning_rate": 9.666571915861287e-06, "loss": 0.9555, "step": 9884 }, { "epoch": 1.0518195360715046, "grad_norm": 11.9921875, "learning_rate": 9.666536384309268e-06, "loss": 1.0915, "step": 9885 }, { "epoch": 1.0519259416897213, "grad_norm": 9.5546875, "learning_rate": 9.66650085275725e-06, "loss": 1.1373, "step": 9886 }, { "epoch": 1.0520323473079378, "grad_norm": 11.1875, "learning_rate": 9.666465321205231e-06, "loss": 0.8245, "step": 9887 }, { "epoch": 1.0521387529261546, "grad_norm": 9.7421875, "learning_rate": 9.666429789653213e-06, "loss": 1.2741, "step": 9888 }, { "epoch": 1.052245158544371, "grad_norm": 13.25, "learning_rate": 9.666394258101195e-06, "loss": 0.8388, "step": 9889 }, { "epoch": 1.0523515641625878, "grad_norm": 13.09375, "learning_rate": 9.666358726549176e-06, "loss": 1.0883, "step": 9890 }, { "epoch": 1.0524579697808045, "grad_norm": 13.0078125, "learning_rate": 9.666323194997158e-06, "loss": 1.1406, "step": 9891 }, { "epoch": 1.052564375399021, "grad_norm": 11.4765625, "learning_rate": 9.66628766344514e-06, "loss": 0.9157, "step": 9892 }, { "epoch": 1.0526707810172378, "grad_norm": 15.625, "learning_rate": 9.666252131893123e-06, "loss": 1.0803, "step": 9893 }, { "epoch": 1.0527771866354543, "grad_norm": 11.2421875, "learning_rate": 9.666216600341104e-06, "loss": 0.6437, "step": 9894 }, { "epoch": 1.052883592253671, "grad_norm": 9.5, "learning_rate": 9.666181068789086e-06, "loss": 1.0918, "step": 9895 }, { "epoch": 1.0529899978718877, "grad_norm": 13.1328125, "learning_rate": 9.666145537237068e-06, "loss": 1.1695, "step": 9896 }, { "epoch": 1.0530964034901042, "grad_norm": 15.1171875, "learning_rate": 9.66611000568505e-06, "loss": 1.0944, "step": 9897 }, { "epoch": 1.053202809108321, "grad_norm": 14.0859375, "learning_rate": 9.66607447413303e-06, "loss": 1.1729, "step": 9898 }, { "epoch": 1.0533092147265375, "grad_norm": 14.2578125, "learning_rate": 9.666038942581012e-06, "loss": 1.1427, "step": 9899 }, { "epoch": 1.0534156203447542, "grad_norm": 10.59375, "learning_rate": 9.666003411028994e-06, "loss": 1.1923, "step": 9900 }, { "epoch": 1.053522025962971, "grad_norm": 11.3671875, "learning_rate": 9.665967879476976e-06, "loss": 1.2083, "step": 9901 }, { "epoch": 1.0536284315811875, "grad_norm": 14.359375, "learning_rate": 9.665932347924957e-06, "loss": 0.9471, "step": 9902 }, { "epoch": 1.0537348371994042, "grad_norm": 10.6171875, "learning_rate": 9.665896816372939e-06, "loss": 0.8848, "step": 9903 }, { "epoch": 1.0538412428176207, "grad_norm": 11.0546875, "learning_rate": 9.665861284820922e-06, "loss": 0.9077, "step": 9904 }, { "epoch": 1.0539476484358374, "grad_norm": 10.875, "learning_rate": 9.665825753268904e-06, "loss": 1.0786, "step": 9905 }, { "epoch": 1.054054054054054, "grad_norm": 15.2421875, "learning_rate": 9.665790221716885e-06, "loss": 1.2348, "step": 9906 }, { "epoch": 1.0541604596722707, "grad_norm": 14.390625, "learning_rate": 9.665754690164867e-06, "loss": 0.9372, "step": 9907 }, { "epoch": 1.0542668652904874, "grad_norm": 16.71875, "learning_rate": 9.665719158612849e-06, "loss": 1.0719, "step": 9908 }, { "epoch": 1.054373270908704, "grad_norm": 13.1171875, "learning_rate": 9.665683627060832e-06, "loss": 1.031, "step": 9909 }, { "epoch": 1.0544796765269207, "grad_norm": 12.1015625, "learning_rate": 9.665648095508813e-06, "loss": 1.0432, "step": 9910 }, { "epoch": 1.0545860821451372, "grad_norm": 17.0, "learning_rate": 9.665612563956795e-06, "loss": 0.8311, "step": 9911 }, { "epoch": 1.054692487763354, "grad_norm": 15.1015625, "learning_rate": 9.665577032404777e-06, "loss": 1.18, "step": 9912 }, { "epoch": 1.0547988933815706, "grad_norm": 11.4140625, "learning_rate": 9.665541500852758e-06, "loss": 1.0656, "step": 9913 }, { "epoch": 1.0549052989997871, "grad_norm": 11.046875, "learning_rate": 9.66550596930074e-06, "loss": 1.2739, "step": 9914 }, { "epoch": 1.0550117046180039, "grad_norm": 21.03125, "learning_rate": 9.665470437748721e-06, "loss": 1.3678, "step": 9915 }, { "epoch": 1.0551181102362204, "grad_norm": 13.9296875, "learning_rate": 9.665434906196703e-06, "loss": 1.0235, "step": 9916 }, { "epoch": 1.055224515854437, "grad_norm": 23.078125, "learning_rate": 9.665399374644685e-06, "loss": 0.8208, "step": 9917 }, { "epoch": 1.0553309214726538, "grad_norm": 10.8125, "learning_rate": 9.665363843092666e-06, "loss": 1.0885, "step": 9918 }, { "epoch": 1.0554373270908703, "grad_norm": 20.359375, "learning_rate": 9.66532831154065e-06, "loss": 1.1199, "step": 9919 }, { "epoch": 1.055543732709087, "grad_norm": 12.359375, "learning_rate": 9.665292779988631e-06, "loss": 1.3866, "step": 9920 }, { "epoch": 1.0556501383273036, "grad_norm": 12.0, "learning_rate": 9.665257248436613e-06, "loss": 1.2141, "step": 9921 }, { "epoch": 1.0557565439455203, "grad_norm": 14.40625, "learning_rate": 9.665221716884594e-06, "loss": 0.8731, "step": 9922 }, { "epoch": 1.055862949563737, "grad_norm": 10.3828125, "learning_rate": 9.665186185332576e-06, "loss": 1.1415, "step": 9923 }, { "epoch": 1.0559693551819536, "grad_norm": 11.8203125, "learning_rate": 9.665150653780558e-06, "loss": 1.1674, "step": 9924 }, { "epoch": 1.0560757608001703, "grad_norm": 10.9140625, "learning_rate": 9.66511512222854e-06, "loss": 0.9682, "step": 9925 }, { "epoch": 1.0561821664183868, "grad_norm": 13.1484375, "learning_rate": 9.665079590676521e-06, "loss": 1.2352, "step": 9926 }, { "epoch": 1.0562885720366035, "grad_norm": 19.171875, "learning_rate": 9.665044059124502e-06, "loss": 1.0166, "step": 9927 }, { "epoch": 1.0563949776548203, "grad_norm": 11.6171875, "learning_rate": 9.665008527572484e-06, "loss": 1.0533, "step": 9928 }, { "epoch": 1.0565013832730368, "grad_norm": 11.890625, "learning_rate": 9.664972996020467e-06, "loss": 1.052, "step": 9929 }, { "epoch": 1.0566077888912535, "grad_norm": 11.453125, "learning_rate": 9.664937464468449e-06, "loss": 1.0378, "step": 9930 }, { "epoch": 1.05671419450947, "grad_norm": 10.1875, "learning_rate": 9.66490193291643e-06, "loss": 1.0571, "step": 9931 }, { "epoch": 1.0568206001276867, "grad_norm": 8.578125, "learning_rate": 9.664866401364412e-06, "loss": 1.0027, "step": 9932 }, { "epoch": 1.0569270057459035, "grad_norm": 9.6796875, "learning_rate": 9.664830869812394e-06, "loss": 1.0441, "step": 9933 }, { "epoch": 1.05703341136412, "grad_norm": 12.765625, "learning_rate": 9.664795338260377e-06, "loss": 1.0533, "step": 9934 }, { "epoch": 1.0571398169823367, "grad_norm": 14.2734375, "learning_rate": 9.664759806708359e-06, "loss": 1.0373, "step": 9935 }, { "epoch": 1.0572462226005532, "grad_norm": 12.0703125, "learning_rate": 9.66472427515634e-06, "loss": 1.0325, "step": 9936 }, { "epoch": 1.05735262821877, "grad_norm": 13.5625, "learning_rate": 9.664688743604322e-06, "loss": 0.9649, "step": 9937 }, { "epoch": 1.0574590338369867, "grad_norm": 10.2109375, "learning_rate": 9.664653212052304e-06, "loss": 0.942, "step": 9938 }, { "epoch": 1.0575654394552032, "grad_norm": 10.8359375, "learning_rate": 9.664617680500285e-06, "loss": 1.2528, "step": 9939 }, { "epoch": 1.05767184507342, "grad_norm": 16.96875, "learning_rate": 9.664582148948267e-06, "loss": 1.5052, "step": 9940 }, { "epoch": 1.0577782506916364, "grad_norm": 13.1484375, "learning_rate": 9.664546617396248e-06, "loss": 0.9238, "step": 9941 }, { "epoch": 1.0578846563098532, "grad_norm": 11.8828125, "learning_rate": 9.66451108584423e-06, "loss": 1.1888, "step": 9942 }, { "epoch": 1.05799106192807, "grad_norm": 10.859375, "learning_rate": 9.664475554292212e-06, "loss": 0.9478, "step": 9943 }, { "epoch": 1.0580974675462864, "grad_norm": 12.7109375, "learning_rate": 9.664440022740193e-06, "loss": 0.9373, "step": 9944 }, { "epoch": 1.0582038731645032, "grad_norm": 12.9765625, "learning_rate": 9.664404491188176e-06, "loss": 1.0312, "step": 9945 }, { "epoch": 1.0583102787827197, "grad_norm": 12.90625, "learning_rate": 9.664368959636158e-06, "loss": 1.0494, "step": 9946 }, { "epoch": 1.0584166844009364, "grad_norm": 13.234375, "learning_rate": 9.66433342808414e-06, "loss": 0.9566, "step": 9947 }, { "epoch": 1.0585230900191531, "grad_norm": 15.828125, "learning_rate": 9.664297896532121e-06, "loss": 1.1008, "step": 9948 }, { "epoch": 1.0586294956373696, "grad_norm": 14.296875, "learning_rate": 9.664262364980103e-06, "loss": 1.1241, "step": 9949 }, { "epoch": 1.0587359012555864, "grad_norm": 17.4375, "learning_rate": 9.664226833428084e-06, "loss": 1.1586, "step": 9950 }, { "epoch": 1.0588423068738029, "grad_norm": 15.7734375, "learning_rate": 9.664191301876066e-06, "loss": 0.8061, "step": 9951 }, { "epoch": 1.0589487124920196, "grad_norm": 14.5390625, "learning_rate": 9.664155770324048e-06, "loss": 1.087, "step": 9952 }, { "epoch": 1.0590551181102361, "grad_norm": 12.8359375, "learning_rate": 9.66412023877203e-06, "loss": 1.3142, "step": 9953 }, { "epoch": 1.0591615237284528, "grad_norm": 14.2421875, "learning_rate": 9.664084707220011e-06, "loss": 1.0653, "step": 9954 }, { "epoch": 1.0592679293466696, "grad_norm": 16.375, "learning_rate": 9.664049175667994e-06, "loss": 1.4142, "step": 9955 }, { "epoch": 1.059374334964886, "grad_norm": 13.6015625, "learning_rate": 9.664013644115976e-06, "loss": 1.0061, "step": 9956 }, { "epoch": 1.0594807405831028, "grad_norm": 13.328125, "learning_rate": 9.663978112563957e-06, "loss": 0.8286, "step": 9957 }, { "epoch": 1.0595871462013193, "grad_norm": 13.71875, "learning_rate": 9.663942581011939e-06, "loss": 1.2182, "step": 9958 }, { "epoch": 1.059693551819536, "grad_norm": 16.515625, "learning_rate": 9.66390704945992e-06, "loss": 1.1157, "step": 9959 }, { "epoch": 1.0597999574377528, "grad_norm": 13.4609375, "learning_rate": 9.663871517907904e-06, "loss": 1.2115, "step": 9960 }, { "epoch": 1.0599063630559693, "grad_norm": 10.171875, "learning_rate": 9.663835986355886e-06, "loss": 0.9691, "step": 9961 }, { "epoch": 1.060012768674186, "grad_norm": 13.90625, "learning_rate": 9.663800454803867e-06, "loss": 1.1087, "step": 9962 }, { "epoch": 1.0601191742924025, "grad_norm": 12.15625, "learning_rate": 9.663764923251849e-06, "loss": 1.3297, "step": 9963 }, { "epoch": 1.0602255799106193, "grad_norm": 12.1171875, "learning_rate": 9.66372939169983e-06, "loss": 1.1606, "step": 9964 }, { "epoch": 1.060331985528836, "grad_norm": 14.1640625, "learning_rate": 9.663693860147812e-06, "loss": 1.3027, "step": 9965 }, { "epoch": 1.0604383911470525, "grad_norm": 17.53125, "learning_rate": 9.663658328595794e-06, "loss": 1.1587, "step": 9966 }, { "epoch": 1.0605447967652692, "grad_norm": 20.625, "learning_rate": 9.663622797043775e-06, "loss": 1.6768, "step": 9967 }, { "epoch": 1.0606512023834858, "grad_norm": 11.2890625, "learning_rate": 9.663587265491757e-06, "loss": 1.1824, "step": 9968 }, { "epoch": 1.0607576080017025, "grad_norm": 11.125, "learning_rate": 9.663551733939738e-06, "loss": 1.1501, "step": 9969 }, { "epoch": 1.0608640136199192, "grad_norm": 10.921875, "learning_rate": 9.663516202387722e-06, "loss": 0.9904, "step": 9970 }, { "epoch": 1.0609704192381357, "grad_norm": 12.609375, "learning_rate": 9.663480670835703e-06, "loss": 1.0172, "step": 9971 }, { "epoch": 1.0610768248563525, "grad_norm": 11.4921875, "learning_rate": 9.663445139283685e-06, "loss": 1.0167, "step": 9972 }, { "epoch": 1.061183230474569, "grad_norm": 10.03125, "learning_rate": 9.663409607731667e-06, "loss": 0.9666, "step": 9973 }, { "epoch": 1.0612896360927857, "grad_norm": 11.8203125, "learning_rate": 9.663374076179648e-06, "loss": 0.9863, "step": 9974 }, { "epoch": 1.0613960417110024, "grad_norm": 14.625, "learning_rate": 9.663338544627631e-06, "loss": 1.2944, "step": 9975 }, { "epoch": 1.061502447329219, "grad_norm": 17.40625, "learning_rate": 9.663303013075613e-06, "loss": 1.0694, "step": 9976 }, { "epoch": 1.0616088529474357, "grad_norm": 15.046875, "learning_rate": 9.663267481523595e-06, "loss": 1.269, "step": 9977 }, { "epoch": 1.0617152585656522, "grad_norm": 15.046875, "learning_rate": 9.663231949971576e-06, "loss": 1.0288, "step": 9978 }, { "epoch": 1.061821664183869, "grad_norm": 9.828125, "learning_rate": 9.663196418419558e-06, "loss": 0.8348, "step": 9979 }, { "epoch": 1.0619280698020857, "grad_norm": 10.2578125, "learning_rate": 9.663160886867538e-06, "loss": 0.9255, "step": 9980 }, { "epoch": 1.0620344754203022, "grad_norm": 12.6953125, "learning_rate": 9.663125355315521e-06, "loss": 1.0124, "step": 9981 }, { "epoch": 1.062140881038519, "grad_norm": 12.609375, "learning_rate": 9.663089823763503e-06, "loss": 1.4762, "step": 9982 }, { "epoch": 1.0622472866567354, "grad_norm": 63.875, "learning_rate": 9.663054292211484e-06, "loss": 0.8355, "step": 9983 }, { "epoch": 1.0623536922749521, "grad_norm": 11.6484375, "learning_rate": 9.663018760659466e-06, "loss": 1.1809, "step": 9984 }, { "epoch": 1.0624600978931689, "grad_norm": 12.53125, "learning_rate": 9.662983229107448e-06, "loss": 0.8382, "step": 9985 }, { "epoch": 1.0625665035113854, "grad_norm": 18.78125, "learning_rate": 9.66294769755543e-06, "loss": 1.2152, "step": 9986 }, { "epoch": 1.062672909129602, "grad_norm": 15.703125, "learning_rate": 9.662912166003412e-06, "loss": 1.2059, "step": 9987 }, { "epoch": 1.0627793147478186, "grad_norm": 8.7265625, "learning_rate": 9.662876634451394e-06, "loss": 0.9709, "step": 9988 }, { "epoch": 1.0628857203660353, "grad_norm": 10.5, "learning_rate": 9.662841102899376e-06, "loss": 1.0224, "step": 9989 }, { "epoch": 1.0629921259842519, "grad_norm": 12.2109375, "learning_rate": 9.662805571347357e-06, "loss": 1.1736, "step": 9990 }, { "epoch": 1.0630985316024686, "grad_norm": 14.421875, "learning_rate": 9.662770039795339e-06, "loss": 0.8074, "step": 9991 }, { "epoch": 1.0632049372206853, "grad_norm": 12.078125, "learning_rate": 9.66273450824332e-06, "loss": 1.3291, "step": 9992 }, { "epoch": 1.0633113428389018, "grad_norm": 12.2734375, "learning_rate": 9.662698976691302e-06, "loss": 1.0557, "step": 9993 }, { "epoch": 1.0634177484571186, "grad_norm": 12.421875, "learning_rate": 9.662663445139284e-06, "loss": 1.1503, "step": 9994 }, { "epoch": 1.0635241540753353, "grad_norm": 14.546875, "learning_rate": 9.662627913587265e-06, "loss": 1.2611, "step": 9995 }, { "epoch": 1.0636305596935518, "grad_norm": 11.859375, "learning_rate": 9.662592382035249e-06, "loss": 1.243, "step": 9996 }, { "epoch": 1.0637369653117685, "grad_norm": 9.15625, "learning_rate": 9.66255685048323e-06, "loss": 0.4378, "step": 9997 }, { "epoch": 1.063843370929985, "grad_norm": 14.1875, "learning_rate": 9.662521318931212e-06, "loss": 0.9427, "step": 9998 }, { "epoch": 1.0639497765482018, "grad_norm": 10.546875, "learning_rate": 9.662485787379193e-06, "loss": 1.284, "step": 9999 }, { "epoch": 1.0640561821664183, "grad_norm": 16.28125, "learning_rate": 9.662450255827175e-06, "loss": 1.3182, "step": 10000 }, { "epoch": 1.064162587784635, "grad_norm": 11.734375, "learning_rate": 9.662414724275158e-06, "loss": 0.9437, "step": 10001 }, { "epoch": 1.0642689934028517, "grad_norm": 11.75, "learning_rate": 9.66237919272314e-06, "loss": 1.1624, "step": 10002 }, { "epoch": 1.0643753990210683, "grad_norm": 15.234375, "learning_rate": 9.662343661171122e-06, "loss": 1.1429, "step": 10003 }, { "epoch": 1.064481804639285, "grad_norm": 10.0, "learning_rate": 9.662308129619103e-06, "loss": 1.0344, "step": 10004 }, { "epoch": 1.0645882102575015, "grad_norm": 14.6953125, "learning_rate": 9.662272598067085e-06, "loss": 1.0032, "step": 10005 }, { "epoch": 1.0646946158757182, "grad_norm": 10.5234375, "learning_rate": 9.662237066515066e-06, "loss": 0.9455, "step": 10006 }, { "epoch": 1.064801021493935, "grad_norm": 7.94140625, "learning_rate": 9.662201534963048e-06, "loss": 0.9793, "step": 10007 }, { "epoch": 1.0649074271121515, "grad_norm": 10.421875, "learning_rate": 9.66216600341103e-06, "loss": 0.9601, "step": 10008 }, { "epoch": 1.0650138327303682, "grad_norm": 14.140625, "learning_rate": 9.662130471859011e-06, "loss": 1.0326, "step": 10009 }, { "epoch": 1.0651202383485847, "grad_norm": 19.015625, "learning_rate": 9.662094940306993e-06, "loss": 1.1478, "step": 10010 }, { "epoch": 1.0652266439668014, "grad_norm": 15.015625, "learning_rate": 9.662059408754976e-06, "loss": 1.1812, "step": 10011 }, { "epoch": 1.0653330495850182, "grad_norm": 11.5390625, "learning_rate": 9.662023877202958e-06, "loss": 1.1143, "step": 10012 }, { "epoch": 1.0654394552032347, "grad_norm": 12.09375, "learning_rate": 9.66198834565094e-06, "loss": 1.04, "step": 10013 }, { "epoch": 1.0655458608214514, "grad_norm": 11.0703125, "learning_rate": 9.661952814098921e-06, "loss": 0.9989, "step": 10014 }, { "epoch": 1.065652266439668, "grad_norm": 11.28125, "learning_rate": 9.661917282546902e-06, "loss": 1.0419, "step": 10015 }, { "epoch": 1.0657586720578847, "grad_norm": 11.21875, "learning_rate": 9.661881750994884e-06, "loss": 1.1754, "step": 10016 }, { "epoch": 1.0658650776761014, "grad_norm": 14.390625, "learning_rate": 9.661846219442866e-06, "loss": 1.3865, "step": 10017 }, { "epoch": 1.065971483294318, "grad_norm": 13.3671875, "learning_rate": 9.661810687890847e-06, "loss": 1.2321, "step": 10018 }, { "epoch": 1.0660778889125346, "grad_norm": 14.4765625, "learning_rate": 9.661775156338829e-06, "loss": 1.3057, "step": 10019 }, { "epoch": 1.0661842945307511, "grad_norm": 11.171875, "learning_rate": 9.66173962478681e-06, "loss": 1.4715, "step": 10020 }, { "epoch": 1.0662907001489679, "grad_norm": 16.3125, "learning_rate": 9.661704093234792e-06, "loss": 1.3314, "step": 10021 }, { "epoch": 1.0663971057671846, "grad_norm": 12.6015625, "learning_rate": 9.661668561682775e-06, "loss": 1.0931, "step": 10022 }, { "epoch": 1.0665035113854011, "grad_norm": 8.3046875, "learning_rate": 9.661633030130757e-06, "loss": 1.0993, "step": 10023 }, { "epoch": 1.0666099170036178, "grad_norm": 15.984375, "learning_rate": 9.661597498578739e-06, "loss": 1.1608, "step": 10024 }, { "epoch": 1.0667163226218344, "grad_norm": 14.65625, "learning_rate": 9.66156196702672e-06, "loss": 1.1093, "step": 10025 }, { "epoch": 1.066822728240051, "grad_norm": 9.84375, "learning_rate": 9.661526435474702e-06, "loss": 0.8317, "step": 10026 }, { "epoch": 1.0669291338582678, "grad_norm": 10.5703125, "learning_rate": 9.661490903922685e-06, "loss": 1.1815, "step": 10027 }, { "epoch": 1.0670355394764843, "grad_norm": 13.375, "learning_rate": 9.661455372370667e-06, "loss": 0.9301, "step": 10028 }, { "epoch": 1.067141945094701, "grad_norm": 18.328125, "learning_rate": 9.661419840818648e-06, "loss": 1.2415, "step": 10029 }, { "epoch": 1.0672483507129176, "grad_norm": 13.8828125, "learning_rate": 9.66138430926663e-06, "loss": 1.1905, "step": 10030 }, { "epoch": 1.0673547563311343, "grad_norm": 15.2265625, "learning_rate": 9.661348777714612e-06, "loss": 1.1887, "step": 10031 }, { "epoch": 1.067461161949351, "grad_norm": 15.96875, "learning_rate": 9.661313246162593e-06, "loss": 1.0645, "step": 10032 }, { "epoch": 1.0675675675675675, "grad_norm": 11.7578125, "learning_rate": 9.661277714610575e-06, "loss": 0.925, "step": 10033 }, { "epoch": 1.0676739731857843, "grad_norm": 15.796875, "learning_rate": 9.661242183058556e-06, "loss": 1.003, "step": 10034 }, { "epoch": 1.0677803788040008, "grad_norm": 15.109375, "learning_rate": 9.661206651506538e-06, "loss": 1.0943, "step": 10035 }, { "epoch": 1.0678867844222175, "grad_norm": 9.21875, "learning_rate": 9.66117111995452e-06, "loss": 1.0333, "step": 10036 }, { "epoch": 1.067993190040434, "grad_norm": 10.4140625, "learning_rate": 9.661135588402503e-06, "loss": 1.0924, "step": 10037 }, { "epoch": 1.0680995956586508, "grad_norm": 13.078125, "learning_rate": 9.661100056850485e-06, "loss": 1.1051, "step": 10038 }, { "epoch": 1.0682060012768675, "grad_norm": 13.25, "learning_rate": 9.661064525298466e-06, "loss": 1.4475, "step": 10039 }, { "epoch": 1.068312406895084, "grad_norm": 11.1796875, "learning_rate": 9.661028993746448e-06, "loss": 1.2156, "step": 10040 }, { "epoch": 1.0684188125133007, "grad_norm": 13.34375, "learning_rate": 9.66099346219443e-06, "loss": 1.0048, "step": 10041 }, { "epoch": 1.0685252181315172, "grad_norm": 12.125, "learning_rate": 9.660957930642411e-06, "loss": 1.1776, "step": 10042 }, { "epoch": 1.068631623749734, "grad_norm": 10.4375, "learning_rate": 9.660922399090393e-06, "loss": 1.0584, "step": 10043 }, { "epoch": 1.0687380293679507, "grad_norm": 11.4296875, "learning_rate": 9.660886867538374e-06, "loss": 1.2103, "step": 10044 }, { "epoch": 1.0688444349861672, "grad_norm": 13.15625, "learning_rate": 9.660851335986356e-06, "loss": 1.0857, "step": 10045 }, { "epoch": 1.068950840604384, "grad_norm": 19.90625, "learning_rate": 9.660815804434337e-06, "loss": 1.4049, "step": 10046 }, { "epoch": 1.0690572462226005, "grad_norm": 13.09375, "learning_rate": 9.66078027288232e-06, "loss": 0.9958, "step": 10047 }, { "epoch": 1.0691636518408172, "grad_norm": 14.9765625, "learning_rate": 9.660744741330302e-06, "loss": 1.238, "step": 10048 }, { "epoch": 1.069270057459034, "grad_norm": 13.5546875, "learning_rate": 9.660709209778284e-06, "loss": 1.3145, "step": 10049 }, { "epoch": 1.0693764630772504, "grad_norm": 13.4375, "learning_rate": 9.660673678226265e-06, "loss": 1.0001, "step": 10050 }, { "epoch": 1.0694828686954672, "grad_norm": 11.2578125, "learning_rate": 9.660638146674247e-06, "loss": 0.9453, "step": 10051 }, { "epoch": 1.0695892743136837, "grad_norm": 19.0625, "learning_rate": 9.66060261512223e-06, "loss": 0.9855, "step": 10052 }, { "epoch": 1.0696956799319004, "grad_norm": 16.828125, "learning_rate": 9.660567083570212e-06, "loss": 1.1316, "step": 10053 }, { "epoch": 1.0698020855501171, "grad_norm": 13.8203125, "learning_rate": 9.660531552018194e-06, "loss": 1.147, "step": 10054 }, { "epoch": 1.0699084911683336, "grad_norm": 12.5703125, "learning_rate": 9.660496020466175e-06, "loss": 1.201, "step": 10055 }, { "epoch": 1.0700148967865504, "grad_norm": 11.546875, "learning_rate": 9.660460488914157e-06, "loss": 0.9267, "step": 10056 }, { "epoch": 1.0701213024047669, "grad_norm": 11.3984375, "learning_rate": 9.660424957362138e-06, "loss": 1.061, "step": 10057 }, { "epoch": 1.0702277080229836, "grad_norm": 10.8671875, "learning_rate": 9.66038942581012e-06, "loss": 1.0388, "step": 10058 }, { "epoch": 1.0703341136412003, "grad_norm": 12.234375, "learning_rate": 9.660353894258102e-06, "loss": 1.0864, "step": 10059 }, { "epoch": 1.0704405192594169, "grad_norm": 21.875, "learning_rate": 9.660318362706083e-06, "loss": 1.0812, "step": 10060 }, { "epoch": 1.0705469248776336, "grad_norm": 17.015625, "learning_rate": 9.660282831154065e-06, "loss": 1.5363, "step": 10061 }, { "epoch": 1.07065333049585, "grad_norm": 11.6640625, "learning_rate": 9.660247299602046e-06, "loss": 1.174, "step": 10062 }, { "epoch": 1.0707597361140668, "grad_norm": 12.921875, "learning_rate": 9.66021176805003e-06, "loss": 1.1234, "step": 10063 }, { "epoch": 1.0708661417322836, "grad_norm": 12.34375, "learning_rate": 9.660176236498011e-06, "loss": 1.0715, "step": 10064 }, { "epoch": 1.0709725473505, "grad_norm": 12.4296875, "learning_rate": 9.660140704945993e-06, "loss": 0.9137, "step": 10065 }, { "epoch": 1.0710789529687168, "grad_norm": 20.171875, "learning_rate": 9.660105173393975e-06, "loss": 1.3279, "step": 10066 }, { "epoch": 1.0711853585869333, "grad_norm": 11.8046875, "learning_rate": 9.660069641841956e-06, "loss": 0.9826, "step": 10067 }, { "epoch": 1.07129176420515, "grad_norm": 14.9453125, "learning_rate": 9.66003411028994e-06, "loss": 1.2555, "step": 10068 }, { "epoch": 1.0713981698233668, "grad_norm": 10.7265625, "learning_rate": 9.65999857873792e-06, "loss": 1.1804, "step": 10069 }, { "epoch": 1.0715045754415833, "grad_norm": 11.0703125, "learning_rate": 9.659963047185901e-06, "loss": 0.9371, "step": 10070 }, { "epoch": 1.0716109810598, "grad_norm": 12.71875, "learning_rate": 9.659927515633883e-06, "loss": 1.0133, "step": 10071 }, { "epoch": 1.0717173866780165, "grad_norm": 12.1015625, "learning_rate": 9.659891984081864e-06, "loss": 1.4107, "step": 10072 }, { "epoch": 1.0718237922962333, "grad_norm": 12.7265625, "learning_rate": 9.659856452529848e-06, "loss": 1.1035, "step": 10073 }, { "epoch": 1.0719301979144498, "grad_norm": 12.2421875, "learning_rate": 9.659820920977829e-06, "loss": 0.9615, "step": 10074 }, { "epoch": 1.0720366035326665, "grad_norm": 13.546875, "learning_rate": 9.65978538942581e-06, "loss": 1.1916, "step": 10075 }, { "epoch": 1.0721430091508832, "grad_norm": 10.5234375, "learning_rate": 9.659749857873792e-06, "loss": 1.1128, "step": 10076 }, { "epoch": 1.0722494147690997, "grad_norm": 14.0, "learning_rate": 9.659714326321774e-06, "loss": 1.0007, "step": 10077 }, { "epoch": 1.0723558203873165, "grad_norm": 13.6953125, "learning_rate": 9.659678794769757e-06, "loss": 1.1326, "step": 10078 }, { "epoch": 1.0724622260055332, "grad_norm": 10.046875, "learning_rate": 9.659643263217739e-06, "loss": 1.104, "step": 10079 }, { "epoch": 1.0725686316237497, "grad_norm": 10.90625, "learning_rate": 9.65960773166572e-06, "loss": 1.0719, "step": 10080 }, { "epoch": 1.0726750372419664, "grad_norm": 14.8671875, "learning_rate": 9.659572200113702e-06, "loss": 0.9796, "step": 10081 }, { "epoch": 1.072781442860183, "grad_norm": 15.7890625, "learning_rate": 9.659536668561684e-06, "loss": 1.0905, "step": 10082 }, { "epoch": 1.0728878484783997, "grad_norm": 14.3671875, "learning_rate": 9.659501137009665e-06, "loss": 1.2458, "step": 10083 }, { "epoch": 1.0729942540966162, "grad_norm": 11.5234375, "learning_rate": 9.659465605457647e-06, "loss": 0.9759, "step": 10084 }, { "epoch": 1.073100659714833, "grad_norm": 13.609375, "learning_rate": 9.659430073905629e-06, "loss": 1.295, "step": 10085 }, { "epoch": 1.0732070653330497, "grad_norm": 15.4765625, "learning_rate": 9.65939454235361e-06, "loss": 1.0471, "step": 10086 }, { "epoch": 1.0733134709512662, "grad_norm": 12.2109375, "learning_rate": 9.659359010801592e-06, "loss": 1.2365, "step": 10087 }, { "epoch": 1.073419876569483, "grad_norm": 13.0546875, "learning_rate": 9.659323479249575e-06, "loss": 1.067, "step": 10088 }, { "epoch": 1.0735262821876994, "grad_norm": 11.671875, "learning_rate": 9.659287947697557e-06, "loss": 1.1803, "step": 10089 }, { "epoch": 1.0736326878059161, "grad_norm": 17.328125, "learning_rate": 9.659252416145538e-06, "loss": 0.9895, "step": 10090 }, { "epoch": 1.0737390934241329, "grad_norm": 12.984375, "learning_rate": 9.65921688459352e-06, "loss": 1.0707, "step": 10091 }, { "epoch": 1.0738454990423494, "grad_norm": 9.9453125, "learning_rate": 9.659181353041501e-06, "loss": 1.0017, "step": 10092 }, { "epoch": 1.0739519046605661, "grad_norm": 12.859375, "learning_rate": 9.659145821489485e-06, "loss": 1.1155, "step": 10093 }, { "epoch": 1.0740583102787826, "grad_norm": 13.0625, "learning_rate": 9.659110289937466e-06, "loss": 0.9443, "step": 10094 }, { "epoch": 1.0741647158969994, "grad_norm": 11.8359375, "learning_rate": 9.659074758385448e-06, "loss": 0.9347, "step": 10095 }, { "epoch": 1.074271121515216, "grad_norm": 10.0703125, "learning_rate": 9.65903922683343e-06, "loss": 0.7224, "step": 10096 }, { "epoch": 1.0743775271334326, "grad_norm": 12.6171875, "learning_rate": 9.659003695281411e-06, "loss": 0.937, "step": 10097 }, { "epoch": 1.0744839327516493, "grad_norm": 12.4296875, "learning_rate": 9.658968163729393e-06, "loss": 0.9071, "step": 10098 }, { "epoch": 1.0745903383698658, "grad_norm": 15.5546875, "learning_rate": 9.658932632177374e-06, "loss": 1.0181, "step": 10099 }, { "epoch": 1.0746967439880826, "grad_norm": 10.4453125, "learning_rate": 9.658897100625356e-06, "loss": 0.9623, "step": 10100 }, { "epoch": 1.0748031496062993, "grad_norm": 12.7578125, "learning_rate": 9.658861569073338e-06, "loss": 0.9565, "step": 10101 }, { "epoch": 1.0749095552245158, "grad_norm": 7.53125, "learning_rate": 9.65882603752132e-06, "loss": 0.8005, "step": 10102 }, { "epoch": 1.0750159608427325, "grad_norm": 11.09375, "learning_rate": 9.6587905059693e-06, "loss": 0.9142, "step": 10103 }, { "epoch": 1.075122366460949, "grad_norm": 11.7890625, "learning_rate": 9.658754974417284e-06, "loss": 0.9883, "step": 10104 }, { "epoch": 1.0752287720791658, "grad_norm": 13.0859375, "learning_rate": 9.658719442865266e-06, "loss": 1.328, "step": 10105 }, { "epoch": 1.0753351776973825, "grad_norm": 20.015625, "learning_rate": 9.658683911313247e-06, "loss": 1.0453, "step": 10106 }, { "epoch": 1.075441583315599, "grad_norm": 16.5, "learning_rate": 9.658648379761229e-06, "loss": 1.3176, "step": 10107 }, { "epoch": 1.0755479889338158, "grad_norm": 13.8515625, "learning_rate": 9.65861284820921e-06, "loss": 1.1127, "step": 10108 }, { "epoch": 1.0756543945520323, "grad_norm": 12.40625, "learning_rate": 9.658577316657192e-06, "loss": 1.109, "step": 10109 }, { "epoch": 1.075760800170249, "grad_norm": 9.2734375, "learning_rate": 9.658541785105174e-06, "loss": 0.9752, "step": 10110 }, { "epoch": 1.0758672057884657, "grad_norm": 10.0703125, "learning_rate": 9.658506253553155e-06, "loss": 0.9556, "step": 10111 }, { "epoch": 1.0759736114066822, "grad_norm": 15.6796875, "learning_rate": 9.658470722001137e-06, "loss": 0.9276, "step": 10112 }, { "epoch": 1.076080017024899, "grad_norm": 15.8203125, "learning_rate": 9.658435190449119e-06, "loss": 1.1004, "step": 10113 }, { "epoch": 1.0761864226431155, "grad_norm": 12.3046875, "learning_rate": 9.658399658897102e-06, "loss": 1.1941, "step": 10114 }, { "epoch": 1.0762928282613322, "grad_norm": 12.125, "learning_rate": 9.658364127345083e-06, "loss": 0.9977, "step": 10115 }, { "epoch": 1.076399233879549, "grad_norm": 11.015625, "learning_rate": 9.658328595793065e-06, "loss": 0.9816, "step": 10116 }, { "epoch": 1.0765056394977655, "grad_norm": 10.6640625, "learning_rate": 9.658293064241047e-06, "loss": 0.8918, "step": 10117 }, { "epoch": 1.0766120451159822, "grad_norm": 11.6796875, "learning_rate": 9.658257532689028e-06, "loss": 1.1435, "step": 10118 }, { "epoch": 1.0767184507341987, "grad_norm": 10.7734375, "learning_rate": 9.658222001137012e-06, "loss": 0.8959, "step": 10119 }, { "epoch": 1.0768248563524154, "grad_norm": 14.046875, "learning_rate": 9.658186469584993e-06, "loss": 1.1032, "step": 10120 }, { "epoch": 1.076931261970632, "grad_norm": 9.9453125, "learning_rate": 9.658150938032975e-06, "loss": 1.063, "step": 10121 }, { "epoch": 1.0770376675888487, "grad_norm": 13.3984375, "learning_rate": 9.658115406480956e-06, "loss": 1.011, "step": 10122 }, { "epoch": 1.0771440732070654, "grad_norm": 11.2890625, "learning_rate": 9.658079874928938e-06, "loss": 0.8604, "step": 10123 }, { "epoch": 1.077250478825282, "grad_norm": 15.765625, "learning_rate": 9.65804434337692e-06, "loss": 1.2007, "step": 10124 }, { "epoch": 1.0773568844434986, "grad_norm": 13.875, "learning_rate": 9.658008811824901e-06, "loss": 0.9206, "step": 10125 }, { "epoch": 1.0774632900617154, "grad_norm": 10.53125, "learning_rate": 9.657973280272883e-06, "loss": 0.9777, "step": 10126 }, { "epoch": 1.0775696956799319, "grad_norm": 11.4453125, "learning_rate": 9.657937748720864e-06, "loss": 1.0102, "step": 10127 }, { "epoch": 1.0776761012981486, "grad_norm": 15.28125, "learning_rate": 9.657902217168846e-06, "loss": 0.9899, "step": 10128 }, { "epoch": 1.0777825069163651, "grad_norm": 19.90625, "learning_rate": 9.65786668561683e-06, "loss": 1.5278, "step": 10129 }, { "epoch": 1.0778889125345819, "grad_norm": 11.671875, "learning_rate": 9.657831154064811e-06, "loss": 1.1893, "step": 10130 }, { "epoch": 1.0779953181527984, "grad_norm": 11.1640625, "learning_rate": 9.657795622512793e-06, "loss": 0.7662, "step": 10131 }, { "epoch": 1.078101723771015, "grad_norm": 11.6875, "learning_rate": 9.657760090960774e-06, "loss": 0.9175, "step": 10132 }, { "epoch": 1.0782081293892318, "grad_norm": 15.6484375, "learning_rate": 9.657724559408756e-06, "loss": 1.4157, "step": 10133 }, { "epoch": 1.0783145350074483, "grad_norm": 8.9296875, "learning_rate": 9.657689027856737e-06, "loss": 0.8324, "step": 10134 }, { "epoch": 1.078420940625665, "grad_norm": 15.984375, "learning_rate": 9.657653496304719e-06, "loss": 0.8191, "step": 10135 }, { "epoch": 1.0785273462438816, "grad_norm": 11.4609375, "learning_rate": 9.6576179647527e-06, "loss": 0.7486, "step": 10136 }, { "epoch": 1.0786337518620983, "grad_norm": 12.7890625, "learning_rate": 9.657582433200682e-06, "loss": 1.1455, "step": 10137 }, { "epoch": 1.078740157480315, "grad_norm": 19.390625, "learning_rate": 9.657546901648664e-06, "loss": 1.2237, "step": 10138 }, { "epoch": 1.0788465630985316, "grad_norm": 11.2578125, "learning_rate": 9.657511370096645e-06, "loss": 0.9608, "step": 10139 }, { "epoch": 1.0789529687167483, "grad_norm": 15.8203125, "learning_rate": 9.657475838544629e-06, "loss": 1.3624, "step": 10140 }, { "epoch": 1.0790593743349648, "grad_norm": 11.2421875, "learning_rate": 9.65744030699261e-06, "loss": 1.0433, "step": 10141 }, { "epoch": 1.0791657799531815, "grad_norm": 9.8046875, "learning_rate": 9.657404775440592e-06, "loss": 1.0127, "step": 10142 }, { "epoch": 1.0792721855713983, "grad_norm": 14.1640625, "learning_rate": 9.657369243888574e-06, "loss": 1.0913, "step": 10143 }, { "epoch": 1.0793785911896148, "grad_norm": 15.046875, "learning_rate": 9.657333712336555e-06, "loss": 1.0232, "step": 10144 }, { "epoch": 1.0794849968078315, "grad_norm": 11.015625, "learning_rate": 9.657298180784538e-06, "loss": 0.9632, "step": 10145 }, { "epoch": 1.079591402426048, "grad_norm": 10.703125, "learning_rate": 9.65726264923252e-06, "loss": 0.9651, "step": 10146 }, { "epoch": 1.0796978080442647, "grad_norm": 11.21875, "learning_rate": 9.657227117680502e-06, "loss": 1.1378, "step": 10147 }, { "epoch": 1.0798042136624815, "grad_norm": 14.125, "learning_rate": 9.657191586128483e-06, "loss": 0.8544, "step": 10148 }, { "epoch": 1.079910619280698, "grad_norm": 14.84375, "learning_rate": 9.657156054576465e-06, "loss": 1.3527, "step": 10149 }, { "epoch": 1.0800170248989147, "grad_norm": 12.9921875, "learning_rate": 9.657120523024446e-06, "loss": 1.1798, "step": 10150 }, { "epoch": 1.0801234305171312, "grad_norm": 12.328125, "learning_rate": 9.657084991472428e-06, "loss": 0.9758, "step": 10151 }, { "epoch": 1.080229836135348, "grad_norm": 14.0, "learning_rate": 9.65704945992041e-06, "loss": 1.2465, "step": 10152 }, { "epoch": 1.0803362417535647, "grad_norm": 13.2109375, "learning_rate": 9.657013928368391e-06, "loss": 0.9156, "step": 10153 }, { "epoch": 1.0804426473717812, "grad_norm": 11.7421875, "learning_rate": 9.656978396816373e-06, "loss": 0.945, "step": 10154 }, { "epoch": 1.080549052989998, "grad_norm": 14.8515625, "learning_rate": 9.656942865264356e-06, "loss": 1.4309, "step": 10155 }, { "epoch": 1.0806554586082144, "grad_norm": 13.078125, "learning_rate": 9.656907333712338e-06, "loss": 0.9915, "step": 10156 }, { "epoch": 1.0807618642264312, "grad_norm": 22.0625, "learning_rate": 9.65687180216032e-06, "loss": 1.164, "step": 10157 }, { "epoch": 1.080868269844648, "grad_norm": 14.75, "learning_rate": 9.656836270608301e-06, "loss": 1.2678, "step": 10158 }, { "epoch": 1.0809746754628644, "grad_norm": 14.5546875, "learning_rate": 9.656800739056283e-06, "loss": 1.5542, "step": 10159 }, { "epoch": 1.0810810810810811, "grad_norm": 16.734375, "learning_rate": 9.656765207504264e-06, "loss": 1.3331, "step": 10160 }, { "epoch": 1.0811874866992977, "grad_norm": 14.7578125, "learning_rate": 9.656729675952246e-06, "loss": 0.8524, "step": 10161 }, { "epoch": 1.0812938923175144, "grad_norm": 11.625, "learning_rate": 9.656694144400227e-06, "loss": 1.2123, "step": 10162 }, { "epoch": 1.0814002979357311, "grad_norm": 15.2890625, "learning_rate": 9.656658612848209e-06, "loss": 1.2452, "step": 10163 }, { "epoch": 1.0815067035539476, "grad_norm": 17.109375, "learning_rate": 9.65662308129619e-06, "loss": 0.8118, "step": 10164 }, { "epoch": 1.0816131091721644, "grad_norm": 10.0703125, "learning_rate": 9.656587549744174e-06, "loss": 1.1097, "step": 10165 }, { "epoch": 1.0817195147903809, "grad_norm": 12.109375, "learning_rate": 9.656552018192156e-06, "loss": 1.1938, "step": 10166 }, { "epoch": 1.0818259204085976, "grad_norm": 16.375, "learning_rate": 9.656516486640137e-06, "loss": 1.0534, "step": 10167 }, { "epoch": 1.081932326026814, "grad_norm": 16.34375, "learning_rate": 9.656480955088119e-06, "loss": 1.1461, "step": 10168 }, { "epoch": 1.0820387316450308, "grad_norm": 11.140625, "learning_rate": 9.6564454235361e-06, "loss": 0.9054, "step": 10169 }, { "epoch": 1.0821451372632476, "grad_norm": 15.53125, "learning_rate": 9.656409891984084e-06, "loss": 1.0409, "step": 10170 }, { "epoch": 1.082251542881464, "grad_norm": 13.78125, "learning_rate": 9.656374360432065e-06, "loss": 1.0029, "step": 10171 }, { "epoch": 1.0823579484996808, "grad_norm": 12.34375, "learning_rate": 9.656338828880047e-06, "loss": 0.6751, "step": 10172 }, { "epoch": 1.0824643541178975, "grad_norm": 11.34375, "learning_rate": 9.656303297328029e-06, "loss": 0.9887, "step": 10173 }, { "epoch": 1.082570759736114, "grad_norm": 15.1328125, "learning_rate": 9.65626776577601e-06, "loss": 1.0604, "step": 10174 }, { "epoch": 1.0826771653543308, "grad_norm": 11.1875, "learning_rate": 9.656232234223992e-06, "loss": 0.8303, "step": 10175 }, { "epoch": 1.0827835709725473, "grad_norm": 8.90625, "learning_rate": 9.656196702671973e-06, "loss": 1.0501, "step": 10176 }, { "epoch": 1.082889976590764, "grad_norm": 13.7890625, "learning_rate": 9.656161171119955e-06, "loss": 1.0868, "step": 10177 }, { "epoch": 1.0829963822089805, "grad_norm": 11.2578125, "learning_rate": 9.656125639567937e-06, "loss": 0.9558, "step": 10178 }, { "epoch": 1.0831027878271973, "grad_norm": 12.09375, "learning_rate": 9.656090108015918e-06, "loss": 1.1764, "step": 10179 }, { "epoch": 1.083209193445414, "grad_norm": 12.0078125, "learning_rate": 9.6560545764639e-06, "loss": 0.78, "step": 10180 }, { "epoch": 1.0833155990636305, "grad_norm": 8.140625, "learning_rate": 9.656019044911883e-06, "loss": 0.8947, "step": 10181 }, { "epoch": 1.0834220046818472, "grad_norm": 13.09375, "learning_rate": 9.655983513359865e-06, "loss": 0.9793, "step": 10182 }, { "epoch": 1.0835284103000637, "grad_norm": 10.5859375, "learning_rate": 9.655947981807846e-06, "loss": 1.1545, "step": 10183 }, { "epoch": 1.0836348159182805, "grad_norm": 16.6875, "learning_rate": 9.655912450255828e-06, "loss": 1.2115, "step": 10184 }, { "epoch": 1.0837412215364972, "grad_norm": 10.03125, "learning_rate": 9.65587691870381e-06, "loss": 1.0993, "step": 10185 }, { "epoch": 1.0838476271547137, "grad_norm": 12.75, "learning_rate": 9.655841387151793e-06, "loss": 1.1459, "step": 10186 }, { "epoch": 1.0839540327729305, "grad_norm": 11.5234375, "learning_rate": 9.655805855599774e-06, "loss": 1.1983, "step": 10187 }, { "epoch": 1.084060438391147, "grad_norm": 12.796875, "learning_rate": 9.655770324047754e-06, "loss": 1.0975, "step": 10188 }, { "epoch": 1.0841668440093637, "grad_norm": 14.4921875, "learning_rate": 9.655734792495736e-06, "loss": 1.0028, "step": 10189 }, { "epoch": 1.0842732496275804, "grad_norm": 13.9296875, "learning_rate": 9.655699260943718e-06, "loss": 0.9654, "step": 10190 }, { "epoch": 1.084379655245797, "grad_norm": 13.140625, "learning_rate": 9.6556637293917e-06, "loss": 1.0819, "step": 10191 }, { "epoch": 1.0844860608640137, "grad_norm": 10.5703125, "learning_rate": 9.655628197839682e-06, "loss": 1.2079, "step": 10192 }, { "epoch": 1.0845924664822302, "grad_norm": 15.25, "learning_rate": 9.655592666287664e-06, "loss": 1.4293, "step": 10193 }, { "epoch": 1.084698872100447, "grad_norm": 12.3828125, "learning_rate": 9.655557134735646e-06, "loss": 1.1179, "step": 10194 }, { "epoch": 1.0848052777186636, "grad_norm": 17.46875, "learning_rate": 9.655521603183627e-06, "loss": 0.9904, "step": 10195 }, { "epoch": 1.0849116833368802, "grad_norm": 12.921875, "learning_rate": 9.65548607163161e-06, "loss": 0.9727, "step": 10196 }, { "epoch": 1.0850180889550969, "grad_norm": 15.125, "learning_rate": 9.655450540079592e-06, "loss": 1.5798, "step": 10197 }, { "epoch": 1.0851244945733134, "grad_norm": 12.5625, "learning_rate": 9.655415008527574e-06, "loss": 1.059, "step": 10198 }, { "epoch": 1.0852309001915301, "grad_norm": 11.734375, "learning_rate": 9.655379476975555e-06, "loss": 0.8859, "step": 10199 }, { "epoch": 1.0853373058097469, "grad_norm": 10.921875, "learning_rate": 9.655343945423537e-06, "loss": 1.2578, "step": 10200 }, { "epoch": 1.0854437114279634, "grad_norm": 10.859375, "learning_rate": 9.655308413871519e-06, "loss": 1.303, "step": 10201 }, { "epoch": 1.08555011704618, "grad_norm": 9.6484375, "learning_rate": 9.6552728823195e-06, "loss": 0.8728, "step": 10202 }, { "epoch": 1.0856565226643966, "grad_norm": 12.6171875, "learning_rate": 9.655237350767482e-06, "loss": 1.5162, "step": 10203 }, { "epoch": 1.0857629282826133, "grad_norm": 11.859375, "learning_rate": 9.655201819215463e-06, "loss": 0.9105, "step": 10204 }, { "epoch": 1.0858693339008298, "grad_norm": 12.5625, "learning_rate": 9.655166287663445e-06, "loss": 1.3248, "step": 10205 }, { "epoch": 1.0859757395190466, "grad_norm": 13.1484375, "learning_rate": 9.655130756111428e-06, "loss": 1.2578, "step": 10206 }, { "epoch": 1.0860821451372633, "grad_norm": 9.171875, "learning_rate": 9.65509522455941e-06, "loss": 0.9075, "step": 10207 }, { "epoch": 1.0861885507554798, "grad_norm": 11.390625, "learning_rate": 9.655059693007392e-06, "loss": 1.2133, "step": 10208 }, { "epoch": 1.0862949563736966, "grad_norm": 14.96875, "learning_rate": 9.655024161455373e-06, "loss": 1.0574, "step": 10209 }, { "epoch": 1.0864013619919133, "grad_norm": 12.515625, "learning_rate": 9.654988629903355e-06, "loss": 0.9814, "step": 10210 }, { "epoch": 1.0865077676101298, "grad_norm": 13.0390625, "learning_rate": 9.654953098351338e-06, "loss": 1.1614, "step": 10211 }, { "epoch": 1.0866141732283465, "grad_norm": 10.6796875, "learning_rate": 9.65491756679932e-06, "loss": 0.9845, "step": 10212 }, { "epoch": 1.086720578846563, "grad_norm": 11.8359375, "learning_rate": 9.654882035247301e-06, "loss": 1.1356, "step": 10213 }, { "epoch": 1.0868269844647798, "grad_norm": 11.640625, "learning_rate": 9.654846503695283e-06, "loss": 1.0263, "step": 10214 }, { "epoch": 1.0869333900829963, "grad_norm": 10.3203125, "learning_rate": 9.654810972143264e-06, "loss": 0.9562, "step": 10215 }, { "epoch": 1.087039795701213, "grad_norm": 14.5234375, "learning_rate": 9.654775440591246e-06, "loss": 0.9251, "step": 10216 }, { "epoch": 1.0871462013194297, "grad_norm": 13.328125, "learning_rate": 9.654739909039228e-06, "loss": 0.8939, "step": 10217 }, { "epoch": 1.0872526069376462, "grad_norm": 14.171875, "learning_rate": 9.65470437748721e-06, "loss": 0.9907, "step": 10218 }, { "epoch": 1.087359012555863, "grad_norm": 10.6953125, "learning_rate": 9.654668845935191e-06, "loss": 1.0417, "step": 10219 }, { "epoch": 1.0874654181740795, "grad_norm": 12.2109375, "learning_rate": 9.654633314383173e-06, "loss": 0.9155, "step": 10220 }, { "epoch": 1.0875718237922962, "grad_norm": 11.578125, "learning_rate": 9.654597782831154e-06, "loss": 1.193, "step": 10221 }, { "epoch": 1.087678229410513, "grad_norm": 13.9453125, "learning_rate": 9.654562251279137e-06, "loss": 1.2098, "step": 10222 }, { "epoch": 1.0877846350287295, "grad_norm": 12.546875, "learning_rate": 9.654526719727119e-06, "loss": 0.9171, "step": 10223 }, { "epoch": 1.0878910406469462, "grad_norm": 13.0625, "learning_rate": 9.6544911881751e-06, "loss": 1.1877, "step": 10224 }, { "epoch": 1.0879974462651627, "grad_norm": 15.7265625, "learning_rate": 9.654455656623082e-06, "loss": 1.2348, "step": 10225 }, { "epoch": 1.0881038518833794, "grad_norm": 14.1640625, "learning_rate": 9.654420125071064e-06, "loss": 1.2867, "step": 10226 }, { "epoch": 1.0882102575015962, "grad_norm": 14.5078125, "learning_rate": 9.654384593519045e-06, "loss": 0.9712, "step": 10227 }, { "epoch": 1.0883166631198127, "grad_norm": 11.5625, "learning_rate": 9.654349061967027e-06, "loss": 1.007, "step": 10228 }, { "epoch": 1.0884230687380294, "grad_norm": 13.0625, "learning_rate": 9.654313530415009e-06, "loss": 1.1724, "step": 10229 }, { "epoch": 1.088529474356246, "grad_norm": 12.890625, "learning_rate": 9.65427799886299e-06, "loss": 0.9368, "step": 10230 }, { "epoch": 1.0886358799744627, "grad_norm": 10.8984375, "learning_rate": 9.654242467310972e-06, "loss": 1.1186, "step": 10231 }, { "epoch": 1.0887422855926794, "grad_norm": 12.4765625, "learning_rate": 9.654206935758955e-06, "loss": 1.0157, "step": 10232 }, { "epoch": 1.088848691210896, "grad_norm": 16.5, "learning_rate": 9.654171404206937e-06, "loss": 1.1883, "step": 10233 }, { "epoch": 1.0889550968291126, "grad_norm": 12.796875, "learning_rate": 9.654135872654918e-06, "loss": 1.2784, "step": 10234 }, { "epoch": 1.0890615024473291, "grad_norm": 13.25, "learning_rate": 9.6541003411029e-06, "loss": 1.0756, "step": 10235 }, { "epoch": 1.0891679080655459, "grad_norm": 14.5, "learning_rate": 9.654064809550882e-06, "loss": 1.103, "step": 10236 }, { "epoch": 1.0892743136837626, "grad_norm": 10.8671875, "learning_rate": 9.654029277998865e-06, "loss": 1.05, "step": 10237 }, { "epoch": 1.089380719301979, "grad_norm": 9.40625, "learning_rate": 9.653993746446847e-06, "loss": 1.1426, "step": 10238 }, { "epoch": 1.0894871249201958, "grad_norm": 15.15625, "learning_rate": 9.653958214894828e-06, "loss": 1.1093, "step": 10239 }, { "epoch": 1.0895935305384123, "grad_norm": 13.03125, "learning_rate": 9.65392268334281e-06, "loss": 1.5162, "step": 10240 }, { "epoch": 1.089699936156629, "grad_norm": 15.4140625, "learning_rate": 9.653887151790791e-06, "loss": 1.2166, "step": 10241 }, { "epoch": 1.0898063417748458, "grad_norm": 13.1328125, "learning_rate": 9.653851620238773e-06, "loss": 0.8267, "step": 10242 }, { "epoch": 1.0899127473930623, "grad_norm": 10.328125, "learning_rate": 9.653816088686755e-06, "loss": 1.0439, "step": 10243 }, { "epoch": 1.090019153011279, "grad_norm": 11.8046875, "learning_rate": 9.653780557134736e-06, "loss": 1.0173, "step": 10244 }, { "epoch": 1.0901255586294956, "grad_norm": 11.4453125, "learning_rate": 9.653745025582718e-06, "loss": 1.1089, "step": 10245 }, { "epoch": 1.0902319642477123, "grad_norm": 11.484375, "learning_rate": 9.6537094940307e-06, "loss": 1.1734, "step": 10246 }, { "epoch": 1.090338369865929, "grad_norm": 14.1015625, "learning_rate": 9.653673962478683e-06, "loss": 1.1212, "step": 10247 }, { "epoch": 1.0904447754841455, "grad_norm": 9.28125, "learning_rate": 9.653638430926664e-06, "loss": 1.0001, "step": 10248 }, { "epoch": 1.0905511811023623, "grad_norm": 13.328125, "learning_rate": 9.653602899374646e-06, "loss": 1.2749, "step": 10249 }, { "epoch": 1.0906575867205788, "grad_norm": 10.625, "learning_rate": 9.653567367822627e-06, "loss": 0.9037, "step": 10250 }, { "epoch": 1.0907639923387955, "grad_norm": 14.2265625, "learning_rate": 9.653531836270609e-06, "loss": 1.1128, "step": 10251 }, { "epoch": 1.090870397957012, "grad_norm": 9.84375, "learning_rate": 9.65349630471859e-06, "loss": 0.917, "step": 10252 }, { "epoch": 1.0909768035752287, "grad_norm": 10.484375, "learning_rate": 9.653460773166572e-06, "loss": 1.049, "step": 10253 }, { "epoch": 1.0910832091934455, "grad_norm": 10.421875, "learning_rate": 9.653425241614554e-06, "loss": 0.8205, "step": 10254 }, { "epoch": 1.091189614811662, "grad_norm": 9.9453125, "learning_rate": 9.653389710062536e-06, "loss": 1.0565, "step": 10255 }, { "epoch": 1.0912960204298787, "grad_norm": 11.171875, "learning_rate": 9.653354178510517e-06, "loss": 1.1005, "step": 10256 }, { "epoch": 1.0914024260480955, "grad_norm": 13.9140625, "learning_rate": 9.653318646958499e-06, "loss": 0.9987, "step": 10257 }, { "epoch": 1.091508831666312, "grad_norm": 9.0546875, "learning_rate": 9.653283115406482e-06, "loss": 0.9119, "step": 10258 }, { "epoch": 1.0916152372845287, "grad_norm": 10.0625, "learning_rate": 9.653247583854464e-06, "loss": 0.8395, "step": 10259 }, { "epoch": 1.0917216429027452, "grad_norm": 14.375, "learning_rate": 9.653212052302445e-06, "loss": 1.0262, "step": 10260 }, { "epoch": 1.091828048520962, "grad_norm": 10.515625, "learning_rate": 9.653176520750427e-06, "loss": 1.116, "step": 10261 }, { "epoch": 1.0919344541391784, "grad_norm": 13.96875, "learning_rate": 9.653140989198408e-06, "loss": 0.981, "step": 10262 }, { "epoch": 1.0920408597573952, "grad_norm": 11.0390625, "learning_rate": 9.653105457646392e-06, "loss": 0.8382, "step": 10263 }, { "epoch": 1.092147265375612, "grad_norm": 16.671875, "learning_rate": 9.653069926094373e-06, "loss": 1.0719, "step": 10264 }, { "epoch": 1.0922536709938284, "grad_norm": 15.0078125, "learning_rate": 9.653034394542355e-06, "loss": 1.4104, "step": 10265 }, { "epoch": 1.0923600766120451, "grad_norm": 16.1875, "learning_rate": 9.652998862990337e-06, "loss": 1.0089, "step": 10266 }, { "epoch": 1.0924664822302617, "grad_norm": 13.609375, "learning_rate": 9.652963331438318e-06, "loss": 1.1782, "step": 10267 }, { "epoch": 1.0925728878484784, "grad_norm": 13.0859375, "learning_rate": 9.6529277998863e-06, "loss": 1.0257, "step": 10268 }, { "epoch": 1.0926792934666951, "grad_norm": 13.203125, "learning_rate": 9.652892268334281e-06, "loss": 1.119, "step": 10269 }, { "epoch": 1.0927856990849116, "grad_norm": 12.625, "learning_rate": 9.652856736782263e-06, "loss": 0.9652, "step": 10270 }, { "epoch": 1.0928921047031284, "grad_norm": 12.2421875, "learning_rate": 9.652821205230245e-06, "loss": 0.8627, "step": 10271 }, { "epoch": 1.0929985103213449, "grad_norm": 12.015625, "learning_rate": 9.652785673678226e-06, "loss": 1.1011, "step": 10272 }, { "epoch": 1.0931049159395616, "grad_norm": 12.8828125, "learning_rate": 9.65275014212621e-06, "loss": 1.1562, "step": 10273 }, { "epoch": 1.0932113215577783, "grad_norm": 13.9609375, "learning_rate": 9.652714610574191e-06, "loss": 1.0694, "step": 10274 }, { "epoch": 1.0933177271759948, "grad_norm": 11.4765625, "learning_rate": 9.652679079022173e-06, "loss": 0.9996, "step": 10275 }, { "epoch": 1.0934241327942116, "grad_norm": 13.3515625, "learning_rate": 9.652643547470154e-06, "loss": 1.1437, "step": 10276 }, { "epoch": 1.093530538412428, "grad_norm": 14.53125, "learning_rate": 9.652608015918136e-06, "loss": 1.147, "step": 10277 }, { "epoch": 1.0936369440306448, "grad_norm": 19.640625, "learning_rate": 9.652572484366118e-06, "loss": 1.3703, "step": 10278 }, { "epoch": 1.0937433496488616, "grad_norm": 19.5, "learning_rate": 9.6525369528141e-06, "loss": 0.9947, "step": 10279 }, { "epoch": 1.093849755267078, "grad_norm": 12.6796875, "learning_rate": 9.65250142126208e-06, "loss": 1.0033, "step": 10280 }, { "epoch": 1.0939561608852948, "grad_norm": 14.6796875, "learning_rate": 9.652465889710062e-06, "loss": 0.9463, "step": 10281 }, { "epoch": 1.0940625665035113, "grad_norm": 15.3203125, "learning_rate": 9.652430358158044e-06, "loss": 1.2194, "step": 10282 }, { "epoch": 1.094168972121728, "grad_norm": 15.15625, "learning_rate": 9.652394826606027e-06, "loss": 0.9611, "step": 10283 }, { "epoch": 1.0942753777399448, "grad_norm": 10.5234375, "learning_rate": 9.652359295054009e-06, "loss": 1.1475, "step": 10284 }, { "epoch": 1.0943817833581613, "grad_norm": 11.2890625, "learning_rate": 9.65232376350199e-06, "loss": 1.0471, "step": 10285 }, { "epoch": 1.094488188976378, "grad_norm": 11.7265625, "learning_rate": 9.652288231949972e-06, "loss": 1.1064, "step": 10286 }, { "epoch": 1.0945945945945945, "grad_norm": 15.1875, "learning_rate": 9.652252700397954e-06, "loss": 1.23, "step": 10287 }, { "epoch": 1.0947010002128112, "grad_norm": 11.9296875, "learning_rate": 9.652217168845937e-06, "loss": 1.4259, "step": 10288 }, { "epoch": 1.094807405831028, "grad_norm": 15.1171875, "learning_rate": 9.652181637293919e-06, "loss": 1.0117, "step": 10289 }, { "epoch": 1.0949138114492445, "grad_norm": 10.7109375, "learning_rate": 9.6521461057419e-06, "loss": 1.2545, "step": 10290 }, { "epoch": 1.0950202170674612, "grad_norm": 10.8828125, "learning_rate": 9.652110574189882e-06, "loss": 0.938, "step": 10291 }, { "epoch": 1.0951266226856777, "grad_norm": 10.5, "learning_rate": 9.652075042637863e-06, "loss": 1.0222, "step": 10292 }, { "epoch": 1.0952330283038945, "grad_norm": 11.484375, "learning_rate": 9.652039511085845e-06, "loss": 1.1794, "step": 10293 }, { "epoch": 1.0953394339221112, "grad_norm": 12.4921875, "learning_rate": 9.652003979533827e-06, "loss": 0.8742, "step": 10294 }, { "epoch": 1.0954458395403277, "grad_norm": 10.90625, "learning_rate": 9.651968447981808e-06, "loss": 1.0452, "step": 10295 }, { "epoch": 1.0955522451585444, "grad_norm": 10.75, "learning_rate": 9.65193291642979e-06, "loss": 0.8464, "step": 10296 }, { "epoch": 1.095658650776761, "grad_norm": 18.765625, "learning_rate": 9.651897384877771e-06, "loss": 1.0225, "step": 10297 }, { "epoch": 1.0957650563949777, "grad_norm": 20.21875, "learning_rate": 9.651861853325753e-06, "loss": 1.3218, "step": 10298 }, { "epoch": 1.0958714620131942, "grad_norm": 15.1640625, "learning_rate": 9.651826321773736e-06, "loss": 0.9922, "step": 10299 }, { "epoch": 1.095977867631411, "grad_norm": 13.6640625, "learning_rate": 9.651790790221718e-06, "loss": 1.4084, "step": 10300 }, { "epoch": 1.0960842732496276, "grad_norm": 10.7890625, "learning_rate": 9.6517552586697e-06, "loss": 1.2078, "step": 10301 }, { "epoch": 1.0961906788678442, "grad_norm": 10.5703125, "learning_rate": 9.651719727117681e-06, "loss": 0.9715, "step": 10302 }, { "epoch": 1.096297084486061, "grad_norm": 9.0234375, "learning_rate": 9.651684195565663e-06, "loss": 1.0785, "step": 10303 }, { "epoch": 1.0964034901042776, "grad_norm": 17.359375, "learning_rate": 9.651648664013646e-06, "loss": 0.9408, "step": 10304 }, { "epoch": 1.0965098957224941, "grad_norm": 14.3203125, "learning_rate": 9.651613132461628e-06, "loss": 1.1244, "step": 10305 }, { "epoch": 1.0966163013407109, "grad_norm": 11.921875, "learning_rate": 9.65157760090961e-06, "loss": 1.1033, "step": 10306 }, { "epoch": 1.0967227069589274, "grad_norm": 12.6953125, "learning_rate": 9.651542069357591e-06, "loss": 1.3731, "step": 10307 }, { "epoch": 1.096829112577144, "grad_norm": 11.828125, "learning_rate": 9.65150653780557e-06, "loss": 0.9224, "step": 10308 }, { "epoch": 1.0969355181953606, "grad_norm": 22.46875, "learning_rate": 9.651471006253554e-06, "loss": 1.0585, "step": 10309 }, { "epoch": 1.0970419238135773, "grad_norm": 12.4453125, "learning_rate": 9.651435474701536e-06, "loss": 0.9279, "step": 10310 }, { "epoch": 1.097148329431794, "grad_norm": 15.4375, "learning_rate": 9.651399943149517e-06, "loss": 0.9047, "step": 10311 }, { "epoch": 1.0972547350500106, "grad_norm": 16.1875, "learning_rate": 9.651364411597499e-06, "loss": 1.0604, "step": 10312 }, { "epoch": 1.0973611406682273, "grad_norm": 12.125, "learning_rate": 9.65132888004548e-06, "loss": 1.1144, "step": 10313 }, { "epoch": 1.0974675462864438, "grad_norm": 11.0625, "learning_rate": 9.651293348493464e-06, "loss": 1.0455, "step": 10314 }, { "epoch": 1.0975739519046606, "grad_norm": 11.5859375, "learning_rate": 9.651257816941445e-06, "loss": 1.1976, "step": 10315 }, { "epoch": 1.0976803575228773, "grad_norm": 9.8359375, "learning_rate": 9.651222285389427e-06, "loss": 0.9676, "step": 10316 }, { "epoch": 1.0977867631410938, "grad_norm": 13.5078125, "learning_rate": 9.651186753837409e-06, "loss": 1.101, "step": 10317 }, { "epoch": 1.0978931687593105, "grad_norm": 12.5078125, "learning_rate": 9.65115122228539e-06, "loss": 1.2462, "step": 10318 }, { "epoch": 1.097999574377527, "grad_norm": 8.3203125, "learning_rate": 9.651115690733372e-06, "loss": 0.8764, "step": 10319 }, { "epoch": 1.0981059799957438, "grad_norm": 14.2421875, "learning_rate": 9.651080159181354e-06, "loss": 0.9538, "step": 10320 }, { "epoch": 1.0982123856139605, "grad_norm": 11.421875, "learning_rate": 9.651044627629335e-06, "loss": 0.9042, "step": 10321 }, { "epoch": 1.098318791232177, "grad_norm": 13.8046875, "learning_rate": 9.651009096077317e-06, "loss": 1.2787, "step": 10322 }, { "epoch": 1.0984251968503937, "grad_norm": 10.953125, "learning_rate": 9.650973564525298e-06, "loss": 1.2844, "step": 10323 }, { "epoch": 1.0985316024686103, "grad_norm": 17.0625, "learning_rate": 9.650938032973282e-06, "loss": 1.2776, "step": 10324 }, { "epoch": 1.098638008086827, "grad_norm": 15.8046875, "learning_rate": 9.650902501421263e-06, "loss": 1.3301, "step": 10325 }, { "epoch": 1.0987444137050437, "grad_norm": 15.2890625, "learning_rate": 9.650866969869245e-06, "loss": 1.168, "step": 10326 }, { "epoch": 1.0988508193232602, "grad_norm": 16.03125, "learning_rate": 9.650831438317226e-06, "loss": 1.1921, "step": 10327 }, { "epoch": 1.098957224941477, "grad_norm": 13.6328125, "learning_rate": 9.650795906765208e-06, "loss": 1.1432, "step": 10328 }, { "epoch": 1.0990636305596935, "grad_norm": 9.828125, "learning_rate": 9.650760375213191e-06, "loss": 0.9842, "step": 10329 }, { "epoch": 1.0991700361779102, "grad_norm": 10.5625, "learning_rate": 9.650724843661173e-06, "loss": 0.9694, "step": 10330 }, { "epoch": 1.099276441796127, "grad_norm": 12.953125, "learning_rate": 9.650689312109155e-06, "loss": 1.0117, "step": 10331 }, { "epoch": 1.0993828474143434, "grad_norm": 10.9375, "learning_rate": 9.650653780557136e-06, "loss": 1.1001, "step": 10332 }, { "epoch": 1.0994892530325602, "grad_norm": 11.640625, "learning_rate": 9.650618249005118e-06, "loss": 0.7759, "step": 10333 }, { "epoch": 1.0995956586507767, "grad_norm": 12.5625, "learning_rate": 9.6505827174531e-06, "loss": 1.1223, "step": 10334 }, { "epoch": 1.0997020642689934, "grad_norm": 15.96875, "learning_rate": 9.650547185901081e-06, "loss": 1.3431, "step": 10335 }, { "epoch": 1.09980846988721, "grad_norm": 11.6875, "learning_rate": 9.650511654349063e-06, "loss": 1.1379, "step": 10336 }, { "epoch": 1.0999148755054267, "grad_norm": 16.0, "learning_rate": 9.650476122797044e-06, "loss": 1.2512, "step": 10337 }, { "epoch": 1.1000212811236434, "grad_norm": 19.140625, "learning_rate": 9.650440591245026e-06, "loss": 1.2022, "step": 10338 }, { "epoch": 1.10012768674186, "grad_norm": 13.640625, "learning_rate": 9.650405059693007e-06, "loss": 1.2393, "step": 10339 }, { "epoch": 1.1002340923600766, "grad_norm": 12.0859375, "learning_rate": 9.65036952814099e-06, "loss": 1.0442, "step": 10340 }, { "epoch": 1.1003404979782934, "grad_norm": 12.8125, "learning_rate": 9.650333996588972e-06, "loss": 1.3174, "step": 10341 }, { "epoch": 1.1004469035965099, "grad_norm": 10.875, "learning_rate": 9.650298465036954e-06, "loss": 0.8084, "step": 10342 }, { "epoch": 1.1005533092147266, "grad_norm": 13.671875, "learning_rate": 9.650262933484936e-06, "loss": 1.3178, "step": 10343 }, { "epoch": 1.1006597148329431, "grad_norm": 9.828125, "learning_rate": 9.650227401932917e-06, "loss": 1.1718, "step": 10344 }, { "epoch": 1.1007661204511598, "grad_norm": 13.84375, "learning_rate": 9.650191870380899e-06, "loss": 0.9883, "step": 10345 }, { "epoch": 1.1008725260693764, "grad_norm": 15.6875, "learning_rate": 9.65015633882888e-06, "loss": 1.2111, "step": 10346 }, { "epoch": 1.100978931687593, "grad_norm": 13.3984375, "learning_rate": 9.650120807276862e-06, "loss": 0.992, "step": 10347 }, { "epoch": 1.1010853373058098, "grad_norm": 10.8359375, "learning_rate": 9.650085275724844e-06, "loss": 1.1629, "step": 10348 }, { "epoch": 1.1011917429240263, "grad_norm": 12.359375, "learning_rate": 9.650049744172825e-06, "loss": 1.125, "step": 10349 }, { "epoch": 1.101298148542243, "grad_norm": 12.46875, "learning_rate": 9.650014212620808e-06, "loss": 1.1277, "step": 10350 }, { "epoch": 1.1014045541604596, "grad_norm": 15.2421875, "learning_rate": 9.64997868106879e-06, "loss": 1.0296, "step": 10351 }, { "epoch": 1.1015109597786763, "grad_norm": 10.3984375, "learning_rate": 9.649943149516772e-06, "loss": 1.0595, "step": 10352 }, { "epoch": 1.101617365396893, "grad_norm": 11.3828125, "learning_rate": 9.649907617964753e-06, "loss": 1.2398, "step": 10353 }, { "epoch": 1.1017237710151095, "grad_norm": 17.671875, "learning_rate": 9.649872086412735e-06, "loss": 1.1541, "step": 10354 }, { "epoch": 1.1018301766333263, "grad_norm": 13.0234375, "learning_rate": 9.649836554860718e-06, "loss": 1.0552, "step": 10355 }, { "epoch": 1.1019365822515428, "grad_norm": 12.0859375, "learning_rate": 9.6498010233087e-06, "loss": 0.8545, "step": 10356 }, { "epoch": 1.1020429878697595, "grad_norm": 12.96875, "learning_rate": 9.649765491756681e-06, "loss": 0.9901, "step": 10357 }, { "epoch": 1.1021493934879762, "grad_norm": 15.859375, "learning_rate": 9.649729960204663e-06, "loss": 0.9437, "step": 10358 }, { "epoch": 1.1022557991061928, "grad_norm": 12.8671875, "learning_rate": 9.649694428652645e-06, "loss": 0.953, "step": 10359 }, { "epoch": 1.1023622047244095, "grad_norm": 11.53125, "learning_rate": 9.649658897100626e-06, "loss": 1.3773, "step": 10360 }, { "epoch": 1.102468610342626, "grad_norm": 12.7578125, "learning_rate": 9.649623365548608e-06, "loss": 1.0337, "step": 10361 }, { "epoch": 1.1025750159608427, "grad_norm": 12.15625, "learning_rate": 9.64958783399659e-06, "loss": 1.0301, "step": 10362 }, { "epoch": 1.1026814215790595, "grad_norm": 20.375, "learning_rate": 9.649552302444571e-06, "loss": 1.0523, "step": 10363 }, { "epoch": 1.102787827197276, "grad_norm": 12.015625, "learning_rate": 9.649516770892553e-06, "loss": 1.1023, "step": 10364 }, { "epoch": 1.1028942328154927, "grad_norm": 13.03125, "learning_rate": 9.649481239340536e-06, "loss": 1.117, "step": 10365 }, { "epoch": 1.1030006384337092, "grad_norm": 16.78125, "learning_rate": 9.649445707788518e-06, "loss": 0.8186, "step": 10366 }, { "epoch": 1.103107044051926, "grad_norm": 12.8828125, "learning_rate": 9.6494101762365e-06, "loss": 1.1123, "step": 10367 }, { "epoch": 1.1032134496701427, "grad_norm": 11.265625, "learning_rate": 9.64937464468448e-06, "loss": 1.0979, "step": 10368 }, { "epoch": 1.1033198552883592, "grad_norm": 10.21875, "learning_rate": 9.649339113132462e-06, "loss": 0.9607, "step": 10369 }, { "epoch": 1.103426260906576, "grad_norm": 22.828125, "learning_rate": 9.649303581580444e-06, "loss": 1.1836, "step": 10370 }, { "epoch": 1.1035326665247924, "grad_norm": 13.5625, "learning_rate": 9.649268050028426e-06, "loss": 1.1941, "step": 10371 }, { "epoch": 1.1036390721430092, "grad_norm": 13.09375, "learning_rate": 9.649232518476407e-06, "loss": 1.0666, "step": 10372 }, { "epoch": 1.103745477761226, "grad_norm": 15.78125, "learning_rate": 9.649196986924389e-06, "loss": 1.2663, "step": 10373 }, { "epoch": 1.1038518833794424, "grad_norm": 16.96875, "learning_rate": 9.64916145537237e-06, "loss": 0.9149, "step": 10374 }, { "epoch": 1.1039582889976591, "grad_norm": 16.65625, "learning_rate": 9.649125923820352e-06, "loss": 1.2014, "step": 10375 }, { "epoch": 1.1040646946158756, "grad_norm": 14.734375, "learning_rate": 9.649090392268335e-06, "loss": 0.8859, "step": 10376 }, { "epoch": 1.1041711002340924, "grad_norm": 10.734375, "learning_rate": 9.649054860716317e-06, "loss": 0.9969, "step": 10377 }, { "epoch": 1.104277505852309, "grad_norm": 13.53125, "learning_rate": 9.649019329164299e-06, "loss": 1.1874, "step": 10378 }, { "epoch": 1.1043839114705256, "grad_norm": 13.9453125, "learning_rate": 9.64898379761228e-06, "loss": 1.4779, "step": 10379 }, { "epoch": 1.1044903170887423, "grad_norm": 11.78125, "learning_rate": 9.648948266060262e-06, "loss": 1.0915, "step": 10380 }, { "epoch": 1.1045967227069589, "grad_norm": 10.140625, "learning_rate": 9.648912734508245e-06, "loss": 1.1035, "step": 10381 }, { "epoch": 1.1047031283251756, "grad_norm": 11.359375, "learning_rate": 9.648877202956227e-06, "loss": 1.0353, "step": 10382 }, { "epoch": 1.104809533943392, "grad_norm": 22.703125, "learning_rate": 9.648841671404208e-06, "loss": 1.1383, "step": 10383 }, { "epoch": 1.1049159395616088, "grad_norm": 29.75, "learning_rate": 9.64880613985219e-06, "loss": 1.2752, "step": 10384 }, { "epoch": 1.1050223451798256, "grad_norm": 14.9609375, "learning_rate": 9.648770608300171e-06, "loss": 1.0924, "step": 10385 }, { "epoch": 1.105128750798042, "grad_norm": 11.1015625, "learning_rate": 9.648735076748153e-06, "loss": 0.9117, "step": 10386 }, { "epoch": 1.1052351564162588, "grad_norm": 12.5703125, "learning_rate": 9.648699545196135e-06, "loss": 1.1092, "step": 10387 }, { "epoch": 1.1053415620344755, "grad_norm": 12.3125, "learning_rate": 9.648664013644116e-06, "loss": 1.3611, "step": 10388 }, { "epoch": 1.105447967652692, "grad_norm": 19.1875, "learning_rate": 9.648628482092098e-06, "loss": 1.316, "step": 10389 }, { "epoch": 1.1055543732709088, "grad_norm": 9.328125, "learning_rate": 9.64859295054008e-06, "loss": 0.9763, "step": 10390 }, { "epoch": 1.1056607788891253, "grad_norm": 12.703125, "learning_rate": 9.648557418988063e-06, "loss": 0.9244, "step": 10391 }, { "epoch": 1.105767184507342, "grad_norm": 11.3828125, "learning_rate": 9.648521887436044e-06, "loss": 1.0058, "step": 10392 }, { "epoch": 1.1058735901255585, "grad_norm": 15.5078125, "learning_rate": 9.648486355884026e-06, "loss": 1.1605, "step": 10393 }, { "epoch": 1.1059799957437753, "grad_norm": 13.46875, "learning_rate": 9.648450824332008e-06, "loss": 1.1911, "step": 10394 }, { "epoch": 1.106086401361992, "grad_norm": 14.8828125, "learning_rate": 9.64841529277999e-06, "loss": 0.8497, "step": 10395 }, { "epoch": 1.1061928069802085, "grad_norm": 13.140625, "learning_rate": 9.648379761227973e-06, "loss": 0.916, "step": 10396 }, { "epoch": 1.1062992125984252, "grad_norm": 14.2109375, "learning_rate": 9.648344229675952e-06, "loss": 1.0836, "step": 10397 }, { "epoch": 1.1064056182166417, "grad_norm": 13.046875, "learning_rate": 9.648308698123934e-06, "loss": 1.1154, "step": 10398 }, { "epoch": 1.1065120238348585, "grad_norm": 12.7421875, "learning_rate": 9.648273166571916e-06, "loss": 0.925, "step": 10399 }, { "epoch": 1.1066184294530752, "grad_norm": 15.6875, "learning_rate": 9.648237635019897e-06, "loss": 1.0091, "step": 10400 }, { "epoch": 1.1067248350712917, "grad_norm": 9.0703125, "learning_rate": 9.64820210346788e-06, "loss": 0.8262, "step": 10401 }, { "epoch": 1.1068312406895084, "grad_norm": 16.265625, "learning_rate": 9.648166571915862e-06, "loss": 1.2459, "step": 10402 }, { "epoch": 1.106937646307725, "grad_norm": 12.390625, "learning_rate": 9.648131040363844e-06, "loss": 0.9735, "step": 10403 }, { "epoch": 1.1070440519259417, "grad_norm": 12.9609375, "learning_rate": 9.648095508811825e-06, "loss": 0.9752, "step": 10404 }, { "epoch": 1.1071504575441584, "grad_norm": 17.578125, "learning_rate": 9.648059977259807e-06, "loss": 1.1667, "step": 10405 }, { "epoch": 1.107256863162375, "grad_norm": 13.7265625, "learning_rate": 9.64802444570779e-06, "loss": 1.3694, "step": 10406 }, { "epoch": 1.1073632687805917, "grad_norm": 10.6484375, "learning_rate": 9.647988914155772e-06, "loss": 1.0029, "step": 10407 }, { "epoch": 1.1074696743988082, "grad_norm": 11.0703125, "learning_rate": 9.647953382603754e-06, "loss": 0.9919, "step": 10408 }, { "epoch": 1.107576080017025, "grad_norm": 15.4140625, "learning_rate": 9.647917851051735e-06, "loss": 1.1792, "step": 10409 }, { "epoch": 1.1076824856352416, "grad_norm": 15.3359375, "learning_rate": 9.647882319499717e-06, "loss": 1.4064, "step": 10410 }, { "epoch": 1.1077888912534581, "grad_norm": 12.2265625, "learning_rate": 9.647846787947698e-06, "loss": 1.0829, "step": 10411 }, { "epoch": 1.1078952968716749, "grad_norm": 10.6015625, "learning_rate": 9.64781125639568e-06, "loss": 1.2251, "step": 10412 }, { "epoch": 1.1080017024898914, "grad_norm": 9.0, "learning_rate": 9.647775724843662e-06, "loss": 0.9583, "step": 10413 }, { "epoch": 1.1081081081081081, "grad_norm": 12.71875, "learning_rate": 9.647740193291643e-06, "loss": 1.1547, "step": 10414 }, { "epoch": 1.1082145137263248, "grad_norm": 12.2890625, "learning_rate": 9.647704661739625e-06, "loss": 0.8676, "step": 10415 }, { "epoch": 1.1083209193445414, "grad_norm": 12.2578125, "learning_rate": 9.647669130187606e-06, "loss": 1.1446, "step": 10416 }, { "epoch": 1.108427324962758, "grad_norm": 13.4140625, "learning_rate": 9.64763359863559e-06, "loss": 1.1857, "step": 10417 }, { "epoch": 1.1085337305809746, "grad_norm": 13.109375, "learning_rate": 9.647598067083571e-06, "loss": 1.1526, "step": 10418 }, { "epoch": 1.1086401361991913, "grad_norm": 16.65625, "learning_rate": 9.647562535531553e-06, "loss": 1.2717, "step": 10419 }, { "epoch": 1.108746541817408, "grad_norm": 13.015625, "learning_rate": 9.647527003979534e-06, "loss": 1.1235, "step": 10420 }, { "epoch": 1.1088529474356246, "grad_norm": 15.5546875, "learning_rate": 9.647491472427516e-06, "loss": 1.2284, "step": 10421 }, { "epoch": 1.1089593530538413, "grad_norm": 13.5859375, "learning_rate": 9.6474559408755e-06, "loss": 1.0361, "step": 10422 }, { "epoch": 1.1090657586720578, "grad_norm": 24.953125, "learning_rate": 9.647420409323481e-06, "loss": 1.0789, "step": 10423 }, { "epoch": 1.1091721642902745, "grad_norm": 13.15625, "learning_rate": 9.647384877771463e-06, "loss": 0.9003, "step": 10424 }, { "epoch": 1.1092785699084913, "grad_norm": 11.2421875, "learning_rate": 9.647349346219444e-06, "loss": 1.0783, "step": 10425 }, { "epoch": 1.1093849755267078, "grad_norm": 15.0078125, "learning_rate": 9.647313814667426e-06, "loss": 0.8967, "step": 10426 }, { "epoch": 1.1094913811449245, "grad_norm": 13.8125, "learning_rate": 9.647278283115407e-06, "loss": 1.0088, "step": 10427 }, { "epoch": 1.109597786763141, "grad_norm": 12.7265625, "learning_rate": 9.647242751563389e-06, "loss": 1.3673, "step": 10428 }, { "epoch": 1.1097041923813578, "grad_norm": 12.9453125, "learning_rate": 9.64720722001137e-06, "loss": 1.4149, "step": 10429 }, { "epoch": 1.1098105979995743, "grad_norm": 14.8828125, "learning_rate": 9.647171688459352e-06, "loss": 1.159, "step": 10430 }, { "epoch": 1.109917003617791, "grad_norm": 12.4140625, "learning_rate": 9.647136156907334e-06, "loss": 1.2635, "step": 10431 }, { "epoch": 1.1100234092360077, "grad_norm": 10.9453125, "learning_rate": 9.647100625355317e-06, "loss": 0.9696, "step": 10432 }, { "epoch": 1.1101298148542242, "grad_norm": 13.2109375, "learning_rate": 9.647065093803299e-06, "loss": 1.2814, "step": 10433 }, { "epoch": 1.110236220472441, "grad_norm": 12.2109375, "learning_rate": 9.64702956225128e-06, "loss": 1.2246, "step": 10434 }, { "epoch": 1.1103426260906577, "grad_norm": 14.96875, "learning_rate": 9.646994030699262e-06, "loss": 0.8867, "step": 10435 }, { "epoch": 1.1104490317088742, "grad_norm": 10.6796875, "learning_rate": 9.646958499147244e-06, "loss": 1.0406, "step": 10436 }, { "epoch": 1.110555437327091, "grad_norm": 11.140625, "learning_rate": 9.646922967595225e-06, "loss": 1.2444, "step": 10437 }, { "epoch": 1.1106618429453075, "grad_norm": 13.5546875, "learning_rate": 9.646887436043207e-06, "loss": 1.1582, "step": 10438 }, { "epoch": 1.1107682485635242, "grad_norm": 10.890625, "learning_rate": 9.646851904491188e-06, "loss": 0.9805, "step": 10439 }, { "epoch": 1.1108746541817407, "grad_norm": 11.3515625, "learning_rate": 9.64681637293917e-06, "loss": 1.0802, "step": 10440 }, { "epoch": 1.1109810597999574, "grad_norm": 11.0078125, "learning_rate": 9.646780841387152e-06, "loss": 1.3458, "step": 10441 }, { "epoch": 1.1110874654181742, "grad_norm": 12.28125, "learning_rate": 9.646745309835135e-06, "loss": 1.4926, "step": 10442 }, { "epoch": 1.1111938710363907, "grad_norm": 11.8125, "learning_rate": 9.646709778283117e-06, "loss": 0.9862, "step": 10443 }, { "epoch": 1.1113002766546074, "grad_norm": 10.96875, "learning_rate": 9.646674246731098e-06, "loss": 0.8489, "step": 10444 }, { "epoch": 1.111406682272824, "grad_norm": 14.71875, "learning_rate": 9.64663871517908e-06, "loss": 1.1292, "step": 10445 }, { "epoch": 1.1115130878910406, "grad_norm": 12.375, "learning_rate": 9.646603183627061e-06, "loss": 1.0138, "step": 10446 }, { "epoch": 1.1116194935092574, "grad_norm": 12.046875, "learning_rate": 9.646567652075045e-06, "loss": 1.2067, "step": 10447 }, { "epoch": 1.1117258991274739, "grad_norm": 13.609375, "learning_rate": 9.646532120523026e-06, "loss": 0.9872, "step": 10448 }, { "epoch": 1.1118323047456906, "grad_norm": 12.2734375, "learning_rate": 9.646496588971008e-06, "loss": 1.3284, "step": 10449 }, { "epoch": 1.1119387103639071, "grad_norm": 10.4296875, "learning_rate": 9.64646105741899e-06, "loss": 1.0893, "step": 10450 }, { "epoch": 1.1120451159821239, "grad_norm": 12.640625, "learning_rate": 9.646425525866971e-06, "loss": 1.124, "step": 10451 }, { "epoch": 1.1121515216003406, "grad_norm": 12.1328125, "learning_rate": 9.646389994314953e-06, "loss": 1.0769, "step": 10452 }, { "epoch": 1.112257927218557, "grad_norm": 13.0390625, "learning_rate": 9.646354462762934e-06, "loss": 1.2024, "step": 10453 }, { "epoch": 1.1123643328367738, "grad_norm": 14.359375, "learning_rate": 9.646318931210916e-06, "loss": 1.2943, "step": 10454 }, { "epoch": 1.1124707384549903, "grad_norm": 13.796875, "learning_rate": 9.646283399658898e-06, "loss": 1.3564, "step": 10455 }, { "epoch": 1.112577144073207, "grad_norm": 14.8203125, "learning_rate": 9.646247868106879e-06, "loss": 1.1514, "step": 10456 }, { "epoch": 1.1126835496914238, "grad_norm": 13.0625, "learning_rate": 9.64621233655486e-06, "loss": 1.2576, "step": 10457 }, { "epoch": 1.1127899553096403, "grad_norm": 15.7109375, "learning_rate": 9.646176805002844e-06, "loss": 1.2549, "step": 10458 }, { "epoch": 1.112896360927857, "grad_norm": 17.90625, "learning_rate": 9.646141273450826e-06, "loss": 0.9978, "step": 10459 }, { "epoch": 1.1130027665460736, "grad_norm": 13.328125, "learning_rate": 9.646105741898807e-06, "loss": 1.2772, "step": 10460 }, { "epoch": 1.1131091721642903, "grad_norm": 12.703125, "learning_rate": 9.646070210346789e-06, "loss": 1.1483, "step": 10461 }, { "epoch": 1.113215577782507, "grad_norm": 12.96875, "learning_rate": 9.64603467879477e-06, "loss": 1.2558, "step": 10462 }, { "epoch": 1.1133219834007235, "grad_norm": 15.828125, "learning_rate": 9.645999147242752e-06, "loss": 1.2227, "step": 10463 }, { "epoch": 1.1134283890189403, "grad_norm": 8.4765625, "learning_rate": 9.645963615690734e-06, "loss": 0.9211, "step": 10464 }, { "epoch": 1.1135347946371568, "grad_norm": 12.2890625, "learning_rate": 9.645928084138715e-06, "loss": 1.1943, "step": 10465 }, { "epoch": 1.1136412002553735, "grad_norm": 10.2890625, "learning_rate": 9.645892552586697e-06, "loss": 1.0304, "step": 10466 }, { "epoch": 1.1137476058735902, "grad_norm": 12.015625, "learning_rate": 9.645857021034678e-06, "loss": 1.199, "step": 10467 }, { "epoch": 1.1138540114918067, "grad_norm": 19.171875, "learning_rate": 9.645821489482662e-06, "loss": 1.19, "step": 10468 }, { "epoch": 1.1139604171100235, "grad_norm": 17.359375, "learning_rate": 9.645785957930643e-06, "loss": 0.9501, "step": 10469 }, { "epoch": 1.11406682272824, "grad_norm": 12.7734375, "learning_rate": 9.645750426378625e-06, "loss": 1.3423, "step": 10470 }, { "epoch": 1.1141732283464567, "grad_norm": 9.9765625, "learning_rate": 9.645714894826607e-06, "loss": 1.0719, "step": 10471 }, { "epoch": 1.1142796339646734, "grad_norm": 14.1484375, "learning_rate": 9.645679363274588e-06, "loss": 1.272, "step": 10472 }, { "epoch": 1.11438603958289, "grad_norm": 17.484375, "learning_rate": 9.645643831722572e-06, "loss": 1.082, "step": 10473 }, { "epoch": 1.1144924452011067, "grad_norm": 13.671875, "learning_rate": 9.645608300170553e-06, "loss": 1.221, "step": 10474 }, { "epoch": 1.1145988508193232, "grad_norm": 15.6796875, "learning_rate": 9.645572768618535e-06, "loss": 1.2144, "step": 10475 }, { "epoch": 1.11470525643754, "grad_norm": 9.7109375, "learning_rate": 9.645537237066516e-06, "loss": 1.1217, "step": 10476 }, { "epoch": 1.1148116620557564, "grad_norm": 13.59375, "learning_rate": 9.645501705514498e-06, "loss": 1.063, "step": 10477 }, { "epoch": 1.1149180676739732, "grad_norm": 14.7734375, "learning_rate": 9.64546617396248e-06, "loss": 1.2202, "step": 10478 }, { "epoch": 1.11502447329219, "grad_norm": 8.6640625, "learning_rate": 9.645430642410461e-06, "loss": 0.9766, "step": 10479 }, { "epoch": 1.1151308789104064, "grad_norm": 12.7890625, "learning_rate": 9.645395110858443e-06, "loss": 1.3264, "step": 10480 }, { "epoch": 1.1152372845286231, "grad_norm": 12.484375, "learning_rate": 9.645359579306424e-06, "loss": 1.0882, "step": 10481 }, { "epoch": 1.1153436901468397, "grad_norm": 10.203125, "learning_rate": 9.645324047754406e-06, "loss": 0.9758, "step": 10482 }, { "epoch": 1.1154500957650564, "grad_norm": 11.28125, "learning_rate": 9.64528851620239e-06, "loss": 0.967, "step": 10483 }, { "epoch": 1.1155565013832731, "grad_norm": 10.65625, "learning_rate": 9.645252984650371e-06, "loss": 1.2599, "step": 10484 }, { "epoch": 1.1156629070014896, "grad_norm": 12.0, "learning_rate": 9.645217453098352e-06, "loss": 1.1489, "step": 10485 }, { "epoch": 1.1157693126197064, "grad_norm": 9.84375, "learning_rate": 9.645181921546334e-06, "loss": 1.1429, "step": 10486 }, { "epoch": 1.1158757182379229, "grad_norm": 10.609375, "learning_rate": 9.645146389994316e-06, "loss": 0.9348, "step": 10487 }, { "epoch": 1.1159821238561396, "grad_norm": 10.890625, "learning_rate": 9.645110858442297e-06, "loss": 0.8872, "step": 10488 }, { "epoch": 1.1160885294743563, "grad_norm": 15.5859375, "learning_rate": 9.645075326890279e-06, "loss": 1.0846, "step": 10489 }, { "epoch": 1.1161949350925728, "grad_norm": 14.5859375, "learning_rate": 9.64503979533826e-06, "loss": 0.9766, "step": 10490 }, { "epoch": 1.1163013407107896, "grad_norm": 13.15625, "learning_rate": 9.645004263786242e-06, "loss": 1.3476, "step": 10491 }, { "epoch": 1.116407746329006, "grad_norm": 11.2421875, "learning_rate": 9.644968732234224e-06, "loss": 0.991, "step": 10492 }, { "epoch": 1.1165141519472228, "grad_norm": 11.703125, "learning_rate": 9.644933200682205e-06, "loss": 1.1898, "step": 10493 }, { "epoch": 1.1166205575654395, "grad_norm": 13.2109375, "learning_rate": 9.644897669130189e-06, "loss": 0.9267, "step": 10494 }, { "epoch": 1.116726963183656, "grad_norm": 13.6484375, "learning_rate": 9.64486213757817e-06, "loss": 1.1186, "step": 10495 }, { "epoch": 1.1168333688018728, "grad_norm": 13.484375, "learning_rate": 9.644826606026152e-06, "loss": 1.0478, "step": 10496 }, { "epoch": 1.1169397744200893, "grad_norm": 13.46875, "learning_rate": 9.644791074474133e-06, "loss": 1.0432, "step": 10497 }, { "epoch": 1.117046180038306, "grad_norm": 14.65625, "learning_rate": 9.644755542922115e-06, "loss": 0.9704, "step": 10498 }, { "epoch": 1.1171525856565228, "grad_norm": 14.0234375, "learning_rate": 9.644720011370098e-06, "loss": 1.3179, "step": 10499 }, { "epoch": 1.1172589912747393, "grad_norm": 12.4140625, "learning_rate": 9.64468447981808e-06, "loss": 1.3833, "step": 10500 }, { "epoch": 1.117365396892956, "grad_norm": 12.953125, "learning_rate": 9.644648948266062e-06, "loss": 1.0678, "step": 10501 }, { "epoch": 1.1174718025111725, "grad_norm": 10.953125, "learning_rate": 9.644613416714043e-06, "loss": 0.9612, "step": 10502 }, { "epoch": 1.1175782081293892, "grad_norm": 14.75, "learning_rate": 9.644577885162025e-06, "loss": 1.232, "step": 10503 }, { "epoch": 1.117684613747606, "grad_norm": 15.4609375, "learning_rate": 9.644542353610006e-06, "loss": 1.0346, "step": 10504 }, { "epoch": 1.1177910193658225, "grad_norm": 13.4375, "learning_rate": 9.644506822057988e-06, "loss": 1.0899, "step": 10505 }, { "epoch": 1.1178974249840392, "grad_norm": 16.609375, "learning_rate": 9.64447129050597e-06, "loss": 1.1405, "step": 10506 }, { "epoch": 1.1180038306022557, "grad_norm": 13.6328125, "learning_rate": 9.644435758953951e-06, "loss": 1.218, "step": 10507 }, { "epoch": 1.1181102362204725, "grad_norm": 18.03125, "learning_rate": 9.644400227401933e-06, "loss": 1.0724, "step": 10508 }, { "epoch": 1.1182166418386892, "grad_norm": 12.15625, "learning_rate": 9.644364695849916e-06, "loss": 1.0108, "step": 10509 }, { "epoch": 1.1183230474569057, "grad_norm": 10.9765625, "learning_rate": 9.644329164297898e-06, "loss": 1.2894, "step": 10510 }, { "epoch": 1.1184294530751224, "grad_norm": 13.640625, "learning_rate": 9.64429363274588e-06, "loss": 0.906, "step": 10511 }, { "epoch": 1.118535858693339, "grad_norm": 12.2578125, "learning_rate": 9.644258101193861e-06, "loss": 1.1696, "step": 10512 }, { "epoch": 1.1186422643115557, "grad_norm": 13.2265625, "learning_rate": 9.644222569641843e-06, "loss": 1.2394, "step": 10513 }, { "epoch": 1.1187486699297722, "grad_norm": 14.421875, "learning_rate": 9.644187038089826e-06, "loss": 1.194, "step": 10514 }, { "epoch": 1.118855075547989, "grad_norm": 16.953125, "learning_rate": 9.644151506537807e-06, "loss": 0.8727, "step": 10515 }, { "epoch": 1.1189614811662056, "grad_norm": 13.3203125, "learning_rate": 9.644115974985789e-06, "loss": 1.2205, "step": 10516 }, { "epoch": 1.1190678867844222, "grad_norm": 13.8515625, "learning_rate": 9.644080443433769e-06, "loss": 1.2196, "step": 10517 }, { "epoch": 1.1191742924026389, "grad_norm": 12.1328125, "learning_rate": 9.64404491188175e-06, "loss": 0.9906, "step": 10518 }, { "epoch": 1.1192806980208556, "grad_norm": 11.46875, "learning_rate": 9.644009380329734e-06, "loss": 1.0997, "step": 10519 }, { "epoch": 1.1193871036390721, "grad_norm": 11.609375, "learning_rate": 9.643973848777715e-06, "loss": 1.2355, "step": 10520 }, { "epoch": 1.1194935092572889, "grad_norm": 12.2890625, "learning_rate": 9.643938317225697e-06, "loss": 1.2198, "step": 10521 }, { "epoch": 1.1195999148755054, "grad_norm": 13.21875, "learning_rate": 9.643902785673679e-06, "loss": 0.864, "step": 10522 }, { "epoch": 1.119706320493722, "grad_norm": 11.5859375, "learning_rate": 9.64386725412166e-06, "loss": 1.427, "step": 10523 }, { "epoch": 1.1198127261119386, "grad_norm": 12.546875, "learning_rate": 9.643831722569644e-06, "loss": 1.225, "step": 10524 }, { "epoch": 1.1199191317301553, "grad_norm": 12.2109375, "learning_rate": 9.643796191017625e-06, "loss": 1.0512, "step": 10525 }, { "epoch": 1.120025537348372, "grad_norm": 13.1484375, "learning_rate": 9.643760659465607e-06, "loss": 1.0474, "step": 10526 }, { "epoch": 1.1201319429665886, "grad_norm": 10.4921875, "learning_rate": 9.643725127913588e-06, "loss": 1.0742, "step": 10527 }, { "epoch": 1.1202383485848053, "grad_norm": 11.8046875, "learning_rate": 9.64368959636157e-06, "loss": 1.2419, "step": 10528 }, { "epoch": 1.1203447542030218, "grad_norm": 10.453125, "learning_rate": 9.643654064809552e-06, "loss": 0.8781, "step": 10529 }, { "epoch": 1.1204511598212386, "grad_norm": 17.046875, "learning_rate": 9.643618533257533e-06, "loss": 0.7454, "step": 10530 }, { "epoch": 1.1205575654394553, "grad_norm": 9.5546875, "learning_rate": 9.643583001705515e-06, "loss": 0.831, "step": 10531 }, { "epoch": 1.1206639710576718, "grad_norm": 16.15625, "learning_rate": 9.643547470153496e-06, "loss": 0.9949, "step": 10532 }, { "epoch": 1.1207703766758885, "grad_norm": 11.46875, "learning_rate": 9.643511938601478e-06, "loss": 0.9205, "step": 10533 }, { "epoch": 1.120876782294105, "grad_norm": 9.4375, "learning_rate": 9.64347640704946e-06, "loss": 1.051, "step": 10534 }, { "epoch": 1.1209831879123218, "grad_norm": 9.1953125, "learning_rate": 9.643440875497443e-06, "loss": 0.8944, "step": 10535 }, { "epoch": 1.1210895935305385, "grad_norm": 13.0859375, "learning_rate": 9.643405343945425e-06, "loss": 1.1605, "step": 10536 }, { "epoch": 1.121195999148755, "grad_norm": 13.25, "learning_rate": 9.643369812393406e-06, "loss": 1.2026, "step": 10537 }, { "epoch": 1.1213024047669717, "grad_norm": 10.5546875, "learning_rate": 9.643334280841388e-06, "loss": 1.0763, "step": 10538 }, { "epoch": 1.1214088103851882, "grad_norm": 10.8203125, "learning_rate": 9.64329874928937e-06, "loss": 1.2007, "step": 10539 }, { "epoch": 1.121515216003405, "grad_norm": 13.9375, "learning_rate": 9.643263217737353e-06, "loss": 0.963, "step": 10540 }, { "epoch": 1.1216216216216217, "grad_norm": 11.7578125, "learning_rate": 9.643227686185334e-06, "loss": 0.9541, "step": 10541 }, { "epoch": 1.1217280272398382, "grad_norm": 10.6328125, "learning_rate": 9.643192154633316e-06, "loss": 1.0424, "step": 10542 }, { "epoch": 1.121834432858055, "grad_norm": 8.2578125, "learning_rate": 9.643156623081298e-06, "loss": 0.7278, "step": 10543 }, { "epoch": 1.1219408384762715, "grad_norm": 24.46875, "learning_rate": 9.643121091529279e-06, "loss": 0.9538, "step": 10544 }, { "epoch": 1.1220472440944882, "grad_norm": 11.015625, "learning_rate": 9.64308555997726e-06, "loss": 1.0751, "step": 10545 }, { "epoch": 1.122153649712705, "grad_norm": 13.453125, "learning_rate": 9.643050028425242e-06, "loss": 1.1803, "step": 10546 }, { "epoch": 1.1222600553309214, "grad_norm": 11.296875, "learning_rate": 9.643014496873224e-06, "loss": 0.9517, "step": 10547 }, { "epoch": 1.1223664609491382, "grad_norm": 11.0625, "learning_rate": 9.642978965321206e-06, "loss": 0.9463, "step": 10548 }, { "epoch": 1.1224728665673547, "grad_norm": 15.8125, "learning_rate": 9.642943433769187e-06, "loss": 1.0584, "step": 10549 }, { "epoch": 1.1225792721855714, "grad_norm": 9.9609375, "learning_rate": 9.64290790221717e-06, "loss": 0.9906, "step": 10550 }, { "epoch": 1.1226856778037881, "grad_norm": 12.1328125, "learning_rate": 9.642872370665152e-06, "loss": 1.1179, "step": 10551 }, { "epoch": 1.1227920834220046, "grad_norm": 12.3203125, "learning_rate": 9.642836839113134e-06, "loss": 1.2213, "step": 10552 }, { "epoch": 1.1228984890402214, "grad_norm": 9.4921875, "learning_rate": 9.642801307561115e-06, "loss": 0.8383, "step": 10553 }, { "epoch": 1.123004894658438, "grad_norm": 15.046875, "learning_rate": 9.642765776009097e-06, "loss": 1.067, "step": 10554 }, { "epoch": 1.1231113002766546, "grad_norm": 12.4453125, "learning_rate": 9.642730244457078e-06, "loss": 1.079, "step": 10555 }, { "epoch": 1.1232177058948714, "grad_norm": 12.84375, "learning_rate": 9.64269471290506e-06, "loss": 0.9278, "step": 10556 }, { "epoch": 1.1233241115130879, "grad_norm": 17.703125, "learning_rate": 9.642659181353042e-06, "loss": 1.1534, "step": 10557 }, { "epoch": 1.1234305171313046, "grad_norm": 15.6953125, "learning_rate": 9.642623649801023e-06, "loss": 1.1188, "step": 10558 }, { "epoch": 1.123536922749521, "grad_norm": 14.859375, "learning_rate": 9.642588118249005e-06, "loss": 1.1053, "step": 10559 }, { "epoch": 1.1236433283677378, "grad_norm": 12.84375, "learning_rate": 9.642552586696988e-06, "loss": 1.0299, "step": 10560 }, { "epoch": 1.1237497339859543, "grad_norm": 13.1171875, "learning_rate": 9.64251705514497e-06, "loss": 0.984, "step": 10561 }, { "epoch": 1.123856139604171, "grad_norm": 16.796875, "learning_rate": 9.642481523592951e-06, "loss": 1.3236, "step": 10562 }, { "epoch": 1.1239625452223878, "grad_norm": 17.609375, "learning_rate": 9.642445992040933e-06, "loss": 1.2702, "step": 10563 }, { "epoch": 1.1240689508406043, "grad_norm": 11.6171875, "learning_rate": 9.642410460488915e-06, "loss": 1.2757, "step": 10564 }, { "epoch": 1.124175356458821, "grad_norm": 15.6640625, "learning_rate": 9.642374928936898e-06, "loss": 1.0217, "step": 10565 }, { "epoch": 1.1242817620770378, "grad_norm": 21.734375, "learning_rate": 9.64233939738488e-06, "loss": 1.4621, "step": 10566 }, { "epoch": 1.1243881676952543, "grad_norm": 11.1875, "learning_rate": 9.642303865832861e-06, "loss": 1.025, "step": 10567 }, { "epoch": 1.124494573313471, "grad_norm": 12.859375, "learning_rate": 9.642268334280843e-06, "loss": 0.9868, "step": 10568 }, { "epoch": 1.1246009789316875, "grad_norm": 12.375, "learning_rate": 9.642232802728824e-06, "loss": 1.0158, "step": 10569 }, { "epoch": 1.1247073845499043, "grad_norm": 10.171875, "learning_rate": 9.642197271176806e-06, "loss": 1.3341, "step": 10570 }, { "epoch": 1.1248137901681208, "grad_norm": 19.6875, "learning_rate": 9.642161739624788e-06, "loss": 1.2914, "step": 10571 }, { "epoch": 1.1249201957863375, "grad_norm": 14.1015625, "learning_rate": 9.64212620807277e-06, "loss": 1.4414, "step": 10572 }, { "epoch": 1.1250266014045542, "grad_norm": 17.53125, "learning_rate": 9.64209067652075e-06, "loss": 1.2591, "step": 10573 }, { "epoch": 1.1251330070227707, "grad_norm": 12.7109375, "learning_rate": 9.642055144968732e-06, "loss": 0.8317, "step": 10574 }, { "epoch": 1.1252394126409875, "grad_norm": 15.8984375, "learning_rate": 9.642019613416714e-06, "loss": 0.8838, "step": 10575 }, { "epoch": 1.125345818259204, "grad_norm": 21.75, "learning_rate": 9.641984081864697e-06, "loss": 1.4269, "step": 10576 }, { "epoch": 1.1254522238774207, "grad_norm": 10.6953125, "learning_rate": 9.641948550312679e-06, "loss": 0.9551, "step": 10577 }, { "epoch": 1.1255586294956375, "grad_norm": 14.2890625, "learning_rate": 9.64191301876066e-06, "loss": 0.9726, "step": 10578 }, { "epoch": 1.125665035113854, "grad_norm": 13.015625, "learning_rate": 9.641877487208642e-06, "loss": 0.9281, "step": 10579 }, { "epoch": 1.1257714407320707, "grad_norm": 14.75, "learning_rate": 9.641841955656624e-06, "loss": 0.8993, "step": 10580 }, { "epoch": 1.1258778463502872, "grad_norm": 13.7265625, "learning_rate": 9.641806424104605e-06, "loss": 1.2048, "step": 10581 }, { "epoch": 1.125984251968504, "grad_norm": 10.96875, "learning_rate": 9.641770892552587e-06, "loss": 0.731, "step": 10582 }, { "epoch": 1.1260906575867207, "grad_norm": 12.40625, "learning_rate": 9.641735361000569e-06, "loss": 0.934, "step": 10583 }, { "epoch": 1.1261970632049372, "grad_norm": 11.6484375, "learning_rate": 9.64169982944855e-06, "loss": 1.1865, "step": 10584 }, { "epoch": 1.126303468823154, "grad_norm": 10.46875, "learning_rate": 9.641664297896532e-06, "loss": 1.0118, "step": 10585 }, { "epoch": 1.1264098744413704, "grad_norm": 9.609375, "learning_rate": 9.641628766344515e-06, "loss": 0.9979, "step": 10586 }, { "epoch": 1.1265162800595871, "grad_norm": 13.53125, "learning_rate": 9.641593234792497e-06, "loss": 1.1798, "step": 10587 }, { "epoch": 1.1266226856778039, "grad_norm": 9.1953125, "learning_rate": 9.641557703240478e-06, "loss": 0.9559, "step": 10588 }, { "epoch": 1.1267290912960204, "grad_norm": 13.578125, "learning_rate": 9.64152217168846e-06, "loss": 1.0471, "step": 10589 }, { "epoch": 1.1268354969142371, "grad_norm": 8.5703125, "learning_rate": 9.641486640136442e-06, "loss": 1.062, "step": 10590 }, { "epoch": 1.1269419025324536, "grad_norm": 10.9609375, "learning_rate": 9.641451108584425e-06, "loss": 1.0976, "step": 10591 }, { "epoch": 1.1270483081506704, "grad_norm": 11.890625, "learning_rate": 9.641415577032406e-06, "loss": 1.0919, "step": 10592 }, { "epoch": 1.127154713768887, "grad_norm": 11.546875, "learning_rate": 9.641380045480388e-06, "loss": 0.9553, "step": 10593 }, { "epoch": 1.1272611193871036, "grad_norm": 17.078125, "learning_rate": 9.64134451392837e-06, "loss": 1.2034, "step": 10594 }, { "epoch": 1.1273675250053203, "grad_norm": 12.1796875, "learning_rate": 9.641308982376351e-06, "loss": 1.0958, "step": 10595 }, { "epoch": 1.1274739306235368, "grad_norm": 14.3203125, "learning_rate": 9.641273450824333e-06, "loss": 1.0408, "step": 10596 }, { "epoch": 1.1275803362417536, "grad_norm": 14.8515625, "learning_rate": 9.641237919272314e-06, "loss": 1.0495, "step": 10597 }, { "epoch": 1.12768674185997, "grad_norm": 16.34375, "learning_rate": 9.641202387720296e-06, "loss": 1.0977, "step": 10598 }, { "epoch": 1.1277931474781868, "grad_norm": 11.8671875, "learning_rate": 9.641166856168278e-06, "loss": 1.1947, "step": 10599 }, { "epoch": 1.1278995530964036, "grad_norm": 11.671875, "learning_rate": 9.64113132461626e-06, "loss": 1.1314, "step": 10600 }, { "epoch": 1.12800595871462, "grad_norm": 10.5859375, "learning_rate": 9.641095793064243e-06, "loss": 1.0957, "step": 10601 }, { "epoch": 1.1281123643328368, "grad_norm": 12.734375, "learning_rate": 9.641060261512224e-06, "loss": 1.1367, "step": 10602 }, { "epoch": 1.1282187699510535, "grad_norm": 12.109375, "learning_rate": 9.641024729960206e-06, "loss": 1.2277, "step": 10603 }, { "epoch": 1.12832517556927, "grad_norm": 13.6875, "learning_rate": 9.640989198408187e-06, "loss": 1.024, "step": 10604 }, { "epoch": 1.1284315811874868, "grad_norm": 13.1015625, "learning_rate": 9.640953666856169e-06, "loss": 1.0722, "step": 10605 }, { "epoch": 1.1285379868057033, "grad_norm": 8.25, "learning_rate": 9.64091813530415e-06, "loss": 0.873, "step": 10606 }, { "epoch": 1.12864439242392, "grad_norm": 12.2578125, "learning_rate": 9.640882603752132e-06, "loss": 1.3692, "step": 10607 }, { "epoch": 1.1287507980421365, "grad_norm": 11.890625, "learning_rate": 9.640847072200114e-06, "loss": 1.1312, "step": 10608 }, { "epoch": 1.1288572036603532, "grad_norm": 9.9296875, "learning_rate": 9.640811540648095e-06, "loss": 1.1764, "step": 10609 }, { "epoch": 1.12896360927857, "grad_norm": 12.90625, "learning_rate": 9.640776009096077e-06, "loss": 1.4183, "step": 10610 }, { "epoch": 1.1290700148967865, "grad_norm": 9.6484375, "learning_rate": 9.640740477544059e-06, "loss": 0.8504, "step": 10611 }, { "epoch": 1.1291764205150032, "grad_norm": 12.625, "learning_rate": 9.640704945992042e-06, "loss": 1.2133, "step": 10612 }, { "epoch": 1.12928282613322, "grad_norm": 11.265625, "learning_rate": 9.640669414440024e-06, "loss": 1.1978, "step": 10613 }, { "epoch": 1.1293892317514365, "grad_norm": 10.4765625, "learning_rate": 9.640633882888005e-06, "loss": 1.2174, "step": 10614 }, { "epoch": 1.1294956373696532, "grad_norm": 12.703125, "learning_rate": 9.640598351335987e-06, "loss": 0.9132, "step": 10615 }, { "epoch": 1.1296020429878697, "grad_norm": 10.859375, "learning_rate": 9.640562819783968e-06, "loss": 1.0198, "step": 10616 }, { "epoch": 1.1297084486060864, "grad_norm": 12.9375, "learning_rate": 9.640527288231952e-06, "loss": 1.3182, "step": 10617 }, { "epoch": 1.129814854224303, "grad_norm": 17.375, "learning_rate": 9.640491756679933e-06, "loss": 1.1234, "step": 10618 }, { "epoch": 1.1299212598425197, "grad_norm": 12.984375, "learning_rate": 9.640456225127915e-06, "loss": 1.1792, "step": 10619 }, { "epoch": 1.1300276654607364, "grad_norm": 12.46875, "learning_rate": 9.640420693575896e-06, "loss": 0.9824, "step": 10620 }, { "epoch": 1.130134071078953, "grad_norm": 13.6953125, "learning_rate": 9.640385162023878e-06, "loss": 1.3007, "step": 10621 }, { "epoch": 1.1302404766971696, "grad_norm": 13.03125, "learning_rate": 9.64034963047186e-06, "loss": 1.0659, "step": 10622 }, { "epoch": 1.1303468823153862, "grad_norm": 11.53125, "learning_rate": 9.640314098919841e-06, "loss": 1.1424, "step": 10623 }, { "epoch": 1.130453287933603, "grad_norm": 11.3671875, "learning_rate": 9.640278567367823e-06, "loss": 1.1383, "step": 10624 }, { "epoch": 1.1305596935518196, "grad_norm": 14.7421875, "learning_rate": 9.640243035815805e-06, "loss": 1.1487, "step": 10625 }, { "epoch": 1.1306660991700361, "grad_norm": 10.5859375, "learning_rate": 9.640207504263786e-06, "loss": 1.1226, "step": 10626 }, { "epoch": 1.1307725047882529, "grad_norm": 13.1953125, "learning_rate": 9.64017197271177e-06, "loss": 1.154, "step": 10627 }, { "epoch": 1.1308789104064694, "grad_norm": 11.828125, "learning_rate": 9.640136441159751e-06, "loss": 0.9492, "step": 10628 }, { "epoch": 1.130985316024686, "grad_norm": 14.90625, "learning_rate": 9.640100909607733e-06, "loss": 0.8019, "step": 10629 }, { "epoch": 1.1310917216429028, "grad_norm": 16.09375, "learning_rate": 9.640065378055714e-06, "loss": 0.8947, "step": 10630 }, { "epoch": 1.1311981272611193, "grad_norm": 10.171875, "learning_rate": 9.640029846503696e-06, "loss": 0.8697, "step": 10631 }, { "epoch": 1.131304532879336, "grad_norm": 9.1953125, "learning_rate": 9.63999431495168e-06, "loss": 1.0464, "step": 10632 }, { "epoch": 1.1314109384975526, "grad_norm": 17.34375, "learning_rate": 9.63995878339966e-06, "loss": 1.0042, "step": 10633 }, { "epoch": 1.1315173441157693, "grad_norm": 12.3203125, "learning_rate": 9.639923251847642e-06, "loss": 1.2321, "step": 10634 }, { "epoch": 1.131623749733986, "grad_norm": 13.265625, "learning_rate": 9.639887720295624e-06, "loss": 0.9884, "step": 10635 }, { "epoch": 1.1317301553522026, "grad_norm": 18.78125, "learning_rate": 9.639852188743606e-06, "loss": 1.1701, "step": 10636 }, { "epoch": 1.1318365609704193, "grad_norm": 11.0390625, "learning_rate": 9.639816657191587e-06, "loss": 1.0605, "step": 10637 }, { "epoch": 1.1319429665886358, "grad_norm": 11.609375, "learning_rate": 9.639781125639569e-06, "loss": 1.0563, "step": 10638 }, { "epoch": 1.1320493722068525, "grad_norm": 12.6328125, "learning_rate": 9.63974559408755e-06, "loss": 1.0845, "step": 10639 }, { "epoch": 1.1321557778250693, "grad_norm": 11.0390625, "learning_rate": 9.639710062535532e-06, "loss": 0.9368, "step": 10640 }, { "epoch": 1.1322621834432858, "grad_norm": 11.2109375, "learning_rate": 9.639674530983514e-06, "loss": 1.1242, "step": 10641 }, { "epoch": 1.1323685890615025, "grad_norm": 12.2890625, "learning_rate": 9.639638999431497e-06, "loss": 1.0586, "step": 10642 }, { "epoch": 1.132474994679719, "grad_norm": 13.5234375, "learning_rate": 9.639603467879479e-06, "loss": 1.1094, "step": 10643 }, { "epoch": 1.1325814002979357, "grad_norm": 13.515625, "learning_rate": 9.63956793632746e-06, "loss": 1.2546, "step": 10644 }, { "epoch": 1.1326878059161523, "grad_norm": 13.5390625, "learning_rate": 9.639532404775442e-06, "loss": 1.0041, "step": 10645 }, { "epoch": 1.132794211534369, "grad_norm": 11.203125, "learning_rate": 9.639496873223423e-06, "loss": 1.0893, "step": 10646 }, { "epoch": 1.1329006171525857, "grad_norm": 12.5234375, "learning_rate": 9.639461341671405e-06, "loss": 1.1002, "step": 10647 }, { "epoch": 1.1330070227708022, "grad_norm": 11.1015625, "learning_rate": 9.639425810119387e-06, "loss": 0.8966, "step": 10648 }, { "epoch": 1.133113428389019, "grad_norm": 11.3984375, "learning_rate": 9.639390278567368e-06, "loss": 1.1087, "step": 10649 }, { "epoch": 1.1332198340072357, "grad_norm": 10.921875, "learning_rate": 9.63935474701535e-06, "loss": 0.931, "step": 10650 }, { "epoch": 1.1333262396254522, "grad_norm": 11.6640625, "learning_rate": 9.639319215463331e-06, "loss": 1.1065, "step": 10651 }, { "epoch": 1.133432645243669, "grad_norm": 15.9921875, "learning_rate": 9.639283683911313e-06, "loss": 0.9855, "step": 10652 }, { "epoch": 1.1335390508618854, "grad_norm": 12.8828125, "learning_rate": 9.639248152359296e-06, "loss": 1.142, "step": 10653 }, { "epoch": 1.1336454564801022, "grad_norm": 11.8359375, "learning_rate": 9.639212620807278e-06, "loss": 0.8745, "step": 10654 }, { "epoch": 1.1337518620983187, "grad_norm": 12.765625, "learning_rate": 9.63917708925526e-06, "loss": 1.5124, "step": 10655 }, { "epoch": 1.1338582677165354, "grad_norm": 11.4296875, "learning_rate": 9.639141557703241e-06, "loss": 1.0263, "step": 10656 }, { "epoch": 1.1339646733347521, "grad_norm": 13.1875, "learning_rate": 9.639106026151223e-06, "loss": 1.1395, "step": 10657 }, { "epoch": 1.1340710789529687, "grad_norm": 12.8125, "learning_rate": 9.639070494599206e-06, "loss": 1.1616, "step": 10658 }, { "epoch": 1.1341774845711854, "grad_norm": 12.609375, "learning_rate": 9.639034963047188e-06, "loss": 1.2025, "step": 10659 }, { "epoch": 1.1342838901894021, "grad_norm": 13.421875, "learning_rate": 9.63899943149517e-06, "loss": 1.1706, "step": 10660 }, { "epoch": 1.1343902958076186, "grad_norm": 8.3984375, "learning_rate": 9.63896389994315e-06, "loss": 0.9856, "step": 10661 }, { "epoch": 1.1344967014258354, "grad_norm": 12.125, "learning_rate": 9.638928368391132e-06, "loss": 1.0804, "step": 10662 }, { "epoch": 1.1346031070440519, "grad_norm": 13.0625, "learning_rate": 9.638892836839114e-06, "loss": 1.0569, "step": 10663 }, { "epoch": 1.1347095126622686, "grad_norm": 9.984375, "learning_rate": 9.638857305287096e-06, "loss": 0.9605, "step": 10664 }, { "epoch": 1.1348159182804851, "grad_norm": 15.21875, "learning_rate": 9.638821773735077e-06, "loss": 1.3547, "step": 10665 }, { "epoch": 1.1349223238987018, "grad_norm": 10.0078125, "learning_rate": 9.638786242183059e-06, "loss": 0.9705, "step": 10666 }, { "epoch": 1.1350287295169186, "grad_norm": 12.0546875, "learning_rate": 9.63875071063104e-06, "loss": 1.1195, "step": 10667 }, { "epoch": 1.135135135135135, "grad_norm": 12.859375, "learning_rate": 9.638715179079024e-06, "loss": 1.4829, "step": 10668 }, { "epoch": 1.1352415407533518, "grad_norm": 13.609375, "learning_rate": 9.638679647527005e-06, "loss": 1.1955, "step": 10669 }, { "epoch": 1.1353479463715683, "grad_norm": 14.2734375, "learning_rate": 9.638644115974987e-06, "loss": 1.0448, "step": 10670 }, { "epoch": 1.135454351989785, "grad_norm": 10.3125, "learning_rate": 9.638608584422969e-06, "loss": 1.0241, "step": 10671 }, { "epoch": 1.1355607576080018, "grad_norm": 12.7578125, "learning_rate": 9.63857305287095e-06, "loss": 1.3693, "step": 10672 }, { "epoch": 1.1356671632262183, "grad_norm": 12.0390625, "learning_rate": 9.638537521318932e-06, "loss": 1.2439, "step": 10673 }, { "epoch": 1.135773568844435, "grad_norm": 10.0859375, "learning_rate": 9.638501989766913e-06, "loss": 0.9511, "step": 10674 }, { "epoch": 1.1358799744626515, "grad_norm": 12.796875, "learning_rate": 9.638466458214895e-06, "loss": 1.1536, "step": 10675 }, { "epoch": 1.1359863800808683, "grad_norm": 12.8046875, "learning_rate": 9.638430926662877e-06, "loss": 1.2123, "step": 10676 }, { "epoch": 1.136092785699085, "grad_norm": 14.5703125, "learning_rate": 9.638395395110858e-06, "loss": 1.2603, "step": 10677 }, { "epoch": 1.1361991913173015, "grad_norm": 16.328125, "learning_rate": 9.638359863558842e-06, "loss": 0.968, "step": 10678 }, { "epoch": 1.1363055969355182, "grad_norm": 9.1640625, "learning_rate": 9.638324332006823e-06, "loss": 0.8784, "step": 10679 }, { "epoch": 1.1364120025537348, "grad_norm": 9.7578125, "learning_rate": 9.638288800454805e-06, "loss": 1.1487, "step": 10680 }, { "epoch": 1.1365184081719515, "grad_norm": 10.9765625, "learning_rate": 9.638253268902786e-06, "loss": 0.9281, "step": 10681 }, { "epoch": 1.136624813790168, "grad_norm": 10.375, "learning_rate": 9.638217737350768e-06, "loss": 0.8281, "step": 10682 }, { "epoch": 1.1367312194083847, "grad_norm": 10.0234375, "learning_rate": 9.63818220579875e-06, "loss": 1.0211, "step": 10683 }, { "epoch": 1.1368376250266015, "grad_norm": 14.5546875, "learning_rate": 9.638146674246733e-06, "loss": 1.3164, "step": 10684 }, { "epoch": 1.136944030644818, "grad_norm": 14.8515625, "learning_rate": 9.638111142694714e-06, "loss": 0.9647, "step": 10685 }, { "epoch": 1.1370504362630347, "grad_norm": 14.0859375, "learning_rate": 9.638075611142696e-06, "loss": 1.374, "step": 10686 }, { "epoch": 1.1371568418812514, "grad_norm": 11.90625, "learning_rate": 9.638040079590678e-06, "loss": 1.2691, "step": 10687 }, { "epoch": 1.137263247499468, "grad_norm": 14.859375, "learning_rate": 9.63800454803866e-06, "loss": 1.0391, "step": 10688 }, { "epoch": 1.1373696531176847, "grad_norm": 14.078125, "learning_rate": 9.637969016486641e-06, "loss": 1.0829, "step": 10689 }, { "epoch": 1.1374760587359012, "grad_norm": 11.6953125, "learning_rate": 9.637933484934623e-06, "loss": 1.1293, "step": 10690 }, { "epoch": 1.137582464354118, "grad_norm": 12.03125, "learning_rate": 9.637897953382604e-06, "loss": 1.2758, "step": 10691 }, { "epoch": 1.1376888699723344, "grad_norm": 8.3671875, "learning_rate": 9.637862421830586e-06, "loss": 1.0178, "step": 10692 }, { "epoch": 1.1377952755905512, "grad_norm": 12.4140625, "learning_rate": 9.637826890278567e-06, "loss": 1.1315, "step": 10693 }, { "epoch": 1.137901681208768, "grad_norm": 19.28125, "learning_rate": 9.63779135872655e-06, "loss": 1.2767, "step": 10694 }, { "epoch": 1.1380080868269844, "grad_norm": 15.0078125, "learning_rate": 9.637755827174532e-06, "loss": 1.0114, "step": 10695 }, { "epoch": 1.1381144924452011, "grad_norm": 19.390625, "learning_rate": 9.637720295622514e-06, "loss": 1.2355, "step": 10696 }, { "epoch": 1.1382208980634179, "grad_norm": 12.2421875, "learning_rate": 9.637684764070495e-06, "loss": 0.9516, "step": 10697 }, { "epoch": 1.1383273036816344, "grad_norm": 14.6640625, "learning_rate": 9.637649232518477e-06, "loss": 1.1038, "step": 10698 }, { "epoch": 1.138433709299851, "grad_norm": 14.1015625, "learning_rate": 9.637613700966459e-06, "loss": 1.1845, "step": 10699 }, { "epoch": 1.1385401149180676, "grad_norm": 12.796875, "learning_rate": 9.63757816941444e-06, "loss": 1.1945, "step": 10700 }, { "epoch": 1.1386465205362843, "grad_norm": 13.296875, "learning_rate": 9.637542637862422e-06, "loss": 1.1078, "step": 10701 }, { "epoch": 1.1387529261545009, "grad_norm": 16.515625, "learning_rate": 9.637507106310403e-06, "loss": 0.9234, "step": 10702 }, { "epoch": 1.1388593317727176, "grad_norm": 13.8046875, "learning_rate": 9.637471574758385e-06, "loss": 1.317, "step": 10703 }, { "epoch": 1.1389657373909343, "grad_norm": 14.921875, "learning_rate": 9.637436043206368e-06, "loss": 1.1225, "step": 10704 }, { "epoch": 1.1390721430091508, "grad_norm": 12.8046875, "learning_rate": 9.63740051165435e-06, "loss": 0.9656, "step": 10705 }, { "epoch": 1.1391785486273676, "grad_norm": 10.765625, "learning_rate": 9.637364980102332e-06, "loss": 1.2712, "step": 10706 }, { "epoch": 1.1392849542455843, "grad_norm": 14.2734375, "learning_rate": 9.637329448550313e-06, "loss": 1.3811, "step": 10707 }, { "epoch": 1.1393913598638008, "grad_norm": 14.3203125, "learning_rate": 9.637293916998295e-06, "loss": 1.3687, "step": 10708 }, { "epoch": 1.1394977654820175, "grad_norm": 11.59375, "learning_rate": 9.637258385446278e-06, "loss": 0.9771, "step": 10709 }, { "epoch": 1.139604171100234, "grad_norm": 13.09375, "learning_rate": 9.63722285389426e-06, "loss": 1.3363, "step": 10710 }, { "epoch": 1.1397105767184508, "grad_norm": 11.3125, "learning_rate": 9.637187322342241e-06, "loss": 1.0133, "step": 10711 }, { "epoch": 1.1398169823366673, "grad_norm": 12.3671875, "learning_rate": 9.637151790790223e-06, "loss": 1.1454, "step": 10712 }, { "epoch": 1.139923387954884, "grad_norm": 21.203125, "learning_rate": 9.637116259238205e-06, "loss": 1.0791, "step": 10713 }, { "epoch": 1.1400297935731007, "grad_norm": 11.0, "learning_rate": 9.637080727686186e-06, "loss": 0.921, "step": 10714 }, { "epoch": 1.1401361991913173, "grad_norm": 10.4609375, "learning_rate": 9.637045196134168e-06, "loss": 0.9601, "step": 10715 }, { "epoch": 1.140242604809534, "grad_norm": 11.6875, "learning_rate": 9.63700966458215e-06, "loss": 1.0081, "step": 10716 }, { "epoch": 1.1403490104277505, "grad_norm": 10.9609375, "learning_rate": 9.636974133030131e-06, "loss": 1.0062, "step": 10717 }, { "epoch": 1.1404554160459672, "grad_norm": 8.7421875, "learning_rate": 9.636938601478113e-06, "loss": 0.7987, "step": 10718 }, { "epoch": 1.140561821664184, "grad_norm": 9.609375, "learning_rate": 9.636903069926096e-06, "loss": 1.0049, "step": 10719 }, { "epoch": 1.1406682272824005, "grad_norm": 14.3359375, "learning_rate": 9.636867538374077e-06, "loss": 1.1557, "step": 10720 }, { "epoch": 1.1407746329006172, "grad_norm": 10.3046875, "learning_rate": 9.636832006822059e-06, "loss": 1.1177, "step": 10721 }, { "epoch": 1.1408810385188337, "grad_norm": 18.125, "learning_rate": 9.63679647527004e-06, "loss": 1.182, "step": 10722 }, { "epoch": 1.1409874441370504, "grad_norm": 11.015625, "learning_rate": 9.636760943718022e-06, "loss": 1.4289, "step": 10723 }, { "epoch": 1.1410938497552672, "grad_norm": 11.75, "learning_rate": 9.636725412166004e-06, "loss": 1.0705, "step": 10724 }, { "epoch": 1.1412002553734837, "grad_norm": 17.28125, "learning_rate": 9.636689880613986e-06, "loss": 1.3343, "step": 10725 }, { "epoch": 1.1413066609917004, "grad_norm": 13.5, "learning_rate": 9.636654349061967e-06, "loss": 1.2091, "step": 10726 }, { "epoch": 1.141413066609917, "grad_norm": 9.328125, "learning_rate": 9.636618817509949e-06, "loss": 0.9847, "step": 10727 }, { "epoch": 1.1415194722281337, "grad_norm": 15.296875, "learning_rate": 9.63658328595793e-06, "loss": 1.317, "step": 10728 }, { "epoch": 1.1416258778463502, "grad_norm": 11.25, "learning_rate": 9.636547754405912e-06, "loss": 0.9942, "step": 10729 }, { "epoch": 1.141732283464567, "grad_norm": 12.78125, "learning_rate": 9.636512222853895e-06, "loss": 1.2003, "step": 10730 }, { "epoch": 1.1418386890827836, "grad_norm": 14.84375, "learning_rate": 9.636476691301877e-06, "loss": 1.0208, "step": 10731 }, { "epoch": 1.1419450947010001, "grad_norm": 15.40625, "learning_rate": 9.636441159749858e-06, "loss": 1.2053, "step": 10732 }, { "epoch": 1.1420515003192169, "grad_norm": 9.015625, "learning_rate": 9.63640562819784e-06, "loss": 1.0168, "step": 10733 }, { "epoch": 1.1421579059374336, "grad_norm": 11.515625, "learning_rate": 9.636370096645822e-06, "loss": 1.1963, "step": 10734 }, { "epoch": 1.1422643115556501, "grad_norm": 12.859375, "learning_rate": 9.636334565093805e-06, "loss": 1.3525, "step": 10735 }, { "epoch": 1.1423707171738668, "grad_norm": 15.765625, "learning_rate": 9.636299033541787e-06, "loss": 1.1941, "step": 10736 }, { "epoch": 1.1424771227920834, "grad_norm": 14.75, "learning_rate": 9.636263501989768e-06, "loss": 1.0647, "step": 10737 }, { "epoch": 1.1425835284103, "grad_norm": 15.6171875, "learning_rate": 9.63622797043775e-06, "loss": 0.9656, "step": 10738 }, { "epoch": 1.1426899340285166, "grad_norm": 12.234375, "learning_rate": 9.636192438885731e-06, "loss": 1.0344, "step": 10739 }, { "epoch": 1.1427963396467333, "grad_norm": 12.40625, "learning_rate": 9.636156907333713e-06, "loss": 0.9925, "step": 10740 }, { "epoch": 1.14290274526495, "grad_norm": 12.296875, "learning_rate": 9.636121375781695e-06, "loss": 1.1538, "step": 10741 }, { "epoch": 1.1430091508831666, "grad_norm": 11.109375, "learning_rate": 9.636085844229676e-06, "loss": 1.1196, "step": 10742 }, { "epoch": 1.1431155565013833, "grad_norm": 10.8359375, "learning_rate": 9.636050312677658e-06, "loss": 1.2898, "step": 10743 }, { "epoch": 1.1432219621196, "grad_norm": 10.1171875, "learning_rate": 9.63601478112564e-06, "loss": 1.1862, "step": 10744 }, { "epoch": 1.1433283677378165, "grad_norm": 9.46875, "learning_rate": 9.635979249573623e-06, "loss": 1.2327, "step": 10745 }, { "epoch": 1.1434347733560333, "grad_norm": 11.828125, "learning_rate": 9.635943718021604e-06, "loss": 1.1256, "step": 10746 }, { "epoch": 1.1435411789742498, "grad_norm": 14.3515625, "learning_rate": 9.635908186469586e-06, "loss": 0.9776, "step": 10747 }, { "epoch": 1.1436475845924665, "grad_norm": 9.0078125, "learning_rate": 9.635872654917568e-06, "loss": 0.9991, "step": 10748 }, { "epoch": 1.143753990210683, "grad_norm": 16.390625, "learning_rate": 9.63583712336555e-06, "loss": 0.975, "step": 10749 }, { "epoch": 1.1438603958288998, "grad_norm": 11.0625, "learning_rate": 9.635801591813532e-06, "loss": 1.0183, "step": 10750 }, { "epoch": 1.1439668014471165, "grad_norm": 12.59375, "learning_rate": 9.635766060261514e-06, "loss": 1.4374, "step": 10751 }, { "epoch": 1.144073207065333, "grad_norm": 13.265625, "learning_rate": 9.635730528709496e-06, "loss": 0.9895, "step": 10752 }, { "epoch": 1.1441796126835497, "grad_norm": 10.8359375, "learning_rate": 9.635694997157477e-06, "loss": 0.8506, "step": 10753 }, { "epoch": 1.1442860183017662, "grad_norm": 10.40625, "learning_rate": 9.635659465605459e-06, "loss": 1.0956, "step": 10754 }, { "epoch": 1.144392423919983, "grad_norm": 13.5703125, "learning_rate": 9.63562393405344e-06, "loss": 1.1278, "step": 10755 }, { "epoch": 1.1444988295381997, "grad_norm": 13.578125, "learning_rate": 9.635588402501422e-06, "loss": 1.2993, "step": 10756 }, { "epoch": 1.1446052351564162, "grad_norm": 11.8203125, "learning_rate": 9.635552870949404e-06, "loss": 1.0009, "step": 10757 }, { "epoch": 1.144711640774633, "grad_norm": 17.34375, "learning_rate": 9.635517339397385e-06, "loss": 0.885, "step": 10758 }, { "epoch": 1.1448180463928495, "grad_norm": 20.359375, "learning_rate": 9.635481807845367e-06, "loss": 1.1426, "step": 10759 }, { "epoch": 1.1449244520110662, "grad_norm": 15.375, "learning_rate": 9.63544627629335e-06, "loss": 1.3205, "step": 10760 }, { "epoch": 1.145030857629283, "grad_norm": 9.5859375, "learning_rate": 9.635410744741332e-06, "loss": 1.0568, "step": 10761 }, { "epoch": 1.1451372632474994, "grad_norm": 9.875, "learning_rate": 9.635375213189313e-06, "loss": 0.9784, "step": 10762 }, { "epoch": 1.1452436688657162, "grad_norm": 13.9375, "learning_rate": 9.635339681637295e-06, "loss": 1.1801, "step": 10763 }, { "epoch": 1.1453500744839327, "grad_norm": 12.4140625, "learning_rate": 9.635304150085277e-06, "loss": 0.8867, "step": 10764 }, { "epoch": 1.1454564801021494, "grad_norm": 12.7265625, "learning_rate": 9.635268618533258e-06, "loss": 1.2702, "step": 10765 }, { "epoch": 1.1455628857203661, "grad_norm": 11.0390625, "learning_rate": 9.63523308698124e-06, "loss": 1.1717, "step": 10766 }, { "epoch": 1.1456692913385826, "grad_norm": 9.4921875, "learning_rate": 9.635197555429221e-06, "loss": 0.9084, "step": 10767 }, { "epoch": 1.1457756969567994, "grad_norm": 10.09375, "learning_rate": 9.635162023877203e-06, "loss": 0.9613, "step": 10768 }, { "epoch": 1.1458821025750159, "grad_norm": 13.6484375, "learning_rate": 9.635126492325185e-06, "loss": 1.289, "step": 10769 }, { "epoch": 1.1459885081932326, "grad_norm": 10.6640625, "learning_rate": 9.635090960773166e-06, "loss": 0.9713, "step": 10770 }, { "epoch": 1.1460949138114493, "grad_norm": 17.3125, "learning_rate": 9.63505542922115e-06, "loss": 1.1179, "step": 10771 }, { "epoch": 1.1462013194296659, "grad_norm": 13.4609375, "learning_rate": 9.635019897669131e-06, "loss": 1.1257, "step": 10772 }, { "epoch": 1.1463077250478826, "grad_norm": 10.2578125, "learning_rate": 9.634984366117113e-06, "loss": 0.9648, "step": 10773 }, { "epoch": 1.146414130666099, "grad_norm": 12.5703125, "learning_rate": 9.634948834565094e-06, "loss": 0.8899, "step": 10774 }, { "epoch": 1.1465205362843158, "grad_norm": 10.28125, "learning_rate": 9.634913303013076e-06, "loss": 0.9744, "step": 10775 }, { "epoch": 1.1466269419025323, "grad_norm": 12.1875, "learning_rate": 9.63487777146106e-06, "loss": 0.9635, "step": 10776 }, { "epoch": 1.146733347520749, "grad_norm": 12.1484375, "learning_rate": 9.634842239909041e-06, "loss": 1.0132, "step": 10777 }, { "epoch": 1.1468397531389658, "grad_norm": 10.84375, "learning_rate": 9.634806708357023e-06, "loss": 1.0836, "step": 10778 }, { "epoch": 1.1469461587571823, "grad_norm": 12.125, "learning_rate": 9.634771176805004e-06, "loss": 1.2031, "step": 10779 }, { "epoch": 1.147052564375399, "grad_norm": 10.34375, "learning_rate": 9.634735645252986e-06, "loss": 0.892, "step": 10780 }, { "epoch": 1.1471589699936158, "grad_norm": 10.421875, "learning_rate": 9.634700113700967e-06, "loss": 0.8461, "step": 10781 }, { "epoch": 1.1472653756118323, "grad_norm": 12.6640625, "learning_rate": 9.634664582148949e-06, "loss": 1.1316, "step": 10782 }, { "epoch": 1.147371781230049, "grad_norm": 13.5859375, "learning_rate": 9.63462905059693e-06, "loss": 1.1625, "step": 10783 }, { "epoch": 1.1474781868482655, "grad_norm": 11.9453125, "learning_rate": 9.634593519044912e-06, "loss": 1.133, "step": 10784 }, { "epoch": 1.1475845924664823, "grad_norm": 14.046875, "learning_rate": 9.634557987492894e-06, "loss": 0.8463, "step": 10785 }, { "epoch": 1.1476909980846988, "grad_norm": 11.6328125, "learning_rate": 9.634522455940877e-06, "loss": 1.1834, "step": 10786 }, { "epoch": 1.1477974037029155, "grad_norm": 12.3046875, "learning_rate": 9.634486924388859e-06, "loss": 0.8352, "step": 10787 }, { "epoch": 1.1479038093211322, "grad_norm": 14.921875, "learning_rate": 9.63445139283684e-06, "loss": 1.1992, "step": 10788 }, { "epoch": 1.1480102149393487, "grad_norm": 10.484375, "learning_rate": 9.634415861284822e-06, "loss": 1.241, "step": 10789 }, { "epoch": 1.1481166205575655, "grad_norm": 12.4375, "learning_rate": 9.634380329732803e-06, "loss": 1.3757, "step": 10790 }, { "epoch": 1.1482230261757822, "grad_norm": 12.84375, "learning_rate": 9.634344798180785e-06, "loss": 0.9574, "step": 10791 }, { "epoch": 1.1483294317939987, "grad_norm": 19.359375, "learning_rate": 9.634309266628767e-06, "loss": 1.3119, "step": 10792 }, { "epoch": 1.1484358374122154, "grad_norm": 13.1171875, "learning_rate": 9.634273735076748e-06, "loss": 0.9454, "step": 10793 }, { "epoch": 1.148542243030432, "grad_norm": 12.0546875, "learning_rate": 9.63423820352473e-06, "loss": 0.9802, "step": 10794 }, { "epoch": 1.1486486486486487, "grad_norm": 13.34375, "learning_rate": 9.634202671972712e-06, "loss": 0.9783, "step": 10795 }, { "epoch": 1.1487550542668652, "grad_norm": 17.5625, "learning_rate": 9.634167140420695e-06, "loss": 0.9008, "step": 10796 }, { "epoch": 1.148861459885082, "grad_norm": 11.2890625, "learning_rate": 9.634131608868676e-06, "loss": 1.1531, "step": 10797 }, { "epoch": 1.1489678655032987, "grad_norm": 13.828125, "learning_rate": 9.634096077316658e-06, "loss": 1.0575, "step": 10798 }, { "epoch": 1.1490742711215152, "grad_norm": 10.5, "learning_rate": 9.63406054576464e-06, "loss": 0.9153, "step": 10799 }, { "epoch": 1.149180676739732, "grad_norm": 11.28125, "learning_rate": 9.634025014212621e-06, "loss": 1.0531, "step": 10800 }, { "epoch": 1.1492870823579484, "grad_norm": 11.84375, "learning_rate": 9.633989482660603e-06, "loss": 1.0858, "step": 10801 }, { "epoch": 1.1493934879761651, "grad_norm": 10.6796875, "learning_rate": 9.633953951108586e-06, "loss": 1.1991, "step": 10802 }, { "epoch": 1.1494998935943819, "grad_norm": 14.4921875, "learning_rate": 9.633918419556568e-06, "loss": 1.2985, "step": 10803 }, { "epoch": 1.1496062992125984, "grad_norm": 9.625, "learning_rate": 9.63388288800455e-06, "loss": 0.879, "step": 10804 }, { "epoch": 1.1497127048308151, "grad_norm": 14.6875, "learning_rate": 9.633847356452531e-06, "loss": 1.0537, "step": 10805 }, { "epoch": 1.1498191104490316, "grad_norm": 12.8046875, "learning_rate": 9.633811824900513e-06, "loss": 1.1143, "step": 10806 }, { "epoch": 1.1499255160672484, "grad_norm": 10.78125, "learning_rate": 9.633776293348494e-06, "loss": 1.1605, "step": 10807 }, { "epoch": 1.150031921685465, "grad_norm": 14.984375, "learning_rate": 9.633740761796476e-06, "loss": 1.1663, "step": 10808 }, { "epoch": 1.1501383273036816, "grad_norm": 22.078125, "learning_rate": 9.633705230244457e-06, "loss": 1.0142, "step": 10809 }, { "epoch": 1.1502447329218983, "grad_norm": 15.1875, "learning_rate": 9.633669698692439e-06, "loss": 1.4717, "step": 10810 }, { "epoch": 1.1503511385401148, "grad_norm": 12.625, "learning_rate": 9.63363416714042e-06, "loss": 0.9081, "step": 10811 }, { "epoch": 1.1504575441583316, "grad_norm": 15.984375, "learning_rate": 9.633598635588404e-06, "loss": 1.227, "step": 10812 }, { "epoch": 1.150563949776548, "grad_norm": 10.296875, "learning_rate": 9.633563104036386e-06, "loss": 0.942, "step": 10813 }, { "epoch": 1.1506703553947648, "grad_norm": 12.4921875, "learning_rate": 9.633527572484367e-06, "loss": 1.2066, "step": 10814 }, { "epoch": 1.1507767610129815, "grad_norm": 12.4453125, "learning_rate": 9.633492040932349e-06, "loss": 0.8653, "step": 10815 }, { "epoch": 1.150883166631198, "grad_norm": 10.75, "learning_rate": 9.63345650938033e-06, "loss": 1.0603, "step": 10816 }, { "epoch": 1.1509895722494148, "grad_norm": 11.8046875, "learning_rate": 9.633420977828312e-06, "loss": 0.9388, "step": 10817 }, { "epoch": 1.1510959778676315, "grad_norm": 13.7734375, "learning_rate": 9.633385446276294e-06, "loss": 1.2294, "step": 10818 }, { "epoch": 1.151202383485848, "grad_norm": 14.5546875, "learning_rate": 9.633349914724275e-06, "loss": 0.8295, "step": 10819 }, { "epoch": 1.1513087891040648, "grad_norm": 12.21875, "learning_rate": 9.633314383172257e-06, "loss": 1.2271, "step": 10820 }, { "epoch": 1.1514151947222813, "grad_norm": 14.96875, "learning_rate": 9.633278851620238e-06, "loss": 1.4083, "step": 10821 }, { "epoch": 1.151521600340498, "grad_norm": 16.640625, "learning_rate": 9.633243320068222e-06, "loss": 1.1566, "step": 10822 }, { "epoch": 1.1516280059587145, "grad_norm": 13.046875, "learning_rate": 9.633207788516203e-06, "loss": 1.2064, "step": 10823 }, { "epoch": 1.1517344115769312, "grad_norm": 10.203125, "learning_rate": 9.633172256964185e-06, "loss": 1.0144, "step": 10824 }, { "epoch": 1.151840817195148, "grad_norm": 13.0078125, "learning_rate": 9.633136725412167e-06, "loss": 1.0983, "step": 10825 }, { "epoch": 1.1519472228133645, "grad_norm": 12.4609375, "learning_rate": 9.633101193860148e-06, "loss": 1.0241, "step": 10826 }, { "epoch": 1.1520536284315812, "grad_norm": 15.84375, "learning_rate": 9.633065662308131e-06, "loss": 1.1303, "step": 10827 }, { "epoch": 1.152160034049798, "grad_norm": 13.296875, "learning_rate": 9.633030130756113e-06, "loss": 1.1781, "step": 10828 }, { "epoch": 1.1522664396680145, "grad_norm": 10.515625, "learning_rate": 9.632994599204095e-06, "loss": 0.9839, "step": 10829 }, { "epoch": 1.1523728452862312, "grad_norm": 13.140625, "learning_rate": 9.632959067652076e-06, "loss": 1.024, "step": 10830 }, { "epoch": 1.1524792509044477, "grad_norm": 11.1640625, "learning_rate": 9.632923536100058e-06, "loss": 1.0503, "step": 10831 }, { "epoch": 1.1525856565226644, "grad_norm": 11.4140625, "learning_rate": 9.63288800454804e-06, "loss": 1.091, "step": 10832 }, { "epoch": 1.152692062140881, "grad_norm": 12.5390625, "learning_rate": 9.632852472996021e-06, "loss": 1.1855, "step": 10833 }, { "epoch": 1.1527984677590977, "grad_norm": 12.3125, "learning_rate": 9.632816941444003e-06, "loss": 1.2118, "step": 10834 }, { "epoch": 1.1529048733773144, "grad_norm": 14.1328125, "learning_rate": 9.632781409891984e-06, "loss": 0.9919, "step": 10835 }, { "epoch": 1.153011278995531, "grad_norm": 12.7421875, "learning_rate": 9.632745878339966e-06, "loss": 0.8159, "step": 10836 }, { "epoch": 1.1531176846137476, "grad_norm": 16.0, "learning_rate": 9.63271034678795e-06, "loss": 1.2534, "step": 10837 }, { "epoch": 1.1532240902319644, "grad_norm": 12.8828125, "learning_rate": 9.63267481523593e-06, "loss": 1.1154, "step": 10838 }, { "epoch": 1.1533304958501809, "grad_norm": 10.109375, "learning_rate": 9.632639283683912e-06, "loss": 1.1828, "step": 10839 }, { "epoch": 1.1534369014683976, "grad_norm": 12.515625, "learning_rate": 9.632603752131894e-06, "loss": 1.0279, "step": 10840 }, { "epoch": 1.1535433070866141, "grad_norm": 13.65625, "learning_rate": 9.632568220579876e-06, "loss": 1.2366, "step": 10841 }, { "epoch": 1.1536497127048309, "grad_norm": 11.125, "learning_rate": 9.632532689027857e-06, "loss": 0.9477, "step": 10842 }, { "epoch": 1.1537561183230474, "grad_norm": 12.3125, "learning_rate": 9.63249715747584e-06, "loss": 1.0896, "step": 10843 }, { "epoch": 1.153862523941264, "grad_norm": 10.6875, "learning_rate": 9.632461625923822e-06, "loss": 1.0582, "step": 10844 }, { "epoch": 1.1539689295594808, "grad_norm": 16.53125, "learning_rate": 9.632426094371802e-06, "loss": 1.1892, "step": 10845 }, { "epoch": 1.1540753351776973, "grad_norm": 13.8203125, "learning_rate": 9.632390562819784e-06, "loss": 1.2021, "step": 10846 }, { "epoch": 1.154181740795914, "grad_norm": 11.8046875, "learning_rate": 9.632355031267765e-06, "loss": 1.146, "step": 10847 }, { "epoch": 1.1542881464141306, "grad_norm": 19.046875, "learning_rate": 9.632319499715749e-06, "loss": 1.1899, "step": 10848 }, { "epoch": 1.1543945520323473, "grad_norm": 10.75, "learning_rate": 9.63228396816373e-06, "loss": 1.173, "step": 10849 }, { "epoch": 1.154500957650564, "grad_norm": 12.515625, "learning_rate": 9.632248436611712e-06, "loss": 1.4646, "step": 10850 }, { "epoch": 1.1546073632687806, "grad_norm": 11.015625, "learning_rate": 9.632212905059693e-06, "loss": 0.8886, "step": 10851 }, { "epoch": 1.1547137688869973, "grad_norm": 9.78125, "learning_rate": 9.632177373507675e-06, "loss": 1.2006, "step": 10852 }, { "epoch": 1.1548201745052138, "grad_norm": 15.921875, "learning_rate": 9.632141841955658e-06, "loss": 1.3598, "step": 10853 }, { "epoch": 1.1549265801234305, "grad_norm": 10.296875, "learning_rate": 9.63210631040364e-06, "loss": 1.0312, "step": 10854 }, { "epoch": 1.1550329857416473, "grad_norm": 10.0546875, "learning_rate": 9.632070778851621e-06, "loss": 0.938, "step": 10855 }, { "epoch": 1.1551393913598638, "grad_norm": 13.6015625, "learning_rate": 9.632035247299603e-06, "loss": 0.8498, "step": 10856 }, { "epoch": 1.1552457969780805, "grad_norm": 10.984375, "learning_rate": 9.631999715747585e-06, "loss": 1.0658, "step": 10857 }, { "epoch": 1.155352202596297, "grad_norm": 13.78125, "learning_rate": 9.631964184195566e-06, "loss": 1.053, "step": 10858 }, { "epoch": 1.1554586082145137, "grad_norm": 13.2265625, "learning_rate": 9.631928652643548e-06, "loss": 0.8554, "step": 10859 }, { "epoch": 1.1555650138327302, "grad_norm": 11.4609375, "learning_rate": 9.63189312109153e-06, "loss": 0.9911, "step": 10860 }, { "epoch": 1.155671419450947, "grad_norm": 14.6953125, "learning_rate": 9.631857589539511e-06, "loss": 1.1546, "step": 10861 }, { "epoch": 1.1557778250691637, "grad_norm": 14.328125, "learning_rate": 9.631822057987493e-06, "loss": 1.152, "step": 10862 }, { "epoch": 1.1558842306873802, "grad_norm": 13.84375, "learning_rate": 9.631786526435476e-06, "loss": 1.0482, "step": 10863 }, { "epoch": 1.155990636305597, "grad_norm": 14.7890625, "learning_rate": 9.631750994883458e-06, "loss": 0.965, "step": 10864 }, { "epoch": 1.1560970419238137, "grad_norm": 11.2265625, "learning_rate": 9.63171546333144e-06, "loss": 1.2066, "step": 10865 }, { "epoch": 1.1562034475420302, "grad_norm": 11.703125, "learning_rate": 9.631679931779421e-06, "loss": 1.1373, "step": 10866 }, { "epoch": 1.156309853160247, "grad_norm": 10.3984375, "learning_rate": 9.631644400227402e-06, "loss": 0.9862, "step": 10867 }, { "epoch": 1.1564162587784634, "grad_norm": 10.40625, "learning_rate": 9.631608868675386e-06, "loss": 0.7497, "step": 10868 }, { "epoch": 1.1565226643966802, "grad_norm": 13.78125, "learning_rate": 9.631573337123367e-06, "loss": 1.1456, "step": 10869 }, { "epoch": 1.1566290700148967, "grad_norm": 12.0546875, "learning_rate": 9.631537805571349e-06, "loss": 1.166, "step": 10870 }, { "epoch": 1.1567354756331134, "grad_norm": 11.390625, "learning_rate": 9.63150227401933e-06, "loss": 0.9255, "step": 10871 }, { "epoch": 1.1568418812513301, "grad_norm": 14.8828125, "learning_rate": 9.631466742467312e-06, "loss": 1.1208, "step": 10872 }, { "epoch": 1.1569482868695466, "grad_norm": 13.78125, "learning_rate": 9.631431210915294e-06, "loss": 1.011, "step": 10873 }, { "epoch": 1.1570546924877634, "grad_norm": 12.859375, "learning_rate": 9.631395679363275e-06, "loss": 1.2185, "step": 10874 }, { "epoch": 1.1571610981059801, "grad_norm": 11.4296875, "learning_rate": 9.631360147811257e-06, "loss": 1.0875, "step": 10875 }, { "epoch": 1.1572675037241966, "grad_norm": 13.015625, "learning_rate": 9.631324616259239e-06, "loss": 1.3345, "step": 10876 }, { "epoch": 1.1573739093424134, "grad_norm": 12.71875, "learning_rate": 9.63128908470722e-06, "loss": 1.2128, "step": 10877 }, { "epoch": 1.1574803149606299, "grad_norm": 13.6796875, "learning_rate": 9.631253553155204e-06, "loss": 1.6231, "step": 10878 }, { "epoch": 1.1575867205788466, "grad_norm": 10.3671875, "learning_rate": 9.631218021603185e-06, "loss": 1.0057, "step": 10879 }, { "epoch": 1.157693126197063, "grad_norm": 10.8046875, "learning_rate": 9.631182490051167e-06, "loss": 1.3831, "step": 10880 }, { "epoch": 1.1577995318152798, "grad_norm": 12.8125, "learning_rate": 9.631146958499148e-06, "loss": 1.123, "step": 10881 }, { "epoch": 1.1579059374334966, "grad_norm": 13.3515625, "learning_rate": 9.63111142694713e-06, "loss": 1.3218, "step": 10882 }, { "epoch": 1.158012343051713, "grad_norm": 14.484375, "learning_rate": 9.631075895395112e-06, "loss": 1.3005, "step": 10883 }, { "epoch": 1.1581187486699298, "grad_norm": 11.46875, "learning_rate": 9.631040363843093e-06, "loss": 0.8926, "step": 10884 }, { "epoch": 1.1582251542881463, "grad_norm": 12.8203125, "learning_rate": 9.631004832291075e-06, "loss": 1.06, "step": 10885 }, { "epoch": 1.158331559906363, "grad_norm": 18.796875, "learning_rate": 9.630969300739056e-06, "loss": 1.2029, "step": 10886 }, { "epoch": 1.1584379655245798, "grad_norm": 13.984375, "learning_rate": 9.630933769187038e-06, "loss": 1.1321, "step": 10887 }, { "epoch": 1.1585443711427963, "grad_norm": 14.765625, "learning_rate": 9.63089823763502e-06, "loss": 1.0887, "step": 10888 }, { "epoch": 1.158650776761013, "grad_norm": 11.6875, "learning_rate": 9.630862706083003e-06, "loss": 0.9619, "step": 10889 }, { "epoch": 1.1587571823792295, "grad_norm": 14.828125, "learning_rate": 9.630827174530984e-06, "loss": 0.8975, "step": 10890 }, { "epoch": 1.1588635879974463, "grad_norm": 10.2890625, "learning_rate": 9.630791642978966e-06, "loss": 0.9903, "step": 10891 }, { "epoch": 1.158969993615663, "grad_norm": 10.3984375, "learning_rate": 9.630756111426948e-06, "loss": 1.0805, "step": 10892 }, { "epoch": 1.1590763992338795, "grad_norm": 12.8828125, "learning_rate": 9.63072057987493e-06, "loss": 1.0755, "step": 10893 }, { "epoch": 1.1591828048520962, "grad_norm": 17.25, "learning_rate": 9.630685048322913e-06, "loss": 1.1557, "step": 10894 }, { "epoch": 1.1592892104703127, "grad_norm": 19.359375, "learning_rate": 9.630649516770894e-06, "loss": 1.0499, "step": 10895 }, { "epoch": 1.1593956160885295, "grad_norm": 11.5703125, "learning_rate": 9.630613985218876e-06, "loss": 0.8566, "step": 10896 }, { "epoch": 1.1595020217067462, "grad_norm": 10.7890625, "learning_rate": 9.630578453666857e-06, "loss": 0.9707, "step": 10897 }, { "epoch": 1.1596084273249627, "grad_norm": 10.96875, "learning_rate": 9.630542922114839e-06, "loss": 1.1592, "step": 10898 }, { "epoch": 1.1597148329431795, "grad_norm": 13.125, "learning_rate": 9.63050739056282e-06, "loss": 0.9261, "step": 10899 }, { "epoch": 1.159821238561396, "grad_norm": 12.078125, "learning_rate": 9.630471859010802e-06, "loss": 0.9638, "step": 10900 }, { "epoch": 1.1599276441796127, "grad_norm": 15.21875, "learning_rate": 9.630436327458784e-06, "loss": 1.1685, "step": 10901 }, { "epoch": 1.1600340497978294, "grad_norm": 19.328125, "learning_rate": 9.630400795906765e-06, "loss": 1.1771, "step": 10902 }, { "epoch": 1.160140455416046, "grad_norm": 12.734375, "learning_rate": 9.630365264354747e-06, "loss": 1.0004, "step": 10903 }, { "epoch": 1.1602468610342627, "grad_norm": 12.4921875, "learning_rate": 9.63032973280273e-06, "loss": 1.2391, "step": 10904 }, { "epoch": 1.1603532666524792, "grad_norm": 14.6015625, "learning_rate": 9.630294201250712e-06, "loss": 1.0632, "step": 10905 }, { "epoch": 1.160459672270696, "grad_norm": 19.421875, "learning_rate": 9.630258669698694e-06, "loss": 1.5152, "step": 10906 }, { "epoch": 1.1605660778889124, "grad_norm": 12.4765625, "learning_rate": 9.630223138146675e-06, "loss": 1.0547, "step": 10907 }, { "epoch": 1.1606724835071291, "grad_norm": 18.015625, "learning_rate": 9.630187606594657e-06, "loss": 1.1479, "step": 10908 }, { "epoch": 1.1607788891253459, "grad_norm": 14.6328125, "learning_rate": 9.630152075042638e-06, "loss": 1.1466, "step": 10909 }, { "epoch": 1.1608852947435624, "grad_norm": 13.578125, "learning_rate": 9.63011654349062e-06, "loss": 1.0644, "step": 10910 }, { "epoch": 1.1609917003617791, "grad_norm": 15.15625, "learning_rate": 9.630081011938602e-06, "loss": 0.9817, "step": 10911 }, { "epoch": 1.1610981059799959, "grad_norm": 16.40625, "learning_rate": 9.630045480386583e-06, "loss": 1.1976, "step": 10912 }, { "epoch": 1.1612045115982124, "grad_norm": 11.2578125, "learning_rate": 9.630009948834565e-06, "loss": 0.9578, "step": 10913 }, { "epoch": 1.161310917216429, "grad_norm": 12.3828125, "learning_rate": 9.629974417282548e-06, "loss": 0.9528, "step": 10914 }, { "epoch": 1.1614173228346456, "grad_norm": 12.65625, "learning_rate": 9.62993888573053e-06, "loss": 1.0601, "step": 10915 }, { "epoch": 1.1615237284528623, "grad_norm": 19.109375, "learning_rate": 9.629903354178511e-06, "loss": 1.2567, "step": 10916 }, { "epoch": 1.1616301340710788, "grad_norm": 14.0546875, "learning_rate": 9.629867822626493e-06, "loss": 1.1125, "step": 10917 }, { "epoch": 1.1617365396892956, "grad_norm": 15.1640625, "learning_rate": 9.629832291074475e-06, "loss": 1.3389, "step": 10918 }, { "epoch": 1.1618429453075123, "grad_norm": 13.1640625, "learning_rate": 9.629796759522456e-06, "loss": 1.1162, "step": 10919 }, { "epoch": 1.1619493509257288, "grad_norm": 17.46875, "learning_rate": 9.62976122797044e-06, "loss": 1.0459, "step": 10920 }, { "epoch": 1.1620557565439456, "grad_norm": 12.578125, "learning_rate": 9.629725696418421e-06, "loss": 1.0161, "step": 10921 }, { "epoch": 1.1621621621621623, "grad_norm": 11.4375, "learning_rate": 9.629690164866403e-06, "loss": 1.087, "step": 10922 }, { "epoch": 1.1622685677803788, "grad_norm": 11.4609375, "learning_rate": 9.629654633314384e-06, "loss": 0.951, "step": 10923 }, { "epoch": 1.1623749733985955, "grad_norm": 12.65625, "learning_rate": 9.629619101762366e-06, "loss": 1.2322, "step": 10924 }, { "epoch": 1.162481379016812, "grad_norm": 10.140625, "learning_rate": 9.629583570210348e-06, "loss": 1.0813, "step": 10925 }, { "epoch": 1.1625877846350288, "grad_norm": 10.765625, "learning_rate": 9.629548038658329e-06, "loss": 0.9564, "step": 10926 }, { "epoch": 1.1626941902532453, "grad_norm": 13.578125, "learning_rate": 9.62951250710631e-06, "loss": 1.0587, "step": 10927 }, { "epoch": 1.162800595871462, "grad_norm": 9.953125, "learning_rate": 9.629476975554292e-06, "loss": 0.9182, "step": 10928 }, { "epoch": 1.1629070014896787, "grad_norm": 10.71875, "learning_rate": 9.629441444002274e-06, "loss": 0.8752, "step": 10929 }, { "epoch": 1.1630134071078952, "grad_norm": 11.4609375, "learning_rate": 9.629405912450257e-06, "loss": 0.9508, "step": 10930 }, { "epoch": 1.163119812726112, "grad_norm": 13.4296875, "learning_rate": 9.629370380898239e-06, "loss": 0.9929, "step": 10931 }, { "epoch": 1.1632262183443285, "grad_norm": 11.40625, "learning_rate": 9.62933484934622e-06, "loss": 0.9749, "step": 10932 }, { "epoch": 1.1633326239625452, "grad_norm": 11.6484375, "learning_rate": 9.629299317794202e-06, "loss": 1.1258, "step": 10933 }, { "epoch": 1.163439029580762, "grad_norm": 13.28125, "learning_rate": 9.629263786242184e-06, "loss": 1.1666, "step": 10934 }, { "epoch": 1.1635454351989785, "grad_norm": 12.3359375, "learning_rate": 9.629228254690165e-06, "loss": 1.0359, "step": 10935 }, { "epoch": 1.1636518408171952, "grad_norm": 10.0078125, "learning_rate": 9.629192723138147e-06, "loss": 1.0867, "step": 10936 }, { "epoch": 1.1637582464354117, "grad_norm": 9.8984375, "learning_rate": 9.629157191586128e-06, "loss": 0.9529, "step": 10937 }, { "epoch": 1.1638646520536284, "grad_norm": 13.0546875, "learning_rate": 9.62912166003411e-06, "loss": 1.1281, "step": 10938 }, { "epoch": 1.1639710576718452, "grad_norm": 10.9921875, "learning_rate": 9.629086128482092e-06, "loss": 1.2262, "step": 10939 }, { "epoch": 1.1640774632900617, "grad_norm": 9.5, "learning_rate": 9.629050596930075e-06, "loss": 0.9504, "step": 10940 }, { "epoch": 1.1641838689082784, "grad_norm": 14.4375, "learning_rate": 9.629015065378057e-06, "loss": 0.9701, "step": 10941 }, { "epoch": 1.164290274526495, "grad_norm": 11.4375, "learning_rate": 9.628979533826038e-06, "loss": 1.0387, "step": 10942 }, { "epoch": 1.1643966801447116, "grad_norm": 15.6484375, "learning_rate": 9.62894400227402e-06, "loss": 0.9552, "step": 10943 }, { "epoch": 1.1645030857629282, "grad_norm": 8.9375, "learning_rate": 9.628908470722001e-06, "loss": 1.0007, "step": 10944 }, { "epoch": 1.164609491381145, "grad_norm": 9.2265625, "learning_rate": 9.628872939169985e-06, "loss": 1.0143, "step": 10945 }, { "epoch": 1.1647158969993616, "grad_norm": 13.59375, "learning_rate": 9.628837407617966e-06, "loss": 1.2953, "step": 10946 }, { "epoch": 1.1648223026175781, "grad_norm": 11.0078125, "learning_rate": 9.628801876065948e-06, "loss": 0.8169, "step": 10947 }, { "epoch": 1.1649287082357949, "grad_norm": 12.75, "learning_rate": 9.62876634451393e-06, "loss": 1.1069, "step": 10948 }, { "epoch": 1.1650351138540116, "grad_norm": 16.296875, "learning_rate": 9.628730812961911e-06, "loss": 1.4161, "step": 10949 }, { "epoch": 1.165141519472228, "grad_norm": 11.8984375, "learning_rate": 9.628695281409893e-06, "loss": 1.0676, "step": 10950 }, { "epoch": 1.1652479250904448, "grad_norm": 12.9921875, "learning_rate": 9.628659749857874e-06, "loss": 1.0948, "step": 10951 }, { "epoch": 1.1653543307086613, "grad_norm": 12.2265625, "learning_rate": 9.628624218305856e-06, "loss": 1.1049, "step": 10952 }, { "epoch": 1.165460736326878, "grad_norm": 12.5, "learning_rate": 9.628588686753838e-06, "loss": 1.2335, "step": 10953 }, { "epoch": 1.1655671419450946, "grad_norm": 12.3671875, "learning_rate": 9.62855315520182e-06, "loss": 1.3421, "step": 10954 }, { "epoch": 1.1656735475633113, "grad_norm": 12.4765625, "learning_rate": 9.628517623649802e-06, "loss": 1.0538, "step": 10955 }, { "epoch": 1.165779953181528, "grad_norm": 9.9140625, "learning_rate": 9.628482092097784e-06, "loss": 0.9342, "step": 10956 }, { "epoch": 1.1658863587997446, "grad_norm": 12.1328125, "learning_rate": 9.628446560545766e-06, "loss": 1.1181, "step": 10957 }, { "epoch": 1.1659927644179613, "grad_norm": 14.9296875, "learning_rate": 9.628411028993747e-06, "loss": 1.051, "step": 10958 }, { "epoch": 1.166099170036178, "grad_norm": 11.0234375, "learning_rate": 9.628375497441729e-06, "loss": 1.2815, "step": 10959 }, { "epoch": 1.1662055756543945, "grad_norm": 12.609375, "learning_rate": 9.62833996588971e-06, "loss": 1.0602, "step": 10960 }, { "epoch": 1.1663119812726113, "grad_norm": 13.515625, "learning_rate": 9.628304434337694e-06, "loss": 0.9933, "step": 10961 }, { "epoch": 1.1664183868908278, "grad_norm": 16.1875, "learning_rate": 9.628268902785675e-06, "loss": 1.0005, "step": 10962 }, { "epoch": 1.1665247925090445, "grad_norm": 12.46875, "learning_rate": 9.628233371233657e-06, "loss": 1.0018, "step": 10963 }, { "epoch": 1.166631198127261, "grad_norm": 12.8125, "learning_rate": 9.628197839681639e-06, "loss": 1.2447, "step": 10964 }, { "epoch": 1.1667376037454777, "grad_norm": 12.2578125, "learning_rate": 9.628162308129619e-06, "loss": 0.9879, "step": 10965 }, { "epoch": 1.1668440093636945, "grad_norm": 11.6484375, "learning_rate": 9.628126776577602e-06, "loss": 1.0883, "step": 10966 }, { "epoch": 1.166950414981911, "grad_norm": 12.203125, "learning_rate": 9.628091245025583e-06, "loss": 1.2427, "step": 10967 }, { "epoch": 1.1670568206001277, "grad_norm": 15.0546875, "learning_rate": 9.628055713473565e-06, "loss": 1.105, "step": 10968 }, { "epoch": 1.1671632262183445, "grad_norm": 14.4140625, "learning_rate": 9.628020181921547e-06, "loss": 1.0394, "step": 10969 }, { "epoch": 1.167269631836561, "grad_norm": 12.90625, "learning_rate": 9.627984650369528e-06, "loss": 1.0543, "step": 10970 }, { "epoch": 1.1673760374547777, "grad_norm": 18.0, "learning_rate": 9.627949118817512e-06, "loss": 0.9902, "step": 10971 }, { "epoch": 1.1674824430729942, "grad_norm": 13.546875, "learning_rate": 9.627913587265493e-06, "loss": 1.0839, "step": 10972 }, { "epoch": 1.167588848691211, "grad_norm": 13.734375, "learning_rate": 9.627878055713475e-06, "loss": 1.1248, "step": 10973 }, { "epoch": 1.1676952543094274, "grad_norm": 19.59375, "learning_rate": 9.627842524161456e-06, "loss": 1.0494, "step": 10974 }, { "epoch": 1.1678016599276442, "grad_norm": 10.90625, "learning_rate": 9.627806992609438e-06, "loss": 1.0644, "step": 10975 }, { "epoch": 1.167908065545861, "grad_norm": 10.0390625, "learning_rate": 9.62777146105742e-06, "loss": 1.2255, "step": 10976 }, { "epoch": 1.1680144711640774, "grad_norm": 13.7421875, "learning_rate": 9.627735929505401e-06, "loss": 1.2149, "step": 10977 }, { "epoch": 1.1681208767822941, "grad_norm": 14.875, "learning_rate": 9.627700397953383e-06, "loss": 1.0334, "step": 10978 }, { "epoch": 1.1682272824005107, "grad_norm": 12.1796875, "learning_rate": 9.627664866401364e-06, "loss": 1.0162, "step": 10979 }, { "epoch": 1.1683336880187274, "grad_norm": 13.5078125, "learning_rate": 9.627629334849346e-06, "loss": 1.0748, "step": 10980 }, { "epoch": 1.1684400936369441, "grad_norm": 13.3125, "learning_rate": 9.62759380329733e-06, "loss": 1.183, "step": 10981 }, { "epoch": 1.1685464992551606, "grad_norm": 15.8359375, "learning_rate": 9.627558271745311e-06, "loss": 1.1153, "step": 10982 }, { "epoch": 1.1686529048733774, "grad_norm": 12.4140625, "learning_rate": 9.627522740193293e-06, "loss": 1.23, "step": 10983 }, { "epoch": 1.1687593104915939, "grad_norm": 18.421875, "learning_rate": 9.627487208641274e-06, "loss": 1.2805, "step": 10984 }, { "epoch": 1.1688657161098106, "grad_norm": 11.7421875, "learning_rate": 9.627451677089256e-06, "loss": 1.1326, "step": 10985 }, { "epoch": 1.1689721217280273, "grad_norm": 32.625, "learning_rate": 9.627416145537239e-06, "loss": 1.0502, "step": 10986 }, { "epoch": 1.1690785273462438, "grad_norm": 17.015625, "learning_rate": 9.62738061398522e-06, "loss": 1.178, "step": 10987 }, { "epoch": 1.1691849329644606, "grad_norm": 11.3359375, "learning_rate": 9.627345082433202e-06, "loss": 1.1245, "step": 10988 }, { "epoch": 1.169291338582677, "grad_norm": 12.875, "learning_rate": 9.627309550881184e-06, "loss": 1.2338, "step": 10989 }, { "epoch": 1.1693977442008938, "grad_norm": 21.0, "learning_rate": 9.627274019329165e-06, "loss": 1.4105, "step": 10990 }, { "epoch": 1.1695041498191103, "grad_norm": 13.7265625, "learning_rate": 9.627238487777147e-06, "loss": 1.1999, "step": 10991 }, { "epoch": 1.169610555437327, "grad_norm": 14.84375, "learning_rate": 9.627202956225129e-06, "loss": 1.1036, "step": 10992 }, { "epoch": 1.1697169610555438, "grad_norm": 12.1328125, "learning_rate": 9.62716742467311e-06, "loss": 0.9689, "step": 10993 }, { "epoch": 1.1698233666737603, "grad_norm": 12.90625, "learning_rate": 9.627131893121092e-06, "loss": 1.2518, "step": 10994 }, { "epoch": 1.169929772291977, "grad_norm": 12.8984375, "learning_rate": 9.627096361569074e-06, "loss": 1.108, "step": 10995 }, { "epoch": 1.1700361779101938, "grad_norm": 10.5703125, "learning_rate": 9.627060830017055e-06, "loss": 0.8838, "step": 10996 }, { "epoch": 1.1701425835284103, "grad_norm": 11.4140625, "learning_rate": 9.627025298465038e-06, "loss": 0.9585, "step": 10997 }, { "epoch": 1.170248989146627, "grad_norm": 13.359375, "learning_rate": 9.62698976691302e-06, "loss": 0.9511, "step": 10998 }, { "epoch": 1.1703553947648435, "grad_norm": 11.0546875, "learning_rate": 9.626954235361002e-06, "loss": 1.1191, "step": 10999 }, { "epoch": 1.1704618003830602, "grad_norm": 13.6484375, "learning_rate": 9.626918703808983e-06, "loss": 1.0163, "step": 11000 }, { "epoch": 1.1705682060012768, "grad_norm": 11.3046875, "learning_rate": 9.626883172256965e-06, "loss": 1.0841, "step": 11001 }, { "epoch": 1.1706746116194935, "grad_norm": 12.1953125, "learning_rate": 9.626847640704946e-06, "loss": 1.0188, "step": 11002 }, { "epoch": 1.1707810172377102, "grad_norm": 13.828125, "learning_rate": 9.626812109152928e-06, "loss": 0.9794, "step": 11003 }, { "epoch": 1.1708874228559267, "grad_norm": 9.7890625, "learning_rate": 9.62677657760091e-06, "loss": 1.0407, "step": 11004 }, { "epoch": 1.1709938284741435, "grad_norm": 15.5703125, "learning_rate": 9.626741046048891e-06, "loss": 1.0917, "step": 11005 }, { "epoch": 1.1711002340923602, "grad_norm": 8.4609375, "learning_rate": 9.626705514496873e-06, "loss": 1.0706, "step": 11006 }, { "epoch": 1.1712066397105767, "grad_norm": 10.890625, "learning_rate": 9.626669982944856e-06, "loss": 0.915, "step": 11007 }, { "epoch": 1.1713130453287934, "grad_norm": 15.140625, "learning_rate": 9.626634451392838e-06, "loss": 1.1162, "step": 11008 }, { "epoch": 1.17141945094701, "grad_norm": 8.65625, "learning_rate": 9.62659891984082e-06, "loss": 0.8368, "step": 11009 }, { "epoch": 1.1715258565652267, "grad_norm": 17.6875, "learning_rate": 9.626563388288801e-06, "loss": 1.3155, "step": 11010 }, { "epoch": 1.1716322621834432, "grad_norm": 12.1328125, "learning_rate": 9.626527856736783e-06, "loss": 0.8622, "step": 11011 }, { "epoch": 1.17173866780166, "grad_norm": 12.3515625, "learning_rate": 9.626492325184766e-06, "loss": 1.1237, "step": 11012 }, { "epoch": 1.1718450734198766, "grad_norm": 11.9921875, "learning_rate": 9.626456793632748e-06, "loss": 0.9548, "step": 11013 }, { "epoch": 1.1719514790380932, "grad_norm": 13.0, "learning_rate": 9.626421262080729e-06, "loss": 1.1075, "step": 11014 }, { "epoch": 1.17205788465631, "grad_norm": 10.078125, "learning_rate": 9.62638573052871e-06, "loss": 1.1917, "step": 11015 }, { "epoch": 1.1721642902745264, "grad_norm": 17.4375, "learning_rate": 9.626350198976692e-06, "loss": 0.9636, "step": 11016 }, { "epoch": 1.1722706958927431, "grad_norm": 13.84375, "learning_rate": 9.626314667424674e-06, "loss": 0.9243, "step": 11017 }, { "epoch": 1.1723771015109599, "grad_norm": 13.8671875, "learning_rate": 9.626279135872656e-06, "loss": 1.2981, "step": 11018 }, { "epoch": 1.1724835071291764, "grad_norm": 12.5859375, "learning_rate": 9.626243604320637e-06, "loss": 1.2037, "step": 11019 }, { "epoch": 1.172589912747393, "grad_norm": 12.96875, "learning_rate": 9.626208072768619e-06, "loss": 0.9035, "step": 11020 }, { "epoch": 1.1726963183656096, "grad_norm": 10.1328125, "learning_rate": 9.6261725412166e-06, "loss": 0.9717, "step": 11021 }, { "epoch": 1.1728027239838263, "grad_norm": 12.5546875, "learning_rate": 9.626137009664584e-06, "loss": 1.3878, "step": 11022 }, { "epoch": 1.172909129602043, "grad_norm": 10.4921875, "learning_rate": 9.626101478112565e-06, "loss": 1.0758, "step": 11023 }, { "epoch": 1.1730155352202596, "grad_norm": 11.2421875, "learning_rate": 9.626065946560547e-06, "loss": 0.867, "step": 11024 }, { "epoch": 1.1731219408384763, "grad_norm": 8.59375, "learning_rate": 9.626030415008528e-06, "loss": 0.9117, "step": 11025 }, { "epoch": 1.1732283464566928, "grad_norm": 13.7890625, "learning_rate": 9.62599488345651e-06, "loss": 0.8618, "step": 11026 }, { "epoch": 1.1733347520749096, "grad_norm": 11.125, "learning_rate": 9.625959351904492e-06, "loss": 1.1605, "step": 11027 }, { "epoch": 1.1734411576931263, "grad_norm": 23.71875, "learning_rate": 9.625923820352473e-06, "loss": 1.052, "step": 11028 }, { "epoch": 1.1735475633113428, "grad_norm": 12.0, "learning_rate": 9.625888288800455e-06, "loss": 0.9678, "step": 11029 }, { "epoch": 1.1736539689295595, "grad_norm": 11.6015625, "learning_rate": 9.625852757248437e-06, "loss": 1.0214, "step": 11030 }, { "epoch": 1.173760374547776, "grad_norm": 14.7109375, "learning_rate": 9.625817225696418e-06, "loss": 1.2047, "step": 11031 }, { "epoch": 1.1738667801659928, "grad_norm": 13.4765625, "learning_rate": 9.625781694144401e-06, "loss": 1.0825, "step": 11032 }, { "epoch": 1.1739731857842095, "grad_norm": 14.265625, "learning_rate": 9.625746162592383e-06, "loss": 1.1702, "step": 11033 }, { "epoch": 1.174079591402426, "grad_norm": 19.390625, "learning_rate": 9.625710631040365e-06, "loss": 1.5484, "step": 11034 }, { "epoch": 1.1741859970206427, "grad_norm": 11.4375, "learning_rate": 9.625675099488346e-06, "loss": 0.9086, "step": 11035 }, { "epoch": 1.1742924026388593, "grad_norm": 11.234375, "learning_rate": 9.625639567936328e-06, "loss": 0.8826, "step": 11036 }, { "epoch": 1.174398808257076, "grad_norm": 10.1875, "learning_rate": 9.62560403638431e-06, "loss": 1.0177, "step": 11037 }, { "epoch": 1.1745052138752925, "grad_norm": 12.8984375, "learning_rate": 9.625568504832293e-06, "loss": 1.0341, "step": 11038 }, { "epoch": 1.1746116194935092, "grad_norm": 12.3203125, "learning_rate": 9.625532973280274e-06, "loss": 1.2215, "step": 11039 }, { "epoch": 1.174718025111726, "grad_norm": 20.078125, "learning_rate": 9.625497441728256e-06, "loss": 1.2287, "step": 11040 }, { "epoch": 1.1748244307299425, "grad_norm": 14.1953125, "learning_rate": 9.625461910176238e-06, "loss": 1.3021, "step": 11041 }, { "epoch": 1.1749308363481592, "grad_norm": 14.25, "learning_rate": 9.62542637862422e-06, "loss": 0.713, "step": 11042 }, { "epoch": 1.175037241966376, "grad_norm": 9.921875, "learning_rate": 9.6253908470722e-06, "loss": 1.0255, "step": 11043 }, { "epoch": 1.1751436475845924, "grad_norm": 14.734375, "learning_rate": 9.625355315520182e-06, "loss": 1.1551, "step": 11044 }, { "epoch": 1.1752500532028092, "grad_norm": 13.7265625, "learning_rate": 9.625319783968164e-06, "loss": 1.1618, "step": 11045 }, { "epoch": 1.1753564588210257, "grad_norm": 17.578125, "learning_rate": 9.625284252416146e-06, "loss": 1.0552, "step": 11046 }, { "epoch": 1.1754628644392424, "grad_norm": 18.96875, "learning_rate": 9.625248720864127e-06, "loss": 1.0376, "step": 11047 }, { "epoch": 1.175569270057459, "grad_norm": 16.59375, "learning_rate": 9.62521318931211e-06, "loss": 1.0464, "step": 11048 }, { "epoch": 1.1756756756756757, "grad_norm": 16.34375, "learning_rate": 9.625177657760092e-06, "loss": 1.0396, "step": 11049 }, { "epoch": 1.1757820812938924, "grad_norm": 10.5625, "learning_rate": 9.625142126208074e-06, "loss": 1.3282, "step": 11050 }, { "epoch": 1.175888486912109, "grad_norm": 11.3359375, "learning_rate": 9.625106594656055e-06, "loss": 0.9034, "step": 11051 }, { "epoch": 1.1759948925303256, "grad_norm": 10.7109375, "learning_rate": 9.625071063104037e-06, "loss": 1.2163, "step": 11052 }, { "epoch": 1.1761012981485424, "grad_norm": 11.4453125, "learning_rate": 9.62503553155202e-06, "loss": 1.1056, "step": 11053 }, { "epoch": 1.1762077037667589, "grad_norm": 15.6953125, "learning_rate": 9.625e-06, "loss": 1.21, "step": 11054 }, { "epoch": 1.1763141093849756, "grad_norm": 16.203125, "learning_rate": 9.624964468447982e-06, "loss": 0.9091, "step": 11055 }, { "epoch": 1.1764205150031921, "grad_norm": 13.84375, "learning_rate": 9.624928936895963e-06, "loss": 1.2905, "step": 11056 }, { "epoch": 1.1765269206214088, "grad_norm": 16.53125, "learning_rate": 9.624893405343945e-06, "loss": 1.0192, "step": 11057 }, { "epoch": 1.1766333262396254, "grad_norm": 9.84375, "learning_rate": 9.624857873791928e-06, "loss": 0.9511, "step": 11058 }, { "epoch": 1.176739731857842, "grad_norm": 10.515625, "learning_rate": 9.62482234223991e-06, "loss": 0.9166, "step": 11059 }, { "epoch": 1.1768461374760588, "grad_norm": 14.9140625, "learning_rate": 9.624786810687892e-06, "loss": 0.9567, "step": 11060 }, { "epoch": 1.1769525430942753, "grad_norm": 11.578125, "learning_rate": 9.624751279135873e-06, "loss": 1.4522, "step": 11061 }, { "epoch": 1.177058948712492, "grad_norm": 14.0390625, "learning_rate": 9.624715747583855e-06, "loss": 1.1324, "step": 11062 }, { "epoch": 1.1771653543307086, "grad_norm": 11.015625, "learning_rate": 9.624680216031838e-06, "loss": 1.1025, "step": 11063 }, { "epoch": 1.1772717599489253, "grad_norm": 10.3515625, "learning_rate": 9.62464468447982e-06, "loss": 1.0781, "step": 11064 }, { "epoch": 1.177378165567142, "grad_norm": 12.3671875, "learning_rate": 9.624609152927801e-06, "loss": 0.7034, "step": 11065 }, { "epoch": 1.1774845711853585, "grad_norm": 11.65625, "learning_rate": 9.624573621375783e-06, "loss": 1.0838, "step": 11066 }, { "epoch": 1.1775909768035753, "grad_norm": 17.765625, "learning_rate": 9.624538089823764e-06, "loss": 1.2332, "step": 11067 }, { "epoch": 1.1776973824217918, "grad_norm": 13.1171875, "learning_rate": 9.624502558271746e-06, "loss": 1.2078, "step": 11068 }, { "epoch": 1.1778037880400085, "grad_norm": 18.171875, "learning_rate": 9.624467026719728e-06, "loss": 1.2372, "step": 11069 }, { "epoch": 1.1779101936582252, "grad_norm": 13.328125, "learning_rate": 9.62443149516771e-06, "loss": 1.0306, "step": 11070 }, { "epoch": 1.1780165992764418, "grad_norm": 12.7265625, "learning_rate": 9.624395963615691e-06, "loss": 1.019, "step": 11071 }, { "epoch": 1.1781230048946585, "grad_norm": 12.7734375, "learning_rate": 9.624360432063672e-06, "loss": 1.0003, "step": 11072 }, { "epoch": 1.178229410512875, "grad_norm": 13.03125, "learning_rate": 9.624324900511656e-06, "loss": 0.9722, "step": 11073 }, { "epoch": 1.1783358161310917, "grad_norm": 10.1875, "learning_rate": 9.624289368959637e-06, "loss": 0.9255, "step": 11074 }, { "epoch": 1.1784422217493085, "grad_norm": 9.5625, "learning_rate": 9.624253837407619e-06, "loss": 0.8617, "step": 11075 }, { "epoch": 1.178548627367525, "grad_norm": 10.65625, "learning_rate": 9.6242183058556e-06, "loss": 1.0873, "step": 11076 }, { "epoch": 1.1786550329857417, "grad_norm": 10.703125, "learning_rate": 9.624182774303582e-06, "loss": 0.8019, "step": 11077 }, { "epoch": 1.1787614386039582, "grad_norm": 11.9765625, "learning_rate": 9.624147242751564e-06, "loss": 1.1552, "step": 11078 }, { "epoch": 1.178867844222175, "grad_norm": 11.65625, "learning_rate": 9.624111711199547e-06, "loss": 1.1122, "step": 11079 }, { "epoch": 1.1789742498403917, "grad_norm": 12.03125, "learning_rate": 9.624076179647529e-06, "loss": 1.0354, "step": 11080 }, { "epoch": 1.1790806554586082, "grad_norm": 10.296875, "learning_rate": 9.62404064809551e-06, "loss": 0.9235, "step": 11081 }, { "epoch": 1.179187061076825, "grad_norm": 9.9921875, "learning_rate": 9.624005116543492e-06, "loss": 1.1809, "step": 11082 }, { "epoch": 1.1792934666950414, "grad_norm": 10.609375, "learning_rate": 9.623969584991474e-06, "loss": 0.9977, "step": 11083 }, { "epoch": 1.1793998723132582, "grad_norm": 11.3515625, "learning_rate": 9.623934053439455e-06, "loss": 1.1163, "step": 11084 }, { "epoch": 1.1795062779314747, "grad_norm": 15.953125, "learning_rate": 9.623898521887437e-06, "loss": 1.4465, "step": 11085 }, { "epoch": 1.1796126835496914, "grad_norm": 12.609375, "learning_rate": 9.623862990335418e-06, "loss": 1.1236, "step": 11086 }, { "epoch": 1.1797190891679081, "grad_norm": 9.015625, "learning_rate": 9.6238274587834e-06, "loss": 0.9797, "step": 11087 }, { "epoch": 1.1798254947861246, "grad_norm": 14.4375, "learning_rate": 9.623791927231382e-06, "loss": 1.0018, "step": 11088 }, { "epoch": 1.1799319004043414, "grad_norm": 12.6953125, "learning_rate": 9.623756395679365e-06, "loss": 1.0768, "step": 11089 }, { "epoch": 1.180038306022558, "grad_norm": 13.9140625, "learning_rate": 9.623720864127346e-06, "loss": 1.3847, "step": 11090 }, { "epoch": 1.1801447116407746, "grad_norm": 11.6796875, "learning_rate": 9.623685332575328e-06, "loss": 1.1179, "step": 11091 }, { "epoch": 1.1802511172589913, "grad_norm": 10.796875, "learning_rate": 9.62364980102331e-06, "loss": 0.8485, "step": 11092 }, { "epoch": 1.1803575228772079, "grad_norm": 15.5, "learning_rate": 9.623614269471291e-06, "loss": 1.1363, "step": 11093 }, { "epoch": 1.1804639284954246, "grad_norm": 16.859375, "learning_rate": 9.623578737919273e-06, "loss": 1.0456, "step": 11094 }, { "epoch": 1.180570334113641, "grad_norm": 20.328125, "learning_rate": 9.623543206367255e-06, "loss": 1.2073, "step": 11095 }, { "epoch": 1.1806767397318578, "grad_norm": 21.5, "learning_rate": 9.623507674815236e-06, "loss": 0.9674, "step": 11096 }, { "epoch": 1.1807831453500746, "grad_norm": 9.1015625, "learning_rate": 9.623472143263218e-06, "loss": 0.9594, "step": 11097 }, { "epoch": 1.180889550968291, "grad_norm": 12.953125, "learning_rate": 9.6234366117112e-06, "loss": 1.1032, "step": 11098 }, { "epoch": 1.1809959565865078, "grad_norm": 11.28125, "learning_rate": 9.623401080159183e-06, "loss": 1.362, "step": 11099 }, { "epoch": 1.1811023622047245, "grad_norm": 10.6796875, "learning_rate": 9.623365548607164e-06, "loss": 1.0607, "step": 11100 }, { "epoch": 1.181208767822941, "grad_norm": 18.234375, "learning_rate": 9.623330017055146e-06, "loss": 1.1789, "step": 11101 }, { "epoch": 1.1813151734411578, "grad_norm": 10.96875, "learning_rate": 9.623294485503127e-06, "loss": 1.1441, "step": 11102 }, { "epoch": 1.1814215790593743, "grad_norm": 11.640625, "learning_rate": 9.623258953951109e-06, "loss": 1.0958, "step": 11103 }, { "epoch": 1.181527984677591, "grad_norm": 11.9453125, "learning_rate": 9.623223422399092e-06, "loss": 1.1095, "step": 11104 }, { "epoch": 1.1816343902958075, "grad_norm": 18.015625, "learning_rate": 9.623187890847074e-06, "loss": 1.5578, "step": 11105 }, { "epoch": 1.1817407959140243, "grad_norm": 15.78125, "learning_rate": 9.623152359295056e-06, "loss": 1.0747, "step": 11106 }, { "epoch": 1.181847201532241, "grad_norm": 15.796875, "learning_rate": 9.623116827743037e-06, "loss": 1.0312, "step": 11107 }, { "epoch": 1.1819536071504575, "grad_norm": 11.5546875, "learning_rate": 9.623081296191019e-06, "loss": 1.0882, "step": 11108 }, { "epoch": 1.1820600127686742, "grad_norm": 12.5234375, "learning_rate": 9.623045764639e-06, "loss": 1.2428, "step": 11109 }, { "epoch": 1.1821664183868907, "grad_norm": 13.21875, "learning_rate": 9.623010233086982e-06, "loss": 1.0436, "step": 11110 }, { "epoch": 1.1822728240051075, "grad_norm": 13.171875, "learning_rate": 9.622974701534964e-06, "loss": 1.2184, "step": 11111 }, { "epoch": 1.1823792296233242, "grad_norm": 15.1640625, "learning_rate": 9.622939169982945e-06, "loss": 0.9191, "step": 11112 }, { "epoch": 1.1824856352415407, "grad_norm": 11.9140625, "learning_rate": 9.622903638430927e-06, "loss": 1.1959, "step": 11113 }, { "epoch": 1.1825920408597574, "grad_norm": 12.640625, "learning_rate": 9.622868106878908e-06, "loss": 1.0813, "step": 11114 }, { "epoch": 1.182698446477974, "grad_norm": 9.5859375, "learning_rate": 9.622832575326892e-06, "loss": 0.8487, "step": 11115 }, { "epoch": 1.1828048520961907, "grad_norm": 11.796875, "learning_rate": 9.622797043774873e-06, "loss": 1.2299, "step": 11116 }, { "epoch": 1.1829112577144074, "grad_norm": 13.6171875, "learning_rate": 9.622761512222855e-06, "loss": 0.9017, "step": 11117 }, { "epoch": 1.183017663332624, "grad_norm": 12.0234375, "learning_rate": 9.622725980670837e-06, "loss": 1.0552, "step": 11118 }, { "epoch": 1.1831240689508407, "grad_norm": 14.8828125, "learning_rate": 9.622690449118818e-06, "loss": 1.1407, "step": 11119 }, { "epoch": 1.1832304745690572, "grad_norm": 13.734375, "learning_rate": 9.6226549175668e-06, "loss": 1.2034, "step": 11120 }, { "epoch": 1.183336880187274, "grad_norm": 15.9921875, "learning_rate": 9.622619386014781e-06, "loss": 1.2241, "step": 11121 }, { "epoch": 1.1834432858054904, "grad_norm": 9.8828125, "learning_rate": 9.622583854462763e-06, "loss": 1.1104, "step": 11122 }, { "epoch": 1.1835496914237071, "grad_norm": 10.4140625, "learning_rate": 9.622548322910745e-06, "loss": 1.1319, "step": 11123 }, { "epoch": 1.1836560970419239, "grad_norm": 15.7890625, "learning_rate": 9.622512791358726e-06, "loss": 1.0662, "step": 11124 }, { "epoch": 1.1837625026601404, "grad_norm": 15.2265625, "learning_rate": 9.62247725980671e-06, "loss": 1.205, "step": 11125 }, { "epoch": 1.1838689082783571, "grad_norm": 11.4453125, "learning_rate": 9.622441728254691e-06, "loss": 1.1352, "step": 11126 }, { "epoch": 1.1839753138965738, "grad_norm": 17.125, "learning_rate": 9.622406196702673e-06, "loss": 1.386, "step": 11127 }, { "epoch": 1.1840817195147904, "grad_norm": 11.078125, "learning_rate": 9.622370665150654e-06, "loss": 1.1186, "step": 11128 }, { "epoch": 1.184188125133007, "grad_norm": 11.15625, "learning_rate": 9.622335133598636e-06, "loss": 1.1653, "step": 11129 }, { "epoch": 1.1842945307512236, "grad_norm": 10.171875, "learning_rate": 9.62229960204662e-06, "loss": 1.0246, "step": 11130 }, { "epoch": 1.1844009363694403, "grad_norm": 15.8515625, "learning_rate": 9.6222640704946e-06, "loss": 0.9428, "step": 11131 }, { "epoch": 1.1845073419876568, "grad_norm": 10.296875, "learning_rate": 9.622228538942582e-06, "loss": 0.7159, "step": 11132 }, { "epoch": 1.1846137476058736, "grad_norm": 12.2890625, "learning_rate": 9.622193007390564e-06, "loss": 1.1138, "step": 11133 }, { "epoch": 1.1847201532240903, "grad_norm": 10.453125, "learning_rate": 9.622157475838546e-06, "loss": 0.7571, "step": 11134 }, { "epoch": 1.1848265588423068, "grad_norm": 12.3828125, "learning_rate": 9.622121944286527e-06, "loss": 0.9178, "step": 11135 }, { "epoch": 1.1849329644605235, "grad_norm": 13.6796875, "learning_rate": 9.622086412734509e-06, "loss": 1.1822, "step": 11136 }, { "epoch": 1.1850393700787403, "grad_norm": 13.7421875, "learning_rate": 9.62205088118249e-06, "loss": 1.2788, "step": 11137 }, { "epoch": 1.1851457756969568, "grad_norm": 13.671875, "learning_rate": 9.622015349630472e-06, "loss": 1.241, "step": 11138 }, { "epoch": 1.1852521813151735, "grad_norm": 12.7734375, "learning_rate": 9.621979818078454e-06, "loss": 1.08, "step": 11139 }, { "epoch": 1.18535858693339, "grad_norm": 14.109375, "learning_rate": 9.621944286526437e-06, "loss": 1.2026, "step": 11140 }, { "epoch": 1.1854649925516068, "grad_norm": 12.1171875, "learning_rate": 9.621908754974419e-06, "loss": 1.1093, "step": 11141 }, { "epoch": 1.1855713981698233, "grad_norm": 12.59375, "learning_rate": 9.6218732234224e-06, "loss": 1.0916, "step": 11142 }, { "epoch": 1.18567780378804, "grad_norm": 8.84375, "learning_rate": 9.621837691870382e-06, "loss": 1.2405, "step": 11143 }, { "epoch": 1.1857842094062567, "grad_norm": 9.515625, "learning_rate": 9.621802160318363e-06, "loss": 0.9484, "step": 11144 }, { "epoch": 1.1858906150244732, "grad_norm": 12.15625, "learning_rate": 9.621766628766345e-06, "loss": 1.227, "step": 11145 }, { "epoch": 1.18599702064269, "grad_norm": 19.40625, "learning_rate": 9.621731097214327e-06, "loss": 0.9493, "step": 11146 }, { "epoch": 1.1861034262609065, "grad_norm": 11.4765625, "learning_rate": 9.621695565662308e-06, "loss": 1.1763, "step": 11147 }, { "epoch": 1.1862098318791232, "grad_norm": 12.015625, "learning_rate": 9.62166003411029e-06, "loss": 1.2401, "step": 11148 }, { "epoch": 1.18631623749734, "grad_norm": 8.4921875, "learning_rate": 9.621624502558271e-06, "loss": 1.1133, "step": 11149 }, { "epoch": 1.1864226431155565, "grad_norm": 14.6015625, "learning_rate": 9.621588971006255e-06, "loss": 1.1046, "step": 11150 }, { "epoch": 1.1865290487337732, "grad_norm": 11.84375, "learning_rate": 9.621553439454236e-06, "loss": 1.1426, "step": 11151 }, { "epoch": 1.1866354543519897, "grad_norm": 10.1875, "learning_rate": 9.621517907902218e-06, "loss": 1.0, "step": 11152 }, { "epoch": 1.1867418599702064, "grad_norm": 13.34375, "learning_rate": 9.6214823763502e-06, "loss": 1.3201, "step": 11153 }, { "epoch": 1.1868482655884232, "grad_norm": 13.15625, "learning_rate": 9.621446844798181e-06, "loss": 1.0094, "step": 11154 }, { "epoch": 1.1869546712066397, "grad_norm": 14.8828125, "learning_rate": 9.621411313246163e-06, "loss": 1.415, "step": 11155 }, { "epoch": 1.1870610768248564, "grad_norm": 12.84375, "learning_rate": 9.621375781694146e-06, "loss": 0.8203, "step": 11156 }, { "epoch": 1.187167482443073, "grad_norm": 11.5703125, "learning_rate": 9.621340250142128e-06, "loss": 1.1895, "step": 11157 }, { "epoch": 1.1872738880612896, "grad_norm": 12.4296875, "learning_rate": 9.62130471859011e-06, "loss": 1.1702, "step": 11158 }, { "epoch": 1.1873802936795064, "grad_norm": 13.03125, "learning_rate": 9.621269187038091e-06, "loss": 1.3361, "step": 11159 }, { "epoch": 1.1874866992977229, "grad_norm": 12.25, "learning_rate": 9.621233655486073e-06, "loss": 0.9443, "step": 11160 }, { "epoch": 1.1875931049159396, "grad_norm": 10.1015625, "learning_rate": 9.621198123934054e-06, "loss": 1.0912, "step": 11161 }, { "epoch": 1.1876995105341561, "grad_norm": 9.453125, "learning_rate": 9.621162592382036e-06, "loss": 1.0083, "step": 11162 }, { "epoch": 1.1878059161523729, "grad_norm": 9.8359375, "learning_rate": 9.621127060830017e-06, "loss": 1.0203, "step": 11163 }, { "epoch": 1.1879123217705896, "grad_norm": 11.8046875, "learning_rate": 9.621091529277999e-06, "loss": 1.0332, "step": 11164 }, { "epoch": 1.188018727388806, "grad_norm": 10.3125, "learning_rate": 9.62105599772598e-06, "loss": 0.8426, "step": 11165 }, { "epoch": 1.1881251330070228, "grad_norm": 11.1484375, "learning_rate": 9.621020466173964e-06, "loss": 1.0734, "step": 11166 }, { "epoch": 1.1882315386252393, "grad_norm": 14.1015625, "learning_rate": 9.620984934621945e-06, "loss": 1.5119, "step": 11167 }, { "epoch": 1.188337944243456, "grad_norm": 12.625, "learning_rate": 9.620949403069927e-06, "loss": 1.3346, "step": 11168 }, { "epoch": 1.1884443498616726, "grad_norm": 14.40625, "learning_rate": 9.620913871517909e-06, "loss": 1.3808, "step": 11169 }, { "epoch": 1.1885507554798893, "grad_norm": 15.1640625, "learning_rate": 9.62087833996589e-06, "loss": 1.25, "step": 11170 }, { "epoch": 1.188657161098106, "grad_norm": 16.53125, "learning_rate": 9.620842808413874e-06, "loss": 1.188, "step": 11171 }, { "epoch": 1.1887635667163226, "grad_norm": 10.3515625, "learning_rate": 9.620807276861855e-06, "loss": 0.9405, "step": 11172 }, { "epoch": 1.1888699723345393, "grad_norm": 11.78125, "learning_rate": 9.620771745309837e-06, "loss": 1.1192, "step": 11173 }, { "epoch": 1.188976377952756, "grad_norm": 10.5, "learning_rate": 9.620736213757817e-06, "loss": 1.0884, "step": 11174 }, { "epoch": 1.1890827835709725, "grad_norm": 15.140625, "learning_rate": 9.620700682205798e-06, "loss": 0.9024, "step": 11175 }, { "epoch": 1.1891891891891893, "grad_norm": 11.9296875, "learning_rate": 9.620665150653782e-06, "loss": 1.0572, "step": 11176 }, { "epoch": 1.1892955948074058, "grad_norm": 13.9765625, "learning_rate": 9.620629619101763e-06, "loss": 1.1652, "step": 11177 }, { "epoch": 1.1894020004256225, "grad_norm": 13.4609375, "learning_rate": 9.620594087549745e-06, "loss": 1.331, "step": 11178 }, { "epoch": 1.189508406043839, "grad_norm": 12.015625, "learning_rate": 9.620558555997726e-06, "loss": 1.0508, "step": 11179 }, { "epoch": 1.1896148116620557, "grad_norm": 10.9296875, "learning_rate": 9.620523024445708e-06, "loss": 0.9906, "step": 11180 }, { "epoch": 1.1897212172802725, "grad_norm": 14.375, "learning_rate": 9.620487492893691e-06, "loss": 0.9546, "step": 11181 }, { "epoch": 1.189827622898489, "grad_norm": 10.9140625, "learning_rate": 9.620451961341673e-06, "loss": 1.4049, "step": 11182 }, { "epoch": 1.1899340285167057, "grad_norm": 11.84375, "learning_rate": 9.620416429789655e-06, "loss": 1.0298, "step": 11183 }, { "epoch": 1.1900404341349224, "grad_norm": 10.0859375, "learning_rate": 9.620380898237636e-06, "loss": 1.0491, "step": 11184 }, { "epoch": 1.190146839753139, "grad_norm": 15.5703125, "learning_rate": 9.620345366685618e-06, "loss": 0.9174, "step": 11185 }, { "epoch": 1.1902532453713557, "grad_norm": 14.6484375, "learning_rate": 9.6203098351336e-06, "loss": 1.01, "step": 11186 }, { "epoch": 1.1903596509895722, "grad_norm": 11.6796875, "learning_rate": 9.620274303581581e-06, "loss": 1.2141, "step": 11187 }, { "epoch": 1.190466056607789, "grad_norm": 13.984375, "learning_rate": 9.620238772029563e-06, "loss": 1.233, "step": 11188 }, { "epoch": 1.1905724622260054, "grad_norm": 13.3515625, "learning_rate": 9.620203240477544e-06, "loss": 1.1634, "step": 11189 }, { "epoch": 1.1906788678442222, "grad_norm": 10.5625, "learning_rate": 9.620167708925526e-06, "loss": 1.0181, "step": 11190 }, { "epoch": 1.190785273462439, "grad_norm": 16.75, "learning_rate": 9.620132177373507e-06, "loss": 1.2315, "step": 11191 }, { "epoch": 1.1908916790806554, "grad_norm": 15.0546875, "learning_rate": 9.62009664582149e-06, "loss": 1.3177, "step": 11192 }, { "epoch": 1.1909980846988721, "grad_norm": 14.046875, "learning_rate": 9.620061114269472e-06, "loss": 1.2373, "step": 11193 }, { "epoch": 1.1911044903170886, "grad_norm": 13.484375, "learning_rate": 9.620025582717454e-06, "loss": 1.4514, "step": 11194 }, { "epoch": 1.1912108959353054, "grad_norm": 14.1015625, "learning_rate": 9.619990051165436e-06, "loss": 1.0329, "step": 11195 }, { "epoch": 1.1913173015535221, "grad_norm": 14.3046875, "learning_rate": 9.619954519613417e-06, "loss": 1.0879, "step": 11196 }, { "epoch": 1.1914237071717386, "grad_norm": 13.1875, "learning_rate": 9.6199189880614e-06, "loss": 0.9667, "step": 11197 }, { "epoch": 1.1915301127899554, "grad_norm": 11.6796875, "learning_rate": 9.619883456509382e-06, "loss": 1.1849, "step": 11198 }, { "epoch": 1.1916365184081719, "grad_norm": 16.046875, "learning_rate": 9.619847924957364e-06, "loss": 1.1118, "step": 11199 }, { "epoch": 1.1917429240263886, "grad_norm": 13.703125, "learning_rate": 9.619812393405345e-06, "loss": 1.0514, "step": 11200 }, { "epoch": 1.1918493296446053, "grad_norm": 12.90625, "learning_rate": 9.619776861853327e-06, "loss": 0.8997, "step": 11201 }, { "epoch": 1.1919557352628218, "grad_norm": 13.6328125, "learning_rate": 9.619741330301308e-06, "loss": 1.1937, "step": 11202 }, { "epoch": 1.1920621408810386, "grad_norm": 11.4375, "learning_rate": 9.61970579874929e-06, "loss": 1.0334, "step": 11203 }, { "epoch": 1.192168546499255, "grad_norm": 15.265625, "learning_rate": 9.619670267197272e-06, "loss": 1.0416, "step": 11204 }, { "epoch": 1.1922749521174718, "grad_norm": 17.125, "learning_rate": 9.619634735645253e-06, "loss": 1.1113, "step": 11205 }, { "epoch": 1.1923813577356885, "grad_norm": 12.96875, "learning_rate": 9.619599204093235e-06, "loss": 1.0469, "step": 11206 }, { "epoch": 1.192487763353905, "grad_norm": 11.8046875, "learning_rate": 9.619563672541218e-06, "loss": 0.8242, "step": 11207 }, { "epoch": 1.1925941689721218, "grad_norm": 12.28125, "learning_rate": 9.6195281409892e-06, "loss": 1.1592, "step": 11208 }, { "epoch": 1.1927005745903383, "grad_norm": 15.3515625, "learning_rate": 9.619492609437181e-06, "loss": 0.9237, "step": 11209 }, { "epoch": 1.192806980208555, "grad_norm": 14.6328125, "learning_rate": 9.619457077885163e-06, "loss": 0.8439, "step": 11210 }, { "epoch": 1.1929133858267718, "grad_norm": 10.921875, "learning_rate": 9.619421546333145e-06, "loss": 1.2623, "step": 11211 }, { "epoch": 1.1930197914449883, "grad_norm": 14.6796875, "learning_rate": 9.619386014781126e-06, "loss": 1.1941, "step": 11212 }, { "epoch": 1.193126197063205, "grad_norm": 11.3515625, "learning_rate": 9.619350483229108e-06, "loss": 1.1783, "step": 11213 }, { "epoch": 1.1932326026814215, "grad_norm": 9.8984375, "learning_rate": 9.61931495167709e-06, "loss": 1.1945, "step": 11214 }, { "epoch": 1.1933390082996382, "grad_norm": 14.9296875, "learning_rate": 9.619279420125071e-06, "loss": 1.0395, "step": 11215 }, { "epoch": 1.1934454139178547, "grad_norm": 10.7421875, "learning_rate": 9.619243888573053e-06, "loss": 1.1307, "step": 11216 }, { "epoch": 1.1935518195360715, "grad_norm": 15.3046875, "learning_rate": 9.619208357021036e-06, "loss": 1.1579, "step": 11217 }, { "epoch": 1.1936582251542882, "grad_norm": 9.671875, "learning_rate": 9.619172825469018e-06, "loss": 0.8697, "step": 11218 }, { "epoch": 1.1937646307725047, "grad_norm": 13.2109375, "learning_rate": 9.619137293917e-06, "loss": 1.2714, "step": 11219 }, { "epoch": 1.1938710363907215, "grad_norm": 11.5859375, "learning_rate": 9.61910176236498e-06, "loss": 0.9932, "step": 11220 }, { "epoch": 1.1939774420089382, "grad_norm": 12.5625, "learning_rate": 9.619066230812962e-06, "loss": 1.0709, "step": 11221 }, { "epoch": 1.1940838476271547, "grad_norm": 10.140625, "learning_rate": 9.619030699260946e-06, "loss": 1.1943, "step": 11222 }, { "epoch": 1.1941902532453714, "grad_norm": 12.3671875, "learning_rate": 9.618995167708927e-06, "loss": 1.0128, "step": 11223 }, { "epoch": 1.194296658863588, "grad_norm": 11.8984375, "learning_rate": 9.618959636156909e-06, "loss": 1.1462, "step": 11224 }, { "epoch": 1.1944030644818047, "grad_norm": 11.9140625, "learning_rate": 9.61892410460489e-06, "loss": 0.981, "step": 11225 }, { "epoch": 1.1945094701000212, "grad_norm": 10.2578125, "learning_rate": 9.618888573052872e-06, "loss": 1.0437, "step": 11226 }, { "epoch": 1.194615875718238, "grad_norm": 14.5234375, "learning_rate": 9.618853041500854e-06, "loss": 1.0217, "step": 11227 }, { "epoch": 1.1947222813364546, "grad_norm": 14.5703125, "learning_rate": 9.618817509948835e-06, "loss": 1.0281, "step": 11228 }, { "epoch": 1.1948286869546711, "grad_norm": 25.65625, "learning_rate": 9.618781978396817e-06, "loss": 1.1442, "step": 11229 }, { "epoch": 1.1949350925728879, "grad_norm": 13.1796875, "learning_rate": 9.618746446844799e-06, "loss": 1.1476, "step": 11230 }, { "epoch": 1.1950414981911046, "grad_norm": 13.5625, "learning_rate": 9.61871091529278e-06, "loss": 1.0911, "step": 11231 }, { "epoch": 1.1951479038093211, "grad_norm": 13.6875, "learning_rate": 9.618675383740762e-06, "loss": 1.3595, "step": 11232 }, { "epoch": 1.1952543094275379, "grad_norm": 12.9296875, "learning_rate": 9.618639852188745e-06, "loss": 1.0883, "step": 11233 }, { "epoch": 1.1953607150457544, "grad_norm": 10.6640625, "learning_rate": 9.618604320636727e-06, "loss": 1.1, "step": 11234 }, { "epoch": 1.195467120663971, "grad_norm": 13.546875, "learning_rate": 9.618568789084708e-06, "loss": 1.215, "step": 11235 }, { "epoch": 1.1955735262821876, "grad_norm": 11.2109375, "learning_rate": 9.61853325753269e-06, "loss": 1.0906, "step": 11236 }, { "epoch": 1.1956799319004043, "grad_norm": 10.1796875, "learning_rate": 9.618497725980671e-06, "loss": 1.0956, "step": 11237 }, { "epoch": 1.195786337518621, "grad_norm": 11.3046875, "learning_rate": 9.618462194428653e-06, "loss": 1.3898, "step": 11238 }, { "epoch": 1.1958927431368376, "grad_norm": 13.6953125, "learning_rate": 9.618426662876635e-06, "loss": 1.275, "step": 11239 }, { "epoch": 1.1959991487550543, "grad_norm": 10.6796875, "learning_rate": 9.618391131324616e-06, "loss": 1.2202, "step": 11240 }, { "epoch": 1.1961055543732708, "grad_norm": 9.5859375, "learning_rate": 9.618355599772598e-06, "loss": 1.1912, "step": 11241 }, { "epoch": 1.1962119599914876, "grad_norm": 9.6640625, "learning_rate": 9.61832006822058e-06, "loss": 1.0396, "step": 11242 }, { "epoch": 1.1963183656097043, "grad_norm": 11.90625, "learning_rate": 9.618284536668563e-06, "loss": 0.9755, "step": 11243 }, { "epoch": 1.1964247712279208, "grad_norm": 10.2421875, "learning_rate": 9.618249005116544e-06, "loss": 0.9399, "step": 11244 }, { "epoch": 1.1965311768461375, "grad_norm": 10.3125, "learning_rate": 9.618213473564526e-06, "loss": 1.0073, "step": 11245 }, { "epoch": 1.196637582464354, "grad_norm": 9.7734375, "learning_rate": 9.618177942012508e-06, "loss": 0.9427, "step": 11246 }, { "epoch": 1.1967439880825708, "grad_norm": 12.9375, "learning_rate": 9.61814241046049e-06, "loss": 0.9371, "step": 11247 }, { "epoch": 1.1968503937007875, "grad_norm": 11.984375, "learning_rate": 9.618106878908473e-06, "loss": 1.2681, "step": 11248 }, { "epoch": 1.196956799319004, "grad_norm": 12.984375, "learning_rate": 9.618071347356454e-06, "loss": 1.1728, "step": 11249 }, { "epoch": 1.1970632049372207, "grad_norm": 12.046875, "learning_rate": 9.618035815804436e-06, "loss": 1.099, "step": 11250 }, { "epoch": 1.1971696105554372, "grad_norm": 13.0390625, "learning_rate": 9.618000284252417e-06, "loss": 1.3472, "step": 11251 }, { "epoch": 1.197276016173654, "grad_norm": 15.3359375, "learning_rate": 9.617964752700399e-06, "loss": 1.166, "step": 11252 }, { "epoch": 1.1973824217918705, "grad_norm": 11.03125, "learning_rate": 9.61792922114838e-06, "loss": 1.0037, "step": 11253 }, { "epoch": 1.1974888274100872, "grad_norm": 15.8515625, "learning_rate": 9.617893689596362e-06, "loss": 1.2023, "step": 11254 }, { "epoch": 1.197595233028304, "grad_norm": 15.4921875, "learning_rate": 9.617858158044344e-06, "loss": 0.906, "step": 11255 }, { "epoch": 1.1977016386465205, "grad_norm": 14.4140625, "learning_rate": 9.617822626492325e-06, "loss": 1.103, "step": 11256 }, { "epoch": 1.1978080442647372, "grad_norm": 11.953125, "learning_rate": 9.617787094940307e-06, "loss": 1.0451, "step": 11257 }, { "epoch": 1.197914449882954, "grad_norm": 13.03125, "learning_rate": 9.61775156338829e-06, "loss": 0.9444, "step": 11258 }, { "epoch": 1.1980208555011704, "grad_norm": 13.2265625, "learning_rate": 9.617716031836272e-06, "loss": 0.9094, "step": 11259 }, { "epoch": 1.1981272611193872, "grad_norm": 11.390625, "learning_rate": 9.617680500284253e-06, "loss": 1.0304, "step": 11260 }, { "epoch": 1.1982336667376037, "grad_norm": 16.046875, "learning_rate": 9.617644968732235e-06, "loss": 1.106, "step": 11261 }, { "epoch": 1.1983400723558204, "grad_norm": 14.0, "learning_rate": 9.617609437180217e-06, "loss": 1.059, "step": 11262 }, { "epoch": 1.198446477974037, "grad_norm": 10.0625, "learning_rate": 9.617573905628198e-06, "loss": 1.0533, "step": 11263 }, { "epoch": 1.1985528835922536, "grad_norm": 11.1015625, "learning_rate": 9.61753837407618e-06, "loss": 0.8651, "step": 11264 }, { "epoch": 1.1986592892104704, "grad_norm": 10.3203125, "learning_rate": 9.617502842524162e-06, "loss": 1.1872, "step": 11265 }, { "epoch": 1.198765694828687, "grad_norm": 15.7421875, "learning_rate": 9.617467310972143e-06, "loss": 1.2716, "step": 11266 }, { "epoch": 1.1988721004469036, "grad_norm": 13.0, "learning_rate": 9.617431779420125e-06, "loss": 1.2565, "step": 11267 }, { "epoch": 1.1989785060651204, "grad_norm": 18.078125, "learning_rate": 9.617396247868108e-06, "loss": 1.2138, "step": 11268 }, { "epoch": 1.1990849116833369, "grad_norm": 10.921875, "learning_rate": 9.61736071631609e-06, "loss": 1.1927, "step": 11269 }, { "epoch": 1.1991913173015536, "grad_norm": 15.5390625, "learning_rate": 9.617325184764071e-06, "loss": 1.0275, "step": 11270 }, { "epoch": 1.19929772291977, "grad_norm": 9.578125, "learning_rate": 9.617289653212053e-06, "loss": 1.1221, "step": 11271 }, { "epoch": 1.1994041285379868, "grad_norm": 18.6875, "learning_rate": 9.617254121660034e-06, "loss": 1.0944, "step": 11272 }, { "epoch": 1.1995105341562033, "grad_norm": 11.7265625, "learning_rate": 9.617218590108016e-06, "loss": 1.1426, "step": 11273 }, { "epoch": 1.19961693977442, "grad_norm": 10.640625, "learning_rate": 9.617183058556e-06, "loss": 0.9636, "step": 11274 }, { "epoch": 1.1997233453926368, "grad_norm": 9.5234375, "learning_rate": 9.617147527003981e-06, "loss": 0.9955, "step": 11275 }, { "epoch": 1.1998297510108533, "grad_norm": 12.9453125, "learning_rate": 9.617111995451963e-06, "loss": 1.094, "step": 11276 }, { "epoch": 1.19993615662907, "grad_norm": 12.859375, "learning_rate": 9.617076463899944e-06, "loss": 1.1885, "step": 11277 }, { "epoch": 1.2000425622472868, "grad_norm": 11.4765625, "learning_rate": 9.617040932347926e-06, "loss": 1.0745, "step": 11278 }, { "epoch": 1.2001489678655033, "grad_norm": 15.1171875, "learning_rate": 9.617005400795907e-06, "loss": 1.0903, "step": 11279 }, { "epoch": 1.20025537348372, "grad_norm": 12.8359375, "learning_rate": 9.616969869243889e-06, "loss": 1.2704, "step": 11280 }, { "epoch": 1.2003617791019365, "grad_norm": 10.4453125, "learning_rate": 9.61693433769187e-06, "loss": 0.9999, "step": 11281 }, { "epoch": 1.2004681847201533, "grad_norm": 16.28125, "learning_rate": 9.616898806139852e-06, "loss": 1.3582, "step": 11282 }, { "epoch": 1.2005745903383698, "grad_norm": 11.8046875, "learning_rate": 9.616863274587834e-06, "loss": 1.0426, "step": 11283 }, { "epoch": 1.2006809959565865, "grad_norm": 11.7890625, "learning_rate": 9.616827743035817e-06, "loss": 0.9956, "step": 11284 }, { "epoch": 1.2007874015748032, "grad_norm": 10.03125, "learning_rate": 9.616792211483799e-06, "loss": 1.0742, "step": 11285 }, { "epoch": 1.2008938071930197, "grad_norm": 10.546875, "learning_rate": 9.61675667993178e-06, "loss": 1.2129, "step": 11286 }, { "epoch": 1.2010002128112365, "grad_norm": 11.234375, "learning_rate": 9.616721148379762e-06, "loss": 1.2273, "step": 11287 }, { "epoch": 1.201106618429453, "grad_norm": 10.734375, "learning_rate": 9.616685616827744e-06, "loss": 1.1537, "step": 11288 }, { "epoch": 1.2012130240476697, "grad_norm": 10.0, "learning_rate": 9.616650085275727e-06, "loss": 1.1622, "step": 11289 }, { "epoch": 1.2013194296658865, "grad_norm": 13.3984375, "learning_rate": 9.616614553723708e-06, "loss": 1.2914, "step": 11290 }, { "epoch": 1.201425835284103, "grad_norm": 14.6796875, "learning_rate": 9.61657902217169e-06, "loss": 1.2351, "step": 11291 }, { "epoch": 1.2015322409023197, "grad_norm": 11.875, "learning_rate": 9.616543490619672e-06, "loss": 0.989, "step": 11292 }, { "epoch": 1.2016386465205362, "grad_norm": 13.796875, "learning_rate": 9.616507959067652e-06, "loss": 1.0189, "step": 11293 }, { "epoch": 1.201745052138753, "grad_norm": 11.7578125, "learning_rate": 9.616472427515635e-06, "loss": 1.0982, "step": 11294 }, { "epoch": 1.2018514577569697, "grad_norm": 11.359375, "learning_rate": 9.616436895963617e-06, "loss": 1.0279, "step": 11295 }, { "epoch": 1.2019578633751862, "grad_norm": 19.3125, "learning_rate": 9.616401364411598e-06, "loss": 1.1045, "step": 11296 }, { "epoch": 1.202064268993403, "grad_norm": 11.6953125, "learning_rate": 9.61636583285958e-06, "loss": 1.0199, "step": 11297 }, { "epoch": 1.2021706746116194, "grad_norm": 12.484375, "learning_rate": 9.616330301307561e-06, "loss": 1.254, "step": 11298 }, { "epoch": 1.2022770802298361, "grad_norm": 13.6171875, "learning_rate": 9.616294769755545e-06, "loss": 1.5302, "step": 11299 }, { "epoch": 1.2023834858480527, "grad_norm": 10.921875, "learning_rate": 9.616259238203526e-06, "loss": 1.0772, "step": 11300 }, { "epoch": 1.2024898914662694, "grad_norm": 10.3984375, "learning_rate": 9.616223706651508e-06, "loss": 0.8926, "step": 11301 }, { "epoch": 1.2025962970844861, "grad_norm": 8.703125, "learning_rate": 9.61618817509949e-06, "loss": 0.8486, "step": 11302 }, { "epoch": 1.2027027027027026, "grad_norm": 11.0, "learning_rate": 9.616152643547471e-06, "loss": 0.9274, "step": 11303 }, { "epoch": 1.2028091083209194, "grad_norm": 13.53125, "learning_rate": 9.616117111995453e-06, "loss": 0.9435, "step": 11304 }, { "epoch": 1.202915513939136, "grad_norm": 10.5, "learning_rate": 9.616081580443434e-06, "loss": 1.1366, "step": 11305 }, { "epoch": 1.2030219195573526, "grad_norm": 13.9609375, "learning_rate": 9.616046048891416e-06, "loss": 1.3031, "step": 11306 }, { "epoch": 1.2031283251755693, "grad_norm": 11.4375, "learning_rate": 9.616010517339397e-06, "loss": 1.2278, "step": 11307 }, { "epoch": 1.2032347307937858, "grad_norm": 10.4453125, "learning_rate": 9.615974985787379e-06, "loss": 1.0774, "step": 11308 }, { "epoch": 1.2033411364120026, "grad_norm": 8.5390625, "learning_rate": 9.61593945423536e-06, "loss": 0.7211, "step": 11309 }, { "epoch": 1.203447542030219, "grad_norm": 13.4453125, "learning_rate": 9.615903922683344e-06, "loss": 0.9338, "step": 11310 }, { "epoch": 1.2035539476484358, "grad_norm": 9.1953125, "learning_rate": 9.615868391131326e-06, "loss": 0.7754, "step": 11311 }, { "epoch": 1.2036603532666525, "grad_norm": 16.03125, "learning_rate": 9.615832859579307e-06, "loss": 0.9074, "step": 11312 }, { "epoch": 1.203766758884869, "grad_norm": 10.859375, "learning_rate": 9.615797328027289e-06, "loss": 0.8341, "step": 11313 }, { "epoch": 1.2038731645030858, "grad_norm": 13.0078125, "learning_rate": 9.61576179647527e-06, "loss": 1.2926, "step": 11314 }, { "epoch": 1.2039795701213025, "grad_norm": 11.578125, "learning_rate": 9.615726264923254e-06, "loss": 1.2026, "step": 11315 }, { "epoch": 1.204085975739519, "grad_norm": 9.7734375, "learning_rate": 9.615690733371235e-06, "loss": 1.1408, "step": 11316 }, { "epoch": 1.2041923813577358, "grad_norm": 9.390625, "learning_rate": 9.615655201819217e-06, "loss": 0.8597, "step": 11317 }, { "epoch": 1.2042987869759523, "grad_norm": 11.1953125, "learning_rate": 9.615619670267199e-06, "loss": 1.5489, "step": 11318 }, { "epoch": 1.204405192594169, "grad_norm": 13.53125, "learning_rate": 9.61558413871518e-06, "loss": 1.2908, "step": 11319 }, { "epoch": 1.2045115982123855, "grad_norm": 11.484375, "learning_rate": 9.615548607163162e-06, "loss": 1.2633, "step": 11320 }, { "epoch": 1.2046180038306022, "grad_norm": 13.234375, "learning_rate": 9.615513075611143e-06, "loss": 1.124, "step": 11321 }, { "epoch": 1.204724409448819, "grad_norm": 15.765625, "learning_rate": 9.615477544059125e-06, "loss": 0.8121, "step": 11322 }, { "epoch": 1.2048308150670355, "grad_norm": 16.953125, "learning_rate": 9.615442012507107e-06, "loss": 1.4807, "step": 11323 }, { "epoch": 1.2049372206852522, "grad_norm": 12.8125, "learning_rate": 9.615406480955088e-06, "loss": 1.1765, "step": 11324 }, { "epoch": 1.2050436263034687, "grad_norm": 13.3046875, "learning_rate": 9.615370949403071e-06, "loss": 1.2022, "step": 11325 }, { "epoch": 1.2051500319216855, "grad_norm": 10.28125, "learning_rate": 9.615335417851053e-06, "loss": 0.9408, "step": 11326 }, { "epoch": 1.2052564375399022, "grad_norm": 11.75, "learning_rate": 9.615299886299035e-06, "loss": 1.1085, "step": 11327 }, { "epoch": 1.2053628431581187, "grad_norm": 13.09375, "learning_rate": 9.615264354747016e-06, "loss": 1.3057, "step": 11328 }, { "epoch": 1.2054692487763354, "grad_norm": 10.828125, "learning_rate": 9.615228823194998e-06, "loss": 1.2108, "step": 11329 }, { "epoch": 1.205575654394552, "grad_norm": 9.5859375, "learning_rate": 9.61519329164298e-06, "loss": 0.9737, "step": 11330 }, { "epoch": 1.2056820600127687, "grad_norm": 10.125, "learning_rate": 9.615157760090961e-06, "loss": 1.1763, "step": 11331 }, { "epoch": 1.2057884656309854, "grad_norm": 12.5703125, "learning_rate": 9.615122228538943e-06, "loss": 0.9615, "step": 11332 }, { "epoch": 1.205894871249202, "grad_norm": 18.796875, "learning_rate": 9.615086696986924e-06, "loss": 0.8775, "step": 11333 }, { "epoch": 1.2060012768674186, "grad_norm": 10.9140625, "learning_rate": 9.615051165434906e-06, "loss": 1.0785, "step": 11334 }, { "epoch": 1.2061076824856352, "grad_norm": 11.8203125, "learning_rate": 9.61501563388289e-06, "loss": 1.0865, "step": 11335 }, { "epoch": 1.206214088103852, "grad_norm": 10.8828125, "learning_rate": 9.614980102330871e-06, "loss": 1.1633, "step": 11336 }, { "epoch": 1.2063204937220686, "grad_norm": 14.3046875, "learning_rate": 9.614944570778852e-06, "loss": 1.1969, "step": 11337 }, { "epoch": 1.2064268993402851, "grad_norm": 17.6875, "learning_rate": 9.614909039226834e-06, "loss": 1.2611, "step": 11338 }, { "epoch": 1.2065333049585019, "grad_norm": 11.1875, "learning_rate": 9.614873507674816e-06, "loss": 1.0207, "step": 11339 }, { "epoch": 1.2066397105767184, "grad_norm": 10.5625, "learning_rate": 9.614837976122799e-06, "loss": 1.2819, "step": 11340 }, { "epoch": 1.206746116194935, "grad_norm": 11.4140625, "learning_rate": 9.61480244457078e-06, "loss": 1.0663, "step": 11341 }, { "epoch": 1.2068525218131518, "grad_norm": 18.265625, "learning_rate": 9.614766913018762e-06, "loss": 0.931, "step": 11342 }, { "epoch": 1.2069589274313683, "grad_norm": 15.53125, "learning_rate": 9.614731381466744e-06, "loss": 1.2305, "step": 11343 }, { "epoch": 1.207065333049585, "grad_norm": 16.328125, "learning_rate": 9.614695849914725e-06, "loss": 1.0354, "step": 11344 }, { "epoch": 1.2071717386678016, "grad_norm": 14.046875, "learning_rate": 9.614660318362707e-06, "loss": 1.088, "step": 11345 }, { "epoch": 1.2072781442860183, "grad_norm": 12.6171875, "learning_rate": 9.614624786810689e-06, "loss": 1.0299, "step": 11346 }, { "epoch": 1.2073845499042348, "grad_norm": 13.5625, "learning_rate": 9.61458925525867e-06, "loss": 1.1729, "step": 11347 }, { "epoch": 1.2074909555224516, "grad_norm": 13.7578125, "learning_rate": 9.614553723706652e-06, "loss": 1.1754, "step": 11348 }, { "epoch": 1.2075973611406683, "grad_norm": 11.796875, "learning_rate": 9.614518192154633e-06, "loss": 1.1829, "step": 11349 }, { "epoch": 1.2077037667588848, "grad_norm": 10.84375, "learning_rate": 9.614482660602615e-06, "loss": 1.2736, "step": 11350 }, { "epoch": 1.2078101723771015, "grad_norm": 12.21875, "learning_rate": 9.614447129050598e-06, "loss": 0.9458, "step": 11351 }, { "epoch": 1.2079165779953183, "grad_norm": 11.5859375, "learning_rate": 9.61441159749858e-06, "loss": 1.0082, "step": 11352 }, { "epoch": 1.2080229836135348, "grad_norm": 15.34375, "learning_rate": 9.614376065946562e-06, "loss": 1.0399, "step": 11353 }, { "epoch": 1.2081293892317515, "grad_norm": 14.578125, "learning_rate": 9.614340534394543e-06, "loss": 1.0685, "step": 11354 }, { "epoch": 1.208235794849968, "grad_norm": 16.203125, "learning_rate": 9.614305002842525e-06, "loss": 0.9556, "step": 11355 }, { "epoch": 1.2083422004681847, "grad_norm": 12.4765625, "learning_rate": 9.614269471290506e-06, "loss": 0.9729, "step": 11356 }, { "epoch": 1.2084486060864013, "grad_norm": 13.890625, "learning_rate": 9.614233939738488e-06, "loss": 0.9789, "step": 11357 }, { "epoch": 1.208555011704618, "grad_norm": 13.546875, "learning_rate": 9.61419840818647e-06, "loss": 1.3126, "step": 11358 }, { "epoch": 1.2086614173228347, "grad_norm": 11.203125, "learning_rate": 9.614162876634451e-06, "loss": 1.0463, "step": 11359 }, { "epoch": 1.2087678229410512, "grad_norm": 13.65625, "learning_rate": 9.614127345082433e-06, "loss": 1.0536, "step": 11360 }, { "epoch": 1.208874228559268, "grad_norm": 13.265625, "learning_rate": 9.614091813530416e-06, "loss": 0.9271, "step": 11361 }, { "epoch": 1.2089806341774847, "grad_norm": 23.25, "learning_rate": 9.614056281978398e-06, "loss": 0.9892, "step": 11362 }, { "epoch": 1.2090870397957012, "grad_norm": 14.640625, "learning_rate": 9.61402075042638e-06, "loss": 1.1157, "step": 11363 }, { "epoch": 1.209193445413918, "grad_norm": 13.5546875, "learning_rate": 9.613985218874361e-06, "loss": 1.276, "step": 11364 }, { "epoch": 1.2092998510321344, "grad_norm": 12.515625, "learning_rate": 9.613949687322343e-06, "loss": 1.393, "step": 11365 }, { "epoch": 1.2094062566503512, "grad_norm": 12.5078125, "learning_rate": 9.613914155770326e-06, "loss": 1.4333, "step": 11366 }, { "epoch": 1.2095126622685677, "grad_norm": 13.1640625, "learning_rate": 9.613878624218307e-06, "loss": 1.1177, "step": 11367 }, { "epoch": 1.2096190678867844, "grad_norm": 11.5, "learning_rate": 9.613843092666289e-06, "loss": 1.0049, "step": 11368 }, { "epoch": 1.2097254735050011, "grad_norm": 11.65625, "learning_rate": 9.61380756111427e-06, "loss": 0.9402, "step": 11369 }, { "epoch": 1.2098318791232177, "grad_norm": 18.140625, "learning_rate": 9.613772029562252e-06, "loss": 1.0641, "step": 11370 }, { "epoch": 1.2099382847414344, "grad_norm": 16.828125, "learning_rate": 9.613736498010234e-06, "loss": 1.1812, "step": 11371 }, { "epoch": 1.210044690359651, "grad_norm": 14.1015625, "learning_rate": 9.613700966458215e-06, "loss": 1.1426, "step": 11372 }, { "epoch": 1.2101510959778676, "grad_norm": 15.3828125, "learning_rate": 9.613665434906197e-06, "loss": 1.0413, "step": 11373 }, { "epoch": 1.2102575015960844, "grad_norm": 10.8359375, "learning_rate": 9.613629903354179e-06, "loss": 0.9483, "step": 11374 }, { "epoch": 1.2103639072143009, "grad_norm": 22.078125, "learning_rate": 9.61359437180216e-06, "loss": 1.19, "step": 11375 }, { "epoch": 1.2104703128325176, "grad_norm": 13.65625, "learning_rate": 9.613558840250144e-06, "loss": 1.2154, "step": 11376 }, { "epoch": 1.2105767184507341, "grad_norm": 15.1015625, "learning_rate": 9.613523308698125e-06, "loss": 1.005, "step": 11377 }, { "epoch": 1.2106831240689508, "grad_norm": 13.015625, "learning_rate": 9.613487777146107e-06, "loss": 1.1949, "step": 11378 }, { "epoch": 1.2107895296871676, "grad_norm": 10.328125, "learning_rate": 9.613452245594088e-06, "loss": 1.2495, "step": 11379 }, { "epoch": 1.210895935305384, "grad_norm": 13.1875, "learning_rate": 9.61341671404207e-06, "loss": 1.3273, "step": 11380 }, { "epoch": 1.2110023409236008, "grad_norm": 10.3515625, "learning_rate": 9.613381182490053e-06, "loss": 0.9609, "step": 11381 }, { "epoch": 1.2111087465418173, "grad_norm": 11.140625, "learning_rate": 9.613345650938033e-06, "loss": 1.0446, "step": 11382 }, { "epoch": 1.211215152160034, "grad_norm": 12.609375, "learning_rate": 9.613310119386015e-06, "loss": 1.2596, "step": 11383 }, { "epoch": 1.2113215577782506, "grad_norm": 10.625, "learning_rate": 9.613274587833996e-06, "loss": 1.3202, "step": 11384 }, { "epoch": 1.2114279633964673, "grad_norm": 9.7890625, "learning_rate": 9.613239056281978e-06, "loss": 0.9651, "step": 11385 }, { "epoch": 1.211534369014684, "grad_norm": 14.0859375, "learning_rate": 9.613203524729961e-06, "loss": 1.0495, "step": 11386 }, { "epoch": 1.2116407746329005, "grad_norm": 13.140625, "learning_rate": 9.613167993177943e-06, "loss": 0.9353, "step": 11387 }, { "epoch": 1.2117471802511173, "grad_norm": 10.9765625, "learning_rate": 9.613132461625925e-06, "loss": 1.0315, "step": 11388 }, { "epoch": 1.211853585869334, "grad_norm": 15.96875, "learning_rate": 9.613096930073906e-06, "loss": 1.0304, "step": 11389 }, { "epoch": 1.2119599914875505, "grad_norm": 17.15625, "learning_rate": 9.613061398521888e-06, "loss": 1.1505, "step": 11390 }, { "epoch": 1.2120663971057672, "grad_norm": 11.625, "learning_rate": 9.61302586696987e-06, "loss": 0.9904, "step": 11391 }, { "epoch": 1.2121728027239838, "grad_norm": 13.4375, "learning_rate": 9.612990335417853e-06, "loss": 1.1142, "step": 11392 }, { "epoch": 1.2122792083422005, "grad_norm": 10.5546875, "learning_rate": 9.612954803865834e-06, "loss": 0.9919, "step": 11393 }, { "epoch": 1.212385613960417, "grad_norm": 10.890625, "learning_rate": 9.612919272313816e-06, "loss": 1.1755, "step": 11394 }, { "epoch": 1.2124920195786337, "grad_norm": 14.4140625, "learning_rate": 9.612883740761798e-06, "loss": 1.0805, "step": 11395 }, { "epoch": 1.2125984251968505, "grad_norm": 10.1953125, "learning_rate": 9.612848209209779e-06, "loss": 1.0106, "step": 11396 }, { "epoch": 1.212704830815067, "grad_norm": 8.4453125, "learning_rate": 9.61281267765776e-06, "loss": 0.9289, "step": 11397 }, { "epoch": 1.2128112364332837, "grad_norm": 14.109375, "learning_rate": 9.612777146105742e-06, "loss": 1.1566, "step": 11398 }, { "epoch": 1.2129176420515004, "grad_norm": 11.2890625, "learning_rate": 9.612741614553724e-06, "loss": 0.8966, "step": 11399 }, { "epoch": 1.213024047669717, "grad_norm": 9.6875, "learning_rate": 9.612706083001706e-06, "loss": 1.105, "step": 11400 }, { "epoch": 1.2131304532879337, "grad_norm": 11.0234375, "learning_rate": 9.612670551449687e-06, "loss": 0.9566, "step": 11401 }, { "epoch": 1.2132368589061502, "grad_norm": 12.2578125, "learning_rate": 9.61263501989767e-06, "loss": 1.0505, "step": 11402 }, { "epoch": 1.213343264524367, "grad_norm": 13.9921875, "learning_rate": 9.612599488345652e-06, "loss": 1.5418, "step": 11403 }, { "epoch": 1.2134496701425834, "grad_norm": 21.0, "learning_rate": 9.612563956793634e-06, "loss": 1.1844, "step": 11404 }, { "epoch": 1.2135560757608002, "grad_norm": 15.796875, "learning_rate": 9.612528425241615e-06, "loss": 1.0071, "step": 11405 }, { "epoch": 1.2136624813790169, "grad_norm": 12.2734375, "learning_rate": 9.612492893689597e-06, "loss": 0.9383, "step": 11406 }, { "epoch": 1.2137688869972334, "grad_norm": 10.96875, "learning_rate": 9.61245736213758e-06, "loss": 1.3037, "step": 11407 }, { "epoch": 1.2138752926154501, "grad_norm": 12.234375, "learning_rate": 9.612421830585562e-06, "loss": 1.2786, "step": 11408 }, { "epoch": 1.2139816982336669, "grad_norm": 17.4375, "learning_rate": 9.612386299033543e-06, "loss": 1.0632, "step": 11409 }, { "epoch": 1.2140881038518834, "grad_norm": 10.609375, "learning_rate": 9.612350767481525e-06, "loss": 0.9426, "step": 11410 }, { "epoch": 1.2141945094701, "grad_norm": 13.4765625, "learning_rate": 9.612315235929507e-06, "loss": 1.2808, "step": 11411 }, { "epoch": 1.2143009150883166, "grad_norm": 10.6640625, "learning_rate": 9.612279704377488e-06, "loss": 1.1546, "step": 11412 }, { "epoch": 1.2144073207065333, "grad_norm": 12.1328125, "learning_rate": 9.61224417282547e-06, "loss": 0.9936, "step": 11413 }, { "epoch": 1.2145137263247499, "grad_norm": 15.203125, "learning_rate": 9.612208641273451e-06, "loss": 1.2221, "step": 11414 }, { "epoch": 1.2146201319429666, "grad_norm": 13.09375, "learning_rate": 9.612173109721433e-06, "loss": 0.9917, "step": 11415 }, { "epoch": 1.2147265375611833, "grad_norm": 12.046875, "learning_rate": 9.612137578169415e-06, "loss": 1.0553, "step": 11416 }, { "epoch": 1.2148329431793998, "grad_norm": 14.0, "learning_rate": 9.612102046617398e-06, "loss": 1.2787, "step": 11417 }, { "epoch": 1.2149393487976166, "grad_norm": 11.1796875, "learning_rate": 9.61206651506538e-06, "loss": 0.9813, "step": 11418 }, { "epoch": 1.215045754415833, "grad_norm": 14.8046875, "learning_rate": 9.612030983513361e-06, "loss": 1.2786, "step": 11419 }, { "epoch": 1.2151521600340498, "grad_norm": 15.453125, "learning_rate": 9.611995451961343e-06, "loss": 1.0926, "step": 11420 }, { "epoch": 1.2152585656522665, "grad_norm": 11.046875, "learning_rate": 9.611959920409324e-06, "loss": 0.8354, "step": 11421 }, { "epoch": 1.215364971270483, "grad_norm": 15.5390625, "learning_rate": 9.611924388857306e-06, "loss": 1.2671, "step": 11422 }, { "epoch": 1.2154713768886998, "grad_norm": 10.21875, "learning_rate": 9.611888857305288e-06, "loss": 1.0859, "step": 11423 }, { "epoch": 1.2155777825069163, "grad_norm": 8.34375, "learning_rate": 9.61185332575327e-06, "loss": 0.842, "step": 11424 }, { "epoch": 1.215684188125133, "grad_norm": 15.421875, "learning_rate": 9.61181779420125e-06, "loss": 1.255, "step": 11425 }, { "epoch": 1.2157905937433497, "grad_norm": 10.640625, "learning_rate": 9.611782262649232e-06, "loss": 1.0404, "step": 11426 }, { "epoch": 1.2158969993615663, "grad_norm": 11.515625, "learning_rate": 9.611746731097214e-06, "loss": 1.2817, "step": 11427 }, { "epoch": 1.216003404979783, "grad_norm": 11.09375, "learning_rate": 9.611711199545197e-06, "loss": 0.9204, "step": 11428 }, { "epoch": 1.2161098105979995, "grad_norm": 11.0234375, "learning_rate": 9.611675667993179e-06, "loss": 0.8937, "step": 11429 }, { "epoch": 1.2162162162162162, "grad_norm": 9.0078125, "learning_rate": 9.61164013644116e-06, "loss": 0.7689, "step": 11430 }, { "epoch": 1.2163226218344327, "grad_norm": 12.2578125, "learning_rate": 9.611604604889142e-06, "loss": 0.9213, "step": 11431 }, { "epoch": 1.2164290274526495, "grad_norm": 15.4140625, "learning_rate": 9.611569073337124e-06, "loss": 1.2171, "step": 11432 }, { "epoch": 1.2165354330708662, "grad_norm": 12.546875, "learning_rate": 9.611533541785107e-06, "loss": 1.1597, "step": 11433 }, { "epoch": 1.2166418386890827, "grad_norm": 9.7734375, "learning_rate": 9.611498010233089e-06, "loss": 0.9687, "step": 11434 }, { "epoch": 1.2167482443072994, "grad_norm": 8.828125, "learning_rate": 9.61146247868107e-06, "loss": 0.8736, "step": 11435 }, { "epoch": 1.2168546499255162, "grad_norm": 9.6015625, "learning_rate": 9.611426947129052e-06, "loss": 1.1938, "step": 11436 }, { "epoch": 1.2169610555437327, "grad_norm": 9.296875, "learning_rate": 9.611391415577033e-06, "loss": 0.9715, "step": 11437 }, { "epoch": 1.2170674611619494, "grad_norm": 17.703125, "learning_rate": 9.611355884025015e-06, "loss": 1.1135, "step": 11438 }, { "epoch": 1.217173866780166, "grad_norm": 10.1640625, "learning_rate": 9.611320352472997e-06, "loss": 1.1713, "step": 11439 }, { "epoch": 1.2172802723983827, "grad_norm": 11.4765625, "learning_rate": 9.611284820920978e-06, "loss": 0.9971, "step": 11440 }, { "epoch": 1.2173866780165992, "grad_norm": 15.7734375, "learning_rate": 9.61124928936896e-06, "loss": 1.2072, "step": 11441 }, { "epoch": 1.217493083634816, "grad_norm": 14.328125, "learning_rate": 9.611213757816941e-06, "loss": 1.0669, "step": 11442 }, { "epoch": 1.2175994892530326, "grad_norm": 11.7734375, "learning_rate": 9.611178226264925e-06, "loss": 0.9683, "step": 11443 }, { "epoch": 1.2177058948712491, "grad_norm": 8.9765625, "learning_rate": 9.611142694712906e-06, "loss": 1.008, "step": 11444 }, { "epoch": 1.2178123004894659, "grad_norm": 14.296875, "learning_rate": 9.611107163160888e-06, "loss": 1.0213, "step": 11445 }, { "epoch": 1.2179187061076826, "grad_norm": 11.828125, "learning_rate": 9.61107163160887e-06, "loss": 0.6936, "step": 11446 }, { "epoch": 1.2180251117258991, "grad_norm": 9.421875, "learning_rate": 9.611036100056851e-06, "loss": 0.8598, "step": 11447 }, { "epoch": 1.2181315173441158, "grad_norm": 11.65625, "learning_rate": 9.611000568504833e-06, "loss": 1.1483, "step": 11448 }, { "epoch": 1.2182379229623324, "grad_norm": 12.46875, "learning_rate": 9.610965036952814e-06, "loss": 1.144, "step": 11449 }, { "epoch": 1.218344328580549, "grad_norm": 10.8203125, "learning_rate": 9.610929505400796e-06, "loss": 1.0774, "step": 11450 }, { "epoch": 1.2184507341987656, "grad_norm": 11.859375, "learning_rate": 9.610893973848778e-06, "loss": 0.963, "step": 11451 }, { "epoch": 1.2185571398169823, "grad_norm": 10.8515625, "learning_rate": 9.61085844229676e-06, "loss": 1.1605, "step": 11452 }, { "epoch": 1.218663545435199, "grad_norm": 13.7265625, "learning_rate": 9.610822910744743e-06, "loss": 1.113, "step": 11453 }, { "epoch": 1.2187699510534156, "grad_norm": 12.96875, "learning_rate": 9.610787379192724e-06, "loss": 0.9902, "step": 11454 }, { "epoch": 1.2188763566716323, "grad_norm": 13.9609375, "learning_rate": 9.610751847640706e-06, "loss": 1.1253, "step": 11455 }, { "epoch": 1.2189827622898488, "grad_norm": 12.6796875, "learning_rate": 9.610716316088687e-06, "loss": 0.9291, "step": 11456 }, { "epoch": 1.2190891679080655, "grad_norm": 10.1875, "learning_rate": 9.610680784536669e-06, "loss": 1.0081, "step": 11457 }, { "epoch": 1.2191955735262823, "grad_norm": 11.7421875, "learning_rate": 9.610645252984652e-06, "loss": 1.1989, "step": 11458 }, { "epoch": 1.2193019791444988, "grad_norm": 12.171875, "learning_rate": 9.610609721432634e-06, "loss": 1.0178, "step": 11459 }, { "epoch": 1.2194083847627155, "grad_norm": 9.6328125, "learning_rate": 9.610574189880615e-06, "loss": 0.8613, "step": 11460 }, { "epoch": 1.219514790380932, "grad_norm": 12.015625, "learning_rate": 9.610538658328597e-06, "loss": 1.1702, "step": 11461 }, { "epoch": 1.2196211959991488, "grad_norm": 9.6953125, "learning_rate": 9.610503126776579e-06, "loss": 0.9242, "step": 11462 }, { "epoch": 1.2197276016173655, "grad_norm": 10.6640625, "learning_rate": 9.61046759522456e-06, "loss": 0.7613, "step": 11463 }, { "epoch": 1.219834007235582, "grad_norm": 11.0859375, "learning_rate": 9.610432063672542e-06, "loss": 0.8598, "step": 11464 }, { "epoch": 1.2199404128537987, "grad_norm": 12.78125, "learning_rate": 9.610396532120524e-06, "loss": 1.1939, "step": 11465 }, { "epoch": 1.2200468184720152, "grad_norm": 13.25, "learning_rate": 9.610361000568505e-06, "loss": 1.0955, "step": 11466 }, { "epoch": 1.220153224090232, "grad_norm": 12.640625, "learning_rate": 9.610325469016487e-06, "loss": 1.2188, "step": 11467 }, { "epoch": 1.2202596297084487, "grad_norm": 11.734375, "learning_rate": 9.610289937464468e-06, "loss": 1.3045, "step": 11468 }, { "epoch": 1.2203660353266652, "grad_norm": 11.2265625, "learning_rate": 9.610254405912452e-06, "loss": 0.9895, "step": 11469 }, { "epoch": 1.220472440944882, "grad_norm": 9.7421875, "learning_rate": 9.610218874360433e-06, "loss": 1.146, "step": 11470 }, { "epoch": 1.2205788465630985, "grad_norm": 10.9921875, "learning_rate": 9.610183342808415e-06, "loss": 1.1324, "step": 11471 }, { "epoch": 1.2206852521813152, "grad_norm": 14.953125, "learning_rate": 9.610147811256396e-06, "loss": 1.1135, "step": 11472 }, { "epoch": 1.220791657799532, "grad_norm": 11.3046875, "learning_rate": 9.610112279704378e-06, "loss": 1.0605, "step": 11473 }, { "epoch": 1.2208980634177484, "grad_norm": 13.2421875, "learning_rate": 9.61007674815236e-06, "loss": 1.062, "step": 11474 }, { "epoch": 1.2210044690359652, "grad_norm": 10.671875, "learning_rate": 9.610041216600341e-06, "loss": 1.002, "step": 11475 }, { "epoch": 1.2211108746541817, "grad_norm": 13.0390625, "learning_rate": 9.610005685048323e-06, "loss": 1.0839, "step": 11476 }, { "epoch": 1.2212172802723984, "grad_norm": 10.109375, "learning_rate": 9.609970153496304e-06, "loss": 1.0507, "step": 11477 }, { "epoch": 1.221323685890615, "grad_norm": 12.9140625, "learning_rate": 9.609934621944286e-06, "loss": 1.0133, "step": 11478 }, { "epoch": 1.2214300915088316, "grad_norm": 11.4296875, "learning_rate": 9.60989909039227e-06, "loss": 1.1902, "step": 11479 }, { "epoch": 1.2215364971270484, "grad_norm": 12.65625, "learning_rate": 9.609863558840251e-06, "loss": 1.5158, "step": 11480 }, { "epoch": 1.2216429027452649, "grad_norm": 10.0234375, "learning_rate": 9.609828027288233e-06, "loss": 1.0068, "step": 11481 }, { "epoch": 1.2217493083634816, "grad_norm": 10.5, "learning_rate": 9.609792495736214e-06, "loss": 1.1253, "step": 11482 }, { "epoch": 1.2218557139816983, "grad_norm": 11.875, "learning_rate": 9.609756964184196e-06, "loss": 1.0174, "step": 11483 }, { "epoch": 1.2219621195999149, "grad_norm": 13.4921875, "learning_rate": 9.609721432632179e-06, "loss": 1.515, "step": 11484 }, { "epoch": 1.2220685252181316, "grad_norm": 12.90625, "learning_rate": 9.60968590108016e-06, "loss": 1.3099, "step": 11485 }, { "epoch": 1.222174930836348, "grad_norm": 13.03125, "learning_rate": 9.609650369528142e-06, "loss": 1.224, "step": 11486 }, { "epoch": 1.2222813364545648, "grad_norm": 13.21875, "learning_rate": 9.609614837976124e-06, "loss": 1.3239, "step": 11487 }, { "epoch": 1.2223877420727813, "grad_norm": 10.9296875, "learning_rate": 9.609579306424106e-06, "loss": 1.1361, "step": 11488 }, { "epoch": 1.222494147690998, "grad_norm": 12.6484375, "learning_rate": 9.609543774872087e-06, "loss": 0.9876, "step": 11489 }, { "epoch": 1.2226005533092148, "grad_norm": 16.25, "learning_rate": 9.609508243320069e-06, "loss": 0.9901, "step": 11490 }, { "epoch": 1.2227069589274313, "grad_norm": 12.3203125, "learning_rate": 9.60947271176805e-06, "loss": 1.0837, "step": 11491 }, { "epoch": 1.222813364545648, "grad_norm": 10.4375, "learning_rate": 9.609437180216032e-06, "loss": 0.9984, "step": 11492 }, { "epoch": 1.2229197701638648, "grad_norm": 14.1328125, "learning_rate": 9.609401648664014e-06, "loss": 1.0241, "step": 11493 }, { "epoch": 1.2230261757820813, "grad_norm": 11.53125, "learning_rate": 9.609366117111997e-06, "loss": 1.2822, "step": 11494 }, { "epoch": 1.223132581400298, "grad_norm": 11.203125, "learning_rate": 9.609330585559978e-06, "loss": 0.8867, "step": 11495 }, { "epoch": 1.2232389870185145, "grad_norm": 18.546875, "learning_rate": 9.60929505400796e-06, "loss": 1.0409, "step": 11496 }, { "epoch": 1.2233453926367313, "grad_norm": 11.703125, "learning_rate": 9.609259522455942e-06, "loss": 1.0279, "step": 11497 }, { "epoch": 1.2234517982549478, "grad_norm": 12.296875, "learning_rate": 9.609223990903923e-06, "loss": 1.0549, "step": 11498 }, { "epoch": 1.2235582038731645, "grad_norm": 11.296875, "learning_rate": 9.609188459351907e-06, "loss": 1.0541, "step": 11499 }, { "epoch": 1.2236646094913812, "grad_norm": 13.171875, "learning_rate": 9.609152927799888e-06, "loss": 1.2397, "step": 11500 }, { "epoch": 1.2237710151095977, "grad_norm": 11.453125, "learning_rate": 9.60911739624787e-06, "loss": 0.9337, "step": 11501 }, { "epoch": 1.2238774207278145, "grad_norm": 12.5390625, "learning_rate": 9.60908186469585e-06, "loss": 0.9474, "step": 11502 }, { "epoch": 1.223983826346031, "grad_norm": 13.703125, "learning_rate": 9.609046333143831e-06, "loss": 1.0802, "step": 11503 }, { "epoch": 1.2240902319642477, "grad_norm": 13.1796875, "learning_rate": 9.609010801591813e-06, "loss": 1.186, "step": 11504 }, { "epoch": 1.2241966375824644, "grad_norm": 13.3203125, "learning_rate": 9.608975270039796e-06, "loss": 1.3747, "step": 11505 }, { "epoch": 1.224303043200681, "grad_norm": 11.1953125, "learning_rate": 9.608939738487778e-06, "loss": 1.3781, "step": 11506 }, { "epoch": 1.2244094488188977, "grad_norm": 13.0, "learning_rate": 9.60890420693576e-06, "loss": 1.3148, "step": 11507 }, { "epoch": 1.2245158544371142, "grad_norm": 11.828125, "learning_rate": 9.608868675383741e-06, "loss": 1.1696, "step": 11508 }, { "epoch": 1.224622260055331, "grad_norm": 10.0390625, "learning_rate": 9.608833143831723e-06, "loss": 0.8751, "step": 11509 }, { "epoch": 1.2247286656735477, "grad_norm": 11.1796875, "learning_rate": 9.608797612279706e-06, "loss": 1.427, "step": 11510 }, { "epoch": 1.2248350712917642, "grad_norm": 11.1328125, "learning_rate": 9.608762080727688e-06, "loss": 1.0463, "step": 11511 }, { "epoch": 1.224941476909981, "grad_norm": 12.109375, "learning_rate": 9.60872654917567e-06, "loss": 1.1169, "step": 11512 }, { "epoch": 1.2250478825281974, "grad_norm": 15.4765625, "learning_rate": 9.60869101762365e-06, "loss": 1.0844, "step": 11513 }, { "epoch": 1.2251542881464141, "grad_norm": 10.390625, "learning_rate": 9.608655486071632e-06, "loss": 1.0486, "step": 11514 }, { "epoch": 1.2252606937646306, "grad_norm": 18.484375, "learning_rate": 9.608619954519614e-06, "loss": 1.4679, "step": 11515 }, { "epoch": 1.2253670993828474, "grad_norm": 14.1796875, "learning_rate": 9.608584422967596e-06, "loss": 0.8971, "step": 11516 }, { "epoch": 1.2254735050010641, "grad_norm": 11.5234375, "learning_rate": 9.608548891415577e-06, "loss": 1.1773, "step": 11517 }, { "epoch": 1.2255799106192806, "grad_norm": 13.3203125, "learning_rate": 9.608513359863559e-06, "loss": 1.2992, "step": 11518 }, { "epoch": 1.2256863162374974, "grad_norm": 12.359375, "learning_rate": 9.60847782831154e-06, "loss": 1.4212, "step": 11519 }, { "epoch": 1.225792721855714, "grad_norm": 10.8671875, "learning_rate": 9.608442296759524e-06, "loss": 1.2225, "step": 11520 }, { "epoch": 1.2258991274739306, "grad_norm": 12.1015625, "learning_rate": 9.608406765207505e-06, "loss": 1.0295, "step": 11521 }, { "epoch": 1.2260055330921473, "grad_norm": 14.2578125, "learning_rate": 9.608371233655487e-06, "loss": 0.884, "step": 11522 }, { "epoch": 1.2261119387103638, "grad_norm": 15.3515625, "learning_rate": 9.608335702103469e-06, "loss": 1.1628, "step": 11523 }, { "epoch": 1.2262183443285806, "grad_norm": 11.390625, "learning_rate": 9.60830017055145e-06, "loss": 1.0392, "step": 11524 }, { "epoch": 1.226324749946797, "grad_norm": 10.203125, "learning_rate": 9.608264638999433e-06, "loss": 0.9266, "step": 11525 }, { "epoch": 1.2264311555650138, "grad_norm": 14.765625, "learning_rate": 9.608229107447415e-06, "loss": 1.024, "step": 11526 }, { "epoch": 1.2265375611832305, "grad_norm": 12.5, "learning_rate": 9.608193575895397e-06, "loss": 1.2558, "step": 11527 }, { "epoch": 1.226643966801447, "grad_norm": 15.7890625, "learning_rate": 9.608158044343378e-06, "loss": 1.0461, "step": 11528 }, { "epoch": 1.2267503724196638, "grad_norm": 9.5, "learning_rate": 9.60812251279136e-06, "loss": 0.8912, "step": 11529 }, { "epoch": 1.2268567780378805, "grad_norm": 9.4375, "learning_rate": 9.608086981239342e-06, "loss": 1.0373, "step": 11530 }, { "epoch": 1.226963183656097, "grad_norm": 10.6328125, "learning_rate": 9.608051449687323e-06, "loss": 1.0829, "step": 11531 }, { "epoch": 1.2270695892743138, "grad_norm": 10.0625, "learning_rate": 9.608015918135305e-06, "loss": 1.0958, "step": 11532 }, { "epoch": 1.2271759948925303, "grad_norm": 18.09375, "learning_rate": 9.607980386583286e-06, "loss": 1.1703, "step": 11533 }, { "epoch": 1.227282400510747, "grad_norm": 12.6953125, "learning_rate": 9.607944855031268e-06, "loss": 1.2316, "step": 11534 }, { "epoch": 1.2273888061289635, "grad_norm": 9.84375, "learning_rate": 9.607909323479251e-06, "loss": 1.0162, "step": 11535 }, { "epoch": 1.2274952117471802, "grad_norm": 18.953125, "learning_rate": 9.607873791927233e-06, "loss": 1.2443, "step": 11536 }, { "epoch": 1.227601617365397, "grad_norm": 14.6328125, "learning_rate": 9.607838260375214e-06, "loss": 1.3119, "step": 11537 }, { "epoch": 1.2277080229836135, "grad_norm": 14.7890625, "learning_rate": 9.607802728823196e-06, "loss": 0.9214, "step": 11538 }, { "epoch": 1.2278144286018302, "grad_norm": 13.4609375, "learning_rate": 9.607767197271178e-06, "loss": 1.0591, "step": 11539 }, { "epoch": 1.227920834220047, "grad_norm": 9.046875, "learning_rate": 9.60773166571916e-06, "loss": 1.2192, "step": 11540 }, { "epoch": 1.2280272398382635, "grad_norm": 14.1796875, "learning_rate": 9.607696134167141e-06, "loss": 1.4425, "step": 11541 }, { "epoch": 1.2281336454564802, "grad_norm": 13.453125, "learning_rate": 9.607660602615122e-06, "loss": 0.9031, "step": 11542 }, { "epoch": 1.2282400510746967, "grad_norm": 14.359375, "learning_rate": 9.607625071063104e-06, "loss": 1.1507, "step": 11543 }, { "epoch": 1.2283464566929134, "grad_norm": 15.7265625, "learning_rate": 9.607589539511086e-06, "loss": 1.0121, "step": 11544 }, { "epoch": 1.22845286231113, "grad_norm": 12.5078125, "learning_rate": 9.607554007959067e-06, "loss": 0.9513, "step": 11545 }, { "epoch": 1.2285592679293467, "grad_norm": 8.6640625, "learning_rate": 9.60751847640705e-06, "loss": 1.1298, "step": 11546 }, { "epoch": 1.2286656735475634, "grad_norm": 14.921875, "learning_rate": 9.607482944855032e-06, "loss": 0.9179, "step": 11547 }, { "epoch": 1.22877207916578, "grad_norm": 9.015625, "learning_rate": 9.607447413303014e-06, "loss": 0.8787, "step": 11548 }, { "epoch": 1.2288784847839966, "grad_norm": 19.6875, "learning_rate": 9.607411881750995e-06, "loss": 0.532, "step": 11549 }, { "epoch": 1.2289848904022131, "grad_norm": 13.015625, "learning_rate": 9.607376350198977e-06, "loss": 1.1254, "step": 11550 }, { "epoch": 1.2290912960204299, "grad_norm": 16.09375, "learning_rate": 9.60734081864696e-06, "loss": 1.1705, "step": 11551 }, { "epoch": 1.2291977016386466, "grad_norm": 10.0625, "learning_rate": 9.607305287094942e-06, "loss": 1.0132, "step": 11552 }, { "epoch": 1.2293041072568631, "grad_norm": 13.03125, "learning_rate": 9.607269755542924e-06, "loss": 1.1264, "step": 11553 }, { "epoch": 1.2294105128750799, "grad_norm": 11.6171875, "learning_rate": 9.607234223990905e-06, "loss": 0.9685, "step": 11554 }, { "epoch": 1.2295169184932964, "grad_norm": 14.40625, "learning_rate": 9.607198692438887e-06, "loss": 1.0402, "step": 11555 }, { "epoch": 1.229623324111513, "grad_norm": 24.765625, "learning_rate": 9.607163160886868e-06, "loss": 0.7953, "step": 11556 }, { "epoch": 1.2297297297297298, "grad_norm": 10.5625, "learning_rate": 9.60712762933485e-06, "loss": 1.1567, "step": 11557 }, { "epoch": 1.2298361353479463, "grad_norm": 21.0625, "learning_rate": 9.607092097782832e-06, "loss": 1.4851, "step": 11558 }, { "epoch": 1.229942540966163, "grad_norm": 10.0390625, "learning_rate": 9.607056566230813e-06, "loss": 0.9084, "step": 11559 }, { "epoch": 1.2300489465843796, "grad_norm": 11.21875, "learning_rate": 9.607021034678795e-06, "loss": 1.3968, "step": 11560 }, { "epoch": 1.2301553522025963, "grad_norm": 16.71875, "learning_rate": 9.606985503126778e-06, "loss": 1.1125, "step": 11561 }, { "epoch": 1.2302617578208128, "grad_norm": 11.0703125, "learning_rate": 9.60694997157476e-06, "loss": 1.1452, "step": 11562 }, { "epoch": 1.2303681634390295, "grad_norm": 8.40625, "learning_rate": 9.606914440022741e-06, "loss": 0.9029, "step": 11563 }, { "epoch": 1.2304745690572463, "grad_norm": 11.1875, "learning_rate": 9.606878908470723e-06, "loss": 0.9816, "step": 11564 }, { "epoch": 1.2305809746754628, "grad_norm": 15.6640625, "learning_rate": 9.606843376918705e-06, "loss": 1.1978, "step": 11565 }, { "epoch": 1.2306873802936795, "grad_norm": 10.8828125, "learning_rate": 9.606807845366686e-06, "loss": 1.1242, "step": 11566 }, { "epoch": 1.2307937859118963, "grad_norm": 15.2109375, "learning_rate": 9.606772313814668e-06, "loss": 1.3763, "step": 11567 }, { "epoch": 1.2309001915301128, "grad_norm": 12.875, "learning_rate": 9.60673678226265e-06, "loss": 1.309, "step": 11568 }, { "epoch": 1.2310065971483295, "grad_norm": 11.2109375, "learning_rate": 9.606701250710631e-06, "loss": 1.4363, "step": 11569 }, { "epoch": 1.231113002766546, "grad_norm": 11.796875, "learning_rate": 9.606665719158613e-06, "loss": 1.1477, "step": 11570 }, { "epoch": 1.2312194083847627, "grad_norm": 15.1953125, "learning_rate": 9.606630187606596e-06, "loss": 1.1275, "step": 11571 }, { "epoch": 1.2313258140029792, "grad_norm": 11.6953125, "learning_rate": 9.606594656054577e-06, "loss": 1.1693, "step": 11572 }, { "epoch": 1.231432219621196, "grad_norm": 17.390625, "learning_rate": 9.606559124502559e-06, "loss": 0.7755, "step": 11573 }, { "epoch": 1.2315386252394127, "grad_norm": 11.90625, "learning_rate": 9.60652359295054e-06, "loss": 1.1312, "step": 11574 }, { "epoch": 1.2316450308576292, "grad_norm": 12.265625, "learning_rate": 9.606488061398522e-06, "loss": 0.8611, "step": 11575 }, { "epoch": 1.231751436475846, "grad_norm": 15.640625, "learning_rate": 9.606452529846506e-06, "loss": 1.1882, "step": 11576 }, { "epoch": 1.2318578420940627, "grad_norm": 12.765625, "learning_rate": 9.606416998294487e-06, "loss": 1.4159, "step": 11577 }, { "epoch": 1.2319642477122792, "grad_norm": 10.59375, "learning_rate": 9.606381466742469e-06, "loss": 0.8116, "step": 11578 }, { "epoch": 1.232070653330496, "grad_norm": 13.171875, "learning_rate": 9.60634593519045e-06, "loss": 1.1203, "step": 11579 }, { "epoch": 1.2321770589487124, "grad_norm": 13.625, "learning_rate": 9.606310403638432e-06, "loss": 0.9498, "step": 11580 }, { "epoch": 1.2322834645669292, "grad_norm": 10.4453125, "learning_rate": 9.606274872086414e-06, "loss": 1.2165, "step": 11581 }, { "epoch": 1.2323898701851457, "grad_norm": 11.4375, "learning_rate": 9.606239340534395e-06, "loss": 1.0195, "step": 11582 }, { "epoch": 1.2324962758033624, "grad_norm": 14.46875, "learning_rate": 9.606203808982377e-06, "loss": 1.2348, "step": 11583 }, { "epoch": 1.2326026814215791, "grad_norm": 13.203125, "learning_rate": 9.606168277430358e-06, "loss": 1.2827, "step": 11584 }, { "epoch": 1.2327090870397956, "grad_norm": 12.125, "learning_rate": 9.60613274587834e-06, "loss": 1.2682, "step": 11585 }, { "epoch": 1.2328154926580124, "grad_norm": 9.8359375, "learning_rate": 9.606097214326322e-06, "loss": 1.0412, "step": 11586 }, { "epoch": 1.232921898276229, "grad_norm": 10.3203125, "learning_rate": 9.606061682774305e-06, "loss": 0.8859, "step": 11587 }, { "epoch": 1.2330283038944456, "grad_norm": 17.359375, "learning_rate": 9.606026151222287e-06, "loss": 1.4068, "step": 11588 }, { "epoch": 1.2331347095126624, "grad_norm": 11.0078125, "learning_rate": 9.605990619670268e-06, "loss": 1.075, "step": 11589 }, { "epoch": 1.2332411151308789, "grad_norm": 13.0859375, "learning_rate": 9.60595508811825e-06, "loss": 1.0341, "step": 11590 }, { "epoch": 1.2333475207490956, "grad_norm": 9.9453125, "learning_rate": 9.605919556566231e-06, "loss": 1.0571, "step": 11591 }, { "epoch": 1.233453926367312, "grad_norm": 13.578125, "learning_rate": 9.605884025014213e-06, "loss": 1.3816, "step": 11592 }, { "epoch": 1.2335603319855288, "grad_norm": 9.9140625, "learning_rate": 9.605848493462195e-06, "loss": 1.0014, "step": 11593 }, { "epoch": 1.2336667376037456, "grad_norm": 17.765625, "learning_rate": 9.605812961910176e-06, "loss": 1.1273, "step": 11594 }, { "epoch": 1.233773143221962, "grad_norm": 13.2421875, "learning_rate": 9.605777430358158e-06, "loss": 1.0828, "step": 11595 }, { "epoch": 1.2338795488401788, "grad_norm": 12.28125, "learning_rate": 9.60574189880614e-06, "loss": 0.973, "step": 11596 }, { "epoch": 1.2339859544583953, "grad_norm": 16.21875, "learning_rate": 9.605706367254123e-06, "loss": 0.9543, "step": 11597 }, { "epoch": 1.234092360076612, "grad_norm": 14.125, "learning_rate": 9.605670835702104e-06, "loss": 1.2627, "step": 11598 }, { "epoch": 1.2341987656948288, "grad_norm": 10.703125, "learning_rate": 9.605635304150086e-06, "loss": 1.0239, "step": 11599 }, { "epoch": 1.2343051713130453, "grad_norm": 16.15625, "learning_rate": 9.605599772598068e-06, "loss": 1.1244, "step": 11600 }, { "epoch": 1.234411576931262, "grad_norm": 12.90625, "learning_rate": 9.605564241046049e-06, "loss": 1.0768, "step": 11601 }, { "epoch": 1.2345179825494785, "grad_norm": 11.0, "learning_rate": 9.605528709494032e-06, "loss": 1.1933, "step": 11602 }, { "epoch": 1.2346243881676953, "grad_norm": 11.9375, "learning_rate": 9.605493177942014e-06, "loss": 1.0069, "step": 11603 }, { "epoch": 1.234730793785912, "grad_norm": 12.3671875, "learning_rate": 9.605457646389996e-06, "loss": 1.1784, "step": 11604 }, { "epoch": 1.2348371994041285, "grad_norm": 13.9140625, "learning_rate": 9.605422114837977e-06, "loss": 1.0196, "step": 11605 }, { "epoch": 1.2349436050223452, "grad_norm": 13.5390625, "learning_rate": 9.605386583285959e-06, "loss": 1.2376, "step": 11606 }, { "epoch": 1.2350500106405617, "grad_norm": 15.2109375, "learning_rate": 9.60535105173394e-06, "loss": 1.0782, "step": 11607 }, { "epoch": 1.2351564162587785, "grad_norm": 21.625, "learning_rate": 9.605315520181922e-06, "loss": 1.2893, "step": 11608 }, { "epoch": 1.235262821876995, "grad_norm": 13.671875, "learning_rate": 9.605279988629904e-06, "loss": 0.9961, "step": 11609 }, { "epoch": 1.2353692274952117, "grad_norm": 17.375, "learning_rate": 9.605244457077885e-06, "loss": 0.9929, "step": 11610 }, { "epoch": 1.2354756331134285, "grad_norm": 15.1171875, "learning_rate": 9.605208925525867e-06, "loss": 1.0753, "step": 11611 }, { "epoch": 1.235582038731645, "grad_norm": 14.484375, "learning_rate": 9.60517339397385e-06, "loss": 1.1613, "step": 11612 }, { "epoch": 1.2356884443498617, "grad_norm": 12.8046875, "learning_rate": 9.605137862421832e-06, "loss": 1.062, "step": 11613 }, { "epoch": 1.2357948499680784, "grad_norm": 12.1484375, "learning_rate": 9.605102330869813e-06, "loss": 1.5575, "step": 11614 }, { "epoch": 1.235901255586295, "grad_norm": 14.1953125, "learning_rate": 9.605066799317795e-06, "loss": 1.1726, "step": 11615 }, { "epoch": 1.2360076612045117, "grad_norm": 16.40625, "learning_rate": 9.605031267765777e-06, "loss": 0.8786, "step": 11616 }, { "epoch": 1.2361140668227282, "grad_norm": 12.7421875, "learning_rate": 9.60499573621376e-06, "loss": 1.2357, "step": 11617 }, { "epoch": 1.236220472440945, "grad_norm": 11.015625, "learning_rate": 9.604960204661742e-06, "loss": 1.0304, "step": 11618 }, { "epoch": 1.2363268780591614, "grad_norm": 12.1171875, "learning_rate": 9.604924673109723e-06, "loss": 1.1484, "step": 11619 }, { "epoch": 1.2364332836773781, "grad_norm": 10.921875, "learning_rate": 9.604889141557705e-06, "loss": 1.183, "step": 11620 }, { "epoch": 1.2365396892955949, "grad_norm": 14.71875, "learning_rate": 9.604853610005686e-06, "loss": 1.1625, "step": 11621 }, { "epoch": 1.2366460949138114, "grad_norm": 14.03125, "learning_rate": 9.604818078453666e-06, "loss": 1.2206, "step": 11622 }, { "epoch": 1.2367525005320281, "grad_norm": 17.328125, "learning_rate": 9.60478254690165e-06, "loss": 1.0423, "step": 11623 }, { "epoch": 1.2368589061502449, "grad_norm": 14.9375, "learning_rate": 9.604747015349631e-06, "loss": 1.4491, "step": 11624 }, { "epoch": 1.2369653117684614, "grad_norm": 12.7890625, "learning_rate": 9.604711483797613e-06, "loss": 1.1957, "step": 11625 }, { "epoch": 1.237071717386678, "grad_norm": 12.21875, "learning_rate": 9.604675952245594e-06, "loss": 1.0633, "step": 11626 }, { "epoch": 1.2371781230048946, "grad_norm": 18.21875, "learning_rate": 9.604640420693576e-06, "loss": 1.1753, "step": 11627 }, { "epoch": 1.2372845286231113, "grad_norm": 9.8203125, "learning_rate": 9.60460488914156e-06, "loss": 1.1552, "step": 11628 }, { "epoch": 1.2373909342413278, "grad_norm": 10.375, "learning_rate": 9.604569357589541e-06, "loss": 0.987, "step": 11629 }, { "epoch": 1.2374973398595446, "grad_norm": 10.828125, "learning_rate": 9.604533826037523e-06, "loss": 0.8224, "step": 11630 }, { "epoch": 1.2376037454777613, "grad_norm": 15.765625, "learning_rate": 9.604498294485504e-06, "loss": 1.1155, "step": 11631 }, { "epoch": 1.2377101510959778, "grad_norm": 13.8046875, "learning_rate": 9.604462762933486e-06, "loss": 1.1401, "step": 11632 }, { "epoch": 1.2378165567141945, "grad_norm": 9.4375, "learning_rate": 9.604427231381467e-06, "loss": 1.1189, "step": 11633 }, { "epoch": 1.237922962332411, "grad_norm": 15.265625, "learning_rate": 9.604391699829449e-06, "loss": 1.25, "step": 11634 }, { "epoch": 1.2380293679506278, "grad_norm": 14.953125, "learning_rate": 9.60435616827743e-06, "loss": 0.9721, "step": 11635 }, { "epoch": 1.2381357735688445, "grad_norm": 13.953125, "learning_rate": 9.604320636725412e-06, "loss": 0.9459, "step": 11636 }, { "epoch": 1.238242179187061, "grad_norm": 12.3671875, "learning_rate": 9.604285105173394e-06, "loss": 1.0827, "step": 11637 }, { "epoch": 1.2383485848052778, "grad_norm": 11.71875, "learning_rate": 9.604249573621377e-06, "loss": 1.1654, "step": 11638 }, { "epoch": 1.2384549904234943, "grad_norm": 15.8828125, "learning_rate": 9.604214042069359e-06, "loss": 1.1158, "step": 11639 }, { "epoch": 1.238561396041711, "grad_norm": 14.3203125, "learning_rate": 9.60417851051734e-06, "loss": 1.0554, "step": 11640 }, { "epoch": 1.2386678016599277, "grad_norm": 11.703125, "learning_rate": 9.604142978965322e-06, "loss": 1.2607, "step": 11641 }, { "epoch": 1.2387742072781442, "grad_norm": 11.328125, "learning_rate": 9.604107447413303e-06, "loss": 1.2644, "step": 11642 }, { "epoch": 1.238880612896361, "grad_norm": 9.1796875, "learning_rate": 9.604071915861287e-06, "loss": 1.065, "step": 11643 }, { "epoch": 1.2389870185145775, "grad_norm": 10.703125, "learning_rate": 9.604036384309268e-06, "loss": 1.0646, "step": 11644 }, { "epoch": 1.2390934241327942, "grad_norm": 34.3125, "learning_rate": 9.60400085275725e-06, "loss": 1.6028, "step": 11645 }, { "epoch": 1.2391998297510107, "grad_norm": 9.484375, "learning_rate": 9.603965321205232e-06, "loss": 1.1393, "step": 11646 }, { "epoch": 1.2393062353692275, "grad_norm": 12.9375, "learning_rate": 9.603929789653213e-06, "loss": 1.0541, "step": 11647 }, { "epoch": 1.2394126409874442, "grad_norm": 14.5625, "learning_rate": 9.603894258101195e-06, "loss": 0.9694, "step": 11648 }, { "epoch": 1.2395190466056607, "grad_norm": 11.3359375, "learning_rate": 9.603858726549176e-06, "loss": 1.2292, "step": 11649 }, { "epoch": 1.2396254522238774, "grad_norm": 11.46875, "learning_rate": 9.603823194997158e-06, "loss": 1.07, "step": 11650 }, { "epoch": 1.2397318578420942, "grad_norm": 14.5, "learning_rate": 9.60378766344514e-06, "loss": 1.0961, "step": 11651 }, { "epoch": 1.2398382634603107, "grad_norm": 12.8671875, "learning_rate": 9.603752131893121e-06, "loss": 0.8978, "step": 11652 }, { "epoch": 1.2399446690785274, "grad_norm": 12.1953125, "learning_rate": 9.603716600341105e-06, "loss": 1.11, "step": 11653 }, { "epoch": 1.240051074696744, "grad_norm": 9.359375, "learning_rate": 9.603681068789086e-06, "loss": 0.9258, "step": 11654 }, { "epoch": 1.2401574803149606, "grad_norm": 10.1171875, "learning_rate": 9.603645537237068e-06, "loss": 0.9063, "step": 11655 }, { "epoch": 1.2402638859331772, "grad_norm": 9.5625, "learning_rate": 9.60361000568505e-06, "loss": 1.0073, "step": 11656 }, { "epoch": 1.2403702915513939, "grad_norm": 10.4765625, "learning_rate": 9.603574474133031e-06, "loss": 1.1631, "step": 11657 }, { "epoch": 1.2404766971696106, "grad_norm": 12.84375, "learning_rate": 9.603538942581013e-06, "loss": 1.0254, "step": 11658 }, { "epoch": 1.2405831027878271, "grad_norm": 10.9453125, "learning_rate": 9.603503411028994e-06, "loss": 0.9951, "step": 11659 }, { "epoch": 1.2406895084060439, "grad_norm": 13.171875, "learning_rate": 9.603467879476976e-06, "loss": 0.9954, "step": 11660 }, { "epoch": 1.2407959140242606, "grad_norm": 9.8203125, "learning_rate": 9.603432347924957e-06, "loss": 0.9045, "step": 11661 }, { "epoch": 1.240902319642477, "grad_norm": 9.9765625, "learning_rate": 9.603396816372939e-06, "loss": 0.8371, "step": 11662 }, { "epoch": 1.2410087252606938, "grad_norm": 16.421875, "learning_rate": 9.60336128482092e-06, "loss": 1.0534, "step": 11663 }, { "epoch": 1.2411151308789103, "grad_norm": 14.484375, "learning_rate": 9.603325753268904e-06, "loss": 1.0372, "step": 11664 }, { "epoch": 1.241221536497127, "grad_norm": 14.015625, "learning_rate": 9.603290221716886e-06, "loss": 1.2717, "step": 11665 }, { "epoch": 1.2413279421153436, "grad_norm": 16.234375, "learning_rate": 9.603254690164867e-06, "loss": 1.2431, "step": 11666 }, { "epoch": 1.2414343477335603, "grad_norm": 10.9921875, "learning_rate": 9.603219158612849e-06, "loss": 1.1383, "step": 11667 }, { "epoch": 1.241540753351777, "grad_norm": 14.984375, "learning_rate": 9.60318362706083e-06, "loss": 1.1984, "step": 11668 }, { "epoch": 1.2416471589699936, "grad_norm": 13.359375, "learning_rate": 9.603148095508814e-06, "loss": 1.4475, "step": 11669 }, { "epoch": 1.2417535645882103, "grad_norm": 12.5234375, "learning_rate": 9.603112563956795e-06, "loss": 1.1161, "step": 11670 }, { "epoch": 1.241859970206427, "grad_norm": 18.3125, "learning_rate": 9.603077032404777e-06, "loss": 1.366, "step": 11671 }, { "epoch": 1.2419663758246435, "grad_norm": 16.640625, "learning_rate": 9.603041500852758e-06, "loss": 1.0125, "step": 11672 }, { "epoch": 1.2420727814428603, "grad_norm": 11.125, "learning_rate": 9.60300596930074e-06, "loss": 0.9944, "step": 11673 }, { "epoch": 1.2421791870610768, "grad_norm": 10.1328125, "learning_rate": 9.602970437748722e-06, "loss": 1.0048, "step": 11674 }, { "epoch": 1.2422855926792935, "grad_norm": 10.984375, "learning_rate": 9.602934906196703e-06, "loss": 1.0816, "step": 11675 }, { "epoch": 1.24239199829751, "grad_norm": 14.71875, "learning_rate": 9.602899374644685e-06, "loss": 1.1113, "step": 11676 }, { "epoch": 1.2424984039157267, "grad_norm": 10.640625, "learning_rate": 9.602863843092666e-06, "loss": 1.0281, "step": 11677 }, { "epoch": 1.2426048095339435, "grad_norm": 13.7890625, "learning_rate": 9.602828311540648e-06, "loss": 1.1693, "step": 11678 }, { "epoch": 1.24271121515216, "grad_norm": 15.25, "learning_rate": 9.602792779988631e-06, "loss": 1.2476, "step": 11679 }, { "epoch": 1.2428176207703767, "grad_norm": 12.0703125, "learning_rate": 9.602757248436613e-06, "loss": 1.2595, "step": 11680 }, { "epoch": 1.2429240263885932, "grad_norm": 11.421875, "learning_rate": 9.602721716884595e-06, "loss": 1.0833, "step": 11681 }, { "epoch": 1.24303043200681, "grad_norm": 14.9296875, "learning_rate": 9.602686185332576e-06, "loss": 1.1862, "step": 11682 }, { "epoch": 1.2431368376250267, "grad_norm": 20.828125, "learning_rate": 9.602650653780558e-06, "loss": 0.8062, "step": 11683 }, { "epoch": 1.2432432432432432, "grad_norm": 18.390625, "learning_rate": 9.60261512222854e-06, "loss": 1.3355, "step": 11684 }, { "epoch": 1.24334964886146, "grad_norm": 12.53125, "learning_rate": 9.602579590676521e-06, "loss": 1.3515, "step": 11685 }, { "epoch": 1.2434560544796764, "grad_norm": 14.171875, "learning_rate": 9.602544059124503e-06, "loss": 1.1141, "step": 11686 }, { "epoch": 1.2435624600978932, "grad_norm": 14.9921875, "learning_rate": 9.602508527572484e-06, "loss": 1.2648, "step": 11687 }, { "epoch": 1.24366886571611, "grad_norm": 13.234375, "learning_rate": 9.602472996020466e-06, "loss": 0.9586, "step": 11688 }, { "epoch": 1.2437752713343264, "grad_norm": 14.1640625, "learning_rate": 9.602437464468449e-06, "loss": 0.9724, "step": 11689 }, { "epoch": 1.2438816769525431, "grad_norm": 11.5390625, "learning_rate": 9.60240193291643e-06, "loss": 1.0084, "step": 11690 }, { "epoch": 1.2439880825707597, "grad_norm": 13.140625, "learning_rate": 9.602366401364412e-06, "loss": 1.2799, "step": 11691 }, { "epoch": 1.2440944881889764, "grad_norm": 11.34375, "learning_rate": 9.602330869812394e-06, "loss": 1.1522, "step": 11692 }, { "epoch": 1.244200893807193, "grad_norm": 12.0703125, "learning_rate": 9.602295338260376e-06, "loss": 1.0914, "step": 11693 }, { "epoch": 1.2443072994254096, "grad_norm": 9.0234375, "learning_rate": 9.602259806708359e-06, "loss": 1.0688, "step": 11694 }, { "epoch": 1.2444137050436264, "grad_norm": 12.6015625, "learning_rate": 9.60222427515634e-06, "loss": 1.1215, "step": 11695 }, { "epoch": 1.2445201106618429, "grad_norm": 11.4453125, "learning_rate": 9.602188743604322e-06, "loss": 1.239, "step": 11696 }, { "epoch": 1.2446265162800596, "grad_norm": 11.8203125, "learning_rate": 9.602153212052304e-06, "loss": 1.1881, "step": 11697 }, { "epoch": 1.2447329218982763, "grad_norm": 12.9140625, "learning_rate": 9.602117680500285e-06, "loss": 1.0995, "step": 11698 }, { "epoch": 1.2448393275164928, "grad_norm": 16.046875, "learning_rate": 9.602082148948267e-06, "loss": 1.1319, "step": 11699 }, { "epoch": 1.2449457331347096, "grad_norm": 9.6328125, "learning_rate": 9.602046617396249e-06, "loss": 1.0429, "step": 11700 }, { "epoch": 1.245052138752926, "grad_norm": 11.953125, "learning_rate": 9.60201108584423e-06, "loss": 1.2938, "step": 11701 }, { "epoch": 1.2451585443711428, "grad_norm": 23.328125, "learning_rate": 9.601975554292212e-06, "loss": 1.6482, "step": 11702 }, { "epoch": 1.2452649499893593, "grad_norm": 11.1953125, "learning_rate": 9.601940022740193e-06, "loss": 0.9232, "step": 11703 }, { "epoch": 1.245371355607576, "grad_norm": 10.78125, "learning_rate": 9.601904491188175e-06, "loss": 1.0777, "step": 11704 }, { "epoch": 1.2454777612257928, "grad_norm": 12.875, "learning_rate": 9.601868959636158e-06, "loss": 1.025, "step": 11705 }, { "epoch": 1.2455841668440093, "grad_norm": 13.3046875, "learning_rate": 9.60183342808414e-06, "loss": 0.9521, "step": 11706 }, { "epoch": 1.245690572462226, "grad_norm": 12.2734375, "learning_rate": 9.601797896532121e-06, "loss": 1.1687, "step": 11707 }, { "epoch": 1.2457969780804428, "grad_norm": 12.4375, "learning_rate": 9.601762364980103e-06, "loss": 1.1779, "step": 11708 }, { "epoch": 1.2459033836986593, "grad_norm": 15.3828125, "learning_rate": 9.601726833428085e-06, "loss": 1.0427, "step": 11709 }, { "epoch": 1.246009789316876, "grad_norm": 22.375, "learning_rate": 9.601691301876068e-06, "loss": 1.3908, "step": 11710 }, { "epoch": 1.2461161949350925, "grad_norm": 11.0703125, "learning_rate": 9.601655770324048e-06, "loss": 1.0731, "step": 11711 }, { "epoch": 1.2462226005533092, "grad_norm": 13.9453125, "learning_rate": 9.60162023877203e-06, "loss": 1.3146, "step": 11712 }, { "epoch": 1.2463290061715258, "grad_norm": 10.328125, "learning_rate": 9.601584707220011e-06, "loss": 1.1234, "step": 11713 }, { "epoch": 1.2464354117897425, "grad_norm": 11.265625, "learning_rate": 9.601549175667993e-06, "loss": 1.0162, "step": 11714 }, { "epoch": 1.2465418174079592, "grad_norm": 13.1015625, "learning_rate": 9.601513644115976e-06, "loss": 1.1993, "step": 11715 }, { "epoch": 1.2466482230261757, "grad_norm": 12.3515625, "learning_rate": 9.601478112563958e-06, "loss": 1.0607, "step": 11716 }, { "epoch": 1.2467546286443925, "grad_norm": 15.7578125, "learning_rate": 9.60144258101194e-06, "loss": 1.1266, "step": 11717 }, { "epoch": 1.246861034262609, "grad_norm": 16.265625, "learning_rate": 9.60140704945992e-06, "loss": 1.5908, "step": 11718 }, { "epoch": 1.2469674398808257, "grad_norm": 10.96875, "learning_rate": 9.601371517907902e-06, "loss": 0.8696, "step": 11719 }, { "epoch": 1.2470738454990424, "grad_norm": 11.5703125, "learning_rate": 9.601335986355886e-06, "loss": 1.2837, "step": 11720 }, { "epoch": 1.247180251117259, "grad_norm": 12.15625, "learning_rate": 9.601300454803867e-06, "loss": 0.8916, "step": 11721 }, { "epoch": 1.2472866567354757, "grad_norm": 11.25, "learning_rate": 9.601264923251849e-06, "loss": 0.9445, "step": 11722 }, { "epoch": 1.2473930623536922, "grad_norm": 11.875, "learning_rate": 9.60122939169983e-06, "loss": 0.9019, "step": 11723 }, { "epoch": 1.247499467971909, "grad_norm": 17.046875, "learning_rate": 9.601193860147812e-06, "loss": 0.9538, "step": 11724 }, { "epoch": 1.2476058735901256, "grad_norm": 11.1484375, "learning_rate": 9.601158328595794e-06, "loss": 0.9327, "step": 11725 }, { "epoch": 1.2477122792083422, "grad_norm": 10.9375, "learning_rate": 9.601122797043775e-06, "loss": 1.1184, "step": 11726 }, { "epoch": 1.2478186848265589, "grad_norm": 12.75, "learning_rate": 9.601087265491757e-06, "loss": 1.0361, "step": 11727 }, { "epoch": 1.2479250904447754, "grad_norm": 8.6640625, "learning_rate": 9.601051733939739e-06, "loss": 1.0931, "step": 11728 }, { "epoch": 1.2480314960629921, "grad_norm": 14.1953125, "learning_rate": 9.60101620238772e-06, "loss": 1.0282, "step": 11729 }, { "epoch": 1.2481379016812089, "grad_norm": 9.0859375, "learning_rate": 9.600980670835703e-06, "loss": 0.9006, "step": 11730 }, { "epoch": 1.2482443072994254, "grad_norm": 13.4765625, "learning_rate": 9.600945139283685e-06, "loss": 1.0304, "step": 11731 }, { "epoch": 1.248350712917642, "grad_norm": 12.9296875, "learning_rate": 9.600909607731667e-06, "loss": 0.9677, "step": 11732 }, { "epoch": 1.2484571185358586, "grad_norm": 9.390625, "learning_rate": 9.600874076179648e-06, "loss": 1.2607, "step": 11733 }, { "epoch": 1.2485635241540753, "grad_norm": 17.4375, "learning_rate": 9.60083854462763e-06, "loss": 1.195, "step": 11734 }, { "epoch": 1.248669929772292, "grad_norm": 12.203125, "learning_rate": 9.600803013075613e-06, "loss": 0.9993, "step": 11735 }, { "epoch": 1.2487763353905086, "grad_norm": 14.5, "learning_rate": 9.600767481523595e-06, "loss": 1.2323, "step": 11736 }, { "epoch": 1.2488827410087253, "grad_norm": 14.875, "learning_rate": 9.600731949971576e-06, "loss": 0.9844, "step": 11737 }, { "epoch": 1.2489891466269418, "grad_norm": 9.3515625, "learning_rate": 9.600696418419558e-06, "loss": 1.2546, "step": 11738 }, { "epoch": 1.2490955522451586, "grad_norm": 15.9765625, "learning_rate": 9.60066088686754e-06, "loss": 1.1947, "step": 11739 }, { "epoch": 1.249201957863375, "grad_norm": 11.75, "learning_rate": 9.600625355315521e-06, "loss": 0.9181, "step": 11740 }, { "epoch": 1.2493083634815918, "grad_norm": 11.9375, "learning_rate": 9.600589823763503e-06, "loss": 1.1605, "step": 11741 }, { "epoch": 1.2494147690998085, "grad_norm": 9.5078125, "learning_rate": 9.600554292211484e-06, "loss": 0.9094, "step": 11742 }, { "epoch": 1.249521174718025, "grad_norm": 12.34375, "learning_rate": 9.600518760659466e-06, "loss": 1.2169, "step": 11743 }, { "epoch": 1.2496275803362418, "grad_norm": 11.609375, "learning_rate": 9.600483229107448e-06, "loss": 1.0314, "step": 11744 }, { "epoch": 1.2497339859544585, "grad_norm": 14.7265625, "learning_rate": 9.60044769755543e-06, "loss": 1.3004, "step": 11745 }, { "epoch": 1.249840391572675, "grad_norm": 16.640625, "learning_rate": 9.600412166003413e-06, "loss": 1.0163, "step": 11746 }, { "epoch": 1.2499467971908917, "grad_norm": 15.484375, "learning_rate": 9.600376634451394e-06, "loss": 1.0741, "step": 11747 }, { "epoch": 1.2500532028091083, "grad_norm": 11.21875, "learning_rate": 9.600341102899376e-06, "loss": 1.0724, "step": 11748 }, { "epoch": 1.250159608427325, "grad_norm": 11.3671875, "learning_rate": 9.600305571347357e-06, "loss": 1.3426, "step": 11749 }, { "epoch": 1.2502660140455415, "grad_norm": 15.703125, "learning_rate": 9.600270039795339e-06, "loss": 1.4274, "step": 11750 }, { "epoch": 1.2503724196637582, "grad_norm": 12.484375, "learning_rate": 9.60023450824332e-06, "loss": 0.9322, "step": 11751 }, { "epoch": 1.250478825281975, "grad_norm": 22.53125, "learning_rate": 9.600198976691302e-06, "loss": 1.0481, "step": 11752 }, { "epoch": 1.2505852309001915, "grad_norm": 9.6484375, "learning_rate": 9.600163445139284e-06, "loss": 0.8586, "step": 11753 }, { "epoch": 1.2506916365184082, "grad_norm": 12.5078125, "learning_rate": 9.600127913587265e-06, "loss": 1.1045, "step": 11754 }, { "epoch": 1.250798042136625, "grad_norm": 14.6953125, "learning_rate": 9.600092382035247e-06, "loss": 1.1811, "step": 11755 }, { "epoch": 1.2509044477548414, "grad_norm": 10.046875, "learning_rate": 9.60005685048323e-06, "loss": 1.0426, "step": 11756 }, { "epoch": 1.2510108533730582, "grad_norm": 12.75, "learning_rate": 9.600021318931212e-06, "loss": 1.0836, "step": 11757 }, { "epoch": 1.2511172589912747, "grad_norm": 12.6171875, "learning_rate": 9.599985787379194e-06, "loss": 1.0441, "step": 11758 }, { "epoch": 1.2512236646094914, "grad_norm": 11.609375, "learning_rate": 9.599950255827175e-06, "loss": 1.0516, "step": 11759 }, { "epoch": 1.251330070227708, "grad_norm": 11.15625, "learning_rate": 9.599914724275157e-06, "loss": 1.1152, "step": 11760 }, { "epoch": 1.2514364758459247, "grad_norm": 11.9140625, "learning_rate": 9.59987919272314e-06, "loss": 0.8182, "step": 11761 }, { "epoch": 1.2515428814641414, "grad_norm": 19.5, "learning_rate": 9.599843661171122e-06, "loss": 1.0701, "step": 11762 }, { "epoch": 1.251649287082358, "grad_norm": 10.9296875, "learning_rate": 9.599808129619103e-06, "loss": 0.9394, "step": 11763 }, { "epoch": 1.2517556927005746, "grad_norm": 12.359375, "learning_rate": 9.599772598067085e-06, "loss": 1.1477, "step": 11764 }, { "epoch": 1.2518620983187914, "grad_norm": 11.796875, "learning_rate": 9.599737066515067e-06, "loss": 0.9002, "step": 11765 }, { "epoch": 1.2519685039370079, "grad_norm": 19.1875, "learning_rate": 9.599701534963048e-06, "loss": 1.4681, "step": 11766 }, { "epoch": 1.2520749095552244, "grad_norm": 21.203125, "learning_rate": 9.59966600341103e-06, "loss": 0.7659, "step": 11767 }, { "epoch": 1.2521813151734411, "grad_norm": 17.71875, "learning_rate": 9.599630471859011e-06, "loss": 1.1323, "step": 11768 }, { "epoch": 1.2522877207916578, "grad_norm": 12.9296875, "learning_rate": 9.599594940306993e-06, "loss": 0.9373, "step": 11769 }, { "epoch": 1.2523941264098744, "grad_norm": 11.03125, "learning_rate": 9.599559408754975e-06, "loss": 0.9803, "step": 11770 }, { "epoch": 1.252500532028091, "grad_norm": 9.765625, "learning_rate": 9.599523877202958e-06, "loss": 1.0613, "step": 11771 }, { "epoch": 1.2526069376463078, "grad_norm": 9.109375, "learning_rate": 9.59948834565094e-06, "loss": 0.9159, "step": 11772 }, { "epoch": 1.2527133432645243, "grad_norm": 11.0703125, "learning_rate": 9.599452814098921e-06, "loss": 0.8146, "step": 11773 }, { "epoch": 1.252819748882741, "grad_norm": 9.3671875, "learning_rate": 9.599417282546903e-06, "loss": 1.1245, "step": 11774 }, { "epoch": 1.2529261545009576, "grad_norm": 7.859375, "learning_rate": 9.599381750994884e-06, "loss": 1.0234, "step": 11775 }, { "epoch": 1.2530325601191743, "grad_norm": 13.5859375, "learning_rate": 9.599346219442866e-06, "loss": 1.1415, "step": 11776 }, { "epoch": 1.2531389657373908, "grad_norm": 12.265625, "learning_rate": 9.599310687890847e-06, "loss": 0.9617, "step": 11777 }, { "epoch": 1.2532453713556075, "grad_norm": 13.4140625, "learning_rate": 9.599275156338829e-06, "loss": 0.9372, "step": 11778 }, { "epoch": 1.2533517769738243, "grad_norm": 16.0625, "learning_rate": 9.59923962478681e-06, "loss": 1.0781, "step": 11779 }, { "epoch": 1.2534581825920408, "grad_norm": 10.953125, "learning_rate": 9.599204093234792e-06, "loss": 0.7959, "step": 11780 }, { "epoch": 1.2535645882102575, "grad_norm": 10.5859375, "learning_rate": 9.599168561682774e-06, "loss": 1.1697, "step": 11781 }, { "epoch": 1.2536709938284742, "grad_norm": 18.921875, "learning_rate": 9.599133030130757e-06, "loss": 1.2484, "step": 11782 }, { "epoch": 1.2537773994466908, "grad_norm": 13.921875, "learning_rate": 9.599097498578739e-06, "loss": 1.0343, "step": 11783 }, { "epoch": 1.2538838050649075, "grad_norm": 15.5546875, "learning_rate": 9.59906196702672e-06, "loss": 1.4907, "step": 11784 }, { "epoch": 1.253990210683124, "grad_norm": 12.1484375, "learning_rate": 9.599026435474702e-06, "loss": 1.2777, "step": 11785 }, { "epoch": 1.2540966163013407, "grad_norm": 13.75, "learning_rate": 9.598990903922684e-06, "loss": 1.0287, "step": 11786 }, { "epoch": 1.2542030219195572, "grad_norm": 13.5078125, "learning_rate": 9.598955372370667e-06, "loss": 1.2128, "step": 11787 }, { "epoch": 1.254309427537774, "grad_norm": 10.5, "learning_rate": 9.598919840818649e-06, "loss": 1.0747, "step": 11788 }, { "epoch": 1.2544158331559907, "grad_norm": 13.90625, "learning_rate": 9.59888430926663e-06, "loss": 1.2938, "step": 11789 }, { "epoch": 1.2545222387742072, "grad_norm": 12.6015625, "learning_rate": 9.598848777714612e-06, "loss": 0.9655, "step": 11790 }, { "epoch": 1.254628644392424, "grad_norm": 14.265625, "learning_rate": 9.598813246162593e-06, "loss": 1.3322, "step": 11791 }, { "epoch": 1.2547350500106407, "grad_norm": 9.7578125, "learning_rate": 9.598777714610575e-06, "loss": 0.8194, "step": 11792 }, { "epoch": 1.2548414556288572, "grad_norm": 13.5859375, "learning_rate": 9.598742183058557e-06, "loss": 0.8803, "step": 11793 }, { "epoch": 1.254947861247074, "grad_norm": 11.40625, "learning_rate": 9.598706651506538e-06, "loss": 1.0844, "step": 11794 }, { "epoch": 1.2550542668652904, "grad_norm": 10.6953125, "learning_rate": 9.59867111995452e-06, "loss": 1.0656, "step": 11795 }, { "epoch": 1.2551606724835072, "grad_norm": 13.125, "learning_rate": 9.598635588402501e-06, "loss": 0.919, "step": 11796 }, { "epoch": 1.2552670781017237, "grad_norm": 13.9453125, "learning_rate": 9.598600056850485e-06, "loss": 1.2491, "step": 11797 }, { "epoch": 1.2553734837199404, "grad_norm": 17.15625, "learning_rate": 9.598564525298466e-06, "loss": 1.2225, "step": 11798 }, { "epoch": 1.2554798893381571, "grad_norm": 14.4375, "learning_rate": 9.598528993746448e-06, "loss": 1.2089, "step": 11799 }, { "epoch": 1.2555862949563736, "grad_norm": 12.0859375, "learning_rate": 9.59849346219443e-06, "loss": 1.0153, "step": 11800 }, { "epoch": 1.2556927005745904, "grad_norm": 12.015625, "learning_rate": 9.598457930642411e-06, "loss": 1.3759, "step": 11801 }, { "epoch": 1.255799106192807, "grad_norm": 10.6171875, "learning_rate": 9.598422399090393e-06, "loss": 0.9253, "step": 11802 }, { "epoch": 1.2559055118110236, "grad_norm": 11.140625, "learning_rate": 9.598386867538374e-06, "loss": 1.095, "step": 11803 }, { "epoch": 1.2560119174292403, "grad_norm": 12.125, "learning_rate": 9.598351335986356e-06, "loss": 1.3751, "step": 11804 }, { "epoch": 1.2561183230474569, "grad_norm": 12.4765625, "learning_rate": 9.598315804434338e-06, "loss": 1.124, "step": 11805 }, { "epoch": 1.2562247286656736, "grad_norm": 12.6015625, "learning_rate": 9.59828027288232e-06, "loss": 1.1696, "step": 11806 }, { "epoch": 1.25633113428389, "grad_norm": 13.7890625, "learning_rate": 9.598244741330302e-06, "loss": 1.153, "step": 11807 }, { "epoch": 1.2564375399021068, "grad_norm": 9.4140625, "learning_rate": 9.598209209778284e-06, "loss": 1.0961, "step": 11808 }, { "epoch": 1.2565439455203236, "grad_norm": 8.765625, "learning_rate": 9.598173678226266e-06, "loss": 1.1017, "step": 11809 }, { "epoch": 1.25665035113854, "grad_norm": 17.28125, "learning_rate": 9.598138146674247e-06, "loss": 1.0106, "step": 11810 }, { "epoch": 1.2567567567567568, "grad_norm": 13.0546875, "learning_rate": 9.598102615122229e-06, "loss": 0.8683, "step": 11811 }, { "epoch": 1.2568631623749735, "grad_norm": 11.6171875, "learning_rate": 9.598067083570212e-06, "loss": 1.0285, "step": 11812 }, { "epoch": 1.25696956799319, "grad_norm": 11.875, "learning_rate": 9.598031552018194e-06, "loss": 0.8851, "step": 11813 }, { "epoch": 1.2570759736114065, "grad_norm": 12.6171875, "learning_rate": 9.597996020466175e-06, "loss": 1.3237, "step": 11814 }, { "epoch": 1.2571823792296233, "grad_norm": 14.65625, "learning_rate": 9.597960488914157e-06, "loss": 1.039, "step": 11815 }, { "epoch": 1.25728878484784, "grad_norm": 17.578125, "learning_rate": 9.597924957362139e-06, "loss": 1.2686, "step": 11816 }, { "epoch": 1.2573951904660565, "grad_norm": 9.3828125, "learning_rate": 9.59788942581012e-06, "loss": 0.8197, "step": 11817 }, { "epoch": 1.2575015960842733, "grad_norm": 13.0390625, "learning_rate": 9.597853894258102e-06, "loss": 1.1303, "step": 11818 }, { "epoch": 1.25760800170249, "grad_norm": 10.8203125, "learning_rate": 9.597818362706083e-06, "loss": 0.9477, "step": 11819 }, { "epoch": 1.2577144073207065, "grad_norm": 12.015625, "learning_rate": 9.597782831154065e-06, "loss": 1.169, "step": 11820 }, { "epoch": 1.2578208129389232, "grad_norm": 11.734375, "learning_rate": 9.597747299602047e-06, "loss": 1.2343, "step": 11821 }, { "epoch": 1.2579272185571397, "grad_norm": 10.234375, "learning_rate": 9.597711768050028e-06, "loss": 0.949, "step": 11822 }, { "epoch": 1.2580336241753565, "grad_norm": 12.1015625, "learning_rate": 9.597676236498012e-06, "loss": 1.1138, "step": 11823 }, { "epoch": 1.258140029793573, "grad_norm": 14.875, "learning_rate": 9.597640704945993e-06, "loss": 1.4767, "step": 11824 }, { "epoch": 1.2582464354117897, "grad_norm": 17.140625, "learning_rate": 9.597605173393975e-06, "loss": 0.8855, "step": 11825 }, { "epoch": 1.2583528410300064, "grad_norm": 12.6875, "learning_rate": 9.597569641841956e-06, "loss": 0.986, "step": 11826 }, { "epoch": 1.258459246648223, "grad_norm": 10.34375, "learning_rate": 9.597534110289938e-06, "loss": 0.932, "step": 11827 }, { "epoch": 1.2585656522664397, "grad_norm": 10.1953125, "learning_rate": 9.597498578737921e-06, "loss": 1.1742, "step": 11828 }, { "epoch": 1.2586720578846564, "grad_norm": 12.046875, "learning_rate": 9.597463047185903e-06, "loss": 1.1391, "step": 11829 }, { "epoch": 1.258778463502873, "grad_norm": 12.03125, "learning_rate": 9.597427515633883e-06, "loss": 0.9112, "step": 11830 }, { "epoch": 1.2588848691210897, "grad_norm": 10.421875, "learning_rate": 9.597391984081864e-06, "loss": 1.1169, "step": 11831 }, { "epoch": 1.2589912747393062, "grad_norm": 13.2890625, "learning_rate": 9.597356452529846e-06, "loss": 1.1051, "step": 11832 }, { "epoch": 1.259097680357523, "grad_norm": 10.5390625, "learning_rate": 9.59732092097783e-06, "loss": 1.1638, "step": 11833 }, { "epoch": 1.2592040859757394, "grad_norm": 11.828125, "learning_rate": 9.597285389425811e-06, "loss": 1.0263, "step": 11834 }, { "epoch": 1.2593104915939561, "grad_norm": 12.265625, "learning_rate": 9.597249857873793e-06, "loss": 1.1884, "step": 11835 }, { "epoch": 1.2594168972121729, "grad_norm": 13.890625, "learning_rate": 9.597214326321774e-06, "loss": 0.9784, "step": 11836 }, { "epoch": 1.2595233028303894, "grad_norm": 14.1640625, "learning_rate": 9.597178794769756e-06, "loss": 1.119, "step": 11837 }, { "epoch": 1.2596297084486061, "grad_norm": 11.1015625, "learning_rate": 9.597143263217739e-06, "loss": 1.0468, "step": 11838 }, { "epoch": 1.2597361140668228, "grad_norm": 12.65625, "learning_rate": 9.59710773166572e-06, "loss": 1.0563, "step": 11839 }, { "epoch": 1.2598425196850394, "grad_norm": 18.6875, "learning_rate": 9.597072200113702e-06, "loss": 1.2902, "step": 11840 }, { "epoch": 1.259948925303256, "grad_norm": 11.1796875, "learning_rate": 9.597036668561684e-06, "loss": 1.1703, "step": 11841 }, { "epoch": 1.2600553309214726, "grad_norm": 12.484375, "learning_rate": 9.597001137009665e-06, "loss": 1.0088, "step": 11842 }, { "epoch": 1.2601617365396893, "grad_norm": 11.7734375, "learning_rate": 9.596965605457647e-06, "loss": 1.0751, "step": 11843 }, { "epoch": 1.2602681421579058, "grad_norm": 12.8515625, "learning_rate": 9.596930073905629e-06, "loss": 1.0684, "step": 11844 }, { "epoch": 1.2603745477761226, "grad_norm": 13.65625, "learning_rate": 9.59689454235361e-06, "loss": 1.2221, "step": 11845 }, { "epoch": 1.2604809533943393, "grad_norm": 12.140625, "learning_rate": 9.596859010801592e-06, "loss": 1.0228, "step": 11846 }, { "epoch": 1.2605873590125558, "grad_norm": 14.5, "learning_rate": 9.596823479249573e-06, "loss": 1.119, "step": 11847 }, { "epoch": 1.2606937646307725, "grad_norm": 14.46875, "learning_rate": 9.596787947697557e-06, "loss": 1.5062, "step": 11848 }, { "epoch": 1.2608001702489893, "grad_norm": 15.59375, "learning_rate": 9.596752416145538e-06, "loss": 1.097, "step": 11849 }, { "epoch": 1.2609065758672058, "grad_norm": 11.234375, "learning_rate": 9.59671688459352e-06, "loss": 1.2585, "step": 11850 }, { "epoch": 1.2610129814854225, "grad_norm": 13.234375, "learning_rate": 9.596681353041502e-06, "loss": 1.0977, "step": 11851 }, { "epoch": 1.261119387103639, "grad_norm": 13.5859375, "learning_rate": 9.596645821489483e-06, "loss": 1.2132, "step": 11852 }, { "epoch": 1.2612257927218558, "grad_norm": 11.328125, "learning_rate": 9.596610289937467e-06, "loss": 1.0437, "step": 11853 }, { "epoch": 1.2613321983400723, "grad_norm": 13.1328125, "learning_rate": 9.596574758385448e-06, "loss": 1.0685, "step": 11854 }, { "epoch": 1.261438603958289, "grad_norm": 11.25, "learning_rate": 9.59653922683343e-06, "loss": 0.9967, "step": 11855 }, { "epoch": 1.2615450095765057, "grad_norm": 12.6796875, "learning_rate": 9.596503695281411e-06, "loss": 1.1421, "step": 11856 }, { "epoch": 1.2616514151947222, "grad_norm": 10.96875, "learning_rate": 9.596468163729393e-06, "loss": 1.1155, "step": 11857 }, { "epoch": 1.261757820812939, "grad_norm": 12.140625, "learning_rate": 9.596432632177375e-06, "loss": 1.0483, "step": 11858 }, { "epoch": 1.2618642264311557, "grad_norm": 13.3984375, "learning_rate": 9.596397100625356e-06, "loss": 1.0939, "step": 11859 }, { "epoch": 1.2619706320493722, "grad_norm": 10.6796875, "learning_rate": 9.596361569073338e-06, "loss": 1.0258, "step": 11860 }, { "epoch": 1.2620770376675887, "grad_norm": 19.71875, "learning_rate": 9.59632603752132e-06, "loss": 1.0976, "step": 11861 }, { "epoch": 1.2621834432858055, "grad_norm": 9.5078125, "learning_rate": 9.596290505969301e-06, "loss": 1.0149, "step": 11862 }, { "epoch": 1.2622898489040222, "grad_norm": 12.7109375, "learning_rate": 9.596254974417283e-06, "loss": 0.9898, "step": 11863 }, { "epoch": 1.2623962545222387, "grad_norm": 13.1640625, "learning_rate": 9.596219442865266e-06, "loss": 1.134, "step": 11864 }, { "epoch": 1.2625026601404554, "grad_norm": 18.53125, "learning_rate": 9.596183911313247e-06, "loss": 1.16, "step": 11865 }, { "epoch": 1.2626090657586722, "grad_norm": 12.9140625, "learning_rate": 9.596148379761229e-06, "loss": 1.1817, "step": 11866 }, { "epoch": 1.2627154713768887, "grad_norm": 12.3828125, "learning_rate": 9.59611284820921e-06, "loss": 1.0579, "step": 11867 }, { "epoch": 1.2628218769951054, "grad_norm": 11.2109375, "learning_rate": 9.596077316657192e-06, "loss": 1.2951, "step": 11868 }, { "epoch": 1.262928282613322, "grad_norm": 14.8671875, "learning_rate": 9.596041785105174e-06, "loss": 1.4151, "step": 11869 }, { "epoch": 1.2630346882315386, "grad_norm": 11.1640625, "learning_rate": 9.596006253553156e-06, "loss": 1.2989, "step": 11870 }, { "epoch": 1.2631410938497551, "grad_norm": 9.3828125, "learning_rate": 9.595970722001137e-06, "loss": 1.1744, "step": 11871 }, { "epoch": 1.2632474994679719, "grad_norm": 10.3828125, "learning_rate": 9.595935190449119e-06, "loss": 1.2627, "step": 11872 }, { "epoch": 1.2633539050861886, "grad_norm": 9.15625, "learning_rate": 9.5958996588971e-06, "loss": 0.9823, "step": 11873 }, { "epoch": 1.2634603107044051, "grad_norm": 13.46875, "learning_rate": 9.595864127345084e-06, "loss": 0.9798, "step": 11874 }, { "epoch": 1.2635667163226219, "grad_norm": 14.6796875, "learning_rate": 9.595828595793065e-06, "loss": 1.341, "step": 11875 }, { "epoch": 1.2636731219408386, "grad_norm": 15.375, "learning_rate": 9.595793064241047e-06, "loss": 0.9519, "step": 11876 }, { "epoch": 1.263779527559055, "grad_norm": 12.515625, "learning_rate": 9.595757532689028e-06, "loss": 1.078, "step": 11877 }, { "epoch": 1.2638859331772718, "grad_norm": 12.578125, "learning_rate": 9.59572200113701e-06, "loss": 1.0464, "step": 11878 }, { "epoch": 1.2639923387954883, "grad_norm": 14.375, "learning_rate": 9.595686469584993e-06, "loss": 1.0033, "step": 11879 }, { "epoch": 1.264098744413705, "grad_norm": 10.8046875, "learning_rate": 9.595650938032975e-06, "loss": 1.0568, "step": 11880 }, { "epoch": 1.2642051500319216, "grad_norm": 21.140625, "learning_rate": 9.595615406480957e-06, "loss": 0.8931, "step": 11881 }, { "epoch": 1.2643115556501383, "grad_norm": 15.0625, "learning_rate": 9.595579874928938e-06, "loss": 1.1914, "step": 11882 }, { "epoch": 1.264417961268355, "grad_norm": 14.4140625, "learning_rate": 9.59554434337692e-06, "loss": 1.25, "step": 11883 }, { "epoch": 1.2645243668865715, "grad_norm": 15.5859375, "learning_rate": 9.595508811824901e-06, "loss": 1.1664, "step": 11884 }, { "epoch": 1.2646307725047883, "grad_norm": 15.2265625, "learning_rate": 9.595473280272883e-06, "loss": 1.2169, "step": 11885 }, { "epoch": 1.264737178123005, "grad_norm": 19.765625, "learning_rate": 9.595437748720865e-06, "loss": 0.8578, "step": 11886 }, { "epoch": 1.2648435837412215, "grad_norm": 12.140625, "learning_rate": 9.595402217168846e-06, "loss": 1.1005, "step": 11887 }, { "epoch": 1.2649499893594383, "grad_norm": 11.203125, "learning_rate": 9.595366685616828e-06, "loss": 1.3096, "step": 11888 }, { "epoch": 1.2650563949776548, "grad_norm": 15.40625, "learning_rate": 9.595331154064811e-06, "loss": 0.9479, "step": 11889 }, { "epoch": 1.2651628005958715, "grad_norm": 12.25, "learning_rate": 9.595295622512793e-06, "loss": 1.095, "step": 11890 }, { "epoch": 1.265269206214088, "grad_norm": 13.96875, "learning_rate": 9.595260090960774e-06, "loss": 1.2625, "step": 11891 }, { "epoch": 1.2653756118323047, "grad_norm": 13.796875, "learning_rate": 9.595224559408756e-06, "loss": 1.1682, "step": 11892 }, { "epoch": 1.2654820174505215, "grad_norm": 14.5234375, "learning_rate": 9.595189027856738e-06, "loss": 0.7476, "step": 11893 }, { "epoch": 1.265588423068738, "grad_norm": 10.671875, "learning_rate": 9.59515349630472e-06, "loss": 1.11, "step": 11894 }, { "epoch": 1.2656948286869547, "grad_norm": 10.921875, "learning_rate": 9.5951179647527e-06, "loss": 0.994, "step": 11895 }, { "epoch": 1.2658012343051714, "grad_norm": 11.0390625, "learning_rate": 9.595082433200682e-06, "loss": 1.1001, "step": 11896 }, { "epoch": 1.265907639923388, "grad_norm": 14.4296875, "learning_rate": 9.595046901648664e-06, "loss": 1.0437, "step": 11897 }, { "epoch": 1.2660140455416045, "grad_norm": 12.859375, "learning_rate": 9.595011370096646e-06, "loss": 1.1859, "step": 11898 }, { "epoch": 1.2661204511598212, "grad_norm": 13.1328125, "learning_rate": 9.594975838544627e-06, "loss": 0.8618, "step": 11899 }, { "epoch": 1.266226856778038, "grad_norm": 19.625, "learning_rate": 9.59494030699261e-06, "loss": 1.2105, "step": 11900 }, { "epoch": 1.2663332623962544, "grad_norm": 11.40625, "learning_rate": 9.594904775440592e-06, "loss": 0.7922, "step": 11901 }, { "epoch": 1.2664396680144712, "grad_norm": 14.2578125, "learning_rate": 9.594869243888574e-06, "loss": 1.1981, "step": 11902 }, { "epoch": 1.266546073632688, "grad_norm": 13.3203125, "learning_rate": 9.594833712336555e-06, "loss": 0.9794, "step": 11903 }, { "epoch": 1.2666524792509044, "grad_norm": 11.875, "learning_rate": 9.594798180784537e-06, "loss": 0.9059, "step": 11904 }, { "epoch": 1.2667588848691211, "grad_norm": 12.125, "learning_rate": 9.59476264923252e-06, "loss": 1.1257, "step": 11905 }, { "epoch": 1.2668652904873379, "grad_norm": 14.5234375, "learning_rate": 9.594727117680502e-06, "loss": 1.1682, "step": 11906 }, { "epoch": 1.2669716961055544, "grad_norm": 10.9140625, "learning_rate": 9.594691586128483e-06, "loss": 1.0935, "step": 11907 }, { "epoch": 1.2670781017237709, "grad_norm": 10.3828125, "learning_rate": 9.594656054576465e-06, "loss": 0.8914, "step": 11908 }, { "epoch": 1.2671845073419876, "grad_norm": 14.3984375, "learning_rate": 9.594620523024447e-06, "loss": 1.0686, "step": 11909 }, { "epoch": 1.2672909129602044, "grad_norm": 12.15625, "learning_rate": 9.594584991472428e-06, "loss": 1.3984, "step": 11910 }, { "epoch": 1.2673973185784209, "grad_norm": 13.9140625, "learning_rate": 9.59454945992041e-06, "loss": 1.3066, "step": 11911 }, { "epoch": 1.2675037241966376, "grad_norm": 12.0625, "learning_rate": 9.594513928368391e-06, "loss": 1.4479, "step": 11912 }, { "epoch": 1.2676101298148543, "grad_norm": 12.5703125, "learning_rate": 9.594478396816373e-06, "loss": 1.3278, "step": 11913 }, { "epoch": 1.2677165354330708, "grad_norm": 13.5390625, "learning_rate": 9.594442865264355e-06, "loss": 1.2234, "step": 11914 }, { "epoch": 1.2678229410512876, "grad_norm": 14.3046875, "learning_rate": 9.594407333712338e-06, "loss": 1.2293, "step": 11915 }, { "epoch": 1.267929346669504, "grad_norm": 9.9609375, "learning_rate": 9.59437180216032e-06, "loss": 1.3002, "step": 11916 }, { "epoch": 1.2680357522877208, "grad_norm": 11.6796875, "learning_rate": 9.594336270608301e-06, "loss": 0.817, "step": 11917 }, { "epoch": 1.2681421579059373, "grad_norm": 12.484375, "learning_rate": 9.594300739056283e-06, "loss": 0.9798, "step": 11918 }, { "epoch": 1.268248563524154, "grad_norm": 11.734375, "learning_rate": 9.594265207504264e-06, "loss": 1.0774, "step": 11919 }, { "epoch": 1.2683549691423708, "grad_norm": 14.109375, "learning_rate": 9.594229675952246e-06, "loss": 0.9059, "step": 11920 }, { "epoch": 1.2684613747605873, "grad_norm": 18.0625, "learning_rate": 9.594194144400228e-06, "loss": 1.4082, "step": 11921 }, { "epoch": 1.268567780378804, "grad_norm": 10.6484375, "learning_rate": 9.59415861284821e-06, "loss": 1.0617, "step": 11922 }, { "epoch": 1.2686741859970208, "grad_norm": 12.1484375, "learning_rate": 9.594123081296191e-06, "loss": 1.2311, "step": 11923 }, { "epoch": 1.2687805916152373, "grad_norm": 10.296875, "learning_rate": 9.594087549744172e-06, "loss": 0.8412, "step": 11924 }, { "epoch": 1.268886997233454, "grad_norm": 12.421875, "learning_rate": 9.594052018192156e-06, "loss": 0.9422, "step": 11925 }, { "epoch": 1.2689934028516705, "grad_norm": 13.6640625, "learning_rate": 9.594016486640137e-06, "loss": 1.0129, "step": 11926 }, { "epoch": 1.2690998084698872, "grad_norm": 12.1171875, "learning_rate": 9.593980955088119e-06, "loss": 1.2046, "step": 11927 }, { "epoch": 1.2692062140881037, "grad_norm": 11.8125, "learning_rate": 9.5939454235361e-06, "loss": 1.1656, "step": 11928 }, { "epoch": 1.2693126197063205, "grad_norm": 11.234375, "learning_rate": 9.593909891984082e-06, "loss": 0.9981, "step": 11929 }, { "epoch": 1.2694190253245372, "grad_norm": 15.7734375, "learning_rate": 9.593874360432065e-06, "loss": 1.0244, "step": 11930 }, { "epoch": 1.2695254309427537, "grad_norm": 11.1953125, "learning_rate": 9.593838828880047e-06, "loss": 1.0566, "step": 11931 }, { "epoch": 1.2696318365609705, "grad_norm": 11.9453125, "learning_rate": 9.593803297328029e-06, "loss": 0.8793, "step": 11932 }, { "epoch": 1.2697382421791872, "grad_norm": 13.9921875, "learning_rate": 9.59376776577601e-06, "loss": 1.3051, "step": 11933 }, { "epoch": 1.2698446477974037, "grad_norm": 13.40625, "learning_rate": 9.593732234223992e-06, "loss": 0.9592, "step": 11934 }, { "epoch": 1.2699510534156204, "grad_norm": 24.3125, "learning_rate": 9.593696702671974e-06, "loss": 1.0852, "step": 11935 }, { "epoch": 1.270057459033837, "grad_norm": 11.9453125, "learning_rate": 9.593661171119955e-06, "loss": 1.0544, "step": 11936 }, { "epoch": 1.2701638646520537, "grad_norm": 14.859375, "learning_rate": 9.593625639567937e-06, "loss": 0.8133, "step": 11937 }, { "epoch": 1.2702702702702702, "grad_norm": 15.3671875, "learning_rate": 9.593590108015918e-06, "loss": 1.203, "step": 11938 }, { "epoch": 1.270376675888487, "grad_norm": 9.4609375, "learning_rate": 9.5935545764639e-06, "loss": 1.1161, "step": 11939 }, { "epoch": 1.2704830815067036, "grad_norm": 13.140625, "learning_rate": 9.593519044911882e-06, "loss": 1.038, "step": 11940 }, { "epoch": 1.2705894871249201, "grad_norm": 11.7890625, "learning_rate": 9.593483513359865e-06, "loss": 1.0965, "step": 11941 }, { "epoch": 1.2706958927431369, "grad_norm": 9.6484375, "learning_rate": 9.593447981807846e-06, "loss": 1.1327, "step": 11942 }, { "epoch": 1.2708022983613536, "grad_norm": 14.7578125, "learning_rate": 9.593412450255828e-06, "loss": 1.358, "step": 11943 }, { "epoch": 1.2709087039795701, "grad_norm": 13.2421875, "learning_rate": 9.59337691870381e-06, "loss": 1.191, "step": 11944 }, { "epoch": 1.2710151095977866, "grad_norm": 12.2890625, "learning_rate": 9.593341387151791e-06, "loss": 1.2456, "step": 11945 }, { "epoch": 1.2711215152160034, "grad_norm": 11.140625, "learning_rate": 9.593305855599775e-06, "loss": 1.1669, "step": 11946 }, { "epoch": 1.27122792083422, "grad_norm": 15.7109375, "learning_rate": 9.593270324047756e-06, "loss": 1.0215, "step": 11947 }, { "epoch": 1.2713343264524366, "grad_norm": 18.078125, "learning_rate": 9.593234792495738e-06, "loss": 0.9039, "step": 11948 }, { "epoch": 1.2714407320706533, "grad_norm": 11.6484375, "learning_rate": 9.59319926094372e-06, "loss": 1.2664, "step": 11949 }, { "epoch": 1.27154713768887, "grad_norm": 12.140625, "learning_rate": 9.5931637293917e-06, "loss": 1.3773, "step": 11950 }, { "epoch": 1.2716535433070866, "grad_norm": 13.171875, "learning_rate": 9.593128197839683e-06, "loss": 1.0004, "step": 11951 }, { "epoch": 1.2717599489253033, "grad_norm": 16.4375, "learning_rate": 9.593092666287664e-06, "loss": 1.0452, "step": 11952 }, { "epoch": 1.2718663545435198, "grad_norm": 13.8359375, "learning_rate": 9.593057134735646e-06, "loss": 0.8393, "step": 11953 }, { "epoch": 1.2719727601617365, "grad_norm": 13.234375, "learning_rate": 9.593021603183627e-06, "loss": 1.2266, "step": 11954 }, { "epoch": 1.272079165779953, "grad_norm": 11.3671875, "learning_rate": 9.592986071631609e-06, "loss": 0.842, "step": 11955 }, { "epoch": 1.2721855713981698, "grad_norm": 12.6484375, "learning_rate": 9.592950540079592e-06, "loss": 1.2409, "step": 11956 }, { "epoch": 1.2722919770163865, "grad_norm": 12.6953125, "learning_rate": 9.592915008527574e-06, "loss": 1.1045, "step": 11957 }, { "epoch": 1.272398382634603, "grad_norm": 9.765625, "learning_rate": 9.592879476975556e-06, "loss": 1.0634, "step": 11958 }, { "epoch": 1.2725047882528198, "grad_norm": 11.921875, "learning_rate": 9.592843945423537e-06, "loss": 0.9451, "step": 11959 }, { "epoch": 1.2726111938710365, "grad_norm": 14.2265625, "learning_rate": 9.592808413871519e-06, "loss": 0.9965, "step": 11960 }, { "epoch": 1.272717599489253, "grad_norm": 11.3046875, "learning_rate": 9.5927728823195e-06, "loss": 1.1779, "step": 11961 }, { "epoch": 1.2728240051074697, "grad_norm": 11.0, "learning_rate": 9.592737350767482e-06, "loss": 1.1643, "step": 11962 }, { "epoch": 1.2729304107256862, "grad_norm": 11.0234375, "learning_rate": 9.592701819215464e-06, "loss": 1.0425, "step": 11963 }, { "epoch": 1.273036816343903, "grad_norm": 10.953125, "learning_rate": 9.592666287663445e-06, "loss": 0.9326, "step": 11964 }, { "epoch": 1.2731432219621195, "grad_norm": 8.8671875, "learning_rate": 9.592630756111427e-06, "loss": 1.0415, "step": 11965 }, { "epoch": 1.2732496275803362, "grad_norm": 24.359375, "learning_rate": 9.59259522455941e-06, "loss": 1.082, "step": 11966 }, { "epoch": 1.273356033198553, "grad_norm": 14.4296875, "learning_rate": 9.592559693007392e-06, "loss": 1.0567, "step": 11967 }, { "epoch": 1.2734624388167695, "grad_norm": 17.234375, "learning_rate": 9.592524161455373e-06, "loss": 0.9692, "step": 11968 }, { "epoch": 1.2735688444349862, "grad_norm": 11.1328125, "learning_rate": 9.592488629903355e-06, "loss": 1.1338, "step": 11969 }, { "epoch": 1.273675250053203, "grad_norm": 17.296875, "learning_rate": 9.592453098351337e-06, "loss": 1.2323, "step": 11970 }, { "epoch": 1.2737816556714194, "grad_norm": 9.859375, "learning_rate": 9.59241756679932e-06, "loss": 1.1257, "step": 11971 }, { "epoch": 1.2738880612896362, "grad_norm": 12.6796875, "learning_rate": 9.592382035247301e-06, "loss": 1.2933, "step": 11972 }, { "epoch": 1.2739944669078527, "grad_norm": 14.4921875, "learning_rate": 9.592346503695283e-06, "loss": 0.93, "step": 11973 }, { "epoch": 1.2741008725260694, "grad_norm": 14.3046875, "learning_rate": 9.592310972143265e-06, "loss": 1.1255, "step": 11974 }, { "epoch": 1.274207278144286, "grad_norm": 17.0625, "learning_rate": 9.592275440591246e-06, "loss": 0.9766, "step": 11975 }, { "epoch": 1.2743136837625026, "grad_norm": 12.9609375, "learning_rate": 9.592239909039228e-06, "loss": 1.1767, "step": 11976 }, { "epoch": 1.2744200893807194, "grad_norm": 11.28125, "learning_rate": 9.59220437748721e-06, "loss": 1.0277, "step": 11977 }, { "epoch": 1.2745264949989359, "grad_norm": 13.0859375, "learning_rate": 9.592168845935191e-06, "loss": 1.0667, "step": 11978 }, { "epoch": 1.2746329006171526, "grad_norm": 15.7109375, "learning_rate": 9.592133314383173e-06, "loss": 1.3161, "step": 11979 }, { "epoch": 1.2747393062353694, "grad_norm": 13.0546875, "learning_rate": 9.592097782831154e-06, "loss": 1.0923, "step": 11980 }, { "epoch": 1.2748457118535859, "grad_norm": 12.71875, "learning_rate": 9.592062251279136e-06, "loss": 1.2858, "step": 11981 }, { "epoch": 1.2749521174718026, "grad_norm": 14.6875, "learning_rate": 9.59202671972712e-06, "loss": 0.9295, "step": 11982 }, { "epoch": 1.275058523090019, "grad_norm": 8.6015625, "learning_rate": 9.5919911881751e-06, "loss": 1.0783, "step": 11983 }, { "epoch": 1.2751649287082358, "grad_norm": 14.7734375, "learning_rate": 9.591955656623082e-06, "loss": 1.0218, "step": 11984 }, { "epoch": 1.2752713343264523, "grad_norm": 9.46875, "learning_rate": 9.591920125071064e-06, "loss": 1.0012, "step": 11985 }, { "epoch": 1.275377739944669, "grad_norm": 11.921875, "learning_rate": 9.591884593519046e-06, "loss": 1.1226, "step": 11986 }, { "epoch": 1.2754841455628858, "grad_norm": 13.9921875, "learning_rate": 9.591849061967027e-06, "loss": 0.9857, "step": 11987 }, { "epoch": 1.2755905511811023, "grad_norm": 22.25, "learning_rate": 9.591813530415009e-06, "loss": 1.1345, "step": 11988 }, { "epoch": 1.275696956799319, "grad_norm": 10.796875, "learning_rate": 9.59177799886299e-06, "loss": 1.0545, "step": 11989 }, { "epoch": 1.2758033624175358, "grad_norm": 11.90625, "learning_rate": 9.591742467310972e-06, "loss": 0.9767, "step": 11990 }, { "epoch": 1.2759097680357523, "grad_norm": 10.3828125, "learning_rate": 9.591706935758954e-06, "loss": 0.914, "step": 11991 }, { "epoch": 1.2760161736539688, "grad_norm": 10.1640625, "learning_rate": 9.591671404206937e-06, "loss": 1.089, "step": 11992 }, { "epoch": 1.2761225792721855, "grad_norm": 19.234375, "learning_rate": 9.591635872654919e-06, "loss": 1.0368, "step": 11993 }, { "epoch": 1.2762289848904023, "grad_norm": 10.0, "learning_rate": 9.5916003411029e-06, "loss": 0.9264, "step": 11994 }, { "epoch": 1.2763353905086188, "grad_norm": 15.21875, "learning_rate": 9.591564809550882e-06, "loss": 0.9253, "step": 11995 }, { "epoch": 1.2764417961268355, "grad_norm": 13.8359375, "learning_rate": 9.591529277998863e-06, "loss": 1.1348, "step": 11996 }, { "epoch": 1.2765482017450522, "grad_norm": 11.5859375, "learning_rate": 9.591493746446847e-06, "loss": 1.0431, "step": 11997 }, { "epoch": 1.2766546073632687, "grad_norm": 11.265625, "learning_rate": 9.591458214894828e-06, "loss": 1.0699, "step": 11998 }, { "epoch": 1.2767610129814855, "grad_norm": 17.0625, "learning_rate": 9.59142268334281e-06, "loss": 1.0998, "step": 11999 }, { "epoch": 1.276867418599702, "grad_norm": 17.34375, "learning_rate": 9.591387151790792e-06, "loss": 1.0392, "step": 12000 }, { "epoch": 1.2769738242179187, "grad_norm": 12.640625, "learning_rate": 9.591351620238773e-06, "loss": 1.0852, "step": 12001 }, { "epoch": 1.2770802298361352, "grad_norm": 14.6796875, "learning_rate": 9.591316088686755e-06, "loss": 1.1419, "step": 12002 }, { "epoch": 1.277186635454352, "grad_norm": 15.0234375, "learning_rate": 9.591280557134736e-06, "loss": 1.311, "step": 12003 }, { "epoch": 1.2772930410725687, "grad_norm": 11.328125, "learning_rate": 9.591245025582718e-06, "loss": 1.1217, "step": 12004 }, { "epoch": 1.2773994466907852, "grad_norm": 13.1171875, "learning_rate": 9.5912094940307e-06, "loss": 1.2619, "step": 12005 }, { "epoch": 1.277505852309002, "grad_norm": 16.25, "learning_rate": 9.591173962478681e-06, "loss": 1.1021, "step": 12006 }, { "epoch": 1.2776122579272187, "grad_norm": 9.7421875, "learning_rate": 9.591138430926664e-06, "loss": 1.0689, "step": 12007 }, { "epoch": 1.2777186635454352, "grad_norm": 10.6015625, "learning_rate": 9.591102899374646e-06, "loss": 1.1774, "step": 12008 }, { "epoch": 1.277825069163652, "grad_norm": 12.609375, "learning_rate": 9.591067367822628e-06, "loss": 1.0052, "step": 12009 }, { "epoch": 1.2779314747818684, "grad_norm": 11.9453125, "learning_rate": 9.59103183627061e-06, "loss": 1.2196, "step": 12010 }, { "epoch": 1.2780378804000851, "grad_norm": 9.5703125, "learning_rate": 9.590996304718591e-06, "loss": 1.0908, "step": 12011 }, { "epoch": 1.2781442860183017, "grad_norm": 13.2734375, "learning_rate": 9.590960773166572e-06, "loss": 1.0507, "step": 12012 }, { "epoch": 1.2782506916365184, "grad_norm": 15.7890625, "learning_rate": 9.590925241614554e-06, "loss": 1.215, "step": 12013 }, { "epoch": 1.2783570972547351, "grad_norm": 14.34375, "learning_rate": 9.590889710062536e-06, "loss": 1.2309, "step": 12014 }, { "epoch": 1.2784635028729516, "grad_norm": 10.859375, "learning_rate": 9.590854178510517e-06, "loss": 1.4226, "step": 12015 }, { "epoch": 1.2785699084911684, "grad_norm": 12.8828125, "learning_rate": 9.590818646958499e-06, "loss": 1.1459, "step": 12016 }, { "epoch": 1.278676314109385, "grad_norm": 12.8515625, "learning_rate": 9.59078311540648e-06, "loss": 0.9468, "step": 12017 }, { "epoch": 1.2787827197276016, "grad_norm": 11.3359375, "learning_rate": 9.590747583854464e-06, "loss": 1.1115, "step": 12018 }, { "epoch": 1.2788891253458183, "grad_norm": 14.2265625, "learning_rate": 9.590712052302445e-06, "loss": 1.2335, "step": 12019 }, { "epoch": 1.2789955309640348, "grad_norm": 12.875, "learning_rate": 9.590676520750427e-06, "loss": 1.2678, "step": 12020 }, { "epoch": 1.2791019365822516, "grad_norm": 10.5546875, "learning_rate": 9.590640989198409e-06, "loss": 0.938, "step": 12021 }, { "epoch": 1.279208342200468, "grad_norm": 14.34375, "learning_rate": 9.59060545764639e-06, "loss": 1.2234, "step": 12022 }, { "epoch": 1.2793147478186848, "grad_norm": 16.53125, "learning_rate": 9.590569926094374e-06, "loss": 1.0727, "step": 12023 }, { "epoch": 1.2794211534369015, "grad_norm": 12.84375, "learning_rate": 9.590534394542355e-06, "loss": 1.0829, "step": 12024 }, { "epoch": 1.279527559055118, "grad_norm": 11.078125, "learning_rate": 9.590498862990337e-06, "loss": 0.9295, "step": 12025 }, { "epoch": 1.2796339646733348, "grad_norm": 12.390625, "learning_rate": 9.590463331438318e-06, "loss": 1.0447, "step": 12026 }, { "epoch": 1.2797403702915515, "grad_norm": 13.7890625, "learning_rate": 9.5904277998863e-06, "loss": 1.0675, "step": 12027 }, { "epoch": 1.279846775909768, "grad_norm": 11.296875, "learning_rate": 9.590392268334282e-06, "loss": 1.1811, "step": 12028 }, { "epoch": 1.2799531815279845, "grad_norm": 11.5703125, "learning_rate": 9.590356736782263e-06, "loss": 1.0252, "step": 12029 }, { "epoch": 1.2800595871462013, "grad_norm": 9.0390625, "learning_rate": 9.590321205230245e-06, "loss": 1.0319, "step": 12030 }, { "epoch": 1.280165992764418, "grad_norm": 9.9609375, "learning_rate": 9.590285673678226e-06, "loss": 0.8302, "step": 12031 }, { "epoch": 1.2802723983826345, "grad_norm": 16.75, "learning_rate": 9.590250142126208e-06, "loss": 1.1798, "step": 12032 }, { "epoch": 1.2803788040008512, "grad_norm": 13.9609375, "learning_rate": 9.590214610574191e-06, "loss": 1.0668, "step": 12033 }, { "epoch": 1.280485209619068, "grad_norm": 16.0625, "learning_rate": 9.590179079022173e-06, "loss": 1.2345, "step": 12034 }, { "epoch": 1.2805916152372845, "grad_norm": 13.7109375, "learning_rate": 9.590143547470155e-06, "loss": 1.2025, "step": 12035 }, { "epoch": 1.2806980208555012, "grad_norm": 15.75, "learning_rate": 9.590108015918136e-06, "loss": 1.1034, "step": 12036 }, { "epoch": 1.280804426473718, "grad_norm": 9.765625, "learning_rate": 9.590072484366118e-06, "loss": 1.0935, "step": 12037 }, { "epoch": 1.2809108320919345, "grad_norm": 22.078125, "learning_rate": 9.590036952814101e-06, "loss": 1.1469, "step": 12038 }, { "epoch": 1.281017237710151, "grad_norm": 17.828125, "learning_rate": 9.590001421262081e-06, "loss": 0.9793, "step": 12039 }, { "epoch": 1.2811236433283677, "grad_norm": 13.7734375, "learning_rate": 9.589965889710063e-06, "loss": 1.3274, "step": 12040 }, { "epoch": 1.2812300489465844, "grad_norm": 14.6171875, "learning_rate": 9.589930358158044e-06, "loss": 1.146, "step": 12041 }, { "epoch": 1.281336454564801, "grad_norm": 11.640625, "learning_rate": 9.589894826606026e-06, "loss": 1.3253, "step": 12042 }, { "epoch": 1.2814428601830177, "grad_norm": 13.3515625, "learning_rate": 9.589859295054009e-06, "loss": 0.9416, "step": 12043 }, { "epoch": 1.2815492658012344, "grad_norm": 13.1640625, "learning_rate": 9.58982376350199e-06, "loss": 1.0638, "step": 12044 }, { "epoch": 1.281655671419451, "grad_norm": 12.0546875, "learning_rate": 9.589788231949972e-06, "loss": 1.1283, "step": 12045 }, { "epoch": 1.2817620770376676, "grad_norm": 11.328125, "learning_rate": 9.589752700397954e-06, "loss": 1.0632, "step": 12046 }, { "epoch": 1.2818684826558842, "grad_norm": 13.859375, "learning_rate": 9.589717168845935e-06, "loss": 1.0314, "step": 12047 }, { "epoch": 1.2819748882741009, "grad_norm": 9.8984375, "learning_rate": 9.589681637293919e-06, "loss": 1.1707, "step": 12048 }, { "epoch": 1.2820812938923174, "grad_norm": 10.96875, "learning_rate": 9.5896461057419e-06, "loss": 0.9204, "step": 12049 }, { "epoch": 1.2821876995105341, "grad_norm": 10.859375, "learning_rate": 9.589610574189882e-06, "loss": 1.1145, "step": 12050 }, { "epoch": 1.2822941051287509, "grad_norm": 17.109375, "learning_rate": 9.589575042637864e-06, "loss": 1.0092, "step": 12051 }, { "epoch": 1.2824005107469674, "grad_norm": 11.9296875, "learning_rate": 9.589539511085845e-06, "loss": 0.8652, "step": 12052 }, { "epoch": 1.282506916365184, "grad_norm": 13.453125, "learning_rate": 9.589503979533827e-06, "loss": 1.1464, "step": 12053 }, { "epoch": 1.2826133219834008, "grad_norm": 12.921875, "learning_rate": 9.589468447981808e-06, "loss": 0.9879, "step": 12054 }, { "epoch": 1.2827197276016173, "grad_norm": 17.71875, "learning_rate": 9.58943291642979e-06, "loss": 1.1093, "step": 12055 }, { "epoch": 1.282826133219834, "grad_norm": 12.265625, "learning_rate": 9.589397384877772e-06, "loss": 0.8151, "step": 12056 }, { "epoch": 1.2829325388380506, "grad_norm": 11.6796875, "learning_rate": 9.589361853325753e-06, "loss": 0.9147, "step": 12057 }, { "epoch": 1.2830389444562673, "grad_norm": 15.109375, "learning_rate": 9.589326321773735e-06, "loss": 0.9594, "step": 12058 }, { "epoch": 1.2831453500744838, "grad_norm": 13.296875, "learning_rate": 9.589290790221718e-06, "loss": 1.3156, "step": 12059 }, { "epoch": 1.2832517556927006, "grad_norm": 14.71875, "learning_rate": 9.5892552586697e-06, "loss": 1.2194, "step": 12060 }, { "epoch": 1.2833581613109173, "grad_norm": 11.0703125, "learning_rate": 9.589219727117681e-06, "loss": 0.9936, "step": 12061 }, { "epoch": 1.2834645669291338, "grad_norm": 20.875, "learning_rate": 9.589184195565663e-06, "loss": 1.0577, "step": 12062 }, { "epoch": 1.2835709725473505, "grad_norm": 14.6640625, "learning_rate": 9.589148664013645e-06, "loss": 1.0546, "step": 12063 }, { "epoch": 1.2836773781655673, "grad_norm": 11.5625, "learning_rate": 9.589113132461628e-06, "loss": 1.0199, "step": 12064 }, { "epoch": 1.2837837837837838, "grad_norm": 10.3515625, "learning_rate": 9.58907760090961e-06, "loss": 0.5799, "step": 12065 }, { "epoch": 1.2838901894020005, "grad_norm": 13.9375, "learning_rate": 9.589042069357591e-06, "loss": 0.7872, "step": 12066 }, { "epoch": 1.283996595020217, "grad_norm": 11.8671875, "learning_rate": 9.589006537805573e-06, "loss": 0.9995, "step": 12067 }, { "epoch": 1.2841030006384337, "grad_norm": 13.59375, "learning_rate": 9.588971006253554e-06, "loss": 1.1908, "step": 12068 }, { "epoch": 1.2842094062566503, "grad_norm": 11.875, "learning_rate": 9.588935474701536e-06, "loss": 0.8774, "step": 12069 }, { "epoch": 1.284315811874867, "grad_norm": 9.78125, "learning_rate": 9.588899943149518e-06, "loss": 0.8604, "step": 12070 }, { "epoch": 1.2844222174930837, "grad_norm": 15.4296875, "learning_rate": 9.588864411597499e-06, "loss": 0.9848, "step": 12071 }, { "epoch": 1.2845286231113002, "grad_norm": 12.4765625, "learning_rate": 9.58882888004548e-06, "loss": 1.2038, "step": 12072 }, { "epoch": 1.284635028729517, "grad_norm": 16.140625, "learning_rate": 9.588793348493462e-06, "loss": 1.1871, "step": 12073 }, { "epoch": 1.2847414343477337, "grad_norm": 10.3359375, "learning_rate": 9.588757816941446e-06, "loss": 0.942, "step": 12074 }, { "epoch": 1.2848478399659502, "grad_norm": 9.7734375, "learning_rate": 9.588722285389427e-06, "loss": 0.9968, "step": 12075 }, { "epoch": 1.2849542455841667, "grad_norm": 13.515625, "learning_rate": 9.588686753837409e-06, "loss": 1.0582, "step": 12076 }, { "epoch": 1.2850606512023834, "grad_norm": 10.2734375, "learning_rate": 9.58865122228539e-06, "loss": 0.9502, "step": 12077 }, { "epoch": 1.2851670568206002, "grad_norm": 14.1875, "learning_rate": 9.588615690733372e-06, "loss": 1.1727, "step": 12078 }, { "epoch": 1.2852734624388167, "grad_norm": 11.71875, "learning_rate": 9.588580159181354e-06, "loss": 0.9665, "step": 12079 }, { "epoch": 1.2853798680570334, "grad_norm": 13.484375, "learning_rate": 9.588544627629335e-06, "loss": 0.9268, "step": 12080 }, { "epoch": 1.2854862736752501, "grad_norm": 15.046875, "learning_rate": 9.588509096077317e-06, "loss": 1.0077, "step": 12081 }, { "epoch": 1.2855926792934667, "grad_norm": 15.96875, "learning_rate": 9.588473564525298e-06, "loss": 1.0878, "step": 12082 }, { "epoch": 1.2856990849116834, "grad_norm": 10.4296875, "learning_rate": 9.58843803297328e-06, "loss": 0.9003, "step": 12083 }, { "epoch": 1.2858054905299, "grad_norm": 13.3203125, "learning_rate": 9.588402501421263e-06, "loss": 0.9495, "step": 12084 }, { "epoch": 1.2859118961481166, "grad_norm": 12.984375, "learning_rate": 9.588366969869245e-06, "loss": 1.1048, "step": 12085 }, { "epoch": 1.2860183017663331, "grad_norm": 14.796875, "learning_rate": 9.588331438317227e-06, "loss": 1.2401, "step": 12086 }, { "epoch": 1.2861247073845499, "grad_norm": 13.0546875, "learning_rate": 9.588295906765208e-06, "loss": 1.1573, "step": 12087 }, { "epoch": 1.2862311130027666, "grad_norm": 11.4296875, "learning_rate": 9.58826037521319e-06, "loss": 0.9332, "step": 12088 }, { "epoch": 1.2863375186209831, "grad_norm": 10.9453125, "learning_rate": 9.588224843661173e-06, "loss": 1.3419, "step": 12089 }, { "epoch": 1.2864439242391998, "grad_norm": 11.2890625, "learning_rate": 9.588189312109155e-06, "loss": 1.2654, "step": 12090 }, { "epoch": 1.2865503298574166, "grad_norm": 12.34375, "learning_rate": 9.588153780557136e-06, "loss": 0.9051, "step": 12091 }, { "epoch": 1.286656735475633, "grad_norm": 15.875, "learning_rate": 9.588118249005118e-06, "loss": 1.0392, "step": 12092 }, { "epoch": 1.2867631410938498, "grad_norm": 19.890625, "learning_rate": 9.5880827174531e-06, "loss": 1.2634, "step": 12093 }, { "epoch": 1.2868695467120663, "grad_norm": 11.453125, "learning_rate": 9.588047185901081e-06, "loss": 0.7934, "step": 12094 }, { "epoch": 1.286975952330283, "grad_norm": 13.3125, "learning_rate": 9.588011654349063e-06, "loss": 1.3637, "step": 12095 }, { "epoch": 1.2870823579484996, "grad_norm": 12.796875, "learning_rate": 9.587976122797044e-06, "loss": 1.0901, "step": 12096 }, { "epoch": 1.2871887635667163, "grad_norm": 13.4765625, "learning_rate": 9.587940591245026e-06, "loss": 1.0888, "step": 12097 }, { "epoch": 1.287295169184933, "grad_norm": 15.9921875, "learning_rate": 9.587905059693008e-06, "loss": 1.2975, "step": 12098 }, { "epoch": 1.2874015748031495, "grad_norm": 10.9765625, "learning_rate": 9.58786952814099e-06, "loss": 0.9245, "step": 12099 }, { "epoch": 1.2875079804213663, "grad_norm": 13.765625, "learning_rate": 9.587833996588972e-06, "loss": 1.2164, "step": 12100 }, { "epoch": 1.287614386039583, "grad_norm": 11.8515625, "learning_rate": 9.587798465036954e-06, "loss": 0.9795, "step": 12101 }, { "epoch": 1.2877207916577995, "grad_norm": 17.59375, "learning_rate": 9.587762933484936e-06, "loss": 1.092, "step": 12102 }, { "epoch": 1.2878271972760162, "grad_norm": 27.703125, "learning_rate": 9.587727401932917e-06, "loss": 1.4716, "step": 12103 }, { "epoch": 1.2879336028942328, "grad_norm": 11.5234375, "learning_rate": 9.587691870380899e-06, "loss": 1.1789, "step": 12104 }, { "epoch": 1.2880400085124495, "grad_norm": 11.0078125, "learning_rate": 9.58765633882888e-06, "loss": 0.9584, "step": 12105 }, { "epoch": 1.288146414130666, "grad_norm": 10.3046875, "learning_rate": 9.587620807276862e-06, "loss": 1.1387, "step": 12106 }, { "epoch": 1.2882528197488827, "grad_norm": 11.5859375, "learning_rate": 9.587585275724844e-06, "loss": 1.0984, "step": 12107 }, { "epoch": 1.2883592253670995, "grad_norm": 18.296875, "learning_rate": 9.587549744172825e-06, "loss": 1.6204, "step": 12108 }, { "epoch": 1.288465630985316, "grad_norm": 12.0625, "learning_rate": 9.587514212620807e-06, "loss": 1.2607, "step": 12109 }, { "epoch": 1.2885720366035327, "grad_norm": 16.421875, "learning_rate": 9.58747868106879e-06, "loss": 1.1544, "step": 12110 }, { "epoch": 1.2886784422217494, "grad_norm": 12.15625, "learning_rate": 9.587443149516772e-06, "loss": 1.0781, "step": 12111 }, { "epoch": 1.288784847839966, "grad_norm": 12.296875, "learning_rate": 9.587407617964753e-06, "loss": 1.0217, "step": 12112 }, { "epoch": 1.2888912534581827, "grad_norm": 16.1875, "learning_rate": 9.587372086412735e-06, "loss": 1.0028, "step": 12113 }, { "epoch": 1.2889976590763992, "grad_norm": 17.140625, "learning_rate": 9.587336554860717e-06, "loss": 1.2198, "step": 12114 }, { "epoch": 1.289104064694616, "grad_norm": 11.6796875, "learning_rate": 9.5873010233087e-06, "loss": 0.9057, "step": 12115 }, { "epoch": 1.2892104703128324, "grad_norm": 12.5234375, "learning_rate": 9.587265491756682e-06, "loss": 0.8513, "step": 12116 }, { "epoch": 1.2893168759310492, "grad_norm": 13.3984375, "learning_rate": 9.587229960204663e-06, "loss": 1.1365, "step": 12117 }, { "epoch": 1.2894232815492659, "grad_norm": 14.3515625, "learning_rate": 9.587194428652645e-06, "loss": 0.9669, "step": 12118 }, { "epoch": 1.2895296871674824, "grad_norm": 15.6171875, "learning_rate": 9.587158897100626e-06, "loss": 1.2846, "step": 12119 }, { "epoch": 1.2896360927856991, "grad_norm": 11.3515625, "learning_rate": 9.587123365548608e-06, "loss": 0.6792, "step": 12120 }, { "epoch": 1.2897424984039159, "grad_norm": 10.9609375, "learning_rate": 9.58708783399659e-06, "loss": 0.8919, "step": 12121 }, { "epoch": 1.2898489040221324, "grad_norm": 14.0078125, "learning_rate": 9.587052302444571e-06, "loss": 1.4738, "step": 12122 }, { "epoch": 1.2899553096403489, "grad_norm": 9.78125, "learning_rate": 9.587016770892553e-06, "loss": 1.0903, "step": 12123 }, { "epoch": 1.2900617152585656, "grad_norm": 10.078125, "learning_rate": 9.586981239340534e-06, "loss": 1.1055, "step": 12124 }, { "epoch": 1.2901681208767823, "grad_norm": 12.6953125, "learning_rate": 9.586945707788518e-06, "loss": 1.1927, "step": 12125 }, { "epoch": 1.2902745264949989, "grad_norm": 18.484375, "learning_rate": 9.5869101762365e-06, "loss": 1.5857, "step": 12126 }, { "epoch": 1.2903809321132156, "grad_norm": 13.1484375, "learning_rate": 9.586874644684481e-06, "loss": 1.3819, "step": 12127 }, { "epoch": 1.2904873377314323, "grad_norm": 9.75, "learning_rate": 9.586839113132463e-06, "loss": 1.0583, "step": 12128 }, { "epoch": 1.2905937433496488, "grad_norm": 11.5234375, "learning_rate": 9.586803581580444e-06, "loss": 1.0629, "step": 12129 }, { "epoch": 1.2907001489678656, "grad_norm": 13.484375, "learning_rate": 9.586768050028426e-06, "loss": 0.9865, "step": 12130 }, { "epoch": 1.290806554586082, "grad_norm": 14.4765625, "learning_rate": 9.586732518476407e-06, "loss": 1.1162, "step": 12131 }, { "epoch": 1.2909129602042988, "grad_norm": 10.9453125, "learning_rate": 9.586696986924389e-06, "loss": 1.2003, "step": 12132 }, { "epoch": 1.2910193658225153, "grad_norm": 10.8671875, "learning_rate": 9.58666145537237e-06, "loss": 0.9979, "step": 12133 }, { "epoch": 1.291125771440732, "grad_norm": 18.171875, "learning_rate": 9.586625923820352e-06, "loss": 1.0371, "step": 12134 }, { "epoch": 1.2912321770589488, "grad_norm": 12.6875, "learning_rate": 9.586590392268334e-06, "loss": 0.9319, "step": 12135 }, { "epoch": 1.2913385826771653, "grad_norm": 10.4453125, "learning_rate": 9.586554860716317e-06, "loss": 0.9661, "step": 12136 }, { "epoch": 1.291444988295382, "grad_norm": 10.8984375, "learning_rate": 9.586519329164299e-06, "loss": 1.0574, "step": 12137 }, { "epoch": 1.2915513939135987, "grad_norm": 14.3046875, "learning_rate": 9.58648379761228e-06, "loss": 0.9201, "step": 12138 }, { "epoch": 1.2916577995318153, "grad_norm": 13.21875, "learning_rate": 9.586448266060262e-06, "loss": 0.9999, "step": 12139 }, { "epoch": 1.291764205150032, "grad_norm": 10.0, "learning_rate": 9.586412734508244e-06, "loss": 1.0654, "step": 12140 }, { "epoch": 1.2918706107682485, "grad_norm": 12.328125, "learning_rate": 9.586377202956227e-06, "loss": 1.0515, "step": 12141 }, { "epoch": 1.2919770163864652, "grad_norm": 11.3515625, "learning_rate": 9.586341671404208e-06, "loss": 1.053, "step": 12142 }, { "epoch": 1.2920834220046817, "grad_norm": 9.609375, "learning_rate": 9.58630613985219e-06, "loss": 0.9808, "step": 12143 }, { "epoch": 1.2921898276228985, "grad_norm": 12.8671875, "learning_rate": 9.586270608300172e-06, "loss": 0.9107, "step": 12144 }, { "epoch": 1.2922962332411152, "grad_norm": 13.03125, "learning_rate": 9.586235076748153e-06, "loss": 1.3111, "step": 12145 }, { "epoch": 1.2924026388593317, "grad_norm": 10.796875, "learning_rate": 9.586199545196135e-06, "loss": 0.9926, "step": 12146 }, { "epoch": 1.2925090444775484, "grad_norm": 19.0, "learning_rate": 9.586164013644116e-06, "loss": 0.9741, "step": 12147 }, { "epoch": 1.2926154500957652, "grad_norm": 11.6171875, "learning_rate": 9.586128482092098e-06, "loss": 0.9503, "step": 12148 }, { "epoch": 1.2927218557139817, "grad_norm": 12.296875, "learning_rate": 9.58609295054008e-06, "loss": 0.9406, "step": 12149 }, { "epoch": 1.2928282613321984, "grad_norm": 11.5078125, "learning_rate": 9.586057418988061e-06, "loss": 1.2206, "step": 12150 }, { "epoch": 1.292934666950415, "grad_norm": 12.9296875, "learning_rate": 9.586021887436045e-06, "loss": 0.8543, "step": 12151 }, { "epoch": 1.2930410725686317, "grad_norm": 10.96875, "learning_rate": 9.585986355884026e-06, "loss": 0.8226, "step": 12152 }, { "epoch": 1.2931474781868482, "grad_norm": 14.375, "learning_rate": 9.585950824332008e-06, "loss": 1.0833, "step": 12153 }, { "epoch": 1.293253883805065, "grad_norm": 13.328125, "learning_rate": 9.58591529277999e-06, "loss": 1.173, "step": 12154 }, { "epoch": 1.2933602894232816, "grad_norm": 14.0703125, "learning_rate": 9.585879761227971e-06, "loss": 1.2897, "step": 12155 }, { "epoch": 1.2934666950414981, "grad_norm": 10.5390625, "learning_rate": 9.585844229675954e-06, "loss": 1.0226, "step": 12156 }, { "epoch": 1.2935731006597149, "grad_norm": 14.8203125, "learning_rate": 9.585808698123936e-06, "loss": 1.1226, "step": 12157 }, { "epoch": 1.2936795062779316, "grad_norm": 13.4140625, "learning_rate": 9.585773166571918e-06, "loss": 1.1344, "step": 12158 }, { "epoch": 1.293785911896148, "grad_norm": 14.3671875, "learning_rate": 9.585737635019897e-06, "loss": 1.383, "step": 12159 }, { "epoch": 1.2938923175143646, "grad_norm": 13.828125, "learning_rate": 9.585702103467879e-06, "loss": 1.3264, "step": 12160 }, { "epoch": 1.2939987231325814, "grad_norm": 16.484375, "learning_rate": 9.585666571915862e-06, "loss": 1.3064, "step": 12161 }, { "epoch": 1.294105128750798, "grad_norm": 12.7265625, "learning_rate": 9.585631040363844e-06, "loss": 1.0195, "step": 12162 }, { "epoch": 1.2942115343690146, "grad_norm": 14.3125, "learning_rate": 9.585595508811826e-06, "loss": 1.2054, "step": 12163 }, { "epoch": 1.2943179399872313, "grad_norm": 12.9375, "learning_rate": 9.585559977259807e-06, "loss": 0.8127, "step": 12164 }, { "epoch": 1.294424345605448, "grad_norm": 10.1640625, "learning_rate": 9.585524445707789e-06, "loss": 1.005, "step": 12165 }, { "epoch": 1.2945307512236646, "grad_norm": 10.359375, "learning_rate": 9.585488914155772e-06, "loss": 0.9457, "step": 12166 }, { "epoch": 1.2946371568418813, "grad_norm": 14.21875, "learning_rate": 9.585453382603754e-06, "loss": 1.256, "step": 12167 }, { "epoch": 1.294743562460098, "grad_norm": 16.03125, "learning_rate": 9.585417851051735e-06, "loss": 1.1614, "step": 12168 }, { "epoch": 1.2948499680783145, "grad_norm": 13.8359375, "learning_rate": 9.585382319499717e-06, "loss": 1.4314, "step": 12169 }, { "epoch": 1.294956373696531, "grad_norm": 12.921875, "learning_rate": 9.585346787947699e-06, "loss": 1.0043, "step": 12170 }, { "epoch": 1.2950627793147478, "grad_norm": 11.78125, "learning_rate": 9.58531125639568e-06, "loss": 1.0397, "step": 12171 }, { "epoch": 1.2951691849329645, "grad_norm": 11.015625, "learning_rate": 9.585275724843662e-06, "loss": 1.0043, "step": 12172 }, { "epoch": 1.295275590551181, "grad_norm": 14.625, "learning_rate": 9.585240193291643e-06, "loss": 1.2763, "step": 12173 }, { "epoch": 1.2953819961693978, "grad_norm": 10.0625, "learning_rate": 9.585204661739625e-06, "loss": 1.1916, "step": 12174 }, { "epoch": 1.2954884017876145, "grad_norm": 12.828125, "learning_rate": 9.585169130187607e-06, "loss": 1.5529, "step": 12175 }, { "epoch": 1.295594807405831, "grad_norm": 12.484375, "learning_rate": 9.585133598635588e-06, "loss": 1.3415, "step": 12176 }, { "epoch": 1.2957012130240477, "grad_norm": 19.046875, "learning_rate": 9.585098067083571e-06, "loss": 1.1888, "step": 12177 }, { "epoch": 1.2958076186422642, "grad_norm": 16.5, "learning_rate": 9.585062535531553e-06, "loss": 1.0517, "step": 12178 }, { "epoch": 1.295914024260481, "grad_norm": 11.71875, "learning_rate": 9.585027003979535e-06, "loss": 1.0261, "step": 12179 }, { "epoch": 1.2960204298786975, "grad_norm": 25.90625, "learning_rate": 9.584991472427516e-06, "loss": 1.2325, "step": 12180 }, { "epoch": 1.2961268354969142, "grad_norm": 17.921875, "learning_rate": 9.584955940875498e-06, "loss": 1.193, "step": 12181 }, { "epoch": 1.296233241115131, "grad_norm": 15.15625, "learning_rate": 9.584920409323481e-06, "loss": 1.2521, "step": 12182 }, { "epoch": 1.2963396467333475, "grad_norm": 11.140625, "learning_rate": 9.584884877771463e-06, "loss": 1.0272, "step": 12183 }, { "epoch": 1.2964460523515642, "grad_norm": 13.09375, "learning_rate": 9.584849346219444e-06, "loss": 1.128, "step": 12184 }, { "epoch": 1.296552457969781, "grad_norm": 13.859375, "learning_rate": 9.584813814667426e-06, "loss": 1.3296, "step": 12185 }, { "epoch": 1.2966588635879974, "grad_norm": 12.4765625, "learning_rate": 9.584778283115408e-06, "loss": 1.0388, "step": 12186 }, { "epoch": 1.2967652692062142, "grad_norm": 17.40625, "learning_rate": 9.58474275156339e-06, "loss": 0.8296, "step": 12187 }, { "epoch": 1.2968716748244307, "grad_norm": 13.484375, "learning_rate": 9.58470722001137e-06, "loss": 0.9471, "step": 12188 }, { "epoch": 1.2969780804426474, "grad_norm": 20.15625, "learning_rate": 9.584671688459352e-06, "loss": 1.4492, "step": 12189 }, { "epoch": 1.297084486060864, "grad_norm": 13.1953125, "learning_rate": 9.584636156907334e-06, "loss": 1.2132, "step": 12190 }, { "epoch": 1.2971908916790806, "grad_norm": 12.40625, "learning_rate": 9.584600625355316e-06, "loss": 0.9806, "step": 12191 }, { "epoch": 1.2972972972972974, "grad_norm": 11.7265625, "learning_rate": 9.584565093803299e-06, "loss": 1.2524, "step": 12192 }, { "epoch": 1.2974037029155139, "grad_norm": 13.1796875, "learning_rate": 9.58452956225128e-06, "loss": 1.2849, "step": 12193 }, { "epoch": 1.2975101085337306, "grad_norm": 21.34375, "learning_rate": 9.584494030699262e-06, "loss": 1.16, "step": 12194 }, { "epoch": 1.2976165141519473, "grad_norm": 14.6953125, "learning_rate": 9.584458499147244e-06, "loss": 1.2096, "step": 12195 }, { "epoch": 1.2977229197701639, "grad_norm": 9.6484375, "learning_rate": 9.584422967595225e-06, "loss": 0.9513, "step": 12196 }, { "epoch": 1.2978293253883806, "grad_norm": 19.46875, "learning_rate": 9.584387436043207e-06, "loss": 0.884, "step": 12197 }, { "epoch": 1.297935731006597, "grad_norm": 17.421875, "learning_rate": 9.584351904491189e-06, "loss": 1.1379, "step": 12198 }, { "epoch": 1.2980421366248138, "grad_norm": 10.484375, "learning_rate": 9.58431637293917e-06, "loss": 1.1304, "step": 12199 }, { "epoch": 1.2981485422430303, "grad_norm": 9.125, "learning_rate": 9.584280841387152e-06, "loss": 0.99, "step": 12200 }, { "epoch": 1.298254947861247, "grad_norm": 9.28125, "learning_rate": 9.584245309835133e-06, "loss": 1.0499, "step": 12201 }, { "epoch": 1.2983613534794638, "grad_norm": 11.734375, "learning_rate": 9.584209778283117e-06, "loss": 1.2571, "step": 12202 }, { "epoch": 1.2984677590976803, "grad_norm": 12.921875, "learning_rate": 9.584174246731098e-06, "loss": 1.2857, "step": 12203 }, { "epoch": 1.298574164715897, "grad_norm": 11.2734375, "learning_rate": 9.58413871517908e-06, "loss": 1.0791, "step": 12204 }, { "epoch": 1.2986805703341138, "grad_norm": 12.9296875, "learning_rate": 9.584103183627062e-06, "loss": 0.9784, "step": 12205 }, { "epoch": 1.2987869759523303, "grad_norm": 12.734375, "learning_rate": 9.584067652075043e-06, "loss": 1.2665, "step": 12206 }, { "epoch": 1.2988933815705468, "grad_norm": 15.109375, "learning_rate": 9.584032120523026e-06, "loss": 1.1607, "step": 12207 }, { "epoch": 1.2989997871887635, "grad_norm": 14.0859375, "learning_rate": 9.583996588971008e-06, "loss": 1.0137, "step": 12208 }, { "epoch": 1.2991061928069803, "grad_norm": 12.109375, "learning_rate": 9.58396105741899e-06, "loss": 1.0647, "step": 12209 }, { "epoch": 1.2992125984251968, "grad_norm": 11.8046875, "learning_rate": 9.583925525866971e-06, "loss": 0.9299, "step": 12210 }, { "epoch": 1.2993190040434135, "grad_norm": 15.3984375, "learning_rate": 9.583889994314953e-06, "loss": 1.0365, "step": 12211 }, { "epoch": 1.2994254096616302, "grad_norm": 13.4765625, "learning_rate": 9.583854462762934e-06, "loss": 1.0212, "step": 12212 }, { "epoch": 1.2995318152798467, "grad_norm": 10.4609375, "learning_rate": 9.583818931210916e-06, "loss": 1.2453, "step": 12213 }, { "epoch": 1.2996382208980635, "grad_norm": 12.6015625, "learning_rate": 9.583783399658898e-06, "loss": 0.8991, "step": 12214 }, { "epoch": 1.29974462651628, "grad_norm": 12.328125, "learning_rate": 9.58374786810688e-06, "loss": 0.9428, "step": 12215 }, { "epoch": 1.2998510321344967, "grad_norm": 11.578125, "learning_rate": 9.583712336554861e-06, "loss": 0.8868, "step": 12216 }, { "epoch": 1.2999574377527132, "grad_norm": 14.671875, "learning_rate": 9.583676805002843e-06, "loss": 1.1167, "step": 12217 }, { "epoch": 1.30006384337093, "grad_norm": 13.7421875, "learning_rate": 9.583641273450826e-06, "loss": 1.0562, "step": 12218 }, { "epoch": 1.3001702489891467, "grad_norm": 11.2421875, "learning_rate": 9.583605741898807e-06, "loss": 1.0189, "step": 12219 }, { "epoch": 1.3002766546073632, "grad_norm": 11.6328125, "learning_rate": 9.583570210346789e-06, "loss": 0.9089, "step": 12220 }, { "epoch": 1.30038306022558, "grad_norm": 13.40625, "learning_rate": 9.58353467879477e-06, "loss": 1.1148, "step": 12221 }, { "epoch": 1.3004894658437967, "grad_norm": 12.265625, "learning_rate": 9.583499147242752e-06, "loss": 1.1593, "step": 12222 }, { "epoch": 1.3005958714620132, "grad_norm": 12.8984375, "learning_rate": 9.583463615690734e-06, "loss": 1.6432, "step": 12223 }, { "epoch": 1.30070227708023, "grad_norm": 13.125, "learning_rate": 9.583428084138715e-06, "loss": 1.0021, "step": 12224 }, { "epoch": 1.3008086826984464, "grad_norm": 9.921875, "learning_rate": 9.583392552586697e-06, "loss": 1.0851, "step": 12225 }, { "epoch": 1.3009150883166631, "grad_norm": 12.8125, "learning_rate": 9.583357021034679e-06, "loss": 0.837, "step": 12226 }, { "epoch": 1.3010214939348796, "grad_norm": 9.8359375, "learning_rate": 9.58332148948266e-06, "loss": 0.94, "step": 12227 }, { "epoch": 1.3011278995530964, "grad_norm": 9.328125, "learning_rate": 9.583285957930644e-06, "loss": 1.0157, "step": 12228 }, { "epoch": 1.301234305171313, "grad_norm": 10.4765625, "learning_rate": 9.583250426378625e-06, "loss": 1.0041, "step": 12229 }, { "epoch": 1.3013407107895296, "grad_norm": 14.28125, "learning_rate": 9.583214894826607e-06, "loss": 1.2233, "step": 12230 }, { "epoch": 1.3014471164077464, "grad_norm": 10.9375, "learning_rate": 9.583179363274588e-06, "loss": 0.8969, "step": 12231 }, { "epoch": 1.301553522025963, "grad_norm": 12.03125, "learning_rate": 9.58314383172257e-06, "loss": 1.0118, "step": 12232 }, { "epoch": 1.3016599276441796, "grad_norm": 13.6171875, "learning_rate": 9.583108300170553e-06, "loss": 0.9201, "step": 12233 }, { "epoch": 1.3017663332623963, "grad_norm": 13.6953125, "learning_rate": 9.583072768618535e-06, "loss": 1.2519, "step": 12234 }, { "epoch": 1.3018727388806128, "grad_norm": 10.015625, "learning_rate": 9.583037237066517e-06, "loss": 0.974, "step": 12235 }, { "epoch": 1.3019791444988296, "grad_norm": 10.75, "learning_rate": 9.583001705514498e-06, "loss": 1.0732, "step": 12236 }, { "epoch": 1.302085550117046, "grad_norm": 28.875, "learning_rate": 9.58296617396248e-06, "loss": 1.0312, "step": 12237 }, { "epoch": 1.3021919557352628, "grad_norm": 11.2421875, "learning_rate": 9.582930642410461e-06, "loss": 0.8595, "step": 12238 }, { "epoch": 1.3022983613534795, "grad_norm": 16.75, "learning_rate": 9.582895110858443e-06, "loss": 1.1975, "step": 12239 }, { "epoch": 1.302404766971696, "grad_norm": 22.515625, "learning_rate": 9.582859579306425e-06, "loss": 1.008, "step": 12240 }, { "epoch": 1.3025111725899128, "grad_norm": 11.765625, "learning_rate": 9.582824047754406e-06, "loss": 0.9739, "step": 12241 }, { "epoch": 1.3026175782081295, "grad_norm": 15.3515625, "learning_rate": 9.582788516202388e-06, "loss": 1.0844, "step": 12242 }, { "epoch": 1.302723983826346, "grad_norm": 12.8515625, "learning_rate": 9.582752984650371e-06, "loss": 0.9249, "step": 12243 }, { "epoch": 1.3028303894445628, "grad_norm": 14.4375, "learning_rate": 9.582717453098353e-06, "loss": 1.3423, "step": 12244 }, { "epoch": 1.3029367950627793, "grad_norm": 15.6640625, "learning_rate": 9.582681921546334e-06, "loss": 1.1085, "step": 12245 }, { "epoch": 1.303043200680996, "grad_norm": 16.5625, "learning_rate": 9.582646389994316e-06, "loss": 1.1976, "step": 12246 }, { "epoch": 1.3031496062992125, "grad_norm": 14.6875, "learning_rate": 9.582610858442297e-06, "loss": 1.0371, "step": 12247 }, { "epoch": 1.3032560119174292, "grad_norm": 17.59375, "learning_rate": 9.582575326890279e-06, "loss": 1.0608, "step": 12248 }, { "epoch": 1.303362417535646, "grad_norm": 14.40625, "learning_rate": 9.58253979533826e-06, "loss": 1.2505, "step": 12249 }, { "epoch": 1.3034688231538625, "grad_norm": 18.609375, "learning_rate": 9.582504263786242e-06, "loss": 1.3221, "step": 12250 }, { "epoch": 1.3035752287720792, "grad_norm": 10.7734375, "learning_rate": 9.582468732234224e-06, "loss": 1.1331, "step": 12251 }, { "epoch": 1.303681634390296, "grad_norm": 13.84375, "learning_rate": 9.582433200682206e-06, "loss": 0.8691, "step": 12252 }, { "epoch": 1.3037880400085124, "grad_norm": 11.4609375, "learning_rate": 9.582397669130187e-06, "loss": 1.0769, "step": 12253 }, { "epoch": 1.303894445626729, "grad_norm": 14.546875, "learning_rate": 9.58236213757817e-06, "loss": 1.0434, "step": 12254 }, { "epoch": 1.3040008512449457, "grad_norm": 12.171875, "learning_rate": 9.582326606026152e-06, "loss": 1.1815, "step": 12255 }, { "epoch": 1.3041072568631624, "grad_norm": 15.7890625, "learning_rate": 9.582291074474134e-06, "loss": 1.1123, "step": 12256 }, { "epoch": 1.304213662481379, "grad_norm": 14.640625, "learning_rate": 9.582255542922115e-06, "loss": 1.2712, "step": 12257 }, { "epoch": 1.3043200680995957, "grad_norm": 13.359375, "learning_rate": 9.582220011370097e-06, "loss": 1.138, "step": 12258 }, { "epoch": 1.3044264737178124, "grad_norm": 14.9375, "learning_rate": 9.58218447981808e-06, "loss": 1.0031, "step": 12259 }, { "epoch": 1.304532879336029, "grad_norm": 17.359375, "learning_rate": 9.582148948266062e-06, "loss": 1.0852, "step": 12260 }, { "epoch": 1.3046392849542456, "grad_norm": 11.3046875, "learning_rate": 9.582113416714043e-06, "loss": 0.8387, "step": 12261 }, { "epoch": 1.3047456905724621, "grad_norm": 11.65625, "learning_rate": 9.582077885162025e-06, "loss": 1.0256, "step": 12262 }, { "epoch": 1.3048520961906789, "grad_norm": 11.84375, "learning_rate": 9.582042353610007e-06, "loss": 0.9083, "step": 12263 }, { "epoch": 1.3049585018088954, "grad_norm": 12.9765625, "learning_rate": 9.582006822057988e-06, "loss": 1.1007, "step": 12264 }, { "epoch": 1.3050649074271121, "grad_norm": 16.390625, "learning_rate": 9.58197129050597e-06, "loss": 1.2237, "step": 12265 }, { "epoch": 1.3051713130453289, "grad_norm": 15.515625, "learning_rate": 9.581935758953951e-06, "loss": 1.0358, "step": 12266 }, { "epoch": 1.3052777186635454, "grad_norm": 11.9453125, "learning_rate": 9.581900227401933e-06, "loss": 1.0707, "step": 12267 }, { "epoch": 1.305384124281762, "grad_norm": 14.984375, "learning_rate": 9.581864695849915e-06, "loss": 1.3306, "step": 12268 }, { "epoch": 1.3054905298999788, "grad_norm": 15.6640625, "learning_rate": 9.581829164297898e-06, "loss": 1.0972, "step": 12269 }, { "epoch": 1.3055969355181953, "grad_norm": 11.6171875, "learning_rate": 9.58179363274588e-06, "loss": 0.9127, "step": 12270 }, { "epoch": 1.305703341136412, "grad_norm": 14.15625, "learning_rate": 9.581758101193861e-06, "loss": 0.9827, "step": 12271 }, { "epoch": 1.3058097467546286, "grad_norm": 12.65625, "learning_rate": 9.581722569641843e-06, "loss": 0.9863, "step": 12272 }, { "epoch": 1.3059161523728453, "grad_norm": 10.8203125, "learning_rate": 9.581687038089824e-06, "loss": 0.8151, "step": 12273 }, { "epoch": 1.3060225579910618, "grad_norm": 15.59375, "learning_rate": 9.581651506537808e-06, "loss": 0.9021, "step": 12274 }, { "epoch": 1.3061289636092785, "grad_norm": 10.875, "learning_rate": 9.58161597498579e-06, "loss": 0.9299, "step": 12275 }, { "epoch": 1.3062353692274953, "grad_norm": 30.078125, "learning_rate": 9.58158044343377e-06, "loss": 1.4769, "step": 12276 }, { "epoch": 1.3063417748457118, "grad_norm": 14.3984375, "learning_rate": 9.581544911881752e-06, "loss": 1.1143, "step": 12277 }, { "epoch": 1.3064481804639285, "grad_norm": 13.9609375, "learning_rate": 9.581509380329734e-06, "loss": 1.2291, "step": 12278 }, { "epoch": 1.3065545860821453, "grad_norm": 14.3984375, "learning_rate": 9.581473848777716e-06, "loss": 0.9218, "step": 12279 }, { "epoch": 1.3066609917003618, "grad_norm": 10.5859375, "learning_rate": 9.581438317225697e-06, "loss": 1.1019, "step": 12280 }, { "epoch": 1.3067673973185785, "grad_norm": 12.4375, "learning_rate": 9.581402785673679e-06, "loss": 1.1, "step": 12281 }, { "epoch": 1.306873802936795, "grad_norm": 13.1796875, "learning_rate": 9.58136725412166e-06, "loss": 1.22, "step": 12282 }, { "epoch": 1.3069802085550117, "grad_norm": 11.90625, "learning_rate": 9.581331722569642e-06, "loss": 1.0516, "step": 12283 }, { "epoch": 1.3070866141732282, "grad_norm": 16.546875, "learning_rate": 9.581296191017625e-06, "loss": 1.1829, "step": 12284 }, { "epoch": 1.307193019791445, "grad_norm": 11.7421875, "learning_rate": 9.581260659465607e-06, "loss": 0.9439, "step": 12285 }, { "epoch": 1.3072994254096617, "grad_norm": 11.7109375, "learning_rate": 9.581225127913589e-06, "loss": 1.2224, "step": 12286 }, { "epoch": 1.3074058310278782, "grad_norm": 13.1796875, "learning_rate": 9.58118959636157e-06, "loss": 1.2459, "step": 12287 }, { "epoch": 1.307512236646095, "grad_norm": 11.0859375, "learning_rate": 9.581154064809552e-06, "loss": 1.1375, "step": 12288 }, { "epoch": 1.3076186422643117, "grad_norm": 14.2578125, "learning_rate": 9.581118533257533e-06, "loss": 1.3765, "step": 12289 }, { "epoch": 1.3077250478825282, "grad_norm": 11.3359375, "learning_rate": 9.581083001705515e-06, "loss": 0.9054, "step": 12290 }, { "epoch": 1.3078314535007447, "grad_norm": 11.625, "learning_rate": 9.581047470153497e-06, "loss": 0.9668, "step": 12291 }, { "epoch": 1.3079378591189614, "grad_norm": 19.984375, "learning_rate": 9.581011938601478e-06, "loss": 1.1553, "step": 12292 }, { "epoch": 1.3080442647371782, "grad_norm": 14.4609375, "learning_rate": 9.58097640704946e-06, "loss": 1.2226, "step": 12293 }, { "epoch": 1.3081506703553947, "grad_norm": 11.859375, "learning_rate": 9.580940875497441e-06, "loss": 1.1939, "step": 12294 }, { "epoch": 1.3082570759736114, "grad_norm": 11.3359375, "learning_rate": 9.580905343945425e-06, "loss": 0.9828, "step": 12295 }, { "epoch": 1.3083634815918281, "grad_norm": 13.890625, "learning_rate": 9.580869812393406e-06, "loss": 1.127, "step": 12296 }, { "epoch": 1.3084698872100446, "grad_norm": 14.1171875, "learning_rate": 9.580834280841388e-06, "loss": 1.1358, "step": 12297 }, { "epoch": 1.3085762928282614, "grad_norm": 10.296875, "learning_rate": 9.58079874928937e-06, "loss": 0.9046, "step": 12298 }, { "epoch": 1.308682698446478, "grad_norm": 19.203125, "learning_rate": 9.580763217737351e-06, "loss": 1.1481, "step": 12299 }, { "epoch": 1.3087891040646946, "grad_norm": 14.921875, "learning_rate": 9.580727686185334e-06, "loss": 0.9711, "step": 12300 }, { "epoch": 1.3088955096829111, "grad_norm": 9.1796875, "learning_rate": 9.580692154633316e-06, "loss": 0.871, "step": 12301 }, { "epoch": 1.3090019153011279, "grad_norm": 19.078125, "learning_rate": 9.580656623081298e-06, "loss": 1.0291, "step": 12302 }, { "epoch": 1.3091083209193446, "grad_norm": 16.546875, "learning_rate": 9.58062109152928e-06, "loss": 1.1252, "step": 12303 }, { "epoch": 1.309214726537561, "grad_norm": 14.9140625, "learning_rate": 9.580585559977261e-06, "loss": 1.1935, "step": 12304 }, { "epoch": 1.3093211321557778, "grad_norm": 14.1953125, "learning_rate": 9.580550028425243e-06, "loss": 1.3343, "step": 12305 }, { "epoch": 1.3094275377739946, "grad_norm": 11.78125, "learning_rate": 9.580514496873224e-06, "loss": 1.0981, "step": 12306 }, { "epoch": 1.309533943392211, "grad_norm": 12.90625, "learning_rate": 9.580478965321206e-06, "loss": 1.5405, "step": 12307 }, { "epoch": 1.3096403490104278, "grad_norm": 13.7578125, "learning_rate": 9.580443433769187e-06, "loss": 1.1417, "step": 12308 }, { "epoch": 1.3097467546286443, "grad_norm": 14.3671875, "learning_rate": 9.580407902217169e-06, "loss": 1.0221, "step": 12309 }, { "epoch": 1.309853160246861, "grad_norm": 9.0859375, "learning_rate": 9.580372370665152e-06, "loss": 0.874, "step": 12310 }, { "epoch": 1.3099595658650776, "grad_norm": 15.078125, "learning_rate": 9.580336839113134e-06, "loss": 0.9361, "step": 12311 }, { "epoch": 1.3100659714832943, "grad_norm": 10.7265625, "learning_rate": 9.580301307561115e-06, "loss": 1.0925, "step": 12312 }, { "epoch": 1.310172377101511, "grad_norm": 12.640625, "learning_rate": 9.580265776009097e-06, "loss": 0.9483, "step": 12313 }, { "epoch": 1.3102787827197275, "grad_norm": 16.03125, "learning_rate": 9.580230244457079e-06, "loss": 1.1231, "step": 12314 }, { "epoch": 1.3103851883379443, "grad_norm": 11.7109375, "learning_rate": 9.58019471290506e-06, "loss": 1.1924, "step": 12315 }, { "epoch": 1.310491593956161, "grad_norm": 11.328125, "learning_rate": 9.580159181353042e-06, "loss": 0.8812, "step": 12316 }, { "epoch": 1.3105979995743775, "grad_norm": 15.3984375, "learning_rate": 9.580123649801023e-06, "loss": 1.0689, "step": 12317 }, { "epoch": 1.3107044051925942, "grad_norm": 20.765625, "learning_rate": 9.580088118249005e-06, "loss": 1.0436, "step": 12318 }, { "epoch": 1.3108108108108107, "grad_norm": 12.4453125, "learning_rate": 9.580052586696987e-06, "loss": 1.3442, "step": 12319 }, { "epoch": 1.3109172164290275, "grad_norm": 11.8515625, "learning_rate": 9.58001705514497e-06, "loss": 1.0603, "step": 12320 }, { "epoch": 1.311023622047244, "grad_norm": 12.9375, "learning_rate": 9.579981523592952e-06, "loss": 1.151, "step": 12321 }, { "epoch": 1.3111300276654607, "grad_norm": 16.515625, "learning_rate": 9.579945992040933e-06, "loss": 0.8952, "step": 12322 }, { "epoch": 1.3112364332836774, "grad_norm": 13.53125, "learning_rate": 9.579910460488915e-06, "loss": 0.8439, "step": 12323 }, { "epoch": 1.311342838901894, "grad_norm": 14.296875, "learning_rate": 9.579874928936896e-06, "loss": 0.9958, "step": 12324 }, { "epoch": 1.3114492445201107, "grad_norm": 12.5859375, "learning_rate": 9.579839397384878e-06, "loss": 1.1268, "step": 12325 }, { "epoch": 1.3115556501383274, "grad_norm": 12.171875, "learning_rate": 9.579803865832861e-06, "loss": 1.1951, "step": 12326 }, { "epoch": 1.311662055756544, "grad_norm": 19.890625, "learning_rate": 9.579768334280843e-06, "loss": 0.929, "step": 12327 }, { "epoch": 1.3117684613747607, "grad_norm": 15.1640625, "learning_rate": 9.579732802728825e-06, "loss": 1.0607, "step": 12328 }, { "epoch": 1.3118748669929772, "grad_norm": 21.5, "learning_rate": 9.579697271176806e-06, "loss": 1.0326, "step": 12329 }, { "epoch": 1.311981272611194, "grad_norm": 20.171875, "learning_rate": 9.579661739624788e-06, "loss": 0.9125, "step": 12330 }, { "epoch": 1.3120876782294104, "grad_norm": 11.3203125, "learning_rate": 9.57962620807277e-06, "loss": 0.8067, "step": 12331 }, { "epoch": 1.3121940838476271, "grad_norm": 13.7734375, "learning_rate": 9.579590676520751e-06, "loss": 1.2662, "step": 12332 }, { "epoch": 1.3123004894658439, "grad_norm": 10.578125, "learning_rate": 9.579555144968733e-06, "loss": 1.059, "step": 12333 }, { "epoch": 1.3124068950840604, "grad_norm": 13.3984375, "learning_rate": 9.579519613416714e-06, "loss": 1.0143, "step": 12334 }, { "epoch": 1.3125133007022771, "grad_norm": 11.484375, "learning_rate": 9.579484081864696e-06, "loss": 1.2236, "step": 12335 }, { "epoch": 1.3126197063204939, "grad_norm": 13.671875, "learning_rate": 9.579448550312679e-06, "loss": 1.0792, "step": 12336 }, { "epoch": 1.3127261119387104, "grad_norm": 23.546875, "learning_rate": 9.57941301876066e-06, "loss": 1.2074, "step": 12337 }, { "epoch": 1.3128325175569269, "grad_norm": 10.890625, "learning_rate": 9.579377487208642e-06, "loss": 0.9952, "step": 12338 }, { "epoch": 1.3129389231751436, "grad_norm": 16.265625, "learning_rate": 9.579341955656624e-06, "loss": 1.0729, "step": 12339 }, { "epoch": 1.3130453287933603, "grad_norm": 11.6796875, "learning_rate": 9.579306424104606e-06, "loss": 1.209, "step": 12340 }, { "epoch": 1.3131517344115768, "grad_norm": 11.6640625, "learning_rate": 9.579270892552587e-06, "loss": 1.1965, "step": 12341 }, { "epoch": 1.3132581400297936, "grad_norm": 16.015625, "learning_rate": 9.579235361000569e-06, "loss": 0.7512, "step": 12342 }, { "epoch": 1.3133645456480103, "grad_norm": 10.8125, "learning_rate": 9.57919982944855e-06, "loss": 0.9495, "step": 12343 }, { "epoch": 1.3134709512662268, "grad_norm": 13.53125, "learning_rate": 9.579164297896532e-06, "loss": 0.9215, "step": 12344 }, { "epoch": 1.3135773568844435, "grad_norm": 12.0, "learning_rate": 9.579128766344514e-06, "loss": 1.2576, "step": 12345 }, { "epoch": 1.31368376250266, "grad_norm": 9.8515625, "learning_rate": 9.579093234792497e-06, "loss": 1.0694, "step": 12346 }, { "epoch": 1.3137901681208768, "grad_norm": 12.828125, "learning_rate": 9.579057703240478e-06, "loss": 1.2061, "step": 12347 }, { "epoch": 1.3138965737390933, "grad_norm": 15.0859375, "learning_rate": 9.57902217168846e-06, "loss": 1.0437, "step": 12348 }, { "epoch": 1.31400297935731, "grad_norm": 12.3359375, "learning_rate": 9.578986640136442e-06, "loss": 1.0762, "step": 12349 }, { "epoch": 1.3141093849755268, "grad_norm": 17.390625, "learning_rate": 9.578951108584423e-06, "loss": 1.2467, "step": 12350 }, { "epoch": 1.3142157905937433, "grad_norm": 20.25, "learning_rate": 9.578915577032407e-06, "loss": 1.1353, "step": 12351 }, { "epoch": 1.31432219621196, "grad_norm": 17.40625, "learning_rate": 9.578880045480388e-06, "loss": 1.0302, "step": 12352 }, { "epoch": 1.3144286018301767, "grad_norm": 14.890625, "learning_rate": 9.57884451392837e-06, "loss": 1.412, "step": 12353 }, { "epoch": 1.3145350074483932, "grad_norm": 15.5859375, "learning_rate": 9.578808982376351e-06, "loss": 1.6547, "step": 12354 }, { "epoch": 1.31464141306661, "grad_norm": 13.8671875, "learning_rate": 9.578773450824333e-06, "loss": 1.1076, "step": 12355 }, { "epoch": 1.3147478186848265, "grad_norm": 10.5, "learning_rate": 9.578737919272315e-06, "loss": 1.0177, "step": 12356 }, { "epoch": 1.3148542243030432, "grad_norm": 10.3828125, "learning_rate": 9.578702387720296e-06, "loss": 1.2255, "step": 12357 }, { "epoch": 1.3149606299212597, "grad_norm": 14.96875, "learning_rate": 9.578666856168278e-06, "loss": 1.116, "step": 12358 }, { "epoch": 1.3150670355394765, "grad_norm": 11.4296875, "learning_rate": 9.57863132461626e-06, "loss": 1.0569, "step": 12359 }, { "epoch": 1.3151734411576932, "grad_norm": 9.3984375, "learning_rate": 9.578595793064241e-06, "loss": 1.0131, "step": 12360 }, { "epoch": 1.3152798467759097, "grad_norm": 11.546875, "learning_rate": 9.578560261512224e-06, "loss": 0.9701, "step": 12361 }, { "epoch": 1.3153862523941264, "grad_norm": 10.6171875, "learning_rate": 9.578524729960206e-06, "loss": 0.8713, "step": 12362 }, { "epoch": 1.3154926580123432, "grad_norm": 10.2265625, "learning_rate": 9.578489198408188e-06, "loss": 0.7921, "step": 12363 }, { "epoch": 1.3155990636305597, "grad_norm": 15.9921875, "learning_rate": 9.57845366685617e-06, "loss": 1.1693, "step": 12364 }, { "epoch": 1.3157054692487764, "grad_norm": 11.7109375, "learning_rate": 9.57841813530415e-06, "loss": 0.9072, "step": 12365 }, { "epoch": 1.315811874866993, "grad_norm": 9.5, "learning_rate": 9.578382603752132e-06, "loss": 0.8879, "step": 12366 }, { "epoch": 1.3159182804852096, "grad_norm": 14.9765625, "learning_rate": 9.578347072200114e-06, "loss": 1.0257, "step": 12367 }, { "epoch": 1.3160246861034262, "grad_norm": 12.265625, "learning_rate": 9.578311540648096e-06, "loss": 1.219, "step": 12368 }, { "epoch": 1.3161310917216429, "grad_norm": 13.7578125, "learning_rate": 9.578276009096077e-06, "loss": 0.9962, "step": 12369 }, { "epoch": 1.3162374973398596, "grad_norm": 10.6484375, "learning_rate": 9.578240477544059e-06, "loss": 0.902, "step": 12370 }, { "epoch": 1.3163439029580761, "grad_norm": 13.1171875, "learning_rate": 9.57820494599204e-06, "loss": 1.0547, "step": 12371 }, { "epoch": 1.3164503085762929, "grad_norm": 15.0390625, "learning_rate": 9.578169414440024e-06, "loss": 0.9866, "step": 12372 }, { "epoch": 1.3165567141945096, "grad_norm": 14.75, "learning_rate": 9.578133882888005e-06, "loss": 0.9739, "step": 12373 }, { "epoch": 1.316663119812726, "grad_norm": 13.71875, "learning_rate": 9.578098351335987e-06, "loss": 0.9062, "step": 12374 }, { "epoch": 1.3167695254309428, "grad_norm": 16.15625, "learning_rate": 9.578062819783969e-06, "loss": 1.1233, "step": 12375 }, { "epoch": 1.3168759310491593, "grad_norm": 10.0546875, "learning_rate": 9.57802728823195e-06, "loss": 1.1202, "step": 12376 }, { "epoch": 1.316982336667376, "grad_norm": 11.2890625, "learning_rate": 9.577991756679933e-06, "loss": 1.2113, "step": 12377 }, { "epoch": 1.3170887422855926, "grad_norm": 16.421875, "learning_rate": 9.577956225127915e-06, "loss": 1.1366, "step": 12378 }, { "epoch": 1.3171951479038093, "grad_norm": 10.5625, "learning_rate": 9.577920693575897e-06, "loss": 0.9524, "step": 12379 }, { "epoch": 1.317301553522026, "grad_norm": 13.1484375, "learning_rate": 9.577885162023878e-06, "loss": 1.109, "step": 12380 }, { "epoch": 1.3174079591402426, "grad_norm": 13.4765625, "learning_rate": 9.57784963047186e-06, "loss": 1.1457, "step": 12381 }, { "epoch": 1.3175143647584593, "grad_norm": 9.828125, "learning_rate": 9.577814098919841e-06, "loss": 0.9936, "step": 12382 }, { "epoch": 1.317620770376676, "grad_norm": 10.8203125, "learning_rate": 9.577778567367823e-06, "loss": 1.0948, "step": 12383 }, { "epoch": 1.3177271759948925, "grad_norm": 16.171875, "learning_rate": 9.577743035815805e-06, "loss": 1.1484, "step": 12384 }, { "epoch": 1.317833581613109, "grad_norm": 12.53125, "learning_rate": 9.577707504263786e-06, "loss": 1.2724, "step": 12385 }, { "epoch": 1.3179399872313258, "grad_norm": 10.3125, "learning_rate": 9.577671972711768e-06, "loss": 1.0151, "step": 12386 }, { "epoch": 1.3180463928495425, "grad_norm": 17.359375, "learning_rate": 9.577636441159751e-06, "loss": 1.1471, "step": 12387 }, { "epoch": 1.318152798467759, "grad_norm": 11.1171875, "learning_rate": 9.577600909607733e-06, "loss": 0.8858, "step": 12388 }, { "epoch": 1.3182592040859757, "grad_norm": 10.6796875, "learning_rate": 9.577565378055714e-06, "loss": 0.9956, "step": 12389 }, { "epoch": 1.3183656097041925, "grad_norm": 11.21875, "learning_rate": 9.577529846503696e-06, "loss": 1.2858, "step": 12390 }, { "epoch": 1.318472015322409, "grad_norm": 13.5078125, "learning_rate": 9.577494314951678e-06, "loss": 1.3234, "step": 12391 }, { "epoch": 1.3185784209406257, "grad_norm": 11.7421875, "learning_rate": 9.577458783399661e-06, "loss": 1.1036, "step": 12392 }, { "epoch": 1.3186848265588422, "grad_norm": 11.21875, "learning_rate": 9.577423251847643e-06, "loss": 1.0729, "step": 12393 }, { "epoch": 1.318791232177059, "grad_norm": 11.0625, "learning_rate": 9.577387720295624e-06, "loss": 1.1618, "step": 12394 }, { "epoch": 1.3188976377952755, "grad_norm": 12.921875, "learning_rate": 9.577352188743606e-06, "loss": 1.0369, "step": 12395 }, { "epoch": 1.3190040434134922, "grad_norm": 13.9609375, "learning_rate": 9.577316657191587e-06, "loss": 1.0201, "step": 12396 }, { "epoch": 1.319110449031709, "grad_norm": 14.109375, "learning_rate": 9.577281125639569e-06, "loss": 1.1646, "step": 12397 }, { "epoch": 1.3192168546499254, "grad_norm": 11.1015625, "learning_rate": 9.57724559408755e-06, "loss": 1.2399, "step": 12398 }, { "epoch": 1.3193232602681422, "grad_norm": 14.3515625, "learning_rate": 9.577210062535532e-06, "loss": 1.0936, "step": 12399 }, { "epoch": 1.319429665886359, "grad_norm": 12.859375, "learning_rate": 9.577174530983514e-06, "loss": 1.3742, "step": 12400 }, { "epoch": 1.3195360715045754, "grad_norm": 10.46875, "learning_rate": 9.577138999431495e-06, "loss": 1.3224, "step": 12401 }, { "epoch": 1.3196424771227921, "grad_norm": 15.0859375, "learning_rate": 9.577103467879479e-06, "loss": 1.293, "step": 12402 }, { "epoch": 1.3197488827410087, "grad_norm": 9.515625, "learning_rate": 9.57706793632746e-06, "loss": 1.0615, "step": 12403 }, { "epoch": 1.3198552883592254, "grad_norm": 11.109375, "learning_rate": 9.577032404775442e-06, "loss": 1.1137, "step": 12404 }, { "epoch": 1.319961693977442, "grad_norm": 13.2890625, "learning_rate": 9.576996873223424e-06, "loss": 0.9307, "step": 12405 }, { "epoch": 1.3200680995956586, "grad_norm": 12.7421875, "learning_rate": 9.576961341671405e-06, "loss": 1.1015, "step": 12406 }, { "epoch": 1.3201745052138754, "grad_norm": 11.625, "learning_rate": 9.576925810119387e-06, "loss": 1.0076, "step": 12407 }, { "epoch": 1.3202809108320919, "grad_norm": 11.53125, "learning_rate": 9.576890278567368e-06, "loss": 0.7946, "step": 12408 }, { "epoch": 1.3203873164503086, "grad_norm": 8.40625, "learning_rate": 9.57685474701535e-06, "loss": 1.1956, "step": 12409 }, { "epoch": 1.3204937220685253, "grad_norm": 13.5078125, "learning_rate": 9.576819215463332e-06, "loss": 0.9778, "step": 12410 }, { "epoch": 1.3206001276867418, "grad_norm": 11.3828125, "learning_rate": 9.576783683911313e-06, "loss": 0.801, "step": 12411 }, { "epoch": 1.3207065333049586, "grad_norm": 9.984375, "learning_rate": 9.576748152359295e-06, "loss": 0.9355, "step": 12412 }, { "epoch": 1.320812938923175, "grad_norm": 14.75, "learning_rate": 9.576712620807278e-06, "loss": 1.0841, "step": 12413 }, { "epoch": 1.3209193445413918, "grad_norm": 13.8203125, "learning_rate": 9.57667708925526e-06, "loss": 1.4062, "step": 12414 }, { "epoch": 1.3210257501596083, "grad_norm": 12.125, "learning_rate": 9.576641557703241e-06, "loss": 1.1114, "step": 12415 }, { "epoch": 1.321132155777825, "grad_norm": 14.46875, "learning_rate": 9.576606026151223e-06, "loss": 1.1773, "step": 12416 }, { "epoch": 1.3212385613960418, "grad_norm": 14.9765625, "learning_rate": 9.576570494599204e-06, "loss": 1.4171, "step": 12417 }, { "epoch": 1.3213449670142583, "grad_norm": 13.1015625, "learning_rate": 9.576534963047188e-06, "loss": 0.8712, "step": 12418 }, { "epoch": 1.321451372632475, "grad_norm": 12.3125, "learning_rate": 9.57649943149517e-06, "loss": 1.3243, "step": 12419 }, { "epoch": 1.3215577782506918, "grad_norm": 18.078125, "learning_rate": 9.576463899943151e-06, "loss": 1.1286, "step": 12420 }, { "epoch": 1.3216641838689083, "grad_norm": 14.1015625, "learning_rate": 9.576428368391133e-06, "loss": 1.0185, "step": 12421 }, { "epoch": 1.321770589487125, "grad_norm": 11.0546875, "learning_rate": 9.576392836839114e-06, "loss": 0.914, "step": 12422 }, { "epoch": 1.3218769951053415, "grad_norm": 12.3359375, "learning_rate": 9.576357305287096e-06, "loss": 1.1652, "step": 12423 }, { "epoch": 1.3219834007235582, "grad_norm": 11.796875, "learning_rate": 9.576321773735077e-06, "loss": 1.1659, "step": 12424 }, { "epoch": 1.3220898063417748, "grad_norm": 14.8671875, "learning_rate": 9.576286242183059e-06, "loss": 1.1582, "step": 12425 }, { "epoch": 1.3221962119599915, "grad_norm": 12.8828125, "learning_rate": 9.57625071063104e-06, "loss": 1.1742, "step": 12426 }, { "epoch": 1.3223026175782082, "grad_norm": 14.03125, "learning_rate": 9.576215179079022e-06, "loss": 1.1134, "step": 12427 }, { "epoch": 1.3224090231964247, "grad_norm": 12.2578125, "learning_rate": 9.576179647527006e-06, "loss": 1.211, "step": 12428 }, { "epoch": 1.3225154288146415, "grad_norm": 15.109375, "learning_rate": 9.576144115974987e-06, "loss": 1.0567, "step": 12429 }, { "epoch": 1.3226218344328582, "grad_norm": 15.2421875, "learning_rate": 9.576108584422969e-06, "loss": 1.0635, "step": 12430 }, { "epoch": 1.3227282400510747, "grad_norm": 11.5703125, "learning_rate": 9.57607305287095e-06, "loss": 1.3276, "step": 12431 }, { "epoch": 1.3228346456692912, "grad_norm": 11.75, "learning_rate": 9.576037521318932e-06, "loss": 1.0227, "step": 12432 }, { "epoch": 1.322941051287508, "grad_norm": 11.8203125, "learning_rate": 9.576001989766914e-06, "loss": 0.9735, "step": 12433 }, { "epoch": 1.3230474569057247, "grad_norm": 13.578125, "learning_rate": 9.575966458214895e-06, "loss": 1.3958, "step": 12434 }, { "epoch": 1.3231538625239412, "grad_norm": 13.171875, "learning_rate": 9.575930926662877e-06, "loss": 1.0843, "step": 12435 }, { "epoch": 1.323260268142158, "grad_norm": 13.6328125, "learning_rate": 9.575895395110858e-06, "loss": 0.9823, "step": 12436 }, { "epoch": 1.3233666737603746, "grad_norm": 13.078125, "learning_rate": 9.57585986355884e-06, "loss": 1.1066, "step": 12437 }, { "epoch": 1.3234730793785912, "grad_norm": 12.3515625, "learning_rate": 9.575824332006823e-06, "loss": 1.2928, "step": 12438 }, { "epoch": 1.3235794849968079, "grad_norm": 13.53125, "learning_rate": 9.575788800454805e-06, "loss": 1.1272, "step": 12439 }, { "epoch": 1.3236858906150244, "grad_norm": 14.34375, "learning_rate": 9.575753268902787e-06, "loss": 0.8057, "step": 12440 }, { "epoch": 1.3237922962332411, "grad_norm": 14.015625, "learning_rate": 9.575717737350768e-06, "loss": 1.1118, "step": 12441 }, { "epoch": 1.3238987018514576, "grad_norm": 10.40625, "learning_rate": 9.57568220579875e-06, "loss": 0.9952, "step": 12442 }, { "epoch": 1.3240051074696744, "grad_norm": 12.4296875, "learning_rate": 9.575646674246731e-06, "loss": 1.1758, "step": 12443 }, { "epoch": 1.324111513087891, "grad_norm": 9.5234375, "learning_rate": 9.575611142694715e-06, "loss": 1.0202, "step": 12444 }, { "epoch": 1.3242179187061076, "grad_norm": 16.4375, "learning_rate": 9.575575611142696e-06, "loss": 1.2778, "step": 12445 }, { "epoch": 1.3243243243243243, "grad_norm": 13.328125, "learning_rate": 9.575540079590678e-06, "loss": 1.0176, "step": 12446 }, { "epoch": 1.324430729942541, "grad_norm": 12.21875, "learning_rate": 9.57550454803866e-06, "loss": 1.5519, "step": 12447 }, { "epoch": 1.3245371355607576, "grad_norm": 14.28125, "learning_rate": 9.575469016486641e-06, "loss": 1.0849, "step": 12448 }, { "epoch": 1.3246435411789743, "grad_norm": 14.296875, "learning_rate": 9.575433484934623e-06, "loss": 1.0825, "step": 12449 }, { "epoch": 1.3247499467971908, "grad_norm": 11.65625, "learning_rate": 9.575397953382604e-06, "loss": 0.9539, "step": 12450 }, { "epoch": 1.3248563524154076, "grad_norm": 10.640625, "learning_rate": 9.575362421830586e-06, "loss": 0.9111, "step": 12451 }, { "epoch": 1.324962758033624, "grad_norm": 14.5703125, "learning_rate": 9.575326890278567e-06, "loss": 1.3158, "step": 12452 }, { "epoch": 1.3250691636518408, "grad_norm": 11.015625, "learning_rate": 9.575291358726549e-06, "loss": 1.2658, "step": 12453 }, { "epoch": 1.3251755692700575, "grad_norm": 17.90625, "learning_rate": 9.575255827174532e-06, "loss": 1.3811, "step": 12454 }, { "epoch": 1.325281974888274, "grad_norm": 8.125, "learning_rate": 9.575220295622514e-06, "loss": 0.8669, "step": 12455 }, { "epoch": 1.3253883805064908, "grad_norm": 10.46875, "learning_rate": 9.575184764070496e-06, "loss": 0.9696, "step": 12456 }, { "epoch": 1.3254947861247075, "grad_norm": 12.90625, "learning_rate": 9.575149232518477e-06, "loss": 1.0709, "step": 12457 }, { "epoch": 1.325601191742924, "grad_norm": 18.5625, "learning_rate": 9.575113700966459e-06, "loss": 1.1461, "step": 12458 }, { "epoch": 1.3257075973611407, "grad_norm": 8.9921875, "learning_rate": 9.57507816941444e-06, "loss": 0.9724, "step": 12459 }, { "epoch": 1.3258140029793573, "grad_norm": 10.6171875, "learning_rate": 9.575042637862422e-06, "loss": 0.9815, "step": 12460 }, { "epoch": 1.325920408597574, "grad_norm": 11.53125, "learning_rate": 9.575007106310404e-06, "loss": 1.0642, "step": 12461 }, { "epoch": 1.3260268142157905, "grad_norm": 9.3828125, "learning_rate": 9.574971574758385e-06, "loss": 1.441, "step": 12462 }, { "epoch": 1.3261332198340072, "grad_norm": 15.515625, "learning_rate": 9.574936043206367e-06, "loss": 0.9415, "step": 12463 }, { "epoch": 1.326239625452224, "grad_norm": 10.4375, "learning_rate": 9.57490051165435e-06, "loss": 0.9731, "step": 12464 }, { "epoch": 1.3263460310704405, "grad_norm": 10.1640625, "learning_rate": 9.574864980102332e-06, "loss": 1.04, "step": 12465 }, { "epoch": 1.3264524366886572, "grad_norm": 10.8828125, "learning_rate": 9.574829448550313e-06, "loss": 1.0858, "step": 12466 }, { "epoch": 1.326558842306874, "grad_norm": 10.8359375, "learning_rate": 9.574793916998295e-06, "loss": 0.9019, "step": 12467 }, { "epoch": 1.3266652479250904, "grad_norm": 10.890625, "learning_rate": 9.574758385446277e-06, "loss": 1.0991, "step": 12468 }, { "epoch": 1.326771653543307, "grad_norm": 10.65625, "learning_rate": 9.57472285389426e-06, "loss": 1.0436, "step": 12469 }, { "epoch": 1.3268780591615237, "grad_norm": 15.09375, "learning_rate": 9.574687322342242e-06, "loss": 1.1331, "step": 12470 }, { "epoch": 1.3269844647797404, "grad_norm": 13.4921875, "learning_rate": 9.574651790790223e-06, "loss": 1.1689, "step": 12471 }, { "epoch": 1.327090870397957, "grad_norm": 10.8359375, "learning_rate": 9.574616259238205e-06, "loss": 1.0513, "step": 12472 }, { "epoch": 1.3271972760161737, "grad_norm": 15.7421875, "learning_rate": 9.574580727686186e-06, "loss": 1.0016, "step": 12473 }, { "epoch": 1.3273036816343904, "grad_norm": 13.0703125, "learning_rate": 9.574545196134168e-06, "loss": 1.1377, "step": 12474 }, { "epoch": 1.327410087252607, "grad_norm": 11.671875, "learning_rate": 9.57450966458215e-06, "loss": 1.0172, "step": 12475 }, { "epoch": 1.3275164928708236, "grad_norm": 12.703125, "learning_rate": 9.574474133030131e-06, "loss": 1.0744, "step": 12476 }, { "epoch": 1.3276228984890404, "grad_norm": 12.8203125, "learning_rate": 9.574438601478113e-06, "loss": 1.2515, "step": 12477 }, { "epoch": 1.3277293041072569, "grad_norm": 10.078125, "learning_rate": 9.574403069926094e-06, "loss": 1.1655, "step": 12478 }, { "epoch": 1.3278357097254734, "grad_norm": 13.0859375, "learning_rate": 9.574367538374078e-06, "loss": 0.8429, "step": 12479 }, { "epoch": 1.32794211534369, "grad_norm": 10.984375, "learning_rate": 9.57433200682206e-06, "loss": 0.9408, "step": 12480 }, { "epoch": 1.3280485209619068, "grad_norm": 11.6484375, "learning_rate": 9.574296475270041e-06, "loss": 0.9668, "step": 12481 }, { "epoch": 1.3281549265801234, "grad_norm": 16.359375, "learning_rate": 9.574260943718022e-06, "loss": 0.8348, "step": 12482 }, { "epoch": 1.32826133219834, "grad_norm": 13.65625, "learning_rate": 9.574225412166004e-06, "loss": 1.0937, "step": 12483 }, { "epoch": 1.3283677378165568, "grad_norm": 10.46875, "learning_rate": 9.574189880613986e-06, "loss": 1.145, "step": 12484 }, { "epoch": 1.3284741434347733, "grad_norm": 18.75, "learning_rate": 9.574154349061969e-06, "loss": 1.1228, "step": 12485 }, { "epoch": 1.32858054905299, "grad_norm": 13.3671875, "learning_rate": 9.57411881750995e-06, "loss": 1.1445, "step": 12486 }, { "epoch": 1.3286869546712066, "grad_norm": 9.734375, "learning_rate": 9.57408328595793e-06, "loss": 0.9413, "step": 12487 }, { "epoch": 1.3287933602894233, "grad_norm": 12.7734375, "learning_rate": 9.574047754405912e-06, "loss": 1.1992, "step": 12488 }, { "epoch": 1.3288997659076398, "grad_norm": 11.3125, "learning_rate": 9.574012222853894e-06, "loss": 0.9062, "step": 12489 }, { "epoch": 1.3290061715258565, "grad_norm": 11.8125, "learning_rate": 9.573976691301877e-06, "loss": 1.0079, "step": 12490 }, { "epoch": 1.3291125771440733, "grad_norm": 10.578125, "learning_rate": 9.573941159749859e-06, "loss": 1.1007, "step": 12491 }, { "epoch": 1.3292189827622898, "grad_norm": 10.8359375, "learning_rate": 9.57390562819784e-06, "loss": 0.9351, "step": 12492 }, { "epoch": 1.3293253883805065, "grad_norm": 14.546875, "learning_rate": 9.573870096645822e-06, "loss": 1.2069, "step": 12493 }, { "epoch": 1.3294317939987232, "grad_norm": 19.359375, "learning_rate": 9.573834565093803e-06, "loss": 1.3744, "step": 12494 }, { "epoch": 1.3295381996169398, "grad_norm": 12.78125, "learning_rate": 9.573799033541787e-06, "loss": 1.273, "step": 12495 }, { "epoch": 1.3296446052351565, "grad_norm": 17.0, "learning_rate": 9.573763501989768e-06, "loss": 1.1345, "step": 12496 }, { "epoch": 1.329751010853373, "grad_norm": 10.0546875, "learning_rate": 9.57372797043775e-06, "loss": 1.121, "step": 12497 }, { "epoch": 1.3298574164715897, "grad_norm": 9.7734375, "learning_rate": 9.573692438885732e-06, "loss": 1.1711, "step": 12498 }, { "epoch": 1.3299638220898062, "grad_norm": 10.796875, "learning_rate": 9.573656907333713e-06, "loss": 1.242, "step": 12499 }, { "epoch": 1.330070227708023, "grad_norm": 11.5390625, "learning_rate": 9.573621375781695e-06, "loss": 1.0446, "step": 12500 }, { "epoch": 1.3301766333262397, "grad_norm": 10.5625, "learning_rate": 9.573585844229676e-06, "loss": 1.266, "step": 12501 }, { "epoch": 1.3302830389444562, "grad_norm": 12.109375, "learning_rate": 9.573550312677658e-06, "loss": 1.2691, "step": 12502 }, { "epoch": 1.330389444562673, "grad_norm": 11.671875, "learning_rate": 9.57351478112564e-06, "loss": 1.0969, "step": 12503 }, { "epoch": 1.3304958501808897, "grad_norm": 8.28125, "learning_rate": 9.573479249573621e-06, "loss": 1.1644, "step": 12504 }, { "epoch": 1.3306022557991062, "grad_norm": 8.3359375, "learning_rate": 9.573443718021605e-06, "loss": 1.0388, "step": 12505 }, { "epoch": 1.330708661417323, "grad_norm": 14.3203125, "learning_rate": 9.573408186469586e-06, "loss": 1.2506, "step": 12506 }, { "epoch": 1.3308150670355394, "grad_norm": 13.4609375, "learning_rate": 9.573372654917568e-06, "loss": 1.2356, "step": 12507 }, { "epoch": 1.3309214726537562, "grad_norm": 10.609375, "learning_rate": 9.57333712336555e-06, "loss": 1.2566, "step": 12508 }, { "epoch": 1.3310278782719727, "grad_norm": 19.453125, "learning_rate": 9.573301591813531e-06, "loss": 1.0723, "step": 12509 }, { "epoch": 1.3311342838901894, "grad_norm": 14.0546875, "learning_rate": 9.573266060261514e-06, "loss": 1.0801, "step": 12510 }, { "epoch": 1.3312406895084061, "grad_norm": 11.9765625, "learning_rate": 9.573230528709496e-06, "loss": 0.9409, "step": 12511 }, { "epoch": 1.3313470951266226, "grad_norm": 11.6328125, "learning_rate": 9.573194997157477e-06, "loss": 1.0702, "step": 12512 }, { "epoch": 1.3314535007448394, "grad_norm": 10.25, "learning_rate": 9.573159465605459e-06, "loss": 0.9139, "step": 12513 }, { "epoch": 1.331559906363056, "grad_norm": 13.140625, "learning_rate": 9.57312393405344e-06, "loss": 1.209, "step": 12514 }, { "epoch": 1.3316663119812726, "grad_norm": 14.0703125, "learning_rate": 9.573088402501422e-06, "loss": 0.963, "step": 12515 }, { "epoch": 1.3317727175994891, "grad_norm": 12.484375, "learning_rate": 9.573052870949404e-06, "loss": 1.2224, "step": 12516 }, { "epoch": 1.3318791232177059, "grad_norm": 14.21875, "learning_rate": 9.573017339397385e-06, "loss": 1.1253, "step": 12517 }, { "epoch": 1.3319855288359226, "grad_norm": 11.7578125, "learning_rate": 9.572981807845367e-06, "loss": 1.024, "step": 12518 }, { "epoch": 1.332091934454139, "grad_norm": 11.703125, "learning_rate": 9.572946276293349e-06, "loss": 1.1928, "step": 12519 }, { "epoch": 1.3321983400723558, "grad_norm": 14.265625, "learning_rate": 9.57291074474133e-06, "loss": 1.707, "step": 12520 }, { "epoch": 1.3323047456905726, "grad_norm": 11.796875, "learning_rate": 9.572875213189314e-06, "loss": 0.7842, "step": 12521 }, { "epoch": 1.332411151308789, "grad_norm": 10.703125, "learning_rate": 9.572839681637295e-06, "loss": 0.902, "step": 12522 }, { "epoch": 1.3325175569270058, "grad_norm": 10.5390625, "learning_rate": 9.572804150085277e-06, "loss": 1.0614, "step": 12523 }, { "epoch": 1.3326239625452223, "grad_norm": 14.7578125, "learning_rate": 9.572768618533258e-06, "loss": 0.9546, "step": 12524 }, { "epoch": 1.332730368163439, "grad_norm": 11.7421875, "learning_rate": 9.57273308698124e-06, "loss": 1.4624, "step": 12525 }, { "epoch": 1.3328367737816555, "grad_norm": 9.8515625, "learning_rate": 9.572697555429222e-06, "loss": 1.1564, "step": 12526 }, { "epoch": 1.3329431793998723, "grad_norm": 11.1328125, "learning_rate": 9.572662023877203e-06, "loss": 1.1562, "step": 12527 }, { "epoch": 1.333049585018089, "grad_norm": 13.1953125, "learning_rate": 9.572626492325185e-06, "loss": 1.076, "step": 12528 }, { "epoch": 1.3331559906363055, "grad_norm": 10.6953125, "learning_rate": 9.572590960773166e-06, "loss": 0.9786, "step": 12529 }, { "epoch": 1.3332623962545223, "grad_norm": 11.25, "learning_rate": 9.572555429221148e-06, "loss": 1.1168, "step": 12530 }, { "epoch": 1.333368801872739, "grad_norm": 13.8984375, "learning_rate": 9.572519897669131e-06, "loss": 1.0211, "step": 12531 }, { "epoch": 1.3334752074909555, "grad_norm": 14.046875, "learning_rate": 9.572484366117113e-06, "loss": 1.2342, "step": 12532 }, { "epoch": 1.3335816131091722, "grad_norm": 12.328125, "learning_rate": 9.572448834565095e-06, "loss": 1.3197, "step": 12533 }, { "epoch": 1.3336880187273887, "grad_norm": 9.71875, "learning_rate": 9.572413303013076e-06, "loss": 1.0609, "step": 12534 }, { "epoch": 1.3337944243456055, "grad_norm": 15.453125, "learning_rate": 9.572377771461058e-06, "loss": 1.1271, "step": 12535 }, { "epoch": 1.333900829963822, "grad_norm": 10.59375, "learning_rate": 9.572342239909041e-06, "loss": 1.1163, "step": 12536 }, { "epoch": 1.3340072355820387, "grad_norm": 12.2265625, "learning_rate": 9.572306708357023e-06, "loss": 1.1819, "step": 12537 }, { "epoch": 1.3341136412002554, "grad_norm": 11.046875, "learning_rate": 9.572271176805004e-06, "loss": 0.8338, "step": 12538 }, { "epoch": 1.334220046818472, "grad_norm": 13.5859375, "learning_rate": 9.572235645252986e-06, "loss": 0.8692, "step": 12539 }, { "epoch": 1.3343264524366887, "grad_norm": 13.6328125, "learning_rate": 9.572200113700968e-06, "loss": 0.9978, "step": 12540 }, { "epoch": 1.3344328580549054, "grad_norm": 11.21875, "learning_rate": 9.572164582148949e-06, "loss": 1.1808, "step": 12541 }, { "epoch": 1.334539263673122, "grad_norm": 14.8203125, "learning_rate": 9.57212905059693e-06, "loss": 1.1482, "step": 12542 }, { "epoch": 1.3346456692913387, "grad_norm": 15.3203125, "learning_rate": 9.572093519044912e-06, "loss": 1.1694, "step": 12543 }, { "epoch": 1.3347520749095552, "grad_norm": 14.0546875, "learning_rate": 9.572057987492894e-06, "loss": 1.1562, "step": 12544 }, { "epoch": 1.334858480527772, "grad_norm": 12.390625, "learning_rate": 9.572022455940876e-06, "loss": 0.4912, "step": 12545 }, { "epoch": 1.3349648861459884, "grad_norm": 12.171875, "learning_rate": 9.571986924388859e-06, "loss": 1.0722, "step": 12546 }, { "epoch": 1.3350712917642051, "grad_norm": 13.5546875, "learning_rate": 9.57195139283684e-06, "loss": 1.1089, "step": 12547 }, { "epoch": 1.3351776973824219, "grad_norm": 16.453125, "learning_rate": 9.571915861284822e-06, "loss": 1.2152, "step": 12548 }, { "epoch": 1.3352841030006384, "grad_norm": 13.234375, "learning_rate": 9.571880329732804e-06, "loss": 0.8082, "step": 12549 }, { "epoch": 1.335390508618855, "grad_norm": 10.53125, "learning_rate": 9.571844798180785e-06, "loss": 0.9891, "step": 12550 }, { "epoch": 1.3354969142370718, "grad_norm": 17.0625, "learning_rate": 9.571809266628767e-06, "loss": 1.2749, "step": 12551 }, { "epoch": 1.3356033198552884, "grad_norm": 18.65625, "learning_rate": 9.571773735076748e-06, "loss": 0.8834, "step": 12552 }, { "epoch": 1.335709725473505, "grad_norm": 12.1953125, "learning_rate": 9.57173820352473e-06, "loss": 1.3107, "step": 12553 }, { "epoch": 1.3358161310917216, "grad_norm": 19.03125, "learning_rate": 9.571702671972712e-06, "loss": 1.1458, "step": 12554 }, { "epoch": 1.3359225367099383, "grad_norm": 10.359375, "learning_rate": 9.571667140420693e-06, "loss": 1.0927, "step": 12555 }, { "epoch": 1.3360289423281548, "grad_norm": 11.0859375, "learning_rate": 9.571631608868677e-06, "loss": 1.3364, "step": 12556 }, { "epoch": 1.3361353479463716, "grad_norm": 12.359375, "learning_rate": 9.571596077316658e-06, "loss": 1.0625, "step": 12557 }, { "epoch": 1.3362417535645883, "grad_norm": 12.328125, "learning_rate": 9.57156054576464e-06, "loss": 1.06, "step": 12558 }, { "epoch": 1.3363481591828048, "grad_norm": 11.4140625, "learning_rate": 9.571525014212621e-06, "loss": 0.9853, "step": 12559 }, { "epoch": 1.3364545648010215, "grad_norm": 11.6015625, "learning_rate": 9.571489482660603e-06, "loss": 1.133, "step": 12560 }, { "epoch": 1.3365609704192383, "grad_norm": 14.125, "learning_rate": 9.571453951108585e-06, "loss": 1.2676, "step": 12561 }, { "epoch": 1.3366673760374548, "grad_norm": 17.21875, "learning_rate": 9.571418419556568e-06, "loss": 0.7913, "step": 12562 }, { "epoch": 1.3367737816556713, "grad_norm": 10.765625, "learning_rate": 9.57138288800455e-06, "loss": 1.0544, "step": 12563 }, { "epoch": 1.336880187273888, "grad_norm": 10.015625, "learning_rate": 9.571347356452531e-06, "loss": 1.1855, "step": 12564 }, { "epoch": 1.3369865928921048, "grad_norm": 10.09375, "learning_rate": 9.571311824900513e-06, "loss": 0.8908, "step": 12565 }, { "epoch": 1.3370929985103213, "grad_norm": 9.9921875, "learning_rate": 9.571276293348494e-06, "loss": 1.0628, "step": 12566 }, { "epoch": 1.337199404128538, "grad_norm": 14.453125, "learning_rate": 9.571240761796476e-06, "loss": 1.4291, "step": 12567 }, { "epoch": 1.3373058097467547, "grad_norm": 11.53125, "learning_rate": 9.571205230244458e-06, "loss": 0.8753, "step": 12568 }, { "epoch": 1.3374122153649712, "grad_norm": 20.765625, "learning_rate": 9.57116969869244e-06, "loss": 0.9648, "step": 12569 }, { "epoch": 1.337518620983188, "grad_norm": 27.4375, "learning_rate": 9.57113416714042e-06, "loss": 1.3498, "step": 12570 }, { "epoch": 1.3376250266014045, "grad_norm": 15.21875, "learning_rate": 9.571098635588402e-06, "loss": 1.1261, "step": 12571 }, { "epoch": 1.3377314322196212, "grad_norm": 12.2890625, "learning_rate": 9.571063104036386e-06, "loss": 0.8749, "step": 12572 }, { "epoch": 1.3378378378378377, "grad_norm": 10.1875, "learning_rate": 9.571027572484367e-06, "loss": 1.0236, "step": 12573 }, { "epoch": 1.3379442434560544, "grad_norm": 13.5390625, "learning_rate": 9.570992040932349e-06, "loss": 1.089, "step": 12574 }, { "epoch": 1.3380506490742712, "grad_norm": 11.640625, "learning_rate": 9.57095650938033e-06, "loss": 1.0182, "step": 12575 }, { "epoch": 1.3381570546924877, "grad_norm": 12.328125, "learning_rate": 9.570920977828312e-06, "loss": 1.0391, "step": 12576 }, { "epoch": 1.3382634603107044, "grad_norm": 13.640625, "learning_rate": 9.570885446276294e-06, "loss": 1.0884, "step": 12577 }, { "epoch": 1.3383698659289212, "grad_norm": 11.140625, "learning_rate": 9.570849914724275e-06, "loss": 1.1494, "step": 12578 }, { "epoch": 1.3384762715471377, "grad_norm": 12.890625, "learning_rate": 9.570814383172257e-06, "loss": 1.1421, "step": 12579 }, { "epoch": 1.3385826771653544, "grad_norm": 13.015625, "learning_rate": 9.570778851620239e-06, "loss": 0.9415, "step": 12580 }, { "epoch": 1.338689082783571, "grad_norm": 12.21875, "learning_rate": 9.57074332006822e-06, "loss": 0.8496, "step": 12581 }, { "epoch": 1.3387954884017876, "grad_norm": 13.5, "learning_rate": 9.570707788516203e-06, "loss": 1.1355, "step": 12582 }, { "epoch": 1.3389018940200041, "grad_norm": 9.859375, "learning_rate": 9.570672256964185e-06, "loss": 0.9583, "step": 12583 }, { "epoch": 1.3390082996382209, "grad_norm": 14.6015625, "learning_rate": 9.570636725412167e-06, "loss": 1.1782, "step": 12584 }, { "epoch": 1.3391147052564376, "grad_norm": 12.6171875, "learning_rate": 9.570601193860148e-06, "loss": 1.039, "step": 12585 }, { "epoch": 1.3392211108746541, "grad_norm": 14.5234375, "learning_rate": 9.57056566230813e-06, "loss": 1.4742, "step": 12586 }, { "epoch": 1.3393275164928709, "grad_norm": 20.1875, "learning_rate": 9.570530130756113e-06, "loss": 1.6614, "step": 12587 }, { "epoch": 1.3394339221110876, "grad_norm": 14.2734375, "learning_rate": 9.570494599204095e-06, "loss": 1.0867, "step": 12588 }, { "epoch": 1.339540327729304, "grad_norm": 11.6953125, "learning_rate": 9.570459067652076e-06, "loss": 0.9765, "step": 12589 }, { "epoch": 1.3396467333475208, "grad_norm": 14.21875, "learning_rate": 9.570423536100058e-06, "loss": 1.4069, "step": 12590 }, { "epoch": 1.3397531389657373, "grad_norm": 12.7578125, "learning_rate": 9.57038800454804e-06, "loss": 1.4742, "step": 12591 }, { "epoch": 1.339859544583954, "grad_norm": 13.0703125, "learning_rate": 9.570352472996021e-06, "loss": 1.1296, "step": 12592 }, { "epoch": 1.3399659502021706, "grad_norm": 16.21875, "learning_rate": 9.570316941444003e-06, "loss": 0.9503, "step": 12593 }, { "epoch": 1.3400723558203873, "grad_norm": 11.5546875, "learning_rate": 9.570281409891984e-06, "loss": 1.0326, "step": 12594 }, { "epoch": 1.340178761438604, "grad_norm": 18.640625, "learning_rate": 9.570245878339966e-06, "loss": 1.2774, "step": 12595 }, { "epoch": 1.3402851670568205, "grad_norm": 13.6015625, "learning_rate": 9.570210346787948e-06, "loss": 1.2314, "step": 12596 }, { "epoch": 1.3403915726750373, "grad_norm": 10.8515625, "learning_rate": 9.570174815235931e-06, "loss": 1.0104, "step": 12597 }, { "epoch": 1.340497978293254, "grad_norm": 13.484375, "learning_rate": 9.570139283683913e-06, "loss": 1.0862, "step": 12598 }, { "epoch": 1.3406043839114705, "grad_norm": 16.265625, "learning_rate": 9.570103752131894e-06, "loss": 1.1739, "step": 12599 }, { "epoch": 1.340710789529687, "grad_norm": 19.140625, "learning_rate": 9.570068220579876e-06, "loss": 1.0748, "step": 12600 }, { "epoch": 1.3408171951479038, "grad_norm": 10.6015625, "learning_rate": 9.570032689027857e-06, "loss": 0.9921, "step": 12601 }, { "epoch": 1.3409236007661205, "grad_norm": 10.78125, "learning_rate": 9.569997157475839e-06, "loss": 1.1854, "step": 12602 }, { "epoch": 1.341030006384337, "grad_norm": 11.546875, "learning_rate": 9.569961625923822e-06, "loss": 1.1888, "step": 12603 }, { "epoch": 1.3411364120025537, "grad_norm": 12.6796875, "learning_rate": 9.569926094371804e-06, "loss": 1.3952, "step": 12604 }, { "epoch": 1.3412428176207705, "grad_norm": 15.171875, "learning_rate": 9.569890562819786e-06, "loss": 1.0288, "step": 12605 }, { "epoch": 1.341349223238987, "grad_norm": 10.8125, "learning_rate": 9.569855031267767e-06, "loss": 0.8954, "step": 12606 }, { "epoch": 1.3414556288572037, "grad_norm": 12.7109375, "learning_rate": 9.569819499715747e-06, "loss": 1.208, "step": 12607 }, { "epoch": 1.3415620344754204, "grad_norm": 17.984375, "learning_rate": 9.56978396816373e-06, "loss": 1.1242, "step": 12608 }, { "epoch": 1.341668440093637, "grad_norm": 10.828125, "learning_rate": 9.569748436611712e-06, "loss": 0.7548, "step": 12609 }, { "epoch": 1.3417748457118535, "grad_norm": 19.1875, "learning_rate": 9.569712905059694e-06, "loss": 1.3748, "step": 12610 }, { "epoch": 1.3418812513300702, "grad_norm": 12.8203125, "learning_rate": 9.569677373507675e-06, "loss": 1.1623, "step": 12611 }, { "epoch": 1.341987656948287, "grad_norm": 12.7265625, "learning_rate": 9.569641841955657e-06, "loss": 1.282, "step": 12612 }, { "epoch": 1.3420940625665034, "grad_norm": 11.828125, "learning_rate": 9.56960631040364e-06, "loss": 0.9565, "step": 12613 }, { "epoch": 1.3422004681847202, "grad_norm": 11.84375, "learning_rate": 9.569570778851622e-06, "loss": 0.9518, "step": 12614 }, { "epoch": 1.342306873802937, "grad_norm": 13.984375, "learning_rate": 9.569535247299603e-06, "loss": 1.0015, "step": 12615 }, { "epoch": 1.3424132794211534, "grad_norm": 11.9375, "learning_rate": 9.569499715747585e-06, "loss": 1.1836, "step": 12616 }, { "epoch": 1.3425196850393701, "grad_norm": 12.2109375, "learning_rate": 9.569464184195566e-06, "loss": 0.9653, "step": 12617 }, { "epoch": 1.3426260906575866, "grad_norm": 13.1640625, "learning_rate": 9.569428652643548e-06, "loss": 1.5173, "step": 12618 }, { "epoch": 1.3427324962758034, "grad_norm": 14.828125, "learning_rate": 9.56939312109153e-06, "loss": 0.9549, "step": 12619 }, { "epoch": 1.3428389018940199, "grad_norm": 12.5, "learning_rate": 9.569357589539511e-06, "loss": 0.9869, "step": 12620 }, { "epoch": 1.3429453075122366, "grad_norm": 11.28125, "learning_rate": 9.569322057987493e-06, "loss": 1.0406, "step": 12621 }, { "epoch": 1.3430517131304534, "grad_norm": 11.265625, "learning_rate": 9.569286526435475e-06, "loss": 1.2766, "step": 12622 }, { "epoch": 1.3431581187486699, "grad_norm": 12.1171875, "learning_rate": 9.569250994883458e-06, "loss": 0.9223, "step": 12623 }, { "epoch": 1.3432645243668866, "grad_norm": 9.2109375, "learning_rate": 9.56921546333144e-06, "loss": 0.9083, "step": 12624 }, { "epoch": 1.3433709299851033, "grad_norm": 21.328125, "learning_rate": 9.569179931779421e-06, "loss": 1.624, "step": 12625 }, { "epoch": 1.3434773356033198, "grad_norm": 15.03125, "learning_rate": 9.569144400227403e-06, "loss": 1.0618, "step": 12626 }, { "epoch": 1.3435837412215366, "grad_norm": 10.1875, "learning_rate": 9.569108868675384e-06, "loss": 1.1194, "step": 12627 }, { "epoch": 1.343690146839753, "grad_norm": 10.1484375, "learning_rate": 9.569073337123368e-06, "loss": 1.1007, "step": 12628 }, { "epoch": 1.3437965524579698, "grad_norm": 10.5625, "learning_rate": 9.569037805571349e-06, "loss": 0.9451, "step": 12629 }, { "epoch": 1.3439029580761863, "grad_norm": 10.53125, "learning_rate": 9.56900227401933e-06, "loss": 0.9682, "step": 12630 }, { "epoch": 1.344009363694403, "grad_norm": 10.0703125, "learning_rate": 9.568966742467312e-06, "loss": 0.7927, "step": 12631 }, { "epoch": 1.3441157693126198, "grad_norm": 11.84375, "learning_rate": 9.568931210915294e-06, "loss": 1.2372, "step": 12632 }, { "epoch": 1.3442221749308363, "grad_norm": 11.703125, "learning_rate": 9.568895679363276e-06, "loss": 0.9027, "step": 12633 }, { "epoch": 1.344328580549053, "grad_norm": 12.0078125, "learning_rate": 9.568860147811257e-06, "loss": 0.9379, "step": 12634 }, { "epoch": 1.3444349861672698, "grad_norm": 12.4921875, "learning_rate": 9.568824616259239e-06, "loss": 1.3859, "step": 12635 }, { "epoch": 1.3445413917854863, "grad_norm": 17.625, "learning_rate": 9.56878908470722e-06, "loss": 0.8733, "step": 12636 }, { "epoch": 1.344647797403703, "grad_norm": 12.703125, "learning_rate": 9.568753553155202e-06, "loss": 0.8812, "step": 12637 }, { "epoch": 1.3447542030219195, "grad_norm": 9.9765625, "learning_rate": 9.568718021603184e-06, "loss": 1.0242, "step": 12638 }, { "epoch": 1.3448606086401362, "grad_norm": 20.9375, "learning_rate": 9.568682490051167e-06, "loss": 1.0014, "step": 12639 }, { "epoch": 1.3449670142583527, "grad_norm": 11.3203125, "learning_rate": 9.568646958499149e-06, "loss": 1.0854, "step": 12640 }, { "epoch": 1.3450734198765695, "grad_norm": 12.890625, "learning_rate": 9.56861142694713e-06, "loss": 1.198, "step": 12641 }, { "epoch": 1.3451798254947862, "grad_norm": 13.0078125, "learning_rate": 9.568575895395112e-06, "loss": 1.0372, "step": 12642 }, { "epoch": 1.3452862311130027, "grad_norm": 16.5625, "learning_rate": 9.568540363843093e-06, "loss": 1.0254, "step": 12643 }, { "epoch": 1.3453926367312194, "grad_norm": 12.3125, "learning_rate": 9.568504832291075e-06, "loss": 1.157, "step": 12644 }, { "epoch": 1.3454990423494362, "grad_norm": 17.78125, "learning_rate": 9.568469300739057e-06, "loss": 1.5917, "step": 12645 }, { "epoch": 1.3456054479676527, "grad_norm": 14.890625, "learning_rate": 9.568433769187038e-06, "loss": 1.0697, "step": 12646 }, { "epoch": 1.3457118535858692, "grad_norm": 15.984375, "learning_rate": 9.56839823763502e-06, "loss": 1.1887, "step": 12647 }, { "epoch": 1.345818259204086, "grad_norm": 11.8125, "learning_rate": 9.568362706083001e-06, "loss": 0.9618, "step": 12648 }, { "epoch": 1.3459246648223027, "grad_norm": 12.921875, "learning_rate": 9.568327174530985e-06, "loss": 0.8825, "step": 12649 }, { "epoch": 1.3460310704405192, "grad_norm": 14.234375, "learning_rate": 9.568291642978966e-06, "loss": 1.1358, "step": 12650 }, { "epoch": 1.346137476058736, "grad_norm": 11.5703125, "learning_rate": 9.568256111426948e-06, "loss": 1.0426, "step": 12651 }, { "epoch": 1.3462438816769526, "grad_norm": 14.671875, "learning_rate": 9.56822057987493e-06, "loss": 1.1717, "step": 12652 }, { "epoch": 1.3463502872951691, "grad_norm": 9.0546875, "learning_rate": 9.568185048322911e-06, "loss": 0.9193, "step": 12653 }, { "epoch": 1.3464566929133859, "grad_norm": 11.6953125, "learning_rate": 9.568149516770894e-06, "loss": 1.0182, "step": 12654 }, { "epoch": 1.3465630985316024, "grad_norm": 10.6796875, "learning_rate": 9.568113985218876e-06, "loss": 1.1103, "step": 12655 }, { "epoch": 1.3466695041498191, "grad_norm": 12.0, "learning_rate": 9.568078453666858e-06, "loss": 0.9397, "step": 12656 }, { "epoch": 1.3467759097680356, "grad_norm": 14.8203125, "learning_rate": 9.56804292211484e-06, "loss": 1.4102, "step": 12657 }, { "epoch": 1.3468823153862524, "grad_norm": 16.328125, "learning_rate": 9.56800739056282e-06, "loss": 1.0697, "step": 12658 }, { "epoch": 1.346988721004469, "grad_norm": 11.0390625, "learning_rate": 9.567971859010802e-06, "loss": 1.0424, "step": 12659 }, { "epoch": 1.3470951266226856, "grad_norm": 10.28125, "learning_rate": 9.567936327458784e-06, "loss": 0.8698, "step": 12660 }, { "epoch": 1.3472015322409023, "grad_norm": 11.625, "learning_rate": 9.567900795906766e-06, "loss": 1.2211, "step": 12661 }, { "epoch": 1.347307937859119, "grad_norm": 13.21875, "learning_rate": 9.567865264354747e-06, "loss": 1.022, "step": 12662 }, { "epoch": 1.3474143434773356, "grad_norm": 9.9375, "learning_rate": 9.567829732802729e-06, "loss": 1.0249, "step": 12663 }, { "epoch": 1.3475207490955523, "grad_norm": 14.59375, "learning_rate": 9.567794201250712e-06, "loss": 0.9599, "step": 12664 }, { "epoch": 1.3476271547137688, "grad_norm": 14.8515625, "learning_rate": 9.567758669698694e-06, "loss": 1.6641, "step": 12665 }, { "epoch": 1.3477335603319855, "grad_norm": 13.3984375, "learning_rate": 9.567723138146675e-06, "loss": 0.9094, "step": 12666 }, { "epoch": 1.347839965950202, "grad_norm": 13.0546875, "learning_rate": 9.567687606594657e-06, "loss": 1.1817, "step": 12667 }, { "epoch": 1.3479463715684188, "grad_norm": 9.4921875, "learning_rate": 9.567652075042639e-06, "loss": 1.2171, "step": 12668 }, { "epoch": 1.3480527771866355, "grad_norm": 12.875, "learning_rate": 9.56761654349062e-06, "loss": 1.0915, "step": 12669 }, { "epoch": 1.348159182804852, "grad_norm": 19.390625, "learning_rate": 9.567581011938602e-06, "loss": 0.9313, "step": 12670 }, { "epoch": 1.3482655884230688, "grad_norm": 14.6640625, "learning_rate": 9.567545480386583e-06, "loss": 1.0335, "step": 12671 }, { "epoch": 1.3483719940412855, "grad_norm": 10.859375, "learning_rate": 9.567509948834565e-06, "loss": 0.9022, "step": 12672 }, { "epoch": 1.348478399659502, "grad_norm": 9.6640625, "learning_rate": 9.567474417282547e-06, "loss": 1.0622, "step": 12673 }, { "epoch": 1.3485848052777187, "grad_norm": 9.8515625, "learning_rate": 9.56743888573053e-06, "loss": 1.1671, "step": 12674 }, { "epoch": 1.3486912108959352, "grad_norm": 16.203125, "learning_rate": 9.567403354178512e-06, "loss": 1.097, "step": 12675 }, { "epoch": 1.348797616514152, "grad_norm": 9.71875, "learning_rate": 9.567367822626493e-06, "loss": 0.9268, "step": 12676 }, { "epoch": 1.3489040221323685, "grad_norm": 13.625, "learning_rate": 9.567332291074475e-06, "loss": 1.2178, "step": 12677 }, { "epoch": 1.3490104277505852, "grad_norm": 10.90625, "learning_rate": 9.567296759522456e-06, "loss": 0.9399, "step": 12678 }, { "epoch": 1.349116833368802, "grad_norm": 11.4296875, "learning_rate": 9.567261227970438e-06, "loss": 0.8637, "step": 12679 }, { "epoch": 1.3492232389870185, "grad_norm": 11.828125, "learning_rate": 9.567225696418421e-06, "loss": 0.9216, "step": 12680 }, { "epoch": 1.3493296446052352, "grad_norm": 10.7421875, "learning_rate": 9.567190164866403e-06, "loss": 1.207, "step": 12681 }, { "epoch": 1.349436050223452, "grad_norm": 15.5703125, "learning_rate": 9.567154633314384e-06, "loss": 0.7597, "step": 12682 }, { "epoch": 1.3495424558416684, "grad_norm": 13.2734375, "learning_rate": 9.567119101762366e-06, "loss": 1.3274, "step": 12683 }, { "epoch": 1.3496488614598852, "grad_norm": 11.8125, "learning_rate": 9.567083570210348e-06, "loss": 1.0128, "step": 12684 }, { "epoch": 1.3497552670781017, "grad_norm": 12.1171875, "learning_rate": 9.56704803865833e-06, "loss": 1.0569, "step": 12685 }, { "epoch": 1.3498616726963184, "grad_norm": 8.78125, "learning_rate": 9.567012507106311e-06, "loss": 1.0002, "step": 12686 }, { "epoch": 1.349968078314535, "grad_norm": 16.78125, "learning_rate": 9.566976975554292e-06, "loss": 1.2325, "step": 12687 }, { "epoch": 1.3500744839327516, "grad_norm": 11.7265625, "learning_rate": 9.566941444002274e-06, "loss": 1.3117, "step": 12688 }, { "epoch": 1.3501808895509684, "grad_norm": 10.5, "learning_rate": 9.566905912450256e-06, "loss": 1.0543, "step": 12689 }, { "epoch": 1.3502872951691849, "grad_norm": 13.9765625, "learning_rate": 9.566870380898239e-06, "loss": 1.0209, "step": 12690 }, { "epoch": 1.3503937007874016, "grad_norm": 67.0, "learning_rate": 9.56683484934622e-06, "loss": 0.9667, "step": 12691 }, { "epoch": 1.3505001064056184, "grad_norm": 11.7109375, "learning_rate": 9.566799317794202e-06, "loss": 1.3813, "step": 12692 }, { "epoch": 1.3506065120238349, "grad_norm": 11.0078125, "learning_rate": 9.566763786242184e-06, "loss": 0.9923, "step": 12693 }, { "epoch": 1.3507129176420514, "grad_norm": 14.828125, "learning_rate": 9.566728254690165e-06, "loss": 1.0573, "step": 12694 }, { "epoch": 1.350819323260268, "grad_norm": 12.7578125, "learning_rate": 9.566692723138149e-06, "loss": 1.2092, "step": 12695 }, { "epoch": 1.3509257288784848, "grad_norm": 15.6640625, "learning_rate": 9.566657191586129e-06, "loss": 1.0643, "step": 12696 }, { "epoch": 1.3510321344967013, "grad_norm": 14.1953125, "learning_rate": 9.56662166003411e-06, "loss": 1.1868, "step": 12697 }, { "epoch": 1.351138540114918, "grad_norm": 8.3671875, "learning_rate": 9.566586128482092e-06, "loss": 1.1705, "step": 12698 }, { "epoch": 1.3512449457331348, "grad_norm": 10.8359375, "learning_rate": 9.566550596930073e-06, "loss": 0.9496, "step": 12699 }, { "epoch": 1.3513513513513513, "grad_norm": 15.1015625, "learning_rate": 9.566515065378057e-06, "loss": 0.9275, "step": 12700 }, { "epoch": 1.351457756969568, "grad_norm": 11.375, "learning_rate": 9.566479533826038e-06, "loss": 0.9935, "step": 12701 }, { "epoch": 1.3515641625877846, "grad_norm": 12.5, "learning_rate": 9.56644400227402e-06, "loss": 1.0433, "step": 12702 }, { "epoch": 1.3516705682060013, "grad_norm": 9.546875, "learning_rate": 9.566408470722002e-06, "loss": 0.9312, "step": 12703 }, { "epoch": 1.3517769738242178, "grad_norm": 14.3125, "learning_rate": 9.566372939169983e-06, "loss": 1.3398, "step": 12704 }, { "epoch": 1.3518833794424345, "grad_norm": 13.71875, "learning_rate": 9.566337407617967e-06, "loss": 0.9887, "step": 12705 }, { "epoch": 1.3519897850606513, "grad_norm": 18.296875, "learning_rate": 9.566301876065948e-06, "loss": 1.4385, "step": 12706 }, { "epoch": 1.3520961906788678, "grad_norm": 12.875, "learning_rate": 9.56626634451393e-06, "loss": 1.4533, "step": 12707 }, { "epoch": 1.3522025962970845, "grad_norm": 10.0390625, "learning_rate": 9.566230812961911e-06, "loss": 0.8408, "step": 12708 }, { "epoch": 1.3523090019153012, "grad_norm": 13.328125, "learning_rate": 9.566195281409893e-06, "loss": 1.887, "step": 12709 }, { "epoch": 1.3524154075335177, "grad_norm": 11.0078125, "learning_rate": 9.566159749857875e-06, "loss": 0.8691, "step": 12710 }, { "epoch": 1.3525218131517345, "grad_norm": 12.1328125, "learning_rate": 9.566124218305856e-06, "loss": 1.4578, "step": 12711 }, { "epoch": 1.352628218769951, "grad_norm": 13.359375, "learning_rate": 9.566088686753838e-06, "loss": 1.0588, "step": 12712 }, { "epoch": 1.3527346243881677, "grad_norm": 10.9765625, "learning_rate": 9.56605315520182e-06, "loss": 1.1304, "step": 12713 }, { "epoch": 1.3528410300063842, "grad_norm": 13.171875, "learning_rate": 9.566017623649801e-06, "loss": 1.0559, "step": 12714 }, { "epoch": 1.352947435624601, "grad_norm": 12.375, "learning_rate": 9.565982092097784e-06, "loss": 1.0695, "step": 12715 }, { "epoch": 1.3530538412428177, "grad_norm": 12.0, "learning_rate": 9.565946560545766e-06, "loss": 0.9107, "step": 12716 }, { "epoch": 1.3531602468610342, "grad_norm": 15.8515625, "learning_rate": 9.565911028993747e-06, "loss": 1.0318, "step": 12717 }, { "epoch": 1.353266652479251, "grad_norm": 13.8046875, "learning_rate": 9.565875497441729e-06, "loss": 1.0147, "step": 12718 }, { "epoch": 1.3533730580974677, "grad_norm": 10.828125, "learning_rate": 9.56583996588971e-06, "loss": 1.0664, "step": 12719 }, { "epoch": 1.3534794637156842, "grad_norm": 14.2890625, "learning_rate": 9.565804434337692e-06, "loss": 1.1122, "step": 12720 }, { "epoch": 1.353585869333901, "grad_norm": 11.3828125, "learning_rate": 9.565768902785676e-06, "loss": 1.2666, "step": 12721 }, { "epoch": 1.3536922749521174, "grad_norm": 10.1875, "learning_rate": 9.565733371233657e-06, "loss": 0.9245, "step": 12722 }, { "epoch": 1.3537986805703341, "grad_norm": 13.421875, "learning_rate": 9.565697839681639e-06, "loss": 1.0194, "step": 12723 }, { "epoch": 1.3539050861885507, "grad_norm": 10.078125, "learning_rate": 9.56566230812962e-06, "loss": 1.0037, "step": 12724 }, { "epoch": 1.3540114918067674, "grad_norm": 14.484375, "learning_rate": 9.565626776577602e-06, "loss": 1.0675, "step": 12725 }, { "epoch": 1.3541178974249841, "grad_norm": 14.5390625, "learning_rate": 9.565591245025584e-06, "loss": 1.1779, "step": 12726 }, { "epoch": 1.3542243030432006, "grad_norm": 15.359375, "learning_rate": 9.565555713473565e-06, "loss": 0.8453, "step": 12727 }, { "epoch": 1.3543307086614174, "grad_norm": 8.609375, "learning_rate": 9.565520181921547e-06, "loss": 0.8516, "step": 12728 }, { "epoch": 1.354437114279634, "grad_norm": 15.40625, "learning_rate": 9.565484650369528e-06, "loss": 1.1472, "step": 12729 }, { "epoch": 1.3545435198978506, "grad_norm": 9.5234375, "learning_rate": 9.56544911881751e-06, "loss": 1.1619, "step": 12730 }, { "epoch": 1.354649925516067, "grad_norm": 12.6328125, "learning_rate": 9.565413587265493e-06, "loss": 0.9381, "step": 12731 }, { "epoch": 1.3547563311342838, "grad_norm": 12.421875, "learning_rate": 9.565378055713475e-06, "loss": 1.0301, "step": 12732 }, { "epoch": 1.3548627367525006, "grad_norm": 14.1640625, "learning_rate": 9.565342524161457e-06, "loss": 1.359, "step": 12733 }, { "epoch": 1.354969142370717, "grad_norm": 13.5078125, "learning_rate": 9.565306992609438e-06, "loss": 1.2019, "step": 12734 }, { "epoch": 1.3550755479889338, "grad_norm": 14.015625, "learning_rate": 9.56527146105742e-06, "loss": 1.1847, "step": 12735 }, { "epoch": 1.3551819536071505, "grad_norm": 13.0390625, "learning_rate": 9.565235929505401e-06, "loss": 1.2388, "step": 12736 }, { "epoch": 1.355288359225367, "grad_norm": 15.109375, "learning_rate": 9.565200397953383e-06, "loss": 0.984, "step": 12737 }, { "epoch": 1.3553947648435838, "grad_norm": 16.46875, "learning_rate": 9.565164866401365e-06, "loss": 0.9679, "step": 12738 }, { "epoch": 1.3555011704618005, "grad_norm": 13.421875, "learning_rate": 9.565129334849346e-06, "loss": 1.5452, "step": 12739 }, { "epoch": 1.355607576080017, "grad_norm": 12.6015625, "learning_rate": 9.565093803297328e-06, "loss": 1.0449, "step": 12740 }, { "epoch": 1.3557139816982335, "grad_norm": 9.4375, "learning_rate": 9.565058271745311e-06, "loss": 1.0212, "step": 12741 }, { "epoch": 1.3558203873164503, "grad_norm": 10.640625, "learning_rate": 9.565022740193293e-06, "loss": 1.3354, "step": 12742 }, { "epoch": 1.355926792934667, "grad_norm": 10.359375, "learning_rate": 9.564987208641274e-06, "loss": 0.8013, "step": 12743 }, { "epoch": 1.3560331985528835, "grad_norm": 12.4921875, "learning_rate": 9.564951677089256e-06, "loss": 1.1144, "step": 12744 }, { "epoch": 1.3561396041711002, "grad_norm": 10.0546875, "learning_rate": 9.564916145537238e-06, "loss": 0.9809, "step": 12745 }, { "epoch": 1.356246009789317, "grad_norm": 10.28125, "learning_rate": 9.56488061398522e-06, "loss": 0.8802, "step": 12746 }, { "epoch": 1.3563524154075335, "grad_norm": 10.6640625, "learning_rate": 9.564845082433202e-06, "loss": 0.9949, "step": 12747 }, { "epoch": 1.3564588210257502, "grad_norm": 12.9375, "learning_rate": 9.564809550881184e-06, "loss": 0.925, "step": 12748 }, { "epoch": 1.3565652266439667, "grad_norm": 11.90625, "learning_rate": 9.564774019329166e-06, "loss": 0.9517, "step": 12749 }, { "epoch": 1.3566716322621835, "grad_norm": 15.765625, "learning_rate": 9.564738487777147e-06, "loss": 1.4582, "step": 12750 }, { "epoch": 1.3567780378804, "grad_norm": 8.5078125, "learning_rate": 9.564702956225129e-06, "loss": 0.9331, "step": 12751 }, { "epoch": 1.3568844434986167, "grad_norm": 20.453125, "learning_rate": 9.56466742467311e-06, "loss": 1.441, "step": 12752 }, { "epoch": 1.3569908491168334, "grad_norm": 18.859375, "learning_rate": 9.564631893121092e-06, "loss": 1.1989, "step": 12753 }, { "epoch": 1.35709725473505, "grad_norm": 9.78125, "learning_rate": 9.564596361569074e-06, "loss": 0.9249, "step": 12754 }, { "epoch": 1.3572036603532667, "grad_norm": 11.96875, "learning_rate": 9.564560830017055e-06, "loss": 1.0753, "step": 12755 }, { "epoch": 1.3573100659714834, "grad_norm": 13.7734375, "learning_rate": 9.564525298465037e-06, "loss": 1.0146, "step": 12756 }, { "epoch": 1.3574164715897, "grad_norm": 13.5625, "learning_rate": 9.56448976691302e-06, "loss": 1.1532, "step": 12757 }, { "epoch": 1.3575228772079166, "grad_norm": 11.453125, "learning_rate": 9.564454235361002e-06, "loss": 1.4883, "step": 12758 }, { "epoch": 1.3576292828261332, "grad_norm": 12.03125, "learning_rate": 9.564418703808983e-06, "loss": 1.2796, "step": 12759 }, { "epoch": 1.3577356884443499, "grad_norm": 13.0859375, "learning_rate": 9.564383172256965e-06, "loss": 1.1672, "step": 12760 }, { "epoch": 1.3578420940625664, "grad_norm": 12.1015625, "learning_rate": 9.564347640704947e-06, "loss": 0.9349, "step": 12761 }, { "epoch": 1.3579484996807831, "grad_norm": 13.359375, "learning_rate": 9.564312109152928e-06, "loss": 1.3147, "step": 12762 }, { "epoch": 1.3580549052989999, "grad_norm": 12.8125, "learning_rate": 9.56427657760091e-06, "loss": 1.0188, "step": 12763 }, { "epoch": 1.3581613109172164, "grad_norm": 10.2109375, "learning_rate": 9.564241046048891e-06, "loss": 1.052, "step": 12764 }, { "epoch": 1.358267716535433, "grad_norm": 13.640625, "learning_rate": 9.564205514496873e-06, "loss": 1.311, "step": 12765 }, { "epoch": 1.3583741221536498, "grad_norm": 15.8515625, "learning_rate": 9.564169982944855e-06, "loss": 0.9543, "step": 12766 }, { "epoch": 1.3584805277718663, "grad_norm": 14.5859375, "learning_rate": 9.564134451392838e-06, "loss": 1.577, "step": 12767 }, { "epoch": 1.358586933390083, "grad_norm": 10.328125, "learning_rate": 9.56409891984082e-06, "loss": 0.8885, "step": 12768 }, { "epoch": 1.3586933390082996, "grad_norm": 12.078125, "learning_rate": 9.564063388288801e-06, "loss": 1.2319, "step": 12769 }, { "epoch": 1.3587997446265163, "grad_norm": 13.21875, "learning_rate": 9.564027856736783e-06, "loss": 0.9776, "step": 12770 }, { "epoch": 1.3589061502447328, "grad_norm": 16.125, "learning_rate": 9.563992325184764e-06, "loss": 1.3492, "step": 12771 }, { "epoch": 1.3590125558629496, "grad_norm": 14.2109375, "learning_rate": 9.563956793632748e-06, "loss": 1.1064, "step": 12772 }, { "epoch": 1.3591189614811663, "grad_norm": 13.90625, "learning_rate": 9.56392126208073e-06, "loss": 1.1706, "step": 12773 }, { "epoch": 1.3592253670993828, "grad_norm": 9.1015625, "learning_rate": 9.563885730528711e-06, "loss": 0.97, "step": 12774 }, { "epoch": 1.3593317727175995, "grad_norm": 11.609375, "learning_rate": 9.563850198976693e-06, "loss": 0.8933, "step": 12775 }, { "epoch": 1.3594381783358163, "grad_norm": 10.703125, "learning_rate": 9.563814667424674e-06, "loss": 1.0198, "step": 12776 }, { "epoch": 1.3595445839540328, "grad_norm": 13.5234375, "learning_rate": 9.563779135872656e-06, "loss": 1.0052, "step": 12777 }, { "epoch": 1.3596509895722493, "grad_norm": 13.15625, "learning_rate": 9.563743604320637e-06, "loss": 1.0505, "step": 12778 }, { "epoch": 1.359757395190466, "grad_norm": 11.25, "learning_rate": 9.563708072768619e-06, "loss": 1.2926, "step": 12779 }, { "epoch": 1.3598638008086827, "grad_norm": 13.75, "learning_rate": 9.5636725412166e-06, "loss": 1.0007, "step": 12780 }, { "epoch": 1.3599702064268993, "grad_norm": 9.53125, "learning_rate": 9.563637009664582e-06, "loss": 0.9795, "step": 12781 }, { "epoch": 1.360076612045116, "grad_norm": 14.4921875, "learning_rate": 9.563601478112565e-06, "loss": 0.9133, "step": 12782 }, { "epoch": 1.3601830176633327, "grad_norm": 12.1484375, "learning_rate": 9.563565946560547e-06, "loss": 0.997, "step": 12783 }, { "epoch": 1.3602894232815492, "grad_norm": 14.7109375, "learning_rate": 9.563530415008529e-06, "loss": 1.2686, "step": 12784 }, { "epoch": 1.360395828899766, "grad_norm": 19.359375, "learning_rate": 9.56349488345651e-06, "loss": 1.0662, "step": 12785 }, { "epoch": 1.3605022345179825, "grad_norm": 11.078125, "learning_rate": 9.563459351904492e-06, "loss": 0.8703, "step": 12786 }, { "epoch": 1.3606086401361992, "grad_norm": 14.4375, "learning_rate": 9.563423820352473e-06, "loss": 0.9883, "step": 12787 }, { "epoch": 1.3607150457544157, "grad_norm": 9.8203125, "learning_rate": 9.563388288800455e-06, "loss": 0.8487, "step": 12788 }, { "epoch": 1.3608214513726324, "grad_norm": 15.046875, "learning_rate": 9.563352757248437e-06, "loss": 0.8461, "step": 12789 }, { "epoch": 1.3609278569908492, "grad_norm": 11.8125, "learning_rate": 9.563317225696418e-06, "loss": 0.9816, "step": 12790 }, { "epoch": 1.3610342626090657, "grad_norm": 9.828125, "learning_rate": 9.5632816941444e-06, "loss": 1.0102, "step": 12791 }, { "epoch": 1.3611406682272824, "grad_norm": 19.234375, "learning_rate": 9.563246162592383e-06, "loss": 1.0573, "step": 12792 }, { "epoch": 1.3612470738454991, "grad_norm": 11.828125, "learning_rate": 9.563210631040365e-06, "loss": 0.8959, "step": 12793 }, { "epoch": 1.3613534794637157, "grad_norm": 16.234375, "learning_rate": 9.563175099488346e-06, "loss": 1.3517, "step": 12794 }, { "epoch": 1.3614598850819324, "grad_norm": 10.53125, "learning_rate": 9.563139567936328e-06, "loss": 0.9796, "step": 12795 }, { "epoch": 1.361566290700149, "grad_norm": 11.5546875, "learning_rate": 9.56310403638431e-06, "loss": 1.1793, "step": 12796 }, { "epoch": 1.3616726963183656, "grad_norm": 16.171875, "learning_rate": 9.563068504832291e-06, "loss": 1.4319, "step": 12797 }, { "epoch": 1.3617791019365821, "grad_norm": 14.0, "learning_rate": 9.563032973280275e-06, "loss": 1.0145, "step": 12798 }, { "epoch": 1.3618855075547989, "grad_norm": 16.390625, "learning_rate": 9.562997441728256e-06, "loss": 1.1054, "step": 12799 }, { "epoch": 1.3619919131730156, "grad_norm": 12.3359375, "learning_rate": 9.562961910176238e-06, "loss": 0.9401, "step": 12800 }, { "epoch": 1.362098318791232, "grad_norm": 11.46875, "learning_rate": 9.56292637862422e-06, "loss": 1.0824, "step": 12801 }, { "epoch": 1.3622047244094488, "grad_norm": 12.8828125, "learning_rate": 9.562890847072201e-06, "loss": 0.9644, "step": 12802 }, { "epoch": 1.3623111300276656, "grad_norm": 14.5234375, "learning_rate": 9.562855315520183e-06, "loss": 1.1907, "step": 12803 }, { "epoch": 1.362417535645882, "grad_norm": 10.140625, "learning_rate": 9.562819783968164e-06, "loss": 0.976, "step": 12804 }, { "epoch": 1.3625239412640988, "grad_norm": 10.5078125, "learning_rate": 9.562784252416146e-06, "loss": 1.0615, "step": 12805 }, { "epoch": 1.3626303468823153, "grad_norm": 12.3515625, "learning_rate": 9.562748720864127e-06, "loss": 0.9657, "step": 12806 }, { "epoch": 1.362736752500532, "grad_norm": 10.53125, "learning_rate": 9.562713189312109e-06, "loss": 1.1284, "step": 12807 }, { "epoch": 1.3628431581187486, "grad_norm": 10.65625, "learning_rate": 9.562677657760092e-06, "loss": 0.9757, "step": 12808 }, { "epoch": 1.3629495637369653, "grad_norm": 14.7265625, "learning_rate": 9.562642126208074e-06, "loss": 1.2275, "step": 12809 }, { "epoch": 1.363055969355182, "grad_norm": 12.5078125, "learning_rate": 9.562606594656056e-06, "loss": 1.2728, "step": 12810 }, { "epoch": 1.3631623749733985, "grad_norm": 14.2421875, "learning_rate": 9.562571063104037e-06, "loss": 1.053, "step": 12811 }, { "epoch": 1.3632687805916153, "grad_norm": 16.71875, "learning_rate": 9.562535531552019e-06, "loss": 0.9467, "step": 12812 }, { "epoch": 1.363375186209832, "grad_norm": 11.6953125, "learning_rate": 9.562500000000002e-06, "loss": 1.337, "step": 12813 }, { "epoch": 1.3634815918280485, "grad_norm": 9.53125, "learning_rate": 9.562464468447984e-06, "loss": 0.8456, "step": 12814 }, { "epoch": 1.3635879974462652, "grad_norm": 20.25, "learning_rate": 9.562428936895965e-06, "loss": 1.1842, "step": 12815 }, { "epoch": 1.3636944030644818, "grad_norm": 10.953125, "learning_rate": 9.562393405343945e-06, "loss": 0.9542, "step": 12816 }, { "epoch": 1.3638008086826985, "grad_norm": 13.421875, "learning_rate": 9.562357873791927e-06, "loss": 1.2715, "step": 12817 }, { "epoch": 1.363907214300915, "grad_norm": 10.859375, "learning_rate": 9.56232234223991e-06, "loss": 1.2048, "step": 12818 }, { "epoch": 1.3640136199191317, "grad_norm": 13.859375, "learning_rate": 9.562286810687892e-06, "loss": 1.0435, "step": 12819 }, { "epoch": 1.3641200255373485, "grad_norm": 11.1015625, "learning_rate": 9.562251279135873e-06, "loss": 1.0165, "step": 12820 }, { "epoch": 1.364226431155565, "grad_norm": 11.5703125, "learning_rate": 9.562215747583855e-06, "loss": 1.0473, "step": 12821 }, { "epoch": 1.3643328367737817, "grad_norm": 14.3046875, "learning_rate": 9.562180216031837e-06, "loss": 1.0882, "step": 12822 }, { "epoch": 1.3644392423919984, "grad_norm": 10.6953125, "learning_rate": 9.56214468447982e-06, "loss": 1.122, "step": 12823 }, { "epoch": 1.364545648010215, "grad_norm": 13.03125, "learning_rate": 9.562109152927801e-06, "loss": 1.1429, "step": 12824 }, { "epoch": 1.3646520536284314, "grad_norm": 18.453125, "learning_rate": 9.562073621375783e-06, "loss": 1.0677, "step": 12825 }, { "epoch": 1.3647584592466482, "grad_norm": 9.5234375, "learning_rate": 9.562038089823765e-06, "loss": 0.9943, "step": 12826 }, { "epoch": 1.364864864864865, "grad_norm": 12.625, "learning_rate": 9.562002558271746e-06, "loss": 1.1815, "step": 12827 }, { "epoch": 1.3649712704830814, "grad_norm": 12.3046875, "learning_rate": 9.561967026719728e-06, "loss": 1.0347, "step": 12828 }, { "epoch": 1.3650776761012982, "grad_norm": 10.2109375, "learning_rate": 9.56193149516771e-06, "loss": 1.0015, "step": 12829 }, { "epoch": 1.3651840817195149, "grad_norm": 13.796875, "learning_rate": 9.561895963615691e-06, "loss": 1.0307, "step": 12830 }, { "epoch": 1.3652904873377314, "grad_norm": 11.4453125, "learning_rate": 9.561860432063673e-06, "loss": 0.7651, "step": 12831 }, { "epoch": 1.3653968929559481, "grad_norm": 11.5625, "learning_rate": 9.561824900511654e-06, "loss": 1.0209, "step": 12832 }, { "epoch": 1.3655032985741646, "grad_norm": 11.0234375, "learning_rate": 9.561789368959636e-06, "loss": 0.9891, "step": 12833 }, { "epoch": 1.3656097041923814, "grad_norm": 10.71875, "learning_rate": 9.56175383740762e-06, "loss": 0.9768, "step": 12834 }, { "epoch": 1.3657161098105979, "grad_norm": 13.625, "learning_rate": 9.5617183058556e-06, "loss": 1.0117, "step": 12835 }, { "epoch": 1.3658225154288146, "grad_norm": 12.71875, "learning_rate": 9.561682774303582e-06, "loss": 1.0566, "step": 12836 }, { "epoch": 1.3659289210470313, "grad_norm": 10.5703125, "learning_rate": 9.561647242751564e-06, "loss": 0.9991, "step": 12837 }, { "epoch": 1.3660353266652479, "grad_norm": 12.21875, "learning_rate": 9.561611711199546e-06, "loss": 1.0949, "step": 12838 }, { "epoch": 1.3661417322834646, "grad_norm": 11.6640625, "learning_rate": 9.561576179647529e-06, "loss": 1.0894, "step": 12839 }, { "epoch": 1.3662481379016813, "grad_norm": 15.734375, "learning_rate": 9.56154064809551e-06, "loss": 0.9389, "step": 12840 }, { "epoch": 1.3663545435198978, "grad_norm": 12.3984375, "learning_rate": 9.561505116543492e-06, "loss": 1.1209, "step": 12841 }, { "epoch": 1.3664609491381146, "grad_norm": 14.8359375, "learning_rate": 9.561469584991474e-06, "loss": 1.4568, "step": 12842 }, { "epoch": 1.366567354756331, "grad_norm": 12.6328125, "learning_rate": 9.561434053439455e-06, "loss": 1.1778, "step": 12843 }, { "epoch": 1.3666737603745478, "grad_norm": 11.84375, "learning_rate": 9.561398521887437e-06, "loss": 0.8528, "step": 12844 }, { "epoch": 1.3667801659927643, "grad_norm": 10.453125, "learning_rate": 9.561362990335419e-06, "loss": 0.7937, "step": 12845 }, { "epoch": 1.366886571610981, "grad_norm": 11.3125, "learning_rate": 9.5613274587834e-06, "loss": 1.3309, "step": 12846 }, { "epoch": 1.3669929772291978, "grad_norm": 17.421875, "learning_rate": 9.561291927231382e-06, "loss": 1.1892, "step": 12847 }, { "epoch": 1.3670993828474143, "grad_norm": 14.9375, "learning_rate": 9.561256395679363e-06, "loss": 1.4233, "step": 12848 }, { "epoch": 1.367205788465631, "grad_norm": 10.1796875, "learning_rate": 9.561220864127347e-06, "loss": 1.0223, "step": 12849 }, { "epoch": 1.3673121940838477, "grad_norm": 11.890625, "learning_rate": 9.561185332575328e-06, "loss": 1.1145, "step": 12850 }, { "epoch": 1.3674185997020643, "grad_norm": 12.3984375, "learning_rate": 9.56114980102331e-06, "loss": 1.2176, "step": 12851 }, { "epoch": 1.367525005320281, "grad_norm": 15.8359375, "learning_rate": 9.561114269471291e-06, "loss": 1.1545, "step": 12852 }, { "epoch": 1.3676314109384975, "grad_norm": 10.2578125, "learning_rate": 9.561078737919273e-06, "loss": 1.0175, "step": 12853 }, { "epoch": 1.3677378165567142, "grad_norm": 10.0859375, "learning_rate": 9.561043206367255e-06, "loss": 0.9379, "step": 12854 }, { "epoch": 1.3678442221749307, "grad_norm": 12.5, "learning_rate": 9.561007674815236e-06, "loss": 1.1073, "step": 12855 }, { "epoch": 1.3679506277931475, "grad_norm": 12.9140625, "learning_rate": 9.560972143263218e-06, "loss": 1.1135, "step": 12856 }, { "epoch": 1.3680570334113642, "grad_norm": 13.34375, "learning_rate": 9.5609366117112e-06, "loss": 1.042, "step": 12857 }, { "epoch": 1.3681634390295807, "grad_norm": 9.8359375, "learning_rate": 9.560901080159181e-06, "loss": 0.9773, "step": 12858 }, { "epoch": 1.3682698446477974, "grad_norm": 13.65625, "learning_rate": 9.560865548607164e-06, "loss": 1.1232, "step": 12859 }, { "epoch": 1.3683762502660142, "grad_norm": 12.734375, "learning_rate": 9.560830017055146e-06, "loss": 1.2478, "step": 12860 }, { "epoch": 1.3684826558842307, "grad_norm": 12.2890625, "learning_rate": 9.560794485503128e-06, "loss": 0.9112, "step": 12861 }, { "epoch": 1.3685890615024472, "grad_norm": 12.484375, "learning_rate": 9.56075895395111e-06, "loss": 1.0803, "step": 12862 }, { "epoch": 1.368695467120664, "grad_norm": 10.984375, "learning_rate": 9.56072342239909e-06, "loss": 1.0923, "step": 12863 }, { "epoch": 1.3688018727388807, "grad_norm": 19.234375, "learning_rate": 9.560687890847074e-06, "loss": 0.886, "step": 12864 }, { "epoch": 1.3689082783570972, "grad_norm": 15.171875, "learning_rate": 9.560652359295056e-06, "loss": 1.1115, "step": 12865 }, { "epoch": 1.369014683975314, "grad_norm": 13.296875, "learning_rate": 9.560616827743037e-06, "loss": 1.0554, "step": 12866 }, { "epoch": 1.3691210895935306, "grad_norm": 13.7421875, "learning_rate": 9.560581296191019e-06, "loss": 1.2432, "step": 12867 }, { "epoch": 1.3692274952117471, "grad_norm": 12.015625, "learning_rate": 9.560545764639e-06, "loss": 0.9824, "step": 12868 }, { "epoch": 1.3693339008299639, "grad_norm": 13.2734375, "learning_rate": 9.560510233086982e-06, "loss": 1.29, "step": 12869 }, { "epoch": 1.3694403064481806, "grad_norm": 12.5859375, "learning_rate": 9.560474701534964e-06, "loss": 1.1598, "step": 12870 }, { "epoch": 1.369546712066397, "grad_norm": 11.2109375, "learning_rate": 9.560439169982945e-06, "loss": 1.024, "step": 12871 }, { "epoch": 1.3696531176846136, "grad_norm": 9.6953125, "learning_rate": 9.560403638430927e-06, "loss": 1.153, "step": 12872 }, { "epoch": 1.3697595233028304, "grad_norm": 16.34375, "learning_rate": 9.560368106878909e-06, "loss": 1.1135, "step": 12873 }, { "epoch": 1.369865928921047, "grad_norm": 12.7109375, "learning_rate": 9.56033257532689e-06, "loss": 1.1569, "step": 12874 }, { "epoch": 1.3699723345392636, "grad_norm": 13.8515625, "learning_rate": 9.560297043774874e-06, "loss": 0.9716, "step": 12875 }, { "epoch": 1.3700787401574803, "grad_norm": 11.0390625, "learning_rate": 9.560261512222855e-06, "loss": 0.8973, "step": 12876 }, { "epoch": 1.370185145775697, "grad_norm": 13.4453125, "learning_rate": 9.560225980670837e-06, "loss": 1.3513, "step": 12877 }, { "epoch": 1.3702915513939136, "grad_norm": 14.234375, "learning_rate": 9.560190449118818e-06, "loss": 1.017, "step": 12878 }, { "epoch": 1.3703979570121303, "grad_norm": 10.8359375, "learning_rate": 9.5601549175668e-06, "loss": 1.2044, "step": 12879 }, { "epoch": 1.3705043626303468, "grad_norm": 13.2890625, "learning_rate": 9.560119386014782e-06, "loss": 1.1314, "step": 12880 }, { "epoch": 1.3706107682485635, "grad_norm": 14.2890625, "learning_rate": 9.560083854462763e-06, "loss": 0.9914, "step": 12881 }, { "epoch": 1.37071717386678, "grad_norm": 10.6875, "learning_rate": 9.560048322910745e-06, "loss": 1.0313, "step": 12882 }, { "epoch": 1.3708235794849968, "grad_norm": 11.59375, "learning_rate": 9.560012791358726e-06, "loss": 1.3174, "step": 12883 }, { "epoch": 1.3709299851032135, "grad_norm": 14.3359375, "learning_rate": 9.559977259806708e-06, "loss": 1.2151, "step": 12884 }, { "epoch": 1.37103639072143, "grad_norm": 15.109375, "learning_rate": 9.559941728254691e-06, "loss": 1.0585, "step": 12885 }, { "epoch": 1.3711427963396468, "grad_norm": 12.859375, "learning_rate": 9.559906196702673e-06, "loss": 1.0261, "step": 12886 }, { "epoch": 1.3712492019578635, "grad_norm": 11.9921875, "learning_rate": 9.559870665150654e-06, "loss": 1.3281, "step": 12887 }, { "epoch": 1.37135560757608, "grad_norm": 10.7109375, "learning_rate": 9.559835133598636e-06, "loss": 0.7466, "step": 12888 }, { "epoch": 1.3714620131942967, "grad_norm": 14.296875, "learning_rate": 9.559799602046618e-06, "loss": 0.9913, "step": 12889 }, { "epoch": 1.3715684188125132, "grad_norm": 11.7578125, "learning_rate": 9.559764070494601e-06, "loss": 1.1914, "step": 12890 }, { "epoch": 1.37167482443073, "grad_norm": 17.75, "learning_rate": 9.559728538942583e-06, "loss": 1.4401, "step": 12891 }, { "epoch": 1.3717812300489465, "grad_norm": 11.046875, "learning_rate": 9.559693007390564e-06, "loss": 1.1738, "step": 12892 }, { "epoch": 1.3718876356671632, "grad_norm": 11.6796875, "learning_rate": 9.559657475838546e-06, "loss": 1.0701, "step": 12893 }, { "epoch": 1.37199404128538, "grad_norm": 10.109375, "learning_rate": 9.559621944286527e-06, "loss": 1.13, "step": 12894 }, { "epoch": 1.3721004469035964, "grad_norm": 11.5, "learning_rate": 9.559586412734509e-06, "loss": 0.9848, "step": 12895 }, { "epoch": 1.3722068525218132, "grad_norm": 10.875, "learning_rate": 9.55955088118249e-06, "loss": 1.0096, "step": 12896 }, { "epoch": 1.37231325814003, "grad_norm": 14.828125, "learning_rate": 9.559515349630472e-06, "loss": 0.9364, "step": 12897 }, { "epoch": 1.3724196637582464, "grad_norm": 12.4921875, "learning_rate": 9.559479818078454e-06, "loss": 0.8658, "step": 12898 }, { "epoch": 1.3725260693764632, "grad_norm": 10.171875, "learning_rate": 9.559444286526435e-06, "loss": 1.0408, "step": 12899 }, { "epoch": 1.3726324749946797, "grad_norm": 17.8125, "learning_rate": 9.559408754974419e-06, "loss": 1.1052, "step": 12900 }, { "epoch": 1.3727388806128964, "grad_norm": 14.5078125, "learning_rate": 9.5593732234224e-06, "loss": 1.2346, "step": 12901 }, { "epoch": 1.372845286231113, "grad_norm": 13.5703125, "learning_rate": 9.559337691870382e-06, "loss": 1.1718, "step": 12902 }, { "epoch": 1.3729516918493296, "grad_norm": 14.71875, "learning_rate": 9.559302160318364e-06, "loss": 1.034, "step": 12903 }, { "epoch": 1.3730580974675464, "grad_norm": 10.4609375, "learning_rate": 9.559266628766345e-06, "loss": 0.7806, "step": 12904 }, { "epoch": 1.3731645030857629, "grad_norm": 16.78125, "learning_rate": 9.559231097214327e-06, "loss": 1.2321, "step": 12905 }, { "epoch": 1.3732709087039796, "grad_norm": 10.6953125, "learning_rate": 9.559195565662308e-06, "loss": 1.0175, "step": 12906 }, { "epoch": 1.3733773143221963, "grad_norm": 10.5078125, "learning_rate": 9.55916003411029e-06, "loss": 1.1269, "step": 12907 }, { "epoch": 1.3734837199404129, "grad_norm": 16.5625, "learning_rate": 9.559124502558272e-06, "loss": 0.9076, "step": 12908 }, { "epoch": 1.3735901255586294, "grad_norm": 15.0234375, "learning_rate": 9.559088971006253e-06, "loss": 1.0675, "step": 12909 }, { "epoch": 1.373696531176846, "grad_norm": 11.125, "learning_rate": 9.559053439454237e-06, "loss": 1.0297, "step": 12910 }, { "epoch": 1.3738029367950628, "grad_norm": 13.0703125, "learning_rate": 9.559017907902218e-06, "loss": 1.1041, "step": 12911 }, { "epoch": 1.3739093424132793, "grad_norm": 11.3046875, "learning_rate": 9.5589823763502e-06, "loss": 1.142, "step": 12912 }, { "epoch": 1.374015748031496, "grad_norm": 14.6484375, "learning_rate": 9.558946844798181e-06, "loss": 1.1595, "step": 12913 }, { "epoch": 1.3741221536497128, "grad_norm": 9.9765625, "learning_rate": 9.558911313246163e-06, "loss": 1.319, "step": 12914 }, { "epoch": 1.3742285592679293, "grad_norm": 14.734375, "learning_rate": 9.558875781694145e-06, "loss": 1.2108, "step": 12915 }, { "epoch": 1.374334964886146, "grad_norm": 18.140625, "learning_rate": 9.558840250142128e-06, "loss": 1.0445, "step": 12916 }, { "epoch": 1.3744413705043625, "grad_norm": 13.234375, "learning_rate": 9.55880471859011e-06, "loss": 1.3403, "step": 12917 }, { "epoch": 1.3745477761225793, "grad_norm": 10.6171875, "learning_rate": 9.558769187038091e-06, "loss": 1.0044, "step": 12918 }, { "epoch": 1.3746541817407958, "grad_norm": 12.375, "learning_rate": 9.558733655486073e-06, "loss": 0.8489, "step": 12919 }, { "epoch": 1.3747605873590125, "grad_norm": 14.78125, "learning_rate": 9.558698123934054e-06, "loss": 1.4674, "step": 12920 }, { "epoch": 1.3748669929772293, "grad_norm": 15.046875, "learning_rate": 9.558662592382036e-06, "loss": 1.2119, "step": 12921 }, { "epoch": 1.3749733985954458, "grad_norm": 16.234375, "learning_rate": 9.558627060830017e-06, "loss": 1.257, "step": 12922 }, { "epoch": 1.3750798042136625, "grad_norm": 11.046875, "learning_rate": 9.558591529277999e-06, "loss": 1.1449, "step": 12923 }, { "epoch": 1.3751862098318792, "grad_norm": 12.390625, "learning_rate": 9.55855599772598e-06, "loss": 1.31, "step": 12924 }, { "epoch": 1.3752926154500957, "grad_norm": 13.5234375, "learning_rate": 9.558520466173962e-06, "loss": 1.3861, "step": 12925 }, { "epoch": 1.3753990210683125, "grad_norm": 19.359375, "learning_rate": 9.558484934621946e-06, "loss": 0.8025, "step": 12926 }, { "epoch": 1.375505426686529, "grad_norm": 12.59375, "learning_rate": 9.558449403069927e-06, "loss": 1.0402, "step": 12927 }, { "epoch": 1.3756118323047457, "grad_norm": 14.890625, "learning_rate": 9.558413871517909e-06, "loss": 1.0778, "step": 12928 }, { "epoch": 1.3757182379229622, "grad_norm": 13.953125, "learning_rate": 9.55837833996589e-06, "loss": 1.1404, "step": 12929 }, { "epoch": 1.375824643541179, "grad_norm": 8.8671875, "learning_rate": 9.558342808413872e-06, "loss": 0.9712, "step": 12930 }, { "epoch": 1.3759310491593957, "grad_norm": 10.5703125, "learning_rate": 9.558307276861855e-06, "loss": 0.9572, "step": 12931 }, { "epoch": 1.3760374547776122, "grad_norm": 12.3203125, "learning_rate": 9.558271745309837e-06, "loss": 1.127, "step": 12932 }, { "epoch": 1.376143860395829, "grad_norm": 12.734375, "learning_rate": 9.558236213757819e-06, "loss": 0.8862, "step": 12933 }, { "epoch": 1.3762502660140457, "grad_norm": 12.6953125, "learning_rate": 9.5582006822058e-06, "loss": 0.967, "step": 12934 }, { "epoch": 1.3763566716322622, "grad_norm": 11.09375, "learning_rate": 9.55816515065378e-06, "loss": 1.0561, "step": 12935 }, { "epoch": 1.376463077250479, "grad_norm": 11.6875, "learning_rate": 9.558129619101763e-06, "loss": 1.2372, "step": 12936 }, { "epoch": 1.3765694828686954, "grad_norm": 13.828125, "learning_rate": 9.558094087549745e-06, "loss": 1.1409, "step": 12937 }, { "epoch": 1.3766758884869121, "grad_norm": 11.890625, "learning_rate": 9.558058555997727e-06, "loss": 1.223, "step": 12938 }, { "epoch": 1.3767822941051286, "grad_norm": 11.109375, "learning_rate": 9.558023024445708e-06, "loss": 1.4742, "step": 12939 }, { "epoch": 1.3768886997233454, "grad_norm": 13.0859375, "learning_rate": 9.55798749289369e-06, "loss": 0.734, "step": 12940 }, { "epoch": 1.376995105341562, "grad_norm": 10.4765625, "learning_rate": 9.557951961341673e-06, "loss": 1.1156, "step": 12941 }, { "epoch": 1.3771015109597786, "grad_norm": 9.7578125, "learning_rate": 9.557916429789655e-06, "loss": 1.2406, "step": 12942 }, { "epoch": 1.3772079165779954, "grad_norm": 13.203125, "learning_rate": 9.557880898237636e-06, "loss": 0.9152, "step": 12943 }, { "epoch": 1.377314322196212, "grad_norm": 16.609375, "learning_rate": 9.557845366685618e-06, "loss": 0.8983, "step": 12944 }, { "epoch": 1.3774207278144286, "grad_norm": 20.578125, "learning_rate": 9.5578098351336e-06, "loss": 0.8667, "step": 12945 }, { "epoch": 1.3775271334326453, "grad_norm": 14.5078125, "learning_rate": 9.557774303581581e-06, "loss": 1.1358, "step": 12946 }, { "epoch": 1.3776335390508618, "grad_norm": 12.515625, "learning_rate": 9.557738772029563e-06, "loss": 1.0469, "step": 12947 }, { "epoch": 1.3777399446690786, "grad_norm": 10.7890625, "learning_rate": 9.557703240477544e-06, "loss": 0.9531, "step": 12948 }, { "epoch": 1.377846350287295, "grad_norm": 15.75, "learning_rate": 9.557667708925526e-06, "loss": 1.503, "step": 12949 }, { "epoch": 1.3779527559055118, "grad_norm": 10.953125, "learning_rate": 9.557632177373508e-06, "loss": 1.1609, "step": 12950 }, { "epoch": 1.3780591615237285, "grad_norm": 9.3671875, "learning_rate": 9.55759664582149e-06, "loss": 1.027, "step": 12951 }, { "epoch": 1.378165567141945, "grad_norm": 16.65625, "learning_rate": 9.557561114269472e-06, "loss": 1.1147, "step": 12952 }, { "epoch": 1.3782719727601618, "grad_norm": 10.6875, "learning_rate": 9.557525582717454e-06, "loss": 1.1137, "step": 12953 }, { "epoch": 1.3783783783783785, "grad_norm": 12.3828125, "learning_rate": 9.557490051165436e-06, "loss": 1.1938, "step": 12954 }, { "epoch": 1.378484783996595, "grad_norm": 10.2265625, "learning_rate": 9.557454519613417e-06, "loss": 1.2755, "step": 12955 }, { "epoch": 1.3785911896148115, "grad_norm": 13.6640625, "learning_rate": 9.557418988061399e-06, "loss": 0.9894, "step": 12956 }, { "epoch": 1.3786975952330283, "grad_norm": 11.7890625, "learning_rate": 9.557383456509382e-06, "loss": 0.9911, "step": 12957 }, { "epoch": 1.378804000851245, "grad_norm": 9.734375, "learning_rate": 9.557347924957364e-06, "loss": 1.1659, "step": 12958 }, { "epoch": 1.3789104064694615, "grad_norm": 12.140625, "learning_rate": 9.557312393405345e-06, "loss": 0.9672, "step": 12959 }, { "epoch": 1.3790168120876782, "grad_norm": 16.125, "learning_rate": 9.557276861853327e-06, "loss": 1.2418, "step": 12960 }, { "epoch": 1.379123217705895, "grad_norm": 15.5625, "learning_rate": 9.557241330301309e-06, "loss": 1.0668, "step": 12961 }, { "epoch": 1.3792296233241115, "grad_norm": 14.609375, "learning_rate": 9.55720579874929e-06, "loss": 1.3261, "step": 12962 }, { "epoch": 1.3793360289423282, "grad_norm": 12.9453125, "learning_rate": 9.557170267197272e-06, "loss": 1.1966, "step": 12963 }, { "epoch": 1.3794424345605447, "grad_norm": 15.359375, "learning_rate": 9.557134735645253e-06, "loss": 1.1808, "step": 12964 }, { "epoch": 1.3795488401787614, "grad_norm": 16.4375, "learning_rate": 9.557099204093235e-06, "loss": 1.0364, "step": 12965 }, { "epoch": 1.379655245796978, "grad_norm": 12.4140625, "learning_rate": 9.557063672541217e-06, "loss": 1.0309, "step": 12966 }, { "epoch": 1.3797616514151947, "grad_norm": 14.4453125, "learning_rate": 9.5570281409892e-06, "loss": 1.2652, "step": 12967 }, { "epoch": 1.3798680570334114, "grad_norm": 30.328125, "learning_rate": 9.556992609437182e-06, "loss": 0.7245, "step": 12968 }, { "epoch": 1.379974462651628, "grad_norm": 12.875, "learning_rate": 9.556957077885163e-06, "loss": 1.1124, "step": 12969 }, { "epoch": 1.3800808682698447, "grad_norm": 11.953125, "learning_rate": 9.556921546333145e-06, "loss": 0.9732, "step": 12970 }, { "epoch": 1.3801872738880614, "grad_norm": 15.078125, "learning_rate": 9.556886014781126e-06, "loss": 1.2934, "step": 12971 }, { "epoch": 1.380293679506278, "grad_norm": 21.328125, "learning_rate": 9.556850483229108e-06, "loss": 1.4594, "step": 12972 }, { "epoch": 1.3804000851244946, "grad_norm": 13.25, "learning_rate": 9.55681495167709e-06, "loss": 1.3108, "step": 12973 }, { "epoch": 1.3805064907427111, "grad_norm": 11.375, "learning_rate": 9.556779420125071e-06, "loss": 1.0331, "step": 12974 }, { "epoch": 1.3806128963609279, "grad_norm": 15.7265625, "learning_rate": 9.556743888573053e-06, "loss": 1.0497, "step": 12975 }, { "epoch": 1.3807193019791444, "grad_norm": 15.4140625, "learning_rate": 9.556708357021034e-06, "loss": 1.1017, "step": 12976 }, { "epoch": 1.3808257075973611, "grad_norm": 16.921875, "learning_rate": 9.556672825469018e-06, "loss": 1.3379, "step": 12977 }, { "epoch": 1.3809321132155778, "grad_norm": 8.1171875, "learning_rate": 9.556637293917e-06, "loss": 0.9975, "step": 12978 }, { "epoch": 1.3810385188337944, "grad_norm": 13.4140625, "learning_rate": 9.556601762364981e-06, "loss": 1.0958, "step": 12979 }, { "epoch": 1.381144924452011, "grad_norm": 10.921875, "learning_rate": 9.556566230812963e-06, "loss": 0.8549, "step": 12980 }, { "epoch": 1.3812513300702278, "grad_norm": 12.171875, "learning_rate": 9.556530699260944e-06, "loss": 0.922, "step": 12981 }, { "epoch": 1.3813577356884443, "grad_norm": 9.8125, "learning_rate": 9.556495167708927e-06, "loss": 1.143, "step": 12982 }, { "epoch": 1.381464141306661, "grad_norm": 14.7578125, "learning_rate": 9.556459636156909e-06, "loss": 1.1812, "step": 12983 }, { "epoch": 1.3815705469248776, "grad_norm": 17.765625, "learning_rate": 9.55642410460489e-06, "loss": 1.3455, "step": 12984 }, { "epoch": 1.3816769525430943, "grad_norm": 11.7421875, "learning_rate": 9.556388573052872e-06, "loss": 1.0172, "step": 12985 }, { "epoch": 1.3817833581613108, "grad_norm": 10.6796875, "learning_rate": 9.556353041500854e-06, "loss": 0.8982, "step": 12986 }, { "epoch": 1.3818897637795275, "grad_norm": 12.0, "learning_rate": 9.556317509948835e-06, "loss": 1.1048, "step": 12987 }, { "epoch": 1.3819961693977443, "grad_norm": 11.5078125, "learning_rate": 9.556281978396817e-06, "loss": 0.9664, "step": 12988 }, { "epoch": 1.3821025750159608, "grad_norm": 13.625, "learning_rate": 9.556246446844799e-06, "loss": 0.8219, "step": 12989 }, { "epoch": 1.3822089806341775, "grad_norm": 12.0390625, "learning_rate": 9.55621091529278e-06, "loss": 1.149, "step": 12990 }, { "epoch": 1.3823153862523943, "grad_norm": 11.203125, "learning_rate": 9.556175383740762e-06, "loss": 1.0999, "step": 12991 }, { "epoch": 1.3824217918706108, "grad_norm": 10.4609375, "learning_rate": 9.556139852188744e-06, "loss": 1.0115, "step": 12992 }, { "epoch": 1.3825281974888275, "grad_norm": 10.21875, "learning_rate": 9.556104320636727e-06, "loss": 1.0585, "step": 12993 }, { "epoch": 1.382634603107044, "grad_norm": 11.109375, "learning_rate": 9.556068789084708e-06, "loss": 1.1056, "step": 12994 }, { "epoch": 1.3827410087252607, "grad_norm": 17.015625, "learning_rate": 9.55603325753269e-06, "loss": 1.0146, "step": 12995 }, { "epoch": 1.3828474143434772, "grad_norm": 11.4453125, "learning_rate": 9.555997725980672e-06, "loss": 1.089, "step": 12996 }, { "epoch": 1.382953819961694, "grad_norm": 14.3359375, "learning_rate": 9.555962194428653e-06, "loss": 1.1536, "step": 12997 }, { "epoch": 1.3830602255799107, "grad_norm": 12.8359375, "learning_rate": 9.555926662876635e-06, "loss": 1.0529, "step": 12998 }, { "epoch": 1.3831666311981272, "grad_norm": 13.296875, "learning_rate": 9.555891131324616e-06, "loss": 1.0547, "step": 12999 }, { "epoch": 1.383273036816344, "grad_norm": 10.875, "learning_rate": 9.555855599772598e-06, "loss": 0.9622, "step": 13000 }, { "epoch": 1.3833794424345607, "grad_norm": 10.671875, "learning_rate": 9.55582006822058e-06, "loss": 1.1558, "step": 13001 }, { "epoch": 1.3834858480527772, "grad_norm": 14.09375, "learning_rate": 9.555784536668561e-06, "loss": 1.0121, "step": 13002 }, { "epoch": 1.3835922536709937, "grad_norm": 12.25, "learning_rate": 9.555749005116545e-06, "loss": 1.0074, "step": 13003 }, { "epoch": 1.3836986592892104, "grad_norm": 15.4375, "learning_rate": 9.555713473564526e-06, "loss": 0.9488, "step": 13004 }, { "epoch": 1.3838050649074272, "grad_norm": 10.0234375, "learning_rate": 9.555677942012508e-06, "loss": 0.9076, "step": 13005 }, { "epoch": 1.3839114705256437, "grad_norm": 13.1171875, "learning_rate": 9.55564241046049e-06, "loss": 1.0968, "step": 13006 }, { "epoch": 1.3840178761438604, "grad_norm": 10.671875, "learning_rate": 9.555606878908471e-06, "loss": 1.0404, "step": 13007 }, { "epoch": 1.3841242817620771, "grad_norm": 18.5625, "learning_rate": 9.555571347356454e-06, "loss": 1.0651, "step": 13008 }, { "epoch": 1.3842306873802936, "grad_norm": 13.015625, "learning_rate": 9.555535815804436e-06, "loss": 1.137, "step": 13009 }, { "epoch": 1.3843370929985104, "grad_norm": 11.71875, "learning_rate": 9.555500284252418e-06, "loss": 0.8494, "step": 13010 }, { "epoch": 1.3844434986167269, "grad_norm": 9.46875, "learning_rate": 9.555464752700399e-06, "loss": 1.103, "step": 13011 }, { "epoch": 1.3845499042349436, "grad_norm": 9.1640625, "learning_rate": 9.55542922114838e-06, "loss": 1.2148, "step": 13012 }, { "epoch": 1.3846563098531601, "grad_norm": 11.828125, "learning_rate": 9.555393689596362e-06, "loss": 1.1675, "step": 13013 }, { "epoch": 1.3847627154713769, "grad_norm": 13.09375, "learning_rate": 9.555358158044344e-06, "loss": 0.9913, "step": 13014 }, { "epoch": 1.3848691210895936, "grad_norm": 16.5, "learning_rate": 9.555322626492326e-06, "loss": 1.2111, "step": 13015 }, { "epoch": 1.38497552670781, "grad_norm": 15.5234375, "learning_rate": 9.555287094940307e-06, "loss": 1.4467, "step": 13016 }, { "epoch": 1.3850819323260268, "grad_norm": 11.9765625, "learning_rate": 9.555251563388289e-06, "loss": 1.2181, "step": 13017 }, { "epoch": 1.3851883379442436, "grad_norm": 14.484375, "learning_rate": 9.555216031836272e-06, "loss": 0.839, "step": 13018 }, { "epoch": 1.38529474356246, "grad_norm": 13.3828125, "learning_rate": 9.555180500284254e-06, "loss": 0.9793, "step": 13019 }, { "epoch": 1.3854011491806768, "grad_norm": 10.3671875, "learning_rate": 9.555144968732235e-06, "loss": 0.9689, "step": 13020 }, { "epoch": 1.3855075547988933, "grad_norm": 15.328125, "learning_rate": 9.555109437180217e-06, "loss": 1.4981, "step": 13021 }, { "epoch": 1.38561396041711, "grad_norm": 9.3984375, "learning_rate": 9.555073905628198e-06, "loss": 0.9836, "step": 13022 }, { "epoch": 1.3857203660353266, "grad_norm": 12.8984375, "learning_rate": 9.555038374076182e-06, "loss": 1.2491, "step": 13023 }, { "epoch": 1.3858267716535433, "grad_norm": 9.9765625, "learning_rate": 9.555002842524162e-06, "loss": 1.0411, "step": 13024 }, { "epoch": 1.38593317727176, "grad_norm": 11.640625, "learning_rate": 9.554967310972143e-06, "loss": 1.1479, "step": 13025 }, { "epoch": 1.3860395828899765, "grad_norm": 12.3828125, "learning_rate": 9.554931779420125e-06, "loss": 1.0626, "step": 13026 }, { "epoch": 1.3861459885081933, "grad_norm": 13.84375, "learning_rate": 9.554896247868107e-06, "loss": 1.2846, "step": 13027 }, { "epoch": 1.38625239412641, "grad_norm": 12.296875, "learning_rate": 9.554860716316088e-06, "loss": 1.0254, "step": 13028 }, { "epoch": 1.3863587997446265, "grad_norm": 9.921875, "learning_rate": 9.554825184764071e-06, "loss": 1.156, "step": 13029 }, { "epoch": 1.3864652053628432, "grad_norm": 11.1640625, "learning_rate": 9.554789653212053e-06, "loss": 1.129, "step": 13030 }, { "epoch": 1.3865716109810597, "grad_norm": 16.921875, "learning_rate": 9.554754121660035e-06, "loss": 0.8292, "step": 13031 }, { "epoch": 1.3866780165992765, "grad_norm": 11.0078125, "learning_rate": 9.554718590108016e-06, "loss": 1.0433, "step": 13032 }, { "epoch": 1.386784422217493, "grad_norm": 15.0625, "learning_rate": 9.554683058555998e-06, "loss": 1.0065, "step": 13033 }, { "epoch": 1.3868908278357097, "grad_norm": 14.7109375, "learning_rate": 9.554647527003981e-06, "loss": 1.0284, "step": 13034 }, { "epoch": 1.3869972334539264, "grad_norm": 15.0859375, "learning_rate": 9.554611995451963e-06, "loss": 1.2496, "step": 13035 }, { "epoch": 1.387103639072143, "grad_norm": 10.765625, "learning_rate": 9.554576463899944e-06, "loss": 1.0429, "step": 13036 }, { "epoch": 1.3872100446903597, "grad_norm": 10.84375, "learning_rate": 9.554540932347926e-06, "loss": 0.8744, "step": 13037 }, { "epoch": 1.3873164503085764, "grad_norm": 13.28125, "learning_rate": 9.554505400795908e-06, "loss": 0.9565, "step": 13038 }, { "epoch": 1.387422855926793, "grad_norm": 10.5, "learning_rate": 9.55446986924389e-06, "loss": 0.866, "step": 13039 }, { "epoch": 1.3875292615450094, "grad_norm": 10.0859375, "learning_rate": 9.55443433769187e-06, "loss": 0.9135, "step": 13040 }, { "epoch": 1.3876356671632262, "grad_norm": 12.8984375, "learning_rate": 9.554398806139852e-06, "loss": 1.0981, "step": 13041 }, { "epoch": 1.387742072781443, "grad_norm": 16.953125, "learning_rate": 9.554363274587834e-06, "loss": 1.0804, "step": 13042 }, { "epoch": 1.3878484783996594, "grad_norm": 11.4765625, "learning_rate": 9.554327743035816e-06, "loss": 1.1527, "step": 13043 }, { "epoch": 1.3879548840178761, "grad_norm": 11.640625, "learning_rate": 9.554292211483799e-06, "loss": 1.1578, "step": 13044 }, { "epoch": 1.3880612896360929, "grad_norm": 9.5078125, "learning_rate": 9.55425667993178e-06, "loss": 1.1036, "step": 13045 }, { "epoch": 1.3881676952543094, "grad_norm": 11.71875, "learning_rate": 9.554221148379762e-06, "loss": 0.9392, "step": 13046 }, { "epoch": 1.3882741008725261, "grad_norm": 14.8984375, "learning_rate": 9.554185616827744e-06, "loss": 1.1807, "step": 13047 }, { "epoch": 1.3883805064907428, "grad_norm": 9.5, "learning_rate": 9.554150085275725e-06, "loss": 0.9697, "step": 13048 }, { "epoch": 1.3884869121089594, "grad_norm": 17.328125, "learning_rate": 9.554114553723709e-06, "loss": 0.8214, "step": 13049 }, { "epoch": 1.3885933177271759, "grad_norm": 12.0, "learning_rate": 9.55407902217169e-06, "loss": 0.9658, "step": 13050 }, { "epoch": 1.3886997233453926, "grad_norm": 10.7890625, "learning_rate": 9.554043490619672e-06, "loss": 1.1707, "step": 13051 }, { "epoch": 1.3888061289636093, "grad_norm": 16.0, "learning_rate": 9.554007959067653e-06, "loss": 1.044, "step": 13052 }, { "epoch": 1.3889125345818258, "grad_norm": 12.984375, "learning_rate": 9.553972427515635e-06, "loss": 1.4241, "step": 13053 }, { "epoch": 1.3890189402000426, "grad_norm": 15.9921875, "learning_rate": 9.553936895963617e-06, "loss": 1.2746, "step": 13054 }, { "epoch": 1.3891253458182593, "grad_norm": 10.3125, "learning_rate": 9.553901364411598e-06, "loss": 1.2826, "step": 13055 }, { "epoch": 1.3892317514364758, "grad_norm": 12.3359375, "learning_rate": 9.55386583285958e-06, "loss": 1.1597, "step": 13056 }, { "epoch": 1.3893381570546925, "grad_norm": 10.9140625, "learning_rate": 9.553830301307562e-06, "loss": 1.1803, "step": 13057 }, { "epoch": 1.389444562672909, "grad_norm": 14.5546875, "learning_rate": 9.553794769755543e-06, "loss": 1.0224, "step": 13058 }, { "epoch": 1.3895509682911258, "grad_norm": 11.171875, "learning_rate": 9.553759238203526e-06, "loss": 0.98, "step": 13059 }, { "epoch": 1.3896573739093423, "grad_norm": 11.140625, "learning_rate": 9.553723706651508e-06, "loss": 0.764, "step": 13060 }, { "epoch": 1.389763779527559, "grad_norm": 11.953125, "learning_rate": 9.55368817509949e-06, "loss": 0.9684, "step": 13061 }, { "epoch": 1.3898701851457758, "grad_norm": 11.65625, "learning_rate": 9.553652643547471e-06, "loss": 1.1378, "step": 13062 }, { "epoch": 1.3899765907639923, "grad_norm": 10.328125, "learning_rate": 9.553617111995453e-06, "loss": 1.2124, "step": 13063 }, { "epoch": 1.390082996382209, "grad_norm": 13.1328125, "learning_rate": 9.553581580443434e-06, "loss": 1.1005, "step": 13064 }, { "epoch": 1.3901894020004257, "grad_norm": 11.9375, "learning_rate": 9.553546048891416e-06, "loss": 1.0721, "step": 13065 }, { "epoch": 1.3902958076186422, "grad_norm": 11.4765625, "learning_rate": 9.553510517339398e-06, "loss": 0.9078, "step": 13066 }, { "epoch": 1.390402213236859, "grad_norm": 14.984375, "learning_rate": 9.55347498578738e-06, "loss": 1.0974, "step": 13067 }, { "epoch": 1.3905086188550755, "grad_norm": 12.2109375, "learning_rate": 9.553439454235361e-06, "loss": 1.1839, "step": 13068 }, { "epoch": 1.3906150244732922, "grad_norm": 9.3125, "learning_rate": 9.553403922683342e-06, "loss": 1.1083, "step": 13069 }, { "epoch": 1.3907214300915087, "grad_norm": 13.9140625, "learning_rate": 9.553368391131326e-06, "loss": 0.9243, "step": 13070 }, { "epoch": 1.3908278357097255, "grad_norm": 18.0625, "learning_rate": 9.553332859579307e-06, "loss": 1.2754, "step": 13071 }, { "epoch": 1.3909342413279422, "grad_norm": 11.7109375, "learning_rate": 9.553297328027289e-06, "loss": 0.8892, "step": 13072 }, { "epoch": 1.3910406469461587, "grad_norm": 11.890625, "learning_rate": 9.55326179647527e-06, "loss": 1.0147, "step": 13073 }, { "epoch": 1.3911470525643754, "grad_norm": 14.7890625, "learning_rate": 9.553226264923252e-06, "loss": 1.1505, "step": 13074 }, { "epoch": 1.3912534581825922, "grad_norm": 17.53125, "learning_rate": 9.553190733371236e-06, "loss": 1.1383, "step": 13075 }, { "epoch": 1.3913598638008087, "grad_norm": 11.34375, "learning_rate": 9.553155201819217e-06, "loss": 0.9947, "step": 13076 }, { "epoch": 1.3914662694190254, "grad_norm": 21.25, "learning_rate": 9.553119670267199e-06, "loss": 1.1172, "step": 13077 }, { "epoch": 1.391572675037242, "grad_norm": 8.9375, "learning_rate": 9.55308413871518e-06, "loss": 1.0441, "step": 13078 }, { "epoch": 1.3916790806554586, "grad_norm": 10.7734375, "learning_rate": 9.553048607163162e-06, "loss": 1.0825, "step": 13079 }, { "epoch": 1.3917854862736752, "grad_norm": 12.40625, "learning_rate": 9.553013075611144e-06, "loss": 1.0897, "step": 13080 }, { "epoch": 1.3918918918918919, "grad_norm": 12.2109375, "learning_rate": 9.552977544059125e-06, "loss": 1.1442, "step": 13081 }, { "epoch": 1.3919982975101086, "grad_norm": 11.171875, "learning_rate": 9.552942012507107e-06, "loss": 0.7732, "step": 13082 }, { "epoch": 1.3921047031283251, "grad_norm": 11.7890625, "learning_rate": 9.552906480955088e-06, "loss": 1.2686, "step": 13083 }, { "epoch": 1.3922111087465419, "grad_norm": 13.9140625, "learning_rate": 9.55287094940307e-06, "loss": 1.1223, "step": 13084 }, { "epoch": 1.3923175143647586, "grad_norm": 21.328125, "learning_rate": 9.552835417851053e-06, "loss": 1.1635, "step": 13085 }, { "epoch": 1.392423919982975, "grad_norm": 15.3359375, "learning_rate": 9.552799886299035e-06, "loss": 1.0373, "step": 13086 }, { "epoch": 1.3925303256011916, "grad_norm": 10.5, "learning_rate": 9.552764354747016e-06, "loss": 0.8201, "step": 13087 }, { "epoch": 1.3926367312194083, "grad_norm": 9.890625, "learning_rate": 9.552728823194998e-06, "loss": 1.3571, "step": 13088 }, { "epoch": 1.392743136837625, "grad_norm": 10.9765625, "learning_rate": 9.55269329164298e-06, "loss": 1.1869, "step": 13089 }, { "epoch": 1.3928495424558416, "grad_norm": 13.8515625, "learning_rate": 9.552657760090961e-06, "loss": 1.3269, "step": 13090 }, { "epoch": 1.3929559480740583, "grad_norm": 11.984375, "learning_rate": 9.552622228538943e-06, "loss": 1.1739, "step": 13091 }, { "epoch": 1.393062353692275, "grad_norm": 11.6484375, "learning_rate": 9.552586696986925e-06, "loss": 1.1806, "step": 13092 }, { "epoch": 1.3931687593104916, "grad_norm": 18.109375, "learning_rate": 9.552551165434906e-06, "loss": 1.1057, "step": 13093 }, { "epoch": 1.3932751649287083, "grad_norm": 12.640625, "learning_rate": 9.552515633882888e-06, "loss": 1.0811, "step": 13094 }, { "epoch": 1.3933815705469248, "grad_norm": 13.703125, "learning_rate": 9.552480102330871e-06, "loss": 1.1358, "step": 13095 }, { "epoch": 1.3934879761651415, "grad_norm": 26.359375, "learning_rate": 9.552444570778853e-06, "loss": 1.3922, "step": 13096 }, { "epoch": 1.393594381783358, "grad_norm": 11.140625, "learning_rate": 9.552409039226834e-06, "loss": 1.1871, "step": 13097 }, { "epoch": 1.3937007874015748, "grad_norm": 15.171875, "learning_rate": 9.552373507674816e-06, "loss": 0.9296, "step": 13098 }, { "epoch": 1.3938071930197915, "grad_norm": 13.9140625, "learning_rate": 9.552337976122797e-06, "loss": 1.0393, "step": 13099 }, { "epoch": 1.393913598638008, "grad_norm": 13.4375, "learning_rate": 9.55230244457078e-06, "loss": 1.1387, "step": 13100 }, { "epoch": 1.3940200042562247, "grad_norm": 15.390625, "learning_rate": 9.552266913018762e-06, "loss": 1.1588, "step": 13101 }, { "epoch": 1.3941264098744415, "grad_norm": 11.5078125, "learning_rate": 9.552231381466744e-06, "loss": 0.941, "step": 13102 }, { "epoch": 1.394232815492658, "grad_norm": 11.96875, "learning_rate": 9.552195849914726e-06, "loss": 1.1819, "step": 13103 }, { "epoch": 1.3943392211108747, "grad_norm": 9.8125, "learning_rate": 9.552160318362707e-06, "loss": 1.4787, "step": 13104 }, { "epoch": 1.3944456267290912, "grad_norm": 15.2734375, "learning_rate": 9.552124786810689e-06, "loss": 0.9477, "step": 13105 }, { "epoch": 1.394552032347308, "grad_norm": 9.703125, "learning_rate": 9.55208925525867e-06, "loss": 0.9115, "step": 13106 }, { "epoch": 1.3946584379655245, "grad_norm": 17.09375, "learning_rate": 9.552053723706652e-06, "loss": 1.1845, "step": 13107 }, { "epoch": 1.3947648435837412, "grad_norm": 11.015625, "learning_rate": 9.552018192154634e-06, "loss": 0.9924, "step": 13108 }, { "epoch": 1.394871249201958, "grad_norm": 13.9375, "learning_rate": 9.551982660602615e-06, "loss": 1.0072, "step": 13109 }, { "epoch": 1.3949776548201744, "grad_norm": 11.828125, "learning_rate": 9.551947129050597e-06, "loss": 1.0322, "step": 13110 }, { "epoch": 1.3950840604383912, "grad_norm": 10.765625, "learning_rate": 9.55191159749858e-06, "loss": 1.1946, "step": 13111 }, { "epoch": 1.395190466056608, "grad_norm": 12.4765625, "learning_rate": 9.551876065946562e-06, "loss": 0.9734, "step": 13112 }, { "epoch": 1.3952968716748244, "grad_norm": 10.625, "learning_rate": 9.551840534394543e-06, "loss": 1.1136, "step": 13113 }, { "epoch": 1.3954032772930411, "grad_norm": 15.5546875, "learning_rate": 9.551805002842525e-06, "loss": 1.2103, "step": 13114 }, { "epoch": 1.3955096829112577, "grad_norm": 16.53125, "learning_rate": 9.551769471290507e-06, "loss": 1.282, "step": 13115 }, { "epoch": 1.3956160885294744, "grad_norm": 12.9921875, "learning_rate": 9.551733939738488e-06, "loss": 0.9319, "step": 13116 }, { "epoch": 1.395722494147691, "grad_norm": 15.828125, "learning_rate": 9.55169840818647e-06, "loss": 0.9365, "step": 13117 }, { "epoch": 1.3958288997659076, "grad_norm": 14.0078125, "learning_rate": 9.551662876634451e-06, "loss": 1.0862, "step": 13118 }, { "epoch": 1.3959353053841244, "grad_norm": 14.0390625, "learning_rate": 9.551627345082433e-06, "loss": 1.2074, "step": 13119 }, { "epoch": 1.3960417110023409, "grad_norm": 11.1953125, "learning_rate": 9.551591813530415e-06, "loss": 1.1643, "step": 13120 }, { "epoch": 1.3961481166205576, "grad_norm": 12.734375, "learning_rate": 9.551556281978398e-06, "loss": 1.2721, "step": 13121 }, { "epoch": 1.3962545222387743, "grad_norm": 9.9921875, "learning_rate": 9.55152075042638e-06, "loss": 1.0876, "step": 13122 }, { "epoch": 1.3963609278569908, "grad_norm": 13.1484375, "learning_rate": 9.551485218874361e-06, "loss": 1.0614, "step": 13123 }, { "epoch": 1.3964673334752076, "grad_norm": 12.578125, "learning_rate": 9.551449687322343e-06, "loss": 0.9662, "step": 13124 }, { "epoch": 1.396573739093424, "grad_norm": 12.3984375, "learning_rate": 9.551414155770324e-06, "loss": 1.0656, "step": 13125 }, { "epoch": 1.3966801447116408, "grad_norm": 12.28125, "learning_rate": 9.551378624218308e-06, "loss": 1.0163, "step": 13126 }, { "epoch": 1.3967865503298573, "grad_norm": 16.484375, "learning_rate": 9.55134309266629e-06, "loss": 0.9972, "step": 13127 }, { "epoch": 1.396892955948074, "grad_norm": 11.5625, "learning_rate": 9.55130756111427e-06, "loss": 0.9975, "step": 13128 }, { "epoch": 1.3969993615662908, "grad_norm": 11.2890625, "learning_rate": 9.551272029562252e-06, "loss": 1.2811, "step": 13129 }, { "epoch": 1.3971057671845073, "grad_norm": 14.2578125, "learning_rate": 9.551236498010234e-06, "loss": 1.2092, "step": 13130 }, { "epoch": 1.397212172802724, "grad_norm": 13.2578125, "learning_rate": 9.551200966458216e-06, "loss": 1.2994, "step": 13131 }, { "epoch": 1.3973185784209408, "grad_norm": 13.3984375, "learning_rate": 9.551165434906197e-06, "loss": 1.1677, "step": 13132 }, { "epoch": 1.3974249840391573, "grad_norm": 12.5234375, "learning_rate": 9.551129903354179e-06, "loss": 1.6382, "step": 13133 }, { "epoch": 1.3975313896573738, "grad_norm": 19.75, "learning_rate": 9.55109437180216e-06, "loss": 1.1736, "step": 13134 }, { "epoch": 1.3976377952755905, "grad_norm": 14.359375, "learning_rate": 9.551058840250142e-06, "loss": 0.9217, "step": 13135 }, { "epoch": 1.3977442008938072, "grad_norm": 15.234375, "learning_rate": 9.551023308698125e-06, "loss": 1.1161, "step": 13136 }, { "epoch": 1.3978506065120238, "grad_norm": 13.734375, "learning_rate": 9.550987777146107e-06, "loss": 0.9998, "step": 13137 }, { "epoch": 1.3979570121302405, "grad_norm": 14.15625, "learning_rate": 9.550952245594089e-06, "loss": 1.1876, "step": 13138 }, { "epoch": 1.3980634177484572, "grad_norm": 11.3515625, "learning_rate": 9.55091671404207e-06, "loss": 1.0395, "step": 13139 }, { "epoch": 1.3981698233666737, "grad_norm": 9.0703125, "learning_rate": 9.550881182490052e-06, "loss": 1.11, "step": 13140 }, { "epoch": 1.3982762289848905, "grad_norm": 12.3046875, "learning_rate": 9.550845650938035e-06, "loss": 1.4112, "step": 13141 }, { "epoch": 1.398382634603107, "grad_norm": 11.8984375, "learning_rate": 9.550810119386017e-06, "loss": 0.9851, "step": 13142 }, { "epoch": 1.3984890402213237, "grad_norm": 18.609375, "learning_rate": 9.550774587833998e-06, "loss": 1.0754, "step": 13143 }, { "epoch": 1.3985954458395402, "grad_norm": 9.5859375, "learning_rate": 9.550739056281978e-06, "loss": 1.2104, "step": 13144 }, { "epoch": 1.398701851457757, "grad_norm": 10.1328125, "learning_rate": 9.55070352472996e-06, "loss": 0.9528, "step": 13145 }, { "epoch": 1.3988082570759737, "grad_norm": 11.0546875, "learning_rate": 9.550667993177941e-06, "loss": 0.8578, "step": 13146 }, { "epoch": 1.3989146626941902, "grad_norm": 13.1796875, "learning_rate": 9.550632461625925e-06, "loss": 1.2593, "step": 13147 }, { "epoch": 1.399021068312407, "grad_norm": 12.3125, "learning_rate": 9.550596930073906e-06, "loss": 1.0802, "step": 13148 }, { "epoch": 1.3991274739306236, "grad_norm": 10.25, "learning_rate": 9.550561398521888e-06, "loss": 1.0428, "step": 13149 }, { "epoch": 1.3992338795488402, "grad_norm": 9.0, "learning_rate": 9.55052586696987e-06, "loss": 1.06, "step": 13150 }, { "epoch": 1.3993402851670569, "grad_norm": 10.203125, "learning_rate": 9.550490335417851e-06, "loss": 1.0169, "step": 13151 }, { "epoch": 1.3994466907852734, "grad_norm": 18.1875, "learning_rate": 9.550454803865834e-06, "loss": 1.4666, "step": 13152 }, { "epoch": 1.3995530964034901, "grad_norm": 11.6015625, "learning_rate": 9.550419272313816e-06, "loss": 1.1051, "step": 13153 }, { "epoch": 1.3996595020217066, "grad_norm": 10.9921875, "learning_rate": 9.550383740761798e-06, "loss": 1.2542, "step": 13154 }, { "epoch": 1.3997659076399234, "grad_norm": 9.953125, "learning_rate": 9.55034820920978e-06, "loss": 1.273, "step": 13155 }, { "epoch": 1.39987231325814, "grad_norm": 11.84375, "learning_rate": 9.550312677657761e-06, "loss": 1.0114, "step": 13156 }, { "epoch": 1.3999787188763566, "grad_norm": 12.515625, "learning_rate": 9.550277146105742e-06, "loss": 1.1672, "step": 13157 }, { "epoch": 1.4000851244945733, "grad_norm": 12.71875, "learning_rate": 9.550241614553724e-06, "loss": 1.1767, "step": 13158 }, { "epoch": 1.40019153011279, "grad_norm": 12.7265625, "learning_rate": 9.550206083001706e-06, "loss": 0.9985, "step": 13159 }, { "epoch": 1.4002979357310066, "grad_norm": 11.7578125, "learning_rate": 9.550170551449687e-06, "loss": 1.0736, "step": 13160 }, { "epoch": 1.4004043413492233, "grad_norm": 12.140625, "learning_rate": 9.550135019897669e-06, "loss": 1.0408, "step": 13161 }, { "epoch": 1.4005107469674398, "grad_norm": 11.59375, "learning_rate": 9.550099488345652e-06, "loss": 1.1341, "step": 13162 }, { "epoch": 1.4006171525856566, "grad_norm": 10.828125, "learning_rate": 9.550063956793634e-06, "loss": 1.1047, "step": 13163 }, { "epoch": 1.400723558203873, "grad_norm": 12.3359375, "learning_rate": 9.550028425241615e-06, "loss": 1.443, "step": 13164 }, { "epoch": 1.4008299638220898, "grad_norm": 11.5859375, "learning_rate": 9.549992893689597e-06, "loss": 1.14, "step": 13165 }, { "epoch": 1.4009363694403065, "grad_norm": 14.3515625, "learning_rate": 9.549957362137579e-06, "loss": 1.2493, "step": 13166 }, { "epoch": 1.401042775058523, "grad_norm": 12.671875, "learning_rate": 9.549921830585562e-06, "loss": 0.9664, "step": 13167 }, { "epoch": 1.4011491806767398, "grad_norm": 21.09375, "learning_rate": 9.549886299033544e-06, "loss": 1.3479, "step": 13168 }, { "epoch": 1.4012555862949565, "grad_norm": 13.625, "learning_rate": 9.549850767481525e-06, "loss": 1.2401, "step": 13169 }, { "epoch": 1.401361991913173, "grad_norm": 12.4609375, "learning_rate": 9.549815235929507e-06, "loss": 0.9811, "step": 13170 }, { "epoch": 1.4014683975313895, "grad_norm": 11.4140625, "learning_rate": 9.549779704377488e-06, "loss": 1.312, "step": 13171 }, { "epoch": 1.4015748031496063, "grad_norm": 12.140625, "learning_rate": 9.54974417282547e-06, "loss": 0.9113, "step": 13172 }, { "epoch": 1.401681208767823, "grad_norm": 17.515625, "learning_rate": 9.549708641273452e-06, "loss": 0.897, "step": 13173 }, { "epoch": 1.4017876143860395, "grad_norm": 14.4921875, "learning_rate": 9.549673109721433e-06, "loss": 0.9968, "step": 13174 }, { "epoch": 1.4018940200042562, "grad_norm": 13.890625, "learning_rate": 9.549637578169415e-06, "loss": 1.2549, "step": 13175 }, { "epoch": 1.402000425622473, "grad_norm": 13.6328125, "learning_rate": 9.549602046617396e-06, "loss": 1.2237, "step": 13176 }, { "epoch": 1.4021068312406895, "grad_norm": 10.75, "learning_rate": 9.54956651506538e-06, "loss": 0.8492, "step": 13177 }, { "epoch": 1.4022132368589062, "grad_norm": 15.6171875, "learning_rate": 9.549530983513361e-06, "loss": 1.2043, "step": 13178 }, { "epoch": 1.402319642477123, "grad_norm": 10.9609375, "learning_rate": 9.549495451961343e-06, "loss": 0.9331, "step": 13179 }, { "epoch": 1.4024260480953394, "grad_norm": 14.765625, "learning_rate": 9.549459920409325e-06, "loss": 0.9935, "step": 13180 }, { "epoch": 1.402532453713556, "grad_norm": 12.0546875, "learning_rate": 9.549424388857306e-06, "loss": 1.0428, "step": 13181 }, { "epoch": 1.4026388593317727, "grad_norm": 20.625, "learning_rate": 9.549388857305288e-06, "loss": 1.4134, "step": 13182 }, { "epoch": 1.4027452649499894, "grad_norm": 14.078125, "learning_rate": 9.54935332575327e-06, "loss": 1.2426, "step": 13183 }, { "epoch": 1.402851670568206, "grad_norm": 13.484375, "learning_rate": 9.549317794201251e-06, "loss": 1.3645, "step": 13184 }, { "epoch": 1.4029580761864227, "grad_norm": 14.7265625, "learning_rate": 9.549282262649233e-06, "loss": 1.106, "step": 13185 }, { "epoch": 1.4030644818046394, "grad_norm": 15.3671875, "learning_rate": 9.549246731097214e-06, "loss": 1.1704, "step": 13186 }, { "epoch": 1.403170887422856, "grad_norm": 11.5859375, "learning_rate": 9.549211199545196e-06, "loss": 0.8962, "step": 13187 }, { "epoch": 1.4032772930410726, "grad_norm": 13.9921875, "learning_rate": 9.549175667993179e-06, "loss": 1.1322, "step": 13188 }, { "epoch": 1.4033836986592891, "grad_norm": 12.015625, "learning_rate": 9.54914013644116e-06, "loss": 0.9926, "step": 13189 }, { "epoch": 1.4034901042775059, "grad_norm": 12.875, "learning_rate": 9.549104604889142e-06, "loss": 1.1465, "step": 13190 }, { "epoch": 1.4035965098957224, "grad_norm": 9.515625, "learning_rate": 9.549069073337124e-06, "loss": 1.0395, "step": 13191 }, { "epoch": 1.403702915513939, "grad_norm": 16.671875, "learning_rate": 9.549033541785106e-06, "loss": 1.1874, "step": 13192 }, { "epoch": 1.4038093211321558, "grad_norm": 15.7578125, "learning_rate": 9.548998010233089e-06, "loss": 1.1149, "step": 13193 }, { "epoch": 1.4039157267503724, "grad_norm": 15.96875, "learning_rate": 9.54896247868107e-06, "loss": 1.1847, "step": 13194 }, { "epoch": 1.404022132368589, "grad_norm": 12.015625, "learning_rate": 9.548926947129052e-06, "loss": 1.0042, "step": 13195 }, { "epoch": 1.4041285379868058, "grad_norm": 10.921875, "learning_rate": 9.548891415577034e-06, "loss": 1.0634, "step": 13196 }, { "epoch": 1.4042349436050223, "grad_norm": 10.53125, "learning_rate": 9.548855884025015e-06, "loss": 1.0963, "step": 13197 }, { "epoch": 1.404341349223239, "grad_norm": 12.9765625, "learning_rate": 9.548820352472997e-06, "loss": 1.1075, "step": 13198 }, { "epoch": 1.4044477548414556, "grad_norm": 10.96875, "learning_rate": 9.548784820920978e-06, "loss": 1.1296, "step": 13199 }, { "epoch": 1.4045541604596723, "grad_norm": 21.21875, "learning_rate": 9.54874928936896e-06, "loss": 1.0123, "step": 13200 }, { "epoch": 1.4046605660778888, "grad_norm": 12.65625, "learning_rate": 9.548713757816942e-06, "loss": 1.1147, "step": 13201 }, { "epoch": 1.4047669716961055, "grad_norm": 12.5625, "learning_rate": 9.548678226264923e-06, "loss": 1.2005, "step": 13202 }, { "epoch": 1.4048733773143223, "grad_norm": 10.15625, "learning_rate": 9.548642694712907e-06, "loss": 1.0667, "step": 13203 }, { "epoch": 1.4049797829325388, "grad_norm": 13.90625, "learning_rate": 9.548607163160888e-06, "loss": 1.0066, "step": 13204 }, { "epoch": 1.4050861885507555, "grad_norm": 12.4296875, "learning_rate": 9.54857163160887e-06, "loss": 1.2695, "step": 13205 }, { "epoch": 1.4051925941689722, "grad_norm": 10.9296875, "learning_rate": 9.548536100056851e-06, "loss": 1.2378, "step": 13206 }, { "epoch": 1.4052989997871888, "grad_norm": 10.9765625, "learning_rate": 9.548500568504833e-06, "loss": 0.8954, "step": 13207 }, { "epoch": 1.4054054054054055, "grad_norm": 12.109375, "learning_rate": 9.548465036952815e-06, "loss": 0.9119, "step": 13208 }, { "epoch": 1.405511811023622, "grad_norm": 12.3125, "learning_rate": 9.548429505400796e-06, "loss": 0.8077, "step": 13209 }, { "epoch": 1.4056182166418387, "grad_norm": 9.8125, "learning_rate": 9.548393973848778e-06, "loss": 1.109, "step": 13210 }, { "epoch": 1.4057246222600552, "grad_norm": 11.59375, "learning_rate": 9.54835844229676e-06, "loss": 1.0538, "step": 13211 }, { "epoch": 1.405831027878272, "grad_norm": 10.2109375, "learning_rate": 9.548322910744741e-06, "loss": 1.3295, "step": 13212 }, { "epoch": 1.4059374334964887, "grad_norm": 11.96875, "learning_rate": 9.548287379192724e-06, "loss": 1.0516, "step": 13213 }, { "epoch": 1.4060438391147052, "grad_norm": 8.46875, "learning_rate": 9.548251847640706e-06, "loss": 0.7895, "step": 13214 }, { "epoch": 1.406150244732922, "grad_norm": 11.890625, "learning_rate": 9.548216316088688e-06, "loss": 1.0666, "step": 13215 }, { "epoch": 1.4062566503511387, "grad_norm": 12.640625, "learning_rate": 9.548180784536669e-06, "loss": 1.1997, "step": 13216 }, { "epoch": 1.4063630559693552, "grad_norm": 19.71875, "learning_rate": 9.54814525298465e-06, "loss": 1.036, "step": 13217 }, { "epoch": 1.4064694615875717, "grad_norm": 11.640625, "learning_rate": 9.548109721432634e-06, "loss": 1.0151, "step": 13218 }, { "epoch": 1.4065758672057884, "grad_norm": 24.828125, "learning_rate": 9.548074189880616e-06, "loss": 1.2921, "step": 13219 }, { "epoch": 1.4066822728240052, "grad_norm": 12.9140625, "learning_rate": 9.548038658328597e-06, "loss": 0.9371, "step": 13220 }, { "epoch": 1.4067886784422217, "grad_norm": 12.1171875, "learning_rate": 9.548003126776579e-06, "loss": 1.2413, "step": 13221 }, { "epoch": 1.4068950840604384, "grad_norm": 11.2421875, "learning_rate": 9.54796759522456e-06, "loss": 1.1583, "step": 13222 }, { "epoch": 1.4070014896786551, "grad_norm": 14.828125, "learning_rate": 9.547932063672542e-06, "loss": 1.0288, "step": 13223 }, { "epoch": 1.4071078952968716, "grad_norm": 18.34375, "learning_rate": 9.547896532120524e-06, "loss": 1.5471, "step": 13224 }, { "epoch": 1.4072143009150884, "grad_norm": 16.203125, "learning_rate": 9.547861000568505e-06, "loss": 1.066, "step": 13225 }, { "epoch": 1.4073207065333049, "grad_norm": 12.109375, "learning_rate": 9.547825469016487e-06, "loss": 1.2538, "step": 13226 }, { "epoch": 1.4074271121515216, "grad_norm": 15.78125, "learning_rate": 9.547789937464469e-06, "loss": 1.4548, "step": 13227 }, { "epoch": 1.4075335177697381, "grad_norm": 17.0, "learning_rate": 9.54775440591245e-06, "loss": 1.1876, "step": 13228 }, { "epoch": 1.4076399233879549, "grad_norm": 14.3046875, "learning_rate": 9.547718874360433e-06, "loss": 1.3823, "step": 13229 }, { "epoch": 1.4077463290061716, "grad_norm": 12.625, "learning_rate": 9.547683342808415e-06, "loss": 1.1149, "step": 13230 }, { "epoch": 1.407852734624388, "grad_norm": 11.859375, "learning_rate": 9.547647811256397e-06, "loss": 1.2629, "step": 13231 }, { "epoch": 1.4079591402426048, "grad_norm": 11.6796875, "learning_rate": 9.547612279704378e-06, "loss": 1.223, "step": 13232 }, { "epoch": 1.4080655458608216, "grad_norm": 11.1796875, "learning_rate": 9.54757674815236e-06, "loss": 0.9245, "step": 13233 }, { "epoch": 1.408171951479038, "grad_norm": 11.828125, "learning_rate": 9.547541216600341e-06, "loss": 1.0315, "step": 13234 }, { "epoch": 1.4082783570972548, "grad_norm": 10.90625, "learning_rate": 9.547505685048323e-06, "loss": 1.246, "step": 13235 }, { "epoch": 1.4083847627154713, "grad_norm": 14.078125, "learning_rate": 9.547470153496305e-06, "loss": 1.1207, "step": 13236 }, { "epoch": 1.408491168333688, "grad_norm": 13.0859375, "learning_rate": 9.547434621944286e-06, "loss": 1.3647, "step": 13237 }, { "epoch": 1.4085975739519045, "grad_norm": 11.25, "learning_rate": 9.547399090392268e-06, "loss": 1.1596, "step": 13238 }, { "epoch": 1.4087039795701213, "grad_norm": 11.4921875, "learning_rate": 9.547363558840251e-06, "loss": 1.1556, "step": 13239 }, { "epoch": 1.408810385188338, "grad_norm": 9.34375, "learning_rate": 9.547328027288233e-06, "loss": 1.0655, "step": 13240 }, { "epoch": 1.4089167908065545, "grad_norm": 12.046875, "learning_rate": 9.547292495736214e-06, "loss": 1.0848, "step": 13241 }, { "epoch": 1.4090231964247713, "grad_norm": 11.1484375, "learning_rate": 9.547256964184196e-06, "loss": 1.065, "step": 13242 }, { "epoch": 1.409129602042988, "grad_norm": 13.7265625, "learning_rate": 9.547221432632178e-06, "loss": 1.1353, "step": 13243 }, { "epoch": 1.4092360076612045, "grad_norm": 13.5546875, "learning_rate": 9.547185901080161e-06, "loss": 1.1933, "step": 13244 }, { "epoch": 1.4093424132794212, "grad_norm": 14.203125, "learning_rate": 9.547150369528143e-06, "loss": 0.9582, "step": 13245 }, { "epoch": 1.4094488188976377, "grad_norm": 14.796875, "learning_rate": 9.547114837976124e-06, "loss": 1.0846, "step": 13246 }, { "epoch": 1.4095552245158545, "grad_norm": 10.1953125, "learning_rate": 9.547079306424106e-06, "loss": 0.9863, "step": 13247 }, { "epoch": 1.409661630134071, "grad_norm": 10.09375, "learning_rate": 9.547043774872087e-06, "loss": 1.1358, "step": 13248 }, { "epoch": 1.4097680357522877, "grad_norm": 12.59375, "learning_rate": 9.547008243320069e-06, "loss": 1.0107, "step": 13249 }, { "epoch": 1.4098744413705044, "grad_norm": 15.203125, "learning_rate": 9.54697271176805e-06, "loss": 1.1359, "step": 13250 }, { "epoch": 1.409980846988721, "grad_norm": 13.015625, "learning_rate": 9.546937180216032e-06, "loss": 0.9405, "step": 13251 }, { "epoch": 1.4100872526069377, "grad_norm": 12.7265625, "learning_rate": 9.546901648664014e-06, "loss": 1.1028, "step": 13252 }, { "epoch": 1.4101936582251544, "grad_norm": 12.3125, "learning_rate": 9.546866117111995e-06, "loss": 1.0753, "step": 13253 }, { "epoch": 1.410300063843371, "grad_norm": 14.0078125, "learning_rate": 9.546830585559979e-06, "loss": 1.1136, "step": 13254 }, { "epoch": 1.4104064694615877, "grad_norm": 11.8828125, "learning_rate": 9.54679505400796e-06, "loss": 0.9436, "step": 13255 }, { "epoch": 1.4105128750798042, "grad_norm": 10.1328125, "learning_rate": 9.546759522455942e-06, "loss": 1.0472, "step": 13256 }, { "epoch": 1.410619280698021, "grad_norm": 10.609375, "learning_rate": 9.546723990903923e-06, "loss": 1.2965, "step": 13257 }, { "epoch": 1.4107256863162374, "grad_norm": 11.84375, "learning_rate": 9.546688459351905e-06, "loss": 0.8903, "step": 13258 }, { "epoch": 1.4108320919344541, "grad_norm": 14.953125, "learning_rate": 9.546652927799888e-06, "loss": 1.5865, "step": 13259 }, { "epoch": 1.4109384975526709, "grad_norm": 12.375, "learning_rate": 9.54661739624787e-06, "loss": 1.0743, "step": 13260 }, { "epoch": 1.4110449031708874, "grad_norm": 12.015625, "learning_rate": 9.546581864695852e-06, "loss": 1.1661, "step": 13261 }, { "epoch": 1.411151308789104, "grad_norm": 12.953125, "learning_rate": 9.546546333143833e-06, "loss": 0.9766, "step": 13262 }, { "epoch": 1.4112577144073208, "grad_norm": 12.8203125, "learning_rate": 9.546510801591815e-06, "loss": 0.9713, "step": 13263 }, { "epoch": 1.4113641200255373, "grad_norm": 12.546875, "learning_rate": 9.546475270039795e-06, "loss": 0.9648, "step": 13264 }, { "epoch": 1.4114705256437539, "grad_norm": 17.4375, "learning_rate": 9.546439738487778e-06, "loss": 1.03, "step": 13265 }, { "epoch": 1.4115769312619706, "grad_norm": 9.6875, "learning_rate": 9.54640420693576e-06, "loss": 0.899, "step": 13266 }, { "epoch": 1.4116833368801873, "grad_norm": 10.25, "learning_rate": 9.546368675383741e-06, "loss": 0.9458, "step": 13267 }, { "epoch": 1.4117897424984038, "grad_norm": 12.703125, "learning_rate": 9.546333143831723e-06, "loss": 1.2986, "step": 13268 }, { "epoch": 1.4118961481166206, "grad_norm": 11.9921875, "learning_rate": 9.546297612279704e-06, "loss": 1.0282, "step": 13269 }, { "epoch": 1.4120025537348373, "grad_norm": 10.984375, "learning_rate": 9.546262080727688e-06, "loss": 0.8554, "step": 13270 }, { "epoch": 1.4121089593530538, "grad_norm": 12.4140625, "learning_rate": 9.54622654917567e-06, "loss": 1.134, "step": 13271 }, { "epoch": 1.4122153649712705, "grad_norm": 13.7421875, "learning_rate": 9.546191017623651e-06, "loss": 1.0965, "step": 13272 }, { "epoch": 1.412321770589487, "grad_norm": 11.71875, "learning_rate": 9.546155486071633e-06, "loss": 1.1331, "step": 13273 }, { "epoch": 1.4124281762077038, "grad_norm": 11.203125, "learning_rate": 9.546119954519614e-06, "loss": 1.1586, "step": 13274 }, { "epoch": 1.4125345818259203, "grad_norm": 22.109375, "learning_rate": 9.546084422967596e-06, "loss": 1.0179, "step": 13275 }, { "epoch": 1.412640987444137, "grad_norm": 14.0859375, "learning_rate": 9.546048891415577e-06, "loss": 1.1777, "step": 13276 }, { "epoch": 1.4127473930623538, "grad_norm": 11.9140625, "learning_rate": 9.546013359863559e-06, "loss": 1.2825, "step": 13277 }, { "epoch": 1.4128537986805703, "grad_norm": 10.71875, "learning_rate": 9.54597782831154e-06, "loss": 1.2051, "step": 13278 }, { "epoch": 1.412960204298787, "grad_norm": 11.5546875, "learning_rate": 9.545942296759522e-06, "loss": 1.0554, "step": 13279 }, { "epoch": 1.4130666099170037, "grad_norm": 15.1640625, "learning_rate": 9.545906765207506e-06, "loss": 1.131, "step": 13280 }, { "epoch": 1.4131730155352202, "grad_norm": 15.0234375, "learning_rate": 9.545871233655487e-06, "loss": 1.4983, "step": 13281 }, { "epoch": 1.413279421153437, "grad_norm": 19.53125, "learning_rate": 9.545835702103469e-06, "loss": 1.2451, "step": 13282 }, { "epoch": 1.4133858267716535, "grad_norm": 10.6796875, "learning_rate": 9.54580017055145e-06, "loss": 1.1406, "step": 13283 }, { "epoch": 1.4134922323898702, "grad_norm": 13.9296875, "learning_rate": 9.545764638999432e-06, "loss": 0.9391, "step": 13284 }, { "epoch": 1.4135986380080867, "grad_norm": 12.8671875, "learning_rate": 9.545729107447415e-06, "loss": 1.0703, "step": 13285 }, { "epoch": 1.4137050436263034, "grad_norm": 10.375, "learning_rate": 9.545693575895397e-06, "loss": 1.0799, "step": 13286 }, { "epoch": 1.4138114492445202, "grad_norm": 14.7421875, "learning_rate": 9.545658044343378e-06, "loss": 0.9819, "step": 13287 }, { "epoch": 1.4139178548627367, "grad_norm": 12.703125, "learning_rate": 9.54562251279136e-06, "loss": 0.8653, "step": 13288 }, { "epoch": 1.4140242604809534, "grad_norm": 9.3828125, "learning_rate": 9.545586981239342e-06, "loss": 1.0545, "step": 13289 }, { "epoch": 1.4141306660991702, "grad_norm": 15.265625, "learning_rate": 9.545551449687323e-06, "loss": 1.3922, "step": 13290 }, { "epoch": 1.4142370717173867, "grad_norm": 12.8984375, "learning_rate": 9.545515918135305e-06, "loss": 1.1362, "step": 13291 }, { "epoch": 1.4143434773356034, "grad_norm": 10.953125, "learning_rate": 9.545480386583287e-06, "loss": 1.1762, "step": 13292 }, { "epoch": 1.41444988295382, "grad_norm": 9.84375, "learning_rate": 9.545444855031268e-06, "loss": 1.0308, "step": 13293 }, { "epoch": 1.4145562885720366, "grad_norm": 15.6953125, "learning_rate": 9.54540932347925e-06, "loss": 1.2569, "step": 13294 }, { "epoch": 1.4146626941902531, "grad_norm": 13.046875, "learning_rate": 9.545373791927233e-06, "loss": 1.5012, "step": 13295 }, { "epoch": 1.4147690998084699, "grad_norm": 10.625, "learning_rate": 9.545338260375215e-06, "loss": 1.0706, "step": 13296 }, { "epoch": 1.4148755054266866, "grad_norm": 8.9765625, "learning_rate": 9.545302728823196e-06, "loss": 0.9528, "step": 13297 }, { "epoch": 1.4149819110449031, "grad_norm": 14.421875, "learning_rate": 9.545267197271178e-06, "loss": 1.6418, "step": 13298 }, { "epoch": 1.4150883166631198, "grad_norm": 13.671875, "learning_rate": 9.54523166571916e-06, "loss": 1.652, "step": 13299 }, { "epoch": 1.4151947222813366, "grad_norm": 13.4296875, "learning_rate": 9.545196134167141e-06, "loss": 1.0602, "step": 13300 }, { "epoch": 1.415301127899553, "grad_norm": 10.5078125, "learning_rate": 9.545160602615123e-06, "loss": 1.1006, "step": 13301 }, { "epoch": 1.4154075335177696, "grad_norm": 15.1171875, "learning_rate": 9.545125071063104e-06, "loss": 1.123, "step": 13302 }, { "epoch": 1.4155139391359863, "grad_norm": 10.9375, "learning_rate": 9.545089539511086e-06, "loss": 0.9006, "step": 13303 }, { "epoch": 1.415620344754203, "grad_norm": 17.140625, "learning_rate": 9.545054007959067e-06, "loss": 1.4317, "step": 13304 }, { "epoch": 1.4157267503724196, "grad_norm": 12.015625, "learning_rate": 9.545018476407049e-06, "loss": 1.0149, "step": 13305 }, { "epoch": 1.4158331559906363, "grad_norm": 13.9375, "learning_rate": 9.544982944855032e-06, "loss": 1.229, "step": 13306 }, { "epoch": 1.415939561608853, "grad_norm": 11.71875, "learning_rate": 9.544947413303014e-06, "loss": 1.1566, "step": 13307 }, { "epoch": 1.4160459672270695, "grad_norm": 10.171875, "learning_rate": 9.544911881750996e-06, "loss": 1.2366, "step": 13308 }, { "epoch": 1.4161523728452863, "grad_norm": 17.421875, "learning_rate": 9.544876350198977e-06, "loss": 1.4293, "step": 13309 }, { "epoch": 1.416258778463503, "grad_norm": 9.0078125, "learning_rate": 9.544840818646959e-06, "loss": 0.9239, "step": 13310 }, { "epoch": 1.4163651840817195, "grad_norm": 9.640625, "learning_rate": 9.544805287094942e-06, "loss": 1.194, "step": 13311 }, { "epoch": 1.416471589699936, "grad_norm": 9.75, "learning_rate": 9.544769755542924e-06, "loss": 0.9537, "step": 13312 }, { "epoch": 1.4165779953181528, "grad_norm": 11.9296875, "learning_rate": 9.544734223990905e-06, "loss": 1.4438, "step": 13313 }, { "epoch": 1.4166844009363695, "grad_norm": 9.9453125, "learning_rate": 9.544698692438887e-06, "loss": 1.0204, "step": 13314 }, { "epoch": 1.416790806554586, "grad_norm": 12.4296875, "learning_rate": 9.544663160886869e-06, "loss": 1.2967, "step": 13315 }, { "epoch": 1.4168972121728027, "grad_norm": 10.9765625, "learning_rate": 9.54462762933485e-06, "loss": 0.9384, "step": 13316 }, { "epoch": 1.4170036177910195, "grad_norm": 12.4921875, "learning_rate": 9.544592097782832e-06, "loss": 1.1029, "step": 13317 }, { "epoch": 1.417110023409236, "grad_norm": 13.7109375, "learning_rate": 9.544556566230813e-06, "loss": 0.9705, "step": 13318 }, { "epoch": 1.4172164290274527, "grad_norm": 15.5078125, "learning_rate": 9.544521034678795e-06, "loss": 1.1699, "step": 13319 }, { "epoch": 1.4173228346456692, "grad_norm": 12.5234375, "learning_rate": 9.544485503126777e-06, "loss": 1.0292, "step": 13320 }, { "epoch": 1.417429240263886, "grad_norm": 12.3125, "learning_rate": 9.54444997157476e-06, "loss": 1.0707, "step": 13321 }, { "epoch": 1.4175356458821025, "grad_norm": 12.515625, "learning_rate": 9.544414440022741e-06, "loss": 1.236, "step": 13322 }, { "epoch": 1.4176420515003192, "grad_norm": 14.640625, "learning_rate": 9.544378908470723e-06, "loss": 1.4003, "step": 13323 }, { "epoch": 1.417748457118536, "grad_norm": 14.3828125, "learning_rate": 9.544343376918705e-06, "loss": 1.1304, "step": 13324 }, { "epoch": 1.4178548627367524, "grad_norm": 12.9453125, "learning_rate": 9.544307845366686e-06, "loss": 1.265, "step": 13325 }, { "epoch": 1.4179612683549692, "grad_norm": 15.4296875, "learning_rate": 9.544272313814668e-06, "loss": 0.9503, "step": 13326 }, { "epoch": 1.418067673973186, "grad_norm": 13.6015625, "learning_rate": 9.54423678226265e-06, "loss": 1.0847, "step": 13327 }, { "epoch": 1.4181740795914024, "grad_norm": 10.2265625, "learning_rate": 9.544201250710631e-06, "loss": 1.0678, "step": 13328 }, { "epoch": 1.4182804852096191, "grad_norm": 11.8515625, "learning_rate": 9.544165719158613e-06, "loss": 0.9421, "step": 13329 }, { "epoch": 1.4183868908278356, "grad_norm": 11.2109375, "learning_rate": 9.544130187606594e-06, "loss": 1.3325, "step": 13330 }, { "epoch": 1.4184932964460524, "grad_norm": 15.2421875, "learning_rate": 9.544094656054578e-06, "loss": 1.0536, "step": 13331 }, { "epoch": 1.4185997020642689, "grad_norm": 8.5078125, "learning_rate": 9.54405912450256e-06, "loss": 1.2009, "step": 13332 }, { "epoch": 1.4187061076824856, "grad_norm": 17.046875, "learning_rate": 9.54402359295054e-06, "loss": 1.3048, "step": 13333 }, { "epoch": 1.4188125133007023, "grad_norm": 11.6015625, "learning_rate": 9.543988061398522e-06, "loss": 0.8695, "step": 13334 }, { "epoch": 1.4189189189189189, "grad_norm": 12.9375, "learning_rate": 9.543952529846504e-06, "loss": 1.0797, "step": 13335 }, { "epoch": 1.4190253245371356, "grad_norm": 13.4609375, "learning_rate": 9.543916998294487e-06, "loss": 0.9453, "step": 13336 }, { "epoch": 1.4191317301553523, "grad_norm": 10.84375, "learning_rate": 9.543881466742469e-06, "loss": 0.9451, "step": 13337 }, { "epoch": 1.4192381357735688, "grad_norm": 9.4375, "learning_rate": 9.54384593519045e-06, "loss": 0.915, "step": 13338 }, { "epoch": 1.4193445413917856, "grad_norm": 11.1953125, "learning_rate": 9.543810403638432e-06, "loss": 0.8222, "step": 13339 }, { "epoch": 1.419450947010002, "grad_norm": 10.53125, "learning_rate": 9.543774872086414e-06, "loss": 0.881, "step": 13340 }, { "epoch": 1.4195573526282188, "grad_norm": 16.15625, "learning_rate": 9.543739340534395e-06, "loss": 0.9276, "step": 13341 }, { "epoch": 1.4196637582464353, "grad_norm": 10.2421875, "learning_rate": 9.543703808982377e-06, "loss": 0.8229, "step": 13342 }, { "epoch": 1.419770163864652, "grad_norm": 19.109375, "learning_rate": 9.543668277430359e-06, "loss": 0.9168, "step": 13343 }, { "epoch": 1.4198765694828688, "grad_norm": 10.828125, "learning_rate": 9.54363274587834e-06, "loss": 1.0069, "step": 13344 }, { "epoch": 1.4199829751010853, "grad_norm": 9.7421875, "learning_rate": 9.543597214326322e-06, "loss": 1.1623, "step": 13345 }, { "epoch": 1.420089380719302, "grad_norm": 13.6953125, "learning_rate": 9.543561682774303e-06, "loss": 1.1795, "step": 13346 }, { "epoch": 1.4201957863375188, "grad_norm": 11.796875, "learning_rate": 9.543526151222287e-06, "loss": 1.0026, "step": 13347 }, { "epoch": 1.4203021919557353, "grad_norm": 10.21875, "learning_rate": 9.543490619670268e-06, "loss": 1.1322, "step": 13348 }, { "epoch": 1.4204085975739518, "grad_norm": 15.25, "learning_rate": 9.54345508811825e-06, "loss": 1.1189, "step": 13349 }, { "epoch": 1.4205150031921685, "grad_norm": 13.7265625, "learning_rate": 9.543419556566232e-06, "loss": 1.1305, "step": 13350 }, { "epoch": 1.4206214088103852, "grad_norm": 11.75, "learning_rate": 9.543384025014213e-06, "loss": 0.974, "step": 13351 }, { "epoch": 1.4207278144286017, "grad_norm": 12.4375, "learning_rate": 9.543348493462196e-06, "loss": 1.441, "step": 13352 }, { "epoch": 1.4208342200468185, "grad_norm": 11.9453125, "learning_rate": 9.543312961910176e-06, "loss": 0.9566, "step": 13353 }, { "epoch": 1.4209406256650352, "grad_norm": 11.4765625, "learning_rate": 9.543277430358158e-06, "loss": 1.169, "step": 13354 }, { "epoch": 1.4210470312832517, "grad_norm": 15.71875, "learning_rate": 9.54324189880614e-06, "loss": 1.1916, "step": 13355 }, { "epoch": 1.4211534369014684, "grad_norm": 16.203125, "learning_rate": 9.543206367254121e-06, "loss": 0.9695, "step": 13356 }, { "epoch": 1.421259842519685, "grad_norm": 12.3203125, "learning_rate": 9.543170835702104e-06, "loss": 0.9673, "step": 13357 }, { "epoch": 1.4213662481379017, "grad_norm": 10.9296875, "learning_rate": 9.543135304150086e-06, "loss": 0.9155, "step": 13358 }, { "epoch": 1.4214726537561182, "grad_norm": 15.4296875, "learning_rate": 9.543099772598068e-06, "loss": 0.9117, "step": 13359 }, { "epoch": 1.421579059374335, "grad_norm": 18.21875, "learning_rate": 9.54306424104605e-06, "loss": 0.9539, "step": 13360 }, { "epoch": 1.4216854649925517, "grad_norm": 10.2265625, "learning_rate": 9.543028709494031e-06, "loss": 1.0133, "step": 13361 }, { "epoch": 1.4217918706107682, "grad_norm": 14.2890625, "learning_rate": 9.542993177942014e-06, "loss": 1.1512, "step": 13362 }, { "epoch": 1.421898276228985, "grad_norm": 13.9296875, "learning_rate": 9.542957646389996e-06, "loss": 0.8796, "step": 13363 }, { "epoch": 1.4220046818472016, "grad_norm": 12.3515625, "learning_rate": 9.542922114837977e-06, "loss": 1.2908, "step": 13364 }, { "epoch": 1.4221110874654181, "grad_norm": 13.671875, "learning_rate": 9.542886583285959e-06, "loss": 1.4376, "step": 13365 }, { "epoch": 1.4222174930836349, "grad_norm": 14.875, "learning_rate": 9.54285105173394e-06, "loss": 0.9269, "step": 13366 }, { "epoch": 1.4223238987018514, "grad_norm": 10.2578125, "learning_rate": 9.542815520181922e-06, "loss": 1.1234, "step": 13367 }, { "epoch": 1.4224303043200681, "grad_norm": 12.21875, "learning_rate": 9.542779988629904e-06, "loss": 0.9234, "step": 13368 }, { "epoch": 1.4225367099382846, "grad_norm": 11.71875, "learning_rate": 9.542744457077885e-06, "loss": 1.1558, "step": 13369 }, { "epoch": 1.4226431155565014, "grad_norm": 11.5625, "learning_rate": 9.542708925525867e-06, "loss": 1.0065, "step": 13370 }, { "epoch": 1.422749521174718, "grad_norm": 14.5078125, "learning_rate": 9.542673393973849e-06, "loss": 1.1728, "step": 13371 }, { "epoch": 1.4228559267929346, "grad_norm": 11.4375, "learning_rate": 9.542637862421832e-06, "loss": 1.1119, "step": 13372 }, { "epoch": 1.4229623324111513, "grad_norm": 20.5625, "learning_rate": 9.542602330869814e-06, "loss": 1.1255, "step": 13373 }, { "epoch": 1.423068738029368, "grad_norm": 12.3828125, "learning_rate": 9.542566799317795e-06, "loss": 1.0668, "step": 13374 }, { "epoch": 1.4231751436475846, "grad_norm": 13.5234375, "learning_rate": 9.542531267765777e-06, "loss": 1.095, "step": 13375 }, { "epoch": 1.4232815492658013, "grad_norm": 14.078125, "learning_rate": 9.542495736213758e-06, "loss": 1.3108, "step": 13376 }, { "epoch": 1.4233879548840178, "grad_norm": 11.46875, "learning_rate": 9.542460204661742e-06, "loss": 1.0688, "step": 13377 }, { "epoch": 1.4234943605022345, "grad_norm": 13.609375, "learning_rate": 9.542424673109723e-06, "loss": 1.0368, "step": 13378 }, { "epoch": 1.423600766120451, "grad_norm": 12.890625, "learning_rate": 9.542389141557705e-06, "loss": 0.9613, "step": 13379 }, { "epoch": 1.4237071717386678, "grad_norm": 12.484375, "learning_rate": 9.542353610005687e-06, "loss": 1.1545, "step": 13380 }, { "epoch": 1.4238135773568845, "grad_norm": 10.0234375, "learning_rate": 9.542318078453668e-06, "loss": 1.0553, "step": 13381 }, { "epoch": 1.423919982975101, "grad_norm": 10.296875, "learning_rate": 9.54228254690165e-06, "loss": 0.9912, "step": 13382 }, { "epoch": 1.4240263885933178, "grad_norm": 14.5390625, "learning_rate": 9.542247015349631e-06, "loss": 1.2178, "step": 13383 }, { "epoch": 1.4241327942115345, "grad_norm": 12.546875, "learning_rate": 9.542211483797613e-06, "loss": 1.0967, "step": 13384 }, { "epoch": 1.424239199829751, "grad_norm": 13.4140625, "learning_rate": 9.542175952245595e-06, "loss": 1.2265, "step": 13385 }, { "epoch": 1.4243456054479677, "grad_norm": 14.2421875, "learning_rate": 9.542140420693576e-06, "loss": 0.9312, "step": 13386 }, { "epoch": 1.4244520110661842, "grad_norm": 16.4375, "learning_rate": 9.542104889141558e-06, "loss": 1.2553, "step": 13387 }, { "epoch": 1.424558416684401, "grad_norm": 17.875, "learning_rate": 9.542069357589541e-06, "loss": 1.1726, "step": 13388 }, { "epoch": 1.4246648223026175, "grad_norm": 14.9609375, "learning_rate": 9.542033826037523e-06, "loss": 1.1756, "step": 13389 }, { "epoch": 1.4247712279208342, "grad_norm": 14.65625, "learning_rate": 9.541998294485504e-06, "loss": 1.3582, "step": 13390 }, { "epoch": 1.424877633539051, "grad_norm": 11.7265625, "learning_rate": 9.541962762933486e-06, "loss": 1.2255, "step": 13391 }, { "epoch": 1.4249840391572675, "grad_norm": 10.9921875, "learning_rate": 9.541927231381467e-06, "loss": 0.9026, "step": 13392 }, { "epoch": 1.4250904447754842, "grad_norm": 16.9375, "learning_rate": 9.541891699829449e-06, "loss": 1.2915, "step": 13393 }, { "epoch": 1.425196850393701, "grad_norm": 13.4609375, "learning_rate": 9.54185616827743e-06, "loss": 1.3448, "step": 13394 }, { "epoch": 1.4253032560119174, "grad_norm": 15.859375, "learning_rate": 9.541820636725412e-06, "loss": 1.1509, "step": 13395 }, { "epoch": 1.425409661630134, "grad_norm": 12.4765625, "learning_rate": 9.541785105173394e-06, "loss": 0.9347, "step": 13396 }, { "epoch": 1.4255160672483507, "grad_norm": 17.953125, "learning_rate": 9.541749573621376e-06, "loss": 1.1384, "step": 13397 }, { "epoch": 1.4256224728665674, "grad_norm": 13.625, "learning_rate": 9.541714042069359e-06, "loss": 1.0733, "step": 13398 }, { "epoch": 1.425728878484784, "grad_norm": 13.4296875, "learning_rate": 9.54167851051734e-06, "loss": 1.1066, "step": 13399 }, { "epoch": 1.4258352841030006, "grad_norm": 9.4375, "learning_rate": 9.541642978965322e-06, "loss": 1.0805, "step": 13400 }, { "epoch": 1.4259416897212174, "grad_norm": 9.5234375, "learning_rate": 9.541607447413304e-06, "loss": 0.9543, "step": 13401 }, { "epoch": 1.4260480953394339, "grad_norm": 18.4375, "learning_rate": 9.541571915861285e-06, "loss": 1.0619, "step": 13402 }, { "epoch": 1.4261545009576506, "grad_norm": 12.3359375, "learning_rate": 9.541536384309269e-06, "loss": 1.191, "step": 13403 }, { "epoch": 1.4262609065758671, "grad_norm": 15.5625, "learning_rate": 9.54150085275725e-06, "loss": 1.0353, "step": 13404 }, { "epoch": 1.4263673121940839, "grad_norm": 15.296875, "learning_rate": 9.541465321205232e-06, "loss": 1.6699, "step": 13405 }, { "epoch": 1.4264737178123004, "grad_norm": 14.1796875, "learning_rate": 9.541429789653213e-06, "loss": 1.3076, "step": 13406 }, { "epoch": 1.426580123430517, "grad_norm": 8.734375, "learning_rate": 9.541394258101195e-06, "loss": 0.9202, "step": 13407 }, { "epoch": 1.4266865290487338, "grad_norm": 13.6953125, "learning_rate": 9.541358726549177e-06, "loss": 1.19, "step": 13408 }, { "epoch": 1.4267929346669503, "grad_norm": 11.4609375, "learning_rate": 9.541323194997158e-06, "loss": 0.9042, "step": 13409 }, { "epoch": 1.426899340285167, "grad_norm": 13.6875, "learning_rate": 9.54128766344514e-06, "loss": 1.0961, "step": 13410 }, { "epoch": 1.4270057459033838, "grad_norm": 12.6875, "learning_rate": 9.541252131893121e-06, "loss": 1.1452, "step": 13411 }, { "epoch": 1.4271121515216003, "grad_norm": 16.484375, "learning_rate": 9.541216600341103e-06, "loss": 1.0966, "step": 13412 }, { "epoch": 1.427218557139817, "grad_norm": 12.6953125, "learning_rate": 9.541181068789086e-06, "loss": 0.929, "step": 13413 }, { "epoch": 1.4273249627580336, "grad_norm": 11.359375, "learning_rate": 9.541145537237068e-06, "loss": 1.1333, "step": 13414 }, { "epoch": 1.4274313683762503, "grad_norm": 11.3671875, "learning_rate": 9.54111000568505e-06, "loss": 1.2748, "step": 13415 }, { "epoch": 1.4275377739944668, "grad_norm": 9.671875, "learning_rate": 9.541074474133031e-06, "loss": 1.1216, "step": 13416 }, { "epoch": 1.4276441796126835, "grad_norm": 10.765625, "learning_rate": 9.541038942581013e-06, "loss": 0.9324, "step": 13417 }, { "epoch": 1.4277505852309003, "grad_norm": 9.9375, "learning_rate": 9.541003411028994e-06, "loss": 0.9176, "step": 13418 }, { "epoch": 1.4278569908491168, "grad_norm": 12.9296875, "learning_rate": 9.540967879476976e-06, "loss": 1.1042, "step": 13419 }, { "epoch": 1.4279633964673335, "grad_norm": 12.8984375, "learning_rate": 9.540932347924958e-06, "loss": 1.0048, "step": 13420 }, { "epoch": 1.4280698020855502, "grad_norm": 10.7578125, "learning_rate": 9.54089681637294e-06, "loss": 1.0495, "step": 13421 }, { "epoch": 1.4281762077037667, "grad_norm": 11.59375, "learning_rate": 9.54086128482092e-06, "loss": 0.7665, "step": 13422 }, { "epoch": 1.4282826133219835, "grad_norm": 14.9140625, "learning_rate": 9.540825753268902e-06, "loss": 1.3132, "step": 13423 }, { "epoch": 1.4283890189402, "grad_norm": 12.2734375, "learning_rate": 9.540790221716886e-06, "loss": 1.2594, "step": 13424 }, { "epoch": 1.4284954245584167, "grad_norm": 10.4453125, "learning_rate": 9.540754690164867e-06, "loss": 0.9468, "step": 13425 }, { "epoch": 1.4286018301766332, "grad_norm": 21.96875, "learning_rate": 9.540719158612849e-06, "loss": 1.2967, "step": 13426 }, { "epoch": 1.42870823579485, "grad_norm": 15.3515625, "learning_rate": 9.54068362706083e-06, "loss": 1.2033, "step": 13427 }, { "epoch": 1.4288146414130667, "grad_norm": 9.21875, "learning_rate": 9.540648095508812e-06, "loss": 0.9192, "step": 13428 }, { "epoch": 1.4289210470312832, "grad_norm": 12.734375, "learning_rate": 9.540612563956795e-06, "loss": 1.0391, "step": 13429 }, { "epoch": 1.4290274526495, "grad_norm": 16.0625, "learning_rate": 9.540577032404777e-06, "loss": 0.9535, "step": 13430 }, { "epoch": 1.4291338582677167, "grad_norm": 9.7421875, "learning_rate": 9.540541500852759e-06, "loss": 1.1421, "step": 13431 }, { "epoch": 1.4292402638859332, "grad_norm": 18.640625, "learning_rate": 9.54050596930074e-06, "loss": 1.0386, "step": 13432 }, { "epoch": 1.4293466695041497, "grad_norm": 15.4921875, "learning_rate": 9.540470437748722e-06, "loss": 1.0792, "step": 13433 }, { "epoch": 1.4294530751223664, "grad_norm": 12.7109375, "learning_rate": 9.540434906196703e-06, "loss": 1.0504, "step": 13434 }, { "epoch": 1.4295594807405831, "grad_norm": 13.3828125, "learning_rate": 9.540399374644685e-06, "loss": 0.9687, "step": 13435 }, { "epoch": 1.4296658863587997, "grad_norm": 12.0625, "learning_rate": 9.540363843092667e-06, "loss": 0.9096, "step": 13436 }, { "epoch": 1.4297722919770164, "grad_norm": 10.296875, "learning_rate": 9.540328311540648e-06, "loss": 0.8688, "step": 13437 }, { "epoch": 1.4298786975952331, "grad_norm": 13.0078125, "learning_rate": 9.54029277998863e-06, "loss": 0.8331, "step": 13438 }, { "epoch": 1.4299851032134496, "grad_norm": 12.484375, "learning_rate": 9.540257248436613e-06, "loss": 1.0726, "step": 13439 }, { "epoch": 1.4300915088316664, "grad_norm": 14.421875, "learning_rate": 9.540221716884595e-06, "loss": 1.0995, "step": 13440 }, { "epoch": 1.430197914449883, "grad_norm": 16.28125, "learning_rate": 9.540186185332576e-06, "loss": 1.1773, "step": 13441 }, { "epoch": 1.4303043200680996, "grad_norm": 13.609375, "learning_rate": 9.540150653780558e-06, "loss": 1.1006, "step": 13442 }, { "epoch": 1.430410725686316, "grad_norm": 18.40625, "learning_rate": 9.54011512222854e-06, "loss": 1.0003, "step": 13443 }, { "epoch": 1.4305171313045328, "grad_norm": 23.1875, "learning_rate": 9.540079590676521e-06, "loss": 0.8643, "step": 13444 }, { "epoch": 1.4306235369227496, "grad_norm": 10.3046875, "learning_rate": 9.540044059124503e-06, "loss": 0.9812, "step": 13445 }, { "epoch": 1.430729942540966, "grad_norm": 19.0625, "learning_rate": 9.540008527572484e-06, "loss": 1.0109, "step": 13446 }, { "epoch": 1.4308363481591828, "grad_norm": 17.109375, "learning_rate": 9.539972996020466e-06, "loss": 1.2071, "step": 13447 }, { "epoch": 1.4309427537773995, "grad_norm": 16.078125, "learning_rate": 9.539937464468448e-06, "loss": 0.971, "step": 13448 }, { "epoch": 1.431049159395616, "grad_norm": 14.5078125, "learning_rate": 9.539901932916431e-06, "loss": 1.5219, "step": 13449 }, { "epoch": 1.4311555650138328, "grad_norm": 11.0703125, "learning_rate": 9.539866401364413e-06, "loss": 0.8005, "step": 13450 }, { "epoch": 1.4312619706320493, "grad_norm": 11.421875, "learning_rate": 9.539830869812394e-06, "loss": 0.7929, "step": 13451 }, { "epoch": 1.431368376250266, "grad_norm": 9.875, "learning_rate": 9.539795338260376e-06, "loss": 1.0607, "step": 13452 }, { "epoch": 1.4314747818684825, "grad_norm": 15.1875, "learning_rate": 9.539759806708357e-06, "loss": 1.33, "step": 13453 }, { "epoch": 1.4315811874866993, "grad_norm": 15.2578125, "learning_rate": 9.53972427515634e-06, "loss": 1.1511, "step": 13454 }, { "epoch": 1.431687593104916, "grad_norm": 11.296875, "learning_rate": 9.539688743604322e-06, "loss": 1.0856, "step": 13455 }, { "epoch": 1.4317939987231325, "grad_norm": 13.09375, "learning_rate": 9.539653212052304e-06, "loss": 1.0122, "step": 13456 }, { "epoch": 1.4319004043413492, "grad_norm": 11.09375, "learning_rate": 9.539617680500285e-06, "loss": 0.7033, "step": 13457 }, { "epoch": 1.432006809959566, "grad_norm": 11.0859375, "learning_rate": 9.539582148948267e-06, "loss": 1.1425, "step": 13458 }, { "epoch": 1.4321132155777825, "grad_norm": 12.90625, "learning_rate": 9.539546617396249e-06, "loss": 0.8974, "step": 13459 }, { "epoch": 1.4322196211959992, "grad_norm": 12.5234375, "learning_rate": 9.53951108584423e-06, "loss": 1.2201, "step": 13460 }, { "epoch": 1.4323260268142157, "grad_norm": 14.875, "learning_rate": 9.539475554292212e-06, "loss": 1.2425, "step": 13461 }, { "epoch": 1.4324324324324325, "grad_norm": 13.0546875, "learning_rate": 9.539440022740194e-06, "loss": 1.1147, "step": 13462 }, { "epoch": 1.432538838050649, "grad_norm": 11.4140625, "learning_rate": 9.539404491188175e-06, "loss": 0.9527, "step": 13463 }, { "epoch": 1.4326452436688657, "grad_norm": 13.1875, "learning_rate": 9.539368959636157e-06, "loss": 1.1107, "step": 13464 }, { "epoch": 1.4327516492870824, "grad_norm": 17.609375, "learning_rate": 9.53933342808414e-06, "loss": 1.1149, "step": 13465 }, { "epoch": 1.432858054905299, "grad_norm": 11.8125, "learning_rate": 9.539297896532122e-06, "loss": 1.1848, "step": 13466 }, { "epoch": 1.4329644605235157, "grad_norm": 11.625, "learning_rate": 9.539262364980103e-06, "loss": 0.9665, "step": 13467 }, { "epoch": 1.4330708661417324, "grad_norm": 11.0625, "learning_rate": 9.539226833428085e-06, "loss": 1.1132, "step": 13468 }, { "epoch": 1.433177271759949, "grad_norm": 11.6640625, "learning_rate": 9.539191301876066e-06, "loss": 1.2527, "step": 13469 }, { "epoch": 1.4332836773781656, "grad_norm": 12.515625, "learning_rate": 9.53915577032405e-06, "loss": 0.9428, "step": 13470 }, { "epoch": 1.4333900829963822, "grad_norm": 15.640625, "learning_rate": 9.539120238772031e-06, "loss": 1.0471, "step": 13471 }, { "epoch": 1.4334964886145989, "grad_norm": 12.5390625, "learning_rate": 9.539084707220011e-06, "loss": 1.1965, "step": 13472 }, { "epoch": 1.4336028942328154, "grad_norm": 18.84375, "learning_rate": 9.539049175667993e-06, "loss": 1.2486, "step": 13473 }, { "epoch": 1.4337092998510321, "grad_norm": 15.6640625, "learning_rate": 9.539013644115974e-06, "loss": 1.1837, "step": 13474 }, { "epoch": 1.4338157054692489, "grad_norm": 10.3515625, "learning_rate": 9.538978112563958e-06, "loss": 1.1202, "step": 13475 }, { "epoch": 1.4339221110874654, "grad_norm": 13.1328125, "learning_rate": 9.53894258101194e-06, "loss": 0.9819, "step": 13476 }, { "epoch": 1.434028516705682, "grad_norm": 13.1328125, "learning_rate": 9.538907049459921e-06, "loss": 1.1711, "step": 13477 }, { "epoch": 1.4341349223238988, "grad_norm": 15.015625, "learning_rate": 9.538871517907903e-06, "loss": 1.1976, "step": 13478 }, { "epoch": 1.4342413279421153, "grad_norm": 10.1015625, "learning_rate": 9.538835986355884e-06, "loss": 0.9795, "step": 13479 }, { "epoch": 1.4343477335603319, "grad_norm": 13.09375, "learning_rate": 9.538800454803868e-06, "loss": 1.4573, "step": 13480 }, { "epoch": 1.4344541391785486, "grad_norm": 12.9296875, "learning_rate": 9.538764923251849e-06, "loss": 1.174, "step": 13481 }, { "epoch": 1.4345605447967653, "grad_norm": 12.8125, "learning_rate": 9.53872939169983e-06, "loss": 1.0967, "step": 13482 }, { "epoch": 1.4346669504149818, "grad_norm": 11.671875, "learning_rate": 9.538693860147812e-06, "loss": 0.9963, "step": 13483 }, { "epoch": 1.4347733560331986, "grad_norm": 12.46875, "learning_rate": 9.538658328595794e-06, "loss": 1.0947, "step": 13484 }, { "epoch": 1.4348797616514153, "grad_norm": 15.6484375, "learning_rate": 9.538622797043776e-06, "loss": 1.0419, "step": 13485 }, { "epoch": 1.4349861672696318, "grad_norm": 13.4453125, "learning_rate": 9.538587265491757e-06, "loss": 1.2023, "step": 13486 }, { "epoch": 1.4350925728878485, "grad_norm": 13.09375, "learning_rate": 9.538551733939739e-06, "loss": 0.9672, "step": 13487 }, { "epoch": 1.435198978506065, "grad_norm": 10.8203125, "learning_rate": 9.53851620238772e-06, "loss": 1.2392, "step": 13488 }, { "epoch": 1.4353053841242818, "grad_norm": 12.671875, "learning_rate": 9.538480670835702e-06, "loss": 1.2013, "step": 13489 }, { "epoch": 1.4354117897424983, "grad_norm": 16.84375, "learning_rate": 9.538445139283685e-06, "loss": 1.2153, "step": 13490 }, { "epoch": 1.435518195360715, "grad_norm": 13.7421875, "learning_rate": 9.538409607731667e-06, "loss": 1.0984, "step": 13491 }, { "epoch": 1.4356246009789317, "grad_norm": 13.7578125, "learning_rate": 9.538374076179648e-06, "loss": 1.2397, "step": 13492 }, { "epoch": 1.4357310065971483, "grad_norm": 13.2890625, "learning_rate": 9.53833854462763e-06, "loss": 1.1141, "step": 13493 }, { "epoch": 1.435837412215365, "grad_norm": 12.453125, "learning_rate": 9.538303013075612e-06, "loss": 1.2323, "step": 13494 }, { "epoch": 1.4359438178335817, "grad_norm": 10.5625, "learning_rate": 9.538267481523595e-06, "loss": 1.0025, "step": 13495 }, { "epoch": 1.4360502234517982, "grad_norm": 11.953125, "learning_rate": 9.538231949971577e-06, "loss": 1.0342, "step": 13496 }, { "epoch": 1.436156629070015, "grad_norm": 13.2421875, "learning_rate": 9.538196418419558e-06, "loss": 1.4052, "step": 13497 }, { "epoch": 1.4362630346882315, "grad_norm": 13.140625, "learning_rate": 9.53816088686754e-06, "loss": 1.0863, "step": 13498 }, { "epoch": 1.4363694403064482, "grad_norm": 11.1015625, "learning_rate": 9.538125355315521e-06, "loss": 0.9097, "step": 13499 }, { "epoch": 1.4364758459246647, "grad_norm": 13.609375, "learning_rate": 9.538089823763503e-06, "loss": 1.0471, "step": 13500 }, { "epoch": 1.4365822515428814, "grad_norm": 10.3515625, "learning_rate": 9.538054292211485e-06, "loss": 1.0795, "step": 13501 }, { "epoch": 1.4366886571610982, "grad_norm": 19.1875, "learning_rate": 9.538018760659466e-06, "loss": 1.0062, "step": 13502 }, { "epoch": 1.4367950627793147, "grad_norm": 10.0390625, "learning_rate": 9.537983229107448e-06, "loss": 1.0156, "step": 13503 }, { "epoch": 1.4369014683975314, "grad_norm": 8.5546875, "learning_rate": 9.53794769755543e-06, "loss": 1.0406, "step": 13504 }, { "epoch": 1.4370078740157481, "grad_norm": 18.21875, "learning_rate": 9.537912166003411e-06, "loss": 1.0847, "step": 13505 }, { "epoch": 1.4371142796339647, "grad_norm": 18.59375, "learning_rate": 9.537876634451394e-06, "loss": 1.0437, "step": 13506 }, { "epoch": 1.4372206852521814, "grad_norm": 14.3671875, "learning_rate": 9.537841102899376e-06, "loss": 1.1784, "step": 13507 }, { "epoch": 1.437327090870398, "grad_norm": 18.125, "learning_rate": 9.537805571347358e-06, "loss": 1.0018, "step": 13508 }, { "epoch": 1.4374334964886146, "grad_norm": 12.0546875, "learning_rate": 9.53777003979534e-06, "loss": 0.9696, "step": 13509 }, { "epoch": 1.4375399021068311, "grad_norm": 17.15625, "learning_rate": 9.53773450824332e-06, "loss": 1.2887, "step": 13510 }, { "epoch": 1.4376463077250479, "grad_norm": 13.765625, "learning_rate": 9.537698976691302e-06, "loss": 1.2187, "step": 13511 }, { "epoch": 1.4377527133432646, "grad_norm": 15.40625, "learning_rate": 9.537663445139284e-06, "loss": 1.2156, "step": 13512 }, { "epoch": 1.437859118961481, "grad_norm": 12.125, "learning_rate": 9.537627913587266e-06, "loss": 1.1381, "step": 13513 }, { "epoch": 1.4379655245796978, "grad_norm": 14.6484375, "learning_rate": 9.537592382035247e-06, "loss": 1.1757, "step": 13514 }, { "epoch": 1.4380719301979146, "grad_norm": 9.96875, "learning_rate": 9.537556850483229e-06, "loss": 1.1331, "step": 13515 }, { "epoch": 1.438178335816131, "grad_norm": 18.0625, "learning_rate": 9.537521318931212e-06, "loss": 1.1995, "step": 13516 }, { "epoch": 1.4382847414343478, "grad_norm": 14.03125, "learning_rate": 9.537485787379194e-06, "loss": 1.0511, "step": 13517 }, { "epoch": 1.4383911470525643, "grad_norm": 9.0546875, "learning_rate": 9.537450255827175e-06, "loss": 1.1286, "step": 13518 }, { "epoch": 1.438497552670781, "grad_norm": 9.8359375, "learning_rate": 9.537414724275157e-06, "loss": 1.0475, "step": 13519 }, { "epoch": 1.4386039582889976, "grad_norm": 13.8046875, "learning_rate": 9.537379192723139e-06, "loss": 1.225, "step": 13520 }, { "epoch": 1.4387103639072143, "grad_norm": 11.265625, "learning_rate": 9.537343661171122e-06, "loss": 1.0008, "step": 13521 }, { "epoch": 1.438816769525431, "grad_norm": 11.8203125, "learning_rate": 9.537308129619103e-06, "loss": 1.0537, "step": 13522 }, { "epoch": 1.4389231751436475, "grad_norm": 13.96875, "learning_rate": 9.537272598067085e-06, "loss": 0.9428, "step": 13523 }, { "epoch": 1.4390295807618643, "grad_norm": 15.171875, "learning_rate": 9.537237066515067e-06, "loss": 1.0347, "step": 13524 }, { "epoch": 1.439135986380081, "grad_norm": 13.0703125, "learning_rate": 9.537201534963048e-06, "loss": 1.0181, "step": 13525 }, { "epoch": 1.4392423919982975, "grad_norm": 12.8671875, "learning_rate": 9.53716600341103e-06, "loss": 1.0506, "step": 13526 }, { "epoch": 1.439348797616514, "grad_norm": 9.9453125, "learning_rate": 9.537130471859011e-06, "loss": 1.0447, "step": 13527 }, { "epoch": 1.4394552032347308, "grad_norm": 10.3125, "learning_rate": 9.537094940306993e-06, "loss": 1.0641, "step": 13528 }, { "epoch": 1.4395616088529475, "grad_norm": 12.8359375, "learning_rate": 9.537059408754975e-06, "loss": 0.9358, "step": 13529 }, { "epoch": 1.439668014471164, "grad_norm": 15.0625, "learning_rate": 9.537023877202956e-06, "loss": 1.0735, "step": 13530 }, { "epoch": 1.4397744200893807, "grad_norm": 10.359375, "learning_rate": 9.53698834565094e-06, "loss": 1.1372, "step": 13531 }, { "epoch": 1.4398808257075975, "grad_norm": 9.484375, "learning_rate": 9.536952814098921e-06, "loss": 1.1106, "step": 13532 }, { "epoch": 1.439987231325814, "grad_norm": 16.421875, "learning_rate": 9.536917282546903e-06, "loss": 0.8526, "step": 13533 }, { "epoch": 1.4400936369440307, "grad_norm": 11.0859375, "learning_rate": 9.536881750994884e-06, "loss": 1.0901, "step": 13534 }, { "epoch": 1.4402000425622472, "grad_norm": 18.15625, "learning_rate": 9.536846219442866e-06, "loss": 0.9608, "step": 13535 }, { "epoch": 1.440306448180464, "grad_norm": 10.6796875, "learning_rate": 9.536810687890848e-06, "loss": 1.1037, "step": 13536 }, { "epoch": 1.4404128537986804, "grad_norm": 14.8046875, "learning_rate": 9.53677515633883e-06, "loss": 0.99, "step": 13537 }, { "epoch": 1.4405192594168972, "grad_norm": 13.4765625, "learning_rate": 9.536739624786811e-06, "loss": 1.249, "step": 13538 }, { "epoch": 1.440625665035114, "grad_norm": 9.0625, "learning_rate": 9.536704093234792e-06, "loss": 1.1355, "step": 13539 }, { "epoch": 1.4407320706533304, "grad_norm": 8.6796875, "learning_rate": 9.536668561682774e-06, "loss": 0.9069, "step": 13540 }, { "epoch": 1.4408384762715472, "grad_norm": 13.484375, "learning_rate": 9.536633030130756e-06, "loss": 0.9385, "step": 13541 }, { "epoch": 1.4409448818897639, "grad_norm": 12.96875, "learning_rate": 9.536597498578739e-06, "loss": 1.1627, "step": 13542 }, { "epoch": 1.4410512875079804, "grad_norm": 13.34375, "learning_rate": 9.53656196702672e-06, "loss": 0.9662, "step": 13543 }, { "epoch": 1.4411576931261971, "grad_norm": 17.078125, "learning_rate": 9.536526435474702e-06, "loss": 0.8341, "step": 13544 }, { "epoch": 1.4412640987444136, "grad_norm": 9.421875, "learning_rate": 9.536490903922684e-06, "loss": 0.8954, "step": 13545 }, { "epoch": 1.4413705043626304, "grad_norm": 10.4375, "learning_rate": 9.536455372370665e-06, "loss": 1.1794, "step": 13546 }, { "epoch": 1.4414769099808469, "grad_norm": 17.796875, "learning_rate": 9.536419840818649e-06, "loss": 0.8249, "step": 13547 }, { "epoch": 1.4415833155990636, "grad_norm": 10.875, "learning_rate": 9.53638430926663e-06, "loss": 1.355, "step": 13548 }, { "epoch": 1.4416897212172803, "grad_norm": 12.875, "learning_rate": 9.536348777714612e-06, "loss": 1.0075, "step": 13549 }, { "epoch": 1.4417961268354968, "grad_norm": 13.5234375, "learning_rate": 9.536313246162594e-06, "loss": 0.9919, "step": 13550 }, { "epoch": 1.4419025324537136, "grad_norm": 21.375, "learning_rate": 9.536277714610575e-06, "loss": 1.0477, "step": 13551 }, { "epoch": 1.4420089380719303, "grad_norm": 11.203125, "learning_rate": 9.536242183058557e-06, "loss": 1.1131, "step": 13552 }, { "epoch": 1.4421153436901468, "grad_norm": 13.296875, "learning_rate": 9.536206651506538e-06, "loss": 1.0059, "step": 13553 }, { "epoch": 1.4422217493083636, "grad_norm": 14.5, "learning_rate": 9.53617111995452e-06, "loss": 1.0396, "step": 13554 }, { "epoch": 1.44232815492658, "grad_norm": 12.046875, "learning_rate": 9.536135588402502e-06, "loss": 1.1905, "step": 13555 }, { "epoch": 1.4424345605447968, "grad_norm": 10.015625, "learning_rate": 9.536100056850483e-06, "loss": 1.0881, "step": 13556 }, { "epoch": 1.4425409661630133, "grad_norm": 16.96875, "learning_rate": 9.536064525298466e-06, "loss": 0.8858, "step": 13557 }, { "epoch": 1.44264737178123, "grad_norm": 13.828125, "learning_rate": 9.536028993746448e-06, "loss": 1.1583, "step": 13558 }, { "epoch": 1.4427537773994468, "grad_norm": 18.90625, "learning_rate": 9.53599346219443e-06, "loss": 1.0707, "step": 13559 }, { "epoch": 1.4428601830176633, "grad_norm": 13.03125, "learning_rate": 9.535957930642411e-06, "loss": 1.0574, "step": 13560 }, { "epoch": 1.44296658863588, "grad_norm": 13.125, "learning_rate": 9.535922399090393e-06, "loss": 0.8317, "step": 13561 }, { "epoch": 1.4430729942540967, "grad_norm": 15.15625, "learning_rate": 9.535886867538375e-06, "loss": 1.0477, "step": 13562 }, { "epoch": 1.4431793998723133, "grad_norm": 14.953125, "learning_rate": 9.535851335986356e-06, "loss": 1.0303, "step": 13563 }, { "epoch": 1.44328580549053, "grad_norm": 10.421875, "learning_rate": 9.535815804434338e-06, "loss": 1.0517, "step": 13564 }, { "epoch": 1.4433922111087465, "grad_norm": 12.3125, "learning_rate": 9.53578027288232e-06, "loss": 1.0582, "step": 13565 }, { "epoch": 1.4434986167269632, "grad_norm": 16.140625, "learning_rate": 9.535744741330301e-06, "loss": 0.9395, "step": 13566 }, { "epoch": 1.4436050223451797, "grad_norm": 11.8828125, "learning_rate": 9.535709209778284e-06, "loss": 1.008, "step": 13567 }, { "epoch": 1.4437114279633965, "grad_norm": 9.9453125, "learning_rate": 9.535673678226266e-06, "loss": 1.1043, "step": 13568 }, { "epoch": 1.4438178335816132, "grad_norm": 11.3203125, "learning_rate": 9.535638146674247e-06, "loss": 1.1894, "step": 13569 }, { "epoch": 1.4439242391998297, "grad_norm": 14.8515625, "learning_rate": 9.535602615122229e-06, "loss": 0.9908, "step": 13570 }, { "epoch": 1.4440306448180464, "grad_norm": 16.3125, "learning_rate": 9.53556708357021e-06, "loss": 1.1989, "step": 13571 }, { "epoch": 1.4441370504362632, "grad_norm": 12.2734375, "learning_rate": 9.535531552018194e-06, "loss": 1.0681, "step": 13572 }, { "epoch": 1.4442434560544797, "grad_norm": 12.640625, "learning_rate": 9.535496020466176e-06, "loss": 1.2984, "step": 13573 }, { "epoch": 1.4443498616726962, "grad_norm": 12.671875, "learning_rate": 9.535460488914157e-06, "loss": 0.9963, "step": 13574 }, { "epoch": 1.444456267290913, "grad_norm": 13.2734375, "learning_rate": 9.535424957362139e-06, "loss": 1.1004, "step": 13575 }, { "epoch": 1.4445626729091297, "grad_norm": 9.4765625, "learning_rate": 9.53538942581012e-06, "loss": 0.7958, "step": 13576 }, { "epoch": 1.4446690785273462, "grad_norm": 10.34375, "learning_rate": 9.535353894258102e-06, "loss": 0.9571, "step": 13577 }, { "epoch": 1.444775484145563, "grad_norm": 15.8515625, "learning_rate": 9.535318362706084e-06, "loss": 0.9935, "step": 13578 }, { "epoch": 1.4448818897637796, "grad_norm": 13.28125, "learning_rate": 9.535282831154065e-06, "loss": 0.8907, "step": 13579 }, { "epoch": 1.4449882953819961, "grad_norm": 14.0390625, "learning_rate": 9.535247299602047e-06, "loss": 1.0571, "step": 13580 }, { "epoch": 1.4450947010002129, "grad_norm": 12.2265625, "learning_rate": 9.535211768050028e-06, "loss": 1.1932, "step": 13581 }, { "epoch": 1.4452011066184294, "grad_norm": 9.6328125, "learning_rate": 9.53517623649801e-06, "loss": 1.0226, "step": 13582 }, { "epoch": 1.445307512236646, "grad_norm": 14.171875, "learning_rate": 9.535140704945993e-06, "loss": 1.087, "step": 13583 }, { "epoch": 1.4454139178548626, "grad_norm": 18.3125, "learning_rate": 9.535105173393975e-06, "loss": 1.0054, "step": 13584 }, { "epoch": 1.4455203234730793, "grad_norm": 11.109375, "learning_rate": 9.535069641841957e-06, "loss": 1.1727, "step": 13585 }, { "epoch": 1.445626729091296, "grad_norm": 11.390625, "learning_rate": 9.535034110289938e-06, "loss": 1.0484, "step": 13586 }, { "epoch": 1.4457331347095126, "grad_norm": 9.2890625, "learning_rate": 9.53499857873792e-06, "loss": 1.3165, "step": 13587 }, { "epoch": 1.4458395403277293, "grad_norm": 10.78125, "learning_rate": 9.534963047185903e-06, "loss": 1.0664, "step": 13588 }, { "epoch": 1.445945945945946, "grad_norm": 13.6953125, "learning_rate": 9.534927515633885e-06, "loss": 1.0982, "step": 13589 }, { "epoch": 1.4460523515641626, "grad_norm": 13.3359375, "learning_rate": 9.534891984081866e-06, "loss": 1.0224, "step": 13590 }, { "epoch": 1.4461587571823793, "grad_norm": 15.1875, "learning_rate": 9.534856452529848e-06, "loss": 1.24, "step": 13591 }, { "epoch": 1.4462651628005958, "grad_norm": 11.84375, "learning_rate": 9.534820920977828e-06, "loss": 0.9145, "step": 13592 }, { "epoch": 1.4463715684188125, "grad_norm": 10.7109375, "learning_rate": 9.534785389425811e-06, "loss": 1.0877, "step": 13593 }, { "epoch": 1.446477974037029, "grad_norm": 14.359375, "learning_rate": 9.534749857873793e-06, "loss": 1.326, "step": 13594 }, { "epoch": 1.4465843796552458, "grad_norm": 12.7734375, "learning_rate": 9.534714326321774e-06, "loss": 1.0674, "step": 13595 }, { "epoch": 1.4466907852734625, "grad_norm": 15.71875, "learning_rate": 9.534678794769756e-06, "loss": 1.0982, "step": 13596 }, { "epoch": 1.446797190891679, "grad_norm": 15.6015625, "learning_rate": 9.534643263217738e-06, "loss": 1.3406, "step": 13597 }, { "epoch": 1.4469035965098958, "grad_norm": 12.0625, "learning_rate": 9.53460773166572e-06, "loss": 1.0427, "step": 13598 }, { "epoch": 1.4470100021281125, "grad_norm": 15.3359375, "learning_rate": 9.534572200113702e-06, "loss": 1.2836, "step": 13599 }, { "epoch": 1.447116407746329, "grad_norm": 10.125, "learning_rate": 9.534536668561684e-06, "loss": 1.1461, "step": 13600 }, { "epoch": 1.4472228133645457, "grad_norm": 17.140625, "learning_rate": 9.534501137009666e-06, "loss": 1.0119, "step": 13601 }, { "epoch": 1.4473292189827622, "grad_norm": 14.1171875, "learning_rate": 9.534465605457647e-06, "loss": 1.0125, "step": 13602 }, { "epoch": 1.447435624600979, "grad_norm": 14.890625, "learning_rate": 9.534430073905629e-06, "loss": 1.1483, "step": 13603 }, { "epoch": 1.4475420302191955, "grad_norm": 12.875, "learning_rate": 9.53439454235361e-06, "loss": 1.0759, "step": 13604 }, { "epoch": 1.4476484358374122, "grad_norm": 11.7109375, "learning_rate": 9.534359010801592e-06, "loss": 1.1877, "step": 13605 }, { "epoch": 1.447754841455629, "grad_norm": 13.96875, "learning_rate": 9.534323479249574e-06, "loss": 1.0264, "step": 13606 }, { "epoch": 1.4478612470738454, "grad_norm": 11.0078125, "learning_rate": 9.534287947697555e-06, "loss": 1.1941, "step": 13607 }, { "epoch": 1.4479676526920622, "grad_norm": 12.734375, "learning_rate": 9.534252416145539e-06, "loss": 1.136, "step": 13608 }, { "epoch": 1.448074058310279, "grad_norm": 11.1640625, "learning_rate": 9.53421688459352e-06, "loss": 1.0054, "step": 13609 }, { "epoch": 1.4481804639284954, "grad_norm": 14.71875, "learning_rate": 9.534181353041502e-06, "loss": 1.0615, "step": 13610 }, { "epoch": 1.448286869546712, "grad_norm": 9.109375, "learning_rate": 9.534145821489483e-06, "loss": 0.8669, "step": 13611 }, { "epoch": 1.4483932751649287, "grad_norm": 13.7734375, "learning_rate": 9.534110289937465e-06, "loss": 1.3328, "step": 13612 }, { "epoch": 1.4484996807831454, "grad_norm": 10.2265625, "learning_rate": 9.534074758385448e-06, "loss": 0.8696, "step": 13613 }, { "epoch": 1.448606086401362, "grad_norm": 14.8203125, "learning_rate": 9.53403922683343e-06, "loss": 0.8224, "step": 13614 }, { "epoch": 1.4487124920195786, "grad_norm": 11.359375, "learning_rate": 9.534003695281412e-06, "loss": 1.0288, "step": 13615 }, { "epoch": 1.4488188976377954, "grad_norm": 12.2890625, "learning_rate": 9.533968163729393e-06, "loss": 1.1295, "step": 13616 }, { "epoch": 1.4489253032560119, "grad_norm": 9.765625, "learning_rate": 9.533932632177375e-06, "loss": 0.8947, "step": 13617 }, { "epoch": 1.4490317088742286, "grad_norm": 11.8125, "learning_rate": 9.533897100625356e-06, "loss": 1.2023, "step": 13618 }, { "epoch": 1.4491381144924453, "grad_norm": 13.796875, "learning_rate": 9.533861569073338e-06, "loss": 0.9812, "step": 13619 }, { "epoch": 1.4492445201106618, "grad_norm": 14.109375, "learning_rate": 9.53382603752132e-06, "loss": 0.907, "step": 13620 }, { "epoch": 1.4493509257288784, "grad_norm": 13.75, "learning_rate": 9.533790505969301e-06, "loss": 1.0933, "step": 13621 }, { "epoch": 1.449457331347095, "grad_norm": 9.328125, "learning_rate": 9.533754974417283e-06, "loss": 1.063, "step": 13622 }, { "epoch": 1.4495637369653118, "grad_norm": 12.390625, "learning_rate": 9.533719442865264e-06, "loss": 1.2114, "step": 13623 }, { "epoch": 1.4496701425835283, "grad_norm": 13.0859375, "learning_rate": 9.533683911313248e-06, "loss": 1.3667, "step": 13624 }, { "epoch": 1.449776548201745, "grad_norm": 13.90625, "learning_rate": 9.53364837976123e-06, "loss": 1.0894, "step": 13625 }, { "epoch": 1.4498829538199618, "grad_norm": 11.640625, "learning_rate": 9.533612848209211e-06, "loss": 1.1092, "step": 13626 }, { "epoch": 1.4499893594381783, "grad_norm": 11.046875, "learning_rate": 9.533577316657192e-06, "loss": 1.1614, "step": 13627 }, { "epoch": 1.450095765056395, "grad_norm": 15.4609375, "learning_rate": 9.533541785105174e-06, "loss": 1.1322, "step": 13628 }, { "epoch": 1.4502021706746115, "grad_norm": 12.859375, "learning_rate": 9.533506253553156e-06, "loss": 1.3577, "step": 13629 }, { "epoch": 1.4503085762928283, "grad_norm": 16.171875, "learning_rate": 9.533470722001137e-06, "loss": 0.8205, "step": 13630 }, { "epoch": 1.4504149819110448, "grad_norm": 14.828125, "learning_rate": 9.533435190449119e-06, "loss": 1.2262, "step": 13631 }, { "epoch": 1.4505213875292615, "grad_norm": 13.703125, "learning_rate": 9.5333996588971e-06, "loss": 1.369, "step": 13632 }, { "epoch": 1.4506277931474783, "grad_norm": 11.171875, "learning_rate": 9.533364127345082e-06, "loss": 1.087, "step": 13633 }, { "epoch": 1.4507341987656948, "grad_norm": 14.078125, "learning_rate": 9.533328595793065e-06, "loss": 1.2414, "step": 13634 }, { "epoch": 1.4508406043839115, "grad_norm": 16.125, "learning_rate": 9.533293064241047e-06, "loss": 1.0413, "step": 13635 }, { "epoch": 1.4509470100021282, "grad_norm": 11.7890625, "learning_rate": 9.533257532689029e-06, "loss": 0.9901, "step": 13636 }, { "epoch": 1.4510534156203447, "grad_norm": 12.9765625, "learning_rate": 9.53322200113701e-06, "loss": 0.9502, "step": 13637 }, { "epoch": 1.4511598212385615, "grad_norm": 16.34375, "learning_rate": 9.533186469584992e-06, "loss": 1.0938, "step": 13638 }, { "epoch": 1.451266226856778, "grad_norm": 13.25, "learning_rate": 9.533150938032975e-06, "loss": 0.9368, "step": 13639 }, { "epoch": 1.4513726324749947, "grad_norm": 17.46875, "learning_rate": 9.533115406480957e-06, "loss": 1.0183, "step": 13640 }, { "epoch": 1.4514790380932112, "grad_norm": 10.7109375, "learning_rate": 9.533079874928938e-06, "loss": 0.912, "step": 13641 }, { "epoch": 1.451585443711428, "grad_norm": 13.453125, "learning_rate": 9.53304434337692e-06, "loss": 1.2424, "step": 13642 }, { "epoch": 1.4516918493296447, "grad_norm": 12.171875, "learning_rate": 9.533008811824902e-06, "loss": 1.3358, "step": 13643 }, { "epoch": 1.4517982549478612, "grad_norm": 11.5625, "learning_rate": 9.532973280272883e-06, "loss": 0.8051, "step": 13644 }, { "epoch": 1.451904660566078, "grad_norm": 8.8828125, "learning_rate": 9.532937748720865e-06, "loss": 1.0243, "step": 13645 }, { "epoch": 1.4520110661842947, "grad_norm": 14.953125, "learning_rate": 9.532902217168846e-06, "loss": 1.0752, "step": 13646 }, { "epoch": 1.4521174718025112, "grad_norm": 15.4921875, "learning_rate": 9.532866685616828e-06, "loss": 1.047, "step": 13647 }, { "epoch": 1.452223877420728, "grad_norm": 14.5703125, "learning_rate": 9.53283115406481e-06, "loss": 1.3349, "step": 13648 }, { "epoch": 1.4523302830389444, "grad_norm": 12.7421875, "learning_rate": 9.532795622512793e-06, "loss": 1.0937, "step": 13649 }, { "epoch": 1.4524366886571611, "grad_norm": 12.234375, "learning_rate": 9.532760090960775e-06, "loss": 1.1332, "step": 13650 }, { "epoch": 1.4525430942753776, "grad_norm": 17.71875, "learning_rate": 9.532724559408756e-06, "loss": 1.1952, "step": 13651 }, { "epoch": 1.4526494998935944, "grad_norm": 13.4765625, "learning_rate": 9.532689027856738e-06, "loss": 1.0444, "step": 13652 }, { "epoch": 1.452755905511811, "grad_norm": 17.4375, "learning_rate": 9.53265349630472e-06, "loss": 1.4583, "step": 13653 }, { "epoch": 1.4528623111300276, "grad_norm": 14.09375, "learning_rate": 9.532617964752701e-06, "loss": 1.2117, "step": 13654 }, { "epoch": 1.4529687167482443, "grad_norm": 12.4609375, "learning_rate": 9.532582433200683e-06, "loss": 0.9997, "step": 13655 }, { "epoch": 1.453075122366461, "grad_norm": 11.265625, "learning_rate": 9.532546901648664e-06, "loss": 1.1213, "step": 13656 }, { "epoch": 1.4531815279846776, "grad_norm": 14.828125, "learning_rate": 9.532511370096646e-06, "loss": 0.9851, "step": 13657 }, { "epoch": 1.453287933602894, "grad_norm": 18.9375, "learning_rate": 9.532475838544627e-06, "loss": 1.3002, "step": 13658 }, { "epoch": 1.4533943392211108, "grad_norm": 20.734375, "learning_rate": 9.532440306992609e-06, "loss": 1.0829, "step": 13659 }, { "epoch": 1.4535007448393276, "grad_norm": 14.5390625, "learning_rate": 9.532404775440592e-06, "loss": 1.1662, "step": 13660 }, { "epoch": 1.453607150457544, "grad_norm": 12.59375, "learning_rate": 9.532369243888574e-06, "loss": 1.233, "step": 13661 }, { "epoch": 1.4537135560757608, "grad_norm": 10.21875, "learning_rate": 9.532333712336556e-06, "loss": 0.7532, "step": 13662 }, { "epoch": 1.4538199616939775, "grad_norm": 10.6796875, "learning_rate": 9.532298180784537e-06, "loss": 0.9796, "step": 13663 }, { "epoch": 1.453926367312194, "grad_norm": 15.3125, "learning_rate": 9.532262649232519e-06, "loss": 1.0562, "step": 13664 }, { "epoch": 1.4540327729304108, "grad_norm": 10.234375, "learning_rate": 9.532227117680502e-06, "loss": 1.1353, "step": 13665 }, { "epoch": 1.4541391785486273, "grad_norm": 14.1953125, "learning_rate": 9.532191586128484e-06, "loss": 1.1403, "step": 13666 }, { "epoch": 1.454245584166844, "grad_norm": 10.875, "learning_rate": 9.532156054576465e-06, "loss": 1.1025, "step": 13667 }, { "epoch": 1.4543519897850605, "grad_norm": 16.4375, "learning_rate": 9.532120523024447e-06, "loss": 1.6197, "step": 13668 }, { "epoch": 1.4544583954032773, "grad_norm": 13.6328125, "learning_rate": 9.532084991472428e-06, "loss": 1.0943, "step": 13669 }, { "epoch": 1.454564801021494, "grad_norm": 13.8359375, "learning_rate": 9.53204945992041e-06, "loss": 1.1892, "step": 13670 }, { "epoch": 1.4546712066397105, "grad_norm": 11.8828125, "learning_rate": 9.532013928368392e-06, "loss": 1.2046, "step": 13671 }, { "epoch": 1.4547776122579272, "grad_norm": 10.8203125, "learning_rate": 9.531978396816373e-06, "loss": 1.0191, "step": 13672 }, { "epoch": 1.454884017876144, "grad_norm": 13.9296875, "learning_rate": 9.531942865264355e-06, "loss": 0.9839, "step": 13673 }, { "epoch": 1.4549904234943605, "grad_norm": 11.9375, "learning_rate": 9.531907333712336e-06, "loss": 1.2286, "step": 13674 }, { "epoch": 1.4550968291125772, "grad_norm": 13.15625, "learning_rate": 9.53187180216032e-06, "loss": 1.4643, "step": 13675 }, { "epoch": 1.4552032347307937, "grad_norm": 11.3984375, "learning_rate": 9.531836270608301e-06, "loss": 0.9476, "step": 13676 }, { "epoch": 1.4553096403490104, "grad_norm": 16.265625, "learning_rate": 9.531800739056283e-06, "loss": 1.0624, "step": 13677 }, { "epoch": 1.455416045967227, "grad_norm": 11.796875, "learning_rate": 9.531765207504265e-06, "loss": 1.0199, "step": 13678 }, { "epoch": 1.4555224515854437, "grad_norm": 17.515625, "learning_rate": 9.531729675952246e-06, "loss": 1.001, "step": 13679 }, { "epoch": 1.4556288572036604, "grad_norm": 9.2265625, "learning_rate": 9.53169414440023e-06, "loss": 1.054, "step": 13680 }, { "epoch": 1.455735262821877, "grad_norm": 11.3359375, "learning_rate": 9.53165861284821e-06, "loss": 1.1813, "step": 13681 }, { "epoch": 1.4558416684400937, "grad_norm": 18.25, "learning_rate": 9.531623081296191e-06, "loss": 0.9465, "step": 13682 }, { "epoch": 1.4559480740583104, "grad_norm": 12.8515625, "learning_rate": 9.531587549744173e-06, "loss": 0.9169, "step": 13683 }, { "epoch": 1.456054479676527, "grad_norm": 12.2890625, "learning_rate": 9.531552018192154e-06, "loss": 1.0739, "step": 13684 }, { "epoch": 1.4561608852947436, "grad_norm": 10.9140625, "learning_rate": 9.531516486640138e-06, "loss": 0.9257, "step": 13685 }, { "epoch": 1.4562672909129601, "grad_norm": 12.1328125, "learning_rate": 9.531480955088119e-06, "loss": 1.1448, "step": 13686 }, { "epoch": 1.4563736965311769, "grad_norm": 11.1640625, "learning_rate": 9.5314454235361e-06, "loss": 1.1802, "step": 13687 }, { "epoch": 1.4564801021493934, "grad_norm": 10.359375, "learning_rate": 9.531409891984082e-06, "loss": 0.9375, "step": 13688 }, { "epoch": 1.4565865077676101, "grad_norm": 22.59375, "learning_rate": 9.531374360432064e-06, "loss": 1.6374, "step": 13689 }, { "epoch": 1.4566929133858268, "grad_norm": 9.984375, "learning_rate": 9.531338828880047e-06, "loss": 1.1989, "step": 13690 }, { "epoch": 1.4567993190040434, "grad_norm": 9.3359375, "learning_rate": 9.531303297328029e-06, "loss": 0.999, "step": 13691 }, { "epoch": 1.45690572462226, "grad_norm": 16.125, "learning_rate": 9.53126776577601e-06, "loss": 1.3452, "step": 13692 }, { "epoch": 1.4570121302404768, "grad_norm": 9.34375, "learning_rate": 9.531232234223992e-06, "loss": 0.906, "step": 13693 }, { "epoch": 1.4571185358586933, "grad_norm": 12.0390625, "learning_rate": 9.531196702671974e-06, "loss": 1.0793, "step": 13694 }, { "epoch": 1.45722494147691, "grad_norm": 13.6171875, "learning_rate": 9.531161171119955e-06, "loss": 1.0964, "step": 13695 }, { "epoch": 1.4573313470951266, "grad_norm": 12.3984375, "learning_rate": 9.531125639567937e-06, "loss": 1.2304, "step": 13696 }, { "epoch": 1.4574377527133433, "grad_norm": 17.09375, "learning_rate": 9.531090108015919e-06, "loss": 1.0101, "step": 13697 }, { "epoch": 1.4575441583315598, "grad_norm": 11.0703125, "learning_rate": 9.5310545764639e-06, "loss": 0.8834, "step": 13698 }, { "epoch": 1.4576505639497765, "grad_norm": 13.375, "learning_rate": 9.531019044911882e-06, "loss": 1.0416, "step": 13699 }, { "epoch": 1.4577569695679933, "grad_norm": 15.25, "learning_rate": 9.530983513359863e-06, "loss": 0.9597, "step": 13700 }, { "epoch": 1.4578633751862098, "grad_norm": 18.8125, "learning_rate": 9.530947981807847e-06, "loss": 1.0789, "step": 13701 }, { "epoch": 1.4579697808044265, "grad_norm": 13.46875, "learning_rate": 9.530912450255828e-06, "loss": 1.2394, "step": 13702 }, { "epoch": 1.4580761864226433, "grad_norm": 11.9375, "learning_rate": 9.53087691870381e-06, "loss": 1.318, "step": 13703 }, { "epoch": 1.4581825920408598, "grad_norm": 10.6796875, "learning_rate": 9.530841387151791e-06, "loss": 1.0471, "step": 13704 }, { "epoch": 1.4582889976590763, "grad_norm": 12.40625, "learning_rate": 9.530805855599773e-06, "loss": 1.3484, "step": 13705 }, { "epoch": 1.458395403277293, "grad_norm": 14.0625, "learning_rate": 9.530770324047756e-06, "loss": 1.2418, "step": 13706 }, { "epoch": 1.4585018088955097, "grad_norm": 12.9765625, "learning_rate": 9.530734792495738e-06, "loss": 1.1435, "step": 13707 }, { "epoch": 1.4586082145137262, "grad_norm": 13.6328125, "learning_rate": 9.53069926094372e-06, "loss": 1.1668, "step": 13708 }, { "epoch": 1.458714620131943, "grad_norm": 17.09375, "learning_rate": 9.530663729391701e-06, "loss": 1.2657, "step": 13709 }, { "epoch": 1.4588210257501597, "grad_norm": 10.3515625, "learning_rate": 9.530628197839683e-06, "loss": 0.9455, "step": 13710 }, { "epoch": 1.4589274313683762, "grad_norm": 12.5703125, "learning_rate": 9.530592666287664e-06, "loss": 0.971, "step": 13711 }, { "epoch": 1.459033836986593, "grad_norm": 13.1328125, "learning_rate": 9.530557134735646e-06, "loss": 1.1872, "step": 13712 }, { "epoch": 1.4591402426048095, "grad_norm": 11.609375, "learning_rate": 9.530521603183628e-06, "loss": 0.9922, "step": 13713 }, { "epoch": 1.4592466482230262, "grad_norm": 11.6484375, "learning_rate": 9.53048607163161e-06, "loss": 1.019, "step": 13714 }, { "epoch": 1.4593530538412427, "grad_norm": 12.515625, "learning_rate": 9.53045054007959e-06, "loss": 1.094, "step": 13715 }, { "epoch": 1.4594594594594594, "grad_norm": 14.015625, "learning_rate": 9.530415008527574e-06, "loss": 1.1965, "step": 13716 }, { "epoch": 1.4595658650776762, "grad_norm": 13.4921875, "learning_rate": 9.530379476975556e-06, "loss": 1.0679, "step": 13717 }, { "epoch": 1.4596722706958927, "grad_norm": 11.328125, "learning_rate": 9.530343945423537e-06, "loss": 0.904, "step": 13718 }, { "epoch": 1.4597786763141094, "grad_norm": 10.03125, "learning_rate": 9.530308413871519e-06, "loss": 1.2751, "step": 13719 }, { "epoch": 1.4598850819323261, "grad_norm": 11.4609375, "learning_rate": 9.5302728823195e-06, "loss": 1.0458, "step": 13720 }, { "epoch": 1.4599914875505426, "grad_norm": 11.4453125, "learning_rate": 9.530237350767482e-06, "loss": 1.1573, "step": 13721 }, { "epoch": 1.4600978931687594, "grad_norm": 17.203125, "learning_rate": 9.530201819215464e-06, "loss": 0.9386, "step": 13722 }, { "epoch": 1.4602042987869759, "grad_norm": 14.03125, "learning_rate": 9.530166287663445e-06, "loss": 1.1116, "step": 13723 }, { "epoch": 1.4603107044051926, "grad_norm": 18.4375, "learning_rate": 9.530130756111427e-06, "loss": 1.1956, "step": 13724 }, { "epoch": 1.4604171100234091, "grad_norm": 23.5625, "learning_rate": 9.530095224559409e-06, "loss": 1.1087, "step": 13725 }, { "epoch": 1.4605235156416259, "grad_norm": 10.9765625, "learning_rate": 9.530059693007392e-06, "loss": 1.2804, "step": 13726 }, { "epoch": 1.4606299212598426, "grad_norm": 9.546875, "learning_rate": 9.530024161455373e-06, "loss": 1.0838, "step": 13727 }, { "epoch": 1.460736326878059, "grad_norm": 12.0625, "learning_rate": 9.529988629903355e-06, "loss": 1.5344, "step": 13728 }, { "epoch": 1.4608427324962758, "grad_norm": 16.25, "learning_rate": 9.529953098351337e-06, "loss": 1.0461, "step": 13729 }, { "epoch": 1.4609491381144926, "grad_norm": 14.5625, "learning_rate": 9.529917566799318e-06, "loss": 1.0188, "step": 13730 }, { "epoch": 1.461055543732709, "grad_norm": 12.515625, "learning_rate": 9.529882035247302e-06, "loss": 1.0746, "step": 13731 }, { "epoch": 1.4611619493509258, "grad_norm": 14.1640625, "learning_rate": 9.529846503695283e-06, "loss": 1.1215, "step": 13732 }, { "epoch": 1.4612683549691423, "grad_norm": 12.8125, "learning_rate": 9.529810972143265e-06, "loss": 1.199, "step": 13733 }, { "epoch": 1.461374760587359, "grad_norm": 14.875, "learning_rate": 9.529775440591246e-06, "loss": 0.887, "step": 13734 }, { "epoch": 1.4614811662055756, "grad_norm": 15.6171875, "learning_rate": 9.529739909039228e-06, "loss": 0.8869, "step": 13735 }, { "epoch": 1.4615875718237923, "grad_norm": 10.265625, "learning_rate": 9.52970437748721e-06, "loss": 1.2294, "step": 13736 }, { "epoch": 1.461693977442009, "grad_norm": 10.0, "learning_rate": 9.529668845935191e-06, "loss": 1.0718, "step": 13737 }, { "epoch": 1.4618003830602255, "grad_norm": 12.2578125, "learning_rate": 9.529633314383173e-06, "loss": 1.0617, "step": 13738 }, { "epoch": 1.4619067886784423, "grad_norm": 17.421875, "learning_rate": 9.529597782831154e-06, "loss": 1.099, "step": 13739 }, { "epoch": 1.462013194296659, "grad_norm": 11.8359375, "learning_rate": 9.529562251279136e-06, "loss": 0.9422, "step": 13740 }, { "epoch": 1.4621195999148755, "grad_norm": 12.8203125, "learning_rate": 9.529526719727118e-06, "loss": 1.0687, "step": 13741 }, { "epoch": 1.462226005533092, "grad_norm": 9.9765625, "learning_rate": 9.529491188175101e-06, "loss": 1.0412, "step": 13742 }, { "epoch": 1.4623324111513087, "grad_norm": 9.8515625, "learning_rate": 9.529455656623083e-06, "loss": 1.1756, "step": 13743 }, { "epoch": 1.4624388167695255, "grad_norm": 12.515625, "learning_rate": 9.529420125071064e-06, "loss": 1.1393, "step": 13744 }, { "epoch": 1.462545222387742, "grad_norm": 17.984375, "learning_rate": 9.529384593519046e-06, "loss": 1.3235, "step": 13745 }, { "epoch": 1.4626516280059587, "grad_norm": 10.9921875, "learning_rate": 9.529349061967027e-06, "loss": 1.0669, "step": 13746 }, { "epoch": 1.4627580336241754, "grad_norm": 13.109375, "learning_rate": 9.529313530415009e-06, "loss": 0.9852, "step": 13747 }, { "epoch": 1.462864439242392, "grad_norm": 13.8671875, "learning_rate": 9.52927799886299e-06, "loss": 1.1945, "step": 13748 }, { "epoch": 1.4629708448606087, "grad_norm": 10.1484375, "learning_rate": 9.529242467310972e-06, "loss": 1.0297, "step": 13749 }, { "epoch": 1.4630772504788254, "grad_norm": 13.3359375, "learning_rate": 9.529206935758954e-06, "loss": 1.1673, "step": 13750 }, { "epoch": 1.463183656097042, "grad_norm": 10.3984375, "learning_rate": 9.529171404206935e-06, "loss": 1.0899, "step": 13751 }, { "epoch": 1.4632900617152584, "grad_norm": 17.15625, "learning_rate": 9.529135872654919e-06, "loss": 1.2601, "step": 13752 }, { "epoch": 1.4633964673334752, "grad_norm": 11.09375, "learning_rate": 9.5291003411029e-06, "loss": 0.9196, "step": 13753 }, { "epoch": 1.463502872951692, "grad_norm": 10.2421875, "learning_rate": 9.529064809550882e-06, "loss": 1.0366, "step": 13754 }, { "epoch": 1.4636092785699084, "grad_norm": 14.6328125, "learning_rate": 9.529029277998864e-06, "loss": 1.0223, "step": 13755 }, { "epoch": 1.4637156841881251, "grad_norm": 12.5546875, "learning_rate": 9.528993746446845e-06, "loss": 0.9349, "step": 13756 }, { "epoch": 1.4638220898063419, "grad_norm": 11.140625, "learning_rate": 9.528958214894828e-06, "loss": 1.0571, "step": 13757 }, { "epoch": 1.4639284954245584, "grad_norm": 11.703125, "learning_rate": 9.52892268334281e-06, "loss": 0.9976, "step": 13758 }, { "epoch": 1.4640349010427751, "grad_norm": 13.1484375, "learning_rate": 9.528887151790792e-06, "loss": 1.2307, "step": 13759 }, { "epoch": 1.4641413066609916, "grad_norm": 10.328125, "learning_rate": 9.528851620238773e-06, "loss": 1.0305, "step": 13760 }, { "epoch": 1.4642477122792084, "grad_norm": 13.2265625, "learning_rate": 9.528816088686755e-06, "loss": 1.2362, "step": 13761 }, { "epoch": 1.4643541178974249, "grad_norm": 10.6015625, "learning_rate": 9.528780557134736e-06, "loss": 1.1356, "step": 13762 }, { "epoch": 1.4644605235156416, "grad_norm": 11.4765625, "learning_rate": 9.528745025582718e-06, "loss": 1.0456, "step": 13763 }, { "epoch": 1.4645669291338583, "grad_norm": 15.9453125, "learning_rate": 9.5287094940307e-06, "loss": 1.0982, "step": 13764 }, { "epoch": 1.4646733347520748, "grad_norm": 13.109375, "learning_rate": 9.528673962478681e-06, "loss": 1.04, "step": 13765 }, { "epoch": 1.4647797403702916, "grad_norm": 9.546875, "learning_rate": 9.528638430926663e-06, "loss": 1.2997, "step": 13766 }, { "epoch": 1.4648861459885083, "grad_norm": 14.09375, "learning_rate": 9.528602899374646e-06, "loss": 1.2227, "step": 13767 }, { "epoch": 1.4649925516067248, "grad_norm": 13.0546875, "learning_rate": 9.528567367822628e-06, "loss": 0.9335, "step": 13768 }, { "epoch": 1.4650989572249415, "grad_norm": 13.125, "learning_rate": 9.52853183627061e-06, "loss": 1.1287, "step": 13769 }, { "epoch": 1.465205362843158, "grad_norm": 11.7578125, "learning_rate": 9.528496304718591e-06, "loss": 0.9467, "step": 13770 }, { "epoch": 1.4653117684613748, "grad_norm": 13.0859375, "learning_rate": 9.528460773166573e-06, "loss": 1.1207, "step": 13771 }, { "epoch": 1.4654181740795913, "grad_norm": 10.3125, "learning_rate": 9.528425241614554e-06, "loss": 1.2005, "step": 13772 }, { "epoch": 1.465524579697808, "grad_norm": 11.734375, "learning_rate": 9.528389710062536e-06, "loss": 1.1542, "step": 13773 }, { "epoch": 1.4656309853160248, "grad_norm": 13.4453125, "learning_rate": 9.528354178510517e-06, "loss": 1.117, "step": 13774 }, { "epoch": 1.4657373909342413, "grad_norm": 11.0625, "learning_rate": 9.528318646958499e-06, "loss": 0.8889, "step": 13775 }, { "epoch": 1.465843796552458, "grad_norm": 12.203125, "learning_rate": 9.52828311540648e-06, "loss": 1.0624, "step": 13776 }, { "epoch": 1.4659502021706747, "grad_norm": 14.6796875, "learning_rate": 9.528247583854462e-06, "loss": 1.0527, "step": 13777 }, { "epoch": 1.4660566077888912, "grad_norm": 8.3125, "learning_rate": 9.528212052302446e-06, "loss": 0.5371, "step": 13778 }, { "epoch": 1.466163013407108, "grad_norm": 8.3359375, "learning_rate": 9.528176520750427e-06, "loss": 0.8886, "step": 13779 }, { "epoch": 1.4662694190253245, "grad_norm": 12.1484375, "learning_rate": 9.528140989198409e-06, "loss": 1.1357, "step": 13780 }, { "epoch": 1.4663758246435412, "grad_norm": 12.71875, "learning_rate": 9.52810545764639e-06, "loss": 1.0472, "step": 13781 }, { "epoch": 1.4664822302617577, "grad_norm": 11.5625, "learning_rate": 9.528069926094372e-06, "loss": 1.1383, "step": 13782 }, { "epoch": 1.4665886358799745, "grad_norm": 12.40625, "learning_rate": 9.528034394542355e-06, "loss": 1.1005, "step": 13783 }, { "epoch": 1.4666950414981912, "grad_norm": 18.015625, "learning_rate": 9.527998862990337e-06, "loss": 1.0033, "step": 13784 }, { "epoch": 1.4668014471164077, "grad_norm": 13.0703125, "learning_rate": 9.527963331438319e-06, "loss": 1.1641, "step": 13785 }, { "epoch": 1.4669078527346244, "grad_norm": 17.078125, "learning_rate": 9.5279277998863e-06, "loss": 1.2578, "step": 13786 }, { "epoch": 1.4670142583528412, "grad_norm": 11.3046875, "learning_rate": 9.527892268334282e-06, "loss": 1.1968, "step": 13787 }, { "epoch": 1.4671206639710577, "grad_norm": 19.296875, "learning_rate": 9.527856736782263e-06, "loss": 0.8804, "step": 13788 }, { "epoch": 1.4672270695892742, "grad_norm": 10.8125, "learning_rate": 9.527821205230245e-06, "loss": 1.347, "step": 13789 }, { "epoch": 1.467333475207491, "grad_norm": 15.46875, "learning_rate": 9.527785673678227e-06, "loss": 1.0699, "step": 13790 }, { "epoch": 1.4674398808257076, "grad_norm": 15.53125, "learning_rate": 9.527750142126208e-06, "loss": 0.8831, "step": 13791 }, { "epoch": 1.4675462864439242, "grad_norm": 14.25, "learning_rate": 9.52771461057419e-06, "loss": 1.2047, "step": 13792 }, { "epoch": 1.4676526920621409, "grad_norm": 12.3203125, "learning_rate": 9.527679079022173e-06, "loss": 0.8967, "step": 13793 }, { "epoch": 1.4677590976803576, "grad_norm": 14.625, "learning_rate": 9.527643547470155e-06, "loss": 0.9208, "step": 13794 }, { "epoch": 1.4678655032985741, "grad_norm": 17.5625, "learning_rate": 9.527608015918136e-06, "loss": 1.0652, "step": 13795 }, { "epoch": 1.4679719089167909, "grad_norm": 11.625, "learning_rate": 9.527572484366118e-06, "loss": 0.9043, "step": 13796 }, { "epoch": 1.4680783145350074, "grad_norm": 12.515625, "learning_rate": 9.5275369528141e-06, "loss": 1.3851, "step": 13797 }, { "epoch": 1.468184720153224, "grad_norm": 12.015625, "learning_rate": 9.527501421262083e-06, "loss": 1.1868, "step": 13798 }, { "epoch": 1.4682911257714406, "grad_norm": 15.421875, "learning_rate": 9.527465889710064e-06, "loss": 1.0298, "step": 13799 }, { "epoch": 1.4683975313896573, "grad_norm": 15.03125, "learning_rate": 9.527430358158046e-06, "loss": 1.2294, "step": 13800 }, { "epoch": 1.468503937007874, "grad_norm": 12.890625, "learning_rate": 9.527394826606026e-06, "loss": 1.1448, "step": 13801 }, { "epoch": 1.4686103426260906, "grad_norm": 13.515625, "learning_rate": 9.527359295054008e-06, "loss": 0.8753, "step": 13802 }, { "epoch": 1.4687167482443073, "grad_norm": 14.6875, "learning_rate": 9.52732376350199e-06, "loss": 0.9252, "step": 13803 }, { "epoch": 1.468823153862524, "grad_norm": 20.125, "learning_rate": 9.527288231949972e-06, "loss": 0.9419, "step": 13804 }, { "epoch": 1.4689295594807406, "grad_norm": 12.65625, "learning_rate": 9.527252700397954e-06, "loss": 0.9328, "step": 13805 }, { "epoch": 1.4690359650989573, "grad_norm": 11.265625, "learning_rate": 9.527217168845936e-06, "loss": 0.9763, "step": 13806 }, { "epoch": 1.4691423707171738, "grad_norm": 13.328125, "learning_rate": 9.527181637293917e-06, "loss": 1.2362, "step": 13807 }, { "epoch": 1.4692487763353905, "grad_norm": 13.0234375, "learning_rate": 9.5271461057419e-06, "loss": 0.6352, "step": 13808 }, { "epoch": 1.469355181953607, "grad_norm": 15.046875, "learning_rate": 9.527110574189882e-06, "loss": 1.0593, "step": 13809 }, { "epoch": 1.4694615875718238, "grad_norm": 11.59375, "learning_rate": 9.527075042637864e-06, "loss": 0.9633, "step": 13810 }, { "epoch": 1.4695679931900405, "grad_norm": 11.7265625, "learning_rate": 9.527039511085845e-06, "loss": 0.9675, "step": 13811 }, { "epoch": 1.469674398808257, "grad_norm": 14.5078125, "learning_rate": 9.527003979533827e-06, "loss": 0.9643, "step": 13812 }, { "epoch": 1.4697808044264737, "grad_norm": 11.546875, "learning_rate": 9.526968447981809e-06, "loss": 1.0903, "step": 13813 }, { "epoch": 1.4698872100446905, "grad_norm": 13.03125, "learning_rate": 9.52693291642979e-06, "loss": 1.2619, "step": 13814 }, { "epoch": 1.469993615662907, "grad_norm": 10.65625, "learning_rate": 9.526897384877772e-06, "loss": 1.0959, "step": 13815 }, { "epoch": 1.4701000212811237, "grad_norm": 17.484375, "learning_rate": 9.526861853325753e-06, "loss": 1.0421, "step": 13816 }, { "epoch": 1.4702064268993402, "grad_norm": 13.1953125, "learning_rate": 9.526826321773735e-06, "loss": 0.7547, "step": 13817 }, { "epoch": 1.470312832517557, "grad_norm": 13.5703125, "learning_rate": 9.526790790221717e-06, "loss": 1.0557, "step": 13818 }, { "epoch": 1.4704192381357735, "grad_norm": 12.734375, "learning_rate": 9.5267552586697e-06, "loss": 1.113, "step": 13819 }, { "epoch": 1.4705256437539902, "grad_norm": 9.1171875, "learning_rate": 9.526719727117682e-06, "loss": 1.0335, "step": 13820 }, { "epoch": 1.470632049372207, "grad_norm": 11.328125, "learning_rate": 9.526684195565663e-06, "loss": 1.0506, "step": 13821 }, { "epoch": 1.4707384549904234, "grad_norm": 20.203125, "learning_rate": 9.526648664013645e-06, "loss": 1.1939, "step": 13822 }, { "epoch": 1.4708448606086402, "grad_norm": 16.390625, "learning_rate": 9.526613132461626e-06, "loss": 1.1989, "step": 13823 }, { "epoch": 1.470951266226857, "grad_norm": 15.2109375, "learning_rate": 9.52657760090961e-06, "loss": 1.0696, "step": 13824 }, { "epoch": 1.4710576718450734, "grad_norm": 16.875, "learning_rate": 9.526542069357591e-06, "loss": 1.2605, "step": 13825 }, { "epoch": 1.4711640774632901, "grad_norm": 13.1875, "learning_rate": 9.526506537805573e-06, "loss": 1.2131, "step": 13826 }, { "epoch": 1.4712704830815067, "grad_norm": 12.28125, "learning_rate": 9.526471006253554e-06, "loss": 1.1324, "step": 13827 }, { "epoch": 1.4713768886997234, "grad_norm": 10.9609375, "learning_rate": 9.526435474701536e-06, "loss": 1.1196, "step": 13828 }, { "epoch": 1.47148329431794, "grad_norm": 8.5078125, "learning_rate": 9.526399943149518e-06, "loss": 0.9008, "step": 13829 }, { "epoch": 1.4715896999361566, "grad_norm": 13.6953125, "learning_rate": 9.5263644115975e-06, "loss": 0.9743, "step": 13830 }, { "epoch": 1.4716961055543734, "grad_norm": 20.40625, "learning_rate": 9.526328880045481e-06, "loss": 1.1497, "step": 13831 }, { "epoch": 1.4718025111725899, "grad_norm": 10.9921875, "learning_rate": 9.526293348493463e-06, "loss": 0.9681, "step": 13832 }, { "epoch": 1.4719089167908066, "grad_norm": 8.4453125, "learning_rate": 9.526257816941444e-06, "loss": 0.9687, "step": 13833 }, { "epoch": 1.4720153224090233, "grad_norm": 10.3515625, "learning_rate": 9.526222285389427e-06, "loss": 1.2072, "step": 13834 }, { "epoch": 1.4721217280272398, "grad_norm": 10.984375, "learning_rate": 9.526186753837409e-06, "loss": 1.1159, "step": 13835 }, { "epoch": 1.4722281336454563, "grad_norm": 13.234375, "learning_rate": 9.52615122228539e-06, "loss": 1.2583, "step": 13836 }, { "epoch": 1.472334539263673, "grad_norm": 15.984375, "learning_rate": 9.526115690733372e-06, "loss": 1.0027, "step": 13837 }, { "epoch": 1.4724409448818898, "grad_norm": 14.3671875, "learning_rate": 9.526080159181354e-06, "loss": 0.9435, "step": 13838 }, { "epoch": 1.4725473505001063, "grad_norm": 13.8828125, "learning_rate": 9.526044627629335e-06, "loss": 1.0532, "step": 13839 }, { "epoch": 1.472653756118323, "grad_norm": 9.828125, "learning_rate": 9.526009096077317e-06, "loss": 1.1527, "step": 13840 }, { "epoch": 1.4727601617365398, "grad_norm": 12.4375, "learning_rate": 9.525973564525299e-06, "loss": 0.9289, "step": 13841 }, { "epoch": 1.4728665673547563, "grad_norm": 11.1796875, "learning_rate": 9.52593803297328e-06, "loss": 1.0404, "step": 13842 }, { "epoch": 1.472972972972973, "grad_norm": 10.921875, "learning_rate": 9.525902501421262e-06, "loss": 1.1803, "step": 13843 }, { "epoch": 1.4730793785911895, "grad_norm": 11.6875, "learning_rate": 9.525866969869245e-06, "loss": 0.9866, "step": 13844 }, { "epoch": 1.4731857842094063, "grad_norm": 9.7109375, "learning_rate": 9.525831438317227e-06, "loss": 0.9629, "step": 13845 }, { "epoch": 1.4732921898276228, "grad_norm": 13.1796875, "learning_rate": 9.525795906765208e-06, "loss": 1.1459, "step": 13846 }, { "epoch": 1.4733985954458395, "grad_norm": 12.65625, "learning_rate": 9.52576037521319e-06, "loss": 0.9516, "step": 13847 }, { "epoch": 1.4735050010640562, "grad_norm": 13.7421875, "learning_rate": 9.525724843661172e-06, "loss": 1.0603, "step": 13848 }, { "epoch": 1.4736114066822728, "grad_norm": 10.4140625, "learning_rate": 9.525689312109153e-06, "loss": 1.0322, "step": 13849 }, { "epoch": 1.4737178123004895, "grad_norm": 8.8671875, "learning_rate": 9.525653780557137e-06, "loss": 0.8169, "step": 13850 }, { "epoch": 1.4738242179187062, "grad_norm": 16.78125, "learning_rate": 9.525618249005118e-06, "loss": 1.1659, "step": 13851 }, { "epoch": 1.4739306235369227, "grad_norm": 9.3046875, "learning_rate": 9.5255827174531e-06, "loss": 1.1922, "step": 13852 }, { "epoch": 1.4740370291551395, "grad_norm": 12.9140625, "learning_rate": 9.525547185901081e-06, "loss": 1.0111, "step": 13853 }, { "epoch": 1.474143434773356, "grad_norm": 11.296875, "learning_rate": 9.525511654349063e-06, "loss": 1.0516, "step": 13854 }, { "epoch": 1.4742498403915727, "grad_norm": 14.171875, "learning_rate": 9.525476122797045e-06, "loss": 1.3332, "step": 13855 }, { "epoch": 1.4743562460097892, "grad_norm": 13.375, "learning_rate": 9.525440591245026e-06, "loss": 1.4588, "step": 13856 }, { "epoch": 1.474462651628006, "grad_norm": 10.6796875, "learning_rate": 9.525405059693008e-06, "loss": 0.9298, "step": 13857 }, { "epoch": 1.4745690572462227, "grad_norm": 9.234375, "learning_rate": 9.52536952814099e-06, "loss": 0.9253, "step": 13858 }, { "epoch": 1.4746754628644392, "grad_norm": 9.84375, "learning_rate": 9.525333996588971e-06, "loss": 1.2202, "step": 13859 }, { "epoch": 1.474781868482656, "grad_norm": 11.8515625, "learning_rate": 9.525298465036954e-06, "loss": 1.0484, "step": 13860 }, { "epoch": 1.4748882741008726, "grad_norm": 16.59375, "learning_rate": 9.525262933484936e-06, "loss": 1.2188, "step": 13861 }, { "epoch": 1.4749946797190892, "grad_norm": 13.375, "learning_rate": 9.525227401932917e-06, "loss": 1.011, "step": 13862 }, { "epoch": 1.4751010853373059, "grad_norm": 15.015625, "learning_rate": 9.525191870380899e-06, "loss": 1.3058, "step": 13863 }, { "epoch": 1.4752074909555224, "grad_norm": 12.234375, "learning_rate": 9.52515633882888e-06, "loss": 1.1926, "step": 13864 }, { "epoch": 1.4753138965737391, "grad_norm": 11.09375, "learning_rate": 9.525120807276862e-06, "loss": 1.0299, "step": 13865 }, { "epoch": 1.4754203021919556, "grad_norm": 13.703125, "learning_rate": 9.525085275724844e-06, "loss": 1.1312, "step": 13866 }, { "epoch": 1.4755267078101724, "grad_norm": 17.3125, "learning_rate": 9.525049744172826e-06, "loss": 1.1828, "step": 13867 }, { "epoch": 1.475633113428389, "grad_norm": 15.3984375, "learning_rate": 9.525014212620807e-06, "loss": 0.9322, "step": 13868 }, { "epoch": 1.4757395190466056, "grad_norm": 13.5703125, "learning_rate": 9.524978681068789e-06, "loss": 1.0164, "step": 13869 }, { "epoch": 1.4758459246648223, "grad_norm": 16.1875, "learning_rate": 9.524943149516772e-06, "loss": 1.0729, "step": 13870 }, { "epoch": 1.475952330283039, "grad_norm": 10.8671875, "learning_rate": 9.524907617964754e-06, "loss": 1.0945, "step": 13871 }, { "epoch": 1.4760587359012556, "grad_norm": 11.921875, "learning_rate": 9.524872086412735e-06, "loss": 1.1782, "step": 13872 }, { "epoch": 1.476165141519472, "grad_norm": 12.515625, "learning_rate": 9.524836554860717e-06, "loss": 1.0089, "step": 13873 }, { "epoch": 1.4762715471376888, "grad_norm": 12.0546875, "learning_rate": 9.524801023308698e-06, "loss": 0.7635, "step": 13874 }, { "epoch": 1.4763779527559056, "grad_norm": 13.4765625, "learning_rate": 9.524765491756682e-06, "loss": 1.0314, "step": 13875 }, { "epoch": 1.476484358374122, "grad_norm": 12.453125, "learning_rate": 9.524729960204663e-06, "loss": 0.9813, "step": 13876 }, { "epoch": 1.4765907639923388, "grad_norm": 13.078125, "learning_rate": 9.524694428652645e-06, "loss": 1.1597, "step": 13877 }, { "epoch": 1.4766971696105555, "grad_norm": 16.09375, "learning_rate": 9.524658897100627e-06, "loss": 0.9377, "step": 13878 }, { "epoch": 1.476803575228772, "grad_norm": 19.84375, "learning_rate": 9.524623365548608e-06, "loss": 1.1243, "step": 13879 }, { "epoch": 1.4769099808469888, "grad_norm": 10.6484375, "learning_rate": 9.52458783399659e-06, "loss": 1.0972, "step": 13880 }, { "epoch": 1.4770163864652055, "grad_norm": 11.515625, "learning_rate": 9.524552302444571e-06, "loss": 1.0093, "step": 13881 }, { "epoch": 1.477122792083422, "grad_norm": 16.375, "learning_rate": 9.524516770892553e-06, "loss": 1.0972, "step": 13882 }, { "epoch": 1.4772291977016385, "grad_norm": 11.7734375, "learning_rate": 9.524481239340535e-06, "loss": 1.1838, "step": 13883 }, { "epoch": 1.4773356033198553, "grad_norm": 11.8203125, "learning_rate": 9.524445707788516e-06, "loss": 0.8059, "step": 13884 }, { "epoch": 1.477442008938072, "grad_norm": 12.296875, "learning_rate": 9.5244101762365e-06, "loss": 1.0416, "step": 13885 }, { "epoch": 1.4775484145562885, "grad_norm": 12.8203125, "learning_rate": 9.524374644684481e-06, "loss": 1.0668, "step": 13886 }, { "epoch": 1.4776548201745052, "grad_norm": 8.4609375, "learning_rate": 9.524339113132463e-06, "loss": 0.8712, "step": 13887 }, { "epoch": 1.477761225792722, "grad_norm": 11.5234375, "learning_rate": 9.524303581580444e-06, "loss": 1.3574, "step": 13888 }, { "epoch": 1.4778676314109385, "grad_norm": 13.734375, "learning_rate": 9.524268050028426e-06, "loss": 0.9807, "step": 13889 }, { "epoch": 1.4779740370291552, "grad_norm": 17.46875, "learning_rate": 9.524232518476408e-06, "loss": 1.3416, "step": 13890 }, { "epoch": 1.4780804426473717, "grad_norm": 21.78125, "learning_rate": 9.52419698692439e-06, "loss": 1.1906, "step": 13891 }, { "epoch": 1.4781868482655884, "grad_norm": 14.1640625, "learning_rate": 9.52416145537237e-06, "loss": 0.9148, "step": 13892 }, { "epoch": 1.478293253883805, "grad_norm": 12.2421875, "learning_rate": 9.524125923820352e-06, "loss": 1.0303, "step": 13893 }, { "epoch": 1.4783996595020217, "grad_norm": 10.2109375, "learning_rate": 9.524090392268334e-06, "loss": 0.905, "step": 13894 }, { "epoch": 1.4785060651202384, "grad_norm": 13.09375, "learning_rate": 9.524054860716316e-06, "loss": 0.9918, "step": 13895 }, { "epoch": 1.478612470738455, "grad_norm": 15.7578125, "learning_rate": 9.524019329164299e-06, "loss": 1.1463, "step": 13896 }, { "epoch": 1.4787188763566717, "grad_norm": 9.609375, "learning_rate": 9.52398379761228e-06, "loss": 0.9353, "step": 13897 }, { "epoch": 1.4788252819748884, "grad_norm": 12.234375, "learning_rate": 9.523948266060262e-06, "loss": 1.0316, "step": 13898 }, { "epoch": 1.478931687593105, "grad_norm": 8.6796875, "learning_rate": 9.523912734508244e-06, "loss": 0.9276, "step": 13899 }, { "epoch": 1.4790380932113216, "grad_norm": 11.8125, "learning_rate": 9.523877202956225e-06, "loss": 1.2056, "step": 13900 }, { "epoch": 1.4791444988295381, "grad_norm": 16.671875, "learning_rate": 9.523841671404209e-06, "loss": 0.9333, "step": 13901 }, { "epoch": 1.4792509044477549, "grad_norm": 10.5234375, "learning_rate": 9.52380613985219e-06, "loss": 1.1596, "step": 13902 }, { "epoch": 1.4793573100659714, "grad_norm": 13.7109375, "learning_rate": 9.523770608300172e-06, "loss": 1.4702, "step": 13903 }, { "epoch": 1.479463715684188, "grad_norm": 11.6875, "learning_rate": 9.523735076748153e-06, "loss": 1.2499, "step": 13904 }, { "epoch": 1.4795701213024048, "grad_norm": 13.6640625, "learning_rate": 9.523699545196135e-06, "loss": 1.1753, "step": 13905 }, { "epoch": 1.4796765269206213, "grad_norm": 8.9140625, "learning_rate": 9.523664013644117e-06, "loss": 1.0722, "step": 13906 }, { "epoch": 1.479782932538838, "grad_norm": 11.1015625, "learning_rate": 9.523628482092098e-06, "loss": 0.9905, "step": 13907 }, { "epoch": 1.4798893381570548, "grad_norm": 13.4921875, "learning_rate": 9.52359295054008e-06, "loss": 1.379, "step": 13908 }, { "epoch": 1.4799957437752713, "grad_norm": 10.5, "learning_rate": 9.523557418988061e-06, "loss": 1.0609, "step": 13909 }, { "epoch": 1.480102149393488, "grad_norm": 11.1328125, "learning_rate": 9.523521887436043e-06, "loss": 0.9095, "step": 13910 }, { "epoch": 1.4802085550117046, "grad_norm": 13.84375, "learning_rate": 9.523486355884026e-06, "loss": 1.1434, "step": 13911 }, { "epoch": 1.4803149606299213, "grad_norm": 11.7421875, "learning_rate": 9.523450824332008e-06, "loss": 0.9671, "step": 13912 }, { "epoch": 1.4804213662481378, "grad_norm": 10.5546875, "learning_rate": 9.52341529277999e-06, "loss": 1.1724, "step": 13913 }, { "epoch": 1.4805277718663545, "grad_norm": 12.4375, "learning_rate": 9.523379761227971e-06, "loss": 1.2264, "step": 13914 }, { "epoch": 1.4806341774845713, "grad_norm": 15.25, "learning_rate": 9.523344229675953e-06, "loss": 0.9614, "step": 13915 }, { "epoch": 1.4807405831027878, "grad_norm": 15.0625, "learning_rate": 9.523308698123936e-06, "loss": 1.0497, "step": 13916 }, { "epoch": 1.4808469887210045, "grad_norm": 11.625, "learning_rate": 9.523273166571918e-06, "loss": 1.0047, "step": 13917 }, { "epoch": 1.4809533943392212, "grad_norm": 12.8515625, "learning_rate": 9.5232376350199e-06, "loss": 1.363, "step": 13918 }, { "epoch": 1.4810597999574378, "grad_norm": 12.09375, "learning_rate": 9.523202103467881e-06, "loss": 1.116, "step": 13919 }, { "epoch": 1.4811662055756543, "grad_norm": 9.6328125, "learning_rate": 9.523166571915863e-06, "loss": 1.0603, "step": 13920 }, { "epoch": 1.481272611193871, "grad_norm": 11.9453125, "learning_rate": 9.523131040363844e-06, "loss": 1.1232, "step": 13921 }, { "epoch": 1.4813790168120877, "grad_norm": 11.6328125, "learning_rate": 9.523095508811826e-06, "loss": 1.2321, "step": 13922 }, { "epoch": 1.4814854224303042, "grad_norm": 9.921875, "learning_rate": 9.523059977259807e-06, "loss": 0.9921, "step": 13923 }, { "epoch": 1.481591828048521, "grad_norm": 13.59375, "learning_rate": 9.523024445707789e-06, "loss": 1.056, "step": 13924 }, { "epoch": 1.4816982336667377, "grad_norm": 13.8203125, "learning_rate": 9.52298891415577e-06, "loss": 1.0075, "step": 13925 }, { "epoch": 1.4818046392849542, "grad_norm": 13.0390625, "learning_rate": 9.522953382603754e-06, "loss": 1.3337, "step": 13926 }, { "epoch": 1.481911044903171, "grad_norm": 14.6015625, "learning_rate": 9.522917851051735e-06, "loss": 1.4245, "step": 13927 }, { "epoch": 1.4820174505213874, "grad_norm": 16.5625, "learning_rate": 9.522882319499717e-06, "loss": 1.1168, "step": 13928 }, { "epoch": 1.4821238561396042, "grad_norm": 16.15625, "learning_rate": 9.522846787947699e-06, "loss": 1.084, "step": 13929 }, { "epoch": 1.4822302617578207, "grad_norm": 12.0078125, "learning_rate": 9.52281125639568e-06, "loss": 0.9076, "step": 13930 }, { "epoch": 1.4823366673760374, "grad_norm": 11.546875, "learning_rate": 9.522775724843662e-06, "loss": 1.1384, "step": 13931 }, { "epoch": 1.4824430729942542, "grad_norm": 10.5859375, "learning_rate": 9.522740193291644e-06, "loss": 0.9705, "step": 13932 }, { "epoch": 1.4825494786124707, "grad_norm": 12.609375, "learning_rate": 9.522704661739625e-06, "loss": 0.9851, "step": 13933 }, { "epoch": 1.4826558842306874, "grad_norm": 11.9140625, "learning_rate": 9.522669130187607e-06, "loss": 1.0276, "step": 13934 }, { "epoch": 1.4827622898489041, "grad_norm": 14.25, "learning_rate": 9.522633598635588e-06, "loss": 1.4306, "step": 13935 }, { "epoch": 1.4828686954671206, "grad_norm": 13.015625, "learning_rate": 9.52259806708357e-06, "loss": 1.3724, "step": 13936 }, { "epoch": 1.4829751010853374, "grad_norm": 12.0625, "learning_rate": 9.522562535531553e-06, "loss": 0.9324, "step": 13937 }, { "epoch": 1.4830815067035539, "grad_norm": 13.1171875, "learning_rate": 9.522527003979535e-06, "loss": 1.0928, "step": 13938 }, { "epoch": 1.4831879123217706, "grad_norm": 14.46875, "learning_rate": 9.522491472427516e-06, "loss": 1.3639, "step": 13939 }, { "epoch": 1.4832943179399871, "grad_norm": 18.609375, "learning_rate": 9.522455940875498e-06, "loss": 1.0154, "step": 13940 }, { "epoch": 1.4834007235582038, "grad_norm": 16.484375, "learning_rate": 9.52242040932348e-06, "loss": 1.3686, "step": 13941 }, { "epoch": 1.4835071291764206, "grad_norm": 10.046875, "learning_rate": 9.522384877771463e-06, "loss": 1.1296, "step": 13942 }, { "epoch": 1.483613534794637, "grad_norm": 12.8515625, "learning_rate": 9.522349346219445e-06, "loss": 0.9241, "step": 13943 }, { "epoch": 1.4837199404128538, "grad_norm": 15.6328125, "learning_rate": 9.522313814667426e-06, "loss": 0.993, "step": 13944 }, { "epoch": 1.4838263460310706, "grad_norm": 11.5, "learning_rate": 9.522278283115408e-06, "loss": 1.1058, "step": 13945 }, { "epoch": 1.483932751649287, "grad_norm": 13.5859375, "learning_rate": 9.52224275156339e-06, "loss": 0.9885, "step": 13946 }, { "epoch": 1.4840391572675038, "grad_norm": 13.328125, "learning_rate": 9.522207220011371e-06, "loss": 0.9557, "step": 13947 }, { "epoch": 1.4841455628857203, "grad_norm": 15.34375, "learning_rate": 9.522171688459353e-06, "loss": 1.3099, "step": 13948 }, { "epoch": 1.484251968503937, "grad_norm": 10.625, "learning_rate": 9.522136156907334e-06, "loss": 1.2972, "step": 13949 }, { "epoch": 1.4843583741221535, "grad_norm": 11.4296875, "learning_rate": 9.522100625355316e-06, "loss": 1.1283, "step": 13950 }, { "epoch": 1.4844647797403703, "grad_norm": 12.15625, "learning_rate": 9.522065093803297e-06, "loss": 1.0229, "step": 13951 }, { "epoch": 1.484571185358587, "grad_norm": 11.6484375, "learning_rate": 9.52202956225128e-06, "loss": 0.9536, "step": 13952 }, { "epoch": 1.4846775909768035, "grad_norm": 11.21875, "learning_rate": 9.521994030699262e-06, "loss": 1.0372, "step": 13953 }, { "epoch": 1.4847839965950203, "grad_norm": 13.53125, "learning_rate": 9.521958499147244e-06, "loss": 1.2146, "step": 13954 }, { "epoch": 1.484890402213237, "grad_norm": 11.6640625, "learning_rate": 9.521922967595226e-06, "loss": 0.9788, "step": 13955 }, { "epoch": 1.4849968078314535, "grad_norm": 10.3359375, "learning_rate": 9.521887436043207e-06, "loss": 1.0787, "step": 13956 }, { "epoch": 1.4851032134496702, "grad_norm": 10.5078125, "learning_rate": 9.521851904491189e-06, "loss": 1.3319, "step": 13957 }, { "epoch": 1.4852096190678867, "grad_norm": 10.21875, "learning_rate": 9.52181637293917e-06, "loss": 1.0111, "step": 13958 }, { "epoch": 1.4853160246861035, "grad_norm": 12.46875, "learning_rate": 9.521780841387152e-06, "loss": 1.3659, "step": 13959 }, { "epoch": 1.48542243030432, "grad_norm": 16.421875, "learning_rate": 9.521745309835134e-06, "loss": 0.9527, "step": 13960 }, { "epoch": 1.4855288359225367, "grad_norm": 13.6484375, "learning_rate": 9.521709778283115e-06, "loss": 1.196, "step": 13961 }, { "epoch": 1.4856352415407534, "grad_norm": 12.9375, "learning_rate": 9.521674246731098e-06, "loss": 1.2447, "step": 13962 }, { "epoch": 1.48574164715897, "grad_norm": 11.65625, "learning_rate": 9.52163871517908e-06, "loss": 1.0215, "step": 13963 }, { "epoch": 1.4858480527771867, "grad_norm": 11.75, "learning_rate": 9.521603183627062e-06, "loss": 1.0032, "step": 13964 }, { "epoch": 1.4859544583954034, "grad_norm": 10.875, "learning_rate": 9.521567652075043e-06, "loss": 1.0398, "step": 13965 }, { "epoch": 1.48606086401362, "grad_norm": 13.6171875, "learning_rate": 9.521532120523025e-06, "loss": 1.1172, "step": 13966 }, { "epoch": 1.4861672696318364, "grad_norm": 13.6328125, "learning_rate": 9.521496588971007e-06, "loss": 0.9339, "step": 13967 }, { "epoch": 1.4862736752500532, "grad_norm": 11.1796875, "learning_rate": 9.52146105741899e-06, "loss": 0.7588, "step": 13968 }, { "epoch": 1.48638008086827, "grad_norm": 12.3203125, "learning_rate": 9.521425525866971e-06, "loss": 1.2357, "step": 13969 }, { "epoch": 1.4864864864864864, "grad_norm": 9.21875, "learning_rate": 9.521389994314953e-06, "loss": 1.0217, "step": 13970 }, { "epoch": 1.4865928921047031, "grad_norm": 14.234375, "learning_rate": 9.521354462762935e-06, "loss": 0.8875, "step": 13971 }, { "epoch": 1.4866992977229199, "grad_norm": 13.859375, "learning_rate": 9.521318931210916e-06, "loss": 1.1733, "step": 13972 }, { "epoch": 1.4868057033411364, "grad_norm": 18.578125, "learning_rate": 9.521283399658898e-06, "loss": 1.1995, "step": 13973 }, { "epoch": 1.486912108959353, "grad_norm": 14.3671875, "learning_rate": 9.52124786810688e-06, "loss": 1.321, "step": 13974 }, { "epoch": 1.4870185145775696, "grad_norm": 15.65625, "learning_rate": 9.521212336554861e-06, "loss": 1.0786, "step": 13975 }, { "epoch": 1.4871249201957863, "grad_norm": 11.53125, "learning_rate": 9.521176805002843e-06, "loss": 1.0667, "step": 13976 }, { "epoch": 1.4872313258140029, "grad_norm": 11.109375, "learning_rate": 9.521141273450824e-06, "loss": 1.0263, "step": 13977 }, { "epoch": 1.4873377314322196, "grad_norm": 13.703125, "learning_rate": 9.521105741898808e-06, "loss": 0.9815, "step": 13978 }, { "epoch": 1.4874441370504363, "grad_norm": 14.2109375, "learning_rate": 9.52107021034679e-06, "loss": 1.1802, "step": 13979 }, { "epoch": 1.4875505426686528, "grad_norm": 10.1953125, "learning_rate": 9.52103467879477e-06, "loss": 0.899, "step": 13980 }, { "epoch": 1.4876569482868696, "grad_norm": 13.65625, "learning_rate": 9.520999147242752e-06, "loss": 1.2154, "step": 13981 }, { "epoch": 1.4877633539050863, "grad_norm": 12.765625, "learning_rate": 9.520963615690734e-06, "loss": 1.1858, "step": 13982 }, { "epoch": 1.4878697595233028, "grad_norm": 10.2734375, "learning_rate": 9.520928084138716e-06, "loss": 0.9029, "step": 13983 }, { "epoch": 1.4879761651415195, "grad_norm": 13.921875, "learning_rate": 9.520892552586697e-06, "loss": 0.9534, "step": 13984 }, { "epoch": 1.488082570759736, "grad_norm": 12.2421875, "learning_rate": 9.520857021034679e-06, "loss": 1.1792, "step": 13985 }, { "epoch": 1.4881889763779528, "grad_norm": 14.0625, "learning_rate": 9.52082148948266e-06, "loss": 1.1717, "step": 13986 }, { "epoch": 1.4882953819961693, "grad_norm": 14.0078125, "learning_rate": 9.520785957930642e-06, "loss": 0.9111, "step": 13987 }, { "epoch": 1.488401787614386, "grad_norm": 11.3828125, "learning_rate": 9.520750426378625e-06, "loss": 1.0487, "step": 13988 }, { "epoch": 1.4885081932326027, "grad_norm": 19.359375, "learning_rate": 9.520714894826607e-06, "loss": 1.1014, "step": 13989 }, { "epoch": 1.4886145988508193, "grad_norm": 11.2421875, "learning_rate": 9.520679363274589e-06, "loss": 0.854, "step": 13990 }, { "epoch": 1.488721004469036, "grad_norm": 12.6875, "learning_rate": 9.52064383172257e-06, "loss": 1.1369, "step": 13991 }, { "epoch": 1.4888274100872527, "grad_norm": 11.2109375, "learning_rate": 9.520608300170552e-06, "loss": 1.0055, "step": 13992 }, { "epoch": 1.4889338157054692, "grad_norm": 11.5703125, "learning_rate": 9.520572768618535e-06, "loss": 1.0631, "step": 13993 }, { "epoch": 1.489040221323686, "grad_norm": 15.4453125, "learning_rate": 9.520537237066517e-06, "loss": 1.5122, "step": 13994 }, { "epoch": 1.4891466269419025, "grad_norm": 11.984375, "learning_rate": 9.520501705514498e-06, "loss": 1.1331, "step": 13995 }, { "epoch": 1.4892530325601192, "grad_norm": 14.5234375, "learning_rate": 9.52046617396248e-06, "loss": 1.3318, "step": 13996 }, { "epoch": 1.4893594381783357, "grad_norm": 13.0859375, "learning_rate": 9.520430642410461e-06, "loss": 1.0875, "step": 13997 }, { "epoch": 1.4894658437965524, "grad_norm": 13.1171875, "learning_rate": 9.520395110858443e-06, "loss": 0.816, "step": 13998 }, { "epoch": 1.4895722494147692, "grad_norm": 13.015625, "learning_rate": 9.520359579306425e-06, "loss": 1.2037, "step": 13999 }, { "epoch": 1.4896786550329857, "grad_norm": 16.390625, "learning_rate": 9.520324047754406e-06, "loss": 1.0122, "step": 14000 }, { "epoch": 1.4897850606512024, "grad_norm": 18.890625, "learning_rate": 9.520288516202388e-06, "loss": 0.866, "step": 14001 }, { "epoch": 1.4898914662694192, "grad_norm": 11.6796875, "learning_rate": 9.52025298465037e-06, "loss": 0.9806, "step": 14002 }, { "epoch": 1.4899978718876357, "grad_norm": 12.1640625, "learning_rate": 9.520217453098353e-06, "loss": 1.0282, "step": 14003 }, { "epoch": 1.4901042775058522, "grad_norm": 10.953125, "learning_rate": 9.520181921546334e-06, "loss": 0.9562, "step": 14004 }, { "epoch": 1.490210683124069, "grad_norm": 13.796875, "learning_rate": 9.520146389994316e-06, "loss": 1.1839, "step": 14005 }, { "epoch": 1.4903170887422856, "grad_norm": 13.578125, "learning_rate": 9.520110858442298e-06, "loss": 1.2544, "step": 14006 }, { "epoch": 1.4904234943605021, "grad_norm": 10.5859375, "learning_rate": 9.52007532689028e-06, "loss": 1.1626, "step": 14007 }, { "epoch": 1.4905298999787189, "grad_norm": 12.046875, "learning_rate": 9.520039795338261e-06, "loss": 1.4624, "step": 14008 }, { "epoch": 1.4906363055969356, "grad_norm": 12.0, "learning_rate": 9.520004263786242e-06, "loss": 1.3269, "step": 14009 }, { "epoch": 1.4907427112151521, "grad_norm": 11.4921875, "learning_rate": 9.519968732234224e-06, "loss": 1.1845, "step": 14010 }, { "epoch": 1.4908491168333688, "grad_norm": 11.6875, "learning_rate": 9.519933200682206e-06, "loss": 1.0069, "step": 14011 }, { "epoch": 1.4909555224515856, "grad_norm": 8.1953125, "learning_rate": 9.519897669130187e-06, "loss": 1.0233, "step": 14012 }, { "epoch": 1.491061928069802, "grad_norm": 12.4609375, "learning_rate": 9.519862137578169e-06, "loss": 0.8139, "step": 14013 }, { "epoch": 1.4911683336880186, "grad_norm": 11.3828125, "learning_rate": 9.519826606026152e-06, "loss": 1.1384, "step": 14014 }, { "epoch": 1.4912747393062353, "grad_norm": 11.8203125, "learning_rate": 9.519791074474134e-06, "loss": 1.1565, "step": 14015 }, { "epoch": 1.491381144924452, "grad_norm": 14.546875, "learning_rate": 9.519755542922115e-06, "loss": 1.1192, "step": 14016 }, { "epoch": 1.4914875505426686, "grad_norm": 13.7265625, "learning_rate": 9.519720011370097e-06, "loss": 1.0481, "step": 14017 }, { "epoch": 1.4915939561608853, "grad_norm": 17.28125, "learning_rate": 9.519684479818079e-06, "loss": 0.9277, "step": 14018 }, { "epoch": 1.491700361779102, "grad_norm": 12.4609375, "learning_rate": 9.519648948266062e-06, "loss": 1.0856, "step": 14019 }, { "epoch": 1.4918067673973185, "grad_norm": 14.796875, "learning_rate": 9.519613416714044e-06, "loss": 1.0464, "step": 14020 }, { "epoch": 1.4919131730155353, "grad_norm": 12.2578125, "learning_rate": 9.519577885162025e-06, "loss": 0.9657, "step": 14021 }, { "epoch": 1.4920195786337518, "grad_norm": 13.4609375, "learning_rate": 9.519542353610007e-06, "loss": 1.128, "step": 14022 }, { "epoch": 1.4921259842519685, "grad_norm": 19.21875, "learning_rate": 9.519506822057988e-06, "loss": 0.8823, "step": 14023 }, { "epoch": 1.492232389870185, "grad_norm": 16.03125, "learning_rate": 9.51947129050597e-06, "loss": 0.9678, "step": 14024 }, { "epoch": 1.4923387954884018, "grad_norm": 14.7890625, "learning_rate": 9.519435758953952e-06, "loss": 1.0646, "step": 14025 }, { "epoch": 1.4924452011066185, "grad_norm": 11.6171875, "learning_rate": 9.519400227401933e-06, "loss": 1.1722, "step": 14026 }, { "epoch": 1.492551606724835, "grad_norm": 13.609375, "learning_rate": 9.519364695849915e-06, "loss": 1.3771, "step": 14027 }, { "epoch": 1.4926580123430517, "grad_norm": 14.59375, "learning_rate": 9.519329164297896e-06, "loss": 1.1998, "step": 14028 }, { "epoch": 1.4927644179612685, "grad_norm": 15.5625, "learning_rate": 9.51929363274588e-06, "loss": 1.0681, "step": 14029 }, { "epoch": 1.492870823579485, "grad_norm": 11.53125, "learning_rate": 9.519258101193861e-06, "loss": 1.032, "step": 14030 }, { "epoch": 1.4929772291977017, "grad_norm": 16.328125, "learning_rate": 9.519222569641843e-06, "loss": 1.0074, "step": 14031 }, { "epoch": 1.4930836348159182, "grad_norm": 13.3125, "learning_rate": 9.519187038089825e-06, "loss": 1.153, "step": 14032 }, { "epoch": 1.493190040434135, "grad_norm": 21.96875, "learning_rate": 9.519151506537806e-06, "loss": 0.8591, "step": 14033 }, { "epoch": 1.4932964460523515, "grad_norm": 12.21875, "learning_rate": 9.51911597498579e-06, "loss": 1.0729, "step": 14034 }, { "epoch": 1.4934028516705682, "grad_norm": 15.4140625, "learning_rate": 9.519080443433771e-06, "loss": 1.251, "step": 14035 }, { "epoch": 1.493509257288785, "grad_norm": 11.1015625, "learning_rate": 9.519044911881753e-06, "loss": 1.1983, "step": 14036 }, { "epoch": 1.4936156629070014, "grad_norm": 9.8046875, "learning_rate": 9.519009380329734e-06, "loss": 0.7338, "step": 14037 }, { "epoch": 1.4937220685252182, "grad_norm": 15.3125, "learning_rate": 9.518973848777716e-06, "loss": 0.7348, "step": 14038 }, { "epoch": 1.493828474143435, "grad_norm": 12.84375, "learning_rate": 9.518938317225697e-06, "loss": 1.2835, "step": 14039 }, { "epoch": 1.4939348797616514, "grad_norm": 13.2421875, "learning_rate": 9.518902785673679e-06, "loss": 0.9472, "step": 14040 }, { "epoch": 1.4940412853798681, "grad_norm": 12.3671875, "learning_rate": 9.51886725412166e-06, "loss": 1.0381, "step": 14041 }, { "epoch": 1.4941476909980846, "grad_norm": 14.0859375, "learning_rate": 9.518831722569642e-06, "loss": 1.1194, "step": 14042 }, { "epoch": 1.4942540966163014, "grad_norm": 16.96875, "learning_rate": 9.518796191017624e-06, "loss": 1.0776, "step": 14043 }, { "epoch": 1.4943605022345179, "grad_norm": 11.5859375, "learning_rate": 9.518760659465607e-06, "loss": 1.0567, "step": 14044 }, { "epoch": 1.4944669078527346, "grad_norm": 9.6953125, "learning_rate": 9.518725127913589e-06, "loss": 1.1994, "step": 14045 }, { "epoch": 1.4945733134709513, "grad_norm": 12.046875, "learning_rate": 9.51868959636157e-06, "loss": 1.093, "step": 14046 }, { "epoch": 1.4946797190891679, "grad_norm": 15.84375, "learning_rate": 9.518654064809552e-06, "loss": 1.1828, "step": 14047 }, { "epoch": 1.4947861247073846, "grad_norm": 11.046875, "learning_rate": 9.518618533257534e-06, "loss": 1.0217, "step": 14048 }, { "epoch": 1.4948925303256013, "grad_norm": 11.7109375, "learning_rate": 9.518583001705515e-06, "loss": 1.4952, "step": 14049 }, { "epoch": 1.4949989359438178, "grad_norm": 16.234375, "learning_rate": 9.518547470153497e-06, "loss": 1.2929, "step": 14050 }, { "epoch": 1.4951053415620343, "grad_norm": 15.796875, "learning_rate": 9.518511938601478e-06, "loss": 1.0231, "step": 14051 }, { "epoch": 1.495211747180251, "grad_norm": 8.9921875, "learning_rate": 9.51847640704946e-06, "loss": 0.7894, "step": 14052 }, { "epoch": 1.4953181527984678, "grad_norm": 10.8671875, "learning_rate": 9.518440875497442e-06, "loss": 1.0566, "step": 14053 }, { "epoch": 1.4954245584166843, "grad_norm": 13.0703125, "learning_rate": 9.518405343945423e-06, "loss": 1.1421, "step": 14054 }, { "epoch": 1.495530964034901, "grad_norm": 12.046875, "learning_rate": 9.518369812393407e-06, "loss": 0.8559, "step": 14055 }, { "epoch": 1.4956373696531178, "grad_norm": 22.9375, "learning_rate": 9.518334280841388e-06, "loss": 1.4457, "step": 14056 }, { "epoch": 1.4957437752713343, "grad_norm": 15.0390625, "learning_rate": 9.51829874928937e-06, "loss": 1.2425, "step": 14057 }, { "epoch": 1.495850180889551, "grad_norm": 11.7578125, "learning_rate": 9.518263217737351e-06, "loss": 1.1609, "step": 14058 }, { "epoch": 1.4959565865077675, "grad_norm": 16.59375, "learning_rate": 9.518227686185333e-06, "loss": 1.1967, "step": 14059 }, { "epoch": 1.4960629921259843, "grad_norm": 13.25, "learning_rate": 9.518192154633316e-06, "loss": 1.0838, "step": 14060 }, { "epoch": 1.4961693977442008, "grad_norm": 12.578125, "learning_rate": 9.518156623081298e-06, "loss": 0.9947, "step": 14061 }, { "epoch": 1.4962758033624175, "grad_norm": 11.109375, "learning_rate": 9.51812109152928e-06, "loss": 1.1107, "step": 14062 }, { "epoch": 1.4963822089806342, "grad_norm": 12.453125, "learning_rate": 9.518085559977261e-06, "loss": 1.2858, "step": 14063 }, { "epoch": 1.4964886145988507, "grad_norm": 10.6875, "learning_rate": 9.518050028425243e-06, "loss": 0.8981, "step": 14064 }, { "epoch": 1.4965950202170675, "grad_norm": 17.359375, "learning_rate": 9.518014496873224e-06, "loss": 1.372, "step": 14065 }, { "epoch": 1.4967014258352842, "grad_norm": 10.03125, "learning_rate": 9.517978965321206e-06, "loss": 1.1519, "step": 14066 }, { "epoch": 1.4968078314535007, "grad_norm": 12.234375, "learning_rate": 9.517943433769188e-06, "loss": 1.1285, "step": 14067 }, { "epoch": 1.4969142370717174, "grad_norm": 11.1015625, "learning_rate": 9.517907902217169e-06, "loss": 1.0617, "step": 14068 }, { "epoch": 1.497020642689934, "grad_norm": 17.515625, "learning_rate": 9.51787237066515e-06, "loss": 1.1054, "step": 14069 }, { "epoch": 1.4971270483081507, "grad_norm": 12.46875, "learning_rate": 9.517836839113134e-06, "loss": 1.0454, "step": 14070 }, { "epoch": 1.4972334539263672, "grad_norm": 12.8203125, "learning_rate": 9.517801307561116e-06, "loss": 1.1111, "step": 14071 }, { "epoch": 1.497339859544584, "grad_norm": 16.6875, "learning_rate": 9.517765776009097e-06, "loss": 1.2532, "step": 14072 }, { "epoch": 1.4974462651628007, "grad_norm": 9.640625, "learning_rate": 9.517730244457079e-06, "loss": 0.9178, "step": 14073 }, { "epoch": 1.4975526707810172, "grad_norm": 9.9453125, "learning_rate": 9.51769471290506e-06, "loss": 1.175, "step": 14074 }, { "epoch": 1.497659076399234, "grad_norm": 9.3203125, "learning_rate": 9.517659181353042e-06, "loss": 0.9196, "step": 14075 }, { "epoch": 1.4977654820174506, "grad_norm": 14.296875, "learning_rate": 9.517623649801024e-06, "loss": 0.9907, "step": 14076 }, { "epoch": 1.4978718876356671, "grad_norm": 11.1484375, "learning_rate": 9.517588118249005e-06, "loss": 1.2778, "step": 14077 }, { "epoch": 1.4979782932538839, "grad_norm": 12.6953125, "learning_rate": 9.517552586696987e-06, "loss": 1.0181, "step": 14078 }, { "epoch": 1.4980846988721004, "grad_norm": 19.0, "learning_rate": 9.517517055144968e-06, "loss": 1.2121, "step": 14079 }, { "epoch": 1.4981911044903171, "grad_norm": 11.484375, "learning_rate": 9.517481523592952e-06, "loss": 1.2025, "step": 14080 }, { "epoch": 1.4982975101085336, "grad_norm": 11.34375, "learning_rate": 9.517445992040933e-06, "loss": 0.8974, "step": 14081 }, { "epoch": 1.4984039157267504, "grad_norm": 14.1484375, "learning_rate": 9.517410460488915e-06, "loss": 0.9404, "step": 14082 }, { "epoch": 1.498510321344967, "grad_norm": 18.21875, "learning_rate": 9.517374928936897e-06, "loss": 1.0594, "step": 14083 }, { "epoch": 1.4986167269631836, "grad_norm": 13.5234375, "learning_rate": 9.517339397384878e-06, "loss": 1.2331, "step": 14084 }, { "epoch": 1.4987231325814003, "grad_norm": 13.109375, "learning_rate": 9.51730386583286e-06, "loss": 1.0886, "step": 14085 }, { "epoch": 1.498829538199617, "grad_norm": 14.2890625, "learning_rate": 9.517268334280843e-06, "loss": 0.9099, "step": 14086 }, { "epoch": 1.4989359438178336, "grad_norm": 11.5625, "learning_rate": 9.517232802728825e-06, "loss": 0.9137, "step": 14087 }, { "epoch": 1.4990423494360503, "grad_norm": 12.4765625, "learning_rate": 9.517197271176806e-06, "loss": 1.024, "step": 14088 }, { "epoch": 1.4991487550542668, "grad_norm": 13.1328125, "learning_rate": 9.517161739624788e-06, "loss": 1.3721, "step": 14089 }, { "epoch": 1.4992551606724835, "grad_norm": 10.578125, "learning_rate": 9.51712620807277e-06, "loss": 1.0537, "step": 14090 }, { "epoch": 1.4993615662907, "grad_norm": 16.53125, "learning_rate": 9.517090676520751e-06, "loss": 1.0223, "step": 14091 }, { "epoch": 1.4994679719089168, "grad_norm": 11.09375, "learning_rate": 9.517055144968733e-06, "loss": 0.8546, "step": 14092 }, { "epoch": 1.4995743775271335, "grad_norm": 9.9140625, "learning_rate": 9.517019613416714e-06, "loss": 0.9596, "step": 14093 }, { "epoch": 1.49968078314535, "grad_norm": 13.5546875, "learning_rate": 9.516984081864696e-06, "loss": 1.0602, "step": 14094 }, { "epoch": 1.4997871887635668, "grad_norm": 10.15625, "learning_rate": 9.516948550312678e-06, "loss": 1.0414, "step": 14095 }, { "epoch": 1.4998935943817835, "grad_norm": 11.5234375, "learning_rate": 9.516913018760661e-06, "loss": 0.9414, "step": 14096 }, { "epoch": 1.5, "grad_norm": 12.59375, "learning_rate": 9.516877487208642e-06, "loss": 1.0122, "step": 14097 }, { "epoch": 1.5001064056182165, "grad_norm": 16.25, "learning_rate": 9.516841955656624e-06, "loss": 1.2832, "step": 14098 }, { "epoch": 1.5002128112364332, "grad_norm": 14.3671875, "learning_rate": 9.516806424104606e-06, "loss": 1.1541, "step": 14099 }, { "epoch": 1.50031921685465, "grad_norm": 10.9921875, "learning_rate": 9.516770892552587e-06, "loss": 1.1143, "step": 14100 }, { "epoch": 1.5004256224728665, "grad_norm": 18.484375, "learning_rate": 9.516735361000569e-06, "loss": 1.2521, "step": 14101 }, { "epoch": 1.5005320280910832, "grad_norm": 12.0625, "learning_rate": 9.51669982944855e-06, "loss": 1.0181, "step": 14102 }, { "epoch": 1.5006384337093, "grad_norm": 11.9921875, "learning_rate": 9.516664297896532e-06, "loss": 0.9149, "step": 14103 }, { "epoch": 1.5007448393275165, "grad_norm": 17.78125, "learning_rate": 9.516628766344514e-06, "loss": 0.9892, "step": 14104 }, { "epoch": 1.5008512449457332, "grad_norm": 13.4375, "learning_rate": 9.516593234792495e-06, "loss": 1.0374, "step": 14105 }, { "epoch": 1.50095765056395, "grad_norm": 11.5390625, "learning_rate": 9.516557703240479e-06, "loss": 1.2672, "step": 14106 }, { "epoch": 1.5010640561821664, "grad_norm": 12.8671875, "learning_rate": 9.51652217168846e-06, "loss": 1.367, "step": 14107 }, { "epoch": 1.501170461800383, "grad_norm": 13.0546875, "learning_rate": 9.516486640136442e-06, "loss": 1.0369, "step": 14108 }, { "epoch": 1.5012768674185997, "grad_norm": 13.15625, "learning_rate": 9.516451108584423e-06, "loss": 0.952, "step": 14109 }, { "epoch": 1.5013832730368164, "grad_norm": 13.8671875, "learning_rate": 9.516415577032405e-06, "loss": 0.9415, "step": 14110 }, { "epoch": 1.501489678655033, "grad_norm": 13.1953125, "learning_rate": 9.516380045480388e-06, "loss": 1.1516, "step": 14111 }, { "epoch": 1.5015960842732496, "grad_norm": 12.3046875, "learning_rate": 9.51634451392837e-06, "loss": 1.205, "step": 14112 }, { "epoch": 1.5017024898914664, "grad_norm": 9.8359375, "learning_rate": 9.516308982376352e-06, "loss": 0.7484, "step": 14113 }, { "epoch": 1.5018088955096829, "grad_norm": 10.4453125, "learning_rate": 9.516273450824333e-06, "loss": 0.9285, "step": 14114 }, { "epoch": 1.5019153011278994, "grad_norm": 17.9375, "learning_rate": 9.516237919272315e-06, "loss": 1.0409, "step": 14115 }, { "epoch": 1.5020217067461163, "grad_norm": 14.1875, "learning_rate": 9.516202387720296e-06, "loss": 0.902, "step": 14116 }, { "epoch": 1.5021281123643329, "grad_norm": 14.8671875, "learning_rate": 9.516166856168278e-06, "loss": 1.3498, "step": 14117 }, { "epoch": 1.5022345179825494, "grad_norm": 15.4765625, "learning_rate": 9.51613132461626e-06, "loss": 1.3135, "step": 14118 }, { "epoch": 1.502340923600766, "grad_norm": 13.234375, "learning_rate": 9.516095793064241e-06, "loss": 1.2572, "step": 14119 }, { "epoch": 1.5024473292189828, "grad_norm": 13.125, "learning_rate": 9.516060261512223e-06, "loss": 1.086, "step": 14120 }, { "epoch": 1.5025537348371993, "grad_norm": 12.1484375, "learning_rate": 9.516024729960206e-06, "loss": 1.0462, "step": 14121 }, { "epoch": 1.502660140455416, "grad_norm": 11.3671875, "learning_rate": 9.515989198408188e-06, "loss": 0.8868, "step": 14122 }, { "epoch": 1.5027665460736328, "grad_norm": 14.796875, "learning_rate": 9.51595366685617e-06, "loss": 1.1984, "step": 14123 }, { "epoch": 1.5028729516918493, "grad_norm": 12.109375, "learning_rate": 9.515918135304151e-06, "loss": 1.2447, "step": 14124 }, { "epoch": 1.5029793573100658, "grad_norm": 12.3359375, "learning_rate": 9.515882603752133e-06, "loss": 1.1749, "step": 14125 }, { "epoch": 1.5030857629282828, "grad_norm": 13.078125, "learning_rate": 9.515847072200114e-06, "loss": 0.8999, "step": 14126 }, { "epoch": 1.5031921685464993, "grad_norm": 9.9140625, "learning_rate": 9.515811540648097e-06, "loss": 0.8067, "step": 14127 }, { "epoch": 1.5032985741647158, "grad_norm": 13.453125, "learning_rate": 9.515776009096079e-06, "loss": 1.1526, "step": 14128 }, { "epoch": 1.5034049797829325, "grad_norm": 12.2421875, "learning_rate": 9.515740477544059e-06, "loss": 1.1027, "step": 14129 }, { "epoch": 1.5035113854011493, "grad_norm": 11.796875, "learning_rate": 9.51570494599204e-06, "loss": 1.0204, "step": 14130 }, { "epoch": 1.5036177910193658, "grad_norm": 16.15625, "learning_rate": 9.515669414440022e-06, "loss": 1.2998, "step": 14131 }, { "epoch": 1.5037241966375825, "grad_norm": 69.9375, "learning_rate": 9.515633882888006e-06, "loss": 1.3608, "step": 14132 }, { "epoch": 1.5038306022557992, "grad_norm": 12.5703125, "learning_rate": 9.515598351335987e-06, "loss": 0.9178, "step": 14133 }, { "epoch": 1.5039370078740157, "grad_norm": 11.609375, "learning_rate": 9.515562819783969e-06, "loss": 1.0827, "step": 14134 }, { "epoch": 1.5040434134922323, "grad_norm": 11.6953125, "learning_rate": 9.51552728823195e-06, "loss": 1.1199, "step": 14135 }, { "epoch": 1.504149819110449, "grad_norm": 15.734375, "learning_rate": 9.515491756679932e-06, "loss": 1.0094, "step": 14136 }, { "epoch": 1.5042562247286657, "grad_norm": 11.15625, "learning_rate": 9.515456225127915e-06, "loss": 1.3177, "step": 14137 }, { "epoch": 1.5043626303468822, "grad_norm": 12.296875, "learning_rate": 9.515420693575897e-06, "loss": 1.0279, "step": 14138 }, { "epoch": 1.504469035965099, "grad_norm": 11.84375, "learning_rate": 9.515385162023878e-06, "loss": 1.1335, "step": 14139 }, { "epoch": 1.5045754415833157, "grad_norm": 12.359375, "learning_rate": 9.51534963047186e-06, "loss": 1.2324, "step": 14140 }, { "epoch": 1.5046818472015322, "grad_norm": 9.65625, "learning_rate": 9.515314098919842e-06, "loss": 1.198, "step": 14141 }, { "epoch": 1.504788252819749, "grad_norm": 13.265625, "learning_rate": 9.515278567367823e-06, "loss": 1.1728, "step": 14142 }, { "epoch": 1.5048946584379657, "grad_norm": 49.09375, "learning_rate": 9.515243035815805e-06, "loss": 1.5979, "step": 14143 }, { "epoch": 1.5050010640561822, "grad_norm": 14.5234375, "learning_rate": 9.515207504263786e-06, "loss": 1.1657, "step": 14144 }, { "epoch": 1.5051074696743987, "grad_norm": 11.3515625, "learning_rate": 9.515171972711768e-06, "loss": 1.1674, "step": 14145 }, { "epoch": 1.5052138752926154, "grad_norm": 14.0078125, "learning_rate": 9.51513644115975e-06, "loss": 0.8421, "step": 14146 }, { "epoch": 1.5053202809108321, "grad_norm": 10.2109375, "learning_rate": 9.515100909607733e-06, "loss": 0.8884, "step": 14147 }, { "epoch": 1.5054266865290487, "grad_norm": 13.8203125, "learning_rate": 9.515065378055715e-06, "loss": 1.2251, "step": 14148 }, { "epoch": 1.5055330921472654, "grad_norm": 12.2890625, "learning_rate": 9.515029846503696e-06, "loss": 0.9436, "step": 14149 }, { "epoch": 1.5056394977654821, "grad_norm": 12.0546875, "learning_rate": 9.514994314951678e-06, "loss": 0.9262, "step": 14150 }, { "epoch": 1.5057459033836986, "grad_norm": 9.9375, "learning_rate": 9.51495878339966e-06, "loss": 0.9421, "step": 14151 }, { "epoch": 1.5058523090019151, "grad_norm": 10.1484375, "learning_rate": 9.514923251847643e-06, "loss": 1.0245, "step": 14152 }, { "epoch": 1.505958714620132, "grad_norm": 11.7734375, "learning_rate": 9.514887720295624e-06, "loss": 1.2973, "step": 14153 }, { "epoch": 1.5060651202383486, "grad_norm": 11.765625, "learning_rate": 9.514852188743606e-06, "loss": 1.1573, "step": 14154 }, { "epoch": 1.506171525856565, "grad_norm": 11.1640625, "learning_rate": 9.514816657191588e-06, "loss": 1.1147, "step": 14155 }, { "epoch": 1.5062779314747818, "grad_norm": 14.28125, "learning_rate": 9.514781125639569e-06, "loss": 0.9754, "step": 14156 }, { "epoch": 1.5063843370929986, "grad_norm": 11.7890625, "learning_rate": 9.51474559408755e-06, "loss": 0.8826, "step": 14157 }, { "epoch": 1.506490742711215, "grad_norm": 12.8671875, "learning_rate": 9.514710062535532e-06, "loss": 1.0046, "step": 14158 }, { "epoch": 1.5065971483294318, "grad_norm": 14.21875, "learning_rate": 9.514674530983514e-06, "loss": 0.9486, "step": 14159 }, { "epoch": 1.5067035539476485, "grad_norm": 13.0546875, "learning_rate": 9.514638999431496e-06, "loss": 1.1958, "step": 14160 }, { "epoch": 1.506809959565865, "grad_norm": 11.65625, "learning_rate": 9.514603467879477e-06, "loss": 1.1735, "step": 14161 }, { "epoch": 1.5069163651840816, "grad_norm": 12.203125, "learning_rate": 9.514567936327459e-06, "loss": 1.2432, "step": 14162 }, { "epoch": 1.5070227708022985, "grad_norm": 16.125, "learning_rate": 9.514532404775442e-06, "loss": 0.9289, "step": 14163 }, { "epoch": 1.507129176420515, "grad_norm": 12.453125, "learning_rate": 9.514496873223424e-06, "loss": 1.3864, "step": 14164 }, { "epoch": 1.5072355820387315, "grad_norm": 16.84375, "learning_rate": 9.514461341671405e-06, "loss": 1.2989, "step": 14165 }, { "epoch": 1.5073419876569483, "grad_norm": 11.1484375, "learning_rate": 9.514425810119387e-06, "loss": 1.2842, "step": 14166 }, { "epoch": 1.507448393275165, "grad_norm": 13.7109375, "learning_rate": 9.514390278567369e-06, "loss": 0.9787, "step": 14167 }, { "epoch": 1.5075547988933815, "grad_norm": 13.0078125, "learning_rate": 9.51435474701535e-06, "loss": 1.1929, "step": 14168 }, { "epoch": 1.5076612045115982, "grad_norm": 18.3125, "learning_rate": 9.514319215463332e-06, "loss": 1.1274, "step": 14169 }, { "epoch": 1.507767610129815, "grad_norm": 10.46875, "learning_rate": 9.514283683911313e-06, "loss": 1.0488, "step": 14170 }, { "epoch": 1.5078740157480315, "grad_norm": 13.0546875, "learning_rate": 9.514248152359295e-06, "loss": 1.0833, "step": 14171 }, { "epoch": 1.507980421366248, "grad_norm": 12.6796875, "learning_rate": 9.514212620807277e-06, "loss": 0.9727, "step": 14172 }, { "epoch": 1.508086826984465, "grad_norm": 15.953125, "learning_rate": 9.51417708925526e-06, "loss": 1.031, "step": 14173 }, { "epoch": 1.5081932326026815, "grad_norm": 16.53125, "learning_rate": 9.514141557703241e-06, "loss": 1.3486, "step": 14174 }, { "epoch": 1.508299638220898, "grad_norm": 19.109375, "learning_rate": 9.514106026151223e-06, "loss": 0.9524, "step": 14175 }, { "epoch": 1.5084060438391147, "grad_norm": 13.8203125, "learning_rate": 9.514070494599205e-06, "loss": 1.3825, "step": 14176 }, { "epoch": 1.5085124494573314, "grad_norm": 13.9140625, "learning_rate": 9.514034963047186e-06, "loss": 1.1679, "step": 14177 }, { "epoch": 1.508618855075548, "grad_norm": 9.6796875, "learning_rate": 9.51399943149517e-06, "loss": 0.9001, "step": 14178 }, { "epoch": 1.5087252606937647, "grad_norm": 14.296875, "learning_rate": 9.513963899943151e-06, "loss": 1.1515, "step": 14179 }, { "epoch": 1.5088316663119814, "grad_norm": 9.421875, "learning_rate": 9.513928368391133e-06, "loss": 0.939, "step": 14180 }, { "epoch": 1.508938071930198, "grad_norm": 12.2265625, "learning_rate": 9.513892836839114e-06, "loss": 1.2599, "step": 14181 }, { "epoch": 1.5090444775484144, "grad_norm": 12.1328125, "learning_rate": 9.513857305287096e-06, "loss": 0.9074, "step": 14182 }, { "epoch": 1.5091508831666312, "grad_norm": 14.703125, "learning_rate": 9.513821773735078e-06, "loss": 1.1403, "step": 14183 }, { "epoch": 1.5092572887848479, "grad_norm": 33.5, "learning_rate": 9.51378624218306e-06, "loss": 0.9701, "step": 14184 }, { "epoch": 1.5093636944030644, "grad_norm": 13.9296875, "learning_rate": 9.51375071063104e-06, "loss": 0.9874, "step": 14185 }, { "epoch": 1.5094701000212811, "grad_norm": 14.40625, "learning_rate": 9.513715179079022e-06, "loss": 1.1225, "step": 14186 }, { "epoch": 1.5095765056394979, "grad_norm": 9.875, "learning_rate": 9.513679647527004e-06, "loss": 0.9595, "step": 14187 }, { "epoch": 1.5096829112577144, "grad_norm": 11.0078125, "learning_rate": 9.513644115974987e-06, "loss": 1.085, "step": 14188 }, { "epoch": 1.509789316875931, "grad_norm": 13.171875, "learning_rate": 9.513608584422969e-06, "loss": 1.0553, "step": 14189 }, { "epoch": 1.5098957224941478, "grad_norm": 11.234375, "learning_rate": 9.51357305287095e-06, "loss": 0.7817, "step": 14190 }, { "epoch": 1.5100021281123643, "grad_norm": 20.640625, "learning_rate": 9.513537521318932e-06, "loss": 0.9212, "step": 14191 }, { "epoch": 1.5101085337305808, "grad_norm": 11.078125, "learning_rate": 9.513501989766914e-06, "loss": 0.9402, "step": 14192 }, { "epoch": 1.5102149393487976, "grad_norm": 10.4140625, "learning_rate": 9.513466458214895e-06, "loss": 0.8447, "step": 14193 }, { "epoch": 1.5103213449670143, "grad_norm": 12.0859375, "learning_rate": 9.513430926662877e-06, "loss": 1.1152, "step": 14194 }, { "epoch": 1.5104277505852308, "grad_norm": 11.65625, "learning_rate": 9.513395395110859e-06, "loss": 1.1764, "step": 14195 }, { "epoch": 1.5105341562034476, "grad_norm": 14.0625, "learning_rate": 9.51335986355884e-06, "loss": 1.0093, "step": 14196 }, { "epoch": 1.5106405618216643, "grad_norm": 12.1484375, "learning_rate": 9.513324332006822e-06, "loss": 1.1437, "step": 14197 }, { "epoch": 1.5107469674398808, "grad_norm": 12.8046875, "learning_rate": 9.513288800454805e-06, "loss": 1.1141, "step": 14198 }, { "epoch": 1.5108533730580973, "grad_norm": 16.09375, "learning_rate": 9.513253268902787e-06, "loss": 1.1678, "step": 14199 }, { "epoch": 1.5109597786763143, "grad_norm": 12.8828125, "learning_rate": 9.513217737350768e-06, "loss": 1.2199, "step": 14200 }, { "epoch": 1.5110661842945308, "grad_norm": 13.2421875, "learning_rate": 9.51318220579875e-06, "loss": 1.0905, "step": 14201 }, { "epoch": 1.5111725899127473, "grad_norm": 16.84375, "learning_rate": 9.513146674246732e-06, "loss": 1.0215, "step": 14202 }, { "epoch": 1.511278995530964, "grad_norm": 14.4921875, "learning_rate": 9.513111142694713e-06, "loss": 1.5599, "step": 14203 }, { "epoch": 1.5113854011491807, "grad_norm": 11.46875, "learning_rate": 9.513075611142696e-06, "loss": 1.2381, "step": 14204 }, { "epoch": 1.5114918067673973, "grad_norm": 15.5859375, "learning_rate": 9.513040079590678e-06, "loss": 1.26, "step": 14205 }, { "epoch": 1.511598212385614, "grad_norm": 12.03125, "learning_rate": 9.51300454803866e-06, "loss": 1.0762, "step": 14206 }, { "epoch": 1.5117046180038307, "grad_norm": 15.3203125, "learning_rate": 9.512969016486641e-06, "loss": 1.3066, "step": 14207 }, { "epoch": 1.5118110236220472, "grad_norm": 14.7421875, "learning_rate": 9.512933484934623e-06, "loss": 0.9349, "step": 14208 }, { "epoch": 1.5119174292402637, "grad_norm": 10.6640625, "learning_rate": 9.512897953382604e-06, "loss": 0.9996, "step": 14209 }, { "epoch": 1.5120238348584807, "grad_norm": 12.2890625, "learning_rate": 9.512862421830586e-06, "loss": 1.0158, "step": 14210 }, { "epoch": 1.5121302404766972, "grad_norm": 13.9296875, "learning_rate": 9.512826890278568e-06, "loss": 1.2964, "step": 14211 }, { "epoch": 1.5122366460949137, "grad_norm": 11.140625, "learning_rate": 9.51279135872655e-06, "loss": 1.1423, "step": 14212 }, { "epoch": 1.5123430517131304, "grad_norm": 10.015625, "learning_rate": 9.512755827174531e-06, "loss": 1.1148, "step": 14213 }, { "epoch": 1.5124494573313472, "grad_norm": 11.6328125, "learning_rate": 9.512720295622514e-06, "loss": 1.1612, "step": 14214 }, { "epoch": 1.5125558629495637, "grad_norm": 13.5234375, "learning_rate": 9.512684764070496e-06, "loss": 1.1748, "step": 14215 }, { "epoch": 1.5126622685677804, "grad_norm": 9.21875, "learning_rate": 9.512649232518477e-06, "loss": 0.9469, "step": 14216 }, { "epoch": 1.5127686741859971, "grad_norm": 14.984375, "learning_rate": 9.512613700966459e-06, "loss": 1.0441, "step": 14217 }, { "epoch": 1.5128750798042137, "grad_norm": 11.953125, "learning_rate": 9.51257816941444e-06, "loss": 0.9394, "step": 14218 }, { "epoch": 1.5129814854224302, "grad_norm": 14.5625, "learning_rate": 9.512542637862422e-06, "loss": 1.1927, "step": 14219 }, { "epoch": 1.513087891040647, "grad_norm": 11.1875, "learning_rate": 9.512507106310404e-06, "loss": 1.0461, "step": 14220 }, { "epoch": 1.5131942966588636, "grad_norm": 14.078125, "learning_rate": 9.512471574758385e-06, "loss": 1.0834, "step": 14221 }, { "epoch": 1.5133007022770801, "grad_norm": 14.515625, "learning_rate": 9.512436043206367e-06, "loss": 1.1491, "step": 14222 }, { "epoch": 1.5134071078952969, "grad_norm": 11.6953125, "learning_rate": 9.512400511654349e-06, "loss": 1.0639, "step": 14223 }, { "epoch": 1.5135135135135136, "grad_norm": 18.40625, "learning_rate": 9.512364980102332e-06, "loss": 1.2658, "step": 14224 }, { "epoch": 1.51361991913173, "grad_norm": 14.65625, "learning_rate": 9.512329448550314e-06, "loss": 1.2648, "step": 14225 }, { "epoch": 1.5137263247499468, "grad_norm": 10.1171875, "learning_rate": 9.512293916998295e-06, "loss": 1.0607, "step": 14226 }, { "epoch": 1.5138327303681636, "grad_norm": 12.1640625, "learning_rate": 9.512258385446277e-06, "loss": 1.1863, "step": 14227 }, { "epoch": 1.51393913598638, "grad_norm": 10.7734375, "learning_rate": 9.512222853894258e-06, "loss": 1.0657, "step": 14228 }, { "epoch": 1.5140455416045966, "grad_norm": 10.6875, "learning_rate": 9.512187322342242e-06, "loss": 0.9884, "step": 14229 }, { "epoch": 1.5141519472228133, "grad_norm": 11.5625, "learning_rate": 9.512151790790223e-06, "loss": 1.2653, "step": 14230 }, { "epoch": 1.51425835284103, "grad_norm": 12.953125, "learning_rate": 9.512116259238205e-06, "loss": 0.8345, "step": 14231 }, { "epoch": 1.5143647584592466, "grad_norm": 15.71875, "learning_rate": 9.512080727686186e-06, "loss": 0.8158, "step": 14232 }, { "epoch": 1.5144711640774633, "grad_norm": 13.0703125, "learning_rate": 9.512045196134168e-06, "loss": 1.0748, "step": 14233 }, { "epoch": 1.51457756969568, "grad_norm": 11.28125, "learning_rate": 9.51200966458215e-06, "loss": 1.2328, "step": 14234 }, { "epoch": 1.5146839753138965, "grad_norm": 15.265625, "learning_rate": 9.511974133030131e-06, "loss": 0.9598, "step": 14235 }, { "epoch": 1.5147903809321133, "grad_norm": 21.265625, "learning_rate": 9.511938601478113e-06, "loss": 1.3402, "step": 14236 }, { "epoch": 1.51489678655033, "grad_norm": 16.703125, "learning_rate": 9.511903069926095e-06, "loss": 1.2186, "step": 14237 }, { "epoch": 1.5150031921685465, "grad_norm": 11.703125, "learning_rate": 9.511867538374076e-06, "loss": 1.0127, "step": 14238 }, { "epoch": 1.515109597786763, "grad_norm": 11.546875, "learning_rate": 9.51183200682206e-06, "loss": 1.1519, "step": 14239 }, { "epoch": 1.5152160034049797, "grad_norm": 12.8359375, "learning_rate": 9.511796475270041e-06, "loss": 1.1565, "step": 14240 }, { "epoch": 1.5153224090231965, "grad_norm": 13.28125, "learning_rate": 9.511760943718023e-06, "loss": 1.0879, "step": 14241 }, { "epoch": 1.515428814641413, "grad_norm": 11.4375, "learning_rate": 9.511725412166004e-06, "loss": 1.3241, "step": 14242 }, { "epoch": 1.5155352202596297, "grad_norm": 12.8203125, "learning_rate": 9.511689880613986e-06, "loss": 1.1045, "step": 14243 }, { "epoch": 1.5156416258778465, "grad_norm": 13.421875, "learning_rate": 9.511654349061967e-06, "loss": 1.3167, "step": 14244 }, { "epoch": 1.515748031496063, "grad_norm": 11.0703125, "learning_rate": 9.51161881750995e-06, "loss": 1.2525, "step": 14245 }, { "epoch": 1.5158544371142795, "grad_norm": 11.203125, "learning_rate": 9.511583285957932e-06, "loss": 1.1515, "step": 14246 }, { "epoch": 1.5159608427324964, "grad_norm": 11.671875, "learning_rate": 9.511547754405914e-06, "loss": 1.1175, "step": 14247 }, { "epoch": 1.516067248350713, "grad_norm": 10.53125, "learning_rate": 9.511512222853896e-06, "loss": 1.1215, "step": 14248 }, { "epoch": 1.5161736539689294, "grad_norm": 11.8828125, "learning_rate": 9.511476691301876e-06, "loss": 1.0802, "step": 14249 }, { "epoch": 1.5162800595871462, "grad_norm": 11.5078125, "learning_rate": 9.511441159749859e-06, "loss": 1.0508, "step": 14250 }, { "epoch": 1.516386465205363, "grad_norm": 18.890625, "learning_rate": 9.51140562819784e-06, "loss": 1.2164, "step": 14251 }, { "epoch": 1.5164928708235794, "grad_norm": 14.9609375, "learning_rate": 9.511370096645822e-06, "loss": 1.1457, "step": 14252 }, { "epoch": 1.5165992764417962, "grad_norm": 12.3046875, "learning_rate": 9.511334565093804e-06, "loss": 1.0625, "step": 14253 }, { "epoch": 1.5167056820600129, "grad_norm": 11.75, "learning_rate": 9.511299033541785e-06, "loss": 1.0206, "step": 14254 }, { "epoch": 1.5168120876782294, "grad_norm": 11.1171875, "learning_rate": 9.511263501989769e-06, "loss": 1.0836, "step": 14255 }, { "epoch": 1.516918493296446, "grad_norm": 12.1328125, "learning_rate": 9.51122797043775e-06, "loss": 0.9355, "step": 14256 }, { "epoch": 1.5170248989146629, "grad_norm": 10.2578125, "learning_rate": 9.511192438885732e-06, "loss": 0.8605, "step": 14257 }, { "epoch": 1.5171313045328794, "grad_norm": 13.984375, "learning_rate": 9.511156907333713e-06, "loss": 1.1382, "step": 14258 }, { "epoch": 1.5172377101510959, "grad_norm": 11.453125, "learning_rate": 9.511121375781695e-06, "loss": 0.8583, "step": 14259 }, { "epoch": 1.5173441157693126, "grad_norm": 17.03125, "learning_rate": 9.511085844229677e-06, "loss": 0.9922, "step": 14260 }, { "epoch": 1.5174505213875293, "grad_norm": 11.3515625, "learning_rate": 9.511050312677658e-06, "loss": 0.9723, "step": 14261 }, { "epoch": 1.5175569270057458, "grad_norm": 12.65625, "learning_rate": 9.51101478112564e-06, "loss": 1.1222, "step": 14262 }, { "epoch": 1.5176633326239626, "grad_norm": 15.71875, "learning_rate": 9.510979249573621e-06, "loss": 1.0725, "step": 14263 }, { "epoch": 1.5177697382421793, "grad_norm": 15.5859375, "learning_rate": 9.510943718021603e-06, "loss": 0.9792, "step": 14264 }, { "epoch": 1.5178761438603958, "grad_norm": 20.125, "learning_rate": 9.510908186469586e-06, "loss": 0.9194, "step": 14265 }, { "epoch": 1.5179825494786123, "grad_norm": 14.953125, "learning_rate": 9.510872654917568e-06, "loss": 0.9804, "step": 14266 }, { "epoch": 1.518088955096829, "grad_norm": 13.859375, "learning_rate": 9.51083712336555e-06, "loss": 1.2758, "step": 14267 }, { "epoch": 1.5181953607150458, "grad_norm": 15.71875, "learning_rate": 9.510801591813531e-06, "loss": 1.2438, "step": 14268 }, { "epoch": 1.5183017663332623, "grad_norm": 11.828125, "learning_rate": 9.510766060261513e-06, "loss": 0.9445, "step": 14269 }, { "epoch": 1.518408171951479, "grad_norm": 14.7421875, "learning_rate": 9.510730528709496e-06, "loss": 1.1266, "step": 14270 }, { "epoch": 1.5185145775696958, "grad_norm": 11.8125, "learning_rate": 9.510694997157478e-06, "loss": 1.2791, "step": 14271 }, { "epoch": 1.5186209831879123, "grad_norm": 13.984375, "learning_rate": 9.51065946560546e-06, "loss": 1.115, "step": 14272 }, { "epoch": 1.518727388806129, "grad_norm": 11.328125, "learning_rate": 9.51062393405344e-06, "loss": 0.9319, "step": 14273 }, { "epoch": 1.5188337944243457, "grad_norm": 16.71875, "learning_rate": 9.510588402501422e-06, "loss": 1.2761, "step": 14274 }, { "epoch": 1.5189402000425622, "grad_norm": 18.515625, "learning_rate": 9.510552870949404e-06, "loss": 0.8935, "step": 14275 }, { "epoch": 1.5190466056607788, "grad_norm": 12.5078125, "learning_rate": 9.510517339397386e-06, "loss": 0.8901, "step": 14276 }, { "epoch": 1.5191530112789955, "grad_norm": 10.3984375, "learning_rate": 9.510481807845367e-06, "loss": 1.2663, "step": 14277 }, { "epoch": 1.5192594168972122, "grad_norm": 12.3046875, "learning_rate": 9.510446276293349e-06, "loss": 1.1517, "step": 14278 }, { "epoch": 1.5193658225154287, "grad_norm": 11.9765625, "learning_rate": 9.51041074474133e-06, "loss": 1.1485, "step": 14279 }, { "epoch": 1.5194722281336455, "grad_norm": 12.6171875, "learning_rate": 9.510375213189312e-06, "loss": 1.0056, "step": 14280 }, { "epoch": 1.5195786337518622, "grad_norm": 11.6953125, "learning_rate": 9.510339681637295e-06, "loss": 1.1117, "step": 14281 }, { "epoch": 1.5196850393700787, "grad_norm": 13.265625, "learning_rate": 9.510304150085277e-06, "loss": 1.1892, "step": 14282 }, { "epoch": 1.5197914449882954, "grad_norm": 10.2890625, "learning_rate": 9.510268618533259e-06, "loss": 1.1475, "step": 14283 }, { "epoch": 1.5198978506065122, "grad_norm": 10.96875, "learning_rate": 9.51023308698124e-06, "loss": 1.025, "step": 14284 }, { "epoch": 1.5200042562247287, "grad_norm": 13.4765625, "learning_rate": 9.510197555429222e-06, "loss": 1.2492, "step": 14285 }, { "epoch": 1.5201106618429452, "grad_norm": 16.828125, "learning_rate": 9.510162023877203e-06, "loss": 1.364, "step": 14286 }, { "epoch": 1.520217067461162, "grad_norm": 13.921875, "learning_rate": 9.510126492325185e-06, "loss": 1.2418, "step": 14287 }, { "epoch": 1.5203234730793787, "grad_norm": 18.09375, "learning_rate": 9.510090960773167e-06, "loss": 1.0631, "step": 14288 }, { "epoch": 1.5204298786975952, "grad_norm": 13.546875, "learning_rate": 9.510055429221148e-06, "loss": 1.2342, "step": 14289 }, { "epoch": 1.520536284315812, "grad_norm": 13.7109375, "learning_rate": 9.51001989766913e-06, "loss": 1.2709, "step": 14290 }, { "epoch": 1.5206426899340286, "grad_norm": 12.5859375, "learning_rate": 9.509984366117113e-06, "loss": 1.0164, "step": 14291 }, { "epoch": 1.5207490955522451, "grad_norm": 10.53125, "learning_rate": 9.509948834565095e-06, "loss": 1.2431, "step": 14292 }, { "epoch": 1.5208555011704616, "grad_norm": 11.046875, "learning_rate": 9.509913303013076e-06, "loss": 0.958, "step": 14293 }, { "epoch": 1.5209619067886786, "grad_norm": 13.8046875, "learning_rate": 9.509877771461058e-06, "loss": 0.9132, "step": 14294 }, { "epoch": 1.521068312406895, "grad_norm": 11.515625, "learning_rate": 9.50984223990904e-06, "loss": 1.0779, "step": 14295 }, { "epoch": 1.5211747180251116, "grad_norm": 10.71875, "learning_rate": 9.509806708357023e-06, "loss": 1.0868, "step": 14296 }, { "epoch": 1.5212811236433283, "grad_norm": 13.125, "learning_rate": 9.509771176805004e-06, "loss": 1.1627, "step": 14297 }, { "epoch": 1.521387529261545, "grad_norm": 13.0234375, "learning_rate": 9.509735645252986e-06, "loss": 1.1456, "step": 14298 }, { "epoch": 1.5214939348797616, "grad_norm": 10.421875, "learning_rate": 9.509700113700968e-06, "loss": 1.1258, "step": 14299 }, { "epoch": 1.5216003404979783, "grad_norm": 13.5625, "learning_rate": 9.50966458214895e-06, "loss": 1.1687, "step": 14300 }, { "epoch": 1.521706746116195, "grad_norm": 16.828125, "learning_rate": 9.509629050596931e-06, "loss": 1.2167, "step": 14301 }, { "epoch": 1.5218131517344116, "grad_norm": 13.0, "learning_rate": 9.509593519044913e-06, "loss": 1.0917, "step": 14302 }, { "epoch": 1.521919557352628, "grad_norm": 15.5703125, "learning_rate": 9.509557987492894e-06, "loss": 1.0897, "step": 14303 }, { "epoch": 1.522025962970845, "grad_norm": 14.3828125, "learning_rate": 9.509522455940876e-06, "loss": 0.9806, "step": 14304 }, { "epoch": 1.5221323685890615, "grad_norm": 10.515625, "learning_rate": 9.509486924388857e-06, "loss": 1.1011, "step": 14305 }, { "epoch": 1.522238774207278, "grad_norm": 13.890625, "learning_rate": 9.50945139283684e-06, "loss": 0.9581, "step": 14306 }, { "epoch": 1.5223451798254948, "grad_norm": 9.4140625, "learning_rate": 9.509415861284822e-06, "loss": 1.179, "step": 14307 }, { "epoch": 1.5224515854437115, "grad_norm": 12.265625, "learning_rate": 9.509380329732804e-06, "loss": 1.0999, "step": 14308 }, { "epoch": 1.522557991061928, "grad_norm": 16.453125, "learning_rate": 9.509344798180785e-06, "loss": 1.1524, "step": 14309 }, { "epoch": 1.5226643966801447, "grad_norm": 12.59375, "learning_rate": 9.509309266628767e-06, "loss": 1.0439, "step": 14310 }, { "epoch": 1.5227708022983615, "grad_norm": 13.8359375, "learning_rate": 9.509273735076749e-06, "loss": 0.9386, "step": 14311 }, { "epoch": 1.522877207916578, "grad_norm": 10.2890625, "learning_rate": 9.50923820352473e-06, "loss": 0.8889, "step": 14312 }, { "epoch": 1.5229836135347945, "grad_norm": 10.609375, "learning_rate": 9.509202671972712e-06, "loss": 1.0019, "step": 14313 }, { "epoch": 1.5230900191530112, "grad_norm": 13.078125, "learning_rate": 9.509167140420693e-06, "loss": 1.3137, "step": 14314 }, { "epoch": 1.523196424771228, "grad_norm": 11.7890625, "learning_rate": 9.509131608868675e-06, "loss": 1.3261, "step": 14315 }, { "epoch": 1.5233028303894445, "grad_norm": 13.5546875, "learning_rate": 9.509096077316658e-06, "loss": 0.9865, "step": 14316 }, { "epoch": 1.5234092360076612, "grad_norm": 15.140625, "learning_rate": 9.50906054576464e-06, "loss": 0.92, "step": 14317 }, { "epoch": 1.523515641625878, "grad_norm": 14.9921875, "learning_rate": 9.509025014212622e-06, "loss": 1.4876, "step": 14318 }, { "epoch": 1.5236220472440944, "grad_norm": 12.8203125, "learning_rate": 9.508989482660603e-06, "loss": 1.1377, "step": 14319 }, { "epoch": 1.5237284528623112, "grad_norm": 10.7890625, "learning_rate": 9.508953951108585e-06, "loss": 1.0063, "step": 14320 }, { "epoch": 1.523834858480528, "grad_norm": 9.25, "learning_rate": 9.508918419556566e-06, "loss": 1.1803, "step": 14321 }, { "epoch": 1.5239412640987444, "grad_norm": 13.9140625, "learning_rate": 9.50888288800455e-06, "loss": 1.2212, "step": 14322 }, { "epoch": 1.524047669716961, "grad_norm": 13.203125, "learning_rate": 9.508847356452531e-06, "loss": 1.31, "step": 14323 }, { "epoch": 1.5241540753351777, "grad_norm": 16.1875, "learning_rate": 9.508811824900513e-06, "loss": 1.2638, "step": 14324 }, { "epoch": 1.5242604809533944, "grad_norm": 10.7109375, "learning_rate": 9.508776293348495e-06, "loss": 1.126, "step": 14325 }, { "epoch": 1.524366886571611, "grad_norm": 12.2890625, "learning_rate": 9.508740761796476e-06, "loss": 0.9644, "step": 14326 }, { "epoch": 1.5244732921898276, "grad_norm": 10.984375, "learning_rate": 9.508705230244458e-06, "loss": 1.0847, "step": 14327 }, { "epoch": 1.5245796978080444, "grad_norm": 12.46875, "learning_rate": 9.50866969869244e-06, "loss": 0.9505, "step": 14328 }, { "epoch": 1.5246861034262609, "grad_norm": 15.421875, "learning_rate": 9.508634167140421e-06, "loss": 1.3802, "step": 14329 }, { "epoch": 1.5247925090444774, "grad_norm": 12.625, "learning_rate": 9.508598635588403e-06, "loss": 1.0847, "step": 14330 }, { "epoch": 1.5248989146626943, "grad_norm": 18.640625, "learning_rate": 9.508563104036384e-06, "loss": 1.2529, "step": 14331 }, { "epoch": 1.5250053202809108, "grad_norm": 20.5, "learning_rate": 9.508527572484367e-06, "loss": 1.019, "step": 14332 }, { "epoch": 1.5251117258991274, "grad_norm": 15.6796875, "learning_rate": 9.508492040932349e-06, "loss": 0.9965, "step": 14333 }, { "epoch": 1.525218131517344, "grad_norm": 11.4140625, "learning_rate": 9.50845650938033e-06, "loss": 0.8205, "step": 14334 }, { "epoch": 1.5253245371355608, "grad_norm": 16.359375, "learning_rate": 9.508420977828312e-06, "loss": 1.0576, "step": 14335 }, { "epoch": 1.5254309427537773, "grad_norm": 17.1875, "learning_rate": 9.508385446276294e-06, "loss": 0.8873, "step": 14336 }, { "epoch": 1.525537348371994, "grad_norm": 13.015625, "learning_rate": 9.508349914724277e-06, "loss": 1.1914, "step": 14337 }, { "epoch": 1.5256437539902108, "grad_norm": 13.9765625, "learning_rate": 9.508314383172257e-06, "loss": 1.5008, "step": 14338 }, { "epoch": 1.5257501596084273, "grad_norm": 19.46875, "learning_rate": 9.508278851620239e-06, "loss": 0.9955, "step": 14339 }, { "epoch": 1.5258565652266438, "grad_norm": 12.90625, "learning_rate": 9.50824332006822e-06, "loss": 0.9239, "step": 14340 }, { "epoch": 1.5259629708448608, "grad_norm": 14.9140625, "learning_rate": 9.508207788516202e-06, "loss": 0.9592, "step": 14341 }, { "epoch": 1.5260693764630773, "grad_norm": 15.890625, "learning_rate": 9.508172256964185e-06, "loss": 0.9124, "step": 14342 }, { "epoch": 1.5261757820812938, "grad_norm": 10.28125, "learning_rate": 9.508136725412167e-06, "loss": 0.9758, "step": 14343 }, { "epoch": 1.5262821876995105, "grad_norm": 13.3671875, "learning_rate": 9.508101193860148e-06, "loss": 1.0207, "step": 14344 }, { "epoch": 1.5263885933177272, "grad_norm": 14.6875, "learning_rate": 9.50806566230813e-06, "loss": 1.3114, "step": 14345 }, { "epoch": 1.5264949989359438, "grad_norm": 10.5234375, "learning_rate": 9.508030130756112e-06, "loss": 1.1238, "step": 14346 }, { "epoch": 1.5266014045541605, "grad_norm": 13.6875, "learning_rate": 9.507994599204095e-06, "loss": 0.9996, "step": 14347 }, { "epoch": 1.5267078101723772, "grad_norm": 12.734375, "learning_rate": 9.507959067652077e-06, "loss": 1.2316, "step": 14348 }, { "epoch": 1.5268142157905937, "grad_norm": 13.03125, "learning_rate": 9.507923536100058e-06, "loss": 1.164, "step": 14349 }, { "epoch": 1.5269206214088102, "grad_norm": 12.8046875, "learning_rate": 9.50788800454804e-06, "loss": 1.0635, "step": 14350 }, { "epoch": 1.527027027027027, "grad_norm": 11.6015625, "learning_rate": 9.507852472996021e-06, "loss": 1.1945, "step": 14351 }, { "epoch": 1.5271334326452437, "grad_norm": 11.484375, "learning_rate": 9.507816941444003e-06, "loss": 1.1874, "step": 14352 }, { "epoch": 1.5272398382634602, "grad_norm": 15.4296875, "learning_rate": 9.507781409891985e-06, "loss": 0.9204, "step": 14353 }, { "epoch": 1.527346243881677, "grad_norm": 15.484375, "learning_rate": 9.507745878339966e-06, "loss": 1.3227, "step": 14354 }, { "epoch": 1.5274526494998937, "grad_norm": 11.6953125, "learning_rate": 9.507710346787948e-06, "loss": 1.4137, "step": 14355 }, { "epoch": 1.5275590551181102, "grad_norm": 10.796875, "learning_rate": 9.50767481523593e-06, "loss": 1.281, "step": 14356 }, { "epoch": 1.527665460736327, "grad_norm": 10.0078125, "learning_rate": 9.507639283683913e-06, "loss": 1.017, "step": 14357 }, { "epoch": 1.5277718663545437, "grad_norm": 14.09375, "learning_rate": 9.507603752131894e-06, "loss": 1.0562, "step": 14358 }, { "epoch": 1.5278782719727602, "grad_norm": 10.4609375, "learning_rate": 9.507568220579876e-06, "loss": 1.1961, "step": 14359 }, { "epoch": 1.5279846775909767, "grad_norm": 15.3671875, "learning_rate": 9.507532689027858e-06, "loss": 0.8585, "step": 14360 }, { "epoch": 1.5280910832091934, "grad_norm": 15.7109375, "learning_rate": 9.50749715747584e-06, "loss": 0.9004, "step": 14361 }, { "epoch": 1.5281974888274101, "grad_norm": 10.609375, "learning_rate": 9.50746162592382e-06, "loss": 0.9451, "step": 14362 }, { "epoch": 1.5283038944456266, "grad_norm": 14.2890625, "learning_rate": 9.507426094371804e-06, "loss": 1.2761, "step": 14363 }, { "epoch": 1.5284103000638434, "grad_norm": 11.6875, "learning_rate": 9.507390562819786e-06, "loss": 1.0937, "step": 14364 }, { "epoch": 1.52851670568206, "grad_norm": 12.7109375, "learning_rate": 9.507355031267767e-06, "loss": 1.0547, "step": 14365 }, { "epoch": 1.5286231113002766, "grad_norm": 21.640625, "learning_rate": 9.507319499715749e-06, "loss": 1.263, "step": 14366 }, { "epoch": 1.5287295169184933, "grad_norm": 15.734375, "learning_rate": 9.50728396816373e-06, "loss": 1.0554, "step": 14367 }, { "epoch": 1.52883592253671, "grad_norm": 13.234375, "learning_rate": 9.507248436611712e-06, "loss": 1.0392, "step": 14368 }, { "epoch": 1.5289423281549266, "grad_norm": 14.015625, "learning_rate": 9.507212905059694e-06, "loss": 1.0981, "step": 14369 }, { "epoch": 1.529048733773143, "grad_norm": 12.71875, "learning_rate": 9.507177373507675e-06, "loss": 0.9629, "step": 14370 }, { "epoch": 1.5291551393913598, "grad_norm": 13.6640625, "learning_rate": 9.507141841955657e-06, "loss": 0.9919, "step": 14371 }, { "epoch": 1.5292615450095766, "grad_norm": 10.296875, "learning_rate": 9.507106310403639e-06, "loss": 0.9172, "step": 14372 }, { "epoch": 1.529367950627793, "grad_norm": 10.5703125, "learning_rate": 9.507070778851622e-06, "loss": 1.2483, "step": 14373 }, { "epoch": 1.5294743562460098, "grad_norm": 13.796875, "learning_rate": 9.507035247299603e-06, "loss": 1.023, "step": 14374 }, { "epoch": 1.5295807618642265, "grad_norm": 11.625, "learning_rate": 9.506999715747585e-06, "loss": 0.7708, "step": 14375 }, { "epoch": 1.529687167482443, "grad_norm": 11.296875, "learning_rate": 9.506964184195567e-06, "loss": 1.1479, "step": 14376 }, { "epoch": 1.5297935731006596, "grad_norm": 12.6015625, "learning_rate": 9.506928652643548e-06, "loss": 1.2791, "step": 14377 }, { "epoch": 1.5298999787188765, "grad_norm": 10.53125, "learning_rate": 9.50689312109153e-06, "loss": 1.0789, "step": 14378 }, { "epoch": 1.530006384337093, "grad_norm": 11.6875, "learning_rate": 9.506857589539511e-06, "loss": 1.0745, "step": 14379 }, { "epoch": 1.5301127899553095, "grad_norm": 14.96875, "learning_rate": 9.506822057987493e-06, "loss": 1.5251, "step": 14380 }, { "epoch": 1.5302191955735263, "grad_norm": 15.1796875, "learning_rate": 9.506786526435475e-06, "loss": 1.3173, "step": 14381 }, { "epoch": 1.530325601191743, "grad_norm": 11.609375, "learning_rate": 9.506750994883456e-06, "loss": 1.1943, "step": 14382 }, { "epoch": 1.5304320068099595, "grad_norm": 14.078125, "learning_rate": 9.50671546333144e-06, "loss": 1.2451, "step": 14383 }, { "epoch": 1.5305384124281762, "grad_norm": 11.8828125, "learning_rate": 9.506679931779421e-06, "loss": 0.9156, "step": 14384 }, { "epoch": 1.530644818046393, "grad_norm": 17.71875, "learning_rate": 9.506644400227403e-06, "loss": 1.1646, "step": 14385 }, { "epoch": 1.5307512236646095, "grad_norm": 13.9609375, "learning_rate": 9.506608868675384e-06, "loss": 1.1708, "step": 14386 }, { "epoch": 1.530857629282826, "grad_norm": 10.6796875, "learning_rate": 9.506573337123366e-06, "loss": 1.003, "step": 14387 }, { "epoch": 1.530964034901043, "grad_norm": 15.9609375, "learning_rate": 9.50653780557135e-06, "loss": 0.8695, "step": 14388 }, { "epoch": 1.5310704405192594, "grad_norm": 15.09375, "learning_rate": 9.506502274019331e-06, "loss": 1.0537, "step": 14389 }, { "epoch": 1.531176846137476, "grad_norm": 17.3125, "learning_rate": 9.506466742467313e-06, "loss": 1.1772, "step": 14390 }, { "epoch": 1.5312832517556927, "grad_norm": 13.484375, "learning_rate": 9.506431210915294e-06, "loss": 1.2063, "step": 14391 }, { "epoch": 1.5313896573739094, "grad_norm": 11.7109375, "learning_rate": 9.506395679363276e-06, "loss": 1.071, "step": 14392 }, { "epoch": 1.531496062992126, "grad_norm": 11.09375, "learning_rate": 9.506360147811257e-06, "loss": 1.0755, "step": 14393 }, { "epoch": 1.5316024686103427, "grad_norm": 13.1640625, "learning_rate": 9.506324616259239e-06, "loss": 1.0177, "step": 14394 }, { "epoch": 1.5317088742285594, "grad_norm": 11.359375, "learning_rate": 9.50628908470722e-06, "loss": 1.3016, "step": 14395 }, { "epoch": 1.531815279846776, "grad_norm": 14.0234375, "learning_rate": 9.506253553155202e-06, "loss": 0.9091, "step": 14396 }, { "epoch": 1.5319216854649924, "grad_norm": 12.7109375, "learning_rate": 9.506218021603184e-06, "loss": 1.2996, "step": 14397 }, { "epoch": 1.5320280910832091, "grad_norm": 9.953125, "learning_rate": 9.506182490051165e-06, "loss": 0.8021, "step": 14398 }, { "epoch": 1.5321344967014259, "grad_norm": 12.3828125, "learning_rate": 9.506146958499149e-06, "loss": 1.1166, "step": 14399 }, { "epoch": 1.5322409023196424, "grad_norm": 12.8671875, "learning_rate": 9.50611142694713e-06, "loss": 1.0216, "step": 14400 }, { "epoch": 1.5323473079378591, "grad_norm": 8.5546875, "learning_rate": 9.506075895395112e-06, "loss": 1.0201, "step": 14401 }, { "epoch": 1.5324537135560758, "grad_norm": 10.4765625, "learning_rate": 9.506040363843094e-06, "loss": 1.0888, "step": 14402 }, { "epoch": 1.5325601191742924, "grad_norm": 12.1171875, "learning_rate": 9.506004832291075e-06, "loss": 1.0272, "step": 14403 }, { "epoch": 1.532666524792509, "grad_norm": 12.2734375, "learning_rate": 9.505969300739057e-06, "loss": 0.9122, "step": 14404 }, { "epoch": 1.5327729304107258, "grad_norm": 16.265625, "learning_rate": 9.505933769187038e-06, "loss": 1.18, "step": 14405 }, { "epoch": 1.5328793360289423, "grad_norm": 13.296875, "learning_rate": 9.50589823763502e-06, "loss": 1.1216, "step": 14406 }, { "epoch": 1.5329857416471588, "grad_norm": 11.9609375, "learning_rate": 9.505862706083002e-06, "loss": 1.1266, "step": 14407 }, { "epoch": 1.5330921472653756, "grad_norm": 14.265625, "learning_rate": 9.505827174530983e-06, "loss": 1.0512, "step": 14408 }, { "epoch": 1.5331985528835923, "grad_norm": 10.3515625, "learning_rate": 9.505791642978966e-06, "loss": 1.1088, "step": 14409 }, { "epoch": 1.5333049585018088, "grad_norm": 12.71875, "learning_rate": 9.505756111426948e-06, "loss": 1.026, "step": 14410 }, { "epoch": 1.5334113641200255, "grad_norm": 12.0390625, "learning_rate": 9.50572057987493e-06, "loss": 1.1382, "step": 14411 }, { "epoch": 1.5335177697382423, "grad_norm": 10.4296875, "learning_rate": 9.505685048322911e-06, "loss": 1.1526, "step": 14412 }, { "epoch": 1.5336241753564588, "grad_norm": 12.21875, "learning_rate": 9.505649516770893e-06, "loss": 1.0837, "step": 14413 }, { "epoch": 1.5337305809746755, "grad_norm": 11.1328125, "learning_rate": 9.505613985218876e-06, "loss": 1.1695, "step": 14414 }, { "epoch": 1.5338369865928922, "grad_norm": 16.4375, "learning_rate": 9.505578453666858e-06, "loss": 1.2181, "step": 14415 }, { "epoch": 1.5339433922111088, "grad_norm": 13.1953125, "learning_rate": 9.50554292211484e-06, "loss": 1.2559, "step": 14416 }, { "epoch": 1.5340497978293253, "grad_norm": 22.75, "learning_rate": 9.505507390562821e-06, "loss": 1.0179, "step": 14417 }, { "epoch": 1.534156203447542, "grad_norm": 10.8125, "learning_rate": 9.505471859010803e-06, "loss": 0.9964, "step": 14418 }, { "epoch": 1.5342626090657587, "grad_norm": 12.796875, "learning_rate": 9.505436327458784e-06, "loss": 0.9095, "step": 14419 }, { "epoch": 1.5343690146839752, "grad_norm": 11.609375, "learning_rate": 9.505400795906766e-06, "loss": 1.0351, "step": 14420 }, { "epoch": 1.534475420302192, "grad_norm": 12.5078125, "learning_rate": 9.505365264354747e-06, "loss": 1.0955, "step": 14421 }, { "epoch": 1.5345818259204087, "grad_norm": 13.6875, "learning_rate": 9.505329732802729e-06, "loss": 0.9316, "step": 14422 }, { "epoch": 1.5346882315386252, "grad_norm": 21.859375, "learning_rate": 9.50529420125071e-06, "loss": 1.0782, "step": 14423 }, { "epoch": 1.5347946371568417, "grad_norm": 12.890625, "learning_rate": 9.505258669698694e-06, "loss": 1.2101, "step": 14424 }, { "epoch": 1.5349010427750587, "grad_norm": 13.9375, "learning_rate": 9.505223138146676e-06, "loss": 1.0684, "step": 14425 }, { "epoch": 1.5350074483932752, "grad_norm": 14.515625, "learning_rate": 9.505187606594657e-06, "loss": 1.0464, "step": 14426 }, { "epoch": 1.5351138540114917, "grad_norm": 10.90625, "learning_rate": 9.505152075042639e-06, "loss": 1.2087, "step": 14427 }, { "epoch": 1.5352202596297084, "grad_norm": 14.90625, "learning_rate": 9.50511654349062e-06, "loss": 0.9602, "step": 14428 }, { "epoch": 1.5353266652479252, "grad_norm": 12.109375, "learning_rate": 9.505081011938602e-06, "loss": 1.179, "step": 14429 }, { "epoch": 1.5354330708661417, "grad_norm": 11.2578125, "learning_rate": 9.505045480386584e-06, "loss": 1.2758, "step": 14430 }, { "epoch": 1.5355394764843584, "grad_norm": 14.1796875, "learning_rate": 9.505009948834565e-06, "loss": 1.2774, "step": 14431 }, { "epoch": 1.5356458821025751, "grad_norm": 12.5625, "learning_rate": 9.504974417282547e-06, "loss": 1.1168, "step": 14432 }, { "epoch": 1.5357522877207916, "grad_norm": 12.203125, "learning_rate": 9.504938885730528e-06, "loss": 1.1928, "step": 14433 }, { "epoch": 1.5358586933390082, "grad_norm": 18.078125, "learning_rate": 9.504903354178512e-06, "loss": 1.2392, "step": 14434 }, { "epoch": 1.535965098957225, "grad_norm": 8.5703125, "learning_rate": 9.504867822626493e-06, "loss": 0.9185, "step": 14435 }, { "epoch": 1.5360715045754416, "grad_norm": 9.1171875, "learning_rate": 9.504832291074475e-06, "loss": 0.9541, "step": 14436 }, { "epoch": 1.5361779101936581, "grad_norm": 15.0625, "learning_rate": 9.504796759522457e-06, "loss": 1.0648, "step": 14437 }, { "epoch": 1.5362843158118749, "grad_norm": 13.21875, "learning_rate": 9.504761227970438e-06, "loss": 1.5189, "step": 14438 }, { "epoch": 1.5363907214300916, "grad_norm": 16.078125, "learning_rate": 9.50472569641842e-06, "loss": 1.2487, "step": 14439 }, { "epoch": 1.536497127048308, "grad_norm": 13.8671875, "learning_rate": 9.504690164866403e-06, "loss": 0.8505, "step": 14440 }, { "epoch": 1.5366035326665248, "grad_norm": 9.8671875, "learning_rate": 9.504654633314385e-06, "loss": 1.1586, "step": 14441 }, { "epoch": 1.5367099382847416, "grad_norm": 141.25, "learning_rate": 9.504619101762366e-06, "loss": 0.925, "step": 14442 }, { "epoch": 1.536816343902958, "grad_norm": 10.6796875, "learning_rate": 9.504583570210348e-06, "loss": 1.442, "step": 14443 }, { "epoch": 1.5369227495211746, "grad_norm": 13.5390625, "learning_rate": 9.50454803865833e-06, "loss": 1.2589, "step": 14444 }, { "epoch": 1.5370291551393913, "grad_norm": 12.828125, "learning_rate": 9.504512507106311e-06, "loss": 0.9494, "step": 14445 }, { "epoch": 1.537135560757608, "grad_norm": 12.0703125, "learning_rate": 9.504476975554293e-06, "loss": 1.1534, "step": 14446 }, { "epoch": 1.5372419663758246, "grad_norm": 12.046875, "learning_rate": 9.504441444002274e-06, "loss": 0.8709, "step": 14447 }, { "epoch": 1.5373483719940413, "grad_norm": 9.0546875, "learning_rate": 9.504405912450256e-06, "loss": 1.0045, "step": 14448 }, { "epoch": 1.537454777612258, "grad_norm": 9.75, "learning_rate": 9.504370380898237e-06, "loss": 1.0194, "step": 14449 }, { "epoch": 1.5375611832304745, "grad_norm": 11.3203125, "learning_rate": 9.50433484934622e-06, "loss": 1.0994, "step": 14450 }, { "epoch": 1.5376675888486913, "grad_norm": 14.03125, "learning_rate": 9.504299317794202e-06, "loss": 1.0592, "step": 14451 }, { "epoch": 1.537773994466908, "grad_norm": 13.328125, "learning_rate": 9.504263786242184e-06, "loss": 1.3475, "step": 14452 }, { "epoch": 1.5378804000851245, "grad_norm": 11.3125, "learning_rate": 9.504228254690166e-06, "loss": 1.1567, "step": 14453 }, { "epoch": 1.537986805703341, "grad_norm": 11.328125, "learning_rate": 9.504192723138147e-06, "loss": 0.9744, "step": 14454 }, { "epoch": 1.5380932113215577, "grad_norm": 11.7265625, "learning_rate": 9.50415719158613e-06, "loss": 1.2527, "step": 14455 }, { "epoch": 1.5381996169397745, "grad_norm": 12.5390625, "learning_rate": 9.504121660034112e-06, "loss": 1.1286, "step": 14456 }, { "epoch": 1.538306022557991, "grad_norm": 11.4453125, "learning_rate": 9.504086128482094e-06, "loss": 1.195, "step": 14457 }, { "epoch": 1.5384124281762077, "grad_norm": 13.375, "learning_rate": 9.504050596930074e-06, "loss": 1.2682, "step": 14458 }, { "epoch": 1.5385188337944244, "grad_norm": 9.6796875, "learning_rate": 9.504015065378055e-06, "loss": 1.0541, "step": 14459 }, { "epoch": 1.538625239412641, "grad_norm": 11.046875, "learning_rate": 9.503979533826039e-06, "loss": 1.1629, "step": 14460 }, { "epoch": 1.5387316450308575, "grad_norm": 11.0, "learning_rate": 9.50394400227402e-06, "loss": 1.0393, "step": 14461 }, { "epoch": 1.5388380506490744, "grad_norm": 15.34375, "learning_rate": 9.503908470722002e-06, "loss": 1.1315, "step": 14462 }, { "epoch": 1.538944456267291, "grad_norm": 17.34375, "learning_rate": 9.503872939169983e-06, "loss": 0.9396, "step": 14463 }, { "epoch": 1.5390508618855074, "grad_norm": 8.9609375, "learning_rate": 9.503837407617965e-06, "loss": 0.984, "step": 14464 }, { "epoch": 1.5391572675037242, "grad_norm": 11.640625, "learning_rate": 9.503801876065948e-06, "loss": 0.8312, "step": 14465 }, { "epoch": 1.539263673121941, "grad_norm": 12.375, "learning_rate": 9.50376634451393e-06, "loss": 1.0853, "step": 14466 }, { "epoch": 1.5393700787401574, "grad_norm": 12.640625, "learning_rate": 9.503730812961911e-06, "loss": 1.0558, "step": 14467 }, { "epoch": 1.5394764843583741, "grad_norm": 11.09375, "learning_rate": 9.503695281409893e-06, "loss": 0.8111, "step": 14468 }, { "epoch": 1.5395828899765909, "grad_norm": 9.96875, "learning_rate": 9.503659749857875e-06, "loss": 1.1669, "step": 14469 }, { "epoch": 1.5396892955948074, "grad_norm": 9.7578125, "learning_rate": 9.503624218305856e-06, "loss": 1.1738, "step": 14470 }, { "epoch": 1.539795701213024, "grad_norm": 14.671875, "learning_rate": 9.503588686753838e-06, "loss": 1.1972, "step": 14471 }, { "epoch": 1.5399021068312408, "grad_norm": 15.0390625, "learning_rate": 9.50355315520182e-06, "loss": 1.1323, "step": 14472 }, { "epoch": 1.5400085124494574, "grad_norm": 15.34375, "learning_rate": 9.503517623649801e-06, "loss": 1.0734, "step": 14473 }, { "epoch": 1.5401149180676739, "grad_norm": 11.96875, "learning_rate": 9.503482092097783e-06, "loss": 0.9361, "step": 14474 }, { "epoch": 1.5402213236858906, "grad_norm": 11.6328125, "learning_rate": 9.503446560545764e-06, "loss": 0.9457, "step": 14475 }, { "epoch": 1.5403277293041073, "grad_norm": 12.859375, "learning_rate": 9.503411028993748e-06, "loss": 1.2012, "step": 14476 }, { "epoch": 1.5404341349223238, "grad_norm": 11.421875, "learning_rate": 9.50337549744173e-06, "loss": 1.0214, "step": 14477 }, { "epoch": 1.5405405405405406, "grad_norm": 12.0703125, "learning_rate": 9.503339965889711e-06, "loss": 1.0732, "step": 14478 }, { "epoch": 1.5406469461587573, "grad_norm": 11.0546875, "learning_rate": 9.503304434337692e-06, "loss": 1.0743, "step": 14479 }, { "epoch": 1.5407533517769738, "grad_norm": 11.2109375, "learning_rate": 9.503268902785674e-06, "loss": 1.1586, "step": 14480 }, { "epoch": 1.5408597573951903, "grad_norm": 12.3203125, "learning_rate": 9.503233371233657e-06, "loss": 1.0164, "step": 14481 }, { "epoch": 1.540966163013407, "grad_norm": 19.34375, "learning_rate": 9.503197839681639e-06, "loss": 1.1875, "step": 14482 }, { "epoch": 1.5410725686316238, "grad_norm": 10.09375, "learning_rate": 9.50316230812962e-06, "loss": 0.9559, "step": 14483 }, { "epoch": 1.5411789742498403, "grad_norm": 14.265625, "learning_rate": 9.503126776577602e-06, "loss": 1.2156, "step": 14484 }, { "epoch": 1.541285379868057, "grad_norm": 12.1640625, "learning_rate": 9.503091245025584e-06, "loss": 1.1017, "step": 14485 }, { "epoch": 1.5413917854862738, "grad_norm": 9.40625, "learning_rate": 9.503055713473565e-06, "loss": 1.0119, "step": 14486 }, { "epoch": 1.5414981911044903, "grad_norm": 17.1875, "learning_rate": 9.503020181921547e-06, "loss": 1.0142, "step": 14487 }, { "epoch": 1.541604596722707, "grad_norm": 13.0234375, "learning_rate": 9.502984650369529e-06, "loss": 1.0564, "step": 14488 }, { "epoch": 1.5417110023409237, "grad_norm": 14.7890625, "learning_rate": 9.50294911881751e-06, "loss": 1.195, "step": 14489 }, { "epoch": 1.5418174079591402, "grad_norm": 10.3828125, "learning_rate": 9.502913587265492e-06, "loss": 1.0033, "step": 14490 }, { "epoch": 1.5419238135773567, "grad_norm": 11.4375, "learning_rate": 9.502878055713475e-06, "loss": 0.8074, "step": 14491 }, { "epoch": 1.5420302191955735, "grad_norm": 11.8125, "learning_rate": 9.502842524161457e-06, "loss": 1.0158, "step": 14492 }, { "epoch": 1.5421366248137902, "grad_norm": 11.359375, "learning_rate": 9.502806992609438e-06, "loss": 0.9952, "step": 14493 }, { "epoch": 1.5422430304320067, "grad_norm": 16.953125, "learning_rate": 9.50277146105742e-06, "loss": 1.1983, "step": 14494 }, { "epoch": 1.5423494360502235, "grad_norm": 12.4921875, "learning_rate": 9.502735929505402e-06, "loss": 0.8979, "step": 14495 }, { "epoch": 1.5424558416684402, "grad_norm": 15.9765625, "learning_rate": 9.502700397953383e-06, "loss": 1.135, "step": 14496 }, { "epoch": 1.5425622472866567, "grad_norm": 12.0859375, "learning_rate": 9.502664866401365e-06, "loss": 0.967, "step": 14497 }, { "epoch": 1.5426686529048734, "grad_norm": 13.4140625, "learning_rate": 9.502629334849346e-06, "loss": 0.8151, "step": 14498 }, { "epoch": 1.5427750585230902, "grad_norm": 15.5546875, "learning_rate": 9.502593803297328e-06, "loss": 1.0178, "step": 14499 }, { "epoch": 1.5428814641413067, "grad_norm": 12.296875, "learning_rate": 9.50255827174531e-06, "loss": 1.2317, "step": 14500 }, { "epoch": 1.5429878697595232, "grad_norm": 12.15625, "learning_rate": 9.502522740193293e-06, "loss": 1.0607, "step": 14501 }, { "epoch": 1.54309427537774, "grad_norm": 13.6875, "learning_rate": 9.502487208641275e-06, "loss": 1.258, "step": 14502 }, { "epoch": 1.5432006809959566, "grad_norm": 16.171875, "learning_rate": 9.502451677089256e-06, "loss": 0.9741, "step": 14503 }, { "epoch": 1.5433070866141732, "grad_norm": 11.5859375, "learning_rate": 9.502416145537238e-06, "loss": 1.3553, "step": 14504 }, { "epoch": 1.5434134922323899, "grad_norm": 11.40625, "learning_rate": 9.50238061398522e-06, "loss": 1.2722, "step": 14505 }, { "epoch": 1.5435198978506066, "grad_norm": 12.28125, "learning_rate": 9.502345082433203e-06, "loss": 1.0544, "step": 14506 }, { "epoch": 1.5436263034688231, "grad_norm": 11.9375, "learning_rate": 9.502309550881184e-06, "loss": 1.0301, "step": 14507 }, { "epoch": 1.5437327090870396, "grad_norm": 12.4609375, "learning_rate": 9.502274019329166e-06, "loss": 1.2638, "step": 14508 }, { "epoch": 1.5438391147052566, "grad_norm": 14.390625, "learning_rate": 9.502238487777147e-06, "loss": 1.2124, "step": 14509 }, { "epoch": 1.543945520323473, "grad_norm": 12.796875, "learning_rate": 9.502202956225129e-06, "loss": 1.3922, "step": 14510 }, { "epoch": 1.5440519259416896, "grad_norm": 13.1953125, "learning_rate": 9.50216742467311e-06, "loss": 1.1339, "step": 14511 }, { "epoch": 1.5441583315599063, "grad_norm": 11.4609375, "learning_rate": 9.502131893121092e-06, "loss": 1.201, "step": 14512 }, { "epoch": 1.544264737178123, "grad_norm": 16.5625, "learning_rate": 9.502096361569074e-06, "loss": 1.2801, "step": 14513 }, { "epoch": 1.5443711427963396, "grad_norm": 13.1484375, "learning_rate": 9.502060830017055e-06, "loss": 1.3149, "step": 14514 }, { "epoch": 1.5444775484145563, "grad_norm": 17.203125, "learning_rate": 9.502025298465037e-06, "loss": 1.1409, "step": 14515 }, { "epoch": 1.544583954032773, "grad_norm": 11.328125, "learning_rate": 9.501989766913019e-06, "loss": 0.9462, "step": 14516 }, { "epoch": 1.5446903596509896, "grad_norm": 10.9609375, "learning_rate": 9.501954235361002e-06, "loss": 0.9952, "step": 14517 }, { "epoch": 1.544796765269206, "grad_norm": 10.203125, "learning_rate": 9.501918703808984e-06, "loss": 1.0901, "step": 14518 }, { "epoch": 1.544903170887423, "grad_norm": 13.796875, "learning_rate": 9.501883172256965e-06, "loss": 1.6362, "step": 14519 }, { "epoch": 1.5450095765056395, "grad_norm": 17.46875, "learning_rate": 9.501847640704947e-06, "loss": 1.3899, "step": 14520 }, { "epoch": 1.545115982123856, "grad_norm": 12.9609375, "learning_rate": 9.501812109152928e-06, "loss": 1.2875, "step": 14521 }, { "epoch": 1.5452223877420728, "grad_norm": 10.328125, "learning_rate": 9.50177657760091e-06, "loss": 1.0616, "step": 14522 }, { "epoch": 1.5453287933602895, "grad_norm": 11.8359375, "learning_rate": 9.501741046048892e-06, "loss": 1.0662, "step": 14523 }, { "epoch": 1.545435198978506, "grad_norm": 12.046875, "learning_rate": 9.501705514496873e-06, "loss": 1.2822, "step": 14524 }, { "epoch": 1.5455416045967227, "grad_norm": 23.984375, "learning_rate": 9.501669982944855e-06, "loss": 1.057, "step": 14525 }, { "epoch": 1.5456480102149395, "grad_norm": 13.6953125, "learning_rate": 9.501634451392836e-06, "loss": 0.9337, "step": 14526 }, { "epoch": 1.545754415833156, "grad_norm": 15.6015625, "learning_rate": 9.50159891984082e-06, "loss": 1.1129, "step": 14527 }, { "epoch": 1.5458608214513725, "grad_norm": 16.859375, "learning_rate": 9.501563388288801e-06, "loss": 1.277, "step": 14528 }, { "epoch": 1.5459672270695892, "grad_norm": 10.28125, "learning_rate": 9.501527856736783e-06, "loss": 1.2275, "step": 14529 }, { "epoch": 1.546073632687806, "grad_norm": 13.4453125, "learning_rate": 9.501492325184765e-06, "loss": 1.2149, "step": 14530 }, { "epoch": 1.5461800383060225, "grad_norm": 12.3125, "learning_rate": 9.501456793632746e-06, "loss": 1.0304, "step": 14531 }, { "epoch": 1.5462864439242392, "grad_norm": 9.828125, "learning_rate": 9.50142126208073e-06, "loss": 0.9014, "step": 14532 }, { "epoch": 1.546392849542456, "grad_norm": 11.2578125, "learning_rate": 9.501385730528711e-06, "loss": 1.3097, "step": 14533 }, { "epoch": 1.5464992551606724, "grad_norm": 10.234375, "learning_rate": 9.501350198976693e-06, "loss": 1.1448, "step": 14534 }, { "epoch": 1.5466056607788892, "grad_norm": 14.9921875, "learning_rate": 9.501314667424674e-06, "loss": 1.3166, "step": 14535 }, { "epoch": 1.546712066397106, "grad_norm": 14.8828125, "learning_rate": 9.501279135872656e-06, "loss": 0.8397, "step": 14536 }, { "epoch": 1.5468184720153224, "grad_norm": 13.7265625, "learning_rate": 9.501243604320638e-06, "loss": 1.1329, "step": 14537 }, { "epoch": 1.546924877633539, "grad_norm": 13.3125, "learning_rate": 9.501208072768619e-06, "loss": 1.334, "step": 14538 }, { "epoch": 1.5470312832517557, "grad_norm": 11.0859375, "learning_rate": 9.5011725412166e-06, "loss": 0.9645, "step": 14539 }, { "epoch": 1.5471376888699724, "grad_norm": 12.9921875, "learning_rate": 9.501137009664582e-06, "loss": 0.9921, "step": 14540 }, { "epoch": 1.547244094488189, "grad_norm": 11.3515625, "learning_rate": 9.501101478112564e-06, "loss": 0.9259, "step": 14541 }, { "epoch": 1.5473505001064056, "grad_norm": 16.234375, "learning_rate": 9.501065946560547e-06, "loss": 0.9144, "step": 14542 }, { "epoch": 1.5474569057246224, "grad_norm": 9.3203125, "learning_rate": 9.501030415008529e-06, "loss": 1.1959, "step": 14543 }, { "epoch": 1.5475633113428389, "grad_norm": 11.625, "learning_rate": 9.50099488345651e-06, "loss": 1.3719, "step": 14544 }, { "epoch": 1.5476697169610556, "grad_norm": 12.65625, "learning_rate": 9.500959351904492e-06, "loss": 1.2912, "step": 14545 }, { "epoch": 1.5477761225792723, "grad_norm": 12.1484375, "learning_rate": 9.500923820352474e-06, "loss": 1.0349, "step": 14546 }, { "epoch": 1.5478825281974888, "grad_norm": 13.171875, "learning_rate": 9.500888288800455e-06, "loss": 1.1274, "step": 14547 }, { "epoch": 1.5479889338157053, "grad_norm": 12.1953125, "learning_rate": 9.500852757248437e-06, "loss": 1.0758, "step": 14548 }, { "epoch": 1.548095339433922, "grad_norm": 13.5546875, "learning_rate": 9.500817225696418e-06, "loss": 1.0875, "step": 14549 }, { "epoch": 1.5482017450521388, "grad_norm": 15.0234375, "learning_rate": 9.5007816941444e-06, "loss": 1.1241, "step": 14550 }, { "epoch": 1.5483081506703553, "grad_norm": 13.640625, "learning_rate": 9.500746162592382e-06, "loss": 1.3257, "step": 14551 }, { "epoch": 1.548414556288572, "grad_norm": 10.609375, "learning_rate": 9.500710631040365e-06, "loss": 1.2815, "step": 14552 }, { "epoch": 1.5485209619067888, "grad_norm": 11.953125, "learning_rate": 9.500675099488347e-06, "loss": 1.0175, "step": 14553 }, { "epoch": 1.5486273675250053, "grad_norm": 13.09375, "learning_rate": 9.500639567936328e-06, "loss": 0.8998, "step": 14554 }, { "epoch": 1.5487337731432218, "grad_norm": 9.9375, "learning_rate": 9.50060403638431e-06, "loss": 1.3405, "step": 14555 }, { "epoch": 1.5488401787614388, "grad_norm": 13.125, "learning_rate": 9.500568504832291e-06, "loss": 1.0898, "step": 14556 }, { "epoch": 1.5489465843796553, "grad_norm": 13.9921875, "learning_rate": 9.500532973280273e-06, "loss": 1.3018, "step": 14557 }, { "epoch": 1.5490529899978718, "grad_norm": 11.828125, "learning_rate": 9.500497441728256e-06, "loss": 0.9834, "step": 14558 }, { "epoch": 1.5491593956160885, "grad_norm": 10.109375, "learning_rate": 9.500461910176238e-06, "loss": 1.163, "step": 14559 }, { "epoch": 1.5492658012343052, "grad_norm": 13.5546875, "learning_rate": 9.50042637862422e-06, "loss": 1.2596, "step": 14560 }, { "epoch": 1.5493722068525217, "grad_norm": 12.21875, "learning_rate": 9.500390847072201e-06, "loss": 1.3037, "step": 14561 }, { "epoch": 1.5494786124707385, "grad_norm": 11.9296875, "learning_rate": 9.500355315520183e-06, "loss": 1.096, "step": 14562 }, { "epoch": 1.5495850180889552, "grad_norm": 10.1796875, "learning_rate": 9.500319783968164e-06, "loss": 1.0255, "step": 14563 }, { "epoch": 1.5496914237071717, "grad_norm": 14.5859375, "learning_rate": 9.500284252416146e-06, "loss": 0.9096, "step": 14564 }, { "epoch": 1.5497978293253882, "grad_norm": 11.7734375, "learning_rate": 9.500248720864128e-06, "loss": 1.3483, "step": 14565 }, { "epoch": 1.5499042349436052, "grad_norm": 11.421875, "learning_rate": 9.50021318931211e-06, "loss": 1.0027, "step": 14566 }, { "epoch": 1.5500106405618217, "grad_norm": 8.890625, "learning_rate": 9.50017765776009e-06, "loss": 1.2327, "step": 14567 }, { "epoch": 1.5501170461800382, "grad_norm": 10.0078125, "learning_rate": 9.500142126208074e-06, "loss": 1.0589, "step": 14568 }, { "epoch": 1.550223451798255, "grad_norm": 10.71875, "learning_rate": 9.500106594656056e-06, "loss": 1.3338, "step": 14569 }, { "epoch": 1.5503298574164717, "grad_norm": 12.1953125, "learning_rate": 9.500071063104037e-06, "loss": 0.7894, "step": 14570 }, { "epoch": 1.5504362630346882, "grad_norm": 13.5546875, "learning_rate": 9.500035531552019e-06, "loss": 0.9689, "step": 14571 }, { "epoch": 1.550542668652905, "grad_norm": 16.671875, "learning_rate": 9.5e-06, "loss": 1.4008, "step": 14572 }, { "epoch": 1.5506490742711216, "grad_norm": 15.796875, "learning_rate": 9.499964468447984e-06, "loss": 1.2341, "step": 14573 }, { "epoch": 1.5507554798893382, "grad_norm": 8.7421875, "learning_rate": 9.499928936895965e-06, "loss": 1.1436, "step": 14574 }, { "epoch": 1.5508618855075547, "grad_norm": 10.375, "learning_rate": 9.499893405343947e-06, "loss": 0.9725, "step": 14575 }, { "epoch": 1.5509682911257714, "grad_norm": 8.96875, "learning_rate": 9.499857873791929e-06, "loss": 1.1963, "step": 14576 }, { "epoch": 1.5510746967439881, "grad_norm": 15.5546875, "learning_rate": 9.499822342239909e-06, "loss": 0.9387, "step": 14577 }, { "epoch": 1.5511811023622046, "grad_norm": 13.1796875, "learning_rate": 9.499786810687892e-06, "loss": 1.1687, "step": 14578 }, { "epoch": 1.5512875079804214, "grad_norm": 16.671875, "learning_rate": 9.499751279135873e-06, "loss": 0.917, "step": 14579 }, { "epoch": 1.551393913598638, "grad_norm": 11.578125, "learning_rate": 9.499715747583855e-06, "loss": 1.1263, "step": 14580 }, { "epoch": 1.5515003192168546, "grad_norm": 10.4921875, "learning_rate": 9.499680216031837e-06, "loss": 1.3733, "step": 14581 }, { "epoch": 1.5516067248350713, "grad_norm": 15.2890625, "learning_rate": 9.499644684479818e-06, "loss": 1.1101, "step": 14582 }, { "epoch": 1.551713130453288, "grad_norm": 20.53125, "learning_rate": 9.499609152927802e-06, "loss": 1.2004, "step": 14583 }, { "epoch": 1.5518195360715046, "grad_norm": 8.6796875, "learning_rate": 9.499573621375783e-06, "loss": 0.8708, "step": 14584 }, { "epoch": 1.551925941689721, "grad_norm": 12.59375, "learning_rate": 9.499538089823765e-06, "loss": 0.8681, "step": 14585 }, { "epoch": 1.5520323473079378, "grad_norm": 12.8125, "learning_rate": 9.499502558271746e-06, "loss": 0.8446, "step": 14586 }, { "epoch": 1.5521387529261546, "grad_norm": 10.1875, "learning_rate": 9.499467026719728e-06, "loss": 0.8233, "step": 14587 }, { "epoch": 1.552245158544371, "grad_norm": 15.1640625, "learning_rate": 9.49943149516771e-06, "loss": 1.0954, "step": 14588 }, { "epoch": 1.5523515641625878, "grad_norm": 12.9921875, "learning_rate": 9.499395963615691e-06, "loss": 1.0448, "step": 14589 }, { "epoch": 1.5524579697808045, "grad_norm": 14.8984375, "learning_rate": 9.499360432063673e-06, "loss": 1.2456, "step": 14590 }, { "epoch": 1.552564375399021, "grad_norm": 12.671875, "learning_rate": 9.499324900511654e-06, "loss": 1.3984, "step": 14591 }, { "epoch": 1.5526707810172375, "grad_norm": 16.828125, "learning_rate": 9.499289368959636e-06, "loss": 1.1971, "step": 14592 }, { "epoch": 1.5527771866354545, "grad_norm": 11.8828125, "learning_rate": 9.499253837407618e-06, "loss": 1.1884, "step": 14593 }, { "epoch": 1.552883592253671, "grad_norm": 18.0625, "learning_rate": 9.499218305855601e-06, "loss": 1.1209, "step": 14594 }, { "epoch": 1.5529899978718875, "grad_norm": 15.4375, "learning_rate": 9.499182774303583e-06, "loss": 1.1921, "step": 14595 }, { "epoch": 1.5530964034901042, "grad_norm": 17.859375, "learning_rate": 9.499147242751564e-06, "loss": 1.7462, "step": 14596 }, { "epoch": 1.553202809108321, "grad_norm": 12.546875, "learning_rate": 9.499111711199546e-06, "loss": 0.8582, "step": 14597 }, { "epoch": 1.5533092147265375, "grad_norm": 15.2890625, "learning_rate": 9.499076179647527e-06, "loss": 1.3994, "step": 14598 }, { "epoch": 1.5534156203447542, "grad_norm": 10.8203125, "learning_rate": 9.49904064809551e-06, "loss": 1.1519, "step": 14599 }, { "epoch": 1.553522025962971, "grad_norm": 17.640625, "learning_rate": 9.499005116543492e-06, "loss": 1.0058, "step": 14600 }, { "epoch": 1.5536284315811875, "grad_norm": 16.515625, "learning_rate": 9.498969584991474e-06, "loss": 1.1033, "step": 14601 }, { "epoch": 1.553734837199404, "grad_norm": 10.5234375, "learning_rate": 9.498934053439455e-06, "loss": 1.1237, "step": 14602 }, { "epoch": 1.553841242817621, "grad_norm": 10.234375, "learning_rate": 9.498898521887437e-06, "loss": 1.178, "step": 14603 }, { "epoch": 1.5539476484358374, "grad_norm": 12.3125, "learning_rate": 9.498862990335419e-06, "loss": 1.1928, "step": 14604 }, { "epoch": 1.554054054054054, "grad_norm": 14.7578125, "learning_rate": 9.4988274587834e-06, "loss": 1.1684, "step": 14605 }, { "epoch": 1.5541604596722707, "grad_norm": 9.3515625, "learning_rate": 9.498791927231382e-06, "loss": 0.9721, "step": 14606 }, { "epoch": 1.5542668652904874, "grad_norm": 15.609375, "learning_rate": 9.498756395679364e-06, "loss": 1.3583, "step": 14607 }, { "epoch": 1.554373270908704, "grad_norm": 12.25, "learning_rate": 9.498720864127345e-06, "loss": 1.2618, "step": 14608 }, { "epoch": 1.5544796765269207, "grad_norm": 12.984375, "learning_rate": 9.498685332575328e-06, "loss": 1.0111, "step": 14609 }, { "epoch": 1.5545860821451374, "grad_norm": 10.140625, "learning_rate": 9.49864980102331e-06, "loss": 0.9906, "step": 14610 }, { "epoch": 1.554692487763354, "grad_norm": 9.515625, "learning_rate": 9.498614269471292e-06, "loss": 1.172, "step": 14611 }, { "epoch": 1.5547988933815704, "grad_norm": 10.21875, "learning_rate": 9.498578737919273e-06, "loss": 0.8676, "step": 14612 }, { "epoch": 1.5549052989997874, "grad_norm": 10.6015625, "learning_rate": 9.498543206367255e-06, "loss": 1.0387, "step": 14613 }, { "epoch": 1.5550117046180039, "grad_norm": 12.484375, "learning_rate": 9.498507674815236e-06, "loss": 1.0708, "step": 14614 }, { "epoch": 1.5551181102362204, "grad_norm": 17.46875, "learning_rate": 9.498472143263218e-06, "loss": 1.0935, "step": 14615 }, { "epoch": 1.555224515854437, "grad_norm": 16.3125, "learning_rate": 9.4984366117112e-06, "loss": 1.3228, "step": 14616 }, { "epoch": 1.5553309214726538, "grad_norm": 12.640625, "learning_rate": 9.498401080159181e-06, "loss": 1.0447, "step": 14617 }, { "epoch": 1.5554373270908703, "grad_norm": 12.4609375, "learning_rate": 9.498365548607163e-06, "loss": 0.9739, "step": 14618 }, { "epoch": 1.555543732709087, "grad_norm": 13.875, "learning_rate": 9.498330017055146e-06, "loss": 1.3233, "step": 14619 }, { "epoch": 1.5556501383273038, "grad_norm": 10.984375, "learning_rate": 9.498294485503128e-06, "loss": 1.109, "step": 14620 }, { "epoch": 1.5557565439455203, "grad_norm": 18.90625, "learning_rate": 9.49825895395111e-06, "loss": 1.1511, "step": 14621 }, { "epoch": 1.5558629495637368, "grad_norm": 11.9921875, "learning_rate": 9.498223422399091e-06, "loss": 1.3005, "step": 14622 }, { "epoch": 1.5559693551819536, "grad_norm": 10.0546875, "learning_rate": 9.498187890847073e-06, "loss": 1.0164, "step": 14623 }, { "epoch": 1.5560757608001703, "grad_norm": 11.875, "learning_rate": 9.498152359295056e-06, "loss": 1.2772, "step": 14624 }, { "epoch": 1.5561821664183868, "grad_norm": 9.4453125, "learning_rate": 9.498116827743038e-06, "loss": 1.011, "step": 14625 }, { "epoch": 1.5562885720366035, "grad_norm": 12.4921875, "learning_rate": 9.498081296191019e-06, "loss": 1.221, "step": 14626 }, { "epoch": 1.5563949776548203, "grad_norm": 12.25, "learning_rate": 9.498045764639e-06, "loss": 1.1336, "step": 14627 }, { "epoch": 1.5565013832730368, "grad_norm": 10.3515625, "learning_rate": 9.498010233086982e-06, "loss": 1.0545, "step": 14628 }, { "epoch": 1.5566077888912535, "grad_norm": 13.796875, "learning_rate": 9.497974701534964e-06, "loss": 1.1069, "step": 14629 }, { "epoch": 1.5567141945094702, "grad_norm": 13.4921875, "learning_rate": 9.497939169982946e-06, "loss": 1.3327, "step": 14630 }, { "epoch": 1.5568206001276867, "grad_norm": 17.890625, "learning_rate": 9.497903638430927e-06, "loss": 1.2357, "step": 14631 }, { "epoch": 1.5569270057459033, "grad_norm": 9.0, "learning_rate": 9.497868106878909e-06, "loss": 0.963, "step": 14632 }, { "epoch": 1.55703341136412, "grad_norm": 10.9453125, "learning_rate": 9.49783257532689e-06, "loss": 1.1061, "step": 14633 }, { "epoch": 1.5571398169823367, "grad_norm": 11.8828125, "learning_rate": 9.497797043774872e-06, "loss": 0.8569, "step": 14634 }, { "epoch": 1.5572462226005532, "grad_norm": 12.875, "learning_rate": 9.497761512222855e-06, "loss": 0.8344, "step": 14635 }, { "epoch": 1.55735262821877, "grad_norm": 16.9375, "learning_rate": 9.497725980670837e-06, "loss": 1.331, "step": 14636 }, { "epoch": 1.5574590338369867, "grad_norm": 11.28125, "learning_rate": 9.497690449118819e-06, "loss": 1.1167, "step": 14637 }, { "epoch": 1.5575654394552032, "grad_norm": 11.671875, "learning_rate": 9.4976549175668e-06, "loss": 0.8963, "step": 14638 }, { "epoch": 1.5576718450734197, "grad_norm": 12.3671875, "learning_rate": 9.497619386014782e-06, "loss": 1.3233, "step": 14639 }, { "epoch": 1.5577782506916367, "grad_norm": 13.40625, "learning_rate": 9.497583854462763e-06, "loss": 0.8904, "step": 14640 }, { "epoch": 1.5578846563098532, "grad_norm": 13.546875, "learning_rate": 9.497548322910745e-06, "loss": 0.8738, "step": 14641 }, { "epoch": 1.5579910619280697, "grad_norm": 12.8984375, "learning_rate": 9.497512791358727e-06, "loss": 0.9579, "step": 14642 }, { "epoch": 1.5580974675462864, "grad_norm": 14.640625, "learning_rate": 9.497477259806708e-06, "loss": 1.2466, "step": 14643 }, { "epoch": 1.5582038731645032, "grad_norm": 15.3046875, "learning_rate": 9.49744172825469e-06, "loss": 1.2163, "step": 14644 }, { "epoch": 1.5583102787827197, "grad_norm": 10.2109375, "learning_rate": 9.497406196702673e-06, "loss": 1.2981, "step": 14645 }, { "epoch": 1.5584166844009364, "grad_norm": 12.0546875, "learning_rate": 9.497370665150655e-06, "loss": 1.1749, "step": 14646 }, { "epoch": 1.5585230900191531, "grad_norm": 11.6875, "learning_rate": 9.497335133598636e-06, "loss": 0.9362, "step": 14647 }, { "epoch": 1.5586294956373696, "grad_norm": 11.90625, "learning_rate": 9.497299602046618e-06, "loss": 1.1387, "step": 14648 }, { "epoch": 1.5587359012555861, "grad_norm": 11.8515625, "learning_rate": 9.4972640704946e-06, "loss": 0.9687, "step": 14649 }, { "epoch": 1.558842306873803, "grad_norm": 14.2578125, "learning_rate": 9.497228538942583e-06, "loss": 1.0094, "step": 14650 }, { "epoch": 1.5589487124920196, "grad_norm": 13.75, "learning_rate": 9.497193007390564e-06, "loss": 1.4069, "step": 14651 }, { "epoch": 1.5590551181102361, "grad_norm": 12.9609375, "learning_rate": 9.497157475838546e-06, "loss": 1.105, "step": 14652 }, { "epoch": 1.5591615237284528, "grad_norm": 18.921875, "learning_rate": 9.497121944286528e-06, "loss": 1.1044, "step": 14653 }, { "epoch": 1.5592679293466696, "grad_norm": 14.515625, "learning_rate": 9.49708641273451e-06, "loss": 1.1972, "step": 14654 }, { "epoch": 1.559374334964886, "grad_norm": 12.296875, "learning_rate": 9.49705088118249e-06, "loss": 1.1049, "step": 14655 }, { "epoch": 1.5594807405831028, "grad_norm": 15.09375, "learning_rate": 9.497015349630472e-06, "loss": 1.4712, "step": 14656 }, { "epoch": 1.5595871462013196, "grad_norm": 13.6484375, "learning_rate": 9.496979818078454e-06, "loss": 0.9409, "step": 14657 }, { "epoch": 1.559693551819536, "grad_norm": 12.4375, "learning_rate": 9.496944286526436e-06, "loss": 1.0282, "step": 14658 }, { "epoch": 1.5597999574377526, "grad_norm": 10.3046875, "learning_rate": 9.496908754974417e-06, "loss": 1.0863, "step": 14659 }, { "epoch": 1.5599063630559693, "grad_norm": 13.7265625, "learning_rate": 9.4968732234224e-06, "loss": 1.1575, "step": 14660 }, { "epoch": 1.560012768674186, "grad_norm": 9.21875, "learning_rate": 9.496837691870382e-06, "loss": 0.9031, "step": 14661 }, { "epoch": 1.5601191742924025, "grad_norm": 15.28125, "learning_rate": 9.496802160318364e-06, "loss": 1.034, "step": 14662 }, { "epoch": 1.5602255799106193, "grad_norm": 11.6328125, "learning_rate": 9.496766628766345e-06, "loss": 1.0314, "step": 14663 }, { "epoch": 1.560331985528836, "grad_norm": 14.5390625, "learning_rate": 9.496731097214327e-06, "loss": 1.1946, "step": 14664 }, { "epoch": 1.5604383911470525, "grad_norm": 11.6796875, "learning_rate": 9.49669556566231e-06, "loss": 1.2714, "step": 14665 }, { "epoch": 1.5605447967652692, "grad_norm": 12.140625, "learning_rate": 9.49666003411029e-06, "loss": 1.1002, "step": 14666 }, { "epoch": 1.560651202383486, "grad_norm": 14.1640625, "learning_rate": 9.496624502558272e-06, "loss": 1.1483, "step": 14667 }, { "epoch": 1.5607576080017025, "grad_norm": 14.5390625, "learning_rate": 9.496588971006253e-06, "loss": 1.0693, "step": 14668 }, { "epoch": 1.560864013619919, "grad_norm": 10.7421875, "learning_rate": 9.496553439454235e-06, "loss": 1.3883, "step": 14669 }, { "epoch": 1.5609704192381357, "grad_norm": 10.984375, "learning_rate": 9.496517907902217e-06, "loss": 1.2516, "step": 14670 }, { "epoch": 1.5610768248563525, "grad_norm": 12.609375, "learning_rate": 9.4964823763502e-06, "loss": 0.9719, "step": 14671 }, { "epoch": 1.561183230474569, "grad_norm": 11.6484375, "learning_rate": 9.496446844798182e-06, "loss": 1.2363, "step": 14672 }, { "epoch": 1.5612896360927857, "grad_norm": 12.8984375, "learning_rate": 9.496411313246163e-06, "loss": 0.9508, "step": 14673 }, { "epoch": 1.5613960417110024, "grad_norm": 18.765625, "learning_rate": 9.496375781694145e-06, "loss": 1.204, "step": 14674 }, { "epoch": 1.561502447329219, "grad_norm": 23.078125, "learning_rate": 9.496340250142126e-06, "loss": 1.2404, "step": 14675 }, { "epoch": 1.5616088529474357, "grad_norm": 10.8046875, "learning_rate": 9.49630471859011e-06, "loss": 1.0078, "step": 14676 }, { "epoch": 1.5617152585656524, "grad_norm": 17.46875, "learning_rate": 9.496269187038091e-06, "loss": 1.4304, "step": 14677 }, { "epoch": 1.561821664183869, "grad_norm": 11.2734375, "learning_rate": 9.496233655486073e-06, "loss": 1.2379, "step": 14678 }, { "epoch": 1.5619280698020854, "grad_norm": 16.703125, "learning_rate": 9.496198123934054e-06, "loss": 1.0945, "step": 14679 }, { "epoch": 1.5620344754203022, "grad_norm": 17.734375, "learning_rate": 9.496162592382036e-06, "loss": 1.0933, "step": 14680 }, { "epoch": 1.562140881038519, "grad_norm": 11.203125, "learning_rate": 9.496127060830018e-06, "loss": 1.1499, "step": 14681 }, { "epoch": 1.5622472866567354, "grad_norm": 10.4765625, "learning_rate": 9.496091529278e-06, "loss": 1.0901, "step": 14682 }, { "epoch": 1.5623536922749521, "grad_norm": 13.8671875, "learning_rate": 9.496055997725981e-06, "loss": 1.1172, "step": 14683 }, { "epoch": 1.5624600978931689, "grad_norm": 10.8515625, "learning_rate": 9.496020466173962e-06, "loss": 0.9878, "step": 14684 }, { "epoch": 1.5625665035113854, "grad_norm": 8.8515625, "learning_rate": 9.495984934621944e-06, "loss": 1.1775, "step": 14685 }, { "epoch": 1.5626729091296019, "grad_norm": 13.71875, "learning_rate": 9.495949403069927e-06, "loss": 1.0615, "step": 14686 }, { "epoch": 1.5627793147478188, "grad_norm": 11.9921875, "learning_rate": 9.495913871517909e-06, "loss": 1.0289, "step": 14687 }, { "epoch": 1.5628857203660353, "grad_norm": 12.234375, "learning_rate": 9.49587833996589e-06, "loss": 1.1518, "step": 14688 }, { "epoch": 1.5629921259842519, "grad_norm": 19.71875, "learning_rate": 9.495842808413872e-06, "loss": 1.2092, "step": 14689 }, { "epoch": 1.5630985316024686, "grad_norm": 12.0234375, "learning_rate": 9.495807276861854e-06, "loss": 1.0092, "step": 14690 }, { "epoch": 1.5632049372206853, "grad_norm": 13.5, "learning_rate": 9.495771745309837e-06, "loss": 1.0179, "step": 14691 }, { "epoch": 1.5633113428389018, "grad_norm": 14.609375, "learning_rate": 9.495736213757819e-06, "loss": 1.1126, "step": 14692 }, { "epoch": 1.5634177484571186, "grad_norm": 17.1875, "learning_rate": 9.4957006822058e-06, "loss": 0.8517, "step": 14693 }, { "epoch": 1.5635241540753353, "grad_norm": 12.9453125, "learning_rate": 9.495665150653782e-06, "loss": 1.0271, "step": 14694 }, { "epoch": 1.5636305596935518, "grad_norm": 12.5, "learning_rate": 9.495629619101764e-06, "loss": 1.1659, "step": 14695 }, { "epoch": 1.5637369653117683, "grad_norm": 15.0546875, "learning_rate": 9.495594087549745e-06, "loss": 1.3519, "step": 14696 }, { "epoch": 1.5638433709299853, "grad_norm": 13.3671875, "learning_rate": 9.495558555997727e-06, "loss": 0.9323, "step": 14697 }, { "epoch": 1.5639497765482018, "grad_norm": 18.34375, "learning_rate": 9.495523024445708e-06, "loss": 0.8064, "step": 14698 }, { "epoch": 1.5640561821664183, "grad_norm": 14.5546875, "learning_rate": 9.49548749289369e-06, "loss": 0.974, "step": 14699 }, { "epoch": 1.564162587784635, "grad_norm": 9.828125, "learning_rate": 9.495451961341672e-06, "loss": 1.0657, "step": 14700 }, { "epoch": 1.5642689934028517, "grad_norm": 11.2265625, "learning_rate": 9.495416429789655e-06, "loss": 1.3686, "step": 14701 }, { "epoch": 1.5643753990210683, "grad_norm": 11.859375, "learning_rate": 9.495380898237636e-06, "loss": 1.1338, "step": 14702 }, { "epoch": 1.564481804639285, "grad_norm": 12.828125, "learning_rate": 9.495345366685618e-06, "loss": 1.2017, "step": 14703 }, { "epoch": 1.5645882102575017, "grad_norm": 14.359375, "learning_rate": 9.4953098351336e-06, "loss": 1.1269, "step": 14704 }, { "epoch": 1.5646946158757182, "grad_norm": 10.3671875, "learning_rate": 9.495274303581581e-06, "loss": 1.1146, "step": 14705 }, { "epoch": 1.5648010214939347, "grad_norm": 13.0546875, "learning_rate": 9.495238772029563e-06, "loss": 1.381, "step": 14706 }, { "epoch": 1.5649074271121515, "grad_norm": 26.359375, "learning_rate": 9.495203240477545e-06, "loss": 1.1291, "step": 14707 }, { "epoch": 1.5650138327303682, "grad_norm": 12.6328125, "learning_rate": 9.495167708925526e-06, "loss": 1.1493, "step": 14708 }, { "epoch": 1.5651202383485847, "grad_norm": 13.3984375, "learning_rate": 9.495132177373508e-06, "loss": 0.9358, "step": 14709 }, { "epoch": 1.5652266439668014, "grad_norm": 18.53125, "learning_rate": 9.49509664582149e-06, "loss": 1.3025, "step": 14710 }, { "epoch": 1.5653330495850182, "grad_norm": 9.4765625, "learning_rate": 9.495061114269471e-06, "loss": 1.0425, "step": 14711 }, { "epoch": 1.5654394552032347, "grad_norm": 11.3359375, "learning_rate": 9.495025582717454e-06, "loss": 1.2265, "step": 14712 }, { "epoch": 1.5655458608214514, "grad_norm": 9.265625, "learning_rate": 9.494990051165436e-06, "loss": 1.1523, "step": 14713 }, { "epoch": 1.5656522664396681, "grad_norm": 14.7109375, "learning_rate": 9.494954519613417e-06, "loss": 0.9168, "step": 14714 }, { "epoch": 1.5657586720578847, "grad_norm": 15.3203125, "learning_rate": 9.494918988061399e-06, "loss": 0.8493, "step": 14715 }, { "epoch": 1.5658650776761012, "grad_norm": 9.0390625, "learning_rate": 9.49488345650938e-06, "loss": 0.9194, "step": 14716 }, { "epoch": 1.565971483294318, "grad_norm": 10.859375, "learning_rate": 9.494847924957364e-06, "loss": 1.1625, "step": 14717 }, { "epoch": 1.5660778889125346, "grad_norm": 26.75, "learning_rate": 9.494812393405346e-06, "loss": 1.3886, "step": 14718 }, { "epoch": 1.5661842945307511, "grad_norm": 16.796875, "learning_rate": 9.494776861853327e-06, "loss": 1.0469, "step": 14719 }, { "epoch": 1.5662907001489679, "grad_norm": 13.4296875, "learning_rate": 9.494741330301309e-06, "loss": 1.3914, "step": 14720 }, { "epoch": 1.5663971057671846, "grad_norm": 12.25, "learning_rate": 9.49470579874929e-06, "loss": 1.2562, "step": 14721 }, { "epoch": 1.5665035113854011, "grad_norm": 11.296875, "learning_rate": 9.494670267197272e-06, "loss": 1.067, "step": 14722 }, { "epoch": 1.5666099170036176, "grad_norm": 11.6953125, "learning_rate": 9.494634735645254e-06, "loss": 0.9749, "step": 14723 }, { "epoch": 1.5667163226218346, "grad_norm": 11.8125, "learning_rate": 9.494599204093235e-06, "loss": 0.9034, "step": 14724 }, { "epoch": 1.566822728240051, "grad_norm": 12.984375, "learning_rate": 9.494563672541217e-06, "loss": 1.1159, "step": 14725 }, { "epoch": 1.5669291338582676, "grad_norm": 11.4765625, "learning_rate": 9.494528140989198e-06, "loss": 0.9709, "step": 14726 }, { "epoch": 1.5670355394764843, "grad_norm": 16.421875, "learning_rate": 9.494492609437182e-06, "loss": 0.8469, "step": 14727 }, { "epoch": 1.567141945094701, "grad_norm": 12.109375, "learning_rate": 9.494457077885163e-06, "loss": 1.1077, "step": 14728 }, { "epoch": 1.5672483507129176, "grad_norm": 8.828125, "learning_rate": 9.494421546333145e-06, "loss": 1.0966, "step": 14729 }, { "epoch": 1.5673547563311343, "grad_norm": 9.2734375, "learning_rate": 9.494386014781127e-06, "loss": 1.1036, "step": 14730 }, { "epoch": 1.567461161949351, "grad_norm": 14.015625, "learning_rate": 9.494350483229108e-06, "loss": 1.1424, "step": 14731 }, { "epoch": 1.5675675675675675, "grad_norm": 14.84375, "learning_rate": 9.49431495167709e-06, "loss": 1.1691, "step": 14732 }, { "epoch": 1.567673973185784, "grad_norm": 12.3828125, "learning_rate": 9.494279420125071e-06, "loss": 1.0848, "step": 14733 }, { "epoch": 1.567780378804001, "grad_norm": 11.765625, "learning_rate": 9.494243888573053e-06, "loss": 0.9509, "step": 14734 }, { "epoch": 1.5678867844222175, "grad_norm": 12.4921875, "learning_rate": 9.494208357021035e-06, "loss": 0.8501, "step": 14735 }, { "epoch": 1.567993190040434, "grad_norm": 13.6171875, "learning_rate": 9.494172825469016e-06, "loss": 1.0797, "step": 14736 }, { "epoch": 1.5680995956586508, "grad_norm": 15.59375, "learning_rate": 9.494137293917e-06, "loss": 1.1656, "step": 14737 }, { "epoch": 1.5682060012768675, "grad_norm": 14.671875, "learning_rate": 9.494101762364981e-06, "loss": 1.5311, "step": 14738 }, { "epoch": 1.568312406895084, "grad_norm": 13.15625, "learning_rate": 9.494066230812963e-06, "loss": 0.8287, "step": 14739 }, { "epoch": 1.5684188125133007, "grad_norm": 18.578125, "learning_rate": 9.494030699260944e-06, "loss": 1.2652, "step": 14740 }, { "epoch": 1.5685252181315175, "grad_norm": 10.2734375, "learning_rate": 9.493995167708926e-06, "loss": 1.1393, "step": 14741 }, { "epoch": 1.568631623749734, "grad_norm": 10.8515625, "learning_rate": 9.49395963615691e-06, "loss": 1.2444, "step": 14742 }, { "epoch": 1.5687380293679505, "grad_norm": 21.59375, "learning_rate": 9.49392410460489e-06, "loss": 0.9722, "step": 14743 }, { "epoch": 1.5688444349861674, "grad_norm": 16.984375, "learning_rate": 9.493888573052872e-06, "loss": 1.1488, "step": 14744 }, { "epoch": 1.568950840604384, "grad_norm": 10.1953125, "learning_rate": 9.493853041500854e-06, "loss": 1.1064, "step": 14745 }, { "epoch": 1.5690572462226005, "grad_norm": 12.5625, "learning_rate": 9.493817509948836e-06, "loss": 1.0385, "step": 14746 }, { "epoch": 1.5691636518408172, "grad_norm": 11.2109375, "learning_rate": 9.493781978396817e-06, "loss": 0.8531, "step": 14747 }, { "epoch": 1.569270057459034, "grad_norm": 14.28125, "learning_rate": 9.493746446844799e-06, "loss": 1.1726, "step": 14748 }, { "epoch": 1.5693764630772504, "grad_norm": 11.765625, "learning_rate": 9.49371091529278e-06, "loss": 1.1259, "step": 14749 }, { "epoch": 1.5694828686954672, "grad_norm": 11.4453125, "learning_rate": 9.493675383740762e-06, "loss": 1.0919, "step": 14750 }, { "epoch": 1.569589274313684, "grad_norm": 13.0703125, "learning_rate": 9.493639852188744e-06, "loss": 0.8694, "step": 14751 }, { "epoch": 1.5696956799319004, "grad_norm": 10.2890625, "learning_rate": 9.493604320636725e-06, "loss": 1.1788, "step": 14752 }, { "epoch": 1.569802085550117, "grad_norm": 11.125, "learning_rate": 9.493568789084709e-06, "loss": 1.1871, "step": 14753 }, { "epoch": 1.5699084911683336, "grad_norm": 13.125, "learning_rate": 9.49353325753269e-06, "loss": 1.0407, "step": 14754 }, { "epoch": 1.5700148967865504, "grad_norm": 14.0859375, "learning_rate": 9.493497725980672e-06, "loss": 1.1405, "step": 14755 }, { "epoch": 1.5701213024047669, "grad_norm": 12.5859375, "learning_rate": 9.493462194428653e-06, "loss": 1.1332, "step": 14756 }, { "epoch": 1.5702277080229836, "grad_norm": 13.46875, "learning_rate": 9.493426662876635e-06, "loss": 1.0423, "step": 14757 }, { "epoch": 1.5703341136412003, "grad_norm": 11.75, "learning_rate": 9.493391131324617e-06, "loss": 0.8318, "step": 14758 }, { "epoch": 1.5704405192594169, "grad_norm": 9.6953125, "learning_rate": 9.493355599772598e-06, "loss": 0.8529, "step": 14759 }, { "epoch": 1.5705469248776336, "grad_norm": 12.8359375, "learning_rate": 9.49332006822058e-06, "loss": 1.1062, "step": 14760 }, { "epoch": 1.5706533304958503, "grad_norm": 11.9296875, "learning_rate": 9.493284536668561e-06, "loss": 1.3145, "step": 14761 }, { "epoch": 1.5707597361140668, "grad_norm": 14.421875, "learning_rate": 9.493249005116543e-06, "loss": 1.1161, "step": 14762 }, { "epoch": 1.5708661417322833, "grad_norm": 8.875, "learning_rate": 9.493213473564526e-06, "loss": 0.9717, "step": 14763 }, { "epoch": 1.5709725473505, "grad_norm": 15.5234375, "learning_rate": 9.493177942012508e-06, "loss": 0.9268, "step": 14764 }, { "epoch": 1.5710789529687168, "grad_norm": 9.296875, "learning_rate": 9.49314241046049e-06, "loss": 0.9582, "step": 14765 }, { "epoch": 1.5711853585869333, "grad_norm": 14.0703125, "learning_rate": 9.493106878908471e-06, "loss": 1.0006, "step": 14766 }, { "epoch": 1.57129176420515, "grad_norm": 9.953125, "learning_rate": 9.493071347356453e-06, "loss": 0.9632, "step": 14767 }, { "epoch": 1.5713981698233668, "grad_norm": 12.421875, "learning_rate": 9.493035815804436e-06, "loss": 0.9014, "step": 14768 }, { "epoch": 1.5715045754415833, "grad_norm": 14.453125, "learning_rate": 9.493000284252418e-06, "loss": 0.888, "step": 14769 }, { "epoch": 1.5716109810597998, "grad_norm": 10.9765625, "learning_rate": 9.4929647527004e-06, "loss": 1.0564, "step": 14770 }, { "epoch": 1.5717173866780167, "grad_norm": 9.6484375, "learning_rate": 9.492929221148381e-06, "loss": 1.026, "step": 14771 }, { "epoch": 1.5718237922962333, "grad_norm": 12.296875, "learning_rate": 9.492893689596363e-06, "loss": 0.9903, "step": 14772 }, { "epoch": 1.5719301979144498, "grad_norm": 11.6328125, "learning_rate": 9.492858158044344e-06, "loss": 1.0712, "step": 14773 }, { "epoch": 1.5720366035326665, "grad_norm": 12.1953125, "learning_rate": 9.492822626492326e-06, "loss": 1.2786, "step": 14774 }, { "epoch": 1.5721430091508832, "grad_norm": 15.5, "learning_rate": 9.492787094940307e-06, "loss": 1.21, "step": 14775 }, { "epoch": 1.5722494147690997, "grad_norm": 14.2421875, "learning_rate": 9.492751563388289e-06, "loss": 1.3019, "step": 14776 }, { "epoch": 1.5723558203873165, "grad_norm": 16.78125, "learning_rate": 9.49271603183627e-06, "loss": 1.0405, "step": 14777 }, { "epoch": 1.5724622260055332, "grad_norm": 17.40625, "learning_rate": 9.492680500284254e-06, "loss": 1.3762, "step": 14778 }, { "epoch": 1.5725686316237497, "grad_norm": 11.5625, "learning_rate": 9.492644968732235e-06, "loss": 0.963, "step": 14779 }, { "epoch": 1.5726750372419662, "grad_norm": 17.578125, "learning_rate": 9.492609437180217e-06, "loss": 1.0714, "step": 14780 }, { "epoch": 1.5727814428601832, "grad_norm": 15.0546875, "learning_rate": 9.492573905628199e-06, "loss": 0.9132, "step": 14781 }, { "epoch": 1.5728878484783997, "grad_norm": 12.421875, "learning_rate": 9.49253837407618e-06, "loss": 1.0703, "step": 14782 }, { "epoch": 1.5729942540966162, "grad_norm": 17.046875, "learning_rate": 9.492502842524164e-06, "loss": 1.3795, "step": 14783 }, { "epoch": 1.573100659714833, "grad_norm": 9.9609375, "learning_rate": 9.492467310972145e-06, "loss": 0.8308, "step": 14784 }, { "epoch": 1.5732070653330497, "grad_norm": 13.15625, "learning_rate": 9.492431779420127e-06, "loss": 1.4678, "step": 14785 }, { "epoch": 1.5733134709512662, "grad_norm": 11.03125, "learning_rate": 9.492396247868107e-06, "loss": 0.9642, "step": 14786 }, { "epoch": 1.573419876569483, "grad_norm": 13.8828125, "learning_rate": 9.492360716316088e-06, "loss": 1.241, "step": 14787 }, { "epoch": 1.5735262821876996, "grad_norm": 17.6875, "learning_rate": 9.49232518476407e-06, "loss": 1.048, "step": 14788 }, { "epoch": 1.5736326878059161, "grad_norm": 17.125, "learning_rate": 9.492289653212053e-06, "loss": 1.1444, "step": 14789 }, { "epoch": 1.5737390934241327, "grad_norm": 13.015625, "learning_rate": 9.492254121660035e-06, "loss": 1.0964, "step": 14790 }, { "epoch": 1.5738454990423494, "grad_norm": 12.953125, "learning_rate": 9.492218590108016e-06, "loss": 1.1198, "step": 14791 }, { "epoch": 1.5739519046605661, "grad_norm": 16.875, "learning_rate": 9.492183058555998e-06, "loss": 0.9665, "step": 14792 }, { "epoch": 1.5740583102787826, "grad_norm": 19.546875, "learning_rate": 9.49214752700398e-06, "loss": 0.9324, "step": 14793 }, { "epoch": 1.5741647158969994, "grad_norm": 15.0546875, "learning_rate": 9.492111995451963e-06, "loss": 0.7337, "step": 14794 }, { "epoch": 1.574271121515216, "grad_norm": 12.6796875, "learning_rate": 9.492076463899945e-06, "loss": 0.8794, "step": 14795 }, { "epoch": 1.5743775271334326, "grad_norm": 12.5234375, "learning_rate": 9.492040932347926e-06, "loss": 1.1443, "step": 14796 }, { "epoch": 1.5744839327516493, "grad_norm": 10.140625, "learning_rate": 9.492005400795908e-06, "loss": 0.9746, "step": 14797 }, { "epoch": 1.574590338369866, "grad_norm": 13.125, "learning_rate": 9.49196986924389e-06, "loss": 1.1592, "step": 14798 }, { "epoch": 1.5746967439880826, "grad_norm": 14.8828125, "learning_rate": 9.491934337691871e-06, "loss": 1.2681, "step": 14799 }, { "epoch": 1.574803149606299, "grad_norm": 12.7421875, "learning_rate": 9.491898806139853e-06, "loss": 1.442, "step": 14800 }, { "epoch": 1.5749095552245158, "grad_norm": 17.125, "learning_rate": 9.491863274587834e-06, "loss": 0.9432, "step": 14801 }, { "epoch": 1.5750159608427325, "grad_norm": 14.1171875, "learning_rate": 9.491827743035816e-06, "loss": 0.9226, "step": 14802 }, { "epoch": 1.575122366460949, "grad_norm": 12.046875, "learning_rate": 9.491792211483797e-06, "loss": 0.743, "step": 14803 }, { "epoch": 1.5752287720791658, "grad_norm": 11.328125, "learning_rate": 9.49175667993178e-06, "loss": 1.035, "step": 14804 }, { "epoch": 1.5753351776973825, "grad_norm": 15.3515625, "learning_rate": 9.491721148379762e-06, "loss": 0.9862, "step": 14805 }, { "epoch": 1.575441583315599, "grad_norm": 9.3359375, "learning_rate": 9.491685616827744e-06, "loss": 1.0662, "step": 14806 }, { "epoch": 1.5755479889338158, "grad_norm": 9.921875, "learning_rate": 9.491650085275726e-06, "loss": 0.8429, "step": 14807 }, { "epoch": 1.5756543945520325, "grad_norm": 23.484375, "learning_rate": 9.491614553723707e-06, "loss": 1.189, "step": 14808 }, { "epoch": 1.575760800170249, "grad_norm": 13.1015625, "learning_rate": 9.49157902217169e-06, "loss": 0.9819, "step": 14809 }, { "epoch": 1.5758672057884655, "grad_norm": 13.671875, "learning_rate": 9.491543490619672e-06, "loss": 1.1271, "step": 14810 }, { "epoch": 1.5759736114066822, "grad_norm": 13.8984375, "learning_rate": 9.491507959067654e-06, "loss": 1.019, "step": 14811 }, { "epoch": 1.576080017024899, "grad_norm": 12.890625, "learning_rate": 9.491472427515635e-06, "loss": 1.134, "step": 14812 }, { "epoch": 1.5761864226431155, "grad_norm": 9.765625, "learning_rate": 9.491436895963617e-06, "loss": 0.8735, "step": 14813 }, { "epoch": 1.5762928282613322, "grad_norm": 13.8359375, "learning_rate": 9.491401364411598e-06, "loss": 1.0321, "step": 14814 }, { "epoch": 1.576399233879549, "grad_norm": 10.421875, "learning_rate": 9.49136583285958e-06, "loss": 0.9479, "step": 14815 }, { "epoch": 1.5765056394977655, "grad_norm": 16.34375, "learning_rate": 9.491330301307562e-06, "loss": 1.0523, "step": 14816 }, { "epoch": 1.576612045115982, "grad_norm": 23.171875, "learning_rate": 9.491294769755543e-06, "loss": 1.5213, "step": 14817 }, { "epoch": 1.576718450734199, "grad_norm": 10.0546875, "learning_rate": 9.491259238203525e-06, "loss": 0.9686, "step": 14818 }, { "epoch": 1.5768248563524154, "grad_norm": 16.796875, "learning_rate": 9.491223706651508e-06, "loss": 0.9156, "step": 14819 }, { "epoch": 1.576931261970632, "grad_norm": 13.9453125, "learning_rate": 9.49118817509949e-06, "loss": 1.1115, "step": 14820 }, { "epoch": 1.5770376675888487, "grad_norm": 10.453125, "learning_rate": 9.491152643547471e-06, "loss": 1.1353, "step": 14821 }, { "epoch": 1.5771440732070654, "grad_norm": 14.84375, "learning_rate": 9.491117111995453e-06, "loss": 0.9645, "step": 14822 }, { "epoch": 1.577250478825282, "grad_norm": 12.875, "learning_rate": 9.491081580443435e-06, "loss": 0.9862, "step": 14823 }, { "epoch": 1.5773568844434986, "grad_norm": 13.4296875, "learning_rate": 9.491046048891416e-06, "loss": 1.1964, "step": 14824 }, { "epoch": 1.5774632900617154, "grad_norm": 14.34375, "learning_rate": 9.491010517339398e-06, "loss": 1.1491, "step": 14825 }, { "epoch": 1.5775696956799319, "grad_norm": 13.8125, "learning_rate": 9.49097498578738e-06, "loss": 0.9306, "step": 14826 }, { "epoch": 1.5776761012981484, "grad_norm": 11.9140625, "learning_rate": 9.490939454235361e-06, "loss": 1.205, "step": 14827 }, { "epoch": 1.5777825069163653, "grad_norm": 18.78125, "learning_rate": 9.490903922683343e-06, "loss": 1.1013, "step": 14828 }, { "epoch": 1.5778889125345819, "grad_norm": 12.890625, "learning_rate": 9.490868391131324e-06, "loss": 0.9559, "step": 14829 }, { "epoch": 1.5779953181527984, "grad_norm": 14.1640625, "learning_rate": 9.490832859579308e-06, "loss": 1.195, "step": 14830 }, { "epoch": 1.578101723771015, "grad_norm": 12.640625, "learning_rate": 9.49079732802729e-06, "loss": 1.1952, "step": 14831 }, { "epoch": 1.5782081293892318, "grad_norm": 18.359375, "learning_rate": 9.49076179647527e-06, "loss": 1.0435, "step": 14832 }, { "epoch": 1.5783145350074483, "grad_norm": 10.1328125, "learning_rate": 9.490726264923252e-06, "loss": 1.0549, "step": 14833 }, { "epoch": 1.578420940625665, "grad_norm": 15.296875, "learning_rate": 9.490690733371234e-06, "loss": 1.1981, "step": 14834 }, { "epoch": 1.5785273462438818, "grad_norm": 10.609375, "learning_rate": 9.490655201819217e-06, "loss": 0.9318, "step": 14835 }, { "epoch": 1.5786337518620983, "grad_norm": 15.8125, "learning_rate": 9.490619670267199e-06, "loss": 0.9464, "step": 14836 }, { "epoch": 1.5787401574803148, "grad_norm": 9.9765625, "learning_rate": 9.49058413871518e-06, "loss": 1.0783, "step": 14837 }, { "epoch": 1.5788465630985316, "grad_norm": 10.84375, "learning_rate": 9.490548607163162e-06, "loss": 1.0149, "step": 14838 }, { "epoch": 1.5789529687167483, "grad_norm": 12.0, "learning_rate": 9.490513075611144e-06, "loss": 1.152, "step": 14839 }, { "epoch": 1.5790593743349648, "grad_norm": 15.078125, "learning_rate": 9.490477544059125e-06, "loss": 0.9231, "step": 14840 }, { "epoch": 1.5791657799531815, "grad_norm": 12.2265625, "learning_rate": 9.490442012507107e-06, "loss": 1.0484, "step": 14841 }, { "epoch": 1.5792721855713983, "grad_norm": 12.6640625, "learning_rate": 9.490406480955089e-06, "loss": 1.1678, "step": 14842 }, { "epoch": 1.5793785911896148, "grad_norm": 21.265625, "learning_rate": 9.49037094940307e-06, "loss": 1.2361, "step": 14843 }, { "epoch": 1.5794849968078315, "grad_norm": 12.2109375, "learning_rate": 9.490335417851052e-06, "loss": 1.0178, "step": 14844 }, { "epoch": 1.5795914024260482, "grad_norm": 13.1953125, "learning_rate": 9.490299886299035e-06, "loss": 1.1707, "step": 14845 }, { "epoch": 1.5796978080442647, "grad_norm": 11.1328125, "learning_rate": 9.490264354747017e-06, "loss": 1.2701, "step": 14846 }, { "epoch": 1.5798042136624812, "grad_norm": 12.5390625, "learning_rate": 9.490228823194998e-06, "loss": 1.0172, "step": 14847 }, { "epoch": 1.579910619280698, "grad_norm": 18.140625, "learning_rate": 9.49019329164298e-06, "loss": 0.9239, "step": 14848 }, { "epoch": 1.5800170248989147, "grad_norm": 11.3671875, "learning_rate": 9.490157760090961e-06, "loss": 1.2438, "step": 14849 }, { "epoch": 1.5801234305171312, "grad_norm": 13.921875, "learning_rate": 9.490122228538943e-06, "loss": 0.9813, "step": 14850 }, { "epoch": 1.580229836135348, "grad_norm": 11.59375, "learning_rate": 9.490086696986925e-06, "loss": 1.08, "step": 14851 }, { "epoch": 1.5803362417535647, "grad_norm": 10.0, "learning_rate": 9.490051165434906e-06, "loss": 0.9766, "step": 14852 }, { "epoch": 1.5804426473717812, "grad_norm": 19.234375, "learning_rate": 9.490015633882888e-06, "loss": 1.5914, "step": 14853 }, { "epoch": 1.580549052989998, "grad_norm": 11.2421875, "learning_rate": 9.48998010233087e-06, "loss": 1.0218, "step": 14854 }, { "epoch": 1.5806554586082147, "grad_norm": 12.2578125, "learning_rate": 9.489944570778853e-06, "loss": 1.2066, "step": 14855 }, { "epoch": 1.5807618642264312, "grad_norm": 11.8046875, "learning_rate": 9.489909039226834e-06, "loss": 0.9113, "step": 14856 }, { "epoch": 1.5808682698446477, "grad_norm": 17.0625, "learning_rate": 9.489873507674816e-06, "loss": 0.9489, "step": 14857 }, { "epoch": 1.5809746754628644, "grad_norm": 10.140625, "learning_rate": 9.489837976122798e-06, "loss": 1.0577, "step": 14858 }, { "epoch": 1.5810810810810811, "grad_norm": 12.625, "learning_rate": 9.48980244457078e-06, "loss": 1.1264, "step": 14859 }, { "epoch": 1.5811874866992977, "grad_norm": 9.578125, "learning_rate": 9.489766913018763e-06, "loss": 0.9804, "step": 14860 }, { "epoch": 1.5812938923175144, "grad_norm": 14.25, "learning_rate": 9.489731381466744e-06, "loss": 1.2413, "step": 14861 }, { "epoch": 1.5814002979357311, "grad_norm": 12.3828125, "learning_rate": 9.489695849914726e-06, "loss": 0.9903, "step": 14862 }, { "epoch": 1.5815067035539476, "grad_norm": 14.0390625, "learning_rate": 9.489660318362707e-06, "loss": 0.8739, "step": 14863 }, { "epoch": 1.5816131091721641, "grad_norm": 14.3203125, "learning_rate": 9.489624786810689e-06, "loss": 1.0252, "step": 14864 }, { "epoch": 1.581719514790381, "grad_norm": 11.765625, "learning_rate": 9.48958925525867e-06, "loss": 0.8688, "step": 14865 }, { "epoch": 1.5818259204085976, "grad_norm": 14.84375, "learning_rate": 9.489553723706652e-06, "loss": 1.0176, "step": 14866 }, { "epoch": 1.581932326026814, "grad_norm": 14.203125, "learning_rate": 9.489518192154634e-06, "loss": 1.0383, "step": 14867 }, { "epoch": 1.5820387316450308, "grad_norm": 13.625, "learning_rate": 9.489482660602615e-06, "loss": 0.9982, "step": 14868 }, { "epoch": 1.5821451372632476, "grad_norm": 10.3671875, "learning_rate": 9.489447129050597e-06, "loss": 0.9735, "step": 14869 }, { "epoch": 1.582251542881464, "grad_norm": 13.9765625, "learning_rate": 9.489411597498579e-06, "loss": 1.062, "step": 14870 }, { "epoch": 1.5823579484996808, "grad_norm": 11.8125, "learning_rate": 9.489376065946562e-06, "loss": 1.2898, "step": 14871 }, { "epoch": 1.5824643541178975, "grad_norm": 11.1484375, "learning_rate": 9.489340534394544e-06, "loss": 0.9792, "step": 14872 }, { "epoch": 1.582570759736114, "grad_norm": 12.4296875, "learning_rate": 9.489305002842525e-06, "loss": 1.3211, "step": 14873 }, { "epoch": 1.5826771653543306, "grad_norm": 9.7265625, "learning_rate": 9.489269471290507e-06, "loss": 1.029, "step": 14874 }, { "epoch": 1.5827835709725475, "grad_norm": 12.828125, "learning_rate": 9.489233939738488e-06, "loss": 0.927, "step": 14875 }, { "epoch": 1.582889976590764, "grad_norm": 9.734375, "learning_rate": 9.48919840818647e-06, "loss": 0.9124, "step": 14876 }, { "epoch": 1.5829963822089805, "grad_norm": 14.2421875, "learning_rate": 9.489162876634452e-06, "loss": 1.0125, "step": 14877 }, { "epoch": 1.5831027878271973, "grad_norm": 13.875, "learning_rate": 9.489127345082433e-06, "loss": 1.0668, "step": 14878 }, { "epoch": 1.583209193445414, "grad_norm": 12.9921875, "learning_rate": 9.489091813530415e-06, "loss": 0.9042, "step": 14879 }, { "epoch": 1.5833155990636305, "grad_norm": 12.6640625, "learning_rate": 9.489056281978396e-06, "loss": 1.0162, "step": 14880 }, { "epoch": 1.5834220046818472, "grad_norm": 9.703125, "learning_rate": 9.48902075042638e-06, "loss": 0.7951, "step": 14881 }, { "epoch": 1.583528410300064, "grad_norm": 15.0390625, "learning_rate": 9.488985218874361e-06, "loss": 0.8818, "step": 14882 }, { "epoch": 1.5836348159182805, "grad_norm": 10.6171875, "learning_rate": 9.488949687322343e-06, "loss": 1.1336, "step": 14883 }, { "epoch": 1.583741221536497, "grad_norm": 10.859375, "learning_rate": 9.488914155770324e-06, "loss": 1.0668, "step": 14884 }, { "epoch": 1.5838476271547137, "grad_norm": 15.9296875, "learning_rate": 9.488878624218306e-06, "loss": 1.1819, "step": 14885 }, { "epoch": 1.5839540327729305, "grad_norm": 15.78125, "learning_rate": 9.48884309266629e-06, "loss": 1.3907, "step": 14886 }, { "epoch": 1.584060438391147, "grad_norm": 19.3125, "learning_rate": 9.488807561114271e-06, "loss": 1.395, "step": 14887 }, { "epoch": 1.5841668440093637, "grad_norm": 15.765625, "learning_rate": 9.488772029562253e-06, "loss": 1.2344, "step": 14888 }, { "epoch": 1.5842732496275804, "grad_norm": 20.109375, "learning_rate": 9.488736498010234e-06, "loss": 0.7561, "step": 14889 }, { "epoch": 1.584379655245797, "grad_norm": 10.4453125, "learning_rate": 9.488700966458216e-06, "loss": 0.9996, "step": 14890 }, { "epoch": 1.5844860608640137, "grad_norm": 12.3671875, "learning_rate": 9.488665434906197e-06, "loss": 1.1264, "step": 14891 }, { "epoch": 1.5845924664822304, "grad_norm": 16.84375, "learning_rate": 9.488629903354179e-06, "loss": 1.4361, "step": 14892 }, { "epoch": 1.584698872100447, "grad_norm": 13.1015625, "learning_rate": 9.48859437180216e-06, "loss": 1.0272, "step": 14893 }, { "epoch": 1.5848052777186634, "grad_norm": 18.90625, "learning_rate": 9.488558840250142e-06, "loss": 1.1184, "step": 14894 }, { "epoch": 1.5849116833368802, "grad_norm": 11.828125, "learning_rate": 9.488523308698124e-06, "loss": 1.0478, "step": 14895 }, { "epoch": 1.5850180889550969, "grad_norm": 16.234375, "learning_rate": 9.488487777146107e-06, "loss": 0.7444, "step": 14896 }, { "epoch": 1.5851244945733134, "grad_norm": 11.5078125, "learning_rate": 9.488452245594089e-06, "loss": 0.862, "step": 14897 }, { "epoch": 1.5852309001915301, "grad_norm": 14.796875, "learning_rate": 9.48841671404207e-06, "loss": 0.9537, "step": 14898 }, { "epoch": 1.5853373058097469, "grad_norm": 11.328125, "learning_rate": 9.488381182490052e-06, "loss": 1.2405, "step": 14899 }, { "epoch": 1.5854437114279634, "grad_norm": 13.8359375, "learning_rate": 9.488345650938034e-06, "loss": 1.0597, "step": 14900 }, { "epoch": 1.5855501170461799, "grad_norm": 14.609375, "learning_rate": 9.488310119386017e-06, "loss": 1.2409, "step": 14901 }, { "epoch": 1.5856565226643968, "grad_norm": 10.6796875, "learning_rate": 9.488274587833998e-06, "loss": 1.2185, "step": 14902 }, { "epoch": 1.5857629282826133, "grad_norm": 12.8046875, "learning_rate": 9.48823905628198e-06, "loss": 1.0956, "step": 14903 }, { "epoch": 1.5858693339008298, "grad_norm": 16.359375, "learning_rate": 9.488203524729962e-06, "loss": 1.0492, "step": 14904 }, { "epoch": 1.5859757395190466, "grad_norm": 14.3984375, "learning_rate": 9.488167993177943e-06, "loss": 1.0786, "step": 14905 }, { "epoch": 1.5860821451372633, "grad_norm": 17.84375, "learning_rate": 9.488132461625923e-06, "loss": 1.242, "step": 14906 }, { "epoch": 1.5861885507554798, "grad_norm": 16.390625, "learning_rate": 9.488096930073907e-06, "loss": 1.1307, "step": 14907 }, { "epoch": 1.5862949563736966, "grad_norm": 11.0625, "learning_rate": 9.488061398521888e-06, "loss": 1.0832, "step": 14908 }, { "epoch": 1.5864013619919133, "grad_norm": 13.46875, "learning_rate": 9.48802586696987e-06, "loss": 1.1241, "step": 14909 }, { "epoch": 1.5865077676101298, "grad_norm": 12.1328125, "learning_rate": 9.487990335417851e-06, "loss": 0.9873, "step": 14910 }, { "epoch": 1.5866141732283463, "grad_norm": 9.734375, "learning_rate": 9.487954803865833e-06, "loss": 1.0192, "step": 14911 }, { "epoch": 1.5867205788465633, "grad_norm": 11.8515625, "learning_rate": 9.487919272313816e-06, "loss": 1.0356, "step": 14912 }, { "epoch": 1.5868269844647798, "grad_norm": 15.2421875, "learning_rate": 9.487883740761798e-06, "loss": 1.159, "step": 14913 }, { "epoch": 1.5869333900829963, "grad_norm": 10.5234375, "learning_rate": 9.48784820920978e-06, "loss": 1.0654, "step": 14914 }, { "epoch": 1.587039795701213, "grad_norm": 14.453125, "learning_rate": 9.487812677657761e-06, "loss": 1.2886, "step": 14915 }, { "epoch": 1.5871462013194297, "grad_norm": 12.9140625, "learning_rate": 9.487777146105743e-06, "loss": 1.2594, "step": 14916 }, { "epoch": 1.5872526069376462, "grad_norm": 15.5546875, "learning_rate": 9.487741614553724e-06, "loss": 0.9035, "step": 14917 }, { "epoch": 1.587359012555863, "grad_norm": 14.4609375, "learning_rate": 9.487706083001706e-06, "loss": 1.2168, "step": 14918 }, { "epoch": 1.5874654181740797, "grad_norm": 12.9609375, "learning_rate": 9.487670551449687e-06, "loss": 0.9988, "step": 14919 }, { "epoch": 1.5875718237922962, "grad_norm": 9.3203125, "learning_rate": 9.487635019897669e-06, "loss": 0.7572, "step": 14920 }, { "epoch": 1.5876782294105127, "grad_norm": 16.640625, "learning_rate": 9.48759948834565e-06, "loss": 1.093, "step": 14921 }, { "epoch": 1.5877846350287295, "grad_norm": 12.0703125, "learning_rate": 9.487563956793634e-06, "loss": 1.198, "step": 14922 }, { "epoch": 1.5878910406469462, "grad_norm": 10.4765625, "learning_rate": 9.487528425241616e-06, "loss": 0.9542, "step": 14923 }, { "epoch": 1.5879974462651627, "grad_norm": 12.7109375, "learning_rate": 9.487492893689597e-06, "loss": 1.1967, "step": 14924 }, { "epoch": 1.5881038518833794, "grad_norm": 12.78125, "learning_rate": 9.487457362137579e-06, "loss": 1.0172, "step": 14925 }, { "epoch": 1.5882102575015962, "grad_norm": 12.40625, "learning_rate": 9.48742183058556e-06, "loss": 0.9298, "step": 14926 }, { "epoch": 1.5883166631198127, "grad_norm": 10.9375, "learning_rate": 9.487386299033544e-06, "loss": 1.049, "step": 14927 }, { "epoch": 1.5884230687380294, "grad_norm": 11.484375, "learning_rate": 9.487350767481525e-06, "loss": 1.2179, "step": 14928 }, { "epoch": 1.5885294743562461, "grad_norm": 11.625, "learning_rate": 9.487315235929507e-06, "loss": 0.8648, "step": 14929 }, { "epoch": 1.5886358799744627, "grad_norm": 10.296875, "learning_rate": 9.487279704377489e-06, "loss": 1.1733, "step": 14930 }, { "epoch": 1.5887422855926792, "grad_norm": 15.609375, "learning_rate": 9.48724417282547e-06, "loss": 1.1604, "step": 14931 }, { "epoch": 1.588848691210896, "grad_norm": 11.9609375, "learning_rate": 9.487208641273452e-06, "loss": 1.1207, "step": 14932 }, { "epoch": 1.5889550968291126, "grad_norm": 16.171875, "learning_rate": 9.487173109721433e-06, "loss": 1.0498, "step": 14933 }, { "epoch": 1.5890615024473291, "grad_norm": 12.453125, "learning_rate": 9.487137578169415e-06, "loss": 1.3293, "step": 14934 }, { "epoch": 1.5891679080655459, "grad_norm": 14.1484375, "learning_rate": 9.487102046617397e-06, "loss": 1.1935, "step": 14935 }, { "epoch": 1.5892743136837626, "grad_norm": 12.03125, "learning_rate": 9.487066515065378e-06, "loss": 1.3546, "step": 14936 }, { "epoch": 1.589380719301979, "grad_norm": 11.3515625, "learning_rate": 9.487030983513361e-06, "loss": 1.3258, "step": 14937 }, { "epoch": 1.5894871249201958, "grad_norm": 11.484375, "learning_rate": 9.486995451961343e-06, "loss": 0.9223, "step": 14938 }, { "epoch": 1.5895935305384126, "grad_norm": 13.3515625, "learning_rate": 9.486959920409325e-06, "loss": 1.1024, "step": 14939 }, { "epoch": 1.589699936156629, "grad_norm": 13.5625, "learning_rate": 9.486924388857306e-06, "loss": 1.1761, "step": 14940 }, { "epoch": 1.5898063417748456, "grad_norm": 10.796875, "learning_rate": 9.486888857305288e-06, "loss": 1.2066, "step": 14941 }, { "epoch": 1.5899127473930623, "grad_norm": 11.5078125, "learning_rate": 9.48685332575327e-06, "loss": 1.0602, "step": 14942 }, { "epoch": 1.590019153011279, "grad_norm": 13.046875, "learning_rate": 9.486817794201251e-06, "loss": 1.3018, "step": 14943 }, { "epoch": 1.5901255586294956, "grad_norm": 9.921875, "learning_rate": 9.486782262649233e-06, "loss": 1.184, "step": 14944 }, { "epoch": 1.5902319642477123, "grad_norm": 13.2265625, "learning_rate": 9.486746731097214e-06, "loss": 1.2095, "step": 14945 }, { "epoch": 1.590338369865929, "grad_norm": 14.3046875, "learning_rate": 9.486711199545196e-06, "loss": 0.9003, "step": 14946 }, { "epoch": 1.5904447754841455, "grad_norm": 14.375, "learning_rate": 9.486675667993178e-06, "loss": 1.052, "step": 14947 }, { "epoch": 1.590551181102362, "grad_norm": 11.1796875, "learning_rate": 9.486640136441161e-06, "loss": 1.1239, "step": 14948 }, { "epoch": 1.590657586720579, "grad_norm": 12.7265625, "learning_rate": 9.486604604889142e-06, "loss": 0.9367, "step": 14949 }, { "epoch": 1.5907639923387955, "grad_norm": 9.9296875, "learning_rate": 9.486569073337124e-06, "loss": 0.9445, "step": 14950 }, { "epoch": 1.590870397957012, "grad_norm": 12.484375, "learning_rate": 9.486533541785106e-06, "loss": 0.9974, "step": 14951 }, { "epoch": 1.5909768035752287, "grad_norm": 13.5234375, "learning_rate": 9.486498010233087e-06, "loss": 0.9462, "step": 14952 }, { "epoch": 1.5910832091934455, "grad_norm": 11.78125, "learning_rate": 9.48646247868107e-06, "loss": 1.2505, "step": 14953 }, { "epoch": 1.591189614811662, "grad_norm": 10.671875, "learning_rate": 9.486426947129052e-06, "loss": 1.0868, "step": 14954 }, { "epoch": 1.5912960204298787, "grad_norm": 12.796875, "learning_rate": 9.486391415577034e-06, "loss": 0.987, "step": 14955 }, { "epoch": 1.5914024260480955, "grad_norm": 14.2109375, "learning_rate": 9.486355884025015e-06, "loss": 1.0168, "step": 14956 }, { "epoch": 1.591508831666312, "grad_norm": 16.359375, "learning_rate": 9.486320352472997e-06, "loss": 0.8372, "step": 14957 }, { "epoch": 1.5916152372845285, "grad_norm": 16.171875, "learning_rate": 9.486284820920979e-06, "loss": 1.0882, "step": 14958 }, { "epoch": 1.5917216429027454, "grad_norm": 12.8828125, "learning_rate": 9.48624928936896e-06, "loss": 1.1395, "step": 14959 }, { "epoch": 1.591828048520962, "grad_norm": 14.2265625, "learning_rate": 9.486213757816942e-06, "loss": 1.0151, "step": 14960 }, { "epoch": 1.5919344541391784, "grad_norm": 12.515625, "learning_rate": 9.486178226264923e-06, "loss": 1.2751, "step": 14961 }, { "epoch": 1.5920408597573952, "grad_norm": 14.3828125, "learning_rate": 9.486142694712905e-06, "loss": 1.2207, "step": 14962 }, { "epoch": 1.592147265375612, "grad_norm": 17.71875, "learning_rate": 9.486107163160888e-06, "loss": 1.1728, "step": 14963 }, { "epoch": 1.5922536709938284, "grad_norm": 14.78125, "learning_rate": 9.48607163160887e-06, "loss": 1.1558, "step": 14964 }, { "epoch": 1.5923600766120451, "grad_norm": 14.203125, "learning_rate": 9.486036100056852e-06, "loss": 1.1882, "step": 14965 }, { "epoch": 1.5924664822302619, "grad_norm": 11.03125, "learning_rate": 9.486000568504833e-06, "loss": 1.0915, "step": 14966 }, { "epoch": 1.5925728878484784, "grad_norm": 10.0703125, "learning_rate": 9.485965036952815e-06, "loss": 0.8544, "step": 14967 }, { "epoch": 1.592679293466695, "grad_norm": 10.96875, "learning_rate": 9.485929505400796e-06, "loss": 1.2904, "step": 14968 }, { "epoch": 1.5927856990849116, "grad_norm": 13.6171875, "learning_rate": 9.485893973848778e-06, "loss": 1.5203, "step": 14969 }, { "epoch": 1.5928921047031284, "grad_norm": 13.8515625, "learning_rate": 9.48585844229676e-06, "loss": 0.9813, "step": 14970 }, { "epoch": 1.5929985103213449, "grad_norm": 11.1796875, "learning_rate": 9.485822910744741e-06, "loss": 0.994, "step": 14971 }, { "epoch": 1.5931049159395616, "grad_norm": 17.265625, "learning_rate": 9.485787379192723e-06, "loss": 1.0849, "step": 14972 }, { "epoch": 1.5932113215577783, "grad_norm": 14.1328125, "learning_rate": 9.485751847640706e-06, "loss": 1.0827, "step": 14973 }, { "epoch": 1.5933177271759948, "grad_norm": 10.6171875, "learning_rate": 9.485716316088688e-06, "loss": 1.3339, "step": 14974 }, { "epoch": 1.5934241327942116, "grad_norm": 10.6015625, "learning_rate": 9.48568078453667e-06, "loss": 1.0168, "step": 14975 }, { "epoch": 1.5935305384124283, "grad_norm": 10.6640625, "learning_rate": 9.485645252984651e-06, "loss": 1.1515, "step": 14976 }, { "epoch": 1.5936369440306448, "grad_norm": 8.6875, "learning_rate": 9.485609721432633e-06, "loss": 1.0024, "step": 14977 }, { "epoch": 1.5937433496488613, "grad_norm": 10.796875, "learning_rate": 9.485574189880616e-06, "loss": 0.9768, "step": 14978 }, { "epoch": 1.593849755267078, "grad_norm": 13.3046875, "learning_rate": 9.485538658328597e-06, "loss": 0.9831, "step": 14979 }, { "epoch": 1.5939561608852948, "grad_norm": 9.40625, "learning_rate": 9.485503126776579e-06, "loss": 1.0827, "step": 14980 }, { "epoch": 1.5940625665035113, "grad_norm": 10.2265625, "learning_rate": 9.48546759522456e-06, "loss": 1.1383, "step": 14981 }, { "epoch": 1.594168972121728, "grad_norm": 10.4296875, "learning_rate": 9.485432063672542e-06, "loss": 1.4288, "step": 14982 }, { "epoch": 1.5942753777399448, "grad_norm": 14.6875, "learning_rate": 9.485396532120524e-06, "loss": 1.0617, "step": 14983 }, { "epoch": 1.5943817833581613, "grad_norm": 10.421875, "learning_rate": 9.485361000568505e-06, "loss": 1.179, "step": 14984 }, { "epoch": 1.594488188976378, "grad_norm": 12.953125, "learning_rate": 9.485325469016487e-06, "loss": 1.1516, "step": 14985 }, { "epoch": 1.5945945945945947, "grad_norm": 14.875, "learning_rate": 9.485289937464469e-06, "loss": 1.219, "step": 14986 }, { "epoch": 1.5947010002128112, "grad_norm": 17.1875, "learning_rate": 9.48525440591245e-06, "loss": 0.8732, "step": 14987 }, { "epoch": 1.5948074058310278, "grad_norm": 12.96875, "learning_rate": 9.485218874360432e-06, "loss": 1.0762, "step": 14988 }, { "epoch": 1.5949138114492445, "grad_norm": 10.84375, "learning_rate": 9.485183342808415e-06, "loss": 1.1286, "step": 14989 }, { "epoch": 1.5950202170674612, "grad_norm": 10.7109375, "learning_rate": 9.485147811256397e-06, "loss": 1.0342, "step": 14990 }, { "epoch": 1.5951266226856777, "grad_norm": 11.4921875, "learning_rate": 9.485112279704378e-06, "loss": 1.0545, "step": 14991 }, { "epoch": 1.5952330283038945, "grad_norm": 13.484375, "learning_rate": 9.48507674815236e-06, "loss": 1.0845, "step": 14992 }, { "epoch": 1.5953394339221112, "grad_norm": 13.7265625, "learning_rate": 9.485041216600342e-06, "loss": 1.0617, "step": 14993 }, { "epoch": 1.5954458395403277, "grad_norm": 14.8203125, "learning_rate": 9.485005685048325e-06, "loss": 1.2439, "step": 14994 }, { "epoch": 1.5955522451585442, "grad_norm": 10.484375, "learning_rate": 9.484970153496305e-06, "loss": 1.2012, "step": 14995 }, { "epoch": 1.5956586507767612, "grad_norm": 18.0625, "learning_rate": 9.484934621944286e-06, "loss": 1.0486, "step": 14996 }, { "epoch": 1.5957650563949777, "grad_norm": 13.71875, "learning_rate": 9.484899090392268e-06, "loss": 1.1455, "step": 14997 }, { "epoch": 1.5958714620131942, "grad_norm": 11.2421875, "learning_rate": 9.48486355884025e-06, "loss": 1.1155, "step": 14998 }, { "epoch": 1.595977867631411, "grad_norm": 12.4453125, "learning_rate": 9.484828027288233e-06, "loss": 1.2523, "step": 14999 }, { "epoch": 1.5960842732496276, "grad_norm": 20.15625, "learning_rate": 9.484792495736215e-06, "loss": 1.1467, "step": 15000 }, { "epoch": 1.5961906788678442, "grad_norm": 17.0, "learning_rate": 9.484756964184196e-06, "loss": 1.1266, "step": 15001 }, { "epoch": 1.596297084486061, "grad_norm": 11.4765625, "learning_rate": 9.484721432632178e-06, "loss": 1.2735, "step": 15002 }, { "epoch": 1.5964034901042776, "grad_norm": 11.1015625, "learning_rate": 9.48468590108016e-06, "loss": 1.317, "step": 15003 }, { "epoch": 1.5965098957224941, "grad_norm": 14.6328125, "learning_rate": 9.484650369528143e-06, "loss": 1.1253, "step": 15004 }, { "epoch": 1.5966163013407106, "grad_norm": 15.8046875, "learning_rate": 9.484614837976124e-06, "loss": 0.9714, "step": 15005 }, { "epoch": 1.5967227069589276, "grad_norm": 14.9921875, "learning_rate": 9.484579306424106e-06, "loss": 1.0767, "step": 15006 }, { "epoch": 1.596829112577144, "grad_norm": 13.328125, "learning_rate": 9.484543774872088e-06, "loss": 1.0521, "step": 15007 }, { "epoch": 1.5969355181953606, "grad_norm": 14.2890625, "learning_rate": 9.484508243320069e-06, "loss": 0.6069, "step": 15008 }, { "epoch": 1.5970419238135773, "grad_norm": 11.234375, "learning_rate": 9.48447271176805e-06, "loss": 1.0724, "step": 15009 }, { "epoch": 1.597148329431794, "grad_norm": 12.9375, "learning_rate": 9.484437180216032e-06, "loss": 1.1025, "step": 15010 }, { "epoch": 1.5972547350500106, "grad_norm": 12.671875, "learning_rate": 9.484401648664014e-06, "loss": 1.2676, "step": 15011 }, { "epoch": 1.5973611406682273, "grad_norm": 12.125, "learning_rate": 9.484366117111996e-06, "loss": 0.9242, "step": 15012 }, { "epoch": 1.597467546286444, "grad_norm": 13.7265625, "learning_rate": 9.484330585559977e-06, "loss": 1.2469, "step": 15013 }, { "epoch": 1.5975739519046606, "grad_norm": 17.421875, "learning_rate": 9.48429505400796e-06, "loss": 1.1885, "step": 15014 }, { "epoch": 1.597680357522877, "grad_norm": 13.90625, "learning_rate": 9.484259522455942e-06, "loss": 0.8849, "step": 15015 }, { "epoch": 1.5977867631410938, "grad_norm": 15.21875, "learning_rate": 9.484223990903924e-06, "loss": 1.2668, "step": 15016 }, { "epoch": 1.5978931687593105, "grad_norm": 13.0703125, "learning_rate": 9.484188459351905e-06, "loss": 1.1439, "step": 15017 }, { "epoch": 1.597999574377527, "grad_norm": 14.5, "learning_rate": 9.484152927799887e-06, "loss": 1.0434, "step": 15018 }, { "epoch": 1.5981059799957438, "grad_norm": 11.0078125, "learning_rate": 9.48411739624787e-06, "loss": 1.0959, "step": 15019 }, { "epoch": 1.5982123856139605, "grad_norm": 10.0390625, "learning_rate": 9.484081864695852e-06, "loss": 1.047, "step": 15020 }, { "epoch": 1.598318791232177, "grad_norm": 12.265625, "learning_rate": 9.484046333143833e-06, "loss": 1.1348, "step": 15021 }, { "epoch": 1.5984251968503937, "grad_norm": 26.421875, "learning_rate": 9.484010801591815e-06, "loss": 1.5587, "step": 15022 }, { "epoch": 1.5985316024686105, "grad_norm": 9.6328125, "learning_rate": 9.483975270039797e-06, "loss": 1.019, "step": 15023 }, { "epoch": 1.598638008086827, "grad_norm": 12.875, "learning_rate": 9.483939738487778e-06, "loss": 1.0392, "step": 15024 }, { "epoch": 1.5987444137050435, "grad_norm": 11.7734375, "learning_rate": 9.48390420693576e-06, "loss": 1.0612, "step": 15025 }, { "epoch": 1.5988508193232602, "grad_norm": 10.609375, "learning_rate": 9.483868675383741e-06, "loss": 1.2291, "step": 15026 }, { "epoch": 1.598957224941477, "grad_norm": 13.234375, "learning_rate": 9.483833143831723e-06, "loss": 1.031, "step": 15027 }, { "epoch": 1.5990636305596935, "grad_norm": 11.328125, "learning_rate": 9.483797612279705e-06, "loss": 1.1311, "step": 15028 }, { "epoch": 1.5991700361779102, "grad_norm": 18.28125, "learning_rate": 9.483762080727686e-06, "loss": 0.9472, "step": 15029 }, { "epoch": 1.599276441796127, "grad_norm": 14.109375, "learning_rate": 9.48372654917567e-06, "loss": 1.1779, "step": 15030 }, { "epoch": 1.5993828474143434, "grad_norm": 11.5546875, "learning_rate": 9.483691017623651e-06, "loss": 1.2045, "step": 15031 }, { "epoch": 1.59948925303256, "grad_norm": 13.9921875, "learning_rate": 9.483655486071633e-06, "loss": 0.9712, "step": 15032 }, { "epoch": 1.599595658650777, "grad_norm": 12.234375, "learning_rate": 9.483619954519614e-06, "loss": 1.221, "step": 15033 }, { "epoch": 1.5997020642689934, "grad_norm": 13.3984375, "learning_rate": 9.483584422967596e-06, "loss": 1.0999, "step": 15034 }, { "epoch": 1.59980846988721, "grad_norm": 11.3984375, "learning_rate": 9.483548891415578e-06, "loss": 0.9754, "step": 15035 }, { "epoch": 1.5999148755054267, "grad_norm": 11.3671875, "learning_rate": 9.48351335986356e-06, "loss": 0.949, "step": 15036 }, { "epoch": 1.6000212811236434, "grad_norm": 12.0234375, "learning_rate": 9.48347782831154e-06, "loss": 1.0417, "step": 15037 }, { "epoch": 1.60012768674186, "grad_norm": 14.1171875, "learning_rate": 9.483442296759522e-06, "loss": 1.033, "step": 15038 }, { "epoch": 1.6002340923600766, "grad_norm": 18.421875, "learning_rate": 9.483406765207504e-06, "loss": 1.3039, "step": 15039 }, { "epoch": 1.6003404979782934, "grad_norm": 11.078125, "learning_rate": 9.483371233655487e-06, "loss": 0.9973, "step": 15040 }, { "epoch": 1.6004469035965099, "grad_norm": 11.2890625, "learning_rate": 9.483335702103469e-06, "loss": 1.0592, "step": 15041 }, { "epoch": 1.6005533092147264, "grad_norm": 15.0625, "learning_rate": 9.48330017055145e-06, "loss": 1.1532, "step": 15042 }, { "epoch": 1.6006597148329433, "grad_norm": 13.03125, "learning_rate": 9.483264638999432e-06, "loss": 0.8354, "step": 15043 }, { "epoch": 1.6007661204511598, "grad_norm": 12.4765625, "learning_rate": 9.483229107447414e-06, "loss": 0.9127, "step": 15044 }, { "epoch": 1.6008725260693764, "grad_norm": 13.7109375, "learning_rate": 9.483193575895397e-06, "loss": 1.0862, "step": 15045 }, { "epoch": 1.600978931687593, "grad_norm": 13.2578125, "learning_rate": 9.483158044343379e-06, "loss": 1.1504, "step": 15046 }, { "epoch": 1.6010853373058098, "grad_norm": 13.2109375, "learning_rate": 9.48312251279136e-06, "loss": 1.0466, "step": 15047 }, { "epoch": 1.6011917429240263, "grad_norm": 12.7265625, "learning_rate": 9.483086981239342e-06, "loss": 1.1538, "step": 15048 }, { "epoch": 1.601298148542243, "grad_norm": 11.109375, "learning_rate": 9.483051449687323e-06, "loss": 0.5251, "step": 15049 }, { "epoch": 1.6014045541604598, "grad_norm": 10.5625, "learning_rate": 9.483015918135305e-06, "loss": 0.9867, "step": 15050 }, { "epoch": 1.6015109597786763, "grad_norm": 14.0, "learning_rate": 9.482980386583287e-06, "loss": 1.1484, "step": 15051 }, { "epoch": 1.6016173653968928, "grad_norm": 16.09375, "learning_rate": 9.482944855031268e-06, "loss": 1.2387, "step": 15052 }, { "epoch": 1.6017237710151095, "grad_norm": 11.9296875, "learning_rate": 9.48290932347925e-06, "loss": 0.8783, "step": 15053 }, { "epoch": 1.6018301766333263, "grad_norm": 30.984375, "learning_rate": 9.482873791927231e-06, "loss": 1.1119, "step": 15054 }, { "epoch": 1.6019365822515428, "grad_norm": 12.109375, "learning_rate": 9.482838260375215e-06, "loss": 0.933, "step": 15055 }, { "epoch": 1.6020429878697595, "grad_norm": 13.1328125, "learning_rate": 9.482802728823196e-06, "loss": 1.1645, "step": 15056 }, { "epoch": 1.6021493934879762, "grad_norm": 16.640625, "learning_rate": 9.482767197271178e-06, "loss": 1.5317, "step": 15057 }, { "epoch": 1.6022557991061928, "grad_norm": 11.875, "learning_rate": 9.48273166571916e-06, "loss": 1.0346, "step": 15058 }, { "epoch": 1.6023622047244095, "grad_norm": 12.625, "learning_rate": 9.482696134167141e-06, "loss": 1.1415, "step": 15059 }, { "epoch": 1.6024686103426262, "grad_norm": 12.7265625, "learning_rate": 9.482660602615123e-06, "loss": 1.1509, "step": 15060 }, { "epoch": 1.6025750159608427, "grad_norm": 12.5859375, "learning_rate": 9.482625071063104e-06, "loss": 1.0006, "step": 15061 }, { "epoch": 1.6026814215790592, "grad_norm": 13.9921875, "learning_rate": 9.482589539511086e-06, "loss": 0.8762, "step": 15062 }, { "epoch": 1.602787827197276, "grad_norm": 13.4609375, "learning_rate": 9.482554007959068e-06, "loss": 1.199, "step": 15063 }, { "epoch": 1.6028942328154927, "grad_norm": 10.25, "learning_rate": 9.48251847640705e-06, "loss": 1.048, "step": 15064 }, { "epoch": 1.6030006384337092, "grad_norm": 10.515625, "learning_rate": 9.482482944855031e-06, "loss": 1.1182, "step": 15065 }, { "epoch": 1.603107044051926, "grad_norm": 10.1875, "learning_rate": 9.482447413303014e-06, "loss": 1.3636, "step": 15066 }, { "epoch": 1.6032134496701427, "grad_norm": 14.7734375, "learning_rate": 9.482411881750996e-06, "loss": 1.135, "step": 15067 }, { "epoch": 1.6033198552883592, "grad_norm": 13.203125, "learning_rate": 9.482376350198977e-06, "loss": 1.3353, "step": 15068 }, { "epoch": 1.603426260906576, "grad_norm": 16.34375, "learning_rate": 9.482340818646959e-06, "loss": 1.2233, "step": 15069 }, { "epoch": 1.6035326665247926, "grad_norm": 16.25, "learning_rate": 9.48230528709494e-06, "loss": 1.2542, "step": 15070 }, { "epoch": 1.6036390721430092, "grad_norm": 13.90625, "learning_rate": 9.482269755542924e-06, "loss": 0.9463, "step": 15071 }, { "epoch": 1.6037454777612257, "grad_norm": 11.375, "learning_rate": 9.482234223990905e-06, "loss": 1.2604, "step": 15072 }, { "epoch": 1.6038518833794424, "grad_norm": 13.796875, "learning_rate": 9.482198692438887e-06, "loss": 0.855, "step": 15073 }, { "epoch": 1.6039582889976591, "grad_norm": 13.4921875, "learning_rate": 9.482163160886869e-06, "loss": 1.3072, "step": 15074 }, { "epoch": 1.6040646946158756, "grad_norm": 17.4375, "learning_rate": 9.48212762933485e-06, "loss": 1.0629, "step": 15075 }, { "epoch": 1.6041711002340924, "grad_norm": 11.3515625, "learning_rate": 9.482092097782832e-06, "loss": 1.1967, "step": 15076 }, { "epoch": 1.604277505852309, "grad_norm": 15.765625, "learning_rate": 9.482056566230814e-06, "loss": 1.6436, "step": 15077 }, { "epoch": 1.6043839114705256, "grad_norm": 11.375, "learning_rate": 9.482021034678795e-06, "loss": 1.1245, "step": 15078 }, { "epoch": 1.6044903170887421, "grad_norm": 11.1171875, "learning_rate": 9.481985503126777e-06, "loss": 1.0328, "step": 15079 }, { "epoch": 1.604596722706959, "grad_norm": 13.5703125, "learning_rate": 9.481949971574758e-06, "loss": 1.0428, "step": 15080 }, { "epoch": 1.6047031283251756, "grad_norm": 12.3515625, "learning_rate": 9.481914440022742e-06, "loss": 1.1187, "step": 15081 }, { "epoch": 1.604809533943392, "grad_norm": 12.046875, "learning_rate": 9.481878908470723e-06, "loss": 1.31, "step": 15082 }, { "epoch": 1.6049159395616088, "grad_norm": 14.234375, "learning_rate": 9.481843376918705e-06, "loss": 1.2769, "step": 15083 }, { "epoch": 1.6050223451798256, "grad_norm": 13.1171875, "learning_rate": 9.481807845366686e-06, "loss": 0.7783, "step": 15084 }, { "epoch": 1.605128750798042, "grad_norm": 13.578125, "learning_rate": 9.481772313814668e-06, "loss": 1.0604, "step": 15085 }, { "epoch": 1.6052351564162588, "grad_norm": 10.625, "learning_rate": 9.48173678226265e-06, "loss": 1.1425, "step": 15086 }, { "epoch": 1.6053415620344755, "grad_norm": 13.1015625, "learning_rate": 9.481701250710631e-06, "loss": 0.9018, "step": 15087 }, { "epoch": 1.605447967652692, "grad_norm": 10.5625, "learning_rate": 9.481665719158613e-06, "loss": 1.3524, "step": 15088 }, { "epoch": 1.6055543732709086, "grad_norm": 11.25, "learning_rate": 9.481630187606595e-06, "loss": 1.0593, "step": 15089 }, { "epoch": 1.6056607788891255, "grad_norm": 12.859375, "learning_rate": 9.481594656054576e-06, "loss": 0.8537, "step": 15090 }, { "epoch": 1.605767184507342, "grad_norm": 12.8671875, "learning_rate": 9.48155912450256e-06, "loss": 1.0863, "step": 15091 }, { "epoch": 1.6058735901255585, "grad_norm": 17.421875, "learning_rate": 9.481523592950541e-06, "loss": 1.2446, "step": 15092 }, { "epoch": 1.6059799957437753, "grad_norm": 14.6484375, "learning_rate": 9.481488061398523e-06, "loss": 1.1151, "step": 15093 }, { "epoch": 1.606086401361992, "grad_norm": 11.1015625, "learning_rate": 9.481452529846504e-06, "loss": 1.0402, "step": 15094 }, { "epoch": 1.6061928069802085, "grad_norm": 13.4296875, "learning_rate": 9.481416998294486e-06, "loss": 0.9979, "step": 15095 }, { "epoch": 1.6062992125984252, "grad_norm": 10.8515625, "learning_rate": 9.481381466742469e-06, "loss": 0.9133, "step": 15096 }, { "epoch": 1.606405618216642, "grad_norm": 14.1171875, "learning_rate": 9.48134593519045e-06, "loss": 1.1999, "step": 15097 }, { "epoch": 1.6065120238348585, "grad_norm": 12.046875, "learning_rate": 9.481310403638432e-06, "loss": 0.8249, "step": 15098 }, { "epoch": 1.606618429453075, "grad_norm": 17.125, "learning_rate": 9.481274872086414e-06, "loss": 1.1446, "step": 15099 }, { "epoch": 1.6067248350712917, "grad_norm": 11.828125, "learning_rate": 9.481239340534396e-06, "loss": 1.1944, "step": 15100 }, { "epoch": 1.6068312406895084, "grad_norm": 11.8671875, "learning_rate": 9.481203808982377e-06, "loss": 1.1621, "step": 15101 }, { "epoch": 1.606937646307725, "grad_norm": 13.8984375, "learning_rate": 9.481168277430359e-06, "loss": 0.9954, "step": 15102 }, { "epoch": 1.6070440519259417, "grad_norm": 11.1796875, "learning_rate": 9.48113274587834e-06, "loss": 1.1272, "step": 15103 }, { "epoch": 1.6071504575441584, "grad_norm": 17.140625, "learning_rate": 9.481097214326322e-06, "loss": 0.8753, "step": 15104 }, { "epoch": 1.607256863162375, "grad_norm": 13.0703125, "learning_rate": 9.481061682774304e-06, "loss": 0.985, "step": 15105 }, { "epoch": 1.6073632687805917, "grad_norm": 11.6953125, "learning_rate": 9.481026151222285e-06, "loss": 0.9023, "step": 15106 }, { "epoch": 1.6074696743988084, "grad_norm": 9.5234375, "learning_rate": 9.480990619670269e-06, "loss": 0.9222, "step": 15107 }, { "epoch": 1.607576080017025, "grad_norm": 11.7109375, "learning_rate": 9.48095508811825e-06, "loss": 1.3881, "step": 15108 }, { "epoch": 1.6076824856352414, "grad_norm": 11.046875, "learning_rate": 9.480919556566232e-06, "loss": 1.2592, "step": 15109 }, { "epoch": 1.6077888912534581, "grad_norm": 14.3671875, "learning_rate": 9.480884025014213e-06, "loss": 0.9811, "step": 15110 }, { "epoch": 1.6078952968716749, "grad_norm": 11.953125, "learning_rate": 9.480848493462195e-06, "loss": 1.231, "step": 15111 }, { "epoch": 1.6080017024898914, "grad_norm": 12.828125, "learning_rate": 9.480812961910178e-06, "loss": 1.037, "step": 15112 }, { "epoch": 1.6081081081081081, "grad_norm": 21.625, "learning_rate": 9.48077743035816e-06, "loss": 1.0433, "step": 15113 }, { "epoch": 1.6082145137263248, "grad_norm": 10.5859375, "learning_rate": 9.48074189880614e-06, "loss": 1.1591, "step": 15114 }, { "epoch": 1.6083209193445414, "grad_norm": 20.34375, "learning_rate": 9.480706367254121e-06, "loss": 1.2193, "step": 15115 }, { "epoch": 1.608427324962758, "grad_norm": 9.921875, "learning_rate": 9.480670835702103e-06, "loss": 1.2172, "step": 15116 }, { "epoch": 1.6085337305809748, "grad_norm": 14.546875, "learning_rate": 9.480635304150086e-06, "loss": 1.0847, "step": 15117 }, { "epoch": 1.6086401361991913, "grad_norm": 11.6953125, "learning_rate": 9.480599772598068e-06, "loss": 1.1283, "step": 15118 }, { "epoch": 1.6087465418174078, "grad_norm": 12.78125, "learning_rate": 9.48056424104605e-06, "loss": 1.2008, "step": 15119 }, { "epoch": 1.6088529474356246, "grad_norm": 14.0625, "learning_rate": 9.480528709494031e-06, "loss": 1.2271, "step": 15120 }, { "epoch": 1.6089593530538413, "grad_norm": 13.4609375, "learning_rate": 9.480493177942013e-06, "loss": 0.9623, "step": 15121 }, { "epoch": 1.6090657586720578, "grad_norm": 11.453125, "learning_rate": 9.480457646389996e-06, "loss": 1.2336, "step": 15122 }, { "epoch": 1.6091721642902745, "grad_norm": 15.4140625, "learning_rate": 9.480422114837978e-06, "loss": 1.7392, "step": 15123 }, { "epoch": 1.6092785699084913, "grad_norm": 12.703125, "learning_rate": 9.48038658328596e-06, "loss": 1.0075, "step": 15124 }, { "epoch": 1.6093849755267078, "grad_norm": 14.1875, "learning_rate": 9.48035105173394e-06, "loss": 1.2097, "step": 15125 }, { "epoch": 1.6094913811449243, "grad_norm": 11.9453125, "learning_rate": 9.480315520181922e-06, "loss": 1.064, "step": 15126 }, { "epoch": 1.6095977867631412, "grad_norm": 13.125, "learning_rate": 9.480279988629904e-06, "loss": 1.2038, "step": 15127 }, { "epoch": 1.6097041923813578, "grad_norm": 12.96875, "learning_rate": 9.480244457077886e-06, "loss": 1.0885, "step": 15128 }, { "epoch": 1.6098105979995743, "grad_norm": 16.578125, "learning_rate": 9.480208925525867e-06, "loss": 1.1164, "step": 15129 }, { "epoch": 1.609917003617791, "grad_norm": 12.53125, "learning_rate": 9.480173393973849e-06, "loss": 1.0742, "step": 15130 }, { "epoch": 1.6100234092360077, "grad_norm": 14.7734375, "learning_rate": 9.48013786242183e-06, "loss": 1.0158, "step": 15131 }, { "epoch": 1.6101298148542242, "grad_norm": 10.0546875, "learning_rate": 9.480102330869814e-06, "loss": 1.1291, "step": 15132 }, { "epoch": 1.610236220472441, "grad_norm": 13.53125, "learning_rate": 9.480066799317795e-06, "loss": 1.05, "step": 15133 }, { "epoch": 1.6103426260906577, "grad_norm": 13.9453125, "learning_rate": 9.480031267765777e-06, "loss": 1.0532, "step": 15134 }, { "epoch": 1.6104490317088742, "grad_norm": 10.0703125, "learning_rate": 9.479995736213759e-06, "loss": 0.9851, "step": 15135 }, { "epoch": 1.6105554373270907, "grad_norm": 12.1953125, "learning_rate": 9.47996020466174e-06, "loss": 1.025, "step": 15136 }, { "epoch": 1.6106618429453077, "grad_norm": 11.2578125, "learning_rate": 9.479924673109723e-06, "loss": 1.0147, "step": 15137 }, { "epoch": 1.6107682485635242, "grad_norm": 10.3671875, "learning_rate": 9.479889141557705e-06, "loss": 1.114, "step": 15138 }, { "epoch": 1.6108746541817407, "grad_norm": 10.2421875, "learning_rate": 9.479853610005687e-06, "loss": 1.4888, "step": 15139 }, { "epoch": 1.6109810597999574, "grad_norm": 11.3515625, "learning_rate": 9.479818078453668e-06, "loss": 1.1529, "step": 15140 }, { "epoch": 1.6110874654181742, "grad_norm": 9.0859375, "learning_rate": 9.47978254690165e-06, "loss": 0.865, "step": 15141 }, { "epoch": 1.6111938710363907, "grad_norm": 12.2109375, "learning_rate": 9.479747015349632e-06, "loss": 1.1638, "step": 15142 }, { "epoch": 1.6113002766546074, "grad_norm": 13.640625, "learning_rate": 9.479711483797613e-06, "loss": 1.0896, "step": 15143 }, { "epoch": 1.6114066822728241, "grad_norm": 14.484375, "learning_rate": 9.479675952245595e-06, "loss": 0.9001, "step": 15144 }, { "epoch": 1.6115130878910406, "grad_norm": 12.34375, "learning_rate": 9.479640420693576e-06, "loss": 1.4187, "step": 15145 }, { "epoch": 1.6116194935092572, "grad_norm": 13.0078125, "learning_rate": 9.479604889141558e-06, "loss": 1.376, "step": 15146 }, { "epoch": 1.6117258991274739, "grad_norm": 11.1953125, "learning_rate": 9.47956935758954e-06, "loss": 1.0807, "step": 15147 }, { "epoch": 1.6118323047456906, "grad_norm": 24.75, "learning_rate": 9.479533826037523e-06, "loss": 1.2003, "step": 15148 }, { "epoch": 1.6119387103639071, "grad_norm": 22.15625, "learning_rate": 9.479498294485504e-06, "loss": 1.0265, "step": 15149 }, { "epoch": 1.6120451159821239, "grad_norm": 10.46875, "learning_rate": 9.479462762933486e-06, "loss": 1.0871, "step": 15150 }, { "epoch": 1.6121515216003406, "grad_norm": 11.296875, "learning_rate": 9.479427231381468e-06, "loss": 1.1762, "step": 15151 }, { "epoch": 1.612257927218557, "grad_norm": 13.6484375, "learning_rate": 9.47939169982945e-06, "loss": 1.3491, "step": 15152 }, { "epoch": 1.6123643328367738, "grad_norm": 11.6953125, "learning_rate": 9.479356168277431e-06, "loss": 1.0568, "step": 15153 }, { "epoch": 1.6124707384549906, "grad_norm": 13.640625, "learning_rate": 9.479320636725412e-06, "loss": 1.1503, "step": 15154 }, { "epoch": 1.612577144073207, "grad_norm": 12.890625, "learning_rate": 9.479285105173394e-06, "loss": 1.2989, "step": 15155 }, { "epoch": 1.6126835496914236, "grad_norm": 13.6796875, "learning_rate": 9.479249573621376e-06, "loss": 0.837, "step": 15156 }, { "epoch": 1.6127899553096403, "grad_norm": 10.875, "learning_rate": 9.479214042069357e-06, "loss": 1.0423, "step": 15157 }, { "epoch": 1.612896360927857, "grad_norm": 12.8984375, "learning_rate": 9.47917851051734e-06, "loss": 0.9735, "step": 15158 }, { "epoch": 1.6130027665460736, "grad_norm": 10.0625, "learning_rate": 9.479142978965322e-06, "loss": 1.2623, "step": 15159 }, { "epoch": 1.6131091721642903, "grad_norm": 12.390625, "learning_rate": 9.479107447413304e-06, "loss": 1.1477, "step": 15160 }, { "epoch": 1.613215577782507, "grad_norm": 10.4140625, "learning_rate": 9.479071915861285e-06, "loss": 1.073, "step": 15161 }, { "epoch": 1.6133219834007235, "grad_norm": 9.8515625, "learning_rate": 9.479036384309267e-06, "loss": 0.9983, "step": 15162 }, { "epoch": 1.61342838901894, "grad_norm": 13.2890625, "learning_rate": 9.47900085275725e-06, "loss": 1.1968, "step": 15163 }, { "epoch": 1.613534794637157, "grad_norm": 10.0, "learning_rate": 9.478965321205232e-06, "loss": 1.1237, "step": 15164 }, { "epoch": 1.6136412002553735, "grad_norm": 12.9609375, "learning_rate": 9.478929789653214e-06, "loss": 1.0804, "step": 15165 }, { "epoch": 1.61374760587359, "grad_norm": 11.0, "learning_rate": 9.478894258101195e-06, "loss": 1.2485, "step": 15166 }, { "epoch": 1.6138540114918067, "grad_norm": 11.8828125, "learning_rate": 9.478858726549177e-06, "loss": 0.7516, "step": 15167 }, { "epoch": 1.6139604171100235, "grad_norm": 13.515625, "learning_rate": 9.478823194997158e-06, "loss": 1.2619, "step": 15168 }, { "epoch": 1.61406682272824, "grad_norm": 11.4140625, "learning_rate": 9.47878766344514e-06, "loss": 1.0588, "step": 15169 }, { "epoch": 1.6141732283464567, "grad_norm": 14.53125, "learning_rate": 9.478752131893122e-06, "loss": 1.0109, "step": 15170 }, { "epoch": 1.6142796339646734, "grad_norm": 11.125, "learning_rate": 9.478716600341103e-06, "loss": 1.1349, "step": 15171 }, { "epoch": 1.61438603958289, "grad_norm": 13.390625, "learning_rate": 9.478681068789085e-06, "loss": 1.2297, "step": 15172 }, { "epoch": 1.6144924452011065, "grad_norm": 12.546875, "learning_rate": 9.478645537237068e-06, "loss": 1.0123, "step": 15173 }, { "epoch": 1.6145988508193234, "grad_norm": 12.3984375, "learning_rate": 9.47861000568505e-06, "loss": 1.0626, "step": 15174 }, { "epoch": 1.61470525643754, "grad_norm": 14.109375, "learning_rate": 9.478574474133031e-06, "loss": 1.3382, "step": 15175 }, { "epoch": 1.6148116620557564, "grad_norm": 12.0859375, "learning_rate": 9.478538942581013e-06, "loss": 0.8745, "step": 15176 }, { "epoch": 1.6149180676739732, "grad_norm": 13.4375, "learning_rate": 9.478503411028995e-06, "loss": 1.2456, "step": 15177 }, { "epoch": 1.61502447329219, "grad_norm": 12.3984375, "learning_rate": 9.478467879476976e-06, "loss": 1.053, "step": 15178 }, { "epoch": 1.6151308789104064, "grad_norm": 11.2265625, "learning_rate": 9.478432347924958e-06, "loss": 0.9947, "step": 15179 }, { "epoch": 1.6152372845286231, "grad_norm": 13.546875, "learning_rate": 9.47839681637294e-06, "loss": 0.9099, "step": 15180 }, { "epoch": 1.6153436901468399, "grad_norm": 12.109375, "learning_rate": 9.478361284820921e-06, "loss": 0.6079, "step": 15181 }, { "epoch": 1.6154500957650564, "grad_norm": 13.0859375, "learning_rate": 9.478325753268903e-06, "loss": 1.0213, "step": 15182 }, { "epoch": 1.615556501383273, "grad_norm": 10.296875, "learning_rate": 9.478290221716884e-06, "loss": 1.2165, "step": 15183 }, { "epoch": 1.6156629070014896, "grad_norm": 15.765625, "learning_rate": 9.478254690164867e-06, "loss": 1.3629, "step": 15184 }, { "epoch": 1.6157693126197064, "grad_norm": 16.09375, "learning_rate": 9.478219158612849e-06, "loss": 1.0459, "step": 15185 }, { "epoch": 1.6158757182379229, "grad_norm": 15.1171875, "learning_rate": 9.47818362706083e-06, "loss": 1.2376, "step": 15186 }, { "epoch": 1.6159821238561396, "grad_norm": 10.25, "learning_rate": 9.478148095508812e-06, "loss": 1.3089, "step": 15187 }, { "epoch": 1.6160885294743563, "grad_norm": 11.625, "learning_rate": 9.478112563956794e-06, "loss": 1.1158, "step": 15188 }, { "epoch": 1.6161949350925728, "grad_norm": 11.734375, "learning_rate": 9.478077032404777e-06, "loss": 1.0205, "step": 15189 }, { "epoch": 1.6163013407107896, "grad_norm": 10.8984375, "learning_rate": 9.478041500852759e-06, "loss": 1.0892, "step": 15190 }, { "epoch": 1.6164077463290063, "grad_norm": 11.2890625, "learning_rate": 9.47800596930074e-06, "loss": 0.9707, "step": 15191 }, { "epoch": 1.6165141519472228, "grad_norm": 13.3359375, "learning_rate": 9.477970437748722e-06, "loss": 0.9969, "step": 15192 }, { "epoch": 1.6166205575654393, "grad_norm": 15.65625, "learning_rate": 9.477934906196704e-06, "loss": 1.1512, "step": 15193 }, { "epoch": 1.616726963183656, "grad_norm": 21.890625, "learning_rate": 9.477899374644685e-06, "loss": 1.0349, "step": 15194 }, { "epoch": 1.6168333688018728, "grad_norm": 13.921875, "learning_rate": 9.477863843092667e-06, "loss": 1.1317, "step": 15195 }, { "epoch": 1.6169397744200893, "grad_norm": 11.59375, "learning_rate": 9.477828311540648e-06, "loss": 1.0355, "step": 15196 }, { "epoch": 1.617046180038306, "grad_norm": 11.453125, "learning_rate": 9.47779277998863e-06, "loss": 1.0102, "step": 15197 }, { "epoch": 1.6171525856565228, "grad_norm": 14.703125, "learning_rate": 9.477757248436612e-06, "loss": 1.0057, "step": 15198 }, { "epoch": 1.6172589912747393, "grad_norm": 12.03125, "learning_rate": 9.477721716884595e-06, "loss": 1.6132, "step": 15199 }, { "epoch": 1.617365396892956, "grad_norm": 12.203125, "learning_rate": 9.477686185332577e-06, "loss": 1.0183, "step": 15200 }, { "epoch": 1.6174718025111727, "grad_norm": 13.1328125, "learning_rate": 9.477650653780558e-06, "loss": 1.193, "step": 15201 }, { "epoch": 1.6175782081293892, "grad_norm": 10.5703125, "learning_rate": 9.47761512222854e-06, "loss": 1.1814, "step": 15202 }, { "epoch": 1.6176846137476057, "grad_norm": 19.5625, "learning_rate": 9.477579590676521e-06, "loss": 1.1857, "step": 15203 }, { "epoch": 1.6177910193658225, "grad_norm": 12.421875, "learning_rate": 9.477544059124503e-06, "loss": 1.3563, "step": 15204 }, { "epoch": 1.6178974249840392, "grad_norm": 9.8828125, "learning_rate": 9.477508527572485e-06, "loss": 0.8502, "step": 15205 }, { "epoch": 1.6180038306022557, "grad_norm": 13.296875, "learning_rate": 9.477472996020466e-06, "loss": 1.0584, "step": 15206 }, { "epoch": 1.6181102362204725, "grad_norm": 10.0390625, "learning_rate": 9.477437464468448e-06, "loss": 1.0965, "step": 15207 }, { "epoch": 1.6182166418386892, "grad_norm": 13.9296875, "learning_rate": 9.47740193291643e-06, "loss": 1.2101, "step": 15208 }, { "epoch": 1.6183230474569057, "grad_norm": 15.8671875, "learning_rate": 9.477366401364413e-06, "loss": 0.9712, "step": 15209 }, { "epoch": 1.6184294530751222, "grad_norm": 15.7109375, "learning_rate": 9.477330869812394e-06, "loss": 1.4416, "step": 15210 }, { "epoch": 1.6185358586933392, "grad_norm": 10.84375, "learning_rate": 9.477295338260376e-06, "loss": 0.7187, "step": 15211 }, { "epoch": 1.6186422643115557, "grad_norm": 11.3359375, "learning_rate": 9.477259806708358e-06, "loss": 1.0924, "step": 15212 }, { "epoch": 1.6187486699297722, "grad_norm": 16.484375, "learning_rate": 9.477224275156339e-06, "loss": 1.0697, "step": 15213 }, { "epoch": 1.618855075547989, "grad_norm": 11.9921875, "learning_rate": 9.477188743604322e-06, "loss": 1.042, "step": 15214 }, { "epoch": 1.6189614811662056, "grad_norm": 13.5625, "learning_rate": 9.477153212052304e-06, "loss": 1.3506, "step": 15215 }, { "epoch": 1.6190678867844222, "grad_norm": 13.03125, "learning_rate": 9.477117680500286e-06, "loss": 1.0164, "step": 15216 }, { "epoch": 1.6191742924026389, "grad_norm": 19.078125, "learning_rate": 9.477082148948267e-06, "loss": 1.2047, "step": 15217 }, { "epoch": 1.6192806980208556, "grad_norm": 15.609375, "learning_rate": 9.477046617396249e-06, "loss": 1.2223, "step": 15218 }, { "epoch": 1.6193871036390721, "grad_norm": 16.3125, "learning_rate": 9.47701108584423e-06, "loss": 1.1102, "step": 15219 }, { "epoch": 1.6194935092572886, "grad_norm": 11.828125, "learning_rate": 9.476975554292212e-06, "loss": 1.2179, "step": 15220 }, { "epoch": 1.6195999148755056, "grad_norm": 13.9765625, "learning_rate": 9.476940022740194e-06, "loss": 1.0614, "step": 15221 }, { "epoch": 1.619706320493722, "grad_norm": 12.5234375, "learning_rate": 9.476904491188175e-06, "loss": 1.0377, "step": 15222 }, { "epoch": 1.6198127261119386, "grad_norm": 12.875, "learning_rate": 9.476868959636157e-06, "loss": 1.144, "step": 15223 }, { "epoch": 1.6199191317301553, "grad_norm": 9.9296875, "learning_rate": 9.476833428084139e-06, "loss": 1.0271, "step": 15224 }, { "epoch": 1.620025537348372, "grad_norm": 20.390625, "learning_rate": 9.476797896532122e-06, "loss": 1.0674, "step": 15225 }, { "epoch": 1.6201319429665886, "grad_norm": 12.640625, "learning_rate": 9.476762364980103e-06, "loss": 1.3498, "step": 15226 }, { "epoch": 1.6202383485848053, "grad_norm": 15.0, "learning_rate": 9.476726833428085e-06, "loss": 0.8663, "step": 15227 }, { "epoch": 1.620344754203022, "grad_norm": 12.203125, "learning_rate": 9.476691301876067e-06, "loss": 1.0744, "step": 15228 }, { "epoch": 1.6204511598212386, "grad_norm": 13.3203125, "learning_rate": 9.476655770324048e-06, "loss": 1.3936, "step": 15229 }, { "epoch": 1.620557565439455, "grad_norm": 11.90625, "learning_rate": 9.476620238772032e-06, "loss": 1.1328, "step": 15230 }, { "epoch": 1.6206639710576718, "grad_norm": 14.015625, "learning_rate": 9.476584707220013e-06, "loss": 1.0885, "step": 15231 }, { "epoch": 1.6207703766758885, "grad_norm": 13.2421875, "learning_rate": 9.476549175667995e-06, "loss": 1.1722, "step": 15232 }, { "epoch": 1.620876782294105, "grad_norm": 15.0625, "learning_rate": 9.476513644115976e-06, "loss": 0.9683, "step": 15233 }, { "epoch": 1.6209831879123218, "grad_norm": 13.2265625, "learning_rate": 9.476478112563956e-06, "loss": 1.495, "step": 15234 }, { "epoch": 1.6210895935305385, "grad_norm": 14.40625, "learning_rate": 9.47644258101194e-06, "loss": 1.2502, "step": 15235 }, { "epoch": 1.621195999148755, "grad_norm": 12.328125, "learning_rate": 9.476407049459921e-06, "loss": 1.0387, "step": 15236 }, { "epoch": 1.6213024047669717, "grad_norm": 20.53125, "learning_rate": 9.476371517907903e-06, "loss": 0.9438, "step": 15237 }, { "epoch": 1.6214088103851885, "grad_norm": 13.09375, "learning_rate": 9.476335986355884e-06, "loss": 1.0616, "step": 15238 }, { "epoch": 1.621515216003405, "grad_norm": 16.328125, "learning_rate": 9.476300454803866e-06, "loss": 1.1387, "step": 15239 }, { "epoch": 1.6216216216216215, "grad_norm": 12.140625, "learning_rate": 9.47626492325185e-06, "loss": 1.0798, "step": 15240 }, { "epoch": 1.6217280272398382, "grad_norm": 14.5078125, "learning_rate": 9.476229391699831e-06, "loss": 0.9149, "step": 15241 }, { "epoch": 1.621834432858055, "grad_norm": 10.734375, "learning_rate": 9.476193860147813e-06, "loss": 0.9598, "step": 15242 }, { "epoch": 1.6219408384762715, "grad_norm": 14.09375, "learning_rate": 9.476158328595794e-06, "loss": 1.2326, "step": 15243 }, { "epoch": 1.6220472440944882, "grad_norm": 16.265625, "learning_rate": 9.476122797043776e-06, "loss": 1.2045, "step": 15244 }, { "epoch": 1.622153649712705, "grad_norm": 11.203125, "learning_rate": 9.476087265491757e-06, "loss": 1.1747, "step": 15245 }, { "epoch": 1.6222600553309214, "grad_norm": 10.984375, "learning_rate": 9.476051733939739e-06, "loss": 0.9802, "step": 15246 }, { "epoch": 1.6223664609491382, "grad_norm": 14.3515625, "learning_rate": 9.47601620238772e-06, "loss": 1.2475, "step": 15247 }, { "epoch": 1.622472866567355, "grad_norm": 13.4609375, "learning_rate": 9.475980670835702e-06, "loss": 0.7589, "step": 15248 }, { "epoch": 1.6225792721855714, "grad_norm": 13.59375, "learning_rate": 9.475945139283684e-06, "loss": 1.2343, "step": 15249 }, { "epoch": 1.622685677803788, "grad_norm": 11.203125, "learning_rate": 9.475909607731667e-06, "loss": 1.3069, "step": 15250 }, { "epoch": 1.6227920834220046, "grad_norm": 11.6328125, "learning_rate": 9.475874076179649e-06, "loss": 0.8201, "step": 15251 }, { "epoch": 1.6228984890402214, "grad_norm": 12.4921875, "learning_rate": 9.47583854462763e-06, "loss": 0.9435, "step": 15252 }, { "epoch": 1.623004894658438, "grad_norm": 11.1171875, "learning_rate": 9.475803013075612e-06, "loss": 1.0385, "step": 15253 }, { "epoch": 1.6231113002766546, "grad_norm": 8.078125, "learning_rate": 9.475767481523593e-06, "loss": 1.0195, "step": 15254 }, { "epoch": 1.6232177058948714, "grad_norm": 12.4140625, "learning_rate": 9.475731949971577e-06, "loss": 0.8737, "step": 15255 }, { "epoch": 1.6233241115130879, "grad_norm": 10.015625, "learning_rate": 9.475696418419558e-06, "loss": 1.0946, "step": 15256 }, { "epoch": 1.6234305171313044, "grad_norm": 10.96875, "learning_rate": 9.47566088686754e-06, "loss": 1.0426, "step": 15257 }, { "epoch": 1.6235369227495213, "grad_norm": 13.28125, "learning_rate": 9.475625355315522e-06, "loss": 1.0469, "step": 15258 }, { "epoch": 1.6236433283677378, "grad_norm": 12.0390625, "learning_rate": 9.475589823763503e-06, "loss": 1.1822, "step": 15259 }, { "epoch": 1.6237497339859543, "grad_norm": 15.625, "learning_rate": 9.475554292211485e-06, "loss": 1.167, "step": 15260 }, { "epoch": 1.623856139604171, "grad_norm": 10.828125, "learning_rate": 9.475518760659466e-06, "loss": 0.9431, "step": 15261 }, { "epoch": 1.6239625452223878, "grad_norm": 12.8125, "learning_rate": 9.475483229107448e-06, "loss": 0.9317, "step": 15262 }, { "epoch": 1.6240689508406043, "grad_norm": 12.9609375, "learning_rate": 9.47544769755543e-06, "loss": 1.3406, "step": 15263 }, { "epoch": 1.624175356458821, "grad_norm": 12.609375, "learning_rate": 9.475412166003411e-06, "loss": 1.2033, "step": 15264 }, { "epoch": 1.6242817620770378, "grad_norm": 17.140625, "learning_rate": 9.475376634451393e-06, "loss": 1.3149, "step": 15265 }, { "epoch": 1.6243881676952543, "grad_norm": 13.8125, "learning_rate": 9.475341102899376e-06, "loss": 1.0014, "step": 15266 }, { "epoch": 1.6244945733134708, "grad_norm": 12.5234375, "learning_rate": 9.475305571347358e-06, "loss": 1.1423, "step": 15267 }, { "epoch": 1.6246009789316878, "grad_norm": 12.5234375, "learning_rate": 9.47527003979534e-06, "loss": 1.201, "step": 15268 }, { "epoch": 1.6247073845499043, "grad_norm": 14.1875, "learning_rate": 9.475234508243321e-06, "loss": 1.2607, "step": 15269 }, { "epoch": 1.6248137901681208, "grad_norm": 10.046875, "learning_rate": 9.475198976691303e-06, "loss": 0.6784, "step": 15270 }, { "epoch": 1.6249201957863375, "grad_norm": 14.328125, "learning_rate": 9.475163445139284e-06, "loss": 1.3736, "step": 15271 }, { "epoch": 1.6250266014045542, "grad_norm": 11.0234375, "learning_rate": 9.475127913587266e-06, "loss": 0.9939, "step": 15272 }, { "epoch": 1.6251330070227707, "grad_norm": 13.0546875, "learning_rate": 9.475092382035247e-06, "loss": 0.9962, "step": 15273 }, { "epoch": 1.6252394126409875, "grad_norm": 12.3984375, "learning_rate": 9.475056850483229e-06, "loss": 1.1726, "step": 15274 }, { "epoch": 1.6253458182592042, "grad_norm": 15.984375, "learning_rate": 9.47502131893121e-06, "loss": 1.162, "step": 15275 }, { "epoch": 1.6254522238774207, "grad_norm": 17.328125, "learning_rate": 9.474985787379194e-06, "loss": 1.1393, "step": 15276 }, { "epoch": 1.6255586294956372, "grad_norm": 13.90625, "learning_rate": 9.474950255827176e-06, "loss": 0.8995, "step": 15277 }, { "epoch": 1.625665035113854, "grad_norm": 14.0625, "learning_rate": 9.474914724275157e-06, "loss": 1.0617, "step": 15278 }, { "epoch": 1.6257714407320707, "grad_norm": 15.2265625, "learning_rate": 9.474879192723139e-06, "loss": 1.1095, "step": 15279 }, { "epoch": 1.6258778463502872, "grad_norm": 10.5390625, "learning_rate": 9.47484366117112e-06, "loss": 0.8988, "step": 15280 }, { "epoch": 1.625984251968504, "grad_norm": 15.2421875, "learning_rate": 9.474808129619104e-06, "loss": 1.0603, "step": 15281 }, { "epoch": 1.6260906575867207, "grad_norm": 12.953125, "learning_rate": 9.474772598067085e-06, "loss": 1.0308, "step": 15282 }, { "epoch": 1.6261970632049372, "grad_norm": 12.515625, "learning_rate": 9.474737066515067e-06, "loss": 1.1865, "step": 15283 }, { "epoch": 1.626303468823154, "grad_norm": 13.6484375, "learning_rate": 9.474701534963048e-06, "loss": 1.1746, "step": 15284 }, { "epoch": 1.6264098744413706, "grad_norm": 13.3671875, "learning_rate": 9.47466600341103e-06, "loss": 1.3334, "step": 15285 }, { "epoch": 1.6265162800595871, "grad_norm": 19.265625, "learning_rate": 9.474630471859012e-06, "loss": 1.0517, "step": 15286 }, { "epoch": 1.6266226856778037, "grad_norm": 14.4921875, "learning_rate": 9.474594940306993e-06, "loss": 1.1949, "step": 15287 }, { "epoch": 1.6267290912960204, "grad_norm": 11.7578125, "learning_rate": 9.474559408754975e-06, "loss": 1.1994, "step": 15288 }, { "epoch": 1.6268354969142371, "grad_norm": 11.921875, "learning_rate": 9.474523877202956e-06, "loss": 1.1553, "step": 15289 }, { "epoch": 1.6269419025324536, "grad_norm": 14.296875, "learning_rate": 9.474488345650938e-06, "loss": 1.2602, "step": 15290 }, { "epoch": 1.6270483081506704, "grad_norm": 20.03125, "learning_rate": 9.474452814098921e-06, "loss": 1.0539, "step": 15291 }, { "epoch": 1.627154713768887, "grad_norm": 14.5234375, "learning_rate": 9.474417282546903e-06, "loss": 1.3643, "step": 15292 }, { "epoch": 1.6272611193871036, "grad_norm": 12.1953125, "learning_rate": 9.474381750994885e-06, "loss": 1.0806, "step": 15293 }, { "epoch": 1.6273675250053201, "grad_norm": 13.4453125, "learning_rate": 9.474346219442866e-06, "loss": 1.2768, "step": 15294 }, { "epoch": 1.627473930623537, "grad_norm": 14.3515625, "learning_rate": 9.474310687890848e-06, "loss": 0.9911, "step": 15295 }, { "epoch": 1.6275803362417536, "grad_norm": 11.78125, "learning_rate": 9.47427515633883e-06, "loss": 0.9784, "step": 15296 }, { "epoch": 1.62768674185997, "grad_norm": 11.59375, "learning_rate": 9.474239624786811e-06, "loss": 1.267, "step": 15297 }, { "epoch": 1.6277931474781868, "grad_norm": 12.40625, "learning_rate": 9.474204093234793e-06, "loss": 1.2977, "step": 15298 }, { "epoch": 1.6278995530964036, "grad_norm": 11.328125, "learning_rate": 9.474168561682774e-06, "loss": 0.9233, "step": 15299 }, { "epoch": 1.62800595871462, "grad_norm": 27.90625, "learning_rate": 9.474133030130756e-06, "loss": 1.4588, "step": 15300 }, { "epoch": 1.6281123643328368, "grad_norm": 12.515625, "learning_rate": 9.474097498578737e-06, "loss": 1.2755, "step": 15301 }, { "epoch": 1.6282187699510535, "grad_norm": 9.6171875, "learning_rate": 9.47406196702672e-06, "loss": 1.1236, "step": 15302 }, { "epoch": 1.62832517556927, "grad_norm": 13.828125, "learning_rate": 9.474026435474702e-06, "loss": 0.9497, "step": 15303 }, { "epoch": 1.6284315811874865, "grad_norm": 14.4140625, "learning_rate": 9.473990903922684e-06, "loss": 1.2037, "step": 15304 }, { "epoch": 1.6285379868057035, "grad_norm": 11.515625, "learning_rate": 9.473955372370666e-06, "loss": 1.2403, "step": 15305 }, { "epoch": 1.62864439242392, "grad_norm": 12.59375, "learning_rate": 9.473919840818647e-06, "loss": 1.2117, "step": 15306 }, { "epoch": 1.6287507980421365, "grad_norm": 15.3671875, "learning_rate": 9.47388430926663e-06, "loss": 1.1057, "step": 15307 }, { "epoch": 1.6288572036603532, "grad_norm": 17.890625, "learning_rate": 9.473848777714612e-06, "loss": 1.538, "step": 15308 }, { "epoch": 1.62896360927857, "grad_norm": 13.6953125, "learning_rate": 9.473813246162594e-06, "loss": 1.1345, "step": 15309 }, { "epoch": 1.6290700148967865, "grad_norm": 14.7734375, "learning_rate": 9.473777714610575e-06, "loss": 0.9325, "step": 15310 }, { "epoch": 1.6291764205150032, "grad_norm": 11.3515625, "learning_rate": 9.473742183058557e-06, "loss": 0.9206, "step": 15311 }, { "epoch": 1.62928282613322, "grad_norm": 11.9921875, "learning_rate": 9.473706651506539e-06, "loss": 1.2164, "step": 15312 }, { "epoch": 1.6293892317514365, "grad_norm": 9.75, "learning_rate": 9.47367111995452e-06, "loss": 1.0947, "step": 15313 }, { "epoch": 1.629495637369653, "grad_norm": 12.8671875, "learning_rate": 9.473635588402502e-06, "loss": 1.292, "step": 15314 }, { "epoch": 1.62960204298787, "grad_norm": 11.8203125, "learning_rate": 9.473600056850483e-06, "loss": 0.9286, "step": 15315 }, { "epoch": 1.6297084486060864, "grad_norm": 11.1796875, "learning_rate": 9.473564525298465e-06, "loss": 1.0798, "step": 15316 }, { "epoch": 1.629814854224303, "grad_norm": 11.8828125, "learning_rate": 9.473528993746448e-06, "loss": 1.2439, "step": 15317 }, { "epoch": 1.6299212598425197, "grad_norm": 13.4609375, "learning_rate": 9.47349346219443e-06, "loss": 1.2075, "step": 15318 }, { "epoch": 1.6300276654607364, "grad_norm": 10.890625, "learning_rate": 9.473457930642411e-06, "loss": 1.0935, "step": 15319 }, { "epoch": 1.630134071078953, "grad_norm": 15.375, "learning_rate": 9.473422399090393e-06, "loss": 1.127, "step": 15320 }, { "epoch": 1.6302404766971696, "grad_norm": 10.1796875, "learning_rate": 9.473386867538375e-06, "loss": 1.0754, "step": 15321 }, { "epoch": 1.6303468823153864, "grad_norm": 10.8359375, "learning_rate": 9.473351335986358e-06, "loss": 1.1566, "step": 15322 }, { "epoch": 1.630453287933603, "grad_norm": 11.2890625, "learning_rate": 9.473315804434338e-06, "loss": 1.22, "step": 15323 }, { "epoch": 1.6305596935518194, "grad_norm": 11.140625, "learning_rate": 9.47328027288232e-06, "loss": 0.9869, "step": 15324 }, { "epoch": 1.6306660991700361, "grad_norm": 13.4140625, "learning_rate": 9.473244741330301e-06, "loss": 1.19, "step": 15325 }, { "epoch": 1.6307725047882529, "grad_norm": 12.8125, "learning_rate": 9.473209209778283e-06, "loss": 1.1803, "step": 15326 }, { "epoch": 1.6308789104064694, "grad_norm": 9.40625, "learning_rate": 9.473173678226266e-06, "loss": 0.85, "step": 15327 }, { "epoch": 1.630985316024686, "grad_norm": 12.1484375, "learning_rate": 9.473138146674248e-06, "loss": 1.4724, "step": 15328 }, { "epoch": 1.6310917216429028, "grad_norm": 10.2109375, "learning_rate": 9.47310261512223e-06, "loss": 1.0397, "step": 15329 }, { "epoch": 1.6311981272611193, "grad_norm": 12.7734375, "learning_rate": 9.47306708357021e-06, "loss": 1.1391, "step": 15330 }, { "epoch": 1.631304532879336, "grad_norm": 12.3046875, "learning_rate": 9.473031552018192e-06, "loss": 1.1219, "step": 15331 }, { "epoch": 1.6314109384975528, "grad_norm": 11.0, "learning_rate": 9.472996020466176e-06, "loss": 1.0371, "step": 15332 }, { "epoch": 1.6315173441157693, "grad_norm": 13.6875, "learning_rate": 9.472960488914157e-06, "loss": 1.1076, "step": 15333 }, { "epoch": 1.6316237497339858, "grad_norm": 13.4453125, "learning_rate": 9.472924957362139e-06, "loss": 1.0956, "step": 15334 }, { "epoch": 1.6317301553522026, "grad_norm": 18.015625, "learning_rate": 9.47288942581012e-06, "loss": 1.2318, "step": 15335 }, { "epoch": 1.6318365609704193, "grad_norm": 11.2578125, "learning_rate": 9.472853894258102e-06, "loss": 1.1467, "step": 15336 }, { "epoch": 1.6319429665886358, "grad_norm": 21.09375, "learning_rate": 9.472818362706084e-06, "loss": 1.1693, "step": 15337 }, { "epoch": 1.6320493722068525, "grad_norm": 13.0625, "learning_rate": 9.472782831154065e-06, "loss": 1.1679, "step": 15338 }, { "epoch": 1.6321557778250693, "grad_norm": 10.34375, "learning_rate": 9.472747299602047e-06, "loss": 1.0268, "step": 15339 }, { "epoch": 1.6322621834432858, "grad_norm": 15.65625, "learning_rate": 9.472711768050029e-06, "loss": 1.3027, "step": 15340 }, { "epoch": 1.6323685890615023, "grad_norm": 10.375, "learning_rate": 9.47267623649801e-06, "loss": 0.9508, "step": 15341 }, { "epoch": 1.6324749946797192, "grad_norm": 14.2421875, "learning_rate": 9.472640704945992e-06, "loss": 1.0295, "step": 15342 }, { "epoch": 1.6325814002979357, "grad_norm": 16.96875, "learning_rate": 9.472605173393975e-06, "loss": 0.8797, "step": 15343 }, { "epoch": 1.6326878059161523, "grad_norm": 16.984375, "learning_rate": 9.472569641841957e-06, "loss": 0.8678, "step": 15344 }, { "epoch": 1.632794211534369, "grad_norm": 15.5859375, "learning_rate": 9.472534110289938e-06, "loss": 0.9881, "step": 15345 }, { "epoch": 1.6329006171525857, "grad_norm": 10.2109375, "learning_rate": 9.47249857873792e-06, "loss": 1.0595, "step": 15346 }, { "epoch": 1.6330070227708022, "grad_norm": 9.046875, "learning_rate": 9.472463047185902e-06, "loss": 0.9813, "step": 15347 }, { "epoch": 1.633113428389019, "grad_norm": 9.296875, "learning_rate": 9.472427515633885e-06, "loss": 1.0567, "step": 15348 }, { "epoch": 1.6332198340072357, "grad_norm": 19.125, "learning_rate": 9.472391984081866e-06, "loss": 1.3008, "step": 15349 }, { "epoch": 1.6333262396254522, "grad_norm": 12.6484375, "learning_rate": 9.472356452529848e-06, "loss": 0.9195, "step": 15350 }, { "epoch": 1.6334326452436687, "grad_norm": 12.8671875, "learning_rate": 9.47232092097783e-06, "loss": 1.1469, "step": 15351 }, { "epoch": 1.6335390508618857, "grad_norm": 13.7890625, "learning_rate": 9.472285389425811e-06, "loss": 1.1165, "step": 15352 }, { "epoch": 1.6336454564801022, "grad_norm": 14.9765625, "learning_rate": 9.472249857873793e-06, "loss": 1.2086, "step": 15353 }, { "epoch": 1.6337518620983187, "grad_norm": 21.84375, "learning_rate": 9.472214326321774e-06, "loss": 1.2181, "step": 15354 }, { "epoch": 1.6338582677165354, "grad_norm": 10.6171875, "learning_rate": 9.472178794769756e-06, "loss": 0.93, "step": 15355 }, { "epoch": 1.6339646733347521, "grad_norm": 12.078125, "learning_rate": 9.472143263217738e-06, "loss": 0.9154, "step": 15356 }, { "epoch": 1.6340710789529687, "grad_norm": 24.734375, "learning_rate": 9.47210773166572e-06, "loss": 1.0894, "step": 15357 }, { "epoch": 1.6341774845711854, "grad_norm": 11.8984375, "learning_rate": 9.472072200113703e-06, "loss": 1.0924, "step": 15358 }, { "epoch": 1.6342838901894021, "grad_norm": 14.140625, "learning_rate": 9.472036668561684e-06, "loss": 1.1659, "step": 15359 }, { "epoch": 1.6343902958076186, "grad_norm": 13.671875, "learning_rate": 9.472001137009666e-06, "loss": 1.0984, "step": 15360 }, { "epoch": 1.6344967014258351, "grad_norm": 16.859375, "learning_rate": 9.471965605457647e-06, "loss": 1.0166, "step": 15361 }, { "epoch": 1.6346031070440519, "grad_norm": 13.0703125, "learning_rate": 9.471930073905629e-06, "loss": 1.1786, "step": 15362 }, { "epoch": 1.6347095126622686, "grad_norm": 12.6171875, "learning_rate": 9.47189454235361e-06, "loss": 1.1536, "step": 15363 }, { "epoch": 1.6348159182804851, "grad_norm": 13.1640625, "learning_rate": 9.471859010801592e-06, "loss": 1.1372, "step": 15364 }, { "epoch": 1.6349223238987018, "grad_norm": 18.015625, "learning_rate": 9.471823479249574e-06, "loss": 1.1413, "step": 15365 }, { "epoch": 1.6350287295169186, "grad_norm": 14.6328125, "learning_rate": 9.471787947697555e-06, "loss": 1.3693, "step": 15366 }, { "epoch": 1.635135135135135, "grad_norm": 11.0546875, "learning_rate": 9.471752416145537e-06, "loss": 0.8188, "step": 15367 }, { "epoch": 1.6352415407533518, "grad_norm": 16.375, "learning_rate": 9.47171688459352e-06, "loss": 1.2609, "step": 15368 }, { "epoch": 1.6353479463715686, "grad_norm": 15.6328125, "learning_rate": 9.471681353041502e-06, "loss": 1.3042, "step": 15369 }, { "epoch": 1.635454351989785, "grad_norm": 13.75, "learning_rate": 9.471645821489484e-06, "loss": 1.0363, "step": 15370 }, { "epoch": 1.6355607576080016, "grad_norm": 9.1875, "learning_rate": 9.471610289937465e-06, "loss": 0.9242, "step": 15371 }, { "epoch": 1.6356671632262183, "grad_norm": 14.96875, "learning_rate": 9.471574758385447e-06, "loss": 0.9658, "step": 15372 }, { "epoch": 1.635773568844435, "grad_norm": 12.5625, "learning_rate": 9.47153922683343e-06, "loss": 0.7725, "step": 15373 }, { "epoch": 1.6358799744626515, "grad_norm": 13.4609375, "learning_rate": 9.471503695281412e-06, "loss": 1.2336, "step": 15374 }, { "epoch": 1.6359863800808683, "grad_norm": 9.6953125, "learning_rate": 9.471468163729393e-06, "loss": 1.1162, "step": 15375 }, { "epoch": 1.636092785699085, "grad_norm": 14.6953125, "learning_rate": 9.471432632177375e-06, "loss": 1.273, "step": 15376 }, { "epoch": 1.6361991913173015, "grad_norm": 20.0625, "learning_rate": 9.471397100625357e-06, "loss": 1.1158, "step": 15377 }, { "epoch": 1.6363055969355182, "grad_norm": 14.5078125, "learning_rate": 9.471361569073338e-06, "loss": 0.9484, "step": 15378 }, { "epoch": 1.636412002553735, "grad_norm": 12.0234375, "learning_rate": 9.47132603752132e-06, "loss": 1.2593, "step": 15379 }, { "epoch": 1.6365184081719515, "grad_norm": 15.4375, "learning_rate": 9.471290505969301e-06, "loss": 0.9932, "step": 15380 }, { "epoch": 1.636624813790168, "grad_norm": 11.78125, "learning_rate": 9.471254974417283e-06, "loss": 0.9171, "step": 15381 }, { "epoch": 1.6367312194083847, "grad_norm": 11.3046875, "learning_rate": 9.471219442865265e-06, "loss": 1.3185, "step": 15382 }, { "epoch": 1.6368376250266015, "grad_norm": 15.28125, "learning_rate": 9.471183911313246e-06, "loss": 1.1296, "step": 15383 }, { "epoch": 1.636944030644818, "grad_norm": 14.9921875, "learning_rate": 9.47114837976123e-06, "loss": 1.1347, "step": 15384 }, { "epoch": 1.6370504362630347, "grad_norm": 11.7578125, "learning_rate": 9.471112848209211e-06, "loss": 1.0043, "step": 15385 }, { "epoch": 1.6371568418812514, "grad_norm": 15.328125, "learning_rate": 9.471077316657193e-06, "loss": 1.2354, "step": 15386 }, { "epoch": 1.637263247499468, "grad_norm": 13.4375, "learning_rate": 9.471041785105174e-06, "loss": 1.2787, "step": 15387 }, { "epoch": 1.6373696531176845, "grad_norm": 13.96875, "learning_rate": 9.471006253553156e-06, "loss": 0.8421, "step": 15388 }, { "epoch": 1.6374760587359014, "grad_norm": 12.96875, "learning_rate": 9.470970722001137e-06, "loss": 0.7322, "step": 15389 }, { "epoch": 1.637582464354118, "grad_norm": 11.6484375, "learning_rate": 9.470935190449119e-06, "loss": 1.1211, "step": 15390 }, { "epoch": 1.6376888699723344, "grad_norm": 11.21875, "learning_rate": 9.4708996588971e-06, "loss": 0.9292, "step": 15391 }, { "epoch": 1.6377952755905512, "grad_norm": 14.5078125, "learning_rate": 9.470864127345082e-06, "loss": 1.1327, "step": 15392 }, { "epoch": 1.637901681208768, "grad_norm": 10.484375, "learning_rate": 9.470828595793064e-06, "loss": 0.7147, "step": 15393 }, { "epoch": 1.6380080868269844, "grad_norm": 14.53125, "learning_rate": 9.470793064241047e-06, "loss": 1.2738, "step": 15394 }, { "epoch": 1.6381144924452011, "grad_norm": 13.4765625, "learning_rate": 9.470757532689029e-06, "loss": 1.0935, "step": 15395 }, { "epoch": 1.6382208980634179, "grad_norm": 15.0234375, "learning_rate": 9.47072200113701e-06, "loss": 1.6042, "step": 15396 }, { "epoch": 1.6383273036816344, "grad_norm": 10.984375, "learning_rate": 9.470686469584992e-06, "loss": 1.059, "step": 15397 }, { "epoch": 1.6384337092998509, "grad_norm": 10.7578125, "learning_rate": 9.470650938032974e-06, "loss": 1.1368, "step": 15398 }, { "epoch": 1.6385401149180678, "grad_norm": 9.0234375, "learning_rate": 9.470615406480957e-06, "loss": 0.9114, "step": 15399 }, { "epoch": 1.6386465205362843, "grad_norm": 13.5, "learning_rate": 9.470579874928939e-06, "loss": 1.0699, "step": 15400 }, { "epoch": 1.6387529261545009, "grad_norm": 14.328125, "learning_rate": 9.47054434337692e-06, "loss": 1.1728, "step": 15401 }, { "epoch": 1.6388593317727176, "grad_norm": 13.46875, "learning_rate": 9.470508811824902e-06, "loss": 0.8577, "step": 15402 }, { "epoch": 1.6389657373909343, "grad_norm": 12.171875, "learning_rate": 9.470473280272883e-06, "loss": 1.0758, "step": 15403 }, { "epoch": 1.6390721430091508, "grad_norm": 15.0859375, "learning_rate": 9.470437748720865e-06, "loss": 1.131, "step": 15404 }, { "epoch": 1.6391785486273676, "grad_norm": 17.171875, "learning_rate": 9.470402217168847e-06, "loss": 1.3211, "step": 15405 }, { "epoch": 1.6392849542455843, "grad_norm": 9.78125, "learning_rate": 9.470366685616828e-06, "loss": 0.791, "step": 15406 }, { "epoch": 1.6393913598638008, "grad_norm": 14.71875, "learning_rate": 9.47033115406481e-06, "loss": 1.0174, "step": 15407 }, { "epoch": 1.6394977654820173, "grad_norm": 11.2890625, "learning_rate": 9.470295622512791e-06, "loss": 1.1454, "step": 15408 }, { "epoch": 1.639604171100234, "grad_norm": 9.7421875, "learning_rate": 9.470260090960775e-06, "loss": 0.9239, "step": 15409 }, { "epoch": 1.6397105767184508, "grad_norm": 10.4609375, "learning_rate": 9.470224559408756e-06, "loss": 0.8253, "step": 15410 }, { "epoch": 1.6398169823366673, "grad_norm": 10.734375, "learning_rate": 9.470189027856738e-06, "loss": 0.9862, "step": 15411 }, { "epoch": 1.639923387954884, "grad_norm": 9.453125, "learning_rate": 9.47015349630472e-06, "loss": 1.0595, "step": 15412 }, { "epoch": 1.6400297935731007, "grad_norm": 11.5859375, "learning_rate": 9.470117964752701e-06, "loss": 1.171, "step": 15413 }, { "epoch": 1.6401361991913173, "grad_norm": 11.2421875, "learning_rate": 9.470082433200683e-06, "loss": 1.1613, "step": 15414 }, { "epoch": 1.640242604809534, "grad_norm": 9.6484375, "learning_rate": 9.470046901648664e-06, "loss": 1.0165, "step": 15415 }, { "epoch": 1.6403490104277507, "grad_norm": 12.7890625, "learning_rate": 9.470011370096646e-06, "loss": 0.9643, "step": 15416 }, { "epoch": 1.6404554160459672, "grad_norm": 17.015625, "learning_rate": 9.469975838544628e-06, "loss": 1.3683, "step": 15417 }, { "epoch": 1.6405618216641837, "grad_norm": 12.8828125, "learning_rate": 9.46994030699261e-06, "loss": 0.8308, "step": 15418 }, { "epoch": 1.6406682272824005, "grad_norm": 11.84375, "learning_rate": 9.46990477544059e-06, "loss": 1.1842, "step": 15419 }, { "epoch": 1.6407746329006172, "grad_norm": 14.9765625, "learning_rate": 9.469869243888574e-06, "loss": 1.4183, "step": 15420 }, { "epoch": 1.6408810385188337, "grad_norm": 9.6640625, "learning_rate": 9.469833712336556e-06, "loss": 0.881, "step": 15421 }, { "epoch": 1.6409874441370504, "grad_norm": 14.7734375, "learning_rate": 9.469798180784537e-06, "loss": 0.8377, "step": 15422 }, { "epoch": 1.6410938497552672, "grad_norm": 10.9375, "learning_rate": 9.469762649232519e-06, "loss": 1.1897, "step": 15423 }, { "epoch": 1.6412002553734837, "grad_norm": 11.015625, "learning_rate": 9.4697271176805e-06, "loss": 1.4947, "step": 15424 }, { "epoch": 1.6413066609917004, "grad_norm": 11.984375, "learning_rate": 9.469691586128484e-06, "loss": 1.1785, "step": 15425 }, { "epoch": 1.6414130666099171, "grad_norm": 12.546875, "learning_rate": 9.469656054576465e-06, "loss": 1.1932, "step": 15426 }, { "epoch": 1.6415194722281337, "grad_norm": 11.359375, "learning_rate": 9.469620523024447e-06, "loss": 1.2891, "step": 15427 }, { "epoch": 1.6416258778463502, "grad_norm": 10.671875, "learning_rate": 9.469584991472429e-06, "loss": 1.0417, "step": 15428 }, { "epoch": 1.641732283464567, "grad_norm": 15.265625, "learning_rate": 9.46954945992041e-06, "loss": 1.1006, "step": 15429 }, { "epoch": 1.6418386890827836, "grad_norm": 13.4921875, "learning_rate": 9.469513928368392e-06, "loss": 1.0533, "step": 15430 }, { "epoch": 1.6419450947010001, "grad_norm": 13.9296875, "learning_rate": 9.469478396816373e-06, "loss": 0.9679, "step": 15431 }, { "epoch": 1.6420515003192169, "grad_norm": 11.6328125, "learning_rate": 9.469442865264355e-06, "loss": 0.9743, "step": 15432 }, { "epoch": 1.6421579059374336, "grad_norm": 16.84375, "learning_rate": 9.469407333712337e-06, "loss": 1.1213, "step": 15433 }, { "epoch": 1.6422643115556501, "grad_norm": 17.109375, "learning_rate": 9.469371802160318e-06, "loss": 1.236, "step": 15434 }, { "epoch": 1.6423707171738666, "grad_norm": 12.8203125, "learning_rate": 9.469336270608302e-06, "loss": 1.1089, "step": 15435 }, { "epoch": 1.6424771227920836, "grad_norm": 12.625, "learning_rate": 9.469300739056283e-06, "loss": 1.217, "step": 15436 }, { "epoch": 1.6425835284103, "grad_norm": 14.7421875, "learning_rate": 9.469265207504265e-06, "loss": 0.939, "step": 15437 }, { "epoch": 1.6426899340285166, "grad_norm": 11.2421875, "learning_rate": 9.469229675952246e-06, "loss": 1.1019, "step": 15438 }, { "epoch": 1.6427963396467333, "grad_norm": 13.78125, "learning_rate": 9.469194144400228e-06, "loss": 1.1102, "step": 15439 }, { "epoch": 1.64290274526495, "grad_norm": 13.0625, "learning_rate": 9.469158612848211e-06, "loss": 1.1276, "step": 15440 }, { "epoch": 1.6430091508831666, "grad_norm": 14.5390625, "learning_rate": 9.469123081296193e-06, "loss": 1.0569, "step": 15441 }, { "epoch": 1.6431155565013833, "grad_norm": 12.34375, "learning_rate": 9.469087549744175e-06, "loss": 1.2848, "step": 15442 }, { "epoch": 1.6432219621196, "grad_norm": 11.8828125, "learning_rate": 9.469052018192154e-06, "loss": 1.0014, "step": 15443 }, { "epoch": 1.6433283677378165, "grad_norm": 14.5, "learning_rate": 9.469016486640136e-06, "loss": 1.2167, "step": 15444 }, { "epoch": 1.643434773356033, "grad_norm": 13.703125, "learning_rate": 9.46898095508812e-06, "loss": 1.0359, "step": 15445 }, { "epoch": 1.64354117897425, "grad_norm": 13.3515625, "learning_rate": 9.468945423536101e-06, "loss": 1.3325, "step": 15446 }, { "epoch": 1.6436475845924665, "grad_norm": 12.015625, "learning_rate": 9.468909891984083e-06, "loss": 1.0616, "step": 15447 }, { "epoch": 1.643753990210683, "grad_norm": 13.7890625, "learning_rate": 9.468874360432064e-06, "loss": 1.204, "step": 15448 }, { "epoch": 1.6438603958288998, "grad_norm": 12.96875, "learning_rate": 9.468838828880046e-06, "loss": 1.0643, "step": 15449 }, { "epoch": 1.6439668014471165, "grad_norm": 11.0234375, "learning_rate": 9.468803297328029e-06, "loss": 1.1554, "step": 15450 }, { "epoch": 1.644073207065333, "grad_norm": 12.21875, "learning_rate": 9.46876776577601e-06, "loss": 1.172, "step": 15451 }, { "epoch": 1.6441796126835497, "grad_norm": 12.5, "learning_rate": 9.468732234223992e-06, "loss": 0.6935, "step": 15452 }, { "epoch": 1.6442860183017665, "grad_norm": 12.6640625, "learning_rate": 9.468696702671974e-06, "loss": 1.1868, "step": 15453 }, { "epoch": 1.644392423919983, "grad_norm": 10.78125, "learning_rate": 9.468661171119955e-06, "loss": 0.8612, "step": 15454 }, { "epoch": 1.6444988295381995, "grad_norm": 10.7109375, "learning_rate": 9.468625639567937e-06, "loss": 1.1237, "step": 15455 }, { "epoch": 1.6446052351564162, "grad_norm": 14.5859375, "learning_rate": 9.468590108015919e-06, "loss": 1.1805, "step": 15456 }, { "epoch": 1.644711640774633, "grad_norm": 9.3359375, "learning_rate": 9.4685545764639e-06, "loss": 1.1917, "step": 15457 }, { "epoch": 1.6448180463928495, "grad_norm": 8.75, "learning_rate": 9.468519044911882e-06, "loss": 1.3195, "step": 15458 }, { "epoch": 1.6449244520110662, "grad_norm": 12.0703125, "learning_rate": 9.468483513359864e-06, "loss": 0.915, "step": 15459 }, { "epoch": 1.645030857629283, "grad_norm": 15.625, "learning_rate": 9.468447981807845e-06, "loss": 1.0379, "step": 15460 }, { "epoch": 1.6451372632474994, "grad_norm": 23.828125, "learning_rate": 9.468412450255828e-06, "loss": 1.3478, "step": 15461 }, { "epoch": 1.6452436688657162, "grad_norm": 19.46875, "learning_rate": 9.46837691870381e-06, "loss": 1.1428, "step": 15462 }, { "epoch": 1.645350074483933, "grad_norm": 12.390625, "learning_rate": 9.468341387151792e-06, "loss": 0.9966, "step": 15463 }, { "epoch": 1.6454564801021494, "grad_norm": 16.484375, "learning_rate": 9.468305855599773e-06, "loss": 1.1061, "step": 15464 }, { "epoch": 1.645562885720366, "grad_norm": 13.328125, "learning_rate": 9.468270324047755e-06, "loss": 1.1171, "step": 15465 }, { "epoch": 1.6456692913385826, "grad_norm": 9.078125, "learning_rate": 9.468234792495738e-06, "loss": 0.9564, "step": 15466 }, { "epoch": 1.6457756969567994, "grad_norm": 17.78125, "learning_rate": 9.46819926094372e-06, "loss": 1.1703, "step": 15467 }, { "epoch": 1.6458821025750159, "grad_norm": 10.7265625, "learning_rate": 9.468163729391701e-06, "loss": 1.14, "step": 15468 }, { "epoch": 1.6459885081932326, "grad_norm": 15.453125, "learning_rate": 9.468128197839683e-06, "loss": 1.1156, "step": 15469 }, { "epoch": 1.6460949138114493, "grad_norm": 14.1796875, "learning_rate": 9.468092666287665e-06, "loss": 0.8187, "step": 15470 }, { "epoch": 1.6462013194296659, "grad_norm": 9.1953125, "learning_rate": 9.468057134735646e-06, "loss": 0.9992, "step": 15471 }, { "epoch": 1.6463077250478824, "grad_norm": 11.1640625, "learning_rate": 9.468021603183628e-06, "loss": 0.9779, "step": 15472 }, { "epoch": 1.6464141306660993, "grad_norm": 10.3359375, "learning_rate": 9.46798607163161e-06, "loss": 0.9776, "step": 15473 }, { "epoch": 1.6465205362843158, "grad_norm": 12.3984375, "learning_rate": 9.467950540079591e-06, "loss": 0.8903, "step": 15474 }, { "epoch": 1.6466269419025323, "grad_norm": 16.203125, "learning_rate": 9.467915008527573e-06, "loss": 1.3336, "step": 15475 }, { "epoch": 1.646733347520749, "grad_norm": 14.6171875, "learning_rate": 9.467879476975556e-06, "loss": 0.9296, "step": 15476 }, { "epoch": 1.6468397531389658, "grad_norm": 9.328125, "learning_rate": 9.467843945423538e-06, "loss": 0.8552, "step": 15477 }, { "epoch": 1.6469461587571823, "grad_norm": 11.2265625, "learning_rate": 9.467808413871519e-06, "loss": 0.9006, "step": 15478 }, { "epoch": 1.647052564375399, "grad_norm": 14.765625, "learning_rate": 9.4677728823195e-06, "loss": 0.8621, "step": 15479 }, { "epoch": 1.6471589699936158, "grad_norm": 18.46875, "learning_rate": 9.467737350767482e-06, "loss": 1.246, "step": 15480 }, { "epoch": 1.6472653756118323, "grad_norm": 12.2421875, "learning_rate": 9.467701819215464e-06, "loss": 1.4275, "step": 15481 }, { "epoch": 1.6473717812300488, "grad_norm": 12.9453125, "learning_rate": 9.467666287663446e-06, "loss": 1.0568, "step": 15482 }, { "epoch": 1.6474781868482657, "grad_norm": 10.84375, "learning_rate": 9.467630756111427e-06, "loss": 1.1744, "step": 15483 }, { "epoch": 1.6475845924664823, "grad_norm": 12.625, "learning_rate": 9.467595224559409e-06, "loss": 1.0, "step": 15484 }, { "epoch": 1.6476909980846988, "grad_norm": 13.9375, "learning_rate": 9.46755969300739e-06, "loss": 1.4914, "step": 15485 }, { "epoch": 1.6477974037029155, "grad_norm": 17.640625, "learning_rate": 9.467524161455374e-06, "loss": 0.8654, "step": 15486 }, { "epoch": 1.6479038093211322, "grad_norm": 11.0859375, "learning_rate": 9.467488629903355e-06, "loss": 1.1927, "step": 15487 }, { "epoch": 1.6480102149393487, "grad_norm": 13.359375, "learning_rate": 9.467453098351337e-06, "loss": 1.0314, "step": 15488 }, { "epoch": 1.6481166205575655, "grad_norm": 14.3125, "learning_rate": 9.467417566799318e-06, "loss": 0.8517, "step": 15489 }, { "epoch": 1.6482230261757822, "grad_norm": 11.8125, "learning_rate": 9.4673820352473e-06, "loss": 1.2779, "step": 15490 }, { "epoch": 1.6483294317939987, "grad_norm": 12.46875, "learning_rate": 9.467346503695282e-06, "loss": 1.2062, "step": 15491 }, { "epoch": 1.6484358374122152, "grad_norm": 12.140625, "learning_rate": 9.467310972143265e-06, "loss": 1.2003, "step": 15492 }, { "epoch": 1.648542243030432, "grad_norm": 13.3515625, "learning_rate": 9.467275440591247e-06, "loss": 1.2357, "step": 15493 }, { "epoch": 1.6486486486486487, "grad_norm": 12.625, "learning_rate": 9.467239909039228e-06, "loss": 1.0553, "step": 15494 }, { "epoch": 1.6487550542668652, "grad_norm": 8.671875, "learning_rate": 9.46720437748721e-06, "loss": 0.6809, "step": 15495 }, { "epoch": 1.648861459885082, "grad_norm": 13.046875, "learning_rate": 9.467168845935191e-06, "loss": 1.14, "step": 15496 }, { "epoch": 1.6489678655032987, "grad_norm": 12.0078125, "learning_rate": 9.467133314383173e-06, "loss": 1.0215, "step": 15497 }, { "epoch": 1.6490742711215152, "grad_norm": 14.875, "learning_rate": 9.467097782831155e-06, "loss": 1.1608, "step": 15498 }, { "epoch": 1.649180676739732, "grad_norm": 13.59375, "learning_rate": 9.467062251279136e-06, "loss": 1.165, "step": 15499 }, { "epoch": 1.6492870823579486, "grad_norm": 11.59375, "learning_rate": 9.467026719727118e-06, "loss": 1.2275, "step": 15500 }, { "epoch": 1.6493934879761651, "grad_norm": 10.625, "learning_rate": 9.4669911881751e-06, "loss": 1.1102, "step": 15501 }, { "epoch": 1.6494998935943816, "grad_norm": 13.2421875, "learning_rate": 9.466955656623083e-06, "loss": 1.3028, "step": 15502 }, { "epoch": 1.6496062992125984, "grad_norm": 12.109375, "learning_rate": 9.466920125071064e-06, "loss": 0.8679, "step": 15503 }, { "epoch": 1.6497127048308151, "grad_norm": 11.1875, "learning_rate": 9.466884593519046e-06, "loss": 1.0179, "step": 15504 }, { "epoch": 1.6498191104490316, "grad_norm": 10.9609375, "learning_rate": 9.466849061967028e-06, "loss": 1.1142, "step": 15505 }, { "epoch": 1.6499255160672484, "grad_norm": 14.421875, "learning_rate": 9.46681353041501e-06, "loss": 1.2567, "step": 15506 }, { "epoch": 1.650031921685465, "grad_norm": 14.9921875, "learning_rate": 9.46677799886299e-06, "loss": 1.2359, "step": 15507 }, { "epoch": 1.6501383273036816, "grad_norm": 9.0390625, "learning_rate": 9.466742467310972e-06, "loss": 1.196, "step": 15508 }, { "epoch": 1.6502447329218983, "grad_norm": 8.6484375, "learning_rate": 9.466706935758954e-06, "loss": 0.9896, "step": 15509 }, { "epoch": 1.650351138540115, "grad_norm": 13.75, "learning_rate": 9.466671404206936e-06, "loss": 1.0425, "step": 15510 }, { "epoch": 1.6504575441583316, "grad_norm": 11.8984375, "learning_rate": 9.466635872654917e-06, "loss": 1.4557, "step": 15511 }, { "epoch": 1.650563949776548, "grad_norm": 12.125, "learning_rate": 9.4666003411029e-06, "loss": 1.0932, "step": 15512 }, { "epoch": 1.6506703553947648, "grad_norm": 14.4765625, "learning_rate": 9.466564809550882e-06, "loss": 1.4027, "step": 15513 }, { "epoch": 1.6507767610129815, "grad_norm": 17.265625, "learning_rate": 9.466529277998864e-06, "loss": 1.2792, "step": 15514 }, { "epoch": 1.650883166631198, "grad_norm": 13.2421875, "learning_rate": 9.466493746446845e-06, "loss": 0.9149, "step": 15515 }, { "epoch": 1.6509895722494148, "grad_norm": 11.84375, "learning_rate": 9.466458214894827e-06, "loss": 1.0919, "step": 15516 }, { "epoch": 1.6510959778676315, "grad_norm": 11.3515625, "learning_rate": 9.46642268334281e-06, "loss": 0.7556, "step": 15517 }, { "epoch": 1.651202383485848, "grad_norm": 14.796875, "learning_rate": 9.466387151790792e-06, "loss": 1.2241, "step": 15518 }, { "epoch": 1.6513087891040645, "grad_norm": 11.0234375, "learning_rate": 9.466351620238773e-06, "loss": 1.0712, "step": 15519 }, { "epoch": 1.6514151947222815, "grad_norm": 11.6796875, "learning_rate": 9.466316088686755e-06, "loss": 0.9361, "step": 15520 }, { "epoch": 1.651521600340498, "grad_norm": 9.8125, "learning_rate": 9.466280557134737e-06, "loss": 1.05, "step": 15521 }, { "epoch": 1.6516280059587145, "grad_norm": 10.328125, "learning_rate": 9.466245025582718e-06, "loss": 1.2007, "step": 15522 }, { "epoch": 1.6517344115769312, "grad_norm": 16.9375, "learning_rate": 9.4662094940307e-06, "loss": 1.0178, "step": 15523 }, { "epoch": 1.651840817195148, "grad_norm": 13.2265625, "learning_rate": 9.466173962478681e-06, "loss": 0.9323, "step": 15524 }, { "epoch": 1.6519472228133645, "grad_norm": 13.53125, "learning_rate": 9.466138430926663e-06, "loss": 0.8534, "step": 15525 }, { "epoch": 1.6520536284315812, "grad_norm": 18.375, "learning_rate": 9.466102899374645e-06, "loss": 1.0337, "step": 15526 }, { "epoch": 1.652160034049798, "grad_norm": 22.796875, "learning_rate": 9.466067367822628e-06, "loss": 1.1179, "step": 15527 }, { "epoch": 1.6522664396680145, "grad_norm": 10.6640625, "learning_rate": 9.46603183627061e-06, "loss": 1.1719, "step": 15528 }, { "epoch": 1.652372845286231, "grad_norm": 10.921875, "learning_rate": 9.465996304718591e-06, "loss": 0.9003, "step": 15529 }, { "epoch": 1.652479250904448, "grad_norm": 9.671875, "learning_rate": 9.465960773166573e-06, "loss": 0.8714, "step": 15530 }, { "epoch": 1.6525856565226644, "grad_norm": 16.6875, "learning_rate": 9.465925241614554e-06, "loss": 1.4154, "step": 15531 }, { "epoch": 1.652692062140881, "grad_norm": 12.625, "learning_rate": 9.465889710062536e-06, "loss": 1.5973, "step": 15532 }, { "epoch": 1.6527984677590977, "grad_norm": 10.71875, "learning_rate": 9.465854178510518e-06, "loss": 1.1442, "step": 15533 }, { "epoch": 1.6529048733773144, "grad_norm": 11.4296875, "learning_rate": 9.4658186469585e-06, "loss": 0.9738, "step": 15534 }, { "epoch": 1.653011278995531, "grad_norm": 11.3828125, "learning_rate": 9.465783115406481e-06, "loss": 1.0449, "step": 15535 }, { "epoch": 1.6531176846137476, "grad_norm": 11.2109375, "learning_rate": 9.465747583854462e-06, "loss": 1.0064, "step": 15536 }, { "epoch": 1.6532240902319644, "grad_norm": 10.28125, "learning_rate": 9.465712052302444e-06, "loss": 1.2297, "step": 15537 }, { "epoch": 1.6533304958501809, "grad_norm": 11.3125, "learning_rate": 9.465676520750427e-06, "loss": 0.9326, "step": 15538 }, { "epoch": 1.6534369014683974, "grad_norm": 11.1484375, "learning_rate": 9.465640989198409e-06, "loss": 1.1258, "step": 15539 }, { "epoch": 1.6535433070866141, "grad_norm": 31.671875, "learning_rate": 9.46560545764639e-06, "loss": 0.7264, "step": 15540 }, { "epoch": 1.6536497127048309, "grad_norm": 12.359375, "learning_rate": 9.465569926094372e-06, "loss": 1.1657, "step": 15541 }, { "epoch": 1.6537561183230474, "grad_norm": 21.28125, "learning_rate": 9.465534394542354e-06, "loss": 1.0043, "step": 15542 }, { "epoch": 1.653862523941264, "grad_norm": 12.0078125, "learning_rate": 9.465498862990337e-06, "loss": 0.9826, "step": 15543 }, { "epoch": 1.6539689295594808, "grad_norm": 11.734375, "learning_rate": 9.465463331438319e-06, "loss": 1.1358, "step": 15544 }, { "epoch": 1.6540753351776973, "grad_norm": 8.46875, "learning_rate": 9.4654277998863e-06, "loss": 0.3654, "step": 15545 }, { "epoch": 1.654181740795914, "grad_norm": 15.71875, "learning_rate": 9.465392268334282e-06, "loss": 1.2892, "step": 15546 }, { "epoch": 1.6542881464141308, "grad_norm": 15.5078125, "learning_rate": 9.465356736782264e-06, "loss": 1.0896, "step": 15547 }, { "epoch": 1.6543945520323473, "grad_norm": 29.3125, "learning_rate": 9.465321205230245e-06, "loss": 1.0241, "step": 15548 }, { "epoch": 1.6545009576505638, "grad_norm": 11.7421875, "learning_rate": 9.465285673678227e-06, "loss": 1.2421, "step": 15549 }, { "epoch": 1.6546073632687806, "grad_norm": 13.1796875, "learning_rate": 9.465250142126208e-06, "loss": 1.1303, "step": 15550 }, { "epoch": 1.6547137688869973, "grad_norm": 11.8671875, "learning_rate": 9.46521461057419e-06, "loss": 1.0617, "step": 15551 }, { "epoch": 1.6548201745052138, "grad_norm": 13.9375, "learning_rate": 9.465179079022172e-06, "loss": 1.2226, "step": 15552 }, { "epoch": 1.6549265801234305, "grad_norm": 11.75, "learning_rate": 9.465143547470155e-06, "loss": 1.2815, "step": 15553 }, { "epoch": 1.6550329857416473, "grad_norm": 12.875, "learning_rate": 9.465108015918136e-06, "loss": 1.258, "step": 15554 }, { "epoch": 1.6551393913598638, "grad_norm": 11.8984375, "learning_rate": 9.465072484366118e-06, "loss": 0.9563, "step": 15555 }, { "epoch": 1.6552457969780805, "grad_norm": 15.71875, "learning_rate": 9.4650369528141e-06, "loss": 1.214, "step": 15556 }, { "epoch": 1.6553522025962972, "grad_norm": 13.6171875, "learning_rate": 9.465001421262081e-06, "loss": 1.0325, "step": 15557 }, { "epoch": 1.6554586082145137, "grad_norm": 18.125, "learning_rate": 9.464965889710065e-06, "loss": 1.3452, "step": 15558 }, { "epoch": 1.6555650138327302, "grad_norm": 11.671875, "learning_rate": 9.464930358158046e-06, "loss": 1.0114, "step": 15559 }, { "epoch": 1.655671419450947, "grad_norm": 13.59375, "learning_rate": 9.464894826606028e-06, "loss": 0.9231, "step": 15560 }, { "epoch": 1.6557778250691637, "grad_norm": 13.1171875, "learning_rate": 9.46485929505401e-06, "loss": 1.0983, "step": 15561 }, { "epoch": 1.6558842306873802, "grad_norm": 12.3671875, "learning_rate": 9.464823763501991e-06, "loss": 0.974, "step": 15562 }, { "epoch": 1.655990636305597, "grad_norm": 10.109375, "learning_rate": 9.464788231949973e-06, "loss": 0.9479, "step": 15563 }, { "epoch": 1.6560970419238137, "grad_norm": 10.546875, "learning_rate": 9.464752700397954e-06, "loss": 0.948, "step": 15564 }, { "epoch": 1.6562034475420302, "grad_norm": 17.640625, "learning_rate": 9.464717168845936e-06, "loss": 1.168, "step": 15565 }, { "epoch": 1.6563098531602467, "grad_norm": 17.671875, "learning_rate": 9.464681637293917e-06, "loss": 1.0028, "step": 15566 }, { "epoch": 1.6564162587784637, "grad_norm": 20.3125, "learning_rate": 9.464646105741899e-06, "loss": 0.9229, "step": 15567 }, { "epoch": 1.6565226643966802, "grad_norm": 10.1015625, "learning_rate": 9.464610574189882e-06, "loss": 1.0684, "step": 15568 }, { "epoch": 1.6566290700148967, "grad_norm": 14.390625, "learning_rate": 9.464575042637864e-06, "loss": 1.0703, "step": 15569 }, { "epoch": 1.6567354756331134, "grad_norm": 13.0, "learning_rate": 9.464539511085846e-06, "loss": 0.8563, "step": 15570 }, { "epoch": 1.6568418812513301, "grad_norm": 11.625, "learning_rate": 9.464503979533827e-06, "loss": 1.2588, "step": 15571 }, { "epoch": 1.6569482868695466, "grad_norm": 12.625, "learning_rate": 9.464468447981809e-06, "loss": 1.3174, "step": 15572 }, { "epoch": 1.6570546924877634, "grad_norm": 14.328125, "learning_rate": 9.46443291642979e-06, "loss": 1.1556, "step": 15573 }, { "epoch": 1.6571610981059801, "grad_norm": 10.828125, "learning_rate": 9.464397384877772e-06, "loss": 1.1088, "step": 15574 }, { "epoch": 1.6572675037241966, "grad_norm": 29.46875, "learning_rate": 9.464361853325754e-06, "loss": 1.1898, "step": 15575 }, { "epoch": 1.6573739093424131, "grad_norm": 14.15625, "learning_rate": 9.464326321773735e-06, "loss": 1.2029, "step": 15576 }, { "epoch": 1.65748031496063, "grad_norm": 18.265625, "learning_rate": 9.464290790221717e-06, "loss": 1.3763, "step": 15577 }, { "epoch": 1.6575867205788466, "grad_norm": 12.0, "learning_rate": 9.464255258669698e-06, "loss": 1.1853, "step": 15578 }, { "epoch": 1.657693126197063, "grad_norm": 12.2109375, "learning_rate": 9.464219727117682e-06, "loss": 1.0653, "step": 15579 }, { "epoch": 1.6577995318152798, "grad_norm": 15.3359375, "learning_rate": 9.464184195565663e-06, "loss": 1.0727, "step": 15580 }, { "epoch": 1.6579059374334966, "grad_norm": 13.4765625, "learning_rate": 9.464148664013645e-06, "loss": 0.9751, "step": 15581 }, { "epoch": 1.658012343051713, "grad_norm": 12.8671875, "learning_rate": 9.464113132461627e-06, "loss": 1.2508, "step": 15582 }, { "epoch": 1.6581187486699298, "grad_norm": 10.6796875, "learning_rate": 9.464077600909608e-06, "loss": 1.3183, "step": 15583 }, { "epoch": 1.6582251542881465, "grad_norm": 11.78125, "learning_rate": 9.464042069357591e-06, "loss": 1.1706, "step": 15584 }, { "epoch": 1.658331559906363, "grad_norm": 11.46875, "learning_rate": 9.464006537805573e-06, "loss": 1.0598, "step": 15585 }, { "epoch": 1.6584379655245796, "grad_norm": 19.546875, "learning_rate": 9.463971006253555e-06, "loss": 0.769, "step": 15586 }, { "epoch": 1.6585443711427963, "grad_norm": 13.6875, "learning_rate": 9.463935474701536e-06, "loss": 0.9472, "step": 15587 }, { "epoch": 1.658650776761013, "grad_norm": 14.9375, "learning_rate": 9.463899943149518e-06, "loss": 1.4817, "step": 15588 }, { "epoch": 1.6587571823792295, "grad_norm": 9.828125, "learning_rate": 9.4638644115975e-06, "loss": 1.0938, "step": 15589 }, { "epoch": 1.6588635879974463, "grad_norm": 17.734375, "learning_rate": 9.463828880045481e-06, "loss": 1.1105, "step": 15590 }, { "epoch": 1.658969993615663, "grad_norm": 14.3125, "learning_rate": 9.463793348493463e-06, "loss": 1.032, "step": 15591 }, { "epoch": 1.6590763992338795, "grad_norm": 9.59375, "learning_rate": 9.463757816941444e-06, "loss": 1.0107, "step": 15592 }, { "epoch": 1.6591828048520962, "grad_norm": 23.234375, "learning_rate": 9.463722285389426e-06, "loss": 1.1613, "step": 15593 }, { "epoch": 1.659289210470313, "grad_norm": 11.0703125, "learning_rate": 9.46368675383741e-06, "loss": 1.1152, "step": 15594 }, { "epoch": 1.6593956160885295, "grad_norm": 12.9296875, "learning_rate": 9.46365122228539e-06, "loss": 0.9794, "step": 15595 }, { "epoch": 1.659502021706746, "grad_norm": 14.15625, "learning_rate": 9.463615690733372e-06, "loss": 1.2032, "step": 15596 }, { "epoch": 1.6596084273249627, "grad_norm": 12.1796875, "learning_rate": 9.463580159181354e-06, "loss": 1.1998, "step": 15597 }, { "epoch": 1.6597148329431795, "grad_norm": 10.75, "learning_rate": 9.463544627629336e-06, "loss": 1.2326, "step": 15598 }, { "epoch": 1.659821238561396, "grad_norm": 13.5625, "learning_rate": 9.463509096077317e-06, "loss": 1.2361, "step": 15599 }, { "epoch": 1.6599276441796127, "grad_norm": 14.59375, "learning_rate": 9.463473564525299e-06, "loss": 1.3427, "step": 15600 }, { "epoch": 1.6600340497978294, "grad_norm": 14.7421875, "learning_rate": 9.46343803297328e-06, "loss": 1.1074, "step": 15601 }, { "epoch": 1.660140455416046, "grad_norm": 12.8046875, "learning_rate": 9.463402501421262e-06, "loss": 0.9711, "step": 15602 }, { "epoch": 1.6602468610342624, "grad_norm": 11.8984375, "learning_rate": 9.463366969869244e-06, "loss": 0.9295, "step": 15603 }, { "epoch": 1.6603532666524794, "grad_norm": 11.2265625, "learning_rate": 9.463331438317227e-06, "loss": 1.0612, "step": 15604 }, { "epoch": 1.660459672270696, "grad_norm": 13.6796875, "learning_rate": 9.463295906765209e-06, "loss": 0.9745, "step": 15605 }, { "epoch": 1.6605660778889124, "grad_norm": 12.1953125, "learning_rate": 9.46326037521319e-06, "loss": 1.067, "step": 15606 }, { "epoch": 1.6606724835071291, "grad_norm": 10.7265625, "learning_rate": 9.463224843661172e-06, "loss": 1.222, "step": 15607 }, { "epoch": 1.6607788891253459, "grad_norm": 14.4140625, "learning_rate": 9.463189312109153e-06, "loss": 1.0124, "step": 15608 }, { "epoch": 1.6608852947435624, "grad_norm": 8.9921875, "learning_rate": 9.463153780557135e-06, "loss": 1.1708, "step": 15609 }, { "epoch": 1.6609917003617791, "grad_norm": 12.0078125, "learning_rate": 9.463118249005118e-06, "loss": 1.026, "step": 15610 }, { "epoch": 1.6610981059799959, "grad_norm": 8.203125, "learning_rate": 9.4630827174531e-06, "loss": 1.0132, "step": 15611 }, { "epoch": 1.6612045115982124, "grad_norm": 15.234375, "learning_rate": 9.463047185901082e-06, "loss": 1.1137, "step": 15612 }, { "epoch": 1.6613109172164289, "grad_norm": 12.71875, "learning_rate": 9.463011654349063e-06, "loss": 1.206, "step": 15613 }, { "epoch": 1.6614173228346458, "grad_norm": 12.4609375, "learning_rate": 9.462976122797045e-06, "loss": 1.102, "step": 15614 }, { "epoch": 1.6615237284528623, "grad_norm": 11.53125, "learning_rate": 9.462940591245026e-06, "loss": 1.1168, "step": 15615 }, { "epoch": 1.6616301340710788, "grad_norm": 12.6796875, "learning_rate": 9.462905059693008e-06, "loss": 1.0866, "step": 15616 }, { "epoch": 1.6617365396892956, "grad_norm": 15.0859375, "learning_rate": 9.46286952814099e-06, "loss": 1.0209, "step": 15617 }, { "epoch": 1.6618429453075123, "grad_norm": 9.71875, "learning_rate": 9.462833996588971e-06, "loss": 1.1979, "step": 15618 }, { "epoch": 1.6619493509257288, "grad_norm": 17.890625, "learning_rate": 9.462798465036953e-06, "loss": 1.0697, "step": 15619 }, { "epoch": 1.6620557565439456, "grad_norm": 11.53125, "learning_rate": 9.462762933484936e-06, "loss": 1.2336, "step": 15620 }, { "epoch": 1.6621621621621623, "grad_norm": 9.3828125, "learning_rate": 9.462727401932918e-06, "loss": 1.0062, "step": 15621 }, { "epoch": 1.6622685677803788, "grad_norm": 14.3671875, "learning_rate": 9.4626918703809e-06, "loss": 1.2054, "step": 15622 }, { "epoch": 1.6623749733985953, "grad_norm": 13.15625, "learning_rate": 9.462656338828881e-06, "loss": 1.0295, "step": 15623 }, { "epoch": 1.662481379016812, "grad_norm": 12.078125, "learning_rate": 9.462620807276862e-06, "loss": 1.1363, "step": 15624 }, { "epoch": 1.6625877846350288, "grad_norm": 10.875, "learning_rate": 9.462585275724844e-06, "loss": 0.9004, "step": 15625 }, { "epoch": 1.6626941902532453, "grad_norm": 11.4296875, "learning_rate": 9.462549744172826e-06, "loss": 1.0348, "step": 15626 }, { "epoch": 1.662800595871462, "grad_norm": 15.078125, "learning_rate": 9.462514212620807e-06, "loss": 1.1144, "step": 15627 }, { "epoch": 1.6629070014896787, "grad_norm": 11.875, "learning_rate": 9.462478681068789e-06, "loss": 1.0647, "step": 15628 }, { "epoch": 1.6630134071078952, "grad_norm": 11.3359375, "learning_rate": 9.46244314951677e-06, "loss": 1.1742, "step": 15629 }, { "epoch": 1.663119812726112, "grad_norm": 11.828125, "learning_rate": 9.462407617964754e-06, "loss": 0.859, "step": 15630 }, { "epoch": 1.6632262183443287, "grad_norm": 17.375, "learning_rate": 9.462372086412735e-06, "loss": 1.1022, "step": 15631 }, { "epoch": 1.6633326239625452, "grad_norm": 13.171875, "learning_rate": 9.462336554860717e-06, "loss": 1.0519, "step": 15632 }, { "epoch": 1.6634390295807617, "grad_norm": 12.1640625, "learning_rate": 9.462301023308699e-06, "loss": 0.9365, "step": 15633 }, { "epoch": 1.6635454351989785, "grad_norm": 10.921875, "learning_rate": 9.46226549175668e-06, "loss": 1.1457, "step": 15634 }, { "epoch": 1.6636518408171952, "grad_norm": 11.4375, "learning_rate": 9.462229960204664e-06, "loss": 0.8807, "step": 15635 }, { "epoch": 1.6637582464354117, "grad_norm": 12.9765625, "learning_rate": 9.462194428652645e-06, "loss": 1.2316, "step": 15636 }, { "epoch": 1.6638646520536284, "grad_norm": 16.859375, "learning_rate": 9.462158897100627e-06, "loss": 1.0125, "step": 15637 }, { "epoch": 1.6639710576718452, "grad_norm": 15.7734375, "learning_rate": 9.462123365548608e-06, "loss": 1.0664, "step": 15638 }, { "epoch": 1.6640774632900617, "grad_norm": 10.765625, "learning_rate": 9.46208783399659e-06, "loss": 0.8853, "step": 15639 }, { "epoch": 1.6641838689082784, "grad_norm": 12.9921875, "learning_rate": 9.462052302444572e-06, "loss": 0.9228, "step": 15640 }, { "epoch": 1.6642902745264951, "grad_norm": 20.25, "learning_rate": 9.462016770892553e-06, "loss": 1.1455, "step": 15641 }, { "epoch": 1.6643966801447116, "grad_norm": 10.5546875, "learning_rate": 9.461981239340535e-06, "loss": 1.1181, "step": 15642 }, { "epoch": 1.6645030857629282, "grad_norm": 11.5, "learning_rate": 9.461945707788516e-06, "loss": 1.2141, "step": 15643 }, { "epoch": 1.664609491381145, "grad_norm": 10.4140625, "learning_rate": 9.461910176236498e-06, "loss": 1.162, "step": 15644 }, { "epoch": 1.6647158969993616, "grad_norm": 14.8828125, "learning_rate": 9.461874644684481e-06, "loss": 1.328, "step": 15645 }, { "epoch": 1.6648223026175781, "grad_norm": 10.2265625, "learning_rate": 9.461839113132463e-06, "loss": 0.9661, "step": 15646 }, { "epoch": 1.6649287082357949, "grad_norm": 13.125, "learning_rate": 9.461803581580445e-06, "loss": 1.267, "step": 15647 }, { "epoch": 1.6650351138540116, "grad_norm": 12.28125, "learning_rate": 9.461768050028426e-06, "loss": 1.3788, "step": 15648 }, { "epoch": 1.665141519472228, "grad_norm": 12.71875, "learning_rate": 9.461732518476408e-06, "loss": 0.946, "step": 15649 }, { "epoch": 1.6652479250904446, "grad_norm": 15.84375, "learning_rate": 9.46169698692439e-06, "loss": 1.006, "step": 15650 }, { "epoch": 1.6653543307086616, "grad_norm": 14.9296875, "learning_rate": 9.461661455372371e-06, "loss": 0.9017, "step": 15651 }, { "epoch": 1.665460736326878, "grad_norm": 12.1875, "learning_rate": 9.461625923820353e-06, "loss": 0.9681, "step": 15652 }, { "epoch": 1.6655671419450946, "grad_norm": 10.5625, "learning_rate": 9.461590392268334e-06, "loss": 1.1582, "step": 15653 }, { "epoch": 1.6656735475633113, "grad_norm": 14.078125, "learning_rate": 9.461554860716316e-06, "loss": 1.0918, "step": 15654 }, { "epoch": 1.665779953181528, "grad_norm": 14.1953125, "learning_rate": 9.461519329164297e-06, "loss": 1.515, "step": 15655 }, { "epoch": 1.6658863587997446, "grad_norm": 15.328125, "learning_rate": 9.46148379761228e-06, "loss": 0.8526, "step": 15656 }, { "epoch": 1.6659927644179613, "grad_norm": 15.328125, "learning_rate": 9.461448266060262e-06, "loss": 1.1075, "step": 15657 }, { "epoch": 1.666099170036178, "grad_norm": 13.2734375, "learning_rate": 9.461412734508244e-06, "loss": 1.1746, "step": 15658 }, { "epoch": 1.6662055756543945, "grad_norm": 11.375, "learning_rate": 9.461377202956225e-06, "loss": 1.0956, "step": 15659 }, { "epoch": 1.666311981272611, "grad_norm": 11.5703125, "learning_rate": 9.461341671404207e-06, "loss": 1.0521, "step": 15660 }, { "epoch": 1.666418386890828, "grad_norm": 13.1953125, "learning_rate": 9.46130613985219e-06, "loss": 0.9678, "step": 15661 }, { "epoch": 1.6665247925090445, "grad_norm": 13.015625, "learning_rate": 9.461270608300172e-06, "loss": 1.3569, "step": 15662 }, { "epoch": 1.666631198127261, "grad_norm": 15.3984375, "learning_rate": 9.461235076748154e-06, "loss": 1.3771, "step": 15663 }, { "epoch": 1.6667376037454777, "grad_norm": 14.484375, "learning_rate": 9.461199545196135e-06, "loss": 0.9096, "step": 15664 }, { "epoch": 1.6668440093636945, "grad_norm": 13.4140625, "learning_rate": 9.461164013644117e-06, "loss": 1.2011, "step": 15665 }, { "epoch": 1.666950414981911, "grad_norm": 20.90625, "learning_rate": 9.461128482092098e-06, "loss": 0.944, "step": 15666 }, { "epoch": 1.6670568206001277, "grad_norm": 12.484375, "learning_rate": 9.46109295054008e-06, "loss": 1.0679, "step": 15667 }, { "epoch": 1.6671632262183445, "grad_norm": 11.9453125, "learning_rate": 9.461057418988062e-06, "loss": 1.125, "step": 15668 }, { "epoch": 1.667269631836561, "grad_norm": 13.4375, "learning_rate": 9.461021887436043e-06, "loss": 1.0707, "step": 15669 }, { "epoch": 1.6673760374547775, "grad_norm": 9.765625, "learning_rate": 9.460986355884025e-06, "loss": 0.9518, "step": 15670 }, { "epoch": 1.6674824430729942, "grad_norm": 10.453125, "learning_rate": 9.460950824332008e-06, "loss": 1.2395, "step": 15671 }, { "epoch": 1.667588848691211, "grad_norm": 18.875, "learning_rate": 9.46091529277999e-06, "loss": 0.9097, "step": 15672 }, { "epoch": 1.6676952543094274, "grad_norm": 13.84375, "learning_rate": 9.460879761227971e-06, "loss": 1.1662, "step": 15673 }, { "epoch": 1.6678016599276442, "grad_norm": 11.5859375, "learning_rate": 9.460844229675953e-06, "loss": 1.0356, "step": 15674 }, { "epoch": 1.667908065545861, "grad_norm": 10.6953125, "learning_rate": 9.460808698123935e-06, "loss": 1.0692, "step": 15675 }, { "epoch": 1.6680144711640774, "grad_norm": 10.578125, "learning_rate": 9.460773166571918e-06, "loss": 1.1433, "step": 15676 }, { "epoch": 1.6681208767822941, "grad_norm": 12.484375, "learning_rate": 9.4607376350199e-06, "loss": 1.2686, "step": 15677 }, { "epoch": 1.6682272824005109, "grad_norm": 11.65625, "learning_rate": 9.460702103467881e-06, "loss": 1.0433, "step": 15678 }, { "epoch": 1.6683336880187274, "grad_norm": 11.796875, "learning_rate": 9.460666571915863e-06, "loss": 1.0412, "step": 15679 }, { "epoch": 1.668440093636944, "grad_norm": 16.453125, "learning_rate": 9.460631040363844e-06, "loss": 1.6844, "step": 15680 }, { "epoch": 1.6685464992551606, "grad_norm": 11.1640625, "learning_rate": 9.460595508811826e-06, "loss": 1.0673, "step": 15681 }, { "epoch": 1.6686529048733774, "grad_norm": 10.6171875, "learning_rate": 9.460559977259808e-06, "loss": 0.9016, "step": 15682 }, { "epoch": 1.6687593104915939, "grad_norm": 15.9375, "learning_rate": 9.460524445707789e-06, "loss": 1.0425, "step": 15683 }, { "epoch": 1.6688657161098106, "grad_norm": 12.203125, "learning_rate": 9.46048891415577e-06, "loss": 1.2722, "step": 15684 }, { "epoch": 1.6689721217280273, "grad_norm": 10.6640625, "learning_rate": 9.460453382603752e-06, "loss": 1.0945, "step": 15685 }, { "epoch": 1.6690785273462438, "grad_norm": 9.515625, "learning_rate": 9.460417851051736e-06, "loss": 1.0023, "step": 15686 }, { "epoch": 1.6691849329644606, "grad_norm": 21.40625, "learning_rate": 9.460382319499717e-06, "loss": 1.0998, "step": 15687 }, { "epoch": 1.6692913385826773, "grad_norm": 19.8125, "learning_rate": 9.460346787947699e-06, "loss": 1.0302, "step": 15688 }, { "epoch": 1.6693977442008938, "grad_norm": 12.859375, "learning_rate": 9.46031125639568e-06, "loss": 1.22, "step": 15689 }, { "epoch": 1.6695041498191103, "grad_norm": 10.421875, "learning_rate": 9.460275724843662e-06, "loss": 1.0843, "step": 15690 }, { "epoch": 1.669610555437327, "grad_norm": 10.8671875, "learning_rate": 9.460240193291644e-06, "loss": 1.0617, "step": 15691 }, { "epoch": 1.6697169610555438, "grad_norm": 10.0625, "learning_rate": 9.460204661739625e-06, "loss": 1.2982, "step": 15692 }, { "epoch": 1.6698233666737603, "grad_norm": 13.4140625, "learning_rate": 9.460169130187607e-06, "loss": 1.2807, "step": 15693 }, { "epoch": 1.669929772291977, "grad_norm": 10.640625, "learning_rate": 9.460133598635589e-06, "loss": 1.012, "step": 15694 }, { "epoch": 1.6700361779101938, "grad_norm": 10.65625, "learning_rate": 9.46009806708357e-06, "loss": 0.9719, "step": 15695 }, { "epoch": 1.6701425835284103, "grad_norm": 16.015625, "learning_rate": 9.460062535531552e-06, "loss": 0.8537, "step": 15696 }, { "epoch": 1.6702489891466268, "grad_norm": 15.4140625, "learning_rate": 9.460027003979535e-06, "loss": 1.0425, "step": 15697 }, { "epoch": 1.6703553947648437, "grad_norm": 11.1875, "learning_rate": 9.459991472427517e-06, "loss": 1.0096, "step": 15698 }, { "epoch": 1.6704618003830602, "grad_norm": 11.609375, "learning_rate": 9.459955940875498e-06, "loss": 1.3404, "step": 15699 }, { "epoch": 1.6705682060012768, "grad_norm": 12.7734375, "learning_rate": 9.45992040932348e-06, "loss": 0.9187, "step": 15700 }, { "epoch": 1.6706746116194935, "grad_norm": 10.6640625, "learning_rate": 9.459884877771461e-06, "loss": 1.0817, "step": 15701 }, { "epoch": 1.6707810172377102, "grad_norm": 13.8515625, "learning_rate": 9.459849346219445e-06, "loss": 1.1457, "step": 15702 }, { "epoch": 1.6708874228559267, "grad_norm": 13.5, "learning_rate": 9.459813814667426e-06, "loss": 1.0133, "step": 15703 }, { "epoch": 1.6709938284741435, "grad_norm": 11.265625, "learning_rate": 9.459778283115408e-06, "loss": 0.8199, "step": 15704 }, { "epoch": 1.6711002340923602, "grad_norm": 15.0234375, "learning_rate": 9.45974275156339e-06, "loss": 0.9489, "step": 15705 }, { "epoch": 1.6712066397105767, "grad_norm": 12.703125, "learning_rate": 9.459707220011371e-06, "loss": 1.0442, "step": 15706 }, { "epoch": 1.6713130453287932, "grad_norm": 18.25, "learning_rate": 9.459671688459353e-06, "loss": 1.1098, "step": 15707 }, { "epoch": 1.6714194509470102, "grad_norm": 13.984375, "learning_rate": 9.459636156907334e-06, "loss": 0.9001, "step": 15708 }, { "epoch": 1.6715258565652267, "grad_norm": 13.3203125, "learning_rate": 9.459600625355316e-06, "loss": 0.9559, "step": 15709 }, { "epoch": 1.6716322621834432, "grad_norm": 17.53125, "learning_rate": 9.459565093803298e-06, "loss": 0.9179, "step": 15710 }, { "epoch": 1.67173866780166, "grad_norm": 10.1796875, "learning_rate": 9.45952956225128e-06, "loss": 1.3184, "step": 15711 }, { "epoch": 1.6718450734198766, "grad_norm": 12.5234375, "learning_rate": 9.459494030699263e-06, "loss": 1.0722, "step": 15712 }, { "epoch": 1.6719514790380932, "grad_norm": 24.046875, "learning_rate": 9.459458499147244e-06, "loss": 1.2481, "step": 15713 }, { "epoch": 1.67205788465631, "grad_norm": 10.671875, "learning_rate": 9.459422967595226e-06, "loss": 1.3134, "step": 15714 }, { "epoch": 1.6721642902745266, "grad_norm": 13.765625, "learning_rate": 9.459387436043207e-06, "loss": 1.256, "step": 15715 }, { "epoch": 1.6722706958927431, "grad_norm": 16.546875, "learning_rate": 9.459351904491189e-06, "loss": 1.0112, "step": 15716 }, { "epoch": 1.6723771015109596, "grad_norm": 12.8359375, "learning_rate": 9.45931637293917e-06, "loss": 1.38, "step": 15717 }, { "epoch": 1.6724835071291764, "grad_norm": 13.328125, "learning_rate": 9.459280841387152e-06, "loss": 1.4161, "step": 15718 }, { "epoch": 1.672589912747393, "grad_norm": 12.5, "learning_rate": 9.459245309835134e-06, "loss": 1.0201, "step": 15719 }, { "epoch": 1.6726963183656096, "grad_norm": 12.7578125, "learning_rate": 9.459209778283115e-06, "loss": 1.0154, "step": 15720 }, { "epoch": 1.6728027239838263, "grad_norm": 11.9453125, "learning_rate": 9.459174246731097e-06, "loss": 0.9501, "step": 15721 }, { "epoch": 1.672909129602043, "grad_norm": 13.6484375, "learning_rate": 9.45913871517908e-06, "loss": 1.3855, "step": 15722 }, { "epoch": 1.6730155352202596, "grad_norm": 11.5625, "learning_rate": 9.459103183627062e-06, "loss": 1.1012, "step": 15723 }, { "epoch": 1.6731219408384763, "grad_norm": 11.1875, "learning_rate": 9.459067652075043e-06, "loss": 0.9852, "step": 15724 }, { "epoch": 1.673228346456693, "grad_norm": 12.578125, "learning_rate": 9.459032120523025e-06, "loss": 0.9674, "step": 15725 }, { "epoch": 1.6733347520749096, "grad_norm": 11.75, "learning_rate": 9.458996588971007e-06, "loss": 0.8827, "step": 15726 }, { "epoch": 1.673441157693126, "grad_norm": 15.4609375, "learning_rate": 9.458961057418988e-06, "loss": 0.9783, "step": 15727 }, { "epoch": 1.6735475633113428, "grad_norm": 14.9921875, "learning_rate": 9.458925525866972e-06, "loss": 1.5453, "step": 15728 }, { "epoch": 1.6736539689295595, "grad_norm": 10.21875, "learning_rate": 9.458889994314953e-06, "loss": 0.9161, "step": 15729 }, { "epoch": 1.673760374547776, "grad_norm": 16.9375, "learning_rate": 9.458854462762935e-06, "loss": 1.2837, "step": 15730 }, { "epoch": 1.6738667801659928, "grad_norm": 19.3125, "learning_rate": 9.458818931210916e-06, "loss": 0.9933, "step": 15731 }, { "epoch": 1.6739731857842095, "grad_norm": 11.2421875, "learning_rate": 9.458783399658898e-06, "loss": 0.9072, "step": 15732 }, { "epoch": 1.674079591402426, "grad_norm": 11.9453125, "learning_rate": 9.45874786810688e-06, "loss": 1.1107, "step": 15733 }, { "epoch": 1.6741859970206425, "grad_norm": 11.8046875, "learning_rate": 9.458712336554861e-06, "loss": 1.0922, "step": 15734 }, { "epoch": 1.6742924026388595, "grad_norm": 10.015625, "learning_rate": 9.458676805002843e-06, "loss": 0.8809, "step": 15735 }, { "epoch": 1.674398808257076, "grad_norm": 11.2578125, "learning_rate": 9.458641273450824e-06, "loss": 0.9797, "step": 15736 }, { "epoch": 1.6745052138752925, "grad_norm": 15.5078125, "learning_rate": 9.458605741898806e-06, "loss": 1.1585, "step": 15737 }, { "epoch": 1.6746116194935092, "grad_norm": 13.09375, "learning_rate": 9.45857021034679e-06, "loss": 1.2913, "step": 15738 }, { "epoch": 1.674718025111726, "grad_norm": 12.3828125, "learning_rate": 9.458534678794771e-06, "loss": 1.1236, "step": 15739 }, { "epoch": 1.6748244307299425, "grad_norm": 14.59375, "learning_rate": 9.458499147242753e-06, "loss": 0.937, "step": 15740 }, { "epoch": 1.6749308363481592, "grad_norm": 9.890625, "learning_rate": 9.458463615690734e-06, "loss": 0.9802, "step": 15741 }, { "epoch": 1.675037241966376, "grad_norm": 14.5859375, "learning_rate": 9.458428084138716e-06, "loss": 1.1029, "step": 15742 }, { "epoch": 1.6751436475845924, "grad_norm": 13.9921875, "learning_rate": 9.458392552586697e-06, "loss": 1.3437, "step": 15743 }, { "epoch": 1.675250053202809, "grad_norm": 10.390625, "learning_rate": 9.458357021034679e-06, "loss": 0.961, "step": 15744 }, { "epoch": 1.675356458821026, "grad_norm": 13.9453125, "learning_rate": 9.45832148948266e-06, "loss": 1.0773, "step": 15745 }, { "epoch": 1.6754628644392424, "grad_norm": 14.0625, "learning_rate": 9.458285957930642e-06, "loss": 1.0768, "step": 15746 }, { "epoch": 1.675569270057459, "grad_norm": 10.921875, "learning_rate": 9.458250426378624e-06, "loss": 1.0045, "step": 15747 }, { "epoch": 1.6756756756756757, "grad_norm": 14.890625, "learning_rate": 9.458214894826607e-06, "loss": 1.217, "step": 15748 }, { "epoch": 1.6757820812938924, "grad_norm": 10.890625, "learning_rate": 9.458179363274589e-06, "loss": 0.9724, "step": 15749 }, { "epoch": 1.675888486912109, "grad_norm": 13.0546875, "learning_rate": 9.45814383172257e-06, "loss": 1.0884, "step": 15750 }, { "epoch": 1.6759948925303256, "grad_norm": 15.28125, "learning_rate": 9.458108300170552e-06, "loss": 1.1741, "step": 15751 }, { "epoch": 1.6761012981485424, "grad_norm": 12.40625, "learning_rate": 9.458072768618534e-06, "loss": 1.1015, "step": 15752 }, { "epoch": 1.6762077037667589, "grad_norm": 14.78125, "learning_rate": 9.458037237066517e-06, "loss": 1.2716, "step": 15753 }, { "epoch": 1.6763141093849754, "grad_norm": 10.3671875, "learning_rate": 9.458001705514498e-06, "loss": 0.9067, "step": 15754 }, { "epoch": 1.6764205150031921, "grad_norm": 23.203125, "learning_rate": 9.45796617396248e-06, "loss": 0.8136, "step": 15755 }, { "epoch": 1.6765269206214088, "grad_norm": 18.625, "learning_rate": 9.457930642410462e-06, "loss": 0.9511, "step": 15756 }, { "epoch": 1.6766333262396254, "grad_norm": 14.109375, "learning_rate": 9.457895110858443e-06, "loss": 1.0426, "step": 15757 }, { "epoch": 1.676739731857842, "grad_norm": 9.5703125, "learning_rate": 9.457859579306425e-06, "loss": 0.9972, "step": 15758 }, { "epoch": 1.6768461374760588, "grad_norm": 13.3359375, "learning_rate": 9.457824047754406e-06, "loss": 1.2755, "step": 15759 }, { "epoch": 1.6769525430942753, "grad_norm": 11.8046875, "learning_rate": 9.457788516202388e-06, "loss": 1.035, "step": 15760 }, { "epoch": 1.677058948712492, "grad_norm": 10.421875, "learning_rate": 9.45775298465037e-06, "loss": 1.1574, "step": 15761 }, { "epoch": 1.6771653543307088, "grad_norm": 19.140625, "learning_rate": 9.457717453098351e-06, "loss": 1.2494, "step": 15762 }, { "epoch": 1.6772717599489253, "grad_norm": 13.34375, "learning_rate": 9.457681921546335e-06, "loss": 1.0798, "step": 15763 }, { "epoch": 1.6773781655671418, "grad_norm": 18.234375, "learning_rate": 9.457646389994316e-06, "loss": 1.1757, "step": 15764 }, { "epoch": 1.6774845711853585, "grad_norm": 9.5859375, "learning_rate": 9.457610858442298e-06, "loss": 1.043, "step": 15765 }, { "epoch": 1.6775909768035753, "grad_norm": 13.765625, "learning_rate": 9.45757532689028e-06, "loss": 0.9944, "step": 15766 }, { "epoch": 1.6776973824217918, "grad_norm": 10.78125, "learning_rate": 9.457539795338261e-06, "loss": 1.0028, "step": 15767 }, { "epoch": 1.6778037880400085, "grad_norm": 11.2890625, "learning_rate": 9.457504263786243e-06, "loss": 0.9547, "step": 15768 }, { "epoch": 1.6779101936582252, "grad_norm": 12.8828125, "learning_rate": 9.457468732234226e-06, "loss": 1.0878, "step": 15769 }, { "epoch": 1.6780165992764418, "grad_norm": 20.90625, "learning_rate": 9.457433200682208e-06, "loss": 0.9484, "step": 15770 }, { "epoch": 1.6781230048946585, "grad_norm": 16.640625, "learning_rate": 9.457397669130187e-06, "loss": 0.9457, "step": 15771 }, { "epoch": 1.6782294105128752, "grad_norm": 25.578125, "learning_rate": 9.457362137578169e-06, "loss": 1.3015, "step": 15772 }, { "epoch": 1.6783358161310917, "grad_norm": 10.7890625, "learning_rate": 9.45732660602615e-06, "loss": 1.0078, "step": 15773 }, { "epoch": 1.6784422217493082, "grad_norm": 12.2265625, "learning_rate": 9.457291074474134e-06, "loss": 1.2898, "step": 15774 }, { "epoch": 1.678548627367525, "grad_norm": 14.2734375, "learning_rate": 9.457255542922116e-06, "loss": 1.0509, "step": 15775 }, { "epoch": 1.6786550329857417, "grad_norm": 9.4453125, "learning_rate": 9.457220011370097e-06, "loss": 1.132, "step": 15776 }, { "epoch": 1.6787614386039582, "grad_norm": 19.046875, "learning_rate": 9.457184479818079e-06, "loss": 1.1852, "step": 15777 }, { "epoch": 1.678867844222175, "grad_norm": 11.6640625, "learning_rate": 9.45714894826606e-06, "loss": 1.1264, "step": 15778 }, { "epoch": 1.6789742498403917, "grad_norm": 14.4140625, "learning_rate": 9.457113416714044e-06, "loss": 1.2469, "step": 15779 }, { "epoch": 1.6790806554586082, "grad_norm": 11.6875, "learning_rate": 9.457077885162025e-06, "loss": 0.998, "step": 15780 }, { "epoch": 1.6791870610768247, "grad_norm": 13.28125, "learning_rate": 9.457042353610007e-06, "loss": 1.1386, "step": 15781 }, { "epoch": 1.6792934666950416, "grad_norm": 11.828125, "learning_rate": 9.457006822057989e-06, "loss": 1.0226, "step": 15782 }, { "epoch": 1.6793998723132582, "grad_norm": 11.8046875, "learning_rate": 9.45697129050597e-06, "loss": 1.3586, "step": 15783 }, { "epoch": 1.6795062779314747, "grad_norm": 8.9140625, "learning_rate": 9.456935758953952e-06, "loss": 0.9584, "step": 15784 }, { "epoch": 1.6796126835496914, "grad_norm": 10.5390625, "learning_rate": 9.456900227401933e-06, "loss": 1.1129, "step": 15785 }, { "epoch": 1.6797190891679081, "grad_norm": 13.46875, "learning_rate": 9.456864695849915e-06, "loss": 1.0802, "step": 15786 }, { "epoch": 1.6798254947861246, "grad_norm": 10.03125, "learning_rate": 9.456829164297897e-06, "loss": 0.9127, "step": 15787 }, { "epoch": 1.6799319004043414, "grad_norm": 10.5859375, "learning_rate": 9.456793632745878e-06, "loss": 0.9618, "step": 15788 }, { "epoch": 1.680038306022558, "grad_norm": 13.7890625, "learning_rate": 9.456758101193861e-06, "loss": 1.1398, "step": 15789 }, { "epoch": 1.6801447116407746, "grad_norm": 13.1484375, "learning_rate": 9.456722569641843e-06, "loss": 1.0911, "step": 15790 }, { "epoch": 1.6802511172589911, "grad_norm": 29.609375, "learning_rate": 9.456687038089825e-06, "loss": 1.2822, "step": 15791 }, { "epoch": 1.680357522877208, "grad_norm": 16.09375, "learning_rate": 9.456651506537806e-06, "loss": 0.8814, "step": 15792 }, { "epoch": 1.6804639284954246, "grad_norm": 11.3046875, "learning_rate": 9.456615974985788e-06, "loss": 1.1265, "step": 15793 }, { "epoch": 1.680570334113641, "grad_norm": 14.1328125, "learning_rate": 9.456580443433771e-06, "loss": 1.2726, "step": 15794 }, { "epoch": 1.6806767397318578, "grad_norm": 13.1171875, "learning_rate": 9.456544911881753e-06, "loss": 1.5399, "step": 15795 }, { "epoch": 1.6807831453500746, "grad_norm": 9.390625, "learning_rate": 9.456509380329734e-06, "loss": 1.0038, "step": 15796 }, { "epoch": 1.680889550968291, "grad_norm": 10.96875, "learning_rate": 9.456473848777716e-06, "loss": 0.9984, "step": 15797 }, { "epoch": 1.6809959565865078, "grad_norm": 12.7890625, "learning_rate": 9.456438317225698e-06, "loss": 1.1215, "step": 15798 }, { "epoch": 1.6811023622047245, "grad_norm": 12.140625, "learning_rate": 9.45640278567368e-06, "loss": 1.2068, "step": 15799 }, { "epoch": 1.681208767822941, "grad_norm": 10.265625, "learning_rate": 9.45636725412166e-06, "loss": 0.608, "step": 15800 }, { "epoch": 1.6813151734411576, "grad_norm": 16.625, "learning_rate": 9.456331722569642e-06, "loss": 1.0457, "step": 15801 }, { "epoch": 1.6814215790593743, "grad_norm": 15.4765625, "learning_rate": 9.456296191017624e-06, "loss": 1.1794, "step": 15802 }, { "epoch": 1.681527984677591, "grad_norm": 13.265625, "learning_rate": 9.456260659465606e-06, "loss": 1.3065, "step": 15803 }, { "epoch": 1.6816343902958075, "grad_norm": 10.6328125, "learning_rate": 9.456225127913587e-06, "loss": 1.0903, "step": 15804 }, { "epoch": 1.6817407959140243, "grad_norm": 11.015625, "learning_rate": 9.45618959636157e-06, "loss": 0.9477, "step": 15805 }, { "epoch": 1.681847201532241, "grad_norm": 11.8515625, "learning_rate": 9.456154064809552e-06, "loss": 1.0665, "step": 15806 }, { "epoch": 1.6819536071504575, "grad_norm": 11.234375, "learning_rate": 9.456118533257534e-06, "loss": 1.2508, "step": 15807 }, { "epoch": 1.6820600127686742, "grad_norm": 13.1328125, "learning_rate": 9.456083001705515e-06, "loss": 1.2018, "step": 15808 }, { "epoch": 1.682166418386891, "grad_norm": 13.328125, "learning_rate": 9.456047470153497e-06, "loss": 1.3487, "step": 15809 }, { "epoch": 1.6822728240051075, "grad_norm": 13.34375, "learning_rate": 9.456011938601479e-06, "loss": 1.0069, "step": 15810 }, { "epoch": 1.682379229623324, "grad_norm": 11.5546875, "learning_rate": 9.45597640704946e-06, "loss": 1.0831, "step": 15811 }, { "epoch": 1.6824856352415407, "grad_norm": 11.71875, "learning_rate": 9.455940875497442e-06, "loss": 0.9264, "step": 15812 }, { "epoch": 1.6825920408597574, "grad_norm": 9.546875, "learning_rate": 9.455905343945423e-06, "loss": 0.972, "step": 15813 }, { "epoch": 1.682698446477974, "grad_norm": 11.75, "learning_rate": 9.455869812393405e-06, "loss": 1.1766, "step": 15814 }, { "epoch": 1.6828048520961907, "grad_norm": 11.625, "learning_rate": 9.455834280841388e-06, "loss": 0.9521, "step": 15815 }, { "epoch": 1.6829112577144074, "grad_norm": 17.203125, "learning_rate": 9.45579874928937e-06, "loss": 1.3235, "step": 15816 }, { "epoch": 1.683017663332624, "grad_norm": 13.3046875, "learning_rate": 9.455763217737352e-06, "loss": 1.0426, "step": 15817 }, { "epoch": 1.6831240689508407, "grad_norm": 13.46875, "learning_rate": 9.455727686185333e-06, "loss": 0.9572, "step": 15818 }, { "epoch": 1.6832304745690574, "grad_norm": 13.359375, "learning_rate": 9.455692154633315e-06, "loss": 1.085, "step": 15819 }, { "epoch": 1.683336880187274, "grad_norm": 13.421875, "learning_rate": 9.455656623081298e-06, "loss": 0.9134, "step": 15820 }, { "epoch": 1.6834432858054904, "grad_norm": 9.6328125, "learning_rate": 9.45562109152928e-06, "loss": 0.9337, "step": 15821 }, { "epoch": 1.6835496914237071, "grad_norm": 9.671875, "learning_rate": 9.455585559977261e-06, "loss": 1.1428, "step": 15822 }, { "epoch": 1.6836560970419239, "grad_norm": 13.8359375, "learning_rate": 9.455550028425243e-06, "loss": 1.1237, "step": 15823 }, { "epoch": 1.6837625026601404, "grad_norm": 11.0390625, "learning_rate": 9.455514496873224e-06, "loss": 0.9358, "step": 15824 }, { "epoch": 1.6838689082783571, "grad_norm": 11.5703125, "learning_rate": 9.455478965321206e-06, "loss": 1.1347, "step": 15825 }, { "epoch": 1.6839753138965738, "grad_norm": 12.078125, "learning_rate": 9.455443433769188e-06, "loss": 1.0216, "step": 15826 }, { "epoch": 1.6840817195147904, "grad_norm": 10.296875, "learning_rate": 9.45540790221717e-06, "loss": 1.0602, "step": 15827 }, { "epoch": 1.6841881251330069, "grad_norm": 13.3125, "learning_rate": 9.455372370665151e-06, "loss": 0.8361, "step": 15828 }, { "epoch": 1.6842945307512238, "grad_norm": 12.390625, "learning_rate": 9.455336839113133e-06, "loss": 1.0276, "step": 15829 }, { "epoch": 1.6844009363694403, "grad_norm": 15.3046875, "learning_rate": 9.455301307561116e-06, "loss": 0.9884, "step": 15830 }, { "epoch": 1.6845073419876568, "grad_norm": 10.8359375, "learning_rate": 9.455265776009097e-06, "loss": 1.3093, "step": 15831 }, { "epoch": 1.6846137476058736, "grad_norm": 12.6640625, "learning_rate": 9.455230244457079e-06, "loss": 1.3178, "step": 15832 }, { "epoch": 1.6847201532240903, "grad_norm": 11.84375, "learning_rate": 9.45519471290506e-06, "loss": 1.098, "step": 15833 }, { "epoch": 1.6848265588423068, "grad_norm": 12.3046875, "learning_rate": 9.455159181353042e-06, "loss": 1.0277, "step": 15834 }, { "epoch": 1.6849329644605235, "grad_norm": 9.578125, "learning_rate": 9.455123649801024e-06, "loss": 0.9834, "step": 15835 }, { "epoch": 1.6850393700787403, "grad_norm": 12.9609375, "learning_rate": 9.455088118249005e-06, "loss": 1.1935, "step": 15836 }, { "epoch": 1.6851457756969568, "grad_norm": 14.0703125, "learning_rate": 9.455052586696987e-06, "loss": 1.3477, "step": 15837 }, { "epoch": 1.6852521813151733, "grad_norm": 14.8359375, "learning_rate": 9.455017055144969e-06, "loss": 1.1553, "step": 15838 }, { "epoch": 1.6853585869333902, "grad_norm": 11.484375, "learning_rate": 9.45498152359295e-06, "loss": 0.9922, "step": 15839 }, { "epoch": 1.6854649925516068, "grad_norm": 13.703125, "learning_rate": 9.454945992040934e-06, "loss": 1.0166, "step": 15840 }, { "epoch": 1.6855713981698233, "grad_norm": 12.484375, "learning_rate": 9.454910460488915e-06, "loss": 1.1261, "step": 15841 }, { "epoch": 1.68567780378804, "grad_norm": 17.96875, "learning_rate": 9.454874928936897e-06, "loss": 1.1107, "step": 15842 }, { "epoch": 1.6857842094062567, "grad_norm": 14.28125, "learning_rate": 9.454839397384878e-06, "loss": 1.0836, "step": 15843 }, { "epoch": 1.6858906150244732, "grad_norm": 14.609375, "learning_rate": 9.45480386583286e-06, "loss": 1.1076, "step": 15844 }, { "epoch": 1.68599702064269, "grad_norm": 10.90625, "learning_rate": 9.454768334280842e-06, "loss": 0.9391, "step": 15845 }, { "epoch": 1.6861034262609067, "grad_norm": 11.1015625, "learning_rate": 9.454732802728825e-06, "loss": 1.3381, "step": 15846 }, { "epoch": 1.6862098318791232, "grad_norm": 9.375, "learning_rate": 9.454697271176807e-06, "loss": 1.068, "step": 15847 }, { "epoch": 1.6863162374973397, "grad_norm": 10.4375, "learning_rate": 9.454661739624788e-06, "loss": 1.2853, "step": 15848 }, { "epoch": 1.6864226431155565, "grad_norm": 15.8359375, "learning_rate": 9.45462620807277e-06, "loss": 1.1349, "step": 15849 }, { "epoch": 1.6865290487337732, "grad_norm": 12.4453125, "learning_rate": 9.454590676520751e-06, "loss": 1.0696, "step": 15850 }, { "epoch": 1.6866354543519897, "grad_norm": 11.15625, "learning_rate": 9.454555144968733e-06, "loss": 0.9353, "step": 15851 }, { "epoch": 1.6867418599702064, "grad_norm": 11.484375, "learning_rate": 9.454519613416715e-06, "loss": 1.0711, "step": 15852 }, { "epoch": 1.6868482655884232, "grad_norm": 11.5703125, "learning_rate": 9.454484081864696e-06, "loss": 1.4401, "step": 15853 }, { "epoch": 1.6869546712066397, "grad_norm": 11.5234375, "learning_rate": 9.454448550312678e-06, "loss": 0.8039, "step": 15854 }, { "epoch": 1.6870610768248564, "grad_norm": 10.4375, "learning_rate": 9.45441301876066e-06, "loss": 0.8427, "step": 15855 }, { "epoch": 1.6871674824430731, "grad_norm": 21.375, "learning_rate": 9.454377487208643e-06, "loss": 0.8121, "step": 15856 }, { "epoch": 1.6872738880612896, "grad_norm": 16.0, "learning_rate": 9.454341955656624e-06, "loss": 1.1475, "step": 15857 }, { "epoch": 1.6873802936795061, "grad_norm": 11.5625, "learning_rate": 9.454306424104606e-06, "loss": 0.882, "step": 15858 }, { "epoch": 1.6874866992977229, "grad_norm": 12.8125, "learning_rate": 9.454270892552587e-06, "loss": 1.1952, "step": 15859 }, { "epoch": 1.6875931049159396, "grad_norm": 12.5390625, "learning_rate": 9.454235361000569e-06, "loss": 1.0005, "step": 15860 }, { "epoch": 1.6876995105341561, "grad_norm": 10.703125, "learning_rate": 9.45419982944855e-06, "loss": 1.0262, "step": 15861 }, { "epoch": 1.6878059161523729, "grad_norm": 13.6171875, "learning_rate": 9.454164297896532e-06, "loss": 1.3823, "step": 15862 }, { "epoch": 1.6879123217705896, "grad_norm": 11.4140625, "learning_rate": 9.454128766344514e-06, "loss": 0.9474, "step": 15863 }, { "epoch": 1.688018727388806, "grad_norm": 11.1015625, "learning_rate": 9.454093234792496e-06, "loss": 0.847, "step": 15864 }, { "epoch": 1.6881251330070226, "grad_norm": 10.7265625, "learning_rate": 9.454057703240477e-06, "loss": 0.9886, "step": 15865 }, { "epoch": 1.6882315386252396, "grad_norm": 13.125, "learning_rate": 9.45402217168846e-06, "loss": 1.0678, "step": 15866 }, { "epoch": 1.688337944243456, "grad_norm": 11.375, "learning_rate": 9.453986640136442e-06, "loss": 0.9563, "step": 15867 }, { "epoch": 1.6884443498616726, "grad_norm": 13.21875, "learning_rate": 9.453951108584424e-06, "loss": 1.1064, "step": 15868 }, { "epoch": 1.6885507554798893, "grad_norm": 11.1953125, "learning_rate": 9.453915577032405e-06, "loss": 1.157, "step": 15869 }, { "epoch": 1.688657161098106, "grad_norm": 12.2578125, "learning_rate": 9.453880045480387e-06, "loss": 1.2647, "step": 15870 }, { "epoch": 1.6887635667163226, "grad_norm": 11.3125, "learning_rate": 9.45384451392837e-06, "loss": 1.0832, "step": 15871 }, { "epoch": 1.6888699723345393, "grad_norm": 12.9609375, "learning_rate": 9.453808982376352e-06, "loss": 1.004, "step": 15872 }, { "epoch": 1.688976377952756, "grad_norm": 10.1015625, "learning_rate": 9.453773450824333e-06, "loss": 0.9253, "step": 15873 }, { "epoch": 1.6890827835709725, "grad_norm": 19.1875, "learning_rate": 9.453737919272315e-06, "loss": 1.2389, "step": 15874 }, { "epoch": 1.689189189189189, "grad_norm": 16.640625, "learning_rate": 9.453702387720297e-06, "loss": 1.1358, "step": 15875 }, { "epoch": 1.689295594807406, "grad_norm": 12.6328125, "learning_rate": 9.453666856168278e-06, "loss": 1.1223, "step": 15876 }, { "epoch": 1.6894020004256225, "grad_norm": 12.1875, "learning_rate": 9.45363132461626e-06, "loss": 1.0753, "step": 15877 }, { "epoch": 1.689508406043839, "grad_norm": 11.78125, "learning_rate": 9.453595793064241e-06, "loss": 1.2356, "step": 15878 }, { "epoch": 1.6896148116620557, "grad_norm": 72.1875, "learning_rate": 9.453560261512223e-06, "loss": 1.3756, "step": 15879 }, { "epoch": 1.6897212172802725, "grad_norm": 12.3828125, "learning_rate": 9.453524729960205e-06, "loss": 1.2035, "step": 15880 }, { "epoch": 1.689827622898489, "grad_norm": 13.1640625, "learning_rate": 9.453489198408188e-06, "loss": 1.0401, "step": 15881 }, { "epoch": 1.6899340285167057, "grad_norm": 10.7265625, "learning_rate": 9.45345366685617e-06, "loss": 0.8262, "step": 15882 }, { "epoch": 1.6900404341349224, "grad_norm": 12.8203125, "learning_rate": 9.453418135304151e-06, "loss": 1.2786, "step": 15883 }, { "epoch": 1.690146839753139, "grad_norm": 13.5078125, "learning_rate": 9.453382603752133e-06, "loss": 1.3731, "step": 15884 }, { "epoch": 1.6902532453713555, "grad_norm": 11.75, "learning_rate": 9.453347072200114e-06, "loss": 0.9336, "step": 15885 }, { "epoch": 1.6903596509895724, "grad_norm": 13.5234375, "learning_rate": 9.453311540648096e-06, "loss": 1.171, "step": 15886 }, { "epoch": 1.690466056607789, "grad_norm": 10.3359375, "learning_rate": 9.45327600909608e-06, "loss": 1.1719, "step": 15887 }, { "epoch": 1.6905724622260054, "grad_norm": 11.1015625, "learning_rate": 9.453240477544061e-06, "loss": 1.2183, "step": 15888 }, { "epoch": 1.6906788678442222, "grad_norm": 14.953125, "learning_rate": 9.453204945992042e-06, "loss": 1.1621, "step": 15889 }, { "epoch": 1.690785273462439, "grad_norm": 10.046875, "learning_rate": 9.453169414440024e-06, "loss": 1.0661, "step": 15890 }, { "epoch": 1.6908916790806554, "grad_norm": 21.6875, "learning_rate": 9.453133882888004e-06, "loss": 1.1438, "step": 15891 }, { "epoch": 1.6909980846988721, "grad_norm": 12.1953125, "learning_rate": 9.453098351335987e-06, "loss": 1.0907, "step": 15892 }, { "epoch": 1.6911044903170889, "grad_norm": 15.9453125, "learning_rate": 9.453062819783969e-06, "loss": 1.2174, "step": 15893 }, { "epoch": 1.6912108959353054, "grad_norm": 14.1484375, "learning_rate": 9.45302728823195e-06, "loss": 1.1918, "step": 15894 }, { "epoch": 1.691317301553522, "grad_norm": 10.953125, "learning_rate": 9.452991756679932e-06, "loss": 0.9291, "step": 15895 }, { "epoch": 1.6914237071717386, "grad_norm": 15.640625, "learning_rate": 9.452956225127914e-06, "loss": 1.3324, "step": 15896 }, { "epoch": 1.6915301127899554, "grad_norm": 9.78125, "learning_rate": 9.452920693575897e-06, "loss": 0.9505, "step": 15897 }, { "epoch": 1.6916365184081719, "grad_norm": 13.7421875, "learning_rate": 9.452885162023879e-06, "loss": 1.0183, "step": 15898 }, { "epoch": 1.6917429240263886, "grad_norm": 17.390625, "learning_rate": 9.45284963047186e-06, "loss": 1.2948, "step": 15899 }, { "epoch": 1.6918493296446053, "grad_norm": 16.125, "learning_rate": 9.452814098919842e-06, "loss": 0.8293, "step": 15900 }, { "epoch": 1.6919557352628218, "grad_norm": 12.46875, "learning_rate": 9.452778567367823e-06, "loss": 1.1607, "step": 15901 }, { "epoch": 1.6920621408810386, "grad_norm": 16.140625, "learning_rate": 9.452743035815805e-06, "loss": 1.0392, "step": 15902 }, { "epoch": 1.6921685464992553, "grad_norm": 19.265625, "learning_rate": 9.452707504263787e-06, "loss": 1.1181, "step": 15903 }, { "epoch": 1.6922749521174718, "grad_norm": 11.359375, "learning_rate": 9.452671972711768e-06, "loss": 1.0308, "step": 15904 }, { "epoch": 1.6923813577356883, "grad_norm": 11.2890625, "learning_rate": 9.45263644115975e-06, "loss": 1.0627, "step": 15905 }, { "epoch": 1.692487763353905, "grad_norm": 13.7109375, "learning_rate": 9.452600909607731e-06, "loss": 1.0281, "step": 15906 }, { "epoch": 1.6925941689721218, "grad_norm": 14.96875, "learning_rate": 9.452565378055715e-06, "loss": 0.9509, "step": 15907 }, { "epoch": 1.6927005745903383, "grad_norm": 15.453125, "learning_rate": 9.452529846503696e-06, "loss": 1.0198, "step": 15908 }, { "epoch": 1.692806980208555, "grad_norm": 13.4296875, "learning_rate": 9.452494314951678e-06, "loss": 1.0576, "step": 15909 }, { "epoch": 1.6929133858267718, "grad_norm": 9.4453125, "learning_rate": 9.45245878339966e-06, "loss": 0.782, "step": 15910 }, { "epoch": 1.6930197914449883, "grad_norm": 10.2109375, "learning_rate": 9.452423251847641e-06, "loss": 1.0153, "step": 15911 }, { "epoch": 1.6931261970632048, "grad_norm": 11.15625, "learning_rate": 9.452387720295624e-06, "loss": 1.0932, "step": 15912 }, { "epoch": 1.6932326026814217, "grad_norm": 19.03125, "learning_rate": 9.452352188743606e-06, "loss": 0.9009, "step": 15913 }, { "epoch": 1.6933390082996382, "grad_norm": 12.640625, "learning_rate": 9.452316657191588e-06, "loss": 0.9776, "step": 15914 }, { "epoch": 1.6934454139178547, "grad_norm": 10.875, "learning_rate": 9.45228112563957e-06, "loss": 1.0913, "step": 15915 }, { "epoch": 1.6935518195360715, "grad_norm": 17.859375, "learning_rate": 9.452245594087551e-06, "loss": 1.1011, "step": 15916 }, { "epoch": 1.6936582251542882, "grad_norm": 13.2578125, "learning_rate": 9.452210062535533e-06, "loss": 1.1531, "step": 15917 }, { "epoch": 1.6937646307725047, "grad_norm": 14.0234375, "learning_rate": 9.452174530983514e-06, "loss": 1.1226, "step": 15918 }, { "epoch": 1.6938710363907215, "grad_norm": 11.375, "learning_rate": 9.452138999431496e-06, "loss": 1.1736, "step": 15919 }, { "epoch": 1.6939774420089382, "grad_norm": 16.15625, "learning_rate": 9.452103467879477e-06, "loss": 0.9807, "step": 15920 }, { "epoch": 1.6940838476271547, "grad_norm": 12.9296875, "learning_rate": 9.452067936327459e-06, "loss": 1.2077, "step": 15921 }, { "epoch": 1.6941902532453712, "grad_norm": 15.8359375, "learning_rate": 9.45203240477544e-06, "loss": 1.0965, "step": 15922 }, { "epoch": 1.6942966588635882, "grad_norm": 17.046875, "learning_rate": 9.451996873223424e-06, "loss": 1.1061, "step": 15923 }, { "epoch": 1.6944030644818047, "grad_norm": 12.234375, "learning_rate": 9.451961341671405e-06, "loss": 1.024, "step": 15924 }, { "epoch": 1.6945094701000212, "grad_norm": 12.8046875, "learning_rate": 9.451925810119387e-06, "loss": 0.9989, "step": 15925 }, { "epoch": 1.694615875718238, "grad_norm": 11.015625, "learning_rate": 9.451890278567369e-06, "loss": 0.9419, "step": 15926 }, { "epoch": 1.6947222813364546, "grad_norm": 14.984375, "learning_rate": 9.45185474701535e-06, "loss": 1.2051, "step": 15927 }, { "epoch": 1.6948286869546711, "grad_norm": 13.421875, "learning_rate": 9.451819215463332e-06, "loss": 1.4128, "step": 15928 }, { "epoch": 1.6949350925728879, "grad_norm": 19.5, "learning_rate": 9.451783683911314e-06, "loss": 1.1256, "step": 15929 }, { "epoch": 1.6950414981911046, "grad_norm": 21.8125, "learning_rate": 9.451748152359295e-06, "loss": 1.0679, "step": 15930 }, { "epoch": 1.6951479038093211, "grad_norm": 18.15625, "learning_rate": 9.451712620807277e-06, "loss": 1.1006, "step": 15931 }, { "epoch": 1.6952543094275376, "grad_norm": 13.546875, "learning_rate": 9.451677089255258e-06, "loss": 1.3345, "step": 15932 }, { "epoch": 1.6953607150457544, "grad_norm": 12.734375, "learning_rate": 9.451641557703242e-06, "loss": 0.9466, "step": 15933 }, { "epoch": 1.695467120663971, "grad_norm": 10.65625, "learning_rate": 9.451606026151223e-06, "loss": 1.0127, "step": 15934 }, { "epoch": 1.6955735262821876, "grad_norm": 10.3984375, "learning_rate": 9.451570494599205e-06, "loss": 1.2395, "step": 15935 }, { "epoch": 1.6956799319004043, "grad_norm": 15.1171875, "learning_rate": 9.451534963047186e-06, "loss": 1.0866, "step": 15936 }, { "epoch": 1.695786337518621, "grad_norm": 14.3984375, "learning_rate": 9.451499431495168e-06, "loss": 1.2756, "step": 15937 }, { "epoch": 1.6958927431368376, "grad_norm": 14.2734375, "learning_rate": 9.451463899943151e-06, "loss": 1.1787, "step": 15938 }, { "epoch": 1.6959991487550543, "grad_norm": 18.28125, "learning_rate": 9.451428368391133e-06, "loss": 1.1369, "step": 15939 }, { "epoch": 1.696105554373271, "grad_norm": 12.515625, "learning_rate": 9.451392836839115e-06, "loss": 1.1866, "step": 15940 }, { "epoch": 1.6962119599914876, "grad_norm": 12.0390625, "learning_rate": 9.451357305287096e-06, "loss": 1.1613, "step": 15941 }, { "epoch": 1.696318365609704, "grad_norm": 11.5859375, "learning_rate": 9.451321773735078e-06, "loss": 1.0939, "step": 15942 }, { "epoch": 1.6964247712279208, "grad_norm": 10.859375, "learning_rate": 9.45128624218306e-06, "loss": 1.2893, "step": 15943 }, { "epoch": 1.6965311768461375, "grad_norm": 13.640625, "learning_rate": 9.451250710631041e-06, "loss": 1.2879, "step": 15944 }, { "epoch": 1.696637582464354, "grad_norm": 13.6328125, "learning_rate": 9.451215179079023e-06, "loss": 1.2501, "step": 15945 }, { "epoch": 1.6967439880825708, "grad_norm": 10.1484375, "learning_rate": 9.451179647527004e-06, "loss": 0.7924, "step": 15946 }, { "epoch": 1.6968503937007875, "grad_norm": 9.296875, "learning_rate": 9.451144115974986e-06, "loss": 1.1411, "step": 15947 }, { "epoch": 1.696956799319004, "grad_norm": 12.359375, "learning_rate": 9.451108584422969e-06, "loss": 0.9708, "step": 15948 }, { "epoch": 1.6970632049372207, "grad_norm": 15.890625, "learning_rate": 9.45107305287095e-06, "loss": 0.9813, "step": 15949 }, { "epoch": 1.6971696105554375, "grad_norm": 15.625, "learning_rate": 9.451037521318932e-06, "loss": 1.1554, "step": 15950 }, { "epoch": 1.697276016173654, "grad_norm": 12.46875, "learning_rate": 9.451001989766914e-06, "loss": 1.2375, "step": 15951 }, { "epoch": 1.6973824217918705, "grad_norm": 11.9296875, "learning_rate": 9.450966458214896e-06, "loss": 1.0411, "step": 15952 }, { "epoch": 1.6974888274100872, "grad_norm": 10.5625, "learning_rate": 9.450930926662877e-06, "loss": 1.0758, "step": 15953 }, { "epoch": 1.697595233028304, "grad_norm": 7.99609375, "learning_rate": 9.450895395110859e-06, "loss": 0.9827, "step": 15954 }, { "epoch": 1.6977016386465205, "grad_norm": 14.609375, "learning_rate": 9.45085986355884e-06, "loss": 1.2949, "step": 15955 }, { "epoch": 1.6978080442647372, "grad_norm": 19.15625, "learning_rate": 9.450824332006822e-06, "loss": 1.0727, "step": 15956 }, { "epoch": 1.697914449882954, "grad_norm": 10.78125, "learning_rate": 9.450788800454804e-06, "loss": 1.1111, "step": 15957 }, { "epoch": 1.6980208555011704, "grad_norm": 13.09375, "learning_rate": 9.450753268902787e-06, "loss": 0.9872, "step": 15958 }, { "epoch": 1.698127261119387, "grad_norm": 15.828125, "learning_rate": 9.450717737350768e-06, "loss": 1.1352, "step": 15959 }, { "epoch": 1.698233666737604, "grad_norm": 11.0078125, "learning_rate": 9.45068220579875e-06, "loss": 1.1784, "step": 15960 }, { "epoch": 1.6983400723558204, "grad_norm": 13.625, "learning_rate": 9.450646674246732e-06, "loss": 1.0858, "step": 15961 }, { "epoch": 1.698446477974037, "grad_norm": 13.34375, "learning_rate": 9.450611142694713e-06, "loss": 1.0267, "step": 15962 }, { "epoch": 1.6985528835922536, "grad_norm": 10.890625, "learning_rate": 9.450575611142695e-06, "loss": 1.1897, "step": 15963 }, { "epoch": 1.6986592892104704, "grad_norm": 11.0546875, "learning_rate": 9.450540079590678e-06, "loss": 1.0212, "step": 15964 }, { "epoch": 1.698765694828687, "grad_norm": 8.984375, "learning_rate": 9.45050454803866e-06, "loss": 1.0703, "step": 15965 }, { "epoch": 1.6988721004469036, "grad_norm": 11.4375, "learning_rate": 9.450469016486641e-06, "loss": 1.19, "step": 15966 }, { "epoch": 1.6989785060651204, "grad_norm": 12.671875, "learning_rate": 9.450433484934623e-06, "loss": 1.2721, "step": 15967 }, { "epoch": 1.6990849116833369, "grad_norm": 14.09375, "learning_rate": 9.450397953382605e-06, "loss": 1.3423, "step": 15968 }, { "epoch": 1.6991913173015534, "grad_norm": 11.4609375, "learning_rate": 9.450362421830586e-06, "loss": 0.9662, "step": 15969 }, { "epoch": 1.6992977229197703, "grad_norm": 13.7578125, "learning_rate": 9.450326890278568e-06, "loss": 1.2014, "step": 15970 }, { "epoch": 1.6994041285379868, "grad_norm": 14.140625, "learning_rate": 9.45029135872655e-06, "loss": 1.0774, "step": 15971 }, { "epoch": 1.6995105341562033, "grad_norm": 11.5078125, "learning_rate": 9.450255827174531e-06, "loss": 1.1821, "step": 15972 }, { "epoch": 1.69961693977442, "grad_norm": 16.40625, "learning_rate": 9.450220295622513e-06, "loss": 1.1782, "step": 15973 }, { "epoch": 1.6997233453926368, "grad_norm": 11.9609375, "learning_rate": 9.450184764070496e-06, "loss": 1.0911, "step": 15974 }, { "epoch": 1.6998297510108533, "grad_norm": 10.4296875, "learning_rate": 9.450149232518478e-06, "loss": 1.2562, "step": 15975 }, { "epoch": 1.69993615662907, "grad_norm": 9.9375, "learning_rate": 9.45011370096646e-06, "loss": 1.0674, "step": 15976 }, { "epoch": 1.7000425622472868, "grad_norm": 15.515625, "learning_rate": 9.45007816941444e-06, "loss": 1.4483, "step": 15977 }, { "epoch": 1.7001489678655033, "grad_norm": 11.2578125, "learning_rate": 9.450042637862422e-06, "loss": 1.0748, "step": 15978 }, { "epoch": 1.7002553734837198, "grad_norm": 11.3203125, "learning_rate": 9.450007106310406e-06, "loss": 1.1739, "step": 15979 }, { "epoch": 1.7003617791019365, "grad_norm": 10.984375, "learning_rate": 9.449971574758386e-06, "loss": 1.1389, "step": 15980 }, { "epoch": 1.7004681847201533, "grad_norm": 14.828125, "learning_rate": 9.449936043206367e-06, "loss": 0.8839, "step": 15981 }, { "epoch": 1.7005745903383698, "grad_norm": 13.515625, "learning_rate": 9.449900511654349e-06, "loss": 1.1187, "step": 15982 }, { "epoch": 1.7006809959565865, "grad_norm": 12.6484375, "learning_rate": 9.44986498010233e-06, "loss": 1.3026, "step": 15983 }, { "epoch": 1.7007874015748032, "grad_norm": 11.828125, "learning_rate": 9.449829448550314e-06, "loss": 1.03, "step": 15984 }, { "epoch": 1.7008938071930197, "grad_norm": 11.890625, "learning_rate": 9.449793916998295e-06, "loss": 1.086, "step": 15985 }, { "epoch": 1.7010002128112365, "grad_norm": 28.5625, "learning_rate": 9.449758385446277e-06, "loss": 1.1103, "step": 15986 }, { "epoch": 1.7011066184294532, "grad_norm": 14.1875, "learning_rate": 9.449722853894259e-06, "loss": 1.2789, "step": 15987 }, { "epoch": 1.7012130240476697, "grad_norm": 13.6953125, "learning_rate": 9.44968732234224e-06, "loss": 1.0701, "step": 15988 }, { "epoch": 1.7013194296658862, "grad_norm": 17.4375, "learning_rate": 9.449651790790223e-06, "loss": 1.0221, "step": 15989 }, { "epoch": 1.701425835284103, "grad_norm": 11.75, "learning_rate": 9.449616259238205e-06, "loss": 1.0329, "step": 15990 }, { "epoch": 1.7015322409023197, "grad_norm": 16.296875, "learning_rate": 9.449580727686187e-06, "loss": 1.2803, "step": 15991 }, { "epoch": 1.7016386465205362, "grad_norm": 14.3203125, "learning_rate": 9.449545196134168e-06, "loss": 0.9858, "step": 15992 }, { "epoch": 1.701745052138753, "grad_norm": 11.7421875, "learning_rate": 9.44950966458215e-06, "loss": 1.0494, "step": 15993 }, { "epoch": 1.7018514577569697, "grad_norm": 17.828125, "learning_rate": 9.449474133030131e-06, "loss": 1.2312, "step": 15994 }, { "epoch": 1.7019578633751862, "grad_norm": 10.9140625, "learning_rate": 9.449438601478113e-06, "loss": 0.7315, "step": 15995 }, { "epoch": 1.702064268993403, "grad_norm": 14.8359375, "learning_rate": 9.449403069926095e-06, "loss": 1.3801, "step": 15996 }, { "epoch": 1.7021706746116196, "grad_norm": 12.71875, "learning_rate": 9.449367538374076e-06, "loss": 1.2196, "step": 15997 }, { "epoch": 1.7022770802298361, "grad_norm": 13.0234375, "learning_rate": 9.449332006822058e-06, "loss": 1.0737, "step": 15998 }, { "epoch": 1.7023834858480527, "grad_norm": 11.9609375, "learning_rate": 9.44929647527004e-06, "loss": 1.0379, "step": 15999 }, { "epoch": 1.7024898914662694, "grad_norm": 12.09375, "learning_rate": 9.449260943718023e-06, "loss": 0.9403, "step": 16000 }, { "epoch": 1.7025962970844861, "grad_norm": 15.28125, "learning_rate": 9.449225412166004e-06, "loss": 1.1439, "step": 16001 }, { "epoch": 1.7027027027027026, "grad_norm": 12.7109375, "learning_rate": 9.449189880613986e-06, "loss": 1.3462, "step": 16002 }, { "epoch": 1.7028091083209194, "grad_norm": 15.640625, "learning_rate": 9.449154349061968e-06, "loss": 1.0598, "step": 16003 }, { "epoch": 1.702915513939136, "grad_norm": 12.5703125, "learning_rate": 9.44911881750995e-06, "loss": 0.9812, "step": 16004 }, { "epoch": 1.7030219195573526, "grad_norm": 9.9609375, "learning_rate": 9.449083285957933e-06, "loss": 1.0594, "step": 16005 }, { "epoch": 1.7031283251755691, "grad_norm": 13.046875, "learning_rate": 9.449047754405914e-06, "loss": 1.117, "step": 16006 }, { "epoch": 1.703234730793786, "grad_norm": 13.1875, "learning_rate": 9.449012222853896e-06, "loss": 1.3237, "step": 16007 }, { "epoch": 1.7033411364120026, "grad_norm": 13.359375, "learning_rate": 9.448976691301877e-06, "loss": 1.1156, "step": 16008 }, { "epoch": 1.703447542030219, "grad_norm": 15.8125, "learning_rate": 9.448941159749859e-06, "loss": 0.9387, "step": 16009 }, { "epoch": 1.7035539476484358, "grad_norm": 15.7578125, "learning_rate": 9.44890562819784e-06, "loss": 1.0664, "step": 16010 }, { "epoch": 1.7036603532666525, "grad_norm": 12.5390625, "learning_rate": 9.448870096645822e-06, "loss": 1.1952, "step": 16011 }, { "epoch": 1.703766758884869, "grad_norm": 11.484375, "learning_rate": 9.448834565093804e-06, "loss": 1.3607, "step": 16012 }, { "epoch": 1.7038731645030858, "grad_norm": 14.0234375, "learning_rate": 9.448799033541785e-06, "loss": 1.0899, "step": 16013 }, { "epoch": 1.7039795701213025, "grad_norm": 16.796875, "learning_rate": 9.448763501989767e-06, "loss": 1.2018, "step": 16014 }, { "epoch": 1.704085975739519, "grad_norm": 15.8984375, "learning_rate": 9.44872797043775e-06, "loss": 1.1856, "step": 16015 }, { "epoch": 1.7041923813577355, "grad_norm": 12.7265625, "learning_rate": 9.448692438885732e-06, "loss": 1.1374, "step": 16016 }, { "epoch": 1.7042987869759525, "grad_norm": 11.8046875, "learning_rate": 9.448656907333714e-06, "loss": 1.2374, "step": 16017 }, { "epoch": 1.704405192594169, "grad_norm": 9.7265625, "learning_rate": 9.448621375781695e-06, "loss": 1.0847, "step": 16018 }, { "epoch": 1.7045115982123855, "grad_norm": 10.890625, "learning_rate": 9.448585844229677e-06, "loss": 0.9738, "step": 16019 }, { "epoch": 1.7046180038306022, "grad_norm": 11.90625, "learning_rate": 9.448550312677658e-06, "loss": 1.2233, "step": 16020 }, { "epoch": 1.704724409448819, "grad_norm": 15.3984375, "learning_rate": 9.44851478112564e-06, "loss": 1.619, "step": 16021 }, { "epoch": 1.7048308150670355, "grad_norm": 12.8203125, "learning_rate": 9.448479249573622e-06, "loss": 1.1744, "step": 16022 }, { "epoch": 1.7049372206852522, "grad_norm": 14.8046875, "learning_rate": 9.448443718021603e-06, "loss": 1.3543, "step": 16023 }, { "epoch": 1.705043626303469, "grad_norm": 16.015625, "learning_rate": 9.448408186469585e-06, "loss": 1.0756, "step": 16024 }, { "epoch": 1.7051500319216855, "grad_norm": 13.5234375, "learning_rate": 9.448372654917568e-06, "loss": 1.2436, "step": 16025 }, { "epoch": 1.705256437539902, "grad_norm": 12.6484375, "learning_rate": 9.44833712336555e-06, "loss": 1.0361, "step": 16026 }, { "epoch": 1.7053628431581187, "grad_norm": 11.96875, "learning_rate": 9.448301591813531e-06, "loss": 0.9434, "step": 16027 }, { "epoch": 1.7054692487763354, "grad_norm": 13.0, "learning_rate": 9.448266060261513e-06, "loss": 0.8341, "step": 16028 }, { "epoch": 1.705575654394552, "grad_norm": 10.7265625, "learning_rate": 9.448230528709495e-06, "loss": 1.0799, "step": 16029 }, { "epoch": 1.7056820600127687, "grad_norm": 12.28125, "learning_rate": 9.448194997157478e-06, "loss": 1.2013, "step": 16030 }, { "epoch": 1.7057884656309854, "grad_norm": 10.9765625, "learning_rate": 9.44815946560546e-06, "loss": 0.9956, "step": 16031 }, { "epoch": 1.705894871249202, "grad_norm": 10.9453125, "learning_rate": 9.448123934053441e-06, "loss": 1.0281, "step": 16032 }, { "epoch": 1.7060012768674186, "grad_norm": 16.046875, "learning_rate": 9.448088402501423e-06, "loss": 0.9897, "step": 16033 }, { "epoch": 1.7061076824856354, "grad_norm": 9.546875, "learning_rate": 9.448052870949404e-06, "loss": 1.0397, "step": 16034 }, { "epoch": 1.706214088103852, "grad_norm": 13.140625, "learning_rate": 9.448017339397386e-06, "loss": 1.0186, "step": 16035 }, { "epoch": 1.7063204937220684, "grad_norm": 12.203125, "learning_rate": 9.447981807845367e-06, "loss": 1.2278, "step": 16036 }, { "epoch": 1.7064268993402851, "grad_norm": 17.3125, "learning_rate": 9.447946276293349e-06, "loss": 1.2696, "step": 16037 }, { "epoch": 1.7065333049585019, "grad_norm": 14.2734375, "learning_rate": 9.44791074474133e-06, "loss": 1.0021, "step": 16038 }, { "epoch": 1.7066397105767184, "grad_norm": 10.2578125, "learning_rate": 9.447875213189312e-06, "loss": 0.9066, "step": 16039 }, { "epoch": 1.706746116194935, "grad_norm": 18.78125, "learning_rate": 9.447839681637294e-06, "loss": 0.9995, "step": 16040 }, { "epoch": 1.7068525218131518, "grad_norm": 12.2421875, "learning_rate": 9.447804150085277e-06, "loss": 0.9097, "step": 16041 }, { "epoch": 1.7069589274313683, "grad_norm": 12.40625, "learning_rate": 9.447768618533259e-06, "loss": 1.4983, "step": 16042 }, { "epoch": 1.7070653330495849, "grad_norm": 14.9921875, "learning_rate": 9.44773308698124e-06, "loss": 1.3006, "step": 16043 }, { "epoch": 1.7071717386678018, "grad_norm": 9.65625, "learning_rate": 9.447697555429222e-06, "loss": 0.9408, "step": 16044 }, { "epoch": 1.7072781442860183, "grad_norm": 10.359375, "learning_rate": 9.447662023877204e-06, "loss": 1.2056, "step": 16045 }, { "epoch": 1.7073845499042348, "grad_norm": 13.09375, "learning_rate": 9.447626492325185e-06, "loss": 0.8988, "step": 16046 }, { "epoch": 1.7074909555224516, "grad_norm": 13.859375, "learning_rate": 9.447590960773167e-06, "loss": 1.3386, "step": 16047 }, { "epoch": 1.7075973611406683, "grad_norm": 14.140625, "learning_rate": 9.447555429221148e-06, "loss": 1.089, "step": 16048 }, { "epoch": 1.7077037667588848, "grad_norm": 15.6640625, "learning_rate": 9.44751989766913e-06, "loss": 1.0448, "step": 16049 }, { "epoch": 1.7078101723771015, "grad_norm": 11.4765625, "learning_rate": 9.447484366117112e-06, "loss": 0.985, "step": 16050 }, { "epoch": 1.7079165779953183, "grad_norm": 10.625, "learning_rate": 9.447448834565095e-06, "loss": 1.1222, "step": 16051 }, { "epoch": 1.7080229836135348, "grad_norm": 10.6875, "learning_rate": 9.447413303013077e-06, "loss": 1.1488, "step": 16052 }, { "epoch": 1.7081293892317513, "grad_norm": 17.3125, "learning_rate": 9.447377771461058e-06, "loss": 1.1334, "step": 16053 }, { "epoch": 1.7082357948499682, "grad_norm": 13.2265625, "learning_rate": 9.44734223990904e-06, "loss": 1.1778, "step": 16054 }, { "epoch": 1.7083422004681847, "grad_norm": 14.5078125, "learning_rate": 9.447306708357021e-06, "loss": 1.0376, "step": 16055 }, { "epoch": 1.7084486060864013, "grad_norm": 9.7265625, "learning_rate": 9.447271176805005e-06, "loss": 0.9289, "step": 16056 }, { "epoch": 1.708555011704618, "grad_norm": 14.59375, "learning_rate": 9.447235645252986e-06, "loss": 1.3388, "step": 16057 }, { "epoch": 1.7086614173228347, "grad_norm": 14.7734375, "learning_rate": 9.447200113700968e-06, "loss": 1.0594, "step": 16058 }, { "epoch": 1.7087678229410512, "grad_norm": 14.3984375, "learning_rate": 9.44716458214895e-06, "loss": 1.2235, "step": 16059 }, { "epoch": 1.708874228559268, "grad_norm": 12.2109375, "learning_rate": 9.447129050596931e-06, "loss": 1.1289, "step": 16060 }, { "epoch": 1.7089806341774847, "grad_norm": 13.09375, "learning_rate": 9.447093519044913e-06, "loss": 1.2529, "step": 16061 }, { "epoch": 1.7090870397957012, "grad_norm": 14.7109375, "learning_rate": 9.447057987492894e-06, "loss": 1.3101, "step": 16062 }, { "epoch": 1.7091934454139177, "grad_norm": 10.7265625, "learning_rate": 9.447022455940876e-06, "loss": 1.0848, "step": 16063 }, { "epoch": 1.7092998510321344, "grad_norm": 14.4296875, "learning_rate": 9.446986924388858e-06, "loss": 1.1019, "step": 16064 }, { "epoch": 1.7094062566503512, "grad_norm": 12.5078125, "learning_rate": 9.446951392836839e-06, "loss": 0.9093, "step": 16065 }, { "epoch": 1.7095126622685677, "grad_norm": 13.1640625, "learning_rate": 9.446915861284822e-06, "loss": 0.7354, "step": 16066 }, { "epoch": 1.7096190678867844, "grad_norm": 13.96875, "learning_rate": 9.446880329732804e-06, "loss": 1.0032, "step": 16067 }, { "epoch": 1.7097254735050011, "grad_norm": 11.0625, "learning_rate": 9.446844798180786e-06, "loss": 1.1928, "step": 16068 }, { "epoch": 1.7098318791232177, "grad_norm": 10.625, "learning_rate": 9.446809266628767e-06, "loss": 1.0896, "step": 16069 }, { "epoch": 1.7099382847414344, "grad_norm": 17.0625, "learning_rate": 9.446773735076749e-06, "loss": 1.1177, "step": 16070 }, { "epoch": 1.7100446903596511, "grad_norm": 11.5078125, "learning_rate": 9.44673820352473e-06, "loss": 0.9972, "step": 16071 }, { "epoch": 1.7101510959778676, "grad_norm": 14.1796875, "learning_rate": 9.446702671972712e-06, "loss": 1.0653, "step": 16072 }, { "epoch": 1.7102575015960841, "grad_norm": 9.140625, "learning_rate": 9.446667140420694e-06, "loss": 0.8738, "step": 16073 }, { "epoch": 1.7103639072143009, "grad_norm": 14.09375, "learning_rate": 9.446631608868675e-06, "loss": 1.2491, "step": 16074 }, { "epoch": 1.7104703128325176, "grad_norm": 12.921875, "learning_rate": 9.446596077316657e-06, "loss": 0.9221, "step": 16075 }, { "epoch": 1.7105767184507341, "grad_norm": 11.625, "learning_rate": 9.44656054576464e-06, "loss": 1.3808, "step": 16076 }, { "epoch": 1.7106831240689508, "grad_norm": 16.375, "learning_rate": 9.446525014212622e-06, "loss": 1.0483, "step": 16077 }, { "epoch": 1.7107895296871676, "grad_norm": 10.2109375, "learning_rate": 9.446489482660603e-06, "loss": 0.9926, "step": 16078 }, { "epoch": 1.710895935305384, "grad_norm": 12.640625, "learning_rate": 9.446453951108585e-06, "loss": 1.1257, "step": 16079 }, { "epoch": 1.7110023409236008, "grad_norm": 17.125, "learning_rate": 9.446418419556567e-06, "loss": 1.2102, "step": 16080 }, { "epoch": 1.7111087465418175, "grad_norm": 10.5390625, "learning_rate": 9.446382888004548e-06, "loss": 1.1127, "step": 16081 }, { "epoch": 1.711215152160034, "grad_norm": 10.9765625, "learning_rate": 9.446347356452532e-06, "loss": 1.1571, "step": 16082 }, { "epoch": 1.7113215577782506, "grad_norm": 14.28125, "learning_rate": 9.446311824900513e-06, "loss": 1.1194, "step": 16083 }, { "epoch": 1.7114279633964673, "grad_norm": 12.359375, "learning_rate": 9.446276293348495e-06, "loss": 1.1134, "step": 16084 }, { "epoch": 1.711534369014684, "grad_norm": 14.6484375, "learning_rate": 9.446240761796476e-06, "loss": 0.9319, "step": 16085 }, { "epoch": 1.7116407746329005, "grad_norm": 16.625, "learning_rate": 9.446205230244458e-06, "loss": 1.303, "step": 16086 }, { "epoch": 1.7117471802511173, "grad_norm": 11.1171875, "learning_rate": 9.44616969869244e-06, "loss": 1.2266, "step": 16087 }, { "epoch": 1.711853585869334, "grad_norm": 16.25, "learning_rate": 9.446134167140421e-06, "loss": 1.0183, "step": 16088 }, { "epoch": 1.7119599914875505, "grad_norm": 13.359375, "learning_rate": 9.446098635588403e-06, "loss": 1.0248, "step": 16089 }, { "epoch": 1.712066397105767, "grad_norm": 8.734375, "learning_rate": 9.446063104036384e-06, "loss": 0.8869, "step": 16090 }, { "epoch": 1.712172802723984, "grad_norm": 12.4453125, "learning_rate": 9.446027572484366e-06, "loss": 1.2196, "step": 16091 }, { "epoch": 1.7122792083422005, "grad_norm": 15.109375, "learning_rate": 9.44599204093235e-06, "loss": 0.6655, "step": 16092 }, { "epoch": 1.712385613960417, "grad_norm": 10.125, "learning_rate": 9.445956509380331e-06, "loss": 1.1464, "step": 16093 }, { "epoch": 1.7124920195786337, "grad_norm": 11.5546875, "learning_rate": 9.445920977828312e-06, "loss": 1.1683, "step": 16094 }, { "epoch": 1.7125984251968505, "grad_norm": 11.953125, "learning_rate": 9.445885446276294e-06, "loss": 1.1745, "step": 16095 }, { "epoch": 1.712704830815067, "grad_norm": 16.5, "learning_rate": 9.445849914724276e-06, "loss": 1.1307, "step": 16096 }, { "epoch": 1.7128112364332837, "grad_norm": 14.4453125, "learning_rate": 9.445814383172259e-06, "loss": 1.155, "step": 16097 }, { "epoch": 1.7129176420515004, "grad_norm": 12.5625, "learning_rate": 9.44577885162024e-06, "loss": 1.1713, "step": 16098 }, { "epoch": 1.713024047669717, "grad_norm": 13.921875, "learning_rate": 9.445743320068222e-06, "loss": 1.1737, "step": 16099 }, { "epoch": 1.7131304532879335, "grad_norm": 16.640625, "learning_rate": 9.445707788516202e-06, "loss": 0.8947, "step": 16100 }, { "epoch": 1.7132368589061504, "grad_norm": 11.7578125, "learning_rate": 9.445672256964184e-06, "loss": 1.2004, "step": 16101 }, { "epoch": 1.713343264524367, "grad_norm": 21.21875, "learning_rate": 9.445636725412167e-06, "loss": 1.4363, "step": 16102 }, { "epoch": 1.7134496701425834, "grad_norm": 8.8046875, "learning_rate": 9.445601193860149e-06, "loss": 1.0461, "step": 16103 }, { "epoch": 1.7135560757608002, "grad_norm": 15.625, "learning_rate": 9.44556566230813e-06, "loss": 1.0895, "step": 16104 }, { "epoch": 1.7136624813790169, "grad_norm": 11.84375, "learning_rate": 9.445530130756112e-06, "loss": 1.2081, "step": 16105 }, { "epoch": 1.7137688869972334, "grad_norm": 17.203125, "learning_rate": 9.445494599204093e-06, "loss": 1.1522, "step": 16106 }, { "epoch": 1.7138752926154501, "grad_norm": 16.671875, "learning_rate": 9.445459067652077e-06, "loss": 0.9322, "step": 16107 }, { "epoch": 1.7139816982336669, "grad_norm": 9.78125, "learning_rate": 9.445423536100058e-06, "loss": 0.9225, "step": 16108 }, { "epoch": 1.7140881038518834, "grad_norm": 18.453125, "learning_rate": 9.44538800454804e-06, "loss": 0.9727, "step": 16109 }, { "epoch": 1.7141945094700999, "grad_norm": 11.9609375, "learning_rate": 9.445352472996022e-06, "loss": 1.1378, "step": 16110 }, { "epoch": 1.7143009150883166, "grad_norm": 11.2890625, "learning_rate": 9.445316941444003e-06, "loss": 1.0989, "step": 16111 }, { "epoch": 1.7144073207065333, "grad_norm": 11.2265625, "learning_rate": 9.445281409891985e-06, "loss": 1.1605, "step": 16112 }, { "epoch": 1.7145137263247499, "grad_norm": 10.6640625, "learning_rate": 9.445245878339966e-06, "loss": 1.1513, "step": 16113 }, { "epoch": 1.7146201319429666, "grad_norm": 15.375, "learning_rate": 9.445210346787948e-06, "loss": 1.1731, "step": 16114 }, { "epoch": 1.7147265375611833, "grad_norm": 10.75, "learning_rate": 9.44517481523593e-06, "loss": 1.3308, "step": 16115 }, { "epoch": 1.7148329431793998, "grad_norm": 10.671875, "learning_rate": 9.445139283683911e-06, "loss": 1.2388, "step": 16116 }, { "epoch": 1.7149393487976166, "grad_norm": 9.578125, "learning_rate": 9.445103752131893e-06, "loss": 0.9897, "step": 16117 }, { "epoch": 1.7150457544158333, "grad_norm": 14.4375, "learning_rate": 9.445068220579876e-06, "loss": 0.8787, "step": 16118 }, { "epoch": 1.7151521600340498, "grad_norm": 15.5234375, "learning_rate": 9.445032689027858e-06, "loss": 1.0399, "step": 16119 }, { "epoch": 1.7152585656522663, "grad_norm": 14.671875, "learning_rate": 9.44499715747584e-06, "loss": 1.0875, "step": 16120 }, { "epoch": 1.715364971270483, "grad_norm": 11.625, "learning_rate": 9.444961625923821e-06, "loss": 1.1248, "step": 16121 }, { "epoch": 1.7154713768886998, "grad_norm": 12.125, "learning_rate": 9.444926094371803e-06, "loss": 1.1055, "step": 16122 }, { "epoch": 1.7155777825069163, "grad_norm": 12.7109375, "learning_rate": 9.444890562819786e-06, "loss": 1.2068, "step": 16123 }, { "epoch": 1.715684188125133, "grad_norm": 12.3828125, "learning_rate": 9.444855031267767e-06, "loss": 0.9944, "step": 16124 }, { "epoch": 1.7157905937433497, "grad_norm": 10.0390625, "learning_rate": 9.444819499715749e-06, "loss": 1.0492, "step": 16125 }, { "epoch": 1.7158969993615663, "grad_norm": 10.6640625, "learning_rate": 9.44478396816373e-06, "loss": 0.8807, "step": 16126 }, { "epoch": 1.716003404979783, "grad_norm": 11.9609375, "learning_rate": 9.444748436611712e-06, "loss": 1.5559, "step": 16127 }, { "epoch": 1.7161098105979997, "grad_norm": 12.1640625, "learning_rate": 9.444712905059694e-06, "loss": 1.0156, "step": 16128 }, { "epoch": 1.7162162162162162, "grad_norm": 17.125, "learning_rate": 9.444677373507675e-06, "loss": 1.1641, "step": 16129 }, { "epoch": 1.7163226218344327, "grad_norm": 13.609375, "learning_rate": 9.444641841955657e-06, "loss": 1.0372, "step": 16130 }, { "epoch": 1.7164290274526495, "grad_norm": 9.625, "learning_rate": 9.444606310403639e-06, "loss": 1.2668, "step": 16131 }, { "epoch": 1.7165354330708662, "grad_norm": 9.5703125, "learning_rate": 9.44457077885162e-06, "loss": 1.1593, "step": 16132 }, { "epoch": 1.7166418386890827, "grad_norm": 10.6171875, "learning_rate": 9.444535247299604e-06, "loss": 1.1564, "step": 16133 }, { "epoch": 1.7167482443072994, "grad_norm": 11.3203125, "learning_rate": 9.444499715747585e-06, "loss": 0.9548, "step": 16134 }, { "epoch": 1.7168546499255162, "grad_norm": 15.1015625, "learning_rate": 9.444464184195567e-06, "loss": 1.1165, "step": 16135 }, { "epoch": 1.7169610555437327, "grad_norm": 11.84375, "learning_rate": 9.444428652643548e-06, "loss": 1.0226, "step": 16136 }, { "epoch": 1.7170674611619492, "grad_norm": 20.203125, "learning_rate": 9.44439312109153e-06, "loss": 1.1054, "step": 16137 }, { "epoch": 1.7171738667801661, "grad_norm": 11.6875, "learning_rate": 9.444357589539512e-06, "loss": 0.9599, "step": 16138 }, { "epoch": 1.7172802723983827, "grad_norm": 10.109375, "learning_rate": 9.444322057987493e-06, "loss": 0.9818, "step": 16139 }, { "epoch": 1.7173866780165992, "grad_norm": 14.6796875, "learning_rate": 9.444286526435475e-06, "loss": 1.1122, "step": 16140 }, { "epoch": 1.717493083634816, "grad_norm": 11.9921875, "learning_rate": 9.444250994883456e-06, "loss": 1.0787, "step": 16141 }, { "epoch": 1.7175994892530326, "grad_norm": 12.3984375, "learning_rate": 9.444215463331438e-06, "loss": 1.187, "step": 16142 }, { "epoch": 1.7177058948712491, "grad_norm": 15.796875, "learning_rate": 9.444179931779421e-06, "loss": 1.097, "step": 16143 }, { "epoch": 1.7178123004894659, "grad_norm": 13.265625, "learning_rate": 9.444144400227403e-06, "loss": 1.0661, "step": 16144 }, { "epoch": 1.7179187061076826, "grad_norm": 11.03125, "learning_rate": 9.444108868675385e-06, "loss": 1.2021, "step": 16145 }, { "epoch": 1.7180251117258991, "grad_norm": 10.40625, "learning_rate": 9.444073337123366e-06, "loss": 1.0586, "step": 16146 }, { "epoch": 1.7181315173441156, "grad_norm": 16.5, "learning_rate": 9.444037805571348e-06, "loss": 1.2828, "step": 16147 }, { "epoch": 1.7182379229623326, "grad_norm": 10.6015625, "learning_rate": 9.444002274019331e-06, "loss": 1.0233, "step": 16148 }, { "epoch": 1.718344328580549, "grad_norm": 25.8125, "learning_rate": 9.443966742467313e-06, "loss": 1.2203, "step": 16149 }, { "epoch": 1.7184507341987656, "grad_norm": 10.625, "learning_rate": 9.443931210915294e-06, "loss": 0.9142, "step": 16150 }, { "epoch": 1.7185571398169823, "grad_norm": 14.7890625, "learning_rate": 9.443895679363276e-06, "loss": 0.927, "step": 16151 }, { "epoch": 1.718663545435199, "grad_norm": 12.6484375, "learning_rate": 9.443860147811258e-06, "loss": 1.0368, "step": 16152 }, { "epoch": 1.7187699510534156, "grad_norm": 14.875, "learning_rate": 9.443824616259239e-06, "loss": 1.0036, "step": 16153 }, { "epoch": 1.7188763566716323, "grad_norm": 9.75, "learning_rate": 9.44378908470722e-06, "loss": 1.0015, "step": 16154 }, { "epoch": 1.718982762289849, "grad_norm": 11.9765625, "learning_rate": 9.443753553155202e-06, "loss": 1.0666, "step": 16155 }, { "epoch": 1.7190891679080655, "grad_norm": 12.2421875, "learning_rate": 9.443718021603184e-06, "loss": 1.2426, "step": 16156 }, { "epoch": 1.719195573526282, "grad_norm": 9.4375, "learning_rate": 9.443682490051166e-06, "loss": 1.1002, "step": 16157 }, { "epoch": 1.7193019791444988, "grad_norm": 9.0390625, "learning_rate": 9.443646958499147e-06, "loss": 0.8898, "step": 16158 }, { "epoch": 1.7194083847627155, "grad_norm": 11.1015625, "learning_rate": 9.44361142694713e-06, "loss": 1.0783, "step": 16159 }, { "epoch": 1.719514790380932, "grad_norm": 12.609375, "learning_rate": 9.443575895395112e-06, "loss": 0.918, "step": 16160 }, { "epoch": 1.7196211959991488, "grad_norm": 18.1875, "learning_rate": 9.443540363843094e-06, "loss": 1.0274, "step": 16161 }, { "epoch": 1.7197276016173655, "grad_norm": 12.25, "learning_rate": 9.443504832291075e-06, "loss": 1.2277, "step": 16162 }, { "epoch": 1.719834007235582, "grad_norm": 13.8671875, "learning_rate": 9.443469300739057e-06, "loss": 1.1714, "step": 16163 }, { "epoch": 1.7199404128537987, "grad_norm": 15.984375, "learning_rate": 9.443433769187039e-06, "loss": 1.2206, "step": 16164 }, { "epoch": 1.7200468184720155, "grad_norm": 18.484375, "learning_rate": 9.44339823763502e-06, "loss": 1.2415, "step": 16165 }, { "epoch": 1.720153224090232, "grad_norm": 12.453125, "learning_rate": 9.443362706083002e-06, "loss": 1.1098, "step": 16166 }, { "epoch": 1.7202596297084485, "grad_norm": 13.5703125, "learning_rate": 9.443327174530983e-06, "loss": 1.553, "step": 16167 }, { "epoch": 1.7203660353266652, "grad_norm": 10.8359375, "learning_rate": 9.443291642978965e-06, "loss": 1.017, "step": 16168 }, { "epoch": 1.720472440944882, "grad_norm": 11.03125, "learning_rate": 9.443256111426948e-06, "loss": 1.0904, "step": 16169 }, { "epoch": 1.7205788465630985, "grad_norm": 9.0078125, "learning_rate": 9.44322057987493e-06, "loss": 0.8773, "step": 16170 }, { "epoch": 1.7206852521813152, "grad_norm": 17.84375, "learning_rate": 9.443185048322911e-06, "loss": 1.0356, "step": 16171 }, { "epoch": 1.720791657799532, "grad_norm": 16.953125, "learning_rate": 9.443149516770893e-06, "loss": 1.1109, "step": 16172 }, { "epoch": 1.7208980634177484, "grad_norm": 10.515625, "learning_rate": 9.443113985218875e-06, "loss": 0.9248, "step": 16173 }, { "epoch": 1.721004469035965, "grad_norm": 11.9765625, "learning_rate": 9.443078453666858e-06, "loss": 0.9951, "step": 16174 }, { "epoch": 1.7211108746541819, "grad_norm": 15.078125, "learning_rate": 9.44304292211484e-06, "loss": 1.0504, "step": 16175 }, { "epoch": 1.7212172802723984, "grad_norm": 22.484375, "learning_rate": 9.443007390562821e-06, "loss": 1.224, "step": 16176 }, { "epoch": 1.721323685890615, "grad_norm": 17.9375, "learning_rate": 9.442971859010803e-06, "loss": 1.2153, "step": 16177 }, { "epoch": 1.7214300915088316, "grad_norm": 10.6328125, "learning_rate": 9.442936327458784e-06, "loss": 1.1032, "step": 16178 }, { "epoch": 1.7215364971270484, "grad_norm": 13.5234375, "learning_rate": 9.442900795906766e-06, "loss": 0.9974, "step": 16179 }, { "epoch": 1.7216429027452649, "grad_norm": 13.0078125, "learning_rate": 9.442865264354748e-06, "loss": 0.9177, "step": 16180 }, { "epoch": 1.7217493083634816, "grad_norm": 12.984375, "learning_rate": 9.44282973280273e-06, "loss": 0.9835, "step": 16181 }, { "epoch": 1.7218557139816983, "grad_norm": 19.890625, "learning_rate": 9.44279420125071e-06, "loss": 0.7471, "step": 16182 }, { "epoch": 1.7219621195999149, "grad_norm": 10.4375, "learning_rate": 9.442758669698692e-06, "loss": 1.1532, "step": 16183 }, { "epoch": 1.7220685252181314, "grad_norm": 13.3125, "learning_rate": 9.442723138146676e-06, "loss": 1.0346, "step": 16184 }, { "epoch": 1.7221749308363483, "grad_norm": 13.4296875, "learning_rate": 9.442687606594657e-06, "loss": 1.0993, "step": 16185 }, { "epoch": 1.7222813364545648, "grad_norm": 12.3515625, "learning_rate": 9.442652075042639e-06, "loss": 1.1691, "step": 16186 }, { "epoch": 1.7223877420727813, "grad_norm": 15.15625, "learning_rate": 9.44261654349062e-06, "loss": 1.1176, "step": 16187 }, { "epoch": 1.722494147690998, "grad_norm": 12.0078125, "learning_rate": 9.442581011938602e-06, "loss": 0.8997, "step": 16188 }, { "epoch": 1.7226005533092148, "grad_norm": 12.125, "learning_rate": 9.442545480386584e-06, "loss": 1.3416, "step": 16189 }, { "epoch": 1.7227069589274313, "grad_norm": 11.578125, "learning_rate": 9.442509948834565e-06, "loss": 1.0916, "step": 16190 }, { "epoch": 1.722813364545648, "grad_norm": 12.1953125, "learning_rate": 9.442474417282547e-06, "loss": 0.9313, "step": 16191 }, { "epoch": 1.7229197701638648, "grad_norm": 11.8671875, "learning_rate": 9.442438885730529e-06, "loss": 1.1324, "step": 16192 }, { "epoch": 1.7230261757820813, "grad_norm": 18.53125, "learning_rate": 9.44240335417851e-06, "loss": 1.0864, "step": 16193 }, { "epoch": 1.7231325814002978, "grad_norm": 20.8125, "learning_rate": 9.442367822626493e-06, "loss": 1.3844, "step": 16194 }, { "epoch": 1.7232389870185145, "grad_norm": 12.5, "learning_rate": 9.442332291074475e-06, "loss": 0.9312, "step": 16195 }, { "epoch": 1.7233453926367313, "grad_norm": 10.8515625, "learning_rate": 9.442296759522457e-06, "loss": 1.1278, "step": 16196 }, { "epoch": 1.7234517982549478, "grad_norm": 17.84375, "learning_rate": 9.442261227970438e-06, "loss": 1.1235, "step": 16197 }, { "epoch": 1.7235582038731645, "grad_norm": 10.6484375, "learning_rate": 9.44222569641842e-06, "loss": 0.96, "step": 16198 }, { "epoch": 1.7236646094913812, "grad_norm": 13.6328125, "learning_rate": 9.442190164866402e-06, "loss": 1.0332, "step": 16199 }, { "epoch": 1.7237710151095977, "grad_norm": 10.3671875, "learning_rate": 9.442154633314385e-06, "loss": 1.2615, "step": 16200 }, { "epoch": 1.7238774207278145, "grad_norm": 14.9921875, "learning_rate": 9.442119101762366e-06, "loss": 1.1699, "step": 16201 }, { "epoch": 1.7239838263460312, "grad_norm": 19.625, "learning_rate": 9.442083570210348e-06, "loss": 0.6273, "step": 16202 }, { "epoch": 1.7240902319642477, "grad_norm": 11.0703125, "learning_rate": 9.44204803865833e-06, "loss": 1.0068, "step": 16203 }, { "epoch": 1.7241966375824642, "grad_norm": 21.40625, "learning_rate": 9.442012507106311e-06, "loss": 1.2417, "step": 16204 }, { "epoch": 1.724303043200681, "grad_norm": 13.0546875, "learning_rate": 9.441976975554293e-06, "loss": 0.853, "step": 16205 }, { "epoch": 1.7244094488188977, "grad_norm": 14.5859375, "learning_rate": 9.441941444002274e-06, "loss": 1.0851, "step": 16206 }, { "epoch": 1.7245158544371142, "grad_norm": 11.421875, "learning_rate": 9.441905912450256e-06, "loss": 1.0657, "step": 16207 }, { "epoch": 1.724622260055331, "grad_norm": 12.109375, "learning_rate": 9.441870380898238e-06, "loss": 1.1099, "step": 16208 }, { "epoch": 1.7247286656735477, "grad_norm": 13.09375, "learning_rate": 9.44183484934622e-06, "loss": 1.2322, "step": 16209 }, { "epoch": 1.7248350712917642, "grad_norm": 13.7421875, "learning_rate": 9.441799317794203e-06, "loss": 1.0264, "step": 16210 }, { "epoch": 1.724941476909981, "grad_norm": 13.640625, "learning_rate": 9.441763786242184e-06, "loss": 1.0515, "step": 16211 }, { "epoch": 1.7250478825281976, "grad_norm": 11.4453125, "learning_rate": 9.441728254690166e-06, "loss": 0.9294, "step": 16212 }, { "epoch": 1.7251542881464141, "grad_norm": 18.453125, "learning_rate": 9.441692723138147e-06, "loss": 0.9093, "step": 16213 }, { "epoch": 1.7252606937646306, "grad_norm": 12.8671875, "learning_rate": 9.441657191586129e-06, "loss": 1.2281, "step": 16214 }, { "epoch": 1.7253670993828474, "grad_norm": 15.1953125, "learning_rate": 9.441621660034112e-06, "loss": 1.0792, "step": 16215 }, { "epoch": 1.7254735050010641, "grad_norm": 10.8671875, "learning_rate": 9.441586128482094e-06, "loss": 0.9978, "step": 16216 }, { "epoch": 1.7255799106192806, "grad_norm": 11.796875, "learning_rate": 9.441550596930076e-06, "loss": 0.8671, "step": 16217 }, { "epoch": 1.7256863162374974, "grad_norm": 15.359375, "learning_rate": 9.441515065378057e-06, "loss": 1.3895, "step": 16218 }, { "epoch": 1.725792721855714, "grad_norm": 17.515625, "learning_rate": 9.441479533826037e-06, "loss": 0.8945, "step": 16219 }, { "epoch": 1.7258991274739306, "grad_norm": 13.046875, "learning_rate": 9.44144400227402e-06, "loss": 1.1301, "step": 16220 }, { "epoch": 1.726005533092147, "grad_norm": 15.796875, "learning_rate": 9.441408470722002e-06, "loss": 1.0339, "step": 16221 }, { "epoch": 1.726111938710364, "grad_norm": 11.3125, "learning_rate": 9.441372939169984e-06, "loss": 1.2574, "step": 16222 }, { "epoch": 1.7262183443285806, "grad_norm": 13.984375, "learning_rate": 9.441337407617965e-06, "loss": 1.1469, "step": 16223 }, { "epoch": 1.726324749946797, "grad_norm": 10.4609375, "learning_rate": 9.441301876065947e-06, "loss": 1.0711, "step": 16224 }, { "epoch": 1.7264311555650138, "grad_norm": 12.3671875, "learning_rate": 9.44126634451393e-06, "loss": 1.1891, "step": 16225 }, { "epoch": 1.7265375611832305, "grad_norm": 12.1796875, "learning_rate": 9.441230812961912e-06, "loss": 1.4122, "step": 16226 }, { "epoch": 1.726643966801447, "grad_norm": 19.5625, "learning_rate": 9.441195281409893e-06, "loss": 0.9727, "step": 16227 }, { "epoch": 1.7267503724196638, "grad_norm": 10.0234375, "learning_rate": 9.441159749857875e-06, "loss": 1.079, "step": 16228 }, { "epoch": 1.7268567780378805, "grad_norm": 14.9375, "learning_rate": 9.441124218305856e-06, "loss": 0.9904, "step": 16229 }, { "epoch": 1.726963183656097, "grad_norm": 10.7734375, "learning_rate": 9.441088686753838e-06, "loss": 1.0737, "step": 16230 }, { "epoch": 1.7270695892743135, "grad_norm": 10.8203125, "learning_rate": 9.44105315520182e-06, "loss": 0.8569, "step": 16231 }, { "epoch": 1.7271759948925305, "grad_norm": 14.21875, "learning_rate": 9.441017623649801e-06, "loss": 1.3231, "step": 16232 }, { "epoch": 1.727282400510747, "grad_norm": 13.5859375, "learning_rate": 9.440982092097783e-06, "loss": 1.2111, "step": 16233 }, { "epoch": 1.7273888061289635, "grad_norm": 11.1640625, "learning_rate": 9.440946560545765e-06, "loss": 1.1307, "step": 16234 }, { "epoch": 1.7274952117471802, "grad_norm": 14.953125, "learning_rate": 9.440911028993746e-06, "loss": 1.0747, "step": 16235 }, { "epoch": 1.727601617365397, "grad_norm": 15.9765625, "learning_rate": 9.44087549744173e-06, "loss": 0.9896, "step": 16236 }, { "epoch": 1.7277080229836135, "grad_norm": 19.75, "learning_rate": 9.440839965889711e-06, "loss": 1.0305, "step": 16237 }, { "epoch": 1.7278144286018302, "grad_norm": 13.859375, "learning_rate": 9.440804434337693e-06, "loss": 0.997, "step": 16238 }, { "epoch": 1.727920834220047, "grad_norm": 12.6796875, "learning_rate": 9.440768902785674e-06, "loss": 1.2117, "step": 16239 }, { "epoch": 1.7280272398382635, "grad_norm": 11.578125, "learning_rate": 9.440733371233656e-06, "loss": 0.931, "step": 16240 }, { "epoch": 1.72813364545648, "grad_norm": 8.390625, "learning_rate": 9.44069783968164e-06, "loss": 1.1012, "step": 16241 }, { "epoch": 1.7282400510746967, "grad_norm": 10.21875, "learning_rate": 9.44066230812962e-06, "loss": 1.1723, "step": 16242 }, { "epoch": 1.7283464566929134, "grad_norm": 9.609375, "learning_rate": 9.440626776577602e-06, "loss": 1.077, "step": 16243 }, { "epoch": 1.72845286231113, "grad_norm": 13.4921875, "learning_rate": 9.440591245025584e-06, "loss": 1.1555, "step": 16244 }, { "epoch": 1.7285592679293467, "grad_norm": 17.578125, "learning_rate": 9.440555713473566e-06, "loss": 0.9657, "step": 16245 }, { "epoch": 1.7286656735475634, "grad_norm": 20.234375, "learning_rate": 9.440520181921547e-06, "loss": 0.9562, "step": 16246 }, { "epoch": 1.72877207916578, "grad_norm": 14.0546875, "learning_rate": 9.440484650369529e-06, "loss": 0.8543, "step": 16247 }, { "epoch": 1.7288784847839966, "grad_norm": 18.953125, "learning_rate": 9.44044911881751e-06, "loss": 1.0271, "step": 16248 }, { "epoch": 1.7289848904022134, "grad_norm": 12.1640625, "learning_rate": 9.440413587265492e-06, "loss": 1.1309, "step": 16249 }, { "epoch": 1.7290912960204299, "grad_norm": 16.140625, "learning_rate": 9.440378055713474e-06, "loss": 1.4427, "step": 16250 }, { "epoch": 1.7291977016386464, "grad_norm": 9.203125, "learning_rate": 9.440342524161457e-06, "loss": 1.0226, "step": 16251 }, { "epoch": 1.7293041072568631, "grad_norm": 18.78125, "learning_rate": 9.440306992609439e-06, "loss": 0.96, "step": 16252 }, { "epoch": 1.7294105128750799, "grad_norm": 16.328125, "learning_rate": 9.44027146105742e-06, "loss": 1.0734, "step": 16253 }, { "epoch": 1.7295169184932964, "grad_norm": 15.171875, "learning_rate": 9.440235929505402e-06, "loss": 1.0378, "step": 16254 }, { "epoch": 1.729623324111513, "grad_norm": 16.25, "learning_rate": 9.440200397953383e-06, "loss": 1.1779, "step": 16255 }, { "epoch": 1.7297297297297298, "grad_norm": 11.640625, "learning_rate": 9.440164866401365e-06, "loss": 1.0042, "step": 16256 }, { "epoch": 1.7298361353479463, "grad_norm": 10.4609375, "learning_rate": 9.440129334849347e-06, "loss": 1.2561, "step": 16257 }, { "epoch": 1.729942540966163, "grad_norm": 17.0, "learning_rate": 9.440093803297328e-06, "loss": 0.8655, "step": 16258 }, { "epoch": 1.7300489465843798, "grad_norm": 11.421875, "learning_rate": 9.44005827174531e-06, "loss": 1.1485, "step": 16259 }, { "epoch": 1.7301553522025963, "grad_norm": 11.5078125, "learning_rate": 9.440022740193291e-06, "loss": 1.2072, "step": 16260 }, { "epoch": 1.7302617578208128, "grad_norm": 10.3046875, "learning_rate": 9.439987208641275e-06, "loss": 1.0415, "step": 16261 }, { "epoch": 1.7303681634390295, "grad_norm": 13.578125, "learning_rate": 9.439951677089256e-06, "loss": 1.0465, "step": 16262 }, { "epoch": 1.7304745690572463, "grad_norm": 13.8046875, "learning_rate": 9.439916145537238e-06, "loss": 1.1399, "step": 16263 }, { "epoch": 1.7305809746754628, "grad_norm": 11.25, "learning_rate": 9.43988061398522e-06, "loss": 0.972, "step": 16264 }, { "epoch": 1.7306873802936795, "grad_norm": 12.3203125, "learning_rate": 9.439845082433201e-06, "loss": 1.2839, "step": 16265 }, { "epoch": 1.7307937859118963, "grad_norm": 13.703125, "learning_rate": 9.439809550881184e-06, "loss": 1.2111, "step": 16266 }, { "epoch": 1.7309001915301128, "grad_norm": 16.546875, "learning_rate": 9.439774019329166e-06, "loss": 0.9184, "step": 16267 }, { "epoch": 1.7310065971483293, "grad_norm": 11.8515625, "learning_rate": 9.439738487777148e-06, "loss": 0.9551, "step": 16268 }, { "epoch": 1.7311130027665462, "grad_norm": 19.390625, "learning_rate": 9.43970295622513e-06, "loss": 0.9792, "step": 16269 }, { "epoch": 1.7312194083847627, "grad_norm": 10.1484375, "learning_rate": 9.43966742467311e-06, "loss": 1.18, "step": 16270 }, { "epoch": 1.7313258140029792, "grad_norm": 11.4140625, "learning_rate": 9.439631893121092e-06, "loss": 1.0349, "step": 16271 }, { "epoch": 1.731432219621196, "grad_norm": 14.5859375, "learning_rate": 9.439596361569074e-06, "loss": 1.3262, "step": 16272 }, { "epoch": 1.7315386252394127, "grad_norm": 12.953125, "learning_rate": 9.439560830017056e-06, "loss": 1.1884, "step": 16273 }, { "epoch": 1.7316450308576292, "grad_norm": 12.7890625, "learning_rate": 9.439525298465037e-06, "loss": 0.8878, "step": 16274 }, { "epoch": 1.731751436475846, "grad_norm": 16.140625, "learning_rate": 9.439489766913019e-06, "loss": 1.1746, "step": 16275 }, { "epoch": 1.7318578420940627, "grad_norm": 10.8203125, "learning_rate": 9.439454235361e-06, "loss": 1.1346, "step": 16276 }, { "epoch": 1.7319642477122792, "grad_norm": 17.640625, "learning_rate": 9.439418703808984e-06, "loss": 1.063, "step": 16277 }, { "epoch": 1.7320706533304957, "grad_norm": 15.1796875, "learning_rate": 9.439383172256965e-06, "loss": 1.0672, "step": 16278 }, { "epoch": 1.7321770589487127, "grad_norm": 11.6484375, "learning_rate": 9.439347640704947e-06, "loss": 1.012, "step": 16279 }, { "epoch": 1.7322834645669292, "grad_norm": 11.9765625, "learning_rate": 9.439312109152929e-06, "loss": 1.3579, "step": 16280 }, { "epoch": 1.7323898701851457, "grad_norm": 15.2109375, "learning_rate": 9.43927657760091e-06, "loss": 1.0393, "step": 16281 }, { "epoch": 1.7324962758033624, "grad_norm": 14.6640625, "learning_rate": 9.439241046048892e-06, "loss": 0.979, "step": 16282 }, { "epoch": 1.7326026814215791, "grad_norm": 12.671875, "learning_rate": 9.439205514496873e-06, "loss": 1.3508, "step": 16283 }, { "epoch": 1.7327090870397956, "grad_norm": 13.0, "learning_rate": 9.439169982944855e-06, "loss": 1.0285, "step": 16284 }, { "epoch": 1.7328154926580124, "grad_norm": 15.1484375, "learning_rate": 9.439134451392837e-06, "loss": 1.8211, "step": 16285 }, { "epoch": 1.7329218982762291, "grad_norm": 13.390625, "learning_rate": 9.439098919840818e-06, "loss": 0.9573, "step": 16286 }, { "epoch": 1.7330283038944456, "grad_norm": 10.9609375, "learning_rate": 9.439063388288802e-06, "loss": 1.103, "step": 16287 }, { "epoch": 1.7331347095126621, "grad_norm": 17.1875, "learning_rate": 9.439027856736783e-06, "loss": 1.1589, "step": 16288 }, { "epoch": 1.7332411151308789, "grad_norm": 12.390625, "learning_rate": 9.438992325184765e-06, "loss": 0.955, "step": 16289 }, { "epoch": 1.7333475207490956, "grad_norm": 9.578125, "learning_rate": 9.438956793632746e-06, "loss": 0.8147, "step": 16290 }, { "epoch": 1.733453926367312, "grad_norm": 10.765625, "learning_rate": 9.438921262080728e-06, "loss": 1.1092, "step": 16291 }, { "epoch": 1.7335603319855288, "grad_norm": 10.1328125, "learning_rate": 9.438885730528711e-06, "loss": 1.1072, "step": 16292 }, { "epoch": 1.7336667376037456, "grad_norm": 13.0625, "learning_rate": 9.438850198976693e-06, "loss": 1.1599, "step": 16293 }, { "epoch": 1.733773143221962, "grad_norm": 9.328125, "learning_rate": 9.438814667424674e-06, "loss": 0.9477, "step": 16294 }, { "epoch": 1.7338795488401788, "grad_norm": 12.8125, "learning_rate": 9.438779135872656e-06, "loss": 1.0137, "step": 16295 }, { "epoch": 1.7339859544583955, "grad_norm": 11.921875, "learning_rate": 9.438743604320638e-06, "loss": 1.0834, "step": 16296 }, { "epoch": 1.734092360076612, "grad_norm": 12.8046875, "learning_rate": 9.43870807276862e-06, "loss": 1.1205, "step": 16297 }, { "epoch": 1.7341987656948286, "grad_norm": 10.5625, "learning_rate": 9.438672541216601e-06, "loss": 1.3322, "step": 16298 }, { "epoch": 1.7343051713130453, "grad_norm": 12.25, "learning_rate": 9.438637009664583e-06, "loss": 0.7566, "step": 16299 }, { "epoch": 1.734411576931262, "grad_norm": 9.46875, "learning_rate": 9.438601478112564e-06, "loss": 0.9477, "step": 16300 }, { "epoch": 1.7345179825494785, "grad_norm": 15.84375, "learning_rate": 9.438565946560546e-06, "loss": 1.0613, "step": 16301 }, { "epoch": 1.7346243881676953, "grad_norm": 14.6015625, "learning_rate": 9.438530415008529e-06, "loss": 0.9706, "step": 16302 }, { "epoch": 1.734730793785912, "grad_norm": 13.2734375, "learning_rate": 9.43849488345651e-06, "loss": 1.3321, "step": 16303 }, { "epoch": 1.7348371994041285, "grad_norm": 12.0546875, "learning_rate": 9.438459351904492e-06, "loss": 0.9774, "step": 16304 }, { "epoch": 1.734943605022345, "grad_norm": 10.2578125, "learning_rate": 9.438423820352474e-06, "loss": 1.0972, "step": 16305 }, { "epoch": 1.735050010640562, "grad_norm": 12.3125, "learning_rate": 9.438388288800455e-06, "loss": 1.2346, "step": 16306 }, { "epoch": 1.7351564162587785, "grad_norm": 12.9375, "learning_rate": 9.438352757248439e-06, "loss": 1.3862, "step": 16307 }, { "epoch": 1.735262821876995, "grad_norm": 10.4375, "learning_rate": 9.438317225696419e-06, "loss": 0.8935, "step": 16308 }, { "epoch": 1.7353692274952117, "grad_norm": 14.859375, "learning_rate": 9.4382816941444e-06, "loss": 1.1516, "step": 16309 }, { "epoch": 1.7354756331134285, "grad_norm": 9.890625, "learning_rate": 9.438246162592382e-06, "loss": 1.1179, "step": 16310 }, { "epoch": 1.735582038731645, "grad_norm": 13.765625, "learning_rate": 9.438210631040363e-06, "loss": 1.0748, "step": 16311 }, { "epoch": 1.7356884443498617, "grad_norm": 13.8203125, "learning_rate": 9.438175099488345e-06, "loss": 1.1057, "step": 16312 }, { "epoch": 1.7357948499680784, "grad_norm": 9.546875, "learning_rate": 9.438139567936328e-06, "loss": 0.8778, "step": 16313 }, { "epoch": 1.735901255586295, "grad_norm": 9.9375, "learning_rate": 9.43810403638431e-06, "loss": 1.0154, "step": 16314 }, { "epoch": 1.7360076612045114, "grad_norm": 12.8828125, "learning_rate": 9.438068504832292e-06, "loss": 1.2209, "step": 16315 }, { "epoch": 1.7361140668227284, "grad_norm": 14.9765625, "learning_rate": 9.438032973280273e-06, "loss": 1.1719, "step": 16316 }, { "epoch": 1.736220472440945, "grad_norm": 16.90625, "learning_rate": 9.437997441728255e-06, "loss": 1.0584, "step": 16317 }, { "epoch": 1.7363268780591614, "grad_norm": 12.421875, "learning_rate": 9.437961910176238e-06, "loss": 1.2096, "step": 16318 }, { "epoch": 1.7364332836773781, "grad_norm": 18.21875, "learning_rate": 9.43792637862422e-06, "loss": 0.822, "step": 16319 }, { "epoch": 1.7365396892955949, "grad_norm": 19.234375, "learning_rate": 9.437890847072201e-06, "loss": 0.9399, "step": 16320 }, { "epoch": 1.7366460949138114, "grad_norm": 13.7890625, "learning_rate": 9.437855315520183e-06, "loss": 1.2428, "step": 16321 }, { "epoch": 1.7367525005320281, "grad_norm": 14.0234375, "learning_rate": 9.437819783968165e-06, "loss": 1.0946, "step": 16322 }, { "epoch": 1.7368589061502449, "grad_norm": 13.6484375, "learning_rate": 9.437784252416146e-06, "loss": 1.1207, "step": 16323 }, { "epoch": 1.7369653117684614, "grad_norm": 9.5703125, "learning_rate": 9.437748720864128e-06, "loss": 0.828, "step": 16324 }, { "epoch": 1.7370717173866779, "grad_norm": 21.09375, "learning_rate": 9.43771318931211e-06, "loss": 1.1803, "step": 16325 }, { "epoch": 1.7371781230048946, "grad_norm": 12.75, "learning_rate": 9.437677657760091e-06, "loss": 1.513, "step": 16326 }, { "epoch": 1.7372845286231113, "grad_norm": 10.75, "learning_rate": 9.437642126208073e-06, "loss": 1.1574, "step": 16327 }, { "epoch": 1.7373909342413278, "grad_norm": 11.5546875, "learning_rate": 9.437606594656056e-06, "loss": 1.0502, "step": 16328 }, { "epoch": 1.7374973398595446, "grad_norm": 13.78125, "learning_rate": 9.437571063104037e-06, "loss": 1.2063, "step": 16329 }, { "epoch": 1.7376037454777613, "grad_norm": 10.65625, "learning_rate": 9.437535531552019e-06, "loss": 1.033, "step": 16330 }, { "epoch": 1.7377101510959778, "grad_norm": 15.3828125, "learning_rate": 9.4375e-06, "loss": 1.0706, "step": 16331 }, { "epoch": 1.7378165567141945, "grad_norm": 12.03125, "learning_rate": 9.437464468447982e-06, "loss": 1.0094, "step": 16332 }, { "epoch": 1.7379229623324113, "grad_norm": 15.0, "learning_rate": 9.437428936895966e-06, "loss": 1.1451, "step": 16333 }, { "epoch": 1.7380293679506278, "grad_norm": 11.109375, "learning_rate": 9.437393405343947e-06, "loss": 1.3272, "step": 16334 }, { "epoch": 1.7381357735688443, "grad_norm": 14.65625, "learning_rate": 9.437357873791929e-06, "loss": 1.2749, "step": 16335 }, { "epoch": 1.738242179187061, "grad_norm": 10.796875, "learning_rate": 9.43732234223991e-06, "loss": 1.1605, "step": 16336 }, { "epoch": 1.7383485848052778, "grad_norm": 13.2421875, "learning_rate": 9.437286810687892e-06, "loss": 1.1816, "step": 16337 }, { "epoch": 1.7384549904234943, "grad_norm": 14.9453125, "learning_rate": 9.437251279135874e-06, "loss": 1.2712, "step": 16338 }, { "epoch": 1.738561396041711, "grad_norm": 14.0390625, "learning_rate": 9.437215747583855e-06, "loss": 1.1895, "step": 16339 }, { "epoch": 1.7386678016599277, "grad_norm": 13.21875, "learning_rate": 9.437180216031837e-06, "loss": 0.9994, "step": 16340 }, { "epoch": 1.7387742072781442, "grad_norm": 11.3203125, "learning_rate": 9.437144684479818e-06, "loss": 1.0767, "step": 16341 }, { "epoch": 1.738880612896361, "grad_norm": 13.7734375, "learning_rate": 9.4371091529278e-06, "loss": 1.1124, "step": 16342 }, { "epoch": 1.7389870185145777, "grad_norm": 13.3359375, "learning_rate": 9.437073621375783e-06, "loss": 1.4195, "step": 16343 }, { "epoch": 1.7390934241327942, "grad_norm": 13.703125, "learning_rate": 9.437038089823765e-06, "loss": 1.2291, "step": 16344 }, { "epoch": 1.7391998297510107, "grad_norm": 13.1328125, "learning_rate": 9.437002558271747e-06, "loss": 1.335, "step": 16345 }, { "epoch": 1.7393062353692275, "grad_norm": 12.4375, "learning_rate": 9.436967026719728e-06, "loss": 1.1969, "step": 16346 }, { "epoch": 1.7394126409874442, "grad_norm": 17.5625, "learning_rate": 9.43693149516771e-06, "loss": 1.1033, "step": 16347 }, { "epoch": 1.7395190466056607, "grad_norm": 12.296875, "learning_rate": 9.436895963615691e-06, "loss": 0.9632, "step": 16348 }, { "epoch": 1.7396254522238774, "grad_norm": 19.6875, "learning_rate": 9.436860432063673e-06, "loss": 1.048, "step": 16349 }, { "epoch": 1.7397318578420942, "grad_norm": 13.6796875, "learning_rate": 9.436824900511655e-06, "loss": 1.143, "step": 16350 }, { "epoch": 1.7398382634603107, "grad_norm": 17.515625, "learning_rate": 9.436789368959636e-06, "loss": 1.2338, "step": 16351 }, { "epoch": 1.7399446690785272, "grad_norm": 16.28125, "learning_rate": 9.436753837407618e-06, "loss": 0.9893, "step": 16352 }, { "epoch": 1.7400510746967441, "grad_norm": 12.6171875, "learning_rate": 9.4367183058556e-06, "loss": 1.3607, "step": 16353 }, { "epoch": 1.7401574803149606, "grad_norm": 12.8046875, "learning_rate": 9.436682774303583e-06, "loss": 1.2263, "step": 16354 }, { "epoch": 1.7402638859331772, "grad_norm": 18.03125, "learning_rate": 9.436647242751564e-06, "loss": 1.1158, "step": 16355 }, { "epoch": 1.7403702915513939, "grad_norm": 16.28125, "learning_rate": 9.436611711199546e-06, "loss": 1.0793, "step": 16356 }, { "epoch": 1.7404766971696106, "grad_norm": 15.1171875, "learning_rate": 9.436576179647528e-06, "loss": 1.2747, "step": 16357 }, { "epoch": 1.7405831027878271, "grad_norm": 12.65625, "learning_rate": 9.43654064809551e-06, "loss": 1.0733, "step": 16358 }, { "epoch": 1.7406895084060439, "grad_norm": 14.65625, "learning_rate": 9.436505116543492e-06, "loss": 1.3363, "step": 16359 }, { "epoch": 1.7407959140242606, "grad_norm": 14.1484375, "learning_rate": 9.436469584991474e-06, "loss": 1.0409, "step": 16360 }, { "epoch": 1.740902319642477, "grad_norm": 14.5234375, "learning_rate": 9.436434053439456e-06, "loss": 1.2774, "step": 16361 }, { "epoch": 1.7410087252606936, "grad_norm": 16.40625, "learning_rate": 9.436398521887437e-06, "loss": 1.0258, "step": 16362 }, { "epoch": 1.7411151308789106, "grad_norm": 12.25, "learning_rate": 9.436362990335419e-06, "loss": 1.3763, "step": 16363 }, { "epoch": 1.741221536497127, "grad_norm": 11.1171875, "learning_rate": 9.4363274587834e-06, "loss": 1.0243, "step": 16364 }, { "epoch": 1.7413279421153436, "grad_norm": 16.21875, "learning_rate": 9.436291927231382e-06, "loss": 0.6456, "step": 16365 }, { "epoch": 1.7414343477335603, "grad_norm": 10.265625, "learning_rate": 9.436256395679364e-06, "loss": 1.2359, "step": 16366 }, { "epoch": 1.741540753351777, "grad_norm": 16.59375, "learning_rate": 9.436220864127345e-06, "loss": 1.0287, "step": 16367 }, { "epoch": 1.7416471589699936, "grad_norm": 16.09375, "learning_rate": 9.436185332575327e-06, "loss": 1.1435, "step": 16368 }, { "epoch": 1.7417535645882103, "grad_norm": 11.8828125, "learning_rate": 9.43614980102331e-06, "loss": 0.9436, "step": 16369 }, { "epoch": 1.741859970206427, "grad_norm": 14.7421875, "learning_rate": 9.436114269471292e-06, "loss": 0.9951, "step": 16370 }, { "epoch": 1.7419663758246435, "grad_norm": 11.515625, "learning_rate": 9.436078737919273e-06, "loss": 1.1294, "step": 16371 }, { "epoch": 1.74207278144286, "grad_norm": 13.1640625, "learning_rate": 9.436043206367255e-06, "loss": 0.972, "step": 16372 }, { "epoch": 1.7421791870610768, "grad_norm": 12.5546875, "learning_rate": 9.436007674815237e-06, "loss": 1.0639, "step": 16373 }, { "epoch": 1.7422855926792935, "grad_norm": 15.9609375, "learning_rate": 9.435972143263218e-06, "loss": 1.2068, "step": 16374 }, { "epoch": 1.74239199829751, "grad_norm": 14.0078125, "learning_rate": 9.4359366117112e-06, "loss": 0.7962, "step": 16375 }, { "epoch": 1.7424984039157267, "grad_norm": 12.328125, "learning_rate": 9.435901080159181e-06, "loss": 1.0877, "step": 16376 }, { "epoch": 1.7426048095339435, "grad_norm": 11.515625, "learning_rate": 9.435865548607163e-06, "loss": 1.3313, "step": 16377 }, { "epoch": 1.74271121515216, "grad_norm": 14.4921875, "learning_rate": 9.435830017055145e-06, "loss": 1.3498, "step": 16378 }, { "epoch": 1.7428176207703767, "grad_norm": 10.578125, "learning_rate": 9.435794485503128e-06, "loss": 1.2459, "step": 16379 }, { "epoch": 1.7429240263885935, "grad_norm": 13.84375, "learning_rate": 9.43575895395111e-06, "loss": 1.1345, "step": 16380 }, { "epoch": 1.74303043200681, "grad_norm": 11.7421875, "learning_rate": 9.435723422399091e-06, "loss": 1.0387, "step": 16381 }, { "epoch": 1.7431368376250265, "grad_norm": 13.359375, "learning_rate": 9.435687890847073e-06, "loss": 1.1128, "step": 16382 }, { "epoch": 1.7432432432432432, "grad_norm": 11.6015625, "learning_rate": 9.435652359295054e-06, "loss": 0.833, "step": 16383 }, { "epoch": 1.74334964886146, "grad_norm": 12.265625, "learning_rate": 9.435616827743038e-06, "loss": 0.9424, "step": 16384 }, { "epoch": 1.7434560544796764, "grad_norm": 11.3046875, "learning_rate": 9.43558129619102e-06, "loss": 0.9568, "step": 16385 }, { "epoch": 1.7435624600978932, "grad_norm": 9.5, "learning_rate": 9.435545764639001e-06, "loss": 0.8169, "step": 16386 }, { "epoch": 1.74366886571611, "grad_norm": 11.7890625, "learning_rate": 9.435510233086983e-06, "loss": 1.0288, "step": 16387 }, { "epoch": 1.7437752713343264, "grad_norm": 16.09375, "learning_rate": 9.435474701534964e-06, "loss": 0.8471, "step": 16388 }, { "epoch": 1.7438816769525431, "grad_norm": 11.75, "learning_rate": 9.435439169982946e-06, "loss": 1.0982, "step": 16389 }, { "epoch": 1.7439880825707599, "grad_norm": 11.9296875, "learning_rate": 9.435403638430927e-06, "loss": 0.956, "step": 16390 }, { "epoch": 1.7440944881889764, "grad_norm": 13.5625, "learning_rate": 9.435368106878909e-06, "loss": 1.1274, "step": 16391 }, { "epoch": 1.744200893807193, "grad_norm": 14.4609375, "learning_rate": 9.43533257532689e-06, "loss": 0.9544, "step": 16392 }, { "epoch": 1.7443072994254096, "grad_norm": 9.46875, "learning_rate": 9.435297043774872e-06, "loss": 1.2115, "step": 16393 }, { "epoch": 1.7444137050436264, "grad_norm": 14.3203125, "learning_rate": 9.435261512222854e-06, "loss": 1.1078, "step": 16394 }, { "epoch": 1.7445201106618429, "grad_norm": 10.875, "learning_rate": 9.435225980670837e-06, "loss": 1.0253, "step": 16395 }, { "epoch": 1.7446265162800596, "grad_norm": 12.65625, "learning_rate": 9.435190449118819e-06, "loss": 1.3516, "step": 16396 }, { "epoch": 1.7447329218982763, "grad_norm": 15.1015625, "learning_rate": 9.4351549175668e-06, "loss": 1.3843, "step": 16397 }, { "epoch": 1.7448393275164928, "grad_norm": 11.6171875, "learning_rate": 9.435119386014782e-06, "loss": 0.8371, "step": 16398 }, { "epoch": 1.7449457331347094, "grad_norm": 11.9375, "learning_rate": 9.435083854462764e-06, "loss": 1.1527, "step": 16399 }, { "epoch": 1.7450521387529263, "grad_norm": 10.6875, "learning_rate": 9.435048322910745e-06, "loss": 1.0928, "step": 16400 }, { "epoch": 1.7451585443711428, "grad_norm": 19.890625, "learning_rate": 9.435012791358727e-06, "loss": 1.2089, "step": 16401 }, { "epoch": 1.7452649499893593, "grad_norm": 14.5703125, "learning_rate": 9.434977259806708e-06, "loss": 1.3998, "step": 16402 }, { "epoch": 1.745371355607576, "grad_norm": 18.1875, "learning_rate": 9.43494172825469e-06, "loss": 1.4204, "step": 16403 }, { "epoch": 1.7454777612257928, "grad_norm": 16.75, "learning_rate": 9.434906196702672e-06, "loss": 1.0299, "step": 16404 }, { "epoch": 1.7455841668440093, "grad_norm": 14.5, "learning_rate": 9.434870665150655e-06, "loss": 1.1093, "step": 16405 }, { "epoch": 1.745690572462226, "grad_norm": 13.578125, "learning_rate": 9.434835133598636e-06, "loss": 1.0936, "step": 16406 }, { "epoch": 1.7457969780804428, "grad_norm": 11.1796875, "learning_rate": 9.434799602046618e-06, "loss": 1.2539, "step": 16407 }, { "epoch": 1.7459033836986593, "grad_norm": 13.6953125, "learning_rate": 9.4347640704946e-06, "loss": 1.0099, "step": 16408 }, { "epoch": 1.7460097893168758, "grad_norm": 10.015625, "learning_rate": 9.434728538942581e-06, "loss": 1.0105, "step": 16409 }, { "epoch": 1.7461161949350927, "grad_norm": 14.671875, "learning_rate": 9.434693007390565e-06, "loss": 1.0887, "step": 16410 }, { "epoch": 1.7462226005533092, "grad_norm": 15.5078125, "learning_rate": 9.434657475838546e-06, "loss": 1.154, "step": 16411 }, { "epoch": 1.7463290061715258, "grad_norm": 9.2578125, "learning_rate": 9.434621944286528e-06, "loss": 0.9231, "step": 16412 }, { "epoch": 1.7464354117897425, "grad_norm": 13.1953125, "learning_rate": 9.43458641273451e-06, "loss": 0.9171, "step": 16413 }, { "epoch": 1.7465418174079592, "grad_norm": 11.53125, "learning_rate": 9.434550881182491e-06, "loss": 1.0438, "step": 16414 }, { "epoch": 1.7466482230261757, "grad_norm": 13.546875, "learning_rate": 9.434515349630473e-06, "loss": 0.9812, "step": 16415 }, { "epoch": 1.7467546286443925, "grad_norm": 10.625, "learning_rate": 9.434479818078454e-06, "loss": 1.1349, "step": 16416 }, { "epoch": 1.7468610342626092, "grad_norm": 17.328125, "learning_rate": 9.434444286526436e-06, "loss": 0.8672, "step": 16417 }, { "epoch": 1.7469674398808257, "grad_norm": 10.5703125, "learning_rate": 9.434408754974417e-06, "loss": 1.1433, "step": 16418 }, { "epoch": 1.7470738454990422, "grad_norm": 11.546875, "learning_rate": 9.434373223422399e-06, "loss": 0.878, "step": 16419 }, { "epoch": 1.747180251117259, "grad_norm": 13.21875, "learning_rate": 9.434337691870382e-06, "loss": 1.1109, "step": 16420 }, { "epoch": 1.7472866567354757, "grad_norm": 12.921875, "learning_rate": 9.434302160318364e-06, "loss": 1.1273, "step": 16421 }, { "epoch": 1.7473930623536922, "grad_norm": 13.8515625, "learning_rate": 9.434266628766346e-06, "loss": 1.2557, "step": 16422 }, { "epoch": 1.747499467971909, "grad_norm": 19.375, "learning_rate": 9.434231097214327e-06, "loss": 1.1461, "step": 16423 }, { "epoch": 1.7476058735901256, "grad_norm": 14.4765625, "learning_rate": 9.434195565662309e-06, "loss": 1.1994, "step": 16424 }, { "epoch": 1.7477122792083422, "grad_norm": 11.03125, "learning_rate": 9.434160034110292e-06, "loss": 1.2733, "step": 16425 }, { "epoch": 1.7478186848265589, "grad_norm": 13.9765625, "learning_rate": 9.434124502558274e-06, "loss": 1.0562, "step": 16426 }, { "epoch": 1.7479250904447756, "grad_norm": 17.796875, "learning_rate": 9.434088971006255e-06, "loss": 1.1512, "step": 16427 }, { "epoch": 1.7480314960629921, "grad_norm": 12.1484375, "learning_rate": 9.434053439454235e-06, "loss": 1.0265, "step": 16428 }, { "epoch": 1.7481379016812086, "grad_norm": 18.171875, "learning_rate": 9.434017907902217e-06, "loss": 1.1176, "step": 16429 }, { "epoch": 1.7482443072994254, "grad_norm": 12.1171875, "learning_rate": 9.433982376350198e-06, "loss": 1.2098, "step": 16430 }, { "epoch": 1.748350712917642, "grad_norm": 12.15625, "learning_rate": 9.433946844798182e-06, "loss": 1.0743, "step": 16431 }, { "epoch": 1.7484571185358586, "grad_norm": 14.3515625, "learning_rate": 9.433911313246163e-06, "loss": 0.9706, "step": 16432 }, { "epoch": 1.7485635241540753, "grad_norm": 14.140625, "learning_rate": 9.433875781694145e-06, "loss": 1.0357, "step": 16433 }, { "epoch": 1.748669929772292, "grad_norm": 11.0625, "learning_rate": 9.433840250142127e-06, "loss": 1.0234, "step": 16434 }, { "epoch": 1.7487763353905086, "grad_norm": 13.1484375, "learning_rate": 9.433804718590108e-06, "loss": 1.4302, "step": 16435 }, { "epoch": 1.748882741008725, "grad_norm": 15.8125, "learning_rate": 9.433769187038091e-06, "loss": 1.1138, "step": 16436 }, { "epoch": 1.748989146626942, "grad_norm": 17.40625, "learning_rate": 9.433733655486073e-06, "loss": 1.3931, "step": 16437 }, { "epoch": 1.7490955522451586, "grad_norm": 15.8046875, "learning_rate": 9.433698123934055e-06, "loss": 1.1729, "step": 16438 }, { "epoch": 1.749201957863375, "grad_norm": 9.28125, "learning_rate": 9.433662592382036e-06, "loss": 1.1037, "step": 16439 }, { "epoch": 1.7493083634815918, "grad_norm": 12.1953125, "learning_rate": 9.433627060830018e-06, "loss": 1.031, "step": 16440 }, { "epoch": 1.7494147690998085, "grad_norm": 11.046875, "learning_rate": 9.433591529278e-06, "loss": 1.0791, "step": 16441 }, { "epoch": 1.749521174718025, "grad_norm": 13.203125, "learning_rate": 9.433555997725981e-06, "loss": 1.1428, "step": 16442 }, { "epoch": 1.7496275803362418, "grad_norm": 10.4609375, "learning_rate": 9.433520466173963e-06, "loss": 1.1528, "step": 16443 }, { "epoch": 1.7497339859544585, "grad_norm": 9.515625, "learning_rate": 9.433484934621944e-06, "loss": 1.0395, "step": 16444 }, { "epoch": 1.749840391572675, "grad_norm": 11.140625, "learning_rate": 9.433449403069926e-06, "loss": 0.9082, "step": 16445 }, { "epoch": 1.7499467971908915, "grad_norm": 11.1640625, "learning_rate": 9.43341387151791e-06, "loss": 0.7353, "step": 16446 }, { "epoch": 1.7500532028091085, "grad_norm": 11.671875, "learning_rate": 9.43337833996589e-06, "loss": 1.1959, "step": 16447 }, { "epoch": 1.750159608427325, "grad_norm": 11.96875, "learning_rate": 9.433342808413872e-06, "loss": 1.2633, "step": 16448 }, { "epoch": 1.7502660140455415, "grad_norm": 11.6640625, "learning_rate": 9.433307276861854e-06, "loss": 1.0285, "step": 16449 }, { "epoch": 1.7503724196637582, "grad_norm": 12.015625, "learning_rate": 9.433271745309836e-06, "loss": 1.0443, "step": 16450 }, { "epoch": 1.750478825281975, "grad_norm": 13.015625, "learning_rate": 9.433236213757819e-06, "loss": 1.3557, "step": 16451 }, { "epoch": 1.7505852309001915, "grad_norm": 13.0390625, "learning_rate": 9.4332006822058e-06, "loss": 1.0356, "step": 16452 }, { "epoch": 1.7506916365184082, "grad_norm": 16.34375, "learning_rate": 9.433165150653782e-06, "loss": 1.2234, "step": 16453 }, { "epoch": 1.750798042136625, "grad_norm": 12.40625, "learning_rate": 9.433129619101764e-06, "loss": 0.9275, "step": 16454 }, { "epoch": 1.7509044477548414, "grad_norm": 13.3828125, "learning_rate": 9.433094087549745e-06, "loss": 0.8524, "step": 16455 }, { "epoch": 1.751010853373058, "grad_norm": 13.734375, "learning_rate": 9.433058555997727e-06, "loss": 1.0582, "step": 16456 }, { "epoch": 1.751117258991275, "grad_norm": 12.1171875, "learning_rate": 9.433023024445709e-06, "loss": 1.1921, "step": 16457 }, { "epoch": 1.7512236646094914, "grad_norm": 13.1328125, "learning_rate": 9.43298749289369e-06, "loss": 1.1123, "step": 16458 }, { "epoch": 1.751330070227708, "grad_norm": 19.0625, "learning_rate": 9.432951961341672e-06, "loss": 1.258, "step": 16459 }, { "epoch": 1.7514364758459247, "grad_norm": 12.4375, "learning_rate": 9.432916429789653e-06, "loss": 0.9331, "step": 16460 }, { "epoch": 1.7515428814641414, "grad_norm": 12.71875, "learning_rate": 9.432880898237637e-06, "loss": 1.1063, "step": 16461 }, { "epoch": 1.751649287082358, "grad_norm": 13.0078125, "learning_rate": 9.432845366685618e-06, "loss": 0.9196, "step": 16462 }, { "epoch": 1.7517556927005746, "grad_norm": 12.1796875, "learning_rate": 9.4328098351336e-06, "loss": 1.1848, "step": 16463 }, { "epoch": 1.7518620983187914, "grad_norm": 16.25, "learning_rate": 9.432774303581581e-06, "loss": 0.935, "step": 16464 }, { "epoch": 1.7519685039370079, "grad_norm": 11.265625, "learning_rate": 9.432738772029563e-06, "loss": 1.0929, "step": 16465 }, { "epoch": 1.7520749095552244, "grad_norm": 13.1640625, "learning_rate": 9.432703240477545e-06, "loss": 1.1762, "step": 16466 }, { "epoch": 1.7521813151734411, "grad_norm": 12.9765625, "learning_rate": 9.432667708925526e-06, "loss": 1.0597, "step": 16467 }, { "epoch": 1.7522877207916578, "grad_norm": 11.140625, "learning_rate": 9.432632177373508e-06, "loss": 1.0834, "step": 16468 }, { "epoch": 1.7523941264098744, "grad_norm": 11.8203125, "learning_rate": 9.43259664582149e-06, "loss": 1.2458, "step": 16469 }, { "epoch": 1.752500532028091, "grad_norm": 13.6015625, "learning_rate": 9.432561114269471e-06, "loss": 1.0723, "step": 16470 }, { "epoch": 1.7526069376463078, "grad_norm": 13.0078125, "learning_rate": 9.432525582717453e-06, "loss": 1.3312, "step": 16471 }, { "epoch": 1.7527133432645243, "grad_norm": 19.015625, "learning_rate": 9.432490051165436e-06, "loss": 1.3276, "step": 16472 }, { "epoch": 1.752819748882741, "grad_norm": 13.15625, "learning_rate": 9.432454519613418e-06, "loss": 1.781, "step": 16473 }, { "epoch": 1.7529261545009578, "grad_norm": 12.734375, "learning_rate": 9.4324189880614e-06, "loss": 1.0167, "step": 16474 }, { "epoch": 1.7530325601191743, "grad_norm": 12.9375, "learning_rate": 9.432383456509381e-06, "loss": 0.981, "step": 16475 }, { "epoch": 1.7531389657373908, "grad_norm": 10.03125, "learning_rate": 9.432347924957362e-06, "loss": 0.8413, "step": 16476 }, { "epoch": 1.7532453713556075, "grad_norm": 7.91796875, "learning_rate": 9.432312393405346e-06, "loss": 0.8417, "step": 16477 }, { "epoch": 1.7533517769738243, "grad_norm": 30.5625, "learning_rate": 9.432276861853327e-06, "loss": 1.5631, "step": 16478 }, { "epoch": 1.7534581825920408, "grad_norm": 11.5703125, "learning_rate": 9.432241330301309e-06, "loss": 1.1196, "step": 16479 }, { "epoch": 1.7535645882102575, "grad_norm": 12.4140625, "learning_rate": 9.43220579874929e-06, "loss": 1.0897, "step": 16480 }, { "epoch": 1.7536709938284742, "grad_norm": 12.9765625, "learning_rate": 9.432170267197272e-06, "loss": 1.0452, "step": 16481 }, { "epoch": 1.7537773994466908, "grad_norm": 12.40625, "learning_rate": 9.432134735645254e-06, "loss": 0.9938, "step": 16482 }, { "epoch": 1.7538838050649073, "grad_norm": 14.9296875, "learning_rate": 9.432099204093235e-06, "loss": 1.0302, "step": 16483 }, { "epoch": 1.7539902106831242, "grad_norm": 14.625, "learning_rate": 9.432063672541217e-06, "loss": 0.9504, "step": 16484 }, { "epoch": 1.7540966163013407, "grad_norm": 9.8203125, "learning_rate": 9.432028140989199e-06, "loss": 1.0978, "step": 16485 }, { "epoch": 1.7542030219195572, "grad_norm": 14.2578125, "learning_rate": 9.43199260943718e-06, "loss": 1.3252, "step": 16486 }, { "epoch": 1.754309427537774, "grad_norm": 11.125, "learning_rate": 9.431957077885164e-06, "loss": 1.2844, "step": 16487 }, { "epoch": 1.7544158331559907, "grad_norm": 15.7578125, "learning_rate": 9.431921546333145e-06, "loss": 0.9425, "step": 16488 }, { "epoch": 1.7545222387742072, "grad_norm": 14.25, "learning_rate": 9.431886014781127e-06, "loss": 1.005, "step": 16489 }, { "epoch": 1.754628644392424, "grad_norm": 14.546875, "learning_rate": 9.431850483229108e-06, "loss": 1.1592, "step": 16490 }, { "epoch": 1.7547350500106407, "grad_norm": 11.1484375, "learning_rate": 9.43181495167709e-06, "loss": 1.0382, "step": 16491 }, { "epoch": 1.7548414556288572, "grad_norm": 14.203125, "learning_rate": 9.431779420125072e-06, "loss": 1.193, "step": 16492 }, { "epoch": 1.7549478612470737, "grad_norm": 11.9453125, "learning_rate": 9.431743888573053e-06, "loss": 0.8623, "step": 16493 }, { "epoch": 1.7550542668652906, "grad_norm": 9.78125, "learning_rate": 9.431708357021035e-06, "loss": 0.9, "step": 16494 }, { "epoch": 1.7551606724835072, "grad_norm": 10.125, "learning_rate": 9.431672825469016e-06, "loss": 1.0419, "step": 16495 }, { "epoch": 1.7552670781017237, "grad_norm": 10.46875, "learning_rate": 9.431637293916998e-06, "loss": 1.1861, "step": 16496 }, { "epoch": 1.7553734837199404, "grad_norm": 11.1015625, "learning_rate": 9.431601762364981e-06, "loss": 1.1609, "step": 16497 }, { "epoch": 1.7554798893381571, "grad_norm": 12.140625, "learning_rate": 9.431566230812963e-06, "loss": 0.947, "step": 16498 }, { "epoch": 1.7555862949563736, "grad_norm": 11.1171875, "learning_rate": 9.431530699260944e-06, "loss": 1.0898, "step": 16499 }, { "epoch": 1.7556927005745904, "grad_norm": 10.7578125, "learning_rate": 9.431495167708926e-06, "loss": 0.9102, "step": 16500 }, { "epoch": 1.755799106192807, "grad_norm": 10.515625, "learning_rate": 9.431459636156908e-06, "loss": 0.9417, "step": 16501 }, { "epoch": 1.7559055118110236, "grad_norm": 19.34375, "learning_rate": 9.431424104604891e-06, "loss": 1.0266, "step": 16502 }, { "epoch": 1.7560119174292401, "grad_norm": 11.6875, "learning_rate": 9.431388573052873e-06, "loss": 1.052, "step": 16503 }, { "epoch": 1.7561183230474569, "grad_norm": 11.0859375, "learning_rate": 9.431353041500854e-06, "loss": 1.2726, "step": 16504 }, { "epoch": 1.7562247286656736, "grad_norm": 13.0078125, "learning_rate": 9.431317509948836e-06, "loss": 1.0495, "step": 16505 }, { "epoch": 1.75633113428389, "grad_norm": 12.203125, "learning_rate": 9.431281978396817e-06, "loss": 1.1577, "step": 16506 }, { "epoch": 1.7564375399021068, "grad_norm": 12.515625, "learning_rate": 9.431246446844799e-06, "loss": 0.9048, "step": 16507 }, { "epoch": 1.7565439455203236, "grad_norm": 11.9765625, "learning_rate": 9.43121091529278e-06, "loss": 1.0153, "step": 16508 }, { "epoch": 1.75665035113854, "grad_norm": 11.2265625, "learning_rate": 9.431175383740762e-06, "loss": 1.1451, "step": 16509 }, { "epoch": 1.7567567567567568, "grad_norm": 13.2890625, "learning_rate": 9.431139852188744e-06, "loss": 1.0796, "step": 16510 }, { "epoch": 1.7568631623749735, "grad_norm": 12.1953125, "learning_rate": 9.431104320636725e-06, "loss": 0.7643, "step": 16511 }, { "epoch": 1.75696956799319, "grad_norm": 19.234375, "learning_rate": 9.431068789084707e-06, "loss": 0.8905, "step": 16512 }, { "epoch": 1.7570759736114065, "grad_norm": 15.8984375, "learning_rate": 9.43103325753269e-06, "loss": 1.1467, "step": 16513 }, { "epoch": 1.7571823792296233, "grad_norm": 22.109375, "learning_rate": 9.430997725980672e-06, "loss": 1.2405, "step": 16514 }, { "epoch": 1.75728878484784, "grad_norm": 16.921875, "learning_rate": 9.430962194428654e-06, "loss": 1.3828, "step": 16515 }, { "epoch": 1.7573951904660565, "grad_norm": 13.6171875, "learning_rate": 9.430926662876635e-06, "loss": 1.0875, "step": 16516 }, { "epoch": 1.7575015960842733, "grad_norm": 10.46875, "learning_rate": 9.430891131324617e-06, "loss": 1.0947, "step": 16517 }, { "epoch": 1.75760800170249, "grad_norm": 13.953125, "learning_rate": 9.430855599772598e-06, "loss": 0.9573, "step": 16518 }, { "epoch": 1.7577144073207065, "grad_norm": 12.359375, "learning_rate": 9.43082006822058e-06, "loss": 0.9678, "step": 16519 }, { "epoch": 1.7578208129389232, "grad_norm": 11.9921875, "learning_rate": 9.430784536668562e-06, "loss": 1.1078, "step": 16520 }, { "epoch": 1.75792721855714, "grad_norm": 14.6875, "learning_rate": 9.430749005116543e-06, "loss": 1.382, "step": 16521 }, { "epoch": 1.7580336241753565, "grad_norm": 9.640625, "learning_rate": 9.430713473564525e-06, "loss": 0.964, "step": 16522 }, { "epoch": 1.758140029793573, "grad_norm": 10.09375, "learning_rate": 9.430677942012508e-06, "loss": 1.0386, "step": 16523 }, { "epoch": 1.7582464354117897, "grad_norm": 19.0625, "learning_rate": 9.43064241046049e-06, "loss": 1.3966, "step": 16524 }, { "epoch": 1.7583528410300064, "grad_norm": 8.8359375, "learning_rate": 9.430606878908471e-06, "loss": 0.9448, "step": 16525 }, { "epoch": 1.758459246648223, "grad_norm": 16.59375, "learning_rate": 9.430571347356453e-06, "loss": 1.135, "step": 16526 }, { "epoch": 1.7585656522664397, "grad_norm": 14.5703125, "learning_rate": 9.430535815804435e-06, "loss": 0.9164, "step": 16527 }, { "epoch": 1.7586720578846564, "grad_norm": 11.4140625, "learning_rate": 9.430500284252418e-06, "loss": 1.0813, "step": 16528 }, { "epoch": 1.758778463502873, "grad_norm": 11.8515625, "learning_rate": 9.4304647527004e-06, "loss": 1.0862, "step": 16529 }, { "epoch": 1.7588848691210894, "grad_norm": 10.5703125, "learning_rate": 9.430429221148381e-06, "loss": 1.0443, "step": 16530 }, { "epoch": 1.7589912747393064, "grad_norm": 10.1484375, "learning_rate": 9.430393689596363e-06, "loss": 1.0562, "step": 16531 }, { "epoch": 1.759097680357523, "grad_norm": 16.375, "learning_rate": 9.430358158044344e-06, "loss": 1.0831, "step": 16532 }, { "epoch": 1.7592040859757394, "grad_norm": 10.03125, "learning_rate": 9.430322626492326e-06, "loss": 1.0435, "step": 16533 }, { "epoch": 1.7593104915939561, "grad_norm": 13.7109375, "learning_rate": 9.430287094940308e-06, "loss": 0.9442, "step": 16534 }, { "epoch": 1.7594168972121729, "grad_norm": 8.890625, "learning_rate": 9.430251563388289e-06, "loss": 0.8775, "step": 16535 }, { "epoch": 1.7595233028303894, "grad_norm": 10.234375, "learning_rate": 9.43021603183627e-06, "loss": 0.9548, "step": 16536 }, { "epoch": 1.7596297084486061, "grad_norm": 11.0546875, "learning_rate": 9.430180500284252e-06, "loss": 1.198, "step": 16537 }, { "epoch": 1.7597361140668228, "grad_norm": 10.8125, "learning_rate": 9.430144968732236e-06, "loss": 0.9916, "step": 16538 }, { "epoch": 1.7598425196850394, "grad_norm": 10.8671875, "learning_rate": 9.430109437180217e-06, "loss": 1.0514, "step": 16539 }, { "epoch": 1.7599489253032559, "grad_norm": 11.7734375, "learning_rate": 9.430073905628199e-06, "loss": 1.2252, "step": 16540 }, { "epoch": 1.7600553309214728, "grad_norm": 12.1640625, "learning_rate": 9.43003837407618e-06, "loss": 0.9152, "step": 16541 }, { "epoch": 1.7601617365396893, "grad_norm": 16.328125, "learning_rate": 9.430002842524162e-06, "loss": 1.248, "step": 16542 }, { "epoch": 1.7602681421579058, "grad_norm": 14.59375, "learning_rate": 9.429967310972145e-06, "loss": 1.2249, "step": 16543 }, { "epoch": 1.7603745477761226, "grad_norm": 9.578125, "learning_rate": 9.429931779420127e-06, "loss": 0.8569, "step": 16544 }, { "epoch": 1.7604809533943393, "grad_norm": 13.5, "learning_rate": 9.429896247868109e-06, "loss": 1.1347, "step": 16545 }, { "epoch": 1.7605873590125558, "grad_norm": 15.4921875, "learning_rate": 9.42986071631609e-06, "loss": 1.5145, "step": 16546 }, { "epoch": 1.7606937646307725, "grad_norm": 13.234375, "learning_rate": 9.429825184764072e-06, "loss": 1.1036, "step": 16547 }, { "epoch": 1.7608001702489893, "grad_norm": 10.921875, "learning_rate": 9.429789653212052e-06, "loss": 0.8884, "step": 16548 }, { "epoch": 1.7609065758672058, "grad_norm": 12.1953125, "learning_rate": 9.429754121660035e-06, "loss": 1.1161, "step": 16549 }, { "epoch": 1.7610129814854223, "grad_norm": 12.671875, "learning_rate": 9.429718590108017e-06, "loss": 1.0263, "step": 16550 }, { "epoch": 1.761119387103639, "grad_norm": 12.578125, "learning_rate": 9.429683058555998e-06, "loss": 1.3185, "step": 16551 }, { "epoch": 1.7612257927218558, "grad_norm": 12.2265625, "learning_rate": 9.42964752700398e-06, "loss": 1.072, "step": 16552 }, { "epoch": 1.7613321983400723, "grad_norm": 12.6015625, "learning_rate": 9.429611995451961e-06, "loss": 1.152, "step": 16553 }, { "epoch": 1.761438603958289, "grad_norm": 12.859375, "learning_rate": 9.429576463899945e-06, "loss": 1.0349, "step": 16554 }, { "epoch": 1.7615450095765057, "grad_norm": 15.28125, "learning_rate": 9.429540932347926e-06, "loss": 0.9308, "step": 16555 }, { "epoch": 1.7616514151947222, "grad_norm": 13.7421875, "learning_rate": 9.429505400795908e-06, "loss": 1.0697, "step": 16556 }, { "epoch": 1.761757820812939, "grad_norm": 13.6015625, "learning_rate": 9.42946986924389e-06, "loss": 0.8995, "step": 16557 }, { "epoch": 1.7618642264311557, "grad_norm": 17.4375, "learning_rate": 9.429434337691871e-06, "loss": 1.0207, "step": 16558 }, { "epoch": 1.7619706320493722, "grad_norm": 11.046875, "learning_rate": 9.429398806139853e-06, "loss": 0.9315, "step": 16559 }, { "epoch": 1.7620770376675887, "grad_norm": 12.3984375, "learning_rate": 9.429363274587834e-06, "loss": 1.0727, "step": 16560 }, { "epoch": 1.7621834432858055, "grad_norm": 11.578125, "learning_rate": 9.429327743035816e-06, "loss": 0.9771, "step": 16561 }, { "epoch": 1.7622898489040222, "grad_norm": 11.4609375, "learning_rate": 9.429292211483798e-06, "loss": 1.0807, "step": 16562 }, { "epoch": 1.7623962545222387, "grad_norm": 12.8984375, "learning_rate": 9.42925667993178e-06, "loss": 1.1381, "step": 16563 }, { "epoch": 1.7625026601404554, "grad_norm": 11.5390625, "learning_rate": 9.429221148379762e-06, "loss": 1.1454, "step": 16564 }, { "epoch": 1.7626090657586722, "grad_norm": 10.3359375, "learning_rate": 9.429185616827744e-06, "loss": 0.8619, "step": 16565 }, { "epoch": 1.7627154713768887, "grad_norm": 14.78125, "learning_rate": 9.429150085275726e-06, "loss": 0.8935, "step": 16566 }, { "epoch": 1.7628218769951054, "grad_norm": 11.78125, "learning_rate": 9.429114553723707e-06, "loss": 1.3063, "step": 16567 }, { "epoch": 1.7629282826133221, "grad_norm": 12.84375, "learning_rate": 9.429079022171689e-06, "loss": 1.2846, "step": 16568 }, { "epoch": 1.7630346882315386, "grad_norm": 10.3203125, "learning_rate": 9.429043490619672e-06, "loss": 1.1502, "step": 16569 }, { "epoch": 1.7631410938497551, "grad_norm": 15.96875, "learning_rate": 9.429007959067654e-06, "loss": 1.0614, "step": 16570 }, { "epoch": 1.7632474994679719, "grad_norm": 11.8046875, "learning_rate": 9.428972427515635e-06, "loss": 0.9903, "step": 16571 }, { "epoch": 1.7633539050861886, "grad_norm": 15.46875, "learning_rate": 9.428936895963617e-06, "loss": 0.9035, "step": 16572 }, { "epoch": 1.7634603107044051, "grad_norm": 10.3203125, "learning_rate": 9.428901364411599e-06, "loss": 0.9779, "step": 16573 }, { "epoch": 1.7635667163226219, "grad_norm": 13.8671875, "learning_rate": 9.42886583285958e-06, "loss": 1.0764, "step": 16574 }, { "epoch": 1.7636731219408386, "grad_norm": 13.765625, "learning_rate": 9.428830301307562e-06, "loss": 1.1201, "step": 16575 }, { "epoch": 1.763779527559055, "grad_norm": 14.796875, "learning_rate": 9.428794769755543e-06, "loss": 1.0415, "step": 16576 }, { "epoch": 1.7638859331772716, "grad_norm": 11.6640625, "learning_rate": 9.428759238203525e-06, "loss": 1.075, "step": 16577 }, { "epoch": 1.7639923387954886, "grad_norm": 12.2109375, "learning_rate": 9.428723706651507e-06, "loss": 1.2317, "step": 16578 }, { "epoch": 1.764098744413705, "grad_norm": 14.78125, "learning_rate": 9.42868817509949e-06, "loss": 0.9469, "step": 16579 }, { "epoch": 1.7642051500319216, "grad_norm": 11.90625, "learning_rate": 9.428652643547472e-06, "loss": 0.9055, "step": 16580 }, { "epoch": 1.7643115556501383, "grad_norm": 12.78125, "learning_rate": 9.428617111995453e-06, "loss": 1.2372, "step": 16581 }, { "epoch": 1.764417961268355, "grad_norm": 13.40625, "learning_rate": 9.428581580443435e-06, "loss": 0.9205, "step": 16582 }, { "epoch": 1.7645243668865715, "grad_norm": 10.328125, "learning_rate": 9.428546048891416e-06, "loss": 1.2328, "step": 16583 }, { "epoch": 1.7646307725047883, "grad_norm": 11.4921875, "learning_rate": 9.428510517339398e-06, "loss": 1.0593, "step": 16584 }, { "epoch": 1.764737178123005, "grad_norm": 13.0234375, "learning_rate": 9.42847498578738e-06, "loss": 1.1074, "step": 16585 }, { "epoch": 1.7648435837412215, "grad_norm": 12.8671875, "learning_rate": 9.428439454235361e-06, "loss": 1.1934, "step": 16586 }, { "epoch": 1.764949989359438, "grad_norm": 19.359375, "learning_rate": 9.428403922683343e-06, "loss": 0.9808, "step": 16587 }, { "epoch": 1.765056394977655, "grad_norm": 13.15625, "learning_rate": 9.428368391131324e-06, "loss": 1.0707, "step": 16588 }, { "epoch": 1.7651628005958715, "grad_norm": 10.328125, "learning_rate": 9.428332859579306e-06, "loss": 0.9938, "step": 16589 }, { "epoch": 1.765269206214088, "grad_norm": 13.984375, "learning_rate": 9.42829732802729e-06, "loss": 1.1523, "step": 16590 }, { "epoch": 1.7653756118323047, "grad_norm": 12.8203125, "learning_rate": 9.428261796475271e-06, "loss": 0.9807, "step": 16591 }, { "epoch": 1.7654820174505215, "grad_norm": 12.8046875, "learning_rate": 9.428226264923253e-06, "loss": 1.1105, "step": 16592 }, { "epoch": 1.765588423068738, "grad_norm": 10.546875, "learning_rate": 9.428190733371234e-06, "loss": 0.9872, "step": 16593 }, { "epoch": 1.7656948286869547, "grad_norm": 14.84375, "learning_rate": 9.428155201819216e-06, "loss": 1.0687, "step": 16594 }, { "epoch": 1.7658012343051714, "grad_norm": 14.4375, "learning_rate": 9.428119670267199e-06, "loss": 0.7411, "step": 16595 }, { "epoch": 1.765907639923388, "grad_norm": 14.09375, "learning_rate": 9.42808413871518e-06, "loss": 1.1197, "step": 16596 }, { "epoch": 1.7660140455416045, "grad_norm": 10.3046875, "learning_rate": 9.428048607163162e-06, "loss": 1.0267, "step": 16597 }, { "epoch": 1.7661204511598212, "grad_norm": 18.15625, "learning_rate": 9.428013075611144e-06, "loss": 1.0962, "step": 16598 }, { "epoch": 1.766226856778038, "grad_norm": 22.625, "learning_rate": 9.427977544059125e-06, "loss": 0.941, "step": 16599 }, { "epoch": 1.7663332623962544, "grad_norm": 10.2890625, "learning_rate": 9.427942012507107e-06, "loss": 0.9608, "step": 16600 }, { "epoch": 1.7664396680144712, "grad_norm": 15.8125, "learning_rate": 9.427906480955089e-06, "loss": 1.4965, "step": 16601 }, { "epoch": 1.766546073632688, "grad_norm": 14.1875, "learning_rate": 9.42787094940307e-06, "loss": 1.0122, "step": 16602 }, { "epoch": 1.7666524792509044, "grad_norm": 11.890625, "learning_rate": 9.427835417851052e-06, "loss": 1.2483, "step": 16603 }, { "epoch": 1.7667588848691211, "grad_norm": 15.0625, "learning_rate": 9.427799886299034e-06, "loss": 1.1749, "step": 16604 }, { "epoch": 1.7668652904873379, "grad_norm": 11.9921875, "learning_rate": 9.427764354747017e-06, "loss": 1.0339, "step": 16605 }, { "epoch": 1.7669716961055544, "grad_norm": 15.890625, "learning_rate": 9.427728823194998e-06, "loss": 1.1391, "step": 16606 }, { "epoch": 1.7670781017237709, "grad_norm": 8.5390625, "learning_rate": 9.42769329164298e-06, "loss": 0.8775, "step": 16607 }, { "epoch": 1.7671845073419876, "grad_norm": 30.0, "learning_rate": 9.427657760090962e-06, "loss": 1.1021, "step": 16608 }, { "epoch": 1.7672909129602044, "grad_norm": 11.6953125, "learning_rate": 9.427622228538943e-06, "loss": 1.1203, "step": 16609 }, { "epoch": 1.7673973185784209, "grad_norm": 12.2890625, "learning_rate": 9.427586696986925e-06, "loss": 1.0728, "step": 16610 }, { "epoch": 1.7675037241966376, "grad_norm": 12.828125, "learning_rate": 9.427551165434906e-06, "loss": 0.9711, "step": 16611 }, { "epoch": 1.7676101298148543, "grad_norm": 14.734375, "learning_rate": 9.427515633882888e-06, "loss": 1.1297, "step": 16612 }, { "epoch": 1.7677165354330708, "grad_norm": 14.359375, "learning_rate": 9.42748010233087e-06, "loss": 1.0356, "step": 16613 }, { "epoch": 1.7678229410512873, "grad_norm": 10.7734375, "learning_rate": 9.427444570778851e-06, "loss": 1.1297, "step": 16614 }, { "epoch": 1.7679293466695043, "grad_norm": 12.4296875, "learning_rate": 9.427409039226835e-06, "loss": 1.2184, "step": 16615 }, { "epoch": 1.7680357522877208, "grad_norm": 9.5234375, "learning_rate": 9.427373507674816e-06, "loss": 0.9226, "step": 16616 }, { "epoch": 1.7681421579059373, "grad_norm": 12.875, "learning_rate": 9.427337976122798e-06, "loss": 1.0811, "step": 16617 }, { "epoch": 1.768248563524154, "grad_norm": 18.90625, "learning_rate": 9.42730244457078e-06, "loss": 1.1181, "step": 16618 }, { "epoch": 1.7683549691423708, "grad_norm": 9.609375, "learning_rate": 9.427266913018761e-06, "loss": 1.1023, "step": 16619 }, { "epoch": 1.7684613747605873, "grad_norm": 12.4453125, "learning_rate": 9.427231381466744e-06, "loss": 1.2734, "step": 16620 }, { "epoch": 1.768567780378804, "grad_norm": 10.9296875, "learning_rate": 9.427195849914726e-06, "loss": 1.0358, "step": 16621 }, { "epoch": 1.7686741859970208, "grad_norm": 15.15625, "learning_rate": 9.427160318362708e-06, "loss": 1.1177, "step": 16622 }, { "epoch": 1.7687805916152373, "grad_norm": 10.2265625, "learning_rate": 9.427124786810689e-06, "loss": 0.9537, "step": 16623 }, { "epoch": 1.7688869972334538, "grad_norm": 13.6484375, "learning_rate": 9.42708925525867e-06, "loss": 1.4571, "step": 16624 }, { "epoch": 1.7689934028516707, "grad_norm": 16.125, "learning_rate": 9.427053723706652e-06, "loss": 1.2647, "step": 16625 }, { "epoch": 1.7690998084698872, "grad_norm": 11.859375, "learning_rate": 9.427018192154634e-06, "loss": 1.132, "step": 16626 }, { "epoch": 1.7692062140881037, "grad_norm": 13.1171875, "learning_rate": 9.426982660602616e-06, "loss": 0.9955, "step": 16627 }, { "epoch": 1.7693126197063205, "grad_norm": 12.8125, "learning_rate": 9.426947129050597e-06, "loss": 1.5203, "step": 16628 }, { "epoch": 1.7694190253245372, "grad_norm": 10.6875, "learning_rate": 9.426911597498579e-06, "loss": 0.9462, "step": 16629 }, { "epoch": 1.7695254309427537, "grad_norm": 17.15625, "learning_rate": 9.42687606594656e-06, "loss": 1.2095, "step": 16630 }, { "epoch": 1.7696318365609705, "grad_norm": 11.9375, "learning_rate": 9.426840534394544e-06, "loss": 1.2308, "step": 16631 }, { "epoch": 1.7697382421791872, "grad_norm": 12.515625, "learning_rate": 9.426805002842525e-06, "loss": 1.3014, "step": 16632 }, { "epoch": 1.7698446477974037, "grad_norm": 15.53125, "learning_rate": 9.426769471290507e-06, "loss": 1.1527, "step": 16633 }, { "epoch": 1.7699510534156202, "grad_norm": 13.8125, "learning_rate": 9.426733939738489e-06, "loss": 1.3297, "step": 16634 }, { "epoch": 1.770057459033837, "grad_norm": 9.1875, "learning_rate": 9.42669840818647e-06, "loss": 0.9279, "step": 16635 }, { "epoch": 1.7701638646520537, "grad_norm": 17.640625, "learning_rate": 9.426662876634453e-06, "loss": 0.8744, "step": 16636 }, { "epoch": 1.7702702702702702, "grad_norm": 12.8125, "learning_rate": 9.426627345082433e-06, "loss": 0.9115, "step": 16637 }, { "epoch": 1.770376675888487, "grad_norm": 13.25, "learning_rate": 9.426591813530415e-06, "loss": 1.1111, "step": 16638 }, { "epoch": 1.7704830815067036, "grad_norm": 20.34375, "learning_rate": 9.426556281978397e-06, "loss": 1.1166, "step": 16639 }, { "epoch": 1.7705894871249201, "grad_norm": 12.546875, "learning_rate": 9.426520750426378e-06, "loss": 1.1615, "step": 16640 }, { "epoch": 1.7706958927431369, "grad_norm": 15.28125, "learning_rate": 9.426485218874361e-06, "loss": 1.0665, "step": 16641 }, { "epoch": 1.7708022983613536, "grad_norm": 13.3671875, "learning_rate": 9.426449687322343e-06, "loss": 1.1637, "step": 16642 }, { "epoch": 1.7709087039795701, "grad_norm": 13.078125, "learning_rate": 9.426414155770325e-06, "loss": 0.9247, "step": 16643 }, { "epoch": 1.7710151095977866, "grad_norm": 14.2109375, "learning_rate": 9.426378624218306e-06, "loss": 0.9542, "step": 16644 }, { "epoch": 1.7711215152160034, "grad_norm": 15.4140625, "learning_rate": 9.426343092666288e-06, "loss": 1.0753, "step": 16645 }, { "epoch": 1.77122792083422, "grad_norm": 10.6328125, "learning_rate": 9.426307561114271e-06, "loss": 0.9594, "step": 16646 }, { "epoch": 1.7713343264524366, "grad_norm": 10.7109375, "learning_rate": 9.426272029562253e-06, "loss": 1.0455, "step": 16647 }, { "epoch": 1.7714407320706533, "grad_norm": 13.78125, "learning_rate": 9.426236498010234e-06, "loss": 1.297, "step": 16648 }, { "epoch": 1.77154713768887, "grad_norm": 20.0625, "learning_rate": 9.426200966458216e-06, "loss": 1.1431, "step": 16649 }, { "epoch": 1.7716535433070866, "grad_norm": 11.3125, "learning_rate": 9.426165434906198e-06, "loss": 1.1, "step": 16650 }, { "epoch": 1.7717599489253033, "grad_norm": 14.6953125, "learning_rate": 9.42612990335418e-06, "loss": 0.9984, "step": 16651 }, { "epoch": 1.77186635454352, "grad_norm": 18.625, "learning_rate": 9.42609437180216e-06, "loss": 0.9838, "step": 16652 }, { "epoch": 1.7719727601617365, "grad_norm": 13.140625, "learning_rate": 9.426058840250142e-06, "loss": 0.8531, "step": 16653 }, { "epoch": 1.772079165779953, "grad_norm": 11.71875, "learning_rate": 9.426023308698124e-06, "loss": 1.1802, "step": 16654 }, { "epoch": 1.7721855713981698, "grad_norm": 10.7265625, "learning_rate": 9.425987777146106e-06, "loss": 1.0502, "step": 16655 }, { "epoch": 1.7722919770163865, "grad_norm": 17.046875, "learning_rate": 9.425952245594089e-06, "loss": 0.9417, "step": 16656 }, { "epoch": 1.772398382634603, "grad_norm": 11.7265625, "learning_rate": 9.42591671404207e-06, "loss": 0.697, "step": 16657 }, { "epoch": 1.7725047882528198, "grad_norm": 17.65625, "learning_rate": 9.425881182490052e-06, "loss": 1.0415, "step": 16658 }, { "epoch": 1.7726111938710365, "grad_norm": 11.375, "learning_rate": 9.425845650938034e-06, "loss": 1.2737, "step": 16659 }, { "epoch": 1.772717599489253, "grad_norm": 9.140625, "learning_rate": 9.425810119386015e-06, "loss": 0.9869, "step": 16660 }, { "epoch": 1.7728240051074695, "grad_norm": 14.1484375, "learning_rate": 9.425774587833999e-06, "loss": 1.2994, "step": 16661 }, { "epoch": 1.7729304107256865, "grad_norm": 17.1875, "learning_rate": 9.42573905628198e-06, "loss": 0.76, "step": 16662 }, { "epoch": 1.773036816343903, "grad_norm": 21.328125, "learning_rate": 9.425703524729962e-06, "loss": 1.1605, "step": 16663 }, { "epoch": 1.7731432219621195, "grad_norm": 12.59375, "learning_rate": 9.425667993177943e-06, "loss": 1.3036, "step": 16664 }, { "epoch": 1.7732496275803362, "grad_norm": 21.828125, "learning_rate": 9.425632461625925e-06, "loss": 0.9935, "step": 16665 }, { "epoch": 1.773356033198553, "grad_norm": 13.4296875, "learning_rate": 9.425596930073907e-06, "loss": 0.9923, "step": 16666 }, { "epoch": 1.7734624388167695, "grad_norm": 13.8671875, "learning_rate": 9.425561398521888e-06, "loss": 0.8882, "step": 16667 }, { "epoch": 1.7735688444349862, "grad_norm": 11.8203125, "learning_rate": 9.42552586696987e-06, "loss": 0.9799, "step": 16668 }, { "epoch": 1.773675250053203, "grad_norm": 14.4609375, "learning_rate": 9.425490335417852e-06, "loss": 1.2742, "step": 16669 }, { "epoch": 1.7737816556714194, "grad_norm": 11.421875, "learning_rate": 9.425454803865833e-06, "loss": 1.0048, "step": 16670 }, { "epoch": 1.773888061289636, "grad_norm": 13.828125, "learning_rate": 9.425419272313815e-06, "loss": 0.9563, "step": 16671 }, { "epoch": 1.773994466907853, "grad_norm": 13.0703125, "learning_rate": 9.425383740761798e-06, "loss": 0.8129, "step": 16672 }, { "epoch": 1.7741008725260694, "grad_norm": 14.0234375, "learning_rate": 9.42534820920978e-06, "loss": 1.5516, "step": 16673 }, { "epoch": 1.774207278144286, "grad_norm": 16.015625, "learning_rate": 9.425312677657761e-06, "loss": 1.1321, "step": 16674 }, { "epoch": 1.7743136837625026, "grad_norm": 13.8515625, "learning_rate": 9.425277146105743e-06, "loss": 0.9808, "step": 16675 }, { "epoch": 1.7744200893807194, "grad_norm": 15.125, "learning_rate": 9.425241614553724e-06, "loss": 1.1261, "step": 16676 }, { "epoch": 1.7745264949989359, "grad_norm": 14.9453125, "learning_rate": 9.425206083001706e-06, "loss": 1.1626, "step": 16677 }, { "epoch": 1.7746329006171526, "grad_norm": 15.8515625, "learning_rate": 9.425170551449688e-06, "loss": 1.2346, "step": 16678 }, { "epoch": 1.7747393062353694, "grad_norm": 16.765625, "learning_rate": 9.42513501989767e-06, "loss": 1.0176, "step": 16679 }, { "epoch": 1.7748457118535859, "grad_norm": 14.1328125, "learning_rate": 9.425099488345651e-06, "loss": 1.2028, "step": 16680 }, { "epoch": 1.7749521174718024, "grad_norm": 10.296875, "learning_rate": 9.425063956793632e-06, "loss": 1.2386, "step": 16681 }, { "epoch": 1.775058523090019, "grad_norm": 13.328125, "learning_rate": 9.425028425241616e-06, "loss": 1.0712, "step": 16682 }, { "epoch": 1.7751649287082358, "grad_norm": 15.328125, "learning_rate": 9.424992893689597e-06, "loss": 0.9335, "step": 16683 }, { "epoch": 1.7752713343264523, "grad_norm": 11.9765625, "learning_rate": 9.424957362137579e-06, "loss": 1.0304, "step": 16684 }, { "epoch": 1.775377739944669, "grad_norm": 15.1484375, "learning_rate": 9.42492183058556e-06, "loss": 0.9656, "step": 16685 }, { "epoch": 1.7754841455628858, "grad_norm": 11.7578125, "learning_rate": 9.424886299033542e-06, "loss": 1.0811, "step": 16686 }, { "epoch": 1.7755905511811023, "grad_norm": 11.96875, "learning_rate": 9.424850767481526e-06, "loss": 0.9909, "step": 16687 }, { "epoch": 1.775696956799319, "grad_norm": 11.140625, "learning_rate": 9.424815235929507e-06, "loss": 1.0725, "step": 16688 }, { "epoch": 1.7758033624175358, "grad_norm": 13.3984375, "learning_rate": 9.424779704377489e-06, "loss": 1.1057, "step": 16689 }, { "epoch": 1.7759097680357523, "grad_norm": 9.3984375, "learning_rate": 9.42474417282547e-06, "loss": 1.1136, "step": 16690 }, { "epoch": 1.7760161736539688, "grad_norm": 11.5625, "learning_rate": 9.424708641273452e-06, "loss": 1.1133, "step": 16691 }, { "epoch": 1.7761225792721855, "grad_norm": 16.171875, "learning_rate": 9.424673109721434e-06, "loss": 1.0426, "step": 16692 }, { "epoch": 1.7762289848904023, "grad_norm": 13.6015625, "learning_rate": 9.424637578169415e-06, "loss": 1.212, "step": 16693 }, { "epoch": 1.7763353905086188, "grad_norm": 12.9921875, "learning_rate": 9.424602046617397e-06, "loss": 1.2698, "step": 16694 }, { "epoch": 1.7764417961268355, "grad_norm": 11.625, "learning_rate": 9.424566515065378e-06, "loss": 1.0714, "step": 16695 }, { "epoch": 1.7765482017450522, "grad_norm": 10.2421875, "learning_rate": 9.42453098351336e-06, "loss": 1.1117, "step": 16696 }, { "epoch": 1.7766546073632687, "grad_norm": 13.109375, "learning_rate": 9.424495451961343e-06, "loss": 0.9715, "step": 16697 }, { "epoch": 1.7767610129814855, "grad_norm": 18.625, "learning_rate": 9.424459920409325e-06, "loss": 1.4883, "step": 16698 }, { "epoch": 1.7768674185997022, "grad_norm": 11.0859375, "learning_rate": 9.424424388857306e-06, "loss": 1.0146, "step": 16699 }, { "epoch": 1.7769738242179187, "grad_norm": 13.5078125, "learning_rate": 9.424388857305288e-06, "loss": 1.1073, "step": 16700 }, { "epoch": 1.7770802298361352, "grad_norm": 12.25, "learning_rate": 9.42435332575327e-06, "loss": 1.0179, "step": 16701 }, { "epoch": 1.777186635454352, "grad_norm": 16.59375, "learning_rate": 9.424317794201251e-06, "loss": 1.2902, "step": 16702 }, { "epoch": 1.7772930410725687, "grad_norm": 13.0, "learning_rate": 9.424282262649233e-06, "loss": 1.4411, "step": 16703 }, { "epoch": 1.7773994466907852, "grad_norm": 15.6328125, "learning_rate": 9.424246731097215e-06, "loss": 1.1203, "step": 16704 }, { "epoch": 1.777505852309002, "grad_norm": 12.0546875, "learning_rate": 9.424211199545196e-06, "loss": 1.2364, "step": 16705 }, { "epoch": 1.7776122579272187, "grad_norm": 11.0546875, "learning_rate": 9.424175667993178e-06, "loss": 1.1279, "step": 16706 }, { "epoch": 1.7777186635454352, "grad_norm": 10.9765625, "learning_rate": 9.42414013644116e-06, "loss": 1.0703, "step": 16707 }, { "epoch": 1.7778250691636517, "grad_norm": 10.1015625, "learning_rate": 9.424104604889143e-06, "loss": 1.0063, "step": 16708 }, { "epoch": 1.7779314747818686, "grad_norm": 13.921875, "learning_rate": 9.424069073337124e-06, "loss": 0.9428, "step": 16709 }, { "epoch": 1.7780378804000851, "grad_norm": 13.375, "learning_rate": 9.424033541785106e-06, "loss": 0.9854, "step": 16710 }, { "epoch": 1.7781442860183017, "grad_norm": 13.9296875, "learning_rate": 9.423998010233087e-06, "loss": 1.653, "step": 16711 }, { "epoch": 1.7782506916365184, "grad_norm": 12.3359375, "learning_rate": 9.423962478681069e-06, "loss": 1.2772, "step": 16712 }, { "epoch": 1.7783570972547351, "grad_norm": 16.828125, "learning_rate": 9.423926947129052e-06, "loss": 1.1709, "step": 16713 }, { "epoch": 1.7784635028729516, "grad_norm": 11.578125, "learning_rate": 9.423891415577034e-06, "loss": 1.1814, "step": 16714 }, { "epoch": 1.7785699084911684, "grad_norm": 10.140625, "learning_rate": 9.423855884025016e-06, "loss": 0.9783, "step": 16715 }, { "epoch": 1.778676314109385, "grad_norm": 12.3125, "learning_rate": 9.423820352472997e-06, "loss": 1.1621, "step": 16716 }, { "epoch": 1.7787827197276016, "grad_norm": 11.296875, "learning_rate": 9.423784820920979e-06, "loss": 0.9242, "step": 16717 }, { "epoch": 1.7788891253458181, "grad_norm": 11.28125, "learning_rate": 9.42374928936896e-06, "loss": 0.8954, "step": 16718 }, { "epoch": 1.778995530964035, "grad_norm": 16.15625, "learning_rate": 9.423713757816942e-06, "loss": 1.4255, "step": 16719 }, { "epoch": 1.7791019365822516, "grad_norm": 11.0625, "learning_rate": 9.423678226264924e-06, "loss": 1.2998, "step": 16720 }, { "epoch": 1.779208342200468, "grad_norm": 15.40625, "learning_rate": 9.423642694712905e-06, "loss": 1.3147, "step": 16721 }, { "epoch": 1.7793147478186848, "grad_norm": 21.359375, "learning_rate": 9.423607163160887e-06, "loss": 0.9458, "step": 16722 }, { "epoch": 1.7794211534369015, "grad_norm": 13.578125, "learning_rate": 9.42357163160887e-06, "loss": 1.1781, "step": 16723 }, { "epoch": 1.779527559055118, "grad_norm": 15.453125, "learning_rate": 9.423536100056852e-06, "loss": 1.3149, "step": 16724 }, { "epoch": 1.7796339646733348, "grad_norm": 11.8828125, "learning_rate": 9.423500568504833e-06, "loss": 1.2792, "step": 16725 }, { "epoch": 1.7797403702915515, "grad_norm": 11.0390625, "learning_rate": 9.423465036952815e-06, "loss": 0.9451, "step": 16726 }, { "epoch": 1.779846775909768, "grad_norm": 11.2109375, "learning_rate": 9.423429505400797e-06, "loss": 1.2281, "step": 16727 }, { "epoch": 1.7799531815279845, "grad_norm": 10.5, "learning_rate": 9.423393973848778e-06, "loss": 1.1869, "step": 16728 }, { "epoch": 1.7800595871462013, "grad_norm": 9.3046875, "learning_rate": 9.42335844229676e-06, "loss": 1.079, "step": 16729 }, { "epoch": 1.780165992764418, "grad_norm": 12.140625, "learning_rate": 9.423322910744741e-06, "loss": 1.1547, "step": 16730 }, { "epoch": 1.7802723983826345, "grad_norm": 10.4140625, "learning_rate": 9.423287379192723e-06, "loss": 1.0178, "step": 16731 }, { "epoch": 1.7803788040008512, "grad_norm": 11.359375, "learning_rate": 9.423251847640705e-06, "loss": 1.028, "step": 16732 }, { "epoch": 1.780485209619068, "grad_norm": 16.515625, "learning_rate": 9.423216316088688e-06, "loss": 1.1715, "step": 16733 }, { "epoch": 1.7805916152372845, "grad_norm": 13.40625, "learning_rate": 9.42318078453667e-06, "loss": 1.4576, "step": 16734 }, { "epoch": 1.7806980208555012, "grad_norm": 13.4609375, "learning_rate": 9.423145252984651e-06, "loss": 1.0701, "step": 16735 }, { "epoch": 1.780804426473718, "grad_norm": 17.53125, "learning_rate": 9.423109721432633e-06, "loss": 1.0138, "step": 16736 }, { "epoch": 1.7809108320919345, "grad_norm": 15.6328125, "learning_rate": 9.423074189880614e-06, "loss": 0.9513, "step": 16737 }, { "epoch": 1.781017237710151, "grad_norm": 13.3125, "learning_rate": 9.423038658328598e-06, "loss": 1.268, "step": 16738 }, { "epoch": 1.7811236433283677, "grad_norm": 13.1328125, "learning_rate": 9.42300312677658e-06, "loss": 0.9387, "step": 16739 }, { "epoch": 1.7812300489465844, "grad_norm": 13.21875, "learning_rate": 9.42296759522456e-06, "loss": 1.1615, "step": 16740 }, { "epoch": 1.781336454564801, "grad_norm": 13.3359375, "learning_rate": 9.422932063672542e-06, "loss": 1.3769, "step": 16741 }, { "epoch": 1.7814428601830177, "grad_norm": 12.3984375, "learning_rate": 9.422896532120524e-06, "loss": 0.9892, "step": 16742 }, { "epoch": 1.7815492658012344, "grad_norm": 9.234375, "learning_rate": 9.422861000568506e-06, "loss": 0.9486, "step": 16743 }, { "epoch": 1.781655671419451, "grad_norm": 18.1875, "learning_rate": 9.422825469016487e-06, "loss": 0.994, "step": 16744 }, { "epoch": 1.7817620770376674, "grad_norm": 18.09375, "learning_rate": 9.422789937464469e-06, "loss": 1.015, "step": 16745 }, { "epoch": 1.7818684826558844, "grad_norm": 12.3046875, "learning_rate": 9.42275440591245e-06, "loss": 1.2004, "step": 16746 }, { "epoch": 1.7819748882741009, "grad_norm": 11.5703125, "learning_rate": 9.422718874360432e-06, "loss": 0.9675, "step": 16747 }, { "epoch": 1.7820812938923174, "grad_norm": 12.4765625, "learning_rate": 9.422683342808414e-06, "loss": 1.0915, "step": 16748 }, { "epoch": 1.7821876995105341, "grad_norm": 12.4921875, "learning_rate": 9.422647811256397e-06, "loss": 0.8417, "step": 16749 }, { "epoch": 1.7822941051287509, "grad_norm": 18.609375, "learning_rate": 9.422612279704379e-06, "loss": 0.7875, "step": 16750 }, { "epoch": 1.7824005107469674, "grad_norm": 10.703125, "learning_rate": 9.42257674815236e-06, "loss": 0.9195, "step": 16751 }, { "epoch": 1.782506916365184, "grad_norm": 12.6171875, "learning_rate": 9.422541216600342e-06, "loss": 0.9616, "step": 16752 }, { "epoch": 1.7826133219834008, "grad_norm": 10.921875, "learning_rate": 9.422505685048323e-06, "loss": 0.9509, "step": 16753 }, { "epoch": 1.7827197276016173, "grad_norm": 14.59375, "learning_rate": 9.422470153496307e-06, "loss": 1.381, "step": 16754 }, { "epoch": 1.7828261332198339, "grad_norm": 12.359375, "learning_rate": 9.422434621944288e-06, "loss": 0.8937, "step": 16755 }, { "epoch": 1.7829325388380508, "grad_norm": 8.546875, "learning_rate": 9.422399090392268e-06, "loss": 0.875, "step": 16756 }, { "epoch": 1.7830389444562673, "grad_norm": 9.3984375, "learning_rate": 9.42236355884025e-06, "loss": 1.0596, "step": 16757 }, { "epoch": 1.7831453500744838, "grad_norm": 12.28125, "learning_rate": 9.422328027288231e-06, "loss": 1.2534, "step": 16758 }, { "epoch": 1.7832517556927006, "grad_norm": 16.515625, "learning_rate": 9.422292495736215e-06, "loss": 1.1699, "step": 16759 }, { "epoch": 1.7833581613109173, "grad_norm": 19.34375, "learning_rate": 9.422256964184196e-06, "loss": 0.9498, "step": 16760 }, { "epoch": 1.7834645669291338, "grad_norm": 13.4921875, "learning_rate": 9.422221432632178e-06, "loss": 1.1654, "step": 16761 }, { "epoch": 1.7835709725473505, "grad_norm": 10.8515625, "learning_rate": 9.42218590108016e-06, "loss": 1.1592, "step": 16762 }, { "epoch": 1.7836773781655673, "grad_norm": 12.15625, "learning_rate": 9.422150369528141e-06, "loss": 1.0145, "step": 16763 }, { "epoch": 1.7837837837837838, "grad_norm": 11.1484375, "learning_rate": 9.422114837976124e-06, "loss": 1.0227, "step": 16764 }, { "epoch": 1.7838901894020003, "grad_norm": 11.9140625, "learning_rate": 9.422079306424106e-06, "loss": 1.3327, "step": 16765 }, { "epoch": 1.783996595020217, "grad_norm": 11.4296875, "learning_rate": 9.422043774872088e-06, "loss": 1.292, "step": 16766 }, { "epoch": 1.7841030006384337, "grad_norm": 11.359375, "learning_rate": 9.42200824332007e-06, "loss": 1.1493, "step": 16767 }, { "epoch": 1.7842094062566503, "grad_norm": 10.875, "learning_rate": 9.421972711768051e-06, "loss": 1.1237, "step": 16768 }, { "epoch": 1.784315811874867, "grad_norm": 17.046875, "learning_rate": 9.421937180216033e-06, "loss": 1.0819, "step": 16769 }, { "epoch": 1.7844222174930837, "grad_norm": 11.6953125, "learning_rate": 9.421901648664014e-06, "loss": 1.1722, "step": 16770 }, { "epoch": 1.7845286231113002, "grad_norm": 10.6640625, "learning_rate": 9.421866117111996e-06, "loss": 1.1146, "step": 16771 }, { "epoch": 1.784635028729517, "grad_norm": 10.7578125, "learning_rate": 9.421830585559977e-06, "loss": 1.1218, "step": 16772 }, { "epoch": 1.7847414343477337, "grad_norm": 9.7109375, "learning_rate": 9.421795054007959e-06, "loss": 1.0974, "step": 16773 }, { "epoch": 1.7848478399659502, "grad_norm": 10.2265625, "learning_rate": 9.421759522455942e-06, "loss": 1.072, "step": 16774 }, { "epoch": 1.7849542455841667, "grad_norm": 15.03125, "learning_rate": 9.421723990903924e-06, "loss": 1.104, "step": 16775 }, { "epoch": 1.7850606512023834, "grad_norm": 12.3359375, "learning_rate": 9.421688459351905e-06, "loss": 1.0336, "step": 16776 }, { "epoch": 1.7851670568206002, "grad_norm": 11.375, "learning_rate": 9.421652927799887e-06, "loss": 1.1565, "step": 16777 }, { "epoch": 1.7852734624388167, "grad_norm": 13.46875, "learning_rate": 9.421617396247869e-06, "loss": 1.3851, "step": 16778 }, { "epoch": 1.7853798680570334, "grad_norm": 15.21875, "learning_rate": 9.421581864695852e-06, "loss": 1.7389, "step": 16779 }, { "epoch": 1.7854862736752501, "grad_norm": 11.09375, "learning_rate": 9.421546333143834e-06, "loss": 1.2446, "step": 16780 }, { "epoch": 1.7855926792934667, "grad_norm": 11.1015625, "learning_rate": 9.421510801591815e-06, "loss": 1.3012, "step": 16781 }, { "epoch": 1.7856990849116834, "grad_norm": 11.8046875, "learning_rate": 9.421475270039797e-06, "loss": 1.0522, "step": 16782 }, { "epoch": 1.7858054905299001, "grad_norm": 15.390625, "learning_rate": 9.421439738487778e-06, "loss": 0.9176, "step": 16783 }, { "epoch": 1.7859118961481166, "grad_norm": 15.375, "learning_rate": 9.42140420693576e-06, "loss": 0.8461, "step": 16784 }, { "epoch": 1.7860183017663331, "grad_norm": 16.625, "learning_rate": 9.421368675383742e-06, "loss": 1.4153, "step": 16785 }, { "epoch": 1.7861247073845499, "grad_norm": 13.2890625, "learning_rate": 9.421333143831723e-06, "loss": 1.0099, "step": 16786 }, { "epoch": 1.7862311130027666, "grad_norm": 16.53125, "learning_rate": 9.421297612279705e-06, "loss": 1.1373, "step": 16787 }, { "epoch": 1.7863375186209831, "grad_norm": 13.0, "learning_rate": 9.421262080727686e-06, "loss": 0.9511, "step": 16788 }, { "epoch": 1.7864439242391998, "grad_norm": 14.875, "learning_rate": 9.421226549175668e-06, "loss": 1.0952, "step": 16789 }, { "epoch": 1.7865503298574166, "grad_norm": 9.1796875, "learning_rate": 9.421191017623651e-06, "loss": 1.0097, "step": 16790 }, { "epoch": 1.786656735475633, "grad_norm": 11.671875, "learning_rate": 9.421155486071633e-06, "loss": 1.267, "step": 16791 }, { "epoch": 1.7867631410938496, "grad_norm": 14.453125, "learning_rate": 9.421119954519615e-06, "loss": 1.1451, "step": 16792 }, { "epoch": 1.7868695467120665, "grad_norm": 10.7890625, "learning_rate": 9.421084422967596e-06, "loss": 1.1231, "step": 16793 }, { "epoch": 1.786975952330283, "grad_norm": 14.828125, "learning_rate": 9.421048891415578e-06, "loss": 0.8949, "step": 16794 }, { "epoch": 1.7870823579484996, "grad_norm": 11.5234375, "learning_rate": 9.42101335986356e-06, "loss": 1.1551, "step": 16795 }, { "epoch": 1.7871887635667163, "grad_norm": 11.515625, "learning_rate": 9.420977828311541e-06, "loss": 0.9613, "step": 16796 }, { "epoch": 1.787295169184933, "grad_norm": 11.4765625, "learning_rate": 9.420942296759523e-06, "loss": 1.0384, "step": 16797 }, { "epoch": 1.7874015748031495, "grad_norm": 10.921875, "learning_rate": 9.420906765207504e-06, "loss": 1.2336, "step": 16798 }, { "epoch": 1.7875079804213663, "grad_norm": 11.328125, "learning_rate": 9.420871233655486e-06, "loss": 0.7813, "step": 16799 }, { "epoch": 1.787614386039583, "grad_norm": 14.9453125, "learning_rate": 9.420835702103469e-06, "loss": 1.1265, "step": 16800 }, { "epoch": 1.7877207916577995, "grad_norm": 13.6796875, "learning_rate": 9.42080017055145e-06, "loss": 1.1229, "step": 16801 }, { "epoch": 1.787827197276016, "grad_norm": 9.734375, "learning_rate": 9.420764638999432e-06, "loss": 1.0342, "step": 16802 }, { "epoch": 1.787933602894233, "grad_norm": 11.5625, "learning_rate": 9.420729107447414e-06, "loss": 0.9764, "step": 16803 }, { "epoch": 1.7880400085124495, "grad_norm": 10.5625, "learning_rate": 9.420693575895396e-06, "loss": 1.2808, "step": 16804 }, { "epoch": 1.788146414130666, "grad_norm": 12.7734375, "learning_rate": 9.420658044343379e-06, "loss": 1.1244, "step": 16805 }, { "epoch": 1.7882528197488827, "grad_norm": 16.25, "learning_rate": 9.42062251279136e-06, "loss": 1.2284, "step": 16806 }, { "epoch": 1.7883592253670995, "grad_norm": 10.03125, "learning_rate": 9.420586981239342e-06, "loss": 0.8598, "step": 16807 }, { "epoch": 1.788465630985316, "grad_norm": 13.96875, "learning_rate": 9.420551449687324e-06, "loss": 1.1312, "step": 16808 }, { "epoch": 1.7885720366035327, "grad_norm": 11.03125, "learning_rate": 9.420515918135305e-06, "loss": 1.0646, "step": 16809 }, { "epoch": 1.7886784422217494, "grad_norm": 10.3828125, "learning_rate": 9.420480386583287e-06, "loss": 1.1601, "step": 16810 }, { "epoch": 1.788784847839966, "grad_norm": 11.4453125, "learning_rate": 9.420444855031268e-06, "loss": 0.8947, "step": 16811 }, { "epoch": 1.7888912534581825, "grad_norm": 13.171875, "learning_rate": 9.42040932347925e-06, "loss": 1.1834, "step": 16812 }, { "epoch": 1.7889976590763992, "grad_norm": 14.3515625, "learning_rate": 9.420373791927232e-06, "loss": 1.0039, "step": 16813 }, { "epoch": 1.789104064694616, "grad_norm": 15.1640625, "learning_rate": 9.420338260375213e-06, "loss": 1.3198, "step": 16814 }, { "epoch": 1.7892104703128324, "grad_norm": 15.5703125, "learning_rate": 9.420302728823197e-06, "loss": 0.9042, "step": 16815 }, { "epoch": 1.7893168759310492, "grad_norm": 11.546875, "learning_rate": 9.420267197271178e-06, "loss": 0.9583, "step": 16816 }, { "epoch": 1.7894232815492659, "grad_norm": 12.1328125, "learning_rate": 9.42023166571916e-06, "loss": 0.9416, "step": 16817 }, { "epoch": 1.7895296871674824, "grad_norm": 14.015625, "learning_rate": 9.420196134167141e-06, "loss": 1.1384, "step": 16818 }, { "epoch": 1.7896360927856991, "grad_norm": 23.703125, "learning_rate": 9.420160602615123e-06, "loss": 1.17, "step": 16819 }, { "epoch": 1.7897424984039159, "grad_norm": 11.0390625, "learning_rate": 9.420125071063105e-06, "loss": 1.0012, "step": 16820 }, { "epoch": 1.7898489040221324, "grad_norm": 17.9375, "learning_rate": 9.420089539511086e-06, "loss": 0.9695, "step": 16821 }, { "epoch": 1.7899553096403489, "grad_norm": 14.5, "learning_rate": 9.420054007959068e-06, "loss": 0.9645, "step": 16822 }, { "epoch": 1.7900617152585656, "grad_norm": 16.015625, "learning_rate": 9.42001847640705e-06, "loss": 1.1653, "step": 16823 }, { "epoch": 1.7901681208767823, "grad_norm": 11.84375, "learning_rate": 9.419982944855031e-06, "loss": 1.3347, "step": 16824 }, { "epoch": 1.7902745264949989, "grad_norm": 12.0390625, "learning_rate": 9.419947413303013e-06, "loss": 1.033, "step": 16825 }, { "epoch": 1.7903809321132156, "grad_norm": 13.9140625, "learning_rate": 9.419911881750996e-06, "loss": 0.8699, "step": 16826 }, { "epoch": 1.7904873377314323, "grad_norm": 12.0625, "learning_rate": 9.419876350198978e-06, "loss": 1.2966, "step": 16827 }, { "epoch": 1.7905937433496488, "grad_norm": 9.8203125, "learning_rate": 9.41984081864696e-06, "loss": 0.9679, "step": 16828 }, { "epoch": 1.7907001489678656, "grad_norm": 11.6875, "learning_rate": 9.41980528709494e-06, "loss": 0.9962, "step": 16829 }, { "epoch": 1.7908065545860823, "grad_norm": 12.671875, "learning_rate": 9.419769755542922e-06, "loss": 1.0084, "step": 16830 }, { "epoch": 1.7909129602042988, "grad_norm": 10.171875, "learning_rate": 9.419734223990906e-06, "loss": 0.8829, "step": 16831 }, { "epoch": 1.7910193658225153, "grad_norm": 16.46875, "learning_rate": 9.419698692438887e-06, "loss": 0.9717, "step": 16832 }, { "epoch": 1.791125771440732, "grad_norm": 10.953125, "learning_rate": 9.419663160886869e-06, "loss": 1.3345, "step": 16833 }, { "epoch": 1.7912321770589488, "grad_norm": 16.578125, "learning_rate": 9.41962762933485e-06, "loss": 1.0934, "step": 16834 }, { "epoch": 1.7913385826771653, "grad_norm": 16.390625, "learning_rate": 9.419592097782832e-06, "loss": 0.9328, "step": 16835 }, { "epoch": 1.791444988295382, "grad_norm": 13.3671875, "learning_rate": 9.419556566230814e-06, "loss": 1.0371, "step": 16836 }, { "epoch": 1.7915513939135987, "grad_norm": 9.1640625, "learning_rate": 9.419521034678795e-06, "loss": 1.1653, "step": 16837 }, { "epoch": 1.7916577995318153, "grad_norm": 17.140625, "learning_rate": 9.419485503126777e-06, "loss": 0.9823, "step": 16838 }, { "epoch": 1.7917642051500318, "grad_norm": 15.9921875, "learning_rate": 9.419449971574759e-06, "loss": 1.1785, "step": 16839 }, { "epoch": 1.7918706107682487, "grad_norm": 12.3203125, "learning_rate": 9.41941444002274e-06, "loss": 1.0046, "step": 16840 }, { "epoch": 1.7919770163864652, "grad_norm": 16.28125, "learning_rate": 9.419378908470723e-06, "loss": 1.1028, "step": 16841 }, { "epoch": 1.7920834220046817, "grad_norm": 19.578125, "learning_rate": 9.419343376918705e-06, "loss": 1.2929, "step": 16842 }, { "epoch": 1.7921898276228985, "grad_norm": 12.5703125, "learning_rate": 9.419307845366687e-06, "loss": 1.2326, "step": 16843 }, { "epoch": 1.7922962332411152, "grad_norm": 9.40625, "learning_rate": 9.419272313814668e-06, "loss": 0.9691, "step": 16844 }, { "epoch": 1.7924026388593317, "grad_norm": 12.0703125, "learning_rate": 9.41923678226265e-06, "loss": 1.0346, "step": 16845 }, { "epoch": 1.7925090444775484, "grad_norm": 13.796875, "learning_rate": 9.419201250710631e-06, "loss": 1.1486, "step": 16846 }, { "epoch": 1.7926154500957652, "grad_norm": 15.3984375, "learning_rate": 9.419165719158613e-06, "loss": 1.1348, "step": 16847 }, { "epoch": 1.7927218557139817, "grad_norm": 11.7421875, "learning_rate": 9.419130187606595e-06, "loss": 1.2446, "step": 16848 }, { "epoch": 1.7928282613321982, "grad_norm": 11.7734375, "learning_rate": 9.419094656054576e-06, "loss": 1.5471, "step": 16849 }, { "epoch": 1.7929346669504151, "grad_norm": 20.171875, "learning_rate": 9.419059124502558e-06, "loss": 0.9837, "step": 16850 }, { "epoch": 1.7930410725686317, "grad_norm": 18.4375, "learning_rate": 9.419023592950541e-06, "loss": 1.1311, "step": 16851 }, { "epoch": 1.7931474781868482, "grad_norm": 14.5546875, "learning_rate": 9.418988061398523e-06, "loss": 1.645, "step": 16852 }, { "epoch": 1.793253883805065, "grad_norm": 13.4375, "learning_rate": 9.418952529846504e-06, "loss": 0.9229, "step": 16853 }, { "epoch": 1.7933602894232816, "grad_norm": 18.8125, "learning_rate": 9.418916998294486e-06, "loss": 1.2156, "step": 16854 }, { "epoch": 1.7934666950414981, "grad_norm": 11.4140625, "learning_rate": 9.418881466742468e-06, "loss": 1.2276, "step": 16855 }, { "epoch": 1.7935731006597149, "grad_norm": 10.9140625, "learning_rate": 9.418845935190451e-06, "loss": 1.1734, "step": 16856 }, { "epoch": 1.7936795062779316, "grad_norm": 13.9140625, "learning_rate": 9.418810403638433e-06, "loss": 0.8938, "step": 16857 }, { "epoch": 1.793785911896148, "grad_norm": 13.359375, "learning_rate": 9.418774872086414e-06, "loss": 1.1251, "step": 16858 }, { "epoch": 1.7938923175143646, "grad_norm": 13.8828125, "learning_rate": 9.418739340534396e-06, "loss": 1.0987, "step": 16859 }, { "epoch": 1.7939987231325814, "grad_norm": 12.7578125, "learning_rate": 9.418703808982377e-06, "loss": 1.1963, "step": 16860 }, { "epoch": 1.794105128750798, "grad_norm": 12.1875, "learning_rate": 9.418668277430359e-06, "loss": 1.3354, "step": 16861 }, { "epoch": 1.7942115343690146, "grad_norm": 14.265625, "learning_rate": 9.41863274587834e-06, "loss": 0.9865, "step": 16862 }, { "epoch": 1.7943179399872313, "grad_norm": 9.2578125, "learning_rate": 9.418597214326322e-06, "loss": 0.5882, "step": 16863 }, { "epoch": 1.794424345605448, "grad_norm": 10.6875, "learning_rate": 9.418561682774304e-06, "loss": 0.996, "step": 16864 }, { "epoch": 1.7945307512236646, "grad_norm": 12.3828125, "learning_rate": 9.418526151222285e-06, "loss": 1.2954, "step": 16865 }, { "epoch": 1.7946371568418813, "grad_norm": 12.7109375, "learning_rate": 9.418490619670267e-06, "loss": 1.2364, "step": 16866 }, { "epoch": 1.794743562460098, "grad_norm": 10.765625, "learning_rate": 9.41845508811825e-06, "loss": 1.2421, "step": 16867 }, { "epoch": 1.7948499680783145, "grad_norm": 10.4921875, "learning_rate": 9.418419556566232e-06, "loss": 0.94, "step": 16868 }, { "epoch": 1.794956373696531, "grad_norm": 18.609375, "learning_rate": 9.418384025014214e-06, "loss": 0.8313, "step": 16869 }, { "epoch": 1.7950627793147478, "grad_norm": 11.1484375, "learning_rate": 9.418348493462195e-06, "loss": 0.9892, "step": 16870 }, { "epoch": 1.7951691849329645, "grad_norm": 10.796875, "learning_rate": 9.418312961910177e-06, "loss": 1.1384, "step": 16871 }, { "epoch": 1.795275590551181, "grad_norm": 12.546875, "learning_rate": 9.41827743035816e-06, "loss": 1.1399, "step": 16872 }, { "epoch": 1.7953819961693978, "grad_norm": 15.7734375, "learning_rate": 9.418241898806142e-06, "loss": 1.058, "step": 16873 }, { "epoch": 1.7954884017876145, "grad_norm": 13.21875, "learning_rate": 9.418206367254123e-06, "loss": 1.1175, "step": 16874 }, { "epoch": 1.795594807405831, "grad_norm": 11.0234375, "learning_rate": 9.418170835702105e-06, "loss": 0.9893, "step": 16875 }, { "epoch": 1.7957012130240475, "grad_norm": 8.5625, "learning_rate": 9.418135304150085e-06, "loss": 0.8035, "step": 16876 }, { "epoch": 1.7958076186422645, "grad_norm": 16.09375, "learning_rate": 9.418099772598068e-06, "loss": 1.2463, "step": 16877 }, { "epoch": 1.795914024260481, "grad_norm": 12.5625, "learning_rate": 9.41806424104605e-06, "loss": 0.9871, "step": 16878 }, { "epoch": 1.7960204298786975, "grad_norm": 11.84375, "learning_rate": 9.418028709494031e-06, "loss": 1.2361, "step": 16879 }, { "epoch": 1.7961268354969142, "grad_norm": 12.1328125, "learning_rate": 9.417993177942013e-06, "loss": 1.4122, "step": 16880 }, { "epoch": 1.796233241115131, "grad_norm": 14.0546875, "learning_rate": 9.417957646389994e-06, "loss": 1.3619, "step": 16881 }, { "epoch": 1.7963396467333475, "grad_norm": 10.7421875, "learning_rate": 9.417922114837978e-06, "loss": 0.9496, "step": 16882 }, { "epoch": 1.7964460523515642, "grad_norm": 12.0078125, "learning_rate": 9.41788658328596e-06, "loss": 1.0212, "step": 16883 }, { "epoch": 1.796552457969781, "grad_norm": 11.9609375, "learning_rate": 9.417851051733941e-06, "loss": 1.1524, "step": 16884 }, { "epoch": 1.7966588635879974, "grad_norm": 13.4453125, "learning_rate": 9.417815520181923e-06, "loss": 1.1401, "step": 16885 }, { "epoch": 1.796765269206214, "grad_norm": 12.8515625, "learning_rate": 9.417779988629904e-06, "loss": 1.0909, "step": 16886 }, { "epoch": 1.7968716748244309, "grad_norm": 10.0546875, "learning_rate": 9.417744457077886e-06, "loss": 1.1274, "step": 16887 }, { "epoch": 1.7969780804426474, "grad_norm": 13.5234375, "learning_rate": 9.417708925525867e-06, "loss": 1.321, "step": 16888 }, { "epoch": 1.797084486060864, "grad_norm": 14.578125, "learning_rate": 9.417673393973849e-06, "loss": 1.2077, "step": 16889 }, { "epoch": 1.7971908916790806, "grad_norm": 14.90625, "learning_rate": 9.41763786242183e-06, "loss": 1.0542, "step": 16890 }, { "epoch": 1.7972972972972974, "grad_norm": 11.109375, "learning_rate": 9.417602330869812e-06, "loss": 1.2771, "step": 16891 }, { "epoch": 1.7974037029155139, "grad_norm": 11.5546875, "learning_rate": 9.417566799317796e-06, "loss": 0.9816, "step": 16892 }, { "epoch": 1.7975101085337306, "grad_norm": 14.3828125, "learning_rate": 9.417531267765777e-06, "loss": 1.0327, "step": 16893 }, { "epoch": 1.7976165141519473, "grad_norm": 10.1953125, "learning_rate": 9.417495736213759e-06, "loss": 1.2572, "step": 16894 }, { "epoch": 1.7977229197701639, "grad_norm": 11.8203125, "learning_rate": 9.41746020466174e-06, "loss": 1.1375, "step": 16895 }, { "epoch": 1.7978293253883804, "grad_norm": 12.0390625, "learning_rate": 9.417424673109722e-06, "loss": 1.135, "step": 16896 }, { "epoch": 1.797935731006597, "grad_norm": 14.140625, "learning_rate": 9.417389141557705e-06, "loss": 0.9939, "step": 16897 }, { "epoch": 1.7980421366248138, "grad_norm": 13.4140625, "learning_rate": 9.417353610005687e-06, "loss": 1.1869, "step": 16898 }, { "epoch": 1.7981485422430303, "grad_norm": 11.4296875, "learning_rate": 9.417318078453668e-06, "loss": 1.0101, "step": 16899 }, { "epoch": 1.798254947861247, "grad_norm": 8.859375, "learning_rate": 9.41728254690165e-06, "loss": 0.8848, "step": 16900 }, { "epoch": 1.7983613534794638, "grad_norm": 10.6015625, "learning_rate": 9.417247015349632e-06, "loss": 1.2157, "step": 16901 }, { "epoch": 1.7984677590976803, "grad_norm": 9.875, "learning_rate": 9.417211483797613e-06, "loss": 0.8988, "step": 16902 }, { "epoch": 1.798574164715897, "grad_norm": 10.8671875, "learning_rate": 9.417175952245595e-06, "loss": 1.0586, "step": 16903 }, { "epoch": 1.7986805703341138, "grad_norm": 9.65625, "learning_rate": 9.417140420693577e-06, "loss": 0.9495, "step": 16904 }, { "epoch": 1.7987869759523303, "grad_norm": 9.421875, "learning_rate": 9.417104889141558e-06, "loss": 1.0585, "step": 16905 }, { "epoch": 1.7988933815705468, "grad_norm": 121.9375, "learning_rate": 9.41706935758954e-06, "loss": 1.0725, "step": 16906 }, { "epoch": 1.7989997871887635, "grad_norm": 11.4921875, "learning_rate": 9.417033826037521e-06, "loss": 1.0182, "step": 16907 }, { "epoch": 1.7991061928069803, "grad_norm": 9.703125, "learning_rate": 9.416998294485505e-06, "loss": 0.943, "step": 16908 }, { "epoch": 1.7992125984251968, "grad_norm": 13.515625, "learning_rate": 9.416962762933486e-06, "loss": 1.3573, "step": 16909 }, { "epoch": 1.7993190040434135, "grad_norm": 12.171875, "learning_rate": 9.416927231381468e-06, "loss": 1.1537, "step": 16910 }, { "epoch": 1.7994254096616302, "grad_norm": 12.53125, "learning_rate": 9.41689169982945e-06, "loss": 1.1591, "step": 16911 }, { "epoch": 1.7995318152798467, "grad_norm": 11.5390625, "learning_rate": 9.416856168277431e-06, "loss": 1.0339, "step": 16912 }, { "epoch": 1.7996382208980635, "grad_norm": 14.46875, "learning_rate": 9.416820636725413e-06, "loss": 1.0014, "step": 16913 }, { "epoch": 1.7997446265162802, "grad_norm": 10.6953125, "learning_rate": 9.416785105173394e-06, "loss": 1.0079, "step": 16914 }, { "epoch": 1.7998510321344967, "grad_norm": 15.4140625, "learning_rate": 9.416749573621376e-06, "loss": 1.1078, "step": 16915 }, { "epoch": 1.7999574377527132, "grad_norm": 19.234375, "learning_rate": 9.416714042069357e-06, "loss": 1.2814, "step": 16916 }, { "epoch": 1.80006384337093, "grad_norm": 12.8203125, "learning_rate": 9.416678510517339e-06, "loss": 1.1089, "step": 16917 }, { "epoch": 1.8001702489891467, "grad_norm": 11.6953125, "learning_rate": 9.416642978965322e-06, "loss": 1.0588, "step": 16918 }, { "epoch": 1.8002766546073632, "grad_norm": 12.6171875, "learning_rate": 9.416607447413304e-06, "loss": 1.0529, "step": 16919 }, { "epoch": 1.80038306022558, "grad_norm": 9.828125, "learning_rate": 9.416571915861286e-06, "loss": 1.3364, "step": 16920 }, { "epoch": 1.8004894658437967, "grad_norm": 10.140625, "learning_rate": 9.416536384309267e-06, "loss": 1.0332, "step": 16921 }, { "epoch": 1.8005958714620132, "grad_norm": 12.8359375, "learning_rate": 9.416500852757249e-06, "loss": 0.9985, "step": 16922 }, { "epoch": 1.8007022770802297, "grad_norm": 13.875, "learning_rate": 9.416465321205232e-06, "loss": 1.0514, "step": 16923 }, { "epoch": 1.8008086826984466, "grad_norm": 13.453125, "learning_rate": 9.416429789653214e-06, "loss": 1.1564, "step": 16924 }, { "epoch": 1.8009150883166631, "grad_norm": 13.515625, "learning_rate": 9.416394258101195e-06, "loss": 1.1089, "step": 16925 }, { "epoch": 1.8010214939348796, "grad_norm": 17.09375, "learning_rate": 9.416358726549177e-06, "loss": 1.045, "step": 16926 }, { "epoch": 1.8011278995530964, "grad_norm": 18.203125, "learning_rate": 9.416323194997159e-06, "loss": 1.2001, "step": 16927 }, { "epoch": 1.801234305171313, "grad_norm": 10.7578125, "learning_rate": 9.41628766344514e-06, "loss": 1.0289, "step": 16928 }, { "epoch": 1.8013407107895296, "grad_norm": 13.953125, "learning_rate": 9.416252131893122e-06, "loss": 0.9479, "step": 16929 }, { "epoch": 1.8014471164077464, "grad_norm": 10.1640625, "learning_rate": 9.416216600341103e-06, "loss": 0.9895, "step": 16930 }, { "epoch": 1.801553522025963, "grad_norm": 11.546875, "learning_rate": 9.416181068789085e-06, "loss": 1.0754, "step": 16931 }, { "epoch": 1.8016599276441796, "grad_norm": 29.59375, "learning_rate": 9.416145537237067e-06, "loss": 0.7844, "step": 16932 }, { "epoch": 1.801766333262396, "grad_norm": 11.0078125, "learning_rate": 9.41611000568505e-06, "loss": 1.1035, "step": 16933 }, { "epoch": 1.801872738880613, "grad_norm": 11.8671875, "learning_rate": 9.416074474133031e-06, "loss": 1.06, "step": 16934 }, { "epoch": 1.8019791444988296, "grad_norm": 14.890625, "learning_rate": 9.416038942581013e-06, "loss": 1.0828, "step": 16935 }, { "epoch": 1.802085550117046, "grad_norm": 11.109375, "learning_rate": 9.416003411028995e-06, "loss": 0.9773, "step": 16936 }, { "epoch": 1.8021919557352628, "grad_norm": 12.1640625, "learning_rate": 9.415967879476976e-06, "loss": 1.2111, "step": 16937 }, { "epoch": 1.8022983613534795, "grad_norm": 12.703125, "learning_rate": 9.415932347924958e-06, "loss": 1.1234, "step": 16938 }, { "epoch": 1.802404766971696, "grad_norm": 13.546875, "learning_rate": 9.41589681637294e-06, "loss": 1.0218, "step": 16939 }, { "epoch": 1.8025111725899128, "grad_norm": 12.9921875, "learning_rate": 9.415861284820921e-06, "loss": 1.1394, "step": 16940 }, { "epoch": 1.8026175782081295, "grad_norm": 10.6015625, "learning_rate": 9.415825753268903e-06, "loss": 0.8226, "step": 16941 }, { "epoch": 1.802723983826346, "grad_norm": 14.4375, "learning_rate": 9.415790221716884e-06, "loss": 0.9657, "step": 16942 }, { "epoch": 1.8028303894445625, "grad_norm": 10.015625, "learning_rate": 9.415754690164866e-06, "loss": 1.1025, "step": 16943 }, { "epoch": 1.8029367950627793, "grad_norm": 11.625, "learning_rate": 9.41571915861285e-06, "loss": 1.5079, "step": 16944 }, { "epoch": 1.803043200680996, "grad_norm": 11.5625, "learning_rate": 9.415683627060831e-06, "loss": 1.1025, "step": 16945 }, { "epoch": 1.8031496062992125, "grad_norm": 9.984375, "learning_rate": 9.415648095508812e-06, "loss": 1.11, "step": 16946 }, { "epoch": 1.8032560119174292, "grad_norm": 14.515625, "learning_rate": 9.415612563956794e-06, "loss": 1.142, "step": 16947 }, { "epoch": 1.803362417535646, "grad_norm": 17.015625, "learning_rate": 9.415577032404776e-06, "loss": 1.1181, "step": 16948 }, { "epoch": 1.8034688231538625, "grad_norm": 15.484375, "learning_rate": 9.415541500852759e-06, "loss": 0.9899, "step": 16949 }, { "epoch": 1.8035752287720792, "grad_norm": 13.90625, "learning_rate": 9.41550596930074e-06, "loss": 1.1017, "step": 16950 }, { "epoch": 1.803681634390296, "grad_norm": 14.2578125, "learning_rate": 9.415470437748722e-06, "loss": 1.1547, "step": 16951 }, { "epoch": 1.8037880400085124, "grad_norm": 10.65625, "learning_rate": 9.415434906196704e-06, "loss": 1.3495, "step": 16952 }, { "epoch": 1.803894445626729, "grad_norm": 11.5078125, "learning_rate": 9.415399374644685e-06, "loss": 1.2631, "step": 16953 }, { "epoch": 1.8040008512449457, "grad_norm": 11.671875, "learning_rate": 9.415363843092667e-06, "loss": 1.0177, "step": 16954 }, { "epoch": 1.8041072568631624, "grad_norm": 14.5703125, "learning_rate": 9.415328311540649e-06, "loss": 1.0534, "step": 16955 }, { "epoch": 1.804213662481379, "grad_norm": 11.8359375, "learning_rate": 9.41529277998863e-06, "loss": 1.0652, "step": 16956 }, { "epoch": 1.8043200680995957, "grad_norm": 17.40625, "learning_rate": 9.415257248436612e-06, "loss": 1.1873, "step": 16957 }, { "epoch": 1.8044264737178124, "grad_norm": 16.609375, "learning_rate": 9.415221716884593e-06, "loss": 1.1349, "step": 16958 }, { "epoch": 1.804532879336029, "grad_norm": 16.796875, "learning_rate": 9.415186185332577e-06, "loss": 0.9804, "step": 16959 }, { "epoch": 1.8046392849542456, "grad_norm": 93.0, "learning_rate": 9.415150653780558e-06, "loss": 1.1945, "step": 16960 }, { "epoch": 1.8047456905724624, "grad_norm": 17.515625, "learning_rate": 9.41511512222854e-06, "loss": 1.125, "step": 16961 }, { "epoch": 1.8048520961906789, "grad_norm": 13.484375, "learning_rate": 9.415079590676522e-06, "loss": 0.8922, "step": 16962 }, { "epoch": 1.8049585018088954, "grad_norm": 14.453125, "learning_rate": 9.415044059124503e-06, "loss": 1.1994, "step": 16963 }, { "epoch": 1.8050649074271121, "grad_norm": 16.921875, "learning_rate": 9.415008527572486e-06, "loss": 1.2633, "step": 16964 }, { "epoch": 1.8051713130453289, "grad_norm": 10.1875, "learning_rate": 9.414972996020466e-06, "loss": 0.8381, "step": 16965 }, { "epoch": 1.8052777186635454, "grad_norm": 17.6875, "learning_rate": 9.414937464468448e-06, "loss": 1.0826, "step": 16966 }, { "epoch": 1.805384124281762, "grad_norm": 13.3671875, "learning_rate": 9.41490193291643e-06, "loss": 0.8887, "step": 16967 }, { "epoch": 1.8054905298999788, "grad_norm": 16.75, "learning_rate": 9.414866401364411e-06, "loss": 1.456, "step": 16968 }, { "epoch": 1.8055969355181953, "grad_norm": 12.6953125, "learning_rate": 9.414830869812394e-06, "loss": 1.1973, "step": 16969 }, { "epoch": 1.8057033411364118, "grad_norm": 11.9921875, "learning_rate": 9.414795338260376e-06, "loss": 1.0637, "step": 16970 }, { "epoch": 1.8058097467546288, "grad_norm": 18.0, "learning_rate": 9.414759806708358e-06, "loss": 0.9838, "step": 16971 }, { "epoch": 1.8059161523728453, "grad_norm": 12.8515625, "learning_rate": 9.41472427515634e-06, "loss": 1.0792, "step": 16972 }, { "epoch": 1.8060225579910618, "grad_norm": 11.0390625, "learning_rate": 9.414688743604321e-06, "loss": 1.1619, "step": 16973 }, { "epoch": 1.8061289636092785, "grad_norm": 12.90625, "learning_rate": 9.414653212052304e-06, "loss": 1.2836, "step": 16974 }, { "epoch": 1.8062353692274953, "grad_norm": 16.328125, "learning_rate": 9.414617680500286e-06, "loss": 1.1055, "step": 16975 }, { "epoch": 1.8063417748457118, "grad_norm": 14.1875, "learning_rate": 9.414582148948267e-06, "loss": 1.1537, "step": 16976 }, { "epoch": 1.8064481804639285, "grad_norm": 12.578125, "learning_rate": 9.414546617396249e-06, "loss": 1.0666, "step": 16977 }, { "epoch": 1.8065545860821453, "grad_norm": 12.765625, "learning_rate": 9.41451108584423e-06, "loss": 1.3831, "step": 16978 }, { "epoch": 1.8066609917003618, "grad_norm": 12.265625, "learning_rate": 9.414475554292212e-06, "loss": 1.3445, "step": 16979 }, { "epoch": 1.8067673973185783, "grad_norm": 11.8984375, "learning_rate": 9.414440022740194e-06, "loss": 0.9471, "step": 16980 }, { "epoch": 1.8068738029367952, "grad_norm": 12.671875, "learning_rate": 9.414404491188175e-06, "loss": 1.2373, "step": 16981 }, { "epoch": 1.8069802085550117, "grad_norm": 11.265625, "learning_rate": 9.414368959636157e-06, "loss": 1.2335, "step": 16982 }, { "epoch": 1.8070866141732282, "grad_norm": 11.609375, "learning_rate": 9.414333428084139e-06, "loss": 1.1273, "step": 16983 }, { "epoch": 1.807193019791445, "grad_norm": 8.9765625, "learning_rate": 9.41429789653212e-06, "loss": 0.8624, "step": 16984 }, { "epoch": 1.8072994254096617, "grad_norm": 10.640625, "learning_rate": 9.414262364980104e-06, "loss": 1.0623, "step": 16985 }, { "epoch": 1.8074058310278782, "grad_norm": 11.171875, "learning_rate": 9.414226833428085e-06, "loss": 1.1807, "step": 16986 }, { "epoch": 1.807512236646095, "grad_norm": 12.140625, "learning_rate": 9.414191301876067e-06, "loss": 1.1374, "step": 16987 }, { "epoch": 1.8076186422643117, "grad_norm": 17.453125, "learning_rate": 9.414155770324048e-06, "loss": 0.9912, "step": 16988 }, { "epoch": 1.8077250478825282, "grad_norm": 12.9375, "learning_rate": 9.41412023877203e-06, "loss": 1.1702, "step": 16989 }, { "epoch": 1.8078314535007447, "grad_norm": 13.0859375, "learning_rate": 9.414084707220013e-06, "loss": 1.0611, "step": 16990 }, { "epoch": 1.8079378591189614, "grad_norm": 11.6171875, "learning_rate": 9.414049175667995e-06, "loss": 1.2688, "step": 16991 }, { "epoch": 1.8080442647371782, "grad_norm": 12.734375, "learning_rate": 9.414013644115977e-06, "loss": 0.8809, "step": 16992 }, { "epoch": 1.8081506703553947, "grad_norm": 11.984375, "learning_rate": 9.413978112563958e-06, "loss": 0.9203, "step": 16993 }, { "epoch": 1.8082570759736114, "grad_norm": 11.390625, "learning_rate": 9.41394258101194e-06, "loss": 0.8489, "step": 16994 }, { "epoch": 1.8083634815918281, "grad_norm": 13.34375, "learning_rate": 9.413907049459921e-06, "loss": 1.2655, "step": 16995 }, { "epoch": 1.8084698872100446, "grad_norm": 14.6875, "learning_rate": 9.413871517907903e-06, "loss": 1.1506, "step": 16996 }, { "epoch": 1.8085762928282614, "grad_norm": 13.1796875, "learning_rate": 9.413835986355885e-06, "loss": 0.9333, "step": 16997 }, { "epoch": 1.808682698446478, "grad_norm": 16.359375, "learning_rate": 9.413800454803866e-06, "loss": 1.048, "step": 16998 }, { "epoch": 1.8087891040646946, "grad_norm": 11.59375, "learning_rate": 9.413764923251848e-06, "loss": 1.1257, "step": 16999 }, { "epoch": 1.8088955096829111, "grad_norm": 13.3984375, "learning_rate": 9.413729391699831e-06, "loss": 1.0032, "step": 17000 }, { "epoch": 1.8090019153011279, "grad_norm": 12.5078125, "learning_rate": 9.413693860147813e-06, "loss": 0.9695, "step": 17001 }, { "epoch": 1.8091083209193446, "grad_norm": 11.765625, "learning_rate": 9.413658328595794e-06, "loss": 1.1877, "step": 17002 }, { "epoch": 1.809214726537561, "grad_norm": 16.28125, "learning_rate": 9.413622797043776e-06, "loss": 1.3172, "step": 17003 }, { "epoch": 1.8093211321557778, "grad_norm": 11.1796875, "learning_rate": 9.413587265491758e-06, "loss": 1.0645, "step": 17004 }, { "epoch": 1.8094275377739946, "grad_norm": 8.703125, "learning_rate": 9.413551733939739e-06, "loss": 0.9733, "step": 17005 }, { "epoch": 1.809533943392211, "grad_norm": 10.6796875, "learning_rate": 9.41351620238772e-06, "loss": 1.0099, "step": 17006 }, { "epoch": 1.8096403490104276, "grad_norm": 10.1875, "learning_rate": 9.413480670835702e-06, "loss": 1.2163, "step": 17007 }, { "epoch": 1.8097467546286445, "grad_norm": 11.1875, "learning_rate": 9.413445139283684e-06, "loss": 1.2332, "step": 17008 }, { "epoch": 1.809853160246861, "grad_norm": 12.3828125, "learning_rate": 9.413409607731666e-06, "loss": 0.9191, "step": 17009 }, { "epoch": 1.8099595658650776, "grad_norm": 10.765625, "learning_rate": 9.413374076179649e-06, "loss": 1.0887, "step": 17010 }, { "epoch": 1.8100659714832943, "grad_norm": 11.921875, "learning_rate": 9.41333854462763e-06, "loss": 0.8313, "step": 17011 }, { "epoch": 1.810172377101511, "grad_norm": 9.15625, "learning_rate": 9.413303013075612e-06, "loss": 0.9292, "step": 17012 }, { "epoch": 1.8102787827197275, "grad_norm": 12.453125, "learning_rate": 9.413267481523594e-06, "loss": 1.0853, "step": 17013 }, { "epoch": 1.8103851883379443, "grad_norm": 11.5390625, "learning_rate": 9.413231949971575e-06, "loss": 0.9713, "step": 17014 }, { "epoch": 1.810491593956161, "grad_norm": 18.0, "learning_rate": 9.413196418419559e-06, "loss": 0.9771, "step": 17015 }, { "epoch": 1.8105979995743775, "grad_norm": 9.0, "learning_rate": 9.41316088686754e-06, "loss": 0.9158, "step": 17016 }, { "epoch": 1.810704405192594, "grad_norm": 18.78125, "learning_rate": 9.413125355315522e-06, "loss": 0.8705, "step": 17017 }, { "epoch": 1.810810810810811, "grad_norm": 16.96875, "learning_rate": 9.413089823763503e-06, "loss": 1.133, "step": 17018 }, { "epoch": 1.8109172164290275, "grad_norm": 13.2890625, "learning_rate": 9.413054292211485e-06, "loss": 1.1628, "step": 17019 }, { "epoch": 1.811023622047244, "grad_norm": 16.5625, "learning_rate": 9.413018760659467e-06, "loss": 1.0315, "step": 17020 }, { "epoch": 1.8111300276654607, "grad_norm": 10.8671875, "learning_rate": 9.412983229107448e-06, "loss": 0.9377, "step": 17021 }, { "epoch": 1.8112364332836774, "grad_norm": 16.25, "learning_rate": 9.41294769755543e-06, "loss": 1.1983, "step": 17022 }, { "epoch": 1.811342838901894, "grad_norm": 9.1640625, "learning_rate": 9.412912166003411e-06, "loss": 1.112, "step": 17023 }, { "epoch": 1.8114492445201107, "grad_norm": 13.8359375, "learning_rate": 9.412876634451393e-06, "loss": 1.1774, "step": 17024 }, { "epoch": 1.8115556501383274, "grad_norm": 12.7265625, "learning_rate": 9.412841102899375e-06, "loss": 1.3375, "step": 17025 }, { "epoch": 1.811662055756544, "grad_norm": 12.53125, "learning_rate": 9.412805571347358e-06, "loss": 0.9604, "step": 17026 }, { "epoch": 1.8117684613747604, "grad_norm": 10.3203125, "learning_rate": 9.41277003979534e-06, "loss": 1.1817, "step": 17027 }, { "epoch": 1.8118748669929774, "grad_norm": 11.3828125, "learning_rate": 9.412734508243321e-06, "loss": 0.8824, "step": 17028 }, { "epoch": 1.811981272611194, "grad_norm": 14.890625, "learning_rate": 9.412698976691303e-06, "loss": 0.9221, "step": 17029 }, { "epoch": 1.8120876782294104, "grad_norm": 14.484375, "learning_rate": 9.412663445139284e-06, "loss": 1.2598, "step": 17030 }, { "epoch": 1.8121940838476271, "grad_norm": 11.9140625, "learning_rate": 9.412627913587266e-06, "loss": 1.0583, "step": 17031 }, { "epoch": 1.8123004894658439, "grad_norm": 17.890625, "learning_rate": 9.412592382035248e-06, "loss": 0.9456, "step": 17032 }, { "epoch": 1.8124068950840604, "grad_norm": 9.3203125, "learning_rate": 9.41255685048323e-06, "loss": 0.9077, "step": 17033 }, { "epoch": 1.8125133007022771, "grad_norm": 11.8671875, "learning_rate": 9.41252131893121e-06, "loss": 0.945, "step": 17034 }, { "epoch": 1.8126197063204939, "grad_norm": 10.234375, "learning_rate": 9.412485787379192e-06, "loss": 1.191, "step": 17035 }, { "epoch": 1.8127261119387104, "grad_norm": 15.2578125, "learning_rate": 9.412450255827176e-06, "loss": 1.1212, "step": 17036 }, { "epoch": 1.8128325175569269, "grad_norm": 16.28125, "learning_rate": 9.412414724275157e-06, "loss": 0.9287, "step": 17037 }, { "epoch": 1.8129389231751436, "grad_norm": 16.890625, "learning_rate": 9.412379192723139e-06, "loss": 1.0932, "step": 17038 }, { "epoch": 1.8130453287933603, "grad_norm": 13.75, "learning_rate": 9.41234366117112e-06, "loss": 1.0659, "step": 17039 }, { "epoch": 1.8131517344115768, "grad_norm": 12.296875, "learning_rate": 9.412308129619102e-06, "loss": 1.4022, "step": 17040 }, { "epoch": 1.8132581400297936, "grad_norm": 13.5234375, "learning_rate": 9.412272598067085e-06, "loss": 0.9686, "step": 17041 }, { "epoch": 1.8133645456480103, "grad_norm": 12.25, "learning_rate": 9.412237066515067e-06, "loss": 1.4327, "step": 17042 }, { "epoch": 1.8134709512662268, "grad_norm": 13.3515625, "learning_rate": 9.412201534963049e-06, "loss": 1.2629, "step": 17043 }, { "epoch": 1.8135773568844435, "grad_norm": 13.765625, "learning_rate": 9.41216600341103e-06, "loss": 1.3019, "step": 17044 }, { "epoch": 1.8136837625026603, "grad_norm": 13.765625, "learning_rate": 9.412130471859012e-06, "loss": 0.9417, "step": 17045 }, { "epoch": 1.8137901681208768, "grad_norm": 12.3671875, "learning_rate": 9.412094940306993e-06, "loss": 1.1406, "step": 17046 }, { "epoch": 1.8138965737390933, "grad_norm": 14.0703125, "learning_rate": 9.412059408754975e-06, "loss": 0.8275, "step": 17047 }, { "epoch": 1.81400297935731, "grad_norm": 16.3125, "learning_rate": 9.412023877202957e-06, "loss": 0.9695, "step": 17048 }, { "epoch": 1.8141093849755268, "grad_norm": 13.5, "learning_rate": 9.411988345650938e-06, "loss": 1.0326, "step": 17049 }, { "epoch": 1.8142157905937433, "grad_norm": 13.171875, "learning_rate": 9.41195281409892e-06, "loss": 1.109, "step": 17050 }, { "epoch": 1.81432219621196, "grad_norm": 10.5078125, "learning_rate": 9.411917282546903e-06, "loss": 1.1906, "step": 17051 }, { "epoch": 1.8144286018301767, "grad_norm": 11.2109375, "learning_rate": 9.411881750994885e-06, "loss": 1.0652, "step": 17052 }, { "epoch": 1.8145350074483932, "grad_norm": 9.6015625, "learning_rate": 9.411846219442866e-06, "loss": 0.9503, "step": 17053 }, { "epoch": 1.8146414130666098, "grad_norm": 19.09375, "learning_rate": 9.411810687890848e-06, "loss": 1.0535, "step": 17054 }, { "epoch": 1.8147478186848267, "grad_norm": 13.484375, "learning_rate": 9.41177515633883e-06, "loss": 0.9407, "step": 17055 }, { "epoch": 1.8148542243030432, "grad_norm": 10.296875, "learning_rate": 9.411739624786811e-06, "loss": 1.0838, "step": 17056 }, { "epoch": 1.8149606299212597, "grad_norm": 13.8671875, "learning_rate": 9.411704093234793e-06, "loss": 1.3041, "step": 17057 }, { "epoch": 1.8150670355394765, "grad_norm": 11.984375, "learning_rate": 9.411668561682774e-06, "loss": 0.9998, "step": 17058 }, { "epoch": 1.8151734411576932, "grad_norm": 15.4296875, "learning_rate": 9.411633030130756e-06, "loss": 1.0962, "step": 17059 }, { "epoch": 1.8152798467759097, "grad_norm": 13.546875, "learning_rate": 9.411597498578738e-06, "loss": 1.0347, "step": 17060 }, { "epoch": 1.8153862523941264, "grad_norm": 9.9921875, "learning_rate": 9.41156196702672e-06, "loss": 1.0553, "step": 17061 }, { "epoch": 1.8154926580123432, "grad_norm": 17.78125, "learning_rate": 9.411526435474703e-06, "loss": 1.4072, "step": 17062 }, { "epoch": 1.8155990636305597, "grad_norm": 13.171875, "learning_rate": 9.411490903922684e-06, "loss": 1.2176, "step": 17063 }, { "epoch": 1.8157054692487762, "grad_norm": 10.796875, "learning_rate": 9.411455372370666e-06, "loss": 1.1602, "step": 17064 }, { "epoch": 1.8158118748669931, "grad_norm": 11.9921875, "learning_rate": 9.411419840818647e-06, "loss": 0.9568, "step": 17065 }, { "epoch": 1.8159182804852096, "grad_norm": 10.2421875, "learning_rate": 9.411384309266629e-06, "loss": 1.2285, "step": 17066 }, { "epoch": 1.8160246861034262, "grad_norm": 7.98046875, "learning_rate": 9.411348777714612e-06, "loss": 1.0269, "step": 17067 }, { "epoch": 1.8161310917216429, "grad_norm": 10.0703125, "learning_rate": 9.411313246162594e-06, "loss": 1.1086, "step": 17068 }, { "epoch": 1.8162374973398596, "grad_norm": 11.4375, "learning_rate": 9.411277714610575e-06, "loss": 0.8642, "step": 17069 }, { "epoch": 1.8163439029580761, "grad_norm": 11.5703125, "learning_rate": 9.411242183058557e-06, "loss": 0.7896, "step": 17070 }, { "epoch": 1.8164503085762929, "grad_norm": 14.375, "learning_rate": 9.411206651506539e-06, "loss": 1.3623, "step": 17071 }, { "epoch": 1.8165567141945096, "grad_norm": 12.1875, "learning_rate": 9.41117111995452e-06, "loss": 0.9025, "step": 17072 }, { "epoch": 1.816663119812726, "grad_norm": 12.21875, "learning_rate": 9.411135588402502e-06, "loss": 1.1933, "step": 17073 }, { "epoch": 1.8167695254309426, "grad_norm": 14.0, "learning_rate": 9.411100056850484e-06, "loss": 1.2005, "step": 17074 }, { "epoch": 1.8168759310491593, "grad_norm": 10.453125, "learning_rate": 9.411064525298465e-06, "loss": 1.2852, "step": 17075 }, { "epoch": 1.816982336667376, "grad_norm": 14.53125, "learning_rate": 9.411028993746447e-06, "loss": 1.0671, "step": 17076 }, { "epoch": 1.8170887422855926, "grad_norm": 11.578125, "learning_rate": 9.41099346219443e-06, "loss": 0.948, "step": 17077 }, { "epoch": 1.8171951479038093, "grad_norm": 14.375, "learning_rate": 9.410957930642412e-06, "loss": 0.9908, "step": 17078 }, { "epoch": 1.817301553522026, "grad_norm": 11.28125, "learning_rate": 9.410922399090393e-06, "loss": 1.1543, "step": 17079 }, { "epoch": 1.8174079591402426, "grad_norm": 13.1484375, "learning_rate": 9.410886867538375e-06, "loss": 1.3334, "step": 17080 }, { "epoch": 1.8175143647584593, "grad_norm": 10.671875, "learning_rate": 9.410851335986356e-06, "loss": 1.2334, "step": 17081 }, { "epoch": 1.817620770376676, "grad_norm": 8.5703125, "learning_rate": 9.41081580443434e-06, "loss": 1.0347, "step": 17082 }, { "epoch": 1.8177271759948925, "grad_norm": 20.25, "learning_rate": 9.410780272882321e-06, "loss": 1.13, "step": 17083 }, { "epoch": 1.817833581613109, "grad_norm": 15.9765625, "learning_rate": 9.410744741330303e-06, "loss": 1.2421, "step": 17084 }, { "epoch": 1.8179399872313258, "grad_norm": 10.71875, "learning_rate": 9.410709209778283e-06, "loss": 1.3364, "step": 17085 }, { "epoch": 1.8180463928495425, "grad_norm": 16.1875, "learning_rate": 9.410673678226264e-06, "loss": 0.9544, "step": 17086 }, { "epoch": 1.818152798467759, "grad_norm": 12.4140625, "learning_rate": 9.410638146674248e-06, "loss": 1.0602, "step": 17087 }, { "epoch": 1.8182592040859757, "grad_norm": 10.5859375, "learning_rate": 9.41060261512223e-06, "loss": 1.093, "step": 17088 }, { "epoch": 1.8183656097041925, "grad_norm": 10.6015625, "learning_rate": 9.410567083570211e-06, "loss": 1.0187, "step": 17089 }, { "epoch": 1.818472015322409, "grad_norm": 10.625, "learning_rate": 9.410531552018193e-06, "loss": 0.9972, "step": 17090 }, { "epoch": 1.8185784209406257, "grad_norm": 13.03125, "learning_rate": 9.410496020466174e-06, "loss": 1.3391, "step": 17091 }, { "epoch": 1.8186848265588424, "grad_norm": 11.6015625, "learning_rate": 9.410460488914158e-06, "loss": 1.4455, "step": 17092 }, { "epoch": 1.818791232177059, "grad_norm": 11.3125, "learning_rate": 9.410424957362139e-06, "loss": 0.9793, "step": 17093 }, { "epoch": 1.8188976377952755, "grad_norm": 9.84375, "learning_rate": 9.41038942581012e-06, "loss": 0.8745, "step": 17094 }, { "epoch": 1.8190040434134922, "grad_norm": 9.625, "learning_rate": 9.410353894258102e-06, "loss": 0.8167, "step": 17095 }, { "epoch": 1.819110449031709, "grad_norm": 11.921875, "learning_rate": 9.410318362706084e-06, "loss": 1.1621, "step": 17096 }, { "epoch": 1.8192168546499254, "grad_norm": 12.3125, "learning_rate": 9.410282831154066e-06, "loss": 1.2061, "step": 17097 }, { "epoch": 1.8193232602681422, "grad_norm": 11.7265625, "learning_rate": 9.410247299602047e-06, "loss": 0.6709, "step": 17098 }, { "epoch": 1.819429665886359, "grad_norm": 18.78125, "learning_rate": 9.410211768050029e-06, "loss": 1.783, "step": 17099 }, { "epoch": 1.8195360715045754, "grad_norm": 21.484375, "learning_rate": 9.41017623649801e-06, "loss": 1.3536, "step": 17100 }, { "epoch": 1.819642477122792, "grad_norm": 13.9375, "learning_rate": 9.410140704945992e-06, "loss": 1.1468, "step": 17101 }, { "epoch": 1.8197488827410089, "grad_norm": 13.6796875, "learning_rate": 9.410105173393974e-06, "loss": 1.2068, "step": 17102 }, { "epoch": 1.8198552883592254, "grad_norm": 13.0078125, "learning_rate": 9.410069641841957e-06, "loss": 1.2813, "step": 17103 }, { "epoch": 1.819961693977442, "grad_norm": 13.28125, "learning_rate": 9.410034110289939e-06, "loss": 1.0709, "step": 17104 }, { "epoch": 1.8200680995956586, "grad_norm": 13.8515625, "learning_rate": 9.40999857873792e-06, "loss": 0.8649, "step": 17105 }, { "epoch": 1.8201745052138754, "grad_norm": 10.328125, "learning_rate": 9.409963047185902e-06, "loss": 1.0122, "step": 17106 }, { "epoch": 1.8202809108320919, "grad_norm": 13.796875, "learning_rate": 9.409927515633883e-06, "loss": 1.0461, "step": 17107 }, { "epoch": 1.8203873164503086, "grad_norm": 10.078125, "learning_rate": 9.409891984081867e-06, "loss": 1.0998, "step": 17108 }, { "epoch": 1.8204937220685253, "grad_norm": 11.6015625, "learning_rate": 9.409856452529848e-06, "loss": 1.0938, "step": 17109 }, { "epoch": 1.8206001276867418, "grad_norm": 16.53125, "learning_rate": 9.40982092097783e-06, "loss": 0.9903, "step": 17110 }, { "epoch": 1.8207065333049584, "grad_norm": 11.046875, "learning_rate": 9.409785389425811e-06, "loss": 1.0991, "step": 17111 }, { "epoch": 1.8208129389231753, "grad_norm": 11.8515625, "learning_rate": 9.409749857873793e-06, "loss": 1.0843, "step": 17112 }, { "epoch": 1.8209193445413918, "grad_norm": 14.3046875, "learning_rate": 9.409714326321775e-06, "loss": 1.308, "step": 17113 }, { "epoch": 1.8210257501596083, "grad_norm": 9.8828125, "learning_rate": 9.409678794769756e-06, "loss": 0.9224, "step": 17114 }, { "epoch": 1.821132155777825, "grad_norm": 9.453125, "learning_rate": 9.409643263217738e-06, "loss": 1.1387, "step": 17115 }, { "epoch": 1.8212385613960418, "grad_norm": 12.2734375, "learning_rate": 9.40960773166572e-06, "loss": 0.9547, "step": 17116 }, { "epoch": 1.8213449670142583, "grad_norm": 16.390625, "learning_rate": 9.409572200113701e-06, "loss": 1.2277, "step": 17117 }, { "epoch": 1.821451372632475, "grad_norm": 15.90625, "learning_rate": 9.409536668561684e-06, "loss": 1.0574, "step": 17118 }, { "epoch": 1.8215577782506918, "grad_norm": 13.25, "learning_rate": 9.409501137009666e-06, "loss": 1.0898, "step": 17119 }, { "epoch": 1.8216641838689083, "grad_norm": 11.3046875, "learning_rate": 9.409465605457648e-06, "loss": 1.1725, "step": 17120 }, { "epoch": 1.8217705894871248, "grad_norm": 13.7265625, "learning_rate": 9.40943007390563e-06, "loss": 1.0052, "step": 17121 }, { "epoch": 1.8218769951053415, "grad_norm": 11.2421875, "learning_rate": 9.40939454235361e-06, "loss": 1.1964, "step": 17122 }, { "epoch": 1.8219834007235582, "grad_norm": 9.5078125, "learning_rate": 9.409359010801592e-06, "loss": 1.0204, "step": 17123 }, { "epoch": 1.8220898063417748, "grad_norm": 12.015625, "learning_rate": 9.409323479249574e-06, "loss": 1.0807, "step": 17124 }, { "epoch": 1.8221962119599915, "grad_norm": 15.1171875, "learning_rate": 9.409287947697556e-06, "loss": 1.2065, "step": 17125 }, { "epoch": 1.8223026175782082, "grad_norm": 12.6796875, "learning_rate": 9.409252416145537e-06, "loss": 0.8851, "step": 17126 }, { "epoch": 1.8224090231964247, "grad_norm": 11.5859375, "learning_rate": 9.409216884593519e-06, "loss": 1.3134, "step": 17127 }, { "epoch": 1.8225154288146415, "grad_norm": 14.84375, "learning_rate": 9.409181353041502e-06, "loss": 0.8684, "step": 17128 }, { "epoch": 1.8226218344328582, "grad_norm": 11.859375, "learning_rate": 9.409145821489484e-06, "loss": 1.1681, "step": 17129 }, { "epoch": 1.8227282400510747, "grad_norm": 12.5078125, "learning_rate": 9.409110289937465e-06, "loss": 1.1721, "step": 17130 }, { "epoch": 1.8228346456692912, "grad_norm": 12.5703125, "learning_rate": 9.409074758385447e-06, "loss": 1.0663, "step": 17131 }, { "epoch": 1.822941051287508, "grad_norm": 11.21875, "learning_rate": 9.409039226833429e-06, "loss": 1.161, "step": 17132 }, { "epoch": 1.8230474569057247, "grad_norm": 14.40625, "learning_rate": 9.40900369528141e-06, "loss": 1.1267, "step": 17133 }, { "epoch": 1.8231538625239412, "grad_norm": 11.203125, "learning_rate": 9.408968163729393e-06, "loss": 1.1329, "step": 17134 }, { "epoch": 1.823260268142158, "grad_norm": 9.6640625, "learning_rate": 9.408932632177375e-06, "loss": 1.033, "step": 17135 }, { "epoch": 1.8233666737603746, "grad_norm": 15.140625, "learning_rate": 9.408897100625357e-06, "loss": 1.0726, "step": 17136 }, { "epoch": 1.8234730793785912, "grad_norm": 16.3125, "learning_rate": 9.408861569073338e-06, "loss": 0.9889, "step": 17137 }, { "epoch": 1.8235794849968079, "grad_norm": 13.4140625, "learning_rate": 9.40882603752132e-06, "loss": 1.2278, "step": 17138 }, { "epoch": 1.8236858906150246, "grad_norm": 13.375, "learning_rate": 9.408790505969302e-06, "loss": 0.9697, "step": 17139 }, { "epoch": 1.8237922962332411, "grad_norm": 15.2421875, "learning_rate": 9.408754974417283e-06, "loss": 0.8506, "step": 17140 }, { "epoch": 1.8238987018514576, "grad_norm": 15.4140625, "learning_rate": 9.408719442865265e-06, "loss": 1.179, "step": 17141 }, { "epoch": 1.8240051074696744, "grad_norm": 16.03125, "learning_rate": 9.408683911313246e-06, "loss": 1.0424, "step": 17142 }, { "epoch": 1.824111513087891, "grad_norm": 12.015625, "learning_rate": 9.408648379761228e-06, "loss": 1.0563, "step": 17143 }, { "epoch": 1.8242179187061076, "grad_norm": 17.359375, "learning_rate": 9.408612848209211e-06, "loss": 1.1728, "step": 17144 }, { "epoch": 1.8243243243243243, "grad_norm": 19.90625, "learning_rate": 9.408577316657193e-06, "loss": 1.4981, "step": 17145 }, { "epoch": 1.824430729942541, "grad_norm": 15.765625, "learning_rate": 9.408541785105174e-06, "loss": 0.9865, "step": 17146 }, { "epoch": 1.8245371355607576, "grad_norm": 13.2109375, "learning_rate": 9.408506253553156e-06, "loss": 1.0123, "step": 17147 }, { "epoch": 1.824643541178974, "grad_norm": 16.609375, "learning_rate": 9.408470722001138e-06, "loss": 1.1872, "step": 17148 }, { "epoch": 1.824749946797191, "grad_norm": 13.25, "learning_rate": 9.40843519044912e-06, "loss": 0.9907, "step": 17149 }, { "epoch": 1.8248563524154076, "grad_norm": 15.578125, "learning_rate": 9.408399658897101e-06, "loss": 0.9202, "step": 17150 }, { "epoch": 1.824962758033624, "grad_norm": 14.234375, "learning_rate": 9.408364127345082e-06, "loss": 1.0918, "step": 17151 }, { "epoch": 1.8250691636518408, "grad_norm": 15.0390625, "learning_rate": 9.408328595793064e-06, "loss": 1.1449, "step": 17152 }, { "epoch": 1.8251755692700575, "grad_norm": 11.8203125, "learning_rate": 9.408293064241046e-06, "loss": 0.9563, "step": 17153 }, { "epoch": 1.825281974888274, "grad_norm": 12.8515625, "learning_rate": 9.408257532689029e-06, "loss": 1.1016, "step": 17154 }, { "epoch": 1.8253883805064908, "grad_norm": 17.53125, "learning_rate": 9.40822200113701e-06, "loss": 1.2683, "step": 17155 }, { "epoch": 1.8254947861247075, "grad_norm": 15.265625, "learning_rate": 9.408186469584992e-06, "loss": 1.1545, "step": 17156 }, { "epoch": 1.825601191742924, "grad_norm": 17.40625, "learning_rate": 9.408150938032974e-06, "loss": 1.1192, "step": 17157 }, { "epoch": 1.8257075973611405, "grad_norm": 13.5703125, "learning_rate": 9.408115406480955e-06, "loss": 0.9881, "step": 17158 }, { "epoch": 1.8258140029793575, "grad_norm": 10.7578125, "learning_rate": 9.408079874928939e-06, "loss": 0.9932, "step": 17159 }, { "epoch": 1.825920408597574, "grad_norm": 17.296875, "learning_rate": 9.40804434337692e-06, "loss": 1.1306, "step": 17160 }, { "epoch": 1.8260268142157905, "grad_norm": 10.7265625, "learning_rate": 9.408008811824902e-06, "loss": 1.275, "step": 17161 }, { "epoch": 1.8261332198340072, "grad_norm": 9.3671875, "learning_rate": 9.407973280272884e-06, "loss": 1.0627, "step": 17162 }, { "epoch": 1.826239625452224, "grad_norm": 14.421875, "learning_rate": 9.407937748720865e-06, "loss": 0.9864, "step": 17163 }, { "epoch": 1.8263460310704405, "grad_norm": 9.34375, "learning_rate": 9.407902217168847e-06, "loss": 0.8915, "step": 17164 }, { "epoch": 1.8264524366886572, "grad_norm": 12.1328125, "learning_rate": 9.407866685616828e-06, "loss": 1.2494, "step": 17165 }, { "epoch": 1.826558842306874, "grad_norm": 17.359375, "learning_rate": 9.40783115406481e-06, "loss": 0.671, "step": 17166 }, { "epoch": 1.8266652479250904, "grad_norm": 10.5, "learning_rate": 9.407795622512792e-06, "loss": 0.9774, "step": 17167 }, { "epoch": 1.826771653543307, "grad_norm": 9.8828125, "learning_rate": 9.407760090960773e-06, "loss": 0.9, "step": 17168 }, { "epoch": 1.8268780591615237, "grad_norm": 9.6875, "learning_rate": 9.407724559408756e-06, "loss": 1.1466, "step": 17169 }, { "epoch": 1.8269844647797404, "grad_norm": 11.390625, "learning_rate": 9.407689027856738e-06, "loss": 1.0503, "step": 17170 }, { "epoch": 1.827090870397957, "grad_norm": 13.1640625, "learning_rate": 9.40765349630472e-06, "loss": 0.8341, "step": 17171 }, { "epoch": 1.8271972760161737, "grad_norm": 20.390625, "learning_rate": 9.407617964752701e-06, "loss": 1.1669, "step": 17172 }, { "epoch": 1.8273036816343904, "grad_norm": 12.734375, "learning_rate": 9.407582433200683e-06, "loss": 0.9958, "step": 17173 }, { "epoch": 1.827410087252607, "grad_norm": 16.171875, "learning_rate": 9.407546901648665e-06, "loss": 0.7892, "step": 17174 }, { "epoch": 1.8275164928708236, "grad_norm": 10.09375, "learning_rate": 9.407511370096646e-06, "loss": 1.1451, "step": 17175 }, { "epoch": 1.8276228984890404, "grad_norm": 13.8046875, "learning_rate": 9.407475838544628e-06, "loss": 1.0261, "step": 17176 }, { "epoch": 1.8277293041072569, "grad_norm": 15.140625, "learning_rate": 9.40744030699261e-06, "loss": 0.8548, "step": 17177 }, { "epoch": 1.8278357097254734, "grad_norm": 14.40625, "learning_rate": 9.407404775440591e-06, "loss": 1.1893, "step": 17178 }, { "epoch": 1.82794211534369, "grad_norm": 15.0625, "learning_rate": 9.407369243888573e-06, "loss": 0.7834, "step": 17179 }, { "epoch": 1.8280485209619068, "grad_norm": 14.28125, "learning_rate": 9.407333712336556e-06, "loss": 1.0768, "step": 17180 }, { "epoch": 1.8281549265801234, "grad_norm": 14.9453125, "learning_rate": 9.407298180784537e-06, "loss": 1.0034, "step": 17181 }, { "epoch": 1.82826133219834, "grad_norm": 11.65625, "learning_rate": 9.407262649232519e-06, "loss": 1.2341, "step": 17182 }, { "epoch": 1.8283677378165568, "grad_norm": 25.890625, "learning_rate": 9.4072271176805e-06, "loss": 1.1209, "step": 17183 }, { "epoch": 1.8284741434347733, "grad_norm": 11.296875, "learning_rate": 9.407191586128482e-06, "loss": 0.9939, "step": 17184 }, { "epoch": 1.8285805490529898, "grad_norm": 10.703125, "learning_rate": 9.407156054576466e-06, "loss": 1.1391, "step": 17185 }, { "epoch": 1.8286869546712068, "grad_norm": 11.0078125, "learning_rate": 9.407120523024447e-06, "loss": 1.0427, "step": 17186 }, { "epoch": 1.8287933602894233, "grad_norm": 13.6796875, "learning_rate": 9.407084991472429e-06, "loss": 1.1012, "step": 17187 }, { "epoch": 1.8288997659076398, "grad_norm": 14.9921875, "learning_rate": 9.40704945992041e-06, "loss": 1.2607, "step": 17188 }, { "epoch": 1.8290061715258565, "grad_norm": 14.1953125, "learning_rate": 9.407013928368392e-06, "loss": 1.323, "step": 17189 }, { "epoch": 1.8291125771440733, "grad_norm": 8.78125, "learning_rate": 9.406978396816374e-06, "loss": 0.9379, "step": 17190 }, { "epoch": 1.8292189827622898, "grad_norm": 9.09375, "learning_rate": 9.406942865264355e-06, "loss": 0.9206, "step": 17191 }, { "epoch": 1.8293253883805065, "grad_norm": 13.3359375, "learning_rate": 9.406907333712337e-06, "loss": 1.1846, "step": 17192 }, { "epoch": 1.8294317939987232, "grad_norm": 10.03125, "learning_rate": 9.406871802160318e-06, "loss": 0.9999, "step": 17193 }, { "epoch": 1.8295381996169398, "grad_norm": 12.6953125, "learning_rate": 9.4068362706083e-06, "loss": 1.2495, "step": 17194 }, { "epoch": 1.8296446052351563, "grad_norm": 11.1328125, "learning_rate": 9.406800739056283e-06, "loss": 1.1417, "step": 17195 }, { "epoch": 1.8297510108533732, "grad_norm": 15.953125, "learning_rate": 9.406765207504265e-06, "loss": 0.8941, "step": 17196 }, { "epoch": 1.8298574164715897, "grad_norm": 11.84375, "learning_rate": 9.406729675952247e-06, "loss": 1.0917, "step": 17197 }, { "epoch": 1.8299638220898062, "grad_norm": 10.734375, "learning_rate": 9.406694144400228e-06, "loss": 1.1661, "step": 17198 }, { "epoch": 1.830070227708023, "grad_norm": 9.40625, "learning_rate": 9.40665861284821e-06, "loss": 1.2521, "step": 17199 }, { "epoch": 1.8301766333262397, "grad_norm": 8.5390625, "learning_rate": 9.406623081296193e-06, "loss": 0.8202, "step": 17200 }, { "epoch": 1.8302830389444562, "grad_norm": 12.2109375, "learning_rate": 9.406587549744175e-06, "loss": 1.2158, "step": 17201 }, { "epoch": 1.830389444562673, "grad_norm": 13.4765625, "learning_rate": 9.406552018192156e-06, "loss": 0.9461, "step": 17202 }, { "epoch": 1.8304958501808897, "grad_norm": 22.390625, "learning_rate": 9.406516486640138e-06, "loss": 0.9684, "step": 17203 }, { "epoch": 1.8306022557991062, "grad_norm": 11.9375, "learning_rate": 9.40648095508812e-06, "loss": 1.0243, "step": 17204 }, { "epoch": 1.8307086614173227, "grad_norm": 12.1328125, "learning_rate": 9.406445423536101e-06, "loss": 0.777, "step": 17205 }, { "epoch": 1.8308150670355394, "grad_norm": 14.4921875, "learning_rate": 9.406409891984083e-06, "loss": 0.9877, "step": 17206 }, { "epoch": 1.8309214726537562, "grad_norm": 12.828125, "learning_rate": 9.406374360432064e-06, "loss": 1.0612, "step": 17207 }, { "epoch": 1.8310278782719727, "grad_norm": 12.28125, "learning_rate": 9.406338828880046e-06, "loss": 0.9929, "step": 17208 }, { "epoch": 1.8311342838901894, "grad_norm": 10.2109375, "learning_rate": 9.406303297328028e-06, "loss": 1.1184, "step": 17209 }, { "epoch": 1.8312406895084061, "grad_norm": 11.125, "learning_rate": 9.40626776577601e-06, "loss": 0.9561, "step": 17210 }, { "epoch": 1.8313470951266226, "grad_norm": 10.2109375, "learning_rate": 9.406232234223992e-06, "loss": 1.1294, "step": 17211 }, { "epoch": 1.8314535007448394, "grad_norm": 12.59375, "learning_rate": 9.406196702671974e-06, "loss": 0.9768, "step": 17212 }, { "epoch": 1.831559906363056, "grad_norm": 11.40625, "learning_rate": 9.406161171119956e-06, "loss": 0.906, "step": 17213 }, { "epoch": 1.8316663119812726, "grad_norm": 12.1640625, "learning_rate": 9.406125639567937e-06, "loss": 0.9217, "step": 17214 }, { "epoch": 1.8317727175994891, "grad_norm": 13.5703125, "learning_rate": 9.406090108015919e-06, "loss": 1.0588, "step": 17215 }, { "epoch": 1.8318791232177059, "grad_norm": 14.1015625, "learning_rate": 9.4060545764639e-06, "loss": 1.2136, "step": 17216 }, { "epoch": 1.8319855288359226, "grad_norm": 12.75, "learning_rate": 9.406019044911882e-06, "loss": 1.0025, "step": 17217 }, { "epoch": 1.832091934454139, "grad_norm": 14.234375, "learning_rate": 9.405983513359864e-06, "loss": 1.1733, "step": 17218 }, { "epoch": 1.8321983400723558, "grad_norm": 14.125, "learning_rate": 9.405947981807845e-06, "loss": 1.1762, "step": 17219 }, { "epoch": 1.8323047456905726, "grad_norm": 16.5, "learning_rate": 9.405912450255827e-06, "loss": 1.0186, "step": 17220 }, { "epoch": 1.832411151308789, "grad_norm": 13.6796875, "learning_rate": 9.40587691870381e-06, "loss": 1.2155, "step": 17221 }, { "epoch": 1.8325175569270058, "grad_norm": 11.1640625, "learning_rate": 9.405841387151792e-06, "loss": 1.0398, "step": 17222 }, { "epoch": 1.8326239625452225, "grad_norm": 15.3984375, "learning_rate": 9.405805855599773e-06, "loss": 0.98, "step": 17223 }, { "epoch": 1.832730368163439, "grad_norm": 14.34375, "learning_rate": 9.405770324047755e-06, "loss": 1.2549, "step": 17224 }, { "epoch": 1.8328367737816555, "grad_norm": 23.53125, "learning_rate": 9.405734792495737e-06, "loss": 0.988, "step": 17225 }, { "epoch": 1.8329431793998723, "grad_norm": 18.75, "learning_rate": 9.40569926094372e-06, "loss": 1.03, "step": 17226 }, { "epoch": 1.833049585018089, "grad_norm": 16.09375, "learning_rate": 9.405663729391702e-06, "loss": 1.1017, "step": 17227 }, { "epoch": 1.8331559906363055, "grad_norm": 12.59375, "learning_rate": 9.405628197839683e-06, "loss": 1.0803, "step": 17228 }, { "epoch": 1.8332623962545223, "grad_norm": 14.8046875, "learning_rate": 9.405592666287665e-06, "loss": 1.0946, "step": 17229 }, { "epoch": 1.833368801872739, "grad_norm": 12.9765625, "learning_rate": 9.405557134735646e-06, "loss": 1.3283, "step": 17230 }, { "epoch": 1.8334752074909555, "grad_norm": 9.65625, "learning_rate": 9.405521603183628e-06, "loss": 0.844, "step": 17231 }, { "epoch": 1.833581613109172, "grad_norm": 22.375, "learning_rate": 9.40548607163161e-06, "loss": 1.3141, "step": 17232 }, { "epoch": 1.833688018727389, "grad_norm": 12.09375, "learning_rate": 9.405450540079591e-06, "loss": 1.0999, "step": 17233 }, { "epoch": 1.8337944243456055, "grad_norm": 20.640625, "learning_rate": 9.405415008527573e-06, "loss": 1.2834, "step": 17234 }, { "epoch": 1.833900829963822, "grad_norm": 16.484375, "learning_rate": 9.405379476975554e-06, "loss": 1.0511, "step": 17235 }, { "epoch": 1.8340072355820387, "grad_norm": 17.34375, "learning_rate": 9.405343945423538e-06, "loss": 0.9232, "step": 17236 }, { "epoch": 1.8341136412002554, "grad_norm": 12.0078125, "learning_rate": 9.40530841387152e-06, "loss": 0.7784, "step": 17237 }, { "epoch": 1.834220046818472, "grad_norm": 11.46875, "learning_rate": 9.405272882319501e-06, "loss": 1.0753, "step": 17238 }, { "epoch": 1.8343264524366887, "grad_norm": 15.71875, "learning_rate": 9.405237350767483e-06, "loss": 1.3623, "step": 17239 }, { "epoch": 1.8344328580549054, "grad_norm": 14.4140625, "learning_rate": 9.405201819215464e-06, "loss": 1.3009, "step": 17240 }, { "epoch": 1.834539263673122, "grad_norm": 10.7265625, "learning_rate": 9.405166287663446e-06, "loss": 1.1173, "step": 17241 }, { "epoch": 1.8346456692913384, "grad_norm": 14.375, "learning_rate": 9.405130756111427e-06, "loss": 1.1932, "step": 17242 }, { "epoch": 1.8347520749095554, "grad_norm": 11.6875, "learning_rate": 9.405095224559409e-06, "loss": 1.164, "step": 17243 }, { "epoch": 1.834858480527772, "grad_norm": 11.03125, "learning_rate": 9.40505969300739e-06, "loss": 1.3317, "step": 17244 }, { "epoch": 1.8349648861459884, "grad_norm": 12.546875, "learning_rate": 9.405024161455372e-06, "loss": 0.851, "step": 17245 }, { "epoch": 1.8350712917642051, "grad_norm": 15.125, "learning_rate": 9.404988629903355e-06, "loss": 1.254, "step": 17246 }, { "epoch": 1.8351776973824219, "grad_norm": 11.6015625, "learning_rate": 9.404953098351337e-06, "loss": 1.1102, "step": 17247 }, { "epoch": 1.8352841030006384, "grad_norm": 17.171875, "learning_rate": 9.404917566799319e-06, "loss": 1.1004, "step": 17248 }, { "epoch": 1.835390508618855, "grad_norm": 12.4140625, "learning_rate": 9.4048820352473e-06, "loss": 1.2661, "step": 17249 }, { "epoch": 1.8354969142370718, "grad_norm": 14.375, "learning_rate": 9.404846503695282e-06, "loss": 0.9307, "step": 17250 }, { "epoch": 1.8356033198552884, "grad_norm": 14.3828125, "learning_rate": 9.404810972143263e-06, "loss": 1.3267, "step": 17251 }, { "epoch": 1.8357097254735049, "grad_norm": 10.84375, "learning_rate": 9.404775440591247e-06, "loss": 1.0322, "step": 17252 }, { "epoch": 1.8358161310917216, "grad_norm": 12.28125, "learning_rate": 9.404739909039228e-06, "loss": 0.834, "step": 17253 }, { "epoch": 1.8359225367099383, "grad_norm": 8.8125, "learning_rate": 9.40470437748721e-06, "loss": 1.1809, "step": 17254 }, { "epoch": 1.8360289423281548, "grad_norm": 12.296875, "learning_rate": 9.404668845935192e-06, "loss": 1.0599, "step": 17255 }, { "epoch": 1.8361353479463716, "grad_norm": 8.1953125, "learning_rate": 9.404633314383173e-06, "loss": 1.0374, "step": 17256 }, { "epoch": 1.8362417535645883, "grad_norm": 7.75390625, "learning_rate": 9.404597782831155e-06, "loss": 0.8543, "step": 17257 }, { "epoch": 1.8363481591828048, "grad_norm": 10.6484375, "learning_rate": 9.404562251279136e-06, "loss": 0.9441, "step": 17258 }, { "epoch": 1.8364545648010215, "grad_norm": 17.984375, "learning_rate": 9.404526719727118e-06, "loss": 1.2966, "step": 17259 }, { "epoch": 1.8365609704192383, "grad_norm": 14.484375, "learning_rate": 9.4044911881751e-06, "loss": 1.3555, "step": 17260 }, { "epoch": 1.8366673760374548, "grad_norm": 10.3515625, "learning_rate": 9.404455656623081e-06, "loss": 1.2856, "step": 17261 }, { "epoch": 1.8367737816556713, "grad_norm": 12.34375, "learning_rate": 9.404420125071065e-06, "loss": 0.9104, "step": 17262 }, { "epoch": 1.836880187273888, "grad_norm": 12.765625, "learning_rate": 9.404384593519046e-06, "loss": 1.0714, "step": 17263 }, { "epoch": 1.8369865928921048, "grad_norm": 12.9609375, "learning_rate": 9.404349061967028e-06, "loss": 1.3818, "step": 17264 }, { "epoch": 1.8370929985103213, "grad_norm": 12.1953125, "learning_rate": 9.40431353041501e-06, "loss": 1.3254, "step": 17265 }, { "epoch": 1.837199404128538, "grad_norm": 22.09375, "learning_rate": 9.404277998862991e-06, "loss": 1.1546, "step": 17266 }, { "epoch": 1.8373058097467547, "grad_norm": 10.6171875, "learning_rate": 9.404242467310973e-06, "loss": 1.0352, "step": 17267 }, { "epoch": 1.8374122153649712, "grad_norm": 11.9296875, "learning_rate": 9.404206935758954e-06, "loss": 1.0367, "step": 17268 }, { "epoch": 1.837518620983188, "grad_norm": 17.0625, "learning_rate": 9.404171404206936e-06, "loss": 0.9911, "step": 17269 }, { "epoch": 1.8376250266014047, "grad_norm": 12.046875, "learning_rate": 9.404135872654917e-06, "loss": 1.1047, "step": 17270 }, { "epoch": 1.8377314322196212, "grad_norm": 17.234375, "learning_rate": 9.404100341102899e-06, "loss": 1.1554, "step": 17271 }, { "epoch": 1.8378378378378377, "grad_norm": 10.6796875, "learning_rate": 9.404064809550882e-06, "loss": 0.9625, "step": 17272 }, { "epoch": 1.8379442434560544, "grad_norm": 10.875, "learning_rate": 9.404029277998864e-06, "loss": 1.1819, "step": 17273 }, { "epoch": 1.8380506490742712, "grad_norm": 11.7578125, "learning_rate": 9.403993746446846e-06, "loss": 1.2091, "step": 17274 }, { "epoch": 1.8381570546924877, "grad_norm": 14.84375, "learning_rate": 9.403958214894827e-06, "loss": 1.1968, "step": 17275 }, { "epoch": 1.8382634603107044, "grad_norm": 14.59375, "learning_rate": 9.403922683342809e-06, "loss": 1.489, "step": 17276 }, { "epoch": 1.8383698659289212, "grad_norm": 14.7265625, "learning_rate": 9.403887151790792e-06, "loss": 1.1322, "step": 17277 }, { "epoch": 1.8384762715471377, "grad_norm": 11.25, "learning_rate": 9.403851620238774e-06, "loss": 1.1697, "step": 17278 }, { "epoch": 1.8385826771653542, "grad_norm": 12.125, "learning_rate": 9.403816088686755e-06, "loss": 1.212, "step": 17279 }, { "epoch": 1.8386890827835711, "grad_norm": 13.40625, "learning_rate": 9.403780557134737e-06, "loss": 1.1301, "step": 17280 }, { "epoch": 1.8387954884017876, "grad_norm": 12.0625, "learning_rate": 9.403745025582718e-06, "loss": 1.1873, "step": 17281 }, { "epoch": 1.8389018940200041, "grad_norm": 8.921875, "learning_rate": 9.4037094940307e-06, "loss": 0.9644, "step": 17282 }, { "epoch": 1.8390082996382209, "grad_norm": 10.8203125, "learning_rate": 9.403673962478682e-06, "loss": 0.9656, "step": 17283 }, { "epoch": 1.8391147052564376, "grad_norm": 15.9140625, "learning_rate": 9.403638430926663e-06, "loss": 1.0564, "step": 17284 }, { "epoch": 1.8392211108746541, "grad_norm": 10.578125, "learning_rate": 9.403602899374645e-06, "loss": 1.1319, "step": 17285 }, { "epoch": 1.8393275164928709, "grad_norm": 11.1875, "learning_rate": 9.403567367822626e-06, "loss": 0.9825, "step": 17286 }, { "epoch": 1.8394339221110876, "grad_norm": 12.40625, "learning_rate": 9.40353183627061e-06, "loss": 1.0305, "step": 17287 }, { "epoch": 1.839540327729304, "grad_norm": 10.1953125, "learning_rate": 9.403496304718591e-06, "loss": 1.0022, "step": 17288 }, { "epoch": 1.8396467333475206, "grad_norm": 13.3203125, "learning_rate": 9.403460773166573e-06, "loss": 1.1715, "step": 17289 }, { "epoch": 1.8397531389657376, "grad_norm": 11.0546875, "learning_rate": 9.403425241614555e-06, "loss": 0.9835, "step": 17290 }, { "epoch": 1.839859544583954, "grad_norm": 14.6640625, "learning_rate": 9.403389710062536e-06, "loss": 0.8339, "step": 17291 }, { "epoch": 1.8399659502021706, "grad_norm": 13.2734375, "learning_rate": 9.403354178510518e-06, "loss": 0.9634, "step": 17292 }, { "epoch": 1.8400723558203873, "grad_norm": 12.9140625, "learning_rate": 9.4033186469585e-06, "loss": 0.9918, "step": 17293 }, { "epoch": 1.840178761438604, "grad_norm": 12.2734375, "learning_rate": 9.403283115406481e-06, "loss": 0.8365, "step": 17294 }, { "epoch": 1.8402851670568205, "grad_norm": 19.125, "learning_rate": 9.403247583854463e-06, "loss": 1.0245, "step": 17295 }, { "epoch": 1.8403915726750373, "grad_norm": 16.46875, "learning_rate": 9.403212052302444e-06, "loss": 1.0725, "step": 17296 }, { "epoch": 1.840497978293254, "grad_norm": 14.6640625, "learning_rate": 9.403176520750426e-06, "loss": 1.028, "step": 17297 }, { "epoch": 1.8406043839114705, "grad_norm": 14.3828125, "learning_rate": 9.40314098919841e-06, "loss": 0.9976, "step": 17298 }, { "epoch": 1.840710789529687, "grad_norm": 11.328125, "learning_rate": 9.40310545764639e-06, "loss": 1.0227, "step": 17299 }, { "epoch": 1.8408171951479038, "grad_norm": 13.46875, "learning_rate": 9.403069926094372e-06, "loss": 1.2463, "step": 17300 }, { "epoch": 1.8409236007661205, "grad_norm": 13.9765625, "learning_rate": 9.403034394542354e-06, "loss": 1.4752, "step": 17301 }, { "epoch": 1.841030006384337, "grad_norm": 12.234375, "learning_rate": 9.402998862990336e-06, "loss": 1.2555, "step": 17302 }, { "epoch": 1.8411364120025537, "grad_norm": 15.3359375, "learning_rate": 9.402963331438319e-06, "loss": 1.4164, "step": 17303 }, { "epoch": 1.8412428176207705, "grad_norm": 11.875, "learning_rate": 9.4029277998863e-06, "loss": 1.2822, "step": 17304 }, { "epoch": 1.841349223238987, "grad_norm": 14.109375, "learning_rate": 9.402892268334282e-06, "loss": 1.0894, "step": 17305 }, { "epoch": 1.8414556288572037, "grad_norm": 13.640625, "learning_rate": 9.402856736782264e-06, "loss": 0.9937, "step": 17306 }, { "epoch": 1.8415620344754204, "grad_norm": 12.8125, "learning_rate": 9.402821205230245e-06, "loss": 1.0885, "step": 17307 }, { "epoch": 1.841668440093637, "grad_norm": 15.8515625, "learning_rate": 9.402785673678227e-06, "loss": 0.9264, "step": 17308 }, { "epoch": 1.8417748457118535, "grad_norm": 14.3046875, "learning_rate": 9.402750142126209e-06, "loss": 0.802, "step": 17309 }, { "epoch": 1.8418812513300702, "grad_norm": 13.8828125, "learning_rate": 9.40271461057419e-06, "loss": 0.9477, "step": 17310 }, { "epoch": 1.841987656948287, "grad_norm": 14.140625, "learning_rate": 9.402679079022172e-06, "loss": 0.8709, "step": 17311 }, { "epoch": 1.8420940625665034, "grad_norm": 13.28125, "learning_rate": 9.402643547470153e-06, "loss": 1.2518, "step": 17312 }, { "epoch": 1.8422004681847202, "grad_norm": 10.0234375, "learning_rate": 9.402608015918137e-06, "loss": 1.0706, "step": 17313 }, { "epoch": 1.842306873802937, "grad_norm": 10.703125, "learning_rate": 9.402572484366118e-06, "loss": 1.2888, "step": 17314 }, { "epoch": 1.8424132794211534, "grad_norm": 21.953125, "learning_rate": 9.4025369528141e-06, "loss": 0.9586, "step": 17315 }, { "epoch": 1.84251968503937, "grad_norm": 12.5390625, "learning_rate": 9.402501421262081e-06, "loss": 1.1292, "step": 17316 }, { "epoch": 1.8426260906575869, "grad_norm": 18.34375, "learning_rate": 9.402465889710063e-06, "loss": 1.0695, "step": 17317 }, { "epoch": 1.8427324962758034, "grad_norm": 13.1875, "learning_rate": 9.402430358158046e-06, "loss": 1.0403, "step": 17318 }, { "epoch": 1.8428389018940199, "grad_norm": 12.046875, "learning_rate": 9.402394826606028e-06, "loss": 0.9453, "step": 17319 }, { "epoch": 1.8429453075122366, "grad_norm": 10.5390625, "learning_rate": 9.40235929505401e-06, "loss": 0.9069, "step": 17320 }, { "epoch": 1.8430517131304534, "grad_norm": 9.3203125, "learning_rate": 9.402323763501991e-06, "loss": 1.1036, "step": 17321 }, { "epoch": 1.8431581187486699, "grad_norm": 14.421875, "learning_rate": 9.402288231949973e-06, "loss": 1.1581, "step": 17322 }, { "epoch": 1.8432645243668866, "grad_norm": 11.6015625, "learning_rate": 9.402252700397954e-06, "loss": 1.2619, "step": 17323 }, { "epoch": 1.8433709299851033, "grad_norm": 16.09375, "learning_rate": 9.402217168845936e-06, "loss": 1.3979, "step": 17324 }, { "epoch": 1.8434773356033198, "grad_norm": 16.234375, "learning_rate": 9.402181637293918e-06, "loss": 0.9801, "step": 17325 }, { "epoch": 1.8435837412215363, "grad_norm": 13.0, "learning_rate": 9.4021461057419e-06, "loss": 0.8681, "step": 17326 }, { "epoch": 1.8436901468397533, "grad_norm": 12.328125, "learning_rate": 9.40211057418988e-06, "loss": 1.0024, "step": 17327 }, { "epoch": 1.8437965524579698, "grad_norm": 12.859375, "learning_rate": 9.402075042637862e-06, "loss": 1.125, "step": 17328 }, { "epoch": 1.8439029580761863, "grad_norm": 16.71875, "learning_rate": 9.402039511085846e-06, "loss": 1.6783, "step": 17329 }, { "epoch": 1.844009363694403, "grad_norm": 10.25, "learning_rate": 9.402003979533827e-06, "loss": 0.9544, "step": 17330 }, { "epoch": 1.8441157693126198, "grad_norm": 8.984375, "learning_rate": 9.401968447981809e-06, "loss": 1.0213, "step": 17331 }, { "epoch": 1.8442221749308363, "grad_norm": 13.0859375, "learning_rate": 9.40193291642979e-06, "loss": 1.0008, "step": 17332 }, { "epoch": 1.844328580549053, "grad_norm": 15.0390625, "learning_rate": 9.401897384877772e-06, "loss": 0.8872, "step": 17333 }, { "epoch": 1.8444349861672698, "grad_norm": 13.1953125, "learning_rate": 9.401861853325754e-06, "loss": 0.9877, "step": 17334 }, { "epoch": 1.8445413917854863, "grad_norm": 12.4375, "learning_rate": 9.401826321773735e-06, "loss": 1.1525, "step": 17335 }, { "epoch": 1.8446477974037028, "grad_norm": 11.6328125, "learning_rate": 9.401790790221717e-06, "loss": 1.0624, "step": 17336 }, { "epoch": 1.8447542030219195, "grad_norm": 13.4453125, "learning_rate": 9.401755258669699e-06, "loss": 1.6717, "step": 17337 }, { "epoch": 1.8448606086401362, "grad_norm": 13.3046875, "learning_rate": 9.40171972711768e-06, "loss": 1.033, "step": 17338 }, { "epoch": 1.8449670142583527, "grad_norm": 11.15625, "learning_rate": 9.401684195565663e-06, "loss": 1.0234, "step": 17339 }, { "epoch": 1.8450734198765695, "grad_norm": 16.859375, "learning_rate": 9.401648664013645e-06, "loss": 1.3612, "step": 17340 }, { "epoch": 1.8451798254947862, "grad_norm": 12.1640625, "learning_rate": 9.401613132461627e-06, "loss": 0.7332, "step": 17341 }, { "epoch": 1.8452862311130027, "grad_norm": 14.25, "learning_rate": 9.401577600909608e-06, "loss": 1.0683, "step": 17342 }, { "epoch": 1.8453926367312194, "grad_norm": 12.1171875, "learning_rate": 9.40154206935759e-06, "loss": 1.2815, "step": 17343 }, { "epoch": 1.8454990423494362, "grad_norm": 15.8828125, "learning_rate": 9.401506537805573e-06, "loss": 1.233, "step": 17344 }, { "epoch": 1.8456054479676527, "grad_norm": 12.859375, "learning_rate": 9.401471006253555e-06, "loss": 1.0981, "step": 17345 }, { "epoch": 1.8457118535858692, "grad_norm": 10.640625, "learning_rate": 9.401435474701536e-06, "loss": 0.9426, "step": 17346 }, { "epoch": 1.845818259204086, "grad_norm": 11.71875, "learning_rate": 9.401399943149518e-06, "loss": 1.4706, "step": 17347 }, { "epoch": 1.8459246648223027, "grad_norm": 9.34375, "learning_rate": 9.4013644115975e-06, "loss": 1.0167, "step": 17348 }, { "epoch": 1.8460310704405192, "grad_norm": 9.875, "learning_rate": 9.401328880045481e-06, "loss": 1.1176, "step": 17349 }, { "epoch": 1.846137476058736, "grad_norm": 12.125, "learning_rate": 9.401293348493463e-06, "loss": 0.9378, "step": 17350 }, { "epoch": 1.8462438816769526, "grad_norm": 11.1328125, "learning_rate": 9.401257816941444e-06, "loss": 1.1811, "step": 17351 }, { "epoch": 1.8463502872951691, "grad_norm": 10.0, "learning_rate": 9.401222285389426e-06, "loss": 1.1672, "step": 17352 }, { "epoch": 1.8464566929133859, "grad_norm": 10.203125, "learning_rate": 9.401186753837408e-06, "loss": 1.2212, "step": 17353 }, { "epoch": 1.8465630985316026, "grad_norm": 13.578125, "learning_rate": 9.401151222285391e-06, "loss": 1.1222, "step": 17354 }, { "epoch": 1.8466695041498191, "grad_norm": 18.375, "learning_rate": 9.401115690733373e-06, "loss": 1.2751, "step": 17355 }, { "epoch": 1.8467759097680356, "grad_norm": 12.4375, "learning_rate": 9.401080159181354e-06, "loss": 1.1888, "step": 17356 }, { "epoch": 1.8468823153862524, "grad_norm": 15.6640625, "learning_rate": 9.401044627629336e-06, "loss": 1.2522, "step": 17357 }, { "epoch": 1.846988721004469, "grad_norm": 11.4453125, "learning_rate": 9.401009096077317e-06, "loss": 0.9077, "step": 17358 }, { "epoch": 1.8470951266226856, "grad_norm": 21.8125, "learning_rate": 9.400973564525299e-06, "loss": 0.9089, "step": 17359 }, { "epoch": 1.8472015322409023, "grad_norm": 16.21875, "learning_rate": 9.40093803297328e-06, "loss": 1.3327, "step": 17360 }, { "epoch": 1.847307937859119, "grad_norm": 15.140625, "learning_rate": 9.400902501421262e-06, "loss": 1.3226, "step": 17361 }, { "epoch": 1.8474143434773356, "grad_norm": 13.390625, "learning_rate": 9.400866969869244e-06, "loss": 1.1484, "step": 17362 }, { "epoch": 1.847520749095552, "grad_norm": 11.8515625, "learning_rate": 9.400831438317225e-06, "loss": 1.032, "step": 17363 }, { "epoch": 1.847627154713769, "grad_norm": 13.34375, "learning_rate": 9.400795906765209e-06, "loss": 1.1828, "step": 17364 }, { "epoch": 1.8477335603319855, "grad_norm": 11.03125, "learning_rate": 9.40076037521319e-06, "loss": 0.9853, "step": 17365 }, { "epoch": 1.847839965950202, "grad_norm": 16.109375, "learning_rate": 9.400724843661172e-06, "loss": 1.0271, "step": 17366 }, { "epoch": 1.8479463715684188, "grad_norm": 13.96875, "learning_rate": 9.400689312109154e-06, "loss": 1.2141, "step": 17367 }, { "epoch": 1.8480527771866355, "grad_norm": 14.28125, "learning_rate": 9.400653780557135e-06, "loss": 1.1663, "step": 17368 }, { "epoch": 1.848159182804852, "grad_norm": 12.7734375, "learning_rate": 9.400618249005117e-06, "loss": 1.3584, "step": 17369 }, { "epoch": 1.8482655884230688, "grad_norm": 17.953125, "learning_rate": 9.4005827174531e-06, "loss": 1.2008, "step": 17370 }, { "epoch": 1.8483719940412855, "grad_norm": 11.953125, "learning_rate": 9.400547185901082e-06, "loss": 0.8882, "step": 17371 }, { "epoch": 1.848478399659502, "grad_norm": 11.359375, "learning_rate": 9.400511654349063e-06, "loss": 1.2072, "step": 17372 }, { "epoch": 1.8485848052777185, "grad_norm": 9.515625, "learning_rate": 9.400476122797045e-06, "loss": 0.8647, "step": 17373 }, { "epoch": 1.8486912108959355, "grad_norm": 10.5625, "learning_rate": 9.400440591245027e-06, "loss": 1.1985, "step": 17374 }, { "epoch": 1.848797616514152, "grad_norm": 9.421875, "learning_rate": 9.400405059693008e-06, "loss": 0.86, "step": 17375 }, { "epoch": 1.8489040221323685, "grad_norm": 12.2265625, "learning_rate": 9.40036952814099e-06, "loss": 1.045, "step": 17376 }, { "epoch": 1.8490104277505852, "grad_norm": 16.9375, "learning_rate": 9.400333996588971e-06, "loss": 1.0956, "step": 17377 }, { "epoch": 1.849116833368802, "grad_norm": 14.3515625, "learning_rate": 9.400298465036953e-06, "loss": 1.0123, "step": 17378 }, { "epoch": 1.8492232389870185, "grad_norm": 14.1640625, "learning_rate": 9.400262933484935e-06, "loss": 1.2262, "step": 17379 }, { "epoch": 1.8493296446052352, "grad_norm": 12.703125, "learning_rate": 9.400227401932918e-06, "loss": 1.0402, "step": 17380 }, { "epoch": 1.849436050223452, "grad_norm": 39.28125, "learning_rate": 9.4001918703809e-06, "loss": 1.041, "step": 17381 }, { "epoch": 1.8495424558416684, "grad_norm": 9.65625, "learning_rate": 9.400156338828881e-06, "loss": 1.1161, "step": 17382 }, { "epoch": 1.849648861459885, "grad_norm": 22.6875, "learning_rate": 9.400120807276863e-06, "loss": 0.9347, "step": 17383 }, { "epoch": 1.8497552670781017, "grad_norm": 14.3984375, "learning_rate": 9.400085275724844e-06, "loss": 0.9261, "step": 17384 }, { "epoch": 1.8498616726963184, "grad_norm": 10.7109375, "learning_rate": 9.400049744172826e-06, "loss": 0.9399, "step": 17385 }, { "epoch": 1.849968078314535, "grad_norm": 13.5078125, "learning_rate": 9.400014212620807e-06, "loss": 1.0894, "step": 17386 }, { "epoch": 1.8500744839327516, "grad_norm": 22.140625, "learning_rate": 9.399978681068789e-06, "loss": 0.9061, "step": 17387 }, { "epoch": 1.8501808895509684, "grad_norm": 11.46875, "learning_rate": 9.39994314951677e-06, "loss": 1.0416, "step": 17388 }, { "epoch": 1.8502872951691849, "grad_norm": 10.5, "learning_rate": 9.399907617964752e-06, "loss": 1.0945, "step": 17389 }, { "epoch": 1.8503937007874016, "grad_norm": 11.0859375, "learning_rate": 9.399872086412736e-06, "loss": 1.0544, "step": 17390 }, { "epoch": 1.8505001064056184, "grad_norm": 13.921875, "learning_rate": 9.399836554860717e-06, "loss": 1.2906, "step": 17391 }, { "epoch": 1.8506065120238349, "grad_norm": 11.2890625, "learning_rate": 9.399801023308699e-06, "loss": 0.9456, "step": 17392 }, { "epoch": 1.8507129176420514, "grad_norm": 14.8359375, "learning_rate": 9.39976549175668e-06, "loss": 1.1781, "step": 17393 }, { "epoch": 1.850819323260268, "grad_norm": 16.609375, "learning_rate": 9.399729960204662e-06, "loss": 0.9525, "step": 17394 }, { "epoch": 1.8509257288784848, "grad_norm": 14.1015625, "learning_rate": 9.399694428652645e-06, "loss": 0.9873, "step": 17395 }, { "epoch": 1.8510321344967013, "grad_norm": 21.84375, "learning_rate": 9.399658897100627e-06, "loss": 1.1956, "step": 17396 }, { "epoch": 1.851138540114918, "grad_norm": 15.6328125, "learning_rate": 9.399623365548609e-06, "loss": 1.3699, "step": 17397 }, { "epoch": 1.8512449457331348, "grad_norm": 9.6875, "learning_rate": 9.39958783399659e-06, "loss": 1.1108, "step": 17398 }, { "epoch": 1.8513513513513513, "grad_norm": 16.78125, "learning_rate": 9.399552302444572e-06, "loss": 0.9648, "step": 17399 }, { "epoch": 1.851457756969568, "grad_norm": 11.5078125, "learning_rate": 9.399516770892553e-06, "loss": 0.9056, "step": 17400 }, { "epoch": 1.8515641625877848, "grad_norm": 13.9140625, "learning_rate": 9.399481239340535e-06, "loss": 1.0818, "step": 17401 }, { "epoch": 1.8516705682060013, "grad_norm": 11.765625, "learning_rate": 9.399445707788517e-06, "loss": 0.8595, "step": 17402 }, { "epoch": 1.8517769738242178, "grad_norm": 13.328125, "learning_rate": 9.399410176236498e-06, "loss": 0.7987, "step": 17403 }, { "epoch": 1.8518833794424345, "grad_norm": 14.2265625, "learning_rate": 9.39937464468448e-06, "loss": 1.2014, "step": 17404 }, { "epoch": 1.8519897850606513, "grad_norm": 11.71875, "learning_rate": 9.399339113132463e-06, "loss": 0.9337, "step": 17405 }, { "epoch": 1.8520961906788678, "grad_norm": 12.3203125, "learning_rate": 9.399303581580445e-06, "loss": 1.0488, "step": 17406 }, { "epoch": 1.8522025962970845, "grad_norm": 16.984375, "learning_rate": 9.399268050028426e-06, "loss": 1.0768, "step": 17407 }, { "epoch": 1.8523090019153012, "grad_norm": 22.828125, "learning_rate": 9.399232518476408e-06, "loss": 1.1721, "step": 17408 }, { "epoch": 1.8524154075335177, "grad_norm": 16.984375, "learning_rate": 9.39919698692439e-06, "loss": 0.9459, "step": 17409 }, { "epoch": 1.8525218131517343, "grad_norm": 14.1640625, "learning_rate": 9.399161455372371e-06, "loss": 0.917, "step": 17410 }, { "epoch": 1.8526282187699512, "grad_norm": 15.5625, "learning_rate": 9.399125923820354e-06, "loss": 1.3187, "step": 17411 }, { "epoch": 1.8527346243881677, "grad_norm": 12.7109375, "learning_rate": 9.399090392268336e-06, "loss": 1.3174, "step": 17412 }, { "epoch": 1.8528410300063842, "grad_norm": 16.546875, "learning_rate": 9.399054860716316e-06, "loss": 1.0153, "step": 17413 }, { "epoch": 1.852947435624601, "grad_norm": 18.390625, "learning_rate": 9.399019329164298e-06, "loss": 1.0209, "step": 17414 }, { "epoch": 1.8530538412428177, "grad_norm": 14.4375, "learning_rate": 9.39898379761228e-06, "loss": 0.954, "step": 17415 }, { "epoch": 1.8531602468610342, "grad_norm": 9.203125, "learning_rate": 9.398948266060262e-06, "loss": 1.0364, "step": 17416 }, { "epoch": 1.853266652479251, "grad_norm": 14.7109375, "learning_rate": 9.398912734508244e-06, "loss": 1.0219, "step": 17417 }, { "epoch": 1.8533730580974677, "grad_norm": 11.234375, "learning_rate": 9.398877202956226e-06, "loss": 0.9491, "step": 17418 }, { "epoch": 1.8534794637156842, "grad_norm": 14.28125, "learning_rate": 9.398841671404207e-06, "loss": 0.9408, "step": 17419 }, { "epoch": 1.8535858693339007, "grad_norm": 12.8515625, "learning_rate": 9.398806139852189e-06, "loss": 1.0966, "step": 17420 }, { "epoch": 1.8536922749521176, "grad_norm": 12.0859375, "learning_rate": 9.398770608300172e-06, "loss": 0.9253, "step": 17421 }, { "epoch": 1.8537986805703341, "grad_norm": 10.71875, "learning_rate": 9.398735076748154e-06, "loss": 1.1444, "step": 17422 }, { "epoch": 1.8539050861885507, "grad_norm": 9.6484375, "learning_rate": 9.398699545196135e-06, "loss": 1.1971, "step": 17423 }, { "epoch": 1.8540114918067674, "grad_norm": 13.5546875, "learning_rate": 9.398664013644117e-06, "loss": 1.1864, "step": 17424 }, { "epoch": 1.8541178974249841, "grad_norm": 8.5390625, "learning_rate": 9.398628482092099e-06, "loss": 1.0128, "step": 17425 }, { "epoch": 1.8542243030432006, "grad_norm": 22.921875, "learning_rate": 9.39859295054008e-06, "loss": 1.4719, "step": 17426 }, { "epoch": 1.8543307086614174, "grad_norm": 13.09375, "learning_rate": 9.398557418988062e-06, "loss": 1.3487, "step": 17427 }, { "epoch": 1.854437114279634, "grad_norm": 9.2890625, "learning_rate": 9.398521887436043e-06, "loss": 1.0252, "step": 17428 }, { "epoch": 1.8545435198978506, "grad_norm": 17.078125, "learning_rate": 9.398486355884025e-06, "loss": 1.1966, "step": 17429 }, { "epoch": 1.854649925516067, "grad_norm": 15.2265625, "learning_rate": 9.398450824332007e-06, "loss": 1.1054, "step": 17430 }, { "epoch": 1.8547563311342838, "grad_norm": 18.578125, "learning_rate": 9.39841529277999e-06, "loss": 1.0003, "step": 17431 }, { "epoch": 1.8548627367525006, "grad_norm": 11.3046875, "learning_rate": 9.398379761227972e-06, "loss": 1.158, "step": 17432 }, { "epoch": 1.854969142370717, "grad_norm": 8.828125, "learning_rate": 9.398344229675953e-06, "loss": 0.882, "step": 17433 }, { "epoch": 1.8550755479889338, "grad_norm": 13.4921875, "learning_rate": 9.398308698123935e-06, "loss": 0.989, "step": 17434 }, { "epoch": 1.8551819536071505, "grad_norm": 12.3984375, "learning_rate": 9.398273166571916e-06, "loss": 0.9522, "step": 17435 }, { "epoch": 1.855288359225367, "grad_norm": 10.03125, "learning_rate": 9.3982376350199e-06, "loss": 1.3127, "step": 17436 }, { "epoch": 1.8553947648435838, "grad_norm": 14.0078125, "learning_rate": 9.398202103467881e-06, "loss": 1.0033, "step": 17437 }, { "epoch": 1.8555011704618005, "grad_norm": 13.4296875, "learning_rate": 9.398166571915863e-06, "loss": 1.2425, "step": 17438 }, { "epoch": 1.855607576080017, "grad_norm": 19.65625, "learning_rate": 9.398131040363844e-06, "loss": 1.005, "step": 17439 }, { "epoch": 1.8557139816982335, "grad_norm": 18.421875, "learning_rate": 9.398095508811826e-06, "loss": 1.0546, "step": 17440 }, { "epoch": 1.8558203873164503, "grad_norm": 12.4140625, "learning_rate": 9.398059977259808e-06, "loss": 1.2297, "step": 17441 }, { "epoch": 1.855926792934667, "grad_norm": 10.6328125, "learning_rate": 9.39802444570779e-06, "loss": 0.9458, "step": 17442 }, { "epoch": 1.8560331985528835, "grad_norm": 10.3125, "learning_rate": 9.397988914155771e-06, "loss": 1.1847, "step": 17443 }, { "epoch": 1.8561396041711002, "grad_norm": 15.109375, "learning_rate": 9.397953382603753e-06, "loss": 1.1568, "step": 17444 }, { "epoch": 1.856246009789317, "grad_norm": 13.1796875, "learning_rate": 9.397917851051734e-06, "loss": 1.2422, "step": 17445 }, { "epoch": 1.8563524154075335, "grad_norm": 10.5, "learning_rate": 9.397882319499716e-06, "loss": 1.1279, "step": 17446 }, { "epoch": 1.85645882102575, "grad_norm": 15.390625, "learning_rate": 9.397846787947699e-06, "loss": 1.0404, "step": 17447 }, { "epoch": 1.856565226643967, "grad_norm": 9.4921875, "learning_rate": 9.39781125639568e-06, "loss": 0.9735, "step": 17448 }, { "epoch": 1.8566716322621835, "grad_norm": 15.7890625, "learning_rate": 9.397775724843662e-06, "loss": 1.0398, "step": 17449 }, { "epoch": 1.8567780378804, "grad_norm": 13.8515625, "learning_rate": 9.397740193291644e-06, "loss": 1.0495, "step": 17450 }, { "epoch": 1.8568844434986167, "grad_norm": 15.515625, "learning_rate": 9.397704661739625e-06, "loss": 1.2331, "step": 17451 }, { "epoch": 1.8569908491168334, "grad_norm": 12.0703125, "learning_rate": 9.397669130187607e-06, "loss": 1.0525, "step": 17452 }, { "epoch": 1.85709725473505, "grad_norm": 17.21875, "learning_rate": 9.397633598635589e-06, "loss": 1.0881, "step": 17453 }, { "epoch": 1.8572036603532667, "grad_norm": 12.9140625, "learning_rate": 9.39759806708357e-06, "loss": 1.2186, "step": 17454 }, { "epoch": 1.8573100659714834, "grad_norm": 10.0234375, "learning_rate": 9.397562535531552e-06, "loss": 1.0679, "step": 17455 }, { "epoch": 1.8574164715897, "grad_norm": 13.203125, "learning_rate": 9.397527003979534e-06, "loss": 1.122, "step": 17456 }, { "epoch": 1.8575228772079164, "grad_norm": 15.015625, "learning_rate": 9.397491472427517e-06, "loss": 1.1094, "step": 17457 }, { "epoch": 1.8576292828261334, "grad_norm": 10.3359375, "learning_rate": 9.397455940875498e-06, "loss": 0.9994, "step": 17458 }, { "epoch": 1.8577356884443499, "grad_norm": 12.4765625, "learning_rate": 9.39742040932348e-06, "loss": 1.404, "step": 17459 }, { "epoch": 1.8578420940625664, "grad_norm": 13.5703125, "learning_rate": 9.397384877771462e-06, "loss": 1.0584, "step": 17460 }, { "epoch": 1.8579484996807831, "grad_norm": 11.3671875, "learning_rate": 9.397349346219443e-06, "loss": 0.987, "step": 17461 }, { "epoch": 1.8580549052989999, "grad_norm": 10.6015625, "learning_rate": 9.397313814667427e-06, "loss": 1.0481, "step": 17462 }, { "epoch": 1.8581613109172164, "grad_norm": 10.21875, "learning_rate": 9.397278283115408e-06, "loss": 1.048, "step": 17463 }, { "epoch": 1.858267716535433, "grad_norm": 11.6171875, "learning_rate": 9.39724275156339e-06, "loss": 1.1576, "step": 17464 }, { "epoch": 1.8583741221536498, "grad_norm": 10.78125, "learning_rate": 9.397207220011371e-06, "loss": 1.0116, "step": 17465 }, { "epoch": 1.8584805277718663, "grad_norm": 10.9453125, "learning_rate": 9.397171688459353e-06, "loss": 1.09, "step": 17466 }, { "epoch": 1.8585869333900829, "grad_norm": 12.6640625, "learning_rate": 9.397136156907335e-06, "loss": 1.1329, "step": 17467 }, { "epoch": 1.8586933390082996, "grad_norm": 15.59375, "learning_rate": 9.397100625355316e-06, "loss": 1.1652, "step": 17468 }, { "epoch": 1.8587997446265163, "grad_norm": 15.2109375, "learning_rate": 9.397065093803298e-06, "loss": 0.9388, "step": 17469 }, { "epoch": 1.8589061502447328, "grad_norm": 18.234375, "learning_rate": 9.39702956225128e-06, "loss": 0.8768, "step": 17470 }, { "epoch": 1.8590125558629496, "grad_norm": 16.078125, "learning_rate": 9.396994030699261e-06, "loss": 1.0167, "step": 17471 }, { "epoch": 1.8591189614811663, "grad_norm": 14.03125, "learning_rate": 9.396958499147244e-06, "loss": 1.018, "step": 17472 }, { "epoch": 1.8592253670993828, "grad_norm": 12.390625, "learning_rate": 9.396922967595226e-06, "loss": 1.2277, "step": 17473 }, { "epoch": 1.8593317727175995, "grad_norm": 13.0390625, "learning_rate": 9.396887436043208e-06, "loss": 1.1454, "step": 17474 }, { "epoch": 1.8594381783358163, "grad_norm": 14.5390625, "learning_rate": 9.396851904491189e-06, "loss": 0.9509, "step": 17475 }, { "epoch": 1.8595445839540328, "grad_norm": 15.9375, "learning_rate": 9.39681637293917e-06, "loss": 1.0081, "step": 17476 }, { "epoch": 1.8596509895722493, "grad_norm": 15.640625, "learning_rate": 9.396780841387152e-06, "loss": 1.2435, "step": 17477 }, { "epoch": 1.859757395190466, "grad_norm": 20.109375, "learning_rate": 9.396745309835134e-06, "loss": 1.1179, "step": 17478 }, { "epoch": 1.8598638008086827, "grad_norm": 14.1484375, "learning_rate": 9.396709778283116e-06, "loss": 1.0976, "step": 17479 }, { "epoch": 1.8599702064268993, "grad_norm": 22.71875, "learning_rate": 9.396674246731097e-06, "loss": 1.1585, "step": 17480 }, { "epoch": 1.860076612045116, "grad_norm": 16.46875, "learning_rate": 9.396638715179079e-06, "loss": 1.1509, "step": 17481 }, { "epoch": 1.8601830176633327, "grad_norm": 10.8203125, "learning_rate": 9.396603183627062e-06, "loss": 1.0917, "step": 17482 }, { "epoch": 1.8602894232815492, "grad_norm": 14.2890625, "learning_rate": 9.396567652075044e-06, "loss": 0.9962, "step": 17483 }, { "epoch": 1.860395828899766, "grad_norm": 12.8359375, "learning_rate": 9.396532120523025e-06, "loss": 1.1472, "step": 17484 }, { "epoch": 1.8605022345179827, "grad_norm": 18.03125, "learning_rate": 9.396496588971007e-06, "loss": 1.1519, "step": 17485 }, { "epoch": 1.8606086401361992, "grad_norm": 14.296875, "learning_rate": 9.396461057418988e-06, "loss": 0.8931, "step": 17486 }, { "epoch": 1.8607150457544157, "grad_norm": 10.9140625, "learning_rate": 9.39642552586697e-06, "loss": 0.9138, "step": 17487 }, { "epoch": 1.8608214513726324, "grad_norm": 13.34375, "learning_rate": 9.396389994314953e-06, "loss": 1.2012, "step": 17488 }, { "epoch": 1.8609278569908492, "grad_norm": 12.21875, "learning_rate": 9.396354462762935e-06, "loss": 1.1253, "step": 17489 }, { "epoch": 1.8610342626090657, "grad_norm": 12.421875, "learning_rate": 9.396318931210917e-06, "loss": 0.9882, "step": 17490 }, { "epoch": 1.8611406682272824, "grad_norm": 10.5546875, "learning_rate": 9.396283399658898e-06, "loss": 1.1726, "step": 17491 }, { "epoch": 1.8612470738454991, "grad_norm": 13.8984375, "learning_rate": 9.39624786810688e-06, "loss": 1.2006, "step": 17492 }, { "epoch": 1.8613534794637157, "grad_norm": 12.671875, "learning_rate": 9.396212336554861e-06, "loss": 1.2147, "step": 17493 }, { "epoch": 1.8614598850819322, "grad_norm": 14.0390625, "learning_rate": 9.396176805002843e-06, "loss": 1.0573, "step": 17494 }, { "epoch": 1.8615662907001491, "grad_norm": 12.1171875, "learning_rate": 9.396141273450825e-06, "loss": 1.2571, "step": 17495 }, { "epoch": 1.8616726963183656, "grad_norm": 9.28125, "learning_rate": 9.396105741898806e-06, "loss": 1.1136, "step": 17496 }, { "epoch": 1.8617791019365821, "grad_norm": 11.828125, "learning_rate": 9.396070210346788e-06, "loss": 0.966, "step": 17497 }, { "epoch": 1.8618855075547989, "grad_norm": 19.125, "learning_rate": 9.396034678794771e-06, "loss": 1.2568, "step": 17498 }, { "epoch": 1.8619919131730156, "grad_norm": 11.9765625, "learning_rate": 9.395999147242753e-06, "loss": 0.9244, "step": 17499 }, { "epoch": 1.862098318791232, "grad_norm": 14.0859375, "learning_rate": 9.395963615690734e-06, "loss": 1.0288, "step": 17500 }, { "epoch": 1.8622047244094488, "grad_norm": 10.7578125, "learning_rate": 9.395928084138716e-06, "loss": 1.0209, "step": 17501 }, { "epoch": 1.8623111300276656, "grad_norm": 10.9375, "learning_rate": 9.395892552586698e-06, "loss": 0.9339, "step": 17502 }, { "epoch": 1.862417535645882, "grad_norm": 17.203125, "learning_rate": 9.39585702103468e-06, "loss": 1.0874, "step": 17503 }, { "epoch": 1.8625239412640986, "grad_norm": 34.5, "learning_rate": 9.39582148948266e-06, "loss": 1.2278, "step": 17504 }, { "epoch": 1.8626303468823155, "grad_norm": 11.8046875, "learning_rate": 9.395785957930642e-06, "loss": 1.0736, "step": 17505 }, { "epoch": 1.862736752500532, "grad_norm": 12.9765625, "learning_rate": 9.395750426378624e-06, "loss": 1.3262, "step": 17506 }, { "epoch": 1.8628431581187486, "grad_norm": 13.2890625, "learning_rate": 9.395714894826606e-06, "loss": 1.3973, "step": 17507 }, { "epoch": 1.8629495637369653, "grad_norm": 15.2109375, "learning_rate": 9.395679363274589e-06, "loss": 0.9214, "step": 17508 }, { "epoch": 1.863055969355182, "grad_norm": 13.8046875, "learning_rate": 9.39564383172257e-06, "loss": 0.9466, "step": 17509 }, { "epoch": 1.8631623749733985, "grad_norm": 11.140625, "learning_rate": 9.395608300170552e-06, "loss": 0.994, "step": 17510 }, { "epoch": 1.8632687805916153, "grad_norm": 14.2421875, "learning_rate": 9.395572768618534e-06, "loss": 1.153, "step": 17511 }, { "epoch": 1.863375186209832, "grad_norm": 16.5, "learning_rate": 9.395537237066515e-06, "loss": 0.9973, "step": 17512 }, { "epoch": 1.8634815918280485, "grad_norm": 27.125, "learning_rate": 9.395501705514499e-06, "loss": 0.7989, "step": 17513 }, { "epoch": 1.863587997446265, "grad_norm": 16.6875, "learning_rate": 9.39546617396248e-06, "loss": 1.1613, "step": 17514 }, { "epoch": 1.8636944030644818, "grad_norm": 15.2578125, "learning_rate": 9.395430642410462e-06, "loss": 1.2026, "step": 17515 }, { "epoch": 1.8638008086826985, "grad_norm": 13.828125, "learning_rate": 9.395395110858443e-06, "loss": 1.4358, "step": 17516 }, { "epoch": 1.863907214300915, "grad_norm": 11.28125, "learning_rate": 9.395359579306425e-06, "loss": 1.2163, "step": 17517 }, { "epoch": 1.8640136199191317, "grad_norm": 13.3984375, "learning_rate": 9.395324047754407e-06, "loss": 1.0841, "step": 17518 }, { "epoch": 1.8641200255373485, "grad_norm": 11.6796875, "learning_rate": 9.395288516202388e-06, "loss": 1.3085, "step": 17519 }, { "epoch": 1.864226431155565, "grad_norm": 9.6484375, "learning_rate": 9.39525298465037e-06, "loss": 1.2014, "step": 17520 }, { "epoch": 1.8643328367737817, "grad_norm": 12.421875, "learning_rate": 9.395217453098351e-06, "loss": 1.0715, "step": 17521 }, { "epoch": 1.8644392423919984, "grad_norm": 19.140625, "learning_rate": 9.395181921546333e-06, "loss": 0.889, "step": 17522 }, { "epoch": 1.864545648010215, "grad_norm": 12.2890625, "learning_rate": 9.395146389994316e-06, "loss": 1.0375, "step": 17523 }, { "epoch": 1.8646520536284314, "grad_norm": 12.703125, "learning_rate": 9.395110858442298e-06, "loss": 1.1282, "step": 17524 }, { "epoch": 1.8647584592466482, "grad_norm": 14.09375, "learning_rate": 9.39507532689028e-06, "loss": 0.9657, "step": 17525 }, { "epoch": 1.864864864864865, "grad_norm": 17.890625, "learning_rate": 9.395039795338261e-06, "loss": 1.278, "step": 17526 }, { "epoch": 1.8649712704830814, "grad_norm": 21.65625, "learning_rate": 9.395004263786243e-06, "loss": 1.0809, "step": 17527 }, { "epoch": 1.8650776761012982, "grad_norm": 15.3828125, "learning_rate": 9.394968732234224e-06, "loss": 1.216, "step": 17528 }, { "epoch": 1.8651840817195149, "grad_norm": 14.0078125, "learning_rate": 9.394933200682208e-06, "loss": 1.2852, "step": 17529 }, { "epoch": 1.8652904873377314, "grad_norm": 13.15625, "learning_rate": 9.39489766913019e-06, "loss": 1.1179, "step": 17530 }, { "epoch": 1.8653968929559481, "grad_norm": 10.8828125, "learning_rate": 9.394862137578171e-06, "loss": 1.055, "step": 17531 }, { "epoch": 1.8655032985741649, "grad_norm": 14.984375, "learning_rate": 9.394826606026153e-06, "loss": 1.1418, "step": 17532 }, { "epoch": 1.8656097041923814, "grad_norm": 10.40625, "learning_rate": 9.394791074474132e-06, "loss": 1.0882, "step": 17533 }, { "epoch": 1.8657161098105979, "grad_norm": 8.1640625, "learning_rate": 9.394755542922116e-06, "loss": 0.8018, "step": 17534 }, { "epoch": 1.8658225154288146, "grad_norm": 18.328125, "learning_rate": 9.394720011370097e-06, "loss": 1.3647, "step": 17535 }, { "epoch": 1.8659289210470313, "grad_norm": 14.46875, "learning_rate": 9.394684479818079e-06, "loss": 0.9812, "step": 17536 }, { "epoch": 1.8660353266652479, "grad_norm": 12.2890625, "learning_rate": 9.39464894826606e-06, "loss": 0.8417, "step": 17537 }, { "epoch": 1.8661417322834646, "grad_norm": 11.9140625, "learning_rate": 9.394613416714042e-06, "loss": 1.239, "step": 17538 }, { "epoch": 1.8662481379016813, "grad_norm": 11.515625, "learning_rate": 9.394577885162025e-06, "loss": 1.0487, "step": 17539 }, { "epoch": 1.8663545435198978, "grad_norm": 9.84375, "learning_rate": 9.394542353610007e-06, "loss": 1.2169, "step": 17540 }, { "epoch": 1.8664609491381143, "grad_norm": 11.453125, "learning_rate": 9.394506822057989e-06, "loss": 1.1608, "step": 17541 }, { "epoch": 1.8665673547563313, "grad_norm": 11.3828125, "learning_rate": 9.39447129050597e-06, "loss": 1.0805, "step": 17542 }, { "epoch": 1.8666737603745478, "grad_norm": 8.84375, "learning_rate": 9.394435758953952e-06, "loss": 0.8499, "step": 17543 }, { "epoch": 1.8667801659927643, "grad_norm": 10.046875, "learning_rate": 9.394400227401934e-06, "loss": 1.011, "step": 17544 }, { "epoch": 1.866886571610981, "grad_norm": 9.96875, "learning_rate": 9.394364695849915e-06, "loss": 1.29, "step": 17545 }, { "epoch": 1.8669929772291978, "grad_norm": 15.4296875, "learning_rate": 9.394329164297897e-06, "loss": 1.1361, "step": 17546 }, { "epoch": 1.8670993828474143, "grad_norm": 14.4375, "learning_rate": 9.394293632745878e-06, "loss": 1.3637, "step": 17547 }, { "epoch": 1.867205788465631, "grad_norm": 10.75, "learning_rate": 9.39425810119386e-06, "loss": 1.0786, "step": 17548 }, { "epoch": 1.8673121940838477, "grad_norm": 12.890625, "learning_rate": 9.394222569641843e-06, "loss": 1.1902, "step": 17549 }, { "epoch": 1.8674185997020643, "grad_norm": 10.65625, "learning_rate": 9.394187038089825e-06, "loss": 1.1536, "step": 17550 }, { "epoch": 1.8675250053202808, "grad_norm": 17.015625, "learning_rate": 9.394151506537806e-06, "loss": 0.8358, "step": 17551 }, { "epoch": 1.8676314109384977, "grad_norm": 18.25, "learning_rate": 9.394115974985788e-06, "loss": 1.0318, "step": 17552 }, { "epoch": 1.8677378165567142, "grad_norm": 14.2421875, "learning_rate": 9.39408044343377e-06, "loss": 1.06, "step": 17553 }, { "epoch": 1.8678442221749307, "grad_norm": 11.3125, "learning_rate": 9.394044911881753e-06, "loss": 0.7955, "step": 17554 }, { "epoch": 1.8679506277931475, "grad_norm": 12.7890625, "learning_rate": 9.394009380329735e-06, "loss": 1.2505, "step": 17555 }, { "epoch": 1.8680570334113642, "grad_norm": 10.421875, "learning_rate": 9.393973848777716e-06, "loss": 1.2401, "step": 17556 }, { "epoch": 1.8681634390295807, "grad_norm": 12.1640625, "learning_rate": 9.393938317225698e-06, "loss": 1.2631, "step": 17557 }, { "epoch": 1.8682698446477974, "grad_norm": 18.234375, "learning_rate": 9.39390278567368e-06, "loss": 1.0612, "step": 17558 }, { "epoch": 1.8683762502660142, "grad_norm": 13.015625, "learning_rate": 9.393867254121661e-06, "loss": 1.1332, "step": 17559 }, { "epoch": 1.8684826558842307, "grad_norm": 14.734375, "learning_rate": 9.393831722569643e-06, "loss": 1.1111, "step": 17560 }, { "epoch": 1.8685890615024472, "grad_norm": 13.125, "learning_rate": 9.393796191017624e-06, "loss": 1.0542, "step": 17561 }, { "epoch": 1.868695467120664, "grad_norm": 10.28125, "learning_rate": 9.393760659465606e-06, "loss": 1.0889, "step": 17562 }, { "epoch": 1.8688018727388807, "grad_norm": 13.46875, "learning_rate": 9.393725127913587e-06, "loss": 1.0219, "step": 17563 }, { "epoch": 1.8689082783570972, "grad_norm": 11.5, "learning_rate": 9.393689596361569e-06, "loss": 1.0176, "step": 17564 }, { "epoch": 1.869014683975314, "grad_norm": 9.7890625, "learning_rate": 9.393654064809552e-06, "loss": 1.0844, "step": 17565 }, { "epoch": 1.8691210895935306, "grad_norm": 10.296875, "learning_rate": 9.393618533257534e-06, "loss": 0.9835, "step": 17566 }, { "epoch": 1.8692274952117471, "grad_norm": 16.015625, "learning_rate": 9.393583001705516e-06, "loss": 1.2037, "step": 17567 }, { "epoch": 1.8693339008299639, "grad_norm": 11.7421875, "learning_rate": 9.393547470153497e-06, "loss": 1.0225, "step": 17568 }, { "epoch": 1.8694403064481806, "grad_norm": 17.75, "learning_rate": 9.393511938601479e-06, "loss": 1.1088, "step": 17569 }, { "epoch": 1.869546712066397, "grad_norm": 12.6484375, "learning_rate": 9.39347640704946e-06, "loss": 1.1209, "step": 17570 }, { "epoch": 1.8696531176846136, "grad_norm": 12.0234375, "learning_rate": 9.393440875497442e-06, "loss": 1.0232, "step": 17571 }, { "epoch": 1.8697595233028304, "grad_norm": 12.921875, "learning_rate": 9.393405343945424e-06, "loss": 0.9811, "step": 17572 }, { "epoch": 1.869865928921047, "grad_norm": 14.203125, "learning_rate": 9.393369812393405e-06, "loss": 0.9889, "step": 17573 }, { "epoch": 1.8699723345392636, "grad_norm": 12.75, "learning_rate": 9.393334280841387e-06, "loss": 0.7841, "step": 17574 }, { "epoch": 1.8700787401574803, "grad_norm": 13.109375, "learning_rate": 9.39329874928937e-06, "loss": 0.9203, "step": 17575 }, { "epoch": 1.870185145775697, "grad_norm": 11.703125, "learning_rate": 9.393263217737352e-06, "loss": 1.0225, "step": 17576 }, { "epoch": 1.8702915513939136, "grad_norm": 11.0703125, "learning_rate": 9.393227686185333e-06, "loss": 1.1799, "step": 17577 }, { "epoch": 1.87039795701213, "grad_norm": 11.953125, "learning_rate": 9.393192154633315e-06, "loss": 0.9704, "step": 17578 }, { "epoch": 1.870504362630347, "grad_norm": 22.078125, "learning_rate": 9.393156623081297e-06, "loss": 0.8864, "step": 17579 }, { "epoch": 1.8706107682485635, "grad_norm": 11.7890625, "learning_rate": 9.39312109152928e-06, "loss": 1.0717, "step": 17580 }, { "epoch": 1.87071717386678, "grad_norm": 9.5234375, "learning_rate": 9.393085559977261e-06, "loss": 1.0865, "step": 17581 }, { "epoch": 1.8708235794849968, "grad_norm": 17.125, "learning_rate": 9.393050028425243e-06, "loss": 1.0309, "step": 17582 }, { "epoch": 1.8709299851032135, "grad_norm": 10.7578125, "learning_rate": 9.393014496873225e-06, "loss": 1.1134, "step": 17583 }, { "epoch": 1.87103639072143, "grad_norm": 13.1015625, "learning_rate": 9.392978965321206e-06, "loss": 1.2341, "step": 17584 }, { "epoch": 1.8711427963396468, "grad_norm": 10.3203125, "learning_rate": 9.392943433769188e-06, "loss": 1.0478, "step": 17585 }, { "epoch": 1.8712492019578635, "grad_norm": 11.0703125, "learning_rate": 9.39290790221717e-06, "loss": 0.8938, "step": 17586 }, { "epoch": 1.87135560757608, "grad_norm": 10.0546875, "learning_rate": 9.392872370665151e-06, "loss": 1.0858, "step": 17587 }, { "epoch": 1.8714620131942965, "grad_norm": 14.1484375, "learning_rate": 9.392836839113133e-06, "loss": 1.044, "step": 17588 }, { "epoch": 1.8715684188125135, "grad_norm": 14.890625, "learning_rate": 9.392801307561114e-06, "loss": 1.4576, "step": 17589 }, { "epoch": 1.87167482443073, "grad_norm": 12.5703125, "learning_rate": 9.392765776009098e-06, "loss": 1.4156, "step": 17590 }, { "epoch": 1.8717812300489465, "grad_norm": 14.2109375, "learning_rate": 9.39273024445708e-06, "loss": 1.0549, "step": 17591 }, { "epoch": 1.8718876356671632, "grad_norm": 14.2578125, "learning_rate": 9.39269471290506e-06, "loss": 1.1227, "step": 17592 }, { "epoch": 1.87199404128538, "grad_norm": 16.578125, "learning_rate": 9.392659181353042e-06, "loss": 1.2253, "step": 17593 }, { "epoch": 1.8721004469035964, "grad_norm": 12.0234375, "learning_rate": 9.392623649801024e-06, "loss": 1.2979, "step": 17594 }, { "epoch": 1.8722068525218132, "grad_norm": 12.484375, "learning_rate": 9.392588118249006e-06, "loss": 0.926, "step": 17595 }, { "epoch": 1.87231325814003, "grad_norm": 9.421875, "learning_rate": 9.392552586696987e-06, "loss": 1.0284, "step": 17596 }, { "epoch": 1.8724196637582464, "grad_norm": 9.75, "learning_rate": 9.392517055144969e-06, "loss": 1.2049, "step": 17597 }, { "epoch": 1.872526069376463, "grad_norm": 12.6796875, "learning_rate": 9.39248152359295e-06, "loss": 1.2452, "step": 17598 }, { "epoch": 1.8726324749946799, "grad_norm": 21.390625, "learning_rate": 9.392445992040932e-06, "loss": 0.9934, "step": 17599 }, { "epoch": 1.8727388806128964, "grad_norm": 9.7890625, "learning_rate": 9.392410460488915e-06, "loss": 1.1384, "step": 17600 }, { "epoch": 1.872845286231113, "grad_norm": 10.7734375, "learning_rate": 9.392374928936897e-06, "loss": 1.1469, "step": 17601 }, { "epoch": 1.8729516918493296, "grad_norm": 13.5, "learning_rate": 9.392339397384879e-06, "loss": 1.1103, "step": 17602 }, { "epoch": 1.8730580974675464, "grad_norm": 11.734375, "learning_rate": 9.39230386583286e-06, "loss": 1.1649, "step": 17603 }, { "epoch": 1.8731645030857629, "grad_norm": 20.296875, "learning_rate": 9.392268334280842e-06, "loss": 1.0161, "step": 17604 }, { "epoch": 1.8732709087039796, "grad_norm": 11.6328125, "learning_rate": 9.392232802728823e-06, "loss": 1.1665, "step": 17605 }, { "epoch": 1.8733773143221963, "grad_norm": 13.5078125, "learning_rate": 9.392197271176807e-06, "loss": 0.9682, "step": 17606 }, { "epoch": 1.8734837199404129, "grad_norm": 16.046875, "learning_rate": 9.392161739624788e-06, "loss": 1.2315, "step": 17607 }, { "epoch": 1.8735901255586294, "grad_norm": 14.5625, "learning_rate": 9.39212620807277e-06, "loss": 1.0425, "step": 17608 }, { "epoch": 1.873696531176846, "grad_norm": 16.40625, "learning_rate": 9.392090676520752e-06, "loss": 0.9604, "step": 17609 }, { "epoch": 1.8738029367950628, "grad_norm": 11.171875, "learning_rate": 9.392055144968733e-06, "loss": 0.8404, "step": 17610 }, { "epoch": 1.8739093424132793, "grad_norm": 11.328125, "learning_rate": 9.392019613416715e-06, "loss": 0.9544, "step": 17611 }, { "epoch": 1.874015748031496, "grad_norm": 11.140625, "learning_rate": 9.391984081864696e-06, "loss": 0.9726, "step": 17612 }, { "epoch": 1.8741221536497128, "grad_norm": 14.0234375, "learning_rate": 9.391948550312678e-06, "loss": 1.3026, "step": 17613 }, { "epoch": 1.8742285592679293, "grad_norm": 13.2578125, "learning_rate": 9.39191301876066e-06, "loss": 1.2502, "step": 17614 }, { "epoch": 1.874334964886146, "grad_norm": 8.6875, "learning_rate": 9.391877487208641e-06, "loss": 0.9339, "step": 17615 }, { "epoch": 1.8744413705043628, "grad_norm": 18.140625, "learning_rate": 9.391841955656624e-06, "loss": 1.0602, "step": 17616 }, { "epoch": 1.8745477761225793, "grad_norm": 13.3359375, "learning_rate": 9.391806424104606e-06, "loss": 0.978, "step": 17617 }, { "epoch": 1.8746541817407958, "grad_norm": 13.1171875, "learning_rate": 9.391770892552588e-06, "loss": 1.1775, "step": 17618 }, { "epoch": 1.8747605873590125, "grad_norm": 17.75, "learning_rate": 9.39173536100057e-06, "loss": 1.1295, "step": 17619 }, { "epoch": 1.8748669929772293, "grad_norm": 16.890625, "learning_rate": 9.391699829448551e-06, "loss": 0.9317, "step": 17620 }, { "epoch": 1.8749733985954458, "grad_norm": 11.9453125, "learning_rate": 9.391664297896534e-06, "loss": 1.1839, "step": 17621 }, { "epoch": 1.8750798042136625, "grad_norm": 12.765625, "learning_rate": 9.391628766344514e-06, "loss": 1.0518, "step": 17622 }, { "epoch": 1.8751862098318792, "grad_norm": 16.59375, "learning_rate": 9.391593234792496e-06, "loss": 0.9861, "step": 17623 }, { "epoch": 1.8752926154500957, "grad_norm": 14.046875, "learning_rate": 9.391557703240477e-06, "loss": 1.025, "step": 17624 }, { "epoch": 1.8753990210683122, "grad_norm": 14.71875, "learning_rate": 9.391522171688459e-06, "loss": 1.0305, "step": 17625 }, { "epoch": 1.8755054266865292, "grad_norm": 13.9140625, "learning_rate": 9.391486640136442e-06, "loss": 1.1519, "step": 17626 }, { "epoch": 1.8756118323047457, "grad_norm": 11.4453125, "learning_rate": 9.391451108584424e-06, "loss": 0.923, "step": 17627 }, { "epoch": 1.8757182379229622, "grad_norm": 15.8515625, "learning_rate": 9.391415577032405e-06, "loss": 0.9562, "step": 17628 }, { "epoch": 1.875824643541179, "grad_norm": 11.3125, "learning_rate": 9.391380045480387e-06, "loss": 1.1154, "step": 17629 }, { "epoch": 1.8759310491593957, "grad_norm": 10.171875, "learning_rate": 9.391344513928369e-06, "loss": 1.1014, "step": 17630 }, { "epoch": 1.8760374547776122, "grad_norm": 15.53125, "learning_rate": 9.391308982376352e-06, "loss": 1.0075, "step": 17631 }, { "epoch": 1.876143860395829, "grad_norm": 9.484375, "learning_rate": 9.391273450824334e-06, "loss": 1.1925, "step": 17632 }, { "epoch": 1.8762502660140457, "grad_norm": 14.5234375, "learning_rate": 9.391237919272315e-06, "loss": 1.238, "step": 17633 }, { "epoch": 1.8763566716322622, "grad_norm": 12.03125, "learning_rate": 9.391202387720297e-06, "loss": 1.0926, "step": 17634 }, { "epoch": 1.8764630772504787, "grad_norm": 16.390625, "learning_rate": 9.391166856168278e-06, "loss": 1.6049, "step": 17635 }, { "epoch": 1.8765694828686956, "grad_norm": 10.609375, "learning_rate": 9.39113132461626e-06, "loss": 0.9809, "step": 17636 }, { "epoch": 1.8766758884869121, "grad_norm": 13.75, "learning_rate": 9.391095793064242e-06, "loss": 1.1039, "step": 17637 }, { "epoch": 1.8767822941051286, "grad_norm": 15.390625, "learning_rate": 9.391060261512223e-06, "loss": 0.9461, "step": 17638 }, { "epoch": 1.8768886997233454, "grad_norm": 12.0390625, "learning_rate": 9.391024729960205e-06, "loss": 1.35, "step": 17639 }, { "epoch": 1.876995105341562, "grad_norm": 12.0234375, "learning_rate": 9.390989198408186e-06, "loss": 1.0384, "step": 17640 }, { "epoch": 1.8771015109597786, "grad_norm": 12.484375, "learning_rate": 9.390953666856168e-06, "loss": 1.2654, "step": 17641 }, { "epoch": 1.8772079165779954, "grad_norm": 15.0625, "learning_rate": 9.390918135304151e-06, "loss": 1.2386, "step": 17642 }, { "epoch": 1.877314322196212, "grad_norm": 12.1171875, "learning_rate": 9.390882603752133e-06, "loss": 1.1497, "step": 17643 }, { "epoch": 1.8774207278144286, "grad_norm": 11.1328125, "learning_rate": 9.390847072200115e-06, "loss": 1.0736, "step": 17644 }, { "epoch": 1.877527133432645, "grad_norm": 11.0703125, "learning_rate": 9.390811540648096e-06, "loss": 1.1549, "step": 17645 }, { "epoch": 1.8776335390508618, "grad_norm": 9.75, "learning_rate": 9.390776009096078e-06, "loss": 0.9323, "step": 17646 }, { "epoch": 1.8777399446690786, "grad_norm": 16.125, "learning_rate": 9.390740477544061e-06, "loss": 1.2132, "step": 17647 }, { "epoch": 1.877846350287295, "grad_norm": 12.9375, "learning_rate": 9.390704945992043e-06, "loss": 1.071, "step": 17648 }, { "epoch": 1.8779527559055118, "grad_norm": 14.21875, "learning_rate": 9.390669414440024e-06, "loss": 0.9788, "step": 17649 }, { "epoch": 1.8780591615237285, "grad_norm": 11.9765625, "learning_rate": 9.390633882888006e-06, "loss": 1.2648, "step": 17650 }, { "epoch": 1.878165567141945, "grad_norm": 12.0, "learning_rate": 9.390598351335987e-06, "loss": 1.0736, "step": 17651 }, { "epoch": 1.8782719727601618, "grad_norm": 9.5703125, "learning_rate": 9.390562819783969e-06, "loss": 0.915, "step": 17652 }, { "epoch": 1.8783783783783785, "grad_norm": 10.4375, "learning_rate": 9.39052728823195e-06, "loss": 1.0399, "step": 17653 }, { "epoch": 1.878484783996595, "grad_norm": 11.671875, "learning_rate": 9.390491756679932e-06, "loss": 1.178, "step": 17654 }, { "epoch": 1.8785911896148115, "grad_norm": 14.375, "learning_rate": 9.390456225127914e-06, "loss": 0.9928, "step": 17655 }, { "epoch": 1.8786975952330283, "grad_norm": 21.140625, "learning_rate": 9.390420693575895e-06, "loss": 1.1553, "step": 17656 }, { "epoch": 1.878804000851245, "grad_norm": 9.40625, "learning_rate": 9.390385162023879e-06, "loss": 0.9603, "step": 17657 }, { "epoch": 1.8789104064694615, "grad_norm": 15.8671875, "learning_rate": 9.39034963047186e-06, "loss": 0.9388, "step": 17658 }, { "epoch": 1.8790168120876782, "grad_norm": 13.5, "learning_rate": 9.390314098919842e-06, "loss": 0.9824, "step": 17659 }, { "epoch": 1.879123217705895, "grad_norm": 13.5703125, "learning_rate": 9.390278567367824e-06, "loss": 1.0784, "step": 17660 }, { "epoch": 1.8792296233241115, "grad_norm": 14.84375, "learning_rate": 9.390243035815805e-06, "loss": 1.3982, "step": 17661 }, { "epoch": 1.8793360289423282, "grad_norm": 12.1953125, "learning_rate": 9.390207504263787e-06, "loss": 1.0058, "step": 17662 }, { "epoch": 1.879442434560545, "grad_norm": 16.78125, "learning_rate": 9.390171972711768e-06, "loss": 1.0133, "step": 17663 }, { "epoch": 1.8795488401787614, "grad_norm": 15.515625, "learning_rate": 9.39013644115975e-06, "loss": 0.9737, "step": 17664 }, { "epoch": 1.879655245796978, "grad_norm": 12.875, "learning_rate": 9.390100909607732e-06, "loss": 1.0717, "step": 17665 }, { "epoch": 1.8797616514151947, "grad_norm": 12.2109375, "learning_rate": 9.390065378055713e-06, "loss": 1.0464, "step": 17666 }, { "epoch": 1.8798680570334114, "grad_norm": 10.90625, "learning_rate": 9.390029846503697e-06, "loss": 0.9725, "step": 17667 }, { "epoch": 1.879974462651628, "grad_norm": 11.5234375, "learning_rate": 9.389994314951678e-06, "loss": 1.0047, "step": 17668 }, { "epoch": 1.8800808682698447, "grad_norm": 11.9921875, "learning_rate": 9.38995878339966e-06, "loss": 1.0904, "step": 17669 }, { "epoch": 1.8801872738880614, "grad_norm": 15.1484375, "learning_rate": 9.389923251847641e-06, "loss": 1.0311, "step": 17670 }, { "epoch": 1.880293679506278, "grad_norm": 9.4375, "learning_rate": 9.389887720295623e-06, "loss": 0.9576, "step": 17671 }, { "epoch": 1.8804000851244944, "grad_norm": 16.5625, "learning_rate": 9.389852188743606e-06, "loss": 1.0423, "step": 17672 }, { "epoch": 1.8805064907427114, "grad_norm": 12.734375, "learning_rate": 9.389816657191588e-06, "loss": 1.0378, "step": 17673 }, { "epoch": 1.8806128963609279, "grad_norm": 11.0234375, "learning_rate": 9.38978112563957e-06, "loss": 1.1303, "step": 17674 }, { "epoch": 1.8807193019791444, "grad_norm": 10.6796875, "learning_rate": 9.389745594087551e-06, "loss": 1.1353, "step": 17675 }, { "epoch": 1.8808257075973611, "grad_norm": 15.6015625, "learning_rate": 9.389710062535533e-06, "loss": 1.2928, "step": 17676 }, { "epoch": 1.8809321132155778, "grad_norm": 14.9765625, "learning_rate": 9.389674530983514e-06, "loss": 1.196, "step": 17677 }, { "epoch": 1.8810385188337944, "grad_norm": 13.9296875, "learning_rate": 9.389638999431496e-06, "loss": 1.048, "step": 17678 }, { "epoch": 1.881144924452011, "grad_norm": 11.4609375, "learning_rate": 9.389603467879478e-06, "loss": 1.2848, "step": 17679 }, { "epoch": 1.8812513300702278, "grad_norm": 12.6640625, "learning_rate": 9.389567936327459e-06, "loss": 0.9642, "step": 17680 }, { "epoch": 1.8813577356884443, "grad_norm": 15.203125, "learning_rate": 9.38953240477544e-06, "loss": 1.0941, "step": 17681 }, { "epoch": 1.8814641413066608, "grad_norm": 11.15625, "learning_rate": 9.389496873223422e-06, "loss": 0.9728, "step": 17682 }, { "epoch": 1.8815705469248778, "grad_norm": 13.703125, "learning_rate": 9.389461341671406e-06, "loss": 1.5574, "step": 17683 }, { "epoch": 1.8816769525430943, "grad_norm": 12.2890625, "learning_rate": 9.389425810119387e-06, "loss": 1.0563, "step": 17684 }, { "epoch": 1.8817833581613108, "grad_norm": 14.0078125, "learning_rate": 9.389390278567369e-06, "loss": 1.4763, "step": 17685 }, { "epoch": 1.8818897637795275, "grad_norm": 13.0703125, "learning_rate": 9.38935474701535e-06, "loss": 1.1479, "step": 17686 }, { "epoch": 1.8819961693977443, "grad_norm": 12.1171875, "learning_rate": 9.389319215463332e-06, "loss": 1.2407, "step": 17687 }, { "epoch": 1.8821025750159608, "grad_norm": 11.0390625, "learning_rate": 9.389283683911314e-06, "loss": 1.3927, "step": 17688 }, { "epoch": 1.8822089806341775, "grad_norm": 12.3515625, "learning_rate": 9.389248152359295e-06, "loss": 0.785, "step": 17689 }, { "epoch": 1.8823153862523943, "grad_norm": 13.1875, "learning_rate": 9.389212620807277e-06, "loss": 1.2045, "step": 17690 }, { "epoch": 1.8824217918706108, "grad_norm": 12.0703125, "learning_rate": 9.389177089255259e-06, "loss": 0.9467, "step": 17691 }, { "epoch": 1.8825281974888273, "grad_norm": 15.5078125, "learning_rate": 9.38914155770324e-06, "loss": 1.1697, "step": 17692 }, { "epoch": 1.882634603107044, "grad_norm": 14.953125, "learning_rate": 9.389106026151223e-06, "loss": 1.0896, "step": 17693 }, { "epoch": 1.8827410087252607, "grad_norm": 15.78125, "learning_rate": 9.389070494599205e-06, "loss": 1.065, "step": 17694 }, { "epoch": 1.8828474143434772, "grad_norm": 12.6640625, "learning_rate": 9.389034963047187e-06, "loss": 0.9238, "step": 17695 }, { "epoch": 1.882953819961694, "grad_norm": 13.828125, "learning_rate": 9.388999431495168e-06, "loss": 0.955, "step": 17696 }, { "epoch": 1.8830602255799107, "grad_norm": 18.6875, "learning_rate": 9.38896389994315e-06, "loss": 1.4321, "step": 17697 }, { "epoch": 1.8831666311981272, "grad_norm": 13.7109375, "learning_rate": 9.388928368391133e-06, "loss": 0.994, "step": 17698 }, { "epoch": 1.883273036816344, "grad_norm": 12.0859375, "learning_rate": 9.388892836839115e-06, "loss": 1.1089, "step": 17699 }, { "epoch": 1.8833794424345607, "grad_norm": 13.1015625, "learning_rate": 9.388857305287096e-06, "loss": 1.0371, "step": 17700 }, { "epoch": 1.8834858480527772, "grad_norm": 13.703125, "learning_rate": 9.388821773735078e-06, "loss": 1.2182, "step": 17701 }, { "epoch": 1.8835922536709937, "grad_norm": 13.1328125, "learning_rate": 9.38878624218306e-06, "loss": 1.001, "step": 17702 }, { "epoch": 1.8836986592892104, "grad_norm": 10.6953125, "learning_rate": 9.388750710631041e-06, "loss": 1.0355, "step": 17703 }, { "epoch": 1.8838050649074272, "grad_norm": 17.453125, "learning_rate": 9.388715179079023e-06, "loss": 0.8671, "step": 17704 }, { "epoch": 1.8839114705256437, "grad_norm": 15.84375, "learning_rate": 9.388679647527004e-06, "loss": 1.1058, "step": 17705 }, { "epoch": 1.8840178761438604, "grad_norm": 16.859375, "learning_rate": 9.388644115974986e-06, "loss": 1.1392, "step": 17706 }, { "epoch": 1.8841242817620771, "grad_norm": 11.4609375, "learning_rate": 9.388608584422968e-06, "loss": 0.9979, "step": 17707 }, { "epoch": 1.8842306873802936, "grad_norm": 8.71875, "learning_rate": 9.388573052870951e-06, "loss": 0.8918, "step": 17708 }, { "epoch": 1.8843370929985104, "grad_norm": 12.3125, "learning_rate": 9.388537521318933e-06, "loss": 1.0672, "step": 17709 }, { "epoch": 1.884443498616727, "grad_norm": 9.5234375, "learning_rate": 9.388501989766914e-06, "loss": 0.9354, "step": 17710 }, { "epoch": 1.8845499042349436, "grad_norm": 10.1640625, "learning_rate": 9.388466458214896e-06, "loss": 1.0841, "step": 17711 }, { "epoch": 1.8846563098531601, "grad_norm": 10.46875, "learning_rate": 9.388430926662877e-06, "loss": 1.1498, "step": 17712 }, { "epoch": 1.8847627154713769, "grad_norm": 10.25, "learning_rate": 9.388395395110859e-06, "loss": 1.0844, "step": 17713 }, { "epoch": 1.8848691210895936, "grad_norm": 10.296875, "learning_rate": 9.38835986355884e-06, "loss": 1.3749, "step": 17714 }, { "epoch": 1.88497552670781, "grad_norm": 11.0546875, "learning_rate": 9.388324332006822e-06, "loss": 1.0929, "step": 17715 }, { "epoch": 1.8850819323260268, "grad_norm": 12.4296875, "learning_rate": 9.388288800454804e-06, "loss": 0.9528, "step": 17716 }, { "epoch": 1.8851883379442436, "grad_norm": 15.640625, "learning_rate": 9.388253268902785e-06, "loss": 1.1249, "step": 17717 }, { "epoch": 1.88529474356246, "grad_norm": 17.0, "learning_rate": 9.388217737350769e-06, "loss": 0.9389, "step": 17718 }, { "epoch": 1.8854011491806766, "grad_norm": 16.34375, "learning_rate": 9.38818220579875e-06, "loss": 1.1161, "step": 17719 }, { "epoch": 1.8855075547988935, "grad_norm": 16.984375, "learning_rate": 9.388146674246732e-06, "loss": 1.2545, "step": 17720 }, { "epoch": 1.88561396041711, "grad_norm": 13.734375, "learning_rate": 9.388111142694713e-06, "loss": 1.2849, "step": 17721 }, { "epoch": 1.8857203660353266, "grad_norm": 13.0625, "learning_rate": 9.388075611142695e-06, "loss": 1.0449, "step": 17722 }, { "epoch": 1.8858267716535433, "grad_norm": 14.1640625, "learning_rate": 9.388040079590677e-06, "loss": 1.1086, "step": 17723 }, { "epoch": 1.88593317727176, "grad_norm": 11.2109375, "learning_rate": 9.38800454803866e-06, "loss": 1.0353, "step": 17724 }, { "epoch": 1.8860395828899765, "grad_norm": 12.5078125, "learning_rate": 9.387969016486642e-06, "loss": 1.0143, "step": 17725 }, { "epoch": 1.8861459885081933, "grad_norm": 17.546875, "learning_rate": 9.387933484934623e-06, "loss": 1.1795, "step": 17726 }, { "epoch": 1.88625239412641, "grad_norm": 17.125, "learning_rate": 9.387897953382605e-06, "loss": 0.9737, "step": 17727 }, { "epoch": 1.8863587997446265, "grad_norm": 12.0390625, "learning_rate": 9.387862421830586e-06, "loss": 1.2891, "step": 17728 }, { "epoch": 1.886465205362843, "grad_norm": 15.3671875, "learning_rate": 9.387826890278568e-06, "loss": 0.9667, "step": 17729 }, { "epoch": 1.88657161098106, "grad_norm": 16.953125, "learning_rate": 9.38779135872655e-06, "loss": 1.2031, "step": 17730 }, { "epoch": 1.8866780165992765, "grad_norm": 10.75, "learning_rate": 9.387755827174531e-06, "loss": 1.0879, "step": 17731 }, { "epoch": 1.886784422217493, "grad_norm": 12.4375, "learning_rate": 9.387720295622513e-06, "loss": 1.1491, "step": 17732 }, { "epoch": 1.8868908278357097, "grad_norm": 12.109375, "learning_rate": 9.387684764070494e-06, "loss": 1.0055, "step": 17733 }, { "epoch": 1.8869972334539264, "grad_norm": 17.078125, "learning_rate": 9.387649232518478e-06, "loss": 1.091, "step": 17734 }, { "epoch": 1.887103639072143, "grad_norm": 11.7890625, "learning_rate": 9.38761370096646e-06, "loss": 0.9426, "step": 17735 }, { "epoch": 1.8872100446903597, "grad_norm": 14.5859375, "learning_rate": 9.387578169414441e-06, "loss": 1.0886, "step": 17736 }, { "epoch": 1.8873164503085764, "grad_norm": 10.2421875, "learning_rate": 9.387542637862423e-06, "loss": 0.9201, "step": 17737 }, { "epoch": 1.887422855926793, "grad_norm": 17.140625, "learning_rate": 9.387507106310404e-06, "loss": 1.2366, "step": 17738 }, { "epoch": 1.8875292615450094, "grad_norm": 12.3359375, "learning_rate": 9.387471574758387e-06, "loss": 0.9237, "step": 17739 }, { "epoch": 1.8876356671632262, "grad_norm": 13.84375, "learning_rate": 9.387436043206369e-06, "loss": 0.9974, "step": 17740 }, { "epoch": 1.887742072781443, "grad_norm": 12.734375, "learning_rate": 9.38740051165435e-06, "loss": 1.1624, "step": 17741 }, { "epoch": 1.8878484783996594, "grad_norm": 15.4140625, "learning_rate": 9.38736498010233e-06, "loss": 1.079, "step": 17742 }, { "epoch": 1.8879548840178761, "grad_norm": 12.1015625, "learning_rate": 9.387329448550312e-06, "loss": 1.0679, "step": 17743 }, { "epoch": 1.8880612896360929, "grad_norm": 14.578125, "learning_rate": 9.387293916998296e-06, "loss": 1.5837, "step": 17744 }, { "epoch": 1.8881676952543094, "grad_norm": 13.6484375, "learning_rate": 9.387258385446277e-06, "loss": 0.9438, "step": 17745 }, { "epoch": 1.8882741008725261, "grad_norm": 15.2890625, "learning_rate": 9.387222853894259e-06, "loss": 0.738, "step": 17746 }, { "epoch": 1.8883805064907428, "grad_norm": 13.1328125, "learning_rate": 9.38718732234224e-06, "loss": 1.1028, "step": 17747 }, { "epoch": 1.8884869121089594, "grad_norm": 12.734375, "learning_rate": 9.387151790790222e-06, "loss": 1.2503, "step": 17748 }, { "epoch": 1.8885933177271759, "grad_norm": 11.1015625, "learning_rate": 9.387116259238205e-06, "loss": 0.9129, "step": 17749 }, { "epoch": 1.8886997233453926, "grad_norm": 12.3203125, "learning_rate": 9.387080727686187e-06, "loss": 1.1065, "step": 17750 }, { "epoch": 1.8888061289636093, "grad_norm": 10.9453125, "learning_rate": 9.387045196134168e-06, "loss": 0.9309, "step": 17751 }, { "epoch": 1.8889125345818258, "grad_norm": 16.84375, "learning_rate": 9.38700966458215e-06, "loss": 1.0204, "step": 17752 }, { "epoch": 1.8890189402000426, "grad_norm": 13.7734375, "learning_rate": 9.386974133030132e-06, "loss": 0.9502, "step": 17753 }, { "epoch": 1.8891253458182593, "grad_norm": 12.34375, "learning_rate": 9.386938601478113e-06, "loss": 1.032, "step": 17754 }, { "epoch": 1.8892317514364758, "grad_norm": 12.9375, "learning_rate": 9.386903069926095e-06, "loss": 1.0877, "step": 17755 }, { "epoch": 1.8893381570546923, "grad_norm": 12.46875, "learning_rate": 9.386867538374076e-06, "loss": 1.0523, "step": 17756 }, { "epoch": 1.8894445626729093, "grad_norm": 12.15625, "learning_rate": 9.386832006822058e-06, "loss": 0.9427, "step": 17757 }, { "epoch": 1.8895509682911258, "grad_norm": 12.03125, "learning_rate": 9.38679647527004e-06, "loss": 1.0546, "step": 17758 }, { "epoch": 1.8896573739093423, "grad_norm": 10.8203125, "learning_rate": 9.386760943718021e-06, "loss": 0.8982, "step": 17759 }, { "epoch": 1.889763779527559, "grad_norm": 12.3515625, "learning_rate": 9.386725412166005e-06, "loss": 1.0266, "step": 17760 }, { "epoch": 1.8898701851457758, "grad_norm": 17.421875, "learning_rate": 9.386689880613986e-06, "loss": 1.3576, "step": 17761 }, { "epoch": 1.8899765907639923, "grad_norm": 9.5859375, "learning_rate": 9.386654349061968e-06, "loss": 0.9894, "step": 17762 }, { "epoch": 1.890082996382209, "grad_norm": 12.328125, "learning_rate": 9.38661881750995e-06, "loss": 1.4855, "step": 17763 }, { "epoch": 1.8901894020004257, "grad_norm": 14.7890625, "learning_rate": 9.386583285957931e-06, "loss": 0.9987, "step": 17764 }, { "epoch": 1.8902958076186422, "grad_norm": 14.5078125, "learning_rate": 9.386547754405914e-06, "loss": 1.3462, "step": 17765 }, { "epoch": 1.8904022132368588, "grad_norm": 18.125, "learning_rate": 9.386512222853896e-06, "loss": 1.0492, "step": 17766 }, { "epoch": 1.8905086188550757, "grad_norm": 14.15625, "learning_rate": 9.386476691301878e-06, "loss": 1.0456, "step": 17767 }, { "epoch": 1.8906150244732922, "grad_norm": 13.3515625, "learning_rate": 9.38644115974986e-06, "loss": 1.1078, "step": 17768 }, { "epoch": 1.8907214300915087, "grad_norm": 12.7890625, "learning_rate": 9.38640562819784e-06, "loss": 0.8975, "step": 17769 }, { "epoch": 1.8908278357097255, "grad_norm": 11.4609375, "learning_rate": 9.386370096645822e-06, "loss": 0.8495, "step": 17770 }, { "epoch": 1.8909342413279422, "grad_norm": 14.8671875, "learning_rate": 9.386334565093804e-06, "loss": 0.9493, "step": 17771 }, { "epoch": 1.8910406469461587, "grad_norm": 18.375, "learning_rate": 9.386299033541786e-06, "loss": 0.7686, "step": 17772 }, { "epoch": 1.8911470525643754, "grad_norm": 14.78125, "learning_rate": 9.386263501989767e-06, "loss": 1.4518, "step": 17773 }, { "epoch": 1.8912534581825922, "grad_norm": 12.0390625, "learning_rate": 9.386227970437749e-06, "loss": 1.053, "step": 17774 }, { "epoch": 1.8913598638008087, "grad_norm": 14.1796875, "learning_rate": 9.386192438885732e-06, "loss": 1.042, "step": 17775 }, { "epoch": 1.8914662694190252, "grad_norm": 12.125, "learning_rate": 9.386156907333714e-06, "loss": 0.8933, "step": 17776 }, { "epoch": 1.891572675037242, "grad_norm": 11.6171875, "learning_rate": 9.386121375781695e-06, "loss": 0.776, "step": 17777 }, { "epoch": 1.8916790806554586, "grad_norm": 12.3203125, "learning_rate": 9.386085844229677e-06, "loss": 1.1056, "step": 17778 }, { "epoch": 1.8917854862736752, "grad_norm": 10.8515625, "learning_rate": 9.386050312677659e-06, "loss": 1.0129, "step": 17779 }, { "epoch": 1.8918918918918919, "grad_norm": 12.046875, "learning_rate": 9.38601478112564e-06, "loss": 1.0134, "step": 17780 }, { "epoch": 1.8919982975101086, "grad_norm": 14.4375, "learning_rate": 9.385979249573622e-06, "loss": 1.2618, "step": 17781 }, { "epoch": 1.8921047031283251, "grad_norm": 11.7265625, "learning_rate": 9.385943718021603e-06, "loss": 0.8362, "step": 17782 }, { "epoch": 1.8922111087465419, "grad_norm": 12.84375, "learning_rate": 9.385908186469585e-06, "loss": 1.1592, "step": 17783 }, { "epoch": 1.8923175143647586, "grad_norm": 12.15625, "learning_rate": 9.385872654917567e-06, "loss": 1.1669, "step": 17784 }, { "epoch": 1.892423919982975, "grad_norm": 13.0234375, "learning_rate": 9.38583712336555e-06, "loss": 1.2388, "step": 17785 }, { "epoch": 1.8925303256011916, "grad_norm": 16.46875, "learning_rate": 9.385801591813531e-06, "loss": 1.0826, "step": 17786 }, { "epoch": 1.8926367312194083, "grad_norm": 9.71875, "learning_rate": 9.385766060261513e-06, "loss": 1.1262, "step": 17787 }, { "epoch": 1.892743136837625, "grad_norm": 20.921875, "learning_rate": 9.385730528709495e-06, "loss": 1.169, "step": 17788 }, { "epoch": 1.8928495424558416, "grad_norm": 13.4140625, "learning_rate": 9.385694997157476e-06, "loss": 1.0807, "step": 17789 }, { "epoch": 1.8929559480740583, "grad_norm": 61.28125, "learning_rate": 9.38565946560546e-06, "loss": 0.8311, "step": 17790 }, { "epoch": 1.893062353692275, "grad_norm": 9.9921875, "learning_rate": 9.385623934053441e-06, "loss": 1.1799, "step": 17791 }, { "epoch": 1.8931687593104916, "grad_norm": 11.640625, "learning_rate": 9.385588402501423e-06, "loss": 1.1636, "step": 17792 }, { "epoch": 1.8932751649287083, "grad_norm": 9.9765625, "learning_rate": 9.385552870949404e-06, "loss": 0.8944, "step": 17793 }, { "epoch": 1.893381570546925, "grad_norm": 11.453125, "learning_rate": 9.385517339397386e-06, "loss": 1.0155, "step": 17794 }, { "epoch": 1.8934879761651415, "grad_norm": 9.921875, "learning_rate": 9.385481807845368e-06, "loss": 1.0463, "step": 17795 }, { "epoch": 1.893594381783358, "grad_norm": 13.59375, "learning_rate": 9.38544627629335e-06, "loss": 0.9008, "step": 17796 }, { "epoch": 1.8937007874015748, "grad_norm": 9.171875, "learning_rate": 9.38541074474133e-06, "loss": 1.1712, "step": 17797 }, { "epoch": 1.8938071930197915, "grad_norm": 10.65625, "learning_rate": 9.385375213189312e-06, "loss": 0.9511, "step": 17798 }, { "epoch": 1.893913598638008, "grad_norm": 22.96875, "learning_rate": 9.385339681637294e-06, "loss": 1.0067, "step": 17799 }, { "epoch": 1.8940200042562247, "grad_norm": 11.6953125, "learning_rate": 9.385304150085276e-06, "loss": 1.2315, "step": 17800 }, { "epoch": 1.8941264098744415, "grad_norm": 14.4296875, "learning_rate": 9.385268618533259e-06, "loss": 1.3682, "step": 17801 }, { "epoch": 1.894232815492658, "grad_norm": 11.5859375, "learning_rate": 9.38523308698124e-06, "loss": 1.3063, "step": 17802 }, { "epoch": 1.8943392211108745, "grad_norm": 10.875, "learning_rate": 9.385197555429222e-06, "loss": 1.0289, "step": 17803 }, { "epoch": 1.8944456267290914, "grad_norm": 10.859375, "learning_rate": 9.385162023877204e-06, "loss": 0.8758, "step": 17804 }, { "epoch": 1.894552032347308, "grad_norm": 12.453125, "learning_rate": 9.385126492325185e-06, "loss": 1.262, "step": 17805 }, { "epoch": 1.8946584379655245, "grad_norm": 12.1328125, "learning_rate": 9.385090960773167e-06, "loss": 0.9725, "step": 17806 }, { "epoch": 1.8947648435837412, "grad_norm": 12.2578125, "learning_rate": 9.385055429221149e-06, "loss": 1.3572, "step": 17807 }, { "epoch": 1.894871249201958, "grad_norm": 15.4375, "learning_rate": 9.38501989766913e-06, "loss": 1.1102, "step": 17808 }, { "epoch": 1.8949776548201744, "grad_norm": 11.2734375, "learning_rate": 9.384984366117112e-06, "loss": 1.1301, "step": 17809 }, { "epoch": 1.8950840604383912, "grad_norm": 9.2265625, "learning_rate": 9.384948834565093e-06, "loss": 0.6189, "step": 17810 }, { "epoch": 1.895190466056608, "grad_norm": 16.390625, "learning_rate": 9.384913303013077e-06, "loss": 1.2541, "step": 17811 }, { "epoch": 1.8952968716748244, "grad_norm": 11.4296875, "learning_rate": 9.384877771461058e-06, "loss": 0.9015, "step": 17812 }, { "epoch": 1.895403277293041, "grad_norm": 10.65625, "learning_rate": 9.38484223990904e-06, "loss": 1.1086, "step": 17813 }, { "epoch": 1.8955096829112579, "grad_norm": 11.2421875, "learning_rate": 9.384806708357022e-06, "loss": 0.9957, "step": 17814 }, { "epoch": 1.8956160885294744, "grad_norm": 8.875, "learning_rate": 9.384771176805003e-06, "loss": 1.0174, "step": 17815 }, { "epoch": 1.895722494147691, "grad_norm": 11.4453125, "learning_rate": 9.384735645252986e-06, "loss": 0.9149, "step": 17816 }, { "epoch": 1.8958288997659076, "grad_norm": 11.7890625, "learning_rate": 9.384700113700968e-06, "loss": 1.2738, "step": 17817 }, { "epoch": 1.8959353053841244, "grad_norm": 13.3671875, "learning_rate": 9.38466458214895e-06, "loss": 1.0311, "step": 17818 }, { "epoch": 1.8960417110023409, "grad_norm": 10.0546875, "learning_rate": 9.384629050596931e-06, "loss": 1.151, "step": 17819 }, { "epoch": 1.8961481166205576, "grad_norm": 18.3125, "learning_rate": 9.384593519044913e-06, "loss": 1.0844, "step": 17820 }, { "epoch": 1.8962545222387743, "grad_norm": 11.75, "learning_rate": 9.384557987492894e-06, "loss": 1.1391, "step": 17821 }, { "epoch": 1.8963609278569908, "grad_norm": 14.0, "learning_rate": 9.384522455940876e-06, "loss": 1.2305, "step": 17822 }, { "epoch": 1.8964673334752074, "grad_norm": 14.2890625, "learning_rate": 9.384486924388858e-06, "loss": 1.3889, "step": 17823 }, { "epoch": 1.896573739093424, "grad_norm": 11.453125, "learning_rate": 9.38445139283684e-06, "loss": 1.302, "step": 17824 }, { "epoch": 1.8966801447116408, "grad_norm": 13.8828125, "learning_rate": 9.384415861284821e-06, "loss": 0.8565, "step": 17825 }, { "epoch": 1.8967865503298573, "grad_norm": 10.0859375, "learning_rate": 9.384380329732804e-06, "loss": 1.1424, "step": 17826 }, { "epoch": 1.896892955948074, "grad_norm": 15.546875, "learning_rate": 9.384344798180786e-06, "loss": 0.9073, "step": 17827 }, { "epoch": 1.8969993615662908, "grad_norm": 14.9921875, "learning_rate": 9.384309266628767e-06, "loss": 1.3139, "step": 17828 }, { "epoch": 1.8971057671845073, "grad_norm": 11.7734375, "learning_rate": 9.384273735076749e-06, "loss": 1.0542, "step": 17829 }, { "epoch": 1.897212172802724, "grad_norm": 13.5, "learning_rate": 9.38423820352473e-06, "loss": 0.9827, "step": 17830 }, { "epoch": 1.8973185784209408, "grad_norm": 11.09375, "learning_rate": 9.384202671972712e-06, "loss": 1.1592, "step": 17831 }, { "epoch": 1.8974249840391573, "grad_norm": 13.3046875, "learning_rate": 9.384167140420694e-06, "loss": 1.0677, "step": 17832 }, { "epoch": 1.8975313896573738, "grad_norm": 9.0546875, "learning_rate": 9.384131608868675e-06, "loss": 0.8451, "step": 17833 }, { "epoch": 1.8976377952755905, "grad_norm": 10.3046875, "learning_rate": 9.384096077316657e-06, "loss": 1.0437, "step": 17834 }, { "epoch": 1.8977442008938072, "grad_norm": 10.3984375, "learning_rate": 9.384060545764639e-06, "loss": 1.26, "step": 17835 }, { "epoch": 1.8978506065120238, "grad_norm": 12.4296875, "learning_rate": 9.38402501421262e-06, "loss": 1.1335, "step": 17836 }, { "epoch": 1.8979570121302405, "grad_norm": 13.15625, "learning_rate": 9.383989482660604e-06, "loss": 1.3162, "step": 17837 }, { "epoch": 1.8980634177484572, "grad_norm": 17.234375, "learning_rate": 9.383953951108585e-06, "loss": 1.1971, "step": 17838 }, { "epoch": 1.8981698233666737, "grad_norm": 14.53125, "learning_rate": 9.383918419556567e-06, "loss": 1.0815, "step": 17839 }, { "epoch": 1.8982762289848905, "grad_norm": 10.5234375, "learning_rate": 9.383882888004548e-06, "loss": 0.8893, "step": 17840 }, { "epoch": 1.8983826346031072, "grad_norm": 12.15625, "learning_rate": 9.38384735645253e-06, "loss": 0.9397, "step": 17841 }, { "epoch": 1.8984890402213237, "grad_norm": 13.5859375, "learning_rate": 9.383811824900513e-06, "loss": 1.261, "step": 17842 }, { "epoch": 1.8985954458395402, "grad_norm": 12.1484375, "learning_rate": 9.383776293348495e-06, "loss": 1.3775, "step": 17843 }, { "epoch": 1.898701851457757, "grad_norm": 15.59375, "learning_rate": 9.383740761796477e-06, "loss": 1.0029, "step": 17844 }, { "epoch": 1.8988082570759737, "grad_norm": 9.140625, "learning_rate": 9.383705230244458e-06, "loss": 0.9465, "step": 17845 }, { "epoch": 1.8989146626941902, "grad_norm": 19.984375, "learning_rate": 9.38366969869244e-06, "loss": 1.3067, "step": 17846 }, { "epoch": 1.899021068312407, "grad_norm": 17.515625, "learning_rate": 9.383634167140421e-06, "loss": 1.0547, "step": 17847 }, { "epoch": 1.8991274739306236, "grad_norm": 10.1171875, "learning_rate": 9.383598635588403e-06, "loss": 1.1147, "step": 17848 }, { "epoch": 1.8992338795488402, "grad_norm": 13.6328125, "learning_rate": 9.383563104036385e-06, "loss": 0.9218, "step": 17849 }, { "epoch": 1.8993402851670567, "grad_norm": 11.3984375, "learning_rate": 9.383527572484366e-06, "loss": 1.2357, "step": 17850 }, { "epoch": 1.8994466907852736, "grad_norm": 12.8828125, "learning_rate": 9.383492040932348e-06, "loss": 1.1539, "step": 17851 }, { "epoch": 1.8995530964034901, "grad_norm": 12.625, "learning_rate": 9.383456509380331e-06, "loss": 0.8229, "step": 17852 }, { "epoch": 1.8996595020217066, "grad_norm": 18.71875, "learning_rate": 9.383420977828313e-06, "loss": 1.1558, "step": 17853 }, { "epoch": 1.8997659076399234, "grad_norm": 16.421875, "learning_rate": 9.383385446276294e-06, "loss": 0.9841, "step": 17854 }, { "epoch": 1.89987231325814, "grad_norm": 18.078125, "learning_rate": 9.383349914724276e-06, "loss": 0.9459, "step": 17855 }, { "epoch": 1.8999787188763566, "grad_norm": 15.0546875, "learning_rate": 9.383314383172257e-06, "loss": 1.0913, "step": 17856 }, { "epoch": 1.9000851244945733, "grad_norm": 14.34375, "learning_rate": 9.38327885162024e-06, "loss": 1.0312, "step": 17857 }, { "epoch": 1.90019153011279, "grad_norm": 13.0859375, "learning_rate": 9.383243320068222e-06, "loss": 1.0276, "step": 17858 }, { "epoch": 1.9002979357310066, "grad_norm": 9.578125, "learning_rate": 9.383207788516204e-06, "loss": 1.1326, "step": 17859 }, { "epoch": 1.900404341349223, "grad_norm": 13.84375, "learning_rate": 9.383172256964186e-06, "loss": 1.2201, "step": 17860 }, { "epoch": 1.90051074696744, "grad_norm": 9.9921875, "learning_rate": 9.383136725412166e-06, "loss": 1.063, "step": 17861 }, { "epoch": 1.9006171525856566, "grad_norm": 15.65625, "learning_rate": 9.383101193860149e-06, "loss": 1.0995, "step": 17862 }, { "epoch": 1.900723558203873, "grad_norm": 9.7578125, "learning_rate": 9.38306566230813e-06, "loss": 0.9853, "step": 17863 }, { "epoch": 1.9008299638220898, "grad_norm": 13.9375, "learning_rate": 9.383030130756112e-06, "loss": 1.008, "step": 17864 }, { "epoch": 1.9009363694403065, "grad_norm": 14.4609375, "learning_rate": 9.382994599204094e-06, "loss": 1.0641, "step": 17865 }, { "epoch": 1.901042775058523, "grad_norm": 18.40625, "learning_rate": 9.382959067652075e-06, "loss": 0.9498, "step": 17866 }, { "epoch": 1.9011491806767398, "grad_norm": 20.0625, "learning_rate": 9.382923536100059e-06, "loss": 1.1865, "step": 17867 }, { "epoch": 1.9012555862949565, "grad_norm": 11.4296875, "learning_rate": 9.38288800454804e-06, "loss": 1.0604, "step": 17868 }, { "epoch": 1.901361991913173, "grad_norm": 13.2421875, "learning_rate": 9.382852472996022e-06, "loss": 1.0342, "step": 17869 }, { "epoch": 1.9014683975313895, "grad_norm": 11.5859375, "learning_rate": 9.382816941444003e-06, "loss": 1.0743, "step": 17870 }, { "epoch": 1.9015748031496063, "grad_norm": 14.0546875, "learning_rate": 9.382781409891985e-06, "loss": 1.0337, "step": 17871 }, { "epoch": 1.901681208767823, "grad_norm": 11.53125, "learning_rate": 9.382745878339967e-06, "loss": 1.1001, "step": 17872 }, { "epoch": 1.9017876143860395, "grad_norm": 10.5078125, "learning_rate": 9.382710346787948e-06, "loss": 1.0024, "step": 17873 }, { "epoch": 1.9018940200042562, "grad_norm": 11.46875, "learning_rate": 9.38267481523593e-06, "loss": 1.0607, "step": 17874 }, { "epoch": 1.902000425622473, "grad_norm": 10.7421875, "learning_rate": 9.382639283683911e-06, "loss": 0.9749, "step": 17875 }, { "epoch": 1.9021068312406895, "grad_norm": 13.171875, "learning_rate": 9.382603752131893e-06, "loss": 0.9715, "step": 17876 }, { "epoch": 1.9022132368589062, "grad_norm": 15.4765625, "learning_rate": 9.382568220579875e-06, "loss": 1.0616, "step": 17877 }, { "epoch": 1.902319642477123, "grad_norm": 12.828125, "learning_rate": 9.382532689027858e-06, "loss": 1.2644, "step": 17878 }, { "epoch": 1.9024260480953394, "grad_norm": 12.578125, "learning_rate": 9.38249715747584e-06, "loss": 1.1775, "step": 17879 }, { "epoch": 1.902532453713556, "grad_norm": 14.578125, "learning_rate": 9.382461625923821e-06, "loss": 0.9461, "step": 17880 }, { "epoch": 1.9026388593317727, "grad_norm": 11.1171875, "learning_rate": 9.382426094371803e-06, "loss": 1.1601, "step": 17881 }, { "epoch": 1.9027452649499894, "grad_norm": 16.734375, "learning_rate": 9.382390562819784e-06, "loss": 1.1424, "step": 17882 }, { "epoch": 1.902851670568206, "grad_norm": 16.0625, "learning_rate": 9.382355031267768e-06, "loss": 1.394, "step": 17883 }, { "epoch": 1.9029580761864227, "grad_norm": 11.0703125, "learning_rate": 9.38231949971575e-06, "loss": 0.9652, "step": 17884 }, { "epoch": 1.9030644818046394, "grad_norm": 15.2578125, "learning_rate": 9.382283968163731e-06, "loss": 1.0764, "step": 17885 }, { "epoch": 1.903170887422856, "grad_norm": 10.53125, "learning_rate": 9.382248436611712e-06, "loss": 0.9728, "step": 17886 }, { "epoch": 1.9032772930410724, "grad_norm": 12.203125, "learning_rate": 9.382212905059694e-06, "loss": 1.0278, "step": 17887 }, { "epoch": 1.9033836986592894, "grad_norm": 11.9609375, "learning_rate": 9.382177373507676e-06, "loss": 1.1429, "step": 17888 }, { "epoch": 1.9034901042775059, "grad_norm": 10.4375, "learning_rate": 9.382141841955657e-06, "loss": 0.9994, "step": 17889 }, { "epoch": 1.9035965098957224, "grad_norm": 14.2421875, "learning_rate": 9.382106310403639e-06, "loss": 1.232, "step": 17890 }, { "epoch": 1.903702915513939, "grad_norm": 15.8828125, "learning_rate": 9.38207077885162e-06, "loss": 1.256, "step": 17891 }, { "epoch": 1.9038093211321558, "grad_norm": 13.9296875, "learning_rate": 9.382035247299602e-06, "loss": 1.0589, "step": 17892 }, { "epoch": 1.9039157267503724, "grad_norm": 11.78125, "learning_rate": 9.381999715747585e-06, "loss": 1.3438, "step": 17893 }, { "epoch": 1.904022132368589, "grad_norm": 11.671875, "learning_rate": 9.381964184195567e-06, "loss": 1.2734, "step": 17894 }, { "epoch": 1.9041285379868058, "grad_norm": 14.4921875, "learning_rate": 9.381928652643549e-06, "loss": 1.3043, "step": 17895 }, { "epoch": 1.9042349436050223, "grad_norm": 15.3203125, "learning_rate": 9.38189312109153e-06, "loss": 1.1459, "step": 17896 }, { "epoch": 1.9043413492232388, "grad_norm": 11.15625, "learning_rate": 9.381857589539512e-06, "loss": 1.2402, "step": 17897 }, { "epoch": 1.9044477548414558, "grad_norm": 20.953125, "learning_rate": 9.381822057987493e-06, "loss": 1.1302, "step": 17898 }, { "epoch": 1.9045541604596723, "grad_norm": 11.890625, "learning_rate": 9.381786526435475e-06, "loss": 0.962, "step": 17899 }, { "epoch": 1.9046605660778888, "grad_norm": 14.71875, "learning_rate": 9.381750994883457e-06, "loss": 1.1198, "step": 17900 }, { "epoch": 1.9047669716961055, "grad_norm": 10.0703125, "learning_rate": 9.381715463331438e-06, "loss": 0.914, "step": 17901 }, { "epoch": 1.9048733773143223, "grad_norm": 13.3203125, "learning_rate": 9.38167993177942e-06, "loss": 1.3998, "step": 17902 }, { "epoch": 1.9049797829325388, "grad_norm": 12.984375, "learning_rate": 9.381644400227403e-06, "loss": 1.1887, "step": 17903 }, { "epoch": 1.9050861885507555, "grad_norm": 10.4765625, "learning_rate": 9.381608868675385e-06, "loss": 1.007, "step": 17904 }, { "epoch": 1.9051925941689722, "grad_norm": 11.1875, "learning_rate": 9.381573337123366e-06, "loss": 1.0469, "step": 17905 }, { "epoch": 1.9052989997871888, "grad_norm": 16.78125, "learning_rate": 9.381537805571348e-06, "loss": 1.3677, "step": 17906 }, { "epoch": 1.9054054054054053, "grad_norm": 17.515625, "learning_rate": 9.38150227401933e-06, "loss": 0.9652, "step": 17907 }, { "epoch": 1.905511811023622, "grad_norm": 13.28125, "learning_rate": 9.381466742467313e-06, "loss": 1.1231, "step": 17908 }, { "epoch": 1.9056182166418387, "grad_norm": 16.890625, "learning_rate": 9.381431210915294e-06, "loss": 0.996, "step": 17909 }, { "epoch": 1.9057246222600552, "grad_norm": 14.9375, "learning_rate": 9.381395679363276e-06, "loss": 1.0691, "step": 17910 }, { "epoch": 1.905831027878272, "grad_norm": 14.3203125, "learning_rate": 9.381360147811258e-06, "loss": 0.8956, "step": 17911 }, { "epoch": 1.9059374334964887, "grad_norm": 10.46875, "learning_rate": 9.38132461625924e-06, "loss": 1.0788, "step": 17912 }, { "epoch": 1.9060438391147052, "grad_norm": 14.5859375, "learning_rate": 9.381289084707221e-06, "loss": 0.9417, "step": 17913 }, { "epoch": 1.906150244732922, "grad_norm": 11.8984375, "learning_rate": 9.381253553155203e-06, "loss": 0.9384, "step": 17914 }, { "epoch": 1.9062566503511387, "grad_norm": 12.4296875, "learning_rate": 9.381218021603184e-06, "loss": 1.3348, "step": 17915 }, { "epoch": 1.9063630559693552, "grad_norm": 12.8828125, "learning_rate": 9.381182490051166e-06, "loss": 1.1759, "step": 17916 }, { "epoch": 1.9064694615875717, "grad_norm": 10.953125, "learning_rate": 9.381146958499147e-06, "loss": 1.1124, "step": 17917 }, { "epoch": 1.9065758672057884, "grad_norm": 13.5703125, "learning_rate": 9.381111426947129e-06, "loss": 1.0234, "step": 17918 }, { "epoch": 1.9066822728240052, "grad_norm": 16.953125, "learning_rate": 9.381075895395112e-06, "loss": 0.8806, "step": 17919 }, { "epoch": 1.9067886784422217, "grad_norm": 24.21875, "learning_rate": 9.381040363843094e-06, "loss": 1.2787, "step": 17920 }, { "epoch": 1.9068950840604384, "grad_norm": 12.8125, "learning_rate": 9.381004832291075e-06, "loss": 1.3038, "step": 17921 }, { "epoch": 1.9070014896786551, "grad_norm": 14.7578125, "learning_rate": 9.380969300739057e-06, "loss": 1.0841, "step": 17922 }, { "epoch": 1.9071078952968716, "grad_norm": 10.8828125, "learning_rate": 9.380933769187039e-06, "loss": 1.1944, "step": 17923 }, { "epoch": 1.9072143009150884, "grad_norm": 12.59375, "learning_rate": 9.38089823763502e-06, "loss": 1.4763, "step": 17924 }, { "epoch": 1.907320706533305, "grad_norm": 11.2109375, "learning_rate": 9.380862706083002e-06, "loss": 0.997, "step": 17925 }, { "epoch": 1.9074271121515216, "grad_norm": 9.8359375, "learning_rate": 9.380827174530983e-06, "loss": 0.7463, "step": 17926 }, { "epoch": 1.9075335177697381, "grad_norm": 11.9140625, "learning_rate": 9.380791642978965e-06, "loss": 1.1929, "step": 17927 }, { "epoch": 1.9076399233879549, "grad_norm": 11.046875, "learning_rate": 9.380756111426947e-06, "loss": 1.0961, "step": 17928 }, { "epoch": 1.9077463290061716, "grad_norm": 15.640625, "learning_rate": 9.38072057987493e-06, "loss": 1.2797, "step": 17929 }, { "epoch": 1.907852734624388, "grad_norm": 11.765625, "learning_rate": 9.380685048322912e-06, "loss": 1.1963, "step": 17930 }, { "epoch": 1.9079591402426048, "grad_norm": 12.0, "learning_rate": 9.380649516770893e-06, "loss": 1.2222, "step": 17931 }, { "epoch": 1.9080655458608216, "grad_norm": 15.296875, "learning_rate": 9.380613985218875e-06, "loss": 1.0817, "step": 17932 }, { "epoch": 1.908171951479038, "grad_norm": 13.828125, "learning_rate": 9.380578453666856e-06, "loss": 1.3711, "step": 17933 }, { "epoch": 1.9082783570972546, "grad_norm": 13.6640625, "learning_rate": 9.38054292211484e-06, "loss": 1.1037, "step": 17934 }, { "epoch": 1.9083847627154715, "grad_norm": 10.765625, "learning_rate": 9.380507390562821e-06, "loss": 1.0702, "step": 17935 }, { "epoch": 1.908491168333688, "grad_norm": 13.453125, "learning_rate": 9.380471859010803e-06, "loss": 0.9731, "step": 17936 }, { "epoch": 1.9085975739519045, "grad_norm": 11.7109375, "learning_rate": 9.380436327458785e-06, "loss": 1.0467, "step": 17937 }, { "epoch": 1.9087039795701213, "grad_norm": 11.0703125, "learning_rate": 9.380400795906766e-06, "loss": 0.949, "step": 17938 }, { "epoch": 1.908810385188338, "grad_norm": 14.2734375, "learning_rate": 9.380365264354748e-06, "loss": 0.8687, "step": 17939 }, { "epoch": 1.9089167908065545, "grad_norm": 12.3359375, "learning_rate": 9.38032973280273e-06, "loss": 0.9879, "step": 17940 }, { "epoch": 1.9090231964247713, "grad_norm": 15.5390625, "learning_rate": 9.380294201250711e-06, "loss": 1.1681, "step": 17941 }, { "epoch": 1.909129602042988, "grad_norm": 12.1875, "learning_rate": 9.380258669698693e-06, "loss": 0.9486, "step": 17942 }, { "epoch": 1.9092360076612045, "grad_norm": 11.9921875, "learning_rate": 9.380223138146674e-06, "loss": 1.1305, "step": 17943 }, { "epoch": 1.909342413279421, "grad_norm": 11.9921875, "learning_rate": 9.380187606594658e-06, "loss": 1.0214, "step": 17944 }, { "epoch": 1.909448818897638, "grad_norm": 12.6875, "learning_rate": 9.380152075042639e-06, "loss": 1.0899, "step": 17945 }, { "epoch": 1.9095552245158545, "grad_norm": 14.6171875, "learning_rate": 9.38011654349062e-06, "loss": 1.0484, "step": 17946 }, { "epoch": 1.909661630134071, "grad_norm": 11.84375, "learning_rate": 9.380081011938602e-06, "loss": 0.9986, "step": 17947 }, { "epoch": 1.9097680357522877, "grad_norm": 9.8046875, "learning_rate": 9.380045480386584e-06, "loss": 0.9529, "step": 17948 }, { "epoch": 1.9098744413705044, "grad_norm": 12.8515625, "learning_rate": 9.380009948834567e-06, "loss": 1.1717, "step": 17949 }, { "epoch": 1.909980846988721, "grad_norm": 12.1484375, "learning_rate": 9.379974417282547e-06, "loss": 1.174, "step": 17950 }, { "epoch": 1.9100872526069377, "grad_norm": 13.296875, "learning_rate": 9.379938885730529e-06, "loss": 1.264, "step": 17951 }, { "epoch": 1.9101936582251544, "grad_norm": 11.21875, "learning_rate": 9.37990335417851e-06, "loss": 0.759, "step": 17952 }, { "epoch": 1.910300063843371, "grad_norm": 15.625, "learning_rate": 9.379867822626492e-06, "loss": 0.9398, "step": 17953 }, { "epoch": 1.9104064694615874, "grad_norm": 10.7734375, "learning_rate": 9.379832291074474e-06, "loss": 1.2251, "step": 17954 }, { "epoch": 1.9105128750798042, "grad_norm": 14.3125, "learning_rate": 9.379796759522457e-06, "loss": 1.1648, "step": 17955 }, { "epoch": 1.910619280698021, "grad_norm": 21.015625, "learning_rate": 9.379761227970438e-06, "loss": 1.3986, "step": 17956 }, { "epoch": 1.9107256863162374, "grad_norm": 12.7578125, "learning_rate": 9.37972569641842e-06, "loss": 1.1446, "step": 17957 }, { "epoch": 1.9108320919344541, "grad_norm": 17.03125, "learning_rate": 9.379690164866402e-06, "loss": 1.1054, "step": 17958 }, { "epoch": 1.9109384975526709, "grad_norm": 11.3125, "learning_rate": 9.379654633314383e-06, "loss": 1.1377, "step": 17959 }, { "epoch": 1.9110449031708874, "grad_norm": 15.3671875, "learning_rate": 9.379619101762367e-06, "loss": 0.9844, "step": 17960 }, { "epoch": 1.911151308789104, "grad_norm": 10.6328125, "learning_rate": 9.379583570210348e-06, "loss": 1.0644, "step": 17961 }, { "epoch": 1.9112577144073208, "grad_norm": 10.125, "learning_rate": 9.37954803865833e-06, "loss": 1.014, "step": 17962 }, { "epoch": 1.9113641200255373, "grad_norm": 10.265625, "learning_rate": 9.379512507106311e-06, "loss": 1.0116, "step": 17963 }, { "epoch": 1.9114705256437539, "grad_norm": 11.796875, "learning_rate": 9.379476975554293e-06, "loss": 0.9531, "step": 17964 }, { "epoch": 1.9115769312619706, "grad_norm": 12.921875, "learning_rate": 9.379441444002275e-06, "loss": 1.2138, "step": 17965 }, { "epoch": 1.9116833368801873, "grad_norm": 11.0078125, "learning_rate": 9.379405912450256e-06, "loss": 1.1112, "step": 17966 }, { "epoch": 1.9117897424984038, "grad_norm": 12.4921875, "learning_rate": 9.379370380898238e-06, "loss": 1.2615, "step": 17967 }, { "epoch": 1.9118961481166206, "grad_norm": 11.28125, "learning_rate": 9.37933484934622e-06, "loss": 1.3466, "step": 17968 }, { "epoch": 1.9120025537348373, "grad_norm": 13.1171875, "learning_rate": 9.379299317794201e-06, "loss": 0.9816, "step": 17969 }, { "epoch": 1.9121089593530538, "grad_norm": 12.8515625, "learning_rate": 9.379263786242184e-06, "loss": 1.2427, "step": 17970 }, { "epoch": 1.9122153649712705, "grad_norm": 10.5234375, "learning_rate": 9.379228254690166e-06, "loss": 0.9518, "step": 17971 }, { "epoch": 1.9123217705894873, "grad_norm": 13.046875, "learning_rate": 9.379192723138148e-06, "loss": 1.1981, "step": 17972 }, { "epoch": 1.9124281762077038, "grad_norm": 12.828125, "learning_rate": 9.37915719158613e-06, "loss": 0.9693, "step": 17973 }, { "epoch": 1.9125345818259203, "grad_norm": 15.7890625, "learning_rate": 9.37912166003411e-06, "loss": 1.0737, "step": 17974 }, { "epoch": 1.912640987444137, "grad_norm": 13.2890625, "learning_rate": 9.379086128482094e-06, "loss": 1.1404, "step": 17975 }, { "epoch": 1.9127473930623538, "grad_norm": 15.8515625, "learning_rate": 9.379050596930076e-06, "loss": 0.9857, "step": 17976 }, { "epoch": 1.9128537986805703, "grad_norm": 11.3125, "learning_rate": 9.379015065378057e-06, "loss": 0.9135, "step": 17977 }, { "epoch": 1.912960204298787, "grad_norm": 13.5625, "learning_rate": 9.378979533826039e-06, "loss": 1.147, "step": 17978 }, { "epoch": 1.9130666099170037, "grad_norm": 10.4296875, "learning_rate": 9.37894400227402e-06, "loss": 1.0909, "step": 17979 }, { "epoch": 1.9131730155352202, "grad_norm": 12.84375, "learning_rate": 9.378908470722002e-06, "loss": 1.0721, "step": 17980 }, { "epoch": 1.9132794211534367, "grad_norm": 10.3046875, "learning_rate": 9.378872939169984e-06, "loss": 0.8825, "step": 17981 }, { "epoch": 1.9133858267716537, "grad_norm": 10.7578125, "learning_rate": 9.378837407617965e-06, "loss": 1.0495, "step": 17982 }, { "epoch": 1.9134922323898702, "grad_norm": 15.046875, "learning_rate": 9.378801876065947e-06, "loss": 1.2319, "step": 17983 }, { "epoch": 1.9135986380080867, "grad_norm": 16.421875, "learning_rate": 9.378766344513929e-06, "loss": 1.0782, "step": 17984 }, { "epoch": 1.9137050436263034, "grad_norm": 12.9765625, "learning_rate": 9.378730812961912e-06, "loss": 0.8663, "step": 17985 }, { "epoch": 1.9138114492445202, "grad_norm": 11.890625, "learning_rate": 9.378695281409893e-06, "loss": 1.2126, "step": 17986 }, { "epoch": 1.9139178548627367, "grad_norm": 22.25, "learning_rate": 9.378659749857875e-06, "loss": 1.2418, "step": 17987 }, { "epoch": 1.9140242604809534, "grad_norm": 11.21875, "learning_rate": 9.378624218305857e-06, "loss": 1.0675, "step": 17988 }, { "epoch": 1.9141306660991702, "grad_norm": 11.9921875, "learning_rate": 9.378588686753838e-06, "loss": 1.0945, "step": 17989 }, { "epoch": 1.9142370717173867, "grad_norm": 15.21875, "learning_rate": 9.37855315520182e-06, "loss": 0.8727, "step": 17990 }, { "epoch": 1.9143434773356032, "grad_norm": 10.828125, "learning_rate": 9.378517623649801e-06, "loss": 0.989, "step": 17991 }, { "epoch": 1.9144498829538201, "grad_norm": 11.53125, "learning_rate": 9.378482092097783e-06, "loss": 0.981, "step": 17992 }, { "epoch": 1.9145562885720366, "grad_norm": 10.7109375, "learning_rate": 9.378446560545765e-06, "loss": 0.9512, "step": 17993 }, { "epoch": 1.9146626941902531, "grad_norm": 15.9453125, "learning_rate": 9.378411028993746e-06, "loss": 1.2109, "step": 17994 }, { "epoch": 1.9147690998084699, "grad_norm": 9.9921875, "learning_rate": 9.378375497441728e-06, "loss": 1.0532, "step": 17995 }, { "epoch": 1.9148755054266866, "grad_norm": 9.578125, "learning_rate": 9.378339965889711e-06, "loss": 1.233, "step": 17996 }, { "epoch": 1.9149819110449031, "grad_norm": 15.140625, "learning_rate": 9.378304434337693e-06, "loss": 0.8686, "step": 17997 }, { "epoch": 1.9150883166631198, "grad_norm": 11.8046875, "learning_rate": 9.378268902785674e-06, "loss": 1.0558, "step": 17998 }, { "epoch": 1.9151947222813366, "grad_norm": 14.1875, "learning_rate": 9.378233371233656e-06, "loss": 1.0906, "step": 17999 }, { "epoch": 1.915301127899553, "grad_norm": 20.390625, "learning_rate": 9.378197839681638e-06, "loss": 1.025, "step": 18000 }, { "epoch": 1.9154075335177696, "grad_norm": 10.3203125, "learning_rate": 9.378162308129621e-06, "loss": 0.9566, "step": 18001 }, { "epoch": 1.9155139391359863, "grad_norm": 10.0, "learning_rate": 9.378126776577603e-06, "loss": 0.9911, "step": 18002 }, { "epoch": 1.915620344754203, "grad_norm": 19.9375, "learning_rate": 9.378091245025584e-06, "loss": 1.0294, "step": 18003 }, { "epoch": 1.9157267503724196, "grad_norm": 11.8515625, "learning_rate": 9.378055713473566e-06, "loss": 0.9788, "step": 18004 }, { "epoch": 1.9158331559906363, "grad_norm": 14.8515625, "learning_rate": 9.378020181921547e-06, "loss": 1.0968, "step": 18005 }, { "epoch": 1.915939561608853, "grad_norm": 16.125, "learning_rate": 9.377984650369529e-06, "loss": 1.1517, "step": 18006 }, { "epoch": 1.9160459672270695, "grad_norm": 15.0, "learning_rate": 9.37794911881751e-06, "loss": 1.309, "step": 18007 }, { "epoch": 1.9161523728452863, "grad_norm": 14.3046875, "learning_rate": 9.377913587265492e-06, "loss": 1.4294, "step": 18008 }, { "epoch": 1.916258778463503, "grad_norm": 10.9921875, "learning_rate": 9.377878055713474e-06, "loss": 1.2055, "step": 18009 }, { "epoch": 1.9163651840817195, "grad_norm": 12.328125, "learning_rate": 9.377842524161455e-06, "loss": 1.0063, "step": 18010 }, { "epoch": 1.916471589699936, "grad_norm": 15.25, "learning_rate": 9.377806992609439e-06, "loss": 0.9307, "step": 18011 }, { "epoch": 1.9165779953181528, "grad_norm": 13.3671875, "learning_rate": 9.37777146105742e-06, "loss": 1.1113, "step": 18012 }, { "epoch": 1.9166844009363695, "grad_norm": 11.7421875, "learning_rate": 9.377735929505402e-06, "loss": 1.2664, "step": 18013 }, { "epoch": 1.916790806554586, "grad_norm": 10.9921875, "learning_rate": 9.377700397953384e-06, "loss": 1.1248, "step": 18014 }, { "epoch": 1.9168972121728027, "grad_norm": 17.390625, "learning_rate": 9.377664866401365e-06, "loss": 0.9374, "step": 18015 }, { "epoch": 1.9170036177910195, "grad_norm": 12.3828125, "learning_rate": 9.377629334849347e-06, "loss": 0.915, "step": 18016 }, { "epoch": 1.917110023409236, "grad_norm": 13.421875, "learning_rate": 9.377593803297328e-06, "loss": 1.0049, "step": 18017 }, { "epoch": 1.9172164290274525, "grad_norm": 13.09375, "learning_rate": 9.37755827174531e-06, "loss": 1.2659, "step": 18018 }, { "epoch": 1.9173228346456694, "grad_norm": 16.984375, "learning_rate": 9.377522740193292e-06, "loss": 1.099, "step": 18019 }, { "epoch": 1.917429240263886, "grad_norm": 13.125, "learning_rate": 9.377487208641273e-06, "loss": 1.1168, "step": 18020 }, { "epoch": 1.9175356458821025, "grad_norm": 11.8828125, "learning_rate": 9.377451677089256e-06, "loss": 0.9185, "step": 18021 }, { "epoch": 1.9176420515003192, "grad_norm": 18.34375, "learning_rate": 9.377416145537238e-06, "loss": 1.3195, "step": 18022 }, { "epoch": 1.917748457118536, "grad_norm": 14.2109375, "learning_rate": 9.37738061398522e-06, "loss": 1.2667, "step": 18023 }, { "epoch": 1.9178548627367524, "grad_norm": 16.59375, "learning_rate": 9.377345082433201e-06, "loss": 1.0562, "step": 18024 }, { "epoch": 1.9179612683549692, "grad_norm": 13.1953125, "learning_rate": 9.377309550881183e-06, "loss": 1.0393, "step": 18025 }, { "epoch": 1.918067673973186, "grad_norm": 14.3203125, "learning_rate": 9.377274019329166e-06, "loss": 1.1247, "step": 18026 }, { "epoch": 1.9181740795914024, "grad_norm": 12.0703125, "learning_rate": 9.377238487777148e-06, "loss": 1.0209, "step": 18027 }, { "epoch": 1.918280485209619, "grad_norm": 12.4921875, "learning_rate": 9.37720295622513e-06, "loss": 0.9879, "step": 18028 }, { "epoch": 1.9183868908278359, "grad_norm": 9.8828125, "learning_rate": 9.377167424673111e-06, "loss": 1.1997, "step": 18029 }, { "epoch": 1.9184932964460524, "grad_norm": 16.0625, "learning_rate": 9.377131893121093e-06, "loss": 1.0848, "step": 18030 }, { "epoch": 1.9185997020642689, "grad_norm": 17.546875, "learning_rate": 9.377096361569074e-06, "loss": 1.0435, "step": 18031 }, { "epoch": 1.9187061076824856, "grad_norm": 12.109375, "learning_rate": 9.377060830017056e-06, "loss": 1.1644, "step": 18032 }, { "epoch": 1.9188125133007023, "grad_norm": 13.0, "learning_rate": 9.377025298465037e-06, "loss": 1.0336, "step": 18033 }, { "epoch": 1.9189189189189189, "grad_norm": 20.46875, "learning_rate": 9.376989766913019e-06, "loss": 1.2136, "step": 18034 }, { "epoch": 1.9190253245371356, "grad_norm": 13.953125, "learning_rate": 9.376954235361e-06, "loss": 1.1204, "step": 18035 }, { "epoch": 1.9191317301553523, "grad_norm": 14.453125, "learning_rate": 9.376918703808982e-06, "loss": 1.0495, "step": 18036 }, { "epoch": 1.9192381357735688, "grad_norm": 15.53125, "learning_rate": 9.376883172256966e-06, "loss": 0.9935, "step": 18037 }, { "epoch": 1.9193445413917853, "grad_norm": 14.1796875, "learning_rate": 9.376847640704947e-06, "loss": 0.9669, "step": 18038 }, { "epoch": 1.919450947010002, "grad_norm": 15.5078125, "learning_rate": 9.376812109152929e-06, "loss": 1.1542, "step": 18039 }, { "epoch": 1.9195573526282188, "grad_norm": 11.484375, "learning_rate": 9.37677657760091e-06, "loss": 1.0491, "step": 18040 }, { "epoch": 1.9196637582464353, "grad_norm": 10.484375, "learning_rate": 9.376741046048892e-06, "loss": 1.1238, "step": 18041 }, { "epoch": 1.919770163864652, "grad_norm": 14.109375, "learning_rate": 9.376705514496874e-06, "loss": 1.1889, "step": 18042 }, { "epoch": 1.9198765694828688, "grad_norm": 12.1015625, "learning_rate": 9.376669982944855e-06, "loss": 1.127, "step": 18043 }, { "epoch": 1.9199829751010853, "grad_norm": 13.1953125, "learning_rate": 9.376634451392837e-06, "loss": 1.1237, "step": 18044 }, { "epoch": 1.920089380719302, "grad_norm": 13.6171875, "learning_rate": 9.376598919840818e-06, "loss": 1.0351, "step": 18045 }, { "epoch": 1.9201957863375188, "grad_norm": 11.4765625, "learning_rate": 9.3765633882888e-06, "loss": 1.0097, "step": 18046 }, { "epoch": 1.9203021919557353, "grad_norm": 15.9765625, "learning_rate": 9.376527856736783e-06, "loss": 1.0545, "step": 18047 }, { "epoch": 1.9204085975739518, "grad_norm": 14.4765625, "learning_rate": 9.376492325184765e-06, "loss": 1.2625, "step": 18048 }, { "epoch": 1.9205150031921685, "grad_norm": 14.3515625, "learning_rate": 9.376456793632747e-06, "loss": 1.0213, "step": 18049 }, { "epoch": 1.9206214088103852, "grad_norm": 13.1015625, "learning_rate": 9.376421262080728e-06, "loss": 1.2044, "step": 18050 }, { "epoch": 1.9207278144286017, "grad_norm": 15.859375, "learning_rate": 9.37638573052871e-06, "loss": 0.9702, "step": 18051 }, { "epoch": 1.9208342200468185, "grad_norm": 10.7109375, "learning_rate": 9.376350198976693e-06, "loss": 1.1257, "step": 18052 }, { "epoch": 1.9209406256650352, "grad_norm": 13.0703125, "learning_rate": 9.376314667424675e-06, "loss": 0.964, "step": 18053 }, { "epoch": 1.9210470312832517, "grad_norm": 12.7578125, "learning_rate": 9.376279135872656e-06, "loss": 1.0788, "step": 18054 }, { "epoch": 1.9211534369014684, "grad_norm": 12.6328125, "learning_rate": 9.376243604320638e-06, "loss": 1.0686, "step": 18055 }, { "epoch": 1.9212598425196852, "grad_norm": 15.265625, "learning_rate": 9.37620807276862e-06, "loss": 1.1568, "step": 18056 }, { "epoch": 1.9213662481379017, "grad_norm": 10.3359375, "learning_rate": 9.376172541216601e-06, "loss": 1.0586, "step": 18057 }, { "epoch": 1.9214726537561182, "grad_norm": 12.2578125, "learning_rate": 9.376137009664583e-06, "loss": 1.0575, "step": 18058 }, { "epoch": 1.921579059374335, "grad_norm": 11.421875, "learning_rate": 9.376101478112564e-06, "loss": 0.9506, "step": 18059 }, { "epoch": 1.9216854649925517, "grad_norm": 11.40625, "learning_rate": 9.376065946560546e-06, "loss": 1.2091, "step": 18060 }, { "epoch": 1.9217918706107682, "grad_norm": 11.640625, "learning_rate": 9.376030415008528e-06, "loss": 0.9612, "step": 18061 }, { "epoch": 1.921898276228985, "grad_norm": 15.5, "learning_rate": 9.37599488345651e-06, "loss": 1.0105, "step": 18062 }, { "epoch": 1.9220046818472016, "grad_norm": 9.515625, "learning_rate": 9.375959351904492e-06, "loss": 0.9974, "step": 18063 }, { "epoch": 1.9221110874654181, "grad_norm": 10.265625, "learning_rate": 9.375923820352474e-06, "loss": 1.2693, "step": 18064 }, { "epoch": 1.9222174930836347, "grad_norm": 20.4375, "learning_rate": 9.375888288800456e-06, "loss": 1.0827, "step": 18065 }, { "epoch": 1.9223238987018516, "grad_norm": 12.84375, "learning_rate": 9.375852757248437e-06, "loss": 1.0623, "step": 18066 }, { "epoch": 1.9224303043200681, "grad_norm": 11.5703125, "learning_rate": 9.37581722569642e-06, "loss": 1.0931, "step": 18067 }, { "epoch": 1.9225367099382846, "grad_norm": 11.296875, "learning_rate": 9.375781694144402e-06, "loss": 1.2646, "step": 18068 }, { "epoch": 1.9226431155565014, "grad_norm": 14.546875, "learning_rate": 9.375746162592384e-06, "loss": 1.0774, "step": 18069 }, { "epoch": 1.922749521174718, "grad_norm": 10.5, "learning_rate": 9.375710631040364e-06, "loss": 1.2137, "step": 18070 }, { "epoch": 1.9228559267929346, "grad_norm": 16.625, "learning_rate": 9.375675099488345e-06, "loss": 0.9436, "step": 18071 }, { "epoch": 1.9229623324111513, "grad_norm": 14.875, "learning_rate": 9.375639567936327e-06, "loss": 1.169, "step": 18072 }, { "epoch": 1.923068738029368, "grad_norm": 11.21875, "learning_rate": 9.37560403638431e-06, "loss": 0.9872, "step": 18073 }, { "epoch": 1.9231751436475846, "grad_norm": 10.2109375, "learning_rate": 9.375568504832292e-06, "loss": 1.0039, "step": 18074 }, { "epoch": 1.923281549265801, "grad_norm": 21.34375, "learning_rate": 9.375532973280273e-06, "loss": 1.3223, "step": 18075 }, { "epoch": 1.923387954884018, "grad_norm": 14.0546875, "learning_rate": 9.375497441728255e-06, "loss": 1.154, "step": 18076 }, { "epoch": 1.9234943605022345, "grad_norm": 17.53125, "learning_rate": 9.375461910176237e-06, "loss": 1.2936, "step": 18077 }, { "epoch": 1.923600766120451, "grad_norm": 14.875, "learning_rate": 9.37542637862422e-06, "loss": 1.3396, "step": 18078 }, { "epoch": 1.9237071717386678, "grad_norm": 10.859375, "learning_rate": 9.375390847072202e-06, "loss": 0.9892, "step": 18079 }, { "epoch": 1.9238135773568845, "grad_norm": 13.5859375, "learning_rate": 9.375355315520183e-06, "loss": 0.9892, "step": 18080 }, { "epoch": 1.923919982975101, "grad_norm": 16.765625, "learning_rate": 9.375319783968165e-06, "loss": 0.8661, "step": 18081 }, { "epoch": 1.9240263885933178, "grad_norm": 13.2265625, "learning_rate": 9.375284252416146e-06, "loss": 1.1056, "step": 18082 }, { "epoch": 1.9241327942115345, "grad_norm": 12.875, "learning_rate": 9.375248720864128e-06, "loss": 0.9938, "step": 18083 }, { "epoch": 1.924239199829751, "grad_norm": 12.5234375, "learning_rate": 9.37521318931211e-06, "loss": 1.0759, "step": 18084 }, { "epoch": 1.9243456054479675, "grad_norm": 20.125, "learning_rate": 9.375177657760091e-06, "loss": 1.1288, "step": 18085 }, { "epoch": 1.9244520110661842, "grad_norm": 12.484375, "learning_rate": 9.375142126208073e-06, "loss": 1.0966, "step": 18086 }, { "epoch": 1.924558416684401, "grad_norm": 10.4453125, "learning_rate": 9.375106594656054e-06, "loss": 0.8563, "step": 18087 }, { "epoch": 1.9246648223026175, "grad_norm": 11.2421875, "learning_rate": 9.375071063104038e-06, "loss": 0.998, "step": 18088 }, { "epoch": 1.9247712279208342, "grad_norm": 14.3984375, "learning_rate": 9.37503553155202e-06, "loss": 1.2326, "step": 18089 }, { "epoch": 1.924877633539051, "grad_norm": 17.359375, "learning_rate": 9.375000000000001e-06, "loss": 1.2083, "step": 18090 }, { "epoch": 1.9249840391572675, "grad_norm": 9.8359375, "learning_rate": 9.374964468447982e-06, "loss": 1.1819, "step": 18091 }, { "epoch": 1.9250904447754842, "grad_norm": 12.328125, "learning_rate": 9.374928936895964e-06, "loss": 1.2485, "step": 18092 }, { "epoch": 1.925196850393701, "grad_norm": 13.3828125, "learning_rate": 9.374893405343947e-06, "loss": 1.0592, "step": 18093 }, { "epoch": 1.9253032560119174, "grad_norm": 12.6171875, "learning_rate": 9.374857873791929e-06, "loss": 0.9374, "step": 18094 }, { "epoch": 1.925409661630134, "grad_norm": 11.359375, "learning_rate": 9.37482234223991e-06, "loss": 1.215, "step": 18095 }, { "epoch": 1.9255160672483507, "grad_norm": 12.0, "learning_rate": 9.374786810687892e-06, "loss": 0.9783, "step": 18096 }, { "epoch": 1.9256224728665674, "grad_norm": 9.765625, "learning_rate": 9.374751279135874e-06, "loss": 0.9922, "step": 18097 }, { "epoch": 1.925728878484784, "grad_norm": 13.0859375, "learning_rate": 9.374715747583855e-06, "loss": 1.0323, "step": 18098 }, { "epoch": 1.9258352841030006, "grad_norm": 11.640625, "learning_rate": 9.374680216031837e-06, "loss": 1.003, "step": 18099 }, { "epoch": 1.9259416897212174, "grad_norm": 12.7578125, "learning_rate": 9.374644684479819e-06, "loss": 1.0535, "step": 18100 }, { "epoch": 1.9260480953394339, "grad_norm": 12.15625, "learning_rate": 9.3746091529278e-06, "loss": 1.343, "step": 18101 }, { "epoch": 1.9261545009576506, "grad_norm": 11.015625, "learning_rate": 9.374573621375782e-06, "loss": 0.9474, "step": 18102 }, { "epoch": 1.9262609065758673, "grad_norm": 11.453125, "learning_rate": 9.374538089823765e-06, "loss": 0.9252, "step": 18103 }, { "epoch": 1.9263673121940839, "grad_norm": 11.1796875, "learning_rate": 9.374502558271747e-06, "loss": 1.0972, "step": 18104 }, { "epoch": 1.9264737178123004, "grad_norm": 12.40625, "learning_rate": 9.374467026719728e-06, "loss": 1.5378, "step": 18105 }, { "epoch": 1.926580123430517, "grad_norm": 14.5, "learning_rate": 9.37443149516771e-06, "loss": 1.1344, "step": 18106 }, { "epoch": 1.9266865290487338, "grad_norm": 9.75, "learning_rate": 9.374395963615692e-06, "loss": 1.0999, "step": 18107 }, { "epoch": 1.9267929346669503, "grad_norm": 11.0859375, "learning_rate": 9.374360432063673e-06, "loss": 1.0444, "step": 18108 }, { "epoch": 1.926899340285167, "grad_norm": 11.0625, "learning_rate": 9.374324900511655e-06, "loss": 1.0773, "step": 18109 }, { "epoch": 1.9270057459033838, "grad_norm": 9.9296875, "learning_rate": 9.374289368959636e-06, "loss": 1.1318, "step": 18110 }, { "epoch": 1.9271121515216003, "grad_norm": 10.25, "learning_rate": 9.374253837407618e-06, "loss": 0.9673, "step": 18111 }, { "epoch": 1.9272185571398168, "grad_norm": 13.9453125, "learning_rate": 9.3742183058556e-06, "loss": 0.8129, "step": 18112 }, { "epoch": 1.9273249627580338, "grad_norm": 10.5625, "learning_rate": 9.374182774303581e-06, "loss": 1.1975, "step": 18113 }, { "epoch": 1.9274313683762503, "grad_norm": 11.9453125, "learning_rate": 9.374147242751565e-06, "loss": 1.03, "step": 18114 }, { "epoch": 1.9275377739944668, "grad_norm": 14.171875, "learning_rate": 9.374111711199546e-06, "loss": 1.0985, "step": 18115 }, { "epoch": 1.9276441796126835, "grad_norm": 9.5546875, "learning_rate": 9.374076179647528e-06, "loss": 0.9954, "step": 18116 }, { "epoch": 1.9277505852309003, "grad_norm": 16.46875, "learning_rate": 9.37404064809551e-06, "loss": 0.9465, "step": 18117 }, { "epoch": 1.9278569908491168, "grad_norm": 10.234375, "learning_rate": 9.374005116543491e-06, "loss": 1.0101, "step": 18118 }, { "epoch": 1.9279633964673335, "grad_norm": 9.5546875, "learning_rate": 9.373969584991474e-06, "loss": 1.0502, "step": 18119 }, { "epoch": 1.9280698020855502, "grad_norm": 11.9375, "learning_rate": 9.373934053439456e-06, "loss": 1.2774, "step": 18120 }, { "epoch": 1.9281762077037667, "grad_norm": 12.7265625, "learning_rate": 9.373898521887437e-06, "loss": 1.1397, "step": 18121 }, { "epoch": 1.9282826133219833, "grad_norm": 23.328125, "learning_rate": 9.373862990335419e-06, "loss": 1.1925, "step": 18122 }, { "epoch": 1.9283890189402002, "grad_norm": 21.71875, "learning_rate": 9.3738274587834e-06, "loss": 0.9916, "step": 18123 }, { "epoch": 1.9284954245584167, "grad_norm": 10.8671875, "learning_rate": 9.373791927231382e-06, "loss": 0.7499, "step": 18124 }, { "epoch": 1.9286018301766332, "grad_norm": 16.421875, "learning_rate": 9.373756395679364e-06, "loss": 1.364, "step": 18125 }, { "epoch": 1.92870823579485, "grad_norm": 17.171875, "learning_rate": 9.373720864127345e-06, "loss": 1.087, "step": 18126 }, { "epoch": 1.9288146414130667, "grad_norm": 19.5, "learning_rate": 9.373685332575327e-06, "loss": 1.1435, "step": 18127 }, { "epoch": 1.9289210470312832, "grad_norm": 10.484375, "learning_rate": 9.373649801023309e-06, "loss": 1.1911, "step": 18128 }, { "epoch": 1.9290274526495, "grad_norm": 12.265625, "learning_rate": 9.373614269471292e-06, "loss": 1.1413, "step": 18129 }, { "epoch": 1.9291338582677167, "grad_norm": 11.4140625, "learning_rate": 9.373578737919274e-06, "loss": 1.054, "step": 18130 }, { "epoch": 1.9292402638859332, "grad_norm": 12.984375, "learning_rate": 9.373543206367255e-06, "loss": 0.9096, "step": 18131 }, { "epoch": 1.9293466695041497, "grad_norm": 10.5703125, "learning_rate": 9.373507674815237e-06, "loss": 1.0454, "step": 18132 }, { "epoch": 1.9294530751223664, "grad_norm": 17.8125, "learning_rate": 9.373472143263218e-06, "loss": 1.0057, "step": 18133 }, { "epoch": 1.9295594807405831, "grad_norm": 13.859375, "learning_rate": 9.3734366117112e-06, "loss": 1.5125, "step": 18134 }, { "epoch": 1.9296658863587997, "grad_norm": 14.5, "learning_rate": 9.373401080159182e-06, "loss": 0.8659, "step": 18135 }, { "epoch": 1.9297722919770164, "grad_norm": 14.6015625, "learning_rate": 9.373365548607163e-06, "loss": 1.0987, "step": 18136 }, { "epoch": 1.9298786975952331, "grad_norm": 10.234375, "learning_rate": 9.373330017055145e-06, "loss": 1.1062, "step": 18137 }, { "epoch": 1.9299851032134496, "grad_norm": 14.5390625, "learning_rate": 9.373294485503126e-06, "loss": 1.3434, "step": 18138 }, { "epoch": 1.9300915088316664, "grad_norm": 11.4609375, "learning_rate": 9.37325895395111e-06, "loss": 1.1368, "step": 18139 }, { "epoch": 1.930197914449883, "grad_norm": 14.7578125, "learning_rate": 9.373223422399091e-06, "loss": 0.9015, "step": 18140 }, { "epoch": 1.9303043200680996, "grad_norm": 15.125, "learning_rate": 9.373187890847073e-06, "loss": 1.0547, "step": 18141 }, { "epoch": 1.930410725686316, "grad_norm": 15.15625, "learning_rate": 9.373152359295055e-06, "loss": 0.9767, "step": 18142 }, { "epoch": 1.9305171313045328, "grad_norm": 16.296875, "learning_rate": 9.373116827743036e-06, "loss": 1.1808, "step": 18143 }, { "epoch": 1.9306235369227496, "grad_norm": 21.328125, "learning_rate": 9.37308129619102e-06, "loss": 1.054, "step": 18144 }, { "epoch": 1.930729942540966, "grad_norm": 14.4296875, "learning_rate": 9.373045764639001e-06, "loss": 1.276, "step": 18145 }, { "epoch": 1.9308363481591828, "grad_norm": 12.09375, "learning_rate": 9.373010233086983e-06, "loss": 1.0285, "step": 18146 }, { "epoch": 1.9309427537773995, "grad_norm": 10.421875, "learning_rate": 9.372974701534964e-06, "loss": 1.1732, "step": 18147 }, { "epoch": 1.931049159395616, "grad_norm": 11.9296875, "learning_rate": 9.372939169982946e-06, "loss": 1.233, "step": 18148 }, { "epoch": 1.9311555650138326, "grad_norm": 12.375, "learning_rate": 9.372903638430928e-06, "loss": 1.2418, "step": 18149 }, { "epoch": 1.9312619706320495, "grad_norm": 10.9140625, "learning_rate": 9.372868106878909e-06, "loss": 0.866, "step": 18150 }, { "epoch": 1.931368376250266, "grad_norm": 15.2578125, "learning_rate": 9.37283257532689e-06, "loss": 1.1263, "step": 18151 }, { "epoch": 1.9314747818684825, "grad_norm": 10.890625, "learning_rate": 9.372797043774872e-06, "loss": 0.9416, "step": 18152 }, { "epoch": 1.9315811874866993, "grad_norm": 10.734375, "learning_rate": 9.372761512222854e-06, "loss": 1.2219, "step": 18153 }, { "epoch": 1.931687593104916, "grad_norm": 10.6953125, "learning_rate": 9.372725980670836e-06, "loss": 1.0883, "step": 18154 }, { "epoch": 1.9317939987231325, "grad_norm": 10.53125, "learning_rate": 9.372690449118819e-06, "loss": 1.1628, "step": 18155 }, { "epoch": 1.9319004043413492, "grad_norm": 15.875, "learning_rate": 9.3726549175668e-06, "loss": 1.1196, "step": 18156 }, { "epoch": 1.932006809959566, "grad_norm": 16.765625, "learning_rate": 9.372619386014782e-06, "loss": 0.9498, "step": 18157 }, { "epoch": 1.9321132155777825, "grad_norm": 14.53125, "learning_rate": 9.372583854462764e-06, "loss": 1.1219, "step": 18158 }, { "epoch": 1.932219621195999, "grad_norm": 17.265625, "learning_rate": 9.372548322910745e-06, "loss": 1.1057, "step": 18159 }, { "epoch": 1.932326026814216, "grad_norm": 15.9765625, "learning_rate": 9.372512791358727e-06, "loss": 1.2266, "step": 18160 }, { "epoch": 1.9324324324324325, "grad_norm": 11.625, "learning_rate": 9.372477259806708e-06, "loss": 1.2023, "step": 18161 }, { "epoch": 1.932538838050649, "grad_norm": 10.140625, "learning_rate": 9.37244172825469e-06, "loss": 0.8861, "step": 18162 }, { "epoch": 1.9326452436688657, "grad_norm": 15.484375, "learning_rate": 9.372406196702672e-06, "loss": 1.1314, "step": 18163 }, { "epoch": 1.9327516492870824, "grad_norm": 16.796875, "learning_rate": 9.372370665150653e-06, "loss": 1.2012, "step": 18164 }, { "epoch": 1.932858054905299, "grad_norm": 16.53125, "learning_rate": 9.372335133598637e-06, "loss": 1.0756, "step": 18165 }, { "epoch": 1.9329644605235157, "grad_norm": 15.0546875, "learning_rate": 9.372299602046618e-06, "loss": 1.092, "step": 18166 }, { "epoch": 1.9330708661417324, "grad_norm": 15.921875, "learning_rate": 9.3722640704946e-06, "loss": 0.9597, "step": 18167 }, { "epoch": 1.933177271759949, "grad_norm": 14.71875, "learning_rate": 9.372228538942581e-06, "loss": 1.155, "step": 18168 }, { "epoch": 1.9332836773781654, "grad_norm": 10.1328125, "learning_rate": 9.372193007390563e-06, "loss": 0.8379, "step": 18169 }, { "epoch": 1.9333900829963824, "grad_norm": 10.5859375, "learning_rate": 9.372157475838546e-06, "loss": 1.1543, "step": 18170 }, { "epoch": 1.9334964886145989, "grad_norm": 15.8515625, "learning_rate": 9.372121944286528e-06, "loss": 0.9181, "step": 18171 }, { "epoch": 1.9336028942328154, "grad_norm": 10.890625, "learning_rate": 9.37208641273451e-06, "loss": 1.0813, "step": 18172 }, { "epoch": 1.9337092998510321, "grad_norm": 10.4375, "learning_rate": 9.372050881182491e-06, "loss": 1.2764, "step": 18173 }, { "epoch": 1.9338157054692489, "grad_norm": 10.671875, "learning_rate": 9.372015349630473e-06, "loss": 1.0392, "step": 18174 }, { "epoch": 1.9339221110874654, "grad_norm": 12.78125, "learning_rate": 9.371979818078454e-06, "loss": 1.0865, "step": 18175 }, { "epoch": 1.934028516705682, "grad_norm": 13.6796875, "learning_rate": 9.371944286526436e-06, "loss": 1.3505, "step": 18176 }, { "epoch": 1.9341349223238988, "grad_norm": 15.125, "learning_rate": 9.371908754974418e-06, "loss": 1.4192, "step": 18177 }, { "epoch": 1.9342413279421153, "grad_norm": 11.59375, "learning_rate": 9.3718732234224e-06, "loss": 1.2209, "step": 18178 }, { "epoch": 1.9343477335603319, "grad_norm": 19.46875, "learning_rate": 9.37183769187038e-06, "loss": 1.0301, "step": 18179 }, { "epoch": 1.9344541391785486, "grad_norm": 16.625, "learning_rate": 9.371802160318364e-06, "loss": 0.9436, "step": 18180 }, { "epoch": 1.9345605447967653, "grad_norm": 10.21875, "learning_rate": 9.371766628766346e-06, "loss": 1.0575, "step": 18181 }, { "epoch": 1.9346669504149818, "grad_norm": 11.71875, "learning_rate": 9.371731097214327e-06, "loss": 1.1613, "step": 18182 }, { "epoch": 1.9347733560331986, "grad_norm": 13.7734375, "learning_rate": 9.371695565662309e-06, "loss": 1.1387, "step": 18183 }, { "epoch": 1.9348797616514153, "grad_norm": 12.7421875, "learning_rate": 9.37166003411029e-06, "loss": 1.0885, "step": 18184 }, { "epoch": 1.9349861672696318, "grad_norm": 12.0859375, "learning_rate": 9.371624502558274e-06, "loss": 1.3766, "step": 18185 }, { "epoch": 1.9350925728878485, "grad_norm": 15.265625, "learning_rate": 9.371588971006255e-06, "loss": 1.1528, "step": 18186 }, { "epoch": 1.9351989785060653, "grad_norm": 13.0703125, "learning_rate": 9.371553439454237e-06, "loss": 1.1251, "step": 18187 }, { "epoch": 1.9353053841242818, "grad_norm": 12.6015625, "learning_rate": 9.371517907902219e-06, "loss": 1.2744, "step": 18188 }, { "epoch": 1.9354117897424983, "grad_norm": 12.3828125, "learning_rate": 9.3714823763502e-06, "loss": 1.2374, "step": 18189 }, { "epoch": 1.935518195360715, "grad_norm": 12.828125, "learning_rate": 9.37144684479818e-06, "loss": 1.119, "step": 18190 }, { "epoch": 1.9356246009789317, "grad_norm": 13.2578125, "learning_rate": 9.371411313246163e-06, "loss": 1.2373, "step": 18191 }, { "epoch": 1.9357310065971483, "grad_norm": 15.4609375, "learning_rate": 9.371375781694145e-06, "loss": 0.9379, "step": 18192 }, { "epoch": 1.935837412215365, "grad_norm": 10.0859375, "learning_rate": 9.371340250142127e-06, "loss": 0.9892, "step": 18193 }, { "epoch": 1.9359438178335817, "grad_norm": 14.0078125, "learning_rate": 9.371304718590108e-06, "loss": 1.2328, "step": 18194 }, { "epoch": 1.9360502234517982, "grad_norm": 14.71875, "learning_rate": 9.37126918703809e-06, "loss": 0.9811, "step": 18195 }, { "epoch": 1.9361566290700147, "grad_norm": 10.46875, "learning_rate": 9.371233655486073e-06, "loss": 1.2734, "step": 18196 }, { "epoch": 1.9362630346882317, "grad_norm": 11.6171875, "learning_rate": 9.371198123934055e-06, "loss": 0.8858, "step": 18197 }, { "epoch": 1.9363694403064482, "grad_norm": 14.3203125, "learning_rate": 9.371162592382036e-06, "loss": 1.0522, "step": 18198 }, { "epoch": 1.9364758459246647, "grad_norm": 14.3515625, "learning_rate": 9.371127060830018e-06, "loss": 1.1194, "step": 18199 }, { "epoch": 1.9365822515428814, "grad_norm": 16.171875, "learning_rate": 9.371091529278e-06, "loss": 1.0183, "step": 18200 }, { "epoch": 1.9366886571610982, "grad_norm": 16.4375, "learning_rate": 9.371055997725981e-06, "loss": 1.1099, "step": 18201 }, { "epoch": 1.9367950627793147, "grad_norm": 10.140625, "learning_rate": 9.371020466173963e-06, "loss": 1.0455, "step": 18202 }, { "epoch": 1.9369014683975314, "grad_norm": 17.46875, "learning_rate": 9.370984934621944e-06, "loss": 1.3238, "step": 18203 }, { "epoch": 1.9370078740157481, "grad_norm": 15.5234375, "learning_rate": 9.370949403069926e-06, "loss": 0.9886, "step": 18204 }, { "epoch": 1.9371142796339647, "grad_norm": 10.25, "learning_rate": 9.370913871517908e-06, "loss": 1.135, "step": 18205 }, { "epoch": 1.9372206852521812, "grad_norm": 13.2734375, "learning_rate": 9.370878339965891e-06, "loss": 1.1944, "step": 18206 }, { "epoch": 1.9373270908703981, "grad_norm": 13.84375, "learning_rate": 9.370842808413873e-06, "loss": 1.4038, "step": 18207 }, { "epoch": 1.9374334964886146, "grad_norm": 13.5, "learning_rate": 9.370807276861854e-06, "loss": 1.1121, "step": 18208 }, { "epoch": 1.9375399021068311, "grad_norm": 15.328125, "learning_rate": 9.370771745309836e-06, "loss": 1.0457, "step": 18209 }, { "epoch": 1.9376463077250479, "grad_norm": 16.421875, "learning_rate": 9.370736213757817e-06, "loss": 0.986, "step": 18210 }, { "epoch": 1.9377527133432646, "grad_norm": 11.7109375, "learning_rate": 9.3707006822058e-06, "loss": 1.0912, "step": 18211 }, { "epoch": 1.937859118961481, "grad_norm": 10.28125, "learning_rate": 9.370665150653782e-06, "loss": 0.8501, "step": 18212 }, { "epoch": 1.9379655245796978, "grad_norm": 12.4609375, "learning_rate": 9.370629619101764e-06, "loss": 1.1527, "step": 18213 }, { "epoch": 1.9380719301979146, "grad_norm": 10.640625, "learning_rate": 9.370594087549746e-06, "loss": 1.133, "step": 18214 }, { "epoch": 1.938178335816131, "grad_norm": 16.359375, "learning_rate": 9.370558555997727e-06, "loss": 1.1961, "step": 18215 }, { "epoch": 1.9382847414343476, "grad_norm": 9.0234375, "learning_rate": 9.370523024445709e-06, "loss": 0.9375, "step": 18216 }, { "epoch": 1.9383911470525643, "grad_norm": 13.2421875, "learning_rate": 9.37048749289369e-06, "loss": 1.1162, "step": 18217 }, { "epoch": 1.938497552670781, "grad_norm": 10.859375, "learning_rate": 9.370451961341672e-06, "loss": 0.9246, "step": 18218 }, { "epoch": 1.9386039582889976, "grad_norm": 12.484375, "learning_rate": 9.370416429789654e-06, "loss": 0.9108, "step": 18219 }, { "epoch": 1.9387103639072143, "grad_norm": 15.96875, "learning_rate": 9.370380898237635e-06, "loss": 1.1546, "step": 18220 }, { "epoch": 1.938816769525431, "grad_norm": 10.109375, "learning_rate": 9.370345366685618e-06, "loss": 1.1342, "step": 18221 }, { "epoch": 1.9389231751436475, "grad_norm": 17.453125, "learning_rate": 9.3703098351336e-06, "loss": 1.1004, "step": 18222 }, { "epoch": 1.9390295807618643, "grad_norm": 14.890625, "learning_rate": 9.370274303581582e-06, "loss": 1.1843, "step": 18223 }, { "epoch": 1.939135986380081, "grad_norm": 11.6796875, "learning_rate": 9.370238772029563e-06, "loss": 0.928, "step": 18224 }, { "epoch": 1.9392423919982975, "grad_norm": 13.8125, "learning_rate": 9.370203240477545e-06, "loss": 0.7062, "step": 18225 }, { "epoch": 1.939348797616514, "grad_norm": 12.2734375, "learning_rate": 9.370167708925526e-06, "loss": 1.1081, "step": 18226 }, { "epoch": 1.9394552032347308, "grad_norm": 12.375, "learning_rate": 9.370132177373508e-06, "loss": 0.8126, "step": 18227 }, { "epoch": 1.9395616088529475, "grad_norm": 13.640625, "learning_rate": 9.37009664582149e-06, "loss": 0.9894, "step": 18228 }, { "epoch": 1.939668014471164, "grad_norm": 11.3125, "learning_rate": 9.370061114269471e-06, "loss": 0.8509, "step": 18229 }, { "epoch": 1.9397744200893807, "grad_norm": 11.1796875, "learning_rate": 9.370025582717453e-06, "loss": 0.9409, "step": 18230 }, { "epoch": 1.9398808257075975, "grad_norm": 15.2578125, "learning_rate": 9.369990051165435e-06, "loss": 0.92, "step": 18231 }, { "epoch": 1.939987231325814, "grad_norm": 16.40625, "learning_rate": 9.369954519613418e-06, "loss": 1.0398, "step": 18232 }, { "epoch": 1.9400936369440307, "grad_norm": 13.1484375, "learning_rate": 9.3699189880614e-06, "loss": 1.1833, "step": 18233 }, { "epoch": 1.9402000425622474, "grad_norm": 11.578125, "learning_rate": 9.369883456509381e-06, "loss": 0.9162, "step": 18234 }, { "epoch": 1.940306448180464, "grad_norm": 9.1015625, "learning_rate": 9.369847924957363e-06, "loss": 1.0592, "step": 18235 }, { "epoch": 1.9404128537986804, "grad_norm": 10.25, "learning_rate": 9.369812393405344e-06, "loss": 1.2603, "step": 18236 }, { "epoch": 1.9405192594168972, "grad_norm": 9.2109375, "learning_rate": 9.369776861853328e-06, "loss": 1.048, "step": 18237 }, { "epoch": 1.940625665035114, "grad_norm": 13.3046875, "learning_rate": 9.369741330301309e-06, "loss": 0.8207, "step": 18238 }, { "epoch": 1.9407320706533304, "grad_norm": 11.5859375, "learning_rate": 9.36970579874929e-06, "loss": 1.1505, "step": 18239 }, { "epoch": 1.9408384762715472, "grad_norm": 10.609375, "learning_rate": 9.369670267197272e-06, "loss": 1.2538, "step": 18240 }, { "epoch": 1.9409448818897639, "grad_norm": 15.0859375, "learning_rate": 9.369634735645254e-06, "loss": 0.91, "step": 18241 }, { "epoch": 1.9410512875079804, "grad_norm": 10.9921875, "learning_rate": 9.369599204093236e-06, "loss": 1.4721, "step": 18242 }, { "epoch": 1.941157693126197, "grad_norm": 12.765625, "learning_rate": 9.369563672541217e-06, "loss": 0.8689, "step": 18243 }, { "epoch": 1.9412640987444139, "grad_norm": 14.7578125, "learning_rate": 9.369528140989199e-06, "loss": 1.3322, "step": 18244 }, { "epoch": 1.9413705043626304, "grad_norm": 13.34375, "learning_rate": 9.36949260943718e-06, "loss": 1.0612, "step": 18245 }, { "epoch": 1.9414769099808469, "grad_norm": 9.3125, "learning_rate": 9.369457077885162e-06, "loss": 0.8912, "step": 18246 }, { "epoch": 1.9415833155990636, "grad_norm": 9.109375, "learning_rate": 9.369421546333145e-06, "loss": 0.8753, "step": 18247 }, { "epoch": 1.9416897212172803, "grad_norm": 12.0859375, "learning_rate": 9.369386014781127e-06, "loss": 1.2769, "step": 18248 }, { "epoch": 1.9417961268354968, "grad_norm": 12.6796875, "learning_rate": 9.369350483229109e-06, "loss": 1.335, "step": 18249 }, { "epoch": 1.9419025324537136, "grad_norm": 10.46875, "learning_rate": 9.36931495167709e-06, "loss": 0.936, "step": 18250 }, { "epoch": 1.9420089380719303, "grad_norm": 12.53125, "learning_rate": 9.369279420125072e-06, "loss": 0.8654, "step": 18251 }, { "epoch": 1.9421153436901468, "grad_norm": 10.6484375, "learning_rate": 9.369243888573053e-06, "loss": 1.0927, "step": 18252 }, { "epoch": 1.9422217493083633, "grad_norm": 14.015625, "learning_rate": 9.369208357021035e-06, "loss": 1.0709, "step": 18253 }, { "epoch": 1.9423281549265803, "grad_norm": 13.0234375, "learning_rate": 9.369172825469017e-06, "loss": 0.9844, "step": 18254 }, { "epoch": 1.9424345605447968, "grad_norm": 12.4609375, "learning_rate": 9.369137293916998e-06, "loss": 0.9353, "step": 18255 }, { "epoch": 1.9425409661630133, "grad_norm": 14.5234375, "learning_rate": 9.36910176236498e-06, "loss": 1.19, "step": 18256 }, { "epoch": 1.94264737178123, "grad_norm": 12.453125, "learning_rate": 9.369066230812963e-06, "loss": 0.9067, "step": 18257 }, { "epoch": 1.9427537773994468, "grad_norm": 10.9296875, "learning_rate": 9.369030699260945e-06, "loss": 1.1693, "step": 18258 }, { "epoch": 1.9428601830176633, "grad_norm": 11.15625, "learning_rate": 9.368995167708926e-06, "loss": 1.0223, "step": 18259 }, { "epoch": 1.94296658863588, "grad_norm": 10.96875, "learning_rate": 9.368959636156908e-06, "loss": 1.2704, "step": 18260 }, { "epoch": 1.9430729942540967, "grad_norm": 9.4296875, "learning_rate": 9.36892410460489e-06, "loss": 0.8958, "step": 18261 }, { "epoch": 1.9431793998723133, "grad_norm": 14.03125, "learning_rate": 9.368888573052873e-06, "loss": 1.1202, "step": 18262 }, { "epoch": 1.9432858054905298, "grad_norm": 11.1015625, "learning_rate": 9.368853041500854e-06, "loss": 1.3573, "step": 18263 }, { "epoch": 1.9433922111087465, "grad_norm": 15.5078125, "learning_rate": 9.368817509948836e-06, "loss": 1.4922, "step": 18264 }, { "epoch": 1.9434986167269632, "grad_norm": 9.7890625, "learning_rate": 9.368781978396818e-06, "loss": 0.8397, "step": 18265 }, { "epoch": 1.9436050223451797, "grad_norm": 12.390625, "learning_rate": 9.3687464468448e-06, "loss": 1.0274, "step": 18266 }, { "epoch": 1.9437114279633965, "grad_norm": 10.9296875, "learning_rate": 9.36871091529278e-06, "loss": 1.1769, "step": 18267 }, { "epoch": 1.9438178335816132, "grad_norm": 13.3515625, "learning_rate": 9.368675383740762e-06, "loss": 1.0846, "step": 18268 }, { "epoch": 1.9439242391998297, "grad_norm": 11.2890625, "learning_rate": 9.368639852188744e-06, "loss": 1.0679, "step": 18269 }, { "epoch": 1.9440306448180464, "grad_norm": 14.5546875, "learning_rate": 9.368604320636726e-06, "loss": 1.0288, "step": 18270 }, { "epoch": 1.9441370504362632, "grad_norm": 11.296875, "learning_rate": 9.368568789084707e-06, "loss": 1.0302, "step": 18271 }, { "epoch": 1.9442434560544797, "grad_norm": 15.65625, "learning_rate": 9.368533257532689e-06, "loss": 1.0241, "step": 18272 }, { "epoch": 1.9443498616726962, "grad_norm": 14.5390625, "learning_rate": 9.368497725980672e-06, "loss": 1.2607, "step": 18273 }, { "epoch": 1.944456267290913, "grad_norm": 13.7890625, "learning_rate": 9.368462194428654e-06, "loss": 0.9305, "step": 18274 }, { "epoch": 1.9445626729091297, "grad_norm": 11.2890625, "learning_rate": 9.368426662876635e-06, "loss": 1.257, "step": 18275 }, { "epoch": 1.9446690785273462, "grad_norm": 12.1640625, "learning_rate": 9.368391131324617e-06, "loss": 1.2813, "step": 18276 }, { "epoch": 1.944775484145563, "grad_norm": 11.609375, "learning_rate": 9.368355599772599e-06, "loss": 1.0873, "step": 18277 }, { "epoch": 1.9448818897637796, "grad_norm": 11.1953125, "learning_rate": 9.368320068220582e-06, "loss": 1.1706, "step": 18278 }, { "epoch": 1.9449882953819961, "grad_norm": 10.9609375, "learning_rate": 9.368284536668562e-06, "loss": 1.275, "step": 18279 }, { "epoch": 1.9450947010002126, "grad_norm": 9.7734375, "learning_rate": 9.368249005116543e-06, "loss": 1.2707, "step": 18280 }, { "epoch": 1.9452011066184296, "grad_norm": 15.203125, "learning_rate": 9.368213473564525e-06, "loss": 1.0, "step": 18281 }, { "epoch": 1.945307512236646, "grad_norm": 9.703125, "learning_rate": 9.368177942012507e-06, "loss": 1.1321, "step": 18282 }, { "epoch": 1.9454139178548626, "grad_norm": 11.5, "learning_rate": 9.36814241046049e-06, "loss": 0.948, "step": 18283 }, { "epoch": 1.9455203234730793, "grad_norm": 12.6875, "learning_rate": 9.368106878908472e-06, "loss": 1.0532, "step": 18284 }, { "epoch": 1.945626729091296, "grad_norm": 10.0, "learning_rate": 9.368071347356453e-06, "loss": 1.1414, "step": 18285 }, { "epoch": 1.9457331347095126, "grad_norm": 14.5859375, "learning_rate": 9.368035815804435e-06, "loss": 0.9825, "step": 18286 }, { "epoch": 1.9458395403277293, "grad_norm": 12.3125, "learning_rate": 9.368000284252416e-06, "loss": 1.0661, "step": 18287 }, { "epoch": 1.945945945945946, "grad_norm": 13.7265625, "learning_rate": 9.3679647527004e-06, "loss": 1.1646, "step": 18288 }, { "epoch": 1.9460523515641626, "grad_norm": 13.1328125, "learning_rate": 9.367929221148381e-06, "loss": 1.2219, "step": 18289 }, { "epoch": 1.946158757182379, "grad_norm": 15.3671875, "learning_rate": 9.367893689596363e-06, "loss": 1.2845, "step": 18290 }, { "epoch": 1.946265162800596, "grad_norm": 15.453125, "learning_rate": 9.367858158044344e-06, "loss": 1.1456, "step": 18291 }, { "epoch": 1.9463715684188125, "grad_norm": 21.671875, "learning_rate": 9.367822626492326e-06, "loss": 1.0617, "step": 18292 }, { "epoch": 1.946477974037029, "grad_norm": 9.9140625, "learning_rate": 9.367787094940308e-06, "loss": 1.0351, "step": 18293 }, { "epoch": 1.9465843796552458, "grad_norm": 16.578125, "learning_rate": 9.36775156338829e-06, "loss": 0.9414, "step": 18294 }, { "epoch": 1.9466907852734625, "grad_norm": 11.5625, "learning_rate": 9.367716031836271e-06, "loss": 1.3133, "step": 18295 }, { "epoch": 1.946797190891679, "grad_norm": 14.0625, "learning_rate": 9.367680500284253e-06, "loss": 0.9148, "step": 18296 }, { "epoch": 1.9469035965098958, "grad_norm": 11.71875, "learning_rate": 9.367644968732234e-06, "loss": 0.9351, "step": 18297 }, { "epoch": 1.9470100021281125, "grad_norm": 14.46875, "learning_rate": 9.367609437180217e-06, "loss": 1.1131, "step": 18298 }, { "epoch": 1.947116407746329, "grad_norm": 14.5703125, "learning_rate": 9.367573905628199e-06, "loss": 0.9818, "step": 18299 }, { "epoch": 1.9472228133645455, "grad_norm": 12.6328125, "learning_rate": 9.36753837407618e-06, "loss": 0.9674, "step": 18300 }, { "epoch": 1.9473292189827625, "grad_norm": 20.75, "learning_rate": 9.367502842524162e-06, "loss": 1.1109, "step": 18301 }, { "epoch": 1.947435624600979, "grad_norm": 12.0390625, "learning_rate": 9.367467310972144e-06, "loss": 1.2472, "step": 18302 }, { "epoch": 1.9475420302191955, "grad_norm": 11.4453125, "learning_rate": 9.367431779420127e-06, "loss": 0.8873, "step": 18303 }, { "epoch": 1.9476484358374122, "grad_norm": 11.09375, "learning_rate": 9.367396247868109e-06, "loss": 1.3478, "step": 18304 }, { "epoch": 1.947754841455629, "grad_norm": 12.4296875, "learning_rate": 9.36736071631609e-06, "loss": 1.0344, "step": 18305 }, { "epoch": 1.9478612470738454, "grad_norm": 13.109375, "learning_rate": 9.367325184764072e-06, "loss": 1.3211, "step": 18306 }, { "epoch": 1.9479676526920622, "grad_norm": 11.4375, "learning_rate": 9.367289653212054e-06, "loss": 1.0087, "step": 18307 }, { "epoch": 1.948074058310279, "grad_norm": 11.65625, "learning_rate": 9.367254121660035e-06, "loss": 0.7168, "step": 18308 }, { "epoch": 1.9481804639284954, "grad_norm": 13.296875, "learning_rate": 9.367218590108017e-06, "loss": 1.0416, "step": 18309 }, { "epoch": 1.948286869546712, "grad_norm": 12.4765625, "learning_rate": 9.367183058555998e-06, "loss": 0.9509, "step": 18310 }, { "epoch": 1.9483932751649287, "grad_norm": 12.3828125, "learning_rate": 9.36714752700398e-06, "loss": 1.0103, "step": 18311 }, { "epoch": 1.9484996807831454, "grad_norm": 9.734375, "learning_rate": 9.367111995451962e-06, "loss": 1.0181, "step": 18312 }, { "epoch": 1.948606086401362, "grad_norm": 11.71875, "learning_rate": 9.367076463899943e-06, "loss": 0.7912, "step": 18313 }, { "epoch": 1.9487124920195786, "grad_norm": 12.71875, "learning_rate": 9.367040932347927e-06, "loss": 1.0428, "step": 18314 }, { "epoch": 1.9488188976377954, "grad_norm": 12.5, "learning_rate": 9.367005400795908e-06, "loss": 0.8572, "step": 18315 }, { "epoch": 1.9489253032560119, "grad_norm": 14.4453125, "learning_rate": 9.36696986924389e-06, "loss": 1.0846, "step": 18316 }, { "epoch": 1.9490317088742286, "grad_norm": 15.0859375, "learning_rate": 9.366934337691871e-06, "loss": 1.0776, "step": 18317 }, { "epoch": 1.9491381144924453, "grad_norm": 11.59375, "learning_rate": 9.366898806139853e-06, "loss": 1.1089, "step": 18318 }, { "epoch": 1.9492445201106618, "grad_norm": 13.8671875, "learning_rate": 9.366863274587835e-06, "loss": 1.3541, "step": 18319 }, { "epoch": 1.9493509257288784, "grad_norm": 14.734375, "learning_rate": 9.366827743035816e-06, "loss": 1.0312, "step": 18320 }, { "epoch": 1.949457331347095, "grad_norm": 17.265625, "learning_rate": 9.366792211483798e-06, "loss": 1.02, "step": 18321 }, { "epoch": 1.9495637369653118, "grad_norm": 13.5859375, "learning_rate": 9.36675667993178e-06, "loss": 1.095, "step": 18322 }, { "epoch": 1.9496701425835283, "grad_norm": 12.734375, "learning_rate": 9.366721148379761e-06, "loss": 1.0916, "step": 18323 }, { "epoch": 1.949776548201745, "grad_norm": 17.34375, "learning_rate": 9.366685616827744e-06, "loss": 0.9371, "step": 18324 }, { "epoch": 1.9498829538199618, "grad_norm": 10.359375, "learning_rate": 9.366650085275726e-06, "loss": 1.0669, "step": 18325 }, { "epoch": 1.9499893594381783, "grad_norm": 16.3125, "learning_rate": 9.366614553723707e-06, "loss": 1.0694, "step": 18326 }, { "epoch": 1.9500957650563948, "grad_norm": 9.75, "learning_rate": 9.366579022171689e-06, "loss": 1.0411, "step": 18327 }, { "epoch": 1.9502021706746118, "grad_norm": 10.6796875, "learning_rate": 9.36654349061967e-06, "loss": 0.8775, "step": 18328 }, { "epoch": 1.9503085762928283, "grad_norm": 15.25, "learning_rate": 9.366507959067654e-06, "loss": 0.9424, "step": 18329 }, { "epoch": 1.9504149819110448, "grad_norm": 13.5390625, "learning_rate": 9.366472427515636e-06, "loss": 1.141, "step": 18330 }, { "epoch": 1.9505213875292615, "grad_norm": 9.453125, "learning_rate": 9.366436895963617e-06, "loss": 1.1758, "step": 18331 }, { "epoch": 1.9506277931474783, "grad_norm": 10.78125, "learning_rate": 9.366401364411599e-06, "loss": 1.0041, "step": 18332 }, { "epoch": 1.9507341987656948, "grad_norm": 9.9609375, "learning_rate": 9.36636583285958e-06, "loss": 1.0737, "step": 18333 }, { "epoch": 1.9508406043839115, "grad_norm": 15.234375, "learning_rate": 9.366330301307562e-06, "loss": 0.9838, "step": 18334 }, { "epoch": 1.9509470100021282, "grad_norm": 13.2890625, "learning_rate": 9.366294769755544e-06, "loss": 0.9638, "step": 18335 }, { "epoch": 1.9510534156203447, "grad_norm": 14.578125, "learning_rate": 9.366259238203525e-06, "loss": 1.3864, "step": 18336 }, { "epoch": 1.9511598212385612, "grad_norm": 19.640625, "learning_rate": 9.366223706651507e-06, "loss": 0.8468, "step": 18337 }, { "epoch": 1.9512662268567782, "grad_norm": 15.4140625, "learning_rate": 9.366188175099488e-06, "loss": 1.159, "step": 18338 }, { "epoch": 1.9513726324749947, "grad_norm": 10.8984375, "learning_rate": 9.366152643547472e-06, "loss": 0.9616, "step": 18339 }, { "epoch": 1.9514790380932112, "grad_norm": 11.2734375, "learning_rate": 9.366117111995453e-06, "loss": 0.7758, "step": 18340 }, { "epoch": 1.951585443711428, "grad_norm": 13.8125, "learning_rate": 9.366081580443435e-06, "loss": 1.0779, "step": 18341 }, { "epoch": 1.9516918493296447, "grad_norm": 14.2265625, "learning_rate": 9.366046048891417e-06, "loss": 1.088, "step": 18342 }, { "epoch": 1.9517982549478612, "grad_norm": 15.7109375, "learning_rate": 9.366010517339398e-06, "loss": 1.0211, "step": 18343 }, { "epoch": 1.951904660566078, "grad_norm": 14.1015625, "learning_rate": 9.36597498578738e-06, "loss": 0.7889, "step": 18344 }, { "epoch": 1.9520110661842947, "grad_norm": 16.734375, "learning_rate": 9.365939454235361e-06, "loss": 1.1333, "step": 18345 }, { "epoch": 1.9521174718025112, "grad_norm": 11.3515625, "learning_rate": 9.365903922683343e-06, "loss": 0.9386, "step": 18346 }, { "epoch": 1.9522238774207277, "grad_norm": 12.1640625, "learning_rate": 9.365868391131325e-06, "loss": 1.0347, "step": 18347 }, { "epoch": 1.9523302830389444, "grad_norm": 13.390625, "learning_rate": 9.365832859579306e-06, "loss": 1.1964, "step": 18348 }, { "epoch": 1.9524366886571611, "grad_norm": 11.5703125, "learning_rate": 9.365797328027288e-06, "loss": 1.224, "step": 18349 }, { "epoch": 1.9525430942753776, "grad_norm": 16.328125, "learning_rate": 9.365761796475271e-06, "loss": 1.4948, "step": 18350 }, { "epoch": 1.9526494998935944, "grad_norm": 11.3359375, "learning_rate": 9.365726264923253e-06, "loss": 1.2179, "step": 18351 }, { "epoch": 1.952755905511811, "grad_norm": 15.21875, "learning_rate": 9.365690733371234e-06, "loss": 1.0463, "step": 18352 }, { "epoch": 1.9528623111300276, "grad_norm": 18.3125, "learning_rate": 9.365655201819216e-06, "loss": 1.0838, "step": 18353 }, { "epoch": 1.9529687167482443, "grad_norm": 15.875, "learning_rate": 9.365619670267198e-06, "loss": 1.1939, "step": 18354 }, { "epoch": 1.953075122366461, "grad_norm": 16.390625, "learning_rate": 9.365584138715181e-06, "loss": 1.2362, "step": 18355 }, { "epoch": 1.9531815279846776, "grad_norm": 11.0, "learning_rate": 9.365548607163162e-06, "loss": 0.9191, "step": 18356 }, { "epoch": 1.953287933602894, "grad_norm": 11.4140625, "learning_rate": 9.365513075611144e-06, "loss": 0.7934, "step": 18357 }, { "epoch": 1.9533943392211108, "grad_norm": 14.9375, "learning_rate": 9.365477544059126e-06, "loss": 0.9568, "step": 18358 }, { "epoch": 1.9535007448393276, "grad_norm": 11.921875, "learning_rate": 9.365442012507107e-06, "loss": 1.1792, "step": 18359 }, { "epoch": 1.953607150457544, "grad_norm": 9.4375, "learning_rate": 9.365406480955089e-06, "loss": 0.9259, "step": 18360 }, { "epoch": 1.9537135560757608, "grad_norm": 13.7734375, "learning_rate": 9.36537094940307e-06, "loss": 1.1206, "step": 18361 }, { "epoch": 1.9538199616939775, "grad_norm": 14.828125, "learning_rate": 9.365335417851052e-06, "loss": 0.9979, "step": 18362 }, { "epoch": 1.953926367312194, "grad_norm": 11.1171875, "learning_rate": 9.365299886299034e-06, "loss": 0.9743, "step": 18363 }, { "epoch": 1.9540327729304108, "grad_norm": 10.90625, "learning_rate": 9.365264354747015e-06, "loss": 1.2394, "step": 18364 }, { "epoch": 1.9541391785486275, "grad_norm": 12.1640625, "learning_rate": 9.365228823194999e-06, "loss": 0.9036, "step": 18365 }, { "epoch": 1.954245584166844, "grad_norm": 14.4375, "learning_rate": 9.36519329164298e-06, "loss": 1.0393, "step": 18366 }, { "epoch": 1.9543519897850605, "grad_norm": 15.375, "learning_rate": 9.365157760090962e-06, "loss": 0.9991, "step": 18367 }, { "epoch": 1.9544583954032773, "grad_norm": 10.7734375, "learning_rate": 9.365122228538943e-06, "loss": 1.2801, "step": 18368 }, { "epoch": 1.954564801021494, "grad_norm": 14.703125, "learning_rate": 9.365086696986925e-06, "loss": 1.2516, "step": 18369 }, { "epoch": 1.9546712066397105, "grad_norm": 16.859375, "learning_rate": 9.365051165434907e-06, "loss": 1.1391, "step": 18370 }, { "epoch": 1.9547776122579272, "grad_norm": 11.671875, "learning_rate": 9.365015633882888e-06, "loss": 1.1192, "step": 18371 }, { "epoch": 1.954884017876144, "grad_norm": 18.703125, "learning_rate": 9.36498010233087e-06, "loss": 1.0013, "step": 18372 }, { "epoch": 1.9549904234943605, "grad_norm": 12.9140625, "learning_rate": 9.364944570778851e-06, "loss": 1.3379, "step": 18373 }, { "epoch": 1.955096829112577, "grad_norm": 12.3671875, "learning_rate": 9.364909039226833e-06, "loss": 1.0969, "step": 18374 }, { "epoch": 1.955203234730794, "grad_norm": 10.7734375, "learning_rate": 9.364873507674816e-06, "loss": 1.3981, "step": 18375 }, { "epoch": 1.9553096403490104, "grad_norm": 13.03125, "learning_rate": 9.364837976122798e-06, "loss": 1.4752, "step": 18376 }, { "epoch": 1.955416045967227, "grad_norm": 20.890625, "learning_rate": 9.36480244457078e-06, "loss": 1.0644, "step": 18377 }, { "epoch": 1.9555224515854437, "grad_norm": 12.1875, "learning_rate": 9.364766913018761e-06, "loss": 1.1469, "step": 18378 }, { "epoch": 1.9556288572036604, "grad_norm": 9.265625, "learning_rate": 9.364731381466743e-06, "loss": 1.1843, "step": 18379 }, { "epoch": 1.955735262821877, "grad_norm": 12.796875, "learning_rate": 9.364695849914726e-06, "loss": 1.2599, "step": 18380 }, { "epoch": 1.9558416684400937, "grad_norm": 10.7734375, "learning_rate": 9.364660318362708e-06, "loss": 1.0949, "step": 18381 }, { "epoch": 1.9559480740583104, "grad_norm": 16.578125, "learning_rate": 9.36462478681069e-06, "loss": 1.0692, "step": 18382 }, { "epoch": 1.956054479676527, "grad_norm": 10.3203125, "learning_rate": 9.364589255258671e-06, "loss": 0.9521, "step": 18383 }, { "epoch": 1.9561608852947434, "grad_norm": 13.6640625, "learning_rate": 9.364553723706653e-06, "loss": 1.0462, "step": 18384 }, { "epoch": 1.9562672909129604, "grad_norm": 15.7734375, "learning_rate": 9.364518192154634e-06, "loss": 0.9255, "step": 18385 }, { "epoch": 1.9563736965311769, "grad_norm": 13.2265625, "learning_rate": 9.364482660602616e-06, "loss": 1.2558, "step": 18386 }, { "epoch": 1.9564801021493934, "grad_norm": 12.9296875, "learning_rate": 9.364447129050597e-06, "loss": 1.0672, "step": 18387 }, { "epoch": 1.9565865077676101, "grad_norm": 20.953125, "learning_rate": 9.364411597498579e-06, "loss": 1.1574, "step": 18388 }, { "epoch": 1.9566929133858268, "grad_norm": 11.8828125, "learning_rate": 9.36437606594656e-06, "loss": 1.1589, "step": 18389 }, { "epoch": 1.9567993190040434, "grad_norm": 10.609375, "learning_rate": 9.364340534394542e-06, "loss": 1.1335, "step": 18390 }, { "epoch": 1.95690572462226, "grad_norm": 11.875, "learning_rate": 9.364305002842525e-06, "loss": 0.9709, "step": 18391 }, { "epoch": 1.9570121302404768, "grad_norm": 16.046875, "learning_rate": 9.364269471290507e-06, "loss": 0.9126, "step": 18392 }, { "epoch": 1.9571185358586933, "grad_norm": 10.40625, "learning_rate": 9.364233939738489e-06, "loss": 1.2066, "step": 18393 }, { "epoch": 1.9572249414769098, "grad_norm": 9.625, "learning_rate": 9.36419840818647e-06, "loss": 0.8778, "step": 18394 }, { "epoch": 1.9573313470951266, "grad_norm": 12.3359375, "learning_rate": 9.364162876634452e-06, "loss": 0.9773, "step": 18395 }, { "epoch": 1.9574377527133433, "grad_norm": 13.7265625, "learning_rate": 9.364127345082435e-06, "loss": 1.1466, "step": 18396 }, { "epoch": 1.9575441583315598, "grad_norm": 10.328125, "learning_rate": 9.364091813530417e-06, "loss": 0.9168, "step": 18397 }, { "epoch": 1.9576505639497765, "grad_norm": 20.40625, "learning_rate": 9.364056281978397e-06, "loss": 0.9743, "step": 18398 }, { "epoch": 1.9577569695679933, "grad_norm": 10.8125, "learning_rate": 9.364020750426378e-06, "loss": 1.1446, "step": 18399 }, { "epoch": 1.9578633751862098, "grad_norm": 13.015625, "learning_rate": 9.36398521887436e-06, "loss": 1.0941, "step": 18400 }, { "epoch": 1.9579697808044265, "grad_norm": 11.1015625, "learning_rate": 9.363949687322343e-06, "loss": 1.0767, "step": 18401 }, { "epoch": 1.9580761864226433, "grad_norm": 14.2421875, "learning_rate": 9.363914155770325e-06, "loss": 1.227, "step": 18402 }, { "epoch": 1.9581825920408598, "grad_norm": 13.109375, "learning_rate": 9.363878624218306e-06, "loss": 1.0018, "step": 18403 }, { "epoch": 1.9582889976590763, "grad_norm": 11.8203125, "learning_rate": 9.363843092666288e-06, "loss": 1.1027, "step": 18404 }, { "epoch": 1.958395403277293, "grad_norm": 13.3125, "learning_rate": 9.36380756111427e-06, "loss": 1.0267, "step": 18405 }, { "epoch": 1.9585018088955097, "grad_norm": 12.8203125, "learning_rate": 9.363772029562253e-06, "loss": 1.1765, "step": 18406 }, { "epoch": 1.9586082145137262, "grad_norm": 15.0078125, "learning_rate": 9.363736498010235e-06, "loss": 1.1373, "step": 18407 }, { "epoch": 1.958714620131943, "grad_norm": 13.0390625, "learning_rate": 9.363700966458216e-06, "loss": 1.3692, "step": 18408 }, { "epoch": 1.9588210257501597, "grad_norm": 14.2421875, "learning_rate": 9.363665434906198e-06, "loss": 0.9788, "step": 18409 }, { "epoch": 1.9589274313683762, "grad_norm": 10.6640625, "learning_rate": 9.36362990335418e-06, "loss": 1.0411, "step": 18410 }, { "epoch": 1.959033836986593, "grad_norm": 11.34375, "learning_rate": 9.363594371802161e-06, "loss": 1.0004, "step": 18411 }, { "epoch": 1.9591402426048097, "grad_norm": 15.625, "learning_rate": 9.363558840250143e-06, "loss": 0.9686, "step": 18412 }, { "epoch": 1.9592466482230262, "grad_norm": 10.4921875, "learning_rate": 9.363523308698124e-06, "loss": 0.9254, "step": 18413 }, { "epoch": 1.9593530538412427, "grad_norm": 12.8125, "learning_rate": 9.363487777146106e-06, "loss": 1.0563, "step": 18414 }, { "epoch": 1.9594594594594594, "grad_norm": 11.1015625, "learning_rate": 9.363452245594087e-06, "loss": 1.1745, "step": 18415 }, { "epoch": 1.9595658650776762, "grad_norm": 16.5, "learning_rate": 9.36341671404207e-06, "loss": 1.0524, "step": 18416 }, { "epoch": 1.9596722706958927, "grad_norm": 12.859375, "learning_rate": 9.363381182490052e-06, "loss": 1.2378, "step": 18417 }, { "epoch": 1.9597786763141094, "grad_norm": 12.9375, "learning_rate": 9.363345650938034e-06, "loss": 1.3381, "step": 18418 }, { "epoch": 1.9598850819323261, "grad_norm": 14.640625, "learning_rate": 9.363310119386016e-06, "loss": 0.9387, "step": 18419 }, { "epoch": 1.9599914875505426, "grad_norm": 13.4453125, "learning_rate": 9.363274587833997e-06, "loss": 1.1821, "step": 18420 }, { "epoch": 1.9600978931687592, "grad_norm": 14.8671875, "learning_rate": 9.36323905628198e-06, "loss": 1.1099, "step": 18421 }, { "epoch": 1.960204298786976, "grad_norm": 12.265625, "learning_rate": 9.363203524729962e-06, "loss": 0.9892, "step": 18422 }, { "epoch": 1.9603107044051926, "grad_norm": 12.3359375, "learning_rate": 9.363167993177944e-06, "loss": 0.9999, "step": 18423 }, { "epoch": 1.9604171100234091, "grad_norm": 13.4140625, "learning_rate": 9.363132461625925e-06, "loss": 0.9421, "step": 18424 }, { "epoch": 1.9605235156416259, "grad_norm": 18.5625, "learning_rate": 9.363096930073907e-06, "loss": 1.0591, "step": 18425 }, { "epoch": 1.9606299212598426, "grad_norm": 15.9765625, "learning_rate": 9.363061398521888e-06, "loss": 1.0775, "step": 18426 }, { "epoch": 1.960736326878059, "grad_norm": 9.6484375, "learning_rate": 9.36302586696987e-06, "loss": 0.9684, "step": 18427 }, { "epoch": 1.9608427324962758, "grad_norm": 10.3984375, "learning_rate": 9.362990335417852e-06, "loss": 1.1554, "step": 18428 }, { "epoch": 1.9609491381144926, "grad_norm": 13.953125, "learning_rate": 9.362954803865833e-06, "loss": 0.8258, "step": 18429 }, { "epoch": 1.961055543732709, "grad_norm": 14.5078125, "learning_rate": 9.362919272313815e-06, "loss": 0.99, "step": 18430 }, { "epoch": 1.9611619493509256, "grad_norm": 13.421875, "learning_rate": 9.362883740761797e-06, "loss": 1.1702, "step": 18431 }, { "epoch": 1.9612683549691425, "grad_norm": 13.9296875, "learning_rate": 9.36284820920978e-06, "loss": 1.2814, "step": 18432 }, { "epoch": 1.961374760587359, "grad_norm": 9.5625, "learning_rate": 9.362812677657761e-06, "loss": 0.8151, "step": 18433 }, { "epoch": 1.9614811662055756, "grad_norm": 14.4453125, "learning_rate": 9.362777146105743e-06, "loss": 1.1231, "step": 18434 }, { "epoch": 1.9615875718237923, "grad_norm": 8.75, "learning_rate": 9.362741614553725e-06, "loss": 0.7271, "step": 18435 }, { "epoch": 1.961693977442009, "grad_norm": 9.8828125, "learning_rate": 9.362706083001706e-06, "loss": 0.9985, "step": 18436 }, { "epoch": 1.9618003830602255, "grad_norm": 13.0, "learning_rate": 9.362670551449688e-06, "loss": 1.2778, "step": 18437 }, { "epoch": 1.9619067886784423, "grad_norm": 12.3828125, "learning_rate": 9.36263501989767e-06, "loss": 1.1909, "step": 18438 }, { "epoch": 1.962013194296659, "grad_norm": 10.984375, "learning_rate": 9.362599488345651e-06, "loss": 1.1687, "step": 18439 }, { "epoch": 1.9621195999148755, "grad_norm": 9.9765625, "learning_rate": 9.362563956793633e-06, "loss": 1.132, "step": 18440 }, { "epoch": 1.962226005533092, "grad_norm": 13.9140625, "learning_rate": 9.362528425241614e-06, "loss": 0.9423, "step": 18441 }, { "epoch": 1.9623324111513087, "grad_norm": 10.2890625, "learning_rate": 9.362492893689598e-06, "loss": 1.0142, "step": 18442 }, { "epoch": 1.9624388167695255, "grad_norm": 9.453125, "learning_rate": 9.36245736213758e-06, "loss": 0.9356, "step": 18443 }, { "epoch": 1.962545222387742, "grad_norm": 10.2578125, "learning_rate": 9.36242183058556e-06, "loss": 1.2346, "step": 18444 }, { "epoch": 1.9626516280059587, "grad_norm": 8.5390625, "learning_rate": 9.362386299033542e-06, "loss": 1.2889, "step": 18445 }, { "epoch": 1.9627580336241754, "grad_norm": 10.6953125, "learning_rate": 9.362350767481524e-06, "loss": 1.3322, "step": 18446 }, { "epoch": 1.962864439242392, "grad_norm": 17.390625, "learning_rate": 9.362315235929507e-06, "loss": 1.172, "step": 18447 }, { "epoch": 1.9629708448606087, "grad_norm": 10.4453125, "learning_rate": 9.362279704377489e-06, "loss": 0.9751, "step": 18448 }, { "epoch": 1.9630772504788254, "grad_norm": 12.65625, "learning_rate": 9.36224417282547e-06, "loss": 1.1941, "step": 18449 }, { "epoch": 1.963183656097042, "grad_norm": 12.0078125, "learning_rate": 9.362208641273452e-06, "loss": 1.2637, "step": 18450 }, { "epoch": 1.9632900617152584, "grad_norm": 14.453125, "learning_rate": 9.362173109721434e-06, "loss": 0.9898, "step": 18451 }, { "epoch": 1.9633964673334752, "grad_norm": 14.6484375, "learning_rate": 9.362137578169415e-06, "loss": 0.8501, "step": 18452 }, { "epoch": 1.963502872951692, "grad_norm": 12.5546875, "learning_rate": 9.362102046617397e-06, "loss": 0.9149, "step": 18453 }, { "epoch": 1.9636092785699084, "grad_norm": 12.71875, "learning_rate": 9.362066515065379e-06, "loss": 1.2024, "step": 18454 }, { "epoch": 1.9637156841881251, "grad_norm": 11.84375, "learning_rate": 9.36203098351336e-06, "loss": 1.1208, "step": 18455 }, { "epoch": 1.9638220898063419, "grad_norm": 10.9765625, "learning_rate": 9.361995451961342e-06, "loss": 1.085, "step": 18456 }, { "epoch": 1.9639284954245584, "grad_norm": 10.6953125, "learning_rate": 9.361959920409325e-06, "loss": 1.3543, "step": 18457 }, { "epoch": 1.964034901042775, "grad_norm": 11.359375, "learning_rate": 9.361924388857307e-06, "loss": 1.1242, "step": 18458 }, { "epoch": 1.9641413066609918, "grad_norm": 11.0390625, "learning_rate": 9.361888857305288e-06, "loss": 1.3249, "step": 18459 }, { "epoch": 1.9642477122792084, "grad_norm": 10.6171875, "learning_rate": 9.36185332575327e-06, "loss": 0.9413, "step": 18460 }, { "epoch": 1.9643541178974249, "grad_norm": 11.78125, "learning_rate": 9.361817794201251e-06, "loss": 1.0427, "step": 18461 }, { "epoch": 1.9644605235156416, "grad_norm": 14.359375, "learning_rate": 9.361782262649233e-06, "loss": 1.2051, "step": 18462 }, { "epoch": 1.9645669291338583, "grad_norm": 14.59375, "learning_rate": 9.361746731097215e-06, "loss": 1.2762, "step": 18463 }, { "epoch": 1.9646733347520748, "grad_norm": 11.78125, "learning_rate": 9.361711199545196e-06, "loss": 1.057, "step": 18464 }, { "epoch": 1.9647797403702916, "grad_norm": 11.1640625, "learning_rate": 9.361675667993178e-06, "loss": 1.0915, "step": 18465 }, { "epoch": 1.9648861459885083, "grad_norm": 10.921875, "learning_rate": 9.36164013644116e-06, "loss": 0.9231, "step": 18466 }, { "epoch": 1.9649925516067248, "grad_norm": 13.890625, "learning_rate": 9.361604604889141e-06, "loss": 1.2194, "step": 18467 }, { "epoch": 1.9650989572249413, "grad_norm": 11.3046875, "learning_rate": 9.361569073337124e-06, "loss": 1.016, "step": 18468 }, { "epoch": 1.9652053628431583, "grad_norm": 8.7890625, "learning_rate": 9.361533541785106e-06, "loss": 0.9923, "step": 18469 }, { "epoch": 1.9653117684613748, "grad_norm": 14.1875, "learning_rate": 9.361498010233088e-06, "loss": 0.9203, "step": 18470 }, { "epoch": 1.9654181740795913, "grad_norm": 11.390625, "learning_rate": 9.36146247868107e-06, "loss": 1.108, "step": 18471 }, { "epoch": 1.965524579697808, "grad_norm": 18.125, "learning_rate": 9.361426947129051e-06, "loss": 1.2331, "step": 18472 }, { "epoch": 1.9656309853160248, "grad_norm": 11.671875, "learning_rate": 9.361391415577034e-06, "loss": 1.0131, "step": 18473 }, { "epoch": 1.9657373909342413, "grad_norm": 14.9296875, "learning_rate": 9.361355884025016e-06, "loss": 0.9886, "step": 18474 }, { "epoch": 1.965843796552458, "grad_norm": 10.9765625, "learning_rate": 9.361320352472997e-06, "loss": 0.9967, "step": 18475 }, { "epoch": 1.9659502021706747, "grad_norm": 12.5625, "learning_rate": 9.361284820920979e-06, "loss": 0.8809, "step": 18476 }, { "epoch": 1.9660566077888912, "grad_norm": 11.7734375, "learning_rate": 9.36124928936896e-06, "loss": 0.9556, "step": 18477 }, { "epoch": 1.9661630134071078, "grad_norm": 11.1796875, "learning_rate": 9.361213757816942e-06, "loss": 1.1351, "step": 18478 }, { "epoch": 1.9662694190253245, "grad_norm": 9.3515625, "learning_rate": 9.361178226264924e-06, "loss": 0.7359, "step": 18479 }, { "epoch": 1.9663758246435412, "grad_norm": 14.0859375, "learning_rate": 9.361142694712905e-06, "loss": 1.0598, "step": 18480 }, { "epoch": 1.9664822302617577, "grad_norm": 16.46875, "learning_rate": 9.361107163160887e-06, "loss": 1.0312, "step": 18481 }, { "epoch": 1.9665886358799745, "grad_norm": 13.625, "learning_rate": 9.361071631608869e-06, "loss": 1.0619, "step": 18482 }, { "epoch": 1.9666950414981912, "grad_norm": 14.1015625, "learning_rate": 9.361036100056852e-06, "loss": 1.3119, "step": 18483 }, { "epoch": 1.9668014471164077, "grad_norm": 11.9453125, "learning_rate": 9.361000568504834e-06, "loss": 1.2913, "step": 18484 }, { "epoch": 1.9669078527346244, "grad_norm": 12.515625, "learning_rate": 9.360965036952815e-06, "loss": 1.1604, "step": 18485 }, { "epoch": 1.9670142583528412, "grad_norm": 9.71875, "learning_rate": 9.360929505400797e-06, "loss": 1.1708, "step": 18486 }, { "epoch": 1.9671206639710577, "grad_norm": 11.765625, "learning_rate": 9.360893973848778e-06, "loss": 1.2594, "step": 18487 }, { "epoch": 1.9672270695892742, "grad_norm": 14.765625, "learning_rate": 9.36085844229676e-06, "loss": 1.2132, "step": 18488 }, { "epoch": 1.967333475207491, "grad_norm": 11.125, "learning_rate": 9.360822910744742e-06, "loss": 1.1106, "step": 18489 }, { "epoch": 1.9674398808257076, "grad_norm": 10.5859375, "learning_rate": 9.360787379192723e-06, "loss": 0.9204, "step": 18490 }, { "epoch": 1.9675462864439242, "grad_norm": 21.765625, "learning_rate": 9.360751847640705e-06, "loss": 1.1952, "step": 18491 }, { "epoch": 1.9676526920621409, "grad_norm": 11.1328125, "learning_rate": 9.360716316088686e-06, "loss": 1.1477, "step": 18492 }, { "epoch": 1.9677590976803576, "grad_norm": 11.4921875, "learning_rate": 9.36068078453667e-06, "loss": 1.1947, "step": 18493 }, { "epoch": 1.9678655032985741, "grad_norm": 13.2890625, "learning_rate": 9.360645252984651e-06, "loss": 1.648, "step": 18494 }, { "epoch": 1.9679719089167909, "grad_norm": 14.28125, "learning_rate": 9.360609721432633e-06, "loss": 1.2347, "step": 18495 }, { "epoch": 1.9680783145350076, "grad_norm": 13.515625, "learning_rate": 9.360574189880614e-06, "loss": 1.0033, "step": 18496 }, { "epoch": 1.968184720153224, "grad_norm": 22.78125, "learning_rate": 9.360538658328596e-06, "loss": 1.0567, "step": 18497 }, { "epoch": 1.9682911257714406, "grad_norm": 11.875, "learning_rate": 9.36050312677658e-06, "loss": 0.9426, "step": 18498 }, { "epoch": 1.9683975313896573, "grad_norm": 12.7109375, "learning_rate": 9.360467595224561e-06, "loss": 0.9756, "step": 18499 }, { "epoch": 1.968503937007874, "grad_norm": 16.734375, "learning_rate": 9.360432063672543e-06, "loss": 1.0572, "step": 18500 }, { "epoch": 1.9686103426260906, "grad_norm": 15.0703125, "learning_rate": 9.360396532120524e-06, "loss": 0.8882, "step": 18501 }, { "epoch": 1.9687167482443073, "grad_norm": 16.15625, "learning_rate": 9.360361000568506e-06, "loss": 0.9301, "step": 18502 }, { "epoch": 1.968823153862524, "grad_norm": 11.5234375, "learning_rate": 9.360325469016487e-06, "loss": 1.2152, "step": 18503 }, { "epoch": 1.9689295594807406, "grad_norm": 13.8046875, "learning_rate": 9.360289937464469e-06, "loss": 0.7599, "step": 18504 }, { "epoch": 1.969035965098957, "grad_norm": 11.1953125, "learning_rate": 9.36025440591245e-06, "loss": 1.2577, "step": 18505 }, { "epoch": 1.969142370717174, "grad_norm": 16.625, "learning_rate": 9.360218874360432e-06, "loss": 1.1032, "step": 18506 }, { "epoch": 1.9692487763353905, "grad_norm": 11.6796875, "learning_rate": 9.360183342808414e-06, "loss": 1.27, "step": 18507 }, { "epoch": 1.969355181953607, "grad_norm": 10.15625, "learning_rate": 9.360147811256395e-06, "loss": 0.9497, "step": 18508 }, { "epoch": 1.9694615875718238, "grad_norm": 11.7734375, "learning_rate": 9.360112279704379e-06, "loss": 0.9539, "step": 18509 }, { "epoch": 1.9695679931900405, "grad_norm": 11.2734375, "learning_rate": 9.36007674815236e-06, "loss": 1.2027, "step": 18510 }, { "epoch": 1.969674398808257, "grad_norm": 13.3828125, "learning_rate": 9.360041216600342e-06, "loss": 1.1068, "step": 18511 }, { "epoch": 1.9697808044264737, "grad_norm": 11.8828125, "learning_rate": 9.360005685048324e-06, "loss": 1.3457, "step": 18512 }, { "epoch": 1.9698872100446905, "grad_norm": 12.59375, "learning_rate": 9.359970153496305e-06, "loss": 1.0221, "step": 18513 }, { "epoch": 1.969993615662907, "grad_norm": 9.7421875, "learning_rate": 9.359934621944288e-06, "loss": 1.0513, "step": 18514 }, { "epoch": 1.9701000212811235, "grad_norm": 10.734375, "learning_rate": 9.35989909039227e-06, "loss": 0.9698, "step": 18515 }, { "epoch": 1.9702064268993404, "grad_norm": 11.4296875, "learning_rate": 9.359863558840252e-06, "loss": 1.3317, "step": 18516 }, { "epoch": 1.970312832517557, "grad_norm": 13.140625, "learning_rate": 9.359828027288233e-06, "loss": 1.3019, "step": 18517 }, { "epoch": 1.9704192381357735, "grad_norm": 13.5, "learning_rate": 9.359792495736213e-06, "loss": 1.0333, "step": 18518 }, { "epoch": 1.9705256437539902, "grad_norm": 12.4296875, "learning_rate": 9.359756964184197e-06, "loss": 0.8879, "step": 18519 }, { "epoch": 1.970632049372207, "grad_norm": 12.0390625, "learning_rate": 9.359721432632178e-06, "loss": 1.1168, "step": 18520 }, { "epoch": 1.9707384549904234, "grad_norm": 11.7109375, "learning_rate": 9.35968590108016e-06, "loss": 1.2132, "step": 18521 }, { "epoch": 1.9708448606086402, "grad_norm": 10.5078125, "learning_rate": 9.359650369528141e-06, "loss": 0.9549, "step": 18522 }, { "epoch": 1.970951266226857, "grad_norm": 11.2265625, "learning_rate": 9.359614837976123e-06, "loss": 1.0082, "step": 18523 }, { "epoch": 1.9710576718450734, "grad_norm": 11.2734375, "learning_rate": 9.359579306424106e-06, "loss": 1.1722, "step": 18524 }, { "epoch": 1.97116407746329, "grad_norm": 14.046875, "learning_rate": 9.359543774872088e-06, "loss": 1.1422, "step": 18525 }, { "epoch": 1.9712704830815067, "grad_norm": 11.234375, "learning_rate": 9.35950824332007e-06, "loss": 1.0926, "step": 18526 }, { "epoch": 1.9713768886997234, "grad_norm": 12.984375, "learning_rate": 9.359472711768051e-06, "loss": 1.0184, "step": 18527 }, { "epoch": 1.97148329431794, "grad_norm": 11.21875, "learning_rate": 9.359437180216033e-06, "loss": 1.0371, "step": 18528 }, { "epoch": 1.9715896999361566, "grad_norm": 16.453125, "learning_rate": 9.359401648664014e-06, "loss": 1.0586, "step": 18529 }, { "epoch": 1.9716961055543734, "grad_norm": 11.8515625, "learning_rate": 9.359366117111996e-06, "loss": 1.1167, "step": 18530 }, { "epoch": 1.9718025111725899, "grad_norm": 10.7578125, "learning_rate": 9.359330585559978e-06, "loss": 1.2257, "step": 18531 }, { "epoch": 1.9719089167908066, "grad_norm": 12.2890625, "learning_rate": 9.359295054007959e-06, "loss": 1.1988, "step": 18532 }, { "epoch": 1.9720153224090233, "grad_norm": 20.765625, "learning_rate": 9.35925952245594e-06, "loss": 0.9932, "step": 18533 }, { "epoch": 1.9721217280272398, "grad_norm": 12.421875, "learning_rate": 9.359223990903924e-06, "loss": 1.1814, "step": 18534 }, { "epoch": 1.9722281336454563, "grad_norm": 12.6328125, "learning_rate": 9.359188459351906e-06, "loss": 1.388, "step": 18535 }, { "epoch": 1.972334539263673, "grad_norm": 12.703125, "learning_rate": 9.359152927799887e-06, "loss": 1.1933, "step": 18536 }, { "epoch": 1.9724409448818898, "grad_norm": 15.734375, "learning_rate": 9.359117396247869e-06, "loss": 0.9872, "step": 18537 }, { "epoch": 1.9725473505001063, "grad_norm": 13.1953125, "learning_rate": 9.35908186469585e-06, "loss": 1.1151, "step": 18538 }, { "epoch": 1.972653756118323, "grad_norm": 10.3515625, "learning_rate": 9.359046333143834e-06, "loss": 1.467, "step": 18539 }, { "epoch": 1.9727601617365398, "grad_norm": 15.1875, "learning_rate": 9.359010801591815e-06, "loss": 0.9991, "step": 18540 }, { "epoch": 1.9728665673547563, "grad_norm": 16.828125, "learning_rate": 9.358975270039797e-06, "loss": 0.8294, "step": 18541 }, { "epoch": 1.972972972972973, "grad_norm": 11.71875, "learning_rate": 9.358939738487779e-06, "loss": 1.2737, "step": 18542 }, { "epoch": 1.9730793785911898, "grad_norm": 15.8046875, "learning_rate": 9.35890420693576e-06, "loss": 0.9668, "step": 18543 }, { "epoch": 1.9731857842094063, "grad_norm": 13.2890625, "learning_rate": 9.358868675383742e-06, "loss": 0.9323, "step": 18544 }, { "epoch": 1.9732921898276228, "grad_norm": 13.1640625, "learning_rate": 9.358833143831723e-06, "loss": 1.1105, "step": 18545 }, { "epoch": 1.9733985954458395, "grad_norm": 18.203125, "learning_rate": 9.358797612279705e-06, "loss": 1.1004, "step": 18546 }, { "epoch": 1.9735050010640562, "grad_norm": 12.1328125, "learning_rate": 9.358762080727687e-06, "loss": 1.2666, "step": 18547 }, { "epoch": 1.9736114066822728, "grad_norm": 13.6171875, "learning_rate": 9.358726549175668e-06, "loss": 1.0063, "step": 18548 }, { "epoch": 1.9737178123004895, "grad_norm": 11.8515625, "learning_rate": 9.35869101762365e-06, "loss": 1.0544, "step": 18549 }, { "epoch": 1.9738242179187062, "grad_norm": 15.71875, "learning_rate": 9.358655486071633e-06, "loss": 1.1174, "step": 18550 }, { "epoch": 1.9739306235369227, "grad_norm": 13.2109375, "learning_rate": 9.358619954519615e-06, "loss": 1.1308, "step": 18551 }, { "epoch": 1.9740370291551392, "grad_norm": 10.7578125, "learning_rate": 9.358584422967596e-06, "loss": 1.253, "step": 18552 }, { "epoch": 1.9741434347733562, "grad_norm": 13.6015625, "learning_rate": 9.358548891415578e-06, "loss": 0.7888, "step": 18553 }, { "epoch": 1.9742498403915727, "grad_norm": 14.328125, "learning_rate": 9.35851335986356e-06, "loss": 0.9296, "step": 18554 }, { "epoch": 1.9743562460097892, "grad_norm": 9.71875, "learning_rate": 9.358477828311541e-06, "loss": 0.869, "step": 18555 }, { "epoch": 1.974462651628006, "grad_norm": 14.7890625, "learning_rate": 9.358442296759523e-06, "loss": 1.2199, "step": 18556 }, { "epoch": 1.9745690572462227, "grad_norm": 10.4765625, "learning_rate": 9.358406765207504e-06, "loss": 1.2663, "step": 18557 }, { "epoch": 1.9746754628644392, "grad_norm": 14.2265625, "learning_rate": 9.358371233655486e-06, "loss": 1.2045, "step": 18558 }, { "epoch": 1.974781868482656, "grad_norm": 12.7265625, "learning_rate": 9.358335702103468e-06, "loss": 1.0979, "step": 18559 }, { "epoch": 1.9748882741008726, "grad_norm": 16.59375, "learning_rate": 9.358300170551451e-06, "loss": 0.9671, "step": 18560 }, { "epoch": 1.9749946797190892, "grad_norm": 14.3359375, "learning_rate": 9.358264638999432e-06, "loss": 1.0702, "step": 18561 }, { "epoch": 1.9751010853373057, "grad_norm": 24.59375, "learning_rate": 9.358229107447414e-06, "loss": 0.853, "step": 18562 }, { "epoch": 1.9752074909555226, "grad_norm": 12.9140625, "learning_rate": 9.358193575895396e-06, "loss": 1.4197, "step": 18563 }, { "epoch": 1.9753138965737391, "grad_norm": 13.703125, "learning_rate": 9.358158044343377e-06, "loss": 1.077, "step": 18564 }, { "epoch": 1.9754203021919556, "grad_norm": 15.3671875, "learning_rate": 9.35812251279136e-06, "loss": 1.1395, "step": 18565 }, { "epoch": 1.9755267078101724, "grad_norm": 12.4609375, "learning_rate": 9.358086981239342e-06, "loss": 1.0854, "step": 18566 }, { "epoch": 1.975633113428389, "grad_norm": 16.40625, "learning_rate": 9.358051449687324e-06, "loss": 1.2076, "step": 18567 }, { "epoch": 1.9757395190466056, "grad_norm": 10.921875, "learning_rate": 9.358015918135305e-06, "loss": 0.9903, "step": 18568 }, { "epoch": 1.9758459246648223, "grad_norm": 18.828125, "learning_rate": 9.357980386583287e-06, "loss": 1.1032, "step": 18569 }, { "epoch": 1.975952330283039, "grad_norm": 16.828125, "learning_rate": 9.357944855031269e-06, "loss": 0.9189, "step": 18570 }, { "epoch": 1.9760587359012556, "grad_norm": 13.140625, "learning_rate": 9.35790932347925e-06, "loss": 0.9483, "step": 18571 }, { "epoch": 1.976165141519472, "grad_norm": 12.015625, "learning_rate": 9.357873791927232e-06, "loss": 0.9811, "step": 18572 }, { "epoch": 1.9762715471376888, "grad_norm": 10.1484375, "learning_rate": 9.357838260375213e-06, "loss": 0.7099, "step": 18573 }, { "epoch": 1.9763779527559056, "grad_norm": 16.59375, "learning_rate": 9.357802728823195e-06, "loss": 1.1952, "step": 18574 }, { "epoch": 1.976484358374122, "grad_norm": 12.546875, "learning_rate": 9.357767197271178e-06, "loss": 1.106, "step": 18575 }, { "epoch": 1.9765907639923388, "grad_norm": 13.1484375, "learning_rate": 9.35773166571916e-06, "loss": 0.9901, "step": 18576 }, { "epoch": 1.9766971696105555, "grad_norm": 17.34375, "learning_rate": 9.357696134167142e-06, "loss": 1.1181, "step": 18577 }, { "epoch": 1.976803575228772, "grad_norm": 11.578125, "learning_rate": 9.357660602615123e-06, "loss": 1.2308, "step": 18578 }, { "epoch": 1.9769099808469888, "grad_norm": 13.9296875, "learning_rate": 9.357625071063105e-06, "loss": 0.9949, "step": 18579 }, { "epoch": 1.9770163864652055, "grad_norm": 17.359375, "learning_rate": 9.357589539511086e-06, "loss": 1.1727, "step": 18580 }, { "epoch": 1.977122792083422, "grad_norm": 12.8125, "learning_rate": 9.357554007959068e-06, "loss": 0.8973, "step": 18581 }, { "epoch": 1.9772291977016385, "grad_norm": 13.15625, "learning_rate": 9.35751847640705e-06, "loss": 1.1425, "step": 18582 }, { "epoch": 1.9773356033198553, "grad_norm": 15.4609375, "learning_rate": 9.357482944855031e-06, "loss": 1.0639, "step": 18583 }, { "epoch": 1.977442008938072, "grad_norm": 15.125, "learning_rate": 9.357447413303013e-06, "loss": 1.0118, "step": 18584 }, { "epoch": 1.9775484145562885, "grad_norm": 13.5703125, "learning_rate": 9.357411881750994e-06, "loss": 1.0236, "step": 18585 }, { "epoch": 1.9776548201745052, "grad_norm": 11.0546875, "learning_rate": 9.357376350198978e-06, "loss": 1.1057, "step": 18586 }, { "epoch": 1.977761225792722, "grad_norm": 12.03125, "learning_rate": 9.35734081864696e-06, "loss": 0.9726, "step": 18587 }, { "epoch": 1.9778676314109385, "grad_norm": 12.390625, "learning_rate": 9.357305287094941e-06, "loss": 1.1188, "step": 18588 }, { "epoch": 1.977974037029155, "grad_norm": 12.5390625, "learning_rate": 9.357269755542923e-06, "loss": 0.8056, "step": 18589 }, { "epoch": 1.978080442647372, "grad_norm": 14.96875, "learning_rate": 9.357234223990904e-06, "loss": 1.0241, "step": 18590 }, { "epoch": 1.9781868482655884, "grad_norm": 12.140625, "learning_rate": 9.357198692438887e-06, "loss": 0.9059, "step": 18591 }, { "epoch": 1.978293253883805, "grad_norm": 12.1015625, "learning_rate": 9.357163160886869e-06, "loss": 0.9496, "step": 18592 }, { "epoch": 1.9783996595020217, "grad_norm": 10.859375, "learning_rate": 9.35712762933485e-06, "loss": 0.9487, "step": 18593 }, { "epoch": 1.9785060651202384, "grad_norm": 15.2421875, "learning_rate": 9.357092097782832e-06, "loss": 1.1199, "step": 18594 }, { "epoch": 1.978612470738455, "grad_norm": 12.9921875, "learning_rate": 9.357056566230814e-06, "loss": 0.9552, "step": 18595 }, { "epoch": 1.9787188763566717, "grad_norm": 12.7578125, "learning_rate": 9.357021034678795e-06, "loss": 0.945, "step": 18596 }, { "epoch": 1.9788252819748884, "grad_norm": 9.1171875, "learning_rate": 9.356985503126777e-06, "loss": 0.9003, "step": 18597 }, { "epoch": 1.978931687593105, "grad_norm": 15.65625, "learning_rate": 9.356949971574759e-06, "loss": 1.1633, "step": 18598 }, { "epoch": 1.9790380932113214, "grad_norm": 14.3515625, "learning_rate": 9.35691444002274e-06, "loss": 1.1144, "step": 18599 }, { "epoch": 1.9791444988295384, "grad_norm": 12.1875, "learning_rate": 9.356878908470722e-06, "loss": 1.1589, "step": 18600 }, { "epoch": 1.9792509044477549, "grad_norm": 10.0703125, "learning_rate": 9.356843376918705e-06, "loss": 0.9444, "step": 18601 }, { "epoch": 1.9793573100659714, "grad_norm": 14.1953125, "learning_rate": 9.356807845366687e-06, "loss": 1.0942, "step": 18602 }, { "epoch": 1.979463715684188, "grad_norm": 14.1796875, "learning_rate": 9.356772313814668e-06, "loss": 1.1581, "step": 18603 }, { "epoch": 1.9795701213024048, "grad_norm": 9.71875, "learning_rate": 9.35673678226265e-06, "loss": 0.9377, "step": 18604 }, { "epoch": 1.9796765269206213, "grad_norm": 16.984375, "learning_rate": 9.356701250710632e-06, "loss": 1.0732, "step": 18605 }, { "epoch": 1.979782932538838, "grad_norm": 11.7265625, "learning_rate": 9.356665719158615e-06, "loss": 1.2874, "step": 18606 }, { "epoch": 1.9798893381570548, "grad_norm": 9.4140625, "learning_rate": 9.356630187606595e-06, "loss": 1.0247, "step": 18607 }, { "epoch": 1.9799957437752713, "grad_norm": 10.328125, "learning_rate": 9.356594656054576e-06, "loss": 0.9669, "step": 18608 }, { "epoch": 1.9801021493934878, "grad_norm": 11.7421875, "learning_rate": 9.356559124502558e-06, "loss": 1.0946, "step": 18609 }, { "epoch": 1.9802085550117046, "grad_norm": 12.890625, "learning_rate": 9.35652359295054e-06, "loss": 1.1672, "step": 18610 }, { "epoch": 1.9803149606299213, "grad_norm": 15.9609375, "learning_rate": 9.356488061398523e-06, "loss": 1.2856, "step": 18611 }, { "epoch": 1.9804213662481378, "grad_norm": 9.5859375, "learning_rate": 9.356452529846505e-06, "loss": 1.172, "step": 18612 }, { "epoch": 1.9805277718663545, "grad_norm": 16.59375, "learning_rate": 9.356416998294486e-06, "loss": 1.3639, "step": 18613 }, { "epoch": 1.9806341774845713, "grad_norm": 17.109375, "learning_rate": 9.356381466742468e-06, "loss": 1.2294, "step": 18614 }, { "epoch": 1.9807405831027878, "grad_norm": 14.71875, "learning_rate": 9.35634593519045e-06, "loss": 1.0402, "step": 18615 }, { "epoch": 1.9808469887210045, "grad_norm": 11.5078125, "learning_rate": 9.356310403638433e-06, "loss": 1.0643, "step": 18616 }, { "epoch": 1.9809533943392212, "grad_norm": 12.1953125, "learning_rate": 9.356274872086414e-06, "loss": 0.8866, "step": 18617 }, { "epoch": 1.9810597999574378, "grad_norm": 16.484375, "learning_rate": 9.356239340534396e-06, "loss": 1.272, "step": 18618 }, { "epoch": 1.9811662055756543, "grad_norm": 10.828125, "learning_rate": 9.356203808982378e-06, "loss": 0.9391, "step": 18619 }, { "epoch": 1.981272611193871, "grad_norm": 11.6484375, "learning_rate": 9.356168277430359e-06, "loss": 0.9778, "step": 18620 }, { "epoch": 1.9813790168120877, "grad_norm": 12.3828125, "learning_rate": 9.35613274587834e-06, "loss": 1.155, "step": 18621 }, { "epoch": 1.9814854224303042, "grad_norm": 21.21875, "learning_rate": 9.356097214326322e-06, "loss": 1.0635, "step": 18622 }, { "epoch": 1.981591828048521, "grad_norm": 17.546875, "learning_rate": 9.356061682774304e-06, "loss": 1.0461, "step": 18623 }, { "epoch": 1.9816982336667377, "grad_norm": 13.2890625, "learning_rate": 9.356026151222286e-06, "loss": 1.0165, "step": 18624 }, { "epoch": 1.9818046392849542, "grad_norm": 16.859375, "learning_rate": 9.355990619670267e-06, "loss": 1.0751, "step": 18625 }, { "epoch": 1.981911044903171, "grad_norm": 12.65625, "learning_rate": 9.355955088118249e-06, "loss": 1.1144, "step": 18626 }, { "epoch": 1.9820174505213877, "grad_norm": 12.0703125, "learning_rate": 9.355919556566232e-06, "loss": 0.9511, "step": 18627 }, { "epoch": 1.9821238561396042, "grad_norm": 11.96875, "learning_rate": 9.355884025014214e-06, "loss": 1.0904, "step": 18628 }, { "epoch": 1.9822302617578207, "grad_norm": 11.71875, "learning_rate": 9.355848493462195e-06, "loss": 1.1004, "step": 18629 }, { "epoch": 1.9823366673760374, "grad_norm": 14.375, "learning_rate": 9.355812961910177e-06, "loss": 1.1091, "step": 18630 }, { "epoch": 1.9824430729942542, "grad_norm": 16.1875, "learning_rate": 9.355777430358158e-06, "loss": 1.0719, "step": 18631 }, { "epoch": 1.9825494786124707, "grad_norm": 11.015625, "learning_rate": 9.355741898806142e-06, "loss": 1.1646, "step": 18632 }, { "epoch": 1.9826558842306874, "grad_norm": 12.28125, "learning_rate": 9.355706367254123e-06, "loss": 0.9332, "step": 18633 }, { "epoch": 1.9827622898489041, "grad_norm": 13.78125, "learning_rate": 9.355670835702105e-06, "loss": 0.9936, "step": 18634 }, { "epoch": 1.9828686954671206, "grad_norm": 15.9296875, "learning_rate": 9.355635304150087e-06, "loss": 1.0898, "step": 18635 }, { "epoch": 1.9829751010853371, "grad_norm": 17.171875, "learning_rate": 9.355599772598068e-06, "loss": 1.143, "step": 18636 }, { "epoch": 1.983081506703554, "grad_norm": 11.171875, "learning_rate": 9.35556424104605e-06, "loss": 1.1951, "step": 18637 }, { "epoch": 1.9831879123217706, "grad_norm": 12.890625, "learning_rate": 9.355528709494031e-06, "loss": 1.2233, "step": 18638 }, { "epoch": 1.9832943179399871, "grad_norm": 14.453125, "learning_rate": 9.355493177942013e-06, "loss": 1.0186, "step": 18639 }, { "epoch": 1.9834007235582038, "grad_norm": 15.515625, "learning_rate": 9.355457646389995e-06, "loss": 1.2635, "step": 18640 }, { "epoch": 1.9835071291764206, "grad_norm": 12.4140625, "learning_rate": 9.355422114837976e-06, "loss": 1.1254, "step": 18641 }, { "epoch": 1.983613534794637, "grad_norm": 13.2265625, "learning_rate": 9.35538658328596e-06, "loss": 1.0123, "step": 18642 }, { "epoch": 1.9837199404128538, "grad_norm": 9.5, "learning_rate": 9.355351051733941e-06, "loss": 1.0194, "step": 18643 }, { "epoch": 1.9838263460310706, "grad_norm": 13.765625, "learning_rate": 9.355315520181923e-06, "loss": 1.0948, "step": 18644 }, { "epoch": 1.983932751649287, "grad_norm": 11.6015625, "learning_rate": 9.355279988629904e-06, "loss": 1.1027, "step": 18645 }, { "epoch": 1.9840391572675036, "grad_norm": 11.4140625, "learning_rate": 9.355244457077886e-06, "loss": 0.9749, "step": 18646 }, { "epoch": 1.9841455628857205, "grad_norm": 10.921875, "learning_rate": 9.355208925525868e-06, "loss": 1.1021, "step": 18647 }, { "epoch": 1.984251968503937, "grad_norm": 11.5859375, "learning_rate": 9.35517339397385e-06, "loss": 1.1169, "step": 18648 }, { "epoch": 1.9843583741221535, "grad_norm": 23.84375, "learning_rate": 9.35513786242183e-06, "loss": 1.0201, "step": 18649 }, { "epoch": 1.9844647797403703, "grad_norm": 13.65625, "learning_rate": 9.355102330869812e-06, "loss": 1.0485, "step": 18650 }, { "epoch": 1.984571185358587, "grad_norm": 14.671875, "learning_rate": 9.355066799317794e-06, "loss": 1.2169, "step": 18651 }, { "epoch": 1.9846775909768035, "grad_norm": 22.4375, "learning_rate": 9.355031267765777e-06, "loss": 1.1572, "step": 18652 }, { "epoch": 1.9847839965950203, "grad_norm": 10.71875, "learning_rate": 9.354995736213759e-06, "loss": 1.3136, "step": 18653 }, { "epoch": 1.984890402213237, "grad_norm": 11.421875, "learning_rate": 9.35496020466174e-06, "loss": 1.0178, "step": 18654 }, { "epoch": 1.9849968078314535, "grad_norm": 12.984375, "learning_rate": 9.354924673109722e-06, "loss": 1.2007, "step": 18655 }, { "epoch": 1.98510321344967, "grad_norm": 9.28125, "learning_rate": 9.354889141557704e-06, "loss": 1.0767, "step": 18656 }, { "epoch": 1.9852096190678867, "grad_norm": 13.71875, "learning_rate": 9.354853610005685e-06, "loss": 0.9813, "step": 18657 }, { "epoch": 1.9853160246861035, "grad_norm": 11.578125, "learning_rate": 9.354818078453669e-06, "loss": 1.1247, "step": 18658 }, { "epoch": 1.98542243030432, "grad_norm": 11.53125, "learning_rate": 9.35478254690165e-06, "loss": 0.9474, "step": 18659 }, { "epoch": 1.9855288359225367, "grad_norm": 9.921875, "learning_rate": 9.354747015349632e-06, "loss": 1.009, "step": 18660 }, { "epoch": 1.9856352415407534, "grad_norm": 15.5546875, "learning_rate": 9.354711483797613e-06, "loss": 1.1044, "step": 18661 }, { "epoch": 1.98574164715897, "grad_norm": 10.9296875, "learning_rate": 9.354675952245595e-06, "loss": 0.9765, "step": 18662 }, { "epoch": 1.9858480527771867, "grad_norm": 12.2421875, "learning_rate": 9.354640420693577e-06, "loss": 1.1753, "step": 18663 }, { "epoch": 1.9859544583954034, "grad_norm": 12.5625, "learning_rate": 9.354604889141558e-06, "loss": 1.156, "step": 18664 }, { "epoch": 1.98606086401362, "grad_norm": 12.0546875, "learning_rate": 9.35456935758954e-06, "loss": 1.1046, "step": 18665 }, { "epoch": 1.9861672696318364, "grad_norm": 15.53125, "learning_rate": 9.354533826037522e-06, "loss": 1.067, "step": 18666 }, { "epoch": 1.9862736752500532, "grad_norm": 13.375, "learning_rate": 9.354498294485503e-06, "loss": 1.1428, "step": 18667 }, { "epoch": 1.98638008086827, "grad_norm": 11.1796875, "learning_rate": 9.354462762933486e-06, "loss": 0.9152, "step": 18668 }, { "epoch": 1.9864864864864864, "grad_norm": 14.71875, "learning_rate": 9.354427231381468e-06, "loss": 1.0729, "step": 18669 }, { "epoch": 1.9865928921047031, "grad_norm": 14.8359375, "learning_rate": 9.35439169982945e-06, "loss": 1.2034, "step": 18670 }, { "epoch": 1.9866992977229199, "grad_norm": 10.0625, "learning_rate": 9.354356168277431e-06, "loss": 1.0985, "step": 18671 }, { "epoch": 1.9868057033411364, "grad_norm": 12.84375, "learning_rate": 9.354320636725413e-06, "loss": 0.8927, "step": 18672 }, { "epoch": 1.986912108959353, "grad_norm": 16.796875, "learning_rate": 9.354285105173394e-06, "loss": 0.7541, "step": 18673 }, { "epoch": 1.9870185145775698, "grad_norm": 11.078125, "learning_rate": 9.354249573621376e-06, "loss": 1.0375, "step": 18674 }, { "epoch": 1.9871249201957863, "grad_norm": 9.84375, "learning_rate": 9.354214042069358e-06, "loss": 1.0191, "step": 18675 }, { "epoch": 1.9872313258140029, "grad_norm": 11.75, "learning_rate": 9.35417851051734e-06, "loss": 0.8556, "step": 18676 }, { "epoch": 1.9873377314322196, "grad_norm": 10.625, "learning_rate": 9.354142978965321e-06, "loss": 1.1722, "step": 18677 }, { "epoch": 1.9874441370504363, "grad_norm": 15.6953125, "learning_rate": 9.354107447413304e-06, "loss": 1.2852, "step": 18678 }, { "epoch": 1.9875505426686528, "grad_norm": 12.9765625, "learning_rate": 9.354071915861286e-06, "loss": 1.1519, "step": 18679 }, { "epoch": 1.9876569482868696, "grad_norm": 9.7578125, "learning_rate": 9.354036384309267e-06, "loss": 0.9982, "step": 18680 }, { "epoch": 1.9877633539050863, "grad_norm": 14.6171875, "learning_rate": 9.354000852757249e-06, "loss": 0.9506, "step": 18681 }, { "epoch": 1.9878697595233028, "grad_norm": 11.3359375, "learning_rate": 9.35396532120523e-06, "loss": 1.1018, "step": 18682 }, { "epoch": 1.9879761651415193, "grad_norm": 15.5390625, "learning_rate": 9.353929789653214e-06, "loss": 1.0496, "step": 18683 }, { "epoch": 1.9880825707597363, "grad_norm": 9.7421875, "learning_rate": 9.353894258101196e-06, "loss": 1.0904, "step": 18684 }, { "epoch": 1.9881889763779528, "grad_norm": 11.8515625, "learning_rate": 9.353858726549177e-06, "loss": 0.9489, "step": 18685 }, { "epoch": 1.9882953819961693, "grad_norm": 14.0078125, "learning_rate": 9.353823194997159e-06, "loss": 1.2393, "step": 18686 }, { "epoch": 1.988401787614386, "grad_norm": 10.3515625, "learning_rate": 9.35378766344514e-06, "loss": 1.0972, "step": 18687 }, { "epoch": 1.9885081932326027, "grad_norm": 12.953125, "learning_rate": 9.353752131893122e-06, "loss": 1.1223, "step": 18688 }, { "epoch": 1.9886145988508193, "grad_norm": 12.484375, "learning_rate": 9.353716600341104e-06, "loss": 1.1283, "step": 18689 }, { "epoch": 1.988721004469036, "grad_norm": 12.828125, "learning_rate": 9.353681068789085e-06, "loss": 1.0244, "step": 18690 }, { "epoch": 1.9888274100872527, "grad_norm": 10.640625, "learning_rate": 9.353645537237067e-06, "loss": 1.0371, "step": 18691 }, { "epoch": 1.9889338157054692, "grad_norm": 10.78125, "learning_rate": 9.353610005685048e-06, "loss": 0.9958, "step": 18692 }, { "epoch": 1.9890402213236857, "grad_norm": 12.6875, "learning_rate": 9.353574474133032e-06, "loss": 1.0997, "step": 18693 }, { "epoch": 1.9891466269419027, "grad_norm": 9.875, "learning_rate": 9.353538942581013e-06, "loss": 1.0615, "step": 18694 }, { "epoch": 1.9892530325601192, "grad_norm": 14.7578125, "learning_rate": 9.353503411028995e-06, "loss": 1.0085, "step": 18695 }, { "epoch": 1.9893594381783357, "grad_norm": 12.1328125, "learning_rate": 9.353467879476976e-06, "loss": 1.1134, "step": 18696 }, { "epoch": 1.9894658437965524, "grad_norm": 9.8515625, "learning_rate": 9.353432347924958e-06, "loss": 0.9666, "step": 18697 }, { "epoch": 1.9895722494147692, "grad_norm": 10.8203125, "learning_rate": 9.35339681637294e-06, "loss": 0.9939, "step": 18698 }, { "epoch": 1.9896786550329857, "grad_norm": 13.8984375, "learning_rate": 9.353361284820921e-06, "loss": 1.115, "step": 18699 }, { "epoch": 1.9897850606512024, "grad_norm": 11.640625, "learning_rate": 9.353325753268903e-06, "loss": 1.0796, "step": 18700 }, { "epoch": 1.9898914662694192, "grad_norm": 14.140625, "learning_rate": 9.353290221716885e-06, "loss": 1.0439, "step": 18701 }, { "epoch": 1.9899978718876357, "grad_norm": 12.8046875, "learning_rate": 9.353254690164866e-06, "loss": 1.299, "step": 18702 }, { "epoch": 1.9901042775058522, "grad_norm": 18.71875, "learning_rate": 9.353219158612848e-06, "loss": 0.9204, "step": 18703 }, { "epoch": 1.990210683124069, "grad_norm": 13.203125, "learning_rate": 9.353183627060831e-06, "loss": 1.2389, "step": 18704 }, { "epoch": 1.9903170887422856, "grad_norm": 12.625, "learning_rate": 9.353148095508813e-06, "loss": 1.3936, "step": 18705 }, { "epoch": 1.9904234943605021, "grad_norm": 10.5546875, "learning_rate": 9.353112563956794e-06, "loss": 1.1089, "step": 18706 }, { "epoch": 1.9905298999787189, "grad_norm": 12.0546875, "learning_rate": 9.353077032404776e-06, "loss": 1.0363, "step": 18707 }, { "epoch": 1.9906363055969356, "grad_norm": 12.6796875, "learning_rate": 9.353041500852757e-06, "loss": 0.8946, "step": 18708 }, { "epoch": 1.9907427112151521, "grad_norm": 18.796875, "learning_rate": 9.35300596930074e-06, "loss": 1.2082, "step": 18709 }, { "epoch": 1.9908491168333688, "grad_norm": 10.375, "learning_rate": 9.352970437748722e-06, "loss": 0.8572, "step": 18710 }, { "epoch": 1.9909555224515856, "grad_norm": 20.28125, "learning_rate": 9.352934906196704e-06, "loss": 0.8946, "step": 18711 }, { "epoch": 1.991061928069802, "grad_norm": 11.1796875, "learning_rate": 9.352899374644686e-06, "loss": 1.3467, "step": 18712 }, { "epoch": 1.9911683336880186, "grad_norm": 14.5, "learning_rate": 9.352863843092667e-06, "loss": 0.9394, "step": 18713 }, { "epoch": 1.9912747393062353, "grad_norm": 13.328125, "learning_rate": 9.352828311540649e-06, "loss": 0.843, "step": 18714 }, { "epoch": 1.991381144924452, "grad_norm": 10.8203125, "learning_rate": 9.35279277998863e-06, "loss": 0.9592, "step": 18715 }, { "epoch": 1.9914875505426686, "grad_norm": 9.3671875, "learning_rate": 9.352757248436612e-06, "loss": 1.0904, "step": 18716 }, { "epoch": 1.9915939561608853, "grad_norm": 18.8125, "learning_rate": 9.352721716884594e-06, "loss": 1.4103, "step": 18717 }, { "epoch": 1.991700361779102, "grad_norm": 12.953125, "learning_rate": 9.352686185332575e-06, "loss": 1.0506, "step": 18718 }, { "epoch": 1.9918067673973185, "grad_norm": 11.734375, "learning_rate": 9.352650653780559e-06, "loss": 0.8119, "step": 18719 }, { "epoch": 1.991913173015535, "grad_norm": 24.21875, "learning_rate": 9.35261512222854e-06, "loss": 0.9228, "step": 18720 }, { "epoch": 1.992019578633752, "grad_norm": 13.40625, "learning_rate": 9.352579590676522e-06, "loss": 0.9842, "step": 18721 }, { "epoch": 1.9921259842519685, "grad_norm": 10.5546875, "learning_rate": 9.352544059124503e-06, "loss": 1.0938, "step": 18722 }, { "epoch": 1.992232389870185, "grad_norm": 12.2734375, "learning_rate": 9.352508527572485e-06, "loss": 1.1553, "step": 18723 }, { "epoch": 1.9923387954884018, "grad_norm": 11.9375, "learning_rate": 9.352472996020468e-06, "loss": 1.0311, "step": 18724 }, { "epoch": 1.9924452011066185, "grad_norm": 14.9453125, "learning_rate": 9.35243746446845e-06, "loss": 0.8103, "step": 18725 }, { "epoch": 1.992551606724835, "grad_norm": 15.953125, "learning_rate": 9.352401932916431e-06, "loss": 1.2171, "step": 18726 }, { "epoch": 1.9926580123430517, "grad_norm": 13.234375, "learning_rate": 9.352366401364411e-06, "loss": 0.9718, "step": 18727 }, { "epoch": 1.9927644179612685, "grad_norm": 12.421875, "learning_rate": 9.352330869812393e-06, "loss": 1.0248, "step": 18728 }, { "epoch": 1.992870823579485, "grad_norm": 15.5, "learning_rate": 9.352295338260376e-06, "loss": 1.1042, "step": 18729 }, { "epoch": 1.9929772291977015, "grad_norm": 12.9140625, "learning_rate": 9.352259806708358e-06, "loss": 0.8197, "step": 18730 }, { "epoch": 1.9930836348159184, "grad_norm": 12.1171875, "learning_rate": 9.35222427515634e-06, "loss": 1.1323, "step": 18731 }, { "epoch": 1.993190040434135, "grad_norm": 16.390625, "learning_rate": 9.352188743604321e-06, "loss": 1.2433, "step": 18732 }, { "epoch": 1.9932964460523515, "grad_norm": 13.5703125, "learning_rate": 9.352153212052303e-06, "loss": 0.7722, "step": 18733 }, { "epoch": 1.9934028516705682, "grad_norm": 13.0, "learning_rate": 9.352117680500286e-06, "loss": 0.9994, "step": 18734 }, { "epoch": 1.993509257288785, "grad_norm": 17.625, "learning_rate": 9.352082148948268e-06, "loss": 0.8457, "step": 18735 }, { "epoch": 1.9936156629070014, "grad_norm": 13.140625, "learning_rate": 9.35204661739625e-06, "loss": 1.2648, "step": 18736 }, { "epoch": 1.9937220685252182, "grad_norm": 11.203125, "learning_rate": 9.35201108584423e-06, "loss": 0.867, "step": 18737 }, { "epoch": 1.993828474143435, "grad_norm": 14.921875, "learning_rate": 9.351975554292212e-06, "loss": 0.9331, "step": 18738 }, { "epoch": 1.9939348797616514, "grad_norm": 13.2421875, "learning_rate": 9.351940022740194e-06, "loss": 1.0383, "step": 18739 }, { "epoch": 1.994041285379868, "grad_norm": 19.25, "learning_rate": 9.351904491188176e-06, "loss": 0.9952, "step": 18740 }, { "epoch": 1.9941476909980849, "grad_norm": 11.890625, "learning_rate": 9.351868959636157e-06, "loss": 0.8762, "step": 18741 }, { "epoch": 1.9942540966163014, "grad_norm": 11.328125, "learning_rate": 9.351833428084139e-06, "loss": 0.9737, "step": 18742 }, { "epoch": 1.9943605022345179, "grad_norm": 14.1171875, "learning_rate": 9.35179789653212e-06, "loss": 0.9789, "step": 18743 }, { "epoch": 1.9944669078527346, "grad_norm": 14.09375, "learning_rate": 9.351762364980102e-06, "loss": 0.987, "step": 18744 }, { "epoch": 1.9945733134709513, "grad_norm": 10.765625, "learning_rate": 9.351726833428085e-06, "loss": 0.9849, "step": 18745 }, { "epoch": 1.9946797190891679, "grad_norm": 13.5, "learning_rate": 9.351691301876067e-06, "loss": 0.8905, "step": 18746 }, { "epoch": 1.9947861247073846, "grad_norm": 15.4140625, "learning_rate": 9.351655770324049e-06, "loss": 1.2345, "step": 18747 }, { "epoch": 1.9948925303256013, "grad_norm": 11.0625, "learning_rate": 9.35162023877203e-06, "loss": 1.105, "step": 18748 }, { "epoch": 1.9949989359438178, "grad_norm": 13.4453125, "learning_rate": 9.351584707220012e-06, "loss": 1.1866, "step": 18749 }, { "epoch": 1.9951053415620343, "grad_norm": 11.5546875, "learning_rate": 9.351549175667995e-06, "loss": 0.8938, "step": 18750 }, { "epoch": 1.995211747180251, "grad_norm": 12.7109375, "learning_rate": 9.351513644115977e-06, "loss": 1.4006, "step": 18751 }, { "epoch": 1.9953181527984678, "grad_norm": 17.34375, "learning_rate": 9.351478112563958e-06, "loss": 1.3758, "step": 18752 }, { "epoch": 1.9954245584166843, "grad_norm": 12.78125, "learning_rate": 9.35144258101194e-06, "loss": 1.1631, "step": 18753 }, { "epoch": 1.995530964034901, "grad_norm": 15.640625, "learning_rate": 9.351407049459922e-06, "loss": 1.337, "step": 18754 }, { "epoch": 1.9956373696531178, "grad_norm": 14.484375, "learning_rate": 9.351371517907903e-06, "loss": 1.2589, "step": 18755 }, { "epoch": 1.9957437752713343, "grad_norm": 10.578125, "learning_rate": 9.351335986355885e-06, "loss": 1.0197, "step": 18756 }, { "epoch": 1.995850180889551, "grad_norm": 10.625, "learning_rate": 9.351300454803866e-06, "loss": 0.9463, "step": 18757 }, { "epoch": 1.9959565865077677, "grad_norm": 11.453125, "learning_rate": 9.351264923251848e-06, "loss": 1.1103, "step": 18758 }, { "epoch": 1.9960629921259843, "grad_norm": 9.875, "learning_rate": 9.35122939169983e-06, "loss": 1.1849, "step": 18759 }, { "epoch": 1.9961693977442008, "grad_norm": 9.9140625, "learning_rate": 9.351193860147813e-06, "loss": 1.2244, "step": 18760 }, { "epoch": 1.9962758033624175, "grad_norm": 12.484375, "learning_rate": 9.351158328595794e-06, "loss": 1.1301, "step": 18761 }, { "epoch": 1.9963822089806342, "grad_norm": 8.7109375, "learning_rate": 9.351122797043776e-06, "loss": 1.0385, "step": 18762 }, { "epoch": 1.9964886145988507, "grad_norm": 11.5625, "learning_rate": 9.351087265491758e-06, "loss": 0.9655, "step": 18763 }, { "epoch": 1.9965950202170675, "grad_norm": 9.4609375, "learning_rate": 9.35105173393974e-06, "loss": 1.0395, "step": 18764 }, { "epoch": 1.9967014258352842, "grad_norm": 11.8203125, "learning_rate": 9.351016202387721e-06, "loss": 1.1647, "step": 18765 }, { "epoch": 1.9968078314535007, "grad_norm": 12.9296875, "learning_rate": 9.350980670835703e-06, "loss": 0.8721, "step": 18766 }, { "epoch": 1.9969142370717172, "grad_norm": 12.296875, "learning_rate": 9.350945139283684e-06, "loss": 0.9718, "step": 18767 }, { "epoch": 1.9970206426899342, "grad_norm": 12.21875, "learning_rate": 9.350909607731666e-06, "loss": 1.5154, "step": 18768 }, { "epoch": 1.9971270483081507, "grad_norm": 12.421875, "learning_rate": 9.350874076179647e-06, "loss": 1.2783, "step": 18769 }, { "epoch": 1.9972334539263672, "grad_norm": 12.9296875, "learning_rate": 9.35083854462763e-06, "loss": 0.985, "step": 18770 }, { "epoch": 1.997339859544584, "grad_norm": 10.90625, "learning_rate": 9.350803013075612e-06, "loss": 1.0913, "step": 18771 }, { "epoch": 1.9974462651628007, "grad_norm": 17.9375, "learning_rate": 9.350767481523594e-06, "loss": 1.1154, "step": 18772 }, { "epoch": 1.9975526707810172, "grad_norm": 13.84375, "learning_rate": 9.350731949971575e-06, "loss": 1.1978, "step": 18773 }, { "epoch": 1.997659076399234, "grad_norm": 15.390625, "learning_rate": 9.350696418419557e-06, "loss": 1.1358, "step": 18774 }, { "epoch": 1.9977654820174506, "grad_norm": 11.109375, "learning_rate": 9.350660886867539e-06, "loss": 1.0266, "step": 18775 }, { "epoch": 1.9978718876356671, "grad_norm": 13.765625, "learning_rate": 9.350625355315522e-06, "loss": 1.1444, "step": 18776 }, { "epoch": 1.9979782932538837, "grad_norm": 10.046875, "learning_rate": 9.350589823763504e-06, "loss": 1.1007, "step": 18777 }, { "epoch": 1.9980846988721006, "grad_norm": 10.984375, "learning_rate": 9.350554292211485e-06, "loss": 1.308, "step": 18778 }, { "epoch": 1.9981911044903171, "grad_norm": 11.4921875, "learning_rate": 9.350518760659467e-06, "loss": 1.0337, "step": 18779 }, { "epoch": 1.9982975101085336, "grad_norm": 19.84375, "learning_rate": 9.350483229107448e-06, "loss": 1.3624, "step": 18780 }, { "epoch": 1.9984039157267504, "grad_norm": 11.640625, "learning_rate": 9.35044769755543e-06, "loss": 1.1517, "step": 18781 }, { "epoch": 1.998510321344967, "grad_norm": 12.7890625, "learning_rate": 9.350412166003412e-06, "loss": 1.3389, "step": 18782 }, { "epoch": 1.9986167269631836, "grad_norm": 8.3125, "learning_rate": 9.350376634451393e-06, "loss": 0.777, "step": 18783 }, { "epoch": 1.9987231325814003, "grad_norm": 22.09375, "learning_rate": 9.350341102899375e-06, "loss": 0.9243, "step": 18784 }, { "epoch": 1.998829538199617, "grad_norm": 12.8203125, "learning_rate": 9.350305571347356e-06, "loss": 1.0781, "step": 18785 }, { "epoch": 1.9989359438178336, "grad_norm": 17.671875, "learning_rate": 9.35027003979534e-06, "loss": 1.3373, "step": 18786 }, { "epoch": 1.99904234943605, "grad_norm": 12.21875, "learning_rate": 9.350234508243321e-06, "loss": 1.0525, "step": 18787 }, { "epoch": 1.9991487550542668, "grad_norm": 21.125, "learning_rate": 9.350198976691303e-06, "loss": 1.2073, "step": 18788 }, { "epoch": 1.9992551606724835, "grad_norm": 14.6640625, "learning_rate": 9.350163445139285e-06, "loss": 0.9534, "step": 18789 }, { "epoch": 1.9993615662907, "grad_norm": 8.578125, "learning_rate": 9.350127913587266e-06, "loss": 1.0423, "step": 18790 }, { "epoch": 1.9994679719089168, "grad_norm": 9.375, "learning_rate": 9.350092382035248e-06, "loss": 0.8568, "step": 18791 }, { "epoch": 1.9995743775271335, "grad_norm": 37.46875, "learning_rate": 9.35005685048323e-06, "loss": 1.1151, "step": 18792 }, { "epoch": 1.99968078314535, "grad_norm": 11.796875, "learning_rate": 9.350021318931211e-06, "loss": 1.335, "step": 18793 }, { "epoch": 1.9997871887635668, "grad_norm": 10.8203125, "learning_rate": 9.349985787379193e-06, "loss": 0.9553, "step": 18794 }, { "epoch": 1.9998935943817835, "grad_norm": 9.9140625, "learning_rate": 9.349950255827174e-06, "loss": 0.8979, "step": 18795 }, { "epoch": 2.0, "grad_norm": 29.53125, "learning_rate": 9.349914724275157e-06, "loss": 0.9216, "step": 18796 }, { "epoch": 2.0001064056182165, "grad_norm": 11.9296875, "learning_rate": 9.349879192723139e-06, "loss": 0.8389, "step": 18797 }, { "epoch": 2.0002128112364335, "grad_norm": 9.953125, "learning_rate": 9.34984366117112e-06, "loss": 0.9845, "step": 18798 }, { "epoch": 2.00031921685465, "grad_norm": 15.09375, "learning_rate": 9.349808129619102e-06, "loss": 0.9588, "step": 18799 }, { "epoch": 2.0004256224728665, "grad_norm": 13.09375, "learning_rate": 9.349772598067084e-06, "loss": 0.9041, "step": 18800 }, { "epoch": 2.000532028091083, "grad_norm": 11.078125, "learning_rate": 9.349737066515067e-06, "loss": 0.7826, "step": 18801 }, { "epoch": 2.0006384337093, "grad_norm": 9.375, "learning_rate": 9.349701534963049e-06, "loss": 0.7367, "step": 18802 }, { "epoch": 2.0007448393275165, "grad_norm": 12.4609375, "learning_rate": 9.34966600341103e-06, "loss": 0.8888, "step": 18803 }, { "epoch": 2.000851244945733, "grad_norm": 14.3984375, "learning_rate": 9.349630471859012e-06, "loss": 1.0579, "step": 18804 }, { "epoch": 2.00095765056395, "grad_norm": 15.125, "learning_rate": 9.349594940306994e-06, "loss": 0.9548, "step": 18805 }, { "epoch": 2.0010640561821664, "grad_norm": 13.53125, "learning_rate": 9.349559408754975e-06, "loss": 0.8422, "step": 18806 }, { "epoch": 2.001170461800383, "grad_norm": 10.9296875, "learning_rate": 9.349523877202957e-06, "loss": 1.0414, "step": 18807 }, { "epoch": 2.0012768674186, "grad_norm": 13.046875, "learning_rate": 9.349488345650938e-06, "loss": 0.712, "step": 18808 }, { "epoch": 2.0013832730368164, "grad_norm": 13.78125, "learning_rate": 9.34945281409892e-06, "loss": 0.8588, "step": 18809 }, { "epoch": 2.001489678655033, "grad_norm": 12.2421875, "learning_rate": 9.349417282546902e-06, "loss": 1.1432, "step": 18810 }, { "epoch": 2.0015960842732494, "grad_norm": 11.4453125, "learning_rate": 9.349381750994885e-06, "loss": 0.8394, "step": 18811 }, { "epoch": 2.0017024898914664, "grad_norm": 14.25, "learning_rate": 9.349346219442867e-06, "loss": 0.7374, "step": 18812 }, { "epoch": 2.001808895509683, "grad_norm": 8.90625, "learning_rate": 9.349310687890848e-06, "loss": 1.0121, "step": 18813 }, { "epoch": 2.0019153011278994, "grad_norm": 10.1953125, "learning_rate": 9.34927515633883e-06, "loss": 0.7014, "step": 18814 }, { "epoch": 2.0020217067461163, "grad_norm": 11.2578125, "learning_rate": 9.349239624786811e-06, "loss": 0.7837, "step": 18815 }, { "epoch": 2.002128112364333, "grad_norm": 11.0234375, "learning_rate": 9.349204093234793e-06, "loss": 0.93, "step": 18816 }, { "epoch": 2.0022345179825494, "grad_norm": 11.3671875, "learning_rate": 9.349168561682775e-06, "loss": 0.8884, "step": 18817 }, { "epoch": 2.0023409236007663, "grad_norm": 10.046875, "learning_rate": 9.349133030130756e-06, "loss": 0.8939, "step": 18818 }, { "epoch": 2.002447329218983, "grad_norm": 12.2109375, "learning_rate": 9.349097498578738e-06, "loss": 1.0855, "step": 18819 }, { "epoch": 2.0025537348371993, "grad_norm": 9.9921875, "learning_rate": 9.34906196702672e-06, "loss": 0.8379, "step": 18820 }, { "epoch": 2.002660140455416, "grad_norm": 11.0625, "learning_rate": 9.349026435474701e-06, "loss": 0.8347, "step": 18821 }, { "epoch": 2.002766546073633, "grad_norm": 10.4453125, "learning_rate": 9.348990903922684e-06, "loss": 0.864, "step": 18822 }, { "epoch": 2.0028729516918493, "grad_norm": 10.8515625, "learning_rate": 9.348955372370666e-06, "loss": 0.9961, "step": 18823 }, { "epoch": 2.002979357310066, "grad_norm": 9.65625, "learning_rate": 9.348919840818648e-06, "loss": 0.7833, "step": 18824 }, { "epoch": 2.0030857629282828, "grad_norm": 12.0859375, "learning_rate": 9.348884309266629e-06, "loss": 0.7263, "step": 18825 }, { "epoch": 2.0031921685464993, "grad_norm": 11.8359375, "learning_rate": 9.34884877771461e-06, "loss": 0.9772, "step": 18826 }, { "epoch": 2.003298574164716, "grad_norm": 13.203125, "learning_rate": 9.348813246162594e-06, "loss": 0.7604, "step": 18827 }, { "epoch": 2.0034049797829327, "grad_norm": 13.96875, "learning_rate": 9.348777714610576e-06, "loss": 0.9133, "step": 18828 }, { "epoch": 2.0035113854011493, "grad_norm": 23.28125, "learning_rate": 9.348742183058557e-06, "loss": 0.991, "step": 18829 }, { "epoch": 2.0036177910193658, "grad_norm": 12.3515625, "learning_rate": 9.348706651506539e-06, "loss": 0.8194, "step": 18830 }, { "epoch": 2.0037241966375823, "grad_norm": 13.0546875, "learning_rate": 9.34867111995452e-06, "loss": 0.8706, "step": 18831 }, { "epoch": 2.0038306022557992, "grad_norm": 13.765625, "learning_rate": 9.348635588402502e-06, "loss": 0.9294, "step": 18832 }, { "epoch": 2.0039370078740157, "grad_norm": 11.5703125, "learning_rate": 9.348600056850484e-06, "loss": 0.8281, "step": 18833 }, { "epoch": 2.0040434134922323, "grad_norm": 12.359375, "learning_rate": 9.348564525298465e-06, "loss": 0.7999, "step": 18834 }, { "epoch": 2.004149819110449, "grad_norm": 13.171875, "learning_rate": 9.348528993746447e-06, "loss": 1.0669, "step": 18835 }, { "epoch": 2.0042562247286657, "grad_norm": 12.1640625, "learning_rate": 9.348493462194429e-06, "loss": 0.8087, "step": 18836 }, { "epoch": 2.0043626303468822, "grad_norm": 10.890625, "learning_rate": 9.348457930642412e-06, "loss": 0.8685, "step": 18837 }, { "epoch": 2.004469035965099, "grad_norm": 18.21875, "learning_rate": 9.348422399090393e-06, "loss": 0.9605, "step": 18838 }, { "epoch": 2.0045754415833157, "grad_norm": 9.1796875, "learning_rate": 9.348386867538375e-06, "loss": 0.8661, "step": 18839 }, { "epoch": 2.004681847201532, "grad_norm": 13.7890625, "learning_rate": 9.348351335986357e-06, "loss": 0.9349, "step": 18840 }, { "epoch": 2.0047882528197487, "grad_norm": 13.3515625, "learning_rate": 9.348315804434338e-06, "loss": 0.9915, "step": 18841 }, { "epoch": 2.0048946584379657, "grad_norm": 17.625, "learning_rate": 9.348280272882322e-06, "loss": 0.8557, "step": 18842 }, { "epoch": 2.005001064056182, "grad_norm": 13.859375, "learning_rate": 9.348244741330303e-06, "loss": 0.8764, "step": 18843 }, { "epoch": 2.0051074696743987, "grad_norm": 9.984375, "learning_rate": 9.348209209778285e-06, "loss": 0.6973, "step": 18844 }, { "epoch": 2.0052138752926156, "grad_norm": 12.28125, "learning_rate": 9.348173678226266e-06, "loss": 0.7892, "step": 18845 }, { "epoch": 2.005320280910832, "grad_norm": 12.890625, "learning_rate": 9.348138146674248e-06, "loss": 0.8822, "step": 18846 }, { "epoch": 2.0054266865290487, "grad_norm": 16.625, "learning_rate": 9.34810261512223e-06, "loss": 1.0812, "step": 18847 }, { "epoch": 2.005533092147265, "grad_norm": 13.1640625, "learning_rate": 9.348067083570211e-06, "loss": 0.787, "step": 18848 }, { "epoch": 2.005639497765482, "grad_norm": 12.0859375, "learning_rate": 9.348031552018193e-06, "loss": 0.8776, "step": 18849 }, { "epoch": 2.0057459033836986, "grad_norm": 13.4296875, "learning_rate": 9.347996020466174e-06, "loss": 1.14, "step": 18850 }, { "epoch": 2.005852309001915, "grad_norm": 16.453125, "learning_rate": 9.347960488914156e-06, "loss": 0.9832, "step": 18851 }, { "epoch": 2.005958714620132, "grad_norm": 14.96875, "learning_rate": 9.34792495736214e-06, "loss": 1.1246, "step": 18852 }, { "epoch": 2.0060651202383486, "grad_norm": 12.453125, "learning_rate": 9.347889425810121e-06, "loss": 1.0559, "step": 18853 }, { "epoch": 2.006171525856565, "grad_norm": 13.1328125, "learning_rate": 9.347853894258103e-06, "loss": 0.8302, "step": 18854 }, { "epoch": 2.006277931474782, "grad_norm": 15.21875, "learning_rate": 9.347818362706084e-06, "loss": 1.0658, "step": 18855 }, { "epoch": 2.0063843370929986, "grad_norm": 14.984375, "learning_rate": 9.347782831154066e-06, "loss": 0.8337, "step": 18856 }, { "epoch": 2.006490742711215, "grad_norm": 10.9609375, "learning_rate": 9.347747299602047e-06, "loss": 0.8507, "step": 18857 }, { "epoch": 2.0065971483294316, "grad_norm": 14.171875, "learning_rate": 9.347711768050029e-06, "loss": 0.8736, "step": 18858 }, { "epoch": 2.0067035539476485, "grad_norm": 15.2421875, "learning_rate": 9.34767623649801e-06, "loss": 0.7629, "step": 18859 }, { "epoch": 2.006809959565865, "grad_norm": 10.9453125, "learning_rate": 9.347640704945992e-06, "loss": 0.84, "step": 18860 }, { "epoch": 2.0069163651840816, "grad_norm": 12.5859375, "learning_rate": 9.347605173393974e-06, "loss": 0.9825, "step": 18861 }, { "epoch": 2.0070227708022985, "grad_norm": 10.6640625, "learning_rate": 9.347569641841955e-06, "loss": 0.7935, "step": 18862 }, { "epoch": 2.007129176420515, "grad_norm": 10.671875, "learning_rate": 9.347534110289939e-06, "loss": 0.8723, "step": 18863 }, { "epoch": 2.0072355820387315, "grad_norm": 12.96875, "learning_rate": 9.34749857873792e-06, "loss": 1.0357, "step": 18864 }, { "epoch": 2.0073419876569485, "grad_norm": 11.5859375, "learning_rate": 9.347463047185902e-06, "loss": 0.9049, "step": 18865 }, { "epoch": 2.007448393275165, "grad_norm": 12.015625, "learning_rate": 9.347427515633883e-06, "loss": 1.0416, "step": 18866 }, { "epoch": 2.0075547988933815, "grad_norm": 10.1953125, "learning_rate": 9.347391984081865e-06, "loss": 0.8142, "step": 18867 }, { "epoch": 2.007661204511598, "grad_norm": 16.421875, "learning_rate": 9.347356452529848e-06, "loss": 0.7896, "step": 18868 }, { "epoch": 2.007767610129815, "grad_norm": 9.71875, "learning_rate": 9.34732092097783e-06, "loss": 0.7624, "step": 18869 }, { "epoch": 2.0078740157480315, "grad_norm": 14.3984375, "learning_rate": 9.347285389425812e-06, "loss": 0.8741, "step": 18870 }, { "epoch": 2.007980421366248, "grad_norm": 15.21875, "learning_rate": 9.347249857873793e-06, "loss": 0.8281, "step": 18871 }, { "epoch": 2.008086826984465, "grad_norm": 14.5859375, "learning_rate": 9.347214326321775e-06, "loss": 0.9403, "step": 18872 }, { "epoch": 2.0081932326026815, "grad_norm": 11.671875, "learning_rate": 9.347178794769756e-06, "loss": 1.0, "step": 18873 }, { "epoch": 2.008299638220898, "grad_norm": 17.453125, "learning_rate": 9.347143263217738e-06, "loss": 1.1721, "step": 18874 }, { "epoch": 2.008406043839115, "grad_norm": 15.609375, "learning_rate": 9.34710773166572e-06, "loss": 0.8643, "step": 18875 }, { "epoch": 2.0085124494573314, "grad_norm": 11.484375, "learning_rate": 9.347072200113701e-06, "loss": 0.836, "step": 18876 }, { "epoch": 2.008618855075548, "grad_norm": 13.8203125, "learning_rate": 9.347036668561683e-06, "loss": 0.794, "step": 18877 }, { "epoch": 2.0087252606937644, "grad_norm": 10.7578125, "learning_rate": 9.347001137009666e-06, "loss": 0.7799, "step": 18878 }, { "epoch": 2.0088316663119814, "grad_norm": 10.3828125, "learning_rate": 9.346965605457648e-06, "loss": 0.8482, "step": 18879 }, { "epoch": 2.008938071930198, "grad_norm": 10.5546875, "learning_rate": 9.34693007390563e-06, "loss": 0.7832, "step": 18880 }, { "epoch": 2.0090444775484144, "grad_norm": 10.0234375, "learning_rate": 9.346894542353611e-06, "loss": 0.8698, "step": 18881 }, { "epoch": 2.0091508831666314, "grad_norm": 12.2734375, "learning_rate": 9.346859010801593e-06, "loss": 0.8596, "step": 18882 }, { "epoch": 2.009257288784848, "grad_norm": 11.4453125, "learning_rate": 9.346823479249574e-06, "loss": 0.8379, "step": 18883 }, { "epoch": 2.0093636944030644, "grad_norm": 16.125, "learning_rate": 9.346787947697556e-06, "loss": 0.9348, "step": 18884 }, { "epoch": 2.009470100021281, "grad_norm": 12.7734375, "learning_rate": 9.346752416145537e-06, "loss": 0.7809, "step": 18885 }, { "epoch": 2.009576505639498, "grad_norm": 10.265625, "learning_rate": 9.346716884593519e-06, "loss": 0.7857, "step": 18886 }, { "epoch": 2.0096829112577144, "grad_norm": 12.0859375, "learning_rate": 9.3466813530415e-06, "loss": 0.9355, "step": 18887 }, { "epoch": 2.009789316875931, "grad_norm": 8.734375, "learning_rate": 9.346645821489484e-06, "loss": 0.7143, "step": 18888 }, { "epoch": 2.009895722494148, "grad_norm": 13.28125, "learning_rate": 9.346610289937466e-06, "loss": 0.9787, "step": 18889 }, { "epoch": 2.0100021281123643, "grad_norm": 9.984375, "learning_rate": 9.346574758385447e-06, "loss": 0.8091, "step": 18890 }, { "epoch": 2.010108533730581, "grad_norm": 12.7421875, "learning_rate": 9.346539226833429e-06, "loss": 0.8098, "step": 18891 }, { "epoch": 2.010214939348798, "grad_norm": 11.3125, "learning_rate": 9.34650369528141e-06, "loss": 0.917, "step": 18892 }, { "epoch": 2.0103213449670143, "grad_norm": 10.3828125, "learning_rate": 9.346468163729392e-06, "loss": 0.8905, "step": 18893 }, { "epoch": 2.010427750585231, "grad_norm": 10.9453125, "learning_rate": 9.346432632177375e-06, "loss": 0.9401, "step": 18894 }, { "epoch": 2.0105341562034473, "grad_norm": 11.8046875, "learning_rate": 9.346397100625357e-06, "loss": 0.9035, "step": 18895 }, { "epoch": 2.0106405618216643, "grad_norm": 14.015625, "learning_rate": 9.346361569073338e-06, "loss": 1.0665, "step": 18896 }, { "epoch": 2.010746967439881, "grad_norm": 11.8984375, "learning_rate": 9.34632603752132e-06, "loss": 0.8464, "step": 18897 }, { "epoch": 2.0108533730580973, "grad_norm": 14.0859375, "learning_rate": 9.346290505969302e-06, "loss": 0.9391, "step": 18898 }, { "epoch": 2.0109597786763143, "grad_norm": 9.9921875, "learning_rate": 9.346254974417283e-06, "loss": 0.9763, "step": 18899 }, { "epoch": 2.0110661842945308, "grad_norm": 13.7421875, "learning_rate": 9.346219442865265e-06, "loss": 0.9988, "step": 18900 }, { "epoch": 2.0111725899127473, "grad_norm": 13.6640625, "learning_rate": 9.346183911313247e-06, "loss": 0.9614, "step": 18901 }, { "epoch": 2.0112789955309642, "grad_norm": 16.21875, "learning_rate": 9.346148379761228e-06, "loss": 0.8353, "step": 18902 }, { "epoch": 2.0113854011491807, "grad_norm": 15.703125, "learning_rate": 9.34611284820921e-06, "loss": 1.0124, "step": 18903 }, { "epoch": 2.0114918067673973, "grad_norm": 14.1015625, "learning_rate": 9.346077316657193e-06, "loss": 0.7164, "step": 18904 }, { "epoch": 2.0115982123856138, "grad_norm": 10.1328125, "learning_rate": 9.346041785105175e-06, "loss": 0.8226, "step": 18905 }, { "epoch": 2.0117046180038307, "grad_norm": 14.5546875, "learning_rate": 9.346006253553156e-06, "loss": 0.9872, "step": 18906 }, { "epoch": 2.0118110236220472, "grad_norm": 11.5, "learning_rate": 9.345970722001138e-06, "loss": 0.9701, "step": 18907 }, { "epoch": 2.0119174292402637, "grad_norm": 10.34375, "learning_rate": 9.34593519044912e-06, "loss": 0.7398, "step": 18908 }, { "epoch": 2.0120238348584807, "grad_norm": 11.8359375, "learning_rate": 9.345899658897101e-06, "loss": 0.8921, "step": 18909 }, { "epoch": 2.012130240476697, "grad_norm": 13.109375, "learning_rate": 9.345864127345083e-06, "loss": 1.0796, "step": 18910 }, { "epoch": 2.0122366460949137, "grad_norm": 14.4140625, "learning_rate": 9.345828595793064e-06, "loss": 0.949, "step": 18911 }, { "epoch": 2.0123430517131307, "grad_norm": 10.7109375, "learning_rate": 9.345793064241046e-06, "loss": 0.8891, "step": 18912 }, { "epoch": 2.012449457331347, "grad_norm": 15.1953125, "learning_rate": 9.345757532689027e-06, "loss": 0.8091, "step": 18913 }, { "epoch": 2.0125558629495637, "grad_norm": 10.296875, "learning_rate": 9.34572200113701e-06, "loss": 0.881, "step": 18914 }, { "epoch": 2.01266226856778, "grad_norm": 12.6171875, "learning_rate": 9.345686469584992e-06, "loss": 0.9569, "step": 18915 }, { "epoch": 2.012768674185997, "grad_norm": 11.953125, "learning_rate": 9.345650938032974e-06, "loss": 0.8178, "step": 18916 }, { "epoch": 2.0128750798042137, "grad_norm": 12.2109375, "learning_rate": 9.345615406480956e-06, "loss": 0.7763, "step": 18917 }, { "epoch": 2.01298148542243, "grad_norm": 13.9609375, "learning_rate": 9.345579874928937e-06, "loss": 1.172, "step": 18918 }, { "epoch": 2.013087891040647, "grad_norm": 10.6640625, "learning_rate": 9.34554434337692e-06, "loss": 0.8902, "step": 18919 }, { "epoch": 2.0131942966588636, "grad_norm": 12.265625, "learning_rate": 9.345508811824902e-06, "loss": 0.7162, "step": 18920 }, { "epoch": 2.01330070227708, "grad_norm": 12.2890625, "learning_rate": 9.345473280272884e-06, "loss": 0.7334, "step": 18921 }, { "epoch": 2.013407107895297, "grad_norm": 10.4140625, "learning_rate": 9.345437748720865e-06, "loss": 0.6533, "step": 18922 }, { "epoch": 2.0135135135135136, "grad_norm": 10.859375, "learning_rate": 9.345402217168847e-06, "loss": 0.7556, "step": 18923 }, { "epoch": 2.01361991913173, "grad_norm": 14.75, "learning_rate": 9.345366685616829e-06, "loss": 0.8338, "step": 18924 }, { "epoch": 2.0137263247499466, "grad_norm": 11.84375, "learning_rate": 9.34533115406481e-06, "loss": 0.9094, "step": 18925 }, { "epoch": 2.0138327303681636, "grad_norm": 10.484375, "learning_rate": 9.345295622512792e-06, "loss": 0.8791, "step": 18926 }, { "epoch": 2.01393913598638, "grad_norm": 15.5703125, "learning_rate": 9.345260090960773e-06, "loss": 0.8801, "step": 18927 }, { "epoch": 2.0140455416045966, "grad_norm": 13.515625, "learning_rate": 9.345224559408755e-06, "loss": 1.1908, "step": 18928 }, { "epoch": 2.0141519472228135, "grad_norm": 12.6953125, "learning_rate": 9.345189027856738e-06, "loss": 0.7497, "step": 18929 }, { "epoch": 2.01425835284103, "grad_norm": 10.03125, "learning_rate": 9.34515349630472e-06, "loss": 0.919, "step": 18930 }, { "epoch": 2.0143647584592466, "grad_norm": 16.09375, "learning_rate": 9.345117964752701e-06, "loss": 0.8352, "step": 18931 }, { "epoch": 2.014471164077463, "grad_norm": 15.4609375, "learning_rate": 9.345082433200683e-06, "loss": 1.0394, "step": 18932 }, { "epoch": 2.01457756969568, "grad_norm": 16.1875, "learning_rate": 9.345046901648665e-06, "loss": 1.0389, "step": 18933 }, { "epoch": 2.0146839753138965, "grad_norm": 12.8203125, "learning_rate": 9.345011370096646e-06, "loss": 1.0755, "step": 18934 }, { "epoch": 2.014790380932113, "grad_norm": 12.796875, "learning_rate": 9.344975838544628e-06, "loss": 0.8567, "step": 18935 }, { "epoch": 2.01489678655033, "grad_norm": 15.1015625, "learning_rate": 9.34494030699261e-06, "loss": 1.0034, "step": 18936 }, { "epoch": 2.0150031921685465, "grad_norm": 9.96875, "learning_rate": 9.344904775440591e-06, "loss": 0.8741, "step": 18937 }, { "epoch": 2.015109597786763, "grad_norm": 10.2734375, "learning_rate": 9.344869243888573e-06, "loss": 0.8889, "step": 18938 }, { "epoch": 2.01521600340498, "grad_norm": 16.1875, "learning_rate": 9.344833712336554e-06, "loss": 0.807, "step": 18939 }, { "epoch": 2.0153224090231965, "grad_norm": 11.2578125, "learning_rate": 9.344798180784538e-06, "loss": 0.9249, "step": 18940 }, { "epoch": 2.015428814641413, "grad_norm": 13.640625, "learning_rate": 9.34476264923252e-06, "loss": 0.9157, "step": 18941 }, { "epoch": 2.0155352202596295, "grad_norm": 11.4375, "learning_rate": 9.344727117680501e-06, "loss": 0.8555, "step": 18942 }, { "epoch": 2.0156416258778465, "grad_norm": 11.640625, "learning_rate": 9.344691586128482e-06, "loss": 0.9679, "step": 18943 }, { "epoch": 2.015748031496063, "grad_norm": 13.234375, "learning_rate": 9.344656054576464e-06, "loss": 0.8223, "step": 18944 }, { "epoch": 2.0158544371142795, "grad_norm": 12.9375, "learning_rate": 9.344620523024447e-06, "loss": 0.8962, "step": 18945 }, { "epoch": 2.0159608427324964, "grad_norm": 13.6015625, "learning_rate": 9.344584991472429e-06, "loss": 0.9929, "step": 18946 }, { "epoch": 2.016067248350713, "grad_norm": 15.5625, "learning_rate": 9.34454945992041e-06, "loss": 0.8965, "step": 18947 }, { "epoch": 2.0161736539689294, "grad_norm": 12.1015625, "learning_rate": 9.344513928368392e-06, "loss": 0.8147, "step": 18948 }, { "epoch": 2.0162800595871464, "grad_norm": 12.09375, "learning_rate": 9.344478396816374e-06, "loss": 0.8989, "step": 18949 }, { "epoch": 2.016386465205363, "grad_norm": 10.296875, "learning_rate": 9.344442865264355e-06, "loss": 0.7538, "step": 18950 }, { "epoch": 2.0164928708235794, "grad_norm": 14.8671875, "learning_rate": 9.344407333712337e-06, "loss": 0.9124, "step": 18951 }, { "epoch": 2.016599276441796, "grad_norm": 12.0390625, "learning_rate": 9.344371802160319e-06, "loss": 0.7735, "step": 18952 }, { "epoch": 2.016705682060013, "grad_norm": 11.53125, "learning_rate": 9.3443362706083e-06, "loss": 0.8155, "step": 18953 }, { "epoch": 2.0168120876782294, "grad_norm": 9.5703125, "learning_rate": 9.344300739056282e-06, "loss": 0.8956, "step": 18954 }, { "epoch": 2.016918493296446, "grad_norm": 11.75, "learning_rate": 9.344265207504265e-06, "loss": 0.8632, "step": 18955 }, { "epoch": 2.017024898914663, "grad_norm": 10.90625, "learning_rate": 9.344229675952247e-06, "loss": 0.7797, "step": 18956 }, { "epoch": 2.0171313045328794, "grad_norm": 12.9140625, "learning_rate": 9.344194144400228e-06, "loss": 0.8438, "step": 18957 }, { "epoch": 2.017237710151096, "grad_norm": 12.796875, "learning_rate": 9.34415861284821e-06, "loss": 0.8876, "step": 18958 }, { "epoch": 2.017344115769313, "grad_norm": 16.15625, "learning_rate": 9.344123081296192e-06, "loss": 0.8394, "step": 18959 }, { "epoch": 2.0174505213875293, "grad_norm": 12.40625, "learning_rate": 9.344087549744175e-06, "loss": 0.9602, "step": 18960 }, { "epoch": 2.017556927005746, "grad_norm": 10.6484375, "learning_rate": 9.344052018192156e-06, "loss": 0.9575, "step": 18961 }, { "epoch": 2.0176633326239624, "grad_norm": 10.609375, "learning_rate": 9.344016486640138e-06, "loss": 1.0934, "step": 18962 }, { "epoch": 2.0177697382421793, "grad_norm": 11.4296875, "learning_rate": 9.34398095508812e-06, "loss": 0.9349, "step": 18963 }, { "epoch": 2.017876143860396, "grad_norm": 13.1328125, "learning_rate": 9.343945423536101e-06, "loss": 0.7831, "step": 18964 }, { "epoch": 2.0179825494786123, "grad_norm": 18.203125, "learning_rate": 9.343909891984083e-06, "loss": 0.9359, "step": 18965 }, { "epoch": 2.0180889550968293, "grad_norm": 15.03125, "learning_rate": 9.343874360432064e-06, "loss": 0.7921, "step": 18966 }, { "epoch": 2.018195360715046, "grad_norm": 10.484375, "learning_rate": 9.343838828880046e-06, "loss": 0.7597, "step": 18967 }, { "epoch": 2.0183017663332623, "grad_norm": 13.0546875, "learning_rate": 9.343803297328028e-06, "loss": 0.8614, "step": 18968 }, { "epoch": 2.0184081719514793, "grad_norm": 9.3359375, "learning_rate": 9.34376776577601e-06, "loss": 0.7139, "step": 18969 }, { "epoch": 2.0185145775696958, "grad_norm": 13.0234375, "learning_rate": 9.343732234223991e-06, "loss": 0.8374, "step": 18970 }, { "epoch": 2.0186209831879123, "grad_norm": 13.8125, "learning_rate": 9.343696702671974e-06, "loss": 1.1055, "step": 18971 }, { "epoch": 2.018727388806129, "grad_norm": 12.0078125, "learning_rate": 9.343661171119956e-06, "loss": 0.8522, "step": 18972 }, { "epoch": 2.0188337944243457, "grad_norm": 11.828125, "learning_rate": 9.343625639567937e-06, "loss": 0.8402, "step": 18973 }, { "epoch": 2.0189402000425622, "grad_norm": 10.8671875, "learning_rate": 9.343590108015919e-06, "loss": 0.8883, "step": 18974 }, { "epoch": 2.0190466056607788, "grad_norm": 14.5859375, "learning_rate": 9.3435545764639e-06, "loss": 1.0269, "step": 18975 }, { "epoch": 2.0191530112789957, "grad_norm": 18.671875, "learning_rate": 9.343519044911882e-06, "loss": 0.8262, "step": 18976 }, { "epoch": 2.0192594168972122, "grad_norm": 17.0, "learning_rate": 9.343483513359864e-06, "loss": 0.8402, "step": 18977 }, { "epoch": 2.0193658225154287, "grad_norm": 13.390625, "learning_rate": 9.343447981807845e-06, "loss": 0.8091, "step": 18978 }, { "epoch": 2.0194722281336452, "grad_norm": 13.5234375, "learning_rate": 9.343412450255827e-06, "loss": 0.959, "step": 18979 }, { "epoch": 2.019578633751862, "grad_norm": 15.1875, "learning_rate": 9.343376918703809e-06, "loss": 0.9402, "step": 18980 }, { "epoch": 2.0196850393700787, "grad_norm": 12.5, "learning_rate": 9.343341387151792e-06, "loss": 1.0578, "step": 18981 }, { "epoch": 2.019791444988295, "grad_norm": 12.9921875, "learning_rate": 9.343305855599774e-06, "loss": 0.7823, "step": 18982 }, { "epoch": 2.019897850606512, "grad_norm": 13.7109375, "learning_rate": 9.343270324047755e-06, "loss": 0.8445, "step": 18983 }, { "epoch": 2.0200042562247287, "grad_norm": 15.203125, "learning_rate": 9.343234792495737e-06, "loss": 1.0617, "step": 18984 }, { "epoch": 2.020110661842945, "grad_norm": 13.5859375, "learning_rate": 9.343199260943718e-06, "loss": 0.9182, "step": 18985 }, { "epoch": 2.020217067461162, "grad_norm": 17.1875, "learning_rate": 9.343163729391702e-06, "loss": 0.9758, "step": 18986 }, { "epoch": 2.0203234730793787, "grad_norm": 14.84375, "learning_rate": 9.343128197839683e-06, "loss": 1.1753, "step": 18987 }, { "epoch": 2.020429878697595, "grad_norm": 11.78125, "learning_rate": 9.343092666287665e-06, "loss": 1.0063, "step": 18988 }, { "epoch": 2.0205362843158117, "grad_norm": 14.546875, "learning_rate": 9.343057134735647e-06, "loss": 0.8894, "step": 18989 }, { "epoch": 2.0206426899340286, "grad_norm": 13.8359375, "learning_rate": 9.343021603183628e-06, "loss": 0.7333, "step": 18990 }, { "epoch": 2.020749095552245, "grad_norm": 12.796875, "learning_rate": 9.34298607163161e-06, "loss": 0.8282, "step": 18991 }, { "epoch": 2.0208555011704616, "grad_norm": 18.015625, "learning_rate": 9.342950540079591e-06, "loss": 0.8947, "step": 18992 }, { "epoch": 2.0209619067886786, "grad_norm": 18.546875, "learning_rate": 9.342915008527573e-06, "loss": 0.8625, "step": 18993 }, { "epoch": 2.021068312406895, "grad_norm": 14.3046875, "learning_rate": 9.342879476975555e-06, "loss": 0.8567, "step": 18994 }, { "epoch": 2.0211747180251116, "grad_norm": 17.59375, "learning_rate": 9.342843945423536e-06, "loss": 0.9204, "step": 18995 }, { "epoch": 2.0212811236433286, "grad_norm": 15.265625, "learning_rate": 9.34280841387152e-06, "loss": 0.9425, "step": 18996 }, { "epoch": 2.021387529261545, "grad_norm": 12.8671875, "learning_rate": 9.342772882319501e-06, "loss": 0.9026, "step": 18997 }, { "epoch": 2.0214939348797616, "grad_norm": 13.171875, "learning_rate": 9.342737350767483e-06, "loss": 0.868, "step": 18998 }, { "epoch": 2.021600340497978, "grad_norm": 13.171875, "learning_rate": 9.342701819215464e-06, "loss": 0.9861, "step": 18999 }, { "epoch": 2.021706746116195, "grad_norm": 19.59375, "learning_rate": 9.342666287663446e-06, "loss": 0.9445, "step": 19000 }, { "epoch": 2.0218131517344116, "grad_norm": 18.453125, "learning_rate": 9.342630756111427e-06, "loss": 0.9317, "step": 19001 }, { "epoch": 2.021919557352628, "grad_norm": 15.21875, "learning_rate": 9.342595224559409e-06, "loss": 1.0145, "step": 19002 }, { "epoch": 2.022025962970845, "grad_norm": 11.78125, "learning_rate": 9.34255969300739e-06, "loss": 1.0022, "step": 19003 }, { "epoch": 2.0221323685890615, "grad_norm": 13.03125, "learning_rate": 9.342524161455372e-06, "loss": 0.7635, "step": 19004 }, { "epoch": 2.022238774207278, "grad_norm": 11.015625, "learning_rate": 9.342488629903354e-06, "loss": 0.856, "step": 19005 }, { "epoch": 2.022345179825495, "grad_norm": 11.6640625, "learning_rate": 9.342453098351337e-06, "loss": 0.8738, "step": 19006 }, { "epoch": 2.0224515854437115, "grad_norm": 11.1171875, "learning_rate": 9.342417566799319e-06, "loss": 0.8828, "step": 19007 }, { "epoch": 2.022557991061928, "grad_norm": 13.1875, "learning_rate": 9.3423820352473e-06, "loss": 1.0149, "step": 19008 }, { "epoch": 2.0226643966801445, "grad_norm": 12.390625, "learning_rate": 9.342346503695282e-06, "loss": 0.8418, "step": 19009 }, { "epoch": 2.0227708022983615, "grad_norm": 15.0546875, "learning_rate": 9.342310972143264e-06, "loss": 1.0676, "step": 19010 }, { "epoch": 2.022877207916578, "grad_norm": 13.9296875, "learning_rate": 9.342275440591245e-06, "loss": 0.928, "step": 19011 }, { "epoch": 2.0229836135347945, "grad_norm": 15.546875, "learning_rate": 9.342239909039229e-06, "loss": 0.9921, "step": 19012 }, { "epoch": 2.0230900191530115, "grad_norm": 14.7578125, "learning_rate": 9.34220437748721e-06, "loss": 0.9308, "step": 19013 }, { "epoch": 2.023196424771228, "grad_norm": 9.859375, "learning_rate": 9.342168845935192e-06, "loss": 0.9188, "step": 19014 }, { "epoch": 2.0233028303894445, "grad_norm": 15.390625, "learning_rate": 9.342133314383173e-06, "loss": 0.885, "step": 19015 }, { "epoch": 2.0234092360076614, "grad_norm": 13.125, "learning_rate": 9.342097782831155e-06, "loss": 0.9978, "step": 19016 }, { "epoch": 2.023515641625878, "grad_norm": 13.046875, "learning_rate": 9.342062251279137e-06, "loss": 0.9145, "step": 19017 }, { "epoch": 2.0236220472440944, "grad_norm": 18.234375, "learning_rate": 9.342026719727118e-06, "loss": 1.0159, "step": 19018 }, { "epoch": 2.023728452862311, "grad_norm": 11.9921875, "learning_rate": 9.3419911881751e-06, "loss": 0.7899, "step": 19019 }, { "epoch": 2.023834858480528, "grad_norm": 11.7265625, "learning_rate": 9.341955656623081e-06, "loss": 0.8497, "step": 19020 }, { "epoch": 2.0239412640987444, "grad_norm": 10.6328125, "learning_rate": 9.341920125071063e-06, "loss": 0.7807, "step": 19021 }, { "epoch": 2.024047669716961, "grad_norm": 13.6015625, "learning_rate": 9.341884593519046e-06, "loss": 0.9714, "step": 19022 }, { "epoch": 2.024154075335178, "grad_norm": 17.21875, "learning_rate": 9.341849061967028e-06, "loss": 0.7665, "step": 19023 }, { "epoch": 2.0242604809533944, "grad_norm": 14.2421875, "learning_rate": 9.34181353041501e-06, "loss": 0.8583, "step": 19024 }, { "epoch": 2.024366886571611, "grad_norm": 13.2421875, "learning_rate": 9.341777998862991e-06, "loss": 0.9033, "step": 19025 }, { "epoch": 2.0244732921898274, "grad_norm": 9.5390625, "learning_rate": 9.341742467310973e-06, "loss": 0.8853, "step": 19026 }, { "epoch": 2.0245796978080444, "grad_norm": 13.15625, "learning_rate": 9.341706935758954e-06, "loss": 0.829, "step": 19027 }, { "epoch": 2.024686103426261, "grad_norm": 13.453125, "learning_rate": 9.341671404206936e-06, "loss": 1.1139, "step": 19028 }, { "epoch": 2.0247925090444774, "grad_norm": 10.5, "learning_rate": 9.341635872654918e-06, "loss": 0.8582, "step": 19029 }, { "epoch": 2.0248989146626943, "grad_norm": 12.6015625, "learning_rate": 9.3416003411029e-06, "loss": 0.8339, "step": 19030 }, { "epoch": 2.025005320280911, "grad_norm": 9.3125, "learning_rate": 9.34156480955088e-06, "loss": 0.8718, "step": 19031 }, { "epoch": 2.0251117258991274, "grad_norm": 15.3984375, "learning_rate": 9.341529277998864e-06, "loss": 0.9395, "step": 19032 }, { "epoch": 2.0252181315173443, "grad_norm": 10.09375, "learning_rate": 9.341493746446846e-06, "loss": 0.8249, "step": 19033 }, { "epoch": 2.025324537135561, "grad_norm": 12.09375, "learning_rate": 9.341458214894827e-06, "loss": 0.8807, "step": 19034 }, { "epoch": 2.0254309427537773, "grad_norm": 12.6875, "learning_rate": 9.341422683342809e-06, "loss": 0.9606, "step": 19035 }, { "epoch": 2.025537348371994, "grad_norm": 14.21875, "learning_rate": 9.34138715179079e-06, "loss": 0.9602, "step": 19036 }, { "epoch": 2.025643753990211, "grad_norm": 9.0234375, "learning_rate": 9.341351620238774e-06, "loss": 0.716, "step": 19037 }, { "epoch": 2.0257501596084273, "grad_norm": 12.21875, "learning_rate": 9.341316088686755e-06, "loss": 0.9617, "step": 19038 }, { "epoch": 2.025856565226644, "grad_norm": 11.4140625, "learning_rate": 9.341280557134737e-06, "loss": 0.9318, "step": 19039 }, { "epoch": 2.0259629708448608, "grad_norm": 12.453125, "learning_rate": 9.341245025582719e-06, "loss": 0.8988, "step": 19040 }, { "epoch": 2.0260693764630773, "grad_norm": 12.1796875, "learning_rate": 9.3412094940307e-06, "loss": 0.7351, "step": 19041 }, { "epoch": 2.026175782081294, "grad_norm": 9.421875, "learning_rate": 9.341173962478682e-06, "loss": 0.7008, "step": 19042 }, { "epoch": 2.0262821876995107, "grad_norm": 16.453125, "learning_rate": 9.341138430926663e-06, "loss": 0.8148, "step": 19043 }, { "epoch": 2.0263885933177272, "grad_norm": 11.3359375, "learning_rate": 9.341102899374645e-06, "loss": 0.8436, "step": 19044 }, { "epoch": 2.0264949989359438, "grad_norm": 20.25, "learning_rate": 9.341067367822627e-06, "loss": 0.8133, "step": 19045 }, { "epoch": 2.0266014045541603, "grad_norm": 15.0625, "learning_rate": 9.341031836270608e-06, "loss": 0.9539, "step": 19046 }, { "epoch": 2.0267078101723772, "grad_norm": 14.9140625, "learning_rate": 9.340996304718592e-06, "loss": 1.0258, "step": 19047 }, { "epoch": 2.0268142157905937, "grad_norm": 17.953125, "learning_rate": 9.340960773166573e-06, "loss": 1.1714, "step": 19048 }, { "epoch": 2.0269206214088102, "grad_norm": 9.1953125, "learning_rate": 9.340925241614555e-06, "loss": 0.6446, "step": 19049 }, { "epoch": 2.027027027027027, "grad_norm": 15.34375, "learning_rate": 9.340889710062536e-06, "loss": 1.0059, "step": 19050 }, { "epoch": 2.0271334326452437, "grad_norm": 13.1328125, "learning_rate": 9.340854178510518e-06, "loss": 0.813, "step": 19051 }, { "epoch": 2.02723983826346, "grad_norm": 9.3515625, "learning_rate": 9.3408186469585e-06, "loss": 0.9042, "step": 19052 }, { "epoch": 2.027346243881677, "grad_norm": 18.46875, "learning_rate": 9.340783115406483e-06, "loss": 0.9668, "step": 19053 }, { "epoch": 2.0274526494998937, "grad_norm": 15.34375, "learning_rate": 9.340747583854465e-06, "loss": 0.9654, "step": 19054 }, { "epoch": 2.02755905511811, "grad_norm": 20.796875, "learning_rate": 9.340712052302444e-06, "loss": 0.7237, "step": 19055 }, { "epoch": 2.0276654607363267, "grad_norm": 14.859375, "learning_rate": 9.340676520750426e-06, "loss": 0.831, "step": 19056 }, { "epoch": 2.0277718663545437, "grad_norm": 11.1953125, "learning_rate": 9.340640989198408e-06, "loss": 0.9681, "step": 19057 }, { "epoch": 2.02787827197276, "grad_norm": 9.328125, "learning_rate": 9.340605457646391e-06, "loss": 0.8102, "step": 19058 }, { "epoch": 2.0279846775909767, "grad_norm": 9.09375, "learning_rate": 9.340569926094373e-06, "loss": 0.6909, "step": 19059 }, { "epoch": 2.0280910832091936, "grad_norm": 12.234375, "learning_rate": 9.340534394542354e-06, "loss": 0.9599, "step": 19060 }, { "epoch": 2.02819748882741, "grad_norm": 16.09375, "learning_rate": 9.340498862990336e-06, "loss": 1.1584, "step": 19061 }, { "epoch": 2.0283038944456266, "grad_norm": 12.140625, "learning_rate": 9.340463331438317e-06, "loss": 0.8964, "step": 19062 }, { "epoch": 2.028410300063843, "grad_norm": 16.765625, "learning_rate": 9.3404277998863e-06, "loss": 0.8734, "step": 19063 }, { "epoch": 2.02851670568206, "grad_norm": 11.65625, "learning_rate": 9.340392268334282e-06, "loss": 0.8329, "step": 19064 }, { "epoch": 2.0286231113002766, "grad_norm": 11.9140625, "learning_rate": 9.340356736782264e-06, "loss": 0.9261, "step": 19065 }, { "epoch": 2.028729516918493, "grad_norm": 14.2109375, "learning_rate": 9.340321205230245e-06, "loss": 0.7408, "step": 19066 }, { "epoch": 2.02883592253671, "grad_norm": 12.2265625, "learning_rate": 9.340285673678227e-06, "loss": 0.7451, "step": 19067 }, { "epoch": 2.0289423281549266, "grad_norm": 12.6484375, "learning_rate": 9.340250142126209e-06, "loss": 0.8585, "step": 19068 }, { "epoch": 2.029048733773143, "grad_norm": 12.8671875, "learning_rate": 9.34021461057419e-06, "loss": 0.9211, "step": 19069 }, { "epoch": 2.02915513939136, "grad_norm": 12.25, "learning_rate": 9.340179079022172e-06, "loss": 0.7625, "step": 19070 }, { "epoch": 2.0292615450095766, "grad_norm": 10.6328125, "learning_rate": 9.340143547470154e-06, "loss": 0.8519, "step": 19071 }, { "epoch": 2.029367950627793, "grad_norm": 9.8125, "learning_rate": 9.340108015918135e-06, "loss": 0.8168, "step": 19072 }, { "epoch": 2.0294743562460096, "grad_norm": 14.9609375, "learning_rate": 9.340072484366118e-06, "loss": 0.9517, "step": 19073 }, { "epoch": 2.0295807618642265, "grad_norm": 14.0625, "learning_rate": 9.3400369528141e-06, "loss": 0.9417, "step": 19074 }, { "epoch": 2.029687167482443, "grad_norm": 10.9765625, "learning_rate": 9.340001421262082e-06, "loss": 0.9342, "step": 19075 }, { "epoch": 2.0297935731006596, "grad_norm": 12.9296875, "learning_rate": 9.339965889710063e-06, "loss": 0.932, "step": 19076 }, { "epoch": 2.0298999787188765, "grad_norm": 14.2734375, "learning_rate": 9.339930358158045e-06, "loss": 1.0508, "step": 19077 }, { "epoch": 2.030006384337093, "grad_norm": 14.40625, "learning_rate": 9.339894826606028e-06, "loss": 0.8288, "step": 19078 }, { "epoch": 2.0301127899553095, "grad_norm": 15.625, "learning_rate": 9.33985929505401e-06, "loss": 0.7096, "step": 19079 }, { "epoch": 2.0302191955735265, "grad_norm": 11.2890625, "learning_rate": 9.339823763501991e-06, "loss": 0.7597, "step": 19080 }, { "epoch": 2.030325601191743, "grad_norm": 12.8125, "learning_rate": 9.339788231949973e-06, "loss": 0.8484, "step": 19081 }, { "epoch": 2.0304320068099595, "grad_norm": 16.921875, "learning_rate": 9.339752700397955e-06, "loss": 0.9022, "step": 19082 }, { "epoch": 2.030538412428176, "grad_norm": 11.9609375, "learning_rate": 9.339717168845936e-06, "loss": 0.8191, "step": 19083 }, { "epoch": 2.030644818046393, "grad_norm": 16.546875, "learning_rate": 9.339681637293918e-06, "loss": 0.7858, "step": 19084 }, { "epoch": 2.0307512236646095, "grad_norm": 12.2734375, "learning_rate": 9.3396461057419e-06, "loss": 0.9344, "step": 19085 }, { "epoch": 2.030857629282826, "grad_norm": 11.734375, "learning_rate": 9.339610574189881e-06, "loss": 0.8237, "step": 19086 }, { "epoch": 2.030964034901043, "grad_norm": 10.9921875, "learning_rate": 9.339575042637863e-06, "loss": 0.7519, "step": 19087 }, { "epoch": 2.0310704405192594, "grad_norm": 10.8984375, "learning_rate": 9.339539511085844e-06, "loss": 0.7264, "step": 19088 }, { "epoch": 2.031176846137476, "grad_norm": 10.8671875, "learning_rate": 9.339503979533828e-06, "loss": 0.9072, "step": 19089 }, { "epoch": 2.031283251755693, "grad_norm": 13.109375, "learning_rate": 9.339468447981809e-06, "loss": 0.8912, "step": 19090 }, { "epoch": 2.0313896573739094, "grad_norm": 10.1328125, "learning_rate": 9.33943291642979e-06, "loss": 0.6386, "step": 19091 }, { "epoch": 2.031496062992126, "grad_norm": 17.171875, "learning_rate": 9.339397384877772e-06, "loss": 0.8993, "step": 19092 }, { "epoch": 2.0316024686103424, "grad_norm": 14.515625, "learning_rate": 9.339361853325754e-06, "loss": 1.0371, "step": 19093 }, { "epoch": 2.0317088742285594, "grad_norm": 12.078125, "learning_rate": 9.339326321773736e-06, "loss": 0.88, "step": 19094 }, { "epoch": 2.031815279846776, "grad_norm": 15.859375, "learning_rate": 9.339290790221717e-06, "loss": 0.8219, "step": 19095 }, { "epoch": 2.0319216854649924, "grad_norm": 14.953125, "learning_rate": 9.339255258669699e-06, "loss": 0.9804, "step": 19096 }, { "epoch": 2.0320280910832094, "grad_norm": 14.1171875, "learning_rate": 9.33921972711768e-06, "loss": 0.9787, "step": 19097 }, { "epoch": 2.032134496701426, "grad_norm": 9.78125, "learning_rate": 9.339184195565662e-06, "loss": 0.9353, "step": 19098 }, { "epoch": 2.0322409023196424, "grad_norm": 11.515625, "learning_rate": 9.339148664013645e-06, "loss": 0.6702, "step": 19099 }, { "epoch": 2.032347307937859, "grad_norm": 10.7890625, "learning_rate": 9.339113132461627e-06, "loss": 0.7789, "step": 19100 }, { "epoch": 2.032453713556076, "grad_norm": 13.9765625, "learning_rate": 9.339077600909608e-06, "loss": 1.0865, "step": 19101 }, { "epoch": 2.0325601191742924, "grad_norm": 18.234375, "learning_rate": 9.33904206935759e-06, "loss": 0.8086, "step": 19102 }, { "epoch": 2.032666524792509, "grad_norm": 14.8828125, "learning_rate": 9.339006537805572e-06, "loss": 0.9958, "step": 19103 }, { "epoch": 2.032772930410726, "grad_norm": 11.59375, "learning_rate": 9.338971006253555e-06, "loss": 0.9599, "step": 19104 }, { "epoch": 2.0328793360289423, "grad_norm": 16.875, "learning_rate": 9.338935474701537e-06, "loss": 1.1397, "step": 19105 }, { "epoch": 2.032985741647159, "grad_norm": 12.6640625, "learning_rate": 9.338899943149518e-06, "loss": 0.877, "step": 19106 }, { "epoch": 2.033092147265376, "grad_norm": 15.265625, "learning_rate": 9.3388644115975e-06, "loss": 0.9459, "step": 19107 }, { "epoch": 2.0331985528835923, "grad_norm": 12.984375, "learning_rate": 9.338828880045481e-06, "loss": 0.8538, "step": 19108 }, { "epoch": 2.033304958501809, "grad_norm": 16.375, "learning_rate": 9.338793348493463e-06, "loss": 0.9848, "step": 19109 }, { "epoch": 2.0334113641200253, "grad_norm": 11.3203125, "learning_rate": 9.338757816941445e-06, "loss": 0.808, "step": 19110 }, { "epoch": 2.0335177697382423, "grad_norm": 10.5859375, "learning_rate": 9.338722285389426e-06, "loss": 0.846, "step": 19111 }, { "epoch": 2.033624175356459, "grad_norm": 8.6875, "learning_rate": 9.338686753837408e-06, "loss": 0.6545, "step": 19112 }, { "epoch": 2.0337305809746753, "grad_norm": 10.4921875, "learning_rate": 9.33865122228539e-06, "loss": 0.7998, "step": 19113 }, { "epoch": 2.0338369865928922, "grad_norm": 11.140625, "learning_rate": 9.338615690733373e-06, "loss": 0.7724, "step": 19114 }, { "epoch": 2.0339433922111088, "grad_norm": 14.6875, "learning_rate": 9.338580159181354e-06, "loss": 0.7703, "step": 19115 }, { "epoch": 2.0340497978293253, "grad_norm": 14.7578125, "learning_rate": 9.338544627629336e-06, "loss": 0.922, "step": 19116 }, { "epoch": 2.034156203447542, "grad_norm": 17.65625, "learning_rate": 9.338509096077318e-06, "loss": 0.9903, "step": 19117 }, { "epoch": 2.0342626090657587, "grad_norm": 11.8984375, "learning_rate": 9.3384735645253e-06, "loss": 0.806, "step": 19118 }, { "epoch": 2.0343690146839752, "grad_norm": 14.0625, "learning_rate": 9.33843803297328e-06, "loss": 0.9678, "step": 19119 }, { "epoch": 2.0344754203021918, "grad_norm": 10.0703125, "learning_rate": 9.338402501421262e-06, "loss": 0.8798, "step": 19120 }, { "epoch": 2.0345818259204087, "grad_norm": 10.140625, "learning_rate": 9.338366969869244e-06, "loss": 0.8886, "step": 19121 }, { "epoch": 2.034688231538625, "grad_norm": 13.3984375, "learning_rate": 9.338331438317226e-06, "loss": 0.8714, "step": 19122 }, { "epoch": 2.0347946371568417, "grad_norm": 11.3359375, "learning_rate": 9.338295906765207e-06, "loss": 0.8655, "step": 19123 }, { "epoch": 2.0349010427750587, "grad_norm": 14.0078125, "learning_rate": 9.33826037521319e-06, "loss": 0.9261, "step": 19124 }, { "epoch": 2.035007448393275, "grad_norm": 11.671875, "learning_rate": 9.338224843661172e-06, "loss": 0.7831, "step": 19125 }, { "epoch": 2.0351138540114917, "grad_norm": 15.421875, "learning_rate": 9.338189312109154e-06, "loss": 0.7661, "step": 19126 }, { "epoch": 2.0352202596297087, "grad_norm": 16.53125, "learning_rate": 9.338153780557135e-06, "loss": 0.9648, "step": 19127 }, { "epoch": 2.035326665247925, "grad_norm": 10.9765625, "learning_rate": 9.338118249005117e-06, "loss": 0.9589, "step": 19128 }, { "epoch": 2.0354330708661417, "grad_norm": 16.15625, "learning_rate": 9.338082717453099e-06, "loss": 0.8277, "step": 19129 }, { "epoch": 2.035539476484358, "grad_norm": 11.5390625, "learning_rate": 9.338047185901082e-06, "loss": 0.8051, "step": 19130 }, { "epoch": 2.035645882102575, "grad_norm": 12.2109375, "learning_rate": 9.338011654349063e-06, "loss": 0.8382, "step": 19131 }, { "epoch": 2.0357522877207916, "grad_norm": 13.8046875, "learning_rate": 9.337976122797045e-06, "loss": 0.6523, "step": 19132 }, { "epoch": 2.035858693339008, "grad_norm": 11.9765625, "learning_rate": 9.337940591245027e-06, "loss": 0.9521, "step": 19133 }, { "epoch": 2.035965098957225, "grad_norm": 13.765625, "learning_rate": 9.337905059693008e-06, "loss": 0.8854, "step": 19134 }, { "epoch": 2.0360715045754416, "grad_norm": 12.7421875, "learning_rate": 9.33786952814099e-06, "loss": 0.9047, "step": 19135 }, { "epoch": 2.036177910193658, "grad_norm": 10.6171875, "learning_rate": 9.337833996588972e-06, "loss": 0.9382, "step": 19136 }, { "epoch": 2.036284315811875, "grad_norm": 11.65625, "learning_rate": 9.337798465036953e-06, "loss": 0.8828, "step": 19137 }, { "epoch": 2.0363907214300916, "grad_norm": 10.359375, "learning_rate": 9.337762933484935e-06, "loss": 0.7749, "step": 19138 }, { "epoch": 2.036497127048308, "grad_norm": 16.984375, "learning_rate": 9.337727401932916e-06, "loss": 0.9687, "step": 19139 }, { "epoch": 2.0366035326665246, "grad_norm": 11.7109375, "learning_rate": 9.3376918703809e-06, "loss": 0.8998, "step": 19140 }, { "epoch": 2.0367099382847416, "grad_norm": 20.5, "learning_rate": 9.337656338828881e-06, "loss": 0.9898, "step": 19141 }, { "epoch": 2.036816343902958, "grad_norm": 10.078125, "learning_rate": 9.337620807276863e-06, "loss": 0.7424, "step": 19142 }, { "epoch": 2.0369227495211746, "grad_norm": 11.125, "learning_rate": 9.337585275724844e-06, "loss": 0.9245, "step": 19143 }, { "epoch": 2.0370291551393915, "grad_norm": 10.796875, "learning_rate": 9.337549744172826e-06, "loss": 0.8787, "step": 19144 }, { "epoch": 2.037135560757608, "grad_norm": 15.484375, "learning_rate": 9.337514212620808e-06, "loss": 0.9326, "step": 19145 }, { "epoch": 2.0372419663758246, "grad_norm": 13.1796875, "learning_rate": 9.33747868106879e-06, "loss": 0.9652, "step": 19146 }, { "epoch": 2.0373483719940415, "grad_norm": 10.4140625, "learning_rate": 9.337443149516771e-06, "loss": 0.9009, "step": 19147 }, { "epoch": 2.037454777612258, "grad_norm": 16.265625, "learning_rate": 9.337407617964752e-06, "loss": 0.8924, "step": 19148 }, { "epoch": 2.0375611832304745, "grad_norm": 12.875, "learning_rate": 9.337372086412734e-06, "loss": 0.7663, "step": 19149 }, { "epoch": 2.037667588848691, "grad_norm": 12.5234375, "learning_rate": 9.337336554860717e-06, "loss": 0.8527, "step": 19150 }, { "epoch": 2.037773994466908, "grad_norm": 12.359375, "learning_rate": 9.337301023308699e-06, "loss": 1.0046, "step": 19151 }, { "epoch": 2.0378804000851245, "grad_norm": 13.5703125, "learning_rate": 9.33726549175668e-06, "loss": 0.7983, "step": 19152 }, { "epoch": 2.037986805703341, "grad_norm": 18.96875, "learning_rate": 9.337229960204662e-06, "loss": 1.047, "step": 19153 }, { "epoch": 2.038093211321558, "grad_norm": 14.3359375, "learning_rate": 9.337194428652644e-06, "loss": 0.7837, "step": 19154 }, { "epoch": 2.0381996169397745, "grad_norm": 12.15625, "learning_rate": 9.337158897100627e-06, "loss": 1.061, "step": 19155 }, { "epoch": 2.038306022557991, "grad_norm": 14.0703125, "learning_rate": 9.337123365548609e-06, "loss": 0.9623, "step": 19156 }, { "epoch": 2.0384124281762075, "grad_norm": 15.1796875, "learning_rate": 9.33708783399659e-06, "loss": 1.0092, "step": 19157 }, { "epoch": 2.0385188337944244, "grad_norm": 12.4375, "learning_rate": 9.337052302444572e-06, "loss": 0.9012, "step": 19158 }, { "epoch": 2.038625239412641, "grad_norm": 13.328125, "learning_rate": 9.337016770892554e-06, "loss": 0.8871, "step": 19159 }, { "epoch": 2.0387316450308575, "grad_norm": 12.15625, "learning_rate": 9.336981239340535e-06, "loss": 0.788, "step": 19160 }, { "epoch": 2.0388380506490744, "grad_norm": 14.59375, "learning_rate": 9.336945707788517e-06, "loss": 1.0043, "step": 19161 }, { "epoch": 2.038944456267291, "grad_norm": 10.171875, "learning_rate": 9.336910176236498e-06, "loss": 0.8525, "step": 19162 }, { "epoch": 2.0390508618855074, "grad_norm": 11.4140625, "learning_rate": 9.33687464468448e-06, "loss": 0.8206, "step": 19163 }, { "epoch": 2.0391572675037244, "grad_norm": 13.6171875, "learning_rate": 9.336839113132462e-06, "loss": 0.763, "step": 19164 }, { "epoch": 2.039263673121941, "grad_norm": 12.3125, "learning_rate": 9.336803581580443e-06, "loss": 0.8907, "step": 19165 }, { "epoch": 2.0393700787401574, "grad_norm": 12.5859375, "learning_rate": 9.336768050028426e-06, "loss": 0.8807, "step": 19166 }, { "epoch": 2.039476484358374, "grad_norm": 12.40625, "learning_rate": 9.336732518476408e-06, "loss": 0.9624, "step": 19167 }, { "epoch": 2.039582889976591, "grad_norm": 13.953125, "learning_rate": 9.33669698692439e-06, "loss": 0.9439, "step": 19168 }, { "epoch": 2.0396892955948074, "grad_norm": 10.4921875, "learning_rate": 9.336661455372371e-06, "loss": 0.7841, "step": 19169 }, { "epoch": 2.039795701213024, "grad_norm": 16.234375, "learning_rate": 9.336625923820353e-06, "loss": 0.7699, "step": 19170 }, { "epoch": 2.039902106831241, "grad_norm": 13.125, "learning_rate": 9.336590392268336e-06, "loss": 1.0638, "step": 19171 }, { "epoch": 2.0400085124494574, "grad_norm": 19.515625, "learning_rate": 9.336554860716318e-06, "loss": 0.9749, "step": 19172 }, { "epoch": 2.040114918067674, "grad_norm": 13.234375, "learning_rate": 9.3365193291643e-06, "loss": 0.8574, "step": 19173 }, { "epoch": 2.040221323685891, "grad_norm": 13.9921875, "learning_rate": 9.336483797612281e-06, "loss": 1.0705, "step": 19174 }, { "epoch": 2.0403277293041073, "grad_norm": 13.6328125, "learning_rate": 9.336448266060261e-06, "loss": 0.8019, "step": 19175 }, { "epoch": 2.040434134922324, "grad_norm": 13.7109375, "learning_rate": 9.336412734508244e-06, "loss": 0.8978, "step": 19176 }, { "epoch": 2.0405405405405403, "grad_norm": 15.4765625, "learning_rate": 9.336377202956226e-06, "loss": 0.7939, "step": 19177 }, { "epoch": 2.0406469461587573, "grad_norm": 12.5546875, "learning_rate": 9.336341671404207e-06, "loss": 0.9393, "step": 19178 }, { "epoch": 2.040753351776974, "grad_norm": 13.3515625, "learning_rate": 9.336306139852189e-06, "loss": 0.7728, "step": 19179 }, { "epoch": 2.0408597573951903, "grad_norm": 12.4453125, "learning_rate": 9.33627060830017e-06, "loss": 0.9149, "step": 19180 }, { "epoch": 2.0409661630134073, "grad_norm": 12.984375, "learning_rate": 9.336235076748154e-06, "loss": 0.7219, "step": 19181 }, { "epoch": 2.041072568631624, "grad_norm": 13.15625, "learning_rate": 9.336199545196136e-06, "loss": 0.9632, "step": 19182 }, { "epoch": 2.0411789742498403, "grad_norm": 14.609375, "learning_rate": 9.336164013644117e-06, "loss": 0.9379, "step": 19183 }, { "epoch": 2.0412853798680572, "grad_norm": 16.21875, "learning_rate": 9.336128482092099e-06, "loss": 0.7972, "step": 19184 }, { "epoch": 2.0413917854862738, "grad_norm": 11.34375, "learning_rate": 9.33609295054008e-06, "loss": 0.7999, "step": 19185 }, { "epoch": 2.0414981911044903, "grad_norm": 15.46875, "learning_rate": 9.336057418988062e-06, "loss": 0.7333, "step": 19186 }, { "epoch": 2.0416045967227068, "grad_norm": 14.8203125, "learning_rate": 9.336021887436044e-06, "loss": 0.8859, "step": 19187 }, { "epoch": 2.0417110023409237, "grad_norm": 11.890625, "learning_rate": 9.335986355884025e-06, "loss": 1.0953, "step": 19188 }, { "epoch": 2.0418174079591402, "grad_norm": 11.515625, "learning_rate": 9.335950824332007e-06, "loss": 0.8201, "step": 19189 }, { "epoch": 2.0419238135773567, "grad_norm": 11.4765625, "learning_rate": 9.335915292779988e-06, "loss": 0.7352, "step": 19190 }, { "epoch": 2.0420302191955737, "grad_norm": 18.328125, "learning_rate": 9.335879761227972e-06, "loss": 0.785, "step": 19191 }, { "epoch": 2.04213662481379, "grad_norm": 11.671875, "learning_rate": 9.335844229675953e-06, "loss": 0.8409, "step": 19192 }, { "epoch": 2.0422430304320067, "grad_norm": 16.75, "learning_rate": 9.335808698123935e-06, "loss": 0.9456, "step": 19193 }, { "epoch": 2.0423494360502232, "grad_norm": 11.8671875, "learning_rate": 9.335773166571917e-06, "loss": 0.8644, "step": 19194 }, { "epoch": 2.04245584166844, "grad_norm": 10.875, "learning_rate": 9.335737635019898e-06, "loss": 0.882, "step": 19195 }, { "epoch": 2.0425622472866567, "grad_norm": 19.75, "learning_rate": 9.335702103467881e-06, "loss": 0.9112, "step": 19196 }, { "epoch": 2.042668652904873, "grad_norm": 15.296875, "learning_rate": 9.335666571915863e-06, "loss": 0.9541, "step": 19197 }, { "epoch": 2.04277505852309, "grad_norm": 17.890625, "learning_rate": 9.335631040363845e-06, "loss": 0.827, "step": 19198 }, { "epoch": 2.0428814641413067, "grad_norm": 15.8515625, "learning_rate": 9.335595508811826e-06, "loss": 0.8999, "step": 19199 }, { "epoch": 2.042987869759523, "grad_norm": 15.5703125, "learning_rate": 9.335559977259808e-06, "loss": 0.92, "step": 19200 }, { "epoch": 2.04309427537774, "grad_norm": 15.1171875, "learning_rate": 9.33552444570779e-06, "loss": 1.0108, "step": 19201 }, { "epoch": 2.0432006809959566, "grad_norm": 9.046875, "learning_rate": 9.335488914155771e-06, "loss": 0.658, "step": 19202 }, { "epoch": 2.043307086614173, "grad_norm": 10.484375, "learning_rate": 9.335453382603753e-06, "loss": 0.8321, "step": 19203 }, { "epoch": 2.0434134922323897, "grad_norm": 16.5, "learning_rate": 9.335417851051734e-06, "loss": 1.0249, "step": 19204 }, { "epoch": 2.0435198978506066, "grad_norm": 14.296875, "learning_rate": 9.335382319499716e-06, "loss": 0.8582, "step": 19205 }, { "epoch": 2.043626303468823, "grad_norm": 11.0234375, "learning_rate": 9.335346787947698e-06, "loss": 0.6959, "step": 19206 }, { "epoch": 2.0437327090870396, "grad_norm": 11.5703125, "learning_rate": 9.33531125639568e-06, "loss": 0.9445, "step": 19207 }, { "epoch": 2.0438391147052566, "grad_norm": 15.96875, "learning_rate": 9.335275724843662e-06, "loss": 0.8742, "step": 19208 }, { "epoch": 2.043945520323473, "grad_norm": 18.859375, "learning_rate": 9.335240193291644e-06, "loss": 1.0305, "step": 19209 }, { "epoch": 2.0440519259416896, "grad_norm": 14.03125, "learning_rate": 9.335204661739626e-06, "loss": 0.9562, "step": 19210 }, { "epoch": 2.0441583315599066, "grad_norm": 10.5546875, "learning_rate": 9.335169130187607e-06, "loss": 0.7709, "step": 19211 }, { "epoch": 2.044264737178123, "grad_norm": 14.28125, "learning_rate": 9.335133598635589e-06, "loss": 0.9495, "step": 19212 }, { "epoch": 2.0443711427963396, "grad_norm": 18.15625, "learning_rate": 9.33509806708357e-06, "loss": 1.0353, "step": 19213 }, { "epoch": 2.044477548414556, "grad_norm": 10.7578125, "learning_rate": 9.335062535531552e-06, "loss": 0.7887, "step": 19214 }, { "epoch": 2.044583954032773, "grad_norm": 11.203125, "learning_rate": 9.335027003979534e-06, "loss": 0.8527, "step": 19215 }, { "epoch": 2.0446903596509896, "grad_norm": 11.3125, "learning_rate": 9.334991472427515e-06, "loss": 0.8365, "step": 19216 }, { "epoch": 2.044796765269206, "grad_norm": 13.0078125, "learning_rate": 9.334955940875499e-06, "loss": 0.8472, "step": 19217 }, { "epoch": 2.044903170887423, "grad_norm": 13.953125, "learning_rate": 9.33492040932348e-06, "loss": 1.0149, "step": 19218 }, { "epoch": 2.0450095765056395, "grad_norm": 16.640625, "learning_rate": 9.334884877771462e-06, "loss": 1.1319, "step": 19219 }, { "epoch": 2.045115982123856, "grad_norm": 13.71875, "learning_rate": 9.334849346219443e-06, "loss": 1.0039, "step": 19220 }, { "epoch": 2.045222387742073, "grad_norm": 11.8828125, "learning_rate": 9.334813814667425e-06, "loss": 0.9655, "step": 19221 }, { "epoch": 2.0453287933602895, "grad_norm": 10.359375, "learning_rate": 9.334778283115408e-06, "loss": 0.8204, "step": 19222 }, { "epoch": 2.045435198978506, "grad_norm": 16.78125, "learning_rate": 9.33474275156339e-06, "loss": 0.7285, "step": 19223 }, { "epoch": 2.0455416045967225, "grad_norm": 14.71875, "learning_rate": 9.334707220011372e-06, "loss": 1.0674, "step": 19224 }, { "epoch": 2.0456480102149395, "grad_norm": 13.3984375, "learning_rate": 9.334671688459353e-06, "loss": 0.6482, "step": 19225 }, { "epoch": 2.045754415833156, "grad_norm": 14.796875, "learning_rate": 9.334636156907335e-06, "loss": 0.9704, "step": 19226 }, { "epoch": 2.0458608214513725, "grad_norm": 10.2421875, "learning_rate": 9.334600625355316e-06, "loss": 0.9159, "step": 19227 }, { "epoch": 2.0459672270695894, "grad_norm": 12.34375, "learning_rate": 9.334565093803298e-06, "loss": 1.2033, "step": 19228 }, { "epoch": 2.046073632687806, "grad_norm": 14.4140625, "learning_rate": 9.33452956225128e-06, "loss": 1.0175, "step": 19229 }, { "epoch": 2.0461800383060225, "grad_norm": 21.0, "learning_rate": 9.334494030699261e-06, "loss": 1.1522, "step": 19230 }, { "epoch": 2.0462864439242394, "grad_norm": 13.8046875, "learning_rate": 9.334458499147243e-06, "loss": 0.9404, "step": 19231 }, { "epoch": 2.046392849542456, "grad_norm": 14.0078125, "learning_rate": 9.334422967595226e-06, "loss": 0.936, "step": 19232 }, { "epoch": 2.0464992551606724, "grad_norm": 17.65625, "learning_rate": 9.334387436043208e-06, "loss": 1.0713, "step": 19233 }, { "epoch": 2.046605660778889, "grad_norm": 12.6640625, "learning_rate": 9.33435190449119e-06, "loss": 0.9461, "step": 19234 }, { "epoch": 2.046712066397106, "grad_norm": 16.53125, "learning_rate": 9.334316372939171e-06, "loss": 0.7608, "step": 19235 }, { "epoch": 2.0468184720153224, "grad_norm": 11.90625, "learning_rate": 9.334280841387152e-06, "loss": 0.7718, "step": 19236 }, { "epoch": 2.046924877633539, "grad_norm": 10.4609375, "learning_rate": 9.334245309835134e-06, "loss": 0.9695, "step": 19237 }, { "epoch": 2.047031283251756, "grad_norm": 14.65625, "learning_rate": 9.334209778283116e-06, "loss": 0.9682, "step": 19238 }, { "epoch": 2.0471376888699724, "grad_norm": 10.6875, "learning_rate": 9.334174246731097e-06, "loss": 0.8783, "step": 19239 }, { "epoch": 2.047244094488189, "grad_norm": 12.140625, "learning_rate": 9.334138715179079e-06, "loss": 0.9152, "step": 19240 }, { "epoch": 2.0473505001064054, "grad_norm": 15.3125, "learning_rate": 9.33410318362706e-06, "loss": 0.9806, "step": 19241 }, { "epoch": 2.0474569057246224, "grad_norm": 14.0390625, "learning_rate": 9.334067652075044e-06, "loss": 0.8802, "step": 19242 }, { "epoch": 2.047563311342839, "grad_norm": 12.6875, "learning_rate": 9.334032120523025e-06, "loss": 0.8337, "step": 19243 }, { "epoch": 2.0476697169610554, "grad_norm": 14.625, "learning_rate": 9.333996588971007e-06, "loss": 1.0512, "step": 19244 }, { "epoch": 2.0477761225792723, "grad_norm": 12.9296875, "learning_rate": 9.333961057418989e-06, "loss": 0.89, "step": 19245 }, { "epoch": 2.047882528197489, "grad_norm": 14.4375, "learning_rate": 9.33392552586697e-06, "loss": 0.8782, "step": 19246 }, { "epoch": 2.0479889338157053, "grad_norm": 14.125, "learning_rate": 9.333889994314952e-06, "loss": 0.9273, "step": 19247 }, { "epoch": 2.0480953394339223, "grad_norm": 10.6171875, "learning_rate": 9.333854462762935e-06, "loss": 0.929, "step": 19248 }, { "epoch": 2.048201745052139, "grad_norm": 12.03125, "learning_rate": 9.333818931210917e-06, "loss": 0.7964, "step": 19249 }, { "epoch": 2.0483081506703553, "grad_norm": 18.078125, "learning_rate": 9.333783399658898e-06, "loss": 0.8869, "step": 19250 }, { "epoch": 2.048414556288572, "grad_norm": 10.2734375, "learning_rate": 9.33374786810688e-06, "loss": 0.6622, "step": 19251 }, { "epoch": 2.048520961906789, "grad_norm": 12.734375, "learning_rate": 9.333712336554862e-06, "loss": 0.9516, "step": 19252 }, { "epoch": 2.0486273675250053, "grad_norm": 13.53125, "learning_rate": 9.333676805002843e-06, "loss": 0.82, "step": 19253 }, { "epoch": 2.048733773143222, "grad_norm": 14.515625, "learning_rate": 9.333641273450825e-06, "loss": 1.0749, "step": 19254 }, { "epoch": 2.0488401787614388, "grad_norm": 10.875, "learning_rate": 9.333605741898806e-06, "loss": 0.9308, "step": 19255 }, { "epoch": 2.0489465843796553, "grad_norm": 16.296875, "learning_rate": 9.333570210346788e-06, "loss": 0.7932, "step": 19256 }, { "epoch": 2.0490529899978718, "grad_norm": 12.8359375, "learning_rate": 9.33353467879477e-06, "loss": 0.9624, "step": 19257 }, { "epoch": 2.0491593956160887, "grad_norm": 14.390625, "learning_rate": 9.333499147242753e-06, "loss": 0.8829, "step": 19258 }, { "epoch": 2.0492658012343052, "grad_norm": 12.140625, "learning_rate": 9.333463615690735e-06, "loss": 0.7527, "step": 19259 }, { "epoch": 2.0493722068525217, "grad_norm": 15.4453125, "learning_rate": 9.333428084138716e-06, "loss": 0.7611, "step": 19260 }, { "epoch": 2.0494786124707383, "grad_norm": 16.21875, "learning_rate": 9.333392552586698e-06, "loss": 0.7853, "step": 19261 }, { "epoch": 2.049585018088955, "grad_norm": 14.125, "learning_rate": 9.33335702103468e-06, "loss": 1.0619, "step": 19262 }, { "epoch": 2.0496914237071717, "grad_norm": 8.9140625, "learning_rate": 9.333321489482663e-06, "loss": 0.7047, "step": 19263 }, { "epoch": 2.0497978293253882, "grad_norm": 19.609375, "learning_rate": 9.333285957930643e-06, "loss": 0.8823, "step": 19264 }, { "epoch": 2.049904234943605, "grad_norm": 13.3046875, "learning_rate": 9.333250426378624e-06, "loss": 1.0566, "step": 19265 }, { "epoch": 2.0500106405618217, "grad_norm": 11.9296875, "learning_rate": 9.333214894826606e-06, "loss": 0.9796, "step": 19266 }, { "epoch": 2.050117046180038, "grad_norm": 14.3359375, "learning_rate": 9.333179363274587e-06, "loss": 0.9797, "step": 19267 }, { "epoch": 2.050223451798255, "grad_norm": 11.9140625, "learning_rate": 9.33314383172257e-06, "loss": 0.7457, "step": 19268 }, { "epoch": 2.0503298574164717, "grad_norm": 11.484375, "learning_rate": 9.333108300170552e-06, "loss": 0.7614, "step": 19269 }, { "epoch": 2.050436263034688, "grad_norm": 11.484375, "learning_rate": 9.333072768618534e-06, "loss": 0.806, "step": 19270 }, { "epoch": 2.0505426686529047, "grad_norm": 10.1484375, "learning_rate": 9.333037237066516e-06, "loss": 0.8854, "step": 19271 }, { "epoch": 2.0506490742711216, "grad_norm": 10.140625, "learning_rate": 9.333001705514497e-06, "loss": 0.8664, "step": 19272 }, { "epoch": 2.050755479889338, "grad_norm": 13.3828125, "learning_rate": 9.33296617396248e-06, "loss": 1.0155, "step": 19273 }, { "epoch": 2.0508618855075547, "grad_norm": 24.3125, "learning_rate": 9.332930642410462e-06, "loss": 1.1571, "step": 19274 }, { "epoch": 2.0509682911257716, "grad_norm": 12.375, "learning_rate": 9.332895110858444e-06, "loss": 0.9105, "step": 19275 }, { "epoch": 2.051074696743988, "grad_norm": 12.9609375, "learning_rate": 9.332859579306425e-06, "loss": 0.8084, "step": 19276 }, { "epoch": 2.0511811023622046, "grad_norm": 14.859375, "learning_rate": 9.332824047754407e-06, "loss": 0.8564, "step": 19277 }, { "epoch": 2.0512875079804216, "grad_norm": 14.3671875, "learning_rate": 9.332788516202388e-06, "loss": 0.8369, "step": 19278 }, { "epoch": 2.051393913598638, "grad_norm": 10.6640625, "learning_rate": 9.33275298465037e-06, "loss": 0.7463, "step": 19279 }, { "epoch": 2.0515003192168546, "grad_norm": 12.1953125, "learning_rate": 9.332717453098352e-06, "loss": 1.042, "step": 19280 }, { "epoch": 2.051606724835071, "grad_norm": 15.765625, "learning_rate": 9.332681921546333e-06, "loss": 0.9474, "step": 19281 }, { "epoch": 2.051713130453288, "grad_norm": 8.96875, "learning_rate": 9.332646389994315e-06, "loss": 0.875, "step": 19282 }, { "epoch": 2.0518195360715046, "grad_norm": 9.921875, "learning_rate": 9.332610858442296e-06, "loss": 0.7853, "step": 19283 }, { "epoch": 2.051925941689721, "grad_norm": 12.078125, "learning_rate": 9.33257532689028e-06, "loss": 0.8663, "step": 19284 }, { "epoch": 2.052032347307938, "grad_norm": 16.625, "learning_rate": 9.332539795338261e-06, "loss": 0.8953, "step": 19285 }, { "epoch": 2.0521387529261546, "grad_norm": 15.0859375, "learning_rate": 9.332504263786243e-06, "loss": 0.9704, "step": 19286 }, { "epoch": 2.052245158544371, "grad_norm": 13.1015625, "learning_rate": 9.332468732234225e-06, "loss": 0.9287, "step": 19287 }, { "epoch": 2.0523515641625876, "grad_norm": 13.1015625, "learning_rate": 9.332433200682206e-06, "loss": 0.9255, "step": 19288 }, { "epoch": 2.0524579697808045, "grad_norm": 9.8671875, "learning_rate": 9.33239766913019e-06, "loss": 0.7248, "step": 19289 }, { "epoch": 2.052564375399021, "grad_norm": 11.6640625, "learning_rate": 9.332362137578171e-06, "loss": 0.9966, "step": 19290 }, { "epoch": 2.0526707810172375, "grad_norm": 11.4140625, "learning_rate": 9.332326606026153e-06, "loss": 0.8364, "step": 19291 }, { "epoch": 2.0527771866354545, "grad_norm": 12.9609375, "learning_rate": 9.332291074474134e-06, "loss": 0.9218, "step": 19292 }, { "epoch": 2.052883592253671, "grad_norm": 11.3671875, "learning_rate": 9.332255542922116e-06, "loss": 0.9183, "step": 19293 }, { "epoch": 2.0529899978718875, "grad_norm": 10.078125, "learning_rate": 9.332220011370098e-06, "loss": 0.8734, "step": 19294 }, { "epoch": 2.0530964034901045, "grad_norm": 9.546875, "learning_rate": 9.332184479818079e-06, "loss": 0.6742, "step": 19295 }, { "epoch": 2.053202809108321, "grad_norm": 15.25, "learning_rate": 9.33214894826606e-06, "loss": 0.918, "step": 19296 }, { "epoch": 2.0533092147265375, "grad_norm": 16.640625, "learning_rate": 9.332113416714042e-06, "loss": 0.7668, "step": 19297 }, { "epoch": 2.053415620344754, "grad_norm": 17.0625, "learning_rate": 9.332077885162024e-06, "loss": 0.8933, "step": 19298 }, { "epoch": 2.053522025962971, "grad_norm": 12.71875, "learning_rate": 9.332042353610007e-06, "loss": 0.928, "step": 19299 }, { "epoch": 2.0536284315811875, "grad_norm": 10.34375, "learning_rate": 9.332006822057989e-06, "loss": 0.7256, "step": 19300 }, { "epoch": 2.053734837199404, "grad_norm": 13.3984375, "learning_rate": 9.33197129050597e-06, "loss": 0.7795, "step": 19301 }, { "epoch": 2.053841242817621, "grad_norm": 15.0625, "learning_rate": 9.331935758953952e-06, "loss": 0.9596, "step": 19302 }, { "epoch": 2.0539476484358374, "grad_norm": 9.7109375, "learning_rate": 9.331900227401934e-06, "loss": 0.6835, "step": 19303 }, { "epoch": 2.054054054054054, "grad_norm": 9.1328125, "learning_rate": 9.331864695849915e-06, "loss": 0.7294, "step": 19304 }, { "epoch": 2.054160459672271, "grad_norm": 14.25, "learning_rate": 9.331829164297897e-06, "loss": 0.9281, "step": 19305 }, { "epoch": 2.0542668652904874, "grad_norm": 13.2890625, "learning_rate": 9.331793632745879e-06, "loss": 0.8298, "step": 19306 }, { "epoch": 2.054373270908704, "grad_norm": 11.6796875, "learning_rate": 9.33175810119386e-06, "loss": 0.8346, "step": 19307 }, { "epoch": 2.0544796765269204, "grad_norm": 12.8515625, "learning_rate": 9.331722569641842e-06, "loss": 0.8781, "step": 19308 }, { "epoch": 2.0545860821451374, "grad_norm": 11.5390625, "learning_rate": 9.331687038089825e-06, "loss": 0.9446, "step": 19309 }, { "epoch": 2.054692487763354, "grad_norm": 15.4609375, "learning_rate": 9.331651506537807e-06, "loss": 0.9289, "step": 19310 }, { "epoch": 2.0547988933815704, "grad_norm": 11.2890625, "learning_rate": 9.331615974985788e-06, "loss": 0.8849, "step": 19311 }, { "epoch": 2.0549052989997874, "grad_norm": 13.5, "learning_rate": 9.33158044343377e-06, "loss": 0.9834, "step": 19312 }, { "epoch": 2.055011704618004, "grad_norm": 12.3046875, "learning_rate": 9.331544911881751e-06, "loss": 0.908, "step": 19313 }, { "epoch": 2.0551181102362204, "grad_norm": 12.71875, "learning_rate": 9.331509380329735e-06, "loss": 0.8328, "step": 19314 }, { "epoch": 2.0552245158544373, "grad_norm": 9.984375, "learning_rate": 9.331473848777716e-06, "loss": 0.7824, "step": 19315 }, { "epoch": 2.055330921472654, "grad_norm": 12.21875, "learning_rate": 9.331438317225698e-06, "loss": 0.8646, "step": 19316 }, { "epoch": 2.0554373270908703, "grad_norm": 16.90625, "learning_rate": 9.33140278567368e-06, "loss": 0.9082, "step": 19317 }, { "epoch": 2.055543732709087, "grad_norm": 17.421875, "learning_rate": 9.331367254121661e-06, "loss": 0.879, "step": 19318 }, { "epoch": 2.055650138327304, "grad_norm": 11.8359375, "learning_rate": 9.331331722569643e-06, "loss": 0.7935, "step": 19319 }, { "epoch": 2.0557565439455203, "grad_norm": 9.578125, "learning_rate": 9.331296191017624e-06, "loss": 0.7406, "step": 19320 }, { "epoch": 2.055862949563737, "grad_norm": 16.765625, "learning_rate": 9.331260659465606e-06, "loss": 1.082, "step": 19321 }, { "epoch": 2.055969355181954, "grad_norm": 9.171875, "learning_rate": 9.331225127913588e-06, "loss": 0.7684, "step": 19322 }, { "epoch": 2.0560757608001703, "grad_norm": 13.3203125, "learning_rate": 9.33118959636157e-06, "loss": 0.9938, "step": 19323 }, { "epoch": 2.056182166418387, "grad_norm": 11.6875, "learning_rate": 9.33115406480955e-06, "loss": 0.839, "step": 19324 }, { "epoch": 2.0562885720366033, "grad_norm": 12.3828125, "learning_rate": 9.331118533257534e-06, "loss": 0.8288, "step": 19325 }, { "epoch": 2.0563949776548203, "grad_norm": 16.21875, "learning_rate": 9.331083001705516e-06, "loss": 0.8517, "step": 19326 }, { "epoch": 2.0565013832730368, "grad_norm": 18.265625, "learning_rate": 9.331047470153497e-06, "loss": 0.9544, "step": 19327 }, { "epoch": 2.0566077888912533, "grad_norm": 29.453125, "learning_rate": 9.331011938601479e-06, "loss": 0.948, "step": 19328 }, { "epoch": 2.0567141945094702, "grad_norm": 15.40625, "learning_rate": 9.33097640704946e-06, "loss": 0.9637, "step": 19329 }, { "epoch": 2.0568206001276867, "grad_norm": 15.53125, "learning_rate": 9.330940875497442e-06, "loss": 0.9246, "step": 19330 }, { "epoch": 2.0569270057459033, "grad_norm": 12.5859375, "learning_rate": 9.330905343945424e-06, "loss": 1.0085, "step": 19331 }, { "epoch": 2.05703341136412, "grad_norm": 10.46875, "learning_rate": 9.330869812393405e-06, "loss": 0.8169, "step": 19332 }, { "epoch": 2.0571398169823367, "grad_norm": 13.625, "learning_rate": 9.330834280841387e-06, "loss": 0.9775, "step": 19333 }, { "epoch": 2.0572462226005532, "grad_norm": 13.1953125, "learning_rate": 9.330798749289369e-06, "loss": 0.895, "step": 19334 }, { "epoch": 2.0573526282187697, "grad_norm": 21.765625, "learning_rate": 9.330763217737352e-06, "loss": 1.0983, "step": 19335 }, { "epoch": 2.0574590338369867, "grad_norm": 17.59375, "learning_rate": 9.330727686185333e-06, "loss": 0.8821, "step": 19336 }, { "epoch": 2.057565439455203, "grad_norm": 11.2109375, "learning_rate": 9.330692154633315e-06, "loss": 0.8278, "step": 19337 }, { "epoch": 2.0576718450734197, "grad_norm": 9.40625, "learning_rate": 9.330656623081297e-06, "loss": 0.7829, "step": 19338 }, { "epoch": 2.0577782506916367, "grad_norm": 18.296875, "learning_rate": 9.330621091529278e-06, "loss": 0.8338, "step": 19339 }, { "epoch": 2.057884656309853, "grad_norm": 11.171875, "learning_rate": 9.330585559977262e-06, "loss": 0.8478, "step": 19340 }, { "epoch": 2.0579910619280697, "grad_norm": 13.84375, "learning_rate": 9.330550028425243e-06, "loss": 0.9077, "step": 19341 }, { "epoch": 2.0580974675462866, "grad_norm": 14.203125, "learning_rate": 9.330514496873225e-06, "loss": 1.097, "step": 19342 }, { "epoch": 2.058203873164503, "grad_norm": 9.4453125, "learning_rate": 9.330478965321206e-06, "loss": 0.8908, "step": 19343 }, { "epoch": 2.0583102787827197, "grad_norm": 11.640625, "learning_rate": 9.330443433769188e-06, "loss": 1.036, "step": 19344 }, { "epoch": 2.058416684400936, "grad_norm": 11.3515625, "learning_rate": 9.33040790221717e-06, "loss": 0.7969, "step": 19345 }, { "epoch": 2.058523090019153, "grad_norm": 15.0234375, "learning_rate": 9.330372370665151e-06, "loss": 0.9029, "step": 19346 }, { "epoch": 2.0586294956373696, "grad_norm": 12.4296875, "learning_rate": 9.330336839113133e-06, "loss": 0.7554, "step": 19347 }, { "epoch": 2.058735901255586, "grad_norm": 11.9765625, "learning_rate": 9.330301307561114e-06, "loss": 1.0339, "step": 19348 }, { "epoch": 2.058842306873803, "grad_norm": 11.9140625, "learning_rate": 9.330265776009096e-06, "loss": 0.6865, "step": 19349 }, { "epoch": 2.0589487124920196, "grad_norm": 29.203125, "learning_rate": 9.33023024445708e-06, "loss": 0.869, "step": 19350 }, { "epoch": 2.059055118110236, "grad_norm": 11.1796875, "learning_rate": 9.330194712905061e-06, "loss": 0.7707, "step": 19351 }, { "epoch": 2.059161523728453, "grad_norm": 11.7890625, "learning_rate": 9.330159181353043e-06, "loss": 0.8972, "step": 19352 }, { "epoch": 2.0592679293466696, "grad_norm": 13.421875, "learning_rate": 9.330123649801024e-06, "loss": 0.9166, "step": 19353 }, { "epoch": 2.059374334964886, "grad_norm": 11.1953125, "learning_rate": 9.330088118249006e-06, "loss": 0.9867, "step": 19354 }, { "epoch": 2.0594807405831026, "grad_norm": 9.6640625, "learning_rate": 9.330052586696987e-06, "loss": 0.6974, "step": 19355 }, { "epoch": 2.0595871462013196, "grad_norm": 12.7734375, "learning_rate": 9.330017055144969e-06, "loss": 1.0063, "step": 19356 }, { "epoch": 2.059693551819536, "grad_norm": 16.359375, "learning_rate": 9.32998152359295e-06, "loss": 0.8831, "step": 19357 }, { "epoch": 2.0597999574377526, "grad_norm": 14.859375, "learning_rate": 9.329945992040932e-06, "loss": 0.9785, "step": 19358 }, { "epoch": 2.0599063630559695, "grad_norm": 12.5078125, "learning_rate": 9.329910460488914e-06, "loss": 0.8453, "step": 19359 }, { "epoch": 2.060012768674186, "grad_norm": 15.1015625, "learning_rate": 9.329874928936897e-06, "loss": 1.0611, "step": 19360 }, { "epoch": 2.0601191742924025, "grad_norm": 13.96875, "learning_rate": 9.329839397384879e-06, "loss": 0.9459, "step": 19361 }, { "epoch": 2.0602255799106195, "grad_norm": 15.015625, "learning_rate": 9.32980386583286e-06, "loss": 0.8184, "step": 19362 }, { "epoch": 2.060331985528836, "grad_norm": 10.7421875, "learning_rate": 9.329768334280842e-06, "loss": 0.7858, "step": 19363 }, { "epoch": 2.0604383911470525, "grad_norm": 11.2890625, "learning_rate": 9.329732802728824e-06, "loss": 0.9543, "step": 19364 }, { "epoch": 2.060544796765269, "grad_norm": 11.21875, "learning_rate": 9.329697271176805e-06, "loss": 0.7887, "step": 19365 }, { "epoch": 2.060651202383486, "grad_norm": 16.265625, "learning_rate": 9.329661739624788e-06, "loss": 0.8934, "step": 19366 }, { "epoch": 2.0607576080017025, "grad_norm": 11.4765625, "learning_rate": 9.32962620807277e-06, "loss": 0.9119, "step": 19367 }, { "epoch": 2.060864013619919, "grad_norm": 20.75, "learning_rate": 9.329590676520752e-06, "loss": 1.0752, "step": 19368 }, { "epoch": 2.060970419238136, "grad_norm": 10.7265625, "learning_rate": 9.329555144968733e-06, "loss": 0.8776, "step": 19369 }, { "epoch": 2.0610768248563525, "grad_norm": 12.4765625, "learning_rate": 9.329519613416715e-06, "loss": 0.932, "step": 19370 }, { "epoch": 2.061183230474569, "grad_norm": 10.9375, "learning_rate": 9.329484081864697e-06, "loss": 0.8753, "step": 19371 }, { "epoch": 2.0612896360927855, "grad_norm": 10.9765625, "learning_rate": 9.329448550312678e-06, "loss": 0.9614, "step": 19372 }, { "epoch": 2.0613960417110024, "grad_norm": 11.484375, "learning_rate": 9.32941301876066e-06, "loss": 0.8764, "step": 19373 }, { "epoch": 2.061502447329219, "grad_norm": 13.9296875, "learning_rate": 9.329377487208641e-06, "loss": 0.7786, "step": 19374 }, { "epoch": 2.0616088529474355, "grad_norm": 13.8671875, "learning_rate": 9.329341955656623e-06, "loss": 0.9267, "step": 19375 }, { "epoch": 2.0617152585656524, "grad_norm": 10.2578125, "learning_rate": 9.329306424104606e-06, "loss": 0.7984, "step": 19376 }, { "epoch": 2.061821664183869, "grad_norm": 13.5546875, "learning_rate": 9.329270892552588e-06, "loss": 0.8858, "step": 19377 }, { "epoch": 2.0619280698020854, "grad_norm": 13.234375, "learning_rate": 9.32923536100057e-06, "loss": 0.9988, "step": 19378 }, { "epoch": 2.0620344754203024, "grad_norm": 10.7109375, "learning_rate": 9.329199829448551e-06, "loss": 0.8175, "step": 19379 }, { "epoch": 2.062140881038519, "grad_norm": 18.6875, "learning_rate": 9.329164297896533e-06, "loss": 0.8978, "step": 19380 }, { "epoch": 2.0622472866567354, "grad_norm": 13.328125, "learning_rate": 9.329128766344516e-06, "loss": 0.8983, "step": 19381 }, { "epoch": 2.062353692274952, "grad_norm": 16.484375, "learning_rate": 9.329093234792498e-06, "loss": 0.8716, "step": 19382 }, { "epoch": 2.062460097893169, "grad_norm": 14.4296875, "learning_rate": 9.32905770324048e-06, "loss": 0.9968, "step": 19383 }, { "epoch": 2.0625665035113854, "grad_norm": 11.8125, "learning_rate": 9.329022171688459e-06, "loss": 0.9566, "step": 19384 }, { "epoch": 2.062672909129602, "grad_norm": 8.890625, "learning_rate": 9.32898664013644e-06, "loss": 0.6882, "step": 19385 }, { "epoch": 2.062779314747819, "grad_norm": 13.0, "learning_rate": 9.328951108584424e-06, "loss": 1.1406, "step": 19386 }, { "epoch": 2.0628857203660353, "grad_norm": 16.234375, "learning_rate": 9.328915577032406e-06, "loss": 0.9765, "step": 19387 }, { "epoch": 2.062992125984252, "grad_norm": 14.8046875, "learning_rate": 9.328880045480387e-06, "loss": 0.871, "step": 19388 }, { "epoch": 2.063098531602469, "grad_norm": 13.7421875, "learning_rate": 9.328844513928369e-06, "loss": 0.7848, "step": 19389 }, { "epoch": 2.0632049372206853, "grad_norm": 14.046875, "learning_rate": 9.32880898237635e-06, "loss": 0.8578, "step": 19390 }, { "epoch": 2.063311342838902, "grad_norm": 7.74609375, "learning_rate": 9.328773450824334e-06, "loss": 0.7694, "step": 19391 }, { "epoch": 2.0634177484571183, "grad_norm": 15.5546875, "learning_rate": 9.328737919272315e-06, "loss": 0.6953, "step": 19392 }, { "epoch": 2.0635241540753353, "grad_norm": 10.3359375, "learning_rate": 9.328702387720297e-06, "loss": 0.8969, "step": 19393 }, { "epoch": 2.063630559693552, "grad_norm": 9.5234375, "learning_rate": 9.328666856168279e-06, "loss": 0.863, "step": 19394 }, { "epoch": 2.0637369653117683, "grad_norm": 11.046875, "learning_rate": 9.32863132461626e-06, "loss": 0.7145, "step": 19395 }, { "epoch": 2.0638433709299853, "grad_norm": 15.375, "learning_rate": 9.328595793064242e-06, "loss": 1.0508, "step": 19396 }, { "epoch": 2.0639497765482018, "grad_norm": 19.5, "learning_rate": 9.328560261512223e-06, "loss": 1.1005, "step": 19397 }, { "epoch": 2.0640561821664183, "grad_norm": 13.5859375, "learning_rate": 9.328524729960205e-06, "loss": 0.9086, "step": 19398 }, { "epoch": 2.0641625877846352, "grad_norm": 10.21875, "learning_rate": 9.328489198408187e-06, "loss": 0.812, "step": 19399 }, { "epoch": 2.0642689934028517, "grad_norm": 13.921875, "learning_rate": 9.328453666856168e-06, "loss": 0.6907, "step": 19400 }, { "epoch": 2.0643753990210683, "grad_norm": 15.078125, "learning_rate": 9.32841813530415e-06, "loss": 1.1113, "step": 19401 }, { "epoch": 2.0644818046392848, "grad_norm": 10.0625, "learning_rate": 9.328382603752133e-06, "loss": 0.8464, "step": 19402 }, { "epoch": 2.0645882102575017, "grad_norm": 12.6796875, "learning_rate": 9.328347072200115e-06, "loss": 0.7173, "step": 19403 }, { "epoch": 2.0646946158757182, "grad_norm": 14.0703125, "learning_rate": 9.328311540648096e-06, "loss": 0.9555, "step": 19404 }, { "epoch": 2.0648010214939347, "grad_norm": 16.203125, "learning_rate": 9.328276009096078e-06, "loss": 0.9936, "step": 19405 }, { "epoch": 2.0649074271121517, "grad_norm": 14.921875, "learning_rate": 9.32824047754406e-06, "loss": 0.9519, "step": 19406 }, { "epoch": 2.065013832730368, "grad_norm": 11.765625, "learning_rate": 9.328204945992043e-06, "loss": 0.9974, "step": 19407 }, { "epoch": 2.0651202383485847, "grad_norm": 12.71875, "learning_rate": 9.328169414440024e-06, "loss": 0.8789, "step": 19408 }, { "epoch": 2.0652266439668017, "grad_norm": 12.6484375, "learning_rate": 9.328133882888006e-06, "loss": 0.8828, "step": 19409 }, { "epoch": 2.065333049585018, "grad_norm": 14.3125, "learning_rate": 9.328098351335988e-06, "loss": 0.8851, "step": 19410 }, { "epoch": 2.0654394552032347, "grad_norm": 10.8515625, "learning_rate": 9.32806281978397e-06, "loss": 0.7953, "step": 19411 }, { "epoch": 2.065545860821451, "grad_norm": 13.46875, "learning_rate": 9.32802728823195e-06, "loss": 0.857, "step": 19412 }, { "epoch": 2.065652266439668, "grad_norm": 11.8125, "learning_rate": 9.327991756679932e-06, "loss": 0.8241, "step": 19413 }, { "epoch": 2.0657586720578847, "grad_norm": 14.65625, "learning_rate": 9.327956225127914e-06, "loss": 0.9993, "step": 19414 }, { "epoch": 2.065865077676101, "grad_norm": 15.5625, "learning_rate": 9.327920693575896e-06, "loss": 0.8203, "step": 19415 }, { "epoch": 2.065971483294318, "grad_norm": 12.265625, "learning_rate": 9.327885162023877e-06, "loss": 0.8232, "step": 19416 }, { "epoch": 2.0660778889125346, "grad_norm": 10.578125, "learning_rate": 9.32784963047186e-06, "loss": 0.8148, "step": 19417 }, { "epoch": 2.066184294530751, "grad_norm": 13.4375, "learning_rate": 9.327814098919842e-06, "loss": 0.833, "step": 19418 }, { "epoch": 2.0662907001489677, "grad_norm": 15.1953125, "learning_rate": 9.327778567367824e-06, "loss": 0.6923, "step": 19419 }, { "epoch": 2.0663971057671846, "grad_norm": 16.96875, "learning_rate": 9.327743035815805e-06, "loss": 1.1445, "step": 19420 }, { "epoch": 2.066503511385401, "grad_norm": 12.8515625, "learning_rate": 9.327707504263787e-06, "loss": 0.9003, "step": 19421 }, { "epoch": 2.0666099170036176, "grad_norm": 14.21875, "learning_rate": 9.327671972711769e-06, "loss": 1.0954, "step": 19422 }, { "epoch": 2.0667163226218346, "grad_norm": 9.859375, "learning_rate": 9.32763644115975e-06, "loss": 0.6748, "step": 19423 }, { "epoch": 2.066822728240051, "grad_norm": 16.703125, "learning_rate": 9.327600909607732e-06, "loss": 1.0166, "step": 19424 }, { "epoch": 2.0669291338582676, "grad_norm": 12.2109375, "learning_rate": 9.327565378055713e-06, "loss": 0.9366, "step": 19425 }, { "epoch": 2.0670355394764846, "grad_norm": 14.28125, "learning_rate": 9.327529846503695e-06, "loss": 0.9695, "step": 19426 }, { "epoch": 2.067141945094701, "grad_norm": 15.171875, "learning_rate": 9.327494314951678e-06, "loss": 1.0144, "step": 19427 }, { "epoch": 2.0672483507129176, "grad_norm": 14.4375, "learning_rate": 9.32745878339966e-06, "loss": 0.8264, "step": 19428 }, { "epoch": 2.067354756331134, "grad_norm": 9.375, "learning_rate": 9.327423251847642e-06, "loss": 0.8904, "step": 19429 }, { "epoch": 2.067461161949351, "grad_norm": 11.40625, "learning_rate": 9.327387720295623e-06, "loss": 0.8898, "step": 19430 }, { "epoch": 2.0675675675675675, "grad_norm": 14.8359375, "learning_rate": 9.327352188743605e-06, "loss": 1.0126, "step": 19431 }, { "epoch": 2.067673973185784, "grad_norm": 12.0, "learning_rate": 9.327316657191588e-06, "loss": 0.8724, "step": 19432 }, { "epoch": 2.067780378804001, "grad_norm": 11.7265625, "learning_rate": 9.32728112563957e-06, "loss": 0.8488, "step": 19433 }, { "epoch": 2.0678867844222175, "grad_norm": 14.53125, "learning_rate": 9.327245594087551e-06, "loss": 0.8328, "step": 19434 }, { "epoch": 2.067993190040434, "grad_norm": 10.75, "learning_rate": 9.327210062535533e-06, "loss": 0.862, "step": 19435 }, { "epoch": 2.068099595658651, "grad_norm": 12.5703125, "learning_rate": 9.327174530983514e-06, "loss": 0.8925, "step": 19436 }, { "epoch": 2.0682060012768675, "grad_norm": 13.65625, "learning_rate": 9.327138999431496e-06, "loss": 0.9185, "step": 19437 }, { "epoch": 2.068312406895084, "grad_norm": 9.4453125, "learning_rate": 9.327103467879478e-06, "loss": 0.7717, "step": 19438 }, { "epoch": 2.0684188125133005, "grad_norm": 14.4375, "learning_rate": 9.32706793632746e-06, "loss": 0.9566, "step": 19439 }, { "epoch": 2.0685252181315175, "grad_norm": 13.28125, "learning_rate": 9.327032404775441e-06, "loss": 1.0159, "step": 19440 }, { "epoch": 2.068631623749734, "grad_norm": 9.734375, "learning_rate": 9.326996873223423e-06, "loss": 0.8197, "step": 19441 }, { "epoch": 2.0687380293679505, "grad_norm": 12.703125, "learning_rate": 9.326961341671404e-06, "loss": 0.985, "step": 19442 }, { "epoch": 2.0688444349861674, "grad_norm": 16.640625, "learning_rate": 9.326925810119387e-06, "loss": 1.0667, "step": 19443 }, { "epoch": 2.068950840604384, "grad_norm": 11.1640625, "learning_rate": 9.326890278567369e-06, "loss": 0.8012, "step": 19444 }, { "epoch": 2.0690572462226005, "grad_norm": 12.9140625, "learning_rate": 9.32685474701535e-06, "loss": 0.7686, "step": 19445 }, { "epoch": 2.0691636518408174, "grad_norm": 14.640625, "learning_rate": 9.326819215463332e-06, "loss": 0.9166, "step": 19446 }, { "epoch": 2.069270057459034, "grad_norm": 12.6875, "learning_rate": 9.326783683911314e-06, "loss": 0.8746, "step": 19447 }, { "epoch": 2.0693764630772504, "grad_norm": 13.4609375, "learning_rate": 9.326748152359295e-06, "loss": 0.9802, "step": 19448 }, { "epoch": 2.069482868695467, "grad_norm": 12.6328125, "learning_rate": 9.326712620807277e-06, "loss": 0.8852, "step": 19449 }, { "epoch": 2.069589274313684, "grad_norm": 14.359375, "learning_rate": 9.326677089255259e-06, "loss": 1.1311, "step": 19450 }, { "epoch": 2.0696956799319004, "grad_norm": 10.6484375, "learning_rate": 9.32664155770324e-06, "loss": 0.7359, "step": 19451 }, { "epoch": 2.069802085550117, "grad_norm": 14.984375, "learning_rate": 9.326606026151222e-06, "loss": 1.1547, "step": 19452 }, { "epoch": 2.069908491168334, "grad_norm": 14.0, "learning_rate": 9.326570494599205e-06, "loss": 0.9684, "step": 19453 }, { "epoch": 2.0700148967865504, "grad_norm": 13.0234375, "learning_rate": 9.326534963047187e-06, "loss": 0.8652, "step": 19454 }, { "epoch": 2.070121302404767, "grad_norm": 12.15625, "learning_rate": 9.326499431495168e-06, "loss": 0.7401, "step": 19455 }, { "epoch": 2.0702277080229834, "grad_norm": 13.96875, "learning_rate": 9.32646389994315e-06, "loss": 0.9203, "step": 19456 }, { "epoch": 2.0703341136412003, "grad_norm": 12.359375, "learning_rate": 9.326428368391132e-06, "loss": 0.8399, "step": 19457 }, { "epoch": 2.070440519259417, "grad_norm": 11.265625, "learning_rate": 9.326392836839115e-06, "loss": 0.9924, "step": 19458 }, { "epoch": 2.0705469248776334, "grad_norm": 12.265625, "learning_rate": 9.326357305287097e-06, "loss": 0.9578, "step": 19459 }, { "epoch": 2.0706533304958503, "grad_norm": 15.078125, "learning_rate": 9.326321773735078e-06, "loss": 1.0498, "step": 19460 }, { "epoch": 2.070759736114067, "grad_norm": 10.9609375, "learning_rate": 9.32628624218306e-06, "loss": 0.8522, "step": 19461 }, { "epoch": 2.0708661417322833, "grad_norm": 10.1484375, "learning_rate": 9.326250710631041e-06, "loss": 0.8045, "step": 19462 }, { "epoch": 2.0709725473505003, "grad_norm": 234.375, "learning_rate": 9.326215179079023e-06, "loss": 1.3905, "step": 19463 }, { "epoch": 2.071078952968717, "grad_norm": 11.984375, "learning_rate": 9.326179647527005e-06, "loss": 0.77, "step": 19464 }, { "epoch": 2.0711853585869333, "grad_norm": 14.328125, "learning_rate": 9.326144115974986e-06, "loss": 0.8364, "step": 19465 }, { "epoch": 2.07129176420515, "grad_norm": 11.703125, "learning_rate": 9.326108584422968e-06, "loss": 0.975, "step": 19466 }, { "epoch": 2.0713981698233668, "grad_norm": 15.515625, "learning_rate": 9.32607305287095e-06, "loss": 0.9912, "step": 19467 }, { "epoch": 2.0715045754415833, "grad_norm": 10.46875, "learning_rate": 9.326037521318933e-06, "loss": 0.6889, "step": 19468 }, { "epoch": 2.0716109810598, "grad_norm": 9.6796875, "learning_rate": 9.326001989766914e-06, "loss": 0.9017, "step": 19469 }, { "epoch": 2.0717173866780167, "grad_norm": 12.7890625, "learning_rate": 9.325966458214896e-06, "loss": 0.9502, "step": 19470 }, { "epoch": 2.0718237922962333, "grad_norm": 13.2734375, "learning_rate": 9.325930926662877e-06, "loss": 1.0264, "step": 19471 }, { "epoch": 2.0719301979144498, "grad_norm": 13.0234375, "learning_rate": 9.325895395110859e-06, "loss": 0.765, "step": 19472 }, { "epoch": 2.0720366035326667, "grad_norm": 14.3984375, "learning_rate": 9.32585986355884e-06, "loss": 1.0782, "step": 19473 }, { "epoch": 2.0721430091508832, "grad_norm": 13.9140625, "learning_rate": 9.325824332006822e-06, "loss": 0.8133, "step": 19474 }, { "epoch": 2.0722494147690997, "grad_norm": 11.3359375, "learning_rate": 9.325788800454804e-06, "loss": 0.752, "step": 19475 }, { "epoch": 2.0723558203873162, "grad_norm": 18.90625, "learning_rate": 9.325753268902786e-06, "loss": 0.9563, "step": 19476 }, { "epoch": 2.072462226005533, "grad_norm": 12.6796875, "learning_rate": 9.325717737350767e-06, "loss": 0.872, "step": 19477 }, { "epoch": 2.0725686316237497, "grad_norm": 14.109375, "learning_rate": 9.325682205798749e-06, "loss": 0.8014, "step": 19478 }, { "epoch": 2.0726750372419662, "grad_norm": 15.859375, "learning_rate": 9.325646674246732e-06, "loss": 1.0016, "step": 19479 }, { "epoch": 2.072781442860183, "grad_norm": 14.5703125, "learning_rate": 9.325611142694714e-06, "loss": 1.0516, "step": 19480 }, { "epoch": 2.0728878484783997, "grad_norm": 12.6640625, "learning_rate": 9.325575611142695e-06, "loss": 0.8142, "step": 19481 }, { "epoch": 2.072994254096616, "grad_norm": 10.65625, "learning_rate": 9.325540079590677e-06, "loss": 0.8912, "step": 19482 }, { "epoch": 2.073100659714833, "grad_norm": 11.1015625, "learning_rate": 9.325504548038658e-06, "loss": 0.7109, "step": 19483 }, { "epoch": 2.0732070653330497, "grad_norm": 14.5234375, "learning_rate": 9.325469016486642e-06, "loss": 0.8665, "step": 19484 }, { "epoch": 2.073313470951266, "grad_norm": 12.0078125, "learning_rate": 9.325433484934623e-06, "loss": 0.7878, "step": 19485 }, { "epoch": 2.0734198765694827, "grad_norm": 10.3828125, "learning_rate": 9.325397953382605e-06, "loss": 0.7542, "step": 19486 }, { "epoch": 2.0735262821876996, "grad_norm": 15.265625, "learning_rate": 9.325362421830587e-06, "loss": 0.8457, "step": 19487 }, { "epoch": 2.073632687805916, "grad_norm": 15.5078125, "learning_rate": 9.325326890278568e-06, "loss": 1.1989, "step": 19488 }, { "epoch": 2.0737390934241327, "grad_norm": 14.3203125, "learning_rate": 9.32529135872655e-06, "loss": 1.0345, "step": 19489 }, { "epoch": 2.0738454990423496, "grad_norm": 12.6171875, "learning_rate": 9.325255827174531e-06, "loss": 1.1025, "step": 19490 }, { "epoch": 2.073951904660566, "grad_norm": 11.4765625, "learning_rate": 9.325220295622513e-06, "loss": 0.9377, "step": 19491 }, { "epoch": 2.0740583102787826, "grad_norm": 12.8359375, "learning_rate": 9.325184764070495e-06, "loss": 0.9175, "step": 19492 }, { "epoch": 2.074164715896999, "grad_norm": 16.375, "learning_rate": 9.325149232518476e-06, "loss": 0.89, "step": 19493 }, { "epoch": 2.074271121515216, "grad_norm": 13.0078125, "learning_rate": 9.32511370096646e-06, "loss": 0.8089, "step": 19494 }, { "epoch": 2.0743775271334326, "grad_norm": 20.9375, "learning_rate": 9.325078169414441e-06, "loss": 0.9155, "step": 19495 }, { "epoch": 2.074483932751649, "grad_norm": 17.796875, "learning_rate": 9.325042637862423e-06, "loss": 0.9203, "step": 19496 }, { "epoch": 2.074590338369866, "grad_norm": 10.546875, "learning_rate": 9.325007106310404e-06, "loss": 0.7989, "step": 19497 }, { "epoch": 2.0746967439880826, "grad_norm": 11.7109375, "learning_rate": 9.324971574758386e-06, "loss": 0.8503, "step": 19498 }, { "epoch": 2.074803149606299, "grad_norm": 11.4921875, "learning_rate": 9.32493604320637e-06, "loss": 0.874, "step": 19499 }, { "epoch": 2.074909555224516, "grad_norm": 13.5625, "learning_rate": 9.324900511654351e-06, "loss": 0.9356, "step": 19500 }, { "epoch": 2.0750159608427325, "grad_norm": 10.3203125, "learning_rate": 9.324864980102332e-06, "loss": 0.9395, "step": 19501 }, { "epoch": 2.075122366460949, "grad_norm": 19.109375, "learning_rate": 9.324829448550314e-06, "loss": 1.0454, "step": 19502 }, { "epoch": 2.075228772079166, "grad_norm": 12.3359375, "learning_rate": 9.324793916998294e-06, "loss": 0.8749, "step": 19503 }, { "epoch": 2.0753351776973825, "grad_norm": 16.625, "learning_rate": 9.324758385446277e-06, "loss": 0.8909, "step": 19504 }, { "epoch": 2.075441583315599, "grad_norm": 15.2734375, "learning_rate": 9.324722853894259e-06, "loss": 0.9808, "step": 19505 }, { "epoch": 2.0755479889338155, "grad_norm": 14.203125, "learning_rate": 9.32468732234224e-06, "loss": 0.8533, "step": 19506 }, { "epoch": 2.0756543945520325, "grad_norm": 18.78125, "learning_rate": 9.324651790790222e-06, "loss": 0.743, "step": 19507 }, { "epoch": 2.075760800170249, "grad_norm": 14.4375, "learning_rate": 9.324616259238204e-06, "loss": 0.9535, "step": 19508 }, { "epoch": 2.0758672057884655, "grad_norm": 11.609375, "learning_rate": 9.324580727686187e-06, "loss": 0.9073, "step": 19509 }, { "epoch": 2.0759736114066825, "grad_norm": 10.1015625, "learning_rate": 9.324545196134169e-06, "loss": 0.7801, "step": 19510 }, { "epoch": 2.076080017024899, "grad_norm": 12.671875, "learning_rate": 9.32450966458215e-06, "loss": 0.8764, "step": 19511 }, { "epoch": 2.0761864226431155, "grad_norm": 17.15625, "learning_rate": 9.324474133030132e-06, "loss": 0.7998, "step": 19512 }, { "epoch": 2.076292828261332, "grad_norm": 11.453125, "learning_rate": 9.324438601478113e-06, "loss": 0.8129, "step": 19513 }, { "epoch": 2.076399233879549, "grad_norm": 11.4453125, "learning_rate": 9.324403069926095e-06, "loss": 0.76, "step": 19514 }, { "epoch": 2.0765056394977655, "grad_norm": 11.6328125, "learning_rate": 9.324367538374077e-06, "loss": 0.8395, "step": 19515 }, { "epoch": 2.076612045115982, "grad_norm": 11.203125, "learning_rate": 9.324332006822058e-06, "loss": 0.6966, "step": 19516 }, { "epoch": 2.076718450734199, "grad_norm": 13.8515625, "learning_rate": 9.32429647527004e-06, "loss": 0.8991, "step": 19517 }, { "epoch": 2.0768248563524154, "grad_norm": 12.9765625, "learning_rate": 9.324260943718021e-06, "loss": 0.878, "step": 19518 }, { "epoch": 2.076931261970632, "grad_norm": 13.59375, "learning_rate": 9.324225412166003e-06, "loss": 1.004, "step": 19519 }, { "epoch": 2.077037667588849, "grad_norm": 14.2578125, "learning_rate": 9.324189880613986e-06, "loss": 1.0374, "step": 19520 }, { "epoch": 2.0771440732070654, "grad_norm": 13.9921875, "learning_rate": 9.324154349061968e-06, "loss": 0.8073, "step": 19521 }, { "epoch": 2.077250478825282, "grad_norm": 13.078125, "learning_rate": 9.32411881750995e-06, "loss": 0.8696, "step": 19522 }, { "epoch": 2.0773568844434984, "grad_norm": 10.125, "learning_rate": 9.324083285957931e-06, "loss": 0.8922, "step": 19523 }, { "epoch": 2.0774632900617154, "grad_norm": 11.1953125, "learning_rate": 9.324047754405913e-06, "loss": 0.8985, "step": 19524 }, { "epoch": 2.077569695679932, "grad_norm": 13.8046875, "learning_rate": 9.324012222853896e-06, "loss": 0.9643, "step": 19525 }, { "epoch": 2.0776761012981484, "grad_norm": 16.828125, "learning_rate": 9.323976691301878e-06, "loss": 1.0333, "step": 19526 }, { "epoch": 2.0777825069163653, "grad_norm": 17.359375, "learning_rate": 9.32394115974986e-06, "loss": 0.9272, "step": 19527 }, { "epoch": 2.077888912534582, "grad_norm": 11.0078125, "learning_rate": 9.323905628197841e-06, "loss": 0.9702, "step": 19528 }, { "epoch": 2.0779953181527984, "grad_norm": 13.359375, "learning_rate": 9.323870096645823e-06, "loss": 0.8617, "step": 19529 }, { "epoch": 2.0781017237710153, "grad_norm": 13.5546875, "learning_rate": 9.323834565093804e-06, "loss": 0.8795, "step": 19530 }, { "epoch": 2.078208129389232, "grad_norm": 11.4296875, "learning_rate": 9.323799033541786e-06, "loss": 0.9561, "step": 19531 }, { "epoch": 2.0783145350074483, "grad_norm": 12.1328125, "learning_rate": 9.323763501989767e-06, "loss": 0.8248, "step": 19532 }, { "epoch": 2.078420940625665, "grad_norm": 16.1875, "learning_rate": 9.323727970437749e-06, "loss": 0.9128, "step": 19533 }, { "epoch": 2.078527346243882, "grad_norm": 11.390625, "learning_rate": 9.32369243888573e-06, "loss": 0.7622, "step": 19534 }, { "epoch": 2.0786337518620983, "grad_norm": 11.09375, "learning_rate": 9.323656907333714e-06, "loss": 0.9203, "step": 19535 }, { "epoch": 2.078740157480315, "grad_norm": 13.2421875, "learning_rate": 9.323621375781695e-06, "loss": 0.9982, "step": 19536 }, { "epoch": 2.0788465630985318, "grad_norm": 12.3984375, "learning_rate": 9.323585844229677e-06, "loss": 0.9907, "step": 19537 }, { "epoch": 2.0789529687167483, "grad_norm": 8.90625, "learning_rate": 9.323550312677659e-06, "loss": 0.7808, "step": 19538 }, { "epoch": 2.079059374334965, "grad_norm": 9.3515625, "learning_rate": 9.32351478112564e-06, "loss": 0.8045, "step": 19539 }, { "epoch": 2.0791657799531817, "grad_norm": 10.984375, "learning_rate": 9.323479249573622e-06, "loss": 0.9801, "step": 19540 }, { "epoch": 2.0792721855713983, "grad_norm": 11.125, "learning_rate": 9.323443718021604e-06, "loss": 0.8361, "step": 19541 }, { "epoch": 2.0793785911896148, "grad_norm": 12.375, "learning_rate": 9.323408186469585e-06, "loss": 1.0224, "step": 19542 }, { "epoch": 2.0794849968078313, "grad_norm": 13.609375, "learning_rate": 9.323372654917567e-06, "loss": 0.9294, "step": 19543 }, { "epoch": 2.0795914024260482, "grad_norm": 17.046875, "learning_rate": 9.323337123365548e-06, "loss": 0.8639, "step": 19544 }, { "epoch": 2.0796978080442647, "grad_norm": 13.0078125, "learning_rate": 9.323301591813532e-06, "loss": 0.9625, "step": 19545 }, { "epoch": 2.0798042136624812, "grad_norm": 12.5859375, "learning_rate": 9.323266060261513e-06, "loss": 0.9335, "step": 19546 }, { "epoch": 2.079910619280698, "grad_norm": 13.0234375, "learning_rate": 9.323230528709495e-06, "loss": 1.033, "step": 19547 }, { "epoch": 2.0800170248989147, "grad_norm": 15.0859375, "learning_rate": 9.323194997157476e-06, "loss": 1.0343, "step": 19548 }, { "epoch": 2.0801234305171312, "grad_norm": 21.5, "learning_rate": 9.323159465605458e-06, "loss": 1.3246, "step": 19549 }, { "epoch": 2.0802298361353477, "grad_norm": 11.1640625, "learning_rate": 9.323123934053441e-06, "loss": 0.9029, "step": 19550 }, { "epoch": 2.0803362417535647, "grad_norm": 10.375, "learning_rate": 9.323088402501423e-06, "loss": 0.7338, "step": 19551 }, { "epoch": 2.080442647371781, "grad_norm": 10.4765625, "learning_rate": 9.323052870949405e-06, "loss": 0.8209, "step": 19552 }, { "epoch": 2.0805490529899977, "grad_norm": 9.2734375, "learning_rate": 9.323017339397386e-06, "loss": 0.7552, "step": 19553 }, { "epoch": 2.0806554586082147, "grad_norm": 10.3671875, "learning_rate": 9.322981807845368e-06, "loss": 0.8514, "step": 19554 }, { "epoch": 2.080761864226431, "grad_norm": 18.125, "learning_rate": 9.32294627629335e-06, "loss": 0.7824, "step": 19555 }, { "epoch": 2.0808682698446477, "grad_norm": 12.7265625, "learning_rate": 9.322910744741331e-06, "loss": 0.9435, "step": 19556 }, { "epoch": 2.0809746754628646, "grad_norm": 14.5703125, "learning_rate": 9.322875213189313e-06, "loss": 0.8749, "step": 19557 }, { "epoch": 2.081081081081081, "grad_norm": 10.71875, "learning_rate": 9.322839681637294e-06, "loss": 0.6262, "step": 19558 }, { "epoch": 2.0811874866992977, "grad_norm": 14.0390625, "learning_rate": 9.322804150085276e-06, "loss": 0.883, "step": 19559 }, { "epoch": 2.081293892317514, "grad_norm": 11.4453125, "learning_rate": 9.322768618533257e-06, "loss": 0.9765, "step": 19560 }, { "epoch": 2.081400297935731, "grad_norm": 14.9609375, "learning_rate": 9.32273308698124e-06, "loss": 0.9296, "step": 19561 }, { "epoch": 2.0815067035539476, "grad_norm": 18.546875, "learning_rate": 9.322697555429222e-06, "loss": 0.8998, "step": 19562 }, { "epoch": 2.081613109172164, "grad_norm": 13.6171875, "learning_rate": 9.322662023877204e-06, "loss": 0.9643, "step": 19563 }, { "epoch": 2.081719514790381, "grad_norm": 11.0390625, "learning_rate": 9.322626492325186e-06, "loss": 0.8368, "step": 19564 }, { "epoch": 2.0818259204085976, "grad_norm": 14.484375, "learning_rate": 9.322590960773167e-06, "loss": 1.0645, "step": 19565 }, { "epoch": 2.081932326026814, "grad_norm": 13.75, "learning_rate": 9.322555429221149e-06, "loss": 1.1236, "step": 19566 }, { "epoch": 2.082038731645031, "grad_norm": 12.34375, "learning_rate": 9.32251989766913e-06, "loss": 1.0629, "step": 19567 }, { "epoch": 2.0821451372632476, "grad_norm": 16.203125, "learning_rate": 9.322484366117112e-06, "loss": 0.7433, "step": 19568 }, { "epoch": 2.082251542881464, "grad_norm": 11.90625, "learning_rate": 9.322448834565094e-06, "loss": 0.901, "step": 19569 }, { "epoch": 2.0823579484996806, "grad_norm": 16.109375, "learning_rate": 9.322413303013075e-06, "loss": 1.1895, "step": 19570 }, { "epoch": 2.0824643541178975, "grad_norm": 14.03125, "learning_rate": 9.322377771461058e-06, "loss": 0.7882, "step": 19571 }, { "epoch": 2.082570759736114, "grad_norm": 13.53125, "learning_rate": 9.32234223990904e-06, "loss": 0.9191, "step": 19572 }, { "epoch": 2.0826771653543306, "grad_norm": 11.625, "learning_rate": 9.322306708357022e-06, "loss": 0.8444, "step": 19573 }, { "epoch": 2.0827835709725475, "grad_norm": 12.6328125, "learning_rate": 9.322271176805003e-06, "loss": 0.9104, "step": 19574 }, { "epoch": 2.082889976590764, "grad_norm": 14.1328125, "learning_rate": 9.322235645252985e-06, "loss": 0.9135, "step": 19575 }, { "epoch": 2.0829963822089805, "grad_norm": 11.609375, "learning_rate": 9.322200113700968e-06, "loss": 0.7663, "step": 19576 }, { "epoch": 2.0831027878271975, "grad_norm": 10.671875, "learning_rate": 9.32216458214895e-06, "loss": 0.7687, "step": 19577 }, { "epoch": 2.083209193445414, "grad_norm": 11.546875, "learning_rate": 9.322129050596931e-06, "loss": 0.8237, "step": 19578 }, { "epoch": 2.0833155990636305, "grad_norm": 10.15625, "learning_rate": 9.322093519044913e-06, "loss": 0.9547, "step": 19579 }, { "epoch": 2.083422004681847, "grad_norm": 12.3515625, "learning_rate": 9.322057987492895e-06, "loss": 0.8779, "step": 19580 }, { "epoch": 2.083528410300064, "grad_norm": 8.7421875, "learning_rate": 9.322022455940876e-06, "loss": 0.8197, "step": 19581 }, { "epoch": 2.0836348159182805, "grad_norm": 14.3984375, "learning_rate": 9.321986924388858e-06, "loss": 1.0013, "step": 19582 }, { "epoch": 2.083741221536497, "grad_norm": 12.9453125, "learning_rate": 9.32195139283684e-06, "loss": 0.9039, "step": 19583 }, { "epoch": 2.083847627154714, "grad_norm": 9.9140625, "learning_rate": 9.321915861284821e-06, "loss": 0.8105, "step": 19584 }, { "epoch": 2.0839540327729305, "grad_norm": 8.9453125, "learning_rate": 9.321880329732803e-06, "loss": 0.7469, "step": 19585 }, { "epoch": 2.084060438391147, "grad_norm": 14.1015625, "learning_rate": 9.321844798180786e-06, "loss": 0.7648, "step": 19586 }, { "epoch": 2.0841668440093635, "grad_norm": 10.15625, "learning_rate": 9.321809266628768e-06, "loss": 0.6366, "step": 19587 }, { "epoch": 2.0842732496275804, "grad_norm": 12.2109375, "learning_rate": 9.32177373507675e-06, "loss": 0.9597, "step": 19588 }, { "epoch": 2.084379655245797, "grad_norm": 18.71875, "learning_rate": 9.32173820352473e-06, "loss": 1.0905, "step": 19589 }, { "epoch": 2.0844860608640134, "grad_norm": 11.4765625, "learning_rate": 9.321702671972712e-06, "loss": 0.8412, "step": 19590 }, { "epoch": 2.0845924664822304, "grad_norm": 8.96875, "learning_rate": 9.321667140420696e-06, "loss": 0.6692, "step": 19591 }, { "epoch": 2.084698872100447, "grad_norm": 15.3046875, "learning_rate": 9.321631608868676e-06, "loss": 0.8792, "step": 19592 }, { "epoch": 2.0848052777186634, "grad_norm": 16.625, "learning_rate": 9.321596077316657e-06, "loss": 0.7841, "step": 19593 }, { "epoch": 2.0849116833368804, "grad_norm": 11.796875, "learning_rate": 9.321560545764639e-06, "loss": 0.9088, "step": 19594 }, { "epoch": 2.085018088955097, "grad_norm": 13.4140625, "learning_rate": 9.32152501421262e-06, "loss": 0.8852, "step": 19595 }, { "epoch": 2.0851244945733134, "grad_norm": 13.71875, "learning_rate": 9.321489482660602e-06, "loss": 0.8782, "step": 19596 }, { "epoch": 2.08523090019153, "grad_norm": 11.2265625, "learning_rate": 9.321453951108585e-06, "loss": 0.8049, "step": 19597 }, { "epoch": 2.085337305809747, "grad_norm": 14.2421875, "learning_rate": 9.321418419556567e-06, "loss": 0.881, "step": 19598 }, { "epoch": 2.0854437114279634, "grad_norm": 13.4140625, "learning_rate": 9.321382888004549e-06, "loss": 0.872, "step": 19599 }, { "epoch": 2.08555011704618, "grad_norm": 10.6328125, "learning_rate": 9.32134735645253e-06, "loss": 0.8881, "step": 19600 }, { "epoch": 2.085656522664397, "grad_norm": 13.53125, "learning_rate": 9.321311824900512e-06, "loss": 0.9747, "step": 19601 }, { "epoch": 2.0857629282826133, "grad_norm": 12.984375, "learning_rate": 9.321276293348495e-06, "loss": 1.0668, "step": 19602 }, { "epoch": 2.08586933390083, "grad_norm": 11.2578125, "learning_rate": 9.321240761796477e-06, "loss": 0.888, "step": 19603 }, { "epoch": 2.085975739519047, "grad_norm": 13.75, "learning_rate": 9.321205230244458e-06, "loss": 0.7242, "step": 19604 }, { "epoch": 2.0860821451372633, "grad_norm": 10.984375, "learning_rate": 9.32116969869244e-06, "loss": 0.7976, "step": 19605 }, { "epoch": 2.08618855075548, "grad_norm": 14.3515625, "learning_rate": 9.321134167140422e-06, "loss": 1.0553, "step": 19606 }, { "epoch": 2.0862949563736963, "grad_norm": 16.953125, "learning_rate": 9.321098635588403e-06, "loss": 0.864, "step": 19607 }, { "epoch": 2.0864013619919133, "grad_norm": 12.5546875, "learning_rate": 9.321063104036385e-06, "loss": 1.0563, "step": 19608 }, { "epoch": 2.08650776761013, "grad_norm": 10.1015625, "learning_rate": 9.321027572484366e-06, "loss": 0.8104, "step": 19609 }, { "epoch": 2.0866141732283463, "grad_norm": 15.0546875, "learning_rate": 9.320992040932348e-06, "loss": 0.9362, "step": 19610 }, { "epoch": 2.0867205788465633, "grad_norm": 10.96875, "learning_rate": 9.32095650938033e-06, "loss": 0.9978, "step": 19611 }, { "epoch": 2.0868269844647798, "grad_norm": 15.5, "learning_rate": 9.320920977828313e-06, "loss": 1.0047, "step": 19612 }, { "epoch": 2.0869333900829963, "grad_norm": 11.6171875, "learning_rate": 9.320885446276294e-06, "loss": 0.9276, "step": 19613 }, { "epoch": 2.0870397957012132, "grad_norm": 14.3125, "learning_rate": 9.320849914724276e-06, "loss": 0.9152, "step": 19614 }, { "epoch": 2.0871462013194297, "grad_norm": 11.65625, "learning_rate": 9.320814383172258e-06, "loss": 0.9386, "step": 19615 }, { "epoch": 2.0872526069376462, "grad_norm": 16.859375, "learning_rate": 9.32077885162024e-06, "loss": 0.9101, "step": 19616 }, { "epoch": 2.0873590125558628, "grad_norm": 11.84375, "learning_rate": 9.320743320068223e-06, "loss": 1.0248, "step": 19617 }, { "epoch": 2.0874654181740797, "grad_norm": 18.96875, "learning_rate": 9.320707788516204e-06, "loss": 1.0606, "step": 19618 }, { "epoch": 2.087571823792296, "grad_norm": 10.859375, "learning_rate": 9.320672256964186e-06, "loss": 0.846, "step": 19619 }, { "epoch": 2.0876782294105127, "grad_norm": 11.6328125, "learning_rate": 9.320636725412167e-06, "loss": 0.94, "step": 19620 }, { "epoch": 2.0877846350287297, "grad_norm": 11.3203125, "learning_rate": 9.320601193860149e-06, "loss": 0.9293, "step": 19621 }, { "epoch": 2.087891040646946, "grad_norm": 12.2578125, "learning_rate": 9.32056566230813e-06, "loss": 0.8198, "step": 19622 }, { "epoch": 2.0879974462651627, "grad_norm": 13.9296875, "learning_rate": 9.320530130756112e-06, "loss": 0.8475, "step": 19623 }, { "epoch": 2.088103851883379, "grad_norm": 13.3359375, "learning_rate": 9.320494599204094e-06, "loss": 0.9128, "step": 19624 }, { "epoch": 2.088210257501596, "grad_norm": 13.1328125, "learning_rate": 9.320459067652075e-06, "loss": 1.0133, "step": 19625 }, { "epoch": 2.0883166631198127, "grad_norm": 16.140625, "learning_rate": 9.320423536100057e-06, "loss": 0.942, "step": 19626 }, { "epoch": 2.088423068738029, "grad_norm": 14.015625, "learning_rate": 9.32038800454804e-06, "loss": 0.9518, "step": 19627 }, { "epoch": 2.088529474356246, "grad_norm": 12.859375, "learning_rate": 9.320352472996022e-06, "loss": 0.9687, "step": 19628 }, { "epoch": 2.0886358799744627, "grad_norm": 12.9296875, "learning_rate": 9.320316941444004e-06, "loss": 0.7457, "step": 19629 }, { "epoch": 2.088742285592679, "grad_norm": 11.28125, "learning_rate": 9.320281409891985e-06, "loss": 0.9383, "step": 19630 }, { "epoch": 2.088848691210896, "grad_norm": 12.484375, "learning_rate": 9.320245878339967e-06, "loss": 0.8519, "step": 19631 }, { "epoch": 2.0889550968291126, "grad_norm": 18.25, "learning_rate": 9.320210346787948e-06, "loss": 0.9369, "step": 19632 }, { "epoch": 2.089061502447329, "grad_norm": 8.015625, "learning_rate": 9.32017481523593e-06, "loss": 0.7803, "step": 19633 }, { "epoch": 2.089167908065546, "grad_norm": 13.2421875, "learning_rate": 9.320139283683912e-06, "loss": 0.8971, "step": 19634 }, { "epoch": 2.0892743136837626, "grad_norm": 15.265625, "learning_rate": 9.320103752131893e-06, "loss": 0.928, "step": 19635 }, { "epoch": 2.089380719301979, "grad_norm": 14.1015625, "learning_rate": 9.320068220579875e-06, "loss": 0.8084, "step": 19636 }, { "epoch": 2.0894871249201956, "grad_norm": 12.09375, "learning_rate": 9.320032689027856e-06, "loss": 0.777, "step": 19637 }, { "epoch": 2.0895935305384126, "grad_norm": 13.8515625, "learning_rate": 9.31999715747584e-06, "loss": 1.1973, "step": 19638 }, { "epoch": 2.089699936156629, "grad_norm": 12.1484375, "learning_rate": 9.319961625923821e-06, "loss": 0.84, "step": 19639 }, { "epoch": 2.0898063417748456, "grad_norm": 16.859375, "learning_rate": 9.319926094371803e-06, "loss": 0.9774, "step": 19640 }, { "epoch": 2.0899127473930625, "grad_norm": 17.4375, "learning_rate": 9.319890562819785e-06, "loss": 0.7502, "step": 19641 }, { "epoch": 2.090019153011279, "grad_norm": 12.234375, "learning_rate": 9.319855031267766e-06, "loss": 0.8809, "step": 19642 }, { "epoch": 2.0901255586294956, "grad_norm": 12.390625, "learning_rate": 9.31981949971575e-06, "loss": 1.02, "step": 19643 }, { "epoch": 2.090231964247712, "grad_norm": 15.0703125, "learning_rate": 9.319783968163731e-06, "loss": 0.9632, "step": 19644 }, { "epoch": 2.090338369865929, "grad_norm": 17.140625, "learning_rate": 9.319748436611713e-06, "loss": 1.0255, "step": 19645 }, { "epoch": 2.0904447754841455, "grad_norm": 11.4609375, "learning_rate": 9.319712905059694e-06, "loss": 0.698, "step": 19646 }, { "epoch": 2.090551181102362, "grad_norm": 12.265625, "learning_rate": 9.319677373507676e-06, "loss": 0.8121, "step": 19647 }, { "epoch": 2.090657586720579, "grad_norm": 10.46875, "learning_rate": 9.319641841955657e-06, "loss": 0.7465, "step": 19648 }, { "epoch": 2.0907639923387955, "grad_norm": 14.015625, "learning_rate": 9.319606310403639e-06, "loss": 1.0596, "step": 19649 }, { "epoch": 2.090870397957012, "grad_norm": 10.1484375, "learning_rate": 9.31957077885162e-06, "loss": 0.8506, "step": 19650 }, { "epoch": 2.090976803575229, "grad_norm": 12.09375, "learning_rate": 9.319535247299602e-06, "loss": 1.0289, "step": 19651 }, { "epoch": 2.0910832091934455, "grad_norm": 11.203125, "learning_rate": 9.319499715747584e-06, "loss": 0.8126, "step": 19652 }, { "epoch": 2.091189614811662, "grad_norm": 15.6640625, "learning_rate": 9.319464184195567e-06, "loss": 0.9116, "step": 19653 }, { "epoch": 2.0912960204298785, "grad_norm": 10.6171875, "learning_rate": 9.319428652643549e-06, "loss": 0.7571, "step": 19654 }, { "epoch": 2.0914024260480955, "grad_norm": 10.0, "learning_rate": 9.31939312109153e-06, "loss": 1.0118, "step": 19655 }, { "epoch": 2.091508831666312, "grad_norm": 13.3515625, "learning_rate": 9.319357589539512e-06, "loss": 0.8564, "step": 19656 }, { "epoch": 2.0916152372845285, "grad_norm": 8.6953125, "learning_rate": 9.319322057987494e-06, "loss": 0.8356, "step": 19657 }, { "epoch": 2.0917216429027454, "grad_norm": 10.9453125, "learning_rate": 9.319286526435475e-06, "loss": 0.8462, "step": 19658 }, { "epoch": 2.091828048520962, "grad_norm": 12.5, "learning_rate": 9.319250994883457e-06, "loss": 0.8518, "step": 19659 }, { "epoch": 2.0919344541391784, "grad_norm": 11.921875, "learning_rate": 9.319215463331438e-06, "loss": 0.9185, "step": 19660 }, { "epoch": 2.0920408597573954, "grad_norm": 18.0625, "learning_rate": 9.31917993177942e-06, "loss": 1.0056, "step": 19661 }, { "epoch": 2.092147265375612, "grad_norm": 13.2109375, "learning_rate": 9.319144400227402e-06, "loss": 0.8673, "step": 19662 }, { "epoch": 2.0922536709938284, "grad_norm": 16.015625, "learning_rate": 9.319108868675385e-06, "loss": 0.7639, "step": 19663 }, { "epoch": 2.092360076612045, "grad_norm": 12.9296875, "learning_rate": 9.319073337123367e-06, "loss": 0.8787, "step": 19664 }, { "epoch": 2.092466482230262, "grad_norm": 13.4765625, "learning_rate": 9.319037805571348e-06, "loss": 0.8935, "step": 19665 }, { "epoch": 2.0925728878484784, "grad_norm": 13.6796875, "learning_rate": 9.31900227401933e-06, "loss": 0.9057, "step": 19666 }, { "epoch": 2.092679293466695, "grad_norm": 10.9609375, "learning_rate": 9.318966742467311e-06, "loss": 0.8807, "step": 19667 }, { "epoch": 2.092785699084912, "grad_norm": 12.9140625, "learning_rate": 9.318931210915295e-06, "loss": 0.8793, "step": 19668 }, { "epoch": 2.0928921047031284, "grad_norm": 19.609375, "learning_rate": 9.318895679363276e-06, "loss": 1.1164, "step": 19669 }, { "epoch": 2.092998510321345, "grad_norm": 10.53125, "learning_rate": 9.318860147811258e-06, "loss": 0.7935, "step": 19670 }, { "epoch": 2.093104915939562, "grad_norm": 15.9140625, "learning_rate": 9.31882461625924e-06, "loss": 1.236, "step": 19671 }, { "epoch": 2.0932113215577783, "grad_norm": 12.4609375, "learning_rate": 9.318789084707221e-06, "loss": 0.8963, "step": 19672 }, { "epoch": 2.093317727175995, "grad_norm": 13.4609375, "learning_rate": 9.318753553155203e-06, "loss": 0.8569, "step": 19673 }, { "epoch": 2.0934241327942114, "grad_norm": 18.5, "learning_rate": 9.318718021603184e-06, "loss": 0.894, "step": 19674 }, { "epoch": 2.0935305384124283, "grad_norm": 14.0625, "learning_rate": 9.318682490051166e-06, "loss": 0.8496, "step": 19675 }, { "epoch": 2.093636944030645, "grad_norm": 17.4375, "learning_rate": 9.318646958499148e-06, "loss": 1.0018, "step": 19676 }, { "epoch": 2.0937433496488613, "grad_norm": 13.8828125, "learning_rate": 9.318611426947129e-06, "loss": 1.1437, "step": 19677 }, { "epoch": 2.0938497552670783, "grad_norm": 15.015625, "learning_rate": 9.31857589539511e-06, "loss": 0.8527, "step": 19678 }, { "epoch": 2.093956160885295, "grad_norm": 11.546875, "learning_rate": 9.318540363843094e-06, "loss": 0.7354, "step": 19679 }, { "epoch": 2.0940625665035113, "grad_norm": 10.828125, "learning_rate": 9.318504832291076e-06, "loss": 0.7544, "step": 19680 }, { "epoch": 2.094168972121728, "grad_norm": 15.9609375, "learning_rate": 9.318469300739057e-06, "loss": 0.9948, "step": 19681 }, { "epoch": 2.0942753777399448, "grad_norm": 11.984375, "learning_rate": 9.318433769187039e-06, "loss": 0.7209, "step": 19682 }, { "epoch": 2.0943817833581613, "grad_norm": 10.8984375, "learning_rate": 9.31839823763502e-06, "loss": 0.8662, "step": 19683 }, { "epoch": 2.094488188976378, "grad_norm": 10.6484375, "learning_rate": 9.318362706083002e-06, "loss": 0.881, "step": 19684 }, { "epoch": 2.0945945945945947, "grad_norm": 11.2734375, "learning_rate": 9.318327174530984e-06, "loss": 0.8896, "step": 19685 }, { "epoch": 2.0947010002128112, "grad_norm": 10.5390625, "learning_rate": 9.318291642978965e-06, "loss": 0.6475, "step": 19686 }, { "epoch": 2.0948074058310278, "grad_norm": 11.640625, "learning_rate": 9.318256111426947e-06, "loss": 0.7066, "step": 19687 }, { "epoch": 2.0949138114492447, "grad_norm": 11.015625, "learning_rate": 9.318220579874928e-06, "loss": 0.7975, "step": 19688 }, { "epoch": 2.095020217067461, "grad_norm": 10.8671875, "learning_rate": 9.318185048322912e-06, "loss": 0.9767, "step": 19689 }, { "epoch": 2.0951266226856777, "grad_norm": 9.3203125, "learning_rate": 9.318149516770893e-06, "loss": 0.8572, "step": 19690 }, { "epoch": 2.0952330283038942, "grad_norm": 14.2265625, "learning_rate": 9.318113985218875e-06, "loss": 0.8185, "step": 19691 }, { "epoch": 2.095339433922111, "grad_norm": 10.3515625, "learning_rate": 9.318078453666857e-06, "loss": 0.8397, "step": 19692 }, { "epoch": 2.0954458395403277, "grad_norm": 13.09375, "learning_rate": 9.318042922114838e-06, "loss": 1.0121, "step": 19693 }, { "epoch": 2.095552245158544, "grad_norm": 10.7578125, "learning_rate": 9.318007390562822e-06, "loss": 0.8045, "step": 19694 }, { "epoch": 2.095658650776761, "grad_norm": 10.453125, "learning_rate": 9.317971859010803e-06, "loss": 0.8062, "step": 19695 }, { "epoch": 2.0957650563949777, "grad_norm": 15.5859375, "learning_rate": 9.317936327458785e-06, "loss": 0.9021, "step": 19696 }, { "epoch": 2.095871462013194, "grad_norm": 17.046875, "learning_rate": 9.317900795906766e-06, "loss": 0.9057, "step": 19697 }, { "epoch": 2.095977867631411, "grad_norm": 11.53125, "learning_rate": 9.317865264354748e-06, "loss": 0.7124, "step": 19698 }, { "epoch": 2.0960842732496276, "grad_norm": 12.5625, "learning_rate": 9.31782973280273e-06, "loss": 0.9863, "step": 19699 }, { "epoch": 2.096190678867844, "grad_norm": 10.7734375, "learning_rate": 9.317794201250711e-06, "loss": 0.702, "step": 19700 }, { "epoch": 2.0962970844860607, "grad_norm": 13.484375, "learning_rate": 9.317758669698693e-06, "loss": 0.8989, "step": 19701 }, { "epoch": 2.0964034901042776, "grad_norm": 11.9609375, "learning_rate": 9.317723138146674e-06, "loss": 0.9175, "step": 19702 }, { "epoch": 2.096509895722494, "grad_norm": 10.1484375, "learning_rate": 9.317687606594656e-06, "loss": 0.7406, "step": 19703 }, { "epoch": 2.0966163013407106, "grad_norm": 17.21875, "learning_rate": 9.31765207504264e-06, "loss": 0.8588, "step": 19704 }, { "epoch": 2.0967227069589276, "grad_norm": 12.8515625, "learning_rate": 9.317616543490621e-06, "loss": 0.7875, "step": 19705 }, { "epoch": 2.096829112577144, "grad_norm": 17.046875, "learning_rate": 9.317581011938602e-06, "loss": 0.8857, "step": 19706 }, { "epoch": 2.0969355181953606, "grad_norm": 17.65625, "learning_rate": 9.317545480386584e-06, "loss": 0.8866, "step": 19707 }, { "epoch": 2.0970419238135776, "grad_norm": 13.4140625, "learning_rate": 9.317509948834566e-06, "loss": 0.7913, "step": 19708 }, { "epoch": 2.097148329431794, "grad_norm": 14.03125, "learning_rate": 9.317474417282549e-06, "loss": 0.9726, "step": 19709 }, { "epoch": 2.0972547350500106, "grad_norm": 10.6640625, "learning_rate": 9.31743888573053e-06, "loss": 0.8401, "step": 19710 }, { "epoch": 2.097361140668227, "grad_norm": 12.6328125, "learning_rate": 9.317403354178512e-06, "loss": 0.7072, "step": 19711 }, { "epoch": 2.097467546286444, "grad_norm": 13.234375, "learning_rate": 9.317367822626492e-06, "loss": 0.892, "step": 19712 }, { "epoch": 2.0975739519046606, "grad_norm": 12.0234375, "learning_rate": 9.317332291074474e-06, "loss": 0.8952, "step": 19713 }, { "epoch": 2.097680357522877, "grad_norm": 10.4609375, "learning_rate": 9.317296759522455e-06, "loss": 0.9288, "step": 19714 }, { "epoch": 2.097786763141094, "grad_norm": 16.03125, "learning_rate": 9.317261227970439e-06, "loss": 1.1542, "step": 19715 }, { "epoch": 2.0978931687593105, "grad_norm": 13.1875, "learning_rate": 9.31722569641842e-06, "loss": 0.9591, "step": 19716 }, { "epoch": 2.097999574377527, "grad_norm": 12.1484375, "learning_rate": 9.317190164866402e-06, "loss": 0.7705, "step": 19717 }, { "epoch": 2.0981059799957436, "grad_norm": 14.4765625, "learning_rate": 9.317154633314383e-06, "loss": 0.849, "step": 19718 }, { "epoch": 2.0982123856139605, "grad_norm": 12.15625, "learning_rate": 9.317119101762365e-06, "loss": 0.9184, "step": 19719 }, { "epoch": 2.098318791232177, "grad_norm": 14.0078125, "learning_rate": 9.317083570210348e-06, "loss": 0.7542, "step": 19720 }, { "epoch": 2.0984251968503935, "grad_norm": 14.453125, "learning_rate": 9.31704803865833e-06, "loss": 0.8875, "step": 19721 }, { "epoch": 2.0985316024686105, "grad_norm": 12.4296875, "learning_rate": 9.317012507106312e-06, "loss": 0.9442, "step": 19722 }, { "epoch": 2.098638008086827, "grad_norm": 10.515625, "learning_rate": 9.316976975554293e-06, "loss": 0.8843, "step": 19723 }, { "epoch": 2.0987444137050435, "grad_norm": 19.984375, "learning_rate": 9.316941444002275e-06, "loss": 0.9974, "step": 19724 }, { "epoch": 2.0988508193232605, "grad_norm": 13.625, "learning_rate": 9.316905912450256e-06, "loss": 0.8307, "step": 19725 }, { "epoch": 2.098957224941477, "grad_norm": 13.8671875, "learning_rate": 9.316870380898238e-06, "loss": 0.9257, "step": 19726 }, { "epoch": 2.0990636305596935, "grad_norm": 13.078125, "learning_rate": 9.31683484934622e-06, "loss": 0.8624, "step": 19727 }, { "epoch": 2.09917003617791, "grad_norm": 11.6171875, "learning_rate": 9.316799317794201e-06, "loss": 0.983, "step": 19728 }, { "epoch": 2.099276441796127, "grad_norm": 15.453125, "learning_rate": 9.316763786242183e-06, "loss": 0.9787, "step": 19729 }, { "epoch": 2.0993828474143434, "grad_norm": 12.6640625, "learning_rate": 9.316728254690166e-06, "loss": 0.866, "step": 19730 }, { "epoch": 2.09948925303256, "grad_norm": 16.5625, "learning_rate": 9.316692723138148e-06, "loss": 0.8547, "step": 19731 }, { "epoch": 2.099595658650777, "grad_norm": 12.4375, "learning_rate": 9.31665719158613e-06, "loss": 0.8814, "step": 19732 }, { "epoch": 2.0997020642689934, "grad_norm": 13.484375, "learning_rate": 9.316621660034111e-06, "loss": 0.9156, "step": 19733 }, { "epoch": 2.09980846988721, "grad_norm": 12.765625, "learning_rate": 9.316586128482093e-06, "loss": 0.8441, "step": 19734 }, { "epoch": 2.099914875505427, "grad_norm": 12.28125, "learning_rate": 9.316550596930076e-06, "loss": 0.9373, "step": 19735 }, { "epoch": 2.1000212811236434, "grad_norm": 13.671875, "learning_rate": 9.316515065378057e-06, "loss": 0.7044, "step": 19736 }, { "epoch": 2.10012768674186, "grad_norm": 18.578125, "learning_rate": 9.316479533826039e-06, "loss": 0.9633, "step": 19737 }, { "epoch": 2.1002340923600764, "grad_norm": 13.1484375, "learning_rate": 9.31644400227402e-06, "loss": 1.0155, "step": 19738 }, { "epoch": 2.1003404979782934, "grad_norm": 14.9765625, "learning_rate": 9.316408470722002e-06, "loss": 1.0416, "step": 19739 }, { "epoch": 2.10044690359651, "grad_norm": 12.4921875, "learning_rate": 9.316372939169984e-06, "loss": 0.8679, "step": 19740 }, { "epoch": 2.1005533092147264, "grad_norm": 11.3125, "learning_rate": 9.316337407617966e-06, "loss": 0.8622, "step": 19741 }, { "epoch": 2.1006597148329433, "grad_norm": 11.03125, "learning_rate": 9.316301876065947e-06, "loss": 0.8013, "step": 19742 }, { "epoch": 2.10076612045116, "grad_norm": 14.515625, "learning_rate": 9.316266344513929e-06, "loss": 0.8214, "step": 19743 }, { "epoch": 2.1008725260693764, "grad_norm": 17.84375, "learning_rate": 9.31623081296191e-06, "loss": 0.9234, "step": 19744 }, { "epoch": 2.1009789316875933, "grad_norm": 15.375, "learning_rate": 9.316195281409894e-06, "loss": 0.8817, "step": 19745 }, { "epoch": 2.10108533730581, "grad_norm": 12.3515625, "learning_rate": 9.316159749857875e-06, "loss": 0.793, "step": 19746 }, { "epoch": 2.1011917429240263, "grad_norm": 12.53125, "learning_rate": 9.316124218305857e-06, "loss": 0.9863, "step": 19747 }, { "epoch": 2.101298148542243, "grad_norm": 14.5390625, "learning_rate": 9.316088686753838e-06, "loss": 1.1545, "step": 19748 }, { "epoch": 2.10140455416046, "grad_norm": 11.71875, "learning_rate": 9.31605315520182e-06, "loss": 0.797, "step": 19749 }, { "epoch": 2.1015109597786763, "grad_norm": 13.015625, "learning_rate": 9.316017623649802e-06, "loss": 1.0258, "step": 19750 }, { "epoch": 2.101617365396893, "grad_norm": 12.921875, "learning_rate": 9.315982092097783e-06, "loss": 0.8365, "step": 19751 }, { "epoch": 2.1017237710151098, "grad_norm": 16.09375, "learning_rate": 9.315946560545765e-06, "loss": 1.2846, "step": 19752 }, { "epoch": 2.1018301766333263, "grad_norm": 13.8671875, "learning_rate": 9.315911028993746e-06, "loss": 0.9473, "step": 19753 }, { "epoch": 2.101936582251543, "grad_norm": 11.390625, "learning_rate": 9.315875497441728e-06, "loss": 0.9736, "step": 19754 }, { "epoch": 2.1020429878697597, "grad_norm": 11.109375, "learning_rate": 9.31583996588971e-06, "loss": 0.9386, "step": 19755 }, { "epoch": 2.1021493934879762, "grad_norm": 9.3515625, "learning_rate": 9.315804434337693e-06, "loss": 0.8347, "step": 19756 }, { "epoch": 2.1022557991061928, "grad_norm": 12.234375, "learning_rate": 9.315768902785675e-06, "loss": 0.8786, "step": 19757 }, { "epoch": 2.1023622047244093, "grad_norm": 9.5546875, "learning_rate": 9.315733371233656e-06, "loss": 0.7662, "step": 19758 }, { "epoch": 2.102468610342626, "grad_norm": 16.859375, "learning_rate": 9.315697839681638e-06, "loss": 0.8892, "step": 19759 }, { "epoch": 2.1025750159608427, "grad_norm": 12.8671875, "learning_rate": 9.31566230812962e-06, "loss": 0.9092, "step": 19760 }, { "epoch": 2.1026814215790592, "grad_norm": 12.203125, "learning_rate": 9.315626776577603e-06, "loss": 0.7312, "step": 19761 }, { "epoch": 2.102787827197276, "grad_norm": 10.6796875, "learning_rate": 9.315591245025584e-06, "loss": 0.6917, "step": 19762 }, { "epoch": 2.1028942328154927, "grad_norm": 14.3125, "learning_rate": 9.315555713473566e-06, "loss": 0.9942, "step": 19763 }, { "epoch": 2.103000638433709, "grad_norm": 11.3203125, "learning_rate": 9.315520181921548e-06, "loss": 0.8978, "step": 19764 }, { "epoch": 2.103107044051926, "grad_norm": 15.7265625, "learning_rate": 9.315484650369529e-06, "loss": 1.1602, "step": 19765 }, { "epoch": 2.1032134496701427, "grad_norm": 9.6875, "learning_rate": 9.31544911881751e-06, "loss": 0.7475, "step": 19766 }, { "epoch": 2.103319855288359, "grad_norm": 12.8828125, "learning_rate": 9.315413587265492e-06, "loss": 0.9594, "step": 19767 }, { "epoch": 2.1034262609065757, "grad_norm": 14.90625, "learning_rate": 9.315378055713474e-06, "loss": 0.6331, "step": 19768 }, { "epoch": 2.1035326665247926, "grad_norm": 10.7421875, "learning_rate": 9.315342524161456e-06, "loss": 0.7672, "step": 19769 }, { "epoch": 2.103639072143009, "grad_norm": 11.3515625, "learning_rate": 9.315306992609437e-06, "loss": 0.814, "step": 19770 }, { "epoch": 2.1037454777612257, "grad_norm": 11.421875, "learning_rate": 9.31527146105742e-06, "loss": 0.8423, "step": 19771 }, { "epoch": 2.1038518833794426, "grad_norm": 13.671875, "learning_rate": 9.315235929505402e-06, "loss": 0.7248, "step": 19772 }, { "epoch": 2.103958288997659, "grad_norm": 13.9765625, "learning_rate": 9.315200397953384e-06, "loss": 1.1003, "step": 19773 }, { "epoch": 2.1040646946158756, "grad_norm": 16.484375, "learning_rate": 9.315164866401365e-06, "loss": 1.0926, "step": 19774 }, { "epoch": 2.104171100234092, "grad_norm": 10.7109375, "learning_rate": 9.315129334849347e-06, "loss": 0.8543, "step": 19775 }, { "epoch": 2.104277505852309, "grad_norm": 14.5546875, "learning_rate": 9.315093803297329e-06, "loss": 0.8279, "step": 19776 }, { "epoch": 2.1043839114705256, "grad_norm": 18.171875, "learning_rate": 9.31505827174531e-06, "loss": 0.8506, "step": 19777 }, { "epoch": 2.104490317088742, "grad_norm": 14.1796875, "learning_rate": 9.315022740193292e-06, "loss": 0.8678, "step": 19778 }, { "epoch": 2.104596722706959, "grad_norm": 12.5078125, "learning_rate": 9.314987208641273e-06, "loss": 0.9285, "step": 19779 }, { "epoch": 2.1047031283251756, "grad_norm": 15.25, "learning_rate": 9.314951677089255e-06, "loss": 0.9327, "step": 19780 }, { "epoch": 2.104809533943392, "grad_norm": 11.640625, "learning_rate": 9.314916145537238e-06, "loss": 0.8469, "step": 19781 }, { "epoch": 2.104915939561609, "grad_norm": 12.953125, "learning_rate": 9.31488061398522e-06, "loss": 0.9865, "step": 19782 }, { "epoch": 2.1050223451798256, "grad_norm": 11.0, "learning_rate": 9.314845082433201e-06, "loss": 0.792, "step": 19783 }, { "epoch": 2.105128750798042, "grad_norm": 10.6953125, "learning_rate": 9.314809550881183e-06, "loss": 0.8027, "step": 19784 }, { "epoch": 2.1052351564162586, "grad_norm": 12.5390625, "learning_rate": 9.314774019329165e-06, "loss": 0.8647, "step": 19785 }, { "epoch": 2.1053415620344755, "grad_norm": 15.96875, "learning_rate": 9.314738487777148e-06, "loss": 1.0244, "step": 19786 }, { "epoch": 2.105447967652692, "grad_norm": 15.703125, "learning_rate": 9.31470295622513e-06, "loss": 1.0428, "step": 19787 }, { "epoch": 2.1055543732709086, "grad_norm": 19.203125, "learning_rate": 9.314667424673111e-06, "loss": 0.8537, "step": 19788 }, { "epoch": 2.1056607788891255, "grad_norm": 10.8359375, "learning_rate": 9.314631893121093e-06, "loss": 0.8303, "step": 19789 }, { "epoch": 2.105767184507342, "grad_norm": 13.0625, "learning_rate": 9.314596361569074e-06, "loss": 0.7833, "step": 19790 }, { "epoch": 2.1058735901255585, "grad_norm": 9.8046875, "learning_rate": 9.314560830017056e-06, "loss": 0.8723, "step": 19791 }, { "epoch": 2.1059799957437755, "grad_norm": 14.953125, "learning_rate": 9.314525298465038e-06, "loss": 0.9689, "step": 19792 }, { "epoch": 2.106086401361992, "grad_norm": 11.4296875, "learning_rate": 9.31448976691302e-06, "loss": 0.7976, "step": 19793 }, { "epoch": 2.1061928069802085, "grad_norm": 14.1796875, "learning_rate": 9.314454235361e-06, "loss": 0.785, "step": 19794 }, { "epoch": 2.106299212598425, "grad_norm": 11.1328125, "learning_rate": 9.314418703808982e-06, "loss": 0.8486, "step": 19795 }, { "epoch": 2.106405618216642, "grad_norm": 11.4921875, "learning_rate": 9.314383172256964e-06, "loss": 0.9311, "step": 19796 }, { "epoch": 2.1065120238348585, "grad_norm": 15.609375, "learning_rate": 9.314347640704947e-06, "loss": 0.9938, "step": 19797 }, { "epoch": 2.106618429453075, "grad_norm": 18.421875, "learning_rate": 9.314312109152929e-06, "loss": 0.8178, "step": 19798 }, { "epoch": 2.106724835071292, "grad_norm": 11.2109375, "learning_rate": 9.31427657760091e-06, "loss": 0.8788, "step": 19799 }, { "epoch": 2.1068312406895084, "grad_norm": 9.8515625, "learning_rate": 9.314241046048892e-06, "loss": 0.738, "step": 19800 }, { "epoch": 2.106937646307725, "grad_norm": 14.328125, "learning_rate": 9.314205514496874e-06, "loss": 0.862, "step": 19801 }, { "epoch": 2.107044051925942, "grad_norm": 13.8515625, "learning_rate": 9.314169982944855e-06, "loss": 1.0512, "step": 19802 }, { "epoch": 2.1071504575441584, "grad_norm": 19.5, "learning_rate": 9.314134451392837e-06, "loss": 0.972, "step": 19803 }, { "epoch": 2.107256863162375, "grad_norm": 16.046875, "learning_rate": 9.314098919840819e-06, "loss": 0.8162, "step": 19804 }, { "epoch": 2.1073632687805914, "grad_norm": 11.265625, "learning_rate": 9.3140633882888e-06, "loss": 0.8257, "step": 19805 }, { "epoch": 2.1074696743988084, "grad_norm": 12.828125, "learning_rate": 9.314027856736782e-06, "loss": 0.8901, "step": 19806 }, { "epoch": 2.107576080017025, "grad_norm": 10.828125, "learning_rate": 9.313992325184765e-06, "loss": 0.9902, "step": 19807 }, { "epoch": 2.1076824856352414, "grad_norm": 16.234375, "learning_rate": 9.313956793632747e-06, "loss": 0.7663, "step": 19808 }, { "epoch": 2.1077888912534584, "grad_norm": 9.265625, "learning_rate": 9.313921262080728e-06, "loss": 0.7945, "step": 19809 }, { "epoch": 2.107895296871675, "grad_norm": 15.109375, "learning_rate": 9.31388573052871e-06, "loss": 0.9747, "step": 19810 }, { "epoch": 2.1080017024898914, "grad_norm": 17.0625, "learning_rate": 9.313850198976692e-06, "loss": 0.8528, "step": 19811 }, { "epoch": 2.108108108108108, "grad_norm": 14.1328125, "learning_rate": 9.313814667424675e-06, "loss": 0.8922, "step": 19812 }, { "epoch": 2.108214513726325, "grad_norm": 10.84375, "learning_rate": 9.313779135872656e-06, "loss": 0.8412, "step": 19813 }, { "epoch": 2.1083209193445414, "grad_norm": 12.96875, "learning_rate": 9.313743604320638e-06, "loss": 0.9472, "step": 19814 }, { "epoch": 2.108427324962758, "grad_norm": 11.265625, "learning_rate": 9.31370807276862e-06, "loss": 0.7874, "step": 19815 }, { "epoch": 2.108533730580975, "grad_norm": 13.5078125, "learning_rate": 9.313672541216601e-06, "loss": 0.9951, "step": 19816 }, { "epoch": 2.1086401361991913, "grad_norm": 9.3203125, "learning_rate": 9.313637009664583e-06, "loss": 0.4751, "step": 19817 }, { "epoch": 2.108746541817408, "grad_norm": 18.21875, "learning_rate": 9.313601478112564e-06, "loss": 0.8444, "step": 19818 }, { "epoch": 2.108852947435625, "grad_norm": 11.7265625, "learning_rate": 9.313565946560546e-06, "loss": 0.8475, "step": 19819 }, { "epoch": 2.1089593530538413, "grad_norm": 16.609375, "learning_rate": 9.313530415008528e-06, "loss": 1.205, "step": 19820 }, { "epoch": 2.109065758672058, "grad_norm": 13.6171875, "learning_rate": 9.31349488345651e-06, "loss": 1.0448, "step": 19821 }, { "epoch": 2.1091721642902743, "grad_norm": 12.5078125, "learning_rate": 9.313459351904493e-06, "loss": 0.8001, "step": 19822 }, { "epoch": 2.1092785699084913, "grad_norm": 16.234375, "learning_rate": 9.313423820352474e-06, "loss": 1.0274, "step": 19823 }, { "epoch": 2.109384975526708, "grad_norm": 14.34375, "learning_rate": 9.313388288800456e-06, "loss": 1.151, "step": 19824 }, { "epoch": 2.1094913811449243, "grad_norm": 14.7734375, "learning_rate": 9.313352757248437e-06, "loss": 0.9715, "step": 19825 }, { "epoch": 2.1095977867631412, "grad_norm": 14.7734375, "learning_rate": 9.313317225696419e-06, "loss": 0.9341, "step": 19826 }, { "epoch": 2.1097041923813578, "grad_norm": 10.984375, "learning_rate": 9.313281694144402e-06, "loss": 0.8725, "step": 19827 }, { "epoch": 2.1098105979995743, "grad_norm": 9.046875, "learning_rate": 9.313246162592384e-06, "loss": 0.8303, "step": 19828 }, { "epoch": 2.109917003617791, "grad_norm": 13.140625, "learning_rate": 9.313210631040366e-06, "loss": 0.9916, "step": 19829 }, { "epoch": 2.1100234092360077, "grad_norm": 12.90625, "learning_rate": 9.313175099488347e-06, "loss": 0.7925, "step": 19830 }, { "epoch": 2.1101298148542242, "grad_norm": 14.6953125, "learning_rate": 9.313139567936329e-06, "loss": 0.8402, "step": 19831 }, { "epoch": 2.1102362204724407, "grad_norm": 10.578125, "learning_rate": 9.313104036384309e-06, "loss": 0.8438, "step": 19832 }, { "epoch": 2.1103426260906577, "grad_norm": 11.8515625, "learning_rate": 9.313068504832292e-06, "loss": 0.844, "step": 19833 }, { "epoch": 2.110449031708874, "grad_norm": 14.6484375, "learning_rate": 9.313032973280274e-06, "loss": 1.0274, "step": 19834 }, { "epoch": 2.1105554373270907, "grad_norm": 11.640625, "learning_rate": 9.312997441728255e-06, "loss": 0.8802, "step": 19835 }, { "epoch": 2.1106618429453077, "grad_norm": 9.7578125, "learning_rate": 9.312961910176237e-06, "loss": 0.8572, "step": 19836 }, { "epoch": 2.110768248563524, "grad_norm": 10.1328125, "learning_rate": 9.312926378624218e-06, "loss": 0.8568, "step": 19837 }, { "epoch": 2.1108746541817407, "grad_norm": 11.0078125, "learning_rate": 9.312890847072202e-06, "loss": 0.8232, "step": 19838 }, { "epoch": 2.1109810597999576, "grad_norm": 13.71875, "learning_rate": 9.312855315520183e-06, "loss": 0.8779, "step": 19839 }, { "epoch": 2.111087465418174, "grad_norm": 16.328125, "learning_rate": 9.312819783968165e-06, "loss": 1.0328, "step": 19840 }, { "epoch": 2.1111938710363907, "grad_norm": 12.5546875, "learning_rate": 9.312784252416147e-06, "loss": 1.0127, "step": 19841 }, { "epoch": 2.111300276654607, "grad_norm": 13.0546875, "learning_rate": 9.312748720864128e-06, "loss": 0.7413, "step": 19842 }, { "epoch": 2.111406682272824, "grad_norm": 11.3515625, "learning_rate": 9.31271318931211e-06, "loss": 0.639, "step": 19843 }, { "epoch": 2.1115130878910406, "grad_norm": 10.6875, "learning_rate": 9.312677657760091e-06, "loss": 0.8061, "step": 19844 }, { "epoch": 2.111619493509257, "grad_norm": 12.6015625, "learning_rate": 9.312642126208073e-06, "loss": 1.0061, "step": 19845 }, { "epoch": 2.111725899127474, "grad_norm": 11.734375, "learning_rate": 9.312606594656055e-06, "loss": 0.9155, "step": 19846 }, { "epoch": 2.1118323047456906, "grad_norm": 16.34375, "learning_rate": 9.312571063104036e-06, "loss": 1.0066, "step": 19847 }, { "epoch": 2.111938710363907, "grad_norm": 10.6484375, "learning_rate": 9.31253553155202e-06, "loss": 0.7934, "step": 19848 }, { "epoch": 2.1120451159821236, "grad_norm": 9.8984375, "learning_rate": 9.312500000000001e-06, "loss": 0.823, "step": 19849 }, { "epoch": 2.1121515216003406, "grad_norm": 10.6875, "learning_rate": 9.312464468447983e-06, "loss": 0.9568, "step": 19850 }, { "epoch": 2.112257927218557, "grad_norm": 13.390625, "learning_rate": 9.312428936895964e-06, "loss": 0.9719, "step": 19851 }, { "epoch": 2.1123643328367736, "grad_norm": 12.9609375, "learning_rate": 9.312393405343946e-06, "loss": 0.9369, "step": 19852 }, { "epoch": 2.1124707384549906, "grad_norm": 15.5546875, "learning_rate": 9.31235787379193e-06, "loss": 0.8498, "step": 19853 }, { "epoch": 2.112577144073207, "grad_norm": 12.8828125, "learning_rate": 9.31232234223991e-06, "loss": 1.193, "step": 19854 }, { "epoch": 2.1126835496914236, "grad_norm": 17.84375, "learning_rate": 9.312286810687892e-06, "loss": 1.048, "step": 19855 }, { "epoch": 2.1127899553096405, "grad_norm": 13.21875, "learning_rate": 9.312251279135874e-06, "loss": 0.7412, "step": 19856 }, { "epoch": 2.112896360927857, "grad_norm": 21.84375, "learning_rate": 9.312215747583856e-06, "loss": 1.1193, "step": 19857 }, { "epoch": 2.1130027665460736, "grad_norm": 13.328125, "learning_rate": 9.312180216031837e-06, "loss": 0.6819, "step": 19858 }, { "epoch": 2.11310917216429, "grad_norm": 12.640625, "learning_rate": 9.312144684479819e-06, "loss": 1.005, "step": 19859 }, { "epoch": 2.113215577782507, "grad_norm": 17.59375, "learning_rate": 9.3121091529278e-06, "loss": 0.9072, "step": 19860 }, { "epoch": 2.1133219834007235, "grad_norm": 14.25, "learning_rate": 9.312073621375782e-06, "loss": 1.065, "step": 19861 }, { "epoch": 2.11342838901894, "grad_norm": 15.765625, "learning_rate": 9.312038089823764e-06, "loss": 0.8926, "step": 19862 }, { "epoch": 2.113534794637157, "grad_norm": 12.9921875, "learning_rate": 9.312002558271747e-06, "loss": 0.9536, "step": 19863 }, { "epoch": 2.1136412002553735, "grad_norm": 13.0625, "learning_rate": 9.311967026719729e-06, "loss": 0.8747, "step": 19864 }, { "epoch": 2.11374760587359, "grad_norm": 14.1171875, "learning_rate": 9.31193149516771e-06, "loss": 0.7432, "step": 19865 }, { "epoch": 2.113854011491807, "grad_norm": 13.1796875, "learning_rate": 9.311895963615692e-06, "loss": 0.9906, "step": 19866 }, { "epoch": 2.1139604171100235, "grad_norm": 12.0546875, "learning_rate": 9.311860432063673e-06, "loss": 0.916, "step": 19867 }, { "epoch": 2.11406682272824, "grad_norm": 10.1484375, "learning_rate": 9.311824900511655e-06, "loss": 0.8496, "step": 19868 }, { "epoch": 2.1141732283464565, "grad_norm": 11.828125, "learning_rate": 9.311789368959637e-06, "loss": 0.9381, "step": 19869 }, { "epoch": 2.1142796339646734, "grad_norm": 20.15625, "learning_rate": 9.311753837407618e-06, "loss": 0.8898, "step": 19870 }, { "epoch": 2.11438603958289, "grad_norm": 15.4765625, "learning_rate": 9.3117183058556e-06, "loss": 0.9853, "step": 19871 }, { "epoch": 2.1144924452011065, "grad_norm": 12.6328125, "learning_rate": 9.311682774303581e-06, "loss": 0.9694, "step": 19872 }, { "epoch": 2.1145988508193234, "grad_norm": 11.8046875, "learning_rate": 9.311647242751563e-06, "loss": 0.951, "step": 19873 }, { "epoch": 2.11470525643754, "grad_norm": 13.3125, "learning_rate": 9.311611711199546e-06, "loss": 0.9771, "step": 19874 }, { "epoch": 2.1148116620557564, "grad_norm": 10.8203125, "learning_rate": 9.311576179647528e-06, "loss": 0.7775, "step": 19875 }, { "epoch": 2.1149180676739734, "grad_norm": 14.1640625, "learning_rate": 9.31154064809551e-06, "loss": 0.9019, "step": 19876 }, { "epoch": 2.11502447329219, "grad_norm": 14.6328125, "learning_rate": 9.311505116543491e-06, "loss": 0.9238, "step": 19877 }, { "epoch": 2.1151308789104064, "grad_norm": 13.8046875, "learning_rate": 9.311469584991473e-06, "loss": 0.8967, "step": 19878 }, { "epoch": 2.115237284528623, "grad_norm": 16.078125, "learning_rate": 9.311434053439456e-06, "loss": 0.9885, "step": 19879 }, { "epoch": 2.11534369014684, "grad_norm": 11.6484375, "learning_rate": 9.311398521887438e-06, "loss": 0.9647, "step": 19880 }, { "epoch": 2.1154500957650564, "grad_norm": 14.0078125, "learning_rate": 9.31136299033542e-06, "loss": 1.2096, "step": 19881 }, { "epoch": 2.115556501383273, "grad_norm": 16.53125, "learning_rate": 9.3113274587834e-06, "loss": 0.9012, "step": 19882 }, { "epoch": 2.11566290700149, "grad_norm": 10.640625, "learning_rate": 9.311291927231382e-06, "loss": 0.835, "step": 19883 }, { "epoch": 2.1157693126197064, "grad_norm": 13.1015625, "learning_rate": 9.311256395679364e-06, "loss": 0.9741, "step": 19884 }, { "epoch": 2.115875718237923, "grad_norm": 11.46875, "learning_rate": 9.311220864127346e-06, "loss": 0.8048, "step": 19885 }, { "epoch": 2.11598212385614, "grad_norm": 13.109375, "learning_rate": 9.311185332575327e-06, "loss": 0.7538, "step": 19886 }, { "epoch": 2.1160885294743563, "grad_norm": 13.78125, "learning_rate": 9.311149801023309e-06, "loss": 0.8357, "step": 19887 }, { "epoch": 2.116194935092573, "grad_norm": 15.4296875, "learning_rate": 9.31111426947129e-06, "loss": 0.943, "step": 19888 }, { "epoch": 2.1163013407107893, "grad_norm": 12.5703125, "learning_rate": 9.311078737919274e-06, "loss": 0.8006, "step": 19889 }, { "epoch": 2.1164077463290063, "grad_norm": 10.3828125, "learning_rate": 9.311043206367255e-06, "loss": 0.8688, "step": 19890 }, { "epoch": 2.116514151947223, "grad_norm": 16.109375, "learning_rate": 9.311007674815237e-06, "loss": 0.8479, "step": 19891 }, { "epoch": 2.1166205575654393, "grad_norm": 10.53125, "learning_rate": 9.310972143263219e-06, "loss": 0.8826, "step": 19892 }, { "epoch": 2.1167269631836563, "grad_norm": 11.0625, "learning_rate": 9.3109366117112e-06, "loss": 0.9208, "step": 19893 }, { "epoch": 2.116833368801873, "grad_norm": 13.9921875, "learning_rate": 9.310901080159182e-06, "loss": 0.9701, "step": 19894 }, { "epoch": 2.1169397744200893, "grad_norm": 10.4140625, "learning_rate": 9.310865548607163e-06, "loss": 0.8105, "step": 19895 }, { "epoch": 2.1170461800383062, "grad_norm": 12.5625, "learning_rate": 9.310830017055145e-06, "loss": 0.8228, "step": 19896 }, { "epoch": 2.1171525856565228, "grad_norm": 13.2890625, "learning_rate": 9.310794485503127e-06, "loss": 1.031, "step": 19897 }, { "epoch": 2.1172589912747393, "grad_norm": 14.9765625, "learning_rate": 9.310758953951108e-06, "loss": 0.9507, "step": 19898 }, { "epoch": 2.1173653968929558, "grad_norm": 14.2421875, "learning_rate": 9.310723422399092e-06, "loss": 1.0514, "step": 19899 }, { "epoch": 2.1174718025111727, "grad_norm": 13.4765625, "learning_rate": 9.310687890847073e-06, "loss": 0.8494, "step": 19900 }, { "epoch": 2.1175782081293892, "grad_norm": 10.3984375, "learning_rate": 9.310652359295055e-06, "loss": 0.8513, "step": 19901 }, { "epoch": 2.1176846137476057, "grad_norm": 14.421875, "learning_rate": 9.310616827743036e-06, "loss": 1.1158, "step": 19902 }, { "epoch": 2.1177910193658227, "grad_norm": 13.1640625, "learning_rate": 9.310581296191018e-06, "loss": 0.7147, "step": 19903 }, { "epoch": 2.117897424984039, "grad_norm": 13.6640625, "learning_rate": 9.310545764639001e-06, "loss": 0.8648, "step": 19904 }, { "epoch": 2.1180038306022557, "grad_norm": 12.796875, "learning_rate": 9.310510233086983e-06, "loss": 0.9188, "step": 19905 }, { "epoch": 2.1181102362204722, "grad_norm": 13.78125, "learning_rate": 9.310474701534964e-06, "loss": 0.8133, "step": 19906 }, { "epoch": 2.118216641838689, "grad_norm": 14.890625, "learning_rate": 9.310439169982946e-06, "loss": 0.7648, "step": 19907 }, { "epoch": 2.1183230474569057, "grad_norm": 12.3359375, "learning_rate": 9.310403638430928e-06, "loss": 0.8866, "step": 19908 }, { "epoch": 2.118429453075122, "grad_norm": 11.71875, "learning_rate": 9.31036810687891e-06, "loss": 0.7674, "step": 19909 }, { "epoch": 2.118535858693339, "grad_norm": 17.453125, "learning_rate": 9.310332575326891e-06, "loss": 1.0901, "step": 19910 }, { "epoch": 2.1186422643115557, "grad_norm": 11.6171875, "learning_rate": 9.310297043774873e-06, "loss": 0.9103, "step": 19911 }, { "epoch": 2.118748669929772, "grad_norm": 12.296875, "learning_rate": 9.310261512222854e-06, "loss": 0.8621, "step": 19912 }, { "epoch": 2.118855075547989, "grad_norm": 11.5078125, "learning_rate": 9.310225980670836e-06, "loss": 0.8309, "step": 19913 }, { "epoch": 2.1189614811662056, "grad_norm": 11.0078125, "learning_rate": 9.310190449118817e-06, "loss": 0.8592, "step": 19914 }, { "epoch": 2.119067886784422, "grad_norm": 12.0390625, "learning_rate": 9.3101549175668e-06, "loss": 0.7989, "step": 19915 }, { "epoch": 2.1191742924026387, "grad_norm": 13.546875, "learning_rate": 9.310119386014782e-06, "loss": 0.874, "step": 19916 }, { "epoch": 2.1192806980208556, "grad_norm": 11.6328125, "learning_rate": 9.310083854462764e-06, "loss": 0.9949, "step": 19917 }, { "epoch": 2.119387103639072, "grad_norm": 12.40625, "learning_rate": 9.310048322910745e-06, "loss": 0.7276, "step": 19918 }, { "epoch": 2.1194935092572886, "grad_norm": 11.8515625, "learning_rate": 9.310012791358727e-06, "loss": 0.9667, "step": 19919 }, { "epoch": 2.1195999148755056, "grad_norm": 18.84375, "learning_rate": 9.30997725980671e-06, "loss": 0.9984, "step": 19920 }, { "epoch": 2.119706320493722, "grad_norm": 12.3046875, "learning_rate": 9.30994172825469e-06, "loss": 0.857, "step": 19921 }, { "epoch": 2.1198127261119386, "grad_norm": 15.15625, "learning_rate": 9.309906196702672e-06, "loss": 0.8051, "step": 19922 }, { "epoch": 2.1199191317301556, "grad_norm": 15.828125, "learning_rate": 9.309870665150653e-06, "loss": 0.8423, "step": 19923 }, { "epoch": 2.120025537348372, "grad_norm": 9.3046875, "learning_rate": 9.309835133598635e-06, "loss": 0.8434, "step": 19924 }, { "epoch": 2.1201319429665886, "grad_norm": 10.5390625, "learning_rate": 9.309799602046618e-06, "loss": 0.726, "step": 19925 }, { "epoch": 2.120238348584805, "grad_norm": 15.125, "learning_rate": 9.3097640704946e-06, "loss": 0.8505, "step": 19926 }, { "epoch": 2.120344754203022, "grad_norm": 12.8125, "learning_rate": 9.309728538942582e-06, "loss": 1.0296, "step": 19927 }, { "epoch": 2.1204511598212386, "grad_norm": 14.1640625, "learning_rate": 9.309693007390563e-06, "loss": 1.0124, "step": 19928 }, { "epoch": 2.120557565439455, "grad_norm": 11.15625, "learning_rate": 9.309657475838545e-06, "loss": 0.7762, "step": 19929 }, { "epoch": 2.120663971057672, "grad_norm": 15.7578125, "learning_rate": 9.309621944286528e-06, "loss": 1.0516, "step": 19930 }, { "epoch": 2.1207703766758885, "grad_norm": 11.171875, "learning_rate": 9.30958641273451e-06, "loss": 0.8438, "step": 19931 }, { "epoch": 2.120876782294105, "grad_norm": 11.375, "learning_rate": 9.309550881182491e-06, "loss": 0.8762, "step": 19932 }, { "epoch": 2.120983187912322, "grad_norm": 15.203125, "learning_rate": 9.309515349630473e-06, "loss": 0.9771, "step": 19933 }, { "epoch": 2.1210895935305385, "grad_norm": 13.890625, "learning_rate": 9.309479818078455e-06, "loss": 1.0619, "step": 19934 }, { "epoch": 2.121195999148755, "grad_norm": 12.3125, "learning_rate": 9.309444286526436e-06, "loss": 0.8406, "step": 19935 }, { "epoch": 2.1213024047669715, "grad_norm": 13.21875, "learning_rate": 9.309408754974418e-06, "loss": 0.923, "step": 19936 }, { "epoch": 2.1214088103851885, "grad_norm": 11.5546875, "learning_rate": 9.3093732234224e-06, "loss": 0.8334, "step": 19937 }, { "epoch": 2.121515216003405, "grad_norm": 11.421875, "learning_rate": 9.309337691870381e-06, "loss": 0.9589, "step": 19938 }, { "epoch": 2.1216216216216215, "grad_norm": 11.59375, "learning_rate": 9.309302160318363e-06, "loss": 0.8637, "step": 19939 }, { "epoch": 2.1217280272398384, "grad_norm": 11.09375, "learning_rate": 9.309266628766346e-06, "loss": 0.9179, "step": 19940 }, { "epoch": 2.121834432858055, "grad_norm": 11.0, "learning_rate": 9.309231097214327e-06, "loss": 0.9444, "step": 19941 }, { "epoch": 2.1219408384762715, "grad_norm": 12.1015625, "learning_rate": 9.309195565662309e-06, "loss": 0.8509, "step": 19942 }, { "epoch": 2.122047244094488, "grad_norm": 14.375, "learning_rate": 9.30916003411029e-06, "loss": 0.8448, "step": 19943 }, { "epoch": 2.122153649712705, "grad_norm": 11.3125, "learning_rate": 9.309124502558272e-06, "loss": 1.0583, "step": 19944 }, { "epoch": 2.1222600553309214, "grad_norm": 12.875, "learning_rate": 9.309088971006256e-06, "loss": 1.0273, "step": 19945 }, { "epoch": 2.122366460949138, "grad_norm": 10.5625, "learning_rate": 9.309053439454237e-06, "loss": 0.8096, "step": 19946 }, { "epoch": 2.122472866567355, "grad_norm": 17.625, "learning_rate": 9.309017907902219e-06, "loss": 0.9083, "step": 19947 }, { "epoch": 2.1225792721855714, "grad_norm": 17.1875, "learning_rate": 9.3089823763502e-06, "loss": 0.9148, "step": 19948 }, { "epoch": 2.122685677803788, "grad_norm": 11.9375, "learning_rate": 9.308946844798182e-06, "loss": 0.7385, "step": 19949 }, { "epoch": 2.122792083422005, "grad_norm": 14.296875, "learning_rate": 9.308911313246164e-06, "loss": 0.9428, "step": 19950 }, { "epoch": 2.1228984890402214, "grad_norm": 14.453125, "learning_rate": 9.308875781694145e-06, "loss": 0.9466, "step": 19951 }, { "epoch": 2.123004894658438, "grad_norm": 16.5625, "learning_rate": 9.308840250142127e-06, "loss": 0.87, "step": 19952 }, { "epoch": 2.1231113002766544, "grad_norm": 17.1875, "learning_rate": 9.308804718590108e-06, "loss": 0.8431, "step": 19953 }, { "epoch": 2.1232177058948714, "grad_norm": 12.90625, "learning_rate": 9.30876918703809e-06, "loss": 0.8243, "step": 19954 }, { "epoch": 2.123324111513088, "grad_norm": 12.96875, "learning_rate": 9.308733655486072e-06, "loss": 0.8495, "step": 19955 }, { "epoch": 2.1234305171313044, "grad_norm": 10.8203125, "learning_rate": 9.308698123934055e-06, "loss": 0.7831, "step": 19956 }, { "epoch": 2.1235369227495213, "grad_norm": 11.8828125, "learning_rate": 9.308662592382037e-06, "loss": 1.0472, "step": 19957 }, { "epoch": 2.123643328367738, "grad_norm": 13.1953125, "learning_rate": 9.308627060830018e-06, "loss": 0.8582, "step": 19958 }, { "epoch": 2.1237497339859543, "grad_norm": 12.203125, "learning_rate": 9.308591529278e-06, "loss": 1.1591, "step": 19959 }, { "epoch": 2.1238561396041713, "grad_norm": 14.9296875, "learning_rate": 9.308555997725981e-06, "loss": 0.9275, "step": 19960 }, { "epoch": 2.123962545222388, "grad_norm": 13.515625, "learning_rate": 9.308520466173963e-06, "loss": 0.7694, "step": 19961 }, { "epoch": 2.1240689508406043, "grad_norm": 13.15625, "learning_rate": 9.308484934621945e-06, "loss": 0.8497, "step": 19962 }, { "epoch": 2.124175356458821, "grad_norm": 11.6015625, "learning_rate": 9.308449403069926e-06, "loss": 0.9019, "step": 19963 }, { "epoch": 2.124281762077038, "grad_norm": 9.9609375, "learning_rate": 9.308413871517908e-06, "loss": 0.8172, "step": 19964 }, { "epoch": 2.1243881676952543, "grad_norm": 24.046875, "learning_rate": 9.30837833996589e-06, "loss": 1.0488, "step": 19965 }, { "epoch": 2.124494573313471, "grad_norm": 10.046875, "learning_rate": 9.308342808413873e-06, "loss": 0.7174, "step": 19966 }, { "epoch": 2.1246009789316878, "grad_norm": 12.0, "learning_rate": 9.308307276861854e-06, "loss": 0.749, "step": 19967 }, { "epoch": 2.1247073845499043, "grad_norm": 15.3125, "learning_rate": 9.308271745309836e-06, "loss": 0.849, "step": 19968 }, { "epoch": 2.1248137901681208, "grad_norm": 16.453125, "learning_rate": 9.308236213757818e-06, "loss": 0.9604, "step": 19969 }, { "epoch": 2.1249201957863377, "grad_norm": 11.90625, "learning_rate": 9.3082006822058e-06, "loss": 0.9191, "step": 19970 }, { "epoch": 2.1250266014045542, "grad_norm": 17.9375, "learning_rate": 9.308165150653782e-06, "loss": 0.8777, "step": 19971 }, { "epoch": 2.1251330070227707, "grad_norm": 11.453125, "learning_rate": 9.308129619101764e-06, "loss": 0.7598, "step": 19972 }, { "epoch": 2.1252394126409873, "grad_norm": 12.5390625, "learning_rate": 9.308094087549746e-06, "loss": 1.004, "step": 19973 }, { "epoch": 2.125345818259204, "grad_norm": 14.6015625, "learning_rate": 9.308058555997727e-06, "loss": 0.8832, "step": 19974 }, { "epoch": 2.1254522238774207, "grad_norm": 12.265625, "learning_rate": 9.308023024445709e-06, "loss": 0.8912, "step": 19975 }, { "epoch": 2.1255586294956372, "grad_norm": 11.828125, "learning_rate": 9.30798749289369e-06, "loss": 0.8659, "step": 19976 }, { "epoch": 2.125665035113854, "grad_norm": 14.1953125, "learning_rate": 9.307951961341672e-06, "loss": 0.9133, "step": 19977 }, { "epoch": 2.1257714407320707, "grad_norm": 10.6796875, "learning_rate": 9.307916429789654e-06, "loss": 0.9125, "step": 19978 }, { "epoch": 2.125877846350287, "grad_norm": 12.203125, "learning_rate": 9.307880898237635e-06, "loss": 0.8235, "step": 19979 }, { "epoch": 2.1259842519685037, "grad_norm": 18.015625, "learning_rate": 9.307845366685617e-06, "loss": 1.0157, "step": 19980 }, { "epoch": 2.1260906575867207, "grad_norm": 13.3125, "learning_rate": 9.3078098351336e-06, "loss": 0.8068, "step": 19981 }, { "epoch": 2.126197063204937, "grad_norm": 10.0625, "learning_rate": 9.307774303581582e-06, "loss": 0.7457, "step": 19982 }, { "epoch": 2.1263034688231537, "grad_norm": 15.5546875, "learning_rate": 9.307738772029563e-06, "loss": 0.9912, "step": 19983 }, { "epoch": 2.1264098744413706, "grad_norm": 14.0625, "learning_rate": 9.307703240477545e-06, "loss": 0.9542, "step": 19984 }, { "epoch": 2.126516280059587, "grad_norm": 13.015625, "learning_rate": 9.307667708925527e-06, "loss": 0.9754, "step": 19985 }, { "epoch": 2.1266226856778037, "grad_norm": 10.265625, "learning_rate": 9.307632177373508e-06, "loss": 0.6876, "step": 19986 }, { "epoch": 2.1267290912960206, "grad_norm": 14.5078125, "learning_rate": 9.30759664582149e-06, "loss": 0.8904, "step": 19987 }, { "epoch": 2.126835496914237, "grad_norm": 12.1171875, "learning_rate": 9.307561114269471e-06, "loss": 0.7792, "step": 19988 }, { "epoch": 2.1269419025324536, "grad_norm": 13.7578125, "learning_rate": 9.307525582717453e-06, "loss": 0.8274, "step": 19989 }, { "epoch": 2.1270483081506706, "grad_norm": 10.234375, "learning_rate": 9.307490051165435e-06, "loss": 0.7953, "step": 19990 }, { "epoch": 2.127154713768887, "grad_norm": 14.28125, "learning_rate": 9.307454519613416e-06, "loss": 1.0104, "step": 19991 }, { "epoch": 2.1272611193871036, "grad_norm": 9.140625, "learning_rate": 9.3074189880614e-06, "loss": 0.7863, "step": 19992 }, { "epoch": 2.12736752500532, "grad_norm": 16.8125, "learning_rate": 9.307383456509381e-06, "loss": 0.9079, "step": 19993 }, { "epoch": 2.127473930623537, "grad_norm": 11.421875, "learning_rate": 9.307347924957363e-06, "loss": 0.8007, "step": 19994 }, { "epoch": 2.1275803362417536, "grad_norm": 10.7578125, "learning_rate": 9.307312393405344e-06, "loss": 0.7891, "step": 19995 }, { "epoch": 2.12768674185997, "grad_norm": 12.515625, "learning_rate": 9.307276861853326e-06, "loss": 0.7547, "step": 19996 }, { "epoch": 2.127793147478187, "grad_norm": 18.171875, "learning_rate": 9.30724133030131e-06, "loss": 1.0284, "step": 19997 }, { "epoch": 2.1278995530964036, "grad_norm": 13.1953125, "learning_rate": 9.307205798749291e-06, "loss": 1.0978, "step": 19998 }, { "epoch": 2.12800595871462, "grad_norm": 14.0234375, "learning_rate": 9.307170267197273e-06, "loss": 1.0177, "step": 19999 }, { "epoch": 2.1281123643328366, "grad_norm": 12.1328125, "learning_rate": 9.307134735645254e-06, "loss": 0.9078, "step": 20000 }, { "epoch": 2.1282187699510535, "grad_norm": 11.0703125, "learning_rate": 9.307099204093236e-06, "loss": 0.6458, "step": 20001 }, { "epoch": 2.12832517556927, "grad_norm": 12.375, "learning_rate": 9.307063672541217e-06, "loss": 0.9802, "step": 20002 }, { "epoch": 2.1284315811874865, "grad_norm": 18.375, "learning_rate": 9.307028140989199e-06, "loss": 1.1263, "step": 20003 }, { "epoch": 2.1285379868057035, "grad_norm": 12.8046875, "learning_rate": 9.30699260943718e-06, "loss": 0.8815, "step": 20004 }, { "epoch": 2.12864439242392, "grad_norm": 12.3828125, "learning_rate": 9.306957077885162e-06, "loss": 0.8001, "step": 20005 }, { "epoch": 2.1287507980421365, "grad_norm": 11.5703125, "learning_rate": 9.306921546333144e-06, "loss": 0.9698, "step": 20006 }, { "epoch": 2.1288572036603535, "grad_norm": 15.59375, "learning_rate": 9.306886014781127e-06, "loss": 1.3867, "step": 20007 }, { "epoch": 2.12896360927857, "grad_norm": 11.609375, "learning_rate": 9.306850483229109e-06, "loss": 0.9321, "step": 20008 }, { "epoch": 2.1290700148967865, "grad_norm": 13.234375, "learning_rate": 9.30681495167709e-06, "loss": 0.846, "step": 20009 }, { "epoch": 2.129176420515003, "grad_norm": 15.6015625, "learning_rate": 9.306779420125072e-06, "loss": 0.9815, "step": 20010 }, { "epoch": 2.12928282613322, "grad_norm": 18.484375, "learning_rate": 9.306743888573054e-06, "loss": 0.9083, "step": 20011 }, { "epoch": 2.1293892317514365, "grad_norm": 11.765625, "learning_rate": 9.306708357021035e-06, "loss": 0.7689, "step": 20012 }, { "epoch": 2.129495637369653, "grad_norm": 13.015625, "learning_rate": 9.306672825469017e-06, "loss": 0.9103, "step": 20013 }, { "epoch": 2.12960204298787, "grad_norm": 10.9140625, "learning_rate": 9.306637293916998e-06, "loss": 0.9084, "step": 20014 }, { "epoch": 2.1297084486060864, "grad_norm": 23.703125, "learning_rate": 9.30660176236498e-06, "loss": 0.9006, "step": 20015 }, { "epoch": 2.129814854224303, "grad_norm": 11.6015625, "learning_rate": 9.306566230812962e-06, "loss": 0.9427, "step": 20016 }, { "epoch": 2.1299212598425195, "grad_norm": 10.8203125, "learning_rate": 9.306530699260945e-06, "loss": 0.8651, "step": 20017 }, { "epoch": 2.1300276654607364, "grad_norm": 11.0234375, "learning_rate": 9.306495167708926e-06, "loss": 0.907, "step": 20018 }, { "epoch": 2.130134071078953, "grad_norm": 9.3984375, "learning_rate": 9.306459636156908e-06, "loss": 0.6317, "step": 20019 }, { "epoch": 2.1302404766971694, "grad_norm": 15.9296875, "learning_rate": 9.30642410460489e-06, "loss": 0.9996, "step": 20020 }, { "epoch": 2.1303468823153864, "grad_norm": 13.8046875, "learning_rate": 9.306388573052871e-06, "loss": 0.8898, "step": 20021 }, { "epoch": 2.130453287933603, "grad_norm": 12.625, "learning_rate": 9.306353041500855e-06, "loss": 0.731, "step": 20022 }, { "epoch": 2.1305596935518194, "grad_norm": 13.875, "learning_rate": 9.306317509948836e-06, "loss": 0.9457, "step": 20023 }, { "epoch": 2.1306660991700364, "grad_norm": 12.8828125, "learning_rate": 9.306281978396818e-06, "loss": 0.7956, "step": 20024 }, { "epoch": 2.130772504788253, "grad_norm": 11.6953125, "learning_rate": 9.3062464468448e-06, "loss": 1.0572, "step": 20025 }, { "epoch": 2.1308789104064694, "grad_norm": 15.484375, "learning_rate": 9.306210915292781e-06, "loss": 1.0197, "step": 20026 }, { "epoch": 2.1309853160246863, "grad_norm": 11.984375, "learning_rate": 9.306175383740763e-06, "loss": 0.8833, "step": 20027 }, { "epoch": 2.131091721642903, "grad_norm": 17.1875, "learning_rate": 9.306139852188744e-06, "loss": 0.9413, "step": 20028 }, { "epoch": 2.1311981272611193, "grad_norm": 11.6875, "learning_rate": 9.306104320636726e-06, "loss": 0.7554, "step": 20029 }, { "epoch": 2.131304532879336, "grad_norm": 15.71875, "learning_rate": 9.306068789084707e-06, "loss": 0.9208, "step": 20030 }, { "epoch": 2.131410938497553, "grad_norm": 17.078125, "learning_rate": 9.306033257532689e-06, "loss": 0.9264, "step": 20031 }, { "epoch": 2.1315173441157693, "grad_norm": 11.5546875, "learning_rate": 9.30599772598067e-06, "loss": 0.888, "step": 20032 }, { "epoch": 2.131623749733986, "grad_norm": 11.640625, "learning_rate": 9.305962194428654e-06, "loss": 0.8112, "step": 20033 }, { "epoch": 2.131730155352203, "grad_norm": 11.234375, "learning_rate": 9.305926662876636e-06, "loss": 0.7428, "step": 20034 }, { "epoch": 2.1318365609704193, "grad_norm": 12.6640625, "learning_rate": 9.305891131324617e-06, "loss": 0.8843, "step": 20035 }, { "epoch": 2.131942966588636, "grad_norm": 14.90625, "learning_rate": 9.305855599772599e-06, "loss": 0.8701, "step": 20036 }, { "epoch": 2.1320493722068523, "grad_norm": 18.09375, "learning_rate": 9.30582006822058e-06, "loss": 0.9667, "step": 20037 }, { "epoch": 2.1321557778250693, "grad_norm": 16.015625, "learning_rate": 9.305784536668564e-06, "loss": 0.8557, "step": 20038 }, { "epoch": 2.1322621834432858, "grad_norm": 14.2265625, "learning_rate": 9.305749005116545e-06, "loss": 1.0956, "step": 20039 }, { "epoch": 2.1323685890615023, "grad_norm": 13.8828125, "learning_rate": 9.305713473564525e-06, "loss": 0.8214, "step": 20040 }, { "epoch": 2.1324749946797192, "grad_norm": 16.09375, "learning_rate": 9.305677942012507e-06, "loss": 0.9397, "step": 20041 }, { "epoch": 2.1325814002979357, "grad_norm": 12.8359375, "learning_rate": 9.305642410460488e-06, "loss": 0.7666, "step": 20042 }, { "epoch": 2.1326878059161523, "grad_norm": 16.546875, "learning_rate": 9.305606878908472e-06, "loss": 1.0062, "step": 20043 }, { "epoch": 2.132794211534369, "grad_norm": 12.484375, "learning_rate": 9.305571347356453e-06, "loss": 0.7976, "step": 20044 }, { "epoch": 2.1329006171525857, "grad_norm": 13.484375, "learning_rate": 9.305535815804435e-06, "loss": 0.8689, "step": 20045 }, { "epoch": 2.1330070227708022, "grad_norm": 15.3046875, "learning_rate": 9.305500284252417e-06, "loss": 0.9039, "step": 20046 }, { "epoch": 2.1331134283890187, "grad_norm": 11.34375, "learning_rate": 9.305464752700398e-06, "loss": 0.8049, "step": 20047 }, { "epoch": 2.1332198340072357, "grad_norm": 11.140625, "learning_rate": 9.305429221148381e-06, "loss": 0.9444, "step": 20048 }, { "epoch": 2.133326239625452, "grad_norm": 14.171875, "learning_rate": 9.305393689596363e-06, "loss": 0.8294, "step": 20049 }, { "epoch": 2.1334326452436687, "grad_norm": 15.3046875, "learning_rate": 9.305358158044345e-06, "loss": 0.9399, "step": 20050 }, { "epoch": 2.1335390508618857, "grad_norm": 12.328125, "learning_rate": 9.305322626492326e-06, "loss": 0.7867, "step": 20051 }, { "epoch": 2.133645456480102, "grad_norm": 15.0625, "learning_rate": 9.305287094940308e-06, "loss": 0.8928, "step": 20052 }, { "epoch": 2.1337518620983187, "grad_norm": 9.578125, "learning_rate": 9.30525156338829e-06, "loss": 0.8019, "step": 20053 }, { "epoch": 2.1338582677165356, "grad_norm": 14.7109375, "learning_rate": 9.305216031836271e-06, "loss": 0.7957, "step": 20054 }, { "epoch": 2.133964673334752, "grad_norm": 9.5703125, "learning_rate": 9.305180500284253e-06, "loss": 0.8943, "step": 20055 }, { "epoch": 2.1340710789529687, "grad_norm": 17.0, "learning_rate": 9.305144968732234e-06, "loss": 0.9311, "step": 20056 }, { "epoch": 2.134177484571185, "grad_norm": 18.0, "learning_rate": 9.305109437180216e-06, "loss": 1.0765, "step": 20057 }, { "epoch": 2.134283890189402, "grad_norm": 12.0625, "learning_rate": 9.3050739056282e-06, "loss": 0.9112, "step": 20058 }, { "epoch": 2.1343902958076186, "grad_norm": 12.7109375, "learning_rate": 9.30503837407618e-06, "loss": 0.8955, "step": 20059 }, { "epoch": 2.134496701425835, "grad_norm": 11.2109375, "learning_rate": 9.305002842524162e-06, "loss": 0.8599, "step": 20060 }, { "epoch": 2.134603107044052, "grad_norm": 11.46875, "learning_rate": 9.304967310972144e-06, "loss": 0.8038, "step": 20061 }, { "epoch": 2.1347095126622686, "grad_norm": 12.25, "learning_rate": 9.304931779420126e-06, "loss": 0.9641, "step": 20062 }, { "epoch": 2.134815918280485, "grad_norm": 14.3828125, "learning_rate": 9.304896247868109e-06, "loss": 0.8892, "step": 20063 }, { "epoch": 2.134922323898702, "grad_norm": 10.0546875, "learning_rate": 9.30486071631609e-06, "loss": 0.9133, "step": 20064 }, { "epoch": 2.1350287295169186, "grad_norm": 14.3125, "learning_rate": 9.304825184764072e-06, "loss": 0.8161, "step": 20065 }, { "epoch": 2.135135135135135, "grad_norm": 12.9609375, "learning_rate": 9.304789653212054e-06, "loss": 0.8327, "step": 20066 }, { "epoch": 2.1352415407533516, "grad_norm": 10.640625, "learning_rate": 9.304754121660035e-06, "loss": 0.7608, "step": 20067 }, { "epoch": 2.1353479463715686, "grad_norm": 13.3046875, "learning_rate": 9.304718590108017e-06, "loss": 0.8209, "step": 20068 }, { "epoch": 2.135454351989785, "grad_norm": 18.5, "learning_rate": 9.304683058555999e-06, "loss": 1.1245, "step": 20069 }, { "epoch": 2.1355607576080016, "grad_norm": 10.5546875, "learning_rate": 9.30464752700398e-06, "loss": 0.8283, "step": 20070 }, { "epoch": 2.1356671632262185, "grad_norm": 13.328125, "learning_rate": 9.304611995451962e-06, "loss": 1.0276, "step": 20071 }, { "epoch": 2.135773568844435, "grad_norm": 14.4765625, "learning_rate": 9.304576463899943e-06, "loss": 0.9441, "step": 20072 }, { "epoch": 2.1358799744626515, "grad_norm": 11.8046875, "learning_rate": 9.304540932347925e-06, "loss": 0.9039, "step": 20073 }, { "epoch": 2.135986380080868, "grad_norm": 11.515625, "learning_rate": 9.304505400795908e-06, "loss": 0.9599, "step": 20074 }, { "epoch": 2.136092785699085, "grad_norm": 14.359375, "learning_rate": 9.30446986924389e-06, "loss": 1.1373, "step": 20075 }, { "epoch": 2.1361991913173015, "grad_norm": 13.046875, "learning_rate": 9.304434337691871e-06, "loss": 0.8855, "step": 20076 }, { "epoch": 2.136305596935518, "grad_norm": 18.15625, "learning_rate": 9.304398806139853e-06, "loss": 0.9358, "step": 20077 }, { "epoch": 2.136412002553735, "grad_norm": 14.8984375, "learning_rate": 9.304363274587835e-06, "loss": 0.7801, "step": 20078 }, { "epoch": 2.1365184081719515, "grad_norm": 11.875, "learning_rate": 9.304327743035816e-06, "loss": 0.8967, "step": 20079 }, { "epoch": 2.136624813790168, "grad_norm": 11.7890625, "learning_rate": 9.304292211483798e-06, "loss": 0.9121, "step": 20080 }, { "epoch": 2.136731219408385, "grad_norm": 15.0078125, "learning_rate": 9.30425667993178e-06, "loss": 1.0266, "step": 20081 }, { "epoch": 2.1368376250266015, "grad_norm": 14.640625, "learning_rate": 9.304221148379761e-06, "loss": 1.096, "step": 20082 }, { "epoch": 2.136944030644818, "grad_norm": 14.65625, "learning_rate": 9.304185616827743e-06, "loss": 0.8738, "step": 20083 }, { "epoch": 2.1370504362630345, "grad_norm": 11.328125, "learning_rate": 9.304150085275726e-06, "loss": 0.922, "step": 20084 }, { "epoch": 2.1371568418812514, "grad_norm": 18.34375, "learning_rate": 9.304114553723708e-06, "loss": 1.0317, "step": 20085 }, { "epoch": 2.137263247499468, "grad_norm": 12.71875, "learning_rate": 9.30407902217169e-06, "loss": 0.9871, "step": 20086 }, { "epoch": 2.1373696531176845, "grad_norm": 13.6953125, "learning_rate": 9.304043490619671e-06, "loss": 0.9672, "step": 20087 }, { "epoch": 2.1374760587359014, "grad_norm": 12.296875, "learning_rate": 9.304007959067652e-06, "loss": 0.8032, "step": 20088 }, { "epoch": 2.137582464354118, "grad_norm": 10.1796875, "learning_rate": 9.303972427515636e-06, "loss": 0.896, "step": 20089 }, { "epoch": 2.1376888699723344, "grad_norm": 10.15625, "learning_rate": 9.303936895963617e-06, "loss": 0.7428, "step": 20090 }, { "epoch": 2.1377952755905514, "grad_norm": 11.9140625, "learning_rate": 9.303901364411599e-06, "loss": 0.8879, "step": 20091 }, { "epoch": 2.137901681208768, "grad_norm": 12.4765625, "learning_rate": 9.30386583285958e-06, "loss": 0.9681, "step": 20092 }, { "epoch": 2.1380080868269844, "grad_norm": 13.4765625, "learning_rate": 9.303830301307562e-06, "loss": 0.7119, "step": 20093 }, { "epoch": 2.138114492445201, "grad_norm": 12.78125, "learning_rate": 9.303794769755544e-06, "loss": 0.9045, "step": 20094 }, { "epoch": 2.138220898063418, "grad_norm": 13.140625, "learning_rate": 9.303759238203525e-06, "loss": 0.8842, "step": 20095 }, { "epoch": 2.1383273036816344, "grad_norm": 14.0859375, "learning_rate": 9.303723706651507e-06, "loss": 0.8895, "step": 20096 }, { "epoch": 2.138433709299851, "grad_norm": 16.25, "learning_rate": 9.303688175099489e-06, "loss": 1.032, "step": 20097 }, { "epoch": 2.138540114918068, "grad_norm": 12.4453125, "learning_rate": 9.30365264354747e-06, "loss": 0.8894, "step": 20098 }, { "epoch": 2.1386465205362843, "grad_norm": 16.578125, "learning_rate": 9.303617111995454e-06, "loss": 0.9392, "step": 20099 }, { "epoch": 2.138752926154501, "grad_norm": 12.3828125, "learning_rate": 9.303581580443435e-06, "loss": 0.9111, "step": 20100 }, { "epoch": 2.138859331772718, "grad_norm": 13.84375, "learning_rate": 9.303546048891417e-06, "loss": 0.8259, "step": 20101 }, { "epoch": 2.1389657373909343, "grad_norm": 13.765625, "learning_rate": 9.303510517339398e-06, "loss": 0.7608, "step": 20102 }, { "epoch": 2.139072143009151, "grad_norm": 14.5859375, "learning_rate": 9.30347498578738e-06, "loss": 0.8632, "step": 20103 }, { "epoch": 2.1391785486273673, "grad_norm": 12.1796875, "learning_rate": 9.303439454235362e-06, "loss": 0.9095, "step": 20104 }, { "epoch": 2.1392849542455843, "grad_norm": 9.53125, "learning_rate": 9.303403922683343e-06, "loss": 0.8376, "step": 20105 }, { "epoch": 2.139391359863801, "grad_norm": 10.6640625, "learning_rate": 9.303368391131325e-06, "loss": 1.038, "step": 20106 }, { "epoch": 2.1394977654820173, "grad_norm": 12.1875, "learning_rate": 9.303332859579306e-06, "loss": 0.8816, "step": 20107 }, { "epoch": 2.1396041711002343, "grad_norm": 12.828125, "learning_rate": 9.303297328027288e-06, "loss": 0.9022, "step": 20108 }, { "epoch": 2.1397105767184508, "grad_norm": 11.8046875, "learning_rate": 9.30326179647527e-06, "loss": 0.8858, "step": 20109 }, { "epoch": 2.1398169823366673, "grad_norm": 12.09375, "learning_rate": 9.303226264923253e-06, "loss": 1.0094, "step": 20110 }, { "epoch": 2.139923387954884, "grad_norm": 12.734375, "learning_rate": 9.303190733371235e-06, "loss": 1.0288, "step": 20111 }, { "epoch": 2.1400297935731007, "grad_norm": 14.0234375, "learning_rate": 9.303155201819216e-06, "loss": 0.6754, "step": 20112 }, { "epoch": 2.1401361991913173, "grad_norm": 12.59375, "learning_rate": 9.303119670267198e-06, "loss": 0.776, "step": 20113 }, { "epoch": 2.1402426048095338, "grad_norm": 12.109375, "learning_rate": 9.30308413871518e-06, "loss": 0.9021, "step": 20114 }, { "epoch": 2.1403490104277507, "grad_norm": 16.53125, "learning_rate": 9.303048607163163e-06, "loss": 0.9262, "step": 20115 }, { "epoch": 2.1404554160459672, "grad_norm": 12.375, "learning_rate": 9.303013075611144e-06, "loss": 0.8592, "step": 20116 }, { "epoch": 2.1405618216641837, "grad_norm": 10.546875, "learning_rate": 9.302977544059126e-06, "loss": 0.9051, "step": 20117 }, { "epoch": 2.1406682272824007, "grad_norm": 8.4140625, "learning_rate": 9.302942012507107e-06, "loss": 0.7404, "step": 20118 }, { "epoch": 2.140774632900617, "grad_norm": 14.9453125, "learning_rate": 9.302906480955089e-06, "loss": 1.0001, "step": 20119 }, { "epoch": 2.1408810385188337, "grad_norm": 12.7734375, "learning_rate": 9.30287094940307e-06, "loss": 0.7415, "step": 20120 }, { "epoch": 2.1409874441370507, "grad_norm": 19.8125, "learning_rate": 9.302835417851052e-06, "loss": 1.1223, "step": 20121 }, { "epoch": 2.141093849755267, "grad_norm": 14.171875, "learning_rate": 9.302799886299034e-06, "loss": 0.7799, "step": 20122 }, { "epoch": 2.1412002553734837, "grad_norm": 28.671875, "learning_rate": 9.302764354747015e-06, "loss": 1.0345, "step": 20123 }, { "epoch": 2.1413066609917, "grad_norm": 11.5234375, "learning_rate": 9.302728823194997e-06, "loss": 0.7443, "step": 20124 }, { "epoch": 2.141413066609917, "grad_norm": 11.6171875, "learning_rate": 9.30269329164298e-06, "loss": 0.9548, "step": 20125 }, { "epoch": 2.1415194722281337, "grad_norm": 14.1796875, "learning_rate": 9.302657760090962e-06, "loss": 0.9909, "step": 20126 }, { "epoch": 2.14162587784635, "grad_norm": 11.8671875, "learning_rate": 9.302622228538944e-06, "loss": 1.0998, "step": 20127 }, { "epoch": 2.141732283464567, "grad_norm": 16.109375, "learning_rate": 9.302586696986925e-06, "loss": 0.7648, "step": 20128 }, { "epoch": 2.1418386890827836, "grad_norm": 13.8984375, "learning_rate": 9.302551165434907e-06, "loss": 0.8387, "step": 20129 }, { "epoch": 2.141945094701, "grad_norm": 12.5078125, "learning_rate": 9.302515633882888e-06, "loss": 0.8537, "step": 20130 }, { "epoch": 2.1420515003192167, "grad_norm": 11.0859375, "learning_rate": 9.30248010233087e-06, "loss": 0.9723, "step": 20131 }, { "epoch": 2.1421579059374336, "grad_norm": 11.5703125, "learning_rate": 9.302444570778852e-06, "loss": 0.9929, "step": 20132 }, { "epoch": 2.14226431155565, "grad_norm": 14.140625, "learning_rate": 9.302409039226833e-06, "loss": 0.9747, "step": 20133 }, { "epoch": 2.1423707171738666, "grad_norm": 17.703125, "learning_rate": 9.302373507674815e-06, "loss": 0.8104, "step": 20134 }, { "epoch": 2.1424771227920836, "grad_norm": 11.046875, "learning_rate": 9.302337976122798e-06, "loss": 0.9256, "step": 20135 }, { "epoch": 2.1425835284103, "grad_norm": 8.6796875, "learning_rate": 9.30230244457078e-06, "loss": 0.7405, "step": 20136 }, { "epoch": 2.1426899340285166, "grad_norm": 18.421875, "learning_rate": 9.302266913018761e-06, "loss": 1.0309, "step": 20137 }, { "epoch": 2.1427963396467335, "grad_norm": 12.125, "learning_rate": 9.302231381466743e-06, "loss": 0.7857, "step": 20138 }, { "epoch": 2.14290274526495, "grad_norm": 11.1171875, "learning_rate": 9.302195849914725e-06, "loss": 0.9634, "step": 20139 }, { "epoch": 2.1430091508831666, "grad_norm": 11.0625, "learning_rate": 9.302160318362708e-06, "loss": 0.8428, "step": 20140 }, { "epoch": 2.143115556501383, "grad_norm": 16.65625, "learning_rate": 9.30212478681069e-06, "loss": 0.6392, "step": 20141 }, { "epoch": 2.1432219621196, "grad_norm": 13.15625, "learning_rate": 9.302089255258671e-06, "loss": 0.9372, "step": 20142 }, { "epoch": 2.1433283677378165, "grad_norm": 15.90625, "learning_rate": 9.302053723706653e-06, "loss": 0.9985, "step": 20143 }, { "epoch": 2.143434773356033, "grad_norm": 11.65625, "learning_rate": 9.302018192154634e-06, "loss": 0.9703, "step": 20144 }, { "epoch": 2.14354117897425, "grad_norm": 12.8359375, "learning_rate": 9.301982660602616e-06, "loss": 0.9687, "step": 20145 }, { "epoch": 2.1436475845924665, "grad_norm": 10.453125, "learning_rate": 9.301947129050598e-06, "loss": 0.8003, "step": 20146 }, { "epoch": 2.143753990210683, "grad_norm": 9.5, "learning_rate": 9.301911597498579e-06, "loss": 0.7748, "step": 20147 }, { "epoch": 2.1438603958288995, "grad_norm": 9.671875, "learning_rate": 9.30187606594656e-06, "loss": 0.8928, "step": 20148 }, { "epoch": 2.1439668014471165, "grad_norm": 12.015625, "learning_rate": 9.301840534394542e-06, "loss": 1.0971, "step": 20149 }, { "epoch": 2.144073207065333, "grad_norm": 12.5390625, "learning_rate": 9.301805002842524e-06, "loss": 0.8705, "step": 20150 }, { "epoch": 2.1441796126835495, "grad_norm": 10.21875, "learning_rate": 9.301769471290507e-06, "loss": 0.8659, "step": 20151 }, { "epoch": 2.1442860183017665, "grad_norm": 10.0703125, "learning_rate": 9.301733939738489e-06, "loss": 0.8389, "step": 20152 }, { "epoch": 2.144392423919983, "grad_norm": 12.4921875, "learning_rate": 9.30169840818647e-06, "loss": 1.0466, "step": 20153 }, { "epoch": 2.1444988295381995, "grad_norm": 11.96875, "learning_rate": 9.301662876634452e-06, "loss": 0.9647, "step": 20154 }, { "epoch": 2.1446052351564164, "grad_norm": 17.75, "learning_rate": 9.301627345082434e-06, "loss": 0.9464, "step": 20155 }, { "epoch": 2.144711640774633, "grad_norm": 13.0546875, "learning_rate": 9.301591813530417e-06, "loss": 1.1694, "step": 20156 }, { "epoch": 2.1448180463928495, "grad_norm": 13.984375, "learning_rate": 9.301556281978399e-06, "loss": 1.0527, "step": 20157 }, { "epoch": 2.1449244520110664, "grad_norm": 13.7578125, "learning_rate": 9.30152075042638e-06, "loss": 0.8519, "step": 20158 }, { "epoch": 2.145030857629283, "grad_norm": 15.3515625, "learning_rate": 9.301485218874362e-06, "loss": 0.8495, "step": 20159 }, { "epoch": 2.1451372632474994, "grad_norm": 13.921875, "learning_rate": 9.301449687322342e-06, "loss": 0.7859, "step": 20160 }, { "epoch": 2.145243668865716, "grad_norm": 9.4609375, "learning_rate": 9.301414155770325e-06, "loss": 0.7295, "step": 20161 }, { "epoch": 2.145350074483933, "grad_norm": 12.2890625, "learning_rate": 9.301378624218307e-06, "loss": 1.0809, "step": 20162 }, { "epoch": 2.1454564801021494, "grad_norm": 9.78125, "learning_rate": 9.301343092666288e-06, "loss": 0.8726, "step": 20163 }, { "epoch": 2.145562885720366, "grad_norm": 12.890625, "learning_rate": 9.30130756111427e-06, "loss": 0.9133, "step": 20164 }, { "epoch": 2.145669291338583, "grad_norm": 12.2734375, "learning_rate": 9.301272029562251e-06, "loss": 0.8496, "step": 20165 }, { "epoch": 2.1457756969567994, "grad_norm": 16.015625, "learning_rate": 9.301236498010235e-06, "loss": 1.0576, "step": 20166 }, { "epoch": 2.145882102575016, "grad_norm": 11.625, "learning_rate": 9.301200966458216e-06, "loss": 0.764, "step": 20167 }, { "epoch": 2.1459885081932324, "grad_norm": 14.609375, "learning_rate": 9.301165434906198e-06, "loss": 0.9438, "step": 20168 }, { "epoch": 2.1460949138114493, "grad_norm": 15.3046875, "learning_rate": 9.30112990335418e-06, "loss": 0.8647, "step": 20169 }, { "epoch": 2.146201319429666, "grad_norm": 13.015625, "learning_rate": 9.301094371802161e-06, "loss": 0.9282, "step": 20170 }, { "epoch": 2.1463077250478824, "grad_norm": 13.5390625, "learning_rate": 9.301058840250143e-06, "loss": 0.9331, "step": 20171 }, { "epoch": 2.1464141306660993, "grad_norm": 11.625, "learning_rate": 9.301023308698124e-06, "loss": 0.867, "step": 20172 }, { "epoch": 2.146520536284316, "grad_norm": 12.3515625, "learning_rate": 9.300987777146106e-06, "loss": 0.8026, "step": 20173 }, { "epoch": 2.1466269419025323, "grad_norm": 11.84375, "learning_rate": 9.300952245594088e-06, "loss": 0.9098, "step": 20174 }, { "epoch": 2.1467333475207493, "grad_norm": 9.8984375, "learning_rate": 9.30091671404207e-06, "loss": 0.7713, "step": 20175 }, { "epoch": 2.146839753138966, "grad_norm": 12.40625, "learning_rate": 9.300881182490052e-06, "loss": 0.8273, "step": 20176 }, { "epoch": 2.1469461587571823, "grad_norm": 12.96875, "learning_rate": 9.300845650938034e-06, "loss": 0.8845, "step": 20177 }, { "epoch": 2.147052564375399, "grad_norm": 16.515625, "learning_rate": 9.300810119386016e-06, "loss": 0.7799, "step": 20178 }, { "epoch": 2.1471589699936158, "grad_norm": 11.5625, "learning_rate": 9.300774587833997e-06, "loss": 0.8675, "step": 20179 }, { "epoch": 2.1472653756118323, "grad_norm": 12.34375, "learning_rate": 9.300739056281979e-06, "loss": 0.9392, "step": 20180 }, { "epoch": 2.147371781230049, "grad_norm": 11.125, "learning_rate": 9.300703524729962e-06, "loss": 0.6042, "step": 20181 }, { "epoch": 2.1474781868482657, "grad_norm": 15.6484375, "learning_rate": 9.300667993177944e-06, "loss": 0.8866, "step": 20182 }, { "epoch": 2.1475845924664823, "grad_norm": 9.96875, "learning_rate": 9.300632461625925e-06, "loss": 0.8642, "step": 20183 }, { "epoch": 2.1476909980846988, "grad_norm": 15.2109375, "learning_rate": 9.300596930073907e-06, "loss": 0.7279, "step": 20184 }, { "epoch": 2.1477974037029157, "grad_norm": 12.7109375, "learning_rate": 9.300561398521889e-06, "loss": 0.8622, "step": 20185 }, { "epoch": 2.1479038093211322, "grad_norm": 11.3203125, "learning_rate": 9.30052586696987e-06, "loss": 0.8496, "step": 20186 }, { "epoch": 2.1480102149393487, "grad_norm": 12.9453125, "learning_rate": 9.300490335417852e-06, "loss": 0.8174, "step": 20187 }, { "epoch": 2.1481166205575652, "grad_norm": 14.5, "learning_rate": 9.300454803865833e-06, "loss": 0.8674, "step": 20188 }, { "epoch": 2.148223026175782, "grad_norm": 10.1484375, "learning_rate": 9.300419272313815e-06, "loss": 0.831, "step": 20189 }, { "epoch": 2.1483294317939987, "grad_norm": 10.8046875, "learning_rate": 9.300383740761797e-06, "loss": 0.9263, "step": 20190 }, { "epoch": 2.148435837412215, "grad_norm": 13.46875, "learning_rate": 9.300348209209778e-06, "loss": 0.8698, "step": 20191 }, { "epoch": 2.148542243030432, "grad_norm": 20.96875, "learning_rate": 9.300312677657762e-06, "loss": 0.9397, "step": 20192 }, { "epoch": 2.1486486486486487, "grad_norm": 11.796875, "learning_rate": 9.300277146105743e-06, "loss": 0.9066, "step": 20193 }, { "epoch": 2.148755054266865, "grad_norm": 14.484375, "learning_rate": 9.300241614553725e-06, "loss": 0.8026, "step": 20194 }, { "epoch": 2.148861459885082, "grad_norm": 13.1796875, "learning_rate": 9.300206083001706e-06, "loss": 0.8515, "step": 20195 }, { "epoch": 2.1489678655032987, "grad_norm": 15.4375, "learning_rate": 9.300170551449688e-06, "loss": 1.0069, "step": 20196 }, { "epoch": 2.149074271121515, "grad_norm": 12.203125, "learning_rate": 9.30013501989767e-06, "loss": 0.9415, "step": 20197 }, { "epoch": 2.1491806767397317, "grad_norm": 11.140625, "learning_rate": 9.300099488345651e-06, "loss": 0.8536, "step": 20198 }, { "epoch": 2.1492870823579486, "grad_norm": 11.2890625, "learning_rate": 9.300063956793633e-06, "loss": 0.8453, "step": 20199 }, { "epoch": 2.149393487976165, "grad_norm": 13.421875, "learning_rate": 9.300028425241614e-06, "loss": 0.9019, "step": 20200 }, { "epoch": 2.1494998935943816, "grad_norm": 12.5390625, "learning_rate": 9.299992893689596e-06, "loss": 0.8953, "step": 20201 }, { "epoch": 2.1496062992125986, "grad_norm": 15.5, "learning_rate": 9.29995736213758e-06, "loss": 0.8545, "step": 20202 }, { "epoch": 2.149712704830815, "grad_norm": 12.453125, "learning_rate": 9.299921830585561e-06, "loss": 0.7562, "step": 20203 }, { "epoch": 2.1498191104490316, "grad_norm": 14.734375, "learning_rate": 9.299886299033543e-06, "loss": 1.2057, "step": 20204 }, { "epoch": 2.149925516067248, "grad_norm": 13.4609375, "learning_rate": 9.299850767481524e-06, "loss": 1.1297, "step": 20205 }, { "epoch": 2.150031921685465, "grad_norm": 11.9765625, "learning_rate": 9.299815235929506e-06, "loss": 0.8385, "step": 20206 }, { "epoch": 2.1501383273036816, "grad_norm": 11.515625, "learning_rate": 9.299779704377489e-06, "loss": 0.9499, "step": 20207 }, { "epoch": 2.150244732921898, "grad_norm": 13.0546875, "learning_rate": 9.29974417282547e-06, "loss": 0.8651, "step": 20208 }, { "epoch": 2.150351138540115, "grad_norm": 11.0546875, "learning_rate": 9.299708641273452e-06, "loss": 0.6859, "step": 20209 }, { "epoch": 2.1504575441583316, "grad_norm": 16.734375, "learning_rate": 9.299673109721434e-06, "loss": 0.7706, "step": 20210 }, { "epoch": 2.150563949776548, "grad_norm": 11.7734375, "learning_rate": 9.299637578169416e-06, "loss": 0.7741, "step": 20211 }, { "epoch": 2.150670355394765, "grad_norm": 12.8828125, "learning_rate": 9.299602046617397e-06, "loss": 0.8697, "step": 20212 }, { "epoch": 2.1507767610129815, "grad_norm": 13.5859375, "learning_rate": 9.299566515065379e-06, "loss": 0.8042, "step": 20213 }, { "epoch": 2.150883166631198, "grad_norm": 15.1328125, "learning_rate": 9.29953098351336e-06, "loss": 1.0284, "step": 20214 }, { "epoch": 2.150989572249415, "grad_norm": 11.765625, "learning_rate": 9.299495451961342e-06, "loss": 0.9772, "step": 20215 }, { "epoch": 2.1510959778676315, "grad_norm": 20.4375, "learning_rate": 9.299459920409324e-06, "loss": 0.9553, "step": 20216 }, { "epoch": 2.151202383485848, "grad_norm": 12.53125, "learning_rate": 9.299424388857307e-06, "loss": 0.9086, "step": 20217 }, { "epoch": 2.1513087891040645, "grad_norm": 12.9453125, "learning_rate": 9.299388857305288e-06, "loss": 0.8673, "step": 20218 }, { "epoch": 2.1514151947222815, "grad_norm": 15.2265625, "learning_rate": 9.29935332575327e-06, "loss": 0.8766, "step": 20219 }, { "epoch": 2.151521600340498, "grad_norm": 15.5703125, "learning_rate": 9.299317794201252e-06, "loss": 0.9806, "step": 20220 }, { "epoch": 2.1516280059587145, "grad_norm": 12.4140625, "learning_rate": 9.299282262649233e-06, "loss": 0.9343, "step": 20221 }, { "epoch": 2.1517344115769315, "grad_norm": 14.03125, "learning_rate": 9.299246731097215e-06, "loss": 0.9927, "step": 20222 }, { "epoch": 2.151840817195148, "grad_norm": 15.9140625, "learning_rate": 9.299211199545196e-06, "loss": 0.8404, "step": 20223 }, { "epoch": 2.1519472228133645, "grad_norm": 11.328125, "learning_rate": 9.299175667993178e-06, "loss": 0.8721, "step": 20224 }, { "epoch": 2.152053628431581, "grad_norm": 14.9765625, "learning_rate": 9.29914013644116e-06, "loss": 0.8567, "step": 20225 }, { "epoch": 2.152160034049798, "grad_norm": 15.4296875, "learning_rate": 9.299104604889141e-06, "loss": 0.8348, "step": 20226 }, { "epoch": 2.1522664396680145, "grad_norm": 15.9609375, "learning_rate": 9.299069073337123e-06, "loss": 0.9496, "step": 20227 }, { "epoch": 2.152372845286231, "grad_norm": 12.296875, "learning_rate": 9.299033541785106e-06, "loss": 0.9706, "step": 20228 }, { "epoch": 2.152479250904448, "grad_norm": 13.6953125, "learning_rate": 9.298998010233088e-06, "loss": 0.8514, "step": 20229 }, { "epoch": 2.1525856565226644, "grad_norm": 11.890625, "learning_rate": 9.29896247868107e-06, "loss": 0.9067, "step": 20230 }, { "epoch": 2.152692062140881, "grad_norm": 14.296875, "learning_rate": 9.298926947129051e-06, "loss": 0.9353, "step": 20231 }, { "epoch": 2.152798467759098, "grad_norm": 15.1875, "learning_rate": 9.298891415577033e-06, "loss": 0.9274, "step": 20232 }, { "epoch": 2.1529048733773144, "grad_norm": 13.2265625, "learning_rate": 9.298855884025016e-06, "loss": 0.8396, "step": 20233 }, { "epoch": 2.153011278995531, "grad_norm": 12.1171875, "learning_rate": 9.298820352472998e-06, "loss": 0.7386, "step": 20234 }, { "epoch": 2.1531176846137474, "grad_norm": 11.03125, "learning_rate": 9.298784820920979e-06, "loss": 0.8724, "step": 20235 }, { "epoch": 2.1532240902319644, "grad_norm": 13.515625, "learning_rate": 9.29874928936896e-06, "loss": 1.0362, "step": 20236 }, { "epoch": 2.153330495850181, "grad_norm": 11.0625, "learning_rate": 9.298713757816942e-06, "loss": 0.8833, "step": 20237 }, { "epoch": 2.1534369014683974, "grad_norm": 12.1875, "learning_rate": 9.298678226264924e-06, "loss": 0.8647, "step": 20238 }, { "epoch": 2.1535433070866143, "grad_norm": 14.6875, "learning_rate": 9.298642694712906e-06, "loss": 0.7743, "step": 20239 }, { "epoch": 2.153649712704831, "grad_norm": 18.25, "learning_rate": 9.298607163160887e-06, "loss": 0.8886, "step": 20240 }, { "epoch": 2.1537561183230474, "grad_norm": 15.8984375, "learning_rate": 9.298571631608869e-06, "loss": 0.8488, "step": 20241 }, { "epoch": 2.153862523941264, "grad_norm": 13.421875, "learning_rate": 9.29853610005685e-06, "loss": 0.974, "step": 20242 }, { "epoch": 2.153968929559481, "grad_norm": 15.0, "learning_rate": 9.298500568504834e-06, "loss": 0.8894, "step": 20243 }, { "epoch": 2.1540753351776973, "grad_norm": 8.859375, "learning_rate": 9.298465036952815e-06, "loss": 0.7912, "step": 20244 }, { "epoch": 2.154181740795914, "grad_norm": 10.390625, "learning_rate": 9.298429505400797e-06, "loss": 0.6972, "step": 20245 }, { "epoch": 2.154288146414131, "grad_norm": 12.9140625, "learning_rate": 9.298393973848779e-06, "loss": 0.8748, "step": 20246 }, { "epoch": 2.1543945520323473, "grad_norm": 11.03125, "learning_rate": 9.29835844229676e-06, "loss": 0.8468, "step": 20247 }, { "epoch": 2.154500957650564, "grad_norm": 11.328125, "learning_rate": 9.298322910744743e-06, "loss": 0.909, "step": 20248 }, { "epoch": 2.1546073632687808, "grad_norm": 12.125, "learning_rate": 9.298287379192723e-06, "loss": 0.8948, "step": 20249 }, { "epoch": 2.1547137688869973, "grad_norm": 12.25, "learning_rate": 9.298251847640705e-06, "loss": 1.0379, "step": 20250 }, { "epoch": 2.154820174505214, "grad_norm": 9.5390625, "learning_rate": 9.298216316088687e-06, "loss": 0.906, "step": 20251 }, { "epoch": 2.1549265801234307, "grad_norm": 17.28125, "learning_rate": 9.298180784536668e-06, "loss": 0.991, "step": 20252 }, { "epoch": 2.1550329857416473, "grad_norm": 10.4453125, "learning_rate": 9.298145252984651e-06, "loss": 0.852, "step": 20253 }, { "epoch": 2.1551393913598638, "grad_norm": 12.0859375, "learning_rate": 9.298109721432633e-06, "loss": 0.7633, "step": 20254 }, { "epoch": 2.1552457969780803, "grad_norm": 14.953125, "learning_rate": 9.298074189880615e-06, "loss": 0.8374, "step": 20255 }, { "epoch": 2.1553522025962972, "grad_norm": 11.8984375, "learning_rate": 9.298038658328596e-06, "loss": 0.6974, "step": 20256 }, { "epoch": 2.1554586082145137, "grad_norm": 9.7890625, "learning_rate": 9.298003126776578e-06, "loss": 0.8002, "step": 20257 }, { "epoch": 2.1555650138327302, "grad_norm": 12.9375, "learning_rate": 9.297967595224561e-06, "loss": 0.8937, "step": 20258 }, { "epoch": 2.155671419450947, "grad_norm": 15.9296875, "learning_rate": 9.297932063672543e-06, "loss": 1.0755, "step": 20259 }, { "epoch": 2.1557778250691637, "grad_norm": 12.359375, "learning_rate": 9.297896532120524e-06, "loss": 0.8956, "step": 20260 }, { "epoch": 2.15588423068738, "grad_norm": 17.359375, "learning_rate": 9.297861000568506e-06, "loss": 1.0485, "step": 20261 }, { "epoch": 2.1559906363055967, "grad_norm": 15.578125, "learning_rate": 9.297825469016488e-06, "loss": 0.9131, "step": 20262 }, { "epoch": 2.1560970419238137, "grad_norm": 12.0859375, "learning_rate": 9.29778993746447e-06, "loss": 0.8159, "step": 20263 }, { "epoch": 2.15620344754203, "grad_norm": 17.375, "learning_rate": 9.29775440591245e-06, "loss": 1.0123, "step": 20264 }, { "epoch": 2.1563098531602467, "grad_norm": 15.8203125, "learning_rate": 9.297718874360432e-06, "loss": 0.9214, "step": 20265 }, { "epoch": 2.1564162587784637, "grad_norm": 11.8671875, "learning_rate": 9.297683342808414e-06, "loss": 0.8713, "step": 20266 }, { "epoch": 2.15652266439668, "grad_norm": 14.671875, "learning_rate": 9.297647811256396e-06, "loss": 0.9026, "step": 20267 }, { "epoch": 2.1566290700148967, "grad_norm": 13.34375, "learning_rate": 9.297612279704377e-06, "loss": 0.8862, "step": 20268 }, { "epoch": 2.1567354756331136, "grad_norm": 15.0625, "learning_rate": 9.29757674815236e-06, "loss": 0.9183, "step": 20269 }, { "epoch": 2.15684188125133, "grad_norm": 11.1015625, "learning_rate": 9.297541216600342e-06, "loss": 0.8473, "step": 20270 }, { "epoch": 2.1569482868695466, "grad_norm": 12.7734375, "learning_rate": 9.297505685048324e-06, "loss": 0.7277, "step": 20271 }, { "epoch": 2.157054692487763, "grad_norm": 15.6328125, "learning_rate": 9.297470153496305e-06, "loss": 0.7252, "step": 20272 }, { "epoch": 2.15716109810598, "grad_norm": 20.796875, "learning_rate": 9.297434621944287e-06, "loss": 1.1893, "step": 20273 }, { "epoch": 2.1572675037241966, "grad_norm": 11.2578125, "learning_rate": 9.29739909039227e-06, "loss": 0.7931, "step": 20274 }, { "epoch": 2.157373909342413, "grad_norm": 15.28125, "learning_rate": 9.297363558840252e-06, "loss": 0.8656, "step": 20275 }, { "epoch": 2.15748031496063, "grad_norm": 11.890625, "learning_rate": 9.297328027288233e-06, "loss": 0.8343, "step": 20276 }, { "epoch": 2.1575867205788466, "grad_norm": 12.90625, "learning_rate": 9.297292495736215e-06, "loss": 0.7694, "step": 20277 }, { "epoch": 2.157693126197063, "grad_norm": 11.5546875, "learning_rate": 9.297256964184197e-06, "loss": 0.7752, "step": 20278 }, { "epoch": 2.1577995318152796, "grad_norm": 13.390625, "learning_rate": 9.297221432632178e-06, "loss": 0.8801, "step": 20279 }, { "epoch": 2.1579059374334966, "grad_norm": 12.1015625, "learning_rate": 9.29718590108016e-06, "loss": 0.9758, "step": 20280 }, { "epoch": 2.158012343051713, "grad_norm": 9.96875, "learning_rate": 9.297150369528142e-06, "loss": 0.691, "step": 20281 }, { "epoch": 2.1581187486699296, "grad_norm": 13.0625, "learning_rate": 9.297114837976123e-06, "loss": 0.8891, "step": 20282 }, { "epoch": 2.1582251542881465, "grad_norm": 11.6953125, "learning_rate": 9.297079306424105e-06, "loss": 0.9241, "step": 20283 }, { "epoch": 2.158331559906363, "grad_norm": 19.46875, "learning_rate": 9.297043774872088e-06, "loss": 0.9696, "step": 20284 }, { "epoch": 2.1584379655245796, "grad_norm": 13.1328125, "learning_rate": 9.29700824332007e-06, "loss": 0.8633, "step": 20285 }, { "epoch": 2.1585443711427965, "grad_norm": 18.5625, "learning_rate": 9.296972711768051e-06, "loss": 0.8575, "step": 20286 }, { "epoch": 2.158650776761013, "grad_norm": 19.375, "learning_rate": 9.296937180216033e-06, "loss": 1.0483, "step": 20287 }, { "epoch": 2.1587571823792295, "grad_norm": 12.34375, "learning_rate": 9.296901648664014e-06, "loss": 0.9419, "step": 20288 }, { "epoch": 2.1588635879974465, "grad_norm": 15.6953125, "learning_rate": 9.296866117111996e-06, "loss": 0.9732, "step": 20289 }, { "epoch": 2.158969993615663, "grad_norm": 10.0234375, "learning_rate": 9.296830585559978e-06, "loss": 0.9075, "step": 20290 }, { "epoch": 2.1590763992338795, "grad_norm": 14.96875, "learning_rate": 9.29679505400796e-06, "loss": 0.9685, "step": 20291 }, { "epoch": 2.159182804852096, "grad_norm": 14.53125, "learning_rate": 9.296759522455941e-06, "loss": 0.9022, "step": 20292 }, { "epoch": 2.159289210470313, "grad_norm": 12.3828125, "learning_rate": 9.296723990903922e-06, "loss": 0.9432, "step": 20293 }, { "epoch": 2.1593956160885295, "grad_norm": 17.109375, "learning_rate": 9.296688459351906e-06, "loss": 0.7435, "step": 20294 }, { "epoch": 2.159502021706746, "grad_norm": 9.6875, "learning_rate": 9.296652927799887e-06, "loss": 0.8865, "step": 20295 }, { "epoch": 2.159608427324963, "grad_norm": 14.3515625, "learning_rate": 9.296617396247869e-06, "loss": 0.7708, "step": 20296 }, { "epoch": 2.1597148329431795, "grad_norm": 12.65625, "learning_rate": 9.29658186469585e-06, "loss": 0.9921, "step": 20297 }, { "epoch": 2.159821238561396, "grad_norm": 15.171875, "learning_rate": 9.296546333143832e-06, "loss": 0.9878, "step": 20298 }, { "epoch": 2.1599276441796125, "grad_norm": 11.8359375, "learning_rate": 9.296510801591814e-06, "loss": 0.9279, "step": 20299 }, { "epoch": 2.1600340497978294, "grad_norm": 10.3359375, "learning_rate": 9.296475270039797e-06, "loss": 0.766, "step": 20300 }, { "epoch": 2.160140455416046, "grad_norm": 19.578125, "learning_rate": 9.296439738487779e-06, "loss": 0.8664, "step": 20301 }, { "epoch": 2.1602468610342624, "grad_norm": 14.5546875, "learning_rate": 9.29640420693576e-06, "loss": 0.918, "step": 20302 }, { "epoch": 2.1603532666524794, "grad_norm": 17.078125, "learning_rate": 9.296368675383742e-06, "loss": 1.0382, "step": 20303 }, { "epoch": 2.160459672270696, "grad_norm": 15.5859375, "learning_rate": 9.296333143831724e-06, "loss": 0.9666, "step": 20304 }, { "epoch": 2.1605660778889124, "grad_norm": 14.625, "learning_rate": 9.296297612279705e-06, "loss": 0.8473, "step": 20305 }, { "epoch": 2.1606724835071294, "grad_norm": 22.40625, "learning_rate": 9.296262080727687e-06, "loss": 0.9985, "step": 20306 }, { "epoch": 2.160778889125346, "grad_norm": 12.53125, "learning_rate": 9.296226549175668e-06, "loss": 0.9824, "step": 20307 }, { "epoch": 2.1608852947435624, "grad_norm": 13.8671875, "learning_rate": 9.29619101762365e-06, "loss": 1.0333, "step": 20308 }, { "epoch": 2.160991700361779, "grad_norm": 12.734375, "learning_rate": 9.296155486071632e-06, "loss": 0.852, "step": 20309 }, { "epoch": 2.161098105979996, "grad_norm": 15.3359375, "learning_rate": 9.296119954519615e-06, "loss": 0.8905, "step": 20310 }, { "epoch": 2.1612045115982124, "grad_norm": 15.9765625, "learning_rate": 9.296084422967596e-06, "loss": 0.8626, "step": 20311 }, { "epoch": 2.161310917216429, "grad_norm": 9.6171875, "learning_rate": 9.296048891415578e-06, "loss": 0.7608, "step": 20312 }, { "epoch": 2.161417322834646, "grad_norm": 15.4765625, "learning_rate": 9.29601335986356e-06, "loss": 0.8882, "step": 20313 }, { "epoch": 2.1615237284528623, "grad_norm": 12.5703125, "learning_rate": 9.295977828311541e-06, "loss": 1.0394, "step": 20314 }, { "epoch": 2.161630134071079, "grad_norm": 17.34375, "learning_rate": 9.295942296759523e-06, "loss": 1.1002, "step": 20315 }, { "epoch": 2.161736539689296, "grad_norm": 16.84375, "learning_rate": 9.295906765207505e-06, "loss": 0.7678, "step": 20316 }, { "epoch": 2.1618429453075123, "grad_norm": 15.578125, "learning_rate": 9.295871233655486e-06, "loss": 0.9903, "step": 20317 }, { "epoch": 2.161949350925729, "grad_norm": 12.390625, "learning_rate": 9.295835702103468e-06, "loss": 0.9959, "step": 20318 }, { "epoch": 2.1620557565439453, "grad_norm": 16.421875, "learning_rate": 9.29580017055145e-06, "loss": 0.7154, "step": 20319 }, { "epoch": 2.1621621621621623, "grad_norm": 15.875, "learning_rate": 9.295764638999433e-06, "loss": 0.9743, "step": 20320 }, { "epoch": 2.162268567780379, "grad_norm": 18.0625, "learning_rate": 9.295729107447414e-06, "loss": 0.9775, "step": 20321 }, { "epoch": 2.1623749733985953, "grad_norm": 11.390625, "learning_rate": 9.295693575895396e-06, "loss": 0.7934, "step": 20322 }, { "epoch": 2.1624813790168123, "grad_norm": 15.140625, "learning_rate": 9.295658044343377e-06, "loss": 0.8443, "step": 20323 }, { "epoch": 2.1625877846350288, "grad_norm": 10.796875, "learning_rate": 9.295622512791359e-06, "loss": 0.8732, "step": 20324 }, { "epoch": 2.1626941902532453, "grad_norm": 16.109375, "learning_rate": 9.295586981239342e-06, "loss": 0.8089, "step": 20325 }, { "epoch": 2.1628005958714622, "grad_norm": 15.5390625, "learning_rate": 9.295551449687324e-06, "loss": 0.8837, "step": 20326 }, { "epoch": 2.1629070014896787, "grad_norm": 15.109375, "learning_rate": 9.295515918135306e-06, "loss": 1.016, "step": 20327 }, { "epoch": 2.1630134071078952, "grad_norm": 11.4453125, "learning_rate": 9.295480386583287e-06, "loss": 0.9206, "step": 20328 }, { "epoch": 2.1631198127261118, "grad_norm": 9.46875, "learning_rate": 9.295444855031269e-06, "loss": 0.7229, "step": 20329 }, { "epoch": 2.1632262183443287, "grad_norm": 13.4375, "learning_rate": 9.29540932347925e-06, "loss": 0.8675, "step": 20330 }, { "epoch": 2.163332623962545, "grad_norm": 13.3125, "learning_rate": 9.295373791927232e-06, "loss": 0.9984, "step": 20331 }, { "epoch": 2.1634390295807617, "grad_norm": 11.453125, "learning_rate": 9.295338260375214e-06, "loss": 0.9875, "step": 20332 }, { "epoch": 2.1635454351989787, "grad_norm": 12.5625, "learning_rate": 9.295302728823195e-06, "loss": 0.8184, "step": 20333 }, { "epoch": 2.163651840817195, "grad_norm": 10.6328125, "learning_rate": 9.295267197271177e-06, "loss": 0.9263, "step": 20334 }, { "epoch": 2.1637582464354117, "grad_norm": 14.546875, "learning_rate": 9.29523166571916e-06, "loss": 0.9836, "step": 20335 }, { "epoch": 2.163864652053628, "grad_norm": 11.7890625, "learning_rate": 9.295196134167142e-06, "loss": 0.9422, "step": 20336 }, { "epoch": 2.163971057671845, "grad_norm": 12.3359375, "learning_rate": 9.295160602615123e-06, "loss": 0.9107, "step": 20337 }, { "epoch": 2.1640774632900617, "grad_norm": 13.75, "learning_rate": 9.295125071063105e-06, "loss": 0.9353, "step": 20338 }, { "epoch": 2.164183868908278, "grad_norm": 13.9609375, "learning_rate": 9.295089539511087e-06, "loss": 0.8419, "step": 20339 }, { "epoch": 2.164290274526495, "grad_norm": 13.5390625, "learning_rate": 9.295054007959068e-06, "loss": 0.8302, "step": 20340 }, { "epoch": 2.1643966801447116, "grad_norm": 19.0625, "learning_rate": 9.29501847640705e-06, "loss": 1.3176, "step": 20341 }, { "epoch": 2.164503085762928, "grad_norm": 14.125, "learning_rate": 9.294982944855031e-06, "loss": 0.7575, "step": 20342 }, { "epoch": 2.164609491381145, "grad_norm": 15.609375, "learning_rate": 9.294947413303013e-06, "loss": 0.8629, "step": 20343 }, { "epoch": 2.1647158969993616, "grad_norm": 12.2265625, "learning_rate": 9.294911881750995e-06, "loss": 0.8649, "step": 20344 }, { "epoch": 2.164822302617578, "grad_norm": 10.875, "learning_rate": 9.294876350198976e-06, "loss": 0.8234, "step": 20345 }, { "epoch": 2.164928708235795, "grad_norm": 12.7890625, "learning_rate": 9.29484081864696e-06, "loss": 0.9124, "step": 20346 }, { "epoch": 2.1650351138540116, "grad_norm": 15.0, "learning_rate": 9.294805287094941e-06, "loss": 1.0422, "step": 20347 }, { "epoch": 2.165141519472228, "grad_norm": 11.109375, "learning_rate": 9.294769755542923e-06, "loss": 0.8504, "step": 20348 }, { "epoch": 2.1652479250904446, "grad_norm": 18.484375, "learning_rate": 9.294734223990904e-06, "loss": 0.9772, "step": 20349 }, { "epoch": 2.1653543307086616, "grad_norm": 15.78125, "learning_rate": 9.294698692438886e-06, "loss": 0.9878, "step": 20350 }, { "epoch": 2.165460736326878, "grad_norm": 13.4453125, "learning_rate": 9.29466316088687e-06, "loss": 1.0489, "step": 20351 }, { "epoch": 2.1655671419450946, "grad_norm": 11.34375, "learning_rate": 9.29462762933485e-06, "loss": 1.0342, "step": 20352 }, { "epoch": 2.1656735475633115, "grad_norm": 16.71875, "learning_rate": 9.294592097782832e-06, "loss": 0.7929, "step": 20353 }, { "epoch": 2.165779953181528, "grad_norm": 15.2109375, "learning_rate": 9.294556566230814e-06, "loss": 0.9729, "step": 20354 }, { "epoch": 2.1658863587997446, "grad_norm": 9.859375, "learning_rate": 9.294521034678796e-06, "loss": 0.7868, "step": 20355 }, { "epoch": 2.165992764417961, "grad_norm": 14.9609375, "learning_rate": 9.294485503126777e-06, "loss": 0.7059, "step": 20356 }, { "epoch": 2.166099170036178, "grad_norm": 10.3203125, "learning_rate": 9.294449971574759e-06, "loss": 0.7438, "step": 20357 }, { "epoch": 2.1662055756543945, "grad_norm": 14.5390625, "learning_rate": 9.29441444002274e-06, "loss": 0.8297, "step": 20358 }, { "epoch": 2.166311981272611, "grad_norm": 13.484375, "learning_rate": 9.294378908470722e-06, "loss": 0.793, "step": 20359 }, { "epoch": 2.166418386890828, "grad_norm": 13.9296875, "learning_rate": 9.294343376918704e-06, "loss": 0.9716, "step": 20360 }, { "epoch": 2.1665247925090445, "grad_norm": 12.28125, "learning_rate": 9.294307845366687e-06, "loss": 0.8489, "step": 20361 }, { "epoch": 2.166631198127261, "grad_norm": 13.1953125, "learning_rate": 9.294272313814669e-06, "loss": 0.8135, "step": 20362 }, { "epoch": 2.166737603745478, "grad_norm": 17.515625, "learning_rate": 9.29423678226265e-06, "loss": 0.9343, "step": 20363 }, { "epoch": 2.1668440093636945, "grad_norm": 12.6953125, "learning_rate": 9.294201250710632e-06, "loss": 0.8578, "step": 20364 }, { "epoch": 2.166950414981911, "grad_norm": 19.921875, "learning_rate": 9.294165719158613e-06, "loss": 1.0192, "step": 20365 }, { "epoch": 2.1670568206001275, "grad_norm": 14.140625, "learning_rate": 9.294130187606597e-06, "loss": 0.879, "step": 20366 }, { "epoch": 2.1671632262183445, "grad_norm": 12.375, "learning_rate": 9.294094656054578e-06, "loss": 0.8383, "step": 20367 }, { "epoch": 2.167269631836561, "grad_norm": 11.5, "learning_rate": 9.29405912450256e-06, "loss": 0.7972, "step": 20368 }, { "epoch": 2.1673760374547775, "grad_norm": 15.0546875, "learning_rate": 9.29402359295054e-06, "loss": 0.8311, "step": 20369 }, { "epoch": 2.1674824430729944, "grad_norm": 13.46875, "learning_rate": 9.293988061398521e-06, "loss": 0.7302, "step": 20370 }, { "epoch": 2.167588848691211, "grad_norm": 15.234375, "learning_rate": 9.293952529846505e-06, "loss": 0.8411, "step": 20371 }, { "epoch": 2.1676952543094274, "grad_norm": 11.0, "learning_rate": 9.293916998294486e-06, "loss": 0.808, "step": 20372 }, { "epoch": 2.167801659927644, "grad_norm": 18.90625, "learning_rate": 9.293881466742468e-06, "loss": 0.866, "step": 20373 }, { "epoch": 2.167908065545861, "grad_norm": 12.375, "learning_rate": 9.29384593519045e-06, "loss": 0.6326, "step": 20374 }, { "epoch": 2.1680144711640774, "grad_norm": 13.9609375, "learning_rate": 9.293810403638431e-06, "loss": 0.9701, "step": 20375 }, { "epoch": 2.168120876782294, "grad_norm": 14.0078125, "learning_rate": 9.293774872086414e-06, "loss": 0.9231, "step": 20376 }, { "epoch": 2.168227282400511, "grad_norm": 11.1875, "learning_rate": 9.293739340534396e-06, "loss": 0.6827, "step": 20377 }, { "epoch": 2.1683336880187274, "grad_norm": 10.1640625, "learning_rate": 9.293703808982378e-06, "loss": 0.8059, "step": 20378 }, { "epoch": 2.168440093636944, "grad_norm": 16.796875, "learning_rate": 9.29366827743036e-06, "loss": 0.9535, "step": 20379 }, { "epoch": 2.168546499255161, "grad_norm": 12.1171875, "learning_rate": 9.293632745878341e-06, "loss": 0.8461, "step": 20380 }, { "epoch": 2.1686529048733774, "grad_norm": 14.5703125, "learning_rate": 9.293597214326323e-06, "loss": 0.8866, "step": 20381 }, { "epoch": 2.168759310491594, "grad_norm": 14.625, "learning_rate": 9.293561682774304e-06, "loss": 1.0102, "step": 20382 }, { "epoch": 2.168865716109811, "grad_norm": 10.375, "learning_rate": 9.293526151222286e-06, "loss": 0.7736, "step": 20383 }, { "epoch": 2.1689721217280273, "grad_norm": 14.859375, "learning_rate": 9.293490619670267e-06, "loss": 1.1352, "step": 20384 }, { "epoch": 2.169078527346244, "grad_norm": 14.7578125, "learning_rate": 9.293455088118249e-06, "loss": 0.8197, "step": 20385 }, { "epoch": 2.1691849329644604, "grad_norm": 11.1796875, "learning_rate": 9.29341955656623e-06, "loss": 0.7234, "step": 20386 }, { "epoch": 2.1692913385826773, "grad_norm": 14.71875, "learning_rate": 9.293384025014214e-06, "loss": 1.0331, "step": 20387 }, { "epoch": 2.169397744200894, "grad_norm": 9.90625, "learning_rate": 9.293348493462195e-06, "loss": 0.8486, "step": 20388 }, { "epoch": 2.1695041498191103, "grad_norm": 18.453125, "learning_rate": 9.293312961910177e-06, "loss": 0.8489, "step": 20389 }, { "epoch": 2.1696105554373273, "grad_norm": 15.171875, "learning_rate": 9.293277430358159e-06, "loss": 0.8976, "step": 20390 }, { "epoch": 2.169716961055544, "grad_norm": 10.9609375, "learning_rate": 9.29324189880614e-06, "loss": 0.678, "step": 20391 }, { "epoch": 2.1698233666737603, "grad_norm": 13.296875, "learning_rate": 9.293206367254124e-06, "loss": 0.8547, "step": 20392 }, { "epoch": 2.169929772291977, "grad_norm": 10.21875, "learning_rate": 9.293170835702105e-06, "loss": 0.8007, "step": 20393 }, { "epoch": 2.1700361779101938, "grad_norm": 13.640625, "learning_rate": 9.293135304150087e-06, "loss": 1.0094, "step": 20394 }, { "epoch": 2.1701425835284103, "grad_norm": 11.1640625, "learning_rate": 9.293099772598068e-06, "loss": 0.8541, "step": 20395 }, { "epoch": 2.170248989146627, "grad_norm": 11.1640625, "learning_rate": 9.29306424104605e-06, "loss": 0.9437, "step": 20396 }, { "epoch": 2.1703553947648437, "grad_norm": 15.390625, "learning_rate": 9.293028709494032e-06, "loss": 0.8781, "step": 20397 }, { "epoch": 2.1704618003830602, "grad_norm": 14.1796875, "learning_rate": 9.292993177942013e-06, "loss": 0.7698, "step": 20398 }, { "epoch": 2.1705682060012768, "grad_norm": 22.5, "learning_rate": 9.292957646389995e-06, "loss": 0.7925, "step": 20399 }, { "epoch": 2.1706746116194937, "grad_norm": 12.5078125, "learning_rate": 9.292922114837976e-06, "loss": 0.9177, "step": 20400 }, { "epoch": 2.17078101723771, "grad_norm": 9.640625, "learning_rate": 9.292886583285958e-06, "loss": 0.767, "step": 20401 }, { "epoch": 2.1708874228559267, "grad_norm": 12.5546875, "learning_rate": 9.292851051733941e-06, "loss": 0.9751, "step": 20402 }, { "epoch": 2.1709938284741432, "grad_norm": 9.9921875, "learning_rate": 9.292815520181923e-06, "loss": 0.8806, "step": 20403 }, { "epoch": 2.17110023409236, "grad_norm": 12.40625, "learning_rate": 9.292779988629905e-06, "loss": 0.8087, "step": 20404 }, { "epoch": 2.1712066397105767, "grad_norm": 15.1015625, "learning_rate": 9.292744457077886e-06, "loss": 0.8819, "step": 20405 }, { "epoch": 2.171313045328793, "grad_norm": 12.8359375, "learning_rate": 9.292708925525868e-06, "loss": 1.0016, "step": 20406 }, { "epoch": 2.17141945094701, "grad_norm": 13.9140625, "learning_rate": 9.29267339397385e-06, "loss": 0.8417, "step": 20407 }, { "epoch": 2.1715258565652267, "grad_norm": 10.2109375, "learning_rate": 9.292637862421831e-06, "loss": 0.7226, "step": 20408 }, { "epoch": 2.171632262183443, "grad_norm": 11.65625, "learning_rate": 9.292602330869813e-06, "loss": 0.8023, "step": 20409 }, { "epoch": 2.1717386678016597, "grad_norm": 14.3046875, "learning_rate": 9.292566799317794e-06, "loss": 0.9536, "step": 20410 }, { "epoch": 2.1718450734198766, "grad_norm": 24.390625, "learning_rate": 9.292531267765776e-06, "loss": 1.2422, "step": 20411 }, { "epoch": 2.171951479038093, "grad_norm": 16.71875, "learning_rate": 9.292495736213759e-06, "loss": 0.8497, "step": 20412 }, { "epoch": 2.1720578846563097, "grad_norm": 14.2265625, "learning_rate": 9.29246020466174e-06, "loss": 0.8102, "step": 20413 }, { "epoch": 2.1721642902745266, "grad_norm": 13.2109375, "learning_rate": 9.292424673109722e-06, "loss": 1.0084, "step": 20414 }, { "epoch": 2.172270695892743, "grad_norm": 11.859375, "learning_rate": 9.292389141557704e-06, "loss": 0.7565, "step": 20415 }, { "epoch": 2.1723771015109596, "grad_norm": 18.78125, "learning_rate": 9.292353610005686e-06, "loss": 0.7889, "step": 20416 }, { "epoch": 2.1724835071291766, "grad_norm": 11.8984375, "learning_rate": 9.292318078453667e-06, "loss": 0.8071, "step": 20417 }, { "epoch": 2.172589912747393, "grad_norm": 13.390625, "learning_rate": 9.29228254690165e-06, "loss": 0.877, "step": 20418 }, { "epoch": 2.1726963183656096, "grad_norm": 18.484375, "learning_rate": 9.292247015349632e-06, "loss": 0.8854, "step": 20419 }, { "epoch": 2.1728027239838266, "grad_norm": 15.40625, "learning_rate": 9.292211483797614e-06, "loss": 0.8607, "step": 20420 }, { "epoch": 2.172909129602043, "grad_norm": 11.9140625, "learning_rate": 9.292175952245595e-06, "loss": 0.8669, "step": 20421 }, { "epoch": 2.1730155352202596, "grad_norm": 16.40625, "learning_rate": 9.292140420693577e-06, "loss": 0.7673, "step": 20422 }, { "epoch": 2.173121940838476, "grad_norm": 13.453125, "learning_rate": 9.292104889141558e-06, "loss": 0.8957, "step": 20423 }, { "epoch": 2.173228346456693, "grad_norm": 15.65625, "learning_rate": 9.29206935758954e-06, "loss": 1.044, "step": 20424 }, { "epoch": 2.1733347520749096, "grad_norm": 15.546875, "learning_rate": 9.292033826037522e-06, "loss": 0.7885, "step": 20425 }, { "epoch": 2.173441157693126, "grad_norm": 14.8984375, "learning_rate": 9.291998294485503e-06, "loss": 1.0132, "step": 20426 }, { "epoch": 2.173547563311343, "grad_norm": 12.046875, "learning_rate": 9.291962762933485e-06, "loss": 0.7141, "step": 20427 }, { "epoch": 2.1736539689295595, "grad_norm": 13.8671875, "learning_rate": 9.291927231381468e-06, "loss": 0.8561, "step": 20428 }, { "epoch": 2.173760374547776, "grad_norm": 14.4609375, "learning_rate": 9.29189169982945e-06, "loss": 0.927, "step": 20429 }, { "epoch": 2.1738667801659926, "grad_norm": 13.1953125, "learning_rate": 9.291856168277431e-06, "loss": 0.7677, "step": 20430 }, { "epoch": 2.1739731857842095, "grad_norm": 12.7578125, "learning_rate": 9.291820636725413e-06, "loss": 1.1516, "step": 20431 }, { "epoch": 2.174079591402426, "grad_norm": 12.28125, "learning_rate": 9.291785105173395e-06, "loss": 0.8659, "step": 20432 }, { "epoch": 2.1741859970206425, "grad_norm": 12.0078125, "learning_rate": 9.291749573621376e-06, "loss": 0.8449, "step": 20433 }, { "epoch": 2.1742924026388595, "grad_norm": 34.9375, "learning_rate": 9.291714042069358e-06, "loss": 0.6489, "step": 20434 }, { "epoch": 2.174398808257076, "grad_norm": 9.2109375, "learning_rate": 9.29167851051734e-06, "loss": 0.7633, "step": 20435 }, { "epoch": 2.1745052138752925, "grad_norm": 12.3203125, "learning_rate": 9.291642978965321e-06, "loss": 0.8489, "step": 20436 }, { "epoch": 2.1746116194935095, "grad_norm": 13.671875, "learning_rate": 9.291607447413303e-06, "loss": 0.8491, "step": 20437 }, { "epoch": 2.174718025111726, "grad_norm": 14.1875, "learning_rate": 9.291571915861286e-06, "loss": 0.9077, "step": 20438 }, { "epoch": 2.1748244307299425, "grad_norm": 15.75, "learning_rate": 9.291536384309268e-06, "loss": 0.8236, "step": 20439 }, { "epoch": 2.174930836348159, "grad_norm": 15.078125, "learning_rate": 9.29150085275725e-06, "loss": 1.0097, "step": 20440 }, { "epoch": 2.175037241966376, "grad_norm": 11.421875, "learning_rate": 9.29146532120523e-06, "loss": 0.9006, "step": 20441 }, { "epoch": 2.1751436475845924, "grad_norm": 13.8125, "learning_rate": 9.291429789653212e-06, "loss": 0.9708, "step": 20442 }, { "epoch": 2.175250053202809, "grad_norm": 11.9296875, "learning_rate": 9.291394258101196e-06, "loss": 0.8256, "step": 20443 }, { "epoch": 2.175356458821026, "grad_norm": 12.21875, "learning_rate": 9.291358726549177e-06, "loss": 0.6496, "step": 20444 }, { "epoch": 2.1754628644392424, "grad_norm": 9.203125, "learning_rate": 9.291323194997159e-06, "loss": 0.7711, "step": 20445 }, { "epoch": 2.175569270057459, "grad_norm": 13.3828125, "learning_rate": 9.29128766344514e-06, "loss": 0.7798, "step": 20446 }, { "epoch": 2.175675675675676, "grad_norm": 11.6796875, "learning_rate": 9.291252131893122e-06, "loss": 0.7873, "step": 20447 }, { "epoch": 2.1757820812938924, "grad_norm": 15.1953125, "learning_rate": 9.291216600341104e-06, "loss": 0.9872, "step": 20448 }, { "epoch": 2.175888486912109, "grad_norm": 11.0546875, "learning_rate": 9.291181068789085e-06, "loss": 0.7112, "step": 20449 }, { "epoch": 2.1759948925303254, "grad_norm": 16.21875, "learning_rate": 9.291145537237067e-06, "loss": 1.2746, "step": 20450 }, { "epoch": 2.1761012981485424, "grad_norm": 15.6796875, "learning_rate": 9.291110005685049e-06, "loss": 0.9399, "step": 20451 }, { "epoch": 2.176207703766759, "grad_norm": 11.5078125, "learning_rate": 9.29107447413303e-06, "loss": 0.851, "step": 20452 }, { "epoch": 2.1763141093849754, "grad_norm": 9.9921875, "learning_rate": 9.291038942581013e-06, "loss": 0.8784, "step": 20453 }, { "epoch": 2.1764205150031923, "grad_norm": 12.515625, "learning_rate": 9.291003411028995e-06, "loss": 0.856, "step": 20454 }, { "epoch": 2.176526920621409, "grad_norm": 11.8828125, "learning_rate": 9.290967879476977e-06, "loss": 1.0337, "step": 20455 }, { "epoch": 2.1766333262396254, "grad_norm": 21.453125, "learning_rate": 9.290932347924958e-06, "loss": 0.8881, "step": 20456 }, { "epoch": 2.1767397318578423, "grad_norm": 10.8984375, "learning_rate": 9.29089681637294e-06, "loss": 1.0089, "step": 20457 }, { "epoch": 2.176846137476059, "grad_norm": 9.75, "learning_rate": 9.290861284820921e-06, "loss": 0.743, "step": 20458 }, { "epoch": 2.1769525430942753, "grad_norm": 9.9375, "learning_rate": 9.290825753268903e-06, "loss": 0.8488, "step": 20459 }, { "epoch": 2.177058948712492, "grad_norm": 14.4453125, "learning_rate": 9.290790221716885e-06, "loss": 0.8528, "step": 20460 }, { "epoch": 2.177165354330709, "grad_norm": 10.0234375, "learning_rate": 9.290754690164866e-06, "loss": 0.7919, "step": 20461 }, { "epoch": 2.1772717599489253, "grad_norm": 14.875, "learning_rate": 9.290719158612848e-06, "loss": 1.0839, "step": 20462 }, { "epoch": 2.177378165567142, "grad_norm": 12.34375, "learning_rate": 9.29068362706083e-06, "loss": 0.9167, "step": 20463 }, { "epoch": 2.1774845711853588, "grad_norm": 11.4453125, "learning_rate": 9.290648095508813e-06, "loss": 0.9435, "step": 20464 }, { "epoch": 2.1775909768035753, "grad_norm": 18.5, "learning_rate": 9.290612563956794e-06, "loss": 1.0637, "step": 20465 }, { "epoch": 2.177697382421792, "grad_norm": 14.34375, "learning_rate": 9.290577032404776e-06, "loss": 0.9722, "step": 20466 }, { "epoch": 2.1778037880400083, "grad_norm": 12.515625, "learning_rate": 9.290541500852758e-06, "loss": 0.8094, "step": 20467 }, { "epoch": 2.1779101936582252, "grad_norm": 9.625, "learning_rate": 9.29050596930074e-06, "loss": 0.7484, "step": 20468 }, { "epoch": 2.1780165992764418, "grad_norm": 14.546875, "learning_rate": 9.290470437748723e-06, "loss": 0.8348, "step": 20469 }, { "epoch": 2.1781230048946583, "grad_norm": 13.3984375, "learning_rate": 9.290434906196704e-06, "loss": 0.753, "step": 20470 }, { "epoch": 2.178229410512875, "grad_norm": 13.7109375, "learning_rate": 9.290399374644686e-06, "loss": 0.8841, "step": 20471 }, { "epoch": 2.1783358161310917, "grad_norm": 11.421875, "learning_rate": 9.290363843092667e-06, "loss": 0.8229, "step": 20472 }, { "epoch": 2.1784422217493082, "grad_norm": 13.484375, "learning_rate": 9.290328311540649e-06, "loss": 1.1066, "step": 20473 }, { "epoch": 2.178548627367525, "grad_norm": 10.4609375, "learning_rate": 9.29029277998863e-06, "loss": 0.7444, "step": 20474 }, { "epoch": 2.1786550329857417, "grad_norm": 11.359375, "learning_rate": 9.290257248436612e-06, "loss": 0.7767, "step": 20475 }, { "epoch": 2.178761438603958, "grad_norm": 11.7578125, "learning_rate": 9.290221716884594e-06, "loss": 0.9981, "step": 20476 }, { "epoch": 2.178867844222175, "grad_norm": 12.6796875, "learning_rate": 9.290186185332575e-06, "loss": 0.9011, "step": 20477 }, { "epoch": 2.1789742498403917, "grad_norm": 16.828125, "learning_rate": 9.290150653780557e-06, "loss": 0.9138, "step": 20478 }, { "epoch": 2.179080655458608, "grad_norm": 10.21875, "learning_rate": 9.29011512222854e-06, "loss": 0.7656, "step": 20479 }, { "epoch": 2.1791870610768247, "grad_norm": 13.046875, "learning_rate": 9.290079590676522e-06, "loss": 1.0613, "step": 20480 }, { "epoch": 2.1792934666950416, "grad_norm": 13.96875, "learning_rate": 9.290044059124504e-06, "loss": 0.8521, "step": 20481 }, { "epoch": 2.179399872313258, "grad_norm": 13.5078125, "learning_rate": 9.290008527572485e-06, "loss": 1.0325, "step": 20482 }, { "epoch": 2.1795062779314747, "grad_norm": 12.703125, "learning_rate": 9.289972996020467e-06, "loss": 0.9439, "step": 20483 }, { "epoch": 2.1796126835496916, "grad_norm": 13.234375, "learning_rate": 9.28993746446845e-06, "loss": 0.898, "step": 20484 }, { "epoch": 2.179719089167908, "grad_norm": 13.5859375, "learning_rate": 9.289901932916432e-06, "loss": 0.8667, "step": 20485 }, { "epoch": 2.1798254947861246, "grad_norm": 10.453125, "learning_rate": 9.289866401364413e-06, "loss": 0.8839, "step": 20486 }, { "epoch": 2.179931900404341, "grad_norm": 11.5078125, "learning_rate": 9.289830869812395e-06, "loss": 0.7506, "step": 20487 }, { "epoch": 2.180038306022558, "grad_norm": 11.1015625, "learning_rate": 9.289795338260376e-06, "loss": 0.9198, "step": 20488 }, { "epoch": 2.1801447116407746, "grad_norm": 14.6953125, "learning_rate": 9.289759806708358e-06, "loss": 0.9459, "step": 20489 }, { "epoch": 2.180251117258991, "grad_norm": 10.890625, "learning_rate": 9.28972427515634e-06, "loss": 0.839, "step": 20490 }, { "epoch": 2.180357522877208, "grad_norm": 11.8984375, "learning_rate": 9.289688743604321e-06, "loss": 0.876, "step": 20491 }, { "epoch": 2.1804639284954246, "grad_norm": 12.28125, "learning_rate": 9.289653212052303e-06, "loss": 0.9972, "step": 20492 }, { "epoch": 2.180570334113641, "grad_norm": 13.8359375, "learning_rate": 9.289617680500284e-06, "loss": 0.9015, "step": 20493 }, { "epoch": 2.180676739731858, "grad_norm": 14.046875, "learning_rate": 9.289582148948266e-06, "loss": 0.8723, "step": 20494 }, { "epoch": 2.1807831453500746, "grad_norm": 12.875, "learning_rate": 9.28954661739625e-06, "loss": 0.7722, "step": 20495 }, { "epoch": 2.180889550968291, "grad_norm": 11.4921875, "learning_rate": 9.289511085844231e-06, "loss": 0.8546, "step": 20496 }, { "epoch": 2.1809959565865076, "grad_norm": 15.375, "learning_rate": 9.289475554292213e-06, "loss": 0.8481, "step": 20497 }, { "epoch": 2.1811023622047245, "grad_norm": 13.4765625, "learning_rate": 9.289440022740194e-06, "loss": 1.0251, "step": 20498 }, { "epoch": 2.181208767822941, "grad_norm": 14.171875, "learning_rate": 9.289404491188176e-06, "loss": 1.0996, "step": 20499 }, { "epoch": 2.1813151734411576, "grad_norm": 11.3984375, "learning_rate": 9.289368959636157e-06, "loss": 0.8155, "step": 20500 }, { "epoch": 2.1814215790593745, "grad_norm": 11.234375, "learning_rate": 9.289333428084139e-06, "loss": 0.9091, "step": 20501 }, { "epoch": 2.181527984677591, "grad_norm": 14.1953125, "learning_rate": 9.28929789653212e-06, "loss": 0.9387, "step": 20502 }, { "epoch": 2.1816343902958075, "grad_norm": 11.1640625, "learning_rate": 9.289262364980102e-06, "loss": 0.9303, "step": 20503 }, { "epoch": 2.181740795914024, "grad_norm": 17.59375, "learning_rate": 9.289226833428084e-06, "loss": 0.9702, "step": 20504 }, { "epoch": 2.181847201532241, "grad_norm": 12.4140625, "learning_rate": 9.289191301876067e-06, "loss": 0.8331, "step": 20505 }, { "epoch": 2.1819536071504575, "grad_norm": 10.9296875, "learning_rate": 9.289155770324049e-06, "loss": 0.8826, "step": 20506 }, { "epoch": 2.182060012768674, "grad_norm": 16.375, "learning_rate": 9.28912023877203e-06, "loss": 0.8725, "step": 20507 }, { "epoch": 2.182166418386891, "grad_norm": 13.859375, "learning_rate": 9.289084707220012e-06, "loss": 1.0763, "step": 20508 }, { "epoch": 2.1822728240051075, "grad_norm": 17.515625, "learning_rate": 9.289049175667994e-06, "loss": 0.9381, "step": 20509 }, { "epoch": 2.182379229623324, "grad_norm": 11.109375, "learning_rate": 9.289013644115977e-06, "loss": 0.8671, "step": 20510 }, { "epoch": 2.182485635241541, "grad_norm": 13.5703125, "learning_rate": 9.288978112563958e-06, "loss": 0.8199, "step": 20511 }, { "epoch": 2.1825920408597574, "grad_norm": 9.3515625, "learning_rate": 9.28894258101194e-06, "loss": 0.7631, "step": 20512 }, { "epoch": 2.182698446477974, "grad_norm": 13.953125, "learning_rate": 9.288907049459922e-06, "loss": 0.9692, "step": 20513 }, { "epoch": 2.182804852096191, "grad_norm": 12.171875, "learning_rate": 9.288871517907903e-06, "loss": 0.9674, "step": 20514 }, { "epoch": 2.1829112577144074, "grad_norm": 13.6953125, "learning_rate": 9.288835986355885e-06, "loss": 0.8723, "step": 20515 }, { "epoch": 2.183017663332624, "grad_norm": 13.4609375, "learning_rate": 9.288800454803867e-06, "loss": 0.8625, "step": 20516 }, { "epoch": 2.1831240689508404, "grad_norm": 15.0234375, "learning_rate": 9.288764923251848e-06, "loss": 0.8408, "step": 20517 }, { "epoch": 2.1832304745690574, "grad_norm": 14.21875, "learning_rate": 9.28872939169983e-06, "loss": 0.9847, "step": 20518 }, { "epoch": 2.183336880187274, "grad_norm": 11.0546875, "learning_rate": 9.288693860147811e-06, "loss": 0.9779, "step": 20519 }, { "epoch": 2.1834432858054904, "grad_norm": 12.3125, "learning_rate": 9.288658328595795e-06, "loss": 0.875, "step": 20520 }, { "epoch": 2.1835496914237074, "grad_norm": 13.1015625, "learning_rate": 9.288622797043776e-06, "loss": 0.9775, "step": 20521 }, { "epoch": 2.183656097041924, "grad_norm": 8.4921875, "learning_rate": 9.288587265491758e-06, "loss": 0.734, "step": 20522 }, { "epoch": 2.1837625026601404, "grad_norm": 8.6171875, "learning_rate": 9.28855173393974e-06, "loss": 0.8335, "step": 20523 }, { "epoch": 2.183868908278357, "grad_norm": 11.2109375, "learning_rate": 9.288516202387721e-06, "loss": 1.0375, "step": 20524 }, { "epoch": 2.183975313896574, "grad_norm": 10.8125, "learning_rate": 9.288480670835703e-06, "loss": 1.0095, "step": 20525 }, { "epoch": 2.1840817195147904, "grad_norm": 13.234375, "learning_rate": 9.288445139283684e-06, "loss": 0.8584, "step": 20526 }, { "epoch": 2.184188125133007, "grad_norm": 15.453125, "learning_rate": 9.288409607731666e-06, "loss": 0.8048, "step": 20527 }, { "epoch": 2.184294530751224, "grad_norm": 13.8984375, "learning_rate": 9.288374076179647e-06, "loss": 0.9513, "step": 20528 }, { "epoch": 2.1844009363694403, "grad_norm": 12.640625, "learning_rate": 9.288338544627629e-06, "loss": 0.9319, "step": 20529 }, { "epoch": 2.184507341987657, "grad_norm": 9.4453125, "learning_rate": 9.288303013075612e-06, "loss": 0.8133, "step": 20530 }, { "epoch": 2.184613747605874, "grad_norm": 17.359375, "learning_rate": 9.288267481523594e-06, "loss": 0.9595, "step": 20531 }, { "epoch": 2.1847201532240903, "grad_norm": 11.4921875, "learning_rate": 9.288231949971576e-06, "loss": 0.863, "step": 20532 }, { "epoch": 2.184826558842307, "grad_norm": 15.7265625, "learning_rate": 9.288196418419557e-06, "loss": 0.9258, "step": 20533 }, { "epoch": 2.1849329644605233, "grad_norm": 10.2890625, "learning_rate": 9.288160886867539e-06, "loss": 0.858, "step": 20534 }, { "epoch": 2.1850393700787403, "grad_norm": 11.5390625, "learning_rate": 9.28812535531552e-06, "loss": 0.887, "step": 20535 }, { "epoch": 2.185145775696957, "grad_norm": 17.84375, "learning_rate": 9.288089823763504e-06, "loss": 0.9647, "step": 20536 }, { "epoch": 2.1852521813151733, "grad_norm": 13.9609375, "learning_rate": 9.288054292211485e-06, "loss": 0.9037, "step": 20537 }, { "epoch": 2.1853585869333902, "grad_norm": 17.375, "learning_rate": 9.288018760659467e-06, "loss": 0.8428, "step": 20538 }, { "epoch": 2.1854649925516068, "grad_norm": 11.328125, "learning_rate": 9.287983229107449e-06, "loss": 0.7984, "step": 20539 }, { "epoch": 2.1855713981698233, "grad_norm": 14.9375, "learning_rate": 9.28794769755543e-06, "loss": 0.8154, "step": 20540 }, { "epoch": 2.1856778037880398, "grad_norm": 12.3515625, "learning_rate": 9.287912166003412e-06, "loss": 0.7465, "step": 20541 }, { "epoch": 2.1857842094062567, "grad_norm": 10.609375, "learning_rate": 9.287876634451393e-06, "loss": 0.76, "step": 20542 }, { "epoch": 2.1858906150244732, "grad_norm": 13.8359375, "learning_rate": 9.287841102899375e-06, "loss": 0.992, "step": 20543 }, { "epoch": 2.1859970206426897, "grad_norm": 13.5703125, "learning_rate": 9.287805571347357e-06, "loss": 1.066, "step": 20544 }, { "epoch": 2.1861034262609067, "grad_norm": 16.125, "learning_rate": 9.287770039795338e-06, "loss": 0.8079, "step": 20545 }, { "epoch": 2.186209831879123, "grad_norm": 10.96875, "learning_rate": 9.287734508243321e-06, "loss": 0.8335, "step": 20546 }, { "epoch": 2.1863162374973397, "grad_norm": 12.9453125, "learning_rate": 9.287698976691303e-06, "loss": 0.9757, "step": 20547 }, { "epoch": 2.1864226431155567, "grad_norm": 11.1640625, "learning_rate": 9.287663445139285e-06, "loss": 0.8233, "step": 20548 }, { "epoch": 2.186529048733773, "grad_norm": 13.7421875, "learning_rate": 9.287627913587266e-06, "loss": 0.8831, "step": 20549 }, { "epoch": 2.1866354543519897, "grad_norm": 11.6328125, "learning_rate": 9.287592382035248e-06, "loss": 0.8992, "step": 20550 }, { "epoch": 2.1867418599702066, "grad_norm": 13.2578125, "learning_rate": 9.28755685048323e-06, "loss": 1.1258, "step": 20551 }, { "epoch": 2.186848265588423, "grad_norm": 13.2734375, "learning_rate": 9.287521318931211e-06, "loss": 0.8725, "step": 20552 }, { "epoch": 2.1869546712066397, "grad_norm": 12.0390625, "learning_rate": 9.287485787379193e-06, "loss": 0.8314, "step": 20553 }, { "epoch": 2.187061076824856, "grad_norm": 9.6953125, "learning_rate": 9.287450255827174e-06, "loss": 0.7741, "step": 20554 }, { "epoch": 2.187167482443073, "grad_norm": 14.3828125, "learning_rate": 9.287414724275156e-06, "loss": 0.8424, "step": 20555 }, { "epoch": 2.1872738880612896, "grad_norm": 11.1484375, "learning_rate": 9.28737919272314e-06, "loss": 0.7368, "step": 20556 }, { "epoch": 2.187380293679506, "grad_norm": 12.6640625, "learning_rate": 9.287343661171121e-06, "loss": 0.8006, "step": 20557 }, { "epoch": 2.187486699297723, "grad_norm": 33.25, "learning_rate": 9.287308129619102e-06, "loss": 1.1195, "step": 20558 }, { "epoch": 2.1875931049159396, "grad_norm": 12.65625, "learning_rate": 9.287272598067084e-06, "loss": 0.8442, "step": 20559 }, { "epoch": 2.187699510534156, "grad_norm": 14.4921875, "learning_rate": 9.287237066515066e-06, "loss": 0.8651, "step": 20560 }, { "epoch": 2.1878059161523726, "grad_norm": 10.5, "learning_rate": 9.287201534963049e-06, "loss": 0.7806, "step": 20561 }, { "epoch": 2.1879123217705896, "grad_norm": 14.5234375, "learning_rate": 9.28716600341103e-06, "loss": 0.9132, "step": 20562 }, { "epoch": 2.188018727388806, "grad_norm": 14.265625, "learning_rate": 9.287130471859012e-06, "loss": 0.8894, "step": 20563 }, { "epoch": 2.1881251330070226, "grad_norm": 13.6953125, "learning_rate": 9.287094940306994e-06, "loss": 0.871, "step": 20564 }, { "epoch": 2.1882315386252396, "grad_norm": 13.2109375, "learning_rate": 9.287059408754975e-06, "loss": 0.9996, "step": 20565 }, { "epoch": 2.188337944243456, "grad_norm": 11.59375, "learning_rate": 9.287023877202957e-06, "loss": 0.9786, "step": 20566 }, { "epoch": 2.1884443498616726, "grad_norm": 14.90625, "learning_rate": 9.286988345650939e-06, "loss": 0.6778, "step": 20567 }, { "epoch": 2.1885507554798895, "grad_norm": 9.984375, "learning_rate": 9.28695281409892e-06, "loss": 0.7655, "step": 20568 }, { "epoch": 2.188657161098106, "grad_norm": 15.7109375, "learning_rate": 9.286917282546902e-06, "loss": 1.0745, "step": 20569 }, { "epoch": 2.1887635667163226, "grad_norm": 11.1171875, "learning_rate": 9.286881750994883e-06, "loss": 0.8925, "step": 20570 }, { "epoch": 2.188869972334539, "grad_norm": 13.6171875, "learning_rate": 9.286846219442867e-06, "loss": 0.7939, "step": 20571 }, { "epoch": 2.188976377952756, "grad_norm": 11.203125, "learning_rate": 9.286810687890848e-06, "loss": 0.7661, "step": 20572 }, { "epoch": 2.1890827835709725, "grad_norm": 10.5390625, "learning_rate": 9.28677515633883e-06, "loss": 0.7868, "step": 20573 }, { "epoch": 2.189189189189189, "grad_norm": 11.59375, "learning_rate": 9.286739624786812e-06, "loss": 0.8724, "step": 20574 }, { "epoch": 2.189295594807406, "grad_norm": 14.5546875, "learning_rate": 9.286704093234793e-06, "loss": 0.9656, "step": 20575 }, { "epoch": 2.1894020004256225, "grad_norm": 11.7421875, "learning_rate": 9.286668561682775e-06, "loss": 0.7582, "step": 20576 }, { "epoch": 2.189508406043839, "grad_norm": 10.703125, "learning_rate": 9.286633030130756e-06, "loss": 0.8142, "step": 20577 }, { "epoch": 2.189614811662056, "grad_norm": 9.515625, "learning_rate": 9.286597498578738e-06, "loss": 0.8844, "step": 20578 }, { "epoch": 2.1897212172802725, "grad_norm": 15.234375, "learning_rate": 9.28656196702672e-06, "loss": 1.004, "step": 20579 }, { "epoch": 2.189827622898489, "grad_norm": 10.703125, "learning_rate": 9.286526435474701e-06, "loss": 0.8093, "step": 20580 }, { "epoch": 2.1899340285167055, "grad_norm": 10.15625, "learning_rate": 9.286490903922683e-06, "loss": 0.9497, "step": 20581 }, { "epoch": 2.1900404341349224, "grad_norm": 15.0546875, "learning_rate": 9.286455372370666e-06, "loss": 0.9882, "step": 20582 }, { "epoch": 2.190146839753139, "grad_norm": 11.7578125, "learning_rate": 9.286419840818648e-06, "loss": 0.8137, "step": 20583 }, { "epoch": 2.1902532453713555, "grad_norm": 11.5234375, "learning_rate": 9.28638430926663e-06, "loss": 0.8517, "step": 20584 }, { "epoch": 2.1903596509895724, "grad_norm": 11.796875, "learning_rate": 9.286348777714611e-06, "loss": 0.9906, "step": 20585 }, { "epoch": 2.190466056607789, "grad_norm": 16.015625, "learning_rate": 9.286313246162593e-06, "loss": 0.7561, "step": 20586 }, { "epoch": 2.1905724622260054, "grad_norm": 13.984375, "learning_rate": 9.286277714610576e-06, "loss": 0.8919, "step": 20587 }, { "epoch": 2.1906788678442224, "grad_norm": 15.609375, "learning_rate": 9.286242183058557e-06, "loss": 0.9272, "step": 20588 }, { "epoch": 2.190785273462439, "grad_norm": 15.7265625, "learning_rate": 9.286206651506539e-06, "loss": 0.9228, "step": 20589 }, { "epoch": 2.1908916790806554, "grad_norm": 10.5703125, "learning_rate": 9.28617111995452e-06, "loss": 0.6913, "step": 20590 }, { "epoch": 2.190998084698872, "grad_norm": 11.2734375, "learning_rate": 9.286135588402502e-06, "loss": 0.6948, "step": 20591 }, { "epoch": 2.191104490317089, "grad_norm": 12.125, "learning_rate": 9.286100056850484e-06, "loss": 0.8392, "step": 20592 }, { "epoch": 2.1912108959353054, "grad_norm": 18.59375, "learning_rate": 9.286064525298465e-06, "loss": 1.1423, "step": 20593 }, { "epoch": 2.191317301553522, "grad_norm": 14.015625, "learning_rate": 9.286028993746447e-06, "loss": 0.8276, "step": 20594 }, { "epoch": 2.191423707171739, "grad_norm": 13.046875, "learning_rate": 9.285993462194429e-06, "loss": 0.8941, "step": 20595 }, { "epoch": 2.1915301127899554, "grad_norm": 16.859375, "learning_rate": 9.28595793064241e-06, "loss": 1.1211, "step": 20596 }, { "epoch": 2.191636518408172, "grad_norm": 13.5, "learning_rate": 9.285922399090394e-06, "loss": 0.82, "step": 20597 }, { "epoch": 2.1917429240263884, "grad_norm": 13.09375, "learning_rate": 9.285886867538375e-06, "loss": 1.0406, "step": 20598 }, { "epoch": 2.1918493296446053, "grad_norm": 13.453125, "learning_rate": 9.285851335986357e-06, "loss": 0.9595, "step": 20599 }, { "epoch": 2.191955735262822, "grad_norm": 12.8203125, "learning_rate": 9.285815804434338e-06, "loss": 0.9366, "step": 20600 }, { "epoch": 2.1920621408810383, "grad_norm": 14.640625, "learning_rate": 9.28578027288232e-06, "loss": 0.8895, "step": 20601 }, { "epoch": 2.1921685464992553, "grad_norm": 14.2890625, "learning_rate": 9.285744741330303e-06, "loss": 1.0015, "step": 20602 }, { "epoch": 2.192274952117472, "grad_norm": 15.78125, "learning_rate": 9.285709209778285e-06, "loss": 0.9395, "step": 20603 }, { "epoch": 2.1923813577356883, "grad_norm": 16.5, "learning_rate": 9.285673678226267e-06, "loss": 0.8437, "step": 20604 }, { "epoch": 2.1924877633539053, "grad_norm": 10.6953125, "learning_rate": 9.285638146674248e-06, "loss": 0.8132, "step": 20605 }, { "epoch": 2.192594168972122, "grad_norm": 10.6171875, "learning_rate": 9.28560261512223e-06, "loss": 0.6874, "step": 20606 }, { "epoch": 2.1927005745903383, "grad_norm": 11.3984375, "learning_rate": 9.285567083570211e-06, "loss": 0.8796, "step": 20607 }, { "epoch": 2.1928069802085552, "grad_norm": 14.1328125, "learning_rate": 9.285531552018193e-06, "loss": 0.962, "step": 20608 }, { "epoch": 2.1929133858267718, "grad_norm": 13.4453125, "learning_rate": 9.285496020466175e-06, "loss": 0.9829, "step": 20609 }, { "epoch": 2.1930197914449883, "grad_norm": 14.0703125, "learning_rate": 9.285460488914156e-06, "loss": 0.6604, "step": 20610 }, { "epoch": 2.1931261970632048, "grad_norm": 19.34375, "learning_rate": 9.285424957362138e-06, "loss": 0.9695, "step": 20611 }, { "epoch": 2.1932326026814217, "grad_norm": 12.046875, "learning_rate": 9.28538942581012e-06, "loss": 0.7133, "step": 20612 }, { "epoch": 2.1933390082996382, "grad_norm": 20.046875, "learning_rate": 9.285353894258103e-06, "loss": 0.9902, "step": 20613 }, { "epoch": 2.1934454139178547, "grad_norm": 13.6953125, "learning_rate": 9.285318362706084e-06, "loss": 0.8443, "step": 20614 }, { "epoch": 2.1935518195360717, "grad_norm": 11.734375, "learning_rate": 9.285282831154066e-06, "loss": 0.9562, "step": 20615 }, { "epoch": 2.193658225154288, "grad_norm": 10.8125, "learning_rate": 9.285247299602048e-06, "loss": 0.8153, "step": 20616 }, { "epoch": 2.1937646307725047, "grad_norm": 9.7421875, "learning_rate": 9.285211768050029e-06, "loss": 0.6809, "step": 20617 }, { "epoch": 2.1938710363907212, "grad_norm": 14.1328125, "learning_rate": 9.28517623649801e-06, "loss": 0.8295, "step": 20618 }, { "epoch": 2.193977442008938, "grad_norm": 12.6953125, "learning_rate": 9.285140704945992e-06, "loss": 1.0189, "step": 20619 }, { "epoch": 2.1940838476271547, "grad_norm": 15.5625, "learning_rate": 9.285105173393974e-06, "loss": 0.7829, "step": 20620 }, { "epoch": 2.194190253245371, "grad_norm": 13.875, "learning_rate": 9.285069641841956e-06, "loss": 1.0373, "step": 20621 }, { "epoch": 2.194296658863588, "grad_norm": 12.4765625, "learning_rate": 9.285034110289937e-06, "loss": 0.9315, "step": 20622 }, { "epoch": 2.1944030644818047, "grad_norm": 22.15625, "learning_rate": 9.28499857873792e-06, "loss": 0.9046, "step": 20623 }, { "epoch": 2.194509470100021, "grad_norm": 11.8828125, "learning_rate": 9.284963047185902e-06, "loss": 0.7667, "step": 20624 }, { "epoch": 2.194615875718238, "grad_norm": 14.4921875, "learning_rate": 9.284927515633884e-06, "loss": 0.8498, "step": 20625 }, { "epoch": 2.1947222813364546, "grad_norm": 11.1015625, "learning_rate": 9.284891984081865e-06, "loss": 0.8415, "step": 20626 }, { "epoch": 2.194828686954671, "grad_norm": 16.296875, "learning_rate": 9.284856452529847e-06, "loss": 1.0763, "step": 20627 }, { "epoch": 2.1949350925728877, "grad_norm": 14.5390625, "learning_rate": 9.28482092097783e-06, "loss": 0.9494, "step": 20628 }, { "epoch": 2.1950414981911046, "grad_norm": 11.5078125, "learning_rate": 9.284785389425812e-06, "loss": 0.8216, "step": 20629 }, { "epoch": 2.195147903809321, "grad_norm": 10.015625, "learning_rate": 9.284749857873793e-06, "loss": 0.8443, "step": 20630 }, { "epoch": 2.1952543094275376, "grad_norm": 12.53125, "learning_rate": 9.284714326321775e-06, "loss": 0.8437, "step": 20631 }, { "epoch": 2.1953607150457546, "grad_norm": 16.78125, "learning_rate": 9.284678794769757e-06, "loss": 0.823, "step": 20632 }, { "epoch": 2.195467120663971, "grad_norm": 10.65625, "learning_rate": 9.284643263217738e-06, "loss": 0.8402, "step": 20633 }, { "epoch": 2.1955735262821876, "grad_norm": 9.984375, "learning_rate": 9.28460773166572e-06, "loss": 0.9171, "step": 20634 }, { "epoch": 2.195679931900404, "grad_norm": 16.5625, "learning_rate": 9.284572200113701e-06, "loss": 0.8318, "step": 20635 }, { "epoch": 2.195786337518621, "grad_norm": 12.4296875, "learning_rate": 9.284536668561683e-06, "loss": 0.9403, "step": 20636 }, { "epoch": 2.1958927431368376, "grad_norm": 12.2734375, "learning_rate": 9.284501137009665e-06, "loss": 0.7061, "step": 20637 }, { "epoch": 2.195999148755054, "grad_norm": 14.25, "learning_rate": 9.284465605457648e-06, "loss": 1.0221, "step": 20638 }, { "epoch": 2.196105554373271, "grad_norm": 15.375, "learning_rate": 9.28443007390563e-06, "loss": 0.9306, "step": 20639 }, { "epoch": 2.1962119599914876, "grad_norm": 12.8125, "learning_rate": 9.284394542353611e-06, "loss": 0.8166, "step": 20640 }, { "epoch": 2.196318365609704, "grad_norm": 11.890625, "learning_rate": 9.284359010801593e-06, "loss": 0.7605, "step": 20641 }, { "epoch": 2.196424771227921, "grad_norm": 12.1015625, "learning_rate": 9.284323479249574e-06, "loss": 0.8664, "step": 20642 }, { "epoch": 2.1965311768461375, "grad_norm": 11.3046875, "learning_rate": 9.284287947697556e-06, "loss": 0.8902, "step": 20643 }, { "epoch": 2.196637582464354, "grad_norm": 13.8671875, "learning_rate": 9.284252416145538e-06, "loss": 0.9746, "step": 20644 }, { "epoch": 2.196743988082571, "grad_norm": 11.546875, "learning_rate": 9.28421688459352e-06, "loss": 0.976, "step": 20645 }, { "epoch": 2.1968503937007875, "grad_norm": 13.296875, "learning_rate": 9.2841813530415e-06, "loss": 0.7786, "step": 20646 }, { "epoch": 2.196956799319004, "grad_norm": 9.578125, "learning_rate": 9.284145821489482e-06, "loss": 0.7878, "step": 20647 }, { "epoch": 2.1970632049372205, "grad_norm": 13.84375, "learning_rate": 9.284110289937466e-06, "loss": 1.0453, "step": 20648 }, { "epoch": 2.1971696105554375, "grad_norm": 11.484375, "learning_rate": 9.284074758385447e-06, "loss": 0.82, "step": 20649 }, { "epoch": 2.197276016173654, "grad_norm": 13.359375, "learning_rate": 9.284039226833429e-06, "loss": 0.8109, "step": 20650 }, { "epoch": 2.1973824217918705, "grad_norm": 15.890625, "learning_rate": 9.28400369528141e-06, "loss": 0.8941, "step": 20651 }, { "epoch": 2.1974888274100874, "grad_norm": 10.8671875, "learning_rate": 9.283968163729392e-06, "loss": 0.7392, "step": 20652 }, { "epoch": 2.197595233028304, "grad_norm": 15.125, "learning_rate": 9.283932632177374e-06, "loss": 0.8652, "step": 20653 }, { "epoch": 2.1977016386465205, "grad_norm": 15.9921875, "learning_rate": 9.283897100625357e-06, "loss": 0.7947, "step": 20654 }, { "epoch": 2.197808044264737, "grad_norm": 11.140625, "learning_rate": 9.283861569073339e-06, "loss": 0.873, "step": 20655 }, { "epoch": 2.197914449882954, "grad_norm": 9.9375, "learning_rate": 9.28382603752132e-06, "loss": 0.8353, "step": 20656 }, { "epoch": 2.1980208555011704, "grad_norm": 13.7265625, "learning_rate": 9.283790505969302e-06, "loss": 0.9197, "step": 20657 }, { "epoch": 2.198127261119387, "grad_norm": 16.015625, "learning_rate": 9.283754974417283e-06, "loss": 0.9517, "step": 20658 }, { "epoch": 2.198233666737604, "grad_norm": 12.3359375, "learning_rate": 9.283719442865265e-06, "loss": 0.8319, "step": 20659 }, { "epoch": 2.1983400723558204, "grad_norm": 14.2890625, "learning_rate": 9.283683911313247e-06, "loss": 0.9914, "step": 20660 }, { "epoch": 2.198446477974037, "grad_norm": 16.96875, "learning_rate": 9.283648379761228e-06, "loss": 0.8413, "step": 20661 }, { "epoch": 2.198552883592254, "grad_norm": 20.015625, "learning_rate": 9.28361284820921e-06, "loss": 0.7907, "step": 20662 }, { "epoch": 2.1986592892104704, "grad_norm": 15.9375, "learning_rate": 9.283577316657191e-06, "loss": 0.8313, "step": 20663 }, { "epoch": 2.198765694828687, "grad_norm": 12.015625, "learning_rate": 9.283541785105175e-06, "loss": 0.9574, "step": 20664 }, { "epoch": 2.1988721004469034, "grad_norm": 13.421875, "learning_rate": 9.283506253553156e-06, "loss": 1.1978, "step": 20665 }, { "epoch": 2.1989785060651204, "grad_norm": 15.2890625, "learning_rate": 9.283470722001138e-06, "loss": 1.0808, "step": 20666 }, { "epoch": 2.199084911683337, "grad_norm": 14.09375, "learning_rate": 9.28343519044912e-06, "loss": 0.8741, "step": 20667 }, { "epoch": 2.1991913173015534, "grad_norm": 12.890625, "learning_rate": 9.283399658897101e-06, "loss": 0.9387, "step": 20668 }, { "epoch": 2.1992977229197703, "grad_norm": 13.671875, "learning_rate": 9.283364127345083e-06, "loss": 1.06, "step": 20669 }, { "epoch": 2.199404128537987, "grad_norm": 17.25, "learning_rate": 9.283328595793064e-06, "loss": 0.8879, "step": 20670 }, { "epoch": 2.1995105341562033, "grad_norm": 10.96875, "learning_rate": 9.283293064241046e-06, "loss": 0.9388, "step": 20671 }, { "epoch": 2.19961693977442, "grad_norm": 16.21875, "learning_rate": 9.283257532689028e-06, "loss": 1.0467, "step": 20672 }, { "epoch": 2.199723345392637, "grad_norm": 13.8984375, "learning_rate": 9.28322200113701e-06, "loss": 1.2963, "step": 20673 }, { "epoch": 2.1998297510108533, "grad_norm": 13.765625, "learning_rate": 9.283186469584993e-06, "loss": 0.8826, "step": 20674 }, { "epoch": 2.19993615662907, "grad_norm": 13.0234375, "learning_rate": 9.283150938032974e-06, "loss": 1.0063, "step": 20675 }, { "epoch": 2.200042562247287, "grad_norm": 17.640625, "learning_rate": 9.283115406480956e-06, "loss": 0.8751, "step": 20676 }, { "epoch": 2.2001489678655033, "grad_norm": 13.0, "learning_rate": 9.283079874928937e-06, "loss": 1.034, "step": 20677 }, { "epoch": 2.20025537348372, "grad_norm": 12.6640625, "learning_rate": 9.283044343376919e-06, "loss": 0.9842, "step": 20678 }, { "epoch": 2.2003617791019368, "grad_norm": 14.5234375, "learning_rate": 9.283008811824902e-06, "loss": 0.8729, "step": 20679 }, { "epoch": 2.2004681847201533, "grad_norm": 11.6953125, "learning_rate": 9.282973280272884e-06, "loss": 0.9833, "step": 20680 }, { "epoch": 2.2005745903383698, "grad_norm": 13.3359375, "learning_rate": 9.282937748720866e-06, "loss": 0.9938, "step": 20681 }, { "epoch": 2.2006809959565867, "grad_norm": 20.34375, "learning_rate": 9.282902217168847e-06, "loss": 0.9087, "step": 20682 }, { "epoch": 2.2007874015748032, "grad_norm": 11.90625, "learning_rate": 9.282866685616829e-06, "loss": 0.8688, "step": 20683 }, { "epoch": 2.2008938071930197, "grad_norm": 17.28125, "learning_rate": 9.28283115406481e-06, "loss": 0.9642, "step": 20684 }, { "epoch": 2.2010002128112363, "grad_norm": 12.9296875, "learning_rate": 9.282795622512792e-06, "loss": 0.8481, "step": 20685 }, { "epoch": 2.201106618429453, "grad_norm": 10.2265625, "learning_rate": 9.282760090960774e-06, "loss": 0.8448, "step": 20686 }, { "epoch": 2.2012130240476697, "grad_norm": 8.921875, "learning_rate": 9.282724559408755e-06, "loss": 0.7296, "step": 20687 }, { "epoch": 2.2013194296658862, "grad_norm": 15.2109375, "learning_rate": 9.282689027856737e-06, "loss": 0.8622, "step": 20688 }, { "epoch": 2.201425835284103, "grad_norm": 17.046875, "learning_rate": 9.28265349630472e-06, "loss": 1.0333, "step": 20689 }, { "epoch": 2.2015322409023197, "grad_norm": 17.765625, "learning_rate": 9.282617964752702e-06, "loss": 0.9798, "step": 20690 }, { "epoch": 2.201638646520536, "grad_norm": 14.6484375, "learning_rate": 9.282582433200683e-06, "loss": 0.8453, "step": 20691 }, { "epoch": 2.2017450521387527, "grad_norm": 14.375, "learning_rate": 9.282546901648665e-06, "loss": 0.805, "step": 20692 }, { "epoch": 2.2018514577569697, "grad_norm": 10.75, "learning_rate": 9.282511370096646e-06, "loss": 0.8274, "step": 20693 }, { "epoch": 2.201957863375186, "grad_norm": 12.0859375, "learning_rate": 9.282475838544628e-06, "loss": 0.8341, "step": 20694 }, { "epoch": 2.2020642689934027, "grad_norm": 15.7421875, "learning_rate": 9.282440306992611e-06, "loss": 0.9615, "step": 20695 }, { "epoch": 2.2021706746116196, "grad_norm": 11.3046875, "learning_rate": 9.282404775440593e-06, "loss": 0.9056, "step": 20696 }, { "epoch": 2.202277080229836, "grad_norm": 18.296875, "learning_rate": 9.282369243888573e-06, "loss": 0.8956, "step": 20697 }, { "epoch": 2.2023834858480527, "grad_norm": 11.7421875, "learning_rate": 9.282333712336555e-06, "loss": 0.9123, "step": 20698 }, { "epoch": 2.2024898914662696, "grad_norm": 8.984375, "learning_rate": 9.282298180784536e-06, "loss": 0.8449, "step": 20699 }, { "epoch": 2.202596297084486, "grad_norm": 16.953125, "learning_rate": 9.28226264923252e-06, "loss": 0.8305, "step": 20700 }, { "epoch": 2.2027027027027026, "grad_norm": 13.8671875, "learning_rate": 9.282227117680501e-06, "loss": 0.9822, "step": 20701 }, { "epoch": 2.202809108320919, "grad_norm": 12.171875, "learning_rate": 9.282191586128483e-06, "loss": 1.0043, "step": 20702 }, { "epoch": 2.202915513939136, "grad_norm": 12.0390625, "learning_rate": 9.282156054576464e-06, "loss": 0.8488, "step": 20703 }, { "epoch": 2.2030219195573526, "grad_norm": 7.15234375, "learning_rate": 9.282120523024446e-06, "loss": 0.5669, "step": 20704 }, { "epoch": 2.203128325175569, "grad_norm": 10.9140625, "learning_rate": 9.282084991472429e-06, "loss": 0.8532, "step": 20705 }, { "epoch": 2.203234730793786, "grad_norm": 10.703125, "learning_rate": 9.28204945992041e-06, "loss": 0.7541, "step": 20706 }, { "epoch": 2.2033411364120026, "grad_norm": 20.015625, "learning_rate": 9.282013928368392e-06, "loss": 1.0021, "step": 20707 }, { "epoch": 2.203447542030219, "grad_norm": 12.65625, "learning_rate": 9.281978396816374e-06, "loss": 0.9079, "step": 20708 }, { "epoch": 2.203553947648436, "grad_norm": 12.296875, "learning_rate": 9.281942865264356e-06, "loss": 0.8303, "step": 20709 }, { "epoch": 2.2036603532666525, "grad_norm": 12.3671875, "learning_rate": 9.281907333712337e-06, "loss": 0.7988, "step": 20710 }, { "epoch": 2.203766758884869, "grad_norm": 13.65625, "learning_rate": 9.281871802160319e-06, "loss": 0.8565, "step": 20711 }, { "epoch": 2.2038731645030856, "grad_norm": 9.3984375, "learning_rate": 9.2818362706083e-06, "loss": 0.7359, "step": 20712 }, { "epoch": 2.2039795701213025, "grad_norm": 9.7578125, "learning_rate": 9.281800739056282e-06, "loss": 0.7458, "step": 20713 }, { "epoch": 2.204085975739519, "grad_norm": 14.3125, "learning_rate": 9.281765207504264e-06, "loss": 1.1614, "step": 20714 }, { "epoch": 2.2041923813577355, "grad_norm": 11.28125, "learning_rate": 9.281729675952247e-06, "loss": 0.7662, "step": 20715 }, { "epoch": 2.2042987869759525, "grad_norm": 10.6328125, "learning_rate": 9.281694144400229e-06, "loss": 0.849, "step": 20716 }, { "epoch": 2.204405192594169, "grad_norm": 10.6796875, "learning_rate": 9.28165861284821e-06, "loss": 0.707, "step": 20717 }, { "epoch": 2.2045115982123855, "grad_norm": 11.6328125, "learning_rate": 9.281623081296192e-06, "loss": 0.7917, "step": 20718 }, { "epoch": 2.2046180038306025, "grad_norm": 18.625, "learning_rate": 9.281587549744173e-06, "loss": 1.1003, "step": 20719 }, { "epoch": 2.204724409448819, "grad_norm": 11.2734375, "learning_rate": 9.281552018192157e-06, "loss": 0.9398, "step": 20720 }, { "epoch": 2.2048308150670355, "grad_norm": 13.0390625, "learning_rate": 9.281516486640138e-06, "loss": 0.8683, "step": 20721 }, { "epoch": 2.204937220685252, "grad_norm": 9.515625, "learning_rate": 9.28148095508812e-06, "loss": 0.8479, "step": 20722 }, { "epoch": 2.205043626303469, "grad_norm": 12.296875, "learning_rate": 9.281445423536101e-06, "loss": 0.7859, "step": 20723 }, { "epoch": 2.2051500319216855, "grad_norm": 13.96875, "learning_rate": 9.281409891984083e-06, "loss": 0.9961, "step": 20724 }, { "epoch": 2.205256437539902, "grad_norm": 17.796875, "learning_rate": 9.281374360432065e-06, "loss": 0.9326, "step": 20725 }, { "epoch": 2.205362843158119, "grad_norm": 17.8125, "learning_rate": 9.281338828880046e-06, "loss": 0.9071, "step": 20726 }, { "epoch": 2.2054692487763354, "grad_norm": 11.6015625, "learning_rate": 9.281303297328028e-06, "loss": 0.7965, "step": 20727 }, { "epoch": 2.205575654394552, "grad_norm": 13.109375, "learning_rate": 9.28126776577601e-06, "loss": 0.9937, "step": 20728 }, { "epoch": 2.2056820600127685, "grad_norm": 10.484375, "learning_rate": 9.281232234223991e-06, "loss": 0.9449, "step": 20729 }, { "epoch": 2.2057884656309854, "grad_norm": 10.7890625, "learning_rate": 9.281196702671973e-06, "loss": 0.9031, "step": 20730 }, { "epoch": 2.205894871249202, "grad_norm": 16.203125, "learning_rate": 9.281161171119956e-06, "loss": 0.843, "step": 20731 }, { "epoch": 2.2060012768674184, "grad_norm": 12.9140625, "learning_rate": 9.281125639567938e-06, "loss": 0.9584, "step": 20732 }, { "epoch": 2.2061076824856354, "grad_norm": 13.0078125, "learning_rate": 9.28109010801592e-06, "loss": 1.0781, "step": 20733 }, { "epoch": 2.206214088103852, "grad_norm": 12.6640625, "learning_rate": 9.2810545764639e-06, "loss": 1.1648, "step": 20734 }, { "epoch": 2.2063204937220684, "grad_norm": 11.21875, "learning_rate": 9.281019044911882e-06, "loss": 0.7355, "step": 20735 }, { "epoch": 2.2064268993402854, "grad_norm": 12.390625, "learning_rate": 9.280983513359864e-06, "loss": 1.0615, "step": 20736 }, { "epoch": 2.206533304958502, "grad_norm": 10.75, "learning_rate": 9.280947981807846e-06, "loss": 0.8223, "step": 20737 }, { "epoch": 2.2066397105767184, "grad_norm": 15.0625, "learning_rate": 9.280912450255827e-06, "loss": 1.0203, "step": 20738 }, { "epoch": 2.2067461161949353, "grad_norm": 14.7890625, "learning_rate": 9.280876918703809e-06, "loss": 1.1159, "step": 20739 }, { "epoch": 2.206852521813152, "grad_norm": 10.671875, "learning_rate": 9.28084138715179e-06, "loss": 0.7682, "step": 20740 }, { "epoch": 2.2069589274313683, "grad_norm": 13.859375, "learning_rate": 9.280805855599774e-06, "loss": 0.892, "step": 20741 }, { "epoch": 2.207065333049585, "grad_norm": 15.9921875, "learning_rate": 9.280770324047755e-06, "loss": 0.8342, "step": 20742 }, { "epoch": 2.207171738667802, "grad_norm": 13.7890625, "learning_rate": 9.280734792495737e-06, "loss": 0.9777, "step": 20743 }, { "epoch": 2.2072781442860183, "grad_norm": 20.078125, "learning_rate": 9.280699260943719e-06, "loss": 0.9593, "step": 20744 }, { "epoch": 2.207384549904235, "grad_norm": 11.1328125, "learning_rate": 9.2806637293917e-06, "loss": 0.8252, "step": 20745 }, { "epoch": 2.207490955522452, "grad_norm": 12.4921875, "learning_rate": 9.280628197839683e-06, "loss": 0.8161, "step": 20746 }, { "epoch": 2.2075973611406683, "grad_norm": 14.5, "learning_rate": 9.280592666287665e-06, "loss": 0.8715, "step": 20747 }, { "epoch": 2.207703766758885, "grad_norm": 17.96875, "learning_rate": 9.280557134735647e-06, "loss": 0.8871, "step": 20748 }, { "epoch": 2.2078101723771013, "grad_norm": 27.015625, "learning_rate": 9.280521603183628e-06, "loss": 1.0113, "step": 20749 }, { "epoch": 2.2079165779953183, "grad_norm": 14.140625, "learning_rate": 9.28048607163161e-06, "loss": 0.821, "step": 20750 }, { "epoch": 2.2080229836135348, "grad_norm": 17.5625, "learning_rate": 9.280450540079592e-06, "loss": 0.9041, "step": 20751 }, { "epoch": 2.2081293892317513, "grad_norm": 13.8125, "learning_rate": 9.280415008527573e-06, "loss": 0.9007, "step": 20752 }, { "epoch": 2.2082357948499682, "grad_norm": 20.609375, "learning_rate": 9.280379476975555e-06, "loss": 1.2457, "step": 20753 }, { "epoch": 2.2083422004681847, "grad_norm": 12.890625, "learning_rate": 9.280343945423536e-06, "loss": 0.9022, "step": 20754 }, { "epoch": 2.2084486060864013, "grad_norm": 14.625, "learning_rate": 9.280308413871518e-06, "loss": 0.9254, "step": 20755 }, { "epoch": 2.208555011704618, "grad_norm": 11.0, "learning_rate": 9.280272882319501e-06, "loss": 0.9218, "step": 20756 }, { "epoch": 2.2086614173228347, "grad_norm": 14.9453125, "learning_rate": 9.280237350767483e-06, "loss": 0.9873, "step": 20757 }, { "epoch": 2.2087678229410512, "grad_norm": 10.25, "learning_rate": 9.280201819215464e-06, "loss": 0.7984, "step": 20758 }, { "epoch": 2.2088742285592677, "grad_norm": 12.2890625, "learning_rate": 9.280166287663446e-06, "loss": 1.1145, "step": 20759 }, { "epoch": 2.2089806341774847, "grad_norm": 15.6171875, "learning_rate": 9.280130756111428e-06, "loss": 0.9827, "step": 20760 }, { "epoch": 2.209087039795701, "grad_norm": 13.25, "learning_rate": 9.28009522455941e-06, "loss": 0.8703, "step": 20761 }, { "epoch": 2.2091934454139177, "grad_norm": 12.6171875, "learning_rate": 9.280059693007391e-06, "loss": 0.9982, "step": 20762 }, { "epoch": 2.2092998510321347, "grad_norm": 16.09375, "learning_rate": 9.280024161455372e-06, "loss": 0.8336, "step": 20763 }, { "epoch": 2.209406256650351, "grad_norm": 12.296875, "learning_rate": 9.279988629903354e-06, "loss": 0.8401, "step": 20764 }, { "epoch": 2.2095126622685677, "grad_norm": 11.921875, "learning_rate": 9.279953098351336e-06, "loss": 0.8674, "step": 20765 }, { "epoch": 2.209619067886784, "grad_norm": 11.3984375, "learning_rate": 9.279917566799319e-06, "loss": 0.9181, "step": 20766 }, { "epoch": 2.209725473505001, "grad_norm": 12.609375, "learning_rate": 9.2798820352473e-06, "loss": 0.9052, "step": 20767 }, { "epoch": 2.2098318791232177, "grad_norm": 9.5, "learning_rate": 9.279846503695282e-06, "loss": 0.8287, "step": 20768 }, { "epoch": 2.209938284741434, "grad_norm": 11.3984375, "learning_rate": 9.279810972143264e-06, "loss": 0.881, "step": 20769 }, { "epoch": 2.210044690359651, "grad_norm": 14.5859375, "learning_rate": 9.279775440591245e-06, "loss": 0.9278, "step": 20770 }, { "epoch": 2.2101510959778676, "grad_norm": 11.3671875, "learning_rate": 9.279739909039227e-06, "loss": 0.8616, "step": 20771 }, { "epoch": 2.210257501596084, "grad_norm": 13.21875, "learning_rate": 9.27970437748721e-06, "loss": 0.9119, "step": 20772 }, { "epoch": 2.210363907214301, "grad_norm": 12.6484375, "learning_rate": 9.279668845935192e-06, "loss": 0.8731, "step": 20773 }, { "epoch": 2.2104703128325176, "grad_norm": 14.734375, "learning_rate": 9.279633314383174e-06, "loss": 0.8202, "step": 20774 }, { "epoch": 2.210576718450734, "grad_norm": 9.6171875, "learning_rate": 9.279597782831155e-06, "loss": 0.7253, "step": 20775 }, { "epoch": 2.210683124068951, "grad_norm": 19.359375, "learning_rate": 9.279562251279137e-06, "loss": 0.8452, "step": 20776 }, { "epoch": 2.2107895296871676, "grad_norm": 10.65625, "learning_rate": 9.279526719727118e-06, "loss": 0.7958, "step": 20777 }, { "epoch": 2.210895935305384, "grad_norm": 12.53125, "learning_rate": 9.2794911881751e-06, "loss": 0.8249, "step": 20778 }, { "epoch": 2.2110023409236006, "grad_norm": 11.7890625, "learning_rate": 9.279455656623082e-06, "loss": 0.7445, "step": 20779 }, { "epoch": 2.2111087465418175, "grad_norm": 12.9921875, "learning_rate": 9.279420125071063e-06, "loss": 0.7912, "step": 20780 }, { "epoch": 2.211215152160034, "grad_norm": 15.9140625, "learning_rate": 9.279384593519045e-06, "loss": 0.9084, "step": 20781 }, { "epoch": 2.2113215577782506, "grad_norm": 15.7421875, "learning_rate": 9.279349061967028e-06, "loss": 0.9668, "step": 20782 }, { "epoch": 2.2114279633964675, "grad_norm": 10.921875, "learning_rate": 9.27931353041501e-06, "loss": 0.8793, "step": 20783 }, { "epoch": 2.211534369014684, "grad_norm": 10.734375, "learning_rate": 9.279277998862991e-06, "loss": 0.9435, "step": 20784 }, { "epoch": 2.2116407746329005, "grad_norm": 10.9765625, "learning_rate": 9.279242467310973e-06, "loss": 0.9074, "step": 20785 }, { "epoch": 2.211747180251117, "grad_norm": 14.0078125, "learning_rate": 9.279206935758955e-06, "loss": 0.8378, "step": 20786 }, { "epoch": 2.211853585869334, "grad_norm": 14.1796875, "learning_rate": 9.279171404206936e-06, "loss": 0.8827, "step": 20787 }, { "epoch": 2.2119599914875505, "grad_norm": 11.9453125, "learning_rate": 9.279135872654918e-06, "loss": 0.8911, "step": 20788 }, { "epoch": 2.212066397105767, "grad_norm": 12.1953125, "learning_rate": 9.2791003411029e-06, "loss": 0.8085, "step": 20789 }, { "epoch": 2.212172802723984, "grad_norm": 13.7578125, "learning_rate": 9.279064809550881e-06, "loss": 1.0872, "step": 20790 }, { "epoch": 2.2122792083422005, "grad_norm": 16.765625, "learning_rate": 9.279029277998863e-06, "loss": 0.9547, "step": 20791 }, { "epoch": 2.212385613960417, "grad_norm": 16.609375, "learning_rate": 9.278993746446846e-06, "loss": 0.9798, "step": 20792 }, { "epoch": 2.212492019578634, "grad_norm": 13.4609375, "learning_rate": 9.278958214894827e-06, "loss": 1.0228, "step": 20793 }, { "epoch": 2.2125984251968505, "grad_norm": 14.015625, "learning_rate": 9.278922683342809e-06, "loss": 0.871, "step": 20794 }, { "epoch": 2.212704830815067, "grad_norm": 12.3671875, "learning_rate": 9.27888715179079e-06, "loss": 0.7395, "step": 20795 }, { "epoch": 2.2128112364332835, "grad_norm": 15.8046875, "learning_rate": 9.278851620238772e-06, "loss": 0.8823, "step": 20796 }, { "epoch": 2.2129176420515004, "grad_norm": 11.5234375, "learning_rate": 9.278816088686756e-06, "loss": 0.8068, "step": 20797 }, { "epoch": 2.213024047669717, "grad_norm": 10.9140625, "learning_rate": 9.278780557134737e-06, "loss": 0.782, "step": 20798 }, { "epoch": 2.2131304532879335, "grad_norm": 14.125, "learning_rate": 9.278745025582719e-06, "loss": 1.0093, "step": 20799 }, { "epoch": 2.2132368589061504, "grad_norm": 10.8359375, "learning_rate": 9.2787094940307e-06, "loss": 0.8966, "step": 20800 }, { "epoch": 2.213343264524367, "grad_norm": 13.3828125, "learning_rate": 9.278673962478682e-06, "loss": 0.9701, "step": 20801 }, { "epoch": 2.2134496701425834, "grad_norm": 20.703125, "learning_rate": 9.278638430926664e-06, "loss": 0.8582, "step": 20802 }, { "epoch": 2.2135560757608, "grad_norm": 11.9921875, "learning_rate": 9.278602899374645e-06, "loss": 0.7959, "step": 20803 }, { "epoch": 2.213662481379017, "grad_norm": 15.859375, "learning_rate": 9.278567367822627e-06, "loss": 1.1713, "step": 20804 }, { "epoch": 2.2137688869972334, "grad_norm": 14.9140625, "learning_rate": 9.278531836270608e-06, "loss": 0.9756, "step": 20805 }, { "epoch": 2.21387529261545, "grad_norm": 10.8046875, "learning_rate": 9.27849630471859e-06, "loss": 0.7775, "step": 20806 }, { "epoch": 2.213981698233667, "grad_norm": 15.734375, "learning_rate": 9.278460773166572e-06, "loss": 0.9609, "step": 20807 }, { "epoch": 2.2140881038518834, "grad_norm": 17.578125, "learning_rate": 9.278425241614555e-06, "loss": 0.9952, "step": 20808 }, { "epoch": 2.2141945094701, "grad_norm": 12.90625, "learning_rate": 9.278389710062537e-06, "loss": 0.8589, "step": 20809 }, { "epoch": 2.214300915088317, "grad_norm": 11.03125, "learning_rate": 9.278354178510518e-06, "loss": 0.9155, "step": 20810 }, { "epoch": 2.2144073207065333, "grad_norm": 11.9375, "learning_rate": 9.2783186469585e-06, "loss": 0.8417, "step": 20811 }, { "epoch": 2.21451372632475, "grad_norm": 12.40625, "learning_rate": 9.278283115406481e-06, "loss": 0.9238, "step": 20812 }, { "epoch": 2.214620131942967, "grad_norm": 11.9609375, "learning_rate": 9.278247583854465e-06, "loss": 0.8213, "step": 20813 }, { "epoch": 2.2147265375611833, "grad_norm": 13.78125, "learning_rate": 9.278212052302446e-06, "loss": 0.8864, "step": 20814 }, { "epoch": 2.2148329431794, "grad_norm": 10.34375, "learning_rate": 9.278176520750428e-06, "loss": 0.9214, "step": 20815 }, { "epoch": 2.2149393487976163, "grad_norm": 13.7890625, "learning_rate": 9.27814098919841e-06, "loss": 0.9868, "step": 20816 }, { "epoch": 2.2150457544158333, "grad_norm": 13.046875, "learning_rate": 9.27810545764639e-06, "loss": 0.9474, "step": 20817 }, { "epoch": 2.21515216003405, "grad_norm": 11.546875, "learning_rate": 9.278069926094373e-06, "loss": 0.6972, "step": 20818 }, { "epoch": 2.2152585656522663, "grad_norm": 11.1953125, "learning_rate": 9.278034394542354e-06, "loss": 0.7035, "step": 20819 }, { "epoch": 2.2153649712704833, "grad_norm": 14.5, "learning_rate": 9.277998862990336e-06, "loss": 0.8989, "step": 20820 }, { "epoch": 2.2154713768886998, "grad_norm": 13.234375, "learning_rate": 9.277963331438318e-06, "loss": 0.9083, "step": 20821 }, { "epoch": 2.2155777825069163, "grad_norm": 17.40625, "learning_rate": 9.277927799886299e-06, "loss": 0.8662, "step": 20822 }, { "epoch": 2.215684188125133, "grad_norm": 12.09375, "learning_rate": 9.277892268334282e-06, "loss": 0.8758, "step": 20823 }, { "epoch": 2.2157905937433497, "grad_norm": 12.8828125, "learning_rate": 9.277856736782264e-06, "loss": 0.8467, "step": 20824 }, { "epoch": 2.2158969993615663, "grad_norm": 10.6640625, "learning_rate": 9.277821205230246e-06, "loss": 0.8615, "step": 20825 }, { "epoch": 2.2160034049797828, "grad_norm": 12.8046875, "learning_rate": 9.277785673678227e-06, "loss": 0.918, "step": 20826 }, { "epoch": 2.2161098105979997, "grad_norm": 11.2265625, "learning_rate": 9.277750142126209e-06, "loss": 0.9723, "step": 20827 }, { "epoch": 2.2162162162162162, "grad_norm": 14.0234375, "learning_rate": 9.27771461057419e-06, "loss": 0.9527, "step": 20828 }, { "epoch": 2.2163226218344327, "grad_norm": 14.3984375, "learning_rate": 9.277679079022172e-06, "loss": 0.9096, "step": 20829 }, { "epoch": 2.2164290274526497, "grad_norm": 17.640625, "learning_rate": 9.277643547470154e-06, "loss": 0.9839, "step": 20830 }, { "epoch": 2.216535433070866, "grad_norm": 10.6640625, "learning_rate": 9.277608015918135e-06, "loss": 0.9094, "step": 20831 }, { "epoch": 2.2166418386890827, "grad_norm": 14.640625, "learning_rate": 9.277572484366117e-06, "loss": 1.2211, "step": 20832 }, { "epoch": 2.216748244307299, "grad_norm": 10.4453125, "learning_rate": 9.2775369528141e-06, "loss": 0.769, "step": 20833 }, { "epoch": 2.216854649925516, "grad_norm": 11.546875, "learning_rate": 9.277501421262082e-06, "loss": 0.8273, "step": 20834 }, { "epoch": 2.2169610555437327, "grad_norm": 13.015625, "learning_rate": 9.277465889710063e-06, "loss": 0.9336, "step": 20835 }, { "epoch": 2.217067461161949, "grad_norm": 10.453125, "learning_rate": 9.277430358158045e-06, "loss": 0.7951, "step": 20836 }, { "epoch": 2.217173866780166, "grad_norm": 9.890625, "learning_rate": 9.277394826606027e-06, "loss": 0.7727, "step": 20837 }, { "epoch": 2.2172802723983827, "grad_norm": 10.8046875, "learning_rate": 9.27735929505401e-06, "loss": 0.7249, "step": 20838 }, { "epoch": 2.217386678016599, "grad_norm": 24.234375, "learning_rate": 9.277323763501992e-06, "loss": 1.2067, "step": 20839 }, { "epoch": 2.217493083634816, "grad_norm": 13.2421875, "learning_rate": 9.277288231949973e-06, "loss": 0.7904, "step": 20840 }, { "epoch": 2.2175994892530326, "grad_norm": 11.984375, "learning_rate": 9.277252700397955e-06, "loss": 0.846, "step": 20841 }, { "epoch": 2.217705894871249, "grad_norm": 14.4140625, "learning_rate": 9.277217168845936e-06, "loss": 1.0626, "step": 20842 }, { "epoch": 2.2178123004894656, "grad_norm": 13.59375, "learning_rate": 9.277181637293918e-06, "loss": 0.7903, "step": 20843 }, { "epoch": 2.2179187061076826, "grad_norm": 10.4921875, "learning_rate": 9.2771461057419e-06, "loss": 0.8743, "step": 20844 }, { "epoch": 2.218025111725899, "grad_norm": 14.9921875, "learning_rate": 9.277110574189881e-06, "loss": 1.0771, "step": 20845 }, { "epoch": 2.2181315173441156, "grad_norm": 12.9765625, "learning_rate": 9.277075042637863e-06, "loss": 0.9365, "step": 20846 }, { "epoch": 2.2182379229623326, "grad_norm": 11.4609375, "learning_rate": 9.277039511085844e-06, "loss": 0.9302, "step": 20847 }, { "epoch": 2.218344328580549, "grad_norm": 13.828125, "learning_rate": 9.277003979533826e-06, "loss": 1.0983, "step": 20848 }, { "epoch": 2.2184507341987656, "grad_norm": 14.1015625, "learning_rate": 9.27696844798181e-06, "loss": 0.8654, "step": 20849 }, { "epoch": 2.2185571398169825, "grad_norm": 15.0078125, "learning_rate": 9.276932916429791e-06, "loss": 0.8191, "step": 20850 }, { "epoch": 2.218663545435199, "grad_norm": 13.5234375, "learning_rate": 9.276897384877773e-06, "loss": 1.0055, "step": 20851 }, { "epoch": 2.2187699510534156, "grad_norm": 14.1953125, "learning_rate": 9.276861853325754e-06, "loss": 0.8681, "step": 20852 }, { "epoch": 2.218876356671632, "grad_norm": 10.609375, "learning_rate": 9.276826321773736e-06, "loss": 0.922, "step": 20853 }, { "epoch": 2.218982762289849, "grad_norm": 14.4609375, "learning_rate": 9.276790790221717e-06, "loss": 0.9953, "step": 20854 }, { "epoch": 2.2190891679080655, "grad_norm": 13.2578125, "learning_rate": 9.276755258669699e-06, "loss": 0.9163, "step": 20855 }, { "epoch": 2.219195573526282, "grad_norm": 18.46875, "learning_rate": 9.27671972711768e-06, "loss": 1.1554, "step": 20856 }, { "epoch": 2.219301979144499, "grad_norm": 12.46875, "learning_rate": 9.276684195565662e-06, "loss": 0.7497, "step": 20857 }, { "epoch": 2.2194083847627155, "grad_norm": 14.7890625, "learning_rate": 9.276648664013644e-06, "loss": 1.0587, "step": 20858 }, { "epoch": 2.219514790380932, "grad_norm": 9.7265625, "learning_rate": 9.276613132461627e-06, "loss": 0.9084, "step": 20859 }, { "epoch": 2.2196211959991485, "grad_norm": 9.890625, "learning_rate": 9.276577600909609e-06, "loss": 0.7921, "step": 20860 }, { "epoch": 2.2197276016173655, "grad_norm": 17.671875, "learning_rate": 9.27654206935759e-06, "loss": 0.9731, "step": 20861 }, { "epoch": 2.219834007235582, "grad_norm": 11.9375, "learning_rate": 9.276506537805572e-06, "loss": 0.9855, "step": 20862 }, { "epoch": 2.2199404128537985, "grad_norm": 12.71875, "learning_rate": 9.276471006253553e-06, "loss": 0.8417, "step": 20863 }, { "epoch": 2.2200468184720155, "grad_norm": 10.765625, "learning_rate": 9.276435474701537e-06, "loss": 0.9166, "step": 20864 }, { "epoch": 2.220153224090232, "grad_norm": 9.3671875, "learning_rate": 9.276399943149518e-06, "loss": 0.9907, "step": 20865 }, { "epoch": 2.2202596297084485, "grad_norm": 14.7578125, "learning_rate": 9.2763644115975e-06, "loss": 0.9881, "step": 20866 }, { "epoch": 2.2203660353266654, "grad_norm": 10.890625, "learning_rate": 9.276328880045482e-06, "loss": 0.947, "step": 20867 }, { "epoch": 2.220472440944882, "grad_norm": 20.078125, "learning_rate": 9.276293348493463e-06, "loss": 0.9321, "step": 20868 }, { "epoch": 2.2205788465630985, "grad_norm": 12.546875, "learning_rate": 9.276257816941445e-06, "loss": 0.8462, "step": 20869 }, { "epoch": 2.2206852521813154, "grad_norm": 12.4765625, "learning_rate": 9.276222285389426e-06, "loss": 0.8476, "step": 20870 }, { "epoch": 2.220791657799532, "grad_norm": 15.09375, "learning_rate": 9.276186753837408e-06, "loss": 0.8455, "step": 20871 }, { "epoch": 2.2208980634177484, "grad_norm": 14.0078125, "learning_rate": 9.27615122228539e-06, "loss": 0.9352, "step": 20872 }, { "epoch": 2.221004469035965, "grad_norm": 11.8046875, "learning_rate": 9.276115690733371e-06, "loss": 0.8942, "step": 20873 }, { "epoch": 2.221110874654182, "grad_norm": 14.0703125, "learning_rate": 9.276080159181355e-06, "loss": 0.9593, "step": 20874 }, { "epoch": 2.2212172802723984, "grad_norm": 11.0546875, "learning_rate": 9.276044627629336e-06, "loss": 0.8845, "step": 20875 }, { "epoch": 2.221323685890615, "grad_norm": 13.5390625, "learning_rate": 9.276009096077318e-06, "loss": 0.8334, "step": 20876 }, { "epoch": 2.221430091508832, "grad_norm": 16.828125, "learning_rate": 9.2759735645253e-06, "loss": 0.9841, "step": 20877 }, { "epoch": 2.2215364971270484, "grad_norm": 13.3046875, "learning_rate": 9.275938032973281e-06, "loss": 1.0396, "step": 20878 }, { "epoch": 2.221642902745265, "grad_norm": 11.921875, "learning_rate": 9.275902501421263e-06, "loss": 0.895, "step": 20879 }, { "epoch": 2.2217493083634814, "grad_norm": 14.859375, "learning_rate": 9.275866969869244e-06, "loss": 0.8754, "step": 20880 }, { "epoch": 2.2218557139816983, "grad_norm": 13.59375, "learning_rate": 9.275831438317226e-06, "loss": 0.8568, "step": 20881 }, { "epoch": 2.221962119599915, "grad_norm": 13.5703125, "learning_rate": 9.275795906765207e-06, "loss": 0.9238, "step": 20882 }, { "epoch": 2.2220685252181314, "grad_norm": 11.28125, "learning_rate": 9.275760375213189e-06, "loss": 0.7527, "step": 20883 }, { "epoch": 2.2221749308363483, "grad_norm": 10.703125, "learning_rate": 9.275724843661172e-06, "loss": 0.7213, "step": 20884 }, { "epoch": 2.222281336454565, "grad_norm": 14.8828125, "learning_rate": 9.275689312109154e-06, "loss": 0.9457, "step": 20885 }, { "epoch": 2.2223877420727813, "grad_norm": 12.4140625, "learning_rate": 9.275653780557136e-06, "loss": 0.9382, "step": 20886 }, { "epoch": 2.2224941476909983, "grad_norm": 14.1484375, "learning_rate": 9.275618249005117e-06, "loss": 0.9448, "step": 20887 }, { "epoch": 2.222600553309215, "grad_norm": 15.5546875, "learning_rate": 9.275582717453099e-06, "loss": 0.9474, "step": 20888 }, { "epoch": 2.2227069589274313, "grad_norm": 10.953125, "learning_rate": 9.27554718590108e-06, "loss": 0.9088, "step": 20889 }, { "epoch": 2.222813364545648, "grad_norm": 15.453125, "learning_rate": 9.275511654349064e-06, "loss": 0.8231, "step": 20890 }, { "epoch": 2.2229197701638648, "grad_norm": 11.3203125, "learning_rate": 9.275476122797045e-06, "loss": 0.8631, "step": 20891 }, { "epoch": 2.2230261757820813, "grad_norm": 10.6640625, "learning_rate": 9.275440591245027e-06, "loss": 0.9521, "step": 20892 }, { "epoch": 2.223132581400298, "grad_norm": 23.59375, "learning_rate": 9.275405059693008e-06, "loss": 0.8889, "step": 20893 }, { "epoch": 2.2232389870185147, "grad_norm": 12.1953125, "learning_rate": 9.27536952814099e-06, "loss": 1.006, "step": 20894 }, { "epoch": 2.2233453926367313, "grad_norm": 14.875, "learning_rate": 9.275333996588972e-06, "loss": 0.6251, "step": 20895 }, { "epoch": 2.2234517982549478, "grad_norm": 15.09375, "learning_rate": 9.275298465036953e-06, "loss": 1.0544, "step": 20896 }, { "epoch": 2.2235582038731643, "grad_norm": 15.171875, "learning_rate": 9.275262933484935e-06, "loss": 0.9982, "step": 20897 }, { "epoch": 2.2236646094913812, "grad_norm": 14.109375, "learning_rate": 9.275227401932916e-06, "loss": 0.9338, "step": 20898 }, { "epoch": 2.2237710151095977, "grad_norm": 11.2421875, "learning_rate": 9.275191870380898e-06, "loss": 0.937, "step": 20899 }, { "epoch": 2.2238774207278142, "grad_norm": 18.578125, "learning_rate": 9.275156338828881e-06, "loss": 1.1379, "step": 20900 }, { "epoch": 2.223983826346031, "grad_norm": 14.828125, "learning_rate": 9.275120807276863e-06, "loss": 0.8164, "step": 20901 }, { "epoch": 2.2240902319642477, "grad_norm": 13.3828125, "learning_rate": 9.275085275724845e-06, "loss": 0.8248, "step": 20902 }, { "epoch": 2.224196637582464, "grad_norm": 10.4140625, "learning_rate": 9.275049744172826e-06, "loss": 0.8243, "step": 20903 }, { "epoch": 2.224303043200681, "grad_norm": 15.609375, "learning_rate": 9.275014212620808e-06, "loss": 0.7787, "step": 20904 }, { "epoch": 2.2244094488188977, "grad_norm": 13.2421875, "learning_rate": 9.274978681068791e-06, "loss": 0.7408, "step": 20905 }, { "epoch": 2.224515854437114, "grad_norm": 14.3515625, "learning_rate": 9.274943149516771e-06, "loss": 0.9744, "step": 20906 }, { "epoch": 2.224622260055331, "grad_norm": 16.640625, "learning_rate": 9.274907617964753e-06, "loss": 0.9696, "step": 20907 }, { "epoch": 2.2247286656735477, "grad_norm": 9.890625, "learning_rate": 9.274872086412734e-06, "loss": 0.7238, "step": 20908 }, { "epoch": 2.224835071291764, "grad_norm": 10.4140625, "learning_rate": 9.274836554860716e-06, "loss": 1.0176, "step": 20909 }, { "epoch": 2.2249414769099807, "grad_norm": 16.703125, "learning_rate": 9.2748010233087e-06, "loss": 1.106, "step": 20910 }, { "epoch": 2.2250478825281976, "grad_norm": 14.0234375, "learning_rate": 9.27476549175668e-06, "loss": 0.7165, "step": 20911 }, { "epoch": 2.225154288146414, "grad_norm": 10.1015625, "learning_rate": 9.274729960204662e-06, "loss": 0.795, "step": 20912 }, { "epoch": 2.2252606937646306, "grad_norm": 16.078125, "learning_rate": 9.274694428652644e-06, "loss": 1.1225, "step": 20913 }, { "epoch": 2.2253670993828476, "grad_norm": 14.1328125, "learning_rate": 9.274658897100626e-06, "loss": 0.8167, "step": 20914 }, { "epoch": 2.225473505001064, "grad_norm": 12.4453125, "learning_rate": 9.274623365548609e-06, "loss": 0.8861, "step": 20915 }, { "epoch": 2.2255799106192806, "grad_norm": 14.390625, "learning_rate": 9.27458783399659e-06, "loss": 1.0701, "step": 20916 }, { "epoch": 2.225686316237497, "grad_norm": 18.5, "learning_rate": 9.274552302444572e-06, "loss": 0.9715, "step": 20917 }, { "epoch": 2.225792721855714, "grad_norm": 11.625, "learning_rate": 9.274516770892554e-06, "loss": 0.8487, "step": 20918 }, { "epoch": 2.2258991274739306, "grad_norm": 12.828125, "learning_rate": 9.274481239340535e-06, "loss": 0.9431, "step": 20919 }, { "epoch": 2.226005533092147, "grad_norm": 11.953125, "learning_rate": 9.274445707788517e-06, "loss": 0.9781, "step": 20920 }, { "epoch": 2.226111938710364, "grad_norm": 11.375, "learning_rate": 9.274410176236499e-06, "loss": 0.8266, "step": 20921 }, { "epoch": 2.2262183443285806, "grad_norm": 10.640625, "learning_rate": 9.27437464468448e-06, "loss": 0.8854, "step": 20922 }, { "epoch": 2.226324749946797, "grad_norm": 12.4296875, "learning_rate": 9.274339113132462e-06, "loss": 1.02, "step": 20923 }, { "epoch": 2.226431155565014, "grad_norm": 14.6875, "learning_rate": 9.274303581580443e-06, "loss": 1.1049, "step": 20924 }, { "epoch": 2.2265375611832305, "grad_norm": 13.078125, "learning_rate": 9.274268050028425e-06, "loss": 0.8207, "step": 20925 }, { "epoch": 2.226643966801447, "grad_norm": 12.015625, "learning_rate": 9.274232518476408e-06, "loss": 0.8185, "step": 20926 }, { "epoch": 2.2267503724196636, "grad_norm": 12.46875, "learning_rate": 9.27419698692439e-06, "loss": 0.9496, "step": 20927 }, { "epoch": 2.2268567780378805, "grad_norm": 9.8125, "learning_rate": 9.274161455372371e-06, "loss": 0.8445, "step": 20928 }, { "epoch": 2.226963183656097, "grad_norm": 21.234375, "learning_rate": 9.274125923820353e-06, "loss": 0.9127, "step": 20929 }, { "epoch": 2.2270695892743135, "grad_norm": 19.484375, "learning_rate": 9.274090392268335e-06, "loss": 0.9485, "step": 20930 }, { "epoch": 2.2271759948925305, "grad_norm": 13.03125, "learning_rate": 9.274054860716318e-06, "loss": 1.0026, "step": 20931 }, { "epoch": 2.227282400510747, "grad_norm": 14.1015625, "learning_rate": 9.2740193291643e-06, "loss": 0.9111, "step": 20932 }, { "epoch": 2.2273888061289635, "grad_norm": 11.4296875, "learning_rate": 9.273983797612281e-06, "loss": 0.9285, "step": 20933 }, { "epoch": 2.2274952117471805, "grad_norm": 12.65625, "learning_rate": 9.273948266060263e-06, "loss": 1.0082, "step": 20934 }, { "epoch": 2.227601617365397, "grad_norm": 14.609375, "learning_rate": 9.273912734508244e-06, "loss": 0.9088, "step": 20935 }, { "epoch": 2.2277080229836135, "grad_norm": 15.0078125, "learning_rate": 9.273877202956226e-06, "loss": 0.9701, "step": 20936 }, { "epoch": 2.22781442860183, "grad_norm": 12.984375, "learning_rate": 9.273841671404208e-06, "loss": 0.9124, "step": 20937 }, { "epoch": 2.227920834220047, "grad_norm": 11.1640625, "learning_rate": 9.27380613985219e-06, "loss": 0.7788, "step": 20938 }, { "epoch": 2.2280272398382635, "grad_norm": 18.015625, "learning_rate": 9.27377060830017e-06, "loss": 1.068, "step": 20939 }, { "epoch": 2.22813364545648, "grad_norm": 11.1328125, "learning_rate": 9.273735076748152e-06, "loss": 0.8746, "step": 20940 }, { "epoch": 2.228240051074697, "grad_norm": 16.5, "learning_rate": 9.273699545196136e-06, "loss": 0.6987, "step": 20941 }, { "epoch": 2.2283464566929134, "grad_norm": 11.609375, "learning_rate": 9.273664013644117e-06, "loss": 0.9753, "step": 20942 }, { "epoch": 2.22845286231113, "grad_norm": 12.1484375, "learning_rate": 9.273628482092099e-06, "loss": 0.8983, "step": 20943 }, { "epoch": 2.228559267929347, "grad_norm": 14.65625, "learning_rate": 9.27359295054008e-06, "loss": 0.8428, "step": 20944 }, { "epoch": 2.2286656735475634, "grad_norm": 11.984375, "learning_rate": 9.273557418988062e-06, "loss": 0.9428, "step": 20945 }, { "epoch": 2.22877207916578, "grad_norm": 15.4609375, "learning_rate": 9.273521887436044e-06, "loss": 0.91, "step": 20946 }, { "epoch": 2.2288784847839964, "grad_norm": 13.59375, "learning_rate": 9.273486355884025e-06, "loss": 0.8826, "step": 20947 }, { "epoch": 2.2289848904022134, "grad_norm": 13.953125, "learning_rate": 9.273450824332007e-06, "loss": 0.9659, "step": 20948 }, { "epoch": 2.22909129602043, "grad_norm": 14.84375, "learning_rate": 9.273415292779989e-06, "loss": 0.8446, "step": 20949 }, { "epoch": 2.2291977016386464, "grad_norm": 12.390625, "learning_rate": 9.27337976122797e-06, "loss": 0.7798, "step": 20950 }, { "epoch": 2.2293041072568633, "grad_norm": 10.5546875, "learning_rate": 9.273344229675954e-06, "loss": 0.7675, "step": 20951 }, { "epoch": 2.22941051287508, "grad_norm": 12.4140625, "learning_rate": 9.273308698123935e-06, "loss": 0.8546, "step": 20952 }, { "epoch": 2.2295169184932964, "grad_norm": 15.4296875, "learning_rate": 9.273273166571917e-06, "loss": 0.9137, "step": 20953 }, { "epoch": 2.229623324111513, "grad_norm": 12.984375, "learning_rate": 9.273237635019898e-06, "loss": 0.8602, "step": 20954 }, { "epoch": 2.22972972972973, "grad_norm": 12.625, "learning_rate": 9.27320210346788e-06, "loss": 0.9712, "step": 20955 }, { "epoch": 2.2298361353479463, "grad_norm": 11.203125, "learning_rate": 9.273166571915863e-06, "loss": 0.7907, "step": 20956 }, { "epoch": 2.229942540966163, "grad_norm": 13.90625, "learning_rate": 9.273131040363845e-06, "loss": 0.9398, "step": 20957 }, { "epoch": 2.23004894658438, "grad_norm": 13.4453125, "learning_rate": 9.273095508811826e-06, "loss": 0.9903, "step": 20958 }, { "epoch": 2.2301553522025963, "grad_norm": 13.1796875, "learning_rate": 9.273059977259808e-06, "loss": 0.8165, "step": 20959 }, { "epoch": 2.230261757820813, "grad_norm": 10.3984375, "learning_rate": 9.27302444570779e-06, "loss": 0.7252, "step": 20960 }, { "epoch": 2.2303681634390298, "grad_norm": 12.5546875, "learning_rate": 9.272988914155771e-06, "loss": 0.8279, "step": 20961 }, { "epoch": 2.2304745690572463, "grad_norm": 11.78125, "learning_rate": 9.272953382603753e-06, "loss": 0.9408, "step": 20962 }, { "epoch": 2.230580974675463, "grad_norm": 14.640625, "learning_rate": 9.272917851051734e-06, "loss": 1.0082, "step": 20963 }, { "epoch": 2.2306873802936793, "grad_norm": 11.9609375, "learning_rate": 9.272882319499716e-06, "loss": 0.7881, "step": 20964 }, { "epoch": 2.2307937859118963, "grad_norm": 13.5234375, "learning_rate": 9.272846787947698e-06, "loss": 0.9098, "step": 20965 }, { "epoch": 2.2309001915301128, "grad_norm": 14.3203125, "learning_rate": 9.27281125639568e-06, "loss": 1.0076, "step": 20966 }, { "epoch": 2.2310065971483293, "grad_norm": 11.40625, "learning_rate": 9.272775724843663e-06, "loss": 0.8417, "step": 20967 }, { "epoch": 2.2311130027665462, "grad_norm": 19.453125, "learning_rate": 9.272740193291644e-06, "loss": 0.9297, "step": 20968 }, { "epoch": 2.2312194083847627, "grad_norm": 19.125, "learning_rate": 9.272704661739626e-06, "loss": 1.0283, "step": 20969 }, { "epoch": 2.2313258140029792, "grad_norm": 17.4375, "learning_rate": 9.272669130187607e-06, "loss": 1.0608, "step": 20970 }, { "epoch": 2.231432219621196, "grad_norm": 14.46875, "learning_rate": 9.272633598635589e-06, "loss": 0.9959, "step": 20971 }, { "epoch": 2.2315386252394127, "grad_norm": 14.2578125, "learning_rate": 9.27259806708357e-06, "loss": 0.9302, "step": 20972 }, { "epoch": 2.231645030857629, "grad_norm": 9.9765625, "learning_rate": 9.272562535531552e-06, "loss": 0.7501, "step": 20973 }, { "epoch": 2.2317514364758457, "grad_norm": 11.59375, "learning_rate": 9.272527003979534e-06, "loss": 0.8407, "step": 20974 }, { "epoch": 2.2318578420940627, "grad_norm": 20.1875, "learning_rate": 9.272491472427515e-06, "loss": 0.8261, "step": 20975 }, { "epoch": 2.231964247712279, "grad_norm": 18.125, "learning_rate": 9.272455940875497e-06, "loss": 0.8527, "step": 20976 }, { "epoch": 2.2320706533304957, "grad_norm": 19.75, "learning_rate": 9.27242040932348e-06, "loss": 0.9395, "step": 20977 }, { "epoch": 2.2321770589487127, "grad_norm": 12.21875, "learning_rate": 9.272384877771462e-06, "loss": 0.8996, "step": 20978 }, { "epoch": 2.232283464566929, "grad_norm": 12.7734375, "learning_rate": 9.272349346219444e-06, "loss": 0.794, "step": 20979 }, { "epoch": 2.2323898701851457, "grad_norm": 12.140625, "learning_rate": 9.272313814667425e-06, "loss": 1.0437, "step": 20980 }, { "epoch": 2.2324962758033626, "grad_norm": 10.78125, "learning_rate": 9.272278283115407e-06, "loss": 0.8569, "step": 20981 }, { "epoch": 2.232602681421579, "grad_norm": 13.296875, "learning_rate": 9.27224275156339e-06, "loss": 0.9996, "step": 20982 }, { "epoch": 2.2327090870397956, "grad_norm": 16.078125, "learning_rate": 9.272207220011372e-06, "loss": 0.9582, "step": 20983 }, { "epoch": 2.232815492658012, "grad_norm": 12.984375, "learning_rate": 9.272171688459353e-06, "loss": 0.7856, "step": 20984 }, { "epoch": 2.232921898276229, "grad_norm": 11.4140625, "learning_rate": 9.272136156907335e-06, "loss": 0.8422, "step": 20985 }, { "epoch": 2.2330283038944456, "grad_norm": 11.78125, "learning_rate": 9.272100625355317e-06, "loss": 0.8719, "step": 20986 }, { "epoch": 2.233134709512662, "grad_norm": 14.078125, "learning_rate": 9.272065093803298e-06, "loss": 0.954, "step": 20987 }, { "epoch": 2.233241115130879, "grad_norm": 8.671875, "learning_rate": 9.27202956225128e-06, "loss": 0.7184, "step": 20988 }, { "epoch": 2.2333475207490956, "grad_norm": 9.6953125, "learning_rate": 9.271994030699261e-06, "loss": 0.7785, "step": 20989 }, { "epoch": 2.233453926367312, "grad_norm": 11.1875, "learning_rate": 9.271958499147243e-06, "loss": 0.8775, "step": 20990 }, { "epoch": 2.2335603319855286, "grad_norm": 12.3046875, "learning_rate": 9.271922967595225e-06, "loss": 0.8943, "step": 20991 }, { "epoch": 2.2336667376037456, "grad_norm": 14.609375, "learning_rate": 9.271887436043208e-06, "loss": 0.9525, "step": 20992 }, { "epoch": 2.233773143221962, "grad_norm": 9.7109375, "learning_rate": 9.27185190449119e-06, "loss": 0.7777, "step": 20993 }, { "epoch": 2.2338795488401786, "grad_norm": 10.84375, "learning_rate": 9.271816372939171e-06, "loss": 0.8006, "step": 20994 }, { "epoch": 2.2339859544583955, "grad_norm": 12.875, "learning_rate": 9.271780841387153e-06, "loss": 0.9053, "step": 20995 }, { "epoch": 2.234092360076612, "grad_norm": 11.75, "learning_rate": 9.271745309835134e-06, "loss": 0.693, "step": 20996 }, { "epoch": 2.2341987656948286, "grad_norm": 12.0859375, "learning_rate": 9.271709778283116e-06, "loss": 0.8239, "step": 20997 }, { "epoch": 2.2343051713130455, "grad_norm": 18.296875, "learning_rate": 9.271674246731097e-06, "loss": 1.1122, "step": 20998 }, { "epoch": 2.234411576931262, "grad_norm": 14.40625, "learning_rate": 9.271638715179079e-06, "loss": 0.8693, "step": 20999 }, { "epoch": 2.2345179825494785, "grad_norm": 11.8515625, "learning_rate": 9.27160318362706e-06, "loss": 1.0106, "step": 21000 }, { "epoch": 2.2346243881676955, "grad_norm": 18.125, "learning_rate": 9.271567652075042e-06, "loss": 0.96, "step": 21001 }, { "epoch": 2.234730793785912, "grad_norm": 13.203125, "learning_rate": 9.271532120523026e-06, "loss": 1.1643, "step": 21002 }, { "epoch": 2.2348371994041285, "grad_norm": 14.171875, "learning_rate": 9.271496588971007e-06, "loss": 0.7766, "step": 21003 }, { "epoch": 2.234943605022345, "grad_norm": 13.5625, "learning_rate": 9.271461057418989e-06, "loss": 0.8286, "step": 21004 }, { "epoch": 2.235050010640562, "grad_norm": 14.6796875, "learning_rate": 9.27142552586697e-06, "loss": 0.9971, "step": 21005 }, { "epoch": 2.2351564162587785, "grad_norm": 14.046875, "learning_rate": 9.271389994314952e-06, "loss": 0.9103, "step": 21006 }, { "epoch": 2.235262821876995, "grad_norm": 12.5390625, "learning_rate": 9.271354462762934e-06, "loss": 0.9944, "step": 21007 }, { "epoch": 2.235369227495212, "grad_norm": 15.6171875, "learning_rate": 9.271318931210917e-06, "loss": 1.1491, "step": 21008 }, { "epoch": 2.2354756331134285, "grad_norm": 11.375, "learning_rate": 9.271283399658899e-06, "loss": 0.9164, "step": 21009 }, { "epoch": 2.235582038731645, "grad_norm": 15.4296875, "learning_rate": 9.27124786810688e-06, "loss": 0.8311, "step": 21010 }, { "epoch": 2.2356884443498615, "grad_norm": 13.015625, "learning_rate": 9.271212336554862e-06, "loss": 0.8712, "step": 21011 }, { "epoch": 2.2357948499680784, "grad_norm": 13.7265625, "learning_rate": 9.271176805002843e-06, "loss": 0.9268, "step": 21012 }, { "epoch": 2.235901255586295, "grad_norm": 11.46875, "learning_rate": 9.271141273450825e-06, "loss": 0.7385, "step": 21013 }, { "epoch": 2.2360076612045114, "grad_norm": 16.140625, "learning_rate": 9.271105741898807e-06, "loss": 1.0342, "step": 21014 }, { "epoch": 2.2361140668227284, "grad_norm": 14.484375, "learning_rate": 9.271070210346788e-06, "loss": 1.0336, "step": 21015 }, { "epoch": 2.236220472440945, "grad_norm": 14.2421875, "learning_rate": 9.27103467879477e-06, "loss": 1.0085, "step": 21016 }, { "epoch": 2.2363268780591614, "grad_norm": 12.8046875, "learning_rate": 9.270999147242751e-06, "loss": 1.1207, "step": 21017 }, { "epoch": 2.2364332836773784, "grad_norm": 9.9296875, "learning_rate": 9.270963615690735e-06, "loss": 0.7992, "step": 21018 }, { "epoch": 2.236539689295595, "grad_norm": 10.9609375, "learning_rate": 9.270928084138716e-06, "loss": 0.9346, "step": 21019 }, { "epoch": 2.2366460949138114, "grad_norm": 12.8203125, "learning_rate": 9.270892552586698e-06, "loss": 0.7565, "step": 21020 }, { "epoch": 2.236752500532028, "grad_norm": 15.609375, "learning_rate": 9.27085702103468e-06, "loss": 0.9094, "step": 21021 }, { "epoch": 2.236858906150245, "grad_norm": 14.1015625, "learning_rate": 9.270821489482661e-06, "loss": 0.9595, "step": 21022 }, { "epoch": 2.2369653117684614, "grad_norm": 12.4609375, "learning_rate": 9.270785957930644e-06, "loss": 0.8387, "step": 21023 }, { "epoch": 2.237071717386678, "grad_norm": 12.8359375, "learning_rate": 9.270750426378626e-06, "loss": 0.8404, "step": 21024 }, { "epoch": 2.237178123004895, "grad_norm": 13.5546875, "learning_rate": 9.270714894826608e-06, "loss": 0.8892, "step": 21025 }, { "epoch": 2.2372845286231113, "grad_norm": 15.0390625, "learning_rate": 9.270679363274588e-06, "loss": 0.999, "step": 21026 }, { "epoch": 2.237390934241328, "grad_norm": 10.765625, "learning_rate": 9.27064383172257e-06, "loss": 0.9437, "step": 21027 }, { "epoch": 2.2374973398595444, "grad_norm": 11.65625, "learning_rate": 9.270608300170552e-06, "loss": 0.7806, "step": 21028 }, { "epoch": 2.2376037454777613, "grad_norm": 10.125, "learning_rate": 9.270572768618534e-06, "loss": 0.8702, "step": 21029 }, { "epoch": 2.237710151095978, "grad_norm": 11.90625, "learning_rate": 9.270537237066516e-06, "loss": 1.0895, "step": 21030 }, { "epoch": 2.2378165567141943, "grad_norm": 12.515625, "learning_rate": 9.270501705514497e-06, "loss": 0.8565, "step": 21031 }, { "epoch": 2.2379229623324113, "grad_norm": 13.640625, "learning_rate": 9.270466173962479e-06, "loss": 1.084, "step": 21032 }, { "epoch": 2.238029367950628, "grad_norm": 12.2734375, "learning_rate": 9.270430642410462e-06, "loss": 0.8844, "step": 21033 }, { "epoch": 2.2381357735688443, "grad_norm": 11.5625, "learning_rate": 9.270395110858444e-06, "loss": 0.918, "step": 21034 }, { "epoch": 2.2382421791870613, "grad_norm": 10.8203125, "learning_rate": 9.270359579306425e-06, "loss": 0.935, "step": 21035 }, { "epoch": 2.2383485848052778, "grad_norm": 23.671875, "learning_rate": 9.270324047754407e-06, "loss": 1.0857, "step": 21036 }, { "epoch": 2.2384549904234943, "grad_norm": 17.078125, "learning_rate": 9.270288516202389e-06, "loss": 1.018, "step": 21037 }, { "epoch": 2.2385613960417112, "grad_norm": 11.9453125, "learning_rate": 9.27025298465037e-06, "loss": 0.9771, "step": 21038 }, { "epoch": 2.2386678016599277, "grad_norm": 14.5546875, "learning_rate": 9.270217453098352e-06, "loss": 0.9695, "step": 21039 }, { "epoch": 2.2387742072781442, "grad_norm": 10.734375, "learning_rate": 9.270181921546333e-06, "loss": 0.6104, "step": 21040 }, { "epoch": 2.2388806128963608, "grad_norm": 10.984375, "learning_rate": 9.270146389994315e-06, "loss": 0.7905, "step": 21041 }, { "epoch": 2.2389870185145777, "grad_norm": 18.796875, "learning_rate": 9.270110858442297e-06, "loss": 0.9092, "step": 21042 }, { "epoch": 2.239093424132794, "grad_norm": 11.734375, "learning_rate": 9.270075326890278e-06, "loss": 0.8117, "step": 21043 }, { "epoch": 2.2391998297510107, "grad_norm": 12.421875, "learning_rate": 9.270039795338262e-06, "loss": 0.9531, "step": 21044 }, { "epoch": 2.2393062353692277, "grad_norm": 11.5234375, "learning_rate": 9.270004263786243e-06, "loss": 0.9088, "step": 21045 }, { "epoch": 2.239412640987444, "grad_norm": 15.1484375, "learning_rate": 9.269968732234225e-06, "loss": 1.0675, "step": 21046 }, { "epoch": 2.2395190466056607, "grad_norm": 14.515625, "learning_rate": 9.269933200682206e-06, "loss": 1.0371, "step": 21047 }, { "epoch": 2.239625452223877, "grad_norm": 12.5625, "learning_rate": 9.269897669130188e-06, "loss": 0.908, "step": 21048 }, { "epoch": 2.239731857842094, "grad_norm": 12.0546875, "learning_rate": 9.269862137578171e-06, "loss": 0.9408, "step": 21049 }, { "epoch": 2.2398382634603107, "grad_norm": 11.0703125, "learning_rate": 9.269826606026153e-06, "loss": 0.8189, "step": 21050 }, { "epoch": 2.239944669078527, "grad_norm": 12.359375, "learning_rate": 9.269791074474135e-06, "loss": 0.9743, "step": 21051 }, { "epoch": 2.240051074696744, "grad_norm": 12.546875, "learning_rate": 9.269755542922116e-06, "loss": 0.8948, "step": 21052 }, { "epoch": 2.2401574803149606, "grad_norm": 13.4296875, "learning_rate": 9.269720011370098e-06, "loss": 0.8487, "step": 21053 }, { "epoch": 2.240263885933177, "grad_norm": 10.6015625, "learning_rate": 9.26968447981808e-06, "loss": 0.8877, "step": 21054 }, { "epoch": 2.240370291551394, "grad_norm": 10.734375, "learning_rate": 9.269648948266061e-06, "loss": 0.8186, "step": 21055 }, { "epoch": 2.2404766971696106, "grad_norm": 12.0859375, "learning_rate": 9.269613416714043e-06, "loss": 0.8652, "step": 21056 }, { "epoch": 2.240583102787827, "grad_norm": 12.1484375, "learning_rate": 9.269577885162024e-06, "loss": 0.8605, "step": 21057 }, { "epoch": 2.2406895084060436, "grad_norm": 14.5, "learning_rate": 9.269542353610006e-06, "loss": 0.9936, "step": 21058 }, { "epoch": 2.2407959140242606, "grad_norm": 13.453125, "learning_rate": 9.269506822057989e-06, "loss": 0.7776, "step": 21059 }, { "epoch": 2.240902319642477, "grad_norm": 13.4140625, "learning_rate": 9.26947129050597e-06, "loss": 0.8311, "step": 21060 }, { "epoch": 2.2410087252606936, "grad_norm": 15.53125, "learning_rate": 9.269435758953952e-06, "loss": 1.0011, "step": 21061 }, { "epoch": 2.2411151308789106, "grad_norm": 11.4921875, "learning_rate": 9.269400227401934e-06, "loss": 0.7849, "step": 21062 }, { "epoch": 2.241221536497127, "grad_norm": 13.3984375, "learning_rate": 9.269364695849915e-06, "loss": 1.0085, "step": 21063 }, { "epoch": 2.2413279421153436, "grad_norm": 12.5234375, "learning_rate": 9.269329164297897e-06, "loss": 0.8271, "step": 21064 }, { "epoch": 2.2414343477335605, "grad_norm": 13.2890625, "learning_rate": 9.269293632745879e-06, "loss": 0.8747, "step": 21065 }, { "epoch": 2.241540753351777, "grad_norm": 12.734375, "learning_rate": 9.26925810119386e-06, "loss": 0.8312, "step": 21066 }, { "epoch": 2.2416471589699936, "grad_norm": 12.4140625, "learning_rate": 9.269222569641842e-06, "loss": 0.7787, "step": 21067 }, { "epoch": 2.24175356458821, "grad_norm": 13.8046875, "learning_rate": 9.269187038089824e-06, "loss": 0.9178, "step": 21068 }, { "epoch": 2.241859970206427, "grad_norm": 17.40625, "learning_rate": 9.269151506537807e-06, "loss": 1.0319, "step": 21069 }, { "epoch": 2.2419663758246435, "grad_norm": 12.796875, "learning_rate": 9.269115974985788e-06, "loss": 0.8624, "step": 21070 }, { "epoch": 2.24207278144286, "grad_norm": 11.46875, "learning_rate": 9.26908044343377e-06, "loss": 0.7903, "step": 21071 }, { "epoch": 2.242179187061077, "grad_norm": 11.453125, "learning_rate": 9.269044911881752e-06, "loss": 0.8696, "step": 21072 }, { "epoch": 2.2422855926792935, "grad_norm": 15.9453125, "learning_rate": 9.269009380329733e-06, "loss": 0.7196, "step": 21073 }, { "epoch": 2.24239199829751, "grad_norm": 13.8828125, "learning_rate": 9.268973848777717e-06, "loss": 0.92, "step": 21074 }, { "epoch": 2.242498403915727, "grad_norm": 15.4296875, "learning_rate": 9.268938317225698e-06, "loss": 1.0615, "step": 21075 }, { "epoch": 2.2426048095339435, "grad_norm": 14.1796875, "learning_rate": 9.26890278567368e-06, "loss": 0.8237, "step": 21076 }, { "epoch": 2.24271121515216, "grad_norm": 12.78125, "learning_rate": 9.268867254121661e-06, "loss": 0.7869, "step": 21077 }, { "epoch": 2.2428176207703765, "grad_norm": 12.28125, "learning_rate": 9.268831722569643e-06, "loss": 0.8901, "step": 21078 }, { "epoch": 2.2429240263885935, "grad_norm": 16.375, "learning_rate": 9.268796191017625e-06, "loss": 0.839, "step": 21079 }, { "epoch": 2.24303043200681, "grad_norm": 15.5703125, "learning_rate": 9.268760659465606e-06, "loss": 0.9088, "step": 21080 }, { "epoch": 2.2431368376250265, "grad_norm": 12.8984375, "learning_rate": 9.268725127913588e-06, "loss": 0.9772, "step": 21081 }, { "epoch": 2.2432432432432434, "grad_norm": 17.875, "learning_rate": 9.26868959636157e-06, "loss": 1.0279, "step": 21082 }, { "epoch": 2.24334964886146, "grad_norm": 13.8828125, "learning_rate": 9.268654064809551e-06, "loss": 0.7391, "step": 21083 }, { "epoch": 2.2434560544796764, "grad_norm": 14.171875, "learning_rate": 9.268618533257533e-06, "loss": 0.811, "step": 21084 }, { "epoch": 2.243562460097893, "grad_norm": 14.71875, "learning_rate": 9.268583001705516e-06, "loss": 0.9951, "step": 21085 }, { "epoch": 2.24366886571611, "grad_norm": 12.203125, "learning_rate": 9.268547470153498e-06, "loss": 0.9724, "step": 21086 }, { "epoch": 2.2437752713343264, "grad_norm": 18.765625, "learning_rate": 9.268511938601479e-06, "loss": 1.1936, "step": 21087 }, { "epoch": 2.243881676952543, "grad_norm": 14.0546875, "learning_rate": 9.26847640704946e-06, "loss": 0.8461, "step": 21088 }, { "epoch": 2.24398808257076, "grad_norm": 15.515625, "learning_rate": 9.268440875497442e-06, "loss": 0.9297, "step": 21089 }, { "epoch": 2.2440944881889764, "grad_norm": 17.015625, "learning_rate": 9.268405343945424e-06, "loss": 0.9104, "step": 21090 }, { "epoch": 2.244200893807193, "grad_norm": 13.6796875, "learning_rate": 9.268369812393406e-06, "loss": 1.0072, "step": 21091 }, { "epoch": 2.24430729942541, "grad_norm": 14.0234375, "learning_rate": 9.268334280841387e-06, "loss": 1.0024, "step": 21092 }, { "epoch": 2.2444137050436264, "grad_norm": 13.5, "learning_rate": 9.268298749289369e-06, "loss": 0.9424, "step": 21093 }, { "epoch": 2.244520110661843, "grad_norm": 19.1875, "learning_rate": 9.26826321773735e-06, "loss": 0.7854, "step": 21094 }, { "epoch": 2.2446265162800594, "grad_norm": 14.9140625, "learning_rate": 9.268227686185334e-06, "loss": 0.9512, "step": 21095 }, { "epoch": 2.2447329218982763, "grad_norm": 14.7265625, "learning_rate": 9.268192154633315e-06, "loss": 0.79, "step": 21096 }, { "epoch": 2.244839327516493, "grad_norm": 16.25, "learning_rate": 9.268156623081297e-06, "loss": 0.9666, "step": 21097 }, { "epoch": 2.2449457331347094, "grad_norm": 11.3359375, "learning_rate": 9.268121091529278e-06, "loss": 0.8842, "step": 21098 }, { "epoch": 2.2450521387529263, "grad_norm": 11.4765625, "learning_rate": 9.26808555997726e-06, "loss": 0.8009, "step": 21099 }, { "epoch": 2.245158544371143, "grad_norm": 10.3515625, "learning_rate": 9.268050028425243e-06, "loss": 0.793, "step": 21100 }, { "epoch": 2.2452649499893593, "grad_norm": 16.390625, "learning_rate": 9.268014496873225e-06, "loss": 1.095, "step": 21101 }, { "epoch": 2.2453713556075763, "grad_norm": 10.984375, "learning_rate": 9.267978965321207e-06, "loss": 0.8392, "step": 21102 }, { "epoch": 2.245477761225793, "grad_norm": 13.0703125, "learning_rate": 9.267943433769188e-06, "loss": 0.9292, "step": 21103 }, { "epoch": 2.2455841668440093, "grad_norm": 11.59375, "learning_rate": 9.26790790221717e-06, "loss": 0.957, "step": 21104 }, { "epoch": 2.245690572462226, "grad_norm": 15.4140625, "learning_rate": 9.267872370665151e-06, "loss": 0.8758, "step": 21105 }, { "epoch": 2.2457969780804428, "grad_norm": 15.3984375, "learning_rate": 9.267836839113133e-06, "loss": 0.7771, "step": 21106 }, { "epoch": 2.2459033836986593, "grad_norm": 10.7578125, "learning_rate": 9.267801307561115e-06, "loss": 0.9549, "step": 21107 }, { "epoch": 2.246009789316876, "grad_norm": 11.8515625, "learning_rate": 9.267765776009096e-06, "loss": 1.039, "step": 21108 }, { "epoch": 2.2461161949350927, "grad_norm": 12.34375, "learning_rate": 9.267730244457078e-06, "loss": 0.8793, "step": 21109 }, { "epoch": 2.2462226005533092, "grad_norm": 14.859375, "learning_rate": 9.267694712905061e-06, "loss": 0.8255, "step": 21110 }, { "epoch": 2.2463290061715258, "grad_norm": 12.8828125, "learning_rate": 9.267659181353043e-06, "loss": 0.8819, "step": 21111 }, { "epoch": 2.2464354117897427, "grad_norm": 15.1015625, "learning_rate": 9.267623649801024e-06, "loss": 0.9606, "step": 21112 }, { "epoch": 2.246541817407959, "grad_norm": 11.8984375, "learning_rate": 9.267588118249006e-06, "loss": 0.9159, "step": 21113 }, { "epoch": 2.2466482230261757, "grad_norm": 14.8515625, "learning_rate": 9.267552586696988e-06, "loss": 0.8272, "step": 21114 }, { "epoch": 2.2467546286443922, "grad_norm": 13.6875, "learning_rate": 9.26751705514497e-06, "loss": 1.1205, "step": 21115 }, { "epoch": 2.246861034262609, "grad_norm": 15.078125, "learning_rate": 9.26748152359295e-06, "loss": 0.7341, "step": 21116 }, { "epoch": 2.2469674398808257, "grad_norm": 15.9921875, "learning_rate": 9.267445992040932e-06, "loss": 0.9163, "step": 21117 }, { "epoch": 2.247073845499042, "grad_norm": 12.171875, "learning_rate": 9.267410460488914e-06, "loss": 0.9212, "step": 21118 }, { "epoch": 2.247180251117259, "grad_norm": 13.1953125, "learning_rate": 9.267374928936896e-06, "loss": 1.0125, "step": 21119 }, { "epoch": 2.2472866567354757, "grad_norm": 10.875, "learning_rate": 9.267339397384877e-06, "loss": 0.8221, "step": 21120 }, { "epoch": 2.247393062353692, "grad_norm": 15.3828125, "learning_rate": 9.26730386583286e-06, "loss": 0.8148, "step": 21121 }, { "epoch": 2.2474994679719087, "grad_norm": 9.90625, "learning_rate": 9.267268334280842e-06, "loss": 0.8127, "step": 21122 }, { "epoch": 2.2476058735901256, "grad_norm": 10.1484375, "learning_rate": 9.267232802728824e-06, "loss": 0.9426, "step": 21123 }, { "epoch": 2.247712279208342, "grad_norm": 13.8046875, "learning_rate": 9.267197271176805e-06, "loss": 0.92, "step": 21124 }, { "epoch": 2.2478186848265587, "grad_norm": 11.6875, "learning_rate": 9.267161739624787e-06, "loss": 0.7831, "step": 21125 }, { "epoch": 2.2479250904447756, "grad_norm": 14.3359375, "learning_rate": 9.26712620807277e-06, "loss": 0.8759, "step": 21126 }, { "epoch": 2.248031496062992, "grad_norm": 11.8984375, "learning_rate": 9.267090676520752e-06, "loss": 0.8567, "step": 21127 }, { "epoch": 2.2481379016812086, "grad_norm": 12.6484375, "learning_rate": 9.267055144968733e-06, "loss": 0.8411, "step": 21128 }, { "epoch": 2.2482443072994256, "grad_norm": 10.59375, "learning_rate": 9.267019613416715e-06, "loss": 0.6424, "step": 21129 }, { "epoch": 2.248350712917642, "grad_norm": 11.4140625, "learning_rate": 9.266984081864697e-06, "loss": 0.8624, "step": 21130 }, { "epoch": 2.2484571185358586, "grad_norm": 12.4453125, "learning_rate": 9.266948550312678e-06, "loss": 0.7643, "step": 21131 }, { "epoch": 2.2485635241540756, "grad_norm": 12.390625, "learning_rate": 9.26691301876066e-06, "loss": 0.765, "step": 21132 }, { "epoch": 2.248669929772292, "grad_norm": 16.03125, "learning_rate": 9.266877487208641e-06, "loss": 1.0987, "step": 21133 }, { "epoch": 2.2487763353905086, "grad_norm": 15.875, "learning_rate": 9.266841955656623e-06, "loss": 1.1169, "step": 21134 }, { "epoch": 2.248882741008725, "grad_norm": 12.34375, "learning_rate": 9.266806424104605e-06, "loss": 0.9408, "step": 21135 }, { "epoch": 2.248989146626942, "grad_norm": 13.296875, "learning_rate": 9.266770892552588e-06, "loss": 0.9441, "step": 21136 }, { "epoch": 2.2490955522451586, "grad_norm": 12.5625, "learning_rate": 9.26673536100057e-06, "loss": 0.8688, "step": 21137 }, { "epoch": 2.249201957863375, "grad_norm": 13.21875, "learning_rate": 9.266699829448551e-06, "loss": 0.815, "step": 21138 }, { "epoch": 2.249308363481592, "grad_norm": 11.9375, "learning_rate": 9.266664297896533e-06, "loss": 0.9635, "step": 21139 }, { "epoch": 2.2494147690998085, "grad_norm": 13.6015625, "learning_rate": 9.266628766344514e-06, "loss": 0.837, "step": 21140 }, { "epoch": 2.249521174718025, "grad_norm": 10.4765625, "learning_rate": 9.266593234792498e-06, "loss": 0.8163, "step": 21141 }, { "epoch": 2.2496275803362416, "grad_norm": 11.125, "learning_rate": 9.26655770324048e-06, "loss": 0.9292, "step": 21142 }, { "epoch": 2.2497339859544585, "grad_norm": 11.8671875, "learning_rate": 9.266522171688461e-06, "loss": 0.9318, "step": 21143 }, { "epoch": 2.249840391572675, "grad_norm": 15.1328125, "learning_rate": 9.266486640136443e-06, "loss": 0.8792, "step": 21144 }, { "epoch": 2.2499467971908915, "grad_norm": 12.3671875, "learning_rate": 9.266451108584422e-06, "loss": 0.8725, "step": 21145 }, { "epoch": 2.2500532028091085, "grad_norm": 11.5234375, "learning_rate": 9.266415577032406e-06, "loss": 0.9598, "step": 21146 }, { "epoch": 2.250159608427325, "grad_norm": 13.375, "learning_rate": 9.266380045480387e-06, "loss": 0.8086, "step": 21147 }, { "epoch": 2.2502660140455415, "grad_norm": 12.59375, "learning_rate": 9.266344513928369e-06, "loss": 0.7924, "step": 21148 }, { "epoch": 2.2503724196637584, "grad_norm": 10.5, "learning_rate": 9.26630898237635e-06, "loss": 0.8209, "step": 21149 }, { "epoch": 2.250478825281975, "grad_norm": 14.6875, "learning_rate": 9.266273450824332e-06, "loss": 0.8701, "step": 21150 }, { "epoch": 2.2505852309001915, "grad_norm": 11.875, "learning_rate": 9.266237919272315e-06, "loss": 0.9371, "step": 21151 }, { "epoch": 2.250691636518408, "grad_norm": 13.1640625, "learning_rate": 9.266202387720297e-06, "loss": 0.8744, "step": 21152 }, { "epoch": 2.250798042136625, "grad_norm": 13.6875, "learning_rate": 9.266166856168279e-06, "loss": 1.0531, "step": 21153 }, { "epoch": 2.2509044477548414, "grad_norm": 12.953125, "learning_rate": 9.26613132461626e-06, "loss": 0.7866, "step": 21154 }, { "epoch": 2.251010853373058, "grad_norm": 14.8984375, "learning_rate": 9.266095793064242e-06, "loss": 1.0089, "step": 21155 }, { "epoch": 2.251117258991275, "grad_norm": 14.859375, "learning_rate": 9.266060261512224e-06, "loss": 1.0209, "step": 21156 }, { "epoch": 2.2512236646094914, "grad_norm": 13.359375, "learning_rate": 9.266024729960205e-06, "loss": 0.8919, "step": 21157 }, { "epoch": 2.251330070227708, "grad_norm": 12.6953125, "learning_rate": 9.265989198408187e-06, "loss": 0.9212, "step": 21158 }, { "epoch": 2.2514364758459244, "grad_norm": 15.40625, "learning_rate": 9.265953666856168e-06, "loss": 0.9005, "step": 21159 }, { "epoch": 2.2515428814641414, "grad_norm": 15.4140625, "learning_rate": 9.26591813530415e-06, "loss": 0.9301, "step": 21160 }, { "epoch": 2.251649287082358, "grad_norm": 12.296875, "learning_rate": 9.265882603752132e-06, "loss": 0.8717, "step": 21161 }, { "epoch": 2.2517556927005744, "grad_norm": 13.15625, "learning_rate": 9.265847072200115e-06, "loss": 0.908, "step": 21162 }, { "epoch": 2.2518620983187914, "grad_norm": 12.46875, "learning_rate": 9.265811540648096e-06, "loss": 0.88, "step": 21163 }, { "epoch": 2.251968503937008, "grad_norm": 12.0390625, "learning_rate": 9.265776009096078e-06, "loss": 0.9843, "step": 21164 }, { "epoch": 2.2520749095552244, "grad_norm": 14.1953125, "learning_rate": 9.26574047754406e-06, "loss": 0.8079, "step": 21165 }, { "epoch": 2.2521813151734413, "grad_norm": 13.2421875, "learning_rate": 9.265704945992041e-06, "loss": 1.001, "step": 21166 }, { "epoch": 2.252287720791658, "grad_norm": 13.5390625, "learning_rate": 9.265669414440025e-06, "loss": 1.0066, "step": 21167 }, { "epoch": 2.2523941264098744, "grad_norm": 10.3125, "learning_rate": 9.265633882888006e-06, "loss": 0.7914, "step": 21168 }, { "epoch": 2.2525005320280913, "grad_norm": 12.5546875, "learning_rate": 9.265598351335988e-06, "loss": 0.8862, "step": 21169 }, { "epoch": 2.252606937646308, "grad_norm": 13.7109375, "learning_rate": 9.26556281978397e-06, "loss": 0.9775, "step": 21170 }, { "epoch": 2.2527133432645243, "grad_norm": 20.703125, "learning_rate": 9.265527288231951e-06, "loss": 0.8484, "step": 21171 }, { "epoch": 2.252819748882741, "grad_norm": 16.203125, "learning_rate": 9.265491756679933e-06, "loss": 1.0729, "step": 21172 }, { "epoch": 2.252926154500958, "grad_norm": 12.1953125, "learning_rate": 9.265456225127914e-06, "loss": 0.9333, "step": 21173 }, { "epoch": 2.2530325601191743, "grad_norm": 14.9921875, "learning_rate": 9.265420693575896e-06, "loss": 1.118, "step": 21174 }, { "epoch": 2.253138965737391, "grad_norm": 13.109375, "learning_rate": 9.265385162023877e-06, "loss": 0.9345, "step": 21175 }, { "epoch": 2.2532453713556078, "grad_norm": 15.0390625, "learning_rate": 9.265349630471859e-06, "loss": 0.8269, "step": 21176 }, { "epoch": 2.2533517769738243, "grad_norm": 11.1484375, "learning_rate": 9.265314098919842e-06, "loss": 0.9084, "step": 21177 }, { "epoch": 2.253458182592041, "grad_norm": 14.7890625, "learning_rate": 9.265278567367824e-06, "loss": 0.8234, "step": 21178 }, { "epoch": 2.2535645882102573, "grad_norm": 39.03125, "learning_rate": 9.265243035815806e-06, "loss": 0.6523, "step": 21179 }, { "epoch": 2.2536709938284742, "grad_norm": 13.59375, "learning_rate": 9.265207504263787e-06, "loss": 0.8307, "step": 21180 }, { "epoch": 2.2537773994466908, "grad_norm": 11.9375, "learning_rate": 9.265171972711769e-06, "loss": 0.8941, "step": 21181 }, { "epoch": 2.2538838050649073, "grad_norm": 15.5625, "learning_rate": 9.26513644115975e-06, "loss": 0.8041, "step": 21182 }, { "epoch": 2.253990210683124, "grad_norm": 10.3359375, "learning_rate": 9.265100909607732e-06, "loss": 1.0753, "step": 21183 }, { "epoch": 2.2540966163013407, "grad_norm": 10.15625, "learning_rate": 9.265065378055714e-06, "loss": 0.8765, "step": 21184 }, { "epoch": 2.2542030219195572, "grad_norm": 12.828125, "learning_rate": 9.265029846503695e-06, "loss": 0.938, "step": 21185 }, { "epoch": 2.254309427537774, "grad_norm": 15.8046875, "learning_rate": 9.264994314951677e-06, "loss": 0.8351, "step": 21186 }, { "epoch": 2.2544158331559907, "grad_norm": 14.4765625, "learning_rate": 9.26495878339966e-06, "loss": 0.8911, "step": 21187 }, { "epoch": 2.254522238774207, "grad_norm": 13.3359375, "learning_rate": 9.264923251847642e-06, "loss": 0.9312, "step": 21188 }, { "epoch": 2.254628644392424, "grad_norm": 11.71875, "learning_rate": 9.264887720295623e-06, "loss": 0.7555, "step": 21189 }, { "epoch": 2.2547350500106407, "grad_norm": 16.96875, "learning_rate": 9.264852188743605e-06, "loss": 0.9406, "step": 21190 }, { "epoch": 2.254841455628857, "grad_norm": 11.3671875, "learning_rate": 9.264816657191587e-06, "loss": 0.8697, "step": 21191 }, { "epoch": 2.2549478612470737, "grad_norm": 13.8359375, "learning_rate": 9.26478112563957e-06, "loss": 0.8753, "step": 21192 }, { "epoch": 2.2550542668652906, "grad_norm": 11.28125, "learning_rate": 9.264745594087551e-06, "loss": 0.9572, "step": 21193 }, { "epoch": 2.255160672483507, "grad_norm": 13.6484375, "learning_rate": 9.264710062535533e-06, "loss": 0.8459, "step": 21194 }, { "epoch": 2.2552670781017237, "grad_norm": 16.125, "learning_rate": 9.264674530983515e-06, "loss": 0.8855, "step": 21195 }, { "epoch": 2.25537348371994, "grad_norm": 12.5234375, "learning_rate": 9.264638999431496e-06, "loss": 1.0525, "step": 21196 }, { "epoch": 2.255479889338157, "grad_norm": 12.171875, "learning_rate": 9.264603467879478e-06, "loss": 0.9157, "step": 21197 }, { "epoch": 2.2555862949563736, "grad_norm": 7.33203125, "learning_rate": 9.26456793632746e-06, "loss": 0.3792, "step": 21198 }, { "epoch": 2.25569270057459, "grad_norm": 11.953125, "learning_rate": 9.264532404775441e-06, "loss": 0.8856, "step": 21199 }, { "epoch": 2.255799106192807, "grad_norm": 17.6875, "learning_rate": 9.264496873223423e-06, "loss": 0.8831, "step": 21200 }, { "epoch": 2.2559055118110236, "grad_norm": 17.546875, "learning_rate": 9.264461341671404e-06, "loss": 0.905, "step": 21201 }, { "epoch": 2.25601191742924, "grad_norm": 15.7890625, "learning_rate": 9.264425810119386e-06, "loss": 0.9765, "step": 21202 }, { "epoch": 2.256118323047457, "grad_norm": 16.046875, "learning_rate": 9.26439027856737e-06, "loss": 1.0526, "step": 21203 }, { "epoch": 2.2562247286656736, "grad_norm": 17.796875, "learning_rate": 9.26435474701535e-06, "loss": 0.9078, "step": 21204 }, { "epoch": 2.25633113428389, "grad_norm": 12.109375, "learning_rate": 9.264319215463332e-06, "loss": 1.066, "step": 21205 }, { "epoch": 2.256437539902107, "grad_norm": 12.65625, "learning_rate": 9.264283683911314e-06, "loss": 0.9574, "step": 21206 }, { "epoch": 2.2565439455203236, "grad_norm": 13.9296875, "learning_rate": 9.264248152359296e-06, "loss": 0.9921, "step": 21207 }, { "epoch": 2.25665035113854, "grad_norm": 15.625, "learning_rate": 9.264212620807277e-06, "loss": 0.9048, "step": 21208 }, { "epoch": 2.2567567567567566, "grad_norm": 11.640625, "learning_rate": 9.264177089255259e-06, "loss": 0.7957, "step": 21209 }, { "epoch": 2.2568631623749735, "grad_norm": 12.0859375, "learning_rate": 9.26414155770324e-06, "loss": 0.8196, "step": 21210 }, { "epoch": 2.25696956799319, "grad_norm": 11.28125, "learning_rate": 9.264106026151222e-06, "loss": 0.8297, "step": 21211 }, { "epoch": 2.2570759736114065, "grad_norm": 12.9765625, "learning_rate": 9.264070494599204e-06, "loss": 0.8572, "step": 21212 }, { "epoch": 2.2571823792296235, "grad_norm": 9.71875, "learning_rate": 9.264034963047187e-06, "loss": 0.8239, "step": 21213 }, { "epoch": 2.25728878484784, "grad_norm": 11.6796875, "learning_rate": 9.263999431495169e-06, "loss": 0.9243, "step": 21214 }, { "epoch": 2.2573951904660565, "grad_norm": 13.4921875, "learning_rate": 9.26396389994315e-06, "loss": 0.9097, "step": 21215 }, { "epoch": 2.257501596084273, "grad_norm": 12.734375, "learning_rate": 9.263928368391132e-06, "loss": 0.9654, "step": 21216 }, { "epoch": 2.25760800170249, "grad_norm": 13.484375, "learning_rate": 9.263892836839113e-06, "loss": 0.9551, "step": 21217 }, { "epoch": 2.2577144073207065, "grad_norm": 14.984375, "learning_rate": 9.263857305287097e-06, "loss": 1.0227, "step": 21218 }, { "epoch": 2.257820812938923, "grad_norm": 11.21875, "learning_rate": 9.263821773735078e-06, "loss": 0.8026, "step": 21219 }, { "epoch": 2.25792721855714, "grad_norm": 15.3671875, "learning_rate": 9.26378624218306e-06, "loss": 1.2262, "step": 21220 }, { "epoch": 2.2580336241753565, "grad_norm": 11.7890625, "learning_rate": 9.263750710631042e-06, "loss": 0.8825, "step": 21221 }, { "epoch": 2.258140029793573, "grad_norm": 14.4140625, "learning_rate": 9.263715179079023e-06, "loss": 0.9317, "step": 21222 }, { "epoch": 2.25824643541179, "grad_norm": 17.40625, "learning_rate": 9.263679647527005e-06, "loss": 0.8733, "step": 21223 }, { "epoch": 2.2583528410300064, "grad_norm": 16.390625, "learning_rate": 9.263644115974986e-06, "loss": 1.0972, "step": 21224 }, { "epoch": 2.258459246648223, "grad_norm": 9.9765625, "learning_rate": 9.263608584422968e-06, "loss": 0.9456, "step": 21225 }, { "epoch": 2.25856565226644, "grad_norm": 12.4921875, "learning_rate": 9.26357305287095e-06, "loss": 0.7255, "step": 21226 }, { "epoch": 2.2586720578846564, "grad_norm": 13.4453125, "learning_rate": 9.263537521318931e-06, "loss": 1.0051, "step": 21227 }, { "epoch": 2.258778463502873, "grad_norm": 19.453125, "learning_rate": 9.263501989766914e-06, "loss": 1.1899, "step": 21228 }, { "epoch": 2.2588848691210894, "grad_norm": 16.890625, "learning_rate": 9.263466458214896e-06, "loss": 0.8568, "step": 21229 }, { "epoch": 2.2589912747393064, "grad_norm": 15.0, "learning_rate": 9.263430926662878e-06, "loss": 0.8556, "step": 21230 }, { "epoch": 2.259097680357523, "grad_norm": 9.6953125, "learning_rate": 9.26339539511086e-06, "loss": 0.8536, "step": 21231 }, { "epoch": 2.2592040859757394, "grad_norm": 15.25, "learning_rate": 9.263359863558841e-06, "loss": 0.9589, "step": 21232 }, { "epoch": 2.259310491593956, "grad_norm": 11.453125, "learning_rate": 9.263324332006824e-06, "loss": 0.8479, "step": 21233 }, { "epoch": 2.259416897212173, "grad_norm": 20.515625, "learning_rate": 9.263288800454804e-06, "loss": 0.9458, "step": 21234 }, { "epoch": 2.2595233028303894, "grad_norm": 15.109375, "learning_rate": 9.263253268902786e-06, "loss": 1.1589, "step": 21235 }, { "epoch": 2.259629708448606, "grad_norm": 27.65625, "learning_rate": 9.263217737350767e-06, "loss": 0.9539, "step": 21236 }, { "epoch": 2.259736114066823, "grad_norm": 15.328125, "learning_rate": 9.263182205798749e-06, "loss": 1.0273, "step": 21237 }, { "epoch": 2.2598425196850394, "grad_norm": 11.21875, "learning_rate": 9.26314667424673e-06, "loss": 0.8861, "step": 21238 }, { "epoch": 2.259948925303256, "grad_norm": 12.6640625, "learning_rate": 9.263111142694714e-06, "loss": 0.9005, "step": 21239 }, { "epoch": 2.260055330921473, "grad_norm": 9.671875, "learning_rate": 9.263075611142695e-06, "loss": 0.6655, "step": 21240 }, { "epoch": 2.2601617365396893, "grad_norm": 12.8671875, "learning_rate": 9.263040079590677e-06, "loss": 0.7833, "step": 21241 }, { "epoch": 2.260268142157906, "grad_norm": 14.765625, "learning_rate": 9.263004548038659e-06, "loss": 1.0029, "step": 21242 }, { "epoch": 2.260374547776123, "grad_norm": 13.40625, "learning_rate": 9.26296901648664e-06, "loss": 0.9124, "step": 21243 }, { "epoch": 2.2604809533943393, "grad_norm": 17.5625, "learning_rate": 9.262933484934624e-06, "loss": 1.083, "step": 21244 }, { "epoch": 2.260587359012556, "grad_norm": 12.5390625, "learning_rate": 9.262897953382605e-06, "loss": 0.9393, "step": 21245 }, { "epoch": 2.2606937646307723, "grad_norm": 22.5625, "learning_rate": 9.262862421830587e-06, "loss": 0.7588, "step": 21246 }, { "epoch": 2.2608001702489893, "grad_norm": 13.9140625, "learning_rate": 9.262826890278568e-06, "loss": 0.9057, "step": 21247 }, { "epoch": 2.260906575867206, "grad_norm": 18.078125, "learning_rate": 9.26279135872655e-06, "loss": 1.0332, "step": 21248 }, { "epoch": 2.2610129814854223, "grad_norm": 16.75, "learning_rate": 9.262755827174532e-06, "loss": 1.0787, "step": 21249 }, { "epoch": 2.2611193871036392, "grad_norm": 11.4453125, "learning_rate": 9.262720295622513e-06, "loss": 0.789, "step": 21250 }, { "epoch": 2.2612257927218558, "grad_norm": 13.1640625, "learning_rate": 9.262684764070495e-06, "loss": 1.0264, "step": 21251 }, { "epoch": 2.2613321983400723, "grad_norm": 15.734375, "learning_rate": 9.262649232518476e-06, "loss": 1.0404, "step": 21252 }, { "epoch": 2.2614386039582888, "grad_norm": 11.7109375, "learning_rate": 9.262613700966458e-06, "loss": 0.9399, "step": 21253 }, { "epoch": 2.2615450095765057, "grad_norm": 9.0859375, "learning_rate": 9.262578169414441e-06, "loss": 0.8309, "step": 21254 }, { "epoch": 2.2616514151947222, "grad_norm": 13.171875, "learning_rate": 9.262542637862423e-06, "loss": 0.8642, "step": 21255 }, { "epoch": 2.2617578208129387, "grad_norm": 13.0078125, "learning_rate": 9.262507106310405e-06, "loss": 0.9892, "step": 21256 }, { "epoch": 2.2618642264311557, "grad_norm": 11.2421875, "learning_rate": 9.262471574758386e-06, "loss": 0.8297, "step": 21257 }, { "epoch": 2.261970632049372, "grad_norm": 13.671875, "learning_rate": 9.262436043206368e-06, "loss": 1.0643, "step": 21258 }, { "epoch": 2.2620770376675887, "grad_norm": 12.8515625, "learning_rate": 9.262400511654351e-06, "loss": 0.8902, "step": 21259 }, { "epoch": 2.2621834432858057, "grad_norm": 35.84375, "learning_rate": 9.262364980102333e-06, "loss": 0.6933, "step": 21260 }, { "epoch": 2.262289848904022, "grad_norm": 11.734375, "learning_rate": 9.262329448550314e-06, "loss": 0.8173, "step": 21261 }, { "epoch": 2.2623962545222387, "grad_norm": 15.5703125, "learning_rate": 9.262293916998296e-06, "loss": 0.9377, "step": 21262 }, { "epoch": 2.2625026601404556, "grad_norm": 14.0234375, "learning_rate": 9.262258385446277e-06, "loss": 0.9875, "step": 21263 }, { "epoch": 2.262609065758672, "grad_norm": 14.734375, "learning_rate": 9.262222853894259e-06, "loss": 0.9828, "step": 21264 }, { "epoch": 2.2627154713768887, "grad_norm": 12.015625, "learning_rate": 9.26218732234224e-06, "loss": 0.9334, "step": 21265 }, { "epoch": 2.262821876995105, "grad_norm": 30.4375, "learning_rate": 9.262151790790222e-06, "loss": 0.6981, "step": 21266 }, { "epoch": 2.262928282613322, "grad_norm": 13.40625, "learning_rate": 9.262116259238204e-06, "loss": 0.8435, "step": 21267 }, { "epoch": 2.2630346882315386, "grad_norm": 14.28125, "learning_rate": 9.262080727686186e-06, "loss": 1.0274, "step": 21268 }, { "epoch": 2.263141093849755, "grad_norm": 11.7578125, "learning_rate": 9.262045196134169e-06, "loss": 0.7592, "step": 21269 }, { "epoch": 2.263247499467972, "grad_norm": 10.2734375, "learning_rate": 9.26200966458215e-06, "loss": 0.9046, "step": 21270 }, { "epoch": 2.2633539050861886, "grad_norm": 13.4609375, "learning_rate": 9.261974133030132e-06, "loss": 0.8804, "step": 21271 }, { "epoch": 2.263460310704405, "grad_norm": 11.9609375, "learning_rate": 9.261938601478114e-06, "loss": 0.8178, "step": 21272 }, { "epoch": 2.2635667163226216, "grad_norm": 19.296875, "learning_rate": 9.261903069926095e-06, "loss": 0.971, "step": 21273 }, { "epoch": 2.2636731219408386, "grad_norm": 11.984375, "learning_rate": 9.261867538374077e-06, "loss": 1.0217, "step": 21274 }, { "epoch": 2.263779527559055, "grad_norm": 12.8671875, "learning_rate": 9.261832006822058e-06, "loss": 0.934, "step": 21275 }, { "epoch": 2.2638859331772716, "grad_norm": 19.15625, "learning_rate": 9.26179647527004e-06, "loss": 0.7698, "step": 21276 }, { "epoch": 2.2639923387954886, "grad_norm": 12.671875, "learning_rate": 9.261760943718022e-06, "loss": 1.0225, "step": 21277 }, { "epoch": 2.264098744413705, "grad_norm": 15.1171875, "learning_rate": 9.261725412166003e-06, "loss": 0.8696, "step": 21278 }, { "epoch": 2.2642051500319216, "grad_norm": 11.2890625, "learning_rate": 9.261689880613985e-06, "loss": 0.7081, "step": 21279 }, { "epoch": 2.2643115556501385, "grad_norm": 19.953125, "learning_rate": 9.261654349061968e-06, "loss": 0.7438, "step": 21280 }, { "epoch": 2.264417961268355, "grad_norm": 12.9375, "learning_rate": 9.26161881750995e-06, "loss": 0.7394, "step": 21281 }, { "epoch": 2.2645243668865715, "grad_norm": 12.1796875, "learning_rate": 9.261583285957931e-06, "loss": 0.6318, "step": 21282 }, { "epoch": 2.264630772504788, "grad_norm": 10.1875, "learning_rate": 9.261547754405913e-06, "loss": 0.899, "step": 21283 }, { "epoch": 2.264737178123005, "grad_norm": 11.2421875, "learning_rate": 9.261512222853895e-06, "loss": 0.931, "step": 21284 }, { "epoch": 2.2648435837412215, "grad_norm": 12.953125, "learning_rate": 9.261476691301878e-06, "loss": 0.9749, "step": 21285 }, { "epoch": 2.264949989359438, "grad_norm": 15.6796875, "learning_rate": 9.26144115974986e-06, "loss": 1.1369, "step": 21286 }, { "epoch": 2.265056394977655, "grad_norm": 15.015625, "learning_rate": 9.261405628197841e-06, "loss": 0.8839, "step": 21287 }, { "epoch": 2.2651628005958715, "grad_norm": 14.7890625, "learning_rate": 9.261370096645823e-06, "loss": 0.9064, "step": 21288 }, { "epoch": 2.265269206214088, "grad_norm": 14.4140625, "learning_rate": 9.261334565093804e-06, "loss": 0.9435, "step": 21289 }, { "epoch": 2.2653756118323045, "grad_norm": 13.7578125, "learning_rate": 9.261299033541786e-06, "loss": 1.0149, "step": 21290 }, { "epoch": 2.2654820174505215, "grad_norm": 9.9921875, "learning_rate": 9.261263501989768e-06, "loss": 0.7362, "step": 21291 }, { "epoch": 2.265588423068738, "grad_norm": 11.1328125, "learning_rate": 9.261227970437749e-06, "loss": 0.8889, "step": 21292 }, { "epoch": 2.2656948286869545, "grad_norm": 12.640625, "learning_rate": 9.26119243888573e-06, "loss": 0.8979, "step": 21293 }, { "epoch": 2.2658012343051714, "grad_norm": 13.2265625, "learning_rate": 9.261156907333712e-06, "loss": 0.768, "step": 21294 }, { "epoch": 2.265907639923388, "grad_norm": 11.609375, "learning_rate": 9.261121375781696e-06, "loss": 0.84, "step": 21295 }, { "epoch": 2.2660140455416045, "grad_norm": 10.1796875, "learning_rate": 9.261085844229677e-06, "loss": 0.8591, "step": 21296 }, { "epoch": 2.2661204511598214, "grad_norm": 12.921875, "learning_rate": 9.261050312677659e-06, "loss": 0.9844, "step": 21297 }, { "epoch": 2.266226856778038, "grad_norm": 10.8984375, "learning_rate": 9.26101478112564e-06, "loss": 0.8028, "step": 21298 }, { "epoch": 2.2663332623962544, "grad_norm": 12.234375, "learning_rate": 9.260979249573622e-06, "loss": 0.9585, "step": 21299 }, { "epoch": 2.2664396680144714, "grad_norm": 11.875, "learning_rate": 9.260943718021604e-06, "loss": 1.0091, "step": 21300 }, { "epoch": 2.266546073632688, "grad_norm": 17.828125, "learning_rate": 9.260908186469585e-06, "loss": 0.7811, "step": 21301 }, { "epoch": 2.2666524792509044, "grad_norm": 17.828125, "learning_rate": 9.260872654917567e-06, "loss": 1.023, "step": 21302 }, { "epoch": 2.266758884869121, "grad_norm": 18.15625, "learning_rate": 9.260837123365549e-06, "loss": 0.8911, "step": 21303 }, { "epoch": 2.266865290487338, "grad_norm": 13.4375, "learning_rate": 9.26080159181353e-06, "loss": 0.8417, "step": 21304 }, { "epoch": 2.2669716961055544, "grad_norm": 18.25, "learning_rate": 9.260766060261513e-06, "loss": 0.987, "step": 21305 }, { "epoch": 2.267078101723771, "grad_norm": 17.625, "learning_rate": 9.260730528709495e-06, "loss": 0.9336, "step": 21306 }, { "epoch": 2.267184507341988, "grad_norm": 12.0703125, "learning_rate": 9.260694997157477e-06, "loss": 0.9449, "step": 21307 }, { "epoch": 2.2672909129602044, "grad_norm": 12.90625, "learning_rate": 9.260659465605458e-06, "loss": 0.8969, "step": 21308 }, { "epoch": 2.267397318578421, "grad_norm": 10.90625, "learning_rate": 9.26062393405344e-06, "loss": 0.7709, "step": 21309 }, { "epoch": 2.2675037241966374, "grad_norm": 12.8671875, "learning_rate": 9.260588402501423e-06, "loss": 0.9306, "step": 21310 }, { "epoch": 2.2676101298148543, "grad_norm": 11.828125, "learning_rate": 9.260552870949405e-06, "loss": 0.8818, "step": 21311 }, { "epoch": 2.267716535433071, "grad_norm": 15.765625, "learning_rate": 9.260517339397386e-06, "loss": 1.0073, "step": 21312 }, { "epoch": 2.2678229410512873, "grad_norm": 11.265625, "learning_rate": 9.260481807845368e-06, "loss": 0.7827, "step": 21313 }, { "epoch": 2.2679293466695043, "grad_norm": 9.4609375, "learning_rate": 9.26044627629335e-06, "loss": 0.8456, "step": 21314 }, { "epoch": 2.268035752287721, "grad_norm": 14.9609375, "learning_rate": 9.260410744741331e-06, "loss": 0.8861, "step": 21315 }, { "epoch": 2.2681421579059373, "grad_norm": 13.7109375, "learning_rate": 9.260375213189313e-06, "loss": 0.9521, "step": 21316 }, { "epoch": 2.2682485635241543, "grad_norm": 13.8203125, "learning_rate": 9.260339681637294e-06, "loss": 0.7951, "step": 21317 }, { "epoch": 2.268354969142371, "grad_norm": 12.640625, "learning_rate": 9.260304150085276e-06, "loss": 0.9164, "step": 21318 }, { "epoch": 2.2684613747605873, "grad_norm": 23.15625, "learning_rate": 9.260268618533258e-06, "loss": 0.9885, "step": 21319 }, { "epoch": 2.2685677803788042, "grad_norm": 10.484375, "learning_rate": 9.26023308698124e-06, "loss": 0.8163, "step": 21320 }, { "epoch": 2.2686741859970208, "grad_norm": 10.6015625, "learning_rate": 9.260197555429223e-06, "loss": 0.7406, "step": 21321 }, { "epoch": 2.2687805916152373, "grad_norm": 11.59375, "learning_rate": 9.260162023877204e-06, "loss": 0.7227, "step": 21322 }, { "epoch": 2.2688869972334538, "grad_norm": 12.09375, "learning_rate": 9.260126492325186e-06, "loss": 0.9537, "step": 21323 }, { "epoch": 2.2689934028516707, "grad_norm": 13.84375, "learning_rate": 9.260090960773167e-06, "loss": 1.0338, "step": 21324 }, { "epoch": 2.2690998084698872, "grad_norm": 10.3515625, "learning_rate": 9.260055429221149e-06, "loss": 0.7987, "step": 21325 }, { "epoch": 2.2692062140881037, "grad_norm": 15.046875, "learning_rate": 9.26001989766913e-06, "loss": 0.9994, "step": 21326 }, { "epoch": 2.2693126197063203, "grad_norm": 10.34375, "learning_rate": 9.259984366117112e-06, "loss": 0.953, "step": 21327 }, { "epoch": 2.269419025324537, "grad_norm": 15.171875, "learning_rate": 9.259948834565094e-06, "loss": 1.0127, "step": 21328 }, { "epoch": 2.2695254309427537, "grad_norm": 11.8203125, "learning_rate": 9.259913303013075e-06, "loss": 0.8629, "step": 21329 }, { "epoch": 2.2696318365609702, "grad_norm": 14.875, "learning_rate": 9.259877771461057e-06, "loss": 0.8686, "step": 21330 }, { "epoch": 2.269738242179187, "grad_norm": 13.4140625, "learning_rate": 9.25984223990904e-06, "loss": 0.8885, "step": 21331 }, { "epoch": 2.2698446477974037, "grad_norm": 12.375, "learning_rate": 9.259806708357022e-06, "loss": 0.9251, "step": 21332 }, { "epoch": 2.26995105341562, "grad_norm": 11.0859375, "learning_rate": 9.259771176805003e-06, "loss": 0.884, "step": 21333 }, { "epoch": 2.270057459033837, "grad_norm": 17.765625, "learning_rate": 9.259735645252985e-06, "loss": 0.9045, "step": 21334 }, { "epoch": 2.2701638646520537, "grad_norm": 14.890625, "learning_rate": 9.259700113700967e-06, "loss": 1.1344, "step": 21335 }, { "epoch": 2.27027027027027, "grad_norm": 11.7421875, "learning_rate": 9.25966458214895e-06, "loss": 0.848, "step": 21336 }, { "epoch": 2.270376675888487, "grad_norm": 15.4453125, "learning_rate": 9.259629050596932e-06, "loss": 0.8628, "step": 21337 }, { "epoch": 2.2704830815067036, "grad_norm": 11.46875, "learning_rate": 9.259593519044913e-06, "loss": 0.8553, "step": 21338 }, { "epoch": 2.27058948712492, "grad_norm": 11.6953125, "learning_rate": 9.259557987492895e-06, "loss": 0.918, "step": 21339 }, { "epoch": 2.2706958927431367, "grad_norm": 13.2578125, "learning_rate": 9.259522455940876e-06, "loss": 1.0463, "step": 21340 }, { "epoch": 2.2708022983613536, "grad_norm": 11.6171875, "learning_rate": 9.259486924388858e-06, "loss": 0.9026, "step": 21341 }, { "epoch": 2.27090870397957, "grad_norm": 20.171875, "learning_rate": 9.25945139283684e-06, "loss": 0.7005, "step": 21342 }, { "epoch": 2.2710151095977866, "grad_norm": 14.7890625, "learning_rate": 9.259415861284821e-06, "loss": 0.7867, "step": 21343 }, { "epoch": 2.2711215152160036, "grad_norm": 12.0625, "learning_rate": 9.259380329732803e-06, "loss": 0.8177, "step": 21344 }, { "epoch": 2.27122792083422, "grad_norm": 13.8359375, "learning_rate": 9.259344798180784e-06, "loss": 0.8716, "step": 21345 }, { "epoch": 2.2713343264524366, "grad_norm": 11.015625, "learning_rate": 9.259309266628768e-06, "loss": 0.9287, "step": 21346 }, { "epoch": 2.271440732070653, "grad_norm": 15.7109375, "learning_rate": 9.25927373507675e-06, "loss": 1.0084, "step": 21347 }, { "epoch": 2.27154713768887, "grad_norm": 13.0546875, "learning_rate": 9.259238203524731e-06, "loss": 0.8593, "step": 21348 }, { "epoch": 2.2716535433070866, "grad_norm": 10.5703125, "learning_rate": 9.259202671972713e-06, "loss": 0.8779, "step": 21349 }, { "epoch": 2.271759948925303, "grad_norm": 10.7734375, "learning_rate": 9.259167140420694e-06, "loss": 0.7413, "step": 21350 }, { "epoch": 2.27186635454352, "grad_norm": 14.3359375, "learning_rate": 9.259131608868677e-06, "loss": 0.8378, "step": 21351 }, { "epoch": 2.2719727601617365, "grad_norm": 12.71875, "learning_rate": 9.259096077316659e-06, "loss": 1.0206, "step": 21352 }, { "epoch": 2.272079165779953, "grad_norm": 12.1015625, "learning_rate": 9.25906054576464e-06, "loss": 0.9491, "step": 21353 }, { "epoch": 2.27218557139817, "grad_norm": 13.3125, "learning_rate": 9.25902501421262e-06, "loss": 0.9652, "step": 21354 }, { "epoch": 2.2722919770163865, "grad_norm": 13.9609375, "learning_rate": 9.258989482660602e-06, "loss": 0.8594, "step": 21355 }, { "epoch": 2.272398382634603, "grad_norm": 11.671875, "learning_rate": 9.258953951108584e-06, "loss": 1.029, "step": 21356 }, { "epoch": 2.27250478825282, "grad_norm": 14.296875, "learning_rate": 9.258918419556567e-06, "loss": 0.9669, "step": 21357 }, { "epoch": 2.2726111938710365, "grad_norm": 12.078125, "learning_rate": 9.258882888004549e-06, "loss": 0.8915, "step": 21358 }, { "epoch": 2.272717599489253, "grad_norm": 10.390625, "learning_rate": 9.25884735645253e-06, "loss": 0.7859, "step": 21359 }, { "epoch": 2.2728240051074695, "grad_norm": 11.7421875, "learning_rate": 9.258811824900512e-06, "loss": 0.8911, "step": 21360 }, { "epoch": 2.2729304107256865, "grad_norm": 12.1640625, "learning_rate": 9.258776293348494e-06, "loss": 1.0535, "step": 21361 }, { "epoch": 2.273036816343903, "grad_norm": 23.5, "learning_rate": 9.258740761796477e-06, "loss": 1.0645, "step": 21362 }, { "epoch": 2.2731432219621195, "grad_norm": 13.1015625, "learning_rate": 9.258705230244458e-06, "loss": 0.7781, "step": 21363 }, { "epoch": 2.273249627580336, "grad_norm": 13.46875, "learning_rate": 9.25866969869244e-06, "loss": 0.9351, "step": 21364 }, { "epoch": 2.273356033198553, "grad_norm": 16.828125, "learning_rate": 9.258634167140422e-06, "loss": 1.0993, "step": 21365 }, { "epoch": 2.2734624388167695, "grad_norm": 15.8515625, "learning_rate": 9.258598635588403e-06, "loss": 1.0541, "step": 21366 }, { "epoch": 2.273568844434986, "grad_norm": 12.9765625, "learning_rate": 9.258563104036385e-06, "loss": 1.1457, "step": 21367 }, { "epoch": 2.273675250053203, "grad_norm": 15.0078125, "learning_rate": 9.258527572484366e-06, "loss": 0.9959, "step": 21368 }, { "epoch": 2.2737816556714194, "grad_norm": 11.5078125, "learning_rate": 9.258492040932348e-06, "loss": 0.8941, "step": 21369 }, { "epoch": 2.273888061289636, "grad_norm": 13.046875, "learning_rate": 9.25845650938033e-06, "loss": 0.9844, "step": 21370 }, { "epoch": 2.273994466907853, "grad_norm": 12.15625, "learning_rate": 9.258420977828311e-06, "loss": 0.9412, "step": 21371 }, { "epoch": 2.2741008725260694, "grad_norm": 10.15625, "learning_rate": 9.258385446276295e-06, "loss": 0.7284, "step": 21372 }, { "epoch": 2.274207278144286, "grad_norm": 13.28125, "learning_rate": 9.258349914724276e-06, "loss": 0.921, "step": 21373 }, { "epoch": 2.274313683762503, "grad_norm": 12.71875, "learning_rate": 9.258314383172258e-06, "loss": 1.009, "step": 21374 }, { "epoch": 2.2744200893807194, "grad_norm": 9.6171875, "learning_rate": 9.25827885162024e-06, "loss": 0.8573, "step": 21375 }, { "epoch": 2.274526494998936, "grad_norm": 16.265625, "learning_rate": 9.258243320068221e-06, "loss": 0.8497, "step": 21376 }, { "epoch": 2.2746329006171524, "grad_norm": 10.0625, "learning_rate": 9.258207788516204e-06, "loss": 0.7478, "step": 21377 }, { "epoch": 2.2747393062353694, "grad_norm": 12.5625, "learning_rate": 9.258172256964186e-06, "loss": 0.9075, "step": 21378 }, { "epoch": 2.274845711853586, "grad_norm": 12.046875, "learning_rate": 9.258136725412168e-06, "loss": 0.901, "step": 21379 }, { "epoch": 2.2749521174718024, "grad_norm": 9.5546875, "learning_rate": 9.25810119386015e-06, "loss": 0.8131, "step": 21380 }, { "epoch": 2.2750585230900193, "grad_norm": 11.625, "learning_rate": 9.25806566230813e-06, "loss": 0.849, "step": 21381 }, { "epoch": 2.275164928708236, "grad_norm": 16.46875, "learning_rate": 9.258030130756112e-06, "loss": 0.9297, "step": 21382 }, { "epoch": 2.2752713343264523, "grad_norm": 26.859375, "learning_rate": 9.257994599204094e-06, "loss": 0.9644, "step": 21383 }, { "epoch": 2.275377739944669, "grad_norm": 13.609375, "learning_rate": 9.257959067652076e-06, "loss": 0.9626, "step": 21384 }, { "epoch": 2.275484145562886, "grad_norm": 20.6875, "learning_rate": 9.257923536100057e-06, "loss": 0.8529, "step": 21385 }, { "epoch": 2.2755905511811023, "grad_norm": 16.125, "learning_rate": 9.257888004548039e-06, "loss": 0.9597, "step": 21386 }, { "epoch": 2.275696956799319, "grad_norm": 12.1875, "learning_rate": 9.257852472996022e-06, "loss": 0.8234, "step": 21387 }, { "epoch": 2.275803362417536, "grad_norm": 22.796875, "learning_rate": 9.257816941444004e-06, "loss": 1.0045, "step": 21388 }, { "epoch": 2.2759097680357523, "grad_norm": 12.9765625, "learning_rate": 9.257781409891985e-06, "loss": 0.8676, "step": 21389 }, { "epoch": 2.276016173653969, "grad_norm": 13.8125, "learning_rate": 9.257745878339967e-06, "loss": 0.8413, "step": 21390 }, { "epoch": 2.2761225792721858, "grad_norm": 12.65625, "learning_rate": 9.257710346787949e-06, "loss": 0.8017, "step": 21391 }, { "epoch": 2.2762289848904023, "grad_norm": 18.09375, "learning_rate": 9.25767481523593e-06, "loss": 1.1468, "step": 21392 }, { "epoch": 2.2763353905086188, "grad_norm": 16.46875, "learning_rate": 9.257639283683912e-06, "loss": 0.7788, "step": 21393 }, { "epoch": 2.2764417961268357, "grad_norm": 16.125, "learning_rate": 9.257603752131893e-06, "loss": 1.0987, "step": 21394 }, { "epoch": 2.2765482017450522, "grad_norm": 15.4140625, "learning_rate": 9.257568220579875e-06, "loss": 0.9233, "step": 21395 }, { "epoch": 2.2766546073632687, "grad_norm": 14.2578125, "learning_rate": 9.257532689027857e-06, "loss": 1.0168, "step": 21396 }, { "epoch": 2.2767610129814853, "grad_norm": 10.7890625, "learning_rate": 9.257497157475838e-06, "loss": 0.7081, "step": 21397 }, { "epoch": 2.276867418599702, "grad_norm": 11.640625, "learning_rate": 9.257461625923821e-06, "loss": 0.8237, "step": 21398 }, { "epoch": 2.2769738242179187, "grad_norm": 12.9453125, "learning_rate": 9.257426094371803e-06, "loss": 1.0247, "step": 21399 }, { "epoch": 2.2770802298361352, "grad_norm": 11.5546875, "learning_rate": 9.257390562819785e-06, "loss": 0.8429, "step": 21400 }, { "epoch": 2.277186635454352, "grad_norm": 13.8125, "learning_rate": 9.257355031267766e-06, "loss": 0.7765, "step": 21401 }, { "epoch": 2.2772930410725687, "grad_norm": 12.3828125, "learning_rate": 9.257319499715748e-06, "loss": 0.9074, "step": 21402 }, { "epoch": 2.277399446690785, "grad_norm": 18.984375, "learning_rate": 9.257283968163731e-06, "loss": 1.1794, "step": 21403 }, { "epoch": 2.2775058523090017, "grad_norm": 10.3125, "learning_rate": 9.257248436611713e-06, "loss": 0.8103, "step": 21404 }, { "epoch": 2.2776122579272187, "grad_norm": 13.6875, "learning_rate": 9.257212905059694e-06, "loss": 0.8448, "step": 21405 }, { "epoch": 2.277718663545435, "grad_norm": 12.4296875, "learning_rate": 9.257177373507676e-06, "loss": 1.1093, "step": 21406 }, { "epoch": 2.2778250691636517, "grad_norm": 11.453125, "learning_rate": 9.257141841955658e-06, "loss": 0.9878, "step": 21407 }, { "epoch": 2.2779314747818686, "grad_norm": 12.109375, "learning_rate": 9.25710631040364e-06, "loss": 0.9025, "step": 21408 }, { "epoch": 2.278037880400085, "grad_norm": 12.8046875, "learning_rate": 9.25707077885162e-06, "loss": 0.9107, "step": 21409 }, { "epoch": 2.2781442860183017, "grad_norm": 12.3671875, "learning_rate": 9.257035247299602e-06, "loss": 0.8674, "step": 21410 }, { "epoch": 2.2782506916365186, "grad_norm": 14.8046875, "learning_rate": 9.256999715747584e-06, "loss": 1.0049, "step": 21411 }, { "epoch": 2.278357097254735, "grad_norm": 11.5234375, "learning_rate": 9.256964184195566e-06, "loss": 0.8541, "step": 21412 }, { "epoch": 2.2784635028729516, "grad_norm": 11.4765625, "learning_rate": 9.256928652643549e-06, "loss": 0.7517, "step": 21413 }, { "epoch": 2.2785699084911686, "grad_norm": 12.71875, "learning_rate": 9.25689312109153e-06, "loss": 0.8302, "step": 21414 }, { "epoch": 2.278676314109385, "grad_norm": 11.3515625, "learning_rate": 9.256857589539512e-06, "loss": 0.7953, "step": 21415 }, { "epoch": 2.2787827197276016, "grad_norm": 12.296875, "learning_rate": 9.256822057987494e-06, "loss": 0.819, "step": 21416 }, { "epoch": 2.278889125345818, "grad_norm": 20.90625, "learning_rate": 9.256786526435475e-06, "loss": 0.8776, "step": 21417 }, { "epoch": 2.278995530964035, "grad_norm": 12.453125, "learning_rate": 9.256750994883457e-06, "loss": 0.7594, "step": 21418 }, { "epoch": 2.2791019365822516, "grad_norm": 13.3828125, "learning_rate": 9.256715463331439e-06, "loss": 0.8557, "step": 21419 }, { "epoch": 2.279208342200468, "grad_norm": 8.734375, "learning_rate": 9.25667993177942e-06, "loss": 0.6517, "step": 21420 }, { "epoch": 2.2793147478186846, "grad_norm": 9.3984375, "learning_rate": 9.256644400227402e-06, "loss": 0.8204, "step": 21421 }, { "epoch": 2.2794211534369015, "grad_norm": 15.90625, "learning_rate": 9.256608868675383e-06, "loss": 1.2173, "step": 21422 }, { "epoch": 2.279527559055118, "grad_norm": 10.4609375, "learning_rate": 9.256573337123367e-06, "loss": 0.8577, "step": 21423 }, { "epoch": 2.2796339646733346, "grad_norm": 12.3125, "learning_rate": 9.256537805571348e-06, "loss": 0.8773, "step": 21424 }, { "epoch": 2.2797403702915515, "grad_norm": 12.6171875, "learning_rate": 9.25650227401933e-06, "loss": 0.8999, "step": 21425 }, { "epoch": 2.279846775909768, "grad_norm": 9.7109375, "learning_rate": 9.256466742467312e-06, "loss": 0.7774, "step": 21426 }, { "epoch": 2.2799531815279845, "grad_norm": 14.53125, "learning_rate": 9.256431210915293e-06, "loss": 0.8565, "step": 21427 }, { "epoch": 2.2800595871462015, "grad_norm": 12.140625, "learning_rate": 9.256395679363276e-06, "loss": 0.8374, "step": 21428 }, { "epoch": 2.280165992764418, "grad_norm": 12.9453125, "learning_rate": 9.256360147811258e-06, "loss": 0.8951, "step": 21429 }, { "epoch": 2.2802723983826345, "grad_norm": 13.1015625, "learning_rate": 9.25632461625924e-06, "loss": 0.9316, "step": 21430 }, { "epoch": 2.2803788040008515, "grad_norm": 15.640625, "learning_rate": 9.256289084707221e-06, "loss": 1.0062, "step": 21431 }, { "epoch": 2.280485209619068, "grad_norm": 10.5234375, "learning_rate": 9.256253553155203e-06, "loss": 0.8874, "step": 21432 }, { "epoch": 2.2805916152372845, "grad_norm": 13.1953125, "learning_rate": 9.256218021603184e-06, "loss": 0.8346, "step": 21433 }, { "epoch": 2.280698020855501, "grad_norm": 14.4140625, "learning_rate": 9.256182490051166e-06, "loss": 0.8728, "step": 21434 }, { "epoch": 2.280804426473718, "grad_norm": 11.9921875, "learning_rate": 9.256146958499148e-06, "loss": 0.7186, "step": 21435 }, { "epoch": 2.2809108320919345, "grad_norm": 17.71875, "learning_rate": 9.25611142694713e-06, "loss": 0.9017, "step": 21436 }, { "epoch": 2.281017237710151, "grad_norm": 12.4609375, "learning_rate": 9.256075895395111e-06, "loss": 0.9063, "step": 21437 }, { "epoch": 2.281123643328368, "grad_norm": 13.8125, "learning_rate": 9.256040363843093e-06, "loss": 0.889, "step": 21438 }, { "epoch": 2.2812300489465844, "grad_norm": 14.65625, "learning_rate": 9.256004832291076e-06, "loss": 1.0422, "step": 21439 }, { "epoch": 2.281336454564801, "grad_norm": 19.421875, "learning_rate": 9.255969300739057e-06, "loss": 0.9893, "step": 21440 }, { "epoch": 2.2814428601830175, "grad_norm": 16.640625, "learning_rate": 9.255933769187039e-06, "loss": 0.9535, "step": 21441 }, { "epoch": 2.2815492658012344, "grad_norm": 19.28125, "learning_rate": 9.25589823763502e-06, "loss": 1.0088, "step": 21442 }, { "epoch": 2.281655671419451, "grad_norm": 16.0625, "learning_rate": 9.255862706083002e-06, "loss": 0.7723, "step": 21443 }, { "epoch": 2.2817620770376674, "grad_norm": 13.828125, "learning_rate": 9.255827174530984e-06, "loss": 0.9131, "step": 21444 }, { "epoch": 2.2818684826558844, "grad_norm": 16.0, "learning_rate": 9.255791642978965e-06, "loss": 0.7819, "step": 21445 }, { "epoch": 2.281974888274101, "grad_norm": 11.1015625, "learning_rate": 9.255756111426947e-06, "loss": 0.767, "step": 21446 }, { "epoch": 2.2820812938923174, "grad_norm": 13.859375, "learning_rate": 9.255720579874929e-06, "loss": 0.9603, "step": 21447 }, { "epoch": 2.2821876995105344, "grad_norm": 10.96875, "learning_rate": 9.25568504832291e-06, "loss": 0.9194, "step": 21448 }, { "epoch": 2.282294105128751, "grad_norm": 16.875, "learning_rate": 9.255649516770894e-06, "loss": 0.9074, "step": 21449 }, { "epoch": 2.2824005107469674, "grad_norm": 14.390625, "learning_rate": 9.255613985218875e-06, "loss": 1.0074, "step": 21450 }, { "epoch": 2.2825069163651843, "grad_norm": 14.9140625, "learning_rate": 9.255578453666857e-06, "loss": 0.9348, "step": 21451 }, { "epoch": 2.282613321983401, "grad_norm": 10.59375, "learning_rate": 9.255542922114838e-06, "loss": 0.8736, "step": 21452 }, { "epoch": 2.2827197276016173, "grad_norm": 18.234375, "learning_rate": 9.25550739056282e-06, "loss": 0.7998, "step": 21453 }, { "epoch": 2.282826133219834, "grad_norm": 17.796875, "learning_rate": 9.255471859010803e-06, "loss": 0.9955, "step": 21454 }, { "epoch": 2.282932538838051, "grad_norm": 20.890625, "learning_rate": 9.255436327458785e-06, "loss": 1.0297, "step": 21455 }, { "epoch": 2.2830389444562673, "grad_norm": 17.859375, "learning_rate": 9.255400795906767e-06, "loss": 0.9502, "step": 21456 }, { "epoch": 2.283145350074484, "grad_norm": 21.484375, "learning_rate": 9.255365264354748e-06, "loss": 0.9105, "step": 21457 }, { "epoch": 2.2832517556927003, "grad_norm": 11.84375, "learning_rate": 9.25532973280273e-06, "loss": 0.8446, "step": 21458 }, { "epoch": 2.2833581613109173, "grad_norm": 17.671875, "learning_rate": 9.255294201250711e-06, "loss": 0.8886, "step": 21459 }, { "epoch": 2.283464566929134, "grad_norm": 12.8203125, "learning_rate": 9.255258669698693e-06, "loss": 0.9434, "step": 21460 }, { "epoch": 2.2835709725473503, "grad_norm": 14.65625, "learning_rate": 9.255223138146675e-06, "loss": 0.8216, "step": 21461 }, { "epoch": 2.2836773781655673, "grad_norm": 15.3359375, "learning_rate": 9.255187606594656e-06, "loss": 0.9233, "step": 21462 }, { "epoch": 2.2837837837837838, "grad_norm": 13.84375, "learning_rate": 9.255152075042638e-06, "loss": 0.9394, "step": 21463 }, { "epoch": 2.2838901894020003, "grad_norm": 15.171875, "learning_rate": 9.255116543490621e-06, "loss": 0.8781, "step": 21464 }, { "epoch": 2.2839965950202172, "grad_norm": 11.0390625, "learning_rate": 9.255081011938603e-06, "loss": 0.7679, "step": 21465 }, { "epoch": 2.2841030006384337, "grad_norm": 14.5625, "learning_rate": 9.255045480386584e-06, "loss": 0.7279, "step": 21466 }, { "epoch": 2.2842094062566503, "grad_norm": 20.171875, "learning_rate": 9.255009948834566e-06, "loss": 0.8571, "step": 21467 }, { "epoch": 2.284315811874867, "grad_norm": 17.078125, "learning_rate": 9.254974417282547e-06, "loss": 0.9951, "step": 21468 }, { "epoch": 2.2844222174930837, "grad_norm": 14.2734375, "learning_rate": 9.25493888573053e-06, "loss": 0.9235, "step": 21469 }, { "epoch": 2.2845286231113002, "grad_norm": 14.375, "learning_rate": 9.254903354178512e-06, "loss": 0.9938, "step": 21470 }, { "epoch": 2.2846350287295167, "grad_norm": 16.484375, "learning_rate": 9.254867822626494e-06, "loss": 0.7737, "step": 21471 }, { "epoch": 2.2847414343477337, "grad_norm": 11.2421875, "learning_rate": 9.254832291074476e-06, "loss": 0.6627, "step": 21472 }, { "epoch": 2.28484783996595, "grad_norm": 15.109375, "learning_rate": 9.254796759522457e-06, "loss": 0.9353, "step": 21473 }, { "epoch": 2.2849542455841667, "grad_norm": 14.625, "learning_rate": 9.254761227970437e-06, "loss": 0.7308, "step": 21474 }, { "epoch": 2.2850606512023837, "grad_norm": 9.421875, "learning_rate": 9.25472569641842e-06, "loss": 0.7819, "step": 21475 }, { "epoch": 2.2851670568206, "grad_norm": 13.140625, "learning_rate": 9.254690164866402e-06, "loss": 0.7985, "step": 21476 }, { "epoch": 2.2852734624388167, "grad_norm": 12.9453125, "learning_rate": 9.254654633314384e-06, "loss": 0.7565, "step": 21477 }, { "epoch": 2.285379868057033, "grad_norm": 14.953125, "learning_rate": 9.254619101762365e-06, "loss": 0.7503, "step": 21478 }, { "epoch": 2.28548627367525, "grad_norm": 9.640625, "learning_rate": 9.254583570210347e-06, "loss": 0.7182, "step": 21479 }, { "epoch": 2.2855926792934667, "grad_norm": 12.484375, "learning_rate": 9.25454803865833e-06, "loss": 1.0139, "step": 21480 }, { "epoch": 2.285699084911683, "grad_norm": 13.1015625, "learning_rate": 9.254512507106312e-06, "loss": 0.9002, "step": 21481 }, { "epoch": 2.2858054905299, "grad_norm": 10.046875, "learning_rate": 9.254476975554293e-06, "loss": 0.9249, "step": 21482 }, { "epoch": 2.2859118961481166, "grad_norm": 10.3203125, "learning_rate": 9.254441444002275e-06, "loss": 0.7831, "step": 21483 }, { "epoch": 2.286018301766333, "grad_norm": 13.3515625, "learning_rate": 9.254405912450257e-06, "loss": 0.874, "step": 21484 }, { "epoch": 2.28612470738455, "grad_norm": 14.265625, "learning_rate": 9.254370380898238e-06, "loss": 0.9662, "step": 21485 }, { "epoch": 2.2862311130027666, "grad_norm": 7.1796875, "learning_rate": 9.25433484934622e-06, "loss": 0.6509, "step": 21486 }, { "epoch": 2.286337518620983, "grad_norm": 18.046875, "learning_rate": 9.254299317794201e-06, "loss": 1.0592, "step": 21487 }, { "epoch": 2.2864439242392, "grad_norm": 13.2421875, "learning_rate": 9.254263786242183e-06, "loss": 0.9113, "step": 21488 }, { "epoch": 2.2865503298574166, "grad_norm": 13.7890625, "learning_rate": 9.254228254690165e-06, "loss": 0.8051, "step": 21489 }, { "epoch": 2.286656735475633, "grad_norm": 16.21875, "learning_rate": 9.254192723138148e-06, "loss": 1.0279, "step": 21490 }, { "epoch": 2.2867631410938496, "grad_norm": 15.4453125, "learning_rate": 9.25415719158613e-06, "loss": 1.0948, "step": 21491 }, { "epoch": 2.2868695467120665, "grad_norm": 12.890625, "learning_rate": 9.254121660034111e-06, "loss": 0.8202, "step": 21492 }, { "epoch": 2.286975952330283, "grad_norm": 12.6328125, "learning_rate": 9.254086128482093e-06, "loss": 1.0489, "step": 21493 }, { "epoch": 2.2870823579484996, "grad_norm": 10.8828125, "learning_rate": 9.254050596930074e-06, "loss": 0.9079, "step": 21494 }, { "epoch": 2.287188763566716, "grad_norm": 15.7578125, "learning_rate": 9.254015065378058e-06, "loss": 0.9164, "step": 21495 }, { "epoch": 2.287295169184933, "grad_norm": 16.796875, "learning_rate": 9.25397953382604e-06, "loss": 0.7702, "step": 21496 }, { "epoch": 2.2874015748031495, "grad_norm": 13.1328125, "learning_rate": 9.253944002274021e-06, "loss": 1.0907, "step": 21497 }, { "epoch": 2.287507980421366, "grad_norm": 10.4375, "learning_rate": 9.253908470722002e-06, "loss": 1.054, "step": 21498 }, { "epoch": 2.287614386039583, "grad_norm": 14.8359375, "learning_rate": 9.253872939169984e-06, "loss": 0.8883, "step": 21499 }, { "epoch": 2.2877207916577995, "grad_norm": 18.828125, "learning_rate": 9.253837407617966e-06, "loss": 0.9379, "step": 21500 }, { "epoch": 2.287827197276016, "grad_norm": 12.765625, "learning_rate": 9.253801876065947e-06, "loss": 0.9325, "step": 21501 }, { "epoch": 2.287933602894233, "grad_norm": 12.2109375, "learning_rate": 9.253766344513929e-06, "loss": 0.9495, "step": 21502 }, { "epoch": 2.2880400085124495, "grad_norm": 12.78125, "learning_rate": 9.25373081296191e-06, "loss": 1.0248, "step": 21503 }, { "epoch": 2.288146414130666, "grad_norm": 12.875, "learning_rate": 9.253695281409892e-06, "loss": 0.9297, "step": 21504 }, { "epoch": 2.288252819748883, "grad_norm": 16.03125, "learning_rate": 9.253659749857875e-06, "loss": 1.0328, "step": 21505 }, { "epoch": 2.2883592253670995, "grad_norm": 13.765625, "learning_rate": 9.253624218305857e-06, "loss": 0.8608, "step": 21506 }, { "epoch": 2.288465630985316, "grad_norm": 17.265625, "learning_rate": 9.253588686753839e-06, "loss": 1.1596, "step": 21507 }, { "epoch": 2.2885720366035325, "grad_norm": 11.671875, "learning_rate": 9.25355315520182e-06, "loss": 0.9285, "step": 21508 }, { "epoch": 2.2886784422217494, "grad_norm": 10.5703125, "learning_rate": 9.253517623649802e-06, "loss": 0.7759, "step": 21509 }, { "epoch": 2.288784847839966, "grad_norm": 9.6015625, "learning_rate": 9.253482092097783e-06, "loss": 0.8781, "step": 21510 }, { "epoch": 2.2888912534581825, "grad_norm": 15.7890625, "learning_rate": 9.253446560545765e-06, "loss": 0.9546, "step": 21511 }, { "epoch": 2.2889976590763994, "grad_norm": 17.921875, "learning_rate": 9.253411028993747e-06, "loss": 0.9846, "step": 21512 }, { "epoch": 2.289104064694616, "grad_norm": 11.8515625, "learning_rate": 9.253375497441728e-06, "loss": 0.9505, "step": 21513 }, { "epoch": 2.2892104703128324, "grad_norm": 11.9375, "learning_rate": 9.25333996588971e-06, "loss": 0.8955, "step": 21514 }, { "epoch": 2.289316875931049, "grad_norm": 15.53125, "learning_rate": 9.253304434337691e-06, "loss": 1.0408, "step": 21515 }, { "epoch": 2.289423281549266, "grad_norm": 19.75, "learning_rate": 9.253268902785675e-06, "loss": 0.9609, "step": 21516 }, { "epoch": 2.2895296871674824, "grad_norm": 15.3671875, "learning_rate": 9.253233371233656e-06, "loss": 0.8085, "step": 21517 }, { "epoch": 2.289636092785699, "grad_norm": 13.59375, "learning_rate": 9.253197839681638e-06, "loss": 0.8816, "step": 21518 }, { "epoch": 2.289742498403916, "grad_norm": 16.375, "learning_rate": 9.25316230812962e-06, "loss": 0.9122, "step": 21519 }, { "epoch": 2.2898489040221324, "grad_norm": 14.390625, "learning_rate": 9.253126776577601e-06, "loss": 0.7593, "step": 21520 }, { "epoch": 2.289955309640349, "grad_norm": 14.0625, "learning_rate": 9.253091245025585e-06, "loss": 0.8445, "step": 21521 }, { "epoch": 2.290061715258566, "grad_norm": 11.8984375, "learning_rate": 9.253055713473566e-06, "loss": 0.8128, "step": 21522 }, { "epoch": 2.2901681208767823, "grad_norm": 13.8671875, "learning_rate": 9.253020181921548e-06, "loss": 0.8626, "step": 21523 }, { "epoch": 2.290274526494999, "grad_norm": 14.3046875, "learning_rate": 9.25298465036953e-06, "loss": 1.1366, "step": 21524 }, { "epoch": 2.290380932113216, "grad_norm": 27.9375, "learning_rate": 9.252949118817511e-06, "loss": 0.8869, "step": 21525 }, { "epoch": 2.2904873377314323, "grad_norm": 18.578125, "learning_rate": 9.252913587265493e-06, "loss": 0.8594, "step": 21526 }, { "epoch": 2.290593743349649, "grad_norm": 11.359375, "learning_rate": 9.252878055713474e-06, "loss": 0.9003, "step": 21527 }, { "epoch": 2.2907001489678653, "grad_norm": 13.0625, "learning_rate": 9.252842524161456e-06, "loss": 0.8381, "step": 21528 }, { "epoch": 2.2908065545860823, "grad_norm": 17.546875, "learning_rate": 9.252806992609437e-06, "loss": 0.9767, "step": 21529 }, { "epoch": 2.290912960204299, "grad_norm": 10.0546875, "learning_rate": 9.252771461057419e-06, "loss": 0.7335, "step": 21530 }, { "epoch": 2.2910193658225153, "grad_norm": 11.953125, "learning_rate": 9.252735929505402e-06, "loss": 1.0027, "step": 21531 }, { "epoch": 2.2911257714407323, "grad_norm": 11.5390625, "learning_rate": 9.252700397953384e-06, "loss": 0.9511, "step": 21532 }, { "epoch": 2.2912321770589488, "grad_norm": 16.765625, "learning_rate": 9.252664866401365e-06, "loss": 0.9598, "step": 21533 }, { "epoch": 2.2913385826771653, "grad_norm": 11.765625, "learning_rate": 9.252629334849347e-06, "loss": 0.834, "step": 21534 }, { "epoch": 2.291444988295382, "grad_norm": 12.0390625, "learning_rate": 9.252593803297329e-06, "loss": 0.9305, "step": 21535 }, { "epoch": 2.2915513939135987, "grad_norm": 14.1796875, "learning_rate": 9.25255827174531e-06, "loss": 0.7686, "step": 21536 }, { "epoch": 2.2916577995318153, "grad_norm": 16.40625, "learning_rate": 9.252522740193292e-06, "loss": 0.9645, "step": 21537 }, { "epoch": 2.2917642051500318, "grad_norm": 12.9609375, "learning_rate": 9.252487208641274e-06, "loss": 0.959, "step": 21538 }, { "epoch": 2.2918706107682487, "grad_norm": 13.4921875, "learning_rate": 9.252451677089255e-06, "loss": 0.9152, "step": 21539 }, { "epoch": 2.2919770163864652, "grad_norm": 11.015625, "learning_rate": 9.252416145537237e-06, "loss": 0.7365, "step": 21540 }, { "epoch": 2.2920834220046817, "grad_norm": 15.0546875, "learning_rate": 9.25238061398522e-06, "loss": 0.8621, "step": 21541 }, { "epoch": 2.2921898276228987, "grad_norm": 11.7265625, "learning_rate": 9.252345082433202e-06, "loss": 0.7768, "step": 21542 }, { "epoch": 2.292296233241115, "grad_norm": 16.875, "learning_rate": 9.252309550881183e-06, "loss": 0.9582, "step": 21543 }, { "epoch": 2.2924026388593317, "grad_norm": 11.921875, "learning_rate": 9.252274019329165e-06, "loss": 1.0069, "step": 21544 }, { "epoch": 2.2925090444775487, "grad_norm": 11.515625, "learning_rate": 9.252238487777146e-06, "loss": 0.8697, "step": 21545 }, { "epoch": 2.292615450095765, "grad_norm": 13.9140625, "learning_rate": 9.25220295622513e-06, "loss": 1.0358, "step": 21546 }, { "epoch": 2.2927218557139817, "grad_norm": 15.9921875, "learning_rate": 9.252167424673111e-06, "loss": 0.8778, "step": 21547 }, { "epoch": 2.292828261332198, "grad_norm": 9.46875, "learning_rate": 9.252131893121093e-06, "loss": 0.7492, "step": 21548 }, { "epoch": 2.292934666950415, "grad_norm": 10.59375, "learning_rate": 9.252096361569075e-06, "loss": 0.9269, "step": 21549 }, { "epoch": 2.2930410725686317, "grad_norm": 15.921875, "learning_rate": 9.252060830017056e-06, "loss": 0.7797, "step": 21550 }, { "epoch": 2.293147478186848, "grad_norm": 12.3359375, "learning_rate": 9.252025298465038e-06, "loss": 0.8375, "step": 21551 }, { "epoch": 2.2932538838050647, "grad_norm": 15.4453125, "learning_rate": 9.25198976691302e-06, "loss": 0.9613, "step": 21552 }, { "epoch": 2.2933602894232816, "grad_norm": 16.53125, "learning_rate": 9.251954235361001e-06, "loss": 1.0693, "step": 21553 }, { "epoch": 2.293466695041498, "grad_norm": 13.4921875, "learning_rate": 9.251918703808983e-06, "loss": 0.9919, "step": 21554 }, { "epoch": 2.2935731006597146, "grad_norm": 18.296875, "learning_rate": 9.251883172256964e-06, "loss": 0.9991, "step": 21555 }, { "epoch": 2.2936795062779316, "grad_norm": 13.71875, "learning_rate": 9.251847640704946e-06, "loss": 0.8354, "step": 21556 }, { "epoch": 2.293785911896148, "grad_norm": 14.234375, "learning_rate": 9.251812109152929e-06, "loss": 0.9501, "step": 21557 }, { "epoch": 2.2938923175143646, "grad_norm": 12.6328125, "learning_rate": 9.25177657760091e-06, "loss": 0.8454, "step": 21558 }, { "epoch": 2.2939987231325816, "grad_norm": 19.75, "learning_rate": 9.251741046048892e-06, "loss": 1.011, "step": 21559 }, { "epoch": 2.294105128750798, "grad_norm": 13.8125, "learning_rate": 9.251705514496874e-06, "loss": 0.8891, "step": 21560 }, { "epoch": 2.2942115343690146, "grad_norm": 11.0234375, "learning_rate": 9.251669982944856e-06, "loss": 0.9135, "step": 21561 }, { "epoch": 2.2943179399872315, "grad_norm": 12.8828125, "learning_rate": 9.251634451392839e-06, "loss": 0.8837, "step": 21562 }, { "epoch": 2.294424345605448, "grad_norm": 18.171875, "learning_rate": 9.251598919840819e-06, "loss": 1.0834, "step": 21563 }, { "epoch": 2.2945307512236646, "grad_norm": 14.625, "learning_rate": 9.2515633882888e-06, "loss": 0.8341, "step": 21564 }, { "epoch": 2.294637156841881, "grad_norm": 13.3359375, "learning_rate": 9.251527856736782e-06, "loss": 0.9537, "step": 21565 }, { "epoch": 2.294743562460098, "grad_norm": 15.5234375, "learning_rate": 9.251492325184764e-06, "loss": 0.8802, "step": 21566 }, { "epoch": 2.2948499680783145, "grad_norm": 15.109375, "learning_rate": 9.251456793632747e-06, "loss": 0.9772, "step": 21567 }, { "epoch": 2.294956373696531, "grad_norm": 11.1796875, "learning_rate": 9.251421262080728e-06, "loss": 0.8309, "step": 21568 }, { "epoch": 2.295062779314748, "grad_norm": 13.3125, "learning_rate": 9.25138573052871e-06, "loss": 0.9388, "step": 21569 }, { "epoch": 2.2951691849329645, "grad_norm": 11.71875, "learning_rate": 9.251350198976692e-06, "loss": 0.8828, "step": 21570 }, { "epoch": 2.295275590551181, "grad_norm": 12.609375, "learning_rate": 9.251314667424673e-06, "loss": 0.7723, "step": 21571 }, { "epoch": 2.2953819961693975, "grad_norm": 11.59375, "learning_rate": 9.251279135872657e-06, "loss": 0.9439, "step": 21572 }, { "epoch": 2.2954884017876145, "grad_norm": 10.703125, "learning_rate": 9.251243604320638e-06, "loss": 0.8532, "step": 21573 }, { "epoch": 2.295594807405831, "grad_norm": 20.890625, "learning_rate": 9.25120807276862e-06, "loss": 1.1981, "step": 21574 }, { "epoch": 2.2957012130240475, "grad_norm": 13.53125, "learning_rate": 9.251172541216601e-06, "loss": 0.9997, "step": 21575 }, { "epoch": 2.2958076186422645, "grad_norm": 11.3828125, "learning_rate": 9.251137009664583e-06, "loss": 0.9761, "step": 21576 }, { "epoch": 2.295914024260481, "grad_norm": 13.0390625, "learning_rate": 9.251101478112565e-06, "loss": 0.9068, "step": 21577 }, { "epoch": 2.2960204298786975, "grad_norm": 11.40625, "learning_rate": 9.251065946560546e-06, "loss": 0.8541, "step": 21578 }, { "epoch": 2.2961268354969144, "grad_norm": 11.625, "learning_rate": 9.251030415008528e-06, "loss": 0.7798, "step": 21579 }, { "epoch": 2.296233241115131, "grad_norm": 19.375, "learning_rate": 9.25099488345651e-06, "loss": 0.8711, "step": 21580 }, { "epoch": 2.2963396467333475, "grad_norm": 8.8515625, "learning_rate": 9.250959351904491e-06, "loss": 0.8399, "step": 21581 }, { "epoch": 2.2964460523515644, "grad_norm": 11.703125, "learning_rate": 9.250923820352474e-06, "loss": 0.7957, "step": 21582 }, { "epoch": 2.296552457969781, "grad_norm": 14.515625, "learning_rate": 9.250888288800456e-06, "loss": 0.8898, "step": 21583 }, { "epoch": 2.2966588635879974, "grad_norm": 10.0234375, "learning_rate": 9.250852757248438e-06, "loss": 0.8438, "step": 21584 }, { "epoch": 2.296765269206214, "grad_norm": 13.078125, "learning_rate": 9.25081722569642e-06, "loss": 0.907, "step": 21585 }, { "epoch": 2.296871674824431, "grad_norm": 9.9296875, "learning_rate": 9.2507816941444e-06, "loss": 0.8574, "step": 21586 }, { "epoch": 2.2969780804426474, "grad_norm": 12.3515625, "learning_rate": 9.250746162592384e-06, "loss": 0.9704, "step": 21587 }, { "epoch": 2.297084486060864, "grad_norm": 16.453125, "learning_rate": 9.250710631040366e-06, "loss": 0.9927, "step": 21588 }, { "epoch": 2.2971908916790804, "grad_norm": 13.9765625, "learning_rate": 9.250675099488347e-06, "loss": 0.832, "step": 21589 }, { "epoch": 2.2972972972972974, "grad_norm": 13.6640625, "learning_rate": 9.250639567936329e-06, "loss": 1.0259, "step": 21590 }, { "epoch": 2.297403702915514, "grad_norm": 15.09375, "learning_rate": 9.25060403638431e-06, "loss": 1.0155, "step": 21591 }, { "epoch": 2.2975101085337304, "grad_norm": 15.65625, "learning_rate": 9.250568504832292e-06, "loss": 0.973, "step": 21592 }, { "epoch": 2.2976165141519473, "grad_norm": 15.9765625, "learning_rate": 9.250532973280274e-06, "loss": 1.1974, "step": 21593 }, { "epoch": 2.297722919770164, "grad_norm": 14.2265625, "learning_rate": 9.250497441728255e-06, "loss": 0.8735, "step": 21594 }, { "epoch": 2.2978293253883804, "grad_norm": 88.875, "learning_rate": 9.250461910176237e-06, "loss": 0.6768, "step": 21595 }, { "epoch": 2.2979357310065973, "grad_norm": 14.734375, "learning_rate": 9.250426378624219e-06, "loss": 0.9918, "step": 21596 }, { "epoch": 2.298042136624814, "grad_norm": 13.171875, "learning_rate": 9.2503908470722e-06, "loss": 1.0479, "step": 21597 }, { "epoch": 2.2981485422430303, "grad_norm": 13.2578125, "learning_rate": 9.250355315520183e-06, "loss": 0.8522, "step": 21598 }, { "epoch": 2.2982549478612473, "grad_norm": 15.0078125, "learning_rate": 9.250319783968165e-06, "loss": 0.8384, "step": 21599 }, { "epoch": 2.298361353479464, "grad_norm": 12.984375, "learning_rate": 9.250284252416147e-06, "loss": 0.9353, "step": 21600 }, { "epoch": 2.2984677590976803, "grad_norm": 11.265625, "learning_rate": 9.250248720864128e-06, "loss": 0.7912, "step": 21601 }, { "epoch": 2.298574164715897, "grad_norm": 16.5, "learning_rate": 9.25021318931211e-06, "loss": 0.9383, "step": 21602 }, { "epoch": 2.2986805703341138, "grad_norm": 9.7734375, "learning_rate": 9.250177657760091e-06, "loss": 0.7785, "step": 21603 }, { "epoch": 2.2987869759523303, "grad_norm": 11.6953125, "learning_rate": 9.250142126208073e-06, "loss": 0.8152, "step": 21604 }, { "epoch": 2.298893381570547, "grad_norm": 10.0390625, "learning_rate": 9.250106594656055e-06, "loss": 0.7178, "step": 21605 }, { "epoch": 2.2989997871887637, "grad_norm": 10.7734375, "learning_rate": 9.250071063104036e-06, "loss": 0.7415, "step": 21606 }, { "epoch": 2.2991061928069803, "grad_norm": 14.7890625, "learning_rate": 9.250035531552018e-06, "loss": 0.9452, "step": 21607 }, { "epoch": 2.2992125984251968, "grad_norm": 13.453125, "learning_rate": 9.250000000000001e-06, "loss": 0.9049, "step": 21608 }, { "epoch": 2.2993190040434133, "grad_norm": 14.2421875, "learning_rate": 9.249964468447983e-06, "loss": 0.8078, "step": 21609 }, { "epoch": 2.2994254096616302, "grad_norm": 9.5, "learning_rate": 9.249928936895964e-06, "loss": 0.8592, "step": 21610 }, { "epoch": 2.2995318152798467, "grad_norm": 12.5, "learning_rate": 9.249893405343946e-06, "loss": 0.9548, "step": 21611 }, { "epoch": 2.2996382208980632, "grad_norm": 17.40625, "learning_rate": 9.249857873791928e-06, "loss": 0.9729, "step": 21612 }, { "epoch": 2.29974462651628, "grad_norm": 13.3125, "learning_rate": 9.249822342239911e-06, "loss": 0.8175, "step": 21613 }, { "epoch": 2.2998510321344967, "grad_norm": 13.21875, "learning_rate": 9.249786810687893e-06, "loss": 1.1227, "step": 21614 }, { "epoch": 2.299957437752713, "grad_norm": 9.359375, "learning_rate": 9.249751279135874e-06, "loss": 0.8503, "step": 21615 }, { "epoch": 2.30006384337093, "grad_norm": 16.140625, "learning_rate": 9.249715747583856e-06, "loss": 1.1388, "step": 21616 }, { "epoch": 2.3001702489891467, "grad_norm": 12.5625, "learning_rate": 9.249680216031837e-06, "loss": 0.839, "step": 21617 }, { "epoch": 2.300276654607363, "grad_norm": 12.7890625, "learning_rate": 9.249644684479819e-06, "loss": 0.8427, "step": 21618 }, { "epoch": 2.30038306022558, "grad_norm": 10.8359375, "learning_rate": 9.2496091529278e-06, "loss": 0.9552, "step": 21619 }, { "epoch": 2.3004894658437967, "grad_norm": 12.953125, "learning_rate": 9.249573621375782e-06, "loss": 0.9006, "step": 21620 }, { "epoch": 2.300595871462013, "grad_norm": 13.03125, "learning_rate": 9.249538089823764e-06, "loss": 0.7968, "step": 21621 }, { "epoch": 2.3007022770802297, "grad_norm": 16.9375, "learning_rate": 9.249502558271745e-06, "loss": 0.8072, "step": 21622 }, { "epoch": 2.3008086826984466, "grad_norm": 14.53125, "learning_rate": 9.249467026719729e-06, "loss": 0.8735, "step": 21623 }, { "epoch": 2.300915088316663, "grad_norm": 11.9453125, "learning_rate": 9.24943149516771e-06, "loss": 0.9708, "step": 21624 }, { "epoch": 2.3010214939348796, "grad_norm": 11.734375, "learning_rate": 9.249395963615692e-06, "loss": 0.9022, "step": 21625 }, { "epoch": 2.301127899553096, "grad_norm": 10.34375, "learning_rate": 9.249360432063674e-06, "loss": 0.8664, "step": 21626 }, { "epoch": 2.301234305171313, "grad_norm": 10.2109375, "learning_rate": 9.249324900511655e-06, "loss": 0.841, "step": 21627 }, { "epoch": 2.3013407107895296, "grad_norm": 10.8359375, "learning_rate": 9.249289368959637e-06, "loss": 0.78, "step": 21628 }, { "epoch": 2.301447116407746, "grad_norm": 12.1015625, "learning_rate": 9.249253837407618e-06, "loss": 0.917, "step": 21629 }, { "epoch": 2.301553522025963, "grad_norm": 12.5546875, "learning_rate": 9.2492183058556e-06, "loss": 0.9034, "step": 21630 }, { "epoch": 2.3016599276441796, "grad_norm": 15.5234375, "learning_rate": 9.249182774303582e-06, "loss": 0.8484, "step": 21631 }, { "epoch": 2.301766333262396, "grad_norm": 10.4921875, "learning_rate": 9.249147242751563e-06, "loss": 0.9465, "step": 21632 }, { "epoch": 2.301872738880613, "grad_norm": 19.515625, "learning_rate": 9.249111711199545e-06, "loss": 0.8909, "step": 21633 }, { "epoch": 2.3019791444988296, "grad_norm": 13.296875, "learning_rate": 9.249076179647528e-06, "loss": 0.9032, "step": 21634 }, { "epoch": 2.302085550117046, "grad_norm": 10.6171875, "learning_rate": 9.24904064809551e-06, "loss": 0.8692, "step": 21635 }, { "epoch": 2.302191955735263, "grad_norm": 10.109375, "learning_rate": 9.249005116543491e-06, "loss": 0.7063, "step": 21636 }, { "epoch": 2.3022983613534795, "grad_norm": 9.40625, "learning_rate": 9.248969584991473e-06, "loss": 0.765, "step": 21637 }, { "epoch": 2.302404766971696, "grad_norm": 13.671875, "learning_rate": 9.248934053439455e-06, "loss": 0.9951, "step": 21638 }, { "epoch": 2.3025111725899126, "grad_norm": 14.5546875, "learning_rate": 9.248898521887438e-06, "loss": 0.9499, "step": 21639 }, { "epoch": 2.3026175782081295, "grad_norm": 12.609375, "learning_rate": 9.24886299033542e-06, "loss": 0.6603, "step": 21640 }, { "epoch": 2.302723983826346, "grad_norm": 12.3984375, "learning_rate": 9.248827458783401e-06, "loss": 0.8908, "step": 21641 }, { "epoch": 2.3028303894445625, "grad_norm": 11.703125, "learning_rate": 9.248791927231383e-06, "loss": 0.9857, "step": 21642 }, { "epoch": 2.3029367950627795, "grad_norm": 13.3828125, "learning_rate": 9.248756395679364e-06, "loss": 0.8597, "step": 21643 }, { "epoch": 2.303043200680996, "grad_norm": 13.90625, "learning_rate": 9.248720864127346e-06, "loss": 0.9199, "step": 21644 }, { "epoch": 2.3031496062992125, "grad_norm": 17.15625, "learning_rate": 9.248685332575327e-06, "loss": 0.9318, "step": 21645 }, { "epoch": 2.303256011917429, "grad_norm": 12.234375, "learning_rate": 9.248649801023309e-06, "loss": 0.845, "step": 21646 }, { "epoch": 2.303362417535646, "grad_norm": 14.40625, "learning_rate": 9.24861426947129e-06, "loss": 0.9486, "step": 21647 }, { "epoch": 2.3034688231538625, "grad_norm": 18.75, "learning_rate": 9.248578737919272e-06, "loss": 0.8444, "step": 21648 }, { "epoch": 2.303575228772079, "grad_norm": 10.578125, "learning_rate": 9.248543206367256e-06, "loss": 0.8591, "step": 21649 }, { "epoch": 2.303681634390296, "grad_norm": 14.0390625, "learning_rate": 9.248507674815237e-06, "loss": 1.0292, "step": 21650 }, { "epoch": 2.3037880400085124, "grad_norm": 13.671875, "learning_rate": 9.248472143263219e-06, "loss": 0.8701, "step": 21651 }, { "epoch": 2.303894445626729, "grad_norm": 10.015625, "learning_rate": 9.2484366117112e-06, "loss": 0.7994, "step": 21652 }, { "epoch": 2.304000851244946, "grad_norm": 16.875, "learning_rate": 9.248401080159182e-06, "loss": 1.0101, "step": 21653 }, { "epoch": 2.3041072568631624, "grad_norm": 12.515625, "learning_rate": 9.248365548607164e-06, "loss": 0.8908, "step": 21654 }, { "epoch": 2.304213662481379, "grad_norm": 12.875, "learning_rate": 9.248330017055145e-06, "loss": 0.914, "step": 21655 }, { "epoch": 2.304320068099596, "grad_norm": 13.40625, "learning_rate": 9.248294485503127e-06, "loss": 1.014, "step": 21656 }, { "epoch": 2.3044264737178124, "grad_norm": 13.4140625, "learning_rate": 9.248258953951108e-06, "loss": 0.9572, "step": 21657 }, { "epoch": 2.304532879336029, "grad_norm": 14.4375, "learning_rate": 9.24822342239909e-06, "loss": 0.873, "step": 21658 }, { "epoch": 2.3046392849542454, "grad_norm": 10.0703125, "learning_rate": 9.248187890847073e-06, "loss": 0.816, "step": 21659 }, { "epoch": 2.3047456905724624, "grad_norm": 12.75, "learning_rate": 9.248152359295055e-06, "loss": 0.8997, "step": 21660 }, { "epoch": 2.304852096190679, "grad_norm": 15.1171875, "learning_rate": 9.248116827743037e-06, "loss": 0.9638, "step": 21661 }, { "epoch": 2.3049585018088954, "grad_norm": 10.96875, "learning_rate": 9.248081296191018e-06, "loss": 0.832, "step": 21662 }, { "epoch": 2.3050649074271123, "grad_norm": 12.9609375, "learning_rate": 9.248045764639e-06, "loss": 1.0509, "step": 21663 }, { "epoch": 2.305171313045329, "grad_norm": 18.6875, "learning_rate": 9.248010233086983e-06, "loss": 1.1722, "step": 21664 }, { "epoch": 2.3052777186635454, "grad_norm": 16.203125, "learning_rate": 9.247974701534965e-06, "loss": 0.8294, "step": 21665 }, { "epoch": 2.305384124281762, "grad_norm": 11.4609375, "learning_rate": 9.247939169982946e-06, "loss": 0.9685, "step": 21666 }, { "epoch": 2.305490529899979, "grad_norm": 12.109375, "learning_rate": 9.247903638430928e-06, "loss": 0.7506, "step": 21667 }, { "epoch": 2.3055969355181953, "grad_norm": 8.515625, "learning_rate": 9.24786810687891e-06, "loss": 0.7819, "step": 21668 }, { "epoch": 2.305703341136412, "grad_norm": 20.734375, "learning_rate": 9.247832575326891e-06, "loss": 0.7823, "step": 21669 }, { "epoch": 2.305809746754629, "grad_norm": 13.9375, "learning_rate": 9.247797043774873e-06, "loss": 0.8204, "step": 21670 }, { "epoch": 2.3059161523728453, "grad_norm": 11.453125, "learning_rate": 9.247761512222854e-06, "loss": 0.8722, "step": 21671 }, { "epoch": 2.306022557991062, "grad_norm": 15.0234375, "learning_rate": 9.247725980670836e-06, "loss": 1.046, "step": 21672 }, { "epoch": 2.3061289636092788, "grad_norm": 12.2421875, "learning_rate": 9.247690449118818e-06, "loss": 0.8397, "step": 21673 }, { "epoch": 2.3062353692274953, "grad_norm": 18.390625, "learning_rate": 9.247654917566799e-06, "loss": 0.8329, "step": 21674 }, { "epoch": 2.306341774845712, "grad_norm": 12.0234375, "learning_rate": 9.247619386014782e-06, "loss": 0.8905, "step": 21675 }, { "epoch": 2.3064481804639287, "grad_norm": 11.53125, "learning_rate": 9.247583854462764e-06, "loss": 0.887, "step": 21676 }, { "epoch": 2.3065545860821453, "grad_norm": 10.6953125, "learning_rate": 9.247548322910746e-06, "loss": 0.6692, "step": 21677 }, { "epoch": 2.3066609917003618, "grad_norm": 11.53125, "learning_rate": 9.247512791358727e-06, "loss": 0.92, "step": 21678 }, { "epoch": 2.3067673973185783, "grad_norm": 13.0703125, "learning_rate": 9.247477259806709e-06, "loss": 0.8498, "step": 21679 }, { "epoch": 2.3068738029367952, "grad_norm": 15.015625, "learning_rate": 9.247441728254692e-06, "loss": 0.7744, "step": 21680 }, { "epoch": 2.3069802085550117, "grad_norm": 10.109375, "learning_rate": 9.247406196702674e-06, "loss": 0.9246, "step": 21681 }, { "epoch": 2.3070866141732282, "grad_norm": 10.4296875, "learning_rate": 9.247370665150654e-06, "loss": 0.8325, "step": 21682 }, { "epoch": 2.3071930197914448, "grad_norm": 14.7734375, "learning_rate": 9.247335133598635e-06, "loss": 0.6538, "step": 21683 }, { "epoch": 2.3072994254096617, "grad_norm": 10.9609375, "learning_rate": 9.247299602046617e-06, "loss": 0.8441, "step": 21684 }, { "epoch": 2.307405831027878, "grad_norm": 9.7578125, "learning_rate": 9.2472640704946e-06, "loss": 0.9171, "step": 21685 }, { "epoch": 2.3075122366460947, "grad_norm": 11.1171875, "learning_rate": 9.247228538942582e-06, "loss": 0.7408, "step": 21686 }, { "epoch": 2.3076186422643117, "grad_norm": 13.3046875, "learning_rate": 9.247193007390563e-06, "loss": 0.7681, "step": 21687 }, { "epoch": 2.307725047882528, "grad_norm": 13.46875, "learning_rate": 9.247157475838545e-06, "loss": 0.9198, "step": 21688 }, { "epoch": 2.3078314535007447, "grad_norm": 10.1328125, "learning_rate": 9.247121944286527e-06, "loss": 0.9053, "step": 21689 }, { "epoch": 2.3079378591189617, "grad_norm": 12.7421875, "learning_rate": 9.24708641273451e-06, "loss": 0.9812, "step": 21690 }, { "epoch": 2.308044264737178, "grad_norm": 13.5390625, "learning_rate": 9.247050881182492e-06, "loss": 0.9194, "step": 21691 }, { "epoch": 2.3081506703553947, "grad_norm": 13.2421875, "learning_rate": 9.247015349630473e-06, "loss": 0.6912, "step": 21692 }, { "epoch": 2.3082570759736116, "grad_norm": 12.1953125, "learning_rate": 9.246979818078455e-06, "loss": 0.9349, "step": 21693 }, { "epoch": 2.308363481591828, "grad_norm": 12.8203125, "learning_rate": 9.246944286526436e-06, "loss": 0.773, "step": 21694 }, { "epoch": 2.3084698872100446, "grad_norm": 18.078125, "learning_rate": 9.246908754974418e-06, "loss": 0.8399, "step": 21695 }, { "epoch": 2.308576292828261, "grad_norm": 14.3515625, "learning_rate": 9.2468732234224e-06, "loss": 0.8826, "step": 21696 }, { "epoch": 2.308682698446478, "grad_norm": 10.7890625, "learning_rate": 9.246837691870381e-06, "loss": 0.7601, "step": 21697 }, { "epoch": 2.3087891040646946, "grad_norm": 15.6875, "learning_rate": 9.246802160318363e-06, "loss": 0.9826, "step": 21698 }, { "epoch": 2.308895509682911, "grad_norm": 13.953125, "learning_rate": 9.246766628766344e-06, "loss": 0.8981, "step": 21699 }, { "epoch": 2.309001915301128, "grad_norm": 10.9765625, "learning_rate": 9.246731097214328e-06, "loss": 0.9237, "step": 21700 }, { "epoch": 2.3091083209193446, "grad_norm": 11.7734375, "learning_rate": 9.24669556566231e-06, "loss": 0.9069, "step": 21701 }, { "epoch": 2.309214726537561, "grad_norm": 13.3515625, "learning_rate": 9.246660034110291e-06, "loss": 0.9325, "step": 21702 }, { "epoch": 2.3093211321557776, "grad_norm": 11.4765625, "learning_rate": 9.246624502558272e-06, "loss": 0.9443, "step": 21703 }, { "epoch": 2.3094275377739946, "grad_norm": 10.84375, "learning_rate": 9.246588971006254e-06, "loss": 0.9426, "step": 21704 }, { "epoch": 2.309533943392211, "grad_norm": 11.203125, "learning_rate": 9.246553439454237e-06, "loss": 0.7329, "step": 21705 }, { "epoch": 2.3096403490104276, "grad_norm": 11.390625, "learning_rate": 9.246517907902219e-06, "loss": 0.7985, "step": 21706 }, { "epoch": 2.3097467546286445, "grad_norm": 14.3984375, "learning_rate": 9.2464823763502e-06, "loss": 1.0748, "step": 21707 }, { "epoch": 2.309853160246861, "grad_norm": 14.546875, "learning_rate": 9.246446844798182e-06, "loss": 0.909, "step": 21708 }, { "epoch": 2.3099595658650776, "grad_norm": 12.03125, "learning_rate": 9.246411313246164e-06, "loss": 0.7274, "step": 21709 }, { "epoch": 2.3100659714832945, "grad_norm": 12.3203125, "learning_rate": 9.246375781694145e-06, "loss": 0.9817, "step": 21710 }, { "epoch": 2.310172377101511, "grad_norm": 13.2421875, "learning_rate": 9.246340250142127e-06, "loss": 0.7751, "step": 21711 }, { "epoch": 2.3102787827197275, "grad_norm": 11.140625, "learning_rate": 9.246304718590109e-06, "loss": 0.9947, "step": 21712 }, { "epoch": 2.3103851883379445, "grad_norm": 9.4765625, "learning_rate": 9.24626918703809e-06, "loss": 0.7059, "step": 21713 }, { "epoch": 2.310491593956161, "grad_norm": 15.1875, "learning_rate": 9.246233655486072e-06, "loss": 0.8921, "step": 21714 }, { "epoch": 2.3105979995743775, "grad_norm": 12.203125, "learning_rate": 9.246198123934053e-06, "loss": 0.8503, "step": 21715 }, { "epoch": 2.310704405192594, "grad_norm": 11.921875, "learning_rate": 9.246162592382037e-06, "loss": 0.7613, "step": 21716 }, { "epoch": 2.310810810810811, "grad_norm": 10.7421875, "learning_rate": 9.246127060830018e-06, "loss": 0.8712, "step": 21717 }, { "epoch": 2.3109172164290275, "grad_norm": 8.796875, "learning_rate": 9.246091529278e-06, "loss": 0.7867, "step": 21718 }, { "epoch": 2.311023622047244, "grad_norm": 13.6328125, "learning_rate": 9.246055997725982e-06, "loss": 0.9029, "step": 21719 }, { "epoch": 2.3111300276654605, "grad_norm": 21.25, "learning_rate": 9.246020466173963e-06, "loss": 1.114, "step": 21720 }, { "epoch": 2.3112364332836774, "grad_norm": 14.953125, "learning_rate": 9.245984934621945e-06, "loss": 0.9419, "step": 21721 }, { "epoch": 2.311342838901894, "grad_norm": 12.5625, "learning_rate": 9.245949403069926e-06, "loss": 0.8704, "step": 21722 }, { "epoch": 2.3114492445201105, "grad_norm": 13.140625, "learning_rate": 9.245913871517908e-06, "loss": 0.716, "step": 21723 }, { "epoch": 2.3115556501383274, "grad_norm": 12.75, "learning_rate": 9.24587833996589e-06, "loss": 0.8068, "step": 21724 }, { "epoch": 2.311662055756544, "grad_norm": 13.640625, "learning_rate": 9.245842808413871e-06, "loss": 0.8888, "step": 21725 }, { "epoch": 2.3117684613747604, "grad_norm": 19.734375, "learning_rate": 9.245807276861855e-06, "loss": 0.5897, "step": 21726 }, { "epoch": 2.3118748669929774, "grad_norm": 12.546875, "learning_rate": 9.245771745309836e-06, "loss": 0.8485, "step": 21727 }, { "epoch": 2.311981272611194, "grad_norm": 15.265625, "learning_rate": 9.245736213757818e-06, "loss": 0.9015, "step": 21728 }, { "epoch": 2.3120876782294104, "grad_norm": 19.671875, "learning_rate": 9.2457006822058e-06, "loss": 0.8944, "step": 21729 }, { "epoch": 2.3121940838476274, "grad_norm": 14.296875, "learning_rate": 9.245665150653781e-06, "loss": 0.7734, "step": 21730 }, { "epoch": 2.312300489465844, "grad_norm": 16.90625, "learning_rate": 9.245629619101764e-06, "loss": 0.9271, "step": 21731 }, { "epoch": 2.3124068950840604, "grad_norm": 12.3046875, "learning_rate": 9.245594087549746e-06, "loss": 0.784, "step": 21732 }, { "epoch": 2.312513300702277, "grad_norm": 13.4765625, "learning_rate": 9.245558555997727e-06, "loss": 0.8703, "step": 21733 }, { "epoch": 2.312619706320494, "grad_norm": 12.4375, "learning_rate": 9.245523024445709e-06, "loss": 0.912, "step": 21734 }, { "epoch": 2.3127261119387104, "grad_norm": 10.0703125, "learning_rate": 9.24548749289369e-06, "loss": 0.7729, "step": 21735 }, { "epoch": 2.312832517556927, "grad_norm": 9.5859375, "learning_rate": 9.245451961341672e-06, "loss": 0.8274, "step": 21736 }, { "epoch": 2.312938923175144, "grad_norm": 12.3125, "learning_rate": 9.245416429789654e-06, "loss": 0.7908, "step": 21737 }, { "epoch": 2.3130453287933603, "grad_norm": 15.515625, "learning_rate": 9.245380898237635e-06, "loss": 0.9475, "step": 21738 }, { "epoch": 2.313151734411577, "grad_norm": 14.578125, "learning_rate": 9.245345366685617e-06, "loss": 0.9635, "step": 21739 }, { "epoch": 2.3132581400297934, "grad_norm": 22.546875, "learning_rate": 9.245309835133599e-06, "loss": 0.7973, "step": 21740 }, { "epoch": 2.3133645456480103, "grad_norm": 14.5546875, "learning_rate": 9.245274303581582e-06, "loss": 0.8318, "step": 21741 }, { "epoch": 2.313470951266227, "grad_norm": 16.375, "learning_rate": 9.245238772029564e-06, "loss": 1.0261, "step": 21742 }, { "epoch": 2.3135773568844433, "grad_norm": 16.03125, "learning_rate": 9.245203240477545e-06, "loss": 0.9213, "step": 21743 }, { "epoch": 2.3136837625026603, "grad_norm": 13.1953125, "learning_rate": 9.245167708925527e-06, "loss": 0.8033, "step": 21744 }, { "epoch": 2.313790168120877, "grad_norm": 9.9609375, "learning_rate": 9.245132177373508e-06, "loss": 0.7448, "step": 21745 }, { "epoch": 2.3138965737390933, "grad_norm": 16.640625, "learning_rate": 9.24509664582149e-06, "loss": 1.1766, "step": 21746 }, { "epoch": 2.3140029793573103, "grad_norm": 11.8984375, "learning_rate": 9.245061114269472e-06, "loss": 0.7706, "step": 21747 }, { "epoch": 2.3141093849755268, "grad_norm": 15.4921875, "learning_rate": 9.245025582717453e-06, "loss": 0.9675, "step": 21748 }, { "epoch": 2.3142157905937433, "grad_norm": 10.4140625, "learning_rate": 9.244990051165435e-06, "loss": 0.8491, "step": 21749 }, { "epoch": 2.3143221962119602, "grad_norm": 11.4375, "learning_rate": 9.244954519613416e-06, "loss": 0.9228, "step": 21750 }, { "epoch": 2.3144286018301767, "grad_norm": 15.9453125, "learning_rate": 9.244918988061398e-06, "loss": 1.0959, "step": 21751 }, { "epoch": 2.3145350074483932, "grad_norm": 14.71875, "learning_rate": 9.244883456509381e-06, "loss": 1.0229, "step": 21752 }, { "epoch": 2.3146414130666098, "grad_norm": 14.15625, "learning_rate": 9.244847924957363e-06, "loss": 1.0014, "step": 21753 }, { "epoch": 2.3147478186848267, "grad_norm": 13.359375, "learning_rate": 9.244812393405345e-06, "loss": 0.8793, "step": 21754 }, { "epoch": 2.314854224303043, "grad_norm": 13.953125, "learning_rate": 9.244776861853326e-06, "loss": 0.8349, "step": 21755 }, { "epoch": 2.3149606299212597, "grad_norm": 14.4140625, "learning_rate": 9.244741330301308e-06, "loss": 0.8968, "step": 21756 }, { "epoch": 2.3150670355394762, "grad_norm": 10.65625, "learning_rate": 9.244705798749291e-06, "loss": 0.8976, "step": 21757 }, { "epoch": 2.315173441157693, "grad_norm": 12.75, "learning_rate": 9.244670267197273e-06, "loss": 0.8325, "step": 21758 }, { "epoch": 2.3152798467759097, "grad_norm": 14.5390625, "learning_rate": 9.244634735645254e-06, "loss": 0.9423, "step": 21759 }, { "epoch": 2.315386252394126, "grad_norm": 11.5234375, "learning_rate": 9.244599204093236e-06, "loss": 0.7795, "step": 21760 }, { "epoch": 2.315492658012343, "grad_norm": 14.046875, "learning_rate": 9.244563672541218e-06, "loss": 0.8838, "step": 21761 }, { "epoch": 2.3155990636305597, "grad_norm": 13.9140625, "learning_rate": 9.244528140989199e-06, "loss": 1.1121, "step": 21762 }, { "epoch": 2.315705469248776, "grad_norm": 9.9296875, "learning_rate": 9.24449260943718e-06, "loss": 0.7995, "step": 21763 }, { "epoch": 2.315811874866993, "grad_norm": 15.7734375, "learning_rate": 9.244457077885162e-06, "loss": 0.8047, "step": 21764 }, { "epoch": 2.3159182804852096, "grad_norm": 12.109375, "learning_rate": 9.244421546333144e-06, "loss": 0.8204, "step": 21765 }, { "epoch": 2.316024686103426, "grad_norm": 16.890625, "learning_rate": 9.244386014781126e-06, "loss": 0.8014, "step": 21766 }, { "epoch": 2.316131091721643, "grad_norm": 16.234375, "learning_rate": 9.244350483229109e-06, "loss": 1.0097, "step": 21767 }, { "epoch": 2.3162374973398596, "grad_norm": 14.0859375, "learning_rate": 9.24431495167709e-06, "loss": 1.0587, "step": 21768 }, { "epoch": 2.316343902958076, "grad_norm": 20.375, "learning_rate": 9.244279420125072e-06, "loss": 1.0756, "step": 21769 }, { "epoch": 2.3164503085762926, "grad_norm": 11.609375, "learning_rate": 9.244243888573054e-06, "loss": 0.7341, "step": 21770 }, { "epoch": 2.3165567141945096, "grad_norm": 10.5546875, "learning_rate": 9.244208357021035e-06, "loss": 0.9127, "step": 21771 }, { "epoch": 2.316663119812726, "grad_norm": 18.546875, "learning_rate": 9.244172825469017e-06, "loss": 0.9634, "step": 21772 }, { "epoch": 2.3167695254309426, "grad_norm": 17.5, "learning_rate": 9.244137293916999e-06, "loss": 1.041, "step": 21773 }, { "epoch": 2.3168759310491596, "grad_norm": 12.859375, "learning_rate": 9.24410176236498e-06, "loss": 0.7247, "step": 21774 }, { "epoch": 2.316982336667376, "grad_norm": 14.3046875, "learning_rate": 9.244066230812962e-06, "loss": 0.9519, "step": 21775 }, { "epoch": 2.3170887422855926, "grad_norm": 14.59375, "learning_rate": 9.244030699260943e-06, "loss": 1.1049, "step": 21776 }, { "epoch": 2.317195147903809, "grad_norm": 11.625, "learning_rate": 9.243995167708927e-06, "loss": 0.8254, "step": 21777 }, { "epoch": 2.317301553522026, "grad_norm": 11.765625, "learning_rate": 9.243959636156908e-06, "loss": 0.7386, "step": 21778 }, { "epoch": 2.3174079591402426, "grad_norm": 12.28125, "learning_rate": 9.24392410460489e-06, "loss": 1.0771, "step": 21779 }, { "epoch": 2.317514364758459, "grad_norm": 16.390625, "learning_rate": 9.243888573052871e-06, "loss": 0.7957, "step": 21780 }, { "epoch": 2.317620770376676, "grad_norm": 12.9140625, "learning_rate": 9.243853041500853e-06, "loss": 0.9562, "step": 21781 }, { "epoch": 2.3177271759948925, "grad_norm": 14.015625, "learning_rate": 9.243817509948836e-06, "loss": 0.9167, "step": 21782 }, { "epoch": 2.317833581613109, "grad_norm": 14.6328125, "learning_rate": 9.243781978396818e-06, "loss": 1.0261, "step": 21783 }, { "epoch": 2.317939987231326, "grad_norm": 16.53125, "learning_rate": 9.2437464468448e-06, "loss": 0.8487, "step": 21784 }, { "epoch": 2.3180463928495425, "grad_norm": 15.2421875, "learning_rate": 9.243710915292781e-06, "loss": 1.066, "step": 21785 }, { "epoch": 2.318152798467759, "grad_norm": 10.921875, "learning_rate": 9.243675383740763e-06, "loss": 0.7669, "step": 21786 }, { "epoch": 2.318259204085976, "grad_norm": 13.703125, "learning_rate": 9.243639852188744e-06, "loss": 0.8565, "step": 21787 }, { "epoch": 2.3183656097041925, "grad_norm": 13.2578125, "learning_rate": 9.243604320636726e-06, "loss": 0.8728, "step": 21788 }, { "epoch": 2.318472015322409, "grad_norm": 12.078125, "learning_rate": 9.243568789084708e-06, "loss": 0.8678, "step": 21789 }, { "epoch": 2.3185784209406255, "grad_norm": 15.0078125, "learning_rate": 9.24353325753269e-06, "loss": 0.9981, "step": 21790 }, { "epoch": 2.3186848265588424, "grad_norm": 12.125, "learning_rate": 9.24349772598067e-06, "loss": 0.9413, "step": 21791 }, { "epoch": 2.318791232177059, "grad_norm": 15.640625, "learning_rate": 9.243462194428652e-06, "loss": 0.7847, "step": 21792 }, { "epoch": 2.3188976377952755, "grad_norm": 11.6171875, "learning_rate": 9.243426662876636e-06, "loss": 0.8528, "step": 21793 }, { "epoch": 2.3190040434134924, "grad_norm": 11.3515625, "learning_rate": 9.243391131324617e-06, "loss": 0.6765, "step": 21794 }, { "epoch": 2.319110449031709, "grad_norm": 16.765625, "learning_rate": 9.243355599772599e-06, "loss": 0.8176, "step": 21795 }, { "epoch": 2.3192168546499254, "grad_norm": 15.453125, "learning_rate": 9.24332006822058e-06, "loss": 1.1191, "step": 21796 }, { "epoch": 2.319323260268142, "grad_norm": 12.4765625, "learning_rate": 9.243284536668562e-06, "loss": 0.9145, "step": 21797 }, { "epoch": 2.319429665886359, "grad_norm": 9.6953125, "learning_rate": 9.243249005116545e-06, "loss": 0.756, "step": 21798 }, { "epoch": 2.3195360715045754, "grad_norm": 12.75, "learning_rate": 9.243213473564527e-06, "loss": 0.8677, "step": 21799 }, { "epoch": 2.319642477122792, "grad_norm": 13.3359375, "learning_rate": 9.243177942012509e-06, "loss": 0.8328, "step": 21800 }, { "epoch": 2.319748882741009, "grad_norm": 17.671875, "learning_rate": 9.24314241046049e-06, "loss": 0.9392, "step": 21801 }, { "epoch": 2.3198552883592254, "grad_norm": 16.65625, "learning_rate": 9.24310687890847e-06, "loss": 0.8652, "step": 21802 }, { "epoch": 2.319961693977442, "grad_norm": 17.28125, "learning_rate": 9.243071347356453e-06, "loss": 0.8193, "step": 21803 }, { "epoch": 2.320068099595659, "grad_norm": 11.59375, "learning_rate": 9.243035815804435e-06, "loss": 0.7225, "step": 21804 }, { "epoch": 2.3201745052138754, "grad_norm": 13.25, "learning_rate": 9.243000284252417e-06, "loss": 0.847, "step": 21805 }, { "epoch": 2.320280910832092, "grad_norm": 12.359375, "learning_rate": 9.242964752700398e-06, "loss": 0.8329, "step": 21806 }, { "epoch": 2.320387316450309, "grad_norm": 13.3984375, "learning_rate": 9.24292922114838e-06, "loss": 0.8689, "step": 21807 }, { "epoch": 2.3204937220685253, "grad_norm": 12.796875, "learning_rate": 9.242893689596363e-06, "loss": 1.0294, "step": 21808 }, { "epoch": 2.320600127686742, "grad_norm": 11.4453125, "learning_rate": 9.242858158044345e-06, "loss": 1.0099, "step": 21809 }, { "epoch": 2.3207065333049584, "grad_norm": 11.578125, "learning_rate": 9.242822626492326e-06, "loss": 0.8661, "step": 21810 }, { "epoch": 2.3208129389231753, "grad_norm": 17.75, "learning_rate": 9.242787094940308e-06, "loss": 0.8834, "step": 21811 }, { "epoch": 2.320919344541392, "grad_norm": 10.890625, "learning_rate": 9.24275156338829e-06, "loss": 0.9684, "step": 21812 }, { "epoch": 2.3210257501596083, "grad_norm": 15.90625, "learning_rate": 9.242716031836271e-06, "loss": 0.9497, "step": 21813 }, { "epoch": 2.321132155777825, "grad_norm": 12.4296875, "learning_rate": 9.242680500284253e-06, "loss": 0.7898, "step": 21814 }, { "epoch": 2.321238561396042, "grad_norm": 12.78125, "learning_rate": 9.242644968732234e-06, "loss": 0.7953, "step": 21815 }, { "epoch": 2.3213449670142583, "grad_norm": 11.453125, "learning_rate": 9.242609437180216e-06, "loss": 0.944, "step": 21816 }, { "epoch": 2.321451372632475, "grad_norm": 14.4453125, "learning_rate": 9.242573905628198e-06, "loss": 1.0341, "step": 21817 }, { "epoch": 2.3215577782506918, "grad_norm": 8.6015625, "learning_rate": 9.242538374076181e-06, "loss": 0.7983, "step": 21818 }, { "epoch": 2.3216641838689083, "grad_norm": 16.15625, "learning_rate": 9.242502842524163e-06, "loss": 0.9099, "step": 21819 }, { "epoch": 2.321770589487125, "grad_norm": 12.0546875, "learning_rate": 9.242467310972144e-06, "loss": 0.9086, "step": 21820 }, { "epoch": 2.3218769951053417, "grad_norm": 15.8828125, "learning_rate": 9.242431779420126e-06, "loss": 0.9366, "step": 21821 }, { "epoch": 2.3219834007235582, "grad_norm": 14.203125, "learning_rate": 9.242396247868107e-06, "loss": 1.0048, "step": 21822 }, { "epoch": 2.3220898063417748, "grad_norm": 15.96875, "learning_rate": 9.242360716316089e-06, "loss": 0.9774, "step": 21823 }, { "epoch": 2.3221962119599917, "grad_norm": 11.6640625, "learning_rate": 9.242325184764072e-06, "loss": 0.8657, "step": 21824 }, { "epoch": 2.322302617578208, "grad_norm": 13.6640625, "learning_rate": 9.242289653212054e-06, "loss": 1.0488, "step": 21825 }, { "epoch": 2.3224090231964247, "grad_norm": 10.515625, "learning_rate": 9.242254121660036e-06, "loss": 0.7757, "step": 21826 }, { "epoch": 2.3225154288146412, "grad_norm": 12.375, "learning_rate": 9.242218590108017e-06, "loss": 0.8441, "step": 21827 }, { "epoch": 2.322621834432858, "grad_norm": 11.9375, "learning_rate": 9.242183058555999e-06, "loss": 0.8859, "step": 21828 }, { "epoch": 2.3227282400510747, "grad_norm": 10.2265625, "learning_rate": 9.24214752700398e-06, "loss": 0.8575, "step": 21829 }, { "epoch": 2.322834645669291, "grad_norm": 17.421875, "learning_rate": 9.242111995451962e-06, "loss": 0.8782, "step": 21830 }, { "epoch": 2.322941051287508, "grad_norm": 11.4453125, "learning_rate": 9.242076463899944e-06, "loss": 0.8391, "step": 21831 }, { "epoch": 2.3230474569057247, "grad_norm": 11.4765625, "learning_rate": 9.242040932347925e-06, "loss": 0.9049, "step": 21832 }, { "epoch": 2.323153862523941, "grad_norm": 19.078125, "learning_rate": 9.242005400795907e-06, "loss": 0.8663, "step": 21833 }, { "epoch": 2.3232602681421577, "grad_norm": 13.484375, "learning_rate": 9.24196986924389e-06, "loss": 0.9521, "step": 21834 }, { "epoch": 2.3233666737603746, "grad_norm": 15.5078125, "learning_rate": 9.241934337691872e-06, "loss": 0.8328, "step": 21835 }, { "epoch": 2.323473079378591, "grad_norm": 10.5078125, "learning_rate": 9.241898806139853e-06, "loss": 0.8986, "step": 21836 }, { "epoch": 2.3235794849968077, "grad_norm": 10.375, "learning_rate": 9.241863274587835e-06, "loss": 0.741, "step": 21837 }, { "epoch": 2.3236858906150246, "grad_norm": 13.90625, "learning_rate": 9.241827743035816e-06, "loss": 0.9981, "step": 21838 }, { "epoch": 2.323792296233241, "grad_norm": 11.890625, "learning_rate": 9.241792211483798e-06, "loss": 0.7443, "step": 21839 }, { "epoch": 2.3238987018514576, "grad_norm": 13.171875, "learning_rate": 9.24175667993178e-06, "loss": 0.7837, "step": 21840 }, { "epoch": 2.3240051074696746, "grad_norm": 13.03125, "learning_rate": 9.241721148379761e-06, "loss": 1.0062, "step": 21841 }, { "epoch": 2.324111513087891, "grad_norm": 16.171875, "learning_rate": 9.241685616827743e-06, "loss": 0.941, "step": 21842 }, { "epoch": 2.3242179187061076, "grad_norm": 12.9921875, "learning_rate": 9.241650085275725e-06, "loss": 0.9291, "step": 21843 }, { "epoch": 2.3243243243243246, "grad_norm": 11.828125, "learning_rate": 9.241614553723708e-06, "loss": 1.0743, "step": 21844 }, { "epoch": 2.324430729942541, "grad_norm": 12.59375, "learning_rate": 9.24157902217169e-06, "loss": 0.8453, "step": 21845 }, { "epoch": 2.3245371355607576, "grad_norm": 13.2890625, "learning_rate": 9.241543490619671e-06, "loss": 0.8177, "step": 21846 }, { "epoch": 2.324643541178974, "grad_norm": 11.6328125, "learning_rate": 9.241507959067653e-06, "loss": 0.9015, "step": 21847 }, { "epoch": 2.324749946797191, "grad_norm": 12.9140625, "learning_rate": 9.241472427515634e-06, "loss": 0.8956, "step": 21848 }, { "epoch": 2.3248563524154076, "grad_norm": 16.0625, "learning_rate": 9.241436895963618e-06, "loss": 1.0009, "step": 21849 }, { "epoch": 2.324962758033624, "grad_norm": 12.6328125, "learning_rate": 9.2414013644116e-06, "loss": 0.9433, "step": 21850 }, { "epoch": 2.3250691636518406, "grad_norm": 11.5625, "learning_rate": 9.24136583285958e-06, "loss": 0.98, "step": 21851 }, { "epoch": 2.3251755692700575, "grad_norm": 14.9375, "learning_rate": 9.241330301307562e-06, "loss": 0.8956, "step": 21852 }, { "epoch": 2.325281974888274, "grad_norm": 11.890625, "learning_rate": 9.241294769755544e-06, "loss": 0.839, "step": 21853 }, { "epoch": 2.3253883805064905, "grad_norm": 13.1171875, "learning_rate": 9.241259238203526e-06, "loss": 0.8887, "step": 21854 }, { "epoch": 2.3254947861247075, "grad_norm": 16.03125, "learning_rate": 9.241223706651507e-06, "loss": 0.9605, "step": 21855 }, { "epoch": 2.325601191742924, "grad_norm": 17.75, "learning_rate": 9.241188175099489e-06, "loss": 0.9931, "step": 21856 }, { "epoch": 2.3257075973611405, "grad_norm": 9.7734375, "learning_rate": 9.24115264354747e-06, "loss": 0.8651, "step": 21857 }, { "epoch": 2.3258140029793575, "grad_norm": 17.03125, "learning_rate": 9.241117111995452e-06, "loss": 0.8659, "step": 21858 }, { "epoch": 2.325920408597574, "grad_norm": 13.671875, "learning_rate": 9.241081580443435e-06, "loss": 0.8982, "step": 21859 }, { "epoch": 2.3260268142157905, "grad_norm": 18.109375, "learning_rate": 9.241046048891417e-06, "loss": 1.0449, "step": 21860 }, { "epoch": 2.3261332198340074, "grad_norm": 13.890625, "learning_rate": 9.241010517339399e-06, "loss": 0.743, "step": 21861 }, { "epoch": 2.326239625452224, "grad_norm": 13.3671875, "learning_rate": 9.24097498578738e-06, "loss": 1.1396, "step": 21862 }, { "epoch": 2.3263460310704405, "grad_norm": 11.0390625, "learning_rate": 9.240939454235362e-06, "loss": 0.9118, "step": 21863 }, { "epoch": 2.326452436688657, "grad_norm": 11.4921875, "learning_rate": 9.240903922683343e-06, "loss": 0.8239, "step": 21864 }, { "epoch": 2.326558842306874, "grad_norm": 18.328125, "learning_rate": 9.240868391131325e-06, "loss": 0.8336, "step": 21865 }, { "epoch": 2.3266652479250904, "grad_norm": 12.5078125, "learning_rate": 9.240832859579307e-06, "loss": 0.8348, "step": 21866 }, { "epoch": 2.326771653543307, "grad_norm": 12.4609375, "learning_rate": 9.240797328027288e-06, "loss": 0.7898, "step": 21867 }, { "epoch": 2.326878059161524, "grad_norm": 12.328125, "learning_rate": 9.24076179647527e-06, "loss": 0.9989, "step": 21868 }, { "epoch": 2.3269844647797404, "grad_norm": 13.0234375, "learning_rate": 9.240726264923251e-06, "loss": 0.9208, "step": 21869 }, { "epoch": 2.327090870397957, "grad_norm": 15.34375, "learning_rate": 9.240690733371235e-06, "loss": 0.9824, "step": 21870 }, { "epoch": 2.3271972760161734, "grad_norm": 15.984375, "learning_rate": 9.240655201819216e-06, "loss": 0.7528, "step": 21871 }, { "epoch": 2.3273036816343904, "grad_norm": 15.78125, "learning_rate": 9.240619670267198e-06, "loss": 0.9896, "step": 21872 }, { "epoch": 2.327410087252607, "grad_norm": 11.7109375, "learning_rate": 9.24058413871518e-06, "loss": 0.8743, "step": 21873 }, { "epoch": 2.3275164928708234, "grad_norm": 13.21875, "learning_rate": 9.240548607163161e-06, "loss": 0.7267, "step": 21874 }, { "epoch": 2.3276228984890404, "grad_norm": 13.28125, "learning_rate": 9.240513075611144e-06, "loss": 0.9426, "step": 21875 }, { "epoch": 2.327729304107257, "grad_norm": 12.2578125, "learning_rate": 9.240477544059126e-06, "loss": 0.9738, "step": 21876 }, { "epoch": 2.3278357097254734, "grad_norm": 14.4765625, "learning_rate": 9.240442012507108e-06, "loss": 0.7872, "step": 21877 }, { "epoch": 2.3279421153436903, "grad_norm": 14.5625, "learning_rate": 9.24040648095509e-06, "loss": 1.0369, "step": 21878 }, { "epoch": 2.328048520961907, "grad_norm": 9.9765625, "learning_rate": 9.24037094940307e-06, "loss": 0.7976, "step": 21879 }, { "epoch": 2.3281549265801234, "grad_norm": 13.3203125, "learning_rate": 9.240335417851052e-06, "loss": 0.834, "step": 21880 }, { "epoch": 2.3282613321983403, "grad_norm": 15.5078125, "learning_rate": 9.240299886299034e-06, "loss": 0.9058, "step": 21881 }, { "epoch": 2.328367737816557, "grad_norm": 13.6484375, "learning_rate": 9.240264354747016e-06, "loss": 1.0742, "step": 21882 }, { "epoch": 2.3284741434347733, "grad_norm": 14.640625, "learning_rate": 9.240228823194997e-06, "loss": 0.9158, "step": 21883 }, { "epoch": 2.32858054905299, "grad_norm": 12.140625, "learning_rate": 9.240193291642979e-06, "loss": 0.9348, "step": 21884 }, { "epoch": 2.328686954671207, "grad_norm": 12.96875, "learning_rate": 9.240157760090962e-06, "loss": 0.9456, "step": 21885 }, { "epoch": 2.3287933602894233, "grad_norm": 12.8671875, "learning_rate": 9.240122228538944e-06, "loss": 0.9359, "step": 21886 }, { "epoch": 2.32889976590764, "grad_norm": 11.4140625, "learning_rate": 9.240086696986925e-06, "loss": 0.9087, "step": 21887 }, { "epoch": 2.3290061715258563, "grad_norm": 11.4765625, "learning_rate": 9.240051165434907e-06, "loss": 0.9191, "step": 21888 }, { "epoch": 2.3291125771440733, "grad_norm": 16.46875, "learning_rate": 9.240015633882889e-06, "loss": 0.7163, "step": 21889 }, { "epoch": 2.32921898276229, "grad_norm": 14.46875, "learning_rate": 9.239980102330872e-06, "loss": 0.8846, "step": 21890 }, { "epoch": 2.3293253883805063, "grad_norm": 13.8046875, "learning_rate": 9.239944570778852e-06, "loss": 0.9526, "step": 21891 }, { "epoch": 2.3294317939987232, "grad_norm": 15.6953125, "learning_rate": 9.239909039226833e-06, "loss": 1.114, "step": 21892 }, { "epoch": 2.3295381996169398, "grad_norm": 13.953125, "learning_rate": 9.239873507674815e-06, "loss": 0.8381, "step": 21893 }, { "epoch": 2.3296446052351563, "grad_norm": 12.03125, "learning_rate": 9.239837976122797e-06, "loss": 0.6828, "step": 21894 }, { "epoch": 2.329751010853373, "grad_norm": 13.6640625, "learning_rate": 9.23980244457078e-06, "loss": 0.8272, "step": 21895 }, { "epoch": 2.3298574164715897, "grad_norm": 15.984375, "learning_rate": 9.239766913018762e-06, "loss": 0.9435, "step": 21896 }, { "epoch": 2.3299638220898062, "grad_norm": 13.28125, "learning_rate": 9.239731381466743e-06, "loss": 0.8712, "step": 21897 }, { "epoch": 2.330070227708023, "grad_norm": 16.625, "learning_rate": 9.239695849914725e-06, "loss": 0.9624, "step": 21898 }, { "epoch": 2.3301766333262397, "grad_norm": 15.953125, "learning_rate": 9.239660318362706e-06, "loss": 0.9164, "step": 21899 }, { "epoch": 2.330283038944456, "grad_norm": 12.75, "learning_rate": 9.23962478681069e-06, "loss": 0.9532, "step": 21900 }, { "epoch": 2.3303894445626727, "grad_norm": 14.046875, "learning_rate": 9.239589255258671e-06, "loss": 0.9304, "step": 21901 }, { "epoch": 2.3304958501808897, "grad_norm": 16.953125, "learning_rate": 9.239553723706653e-06, "loss": 0.8936, "step": 21902 }, { "epoch": 2.330602255799106, "grad_norm": 11.5625, "learning_rate": 9.239518192154634e-06, "loss": 0.8381, "step": 21903 }, { "epoch": 2.3307086614173227, "grad_norm": 12.7265625, "learning_rate": 9.239482660602616e-06, "loss": 0.9138, "step": 21904 }, { "epoch": 2.3308150670355396, "grad_norm": 13.9921875, "learning_rate": 9.239447129050598e-06, "loss": 1.0802, "step": 21905 }, { "epoch": 2.330921472653756, "grad_norm": 11.7890625, "learning_rate": 9.23941159749858e-06, "loss": 0.8628, "step": 21906 }, { "epoch": 2.3310278782719727, "grad_norm": 12.6796875, "learning_rate": 9.239376065946561e-06, "loss": 1.0639, "step": 21907 }, { "epoch": 2.331134283890189, "grad_norm": 17.328125, "learning_rate": 9.239340534394543e-06, "loss": 0.9462, "step": 21908 }, { "epoch": 2.331240689508406, "grad_norm": 11.96875, "learning_rate": 9.239305002842524e-06, "loss": 0.9804, "step": 21909 }, { "epoch": 2.3313470951266226, "grad_norm": 17.71875, "learning_rate": 9.239269471290506e-06, "loss": 0.8435, "step": 21910 }, { "epoch": 2.331453500744839, "grad_norm": 17.421875, "learning_rate": 9.239233939738489e-06, "loss": 0.9968, "step": 21911 }, { "epoch": 2.331559906363056, "grad_norm": 11.265625, "learning_rate": 9.23919840818647e-06, "loss": 0.9027, "step": 21912 }, { "epoch": 2.3316663119812726, "grad_norm": 10.3125, "learning_rate": 9.239162876634452e-06, "loss": 1.0384, "step": 21913 }, { "epoch": 2.331772717599489, "grad_norm": 11.4375, "learning_rate": 9.239127345082434e-06, "loss": 1.01, "step": 21914 }, { "epoch": 2.331879123217706, "grad_norm": 15.5078125, "learning_rate": 9.239091813530415e-06, "loss": 1.1063, "step": 21915 }, { "epoch": 2.3319855288359226, "grad_norm": 14.0546875, "learning_rate": 9.239056281978399e-06, "loss": 0.8641, "step": 21916 }, { "epoch": 2.332091934454139, "grad_norm": 11.5, "learning_rate": 9.23902075042638e-06, "loss": 0.8651, "step": 21917 }, { "epoch": 2.332198340072356, "grad_norm": 25.28125, "learning_rate": 9.238985218874362e-06, "loss": 1.0288, "step": 21918 }, { "epoch": 2.3323047456905726, "grad_norm": 17.75, "learning_rate": 9.238949687322344e-06, "loss": 0.7475, "step": 21919 }, { "epoch": 2.332411151308789, "grad_norm": 9.8984375, "learning_rate": 9.238914155770325e-06, "loss": 0.823, "step": 21920 }, { "epoch": 2.3325175569270056, "grad_norm": 11.265625, "learning_rate": 9.238878624218307e-06, "loss": 0.9343, "step": 21921 }, { "epoch": 2.3326239625452225, "grad_norm": 11.2578125, "learning_rate": 9.238843092666288e-06, "loss": 1.0251, "step": 21922 }, { "epoch": 2.332730368163439, "grad_norm": 14.0859375, "learning_rate": 9.23880756111427e-06, "loss": 0.731, "step": 21923 }, { "epoch": 2.3328367737816555, "grad_norm": 16.921875, "learning_rate": 9.238772029562252e-06, "loss": 0.81, "step": 21924 }, { "epoch": 2.3329431793998725, "grad_norm": 17.015625, "learning_rate": 9.238736498010233e-06, "loss": 1.0546, "step": 21925 }, { "epoch": 2.333049585018089, "grad_norm": 10.796875, "learning_rate": 9.238700966458217e-06, "loss": 0.944, "step": 21926 }, { "epoch": 2.3331559906363055, "grad_norm": 20.671875, "learning_rate": 9.238665434906198e-06, "loss": 0.8852, "step": 21927 }, { "epoch": 2.333262396254522, "grad_norm": 12.6328125, "learning_rate": 9.23862990335418e-06, "loss": 0.8336, "step": 21928 }, { "epoch": 2.333368801872739, "grad_norm": 23.28125, "learning_rate": 9.238594371802161e-06, "loss": 0.8647, "step": 21929 }, { "epoch": 2.3334752074909555, "grad_norm": 12.2421875, "learning_rate": 9.238558840250143e-06, "loss": 0.9119, "step": 21930 }, { "epoch": 2.333581613109172, "grad_norm": 11.546875, "learning_rate": 9.238523308698125e-06, "loss": 0.807, "step": 21931 }, { "epoch": 2.333688018727389, "grad_norm": 12.4140625, "learning_rate": 9.238487777146106e-06, "loss": 0.8505, "step": 21932 }, { "epoch": 2.3337944243456055, "grad_norm": 13.4296875, "learning_rate": 9.238452245594088e-06, "loss": 0.9598, "step": 21933 }, { "epoch": 2.333900829963822, "grad_norm": 12.0546875, "learning_rate": 9.23841671404207e-06, "loss": 1.0792, "step": 21934 }, { "epoch": 2.334007235582039, "grad_norm": 13.9140625, "learning_rate": 9.238381182490051e-06, "loss": 0.8904, "step": 21935 }, { "epoch": 2.3341136412002554, "grad_norm": 13.0625, "learning_rate": 9.238345650938034e-06, "loss": 0.9832, "step": 21936 }, { "epoch": 2.334220046818472, "grad_norm": 12.984375, "learning_rate": 9.238310119386016e-06, "loss": 0.7164, "step": 21937 }, { "epoch": 2.334326452436689, "grad_norm": 11.59375, "learning_rate": 9.238274587833997e-06, "loss": 0.9764, "step": 21938 }, { "epoch": 2.3344328580549054, "grad_norm": 12.3984375, "learning_rate": 9.238239056281979e-06, "loss": 0.8977, "step": 21939 }, { "epoch": 2.334539263673122, "grad_norm": 16.59375, "learning_rate": 9.23820352472996e-06, "loss": 0.9292, "step": 21940 }, { "epoch": 2.3346456692913384, "grad_norm": 15.296875, "learning_rate": 9.238167993177942e-06, "loss": 1.0612, "step": 21941 }, { "epoch": 2.3347520749095554, "grad_norm": 12.5390625, "learning_rate": 9.238132461625926e-06, "loss": 0.8187, "step": 21942 }, { "epoch": 2.334858480527772, "grad_norm": 11.1640625, "learning_rate": 9.238096930073907e-06, "loss": 0.7438, "step": 21943 }, { "epoch": 2.3349648861459884, "grad_norm": 13.5625, "learning_rate": 9.238061398521889e-06, "loss": 0.8681, "step": 21944 }, { "epoch": 2.335071291764205, "grad_norm": 10.421875, "learning_rate": 9.23802586696987e-06, "loss": 0.6939, "step": 21945 }, { "epoch": 2.335177697382422, "grad_norm": 12.6171875, "learning_rate": 9.237990335417852e-06, "loss": 0.9446, "step": 21946 }, { "epoch": 2.3352841030006384, "grad_norm": 14.203125, "learning_rate": 9.237954803865834e-06, "loss": 1.0417, "step": 21947 }, { "epoch": 2.335390508618855, "grad_norm": 14.4921875, "learning_rate": 9.237919272313815e-06, "loss": 0.8349, "step": 21948 }, { "epoch": 2.335496914237072, "grad_norm": 14.1328125, "learning_rate": 9.237883740761797e-06, "loss": 1.1423, "step": 21949 }, { "epoch": 2.3356033198552884, "grad_norm": 11.2421875, "learning_rate": 9.237848209209778e-06, "loss": 0.7314, "step": 21950 }, { "epoch": 2.335709725473505, "grad_norm": 28.28125, "learning_rate": 9.23781267765776e-06, "loss": 1.1572, "step": 21951 }, { "epoch": 2.335816131091722, "grad_norm": 14.59375, "learning_rate": 9.237777146105743e-06, "loss": 1.0442, "step": 21952 }, { "epoch": 2.3359225367099383, "grad_norm": 11.0703125, "learning_rate": 9.237741614553725e-06, "loss": 0.8059, "step": 21953 }, { "epoch": 2.336028942328155, "grad_norm": 9.9921875, "learning_rate": 9.237706083001707e-06, "loss": 0.8657, "step": 21954 }, { "epoch": 2.336135347946372, "grad_norm": 12.90625, "learning_rate": 9.237670551449688e-06, "loss": 0.7302, "step": 21955 }, { "epoch": 2.3362417535645883, "grad_norm": 11.1640625, "learning_rate": 9.23763501989767e-06, "loss": 0.9664, "step": 21956 }, { "epoch": 2.336348159182805, "grad_norm": 13.4140625, "learning_rate": 9.237599488345651e-06, "loss": 0.7944, "step": 21957 }, { "epoch": 2.3364545648010213, "grad_norm": 11.4765625, "learning_rate": 9.237563956793633e-06, "loss": 0.7923, "step": 21958 }, { "epoch": 2.3365609704192383, "grad_norm": 14.1484375, "learning_rate": 9.237528425241615e-06, "loss": 0.8172, "step": 21959 }, { "epoch": 2.336667376037455, "grad_norm": 18.140625, "learning_rate": 9.237492893689596e-06, "loss": 1.1388, "step": 21960 }, { "epoch": 2.3367737816556713, "grad_norm": 14.0, "learning_rate": 9.237457362137578e-06, "loss": 0.9591, "step": 21961 }, { "epoch": 2.3368801872738882, "grad_norm": 10.96875, "learning_rate": 9.237421830585561e-06, "loss": 0.8544, "step": 21962 }, { "epoch": 2.3369865928921048, "grad_norm": 11.5078125, "learning_rate": 9.237386299033543e-06, "loss": 0.8887, "step": 21963 }, { "epoch": 2.3370929985103213, "grad_norm": 9.6640625, "learning_rate": 9.237350767481524e-06, "loss": 0.8328, "step": 21964 }, { "epoch": 2.3371994041285378, "grad_norm": 15.3359375, "learning_rate": 9.237315235929506e-06, "loss": 0.848, "step": 21965 }, { "epoch": 2.3373058097467547, "grad_norm": 12.6328125, "learning_rate": 9.237279704377488e-06, "loss": 0.891, "step": 21966 }, { "epoch": 2.3374122153649712, "grad_norm": 12.46875, "learning_rate": 9.237244172825471e-06, "loss": 0.9414, "step": 21967 }, { "epoch": 2.3375186209831877, "grad_norm": 10.046875, "learning_rate": 9.237208641273452e-06, "loss": 0.8373, "step": 21968 }, { "epoch": 2.3376250266014047, "grad_norm": 11.65625, "learning_rate": 9.237173109721434e-06, "loss": 1.0873, "step": 21969 }, { "epoch": 2.337731432219621, "grad_norm": 12.96875, "learning_rate": 9.237137578169416e-06, "loss": 0.8187, "step": 21970 }, { "epoch": 2.3378378378378377, "grad_norm": 13.5, "learning_rate": 9.237102046617397e-06, "loss": 0.9053, "step": 21971 }, { "epoch": 2.3379442434560547, "grad_norm": 13.2578125, "learning_rate": 9.237066515065379e-06, "loss": 0.8477, "step": 21972 }, { "epoch": 2.338050649074271, "grad_norm": 15.046875, "learning_rate": 9.23703098351336e-06, "loss": 0.9886, "step": 21973 }, { "epoch": 2.3381570546924877, "grad_norm": 19.53125, "learning_rate": 9.236995451961342e-06, "loss": 1.2574, "step": 21974 }, { "epoch": 2.3382634603107046, "grad_norm": 12.0546875, "learning_rate": 9.236959920409324e-06, "loss": 1.0858, "step": 21975 }, { "epoch": 2.338369865928921, "grad_norm": 11.1484375, "learning_rate": 9.236924388857305e-06, "loss": 0.8545, "step": 21976 }, { "epoch": 2.3384762715471377, "grad_norm": 12.0859375, "learning_rate": 9.236888857305289e-06, "loss": 0.9682, "step": 21977 }, { "epoch": 2.338582677165354, "grad_norm": 12.2890625, "learning_rate": 9.23685332575327e-06, "loss": 0.9921, "step": 21978 }, { "epoch": 2.338689082783571, "grad_norm": 13.171875, "learning_rate": 9.236817794201252e-06, "loss": 1.0616, "step": 21979 }, { "epoch": 2.3387954884017876, "grad_norm": 20.125, "learning_rate": 9.236782262649233e-06, "loss": 1.0708, "step": 21980 }, { "epoch": 2.338901894020004, "grad_norm": 10.7421875, "learning_rate": 9.236746731097215e-06, "loss": 0.7181, "step": 21981 }, { "epoch": 2.3390082996382207, "grad_norm": 11.6015625, "learning_rate": 9.236711199545197e-06, "loss": 0.8552, "step": 21982 }, { "epoch": 2.3391147052564376, "grad_norm": 15.65625, "learning_rate": 9.236675667993178e-06, "loss": 0.9001, "step": 21983 }, { "epoch": 2.339221110874654, "grad_norm": 14.3828125, "learning_rate": 9.23664013644116e-06, "loss": 0.9413, "step": 21984 }, { "epoch": 2.3393275164928706, "grad_norm": 11.5234375, "learning_rate": 9.236604604889141e-06, "loss": 0.7619, "step": 21985 }, { "epoch": 2.3394339221110876, "grad_norm": 15.7265625, "learning_rate": 9.236569073337123e-06, "loss": 1.0495, "step": 21986 }, { "epoch": 2.339540327729304, "grad_norm": 13.4765625, "learning_rate": 9.236533541785105e-06, "loss": 0.889, "step": 21987 }, { "epoch": 2.3396467333475206, "grad_norm": 10.609375, "learning_rate": 9.236498010233088e-06, "loss": 0.8262, "step": 21988 }, { "epoch": 2.3397531389657376, "grad_norm": 13.5078125, "learning_rate": 9.23646247868107e-06, "loss": 0.8878, "step": 21989 }, { "epoch": 2.339859544583954, "grad_norm": 14.265625, "learning_rate": 9.236426947129051e-06, "loss": 1.0674, "step": 21990 }, { "epoch": 2.3399659502021706, "grad_norm": 13.140625, "learning_rate": 9.236391415577033e-06, "loss": 0.8495, "step": 21991 }, { "epoch": 2.3400723558203875, "grad_norm": 21.109375, "learning_rate": 9.236355884025014e-06, "loss": 1.0266, "step": 21992 }, { "epoch": 2.340178761438604, "grad_norm": 18.484375, "learning_rate": 9.236320352472998e-06, "loss": 1.1508, "step": 21993 }, { "epoch": 2.3402851670568205, "grad_norm": 12.703125, "learning_rate": 9.23628482092098e-06, "loss": 0.7462, "step": 21994 }, { "epoch": 2.340391572675037, "grad_norm": 14.28125, "learning_rate": 9.236249289368961e-06, "loss": 0.8551, "step": 21995 }, { "epoch": 2.340497978293254, "grad_norm": 12.0390625, "learning_rate": 9.236213757816943e-06, "loss": 0.7849, "step": 21996 }, { "epoch": 2.3406043839114705, "grad_norm": 16.171875, "learning_rate": 9.236178226264924e-06, "loss": 0.9586, "step": 21997 }, { "epoch": 2.340710789529687, "grad_norm": 11.421875, "learning_rate": 9.236142694712906e-06, "loss": 0.7724, "step": 21998 }, { "epoch": 2.340817195147904, "grad_norm": 13.2265625, "learning_rate": 9.236107163160887e-06, "loss": 0.8006, "step": 21999 }, { "epoch": 2.3409236007661205, "grad_norm": 14.03125, "learning_rate": 9.236071631608869e-06, "loss": 0.8764, "step": 22000 }, { "epoch": 2.341030006384337, "grad_norm": 11.2578125, "learning_rate": 9.23603610005685e-06, "loss": 0.9911, "step": 22001 }, { "epoch": 2.3411364120025535, "grad_norm": 14.8828125, "learning_rate": 9.236000568504832e-06, "loss": 1.0271, "step": 22002 }, { "epoch": 2.3412428176207705, "grad_norm": 18.53125, "learning_rate": 9.235965036952815e-06, "loss": 1.0068, "step": 22003 }, { "epoch": 2.341349223238987, "grad_norm": 13.1328125, "learning_rate": 9.235929505400797e-06, "loss": 0.7636, "step": 22004 }, { "epoch": 2.3414556288572035, "grad_norm": 12.234375, "learning_rate": 9.235893973848779e-06, "loss": 0.7526, "step": 22005 }, { "epoch": 2.3415620344754204, "grad_norm": 16.28125, "learning_rate": 9.23585844229676e-06, "loss": 0.9781, "step": 22006 }, { "epoch": 2.341668440093637, "grad_norm": 12.0625, "learning_rate": 9.235822910744742e-06, "loss": 0.8685, "step": 22007 }, { "epoch": 2.3417748457118535, "grad_norm": 11.2578125, "learning_rate": 9.235787379192725e-06, "loss": 0.7618, "step": 22008 }, { "epoch": 2.3418812513300704, "grad_norm": 10.421875, "learning_rate": 9.235751847640707e-06, "loss": 0.9313, "step": 22009 }, { "epoch": 2.341987656948287, "grad_norm": 11.5078125, "learning_rate": 9.235716316088688e-06, "loss": 0.7417, "step": 22010 }, { "epoch": 2.3420940625665034, "grad_norm": 12.78125, "learning_rate": 9.235680784536668e-06, "loss": 0.8682, "step": 22011 }, { "epoch": 2.3422004681847204, "grad_norm": 12.296875, "learning_rate": 9.23564525298465e-06, "loss": 0.9103, "step": 22012 }, { "epoch": 2.342306873802937, "grad_norm": 10.484375, "learning_rate": 9.235609721432633e-06, "loss": 0.8597, "step": 22013 }, { "epoch": 2.3424132794211534, "grad_norm": 8.4453125, "learning_rate": 9.235574189880615e-06, "loss": 0.6433, "step": 22014 }, { "epoch": 2.34251968503937, "grad_norm": 23.6875, "learning_rate": 9.235538658328596e-06, "loss": 0.8486, "step": 22015 }, { "epoch": 2.342626090657587, "grad_norm": 12.2421875, "learning_rate": 9.235503126776578e-06, "loss": 0.8148, "step": 22016 }, { "epoch": 2.3427324962758034, "grad_norm": 8.4453125, "learning_rate": 9.23546759522456e-06, "loss": 0.5475, "step": 22017 }, { "epoch": 2.34283890189402, "grad_norm": 13.734375, "learning_rate": 9.235432063672543e-06, "loss": 0.9517, "step": 22018 }, { "epoch": 2.3429453075122364, "grad_norm": 12.703125, "learning_rate": 9.235396532120525e-06, "loss": 0.9006, "step": 22019 }, { "epoch": 2.3430517131304534, "grad_norm": 10.984375, "learning_rate": 9.235361000568506e-06, "loss": 0.8254, "step": 22020 }, { "epoch": 2.34315811874867, "grad_norm": 13.0859375, "learning_rate": 9.235325469016488e-06, "loss": 1.0659, "step": 22021 }, { "epoch": 2.3432645243668864, "grad_norm": 11.921875, "learning_rate": 9.23528993746447e-06, "loss": 0.884, "step": 22022 }, { "epoch": 2.3433709299851033, "grad_norm": 11.0703125, "learning_rate": 9.235254405912451e-06, "loss": 0.6875, "step": 22023 }, { "epoch": 2.34347733560332, "grad_norm": 10.2265625, "learning_rate": 9.235218874360433e-06, "loss": 0.8638, "step": 22024 }, { "epoch": 2.3435837412215363, "grad_norm": 12.84375, "learning_rate": 9.235183342808414e-06, "loss": 0.9512, "step": 22025 }, { "epoch": 2.3436901468397533, "grad_norm": 18.828125, "learning_rate": 9.235147811256396e-06, "loss": 1.0163, "step": 22026 }, { "epoch": 2.34379655245797, "grad_norm": 11.859375, "learning_rate": 9.235112279704377e-06, "loss": 0.826, "step": 22027 }, { "epoch": 2.3439029580761863, "grad_norm": 12.7421875, "learning_rate": 9.235076748152359e-06, "loss": 0.8716, "step": 22028 }, { "epoch": 2.3440093636944033, "grad_norm": 14.65625, "learning_rate": 9.235041216600342e-06, "loss": 1.1284, "step": 22029 }, { "epoch": 2.34411576931262, "grad_norm": 10.8515625, "learning_rate": 9.235005685048324e-06, "loss": 0.8147, "step": 22030 }, { "epoch": 2.3442221749308363, "grad_norm": 10.546875, "learning_rate": 9.234970153496306e-06, "loss": 0.9434, "step": 22031 }, { "epoch": 2.344328580549053, "grad_norm": 13.421875, "learning_rate": 9.234934621944287e-06, "loss": 0.8979, "step": 22032 }, { "epoch": 2.3444349861672698, "grad_norm": 10.84375, "learning_rate": 9.234899090392269e-06, "loss": 0.6955, "step": 22033 }, { "epoch": 2.3445413917854863, "grad_norm": 20.78125, "learning_rate": 9.234863558840252e-06, "loss": 1.0138, "step": 22034 }, { "epoch": 2.3446477974037028, "grad_norm": 14.4140625, "learning_rate": 9.234828027288234e-06, "loss": 1.0852, "step": 22035 }, { "epoch": 2.3447542030219197, "grad_norm": 13.3828125, "learning_rate": 9.234792495736215e-06, "loss": 0.9216, "step": 22036 }, { "epoch": 2.3448606086401362, "grad_norm": 12.15625, "learning_rate": 9.234756964184197e-06, "loss": 0.6704, "step": 22037 }, { "epoch": 2.3449670142583527, "grad_norm": 12.6484375, "learning_rate": 9.234721432632178e-06, "loss": 1.0468, "step": 22038 }, { "epoch": 2.3450734198765693, "grad_norm": 10.296875, "learning_rate": 9.23468590108016e-06, "loss": 0.6909, "step": 22039 }, { "epoch": 2.345179825494786, "grad_norm": 10.5078125, "learning_rate": 9.234650369528142e-06, "loss": 0.7987, "step": 22040 }, { "epoch": 2.3452862311130027, "grad_norm": 13.390625, "learning_rate": 9.234614837976123e-06, "loss": 0.827, "step": 22041 }, { "epoch": 2.3453926367312192, "grad_norm": 15.8203125, "learning_rate": 9.234579306424105e-06, "loss": 0.9314, "step": 22042 }, { "epoch": 2.345499042349436, "grad_norm": 12.90625, "learning_rate": 9.234543774872087e-06, "loss": 0.8035, "step": 22043 }, { "epoch": 2.3456054479676527, "grad_norm": 11.1484375, "learning_rate": 9.23450824332007e-06, "loss": 0.8277, "step": 22044 }, { "epoch": 2.345711853585869, "grad_norm": 13.5703125, "learning_rate": 9.234472711768051e-06, "loss": 0.9083, "step": 22045 }, { "epoch": 2.345818259204086, "grad_norm": 15.6796875, "learning_rate": 9.234437180216033e-06, "loss": 0.9298, "step": 22046 }, { "epoch": 2.3459246648223027, "grad_norm": 15.25, "learning_rate": 9.234401648664015e-06, "loss": 0.9696, "step": 22047 }, { "epoch": 2.346031070440519, "grad_norm": 12.2421875, "learning_rate": 9.234366117111996e-06, "loss": 0.7623, "step": 22048 }, { "epoch": 2.346137476058736, "grad_norm": 16.03125, "learning_rate": 9.234330585559978e-06, "loss": 1.0145, "step": 22049 }, { "epoch": 2.3462438816769526, "grad_norm": 14.515625, "learning_rate": 9.23429505400796e-06, "loss": 0.747, "step": 22050 }, { "epoch": 2.346350287295169, "grad_norm": 11.5625, "learning_rate": 9.234259522455941e-06, "loss": 0.9001, "step": 22051 }, { "epoch": 2.3464566929133857, "grad_norm": 17.109375, "learning_rate": 9.234223990903923e-06, "loss": 0.9731, "step": 22052 }, { "epoch": 2.3465630985316026, "grad_norm": 22.265625, "learning_rate": 9.234188459351904e-06, "loss": 0.8625, "step": 22053 }, { "epoch": 2.346669504149819, "grad_norm": 13.859375, "learning_rate": 9.234152927799888e-06, "loss": 1.0628, "step": 22054 }, { "epoch": 2.3467759097680356, "grad_norm": 15.7578125, "learning_rate": 9.23411739624787e-06, "loss": 0.7796, "step": 22055 }, { "epoch": 2.3468823153862526, "grad_norm": 14.4453125, "learning_rate": 9.23408186469585e-06, "loss": 1.008, "step": 22056 }, { "epoch": 2.346988721004469, "grad_norm": 13.0859375, "learning_rate": 9.234046333143832e-06, "loss": 0.8717, "step": 22057 }, { "epoch": 2.3470951266226856, "grad_norm": 13.703125, "learning_rate": 9.234010801591814e-06, "loss": 0.7692, "step": 22058 }, { "epoch": 2.347201532240902, "grad_norm": 10.34375, "learning_rate": 9.233975270039796e-06, "loss": 0.7661, "step": 22059 }, { "epoch": 2.347307937859119, "grad_norm": 12.515625, "learning_rate": 9.233939738487779e-06, "loss": 0.8015, "step": 22060 }, { "epoch": 2.3474143434773356, "grad_norm": 13.703125, "learning_rate": 9.23390420693576e-06, "loss": 1.046, "step": 22061 }, { "epoch": 2.347520749095552, "grad_norm": 13.0625, "learning_rate": 9.233868675383742e-06, "loss": 0.7806, "step": 22062 }, { "epoch": 2.347627154713769, "grad_norm": 19.875, "learning_rate": 9.233833143831724e-06, "loss": 0.9968, "step": 22063 }, { "epoch": 2.3477335603319855, "grad_norm": 13.8359375, "learning_rate": 9.233797612279705e-06, "loss": 0.8099, "step": 22064 }, { "epoch": 2.347839965950202, "grad_norm": 13.25, "learning_rate": 9.233762080727687e-06, "loss": 0.9555, "step": 22065 }, { "epoch": 2.347946371568419, "grad_norm": 19.859375, "learning_rate": 9.233726549175669e-06, "loss": 0.9328, "step": 22066 }, { "epoch": 2.3480527771866355, "grad_norm": 11.6875, "learning_rate": 9.23369101762365e-06, "loss": 0.9185, "step": 22067 }, { "epoch": 2.348159182804852, "grad_norm": 17.171875, "learning_rate": 9.233655486071632e-06, "loss": 0.9729, "step": 22068 }, { "epoch": 2.348265588423069, "grad_norm": 14.96875, "learning_rate": 9.233619954519613e-06, "loss": 1.1595, "step": 22069 }, { "epoch": 2.3483719940412855, "grad_norm": 16.109375, "learning_rate": 9.233584422967597e-06, "loss": 1.0532, "step": 22070 }, { "epoch": 2.348478399659502, "grad_norm": 13.875, "learning_rate": 9.233548891415578e-06, "loss": 0.9681, "step": 22071 }, { "epoch": 2.3485848052777185, "grad_norm": 26.84375, "learning_rate": 9.23351335986356e-06, "loss": 1.0891, "step": 22072 }, { "epoch": 2.3486912108959355, "grad_norm": 12.7578125, "learning_rate": 9.233477828311541e-06, "loss": 0.8286, "step": 22073 }, { "epoch": 2.348797616514152, "grad_norm": 14.234375, "learning_rate": 9.233442296759523e-06, "loss": 0.9765, "step": 22074 }, { "epoch": 2.3489040221323685, "grad_norm": 10.84375, "learning_rate": 9.233406765207505e-06, "loss": 0.7838, "step": 22075 }, { "epoch": 2.349010427750585, "grad_norm": 18.515625, "learning_rate": 9.233371233655486e-06, "loss": 0.9474, "step": 22076 }, { "epoch": 2.349116833368802, "grad_norm": 17.28125, "learning_rate": 9.233335702103468e-06, "loss": 0.9102, "step": 22077 }, { "epoch": 2.3492232389870185, "grad_norm": 14.0859375, "learning_rate": 9.23330017055145e-06, "loss": 1.0058, "step": 22078 }, { "epoch": 2.349329644605235, "grad_norm": 12.25, "learning_rate": 9.233264638999431e-06, "loss": 0.9525, "step": 22079 }, { "epoch": 2.349436050223452, "grad_norm": 12.90625, "learning_rate": 9.233229107447414e-06, "loss": 0.8522, "step": 22080 }, { "epoch": 2.3495424558416684, "grad_norm": 11.40625, "learning_rate": 9.233193575895396e-06, "loss": 1.0262, "step": 22081 }, { "epoch": 2.349648861459885, "grad_norm": 10.5234375, "learning_rate": 9.233158044343378e-06, "loss": 0.8472, "step": 22082 }, { "epoch": 2.349755267078102, "grad_norm": 12.3984375, "learning_rate": 9.23312251279136e-06, "loss": 0.7932, "step": 22083 }, { "epoch": 2.3498616726963184, "grad_norm": 29.03125, "learning_rate": 9.233086981239341e-06, "loss": 0.8932, "step": 22084 }, { "epoch": 2.349968078314535, "grad_norm": 10.4765625, "learning_rate": 9.233051449687324e-06, "loss": 0.8888, "step": 22085 }, { "epoch": 2.350074483932752, "grad_norm": 14.2734375, "learning_rate": 9.233015918135306e-06, "loss": 0.9499, "step": 22086 }, { "epoch": 2.3501808895509684, "grad_norm": 11.484375, "learning_rate": 9.232980386583287e-06, "loss": 0.9292, "step": 22087 }, { "epoch": 2.350287295169185, "grad_norm": 13.9453125, "learning_rate": 9.232944855031269e-06, "loss": 0.9529, "step": 22088 }, { "epoch": 2.3503937007874014, "grad_norm": 11.984375, "learning_rate": 9.23290932347925e-06, "loss": 0.9207, "step": 22089 }, { "epoch": 2.3505001064056184, "grad_norm": 13.859375, "learning_rate": 9.232873791927232e-06, "loss": 1.0171, "step": 22090 }, { "epoch": 2.350606512023835, "grad_norm": 10.984375, "learning_rate": 9.232838260375214e-06, "loss": 0.8844, "step": 22091 }, { "epoch": 2.3507129176420514, "grad_norm": 32.1875, "learning_rate": 9.232802728823195e-06, "loss": 0.8984, "step": 22092 }, { "epoch": 2.3508193232602683, "grad_norm": 9.8359375, "learning_rate": 9.232767197271177e-06, "loss": 0.7696, "step": 22093 }, { "epoch": 2.350925728878485, "grad_norm": 13.8828125, "learning_rate": 9.232731665719159e-06, "loss": 0.778, "step": 22094 }, { "epoch": 2.3510321344967013, "grad_norm": 17.125, "learning_rate": 9.232696134167142e-06, "loss": 0.7267, "step": 22095 }, { "epoch": 2.351138540114918, "grad_norm": 19.171875, "learning_rate": 9.232660602615124e-06, "loss": 0.8554, "step": 22096 }, { "epoch": 2.351244945733135, "grad_norm": 11.2421875, "learning_rate": 9.232625071063105e-06, "loss": 0.9145, "step": 22097 }, { "epoch": 2.3513513513513513, "grad_norm": 11.8359375, "learning_rate": 9.232589539511087e-06, "loss": 0.9916, "step": 22098 }, { "epoch": 2.351457756969568, "grad_norm": 15.3515625, "learning_rate": 9.232554007959068e-06, "loss": 1.1063, "step": 22099 }, { "epoch": 2.351564162587785, "grad_norm": 14.8515625, "learning_rate": 9.23251847640705e-06, "loss": 0.8135, "step": 22100 }, { "epoch": 2.3516705682060013, "grad_norm": 14.3828125, "learning_rate": 9.232482944855032e-06, "loss": 0.8309, "step": 22101 }, { "epoch": 2.351776973824218, "grad_norm": 10.796875, "learning_rate": 9.232447413303013e-06, "loss": 0.8488, "step": 22102 }, { "epoch": 2.3518833794424348, "grad_norm": 11.609375, "learning_rate": 9.232411881750995e-06, "loss": 0.8288, "step": 22103 }, { "epoch": 2.3519897850606513, "grad_norm": 11.8515625, "learning_rate": 9.232376350198976e-06, "loss": 0.9473, "step": 22104 }, { "epoch": 2.3520961906788678, "grad_norm": 10.9765625, "learning_rate": 9.232340818646958e-06, "loss": 0.8522, "step": 22105 }, { "epoch": 2.3522025962970847, "grad_norm": 14.8125, "learning_rate": 9.232305287094941e-06, "loss": 0.9739, "step": 22106 }, { "epoch": 2.3523090019153012, "grad_norm": 11.484375, "learning_rate": 9.232269755542923e-06, "loss": 0.8798, "step": 22107 }, { "epoch": 2.3524154075335177, "grad_norm": 9.3125, "learning_rate": 9.232234223990905e-06, "loss": 0.8074, "step": 22108 }, { "epoch": 2.3525218131517343, "grad_norm": 11.3359375, "learning_rate": 9.232198692438886e-06, "loss": 0.7857, "step": 22109 }, { "epoch": 2.352628218769951, "grad_norm": 15.171875, "learning_rate": 9.232163160886868e-06, "loss": 1.2241, "step": 22110 }, { "epoch": 2.3527346243881677, "grad_norm": 12.296875, "learning_rate": 9.232127629334851e-06, "loss": 0.8991, "step": 22111 }, { "epoch": 2.3528410300063842, "grad_norm": 11.1015625, "learning_rate": 9.232092097782833e-06, "loss": 0.9707, "step": 22112 }, { "epoch": 2.3529474356246007, "grad_norm": 10.671875, "learning_rate": 9.232056566230814e-06, "loss": 0.798, "step": 22113 }, { "epoch": 2.3530538412428177, "grad_norm": 15.3046875, "learning_rate": 9.232021034678796e-06, "loss": 0.9023, "step": 22114 }, { "epoch": 2.353160246861034, "grad_norm": 16.1875, "learning_rate": 9.231985503126777e-06, "loss": 1.0433, "step": 22115 }, { "epoch": 2.3532666524792507, "grad_norm": 11.90625, "learning_rate": 9.231949971574759e-06, "loss": 0.8522, "step": 22116 }, { "epoch": 2.3533730580974677, "grad_norm": 17.65625, "learning_rate": 9.23191444002274e-06, "loss": 0.9195, "step": 22117 }, { "epoch": 2.353479463715684, "grad_norm": 12.1484375, "learning_rate": 9.231878908470722e-06, "loss": 0.8044, "step": 22118 }, { "epoch": 2.3535858693339007, "grad_norm": 12.6796875, "learning_rate": 9.231843376918704e-06, "loss": 0.8461, "step": 22119 }, { "epoch": 2.3536922749521176, "grad_norm": 32.46875, "learning_rate": 9.231807845366685e-06, "loss": 0.7986, "step": 22120 }, { "epoch": 2.353798680570334, "grad_norm": 11.2265625, "learning_rate": 9.231772313814669e-06, "loss": 1.045, "step": 22121 }, { "epoch": 2.3539050861885507, "grad_norm": 16.09375, "learning_rate": 9.23173678226265e-06, "loss": 1.0691, "step": 22122 }, { "epoch": 2.3540114918067676, "grad_norm": 25.859375, "learning_rate": 9.231701250710632e-06, "loss": 1.168, "step": 22123 }, { "epoch": 2.354117897424984, "grad_norm": 13.1171875, "learning_rate": 9.231665719158614e-06, "loss": 0.8854, "step": 22124 }, { "epoch": 2.3542243030432006, "grad_norm": 23.703125, "learning_rate": 9.231630187606595e-06, "loss": 1.0769, "step": 22125 }, { "epoch": 2.354330708661417, "grad_norm": 15.8671875, "learning_rate": 9.231594656054579e-06, "loss": 0.9747, "step": 22126 }, { "epoch": 2.354437114279634, "grad_norm": 19.3125, "learning_rate": 9.23155912450256e-06, "loss": 0.9047, "step": 22127 }, { "epoch": 2.3545435198978506, "grad_norm": 10.4765625, "learning_rate": 9.231523592950542e-06, "loss": 0.9487, "step": 22128 }, { "epoch": 2.354649925516067, "grad_norm": 12.984375, "learning_rate": 9.231488061398523e-06, "loss": 0.7711, "step": 22129 }, { "epoch": 2.354756331134284, "grad_norm": 14.953125, "learning_rate": 9.231452529846505e-06, "loss": 1.0109, "step": 22130 }, { "epoch": 2.3548627367525006, "grad_norm": 14.4140625, "learning_rate": 9.231416998294487e-06, "loss": 1.0232, "step": 22131 }, { "epoch": 2.354969142370717, "grad_norm": 10.703125, "learning_rate": 9.231381466742468e-06, "loss": 0.8353, "step": 22132 }, { "epoch": 2.3550755479889336, "grad_norm": 16.9375, "learning_rate": 9.23134593519045e-06, "loss": 1.0179, "step": 22133 }, { "epoch": 2.3551819536071505, "grad_norm": 16.515625, "learning_rate": 9.231310403638431e-06, "loss": 0.8728, "step": 22134 }, { "epoch": 2.355288359225367, "grad_norm": 22.765625, "learning_rate": 9.231274872086413e-06, "loss": 1.1056, "step": 22135 }, { "epoch": 2.3553947648435836, "grad_norm": 20.359375, "learning_rate": 9.231239340534395e-06, "loss": 0.8378, "step": 22136 }, { "epoch": 2.3555011704618005, "grad_norm": 12.2578125, "learning_rate": 9.231203808982378e-06, "loss": 1.1179, "step": 22137 }, { "epoch": 2.355607576080017, "grad_norm": 21.0625, "learning_rate": 9.23116827743036e-06, "loss": 0.8647, "step": 22138 }, { "epoch": 2.3557139816982335, "grad_norm": 10.8046875, "learning_rate": 9.231132745878341e-06, "loss": 0.8199, "step": 22139 }, { "epoch": 2.3558203873164505, "grad_norm": 10.3828125, "learning_rate": 9.231097214326323e-06, "loss": 0.6836, "step": 22140 }, { "epoch": 2.355926792934667, "grad_norm": 13.40625, "learning_rate": 9.231061682774304e-06, "loss": 0.8873, "step": 22141 }, { "epoch": 2.3560331985528835, "grad_norm": 20.578125, "learning_rate": 9.231026151222286e-06, "loss": 0.5719, "step": 22142 }, { "epoch": 2.3561396041711005, "grad_norm": 18.8125, "learning_rate": 9.230990619670268e-06, "loss": 0.9026, "step": 22143 }, { "epoch": 2.356246009789317, "grad_norm": 13.875, "learning_rate": 9.230955088118249e-06, "loss": 1.0157, "step": 22144 }, { "epoch": 2.3563524154075335, "grad_norm": 17.078125, "learning_rate": 9.23091955656623e-06, "loss": 0.9984, "step": 22145 }, { "epoch": 2.35645882102575, "grad_norm": 12.765625, "learning_rate": 9.230884025014212e-06, "loss": 0.9805, "step": 22146 }, { "epoch": 2.356565226643967, "grad_norm": 12.6328125, "learning_rate": 9.230848493462196e-06, "loss": 0.8144, "step": 22147 }, { "epoch": 2.3566716322621835, "grad_norm": 11.5078125, "learning_rate": 9.230812961910177e-06, "loss": 0.8123, "step": 22148 }, { "epoch": 2.3567780378804, "grad_norm": 15.4375, "learning_rate": 9.230777430358159e-06, "loss": 0.7668, "step": 22149 }, { "epoch": 2.356884443498617, "grad_norm": 13.578125, "learning_rate": 9.23074189880614e-06, "loss": 0.8828, "step": 22150 }, { "epoch": 2.3569908491168334, "grad_norm": 12.1328125, "learning_rate": 9.230706367254122e-06, "loss": 0.7797, "step": 22151 }, { "epoch": 2.35709725473505, "grad_norm": 11.3671875, "learning_rate": 9.230670835702105e-06, "loss": 0.8035, "step": 22152 }, { "epoch": 2.3572036603532665, "grad_norm": 13.4140625, "learning_rate": 9.230635304150087e-06, "loss": 1.0943, "step": 22153 }, { "epoch": 2.3573100659714834, "grad_norm": 13.9140625, "learning_rate": 9.230599772598069e-06, "loss": 0.8725, "step": 22154 }, { "epoch": 2.3574164715897, "grad_norm": 11.390625, "learning_rate": 9.23056424104605e-06, "loss": 0.966, "step": 22155 }, { "epoch": 2.3575228772079164, "grad_norm": 10.6953125, "learning_rate": 9.230528709494032e-06, "loss": 0.7293, "step": 22156 }, { "epoch": 2.3576292828261334, "grad_norm": 13.0, "learning_rate": 9.230493177942013e-06, "loss": 0.7141, "step": 22157 }, { "epoch": 2.35773568844435, "grad_norm": 46.375, "learning_rate": 9.230457646389995e-06, "loss": 0.7327, "step": 22158 }, { "epoch": 2.3578420940625664, "grad_norm": 11.6640625, "learning_rate": 9.230422114837977e-06, "loss": 0.8067, "step": 22159 }, { "epoch": 2.3579484996807833, "grad_norm": 10.59375, "learning_rate": 9.230386583285958e-06, "loss": 1.0351, "step": 22160 }, { "epoch": 2.358054905299, "grad_norm": 9.1796875, "learning_rate": 9.23035105173394e-06, "loss": 0.7355, "step": 22161 }, { "epoch": 2.3581613109172164, "grad_norm": 14.296875, "learning_rate": 9.230315520181923e-06, "loss": 0.8432, "step": 22162 }, { "epoch": 2.358267716535433, "grad_norm": 12.90625, "learning_rate": 9.230279988629905e-06, "loss": 0.8577, "step": 22163 }, { "epoch": 2.35837412215365, "grad_norm": 12.2890625, "learning_rate": 9.230244457077886e-06, "loss": 0.7795, "step": 22164 }, { "epoch": 2.3584805277718663, "grad_norm": 13.40625, "learning_rate": 9.230208925525868e-06, "loss": 1.1412, "step": 22165 }, { "epoch": 2.358586933390083, "grad_norm": 16.4375, "learning_rate": 9.23017339397385e-06, "loss": 0.9069, "step": 22166 }, { "epoch": 2.3586933390083, "grad_norm": 14.53125, "learning_rate": 9.230137862421831e-06, "loss": 0.9016, "step": 22167 }, { "epoch": 2.3587997446265163, "grad_norm": 12.5234375, "learning_rate": 9.230102330869813e-06, "loss": 0.7599, "step": 22168 }, { "epoch": 2.358906150244733, "grad_norm": 12.6328125, "learning_rate": 9.230066799317794e-06, "loss": 0.8594, "step": 22169 }, { "epoch": 2.3590125558629493, "grad_norm": 11.8125, "learning_rate": 9.230031267765776e-06, "loss": 0.8612, "step": 22170 }, { "epoch": 2.3591189614811663, "grad_norm": 11.84375, "learning_rate": 9.229995736213758e-06, "loss": 0.9112, "step": 22171 }, { "epoch": 2.359225367099383, "grad_norm": 9.9765625, "learning_rate": 9.229960204661741e-06, "loss": 0.7276, "step": 22172 }, { "epoch": 2.3593317727175993, "grad_norm": 10.5078125, "learning_rate": 9.229924673109722e-06, "loss": 0.9306, "step": 22173 }, { "epoch": 2.3594381783358163, "grad_norm": 13.7109375, "learning_rate": 9.229889141557704e-06, "loss": 0.9794, "step": 22174 }, { "epoch": 2.3595445839540328, "grad_norm": 12.609375, "learning_rate": 9.229853610005686e-06, "loss": 0.8447, "step": 22175 }, { "epoch": 2.3596509895722493, "grad_norm": 10.9140625, "learning_rate": 9.229818078453667e-06, "loss": 0.9118, "step": 22176 }, { "epoch": 2.3597573951904662, "grad_norm": 21.984375, "learning_rate": 9.229782546901649e-06, "loss": 0.9483, "step": 22177 }, { "epoch": 2.3598638008086827, "grad_norm": 18.671875, "learning_rate": 9.229747015349632e-06, "loss": 0.8991, "step": 22178 }, { "epoch": 2.3599702064268993, "grad_norm": 16.328125, "learning_rate": 9.229711483797614e-06, "loss": 0.9449, "step": 22179 }, { "epoch": 2.360076612045116, "grad_norm": 11.6796875, "learning_rate": 9.229675952245595e-06, "loss": 0.9975, "step": 22180 }, { "epoch": 2.3601830176633327, "grad_norm": 10.578125, "learning_rate": 9.229640420693577e-06, "loss": 0.9848, "step": 22181 }, { "epoch": 2.3602894232815492, "grad_norm": 10.0390625, "learning_rate": 9.229604889141559e-06, "loss": 0.7257, "step": 22182 }, { "epoch": 2.3603958288997657, "grad_norm": 13.2421875, "learning_rate": 9.22956935758954e-06, "loss": 0.9572, "step": 22183 }, { "epoch": 2.3605022345179827, "grad_norm": 10.6015625, "learning_rate": 9.229533826037522e-06, "loss": 0.7929, "step": 22184 }, { "epoch": 2.360608640136199, "grad_norm": 14.265625, "learning_rate": 9.229498294485503e-06, "loss": 1.0212, "step": 22185 }, { "epoch": 2.3607150457544157, "grad_norm": 11.3984375, "learning_rate": 9.229462762933485e-06, "loss": 0.7837, "step": 22186 }, { "epoch": 2.3608214513726327, "grad_norm": 17.046875, "learning_rate": 9.229427231381467e-06, "loss": 0.8591, "step": 22187 }, { "epoch": 2.360927856990849, "grad_norm": 16.890625, "learning_rate": 9.22939169982945e-06, "loss": 0.8985, "step": 22188 }, { "epoch": 2.3610342626090657, "grad_norm": 15.890625, "learning_rate": 9.229356168277432e-06, "loss": 0.9781, "step": 22189 }, { "epoch": 2.361140668227282, "grad_norm": 14.3984375, "learning_rate": 9.229320636725413e-06, "loss": 0.8858, "step": 22190 }, { "epoch": 2.361247073845499, "grad_norm": 12.578125, "learning_rate": 9.229285105173395e-06, "loss": 0.7638, "step": 22191 }, { "epoch": 2.3613534794637157, "grad_norm": 9.6484375, "learning_rate": 9.229249573621376e-06, "loss": 0.8442, "step": 22192 }, { "epoch": 2.361459885081932, "grad_norm": 16.234375, "learning_rate": 9.229214042069358e-06, "loss": 0.9387, "step": 22193 }, { "epoch": 2.361566290700149, "grad_norm": 13.796875, "learning_rate": 9.22917851051734e-06, "loss": 1.1343, "step": 22194 }, { "epoch": 2.3616726963183656, "grad_norm": 11.1640625, "learning_rate": 9.229142978965321e-06, "loss": 0.8365, "step": 22195 }, { "epoch": 2.361779101936582, "grad_norm": 13.4765625, "learning_rate": 9.229107447413303e-06, "loss": 0.9657, "step": 22196 }, { "epoch": 2.361885507554799, "grad_norm": 10.3515625, "learning_rate": 9.229071915861284e-06, "loss": 0.7855, "step": 22197 }, { "epoch": 2.3619919131730156, "grad_norm": 9.859375, "learning_rate": 9.229036384309268e-06, "loss": 0.7537, "step": 22198 }, { "epoch": 2.362098318791232, "grad_norm": 16.671875, "learning_rate": 9.22900085275725e-06, "loss": 1.1275, "step": 22199 }, { "epoch": 2.362204724409449, "grad_norm": 12.4375, "learning_rate": 9.228965321205231e-06, "loss": 0.9622, "step": 22200 }, { "epoch": 2.3623111300276656, "grad_norm": 17.5625, "learning_rate": 9.228929789653213e-06, "loss": 0.9154, "step": 22201 }, { "epoch": 2.362417535645882, "grad_norm": 11.5625, "learning_rate": 9.228894258101194e-06, "loss": 0.7842, "step": 22202 }, { "epoch": 2.3625239412640986, "grad_norm": 21.21875, "learning_rate": 9.228858726549177e-06, "loss": 0.845, "step": 22203 }, { "epoch": 2.3626303468823155, "grad_norm": 17.359375, "learning_rate": 9.228823194997159e-06, "loss": 1.2755, "step": 22204 }, { "epoch": 2.362736752500532, "grad_norm": 14.7109375, "learning_rate": 9.22878766344514e-06, "loss": 0.9544, "step": 22205 }, { "epoch": 2.3628431581187486, "grad_norm": 11.8515625, "learning_rate": 9.228752131893122e-06, "loss": 0.9545, "step": 22206 }, { "epoch": 2.362949563736965, "grad_norm": 15.3359375, "learning_rate": 9.228716600341104e-06, "loss": 0.9048, "step": 22207 }, { "epoch": 2.363055969355182, "grad_norm": 11.9609375, "learning_rate": 9.228681068789085e-06, "loss": 0.8575, "step": 22208 }, { "epoch": 2.3631623749733985, "grad_norm": 18.0, "learning_rate": 9.228645537237067e-06, "loss": 0.8434, "step": 22209 }, { "epoch": 2.363268780591615, "grad_norm": 12.3671875, "learning_rate": 9.228610005685049e-06, "loss": 0.8458, "step": 22210 }, { "epoch": 2.363375186209832, "grad_norm": 13.140625, "learning_rate": 9.22857447413303e-06, "loss": 0.8442, "step": 22211 }, { "epoch": 2.3634815918280485, "grad_norm": 12.4296875, "learning_rate": 9.228538942581012e-06, "loss": 1.0248, "step": 22212 }, { "epoch": 2.363587997446265, "grad_norm": 16.75, "learning_rate": 9.228503411028995e-06, "loss": 0.9726, "step": 22213 }, { "epoch": 2.363694403064482, "grad_norm": 11.4375, "learning_rate": 9.228467879476977e-06, "loss": 0.8952, "step": 22214 }, { "epoch": 2.3638008086826985, "grad_norm": 19.28125, "learning_rate": 9.228432347924958e-06, "loss": 0.854, "step": 22215 }, { "epoch": 2.363907214300915, "grad_norm": 12.1796875, "learning_rate": 9.22839681637294e-06, "loss": 1.05, "step": 22216 }, { "epoch": 2.364013619919132, "grad_norm": 12.0546875, "learning_rate": 9.228361284820922e-06, "loss": 0.8193, "step": 22217 }, { "epoch": 2.3641200255373485, "grad_norm": 17.4375, "learning_rate": 9.228325753268903e-06, "loss": 1.023, "step": 22218 }, { "epoch": 2.364226431155565, "grad_norm": 10.375, "learning_rate": 9.228290221716885e-06, "loss": 0.8923, "step": 22219 }, { "epoch": 2.3643328367737815, "grad_norm": 14.5234375, "learning_rate": 9.228254690164866e-06, "loss": 1.1901, "step": 22220 }, { "epoch": 2.3644392423919984, "grad_norm": 12.015625, "learning_rate": 9.228219158612848e-06, "loss": 0.8092, "step": 22221 }, { "epoch": 2.364545648010215, "grad_norm": 10.4375, "learning_rate": 9.22818362706083e-06, "loss": 0.9092, "step": 22222 }, { "epoch": 2.3646520536284314, "grad_norm": 9.4140625, "learning_rate": 9.228148095508811e-06, "loss": 0.7803, "step": 22223 }, { "epoch": 2.3647584592466484, "grad_norm": 15.3359375, "learning_rate": 9.228112563956795e-06, "loss": 0.9145, "step": 22224 }, { "epoch": 2.364864864864865, "grad_norm": 16.078125, "learning_rate": 9.228077032404776e-06, "loss": 1.1749, "step": 22225 }, { "epoch": 2.3649712704830814, "grad_norm": 14.4375, "learning_rate": 9.228041500852758e-06, "loss": 0.965, "step": 22226 }, { "epoch": 2.365077676101298, "grad_norm": 14.703125, "learning_rate": 9.22800596930074e-06, "loss": 0.959, "step": 22227 }, { "epoch": 2.365184081719515, "grad_norm": 12.5625, "learning_rate": 9.227970437748721e-06, "loss": 0.9339, "step": 22228 }, { "epoch": 2.3652904873377314, "grad_norm": 12.0234375, "learning_rate": 9.227934906196704e-06, "loss": 0.8018, "step": 22229 }, { "epoch": 2.365396892955948, "grad_norm": 16.5, "learning_rate": 9.227899374644686e-06, "loss": 0.9796, "step": 22230 }, { "epoch": 2.365503298574165, "grad_norm": 11.3515625, "learning_rate": 9.227863843092668e-06, "loss": 0.7827, "step": 22231 }, { "epoch": 2.3656097041923814, "grad_norm": 11.078125, "learning_rate": 9.227828311540649e-06, "loss": 0.9618, "step": 22232 }, { "epoch": 2.365716109810598, "grad_norm": 12.171875, "learning_rate": 9.22779277998863e-06, "loss": 0.8878, "step": 22233 }, { "epoch": 2.365822515428815, "grad_norm": 14.359375, "learning_rate": 9.227757248436612e-06, "loss": 0.8585, "step": 22234 }, { "epoch": 2.3659289210470313, "grad_norm": 14.8671875, "learning_rate": 9.227721716884594e-06, "loss": 0.8518, "step": 22235 }, { "epoch": 2.366035326665248, "grad_norm": 11.921875, "learning_rate": 9.227686185332576e-06, "loss": 0.7228, "step": 22236 }, { "epoch": 2.366141732283465, "grad_norm": 10.1875, "learning_rate": 9.227650653780557e-06, "loss": 0.9364, "step": 22237 }, { "epoch": 2.3662481379016813, "grad_norm": 9.3046875, "learning_rate": 9.227615122228539e-06, "loss": 0.7879, "step": 22238 }, { "epoch": 2.366354543519898, "grad_norm": 11.40625, "learning_rate": 9.227579590676522e-06, "loss": 0.8146, "step": 22239 }, { "epoch": 2.3664609491381143, "grad_norm": 14.765625, "learning_rate": 9.227544059124504e-06, "loss": 1.0217, "step": 22240 }, { "epoch": 2.3665673547563313, "grad_norm": 10.59375, "learning_rate": 9.227508527572485e-06, "loss": 0.9219, "step": 22241 }, { "epoch": 2.366673760374548, "grad_norm": 17.0, "learning_rate": 9.227472996020467e-06, "loss": 0.9175, "step": 22242 }, { "epoch": 2.3667801659927643, "grad_norm": 14.0703125, "learning_rate": 9.227437464468449e-06, "loss": 0.9621, "step": 22243 }, { "epoch": 2.366886571610981, "grad_norm": 12.2265625, "learning_rate": 9.227401932916432e-06, "loss": 0.8699, "step": 22244 }, { "epoch": 2.3669929772291978, "grad_norm": 13.5703125, "learning_rate": 9.227366401364413e-06, "loss": 0.9172, "step": 22245 }, { "epoch": 2.3670993828474143, "grad_norm": 11.6015625, "learning_rate": 9.227330869812395e-06, "loss": 0.8617, "step": 22246 }, { "epoch": 2.367205788465631, "grad_norm": 11.75, "learning_rate": 9.227295338260377e-06, "loss": 0.9369, "step": 22247 }, { "epoch": 2.3673121940838477, "grad_norm": 13.140625, "learning_rate": 9.227259806708358e-06, "loss": 0.9228, "step": 22248 }, { "epoch": 2.3674185997020643, "grad_norm": 15.0546875, "learning_rate": 9.22722427515634e-06, "loss": 0.9863, "step": 22249 }, { "epoch": 2.3675250053202808, "grad_norm": 16.28125, "learning_rate": 9.227188743604321e-06, "loss": 1.1838, "step": 22250 }, { "epoch": 2.3676314109384977, "grad_norm": 9.6171875, "learning_rate": 9.227153212052303e-06, "loss": 0.8054, "step": 22251 }, { "epoch": 2.3677378165567142, "grad_norm": 12.109375, "learning_rate": 9.227117680500285e-06, "loss": 0.8729, "step": 22252 }, { "epoch": 2.3678442221749307, "grad_norm": 12.125, "learning_rate": 9.227082148948266e-06, "loss": 0.9646, "step": 22253 }, { "epoch": 2.3679506277931477, "grad_norm": 11.859375, "learning_rate": 9.227046617396248e-06, "loss": 0.8918, "step": 22254 }, { "epoch": 2.368057033411364, "grad_norm": 13.1796875, "learning_rate": 9.227011085844231e-06, "loss": 0.9902, "step": 22255 }, { "epoch": 2.3681634390295807, "grad_norm": 13.65625, "learning_rate": 9.226975554292213e-06, "loss": 0.8935, "step": 22256 }, { "epoch": 2.368269844647797, "grad_norm": 12.4375, "learning_rate": 9.226940022740194e-06, "loss": 0.9963, "step": 22257 }, { "epoch": 2.368376250266014, "grad_norm": 11.796875, "learning_rate": 9.226904491188176e-06, "loss": 0.7975, "step": 22258 }, { "epoch": 2.3684826558842307, "grad_norm": 16.09375, "learning_rate": 9.226868959636158e-06, "loss": 0.7957, "step": 22259 }, { "epoch": 2.368589061502447, "grad_norm": 11.6953125, "learning_rate": 9.22683342808414e-06, "loss": 0.8648, "step": 22260 }, { "epoch": 2.368695467120664, "grad_norm": 10.46875, "learning_rate": 9.22679789653212e-06, "loss": 0.7864, "step": 22261 }, { "epoch": 2.3688018727388807, "grad_norm": 14.0234375, "learning_rate": 9.226762364980102e-06, "loss": 0.9491, "step": 22262 }, { "epoch": 2.368908278357097, "grad_norm": 16.265625, "learning_rate": 9.226726833428084e-06, "loss": 1.2267, "step": 22263 }, { "epoch": 2.3690146839753137, "grad_norm": 11.421875, "learning_rate": 9.226691301876066e-06, "loss": 0.8467, "step": 22264 }, { "epoch": 2.3691210895935306, "grad_norm": 9.5703125, "learning_rate": 9.226655770324049e-06, "loss": 0.7297, "step": 22265 }, { "epoch": 2.369227495211747, "grad_norm": 14.28125, "learning_rate": 9.22662023877203e-06, "loss": 0.9751, "step": 22266 }, { "epoch": 2.3693339008299636, "grad_norm": 10.7265625, "learning_rate": 9.226584707220012e-06, "loss": 0.8791, "step": 22267 }, { "epoch": 2.3694403064481806, "grad_norm": 13.125, "learning_rate": 9.226549175667994e-06, "loss": 0.7992, "step": 22268 }, { "epoch": 2.369546712066397, "grad_norm": 14.9140625, "learning_rate": 9.226513644115975e-06, "loss": 0.9234, "step": 22269 }, { "epoch": 2.3696531176846136, "grad_norm": 14.0703125, "learning_rate": 9.226478112563959e-06, "loss": 0.8597, "step": 22270 }, { "epoch": 2.3697595233028306, "grad_norm": 11.9296875, "learning_rate": 9.22644258101194e-06, "loss": 1.0117, "step": 22271 }, { "epoch": 2.369865928921047, "grad_norm": 17.265625, "learning_rate": 9.226407049459922e-06, "loss": 0.8929, "step": 22272 }, { "epoch": 2.3699723345392636, "grad_norm": 9.78125, "learning_rate": 9.226371517907903e-06, "loss": 0.8691, "step": 22273 }, { "epoch": 2.3700787401574805, "grad_norm": 12.3203125, "learning_rate": 9.226335986355885e-06, "loss": 0.9288, "step": 22274 }, { "epoch": 2.370185145775697, "grad_norm": 12.6796875, "learning_rate": 9.226300454803867e-06, "loss": 0.9193, "step": 22275 }, { "epoch": 2.3702915513939136, "grad_norm": 12.4296875, "learning_rate": 9.226264923251848e-06, "loss": 0.81, "step": 22276 }, { "epoch": 2.37039795701213, "grad_norm": 13.21875, "learning_rate": 9.22622939169983e-06, "loss": 0.7609, "step": 22277 }, { "epoch": 2.370504362630347, "grad_norm": 11.6640625, "learning_rate": 9.226193860147812e-06, "loss": 0.7515, "step": 22278 }, { "epoch": 2.3706107682485635, "grad_norm": 11.4609375, "learning_rate": 9.226158328595793e-06, "loss": 0.8438, "step": 22279 }, { "epoch": 2.37071717386678, "grad_norm": 15.171875, "learning_rate": 9.226122797043776e-06, "loss": 1.0199, "step": 22280 }, { "epoch": 2.370823579484997, "grad_norm": 15.9375, "learning_rate": 9.226087265491758e-06, "loss": 0.9974, "step": 22281 }, { "epoch": 2.3709299851032135, "grad_norm": 13.1953125, "learning_rate": 9.22605173393974e-06, "loss": 0.8974, "step": 22282 }, { "epoch": 2.37103639072143, "grad_norm": 16.65625, "learning_rate": 9.226016202387721e-06, "loss": 0.9696, "step": 22283 }, { "epoch": 2.3711427963396465, "grad_norm": 11.5703125, "learning_rate": 9.225980670835703e-06, "loss": 0.6964, "step": 22284 }, { "epoch": 2.3712492019578635, "grad_norm": 10.4453125, "learning_rate": 9.225945139283684e-06, "loss": 0.8804, "step": 22285 }, { "epoch": 2.37135560757608, "grad_norm": 17.5, "learning_rate": 9.225909607731666e-06, "loss": 0.9026, "step": 22286 }, { "epoch": 2.3714620131942965, "grad_norm": 10.8046875, "learning_rate": 9.225874076179648e-06, "loss": 0.802, "step": 22287 }, { "epoch": 2.3715684188125135, "grad_norm": 11.25, "learning_rate": 9.22583854462763e-06, "loss": 0.9547, "step": 22288 }, { "epoch": 2.37167482443073, "grad_norm": 12.8203125, "learning_rate": 9.225803013075611e-06, "loss": 0.7737, "step": 22289 }, { "epoch": 2.3717812300489465, "grad_norm": 15.828125, "learning_rate": 9.225767481523594e-06, "loss": 1.1493, "step": 22290 }, { "epoch": 2.3718876356671634, "grad_norm": 14.7109375, "learning_rate": 9.225731949971576e-06, "loss": 0.9752, "step": 22291 }, { "epoch": 2.37199404128538, "grad_norm": 11.625, "learning_rate": 9.225696418419557e-06, "loss": 0.9153, "step": 22292 }, { "epoch": 2.3721004469035964, "grad_norm": 10.8671875, "learning_rate": 9.225660886867539e-06, "loss": 0.8483, "step": 22293 }, { "epoch": 2.372206852521813, "grad_norm": 11.515625, "learning_rate": 9.22562535531552e-06, "loss": 1.0275, "step": 22294 }, { "epoch": 2.37231325814003, "grad_norm": 9.390625, "learning_rate": 9.225589823763502e-06, "loss": 0.8476, "step": 22295 }, { "epoch": 2.3724196637582464, "grad_norm": 12.0390625, "learning_rate": 9.225554292211486e-06, "loss": 0.8836, "step": 22296 }, { "epoch": 2.372526069376463, "grad_norm": 8.9296875, "learning_rate": 9.225518760659467e-06, "loss": 0.8024, "step": 22297 }, { "epoch": 2.37263247499468, "grad_norm": 12.984375, "learning_rate": 9.225483229107449e-06, "loss": 0.9178, "step": 22298 }, { "epoch": 2.3727388806128964, "grad_norm": 14.71875, "learning_rate": 9.22544769755543e-06, "loss": 0.9591, "step": 22299 }, { "epoch": 2.372845286231113, "grad_norm": 13.46875, "learning_rate": 9.225412166003412e-06, "loss": 0.6718, "step": 22300 }, { "epoch": 2.3729516918493294, "grad_norm": 10.8828125, "learning_rate": 9.225376634451394e-06, "loss": 0.8577, "step": 22301 }, { "epoch": 2.3730580974675464, "grad_norm": 13.390625, "learning_rate": 9.225341102899375e-06, "loss": 0.8539, "step": 22302 }, { "epoch": 2.373164503085763, "grad_norm": 10.5859375, "learning_rate": 9.225305571347357e-06, "loss": 0.7816, "step": 22303 }, { "epoch": 2.3732709087039794, "grad_norm": 13.21875, "learning_rate": 9.225270039795338e-06, "loss": 0.8661, "step": 22304 }, { "epoch": 2.3733773143221963, "grad_norm": 15.1875, "learning_rate": 9.22523450824332e-06, "loss": 0.8217, "step": 22305 }, { "epoch": 2.373483719940413, "grad_norm": 11.15625, "learning_rate": 9.225198976691303e-06, "loss": 0.773, "step": 22306 }, { "epoch": 2.3735901255586294, "grad_norm": 12.3828125, "learning_rate": 9.225163445139285e-06, "loss": 0.8594, "step": 22307 }, { "epoch": 2.3736965311768463, "grad_norm": 19.796875, "learning_rate": 9.225127913587266e-06, "loss": 1.0413, "step": 22308 }, { "epoch": 2.373802936795063, "grad_norm": 11.265625, "learning_rate": 9.225092382035248e-06, "loss": 0.8413, "step": 22309 }, { "epoch": 2.3739093424132793, "grad_norm": 11.5234375, "learning_rate": 9.22505685048323e-06, "loss": 0.8382, "step": 22310 }, { "epoch": 2.3740157480314963, "grad_norm": 10.9921875, "learning_rate": 9.225021318931211e-06, "loss": 0.9699, "step": 22311 }, { "epoch": 2.374122153649713, "grad_norm": 14.1953125, "learning_rate": 9.224985787379193e-06, "loss": 0.8536, "step": 22312 }, { "epoch": 2.3742285592679293, "grad_norm": 12.3046875, "learning_rate": 9.224950255827175e-06, "loss": 0.9137, "step": 22313 }, { "epoch": 2.374334964886146, "grad_norm": 10.765625, "learning_rate": 9.224914724275156e-06, "loss": 0.85, "step": 22314 }, { "epoch": 2.3744413705043628, "grad_norm": 12.1796875, "learning_rate": 9.224879192723138e-06, "loss": 0.7918, "step": 22315 }, { "epoch": 2.3745477761225793, "grad_norm": 9.546875, "learning_rate": 9.224843661171121e-06, "loss": 0.7673, "step": 22316 }, { "epoch": 2.374654181740796, "grad_norm": 18.4375, "learning_rate": 9.224808129619103e-06, "loss": 0.988, "step": 22317 }, { "epoch": 2.3747605873590127, "grad_norm": 11.2421875, "learning_rate": 9.224772598067084e-06, "loss": 1.0446, "step": 22318 }, { "epoch": 2.3748669929772293, "grad_norm": 12.8046875, "learning_rate": 9.224737066515066e-06, "loss": 1.101, "step": 22319 }, { "epoch": 2.3749733985954458, "grad_norm": 8.0703125, "learning_rate": 9.224701534963047e-06, "loss": 0.7513, "step": 22320 }, { "epoch": 2.3750798042136623, "grad_norm": 12.484375, "learning_rate": 9.22466600341103e-06, "loss": 0.8547, "step": 22321 }, { "epoch": 2.3751862098318792, "grad_norm": 13.2421875, "learning_rate": 9.224630471859012e-06, "loss": 0.7545, "step": 22322 }, { "epoch": 2.3752926154500957, "grad_norm": 13.0703125, "learning_rate": 9.224594940306994e-06, "loss": 0.9134, "step": 22323 }, { "epoch": 2.3753990210683122, "grad_norm": 13.2265625, "learning_rate": 9.224559408754976e-06, "loss": 0.7511, "step": 22324 }, { "epoch": 2.375505426686529, "grad_norm": 14.6015625, "learning_rate": 9.224523877202957e-06, "loss": 0.9181, "step": 22325 }, { "epoch": 2.3756118323047457, "grad_norm": 13.046875, "learning_rate": 9.224488345650939e-06, "loss": 0.9758, "step": 22326 }, { "epoch": 2.375718237922962, "grad_norm": 13.5078125, "learning_rate": 9.22445281409892e-06, "loss": 0.9182, "step": 22327 }, { "epoch": 2.375824643541179, "grad_norm": 11.671875, "learning_rate": 9.224417282546902e-06, "loss": 0.854, "step": 22328 }, { "epoch": 2.3759310491593957, "grad_norm": 18.6875, "learning_rate": 9.224381750994884e-06, "loss": 0.8255, "step": 22329 }, { "epoch": 2.376037454777612, "grad_norm": 13.875, "learning_rate": 9.224346219442865e-06, "loss": 0.9735, "step": 22330 }, { "epoch": 2.376143860395829, "grad_norm": 12.34375, "learning_rate": 9.224310687890849e-06, "loss": 0.9295, "step": 22331 }, { "epoch": 2.3762502660140457, "grad_norm": 11.9921875, "learning_rate": 9.22427515633883e-06, "loss": 0.7571, "step": 22332 }, { "epoch": 2.376356671632262, "grad_norm": 13.578125, "learning_rate": 9.224239624786812e-06, "loss": 0.8385, "step": 22333 }, { "epoch": 2.3764630772504787, "grad_norm": 10.9453125, "learning_rate": 9.224204093234793e-06, "loss": 0.9602, "step": 22334 }, { "epoch": 2.3765694828686956, "grad_norm": 14.5390625, "learning_rate": 9.224168561682775e-06, "loss": 0.8255, "step": 22335 }, { "epoch": 2.376675888486912, "grad_norm": 12.3046875, "learning_rate": 9.224133030130757e-06, "loss": 1.0034, "step": 22336 }, { "epoch": 2.3767822941051286, "grad_norm": 12.6796875, "learning_rate": 9.22409749857874e-06, "loss": 0.7707, "step": 22337 }, { "epoch": 2.376888699723345, "grad_norm": 13.75, "learning_rate": 9.224061967026721e-06, "loss": 1.0037, "step": 22338 }, { "epoch": 2.376995105341562, "grad_norm": 15.8671875, "learning_rate": 9.224026435474701e-06, "loss": 0.9436, "step": 22339 }, { "epoch": 2.3771015109597786, "grad_norm": 11.875, "learning_rate": 9.223990903922683e-06, "loss": 1.0612, "step": 22340 }, { "epoch": 2.377207916577995, "grad_norm": 13.203125, "learning_rate": 9.223955372370665e-06, "loss": 0.9823, "step": 22341 }, { "epoch": 2.377314322196212, "grad_norm": 17.1875, "learning_rate": 9.223919840818648e-06, "loss": 1.1077, "step": 22342 }, { "epoch": 2.3774207278144286, "grad_norm": 13.2109375, "learning_rate": 9.22388430926663e-06, "loss": 1.0251, "step": 22343 }, { "epoch": 2.377527133432645, "grad_norm": 17.234375, "learning_rate": 9.223848777714611e-06, "loss": 0.8387, "step": 22344 }, { "epoch": 2.377633539050862, "grad_norm": 13.015625, "learning_rate": 9.223813246162593e-06, "loss": 0.9599, "step": 22345 }, { "epoch": 2.3777399446690786, "grad_norm": 12.6328125, "learning_rate": 9.223777714610574e-06, "loss": 0.9956, "step": 22346 }, { "epoch": 2.377846350287295, "grad_norm": 16.375, "learning_rate": 9.223742183058558e-06, "loss": 0.9312, "step": 22347 }, { "epoch": 2.377952755905512, "grad_norm": 18.796875, "learning_rate": 9.22370665150654e-06, "loss": 0.9665, "step": 22348 }, { "epoch": 2.3780591615237285, "grad_norm": 16.890625, "learning_rate": 9.22367111995452e-06, "loss": 0.7061, "step": 22349 }, { "epoch": 2.378165567141945, "grad_norm": 11.640625, "learning_rate": 9.223635588402502e-06, "loss": 0.9327, "step": 22350 }, { "epoch": 2.3782719727601616, "grad_norm": 11.59375, "learning_rate": 9.223600056850484e-06, "loss": 0.8099, "step": 22351 }, { "epoch": 2.3783783783783785, "grad_norm": 16.78125, "learning_rate": 9.223564525298466e-06, "loss": 0.9374, "step": 22352 }, { "epoch": 2.378484783996595, "grad_norm": 13.6796875, "learning_rate": 9.223528993746447e-06, "loss": 0.9105, "step": 22353 }, { "epoch": 2.3785911896148115, "grad_norm": 14.2421875, "learning_rate": 9.223493462194429e-06, "loss": 0.988, "step": 22354 }, { "epoch": 2.3786975952330285, "grad_norm": 13.875, "learning_rate": 9.22345793064241e-06, "loss": 0.9953, "step": 22355 }, { "epoch": 2.378804000851245, "grad_norm": 14.0859375, "learning_rate": 9.223422399090392e-06, "loss": 0.8732, "step": 22356 }, { "epoch": 2.3789104064694615, "grad_norm": 25.578125, "learning_rate": 9.223386867538375e-06, "loss": 1.0828, "step": 22357 }, { "epoch": 2.379016812087678, "grad_norm": 13.0234375, "learning_rate": 9.223351335986357e-06, "loss": 0.9214, "step": 22358 }, { "epoch": 2.379123217705895, "grad_norm": 10.609375, "learning_rate": 9.223315804434339e-06, "loss": 0.7411, "step": 22359 }, { "epoch": 2.3792296233241115, "grad_norm": 16.15625, "learning_rate": 9.22328027288232e-06, "loss": 0.9162, "step": 22360 }, { "epoch": 2.379336028942328, "grad_norm": 13.3203125, "learning_rate": 9.223244741330302e-06, "loss": 0.7407, "step": 22361 }, { "epoch": 2.379442434560545, "grad_norm": 16.609375, "learning_rate": 9.223209209778285e-06, "loss": 0.9008, "step": 22362 }, { "epoch": 2.3795488401787614, "grad_norm": 13.859375, "learning_rate": 9.223173678226267e-06, "loss": 0.8828, "step": 22363 }, { "epoch": 2.379655245796978, "grad_norm": 12.796875, "learning_rate": 9.223138146674248e-06, "loss": 0.9125, "step": 22364 }, { "epoch": 2.379761651415195, "grad_norm": 15.4609375, "learning_rate": 9.22310261512223e-06, "loss": 0.7819, "step": 22365 }, { "epoch": 2.3798680570334114, "grad_norm": 11.84375, "learning_rate": 9.223067083570212e-06, "loss": 0.8289, "step": 22366 }, { "epoch": 2.379974462651628, "grad_norm": 14.1796875, "learning_rate": 9.223031552018193e-06, "loss": 0.9318, "step": 22367 }, { "epoch": 2.380080868269845, "grad_norm": 11.59375, "learning_rate": 9.222996020466175e-06, "loss": 0.8679, "step": 22368 }, { "epoch": 2.3801872738880614, "grad_norm": 10.828125, "learning_rate": 9.222960488914156e-06, "loss": 0.9131, "step": 22369 }, { "epoch": 2.380293679506278, "grad_norm": 11.6953125, "learning_rate": 9.222924957362138e-06, "loss": 0.9327, "step": 22370 }, { "epoch": 2.3804000851244944, "grad_norm": 10.7421875, "learning_rate": 9.22288942581012e-06, "loss": 0.7685, "step": 22371 }, { "epoch": 2.3805064907427114, "grad_norm": 10.5859375, "learning_rate": 9.222853894258101e-06, "loss": 0.7241, "step": 22372 }, { "epoch": 2.380612896360928, "grad_norm": 11.78125, "learning_rate": 9.222818362706084e-06, "loss": 0.7248, "step": 22373 }, { "epoch": 2.3807193019791444, "grad_norm": 16.265625, "learning_rate": 9.222782831154066e-06, "loss": 0.943, "step": 22374 }, { "epoch": 2.380825707597361, "grad_norm": 14.625, "learning_rate": 9.222747299602048e-06, "loss": 0.8494, "step": 22375 }, { "epoch": 2.380932113215578, "grad_norm": 13.6171875, "learning_rate": 9.22271176805003e-06, "loss": 1.083, "step": 22376 }, { "epoch": 2.3810385188337944, "grad_norm": 11.2421875, "learning_rate": 9.222676236498011e-06, "loss": 0.9044, "step": 22377 }, { "epoch": 2.381144924452011, "grad_norm": 13.5390625, "learning_rate": 9.222640704945993e-06, "loss": 0.9147, "step": 22378 }, { "epoch": 2.381251330070228, "grad_norm": 16.484375, "learning_rate": 9.222605173393974e-06, "loss": 0.9259, "step": 22379 }, { "epoch": 2.3813577356884443, "grad_norm": 12.7109375, "learning_rate": 9.222569641841956e-06, "loss": 0.9207, "step": 22380 }, { "epoch": 2.381464141306661, "grad_norm": 11.7734375, "learning_rate": 9.222534110289937e-06, "loss": 0.7579, "step": 22381 }, { "epoch": 2.381570546924878, "grad_norm": 13.7109375, "learning_rate": 9.222498578737919e-06, "loss": 0.7916, "step": 22382 }, { "epoch": 2.3816769525430943, "grad_norm": 16.359375, "learning_rate": 9.222463047185902e-06, "loss": 0.8931, "step": 22383 }, { "epoch": 2.381783358161311, "grad_norm": 14.609375, "learning_rate": 9.222427515633884e-06, "loss": 0.8881, "step": 22384 }, { "epoch": 2.3818897637795278, "grad_norm": 10.484375, "learning_rate": 9.222391984081865e-06, "loss": 0.8131, "step": 22385 }, { "epoch": 2.3819961693977443, "grad_norm": 13.2890625, "learning_rate": 9.222356452529847e-06, "loss": 0.8702, "step": 22386 }, { "epoch": 2.382102575015961, "grad_norm": 11.8125, "learning_rate": 9.222320920977829e-06, "loss": 0.8554, "step": 22387 }, { "epoch": 2.3822089806341773, "grad_norm": 14.796875, "learning_rate": 9.222285389425812e-06, "loss": 0.8977, "step": 22388 }, { "epoch": 2.3823153862523943, "grad_norm": 12.5546875, "learning_rate": 9.222249857873794e-06, "loss": 0.7562, "step": 22389 }, { "epoch": 2.3824217918706108, "grad_norm": 12.6015625, "learning_rate": 9.222214326321775e-06, "loss": 0.95, "step": 22390 }, { "epoch": 2.3825281974888273, "grad_norm": 11.96875, "learning_rate": 9.222178794769757e-06, "loss": 0.9358, "step": 22391 }, { "epoch": 2.3826346031070442, "grad_norm": 12.265625, "learning_rate": 9.222143263217738e-06, "loss": 0.7872, "step": 22392 }, { "epoch": 2.3827410087252607, "grad_norm": 12.2734375, "learning_rate": 9.22210773166572e-06, "loss": 0.8942, "step": 22393 }, { "epoch": 2.3828474143434772, "grad_norm": 15.75, "learning_rate": 9.222072200113702e-06, "loss": 0.9724, "step": 22394 }, { "epoch": 2.3829538199616938, "grad_norm": 12.84375, "learning_rate": 9.222036668561683e-06, "loss": 1.0057, "step": 22395 }, { "epoch": 2.3830602255799107, "grad_norm": 10.8828125, "learning_rate": 9.222001137009665e-06, "loss": 0.9463, "step": 22396 }, { "epoch": 2.383166631198127, "grad_norm": 15.0078125, "learning_rate": 9.221965605457646e-06, "loss": 0.9552, "step": 22397 }, { "epoch": 2.3832730368163437, "grad_norm": 13.5390625, "learning_rate": 9.22193007390563e-06, "loss": 0.8259, "step": 22398 }, { "epoch": 2.3833794424345607, "grad_norm": 12.2734375, "learning_rate": 9.221894542353611e-06, "loss": 1.0047, "step": 22399 }, { "epoch": 2.383485848052777, "grad_norm": 13.765625, "learning_rate": 9.221859010801593e-06, "loss": 0.7754, "step": 22400 }, { "epoch": 2.3835922536709937, "grad_norm": 12.4296875, "learning_rate": 9.221823479249575e-06, "loss": 1.0334, "step": 22401 }, { "epoch": 2.3836986592892107, "grad_norm": 13.3984375, "learning_rate": 9.221787947697556e-06, "loss": 0.916, "step": 22402 }, { "epoch": 2.383805064907427, "grad_norm": 12.9453125, "learning_rate": 9.221752416145538e-06, "loss": 0.8381, "step": 22403 }, { "epoch": 2.3839114705256437, "grad_norm": 14.7109375, "learning_rate": 9.22171688459352e-06, "loss": 0.8516, "step": 22404 }, { "epoch": 2.3840178761438606, "grad_norm": 13.8515625, "learning_rate": 9.221681353041501e-06, "loss": 0.9803, "step": 22405 }, { "epoch": 2.384124281762077, "grad_norm": 9.4765625, "learning_rate": 9.221645821489483e-06, "loss": 0.9057, "step": 22406 }, { "epoch": 2.3842306873802936, "grad_norm": 17.375, "learning_rate": 9.221610289937464e-06, "loss": 0.8561, "step": 22407 }, { "epoch": 2.38433709299851, "grad_norm": 10.71875, "learning_rate": 9.221574758385447e-06, "loss": 0.9608, "step": 22408 }, { "epoch": 2.384443498616727, "grad_norm": 13.0, "learning_rate": 9.221539226833429e-06, "loss": 0.8946, "step": 22409 }, { "epoch": 2.3845499042349436, "grad_norm": 14.0703125, "learning_rate": 9.22150369528141e-06, "loss": 0.7574, "step": 22410 }, { "epoch": 2.38465630985316, "grad_norm": 9.765625, "learning_rate": 9.221468163729392e-06, "loss": 0.7447, "step": 22411 }, { "epoch": 2.384762715471377, "grad_norm": 12.359375, "learning_rate": 9.221432632177374e-06, "loss": 0.8648, "step": 22412 }, { "epoch": 2.3848691210895936, "grad_norm": 13.4296875, "learning_rate": 9.221397100625356e-06, "loss": 0.8256, "step": 22413 }, { "epoch": 2.38497552670781, "grad_norm": 9.796875, "learning_rate": 9.221361569073339e-06, "loss": 0.8659, "step": 22414 }, { "epoch": 2.3850819323260266, "grad_norm": 15.984375, "learning_rate": 9.22132603752132e-06, "loss": 0.9009, "step": 22415 }, { "epoch": 2.3851883379442436, "grad_norm": 12.4921875, "learning_rate": 9.221290505969302e-06, "loss": 0.8872, "step": 22416 }, { "epoch": 2.38529474356246, "grad_norm": 14.7890625, "learning_rate": 9.221254974417284e-06, "loss": 0.9219, "step": 22417 }, { "epoch": 2.3854011491806766, "grad_norm": 10.78125, "learning_rate": 9.221219442865265e-06, "loss": 0.8987, "step": 22418 }, { "epoch": 2.3855075547988935, "grad_norm": 12.0234375, "learning_rate": 9.221183911313247e-06, "loss": 0.8553, "step": 22419 }, { "epoch": 2.38561396041711, "grad_norm": 9.9765625, "learning_rate": 9.221148379761228e-06, "loss": 0.8289, "step": 22420 }, { "epoch": 2.3857203660353266, "grad_norm": 13.796875, "learning_rate": 9.22111284820921e-06, "loss": 0.8923, "step": 22421 }, { "epoch": 2.3858267716535435, "grad_norm": 10.8203125, "learning_rate": 9.221077316657192e-06, "loss": 0.7903, "step": 22422 }, { "epoch": 2.38593317727176, "grad_norm": 13.6953125, "learning_rate": 9.221041785105173e-06, "loss": 0.9179, "step": 22423 }, { "epoch": 2.3860395828899765, "grad_norm": 14.984375, "learning_rate": 9.221006253553157e-06, "loss": 0.9225, "step": 22424 }, { "epoch": 2.386145988508193, "grad_norm": 8.6875, "learning_rate": 9.220970722001138e-06, "loss": 0.6566, "step": 22425 }, { "epoch": 2.38625239412641, "grad_norm": 12.7890625, "learning_rate": 9.22093519044912e-06, "loss": 0.7578, "step": 22426 }, { "epoch": 2.3863587997446265, "grad_norm": 15.5234375, "learning_rate": 9.220899658897101e-06, "loss": 0.8999, "step": 22427 }, { "epoch": 2.386465205362843, "grad_norm": 12.1171875, "learning_rate": 9.220864127345083e-06, "loss": 1.0368, "step": 22428 }, { "epoch": 2.38657161098106, "grad_norm": 18.59375, "learning_rate": 9.220828595793065e-06, "loss": 0.8603, "step": 22429 }, { "epoch": 2.3866780165992765, "grad_norm": 12.7421875, "learning_rate": 9.220793064241046e-06, "loss": 0.9478, "step": 22430 }, { "epoch": 2.386784422217493, "grad_norm": 15.5703125, "learning_rate": 9.220757532689028e-06, "loss": 1.0093, "step": 22431 }, { "epoch": 2.3868908278357095, "grad_norm": 11.3515625, "learning_rate": 9.22072200113701e-06, "loss": 0.8706, "step": 22432 }, { "epoch": 2.3869972334539264, "grad_norm": 11.2890625, "learning_rate": 9.220686469584991e-06, "loss": 0.9317, "step": 22433 }, { "epoch": 2.387103639072143, "grad_norm": 11.125, "learning_rate": 9.220650938032974e-06, "loss": 0.9435, "step": 22434 }, { "epoch": 2.3872100446903595, "grad_norm": 10.3359375, "learning_rate": 9.220615406480956e-06, "loss": 0.6947, "step": 22435 }, { "epoch": 2.3873164503085764, "grad_norm": 15.5625, "learning_rate": 9.220579874928938e-06, "loss": 1.1455, "step": 22436 }, { "epoch": 2.387422855926793, "grad_norm": 15.2109375, "learning_rate": 9.22054434337692e-06, "loss": 0.945, "step": 22437 }, { "epoch": 2.3875292615450094, "grad_norm": 17.359375, "learning_rate": 9.2205088118249e-06, "loss": 0.8992, "step": 22438 }, { "epoch": 2.3876356671632264, "grad_norm": 12.8046875, "learning_rate": 9.220473280272884e-06, "loss": 0.8328, "step": 22439 }, { "epoch": 2.387742072781443, "grad_norm": 12.6953125, "learning_rate": 9.220437748720866e-06, "loss": 0.9066, "step": 22440 }, { "epoch": 2.3878484783996594, "grad_norm": 15.28125, "learning_rate": 9.220402217168847e-06, "loss": 0.8752, "step": 22441 }, { "epoch": 2.3879548840178764, "grad_norm": 14.9453125, "learning_rate": 9.220366685616829e-06, "loss": 0.8397, "step": 22442 }, { "epoch": 2.388061289636093, "grad_norm": 15.03125, "learning_rate": 9.22033115406481e-06, "loss": 0.9676, "step": 22443 }, { "epoch": 2.3881676952543094, "grad_norm": 15.5234375, "learning_rate": 9.220295622512792e-06, "loss": 1.0173, "step": 22444 }, { "epoch": 2.388274100872526, "grad_norm": 14.21875, "learning_rate": 9.220260090960774e-06, "loss": 0.8884, "step": 22445 }, { "epoch": 2.388380506490743, "grad_norm": 11.6484375, "learning_rate": 9.220224559408755e-06, "loss": 0.7606, "step": 22446 }, { "epoch": 2.3884869121089594, "grad_norm": 20.921875, "learning_rate": 9.220189027856737e-06, "loss": 0.9078, "step": 22447 }, { "epoch": 2.388593317727176, "grad_norm": 11.203125, "learning_rate": 9.220153496304719e-06, "loss": 1.0148, "step": 22448 }, { "epoch": 2.388699723345393, "grad_norm": 10.140625, "learning_rate": 9.2201179647527e-06, "loss": 0.8977, "step": 22449 }, { "epoch": 2.3888061289636093, "grad_norm": 11.65625, "learning_rate": 9.220082433200683e-06, "loss": 0.7997, "step": 22450 }, { "epoch": 2.388912534581826, "grad_norm": 14.203125, "learning_rate": 9.220046901648665e-06, "loss": 1.1709, "step": 22451 }, { "epoch": 2.3890189402000424, "grad_norm": 12.2109375, "learning_rate": 9.220011370096647e-06, "loss": 0.7747, "step": 22452 }, { "epoch": 2.3891253458182593, "grad_norm": 13.890625, "learning_rate": 9.219975838544628e-06, "loss": 1.1172, "step": 22453 }, { "epoch": 2.389231751436476, "grad_norm": 13.0859375, "learning_rate": 9.21994030699261e-06, "loss": 0.994, "step": 22454 }, { "epoch": 2.3893381570546923, "grad_norm": 26.921875, "learning_rate": 9.219904775440593e-06, "loss": 1.1901, "step": 22455 }, { "epoch": 2.3894445626729093, "grad_norm": 11.203125, "learning_rate": 9.219869243888575e-06, "loss": 0.8272, "step": 22456 }, { "epoch": 2.389550968291126, "grad_norm": 11.546875, "learning_rate": 9.219833712336556e-06, "loss": 0.8853, "step": 22457 }, { "epoch": 2.3896573739093423, "grad_norm": 12.1484375, "learning_rate": 9.219798180784538e-06, "loss": 0.972, "step": 22458 }, { "epoch": 2.3897637795275593, "grad_norm": 12.5078125, "learning_rate": 9.219762649232518e-06, "loss": 0.838, "step": 22459 }, { "epoch": 2.3898701851457758, "grad_norm": 15.421875, "learning_rate": 9.219727117680501e-06, "loss": 0.9345, "step": 22460 }, { "epoch": 2.3899765907639923, "grad_norm": 15.3203125, "learning_rate": 9.219691586128483e-06, "loss": 0.8847, "step": 22461 }, { "epoch": 2.3900829963822092, "grad_norm": 12.796875, "learning_rate": 9.219656054576464e-06, "loss": 0.6219, "step": 22462 }, { "epoch": 2.3901894020004257, "grad_norm": 9.515625, "learning_rate": 9.219620523024446e-06, "loss": 0.7915, "step": 22463 }, { "epoch": 2.3902958076186422, "grad_norm": 18.90625, "learning_rate": 9.219584991472428e-06, "loss": 1.4185, "step": 22464 }, { "epoch": 2.3904022132368588, "grad_norm": 14.8984375, "learning_rate": 9.219549459920411e-06, "loss": 0.9467, "step": 22465 }, { "epoch": 2.3905086188550757, "grad_norm": 10.6953125, "learning_rate": 9.219513928368393e-06, "loss": 0.7801, "step": 22466 }, { "epoch": 2.390615024473292, "grad_norm": 14.03125, "learning_rate": 9.219478396816374e-06, "loss": 0.8099, "step": 22467 }, { "epoch": 2.3907214300915087, "grad_norm": 16.125, "learning_rate": 9.219442865264356e-06, "loss": 1.1166, "step": 22468 }, { "epoch": 2.3908278357097252, "grad_norm": 15.2890625, "learning_rate": 9.219407333712337e-06, "loss": 0.7636, "step": 22469 }, { "epoch": 2.390934241327942, "grad_norm": 12.9375, "learning_rate": 9.219371802160319e-06, "loss": 0.8781, "step": 22470 }, { "epoch": 2.3910406469461587, "grad_norm": 13.9296875, "learning_rate": 9.2193362706083e-06, "loss": 1.0504, "step": 22471 }, { "epoch": 2.391147052564375, "grad_norm": 13.453125, "learning_rate": 9.219300739056282e-06, "loss": 1.1445, "step": 22472 }, { "epoch": 2.391253458182592, "grad_norm": 14.7265625, "learning_rate": 9.219265207504264e-06, "loss": 0.8737, "step": 22473 }, { "epoch": 2.3913598638008087, "grad_norm": 16.953125, "learning_rate": 9.219229675952245e-06, "loss": 0.9122, "step": 22474 }, { "epoch": 2.391466269419025, "grad_norm": 33.90625, "learning_rate": 9.219194144400229e-06, "loss": 0.9683, "step": 22475 }, { "epoch": 2.391572675037242, "grad_norm": 9.0703125, "learning_rate": 9.21915861284821e-06, "loss": 0.8103, "step": 22476 }, { "epoch": 2.3916790806554586, "grad_norm": 14.09375, "learning_rate": 9.219123081296192e-06, "loss": 1.0529, "step": 22477 }, { "epoch": 2.391785486273675, "grad_norm": 13.65625, "learning_rate": 9.219087549744174e-06, "loss": 0.9488, "step": 22478 }, { "epoch": 2.391891891891892, "grad_norm": 12.6875, "learning_rate": 9.219052018192155e-06, "loss": 0.8799, "step": 22479 }, { "epoch": 2.3919982975101086, "grad_norm": 13.8515625, "learning_rate": 9.219016486640138e-06, "loss": 0.8662, "step": 22480 }, { "epoch": 2.392104703128325, "grad_norm": 11.875, "learning_rate": 9.21898095508812e-06, "loss": 0.96, "step": 22481 }, { "epoch": 2.3922111087465416, "grad_norm": 13.5390625, "learning_rate": 9.218945423536102e-06, "loss": 0.8791, "step": 22482 }, { "epoch": 2.3923175143647586, "grad_norm": 10.7421875, "learning_rate": 9.218909891984083e-06, "loss": 0.9091, "step": 22483 }, { "epoch": 2.392423919982975, "grad_norm": 11.8984375, "learning_rate": 9.218874360432065e-06, "loss": 0.8371, "step": 22484 }, { "epoch": 2.3925303256011916, "grad_norm": 20.359375, "learning_rate": 9.218838828880046e-06, "loss": 0.686, "step": 22485 }, { "epoch": 2.3926367312194086, "grad_norm": 17.609375, "learning_rate": 9.218803297328028e-06, "loss": 0.8795, "step": 22486 }, { "epoch": 2.392743136837625, "grad_norm": 13.953125, "learning_rate": 9.21876776577601e-06, "loss": 0.8386, "step": 22487 }, { "epoch": 2.3928495424558416, "grad_norm": 12.6953125, "learning_rate": 9.218732234223991e-06, "loss": 0.8942, "step": 22488 }, { "epoch": 2.392955948074058, "grad_norm": 11.9453125, "learning_rate": 9.218696702671973e-06, "loss": 0.8755, "step": 22489 }, { "epoch": 2.393062353692275, "grad_norm": 12.15625, "learning_rate": 9.218661171119954e-06, "loss": 0.8553, "step": 22490 }, { "epoch": 2.3931687593104916, "grad_norm": 14.0546875, "learning_rate": 9.218625639567938e-06, "loss": 0.9525, "step": 22491 }, { "epoch": 2.393275164928708, "grad_norm": 11.265625, "learning_rate": 9.21859010801592e-06, "loss": 0.6959, "step": 22492 }, { "epoch": 2.393381570546925, "grad_norm": 12.15625, "learning_rate": 9.218554576463901e-06, "loss": 0.9894, "step": 22493 }, { "epoch": 2.3934879761651415, "grad_norm": 12.8515625, "learning_rate": 9.218519044911883e-06, "loss": 1.0527, "step": 22494 }, { "epoch": 2.393594381783358, "grad_norm": 10.7421875, "learning_rate": 9.218483513359864e-06, "loss": 0.802, "step": 22495 }, { "epoch": 2.393700787401575, "grad_norm": 11.4453125, "learning_rate": 9.218447981807846e-06, "loss": 0.9317, "step": 22496 }, { "epoch": 2.3938071930197915, "grad_norm": 12.8359375, "learning_rate": 9.218412450255827e-06, "loss": 0.7267, "step": 22497 }, { "epoch": 2.393913598638008, "grad_norm": 13.359375, "learning_rate": 9.218376918703809e-06, "loss": 1.0716, "step": 22498 }, { "epoch": 2.394020004256225, "grad_norm": 13.5625, "learning_rate": 9.21834138715179e-06, "loss": 0.8196, "step": 22499 }, { "epoch": 2.3941264098744415, "grad_norm": 11.2109375, "learning_rate": 9.218305855599772e-06, "loss": 0.8268, "step": 22500 }, { "epoch": 2.394232815492658, "grad_norm": 15.21875, "learning_rate": 9.218270324047756e-06, "loss": 0.8878, "step": 22501 }, { "epoch": 2.3943392211108745, "grad_norm": 14.34375, "learning_rate": 9.218234792495737e-06, "loss": 0.9632, "step": 22502 }, { "epoch": 2.3944456267290914, "grad_norm": 11.5234375, "learning_rate": 9.218199260943719e-06, "loss": 0.8154, "step": 22503 }, { "epoch": 2.394552032347308, "grad_norm": 19.40625, "learning_rate": 9.2181637293917e-06, "loss": 1.2887, "step": 22504 }, { "epoch": 2.3946584379655245, "grad_norm": 13.75, "learning_rate": 9.218128197839682e-06, "loss": 0.9257, "step": 22505 }, { "epoch": 2.394764843583741, "grad_norm": 10.703125, "learning_rate": 9.218092666287665e-06, "loss": 0.9486, "step": 22506 }, { "epoch": 2.394871249201958, "grad_norm": 14.6484375, "learning_rate": 9.218057134735647e-06, "loss": 1.003, "step": 22507 }, { "epoch": 2.3949776548201744, "grad_norm": 8.9765625, "learning_rate": 9.218021603183628e-06, "loss": 0.8559, "step": 22508 }, { "epoch": 2.395084060438391, "grad_norm": 13.328125, "learning_rate": 9.21798607163161e-06, "loss": 0.8976, "step": 22509 }, { "epoch": 2.395190466056608, "grad_norm": 13.7109375, "learning_rate": 9.217950540079592e-06, "loss": 1.0206, "step": 22510 }, { "epoch": 2.3952968716748244, "grad_norm": 15.0, "learning_rate": 9.217915008527573e-06, "loss": 0.9815, "step": 22511 }, { "epoch": 2.395403277293041, "grad_norm": 11.7578125, "learning_rate": 9.217879476975555e-06, "loss": 0.9558, "step": 22512 }, { "epoch": 2.395509682911258, "grad_norm": 12.859375, "learning_rate": 9.217843945423537e-06, "loss": 1.0175, "step": 22513 }, { "epoch": 2.3956160885294744, "grad_norm": 11.203125, "learning_rate": 9.217808413871518e-06, "loss": 0.7444, "step": 22514 }, { "epoch": 2.395722494147691, "grad_norm": 11.9921875, "learning_rate": 9.2177728823195e-06, "loss": 0.8116, "step": 22515 }, { "epoch": 2.395828899765908, "grad_norm": 13.8125, "learning_rate": 9.217737350767483e-06, "loss": 0.9375, "step": 22516 }, { "epoch": 2.3959353053841244, "grad_norm": 12.4140625, "learning_rate": 9.217701819215465e-06, "loss": 0.7939, "step": 22517 }, { "epoch": 2.396041711002341, "grad_norm": 11.0078125, "learning_rate": 9.217666287663446e-06, "loss": 0.89, "step": 22518 }, { "epoch": 2.3961481166205574, "grad_norm": 21.15625, "learning_rate": 9.217630756111428e-06, "loss": 0.9412, "step": 22519 }, { "epoch": 2.3962545222387743, "grad_norm": 11.8828125, "learning_rate": 9.21759522455941e-06, "loss": 0.8775, "step": 22520 }, { "epoch": 2.396360927856991, "grad_norm": 12.234375, "learning_rate": 9.217559693007391e-06, "loss": 0.8237, "step": 22521 }, { "epoch": 2.3964673334752074, "grad_norm": 16.453125, "learning_rate": 9.217524161455373e-06, "loss": 0.9918, "step": 22522 }, { "epoch": 2.3965737390934243, "grad_norm": 15.34375, "learning_rate": 9.217488629903354e-06, "loss": 0.9388, "step": 22523 }, { "epoch": 2.396680144711641, "grad_norm": 13.40625, "learning_rate": 9.217453098351336e-06, "loss": 0.9128, "step": 22524 }, { "epoch": 2.3967865503298573, "grad_norm": 16.203125, "learning_rate": 9.217417566799317e-06, "loss": 1.1291, "step": 22525 }, { "epoch": 2.396892955948074, "grad_norm": 14.6953125, "learning_rate": 9.2173820352473e-06, "loss": 1.0204, "step": 22526 }, { "epoch": 2.396999361566291, "grad_norm": 13.2265625, "learning_rate": 9.217346503695282e-06, "loss": 0.9507, "step": 22527 }, { "epoch": 2.3971057671845073, "grad_norm": 12.3125, "learning_rate": 9.217310972143264e-06, "loss": 0.8395, "step": 22528 }, { "epoch": 2.397212172802724, "grad_norm": 12.7890625, "learning_rate": 9.217275440591246e-06, "loss": 0.8915, "step": 22529 }, { "epoch": 2.3973185784209408, "grad_norm": 10.8984375, "learning_rate": 9.217239909039227e-06, "loss": 0.8352, "step": 22530 }, { "epoch": 2.3974249840391573, "grad_norm": 10.984375, "learning_rate": 9.217204377487209e-06, "loss": 0.9137, "step": 22531 }, { "epoch": 2.397531389657374, "grad_norm": 20.328125, "learning_rate": 9.217168845935192e-06, "loss": 1.098, "step": 22532 }, { "epoch": 2.3976377952755907, "grad_norm": 18.578125, "learning_rate": 9.217133314383174e-06, "loss": 0.961, "step": 22533 }, { "epoch": 2.3977442008938072, "grad_norm": 11.8203125, "learning_rate": 9.217097782831155e-06, "loss": 0.733, "step": 22534 }, { "epoch": 2.3978506065120238, "grad_norm": 10.515625, "learning_rate": 9.217062251279137e-06, "loss": 0.6959, "step": 22535 }, { "epoch": 2.3979570121302407, "grad_norm": 10.921875, "learning_rate": 9.217026719727119e-06, "loss": 0.8108, "step": 22536 }, { "epoch": 2.398063417748457, "grad_norm": 10.9921875, "learning_rate": 9.2169911881751e-06, "loss": 1.018, "step": 22537 }, { "epoch": 2.3981698233666737, "grad_norm": 14.828125, "learning_rate": 9.216955656623082e-06, "loss": 0.898, "step": 22538 }, { "epoch": 2.3982762289848902, "grad_norm": 11.4765625, "learning_rate": 9.216920125071063e-06, "loss": 0.851, "step": 22539 }, { "epoch": 2.398382634603107, "grad_norm": 22.328125, "learning_rate": 9.216884593519045e-06, "loss": 1.0936, "step": 22540 }, { "epoch": 2.3984890402213237, "grad_norm": 14.1328125, "learning_rate": 9.216849061967027e-06, "loss": 0.8801, "step": 22541 }, { "epoch": 2.39859544583954, "grad_norm": 17.09375, "learning_rate": 9.21681353041501e-06, "loss": 1.0479, "step": 22542 }, { "epoch": 2.398701851457757, "grad_norm": 16.0, "learning_rate": 9.216777998862991e-06, "loss": 0.8668, "step": 22543 }, { "epoch": 2.3988082570759737, "grad_norm": 14.578125, "learning_rate": 9.216742467310973e-06, "loss": 0.9566, "step": 22544 }, { "epoch": 2.39891466269419, "grad_norm": 10.8828125, "learning_rate": 9.216706935758955e-06, "loss": 0.8107, "step": 22545 }, { "epoch": 2.3990210683124067, "grad_norm": 13.9453125, "learning_rate": 9.216671404206936e-06, "loss": 1.0057, "step": 22546 }, { "epoch": 2.3991274739306236, "grad_norm": 18.0625, "learning_rate": 9.21663587265492e-06, "loss": 0.9654, "step": 22547 }, { "epoch": 2.39923387954884, "grad_norm": 12.53125, "learning_rate": 9.2166003411029e-06, "loss": 1.0544, "step": 22548 }, { "epoch": 2.3993402851670567, "grad_norm": 15.6796875, "learning_rate": 9.216564809550881e-06, "loss": 1.143, "step": 22549 }, { "epoch": 2.3994466907852736, "grad_norm": 11.046875, "learning_rate": 9.216529277998863e-06, "loss": 0.8315, "step": 22550 }, { "epoch": 2.39955309640349, "grad_norm": 14.84375, "learning_rate": 9.216493746446844e-06, "loss": 0.8596, "step": 22551 }, { "epoch": 2.3996595020217066, "grad_norm": 14.0546875, "learning_rate": 9.216458214894828e-06, "loss": 0.8844, "step": 22552 }, { "epoch": 2.3997659076399236, "grad_norm": 10.9453125, "learning_rate": 9.21642268334281e-06, "loss": 0.7249, "step": 22553 }, { "epoch": 2.39987231325814, "grad_norm": 12.5078125, "learning_rate": 9.216387151790791e-06, "loss": 0.8998, "step": 22554 }, { "epoch": 2.3999787188763566, "grad_norm": 13.46875, "learning_rate": 9.216351620238772e-06, "loss": 0.9593, "step": 22555 }, { "epoch": 2.4000851244945736, "grad_norm": 13.4609375, "learning_rate": 9.216316088686754e-06, "loss": 1.1145, "step": 22556 }, { "epoch": 2.40019153011279, "grad_norm": 10.515625, "learning_rate": 9.216280557134737e-06, "loss": 0.8672, "step": 22557 }, { "epoch": 2.4002979357310066, "grad_norm": 10.6171875, "learning_rate": 9.216245025582719e-06, "loss": 0.818, "step": 22558 }, { "epoch": 2.400404341349223, "grad_norm": 12.28125, "learning_rate": 9.2162094940307e-06, "loss": 0.9364, "step": 22559 }, { "epoch": 2.40051074696744, "grad_norm": 16.78125, "learning_rate": 9.216173962478682e-06, "loss": 1.0261, "step": 22560 }, { "epoch": 2.4006171525856566, "grad_norm": 13.390625, "learning_rate": 9.216138430926664e-06, "loss": 0.9124, "step": 22561 }, { "epoch": 2.400723558203873, "grad_norm": 11.2109375, "learning_rate": 9.216102899374645e-06, "loss": 0.7142, "step": 22562 }, { "epoch": 2.4008299638220896, "grad_norm": 11.296875, "learning_rate": 9.216067367822627e-06, "loss": 0.8085, "step": 22563 }, { "epoch": 2.4009363694403065, "grad_norm": 14.4921875, "learning_rate": 9.216031836270609e-06, "loss": 0.7932, "step": 22564 }, { "epoch": 2.401042775058523, "grad_norm": 14.1328125, "learning_rate": 9.21599630471859e-06, "loss": 0.9156, "step": 22565 }, { "epoch": 2.4011491806767395, "grad_norm": 17.09375, "learning_rate": 9.215960773166572e-06, "loss": 1.1802, "step": 22566 }, { "epoch": 2.4012555862949565, "grad_norm": 10.078125, "learning_rate": 9.215925241614553e-06, "loss": 0.6667, "step": 22567 }, { "epoch": 2.401361991913173, "grad_norm": 13.7578125, "learning_rate": 9.215889710062537e-06, "loss": 0.8832, "step": 22568 }, { "epoch": 2.4014683975313895, "grad_norm": 12.5078125, "learning_rate": 9.215854178510518e-06, "loss": 0.8181, "step": 22569 }, { "epoch": 2.4015748031496065, "grad_norm": 11.15625, "learning_rate": 9.2158186469585e-06, "loss": 0.8103, "step": 22570 }, { "epoch": 2.401681208767823, "grad_norm": 11.359375, "learning_rate": 9.215783115406482e-06, "loss": 0.8484, "step": 22571 }, { "epoch": 2.4017876143860395, "grad_norm": 12.921875, "learning_rate": 9.215747583854463e-06, "loss": 0.7641, "step": 22572 }, { "epoch": 2.4018940200042564, "grad_norm": 16.296875, "learning_rate": 9.215712052302446e-06, "loss": 0.9659, "step": 22573 }, { "epoch": 2.402000425622473, "grad_norm": 16.671875, "learning_rate": 9.215676520750428e-06, "loss": 0.9047, "step": 22574 }, { "epoch": 2.4021068312406895, "grad_norm": 11.671875, "learning_rate": 9.21564098919841e-06, "loss": 0.8795, "step": 22575 }, { "epoch": 2.402213236858906, "grad_norm": 19.25, "learning_rate": 9.215605457646391e-06, "loss": 0.9619, "step": 22576 }, { "epoch": 2.402319642477123, "grad_norm": 11.3984375, "learning_rate": 9.215569926094373e-06, "loss": 1.0355, "step": 22577 }, { "epoch": 2.4024260480953394, "grad_norm": 13.1875, "learning_rate": 9.215534394542355e-06, "loss": 0.8604, "step": 22578 }, { "epoch": 2.402532453713556, "grad_norm": 16.375, "learning_rate": 9.215498862990336e-06, "loss": 0.922, "step": 22579 }, { "epoch": 2.402638859331773, "grad_norm": 11.765625, "learning_rate": 9.215463331438318e-06, "loss": 0.712, "step": 22580 }, { "epoch": 2.4027452649499894, "grad_norm": 13.859375, "learning_rate": 9.2154277998863e-06, "loss": 0.8583, "step": 22581 }, { "epoch": 2.402851670568206, "grad_norm": 11.890625, "learning_rate": 9.215392268334281e-06, "loss": 0.7873, "step": 22582 }, { "epoch": 2.4029580761864224, "grad_norm": 17.28125, "learning_rate": 9.215356736782264e-06, "loss": 0.909, "step": 22583 }, { "epoch": 2.4030644818046394, "grad_norm": 14.0546875, "learning_rate": 9.215321205230246e-06, "loss": 0.8302, "step": 22584 }, { "epoch": 2.403170887422856, "grad_norm": 12.7578125, "learning_rate": 9.215285673678227e-06, "loss": 0.9535, "step": 22585 }, { "epoch": 2.4032772930410724, "grad_norm": 15.40625, "learning_rate": 9.215250142126209e-06, "loss": 1.0729, "step": 22586 }, { "epoch": 2.4033836986592894, "grad_norm": 15.8359375, "learning_rate": 9.21521461057419e-06, "loss": 0.9343, "step": 22587 }, { "epoch": 2.403490104277506, "grad_norm": 16.578125, "learning_rate": 9.215179079022172e-06, "loss": 1.0344, "step": 22588 }, { "epoch": 2.4035965098957224, "grad_norm": 11.1171875, "learning_rate": 9.215143547470154e-06, "loss": 0.898, "step": 22589 }, { "epoch": 2.4037029155139393, "grad_norm": 15.4921875, "learning_rate": 9.215108015918135e-06, "loss": 1.09, "step": 22590 }, { "epoch": 2.403809321132156, "grad_norm": 22.234375, "learning_rate": 9.215072484366117e-06, "loss": 0.8545, "step": 22591 }, { "epoch": 2.4039157267503724, "grad_norm": 13.8515625, "learning_rate": 9.215036952814099e-06, "loss": 0.8207, "step": 22592 }, { "epoch": 2.4040221323685893, "grad_norm": 15.4140625, "learning_rate": 9.215001421262082e-06, "loss": 0.8765, "step": 22593 }, { "epoch": 2.404128537986806, "grad_norm": 13.390625, "learning_rate": 9.214965889710064e-06, "loss": 0.846, "step": 22594 }, { "epoch": 2.4042349436050223, "grad_norm": 12.1640625, "learning_rate": 9.214930358158045e-06, "loss": 1.0521, "step": 22595 }, { "epoch": 2.404341349223239, "grad_norm": 15.4609375, "learning_rate": 9.214894826606027e-06, "loss": 1.0459, "step": 22596 }, { "epoch": 2.404447754841456, "grad_norm": 15.09375, "learning_rate": 9.214859295054008e-06, "loss": 0.8808, "step": 22597 }, { "epoch": 2.4045541604596723, "grad_norm": 15.96875, "learning_rate": 9.214823763501992e-06, "loss": 0.9492, "step": 22598 }, { "epoch": 2.404660566077889, "grad_norm": 13.453125, "learning_rate": 9.214788231949973e-06, "loss": 0.8472, "step": 22599 }, { "epoch": 2.4047669716961053, "grad_norm": 10.875, "learning_rate": 9.214752700397955e-06, "loss": 0.9036, "step": 22600 }, { "epoch": 2.4048733773143223, "grad_norm": 14.125, "learning_rate": 9.214717168845937e-06, "loss": 0.9136, "step": 22601 }, { "epoch": 2.404979782932539, "grad_norm": 15.75, "learning_rate": 9.214681637293918e-06, "loss": 1.1586, "step": 22602 }, { "epoch": 2.4050861885507553, "grad_norm": 12.96875, "learning_rate": 9.2146461057419e-06, "loss": 1.0108, "step": 22603 }, { "epoch": 2.4051925941689722, "grad_norm": 14.7421875, "learning_rate": 9.214610574189881e-06, "loss": 1.0671, "step": 22604 }, { "epoch": 2.4052989997871888, "grad_norm": 13.6171875, "learning_rate": 9.214575042637863e-06, "loss": 0.9128, "step": 22605 }, { "epoch": 2.4054054054054053, "grad_norm": 14.0703125, "learning_rate": 9.214539511085845e-06, "loss": 0.9246, "step": 22606 }, { "epoch": 2.405511811023622, "grad_norm": 13.75, "learning_rate": 9.214503979533826e-06, "loss": 0.9211, "step": 22607 }, { "epoch": 2.4056182166418387, "grad_norm": 20.0, "learning_rate": 9.214468447981808e-06, "loss": 0.7878, "step": 22608 }, { "epoch": 2.4057246222600552, "grad_norm": 17.234375, "learning_rate": 9.214432916429791e-06, "loss": 0.8899, "step": 22609 }, { "epoch": 2.405831027878272, "grad_norm": 10.203125, "learning_rate": 9.214397384877773e-06, "loss": 0.7607, "step": 22610 }, { "epoch": 2.4059374334964887, "grad_norm": 14.0390625, "learning_rate": 9.214361853325754e-06, "loss": 1.004, "step": 22611 }, { "epoch": 2.406043839114705, "grad_norm": 12.2421875, "learning_rate": 9.214326321773736e-06, "loss": 0.8407, "step": 22612 }, { "epoch": 2.4061502447329217, "grad_norm": 12.109375, "learning_rate": 9.214290790221718e-06, "loss": 0.9063, "step": 22613 }, { "epoch": 2.4062566503511387, "grad_norm": 15.3125, "learning_rate": 9.214255258669699e-06, "loss": 0.9464, "step": 22614 }, { "epoch": 2.406363055969355, "grad_norm": 15.3125, "learning_rate": 9.21421972711768e-06, "loss": 0.7783, "step": 22615 }, { "epoch": 2.4064694615875717, "grad_norm": 14.1484375, "learning_rate": 9.214184195565662e-06, "loss": 0.9698, "step": 22616 }, { "epoch": 2.4065758672057886, "grad_norm": 10.046875, "learning_rate": 9.214148664013644e-06, "loss": 0.889, "step": 22617 }, { "epoch": 2.406682272824005, "grad_norm": 20.65625, "learning_rate": 9.214113132461626e-06, "loss": 0.9573, "step": 22618 }, { "epoch": 2.4067886784422217, "grad_norm": 9.1953125, "learning_rate": 9.214077600909609e-06, "loss": 0.5533, "step": 22619 }, { "epoch": 2.406895084060438, "grad_norm": 13.5546875, "learning_rate": 9.21404206935759e-06, "loss": 1.0082, "step": 22620 }, { "epoch": 2.407001489678655, "grad_norm": 20.890625, "learning_rate": 9.214006537805572e-06, "loss": 1.119, "step": 22621 }, { "epoch": 2.4071078952968716, "grad_norm": 13.859375, "learning_rate": 9.213971006253554e-06, "loss": 1.119, "step": 22622 }, { "epoch": 2.407214300915088, "grad_norm": 10.640625, "learning_rate": 9.213935474701535e-06, "loss": 0.6667, "step": 22623 }, { "epoch": 2.407320706533305, "grad_norm": 15.9140625, "learning_rate": 9.213899943149519e-06, "loss": 0.9633, "step": 22624 }, { "epoch": 2.4074271121515216, "grad_norm": 12.5703125, "learning_rate": 9.2138644115975e-06, "loss": 0.8675, "step": 22625 }, { "epoch": 2.407533517769738, "grad_norm": 14.109375, "learning_rate": 9.213828880045482e-06, "loss": 0.8705, "step": 22626 }, { "epoch": 2.407639923387955, "grad_norm": 12.9921875, "learning_rate": 9.213793348493463e-06, "loss": 0.833, "step": 22627 }, { "epoch": 2.4077463290061716, "grad_norm": 10.078125, "learning_rate": 9.213757816941445e-06, "loss": 0.7928, "step": 22628 }, { "epoch": 2.407852734624388, "grad_norm": 16.890625, "learning_rate": 9.213722285389427e-06, "loss": 0.9162, "step": 22629 }, { "epoch": 2.407959140242605, "grad_norm": 13.65625, "learning_rate": 9.213686753837408e-06, "loss": 0.842, "step": 22630 }, { "epoch": 2.4080655458608216, "grad_norm": 11.9609375, "learning_rate": 9.21365122228539e-06, "loss": 0.8656, "step": 22631 }, { "epoch": 2.408171951479038, "grad_norm": 14.4765625, "learning_rate": 9.213615690733371e-06, "loss": 0.9251, "step": 22632 }, { "epoch": 2.4082783570972546, "grad_norm": 17.1875, "learning_rate": 9.213580159181353e-06, "loss": 0.8693, "step": 22633 }, { "epoch": 2.4083847627154715, "grad_norm": 13.2265625, "learning_rate": 9.213544627629336e-06, "loss": 0.8078, "step": 22634 }, { "epoch": 2.408491168333688, "grad_norm": 11.03125, "learning_rate": 9.213509096077318e-06, "loss": 0.8854, "step": 22635 }, { "epoch": 2.4085975739519045, "grad_norm": 12.71875, "learning_rate": 9.2134735645253e-06, "loss": 0.9791, "step": 22636 }, { "epoch": 2.408703979570121, "grad_norm": 12.3125, "learning_rate": 9.213438032973281e-06, "loss": 0.9943, "step": 22637 }, { "epoch": 2.408810385188338, "grad_norm": 11.84375, "learning_rate": 9.213402501421263e-06, "loss": 0.743, "step": 22638 }, { "epoch": 2.4089167908065545, "grad_norm": 12.15625, "learning_rate": 9.213366969869244e-06, "loss": 0.9347, "step": 22639 }, { "epoch": 2.409023196424771, "grad_norm": 12.5546875, "learning_rate": 9.213331438317226e-06, "loss": 0.8789, "step": 22640 }, { "epoch": 2.409129602042988, "grad_norm": 18.609375, "learning_rate": 9.213295906765208e-06, "loss": 0.8947, "step": 22641 }, { "epoch": 2.4092360076612045, "grad_norm": 16.59375, "learning_rate": 9.21326037521319e-06, "loss": 0.9626, "step": 22642 }, { "epoch": 2.409342413279421, "grad_norm": 18.78125, "learning_rate": 9.21322484366117e-06, "loss": 0.9998, "step": 22643 }, { "epoch": 2.409448818897638, "grad_norm": 12.546875, "learning_rate": 9.213189312109152e-06, "loss": 1.0537, "step": 22644 }, { "epoch": 2.4095552245158545, "grad_norm": 9.9921875, "learning_rate": 9.213153780557136e-06, "loss": 0.8063, "step": 22645 }, { "epoch": 2.409661630134071, "grad_norm": 11.9609375, "learning_rate": 9.213118249005117e-06, "loss": 1.0541, "step": 22646 }, { "epoch": 2.409768035752288, "grad_norm": 11.734375, "learning_rate": 9.213082717453099e-06, "loss": 0.7897, "step": 22647 }, { "epoch": 2.4098744413705044, "grad_norm": 12.53125, "learning_rate": 9.21304718590108e-06, "loss": 0.827, "step": 22648 }, { "epoch": 2.409980846988721, "grad_norm": 20.875, "learning_rate": 9.213011654349062e-06, "loss": 0.9257, "step": 22649 }, { "epoch": 2.4100872526069375, "grad_norm": 12.53125, "learning_rate": 9.212976122797045e-06, "loss": 0.8442, "step": 22650 }, { "epoch": 2.4101936582251544, "grad_norm": 14.515625, "learning_rate": 9.212940591245027e-06, "loss": 1.0338, "step": 22651 }, { "epoch": 2.410300063843371, "grad_norm": 12.4765625, "learning_rate": 9.212905059693009e-06, "loss": 0.8532, "step": 22652 }, { "epoch": 2.4104064694615874, "grad_norm": 12.5390625, "learning_rate": 9.21286952814099e-06, "loss": 0.7711, "step": 22653 }, { "epoch": 2.4105128750798044, "grad_norm": 12.796875, "learning_rate": 9.212833996588972e-06, "loss": 0.8457, "step": 22654 }, { "epoch": 2.410619280698021, "grad_norm": 12.5234375, "learning_rate": 9.212798465036953e-06, "loss": 1.0326, "step": 22655 }, { "epoch": 2.4107256863162374, "grad_norm": 12.5859375, "learning_rate": 9.212762933484935e-06, "loss": 0.7924, "step": 22656 }, { "epoch": 2.410832091934454, "grad_norm": 12.5234375, "learning_rate": 9.212727401932917e-06, "loss": 0.7986, "step": 22657 }, { "epoch": 2.410938497552671, "grad_norm": 9.71875, "learning_rate": 9.212691870380898e-06, "loss": 0.8878, "step": 22658 }, { "epoch": 2.4110449031708874, "grad_norm": 14.1640625, "learning_rate": 9.21265633882888e-06, "loss": 0.7748, "step": 22659 }, { "epoch": 2.411151308789104, "grad_norm": 12.4609375, "learning_rate": 9.212620807276863e-06, "loss": 0.8401, "step": 22660 }, { "epoch": 2.411257714407321, "grad_norm": 9.828125, "learning_rate": 9.212585275724845e-06, "loss": 0.9935, "step": 22661 }, { "epoch": 2.4113641200255373, "grad_norm": 13.2578125, "learning_rate": 9.212549744172826e-06, "loss": 0.8585, "step": 22662 }, { "epoch": 2.411470525643754, "grad_norm": 13.3125, "learning_rate": 9.212514212620808e-06, "loss": 0.8027, "step": 22663 }, { "epoch": 2.411576931261971, "grad_norm": 16.265625, "learning_rate": 9.21247868106879e-06, "loss": 0.8521, "step": 22664 }, { "epoch": 2.4116833368801873, "grad_norm": 10.8046875, "learning_rate": 9.212443149516773e-06, "loss": 0.7924, "step": 22665 }, { "epoch": 2.411789742498404, "grad_norm": 12.84375, "learning_rate": 9.212407617964755e-06, "loss": 0.9037, "step": 22666 }, { "epoch": 2.411896148116621, "grad_norm": 15.125, "learning_rate": 9.212372086412736e-06, "loss": 0.8181, "step": 22667 }, { "epoch": 2.4120025537348373, "grad_norm": 11.4921875, "learning_rate": 9.212336554860716e-06, "loss": 0.8593, "step": 22668 }, { "epoch": 2.412108959353054, "grad_norm": 12.9375, "learning_rate": 9.212301023308698e-06, "loss": 0.869, "step": 22669 }, { "epoch": 2.4122153649712703, "grad_norm": 15.1171875, "learning_rate": 9.212265491756681e-06, "loss": 0.7887, "step": 22670 }, { "epoch": 2.4123217705894873, "grad_norm": 12.390625, "learning_rate": 9.212229960204663e-06, "loss": 0.899, "step": 22671 }, { "epoch": 2.412428176207704, "grad_norm": 14.5703125, "learning_rate": 9.212194428652644e-06, "loss": 1.0093, "step": 22672 }, { "epoch": 2.4125345818259203, "grad_norm": 13.7421875, "learning_rate": 9.212158897100626e-06, "loss": 0.9825, "step": 22673 }, { "epoch": 2.4126409874441372, "grad_norm": 11.265625, "learning_rate": 9.212123365548607e-06, "loss": 1.0371, "step": 22674 }, { "epoch": 2.4127473930623538, "grad_norm": 10.7421875, "learning_rate": 9.21208783399659e-06, "loss": 0.8634, "step": 22675 }, { "epoch": 2.4128537986805703, "grad_norm": 13.625, "learning_rate": 9.212052302444572e-06, "loss": 1.0708, "step": 22676 }, { "epoch": 2.4129602042987868, "grad_norm": 15.2109375, "learning_rate": 9.212016770892554e-06, "loss": 0.992, "step": 22677 }, { "epoch": 2.4130666099170037, "grad_norm": 18.109375, "learning_rate": 9.211981239340535e-06, "loss": 0.97, "step": 22678 }, { "epoch": 2.4131730155352202, "grad_norm": 12.2109375, "learning_rate": 9.211945707788517e-06, "loss": 0.841, "step": 22679 }, { "epoch": 2.4132794211534367, "grad_norm": 18.90625, "learning_rate": 9.211910176236499e-06, "loss": 0.7381, "step": 22680 }, { "epoch": 2.4133858267716537, "grad_norm": 11.265625, "learning_rate": 9.21187464468448e-06, "loss": 0.8311, "step": 22681 }, { "epoch": 2.41349223238987, "grad_norm": 15.265625, "learning_rate": 9.211839113132462e-06, "loss": 0.9007, "step": 22682 }, { "epoch": 2.4135986380080867, "grad_norm": 12.28125, "learning_rate": 9.211803581580444e-06, "loss": 0.8142, "step": 22683 }, { "epoch": 2.4137050436263037, "grad_norm": 13.109375, "learning_rate": 9.211768050028425e-06, "loss": 0.8457, "step": 22684 }, { "epoch": 2.41381144924452, "grad_norm": 14.1953125, "learning_rate": 9.211732518476407e-06, "loss": 0.8605, "step": 22685 }, { "epoch": 2.4139178548627367, "grad_norm": 12.734375, "learning_rate": 9.21169698692439e-06, "loss": 0.8226, "step": 22686 }, { "epoch": 2.4140242604809536, "grad_norm": 19.3125, "learning_rate": 9.211661455372372e-06, "loss": 0.9326, "step": 22687 }, { "epoch": 2.41413066609917, "grad_norm": 12.9921875, "learning_rate": 9.211625923820353e-06, "loss": 0.9773, "step": 22688 }, { "epoch": 2.4142370717173867, "grad_norm": 12.71875, "learning_rate": 9.211590392268335e-06, "loss": 0.815, "step": 22689 }, { "epoch": 2.414343477335603, "grad_norm": 14.125, "learning_rate": 9.211554860716316e-06, "loss": 0.8831, "step": 22690 }, { "epoch": 2.41444988295382, "grad_norm": 11.2578125, "learning_rate": 9.2115193291643e-06, "loss": 0.7213, "step": 22691 }, { "epoch": 2.4145562885720366, "grad_norm": 12.3046875, "learning_rate": 9.211483797612281e-06, "loss": 0.8236, "step": 22692 }, { "epoch": 2.414662694190253, "grad_norm": 14.421875, "learning_rate": 9.211448266060263e-06, "loss": 1.299, "step": 22693 }, { "epoch": 2.4147690998084697, "grad_norm": 13.1015625, "learning_rate": 9.211412734508245e-06, "loss": 0.8142, "step": 22694 }, { "epoch": 2.4148755054266866, "grad_norm": 14.75, "learning_rate": 9.211377202956226e-06, "loss": 0.9309, "step": 22695 }, { "epoch": 2.414981911044903, "grad_norm": 12.546875, "learning_rate": 9.211341671404208e-06, "loss": 0.9234, "step": 22696 }, { "epoch": 2.4150883166631196, "grad_norm": 12.40625, "learning_rate": 9.21130613985219e-06, "loss": 0.8038, "step": 22697 }, { "epoch": 2.4151947222813366, "grad_norm": 14.2421875, "learning_rate": 9.211270608300171e-06, "loss": 0.9679, "step": 22698 }, { "epoch": 2.415301127899553, "grad_norm": 10.1640625, "learning_rate": 9.211235076748153e-06, "loss": 0.774, "step": 22699 }, { "epoch": 2.4154075335177696, "grad_norm": 9.578125, "learning_rate": 9.211199545196134e-06, "loss": 0.7938, "step": 22700 }, { "epoch": 2.4155139391359866, "grad_norm": 9.8359375, "learning_rate": 9.211164013644118e-06, "loss": 0.8347, "step": 22701 }, { "epoch": 2.415620344754203, "grad_norm": 10.34375, "learning_rate": 9.211128482092099e-06, "loss": 0.7979, "step": 22702 }, { "epoch": 2.4157267503724196, "grad_norm": 12.15625, "learning_rate": 9.21109295054008e-06, "loss": 0.9923, "step": 22703 }, { "epoch": 2.4158331559906365, "grad_norm": 11.40625, "learning_rate": 9.211057418988062e-06, "loss": 0.8734, "step": 22704 }, { "epoch": 2.415939561608853, "grad_norm": 14.4140625, "learning_rate": 9.211021887436044e-06, "loss": 1.1962, "step": 22705 }, { "epoch": 2.4160459672270695, "grad_norm": 12.734375, "learning_rate": 9.210986355884026e-06, "loss": 0.9832, "step": 22706 }, { "epoch": 2.416152372845286, "grad_norm": 14.1015625, "learning_rate": 9.210950824332007e-06, "loss": 1.096, "step": 22707 }, { "epoch": 2.416258778463503, "grad_norm": 13.03125, "learning_rate": 9.210915292779989e-06, "loss": 0.8145, "step": 22708 }, { "epoch": 2.4163651840817195, "grad_norm": 11.953125, "learning_rate": 9.21087976122797e-06, "loss": 0.8511, "step": 22709 }, { "epoch": 2.416471589699936, "grad_norm": 17.1875, "learning_rate": 9.210844229675952e-06, "loss": 1.1419, "step": 22710 }, { "epoch": 2.416577995318153, "grad_norm": 14.9921875, "learning_rate": 9.210808698123935e-06, "loss": 0.5625, "step": 22711 }, { "epoch": 2.4166844009363695, "grad_norm": 14.71875, "learning_rate": 9.210773166571917e-06, "loss": 1.1612, "step": 22712 }, { "epoch": 2.416790806554586, "grad_norm": 12.75, "learning_rate": 9.210737635019899e-06, "loss": 1.061, "step": 22713 }, { "epoch": 2.4168972121728025, "grad_norm": 13.203125, "learning_rate": 9.21070210346788e-06, "loss": 0.8651, "step": 22714 }, { "epoch": 2.4170036177910195, "grad_norm": 11.9296875, "learning_rate": 9.210666571915862e-06, "loss": 0.8246, "step": 22715 }, { "epoch": 2.417110023409236, "grad_norm": 13.9375, "learning_rate": 9.210631040363845e-06, "loss": 1.0044, "step": 22716 }, { "epoch": 2.4172164290274525, "grad_norm": 16.703125, "learning_rate": 9.210595508811827e-06, "loss": 1.0132, "step": 22717 }, { "epoch": 2.4173228346456694, "grad_norm": 14.5546875, "learning_rate": 9.210559977259808e-06, "loss": 0.995, "step": 22718 }, { "epoch": 2.417429240263886, "grad_norm": 10.9609375, "learning_rate": 9.21052444570779e-06, "loss": 0.9031, "step": 22719 }, { "epoch": 2.4175356458821025, "grad_norm": 17.40625, "learning_rate": 9.210488914155771e-06, "loss": 0.928, "step": 22720 }, { "epoch": 2.4176420515003194, "grad_norm": 14.3671875, "learning_rate": 9.210453382603753e-06, "loss": 0.8713, "step": 22721 }, { "epoch": 2.417748457118536, "grad_norm": 12.875, "learning_rate": 9.210417851051735e-06, "loss": 0.8837, "step": 22722 }, { "epoch": 2.4178548627367524, "grad_norm": 13.0234375, "learning_rate": 9.210382319499716e-06, "loss": 0.9679, "step": 22723 }, { "epoch": 2.4179612683549694, "grad_norm": 11.96875, "learning_rate": 9.210346787947698e-06, "loss": 0.7355, "step": 22724 }, { "epoch": 2.418067673973186, "grad_norm": 16.046875, "learning_rate": 9.21031125639568e-06, "loss": 0.7325, "step": 22725 }, { "epoch": 2.4181740795914024, "grad_norm": 13.1015625, "learning_rate": 9.210275724843661e-06, "loss": 1.0423, "step": 22726 }, { "epoch": 2.418280485209619, "grad_norm": 14.578125, "learning_rate": 9.210240193291644e-06, "loss": 0.9248, "step": 22727 }, { "epoch": 2.418386890827836, "grad_norm": 16.671875, "learning_rate": 9.210204661739626e-06, "loss": 0.9312, "step": 22728 }, { "epoch": 2.4184932964460524, "grad_norm": 13.875, "learning_rate": 9.210169130187608e-06, "loss": 0.9037, "step": 22729 }, { "epoch": 2.418599702064269, "grad_norm": 12.2109375, "learning_rate": 9.21013359863559e-06, "loss": 0.8597, "step": 22730 }, { "epoch": 2.4187061076824854, "grad_norm": 11.90625, "learning_rate": 9.21009806708357e-06, "loss": 0.8309, "step": 22731 }, { "epoch": 2.4188125133007023, "grad_norm": 11.0703125, "learning_rate": 9.210062535531552e-06, "loss": 0.7947, "step": 22732 }, { "epoch": 2.418918918918919, "grad_norm": 11.671875, "learning_rate": 9.210027003979534e-06, "loss": 0.9414, "step": 22733 }, { "epoch": 2.4190253245371354, "grad_norm": 12.9453125, "learning_rate": 9.209991472427516e-06, "loss": 0.8934, "step": 22734 }, { "epoch": 2.4191317301553523, "grad_norm": 18.59375, "learning_rate": 9.209955940875497e-06, "loss": 1.0201, "step": 22735 }, { "epoch": 2.419238135773569, "grad_norm": 11.5078125, "learning_rate": 9.209920409323479e-06, "loss": 0.9936, "step": 22736 }, { "epoch": 2.4193445413917853, "grad_norm": 15.59375, "learning_rate": 9.209884877771462e-06, "loss": 1.1046, "step": 22737 }, { "epoch": 2.4194509470100023, "grad_norm": 13.2109375, "learning_rate": 9.209849346219444e-06, "loss": 0.9663, "step": 22738 }, { "epoch": 2.419557352628219, "grad_norm": 13.40625, "learning_rate": 9.209813814667425e-06, "loss": 0.9686, "step": 22739 }, { "epoch": 2.4196637582464353, "grad_norm": 11.8203125, "learning_rate": 9.209778283115407e-06, "loss": 0.8799, "step": 22740 }, { "epoch": 2.4197701638646523, "grad_norm": 15.90625, "learning_rate": 9.209742751563389e-06, "loss": 0.787, "step": 22741 }, { "epoch": 2.4198765694828688, "grad_norm": 16.28125, "learning_rate": 9.209707220011372e-06, "loss": 0.8989, "step": 22742 }, { "epoch": 2.4199829751010853, "grad_norm": 13.78125, "learning_rate": 9.209671688459353e-06, "loss": 1.046, "step": 22743 }, { "epoch": 2.420089380719302, "grad_norm": 11.328125, "learning_rate": 9.209636156907335e-06, "loss": 0.953, "step": 22744 }, { "epoch": 2.4201957863375188, "grad_norm": 19.546875, "learning_rate": 9.209600625355317e-06, "loss": 0.7131, "step": 22745 }, { "epoch": 2.4203021919557353, "grad_norm": 12.1484375, "learning_rate": 9.209565093803298e-06, "loss": 0.8357, "step": 22746 }, { "epoch": 2.4204085975739518, "grad_norm": 12.421875, "learning_rate": 9.20952956225128e-06, "loss": 1.0997, "step": 22747 }, { "epoch": 2.4205150031921687, "grad_norm": 10.4609375, "learning_rate": 9.209494030699262e-06, "loss": 0.8483, "step": 22748 }, { "epoch": 2.4206214088103852, "grad_norm": 16.28125, "learning_rate": 9.209458499147243e-06, "loss": 0.8609, "step": 22749 }, { "epoch": 2.4207278144286017, "grad_norm": 12.2421875, "learning_rate": 9.209422967595225e-06, "loss": 0.9031, "step": 22750 }, { "epoch": 2.4208342200468183, "grad_norm": 14.515625, "learning_rate": 9.209387436043206e-06, "loss": 1.0319, "step": 22751 }, { "epoch": 2.420940625665035, "grad_norm": 13.578125, "learning_rate": 9.20935190449119e-06, "loss": 0.8605, "step": 22752 }, { "epoch": 2.4210470312832517, "grad_norm": 12.84375, "learning_rate": 9.209316372939171e-06, "loss": 1.0995, "step": 22753 }, { "epoch": 2.4211534369014682, "grad_norm": 14.046875, "learning_rate": 9.209280841387153e-06, "loss": 1.0187, "step": 22754 }, { "epoch": 2.421259842519685, "grad_norm": 13.828125, "learning_rate": 9.209245309835134e-06, "loss": 0.8954, "step": 22755 }, { "epoch": 2.4213662481379017, "grad_norm": 12.375, "learning_rate": 9.209209778283116e-06, "loss": 0.8499, "step": 22756 }, { "epoch": 2.421472653756118, "grad_norm": 9.9765625, "learning_rate": 9.209174246731098e-06, "loss": 0.7795, "step": 22757 }, { "epoch": 2.421579059374335, "grad_norm": 15.328125, "learning_rate": 9.20913871517908e-06, "loss": 1.0868, "step": 22758 }, { "epoch": 2.4216854649925517, "grad_norm": 15.3828125, "learning_rate": 9.209103183627061e-06, "loss": 0.7998, "step": 22759 }, { "epoch": 2.421791870610768, "grad_norm": 14.0703125, "learning_rate": 9.209067652075042e-06, "loss": 0.7946, "step": 22760 }, { "epoch": 2.421898276228985, "grad_norm": 11.921875, "learning_rate": 9.209032120523024e-06, "loss": 0.8163, "step": 22761 }, { "epoch": 2.4220046818472016, "grad_norm": 12.5859375, "learning_rate": 9.208996588971006e-06, "loss": 0.9885, "step": 22762 }, { "epoch": 2.422111087465418, "grad_norm": 12.1796875, "learning_rate": 9.208961057418989e-06, "loss": 0.731, "step": 22763 }, { "epoch": 2.4222174930836347, "grad_norm": 12.4140625, "learning_rate": 9.20892552586697e-06, "loss": 1.0044, "step": 22764 }, { "epoch": 2.4223238987018516, "grad_norm": 11.453125, "learning_rate": 9.208889994314952e-06, "loss": 0.8491, "step": 22765 }, { "epoch": 2.422430304320068, "grad_norm": 13.1328125, "learning_rate": 9.208854462762934e-06, "loss": 0.9624, "step": 22766 }, { "epoch": 2.4225367099382846, "grad_norm": 13.8671875, "learning_rate": 9.208818931210915e-06, "loss": 0.9142, "step": 22767 }, { "epoch": 2.422643115556501, "grad_norm": 15.5625, "learning_rate": 9.208783399658899e-06, "loss": 0.8622, "step": 22768 }, { "epoch": 2.422749521174718, "grad_norm": 14.7890625, "learning_rate": 9.20874786810688e-06, "loss": 0.9118, "step": 22769 }, { "epoch": 2.4228559267929346, "grad_norm": 18.875, "learning_rate": 9.208712336554862e-06, "loss": 0.8827, "step": 22770 }, { "epoch": 2.422962332411151, "grad_norm": 13.8828125, "learning_rate": 9.208676805002844e-06, "loss": 0.8047, "step": 22771 }, { "epoch": 2.423068738029368, "grad_norm": 14.2890625, "learning_rate": 9.208641273450825e-06, "loss": 0.9111, "step": 22772 }, { "epoch": 2.4231751436475846, "grad_norm": 14.6171875, "learning_rate": 9.208605741898807e-06, "loss": 0.9835, "step": 22773 }, { "epoch": 2.423281549265801, "grad_norm": 12.75, "learning_rate": 9.208570210346788e-06, "loss": 0.8714, "step": 22774 }, { "epoch": 2.423387954884018, "grad_norm": 15.1484375, "learning_rate": 9.20853467879477e-06, "loss": 0.9551, "step": 22775 }, { "epoch": 2.4234943605022345, "grad_norm": 9.6953125, "learning_rate": 9.208499147242752e-06, "loss": 0.773, "step": 22776 }, { "epoch": 2.423600766120451, "grad_norm": 12.1953125, "learning_rate": 9.208463615690733e-06, "loss": 0.99, "step": 22777 }, { "epoch": 2.423707171738668, "grad_norm": 13.6015625, "learning_rate": 9.208428084138716e-06, "loss": 0.9657, "step": 22778 }, { "epoch": 2.4238135773568845, "grad_norm": 11.625, "learning_rate": 9.208392552586698e-06, "loss": 0.8573, "step": 22779 }, { "epoch": 2.423919982975101, "grad_norm": 15.5859375, "learning_rate": 9.20835702103468e-06, "loss": 1.0281, "step": 22780 }, { "epoch": 2.4240263885933175, "grad_norm": 12.8203125, "learning_rate": 9.208321489482661e-06, "loss": 0.8583, "step": 22781 }, { "epoch": 2.4241327942115345, "grad_norm": 10.46875, "learning_rate": 9.208285957930643e-06, "loss": 0.8399, "step": 22782 }, { "epoch": 2.424239199829751, "grad_norm": 11.71875, "learning_rate": 9.208250426378626e-06, "loss": 1.0022, "step": 22783 }, { "epoch": 2.4243456054479675, "grad_norm": 12.4609375, "learning_rate": 9.208214894826608e-06, "loss": 0.871, "step": 22784 }, { "epoch": 2.4244520110661845, "grad_norm": 12.1484375, "learning_rate": 9.20817936327459e-06, "loss": 0.891, "step": 22785 }, { "epoch": 2.424558416684401, "grad_norm": 17.640625, "learning_rate": 9.208143831722571e-06, "loss": 0.8226, "step": 22786 }, { "epoch": 2.4246648223026175, "grad_norm": 13.890625, "learning_rate": 9.208108300170551e-06, "loss": 0.999, "step": 22787 }, { "epoch": 2.424771227920834, "grad_norm": 12.6875, "learning_rate": 9.208072768618534e-06, "loss": 1.0131, "step": 22788 }, { "epoch": 2.424877633539051, "grad_norm": 11.28125, "learning_rate": 9.208037237066516e-06, "loss": 0.7649, "step": 22789 }, { "epoch": 2.4249840391572675, "grad_norm": 14.4921875, "learning_rate": 9.208001705514497e-06, "loss": 0.932, "step": 22790 }, { "epoch": 2.425090444775484, "grad_norm": 14.8984375, "learning_rate": 9.207966173962479e-06, "loss": 0.8937, "step": 22791 }, { "epoch": 2.425196850393701, "grad_norm": 14.015625, "learning_rate": 9.20793064241046e-06, "loss": 0.9899, "step": 22792 }, { "epoch": 2.4253032560119174, "grad_norm": 11.140625, "learning_rate": 9.207895110858444e-06, "loss": 0.8112, "step": 22793 }, { "epoch": 2.425409661630134, "grad_norm": 12.609375, "learning_rate": 9.207859579306426e-06, "loss": 0.8046, "step": 22794 }, { "epoch": 2.425516067248351, "grad_norm": 16.03125, "learning_rate": 9.207824047754407e-06, "loss": 1.0256, "step": 22795 }, { "epoch": 2.4256224728665674, "grad_norm": 13.6171875, "learning_rate": 9.207788516202389e-06, "loss": 0.9152, "step": 22796 }, { "epoch": 2.425728878484784, "grad_norm": 11.9140625, "learning_rate": 9.20775298465037e-06, "loss": 0.9682, "step": 22797 }, { "epoch": 2.425835284103001, "grad_norm": 12.3046875, "learning_rate": 9.207717453098352e-06, "loss": 0.868, "step": 22798 }, { "epoch": 2.4259416897212174, "grad_norm": 14.1484375, "learning_rate": 9.207681921546334e-06, "loss": 0.8113, "step": 22799 }, { "epoch": 2.426048095339434, "grad_norm": 10.4140625, "learning_rate": 9.207646389994315e-06, "loss": 0.8251, "step": 22800 }, { "epoch": 2.4261545009576504, "grad_norm": 11.0, "learning_rate": 9.207610858442297e-06, "loss": 0.9403, "step": 22801 }, { "epoch": 2.4262609065758673, "grad_norm": 12.828125, "learning_rate": 9.207575326890278e-06, "loss": 0.867, "step": 22802 }, { "epoch": 2.426367312194084, "grad_norm": 12.4765625, "learning_rate": 9.20753979533826e-06, "loss": 0.9933, "step": 22803 }, { "epoch": 2.4264737178123004, "grad_norm": 11.1328125, "learning_rate": 9.207504263786243e-06, "loss": 0.9471, "step": 22804 }, { "epoch": 2.4265801234305173, "grad_norm": 12.546875, "learning_rate": 9.207468732234225e-06, "loss": 0.8476, "step": 22805 }, { "epoch": 2.426686529048734, "grad_norm": 11.359375, "learning_rate": 9.207433200682207e-06, "loss": 1.0202, "step": 22806 }, { "epoch": 2.4267929346669503, "grad_norm": 12.015625, "learning_rate": 9.207397669130188e-06, "loss": 0.8771, "step": 22807 }, { "epoch": 2.426899340285167, "grad_norm": 12.453125, "learning_rate": 9.20736213757817e-06, "loss": 0.8993, "step": 22808 }, { "epoch": 2.427005745903384, "grad_norm": 12.265625, "learning_rate": 9.207326606026153e-06, "loss": 0.8162, "step": 22809 }, { "epoch": 2.4271121515216003, "grad_norm": 16.390625, "learning_rate": 9.207291074474135e-06, "loss": 1.0207, "step": 22810 }, { "epoch": 2.427218557139817, "grad_norm": 10.546875, "learning_rate": 9.207255542922116e-06, "loss": 0.7694, "step": 22811 }, { "epoch": 2.4273249627580338, "grad_norm": 12.0859375, "learning_rate": 9.207220011370098e-06, "loss": 1.0052, "step": 22812 }, { "epoch": 2.4274313683762503, "grad_norm": 15.375, "learning_rate": 9.20718447981808e-06, "loss": 1.0654, "step": 22813 }, { "epoch": 2.427537773994467, "grad_norm": 18.75, "learning_rate": 9.207148948266061e-06, "loss": 0.828, "step": 22814 }, { "epoch": 2.4276441796126838, "grad_norm": 14.7578125, "learning_rate": 9.207113416714043e-06, "loss": 1.2471, "step": 22815 }, { "epoch": 2.4277505852309003, "grad_norm": 13.9453125, "learning_rate": 9.207077885162024e-06, "loss": 0.8951, "step": 22816 }, { "epoch": 2.4278569908491168, "grad_norm": 11.546875, "learning_rate": 9.207042353610006e-06, "loss": 0.7436, "step": 22817 }, { "epoch": 2.4279633964673337, "grad_norm": 13.5, "learning_rate": 9.207006822057988e-06, "loss": 1.0892, "step": 22818 }, { "epoch": 2.4280698020855502, "grad_norm": 10.53125, "learning_rate": 9.20697129050597e-06, "loss": 0.8432, "step": 22819 }, { "epoch": 2.4281762077037667, "grad_norm": 12.171875, "learning_rate": 9.206935758953952e-06, "loss": 0.8602, "step": 22820 }, { "epoch": 2.4282826133219833, "grad_norm": 14.3828125, "learning_rate": 9.206900227401934e-06, "loss": 1.0722, "step": 22821 }, { "epoch": 2.4283890189402, "grad_norm": 14.7109375, "learning_rate": 9.206864695849916e-06, "loss": 1.1137, "step": 22822 }, { "epoch": 2.4284954245584167, "grad_norm": 13.25, "learning_rate": 9.206829164297897e-06, "loss": 1.0126, "step": 22823 }, { "epoch": 2.4286018301766332, "grad_norm": 13.4453125, "learning_rate": 9.206793632745879e-06, "loss": 0.8184, "step": 22824 }, { "epoch": 2.4287082357948497, "grad_norm": 12.734375, "learning_rate": 9.20675810119386e-06, "loss": 1.0343, "step": 22825 }, { "epoch": 2.4288146414130667, "grad_norm": 16.3125, "learning_rate": 9.206722569641842e-06, "loss": 0.8038, "step": 22826 }, { "epoch": 2.428921047031283, "grad_norm": 12.390625, "learning_rate": 9.206687038089824e-06, "loss": 0.9565, "step": 22827 }, { "epoch": 2.4290274526494997, "grad_norm": 12.21875, "learning_rate": 9.206651506537805e-06, "loss": 1.1145, "step": 22828 }, { "epoch": 2.4291338582677167, "grad_norm": 12.6640625, "learning_rate": 9.206615974985789e-06, "loss": 0.857, "step": 22829 }, { "epoch": 2.429240263885933, "grad_norm": 11.296875, "learning_rate": 9.20658044343377e-06, "loss": 1.0137, "step": 22830 }, { "epoch": 2.4293466695041497, "grad_norm": 13.828125, "learning_rate": 9.206544911881752e-06, "loss": 1.0467, "step": 22831 }, { "epoch": 2.4294530751223666, "grad_norm": 12.921875, "learning_rate": 9.206509380329733e-06, "loss": 0.7506, "step": 22832 }, { "epoch": 2.429559480740583, "grad_norm": 14.40625, "learning_rate": 9.206473848777715e-06, "loss": 0.8751, "step": 22833 }, { "epoch": 2.4296658863587997, "grad_norm": 13.75, "learning_rate": 9.206438317225698e-06, "loss": 0.9419, "step": 22834 }, { "epoch": 2.4297722919770166, "grad_norm": 17.5625, "learning_rate": 9.20640278567368e-06, "loss": 0.9679, "step": 22835 }, { "epoch": 2.429878697595233, "grad_norm": 11.6953125, "learning_rate": 9.206367254121662e-06, "loss": 0.9016, "step": 22836 }, { "epoch": 2.4299851032134496, "grad_norm": 14.0234375, "learning_rate": 9.206331722569643e-06, "loss": 0.86, "step": 22837 }, { "epoch": 2.430091508831666, "grad_norm": 10.078125, "learning_rate": 9.206296191017625e-06, "loss": 0.7574, "step": 22838 }, { "epoch": 2.430197914449883, "grad_norm": 12.671875, "learning_rate": 9.206260659465606e-06, "loss": 1.0495, "step": 22839 }, { "epoch": 2.4303043200680996, "grad_norm": 15.921875, "learning_rate": 9.206225127913588e-06, "loss": 0.8062, "step": 22840 }, { "epoch": 2.430410725686316, "grad_norm": 10.0390625, "learning_rate": 9.20618959636157e-06, "loss": 0.7512, "step": 22841 }, { "epoch": 2.430517131304533, "grad_norm": 11.3203125, "learning_rate": 9.206154064809551e-06, "loss": 0.7632, "step": 22842 }, { "epoch": 2.4306235369227496, "grad_norm": 16.546875, "learning_rate": 9.206118533257533e-06, "loss": 1.1026, "step": 22843 }, { "epoch": 2.430729942540966, "grad_norm": 19.5, "learning_rate": 9.206083001705514e-06, "loss": 0.9594, "step": 22844 }, { "epoch": 2.4308363481591826, "grad_norm": 16.171875, "learning_rate": 9.206047470153498e-06, "loss": 0.9154, "step": 22845 }, { "epoch": 2.4309427537773995, "grad_norm": 11.8671875, "learning_rate": 9.20601193860148e-06, "loss": 0.6953, "step": 22846 }, { "epoch": 2.431049159395616, "grad_norm": 10.6328125, "learning_rate": 9.205976407049461e-06, "loss": 0.8835, "step": 22847 }, { "epoch": 2.4311555650138326, "grad_norm": 12.1484375, "learning_rate": 9.205940875497443e-06, "loss": 0.7801, "step": 22848 }, { "epoch": 2.4312619706320495, "grad_norm": 16.0, "learning_rate": 9.205905343945424e-06, "loss": 1.1544, "step": 22849 }, { "epoch": 2.431368376250266, "grad_norm": 10.4140625, "learning_rate": 9.205869812393406e-06, "loss": 0.8332, "step": 22850 }, { "epoch": 2.4314747818684825, "grad_norm": 13.484375, "learning_rate": 9.205834280841387e-06, "loss": 0.9177, "step": 22851 }, { "epoch": 2.4315811874866995, "grad_norm": 11.6875, "learning_rate": 9.205798749289369e-06, "loss": 0.8472, "step": 22852 }, { "epoch": 2.431687593104916, "grad_norm": 18.0, "learning_rate": 9.20576321773735e-06, "loss": 1.1581, "step": 22853 }, { "epoch": 2.4317939987231325, "grad_norm": 12.96875, "learning_rate": 9.205727686185332e-06, "loss": 0.7635, "step": 22854 }, { "epoch": 2.4319004043413495, "grad_norm": 10.96875, "learning_rate": 9.205692154633315e-06, "loss": 0.8538, "step": 22855 }, { "epoch": 2.432006809959566, "grad_norm": 16.03125, "learning_rate": 9.205656623081297e-06, "loss": 1.1065, "step": 22856 }, { "epoch": 2.4321132155777825, "grad_norm": 14.1015625, "learning_rate": 9.205621091529279e-06, "loss": 0.8998, "step": 22857 }, { "epoch": 2.432219621195999, "grad_norm": 15.28125, "learning_rate": 9.20558555997726e-06, "loss": 1.0286, "step": 22858 }, { "epoch": 2.432326026814216, "grad_norm": 14.609375, "learning_rate": 9.205550028425242e-06, "loss": 0.8104, "step": 22859 }, { "epoch": 2.4324324324324325, "grad_norm": 12.3046875, "learning_rate": 9.205514496873225e-06, "loss": 0.8036, "step": 22860 }, { "epoch": 2.432538838050649, "grad_norm": 14.8125, "learning_rate": 9.205478965321207e-06, "loss": 0.9867, "step": 22861 }, { "epoch": 2.4326452436688655, "grad_norm": 20.09375, "learning_rate": 9.205443433769188e-06, "loss": 0.9072, "step": 22862 }, { "epoch": 2.4327516492870824, "grad_norm": 12.78125, "learning_rate": 9.20540790221717e-06, "loss": 0.9105, "step": 22863 }, { "epoch": 2.432858054905299, "grad_norm": 14.3359375, "learning_rate": 9.205372370665152e-06, "loss": 0.97, "step": 22864 }, { "epoch": 2.4329644605235154, "grad_norm": 13.0625, "learning_rate": 9.205336839113133e-06, "loss": 0.9251, "step": 22865 }, { "epoch": 2.4330708661417324, "grad_norm": 11.7578125, "learning_rate": 9.205301307561115e-06, "loss": 0.7794, "step": 22866 }, { "epoch": 2.433177271759949, "grad_norm": 14.203125, "learning_rate": 9.205265776009096e-06, "loss": 0.8818, "step": 22867 }, { "epoch": 2.4332836773781654, "grad_norm": 18.453125, "learning_rate": 9.205230244457078e-06, "loss": 0.9358, "step": 22868 }, { "epoch": 2.4333900829963824, "grad_norm": 15.3671875, "learning_rate": 9.20519471290506e-06, "loss": 1.0279, "step": 22869 }, { "epoch": 2.433496488614599, "grad_norm": 14.6875, "learning_rate": 9.205159181353043e-06, "loss": 0.8316, "step": 22870 }, { "epoch": 2.4336028942328154, "grad_norm": 11.84375, "learning_rate": 9.205123649801025e-06, "loss": 0.9129, "step": 22871 }, { "epoch": 2.4337092998510323, "grad_norm": 13.5625, "learning_rate": 9.205088118249006e-06, "loss": 0.91, "step": 22872 }, { "epoch": 2.433815705469249, "grad_norm": 13.4609375, "learning_rate": 9.205052586696988e-06, "loss": 0.7637, "step": 22873 }, { "epoch": 2.4339221110874654, "grad_norm": 13.6953125, "learning_rate": 9.20501705514497e-06, "loss": 0.9486, "step": 22874 }, { "epoch": 2.434028516705682, "grad_norm": 13.8359375, "learning_rate": 9.204981523592953e-06, "loss": 0.9688, "step": 22875 }, { "epoch": 2.434134922323899, "grad_norm": 16.90625, "learning_rate": 9.204945992040933e-06, "loss": 0.8983, "step": 22876 }, { "epoch": 2.4342413279421153, "grad_norm": 12.1875, "learning_rate": 9.204910460488914e-06, "loss": 0.7351, "step": 22877 }, { "epoch": 2.434347733560332, "grad_norm": 12.796875, "learning_rate": 9.204874928936896e-06, "loss": 0.7927, "step": 22878 }, { "epoch": 2.434454139178549, "grad_norm": 12.578125, "learning_rate": 9.204839397384877e-06, "loss": 0.9394, "step": 22879 }, { "epoch": 2.4345605447967653, "grad_norm": 12.125, "learning_rate": 9.204803865832859e-06, "loss": 0.9954, "step": 22880 }, { "epoch": 2.434666950414982, "grad_norm": 10.5234375, "learning_rate": 9.204768334280842e-06, "loss": 0.806, "step": 22881 }, { "epoch": 2.4347733560331983, "grad_norm": 13.984375, "learning_rate": 9.204732802728824e-06, "loss": 0.8958, "step": 22882 }, { "epoch": 2.4348797616514153, "grad_norm": 11.9140625, "learning_rate": 9.204697271176806e-06, "loss": 0.9629, "step": 22883 }, { "epoch": 2.434986167269632, "grad_norm": 13.078125, "learning_rate": 9.204661739624787e-06, "loss": 0.8424, "step": 22884 }, { "epoch": 2.4350925728878483, "grad_norm": 10.171875, "learning_rate": 9.204626208072769e-06, "loss": 0.8674, "step": 22885 }, { "epoch": 2.4351989785060653, "grad_norm": 18.25, "learning_rate": 9.204590676520752e-06, "loss": 0.8945, "step": 22886 }, { "epoch": 2.4353053841242818, "grad_norm": 17.5625, "learning_rate": 9.204555144968734e-06, "loss": 0.9569, "step": 22887 }, { "epoch": 2.4354117897424983, "grad_norm": 10.1640625, "learning_rate": 9.204519613416715e-06, "loss": 0.7843, "step": 22888 }, { "epoch": 2.4355181953607152, "grad_norm": 11.5234375, "learning_rate": 9.204484081864697e-06, "loss": 0.8659, "step": 22889 }, { "epoch": 2.4356246009789317, "grad_norm": 13.2734375, "learning_rate": 9.204448550312678e-06, "loss": 0.7423, "step": 22890 }, { "epoch": 2.4357310065971483, "grad_norm": 13.953125, "learning_rate": 9.20441301876066e-06, "loss": 1.2746, "step": 22891 }, { "epoch": 2.435837412215365, "grad_norm": 10.6953125, "learning_rate": 9.204377487208642e-06, "loss": 0.9067, "step": 22892 }, { "epoch": 2.4359438178335817, "grad_norm": 16.046875, "learning_rate": 9.204341955656623e-06, "loss": 1.0014, "step": 22893 }, { "epoch": 2.4360502234517982, "grad_norm": 12.96875, "learning_rate": 9.204306424104605e-06, "loss": 0.6217, "step": 22894 }, { "epoch": 2.4361566290700147, "grad_norm": 10.4609375, "learning_rate": 9.204270892552586e-06, "loss": 0.8709, "step": 22895 }, { "epoch": 2.4362630346882317, "grad_norm": 13.9296875, "learning_rate": 9.20423536100057e-06, "loss": 0.7859, "step": 22896 }, { "epoch": 2.436369440306448, "grad_norm": 13.0625, "learning_rate": 9.204199829448551e-06, "loss": 0.9135, "step": 22897 }, { "epoch": 2.4364758459246647, "grad_norm": 9.8046875, "learning_rate": 9.204164297896533e-06, "loss": 0.8358, "step": 22898 }, { "epoch": 2.436582251542881, "grad_norm": 10.203125, "learning_rate": 9.204128766344515e-06, "loss": 0.7943, "step": 22899 }, { "epoch": 2.436688657161098, "grad_norm": 14.5, "learning_rate": 9.204093234792496e-06, "loss": 0.891, "step": 22900 }, { "epoch": 2.4367950627793147, "grad_norm": 13.2890625, "learning_rate": 9.20405770324048e-06, "loss": 0.7317, "step": 22901 }, { "epoch": 2.436901468397531, "grad_norm": 11.546875, "learning_rate": 9.204022171688461e-06, "loss": 0.8525, "step": 22902 }, { "epoch": 2.437007874015748, "grad_norm": 13.4375, "learning_rate": 9.203986640136443e-06, "loss": 0.8104, "step": 22903 }, { "epoch": 2.4371142796339647, "grad_norm": 13.6640625, "learning_rate": 9.203951108584424e-06, "loss": 0.5524, "step": 22904 }, { "epoch": 2.437220685252181, "grad_norm": 18.84375, "learning_rate": 9.203915577032406e-06, "loss": 0.9117, "step": 22905 }, { "epoch": 2.437327090870398, "grad_norm": 16.890625, "learning_rate": 9.203880045480388e-06, "loss": 0.9536, "step": 22906 }, { "epoch": 2.4374334964886146, "grad_norm": 11.4609375, "learning_rate": 9.20384451392837e-06, "loss": 0.771, "step": 22907 }, { "epoch": 2.437539902106831, "grad_norm": 20.140625, "learning_rate": 9.20380898237635e-06, "loss": 1.0591, "step": 22908 }, { "epoch": 2.437646307725048, "grad_norm": 13.359375, "learning_rate": 9.203773450824332e-06, "loss": 0.8129, "step": 22909 }, { "epoch": 2.4377527133432646, "grad_norm": 18.6875, "learning_rate": 9.203737919272314e-06, "loss": 1.024, "step": 22910 }, { "epoch": 2.437859118961481, "grad_norm": 22.328125, "learning_rate": 9.203702387720297e-06, "loss": 0.9332, "step": 22911 }, { "epoch": 2.4379655245796976, "grad_norm": 16.46875, "learning_rate": 9.203666856168279e-06, "loss": 0.9768, "step": 22912 }, { "epoch": 2.4380719301979146, "grad_norm": 12.7734375, "learning_rate": 9.20363132461626e-06, "loss": 0.7564, "step": 22913 }, { "epoch": 2.438178335816131, "grad_norm": 11.0390625, "learning_rate": 9.203595793064242e-06, "loss": 0.8416, "step": 22914 }, { "epoch": 2.4382847414343476, "grad_norm": 14.53125, "learning_rate": 9.203560261512224e-06, "loss": 0.9995, "step": 22915 }, { "epoch": 2.4383911470525645, "grad_norm": 12.0625, "learning_rate": 9.203524729960205e-06, "loss": 0.7897, "step": 22916 }, { "epoch": 2.438497552670781, "grad_norm": 11.0703125, "learning_rate": 9.203489198408187e-06, "loss": 0.6761, "step": 22917 }, { "epoch": 2.4386039582889976, "grad_norm": 17.265625, "learning_rate": 9.203453666856169e-06, "loss": 0.8353, "step": 22918 }, { "epoch": 2.438710363907214, "grad_norm": 17.4375, "learning_rate": 9.20341813530415e-06, "loss": 0.9182, "step": 22919 }, { "epoch": 2.438816769525431, "grad_norm": 14.1953125, "learning_rate": 9.203382603752132e-06, "loss": 0.8052, "step": 22920 }, { "epoch": 2.4389231751436475, "grad_norm": 15.65625, "learning_rate": 9.203347072200113e-06, "loss": 0.9624, "step": 22921 }, { "epoch": 2.439029580761864, "grad_norm": 14.015625, "learning_rate": 9.203311540648097e-06, "loss": 0.884, "step": 22922 }, { "epoch": 2.439135986380081, "grad_norm": 10.375, "learning_rate": 9.203276009096078e-06, "loss": 0.8484, "step": 22923 }, { "epoch": 2.4392423919982975, "grad_norm": 17.34375, "learning_rate": 9.20324047754406e-06, "loss": 0.9924, "step": 22924 }, { "epoch": 2.439348797616514, "grad_norm": 17.015625, "learning_rate": 9.203204945992041e-06, "loss": 0.8387, "step": 22925 }, { "epoch": 2.439455203234731, "grad_norm": 15.0625, "learning_rate": 9.203169414440023e-06, "loss": 0.9845, "step": 22926 }, { "epoch": 2.4395616088529475, "grad_norm": 15.484375, "learning_rate": 9.203133882888006e-06, "loss": 0.9017, "step": 22927 }, { "epoch": 2.439668014471164, "grad_norm": 15.65625, "learning_rate": 9.203098351335988e-06, "loss": 0.6565, "step": 22928 }, { "epoch": 2.439774420089381, "grad_norm": 16.90625, "learning_rate": 9.20306281978397e-06, "loss": 0.8134, "step": 22929 }, { "epoch": 2.4398808257075975, "grad_norm": 14.328125, "learning_rate": 9.203027288231951e-06, "loss": 0.9396, "step": 22930 }, { "epoch": 2.439987231325814, "grad_norm": 13.8046875, "learning_rate": 9.202991756679933e-06, "loss": 1.0264, "step": 22931 }, { "epoch": 2.4400936369440305, "grad_norm": 13.8671875, "learning_rate": 9.202956225127914e-06, "loss": 1.064, "step": 22932 }, { "epoch": 2.4402000425622474, "grad_norm": 14.21875, "learning_rate": 9.202920693575896e-06, "loss": 1.007, "step": 22933 }, { "epoch": 2.440306448180464, "grad_norm": 21.28125, "learning_rate": 9.202885162023878e-06, "loss": 0.8883, "step": 22934 }, { "epoch": 2.4404128537986804, "grad_norm": 14.765625, "learning_rate": 9.20284963047186e-06, "loss": 0.8326, "step": 22935 }, { "epoch": 2.4405192594168974, "grad_norm": 9.4296875, "learning_rate": 9.20281409891984e-06, "loss": 0.8924, "step": 22936 }, { "epoch": 2.440625665035114, "grad_norm": 13.4140625, "learning_rate": 9.202778567367824e-06, "loss": 0.656, "step": 22937 }, { "epoch": 2.4407320706533304, "grad_norm": 19.453125, "learning_rate": 9.202743035815806e-06, "loss": 1.078, "step": 22938 }, { "epoch": 2.440838476271547, "grad_norm": 15.8671875, "learning_rate": 9.202707504263787e-06, "loss": 0.9987, "step": 22939 }, { "epoch": 2.440944881889764, "grad_norm": 12.8125, "learning_rate": 9.202671972711769e-06, "loss": 0.7068, "step": 22940 }, { "epoch": 2.4410512875079804, "grad_norm": 12.2421875, "learning_rate": 9.20263644115975e-06, "loss": 1.0023, "step": 22941 }, { "epoch": 2.441157693126197, "grad_norm": 13.8203125, "learning_rate": 9.202600909607732e-06, "loss": 0.7885, "step": 22942 }, { "epoch": 2.441264098744414, "grad_norm": 10.671875, "learning_rate": 9.202565378055714e-06, "loss": 0.9192, "step": 22943 }, { "epoch": 2.4413705043626304, "grad_norm": 12.59375, "learning_rate": 9.202529846503695e-06, "loss": 0.9955, "step": 22944 }, { "epoch": 2.441476909980847, "grad_norm": 14.1015625, "learning_rate": 9.202494314951677e-06, "loss": 0.8861, "step": 22945 }, { "epoch": 2.441583315599064, "grad_norm": 13.703125, "learning_rate": 9.202458783399659e-06, "loss": 0.8046, "step": 22946 }, { "epoch": 2.4416897212172803, "grad_norm": 15.8203125, "learning_rate": 9.202423251847642e-06, "loss": 0.8187, "step": 22947 }, { "epoch": 2.441796126835497, "grad_norm": 13.390625, "learning_rate": 9.202387720295624e-06, "loss": 0.8751, "step": 22948 }, { "epoch": 2.441902532453714, "grad_norm": 14.828125, "learning_rate": 9.202352188743605e-06, "loss": 0.8956, "step": 22949 }, { "epoch": 2.4420089380719303, "grad_norm": 16.140625, "learning_rate": 9.202316657191587e-06, "loss": 1.0214, "step": 22950 }, { "epoch": 2.442115343690147, "grad_norm": 10.1875, "learning_rate": 9.202281125639568e-06, "loss": 0.8584, "step": 22951 }, { "epoch": 2.4422217493083633, "grad_norm": 10.09375, "learning_rate": 9.202245594087552e-06, "loss": 0.9074, "step": 22952 }, { "epoch": 2.4423281549265803, "grad_norm": 10.078125, "learning_rate": 9.202210062535533e-06, "loss": 0.7962, "step": 22953 }, { "epoch": 2.442434560544797, "grad_norm": 13.1015625, "learning_rate": 9.202174530983515e-06, "loss": 1.0463, "step": 22954 }, { "epoch": 2.4425409661630133, "grad_norm": 24.15625, "learning_rate": 9.202138999431496e-06, "loss": 1.2088, "step": 22955 }, { "epoch": 2.44264737178123, "grad_norm": 13.6171875, "learning_rate": 9.202103467879478e-06, "loss": 0.8773, "step": 22956 }, { "epoch": 2.4427537773994468, "grad_norm": 11.5703125, "learning_rate": 9.20206793632746e-06, "loss": 0.8754, "step": 22957 }, { "epoch": 2.4428601830176633, "grad_norm": 11.8046875, "learning_rate": 9.202032404775441e-06, "loss": 0.8763, "step": 22958 }, { "epoch": 2.44296658863588, "grad_norm": 11.84375, "learning_rate": 9.201996873223423e-06, "loss": 0.8682, "step": 22959 }, { "epoch": 2.4430729942540967, "grad_norm": 11.4921875, "learning_rate": 9.201961341671404e-06, "loss": 0.8206, "step": 22960 }, { "epoch": 2.4431793998723133, "grad_norm": 16.1875, "learning_rate": 9.201925810119386e-06, "loss": 0.8561, "step": 22961 }, { "epoch": 2.4432858054905298, "grad_norm": 14.1015625, "learning_rate": 9.201890278567368e-06, "loss": 0.9803, "step": 22962 }, { "epoch": 2.4433922111087467, "grad_norm": 14.1328125, "learning_rate": 9.201854747015351e-06, "loss": 0.8988, "step": 22963 }, { "epoch": 2.4434986167269632, "grad_norm": 12.359375, "learning_rate": 9.201819215463333e-06, "loss": 0.7096, "step": 22964 }, { "epoch": 2.4436050223451797, "grad_norm": 15.4921875, "learning_rate": 9.201783683911314e-06, "loss": 0.9396, "step": 22965 }, { "epoch": 2.4437114279633967, "grad_norm": 12.859375, "learning_rate": 9.201748152359296e-06, "loss": 0.9627, "step": 22966 }, { "epoch": 2.443817833581613, "grad_norm": 13.203125, "learning_rate": 9.201712620807277e-06, "loss": 0.6952, "step": 22967 }, { "epoch": 2.4439242391998297, "grad_norm": 11.4921875, "learning_rate": 9.201677089255259e-06, "loss": 0.9962, "step": 22968 }, { "epoch": 2.444030644818046, "grad_norm": 15.625, "learning_rate": 9.20164155770324e-06, "loss": 0.9778, "step": 22969 }, { "epoch": 2.444137050436263, "grad_norm": 17.140625, "learning_rate": 9.201606026151222e-06, "loss": 0.8093, "step": 22970 }, { "epoch": 2.4442434560544797, "grad_norm": 14.8671875, "learning_rate": 9.201570494599204e-06, "loss": 1.0478, "step": 22971 }, { "epoch": 2.444349861672696, "grad_norm": 14.171875, "learning_rate": 9.201534963047185e-06, "loss": 0.97, "step": 22972 }, { "epoch": 2.444456267290913, "grad_norm": 14.640625, "learning_rate": 9.201499431495169e-06, "loss": 0.9291, "step": 22973 }, { "epoch": 2.4445626729091297, "grad_norm": 14.484375, "learning_rate": 9.20146389994315e-06, "loss": 0.8957, "step": 22974 }, { "epoch": 2.444669078527346, "grad_norm": 18.453125, "learning_rate": 9.201428368391132e-06, "loss": 1.2679, "step": 22975 }, { "epoch": 2.4447754841455627, "grad_norm": 12.6171875, "learning_rate": 9.201392836839114e-06, "loss": 0.8017, "step": 22976 }, { "epoch": 2.4448818897637796, "grad_norm": 11.625, "learning_rate": 9.201357305287095e-06, "loss": 0.7487, "step": 22977 }, { "epoch": 2.444988295381996, "grad_norm": 11.921875, "learning_rate": 9.201321773735078e-06, "loss": 0.7407, "step": 22978 }, { "epoch": 2.4450947010002126, "grad_norm": 11.0703125, "learning_rate": 9.20128624218306e-06, "loss": 0.8154, "step": 22979 }, { "epoch": 2.4452011066184296, "grad_norm": 13.1640625, "learning_rate": 9.201250710631042e-06, "loss": 0.8194, "step": 22980 }, { "epoch": 2.445307512236646, "grad_norm": 12.9609375, "learning_rate": 9.201215179079023e-06, "loss": 0.8359, "step": 22981 }, { "epoch": 2.4454139178548626, "grad_norm": 11.65625, "learning_rate": 9.201179647527005e-06, "loss": 0.9766, "step": 22982 }, { "epoch": 2.4455203234730796, "grad_norm": 14.8203125, "learning_rate": 9.201144115974987e-06, "loss": 0.8062, "step": 22983 }, { "epoch": 2.445626729091296, "grad_norm": 12.2265625, "learning_rate": 9.201108584422968e-06, "loss": 0.8286, "step": 22984 }, { "epoch": 2.4457331347095126, "grad_norm": 21.1875, "learning_rate": 9.20107305287095e-06, "loss": 1.0608, "step": 22985 }, { "epoch": 2.4458395403277295, "grad_norm": 15.21875, "learning_rate": 9.201037521318931e-06, "loss": 0.8312, "step": 22986 }, { "epoch": 2.445945945945946, "grad_norm": 12.8046875, "learning_rate": 9.201001989766913e-06, "loss": 0.6908, "step": 22987 }, { "epoch": 2.4460523515641626, "grad_norm": 12.4453125, "learning_rate": 9.200966458214896e-06, "loss": 0.913, "step": 22988 }, { "epoch": 2.446158757182379, "grad_norm": 15.3125, "learning_rate": 9.200930926662878e-06, "loss": 0.7929, "step": 22989 }, { "epoch": 2.446265162800596, "grad_norm": 14.5390625, "learning_rate": 9.20089539511086e-06, "loss": 0.7647, "step": 22990 }, { "epoch": 2.4463715684188125, "grad_norm": 11.828125, "learning_rate": 9.200859863558841e-06, "loss": 0.87, "step": 22991 }, { "epoch": 2.446477974037029, "grad_norm": 11.21875, "learning_rate": 9.200824332006823e-06, "loss": 1.003, "step": 22992 }, { "epoch": 2.4465843796552456, "grad_norm": 12.328125, "learning_rate": 9.200788800454806e-06, "loss": 0.9091, "step": 22993 }, { "epoch": 2.4466907852734625, "grad_norm": 11.4921875, "learning_rate": 9.200753268902788e-06, "loss": 0.836, "step": 22994 }, { "epoch": 2.446797190891679, "grad_norm": 11.65625, "learning_rate": 9.20071773735077e-06, "loss": 0.8854, "step": 22995 }, { "epoch": 2.4469035965098955, "grad_norm": 11.9140625, "learning_rate": 9.200682205798749e-06, "loss": 0.7192, "step": 22996 }, { "epoch": 2.4470100021281125, "grad_norm": 11.75, "learning_rate": 9.20064667424673e-06, "loss": 0.922, "step": 22997 }, { "epoch": 2.447116407746329, "grad_norm": 12.921875, "learning_rate": 9.200611142694712e-06, "loss": 0.791, "step": 22998 }, { "epoch": 2.4472228133645455, "grad_norm": 12.5625, "learning_rate": 9.200575611142696e-06, "loss": 1.042, "step": 22999 }, { "epoch": 2.4473292189827625, "grad_norm": 12.1328125, "learning_rate": 9.200540079590677e-06, "loss": 0.9464, "step": 23000 }, { "epoch": 2.447435624600979, "grad_norm": 13.7265625, "learning_rate": 9.200504548038659e-06, "loss": 1.017, "step": 23001 }, { "epoch": 2.4475420302191955, "grad_norm": 11.3125, "learning_rate": 9.20046901648664e-06, "loss": 0.9099, "step": 23002 }, { "epoch": 2.4476484358374124, "grad_norm": 12.359375, "learning_rate": 9.200433484934622e-06, "loss": 0.8476, "step": 23003 }, { "epoch": 2.447754841455629, "grad_norm": 10.2734375, "learning_rate": 9.200397953382605e-06, "loss": 0.8267, "step": 23004 }, { "epoch": 2.4478612470738454, "grad_norm": 12.4609375, "learning_rate": 9.200362421830587e-06, "loss": 0.8191, "step": 23005 }, { "epoch": 2.447967652692062, "grad_norm": 11.140625, "learning_rate": 9.200326890278569e-06, "loss": 0.9172, "step": 23006 }, { "epoch": 2.448074058310279, "grad_norm": 16.890625, "learning_rate": 9.20029135872655e-06, "loss": 0.8678, "step": 23007 }, { "epoch": 2.4481804639284954, "grad_norm": 12.8984375, "learning_rate": 9.200255827174532e-06, "loss": 0.9451, "step": 23008 }, { "epoch": 2.448286869546712, "grad_norm": 14.28125, "learning_rate": 9.200220295622513e-06, "loss": 0.9798, "step": 23009 }, { "epoch": 2.448393275164929, "grad_norm": 14.703125, "learning_rate": 9.200184764070495e-06, "loss": 0.9516, "step": 23010 }, { "epoch": 2.4484996807831454, "grad_norm": 12.609375, "learning_rate": 9.200149232518477e-06, "loss": 0.7624, "step": 23011 }, { "epoch": 2.448606086401362, "grad_norm": 13.7265625, "learning_rate": 9.200113700966458e-06, "loss": 1.0201, "step": 23012 }, { "epoch": 2.4487124920195784, "grad_norm": 18.375, "learning_rate": 9.20007816941444e-06, "loss": 1.0306, "step": 23013 }, { "epoch": 2.4488188976377954, "grad_norm": 15.0625, "learning_rate": 9.200042637862423e-06, "loss": 0.9029, "step": 23014 }, { "epoch": 2.448925303256012, "grad_norm": 13.953125, "learning_rate": 9.200007106310405e-06, "loss": 0.7459, "step": 23015 }, { "epoch": 2.4490317088742284, "grad_norm": 12.125, "learning_rate": 9.199971574758386e-06, "loss": 0.9761, "step": 23016 }, { "epoch": 2.4491381144924453, "grad_norm": 15.84375, "learning_rate": 9.199936043206368e-06, "loss": 0.9882, "step": 23017 }, { "epoch": 2.449244520110662, "grad_norm": 12.1796875, "learning_rate": 9.19990051165435e-06, "loss": 0.9698, "step": 23018 }, { "epoch": 2.4493509257288784, "grad_norm": 12.796875, "learning_rate": 9.199864980102333e-06, "loss": 1.0773, "step": 23019 }, { "epoch": 2.4494573313470953, "grad_norm": 14.2578125, "learning_rate": 9.199829448550314e-06, "loss": 1.0261, "step": 23020 }, { "epoch": 2.449563736965312, "grad_norm": 14.46875, "learning_rate": 9.199793916998296e-06, "loss": 0.8514, "step": 23021 }, { "epoch": 2.4496701425835283, "grad_norm": 13.703125, "learning_rate": 9.199758385446278e-06, "loss": 0.8499, "step": 23022 }, { "epoch": 2.4497765482017453, "grad_norm": 15.625, "learning_rate": 9.19972285389426e-06, "loss": 0.9829, "step": 23023 }, { "epoch": 2.449882953819962, "grad_norm": 19.1875, "learning_rate": 9.199687322342241e-06, "loss": 0.9784, "step": 23024 }, { "epoch": 2.4499893594381783, "grad_norm": 12.140625, "learning_rate": 9.199651790790222e-06, "loss": 1.0287, "step": 23025 }, { "epoch": 2.450095765056395, "grad_norm": 13.015625, "learning_rate": 9.199616259238204e-06, "loss": 0.9585, "step": 23026 }, { "epoch": 2.4502021706746118, "grad_norm": 17.5, "learning_rate": 9.199580727686186e-06, "loss": 0.9595, "step": 23027 }, { "epoch": 2.4503085762928283, "grad_norm": 20.0625, "learning_rate": 9.199545196134167e-06, "loss": 1.05, "step": 23028 }, { "epoch": 2.450414981911045, "grad_norm": 13.265625, "learning_rate": 9.19950966458215e-06, "loss": 0.8787, "step": 23029 }, { "epoch": 2.4505213875292613, "grad_norm": 13.1484375, "learning_rate": 9.199474133030132e-06, "loss": 1.0133, "step": 23030 }, { "epoch": 2.4506277931474783, "grad_norm": 13.78125, "learning_rate": 9.199438601478114e-06, "loss": 0.945, "step": 23031 }, { "epoch": 2.4507341987656948, "grad_norm": 11.25, "learning_rate": 9.199403069926095e-06, "loss": 0.7443, "step": 23032 }, { "epoch": 2.4508406043839113, "grad_norm": 14.421875, "learning_rate": 9.199367538374077e-06, "loss": 1.0326, "step": 23033 }, { "epoch": 2.4509470100021282, "grad_norm": 13.3203125, "learning_rate": 9.199332006822059e-06, "loss": 0.8881, "step": 23034 }, { "epoch": 2.4510534156203447, "grad_norm": 8.9765625, "learning_rate": 9.19929647527004e-06, "loss": 0.916, "step": 23035 }, { "epoch": 2.4511598212385612, "grad_norm": 11.828125, "learning_rate": 9.199260943718022e-06, "loss": 0.8475, "step": 23036 }, { "epoch": 2.451266226856778, "grad_norm": 15.625, "learning_rate": 9.199225412166003e-06, "loss": 0.8377, "step": 23037 }, { "epoch": 2.4513726324749947, "grad_norm": 12.84375, "learning_rate": 9.199189880613985e-06, "loss": 0.9102, "step": 23038 }, { "epoch": 2.451479038093211, "grad_norm": 11.5, "learning_rate": 9.199154349061967e-06, "loss": 0.8138, "step": 23039 }, { "epoch": 2.451585443711428, "grad_norm": 15.0625, "learning_rate": 9.19911881750995e-06, "loss": 0.9944, "step": 23040 }, { "epoch": 2.4516918493296447, "grad_norm": 12.28125, "learning_rate": 9.199083285957932e-06, "loss": 0.9119, "step": 23041 }, { "epoch": 2.451798254947861, "grad_norm": 22.828125, "learning_rate": 9.199047754405913e-06, "loss": 1.1134, "step": 23042 }, { "epoch": 2.4519046605660777, "grad_norm": 11.6640625, "learning_rate": 9.199012222853895e-06, "loss": 0.7964, "step": 23043 }, { "epoch": 2.4520110661842947, "grad_norm": 14.5703125, "learning_rate": 9.198976691301876e-06, "loss": 1.1275, "step": 23044 }, { "epoch": 2.452117471802511, "grad_norm": 16.21875, "learning_rate": 9.19894115974986e-06, "loss": 0.9305, "step": 23045 }, { "epoch": 2.4522238774207277, "grad_norm": 13.8046875, "learning_rate": 9.198905628197841e-06, "loss": 0.9578, "step": 23046 }, { "epoch": 2.4523302830389446, "grad_norm": 15.9375, "learning_rate": 9.198870096645823e-06, "loss": 0.8497, "step": 23047 }, { "epoch": 2.452436688657161, "grad_norm": 11.21875, "learning_rate": 9.198834565093804e-06, "loss": 0.8648, "step": 23048 }, { "epoch": 2.4525430942753776, "grad_norm": 13.1328125, "learning_rate": 9.198799033541786e-06, "loss": 0.843, "step": 23049 }, { "epoch": 2.452649499893594, "grad_norm": 12.546875, "learning_rate": 9.198763501989768e-06, "loss": 0.7901, "step": 23050 }, { "epoch": 2.452755905511811, "grad_norm": 15.0703125, "learning_rate": 9.19872797043775e-06, "loss": 0.9151, "step": 23051 }, { "epoch": 2.4528623111300276, "grad_norm": 11.5625, "learning_rate": 9.198692438885731e-06, "loss": 0.873, "step": 23052 }, { "epoch": 2.452968716748244, "grad_norm": 12.2109375, "learning_rate": 9.198656907333713e-06, "loss": 0.8388, "step": 23053 }, { "epoch": 2.453075122366461, "grad_norm": 12.640625, "learning_rate": 9.198621375781694e-06, "loss": 0.856, "step": 23054 }, { "epoch": 2.4531815279846776, "grad_norm": 11.59375, "learning_rate": 9.198585844229677e-06, "loss": 0.7727, "step": 23055 }, { "epoch": 2.453287933602894, "grad_norm": 11.6796875, "learning_rate": 9.198550312677659e-06, "loss": 1.0122, "step": 23056 }, { "epoch": 2.453394339221111, "grad_norm": 12.8125, "learning_rate": 9.19851478112564e-06, "loss": 0.6826, "step": 23057 }, { "epoch": 2.4535007448393276, "grad_norm": 17.65625, "learning_rate": 9.198479249573622e-06, "loss": 1.0607, "step": 23058 }, { "epoch": 2.453607150457544, "grad_norm": 11.75, "learning_rate": 9.198443718021604e-06, "loss": 0.9754, "step": 23059 }, { "epoch": 2.453713556075761, "grad_norm": 15.296875, "learning_rate": 9.198408186469585e-06, "loss": 1.041, "step": 23060 }, { "epoch": 2.4538199616939775, "grad_norm": 11.78125, "learning_rate": 9.198372654917567e-06, "loss": 1.1198, "step": 23061 }, { "epoch": 2.453926367312194, "grad_norm": 12.0234375, "learning_rate": 9.198337123365549e-06, "loss": 0.9042, "step": 23062 }, { "epoch": 2.4540327729304106, "grad_norm": 11.7578125, "learning_rate": 9.19830159181353e-06, "loss": 0.81, "step": 23063 }, { "epoch": 2.4541391785486275, "grad_norm": 15.0078125, "learning_rate": 9.198266060261512e-06, "loss": 0.9313, "step": 23064 }, { "epoch": 2.454245584166844, "grad_norm": 13.1640625, "learning_rate": 9.198230528709495e-06, "loss": 0.8514, "step": 23065 }, { "epoch": 2.4543519897850605, "grad_norm": 13.875, "learning_rate": 9.198194997157477e-06, "loss": 0.8351, "step": 23066 }, { "epoch": 2.4544583954032775, "grad_norm": 16.34375, "learning_rate": 9.198159465605458e-06, "loss": 0.9961, "step": 23067 }, { "epoch": 2.454564801021494, "grad_norm": 13.9921875, "learning_rate": 9.19812393405344e-06, "loss": 0.8418, "step": 23068 }, { "epoch": 2.4546712066397105, "grad_norm": 11.0546875, "learning_rate": 9.198088402501422e-06, "loss": 0.8447, "step": 23069 }, { "epoch": 2.454777612257927, "grad_norm": 15.4609375, "learning_rate": 9.198052870949405e-06, "loss": 0.9468, "step": 23070 }, { "epoch": 2.454884017876144, "grad_norm": 13.0625, "learning_rate": 9.198017339397387e-06, "loss": 0.9792, "step": 23071 }, { "epoch": 2.4549904234943605, "grad_norm": 13.1796875, "learning_rate": 9.197981807845368e-06, "loss": 0.9469, "step": 23072 }, { "epoch": 2.455096829112577, "grad_norm": 13.375, "learning_rate": 9.19794627629335e-06, "loss": 0.7355, "step": 23073 }, { "epoch": 2.455203234730794, "grad_norm": 12.765625, "learning_rate": 9.197910744741331e-06, "loss": 1.0119, "step": 23074 }, { "epoch": 2.4553096403490104, "grad_norm": 9.8125, "learning_rate": 9.197875213189313e-06, "loss": 0.8944, "step": 23075 }, { "epoch": 2.455416045967227, "grad_norm": 15.5, "learning_rate": 9.197839681637295e-06, "loss": 1.0759, "step": 23076 }, { "epoch": 2.455522451585444, "grad_norm": 13.3125, "learning_rate": 9.197804150085276e-06, "loss": 0.8187, "step": 23077 }, { "epoch": 2.4556288572036604, "grad_norm": 13.8125, "learning_rate": 9.197768618533258e-06, "loss": 0.8468, "step": 23078 }, { "epoch": 2.455735262821877, "grad_norm": 10.7265625, "learning_rate": 9.19773308698124e-06, "loss": 0.779, "step": 23079 }, { "epoch": 2.455841668440094, "grad_norm": 17.0625, "learning_rate": 9.197697555429221e-06, "loss": 1.0214, "step": 23080 }, { "epoch": 2.4559480740583104, "grad_norm": 16.0, "learning_rate": 9.197662023877204e-06, "loss": 0.8924, "step": 23081 }, { "epoch": 2.456054479676527, "grad_norm": 13.53125, "learning_rate": 9.197626492325186e-06, "loss": 0.9658, "step": 23082 }, { "epoch": 2.4561608852947434, "grad_norm": 16.046875, "learning_rate": 9.197590960773168e-06, "loss": 0.938, "step": 23083 }, { "epoch": 2.4562672909129604, "grad_norm": 15.5703125, "learning_rate": 9.197555429221149e-06, "loss": 0.8493, "step": 23084 }, { "epoch": 2.456373696531177, "grad_norm": 12.0625, "learning_rate": 9.19751989766913e-06, "loss": 0.894, "step": 23085 }, { "epoch": 2.4564801021493934, "grad_norm": 13.5078125, "learning_rate": 9.197484366117112e-06, "loss": 1.0074, "step": 23086 }, { "epoch": 2.45658650776761, "grad_norm": 15.2578125, "learning_rate": 9.197448834565094e-06, "loss": 0.8703, "step": 23087 }, { "epoch": 2.456692913385827, "grad_norm": 12.921875, "learning_rate": 9.197413303013076e-06, "loss": 1.1014, "step": 23088 }, { "epoch": 2.4567993190040434, "grad_norm": 18.84375, "learning_rate": 9.197377771461057e-06, "loss": 0.7593, "step": 23089 }, { "epoch": 2.45690572462226, "grad_norm": 12.5078125, "learning_rate": 9.197342239909039e-06, "loss": 0.9477, "step": 23090 }, { "epoch": 2.457012130240477, "grad_norm": 11.7578125, "learning_rate": 9.197306708357022e-06, "loss": 1.0384, "step": 23091 }, { "epoch": 2.4571185358586933, "grad_norm": 20.078125, "learning_rate": 9.197271176805004e-06, "loss": 0.8936, "step": 23092 }, { "epoch": 2.45722494147691, "grad_norm": 12.984375, "learning_rate": 9.197235645252985e-06, "loss": 0.835, "step": 23093 }, { "epoch": 2.457331347095127, "grad_norm": 16.703125, "learning_rate": 9.197200113700967e-06, "loss": 0.9261, "step": 23094 }, { "epoch": 2.4574377527133433, "grad_norm": 10.3984375, "learning_rate": 9.197164582148948e-06, "loss": 0.9253, "step": 23095 }, { "epoch": 2.45754415833156, "grad_norm": 13.71875, "learning_rate": 9.197129050596932e-06, "loss": 0.797, "step": 23096 }, { "epoch": 2.4576505639497768, "grad_norm": 11.6328125, "learning_rate": 9.197093519044913e-06, "loss": 0.7828, "step": 23097 }, { "epoch": 2.4577569695679933, "grad_norm": 14.4296875, "learning_rate": 9.197057987492895e-06, "loss": 0.9426, "step": 23098 }, { "epoch": 2.45786337518621, "grad_norm": 16.609375, "learning_rate": 9.197022455940877e-06, "loss": 1.0331, "step": 23099 }, { "epoch": 2.4579697808044263, "grad_norm": 9.8515625, "learning_rate": 9.196986924388858e-06, "loss": 0.7582, "step": 23100 }, { "epoch": 2.4580761864226433, "grad_norm": 13.1015625, "learning_rate": 9.19695139283684e-06, "loss": 0.8044, "step": 23101 }, { "epoch": 2.4581825920408598, "grad_norm": 12.6484375, "learning_rate": 9.196915861284821e-06, "loss": 0.9087, "step": 23102 }, { "epoch": 2.4582889976590763, "grad_norm": 16.265625, "learning_rate": 9.196880329732803e-06, "loss": 0.9134, "step": 23103 }, { "epoch": 2.4583954032772932, "grad_norm": 9.171875, "learning_rate": 9.196844798180785e-06, "loss": 0.8283, "step": 23104 }, { "epoch": 2.4585018088955097, "grad_norm": 14.15625, "learning_rate": 9.196809266628766e-06, "loss": 0.8532, "step": 23105 }, { "epoch": 2.4586082145137262, "grad_norm": 11.5703125, "learning_rate": 9.19677373507675e-06, "loss": 0.9937, "step": 23106 }, { "epoch": 2.4587146201319428, "grad_norm": 10.609375, "learning_rate": 9.196738203524731e-06, "loss": 0.8246, "step": 23107 }, { "epoch": 2.4588210257501597, "grad_norm": 15.09375, "learning_rate": 9.196702671972713e-06, "loss": 1.022, "step": 23108 }, { "epoch": 2.458927431368376, "grad_norm": 10.1484375, "learning_rate": 9.196667140420694e-06, "loss": 0.9151, "step": 23109 }, { "epoch": 2.4590338369865927, "grad_norm": 10.1484375, "learning_rate": 9.196631608868676e-06, "loss": 0.7875, "step": 23110 }, { "epoch": 2.4591402426048097, "grad_norm": 15.2734375, "learning_rate": 9.19659607731666e-06, "loss": 1.0861, "step": 23111 }, { "epoch": 2.459246648223026, "grad_norm": 15.4765625, "learning_rate": 9.196560545764641e-06, "loss": 0.8909, "step": 23112 }, { "epoch": 2.4593530538412427, "grad_norm": 12.921875, "learning_rate": 9.196525014212622e-06, "loss": 0.9178, "step": 23113 }, { "epoch": 2.4594594594594597, "grad_norm": 13.890625, "learning_rate": 9.196489482660604e-06, "loss": 0.714, "step": 23114 }, { "epoch": 2.459565865077676, "grad_norm": 17.6875, "learning_rate": 9.196453951108586e-06, "loss": 0.9889, "step": 23115 }, { "epoch": 2.4596722706958927, "grad_norm": 15.2421875, "learning_rate": 9.196418419556566e-06, "loss": 0.9311, "step": 23116 }, { "epoch": 2.4597786763141096, "grad_norm": 12.3203125, "learning_rate": 9.196382888004549e-06, "loss": 0.9056, "step": 23117 }, { "epoch": 2.459885081932326, "grad_norm": 11.2578125, "learning_rate": 9.19634735645253e-06, "loss": 0.9803, "step": 23118 }, { "epoch": 2.4599914875505426, "grad_norm": 18.390625, "learning_rate": 9.196311824900512e-06, "loss": 1.1725, "step": 23119 }, { "epoch": 2.460097893168759, "grad_norm": 10.8515625, "learning_rate": 9.196276293348494e-06, "loss": 0.8587, "step": 23120 }, { "epoch": 2.460204298786976, "grad_norm": 13.546875, "learning_rate": 9.196240761796475e-06, "loss": 0.8888, "step": 23121 }, { "epoch": 2.4603107044051926, "grad_norm": 9.6171875, "learning_rate": 9.196205230244459e-06, "loss": 0.8411, "step": 23122 }, { "epoch": 2.460417110023409, "grad_norm": 14.125, "learning_rate": 9.19616969869244e-06, "loss": 0.8145, "step": 23123 }, { "epoch": 2.4605235156416256, "grad_norm": 20.359375, "learning_rate": 9.196134167140422e-06, "loss": 0.9032, "step": 23124 }, { "epoch": 2.4606299212598426, "grad_norm": 18.125, "learning_rate": 9.196098635588403e-06, "loss": 1.1316, "step": 23125 }, { "epoch": 2.460736326878059, "grad_norm": 14.5390625, "learning_rate": 9.196063104036385e-06, "loss": 0.7983, "step": 23126 }, { "epoch": 2.4608427324962756, "grad_norm": 15.3671875, "learning_rate": 9.196027572484367e-06, "loss": 0.9463, "step": 23127 }, { "epoch": 2.4609491381144926, "grad_norm": 12.15625, "learning_rate": 9.195992040932348e-06, "loss": 0.8115, "step": 23128 }, { "epoch": 2.461055543732709, "grad_norm": 13.21875, "learning_rate": 9.19595650938033e-06, "loss": 0.8142, "step": 23129 }, { "epoch": 2.4611619493509256, "grad_norm": 16.671875, "learning_rate": 9.195920977828311e-06, "loss": 0.9819, "step": 23130 }, { "epoch": 2.4612683549691425, "grad_norm": 11.5546875, "learning_rate": 9.195885446276293e-06, "loss": 0.9005, "step": 23131 }, { "epoch": 2.461374760587359, "grad_norm": 13.625, "learning_rate": 9.195849914724276e-06, "loss": 0.7494, "step": 23132 }, { "epoch": 2.4614811662055756, "grad_norm": 13.1640625, "learning_rate": 9.195814383172258e-06, "loss": 1.1141, "step": 23133 }, { "epoch": 2.4615875718237925, "grad_norm": 11.453125, "learning_rate": 9.19577885162024e-06, "loss": 0.8304, "step": 23134 }, { "epoch": 2.461693977442009, "grad_norm": 14.2890625, "learning_rate": 9.195743320068221e-06, "loss": 0.9634, "step": 23135 }, { "epoch": 2.4618003830602255, "grad_norm": 16.84375, "learning_rate": 9.195707788516203e-06, "loss": 0.9119, "step": 23136 }, { "epoch": 2.461906788678442, "grad_norm": 14.3515625, "learning_rate": 9.195672256964186e-06, "loss": 0.9505, "step": 23137 }, { "epoch": 2.462013194296659, "grad_norm": 16.953125, "learning_rate": 9.195636725412168e-06, "loss": 0.8856, "step": 23138 }, { "epoch": 2.4621195999148755, "grad_norm": 13.25, "learning_rate": 9.19560119386015e-06, "loss": 0.9597, "step": 23139 }, { "epoch": 2.462226005533092, "grad_norm": 14.765625, "learning_rate": 9.195565662308131e-06, "loss": 0.9017, "step": 23140 }, { "epoch": 2.462332411151309, "grad_norm": 14.21875, "learning_rate": 9.195530130756113e-06, "loss": 0.745, "step": 23141 }, { "epoch": 2.4624388167695255, "grad_norm": 9.453125, "learning_rate": 9.195494599204094e-06, "loss": 0.7809, "step": 23142 }, { "epoch": 2.462545222387742, "grad_norm": 16.296875, "learning_rate": 9.195459067652076e-06, "loss": 0.8387, "step": 23143 }, { "epoch": 2.4626516280059585, "grad_norm": 13.875, "learning_rate": 9.195423536100057e-06, "loss": 0.7993, "step": 23144 }, { "epoch": 2.4627580336241754, "grad_norm": 11.8828125, "learning_rate": 9.195388004548039e-06, "loss": 0.9363, "step": 23145 }, { "epoch": 2.462864439242392, "grad_norm": 14.8203125, "learning_rate": 9.19535247299602e-06, "loss": 0.8785, "step": 23146 }, { "epoch": 2.4629708448606085, "grad_norm": 12.3984375, "learning_rate": 9.195316941444004e-06, "loss": 0.9129, "step": 23147 }, { "epoch": 2.4630772504788254, "grad_norm": 13.1328125, "learning_rate": 9.195281409891985e-06, "loss": 0.9008, "step": 23148 }, { "epoch": 2.463183656097042, "grad_norm": 12.25, "learning_rate": 9.195245878339967e-06, "loss": 0.8181, "step": 23149 }, { "epoch": 2.4632900617152584, "grad_norm": 19.125, "learning_rate": 9.195210346787949e-06, "loss": 1.0142, "step": 23150 }, { "epoch": 2.4633964673334754, "grad_norm": 11.5234375, "learning_rate": 9.19517481523593e-06, "loss": 0.8171, "step": 23151 }, { "epoch": 2.463502872951692, "grad_norm": 14.0546875, "learning_rate": 9.195139283683912e-06, "loss": 1.0155, "step": 23152 }, { "epoch": 2.4636092785699084, "grad_norm": 14.171875, "learning_rate": 9.195103752131894e-06, "loss": 1.0102, "step": 23153 }, { "epoch": 2.4637156841881254, "grad_norm": 17.390625, "learning_rate": 9.195068220579875e-06, "loss": 0.9178, "step": 23154 }, { "epoch": 2.463822089806342, "grad_norm": 11.9140625, "learning_rate": 9.195032689027857e-06, "loss": 0.9094, "step": 23155 }, { "epoch": 2.4639284954245584, "grad_norm": 12.3046875, "learning_rate": 9.194997157475838e-06, "loss": 0.7205, "step": 23156 }, { "epoch": 2.464034901042775, "grad_norm": 12.2890625, "learning_rate": 9.19496162592382e-06, "loss": 0.8206, "step": 23157 }, { "epoch": 2.464141306660992, "grad_norm": 14.6953125, "learning_rate": 9.194926094371803e-06, "loss": 0.8908, "step": 23158 }, { "epoch": 2.4642477122792084, "grad_norm": 12.96875, "learning_rate": 9.194890562819785e-06, "loss": 0.9065, "step": 23159 }, { "epoch": 2.464354117897425, "grad_norm": 12.4140625, "learning_rate": 9.194855031267766e-06, "loss": 0.9832, "step": 23160 }, { "epoch": 2.4644605235156414, "grad_norm": 12.7421875, "learning_rate": 9.194819499715748e-06, "loss": 0.8493, "step": 23161 }, { "epoch": 2.4645669291338583, "grad_norm": 11.1328125, "learning_rate": 9.19478396816373e-06, "loss": 0.9171, "step": 23162 }, { "epoch": 2.464673334752075, "grad_norm": 11.28125, "learning_rate": 9.194748436611713e-06, "loss": 0.828, "step": 23163 }, { "epoch": 2.4647797403702913, "grad_norm": 13.0625, "learning_rate": 9.194712905059695e-06, "loss": 0.9409, "step": 23164 }, { "epoch": 2.4648861459885083, "grad_norm": 11.796875, "learning_rate": 9.194677373507676e-06, "loss": 0.7603, "step": 23165 }, { "epoch": 2.464992551606725, "grad_norm": 13.359375, "learning_rate": 9.194641841955658e-06, "loss": 0.9693, "step": 23166 }, { "epoch": 2.4650989572249413, "grad_norm": 13.4375, "learning_rate": 9.19460631040364e-06, "loss": 0.9087, "step": 23167 }, { "epoch": 2.4652053628431583, "grad_norm": 20.75, "learning_rate": 9.194570778851621e-06, "loss": 1.0842, "step": 23168 }, { "epoch": 2.465311768461375, "grad_norm": 13.171875, "learning_rate": 9.194535247299603e-06, "loss": 0.7864, "step": 23169 }, { "epoch": 2.4654181740795913, "grad_norm": 13.8828125, "learning_rate": 9.194499715747584e-06, "loss": 1.0788, "step": 23170 }, { "epoch": 2.4655245796978082, "grad_norm": 10.9296875, "learning_rate": 9.194464184195566e-06, "loss": 0.935, "step": 23171 }, { "epoch": 2.4656309853160248, "grad_norm": 13.0625, "learning_rate": 9.194428652643547e-06, "loss": 0.7153, "step": 23172 }, { "epoch": 2.4657373909342413, "grad_norm": 10.7421875, "learning_rate": 9.19439312109153e-06, "loss": 0.8154, "step": 23173 }, { "epoch": 2.465843796552458, "grad_norm": 16.21875, "learning_rate": 9.194357589539512e-06, "loss": 0.9414, "step": 23174 }, { "epoch": 2.4659502021706747, "grad_norm": 13.3984375, "learning_rate": 9.194322057987494e-06, "loss": 0.9055, "step": 23175 }, { "epoch": 2.4660566077888912, "grad_norm": 12.234375, "learning_rate": 9.194286526435476e-06, "loss": 0.8278, "step": 23176 }, { "epoch": 2.4661630134071078, "grad_norm": 12.703125, "learning_rate": 9.194250994883457e-06, "loss": 0.7656, "step": 23177 }, { "epoch": 2.4662694190253247, "grad_norm": 14.1015625, "learning_rate": 9.194215463331439e-06, "loss": 0.8025, "step": 23178 }, { "epoch": 2.466375824643541, "grad_norm": 11.4921875, "learning_rate": 9.19417993177942e-06, "loss": 0.8041, "step": 23179 }, { "epoch": 2.4664822302617577, "grad_norm": 11.8984375, "learning_rate": 9.194144400227402e-06, "loss": 0.9766, "step": 23180 }, { "epoch": 2.4665886358799742, "grad_norm": 13.359375, "learning_rate": 9.194108868675384e-06, "loss": 0.8083, "step": 23181 }, { "epoch": 2.466695041498191, "grad_norm": 13.9453125, "learning_rate": 9.194073337123365e-06, "loss": 0.8812, "step": 23182 }, { "epoch": 2.4668014471164077, "grad_norm": 12.171875, "learning_rate": 9.194037805571349e-06, "loss": 0.7335, "step": 23183 }, { "epoch": 2.466907852734624, "grad_norm": 15.765625, "learning_rate": 9.19400227401933e-06, "loss": 1.0404, "step": 23184 }, { "epoch": 2.467014258352841, "grad_norm": 10.5390625, "learning_rate": 9.193966742467312e-06, "loss": 0.9318, "step": 23185 }, { "epoch": 2.4671206639710577, "grad_norm": 13.2890625, "learning_rate": 9.193931210915293e-06, "loss": 0.9361, "step": 23186 }, { "epoch": 2.467227069589274, "grad_norm": 11.46875, "learning_rate": 9.193895679363275e-06, "loss": 0.9626, "step": 23187 }, { "epoch": 2.467333475207491, "grad_norm": 15.6796875, "learning_rate": 9.193860147811258e-06, "loss": 0.8985, "step": 23188 }, { "epoch": 2.4674398808257076, "grad_norm": 13.421875, "learning_rate": 9.19382461625924e-06, "loss": 0.8097, "step": 23189 }, { "epoch": 2.467546286443924, "grad_norm": 13.21875, "learning_rate": 9.193789084707221e-06, "loss": 1.0178, "step": 23190 }, { "epoch": 2.467652692062141, "grad_norm": 12.734375, "learning_rate": 9.193753553155203e-06, "loss": 0.9188, "step": 23191 }, { "epoch": 2.4677590976803576, "grad_norm": 11.578125, "learning_rate": 9.193718021603185e-06, "loss": 0.9509, "step": 23192 }, { "epoch": 2.467865503298574, "grad_norm": 10.1171875, "learning_rate": 9.193682490051166e-06, "loss": 0.8807, "step": 23193 }, { "epoch": 2.4679719089167906, "grad_norm": 9.3828125, "learning_rate": 9.193646958499148e-06, "loss": 0.8992, "step": 23194 }, { "epoch": 2.4680783145350076, "grad_norm": 13.1484375, "learning_rate": 9.19361142694713e-06, "loss": 0.9734, "step": 23195 }, { "epoch": 2.468184720153224, "grad_norm": 11.40625, "learning_rate": 9.193575895395111e-06, "loss": 1.0255, "step": 23196 }, { "epoch": 2.4682911257714406, "grad_norm": 14.9296875, "learning_rate": 9.193540363843093e-06, "loss": 1.132, "step": 23197 }, { "epoch": 2.4683975313896576, "grad_norm": 10.7890625, "learning_rate": 9.193504832291074e-06, "loss": 0.849, "step": 23198 }, { "epoch": 2.468503937007874, "grad_norm": 18.046875, "learning_rate": 9.193469300739058e-06, "loss": 0.9024, "step": 23199 }, { "epoch": 2.4686103426260906, "grad_norm": 10.6328125, "learning_rate": 9.19343376918704e-06, "loss": 0.8197, "step": 23200 }, { "epoch": 2.468716748244307, "grad_norm": 12.7890625, "learning_rate": 9.19339823763502e-06, "loss": 0.9941, "step": 23201 }, { "epoch": 2.468823153862524, "grad_norm": 12.9921875, "learning_rate": 9.193362706083002e-06, "loss": 0.8639, "step": 23202 }, { "epoch": 2.4689295594807406, "grad_norm": 18.375, "learning_rate": 9.193327174530984e-06, "loss": 1.0293, "step": 23203 }, { "epoch": 2.469035965098957, "grad_norm": 12.234375, "learning_rate": 9.193291642978967e-06, "loss": 0.9894, "step": 23204 }, { "epoch": 2.469142370717174, "grad_norm": 8.8671875, "learning_rate": 9.193256111426947e-06, "loss": 0.7856, "step": 23205 }, { "epoch": 2.4692487763353905, "grad_norm": 10.5234375, "learning_rate": 9.193220579874929e-06, "loss": 0.8297, "step": 23206 }, { "epoch": 2.469355181953607, "grad_norm": 12.765625, "learning_rate": 9.19318504832291e-06, "loss": 0.8461, "step": 23207 }, { "epoch": 2.469461587571824, "grad_norm": 12.53125, "learning_rate": 9.193149516770892e-06, "loss": 1.0305, "step": 23208 }, { "epoch": 2.4695679931900405, "grad_norm": 9.0703125, "learning_rate": 9.193113985218875e-06, "loss": 0.778, "step": 23209 }, { "epoch": 2.469674398808257, "grad_norm": 12.8515625, "learning_rate": 9.193078453666857e-06, "loss": 0.9368, "step": 23210 }, { "epoch": 2.469780804426474, "grad_norm": 15.9609375, "learning_rate": 9.193042922114839e-06, "loss": 0.931, "step": 23211 }, { "epoch": 2.4698872100446905, "grad_norm": 15.4375, "learning_rate": 9.19300739056282e-06, "loss": 1.0259, "step": 23212 }, { "epoch": 2.469993615662907, "grad_norm": 12.328125, "learning_rate": 9.192971859010802e-06, "loss": 0.959, "step": 23213 }, { "epoch": 2.4701000212811235, "grad_norm": 11.890625, "learning_rate": 9.192936327458785e-06, "loss": 1.0871, "step": 23214 }, { "epoch": 2.4702064268993404, "grad_norm": 12.2421875, "learning_rate": 9.192900795906767e-06, "loss": 0.9701, "step": 23215 }, { "epoch": 2.470312832517557, "grad_norm": 15.15625, "learning_rate": 9.192865264354748e-06, "loss": 1.0864, "step": 23216 }, { "epoch": 2.4704192381357735, "grad_norm": 11.671875, "learning_rate": 9.19282973280273e-06, "loss": 0.7732, "step": 23217 }, { "epoch": 2.47052564375399, "grad_norm": 13.125, "learning_rate": 9.192794201250712e-06, "loss": 0.7901, "step": 23218 }, { "epoch": 2.470632049372207, "grad_norm": 12.140625, "learning_rate": 9.192758669698693e-06, "loss": 0.8008, "step": 23219 }, { "epoch": 2.4707384549904234, "grad_norm": 14.5625, "learning_rate": 9.192723138146675e-06, "loss": 0.7251, "step": 23220 }, { "epoch": 2.47084486060864, "grad_norm": 14.2109375, "learning_rate": 9.192687606594656e-06, "loss": 0.8016, "step": 23221 }, { "epoch": 2.470951266226857, "grad_norm": 13.78125, "learning_rate": 9.192652075042638e-06, "loss": 0.8613, "step": 23222 }, { "epoch": 2.4710576718450734, "grad_norm": 15.5390625, "learning_rate": 9.19261654349062e-06, "loss": 0.8402, "step": 23223 }, { "epoch": 2.47116407746329, "grad_norm": 12.7578125, "learning_rate": 9.192581011938603e-06, "loss": 1.0288, "step": 23224 }, { "epoch": 2.471270483081507, "grad_norm": 14.015625, "learning_rate": 9.192545480386584e-06, "loss": 0.9244, "step": 23225 }, { "epoch": 2.4713768886997234, "grad_norm": 14.796875, "learning_rate": 9.192509948834566e-06, "loss": 0.8981, "step": 23226 }, { "epoch": 2.47148329431794, "grad_norm": 10.6953125, "learning_rate": 9.192474417282548e-06, "loss": 0.9215, "step": 23227 }, { "epoch": 2.471589699936157, "grad_norm": 13.5078125, "learning_rate": 9.19243888573053e-06, "loss": 0.9813, "step": 23228 }, { "epoch": 2.4716961055543734, "grad_norm": 10.2890625, "learning_rate": 9.192403354178513e-06, "loss": 0.9316, "step": 23229 }, { "epoch": 2.47180251117259, "grad_norm": 13.8125, "learning_rate": 9.192367822626494e-06, "loss": 0.9896, "step": 23230 }, { "epoch": 2.4719089167908064, "grad_norm": 14.6171875, "learning_rate": 9.192332291074476e-06, "loss": 0.904, "step": 23231 }, { "epoch": 2.4720153224090233, "grad_norm": 19.265625, "learning_rate": 9.192296759522457e-06, "loss": 0.8984, "step": 23232 }, { "epoch": 2.47212172802724, "grad_norm": 9.8828125, "learning_rate": 9.192261227970439e-06, "loss": 0.8525, "step": 23233 }, { "epoch": 2.4722281336454563, "grad_norm": 17.46875, "learning_rate": 9.19222569641842e-06, "loss": 0.9064, "step": 23234 }, { "epoch": 2.4723345392636733, "grad_norm": 12.3671875, "learning_rate": 9.192190164866402e-06, "loss": 0.9613, "step": 23235 }, { "epoch": 2.47244094488189, "grad_norm": 14.1875, "learning_rate": 9.192154633314384e-06, "loss": 0.8021, "step": 23236 }, { "epoch": 2.4725473505001063, "grad_norm": 11.5859375, "learning_rate": 9.192119101762365e-06, "loss": 0.8018, "step": 23237 }, { "epoch": 2.472653756118323, "grad_norm": 13.8203125, "learning_rate": 9.192083570210347e-06, "loss": 1.0131, "step": 23238 }, { "epoch": 2.47276016173654, "grad_norm": 9.625, "learning_rate": 9.192048038658329e-06, "loss": 0.8063, "step": 23239 }, { "epoch": 2.4728665673547563, "grad_norm": 11.6796875, "learning_rate": 9.192012507106312e-06, "loss": 0.7623, "step": 23240 }, { "epoch": 2.472972972972973, "grad_norm": 16.5, "learning_rate": 9.191976975554294e-06, "loss": 0.6638, "step": 23241 }, { "epoch": 2.4730793785911898, "grad_norm": 19.53125, "learning_rate": 9.191941444002275e-06, "loss": 0.892, "step": 23242 }, { "epoch": 2.4731857842094063, "grad_norm": 14.625, "learning_rate": 9.191905912450257e-06, "loss": 0.8615, "step": 23243 }, { "epoch": 2.4732921898276228, "grad_norm": 11.5234375, "learning_rate": 9.191870380898238e-06, "loss": 0.9157, "step": 23244 }, { "epoch": 2.4733985954458397, "grad_norm": 15.0, "learning_rate": 9.19183484934622e-06, "loss": 0.9191, "step": 23245 }, { "epoch": 2.4735050010640562, "grad_norm": 13.734375, "learning_rate": 9.191799317794202e-06, "loss": 0.9679, "step": 23246 }, { "epoch": 2.4736114066822728, "grad_norm": 11.53125, "learning_rate": 9.191763786242183e-06, "loss": 0.9021, "step": 23247 }, { "epoch": 2.4737178123004897, "grad_norm": 10.625, "learning_rate": 9.191728254690165e-06, "loss": 0.8542, "step": 23248 }, { "epoch": 2.473824217918706, "grad_norm": 13.109375, "learning_rate": 9.191692723138146e-06, "loss": 0.786, "step": 23249 }, { "epoch": 2.4739306235369227, "grad_norm": 16.25, "learning_rate": 9.19165719158613e-06, "loss": 0.9652, "step": 23250 }, { "epoch": 2.4740370291551392, "grad_norm": 18.375, "learning_rate": 9.191621660034111e-06, "loss": 1.1939, "step": 23251 }, { "epoch": 2.474143434773356, "grad_norm": 13.3125, "learning_rate": 9.191586128482093e-06, "loss": 0.857, "step": 23252 }, { "epoch": 2.4742498403915727, "grad_norm": 14.2578125, "learning_rate": 9.191550596930075e-06, "loss": 0.9995, "step": 23253 }, { "epoch": 2.474356246009789, "grad_norm": 11.6796875, "learning_rate": 9.191515065378056e-06, "loss": 0.8868, "step": 23254 }, { "epoch": 2.4744626516280057, "grad_norm": 11.5625, "learning_rate": 9.19147953382604e-06, "loss": 0.9334, "step": 23255 }, { "epoch": 2.4745690572462227, "grad_norm": 12.9921875, "learning_rate": 9.191444002274021e-06, "loss": 0.8655, "step": 23256 }, { "epoch": 2.474675462864439, "grad_norm": 10.6015625, "learning_rate": 9.191408470722003e-06, "loss": 0.7626, "step": 23257 }, { "epoch": 2.4747818684826557, "grad_norm": 14.59375, "learning_rate": 9.191372939169984e-06, "loss": 0.8669, "step": 23258 }, { "epoch": 2.4748882741008726, "grad_norm": 12.53125, "learning_rate": 9.191337407617966e-06, "loss": 0.8557, "step": 23259 }, { "epoch": 2.474994679719089, "grad_norm": 11.953125, "learning_rate": 9.191301876065947e-06, "loss": 0.877, "step": 23260 }, { "epoch": 2.4751010853373057, "grad_norm": 15.03125, "learning_rate": 9.191266344513929e-06, "loss": 0.6972, "step": 23261 }, { "epoch": 2.4752074909555226, "grad_norm": 18.28125, "learning_rate": 9.19123081296191e-06, "loss": 0.9435, "step": 23262 }, { "epoch": 2.475313896573739, "grad_norm": 12.296875, "learning_rate": 9.191195281409892e-06, "loss": 0.7873, "step": 23263 }, { "epoch": 2.4754203021919556, "grad_norm": 12.5859375, "learning_rate": 9.191159749857874e-06, "loss": 0.8526, "step": 23264 }, { "epoch": 2.4755267078101726, "grad_norm": 12.625, "learning_rate": 9.191124218305857e-06, "loss": 0.9137, "step": 23265 }, { "epoch": 2.475633113428389, "grad_norm": 14.8203125, "learning_rate": 9.191088686753839e-06, "loss": 0.8132, "step": 23266 }, { "epoch": 2.4757395190466056, "grad_norm": 9.8125, "learning_rate": 9.19105315520182e-06, "loss": 0.7472, "step": 23267 }, { "epoch": 2.475845924664822, "grad_norm": 15.6796875, "learning_rate": 9.191017623649802e-06, "loss": 0.8991, "step": 23268 }, { "epoch": 2.475952330283039, "grad_norm": 12.8828125, "learning_rate": 9.190982092097784e-06, "loss": 0.7582, "step": 23269 }, { "epoch": 2.4760587359012556, "grad_norm": 20.015625, "learning_rate": 9.190946560545765e-06, "loss": 1.0084, "step": 23270 }, { "epoch": 2.476165141519472, "grad_norm": 13.90625, "learning_rate": 9.190911028993747e-06, "loss": 0.952, "step": 23271 }, { "epoch": 2.476271547137689, "grad_norm": 12.8125, "learning_rate": 9.190875497441728e-06, "loss": 1.027, "step": 23272 }, { "epoch": 2.4763779527559056, "grad_norm": 11.484375, "learning_rate": 9.19083996588971e-06, "loss": 0.8854, "step": 23273 }, { "epoch": 2.476484358374122, "grad_norm": 11.890625, "learning_rate": 9.190804434337692e-06, "loss": 0.9608, "step": 23274 }, { "epoch": 2.4765907639923386, "grad_norm": 10.828125, "learning_rate": 9.190768902785673e-06, "loss": 0.829, "step": 23275 }, { "epoch": 2.4766971696105555, "grad_norm": 13.2578125, "learning_rate": 9.190733371233657e-06, "loss": 0.9887, "step": 23276 }, { "epoch": 2.476803575228772, "grad_norm": 13.0703125, "learning_rate": 9.190697839681638e-06, "loss": 0.6472, "step": 23277 }, { "epoch": 2.4769099808469885, "grad_norm": 11.109375, "learning_rate": 9.19066230812962e-06, "loss": 0.8451, "step": 23278 }, { "epoch": 2.4770163864652055, "grad_norm": 11.90625, "learning_rate": 9.190626776577601e-06, "loss": 0.8904, "step": 23279 }, { "epoch": 2.477122792083422, "grad_norm": 11.3046875, "learning_rate": 9.190591245025583e-06, "loss": 0.8694, "step": 23280 }, { "epoch": 2.4772291977016385, "grad_norm": 13.609375, "learning_rate": 9.190555713473566e-06, "loss": 0.7232, "step": 23281 }, { "epoch": 2.4773356033198555, "grad_norm": 18.390625, "learning_rate": 9.190520181921548e-06, "loss": 0.989, "step": 23282 }, { "epoch": 2.477442008938072, "grad_norm": 14.5390625, "learning_rate": 9.19048465036953e-06, "loss": 0.9337, "step": 23283 }, { "epoch": 2.4775484145562885, "grad_norm": 14.2109375, "learning_rate": 9.190449118817511e-06, "loss": 0.8921, "step": 23284 }, { "epoch": 2.4776548201745054, "grad_norm": 12.5625, "learning_rate": 9.190413587265493e-06, "loss": 1.0236, "step": 23285 }, { "epoch": 2.477761225792722, "grad_norm": 10.2734375, "learning_rate": 9.190378055713474e-06, "loss": 0.8197, "step": 23286 }, { "epoch": 2.4778676314109385, "grad_norm": 12.59375, "learning_rate": 9.190342524161456e-06, "loss": 1.0143, "step": 23287 }, { "epoch": 2.477974037029155, "grad_norm": 13.1484375, "learning_rate": 9.190306992609438e-06, "loss": 0.9127, "step": 23288 }, { "epoch": 2.478080442647372, "grad_norm": 12.1796875, "learning_rate": 9.190271461057419e-06, "loss": 1.0044, "step": 23289 }, { "epoch": 2.4781868482655884, "grad_norm": 10.4609375, "learning_rate": 9.1902359295054e-06, "loss": 0.9304, "step": 23290 }, { "epoch": 2.478293253883805, "grad_norm": 17.515625, "learning_rate": 9.190200397953384e-06, "loss": 1.1348, "step": 23291 }, { "epoch": 2.4783996595020215, "grad_norm": 15.28125, "learning_rate": 9.190164866401366e-06, "loss": 0.7828, "step": 23292 }, { "epoch": 2.4785060651202384, "grad_norm": 14.8203125, "learning_rate": 9.190129334849347e-06, "loss": 0.9147, "step": 23293 }, { "epoch": 2.478612470738455, "grad_norm": 13.9609375, "learning_rate": 9.190093803297329e-06, "loss": 0.9619, "step": 23294 }, { "epoch": 2.4787188763566714, "grad_norm": 17.46875, "learning_rate": 9.19005827174531e-06, "loss": 1.0003, "step": 23295 }, { "epoch": 2.4788252819748884, "grad_norm": 14.8515625, "learning_rate": 9.190022740193292e-06, "loss": 0.8931, "step": 23296 }, { "epoch": 2.478931687593105, "grad_norm": 12.625, "learning_rate": 9.189987208641274e-06, "loss": 0.9226, "step": 23297 }, { "epoch": 2.4790380932113214, "grad_norm": 11.296875, "learning_rate": 9.189951677089255e-06, "loss": 0.9517, "step": 23298 }, { "epoch": 2.4791444988295384, "grad_norm": 13.09375, "learning_rate": 9.189916145537237e-06, "loss": 0.7958, "step": 23299 }, { "epoch": 2.479250904447755, "grad_norm": 12.8203125, "learning_rate": 9.189880613985219e-06, "loss": 0.9276, "step": 23300 }, { "epoch": 2.4793573100659714, "grad_norm": 14.390625, "learning_rate": 9.189845082433202e-06, "loss": 0.7888, "step": 23301 }, { "epoch": 2.4794637156841883, "grad_norm": 10.75, "learning_rate": 9.189809550881183e-06, "loss": 0.9764, "step": 23302 }, { "epoch": 2.479570121302405, "grad_norm": 10.0703125, "learning_rate": 9.189774019329165e-06, "loss": 0.7755, "step": 23303 }, { "epoch": 2.4796765269206213, "grad_norm": 10.8203125, "learning_rate": 9.189738487777147e-06, "loss": 0.828, "step": 23304 }, { "epoch": 2.479782932538838, "grad_norm": 15.296875, "learning_rate": 9.189702956225128e-06, "loss": 1.0812, "step": 23305 }, { "epoch": 2.479889338157055, "grad_norm": 10.875, "learning_rate": 9.189667424673112e-06, "loss": 0.8939, "step": 23306 }, { "epoch": 2.4799957437752713, "grad_norm": 12.265625, "learning_rate": 9.189631893121093e-06, "loss": 0.8205, "step": 23307 }, { "epoch": 2.480102149393488, "grad_norm": 13.359375, "learning_rate": 9.189596361569075e-06, "loss": 0.9296, "step": 23308 }, { "epoch": 2.480208555011705, "grad_norm": 13.734375, "learning_rate": 9.189560830017056e-06, "loss": 0.934, "step": 23309 }, { "epoch": 2.4803149606299213, "grad_norm": 12.8046875, "learning_rate": 9.189525298465038e-06, "loss": 0.822, "step": 23310 }, { "epoch": 2.480421366248138, "grad_norm": 12.203125, "learning_rate": 9.18948976691302e-06, "loss": 0.8226, "step": 23311 }, { "epoch": 2.4805277718663543, "grad_norm": 15.7578125, "learning_rate": 9.189454235361001e-06, "loss": 0.727, "step": 23312 }, { "epoch": 2.4806341774845713, "grad_norm": 12.3359375, "learning_rate": 9.189418703808983e-06, "loss": 0.9642, "step": 23313 }, { "epoch": 2.4807405831027878, "grad_norm": 13.5, "learning_rate": 9.189383172256964e-06, "loss": 0.9297, "step": 23314 }, { "epoch": 2.4808469887210043, "grad_norm": 16.9375, "learning_rate": 9.189347640704946e-06, "loss": 0.8393, "step": 23315 }, { "epoch": 2.4809533943392212, "grad_norm": 14.296875, "learning_rate": 9.189312109152928e-06, "loss": 1.1312, "step": 23316 }, { "epoch": 2.4810597999574378, "grad_norm": 13.359375, "learning_rate": 9.189276577600911e-06, "loss": 0.8048, "step": 23317 }, { "epoch": 2.4811662055756543, "grad_norm": 15.1875, "learning_rate": 9.189241046048893e-06, "loss": 0.8022, "step": 23318 }, { "epoch": 2.481272611193871, "grad_norm": 11.75, "learning_rate": 9.189205514496874e-06, "loss": 0.748, "step": 23319 }, { "epoch": 2.4813790168120877, "grad_norm": 11.921875, "learning_rate": 9.189169982944856e-06, "loss": 0.8277, "step": 23320 }, { "epoch": 2.4814854224303042, "grad_norm": 12.2734375, "learning_rate": 9.189134451392837e-06, "loss": 0.9186, "step": 23321 }, { "epoch": 2.481591828048521, "grad_norm": 9.703125, "learning_rate": 9.18909891984082e-06, "loss": 0.7666, "step": 23322 }, { "epoch": 2.4816982336667377, "grad_norm": 9.53125, "learning_rate": 9.189063388288802e-06, "loss": 0.8429, "step": 23323 }, { "epoch": 2.481804639284954, "grad_norm": 10.15625, "learning_rate": 9.189027856736782e-06, "loss": 0.8549, "step": 23324 }, { "epoch": 2.4819110449031707, "grad_norm": 12.2421875, "learning_rate": 9.188992325184764e-06, "loss": 0.8573, "step": 23325 }, { "epoch": 2.4820174505213877, "grad_norm": 13.96875, "learning_rate": 9.188956793632745e-06, "loss": 0.9829, "step": 23326 }, { "epoch": 2.482123856139604, "grad_norm": 9.9609375, "learning_rate": 9.188921262080729e-06, "loss": 0.8148, "step": 23327 }, { "epoch": 2.4822302617578207, "grad_norm": 11.7265625, "learning_rate": 9.18888573052871e-06, "loss": 0.8353, "step": 23328 }, { "epoch": 2.4823366673760376, "grad_norm": 15.7734375, "learning_rate": 9.188850198976692e-06, "loss": 0.8824, "step": 23329 }, { "epoch": 2.482443072994254, "grad_norm": 14.546875, "learning_rate": 9.188814667424673e-06, "loss": 0.7205, "step": 23330 }, { "epoch": 2.4825494786124707, "grad_norm": 12.515625, "learning_rate": 9.188779135872655e-06, "loss": 0.9674, "step": 23331 }, { "epoch": 2.482655884230687, "grad_norm": 15.265625, "learning_rate": 9.188743604320638e-06, "loss": 1.0462, "step": 23332 }, { "epoch": 2.482762289848904, "grad_norm": 18.234375, "learning_rate": 9.18870807276862e-06, "loss": 0.9827, "step": 23333 }, { "epoch": 2.4828686954671206, "grad_norm": 13.9921875, "learning_rate": 9.188672541216602e-06, "loss": 0.8938, "step": 23334 }, { "epoch": 2.482975101085337, "grad_norm": 17.234375, "learning_rate": 9.188637009664583e-06, "loss": 0.9173, "step": 23335 }, { "epoch": 2.483081506703554, "grad_norm": 11.2734375, "learning_rate": 9.188601478112565e-06, "loss": 0.8574, "step": 23336 }, { "epoch": 2.4831879123217706, "grad_norm": 13.5703125, "learning_rate": 9.188565946560546e-06, "loss": 0.9874, "step": 23337 }, { "epoch": 2.483294317939987, "grad_norm": 11.6953125, "learning_rate": 9.188530415008528e-06, "loss": 0.8257, "step": 23338 }, { "epoch": 2.483400723558204, "grad_norm": 10.34375, "learning_rate": 9.18849488345651e-06, "loss": 0.9101, "step": 23339 }, { "epoch": 2.4835071291764206, "grad_norm": 19.578125, "learning_rate": 9.188459351904491e-06, "loss": 1.0806, "step": 23340 }, { "epoch": 2.483613534794637, "grad_norm": 11.9296875, "learning_rate": 9.188423820352473e-06, "loss": 1.0006, "step": 23341 }, { "epoch": 2.483719940412854, "grad_norm": 15.5078125, "learning_rate": 9.188388288800456e-06, "loss": 1.0552, "step": 23342 }, { "epoch": 2.4838263460310706, "grad_norm": 11.984375, "learning_rate": 9.188352757248438e-06, "loss": 0.9627, "step": 23343 }, { "epoch": 2.483932751649287, "grad_norm": 16.15625, "learning_rate": 9.18831722569642e-06, "loss": 0.8565, "step": 23344 }, { "epoch": 2.4840391572675036, "grad_norm": 14.7109375, "learning_rate": 9.188281694144401e-06, "loss": 0.9015, "step": 23345 }, { "epoch": 2.4841455628857205, "grad_norm": 12.34375, "learning_rate": 9.188246162592383e-06, "loss": 0.9684, "step": 23346 }, { "epoch": 2.484251968503937, "grad_norm": 10.9453125, "learning_rate": 9.188210631040366e-06, "loss": 0.806, "step": 23347 }, { "epoch": 2.4843583741221535, "grad_norm": 10.6796875, "learning_rate": 9.188175099488347e-06, "loss": 0.855, "step": 23348 }, { "epoch": 2.48446477974037, "grad_norm": 27.21875, "learning_rate": 9.188139567936329e-06, "loss": 0.8525, "step": 23349 }, { "epoch": 2.484571185358587, "grad_norm": 13.390625, "learning_rate": 9.18810403638431e-06, "loss": 0.9715, "step": 23350 }, { "epoch": 2.4846775909768035, "grad_norm": 16.640625, "learning_rate": 9.188068504832292e-06, "loss": 0.9772, "step": 23351 }, { "epoch": 2.48478399659502, "grad_norm": 13.40625, "learning_rate": 9.188032973280274e-06, "loss": 1.0699, "step": 23352 }, { "epoch": 2.484890402213237, "grad_norm": 11.3828125, "learning_rate": 9.187997441728256e-06, "loss": 0.7286, "step": 23353 }, { "epoch": 2.4849968078314535, "grad_norm": 12.5859375, "learning_rate": 9.187961910176237e-06, "loss": 0.8034, "step": 23354 }, { "epoch": 2.48510321344967, "grad_norm": 17.75, "learning_rate": 9.187926378624219e-06, "loss": 0.905, "step": 23355 }, { "epoch": 2.485209619067887, "grad_norm": 10.859375, "learning_rate": 9.1878908470722e-06, "loss": 0.7922, "step": 23356 }, { "epoch": 2.4853160246861035, "grad_norm": 13.8359375, "learning_rate": 9.187855315520182e-06, "loss": 0.8129, "step": 23357 }, { "epoch": 2.48542243030432, "grad_norm": 21.78125, "learning_rate": 9.187819783968165e-06, "loss": 0.9801, "step": 23358 }, { "epoch": 2.485528835922537, "grad_norm": 11.734375, "learning_rate": 9.187784252416147e-06, "loss": 0.9419, "step": 23359 }, { "epoch": 2.4856352415407534, "grad_norm": 14.65625, "learning_rate": 9.187748720864128e-06, "loss": 0.9997, "step": 23360 }, { "epoch": 2.48574164715897, "grad_norm": 245.125, "learning_rate": 9.18771318931211e-06, "loss": 0.8949, "step": 23361 }, { "epoch": 2.4858480527771865, "grad_norm": 13.8984375, "learning_rate": 9.187677657760092e-06, "loss": 0.7277, "step": 23362 }, { "epoch": 2.4859544583954034, "grad_norm": 16.109375, "learning_rate": 9.187642126208073e-06, "loss": 1.0484, "step": 23363 }, { "epoch": 2.48606086401362, "grad_norm": 11.6953125, "learning_rate": 9.187606594656055e-06, "loss": 0.9616, "step": 23364 }, { "epoch": 2.4861672696318364, "grad_norm": 10.7421875, "learning_rate": 9.187571063104036e-06, "loss": 0.8235, "step": 23365 }, { "epoch": 2.4862736752500534, "grad_norm": 14.921875, "learning_rate": 9.187535531552018e-06, "loss": 0.8149, "step": 23366 }, { "epoch": 2.48638008086827, "grad_norm": 16.75, "learning_rate": 9.1875e-06, "loss": 1.0824, "step": 23367 }, { "epoch": 2.4864864864864864, "grad_norm": 12.7578125, "learning_rate": 9.187464468447983e-06, "loss": 0.8482, "step": 23368 }, { "epoch": 2.486592892104703, "grad_norm": 9.859375, "learning_rate": 9.187428936895965e-06, "loss": 0.8086, "step": 23369 }, { "epoch": 2.48669929772292, "grad_norm": 16.234375, "learning_rate": 9.187393405343946e-06, "loss": 0.7765, "step": 23370 }, { "epoch": 2.4868057033411364, "grad_norm": 15.5390625, "learning_rate": 9.187357873791928e-06, "loss": 1.074, "step": 23371 }, { "epoch": 2.486912108959353, "grad_norm": 14.234375, "learning_rate": 9.18732234223991e-06, "loss": 0.8597, "step": 23372 }, { "epoch": 2.48701851457757, "grad_norm": 13.6328125, "learning_rate": 9.187286810687893e-06, "loss": 0.8471, "step": 23373 }, { "epoch": 2.4871249201957863, "grad_norm": 13.6953125, "learning_rate": 9.187251279135874e-06, "loss": 0.8288, "step": 23374 }, { "epoch": 2.487231325814003, "grad_norm": 14.0390625, "learning_rate": 9.187215747583856e-06, "loss": 0.7899, "step": 23375 }, { "epoch": 2.48733773143222, "grad_norm": 11.890625, "learning_rate": 9.187180216031838e-06, "loss": 0.8449, "step": 23376 }, { "epoch": 2.4874441370504363, "grad_norm": 13.234375, "learning_rate": 9.18714468447982e-06, "loss": 0.9785, "step": 23377 }, { "epoch": 2.487550542668653, "grad_norm": 12.1953125, "learning_rate": 9.1871091529278e-06, "loss": 0.7193, "step": 23378 }, { "epoch": 2.48765694828687, "grad_norm": 10.9375, "learning_rate": 9.187073621375782e-06, "loss": 0.7791, "step": 23379 }, { "epoch": 2.4877633539050863, "grad_norm": 14.5546875, "learning_rate": 9.187038089823764e-06, "loss": 0.9585, "step": 23380 }, { "epoch": 2.487869759523303, "grad_norm": 15.046875, "learning_rate": 9.187002558271746e-06, "loss": 0.8088, "step": 23381 }, { "epoch": 2.4879761651415193, "grad_norm": 13.65625, "learning_rate": 9.186967026719727e-06, "loss": 0.9019, "step": 23382 }, { "epoch": 2.4880825707597363, "grad_norm": 12.8984375, "learning_rate": 9.18693149516771e-06, "loss": 0.8951, "step": 23383 }, { "epoch": 2.4881889763779528, "grad_norm": 12.0390625, "learning_rate": 9.186895963615692e-06, "loss": 0.9448, "step": 23384 }, { "epoch": 2.4882953819961693, "grad_norm": 13.0546875, "learning_rate": 9.186860432063674e-06, "loss": 0.8591, "step": 23385 }, { "epoch": 2.488401787614386, "grad_norm": 12.7109375, "learning_rate": 9.186824900511655e-06, "loss": 1.0599, "step": 23386 }, { "epoch": 2.4885081932326027, "grad_norm": 15.4921875, "learning_rate": 9.186789368959637e-06, "loss": 0.8291, "step": 23387 }, { "epoch": 2.4886145988508193, "grad_norm": 13.03125, "learning_rate": 9.186753837407619e-06, "loss": 1.079, "step": 23388 }, { "epoch": 2.4887210044690358, "grad_norm": 13.25, "learning_rate": 9.1867183058556e-06, "loss": 0.8971, "step": 23389 }, { "epoch": 2.4888274100872527, "grad_norm": 19.5, "learning_rate": 9.186682774303582e-06, "loss": 0.9957, "step": 23390 }, { "epoch": 2.4889338157054692, "grad_norm": 10.7265625, "learning_rate": 9.186647242751563e-06, "loss": 0.6356, "step": 23391 }, { "epoch": 2.4890402213236857, "grad_norm": 14.203125, "learning_rate": 9.186611711199545e-06, "loss": 0.9285, "step": 23392 }, { "epoch": 2.4891466269419027, "grad_norm": 14.0, "learning_rate": 9.186576179647527e-06, "loss": 1.0585, "step": 23393 }, { "epoch": 2.489253032560119, "grad_norm": 13.828125, "learning_rate": 9.18654064809551e-06, "loss": 0.9956, "step": 23394 }, { "epoch": 2.4893594381783357, "grad_norm": 11.171875, "learning_rate": 9.186505116543491e-06, "loss": 0.7049, "step": 23395 }, { "epoch": 2.4894658437965527, "grad_norm": 14.2109375, "learning_rate": 9.186469584991473e-06, "loss": 1.197, "step": 23396 }, { "epoch": 2.489572249414769, "grad_norm": 10.0234375, "learning_rate": 9.186434053439455e-06, "loss": 0.7924, "step": 23397 }, { "epoch": 2.4896786550329857, "grad_norm": 15.71875, "learning_rate": 9.186398521887436e-06, "loss": 0.7656, "step": 23398 }, { "epoch": 2.489785060651202, "grad_norm": 12.2109375, "learning_rate": 9.18636299033542e-06, "loss": 1.104, "step": 23399 }, { "epoch": 2.489891466269419, "grad_norm": 20.015625, "learning_rate": 9.186327458783401e-06, "loss": 0.8942, "step": 23400 }, { "epoch": 2.4899978718876357, "grad_norm": 16.75, "learning_rate": 9.186291927231383e-06, "loss": 0.9711, "step": 23401 }, { "epoch": 2.490104277505852, "grad_norm": 13.328125, "learning_rate": 9.186256395679364e-06, "loss": 1.0601, "step": 23402 }, { "epoch": 2.490210683124069, "grad_norm": 12.9453125, "learning_rate": 9.186220864127346e-06, "loss": 0.8914, "step": 23403 }, { "epoch": 2.4903170887422856, "grad_norm": 13.9296875, "learning_rate": 9.186185332575328e-06, "loss": 0.832, "step": 23404 }, { "epoch": 2.490423494360502, "grad_norm": 9.953125, "learning_rate": 9.18614980102331e-06, "loss": 0.8797, "step": 23405 }, { "epoch": 2.4905298999787187, "grad_norm": 12.0859375, "learning_rate": 9.18611426947129e-06, "loss": 0.9594, "step": 23406 }, { "epoch": 2.4906363055969356, "grad_norm": 14.109375, "learning_rate": 9.186078737919272e-06, "loss": 1.0512, "step": 23407 }, { "epoch": 2.490742711215152, "grad_norm": 12.4296875, "learning_rate": 9.186043206367254e-06, "loss": 0.9556, "step": 23408 }, { "epoch": 2.4908491168333686, "grad_norm": 14.234375, "learning_rate": 9.186007674815237e-06, "loss": 1.097, "step": 23409 }, { "epoch": 2.4909555224515856, "grad_norm": 12.7890625, "learning_rate": 9.185972143263219e-06, "loss": 0.8413, "step": 23410 }, { "epoch": 2.491061928069802, "grad_norm": 12.421875, "learning_rate": 9.1859366117112e-06, "loss": 1.008, "step": 23411 }, { "epoch": 2.4911683336880186, "grad_norm": 13.5625, "learning_rate": 9.185901080159182e-06, "loss": 0.8999, "step": 23412 }, { "epoch": 2.4912747393062356, "grad_norm": 28.3125, "learning_rate": 9.185865548607164e-06, "loss": 0.994, "step": 23413 }, { "epoch": 2.491381144924452, "grad_norm": 12.1796875, "learning_rate": 9.185830017055145e-06, "loss": 0.8876, "step": 23414 }, { "epoch": 2.4914875505426686, "grad_norm": 18.640625, "learning_rate": 9.185794485503127e-06, "loss": 0.866, "step": 23415 }, { "epoch": 2.4915939561608855, "grad_norm": 12.5625, "learning_rate": 9.185758953951109e-06, "loss": 0.8078, "step": 23416 }, { "epoch": 2.491700361779102, "grad_norm": 13.1484375, "learning_rate": 9.18572342239909e-06, "loss": 0.8986, "step": 23417 }, { "epoch": 2.4918067673973185, "grad_norm": 12.7578125, "learning_rate": 9.185687890847072e-06, "loss": 0.9512, "step": 23418 }, { "epoch": 2.491913173015535, "grad_norm": 14.828125, "learning_rate": 9.185652359295055e-06, "loss": 1.0605, "step": 23419 }, { "epoch": 2.492019578633752, "grad_norm": 17.15625, "learning_rate": 9.185616827743037e-06, "loss": 1.0614, "step": 23420 }, { "epoch": 2.4921259842519685, "grad_norm": 11.9453125, "learning_rate": 9.185581296191018e-06, "loss": 0.7479, "step": 23421 }, { "epoch": 2.492232389870185, "grad_norm": 10.7890625, "learning_rate": 9.185545764639e-06, "loss": 0.8427, "step": 23422 }, { "epoch": 2.492338795488402, "grad_norm": 17.25, "learning_rate": 9.185510233086982e-06, "loss": 0.9873, "step": 23423 }, { "epoch": 2.4924452011066185, "grad_norm": 18.140625, "learning_rate": 9.185474701534965e-06, "loss": 0.9336, "step": 23424 }, { "epoch": 2.492551606724835, "grad_norm": 19.796875, "learning_rate": 9.185439169982946e-06, "loss": 0.9587, "step": 23425 }, { "epoch": 2.4926580123430515, "grad_norm": 13.8359375, "learning_rate": 9.185403638430928e-06, "loss": 0.8816, "step": 23426 }, { "epoch": 2.4927644179612685, "grad_norm": 13.109375, "learning_rate": 9.18536810687891e-06, "loss": 0.9387, "step": 23427 }, { "epoch": 2.492870823579485, "grad_norm": 9.703125, "learning_rate": 9.185332575326891e-06, "loss": 0.7884, "step": 23428 }, { "epoch": 2.4929772291977015, "grad_norm": 12.15625, "learning_rate": 9.185297043774873e-06, "loss": 0.9058, "step": 23429 }, { "epoch": 2.4930836348159184, "grad_norm": 13.5625, "learning_rate": 9.185261512222854e-06, "loss": 0.8311, "step": 23430 }, { "epoch": 2.493190040434135, "grad_norm": 14.140625, "learning_rate": 9.185225980670836e-06, "loss": 0.8189, "step": 23431 }, { "epoch": 2.4932964460523515, "grad_norm": 16.0, "learning_rate": 9.185190449118818e-06, "loss": 1.0406, "step": 23432 }, { "epoch": 2.4934028516705684, "grad_norm": 14.7265625, "learning_rate": 9.1851549175668e-06, "loss": 1.0095, "step": 23433 }, { "epoch": 2.493509257288785, "grad_norm": 11.7109375, "learning_rate": 9.185119386014781e-06, "loss": 0.9327, "step": 23434 }, { "epoch": 2.4936156629070014, "grad_norm": 9.984375, "learning_rate": 9.185083854462764e-06, "loss": 0.8415, "step": 23435 }, { "epoch": 2.493722068525218, "grad_norm": 15.484375, "learning_rate": 9.185048322910746e-06, "loss": 0.9691, "step": 23436 }, { "epoch": 2.493828474143435, "grad_norm": 12.71875, "learning_rate": 9.185012791358727e-06, "loss": 0.8143, "step": 23437 }, { "epoch": 2.4939348797616514, "grad_norm": 14.578125, "learning_rate": 9.184977259806709e-06, "loss": 0.9004, "step": 23438 }, { "epoch": 2.494041285379868, "grad_norm": 12.390625, "learning_rate": 9.18494172825469e-06, "loss": 0.7157, "step": 23439 }, { "epoch": 2.494147690998085, "grad_norm": 12.859375, "learning_rate": 9.184906196702674e-06, "loss": 0.8755, "step": 23440 }, { "epoch": 2.4942540966163014, "grad_norm": 11.734375, "learning_rate": 9.184870665150656e-06, "loss": 0.9853, "step": 23441 }, { "epoch": 2.494360502234518, "grad_norm": 12.3671875, "learning_rate": 9.184835133598637e-06, "loss": 0.7181, "step": 23442 }, { "epoch": 2.4944669078527344, "grad_norm": 11.4921875, "learning_rate": 9.184799602046619e-06, "loss": 0.8341, "step": 23443 }, { "epoch": 2.4945733134709513, "grad_norm": 11.9765625, "learning_rate": 9.184764070494599e-06, "loss": 0.9795, "step": 23444 }, { "epoch": 2.494679719089168, "grad_norm": 15.703125, "learning_rate": 9.184728538942582e-06, "loss": 0.932, "step": 23445 }, { "epoch": 2.4947861247073844, "grad_norm": 22.1875, "learning_rate": 9.184693007390564e-06, "loss": 0.9101, "step": 23446 }, { "epoch": 2.4948925303256013, "grad_norm": 12.3125, "learning_rate": 9.184657475838545e-06, "loss": 0.8917, "step": 23447 }, { "epoch": 2.494998935943818, "grad_norm": 10.015625, "learning_rate": 9.184621944286527e-06, "loss": 0.802, "step": 23448 }, { "epoch": 2.4951053415620343, "grad_norm": 10.5859375, "learning_rate": 9.184586412734508e-06, "loss": 0.838, "step": 23449 }, { "epoch": 2.4952117471802513, "grad_norm": 14.6953125, "learning_rate": 9.184550881182492e-06, "loss": 0.8116, "step": 23450 }, { "epoch": 2.495318152798468, "grad_norm": 14.8828125, "learning_rate": 9.184515349630473e-06, "loss": 1.0425, "step": 23451 }, { "epoch": 2.4954245584166843, "grad_norm": 11.015625, "learning_rate": 9.184479818078455e-06, "loss": 0.7201, "step": 23452 }, { "epoch": 2.4955309640349013, "grad_norm": 13.59375, "learning_rate": 9.184444286526437e-06, "loss": 0.9895, "step": 23453 }, { "epoch": 2.4956373696531178, "grad_norm": 15.1796875, "learning_rate": 9.184408754974418e-06, "loss": 0.8862, "step": 23454 }, { "epoch": 2.4957437752713343, "grad_norm": 12.6171875, "learning_rate": 9.1843732234224e-06, "loss": 0.8636, "step": 23455 }, { "epoch": 2.495850180889551, "grad_norm": 11.90625, "learning_rate": 9.184337691870381e-06, "loss": 0.7381, "step": 23456 }, { "epoch": 2.4959565865077677, "grad_norm": 11.796875, "learning_rate": 9.184302160318363e-06, "loss": 0.7461, "step": 23457 }, { "epoch": 2.4960629921259843, "grad_norm": 15.5859375, "learning_rate": 9.184266628766345e-06, "loss": 0.7715, "step": 23458 }, { "epoch": 2.4961693977442008, "grad_norm": 16.59375, "learning_rate": 9.184231097214326e-06, "loss": 0.9982, "step": 23459 }, { "epoch": 2.4962758033624177, "grad_norm": 18.734375, "learning_rate": 9.18419556566231e-06, "loss": 0.8757, "step": 23460 }, { "epoch": 2.4963822089806342, "grad_norm": 10.453125, "learning_rate": 9.184160034110291e-06, "loss": 0.9245, "step": 23461 }, { "epoch": 2.4964886145988507, "grad_norm": 17.21875, "learning_rate": 9.184124502558273e-06, "loss": 0.9307, "step": 23462 }, { "epoch": 2.4965950202170673, "grad_norm": 12.75, "learning_rate": 9.184088971006254e-06, "loss": 0.7749, "step": 23463 }, { "epoch": 2.496701425835284, "grad_norm": 12.21875, "learning_rate": 9.184053439454236e-06, "loss": 0.7908, "step": 23464 }, { "epoch": 2.4968078314535007, "grad_norm": 12.015625, "learning_rate": 9.184017907902217e-06, "loss": 0.8824, "step": 23465 }, { "epoch": 2.4969142370717172, "grad_norm": 13.4453125, "learning_rate": 9.1839823763502e-06, "loss": 0.8302, "step": 23466 }, { "epoch": 2.497020642689934, "grad_norm": 12.6171875, "learning_rate": 9.183946844798182e-06, "loss": 0.8258, "step": 23467 }, { "epoch": 2.4971270483081507, "grad_norm": 18.34375, "learning_rate": 9.183911313246164e-06, "loss": 0.9529, "step": 23468 }, { "epoch": 2.497233453926367, "grad_norm": 14.4140625, "learning_rate": 9.183875781694146e-06, "loss": 0.832, "step": 23469 }, { "epoch": 2.497339859544584, "grad_norm": 11.765625, "learning_rate": 9.183840250142127e-06, "loss": 1.0142, "step": 23470 }, { "epoch": 2.4974462651628007, "grad_norm": 16.09375, "learning_rate": 9.183804718590109e-06, "loss": 1.0703, "step": 23471 }, { "epoch": 2.497552670781017, "grad_norm": 12.59375, "learning_rate": 9.18376918703809e-06, "loss": 0.7535, "step": 23472 }, { "epoch": 2.497659076399234, "grad_norm": 11.03125, "learning_rate": 9.183733655486072e-06, "loss": 0.8717, "step": 23473 }, { "epoch": 2.4977654820174506, "grad_norm": 15.59375, "learning_rate": 9.183698123934054e-06, "loss": 0.9731, "step": 23474 }, { "epoch": 2.497871887635667, "grad_norm": 13.9765625, "learning_rate": 9.183662592382035e-06, "loss": 0.9762, "step": 23475 }, { "epoch": 2.4979782932538837, "grad_norm": 9.6640625, "learning_rate": 9.183627060830019e-06, "loss": 0.8398, "step": 23476 }, { "epoch": 2.4980846988721006, "grad_norm": 17.078125, "learning_rate": 9.183591529278e-06, "loss": 1.0678, "step": 23477 }, { "epoch": 2.498191104490317, "grad_norm": 12.7109375, "learning_rate": 9.183555997725982e-06, "loss": 1.1046, "step": 23478 }, { "epoch": 2.4982975101085336, "grad_norm": 18.515625, "learning_rate": 9.183520466173963e-06, "loss": 0.9679, "step": 23479 }, { "epoch": 2.49840391572675, "grad_norm": 13.71875, "learning_rate": 9.183484934621945e-06, "loss": 0.8581, "step": 23480 }, { "epoch": 2.498510321344967, "grad_norm": 18.15625, "learning_rate": 9.183449403069927e-06, "loss": 0.8406, "step": 23481 }, { "epoch": 2.4986167269631836, "grad_norm": 15.8125, "learning_rate": 9.183413871517908e-06, "loss": 0.8827, "step": 23482 }, { "epoch": 2.4987231325814, "grad_norm": 11.5703125, "learning_rate": 9.18337833996589e-06, "loss": 0.7685, "step": 23483 }, { "epoch": 2.498829538199617, "grad_norm": 12.0625, "learning_rate": 9.183342808413871e-06, "loss": 0.777, "step": 23484 }, { "epoch": 2.4989359438178336, "grad_norm": 18.8125, "learning_rate": 9.183307276861853e-06, "loss": 0.8563, "step": 23485 }, { "epoch": 2.49904234943605, "grad_norm": 19.109375, "learning_rate": 9.183271745309836e-06, "loss": 1.1297, "step": 23486 }, { "epoch": 2.499148755054267, "grad_norm": 13.75, "learning_rate": 9.183236213757818e-06, "loss": 0.8488, "step": 23487 }, { "epoch": 2.4992551606724835, "grad_norm": 16.90625, "learning_rate": 9.1832006822058e-06, "loss": 0.9688, "step": 23488 }, { "epoch": 2.4993615662907, "grad_norm": 14.0234375, "learning_rate": 9.183165150653781e-06, "loss": 0.9167, "step": 23489 }, { "epoch": 2.499467971908917, "grad_norm": 16.734375, "learning_rate": 9.183129619101763e-06, "loss": 0.9505, "step": 23490 }, { "epoch": 2.4995743775271335, "grad_norm": 10.6171875, "learning_rate": 9.183094087549746e-06, "loss": 0.8312, "step": 23491 }, { "epoch": 2.49968078314535, "grad_norm": 15.203125, "learning_rate": 9.183058555997728e-06, "loss": 0.947, "step": 23492 }, { "epoch": 2.4997871887635665, "grad_norm": 10.6875, "learning_rate": 9.18302302444571e-06, "loss": 0.8545, "step": 23493 }, { "epoch": 2.4998935943817835, "grad_norm": 13.28125, "learning_rate": 9.182987492893691e-06, "loss": 0.7591, "step": 23494 }, { "epoch": 2.5, "grad_norm": 10.890625, "learning_rate": 9.182951961341672e-06, "loss": 0.8871, "step": 23495 }, { "epoch": 2.5001064056182165, "grad_norm": 12.5859375, "learning_rate": 9.182916429789654e-06, "loss": 0.8218, "step": 23496 }, { "epoch": 2.500212811236433, "grad_norm": 9.9453125, "learning_rate": 9.182880898237636e-06, "loss": 0.8302, "step": 23497 }, { "epoch": 2.50031921685465, "grad_norm": 9.9609375, "learning_rate": 9.182845366685617e-06, "loss": 0.7725, "step": 23498 }, { "epoch": 2.5004256224728665, "grad_norm": 12.3125, "learning_rate": 9.182809835133599e-06, "loss": 0.7209, "step": 23499 }, { "epoch": 2.500532028091083, "grad_norm": 14.5703125, "learning_rate": 9.18277430358158e-06, "loss": 0.8693, "step": 23500 }, { "epoch": 2.5006384337093, "grad_norm": 16.078125, "learning_rate": 9.182738772029564e-06, "loss": 0.7751, "step": 23501 }, { "epoch": 2.5007448393275165, "grad_norm": 13.1953125, "learning_rate": 9.182703240477545e-06, "loss": 0.9672, "step": 23502 }, { "epoch": 2.500851244945733, "grad_norm": 11.9453125, "learning_rate": 9.182667708925527e-06, "loss": 0.8279, "step": 23503 }, { "epoch": 2.50095765056395, "grad_norm": 17.875, "learning_rate": 9.182632177373509e-06, "loss": 0.854, "step": 23504 }, { "epoch": 2.5010640561821664, "grad_norm": 12.59375, "learning_rate": 9.18259664582149e-06, "loss": 1.0944, "step": 23505 }, { "epoch": 2.501170461800383, "grad_norm": 19.546875, "learning_rate": 9.182561114269472e-06, "loss": 0.9985, "step": 23506 }, { "epoch": 2.5012768674186, "grad_norm": 18.15625, "learning_rate": 9.182525582717453e-06, "loss": 0.867, "step": 23507 }, { "epoch": 2.5013832730368164, "grad_norm": 12.8125, "learning_rate": 9.182490051165435e-06, "loss": 0.8922, "step": 23508 }, { "epoch": 2.501489678655033, "grad_norm": 13.3125, "learning_rate": 9.182454519613417e-06, "loss": 0.8386, "step": 23509 }, { "epoch": 2.50159608427325, "grad_norm": 11.6796875, "learning_rate": 9.182418988061398e-06, "loss": 0.746, "step": 23510 }, { "epoch": 2.5017024898914664, "grad_norm": 12.2578125, "learning_rate": 9.18238345650938e-06, "loss": 0.9531, "step": 23511 }, { "epoch": 2.501808895509683, "grad_norm": 15.34375, "learning_rate": 9.182347924957363e-06, "loss": 0.9127, "step": 23512 }, { "epoch": 2.5019153011278994, "grad_norm": 12.4140625, "learning_rate": 9.182312393405345e-06, "loss": 0.8542, "step": 23513 }, { "epoch": 2.5020217067461163, "grad_norm": 11.3125, "learning_rate": 9.182276861853326e-06, "loss": 0.8356, "step": 23514 }, { "epoch": 2.502128112364333, "grad_norm": 12.921875, "learning_rate": 9.182241330301308e-06, "loss": 0.8812, "step": 23515 }, { "epoch": 2.5022345179825494, "grad_norm": 10.328125, "learning_rate": 9.18220579874929e-06, "loss": 0.8596, "step": 23516 }, { "epoch": 2.502340923600766, "grad_norm": 11.5625, "learning_rate": 9.182170267197273e-06, "loss": 0.7944, "step": 23517 }, { "epoch": 2.502447329218983, "grad_norm": 12.5546875, "learning_rate": 9.182134735645254e-06, "loss": 0.7768, "step": 23518 }, { "epoch": 2.5025537348371993, "grad_norm": 11.546875, "learning_rate": 9.182099204093236e-06, "loss": 0.767, "step": 23519 }, { "epoch": 2.502660140455416, "grad_norm": 17.671875, "learning_rate": 9.182063672541218e-06, "loss": 0.9336, "step": 23520 }, { "epoch": 2.502766546073633, "grad_norm": 15.59375, "learning_rate": 9.1820281409892e-06, "loss": 1.0055, "step": 23521 }, { "epoch": 2.5028729516918493, "grad_norm": 14.1953125, "learning_rate": 9.181992609437181e-06, "loss": 0.9324, "step": 23522 }, { "epoch": 2.502979357310066, "grad_norm": 9.9765625, "learning_rate": 9.181957077885163e-06, "loss": 0.8792, "step": 23523 }, { "epoch": 2.5030857629282828, "grad_norm": 12.578125, "learning_rate": 9.181921546333144e-06, "loss": 0.779, "step": 23524 }, { "epoch": 2.5031921685464993, "grad_norm": 11.015625, "learning_rate": 9.181886014781126e-06, "loss": 0.8604, "step": 23525 }, { "epoch": 2.503298574164716, "grad_norm": 13.75, "learning_rate": 9.181850483229107e-06, "loss": 0.9804, "step": 23526 }, { "epoch": 2.5034049797829327, "grad_norm": 14.78125, "learning_rate": 9.18181495167709e-06, "loss": 0.9225, "step": 23527 }, { "epoch": 2.5035113854011493, "grad_norm": 13.21875, "learning_rate": 9.181779420125072e-06, "loss": 1.0303, "step": 23528 }, { "epoch": 2.5036177910193658, "grad_norm": 13.3984375, "learning_rate": 9.181743888573054e-06, "loss": 0.8915, "step": 23529 }, { "epoch": 2.5037241966375827, "grad_norm": 13.890625, "learning_rate": 9.181708357021035e-06, "loss": 0.8401, "step": 23530 }, { "epoch": 2.5038306022557992, "grad_norm": 14.6640625, "learning_rate": 9.181672825469017e-06, "loss": 1.0008, "step": 23531 }, { "epoch": 2.5039370078740157, "grad_norm": 15.203125, "learning_rate": 9.181637293917e-06, "loss": 0.8257, "step": 23532 }, { "epoch": 2.5040434134922323, "grad_norm": 10.9765625, "learning_rate": 9.18160176236498e-06, "loss": 0.7515, "step": 23533 }, { "epoch": 2.5041498191104488, "grad_norm": 17.6875, "learning_rate": 9.181566230812962e-06, "loss": 1.0116, "step": 23534 }, { "epoch": 2.5042562247286657, "grad_norm": 14.3828125, "learning_rate": 9.181530699260944e-06, "loss": 0.8718, "step": 23535 }, { "epoch": 2.5043626303468822, "grad_norm": 14.984375, "learning_rate": 9.181495167708925e-06, "loss": 0.929, "step": 23536 }, { "epoch": 2.5044690359650987, "grad_norm": 13.78125, "learning_rate": 9.181459636156908e-06, "loss": 1.1362, "step": 23537 }, { "epoch": 2.5045754415833157, "grad_norm": 11.828125, "learning_rate": 9.18142410460489e-06, "loss": 0.8467, "step": 23538 }, { "epoch": 2.504681847201532, "grad_norm": 17.515625, "learning_rate": 9.181388573052872e-06, "loss": 1.0577, "step": 23539 }, { "epoch": 2.5047882528197487, "grad_norm": 13.0625, "learning_rate": 9.181353041500853e-06, "loss": 1.0852, "step": 23540 }, { "epoch": 2.5048946584379657, "grad_norm": 13.78125, "learning_rate": 9.181317509948835e-06, "loss": 0.865, "step": 23541 }, { "epoch": 2.505001064056182, "grad_norm": 18.875, "learning_rate": 9.181281978396818e-06, "loss": 0.9725, "step": 23542 }, { "epoch": 2.5051074696743987, "grad_norm": 14.53125, "learning_rate": 9.1812464468448e-06, "loss": 0.9275, "step": 23543 }, { "epoch": 2.5052138752926156, "grad_norm": 13.1328125, "learning_rate": 9.181210915292781e-06, "loss": 0.9443, "step": 23544 }, { "epoch": 2.505320280910832, "grad_norm": 15.2265625, "learning_rate": 9.181175383740763e-06, "loss": 0.9475, "step": 23545 }, { "epoch": 2.5054266865290487, "grad_norm": 20.296875, "learning_rate": 9.181139852188745e-06, "loss": 0.9664, "step": 23546 }, { "epoch": 2.5055330921472656, "grad_norm": 10.9453125, "learning_rate": 9.181104320636726e-06, "loss": 0.8139, "step": 23547 }, { "epoch": 2.505639497765482, "grad_norm": 21.5625, "learning_rate": 9.181068789084708e-06, "loss": 1.118, "step": 23548 }, { "epoch": 2.5057459033836986, "grad_norm": 14.5078125, "learning_rate": 9.18103325753269e-06, "loss": 1.0403, "step": 23549 }, { "epoch": 2.505852309001915, "grad_norm": 12.5625, "learning_rate": 9.180997725980671e-06, "loss": 0.739, "step": 23550 }, { "epoch": 2.505958714620132, "grad_norm": 11.7109375, "learning_rate": 9.180962194428653e-06, "loss": 0.976, "step": 23551 }, { "epoch": 2.5060651202383486, "grad_norm": 11.328125, "learning_rate": 9.180926662876634e-06, "loss": 0.8404, "step": 23552 }, { "epoch": 2.506171525856565, "grad_norm": 16.078125, "learning_rate": 9.180891131324618e-06, "loss": 0.835, "step": 23553 }, { "epoch": 2.5062779314747816, "grad_norm": 11.1484375, "learning_rate": 9.180855599772599e-06, "loss": 0.9267, "step": 23554 }, { "epoch": 2.5063843370929986, "grad_norm": 15.796875, "learning_rate": 9.18082006822058e-06, "loss": 0.9175, "step": 23555 }, { "epoch": 2.506490742711215, "grad_norm": 12.40625, "learning_rate": 9.180784536668562e-06, "loss": 0.9297, "step": 23556 }, { "epoch": 2.5065971483294316, "grad_norm": 9.7421875, "learning_rate": 9.180749005116544e-06, "loss": 0.7735, "step": 23557 }, { "epoch": 2.5067035539476485, "grad_norm": 11.9609375, "learning_rate": 9.180713473564527e-06, "loss": 0.9726, "step": 23558 }, { "epoch": 2.506809959565865, "grad_norm": 14.34375, "learning_rate": 9.180677942012509e-06, "loss": 1.077, "step": 23559 }, { "epoch": 2.5069163651840816, "grad_norm": 15.0, "learning_rate": 9.18064241046049e-06, "loss": 0.8935, "step": 23560 }, { "epoch": 2.5070227708022985, "grad_norm": 16.640625, "learning_rate": 9.180606878908472e-06, "loss": 0.9313, "step": 23561 }, { "epoch": 2.507129176420515, "grad_norm": 9.6640625, "learning_rate": 9.180571347356454e-06, "loss": 0.8054, "step": 23562 }, { "epoch": 2.5072355820387315, "grad_norm": 11.859375, "learning_rate": 9.180535815804435e-06, "loss": 0.8665, "step": 23563 }, { "epoch": 2.5073419876569485, "grad_norm": 18.515625, "learning_rate": 9.180500284252417e-06, "loss": 0.9409, "step": 23564 }, { "epoch": 2.507448393275165, "grad_norm": 16.765625, "learning_rate": 9.180464752700398e-06, "loss": 0.9689, "step": 23565 }, { "epoch": 2.5075547988933815, "grad_norm": 12.9140625, "learning_rate": 9.18042922114838e-06, "loss": 0.9966, "step": 23566 }, { "epoch": 2.5076612045115985, "grad_norm": 17.203125, "learning_rate": 9.180393689596362e-06, "loss": 0.8384, "step": 23567 }, { "epoch": 2.507767610129815, "grad_norm": 12.796875, "learning_rate": 9.180358158044345e-06, "loss": 1.0062, "step": 23568 }, { "epoch": 2.5078740157480315, "grad_norm": 14.796875, "learning_rate": 9.180322626492327e-06, "loss": 1.0996, "step": 23569 }, { "epoch": 2.507980421366248, "grad_norm": 15.359375, "learning_rate": 9.180287094940308e-06, "loss": 1.0054, "step": 23570 }, { "epoch": 2.508086826984465, "grad_norm": 16.3125, "learning_rate": 9.18025156338829e-06, "loss": 0.8466, "step": 23571 }, { "epoch": 2.5081932326026815, "grad_norm": 14.2265625, "learning_rate": 9.180216031836271e-06, "loss": 0.7865, "step": 23572 }, { "epoch": 2.508299638220898, "grad_norm": 14.1328125, "learning_rate": 9.180180500284253e-06, "loss": 0.8913, "step": 23573 }, { "epoch": 2.5084060438391145, "grad_norm": 14.5625, "learning_rate": 9.180144968732235e-06, "loss": 1.0993, "step": 23574 }, { "epoch": 2.5085124494573314, "grad_norm": 11.5078125, "learning_rate": 9.180109437180216e-06, "loss": 0.8795, "step": 23575 }, { "epoch": 2.508618855075548, "grad_norm": 11.2421875, "learning_rate": 9.180073905628198e-06, "loss": 0.912, "step": 23576 }, { "epoch": 2.5087252606937644, "grad_norm": 11.7265625, "learning_rate": 9.18003837407618e-06, "loss": 0.919, "step": 23577 }, { "epoch": 2.5088316663119814, "grad_norm": 13.25, "learning_rate": 9.180002842524163e-06, "loss": 0.8177, "step": 23578 }, { "epoch": 2.508938071930198, "grad_norm": 16.3125, "learning_rate": 9.179967310972144e-06, "loss": 1.075, "step": 23579 }, { "epoch": 2.5090444775484144, "grad_norm": 15.625, "learning_rate": 9.179931779420126e-06, "loss": 0.9174, "step": 23580 }, { "epoch": 2.5091508831666314, "grad_norm": 12.8203125, "learning_rate": 9.179896247868108e-06, "loss": 0.8793, "step": 23581 }, { "epoch": 2.509257288784848, "grad_norm": 10.203125, "learning_rate": 9.17986071631609e-06, "loss": 0.8765, "step": 23582 }, { "epoch": 2.5093636944030644, "grad_norm": 11.609375, "learning_rate": 9.17982518476407e-06, "loss": 0.8164, "step": 23583 }, { "epoch": 2.5094701000212813, "grad_norm": 15.5078125, "learning_rate": 9.179789653212054e-06, "loss": 1.084, "step": 23584 }, { "epoch": 2.509576505639498, "grad_norm": 13.1953125, "learning_rate": 9.179754121660036e-06, "loss": 1.0525, "step": 23585 }, { "epoch": 2.5096829112577144, "grad_norm": 12.1484375, "learning_rate": 9.179718590108017e-06, "loss": 0.8556, "step": 23586 }, { "epoch": 2.5097893168759313, "grad_norm": 11.5390625, "learning_rate": 9.179683058555999e-06, "loss": 0.8273, "step": 23587 }, { "epoch": 2.509895722494148, "grad_norm": 13.3203125, "learning_rate": 9.17964752700398e-06, "loss": 0.9311, "step": 23588 }, { "epoch": 2.5100021281123643, "grad_norm": 12.859375, "learning_rate": 9.179611995451962e-06, "loss": 0.9492, "step": 23589 }, { "epoch": 2.510108533730581, "grad_norm": 14.515625, "learning_rate": 9.179576463899944e-06, "loss": 0.7622, "step": 23590 }, { "epoch": 2.5102149393487974, "grad_norm": 12.703125, "learning_rate": 9.179540932347925e-06, "loss": 0.7914, "step": 23591 }, { "epoch": 2.5103213449670143, "grad_norm": 11.4921875, "learning_rate": 9.179505400795907e-06, "loss": 0.9568, "step": 23592 }, { "epoch": 2.510427750585231, "grad_norm": 11.1171875, "learning_rate": 9.179469869243889e-06, "loss": 0.7635, "step": 23593 }, { "epoch": 2.5105341562034473, "grad_norm": 12.7421875, "learning_rate": 9.179434337691872e-06, "loss": 1.0309, "step": 23594 }, { "epoch": 2.5106405618216643, "grad_norm": 13.15625, "learning_rate": 9.179398806139853e-06, "loss": 0.9989, "step": 23595 }, { "epoch": 2.510746967439881, "grad_norm": 13.2421875, "learning_rate": 9.179363274587835e-06, "loss": 1.127, "step": 23596 }, { "epoch": 2.5108533730580973, "grad_norm": 13.1015625, "learning_rate": 9.179327743035817e-06, "loss": 1.004, "step": 23597 }, { "epoch": 2.5109597786763143, "grad_norm": 14.0390625, "learning_rate": 9.179292211483798e-06, "loss": 0.8873, "step": 23598 }, { "epoch": 2.5110661842945308, "grad_norm": 15.453125, "learning_rate": 9.17925667993178e-06, "loss": 1.1442, "step": 23599 }, { "epoch": 2.5111725899127473, "grad_norm": 10.40625, "learning_rate": 9.179221148379761e-06, "loss": 0.9882, "step": 23600 }, { "epoch": 2.5112789955309642, "grad_norm": 13.09375, "learning_rate": 9.179185616827743e-06, "loss": 0.778, "step": 23601 }, { "epoch": 2.5113854011491807, "grad_norm": 16.234375, "learning_rate": 9.179150085275725e-06, "loss": 1.0023, "step": 23602 }, { "epoch": 2.5114918067673973, "grad_norm": 12.453125, "learning_rate": 9.179114553723706e-06, "loss": 0.7896, "step": 23603 }, { "epoch": 2.511598212385614, "grad_norm": 13.1015625, "learning_rate": 9.17907902217169e-06, "loss": 1.2378, "step": 23604 }, { "epoch": 2.5117046180038307, "grad_norm": 14.265625, "learning_rate": 9.179043490619671e-06, "loss": 1.1899, "step": 23605 }, { "epoch": 2.5118110236220472, "grad_norm": 12.1640625, "learning_rate": 9.179007959067653e-06, "loss": 0.8536, "step": 23606 }, { "epoch": 2.5119174292402637, "grad_norm": 13.9765625, "learning_rate": 9.178972427515634e-06, "loss": 0.8219, "step": 23607 }, { "epoch": 2.5120238348584807, "grad_norm": 12.46875, "learning_rate": 9.178936895963616e-06, "loss": 0.7377, "step": 23608 }, { "epoch": 2.512130240476697, "grad_norm": 12.46875, "learning_rate": 9.1789013644116e-06, "loss": 0.8266, "step": 23609 }, { "epoch": 2.5122366460949137, "grad_norm": 11.453125, "learning_rate": 9.178865832859581e-06, "loss": 0.9049, "step": 23610 }, { "epoch": 2.51234305171313, "grad_norm": 12.7890625, "learning_rate": 9.178830301307563e-06, "loss": 0.9438, "step": 23611 }, { "epoch": 2.512449457331347, "grad_norm": 16.640625, "learning_rate": 9.178794769755544e-06, "loss": 0.9148, "step": 23612 }, { "epoch": 2.5125558629495637, "grad_norm": 12.71875, "learning_rate": 9.178759238203526e-06, "loss": 0.8841, "step": 23613 }, { "epoch": 2.51266226856778, "grad_norm": 15.953125, "learning_rate": 9.178723706651507e-06, "loss": 0.9817, "step": 23614 }, { "epoch": 2.512768674185997, "grad_norm": 9.34375, "learning_rate": 9.178688175099489e-06, "loss": 0.7141, "step": 23615 }, { "epoch": 2.5128750798042137, "grad_norm": 9.8359375, "learning_rate": 9.17865264354747e-06, "loss": 0.7759, "step": 23616 }, { "epoch": 2.51298148542243, "grad_norm": 14.1796875, "learning_rate": 9.178617111995452e-06, "loss": 0.9577, "step": 23617 }, { "epoch": 2.513087891040647, "grad_norm": 11.1640625, "learning_rate": 9.178581580443434e-06, "loss": 0.8508, "step": 23618 }, { "epoch": 2.5131942966588636, "grad_norm": 13.359375, "learning_rate": 9.178546048891417e-06, "loss": 0.8983, "step": 23619 }, { "epoch": 2.51330070227708, "grad_norm": 15.34375, "learning_rate": 9.178510517339399e-06, "loss": 0.9018, "step": 23620 }, { "epoch": 2.513407107895297, "grad_norm": 15.0078125, "learning_rate": 9.17847498578738e-06, "loss": 1.0159, "step": 23621 }, { "epoch": 2.5135135135135136, "grad_norm": 12.921875, "learning_rate": 9.178439454235362e-06, "loss": 0.8848, "step": 23622 }, { "epoch": 2.51361991913173, "grad_norm": 12.1484375, "learning_rate": 9.178403922683344e-06, "loss": 0.9208, "step": 23623 }, { "epoch": 2.513726324749947, "grad_norm": 13.796875, "learning_rate": 9.178368391131325e-06, "loss": 1.1395, "step": 23624 }, { "epoch": 2.5138327303681636, "grad_norm": 11.0078125, "learning_rate": 9.178332859579307e-06, "loss": 0.9737, "step": 23625 }, { "epoch": 2.51393913598638, "grad_norm": 13.703125, "learning_rate": 9.178297328027288e-06, "loss": 0.9665, "step": 23626 }, { "epoch": 2.5140455416045966, "grad_norm": 13.765625, "learning_rate": 9.17826179647527e-06, "loss": 1.0413, "step": 23627 }, { "epoch": 2.514151947222813, "grad_norm": 15.09375, "learning_rate": 9.178226264923252e-06, "loss": 1.2163, "step": 23628 }, { "epoch": 2.51425835284103, "grad_norm": 15.1875, "learning_rate": 9.178190733371233e-06, "loss": 0.8897, "step": 23629 }, { "epoch": 2.5143647584592466, "grad_norm": 15.171875, "learning_rate": 9.178155201819216e-06, "loss": 0.885, "step": 23630 }, { "epoch": 2.514471164077463, "grad_norm": 11.2109375, "learning_rate": 9.178119670267198e-06, "loss": 0.8549, "step": 23631 }, { "epoch": 2.51457756969568, "grad_norm": 14.6484375, "learning_rate": 9.17808413871518e-06, "loss": 0.8713, "step": 23632 }, { "epoch": 2.5146839753138965, "grad_norm": 14.9140625, "learning_rate": 9.178048607163161e-06, "loss": 0.7774, "step": 23633 }, { "epoch": 2.514790380932113, "grad_norm": 12.4375, "learning_rate": 9.178013075611143e-06, "loss": 0.9226, "step": 23634 }, { "epoch": 2.51489678655033, "grad_norm": 12.3984375, "learning_rate": 9.177977544059126e-06, "loss": 0.9896, "step": 23635 }, { "epoch": 2.5150031921685465, "grad_norm": 15.8203125, "learning_rate": 9.177942012507108e-06, "loss": 0.9914, "step": 23636 }, { "epoch": 2.515109597786763, "grad_norm": 10.78125, "learning_rate": 9.17790648095509e-06, "loss": 0.6806, "step": 23637 }, { "epoch": 2.51521600340498, "grad_norm": 14.1171875, "learning_rate": 9.177870949403071e-06, "loss": 0.8988, "step": 23638 }, { "epoch": 2.5153224090231965, "grad_norm": 19.375, "learning_rate": 9.177835417851053e-06, "loss": 1.0819, "step": 23639 }, { "epoch": 2.515428814641413, "grad_norm": 13.4375, "learning_rate": 9.177799886299034e-06, "loss": 0.922, "step": 23640 }, { "epoch": 2.51553522025963, "grad_norm": 17.25, "learning_rate": 9.177764354747016e-06, "loss": 0.8917, "step": 23641 }, { "epoch": 2.5156416258778465, "grad_norm": 10.7890625, "learning_rate": 9.177728823194997e-06, "loss": 0.9399, "step": 23642 }, { "epoch": 2.515748031496063, "grad_norm": 17.828125, "learning_rate": 9.177693291642979e-06, "loss": 0.9764, "step": 23643 }, { "epoch": 2.5158544371142795, "grad_norm": 13.546875, "learning_rate": 9.17765776009096e-06, "loss": 0.9557, "step": 23644 }, { "epoch": 2.5159608427324964, "grad_norm": 15.4375, "learning_rate": 9.177622228538944e-06, "loss": 0.8641, "step": 23645 }, { "epoch": 2.516067248350713, "grad_norm": 15.3515625, "learning_rate": 9.177586696986926e-06, "loss": 0.9136, "step": 23646 }, { "epoch": 2.5161736539689294, "grad_norm": 12.890625, "learning_rate": 9.177551165434907e-06, "loss": 0.7854, "step": 23647 }, { "epoch": 2.516280059587146, "grad_norm": 16.5, "learning_rate": 9.177515633882889e-06, "loss": 0.9568, "step": 23648 }, { "epoch": 2.516386465205363, "grad_norm": 16.5, "learning_rate": 9.17748010233087e-06, "loss": 1.0146, "step": 23649 }, { "epoch": 2.5164928708235794, "grad_norm": 13.203125, "learning_rate": 9.177444570778854e-06, "loss": 0.789, "step": 23650 }, { "epoch": 2.516599276441796, "grad_norm": 13.1328125, "learning_rate": 9.177409039226835e-06, "loss": 0.8879, "step": 23651 }, { "epoch": 2.516705682060013, "grad_norm": 10.3046875, "learning_rate": 9.177373507674817e-06, "loss": 0.6941, "step": 23652 }, { "epoch": 2.5168120876782294, "grad_norm": 9.703125, "learning_rate": 9.177337976122797e-06, "loss": 0.7358, "step": 23653 }, { "epoch": 2.516918493296446, "grad_norm": 12.6640625, "learning_rate": 9.177302444570778e-06, "loss": 0.9621, "step": 23654 }, { "epoch": 2.517024898914663, "grad_norm": 18.703125, "learning_rate": 9.177266913018762e-06, "loss": 0.7816, "step": 23655 }, { "epoch": 2.5171313045328794, "grad_norm": 14.5234375, "learning_rate": 9.177231381466743e-06, "loss": 1.0042, "step": 23656 }, { "epoch": 2.517237710151096, "grad_norm": 9.46875, "learning_rate": 9.177195849914725e-06, "loss": 0.7223, "step": 23657 }, { "epoch": 2.517344115769313, "grad_norm": 15.953125, "learning_rate": 9.177160318362707e-06, "loss": 0.95, "step": 23658 }, { "epoch": 2.5174505213875293, "grad_norm": 21.40625, "learning_rate": 9.177124786810688e-06, "loss": 0.8137, "step": 23659 }, { "epoch": 2.517556927005746, "grad_norm": 11.9453125, "learning_rate": 9.177089255258671e-06, "loss": 0.7559, "step": 23660 }, { "epoch": 2.517663332623963, "grad_norm": 10.65625, "learning_rate": 9.177053723706653e-06, "loss": 0.8667, "step": 23661 }, { "epoch": 2.5177697382421793, "grad_norm": 13.9453125, "learning_rate": 9.177018192154635e-06, "loss": 0.9, "step": 23662 }, { "epoch": 2.517876143860396, "grad_norm": 13.5078125, "learning_rate": 9.176982660602616e-06, "loss": 0.8934, "step": 23663 }, { "epoch": 2.5179825494786123, "grad_norm": 12.0703125, "learning_rate": 9.176947129050598e-06, "loss": 1.0286, "step": 23664 }, { "epoch": 2.518088955096829, "grad_norm": 10.28125, "learning_rate": 9.17691159749858e-06, "loss": 0.8882, "step": 23665 }, { "epoch": 2.518195360715046, "grad_norm": 14.9375, "learning_rate": 9.176876065946561e-06, "loss": 0.9117, "step": 23666 }, { "epoch": 2.5183017663332623, "grad_norm": 10.96875, "learning_rate": 9.176840534394543e-06, "loss": 0.9408, "step": 23667 }, { "epoch": 2.518408171951479, "grad_norm": 12.6953125, "learning_rate": 9.176805002842524e-06, "loss": 0.7808, "step": 23668 }, { "epoch": 2.5185145775696958, "grad_norm": 11.265625, "learning_rate": 9.176769471290506e-06, "loss": 0.8518, "step": 23669 }, { "epoch": 2.5186209831879123, "grad_norm": 16.390625, "learning_rate": 9.176733939738488e-06, "loss": 1.0789, "step": 23670 }, { "epoch": 2.518727388806129, "grad_norm": 12.3359375, "learning_rate": 9.17669840818647e-06, "loss": 0.9353, "step": 23671 }, { "epoch": 2.5188337944243457, "grad_norm": 18.6875, "learning_rate": 9.176662876634452e-06, "loss": 0.9163, "step": 23672 }, { "epoch": 2.5189402000425622, "grad_norm": 11.90625, "learning_rate": 9.176627345082434e-06, "loss": 0.9864, "step": 23673 }, { "epoch": 2.5190466056607788, "grad_norm": 16.453125, "learning_rate": 9.176591813530416e-06, "loss": 0.9195, "step": 23674 }, { "epoch": 2.5191530112789957, "grad_norm": 13.28125, "learning_rate": 9.176556281978397e-06, "loss": 0.662, "step": 23675 }, { "epoch": 2.5192594168972122, "grad_norm": 15.625, "learning_rate": 9.17652075042638e-06, "loss": 1.1056, "step": 23676 }, { "epoch": 2.5193658225154287, "grad_norm": 21.0625, "learning_rate": 9.176485218874362e-06, "loss": 0.9168, "step": 23677 }, { "epoch": 2.5194722281336457, "grad_norm": 10.609375, "learning_rate": 9.176449687322344e-06, "loss": 0.987, "step": 23678 }, { "epoch": 2.519578633751862, "grad_norm": 12.8359375, "learning_rate": 9.176414155770325e-06, "loss": 0.9508, "step": 23679 }, { "epoch": 2.5196850393700787, "grad_norm": 8.734375, "learning_rate": 9.176378624218307e-06, "loss": 0.6353, "step": 23680 }, { "epoch": 2.5197914449882957, "grad_norm": 12.5546875, "learning_rate": 9.176343092666289e-06, "loss": 0.8395, "step": 23681 }, { "epoch": 2.519897850606512, "grad_norm": 13.25, "learning_rate": 9.17630756111427e-06, "loss": 0.8567, "step": 23682 }, { "epoch": 2.5200042562247287, "grad_norm": 17.0625, "learning_rate": 9.176272029562252e-06, "loss": 0.9811, "step": 23683 }, { "epoch": 2.520110661842945, "grad_norm": 11.8515625, "learning_rate": 9.176236498010233e-06, "loss": 0.9177, "step": 23684 }, { "epoch": 2.5202170674611617, "grad_norm": 14.9296875, "learning_rate": 9.176200966458215e-06, "loss": 0.9543, "step": 23685 }, { "epoch": 2.5203234730793787, "grad_norm": 11.6171875, "learning_rate": 9.176165434906198e-06, "loss": 0.9457, "step": 23686 }, { "epoch": 2.520429878697595, "grad_norm": 11.1484375, "learning_rate": 9.17612990335418e-06, "loss": 0.8381, "step": 23687 }, { "epoch": 2.5205362843158117, "grad_norm": 12.0390625, "learning_rate": 9.176094371802162e-06, "loss": 0.8131, "step": 23688 }, { "epoch": 2.5206426899340286, "grad_norm": 18.8125, "learning_rate": 9.176058840250143e-06, "loss": 1.1372, "step": 23689 }, { "epoch": 2.520749095552245, "grad_norm": 13.2734375, "learning_rate": 9.176023308698125e-06, "loss": 0.8476, "step": 23690 }, { "epoch": 2.5208555011704616, "grad_norm": 15.109375, "learning_rate": 9.175987777146106e-06, "loss": 0.8706, "step": 23691 }, { "epoch": 2.5209619067886786, "grad_norm": 17.875, "learning_rate": 9.175952245594088e-06, "loss": 1.0435, "step": 23692 }, { "epoch": 2.521068312406895, "grad_norm": 16.015625, "learning_rate": 9.17591671404207e-06, "loss": 0.8842, "step": 23693 }, { "epoch": 2.5211747180251116, "grad_norm": 16.0, "learning_rate": 9.175881182490051e-06, "loss": 0.945, "step": 23694 }, { "epoch": 2.5212811236433286, "grad_norm": 20.515625, "learning_rate": 9.175845650938033e-06, "loss": 1.1487, "step": 23695 }, { "epoch": 2.521387529261545, "grad_norm": 11.671875, "learning_rate": 9.175810119386016e-06, "loss": 0.8187, "step": 23696 }, { "epoch": 2.5214939348797616, "grad_norm": 14.9375, "learning_rate": 9.175774587833998e-06, "loss": 1.0558, "step": 23697 }, { "epoch": 2.5216003404979785, "grad_norm": 13.0546875, "learning_rate": 9.17573905628198e-06, "loss": 0.8116, "step": 23698 }, { "epoch": 2.521706746116195, "grad_norm": 14.4921875, "learning_rate": 9.175703524729961e-06, "loss": 0.7523, "step": 23699 }, { "epoch": 2.5218131517344116, "grad_norm": 15.6328125, "learning_rate": 9.175667993177942e-06, "loss": 0.9374, "step": 23700 }, { "epoch": 2.521919557352628, "grad_norm": 17.25, "learning_rate": 9.175632461625924e-06, "loss": 1.0722, "step": 23701 }, { "epoch": 2.522025962970845, "grad_norm": 12.1953125, "learning_rate": 9.175596930073907e-06, "loss": 0.8422, "step": 23702 }, { "epoch": 2.5221323685890615, "grad_norm": 14.375, "learning_rate": 9.175561398521889e-06, "loss": 0.9749, "step": 23703 }, { "epoch": 2.522238774207278, "grad_norm": 11.4375, "learning_rate": 9.17552586696987e-06, "loss": 0.9329, "step": 23704 }, { "epoch": 2.5223451798254946, "grad_norm": 13.5546875, "learning_rate": 9.175490335417852e-06, "loss": 0.9226, "step": 23705 }, { "epoch": 2.5224515854437115, "grad_norm": 13.6796875, "learning_rate": 9.175454803865834e-06, "loss": 0.7518, "step": 23706 }, { "epoch": 2.522557991061928, "grad_norm": 11.3125, "learning_rate": 9.175419272313815e-06, "loss": 0.9452, "step": 23707 }, { "epoch": 2.5226643966801445, "grad_norm": 11.484375, "learning_rate": 9.175383740761797e-06, "loss": 0.8299, "step": 23708 }, { "epoch": 2.5227708022983615, "grad_norm": 16.046875, "learning_rate": 9.175348209209779e-06, "loss": 0.9817, "step": 23709 }, { "epoch": 2.522877207916578, "grad_norm": 13.65625, "learning_rate": 9.17531267765776e-06, "loss": 0.953, "step": 23710 }, { "epoch": 2.5229836135347945, "grad_norm": 14.5234375, "learning_rate": 9.175277146105742e-06, "loss": 0.9401, "step": 23711 }, { "epoch": 2.5230900191530115, "grad_norm": 18.78125, "learning_rate": 9.175241614553725e-06, "loss": 0.9431, "step": 23712 }, { "epoch": 2.523196424771228, "grad_norm": 15.09375, "learning_rate": 9.175206083001707e-06, "loss": 0.8445, "step": 23713 }, { "epoch": 2.5233028303894445, "grad_norm": 12.1484375, "learning_rate": 9.175170551449688e-06, "loss": 0.8136, "step": 23714 }, { "epoch": 2.5234092360076614, "grad_norm": 15.2734375, "learning_rate": 9.17513501989767e-06, "loss": 0.7574, "step": 23715 }, { "epoch": 2.523515641625878, "grad_norm": 13.875, "learning_rate": 9.175099488345652e-06, "loss": 0.9864, "step": 23716 }, { "epoch": 2.5236220472440944, "grad_norm": 15.4140625, "learning_rate": 9.175063956793633e-06, "loss": 0.8302, "step": 23717 }, { "epoch": 2.5237284528623114, "grad_norm": 17.46875, "learning_rate": 9.175028425241615e-06, "loss": 1.0681, "step": 23718 }, { "epoch": 2.523834858480528, "grad_norm": 18.921875, "learning_rate": 9.174992893689596e-06, "loss": 0.9822, "step": 23719 }, { "epoch": 2.5239412640987444, "grad_norm": 15.4296875, "learning_rate": 9.174957362137578e-06, "loss": 0.9754, "step": 23720 }, { "epoch": 2.524047669716961, "grad_norm": 12.5390625, "learning_rate": 9.17492183058556e-06, "loss": 0.9633, "step": 23721 }, { "epoch": 2.5241540753351774, "grad_norm": 10.8984375, "learning_rate": 9.174886299033543e-06, "loss": 0.9165, "step": 23722 }, { "epoch": 2.5242604809533944, "grad_norm": 11.921875, "learning_rate": 9.174850767481525e-06, "loss": 0.9122, "step": 23723 }, { "epoch": 2.524366886571611, "grad_norm": 14.7734375, "learning_rate": 9.174815235929506e-06, "loss": 0.9382, "step": 23724 }, { "epoch": 2.5244732921898274, "grad_norm": 17.3125, "learning_rate": 9.174779704377488e-06, "loss": 0.9393, "step": 23725 }, { "epoch": 2.5245796978080444, "grad_norm": 15.796875, "learning_rate": 9.17474417282547e-06, "loss": 1.0276, "step": 23726 }, { "epoch": 2.524686103426261, "grad_norm": 14.828125, "learning_rate": 9.174708641273453e-06, "loss": 0.6656, "step": 23727 }, { "epoch": 2.5247925090444774, "grad_norm": 11.28125, "learning_rate": 9.174673109721434e-06, "loss": 0.8342, "step": 23728 }, { "epoch": 2.5248989146626943, "grad_norm": 12.859375, "learning_rate": 9.174637578169416e-06, "loss": 0.9967, "step": 23729 }, { "epoch": 2.525005320280911, "grad_norm": 9.578125, "learning_rate": 9.174602046617397e-06, "loss": 0.7359, "step": 23730 }, { "epoch": 2.5251117258991274, "grad_norm": 11.1875, "learning_rate": 9.174566515065379e-06, "loss": 0.9503, "step": 23731 }, { "epoch": 2.5252181315173443, "grad_norm": 9.1640625, "learning_rate": 9.17453098351336e-06, "loss": 0.8801, "step": 23732 }, { "epoch": 2.525324537135561, "grad_norm": 13.5859375, "learning_rate": 9.174495451961342e-06, "loss": 0.7929, "step": 23733 }, { "epoch": 2.5254309427537773, "grad_norm": 11.6953125, "learning_rate": 9.174459920409324e-06, "loss": 0.8596, "step": 23734 }, { "epoch": 2.5255373483719943, "grad_norm": 11.90625, "learning_rate": 9.174424388857305e-06, "loss": 1.0008, "step": 23735 }, { "epoch": 2.525643753990211, "grad_norm": 10.3515625, "learning_rate": 9.174388857305287e-06, "loss": 0.9115, "step": 23736 }, { "epoch": 2.5257501596084273, "grad_norm": 12.1640625, "learning_rate": 9.17435332575327e-06, "loss": 0.921, "step": 23737 }, { "epoch": 2.525856565226644, "grad_norm": 14.328125, "learning_rate": 9.174317794201252e-06, "loss": 0.9634, "step": 23738 }, { "epoch": 2.5259629708448608, "grad_norm": 9.8671875, "learning_rate": 9.174282262649234e-06, "loss": 0.738, "step": 23739 }, { "epoch": 2.5260693764630773, "grad_norm": 10.0625, "learning_rate": 9.174246731097215e-06, "loss": 0.8243, "step": 23740 }, { "epoch": 2.526175782081294, "grad_norm": 13.3671875, "learning_rate": 9.174211199545197e-06, "loss": 1.009, "step": 23741 }, { "epoch": 2.5262821876995103, "grad_norm": 14.3828125, "learning_rate": 9.174175667993178e-06, "loss": 0.9991, "step": 23742 }, { "epoch": 2.5263885933177272, "grad_norm": 11.8515625, "learning_rate": 9.17414013644116e-06, "loss": 0.9474, "step": 23743 }, { "epoch": 2.5264949989359438, "grad_norm": 10.8671875, "learning_rate": 9.174104604889142e-06, "loss": 0.9152, "step": 23744 }, { "epoch": 2.5266014045541603, "grad_norm": 10.1171875, "learning_rate": 9.174069073337123e-06, "loss": 0.7753, "step": 23745 }, { "epoch": 2.5267078101723772, "grad_norm": 10.8828125, "learning_rate": 9.174033541785105e-06, "loss": 1.0361, "step": 23746 }, { "epoch": 2.5268142157905937, "grad_norm": 15.3203125, "learning_rate": 9.173998010233086e-06, "loss": 0.8452, "step": 23747 }, { "epoch": 2.5269206214088102, "grad_norm": 11.2578125, "learning_rate": 9.17396247868107e-06, "loss": 0.9194, "step": 23748 }, { "epoch": 2.527027027027027, "grad_norm": 13.640625, "learning_rate": 9.173926947129051e-06, "loss": 1.0509, "step": 23749 }, { "epoch": 2.5271334326452437, "grad_norm": 11.046875, "learning_rate": 9.173891415577033e-06, "loss": 0.7645, "step": 23750 }, { "epoch": 2.52723983826346, "grad_norm": 11.765625, "learning_rate": 9.173855884025015e-06, "loss": 0.9662, "step": 23751 }, { "epoch": 2.527346243881677, "grad_norm": 12.7734375, "learning_rate": 9.173820352472996e-06, "loss": 1.1103, "step": 23752 }, { "epoch": 2.5274526494998937, "grad_norm": 9.953125, "learning_rate": 9.17378482092098e-06, "loss": 0.9182, "step": 23753 }, { "epoch": 2.52755905511811, "grad_norm": 14.4453125, "learning_rate": 9.173749289368961e-06, "loss": 0.7976, "step": 23754 }, { "epoch": 2.527665460736327, "grad_norm": 11.515625, "learning_rate": 9.173713757816943e-06, "loss": 0.9607, "step": 23755 }, { "epoch": 2.5277718663545437, "grad_norm": 12.515625, "learning_rate": 9.173678226264924e-06, "loss": 0.8294, "step": 23756 }, { "epoch": 2.52787827197276, "grad_norm": 12.265625, "learning_rate": 9.173642694712906e-06, "loss": 0.8809, "step": 23757 }, { "epoch": 2.5279846775909767, "grad_norm": 13.421875, "learning_rate": 9.173607163160888e-06, "loss": 0.7858, "step": 23758 }, { "epoch": 2.528091083209193, "grad_norm": 16.03125, "learning_rate": 9.173571631608869e-06, "loss": 0.8669, "step": 23759 }, { "epoch": 2.52819748882741, "grad_norm": 18.203125, "learning_rate": 9.17353610005685e-06, "loss": 1.0299, "step": 23760 }, { "epoch": 2.5283038944456266, "grad_norm": 14.484375, "learning_rate": 9.173500568504832e-06, "loss": 1.0273, "step": 23761 }, { "epoch": 2.528410300063843, "grad_norm": 21.5625, "learning_rate": 9.173465036952814e-06, "loss": 0.9611, "step": 23762 }, { "epoch": 2.52851670568206, "grad_norm": 13.0078125, "learning_rate": 9.173429505400797e-06, "loss": 0.9289, "step": 23763 }, { "epoch": 2.5286231113002766, "grad_norm": 16.734375, "learning_rate": 9.173393973848779e-06, "loss": 0.9627, "step": 23764 }, { "epoch": 2.528729516918493, "grad_norm": 16.484375, "learning_rate": 9.17335844229676e-06, "loss": 1.0686, "step": 23765 }, { "epoch": 2.52883592253671, "grad_norm": 15.7265625, "learning_rate": 9.173322910744742e-06, "loss": 0.7896, "step": 23766 }, { "epoch": 2.5289423281549266, "grad_norm": 19.8125, "learning_rate": 9.173287379192724e-06, "loss": 0.8925, "step": 23767 }, { "epoch": 2.529048733773143, "grad_norm": 17.296875, "learning_rate": 9.173251847640707e-06, "loss": 0.9867, "step": 23768 }, { "epoch": 2.52915513939136, "grad_norm": 15.765625, "learning_rate": 9.173216316088689e-06, "loss": 0.935, "step": 23769 }, { "epoch": 2.5292615450095766, "grad_norm": 13.59375, "learning_rate": 9.17318078453667e-06, "loss": 1.0618, "step": 23770 }, { "epoch": 2.529367950627793, "grad_norm": 13.2890625, "learning_rate": 9.173145252984652e-06, "loss": 0.7979, "step": 23771 }, { "epoch": 2.52947435624601, "grad_norm": 12.640625, "learning_rate": 9.173109721432633e-06, "loss": 0.9437, "step": 23772 }, { "epoch": 2.5295807618642265, "grad_norm": 14.046875, "learning_rate": 9.173074189880615e-06, "loss": 0.8957, "step": 23773 }, { "epoch": 2.529687167482443, "grad_norm": 11.1953125, "learning_rate": 9.173038658328597e-06, "loss": 0.6795, "step": 23774 }, { "epoch": 2.5297935731006596, "grad_norm": 14.8203125, "learning_rate": 9.173003126776578e-06, "loss": 0.9427, "step": 23775 }, { "epoch": 2.5298999787188765, "grad_norm": 12.9140625, "learning_rate": 9.17296759522456e-06, "loss": 1.0065, "step": 23776 }, { "epoch": 2.530006384337093, "grad_norm": 17.609375, "learning_rate": 9.172932063672541e-06, "loss": 1.0182, "step": 23777 }, { "epoch": 2.5301127899553095, "grad_norm": 10.609375, "learning_rate": 9.172896532120523e-06, "loss": 0.7954, "step": 23778 }, { "epoch": 2.530219195573526, "grad_norm": 14.890625, "learning_rate": 9.172861000568506e-06, "loss": 0.9035, "step": 23779 }, { "epoch": 2.530325601191743, "grad_norm": 13.875, "learning_rate": 9.172825469016488e-06, "loss": 0.9604, "step": 23780 }, { "epoch": 2.5304320068099595, "grad_norm": 13.8203125, "learning_rate": 9.17278993746447e-06, "loss": 0.9232, "step": 23781 }, { "epoch": 2.530538412428176, "grad_norm": 12.984375, "learning_rate": 9.172754405912451e-06, "loss": 0.8618, "step": 23782 }, { "epoch": 2.530644818046393, "grad_norm": 12.1796875, "learning_rate": 9.172718874360433e-06, "loss": 0.8771, "step": 23783 }, { "epoch": 2.5307512236646095, "grad_norm": 14.5625, "learning_rate": 9.172683342808414e-06, "loss": 0.8834, "step": 23784 }, { "epoch": 2.530857629282826, "grad_norm": 15.2109375, "learning_rate": 9.172647811256396e-06, "loss": 0.9296, "step": 23785 }, { "epoch": 2.530964034901043, "grad_norm": 17.640625, "learning_rate": 9.172612279704378e-06, "loss": 0.9185, "step": 23786 }, { "epoch": 2.5310704405192594, "grad_norm": 14.4921875, "learning_rate": 9.17257674815236e-06, "loss": 1.0084, "step": 23787 }, { "epoch": 2.531176846137476, "grad_norm": 13.7890625, "learning_rate": 9.17254121660034e-06, "loss": 0.8661, "step": 23788 }, { "epoch": 2.531283251755693, "grad_norm": 12.25, "learning_rate": 9.172505685048324e-06, "loss": 0.9921, "step": 23789 }, { "epoch": 2.5313896573739094, "grad_norm": 12.2890625, "learning_rate": 9.172470153496306e-06, "loss": 0.9038, "step": 23790 }, { "epoch": 2.531496062992126, "grad_norm": 13.96875, "learning_rate": 9.172434621944287e-06, "loss": 1.1011, "step": 23791 }, { "epoch": 2.531602468610343, "grad_norm": 10.515625, "learning_rate": 9.172399090392269e-06, "loss": 0.7828, "step": 23792 }, { "epoch": 2.5317088742285594, "grad_norm": 17.34375, "learning_rate": 9.17236355884025e-06, "loss": 1.1386, "step": 23793 }, { "epoch": 2.531815279846776, "grad_norm": 12.59375, "learning_rate": 9.172328027288234e-06, "loss": 1.0511, "step": 23794 }, { "epoch": 2.5319216854649924, "grad_norm": 11.359375, "learning_rate": 9.172292495736215e-06, "loss": 0.8525, "step": 23795 }, { "epoch": 2.532028091083209, "grad_norm": 16.046875, "learning_rate": 9.172256964184197e-06, "loss": 0.8635, "step": 23796 }, { "epoch": 2.532134496701426, "grad_norm": 15.453125, "learning_rate": 9.172221432632179e-06, "loss": 0.8986, "step": 23797 }, { "epoch": 2.5322409023196424, "grad_norm": 14.3359375, "learning_rate": 9.17218590108016e-06, "loss": 1.0292, "step": 23798 }, { "epoch": 2.532347307937859, "grad_norm": 10.5390625, "learning_rate": 9.172150369528142e-06, "loss": 0.7084, "step": 23799 }, { "epoch": 2.532453713556076, "grad_norm": 16.796875, "learning_rate": 9.172114837976123e-06, "loss": 1.049, "step": 23800 }, { "epoch": 2.5325601191742924, "grad_norm": 10.8203125, "learning_rate": 9.172079306424105e-06, "loss": 0.8881, "step": 23801 }, { "epoch": 2.532666524792509, "grad_norm": 13.3125, "learning_rate": 9.172043774872087e-06, "loss": 0.8908, "step": 23802 }, { "epoch": 2.532772930410726, "grad_norm": 13.6953125, "learning_rate": 9.172008243320068e-06, "loss": 0.9328, "step": 23803 }, { "epoch": 2.5328793360289423, "grad_norm": 12.125, "learning_rate": 9.171972711768052e-06, "loss": 0.8834, "step": 23804 }, { "epoch": 2.532985741647159, "grad_norm": 13.296875, "learning_rate": 9.171937180216033e-06, "loss": 0.9581, "step": 23805 }, { "epoch": 2.533092147265376, "grad_norm": 16.90625, "learning_rate": 9.171901648664015e-06, "loss": 0.8226, "step": 23806 }, { "epoch": 2.5331985528835923, "grad_norm": 11.4765625, "learning_rate": 9.171866117111996e-06, "loss": 0.8852, "step": 23807 }, { "epoch": 2.533304958501809, "grad_norm": 11.234375, "learning_rate": 9.171830585559978e-06, "loss": 1.0335, "step": 23808 }, { "epoch": 2.5334113641200258, "grad_norm": 16.9375, "learning_rate": 9.17179505400796e-06, "loss": 0.8992, "step": 23809 }, { "epoch": 2.5335177697382423, "grad_norm": 12.859375, "learning_rate": 9.171759522455941e-06, "loss": 0.9959, "step": 23810 }, { "epoch": 2.533624175356459, "grad_norm": 12.484375, "learning_rate": 9.171723990903923e-06, "loss": 1.0286, "step": 23811 }, { "epoch": 2.5337305809746757, "grad_norm": 12.8125, "learning_rate": 9.171688459351904e-06, "loss": 0.9923, "step": 23812 }, { "epoch": 2.5338369865928922, "grad_norm": 14.4765625, "learning_rate": 9.171652927799886e-06, "loss": 0.7615, "step": 23813 }, { "epoch": 2.5339433922111088, "grad_norm": 15.28125, "learning_rate": 9.17161739624787e-06, "loss": 0.8925, "step": 23814 }, { "epoch": 2.5340497978293253, "grad_norm": 10.265625, "learning_rate": 9.171581864695851e-06, "loss": 0.8206, "step": 23815 }, { "epoch": 2.5341562034475418, "grad_norm": 12.5859375, "learning_rate": 9.171546333143833e-06, "loss": 0.9297, "step": 23816 }, { "epoch": 2.5342626090657587, "grad_norm": 16.875, "learning_rate": 9.171510801591814e-06, "loss": 1.2152, "step": 23817 }, { "epoch": 2.5343690146839752, "grad_norm": 13.46875, "learning_rate": 9.171475270039796e-06, "loss": 0.8378, "step": 23818 }, { "epoch": 2.5344754203021918, "grad_norm": 10.6328125, "learning_rate": 9.171439738487777e-06, "loss": 0.9047, "step": 23819 }, { "epoch": 2.5345818259204087, "grad_norm": 12.8828125, "learning_rate": 9.17140420693576e-06, "loss": 1.0482, "step": 23820 }, { "epoch": 2.534688231538625, "grad_norm": 11.7578125, "learning_rate": 9.171368675383742e-06, "loss": 0.8655, "step": 23821 }, { "epoch": 2.5347946371568417, "grad_norm": 12.484375, "learning_rate": 9.171333143831724e-06, "loss": 0.8572, "step": 23822 }, { "epoch": 2.5349010427750587, "grad_norm": 12.703125, "learning_rate": 9.171297612279706e-06, "loss": 1.0787, "step": 23823 }, { "epoch": 2.535007448393275, "grad_norm": 13.1328125, "learning_rate": 9.171262080727687e-06, "loss": 0.905, "step": 23824 }, { "epoch": 2.5351138540114917, "grad_norm": 15.296875, "learning_rate": 9.171226549175669e-06, "loss": 0.9198, "step": 23825 }, { "epoch": 2.5352202596297087, "grad_norm": 22.5, "learning_rate": 9.17119101762365e-06, "loss": 0.8714, "step": 23826 }, { "epoch": 2.535326665247925, "grad_norm": 14.5078125, "learning_rate": 9.171155486071632e-06, "loss": 0.9735, "step": 23827 }, { "epoch": 2.5354330708661417, "grad_norm": 13.0625, "learning_rate": 9.171119954519614e-06, "loss": 0.9008, "step": 23828 }, { "epoch": 2.5355394764843586, "grad_norm": 13.0703125, "learning_rate": 9.171084422967595e-06, "loss": 0.8943, "step": 23829 }, { "epoch": 2.535645882102575, "grad_norm": 16.578125, "learning_rate": 9.171048891415578e-06, "loss": 1.1306, "step": 23830 }, { "epoch": 2.5357522877207916, "grad_norm": 10.90625, "learning_rate": 9.17101335986356e-06, "loss": 0.8624, "step": 23831 }, { "epoch": 2.535858693339008, "grad_norm": 13.8671875, "learning_rate": 9.170977828311542e-06, "loss": 1.0015, "step": 23832 }, { "epoch": 2.535965098957225, "grad_norm": 13.921875, "learning_rate": 9.170942296759523e-06, "loss": 0.8934, "step": 23833 }, { "epoch": 2.5360715045754416, "grad_norm": 9.8125, "learning_rate": 9.170906765207505e-06, "loss": 0.9736, "step": 23834 }, { "epoch": 2.536177910193658, "grad_norm": 8.9765625, "learning_rate": 9.170871233655486e-06, "loss": 0.9186, "step": 23835 }, { "epoch": 2.5362843158118746, "grad_norm": 18.390625, "learning_rate": 9.170835702103468e-06, "loss": 0.8736, "step": 23836 }, { "epoch": 2.5363907214300916, "grad_norm": 19.078125, "learning_rate": 9.17080017055145e-06, "loss": 1.1941, "step": 23837 }, { "epoch": 2.536497127048308, "grad_norm": 11.578125, "learning_rate": 9.170764638999431e-06, "loss": 0.9129, "step": 23838 }, { "epoch": 2.5366035326665246, "grad_norm": 12.890625, "learning_rate": 9.170729107447413e-06, "loss": 1.1002, "step": 23839 }, { "epoch": 2.5367099382847416, "grad_norm": 13.4765625, "learning_rate": 9.170693575895396e-06, "loss": 0.9484, "step": 23840 }, { "epoch": 2.536816343902958, "grad_norm": 14.53125, "learning_rate": 9.170658044343378e-06, "loss": 0.9309, "step": 23841 }, { "epoch": 2.5369227495211746, "grad_norm": 10.7578125, "learning_rate": 9.17062251279136e-06, "loss": 0.9041, "step": 23842 }, { "epoch": 2.5370291551393915, "grad_norm": 13.171875, "learning_rate": 9.170586981239341e-06, "loss": 0.8248, "step": 23843 }, { "epoch": 2.537135560757608, "grad_norm": 12.7734375, "learning_rate": 9.170551449687323e-06, "loss": 0.8975, "step": 23844 }, { "epoch": 2.5372419663758246, "grad_norm": 11.8515625, "learning_rate": 9.170515918135306e-06, "loss": 0.9031, "step": 23845 }, { "epoch": 2.5373483719940415, "grad_norm": 12.296875, "learning_rate": 9.170480386583288e-06, "loss": 0.6287, "step": 23846 }, { "epoch": 2.537454777612258, "grad_norm": 12.3671875, "learning_rate": 9.17044485503127e-06, "loss": 0.8869, "step": 23847 }, { "epoch": 2.5375611832304745, "grad_norm": 19.75, "learning_rate": 9.17040932347925e-06, "loss": 0.8824, "step": 23848 }, { "epoch": 2.5376675888486915, "grad_norm": 12.6328125, "learning_rate": 9.170373791927232e-06, "loss": 0.8793, "step": 23849 }, { "epoch": 2.537773994466908, "grad_norm": 16.328125, "learning_rate": 9.170338260375214e-06, "loss": 0.8957, "step": 23850 }, { "epoch": 2.5378804000851245, "grad_norm": 12.6328125, "learning_rate": 9.170302728823196e-06, "loss": 0.8022, "step": 23851 }, { "epoch": 2.537986805703341, "grad_norm": 10.59375, "learning_rate": 9.170267197271177e-06, "loss": 0.7879, "step": 23852 }, { "epoch": 2.5380932113215575, "grad_norm": 11.375, "learning_rate": 9.170231665719159e-06, "loss": 0.7663, "step": 23853 }, { "epoch": 2.5381996169397745, "grad_norm": 11.21875, "learning_rate": 9.17019613416714e-06, "loss": 0.8749, "step": 23854 }, { "epoch": 2.538306022557991, "grad_norm": 15.421875, "learning_rate": 9.170160602615124e-06, "loss": 0.9298, "step": 23855 }, { "epoch": 2.5384124281762075, "grad_norm": 14.625, "learning_rate": 9.170125071063105e-06, "loss": 0.8792, "step": 23856 }, { "epoch": 2.5385188337944244, "grad_norm": 20.671875, "learning_rate": 9.170089539511087e-06, "loss": 0.9186, "step": 23857 }, { "epoch": 2.538625239412641, "grad_norm": 14.6484375, "learning_rate": 9.170054007959069e-06, "loss": 0.8697, "step": 23858 }, { "epoch": 2.5387316450308575, "grad_norm": 19.0, "learning_rate": 9.17001847640705e-06, "loss": 0.948, "step": 23859 }, { "epoch": 2.5388380506490744, "grad_norm": 17.890625, "learning_rate": 9.169982944855032e-06, "loss": 1.0826, "step": 23860 }, { "epoch": 2.538944456267291, "grad_norm": 10.984375, "learning_rate": 9.169947413303013e-06, "loss": 0.8303, "step": 23861 }, { "epoch": 2.5390508618855074, "grad_norm": 16.21875, "learning_rate": 9.169911881750995e-06, "loss": 1.1239, "step": 23862 }, { "epoch": 2.5391572675037244, "grad_norm": 19.359375, "learning_rate": 9.169876350198977e-06, "loss": 0.8031, "step": 23863 }, { "epoch": 2.539263673121941, "grad_norm": 9.84375, "learning_rate": 9.169840818646958e-06, "loss": 0.8778, "step": 23864 }, { "epoch": 2.5393700787401574, "grad_norm": 16.734375, "learning_rate": 9.16980528709494e-06, "loss": 1.007, "step": 23865 }, { "epoch": 2.5394764843583744, "grad_norm": 17.46875, "learning_rate": 9.169769755542923e-06, "loss": 0.7736, "step": 23866 }, { "epoch": 2.539582889976591, "grad_norm": 12.5078125, "learning_rate": 9.169734223990905e-06, "loss": 0.9158, "step": 23867 }, { "epoch": 2.5396892955948074, "grad_norm": 13.28125, "learning_rate": 9.169698692438886e-06, "loss": 0.9317, "step": 23868 }, { "epoch": 2.539795701213024, "grad_norm": 9.8359375, "learning_rate": 9.169663160886868e-06, "loss": 0.8491, "step": 23869 }, { "epoch": 2.539902106831241, "grad_norm": 9.609375, "learning_rate": 9.16962762933485e-06, "loss": 0.9053, "step": 23870 }, { "epoch": 2.5400085124494574, "grad_norm": 16.5625, "learning_rate": 9.169592097782833e-06, "loss": 0.975, "step": 23871 }, { "epoch": 2.540114918067674, "grad_norm": 12.484375, "learning_rate": 9.169556566230814e-06, "loss": 0.7138, "step": 23872 }, { "epoch": 2.5402213236858904, "grad_norm": 14.234375, "learning_rate": 9.169521034678796e-06, "loss": 1.0153, "step": 23873 }, { "epoch": 2.5403277293041073, "grad_norm": 13.1640625, "learning_rate": 9.169485503126778e-06, "loss": 1.0843, "step": 23874 }, { "epoch": 2.540434134922324, "grad_norm": 13.8203125, "learning_rate": 9.16944997157476e-06, "loss": 1.0058, "step": 23875 }, { "epoch": 2.5405405405405403, "grad_norm": 15.671875, "learning_rate": 9.16941444002274e-06, "loss": 0.9464, "step": 23876 }, { "epoch": 2.5406469461587573, "grad_norm": 13.609375, "learning_rate": 9.169378908470722e-06, "loss": 1.009, "step": 23877 }, { "epoch": 2.540753351776974, "grad_norm": 11.5625, "learning_rate": 9.169343376918704e-06, "loss": 0.8851, "step": 23878 }, { "epoch": 2.5408597573951903, "grad_norm": 16.328125, "learning_rate": 9.169307845366686e-06, "loss": 1.0609, "step": 23879 }, { "epoch": 2.5409661630134073, "grad_norm": 12.2890625, "learning_rate": 9.169272313814667e-06, "loss": 0.955, "step": 23880 }, { "epoch": 2.541072568631624, "grad_norm": 10.375, "learning_rate": 9.16923678226265e-06, "loss": 0.7503, "step": 23881 }, { "epoch": 2.5411789742498403, "grad_norm": 11.6328125, "learning_rate": 9.169201250710632e-06, "loss": 0.9702, "step": 23882 }, { "epoch": 2.5412853798680572, "grad_norm": 12.8359375, "learning_rate": 9.169165719158614e-06, "loss": 0.9485, "step": 23883 }, { "epoch": 2.5413917854862738, "grad_norm": 10.1015625, "learning_rate": 9.169130187606595e-06, "loss": 0.9523, "step": 23884 }, { "epoch": 2.5414981911044903, "grad_norm": 12.3046875, "learning_rate": 9.169094656054577e-06, "loss": 0.7849, "step": 23885 }, { "epoch": 2.541604596722707, "grad_norm": 14.6015625, "learning_rate": 9.16905912450256e-06, "loss": 1.0196, "step": 23886 }, { "epoch": 2.5417110023409237, "grad_norm": 13.0, "learning_rate": 9.169023592950542e-06, "loss": 0.9611, "step": 23887 }, { "epoch": 2.5418174079591402, "grad_norm": 11.5234375, "learning_rate": 9.168988061398523e-06, "loss": 0.9543, "step": 23888 }, { "epoch": 2.5419238135773567, "grad_norm": 16.1875, "learning_rate": 9.168952529846505e-06, "loss": 0.8174, "step": 23889 }, { "epoch": 2.5420302191955733, "grad_norm": 14.3828125, "learning_rate": 9.168916998294487e-06, "loss": 0.8967, "step": 23890 }, { "epoch": 2.54213662481379, "grad_norm": 11.9375, "learning_rate": 9.168881466742468e-06, "loss": 0.8592, "step": 23891 }, { "epoch": 2.5422430304320067, "grad_norm": 11.046875, "learning_rate": 9.16884593519045e-06, "loss": 0.8938, "step": 23892 }, { "epoch": 2.5423494360502232, "grad_norm": 15.6015625, "learning_rate": 9.168810403638432e-06, "loss": 0.8835, "step": 23893 }, { "epoch": 2.54245584166844, "grad_norm": 18.40625, "learning_rate": 9.168774872086413e-06, "loss": 1.0013, "step": 23894 }, { "epoch": 2.5425622472866567, "grad_norm": 11.7890625, "learning_rate": 9.168739340534395e-06, "loss": 0.8613, "step": 23895 }, { "epoch": 2.542668652904873, "grad_norm": 11.5546875, "learning_rate": 9.168703808982376e-06, "loss": 0.779, "step": 23896 }, { "epoch": 2.54277505852309, "grad_norm": 12.2578125, "learning_rate": 9.16866827743036e-06, "loss": 0.9154, "step": 23897 }, { "epoch": 2.5428814641413067, "grad_norm": 11.6328125, "learning_rate": 9.168632745878341e-06, "loss": 0.782, "step": 23898 }, { "epoch": 2.542987869759523, "grad_norm": 12.375, "learning_rate": 9.168597214326323e-06, "loss": 0.8205, "step": 23899 }, { "epoch": 2.54309427537774, "grad_norm": 13.5546875, "learning_rate": 9.168561682774304e-06, "loss": 1.0056, "step": 23900 }, { "epoch": 2.5432006809959566, "grad_norm": 11.171875, "learning_rate": 9.168526151222286e-06, "loss": 0.9011, "step": 23901 }, { "epoch": 2.543307086614173, "grad_norm": 11.5234375, "learning_rate": 9.168490619670268e-06, "loss": 0.9764, "step": 23902 }, { "epoch": 2.54341349223239, "grad_norm": 17.046875, "learning_rate": 9.16845508811825e-06, "loss": 0.8816, "step": 23903 }, { "epoch": 2.5435198978506066, "grad_norm": 13.28125, "learning_rate": 9.168419556566231e-06, "loss": 0.7874, "step": 23904 }, { "epoch": 2.543626303468823, "grad_norm": 11.3046875, "learning_rate": 9.168384025014213e-06, "loss": 0.833, "step": 23905 }, { "epoch": 2.5437327090870396, "grad_norm": 12.9296875, "learning_rate": 9.168348493462194e-06, "loss": 0.7978, "step": 23906 }, { "epoch": 2.5438391147052566, "grad_norm": 12.359375, "learning_rate": 9.168312961910177e-06, "loss": 0.779, "step": 23907 }, { "epoch": 2.543945520323473, "grad_norm": 12.671875, "learning_rate": 9.168277430358159e-06, "loss": 0.9849, "step": 23908 }, { "epoch": 2.5440519259416896, "grad_norm": 11.640625, "learning_rate": 9.16824189880614e-06, "loss": 0.959, "step": 23909 }, { "epoch": 2.544158331559906, "grad_norm": 16.078125, "learning_rate": 9.168206367254122e-06, "loss": 1.0156, "step": 23910 }, { "epoch": 2.544264737178123, "grad_norm": 16.015625, "learning_rate": 9.168170835702104e-06, "loss": 1.0267, "step": 23911 }, { "epoch": 2.5443711427963396, "grad_norm": 13.78125, "learning_rate": 9.168135304150087e-06, "loss": 1.053, "step": 23912 }, { "epoch": 2.544477548414556, "grad_norm": 19.15625, "learning_rate": 9.168099772598069e-06, "loss": 1.2091, "step": 23913 }, { "epoch": 2.544583954032773, "grad_norm": 15.1796875, "learning_rate": 9.16806424104605e-06, "loss": 0.9013, "step": 23914 }, { "epoch": 2.5446903596509896, "grad_norm": 11.3671875, "learning_rate": 9.168028709494032e-06, "loss": 0.8508, "step": 23915 }, { "epoch": 2.544796765269206, "grad_norm": 12.359375, "learning_rate": 9.167993177942014e-06, "loss": 0.9408, "step": 23916 }, { "epoch": 2.544903170887423, "grad_norm": 15.21875, "learning_rate": 9.167957646389995e-06, "loss": 0.9511, "step": 23917 }, { "epoch": 2.5450095765056395, "grad_norm": 11.09375, "learning_rate": 9.167922114837977e-06, "loss": 0.8138, "step": 23918 }, { "epoch": 2.545115982123856, "grad_norm": 15.9765625, "learning_rate": 9.167886583285958e-06, "loss": 0.944, "step": 23919 }, { "epoch": 2.545222387742073, "grad_norm": 11.890625, "learning_rate": 9.16785105173394e-06, "loss": 1.0098, "step": 23920 }, { "epoch": 2.5453287933602895, "grad_norm": 16.0, "learning_rate": 9.167815520181922e-06, "loss": 0.9218, "step": 23921 }, { "epoch": 2.545435198978506, "grad_norm": 12.7421875, "learning_rate": 9.167779988629905e-06, "loss": 0.9591, "step": 23922 }, { "epoch": 2.545541604596723, "grad_norm": 11.6796875, "learning_rate": 9.167744457077887e-06, "loss": 0.9215, "step": 23923 }, { "epoch": 2.5456480102149395, "grad_norm": 10.9765625, "learning_rate": 9.167708925525868e-06, "loss": 0.9563, "step": 23924 }, { "epoch": 2.545754415833156, "grad_norm": 13.5390625, "learning_rate": 9.16767339397385e-06, "loss": 0.9169, "step": 23925 }, { "epoch": 2.5458608214513725, "grad_norm": 11.3359375, "learning_rate": 9.167637862421831e-06, "loss": 0.8181, "step": 23926 }, { "epoch": 2.545967227069589, "grad_norm": 11.890625, "learning_rate": 9.167602330869813e-06, "loss": 0.8727, "step": 23927 }, { "epoch": 2.546073632687806, "grad_norm": 14.875, "learning_rate": 9.167566799317795e-06, "loss": 0.9015, "step": 23928 }, { "epoch": 2.5461800383060225, "grad_norm": 10.59375, "learning_rate": 9.167531267765776e-06, "loss": 0.9162, "step": 23929 }, { "epoch": 2.546286443924239, "grad_norm": 13.3671875, "learning_rate": 9.167495736213758e-06, "loss": 0.8201, "step": 23930 }, { "epoch": 2.546392849542456, "grad_norm": 11.03125, "learning_rate": 9.16746020466174e-06, "loss": 0.6763, "step": 23931 }, { "epoch": 2.5464992551606724, "grad_norm": 9.1484375, "learning_rate": 9.167424673109723e-06, "loss": 0.7606, "step": 23932 }, { "epoch": 2.546605660778889, "grad_norm": 11.15625, "learning_rate": 9.167389141557704e-06, "loss": 0.782, "step": 23933 }, { "epoch": 2.546712066397106, "grad_norm": 15.828125, "learning_rate": 9.167353610005686e-06, "loss": 0.9751, "step": 23934 }, { "epoch": 2.5468184720153224, "grad_norm": 19.84375, "learning_rate": 9.167318078453667e-06, "loss": 1.3026, "step": 23935 }, { "epoch": 2.546924877633539, "grad_norm": 11.5546875, "learning_rate": 9.167282546901649e-06, "loss": 0.7282, "step": 23936 }, { "epoch": 2.547031283251756, "grad_norm": 12.5390625, "learning_rate": 9.16724701534963e-06, "loss": 0.8669, "step": 23937 }, { "epoch": 2.5471376888699724, "grad_norm": 10.7734375, "learning_rate": 9.167211483797614e-06, "loss": 0.7671, "step": 23938 }, { "epoch": 2.547244094488189, "grad_norm": 11.59375, "learning_rate": 9.167175952245596e-06, "loss": 0.8698, "step": 23939 }, { "epoch": 2.547350500106406, "grad_norm": 10.5859375, "learning_rate": 9.167140420693577e-06, "loss": 0.9382, "step": 23940 }, { "epoch": 2.5474569057246224, "grad_norm": 9.09375, "learning_rate": 9.167104889141559e-06, "loss": 0.8586, "step": 23941 }, { "epoch": 2.547563311342839, "grad_norm": 12.5546875, "learning_rate": 9.16706935758954e-06, "loss": 0.8985, "step": 23942 }, { "epoch": 2.547669716961056, "grad_norm": 11.2578125, "learning_rate": 9.167033826037522e-06, "loss": 0.8329, "step": 23943 }, { "epoch": 2.5477761225792723, "grad_norm": 10.9375, "learning_rate": 9.166998294485504e-06, "loss": 0.8147, "step": 23944 }, { "epoch": 2.547882528197489, "grad_norm": 17.609375, "learning_rate": 9.166962762933485e-06, "loss": 0.9493, "step": 23945 }, { "epoch": 2.5479889338157053, "grad_norm": 14.890625, "learning_rate": 9.166927231381467e-06, "loss": 0.9317, "step": 23946 }, { "epoch": 2.548095339433922, "grad_norm": 15.890625, "learning_rate": 9.166891699829448e-06, "loss": 0.873, "step": 23947 }, { "epoch": 2.548201745052139, "grad_norm": 11.859375, "learning_rate": 9.166856168277432e-06, "loss": 0.8383, "step": 23948 }, { "epoch": 2.5483081506703553, "grad_norm": 13.2890625, "learning_rate": 9.166820636725413e-06, "loss": 0.9272, "step": 23949 }, { "epoch": 2.548414556288572, "grad_norm": 10.6640625, "learning_rate": 9.166785105173395e-06, "loss": 0.852, "step": 23950 }, { "epoch": 2.548520961906789, "grad_norm": 12.6875, "learning_rate": 9.166749573621377e-06, "loss": 0.9269, "step": 23951 }, { "epoch": 2.5486273675250053, "grad_norm": 10.140625, "learning_rate": 9.166714042069358e-06, "loss": 0.93, "step": 23952 }, { "epoch": 2.548733773143222, "grad_norm": 12.3828125, "learning_rate": 9.16667851051734e-06, "loss": 0.8262, "step": 23953 }, { "epoch": 2.5488401787614388, "grad_norm": 18.859375, "learning_rate": 9.166642978965321e-06, "loss": 1.0273, "step": 23954 }, { "epoch": 2.5489465843796553, "grad_norm": 10.8515625, "learning_rate": 9.166607447413303e-06, "loss": 0.9768, "step": 23955 }, { "epoch": 2.5490529899978718, "grad_norm": 13.1796875, "learning_rate": 9.166571915861285e-06, "loss": 0.9201, "step": 23956 }, { "epoch": 2.5491593956160887, "grad_norm": 10.0625, "learning_rate": 9.166536384309266e-06, "loss": 0.8663, "step": 23957 }, { "epoch": 2.5492658012343052, "grad_norm": 14.515625, "learning_rate": 9.16650085275725e-06, "loss": 0.9361, "step": 23958 }, { "epoch": 2.5493722068525217, "grad_norm": 17.796875, "learning_rate": 9.166465321205231e-06, "loss": 0.9637, "step": 23959 }, { "epoch": 2.5494786124707387, "grad_norm": 12.890625, "learning_rate": 9.166429789653213e-06, "loss": 0.9535, "step": 23960 }, { "epoch": 2.549585018088955, "grad_norm": 11.890625, "learning_rate": 9.166394258101194e-06, "loss": 0.917, "step": 23961 }, { "epoch": 2.5496914237071717, "grad_norm": 18.5, "learning_rate": 9.166358726549176e-06, "loss": 1.0739, "step": 23962 }, { "epoch": 2.5497978293253882, "grad_norm": 14.8203125, "learning_rate": 9.16632319499716e-06, "loss": 0.8868, "step": 23963 }, { "epoch": 2.549904234943605, "grad_norm": 14.0, "learning_rate": 9.166287663445141e-06, "loss": 0.9878, "step": 23964 }, { "epoch": 2.5500106405618217, "grad_norm": 12.4375, "learning_rate": 9.166252131893122e-06, "loss": 0.9033, "step": 23965 }, { "epoch": 2.550117046180038, "grad_norm": 16.140625, "learning_rate": 9.166216600341104e-06, "loss": 0.8597, "step": 23966 }, { "epoch": 2.5502234517982547, "grad_norm": 12.7265625, "learning_rate": 9.166181068789086e-06, "loss": 0.6692, "step": 23967 }, { "epoch": 2.5503298574164717, "grad_norm": 15.3125, "learning_rate": 9.166145537237067e-06, "loss": 1.1594, "step": 23968 }, { "epoch": 2.550436263034688, "grad_norm": 16.9375, "learning_rate": 9.166110005685049e-06, "loss": 0.9716, "step": 23969 }, { "epoch": 2.5505426686529047, "grad_norm": 11.671875, "learning_rate": 9.16607447413303e-06, "loss": 0.9247, "step": 23970 }, { "epoch": 2.5506490742711216, "grad_norm": 10.7109375, "learning_rate": 9.166038942581012e-06, "loss": 0.7491, "step": 23971 }, { "epoch": 2.550755479889338, "grad_norm": 13.5703125, "learning_rate": 9.166003411028994e-06, "loss": 1.054, "step": 23972 }, { "epoch": 2.5508618855075547, "grad_norm": 16.21875, "learning_rate": 9.165967879476975e-06, "loss": 0.9889, "step": 23973 }, { "epoch": 2.5509682911257716, "grad_norm": 15.5078125, "learning_rate": 9.165932347924959e-06, "loss": 0.8207, "step": 23974 }, { "epoch": 2.551074696743988, "grad_norm": 11.8203125, "learning_rate": 9.16589681637294e-06, "loss": 0.917, "step": 23975 }, { "epoch": 2.5511811023622046, "grad_norm": 13.1328125, "learning_rate": 9.165861284820922e-06, "loss": 0.9478, "step": 23976 }, { "epoch": 2.5512875079804216, "grad_norm": 12.265625, "learning_rate": 9.165825753268903e-06, "loss": 0.9919, "step": 23977 }, { "epoch": 2.551393913598638, "grad_norm": 10.0390625, "learning_rate": 9.165790221716885e-06, "loss": 0.8404, "step": 23978 }, { "epoch": 2.5515003192168546, "grad_norm": 9.4296875, "learning_rate": 9.165754690164868e-06, "loss": 0.7201, "step": 23979 }, { "epoch": 2.5516067248350716, "grad_norm": 13.3671875, "learning_rate": 9.16571915861285e-06, "loss": 0.8836, "step": 23980 }, { "epoch": 2.551713130453288, "grad_norm": 11.59375, "learning_rate": 9.16568362706083e-06, "loss": 0.916, "step": 23981 }, { "epoch": 2.5518195360715046, "grad_norm": 18.21875, "learning_rate": 9.165648095508811e-06, "loss": 0.8027, "step": 23982 }, { "epoch": 2.551925941689721, "grad_norm": 11.4921875, "learning_rate": 9.165612563956793e-06, "loss": 0.8807, "step": 23983 }, { "epoch": 2.5520323473079376, "grad_norm": 13.28125, "learning_rate": 9.165577032404776e-06, "loss": 1.0739, "step": 23984 }, { "epoch": 2.5521387529261546, "grad_norm": 12.8359375, "learning_rate": 9.165541500852758e-06, "loss": 0.8343, "step": 23985 }, { "epoch": 2.552245158544371, "grad_norm": 11.3515625, "learning_rate": 9.16550596930074e-06, "loss": 0.7266, "step": 23986 }, { "epoch": 2.5523515641625876, "grad_norm": 10.0234375, "learning_rate": 9.165470437748721e-06, "loss": 0.7358, "step": 23987 }, { "epoch": 2.5524579697808045, "grad_norm": 12.734375, "learning_rate": 9.165434906196703e-06, "loss": 0.9281, "step": 23988 }, { "epoch": 2.552564375399021, "grad_norm": 12.4609375, "learning_rate": 9.165399374644686e-06, "loss": 0.9023, "step": 23989 }, { "epoch": 2.5526707810172375, "grad_norm": 11.6484375, "learning_rate": 9.165363843092668e-06, "loss": 0.8511, "step": 23990 }, { "epoch": 2.5527771866354545, "grad_norm": 14.2578125, "learning_rate": 9.16532831154065e-06, "loss": 1.0481, "step": 23991 }, { "epoch": 2.552883592253671, "grad_norm": 11.71875, "learning_rate": 9.165292779988631e-06, "loss": 0.8561, "step": 23992 }, { "epoch": 2.5529899978718875, "grad_norm": 13.3046875, "learning_rate": 9.165257248436613e-06, "loss": 0.8383, "step": 23993 }, { "epoch": 2.5530964034901045, "grad_norm": 10.4375, "learning_rate": 9.165221716884594e-06, "loss": 0.9611, "step": 23994 }, { "epoch": 2.553202809108321, "grad_norm": 13.1640625, "learning_rate": 9.165186185332576e-06, "loss": 1.1256, "step": 23995 }, { "epoch": 2.5533092147265375, "grad_norm": 8.734375, "learning_rate": 9.165150653780557e-06, "loss": 0.7777, "step": 23996 }, { "epoch": 2.5534156203447544, "grad_norm": 11.65625, "learning_rate": 9.165115122228539e-06, "loss": 0.7686, "step": 23997 }, { "epoch": 2.553522025962971, "grad_norm": 14.3828125, "learning_rate": 9.16507959067652e-06, "loss": 0.9561, "step": 23998 }, { "epoch": 2.5536284315811875, "grad_norm": 17.59375, "learning_rate": 9.165044059124504e-06, "loss": 0.9999, "step": 23999 }, { "epoch": 2.553734837199404, "grad_norm": 16.4375, "learning_rate": 9.165008527572485e-06, "loss": 0.9028, "step": 24000 }, { "epoch": 2.553841242817621, "grad_norm": 14.3125, "learning_rate": 9.164972996020467e-06, "loss": 0.997, "step": 24001 }, { "epoch": 2.5539476484358374, "grad_norm": 19.609375, "learning_rate": 9.164937464468449e-06, "loss": 1.0292, "step": 24002 }, { "epoch": 2.554054054054054, "grad_norm": 13.3828125, "learning_rate": 9.16490193291643e-06, "loss": 1.0492, "step": 24003 }, { "epoch": 2.5541604596722705, "grad_norm": 12.0234375, "learning_rate": 9.164866401364414e-06, "loss": 0.8988, "step": 24004 }, { "epoch": 2.5542668652904874, "grad_norm": 16.328125, "learning_rate": 9.164830869812395e-06, "loss": 0.888, "step": 24005 }, { "epoch": 2.554373270908704, "grad_norm": 13.1484375, "learning_rate": 9.164795338260377e-06, "loss": 0.8441, "step": 24006 }, { "epoch": 2.5544796765269204, "grad_norm": 10.3203125, "learning_rate": 9.164759806708358e-06, "loss": 0.8785, "step": 24007 }, { "epoch": 2.5545860821451374, "grad_norm": 13.8203125, "learning_rate": 9.16472427515634e-06, "loss": 0.9428, "step": 24008 }, { "epoch": 2.554692487763354, "grad_norm": 14.75, "learning_rate": 9.164688743604322e-06, "loss": 0.9723, "step": 24009 }, { "epoch": 2.5547988933815704, "grad_norm": 12.5859375, "learning_rate": 9.164653212052303e-06, "loss": 0.9753, "step": 24010 }, { "epoch": 2.5549052989997874, "grad_norm": 10.5078125, "learning_rate": 9.164617680500285e-06, "loss": 0.6318, "step": 24011 }, { "epoch": 2.555011704618004, "grad_norm": 14.71875, "learning_rate": 9.164582148948266e-06, "loss": 0.9472, "step": 24012 }, { "epoch": 2.5551181102362204, "grad_norm": 14.078125, "learning_rate": 9.164546617396248e-06, "loss": 0.9518, "step": 24013 }, { "epoch": 2.5552245158544373, "grad_norm": 16.140625, "learning_rate": 9.16451108584423e-06, "loss": 0.9458, "step": 24014 }, { "epoch": 2.555330921472654, "grad_norm": 10.7109375, "learning_rate": 9.164475554292213e-06, "loss": 0.8565, "step": 24015 }, { "epoch": 2.5554373270908703, "grad_norm": 11.8984375, "learning_rate": 9.164440022740195e-06, "loss": 0.7373, "step": 24016 }, { "epoch": 2.5555437327090873, "grad_norm": 16.53125, "learning_rate": 9.164404491188176e-06, "loss": 1.0577, "step": 24017 }, { "epoch": 2.555650138327304, "grad_norm": 15.1640625, "learning_rate": 9.164368959636158e-06, "loss": 1.0305, "step": 24018 }, { "epoch": 2.5557565439455203, "grad_norm": 19.765625, "learning_rate": 9.16433342808414e-06, "loss": 0.911, "step": 24019 }, { "epoch": 2.555862949563737, "grad_norm": 18.359375, "learning_rate": 9.164297896532121e-06, "loss": 0.8483, "step": 24020 }, { "epoch": 2.5559693551819533, "grad_norm": 9.640625, "learning_rate": 9.164262364980103e-06, "loss": 0.7412, "step": 24021 }, { "epoch": 2.5560757608001703, "grad_norm": 15.578125, "learning_rate": 9.164226833428084e-06, "loss": 1.043, "step": 24022 }, { "epoch": 2.556182166418387, "grad_norm": 11.578125, "learning_rate": 9.164191301876066e-06, "loss": 0.986, "step": 24023 }, { "epoch": 2.5562885720366033, "grad_norm": 11.015625, "learning_rate": 9.164155770324047e-06, "loss": 0.9089, "step": 24024 }, { "epoch": 2.5563949776548203, "grad_norm": 12.96875, "learning_rate": 9.16412023877203e-06, "loss": 0.8189, "step": 24025 }, { "epoch": 2.5565013832730368, "grad_norm": 10.5625, "learning_rate": 9.164084707220012e-06, "loss": 0.9483, "step": 24026 }, { "epoch": 2.5566077888912533, "grad_norm": 13.125, "learning_rate": 9.164049175667994e-06, "loss": 0.9881, "step": 24027 }, { "epoch": 2.5567141945094702, "grad_norm": 13.4453125, "learning_rate": 9.164013644115976e-06, "loss": 0.9616, "step": 24028 }, { "epoch": 2.5568206001276867, "grad_norm": 12.625, "learning_rate": 9.163978112563957e-06, "loss": 0.8722, "step": 24029 }, { "epoch": 2.5569270057459033, "grad_norm": 10.3828125, "learning_rate": 9.16394258101194e-06, "loss": 0.8233, "step": 24030 }, { "epoch": 2.55703341136412, "grad_norm": 15.5234375, "learning_rate": 9.163907049459922e-06, "loss": 0.8575, "step": 24031 }, { "epoch": 2.5571398169823367, "grad_norm": 12.9765625, "learning_rate": 9.163871517907904e-06, "loss": 0.8458, "step": 24032 }, { "epoch": 2.5572462226005532, "grad_norm": 11.609375, "learning_rate": 9.163835986355885e-06, "loss": 0.9777, "step": 24033 }, { "epoch": 2.55735262821877, "grad_norm": 10.640625, "learning_rate": 9.163800454803867e-06, "loss": 0.9654, "step": 24034 }, { "epoch": 2.5574590338369867, "grad_norm": 12.4765625, "learning_rate": 9.163764923251848e-06, "loss": 0.8124, "step": 24035 }, { "epoch": 2.557565439455203, "grad_norm": 16.796875, "learning_rate": 9.16372939169983e-06, "loss": 0.8027, "step": 24036 }, { "epoch": 2.5576718450734197, "grad_norm": 11.4140625, "learning_rate": 9.163693860147812e-06, "loss": 0.8427, "step": 24037 }, { "epoch": 2.5577782506916367, "grad_norm": 13.1875, "learning_rate": 9.163658328595793e-06, "loss": 0.9698, "step": 24038 }, { "epoch": 2.557884656309853, "grad_norm": 14.5546875, "learning_rate": 9.163622797043775e-06, "loss": 1.0182, "step": 24039 }, { "epoch": 2.5579910619280697, "grad_norm": 13.0625, "learning_rate": 9.163587265491758e-06, "loss": 0.9925, "step": 24040 }, { "epoch": 2.558097467546286, "grad_norm": 11.875, "learning_rate": 9.16355173393974e-06, "loss": 0.9578, "step": 24041 }, { "epoch": 2.558203873164503, "grad_norm": 15.4296875, "learning_rate": 9.163516202387721e-06, "loss": 1.0453, "step": 24042 }, { "epoch": 2.5583102787827197, "grad_norm": 15.7265625, "learning_rate": 9.163480670835703e-06, "loss": 1.1192, "step": 24043 }, { "epoch": 2.558416684400936, "grad_norm": 370.0, "learning_rate": 9.163445139283685e-06, "loss": 0.8987, "step": 24044 }, { "epoch": 2.558523090019153, "grad_norm": 9.3984375, "learning_rate": 9.163409607731666e-06, "loss": 0.7681, "step": 24045 }, { "epoch": 2.5586294956373696, "grad_norm": 13.2734375, "learning_rate": 9.163374076179648e-06, "loss": 1.0197, "step": 24046 }, { "epoch": 2.558735901255586, "grad_norm": 16.328125, "learning_rate": 9.16333854462763e-06, "loss": 1.0158, "step": 24047 }, { "epoch": 2.558842306873803, "grad_norm": 9.765625, "learning_rate": 9.163303013075611e-06, "loss": 0.6232, "step": 24048 }, { "epoch": 2.5589487124920196, "grad_norm": 10.5234375, "learning_rate": 9.163267481523593e-06, "loss": 0.6309, "step": 24049 }, { "epoch": 2.559055118110236, "grad_norm": 15.4453125, "learning_rate": 9.163231949971576e-06, "loss": 0.7686, "step": 24050 }, { "epoch": 2.559161523728453, "grad_norm": 21.359375, "learning_rate": 9.163196418419558e-06, "loss": 0.9873, "step": 24051 }, { "epoch": 2.5592679293466696, "grad_norm": 14.734375, "learning_rate": 9.16316088686754e-06, "loss": 0.759, "step": 24052 }, { "epoch": 2.559374334964886, "grad_norm": 15.5703125, "learning_rate": 9.16312535531552e-06, "loss": 1.0251, "step": 24053 }, { "epoch": 2.559480740583103, "grad_norm": 15.8515625, "learning_rate": 9.163089823763502e-06, "loss": 0.9379, "step": 24054 }, { "epoch": 2.5595871462013196, "grad_norm": 19.265625, "learning_rate": 9.163054292211484e-06, "loss": 1.241, "step": 24055 }, { "epoch": 2.559693551819536, "grad_norm": 11.0, "learning_rate": 9.163018760659467e-06, "loss": 0.7121, "step": 24056 }, { "epoch": 2.5597999574377526, "grad_norm": 14.6953125, "learning_rate": 9.162983229107449e-06, "loss": 1.0105, "step": 24057 }, { "epoch": 2.559906363055969, "grad_norm": 13.5234375, "learning_rate": 9.16294769755543e-06, "loss": 0.8892, "step": 24058 }, { "epoch": 2.560012768674186, "grad_norm": 10.5078125, "learning_rate": 9.162912166003412e-06, "loss": 0.7146, "step": 24059 }, { "epoch": 2.5601191742924025, "grad_norm": 11.2734375, "learning_rate": 9.162876634451394e-06, "loss": 0.8547, "step": 24060 }, { "epoch": 2.560225579910619, "grad_norm": 12.9765625, "learning_rate": 9.162841102899375e-06, "loss": 0.9399, "step": 24061 }, { "epoch": 2.560331985528836, "grad_norm": 13.75, "learning_rate": 9.162805571347357e-06, "loss": 0.9837, "step": 24062 }, { "epoch": 2.5604383911470525, "grad_norm": 13.609375, "learning_rate": 9.162770039795339e-06, "loss": 0.8075, "step": 24063 }, { "epoch": 2.560544796765269, "grad_norm": 12.2109375, "learning_rate": 9.16273450824332e-06, "loss": 0.9857, "step": 24064 }, { "epoch": 2.560651202383486, "grad_norm": 13.140625, "learning_rate": 9.162698976691302e-06, "loss": 1.0115, "step": 24065 }, { "epoch": 2.5607576080017025, "grad_norm": 16.953125, "learning_rate": 9.162663445139285e-06, "loss": 1.0562, "step": 24066 }, { "epoch": 2.560864013619919, "grad_norm": 17.234375, "learning_rate": 9.162627913587267e-06, "loss": 0.998, "step": 24067 }, { "epoch": 2.560970419238136, "grad_norm": 12.390625, "learning_rate": 9.162592382035248e-06, "loss": 0.9537, "step": 24068 }, { "epoch": 2.5610768248563525, "grad_norm": 13.4921875, "learning_rate": 9.16255685048323e-06, "loss": 1.1342, "step": 24069 }, { "epoch": 2.561183230474569, "grad_norm": 14.1015625, "learning_rate": 9.162521318931211e-06, "loss": 0.9532, "step": 24070 }, { "epoch": 2.561289636092786, "grad_norm": 12.046875, "learning_rate": 9.162485787379193e-06, "loss": 0.9089, "step": 24071 }, { "epoch": 2.5613960417110024, "grad_norm": 14.453125, "learning_rate": 9.162450255827175e-06, "loss": 1.0826, "step": 24072 }, { "epoch": 2.561502447329219, "grad_norm": 11.7890625, "learning_rate": 9.162414724275156e-06, "loss": 1.0241, "step": 24073 }, { "epoch": 2.561608852947436, "grad_norm": 11.7421875, "learning_rate": 9.162379192723138e-06, "loss": 0.9464, "step": 24074 }, { "epoch": 2.5617152585656524, "grad_norm": 13.1796875, "learning_rate": 9.16234366117112e-06, "loss": 0.9775, "step": 24075 }, { "epoch": 2.561821664183869, "grad_norm": 16.265625, "learning_rate": 9.162308129619103e-06, "loss": 0.9127, "step": 24076 }, { "epoch": 2.5619280698020854, "grad_norm": 13.3671875, "learning_rate": 9.162272598067084e-06, "loss": 0.8722, "step": 24077 }, { "epoch": 2.562034475420302, "grad_norm": 13.625, "learning_rate": 9.162237066515066e-06, "loss": 0.909, "step": 24078 }, { "epoch": 2.562140881038519, "grad_norm": 15.53125, "learning_rate": 9.162201534963048e-06, "loss": 0.9279, "step": 24079 }, { "epoch": 2.5622472866567354, "grad_norm": 14.3671875, "learning_rate": 9.16216600341103e-06, "loss": 0.8683, "step": 24080 }, { "epoch": 2.562353692274952, "grad_norm": 13.2109375, "learning_rate": 9.162130471859013e-06, "loss": 0.8906, "step": 24081 }, { "epoch": 2.562460097893169, "grad_norm": 11.8828125, "learning_rate": 9.162094940306994e-06, "loss": 1.0226, "step": 24082 }, { "epoch": 2.5625665035113854, "grad_norm": 14.421875, "learning_rate": 9.162059408754976e-06, "loss": 1.0012, "step": 24083 }, { "epoch": 2.562672909129602, "grad_norm": 14.9609375, "learning_rate": 9.162023877202957e-06, "loss": 0.9922, "step": 24084 }, { "epoch": 2.562779314747819, "grad_norm": 10.6875, "learning_rate": 9.161988345650939e-06, "loss": 0.7593, "step": 24085 }, { "epoch": 2.5628857203660353, "grad_norm": 18.984375, "learning_rate": 9.16195281409892e-06, "loss": 1.1452, "step": 24086 }, { "epoch": 2.562992125984252, "grad_norm": 15.9140625, "learning_rate": 9.161917282546902e-06, "loss": 0.8354, "step": 24087 }, { "epoch": 2.563098531602469, "grad_norm": 11.15625, "learning_rate": 9.161881750994884e-06, "loss": 0.8731, "step": 24088 }, { "epoch": 2.5632049372206853, "grad_norm": 18.46875, "learning_rate": 9.161846219442865e-06, "loss": 0.8993, "step": 24089 }, { "epoch": 2.563311342838902, "grad_norm": 12.4921875, "learning_rate": 9.161810687890847e-06, "loss": 0.8688, "step": 24090 }, { "epoch": 2.563417748457119, "grad_norm": 16.25, "learning_rate": 9.161775156338829e-06, "loss": 0.8831, "step": 24091 }, { "epoch": 2.5635241540753353, "grad_norm": 9.8359375, "learning_rate": 9.161739624786812e-06, "loss": 0.8895, "step": 24092 }, { "epoch": 2.563630559693552, "grad_norm": 15.6640625, "learning_rate": 9.161704093234794e-06, "loss": 0.912, "step": 24093 }, { "epoch": 2.5637369653117683, "grad_norm": 15.75, "learning_rate": 9.161668561682775e-06, "loss": 0.8957, "step": 24094 }, { "epoch": 2.5638433709299853, "grad_norm": 15.5, "learning_rate": 9.161633030130757e-06, "loss": 0.9637, "step": 24095 }, { "epoch": 2.5639497765482018, "grad_norm": 14.53125, "learning_rate": 9.161597498578738e-06, "loss": 0.9201, "step": 24096 }, { "epoch": 2.5640561821664183, "grad_norm": 16.53125, "learning_rate": 9.161561967026722e-06, "loss": 0.9449, "step": 24097 }, { "epoch": 2.564162587784635, "grad_norm": 13.6875, "learning_rate": 9.161526435474703e-06, "loss": 0.8921, "step": 24098 }, { "epoch": 2.5642689934028517, "grad_norm": 12.71875, "learning_rate": 9.161490903922685e-06, "loss": 0.8245, "step": 24099 }, { "epoch": 2.5643753990210683, "grad_norm": 14.078125, "learning_rate": 9.161455372370666e-06, "loss": 1.0876, "step": 24100 }, { "epoch": 2.5644818046392848, "grad_norm": 11.8359375, "learning_rate": 9.161419840818646e-06, "loss": 0.9007, "step": 24101 }, { "epoch": 2.5645882102575017, "grad_norm": 12.0625, "learning_rate": 9.16138430926663e-06, "loss": 0.8618, "step": 24102 }, { "epoch": 2.5646946158757182, "grad_norm": 17.46875, "learning_rate": 9.161348777714611e-06, "loss": 0.8141, "step": 24103 }, { "epoch": 2.5648010214939347, "grad_norm": 14.734375, "learning_rate": 9.161313246162593e-06, "loss": 0.9834, "step": 24104 }, { "epoch": 2.5649074271121517, "grad_norm": 16.765625, "learning_rate": 9.161277714610574e-06, "loss": 0.8051, "step": 24105 }, { "epoch": 2.565013832730368, "grad_norm": 12.4921875, "learning_rate": 9.161242183058556e-06, "loss": 0.9206, "step": 24106 }, { "epoch": 2.5651202383485847, "grad_norm": 21.296875, "learning_rate": 9.16120665150654e-06, "loss": 0.9655, "step": 24107 }, { "epoch": 2.5652266439668017, "grad_norm": 11.421875, "learning_rate": 9.161171119954521e-06, "loss": 0.846, "step": 24108 }, { "epoch": 2.565333049585018, "grad_norm": 16.921875, "learning_rate": 9.161135588402503e-06, "loss": 0.9444, "step": 24109 }, { "epoch": 2.5654394552032347, "grad_norm": 13.2109375, "learning_rate": 9.161100056850484e-06, "loss": 0.8678, "step": 24110 }, { "epoch": 2.5655458608214516, "grad_norm": 10.6875, "learning_rate": 9.161064525298466e-06, "loss": 0.8194, "step": 24111 }, { "epoch": 2.565652266439668, "grad_norm": 14.375, "learning_rate": 9.161028993746447e-06, "loss": 1.1396, "step": 24112 }, { "epoch": 2.5657586720578847, "grad_norm": 13.265625, "learning_rate": 9.160993462194429e-06, "loss": 1.0361, "step": 24113 }, { "epoch": 2.565865077676101, "grad_norm": 13.734375, "learning_rate": 9.16095793064241e-06, "loss": 1.0349, "step": 24114 }, { "epoch": 2.5659714832943177, "grad_norm": 23.765625, "learning_rate": 9.160922399090392e-06, "loss": 1.0389, "step": 24115 }, { "epoch": 2.5660778889125346, "grad_norm": 13.6171875, "learning_rate": 9.160886867538374e-06, "loss": 0.9095, "step": 24116 }, { "epoch": 2.566184294530751, "grad_norm": 17.6875, "learning_rate": 9.160851335986357e-06, "loss": 0.895, "step": 24117 }, { "epoch": 2.5662907001489677, "grad_norm": 14.8203125, "learning_rate": 9.160815804434339e-06, "loss": 1.0005, "step": 24118 }, { "epoch": 2.5663971057671846, "grad_norm": 12.0234375, "learning_rate": 9.16078027288232e-06, "loss": 0.9386, "step": 24119 }, { "epoch": 2.566503511385401, "grad_norm": 11.203125, "learning_rate": 9.160744741330302e-06, "loss": 0.9096, "step": 24120 }, { "epoch": 2.5666099170036176, "grad_norm": 12.875, "learning_rate": 9.160709209778284e-06, "loss": 0.8124, "step": 24121 }, { "epoch": 2.5667163226218346, "grad_norm": 12.703125, "learning_rate": 9.160673678226267e-06, "loss": 0.8462, "step": 24122 }, { "epoch": 2.566822728240051, "grad_norm": 15.2109375, "learning_rate": 9.160638146674248e-06, "loss": 1.0676, "step": 24123 }, { "epoch": 2.5669291338582676, "grad_norm": 11.0859375, "learning_rate": 9.16060261512223e-06, "loss": 0.7174, "step": 24124 }, { "epoch": 2.5670355394764846, "grad_norm": 16.796875, "learning_rate": 9.160567083570212e-06, "loss": 1.0993, "step": 24125 }, { "epoch": 2.567141945094701, "grad_norm": 14.8203125, "learning_rate": 9.160531552018193e-06, "loss": 0.9128, "step": 24126 }, { "epoch": 2.5672483507129176, "grad_norm": 9.1484375, "learning_rate": 9.160496020466175e-06, "loss": 0.8564, "step": 24127 }, { "epoch": 2.5673547563311345, "grad_norm": 13.0625, "learning_rate": 9.160460488914157e-06, "loss": 0.9937, "step": 24128 }, { "epoch": 2.567461161949351, "grad_norm": 12.9453125, "learning_rate": 9.160424957362138e-06, "loss": 0.8914, "step": 24129 }, { "epoch": 2.5675675675675675, "grad_norm": 13.2109375, "learning_rate": 9.16038942581012e-06, "loss": 0.9504, "step": 24130 }, { "epoch": 2.567673973185784, "grad_norm": 13.2265625, "learning_rate": 9.160353894258101e-06, "loss": 0.9656, "step": 24131 }, { "epoch": 2.567780378804001, "grad_norm": 12.015625, "learning_rate": 9.160318362706083e-06, "loss": 0.9556, "step": 24132 }, { "epoch": 2.5678867844222175, "grad_norm": 15.578125, "learning_rate": 9.160282831154066e-06, "loss": 1.0055, "step": 24133 }, { "epoch": 2.567993190040434, "grad_norm": 13.0625, "learning_rate": 9.160247299602048e-06, "loss": 1.0024, "step": 24134 }, { "epoch": 2.5680995956586505, "grad_norm": 15.0546875, "learning_rate": 9.16021176805003e-06, "loss": 0.6999, "step": 24135 }, { "epoch": 2.5682060012768675, "grad_norm": 17.203125, "learning_rate": 9.160176236498011e-06, "loss": 0.8994, "step": 24136 }, { "epoch": 2.568312406895084, "grad_norm": 17.46875, "learning_rate": 9.160140704945993e-06, "loss": 0.8811, "step": 24137 }, { "epoch": 2.5684188125133005, "grad_norm": 13.5234375, "learning_rate": 9.160105173393974e-06, "loss": 0.9796, "step": 24138 }, { "epoch": 2.5685252181315175, "grad_norm": 14.4609375, "learning_rate": 9.160069641841956e-06, "loss": 0.9406, "step": 24139 }, { "epoch": 2.568631623749734, "grad_norm": 15.3359375, "learning_rate": 9.160034110289938e-06, "loss": 0.9701, "step": 24140 }, { "epoch": 2.5687380293679505, "grad_norm": 13.375, "learning_rate": 9.159998578737919e-06, "loss": 0.9937, "step": 24141 }, { "epoch": 2.5688444349861674, "grad_norm": 12.5625, "learning_rate": 9.1599630471859e-06, "loss": 0.7749, "step": 24142 }, { "epoch": 2.568950840604384, "grad_norm": 13.5703125, "learning_rate": 9.159927515633884e-06, "loss": 0.8292, "step": 24143 }, { "epoch": 2.5690572462226005, "grad_norm": 11.265625, "learning_rate": 9.159891984081866e-06, "loss": 0.8686, "step": 24144 }, { "epoch": 2.5691636518408174, "grad_norm": 14.0703125, "learning_rate": 9.159856452529847e-06, "loss": 0.9688, "step": 24145 }, { "epoch": 2.569270057459034, "grad_norm": 14.2578125, "learning_rate": 9.159820920977829e-06, "loss": 1.0416, "step": 24146 }, { "epoch": 2.5693764630772504, "grad_norm": 17.0, "learning_rate": 9.15978538942581e-06, "loss": 1.0582, "step": 24147 }, { "epoch": 2.5694828686954674, "grad_norm": 11.8828125, "learning_rate": 9.159749857873794e-06, "loss": 0.8874, "step": 24148 }, { "epoch": 2.569589274313684, "grad_norm": 14.6953125, "learning_rate": 9.159714326321775e-06, "loss": 0.9593, "step": 24149 }, { "epoch": 2.5696956799319004, "grad_norm": 12.65625, "learning_rate": 9.159678794769757e-06, "loss": 0.916, "step": 24150 }, { "epoch": 2.569802085550117, "grad_norm": 13.2265625, "learning_rate": 9.159643263217739e-06, "loss": 0.8923, "step": 24151 }, { "epoch": 2.5699084911683334, "grad_norm": 18.28125, "learning_rate": 9.15960773166572e-06, "loss": 1.1709, "step": 24152 }, { "epoch": 2.5700148967865504, "grad_norm": 10.1171875, "learning_rate": 9.159572200113702e-06, "loss": 0.8461, "step": 24153 }, { "epoch": 2.570121302404767, "grad_norm": 10.5, "learning_rate": 9.159536668561683e-06, "loss": 0.8306, "step": 24154 }, { "epoch": 2.5702277080229834, "grad_norm": 16.203125, "learning_rate": 9.159501137009665e-06, "loss": 0.9505, "step": 24155 }, { "epoch": 2.5703341136412003, "grad_norm": 14.3984375, "learning_rate": 9.159465605457647e-06, "loss": 1.0086, "step": 24156 }, { "epoch": 2.570440519259417, "grad_norm": 13.046875, "learning_rate": 9.159430073905628e-06, "loss": 0.9199, "step": 24157 }, { "epoch": 2.5705469248776334, "grad_norm": 13.8671875, "learning_rate": 9.159394542353612e-06, "loss": 0.916, "step": 24158 }, { "epoch": 2.5706533304958503, "grad_norm": 12.3671875, "learning_rate": 9.159359010801593e-06, "loss": 1.0373, "step": 24159 }, { "epoch": 2.570759736114067, "grad_norm": 12.421875, "learning_rate": 9.159323479249575e-06, "loss": 0.8358, "step": 24160 }, { "epoch": 2.5708661417322833, "grad_norm": 11.375, "learning_rate": 9.159287947697556e-06, "loss": 0.8586, "step": 24161 }, { "epoch": 2.5709725473505003, "grad_norm": 31.0625, "learning_rate": 9.159252416145538e-06, "loss": 0.9397, "step": 24162 }, { "epoch": 2.571078952968717, "grad_norm": 11.828125, "learning_rate": 9.15921688459352e-06, "loss": 0.9904, "step": 24163 }, { "epoch": 2.5711853585869333, "grad_norm": 17.3125, "learning_rate": 9.159181353041501e-06, "loss": 0.8248, "step": 24164 }, { "epoch": 2.5712917642051503, "grad_norm": 15.6015625, "learning_rate": 9.159145821489483e-06, "loss": 0.9244, "step": 24165 }, { "epoch": 2.5713981698233668, "grad_norm": 16.21875, "learning_rate": 9.159110289937464e-06, "loss": 1.0179, "step": 24166 }, { "epoch": 2.5715045754415833, "grad_norm": 17.96875, "learning_rate": 9.159074758385446e-06, "loss": 0.8821, "step": 24167 }, { "epoch": 2.5716109810598, "grad_norm": 10.3359375, "learning_rate": 9.15903922683343e-06, "loss": 0.8814, "step": 24168 }, { "epoch": 2.5717173866780167, "grad_norm": 11.0625, "learning_rate": 9.159003695281411e-06, "loss": 0.8362, "step": 24169 }, { "epoch": 2.5718237922962333, "grad_norm": 12.65625, "learning_rate": 9.158968163729392e-06, "loss": 0.8955, "step": 24170 }, { "epoch": 2.5719301979144498, "grad_norm": 13.3984375, "learning_rate": 9.158932632177374e-06, "loss": 0.9188, "step": 24171 }, { "epoch": 2.5720366035326663, "grad_norm": 11.2578125, "learning_rate": 9.158897100625356e-06, "loss": 0.7908, "step": 24172 }, { "epoch": 2.5721430091508832, "grad_norm": 11.2578125, "learning_rate": 9.158861569073337e-06, "loss": 0.8309, "step": 24173 }, { "epoch": 2.5722494147690997, "grad_norm": 11.140625, "learning_rate": 9.15882603752132e-06, "loss": 0.897, "step": 24174 }, { "epoch": 2.5723558203873162, "grad_norm": 11.859375, "learning_rate": 9.158790505969302e-06, "loss": 1.0574, "step": 24175 }, { "epoch": 2.572462226005533, "grad_norm": 13.9609375, "learning_rate": 9.158754974417284e-06, "loss": 0.8523, "step": 24176 }, { "epoch": 2.5725686316237497, "grad_norm": 15.59375, "learning_rate": 9.158719442865265e-06, "loss": 1.0597, "step": 24177 }, { "epoch": 2.5726750372419662, "grad_norm": 13.8515625, "learning_rate": 9.158683911313247e-06, "loss": 1.0087, "step": 24178 }, { "epoch": 2.572781442860183, "grad_norm": 16.21875, "learning_rate": 9.158648379761229e-06, "loss": 1.0686, "step": 24179 }, { "epoch": 2.5728878484783997, "grad_norm": 12.1015625, "learning_rate": 9.15861284820921e-06, "loss": 0.8267, "step": 24180 }, { "epoch": 2.572994254096616, "grad_norm": 13.8515625, "learning_rate": 9.158577316657192e-06, "loss": 1.1742, "step": 24181 }, { "epoch": 2.573100659714833, "grad_norm": 9.71875, "learning_rate": 9.158541785105173e-06, "loss": 0.8059, "step": 24182 }, { "epoch": 2.5732070653330497, "grad_norm": 12.9765625, "learning_rate": 9.158506253553155e-06, "loss": 0.8399, "step": 24183 }, { "epoch": 2.573313470951266, "grad_norm": 14.5390625, "learning_rate": 9.158470722001138e-06, "loss": 0.9245, "step": 24184 }, { "epoch": 2.573419876569483, "grad_norm": 15.0078125, "learning_rate": 9.15843519044912e-06, "loss": 0.87, "step": 24185 }, { "epoch": 2.5735262821876996, "grad_norm": 13.3359375, "learning_rate": 9.158399658897102e-06, "loss": 0.9698, "step": 24186 }, { "epoch": 2.573632687805916, "grad_norm": 12.640625, "learning_rate": 9.158364127345083e-06, "loss": 1.0305, "step": 24187 }, { "epoch": 2.5737390934241327, "grad_norm": 11.9296875, "learning_rate": 9.158328595793065e-06, "loss": 0.8407, "step": 24188 }, { "epoch": 2.573845499042349, "grad_norm": 14.96875, "learning_rate": 9.158293064241048e-06, "loss": 1.0484, "step": 24189 }, { "epoch": 2.573951904660566, "grad_norm": 12.8203125, "learning_rate": 9.158257532689028e-06, "loss": 0.8631, "step": 24190 }, { "epoch": 2.5740583102787826, "grad_norm": 12.40625, "learning_rate": 9.15822200113701e-06, "loss": 0.8524, "step": 24191 }, { "epoch": 2.574164715896999, "grad_norm": 13.671875, "learning_rate": 9.158186469584991e-06, "loss": 0.9562, "step": 24192 }, { "epoch": 2.574271121515216, "grad_norm": 11.375, "learning_rate": 9.158150938032973e-06, "loss": 0.7359, "step": 24193 }, { "epoch": 2.5743775271334326, "grad_norm": 12.96875, "learning_rate": 9.158115406480956e-06, "loss": 0.8851, "step": 24194 }, { "epoch": 2.574483932751649, "grad_norm": 16.546875, "learning_rate": 9.158079874928938e-06, "loss": 0.8062, "step": 24195 }, { "epoch": 2.574590338369866, "grad_norm": 12.515625, "learning_rate": 9.15804434337692e-06, "loss": 0.9765, "step": 24196 }, { "epoch": 2.5746967439880826, "grad_norm": 11.6796875, "learning_rate": 9.158008811824901e-06, "loss": 0.8183, "step": 24197 }, { "epoch": 2.574803149606299, "grad_norm": 11.5, "learning_rate": 9.157973280272883e-06, "loss": 0.8003, "step": 24198 }, { "epoch": 2.574909555224516, "grad_norm": 11.7265625, "learning_rate": 9.157937748720866e-06, "loss": 0.9771, "step": 24199 }, { "epoch": 2.5750159608427325, "grad_norm": 12.6953125, "learning_rate": 9.157902217168847e-06, "loss": 0.9028, "step": 24200 }, { "epoch": 2.575122366460949, "grad_norm": 11.7578125, "learning_rate": 9.157866685616829e-06, "loss": 1.0075, "step": 24201 }, { "epoch": 2.575228772079166, "grad_norm": 22.375, "learning_rate": 9.15783115406481e-06, "loss": 1.0016, "step": 24202 }, { "epoch": 2.5753351776973825, "grad_norm": 17.328125, "learning_rate": 9.157795622512792e-06, "loss": 1.0372, "step": 24203 }, { "epoch": 2.575441583315599, "grad_norm": 13.1796875, "learning_rate": 9.157760090960774e-06, "loss": 1.1096, "step": 24204 }, { "epoch": 2.575547988933816, "grad_norm": 21.5625, "learning_rate": 9.157724559408755e-06, "loss": 1.071, "step": 24205 }, { "epoch": 2.5756543945520325, "grad_norm": 10.9296875, "learning_rate": 9.157689027856737e-06, "loss": 1.0143, "step": 24206 }, { "epoch": 2.575760800170249, "grad_norm": 25.15625, "learning_rate": 9.157653496304719e-06, "loss": 0.8566, "step": 24207 }, { "epoch": 2.5758672057884655, "grad_norm": 14.3203125, "learning_rate": 9.1576179647527e-06, "loss": 0.9365, "step": 24208 }, { "epoch": 2.575973611406682, "grad_norm": 14.203125, "learning_rate": 9.157582433200682e-06, "loss": 0.7608, "step": 24209 }, { "epoch": 2.576080017024899, "grad_norm": 11.953125, "learning_rate": 9.157546901648665e-06, "loss": 0.8449, "step": 24210 }, { "epoch": 2.5761864226431155, "grad_norm": 11.453125, "learning_rate": 9.157511370096647e-06, "loss": 0.9196, "step": 24211 }, { "epoch": 2.576292828261332, "grad_norm": 12.0703125, "learning_rate": 9.157475838544628e-06, "loss": 0.9452, "step": 24212 }, { "epoch": 2.576399233879549, "grad_norm": 13.765625, "learning_rate": 9.15744030699261e-06, "loss": 0.8008, "step": 24213 }, { "epoch": 2.5765056394977655, "grad_norm": 11.4375, "learning_rate": 9.157404775440592e-06, "loss": 0.8607, "step": 24214 }, { "epoch": 2.576612045115982, "grad_norm": 13.3203125, "learning_rate": 9.157369243888575e-06, "loss": 0.8499, "step": 24215 }, { "epoch": 2.576718450734199, "grad_norm": 12.546875, "learning_rate": 9.157333712336557e-06, "loss": 1.0277, "step": 24216 }, { "epoch": 2.5768248563524154, "grad_norm": 11.03125, "learning_rate": 9.157298180784538e-06, "loss": 0.9391, "step": 24217 }, { "epoch": 2.576931261970632, "grad_norm": 13.0703125, "learning_rate": 9.15726264923252e-06, "loss": 0.9929, "step": 24218 }, { "epoch": 2.577037667588849, "grad_norm": 13.2578125, "learning_rate": 9.157227117680501e-06, "loss": 0.7717, "step": 24219 }, { "epoch": 2.5771440732070654, "grad_norm": 12.4765625, "learning_rate": 9.157191586128483e-06, "loss": 0.9144, "step": 24220 }, { "epoch": 2.577250478825282, "grad_norm": 12.8984375, "learning_rate": 9.157156054576465e-06, "loss": 0.7869, "step": 24221 }, { "epoch": 2.577356884443499, "grad_norm": 10.6484375, "learning_rate": 9.157120523024446e-06, "loss": 0.8037, "step": 24222 }, { "epoch": 2.5774632900617154, "grad_norm": 19.359375, "learning_rate": 9.157084991472428e-06, "loss": 1.1303, "step": 24223 }, { "epoch": 2.577569695679932, "grad_norm": 14.875, "learning_rate": 9.15704945992041e-06, "loss": 0.9899, "step": 24224 }, { "epoch": 2.5776761012981484, "grad_norm": 16.109375, "learning_rate": 9.157013928368393e-06, "loss": 1.0213, "step": 24225 }, { "epoch": 2.5777825069163653, "grad_norm": 13.9609375, "learning_rate": 9.156978396816374e-06, "loss": 0.8729, "step": 24226 }, { "epoch": 2.577888912534582, "grad_norm": 12.6953125, "learning_rate": 9.156942865264356e-06, "loss": 1.0484, "step": 24227 }, { "epoch": 2.5779953181527984, "grad_norm": 12.0234375, "learning_rate": 9.156907333712338e-06, "loss": 0.8028, "step": 24228 }, { "epoch": 2.578101723771015, "grad_norm": 13.6875, "learning_rate": 9.156871802160319e-06, "loss": 0.7948, "step": 24229 }, { "epoch": 2.578208129389232, "grad_norm": 10.625, "learning_rate": 9.1568362706083e-06, "loss": 0.8562, "step": 24230 }, { "epoch": 2.5783145350074483, "grad_norm": 12.8828125, "learning_rate": 9.156800739056282e-06, "loss": 0.7317, "step": 24231 }, { "epoch": 2.578420940625665, "grad_norm": 14.359375, "learning_rate": 9.156765207504264e-06, "loss": 0.9875, "step": 24232 }, { "epoch": 2.578527346243882, "grad_norm": 13.7890625, "learning_rate": 9.156729675952246e-06, "loss": 0.9193, "step": 24233 }, { "epoch": 2.5786337518620983, "grad_norm": 13.0390625, "learning_rate": 9.156694144400227e-06, "loss": 0.8357, "step": 24234 }, { "epoch": 2.578740157480315, "grad_norm": 10.2109375, "learning_rate": 9.15665861284821e-06, "loss": 0.7362, "step": 24235 }, { "epoch": 2.5788465630985318, "grad_norm": 16.625, "learning_rate": 9.156623081296192e-06, "loss": 0.94, "step": 24236 }, { "epoch": 2.5789529687167483, "grad_norm": 12.359375, "learning_rate": 9.156587549744174e-06, "loss": 0.8301, "step": 24237 }, { "epoch": 2.579059374334965, "grad_norm": 12.21875, "learning_rate": 9.156552018192155e-06, "loss": 0.7726, "step": 24238 }, { "epoch": 2.5791657799531817, "grad_norm": 18.765625, "learning_rate": 9.156516486640137e-06, "loss": 0.9527, "step": 24239 }, { "epoch": 2.5792721855713983, "grad_norm": 14.3203125, "learning_rate": 9.15648095508812e-06, "loss": 1.0684, "step": 24240 }, { "epoch": 2.5793785911896148, "grad_norm": 15.8359375, "learning_rate": 9.156445423536102e-06, "loss": 1.0429, "step": 24241 }, { "epoch": 2.5794849968078317, "grad_norm": 15.4140625, "learning_rate": 9.156409891984083e-06, "loss": 0.7723, "step": 24242 }, { "epoch": 2.5795914024260482, "grad_norm": 12.234375, "learning_rate": 9.156374360432065e-06, "loss": 0.9338, "step": 24243 }, { "epoch": 2.5796978080442647, "grad_norm": 15.75, "learning_rate": 9.156338828880047e-06, "loss": 0.884, "step": 24244 }, { "epoch": 2.5798042136624812, "grad_norm": 17.5, "learning_rate": 9.156303297328028e-06, "loss": 1.2303, "step": 24245 }, { "epoch": 2.5799106192806978, "grad_norm": 14.3125, "learning_rate": 9.15626776577601e-06, "loss": 0.8619, "step": 24246 }, { "epoch": 2.5800170248989147, "grad_norm": 14.8828125, "learning_rate": 9.156232234223991e-06, "loss": 1.0553, "step": 24247 }, { "epoch": 2.5801234305171312, "grad_norm": 11.6796875, "learning_rate": 9.156196702671973e-06, "loss": 0.8691, "step": 24248 }, { "epoch": 2.5802298361353477, "grad_norm": 12.90625, "learning_rate": 9.156161171119955e-06, "loss": 0.8469, "step": 24249 }, { "epoch": 2.5803362417535647, "grad_norm": 16.78125, "learning_rate": 9.156125639567936e-06, "loss": 0.9615, "step": 24250 }, { "epoch": 2.580442647371781, "grad_norm": 12.390625, "learning_rate": 9.15609010801592e-06, "loss": 0.8031, "step": 24251 }, { "epoch": 2.5805490529899977, "grad_norm": 11.921875, "learning_rate": 9.156054576463901e-06, "loss": 1.0916, "step": 24252 }, { "epoch": 2.5806554586082147, "grad_norm": 12.578125, "learning_rate": 9.156019044911883e-06, "loss": 0.7808, "step": 24253 }, { "epoch": 2.580761864226431, "grad_norm": 12.359375, "learning_rate": 9.155983513359864e-06, "loss": 0.9824, "step": 24254 }, { "epoch": 2.5808682698446477, "grad_norm": 10.8515625, "learning_rate": 9.155947981807846e-06, "loss": 0.9531, "step": 24255 }, { "epoch": 2.5809746754628646, "grad_norm": 11.7890625, "learning_rate": 9.155912450255828e-06, "loss": 0.9385, "step": 24256 }, { "epoch": 2.581081081081081, "grad_norm": 11.3359375, "learning_rate": 9.15587691870381e-06, "loss": 0.9084, "step": 24257 }, { "epoch": 2.5811874866992977, "grad_norm": 16.515625, "learning_rate": 9.15584138715179e-06, "loss": 0.8674, "step": 24258 }, { "epoch": 2.5812938923175146, "grad_norm": 14.390625, "learning_rate": 9.155805855599772e-06, "loss": 0.9896, "step": 24259 }, { "epoch": 2.581400297935731, "grad_norm": 12.0859375, "learning_rate": 9.155770324047754e-06, "loss": 0.9657, "step": 24260 }, { "epoch": 2.5815067035539476, "grad_norm": 12.6875, "learning_rate": 9.155734792495737e-06, "loss": 1.0058, "step": 24261 }, { "epoch": 2.581613109172164, "grad_norm": 9.2109375, "learning_rate": 9.155699260943719e-06, "loss": 0.7332, "step": 24262 }, { "epoch": 2.581719514790381, "grad_norm": 10.6953125, "learning_rate": 9.1556637293917e-06, "loss": 0.8006, "step": 24263 }, { "epoch": 2.5818259204085976, "grad_norm": 15.921875, "learning_rate": 9.155628197839682e-06, "loss": 1.0187, "step": 24264 }, { "epoch": 2.581932326026814, "grad_norm": 13.203125, "learning_rate": 9.155592666287664e-06, "loss": 0.6935, "step": 24265 }, { "epoch": 2.5820387316450306, "grad_norm": 13.2421875, "learning_rate": 9.155557134735647e-06, "loss": 0.8255, "step": 24266 }, { "epoch": 2.5821451372632476, "grad_norm": 14.4765625, "learning_rate": 9.155521603183629e-06, "loss": 0.6129, "step": 24267 }, { "epoch": 2.582251542881464, "grad_norm": 10.4609375, "learning_rate": 9.15548607163161e-06, "loss": 0.8075, "step": 24268 }, { "epoch": 2.5823579484996806, "grad_norm": 10.34375, "learning_rate": 9.155450540079592e-06, "loss": 0.8054, "step": 24269 }, { "epoch": 2.5824643541178975, "grad_norm": 14.9296875, "learning_rate": 9.155415008527573e-06, "loss": 1.0532, "step": 24270 }, { "epoch": 2.582570759736114, "grad_norm": 11.890625, "learning_rate": 9.155379476975555e-06, "loss": 0.9177, "step": 24271 }, { "epoch": 2.5826771653543306, "grad_norm": 11.90625, "learning_rate": 9.155343945423537e-06, "loss": 0.8074, "step": 24272 }, { "epoch": 2.5827835709725475, "grad_norm": 9.546875, "learning_rate": 9.155308413871518e-06, "loss": 0.8047, "step": 24273 }, { "epoch": 2.582889976590764, "grad_norm": 12.1953125, "learning_rate": 9.1552728823195e-06, "loss": 0.8502, "step": 24274 }, { "epoch": 2.5829963822089805, "grad_norm": 11.984375, "learning_rate": 9.155237350767482e-06, "loss": 0.9024, "step": 24275 }, { "epoch": 2.5831027878271975, "grad_norm": 15.7421875, "learning_rate": 9.155201819215465e-06, "loss": 1.051, "step": 24276 }, { "epoch": 2.583209193445414, "grad_norm": 12.7421875, "learning_rate": 9.155166287663446e-06, "loss": 0.9738, "step": 24277 }, { "epoch": 2.5833155990636305, "grad_norm": 11.71875, "learning_rate": 9.155130756111428e-06, "loss": 0.7459, "step": 24278 }, { "epoch": 2.5834220046818475, "grad_norm": 15.078125, "learning_rate": 9.15509522455941e-06, "loss": 0.9173, "step": 24279 }, { "epoch": 2.583528410300064, "grad_norm": 14.6796875, "learning_rate": 9.155059693007391e-06, "loss": 0.8931, "step": 24280 }, { "epoch": 2.5836348159182805, "grad_norm": 12.15625, "learning_rate": 9.155024161455373e-06, "loss": 0.8412, "step": 24281 }, { "epoch": 2.583741221536497, "grad_norm": 17.125, "learning_rate": 9.154988629903354e-06, "loss": 0.9674, "step": 24282 }, { "epoch": 2.5838476271547135, "grad_norm": 13.8671875, "learning_rate": 9.154953098351336e-06, "loss": 0.8581, "step": 24283 }, { "epoch": 2.5839540327729305, "grad_norm": 12.34375, "learning_rate": 9.154917566799318e-06, "loss": 0.8362, "step": 24284 }, { "epoch": 2.584060438391147, "grad_norm": 12.0703125, "learning_rate": 9.1548820352473e-06, "loss": 0.9157, "step": 24285 }, { "epoch": 2.5841668440093635, "grad_norm": 11.21875, "learning_rate": 9.154846503695281e-06, "loss": 0.746, "step": 24286 }, { "epoch": 2.5842732496275804, "grad_norm": 11.3203125, "learning_rate": 9.154810972143264e-06, "loss": 0.8561, "step": 24287 }, { "epoch": 2.584379655245797, "grad_norm": 12.0703125, "learning_rate": 9.154775440591246e-06, "loss": 1.0241, "step": 24288 }, { "epoch": 2.5844860608640134, "grad_norm": 14.6875, "learning_rate": 9.154739909039227e-06, "loss": 0.9857, "step": 24289 }, { "epoch": 2.5845924664822304, "grad_norm": 14.8671875, "learning_rate": 9.154704377487209e-06, "loss": 0.9742, "step": 24290 }, { "epoch": 2.584698872100447, "grad_norm": 11.8203125, "learning_rate": 9.15466884593519e-06, "loss": 0.9401, "step": 24291 }, { "epoch": 2.5848052777186634, "grad_norm": 18.703125, "learning_rate": 9.154633314383174e-06, "loss": 0.957, "step": 24292 }, { "epoch": 2.5849116833368804, "grad_norm": 14.0859375, "learning_rate": 9.154597782831156e-06, "loss": 1.166, "step": 24293 }, { "epoch": 2.585018088955097, "grad_norm": 14.015625, "learning_rate": 9.154562251279137e-06, "loss": 0.9424, "step": 24294 }, { "epoch": 2.5851244945733134, "grad_norm": 12.1640625, "learning_rate": 9.154526719727119e-06, "loss": 0.9034, "step": 24295 }, { "epoch": 2.5852309001915303, "grad_norm": 15.328125, "learning_rate": 9.1544911881751e-06, "loss": 0.9512, "step": 24296 }, { "epoch": 2.585337305809747, "grad_norm": 15.5703125, "learning_rate": 9.154455656623082e-06, "loss": 0.9192, "step": 24297 }, { "epoch": 2.5854437114279634, "grad_norm": 13.390625, "learning_rate": 9.154420125071064e-06, "loss": 0.8041, "step": 24298 }, { "epoch": 2.58555011704618, "grad_norm": 17.71875, "learning_rate": 9.154384593519045e-06, "loss": 0.5222, "step": 24299 }, { "epoch": 2.585656522664397, "grad_norm": 14.515625, "learning_rate": 9.154349061967027e-06, "loss": 0.8019, "step": 24300 }, { "epoch": 2.5857629282826133, "grad_norm": 14.484375, "learning_rate": 9.154313530415008e-06, "loss": 0.8287, "step": 24301 }, { "epoch": 2.58586933390083, "grad_norm": 15.7578125, "learning_rate": 9.154277998862992e-06, "loss": 1.0412, "step": 24302 }, { "epoch": 2.5859757395190464, "grad_norm": 13.6484375, "learning_rate": 9.154242467310973e-06, "loss": 0.8244, "step": 24303 }, { "epoch": 2.5860821451372633, "grad_norm": 15.828125, "learning_rate": 9.154206935758955e-06, "loss": 0.8438, "step": 24304 }, { "epoch": 2.58618855075548, "grad_norm": 15.0390625, "learning_rate": 9.154171404206936e-06, "loss": 0.8812, "step": 24305 }, { "epoch": 2.5862949563736963, "grad_norm": 17.234375, "learning_rate": 9.154135872654918e-06, "loss": 0.9826, "step": 24306 }, { "epoch": 2.5864013619919133, "grad_norm": 24.1875, "learning_rate": 9.154100341102901e-06, "loss": 0.9854, "step": 24307 }, { "epoch": 2.58650776761013, "grad_norm": 16.4375, "learning_rate": 9.154064809550883e-06, "loss": 1.1004, "step": 24308 }, { "epoch": 2.5866141732283463, "grad_norm": 15.7265625, "learning_rate": 9.154029277998865e-06, "loss": 1.1263, "step": 24309 }, { "epoch": 2.5867205788465633, "grad_norm": 14.734375, "learning_rate": 9.153993746446845e-06, "loss": 1.0303, "step": 24310 }, { "epoch": 2.5868269844647798, "grad_norm": 15.9375, "learning_rate": 9.153958214894826e-06, "loss": 0.9591, "step": 24311 }, { "epoch": 2.5869333900829963, "grad_norm": 12.9765625, "learning_rate": 9.15392268334281e-06, "loss": 0.8361, "step": 24312 }, { "epoch": 2.5870397957012132, "grad_norm": 13.5703125, "learning_rate": 9.153887151790791e-06, "loss": 0.8753, "step": 24313 }, { "epoch": 2.5871462013194297, "grad_norm": 13.4921875, "learning_rate": 9.153851620238773e-06, "loss": 0.8524, "step": 24314 }, { "epoch": 2.5872526069376462, "grad_norm": 11.875, "learning_rate": 9.153816088686754e-06, "loss": 1.1278, "step": 24315 }, { "epoch": 2.587359012555863, "grad_norm": 11.1796875, "learning_rate": 9.153780557134736e-06, "loss": 0.8937, "step": 24316 }, { "epoch": 2.5874654181740797, "grad_norm": 19.25, "learning_rate": 9.15374502558272e-06, "loss": 0.9171, "step": 24317 }, { "epoch": 2.587571823792296, "grad_norm": 11.0, "learning_rate": 9.1537094940307e-06, "loss": 0.7601, "step": 24318 }, { "epoch": 2.5876782294105127, "grad_norm": 10.7734375, "learning_rate": 9.153673962478682e-06, "loss": 0.8816, "step": 24319 }, { "epoch": 2.5877846350287292, "grad_norm": 10.984375, "learning_rate": 9.153638430926664e-06, "loss": 0.8792, "step": 24320 }, { "epoch": 2.587891040646946, "grad_norm": 13.578125, "learning_rate": 9.153602899374646e-06, "loss": 0.8915, "step": 24321 }, { "epoch": 2.5879974462651627, "grad_norm": 12.796875, "learning_rate": 9.153567367822627e-06, "loss": 0.6859, "step": 24322 }, { "epoch": 2.588103851883379, "grad_norm": 9.765625, "learning_rate": 9.153531836270609e-06, "loss": 0.8736, "step": 24323 }, { "epoch": 2.588210257501596, "grad_norm": 16.140625, "learning_rate": 9.15349630471859e-06, "loss": 0.9643, "step": 24324 }, { "epoch": 2.5883166631198127, "grad_norm": 15.6875, "learning_rate": 9.153460773166572e-06, "loss": 1.0265, "step": 24325 }, { "epoch": 2.588423068738029, "grad_norm": 12.609375, "learning_rate": 9.153425241614554e-06, "loss": 0.8994, "step": 24326 }, { "epoch": 2.588529474356246, "grad_norm": 14.90625, "learning_rate": 9.153389710062535e-06, "loss": 0.8917, "step": 24327 }, { "epoch": 2.5886358799744627, "grad_norm": 18.5, "learning_rate": 9.153354178510519e-06, "loss": 0.9041, "step": 24328 }, { "epoch": 2.588742285592679, "grad_norm": 13.2890625, "learning_rate": 9.1533186469585e-06, "loss": 0.9039, "step": 24329 }, { "epoch": 2.588848691210896, "grad_norm": 15.0078125, "learning_rate": 9.153283115406482e-06, "loss": 1.0816, "step": 24330 }, { "epoch": 2.5889550968291126, "grad_norm": 15.5625, "learning_rate": 9.153247583854463e-06, "loss": 0.9604, "step": 24331 }, { "epoch": 2.589061502447329, "grad_norm": 10.5625, "learning_rate": 9.153212052302445e-06, "loss": 0.9022, "step": 24332 }, { "epoch": 2.589167908065546, "grad_norm": 11.7734375, "learning_rate": 9.153176520750428e-06, "loss": 0.7697, "step": 24333 }, { "epoch": 2.5892743136837626, "grad_norm": 12.671875, "learning_rate": 9.15314098919841e-06, "loss": 0.8644, "step": 24334 }, { "epoch": 2.589380719301979, "grad_norm": 13.1953125, "learning_rate": 9.153105457646391e-06, "loss": 0.9599, "step": 24335 }, { "epoch": 2.589487124920196, "grad_norm": 13.140625, "learning_rate": 9.153069926094373e-06, "loss": 0.8601, "step": 24336 }, { "epoch": 2.5895935305384126, "grad_norm": 15.8671875, "learning_rate": 9.153034394542355e-06, "loss": 0.7835, "step": 24337 }, { "epoch": 2.589699936156629, "grad_norm": 10.4453125, "learning_rate": 9.152998862990336e-06, "loss": 0.813, "step": 24338 }, { "epoch": 2.5898063417748456, "grad_norm": 15.2890625, "learning_rate": 9.152963331438318e-06, "loss": 0.9363, "step": 24339 }, { "epoch": 2.589912747393062, "grad_norm": 11.9453125, "learning_rate": 9.1529277998863e-06, "loss": 0.8013, "step": 24340 }, { "epoch": 2.590019153011279, "grad_norm": 12.6171875, "learning_rate": 9.152892268334281e-06, "loss": 0.9032, "step": 24341 }, { "epoch": 2.5901255586294956, "grad_norm": 15.96875, "learning_rate": 9.152856736782263e-06, "loss": 1.0361, "step": 24342 }, { "epoch": 2.590231964247712, "grad_norm": 12.7265625, "learning_rate": 9.152821205230246e-06, "loss": 0.916, "step": 24343 }, { "epoch": 2.590338369865929, "grad_norm": 10.515625, "learning_rate": 9.152785673678228e-06, "loss": 0.7551, "step": 24344 }, { "epoch": 2.5904447754841455, "grad_norm": 11.1875, "learning_rate": 9.15275014212621e-06, "loss": 0.8529, "step": 24345 }, { "epoch": 2.590551181102362, "grad_norm": 17.359375, "learning_rate": 9.15271461057419e-06, "loss": 1.0044, "step": 24346 }, { "epoch": 2.590657586720579, "grad_norm": 10.6953125, "learning_rate": 9.152679079022172e-06, "loss": 0.8452, "step": 24347 }, { "epoch": 2.5907639923387955, "grad_norm": 14.0078125, "learning_rate": 9.152643547470154e-06, "loss": 0.7757, "step": 24348 }, { "epoch": 2.590870397957012, "grad_norm": 13.3203125, "learning_rate": 9.152608015918136e-06, "loss": 0.9653, "step": 24349 }, { "epoch": 2.590976803575229, "grad_norm": 12.5625, "learning_rate": 9.152572484366117e-06, "loss": 1.0658, "step": 24350 }, { "epoch": 2.5910832091934455, "grad_norm": 13.3515625, "learning_rate": 9.152536952814099e-06, "loss": 1.0486, "step": 24351 }, { "epoch": 2.591189614811662, "grad_norm": 15.21875, "learning_rate": 9.15250142126208e-06, "loss": 0.9809, "step": 24352 }, { "epoch": 2.591296020429879, "grad_norm": 11.2421875, "learning_rate": 9.152465889710064e-06, "loss": 0.7549, "step": 24353 }, { "epoch": 2.5914024260480955, "grad_norm": 11.5234375, "learning_rate": 9.152430358158045e-06, "loss": 0.8692, "step": 24354 }, { "epoch": 2.591508831666312, "grad_norm": 10.03125, "learning_rate": 9.152394826606027e-06, "loss": 0.7942, "step": 24355 }, { "epoch": 2.5916152372845285, "grad_norm": 11.6015625, "learning_rate": 9.152359295054009e-06, "loss": 0.9465, "step": 24356 }, { "epoch": 2.5917216429027454, "grad_norm": 11.484375, "learning_rate": 9.15232376350199e-06, "loss": 0.8491, "step": 24357 }, { "epoch": 2.591828048520962, "grad_norm": 13.453125, "learning_rate": 9.152288231949973e-06, "loss": 1.0304, "step": 24358 }, { "epoch": 2.5919344541391784, "grad_norm": 12.03125, "learning_rate": 9.152252700397955e-06, "loss": 0.8847, "step": 24359 }, { "epoch": 2.592040859757395, "grad_norm": 11.203125, "learning_rate": 9.152217168845937e-06, "loss": 0.9406, "step": 24360 }, { "epoch": 2.592147265375612, "grad_norm": 13.9296875, "learning_rate": 9.152181637293918e-06, "loss": 0.8963, "step": 24361 }, { "epoch": 2.5922536709938284, "grad_norm": 11.421875, "learning_rate": 9.1521461057419e-06, "loss": 0.8781, "step": 24362 }, { "epoch": 2.592360076612045, "grad_norm": 13.5546875, "learning_rate": 9.152110574189882e-06, "loss": 0.9422, "step": 24363 }, { "epoch": 2.592466482230262, "grad_norm": 12.8828125, "learning_rate": 9.152075042637863e-06, "loss": 0.8682, "step": 24364 }, { "epoch": 2.5925728878484784, "grad_norm": 19.125, "learning_rate": 9.152039511085845e-06, "loss": 1.018, "step": 24365 }, { "epoch": 2.592679293466695, "grad_norm": 17.65625, "learning_rate": 9.152003979533826e-06, "loss": 0.9211, "step": 24366 }, { "epoch": 2.592785699084912, "grad_norm": 16.078125, "learning_rate": 9.151968447981808e-06, "loss": 0.8155, "step": 24367 }, { "epoch": 2.5928921047031284, "grad_norm": 12.203125, "learning_rate": 9.15193291642979e-06, "loss": 0.7765, "step": 24368 }, { "epoch": 2.592998510321345, "grad_norm": 14.1484375, "learning_rate": 9.151897384877773e-06, "loss": 1.0126, "step": 24369 }, { "epoch": 2.593104915939562, "grad_norm": 9.8125, "learning_rate": 9.151861853325754e-06, "loss": 0.799, "step": 24370 }, { "epoch": 2.5932113215577783, "grad_norm": 13.9140625, "learning_rate": 9.151826321773736e-06, "loss": 0.7871, "step": 24371 }, { "epoch": 2.593317727175995, "grad_norm": 14.7265625, "learning_rate": 9.151790790221718e-06, "loss": 0.9484, "step": 24372 }, { "epoch": 2.593424132794212, "grad_norm": 15.6640625, "learning_rate": 9.1517552586697e-06, "loss": 0.8137, "step": 24373 }, { "epoch": 2.5935305384124283, "grad_norm": 13.375, "learning_rate": 9.151719727117681e-06, "loss": 1.0276, "step": 24374 }, { "epoch": 2.593636944030645, "grad_norm": 12.5703125, "learning_rate": 9.151684195565663e-06, "loss": 0.9496, "step": 24375 }, { "epoch": 2.5937433496488613, "grad_norm": 15.2890625, "learning_rate": 9.151648664013644e-06, "loss": 1.0405, "step": 24376 }, { "epoch": 2.593849755267078, "grad_norm": 12.8984375, "learning_rate": 9.151613132461626e-06, "loss": 0.9152, "step": 24377 }, { "epoch": 2.593956160885295, "grad_norm": 13.8984375, "learning_rate": 9.151577600909607e-06, "loss": 1.0651, "step": 24378 }, { "epoch": 2.5940625665035113, "grad_norm": 17.90625, "learning_rate": 9.15154206935759e-06, "loss": 0.9524, "step": 24379 }, { "epoch": 2.594168972121728, "grad_norm": 16.671875, "learning_rate": 9.151506537805572e-06, "loss": 1.1679, "step": 24380 }, { "epoch": 2.5942753777399448, "grad_norm": 11.53125, "learning_rate": 9.151471006253554e-06, "loss": 0.7396, "step": 24381 }, { "epoch": 2.5943817833581613, "grad_norm": 12.546875, "learning_rate": 9.151435474701535e-06, "loss": 0.9393, "step": 24382 }, { "epoch": 2.594488188976378, "grad_norm": 16.234375, "learning_rate": 9.151399943149517e-06, "loss": 1.2301, "step": 24383 }, { "epoch": 2.5945945945945947, "grad_norm": 10.46875, "learning_rate": 9.1513644115975e-06, "loss": 0.9355, "step": 24384 }, { "epoch": 2.5947010002128112, "grad_norm": 10.90625, "learning_rate": 9.151328880045482e-06, "loss": 0.8268, "step": 24385 }, { "epoch": 2.5948074058310278, "grad_norm": 16.515625, "learning_rate": 9.151293348493464e-06, "loss": 1.1268, "step": 24386 }, { "epoch": 2.5949138114492447, "grad_norm": 10.1015625, "learning_rate": 9.151257816941445e-06, "loss": 0.7952, "step": 24387 }, { "epoch": 2.595020217067461, "grad_norm": 13.921875, "learning_rate": 9.151222285389427e-06, "loss": 1.0381, "step": 24388 }, { "epoch": 2.5951266226856777, "grad_norm": 13.34375, "learning_rate": 9.151186753837408e-06, "loss": 0.9785, "step": 24389 }, { "epoch": 2.5952330283038947, "grad_norm": 17.65625, "learning_rate": 9.15115122228539e-06, "loss": 0.8172, "step": 24390 }, { "epoch": 2.595339433922111, "grad_norm": 12.8359375, "learning_rate": 9.151115690733372e-06, "loss": 0.9325, "step": 24391 }, { "epoch": 2.5954458395403277, "grad_norm": 13.0546875, "learning_rate": 9.151080159181353e-06, "loss": 0.8991, "step": 24392 }, { "epoch": 2.595552245158544, "grad_norm": 12.7109375, "learning_rate": 9.151044627629335e-06, "loss": 0.8429, "step": 24393 }, { "epoch": 2.595658650776761, "grad_norm": 11.125, "learning_rate": 9.151009096077318e-06, "loss": 0.8985, "step": 24394 }, { "epoch": 2.5957650563949777, "grad_norm": 12.0234375, "learning_rate": 9.1509735645253e-06, "loss": 0.8636, "step": 24395 }, { "epoch": 2.595871462013194, "grad_norm": 11.6171875, "learning_rate": 9.150938032973281e-06, "loss": 0.9545, "step": 24396 }, { "epoch": 2.5959778676314107, "grad_norm": 12.890625, "learning_rate": 9.150902501421263e-06, "loss": 0.9691, "step": 24397 }, { "epoch": 2.5960842732496276, "grad_norm": 18.25, "learning_rate": 9.150866969869245e-06, "loss": 0.9366, "step": 24398 }, { "epoch": 2.596190678867844, "grad_norm": 10.3515625, "learning_rate": 9.150831438317226e-06, "loss": 0.7656, "step": 24399 }, { "epoch": 2.5962970844860607, "grad_norm": 12.7265625, "learning_rate": 9.150795906765208e-06, "loss": 1.0168, "step": 24400 }, { "epoch": 2.5964034901042776, "grad_norm": 12.8125, "learning_rate": 9.15076037521319e-06, "loss": 0.8785, "step": 24401 }, { "epoch": 2.596509895722494, "grad_norm": 11.6328125, "learning_rate": 9.150724843661171e-06, "loss": 0.8633, "step": 24402 }, { "epoch": 2.5966163013407106, "grad_norm": 14.6640625, "learning_rate": 9.150689312109153e-06, "loss": 0.7848, "step": 24403 }, { "epoch": 2.5967227069589276, "grad_norm": 13.9609375, "learning_rate": 9.150653780557134e-06, "loss": 0.8907, "step": 24404 }, { "epoch": 2.596829112577144, "grad_norm": 9.7109375, "learning_rate": 9.150618249005117e-06, "loss": 0.6776, "step": 24405 }, { "epoch": 2.5969355181953606, "grad_norm": 14.1640625, "learning_rate": 9.150582717453099e-06, "loss": 0.8685, "step": 24406 }, { "epoch": 2.5970419238135776, "grad_norm": 95.6875, "learning_rate": 9.15054718590108e-06, "loss": 1.1707, "step": 24407 }, { "epoch": 2.597148329431794, "grad_norm": 16.375, "learning_rate": 9.150511654349062e-06, "loss": 0.9398, "step": 24408 }, { "epoch": 2.5972547350500106, "grad_norm": 8.34375, "learning_rate": 9.150476122797044e-06, "loss": 0.7618, "step": 24409 }, { "epoch": 2.5973611406682275, "grad_norm": 14.8828125, "learning_rate": 9.150440591245027e-06, "loss": 1.0671, "step": 24410 }, { "epoch": 2.597467546286444, "grad_norm": 11.4921875, "learning_rate": 9.150405059693009e-06, "loss": 0.9039, "step": 24411 }, { "epoch": 2.5975739519046606, "grad_norm": 10.7265625, "learning_rate": 9.15036952814099e-06, "loss": 0.7607, "step": 24412 }, { "epoch": 2.597680357522877, "grad_norm": 12.5625, "learning_rate": 9.150333996588972e-06, "loss": 0.9825, "step": 24413 }, { "epoch": 2.5977867631410936, "grad_norm": 11.390625, "learning_rate": 9.150298465036954e-06, "loss": 0.8433, "step": 24414 }, { "epoch": 2.5978931687593105, "grad_norm": 10.375, "learning_rate": 9.150262933484935e-06, "loss": 0.8056, "step": 24415 }, { "epoch": 2.597999574377527, "grad_norm": 13.703125, "learning_rate": 9.150227401932917e-06, "loss": 0.8238, "step": 24416 }, { "epoch": 2.5981059799957436, "grad_norm": 11.8046875, "learning_rate": 9.150191870380898e-06, "loss": 0.7467, "step": 24417 }, { "epoch": 2.5982123856139605, "grad_norm": 14.296875, "learning_rate": 9.15015633882888e-06, "loss": 0.8705, "step": 24418 }, { "epoch": 2.598318791232177, "grad_norm": 16.453125, "learning_rate": 9.150120807276862e-06, "loss": 1.0316, "step": 24419 }, { "epoch": 2.5984251968503935, "grad_norm": 12.7109375, "learning_rate": 9.150085275724845e-06, "loss": 0.8776, "step": 24420 }, { "epoch": 2.5985316024686105, "grad_norm": 13.0703125, "learning_rate": 9.150049744172827e-06, "loss": 0.9059, "step": 24421 }, { "epoch": 2.598638008086827, "grad_norm": 14.4140625, "learning_rate": 9.150014212620808e-06, "loss": 1.1615, "step": 24422 }, { "epoch": 2.5987444137050435, "grad_norm": 15.1796875, "learning_rate": 9.14997868106879e-06, "loss": 0.9747, "step": 24423 }, { "epoch": 2.5988508193232605, "grad_norm": 20.28125, "learning_rate": 9.149943149516771e-06, "loss": 1.2602, "step": 24424 }, { "epoch": 2.598957224941477, "grad_norm": 11.3828125, "learning_rate": 9.149907617964755e-06, "loss": 0.7604, "step": 24425 }, { "epoch": 2.5990636305596935, "grad_norm": 14.3828125, "learning_rate": 9.149872086412736e-06, "loss": 0.9416, "step": 24426 }, { "epoch": 2.5991700361779104, "grad_norm": 14.1171875, "learning_rate": 9.149836554860718e-06, "loss": 1.0726, "step": 24427 }, { "epoch": 2.599276441796127, "grad_norm": 11.015625, "learning_rate": 9.1498010233087e-06, "loss": 0.9149, "step": 24428 }, { "epoch": 2.5993828474143434, "grad_norm": 11.4609375, "learning_rate": 9.14976549175668e-06, "loss": 0.8975, "step": 24429 }, { "epoch": 2.59948925303256, "grad_norm": 15.2421875, "learning_rate": 9.149729960204663e-06, "loss": 0.7857, "step": 24430 }, { "epoch": 2.599595658650777, "grad_norm": 11.15625, "learning_rate": 9.149694428652644e-06, "loss": 0.8826, "step": 24431 }, { "epoch": 2.5997020642689934, "grad_norm": 15.296875, "learning_rate": 9.149658897100626e-06, "loss": 0.8021, "step": 24432 }, { "epoch": 2.59980846988721, "grad_norm": 14.625, "learning_rate": 9.149623365548608e-06, "loss": 0.9106, "step": 24433 }, { "epoch": 2.5999148755054264, "grad_norm": 19.078125, "learning_rate": 9.14958783399659e-06, "loss": 1.1888, "step": 24434 }, { "epoch": 2.6000212811236434, "grad_norm": 15.953125, "learning_rate": 9.149552302444572e-06, "loss": 1.0515, "step": 24435 }, { "epoch": 2.60012768674186, "grad_norm": 12.9453125, "learning_rate": 9.149516770892554e-06, "loss": 1.0203, "step": 24436 }, { "epoch": 2.6002340923600764, "grad_norm": 13.1015625, "learning_rate": 9.149481239340536e-06, "loss": 1.0364, "step": 24437 }, { "epoch": 2.6003404979782934, "grad_norm": 12.5703125, "learning_rate": 9.149445707788517e-06, "loss": 0.9135, "step": 24438 }, { "epoch": 2.60044690359651, "grad_norm": 12.7578125, "learning_rate": 9.149410176236499e-06, "loss": 0.8965, "step": 24439 }, { "epoch": 2.6005533092147264, "grad_norm": 13.0625, "learning_rate": 9.14937464468448e-06, "loss": 1.0118, "step": 24440 }, { "epoch": 2.6006597148329433, "grad_norm": 13.0390625, "learning_rate": 9.149339113132462e-06, "loss": 0.8718, "step": 24441 }, { "epoch": 2.60076612045116, "grad_norm": 10.5078125, "learning_rate": 9.149303581580444e-06, "loss": 0.9474, "step": 24442 }, { "epoch": 2.6008725260693764, "grad_norm": 12.5390625, "learning_rate": 9.149268050028425e-06, "loss": 0.9809, "step": 24443 }, { "epoch": 2.6009789316875933, "grad_norm": 10.328125, "learning_rate": 9.149232518476407e-06, "loss": 0.8753, "step": 24444 }, { "epoch": 2.60108533730581, "grad_norm": 10.4296875, "learning_rate": 9.149196986924389e-06, "loss": 0.8427, "step": 24445 }, { "epoch": 2.6011917429240263, "grad_norm": 12.796875, "learning_rate": 9.149161455372372e-06, "loss": 0.9315, "step": 24446 }, { "epoch": 2.6012981485422433, "grad_norm": 16.796875, "learning_rate": 9.149125923820353e-06, "loss": 0.9246, "step": 24447 }, { "epoch": 2.60140455416046, "grad_norm": 16.421875, "learning_rate": 9.149090392268335e-06, "loss": 0.7899, "step": 24448 }, { "epoch": 2.6015109597786763, "grad_norm": 13.5078125, "learning_rate": 9.149054860716317e-06, "loss": 0.9791, "step": 24449 }, { "epoch": 2.601617365396893, "grad_norm": 10.9609375, "learning_rate": 9.149019329164298e-06, "loss": 0.966, "step": 24450 }, { "epoch": 2.6017237710151093, "grad_norm": 14.7109375, "learning_rate": 9.148983797612282e-06, "loss": 0.9096, "step": 24451 }, { "epoch": 2.6018301766333263, "grad_norm": 14.34375, "learning_rate": 9.148948266060263e-06, "loss": 0.9327, "step": 24452 }, { "epoch": 2.601936582251543, "grad_norm": 12.1875, "learning_rate": 9.148912734508245e-06, "loss": 1.08, "step": 24453 }, { "epoch": 2.6020429878697593, "grad_norm": 14.3515625, "learning_rate": 9.148877202956226e-06, "loss": 0.8789, "step": 24454 }, { "epoch": 2.6021493934879762, "grad_norm": 14.9296875, "learning_rate": 9.148841671404208e-06, "loss": 0.93, "step": 24455 }, { "epoch": 2.6022557991061928, "grad_norm": 12.6484375, "learning_rate": 9.14880613985219e-06, "loss": 0.8482, "step": 24456 }, { "epoch": 2.6023622047244093, "grad_norm": 14.1796875, "learning_rate": 9.148770608300171e-06, "loss": 0.7618, "step": 24457 }, { "epoch": 2.602468610342626, "grad_norm": 11.3046875, "learning_rate": 9.148735076748153e-06, "loss": 0.8841, "step": 24458 }, { "epoch": 2.6025750159608427, "grad_norm": 21.671875, "learning_rate": 9.148699545196134e-06, "loss": 0.8232, "step": 24459 }, { "epoch": 2.6026814215790592, "grad_norm": 12.2890625, "learning_rate": 9.148664013644116e-06, "loss": 0.8124, "step": 24460 }, { "epoch": 2.602787827197276, "grad_norm": 10.9375, "learning_rate": 9.1486284820921e-06, "loss": 0.8562, "step": 24461 }, { "epoch": 2.6028942328154927, "grad_norm": 12.15625, "learning_rate": 9.148592950540081e-06, "loss": 0.9119, "step": 24462 }, { "epoch": 2.603000638433709, "grad_norm": 14.2265625, "learning_rate": 9.148557418988063e-06, "loss": 0.9291, "step": 24463 }, { "epoch": 2.603107044051926, "grad_norm": 12.6875, "learning_rate": 9.148521887436044e-06, "loss": 0.8806, "step": 24464 }, { "epoch": 2.6032134496701427, "grad_norm": 10.90625, "learning_rate": 9.148486355884026e-06, "loss": 0.8231, "step": 24465 }, { "epoch": 2.603319855288359, "grad_norm": 12.1015625, "learning_rate": 9.148450824332007e-06, "loss": 0.9214, "step": 24466 }, { "epoch": 2.603426260906576, "grad_norm": 13.609375, "learning_rate": 9.148415292779989e-06, "loss": 0.9187, "step": 24467 }, { "epoch": 2.6035326665247926, "grad_norm": 11.2265625, "learning_rate": 9.14837976122797e-06, "loss": 0.9382, "step": 24468 }, { "epoch": 2.603639072143009, "grad_norm": 13.359375, "learning_rate": 9.148344229675952e-06, "loss": 0.869, "step": 24469 }, { "epoch": 2.6037454777612257, "grad_norm": 16.40625, "learning_rate": 9.148308698123934e-06, "loss": 0.9711, "step": 24470 }, { "epoch": 2.603851883379442, "grad_norm": 15.328125, "learning_rate": 9.148273166571917e-06, "loss": 0.9623, "step": 24471 }, { "epoch": 2.603958288997659, "grad_norm": 12.4921875, "learning_rate": 9.148237635019899e-06, "loss": 0.9478, "step": 24472 }, { "epoch": 2.6040646946158756, "grad_norm": 16.671875, "learning_rate": 9.14820210346788e-06, "loss": 0.8473, "step": 24473 }, { "epoch": 2.604171100234092, "grad_norm": 15.9375, "learning_rate": 9.148166571915862e-06, "loss": 1.0411, "step": 24474 }, { "epoch": 2.604277505852309, "grad_norm": 19.40625, "learning_rate": 9.148131040363843e-06, "loss": 1.0322, "step": 24475 }, { "epoch": 2.6043839114705256, "grad_norm": 11.5625, "learning_rate": 9.148095508811827e-06, "loss": 0.8414, "step": 24476 }, { "epoch": 2.604490317088742, "grad_norm": 14.0546875, "learning_rate": 9.148059977259808e-06, "loss": 0.7994, "step": 24477 }, { "epoch": 2.604596722706959, "grad_norm": 14.578125, "learning_rate": 9.14802444570779e-06, "loss": 0.9298, "step": 24478 }, { "epoch": 2.6047031283251756, "grad_norm": 13.390625, "learning_rate": 9.147988914155772e-06, "loss": 0.9199, "step": 24479 }, { "epoch": 2.604809533943392, "grad_norm": 12.859375, "learning_rate": 9.147953382603753e-06, "loss": 1.0329, "step": 24480 }, { "epoch": 2.604915939561609, "grad_norm": 12.9296875, "learning_rate": 9.147917851051735e-06, "loss": 0.9675, "step": 24481 }, { "epoch": 2.6050223451798256, "grad_norm": 10.671875, "learning_rate": 9.147882319499716e-06, "loss": 0.7984, "step": 24482 }, { "epoch": 2.605128750798042, "grad_norm": 11.0625, "learning_rate": 9.147846787947698e-06, "loss": 0.8901, "step": 24483 }, { "epoch": 2.605235156416259, "grad_norm": 12.7421875, "learning_rate": 9.14781125639568e-06, "loss": 0.9243, "step": 24484 }, { "epoch": 2.6053415620344755, "grad_norm": 12.1328125, "learning_rate": 9.147775724843661e-06, "loss": 0.8161, "step": 24485 }, { "epoch": 2.605447967652692, "grad_norm": 13.515625, "learning_rate": 9.147740193291643e-06, "loss": 0.8416, "step": 24486 }, { "epoch": 2.6055543732709086, "grad_norm": 15.546875, "learning_rate": 9.147704661739626e-06, "loss": 0.8796, "step": 24487 }, { "epoch": 2.6056607788891255, "grad_norm": 13.75, "learning_rate": 9.147669130187608e-06, "loss": 1.2464, "step": 24488 }, { "epoch": 2.605767184507342, "grad_norm": 15.546875, "learning_rate": 9.14763359863559e-06, "loss": 0.8454, "step": 24489 }, { "epoch": 2.6058735901255585, "grad_norm": 13.4453125, "learning_rate": 9.147598067083571e-06, "loss": 0.8505, "step": 24490 }, { "epoch": 2.605979995743775, "grad_norm": 16.03125, "learning_rate": 9.147562535531553e-06, "loss": 1.1375, "step": 24491 }, { "epoch": 2.606086401361992, "grad_norm": 12.9140625, "learning_rate": 9.147527003979534e-06, "loss": 0.9772, "step": 24492 }, { "epoch": 2.6061928069802085, "grad_norm": 14.3359375, "learning_rate": 9.147491472427516e-06, "loss": 0.9524, "step": 24493 }, { "epoch": 2.606299212598425, "grad_norm": 12.0546875, "learning_rate": 9.147455940875497e-06, "loss": 0.9468, "step": 24494 }, { "epoch": 2.606405618216642, "grad_norm": 13.3515625, "learning_rate": 9.147420409323479e-06, "loss": 0.8314, "step": 24495 }, { "epoch": 2.6065120238348585, "grad_norm": 11.9921875, "learning_rate": 9.14738487777146e-06, "loss": 0.9148, "step": 24496 }, { "epoch": 2.606618429453075, "grad_norm": 14.984375, "learning_rate": 9.147349346219444e-06, "loss": 1.1297, "step": 24497 }, { "epoch": 2.606724835071292, "grad_norm": 14.8359375, "learning_rate": 9.147313814667426e-06, "loss": 0.9785, "step": 24498 }, { "epoch": 2.6068312406895084, "grad_norm": 14.4296875, "learning_rate": 9.147278283115407e-06, "loss": 0.9225, "step": 24499 }, { "epoch": 2.606937646307725, "grad_norm": 14.515625, "learning_rate": 9.147242751563389e-06, "loss": 0.9318, "step": 24500 }, { "epoch": 2.607044051925942, "grad_norm": 11.171875, "learning_rate": 9.14720722001137e-06, "loss": 0.8031, "step": 24501 }, { "epoch": 2.6071504575441584, "grad_norm": 12.5546875, "learning_rate": 9.147171688459354e-06, "loss": 0.9505, "step": 24502 }, { "epoch": 2.607256863162375, "grad_norm": 16.46875, "learning_rate": 9.147136156907335e-06, "loss": 0.9694, "step": 24503 }, { "epoch": 2.607363268780592, "grad_norm": 12.6484375, "learning_rate": 9.147100625355317e-06, "loss": 0.8313, "step": 24504 }, { "epoch": 2.6074696743988084, "grad_norm": 13.2421875, "learning_rate": 9.147065093803298e-06, "loss": 0.9184, "step": 24505 }, { "epoch": 2.607576080017025, "grad_norm": 11.1953125, "learning_rate": 9.14702956225128e-06, "loss": 0.988, "step": 24506 }, { "epoch": 2.6076824856352414, "grad_norm": 16.90625, "learning_rate": 9.146994030699262e-06, "loss": 1.2114, "step": 24507 }, { "epoch": 2.607788891253458, "grad_norm": 14.3671875, "learning_rate": 9.146958499147243e-06, "loss": 1.0785, "step": 24508 }, { "epoch": 2.607895296871675, "grad_norm": 11.6328125, "learning_rate": 9.146922967595225e-06, "loss": 0.9245, "step": 24509 }, { "epoch": 2.6080017024898914, "grad_norm": 12.34375, "learning_rate": 9.146887436043207e-06, "loss": 1.0407, "step": 24510 }, { "epoch": 2.608108108108108, "grad_norm": 10.6171875, "learning_rate": 9.146851904491188e-06, "loss": 0.8805, "step": 24511 }, { "epoch": 2.608214513726325, "grad_norm": 12.6328125, "learning_rate": 9.146816372939171e-06, "loss": 0.8031, "step": 24512 }, { "epoch": 2.6083209193445414, "grad_norm": 11.796875, "learning_rate": 9.146780841387153e-06, "loss": 0.8882, "step": 24513 }, { "epoch": 2.608427324962758, "grad_norm": 12.5546875, "learning_rate": 9.146745309835135e-06, "loss": 0.8311, "step": 24514 }, { "epoch": 2.608533730580975, "grad_norm": 11.9609375, "learning_rate": 9.146709778283116e-06, "loss": 0.7806, "step": 24515 }, { "epoch": 2.6086401361991913, "grad_norm": 13.859375, "learning_rate": 9.146674246731098e-06, "loss": 0.7481, "step": 24516 }, { "epoch": 2.608746541817408, "grad_norm": 13.0, "learning_rate": 9.146638715179081e-06, "loss": 0.83, "step": 24517 }, { "epoch": 2.608852947435625, "grad_norm": 13.5, "learning_rate": 9.146603183627061e-06, "loss": 0.8116, "step": 24518 }, { "epoch": 2.6089593530538413, "grad_norm": 17.390625, "learning_rate": 9.146567652075043e-06, "loss": 0.9517, "step": 24519 }, { "epoch": 2.609065758672058, "grad_norm": 16.40625, "learning_rate": 9.146532120523024e-06, "loss": 0.9988, "step": 24520 }, { "epoch": 2.6091721642902748, "grad_norm": 14.15625, "learning_rate": 9.146496588971006e-06, "loss": 0.9329, "step": 24521 }, { "epoch": 2.6092785699084913, "grad_norm": 12.7890625, "learning_rate": 9.146461057418987e-06, "loss": 0.9298, "step": 24522 }, { "epoch": 2.609384975526708, "grad_norm": 15.71875, "learning_rate": 9.14642552586697e-06, "loss": 1.0406, "step": 24523 }, { "epoch": 2.6094913811449243, "grad_norm": 10.6015625, "learning_rate": 9.146389994314952e-06, "loss": 0.9371, "step": 24524 }, { "epoch": 2.6095977867631412, "grad_norm": 15.2890625, "learning_rate": 9.146354462762934e-06, "loss": 1.0031, "step": 24525 }, { "epoch": 2.6097041923813578, "grad_norm": 10.3828125, "learning_rate": 9.146318931210916e-06, "loss": 0.8893, "step": 24526 }, { "epoch": 2.6098105979995743, "grad_norm": 18.578125, "learning_rate": 9.146283399658897e-06, "loss": 1.0305, "step": 24527 }, { "epoch": 2.6099170036177908, "grad_norm": 13.4921875, "learning_rate": 9.14624786810688e-06, "loss": 0.8091, "step": 24528 }, { "epoch": 2.6100234092360077, "grad_norm": 14.3671875, "learning_rate": 9.146212336554862e-06, "loss": 0.7483, "step": 24529 }, { "epoch": 2.6101298148542242, "grad_norm": 10.9296875, "learning_rate": 9.146176805002844e-06, "loss": 0.8412, "step": 24530 }, { "epoch": 2.6102362204724407, "grad_norm": 15.8828125, "learning_rate": 9.146141273450825e-06, "loss": 1.1769, "step": 24531 }, { "epoch": 2.6103426260906577, "grad_norm": 13.921875, "learning_rate": 9.146105741898807e-06, "loss": 0.7973, "step": 24532 }, { "epoch": 2.610449031708874, "grad_norm": 11.6640625, "learning_rate": 9.146070210346789e-06, "loss": 0.8728, "step": 24533 }, { "epoch": 2.6105554373270907, "grad_norm": 11.046875, "learning_rate": 9.14603467879477e-06, "loss": 0.8765, "step": 24534 }, { "epoch": 2.6106618429453077, "grad_norm": 13.015625, "learning_rate": 9.145999147242752e-06, "loss": 0.8785, "step": 24535 }, { "epoch": 2.610768248563524, "grad_norm": 11.9140625, "learning_rate": 9.145963615690733e-06, "loss": 0.774, "step": 24536 }, { "epoch": 2.6108746541817407, "grad_norm": 17.734375, "learning_rate": 9.145928084138715e-06, "loss": 1.1053, "step": 24537 }, { "epoch": 2.6109810597999576, "grad_norm": 13.6484375, "learning_rate": 9.145892552586698e-06, "loss": 0.7622, "step": 24538 }, { "epoch": 2.611087465418174, "grad_norm": 16.328125, "learning_rate": 9.14585702103468e-06, "loss": 0.8814, "step": 24539 }, { "epoch": 2.6111938710363907, "grad_norm": 10.7265625, "learning_rate": 9.145821489482661e-06, "loss": 0.956, "step": 24540 }, { "epoch": 2.6113002766546076, "grad_norm": 12.4140625, "learning_rate": 9.145785957930643e-06, "loss": 0.9925, "step": 24541 }, { "epoch": 2.611406682272824, "grad_norm": 14.8828125, "learning_rate": 9.145750426378625e-06, "loss": 1.0667, "step": 24542 }, { "epoch": 2.6115130878910406, "grad_norm": 12.9140625, "learning_rate": 9.145714894826608e-06, "loss": 0.7906, "step": 24543 }, { "epoch": 2.611619493509257, "grad_norm": 15.03125, "learning_rate": 9.14567936327459e-06, "loss": 0.9942, "step": 24544 }, { "epoch": 2.6117258991274737, "grad_norm": 10.6640625, "learning_rate": 9.145643831722571e-06, "loss": 0.7746, "step": 24545 }, { "epoch": 2.6118323047456906, "grad_norm": 11.4765625, "learning_rate": 9.145608300170553e-06, "loss": 0.7562, "step": 24546 }, { "epoch": 2.611938710363907, "grad_norm": 14.1875, "learning_rate": 9.145572768618534e-06, "loss": 0.7665, "step": 24547 }, { "epoch": 2.6120451159821236, "grad_norm": 14.8671875, "learning_rate": 9.145537237066516e-06, "loss": 1.0054, "step": 24548 }, { "epoch": 2.6121515216003406, "grad_norm": 9.640625, "learning_rate": 9.145501705514498e-06, "loss": 0.7882, "step": 24549 }, { "epoch": 2.612257927218557, "grad_norm": 13.796875, "learning_rate": 9.14546617396248e-06, "loss": 0.9339, "step": 24550 }, { "epoch": 2.6123643328367736, "grad_norm": 14.9921875, "learning_rate": 9.145430642410461e-06, "loss": 0.9453, "step": 24551 }, { "epoch": 2.6124707384549906, "grad_norm": 11.8046875, "learning_rate": 9.145395110858442e-06, "loss": 0.7387, "step": 24552 }, { "epoch": 2.612577144073207, "grad_norm": 13.8125, "learning_rate": 9.145359579306426e-06, "loss": 0.7827, "step": 24553 }, { "epoch": 2.6126835496914236, "grad_norm": 20.359375, "learning_rate": 9.145324047754407e-06, "loss": 0.9322, "step": 24554 }, { "epoch": 2.6127899553096405, "grad_norm": 12.9375, "learning_rate": 9.145288516202389e-06, "loss": 0.9411, "step": 24555 }, { "epoch": 2.612896360927857, "grad_norm": 10.5078125, "learning_rate": 9.14525298465037e-06, "loss": 0.8919, "step": 24556 }, { "epoch": 2.6130027665460736, "grad_norm": 13.5859375, "learning_rate": 9.145217453098352e-06, "loss": 0.9268, "step": 24557 }, { "epoch": 2.6131091721642905, "grad_norm": 13.109375, "learning_rate": 9.145181921546334e-06, "loss": 0.8691, "step": 24558 }, { "epoch": 2.613215577782507, "grad_norm": 16.796875, "learning_rate": 9.145146389994315e-06, "loss": 0.8392, "step": 24559 }, { "epoch": 2.6133219834007235, "grad_norm": 12.7578125, "learning_rate": 9.145110858442297e-06, "loss": 0.8579, "step": 24560 }, { "epoch": 2.61342838901894, "grad_norm": 13.296875, "learning_rate": 9.145075326890279e-06, "loss": 0.9572, "step": 24561 }, { "epoch": 2.613534794637157, "grad_norm": 11.03125, "learning_rate": 9.14503979533826e-06, "loss": 0.8276, "step": 24562 }, { "epoch": 2.6136412002553735, "grad_norm": 12.2421875, "learning_rate": 9.145004263786242e-06, "loss": 0.7715, "step": 24563 }, { "epoch": 2.61374760587359, "grad_norm": 8.75, "learning_rate": 9.144968732234225e-06, "loss": 0.8036, "step": 24564 }, { "epoch": 2.6138540114918065, "grad_norm": 16.046875, "learning_rate": 9.144933200682207e-06, "loss": 0.9544, "step": 24565 }, { "epoch": 2.6139604171100235, "grad_norm": 12.4140625, "learning_rate": 9.144897669130188e-06, "loss": 0.8628, "step": 24566 }, { "epoch": 2.61406682272824, "grad_norm": 12.59375, "learning_rate": 9.14486213757817e-06, "loss": 0.6917, "step": 24567 }, { "epoch": 2.6141732283464565, "grad_norm": 15.2734375, "learning_rate": 9.144826606026152e-06, "loss": 0.8589, "step": 24568 }, { "epoch": 2.6142796339646734, "grad_norm": 17.28125, "learning_rate": 9.144791074474135e-06, "loss": 0.8532, "step": 24569 }, { "epoch": 2.61438603958289, "grad_norm": 13.53125, "learning_rate": 9.144755542922116e-06, "loss": 0.9802, "step": 24570 }, { "epoch": 2.6144924452011065, "grad_norm": 14.4609375, "learning_rate": 9.144720011370098e-06, "loss": 0.8975, "step": 24571 }, { "epoch": 2.6145988508193234, "grad_norm": 10.59375, "learning_rate": 9.14468447981808e-06, "loss": 0.8165, "step": 24572 }, { "epoch": 2.61470525643754, "grad_norm": 11.8671875, "learning_rate": 9.144648948266061e-06, "loss": 0.9373, "step": 24573 }, { "epoch": 2.6148116620557564, "grad_norm": 10.6171875, "learning_rate": 9.144613416714043e-06, "loss": 0.8211, "step": 24574 }, { "epoch": 2.6149180676739734, "grad_norm": 10.8828125, "learning_rate": 9.144577885162024e-06, "loss": 0.884, "step": 24575 }, { "epoch": 2.61502447329219, "grad_norm": 12.8984375, "learning_rate": 9.144542353610006e-06, "loss": 1.095, "step": 24576 }, { "epoch": 2.6151308789104064, "grad_norm": 11.515625, "learning_rate": 9.144506822057988e-06, "loss": 0.9215, "step": 24577 }, { "epoch": 2.6152372845286234, "grad_norm": 10.890625, "learning_rate": 9.14447129050597e-06, "loss": 0.9598, "step": 24578 }, { "epoch": 2.61534369014684, "grad_norm": 13.46875, "learning_rate": 9.144435758953953e-06, "loss": 0.7537, "step": 24579 }, { "epoch": 2.6154500957650564, "grad_norm": 14.1328125, "learning_rate": 9.144400227401934e-06, "loss": 0.8577, "step": 24580 }, { "epoch": 2.615556501383273, "grad_norm": 11.6796875, "learning_rate": 9.144364695849916e-06, "loss": 0.9148, "step": 24581 }, { "epoch": 2.6156629070014894, "grad_norm": 13.7890625, "learning_rate": 9.144329164297897e-06, "loss": 0.8992, "step": 24582 }, { "epoch": 2.6157693126197064, "grad_norm": 12.9453125, "learning_rate": 9.144293632745879e-06, "loss": 1.0695, "step": 24583 }, { "epoch": 2.615875718237923, "grad_norm": 17.15625, "learning_rate": 9.14425810119386e-06, "loss": 0.9611, "step": 24584 }, { "epoch": 2.6159821238561394, "grad_norm": 12.2421875, "learning_rate": 9.144222569641842e-06, "loss": 0.8948, "step": 24585 }, { "epoch": 2.6160885294743563, "grad_norm": 10.90625, "learning_rate": 9.144187038089824e-06, "loss": 0.8808, "step": 24586 }, { "epoch": 2.616194935092573, "grad_norm": 13.53125, "learning_rate": 9.144151506537805e-06, "loss": 0.8756, "step": 24587 }, { "epoch": 2.6163013407107893, "grad_norm": 14.2421875, "learning_rate": 9.144115974985787e-06, "loss": 0.831, "step": 24588 }, { "epoch": 2.6164077463290063, "grad_norm": 21.140625, "learning_rate": 9.14408044343377e-06, "loss": 0.9367, "step": 24589 }, { "epoch": 2.616514151947223, "grad_norm": 12.1640625, "learning_rate": 9.144044911881752e-06, "loss": 0.8937, "step": 24590 }, { "epoch": 2.6166205575654393, "grad_norm": 12.0703125, "learning_rate": 9.144009380329734e-06, "loss": 0.8372, "step": 24591 }, { "epoch": 2.6167269631836563, "grad_norm": 13.125, "learning_rate": 9.143973848777715e-06, "loss": 0.8649, "step": 24592 }, { "epoch": 2.616833368801873, "grad_norm": 13.1015625, "learning_rate": 9.143938317225697e-06, "loss": 0.9487, "step": 24593 }, { "epoch": 2.6169397744200893, "grad_norm": 11.7734375, "learning_rate": 9.14390278567368e-06, "loss": 0.7988, "step": 24594 }, { "epoch": 2.6170461800383062, "grad_norm": 7.99609375, "learning_rate": 9.143867254121662e-06, "loss": 0.7351, "step": 24595 }, { "epoch": 2.6171525856565228, "grad_norm": 12.2265625, "learning_rate": 9.143831722569643e-06, "loss": 0.9174, "step": 24596 }, { "epoch": 2.6172589912747393, "grad_norm": 12.2734375, "learning_rate": 9.143796191017625e-06, "loss": 0.8006, "step": 24597 }, { "epoch": 2.617365396892956, "grad_norm": 12.3984375, "learning_rate": 9.143760659465607e-06, "loss": 1.0113, "step": 24598 }, { "epoch": 2.6174718025111727, "grad_norm": 12.2578125, "learning_rate": 9.143725127913588e-06, "loss": 0.8663, "step": 24599 }, { "epoch": 2.6175782081293892, "grad_norm": 14.8984375, "learning_rate": 9.14368959636157e-06, "loss": 0.9226, "step": 24600 }, { "epoch": 2.6176846137476057, "grad_norm": 12.6484375, "learning_rate": 9.143654064809551e-06, "loss": 0.8998, "step": 24601 }, { "epoch": 2.6177910193658223, "grad_norm": 12.0703125, "learning_rate": 9.143618533257533e-06, "loss": 0.8104, "step": 24602 }, { "epoch": 2.617897424984039, "grad_norm": 20.203125, "learning_rate": 9.143583001705515e-06, "loss": 0.9818, "step": 24603 }, { "epoch": 2.6180038306022557, "grad_norm": 14.9375, "learning_rate": 9.143547470153496e-06, "loss": 0.9783, "step": 24604 }, { "epoch": 2.6181102362204722, "grad_norm": 12.859375, "learning_rate": 9.14351193860148e-06, "loss": 0.98, "step": 24605 }, { "epoch": 2.618216641838689, "grad_norm": 25.6875, "learning_rate": 9.143476407049461e-06, "loss": 1.2522, "step": 24606 }, { "epoch": 2.6183230474569057, "grad_norm": 14.6953125, "learning_rate": 9.143440875497443e-06, "loss": 0.8166, "step": 24607 }, { "epoch": 2.618429453075122, "grad_norm": 11.1640625, "learning_rate": 9.143405343945424e-06, "loss": 0.9326, "step": 24608 }, { "epoch": 2.618535858693339, "grad_norm": 11.890625, "learning_rate": 9.143369812393406e-06, "loss": 1.0611, "step": 24609 }, { "epoch": 2.6186422643115557, "grad_norm": 12.9921875, "learning_rate": 9.143334280841388e-06, "loss": 0.9211, "step": 24610 }, { "epoch": 2.618748669929772, "grad_norm": 13.515625, "learning_rate": 9.143298749289369e-06, "loss": 0.757, "step": 24611 }, { "epoch": 2.618855075547989, "grad_norm": 12.7734375, "learning_rate": 9.14326321773735e-06, "loss": 0.8348, "step": 24612 }, { "epoch": 2.6189614811662056, "grad_norm": 17.078125, "learning_rate": 9.143227686185332e-06, "loss": 1.0641, "step": 24613 }, { "epoch": 2.619067886784422, "grad_norm": 10.8984375, "learning_rate": 9.143192154633314e-06, "loss": 0.7755, "step": 24614 }, { "epoch": 2.619174292402639, "grad_norm": 11.8671875, "learning_rate": 9.143156623081297e-06, "loss": 0.9302, "step": 24615 }, { "epoch": 2.6192806980208556, "grad_norm": 13.3125, "learning_rate": 9.143121091529279e-06, "loss": 0.9521, "step": 24616 }, { "epoch": 2.619387103639072, "grad_norm": 12.140625, "learning_rate": 9.14308555997726e-06, "loss": 0.9402, "step": 24617 }, { "epoch": 2.6194935092572886, "grad_norm": 21.4375, "learning_rate": 9.143050028425242e-06, "loss": 0.8665, "step": 24618 }, { "epoch": 2.6195999148755056, "grad_norm": 10.5234375, "learning_rate": 9.143014496873224e-06, "loss": 0.8632, "step": 24619 }, { "epoch": 2.619706320493722, "grad_norm": 12.7734375, "learning_rate": 9.142978965321207e-06, "loss": 0.7965, "step": 24620 }, { "epoch": 2.6198127261119386, "grad_norm": 13.078125, "learning_rate": 9.142943433769189e-06, "loss": 0.8365, "step": 24621 }, { "epoch": 2.619919131730155, "grad_norm": 10.0546875, "learning_rate": 9.14290790221717e-06, "loss": 0.84, "step": 24622 }, { "epoch": 2.620025537348372, "grad_norm": 12.2890625, "learning_rate": 9.142872370665152e-06, "loss": 0.8647, "step": 24623 }, { "epoch": 2.6201319429665886, "grad_norm": 13.140625, "learning_rate": 9.142836839113133e-06, "loss": 0.9162, "step": 24624 }, { "epoch": 2.620238348584805, "grad_norm": 14.1484375, "learning_rate": 9.142801307561115e-06, "loss": 0.8264, "step": 24625 }, { "epoch": 2.620344754203022, "grad_norm": 14.953125, "learning_rate": 9.142765776009097e-06, "loss": 0.9073, "step": 24626 }, { "epoch": 2.6204511598212386, "grad_norm": 16.578125, "learning_rate": 9.142730244457078e-06, "loss": 0.9863, "step": 24627 }, { "epoch": 2.620557565439455, "grad_norm": 13.53125, "learning_rate": 9.14269471290506e-06, "loss": 0.7681, "step": 24628 }, { "epoch": 2.620663971057672, "grad_norm": 10.2734375, "learning_rate": 9.142659181353041e-06, "loss": 0.8919, "step": 24629 }, { "epoch": 2.6207703766758885, "grad_norm": 12.9375, "learning_rate": 9.142623649801025e-06, "loss": 0.9462, "step": 24630 }, { "epoch": 2.620876782294105, "grad_norm": 11.3203125, "learning_rate": 9.142588118249006e-06, "loss": 0.9427, "step": 24631 }, { "epoch": 2.620983187912322, "grad_norm": 14.484375, "learning_rate": 9.142552586696988e-06, "loss": 1.0094, "step": 24632 }, { "epoch": 2.6210895935305385, "grad_norm": 10.5625, "learning_rate": 9.14251705514497e-06, "loss": 0.9995, "step": 24633 }, { "epoch": 2.621195999148755, "grad_norm": 12.5859375, "learning_rate": 9.142481523592951e-06, "loss": 0.969, "step": 24634 }, { "epoch": 2.621302404766972, "grad_norm": 13.46875, "learning_rate": 9.142445992040934e-06, "loss": 0.7894, "step": 24635 }, { "epoch": 2.6214088103851885, "grad_norm": 14.1875, "learning_rate": 9.142410460488916e-06, "loss": 0.8659, "step": 24636 }, { "epoch": 2.621515216003405, "grad_norm": 14.7109375, "learning_rate": 9.142374928936898e-06, "loss": 0.8217, "step": 24637 }, { "epoch": 2.6216216216216215, "grad_norm": 17.1875, "learning_rate": 9.142339397384878e-06, "loss": 1.0515, "step": 24638 }, { "epoch": 2.621728027239838, "grad_norm": 12.671875, "learning_rate": 9.14230386583286e-06, "loss": 0.737, "step": 24639 }, { "epoch": 2.621834432858055, "grad_norm": 13.0390625, "learning_rate": 9.14226833428084e-06, "loss": 1.0118, "step": 24640 }, { "epoch": 2.6219408384762715, "grad_norm": 14.6875, "learning_rate": 9.142232802728824e-06, "loss": 0.656, "step": 24641 }, { "epoch": 2.622047244094488, "grad_norm": 13.234375, "learning_rate": 9.142197271176806e-06, "loss": 1.0565, "step": 24642 }, { "epoch": 2.622153649712705, "grad_norm": 12.71875, "learning_rate": 9.142161739624787e-06, "loss": 0.8444, "step": 24643 }, { "epoch": 2.6222600553309214, "grad_norm": 13.484375, "learning_rate": 9.142126208072769e-06, "loss": 1.0334, "step": 24644 }, { "epoch": 2.622366460949138, "grad_norm": 10.3828125, "learning_rate": 9.14209067652075e-06, "loss": 0.9078, "step": 24645 }, { "epoch": 2.622472866567355, "grad_norm": 12.703125, "learning_rate": 9.142055144968734e-06, "loss": 0.7993, "step": 24646 }, { "epoch": 2.6225792721855714, "grad_norm": 14.8203125, "learning_rate": 9.142019613416715e-06, "loss": 1.0654, "step": 24647 }, { "epoch": 2.622685677803788, "grad_norm": 11.3046875, "learning_rate": 9.141984081864697e-06, "loss": 0.8778, "step": 24648 }, { "epoch": 2.622792083422005, "grad_norm": 12.9765625, "learning_rate": 9.141948550312679e-06, "loss": 0.8463, "step": 24649 }, { "epoch": 2.6228984890402214, "grad_norm": 11.265625, "learning_rate": 9.14191301876066e-06, "loss": 0.7746, "step": 24650 }, { "epoch": 2.623004894658438, "grad_norm": 10.09375, "learning_rate": 9.141877487208642e-06, "loss": 0.903, "step": 24651 }, { "epoch": 2.623111300276655, "grad_norm": 13.8359375, "learning_rate": 9.141841955656623e-06, "loss": 0.9799, "step": 24652 }, { "epoch": 2.6232177058948714, "grad_norm": 11.5, "learning_rate": 9.141806424104605e-06, "loss": 0.8736, "step": 24653 }, { "epoch": 2.623324111513088, "grad_norm": 16.90625, "learning_rate": 9.141770892552587e-06, "loss": 0.9712, "step": 24654 }, { "epoch": 2.6234305171313044, "grad_norm": 13.1171875, "learning_rate": 9.141735361000568e-06, "loss": 0.7955, "step": 24655 }, { "epoch": 2.6235369227495213, "grad_norm": 9.8984375, "learning_rate": 9.141699829448552e-06, "loss": 0.7583, "step": 24656 }, { "epoch": 2.623643328367738, "grad_norm": 11.0625, "learning_rate": 9.141664297896533e-06, "loss": 0.8831, "step": 24657 }, { "epoch": 2.6237497339859543, "grad_norm": 12.5234375, "learning_rate": 9.141628766344515e-06, "loss": 1.0225, "step": 24658 }, { "epoch": 2.623856139604171, "grad_norm": 12.7109375, "learning_rate": 9.141593234792496e-06, "loss": 0.8193, "step": 24659 }, { "epoch": 2.623962545222388, "grad_norm": 10.6640625, "learning_rate": 9.141557703240478e-06, "loss": 0.8073, "step": 24660 }, { "epoch": 2.6240689508406043, "grad_norm": 12.7109375, "learning_rate": 9.141522171688461e-06, "loss": 0.8149, "step": 24661 }, { "epoch": 2.624175356458821, "grad_norm": 15.7109375, "learning_rate": 9.141486640136443e-06, "loss": 0.9373, "step": 24662 }, { "epoch": 2.624281762077038, "grad_norm": 16.375, "learning_rate": 9.141451108584425e-06, "loss": 1.1253, "step": 24663 }, { "epoch": 2.6243881676952543, "grad_norm": 11.8125, "learning_rate": 9.141415577032406e-06, "loss": 0.8497, "step": 24664 }, { "epoch": 2.624494573313471, "grad_norm": 15.734375, "learning_rate": 9.141380045480388e-06, "loss": 0.9443, "step": 24665 }, { "epoch": 2.6246009789316878, "grad_norm": 12.15625, "learning_rate": 9.14134451392837e-06, "loss": 0.9141, "step": 24666 }, { "epoch": 2.6247073845499043, "grad_norm": 10.1796875, "learning_rate": 9.141308982376351e-06, "loss": 0.9993, "step": 24667 }, { "epoch": 2.6248137901681208, "grad_norm": 12.6875, "learning_rate": 9.141273450824333e-06, "loss": 0.9068, "step": 24668 }, { "epoch": 2.6249201957863377, "grad_norm": 11.7578125, "learning_rate": 9.141237919272314e-06, "loss": 1.026, "step": 24669 }, { "epoch": 2.6250266014045542, "grad_norm": 10.5234375, "learning_rate": 9.141202387720296e-06, "loss": 0.911, "step": 24670 }, { "epoch": 2.6251330070227707, "grad_norm": 12.4765625, "learning_rate": 9.141166856168279e-06, "loss": 0.7805, "step": 24671 }, { "epoch": 2.6252394126409877, "grad_norm": 10.71875, "learning_rate": 9.14113132461626e-06, "loss": 0.762, "step": 24672 }, { "epoch": 2.625345818259204, "grad_norm": 13.1328125, "learning_rate": 9.141095793064242e-06, "loss": 0.9303, "step": 24673 }, { "epoch": 2.6254522238774207, "grad_norm": 10.65625, "learning_rate": 9.141060261512224e-06, "loss": 0.9285, "step": 24674 }, { "epoch": 2.6255586294956372, "grad_norm": 12.5625, "learning_rate": 9.141024729960205e-06, "loss": 1.0257, "step": 24675 }, { "epoch": 2.6256650351138537, "grad_norm": 12.2734375, "learning_rate": 9.140989198408187e-06, "loss": 0.8202, "step": 24676 }, { "epoch": 2.6257714407320707, "grad_norm": 21.5625, "learning_rate": 9.140953666856169e-06, "loss": 1.0463, "step": 24677 }, { "epoch": 2.625877846350287, "grad_norm": 11.8046875, "learning_rate": 9.14091813530415e-06, "loss": 1.1031, "step": 24678 }, { "epoch": 2.6259842519685037, "grad_norm": 12.6875, "learning_rate": 9.140882603752132e-06, "loss": 0.9215, "step": 24679 }, { "epoch": 2.6260906575867207, "grad_norm": 12.9296875, "learning_rate": 9.140847072200114e-06, "loss": 0.9463, "step": 24680 }, { "epoch": 2.626197063204937, "grad_norm": 8.734375, "learning_rate": 9.140811540648095e-06, "loss": 0.7802, "step": 24681 }, { "epoch": 2.6263034688231537, "grad_norm": 13.8203125, "learning_rate": 9.140776009096078e-06, "loss": 0.8323, "step": 24682 }, { "epoch": 2.6264098744413706, "grad_norm": 15.125, "learning_rate": 9.14074047754406e-06, "loss": 0.9111, "step": 24683 }, { "epoch": 2.626516280059587, "grad_norm": 11.484375, "learning_rate": 9.140704945992042e-06, "loss": 0.96, "step": 24684 }, { "epoch": 2.6266226856778037, "grad_norm": 10.453125, "learning_rate": 9.140669414440023e-06, "loss": 0.7978, "step": 24685 }, { "epoch": 2.6267290912960206, "grad_norm": 17.484375, "learning_rate": 9.140633882888005e-06, "loss": 1.1021, "step": 24686 }, { "epoch": 2.626835496914237, "grad_norm": 10.6328125, "learning_rate": 9.140598351335988e-06, "loss": 1.0192, "step": 24687 }, { "epoch": 2.6269419025324536, "grad_norm": 19.015625, "learning_rate": 9.14056281978397e-06, "loss": 1.0953, "step": 24688 }, { "epoch": 2.6270483081506706, "grad_norm": 12.6796875, "learning_rate": 9.140527288231951e-06, "loss": 0.9282, "step": 24689 }, { "epoch": 2.627154713768887, "grad_norm": 16.28125, "learning_rate": 9.140491756679933e-06, "loss": 0.6837, "step": 24690 }, { "epoch": 2.6272611193871036, "grad_norm": 12.03125, "learning_rate": 9.140456225127915e-06, "loss": 1.0283, "step": 24691 }, { "epoch": 2.62736752500532, "grad_norm": 9.7734375, "learning_rate": 9.140420693575896e-06, "loss": 0.779, "step": 24692 }, { "epoch": 2.627473930623537, "grad_norm": 10.9453125, "learning_rate": 9.140385162023878e-06, "loss": 1.0099, "step": 24693 }, { "epoch": 2.6275803362417536, "grad_norm": 15.0703125, "learning_rate": 9.14034963047186e-06, "loss": 1.0387, "step": 24694 }, { "epoch": 2.62768674185997, "grad_norm": 14.0234375, "learning_rate": 9.140314098919841e-06, "loss": 1.0406, "step": 24695 }, { "epoch": 2.6277931474781866, "grad_norm": 12.5078125, "learning_rate": 9.140278567367823e-06, "loss": 0.7812, "step": 24696 }, { "epoch": 2.6278995530964036, "grad_norm": 12.59375, "learning_rate": 9.140243035815806e-06, "loss": 0.8533, "step": 24697 }, { "epoch": 2.62800595871462, "grad_norm": 11.6328125, "learning_rate": 9.140207504263788e-06, "loss": 0.8092, "step": 24698 }, { "epoch": 2.6281123643328366, "grad_norm": 22.203125, "learning_rate": 9.140171972711769e-06, "loss": 1.0439, "step": 24699 }, { "epoch": 2.6282187699510535, "grad_norm": 13.2421875, "learning_rate": 9.14013644115975e-06, "loss": 0.9015, "step": 24700 }, { "epoch": 2.62832517556927, "grad_norm": 10.65625, "learning_rate": 9.140100909607732e-06, "loss": 0.7275, "step": 24701 }, { "epoch": 2.6284315811874865, "grad_norm": 10.8984375, "learning_rate": 9.140065378055714e-06, "loss": 0.6825, "step": 24702 }, { "epoch": 2.6285379868057035, "grad_norm": 17.09375, "learning_rate": 9.140029846503696e-06, "loss": 1.0286, "step": 24703 }, { "epoch": 2.62864439242392, "grad_norm": 12.890625, "learning_rate": 9.139994314951677e-06, "loss": 0.8284, "step": 24704 }, { "epoch": 2.6287507980421365, "grad_norm": 12.3671875, "learning_rate": 9.139958783399659e-06, "loss": 0.846, "step": 24705 }, { "epoch": 2.6288572036603535, "grad_norm": 16.296875, "learning_rate": 9.13992325184764e-06, "loss": 1.0274, "step": 24706 }, { "epoch": 2.62896360927857, "grad_norm": 11.9609375, "learning_rate": 9.139887720295624e-06, "loss": 0.9623, "step": 24707 }, { "epoch": 2.6290700148967865, "grad_norm": 11.265625, "learning_rate": 9.139852188743605e-06, "loss": 1.0629, "step": 24708 }, { "epoch": 2.6291764205150034, "grad_norm": 11.5234375, "learning_rate": 9.139816657191587e-06, "loss": 0.952, "step": 24709 }, { "epoch": 2.62928282613322, "grad_norm": 14.9296875, "learning_rate": 9.139781125639568e-06, "loss": 1.0792, "step": 24710 }, { "epoch": 2.6293892317514365, "grad_norm": 12.0390625, "learning_rate": 9.13974559408755e-06, "loss": 1.055, "step": 24711 }, { "epoch": 2.629495637369653, "grad_norm": 11.125, "learning_rate": 9.139710062535533e-06, "loss": 0.9253, "step": 24712 }, { "epoch": 2.62960204298787, "grad_norm": 14.4375, "learning_rate": 9.139674530983515e-06, "loss": 1.0917, "step": 24713 }, { "epoch": 2.6297084486060864, "grad_norm": 11.453125, "learning_rate": 9.139638999431497e-06, "loss": 0.7737, "step": 24714 }, { "epoch": 2.629814854224303, "grad_norm": 12.109375, "learning_rate": 9.139603467879478e-06, "loss": 0.9511, "step": 24715 }, { "epoch": 2.6299212598425195, "grad_norm": 16.765625, "learning_rate": 9.13956793632746e-06, "loss": 1.0367, "step": 24716 }, { "epoch": 2.6300276654607364, "grad_norm": 19.1875, "learning_rate": 9.139532404775441e-06, "loss": 0.9569, "step": 24717 }, { "epoch": 2.630134071078953, "grad_norm": 10.84375, "learning_rate": 9.139496873223423e-06, "loss": 0.8998, "step": 24718 }, { "epoch": 2.6302404766971694, "grad_norm": 10.8359375, "learning_rate": 9.139461341671405e-06, "loss": 0.8776, "step": 24719 }, { "epoch": 2.6303468823153864, "grad_norm": 14.3515625, "learning_rate": 9.139425810119386e-06, "loss": 0.982, "step": 24720 }, { "epoch": 2.630453287933603, "grad_norm": 11.328125, "learning_rate": 9.139390278567368e-06, "loss": 0.842, "step": 24721 }, { "epoch": 2.6305596935518194, "grad_norm": 10.5625, "learning_rate": 9.13935474701535e-06, "loss": 0.779, "step": 24722 }, { "epoch": 2.6306660991700364, "grad_norm": 13.2734375, "learning_rate": 9.139319215463333e-06, "loss": 0.8158, "step": 24723 }, { "epoch": 2.630772504788253, "grad_norm": 10.3671875, "learning_rate": 9.139283683911314e-06, "loss": 0.7503, "step": 24724 }, { "epoch": 2.6308789104064694, "grad_norm": 18.96875, "learning_rate": 9.139248152359296e-06, "loss": 0.851, "step": 24725 }, { "epoch": 2.6309853160246863, "grad_norm": 11.3671875, "learning_rate": 9.139212620807278e-06, "loss": 0.9263, "step": 24726 }, { "epoch": 2.631091721642903, "grad_norm": 14.359375, "learning_rate": 9.13917708925526e-06, "loss": 0.9234, "step": 24727 }, { "epoch": 2.6311981272611193, "grad_norm": 14.40625, "learning_rate": 9.13914155770324e-06, "loss": 0.8404, "step": 24728 }, { "epoch": 2.6313045328793363, "grad_norm": 10.28125, "learning_rate": 9.139106026151222e-06, "loss": 1.0456, "step": 24729 }, { "epoch": 2.631410938497553, "grad_norm": 10.1953125, "learning_rate": 9.139070494599204e-06, "loss": 0.8631, "step": 24730 }, { "epoch": 2.6315173441157693, "grad_norm": 12.8359375, "learning_rate": 9.139034963047186e-06, "loss": 0.9219, "step": 24731 }, { "epoch": 2.631623749733986, "grad_norm": 12.375, "learning_rate": 9.138999431495167e-06, "loss": 0.9842, "step": 24732 }, { "epoch": 2.6317301553522023, "grad_norm": 11.7578125, "learning_rate": 9.13896389994315e-06, "loss": 0.643, "step": 24733 }, { "epoch": 2.6318365609704193, "grad_norm": 13.5078125, "learning_rate": 9.138928368391132e-06, "loss": 0.9989, "step": 24734 }, { "epoch": 2.631942966588636, "grad_norm": 15.046875, "learning_rate": 9.138892836839114e-06, "loss": 0.9625, "step": 24735 }, { "epoch": 2.6320493722068523, "grad_norm": 12.7734375, "learning_rate": 9.138857305287095e-06, "loss": 0.8258, "step": 24736 }, { "epoch": 2.6321557778250693, "grad_norm": 14.328125, "learning_rate": 9.138821773735077e-06, "loss": 0.7671, "step": 24737 }, { "epoch": 2.6322621834432858, "grad_norm": 12.3828125, "learning_rate": 9.13878624218306e-06, "loss": 0.982, "step": 24738 }, { "epoch": 2.6323685890615023, "grad_norm": 23.109375, "learning_rate": 9.138750710631042e-06, "loss": 1.0212, "step": 24739 }, { "epoch": 2.6324749946797192, "grad_norm": 10.078125, "learning_rate": 9.138715179079023e-06, "loss": 0.9103, "step": 24740 }, { "epoch": 2.6325814002979357, "grad_norm": 15.03125, "learning_rate": 9.138679647527005e-06, "loss": 0.898, "step": 24741 }, { "epoch": 2.6326878059161523, "grad_norm": 17.859375, "learning_rate": 9.138644115974987e-06, "loss": 0.7712, "step": 24742 }, { "epoch": 2.632794211534369, "grad_norm": 12.859375, "learning_rate": 9.138608584422968e-06, "loss": 0.8575, "step": 24743 }, { "epoch": 2.6329006171525857, "grad_norm": 15.6328125, "learning_rate": 9.13857305287095e-06, "loss": 0.9485, "step": 24744 }, { "epoch": 2.6330070227708022, "grad_norm": 21.5625, "learning_rate": 9.138537521318932e-06, "loss": 1.1852, "step": 24745 }, { "epoch": 2.633113428389019, "grad_norm": 12.390625, "learning_rate": 9.138501989766913e-06, "loss": 0.89, "step": 24746 }, { "epoch": 2.6332198340072357, "grad_norm": 18.171875, "learning_rate": 9.138466458214895e-06, "loss": 1.0858, "step": 24747 }, { "epoch": 2.633326239625452, "grad_norm": 14.5625, "learning_rate": 9.138430926662878e-06, "loss": 0.852, "step": 24748 }, { "epoch": 2.6334326452436687, "grad_norm": 13.3515625, "learning_rate": 9.13839539511086e-06, "loss": 0.9498, "step": 24749 }, { "epoch": 2.6335390508618857, "grad_norm": 11.1875, "learning_rate": 9.138359863558841e-06, "loss": 0.8101, "step": 24750 }, { "epoch": 2.633645456480102, "grad_norm": 17.15625, "learning_rate": 9.138324332006823e-06, "loss": 0.9714, "step": 24751 }, { "epoch": 2.6337518620983187, "grad_norm": 13.59375, "learning_rate": 9.138288800454804e-06, "loss": 0.8052, "step": 24752 }, { "epoch": 2.633858267716535, "grad_norm": 10.3203125, "learning_rate": 9.138253268902788e-06, "loss": 0.8734, "step": 24753 }, { "epoch": 2.633964673334752, "grad_norm": 12.5703125, "learning_rate": 9.13821773735077e-06, "loss": 1.0107, "step": 24754 }, { "epoch": 2.6340710789529687, "grad_norm": 15.75, "learning_rate": 9.138182205798751e-06, "loss": 0.8786, "step": 24755 }, { "epoch": 2.634177484571185, "grad_norm": 13.78125, "learning_rate": 9.138146674246733e-06, "loss": 0.9455, "step": 24756 }, { "epoch": 2.634283890189402, "grad_norm": 13.53125, "learning_rate": 9.138111142694714e-06, "loss": 0.9238, "step": 24757 }, { "epoch": 2.6343902958076186, "grad_norm": 10.4609375, "learning_rate": 9.138075611142694e-06, "loss": 0.7006, "step": 24758 }, { "epoch": 2.634496701425835, "grad_norm": 11.5234375, "learning_rate": 9.138040079590677e-06, "loss": 0.7436, "step": 24759 }, { "epoch": 2.634603107044052, "grad_norm": 14.578125, "learning_rate": 9.138004548038659e-06, "loss": 0.9257, "step": 24760 }, { "epoch": 2.6347095126622686, "grad_norm": 11.8671875, "learning_rate": 9.13796901648664e-06, "loss": 0.9224, "step": 24761 }, { "epoch": 2.634815918280485, "grad_norm": 15.1484375, "learning_rate": 9.137933484934622e-06, "loss": 0.8528, "step": 24762 }, { "epoch": 2.634922323898702, "grad_norm": 14.3203125, "learning_rate": 9.137897953382604e-06, "loss": 0.9253, "step": 24763 }, { "epoch": 2.6350287295169186, "grad_norm": 9.5390625, "learning_rate": 9.137862421830587e-06, "loss": 0.7212, "step": 24764 }, { "epoch": 2.635135135135135, "grad_norm": 13.34375, "learning_rate": 9.137826890278569e-06, "loss": 0.8752, "step": 24765 }, { "epoch": 2.635241540753352, "grad_norm": 12.8359375, "learning_rate": 9.13779135872655e-06, "loss": 0.994, "step": 24766 }, { "epoch": 2.6353479463715686, "grad_norm": 11.5390625, "learning_rate": 9.137755827174532e-06, "loss": 0.6795, "step": 24767 }, { "epoch": 2.635454351989785, "grad_norm": 15.390625, "learning_rate": 9.137720295622514e-06, "loss": 0.7935, "step": 24768 }, { "epoch": 2.6355607576080016, "grad_norm": 12.53125, "learning_rate": 9.137684764070495e-06, "loss": 0.9197, "step": 24769 }, { "epoch": 2.635667163226218, "grad_norm": 11.6640625, "learning_rate": 9.137649232518477e-06, "loss": 0.8633, "step": 24770 }, { "epoch": 2.635773568844435, "grad_norm": 13.15625, "learning_rate": 9.137613700966458e-06, "loss": 0.9461, "step": 24771 }, { "epoch": 2.6358799744626515, "grad_norm": 15.671875, "learning_rate": 9.13757816941444e-06, "loss": 1.0241, "step": 24772 }, { "epoch": 2.635986380080868, "grad_norm": 10.4921875, "learning_rate": 9.137542637862422e-06, "loss": 0.9485, "step": 24773 }, { "epoch": 2.636092785699085, "grad_norm": 14.859375, "learning_rate": 9.137507106310405e-06, "loss": 0.9087, "step": 24774 }, { "epoch": 2.6361991913173015, "grad_norm": 12.0703125, "learning_rate": 9.137471574758386e-06, "loss": 0.7925, "step": 24775 }, { "epoch": 2.636305596935518, "grad_norm": 11.8359375, "learning_rate": 9.137436043206368e-06, "loss": 0.9351, "step": 24776 }, { "epoch": 2.636412002553735, "grad_norm": 13.1640625, "learning_rate": 9.13740051165435e-06, "loss": 1.0011, "step": 24777 }, { "epoch": 2.6365184081719515, "grad_norm": 12.3359375, "learning_rate": 9.137364980102331e-06, "loss": 0.9224, "step": 24778 }, { "epoch": 2.636624813790168, "grad_norm": 12.234375, "learning_rate": 9.137329448550315e-06, "loss": 1.0748, "step": 24779 }, { "epoch": 2.636731219408385, "grad_norm": 14.265625, "learning_rate": 9.137293916998296e-06, "loss": 0.9252, "step": 24780 }, { "epoch": 2.6368376250266015, "grad_norm": 19.84375, "learning_rate": 9.137258385446278e-06, "loss": 1.0001, "step": 24781 }, { "epoch": 2.636944030644818, "grad_norm": 13.1171875, "learning_rate": 9.13722285389426e-06, "loss": 0.7825, "step": 24782 }, { "epoch": 2.637050436263035, "grad_norm": 12.5625, "learning_rate": 9.137187322342241e-06, "loss": 0.8752, "step": 24783 }, { "epoch": 2.6371568418812514, "grad_norm": 15.484375, "learning_rate": 9.137151790790223e-06, "loss": 0.7229, "step": 24784 }, { "epoch": 2.637263247499468, "grad_norm": 10.1953125, "learning_rate": 9.137116259238204e-06, "loss": 0.7861, "step": 24785 }, { "epoch": 2.6373696531176845, "grad_norm": 13.4921875, "learning_rate": 9.137080727686186e-06, "loss": 0.9042, "step": 24786 }, { "epoch": 2.6374760587359014, "grad_norm": 10.546875, "learning_rate": 9.137045196134167e-06, "loss": 0.8813, "step": 24787 }, { "epoch": 2.637582464354118, "grad_norm": 15.0859375, "learning_rate": 9.137009664582149e-06, "loss": 0.9124, "step": 24788 }, { "epoch": 2.6376888699723344, "grad_norm": 13.8984375, "learning_rate": 9.136974133030132e-06, "loss": 0.8894, "step": 24789 }, { "epoch": 2.637795275590551, "grad_norm": 10.8828125, "learning_rate": 9.136938601478114e-06, "loss": 0.9372, "step": 24790 }, { "epoch": 2.637901681208768, "grad_norm": 14.703125, "learning_rate": 9.136903069926096e-06, "loss": 1.1565, "step": 24791 }, { "epoch": 2.6380080868269844, "grad_norm": 15.375, "learning_rate": 9.136867538374077e-06, "loss": 1.0264, "step": 24792 }, { "epoch": 2.638114492445201, "grad_norm": 11.65625, "learning_rate": 9.136832006822059e-06, "loss": 0.9031, "step": 24793 }, { "epoch": 2.638220898063418, "grad_norm": 11.9453125, "learning_rate": 9.13679647527004e-06, "loss": 0.9538, "step": 24794 }, { "epoch": 2.6383273036816344, "grad_norm": 14.09375, "learning_rate": 9.136760943718022e-06, "loss": 0.7447, "step": 24795 }, { "epoch": 2.638433709299851, "grad_norm": 11.2890625, "learning_rate": 9.136725412166004e-06, "loss": 0.8673, "step": 24796 }, { "epoch": 2.638540114918068, "grad_norm": 14.3359375, "learning_rate": 9.136689880613985e-06, "loss": 0.9455, "step": 24797 }, { "epoch": 2.6386465205362843, "grad_norm": 13.203125, "learning_rate": 9.136654349061967e-06, "loss": 0.9736, "step": 24798 }, { "epoch": 2.638752926154501, "grad_norm": 36.1875, "learning_rate": 9.136618817509948e-06, "loss": 0.7699, "step": 24799 }, { "epoch": 2.638859331772718, "grad_norm": 13.7421875, "learning_rate": 9.136583285957932e-06, "loss": 1.1428, "step": 24800 }, { "epoch": 2.6389657373909343, "grad_norm": 12.375, "learning_rate": 9.136547754405913e-06, "loss": 0.8151, "step": 24801 }, { "epoch": 2.639072143009151, "grad_norm": 12.25, "learning_rate": 9.136512222853895e-06, "loss": 0.7878, "step": 24802 }, { "epoch": 2.639178548627368, "grad_norm": 12.4453125, "learning_rate": 9.136476691301877e-06, "loss": 0.856, "step": 24803 }, { "epoch": 2.6392849542455843, "grad_norm": 11.8671875, "learning_rate": 9.136441159749858e-06, "loss": 0.8748, "step": 24804 }, { "epoch": 2.639391359863801, "grad_norm": 12.84375, "learning_rate": 9.136405628197841e-06, "loss": 1.1138, "step": 24805 }, { "epoch": 2.6394977654820173, "grad_norm": 15.234375, "learning_rate": 9.136370096645823e-06, "loss": 0.863, "step": 24806 }, { "epoch": 2.639604171100234, "grad_norm": 14.734375, "learning_rate": 9.136334565093805e-06, "loss": 0.9513, "step": 24807 }, { "epoch": 2.6397105767184508, "grad_norm": 11.5234375, "learning_rate": 9.136299033541786e-06, "loss": 0.9326, "step": 24808 }, { "epoch": 2.6398169823366673, "grad_norm": 22.78125, "learning_rate": 9.136263501989768e-06, "loss": 0.9029, "step": 24809 }, { "epoch": 2.639923387954884, "grad_norm": 13.296875, "learning_rate": 9.13622797043775e-06, "loss": 0.9301, "step": 24810 }, { "epoch": 2.6400297935731007, "grad_norm": 13.9296875, "learning_rate": 9.136192438885731e-06, "loss": 0.8909, "step": 24811 }, { "epoch": 2.6401361991913173, "grad_norm": 16.640625, "learning_rate": 9.136156907333713e-06, "loss": 1.0098, "step": 24812 }, { "epoch": 2.6402426048095338, "grad_norm": 20.0, "learning_rate": 9.136121375781694e-06, "loss": 0.9832, "step": 24813 }, { "epoch": 2.6403490104277507, "grad_norm": 12.6328125, "learning_rate": 9.136085844229676e-06, "loss": 0.8639, "step": 24814 }, { "epoch": 2.6404554160459672, "grad_norm": 15.4453125, "learning_rate": 9.13605031267766e-06, "loss": 0.8707, "step": 24815 }, { "epoch": 2.6405618216641837, "grad_norm": 12.9609375, "learning_rate": 9.13601478112564e-06, "loss": 0.6714, "step": 24816 }, { "epoch": 2.6406682272824007, "grad_norm": 11.8828125, "learning_rate": 9.135979249573622e-06, "loss": 0.958, "step": 24817 }, { "epoch": 2.640774632900617, "grad_norm": 12.15625, "learning_rate": 9.135943718021604e-06, "loss": 1.1218, "step": 24818 }, { "epoch": 2.6408810385188337, "grad_norm": 15.1875, "learning_rate": 9.135908186469586e-06, "loss": 0.9932, "step": 24819 }, { "epoch": 2.6409874441370507, "grad_norm": 11.234375, "learning_rate": 9.135872654917567e-06, "loss": 0.8455, "step": 24820 }, { "epoch": 2.641093849755267, "grad_norm": 11.8046875, "learning_rate": 9.135837123365549e-06, "loss": 0.956, "step": 24821 }, { "epoch": 2.6412002553734837, "grad_norm": 13.65625, "learning_rate": 9.13580159181353e-06, "loss": 0.8688, "step": 24822 }, { "epoch": 2.6413066609917006, "grad_norm": 12.8828125, "learning_rate": 9.135766060261512e-06, "loss": 0.9123, "step": 24823 }, { "epoch": 2.641413066609917, "grad_norm": 10.828125, "learning_rate": 9.135730528709494e-06, "loss": 0.9092, "step": 24824 }, { "epoch": 2.6415194722281337, "grad_norm": 23.90625, "learning_rate": 9.135694997157477e-06, "loss": 0.931, "step": 24825 }, { "epoch": 2.64162587784635, "grad_norm": 17.5, "learning_rate": 9.135659465605459e-06, "loss": 0.9629, "step": 24826 }, { "epoch": 2.6417322834645667, "grad_norm": 13.75, "learning_rate": 9.13562393405344e-06, "loss": 0.9661, "step": 24827 }, { "epoch": 2.6418386890827836, "grad_norm": 14.140625, "learning_rate": 9.135588402501422e-06, "loss": 0.9117, "step": 24828 }, { "epoch": 2.641945094701, "grad_norm": 15.8125, "learning_rate": 9.135552870949403e-06, "loss": 0.9167, "step": 24829 }, { "epoch": 2.6420515003192167, "grad_norm": 11.4453125, "learning_rate": 9.135517339397387e-06, "loss": 0.826, "step": 24830 }, { "epoch": 2.6421579059374336, "grad_norm": 10.8515625, "learning_rate": 9.135481807845368e-06, "loss": 0.9498, "step": 24831 }, { "epoch": 2.64226431155565, "grad_norm": 15.984375, "learning_rate": 9.13544627629335e-06, "loss": 0.8825, "step": 24832 }, { "epoch": 2.6423707171738666, "grad_norm": 12.265625, "learning_rate": 9.135410744741332e-06, "loss": 0.7005, "step": 24833 }, { "epoch": 2.6424771227920836, "grad_norm": 349.5, "learning_rate": 9.135375213189313e-06, "loss": 1.4551, "step": 24834 }, { "epoch": 2.6425835284103, "grad_norm": 21.15625, "learning_rate": 9.135339681637295e-06, "loss": 0.922, "step": 24835 }, { "epoch": 2.6426899340285166, "grad_norm": 13.15625, "learning_rate": 9.135304150085276e-06, "loss": 0.9471, "step": 24836 }, { "epoch": 2.6427963396467335, "grad_norm": 13.84375, "learning_rate": 9.135268618533258e-06, "loss": 0.8316, "step": 24837 }, { "epoch": 2.64290274526495, "grad_norm": 11.609375, "learning_rate": 9.13523308698124e-06, "loss": 0.8049, "step": 24838 }, { "epoch": 2.6430091508831666, "grad_norm": 20.28125, "learning_rate": 9.135197555429221e-06, "loss": 1.0057, "step": 24839 }, { "epoch": 2.6431155565013835, "grad_norm": 16.15625, "learning_rate": 9.135162023877203e-06, "loss": 0.8782, "step": 24840 }, { "epoch": 2.6432219621196, "grad_norm": 13.5625, "learning_rate": 9.135126492325186e-06, "loss": 0.9498, "step": 24841 }, { "epoch": 2.6433283677378165, "grad_norm": 12.015625, "learning_rate": 9.135090960773168e-06, "loss": 1.021, "step": 24842 }, { "epoch": 2.643434773356033, "grad_norm": 12.796875, "learning_rate": 9.13505542922115e-06, "loss": 0.8033, "step": 24843 }, { "epoch": 2.64354117897425, "grad_norm": 12.296875, "learning_rate": 9.135019897669131e-06, "loss": 0.6873, "step": 24844 }, { "epoch": 2.6436475845924665, "grad_norm": 12.3515625, "learning_rate": 9.134984366117113e-06, "loss": 0.977, "step": 24845 }, { "epoch": 2.643753990210683, "grad_norm": 17.421875, "learning_rate": 9.134948834565096e-06, "loss": 1.0297, "step": 24846 }, { "epoch": 2.6438603958288995, "grad_norm": 16.609375, "learning_rate": 9.134913303013076e-06, "loss": 0.7841, "step": 24847 }, { "epoch": 2.6439668014471165, "grad_norm": 13.5703125, "learning_rate": 9.134877771461057e-06, "loss": 0.8847, "step": 24848 }, { "epoch": 2.644073207065333, "grad_norm": 14.6328125, "learning_rate": 9.134842239909039e-06, "loss": 0.8879, "step": 24849 }, { "epoch": 2.6441796126835495, "grad_norm": 11.09375, "learning_rate": 9.13480670835702e-06, "loss": 0.9865, "step": 24850 }, { "epoch": 2.6442860183017665, "grad_norm": 14.2734375, "learning_rate": 9.134771176805004e-06, "loss": 0.9907, "step": 24851 }, { "epoch": 2.644392423919983, "grad_norm": 11.4140625, "learning_rate": 9.134735645252985e-06, "loss": 0.849, "step": 24852 }, { "epoch": 2.6444988295381995, "grad_norm": 14.75, "learning_rate": 9.134700113700967e-06, "loss": 0.8568, "step": 24853 }, { "epoch": 2.6446052351564164, "grad_norm": 12.1171875, "learning_rate": 9.134664582148949e-06, "loss": 0.9189, "step": 24854 }, { "epoch": 2.644711640774633, "grad_norm": 10.21875, "learning_rate": 9.13462905059693e-06, "loss": 0.8228, "step": 24855 }, { "epoch": 2.6448180463928495, "grad_norm": 11.2421875, "learning_rate": 9.134593519044914e-06, "loss": 0.8293, "step": 24856 }, { "epoch": 2.6449244520110664, "grad_norm": 10.203125, "learning_rate": 9.134557987492895e-06, "loss": 0.8649, "step": 24857 }, { "epoch": 2.645030857629283, "grad_norm": 10.3984375, "learning_rate": 9.134522455940877e-06, "loss": 0.9328, "step": 24858 }, { "epoch": 2.6451372632474994, "grad_norm": 13.453125, "learning_rate": 9.134486924388858e-06, "loss": 1.0674, "step": 24859 }, { "epoch": 2.6452436688657164, "grad_norm": 12.7109375, "learning_rate": 9.13445139283684e-06, "loss": 0.7806, "step": 24860 }, { "epoch": 2.645350074483933, "grad_norm": 12.8125, "learning_rate": 9.134415861284822e-06, "loss": 0.5895, "step": 24861 }, { "epoch": 2.6454564801021494, "grad_norm": 12.609375, "learning_rate": 9.134380329732803e-06, "loss": 0.988, "step": 24862 }, { "epoch": 2.645562885720366, "grad_norm": 11.453125, "learning_rate": 9.134344798180785e-06, "loss": 0.8375, "step": 24863 }, { "epoch": 2.6456692913385824, "grad_norm": 11.2890625, "learning_rate": 9.134309266628766e-06, "loss": 0.7116, "step": 24864 }, { "epoch": 2.6457756969567994, "grad_norm": 10.9296875, "learning_rate": 9.134273735076748e-06, "loss": 0.7453, "step": 24865 }, { "epoch": 2.645882102575016, "grad_norm": 12.9296875, "learning_rate": 9.134238203524731e-06, "loss": 0.9045, "step": 24866 }, { "epoch": 2.6459885081932324, "grad_norm": 11.9921875, "learning_rate": 9.134202671972713e-06, "loss": 0.8673, "step": 24867 }, { "epoch": 2.6460949138114493, "grad_norm": 13.546875, "learning_rate": 9.134167140420695e-06, "loss": 0.9329, "step": 24868 }, { "epoch": 2.646201319429666, "grad_norm": 14.5625, "learning_rate": 9.134131608868676e-06, "loss": 1.1369, "step": 24869 }, { "epoch": 2.6463077250478824, "grad_norm": 11.9453125, "learning_rate": 9.134096077316658e-06, "loss": 0.87, "step": 24870 }, { "epoch": 2.6464141306660993, "grad_norm": 13.4140625, "learning_rate": 9.134060545764641e-06, "loss": 0.9194, "step": 24871 }, { "epoch": 2.646520536284316, "grad_norm": 21.046875, "learning_rate": 9.134025014212623e-06, "loss": 1.0411, "step": 24872 }, { "epoch": 2.6466269419025323, "grad_norm": 11.1640625, "learning_rate": 9.133989482660604e-06, "loss": 0.8856, "step": 24873 }, { "epoch": 2.6467333475207493, "grad_norm": 15.375, "learning_rate": 9.133953951108586e-06, "loss": 0.8817, "step": 24874 }, { "epoch": 2.646839753138966, "grad_norm": 14.1953125, "learning_rate": 9.133918419556567e-06, "loss": 1.1383, "step": 24875 }, { "epoch": 2.6469461587571823, "grad_norm": 11.734375, "learning_rate": 9.133882888004549e-06, "loss": 0.8144, "step": 24876 }, { "epoch": 2.6470525643753993, "grad_norm": 15.328125, "learning_rate": 9.13384735645253e-06, "loss": 0.9616, "step": 24877 }, { "epoch": 2.6471589699936158, "grad_norm": 12.7265625, "learning_rate": 9.133811824900512e-06, "loss": 0.7977, "step": 24878 }, { "epoch": 2.6472653756118323, "grad_norm": 14.953125, "learning_rate": 9.133776293348494e-06, "loss": 1.0592, "step": 24879 }, { "epoch": 2.647371781230049, "grad_norm": 17.125, "learning_rate": 9.133740761796476e-06, "loss": 0.8981, "step": 24880 }, { "epoch": 2.6474781868482657, "grad_norm": 12.59375, "learning_rate": 9.133705230244457e-06, "loss": 0.8702, "step": 24881 }, { "epoch": 2.6475845924664823, "grad_norm": 17.78125, "learning_rate": 9.13366969869244e-06, "loss": 0.9107, "step": 24882 }, { "epoch": 2.6476909980846988, "grad_norm": 16.96875, "learning_rate": 9.133634167140422e-06, "loss": 1.0755, "step": 24883 }, { "epoch": 2.6477974037029153, "grad_norm": 12.0859375, "learning_rate": 9.133598635588404e-06, "loss": 0.8496, "step": 24884 }, { "epoch": 2.6479038093211322, "grad_norm": 12.3984375, "learning_rate": 9.133563104036385e-06, "loss": 1.0354, "step": 24885 }, { "epoch": 2.6480102149393487, "grad_norm": 14.0703125, "learning_rate": 9.133527572484367e-06, "loss": 0.9738, "step": 24886 }, { "epoch": 2.6481166205575652, "grad_norm": 9.375, "learning_rate": 9.133492040932348e-06, "loss": 0.8454, "step": 24887 }, { "epoch": 2.648223026175782, "grad_norm": 16.6875, "learning_rate": 9.13345650938033e-06, "loss": 0.8988, "step": 24888 }, { "epoch": 2.6483294317939987, "grad_norm": 17.59375, "learning_rate": 9.133420977828312e-06, "loss": 0.9632, "step": 24889 }, { "epoch": 2.648435837412215, "grad_norm": 21.65625, "learning_rate": 9.133385446276293e-06, "loss": 1.0395, "step": 24890 }, { "epoch": 2.648542243030432, "grad_norm": 18.46875, "learning_rate": 9.133349914724275e-06, "loss": 1.0515, "step": 24891 }, { "epoch": 2.6486486486486487, "grad_norm": 13.3671875, "learning_rate": 9.133314383172258e-06, "loss": 0.868, "step": 24892 }, { "epoch": 2.648755054266865, "grad_norm": 15.5234375, "learning_rate": 9.13327885162024e-06, "loss": 0.8299, "step": 24893 }, { "epoch": 2.648861459885082, "grad_norm": 11.5, "learning_rate": 9.133243320068221e-06, "loss": 0.708, "step": 24894 }, { "epoch": 2.6489678655032987, "grad_norm": 12.71875, "learning_rate": 9.133207788516203e-06, "loss": 0.9535, "step": 24895 }, { "epoch": 2.649074271121515, "grad_norm": 15.6953125, "learning_rate": 9.133172256964185e-06, "loss": 0.9704, "step": 24896 }, { "epoch": 2.649180676739732, "grad_norm": 12.3203125, "learning_rate": 9.133136725412168e-06, "loss": 0.9131, "step": 24897 }, { "epoch": 2.6492870823579486, "grad_norm": 15.3359375, "learning_rate": 9.13310119386015e-06, "loss": 1.0445, "step": 24898 }, { "epoch": 2.649393487976165, "grad_norm": 10.46875, "learning_rate": 9.133065662308131e-06, "loss": 0.8751, "step": 24899 }, { "epoch": 2.6494998935943816, "grad_norm": 13.9453125, "learning_rate": 9.133030130756113e-06, "loss": 0.9422, "step": 24900 }, { "epoch": 2.649606299212598, "grad_norm": 12.5703125, "learning_rate": 9.132994599204094e-06, "loss": 0.9626, "step": 24901 }, { "epoch": 2.649712704830815, "grad_norm": 15.5, "learning_rate": 9.132959067652076e-06, "loss": 0.8641, "step": 24902 }, { "epoch": 2.6498191104490316, "grad_norm": 16.828125, "learning_rate": 9.132923536100058e-06, "loss": 0.9986, "step": 24903 }, { "epoch": 2.649925516067248, "grad_norm": 12.546875, "learning_rate": 9.13288800454804e-06, "loss": 0.9002, "step": 24904 }, { "epoch": 2.650031921685465, "grad_norm": 10.953125, "learning_rate": 9.13285247299602e-06, "loss": 0.9741, "step": 24905 }, { "epoch": 2.6501383273036816, "grad_norm": 14.2890625, "learning_rate": 9.132816941444002e-06, "loss": 0.8297, "step": 24906 }, { "epoch": 2.650244732921898, "grad_norm": 14.1640625, "learning_rate": 9.132781409891986e-06, "loss": 0.8926, "step": 24907 }, { "epoch": 2.650351138540115, "grad_norm": 16.265625, "learning_rate": 9.132745878339967e-06, "loss": 0.8522, "step": 24908 }, { "epoch": 2.6504575441583316, "grad_norm": 20.03125, "learning_rate": 9.132710346787949e-06, "loss": 0.7823, "step": 24909 }, { "epoch": 2.650563949776548, "grad_norm": 15.8125, "learning_rate": 9.13267481523593e-06, "loss": 0.901, "step": 24910 }, { "epoch": 2.650670355394765, "grad_norm": 9.8984375, "learning_rate": 9.132639283683912e-06, "loss": 0.7304, "step": 24911 }, { "epoch": 2.6507767610129815, "grad_norm": 12.234375, "learning_rate": 9.132603752131894e-06, "loss": 0.7846, "step": 24912 }, { "epoch": 2.650883166631198, "grad_norm": 14.9296875, "learning_rate": 9.132568220579875e-06, "loss": 1.045, "step": 24913 }, { "epoch": 2.650989572249415, "grad_norm": 12.0, "learning_rate": 9.132532689027857e-06, "loss": 1.0571, "step": 24914 }, { "epoch": 2.6510959778676315, "grad_norm": 12.609375, "learning_rate": 9.132497157475839e-06, "loss": 0.8476, "step": 24915 }, { "epoch": 2.651202383485848, "grad_norm": 13.34375, "learning_rate": 9.13246162592382e-06, "loss": 0.8634, "step": 24916 }, { "epoch": 2.6513087891040645, "grad_norm": 14.59375, "learning_rate": 9.132426094371802e-06, "loss": 0.8309, "step": 24917 }, { "epoch": 2.6514151947222815, "grad_norm": 12.84375, "learning_rate": 9.132390562819785e-06, "loss": 0.9697, "step": 24918 }, { "epoch": 2.651521600340498, "grad_norm": 16.0625, "learning_rate": 9.132355031267767e-06, "loss": 1.0262, "step": 24919 }, { "epoch": 2.6516280059587145, "grad_norm": 15.890625, "learning_rate": 9.132319499715748e-06, "loss": 0.933, "step": 24920 }, { "epoch": 2.651734411576931, "grad_norm": 14.71875, "learning_rate": 9.13228396816373e-06, "loss": 0.9322, "step": 24921 }, { "epoch": 2.651840817195148, "grad_norm": 10.8515625, "learning_rate": 9.132248436611711e-06, "loss": 0.8904, "step": 24922 }, { "epoch": 2.6519472228133645, "grad_norm": 13.6328125, "learning_rate": 9.132212905059695e-06, "loss": 0.8977, "step": 24923 }, { "epoch": 2.652053628431581, "grad_norm": 14.421875, "learning_rate": 9.132177373507676e-06, "loss": 0.9353, "step": 24924 }, { "epoch": 2.652160034049798, "grad_norm": 16.78125, "learning_rate": 9.132141841955658e-06, "loss": 0.9364, "step": 24925 }, { "epoch": 2.6522664396680145, "grad_norm": 16.09375, "learning_rate": 9.13210631040364e-06, "loss": 0.8311, "step": 24926 }, { "epoch": 2.652372845286231, "grad_norm": 10.53125, "learning_rate": 9.132070778851621e-06, "loss": 0.7648, "step": 24927 }, { "epoch": 2.652479250904448, "grad_norm": 11.5, "learning_rate": 9.132035247299603e-06, "loss": 0.9428, "step": 24928 }, { "epoch": 2.6525856565226644, "grad_norm": 13.34375, "learning_rate": 9.131999715747584e-06, "loss": 1.0473, "step": 24929 }, { "epoch": 2.652692062140881, "grad_norm": 17.109375, "learning_rate": 9.131964184195566e-06, "loss": 0.9446, "step": 24930 }, { "epoch": 2.652798467759098, "grad_norm": 13.1484375, "learning_rate": 9.131928652643548e-06, "loss": 0.9142, "step": 24931 }, { "epoch": 2.6529048733773144, "grad_norm": 13.6328125, "learning_rate": 9.13189312109153e-06, "loss": 1.0936, "step": 24932 }, { "epoch": 2.653011278995531, "grad_norm": 13.1953125, "learning_rate": 9.131857589539513e-06, "loss": 0.8334, "step": 24933 }, { "epoch": 2.653117684613748, "grad_norm": 9.0625, "learning_rate": 9.131822057987494e-06, "loss": 0.7888, "step": 24934 }, { "epoch": 2.6532240902319644, "grad_norm": 15.4609375, "learning_rate": 9.131786526435476e-06, "loss": 0.9711, "step": 24935 }, { "epoch": 2.653330495850181, "grad_norm": 19.78125, "learning_rate": 9.131750994883457e-06, "loss": 0.9641, "step": 24936 }, { "epoch": 2.6534369014683974, "grad_norm": 13.28125, "learning_rate": 9.131715463331439e-06, "loss": 0.8707, "step": 24937 }, { "epoch": 2.653543307086614, "grad_norm": 15.078125, "learning_rate": 9.13167993177942e-06, "loss": 1.01, "step": 24938 }, { "epoch": 2.653649712704831, "grad_norm": 17.265625, "learning_rate": 9.131644400227402e-06, "loss": 0.8713, "step": 24939 }, { "epoch": 2.6537561183230474, "grad_norm": 11.765625, "learning_rate": 9.131608868675384e-06, "loss": 0.8892, "step": 24940 }, { "epoch": 2.653862523941264, "grad_norm": 9.6796875, "learning_rate": 9.131573337123365e-06, "loss": 0.778, "step": 24941 }, { "epoch": 2.653968929559481, "grad_norm": 13.03125, "learning_rate": 9.131537805571347e-06, "loss": 0.8639, "step": 24942 }, { "epoch": 2.6540753351776973, "grad_norm": 15.1953125, "learning_rate": 9.13150227401933e-06, "loss": 0.9106, "step": 24943 }, { "epoch": 2.654181740795914, "grad_norm": 14.0859375, "learning_rate": 9.131466742467312e-06, "loss": 1.0384, "step": 24944 }, { "epoch": 2.654288146414131, "grad_norm": 16.6875, "learning_rate": 9.131431210915293e-06, "loss": 0.8628, "step": 24945 }, { "epoch": 2.6543945520323473, "grad_norm": 13.90625, "learning_rate": 9.131395679363275e-06, "loss": 1.0867, "step": 24946 }, { "epoch": 2.654500957650564, "grad_norm": 7.65234375, "learning_rate": 9.131360147811257e-06, "loss": 0.6352, "step": 24947 }, { "epoch": 2.6546073632687808, "grad_norm": 9.7734375, "learning_rate": 9.13132461625924e-06, "loss": 0.8037, "step": 24948 }, { "epoch": 2.6547137688869973, "grad_norm": 15.15625, "learning_rate": 9.131289084707222e-06, "loss": 0.8505, "step": 24949 }, { "epoch": 2.654820174505214, "grad_norm": 11.3515625, "learning_rate": 9.131253553155203e-06, "loss": 0.9558, "step": 24950 }, { "epoch": 2.6549265801234307, "grad_norm": 12.0078125, "learning_rate": 9.131218021603185e-06, "loss": 0.8314, "step": 24951 }, { "epoch": 2.6550329857416473, "grad_norm": 14.0546875, "learning_rate": 9.131182490051166e-06, "loss": 0.9595, "step": 24952 }, { "epoch": 2.6551393913598638, "grad_norm": 10.796875, "learning_rate": 9.131146958499148e-06, "loss": 0.8632, "step": 24953 }, { "epoch": 2.6552457969780807, "grad_norm": 12.8671875, "learning_rate": 9.13111142694713e-06, "loss": 0.7823, "step": 24954 }, { "epoch": 2.6553522025962972, "grad_norm": 11.6875, "learning_rate": 9.131075895395111e-06, "loss": 1.0137, "step": 24955 }, { "epoch": 2.6554586082145137, "grad_norm": 14.0546875, "learning_rate": 9.131040363843093e-06, "loss": 1.0587, "step": 24956 }, { "epoch": 2.6555650138327302, "grad_norm": 11.2109375, "learning_rate": 9.131004832291074e-06, "loss": 0.8691, "step": 24957 }, { "epoch": 2.6556714194509468, "grad_norm": 15.2421875, "learning_rate": 9.130969300739056e-06, "loss": 1.0408, "step": 24958 }, { "epoch": 2.6557778250691637, "grad_norm": 19.203125, "learning_rate": 9.13093376918704e-06, "loss": 1.2504, "step": 24959 }, { "epoch": 2.65588423068738, "grad_norm": 15.046875, "learning_rate": 9.130898237635021e-06, "loss": 0.9404, "step": 24960 }, { "epoch": 2.6559906363055967, "grad_norm": 17.15625, "learning_rate": 9.130862706083003e-06, "loss": 1.0833, "step": 24961 }, { "epoch": 2.6560970419238137, "grad_norm": 11.734375, "learning_rate": 9.130827174530984e-06, "loss": 0.8207, "step": 24962 }, { "epoch": 2.65620344754203, "grad_norm": 10.15625, "learning_rate": 9.130791642978966e-06, "loss": 0.7969, "step": 24963 }, { "epoch": 2.6563098531602467, "grad_norm": 11.6796875, "learning_rate": 9.130756111426949e-06, "loss": 0.8456, "step": 24964 }, { "epoch": 2.6564162587784637, "grad_norm": 16.46875, "learning_rate": 9.13072057987493e-06, "loss": 0.7773, "step": 24965 }, { "epoch": 2.65652266439668, "grad_norm": 15.0625, "learning_rate": 9.13068504832291e-06, "loss": 0.8868, "step": 24966 }, { "epoch": 2.6566290700148967, "grad_norm": 15.2578125, "learning_rate": 9.130649516770892e-06, "loss": 1.0966, "step": 24967 }, { "epoch": 2.6567354756331136, "grad_norm": 10.6015625, "learning_rate": 9.130613985218874e-06, "loss": 0.746, "step": 24968 }, { "epoch": 2.65684188125133, "grad_norm": 11.1953125, "learning_rate": 9.130578453666857e-06, "loss": 0.8412, "step": 24969 }, { "epoch": 2.6569482868695466, "grad_norm": 14.890625, "learning_rate": 9.130542922114839e-06, "loss": 0.9012, "step": 24970 }, { "epoch": 2.6570546924877636, "grad_norm": 15.65625, "learning_rate": 9.13050739056282e-06, "loss": 0.8706, "step": 24971 }, { "epoch": 2.65716109810598, "grad_norm": 12.4375, "learning_rate": 9.130471859010802e-06, "loss": 0.8406, "step": 24972 }, { "epoch": 2.6572675037241966, "grad_norm": 11.3046875, "learning_rate": 9.130436327458784e-06, "loss": 0.8635, "step": 24973 }, { "epoch": 2.657373909342413, "grad_norm": 10.3828125, "learning_rate": 9.130400795906767e-06, "loss": 1.0278, "step": 24974 }, { "epoch": 2.65748031496063, "grad_norm": 10.5, "learning_rate": 9.130365264354748e-06, "loss": 0.7337, "step": 24975 }, { "epoch": 2.6575867205788466, "grad_norm": 12.265625, "learning_rate": 9.13032973280273e-06, "loss": 0.9824, "step": 24976 }, { "epoch": 2.657693126197063, "grad_norm": 12.625, "learning_rate": 9.130294201250712e-06, "loss": 0.954, "step": 24977 }, { "epoch": 2.6577995318152796, "grad_norm": 12.2265625, "learning_rate": 9.130258669698693e-06, "loss": 1.0649, "step": 24978 }, { "epoch": 2.6579059374334966, "grad_norm": 14.7734375, "learning_rate": 9.130223138146675e-06, "loss": 1.0302, "step": 24979 }, { "epoch": 2.658012343051713, "grad_norm": 10.984375, "learning_rate": 9.130187606594657e-06, "loss": 0.8868, "step": 24980 }, { "epoch": 2.6581187486699296, "grad_norm": 10.6796875, "learning_rate": 9.130152075042638e-06, "loss": 0.8976, "step": 24981 }, { "epoch": 2.6582251542881465, "grad_norm": 9.2109375, "learning_rate": 9.13011654349062e-06, "loss": 0.9541, "step": 24982 }, { "epoch": 2.658331559906363, "grad_norm": 15.9765625, "learning_rate": 9.130081011938601e-06, "loss": 1.2042, "step": 24983 }, { "epoch": 2.6584379655245796, "grad_norm": 15.1875, "learning_rate": 9.130045480386585e-06, "loss": 1.0411, "step": 24984 }, { "epoch": 2.6585443711427965, "grad_norm": 10.7734375, "learning_rate": 9.130009948834566e-06, "loss": 0.7619, "step": 24985 }, { "epoch": 2.658650776761013, "grad_norm": 10.0078125, "learning_rate": 9.129974417282548e-06, "loss": 0.8118, "step": 24986 }, { "epoch": 2.6587571823792295, "grad_norm": 14.265625, "learning_rate": 9.12993888573053e-06, "loss": 0.93, "step": 24987 }, { "epoch": 2.6588635879974465, "grad_norm": 14.90625, "learning_rate": 9.129903354178511e-06, "loss": 0.9771, "step": 24988 }, { "epoch": 2.658969993615663, "grad_norm": 16.515625, "learning_rate": 9.129867822626494e-06, "loss": 0.7534, "step": 24989 }, { "epoch": 2.6590763992338795, "grad_norm": 12.1953125, "learning_rate": 9.129832291074476e-06, "loss": 0.9891, "step": 24990 }, { "epoch": 2.6591828048520965, "grad_norm": 13.5703125, "learning_rate": 9.129796759522458e-06, "loss": 0.8659, "step": 24991 }, { "epoch": 2.659289210470313, "grad_norm": 13.9765625, "learning_rate": 9.12976122797044e-06, "loss": 1.1808, "step": 24992 }, { "epoch": 2.6593956160885295, "grad_norm": 13.2109375, "learning_rate": 9.12972569641842e-06, "loss": 1.021, "step": 24993 }, { "epoch": 2.659502021706746, "grad_norm": 13.65625, "learning_rate": 9.129690164866402e-06, "loss": 0.8919, "step": 24994 }, { "epoch": 2.6596084273249625, "grad_norm": 12.5234375, "learning_rate": 9.129654633314384e-06, "loss": 0.9296, "step": 24995 }, { "epoch": 2.6597148329431795, "grad_norm": 14.0, "learning_rate": 9.129619101762366e-06, "loss": 0.8063, "step": 24996 }, { "epoch": 2.659821238561396, "grad_norm": 12.734375, "learning_rate": 9.129583570210347e-06, "loss": 0.973, "step": 24997 }, { "epoch": 2.6599276441796125, "grad_norm": 10.8828125, "learning_rate": 9.129548038658329e-06, "loss": 0.9291, "step": 24998 }, { "epoch": 2.6600340497978294, "grad_norm": 15.8671875, "learning_rate": 9.12951250710631e-06, "loss": 1.1238, "step": 24999 }, { "epoch": 2.660140455416046, "grad_norm": 12.5234375, "learning_rate": 9.129476975554294e-06, "loss": 0.9256, "step": 25000 }, { "epoch": 2.6602468610342624, "grad_norm": 9.0390625, "learning_rate": 9.129441444002275e-06, "loss": 0.6666, "step": 25001 }, { "epoch": 2.6603532666524794, "grad_norm": 13.0390625, "learning_rate": 9.129405912450257e-06, "loss": 0.9196, "step": 25002 }, { "epoch": 2.660459672270696, "grad_norm": 13.0, "learning_rate": 9.129370380898239e-06, "loss": 0.9571, "step": 25003 }, { "epoch": 2.6605660778889124, "grad_norm": 12.3515625, "learning_rate": 9.12933484934622e-06, "loss": 0.8797, "step": 25004 }, { "epoch": 2.6606724835071294, "grad_norm": 15.9140625, "learning_rate": 9.129299317794202e-06, "loss": 0.6745, "step": 25005 }, { "epoch": 2.660778889125346, "grad_norm": 12.109375, "learning_rate": 9.129263786242183e-06, "loss": 0.8374, "step": 25006 }, { "epoch": 2.6608852947435624, "grad_norm": 12.0625, "learning_rate": 9.129228254690165e-06, "loss": 0.6505, "step": 25007 }, { "epoch": 2.6609917003617793, "grad_norm": 16.515625, "learning_rate": 9.129192723138147e-06, "loss": 0.7839, "step": 25008 }, { "epoch": 2.661098105979996, "grad_norm": 13.1875, "learning_rate": 9.129157191586128e-06, "loss": 0.8886, "step": 25009 }, { "epoch": 2.6612045115982124, "grad_norm": 11.40625, "learning_rate": 9.129121660034111e-06, "loss": 1.0355, "step": 25010 }, { "epoch": 2.661310917216429, "grad_norm": 12.84375, "learning_rate": 9.129086128482093e-06, "loss": 0.9643, "step": 25011 }, { "epoch": 2.661417322834646, "grad_norm": 11.46875, "learning_rate": 9.129050596930075e-06, "loss": 0.9628, "step": 25012 }, { "epoch": 2.6615237284528623, "grad_norm": 13.5546875, "learning_rate": 9.129015065378056e-06, "loss": 0.9846, "step": 25013 }, { "epoch": 2.661630134071079, "grad_norm": 61.4375, "learning_rate": 9.128979533826038e-06, "loss": 0.9296, "step": 25014 }, { "epoch": 2.6617365396892954, "grad_norm": 12.2421875, "learning_rate": 9.128944002274021e-06, "loss": 0.8781, "step": 25015 }, { "epoch": 2.6618429453075123, "grad_norm": 10.375, "learning_rate": 9.128908470722003e-06, "loss": 0.805, "step": 25016 }, { "epoch": 2.661949350925729, "grad_norm": 15.9453125, "learning_rate": 9.128872939169984e-06, "loss": 1.0464, "step": 25017 }, { "epoch": 2.6620557565439453, "grad_norm": 14.453125, "learning_rate": 9.128837407617966e-06, "loss": 0.9765, "step": 25018 }, { "epoch": 2.6621621621621623, "grad_norm": 12.859375, "learning_rate": 9.128801876065948e-06, "loss": 1.0095, "step": 25019 }, { "epoch": 2.662268567780379, "grad_norm": 15.90625, "learning_rate": 9.12876634451393e-06, "loss": 1.0299, "step": 25020 }, { "epoch": 2.6623749733985953, "grad_norm": 12.2421875, "learning_rate": 9.128730812961911e-06, "loss": 0.7765, "step": 25021 }, { "epoch": 2.6624813790168123, "grad_norm": 16.375, "learning_rate": 9.128695281409892e-06, "loss": 0.9469, "step": 25022 }, { "epoch": 2.6625877846350288, "grad_norm": 15.40625, "learning_rate": 9.128659749857874e-06, "loss": 0.9214, "step": 25023 }, { "epoch": 2.6626941902532453, "grad_norm": 14.921875, "learning_rate": 9.128624218305856e-06, "loss": 0.9231, "step": 25024 }, { "epoch": 2.6628005958714622, "grad_norm": 12.8125, "learning_rate": 9.128588686753839e-06, "loss": 0.9632, "step": 25025 }, { "epoch": 2.6629070014896787, "grad_norm": 20.875, "learning_rate": 9.12855315520182e-06, "loss": 1.026, "step": 25026 }, { "epoch": 2.6630134071078952, "grad_norm": 13.96875, "learning_rate": 9.128517623649802e-06, "loss": 1.0136, "step": 25027 }, { "epoch": 2.663119812726112, "grad_norm": 14.2421875, "learning_rate": 9.128482092097784e-06, "loss": 0.9196, "step": 25028 }, { "epoch": 2.6632262183443287, "grad_norm": 13.015625, "learning_rate": 9.128446560545765e-06, "loss": 0.7725, "step": 25029 }, { "epoch": 2.663332623962545, "grad_norm": 13.71875, "learning_rate": 9.128411028993747e-06, "loss": 1.036, "step": 25030 }, { "epoch": 2.6634390295807617, "grad_norm": 12.4609375, "learning_rate": 9.128375497441729e-06, "loss": 1.0624, "step": 25031 }, { "epoch": 2.6635454351989782, "grad_norm": 12.78125, "learning_rate": 9.12833996588971e-06, "loss": 0.93, "step": 25032 }, { "epoch": 2.663651840817195, "grad_norm": 10.9296875, "learning_rate": 9.128304434337692e-06, "loss": 0.755, "step": 25033 }, { "epoch": 2.6637582464354117, "grad_norm": 12.328125, "learning_rate": 9.128268902785673e-06, "loss": 1.0708, "step": 25034 }, { "epoch": 2.663864652053628, "grad_norm": 14.1484375, "learning_rate": 9.128233371233655e-06, "loss": 0.9223, "step": 25035 }, { "epoch": 2.663971057671845, "grad_norm": 12.8125, "learning_rate": 9.128197839681638e-06, "loss": 1.06, "step": 25036 }, { "epoch": 2.6640774632900617, "grad_norm": 14.2421875, "learning_rate": 9.12816230812962e-06, "loss": 1.0964, "step": 25037 }, { "epoch": 2.664183868908278, "grad_norm": 12.1875, "learning_rate": 9.128126776577602e-06, "loss": 0.9398, "step": 25038 }, { "epoch": 2.664290274526495, "grad_norm": 17.234375, "learning_rate": 9.128091245025583e-06, "loss": 1.0025, "step": 25039 }, { "epoch": 2.6643966801447116, "grad_norm": 15.7734375, "learning_rate": 9.128055713473565e-06, "loss": 0.9422, "step": 25040 }, { "epoch": 2.664503085762928, "grad_norm": 15.7109375, "learning_rate": 9.128020181921548e-06, "loss": 1.0285, "step": 25041 }, { "epoch": 2.664609491381145, "grad_norm": 12.40625, "learning_rate": 9.12798465036953e-06, "loss": 0.8056, "step": 25042 }, { "epoch": 2.6647158969993616, "grad_norm": 11.0703125, "learning_rate": 9.127949118817511e-06, "loss": 0.8712, "step": 25043 }, { "epoch": 2.664822302617578, "grad_norm": 9.75, "learning_rate": 9.127913587265493e-06, "loss": 0.8398, "step": 25044 }, { "epoch": 2.664928708235795, "grad_norm": 13.1328125, "learning_rate": 9.127878055713474e-06, "loss": 0.8595, "step": 25045 }, { "epoch": 2.6650351138540116, "grad_norm": 14.1640625, "learning_rate": 9.127842524161456e-06, "loss": 0.9379, "step": 25046 }, { "epoch": 2.665141519472228, "grad_norm": 14.7578125, "learning_rate": 9.127806992609438e-06, "loss": 1.1405, "step": 25047 }, { "epoch": 2.6652479250904446, "grad_norm": 11.828125, "learning_rate": 9.12777146105742e-06, "loss": 0.8203, "step": 25048 }, { "epoch": 2.6653543307086616, "grad_norm": 13.8515625, "learning_rate": 9.127735929505401e-06, "loss": 1.0145, "step": 25049 }, { "epoch": 2.665460736326878, "grad_norm": 13.5078125, "learning_rate": 9.127700397953383e-06, "loss": 0.7947, "step": 25050 }, { "epoch": 2.6655671419450946, "grad_norm": 12.7109375, "learning_rate": 9.127664866401366e-06, "loss": 0.9208, "step": 25051 }, { "epoch": 2.665673547563311, "grad_norm": 14.125, "learning_rate": 9.127629334849347e-06, "loss": 0.9066, "step": 25052 }, { "epoch": 2.665779953181528, "grad_norm": 15.5078125, "learning_rate": 9.127593803297329e-06, "loss": 0.9654, "step": 25053 }, { "epoch": 2.6658863587997446, "grad_norm": 11.0078125, "learning_rate": 9.12755827174531e-06, "loss": 0.7704, "step": 25054 }, { "epoch": 2.665992764417961, "grad_norm": 12.7578125, "learning_rate": 9.127522740193292e-06, "loss": 0.8732, "step": 25055 }, { "epoch": 2.666099170036178, "grad_norm": 14.609375, "learning_rate": 9.127487208641274e-06, "loss": 0.7663, "step": 25056 }, { "epoch": 2.6662055756543945, "grad_norm": 15.15625, "learning_rate": 9.127451677089255e-06, "loss": 0.9056, "step": 25057 }, { "epoch": 2.666311981272611, "grad_norm": 12.59375, "learning_rate": 9.127416145537237e-06, "loss": 0.8844, "step": 25058 }, { "epoch": 2.666418386890828, "grad_norm": 11.96875, "learning_rate": 9.127380613985219e-06, "loss": 0.8784, "step": 25059 }, { "epoch": 2.6665247925090445, "grad_norm": 21.890625, "learning_rate": 9.1273450824332e-06, "loss": 1.1792, "step": 25060 }, { "epoch": 2.666631198127261, "grad_norm": 16.828125, "learning_rate": 9.127309550881184e-06, "loss": 0.9698, "step": 25061 }, { "epoch": 2.666737603745478, "grad_norm": 11.1171875, "learning_rate": 9.127274019329165e-06, "loss": 0.9662, "step": 25062 }, { "epoch": 2.6668440093636945, "grad_norm": 11.6796875, "learning_rate": 9.127238487777147e-06, "loss": 0.9262, "step": 25063 }, { "epoch": 2.666950414981911, "grad_norm": 11.109375, "learning_rate": 9.127202956225128e-06, "loss": 0.8617, "step": 25064 }, { "epoch": 2.667056820600128, "grad_norm": 11.921875, "learning_rate": 9.12716742467311e-06, "loss": 0.8468, "step": 25065 }, { "epoch": 2.6671632262183445, "grad_norm": 15.46875, "learning_rate": 9.127131893121093e-06, "loss": 0.9134, "step": 25066 }, { "epoch": 2.667269631836561, "grad_norm": 13.9609375, "learning_rate": 9.127096361569075e-06, "loss": 0.7598, "step": 25067 }, { "epoch": 2.6673760374547775, "grad_norm": 19.84375, "learning_rate": 9.127060830017057e-06, "loss": 1.1994, "step": 25068 }, { "epoch": 2.667482443072994, "grad_norm": 14.3671875, "learning_rate": 9.127025298465038e-06, "loss": 0.7807, "step": 25069 }, { "epoch": 2.667588848691211, "grad_norm": 12.125, "learning_rate": 9.12698976691302e-06, "loss": 0.8532, "step": 25070 }, { "epoch": 2.6676952543094274, "grad_norm": 12.6875, "learning_rate": 9.126954235361001e-06, "loss": 0.7826, "step": 25071 }, { "epoch": 2.667801659927644, "grad_norm": 16.328125, "learning_rate": 9.126918703808983e-06, "loss": 0.9233, "step": 25072 }, { "epoch": 2.667908065545861, "grad_norm": 11.3125, "learning_rate": 9.126883172256965e-06, "loss": 0.8408, "step": 25073 }, { "epoch": 2.6680144711640774, "grad_norm": 13.84375, "learning_rate": 9.126847640704946e-06, "loss": 0.8517, "step": 25074 }, { "epoch": 2.668120876782294, "grad_norm": 14.640625, "learning_rate": 9.126812109152928e-06, "loss": 0.829, "step": 25075 }, { "epoch": 2.668227282400511, "grad_norm": 14.8125, "learning_rate": 9.12677657760091e-06, "loss": 1.0286, "step": 25076 }, { "epoch": 2.6683336880187274, "grad_norm": 11.7421875, "learning_rate": 9.126741046048893e-06, "loss": 0.8673, "step": 25077 }, { "epoch": 2.668440093636944, "grad_norm": 14.03125, "learning_rate": 9.126705514496874e-06, "loss": 1.069, "step": 25078 }, { "epoch": 2.668546499255161, "grad_norm": 15.4609375, "learning_rate": 9.126669982944856e-06, "loss": 0.9804, "step": 25079 }, { "epoch": 2.6686529048733774, "grad_norm": 11.296875, "learning_rate": 9.126634451392838e-06, "loss": 0.878, "step": 25080 }, { "epoch": 2.668759310491594, "grad_norm": 16.4375, "learning_rate": 9.126598919840819e-06, "loss": 0.9527, "step": 25081 }, { "epoch": 2.668865716109811, "grad_norm": 18.34375, "learning_rate": 9.126563388288802e-06, "loss": 0.932, "step": 25082 }, { "epoch": 2.6689721217280273, "grad_norm": 14.875, "learning_rate": 9.126527856736784e-06, "loss": 1.0217, "step": 25083 }, { "epoch": 2.669078527346244, "grad_norm": 12.4453125, "learning_rate": 9.126492325184766e-06, "loss": 1.0155, "step": 25084 }, { "epoch": 2.669184932964461, "grad_norm": 18.65625, "learning_rate": 9.126456793632747e-06, "loss": 0.9885, "step": 25085 }, { "epoch": 2.6692913385826773, "grad_norm": 11.2265625, "learning_rate": 9.126421262080727e-06, "loss": 0.9594, "step": 25086 }, { "epoch": 2.669397744200894, "grad_norm": 12.765625, "learning_rate": 9.12638573052871e-06, "loss": 0.7975, "step": 25087 }, { "epoch": 2.6695041498191103, "grad_norm": 14.3671875, "learning_rate": 9.126350198976692e-06, "loss": 1.0283, "step": 25088 }, { "epoch": 2.669610555437327, "grad_norm": 12.921875, "learning_rate": 9.126314667424674e-06, "loss": 1.0655, "step": 25089 }, { "epoch": 2.669716961055544, "grad_norm": 13.265625, "learning_rate": 9.126279135872655e-06, "loss": 0.8604, "step": 25090 }, { "epoch": 2.6698233666737603, "grad_norm": 10.6484375, "learning_rate": 9.126243604320637e-06, "loss": 0.8304, "step": 25091 }, { "epoch": 2.669929772291977, "grad_norm": 18.1875, "learning_rate": 9.12620807276862e-06, "loss": 0.947, "step": 25092 }, { "epoch": 2.6700361779101938, "grad_norm": 12.140625, "learning_rate": 9.126172541216602e-06, "loss": 1.0051, "step": 25093 }, { "epoch": 2.6701425835284103, "grad_norm": 10.0859375, "learning_rate": 9.126137009664583e-06, "loss": 0.7768, "step": 25094 }, { "epoch": 2.670248989146627, "grad_norm": 14.3203125, "learning_rate": 9.126101478112565e-06, "loss": 1.1594, "step": 25095 }, { "epoch": 2.6703553947648437, "grad_norm": 11.25, "learning_rate": 9.126065946560547e-06, "loss": 0.6717, "step": 25096 }, { "epoch": 2.6704618003830602, "grad_norm": 12.9296875, "learning_rate": 9.126030415008528e-06, "loss": 0.7632, "step": 25097 }, { "epoch": 2.6705682060012768, "grad_norm": 14.1953125, "learning_rate": 9.12599488345651e-06, "loss": 0.9277, "step": 25098 }, { "epoch": 2.6706746116194937, "grad_norm": 11.328125, "learning_rate": 9.125959351904491e-06, "loss": 0.6856, "step": 25099 }, { "epoch": 2.67078101723771, "grad_norm": 11.2421875, "learning_rate": 9.125923820352473e-06, "loss": 0.9694, "step": 25100 }, { "epoch": 2.6708874228559267, "grad_norm": 12.2421875, "learning_rate": 9.125888288800455e-06, "loss": 0.6682, "step": 25101 }, { "epoch": 2.6709938284741437, "grad_norm": 13.40625, "learning_rate": 9.125852757248438e-06, "loss": 0.7786, "step": 25102 }, { "epoch": 2.67110023409236, "grad_norm": 14.921875, "learning_rate": 9.12581722569642e-06, "loss": 0.9367, "step": 25103 }, { "epoch": 2.6712066397105767, "grad_norm": 13.7734375, "learning_rate": 9.125781694144401e-06, "loss": 0.8071, "step": 25104 }, { "epoch": 2.671313045328793, "grad_norm": 22.9375, "learning_rate": 9.125746162592383e-06, "loss": 0.7155, "step": 25105 }, { "epoch": 2.67141945094701, "grad_norm": 13.5234375, "learning_rate": 9.125710631040364e-06, "loss": 0.8058, "step": 25106 }, { "epoch": 2.6715258565652267, "grad_norm": 13.0390625, "learning_rate": 9.125675099488346e-06, "loss": 0.93, "step": 25107 }, { "epoch": 2.671632262183443, "grad_norm": 16.59375, "learning_rate": 9.12563956793633e-06, "loss": 1.0174, "step": 25108 }, { "epoch": 2.6717386678016597, "grad_norm": 15.25, "learning_rate": 9.125604036384311e-06, "loss": 0.9993, "step": 25109 }, { "epoch": 2.6718450734198766, "grad_norm": 20.71875, "learning_rate": 9.125568504832292e-06, "loss": 1.0915, "step": 25110 }, { "epoch": 2.671951479038093, "grad_norm": 11.7109375, "learning_rate": 9.125532973280274e-06, "loss": 0.8419, "step": 25111 }, { "epoch": 2.6720578846563097, "grad_norm": 13.6953125, "learning_rate": 9.125497441728256e-06, "loss": 0.864, "step": 25112 }, { "epoch": 2.6721642902745266, "grad_norm": 10.1015625, "learning_rate": 9.125461910176237e-06, "loss": 0.7878, "step": 25113 }, { "epoch": 2.672270695892743, "grad_norm": 11.5390625, "learning_rate": 9.125426378624219e-06, "loss": 0.9989, "step": 25114 }, { "epoch": 2.6723771015109596, "grad_norm": 14.8984375, "learning_rate": 9.1253908470722e-06, "loss": 0.8158, "step": 25115 }, { "epoch": 2.6724835071291766, "grad_norm": 12.203125, "learning_rate": 9.125355315520182e-06, "loss": 0.9853, "step": 25116 }, { "epoch": 2.672589912747393, "grad_norm": 11.71875, "learning_rate": 9.125319783968164e-06, "loss": 0.8692, "step": 25117 }, { "epoch": 2.6726963183656096, "grad_norm": 14.0078125, "learning_rate": 9.125284252416147e-06, "loss": 0.8935, "step": 25118 }, { "epoch": 2.6728027239838266, "grad_norm": 11.0, "learning_rate": 9.125248720864129e-06, "loss": 0.8468, "step": 25119 }, { "epoch": 2.672909129602043, "grad_norm": 14.6484375, "learning_rate": 9.12521318931211e-06, "loss": 0.8414, "step": 25120 }, { "epoch": 2.6730155352202596, "grad_norm": 11.9375, "learning_rate": 9.125177657760092e-06, "loss": 0.8342, "step": 25121 }, { "epoch": 2.6731219408384765, "grad_norm": 11.1171875, "learning_rate": 9.125142126208073e-06, "loss": 0.9311, "step": 25122 }, { "epoch": 2.673228346456693, "grad_norm": 12.015625, "learning_rate": 9.125106594656055e-06, "loss": 0.8891, "step": 25123 }, { "epoch": 2.6733347520749096, "grad_norm": 13.953125, "learning_rate": 9.125071063104037e-06, "loss": 0.9146, "step": 25124 }, { "epoch": 2.673441157693126, "grad_norm": 14.84375, "learning_rate": 9.125035531552018e-06, "loss": 1.079, "step": 25125 }, { "epoch": 2.6735475633113426, "grad_norm": 12.6875, "learning_rate": 9.125e-06, "loss": 0.8783, "step": 25126 }, { "epoch": 2.6736539689295595, "grad_norm": 10.5078125, "learning_rate": 9.124964468447981e-06, "loss": 0.826, "step": 25127 }, { "epoch": 2.673760374547776, "grad_norm": 14.71875, "learning_rate": 9.124928936895965e-06, "loss": 0.7981, "step": 25128 }, { "epoch": 2.6738667801659926, "grad_norm": 13.1484375, "learning_rate": 9.124893405343946e-06, "loss": 0.7834, "step": 25129 }, { "epoch": 2.6739731857842095, "grad_norm": 10.921875, "learning_rate": 9.124857873791928e-06, "loss": 0.7274, "step": 25130 }, { "epoch": 2.674079591402426, "grad_norm": 13.4140625, "learning_rate": 9.12482234223991e-06, "loss": 0.9427, "step": 25131 }, { "epoch": 2.6741859970206425, "grad_norm": 11.7109375, "learning_rate": 9.124786810687891e-06, "loss": 0.9369, "step": 25132 }, { "epoch": 2.6742924026388595, "grad_norm": 13.4296875, "learning_rate": 9.124751279135875e-06, "loss": 0.9753, "step": 25133 }, { "epoch": 2.674398808257076, "grad_norm": 10.796875, "learning_rate": 9.124715747583856e-06, "loss": 0.8536, "step": 25134 }, { "epoch": 2.6745052138752925, "grad_norm": 13.9921875, "learning_rate": 9.124680216031838e-06, "loss": 0.9759, "step": 25135 }, { "epoch": 2.6746116194935095, "grad_norm": 13.3203125, "learning_rate": 9.12464468447982e-06, "loss": 0.7584, "step": 25136 }, { "epoch": 2.674718025111726, "grad_norm": 17.765625, "learning_rate": 9.124609152927801e-06, "loss": 0.9092, "step": 25137 }, { "epoch": 2.6748244307299425, "grad_norm": 13.6796875, "learning_rate": 9.124573621375783e-06, "loss": 0.6688, "step": 25138 }, { "epoch": 2.6749308363481594, "grad_norm": 15.8203125, "learning_rate": 9.124538089823764e-06, "loss": 0.8388, "step": 25139 }, { "epoch": 2.675037241966376, "grad_norm": 13.40625, "learning_rate": 9.124502558271746e-06, "loss": 0.7617, "step": 25140 }, { "epoch": 2.6751436475845924, "grad_norm": 13.109375, "learning_rate": 9.124467026719727e-06, "loss": 0.8789, "step": 25141 }, { "epoch": 2.675250053202809, "grad_norm": 10.40625, "learning_rate": 9.124431495167709e-06, "loss": 0.8219, "step": 25142 }, { "epoch": 2.675356458821026, "grad_norm": 12.984375, "learning_rate": 9.124395963615692e-06, "loss": 0.9483, "step": 25143 }, { "epoch": 2.6754628644392424, "grad_norm": 20.125, "learning_rate": 9.124360432063674e-06, "loss": 0.8829, "step": 25144 }, { "epoch": 2.675569270057459, "grad_norm": 15.625, "learning_rate": 9.124324900511655e-06, "loss": 0.7812, "step": 25145 }, { "epoch": 2.6756756756756754, "grad_norm": 13.375, "learning_rate": 9.124289368959637e-06, "loss": 0.9756, "step": 25146 }, { "epoch": 2.6757820812938924, "grad_norm": 16.421875, "learning_rate": 9.124253837407619e-06, "loss": 0.9825, "step": 25147 }, { "epoch": 2.675888486912109, "grad_norm": 10.921875, "learning_rate": 9.1242183058556e-06, "loss": 0.6323, "step": 25148 }, { "epoch": 2.6759948925303254, "grad_norm": 12.875, "learning_rate": 9.124182774303582e-06, "loss": 0.9395, "step": 25149 }, { "epoch": 2.6761012981485424, "grad_norm": 13.7265625, "learning_rate": 9.124147242751564e-06, "loss": 0.8864, "step": 25150 }, { "epoch": 2.676207703766759, "grad_norm": 15.96875, "learning_rate": 9.124111711199545e-06, "loss": 0.8076, "step": 25151 }, { "epoch": 2.6763141093849754, "grad_norm": 12.25, "learning_rate": 9.124076179647527e-06, "loss": 0.8482, "step": 25152 }, { "epoch": 2.6764205150031923, "grad_norm": 17.578125, "learning_rate": 9.124040648095508e-06, "loss": 0.888, "step": 25153 }, { "epoch": 2.676526920621409, "grad_norm": 12.3203125, "learning_rate": 9.124005116543492e-06, "loss": 0.8508, "step": 25154 }, { "epoch": 2.6766333262396254, "grad_norm": 13.1484375, "learning_rate": 9.123969584991473e-06, "loss": 1.0348, "step": 25155 }, { "epoch": 2.6767397318578423, "grad_norm": 12.140625, "learning_rate": 9.123934053439455e-06, "loss": 0.9648, "step": 25156 }, { "epoch": 2.676846137476059, "grad_norm": 15.3359375, "learning_rate": 9.123898521887436e-06, "loss": 0.9669, "step": 25157 }, { "epoch": 2.6769525430942753, "grad_norm": 17.546875, "learning_rate": 9.123862990335418e-06, "loss": 1.0905, "step": 25158 }, { "epoch": 2.6770589487124923, "grad_norm": 12.1796875, "learning_rate": 9.123827458783401e-06, "loss": 0.8448, "step": 25159 }, { "epoch": 2.677165354330709, "grad_norm": 16.8125, "learning_rate": 9.123791927231383e-06, "loss": 1.0115, "step": 25160 }, { "epoch": 2.6772717599489253, "grad_norm": 10.3515625, "learning_rate": 9.123756395679365e-06, "loss": 0.8846, "step": 25161 }, { "epoch": 2.677378165567142, "grad_norm": 13.28125, "learning_rate": 9.123720864127346e-06, "loss": 0.9064, "step": 25162 }, { "epoch": 2.6774845711853583, "grad_norm": 12.765625, "learning_rate": 9.123685332575328e-06, "loss": 0.8916, "step": 25163 }, { "epoch": 2.6775909768035753, "grad_norm": 15.34375, "learning_rate": 9.12364980102331e-06, "loss": 0.9389, "step": 25164 }, { "epoch": 2.677697382421792, "grad_norm": 15.546875, "learning_rate": 9.123614269471291e-06, "loss": 0.8399, "step": 25165 }, { "epoch": 2.6778037880400083, "grad_norm": 14.71875, "learning_rate": 9.123578737919273e-06, "loss": 0.9682, "step": 25166 }, { "epoch": 2.6779101936582252, "grad_norm": 13.640625, "learning_rate": 9.123543206367254e-06, "loss": 0.8737, "step": 25167 }, { "epoch": 2.6780165992764418, "grad_norm": 15.703125, "learning_rate": 9.123507674815236e-06, "loss": 0.8816, "step": 25168 }, { "epoch": 2.6781230048946583, "grad_norm": 16.4375, "learning_rate": 9.123472143263219e-06, "loss": 1.2165, "step": 25169 }, { "epoch": 2.678229410512875, "grad_norm": 13.921875, "learning_rate": 9.1234366117112e-06, "loss": 1.0653, "step": 25170 }, { "epoch": 2.6783358161310917, "grad_norm": 14.296875, "learning_rate": 9.123401080159182e-06, "loss": 0.9933, "step": 25171 }, { "epoch": 2.6784422217493082, "grad_norm": 14.6796875, "learning_rate": 9.123365548607164e-06, "loss": 1.1691, "step": 25172 }, { "epoch": 2.678548627367525, "grad_norm": 12.9453125, "learning_rate": 9.123330017055146e-06, "loss": 0.8408, "step": 25173 }, { "epoch": 2.6786550329857417, "grad_norm": 10.953125, "learning_rate": 9.123294485503129e-06, "loss": 0.7691, "step": 25174 }, { "epoch": 2.678761438603958, "grad_norm": 12.75, "learning_rate": 9.123258953951109e-06, "loss": 0.9126, "step": 25175 }, { "epoch": 2.678867844222175, "grad_norm": 20.203125, "learning_rate": 9.12322342239909e-06, "loss": 1.124, "step": 25176 }, { "epoch": 2.6789742498403917, "grad_norm": 11.984375, "learning_rate": 9.123187890847072e-06, "loss": 0.9067, "step": 25177 }, { "epoch": 2.679080655458608, "grad_norm": 15.484375, "learning_rate": 9.123152359295054e-06, "loss": 0.9136, "step": 25178 }, { "epoch": 2.6791870610768247, "grad_norm": 15.84375, "learning_rate": 9.123116827743037e-06, "loss": 0.9088, "step": 25179 }, { "epoch": 2.6792934666950416, "grad_norm": 11.203125, "learning_rate": 9.123081296191018e-06, "loss": 0.8622, "step": 25180 }, { "epoch": 2.679399872313258, "grad_norm": 10.1953125, "learning_rate": 9.123045764639e-06, "loss": 0.8369, "step": 25181 }, { "epoch": 2.6795062779314747, "grad_norm": 17.296875, "learning_rate": 9.123010233086982e-06, "loss": 0.9916, "step": 25182 }, { "epoch": 2.679612683549691, "grad_norm": 13.9453125, "learning_rate": 9.122974701534963e-06, "loss": 1.0873, "step": 25183 }, { "epoch": 2.679719089167908, "grad_norm": 14.34375, "learning_rate": 9.122939169982947e-06, "loss": 1.0728, "step": 25184 }, { "epoch": 2.6798254947861246, "grad_norm": 17.46875, "learning_rate": 9.122903638430928e-06, "loss": 0.9962, "step": 25185 }, { "epoch": 2.679931900404341, "grad_norm": 19.375, "learning_rate": 9.12286810687891e-06, "loss": 0.8227, "step": 25186 }, { "epoch": 2.680038306022558, "grad_norm": 12.3515625, "learning_rate": 9.122832575326891e-06, "loss": 0.8356, "step": 25187 }, { "epoch": 2.6801447116407746, "grad_norm": 9.53125, "learning_rate": 9.122797043774873e-06, "loss": 0.6798, "step": 25188 }, { "epoch": 2.680251117258991, "grad_norm": 14.4765625, "learning_rate": 9.122761512222855e-06, "loss": 1.0826, "step": 25189 }, { "epoch": 2.680357522877208, "grad_norm": 11.1484375, "learning_rate": 9.122725980670836e-06, "loss": 0.9859, "step": 25190 }, { "epoch": 2.6804639284954246, "grad_norm": 14.0078125, "learning_rate": 9.122690449118818e-06, "loss": 0.9447, "step": 25191 }, { "epoch": 2.680570334113641, "grad_norm": 13.6015625, "learning_rate": 9.1226549175668e-06, "loss": 0.9251, "step": 25192 }, { "epoch": 2.680676739731858, "grad_norm": 10.15625, "learning_rate": 9.122619386014781e-06, "loss": 0.7347, "step": 25193 }, { "epoch": 2.6807831453500746, "grad_norm": 14.734375, "learning_rate": 9.122583854462763e-06, "loss": 0.8123, "step": 25194 }, { "epoch": 2.680889550968291, "grad_norm": 12.359375, "learning_rate": 9.122548322910746e-06, "loss": 0.9981, "step": 25195 }, { "epoch": 2.680995956586508, "grad_norm": 13.234375, "learning_rate": 9.122512791358728e-06, "loss": 0.9874, "step": 25196 }, { "epoch": 2.6811023622047245, "grad_norm": 14.5546875, "learning_rate": 9.12247725980671e-06, "loss": 0.6517, "step": 25197 }, { "epoch": 2.681208767822941, "grad_norm": 11.1640625, "learning_rate": 9.12244172825469e-06, "loss": 0.8016, "step": 25198 }, { "epoch": 2.6813151734411576, "grad_norm": 13.1953125, "learning_rate": 9.122406196702672e-06, "loss": 0.9521, "step": 25199 }, { "epoch": 2.681421579059374, "grad_norm": 14.625, "learning_rate": 9.122370665150656e-06, "loss": 0.995, "step": 25200 }, { "epoch": 2.681527984677591, "grad_norm": 10.8828125, "learning_rate": 9.122335133598637e-06, "loss": 0.8841, "step": 25201 }, { "epoch": 2.6816343902958075, "grad_norm": 13.546875, "learning_rate": 9.122299602046619e-06, "loss": 0.9515, "step": 25202 }, { "epoch": 2.681740795914024, "grad_norm": 11.296875, "learning_rate": 9.1222640704946e-06, "loss": 0.9117, "step": 25203 }, { "epoch": 2.681847201532241, "grad_norm": 17.640625, "learning_rate": 9.122228538942582e-06, "loss": 1.0544, "step": 25204 }, { "epoch": 2.6819536071504575, "grad_norm": 12.8984375, "learning_rate": 9.122193007390564e-06, "loss": 0.8388, "step": 25205 }, { "epoch": 2.682060012768674, "grad_norm": 10.609375, "learning_rate": 9.122157475838545e-06, "loss": 0.8531, "step": 25206 }, { "epoch": 2.682166418386891, "grad_norm": 15.4453125, "learning_rate": 9.122121944286527e-06, "loss": 0.8983, "step": 25207 }, { "epoch": 2.6822728240051075, "grad_norm": 11.9140625, "learning_rate": 9.122086412734509e-06, "loss": 0.9239, "step": 25208 }, { "epoch": 2.682379229623324, "grad_norm": 14.8203125, "learning_rate": 9.12205088118249e-06, "loss": 0.8037, "step": 25209 }, { "epoch": 2.682485635241541, "grad_norm": 15.0859375, "learning_rate": 9.122015349630473e-06, "loss": 0.9704, "step": 25210 }, { "epoch": 2.6825920408597574, "grad_norm": 11.234375, "learning_rate": 9.121979818078455e-06, "loss": 0.8646, "step": 25211 }, { "epoch": 2.682698446477974, "grad_norm": 18.359375, "learning_rate": 9.121944286526437e-06, "loss": 0.8914, "step": 25212 }, { "epoch": 2.682804852096191, "grad_norm": 19.109375, "learning_rate": 9.121908754974418e-06, "loss": 0.9917, "step": 25213 }, { "epoch": 2.6829112577144074, "grad_norm": 11.3671875, "learning_rate": 9.1218732234224e-06, "loss": 0.6927, "step": 25214 }, { "epoch": 2.683017663332624, "grad_norm": 11.8359375, "learning_rate": 9.121837691870382e-06, "loss": 0.946, "step": 25215 }, { "epoch": 2.683124068950841, "grad_norm": 12.96875, "learning_rate": 9.121802160318363e-06, "loss": 0.8771, "step": 25216 }, { "epoch": 2.6832304745690574, "grad_norm": 11.25, "learning_rate": 9.121766628766345e-06, "loss": 0.7576, "step": 25217 }, { "epoch": 2.683336880187274, "grad_norm": 17.234375, "learning_rate": 9.121731097214326e-06, "loss": 1.0353, "step": 25218 }, { "epoch": 2.6834432858054904, "grad_norm": 12.2734375, "learning_rate": 9.121695565662308e-06, "loss": 0.9535, "step": 25219 }, { "epoch": 2.683549691423707, "grad_norm": 10.7578125, "learning_rate": 9.121660034110291e-06, "loss": 0.7611, "step": 25220 }, { "epoch": 2.683656097041924, "grad_norm": 18.484375, "learning_rate": 9.121624502558273e-06, "loss": 1.0256, "step": 25221 }, { "epoch": 2.6837625026601404, "grad_norm": 12.546875, "learning_rate": 9.121588971006254e-06, "loss": 1.0007, "step": 25222 }, { "epoch": 2.683868908278357, "grad_norm": 13.625, "learning_rate": 9.121553439454236e-06, "loss": 0.9205, "step": 25223 }, { "epoch": 2.683975313896574, "grad_norm": 15.4140625, "learning_rate": 9.121517907902218e-06, "loss": 1.0536, "step": 25224 }, { "epoch": 2.6840817195147904, "grad_norm": 15.5546875, "learning_rate": 9.1214823763502e-06, "loss": 1.2743, "step": 25225 }, { "epoch": 2.684188125133007, "grad_norm": 14.6875, "learning_rate": 9.121446844798183e-06, "loss": 0.7983, "step": 25226 }, { "epoch": 2.684294530751224, "grad_norm": 14.9375, "learning_rate": 9.121411313246164e-06, "loss": 1.0448, "step": 25227 }, { "epoch": 2.6844009363694403, "grad_norm": 11.890625, "learning_rate": 9.121375781694146e-06, "loss": 0.8326, "step": 25228 }, { "epoch": 2.684507341987657, "grad_norm": 16.75, "learning_rate": 9.121340250142127e-06, "loss": 0.9014, "step": 25229 }, { "epoch": 2.684613747605874, "grad_norm": 9.5859375, "learning_rate": 9.121304718590109e-06, "loss": 0.873, "step": 25230 }, { "epoch": 2.6847201532240903, "grad_norm": 11.40625, "learning_rate": 9.12126918703809e-06, "loss": 0.8682, "step": 25231 }, { "epoch": 2.684826558842307, "grad_norm": 16.296875, "learning_rate": 9.121233655486072e-06, "loss": 0.9612, "step": 25232 }, { "epoch": 2.6849329644605238, "grad_norm": 13.2890625, "learning_rate": 9.121198123934054e-06, "loss": 1.0097, "step": 25233 }, { "epoch": 2.6850393700787403, "grad_norm": 18.1875, "learning_rate": 9.121162592382035e-06, "loss": 0.7079, "step": 25234 }, { "epoch": 2.685145775696957, "grad_norm": 14.640625, "learning_rate": 9.121127060830017e-06, "loss": 0.9, "step": 25235 }, { "epoch": 2.6852521813151733, "grad_norm": 13.109375, "learning_rate": 9.121091529278e-06, "loss": 0.9355, "step": 25236 }, { "epoch": 2.6853585869333902, "grad_norm": 18.0, "learning_rate": 9.121055997725982e-06, "loss": 1.0508, "step": 25237 }, { "epoch": 2.6854649925516068, "grad_norm": 12.171875, "learning_rate": 9.121020466173964e-06, "loss": 0.872, "step": 25238 }, { "epoch": 2.6855713981698233, "grad_norm": 12.2421875, "learning_rate": 9.120984934621945e-06, "loss": 0.8168, "step": 25239 }, { "epoch": 2.6856778037880398, "grad_norm": 12.8515625, "learning_rate": 9.120949403069927e-06, "loss": 0.8358, "step": 25240 }, { "epoch": 2.6857842094062567, "grad_norm": 13.46875, "learning_rate": 9.120913871517908e-06, "loss": 0.9132, "step": 25241 }, { "epoch": 2.6858906150244732, "grad_norm": 16.15625, "learning_rate": 9.12087833996589e-06, "loss": 0.94, "step": 25242 }, { "epoch": 2.6859970206426897, "grad_norm": 17.375, "learning_rate": 9.120842808413872e-06, "loss": 1.11, "step": 25243 }, { "epoch": 2.6861034262609067, "grad_norm": 11.5546875, "learning_rate": 9.120807276861853e-06, "loss": 0.8387, "step": 25244 }, { "epoch": 2.686209831879123, "grad_norm": 12.2734375, "learning_rate": 9.120771745309835e-06, "loss": 0.9698, "step": 25245 }, { "epoch": 2.6863162374973397, "grad_norm": 11.3984375, "learning_rate": 9.120736213757818e-06, "loss": 0.7539, "step": 25246 }, { "epoch": 2.6864226431155567, "grad_norm": 17.875, "learning_rate": 9.1207006822058e-06, "loss": 1.0523, "step": 25247 }, { "epoch": 2.686529048733773, "grad_norm": 13.0703125, "learning_rate": 9.120665150653781e-06, "loss": 0.9853, "step": 25248 }, { "epoch": 2.6866354543519897, "grad_norm": 13.53125, "learning_rate": 9.120629619101763e-06, "loss": 0.9174, "step": 25249 }, { "epoch": 2.6867418599702066, "grad_norm": 11.4609375, "learning_rate": 9.120594087549745e-06, "loss": 0.9855, "step": 25250 }, { "epoch": 2.686848265588423, "grad_norm": 13.3671875, "learning_rate": 9.120558555997728e-06, "loss": 1.033, "step": 25251 }, { "epoch": 2.6869546712066397, "grad_norm": 19.703125, "learning_rate": 9.12052302444571e-06, "loss": 0.9394, "step": 25252 }, { "epoch": 2.6870610768248566, "grad_norm": 12.421875, "learning_rate": 9.120487492893691e-06, "loss": 0.868, "step": 25253 }, { "epoch": 2.687167482443073, "grad_norm": 11.2734375, "learning_rate": 9.120451961341673e-06, "loss": 0.9632, "step": 25254 }, { "epoch": 2.6872738880612896, "grad_norm": 13.546875, "learning_rate": 9.120416429789654e-06, "loss": 1.016, "step": 25255 }, { "epoch": 2.687380293679506, "grad_norm": 11.6015625, "learning_rate": 9.120380898237636e-06, "loss": 0.7734, "step": 25256 }, { "epoch": 2.6874866992977227, "grad_norm": 12.4375, "learning_rate": 9.120345366685617e-06, "loss": 0.9316, "step": 25257 }, { "epoch": 2.6875931049159396, "grad_norm": 13.234375, "learning_rate": 9.120309835133599e-06, "loss": 0.9615, "step": 25258 }, { "epoch": 2.687699510534156, "grad_norm": 12.3515625, "learning_rate": 9.12027430358158e-06, "loss": 0.9962, "step": 25259 }, { "epoch": 2.6878059161523726, "grad_norm": 9.890625, "learning_rate": 9.120238772029562e-06, "loss": 0.8218, "step": 25260 }, { "epoch": 2.6879123217705896, "grad_norm": 18.9375, "learning_rate": 9.120203240477546e-06, "loss": 1.0346, "step": 25261 }, { "epoch": 2.688018727388806, "grad_norm": 11.15625, "learning_rate": 9.120167708925527e-06, "loss": 0.9317, "step": 25262 }, { "epoch": 2.6881251330070226, "grad_norm": 15.296875, "learning_rate": 9.120132177373509e-06, "loss": 0.8884, "step": 25263 }, { "epoch": 2.6882315386252396, "grad_norm": 14.375, "learning_rate": 9.12009664582149e-06, "loss": 1.1099, "step": 25264 }, { "epoch": 2.688337944243456, "grad_norm": 10.8515625, "learning_rate": 9.120061114269472e-06, "loss": 0.762, "step": 25265 }, { "epoch": 2.6884443498616726, "grad_norm": 14.8046875, "learning_rate": 9.120025582717454e-06, "loss": 1.106, "step": 25266 }, { "epoch": 2.6885507554798895, "grad_norm": 13.625, "learning_rate": 9.119990051165435e-06, "loss": 1.0733, "step": 25267 }, { "epoch": 2.688657161098106, "grad_norm": 13.21875, "learning_rate": 9.119954519613417e-06, "loss": 0.814, "step": 25268 }, { "epoch": 2.6887635667163226, "grad_norm": 14.3828125, "learning_rate": 9.119918988061398e-06, "loss": 1.0997, "step": 25269 }, { "epoch": 2.6888699723345395, "grad_norm": 13.9296875, "learning_rate": 9.11988345650938e-06, "loss": 1.0834, "step": 25270 }, { "epoch": 2.688976377952756, "grad_norm": 9.375, "learning_rate": 9.119847924957362e-06, "loss": 0.8094, "step": 25271 }, { "epoch": 2.6890827835709725, "grad_norm": 10.953125, "learning_rate": 9.119812393405345e-06, "loss": 0.8054, "step": 25272 }, { "epoch": 2.689189189189189, "grad_norm": 14.953125, "learning_rate": 9.119776861853327e-06, "loss": 1.1504, "step": 25273 }, { "epoch": 2.689295594807406, "grad_norm": 12.1875, "learning_rate": 9.119741330301308e-06, "loss": 1.0229, "step": 25274 }, { "epoch": 2.6894020004256225, "grad_norm": 12.640625, "learning_rate": 9.11970579874929e-06, "loss": 0.7583, "step": 25275 }, { "epoch": 2.689508406043839, "grad_norm": 8.6015625, "learning_rate": 9.119670267197271e-06, "loss": 0.8569, "step": 25276 }, { "epoch": 2.6896148116620555, "grad_norm": 11.125, "learning_rate": 9.119634735645255e-06, "loss": 0.7717, "step": 25277 }, { "epoch": 2.6897212172802725, "grad_norm": 12.984375, "learning_rate": 9.119599204093236e-06, "loss": 0.9741, "step": 25278 }, { "epoch": 2.689827622898489, "grad_norm": 13.7890625, "learning_rate": 9.119563672541218e-06, "loss": 0.9832, "step": 25279 }, { "epoch": 2.6899340285167055, "grad_norm": 15.203125, "learning_rate": 9.1195281409892e-06, "loss": 1.0061, "step": 25280 }, { "epoch": 2.6900404341349224, "grad_norm": 14.484375, "learning_rate": 9.119492609437181e-06, "loss": 0.9133, "step": 25281 }, { "epoch": 2.690146839753139, "grad_norm": 12.109375, "learning_rate": 9.119457077885163e-06, "loss": 0.8991, "step": 25282 }, { "epoch": 2.6902532453713555, "grad_norm": 11.1328125, "learning_rate": 9.119421546333144e-06, "loss": 0.8583, "step": 25283 }, { "epoch": 2.6903596509895724, "grad_norm": 13.9296875, "learning_rate": 9.119386014781126e-06, "loss": 0.9268, "step": 25284 }, { "epoch": 2.690466056607789, "grad_norm": 10.703125, "learning_rate": 9.119350483229108e-06, "loss": 0.797, "step": 25285 }, { "epoch": 2.6905724622260054, "grad_norm": 17.0, "learning_rate": 9.119314951677089e-06, "loss": 0.8979, "step": 25286 }, { "epoch": 2.6906788678442224, "grad_norm": 10.796875, "learning_rate": 9.119279420125072e-06, "loss": 0.9483, "step": 25287 }, { "epoch": 2.690785273462439, "grad_norm": 14.671875, "learning_rate": 9.119243888573054e-06, "loss": 1.1648, "step": 25288 }, { "epoch": 2.6908916790806554, "grad_norm": 12.875, "learning_rate": 9.119208357021036e-06, "loss": 0.8073, "step": 25289 }, { "epoch": 2.6909980846988724, "grad_norm": 11.1328125, "learning_rate": 9.119172825469017e-06, "loss": 0.9303, "step": 25290 }, { "epoch": 2.691104490317089, "grad_norm": 13.703125, "learning_rate": 9.119137293916999e-06, "loss": 1.0099, "step": 25291 }, { "epoch": 2.6912108959353054, "grad_norm": 12.515625, "learning_rate": 9.119101762364982e-06, "loss": 0.9421, "step": 25292 }, { "epoch": 2.691317301553522, "grad_norm": 12.390625, "learning_rate": 9.119066230812964e-06, "loss": 0.9624, "step": 25293 }, { "epoch": 2.6914237071717384, "grad_norm": 12.6171875, "learning_rate": 9.119030699260945e-06, "loss": 0.8258, "step": 25294 }, { "epoch": 2.6915301127899554, "grad_norm": 13.140625, "learning_rate": 9.118995167708925e-06, "loss": 1.097, "step": 25295 }, { "epoch": 2.691636518408172, "grad_norm": 10.21875, "learning_rate": 9.118959636156907e-06, "loss": 0.7172, "step": 25296 }, { "epoch": 2.6917429240263884, "grad_norm": 12.0, "learning_rate": 9.11892410460489e-06, "loss": 0.8947, "step": 25297 }, { "epoch": 2.6918493296446053, "grad_norm": 9.0546875, "learning_rate": 9.118888573052872e-06, "loss": 0.7751, "step": 25298 }, { "epoch": 2.691955735262822, "grad_norm": 13.234375, "learning_rate": 9.118853041500853e-06, "loss": 1.0006, "step": 25299 }, { "epoch": 2.6920621408810383, "grad_norm": 13.8515625, "learning_rate": 9.118817509948835e-06, "loss": 0.9029, "step": 25300 }, { "epoch": 2.6921685464992553, "grad_norm": 14.7734375, "learning_rate": 9.118781978396817e-06, "loss": 0.8099, "step": 25301 }, { "epoch": 2.692274952117472, "grad_norm": 12.109375, "learning_rate": 9.118746446844798e-06, "loss": 0.9715, "step": 25302 }, { "epoch": 2.6923813577356883, "grad_norm": 17.734375, "learning_rate": 9.118710915292782e-06, "loss": 1.1593, "step": 25303 }, { "epoch": 2.6924877633539053, "grad_norm": 15.859375, "learning_rate": 9.118675383740763e-06, "loss": 0.9198, "step": 25304 }, { "epoch": 2.692594168972122, "grad_norm": 10.59375, "learning_rate": 9.118639852188745e-06, "loss": 0.8138, "step": 25305 }, { "epoch": 2.6927005745903383, "grad_norm": 11.171875, "learning_rate": 9.118604320636726e-06, "loss": 0.6039, "step": 25306 }, { "epoch": 2.6928069802085552, "grad_norm": 14.1015625, "learning_rate": 9.118568789084708e-06, "loss": 0.8583, "step": 25307 }, { "epoch": 2.6929133858267718, "grad_norm": 13.171875, "learning_rate": 9.11853325753269e-06, "loss": 1.0315, "step": 25308 }, { "epoch": 2.6930197914449883, "grad_norm": 17.265625, "learning_rate": 9.118497725980671e-06, "loss": 0.9434, "step": 25309 }, { "epoch": 2.6931261970632048, "grad_norm": 15.65625, "learning_rate": 9.118462194428653e-06, "loss": 0.91, "step": 25310 }, { "epoch": 2.6932326026814217, "grad_norm": 11.765625, "learning_rate": 9.118426662876634e-06, "loss": 0.8319, "step": 25311 }, { "epoch": 2.6933390082996382, "grad_norm": 15.15625, "learning_rate": 9.118391131324616e-06, "loss": 0.9451, "step": 25312 }, { "epoch": 2.6934454139178547, "grad_norm": 16.84375, "learning_rate": 9.1183555997726e-06, "loss": 0.8463, "step": 25313 }, { "epoch": 2.6935518195360713, "grad_norm": 13.4375, "learning_rate": 9.118320068220581e-06, "loss": 0.9134, "step": 25314 }, { "epoch": 2.693658225154288, "grad_norm": 13.9375, "learning_rate": 9.118284536668563e-06, "loss": 1.0469, "step": 25315 }, { "epoch": 2.6937646307725047, "grad_norm": 13.7265625, "learning_rate": 9.118249005116544e-06, "loss": 0.941, "step": 25316 }, { "epoch": 2.6938710363907212, "grad_norm": 15.7890625, "learning_rate": 9.118213473564526e-06, "loss": 1.0397, "step": 25317 }, { "epoch": 2.693977442008938, "grad_norm": 9.9296875, "learning_rate": 9.118177942012509e-06, "loss": 0.833, "step": 25318 }, { "epoch": 2.6940838476271547, "grad_norm": 17.546875, "learning_rate": 9.11814241046049e-06, "loss": 0.8107, "step": 25319 }, { "epoch": 2.694190253245371, "grad_norm": 12.03125, "learning_rate": 9.118106878908472e-06, "loss": 0.8538, "step": 25320 }, { "epoch": 2.694296658863588, "grad_norm": 16.921875, "learning_rate": 9.118071347356454e-06, "loss": 0.7915, "step": 25321 }, { "epoch": 2.6944030644818047, "grad_norm": 10.78125, "learning_rate": 9.118035815804435e-06, "loss": 0.9026, "step": 25322 }, { "epoch": 2.694509470100021, "grad_norm": 13.421875, "learning_rate": 9.118000284252417e-06, "loss": 0.7518, "step": 25323 }, { "epoch": 2.694615875718238, "grad_norm": 12.3828125, "learning_rate": 9.117964752700399e-06, "loss": 0.8578, "step": 25324 }, { "epoch": 2.6947222813364546, "grad_norm": 10.4375, "learning_rate": 9.11792922114838e-06, "loss": 0.7503, "step": 25325 }, { "epoch": 2.694828686954671, "grad_norm": 10.5859375, "learning_rate": 9.117893689596362e-06, "loss": 0.8565, "step": 25326 }, { "epoch": 2.694935092572888, "grad_norm": 10.8125, "learning_rate": 9.117858158044343e-06, "loss": 0.9017, "step": 25327 }, { "epoch": 2.6950414981911046, "grad_norm": 12.0078125, "learning_rate": 9.117822626492327e-06, "loss": 0.8495, "step": 25328 }, { "epoch": 2.695147903809321, "grad_norm": 12.9296875, "learning_rate": 9.117787094940308e-06, "loss": 0.9579, "step": 25329 }, { "epoch": 2.6952543094275376, "grad_norm": 12.625, "learning_rate": 9.11775156338829e-06, "loss": 1.0193, "step": 25330 }, { "epoch": 2.695360715045754, "grad_norm": 12.9921875, "learning_rate": 9.117716031836272e-06, "loss": 0.9304, "step": 25331 }, { "epoch": 2.695467120663971, "grad_norm": 19.015625, "learning_rate": 9.117680500284253e-06, "loss": 1.0088, "step": 25332 }, { "epoch": 2.6955735262821876, "grad_norm": 15.9375, "learning_rate": 9.117644968732235e-06, "loss": 0.9075, "step": 25333 }, { "epoch": 2.695679931900404, "grad_norm": 11.5234375, "learning_rate": 9.117609437180216e-06, "loss": 0.7803, "step": 25334 }, { "epoch": 2.695786337518621, "grad_norm": 10.171875, "learning_rate": 9.117573905628198e-06, "loss": 0.7168, "step": 25335 }, { "epoch": 2.6958927431368376, "grad_norm": 11.8203125, "learning_rate": 9.11753837407618e-06, "loss": 0.7434, "step": 25336 }, { "epoch": 2.695999148755054, "grad_norm": 11.1484375, "learning_rate": 9.117502842524161e-06, "loss": 0.9692, "step": 25337 }, { "epoch": 2.696105554373271, "grad_norm": 15.546875, "learning_rate": 9.117467310972145e-06, "loss": 0.8933, "step": 25338 }, { "epoch": 2.6962119599914876, "grad_norm": 12.7890625, "learning_rate": 9.117431779420126e-06, "loss": 0.8366, "step": 25339 }, { "epoch": 2.696318365609704, "grad_norm": 12.125, "learning_rate": 9.117396247868108e-06, "loss": 0.8087, "step": 25340 }, { "epoch": 2.696424771227921, "grad_norm": 13.0234375, "learning_rate": 9.11736071631609e-06, "loss": 1.0995, "step": 25341 }, { "epoch": 2.6965311768461375, "grad_norm": 13.5078125, "learning_rate": 9.117325184764071e-06, "loss": 0.867, "step": 25342 }, { "epoch": 2.696637582464354, "grad_norm": 13.03125, "learning_rate": 9.117289653212053e-06, "loss": 1.0913, "step": 25343 }, { "epoch": 2.696743988082571, "grad_norm": 13.3359375, "learning_rate": 9.117254121660036e-06, "loss": 0.8632, "step": 25344 }, { "epoch": 2.6968503937007875, "grad_norm": 10.0625, "learning_rate": 9.117218590108017e-06, "loss": 0.8436, "step": 25345 }, { "epoch": 2.696956799319004, "grad_norm": 11.578125, "learning_rate": 9.117183058555999e-06, "loss": 0.6771, "step": 25346 }, { "epoch": 2.697063204937221, "grad_norm": 11.265625, "learning_rate": 9.11714752700398e-06, "loss": 0.8666, "step": 25347 }, { "epoch": 2.6971696105554375, "grad_norm": 11.1796875, "learning_rate": 9.117111995451962e-06, "loss": 0.8891, "step": 25348 }, { "epoch": 2.697276016173654, "grad_norm": 14.2421875, "learning_rate": 9.117076463899944e-06, "loss": 0.98, "step": 25349 }, { "epoch": 2.6973824217918705, "grad_norm": 13.09375, "learning_rate": 9.117040932347926e-06, "loss": 0.9159, "step": 25350 }, { "epoch": 2.697488827410087, "grad_norm": 12.125, "learning_rate": 9.117005400795907e-06, "loss": 0.9953, "step": 25351 }, { "epoch": 2.697595233028304, "grad_norm": 11.2109375, "learning_rate": 9.116969869243889e-06, "loss": 0.9095, "step": 25352 }, { "epoch": 2.6977016386465205, "grad_norm": 15.9296875, "learning_rate": 9.11693433769187e-06, "loss": 0.9401, "step": 25353 }, { "epoch": 2.697808044264737, "grad_norm": 12.6328125, "learning_rate": 9.116898806139854e-06, "loss": 0.9432, "step": 25354 }, { "epoch": 2.697914449882954, "grad_norm": 23.28125, "learning_rate": 9.116863274587835e-06, "loss": 1.099, "step": 25355 }, { "epoch": 2.6980208555011704, "grad_norm": 10.2734375, "learning_rate": 9.116827743035817e-06, "loss": 0.8624, "step": 25356 }, { "epoch": 2.698127261119387, "grad_norm": 14.4296875, "learning_rate": 9.116792211483798e-06, "loss": 0.9381, "step": 25357 }, { "epoch": 2.698233666737604, "grad_norm": 13.1796875, "learning_rate": 9.11675667993178e-06, "loss": 0.941, "step": 25358 }, { "epoch": 2.6983400723558204, "grad_norm": 14.546875, "learning_rate": 9.116721148379762e-06, "loss": 1.1806, "step": 25359 }, { "epoch": 2.698446477974037, "grad_norm": 10.6328125, "learning_rate": 9.116685616827743e-06, "loss": 1.0223, "step": 25360 }, { "epoch": 2.698552883592254, "grad_norm": 14.84375, "learning_rate": 9.116650085275725e-06, "loss": 0.8603, "step": 25361 }, { "epoch": 2.6986592892104704, "grad_norm": 10.5859375, "learning_rate": 9.116614553723706e-06, "loss": 0.9086, "step": 25362 }, { "epoch": 2.698765694828687, "grad_norm": 12.0859375, "learning_rate": 9.116579022171688e-06, "loss": 0.8095, "step": 25363 }, { "epoch": 2.698872100446904, "grad_norm": 10.875, "learning_rate": 9.116543490619671e-06, "loss": 1.0625, "step": 25364 }, { "epoch": 2.6989785060651204, "grad_norm": 12.3984375, "learning_rate": 9.116507959067653e-06, "loss": 1.0196, "step": 25365 }, { "epoch": 2.699084911683337, "grad_norm": 11.0703125, "learning_rate": 9.116472427515635e-06, "loss": 0.8302, "step": 25366 }, { "epoch": 2.6991913173015534, "grad_norm": 9.8046875, "learning_rate": 9.116436895963616e-06, "loss": 0.7761, "step": 25367 }, { "epoch": 2.6992977229197703, "grad_norm": 15.9921875, "learning_rate": 9.116401364411598e-06, "loss": 0.9855, "step": 25368 }, { "epoch": 2.699404128537987, "grad_norm": 11.609375, "learning_rate": 9.116365832859581e-06, "loss": 0.9233, "step": 25369 }, { "epoch": 2.6995105341562033, "grad_norm": 11.3046875, "learning_rate": 9.116330301307563e-06, "loss": 0.8957, "step": 25370 }, { "epoch": 2.69961693977442, "grad_norm": 10.8046875, "learning_rate": 9.116294769755544e-06, "loss": 0.8508, "step": 25371 }, { "epoch": 2.699723345392637, "grad_norm": 8.96875, "learning_rate": 9.116259238203526e-06, "loss": 0.9291, "step": 25372 }, { "epoch": 2.6998297510108533, "grad_norm": 13.3203125, "learning_rate": 9.116223706651508e-06, "loss": 0.8209, "step": 25373 }, { "epoch": 2.69993615662907, "grad_norm": 11.5546875, "learning_rate": 9.116188175099489e-06, "loss": 1.0044, "step": 25374 }, { "epoch": 2.700042562247287, "grad_norm": 12.71875, "learning_rate": 9.11615264354747e-06, "loss": 0.8062, "step": 25375 }, { "epoch": 2.7001489678655033, "grad_norm": 19.6875, "learning_rate": 9.116117111995452e-06, "loss": 0.8228, "step": 25376 }, { "epoch": 2.70025537348372, "grad_norm": 13.65625, "learning_rate": 9.116081580443434e-06, "loss": 1.0394, "step": 25377 }, { "epoch": 2.7003617791019368, "grad_norm": 13.796875, "learning_rate": 9.116046048891416e-06, "loss": 0.7977, "step": 25378 }, { "epoch": 2.7004681847201533, "grad_norm": 18.046875, "learning_rate": 9.116010517339399e-06, "loss": 0.9166, "step": 25379 }, { "epoch": 2.7005745903383698, "grad_norm": 16.453125, "learning_rate": 9.11597498578738e-06, "loss": 0.967, "step": 25380 }, { "epoch": 2.7006809959565867, "grad_norm": 11.3984375, "learning_rate": 9.115939454235362e-06, "loss": 0.8296, "step": 25381 }, { "epoch": 2.7007874015748032, "grad_norm": 15.3984375, "learning_rate": 9.115903922683344e-06, "loss": 1.0401, "step": 25382 }, { "epoch": 2.7008938071930197, "grad_norm": 9.546875, "learning_rate": 9.115868391131325e-06, "loss": 0.8028, "step": 25383 }, { "epoch": 2.7010002128112367, "grad_norm": 13.25, "learning_rate": 9.115832859579307e-06, "loss": 0.9698, "step": 25384 }, { "epoch": 2.701106618429453, "grad_norm": 12.84375, "learning_rate": 9.115797328027289e-06, "loss": 0.9328, "step": 25385 }, { "epoch": 2.7012130240476697, "grad_norm": 14.75, "learning_rate": 9.11576179647527e-06, "loss": 0.5862, "step": 25386 }, { "epoch": 2.7013194296658862, "grad_norm": 13.421875, "learning_rate": 9.115726264923252e-06, "loss": 0.9581, "step": 25387 }, { "epoch": 2.7014258352841027, "grad_norm": 13.8828125, "learning_rate": 9.115690733371233e-06, "loss": 1.0098, "step": 25388 }, { "epoch": 2.7015322409023197, "grad_norm": 12.0625, "learning_rate": 9.115655201819215e-06, "loss": 0.8985, "step": 25389 }, { "epoch": 2.701638646520536, "grad_norm": 11.4453125, "learning_rate": 9.115619670267198e-06, "loss": 0.8229, "step": 25390 }, { "epoch": 2.7017450521387527, "grad_norm": 12.4140625, "learning_rate": 9.11558413871518e-06, "loss": 0.8904, "step": 25391 }, { "epoch": 2.7018514577569697, "grad_norm": 14.1640625, "learning_rate": 9.115548607163161e-06, "loss": 1.0076, "step": 25392 }, { "epoch": 2.701957863375186, "grad_norm": 11.5625, "learning_rate": 9.115513075611143e-06, "loss": 0.9372, "step": 25393 }, { "epoch": 2.7020642689934027, "grad_norm": 18.734375, "learning_rate": 9.115477544059125e-06, "loss": 1.0682, "step": 25394 }, { "epoch": 2.7021706746116196, "grad_norm": 11.8671875, "learning_rate": 9.115442012507108e-06, "loss": 0.7207, "step": 25395 }, { "epoch": 2.702277080229836, "grad_norm": 10.8828125, "learning_rate": 9.11540648095509e-06, "loss": 0.9552, "step": 25396 }, { "epoch": 2.7023834858480527, "grad_norm": 15.859375, "learning_rate": 9.115370949403071e-06, "loss": 1.1437, "step": 25397 }, { "epoch": 2.7024898914662696, "grad_norm": 12.8828125, "learning_rate": 9.115335417851053e-06, "loss": 0.6827, "step": 25398 }, { "epoch": 2.702596297084486, "grad_norm": 12.6171875, "learning_rate": 9.115299886299034e-06, "loss": 0.8063, "step": 25399 }, { "epoch": 2.7027027027027026, "grad_norm": 12.2734375, "learning_rate": 9.115264354747016e-06, "loss": 0.8442, "step": 25400 }, { "epoch": 2.7028091083209196, "grad_norm": 15.9765625, "learning_rate": 9.115228823194998e-06, "loss": 0.9213, "step": 25401 }, { "epoch": 2.702915513939136, "grad_norm": 13.140625, "learning_rate": 9.11519329164298e-06, "loss": 1.023, "step": 25402 }, { "epoch": 2.7030219195573526, "grad_norm": 9.078125, "learning_rate": 9.11515776009096e-06, "loss": 0.8786, "step": 25403 }, { "epoch": 2.703128325175569, "grad_norm": 14.546875, "learning_rate": 9.115122228538942e-06, "loss": 1.1071, "step": 25404 }, { "epoch": 2.703234730793786, "grad_norm": 14.1484375, "learning_rate": 9.115086696986926e-06, "loss": 0.7976, "step": 25405 }, { "epoch": 2.7033411364120026, "grad_norm": 13.671875, "learning_rate": 9.115051165434907e-06, "loss": 0.8659, "step": 25406 }, { "epoch": 2.703447542030219, "grad_norm": 14.3828125, "learning_rate": 9.115015633882889e-06, "loss": 1.0998, "step": 25407 }, { "epoch": 2.7035539476484356, "grad_norm": 16.296875, "learning_rate": 9.11498010233087e-06, "loss": 1.0196, "step": 25408 }, { "epoch": 2.7036603532666525, "grad_norm": 15.53125, "learning_rate": 9.114944570778852e-06, "loss": 1.0461, "step": 25409 }, { "epoch": 2.703766758884869, "grad_norm": 14.109375, "learning_rate": 9.114909039226835e-06, "loss": 0.8057, "step": 25410 }, { "epoch": 2.7038731645030856, "grad_norm": 14.4609375, "learning_rate": 9.114873507674817e-06, "loss": 1.1174, "step": 25411 }, { "epoch": 2.7039795701213025, "grad_norm": 12.8359375, "learning_rate": 9.114837976122799e-06, "loss": 0.8189, "step": 25412 }, { "epoch": 2.704085975739519, "grad_norm": 11.4140625, "learning_rate": 9.11480244457078e-06, "loss": 1.045, "step": 25413 }, { "epoch": 2.7041923813577355, "grad_norm": 14.6875, "learning_rate": 9.114766913018762e-06, "loss": 0.739, "step": 25414 }, { "epoch": 2.7042987869759525, "grad_norm": 12.578125, "learning_rate": 9.114731381466743e-06, "loss": 0.9679, "step": 25415 }, { "epoch": 2.704405192594169, "grad_norm": 12.640625, "learning_rate": 9.114695849914725e-06, "loss": 0.7909, "step": 25416 }, { "epoch": 2.7045115982123855, "grad_norm": 15.7265625, "learning_rate": 9.114660318362707e-06, "loss": 1.0017, "step": 25417 }, { "epoch": 2.7046180038306025, "grad_norm": 12.9921875, "learning_rate": 9.114624786810688e-06, "loss": 0.8596, "step": 25418 }, { "epoch": 2.704724409448819, "grad_norm": 11.546875, "learning_rate": 9.11458925525867e-06, "loss": 0.8825, "step": 25419 }, { "epoch": 2.7048308150670355, "grad_norm": 12.578125, "learning_rate": 9.114553723706652e-06, "loss": 0.8931, "step": 25420 }, { "epoch": 2.7049372206852524, "grad_norm": 10.796875, "learning_rate": 9.114518192154635e-06, "loss": 0.8257, "step": 25421 }, { "epoch": 2.705043626303469, "grad_norm": 8.953125, "learning_rate": 9.114482660602616e-06, "loss": 0.6925, "step": 25422 }, { "epoch": 2.7051500319216855, "grad_norm": 18.4375, "learning_rate": 9.114447129050598e-06, "loss": 1.1414, "step": 25423 }, { "epoch": 2.705256437539902, "grad_norm": 10.484375, "learning_rate": 9.11441159749858e-06, "loss": 0.7569, "step": 25424 }, { "epoch": 2.7053628431581185, "grad_norm": 15.734375, "learning_rate": 9.114376065946561e-06, "loss": 0.987, "step": 25425 }, { "epoch": 2.7054692487763354, "grad_norm": 17.625, "learning_rate": 9.114340534394543e-06, "loss": 1.1673, "step": 25426 }, { "epoch": 2.705575654394552, "grad_norm": 12.0703125, "learning_rate": 9.114305002842524e-06, "loss": 1.0215, "step": 25427 }, { "epoch": 2.7056820600127685, "grad_norm": 12.3203125, "learning_rate": 9.114269471290506e-06, "loss": 0.9734, "step": 25428 }, { "epoch": 2.7057884656309854, "grad_norm": 9.9609375, "learning_rate": 9.114233939738488e-06, "loss": 0.7922, "step": 25429 }, { "epoch": 2.705894871249202, "grad_norm": 11.4140625, "learning_rate": 9.11419840818647e-06, "loss": 0.8032, "step": 25430 }, { "epoch": 2.7060012768674184, "grad_norm": 15.015625, "learning_rate": 9.114162876634453e-06, "loss": 0.9009, "step": 25431 }, { "epoch": 2.7061076824856354, "grad_norm": 11.1953125, "learning_rate": 9.114127345082434e-06, "loss": 0.8137, "step": 25432 }, { "epoch": 2.706214088103852, "grad_norm": 11.7109375, "learning_rate": 9.114091813530416e-06, "loss": 0.8344, "step": 25433 }, { "epoch": 2.7063204937220684, "grad_norm": 13.9296875, "learning_rate": 9.114056281978397e-06, "loss": 0.9867, "step": 25434 }, { "epoch": 2.7064268993402854, "grad_norm": 16.78125, "learning_rate": 9.114020750426379e-06, "loss": 1.1772, "step": 25435 }, { "epoch": 2.706533304958502, "grad_norm": 10.703125, "learning_rate": 9.113985218874362e-06, "loss": 1.0114, "step": 25436 }, { "epoch": 2.7066397105767184, "grad_norm": 13.765625, "learning_rate": 9.113949687322344e-06, "loss": 0.9068, "step": 25437 }, { "epoch": 2.7067461161949353, "grad_norm": 12.8984375, "learning_rate": 9.113914155770326e-06, "loss": 0.7802, "step": 25438 }, { "epoch": 2.706852521813152, "grad_norm": 12.4453125, "learning_rate": 9.113878624218307e-06, "loss": 0.9006, "step": 25439 }, { "epoch": 2.7069589274313683, "grad_norm": 13.6328125, "learning_rate": 9.113843092666289e-06, "loss": 0.8022, "step": 25440 }, { "epoch": 2.707065333049585, "grad_norm": 11.15625, "learning_rate": 9.11380756111427e-06, "loss": 0.8794, "step": 25441 }, { "epoch": 2.707171738667802, "grad_norm": 9.4375, "learning_rate": 9.113772029562252e-06, "loss": 0.7417, "step": 25442 }, { "epoch": 2.7072781442860183, "grad_norm": 15.2734375, "learning_rate": 9.113736498010234e-06, "loss": 0.9392, "step": 25443 }, { "epoch": 2.707384549904235, "grad_norm": 16.453125, "learning_rate": 9.113700966458215e-06, "loss": 0.9274, "step": 25444 }, { "epoch": 2.7074909555224513, "grad_norm": 14.6328125, "learning_rate": 9.113665434906197e-06, "loss": 0.8769, "step": 25445 }, { "epoch": 2.7075973611406683, "grad_norm": 11.7265625, "learning_rate": 9.11362990335418e-06, "loss": 1.1787, "step": 25446 }, { "epoch": 2.707703766758885, "grad_norm": 12.4140625, "learning_rate": 9.113594371802162e-06, "loss": 0.9213, "step": 25447 }, { "epoch": 2.7078101723771013, "grad_norm": 11.0625, "learning_rate": 9.113558840250143e-06, "loss": 0.7931, "step": 25448 }, { "epoch": 2.7079165779953183, "grad_norm": 20.953125, "learning_rate": 9.113523308698125e-06, "loss": 0.8302, "step": 25449 }, { "epoch": 2.7080229836135348, "grad_norm": 14.453125, "learning_rate": 9.113487777146107e-06, "loss": 0.8282, "step": 25450 }, { "epoch": 2.7081293892317513, "grad_norm": 12.03125, "learning_rate": 9.113452245594088e-06, "loss": 0.8351, "step": 25451 }, { "epoch": 2.7082357948499682, "grad_norm": 14.625, "learning_rate": 9.11341671404207e-06, "loss": 0.9561, "step": 25452 }, { "epoch": 2.7083422004681847, "grad_norm": 15.25, "learning_rate": 9.113381182490051e-06, "loss": 0.864, "step": 25453 }, { "epoch": 2.7084486060864013, "grad_norm": 16.8125, "learning_rate": 9.113345650938033e-06, "loss": 0.8266, "step": 25454 }, { "epoch": 2.708555011704618, "grad_norm": 13.3203125, "learning_rate": 9.113310119386015e-06, "loss": 0.851, "step": 25455 }, { "epoch": 2.7086614173228347, "grad_norm": 12.171875, "learning_rate": 9.113274587833998e-06, "loss": 0.7238, "step": 25456 }, { "epoch": 2.7087678229410512, "grad_norm": 15.3046875, "learning_rate": 9.11323905628198e-06, "loss": 0.957, "step": 25457 }, { "epoch": 2.708874228559268, "grad_norm": 13.84375, "learning_rate": 9.113203524729961e-06, "loss": 0.8814, "step": 25458 }, { "epoch": 2.7089806341774847, "grad_norm": 16.265625, "learning_rate": 9.113167993177943e-06, "loss": 0.9592, "step": 25459 }, { "epoch": 2.709087039795701, "grad_norm": 13.515625, "learning_rate": 9.113132461625924e-06, "loss": 1.0519, "step": 25460 }, { "epoch": 2.7091934454139177, "grad_norm": 12.7578125, "learning_rate": 9.113096930073906e-06, "loss": 0.9276, "step": 25461 }, { "epoch": 2.709299851032134, "grad_norm": 12.8828125, "learning_rate": 9.11306139852189e-06, "loss": 0.8386, "step": 25462 }, { "epoch": 2.709406256650351, "grad_norm": 10.375, "learning_rate": 9.11302586696987e-06, "loss": 0.8751, "step": 25463 }, { "epoch": 2.7095126622685677, "grad_norm": 13.5546875, "learning_rate": 9.112990335417852e-06, "loss": 0.8543, "step": 25464 }, { "epoch": 2.709619067886784, "grad_norm": 11.984375, "learning_rate": 9.112954803865834e-06, "loss": 0.8915, "step": 25465 }, { "epoch": 2.709725473505001, "grad_norm": 11.9921875, "learning_rate": 9.112919272313816e-06, "loss": 0.9373, "step": 25466 }, { "epoch": 2.7098318791232177, "grad_norm": 12.671875, "learning_rate": 9.112883740761797e-06, "loss": 0.9209, "step": 25467 }, { "epoch": 2.709938284741434, "grad_norm": 15.203125, "learning_rate": 9.112848209209779e-06, "loss": 0.9628, "step": 25468 }, { "epoch": 2.710044690359651, "grad_norm": 20.484375, "learning_rate": 9.11281267765776e-06, "loss": 0.8493, "step": 25469 }, { "epoch": 2.7101510959778676, "grad_norm": 16.78125, "learning_rate": 9.112777146105742e-06, "loss": 0.8377, "step": 25470 }, { "epoch": 2.710257501596084, "grad_norm": 18.765625, "learning_rate": 9.112741614553724e-06, "loss": 1.0861, "step": 25471 }, { "epoch": 2.710363907214301, "grad_norm": 13.6171875, "learning_rate": 9.112706083001707e-06, "loss": 0.8653, "step": 25472 }, { "epoch": 2.7104703128325176, "grad_norm": 16.390625, "learning_rate": 9.112670551449689e-06, "loss": 0.9019, "step": 25473 }, { "epoch": 2.710576718450734, "grad_norm": 13.3515625, "learning_rate": 9.11263501989767e-06, "loss": 1.0288, "step": 25474 }, { "epoch": 2.710683124068951, "grad_norm": 13.5078125, "learning_rate": 9.112599488345652e-06, "loss": 0.9638, "step": 25475 }, { "epoch": 2.7107895296871676, "grad_norm": 15.046875, "learning_rate": 9.112563956793633e-06, "loss": 1.109, "step": 25476 }, { "epoch": 2.710895935305384, "grad_norm": 13.3828125, "learning_rate": 9.112528425241615e-06, "loss": 0.7326, "step": 25477 }, { "epoch": 2.711002340923601, "grad_norm": 14.4609375, "learning_rate": 9.112492893689597e-06, "loss": 0.9718, "step": 25478 }, { "epoch": 2.7111087465418175, "grad_norm": 16.65625, "learning_rate": 9.112457362137578e-06, "loss": 1.0182, "step": 25479 }, { "epoch": 2.711215152160034, "grad_norm": 22.890625, "learning_rate": 9.11242183058556e-06, "loss": 0.9662, "step": 25480 }, { "epoch": 2.7113215577782506, "grad_norm": 11.484375, "learning_rate": 9.112386299033541e-06, "loss": 0.8016, "step": 25481 }, { "epoch": 2.711427963396467, "grad_norm": 12.828125, "learning_rate": 9.112350767481525e-06, "loss": 0.9882, "step": 25482 }, { "epoch": 2.711534369014684, "grad_norm": 12.2421875, "learning_rate": 9.112315235929506e-06, "loss": 0.9167, "step": 25483 }, { "epoch": 2.7116407746329005, "grad_norm": 14.078125, "learning_rate": 9.112279704377488e-06, "loss": 1.0031, "step": 25484 }, { "epoch": 2.711747180251117, "grad_norm": 12.9375, "learning_rate": 9.11224417282547e-06, "loss": 0.9417, "step": 25485 }, { "epoch": 2.711853585869334, "grad_norm": 12.5546875, "learning_rate": 9.112208641273451e-06, "loss": 0.7566, "step": 25486 }, { "epoch": 2.7119599914875505, "grad_norm": 14.015625, "learning_rate": 9.112173109721434e-06, "loss": 0.928, "step": 25487 }, { "epoch": 2.712066397105767, "grad_norm": 12.671875, "learning_rate": 9.112137578169416e-06, "loss": 1.0715, "step": 25488 }, { "epoch": 2.712172802723984, "grad_norm": 23.046875, "learning_rate": 9.112102046617398e-06, "loss": 0.8986, "step": 25489 }, { "epoch": 2.7122792083422005, "grad_norm": 18.765625, "learning_rate": 9.11206651506538e-06, "loss": 0.8414, "step": 25490 }, { "epoch": 2.712385613960417, "grad_norm": 14.609375, "learning_rate": 9.112030983513361e-06, "loss": 0.9611, "step": 25491 }, { "epoch": 2.712492019578634, "grad_norm": 14.1875, "learning_rate": 9.111995451961342e-06, "loss": 1.151, "step": 25492 }, { "epoch": 2.7125984251968505, "grad_norm": 13.609375, "learning_rate": 9.111959920409324e-06, "loss": 0.9576, "step": 25493 }, { "epoch": 2.712704830815067, "grad_norm": 15.78125, "learning_rate": 9.111924388857306e-06, "loss": 0.8539, "step": 25494 }, { "epoch": 2.712811236433284, "grad_norm": 11.46875, "learning_rate": 9.111888857305287e-06, "loss": 0.8587, "step": 25495 }, { "epoch": 2.7129176420515004, "grad_norm": 10.7421875, "learning_rate": 9.111853325753269e-06, "loss": 0.9245, "step": 25496 }, { "epoch": 2.713024047669717, "grad_norm": 11.140625, "learning_rate": 9.111817794201252e-06, "loss": 0.8084, "step": 25497 }, { "epoch": 2.7131304532879335, "grad_norm": 16.0625, "learning_rate": 9.111782262649234e-06, "loss": 0.9221, "step": 25498 }, { "epoch": 2.7132368589061504, "grad_norm": 12.0859375, "learning_rate": 9.111746731097215e-06, "loss": 1.0521, "step": 25499 }, { "epoch": 2.713343264524367, "grad_norm": 13.0078125, "learning_rate": 9.111711199545197e-06, "loss": 0.8423, "step": 25500 }, { "epoch": 2.7134496701425834, "grad_norm": 15.6484375, "learning_rate": 9.111675667993179e-06, "loss": 1.0227, "step": 25501 }, { "epoch": 2.7135560757608, "grad_norm": 16.71875, "learning_rate": 9.11164013644116e-06, "loss": 1.0739, "step": 25502 }, { "epoch": 2.713662481379017, "grad_norm": 12.8359375, "learning_rate": 9.111604604889142e-06, "loss": 0.9051, "step": 25503 }, { "epoch": 2.7137688869972334, "grad_norm": 11.6796875, "learning_rate": 9.111569073337123e-06, "loss": 0.7234, "step": 25504 }, { "epoch": 2.71387529261545, "grad_norm": 21.828125, "learning_rate": 9.111533541785105e-06, "loss": 1.2791, "step": 25505 }, { "epoch": 2.713981698233667, "grad_norm": 16.109375, "learning_rate": 9.111498010233087e-06, "loss": 1.0904, "step": 25506 }, { "epoch": 2.7140881038518834, "grad_norm": 13.8984375, "learning_rate": 9.111462478681068e-06, "loss": 0.9971, "step": 25507 }, { "epoch": 2.7141945094701, "grad_norm": 14.484375, "learning_rate": 9.111426947129052e-06, "loss": 0.9764, "step": 25508 }, { "epoch": 2.714300915088317, "grad_norm": 10.46875, "learning_rate": 9.111391415577033e-06, "loss": 0.9233, "step": 25509 }, { "epoch": 2.7144073207065333, "grad_norm": 17.796875, "learning_rate": 9.111355884025015e-06, "loss": 0.962, "step": 25510 }, { "epoch": 2.71451372632475, "grad_norm": 12.1328125, "learning_rate": 9.111320352472996e-06, "loss": 0.9989, "step": 25511 }, { "epoch": 2.714620131942967, "grad_norm": 13.125, "learning_rate": 9.111284820920978e-06, "loss": 1.0887, "step": 25512 }, { "epoch": 2.7147265375611833, "grad_norm": 13.328125, "learning_rate": 9.111249289368961e-06, "loss": 1.0109, "step": 25513 }, { "epoch": 2.7148329431794, "grad_norm": 15.2734375, "learning_rate": 9.111213757816943e-06, "loss": 0.9552, "step": 25514 }, { "epoch": 2.714939348797617, "grad_norm": 12.5390625, "learning_rate": 9.111178226264924e-06, "loss": 0.5423, "step": 25515 }, { "epoch": 2.7150457544158333, "grad_norm": 23.21875, "learning_rate": 9.111142694712906e-06, "loss": 0.9807, "step": 25516 }, { "epoch": 2.71515216003405, "grad_norm": 12.921875, "learning_rate": 9.111107163160888e-06, "loss": 0.9739, "step": 25517 }, { "epoch": 2.7152585656522663, "grad_norm": 15.3515625, "learning_rate": 9.11107163160887e-06, "loss": 0.9216, "step": 25518 }, { "epoch": 2.715364971270483, "grad_norm": 13.546875, "learning_rate": 9.111036100056851e-06, "loss": 1.0227, "step": 25519 }, { "epoch": 2.7154713768886998, "grad_norm": 12.8671875, "learning_rate": 9.111000568504833e-06, "loss": 0.7131, "step": 25520 }, { "epoch": 2.7155777825069163, "grad_norm": 17.5, "learning_rate": 9.110965036952814e-06, "loss": 1.1023, "step": 25521 }, { "epoch": 2.715684188125133, "grad_norm": 13.0078125, "learning_rate": 9.110929505400796e-06, "loss": 0.8611, "step": 25522 }, { "epoch": 2.7157905937433497, "grad_norm": 12.234375, "learning_rate": 9.110893973848779e-06, "loss": 1.0075, "step": 25523 }, { "epoch": 2.7158969993615663, "grad_norm": 12.4765625, "learning_rate": 9.11085844229676e-06, "loss": 0.9536, "step": 25524 }, { "epoch": 2.7160034049797828, "grad_norm": 14.0546875, "learning_rate": 9.110822910744742e-06, "loss": 0.819, "step": 25525 }, { "epoch": 2.7161098105979997, "grad_norm": 11.7890625, "learning_rate": 9.110787379192724e-06, "loss": 0.7688, "step": 25526 }, { "epoch": 2.7162162162162162, "grad_norm": 21.984375, "learning_rate": 9.110751847640705e-06, "loss": 1.1649, "step": 25527 }, { "epoch": 2.7163226218344327, "grad_norm": 17.328125, "learning_rate": 9.110716316088689e-06, "loss": 0.8639, "step": 25528 }, { "epoch": 2.7164290274526497, "grad_norm": 11.59375, "learning_rate": 9.11068078453667e-06, "loss": 0.8686, "step": 25529 }, { "epoch": 2.716535433070866, "grad_norm": 13.828125, "learning_rate": 9.110645252984652e-06, "loss": 0.9116, "step": 25530 }, { "epoch": 2.7166418386890827, "grad_norm": 13.4140625, "learning_rate": 9.110609721432634e-06, "loss": 0.9764, "step": 25531 }, { "epoch": 2.7167482443072997, "grad_norm": 13.2578125, "learning_rate": 9.110574189880615e-06, "loss": 1.0904, "step": 25532 }, { "epoch": 2.716854649925516, "grad_norm": 11.8515625, "learning_rate": 9.110538658328597e-06, "loss": 1.0879, "step": 25533 }, { "epoch": 2.7169610555437327, "grad_norm": 15.7421875, "learning_rate": 9.110503126776578e-06, "loss": 0.9223, "step": 25534 }, { "epoch": 2.717067461161949, "grad_norm": 12.2109375, "learning_rate": 9.11046759522456e-06, "loss": 0.8663, "step": 25535 }, { "epoch": 2.717173866780166, "grad_norm": 24.296875, "learning_rate": 9.110432063672542e-06, "loss": 0.8456, "step": 25536 }, { "epoch": 2.7172802723983827, "grad_norm": 18.640625, "learning_rate": 9.110396532120523e-06, "loss": 1.0648, "step": 25537 }, { "epoch": 2.717386678016599, "grad_norm": 10.3671875, "learning_rate": 9.110361000568505e-06, "loss": 0.7412, "step": 25538 }, { "epoch": 2.7174930836348157, "grad_norm": 13.71875, "learning_rate": 9.110325469016488e-06, "loss": 0.8307, "step": 25539 }, { "epoch": 2.7175994892530326, "grad_norm": 10.2265625, "learning_rate": 9.11028993746447e-06, "loss": 0.9472, "step": 25540 }, { "epoch": 2.717705894871249, "grad_norm": 17.046875, "learning_rate": 9.110254405912451e-06, "loss": 1.0626, "step": 25541 }, { "epoch": 2.7178123004894656, "grad_norm": 12.46875, "learning_rate": 9.110218874360433e-06, "loss": 0.9161, "step": 25542 }, { "epoch": 2.7179187061076826, "grad_norm": 12.71875, "learning_rate": 9.110183342808415e-06, "loss": 0.9867, "step": 25543 }, { "epoch": 2.718025111725899, "grad_norm": 12.203125, "learning_rate": 9.110147811256396e-06, "loss": 0.8558, "step": 25544 }, { "epoch": 2.7181315173441156, "grad_norm": 14.53125, "learning_rate": 9.110112279704378e-06, "loss": 0.9313, "step": 25545 }, { "epoch": 2.7182379229623326, "grad_norm": 19.609375, "learning_rate": 9.11007674815236e-06, "loss": 0.8841, "step": 25546 }, { "epoch": 2.718344328580549, "grad_norm": 19.515625, "learning_rate": 9.110041216600341e-06, "loss": 1.0496, "step": 25547 }, { "epoch": 2.7184507341987656, "grad_norm": 16.546875, "learning_rate": 9.110005685048323e-06, "loss": 1.0255, "step": 25548 }, { "epoch": 2.7185571398169825, "grad_norm": 13.4609375, "learning_rate": 9.109970153496306e-06, "loss": 1.2063, "step": 25549 }, { "epoch": 2.718663545435199, "grad_norm": 9.6015625, "learning_rate": 9.109934621944287e-06, "loss": 0.7331, "step": 25550 }, { "epoch": 2.7187699510534156, "grad_norm": 13.3203125, "learning_rate": 9.109899090392269e-06, "loss": 0.9697, "step": 25551 }, { "epoch": 2.7188763566716325, "grad_norm": 13.9296875, "learning_rate": 9.10986355884025e-06, "loss": 0.9888, "step": 25552 }, { "epoch": 2.718982762289849, "grad_norm": 15.40625, "learning_rate": 9.109828027288232e-06, "loss": 1.2011, "step": 25553 }, { "epoch": 2.7190891679080655, "grad_norm": 18.078125, "learning_rate": 9.109792495736216e-06, "loss": 1.083, "step": 25554 }, { "epoch": 2.719195573526282, "grad_norm": 10.8828125, "learning_rate": 9.109756964184197e-06, "loss": 0.9302, "step": 25555 }, { "epoch": 2.7193019791444986, "grad_norm": 11.390625, "learning_rate": 9.109721432632179e-06, "loss": 0.9495, "step": 25556 }, { "epoch": 2.7194083847627155, "grad_norm": 17.171875, "learning_rate": 9.10968590108016e-06, "loss": 0.9375, "step": 25557 }, { "epoch": 2.719514790380932, "grad_norm": 11.28125, "learning_rate": 9.109650369528142e-06, "loss": 0.9267, "step": 25558 }, { "epoch": 2.7196211959991485, "grad_norm": 10.5859375, "learning_rate": 9.109614837976124e-06, "loss": 0.8157, "step": 25559 }, { "epoch": 2.7197276016173655, "grad_norm": 10.203125, "learning_rate": 9.109579306424105e-06, "loss": 0.7702, "step": 25560 }, { "epoch": 2.719834007235582, "grad_norm": 12.765625, "learning_rate": 9.109543774872087e-06, "loss": 0.8494, "step": 25561 }, { "epoch": 2.7199404128537985, "grad_norm": 11.078125, "learning_rate": 9.109508243320068e-06, "loss": 0.9054, "step": 25562 }, { "epoch": 2.7200468184720155, "grad_norm": 12.84375, "learning_rate": 9.10947271176805e-06, "loss": 0.9581, "step": 25563 }, { "epoch": 2.720153224090232, "grad_norm": 9.859375, "learning_rate": 9.109437180216033e-06, "loss": 0.9452, "step": 25564 }, { "epoch": 2.7202596297084485, "grad_norm": 9.8828125, "learning_rate": 9.109401648664015e-06, "loss": 0.8561, "step": 25565 }, { "epoch": 2.7203660353266654, "grad_norm": 11.1953125, "learning_rate": 9.109366117111997e-06, "loss": 0.8424, "step": 25566 }, { "epoch": 2.720472440944882, "grad_norm": 14.109375, "learning_rate": 9.109330585559978e-06, "loss": 0.9147, "step": 25567 }, { "epoch": 2.7205788465630985, "grad_norm": 13.46875, "learning_rate": 9.10929505400796e-06, "loss": 0.9261, "step": 25568 }, { "epoch": 2.7206852521813154, "grad_norm": 10.8671875, "learning_rate": 9.109259522455941e-06, "loss": 0.9336, "step": 25569 }, { "epoch": 2.720791657799532, "grad_norm": 9.5234375, "learning_rate": 9.109223990903923e-06, "loss": 0.8433, "step": 25570 }, { "epoch": 2.7208980634177484, "grad_norm": 16.09375, "learning_rate": 9.109188459351905e-06, "loss": 0.94, "step": 25571 }, { "epoch": 2.721004469035965, "grad_norm": 10.375, "learning_rate": 9.109152927799886e-06, "loss": 0.7919, "step": 25572 }, { "epoch": 2.721110874654182, "grad_norm": 12.3984375, "learning_rate": 9.109117396247868e-06, "loss": 0.865, "step": 25573 }, { "epoch": 2.7212172802723984, "grad_norm": 12.3125, "learning_rate": 9.109081864695851e-06, "loss": 0.8001, "step": 25574 }, { "epoch": 2.721323685890615, "grad_norm": 9.609375, "learning_rate": 9.109046333143833e-06, "loss": 0.9249, "step": 25575 }, { "epoch": 2.7214300915088314, "grad_norm": 12.3984375, "learning_rate": 9.109010801591814e-06, "loss": 0.9019, "step": 25576 }, { "epoch": 2.7215364971270484, "grad_norm": 12.359375, "learning_rate": 9.108975270039796e-06, "loss": 0.905, "step": 25577 }, { "epoch": 2.721642902745265, "grad_norm": 15.2421875, "learning_rate": 9.108939738487778e-06, "loss": 1.1495, "step": 25578 }, { "epoch": 2.7217493083634814, "grad_norm": 12.84375, "learning_rate": 9.10890420693576e-06, "loss": 0.9996, "step": 25579 }, { "epoch": 2.7218557139816983, "grad_norm": 12.8359375, "learning_rate": 9.108868675383742e-06, "loss": 0.9771, "step": 25580 }, { "epoch": 2.721962119599915, "grad_norm": 11.75, "learning_rate": 9.108833143831724e-06, "loss": 1.0133, "step": 25581 }, { "epoch": 2.7220685252181314, "grad_norm": 22.4375, "learning_rate": 9.108797612279706e-06, "loss": 0.8451, "step": 25582 }, { "epoch": 2.7221749308363483, "grad_norm": 11.046875, "learning_rate": 9.108762080727687e-06, "loss": 0.8961, "step": 25583 }, { "epoch": 2.722281336454565, "grad_norm": 10.21875, "learning_rate": 9.108726549175669e-06, "loss": 0.8963, "step": 25584 }, { "epoch": 2.7223877420727813, "grad_norm": 13.1015625, "learning_rate": 9.10869101762365e-06, "loss": 0.8701, "step": 25585 }, { "epoch": 2.7224941476909983, "grad_norm": 11.984375, "learning_rate": 9.108655486071632e-06, "loss": 0.8118, "step": 25586 }, { "epoch": 2.722600553309215, "grad_norm": 12.03125, "learning_rate": 9.108619954519614e-06, "loss": 0.9395, "step": 25587 }, { "epoch": 2.7227069589274313, "grad_norm": 12.9140625, "learning_rate": 9.108584422967595e-06, "loss": 0.9574, "step": 25588 }, { "epoch": 2.7228133645456483, "grad_norm": 10.203125, "learning_rate": 9.108548891415577e-06, "loss": 0.8724, "step": 25589 }, { "epoch": 2.7229197701638648, "grad_norm": 14.5, "learning_rate": 9.10851335986356e-06, "loss": 1.0448, "step": 25590 }, { "epoch": 2.7230261757820813, "grad_norm": 15.7578125, "learning_rate": 9.108477828311542e-06, "loss": 0.9065, "step": 25591 }, { "epoch": 2.723132581400298, "grad_norm": 14.96875, "learning_rate": 9.108442296759523e-06, "loss": 0.8917, "step": 25592 }, { "epoch": 2.7232389870185143, "grad_norm": 14.4140625, "learning_rate": 9.108406765207505e-06, "loss": 0.9653, "step": 25593 }, { "epoch": 2.7233453926367313, "grad_norm": 18.734375, "learning_rate": 9.108371233655487e-06, "loss": 0.8604, "step": 25594 }, { "epoch": 2.7234517982549478, "grad_norm": 15.0078125, "learning_rate": 9.108335702103468e-06, "loss": 1.027, "step": 25595 }, { "epoch": 2.7235582038731643, "grad_norm": 13.890625, "learning_rate": 9.10830017055145e-06, "loss": 0.9919, "step": 25596 }, { "epoch": 2.7236646094913812, "grad_norm": 11.375, "learning_rate": 9.108264638999431e-06, "loss": 0.902, "step": 25597 }, { "epoch": 2.7237710151095977, "grad_norm": 13.6484375, "learning_rate": 9.108229107447413e-06, "loss": 0.9521, "step": 25598 }, { "epoch": 2.7238774207278142, "grad_norm": 14.484375, "learning_rate": 9.108193575895395e-06, "loss": 0.8919, "step": 25599 }, { "epoch": 2.723983826346031, "grad_norm": 12.9453125, "learning_rate": 9.108158044343378e-06, "loss": 0.7837, "step": 25600 }, { "epoch": 2.7240902319642477, "grad_norm": 15.1953125, "learning_rate": 9.10812251279136e-06, "loss": 0.9929, "step": 25601 }, { "epoch": 2.724196637582464, "grad_norm": 11.9375, "learning_rate": 9.108086981239341e-06, "loss": 0.8447, "step": 25602 }, { "epoch": 2.724303043200681, "grad_norm": 11.5078125, "learning_rate": 9.108051449687323e-06, "loss": 0.909, "step": 25603 }, { "epoch": 2.7244094488188977, "grad_norm": 9.59375, "learning_rate": 9.108015918135304e-06, "loss": 0.7476, "step": 25604 }, { "epoch": 2.724515854437114, "grad_norm": 10.25, "learning_rate": 9.107980386583288e-06, "loss": 0.8226, "step": 25605 }, { "epoch": 2.724622260055331, "grad_norm": 14.53125, "learning_rate": 9.10794485503127e-06, "loss": 1.0569, "step": 25606 }, { "epoch": 2.7247286656735477, "grad_norm": 16.71875, "learning_rate": 9.107909323479251e-06, "loss": 1.148, "step": 25607 }, { "epoch": 2.724835071291764, "grad_norm": 11.5546875, "learning_rate": 9.107873791927233e-06, "loss": 0.8608, "step": 25608 }, { "epoch": 2.724941476909981, "grad_norm": 12.671875, "learning_rate": 9.107838260375214e-06, "loss": 0.9737, "step": 25609 }, { "epoch": 2.7250478825281976, "grad_norm": 30.390625, "learning_rate": 9.107802728823196e-06, "loss": 1.2098, "step": 25610 }, { "epoch": 2.725154288146414, "grad_norm": 12.3828125, "learning_rate": 9.107767197271177e-06, "loss": 0.8627, "step": 25611 }, { "epoch": 2.7252606937646306, "grad_norm": 13.4921875, "learning_rate": 9.107731665719159e-06, "loss": 0.848, "step": 25612 }, { "epoch": 2.725367099382847, "grad_norm": 14.6640625, "learning_rate": 9.10769613416714e-06, "loss": 0.798, "step": 25613 }, { "epoch": 2.725473505001064, "grad_norm": 12.8828125, "learning_rate": 9.107660602615122e-06, "loss": 0.9661, "step": 25614 }, { "epoch": 2.7255799106192806, "grad_norm": 15.609375, "learning_rate": 9.107625071063104e-06, "loss": 0.9529, "step": 25615 }, { "epoch": 2.725686316237497, "grad_norm": 13.3203125, "learning_rate": 9.107589539511087e-06, "loss": 0.9879, "step": 25616 }, { "epoch": 2.725792721855714, "grad_norm": 10.078125, "learning_rate": 9.107554007959069e-06, "loss": 0.8807, "step": 25617 }, { "epoch": 2.7258991274739306, "grad_norm": 12.921875, "learning_rate": 9.10751847640705e-06, "loss": 1.0533, "step": 25618 }, { "epoch": 2.726005533092147, "grad_norm": 13.71875, "learning_rate": 9.107482944855032e-06, "loss": 0.9397, "step": 25619 }, { "epoch": 2.726111938710364, "grad_norm": 11.421875, "learning_rate": 9.107447413303014e-06, "loss": 0.8816, "step": 25620 }, { "epoch": 2.7262183443285806, "grad_norm": 14.6328125, "learning_rate": 9.107411881750997e-06, "loss": 0.9478, "step": 25621 }, { "epoch": 2.726324749946797, "grad_norm": 13.0234375, "learning_rate": 9.107376350198978e-06, "loss": 0.9506, "step": 25622 }, { "epoch": 2.726431155565014, "grad_norm": 13.46875, "learning_rate": 9.107340818646958e-06, "loss": 1.0543, "step": 25623 }, { "epoch": 2.7265375611832305, "grad_norm": 14.34375, "learning_rate": 9.10730528709494e-06, "loss": 0.8619, "step": 25624 }, { "epoch": 2.726643966801447, "grad_norm": 12.359375, "learning_rate": 9.107269755542922e-06, "loss": 0.9772, "step": 25625 }, { "epoch": 2.726750372419664, "grad_norm": 11.8203125, "learning_rate": 9.107234223990905e-06, "loss": 0.8569, "step": 25626 }, { "epoch": 2.7268567780378805, "grad_norm": 11.2578125, "learning_rate": 9.107198692438886e-06, "loss": 0.7523, "step": 25627 }, { "epoch": 2.726963183656097, "grad_norm": 13.40625, "learning_rate": 9.107163160886868e-06, "loss": 0.7878, "step": 25628 }, { "epoch": 2.7270695892743135, "grad_norm": 13.8671875, "learning_rate": 9.10712762933485e-06, "loss": 0.9205, "step": 25629 }, { "epoch": 2.7271759948925305, "grad_norm": 8.90625, "learning_rate": 9.107092097782831e-06, "loss": 0.8562, "step": 25630 }, { "epoch": 2.727282400510747, "grad_norm": 11.1875, "learning_rate": 9.107056566230815e-06, "loss": 0.8833, "step": 25631 }, { "epoch": 2.7273888061289635, "grad_norm": 12.8828125, "learning_rate": 9.107021034678796e-06, "loss": 0.8408, "step": 25632 }, { "epoch": 2.72749521174718, "grad_norm": 13.3359375, "learning_rate": 9.106985503126778e-06, "loss": 0.8341, "step": 25633 }, { "epoch": 2.727601617365397, "grad_norm": 12.0390625, "learning_rate": 9.10694997157476e-06, "loss": 1.0093, "step": 25634 }, { "epoch": 2.7277080229836135, "grad_norm": 22.25, "learning_rate": 9.106914440022741e-06, "loss": 0.9656, "step": 25635 }, { "epoch": 2.72781442860183, "grad_norm": 9.1015625, "learning_rate": 9.106878908470723e-06, "loss": 0.8575, "step": 25636 }, { "epoch": 2.727920834220047, "grad_norm": 12.90625, "learning_rate": 9.106843376918704e-06, "loss": 1.0317, "step": 25637 }, { "epoch": 2.7280272398382635, "grad_norm": 14.5546875, "learning_rate": 9.106807845366686e-06, "loss": 0.802, "step": 25638 }, { "epoch": 2.72813364545648, "grad_norm": 12.1171875, "learning_rate": 9.106772313814667e-06, "loss": 1.0365, "step": 25639 }, { "epoch": 2.728240051074697, "grad_norm": 12.6015625, "learning_rate": 9.106736782262649e-06, "loss": 1.0737, "step": 25640 }, { "epoch": 2.7283464566929134, "grad_norm": 18.0, "learning_rate": 9.106701250710632e-06, "loss": 1.0292, "step": 25641 }, { "epoch": 2.72845286231113, "grad_norm": 14.9140625, "learning_rate": 9.106665719158614e-06, "loss": 1.0948, "step": 25642 }, { "epoch": 2.728559267929347, "grad_norm": 14.2734375, "learning_rate": 9.106630187606596e-06, "loss": 0.939, "step": 25643 }, { "epoch": 2.7286656735475634, "grad_norm": 14.0078125, "learning_rate": 9.106594656054577e-06, "loss": 1.0027, "step": 25644 }, { "epoch": 2.72877207916578, "grad_norm": 11.6875, "learning_rate": 9.106559124502559e-06, "loss": 0.8713, "step": 25645 }, { "epoch": 2.728878484783997, "grad_norm": 13.015625, "learning_rate": 9.106523592950542e-06, "loss": 1.0684, "step": 25646 }, { "epoch": 2.7289848904022134, "grad_norm": 15.59375, "learning_rate": 9.106488061398524e-06, "loss": 1.0373, "step": 25647 }, { "epoch": 2.72909129602043, "grad_norm": 10.6171875, "learning_rate": 9.106452529846505e-06, "loss": 0.9112, "step": 25648 }, { "epoch": 2.7291977016386464, "grad_norm": 15.7421875, "learning_rate": 9.106416998294487e-06, "loss": 0.9262, "step": 25649 }, { "epoch": 2.729304107256863, "grad_norm": 12.1640625, "learning_rate": 9.106381466742468e-06, "loss": 0.9246, "step": 25650 }, { "epoch": 2.72941051287508, "grad_norm": 17.109375, "learning_rate": 9.10634593519045e-06, "loss": 0.8607, "step": 25651 }, { "epoch": 2.7295169184932964, "grad_norm": 9.9609375, "learning_rate": 9.106310403638432e-06, "loss": 0.789, "step": 25652 }, { "epoch": 2.729623324111513, "grad_norm": 11.9921875, "learning_rate": 9.106274872086413e-06, "loss": 0.9358, "step": 25653 }, { "epoch": 2.72972972972973, "grad_norm": 10.71875, "learning_rate": 9.106239340534395e-06, "loss": 0.9832, "step": 25654 }, { "epoch": 2.7298361353479463, "grad_norm": 12.4453125, "learning_rate": 9.106203808982377e-06, "loss": 0.9444, "step": 25655 }, { "epoch": 2.729942540966163, "grad_norm": 11.4296875, "learning_rate": 9.106168277430358e-06, "loss": 1.0161, "step": 25656 }, { "epoch": 2.73004894658438, "grad_norm": 12.6484375, "learning_rate": 9.106132745878341e-06, "loss": 0.9971, "step": 25657 }, { "epoch": 2.7301553522025963, "grad_norm": 14.2421875, "learning_rate": 9.106097214326323e-06, "loss": 0.8974, "step": 25658 }, { "epoch": 2.730261757820813, "grad_norm": 12.5390625, "learning_rate": 9.106061682774305e-06, "loss": 0.8771, "step": 25659 }, { "epoch": 2.7303681634390298, "grad_norm": 12.078125, "learning_rate": 9.106026151222286e-06, "loss": 0.8793, "step": 25660 }, { "epoch": 2.7304745690572463, "grad_norm": 19.515625, "learning_rate": 9.105990619670268e-06, "loss": 0.9313, "step": 25661 }, { "epoch": 2.730580974675463, "grad_norm": 13.5, "learning_rate": 9.10595508811825e-06, "loss": 0.9082, "step": 25662 }, { "epoch": 2.7306873802936797, "grad_norm": 12.203125, "learning_rate": 9.105919556566231e-06, "loss": 0.8946, "step": 25663 }, { "epoch": 2.7307937859118963, "grad_norm": 15.3515625, "learning_rate": 9.105884025014213e-06, "loss": 1.0338, "step": 25664 }, { "epoch": 2.7309001915301128, "grad_norm": 16.765625, "learning_rate": 9.105848493462194e-06, "loss": 1.0681, "step": 25665 }, { "epoch": 2.7310065971483293, "grad_norm": 14.0234375, "learning_rate": 9.105812961910176e-06, "loss": 0.8552, "step": 25666 }, { "epoch": 2.7311130027665462, "grad_norm": 11.109375, "learning_rate": 9.10577743035816e-06, "loss": 0.8747, "step": 25667 }, { "epoch": 2.7312194083847627, "grad_norm": 10.140625, "learning_rate": 9.10574189880614e-06, "loss": 0.9032, "step": 25668 }, { "epoch": 2.7313258140029792, "grad_norm": 10.34375, "learning_rate": 9.105706367254122e-06, "loss": 0.8696, "step": 25669 }, { "epoch": 2.7314322196211958, "grad_norm": 13.7578125, "learning_rate": 9.105670835702104e-06, "loss": 0.7674, "step": 25670 }, { "epoch": 2.7315386252394127, "grad_norm": 15.3515625, "learning_rate": 9.105635304150086e-06, "loss": 0.9174, "step": 25671 }, { "epoch": 2.731645030857629, "grad_norm": 14.609375, "learning_rate": 9.105599772598069e-06, "loss": 0.9362, "step": 25672 }, { "epoch": 2.7317514364758457, "grad_norm": 12.171875, "learning_rate": 9.10556424104605e-06, "loss": 0.9412, "step": 25673 }, { "epoch": 2.7318578420940627, "grad_norm": 13.0703125, "learning_rate": 9.105528709494032e-06, "loss": 0.8336, "step": 25674 }, { "epoch": 2.731964247712279, "grad_norm": 11.6328125, "learning_rate": 9.105493177942014e-06, "loss": 0.8343, "step": 25675 }, { "epoch": 2.7320706533304957, "grad_norm": 17.0625, "learning_rate": 9.105457646389995e-06, "loss": 0.9327, "step": 25676 }, { "epoch": 2.7321770589487127, "grad_norm": 12.65625, "learning_rate": 9.105422114837977e-06, "loss": 0.8287, "step": 25677 }, { "epoch": 2.732283464566929, "grad_norm": 14.046875, "learning_rate": 9.105386583285959e-06, "loss": 0.6818, "step": 25678 }, { "epoch": 2.7323898701851457, "grad_norm": 9.6796875, "learning_rate": 9.10535105173394e-06, "loss": 0.746, "step": 25679 }, { "epoch": 2.7324962758033626, "grad_norm": 12.0625, "learning_rate": 9.105315520181922e-06, "loss": 0.9725, "step": 25680 }, { "epoch": 2.732602681421579, "grad_norm": 12.9765625, "learning_rate": 9.105279988629903e-06, "loss": 0.9973, "step": 25681 }, { "epoch": 2.7327090870397956, "grad_norm": 14.9609375, "learning_rate": 9.105244457077887e-06, "loss": 0.7549, "step": 25682 }, { "epoch": 2.7328154926580126, "grad_norm": 11.7578125, "learning_rate": 9.105208925525868e-06, "loss": 0.8277, "step": 25683 }, { "epoch": 2.732921898276229, "grad_norm": 12.9765625, "learning_rate": 9.10517339397385e-06, "loss": 0.9702, "step": 25684 }, { "epoch": 2.7330283038944456, "grad_norm": 9.8984375, "learning_rate": 9.105137862421832e-06, "loss": 0.7945, "step": 25685 }, { "epoch": 2.733134709512662, "grad_norm": 13.5390625, "learning_rate": 9.105102330869813e-06, "loss": 0.9181, "step": 25686 }, { "epoch": 2.7332411151308786, "grad_norm": 11.9609375, "learning_rate": 9.105066799317795e-06, "loss": 0.9021, "step": 25687 }, { "epoch": 2.7333475207490956, "grad_norm": 15.3203125, "learning_rate": 9.105031267765776e-06, "loss": 0.761, "step": 25688 }, { "epoch": 2.733453926367312, "grad_norm": 15.2109375, "learning_rate": 9.104995736213758e-06, "loss": 1.022, "step": 25689 }, { "epoch": 2.7335603319855286, "grad_norm": 13.65625, "learning_rate": 9.10496020466174e-06, "loss": 0.9181, "step": 25690 }, { "epoch": 2.7336667376037456, "grad_norm": 11.1640625, "learning_rate": 9.104924673109721e-06, "loss": 0.6277, "step": 25691 }, { "epoch": 2.733773143221962, "grad_norm": 14.8828125, "learning_rate": 9.104889141557704e-06, "loss": 0.7986, "step": 25692 }, { "epoch": 2.7338795488401786, "grad_norm": 13.3828125, "learning_rate": 9.104853610005686e-06, "loss": 0.9782, "step": 25693 }, { "epoch": 2.7339859544583955, "grad_norm": 13.953125, "learning_rate": 9.104818078453668e-06, "loss": 1.102, "step": 25694 }, { "epoch": 2.734092360076612, "grad_norm": 11.2421875, "learning_rate": 9.10478254690165e-06, "loss": 0.7645, "step": 25695 }, { "epoch": 2.7341987656948286, "grad_norm": 11.390625, "learning_rate": 9.104747015349631e-06, "loss": 0.8448, "step": 25696 }, { "epoch": 2.7343051713130455, "grad_norm": 14.046875, "learning_rate": 9.104711483797612e-06, "loss": 1.0169, "step": 25697 }, { "epoch": 2.734411576931262, "grad_norm": 12.0546875, "learning_rate": 9.104675952245596e-06, "loss": 0.9797, "step": 25698 }, { "epoch": 2.7345179825494785, "grad_norm": 12.8515625, "learning_rate": 9.104640420693577e-06, "loss": 0.8954, "step": 25699 }, { "epoch": 2.7346243881676955, "grad_norm": 11.4609375, "learning_rate": 9.104604889141559e-06, "loss": 1.0537, "step": 25700 }, { "epoch": 2.734730793785912, "grad_norm": 12.8828125, "learning_rate": 9.10456935758954e-06, "loss": 0.9425, "step": 25701 }, { "epoch": 2.7348371994041285, "grad_norm": 14.34375, "learning_rate": 9.104533826037522e-06, "loss": 0.9717, "step": 25702 }, { "epoch": 2.734943605022345, "grad_norm": 9.5625, "learning_rate": 9.104498294485504e-06, "loss": 0.628, "step": 25703 }, { "epoch": 2.735050010640562, "grad_norm": 12.3515625, "learning_rate": 9.104462762933485e-06, "loss": 0.9088, "step": 25704 }, { "epoch": 2.7351564162587785, "grad_norm": 15.5390625, "learning_rate": 9.104427231381467e-06, "loss": 0.8908, "step": 25705 }, { "epoch": 2.735262821876995, "grad_norm": 12.8359375, "learning_rate": 9.104391699829449e-06, "loss": 0.8761, "step": 25706 }, { "epoch": 2.7353692274952115, "grad_norm": 12.171875, "learning_rate": 9.10435616827743e-06, "loss": 0.9589, "step": 25707 }, { "epoch": 2.7354756331134285, "grad_norm": 11.7578125, "learning_rate": 9.104320636725414e-06, "loss": 0.9848, "step": 25708 }, { "epoch": 2.735582038731645, "grad_norm": 13.578125, "learning_rate": 9.104285105173395e-06, "loss": 0.89, "step": 25709 }, { "epoch": 2.7356884443498615, "grad_norm": 9.5078125, "learning_rate": 9.104249573621377e-06, "loss": 0.6937, "step": 25710 }, { "epoch": 2.7357948499680784, "grad_norm": 10.9765625, "learning_rate": 9.104214042069358e-06, "loss": 0.8465, "step": 25711 }, { "epoch": 2.735901255586295, "grad_norm": 15.2890625, "learning_rate": 9.10417851051734e-06, "loss": 1.1056, "step": 25712 }, { "epoch": 2.7360076612045114, "grad_norm": 13.046875, "learning_rate": 9.104142978965322e-06, "loss": 0.7589, "step": 25713 }, { "epoch": 2.7361140668227284, "grad_norm": 10.296875, "learning_rate": 9.104107447413303e-06, "loss": 0.8387, "step": 25714 }, { "epoch": 2.736220472440945, "grad_norm": 18.171875, "learning_rate": 9.104071915861285e-06, "loss": 1.0163, "step": 25715 }, { "epoch": 2.7363268780591614, "grad_norm": 11.234375, "learning_rate": 9.104036384309266e-06, "loss": 0.8516, "step": 25716 }, { "epoch": 2.7364332836773784, "grad_norm": 15.34375, "learning_rate": 9.104000852757248e-06, "loss": 1.1072, "step": 25717 }, { "epoch": 2.736539689295595, "grad_norm": 13.2265625, "learning_rate": 9.103965321205231e-06, "loss": 0.8384, "step": 25718 }, { "epoch": 2.7366460949138114, "grad_norm": 11.4453125, "learning_rate": 9.103929789653213e-06, "loss": 0.8478, "step": 25719 }, { "epoch": 2.7367525005320283, "grad_norm": 20.125, "learning_rate": 9.103894258101195e-06, "loss": 0.9383, "step": 25720 }, { "epoch": 2.736858906150245, "grad_norm": 14.90625, "learning_rate": 9.103858726549176e-06, "loss": 1.201, "step": 25721 }, { "epoch": 2.7369653117684614, "grad_norm": 10.78125, "learning_rate": 9.103823194997158e-06, "loss": 0.7349, "step": 25722 }, { "epoch": 2.737071717386678, "grad_norm": 11.9296875, "learning_rate": 9.103787663445141e-06, "loss": 0.834, "step": 25723 }, { "epoch": 2.7371781230048944, "grad_norm": 13.2734375, "learning_rate": 9.103752131893123e-06, "loss": 1.0012, "step": 25724 }, { "epoch": 2.7372845286231113, "grad_norm": 11.9140625, "learning_rate": 9.103716600341104e-06, "loss": 0.8861, "step": 25725 }, { "epoch": 2.737390934241328, "grad_norm": 15.59375, "learning_rate": 9.103681068789086e-06, "loss": 0.9388, "step": 25726 }, { "epoch": 2.7374973398595444, "grad_norm": 15.859375, "learning_rate": 9.103645537237067e-06, "loss": 0.9939, "step": 25727 }, { "epoch": 2.7376037454777613, "grad_norm": 14.2578125, "learning_rate": 9.103610005685049e-06, "loss": 1.0055, "step": 25728 }, { "epoch": 2.737710151095978, "grad_norm": 13.8671875, "learning_rate": 9.10357447413303e-06, "loss": 0.8783, "step": 25729 }, { "epoch": 2.7378165567141943, "grad_norm": 16.96875, "learning_rate": 9.103538942581012e-06, "loss": 0.9843, "step": 25730 }, { "epoch": 2.7379229623324113, "grad_norm": 14.2734375, "learning_rate": 9.103503411028994e-06, "loss": 0.9167, "step": 25731 }, { "epoch": 2.738029367950628, "grad_norm": 11.21875, "learning_rate": 9.103467879476975e-06, "loss": 0.8119, "step": 25732 }, { "epoch": 2.7381357735688443, "grad_norm": 16.8125, "learning_rate": 9.103432347924957e-06, "loss": 0.7685, "step": 25733 }, { "epoch": 2.7382421791870613, "grad_norm": 13.0625, "learning_rate": 9.10339681637294e-06, "loss": 0.8555, "step": 25734 }, { "epoch": 2.7383485848052778, "grad_norm": 9.8203125, "learning_rate": 9.103361284820922e-06, "loss": 0.6607, "step": 25735 }, { "epoch": 2.7384549904234943, "grad_norm": 10.1953125, "learning_rate": 9.103325753268904e-06, "loss": 0.7459, "step": 25736 }, { "epoch": 2.7385613960417112, "grad_norm": 17.921875, "learning_rate": 9.103290221716885e-06, "loss": 0.9531, "step": 25737 }, { "epoch": 2.7386678016599277, "grad_norm": 13.453125, "learning_rate": 9.103254690164867e-06, "loss": 0.9465, "step": 25738 }, { "epoch": 2.7387742072781442, "grad_norm": 15.21875, "learning_rate": 9.10321915861285e-06, "loss": 1.036, "step": 25739 }, { "epoch": 2.738880612896361, "grad_norm": 16.296875, "learning_rate": 9.103183627060832e-06, "loss": 0.9588, "step": 25740 }, { "epoch": 2.7389870185145777, "grad_norm": 16.140625, "learning_rate": 9.103148095508813e-06, "loss": 0.919, "step": 25741 }, { "epoch": 2.739093424132794, "grad_norm": 14.1328125, "learning_rate": 9.103112563956795e-06, "loss": 0.9825, "step": 25742 }, { "epoch": 2.7391998297510107, "grad_norm": 12.6171875, "learning_rate": 9.103077032404775e-06, "loss": 0.9717, "step": 25743 }, { "epoch": 2.7393062353692272, "grad_norm": 14.234375, "learning_rate": 9.103041500852758e-06, "loss": 1.0013, "step": 25744 }, { "epoch": 2.739412640987444, "grad_norm": 17.734375, "learning_rate": 9.10300596930074e-06, "loss": 0.8931, "step": 25745 }, { "epoch": 2.7395190466056607, "grad_norm": 14.109375, "learning_rate": 9.102970437748721e-06, "loss": 1.0202, "step": 25746 }, { "epoch": 2.739625452223877, "grad_norm": 23.5, "learning_rate": 9.102934906196703e-06, "loss": 0.9285, "step": 25747 }, { "epoch": 2.739731857842094, "grad_norm": 11.5546875, "learning_rate": 9.102899374644685e-06, "loss": 0.8174, "step": 25748 }, { "epoch": 2.7398382634603107, "grad_norm": 11.4296875, "learning_rate": 9.102863843092668e-06, "loss": 0.8675, "step": 25749 }, { "epoch": 2.739944669078527, "grad_norm": 10.8046875, "learning_rate": 9.10282831154065e-06, "loss": 0.7906, "step": 25750 }, { "epoch": 2.740051074696744, "grad_norm": 12.2265625, "learning_rate": 9.102792779988631e-06, "loss": 0.9691, "step": 25751 }, { "epoch": 2.7401574803149606, "grad_norm": 14.9375, "learning_rate": 9.102757248436613e-06, "loss": 0.9381, "step": 25752 }, { "epoch": 2.740263885933177, "grad_norm": 15.1328125, "learning_rate": 9.102721716884594e-06, "loss": 1.0191, "step": 25753 }, { "epoch": 2.740370291551394, "grad_norm": 16.671875, "learning_rate": 9.102686185332576e-06, "loss": 0.8993, "step": 25754 }, { "epoch": 2.7404766971696106, "grad_norm": 14.3046875, "learning_rate": 9.102650653780558e-06, "loss": 1.0963, "step": 25755 }, { "epoch": 2.740583102787827, "grad_norm": 15.3828125, "learning_rate": 9.102615122228539e-06, "loss": 0.9239, "step": 25756 }, { "epoch": 2.740689508406044, "grad_norm": 14.4296875, "learning_rate": 9.10257959067652e-06, "loss": 0.9874, "step": 25757 }, { "epoch": 2.7407959140242606, "grad_norm": 9.9453125, "learning_rate": 9.102544059124502e-06, "loss": 0.8234, "step": 25758 }, { "epoch": 2.740902319642477, "grad_norm": 19.1875, "learning_rate": 9.102508527572486e-06, "loss": 0.9535, "step": 25759 }, { "epoch": 2.7410087252606936, "grad_norm": 14.4453125, "learning_rate": 9.102472996020467e-06, "loss": 0.9932, "step": 25760 }, { "epoch": 2.7411151308789106, "grad_norm": 15.0390625, "learning_rate": 9.102437464468449e-06, "loss": 1.1039, "step": 25761 }, { "epoch": 2.741221536497127, "grad_norm": 13.6015625, "learning_rate": 9.10240193291643e-06, "loss": 0.8993, "step": 25762 }, { "epoch": 2.7413279421153436, "grad_norm": 11.65625, "learning_rate": 9.102366401364412e-06, "loss": 0.9433, "step": 25763 }, { "epoch": 2.74143434773356, "grad_norm": 14.3984375, "learning_rate": 9.102330869812395e-06, "loss": 0.8798, "step": 25764 }, { "epoch": 2.741540753351777, "grad_norm": 12.7265625, "learning_rate": 9.102295338260377e-06, "loss": 0.8072, "step": 25765 }, { "epoch": 2.7416471589699936, "grad_norm": 14.078125, "learning_rate": 9.102259806708359e-06, "loss": 0.8675, "step": 25766 }, { "epoch": 2.74175356458821, "grad_norm": 14.265625, "learning_rate": 9.10222427515634e-06, "loss": 1.0988, "step": 25767 }, { "epoch": 2.741859970206427, "grad_norm": 12.09375, "learning_rate": 9.102188743604322e-06, "loss": 0.7957, "step": 25768 }, { "epoch": 2.7419663758246435, "grad_norm": 12.3125, "learning_rate": 9.102153212052303e-06, "loss": 0.9388, "step": 25769 }, { "epoch": 2.74207278144286, "grad_norm": 17.0625, "learning_rate": 9.102117680500285e-06, "loss": 0.9061, "step": 25770 }, { "epoch": 2.742179187061077, "grad_norm": 12.5, "learning_rate": 9.102082148948267e-06, "loss": 0.7821, "step": 25771 }, { "epoch": 2.7422855926792935, "grad_norm": 11.640625, "learning_rate": 9.102046617396248e-06, "loss": 0.9779, "step": 25772 }, { "epoch": 2.74239199829751, "grad_norm": 9.3828125, "learning_rate": 9.10201108584423e-06, "loss": 0.7363, "step": 25773 }, { "epoch": 2.742498403915727, "grad_norm": 11.046875, "learning_rate": 9.101975554292211e-06, "loss": 0.8661, "step": 25774 }, { "epoch": 2.7426048095339435, "grad_norm": 10.34375, "learning_rate": 9.101940022740195e-06, "loss": 0.9504, "step": 25775 }, { "epoch": 2.74271121515216, "grad_norm": 13.2578125, "learning_rate": 9.101904491188176e-06, "loss": 0.9937, "step": 25776 }, { "epoch": 2.742817620770377, "grad_norm": 11.984375, "learning_rate": 9.101868959636158e-06, "loss": 0.961, "step": 25777 }, { "epoch": 2.7429240263885935, "grad_norm": 18.265625, "learning_rate": 9.10183342808414e-06, "loss": 1.0434, "step": 25778 }, { "epoch": 2.74303043200681, "grad_norm": 13.140625, "learning_rate": 9.101797896532121e-06, "loss": 0.9076, "step": 25779 }, { "epoch": 2.7431368376250265, "grad_norm": 10.1015625, "learning_rate": 9.101762364980103e-06, "loss": 0.7342, "step": 25780 }, { "epoch": 2.743243243243243, "grad_norm": 16.734375, "learning_rate": 9.101726833428084e-06, "loss": 0.8986, "step": 25781 }, { "epoch": 2.74334964886146, "grad_norm": 19.375, "learning_rate": 9.101691301876066e-06, "loss": 0.8507, "step": 25782 }, { "epoch": 2.7434560544796764, "grad_norm": 10.1484375, "learning_rate": 9.101655770324048e-06, "loss": 0.9372, "step": 25783 }, { "epoch": 2.743562460097893, "grad_norm": 14.015625, "learning_rate": 9.10162023877203e-06, "loss": 0.8168, "step": 25784 }, { "epoch": 2.74366886571611, "grad_norm": 15.953125, "learning_rate": 9.101584707220012e-06, "loss": 0.8252, "step": 25785 }, { "epoch": 2.7437752713343264, "grad_norm": 12.4375, "learning_rate": 9.101549175667994e-06, "loss": 0.938, "step": 25786 }, { "epoch": 2.743881676952543, "grad_norm": 12.1484375, "learning_rate": 9.101513644115976e-06, "loss": 0.9528, "step": 25787 }, { "epoch": 2.74398808257076, "grad_norm": 16.0625, "learning_rate": 9.101478112563957e-06, "loss": 0.8375, "step": 25788 }, { "epoch": 2.7440944881889764, "grad_norm": 13.2421875, "learning_rate": 9.101442581011939e-06, "loss": 1.0237, "step": 25789 }, { "epoch": 2.744200893807193, "grad_norm": 12.421875, "learning_rate": 9.101407049459922e-06, "loss": 0.8441, "step": 25790 }, { "epoch": 2.74430729942541, "grad_norm": 13.125, "learning_rate": 9.101371517907904e-06, "loss": 0.802, "step": 25791 }, { "epoch": 2.7444137050436264, "grad_norm": 13.5234375, "learning_rate": 9.101335986355885e-06, "loss": 0.9824, "step": 25792 }, { "epoch": 2.744520110661843, "grad_norm": 12.6875, "learning_rate": 9.101300454803867e-06, "loss": 0.8182, "step": 25793 }, { "epoch": 2.74462651628006, "grad_norm": 11.9296875, "learning_rate": 9.101264923251849e-06, "loss": 0.8292, "step": 25794 }, { "epoch": 2.7447329218982763, "grad_norm": 9.9765625, "learning_rate": 9.10122939169983e-06, "loss": 0.8532, "step": 25795 }, { "epoch": 2.744839327516493, "grad_norm": 13.2421875, "learning_rate": 9.101193860147812e-06, "loss": 0.9524, "step": 25796 }, { "epoch": 2.7449457331347094, "grad_norm": 9.6953125, "learning_rate": 9.101158328595793e-06, "loss": 0.8875, "step": 25797 }, { "epoch": 2.7450521387529263, "grad_norm": 16.84375, "learning_rate": 9.101122797043775e-06, "loss": 0.8825, "step": 25798 }, { "epoch": 2.745158544371143, "grad_norm": 13.296875, "learning_rate": 9.101087265491757e-06, "loss": 0.8314, "step": 25799 }, { "epoch": 2.7452649499893593, "grad_norm": 10.671875, "learning_rate": 9.10105173393974e-06, "loss": 0.9025, "step": 25800 }, { "epoch": 2.745371355607576, "grad_norm": 12.640625, "learning_rate": 9.101016202387722e-06, "loss": 0.8874, "step": 25801 }, { "epoch": 2.745477761225793, "grad_norm": 12.7734375, "learning_rate": 9.100980670835703e-06, "loss": 0.8474, "step": 25802 }, { "epoch": 2.7455841668440093, "grad_norm": 11.9296875, "learning_rate": 9.100945139283685e-06, "loss": 1.1162, "step": 25803 }, { "epoch": 2.745690572462226, "grad_norm": 10.96875, "learning_rate": 9.100909607731666e-06, "loss": 0.8435, "step": 25804 }, { "epoch": 2.7457969780804428, "grad_norm": 16.140625, "learning_rate": 9.100874076179648e-06, "loss": 1.0006, "step": 25805 }, { "epoch": 2.7459033836986593, "grad_norm": 17.84375, "learning_rate": 9.10083854462763e-06, "loss": 0.9285, "step": 25806 }, { "epoch": 2.746009789316876, "grad_norm": 11.6953125, "learning_rate": 9.100803013075611e-06, "loss": 0.95, "step": 25807 }, { "epoch": 2.7461161949350927, "grad_norm": 11.0, "learning_rate": 9.100767481523593e-06, "loss": 0.6596, "step": 25808 }, { "epoch": 2.7462226005533092, "grad_norm": 13.5859375, "learning_rate": 9.100731949971574e-06, "loss": 1.0481, "step": 25809 }, { "epoch": 2.7463290061715258, "grad_norm": 12.46875, "learning_rate": 9.100696418419558e-06, "loss": 0.9574, "step": 25810 }, { "epoch": 2.7464354117897427, "grad_norm": 22.578125, "learning_rate": 9.10066088686754e-06, "loss": 0.9184, "step": 25811 }, { "epoch": 2.746541817407959, "grad_norm": 11.1875, "learning_rate": 9.100625355315521e-06, "loss": 0.7588, "step": 25812 }, { "epoch": 2.7466482230261757, "grad_norm": 17.640625, "learning_rate": 9.100589823763503e-06, "loss": 0.9356, "step": 25813 }, { "epoch": 2.7467546286443927, "grad_norm": 20.640625, "learning_rate": 9.100554292211484e-06, "loss": 1.0864, "step": 25814 }, { "epoch": 2.746861034262609, "grad_norm": 18.609375, "learning_rate": 9.100518760659466e-06, "loss": 0.8576, "step": 25815 }, { "epoch": 2.7469674398808257, "grad_norm": 11.09375, "learning_rate": 9.100483229107449e-06, "loss": 0.8706, "step": 25816 }, { "epoch": 2.747073845499042, "grad_norm": 13.1953125, "learning_rate": 9.10044769755543e-06, "loss": 0.9851, "step": 25817 }, { "epoch": 2.7471802511172587, "grad_norm": 13.890625, "learning_rate": 9.100412166003412e-06, "loss": 0.986, "step": 25818 }, { "epoch": 2.7472866567354757, "grad_norm": 11.9765625, "learning_rate": 9.100376634451394e-06, "loss": 0.9354, "step": 25819 }, { "epoch": 2.747393062353692, "grad_norm": 12.6640625, "learning_rate": 9.100341102899376e-06, "loss": 1.0186, "step": 25820 }, { "epoch": 2.7474994679719087, "grad_norm": 14.53125, "learning_rate": 9.100305571347357e-06, "loss": 0.9532, "step": 25821 }, { "epoch": 2.7476058735901256, "grad_norm": 12.0546875, "learning_rate": 9.100270039795339e-06, "loss": 0.9636, "step": 25822 }, { "epoch": 2.747712279208342, "grad_norm": 10.921875, "learning_rate": 9.10023450824332e-06, "loss": 0.8026, "step": 25823 }, { "epoch": 2.7478186848265587, "grad_norm": 18.3125, "learning_rate": 9.100198976691302e-06, "loss": 0.8466, "step": 25824 }, { "epoch": 2.7479250904447756, "grad_norm": 17.34375, "learning_rate": 9.100163445139284e-06, "loss": 0.8923, "step": 25825 }, { "epoch": 2.748031496062992, "grad_norm": 18.75, "learning_rate": 9.100127913587267e-06, "loss": 0.9356, "step": 25826 }, { "epoch": 2.7481379016812086, "grad_norm": 14.59375, "learning_rate": 9.100092382035248e-06, "loss": 1.1619, "step": 25827 }, { "epoch": 2.7482443072994256, "grad_norm": 12.9296875, "learning_rate": 9.10005685048323e-06, "loss": 0.8576, "step": 25828 }, { "epoch": 2.748350712917642, "grad_norm": 10.59375, "learning_rate": 9.100021318931212e-06, "loss": 0.9348, "step": 25829 }, { "epoch": 2.7484571185358586, "grad_norm": 13.78125, "learning_rate": 9.099985787379193e-06, "loss": 1.0946, "step": 25830 }, { "epoch": 2.7485635241540756, "grad_norm": 11.2421875, "learning_rate": 9.099950255827177e-06, "loss": 0.8686, "step": 25831 }, { "epoch": 2.748669929772292, "grad_norm": 12.78125, "learning_rate": 9.099914724275156e-06, "loss": 0.8965, "step": 25832 }, { "epoch": 2.7487763353905086, "grad_norm": 15.7734375, "learning_rate": 9.099879192723138e-06, "loss": 0.9213, "step": 25833 }, { "epoch": 2.748882741008725, "grad_norm": 17.953125, "learning_rate": 9.09984366117112e-06, "loss": 1.0209, "step": 25834 }, { "epoch": 2.748989146626942, "grad_norm": 12.875, "learning_rate": 9.099808129619101e-06, "loss": 0.8442, "step": 25835 }, { "epoch": 2.7490955522451586, "grad_norm": 14.8671875, "learning_rate": 9.099772598067085e-06, "loss": 0.8531, "step": 25836 }, { "epoch": 2.749201957863375, "grad_norm": 12.6640625, "learning_rate": 9.099737066515066e-06, "loss": 0.9014, "step": 25837 }, { "epoch": 2.7493083634815916, "grad_norm": 10.4375, "learning_rate": 9.099701534963048e-06, "loss": 0.731, "step": 25838 }, { "epoch": 2.7494147690998085, "grad_norm": 13.0625, "learning_rate": 9.09966600341103e-06, "loss": 0.9132, "step": 25839 }, { "epoch": 2.749521174718025, "grad_norm": 17.3125, "learning_rate": 9.099630471859011e-06, "loss": 0.9395, "step": 25840 }, { "epoch": 2.7496275803362416, "grad_norm": 14.6640625, "learning_rate": 9.099594940306994e-06, "loss": 0.9532, "step": 25841 }, { "epoch": 2.7497339859544585, "grad_norm": 11.765625, "learning_rate": 9.099559408754976e-06, "loss": 0.9264, "step": 25842 }, { "epoch": 2.749840391572675, "grad_norm": 14.1484375, "learning_rate": 9.099523877202958e-06, "loss": 0.8864, "step": 25843 }, { "epoch": 2.7499467971908915, "grad_norm": 13.1484375, "learning_rate": 9.099488345650939e-06, "loss": 1.1236, "step": 25844 }, { "epoch": 2.7500532028091085, "grad_norm": 13.1328125, "learning_rate": 9.09945281409892e-06, "loss": 0.8696, "step": 25845 }, { "epoch": 2.750159608427325, "grad_norm": 13.90625, "learning_rate": 9.099417282546902e-06, "loss": 0.9668, "step": 25846 }, { "epoch": 2.7502660140455415, "grad_norm": 15.4296875, "learning_rate": 9.099381750994884e-06, "loss": 1.0772, "step": 25847 }, { "epoch": 2.7503724196637584, "grad_norm": 13.140625, "learning_rate": 9.099346219442866e-06, "loss": 1.1637, "step": 25848 }, { "epoch": 2.750478825281975, "grad_norm": 14.34375, "learning_rate": 9.099310687890847e-06, "loss": 0.9121, "step": 25849 }, { "epoch": 2.7505852309001915, "grad_norm": 14.75, "learning_rate": 9.099275156338829e-06, "loss": 0.7408, "step": 25850 }, { "epoch": 2.7506916365184084, "grad_norm": 14.8828125, "learning_rate": 9.09923962478681e-06, "loss": 0.6663, "step": 25851 }, { "epoch": 2.750798042136625, "grad_norm": 14.5078125, "learning_rate": 9.099204093234794e-06, "loss": 0.9602, "step": 25852 }, { "epoch": 2.7509044477548414, "grad_norm": 11.6796875, "learning_rate": 9.099168561682775e-06, "loss": 0.9094, "step": 25853 }, { "epoch": 2.751010853373058, "grad_norm": 14.109375, "learning_rate": 9.099133030130757e-06, "loss": 0.6374, "step": 25854 }, { "epoch": 2.751117258991275, "grad_norm": 13.3203125, "learning_rate": 9.099097498578739e-06, "loss": 0.9163, "step": 25855 }, { "epoch": 2.7512236646094914, "grad_norm": 11.34375, "learning_rate": 9.09906196702672e-06, "loss": 0.9244, "step": 25856 }, { "epoch": 2.751330070227708, "grad_norm": 8.5546875, "learning_rate": 9.099026435474703e-06, "loss": 0.6399, "step": 25857 }, { "epoch": 2.7514364758459244, "grad_norm": 13.5703125, "learning_rate": 9.098990903922685e-06, "loss": 0.8428, "step": 25858 }, { "epoch": 2.7515428814641414, "grad_norm": 13.46875, "learning_rate": 9.098955372370667e-06, "loss": 0.9863, "step": 25859 }, { "epoch": 2.751649287082358, "grad_norm": 12.6015625, "learning_rate": 9.098919840818648e-06, "loss": 0.7988, "step": 25860 }, { "epoch": 2.7517556927005744, "grad_norm": 14.953125, "learning_rate": 9.09888430926663e-06, "loss": 0.8468, "step": 25861 }, { "epoch": 2.7518620983187914, "grad_norm": 13.1328125, "learning_rate": 9.098848777714611e-06, "loss": 0.991, "step": 25862 }, { "epoch": 2.751968503937008, "grad_norm": 11.8671875, "learning_rate": 9.098813246162593e-06, "loss": 0.867, "step": 25863 }, { "epoch": 2.7520749095552244, "grad_norm": 16.609375, "learning_rate": 9.098777714610575e-06, "loss": 0.8472, "step": 25864 }, { "epoch": 2.7521813151734413, "grad_norm": 15.96875, "learning_rate": 9.098742183058556e-06, "loss": 0.9088, "step": 25865 }, { "epoch": 2.752287720791658, "grad_norm": 26.09375, "learning_rate": 9.098706651506538e-06, "loss": 1.0008, "step": 25866 }, { "epoch": 2.7523941264098744, "grad_norm": 11.75, "learning_rate": 9.098671119954521e-06, "loss": 0.7732, "step": 25867 }, { "epoch": 2.7525005320280913, "grad_norm": 10.7734375, "learning_rate": 9.098635588402503e-06, "loss": 0.8239, "step": 25868 }, { "epoch": 2.752606937646308, "grad_norm": 13.0703125, "learning_rate": 9.098600056850484e-06, "loss": 0.8747, "step": 25869 }, { "epoch": 2.7527133432645243, "grad_norm": 10.3671875, "learning_rate": 9.098564525298466e-06, "loss": 0.8902, "step": 25870 }, { "epoch": 2.7528197488827413, "grad_norm": 12.0625, "learning_rate": 9.098528993746448e-06, "loss": 0.8912, "step": 25871 }, { "epoch": 2.752926154500958, "grad_norm": 12.546875, "learning_rate": 9.09849346219443e-06, "loss": 0.8943, "step": 25872 }, { "epoch": 2.7530325601191743, "grad_norm": 13.59375, "learning_rate": 9.09845793064241e-06, "loss": 0.9929, "step": 25873 }, { "epoch": 2.753138965737391, "grad_norm": 11.265625, "learning_rate": 9.098422399090392e-06, "loss": 0.9679, "step": 25874 }, { "epoch": 2.7532453713556073, "grad_norm": 14.5703125, "learning_rate": 9.098386867538374e-06, "loss": 0.934, "step": 25875 }, { "epoch": 2.7533517769738243, "grad_norm": 12.703125, "learning_rate": 9.098351335986356e-06, "loss": 0.9573, "step": 25876 }, { "epoch": 2.753458182592041, "grad_norm": 13.8515625, "learning_rate": 9.098315804434339e-06, "loss": 0.9952, "step": 25877 }, { "epoch": 2.7535645882102573, "grad_norm": 11.765625, "learning_rate": 9.09828027288232e-06, "loss": 0.8812, "step": 25878 }, { "epoch": 2.7536709938284742, "grad_norm": 10.3203125, "learning_rate": 9.098244741330302e-06, "loss": 0.7566, "step": 25879 }, { "epoch": 2.7537773994466908, "grad_norm": 15.546875, "learning_rate": 9.098209209778284e-06, "loss": 0.9437, "step": 25880 }, { "epoch": 2.7538838050649073, "grad_norm": 11.9453125, "learning_rate": 9.098173678226265e-06, "loss": 1.0388, "step": 25881 }, { "epoch": 2.753990210683124, "grad_norm": 9.96875, "learning_rate": 9.098138146674249e-06, "loss": 0.8029, "step": 25882 }, { "epoch": 2.7540966163013407, "grad_norm": 18.171875, "learning_rate": 9.09810261512223e-06, "loss": 0.966, "step": 25883 }, { "epoch": 2.7542030219195572, "grad_norm": 12.65625, "learning_rate": 9.098067083570212e-06, "loss": 0.9841, "step": 25884 }, { "epoch": 2.754309427537774, "grad_norm": 10.609375, "learning_rate": 9.098031552018193e-06, "loss": 0.8721, "step": 25885 }, { "epoch": 2.7544158331559907, "grad_norm": 18.4375, "learning_rate": 9.097996020466175e-06, "loss": 1.0316, "step": 25886 }, { "epoch": 2.754522238774207, "grad_norm": 10.0625, "learning_rate": 9.097960488914157e-06, "loss": 0.7859, "step": 25887 }, { "epoch": 2.754628644392424, "grad_norm": 12.390625, "learning_rate": 9.097924957362138e-06, "loss": 0.8618, "step": 25888 }, { "epoch": 2.7547350500106407, "grad_norm": 12.84375, "learning_rate": 9.09788942581012e-06, "loss": 1.1344, "step": 25889 }, { "epoch": 2.754841455628857, "grad_norm": 12.2578125, "learning_rate": 9.097853894258102e-06, "loss": 0.7852, "step": 25890 }, { "epoch": 2.7549478612470737, "grad_norm": 10.71875, "learning_rate": 9.097818362706083e-06, "loss": 0.9288, "step": 25891 }, { "epoch": 2.7550542668652906, "grad_norm": 11.9375, "learning_rate": 9.097782831154065e-06, "loss": 0.9414, "step": 25892 }, { "epoch": 2.755160672483507, "grad_norm": 14.15625, "learning_rate": 9.097747299602048e-06, "loss": 0.8851, "step": 25893 }, { "epoch": 2.7552670781017237, "grad_norm": 13.6328125, "learning_rate": 9.09771176805003e-06, "loss": 0.6981, "step": 25894 }, { "epoch": 2.75537348371994, "grad_norm": 13.015625, "learning_rate": 9.097676236498011e-06, "loss": 0.8855, "step": 25895 }, { "epoch": 2.755479889338157, "grad_norm": 12.65625, "learning_rate": 9.097640704945993e-06, "loss": 0.859, "step": 25896 }, { "epoch": 2.7555862949563736, "grad_norm": 13.6796875, "learning_rate": 9.097605173393974e-06, "loss": 0.8643, "step": 25897 }, { "epoch": 2.75569270057459, "grad_norm": 11.5625, "learning_rate": 9.097569641841956e-06, "loss": 0.8784, "step": 25898 }, { "epoch": 2.755799106192807, "grad_norm": 14.9921875, "learning_rate": 9.097534110289938e-06, "loss": 0.9947, "step": 25899 }, { "epoch": 2.7559055118110236, "grad_norm": 16.609375, "learning_rate": 9.09749857873792e-06, "loss": 0.8216, "step": 25900 }, { "epoch": 2.75601191742924, "grad_norm": 23.4375, "learning_rate": 9.097463047185901e-06, "loss": 0.9605, "step": 25901 }, { "epoch": 2.756118323047457, "grad_norm": 19.3125, "learning_rate": 9.097427515633883e-06, "loss": 1.3134, "step": 25902 }, { "epoch": 2.7562247286656736, "grad_norm": 14.203125, "learning_rate": 9.097391984081866e-06, "loss": 0.8464, "step": 25903 }, { "epoch": 2.75633113428389, "grad_norm": 13.8203125, "learning_rate": 9.097356452529847e-06, "loss": 1.0586, "step": 25904 }, { "epoch": 2.756437539902107, "grad_norm": 11.015625, "learning_rate": 9.097320920977829e-06, "loss": 0.889, "step": 25905 }, { "epoch": 2.7565439455203236, "grad_norm": 9.3203125, "learning_rate": 9.09728538942581e-06, "loss": 0.8291, "step": 25906 }, { "epoch": 2.75665035113854, "grad_norm": 13.4453125, "learning_rate": 9.097249857873792e-06, "loss": 1.0703, "step": 25907 }, { "epoch": 2.756756756756757, "grad_norm": 11.09375, "learning_rate": 9.097214326321776e-06, "loss": 0.9173, "step": 25908 }, { "epoch": 2.7568631623749735, "grad_norm": 14.2734375, "learning_rate": 9.097178794769757e-06, "loss": 1.01, "step": 25909 }, { "epoch": 2.75696956799319, "grad_norm": 19.0, "learning_rate": 9.097143263217739e-06, "loss": 0.9899, "step": 25910 }, { "epoch": 2.7570759736114065, "grad_norm": 18.625, "learning_rate": 9.09710773166572e-06, "loss": 0.891, "step": 25911 }, { "epoch": 2.757182379229623, "grad_norm": 12.3359375, "learning_rate": 9.097072200113702e-06, "loss": 0.7426, "step": 25912 }, { "epoch": 2.75728878484784, "grad_norm": 12.921875, "learning_rate": 9.097036668561684e-06, "loss": 0.8552, "step": 25913 }, { "epoch": 2.7573951904660565, "grad_norm": 9.8359375, "learning_rate": 9.097001137009665e-06, "loss": 0.8584, "step": 25914 }, { "epoch": 2.757501596084273, "grad_norm": 12.15625, "learning_rate": 9.096965605457647e-06, "loss": 0.8444, "step": 25915 }, { "epoch": 2.75760800170249, "grad_norm": 14.6328125, "learning_rate": 9.096930073905628e-06, "loss": 1.1296, "step": 25916 }, { "epoch": 2.7577144073207065, "grad_norm": 15.4921875, "learning_rate": 9.09689454235361e-06, "loss": 1.0388, "step": 25917 }, { "epoch": 2.757820812938923, "grad_norm": 12.6796875, "learning_rate": 9.096859010801593e-06, "loss": 0.8825, "step": 25918 }, { "epoch": 2.75792721855714, "grad_norm": 14.453125, "learning_rate": 9.096823479249575e-06, "loss": 1.03, "step": 25919 }, { "epoch": 2.7580336241753565, "grad_norm": 13.7265625, "learning_rate": 9.096787947697557e-06, "loss": 0.7602, "step": 25920 }, { "epoch": 2.758140029793573, "grad_norm": 12.1796875, "learning_rate": 9.096752416145538e-06, "loss": 0.7801, "step": 25921 }, { "epoch": 2.75824643541179, "grad_norm": 12.8359375, "learning_rate": 9.09671688459352e-06, "loss": 0.9375, "step": 25922 }, { "epoch": 2.7583528410300064, "grad_norm": 11.578125, "learning_rate": 9.096681353041501e-06, "loss": 0.8916, "step": 25923 }, { "epoch": 2.758459246648223, "grad_norm": 13.2890625, "learning_rate": 9.096645821489483e-06, "loss": 0.9586, "step": 25924 }, { "epoch": 2.75856565226644, "grad_norm": 13.640625, "learning_rate": 9.096610289937465e-06, "loss": 0.8303, "step": 25925 }, { "epoch": 2.7586720578846564, "grad_norm": 10.0703125, "learning_rate": 9.096574758385446e-06, "loss": 0.8889, "step": 25926 }, { "epoch": 2.758778463502873, "grad_norm": 14.9453125, "learning_rate": 9.096539226833428e-06, "loss": 1.1868, "step": 25927 }, { "epoch": 2.7588848691210894, "grad_norm": 15.3671875, "learning_rate": 9.09650369528141e-06, "loss": 1.1441, "step": 25928 }, { "epoch": 2.7589912747393064, "grad_norm": 14.7578125, "learning_rate": 9.096468163729393e-06, "loss": 0.8735, "step": 25929 }, { "epoch": 2.759097680357523, "grad_norm": 14.6015625, "learning_rate": 9.096432632177374e-06, "loss": 0.9268, "step": 25930 }, { "epoch": 2.7592040859757394, "grad_norm": 16.828125, "learning_rate": 9.096397100625356e-06, "loss": 1.0606, "step": 25931 }, { "epoch": 2.759310491593956, "grad_norm": 12.4765625, "learning_rate": 9.096361569073337e-06, "loss": 0.9082, "step": 25932 }, { "epoch": 2.759416897212173, "grad_norm": 10.5546875, "learning_rate": 9.096326037521319e-06, "loss": 0.8378, "step": 25933 }, { "epoch": 2.7595233028303894, "grad_norm": 13.125, "learning_rate": 9.096290505969302e-06, "loss": 0.941, "step": 25934 }, { "epoch": 2.759629708448606, "grad_norm": 13.453125, "learning_rate": 9.096254974417284e-06, "loss": 0.8941, "step": 25935 }, { "epoch": 2.759736114066823, "grad_norm": 9.796875, "learning_rate": 9.096219442865266e-06, "loss": 0.9317, "step": 25936 }, { "epoch": 2.7598425196850394, "grad_norm": 13.0390625, "learning_rate": 9.096183911313247e-06, "loss": 0.9102, "step": 25937 }, { "epoch": 2.759948925303256, "grad_norm": 12.0390625, "learning_rate": 9.096148379761229e-06, "loss": 0.9854, "step": 25938 }, { "epoch": 2.760055330921473, "grad_norm": 15.5625, "learning_rate": 9.09611284820921e-06, "loss": 0.9, "step": 25939 }, { "epoch": 2.7601617365396893, "grad_norm": 10.3046875, "learning_rate": 9.096077316657192e-06, "loss": 0.8605, "step": 25940 }, { "epoch": 2.760268142157906, "grad_norm": 10.2578125, "learning_rate": 9.096041785105174e-06, "loss": 0.8497, "step": 25941 }, { "epoch": 2.760374547776123, "grad_norm": 12.7265625, "learning_rate": 9.096006253553155e-06, "loss": 0.8415, "step": 25942 }, { "epoch": 2.7604809533943393, "grad_norm": 18.75, "learning_rate": 9.095970722001137e-06, "loss": 0.995, "step": 25943 }, { "epoch": 2.760587359012556, "grad_norm": 11.7890625, "learning_rate": 9.09593519044912e-06, "loss": 0.956, "step": 25944 }, { "epoch": 2.7606937646307728, "grad_norm": 14.0234375, "learning_rate": 9.095899658897102e-06, "loss": 0.8796, "step": 25945 }, { "epoch": 2.7608001702489893, "grad_norm": 12.328125, "learning_rate": 9.095864127345083e-06, "loss": 1.0032, "step": 25946 }, { "epoch": 2.760906575867206, "grad_norm": 16.34375, "learning_rate": 9.095828595793065e-06, "loss": 0.9259, "step": 25947 }, { "epoch": 2.7610129814854223, "grad_norm": 12.1015625, "learning_rate": 9.095793064241047e-06, "loss": 0.7757, "step": 25948 }, { "epoch": 2.761119387103639, "grad_norm": 16.25, "learning_rate": 9.09575753268903e-06, "loss": 0.8875, "step": 25949 }, { "epoch": 2.7612257927218558, "grad_norm": 10.84375, "learning_rate": 9.095722001137011e-06, "loss": 0.9307, "step": 25950 }, { "epoch": 2.7613321983400723, "grad_norm": 12.8828125, "learning_rate": 9.095686469584993e-06, "loss": 1.0092, "step": 25951 }, { "epoch": 2.7614386039582888, "grad_norm": 10.796875, "learning_rate": 9.095650938032973e-06, "loss": 0.9093, "step": 25952 }, { "epoch": 2.7615450095765057, "grad_norm": 12.875, "learning_rate": 9.095615406480955e-06, "loss": 0.8201, "step": 25953 }, { "epoch": 2.7616514151947222, "grad_norm": 11.8359375, "learning_rate": 9.095579874928938e-06, "loss": 0.906, "step": 25954 }, { "epoch": 2.7617578208129387, "grad_norm": 12.6484375, "learning_rate": 9.09554434337692e-06, "loss": 0.8444, "step": 25955 }, { "epoch": 2.7618642264311557, "grad_norm": 15.046875, "learning_rate": 9.095508811824901e-06, "loss": 1.0008, "step": 25956 }, { "epoch": 2.761970632049372, "grad_norm": 12.8984375, "learning_rate": 9.095473280272883e-06, "loss": 1.007, "step": 25957 }, { "epoch": 2.7620770376675887, "grad_norm": 11.4140625, "learning_rate": 9.095437748720864e-06, "loss": 1.0516, "step": 25958 }, { "epoch": 2.7621834432858057, "grad_norm": 21.015625, "learning_rate": 9.095402217168848e-06, "loss": 0.8807, "step": 25959 }, { "epoch": 2.762289848904022, "grad_norm": 15.328125, "learning_rate": 9.09536668561683e-06, "loss": 0.9296, "step": 25960 }, { "epoch": 2.7623962545222387, "grad_norm": 11.046875, "learning_rate": 9.09533115406481e-06, "loss": 0.9007, "step": 25961 }, { "epoch": 2.7625026601404556, "grad_norm": 12.0546875, "learning_rate": 9.095295622512792e-06, "loss": 0.9026, "step": 25962 }, { "epoch": 2.762609065758672, "grad_norm": 12.5546875, "learning_rate": 9.095260090960774e-06, "loss": 0.9543, "step": 25963 }, { "epoch": 2.7627154713768887, "grad_norm": 11.8046875, "learning_rate": 9.095224559408756e-06, "loss": 0.9493, "step": 25964 }, { "epoch": 2.7628218769951056, "grad_norm": 10.765625, "learning_rate": 9.095189027856737e-06, "loss": 0.9162, "step": 25965 }, { "epoch": 2.762928282613322, "grad_norm": 13.5078125, "learning_rate": 9.095153496304719e-06, "loss": 1.0111, "step": 25966 }, { "epoch": 2.7630346882315386, "grad_norm": 15.1484375, "learning_rate": 9.0951179647527e-06, "loss": 0.8829, "step": 25967 }, { "epoch": 2.763141093849755, "grad_norm": 11.25, "learning_rate": 9.095082433200682e-06, "loss": 0.798, "step": 25968 }, { "epoch": 2.7632474994679717, "grad_norm": 16.828125, "learning_rate": 9.095046901648664e-06, "loss": 0.9146, "step": 25969 }, { "epoch": 2.7633539050861886, "grad_norm": 18.171875, "learning_rate": 9.095011370096647e-06, "loss": 1.0147, "step": 25970 }, { "epoch": 2.763460310704405, "grad_norm": 10.859375, "learning_rate": 9.094975838544629e-06, "loss": 0.9966, "step": 25971 }, { "epoch": 2.7635667163226216, "grad_norm": 11.7734375, "learning_rate": 9.09494030699261e-06, "loss": 0.8381, "step": 25972 }, { "epoch": 2.7636731219408386, "grad_norm": 13.09375, "learning_rate": 9.094904775440592e-06, "loss": 0.9883, "step": 25973 }, { "epoch": 2.763779527559055, "grad_norm": 12.9375, "learning_rate": 9.094869243888573e-06, "loss": 1.0583, "step": 25974 }, { "epoch": 2.7638859331772716, "grad_norm": 17.046875, "learning_rate": 9.094833712336557e-06, "loss": 1.042, "step": 25975 }, { "epoch": 2.7639923387954886, "grad_norm": 12.4921875, "learning_rate": 9.094798180784538e-06, "loss": 0.869, "step": 25976 }, { "epoch": 2.764098744413705, "grad_norm": 12.1640625, "learning_rate": 9.09476264923252e-06, "loss": 0.6735, "step": 25977 }, { "epoch": 2.7642051500319216, "grad_norm": 13.4140625, "learning_rate": 9.094727117680502e-06, "loss": 0.8238, "step": 25978 }, { "epoch": 2.7643115556501385, "grad_norm": 13.515625, "learning_rate": 9.094691586128483e-06, "loss": 0.9181, "step": 25979 }, { "epoch": 2.764417961268355, "grad_norm": 13.046875, "learning_rate": 9.094656054576465e-06, "loss": 1.0648, "step": 25980 }, { "epoch": 2.7645243668865715, "grad_norm": 14.03125, "learning_rate": 9.094620523024446e-06, "loss": 0.7345, "step": 25981 }, { "epoch": 2.7646307725047885, "grad_norm": 13.203125, "learning_rate": 9.094584991472428e-06, "loss": 0.9598, "step": 25982 }, { "epoch": 2.764737178123005, "grad_norm": 12.4140625, "learning_rate": 9.09454945992041e-06, "loss": 0.885, "step": 25983 }, { "epoch": 2.7648435837412215, "grad_norm": 12.78125, "learning_rate": 9.094513928368391e-06, "loss": 0.9982, "step": 25984 }, { "epoch": 2.764949989359438, "grad_norm": 13.5703125, "learning_rate": 9.094478396816374e-06, "loss": 0.8049, "step": 25985 }, { "epoch": 2.765056394977655, "grad_norm": 10.96875, "learning_rate": 9.094442865264356e-06, "loss": 0.914, "step": 25986 }, { "epoch": 2.7651628005958715, "grad_norm": 10.4375, "learning_rate": 9.094407333712338e-06, "loss": 0.8524, "step": 25987 }, { "epoch": 2.765269206214088, "grad_norm": 11.6796875, "learning_rate": 9.09437180216032e-06, "loss": 1.0021, "step": 25988 }, { "epoch": 2.7653756118323045, "grad_norm": 12.109375, "learning_rate": 9.094336270608301e-06, "loss": 0.9678, "step": 25989 }, { "epoch": 2.7654820174505215, "grad_norm": 13.6875, "learning_rate": 9.094300739056283e-06, "loss": 0.8796, "step": 25990 }, { "epoch": 2.765588423068738, "grad_norm": 13.8203125, "learning_rate": 9.094265207504264e-06, "loss": 0.917, "step": 25991 }, { "epoch": 2.7656948286869545, "grad_norm": 11.90625, "learning_rate": 9.094229675952246e-06, "loss": 0.9919, "step": 25992 }, { "epoch": 2.7658012343051714, "grad_norm": 11.7109375, "learning_rate": 9.094194144400227e-06, "loss": 0.9166, "step": 25993 }, { "epoch": 2.765907639923388, "grad_norm": 10.1953125, "learning_rate": 9.094158612848209e-06, "loss": 0.9249, "step": 25994 }, { "epoch": 2.7660140455416045, "grad_norm": 12.421875, "learning_rate": 9.094123081296192e-06, "loss": 0.9233, "step": 25995 }, { "epoch": 2.7661204511598214, "grad_norm": 13.5, "learning_rate": 9.094087549744174e-06, "loss": 1.0085, "step": 25996 }, { "epoch": 2.766226856778038, "grad_norm": 13.3984375, "learning_rate": 9.094052018192155e-06, "loss": 0.9486, "step": 25997 }, { "epoch": 2.7663332623962544, "grad_norm": 13.421875, "learning_rate": 9.094016486640137e-06, "loss": 0.9179, "step": 25998 }, { "epoch": 2.7664396680144714, "grad_norm": 13.2734375, "learning_rate": 9.093980955088119e-06, "loss": 0.8128, "step": 25999 }, { "epoch": 2.766546073632688, "grad_norm": 13.6875, "learning_rate": 9.093945423536102e-06, "loss": 0.7904, "step": 26000 }, { "epoch": 2.7666524792509044, "grad_norm": 11.859375, "learning_rate": 9.093909891984084e-06, "loss": 1.0121, "step": 26001 }, { "epoch": 2.7667588848691214, "grad_norm": 13.34375, "learning_rate": 9.093874360432065e-06, "loss": 0.972, "step": 26002 }, { "epoch": 2.766865290487338, "grad_norm": 11.53125, "learning_rate": 9.093838828880047e-06, "loss": 0.8035, "step": 26003 }, { "epoch": 2.7669716961055544, "grad_norm": 13.3515625, "learning_rate": 9.093803297328028e-06, "loss": 0.7659, "step": 26004 }, { "epoch": 2.767078101723771, "grad_norm": 16.84375, "learning_rate": 9.09376776577601e-06, "loss": 0.9675, "step": 26005 }, { "epoch": 2.7671845073419874, "grad_norm": 12.6796875, "learning_rate": 9.093732234223992e-06, "loss": 0.9136, "step": 26006 }, { "epoch": 2.7672909129602044, "grad_norm": 10.703125, "learning_rate": 9.093696702671973e-06, "loss": 0.7674, "step": 26007 }, { "epoch": 2.767397318578421, "grad_norm": 15.4609375, "learning_rate": 9.093661171119955e-06, "loss": 0.9525, "step": 26008 }, { "epoch": 2.7675037241966374, "grad_norm": 16.328125, "learning_rate": 9.093625639567936e-06, "loss": 1.1759, "step": 26009 }, { "epoch": 2.7676101298148543, "grad_norm": 14.5, "learning_rate": 9.093590108015918e-06, "loss": 1.2662, "step": 26010 }, { "epoch": 2.767716535433071, "grad_norm": 11.9296875, "learning_rate": 9.093554576463901e-06, "loss": 0.8677, "step": 26011 }, { "epoch": 2.7678229410512873, "grad_norm": 15.6640625, "learning_rate": 9.093519044911883e-06, "loss": 0.9029, "step": 26012 }, { "epoch": 2.7679293466695043, "grad_norm": 19.71875, "learning_rate": 9.093483513359865e-06, "loss": 0.8298, "step": 26013 }, { "epoch": 2.768035752287721, "grad_norm": 20.921875, "learning_rate": 9.093447981807846e-06, "loss": 1.1116, "step": 26014 }, { "epoch": 2.7681421579059373, "grad_norm": 12.3203125, "learning_rate": 9.093412450255828e-06, "loss": 0.8778, "step": 26015 }, { "epoch": 2.7682485635241543, "grad_norm": 12.453125, "learning_rate": 9.09337691870381e-06, "loss": 0.9112, "step": 26016 }, { "epoch": 2.768354969142371, "grad_norm": 10.2109375, "learning_rate": 9.093341387151791e-06, "loss": 0.983, "step": 26017 }, { "epoch": 2.7684613747605873, "grad_norm": 14.703125, "learning_rate": 9.093305855599773e-06, "loss": 0.8038, "step": 26018 }, { "epoch": 2.7685677803788042, "grad_norm": 12.921875, "learning_rate": 9.093270324047754e-06, "loss": 1.0541, "step": 26019 }, { "epoch": 2.7686741859970208, "grad_norm": 13.671875, "learning_rate": 9.093234792495736e-06, "loss": 0.8852, "step": 26020 }, { "epoch": 2.7687805916152373, "grad_norm": 18.53125, "learning_rate": 9.093199260943719e-06, "loss": 1.1158, "step": 26021 }, { "epoch": 2.7688869972334538, "grad_norm": 11.4765625, "learning_rate": 9.0931637293917e-06, "loss": 0.9832, "step": 26022 }, { "epoch": 2.7689934028516707, "grad_norm": 10.265625, "learning_rate": 9.093128197839682e-06, "loss": 0.8635, "step": 26023 }, { "epoch": 2.7690998084698872, "grad_norm": 17.140625, "learning_rate": 9.093092666287664e-06, "loss": 0.8993, "step": 26024 }, { "epoch": 2.7692062140881037, "grad_norm": 13.1171875, "learning_rate": 9.093057134735646e-06, "loss": 0.7432, "step": 26025 }, { "epoch": 2.7693126197063203, "grad_norm": 13.140625, "learning_rate": 9.093021603183629e-06, "loss": 1.0125, "step": 26026 }, { "epoch": 2.769419025324537, "grad_norm": 10.9765625, "learning_rate": 9.09298607163161e-06, "loss": 0.8943, "step": 26027 }, { "epoch": 2.7695254309427537, "grad_norm": 13.6875, "learning_rate": 9.092950540079592e-06, "loss": 1.0135, "step": 26028 }, { "epoch": 2.7696318365609702, "grad_norm": 13.6875, "learning_rate": 9.092915008527574e-06, "loss": 0.8823, "step": 26029 }, { "epoch": 2.769738242179187, "grad_norm": 13.59375, "learning_rate": 9.092879476975555e-06, "loss": 0.8496, "step": 26030 }, { "epoch": 2.7698446477974037, "grad_norm": 14.328125, "learning_rate": 9.092843945423537e-06, "loss": 0.8497, "step": 26031 }, { "epoch": 2.76995105341562, "grad_norm": 20.40625, "learning_rate": 9.092808413871518e-06, "loss": 0.9798, "step": 26032 }, { "epoch": 2.770057459033837, "grad_norm": 18.625, "learning_rate": 9.0927728823195e-06, "loss": 1.2029, "step": 26033 }, { "epoch": 2.7701638646520537, "grad_norm": 15.375, "learning_rate": 9.092737350767482e-06, "loss": 0.8373, "step": 26034 }, { "epoch": 2.77027027027027, "grad_norm": 13.0546875, "learning_rate": 9.092701819215463e-06, "loss": 0.8945, "step": 26035 }, { "epoch": 2.770376675888487, "grad_norm": 12.2109375, "learning_rate": 9.092666287663447e-06, "loss": 1.0209, "step": 26036 }, { "epoch": 2.7704830815067036, "grad_norm": 12.1171875, "learning_rate": 9.092630756111428e-06, "loss": 0.8253, "step": 26037 }, { "epoch": 2.77058948712492, "grad_norm": 12.671875, "learning_rate": 9.09259522455941e-06, "loss": 0.8747, "step": 26038 }, { "epoch": 2.770695892743137, "grad_norm": 13.8984375, "learning_rate": 9.092559693007391e-06, "loss": 0.8959, "step": 26039 }, { "epoch": 2.7708022983613536, "grad_norm": 12.6640625, "learning_rate": 9.092524161455373e-06, "loss": 0.7804, "step": 26040 }, { "epoch": 2.77090870397957, "grad_norm": 13.53125, "learning_rate": 9.092488629903355e-06, "loss": 0.8929, "step": 26041 }, { "epoch": 2.7710151095977866, "grad_norm": 26.96875, "learning_rate": 9.092453098351336e-06, "loss": 1.1276, "step": 26042 }, { "epoch": 2.771121515216003, "grad_norm": 15.0703125, "learning_rate": 9.092417566799318e-06, "loss": 0.8736, "step": 26043 }, { "epoch": 2.77122792083422, "grad_norm": 19.609375, "learning_rate": 9.0923820352473e-06, "loss": 0.864, "step": 26044 }, { "epoch": 2.7713343264524366, "grad_norm": 13.2265625, "learning_rate": 9.092346503695281e-06, "loss": 0.9687, "step": 26045 }, { "epoch": 2.771440732070653, "grad_norm": 12.0390625, "learning_rate": 9.092310972143263e-06, "loss": 0.9946, "step": 26046 }, { "epoch": 2.77154713768887, "grad_norm": 19.40625, "learning_rate": 9.092275440591246e-06, "loss": 0.6526, "step": 26047 }, { "epoch": 2.7716535433070866, "grad_norm": 12.7421875, "learning_rate": 9.092239909039228e-06, "loss": 0.8888, "step": 26048 }, { "epoch": 2.771759948925303, "grad_norm": 13.0703125, "learning_rate": 9.09220437748721e-06, "loss": 1.0621, "step": 26049 }, { "epoch": 2.77186635454352, "grad_norm": 13.515625, "learning_rate": 9.09216884593519e-06, "loss": 0.8573, "step": 26050 }, { "epoch": 2.7719727601617365, "grad_norm": 25.359375, "learning_rate": 9.092133314383172e-06, "loss": 1.0625, "step": 26051 }, { "epoch": 2.772079165779953, "grad_norm": 13.9140625, "learning_rate": 9.092097782831156e-06, "loss": 0.8549, "step": 26052 }, { "epoch": 2.77218557139817, "grad_norm": 9.9140625, "learning_rate": 9.092062251279137e-06, "loss": 0.906, "step": 26053 }, { "epoch": 2.7722919770163865, "grad_norm": 16.453125, "learning_rate": 9.092026719727119e-06, "loss": 0.973, "step": 26054 }, { "epoch": 2.772398382634603, "grad_norm": 12.734375, "learning_rate": 9.0919911881751e-06, "loss": 0.8247, "step": 26055 }, { "epoch": 2.77250478825282, "grad_norm": 13.6953125, "learning_rate": 9.091955656623082e-06, "loss": 0.9691, "step": 26056 }, { "epoch": 2.7726111938710365, "grad_norm": 11.75, "learning_rate": 9.091920125071064e-06, "loss": 0.9491, "step": 26057 }, { "epoch": 2.772717599489253, "grad_norm": 10.8125, "learning_rate": 9.091884593519045e-06, "loss": 0.8723, "step": 26058 }, { "epoch": 2.7728240051074695, "grad_norm": 14.3828125, "learning_rate": 9.091849061967027e-06, "loss": 0.8414, "step": 26059 }, { "epoch": 2.7729304107256865, "grad_norm": 10.15625, "learning_rate": 9.091813530415009e-06, "loss": 0.902, "step": 26060 }, { "epoch": 2.773036816343903, "grad_norm": 14.2421875, "learning_rate": 9.09177799886299e-06, "loss": 1.0156, "step": 26061 }, { "epoch": 2.7731432219621195, "grad_norm": 19.6875, "learning_rate": 9.091742467310973e-06, "loss": 0.9439, "step": 26062 }, { "epoch": 2.773249627580336, "grad_norm": 14.109375, "learning_rate": 9.091706935758955e-06, "loss": 0.8752, "step": 26063 }, { "epoch": 2.773356033198553, "grad_norm": 15.6328125, "learning_rate": 9.091671404206937e-06, "loss": 0.9108, "step": 26064 }, { "epoch": 2.7734624388167695, "grad_norm": 15.6328125, "learning_rate": 9.091635872654918e-06, "loss": 0.8707, "step": 26065 }, { "epoch": 2.773568844434986, "grad_norm": 12.40625, "learning_rate": 9.0916003411029e-06, "loss": 0.9419, "step": 26066 }, { "epoch": 2.773675250053203, "grad_norm": 12.8671875, "learning_rate": 9.091564809550883e-06, "loss": 0.8381, "step": 26067 }, { "epoch": 2.7737816556714194, "grad_norm": 11.5, "learning_rate": 9.091529277998865e-06, "loss": 0.8694, "step": 26068 }, { "epoch": 2.773888061289636, "grad_norm": 9.9140625, "learning_rate": 9.091493746446846e-06, "loss": 0.7588, "step": 26069 }, { "epoch": 2.773994466907853, "grad_norm": 14.9765625, "learning_rate": 9.091458214894828e-06, "loss": 0.7474, "step": 26070 }, { "epoch": 2.7741008725260694, "grad_norm": 14.4453125, "learning_rate": 9.091422683342808e-06, "loss": 0.8448, "step": 26071 }, { "epoch": 2.774207278144286, "grad_norm": 12.5078125, "learning_rate": 9.091387151790791e-06, "loss": 0.7469, "step": 26072 }, { "epoch": 2.774313683762503, "grad_norm": 14.1015625, "learning_rate": 9.091351620238773e-06, "loss": 0.8021, "step": 26073 }, { "epoch": 2.7744200893807194, "grad_norm": 14.78125, "learning_rate": 9.091316088686754e-06, "loss": 0.9027, "step": 26074 }, { "epoch": 2.774526494998936, "grad_norm": 16.796875, "learning_rate": 9.091280557134736e-06, "loss": 0.887, "step": 26075 }, { "epoch": 2.774632900617153, "grad_norm": 13.25, "learning_rate": 9.091245025582718e-06, "loss": 0.8352, "step": 26076 }, { "epoch": 2.7747393062353694, "grad_norm": 12.59375, "learning_rate": 9.091209494030701e-06, "loss": 1.006, "step": 26077 }, { "epoch": 2.774845711853586, "grad_norm": 9.765625, "learning_rate": 9.091173962478683e-06, "loss": 0.8959, "step": 26078 }, { "epoch": 2.7749521174718024, "grad_norm": 11.609375, "learning_rate": 9.091138430926664e-06, "loss": 0.8308, "step": 26079 }, { "epoch": 2.775058523090019, "grad_norm": 12.140625, "learning_rate": 9.091102899374646e-06, "loss": 0.8051, "step": 26080 }, { "epoch": 2.775164928708236, "grad_norm": 11.1640625, "learning_rate": 9.091067367822627e-06, "loss": 0.8419, "step": 26081 }, { "epoch": 2.7752713343264523, "grad_norm": 13.109375, "learning_rate": 9.091031836270609e-06, "loss": 0.7993, "step": 26082 }, { "epoch": 2.775377739944669, "grad_norm": 18.375, "learning_rate": 9.09099630471859e-06, "loss": 1.0688, "step": 26083 }, { "epoch": 2.775484145562886, "grad_norm": 11.609375, "learning_rate": 9.090960773166572e-06, "loss": 0.8192, "step": 26084 }, { "epoch": 2.7755905511811023, "grad_norm": 12.1875, "learning_rate": 9.090925241614554e-06, "loss": 1.0767, "step": 26085 }, { "epoch": 2.775696956799319, "grad_norm": 16.15625, "learning_rate": 9.090889710062535e-06, "loss": 0.8912, "step": 26086 }, { "epoch": 2.775803362417536, "grad_norm": 14.2890625, "learning_rate": 9.090854178510517e-06, "loss": 0.993, "step": 26087 }, { "epoch": 2.7759097680357523, "grad_norm": 16.453125, "learning_rate": 9.0908186469585e-06, "loss": 1.0241, "step": 26088 }, { "epoch": 2.776016173653969, "grad_norm": 18.5625, "learning_rate": 9.090783115406482e-06, "loss": 1.0532, "step": 26089 }, { "epoch": 2.7761225792721858, "grad_norm": 12.84375, "learning_rate": 9.090747583854464e-06, "loss": 0.939, "step": 26090 }, { "epoch": 2.7762289848904023, "grad_norm": 12.546875, "learning_rate": 9.090712052302445e-06, "loss": 0.9956, "step": 26091 }, { "epoch": 2.7763353905086188, "grad_norm": 18.265625, "learning_rate": 9.090676520750427e-06, "loss": 0.7587, "step": 26092 }, { "epoch": 2.7764417961268357, "grad_norm": 16.21875, "learning_rate": 9.09064098919841e-06, "loss": 0.8085, "step": 26093 }, { "epoch": 2.7765482017450522, "grad_norm": 16.0625, "learning_rate": 9.090605457646392e-06, "loss": 0.6645, "step": 26094 }, { "epoch": 2.7766546073632687, "grad_norm": 15.6953125, "learning_rate": 9.090569926094373e-06, "loss": 0.8487, "step": 26095 }, { "epoch": 2.7767610129814857, "grad_norm": 13.234375, "learning_rate": 9.090534394542355e-06, "loss": 0.9112, "step": 26096 }, { "epoch": 2.776867418599702, "grad_norm": 14.5859375, "learning_rate": 9.090498862990336e-06, "loss": 1.0387, "step": 26097 }, { "epoch": 2.7769738242179187, "grad_norm": 11.59375, "learning_rate": 9.090463331438318e-06, "loss": 0.9649, "step": 26098 }, { "epoch": 2.7770802298361352, "grad_norm": 11.8125, "learning_rate": 9.0904277998863e-06, "loss": 0.9357, "step": 26099 }, { "epoch": 2.7771866354543517, "grad_norm": 11.3984375, "learning_rate": 9.090392268334281e-06, "loss": 0.8502, "step": 26100 }, { "epoch": 2.7772930410725687, "grad_norm": 10.6015625, "learning_rate": 9.090356736782263e-06, "loss": 0.8474, "step": 26101 }, { "epoch": 2.777399446690785, "grad_norm": 16.140625, "learning_rate": 9.090321205230244e-06, "loss": 0.9613, "step": 26102 }, { "epoch": 2.7775058523090017, "grad_norm": 17.75, "learning_rate": 9.090285673678228e-06, "loss": 0.958, "step": 26103 }, { "epoch": 2.7776122579272187, "grad_norm": 10.125, "learning_rate": 9.09025014212621e-06, "loss": 0.7819, "step": 26104 }, { "epoch": 2.777718663545435, "grad_norm": 10.3984375, "learning_rate": 9.090214610574191e-06, "loss": 0.935, "step": 26105 }, { "epoch": 2.7778250691636517, "grad_norm": 15.5234375, "learning_rate": 9.090179079022173e-06, "loss": 0.8979, "step": 26106 }, { "epoch": 2.7779314747818686, "grad_norm": 15.109375, "learning_rate": 9.090143547470154e-06, "loss": 1.0521, "step": 26107 }, { "epoch": 2.778037880400085, "grad_norm": 17.765625, "learning_rate": 9.090108015918136e-06, "loss": 0.8447, "step": 26108 }, { "epoch": 2.7781442860183017, "grad_norm": 12.953125, "learning_rate": 9.090072484366117e-06, "loss": 0.8843, "step": 26109 }, { "epoch": 2.7782506916365186, "grad_norm": 17.015625, "learning_rate": 9.090036952814099e-06, "loss": 0.9429, "step": 26110 }, { "epoch": 2.778357097254735, "grad_norm": 10.1328125, "learning_rate": 9.09000142126208e-06, "loss": 0.8838, "step": 26111 }, { "epoch": 2.7784635028729516, "grad_norm": 14.234375, "learning_rate": 9.089965889710062e-06, "loss": 0.9092, "step": 26112 }, { "epoch": 2.7785699084911686, "grad_norm": 13.9453125, "learning_rate": 9.089930358158046e-06, "loss": 0.857, "step": 26113 }, { "epoch": 2.778676314109385, "grad_norm": 12.3125, "learning_rate": 9.089894826606027e-06, "loss": 0.8652, "step": 26114 }, { "epoch": 2.7787827197276016, "grad_norm": 12.1640625, "learning_rate": 9.089859295054009e-06, "loss": 1.0063, "step": 26115 }, { "epoch": 2.778889125345818, "grad_norm": 12.4921875, "learning_rate": 9.08982376350199e-06, "loss": 0.9408, "step": 26116 }, { "epoch": 2.778995530964035, "grad_norm": 17.75, "learning_rate": 9.089788231949972e-06, "loss": 0.9394, "step": 26117 }, { "epoch": 2.7791019365822516, "grad_norm": 11.3515625, "learning_rate": 9.089752700397955e-06, "loss": 0.9482, "step": 26118 }, { "epoch": 2.779208342200468, "grad_norm": 11.671875, "learning_rate": 9.089717168845937e-06, "loss": 1.0354, "step": 26119 }, { "epoch": 2.7793147478186846, "grad_norm": 11.3046875, "learning_rate": 9.089681637293918e-06, "loss": 0.8834, "step": 26120 }, { "epoch": 2.7794211534369015, "grad_norm": 11.6484375, "learning_rate": 9.0896461057419e-06, "loss": 0.7858, "step": 26121 }, { "epoch": 2.779527559055118, "grad_norm": 11.453125, "learning_rate": 9.089610574189882e-06, "loss": 0.7782, "step": 26122 }, { "epoch": 2.7796339646733346, "grad_norm": 13.90625, "learning_rate": 9.089575042637863e-06, "loss": 0.954, "step": 26123 }, { "epoch": 2.7797403702915515, "grad_norm": 12.6015625, "learning_rate": 9.089539511085845e-06, "loss": 0.9153, "step": 26124 }, { "epoch": 2.779846775909768, "grad_norm": 12.6640625, "learning_rate": 9.089503979533827e-06, "loss": 0.8489, "step": 26125 }, { "epoch": 2.7799531815279845, "grad_norm": 13.3515625, "learning_rate": 9.089468447981808e-06, "loss": 0.8545, "step": 26126 }, { "epoch": 2.7800595871462015, "grad_norm": 13.671875, "learning_rate": 9.08943291642979e-06, "loss": 0.8079, "step": 26127 }, { "epoch": 2.780165992764418, "grad_norm": 11.53125, "learning_rate": 9.089397384877771e-06, "loss": 0.9581, "step": 26128 }, { "epoch": 2.7802723983826345, "grad_norm": 11.625, "learning_rate": 9.089361853325755e-06, "loss": 0.9972, "step": 26129 }, { "epoch": 2.7803788040008515, "grad_norm": 11.3515625, "learning_rate": 9.089326321773736e-06, "loss": 0.9468, "step": 26130 }, { "epoch": 2.780485209619068, "grad_norm": 16.109375, "learning_rate": 9.089290790221718e-06, "loss": 0.8689, "step": 26131 }, { "epoch": 2.7805916152372845, "grad_norm": 12.140625, "learning_rate": 9.0892552586697e-06, "loss": 0.8936, "step": 26132 }, { "epoch": 2.7806980208555014, "grad_norm": 15.375, "learning_rate": 9.089219727117681e-06, "loss": 0.9939, "step": 26133 }, { "epoch": 2.780804426473718, "grad_norm": 13.9453125, "learning_rate": 9.089184195565663e-06, "loss": 0.9816, "step": 26134 }, { "epoch": 2.7809108320919345, "grad_norm": 12.953125, "learning_rate": 9.089148664013644e-06, "loss": 0.9213, "step": 26135 }, { "epoch": 2.781017237710151, "grad_norm": 11.890625, "learning_rate": 9.089113132461626e-06, "loss": 0.8523, "step": 26136 }, { "epoch": 2.7811236433283675, "grad_norm": 14.5546875, "learning_rate": 9.089077600909607e-06, "loss": 0.9074, "step": 26137 }, { "epoch": 2.7812300489465844, "grad_norm": 16.0625, "learning_rate": 9.089042069357589e-06, "loss": 1.0631, "step": 26138 }, { "epoch": 2.781336454564801, "grad_norm": 12.390625, "learning_rate": 9.089006537805572e-06, "loss": 0.9373, "step": 26139 }, { "epoch": 2.7814428601830175, "grad_norm": 10.59375, "learning_rate": 9.088971006253554e-06, "loss": 0.9873, "step": 26140 }, { "epoch": 2.7815492658012344, "grad_norm": 10.390625, "learning_rate": 9.088935474701536e-06, "loss": 0.8208, "step": 26141 }, { "epoch": 2.781655671419451, "grad_norm": 13.2890625, "learning_rate": 9.088899943149517e-06, "loss": 1.0888, "step": 26142 }, { "epoch": 2.7817620770376674, "grad_norm": 13.25, "learning_rate": 9.088864411597499e-06, "loss": 0.7758, "step": 26143 }, { "epoch": 2.7818684826558844, "grad_norm": 12.5703125, "learning_rate": 9.088828880045482e-06, "loss": 1.0675, "step": 26144 }, { "epoch": 2.781974888274101, "grad_norm": 11.140625, "learning_rate": 9.088793348493464e-06, "loss": 0.927, "step": 26145 }, { "epoch": 2.7820812938923174, "grad_norm": 13.109375, "learning_rate": 9.088757816941445e-06, "loss": 0.947, "step": 26146 }, { "epoch": 2.7821876995105344, "grad_norm": 16.171875, "learning_rate": 9.088722285389427e-06, "loss": 1.1219, "step": 26147 }, { "epoch": 2.782294105128751, "grad_norm": 10.53125, "learning_rate": 9.088686753837409e-06, "loss": 0.9057, "step": 26148 }, { "epoch": 2.7824005107469674, "grad_norm": 15.9296875, "learning_rate": 9.08865122228539e-06, "loss": 0.9079, "step": 26149 }, { "epoch": 2.7825069163651843, "grad_norm": 15.6875, "learning_rate": 9.088615690733372e-06, "loss": 0.8692, "step": 26150 }, { "epoch": 2.782613321983401, "grad_norm": 12.1484375, "learning_rate": 9.088580159181353e-06, "loss": 0.9912, "step": 26151 }, { "epoch": 2.7827197276016173, "grad_norm": 14.34375, "learning_rate": 9.088544627629335e-06, "loss": 1.0564, "step": 26152 }, { "epoch": 2.782826133219834, "grad_norm": 12.4609375, "learning_rate": 9.088509096077317e-06, "loss": 0.8762, "step": 26153 }, { "epoch": 2.782932538838051, "grad_norm": 13.9921875, "learning_rate": 9.0884735645253e-06, "loss": 0.9667, "step": 26154 }, { "epoch": 2.7830389444562673, "grad_norm": 12.140625, "learning_rate": 9.088438032973282e-06, "loss": 0.8045, "step": 26155 }, { "epoch": 2.783145350074484, "grad_norm": 13.671875, "learning_rate": 9.088402501421263e-06, "loss": 0.7827, "step": 26156 }, { "epoch": 2.7832517556927003, "grad_norm": 11.3515625, "learning_rate": 9.088366969869245e-06, "loss": 1.0269, "step": 26157 }, { "epoch": 2.7833581613109173, "grad_norm": 13.2890625, "learning_rate": 9.088331438317226e-06, "loss": 0.8675, "step": 26158 }, { "epoch": 2.783464566929134, "grad_norm": 14.75, "learning_rate": 9.08829590676521e-06, "loss": 0.9789, "step": 26159 }, { "epoch": 2.7835709725473503, "grad_norm": 16.125, "learning_rate": 9.08826037521319e-06, "loss": 1.1393, "step": 26160 }, { "epoch": 2.7836773781655673, "grad_norm": 13.09375, "learning_rate": 9.088224843661171e-06, "loss": 0.9453, "step": 26161 }, { "epoch": 2.7837837837837838, "grad_norm": 11.796875, "learning_rate": 9.088189312109153e-06, "loss": 0.8199, "step": 26162 }, { "epoch": 2.7838901894020003, "grad_norm": 11.9765625, "learning_rate": 9.088153780557134e-06, "loss": 0.7815, "step": 26163 }, { "epoch": 2.7839965950202172, "grad_norm": 16.421875, "learning_rate": 9.088118249005116e-06, "loss": 0.8388, "step": 26164 }, { "epoch": 2.7841030006384337, "grad_norm": 9.9296875, "learning_rate": 9.0880827174531e-06, "loss": 0.8973, "step": 26165 }, { "epoch": 2.7842094062566503, "grad_norm": 10.8046875, "learning_rate": 9.088047185901081e-06, "loss": 0.9451, "step": 26166 }, { "epoch": 2.784315811874867, "grad_norm": 13.6015625, "learning_rate": 9.088011654349062e-06, "loss": 0.9427, "step": 26167 }, { "epoch": 2.7844222174930837, "grad_norm": 15.953125, "learning_rate": 9.087976122797044e-06, "loss": 1.0175, "step": 26168 }, { "epoch": 2.7845286231113002, "grad_norm": 13.390625, "learning_rate": 9.087940591245026e-06, "loss": 0.8077, "step": 26169 }, { "epoch": 2.784635028729517, "grad_norm": 15.7109375, "learning_rate": 9.087905059693009e-06, "loss": 0.9949, "step": 26170 }, { "epoch": 2.7847414343477337, "grad_norm": 13.671875, "learning_rate": 9.08786952814099e-06, "loss": 0.9208, "step": 26171 }, { "epoch": 2.78484783996595, "grad_norm": 12.1796875, "learning_rate": 9.087833996588972e-06, "loss": 1.1021, "step": 26172 }, { "epoch": 2.7849542455841667, "grad_norm": 12.875, "learning_rate": 9.087798465036954e-06, "loss": 0.8236, "step": 26173 }, { "epoch": 2.785060651202383, "grad_norm": 17.203125, "learning_rate": 9.087762933484935e-06, "loss": 0.9782, "step": 26174 }, { "epoch": 2.7851670568206, "grad_norm": 15.03125, "learning_rate": 9.087727401932917e-06, "loss": 0.8343, "step": 26175 }, { "epoch": 2.7852734624388167, "grad_norm": 12.90625, "learning_rate": 9.087691870380899e-06, "loss": 1.0369, "step": 26176 }, { "epoch": 2.785379868057033, "grad_norm": 9.0078125, "learning_rate": 9.08765633882888e-06, "loss": 0.7341, "step": 26177 }, { "epoch": 2.78548627367525, "grad_norm": 14.53125, "learning_rate": 9.087620807276862e-06, "loss": 0.963, "step": 26178 }, { "epoch": 2.7855926792934667, "grad_norm": 17.09375, "learning_rate": 9.087585275724843e-06, "loss": 0.8631, "step": 26179 }, { "epoch": 2.785699084911683, "grad_norm": 13.453125, "learning_rate": 9.087549744172827e-06, "loss": 0.8833, "step": 26180 }, { "epoch": 2.7858054905299, "grad_norm": 13.6640625, "learning_rate": 9.087514212620808e-06, "loss": 0.9131, "step": 26181 }, { "epoch": 2.7859118961481166, "grad_norm": 14.625, "learning_rate": 9.08747868106879e-06, "loss": 0.9725, "step": 26182 }, { "epoch": 2.786018301766333, "grad_norm": 14.9453125, "learning_rate": 9.087443149516772e-06, "loss": 0.8098, "step": 26183 }, { "epoch": 2.78612470738455, "grad_norm": 14.5234375, "learning_rate": 9.087407617964753e-06, "loss": 0.8734, "step": 26184 }, { "epoch": 2.7862311130027666, "grad_norm": 11.6328125, "learning_rate": 9.087372086412736e-06, "loss": 0.9994, "step": 26185 }, { "epoch": 2.786337518620983, "grad_norm": 11.875, "learning_rate": 9.087336554860718e-06, "loss": 0.8931, "step": 26186 }, { "epoch": 2.7864439242392, "grad_norm": 12.7109375, "learning_rate": 9.0873010233087e-06, "loss": 1.0136, "step": 26187 }, { "epoch": 2.7865503298574166, "grad_norm": 10.984375, "learning_rate": 9.087265491756681e-06, "loss": 0.9058, "step": 26188 }, { "epoch": 2.786656735475633, "grad_norm": 17.921875, "learning_rate": 9.087229960204663e-06, "loss": 1.2122, "step": 26189 }, { "epoch": 2.7867631410938496, "grad_norm": 10.234375, "learning_rate": 9.087194428652645e-06, "loss": 0.7497, "step": 26190 }, { "epoch": 2.7868695467120665, "grad_norm": 15.078125, "learning_rate": 9.087158897100626e-06, "loss": 0.8062, "step": 26191 }, { "epoch": 2.786975952330283, "grad_norm": 14.171875, "learning_rate": 9.087123365548608e-06, "loss": 0.99, "step": 26192 }, { "epoch": 2.7870823579484996, "grad_norm": 13.6015625, "learning_rate": 9.08708783399659e-06, "loss": 0.8785, "step": 26193 }, { "epoch": 2.787188763566716, "grad_norm": 10.671875, "learning_rate": 9.087052302444571e-06, "loss": 0.8974, "step": 26194 }, { "epoch": 2.787295169184933, "grad_norm": 13.28125, "learning_rate": 9.087016770892554e-06, "loss": 0.8325, "step": 26195 }, { "epoch": 2.7874015748031495, "grad_norm": 12.3671875, "learning_rate": 9.086981239340536e-06, "loss": 0.9334, "step": 26196 }, { "epoch": 2.787507980421366, "grad_norm": 11.375, "learning_rate": 9.086945707788517e-06, "loss": 0.8445, "step": 26197 }, { "epoch": 2.787614386039583, "grad_norm": 19.359375, "learning_rate": 9.086910176236499e-06, "loss": 1.0694, "step": 26198 }, { "epoch": 2.7877207916577995, "grad_norm": 13.1796875, "learning_rate": 9.08687464468448e-06, "loss": 1.0139, "step": 26199 }, { "epoch": 2.787827197276016, "grad_norm": 17.984375, "learning_rate": 9.086839113132462e-06, "loss": 0.968, "step": 26200 }, { "epoch": 2.787933602894233, "grad_norm": 15.4609375, "learning_rate": 9.086803581580444e-06, "loss": 0.9399, "step": 26201 }, { "epoch": 2.7880400085124495, "grad_norm": 14.3046875, "learning_rate": 9.086768050028425e-06, "loss": 0.881, "step": 26202 }, { "epoch": 2.788146414130666, "grad_norm": 9.90625, "learning_rate": 9.086732518476407e-06, "loss": 0.8372, "step": 26203 }, { "epoch": 2.788252819748883, "grad_norm": 11.3828125, "learning_rate": 9.086696986924389e-06, "loss": 1.0708, "step": 26204 }, { "epoch": 2.7883592253670995, "grad_norm": 13.1640625, "learning_rate": 9.08666145537237e-06, "loss": 0.8695, "step": 26205 }, { "epoch": 2.788465630985316, "grad_norm": 11.0703125, "learning_rate": 9.086625923820354e-06, "loss": 0.9801, "step": 26206 }, { "epoch": 2.788572036603533, "grad_norm": 15.75, "learning_rate": 9.086590392268335e-06, "loss": 0.8261, "step": 26207 }, { "epoch": 2.7886784422217494, "grad_norm": 12.4453125, "learning_rate": 9.086554860716317e-06, "loss": 0.9952, "step": 26208 }, { "epoch": 2.788784847839966, "grad_norm": 14.296875, "learning_rate": 9.086519329164298e-06, "loss": 1.0372, "step": 26209 }, { "epoch": 2.7888912534581825, "grad_norm": 11.2265625, "learning_rate": 9.08648379761228e-06, "loss": 0.7789, "step": 26210 }, { "epoch": 2.788997659076399, "grad_norm": 14.6796875, "learning_rate": 9.086448266060263e-06, "loss": 1.0659, "step": 26211 }, { "epoch": 2.789104064694616, "grad_norm": 17.671875, "learning_rate": 9.086412734508245e-06, "loss": 0.9036, "step": 26212 }, { "epoch": 2.7892104703128324, "grad_norm": 24.25, "learning_rate": 9.086377202956227e-06, "loss": 1.0729, "step": 26213 }, { "epoch": 2.789316875931049, "grad_norm": 14.1328125, "learning_rate": 9.086341671404208e-06, "loss": 1.0456, "step": 26214 }, { "epoch": 2.789423281549266, "grad_norm": 13.5078125, "learning_rate": 9.08630613985219e-06, "loss": 1.1676, "step": 26215 }, { "epoch": 2.7895296871674824, "grad_norm": 17.0625, "learning_rate": 9.086270608300171e-06, "loss": 0.9433, "step": 26216 }, { "epoch": 2.789636092785699, "grad_norm": 14.5546875, "learning_rate": 9.086235076748153e-06, "loss": 0.8336, "step": 26217 }, { "epoch": 2.789742498403916, "grad_norm": 9.203125, "learning_rate": 9.086199545196135e-06, "loss": 0.6691, "step": 26218 }, { "epoch": 2.7898489040221324, "grad_norm": 11.625, "learning_rate": 9.086164013644116e-06, "loss": 0.9255, "step": 26219 }, { "epoch": 2.789955309640349, "grad_norm": 13.796875, "learning_rate": 9.086128482092098e-06, "loss": 0.9797, "step": 26220 }, { "epoch": 2.790061715258566, "grad_norm": 16.359375, "learning_rate": 9.086092950540081e-06, "loss": 1.1493, "step": 26221 }, { "epoch": 2.7901681208767823, "grad_norm": 15.328125, "learning_rate": 9.086057418988063e-06, "loss": 0.9227, "step": 26222 }, { "epoch": 2.790274526494999, "grad_norm": 12.234375, "learning_rate": 9.086021887436044e-06, "loss": 0.9641, "step": 26223 }, { "epoch": 2.790380932113216, "grad_norm": 12.4296875, "learning_rate": 9.085986355884026e-06, "loss": 0.8563, "step": 26224 }, { "epoch": 2.7904873377314323, "grad_norm": 11.3828125, "learning_rate": 9.085950824332008e-06, "loss": 0.8315, "step": 26225 }, { "epoch": 2.790593743349649, "grad_norm": 12.4921875, "learning_rate": 9.085915292779989e-06, "loss": 0.917, "step": 26226 }, { "epoch": 2.790700148967866, "grad_norm": 14.4375, "learning_rate": 9.08587976122797e-06, "loss": 0.9409, "step": 26227 }, { "epoch": 2.7908065545860823, "grad_norm": 11.75, "learning_rate": 9.085844229675952e-06, "loss": 0.912, "step": 26228 }, { "epoch": 2.790912960204299, "grad_norm": 13.7734375, "learning_rate": 9.085808698123934e-06, "loss": 0.9383, "step": 26229 }, { "epoch": 2.7910193658225153, "grad_norm": 14.1875, "learning_rate": 9.085773166571916e-06, "loss": 0.8661, "step": 26230 }, { "epoch": 2.791125771440732, "grad_norm": 17.234375, "learning_rate": 9.085737635019899e-06, "loss": 1.0281, "step": 26231 }, { "epoch": 2.7912321770589488, "grad_norm": 11.78125, "learning_rate": 9.08570210346788e-06, "loss": 0.9551, "step": 26232 }, { "epoch": 2.7913385826771653, "grad_norm": 17.953125, "learning_rate": 9.085666571915862e-06, "loss": 0.875, "step": 26233 }, { "epoch": 2.791444988295382, "grad_norm": 11.5625, "learning_rate": 9.085631040363844e-06, "loss": 0.8179, "step": 26234 }, { "epoch": 2.7915513939135987, "grad_norm": 10.2109375, "learning_rate": 9.085595508811825e-06, "loss": 0.8323, "step": 26235 }, { "epoch": 2.7916577995318153, "grad_norm": 11.7890625, "learning_rate": 9.085559977259809e-06, "loss": 0.8733, "step": 26236 }, { "epoch": 2.7917642051500318, "grad_norm": 14.4296875, "learning_rate": 9.08552444570779e-06, "loss": 0.9257, "step": 26237 }, { "epoch": 2.7918706107682487, "grad_norm": 10.984375, "learning_rate": 9.085488914155772e-06, "loss": 0.8558, "step": 26238 }, { "epoch": 2.7919770163864652, "grad_norm": 14.8515625, "learning_rate": 9.085453382603753e-06, "loss": 0.8943, "step": 26239 }, { "epoch": 2.7920834220046817, "grad_norm": 12.9140625, "learning_rate": 9.085417851051735e-06, "loss": 1.1046, "step": 26240 }, { "epoch": 2.7921898276228987, "grad_norm": 12.234375, "learning_rate": 9.085382319499717e-06, "loss": 1.0349, "step": 26241 }, { "epoch": 2.792296233241115, "grad_norm": 11.765625, "learning_rate": 9.085346787947698e-06, "loss": 0.8516, "step": 26242 }, { "epoch": 2.7924026388593317, "grad_norm": 16.828125, "learning_rate": 9.08531125639568e-06, "loss": 0.8558, "step": 26243 }, { "epoch": 2.7925090444775487, "grad_norm": 16.90625, "learning_rate": 9.085275724843661e-06, "loss": 1.0348, "step": 26244 }, { "epoch": 2.792615450095765, "grad_norm": 11.03125, "learning_rate": 9.085240193291643e-06, "loss": 0.801, "step": 26245 }, { "epoch": 2.7927218557139817, "grad_norm": 18.90625, "learning_rate": 9.085204661739625e-06, "loss": 0.9522, "step": 26246 }, { "epoch": 2.792828261332198, "grad_norm": 10.6484375, "learning_rate": 9.085169130187608e-06, "loss": 0.873, "step": 26247 }, { "epoch": 2.792934666950415, "grad_norm": 13.8203125, "learning_rate": 9.08513359863559e-06, "loss": 0.8847, "step": 26248 }, { "epoch": 2.7930410725686317, "grad_norm": 14.890625, "learning_rate": 9.085098067083571e-06, "loss": 0.9996, "step": 26249 }, { "epoch": 2.793147478186848, "grad_norm": 13.0, "learning_rate": 9.085062535531553e-06, "loss": 0.8312, "step": 26250 }, { "epoch": 2.7932538838050647, "grad_norm": 12.4609375, "learning_rate": 9.085027003979534e-06, "loss": 0.948, "step": 26251 }, { "epoch": 2.7933602894232816, "grad_norm": 13.703125, "learning_rate": 9.084991472427516e-06, "loss": 1.0344, "step": 26252 }, { "epoch": 2.793466695041498, "grad_norm": 15.3125, "learning_rate": 9.084955940875498e-06, "loss": 0.9587, "step": 26253 }, { "epoch": 2.7935731006597146, "grad_norm": 16.890625, "learning_rate": 9.08492040932348e-06, "loss": 1.1955, "step": 26254 }, { "epoch": 2.7936795062779316, "grad_norm": 13.1953125, "learning_rate": 9.08488487777146e-06, "loss": 1.0498, "step": 26255 }, { "epoch": 2.793785911896148, "grad_norm": 12.09375, "learning_rate": 9.084849346219442e-06, "loss": 0.8381, "step": 26256 }, { "epoch": 2.7938923175143646, "grad_norm": 14.78125, "learning_rate": 9.084813814667426e-06, "loss": 1.0821, "step": 26257 }, { "epoch": 2.7939987231325816, "grad_norm": 11.4765625, "learning_rate": 9.084778283115407e-06, "loss": 0.8976, "step": 26258 }, { "epoch": 2.794105128750798, "grad_norm": 11.7578125, "learning_rate": 9.084742751563389e-06, "loss": 1.0611, "step": 26259 }, { "epoch": 2.7942115343690146, "grad_norm": 14.0703125, "learning_rate": 9.08470722001137e-06, "loss": 0.8992, "step": 26260 }, { "epoch": 2.7943179399872315, "grad_norm": 14.171875, "learning_rate": 9.084671688459352e-06, "loss": 0.923, "step": 26261 }, { "epoch": 2.794424345605448, "grad_norm": 12.8359375, "learning_rate": 9.084636156907335e-06, "loss": 0.9406, "step": 26262 }, { "epoch": 2.7945307512236646, "grad_norm": 14.203125, "learning_rate": 9.084600625355317e-06, "loss": 1.0952, "step": 26263 }, { "epoch": 2.7946371568418815, "grad_norm": 12.0625, "learning_rate": 9.084565093803299e-06, "loss": 0.8854, "step": 26264 }, { "epoch": 2.794743562460098, "grad_norm": 12.796875, "learning_rate": 9.08452956225128e-06, "loss": 0.9633, "step": 26265 }, { "epoch": 2.7948499680783145, "grad_norm": 17.34375, "learning_rate": 9.084494030699262e-06, "loss": 1.0811, "step": 26266 }, { "epoch": 2.794956373696531, "grad_norm": 9.0703125, "learning_rate": 9.084458499147243e-06, "loss": 0.7565, "step": 26267 }, { "epoch": 2.7950627793147476, "grad_norm": 9.2578125, "learning_rate": 9.084422967595225e-06, "loss": 0.9583, "step": 26268 }, { "epoch": 2.7951691849329645, "grad_norm": 13.3984375, "learning_rate": 9.084387436043207e-06, "loss": 0.891, "step": 26269 }, { "epoch": 2.795275590551181, "grad_norm": 10.6796875, "learning_rate": 9.084351904491188e-06, "loss": 0.815, "step": 26270 }, { "epoch": 2.7953819961693975, "grad_norm": 20.234375, "learning_rate": 9.08431637293917e-06, "loss": 0.8528, "step": 26271 }, { "epoch": 2.7954884017876145, "grad_norm": 12.0859375, "learning_rate": 9.084280841387153e-06, "loss": 1.0907, "step": 26272 }, { "epoch": 2.795594807405831, "grad_norm": 12.140625, "learning_rate": 9.084245309835135e-06, "loss": 1.0352, "step": 26273 }, { "epoch": 2.7957012130240475, "grad_norm": 14.6484375, "learning_rate": 9.084209778283116e-06, "loss": 0.9931, "step": 26274 }, { "epoch": 2.7958076186422645, "grad_norm": 14.5078125, "learning_rate": 9.084174246731098e-06, "loss": 1.0506, "step": 26275 }, { "epoch": 2.795914024260481, "grad_norm": 10.0625, "learning_rate": 9.08413871517908e-06, "loss": 0.9817, "step": 26276 }, { "epoch": 2.7960204298786975, "grad_norm": 12.46875, "learning_rate": 9.084103183627063e-06, "loss": 1.0476, "step": 26277 }, { "epoch": 2.7961268354969144, "grad_norm": 11.578125, "learning_rate": 9.084067652075045e-06, "loss": 0.8279, "step": 26278 }, { "epoch": 2.796233241115131, "grad_norm": 12.3359375, "learning_rate": 9.084032120523026e-06, "loss": 1.0114, "step": 26279 }, { "epoch": 2.7963396467333475, "grad_norm": 13.8359375, "learning_rate": 9.083996588971006e-06, "loss": 0.8376, "step": 26280 }, { "epoch": 2.7964460523515644, "grad_norm": 11.7734375, "learning_rate": 9.083961057418988e-06, "loss": 0.9081, "step": 26281 }, { "epoch": 2.796552457969781, "grad_norm": 10.2109375, "learning_rate": 9.08392552586697e-06, "loss": 0.8354, "step": 26282 }, { "epoch": 2.7966588635879974, "grad_norm": 14.2109375, "learning_rate": 9.083889994314953e-06, "loss": 0.7891, "step": 26283 }, { "epoch": 2.796765269206214, "grad_norm": 12.5703125, "learning_rate": 9.083854462762934e-06, "loss": 1.0353, "step": 26284 }, { "epoch": 2.796871674824431, "grad_norm": 19.296875, "learning_rate": 9.083818931210916e-06, "loss": 0.9492, "step": 26285 }, { "epoch": 2.7969780804426474, "grad_norm": 9.6171875, "learning_rate": 9.083783399658897e-06, "loss": 0.783, "step": 26286 }, { "epoch": 2.797084486060864, "grad_norm": 14.2734375, "learning_rate": 9.083747868106879e-06, "loss": 1.0227, "step": 26287 }, { "epoch": 2.7971908916790804, "grad_norm": 10.0234375, "learning_rate": 9.083712336554862e-06, "loss": 0.8452, "step": 26288 }, { "epoch": 2.7972972972972974, "grad_norm": 14.4375, "learning_rate": 9.083676805002844e-06, "loss": 0.9157, "step": 26289 }, { "epoch": 2.797403702915514, "grad_norm": 18.703125, "learning_rate": 9.083641273450826e-06, "loss": 1.0193, "step": 26290 }, { "epoch": 2.7975101085337304, "grad_norm": 11.9609375, "learning_rate": 9.083605741898807e-06, "loss": 0.8953, "step": 26291 }, { "epoch": 2.7976165141519473, "grad_norm": 10.4375, "learning_rate": 9.083570210346789e-06, "loss": 1.0669, "step": 26292 }, { "epoch": 2.797722919770164, "grad_norm": 12.1171875, "learning_rate": 9.08353467879477e-06, "loss": 1.0166, "step": 26293 }, { "epoch": 2.7978293253883804, "grad_norm": 20.203125, "learning_rate": 9.083499147242752e-06, "loss": 1.116, "step": 26294 }, { "epoch": 2.7979357310065973, "grad_norm": 13.90625, "learning_rate": 9.083463615690734e-06, "loss": 0.9593, "step": 26295 }, { "epoch": 2.798042136624814, "grad_norm": 13.796875, "learning_rate": 9.083428084138715e-06, "loss": 1.1297, "step": 26296 }, { "epoch": 2.7981485422430303, "grad_norm": 22.734375, "learning_rate": 9.083392552586697e-06, "loss": 0.8625, "step": 26297 }, { "epoch": 2.7982549478612473, "grad_norm": 14.7578125, "learning_rate": 9.08335702103468e-06, "loss": 0.7698, "step": 26298 }, { "epoch": 2.798361353479464, "grad_norm": 12.0703125, "learning_rate": 9.083321489482662e-06, "loss": 0.8412, "step": 26299 }, { "epoch": 2.7984677590976803, "grad_norm": 10.1796875, "learning_rate": 9.083285957930643e-06, "loss": 0.7135, "step": 26300 }, { "epoch": 2.7985741647158973, "grad_norm": 13.375, "learning_rate": 9.083250426378625e-06, "loss": 0.7964, "step": 26301 }, { "epoch": 2.7986805703341138, "grad_norm": 12.640625, "learning_rate": 9.083214894826606e-06, "loss": 0.9859, "step": 26302 }, { "epoch": 2.7987869759523303, "grad_norm": 10.09375, "learning_rate": 9.08317936327459e-06, "loss": 0.9024, "step": 26303 }, { "epoch": 2.798893381570547, "grad_norm": 12.265625, "learning_rate": 9.083143831722571e-06, "loss": 0.9414, "step": 26304 }, { "epoch": 2.7989997871887633, "grad_norm": 13.4140625, "learning_rate": 9.083108300170553e-06, "loss": 0.8382, "step": 26305 }, { "epoch": 2.7991061928069803, "grad_norm": 12.203125, "learning_rate": 9.083072768618535e-06, "loss": 0.9229, "step": 26306 }, { "epoch": 2.7992125984251968, "grad_norm": 13.5546875, "learning_rate": 9.083037237066516e-06, "loss": 0.9525, "step": 26307 }, { "epoch": 2.7993190040434133, "grad_norm": 14.8515625, "learning_rate": 9.083001705514498e-06, "loss": 0.7982, "step": 26308 }, { "epoch": 2.7994254096616302, "grad_norm": 13.6796875, "learning_rate": 9.08296617396248e-06, "loss": 0.9956, "step": 26309 }, { "epoch": 2.7995318152798467, "grad_norm": 13.1953125, "learning_rate": 9.082930642410461e-06, "loss": 0.9536, "step": 26310 }, { "epoch": 2.7996382208980632, "grad_norm": 12.5625, "learning_rate": 9.082895110858443e-06, "loss": 0.9459, "step": 26311 }, { "epoch": 2.79974462651628, "grad_norm": 12.0546875, "learning_rate": 9.082859579306424e-06, "loss": 0.8097, "step": 26312 }, { "epoch": 2.7998510321344967, "grad_norm": 11.6328125, "learning_rate": 9.082824047754408e-06, "loss": 0.7541, "step": 26313 }, { "epoch": 2.799957437752713, "grad_norm": 14.625, "learning_rate": 9.082788516202389e-06, "loss": 0.7649, "step": 26314 }, { "epoch": 2.80006384337093, "grad_norm": 11.1484375, "learning_rate": 9.08275298465037e-06, "loss": 0.9009, "step": 26315 }, { "epoch": 2.8001702489891467, "grad_norm": 15.0234375, "learning_rate": 9.082717453098352e-06, "loss": 1.122, "step": 26316 }, { "epoch": 2.800276654607363, "grad_norm": 14.265625, "learning_rate": 9.082681921546334e-06, "loss": 1.2372, "step": 26317 }, { "epoch": 2.80038306022558, "grad_norm": 10.640625, "learning_rate": 9.082646389994316e-06, "loss": 0.8657, "step": 26318 }, { "epoch": 2.8004894658437967, "grad_norm": 13.71875, "learning_rate": 9.082610858442297e-06, "loss": 0.9311, "step": 26319 }, { "epoch": 2.800595871462013, "grad_norm": 14.6875, "learning_rate": 9.082575326890279e-06, "loss": 0.9531, "step": 26320 }, { "epoch": 2.8007022770802297, "grad_norm": 11.359375, "learning_rate": 9.08253979533826e-06, "loss": 0.8942, "step": 26321 }, { "epoch": 2.8008086826984466, "grad_norm": 9.8125, "learning_rate": 9.082504263786242e-06, "loss": 0.77, "step": 26322 }, { "epoch": 2.800915088316663, "grad_norm": 13.6875, "learning_rate": 9.082468732234224e-06, "loss": 0.9334, "step": 26323 }, { "epoch": 2.8010214939348796, "grad_norm": 10.578125, "learning_rate": 9.082433200682207e-06, "loss": 0.8865, "step": 26324 }, { "epoch": 2.801127899553096, "grad_norm": 10.375, "learning_rate": 9.082397669130189e-06, "loss": 0.84, "step": 26325 }, { "epoch": 2.801234305171313, "grad_norm": 12.1328125, "learning_rate": 9.08236213757817e-06, "loss": 0.7793, "step": 26326 }, { "epoch": 2.8013407107895296, "grad_norm": 9.515625, "learning_rate": 9.082326606026152e-06, "loss": 0.6972, "step": 26327 }, { "epoch": 2.801447116407746, "grad_norm": 12.734375, "learning_rate": 9.082291074474133e-06, "loss": 1.1227, "step": 26328 }, { "epoch": 2.801553522025963, "grad_norm": 11.7109375, "learning_rate": 9.082255542922117e-06, "loss": 0.8646, "step": 26329 }, { "epoch": 2.8016599276441796, "grad_norm": 11.9296875, "learning_rate": 9.082220011370098e-06, "loss": 0.9529, "step": 26330 }, { "epoch": 2.801766333262396, "grad_norm": 11.8125, "learning_rate": 9.08218447981808e-06, "loss": 0.9897, "step": 26331 }, { "epoch": 2.801872738880613, "grad_norm": 13.34375, "learning_rate": 9.082148948266061e-06, "loss": 1.1644, "step": 26332 }, { "epoch": 2.8019791444988296, "grad_norm": 11.5703125, "learning_rate": 9.082113416714043e-06, "loss": 0.8393, "step": 26333 }, { "epoch": 2.802085550117046, "grad_norm": 8.9375, "learning_rate": 9.082077885162025e-06, "loss": 0.7025, "step": 26334 }, { "epoch": 2.802191955735263, "grad_norm": 19.0, "learning_rate": 9.082042353610006e-06, "loss": 0.8458, "step": 26335 }, { "epoch": 2.8022983613534795, "grad_norm": 11.421875, "learning_rate": 9.082006822057988e-06, "loss": 0.7872, "step": 26336 }, { "epoch": 2.802404766971696, "grad_norm": 11.609375, "learning_rate": 9.08197129050597e-06, "loss": 0.9251, "step": 26337 }, { "epoch": 2.802511172589913, "grad_norm": 16.953125, "learning_rate": 9.081935758953951e-06, "loss": 1.0248, "step": 26338 }, { "epoch": 2.8026175782081295, "grad_norm": 14.015625, "learning_rate": 9.081900227401934e-06, "loss": 0.9037, "step": 26339 }, { "epoch": 2.802723983826346, "grad_norm": 14.6875, "learning_rate": 9.081864695849916e-06, "loss": 0.8559, "step": 26340 }, { "epoch": 2.8028303894445625, "grad_norm": 15.5546875, "learning_rate": 9.081829164297898e-06, "loss": 1.1722, "step": 26341 }, { "epoch": 2.802936795062779, "grad_norm": 11.640625, "learning_rate": 9.08179363274588e-06, "loss": 0.9378, "step": 26342 }, { "epoch": 2.803043200680996, "grad_norm": 14.5078125, "learning_rate": 9.08175810119386e-06, "loss": 0.973, "step": 26343 }, { "epoch": 2.8031496062992125, "grad_norm": 13.328125, "learning_rate": 9.081722569641842e-06, "loss": 0.9378, "step": 26344 }, { "epoch": 2.803256011917429, "grad_norm": 10.265625, "learning_rate": 9.081687038089824e-06, "loss": 0.9423, "step": 26345 }, { "epoch": 2.803362417535646, "grad_norm": 11.8828125, "learning_rate": 9.081651506537806e-06, "loss": 0.9073, "step": 26346 }, { "epoch": 2.8034688231538625, "grad_norm": 12.9375, "learning_rate": 9.081615974985787e-06, "loss": 0.9174, "step": 26347 }, { "epoch": 2.803575228772079, "grad_norm": 9.921875, "learning_rate": 9.081580443433769e-06, "loss": 0.7914, "step": 26348 }, { "epoch": 2.803681634390296, "grad_norm": 16.140625, "learning_rate": 9.081544911881752e-06, "loss": 0.9441, "step": 26349 }, { "epoch": 2.8037880400085124, "grad_norm": 11.640625, "learning_rate": 9.081509380329734e-06, "loss": 0.755, "step": 26350 }, { "epoch": 2.803894445626729, "grad_norm": 13.4375, "learning_rate": 9.081473848777715e-06, "loss": 0.8088, "step": 26351 }, { "epoch": 2.804000851244946, "grad_norm": 11.3515625, "learning_rate": 9.081438317225697e-06, "loss": 0.9206, "step": 26352 }, { "epoch": 2.8041072568631624, "grad_norm": 13.8203125, "learning_rate": 9.081402785673679e-06, "loss": 0.9028, "step": 26353 }, { "epoch": 2.804213662481379, "grad_norm": 20.421875, "learning_rate": 9.081367254121662e-06, "loss": 0.7823, "step": 26354 }, { "epoch": 2.804320068099596, "grad_norm": 12.2109375, "learning_rate": 9.081331722569643e-06, "loss": 0.8088, "step": 26355 }, { "epoch": 2.8044264737178124, "grad_norm": 12.2890625, "learning_rate": 9.081296191017625e-06, "loss": 0.9258, "step": 26356 }, { "epoch": 2.804532879336029, "grad_norm": 12.3515625, "learning_rate": 9.081260659465607e-06, "loss": 0.8211, "step": 26357 }, { "epoch": 2.804639284954246, "grad_norm": 15.125, "learning_rate": 9.081225127913588e-06, "loss": 0.9675, "step": 26358 }, { "epoch": 2.8047456905724624, "grad_norm": 10.484375, "learning_rate": 9.08118959636157e-06, "loss": 0.7912, "step": 26359 }, { "epoch": 2.804852096190679, "grad_norm": 10.53125, "learning_rate": 9.081154064809552e-06, "loss": 0.9387, "step": 26360 }, { "epoch": 2.8049585018088954, "grad_norm": 13.4453125, "learning_rate": 9.081118533257533e-06, "loss": 0.8899, "step": 26361 }, { "epoch": 2.805064907427112, "grad_norm": 11.96875, "learning_rate": 9.081083001705515e-06, "loss": 0.8675, "step": 26362 }, { "epoch": 2.805171313045329, "grad_norm": 11.3984375, "learning_rate": 9.081047470153496e-06, "loss": 0.818, "step": 26363 }, { "epoch": 2.8052777186635454, "grad_norm": 12.9921875, "learning_rate": 9.081011938601478e-06, "loss": 0.9066, "step": 26364 }, { "epoch": 2.805384124281762, "grad_norm": 11.09375, "learning_rate": 9.080976407049461e-06, "loss": 0.9304, "step": 26365 }, { "epoch": 2.805490529899979, "grad_norm": 16.421875, "learning_rate": 9.080940875497443e-06, "loss": 1.0319, "step": 26366 }, { "epoch": 2.8055969355181953, "grad_norm": 13.75, "learning_rate": 9.080905343945424e-06, "loss": 0.9002, "step": 26367 }, { "epoch": 2.805703341136412, "grad_norm": 13.875, "learning_rate": 9.080869812393406e-06, "loss": 1.0056, "step": 26368 }, { "epoch": 2.805809746754629, "grad_norm": 14.28125, "learning_rate": 9.080834280841388e-06, "loss": 0.9489, "step": 26369 }, { "epoch": 2.8059161523728453, "grad_norm": 9.046875, "learning_rate": 9.08079874928937e-06, "loss": 0.8422, "step": 26370 }, { "epoch": 2.806022557991062, "grad_norm": 12.53125, "learning_rate": 9.080763217737351e-06, "loss": 0.8549, "step": 26371 }, { "epoch": 2.8061289636092788, "grad_norm": 12.1796875, "learning_rate": 9.080727686185332e-06, "loss": 0.6281, "step": 26372 }, { "epoch": 2.8062353692274953, "grad_norm": 10.6953125, "learning_rate": 9.080692154633314e-06, "loss": 1.0393, "step": 26373 }, { "epoch": 2.806341774845712, "grad_norm": 10.0, "learning_rate": 9.080656623081296e-06, "loss": 0.8915, "step": 26374 }, { "epoch": 2.8064481804639287, "grad_norm": 12.0390625, "learning_rate": 9.080621091529279e-06, "loss": 1.0525, "step": 26375 }, { "epoch": 2.8065545860821453, "grad_norm": 13.8828125, "learning_rate": 9.08058555997726e-06, "loss": 0.9741, "step": 26376 }, { "epoch": 2.8066609917003618, "grad_norm": 12.9765625, "learning_rate": 9.080550028425242e-06, "loss": 1.2366, "step": 26377 }, { "epoch": 2.8067673973185783, "grad_norm": 12.5546875, "learning_rate": 9.080514496873224e-06, "loss": 0.7731, "step": 26378 }, { "epoch": 2.8068738029367952, "grad_norm": 11.59375, "learning_rate": 9.080478965321205e-06, "loss": 0.9402, "step": 26379 }, { "epoch": 2.8069802085550117, "grad_norm": 11.3203125, "learning_rate": 9.080443433769189e-06, "loss": 0.9288, "step": 26380 }, { "epoch": 2.8070866141732282, "grad_norm": 14.140625, "learning_rate": 9.08040790221717e-06, "loss": 0.8995, "step": 26381 }, { "epoch": 2.8071930197914448, "grad_norm": 14.6640625, "learning_rate": 9.080372370665152e-06, "loss": 1.0828, "step": 26382 }, { "epoch": 2.8072994254096617, "grad_norm": 10.7734375, "learning_rate": 9.080336839113134e-06, "loss": 0.8451, "step": 26383 }, { "epoch": 2.807405831027878, "grad_norm": 12.6171875, "learning_rate": 9.080301307561115e-06, "loss": 0.8468, "step": 26384 }, { "epoch": 2.8075122366460947, "grad_norm": 12.6796875, "learning_rate": 9.080265776009097e-06, "loss": 0.9466, "step": 26385 }, { "epoch": 2.8076186422643117, "grad_norm": 14.6015625, "learning_rate": 9.080230244457078e-06, "loss": 1.158, "step": 26386 }, { "epoch": 2.807725047882528, "grad_norm": 15.140625, "learning_rate": 9.08019471290506e-06, "loss": 1.068, "step": 26387 }, { "epoch": 2.8078314535007447, "grad_norm": 13.4609375, "learning_rate": 9.080159181353042e-06, "loss": 0.9867, "step": 26388 }, { "epoch": 2.8079378591189617, "grad_norm": 10.890625, "learning_rate": 9.080123649801023e-06, "loss": 0.8162, "step": 26389 }, { "epoch": 2.808044264737178, "grad_norm": 11.5, "learning_rate": 9.080088118249007e-06, "loss": 0.8427, "step": 26390 }, { "epoch": 2.8081506703553947, "grad_norm": 15.4609375, "learning_rate": 9.080052586696988e-06, "loss": 1.0083, "step": 26391 }, { "epoch": 2.8082570759736116, "grad_norm": 12.296875, "learning_rate": 9.08001705514497e-06, "loss": 0.8391, "step": 26392 }, { "epoch": 2.808363481591828, "grad_norm": 11.4375, "learning_rate": 9.079981523592951e-06, "loss": 0.8081, "step": 26393 }, { "epoch": 2.8084698872100446, "grad_norm": 12.9453125, "learning_rate": 9.079945992040933e-06, "loss": 0.9261, "step": 26394 }, { "epoch": 2.8085762928282616, "grad_norm": 11.671875, "learning_rate": 9.079910460488916e-06, "loss": 0.9544, "step": 26395 }, { "epoch": 2.808682698446478, "grad_norm": 14.8203125, "learning_rate": 9.079874928936898e-06, "loss": 1.1659, "step": 26396 }, { "epoch": 2.8087891040646946, "grad_norm": 12.6875, "learning_rate": 9.07983939738488e-06, "loss": 0.9248, "step": 26397 }, { "epoch": 2.808895509682911, "grad_norm": 16.171875, "learning_rate": 9.079803865832861e-06, "loss": 0.9785, "step": 26398 }, { "epoch": 2.8090019153011276, "grad_norm": 14.53125, "learning_rate": 9.079768334280843e-06, "loss": 0.9382, "step": 26399 }, { "epoch": 2.8091083209193446, "grad_norm": 12.734375, "learning_rate": 9.079732802728823e-06, "loss": 0.8927, "step": 26400 }, { "epoch": 2.809214726537561, "grad_norm": 13.6171875, "learning_rate": 9.079697271176806e-06, "loss": 0.9616, "step": 26401 }, { "epoch": 2.8093211321557776, "grad_norm": 15.8359375, "learning_rate": 9.079661739624787e-06, "loss": 0.9992, "step": 26402 }, { "epoch": 2.8094275377739946, "grad_norm": 14.3203125, "learning_rate": 9.079626208072769e-06, "loss": 0.9425, "step": 26403 }, { "epoch": 2.809533943392211, "grad_norm": 11.109375, "learning_rate": 9.07959067652075e-06, "loss": 0.7994, "step": 26404 }, { "epoch": 2.8096403490104276, "grad_norm": 10.390625, "learning_rate": 9.079555144968732e-06, "loss": 0.7498, "step": 26405 }, { "epoch": 2.8097467546286445, "grad_norm": 13.4453125, "learning_rate": 9.079519613416716e-06, "loss": 0.9715, "step": 26406 }, { "epoch": 2.809853160246861, "grad_norm": 11.71875, "learning_rate": 9.079484081864697e-06, "loss": 0.9842, "step": 26407 }, { "epoch": 2.8099595658650776, "grad_norm": 14.6875, "learning_rate": 9.079448550312679e-06, "loss": 0.8854, "step": 26408 }, { "epoch": 2.8100659714832945, "grad_norm": 14.5, "learning_rate": 9.07941301876066e-06, "loss": 0.9355, "step": 26409 }, { "epoch": 2.810172377101511, "grad_norm": 11.5625, "learning_rate": 9.079377487208642e-06, "loss": 0.7978, "step": 26410 }, { "epoch": 2.8102787827197275, "grad_norm": 15.765625, "learning_rate": 9.079341955656624e-06, "loss": 1.0759, "step": 26411 }, { "epoch": 2.8103851883379445, "grad_norm": 12.2734375, "learning_rate": 9.079306424104605e-06, "loss": 0.8416, "step": 26412 }, { "epoch": 2.810491593956161, "grad_norm": 14.140625, "learning_rate": 9.079270892552587e-06, "loss": 0.933, "step": 26413 }, { "epoch": 2.8105979995743775, "grad_norm": 9.5625, "learning_rate": 9.079235361000568e-06, "loss": 0.8823, "step": 26414 }, { "epoch": 2.810704405192594, "grad_norm": 15.5234375, "learning_rate": 9.07919982944855e-06, "loss": 0.9909, "step": 26415 }, { "epoch": 2.810810810810811, "grad_norm": 11.15625, "learning_rate": 9.079164297896533e-06, "loss": 0.899, "step": 26416 }, { "epoch": 2.8109172164290275, "grad_norm": 10.90625, "learning_rate": 9.079128766344515e-06, "loss": 0.9805, "step": 26417 }, { "epoch": 2.811023622047244, "grad_norm": 13.0, "learning_rate": 9.079093234792497e-06, "loss": 0.9167, "step": 26418 }, { "epoch": 2.8111300276654605, "grad_norm": 18.015625, "learning_rate": 9.079057703240478e-06, "loss": 0.9242, "step": 26419 }, { "epoch": 2.8112364332836774, "grad_norm": 12.140625, "learning_rate": 9.07902217168846e-06, "loss": 0.8882, "step": 26420 }, { "epoch": 2.811342838901894, "grad_norm": 10.375, "learning_rate": 9.078986640136443e-06, "loss": 0.806, "step": 26421 }, { "epoch": 2.8114492445201105, "grad_norm": 13.6171875, "learning_rate": 9.078951108584425e-06, "loss": 0.8907, "step": 26422 }, { "epoch": 2.8115556501383274, "grad_norm": 13.7578125, "learning_rate": 9.078915577032406e-06, "loss": 0.8346, "step": 26423 }, { "epoch": 2.811662055756544, "grad_norm": 12.1953125, "learning_rate": 9.078880045480388e-06, "loss": 1.0657, "step": 26424 }, { "epoch": 2.8117684613747604, "grad_norm": 13.3828125, "learning_rate": 9.07884451392837e-06, "loss": 0.8235, "step": 26425 }, { "epoch": 2.8118748669929774, "grad_norm": 13.015625, "learning_rate": 9.078808982376351e-06, "loss": 0.8883, "step": 26426 }, { "epoch": 2.811981272611194, "grad_norm": 18.09375, "learning_rate": 9.078773450824333e-06, "loss": 0.996, "step": 26427 }, { "epoch": 2.8120876782294104, "grad_norm": 13.109375, "learning_rate": 9.078737919272314e-06, "loss": 0.82, "step": 26428 }, { "epoch": 2.8121940838476274, "grad_norm": 12.859375, "learning_rate": 9.078702387720296e-06, "loss": 0.9172, "step": 26429 }, { "epoch": 2.812300489465844, "grad_norm": 13.203125, "learning_rate": 9.078666856168278e-06, "loss": 0.8663, "step": 26430 }, { "epoch": 2.8124068950840604, "grad_norm": 13.3359375, "learning_rate": 9.07863132461626e-06, "loss": 0.9879, "step": 26431 }, { "epoch": 2.8125133007022773, "grad_norm": 16.578125, "learning_rate": 9.078595793064242e-06, "loss": 0.884, "step": 26432 }, { "epoch": 2.812619706320494, "grad_norm": 13.0, "learning_rate": 9.078560261512224e-06, "loss": 0.9937, "step": 26433 }, { "epoch": 2.8127261119387104, "grad_norm": 13.0234375, "learning_rate": 9.078524729960206e-06, "loss": 0.9314, "step": 26434 }, { "epoch": 2.812832517556927, "grad_norm": 10.5703125, "learning_rate": 9.078489198408187e-06, "loss": 0.8817, "step": 26435 }, { "epoch": 2.8129389231751434, "grad_norm": 11.8671875, "learning_rate": 9.078453666856169e-06, "loss": 0.9252, "step": 26436 }, { "epoch": 2.8130453287933603, "grad_norm": 12.203125, "learning_rate": 9.07841813530415e-06, "loss": 0.8874, "step": 26437 }, { "epoch": 2.813151734411577, "grad_norm": 14.9453125, "learning_rate": 9.078382603752132e-06, "loss": 0.9426, "step": 26438 }, { "epoch": 2.8132581400297934, "grad_norm": 14.1484375, "learning_rate": 9.078347072200114e-06, "loss": 1.0036, "step": 26439 }, { "epoch": 2.8133645456480103, "grad_norm": 11.9453125, "learning_rate": 9.078311540648095e-06, "loss": 0.8336, "step": 26440 }, { "epoch": 2.813470951266227, "grad_norm": 14.046875, "learning_rate": 9.078276009096077e-06, "loss": 1.0807, "step": 26441 }, { "epoch": 2.8135773568844433, "grad_norm": 12.015625, "learning_rate": 9.07824047754406e-06, "loss": 0.9319, "step": 26442 }, { "epoch": 2.8136837625026603, "grad_norm": 13.3046875, "learning_rate": 9.078204945992042e-06, "loss": 0.9433, "step": 26443 }, { "epoch": 2.813790168120877, "grad_norm": 9.84375, "learning_rate": 9.078169414440023e-06, "loss": 0.9292, "step": 26444 }, { "epoch": 2.8138965737390933, "grad_norm": 8.6171875, "learning_rate": 9.078133882888005e-06, "loss": 0.854, "step": 26445 }, { "epoch": 2.8140029793573103, "grad_norm": 11.34375, "learning_rate": 9.078098351335987e-06, "loss": 0.9214, "step": 26446 }, { "epoch": 2.8141093849755268, "grad_norm": 14.5234375, "learning_rate": 9.07806281978397e-06, "loss": 0.8711, "step": 26447 }, { "epoch": 2.8142157905937433, "grad_norm": 12.9296875, "learning_rate": 9.078027288231952e-06, "loss": 1.0065, "step": 26448 }, { "epoch": 2.8143221962119602, "grad_norm": 14.3671875, "learning_rate": 9.077991756679933e-06, "loss": 0.8142, "step": 26449 }, { "epoch": 2.8144286018301767, "grad_norm": 11.5078125, "learning_rate": 9.077956225127915e-06, "loss": 0.8132, "step": 26450 }, { "epoch": 2.8145350074483932, "grad_norm": 15.2265625, "learning_rate": 9.077920693575896e-06, "loss": 1.0541, "step": 26451 }, { "epoch": 2.8146414130666098, "grad_norm": 17.296875, "learning_rate": 9.077885162023878e-06, "loss": 0.9125, "step": 26452 }, { "epoch": 2.8147478186848267, "grad_norm": 10.0, "learning_rate": 9.07784963047186e-06, "loss": 0.7588, "step": 26453 }, { "epoch": 2.814854224303043, "grad_norm": 13.2890625, "learning_rate": 9.077814098919841e-06, "loss": 1.0543, "step": 26454 }, { "epoch": 2.8149606299212597, "grad_norm": 13.5, "learning_rate": 9.077778567367823e-06, "loss": 1.0757, "step": 26455 }, { "epoch": 2.8150670355394762, "grad_norm": 14.078125, "learning_rate": 9.077743035815804e-06, "loss": 0.9144, "step": 26456 }, { "epoch": 2.815173441157693, "grad_norm": 10.1953125, "learning_rate": 9.077707504263788e-06, "loss": 0.883, "step": 26457 }, { "epoch": 2.8152798467759097, "grad_norm": 13.5625, "learning_rate": 9.07767197271177e-06, "loss": 1.1081, "step": 26458 }, { "epoch": 2.815386252394126, "grad_norm": 14.4453125, "learning_rate": 9.077636441159751e-06, "loss": 1.0647, "step": 26459 }, { "epoch": 2.815492658012343, "grad_norm": 11.109375, "learning_rate": 9.077600909607733e-06, "loss": 0.8853, "step": 26460 }, { "epoch": 2.8155990636305597, "grad_norm": 12.7734375, "learning_rate": 9.077565378055714e-06, "loss": 1.019, "step": 26461 }, { "epoch": 2.815705469248776, "grad_norm": 10.6953125, "learning_rate": 9.077529846503696e-06, "loss": 0.7995, "step": 26462 }, { "epoch": 2.815811874866993, "grad_norm": 10.7265625, "learning_rate": 9.077494314951677e-06, "loss": 0.8342, "step": 26463 }, { "epoch": 2.8159182804852096, "grad_norm": 7.875, "learning_rate": 9.077458783399659e-06, "loss": 0.6172, "step": 26464 }, { "epoch": 2.816024686103426, "grad_norm": 13.7734375, "learning_rate": 9.07742325184764e-06, "loss": 0.9232, "step": 26465 }, { "epoch": 2.816131091721643, "grad_norm": 12.9609375, "learning_rate": 9.077387720295622e-06, "loss": 0.82, "step": 26466 }, { "epoch": 2.8162374973398596, "grad_norm": 13.125, "learning_rate": 9.077352188743605e-06, "loss": 1.038, "step": 26467 }, { "epoch": 2.816343902958076, "grad_norm": 12.578125, "learning_rate": 9.077316657191587e-06, "loss": 0.9053, "step": 26468 }, { "epoch": 2.816450308576293, "grad_norm": 13.78125, "learning_rate": 9.077281125639569e-06, "loss": 1.0452, "step": 26469 }, { "epoch": 2.8165567141945096, "grad_norm": 13.625, "learning_rate": 9.07724559408755e-06, "loss": 0.9407, "step": 26470 }, { "epoch": 2.816663119812726, "grad_norm": 13.234375, "learning_rate": 9.077210062535532e-06, "loss": 1.0029, "step": 26471 }, { "epoch": 2.8167695254309426, "grad_norm": 12.109375, "learning_rate": 9.077174530983515e-06, "loss": 0.8919, "step": 26472 }, { "epoch": 2.816875931049159, "grad_norm": 11.203125, "learning_rate": 9.077138999431497e-06, "loss": 0.8327, "step": 26473 }, { "epoch": 2.816982336667376, "grad_norm": 12.375, "learning_rate": 9.077103467879478e-06, "loss": 0.9906, "step": 26474 }, { "epoch": 2.8170887422855926, "grad_norm": 17.5625, "learning_rate": 9.07706793632746e-06, "loss": 0.8423, "step": 26475 }, { "epoch": 2.817195147903809, "grad_norm": 17.453125, "learning_rate": 9.077032404775442e-06, "loss": 0.9574, "step": 26476 }, { "epoch": 2.817301553522026, "grad_norm": 16.15625, "learning_rate": 9.076996873223423e-06, "loss": 0.9548, "step": 26477 }, { "epoch": 2.8174079591402426, "grad_norm": 11.515625, "learning_rate": 9.076961341671405e-06, "loss": 0.7752, "step": 26478 }, { "epoch": 2.817514364758459, "grad_norm": 14.078125, "learning_rate": 9.076925810119386e-06, "loss": 0.7997, "step": 26479 }, { "epoch": 2.817620770376676, "grad_norm": 14.359375, "learning_rate": 9.076890278567368e-06, "loss": 0.6676, "step": 26480 }, { "epoch": 2.8177271759948925, "grad_norm": 11.2890625, "learning_rate": 9.07685474701535e-06, "loss": 0.9117, "step": 26481 }, { "epoch": 2.817833581613109, "grad_norm": 13.421875, "learning_rate": 9.076819215463331e-06, "loss": 0.9112, "step": 26482 }, { "epoch": 2.817939987231326, "grad_norm": 13.9296875, "learning_rate": 9.076783683911315e-06, "loss": 0.8553, "step": 26483 }, { "epoch": 2.8180463928495425, "grad_norm": 18.28125, "learning_rate": 9.076748152359296e-06, "loss": 0.911, "step": 26484 }, { "epoch": 2.818152798467759, "grad_norm": 11.9921875, "learning_rate": 9.076712620807278e-06, "loss": 0.8605, "step": 26485 }, { "epoch": 2.818259204085976, "grad_norm": 15.7578125, "learning_rate": 9.07667708925526e-06, "loss": 0.8694, "step": 26486 }, { "epoch": 2.8183656097041925, "grad_norm": 13.3125, "learning_rate": 9.076641557703241e-06, "loss": 0.5921, "step": 26487 }, { "epoch": 2.818472015322409, "grad_norm": 11.546875, "learning_rate": 9.076606026151224e-06, "loss": 0.9741, "step": 26488 }, { "epoch": 2.818578420940626, "grad_norm": 21.40625, "learning_rate": 9.076570494599204e-06, "loss": 1.0867, "step": 26489 }, { "epoch": 2.8186848265588424, "grad_norm": 14.2421875, "learning_rate": 9.076534963047186e-06, "loss": 1.0652, "step": 26490 }, { "epoch": 2.818791232177059, "grad_norm": 15.484375, "learning_rate": 9.076499431495167e-06, "loss": 0.8913, "step": 26491 }, { "epoch": 2.8188976377952755, "grad_norm": 10.6484375, "learning_rate": 9.076463899943149e-06, "loss": 0.9014, "step": 26492 }, { "epoch": 2.819004043413492, "grad_norm": 13.0625, "learning_rate": 9.076428368391132e-06, "loss": 1.0318, "step": 26493 }, { "epoch": 2.819110449031709, "grad_norm": 16.265625, "learning_rate": 9.076392836839114e-06, "loss": 1.1316, "step": 26494 }, { "epoch": 2.8192168546499254, "grad_norm": 10.734375, "learning_rate": 9.076357305287096e-06, "loss": 0.5825, "step": 26495 }, { "epoch": 2.819323260268142, "grad_norm": 12.8828125, "learning_rate": 9.076321773735077e-06, "loss": 0.9257, "step": 26496 }, { "epoch": 2.819429665886359, "grad_norm": 15.25, "learning_rate": 9.076286242183059e-06, "loss": 1.0, "step": 26497 }, { "epoch": 2.8195360715045754, "grad_norm": 13.953125, "learning_rate": 9.076250710631042e-06, "loss": 1.0099, "step": 26498 }, { "epoch": 2.819642477122792, "grad_norm": 11.640625, "learning_rate": 9.076215179079024e-06, "loss": 0.9575, "step": 26499 }, { "epoch": 2.819748882741009, "grad_norm": 17.6875, "learning_rate": 9.076179647527005e-06, "loss": 0.7648, "step": 26500 }, { "epoch": 2.8198552883592254, "grad_norm": 13.078125, "learning_rate": 9.076144115974987e-06, "loss": 1.0642, "step": 26501 }, { "epoch": 2.819961693977442, "grad_norm": 15.7421875, "learning_rate": 9.076108584422968e-06, "loss": 0.7842, "step": 26502 }, { "epoch": 2.820068099595659, "grad_norm": 14.1640625, "learning_rate": 9.07607305287095e-06, "loss": 0.8312, "step": 26503 }, { "epoch": 2.8201745052138754, "grad_norm": 12.359375, "learning_rate": 9.076037521318932e-06, "loss": 0.9636, "step": 26504 }, { "epoch": 2.820280910832092, "grad_norm": 17.8125, "learning_rate": 9.076001989766913e-06, "loss": 0.8457, "step": 26505 }, { "epoch": 2.820387316450309, "grad_norm": 12.0703125, "learning_rate": 9.075966458214895e-06, "loss": 0.9033, "step": 26506 }, { "epoch": 2.8204937220685253, "grad_norm": 11.71875, "learning_rate": 9.075930926662877e-06, "loss": 0.7837, "step": 26507 }, { "epoch": 2.820600127686742, "grad_norm": 23.109375, "learning_rate": 9.07589539511086e-06, "loss": 0.8476, "step": 26508 }, { "epoch": 2.8207065333049584, "grad_norm": 14.0703125, "learning_rate": 9.075859863558841e-06, "loss": 0.9168, "step": 26509 }, { "epoch": 2.8208129389231753, "grad_norm": 16.46875, "learning_rate": 9.075824332006823e-06, "loss": 1.0195, "step": 26510 }, { "epoch": 2.820919344541392, "grad_norm": 11.7421875, "learning_rate": 9.075788800454805e-06, "loss": 0.8176, "step": 26511 }, { "epoch": 2.8210257501596083, "grad_norm": 15.578125, "learning_rate": 9.075753268902786e-06, "loss": 0.7912, "step": 26512 }, { "epoch": 2.821132155777825, "grad_norm": 10.8671875, "learning_rate": 9.07571773735077e-06, "loss": 0.8548, "step": 26513 }, { "epoch": 2.821238561396042, "grad_norm": 13.0546875, "learning_rate": 9.075682205798751e-06, "loss": 1.0024, "step": 26514 }, { "epoch": 2.8213449670142583, "grad_norm": 16.671875, "learning_rate": 9.075646674246733e-06, "loss": 0.7723, "step": 26515 }, { "epoch": 2.821451372632475, "grad_norm": 10.8984375, "learning_rate": 9.075611142694714e-06, "loss": 0.8093, "step": 26516 }, { "epoch": 2.8215577782506918, "grad_norm": 11.875, "learning_rate": 9.075575611142696e-06, "loss": 0.8395, "step": 26517 }, { "epoch": 2.8216641838689083, "grad_norm": 11.546875, "learning_rate": 9.075540079590678e-06, "loss": 0.9889, "step": 26518 }, { "epoch": 2.821770589487125, "grad_norm": 13.3671875, "learning_rate": 9.07550454803866e-06, "loss": 0.8336, "step": 26519 }, { "epoch": 2.8218769951053417, "grad_norm": 15.2734375, "learning_rate": 9.07546901648664e-06, "loss": 0.8825, "step": 26520 }, { "epoch": 2.8219834007235582, "grad_norm": 17.0625, "learning_rate": 9.075433484934622e-06, "loss": 1.0111, "step": 26521 }, { "epoch": 2.8220898063417748, "grad_norm": 19.375, "learning_rate": 9.075397953382604e-06, "loss": 1.0871, "step": 26522 }, { "epoch": 2.8221962119599917, "grad_norm": 18.90625, "learning_rate": 9.075362421830586e-06, "loss": 0.8677, "step": 26523 }, { "epoch": 2.822302617578208, "grad_norm": 12.828125, "learning_rate": 9.075326890278569e-06, "loss": 0.8348, "step": 26524 }, { "epoch": 2.8224090231964247, "grad_norm": 15.53125, "learning_rate": 9.07529135872655e-06, "loss": 0.8023, "step": 26525 }, { "epoch": 2.8225154288146417, "grad_norm": 11.5625, "learning_rate": 9.075255827174532e-06, "loss": 0.9085, "step": 26526 }, { "epoch": 2.822621834432858, "grad_norm": 12.65625, "learning_rate": 9.075220295622514e-06, "loss": 1.0553, "step": 26527 }, { "epoch": 2.8227282400510747, "grad_norm": 14.859375, "learning_rate": 9.075184764070495e-06, "loss": 1.0519, "step": 26528 }, { "epoch": 2.822834645669291, "grad_norm": 11.65625, "learning_rate": 9.075149232518477e-06, "loss": 0.8858, "step": 26529 }, { "epoch": 2.8229410512875077, "grad_norm": 16.5625, "learning_rate": 9.075113700966459e-06, "loss": 0.9367, "step": 26530 }, { "epoch": 2.8230474569057247, "grad_norm": 11.0703125, "learning_rate": 9.07507816941444e-06, "loss": 0.9943, "step": 26531 }, { "epoch": 2.823153862523941, "grad_norm": 10.6484375, "learning_rate": 9.075042637862422e-06, "loss": 0.9033, "step": 26532 }, { "epoch": 2.8232602681421577, "grad_norm": 11.875, "learning_rate": 9.075007106310403e-06, "loss": 0.8735, "step": 26533 }, { "epoch": 2.8233666737603746, "grad_norm": 11.3359375, "learning_rate": 9.074971574758387e-06, "loss": 0.8996, "step": 26534 }, { "epoch": 2.823473079378591, "grad_norm": 14.09375, "learning_rate": 9.074936043206368e-06, "loss": 0.9693, "step": 26535 }, { "epoch": 2.8235794849968077, "grad_norm": 15.078125, "learning_rate": 9.07490051165435e-06, "loss": 0.7827, "step": 26536 }, { "epoch": 2.8236858906150246, "grad_norm": 13.734375, "learning_rate": 9.074864980102331e-06, "loss": 0.9575, "step": 26537 }, { "epoch": 2.823792296233241, "grad_norm": 12.15625, "learning_rate": 9.074829448550313e-06, "loss": 0.8992, "step": 26538 }, { "epoch": 2.8238987018514576, "grad_norm": 11.0390625, "learning_rate": 9.074793916998296e-06, "loss": 0.8393, "step": 26539 }, { "epoch": 2.8240051074696746, "grad_norm": 12.984375, "learning_rate": 9.074758385446278e-06, "loss": 0.8972, "step": 26540 }, { "epoch": 2.824111513087891, "grad_norm": 14.8828125, "learning_rate": 9.07472285389426e-06, "loss": 0.826, "step": 26541 }, { "epoch": 2.8242179187061076, "grad_norm": 15.328125, "learning_rate": 9.074687322342241e-06, "loss": 0.9175, "step": 26542 }, { "epoch": 2.8243243243243246, "grad_norm": 10.296875, "learning_rate": 9.074651790790223e-06, "loss": 0.8326, "step": 26543 }, { "epoch": 2.824430729942541, "grad_norm": 10.6171875, "learning_rate": 9.074616259238204e-06, "loss": 0.7268, "step": 26544 }, { "epoch": 2.8245371355607576, "grad_norm": 12.0703125, "learning_rate": 9.074580727686186e-06, "loss": 0.9344, "step": 26545 }, { "epoch": 2.824643541178974, "grad_norm": 17.609375, "learning_rate": 9.074545196134168e-06, "loss": 0.903, "step": 26546 }, { "epoch": 2.824749946797191, "grad_norm": 16.390625, "learning_rate": 9.07450966458215e-06, "loss": 0.7571, "step": 26547 }, { "epoch": 2.8248563524154076, "grad_norm": 12.4921875, "learning_rate": 9.07447413303013e-06, "loss": 1.0043, "step": 26548 }, { "epoch": 2.824962758033624, "grad_norm": 13.0703125, "learning_rate": 9.074438601478114e-06, "loss": 0.9075, "step": 26549 }, { "epoch": 2.8250691636518406, "grad_norm": 11.8359375, "learning_rate": 9.074403069926096e-06, "loss": 1.0351, "step": 26550 }, { "epoch": 2.8251755692700575, "grad_norm": 11.6640625, "learning_rate": 9.074367538374077e-06, "loss": 0.821, "step": 26551 }, { "epoch": 2.825281974888274, "grad_norm": 10.3046875, "learning_rate": 9.074332006822059e-06, "loss": 0.9434, "step": 26552 }, { "epoch": 2.8253883805064905, "grad_norm": 14.9140625, "learning_rate": 9.07429647527004e-06, "loss": 0.9053, "step": 26553 }, { "epoch": 2.8254947861247075, "grad_norm": 12.5234375, "learning_rate": 9.074260943718022e-06, "loss": 1.0051, "step": 26554 }, { "epoch": 2.825601191742924, "grad_norm": 15.84375, "learning_rate": 9.074225412166004e-06, "loss": 0.8666, "step": 26555 }, { "epoch": 2.8257075973611405, "grad_norm": 11.6171875, "learning_rate": 9.074189880613985e-06, "loss": 0.857, "step": 26556 }, { "epoch": 2.8258140029793575, "grad_norm": 13.609375, "learning_rate": 9.074154349061967e-06, "loss": 0.8439, "step": 26557 }, { "epoch": 2.825920408597574, "grad_norm": 13.015625, "learning_rate": 9.074118817509949e-06, "loss": 0.8452, "step": 26558 }, { "epoch": 2.8260268142157905, "grad_norm": 14.1171875, "learning_rate": 9.07408328595793e-06, "loss": 0.9893, "step": 26559 }, { "epoch": 2.8261332198340074, "grad_norm": 12.2265625, "learning_rate": 9.074047754405914e-06, "loss": 0.9195, "step": 26560 }, { "epoch": 2.826239625452224, "grad_norm": 14.390625, "learning_rate": 9.074012222853895e-06, "loss": 0.9766, "step": 26561 }, { "epoch": 2.8263460310704405, "grad_norm": 13.2578125, "learning_rate": 9.073976691301877e-06, "loss": 0.91, "step": 26562 }, { "epoch": 2.8264524366886574, "grad_norm": 14.734375, "learning_rate": 9.073941159749858e-06, "loss": 1.2565, "step": 26563 }, { "epoch": 2.826558842306874, "grad_norm": 12.9921875, "learning_rate": 9.07390562819784e-06, "loss": 0.9521, "step": 26564 }, { "epoch": 2.8266652479250904, "grad_norm": 13.171875, "learning_rate": 9.073870096645823e-06, "loss": 1.1175, "step": 26565 }, { "epoch": 2.826771653543307, "grad_norm": 12.84375, "learning_rate": 9.073834565093805e-06, "loss": 0.9241, "step": 26566 }, { "epoch": 2.8268780591615235, "grad_norm": 13.8046875, "learning_rate": 9.073799033541786e-06, "loss": 0.9937, "step": 26567 }, { "epoch": 2.8269844647797404, "grad_norm": 11.6640625, "learning_rate": 9.073763501989768e-06, "loss": 0.8039, "step": 26568 }, { "epoch": 2.827090870397957, "grad_norm": 15.0, "learning_rate": 9.07372797043775e-06, "loss": 0.9266, "step": 26569 }, { "epoch": 2.8271972760161734, "grad_norm": 14.296875, "learning_rate": 9.073692438885731e-06, "loss": 0.9213, "step": 26570 }, { "epoch": 2.8273036816343904, "grad_norm": 11.9765625, "learning_rate": 9.073656907333713e-06, "loss": 0.8722, "step": 26571 }, { "epoch": 2.827410087252607, "grad_norm": 11.6640625, "learning_rate": 9.073621375781694e-06, "loss": 1.0656, "step": 26572 }, { "epoch": 2.8275164928708234, "grad_norm": 13.6015625, "learning_rate": 9.073585844229676e-06, "loss": 0.9523, "step": 26573 }, { "epoch": 2.8276228984890404, "grad_norm": 12.8046875, "learning_rate": 9.073550312677658e-06, "loss": 0.9036, "step": 26574 }, { "epoch": 2.827729304107257, "grad_norm": 12.03125, "learning_rate": 9.073514781125641e-06, "loss": 0.7222, "step": 26575 }, { "epoch": 2.8278357097254734, "grad_norm": 14.2578125, "learning_rate": 9.073479249573623e-06, "loss": 1.1615, "step": 26576 }, { "epoch": 2.8279421153436903, "grad_norm": 12.75, "learning_rate": 9.073443718021604e-06, "loss": 1.099, "step": 26577 }, { "epoch": 2.828048520961907, "grad_norm": 12.7421875, "learning_rate": 9.073408186469586e-06, "loss": 0.8857, "step": 26578 }, { "epoch": 2.8281549265801234, "grad_norm": 12.7421875, "learning_rate": 9.073372654917567e-06, "loss": 0.8688, "step": 26579 }, { "epoch": 2.8282613321983403, "grad_norm": 12.0234375, "learning_rate": 9.073337123365549e-06, "loss": 0.8634, "step": 26580 }, { "epoch": 2.828367737816557, "grad_norm": 12.7734375, "learning_rate": 9.07330159181353e-06, "loss": 0.835, "step": 26581 }, { "epoch": 2.8284741434347733, "grad_norm": 15.140625, "learning_rate": 9.073266060261512e-06, "loss": 0.7145, "step": 26582 }, { "epoch": 2.82858054905299, "grad_norm": 9.0546875, "learning_rate": 9.073230528709494e-06, "loss": 0.822, "step": 26583 }, { "epoch": 2.828686954671207, "grad_norm": 12.4921875, "learning_rate": 9.073194997157475e-06, "loss": 0.9256, "step": 26584 }, { "epoch": 2.8287933602894233, "grad_norm": 13.5546875, "learning_rate": 9.073159465605459e-06, "loss": 1.0535, "step": 26585 }, { "epoch": 2.82889976590764, "grad_norm": 14.125, "learning_rate": 9.07312393405344e-06, "loss": 0.9134, "step": 26586 }, { "epoch": 2.8290061715258563, "grad_norm": 13.4453125, "learning_rate": 9.073088402501422e-06, "loss": 1.0931, "step": 26587 }, { "epoch": 2.8291125771440733, "grad_norm": 11.2109375, "learning_rate": 9.073052870949404e-06, "loss": 0.858, "step": 26588 }, { "epoch": 2.82921898276229, "grad_norm": 14.0703125, "learning_rate": 9.073017339397385e-06, "loss": 0.934, "step": 26589 }, { "epoch": 2.8293253883805063, "grad_norm": 11.078125, "learning_rate": 9.072981807845368e-06, "loss": 0.8115, "step": 26590 }, { "epoch": 2.8294317939987232, "grad_norm": 11.1484375, "learning_rate": 9.07294627629335e-06, "loss": 0.7444, "step": 26591 }, { "epoch": 2.8295381996169398, "grad_norm": 10.859375, "learning_rate": 9.072910744741332e-06, "loss": 0.9269, "step": 26592 }, { "epoch": 2.8296446052351563, "grad_norm": 14.0703125, "learning_rate": 9.072875213189313e-06, "loss": 1.0519, "step": 26593 }, { "epoch": 2.829751010853373, "grad_norm": 12.046875, "learning_rate": 9.072839681637295e-06, "loss": 0.7116, "step": 26594 }, { "epoch": 2.8298574164715897, "grad_norm": 10.3046875, "learning_rate": 9.072804150085277e-06, "loss": 0.7874, "step": 26595 }, { "epoch": 2.8299638220898062, "grad_norm": 14.234375, "learning_rate": 9.072768618533258e-06, "loss": 0.9115, "step": 26596 }, { "epoch": 2.830070227708023, "grad_norm": 10.7578125, "learning_rate": 9.07273308698124e-06, "loss": 0.9246, "step": 26597 }, { "epoch": 2.8301766333262397, "grad_norm": 11.5546875, "learning_rate": 9.072697555429221e-06, "loss": 0.8982, "step": 26598 }, { "epoch": 2.830283038944456, "grad_norm": 9.0078125, "learning_rate": 9.072662023877203e-06, "loss": 0.763, "step": 26599 }, { "epoch": 2.830389444562673, "grad_norm": 14.8203125, "learning_rate": 9.072626492325185e-06, "loss": 0.9576, "step": 26600 }, { "epoch": 2.8304958501808897, "grad_norm": 12.21875, "learning_rate": 9.072590960773168e-06, "loss": 0.7908, "step": 26601 }, { "epoch": 2.830602255799106, "grad_norm": 13.1171875, "learning_rate": 9.07255542922115e-06, "loss": 0.9192, "step": 26602 }, { "epoch": 2.8307086614173227, "grad_norm": 12.9921875, "learning_rate": 9.072519897669131e-06, "loss": 0.6909, "step": 26603 }, { "epoch": 2.830815067035539, "grad_norm": 10.4921875, "learning_rate": 9.072484366117113e-06, "loss": 0.8899, "step": 26604 }, { "epoch": 2.830921472653756, "grad_norm": 12.1640625, "learning_rate": 9.072448834565094e-06, "loss": 0.9059, "step": 26605 }, { "epoch": 2.8310278782719727, "grad_norm": 12.8046875, "learning_rate": 9.072413303013078e-06, "loss": 1.036, "step": 26606 }, { "epoch": 2.831134283890189, "grad_norm": 12.4453125, "learning_rate": 9.07237777146106e-06, "loss": 1.0099, "step": 26607 }, { "epoch": 2.831240689508406, "grad_norm": 15.4921875, "learning_rate": 9.072342239909039e-06, "loss": 1.0364, "step": 26608 }, { "epoch": 2.8313470951266226, "grad_norm": 14.34375, "learning_rate": 9.07230670835702e-06, "loss": 1.0231, "step": 26609 }, { "epoch": 2.831453500744839, "grad_norm": 10.9765625, "learning_rate": 9.072271176805002e-06, "loss": 0.8356, "step": 26610 }, { "epoch": 2.831559906363056, "grad_norm": 14.9296875, "learning_rate": 9.072235645252986e-06, "loss": 0.877, "step": 26611 }, { "epoch": 2.8316663119812726, "grad_norm": 11.46875, "learning_rate": 9.072200113700967e-06, "loss": 0.8993, "step": 26612 }, { "epoch": 2.831772717599489, "grad_norm": 11.9296875, "learning_rate": 9.072164582148949e-06, "loss": 0.9604, "step": 26613 }, { "epoch": 2.831879123217706, "grad_norm": 13.6015625, "learning_rate": 9.07212905059693e-06, "loss": 0.8242, "step": 26614 }, { "epoch": 2.8319855288359226, "grad_norm": 14.1484375, "learning_rate": 9.072093519044912e-06, "loss": 0.8975, "step": 26615 }, { "epoch": 2.832091934454139, "grad_norm": 18.890625, "learning_rate": 9.072057987492895e-06, "loss": 1.2027, "step": 26616 }, { "epoch": 2.832198340072356, "grad_norm": 12.3984375, "learning_rate": 9.072022455940877e-06, "loss": 1.0, "step": 26617 }, { "epoch": 2.8323047456905726, "grad_norm": 15.28125, "learning_rate": 9.071986924388859e-06, "loss": 0.8866, "step": 26618 }, { "epoch": 2.832411151308789, "grad_norm": 12.5, "learning_rate": 9.07195139283684e-06, "loss": 0.9984, "step": 26619 }, { "epoch": 2.832517556927006, "grad_norm": 12.453125, "learning_rate": 9.071915861284822e-06, "loss": 0.7902, "step": 26620 }, { "epoch": 2.8326239625452225, "grad_norm": 10.296875, "learning_rate": 9.071880329732803e-06, "loss": 0.7302, "step": 26621 }, { "epoch": 2.832730368163439, "grad_norm": 13.921875, "learning_rate": 9.071844798180785e-06, "loss": 1.1051, "step": 26622 }, { "epoch": 2.8328367737816555, "grad_norm": 13.078125, "learning_rate": 9.071809266628767e-06, "loss": 0.9418, "step": 26623 }, { "epoch": 2.832943179399872, "grad_norm": 10.578125, "learning_rate": 9.071773735076748e-06, "loss": 0.7826, "step": 26624 }, { "epoch": 2.833049585018089, "grad_norm": 12.890625, "learning_rate": 9.07173820352473e-06, "loss": 1.181, "step": 26625 }, { "epoch": 2.8331559906363055, "grad_norm": 12.9765625, "learning_rate": 9.071702671972713e-06, "loss": 0.8708, "step": 26626 }, { "epoch": 2.833262396254522, "grad_norm": 12.234375, "learning_rate": 9.071667140420695e-06, "loss": 1.133, "step": 26627 }, { "epoch": 2.833368801872739, "grad_norm": 13.2734375, "learning_rate": 9.071631608868676e-06, "loss": 0.8893, "step": 26628 }, { "epoch": 2.8334752074909555, "grad_norm": 12.7890625, "learning_rate": 9.071596077316658e-06, "loss": 0.797, "step": 26629 }, { "epoch": 2.833581613109172, "grad_norm": 13.484375, "learning_rate": 9.07156054576464e-06, "loss": 0.9197, "step": 26630 }, { "epoch": 2.833688018727389, "grad_norm": 17.234375, "learning_rate": 9.071525014212621e-06, "loss": 0.8733, "step": 26631 }, { "epoch": 2.8337944243456055, "grad_norm": 12.4375, "learning_rate": 9.071489482660604e-06, "loss": 0.9292, "step": 26632 }, { "epoch": 2.833900829963822, "grad_norm": 12.5390625, "learning_rate": 9.071453951108586e-06, "loss": 0.9066, "step": 26633 }, { "epoch": 2.834007235582039, "grad_norm": 14.234375, "learning_rate": 9.071418419556568e-06, "loss": 0.8385, "step": 26634 }, { "epoch": 2.8341136412002554, "grad_norm": 12.296875, "learning_rate": 9.07138288800455e-06, "loss": 1.1263, "step": 26635 }, { "epoch": 2.834220046818472, "grad_norm": 11.5625, "learning_rate": 9.071347356452531e-06, "loss": 0.7791, "step": 26636 }, { "epoch": 2.834326452436689, "grad_norm": 10.796875, "learning_rate": 9.071311824900512e-06, "loss": 0.8886, "step": 26637 }, { "epoch": 2.8344328580549054, "grad_norm": 14.4296875, "learning_rate": 9.071276293348494e-06, "loss": 0.9871, "step": 26638 }, { "epoch": 2.834539263673122, "grad_norm": 14.9765625, "learning_rate": 9.071240761796476e-06, "loss": 1.0604, "step": 26639 }, { "epoch": 2.8346456692913384, "grad_norm": 10.71875, "learning_rate": 9.071205230244457e-06, "loss": 0.7779, "step": 26640 }, { "epoch": 2.8347520749095554, "grad_norm": 11.7890625, "learning_rate": 9.071169698692439e-06, "loss": 0.8241, "step": 26641 }, { "epoch": 2.834858480527772, "grad_norm": 12.2421875, "learning_rate": 9.071134167140422e-06, "loss": 0.9388, "step": 26642 }, { "epoch": 2.8349648861459884, "grad_norm": 14.5, "learning_rate": 9.071098635588404e-06, "loss": 0.9581, "step": 26643 }, { "epoch": 2.835071291764205, "grad_norm": 13.3828125, "learning_rate": 9.071063104036385e-06, "loss": 1.0501, "step": 26644 }, { "epoch": 2.835177697382422, "grad_norm": 21.1875, "learning_rate": 9.071027572484367e-06, "loss": 0.8212, "step": 26645 }, { "epoch": 2.8352841030006384, "grad_norm": 11.640625, "learning_rate": 9.070992040932349e-06, "loss": 0.9825, "step": 26646 }, { "epoch": 2.835390508618855, "grad_norm": 14.6796875, "learning_rate": 9.07095650938033e-06, "loss": 0.9335, "step": 26647 }, { "epoch": 2.835496914237072, "grad_norm": 14.46875, "learning_rate": 9.070920977828312e-06, "loss": 0.8779, "step": 26648 }, { "epoch": 2.8356033198552884, "grad_norm": 13.7109375, "learning_rate": 9.070885446276293e-06, "loss": 0.884, "step": 26649 }, { "epoch": 2.835709725473505, "grad_norm": 13.1640625, "learning_rate": 9.070849914724275e-06, "loss": 0.9929, "step": 26650 }, { "epoch": 2.835816131091722, "grad_norm": 16.90625, "learning_rate": 9.070814383172257e-06, "loss": 0.8982, "step": 26651 }, { "epoch": 2.8359225367099383, "grad_norm": 12.8046875, "learning_rate": 9.07077885162024e-06, "loss": 0.8949, "step": 26652 }, { "epoch": 2.836028942328155, "grad_norm": 10.171875, "learning_rate": 9.070743320068222e-06, "loss": 1.0917, "step": 26653 }, { "epoch": 2.836135347946372, "grad_norm": 11.2265625, "learning_rate": 9.070707788516203e-06, "loss": 0.7404, "step": 26654 }, { "epoch": 2.8362417535645883, "grad_norm": 11.625, "learning_rate": 9.070672256964185e-06, "loss": 0.9007, "step": 26655 }, { "epoch": 2.836348159182805, "grad_norm": 12.578125, "learning_rate": 9.070636725412166e-06, "loss": 1.0693, "step": 26656 }, { "epoch": 2.8364545648010218, "grad_norm": 10.7578125, "learning_rate": 9.07060119386015e-06, "loss": 0.8644, "step": 26657 }, { "epoch": 2.8365609704192383, "grad_norm": 11.140625, "learning_rate": 9.070565662308131e-06, "loss": 0.8015, "step": 26658 }, { "epoch": 2.836667376037455, "grad_norm": 15.296875, "learning_rate": 9.070530130756113e-06, "loss": 1.0923, "step": 26659 }, { "epoch": 2.8367737816556713, "grad_norm": 12.6875, "learning_rate": 9.070494599204095e-06, "loss": 0.8301, "step": 26660 }, { "epoch": 2.836880187273888, "grad_norm": 10.734375, "learning_rate": 9.070459067652076e-06, "loss": 0.8815, "step": 26661 }, { "epoch": 2.8369865928921048, "grad_norm": 11.7109375, "learning_rate": 9.070423536100058e-06, "loss": 1.0588, "step": 26662 }, { "epoch": 2.8370929985103213, "grad_norm": 12.015625, "learning_rate": 9.07038800454804e-06, "loss": 0.9754, "step": 26663 }, { "epoch": 2.8371994041285378, "grad_norm": 13.546875, "learning_rate": 9.070352472996021e-06, "loss": 0.857, "step": 26664 }, { "epoch": 2.8373058097467547, "grad_norm": 10.7890625, "learning_rate": 9.070316941444003e-06, "loss": 0.8158, "step": 26665 }, { "epoch": 2.8374122153649712, "grad_norm": 11.03125, "learning_rate": 9.070281409891984e-06, "loss": 0.7039, "step": 26666 }, { "epoch": 2.8375186209831877, "grad_norm": 13.7578125, "learning_rate": 9.070245878339967e-06, "loss": 0.8652, "step": 26667 }, { "epoch": 2.8376250266014047, "grad_norm": 12.5859375, "learning_rate": 9.070210346787949e-06, "loss": 0.9572, "step": 26668 }, { "epoch": 2.837731432219621, "grad_norm": 15.4609375, "learning_rate": 9.07017481523593e-06, "loss": 0.6361, "step": 26669 }, { "epoch": 2.8378378378378377, "grad_norm": 11.4921875, "learning_rate": 9.070139283683912e-06, "loss": 0.8997, "step": 26670 }, { "epoch": 2.8379442434560547, "grad_norm": 10.6640625, "learning_rate": 9.070103752131894e-06, "loss": 0.8231, "step": 26671 }, { "epoch": 2.838050649074271, "grad_norm": 14.109375, "learning_rate": 9.070068220579875e-06, "loss": 0.8472, "step": 26672 }, { "epoch": 2.8381570546924877, "grad_norm": 14.6484375, "learning_rate": 9.070032689027857e-06, "loss": 1.1188, "step": 26673 }, { "epoch": 2.8382634603107046, "grad_norm": 12.3984375, "learning_rate": 9.069997157475839e-06, "loss": 0.8279, "step": 26674 }, { "epoch": 2.838369865928921, "grad_norm": 12.4765625, "learning_rate": 9.06996162592382e-06, "loss": 0.8604, "step": 26675 }, { "epoch": 2.8384762715471377, "grad_norm": 13.546875, "learning_rate": 9.069926094371802e-06, "loss": 0.9018, "step": 26676 }, { "epoch": 2.838582677165354, "grad_norm": 13.8671875, "learning_rate": 9.069890562819784e-06, "loss": 0.8759, "step": 26677 }, { "epoch": 2.838689082783571, "grad_norm": 11.4921875, "learning_rate": 9.069855031267767e-06, "loss": 1.0308, "step": 26678 }, { "epoch": 2.8387954884017876, "grad_norm": 12.1796875, "learning_rate": 9.069819499715748e-06, "loss": 0.9363, "step": 26679 }, { "epoch": 2.838901894020004, "grad_norm": 16.78125, "learning_rate": 9.06978396816373e-06, "loss": 0.9908, "step": 26680 }, { "epoch": 2.8390082996382207, "grad_norm": 13.078125, "learning_rate": 9.069748436611712e-06, "loss": 0.8795, "step": 26681 }, { "epoch": 2.8391147052564376, "grad_norm": 13.3828125, "learning_rate": 9.069712905059693e-06, "loss": 0.8913, "step": 26682 }, { "epoch": 2.839221110874654, "grad_norm": 10.1015625, "learning_rate": 9.069677373507677e-06, "loss": 0.8561, "step": 26683 }, { "epoch": 2.8393275164928706, "grad_norm": 10.9453125, "learning_rate": 9.069641841955658e-06, "loss": 0.8971, "step": 26684 }, { "epoch": 2.8394339221110876, "grad_norm": 10.2421875, "learning_rate": 9.06960631040364e-06, "loss": 0.9419, "step": 26685 }, { "epoch": 2.839540327729304, "grad_norm": 16.0, "learning_rate": 9.069570778851621e-06, "loss": 0.9282, "step": 26686 }, { "epoch": 2.8396467333475206, "grad_norm": 13.3359375, "learning_rate": 9.069535247299603e-06, "loss": 0.8704, "step": 26687 }, { "epoch": 2.8397531389657376, "grad_norm": 12.2890625, "learning_rate": 9.069499715747585e-06, "loss": 0.9289, "step": 26688 }, { "epoch": 2.839859544583954, "grad_norm": 12.8984375, "learning_rate": 9.069464184195566e-06, "loss": 1.0832, "step": 26689 }, { "epoch": 2.8399659502021706, "grad_norm": 13.5390625, "learning_rate": 9.069428652643548e-06, "loss": 1.1058, "step": 26690 }, { "epoch": 2.8400723558203875, "grad_norm": 13.4375, "learning_rate": 9.06939312109153e-06, "loss": 0.8267, "step": 26691 }, { "epoch": 2.840178761438604, "grad_norm": 15.3671875, "learning_rate": 9.069357589539511e-06, "loss": 0.9704, "step": 26692 }, { "epoch": 2.8402851670568205, "grad_norm": 12.40625, "learning_rate": 9.069322057987494e-06, "loss": 0.8276, "step": 26693 }, { "epoch": 2.8403915726750375, "grad_norm": 12.09375, "learning_rate": 9.069286526435476e-06, "loss": 0.8678, "step": 26694 }, { "epoch": 2.840497978293254, "grad_norm": 11.265625, "learning_rate": 9.069250994883458e-06, "loss": 0.9311, "step": 26695 }, { "epoch": 2.8406043839114705, "grad_norm": 14.75, "learning_rate": 9.069215463331439e-06, "loss": 0.9446, "step": 26696 }, { "epoch": 2.840710789529687, "grad_norm": 13.6953125, "learning_rate": 9.06917993177942e-06, "loss": 0.9092, "step": 26697 }, { "epoch": 2.8408171951479035, "grad_norm": 13.1484375, "learning_rate": 9.069144400227402e-06, "loss": 0.9059, "step": 26698 }, { "epoch": 2.8409236007661205, "grad_norm": 10.7109375, "learning_rate": 9.069108868675384e-06, "loss": 0.902, "step": 26699 }, { "epoch": 2.841030006384337, "grad_norm": 13.625, "learning_rate": 9.069073337123366e-06, "loss": 0.8552, "step": 26700 }, { "epoch": 2.8411364120025535, "grad_norm": 11.7421875, "learning_rate": 9.069037805571347e-06, "loss": 1.0543, "step": 26701 }, { "epoch": 2.8412428176207705, "grad_norm": 14.171875, "learning_rate": 9.069002274019329e-06, "loss": 0.9194, "step": 26702 }, { "epoch": 2.841349223238987, "grad_norm": 12.0, "learning_rate": 9.068966742467312e-06, "loss": 1.0139, "step": 26703 }, { "epoch": 2.8414556288572035, "grad_norm": 13.9140625, "learning_rate": 9.068931210915294e-06, "loss": 0.8754, "step": 26704 }, { "epoch": 2.8415620344754204, "grad_norm": 11.3828125, "learning_rate": 9.068895679363275e-06, "loss": 0.9665, "step": 26705 }, { "epoch": 2.841668440093637, "grad_norm": 13.0390625, "learning_rate": 9.068860147811257e-06, "loss": 0.9379, "step": 26706 }, { "epoch": 2.8417748457118535, "grad_norm": 10.59375, "learning_rate": 9.068824616259238e-06, "loss": 0.8364, "step": 26707 }, { "epoch": 2.8418812513300704, "grad_norm": 12.7109375, "learning_rate": 9.068789084707222e-06, "loss": 1.0262, "step": 26708 }, { "epoch": 2.841987656948287, "grad_norm": 11.1796875, "learning_rate": 9.068753553155203e-06, "loss": 0.8492, "step": 26709 }, { "epoch": 2.8420940625665034, "grad_norm": 12.078125, "learning_rate": 9.068718021603185e-06, "loss": 0.6369, "step": 26710 }, { "epoch": 2.8422004681847204, "grad_norm": 13.4765625, "learning_rate": 9.068682490051167e-06, "loss": 0.784, "step": 26711 }, { "epoch": 2.842306873802937, "grad_norm": 10.28125, "learning_rate": 9.068646958499148e-06, "loss": 0.7806, "step": 26712 }, { "epoch": 2.8424132794211534, "grad_norm": 14.453125, "learning_rate": 9.06861142694713e-06, "loss": 1.0838, "step": 26713 }, { "epoch": 2.84251968503937, "grad_norm": 15.90625, "learning_rate": 9.068575895395111e-06, "loss": 1.0415, "step": 26714 }, { "epoch": 2.842626090657587, "grad_norm": 12.6171875, "learning_rate": 9.068540363843093e-06, "loss": 0.9806, "step": 26715 }, { "epoch": 2.8427324962758034, "grad_norm": 12.625, "learning_rate": 9.068504832291075e-06, "loss": 0.8113, "step": 26716 }, { "epoch": 2.84283890189402, "grad_norm": 17.078125, "learning_rate": 9.068469300739056e-06, "loss": 0.9066, "step": 26717 }, { "epoch": 2.8429453075122364, "grad_norm": 14.0390625, "learning_rate": 9.068433769187038e-06, "loss": 0.8686, "step": 26718 }, { "epoch": 2.8430517131304534, "grad_norm": 11.9296875, "learning_rate": 9.068398237635021e-06, "loss": 0.9695, "step": 26719 }, { "epoch": 2.84315811874867, "grad_norm": 10.03125, "learning_rate": 9.068362706083003e-06, "loss": 0.6498, "step": 26720 }, { "epoch": 2.8432645243668864, "grad_norm": 13.59375, "learning_rate": 9.068327174530984e-06, "loss": 0.7422, "step": 26721 }, { "epoch": 2.8433709299851033, "grad_norm": 14.7421875, "learning_rate": 9.068291642978966e-06, "loss": 0.9432, "step": 26722 }, { "epoch": 2.84347733560332, "grad_norm": 15.4453125, "learning_rate": 9.068256111426948e-06, "loss": 1.1732, "step": 26723 }, { "epoch": 2.8435837412215363, "grad_norm": 14.234375, "learning_rate": 9.068220579874931e-06, "loss": 0.9541, "step": 26724 }, { "epoch": 2.8436901468397533, "grad_norm": 12.3203125, "learning_rate": 9.068185048322912e-06, "loss": 1.0475, "step": 26725 }, { "epoch": 2.84379655245797, "grad_norm": 16.359375, "learning_rate": 9.068149516770894e-06, "loss": 1.0312, "step": 26726 }, { "epoch": 2.8439029580761863, "grad_norm": 17.703125, "learning_rate": 9.068113985218876e-06, "loss": 0.9535, "step": 26727 }, { "epoch": 2.8440093636944033, "grad_norm": 14.75, "learning_rate": 9.068078453666856e-06, "loss": 0.9946, "step": 26728 }, { "epoch": 2.84411576931262, "grad_norm": 10.140625, "learning_rate": 9.068042922114839e-06, "loss": 0.777, "step": 26729 }, { "epoch": 2.8442221749308363, "grad_norm": 17.1875, "learning_rate": 9.06800739056282e-06, "loss": 0.9044, "step": 26730 }, { "epoch": 2.8443285805490532, "grad_norm": 16.25, "learning_rate": 9.067971859010802e-06, "loss": 0.844, "step": 26731 }, { "epoch": 2.8444349861672698, "grad_norm": 12.34375, "learning_rate": 9.067936327458784e-06, "loss": 0.9473, "step": 26732 }, { "epoch": 2.8445413917854863, "grad_norm": 15.140625, "learning_rate": 9.067900795906765e-06, "loss": 0.913, "step": 26733 }, { "epoch": 2.8446477974037028, "grad_norm": 15.0234375, "learning_rate": 9.067865264354749e-06, "loss": 1.0909, "step": 26734 }, { "epoch": 2.8447542030219193, "grad_norm": 11.3671875, "learning_rate": 9.06782973280273e-06, "loss": 0.8822, "step": 26735 }, { "epoch": 2.8448606086401362, "grad_norm": 16.421875, "learning_rate": 9.067794201250712e-06, "loss": 0.9698, "step": 26736 }, { "epoch": 2.8449670142583527, "grad_norm": 15.90625, "learning_rate": 9.067758669698693e-06, "loss": 0.9034, "step": 26737 }, { "epoch": 2.8450734198765693, "grad_norm": 18.6875, "learning_rate": 9.067723138146675e-06, "loss": 0.8732, "step": 26738 }, { "epoch": 2.845179825494786, "grad_norm": 12.375, "learning_rate": 9.067687606594657e-06, "loss": 0.8115, "step": 26739 }, { "epoch": 2.8452862311130027, "grad_norm": 14.0546875, "learning_rate": 9.067652075042638e-06, "loss": 1.0497, "step": 26740 }, { "epoch": 2.8453926367312192, "grad_norm": 14.0234375, "learning_rate": 9.06761654349062e-06, "loss": 0.9826, "step": 26741 }, { "epoch": 2.845499042349436, "grad_norm": 15.578125, "learning_rate": 9.067581011938602e-06, "loss": 0.8835, "step": 26742 }, { "epoch": 2.8456054479676527, "grad_norm": 14.7734375, "learning_rate": 9.067545480386583e-06, "loss": 0.9085, "step": 26743 }, { "epoch": 2.845711853585869, "grad_norm": 13.671875, "learning_rate": 9.067509948834566e-06, "loss": 1.059, "step": 26744 }, { "epoch": 2.845818259204086, "grad_norm": 14.2734375, "learning_rate": 9.067474417282548e-06, "loss": 0.9107, "step": 26745 }, { "epoch": 2.8459246648223027, "grad_norm": 15.3125, "learning_rate": 9.06743888573053e-06, "loss": 0.9251, "step": 26746 }, { "epoch": 2.846031070440519, "grad_norm": 12.71875, "learning_rate": 9.067403354178511e-06, "loss": 1.0038, "step": 26747 }, { "epoch": 2.846137476058736, "grad_norm": 12.9765625, "learning_rate": 9.067367822626493e-06, "loss": 0.7767, "step": 26748 }, { "epoch": 2.8462438816769526, "grad_norm": 11.78125, "learning_rate": 9.067332291074474e-06, "loss": 0.8437, "step": 26749 }, { "epoch": 2.846350287295169, "grad_norm": 12.859375, "learning_rate": 9.067296759522458e-06, "loss": 0.9707, "step": 26750 }, { "epoch": 2.846456692913386, "grad_norm": 15.875, "learning_rate": 9.06726122797044e-06, "loss": 1.1356, "step": 26751 }, { "epoch": 2.8465630985316026, "grad_norm": 16.0, "learning_rate": 9.067225696418421e-06, "loss": 0.9764, "step": 26752 }, { "epoch": 2.846669504149819, "grad_norm": 11.453125, "learning_rate": 9.067190164866403e-06, "loss": 0.9922, "step": 26753 }, { "epoch": 2.8467759097680356, "grad_norm": 16.296875, "learning_rate": 9.067154633314384e-06, "loss": 1.0314, "step": 26754 }, { "epoch": 2.846882315386252, "grad_norm": 14.5625, "learning_rate": 9.067119101762366e-06, "loss": 1.0022, "step": 26755 }, { "epoch": 2.846988721004469, "grad_norm": 12.9765625, "learning_rate": 9.067083570210347e-06, "loss": 0.7721, "step": 26756 }, { "epoch": 2.8470951266226856, "grad_norm": 9.84375, "learning_rate": 9.067048038658329e-06, "loss": 0.8352, "step": 26757 }, { "epoch": 2.847201532240902, "grad_norm": 18.5, "learning_rate": 9.06701250710631e-06, "loss": 0.8742, "step": 26758 }, { "epoch": 2.847307937859119, "grad_norm": 13.4375, "learning_rate": 9.066976975554292e-06, "loss": 0.9987, "step": 26759 }, { "epoch": 2.8474143434773356, "grad_norm": 17.125, "learning_rate": 9.066941444002276e-06, "loss": 1.0142, "step": 26760 }, { "epoch": 2.847520749095552, "grad_norm": 13.96875, "learning_rate": 9.066905912450257e-06, "loss": 1.0502, "step": 26761 }, { "epoch": 2.847627154713769, "grad_norm": 14.3046875, "learning_rate": 9.066870380898239e-06, "loss": 1.01, "step": 26762 }, { "epoch": 2.8477335603319855, "grad_norm": 12.7265625, "learning_rate": 9.06683484934622e-06, "loss": 0.7741, "step": 26763 }, { "epoch": 2.847839965950202, "grad_norm": 12.671875, "learning_rate": 9.066799317794202e-06, "loss": 0.8804, "step": 26764 }, { "epoch": 2.847946371568419, "grad_norm": 13.7109375, "learning_rate": 9.066763786242184e-06, "loss": 0.9877, "step": 26765 }, { "epoch": 2.8480527771866355, "grad_norm": 13.3046875, "learning_rate": 9.066728254690165e-06, "loss": 0.9097, "step": 26766 }, { "epoch": 2.848159182804852, "grad_norm": 14.8203125, "learning_rate": 9.066692723138147e-06, "loss": 0.9273, "step": 26767 }, { "epoch": 2.848265588423069, "grad_norm": 12.421875, "learning_rate": 9.066657191586128e-06, "loss": 0.8834, "step": 26768 }, { "epoch": 2.8483719940412855, "grad_norm": 12.4765625, "learning_rate": 9.06662166003411e-06, "loss": 0.955, "step": 26769 }, { "epoch": 2.848478399659502, "grad_norm": 13.7421875, "learning_rate": 9.066586128482093e-06, "loss": 0.9849, "step": 26770 }, { "epoch": 2.8485848052777185, "grad_norm": 16.8125, "learning_rate": 9.066550596930075e-06, "loss": 0.9765, "step": 26771 }, { "epoch": 2.8486912108959355, "grad_norm": 11.1640625, "learning_rate": 9.066515065378056e-06, "loss": 0.8954, "step": 26772 }, { "epoch": 2.848797616514152, "grad_norm": 18.53125, "learning_rate": 9.066479533826038e-06, "loss": 1.0019, "step": 26773 }, { "epoch": 2.8489040221323685, "grad_norm": 13.7578125, "learning_rate": 9.06644400227402e-06, "loss": 0.9218, "step": 26774 }, { "epoch": 2.849010427750585, "grad_norm": 11.609375, "learning_rate": 9.066408470722003e-06, "loss": 0.9606, "step": 26775 }, { "epoch": 2.849116833368802, "grad_norm": 20.046875, "learning_rate": 9.066372939169985e-06, "loss": 0.9254, "step": 26776 }, { "epoch": 2.8492232389870185, "grad_norm": 14.265625, "learning_rate": 9.066337407617966e-06, "loss": 0.8866, "step": 26777 }, { "epoch": 2.849329644605235, "grad_norm": 17.078125, "learning_rate": 9.066301876065948e-06, "loss": 0.9987, "step": 26778 }, { "epoch": 2.849436050223452, "grad_norm": 10.5234375, "learning_rate": 9.06626634451393e-06, "loss": 0.8208, "step": 26779 }, { "epoch": 2.8495424558416684, "grad_norm": 14.6015625, "learning_rate": 9.066230812961911e-06, "loss": 0.8902, "step": 26780 }, { "epoch": 2.849648861459885, "grad_norm": 16.15625, "learning_rate": 9.066195281409893e-06, "loss": 0.7376, "step": 26781 }, { "epoch": 2.849755267078102, "grad_norm": 14.71875, "learning_rate": 9.066159749857874e-06, "loss": 0.9421, "step": 26782 }, { "epoch": 2.8498616726963184, "grad_norm": 14.3203125, "learning_rate": 9.066124218305856e-06, "loss": 1.0887, "step": 26783 }, { "epoch": 2.849968078314535, "grad_norm": 14.3203125, "learning_rate": 9.066088686753837e-06, "loss": 1.1113, "step": 26784 }, { "epoch": 2.850074483932752, "grad_norm": 11.0234375, "learning_rate": 9.06605315520182e-06, "loss": 0.9037, "step": 26785 }, { "epoch": 2.8501808895509684, "grad_norm": 10.6640625, "learning_rate": 9.066017623649802e-06, "loss": 0.9012, "step": 26786 }, { "epoch": 2.850287295169185, "grad_norm": 11.8671875, "learning_rate": 9.065982092097784e-06, "loss": 0.9124, "step": 26787 }, { "epoch": 2.850393700787402, "grad_norm": 17.71875, "learning_rate": 9.065946560545766e-06, "loss": 0.9563, "step": 26788 }, { "epoch": 2.8505001064056184, "grad_norm": 10.109375, "learning_rate": 9.065911028993747e-06, "loss": 0.8975, "step": 26789 }, { "epoch": 2.850606512023835, "grad_norm": 14.0078125, "learning_rate": 9.065875497441729e-06, "loss": 0.7678, "step": 26790 }, { "epoch": 2.8507129176420514, "grad_norm": 15.6015625, "learning_rate": 9.06583996588971e-06, "loss": 0.9198, "step": 26791 }, { "epoch": 2.850819323260268, "grad_norm": 12.375, "learning_rate": 9.065804434337692e-06, "loss": 0.8431, "step": 26792 }, { "epoch": 2.850925728878485, "grad_norm": 12.796875, "learning_rate": 9.065768902785674e-06, "loss": 0.8598, "step": 26793 }, { "epoch": 2.8510321344967013, "grad_norm": 18.0, "learning_rate": 9.065733371233655e-06, "loss": 0.8447, "step": 26794 }, { "epoch": 2.851138540114918, "grad_norm": 14.890625, "learning_rate": 9.065697839681637e-06, "loss": 0.9376, "step": 26795 }, { "epoch": 2.851244945733135, "grad_norm": 14.1796875, "learning_rate": 9.06566230812962e-06, "loss": 0.8104, "step": 26796 }, { "epoch": 2.8513513513513513, "grad_norm": 12.9140625, "learning_rate": 9.065626776577602e-06, "loss": 0.8582, "step": 26797 }, { "epoch": 2.851457756969568, "grad_norm": 13.8046875, "learning_rate": 9.065591245025583e-06, "loss": 0.8945, "step": 26798 }, { "epoch": 2.851564162587785, "grad_norm": 16.8125, "learning_rate": 9.065555713473565e-06, "loss": 1.0094, "step": 26799 }, { "epoch": 2.8516705682060013, "grad_norm": 12.6640625, "learning_rate": 9.065520181921547e-06, "loss": 1.0174, "step": 26800 }, { "epoch": 2.851776973824218, "grad_norm": 10.8359375, "learning_rate": 9.06548465036953e-06, "loss": 1.0097, "step": 26801 }, { "epoch": 2.8518833794424348, "grad_norm": 15.640625, "learning_rate": 9.065449118817511e-06, "loss": 0.9386, "step": 26802 }, { "epoch": 2.8519897850606513, "grad_norm": 14.2578125, "learning_rate": 9.065413587265493e-06, "loss": 0.769, "step": 26803 }, { "epoch": 2.8520961906788678, "grad_norm": 16.25, "learning_rate": 9.065378055713475e-06, "loss": 0.7831, "step": 26804 }, { "epoch": 2.8522025962970847, "grad_norm": 19.203125, "learning_rate": 9.065342524161456e-06, "loss": 0.9379, "step": 26805 }, { "epoch": 2.8523090019153012, "grad_norm": 12.0390625, "learning_rate": 9.065306992609438e-06, "loss": 0.7762, "step": 26806 }, { "epoch": 2.8524154075335177, "grad_norm": 10.796875, "learning_rate": 9.06527146105742e-06, "loss": 0.7463, "step": 26807 }, { "epoch": 2.8525218131517343, "grad_norm": 17.0, "learning_rate": 9.065235929505401e-06, "loss": 0.9696, "step": 26808 }, { "epoch": 2.852628218769951, "grad_norm": 14.703125, "learning_rate": 9.065200397953383e-06, "loss": 1.1, "step": 26809 }, { "epoch": 2.8527346243881677, "grad_norm": 14.0234375, "learning_rate": 9.065164866401364e-06, "loss": 1.0069, "step": 26810 }, { "epoch": 2.8528410300063842, "grad_norm": 10.390625, "learning_rate": 9.065129334849348e-06, "loss": 0.7824, "step": 26811 }, { "epoch": 2.8529474356246007, "grad_norm": 13.171875, "learning_rate": 9.06509380329733e-06, "loss": 0.9286, "step": 26812 }, { "epoch": 2.8530538412428177, "grad_norm": 12.7265625, "learning_rate": 9.06505827174531e-06, "loss": 0.9737, "step": 26813 }, { "epoch": 2.853160246861034, "grad_norm": 12.3671875, "learning_rate": 9.065022740193292e-06, "loss": 0.7375, "step": 26814 }, { "epoch": 2.8532666524792507, "grad_norm": 19.546875, "learning_rate": 9.064987208641274e-06, "loss": 0.8417, "step": 26815 }, { "epoch": 2.8533730580974677, "grad_norm": 20.78125, "learning_rate": 9.064951677089257e-06, "loss": 1.024, "step": 26816 }, { "epoch": 2.853479463715684, "grad_norm": 11.1015625, "learning_rate": 9.064916145537237e-06, "loss": 0.7929, "step": 26817 }, { "epoch": 2.8535858693339007, "grad_norm": 10.53125, "learning_rate": 9.064880613985219e-06, "loss": 0.8236, "step": 26818 }, { "epoch": 2.8536922749521176, "grad_norm": 11.625, "learning_rate": 9.0648450824332e-06, "loss": 0.8509, "step": 26819 }, { "epoch": 2.853798680570334, "grad_norm": 11.1171875, "learning_rate": 9.064809550881182e-06, "loss": 0.9408, "step": 26820 }, { "epoch": 2.8539050861885507, "grad_norm": 10.21875, "learning_rate": 9.064774019329165e-06, "loss": 0.7875, "step": 26821 }, { "epoch": 2.8540114918067676, "grad_norm": 11.3828125, "learning_rate": 9.064738487777147e-06, "loss": 0.8051, "step": 26822 }, { "epoch": 2.854117897424984, "grad_norm": 13.3828125, "learning_rate": 9.064702956225129e-06, "loss": 0.9967, "step": 26823 }, { "epoch": 2.8542243030432006, "grad_norm": 14.9296875, "learning_rate": 9.06466742467311e-06, "loss": 1.094, "step": 26824 }, { "epoch": 2.8543307086614176, "grad_norm": 12.921875, "learning_rate": 9.064631893121092e-06, "loss": 0.828, "step": 26825 }, { "epoch": 2.854437114279634, "grad_norm": 21.9375, "learning_rate": 9.064596361569075e-06, "loss": 0.9747, "step": 26826 }, { "epoch": 2.8545435198978506, "grad_norm": 13.234375, "learning_rate": 9.064560830017057e-06, "loss": 0.941, "step": 26827 }, { "epoch": 2.854649925516067, "grad_norm": 12.7734375, "learning_rate": 9.064525298465038e-06, "loss": 0.9818, "step": 26828 }, { "epoch": 2.8547563311342836, "grad_norm": 13.546875, "learning_rate": 9.06448976691302e-06, "loss": 0.9349, "step": 26829 }, { "epoch": 2.8548627367525006, "grad_norm": 15.5625, "learning_rate": 9.064454235361002e-06, "loss": 0.9555, "step": 26830 }, { "epoch": 2.854969142370717, "grad_norm": 14.328125, "learning_rate": 9.064418703808983e-06, "loss": 0.8963, "step": 26831 }, { "epoch": 2.8550755479889336, "grad_norm": 10.640625, "learning_rate": 9.064383172256965e-06, "loss": 0.8119, "step": 26832 }, { "epoch": 2.8551819536071505, "grad_norm": 11.3984375, "learning_rate": 9.064347640704946e-06, "loss": 0.9579, "step": 26833 }, { "epoch": 2.855288359225367, "grad_norm": 10.921875, "learning_rate": 9.064312109152928e-06, "loss": 0.7557, "step": 26834 }, { "epoch": 2.8553947648435836, "grad_norm": 23.734375, "learning_rate": 9.06427657760091e-06, "loss": 1.146, "step": 26835 }, { "epoch": 2.8555011704618005, "grad_norm": 10.0546875, "learning_rate": 9.064241046048891e-06, "loss": 0.8007, "step": 26836 }, { "epoch": 2.855607576080017, "grad_norm": 12.53125, "learning_rate": 9.064205514496874e-06, "loss": 1.0933, "step": 26837 }, { "epoch": 2.8557139816982335, "grad_norm": 14.7109375, "learning_rate": 9.064169982944856e-06, "loss": 0.9501, "step": 26838 }, { "epoch": 2.8558203873164505, "grad_norm": 16.609375, "learning_rate": 9.064134451392838e-06, "loss": 0.9836, "step": 26839 }, { "epoch": 2.855926792934667, "grad_norm": 14.65625, "learning_rate": 9.06409891984082e-06, "loss": 1.0858, "step": 26840 }, { "epoch": 2.8560331985528835, "grad_norm": 13.5546875, "learning_rate": 9.064063388288801e-06, "loss": 0.858, "step": 26841 }, { "epoch": 2.8561396041711005, "grad_norm": 14.4921875, "learning_rate": 9.064027856736784e-06, "loss": 1.0034, "step": 26842 }, { "epoch": 2.856246009789317, "grad_norm": 17.5625, "learning_rate": 9.063992325184766e-06, "loss": 0.8998, "step": 26843 }, { "epoch": 2.8563524154075335, "grad_norm": 12.2890625, "learning_rate": 9.063956793632747e-06, "loss": 0.9299, "step": 26844 }, { "epoch": 2.85645882102575, "grad_norm": 13.4765625, "learning_rate": 9.063921262080729e-06, "loss": 0.9237, "step": 26845 }, { "epoch": 2.856565226643967, "grad_norm": 14.15625, "learning_rate": 9.06388573052871e-06, "loss": 0.9313, "step": 26846 }, { "epoch": 2.8566716322621835, "grad_norm": 18.78125, "learning_rate": 9.063850198976692e-06, "loss": 1.1435, "step": 26847 }, { "epoch": 2.8567780378804, "grad_norm": 11.71875, "learning_rate": 9.063814667424674e-06, "loss": 0.8237, "step": 26848 }, { "epoch": 2.8568844434986165, "grad_norm": 16.765625, "learning_rate": 9.063779135872655e-06, "loss": 0.9942, "step": 26849 }, { "epoch": 2.8569908491168334, "grad_norm": 17.734375, "learning_rate": 9.063743604320637e-06, "loss": 1.0177, "step": 26850 }, { "epoch": 2.85709725473505, "grad_norm": 12.15625, "learning_rate": 9.063708072768619e-06, "loss": 1.0577, "step": 26851 }, { "epoch": 2.8572036603532665, "grad_norm": 13.1875, "learning_rate": 9.063672541216602e-06, "loss": 0.9607, "step": 26852 }, { "epoch": 2.8573100659714834, "grad_norm": 12.84375, "learning_rate": 9.063637009664584e-06, "loss": 0.851, "step": 26853 }, { "epoch": 2.8574164715897, "grad_norm": 16.09375, "learning_rate": 9.063601478112565e-06, "loss": 0.9339, "step": 26854 }, { "epoch": 2.8575228772079164, "grad_norm": 17.765625, "learning_rate": 9.063565946560547e-06, "loss": 1.234, "step": 26855 }, { "epoch": 2.8576292828261334, "grad_norm": 14.8359375, "learning_rate": 9.063530415008528e-06, "loss": 1.014, "step": 26856 }, { "epoch": 2.85773568844435, "grad_norm": 15.515625, "learning_rate": 9.06349488345651e-06, "loss": 0.939, "step": 26857 }, { "epoch": 2.8578420940625664, "grad_norm": 12.9140625, "learning_rate": 9.063459351904492e-06, "loss": 0.8502, "step": 26858 }, { "epoch": 2.8579484996807833, "grad_norm": 13.6875, "learning_rate": 9.063423820352473e-06, "loss": 0.8784, "step": 26859 }, { "epoch": 2.858054905299, "grad_norm": 13.8203125, "learning_rate": 9.063388288800455e-06, "loss": 0.9898, "step": 26860 }, { "epoch": 2.8581613109172164, "grad_norm": 16.46875, "learning_rate": 9.063352757248436e-06, "loss": 1.0448, "step": 26861 }, { "epoch": 2.8582677165354333, "grad_norm": 13.328125, "learning_rate": 9.06331722569642e-06, "loss": 0.7687, "step": 26862 }, { "epoch": 2.85837412215365, "grad_norm": 14.1796875, "learning_rate": 9.063281694144401e-06, "loss": 0.8961, "step": 26863 }, { "epoch": 2.8584805277718663, "grad_norm": 14.2109375, "learning_rate": 9.063246162592383e-06, "loss": 0.9719, "step": 26864 }, { "epoch": 2.858586933390083, "grad_norm": 23.59375, "learning_rate": 9.063210631040365e-06, "loss": 1.1679, "step": 26865 }, { "epoch": 2.8586933390082994, "grad_norm": 15.4296875, "learning_rate": 9.063175099488346e-06, "loss": 1.1344, "step": 26866 }, { "epoch": 2.8587997446265163, "grad_norm": 11.1484375, "learning_rate": 9.063139567936328e-06, "loss": 0.965, "step": 26867 }, { "epoch": 2.858906150244733, "grad_norm": 10.46875, "learning_rate": 9.063104036384311e-06, "loss": 0.788, "step": 26868 }, { "epoch": 2.8590125558629493, "grad_norm": 18.140625, "learning_rate": 9.063068504832293e-06, "loss": 0.9544, "step": 26869 }, { "epoch": 2.8591189614811663, "grad_norm": 12.5390625, "learning_rate": 9.063032973280274e-06, "loss": 0.7126, "step": 26870 }, { "epoch": 2.859225367099383, "grad_norm": 14.1953125, "learning_rate": 9.062997441728256e-06, "loss": 0.9534, "step": 26871 }, { "epoch": 2.8593317727175993, "grad_norm": 13.5625, "learning_rate": 9.062961910176237e-06, "loss": 1.0645, "step": 26872 }, { "epoch": 2.8594381783358163, "grad_norm": 9.6875, "learning_rate": 9.062926378624219e-06, "loss": 0.5354, "step": 26873 }, { "epoch": 2.8595445839540328, "grad_norm": 12.421875, "learning_rate": 9.0628908470722e-06, "loss": 0.817, "step": 26874 }, { "epoch": 2.8596509895722493, "grad_norm": 14.5859375, "learning_rate": 9.062855315520182e-06, "loss": 0.9712, "step": 26875 }, { "epoch": 2.8597573951904662, "grad_norm": 17.0, "learning_rate": 9.062819783968164e-06, "loss": 0.9885, "step": 26876 }, { "epoch": 2.8598638008086827, "grad_norm": 15.3984375, "learning_rate": 9.062784252416146e-06, "loss": 1.0517, "step": 26877 }, { "epoch": 2.8599702064268993, "grad_norm": 16.046875, "learning_rate": 9.062748720864129e-06, "loss": 1.0987, "step": 26878 }, { "epoch": 2.860076612045116, "grad_norm": 15.40625, "learning_rate": 9.06271318931211e-06, "loss": 1.1305, "step": 26879 }, { "epoch": 2.8601830176633327, "grad_norm": 12.9921875, "learning_rate": 9.062677657760092e-06, "loss": 0.7907, "step": 26880 }, { "epoch": 2.8602894232815492, "grad_norm": 12.609375, "learning_rate": 9.062642126208074e-06, "loss": 0.9538, "step": 26881 }, { "epoch": 2.860395828899766, "grad_norm": 12.2578125, "learning_rate": 9.062606594656055e-06, "loss": 0.8256, "step": 26882 }, { "epoch": 2.8605022345179827, "grad_norm": 13.4609375, "learning_rate": 9.062571063104037e-06, "loss": 0.973, "step": 26883 }, { "epoch": 2.860608640136199, "grad_norm": 13.6796875, "learning_rate": 9.062535531552018e-06, "loss": 1.0529, "step": 26884 }, { "epoch": 2.8607150457544157, "grad_norm": 15.5859375, "learning_rate": 9.0625e-06, "loss": 0.9738, "step": 26885 }, { "epoch": 2.860821451372632, "grad_norm": 15.578125, "learning_rate": 9.062464468447982e-06, "loss": 1.1941, "step": 26886 }, { "epoch": 2.860927856990849, "grad_norm": 10.515625, "learning_rate": 9.062428936895963e-06, "loss": 0.9159, "step": 26887 }, { "epoch": 2.8610342626090657, "grad_norm": 20.453125, "learning_rate": 9.062393405343947e-06, "loss": 0.8237, "step": 26888 }, { "epoch": 2.861140668227282, "grad_norm": 11.4921875, "learning_rate": 9.062357873791928e-06, "loss": 0.8898, "step": 26889 }, { "epoch": 2.861247073845499, "grad_norm": 14.28125, "learning_rate": 9.06232234223991e-06, "loss": 0.919, "step": 26890 }, { "epoch": 2.8613534794637157, "grad_norm": 12.484375, "learning_rate": 9.062286810687891e-06, "loss": 1.0089, "step": 26891 }, { "epoch": 2.861459885081932, "grad_norm": 11.8984375, "learning_rate": 9.062251279135873e-06, "loss": 0.8903, "step": 26892 }, { "epoch": 2.861566290700149, "grad_norm": 17.3125, "learning_rate": 9.062215747583856e-06, "loss": 0.8068, "step": 26893 }, { "epoch": 2.8616726963183656, "grad_norm": 17.859375, "learning_rate": 9.062180216031838e-06, "loss": 0.9478, "step": 26894 }, { "epoch": 2.861779101936582, "grad_norm": 13.328125, "learning_rate": 9.06214468447982e-06, "loss": 0.9745, "step": 26895 }, { "epoch": 2.861885507554799, "grad_norm": 8.78125, "learning_rate": 9.062109152927801e-06, "loss": 0.7937, "step": 26896 }, { "epoch": 2.8619919131730156, "grad_norm": 12.8203125, "learning_rate": 9.062073621375783e-06, "loss": 0.8344, "step": 26897 }, { "epoch": 2.862098318791232, "grad_norm": 11.46875, "learning_rate": 9.062038089823764e-06, "loss": 0.8332, "step": 26898 }, { "epoch": 2.862204724409449, "grad_norm": 11.0546875, "learning_rate": 9.062002558271746e-06, "loss": 0.9777, "step": 26899 }, { "epoch": 2.8623111300276656, "grad_norm": 15.640625, "learning_rate": 9.061967026719728e-06, "loss": 1.1092, "step": 26900 }, { "epoch": 2.862417535645882, "grad_norm": 11.328125, "learning_rate": 9.061931495167709e-06, "loss": 0.8623, "step": 26901 }, { "epoch": 2.8625239412640986, "grad_norm": 13.96875, "learning_rate": 9.06189596361569e-06, "loss": 1.0074, "step": 26902 }, { "epoch": 2.8626303468823155, "grad_norm": 12.671875, "learning_rate": 9.061860432063674e-06, "loss": 0.8834, "step": 26903 }, { "epoch": 2.862736752500532, "grad_norm": 12.65625, "learning_rate": 9.061824900511656e-06, "loss": 0.9397, "step": 26904 }, { "epoch": 2.8628431581187486, "grad_norm": 14.5546875, "learning_rate": 9.061789368959637e-06, "loss": 0.9513, "step": 26905 }, { "epoch": 2.862949563736965, "grad_norm": 15.9375, "learning_rate": 9.061753837407619e-06, "loss": 0.7727, "step": 26906 }, { "epoch": 2.863055969355182, "grad_norm": 13.375, "learning_rate": 9.0617183058556e-06, "loss": 0.8358, "step": 26907 }, { "epoch": 2.8631623749733985, "grad_norm": 10.8515625, "learning_rate": 9.061682774303582e-06, "loss": 0.7797, "step": 26908 }, { "epoch": 2.863268780591615, "grad_norm": 12.1953125, "learning_rate": 9.061647242751564e-06, "loss": 0.9899, "step": 26909 }, { "epoch": 2.863375186209832, "grad_norm": 24.359375, "learning_rate": 9.061611711199545e-06, "loss": 0.8613, "step": 26910 }, { "epoch": 2.8634815918280485, "grad_norm": 15.9921875, "learning_rate": 9.061576179647527e-06, "loss": 0.8857, "step": 26911 }, { "epoch": 2.863587997446265, "grad_norm": 15.84375, "learning_rate": 9.061540648095509e-06, "loss": 1.1135, "step": 26912 }, { "epoch": 2.863694403064482, "grad_norm": 10.015625, "learning_rate": 9.06150511654349e-06, "loss": 0.6739, "step": 26913 }, { "epoch": 2.8638008086826985, "grad_norm": 12.7578125, "learning_rate": 9.061469584991473e-06, "loss": 0.8783, "step": 26914 }, { "epoch": 2.863907214300915, "grad_norm": 12.5703125, "learning_rate": 9.061434053439455e-06, "loss": 1.0108, "step": 26915 }, { "epoch": 2.864013619919132, "grad_norm": 11.9296875, "learning_rate": 9.061398521887437e-06, "loss": 0.9403, "step": 26916 }, { "epoch": 2.8641200255373485, "grad_norm": 15.390625, "learning_rate": 9.061362990335418e-06, "loss": 1.0554, "step": 26917 }, { "epoch": 2.864226431155565, "grad_norm": 12.6328125, "learning_rate": 9.0613274587834e-06, "loss": 1.0342, "step": 26918 }, { "epoch": 2.864332836773782, "grad_norm": 10.609375, "learning_rate": 9.061291927231383e-06, "loss": 0.6795, "step": 26919 }, { "epoch": 2.8644392423919984, "grad_norm": 12.8828125, "learning_rate": 9.061256395679365e-06, "loss": 0.6856, "step": 26920 }, { "epoch": 2.864545648010215, "grad_norm": 10.9453125, "learning_rate": 9.061220864127346e-06, "loss": 0.8727, "step": 26921 }, { "epoch": 2.8646520536284314, "grad_norm": 24.921875, "learning_rate": 9.061185332575328e-06, "loss": 1.1631, "step": 26922 }, { "epoch": 2.864758459246648, "grad_norm": 11.453125, "learning_rate": 9.06114980102331e-06, "loss": 0.9254, "step": 26923 }, { "epoch": 2.864864864864865, "grad_norm": 13.375, "learning_rate": 9.061114269471291e-06, "loss": 0.9414, "step": 26924 }, { "epoch": 2.8649712704830814, "grad_norm": 15.25, "learning_rate": 9.061078737919273e-06, "loss": 0.8604, "step": 26925 }, { "epoch": 2.865077676101298, "grad_norm": 14.5, "learning_rate": 9.061043206367254e-06, "loss": 1.0007, "step": 26926 }, { "epoch": 2.865184081719515, "grad_norm": 14.359375, "learning_rate": 9.061007674815236e-06, "loss": 0.9728, "step": 26927 }, { "epoch": 2.8652904873377314, "grad_norm": 17.3125, "learning_rate": 9.060972143263218e-06, "loss": 0.8678, "step": 26928 }, { "epoch": 2.865396892955948, "grad_norm": 13.453125, "learning_rate": 9.060936611711201e-06, "loss": 0.8449, "step": 26929 }, { "epoch": 2.865503298574165, "grad_norm": 13.578125, "learning_rate": 9.060901080159183e-06, "loss": 0.8743, "step": 26930 }, { "epoch": 2.8656097041923814, "grad_norm": 13.953125, "learning_rate": 9.060865548607164e-06, "loss": 0.8891, "step": 26931 }, { "epoch": 2.865716109810598, "grad_norm": 16.765625, "learning_rate": 9.060830017055146e-06, "loss": 0.7617, "step": 26932 }, { "epoch": 2.865822515428815, "grad_norm": 14.109375, "learning_rate": 9.060794485503127e-06, "loss": 0.9855, "step": 26933 }, { "epoch": 2.8659289210470313, "grad_norm": 11.8125, "learning_rate": 9.06075895395111e-06, "loss": 0.946, "step": 26934 }, { "epoch": 2.866035326665248, "grad_norm": 13.109375, "learning_rate": 9.060723422399092e-06, "loss": 0.9892, "step": 26935 }, { "epoch": 2.866141732283465, "grad_norm": 10.703125, "learning_rate": 9.060687890847074e-06, "loss": 0.9269, "step": 26936 }, { "epoch": 2.8662481379016813, "grad_norm": 10.1640625, "learning_rate": 9.060652359295054e-06, "loss": 0.881, "step": 26937 }, { "epoch": 2.866354543519898, "grad_norm": 15.515625, "learning_rate": 9.060616827743035e-06, "loss": 0.9573, "step": 26938 }, { "epoch": 2.8664609491381143, "grad_norm": 16.375, "learning_rate": 9.060581296191019e-06, "loss": 1.1847, "step": 26939 }, { "epoch": 2.8665673547563313, "grad_norm": 9.703125, "learning_rate": 9.060545764639e-06, "loss": 0.7836, "step": 26940 }, { "epoch": 2.866673760374548, "grad_norm": 16.765625, "learning_rate": 9.060510233086982e-06, "loss": 0.8719, "step": 26941 }, { "epoch": 2.8667801659927643, "grad_norm": 12.40625, "learning_rate": 9.060474701534963e-06, "loss": 0.9508, "step": 26942 }, { "epoch": 2.866886571610981, "grad_norm": 9.84375, "learning_rate": 9.060439169982945e-06, "loss": 0.7843, "step": 26943 }, { "epoch": 2.8669929772291978, "grad_norm": 11.984375, "learning_rate": 9.060403638430927e-06, "loss": 0.9797, "step": 26944 }, { "epoch": 2.8670993828474143, "grad_norm": 11.9375, "learning_rate": 9.06036810687891e-06, "loss": 0.9246, "step": 26945 }, { "epoch": 2.867205788465631, "grad_norm": 13.75, "learning_rate": 9.060332575326892e-06, "loss": 1.1246, "step": 26946 }, { "epoch": 2.8673121940838477, "grad_norm": 10.7734375, "learning_rate": 9.060297043774873e-06, "loss": 0.8547, "step": 26947 }, { "epoch": 2.8674185997020643, "grad_norm": 12.2265625, "learning_rate": 9.060261512222855e-06, "loss": 1.0115, "step": 26948 }, { "epoch": 2.8675250053202808, "grad_norm": 13.4921875, "learning_rate": 9.060225980670836e-06, "loss": 0.9149, "step": 26949 }, { "epoch": 2.8676314109384977, "grad_norm": 11.5859375, "learning_rate": 9.060190449118818e-06, "loss": 0.8049, "step": 26950 }, { "epoch": 2.8677378165567142, "grad_norm": 14.5546875, "learning_rate": 9.0601549175668e-06, "loss": 1.1028, "step": 26951 }, { "epoch": 2.8678442221749307, "grad_norm": 15.8203125, "learning_rate": 9.060119386014781e-06, "loss": 0.9817, "step": 26952 }, { "epoch": 2.8679506277931477, "grad_norm": 10.71875, "learning_rate": 9.060083854462763e-06, "loss": 0.9401, "step": 26953 }, { "epoch": 2.868057033411364, "grad_norm": 15.9921875, "learning_rate": 9.060048322910744e-06, "loss": 0.8724, "step": 26954 }, { "epoch": 2.8681634390295807, "grad_norm": 13.6484375, "learning_rate": 9.060012791358728e-06, "loss": 0.9817, "step": 26955 }, { "epoch": 2.8682698446477977, "grad_norm": 11.3828125, "learning_rate": 9.05997725980671e-06, "loss": 0.8566, "step": 26956 }, { "epoch": 2.868376250266014, "grad_norm": 14.328125, "learning_rate": 9.059941728254691e-06, "loss": 1.0114, "step": 26957 }, { "epoch": 2.8684826558842307, "grad_norm": 12.765625, "learning_rate": 9.059906196702673e-06, "loss": 0.7483, "step": 26958 }, { "epoch": 2.868589061502447, "grad_norm": 14.359375, "learning_rate": 9.059870665150654e-06, "loss": 1.0972, "step": 26959 }, { "epoch": 2.8686954671206637, "grad_norm": 15.265625, "learning_rate": 9.059835133598637e-06, "loss": 0.8253, "step": 26960 }, { "epoch": 2.8688018727388807, "grad_norm": 13.171875, "learning_rate": 9.059799602046619e-06, "loss": 0.9831, "step": 26961 }, { "epoch": 2.868908278357097, "grad_norm": 13.9921875, "learning_rate": 9.0597640704946e-06, "loss": 0.9785, "step": 26962 }, { "epoch": 2.8690146839753137, "grad_norm": 14.6640625, "learning_rate": 9.059728538942582e-06, "loss": 0.8336, "step": 26963 }, { "epoch": 2.8691210895935306, "grad_norm": 11.296875, "learning_rate": 9.059693007390564e-06, "loss": 0.8357, "step": 26964 }, { "epoch": 2.869227495211747, "grad_norm": 9.0546875, "learning_rate": 9.059657475838546e-06, "loss": 0.8169, "step": 26965 }, { "epoch": 2.8693339008299636, "grad_norm": 10.3203125, "learning_rate": 9.059621944286527e-06, "loss": 0.8105, "step": 26966 }, { "epoch": 2.8694403064481806, "grad_norm": 11.96875, "learning_rate": 9.059586412734509e-06, "loss": 0.7397, "step": 26967 }, { "epoch": 2.869546712066397, "grad_norm": 15.9921875, "learning_rate": 9.05955088118249e-06, "loss": 0.9276, "step": 26968 }, { "epoch": 2.8696531176846136, "grad_norm": 11.1171875, "learning_rate": 9.059515349630472e-06, "loss": 0.9027, "step": 26969 }, { "epoch": 2.8697595233028306, "grad_norm": 11.53125, "learning_rate": 9.059479818078455e-06, "loss": 0.9192, "step": 26970 }, { "epoch": 2.869865928921047, "grad_norm": 13.015625, "learning_rate": 9.059444286526437e-06, "loss": 0.875, "step": 26971 }, { "epoch": 2.8699723345392636, "grad_norm": 10.5625, "learning_rate": 9.059408754974418e-06, "loss": 0.9596, "step": 26972 }, { "epoch": 2.8700787401574805, "grad_norm": 12.3046875, "learning_rate": 9.0593732234224e-06, "loss": 0.9859, "step": 26973 }, { "epoch": 2.870185145775697, "grad_norm": 13.7421875, "learning_rate": 9.059337691870382e-06, "loss": 0.9972, "step": 26974 }, { "epoch": 2.8702915513939136, "grad_norm": 15.3359375, "learning_rate": 9.059302160318363e-06, "loss": 0.7872, "step": 26975 }, { "epoch": 2.87039795701213, "grad_norm": 14.921875, "learning_rate": 9.059266628766345e-06, "loss": 1.1369, "step": 26976 }, { "epoch": 2.870504362630347, "grad_norm": 12.7578125, "learning_rate": 9.059231097214327e-06, "loss": 0.8291, "step": 26977 }, { "epoch": 2.8706107682485635, "grad_norm": 16.671875, "learning_rate": 9.059195565662308e-06, "loss": 0.7225, "step": 26978 }, { "epoch": 2.87071717386678, "grad_norm": 13.9140625, "learning_rate": 9.05916003411029e-06, "loss": 1.0499, "step": 26979 }, { "epoch": 2.8708235794849966, "grad_norm": 21.171875, "learning_rate": 9.059124502558273e-06, "loss": 1.0845, "step": 26980 }, { "epoch": 2.8709299851032135, "grad_norm": 11.1875, "learning_rate": 9.059088971006255e-06, "loss": 0.9481, "step": 26981 }, { "epoch": 2.87103639072143, "grad_norm": 14.5078125, "learning_rate": 9.059053439454236e-06, "loss": 0.9919, "step": 26982 }, { "epoch": 2.8711427963396465, "grad_norm": 11.140625, "learning_rate": 9.059017907902218e-06, "loss": 0.7896, "step": 26983 }, { "epoch": 2.8712492019578635, "grad_norm": 19.84375, "learning_rate": 9.0589823763502e-06, "loss": 0.8488, "step": 26984 }, { "epoch": 2.87135560757608, "grad_norm": 13.5859375, "learning_rate": 9.058946844798181e-06, "loss": 0.9597, "step": 26985 }, { "epoch": 2.8714620131942965, "grad_norm": 13.359375, "learning_rate": 9.058911313246164e-06, "loss": 0.8354, "step": 26986 }, { "epoch": 2.8715684188125135, "grad_norm": 14.2265625, "learning_rate": 9.058875781694146e-06, "loss": 1.064, "step": 26987 }, { "epoch": 2.87167482443073, "grad_norm": 23.921875, "learning_rate": 9.058840250142128e-06, "loss": 1.1404, "step": 26988 }, { "epoch": 2.8717812300489465, "grad_norm": 15.2109375, "learning_rate": 9.05880471859011e-06, "loss": 1.0016, "step": 26989 }, { "epoch": 2.8718876356671634, "grad_norm": 11.46875, "learning_rate": 9.05876918703809e-06, "loss": 0.8966, "step": 26990 }, { "epoch": 2.87199404128538, "grad_norm": 12.5, "learning_rate": 9.058733655486072e-06, "loss": 0.9191, "step": 26991 }, { "epoch": 2.8721004469035964, "grad_norm": 13.796875, "learning_rate": 9.058698123934054e-06, "loss": 0.9452, "step": 26992 }, { "epoch": 2.8722068525218134, "grad_norm": 14.2734375, "learning_rate": 9.058662592382036e-06, "loss": 0.76, "step": 26993 }, { "epoch": 2.87231325814003, "grad_norm": 17.515625, "learning_rate": 9.058627060830017e-06, "loss": 0.8626, "step": 26994 }, { "epoch": 2.8724196637582464, "grad_norm": 16.8125, "learning_rate": 9.058591529277999e-06, "loss": 0.9715, "step": 26995 }, { "epoch": 2.872526069376463, "grad_norm": 11.6015625, "learning_rate": 9.058555997725982e-06, "loss": 0.8338, "step": 26996 }, { "epoch": 2.87263247499468, "grad_norm": 10.0078125, "learning_rate": 9.058520466173964e-06, "loss": 0.8226, "step": 26997 }, { "epoch": 2.8727388806128964, "grad_norm": 10.859375, "learning_rate": 9.058484934621945e-06, "loss": 0.9371, "step": 26998 }, { "epoch": 2.872845286231113, "grad_norm": 11.203125, "learning_rate": 9.058449403069927e-06, "loss": 1.0452, "step": 26999 }, { "epoch": 2.8729516918493294, "grad_norm": 13.03125, "learning_rate": 9.058413871517909e-06, "loss": 0.8758, "step": 27000 }, { "epoch": 2.8730580974675464, "grad_norm": 13.609375, "learning_rate": 9.05837833996589e-06, "loss": 0.8808, "step": 27001 }, { "epoch": 2.873164503085763, "grad_norm": 13.2890625, "learning_rate": 9.058342808413872e-06, "loss": 1.1065, "step": 27002 }, { "epoch": 2.8732709087039794, "grad_norm": 15.578125, "learning_rate": 9.058307276861853e-06, "loss": 0.7539, "step": 27003 }, { "epoch": 2.8733773143221963, "grad_norm": 17.46875, "learning_rate": 9.058271745309835e-06, "loss": 0.9887, "step": 27004 }, { "epoch": 2.873483719940413, "grad_norm": 12.078125, "learning_rate": 9.058236213757817e-06, "loss": 0.9837, "step": 27005 }, { "epoch": 2.8735901255586294, "grad_norm": 12.8125, "learning_rate": 9.0582006822058e-06, "loss": 0.8844, "step": 27006 }, { "epoch": 2.8736965311768463, "grad_norm": 10.4921875, "learning_rate": 9.058165150653781e-06, "loss": 0.9994, "step": 27007 }, { "epoch": 2.873802936795063, "grad_norm": 10.71875, "learning_rate": 9.058129619101763e-06, "loss": 0.9001, "step": 27008 }, { "epoch": 2.8739093424132793, "grad_norm": 13.6171875, "learning_rate": 9.058094087549745e-06, "loss": 0.8905, "step": 27009 }, { "epoch": 2.8740157480314963, "grad_norm": 11.1328125, "learning_rate": 9.058058555997726e-06, "loss": 0.9638, "step": 27010 }, { "epoch": 2.874122153649713, "grad_norm": 11.578125, "learning_rate": 9.05802302444571e-06, "loss": 0.9235, "step": 27011 }, { "epoch": 2.8742285592679293, "grad_norm": 13.6015625, "learning_rate": 9.057987492893691e-06, "loss": 0.8266, "step": 27012 }, { "epoch": 2.8743349648861463, "grad_norm": 13.28125, "learning_rate": 9.057951961341673e-06, "loss": 0.9977, "step": 27013 }, { "epoch": 2.8744413705043628, "grad_norm": 9.8671875, "learning_rate": 9.057916429789654e-06, "loss": 0.7466, "step": 27014 }, { "epoch": 2.8745477761225793, "grad_norm": 9.0390625, "learning_rate": 9.057880898237636e-06, "loss": 0.6321, "step": 27015 }, { "epoch": 2.874654181740796, "grad_norm": 19.671875, "learning_rate": 9.057845366685618e-06, "loss": 0.7444, "step": 27016 }, { "epoch": 2.8747605873590123, "grad_norm": 16.109375, "learning_rate": 9.0578098351336e-06, "loss": 0.861, "step": 27017 }, { "epoch": 2.8748669929772293, "grad_norm": 12.265625, "learning_rate": 9.05777430358158e-06, "loss": 0.9057, "step": 27018 }, { "epoch": 2.8749733985954458, "grad_norm": 14.0078125, "learning_rate": 9.057738772029562e-06, "loss": 0.8173, "step": 27019 }, { "epoch": 2.8750798042136623, "grad_norm": 10.84375, "learning_rate": 9.057703240477544e-06, "loss": 0.8733, "step": 27020 }, { "epoch": 2.8751862098318792, "grad_norm": 12.9765625, "learning_rate": 9.057667708925527e-06, "loss": 0.9632, "step": 27021 }, { "epoch": 2.8752926154500957, "grad_norm": 11.421875, "learning_rate": 9.057632177373509e-06, "loss": 0.9189, "step": 27022 }, { "epoch": 2.8753990210683122, "grad_norm": 16.171875, "learning_rate": 9.05759664582149e-06, "loss": 0.9449, "step": 27023 }, { "epoch": 2.875505426686529, "grad_norm": 12.453125, "learning_rate": 9.057561114269472e-06, "loss": 0.7899, "step": 27024 }, { "epoch": 2.8756118323047457, "grad_norm": 13.8515625, "learning_rate": 9.057525582717454e-06, "loss": 0.9152, "step": 27025 }, { "epoch": 2.875718237922962, "grad_norm": 14.03125, "learning_rate": 9.057490051165435e-06, "loss": 0.9006, "step": 27026 }, { "epoch": 2.875824643541179, "grad_norm": 12.28125, "learning_rate": 9.057454519613417e-06, "loss": 0.9358, "step": 27027 }, { "epoch": 2.8759310491593957, "grad_norm": 14.1484375, "learning_rate": 9.057418988061399e-06, "loss": 1.0349, "step": 27028 }, { "epoch": 2.876037454777612, "grad_norm": 13.890625, "learning_rate": 9.05738345650938e-06, "loss": 0.9144, "step": 27029 }, { "epoch": 2.876143860395829, "grad_norm": 15.5234375, "learning_rate": 9.057347924957362e-06, "loss": 0.8767, "step": 27030 }, { "epoch": 2.8762502660140457, "grad_norm": 19.296875, "learning_rate": 9.057312393405343e-06, "loss": 1.2067, "step": 27031 }, { "epoch": 2.876356671632262, "grad_norm": 19.828125, "learning_rate": 9.057276861853327e-06, "loss": 0.9663, "step": 27032 }, { "epoch": 2.8764630772504787, "grad_norm": 14.03125, "learning_rate": 9.057241330301308e-06, "loss": 0.9606, "step": 27033 }, { "epoch": 2.8765694828686956, "grad_norm": 15.2421875, "learning_rate": 9.05720579874929e-06, "loss": 0.9585, "step": 27034 }, { "epoch": 2.876675888486912, "grad_norm": 12.625, "learning_rate": 9.057170267197272e-06, "loss": 0.9508, "step": 27035 }, { "epoch": 2.8767822941051286, "grad_norm": 26.65625, "learning_rate": 9.057134735645253e-06, "loss": 0.7957, "step": 27036 }, { "epoch": 2.876888699723345, "grad_norm": 13.0078125, "learning_rate": 9.057099204093236e-06, "loss": 0.8745, "step": 27037 }, { "epoch": 2.876995105341562, "grad_norm": 15.546875, "learning_rate": 9.057063672541218e-06, "loss": 1.1593, "step": 27038 }, { "epoch": 2.8771015109597786, "grad_norm": 13.8125, "learning_rate": 9.0570281409892e-06, "loss": 1.0944, "step": 27039 }, { "epoch": 2.877207916577995, "grad_norm": 12.8515625, "learning_rate": 9.056992609437181e-06, "loss": 0.9722, "step": 27040 }, { "epoch": 2.877314322196212, "grad_norm": 10.609375, "learning_rate": 9.056957077885163e-06, "loss": 0.8081, "step": 27041 }, { "epoch": 2.8774207278144286, "grad_norm": 13.296875, "learning_rate": 9.056921546333144e-06, "loss": 0.981, "step": 27042 }, { "epoch": 2.877527133432645, "grad_norm": 13.6015625, "learning_rate": 9.056886014781126e-06, "loss": 0.9695, "step": 27043 }, { "epoch": 2.877633539050862, "grad_norm": 11.03125, "learning_rate": 9.056850483229108e-06, "loss": 0.9049, "step": 27044 }, { "epoch": 2.8777399446690786, "grad_norm": 15.7734375, "learning_rate": 9.05681495167709e-06, "loss": 0.8846, "step": 27045 }, { "epoch": 2.877846350287295, "grad_norm": 14.0703125, "learning_rate": 9.056779420125071e-06, "loss": 0.7863, "step": 27046 }, { "epoch": 2.877952755905512, "grad_norm": 11.890625, "learning_rate": 9.056743888573054e-06, "loss": 0.8776, "step": 27047 }, { "epoch": 2.8780591615237285, "grad_norm": 13.734375, "learning_rate": 9.056708357021036e-06, "loss": 0.8646, "step": 27048 }, { "epoch": 2.878165567141945, "grad_norm": 15.4375, "learning_rate": 9.056672825469017e-06, "loss": 1.0315, "step": 27049 }, { "epoch": 2.878271972760162, "grad_norm": 14.3671875, "learning_rate": 9.056637293916999e-06, "loss": 1.0107, "step": 27050 }, { "epoch": 2.8783783783783785, "grad_norm": 11.5703125, "learning_rate": 9.05660176236498e-06, "loss": 0.9368, "step": 27051 }, { "epoch": 2.878484783996595, "grad_norm": 10.2578125, "learning_rate": 9.056566230812964e-06, "loss": 0.7593, "step": 27052 }, { "epoch": 2.8785911896148115, "grad_norm": 12.0078125, "learning_rate": 9.056530699260946e-06, "loss": 0.8996, "step": 27053 }, { "epoch": 2.878697595233028, "grad_norm": 12.1875, "learning_rate": 9.056495167708927e-06, "loss": 0.9237, "step": 27054 }, { "epoch": 2.878804000851245, "grad_norm": 15.828125, "learning_rate": 9.056459636156909e-06, "loss": 0.9924, "step": 27055 }, { "epoch": 2.8789104064694615, "grad_norm": 13.734375, "learning_rate": 9.05642410460489e-06, "loss": 0.8629, "step": 27056 }, { "epoch": 2.879016812087678, "grad_norm": 12.71875, "learning_rate": 9.056388573052872e-06, "loss": 0.8357, "step": 27057 }, { "epoch": 2.879123217705895, "grad_norm": 12.5, "learning_rate": 9.056353041500854e-06, "loss": 0.7137, "step": 27058 }, { "epoch": 2.8792296233241115, "grad_norm": 11.5, "learning_rate": 9.056317509948835e-06, "loss": 0.8468, "step": 27059 }, { "epoch": 2.879336028942328, "grad_norm": 14.8671875, "learning_rate": 9.056281978396817e-06, "loss": 0.9718, "step": 27060 }, { "epoch": 2.879442434560545, "grad_norm": 14.3046875, "learning_rate": 9.056246446844798e-06, "loss": 0.9753, "step": 27061 }, { "epoch": 2.8795488401787614, "grad_norm": 11.3828125, "learning_rate": 9.05621091529278e-06, "loss": 0.9978, "step": 27062 }, { "epoch": 2.879655245796978, "grad_norm": 12.6015625, "learning_rate": 9.056175383740763e-06, "loss": 0.9571, "step": 27063 }, { "epoch": 2.879761651415195, "grad_norm": 13.8046875, "learning_rate": 9.056139852188745e-06, "loss": 0.7445, "step": 27064 }, { "epoch": 2.8798680570334114, "grad_norm": 7.6328125, "learning_rate": 9.056104320636727e-06, "loss": 0.758, "step": 27065 }, { "epoch": 2.879974462651628, "grad_norm": 14.234375, "learning_rate": 9.056068789084708e-06, "loss": 0.9558, "step": 27066 }, { "epoch": 2.880080868269845, "grad_norm": 11.6953125, "learning_rate": 9.05603325753269e-06, "loss": 0.846, "step": 27067 }, { "epoch": 2.8801872738880614, "grad_norm": 15.765625, "learning_rate": 9.055997725980671e-06, "loss": 1.1168, "step": 27068 }, { "epoch": 2.880293679506278, "grad_norm": 10.921875, "learning_rate": 9.055962194428653e-06, "loss": 0.9572, "step": 27069 }, { "epoch": 2.8804000851244944, "grad_norm": 13.109375, "learning_rate": 9.055926662876635e-06, "loss": 0.8231, "step": 27070 }, { "epoch": 2.8805064907427114, "grad_norm": 11.96875, "learning_rate": 9.055891131324616e-06, "loss": 1.0104, "step": 27071 }, { "epoch": 2.880612896360928, "grad_norm": 9.515625, "learning_rate": 9.055855599772598e-06, "loss": 0.8062, "step": 27072 }, { "epoch": 2.8807193019791444, "grad_norm": 15.3828125, "learning_rate": 9.055820068220581e-06, "loss": 0.9479, "step": 27073 }, { "epoch": 2.880825707597361, "grad_norm": 9.8125, "learning_rate": 9.055784536668563e-06, "loss": 0.9438, "step": 27074 }, { "epoch": 2.880932113215578, "grad_norm": 13.2734375, "learning_rate": 9.055749005116544e-06, "loss": 0.9904, "step": 27075 }, { "epoch": 2.8810385188337944, "grad_norm": 15.9921875, "learning_rate": 9.055713473564526e-06, "loss": 1.1305, "step": 27076 }, { "epoch": 2.881144924452011, "grad_norm": 14.1796875, "learning_rate": 9.055677942012507e-06, "loss": 0.9112, "step": 27077 }, { "epoch": 2.881251330070228, "grad_norm": 12.859375, "learning_rate": 9.05564241046049e-06, "loss": 0.8539, "step": 27078 }, { "epoch": 2.8813577356884443, "grad_norm": 10.96875, "learning_rate": 9.055606878908472e-06, "loss": 0.9147, "step": 27079 }, { "epoch": 2.881464141306661, "grad_norm": 15.0703125, "learning_rate": 9.055571347356454e-06, "loss": 0.9272, "step": 27080 }, { "epoch": 2.881570546924878, "grad_norm": 12.21875, "learning_rate": 9.055535815804436e-06, "loss": 1.0499, "step": 27081 }, { "epoch": 2.8816769525430943, "grad_norm": 13.6171875, "learning_rate": 9.055500284252417e-06, "loss": 0.9992, "step": 27082 }, { "epoch": 2.881783358161311, "grad_norm": 22.515625, "learning_rate": 9.055464752700399e-06, "loss": 0.8927, "step": 27083 }, { "epoch": 2.8818897637795278, "grad_norm": 10.015625, "learning_rate": 9.05542922114838e-06, "loss": 0.7545, "step": 27084 }, { "epoch": 2.8819961693977443, "grad_norm": 11.734375, "learning_rate": 9.055393689596362e-06, "loss": 0.9346, "step": 27085 }, { "epoch": 2.882102575015961, "grad_norm": 14.3125, "learning_rate": 9.055358158044344e-06, "loss": 0.815, "step": 27086 }, { "epoch": 2.8822089806341777, "grad_norm": 14.203125, "learning_rate": 9.055322626492325e-06, "loss": 0.9604, "step": 27087 }, { "epoch": 2.8823153862523943, "grad_norm": 20.09375, "learning_rate": 9.055287094940309e-06, "loss": 0.9449, "step": 27088 }, { "epoch": 2.8824217918706108, "grad_norm": 13.2734375, "learning_rate": 9.05525156338829e-06, "loss": 0.7831, "step": 27089 }, { "epoch": 2.8825281974888273, "grad_norm": 15.5703125, "learning_rate": 9.055216031836272e-06, "loss": 0.8975, "step": 27090 }, { "epoch": 2.882634603107044, "grad_norm": 16.046875, "learning_rate": 9.055180500284253e-06, "loss": 1.0275, "step": 27091 }, { "epoch": 2.8827410087252607, "grad_norm": 15.953125, "learning_rate": 9.055144968732235e-06, "loss": 0.8685, "step": 27092 }, { "epoch": 2.8828474143434772, "grad_norm": 10.9375, "learning_rate": 9.055109437180217e-06, "loss": 0.8332, "step": 27093 }, { "epoch": 2.8829538199616938, "grad_norm": 18.171875, "learning_rate": 9.055073905628198e-06, "loss": 1.1855, "step": 27094 }, { "epoch": 2.8830602255799107, "grad_norm": 16.546875, "learning_rate": 9.05503837407618e-06, "loss": 0.808, "step": 27095 }, { "epoch": 2.883166631198127, "grad_norm": 13.53125, "learning_rate": 9.055002842524161e-06, "loss": 0.9823, "step": 27096 }, { "epoch": 2.8832730368163437, "grad_norm": 11.140625, "learning_rate": 9.054967310972143e-06, "loss": 0.7957, "step": 27097 }, { "epoch": 2.8833794424345607, "grad_norm": 13.8046875, "learning_rate": 9.054931779420126e-06, "loss": 0.937, "step": 27098 }, { "epoch": 2.883485848052777, "grad_norm": 15.0703125, "learning_rate": 9.054896247868108e-06, "loss": 1.101, "step": 27099 }, { "epoch": 2.8835922536709937, "grad_norm": 15.0078125, "learning_rate": 9.05486071631609e-06, "loss": 0.9825, "step": 27100 }, { "epoch": 2.8836986592892107, "grad_norm": 14.1875, "learning_rate": 9.054825184764071e-06, "loss": 0.9614, "step": 27101 }, { "epoch": 2.883805064907427, "grad_norm": 15.3515625, "learning_rate": 9.054789653212053e-06, "loss": 0.8753, "step": 27102 }, { "epoch": 2.8839114705256437, "grad_norm": 18.8125, "learning_rate": 9.054754121660034e-06, "loss": 0.8675, "step": 27103 }, { "epoch": 2.8840178761438606, "grad_norm": 12.3046875, "learning_rate": 9.054718590108018e-06, "loss": 0.8945, "step": 27104 }, { "epoch": 2.884124281762077, "grad_norm": 10.78125, "learning_rate": 9.054683058556e-06, "loss": 0.8885, "step": 27105 }, { "epoch": 2.8842306873802936, "grad_norm": 9.4375, "learning_rate": 9.054647527003981e-06, "loss": 0.788, "step": 27106 }, { "epoch": 2.8843370929985106, "grad_norm": 13.59375, "learning_rate": 9.054611995451962e-06, "loss": 1.1161, "step": 27107 }, { "epoch": 2.884443498616727, "grad_norm": 14.1875, "learning_rate": 9.054576463899944e-06, "loss": 0.8622, "step": 27108 }, { "epoch": 2.8845499042349436, "grad_norm": 15.1015625, "learning_rate": 9.054540932347926e-06, "loss": 0.9221, "step": 27109 }, { "epoch": 2.88465630985316, "grad_norm": 11.390625, "learning_rate": 9.054505400795907e-06, "loss": 0.9045, "step": 27110 }, { "epoch": 2.8847627154713766, "grad_norm": 17.3125, "learning_rate": 9.054469869243889e-06, "loss": 1.0173, "step": 27111 }, { "epoch": 2.8848691210895936, "grad_norm": 11.28125, "learning_rate": 9.05443433769187e-06, "loss": 0.7863, "step": 27112 }, { "epoch": 2.88497552670781, "grad_norm": 13.78125, "learning_rate": 9.054398806139852e-06, "loss": 0.8951, "step": 27113 }, { "epoch": 2.8850819323260266, "grad_norm": 14.9375, "learning_rate": 9.054363274587835e-06, "loss": 1.0312, "step": 27114 }, { "epoch": 2.8851883379442436, "grad_norm": 13.1953125, "learning_rate": 9.054327743035817e-06, "loss": 0.7742, "step": 27115 }, { "epoch": 2.88529474356246, "grad_norm": 12.15625, "learning_rate": 9.054292211483799e-06, "loss": 0.9943, "step": 27116 }, { "epoch": 2.8854011491806766, "grad_norm": 12.734375, "learning_rate": 9.05425667993178e-06, "loss": 1.0558, "step": 27117 }, { "epoch": 2.8855075547988935, "grad_norm": 10.84375, "learning_rate": 9.054221148379762e-06, "loss": 0.8985, "step": 27118 }, { "epoch": 2.88561396041711, "grad_norm": 10.9140625, "learning_rate": 9.054185616827743e-06, "loss": 0.8614, "step": 27119 }, { "epoch": 2.8857203660353266, "grad_norm": 13.28125, "learning_rate": 9.054150085275725e-06, "loss": 0.8199, "step": 27120 }, { "epoch": 2.8858267716535435, "grad_norm": 15.4375, "learning_rate": 9.054114553723707e-06, "loss": 1.2175, "step": 27121 }, { "epoch": 2.88593317727176, "grad_norm": 11.3671875, "learning_rate": 9.054079022171688e-06, "loss": 0.8631, "step": 27122 }, { "epoch": 2.8860395828899765, "grad_norm": 11.15625, "learning_rate": 9.05404349061967e-06, "loss": 0.8728, "step": 27123 }, { "epoch": 2.8861459885081935, "grad_norm": 9.96875, "learning_rate": 9.054007959067653e-06, "loss": 0.9391, "step": 27124 }, { "epoch": 2.88625239412641, "grad_norm": 9.265625, "learning_rate": 9.053972427515635e-06, "loss": 0.7004, "step": 27125 }, { "epoch": 2.8863587997446265, "grad_norm": 12.03125, "learning_rate": 9.053936895963616e-06, "loss": 1.0058, "step": 27126 }, { "epoch": 2.886465205362843, "grad_norm": 9.4296875, "learning_rate": 9.053901364411598e-06, "loss": 0.8769, "step": 27127 }, { "epoch": 2.88657161098106, "grad_norm": 16.453125, "learning_rate": 9.05386583285958e-06, "loss": 0.959, "step": 27128 }, { "epoch": 2.8866780165992765, "grad_norm": 12.9921875, "learning_rate": 9.053830301307563e-06, "loss": 1.0488, "step": 27129 }, { "epoch": 2.886784422217493, "grad_norm": 13.59375, "learning_rate": 9.053794769755545e-06, "loss": 0.8555, "step": 27130 }, { "epoch": 2.8868908278357095, "grad_norm": 10.03125, "learning_rate": 9.053759238203526e-06, "loss": 0.773, "step": 27131 }, { "epoch": 2.8869972334539264, "grad_norm": 14.265625, "learning_rate": 9.053723706651508e-06, "loss": 0.9995, "step": 27132 }, { "epoch": 2.887103639072143, "grad_norm": 14.1953125, "learning_rate": 9.05368817509949e-06, "loss": 1.0441, "step": 27133 }, { "epoch": 2.8872100446903595, "grad_norm": 13.0234375, "learning_rate": 9.053652643547471e-06, "loss": 0.9226, "step": 27134 }, { "epoch": 2.8873164503085764, "grad_norm": 10.4765625, "learning_rate": 9.053617111995453e-06, "loss": 0.8616, "step": 27135 }, { "epoch": 2.887422855926793, "grad_norm": 14.75, "learning_rate": 9.053581580443434e-06, "loss": 0.9942, "step": 27136 }, { "epoch": 2.8875292615450094, "grad_norm": 13.5859375, "learning_rate": 9.053546048891416e-06, "loss": 0.9209, "step": 27137 }, { "epoch": 2.8876356671632264, "grad_norm": 14.8359375, "learning_rate": 9.053510517339397e-06, "loss": 0.9069, "step": 27138 }, { "epoch": 2.887742072781443, "grad_norm": 11.484375, "learning_rate": 9.05347498578738e-06, "loss": 1.0276, "step": 27139 }, { "epoch": 2.8878484783996594, "grad_norm": 14.296875, "learning_rate": 9.053439454235362e-06, "loss": 0.9672, "step": 27140 }, { "epoch": 2.8879548840178764, "grad_norm": 12.6015625, "learning_rate": 9.053403922683344e-06, "loss": 1.0208, "step": 27141 }, { "epoch": 2.888061289636093, "grad_norm": 10.921875, "learning_rate": 9.053368391131325e-06, "loss": 0.8292, "step": 27142 }, { "epoch": 2.8881676952543094, "grad_norm": 14.4296875, "learning_rate": 9.053332859579307e-06, "loss": 0.9003, "step": 27143 }, { "epoch": 2.8882741008725263, "grad_norm": 13.6484375, "learning_rate": 9.053297328027289e-06, "loss": 1.0568, "step": 27144 }, { "epoch": 2.888380506490743, "grad_norm": 13.859375, "learning_rate": 9.05326179647527e-06, "loss": 0.91, "step": 27145 }, { "epoch": 2.8884869121089594, "grad_norm": 14.6171875, "learning_rate": 9.053226264923252e-06, "loss": 0.9177, "step": 27146 }, { "epoch": 2.888593317727176, "grad_norm": 11.7265625, "learning_rate": 9.053190733371234e-06, "loss": 0.5481, "step": 27147 }, { "epoch": 2.8886997233453924, "grad_norm": 11.0859375, "learning_rate": 9.053155201819215e-06, "loss": 0.8637, "step": 27148 }, { "epoch": 2.8888061289636093, "grad_norm": 14.1015625, "learning_rate": 9.053119670267197e-06, "loss": 0.8342, "step": 27149 }, { "epoch": 2.888912534581826, "grad_norm": 15.078125, "learning_rate": 9.05308413871518e-06, "loss": 1.0329, "step": 27150 }, { "epoch": 2.8890189402000424, "grad_norm": 11.703125, "learning_rate": 9.053048607163162e-06, "loss": 0.7655, "step": 27151 }, { "epoch": 2.8891253458182593, "grad_norm": 13.84375, "learning_rate": 9.053013075611143e-06, "loss": 0.9641, "step": 27152 }, { "epoch": 2.889231751436476, "grad_norm": 16.578125, "learning_rate": 9.052977544059125e-06, "loss": 1.0807, "step": 27153 }, { "epoch": 2.8893381570546923, "grad_norm": 12.46875, "learning_rate": 9.052942012507106e-06, "loss": 0.8584, "step": 27154 }, { "epoch": 2.8894445626729093, "grad_norm": 16.03125, "learning_rate": 9.05290648095509e-06, "loss": 1.0216, "step": 27155 }, { "epoch": 2.889550968291126, "grad_norm": 14.2734375, "learning_rate": 9.052870949403071e-06, "loss": 0.9593, "step": 27156 }, { "epoch": 2.8896573739093423, "grad_norm": 23.09375, "learning_rate": 9.052835417851053e-06, "loss": 1.0737, "step": 27157 }, { "epoch": 2.8897637795275593, "grad_norm": 15.5, "learning_rate": 9.052799886299035e-06, "loss": 1.1677, "step": 27158 }, { "epoch": 2.8898701851457758, "grad_norm": 15.2578125, "learning_rate": 9.052764354747016e-06, "loss": 1.0436, "step": 27159 }, { "epoch": 2.8899765907639923, "grad_norm": 17.4375, "learning_rate": 9.052728823194998e-06, "loss": 1.0301, "step": 27160 }, { "epoch": 2.8900829963822092, "grad_norm": 11.9296875, "learning_rate": 9.05269329164298e-06, "loss": 0.8884, "step": 27161 }, { "epoch": 2.8901894020004257, "grad_norm": 11.8203125, "learning_rate": 9.052657760090961e-06, "loss": 0.9822, "step": 27162 }, { "epoch": 2.8902958076186422, "grad_norm": 11.6328125, "learning_rate": 9.052622228538943e-06, "loss": 0.9176, "step": 27163 }, { "epoch": 2.8904022132368588, "grad_norm": 14.65625, "learning_rate": 9.052586696986924e-06, "loss": 1.0567, "step": 27164 }, { "epoch": 2.8905086188550757, "grad_norm": 17.40625, "learning_rate": 9.052551165434908e-06, "loss": 0.7416, "step": 27165 }, { "epoch": 2.890615024473292, "grad_norm": 12.1953125, "learning_rate": 9.052515633882889e-06, "loss": 1.0179, "step": 27166 }, { "epoch": 2.8907214300915087, "grad_norm": 9.828125, "learning_rate": 9.05248010233087e-06, "loss": 0.9892, "step": 27167 }, { "epoch": 2.8908278357097252, "grad_norm": 14.4609375, "learning_rate": 9.052444570778852e-06, "loss": 0.965, "step": 27168 }, { "epoch": 2.890934241327942, "grad_norm": 17.359375, "learning_rate": 9.052409039226834e-06, "loss": 0.908, "step": 27169 }, { "epoch": 2.8910406469461587, "grad_norm": 13.828125, "learning_rate": 9.052373507674817e-06, "loss": 0.7852, "step": 27170 }, { "epoch": 2.891147052564375, "grad_norm": 10.640625, "learning_rate": 9.052337976122799e-06, "loss": 0.7947, "step": 27171 }, { "epoch": 2.891253458182592, "grad_norm": 13.3359375, "learning_rate": 9.05230244457078e-06, "loss": 0.8494, "step": 27172 }, { "epoch": 2.8913598638008087, "grad_norm": 12.4375, "learning_rate": 9.052266913018762e-06, "loss": 0.891, "step": 27173 }, { "epoch": 2.891466269419025, "grad_norm": 13.9921875, "learning_rate": 9.052231381466744e-06, "loss": 0.9865, "step": 27174 }, { "epoch": 2.891572675037242, "grad_norm": 12.59375, "learning_rate": 9.052195849914725e-06, "loss": 0.9449, "step": 27175 }, { "epoch": 2.8916790806554586, "grad_norm": 11.921875, "learning_rate": 9.052160318362707e-06, "loss": 0.9777, "step": 27176 }, { "epoch": 2.891785486273675, "grad_norm": 12.0703125, "learning_rate": 9.052124786810688e-06, "loss": 0.8145, "step": 27177 }, { "epoch": 2.891891891891892, "grad_norm": 11.3671875, "learning_rate": 9.05208925525867e-06, "loss": 0.966, "step": 27178 }, { "epoch": 2.8919982975101086, "grad_norm": 10.8984375, "learning_rate": 9.052053723706652e-06, "loss": 0.8738, "step": 27179 }, { "epoch": 2.892104703128325, "grad_norm": 11.4140625, "learning_rate": 9.052018192154633e-06, "loss": 0.9614, "step": 27180 }, { "epoch": 2.892211108746542, "grad_norm": 17.21875, "learning_rate": 9.051982660602617e-06, "loss": 0.8316, "step": 27181 }, { "epoch": 2.8923175143647586, "grad_norm": 10.875, "learning_rate": 9.051947129050598e-06, "loss": 0.9484, "step": 27182 }, { "epoch": 2.892423919982975, "grad_norm": 12.8359375, "learning_rate": 9.05191159749858e-06, "loss": 0.881, "step": 27183 }, { "epoch": 2.8925303256011916, "grad_norm": 14.328125, "learning_rate": 9.051876065946561e-06, "loss": 0.9396, "step": 27184 }, { "epoch": 2.892636731219408, "grad_norm": 16.0625, "learning_rate": 9.051840534394543e-06, "loss": 1.1714, "step": 27185 }, { "epoch": 2.892743136837625, "grad_norm": 15.5625, "learning_rate": 9.051805002842525e-06, "loss": 1.0345, "step": 27186 }, { "epoch": 2.8928495424558416, "grad_norm": 11.453125, "learning_rate": 9.051769471290506e-06, "loss": 0.9071, "step": 27187 }, { "epoch": 2.892955948074058, "grad_norm": 14.8671875, "learning_rate": 9.051733939738488e-06, "loss": 0.9707, "step": 27188 }, { "epoch": 2.893062353692275, "grad_norm": 16.53125, "learning_rate": 9.05169840818647e-06, "loss": 0.9356, "step": 27189 }, { "epoch": 2.8931687593104916, "grad_norm": 10.578125, "learning_rate": 9.051662876634451e-06, "loss": 0.8035, "step": 27190 }, { "epoch": 2.893275164928708, "grad_norm": 11.953125, "learning_rate": 9.051627345082434e-06, "loss": 0.9467, "step": 27191 }, { "epoch": 2.893381570546925, "grad_norm": 10.7109375, "learning_rate": 9.051591813530416e-06, "loss": 0.861, "step": 27192 }, { "epoch": 2.8934879761651415, "grad_norm": 15.4140625, "learning_rate": 9.051556281978398e-06, "loss": 0.997, "step": 27193 }, { "epoch": 2.893594381783358, "grad_norm": 12.6953125, "learning_rate": 9.05152075042638e-06, "loss": 0.8824, "step": 27194 }, { "epoch": 2.893700787401575, "grad_norm": 15.9765625, "learning_rate": 9.05148521887436e-06, "loss": 0.903, "step": 27195 }, { "epoch": 2.8938071930197915, "grad_norm": 9.6640625, "learning_rate": 9.051449687322344e-06, "loss": 0.8639, "step": 27196 }, { "epoch": 2.893913598638008, "grad_norm": 12.4609375, "learning_rate": 9.051414155770326e-06, "loss": 1.0471, "step": 27197 }, { "epoch": 2.894020004256225, "grad_norm": 10.6953125, "learning_rate": 9.051378624218307e-06, "loss": 0.7623, "step": 27198 }, { "epoch": 2.8941264098744415, "grad_norm": 12.15625, "learning_rate": 9.051343092666289e-06, "loss": 0.9101, "step": 27199 }, { "epoch": 2.894232815492658, "grad_norm": 15.078125, "learning_rate": 9.05130756111427e-06, "loss": 1.2805, "step": 27200 }, { "epoch": 2.8943392211108745, "grad_norm": 11.75, "learning_rate": 9.051272029562252e-06, "loss": 0.8765, "step": 27201 }, { "epoch": 2.8944456267290914, "grad_norm": 13.8984375, "learning_rate": 9.051236498010234e-06, "loss": 0.8904, "step": 27202 }, { "epoch": 2.894552032347308, "grad_norm": 17.46875, "learning_rate": 9.051200966458215e-06, "loss": 0.8483, "step": 27203 }, { "epoch": 2.8946584379655245, "grad_norm": 16.5, "learning_rate": 9.051165434906197e-06, "loss": 0.7759, "step": 27204 }, { "epoch": 2.894764843583741, "grad_norm": 10.1328125, "learning_rate": 9.051129903354179e-06, "loss": 0.8204, "step": 27205 }, { "epoch": 2.894871249201958, "grad_norm": 14.234375, "learning_rate": 9.051094371802162e-06, "loss": 0.9791, "step": 27206 }, { "epoch": 2.8949776548201744, "grad_norm": 19.109375, "learning_rate": 9.051058840250143e-06, "loss": 0.8759, "step": 27207 }, { "epoch": 2.895084060438391, "grad_norm": 12.9609375, "learning_rate": 9.051023308698125e-06, "loss": 0.8934, "step": 27208 }, { "epoch": 2.895190466056608, "grad_norm": 14.15625, "learning_rate": 9.050987777146107e-06, "loss": 0.9797, "step": 27209 }, { "epoch": 2.8952968716748244, "grad_norm": 15.640625, "learning_rate": 9.050952245594088e-06, "loss": 1.0314, "step": 27210 }, { "epoch": 2.895403277293041, "grad_norm": 14.0625, "learning_rate": 9.05091671404207e-06, "loss": 0.766, "step": 27211 }, { "epoch": 2.895509682911258, "grad_norm": 11.296875, "learning_rate": 9.050881182490051e-06, "loss": 0.9314, "step": 27212 }, { "epoch": 2.8956160885294744, "grad_norm": 13.8046875, "learning_rate": 9.050845650938033e-06, "loss": 0.7502, "step": 27213 }, { "epoch": 2.895722494147691, "grad_norm": 14.6015625, "learning_rate": 9.050810119386015e-06, "loss": 1.0132, "step": 27214 }, { "epoch": 2.895828899765908, "grad_norm": 11.65625, "learning_rate": 9.050774587833996e-06, "loss": 0.8206, "step": 27215 }, { "epoch": 2.8959353053841244, "grad_norm": 14.6875, "learning_rate": 9.05073905628198e-06, "loss": 1.0407, "step": 27216 }, { "epoch": 2.896041711002341, "grad_norm": 12.6875, "learning_rate": 9.050703524729961e-06, "loss": 0.9047, "step": 27217 }, { "epoch": 2.896148116620558, "grad_norm": 11.6953125, "learning_rate": 9.050667993177943e-06, "loss": 0.8207, "step": 27218 }, { "epoch": 2.8962545222387743, "grad_norm": 13.109375, "learning_rate": 9.050632461625924e-06, "loss": 0.9703, "step": 27219 }, { "epoch": 2.896360927856991, "grad_norm": 13.234375, "learning_rate": 9.050596930073906e-06, "loss": 0.8953, "step": 27220 }, { "epoch": 2.8964673334752074, "grad_norm": 15.4296875, "learning_rate": 9.050561398521888e-06, "loss": 0.9413, "step": 27221 }, { "epoch": 2.896573739093424, "grad_norm": 18.078125, "learning_rate": 9.050525866969871e-06, "loss": 1.1782, "step": 27222 }, { "epoch": 2.896680144711641, "grad_norm": 10.9609375, "learning_rate": 9.050490335417853e-06, "loss": 0.7771, "step": 27223 }, { "epoch": 2.8967865503298573, "grad_norm": 13.6953125, "learning_rate": 9.050454803865834e-06, "loss": 0.8681, "step": 27224 }, { "epoch": 2.896892955948074, "grad_norm": 18.5625, "learning_rate": 9.050419272313816e-06, "loss": 0.9531, "step": 27225 }, { "epoch": 2.896999361566291, "grad_norm": 18.5, "learning_rate": 9.050383740761797e-06, "loss": 1.0357, "step": 27226 }, { "epoch": 2.8971057671845073, "grad_norm": 15.8828125, "learning_rate": 9.050348209209779e-06, "loss": 0.9876, "step": 27227 }, { "epoch": 2.897212172802724, "grad_norm": 12.8203125, "learning_rate": 9.05031267765776e-06, "loss": 0.8823, "step": 27228 }, { "epoch": 2.8973185784209408, "grad_norm": 10.9296875, "learning_rate": 9.050277146105742e-06, "loss": 0.7827, "step": 27229 }, { "epoch": 2.8974249840391573, "grad_norm": 13.2578125, "learning_rate": 9.050241614553724e-06, "loss": 0.8416, "step": 27230 }, { "epoch": 2.897531389657374, "grad_norm": 12.4375, "learning_rate": 9.050206083001705e-06, "loss": 0.8903, "step": 27231 }, { "epoch": 2.8976377952755907, "grad_norm": 16.828125, "learning_rate": 9.050170551449689e-06, "loss": 1.0233, "step": 27232 }, { "epoch": 2.8977442008938072, "grad_norm": 12.0078125, "learning_rate": 9.05013501989767e-06, "loss": 1.0062, "step": 27233 }, { "epoch": 2.8978506065120238, "grad_norm": 13.9375, "learning_rate": 9.050099488345652e-06, "loss": 0.9346, "step": 27234 }, { "epoch": 2.8979570121302407, "grad_norm": 14.4296875, "learning_rate": 9.050063956793634e-06, "loss": 0.9882, "step": 27235 }, { "epoch": 2.898063417748457, "grad_norm": 16.265625, "learning_rate": 9.050028425241615e-06, "loss": 0.8157, "step": 27236 }, { "epoch": 2.8981698233666737, "grad_norm": 11.1484375, "learning_rate": 9.049992893689597e-06, "loss": 0.824, "step": 27237 }, { "epoch": 2.8982762289848907, "grad_norm": 10.8984375, "learning_rate": 9.049957362137578e-06, "loss": 0.9469, "step": 27238 }, { "epoch": 2.898382634603107, "grad_norm": 11.0859375, "learning_rate": 9.04992183058556e-06, "loss": 0.8094, "step": 27239 }, { "epoch": 2.8984890402213237, "grad_norm": 11.84375, "learning_rate": 9.049886299033542e-06, "loss": 0.6798, "step": 27240 }, { "epoch": 2.89859544583954, "grad_norm": 19.640625, "learning_rate": 9.049850767481523e-06, "loss": 1.1261, "step": 27241 }, { "epoch": 2.8987018514577567, "grad_norm": 12.2265625, "learning_rate": 9.049815235929506e-06, "loss": 0.943, "step": 27242 }, { "epoch": 2.8988082570759737, "grad_norm": 14.296875, "learning_rate": 9.049779704377488e-06, "loss": 1.0325, "step": 27243 }, { "epoch": 2.89891466269419, "grad_norm": 12.2890625, "learning_rate": 9.04974417282547e-06, "loss": 0.9628, "step": 27244 }, { "epoch": 2.8990210683124067, "grad_norm": 11.6171875, "learning_rate": 9.049708641273451e-06, "loss": 0.8575, "step": 27245 }, { "epoch": 2.8991274739306236, "grad_norm": 20.28125, "learning_rate": 9.049673109721433e-06, "loss": 0.9657, "step": 27246 }, { "epoch": 2.89923387954884, "grad_norm": 16.53125, "learning_rate": 9.049637578169416e-06, "loss": 0.8998, "step": 27247 }, { "epoch": 2.8993402851670567, "grad_norm": 17.125, "learning_rate": 9.049602046617398e-06, "loss": 0.8622, "step": 27248 }, { "epoch": 2.8994466907852736, "grad_norm": 11.25, "learning_rate": 9.04956651506538e-06, "loss": 0.8188, "step": 27249 }, { "epoch": 2.89955309640349, "grad_norm": 11.046875, "learning_rate": 9.049530983513361e-06, "loss": 0.7479, "step": 27250 }, { "epoch": 2.8996595020217066, "grad_norm": 12.5390625, "learning_rate": 9.049495451961343e-06, "loss": 0.9257, "step": 27251 }, { "epoch": 2.8997659076399236, "grad_norm": 15.0703125, "learning_rate": 9.049459920409324e-06, "loss": 0.9513, "step": 27252 }, { "epoch": 2.89987231325814, "grad_norm": 14.609375, "learning_rate": 9.049424388857306e-06, "loss": 1.1811, "step": 27253 }, { "epoch": 2.8999787188763566, "grad_norm": 12.8046875, "learning_rate": 9.049388857305287e-06, "loss": 0.9452, "step": 27254 }, { "epoch": 2.9000851244945736, "grad_norm": 16.078125, "learning_rate": 9.049353325753269e-06, "loss": 1.0684, "step": 27255 }, { "epoch": 2.90019153011279, "grad_norm": 12.9921875, "learning_rate": 9.04931779420125e-06, "loss": 0.8957, "step": 27256 }, { "epoch": 2.9002979357310066, "grad_norm": 12.0390625, "learning_rate": 9.049282262649232e-06, "loss": 0.8668, "step": 27257 }, { "epoch": 2.900404341349223, "grad_norm": 14.046875, "learning_rate": 9.049246731097216e-06, "loss": 0.9536, "step": 27258 }, { "epoch": 2.90051074696744, "grad_norm": 12.7109375, "learning_rate": 9.049211199545197e-06, "loss": 0.972, "step": 27259 }, { "epoch": 2.9006171525856566, "grad_norm": 13.7734375, "learning_rate": 9.049175667993179e-06, "loss": 0.9567, "step": 27260 }, { "epoch": 2.900723558203873, "grad_norm": 11.6015625, "learning_rate": 9.04914013644116e-06, "loss": 0.9315, "step": 27261 }, { "epoch": 2.9008299638220896, "grad_norm": 17.84375, "learning_rate": 9.049104604889142e-06, "loss": 1.0729, "step": 27262 }, { "epoch": 2.9009363694403065, "grad_norm": 10.0625, "learning_rate": 9.049069073337125e-06, "loss": 0.8719, "step": 27263 }, { "epoch": 2.901042775058523, "grad_norm": 11.0546875, "learning_rate": 9.049033541785107e-06, "loss": 0.6087, "step": 27264 }, { "epoch": 2.9011491806767395, "grad_norm": 10.3046875, "learning_rate": 9.048998010233087e-06, "loss": 0.8193, "step": 27265 }, { "epoch": 2.9012555862949565, "grad_norm": 10.90625, "learning_rate": 9.048962478681068e-06, "loss": 0.8458, "step": 27266 }, { "epoch": 2.901361991913173, "grad_norm": 9.046875, "learning_rate": 9.04892694712905e-06, "loss": 0.9447, "step": 27267 }, { "epoch": 2.9014683975313895, "grad_norm": 11.1796875, "learning_rate": 9.048891415577033e-06, "loss": 0.9125, "step": 27268 }, { "epoch": 2.9015748031496065, "grad_norm": 11.1953125, "learning_rate": 9.048855884025015e-06, "loss": 0.9282, "step": 27269 }, { "epoch": 2.901681208767823, "grad_norm": 8.6796875, "learning_rate": 9.048820352472997e-06, "loss": 0.7231, "step": 27270 }, { "epoch": 2.9017876143860395, "grad_norm": 10.7578125, "learning_rate": 9.048784820920978e-06, "loss": 1.0096, "step": 27271 }, { "epoch": 2.9018940200042564, "grad_norm": 14.546875, "learning_rate": 9.04874928936896e-06, "loss": 1.2298, "step": 27272 }, { "epoch": 2.902000425622473, "grad_norm": 12.015625, "learning_rate": 9.048713757816943e-06, "loss": 0.6843, "step": 27273 }, { "epoch": 2.9021068312406895, "grad_norm": 10.09375, "learning_rate": 9.048678226264925e-06, "loss": 0.8306, "step": 27274 }, { "epoch": 2.9022132368589064, "grad_norm": 17.421875, "learning_rate": 9.048642694712906e-06, "loss": 0.9464, "step": 27275 }, { "epoch": 2.902319642477123, "grad_norm": 13.8046875, "learning_rate": 9.048607163160888e-06, "loss": 0.9087, "step": 27276 }, { "epoch": 2.9024260480953394, "grad_norm": 12.75, "learning_rate": 9.04857163160887e-06, "loss": 0.9454, "step": 27277 }, { "epoch": 2.902532453713556, "grad_norm": 14.703125, "learning_rate": 9.048536100056851e-06, "loss": 0.9936, "step": 27278 }, { "epoch": 2.9026388593317725, "grad_norm": 10.7109375, "learning_rate": 9.048500568504833e-06, "loss": 0.9602, "step": 27279 }, { "epoch": 2.9027452649499894, "grad_norm": 14.6796875, "learning_rate": 9.048465036952814e-06, "loss": 0.776, "step": 27280 }, { "epoch": 2.902851670568206, "grad_norm": 16.0625, "learning_rate": 9.048429505400796e-06, "loss": 1.0826, "step": 27281 }, { "epoch": 2.9029580761864224, "grad_norm": 10.734375, "learning_rate": 9.048393973848778e-06, "loss": 0.9063, "step": 27282 }, { "epoch": 2.9030644818046394, "grad_norm": 13.3515625, "learning_rate": 9.04835844229676e-06, "loss": 1.0175, "step": 27283 }, { "epoch": 2.903170887422856, "grad_norm": 14.234375, "learning_rate": 9.048322910744742e-06, "loss": 1.1825, "step": 27284 }, { "epoch": 2.9032772930410724, "grad_norm": 17.21875, "learning_rate": 9.048287379192724e-06, "loss": 0.8777, "step": 27285 }, { "epoch": 2.9033836986592894, "grad_norm": 13.2734375, "learning_rate": 9.048251847640706e-06, "loss": 0.844, "step": 27286 }, { "epoch": 2.903490104277506, "grad_norm": 14.609375, "learning_rate": 9.048216316088687e-06, "loss": 0.9298, "step": 27287 }, { "epoch": 2.9035965098957224, "grad_norm": 12.0, "learning_rate": 9.04818078453667e-06, "loss": 1.0192, "step": 27288 }, { "epoch": 2.9037029155139393, "grad_norm": 22.5625, "learning_rate": 9.048145252984652e-06, "loss": 1.0126, "step": 27289 }, { "epoch": 2.903809321132156, "grad_norm": 11.4453125, "learning_rate": 9.048109721432634e-06, "loss": 0.9874, "step": 27290 }, { "epoch": 2.9039157267503724, "grad_norm": 15.84375, "learning_rate": 9.048074189880615e-06, "loss": 0.9275, "step": 27291 }, { "epoch": 2.9040221323685893, "grad_norm": 10.078125, "learning_rate": 9.048038658328597e-06, "loss": 1.0444, "step": 27292 }, { "epoch": 2.904128537986806, "grad_norm": 11.265625, "learning_rate": 9.048003126776579e-06, "loss": 0.8679, "step": 27293 }, { "epoch": 2.9042349436050223, "grad_norm": 14.53125, "learning_rate": 9.04796759522456e-06, "loss": 0.978, "step": 27294 }, { "epoch": 2.904341349223239, "grad_norm": 12.75, "learning_rate": 9.047932063672542e-06, "loss": 1.2147, "step": 27295 }, { "epoch": 2.904447754841456, "grad_norm": 15.46875, "learning_rate": 9.047896532120523e-06, "loss": 1.0846, "step": 27296 }, { "epoch": 2.9045541604596723, "grad_norm": 12.2265625, "learning_rate": 9.047861000568505e-06, "loss": 0.9488, "step": 27297 }, { "epoch": 2.904660566077889, "grad_norm": 18.3125, "learning_rate": 9.047825469016487e-06, "loss": 1.161, "step": 27298 }, { "epoch": 2.9047669716961053, "grad_norm": 13.0234375, "learning_rate": 9.04778993746447e-06, "loss": 0.8501, "step": 27299 }, { "epoch": 2.9048733773143223, "grad_norm": 22.046875, "learning_rate": 9.047754405912452e-06, "loss": 0.8461, "step": 27300 }, { "epoch": 2.904979782932539, "grad_norm": 13.9921875, "learning_rate": 9.047718874360433e-06, "loss": 0.941, "step": 27301 }, { "epoch": 2.9050861885507553, "grad_norm": 18.84375, "learning_rate": 9.047683342808415e-06, "loss": 0.8449, "step": 27302 }, { "epoch": 2.9051925941689722, "grad_norm": 15.28125, "learning_rate": 9.047647811256396e-06, "loss": 0.9361, "step": 27303 }, { "epoch": 2.9052989997871888, "grad_norm": 13.203125, "learning_rate": 9.047612279704378e-06, "loss": 0.9586, "step": 27304 }, { "epoch": 2.9054054054054053, "grad_norm": 16.296875, "learning_rate": 9.04757674815236e-06, "loss": 0.9272, "step": 27305 }, { "epoch": 2.905511811023622, "grad_norm": 18.140625, "learning_rate": 9.047541216600341e-06, "loss": 1.1975, "step": 27306 }, { "epoch": 2.9056182166418387, "grad_norm": 17.09375, "learning_rate": 9.047505685048323e-06, "loss": 1.0244, "step": 27307 }, { "epoch": 2.9057246222600552, "grad_norm": 12.578125, "learning_rate": 9.047470153496304e-06, "loss": 0.8884, "step": 27308 }, { "epoch": 2.905831027878272, "grad_norm": 17.4375, "learning_rate": 9.047434621944288e-06, "loss": 1.058, "step": 27309 }, { "epoch": 2.9059374334964887, "grad_norm": 17.71875, "learning_rate": 9.04739909039227e-06, "loss": 0.913, "step": 27310 }, { "epoch": 2.906043839114705, "grad_norm": 13.796875, "learning_rate": 9.047363558840251e-06, "loss": 0.989, "step": 27311 }, { "epoch": 2.906150244732922, "grad_norm": 11.203125, "learning_rate": 9.047328027288232e-06, "loss": 0.8894, "step": 27312 }, { "epoch": 2.9062566503511387, "grad_norm": 13.4609375, "learning_rate": 9.047292495736214e-06, "loss": 0.9895, "step": 27313 }, { "epoch": 2.906363055969355, "grad_norm": 18.34375, "learning_rate": 9.047256964184197e-06, "loss": 0.91, "step": 27314 }, { "epoch": 2.9064694615875717, "grad_norm": 13.7578125, "learning_rate": 9.047221432632179e-06, "loss": 0.837, "step": 27315 }, { "epoch": 2.906575867205788, "grad_norm": 13.4375, "learning_rate": 9.04718590108016e-06, "loss": 1.0593, "step": 27316 }, { "epoch": 2.906682272824005, "grad_norm": 12.3046875, "learning_rate": 9.047150369528142e-06, "loss": 1.0637, "step": 27317 }, { "epoch": 2.9067886784422217, "grad_norm": 13.5, "learning_rate": 9.047114837976124e-06, "loss": 0.901, "step": 27318 }, { "epoch": 2.906895084060438, "grad_norm": 11.796875, "learning_rate": 9.047079306424105e-06, "loss": 0.9603, "step": 27319 }, { "epoch": 2.907001489678655, "grad_norm": 11.59375, "learning_rate": 9.047043774872087e-06, "loss": 0.8961, "step": 27320 }, { "epoch": 2.9071078952968716, "grad_norm": 12.1171875, "learning_rate": 9.047008243320069e-06, "loss": 0.9552, "step": 27321 }, { "epoch": 2.907214300915088, "grad_norm": 15.4140625, "learning_rate": 9.04697271176805e-06, "loss": 1.0189, "step": 27322 }, { "epoch": 2.907320706533305, "grad_norm": 14.0234375, "learning_rate": 9.046937180216032e-06, "loss": 0.9823, "step": 27323 }, { "epoch": 2.9074271121515216, "grad_norm": 12.203125, "learning_rate": 9.046901648664015e-06, "loss": 0.8201, "step": 27324 }, { "epoch": 2.907533517769738, "grad_norm": 13.4921875, "learning_rate": 9.046866117111997e-06, "loss": 0.6206, "step": 27325 }, { "epoch": 2.907639923387955, "grad_norm": 11.546875, "learning_rate": 9.046830585559978e-06, "loss": 0.7085, "step": 27326 }, { "epoch": 2.9077463290061716, "grad_norm": 12.0625, "learning_rate": 9.04679505400796e-06, "loss": 0.7475, "step": 27327 }, { "epoch": 2.907852734624388, "grad_norm": 11.6875, "learning_rate": 9.046759522455942e-06, "loss": 0.739, "step": 27328 }, { "epoch": 2.907959140242605, "grad_norm": 12.25, "learning_rate": 9.046723990903923e-06, "loss": 0.762, "step": 27329 }, { "epoch": 2.9080655458608216, "grad_norm": 12.3515625, "learning_rate": 9.046688459351905e-06, "loss": 0.8852, "step": 27330 }, { "epoch": 2.908171951479038, "grad_norm": 11.7109375, "learning_rate": 9.046652927799886e-06, "loss": 0.9898, "step": 27331 }, { "epoch": 2.9082783570972546, "grad_norm": 11.4609375, "learning_rate": 9.046617396247868e-06, "loss": 0.7636, "step": 27332 }, { "epoch": 2.9083847627154715, "grad_norm": 10.0859375, "learning_rate": 9.04658186469585e-06, "loss": 0.8094, "step": 27333 }, { "epoch": 2.908491168333688, "grad_norm": 13.6796875, "learning_rate": 9.046546333143833e-06, "loss": 0.8434, "step": 27334 }, { "epoch": 2.9085975739519045, "grad_norm": 13.0078125, "learning_rate": 9.046510801591815e-06, "loss": 0.8555, "step": 27335 }, { "epoch": 2.908703979570121, "grad_norm": 11.046875, "learning_rate": 9.046475270039796e-06, "loss": 0.9902, "step": 27336 }, { "epoch": 2.908810385188338, "grad_norm": 14.59375, "learning_rate": 9.046439738487778e-06, "loss": 0.9035, "step": 27337 }, { "epoch": 2.9089167908065545, "grad_norm": 13.78125, "learning_rate": 9.04640420693576e-06, "loss": 1.1329, "step": 27338 }, { "epoch": 2.909023196424771, "grad_norm": 14.1171875, "learning_rate": 9.046368675383741e-06, "loss": 0.8741, "step": 27339 }, { "epoch": 2.909129602042988, "grad_norm": 13.7734375, "learning_rate": 9.046333143831724e-06, "loss": 0.8353, "step": 27340 }, { "epoch": 2.9092360076612045, "grad_norm": 13.4609375, "learning_rate": 9.046297612279706e-06, "loss": 0.7922, "step": 27341 }, { "epoch": 2.909342413279421, "grad_norm": 16.265625, "learning_rate": 9.046262080727687e-06, "loss": 1.0046, "step": 27342 }, { "epoch": 2.909448818897638, "grad_norm": 12.15625, "learning_rate": 9.046226549175669e-06, "loss": 0.893, "step": 27343 }, { "epoch": 2.9095552245158545, "grad_norm": 10.34375, "learning_rate": 9.04619101762365e-06, "loss": 0.9202, "step": 27344 }, { "epoch": 2.909661630134071, "grad_norm": 15.703125, "learning_rate": 9.046155486071632e-06, "loss": 0.8146, "step": 27345 }, { "epoch": 2.909768035752288, "grad_norm": 9.609375, "learning_rate": 9.046119954519614e-06, "loss": 0.808, "step": 27346 }, { "epoch": 2.9098744413705044, "grad_norm": 11.71875, "learning_rate": 9.046084422967596e-06, "loss": 0.8124, "step": 27347 }, { "epoch": 2.909980846988721, "grad_norm": 13.5078125, "learning_rate": 9.046048891415577e-06, "loss": 0.9987, "step": 27348 }, { "epoch": 2.910087252606938, "grad_norm": 12.421875, "learning_rate": 9.046013359863559e-06, "loss": 0.9977, "step": 27349 }, { "epoch": 2.9101936582251544, "grad_norm": 10.1875, "learning_rate": 9.045977828311542e-06, "loss": 0.9076, "step": 27350 }, { "epoch": 2.910300063843371, "grad_norm": 13.90625, "learning_rate": 9.045942296759524e-06, "loss": 0.8685, "step": 27351 }, { "epoch": 2.9104064694615874, "grad_norm": 15.09375, "learning_rate": 9.045906765207505e-06, "loss": 0.8772, "step": 27352 }, { "epoch": 2.910512875079804, "grad_norm": 11.1015625, "learning_rate": 9.045871233655487e-06, "loss": 0.8919, "step": 27353 }, { "epoch": 2.910619280698021, "grad_norm": 13.5390625, "learning_rate": 9.045835702103468e-06, "loss": 0.9638, "step": 27354 }, { "epoch": 2.9107256863162374, "grad_norm": 14.03125, "learning_rate": 9.04580017055145e-06, "loss": 0.958, "step": 27355 }, { "epoch": 2.910832091934454, "grad_norm": 14.7109375, "learning_rate": 9.045764638999432e-06, "loss": 0.9712, "step": 27356 }, { "epoch": 2.910938497552671, "grad_norm": 14.5859375, "learning_rate": 9.045729107447413e-06, "loss": 0.8469, "step": 27357 }, { "epoch": 2.9110449031708874, "grad_norm": 12.8203125, "learning_rate": 9.045693575895395e-06, "loss": 0.9313, "step": 27358 }, { "epoch": 2.911151308789104, "grad_norm": 12.3046875, "learning_rate": 9.045658044343376e-06, "loss": 0.9392, "step": 27359 }, { "epoch": 2.911257714407321, "grad_norm": 13.6171875, "learning_rate": 9.04562251279136e-06, "loss": 0.896, "step": 27360 }, { "epoch": 2.9113641200255373, "grad_norm": 12.765625, "learning_rate": 9.045586981239341e-06, "loss": 0.796, "step": 27361 }, { "epoch": 2.911470525643754, "grad_norm": 14.0234375, "learning_rate": 9.045551449687323e-06, "loss": 1.0338, "step": 27362 }, { "epoch": 2.911576931261971, "grad_norm": 10.7265625, "learning_rate": 9.045515918135305e-06, "loss": 0.7351, "step": 27363 }, { "epoch": 2.9116833368801873, "grad_norm": 14.8515625, "learning_rate": 9.045480386583286e-06, "loss": 0.8577, "step": 27364 }, { "epoch": 2.911789742498404, "grad_norm": 10.0234375, "learning_rate": 9.04544485503127e-06, "loss": 0.7907, "step": 27365 }, { "epoch": 2.911896148116621, "grad_norm": 15.921875, "learning_rate": 9.045409323479251e-06, "loss": 0.9653, "step": 27366 }, { "epoch": 2.9120025537348373, "grad_norm": 11.453125, "learning_rate": 9.045373791927233e-06, "loss": 0.8808, "step": 27367 }, { "epoch": 2.912108959353054, "grad_norm": 15.1484375, "learning_rate": 9.045338260375214e-06, "loss": 1.0028, "step": 27368 }, { "epoch": 2.9122153649712708, "grad_norm": 11.578125, "learning_rate": 9.045302728823196e-06, "loss": 0.9068, "step": 27369 }, { "epoch": 2.9123217705894873, "grad_norm": 10.5546875, "learning_rate": 9.045267197271178e-06, "loss": 0.7175, "step": 27370 }, { "epoch": 2.912428176207704, "grad_norm": 11.609375, "learning_rate": 9.045231665719159e-06, "loss": 0.9229, "step": 27371 }, { "epoch": 2.9125345818259203, "grad_norm": 11.90625, "learning_rate": 9.04519613416714e-06, "loss": 0.9404, "step": 27372 }, { "epoch": 2.912640987444137, "grad_norm": 16.578125, "learning_rate": 9.045160602615122e-06, "loss": 1.1315, "step": 27373 }, { "epoch": 2.9127473930623538, "grad_norm": 13.40625, "learning_rate": 9.045125071063104e-06, "loss": 0.6782, "step": 27374 }, { "epoch": 2.9128537986805703, "grad_norm": 13.8046875, "learning_rate": 9.045089539511086e-06, "loss": 0.9838, "step": 27375 }, { "epoch": 2.9129602042987868, "grad_norm": 13.890625, "learning_rate": 9.045054007959069e-06, "loss": 0.9211, "step": 27376 }, { "epoch": 2.9130666099170037, "grad_norm": 9.8125, "learning_rate": 9.04501847640705e-06, "loss": 0.8899, "step": 27377 }, { "epoch": 2.9131730155352202, "grad_norm": 12.6640625, "learning_rate": 9.044982944855032e-06, "loss": 0.9062, "step": 27378 }, { "epoch": 2.9132794211534367, "grad_norm": 16.03125, "learning_rate": 9.044947413303014e-06, "loss": 0.7981, "step": 27379 }, { "epoch": 2.9133858267716537, "grad_norm": 13.78125, "learning_rate": 9.044911881750995e-06, "loss": 1.0671, "step": 27380 }, { "epoch": 2.91349223238987, "grad_norm": 10.6796875, "learning_rate": 9.044876350198979e-06, "loss": 0.8135, "step": 27381 }, { "epoch": 2.9135986380080867, "grad_norm": 11.2109375, "learning_rate": 9.04484081864696e-06, "loss": 0.9482, "step": 27382 }, { "epoch": 2.9137050436263037, "grad_norm": 20.09375, "learning_rate": 9.044805287094942e-06, "loss": 0.9559, "step": 27383 }, { "epoch": 2.91381144924452, "grad_norm": 13.59375, "learning_rate": 9.044769755542923e-06, "loss": 0.805, "step": 27384 }, { "epoch": 2.9139178548627367, "grad_norm": 13.4921875, "learning_rate": 9.044734223990903e-06, "loss": 0.9727, "step": 27385 }, { "epoch": 2.9140242604809536, "grad_norm": 9.3046875, "learning_rate": 9.044698692438887e-06, "loss": 0.9403, "step": 27386 }, { "epoch": 2.91413066609917, "grad_norm": 9.96875, "learning_rate": 9.044663160886868e-06, "loss": 0.9105, "step": 27387 }, { "epoch": 2.9142370717173867, "grad_norm": 14.6328125, "learning_rate": 9.04462762933485e-06, "loss": 1.0323, "step": 27388 }, { "epoch": 2.914343477335603, "grad_norm": 9.015625, "learning_rate": 9.044592097782831e-06, "loss": 0.8973, "step": 27389 }, { "epoch": 2.91444988295382, "grad_norm": 15.1484375, "learning_rate": 9.044556566230813e-06, "loss": 1.0777, "step": 27390 }, { "epoch": 2.9145562885720366, "grad_norm": 10.0703125, "learning_rate": 9.044521034678796e-06, "loss": 0.9408, "step": 27391 }, { "epoch": 2.914662694190253, "grad_norm": 12.8203125, "learning_rate": 9.044485503126778e-06, "loss": 0.9278, "step": 27392 }, { "epoch": 2.9147690998084697, "grad_norm": 14.4375, "learning_rate": 9.04444997157476e-06, "loss": 0.7723, "step": 27393 }, { "epoch": 2.9148755054266866, "grad_norm": 16.46875, "learning_rate": 9.044414440022741e-06, "loss": 0.9637, "step": 27394 }, { "epoch": 2.914981911044903, "grad_norm": 11.0, "learning_rate": 9.044378908470723e-06, "loss": 0.8275, "step": 27395 }, { "epoch": 2.9150883166631196, "grad_norm": 11.875, "learning_rate": 9.044343376918704e-06, "loss": 0.8077, "step": 27396 }, { "epoch": 2.9151947222813366, "grad_norm": 12.0703125, "learning_rate": 9.044307845366686e-06, "loss": 0.7571, "step": 27397 }, { "epoch": 2.915301127899553, "grad_norm": 10.3046875, "learning_rate": 9.044272313814668e-06, "loss": 0.9071, "step": 27398 }, { "epoch": 2.9154075335177696, "grad_norm": 13.0078125, "learning_rate": 9.04423678226265e-06, "loss": 0.8276, "step": 27399 }, { "epoch": 2.9155139391359866, "grad_norm": 11.203125, "learning_rate": 9.04420125071063e-06, "loss": 0.8297, "step": 27400 }, { "epoch": 2.915620344754203, "grad_norm": 10.5, "learning_rate": 9.044165719158614e-06, "loss": 0.9171, "step": 27401 }, { "epoch": 2.9157267503724196, "grad_norm": 15.1015625, "learning_rate": 9.044130187606596e-06, "loss": 0.8745, "step": 27402 }, { "epoch": 2.9158331559906365, "grad_norm": 13.8125, "learning_rate": 9.044094656054577e-06, "loss": 1.0979, "step": 27403 }, { "epoch": 2.915939561608853, "grad_norm": 16.78125, "learning_rate": 9.044059124502559e-06, "loss": 0.9154, "step": 27404 }, { "epoch": 2.9160459672270695, "grad_norm": 17.609375, "learning_rate": 9.04402359295054e-06, "loss": 0.7895, "step": 27405 }, { "epoch": 2.9161523728452865, "grad_norm": 10.78125, "learning_rate": 9.043988061398524e-06, "loss": 0.6605, "step": 27406 }, { "epoch": 2.916258778463503, "grad_norm": 11.625, "learning_rate": 9.043952529846505e-06, "loss": 0.891, "step": 27407 }, { "epoch": 2.9163651840817195, "grad_norm": 12.7421875, "learning_rate": 9.043916998294487e-06, "loss": 0.9672, "step": 27408 }, { "epoch": 2.916471589699936, "grad_norm": 12.1484375, "learning_rate": 9.043881466742469e-06, "loss": 0.8809, "step": 27409 }, { "epoch": 2.9165779953181525, "grad_norm": 14.0234375, "learning_rate": 9.04384593519045e-06, "loss": 0.8669, "step": 27410 }, { "epoch": 2.9166844009363695, "grad_norm": 12.2421875, "learning_rate": 9.043810403638432e-06, "loss": 0.935, "step": 27411 }, { "epoch": 2.916790806554586, "grad_norm": 14.4140625, "learning_rate": 9.043774872086413e-06, "loss": 1.0995, "step": 27412 }, { "epoch": 2.9168972121728025, "grad_norm": 12.8125, "learning_rate": 9.043739340534395e-06, "loss": 1.0977, "step": 27413 }, { "epoch": 2.9170036177910195, "grad_norm": 14.0078125, "learning_rate": 9.043703808982377e-06, "loss": 0.939, "step": 27414 }, { "epoch": 2.917110023409236, "grad_norm": 16.515625, "learning_rate": 9.043668277430358e-06, "loss": 0.9943, "step": 27415 }, { "epoch": 2.9172164290274525, "grad_norm": 10.4765625, "learning_rate": 9.04363274587834e-06, "loss": 0.9582, "step": 27416 }, { "epoch": 2.9173228346456694, "grad_norm": 14.1171875, "learning_rate": 9.043597214326323e-06, "loss": 1.0282, "step": 27417 }, { "epoch": 2.917429240263886, "grad_norm": 13.0859375, "learning_rate": 9.043561682774305e-06, "loss": 0.8096, "step": 27418 }, { "epoch": 2.9175356458821025, "grad_norm": 13.6484375, "learning_rate": 9.043526151222286e-06, "loss": 1.049, "step": 27419 }, { "epoch": 2.9176420515003194, "grad_norm": 15.203125, "learning_rate": 9.043490619670268e-06, "loss": 0.8764, "step": 27420 }, { "epoch": 2.917748457118536, "grad_norm": 12.3203125, "learning_rate": 9.04345508811825e-06, "loss": 1.0422, "step": 27421 }, { "epoch": 2.9178548627367524, "grad_norm": 16.34375, "learning_rate": 9.043419556566231e-06, "loss": 0.9267, "step": 27422 }, { "epoch": 2.9179612683549694, "grad_norm": 10.3203125, "learning_rate": 9.043384025014213e-06, "loss": 0.9473, "step": 27423 }, { "epoch": 2.918067673973186, "grad_norm": 11.96875, "learning_rate": 9.043348493462194e-06, "loss": 0.9854, "step": 27424 }, { "epoch": 2.9181740795914024, "grad_norm": 14.4765625, "learning_rate": 9.043312961910176e-06, "loss": 0.9986, "step": 27425 }, { "epoch": 2.918280485209619, "grad_norm": 18.0, "learning_rate": 9.043277430358158e-06, "loss": 1.1926, "step": 27426 }, { "epoch": 2.918386890827836, "grad_norm": 11.5859375, "learning_rate": 9.043241898806141e-06, "loss": 1.0887, "step": 27427 }, { "epoch": 2.9184932964460524, "grad_norm": 16.171875, "learning_rate": 9.043206367254123e-06, "loss": 0.9253, "step": 27428 }, { "epoch": 2.918599702064269, "grad_norm": 16.265625, "learning_rate": 9.043170835702104e-06, "loss": 0.9612, "step": 27429 }, { "epoch": 2.9187061076824854, "grad_norm": 17.078125, "learning_rate": 9.043135304150086e-06, "loss": 0.7967, "step": 27430 }, { "epoch": 2.9188125133007023, "grad_norm": 13.40625, "learning_rate": 9.043099772598067e-06, "loss": 0.8658, "step": 27431 }, { "epoch": 2.918918918918919, "grad_norm": 15.75, "learning_rate": 9.04306424104605e-06, "loss": 0.9575, "step": 27432 }, { "epoch": 2.9190253245371354, "grad_norm": 10.3046875, "learning_rate": 9.043028709494032e-06, "loss": 0.849, "step": 27433 }, { "epoch": 2.9191317301553523, "grad_norm": 12.5859375, "learning_rate": 9.042993177942014e-06, "loss": 1.1163, "step": 27434 }, { "epoch": 2.919238135773569, "grad_norm": 12.046875, "learning_rate": 9.042957646389996e-06, "loss": 0.8857, "step": 27435 }, { "epoch": 2.9193445413917853, "grad_norm": 16.828125, "learning_rate": 9.042922114837977e-06, "loss": 1.0627, "step": 27436 }, { "epoch": 2.9194509470100023, "grad_norm": 13.4609375, "learning_rate": 9.042886583285959e-06, "loss": 0.9951, "step": 27437 }, { "epoch": 2.919557352628219, "grad_norm": 15.0234375, "learning_rate": 9.04285105173394e-06, "loss": 0.8894, "step": 27438 }, { "epoch": 2.9196637582464353, "grad_norm": 16.171875, "learning_rate": 9.042815520181922e-06, "loss": 1.0048, "step": 27439 }, { "epoch": 2.9197701638646523, "grad_norm": 12.75, "learning_rate": 9.042779988629904e-06, "loss": 1.0558, "step": 27440 }, { "epoch": 2.9198765694828688, "grad_norm": 15.25, "learning_rate": 9.042744457077885e-06, "loss": 0.9744, "step": 27441 }, { "epoch": 2.9199829751010853, "grad_norm": 10.40625, "learning_rate": 9.042708925525868e-06, "loss": 0.927, "step": 27442 }, { "epoch": 2.9200893807193022, "grad_norm": 12.59375, "learning_rate": 9.04267339397385e-06, "loss": 0.9811, "step": 27443 }, { "epoch": 2.9201957863375188, "grad_norm": 12.3046875, "learning_rate": 9.042637862421832e-06, "loss": 0.819, "step": 27444 }, { "epoch": 2.9203021919557353, "grad_norm": 18.65625, "learning_rate": 9.042602330869813e-06, "loss": 0.9829, "step": 27445 }, { "epoch": 2.9204085975739518, "grad_norm": 10.703125, "learning_rate": 9.042566799317795e-06, "loss": 0.7556, "step": 27446 }, { "epoch": 2.9205150031921683, "grad_norm": 12.03125, "learning_rate": 9.042531267765776e-06, "loss": 0.8001, "step": 27447 }, { "epoch": 2.9206214088103852, "grad_norm": 16.9375, "learning_rate": 9.042495736213758e-06, "loss": 1.104, "step": 27448 }, { "epoch": 2.9207278144286017, "grad_norm": 17.890625, "learning_rate": 9.04246020466174e-06, "loss": 0.9385, "step": 27449 }, { "epoch": 2.9208342200468183, "grad_norm": 17.578125, "learning_rate": 9.042424673109721e-06, "loss": 1.0033, "step": 27450 }, { "epoch": 2.920940625665035, "grad_norm": 12.375, "learning_rate": 9.042389141557703e-06, "loss": 0.9751, "step": 27451 }, { "epoch": 2.9210470312832517, "grad_norm": 16.5, "learning_rate": 9.042353610005685e-06, "loss": 0.7815, "step": 27452 }, { "epoch": 2.9211534369014682, "grad_norm": 12.6015625, "learning_rate": 9.042318078453668e-06, "loss": 0.9782, "step": 27453 }, { "epoch": 2.921259842519685, "grad_norm": 12.546875, "learning_rate": 9.04228254690165e-06, "loss": 0.8063, "step": 27454 }, { "epoch": 2.9213662481379017, "grad_norm": 11.5234375, "learning_rate": 9.042247015349631e-06, "loss": 0.904, "step": 27455 }, { "epoch": 2.921472653756118, "grad_norm": 14.3671875, "learning_rate": 9.042211483797613e-06, "loss": 0.927, "step": 27456 }, { "epoch": 2.921579059374335, "grad_norm": 14.3828125, "learning_rate": 9.042175952245594e-06, "loss": 0.965, "step": 27457 }, { "epoch": 2.9216854649925517, "grad_norm": 11.3125, "learning_rate": 9.042140420693578e-06, "loss": 0.8754, "step": 27458 }, { "epoch": 2.921791870610768, "grad_norm": 23.3125, "learning_rate": 9.04210488914156e-06, "loss": 1.1137, "step": 27459 }, { "epoch": 2.921898276228985, "grad_norm": 11.9765625, "learning_rate": 9.04206935758954e-06, "loss": 1.0987, "step": 27460 }, { "epoch": 2.9220046818472016, "grad_norm": 24.6875, "learning_rate": 9.042033826037522e-06, "loss": 1.1623, "step": 27461 }, { "epoch": 2.922111087465418, "grad_norm": 13.390625, "learning_rate": 9.041998294485504e-06, "loss": 1.0677, "step": 27462 }, { "epoch": 2.9222174930836347, "grad_norm": 15.6015625, "learning_rate": 9.041962762933486e-06, "loss": 0.987, "step": 27463 }, { "epoch": 2.9223238987018516, "grad_norm": 12.375, "learning_rate": 9.041927231381467e-06, "loss": 0.9278, "step": 27464 }, { "epoch": 2.922430304320068, "grad_norm": 8.453125, "learning_rate": 9.041891699829449e-06, "loss": 0.6723, "step": 27465 }, { "epoch": 2.9225367099382846, "grad_norm": 15.8046875, "learning_rate": 9.04185616827743e-06, "loss": 0.962, "step": 27466 }, { "epoch": 2.922643115556501, "grad_norm": 12.5078125, "learning_rate": 9.041820636725412e-06, "loss": 0.9242, "step": 27467 }, { "epoch": 2.922749521174718, "grad_norm": 10.6171875, "learning_rate": 9.041785105173395e-06, "loss": 0.9813, "step": 27468 }, { "epoch": 2.9228559267929346, "grad_norm": 12.2734375, "learning_rate": 9.041749573621377e-06, "loss": 0.9631, "step": 27469 }, { "epoch": 2.922962332411151, "grad_norm": 13.484375, "learning_rate": 9.041714042069359e-06, "loss": 1.0778, "step": 27470 }, { "epoch": 2.923068738029368, "grad_norm": 11.875, "learning_rate": 9.04167851051734e-06, "loss": 0.9672, "step": 27471 }, { "epoch": 2.9231751436475846, "grad_norm": 18.453125, "learning_rate": 9.041642978965322e-06, "loss": 0.8292, "step": 27472 }, { "epoch": 2.923281549265801, "grad_norm": 11.46875, "learning_rate": 9.041607447413305e-06, "loss": 0.7104, "step": 27473 }, { "epoch": 2.923387954884018, "grad_norm": 13.0859375, "learning_rate": 9.041571915861285e-06, "loss": 0.8289, "step": 27474 }, { "epoch": 2.9234943605022345, "grad_norm": 11.1484375, "learning_rate": 9.041536384309267e-06, "loss": 0.9193, "step": 27475 }, { "epoch": 2.923600766120451, "grad_norm": 15.7734375, "learning_rate": 9.041500852757248e-06, "loss": 0.8871, "step": 27476 }, { "epoch": 2.923707171738668, "grad_norm": 11.8359375, "learning_rate": 9.04146532120523e-06, "loss": 0.7575, "step": 27477 }, { "epoch": 2.9238135773568845, "grad_norm": 13.671875, "learning_rate": 9.041429789653213e-06, "loss": 0.8564, "step": 27478 }, { "epoch": 2.923919982975101, "grad_norm": 16.546875, "learning_rate": 9.041394258101195e-06, "loss": 1.0652, "step": 27479 }, { "epoch": 2.924026388593318, "grad_norm": 15.0234375, "learning_rate": 9.041358726549176e-06, "loss": 1.1051, "step": 27480 }, { "epoch": 2.9241327942115345, "grad_norm": 12.3203125, "learning_rate": 9.041323194997158e-06, "loss": 0.8584, "step": 27481 }, { "epoch": 2.924239199829751, "grad_norm": 14.53125, "learning_rate": 9.04128766344514e-06, "loss": 0.9369, "step": 27482 }, { "epoch": 2.9243456054479675, "grad_norm": 15.578125, "learning_rate": 9.041252131893123e-06, "loss": 0.8466, "step": 27483 }, { "epoch": 2.924452011066184, "grad_norm": 18.859375, "learning_rate": 9.041216600341104e-06, "loss": 0.988, "step": 27484 }, { "epoch": 2.924558416684401, "grad_norm": 12.4765625, "learning_rate": 9.041181068789086e-06, "loss": 0.9146, "step": 27485 }, { "epoch": 2.9246648223026175, "grad_norm": 10.484375, "learning_rate": 9.041145537237068e-06, "loss": 0.7927, "step": 27486 }, { "epoch": 2.924771227920834, "grad_norm": 14.0234375, "learning_rate": 9.04111000568505e-06, "loss": 0.9064, "step": 27487 }, { "epoch": 2.924877633539051, "grad_norm": 11.4140625, "learning_rate": 9.04107447413303e-06, "loss": 0.9409, "step": 27488 }, { "epoch": 2.9249840391572675, "grad_norm": 11.5, "learning_rate": 9.041038942581012e-06, "loss": 0.7751, "step": 27489 }, { "epoch": 2.925090444775484, "grad_norm": 14.203125, "learning_rate": 9.041003411028994e-06, "loss": 1.0551, "step": 27490 }, { "epoch": 2.925196850393701, "grad_norm": 10.015625, "learning_rate": 9.040967879476976e-06, "loss": 0.8671, "step": 27491 }, { "epoch": 2.9253032560119174, "grad_norm": 10.3671875, "learning_rate": 9.040932347924957e-06, "loss": 0.7521, "step": 27492 }, { "epoch": 2.925409661630134, "grad_norm": 18.453125, "learning_rate": 9.040896816372939e-06, "loss": 0.9561, "step": 27493 }, { "epoch": 2.925516067248351, "grad_norm": 12.5546875, "learning_rate": 9.040861284820922e-06, "loss": 0.9851, "step": 27494 }, { "epoch": 2.9256224728665674, "grad_norm": 12.9296875, "learning_rate": 9.040825753268904e-06, "loss": 1.0349, "step": 27495 }, { "epoch": 2.925728878484784, "grad_norm": 12.578125, "learning_rate": 9.040790221716885e-06, "loss": 0.994, "step": 27496 }, { "epoch": 2.925835284103001, "grad_norm": 17.3125, "learning_rate": 9.040754690164867e-06, "loss": 0.8908, "step": 27497 }, { "epoch": 2.9259416897212174, "grad_norm": 12.765625, "learning_rate": 9.040719158612849e-06, "loss": 0.9864, "step": 27498 }, { "epoch": 2.926048095339434, "grad_norm": 9.96875, "learning_rate": 9.040683627060832e-06, "loss": 0.9215, "step": 27499 }, { "epoch": 2.926154500957651, "grad_norm": 21.5625, "learning_rate": 9.040648095508814e-06, "loss": 1.0694, "step": 27500 }, { "epoch": 2.9262609065758673, "grad_norm": 17.0, "learning_rate": 9.040612563956795e-06, "loss": 1.0051, "step": 27501 }, { "epoch": 2.926367312194084, "grad_norm": 16.625, "learning_rate": 9.040577032404777e-06, "loss": 0.9771, "step": 27502 }, { "epoch": 2.9264737178123004, "grad_norm": 11.1875, "learning_rate": 9.040541500852758e-06, "loss": 0.8245, "step": 27503 }, { "epoch": 2.926580123430517, "grad_norm": 11.6328125, "learning_rate": 9.04050596930074e-06, "loss": 1.0206, "step": 27504 }, { "epoch": 2.926686529048734, "grad_norm": 13.4296875, "learning_rate": 9.040470437748722e-06, "loss": 0.9244, "step": 27505 }, { "epoch": 2.9267929346669503, "grad_norm": 12.203125, "learning_rate": 9.040434906196703e-06, "loss": 0.8621, "step": 27506 }, { "epoch": 2.926899340285167, "grad_norm": 14.3984375, "learning_rate": 9.040399374644685e-06, "loss": 1.0764, "step": 27507 }, { "epoch": 2.927005745903384, "grad_norm": 10.9453125, "learning_rate": 9.040363843092666e-06, "loss": 0.7757, "step": 27508 }, { "epoch": 2.9271121515216003, "grad_norm": 11.9921875, "learning_rate": 9.04032831154065e-06, "loss": 0.8335, "step": 27509 }, { "epoch": 2.927218557139817, "grad_norm": 11.1015625, "learning_rate": 9.040292779988631e-06, "loss": 0.8804, "step": 27510 }, { "epoch": 2.9273249627580338, "grad_norm": 14.90625, "learning_rate": 9.040257248436613e-06, "loss": 1.002, "step": 27511 }, { "epoch": 2.9274313683762503, "grad_norm": 11.8984375, "learning_rate": 9.040221716884594e-06, "loss": 0.8699, "step": 27512 }, { "epoch": 2.927537773994467, "grad_norm": 15.21875, "learning_rate": 9.040186185332576e-06, "loss": 1.081, "step": 27513 }, { "epoch": 2.9276441796126838, "grad_norm": 19.5, "learning_rate": 9.040150653780558e-06, "loss": 0.8646, "step": 27514 }, { "epoch": 2.9277505852309003, "grad_norm": 13.7109375, "learning_rate": 9.04011512222854e-06, "loss": 0.9418, "step": 27515 }, { "epoch": 2.9278569908491168, "grad_norm": 12.828125, "learning_rate": 9.040079590676521e-06, "loss": 1.0328, "step": 27516 }, { "epoch": 2.9279633964673337, "grad_norm": 11.9765625, "learning_rate": 9.040044059124503e-06, "loss": 0.983, "step": 27517 }, { "epoch": 2.9280698020855502, "grad_norm": 10.5703125, "learning_rate": 9.040008527572484e-06, "loss": 0.8445, "step": 27518 }, { "epoch": 2.9281762077037667, "grad_norm": 13.0859375, "learning_rate": 9.039972996020467e-06, "loss": 0.9762, "step": 27519 }, { "epoch": 2.9282826133219833, "grad_norm": 12.9921875, "learning_rate": 9.039937464468449e-06, "loss": 0.9044, "step": 27520 }, { "epoch": 2.9283890189402, "grad_norm": 11.2890625, "learning_rate": 9.03990193291643e-06, "loss": 0.7679, "step": 27521 }, { "epoch": 2.9284954245584167, "grad_norm": 14.1953125, "learning_rate": 9.039866401364412e-06, "loss": 1.0856, "step": 27522 }, { "epoch": 2.9286018301766332, "grad_norm": 11.6484375, "learning_rate": 9.039830869812394e-06, "loss": 0.8202, "step": 27523 }, { "epoch": 2.9287082357948497, "grad_norm": 10.0546875, "learning_rate": 9.039795338260377e-06, "loss": 0.8923, "step": 27524 }, { "epoch": 2.9288146414130667, "grad_norm": 17.625, "learning_rate": 9.039759806708359e-06, "loss": 0.9886, "step": 27525 }, { "epoch": 2.928921047031283, "grad_norm": 13.1640625, "learning_rate": 9.03972427515634e-06, "loss": 0.9925, "step": 27526 }, { "epoch": 2.9290274526494997, "grad_norm": 12.9765625, "learning_rate": 9.039688743604322e-06, "loss": 1.0292, "step": 27527 }, { "epoch": 2.9291338582677167, "grad_norm": 10.046875, "learning_rate": 9.039653212052304e-06, "loss": 0.8769, "step": 27528 }, { "epoch": 2.929240263885933, "grad_norm": 10.2109375, "learning_rate": 9.039617680500285e-06, "loss": 0.9113, "step": 27529 }, { "epoch": 2.9293466695041497, "grad_norm": 14.40625, "learning_rate": 9.039582148948267e-06, "loss": 1.0628, "step": 27530 }, { "epoch": 2.9294530751223666, "grad_norm": 14.0390625, "learning_rate": 9.039546617396248e-06, "loss": 0.8698, "step": 27531 }, { "epoch": 2.929559480740583, "grad_norm": 12.9921875, "learning_rate": 9.03951108584423e-06, "loss": 0.9792, "step": 27532 }, { "epoch": 2.9296658863587997, "grad_norm": 12.9453125, "learning_rate": 9.039475554292212e-06, "loss": 0.9922, "step": 27533 }, { "epoch": 2.9297722919770166, "grad_norm": 11.5, "learning_rate": 9.039440022740193e-06, "loss": 0.9259, "step": 27534 }, { "epoch": 2.929878697595233, "grad_norm": 12.609375, "learning_rate": 9.039404491188177e-06, "loss": 1.0009, "step": 27535 }, { "epoch": 2.9299851032134496, "grad_norm": 14.1328125, "learning_rate": 9.039368959636158e-06, "loss": 0.9845, "step": 27536 }, { "epoch": 2.9300915088316666, "grad_norm": 13.65625, "learning_rate": 9.03933342808414e-06, "loss": 0.8271, "step": 27537 }, { "epoch": 2.930197914449883, "grad_norm": 12.34375, "learning_rate": 9.039297896532121e-06, "loss": 0.9536, "step": 27538 }, { "epoch": 2.9303043200680996, "grad_norm": 9.1796875, "learning_rate": 9.039262364980103e-06, "loss": 0.8166, "step": 27539 }, { "epoch": 2.930410725686316, "grad_norm": 12.484375, "learning_rate": 9.039226833428085e-06, "loss": 0.9637, "step": 27540 }, { "epoch": 2.9305171313045326, "grad_norm": 14.6015625, "learning_rate": 9.039191301876066e-06, "loss": 0.9679, "step": 27541 }, { "epoch": 2.9306235369227496, "grad_norm": 15.2890625, "learning_rate": 9.039155770324048e-06, "loss": 1.0245, "step": 27542 }, { "epoch": 2.930729942540966, "grad_norm": 25.0625, "learning_rate": 9.03912023877203e-06, "loss": 1.1851, "step": 27543 }, { "epoch": 2.9308363481591826, "grad_norm": 13.28125, "learning_rate": 9.039084707220011e-06, "loss": 1.1501, "step": 27544 }, { "epoch": 2.9309427537773995, "grad_norm": 14.4765625, "learning_rate": 9.039049175667994e-06, "loss": 0.7981, "step": 27545 }, { "epoch": 2.931049159395616, "grad_norm": 15.5390625, "learning_rate": 9.039013644115976e-06, "loss": 0.7177, "step": 27546 }, { "epoch": 2.9311555650138326, "grad_norm": 15.8984375, "learning_rate": 9.038978112563957e-06, "loss": 1.0303, "step": 27547 }, { "epoch": 2.9312619706320495, "grad_norm": 10.5, "learning_rate": 9.038942581011939e-06, "loss": 1.0037, "step": 27548 }, { "epoch": 2.931368376250266, "grad_norm": 18.078125, "learning_rate": 9.03890704945992e-06, "loss": 0.9388, "step": 27549 }, { "epoch": 2.9314747818684825, "grad_norm": 14.7734375, "learning_rate": 9.038871517907904e-06, "loss": 0.9962, "step": 27550 }, { "epoch": 2.9315811874866995, "grad_norm": 12.390625, "learning_rate": 9.038835986355886e-06, "loss": 0.9711, "step": 27551 }, { "epoch": 2.931687593104916, "grad_norm": 15.765625, "learning_rate": 9.038800454803867e-06, "loss": 1.107, "step": 27552 }, { "epoch": 2.9317939987231325, "grad_norm": 12.3515625, "learning_rate": 9.038764923251849e-06, "loss": 1.126, "step": 27553 }, { "epoch": 2.9319004043413495, "grad_norm": 14.4375, "learning_rate": 9.03872939169983e-06, "loss": 1.1383, "step": 27554 }, { "epoch": 2.932006809959566, "grad_norm": 15.203125, "learning_rate": 9.038693860147812e-06, "loss": 1.1244, "step": 27555 }, { "epoch": 2.9321132155777825, "grad_norm": 13.671875, "learning_rate": 9.038658328595794e-06, "loss": 0.8585, "step": 27556 }, { "epoch": 2.932219621195999, "grad_norm": 10.71875, "learning_rate": 9.038622797043775e-06, "loss": 0.9743, "step": 27557 }, { "epoch": 2.932326026814216, "grad_norm": 10.421875, "learning_rate": 9.038587265491757e-06, "loss": 0.8084, "step": 27558 }, { "epoch": 2.9324324324324325, "grad_norm": 10.015625, "learning_rate": 9.038551733939738e-06, "loss": 0.8998, "step": 27559 }, { "epoch": 2.932538838050649, "grad_norm": 9.515625, "learning_rate": 9.038516202387722e-06, "loss": 0.9526, "step": 27560 }, { "epoch": 2.9326452436688655, "grad_norm": 14.390625, "learning_rate": 9.038480670835703e-06, "loss": 1.19, "step": 27561 }, { "epoch": 2.9327516492870824, "grad_norm": 11.5703125, "learning_rate": 9.038445139283685e-06, "loss": 1.0589, "step": 27562 }, { "epoch": 2.932858054905299, "grad_norm": 15.6328125, "learning_rate": 9.038409607731667e-06, "loss": 1.001, "step": 27563 }, { "epoch": 2.9329644605235154, "grad_norm": 13.5234375, "learning_rate": 9.038374076179648e-06, "loss": 1.1248, "step": 27564 }, { "epoch": 2.9330708661417324, "grad_norm": 11.921875, "learning_rate": 9.03833854462763e-06, "loss": 0.9369, "step": 27565 }, { "epoch": 2.933177271759949, "grad_norm": 19.59375, "learning_rate": 9.038303013075611e-06, "loss": 0.9343, "step": 27566 }, { "epoch": 2.9332836773781654, "grad_norm": 11.234375, "learning_rate": 9.038267481523593e-06, "loss": 0.8737, "step": 27567 }, { "epoch": 2.9333900829963824, "grad_norm": 14.1875, "learning_rate": 9.038231949971575e-06, "loss": 0.8661, "step": 27568 }, { "epoch": 2.933496488614599, "grad_norm": 13.375, "learning_rate": 9.038196418419556e-06, "loss": 1.0213, "step": 27569 }, { "epoch": 2.9336028942328154, "grad_norm": 13.5546875, "learning_rate": 9.038160886867538e-06, "loss": 0.9002, "step": 27570 }, { "epoch": 2.9337092998510323, "grad_norm": 12.609375, "learning_rate": 9.038125355315521e-06, "loss": 0.842, "step": 27571 }, { "epoch": 2.933815705469249, "grad_norm": 12.0390625, "learning_rate": 9.038089823763503e-06, "loss": 0.8252, "step": 27572 }, { "epoch": 2.9339221110874654, "grad_norm": 11.0546875, "learning_rate": 9.038054292211484e-06, "loss": 0.9424, "step": 27573 }, { "epoch": 2.9340285167056823, "grad_norm": 13.5078125, "learning_rate": 9.038018760659466e-06, "loss": 1.0958, "step": 27574 }, { "epoch": 2.934134922323899, "grad_norm": 11.296875, "learning_rate": 9.037983229107448e-06, "loss": 1.0558, "step": 27575 }, { "epoch": 2.9342413279421153, "grad_norm": 18.078125, "learning_rate": 9.037947697555431e-06, "loss": 0.972, "step": 27576 }, { "epoch": 2.934347733560332, "grad_norm": 12.5390625, "learning_rate": 9.037912166003412e-06, "loss": 0.8241, "step": 27577 }, { "epoch": 2.9344541391785484, "grad_norm": 9.9765625, "learning_rate": 9.037876634451394e-06, "loss": 0.7836, "step": 27578 }, { "epoch": 2.9345605447967653, "grad_norm": 12.1640625, "learning_rate": 9.037841102899376e-06, "loss": 1.0641, "step": 27579 }, { "epoch": 2.934666950414982, "grad_norm": 14.703125, "learning_rate": 9.037805571347357e-06, "loss": 0.799, "step": 27580 }, { "epoch": 2.9347733560331983, "grad_norm": 10.09375, "learning_rate": 9.037770039795339e-06, "loss": 0.8181, "step": 27581 }, { "epoch": 2.9348797616514153, "grad_norm": 12.859375, "learning_rate": 9.03773450824332e-06, "loss": 0.8417, "step": 27582 }, { "epoch": 2.934986167269632, "grad_norm": 11.3359375, "learning_rate": 9.037698976691302e-06, "loss": 0.9102, "step": 27583 }, { "epoch": 2.9350925728878483, "grad_norm": 12.9453125, "learning_rate": 9.037663445139284e-06, "loss": 0.8617, "step": 27584 }, { "epoch": 2.9351989785060653, "grad_norm": 14.640625, "learning_rate": 9.037627913587265e-06, "loss": 1.1531, "step": 27585 }, { "epoch": 2.9353053841242818, "grad_norm": 13.03125, "learning_rate": 9.037592382035249e-06, "loss": 0.8695, "step": 27586 }, { "epoch": 2.9354117897424983, "grad_norm": 13.140625, "learning_rate": 9.03755685048323e-06, "loss": 0.8926, "step": 27587 }, { "epoch": 2.9355181953607152, "grad_norm": 14.859375, "learning_rate": 9.037521318931212e-06, "loss": 1.0126, "step": 27588 }, { "epoch": 2.9356246009789317, "grad_norm": 14.6484375, "learning_rate": 9.037485787379193e-06, "loss": 0.9967, "step": 27589 }, { "epoch": 2.9357310065971483, "grad_norm": 13.90625, "learning_rate": 9.037450255827175e-06, "loss": 0.8451, "step": 27590 }, { "epoch": 2.935837412215365, "grad_norm": 13.8515625, "learning_rate": 9.037414724275158e-06, "loss": 0.9946, "step": 27591 }, { "epoch": 2.9359438178335817, "grad_norm": 12.15625, "learning_rate": 9.03737919272314e-06, "loss": 1.0453, "step": 27592 }, { "epoch": 2.9360502234517982, "grad_norm": 13.140625, "learning_rate": 9.037343661171122e-06, "loss": 0.8131, "step": 27593 }, { "epoch": 2.9361566290700147, "grad_norm": 10.9453125, "learning_rate": 9.037308129619101e-06, "loss": 0.6319, "step": 27594 }, { "epoch": 2.9362630346882317, "grad_norm": 12.421875, "learning_rate": 9.037272598067083e-06, "loss": 0.8675, "step": 27595 }, { "epoch": 2.936369440306448, "grad_norm": 10.90625, "learning_rate": 9.037237066515066e-06, "loss": 1.1403, "step": 27596 }, { "epoch": 2.9364758459246647, "grad_norm": 12.3359375, "learning_rate": 9.037201534963048e-06, "loss": 0.8103, "step": 27597 }, { "epoch": 2.936582251542881, "grad_norm": 17.25, "learning_rate": 9.03716600341103e-06, "loss": 1.1029, "step": 27598 }, { "epoch": 2.936688657161098, "grad_norm": 16.203125, "learning_rate": 9.037130471859011e-06, "loss": 0.8801, "step": 27599 }, { "epoch": 2.9367950627793147, "grad_norm": 11.984375, "learning_rate": 9.037094940306993e-06, "loss": 0.9131, "step": 27600 }, { "epoch": 2.936901468397531, "grad_norm": 18.40625, "learning_rate": 9.037059408754976e-06, "loss": 0.9174, "step": 27601 }, { "epoch": 2.937007874015748, "grad_norm": 11.3125, "learning_rate": 9.037023877202958e-06, "loss": 0.867, "step": 27602 }, { "epoch": 2.9371142796339647, "grad_norm": 12.609375, "learning_rate": 9.03698834565094e-06, "loss": 0.8354, "step": 27603 }, { "epoch": 2.937220685252181, "grad_norm": 17.046875, "learning_rate": 9.036952814098921e-06, "loss": 0.8521, "step": 27604 }, { "epoch": 2.937327090870398, "grad_norm": 13.9453125, "learning_rate": 9.036917282546903e-06, "loss": 0.9986, "step": 27605 }, { "epoch": 2.9374334964886146, "grad_norm": 13.0703125, "learning_rate": 9.036881750994884e-06, "loss": 0.9198, "step": 27606 }, { "epoch": 2.937539902106831, "grad_norm": 12.4375, "learning_rate": 9.036846219442866e-06, "loss": 1.054, "step": 27607 }, { "epoch": 2.937646307725048, "grad_norm": 15.3828125, "learning_rate": 9.036810687890847e-06, "loss": 0.8678, "step": 27608 }, { "epoch": 2.9377527133432646, "grad_norm": 15.40625, "learning_rate": 9.036775156338829e-06, "loss": 0.8436, "step": 27609 }, { "epoch": 2.937859118961481, "grad_norm": 12.3984375, "learning_rate": 9.03673962478681e-06, "loss": 0.8425, "step": 27610 }, { "epoch": 2.937965524579698, "grad_norm": 10.3046875, "learning_rate": 9.036704093234792e-06, "loss": 0.8929, "step": 27611 }, { "epoch": 2.9380719301979146, "grad_norm": 12.7734375, "learning_rate": 9.036668561682775e-06, "loss": 1.0093, "step": 27612 }, { "epoch": 2.938178335816131, "grad_norm": 14.96875, "learning_rate": 9.036633030130757e-06, "loss": 1.0581, "step": 27613 }, { "epoch": 2.9382847414343476, "grad_norm": 11.828125, "learning_rate": 9.036597498578739e-06, "loss": 0.8836, "step": 27614 }, { "epoch": 2.938391147052564, "grad_norm": 15.4609375, "learning_rate": 9.03656196702672e-06, "loss": 0.9969, "step": 27615 }, { "epoch": 2.938497552670781, "grad_norm": 11.6171875, "learning_rate": 9.036526435474702e-06, "loss": 0.8848, "step": 27616 }, { "epoch": 2.9386039582889976, "grad_norm": 11.0859375, "learning_rate": 9.036490903922685e-06, "loss": 0.9793, "step": 27617 }, { "epoch": 2.938710363907214, "grad_norm": 12.7578125, "learning_rate": 9.036455372370667e-06, "loss": 0.8863, "step": 27618 }, { "epoch": 2.938816769525431, "grad_norm": 17.453125, "learning_rate": 9.036419840818648e-06, "loss": 0.8034, "step": 27619 }, { "epoch": 2.9389231751436475, "grad_norm": 13.0390625, "learning_rate": 9.03638430926663e-06, "loss": 1.0714, "step": 27620 }, { "epoch": 2.939029580761864, "grad_norm": 12.5546875, "learning_rate": 9.036348777714612e-06, "loss": 0.9381, "step": 27621 }, { "epoch": 2.939135986380081, "grad_norm": 12.8671875, "learning_rate": 9.036313246162593e-06, "loss": 0.8364, "step": 27622 }, { "epoch": 2.9392423919982975, "grad_norm": 14.0, "learning_rate": 9.036277714610575e-06, "loss": 0.6872, "step": 27623 }, { "epoch": 2.939348797616514, "grad_norm": 14.1796875, "learning_rate": 9.036242183058556e-06, "loss": 0.9538, "step": 27624 }, { "epoch": 2.939455203234731, "grad_norm": 14.0859375, "learning_rate": 9.036206651506538e-06, "loss": 0.779, "step": 27625 }, { "epoch": 2.9395616088529475, "grad_norm": 11.34375, "learning_rate": 9.03617111995452e-06, "loss": 0.8206, "step": 27626 }, { "epoch": 2.939668014471164, "grad_norm": 15.1171875, "learning_rate": 9.036135588402503e-06, "loss": 0.9769, "step": 27627 }, { "epoch": 2.939774420089381, "grad_norm": 8.7734375, "learning_rate": 9.036100056850485e-06, "loss": 0.8608, "step": 27628 }, { "epoch": 2.9398808257075975, "grad_norm": 13.046875, "learning_rate": 9.036064525298466e-06, "loss": 1.0371, "step": 27629 }, { "epoch": 2.939987231325814, "grad_norm": 17.359375, "learning_rate": 9.036028993746448e-06, "loss": 0.9913, "step": 27630 }, { "epoch": 2.940093636944031, "grad_norm": 12.9921875, "learning_rate": 9.03599346219443e-06, "loss": 0.935, "step": 27631 }, { "epoch": 2.9402000425622474, "grad_norm": 11.9140625, "learning_rate": 9.035957930642411e-06, "loss": 0.8401, "step": 27632 }, { "epoch": 2.940306448180464, "grad_norm": 12.2734375, "learning_rate": 9.035922399090393e-06, "loss": 0.9302, "step": 27633 }, { "epoch": 2.9404128537986804, "grad_norm": 9.578125, "learning_rate": 9.035886867538374e-06, "loss": 0.8808, "step": 27634 }, { "epoch": 2.940519259416897, "grad_norm": 12.0, "learning_rate": 9.035851335986356e-06, "loss": 0.8331, "step": 27635 }, { "epoch": 2.940625665035114, "grad_norm": 12.609375, "learning_rate": 9.035815804434337e-06, "loss": 0.7488, "step": 27636 }, { "epoch": 2.9407320706533304, "grad_norm": 13.1484375, "learning_rate": 9.03578027288232e-06, "loss": 0.9588, "step": 27637 }, { "epoch": 2.940838476271547, "grad_norm": 12.0546875, "learning_rate": 9.035744741330302e-06, "loss": 0.9486, "step": 27638 }, { "epoch": 2.940944881889764, "grad_norm": 18.0625, "learning_rate": 9.035709209778284e-06, "loss": 1.0431, "step": 27639 }, { "epoch": 2.9410512875079804, "grad_norm": 10.0078125, "learning_rate": 9.035673678226266e-06, "loss": 0.9564, "step": 27640 }, { "epoch": 2.941157693126197, "grad_norm": 13.3671875, "learning_rate": 9.035638146674247e-06, "loss": 0.9281, "step": 27641 }, { "epoch": 2.941264098744414, "grad_norm": 15.078125, "learning_rate": 9.03560261512223e-06, "loss": 1.0906, "step": 27642 }, { "epoch": 2.9413705043626304, "grad_norm": 11.2734375, "learning_rate": 9.035567083570212e-06, "loss": 0.8363, "step": 27643 }, { "epoch": 2.941476909980847, "grad_norm": 10.90625, "learning_rate": 9.035531552018194e-06, "loss": 0.864, "step": 27644 }, { "epoch": 2.941583315599064, "grad_norm": 15.5390625, "learning_rate": 9.035496020466175e-06, "loss": 0.8285, "step": 27645 }, { "epoch": 2.9416897212172803, "grad_norm": 11.7109375, "learning_rate": 9.035460488914157e-06, "loss": 0.8238, "step": 27646 }, { "epoch": 2.941796126835497, "grad_norm": 17.15625, "learning_rate": 9.035424957362138e-06, "loss": 1.1006, "step": 27647 }, { "epoch": 2.941902532453714, "grad_norm": 13.96875, "learning_rate": 9.03538942581012e-06, "loss": 1.059, "step": 27648 }, { "epoch": 2.9420089380719303, "grad_norm": 21.5625, "learning_rate": 9.035353894258102e-06, "loss": 1.0344, "step": 27649 }, { "epoch": 2.942115343690147, "grad_norm": 15.0078125, "learning_rate": 9.035318362706083e-06, "loss": 1.0886, "step": 27650 }, { "epoch": 2.9422217493083633, "grad_norm": 14.34375, "learning_rate": 9.035282831154065e-06, "loss": 0.9105, "step": 27651 }, { "epoch": 2.9423281549265803, "grad_norm": 12.5859375, "learning_rate": 9.035247299602047e-06, "loss": 0.8084, "step": 27652 }, { "epoch": 2.942434560544797, "grad_norm": 15.1171875, "learning_rate": 9.03521176805003e-06, "loss": 0.8337, "step": 27653 }, { "epoch": 2.9425409661630133, "grad_norm": 10.421875, "learning_rate": 9.035176236498011e-06, "loss": 0.8968, "step": 27654 }, { "epoch": 2.94264737178123, "grad_norm": 14.46875, "learning_rate": 9.035140704945993e-06, "loss": 0.8305, "step": 27655 }, { "epoch": 2.9427537773994468, "grad_norm": 13.59375, "learning_rate": 9.035105173393975e-06, "loss": 0.9992, "step": 27656 }, { "epoch": 2.9428601830176633, "grad_norm": 13.796875, "learning_rate": 9.035069641841956e-06, "loss": 0.8864, "step": 27657 }, { "epoch": 2.94296658863588, "grad_norm": 15.5703125, "learning_rate": 9.035034110289938e-06, "loss": 0.8843, "step": 27658 }, { "epoch": 2.9430729942540967, "grad_norm": 10.25, "learning_rate": 9.03499857873792e-06, "loss": 0.9109, "step": 27659 }, { "epoch": 2.9431793998723133, "grad_norm": 12.59375, "learning_rate": 9.034963047185901e-06, "loss": 0.8704, "step": 27660 }, { "epoch": 2.9432858054905298, "grad_norm": 15.40625, "learning_rate": 9.034927515633883e-06, "loss": 0.9696, "step": 27661 }, { "epoch": 2.9433922111087467, "grad_norm": 13.234375, "learning_rate": 9.034891984081864e-06, "loss": 0.9094, "step": 27662 }, { "epoch": 2.9434986167269632, "grad_norm": 12.078125, "learning_rate": 9.034856452529848e-06, "loss": 1.0415, "step": 27663 }, { "epoch": 2.9436050223451797, "grad_norm": 12.953125, "learning_rate": 9.03482092097783e-06, "loss": 0.8855, "step": 27664 }, { "epoch": 2.9437114279633967, "grad_norm": 13.6640625, "learning_rate": 9.03478538942581e-06, "loss": 0.9035, "step": 27665 }, { "epoch": 2.943817833581613, "grad_norm": 12.796875, "learning_rate": 9.034749857873792e-06, "loss": 1.1552, "step": 27666 }, { "epoch": 2.9439242391998297, "grad_norm": 18.328125, "learning_rate": 9.034714326321774e-06, "loss": 1.2102, "step": 27667 }, { "epoch": 2.9440306448180467, "grad_norm": 13.4765625, "learning_rate": 9.034678794769757e-06, "loss": 0.8302, "step": 27668 }, { "epoch": 2.944137050436263, "grad_norm": 14.640625, "learning_rate": 9.034643263217739e-06, "loss": 0.8811, "step": 27669 }, { "epoch": 2.9442434560544797, "grad_norm": 24.203125, "learning_rate": 9.03460773166572e-06, "loss": 0.9009, "step": 27670 }, { "epoch": 2.944349861672696, "grad_norm": 13.59375, "learning_rate": 9.034572200113702e-06, "loss": 0.8594, "step": 27671 }, { "epoch": 2.9444562672909127, "grad_norm": 13.03125, "learning_rate": 9.034536668561684e-06, "loss": 1.0045, "step": 27672 }, { "epoch": 2.9445626729091297, "grad_norm": 15.53125, "learning_rate": 9.034501137009665e-06, "loss": 0.8773, "step": 27673 }, { "epoch": 2.944669078527346, "grad_norm": 20.765625, "learning_rate": 9.034465605457647e-06, "loss": 1.0329, "step": 27674 }, { "epoch": 2.9447754841455627, "grad_norm": 13.3984375, "learning_rate": 9.034430073905629e-06, "loss": 0.9701, "step": 27675 }, { "epoch": 2.9448818897637796, "grad_norm": 14.109375, "learning_rate": 9.03439454235361e-06, "loss": 1.1473, "step": 27676 }, { "epoch": 2.944988295381996, "grad_norm": 12.4765625, "learning_rate": 9.034359010801592e-06, "loss": 0.9483, "step": 27677 }, { "epoch": 2.9450947010002126, "grad_norm": 14.09375, "learning_rate": 9.034323479249575e-06, "loss": 1.0294, "step": 27678 }, { "epoch": 2.9452011066184296, "grad_norm": 11.8984375, "learning_rate": 9.034287947697557e-06, "loss": 0.9787, "step": 27679 }, { "epoch": 2.945307512236646, "grad_norm": 19.390625, "learning_rate": 9.034252416145538e-06, "loss": 1.3878, "step": 27680 }, { "epoch": 2.9454139178548626, "grad_norm": 20.25, "learning_rate": 9.03421688459352e-06, "loss": 0.9086, "step": 27681 }, { "epoch": 2.9455203234730796, "grad_norm": 16.140625, "learning_rate": 9.034181353041501e-06, "loss": 0.961, "step": 27682 }, { "epoch": 2.945626729091296, "grad_norm": 10.5703125, "learning_rate": 9.034145821489483e-06, "loss": 1.0107, "step": 27683 }, { "epoch": 2.9457331347095126, "grad_norm": 14.890625, "learning_rate": 9.034110289937465e-06, "loss": 0.8845, "step": 27684 }, { "epoch": 2.9458395403277295, "grad_norm": 10.6328125, "learning_rate": 9.034074758385446e-06, "loss": 0.8309, "step": 27685 }, { "epoch": 2.945945945945946, "grad_norm": 13.46875, "learning_rate": 9.034039226833428e-06, "loss": 0.9246, "step": 27686 }, { "epoch": 2.9460523515641626, "grad_norm": 10.4453125, "learning_rate": 9.03400369528141e-06, "loss": 0.8639, "step": 27687 }, { "epoch": 2.946158757182379, "grad_norm": 12.25, "learning_rate": 9.033968163729391e-06, "loss": 1.0597, "step": 27688 }, { "epoch": 2.946265162800596, "grad_norm": 11.5546875, "learning_rate": 9.033932632177374e-06, "loss": 1.0497, "step": 27689 }, { "epoch": 2.9463715684188125, "grad_norm": 11.0546875, "learning_rate": 9.033897100625356e-06, "loss": 0.9503, "step": 27690 }, { "epoch": 2.946477974037029, "grad_norm": 13.2421875, "learning_rate": 9.033861569073338e-06, "loss": 0.9879, "step": 27691 }, { "epoch": 2.9465843796552456, "grad_norm": 12.53125, "learning_rate": 9.03382603752132e-06, "loss": 0.8549, "step": 27692 }, { "epoch": 2.9466907852734625, "grad_norm": 13.8984375, "learning_rate": 9.033790505969301e-06, "loss": 1.0599, "step": 27693 }, { "epoch": 2.946797190891679, "grad_norm": 11.4453125, "learning_rate": 9.033754974417284e-06, "loss": 0.8721, "step": 27694 }, { "epoch": 2.9469035965098955, "grad_norm": 13.6796875, "learning_rate": 9.033719442865266e-06, "loss": 0.8021, "step": 27695 }, { "epoch": 2.9470100021281125, "grad_norm": 15.6953125, "learning_rate": 9.033683911313247e-06, "loss": 0.9228, "step": 27696 }, { "epoch": 2.947116407746329, "grad_norm": 11.8984375, "learning_rate": 9.033648379761229e-06, "loss": 0.954, "step": 27697 }, { "epoch": 2.9472228133645455, "grad_norm": 13.1171875, "learning_rate": 9.03361284820921e-06, "loss": 1.1268, "step": 27698 }, { "epoch": 2.9473292189827625, "grad_norm": 13.171875, "learning_rate": 9.033577316657192e-06, "loss": 0.9911, "step": 27699 }, { "epoch": 2.947435624600979, "grad_norm": 11.21875, "learning_rate": 9.033541785105174e-06, "loss": 0.6782, "step": 27700 }, { "epoch": 2.9475420302191955, "grad_norm": 16.3125, "learning_rate": 9.033506253553155e-06, "loss": 1.0101, "step": 27701 }, { "epoch": 2.9476484358374124, "grad_norm": 14.546875, "learning_rate": 9.033470722001137e-06, "loss": 1.116, "step": 27702 }, { "epoch": 2.947754841455629, "grad_norm": 11.3828125, "learning_rate": 9.033435190449119e-06, "loss": 0.9816, "step": 27703 }, { "epoch": 2.9478612470738454, "grad_norm": 12.84375, "learning_rate": 9.033399658897102e-06, "loss": 0.8758, "step": 27704 }, { "epoch": 2.9479676526920624, "grad_norm": 15.609375, "learning_rate": 9.033364127345084e-06, "loss": 0.9247, "step": 27705 }, { "epoch": 2.948074058310279, "grad_norm": 12.9140625, "learning_rate": 9.033328595793065e-06, "loss": 0.777, "step": 27706 }, { "epoch": 2.9481804639284954, "grad_norm": 11.4296875, "learning_rate": 9.033293064241047e-06, "loss": 0.9601, "step": 27707 }, { "epoch": 2.948286869546712, "grad_norm": 10.1484375, "learning_rate": 9.033257532689028e-06, "loss": 0.7604, "step": 27708 }, { "epoch": 2.9483932751649284, "grad_norm": 14.71875, "learning_rate": 9.033222001137012e-06, "loss": 0.9327, "step": 27709 }, { "epoch": 2.9484996807831454, "grad_norm": 14.2578125, "learning_rate": 9.033186469584993e-06, "loss": 1.1076, "step": 27710 }, { "epoch": 2.948606086401362, "grad_norm": 12.09375, "learning_rate": 9.033150938032975e-06, "loss": 0.919, "step": 27711 }, { "epoch": 2.9487124920195784, "grad_norm": 11.9765625, "learning_rate": 9.033115406480956e-06, "loss": 0.9005, "step": 27712 }, { "epoch": 2.9488188976377954, "grad_norm": 15.515625, "learning_rate": 9.033079874928936e-06, "loss": 0.9374, "step": 27713 }, { "epoch": 2.948925303256012, "grad_norm": 13.8515625, "learning_rate": 9.03304434337692e-06, "loss": 0.9239, "step": 27714 }, { "epoch": 2.9490317088742284, "grad_norm": 12.125, "learning_rate": 9.033008811824901e-06, "loss": 0.8716, "step": 27715 }, { "epoch": 2.9491381144924453, "grad_norm": 11.3515625, "learning_rate": 9.032973280272883e-06, "loss": 0.7154, "step": 27716 }, { "epoch": 2.949244520110662, "grad_norm": 12.203125, "learning_rate": 9.032937748720865e-06, "loss": 0.8868, "step": 27717 }, { "epoch": 2.9493509257288784, "grad_norm": 9.9609375, "learning_rate": 9.032902217168846e-06, "loss": 0.7864, "step": 27718 }, { "epoch": 2.9494573313470953, "grad_norm": 15.1484375, "learning_rate": 9.03286668561683e-06, "loss": 0.8905, "step": 27719 }, { "epoch": 2.949563736965312, "grad_norm": 16.96875, "learning_rate": 9.032831154064811e-06, "loss": 0.8457, "step": 27720 }, { "epoch": 2.9496701425835283, "grad_norm": 13.8359375, "learning_rate": 9.032795622512793e-06, "loss": 0.7449, "step": 27721 }, { "epoch": 2.9497765482017453, "grad_norm": 10.3984375, "learning_rate": 9.032760090960774e-06, "loss": 0.8651, "step": 27722 }, { "epoch": 2.949882953819962, "grad_norm": 10.390625, "learning_rate": 9.032724559408756e-06, "loss": 0.8257, "step": 27723 }, { "epoch": 2.9499893594381783, "grad_norm": 13.40625, "learning_rate": 9.032689027856737e-06, "loss": 0.9853, "step": 27724 }, { "epoch": 2.950095765056395, "grad_norm": 15.7734375, "learning_rate": 9.032653496304719e-06, "loss": 1.0933, "step": 27725 }, { "epoch": 2.9502021706746118, "grad_norm": 12.2578125, "learning_rate": 9.0326179647527e-06, "loss": 1.0345, "step": 27726 }, { "epoch": 2.9503085762928283, "grad_norm": 11.65625, "learning_rate": 9.032582433200682e-06, "loss": 0.9566, "step": 27727 }, { "epoch": 2.950414981911045, "grad_norm": 11.6796875, "learning_rate": 9.032546901648664e-06, "loss": 0.8286, "step": 27728 }, { "epoch": 2.9505213875292613, "grad_norm": 10.59375, "learning_rate": 9.032511370096645e-06, "loss": 0.7906, "step": 27729 }, { "epoch": 2.9506277931474783, "grad_norm": 10.4765625, "learning_rate": 9.032475838544629e-06, "loss": 0.7066, "step": 27730 }, { "epoch": 2.9507341987656948, "grad_norm": 10.859375, "learning_rate": 9.03244030699261e-06, "loss": 0.8951, "step": 27731 }, { "epoch": 2.9508406043839113, "grad_norm": 14.3984375, "learning_rate": 9.032404775440592e-06, "loss": 0.9904, "step": 27732 }, { "epoch": 2.9509470100021282, "grad_norm": 11.703125, "learning_rate": 9.032369243888574e-06, "loss": 0.9133, "step": 27733 }, { "epoch": 2.9510534156203447, "grad_norm": 10.5234375, "learning_rate": 9.032333712336555e-06, "loss": 0.893, "step": 27734 }, { "epoch": 2.9511598212385612, "grad_norm": 11.6953125, "learning_rate": 9.032298180784539e-06, "loss": 0.8275, "step": 27735 }, { "epoch": 2.951266226856778, "grad_norm": 18.640625, "learning_rate": 9.03226264923252e-06, "loss": 0.9867, "step": 27736 }, { "epoch": 2.9513726324749947, "grad_norm": 13.0390625, "learning_rate": 9.032227117680502e-06, "loss": 0.8481, "step": 27737 }, { "epoch": 2.951479038093211, "grad_norm": 18.078125, "learning_rate": 9.032191586128483e-06, "loss": 1.0381, "step": 27738 }, { "epoch": 2.951585443711428, "grad_norm": 12.734375, "learning_rate": 9.032156054576465e-06, "loss": 0.8353, "step": 27739 }, { "epoch": 2.9516918493296447, "grad_norm": 11.78125, "learning_rate": 9.032120523024447e-06, "loss": 0.7101, "step": 27740 }, { "epoch": 2.951798254947861, "grad_norm": 15.453125, "learning_rate": 9.032084991472428e-06, "loss": 0.8169, "step": 27741 }, { "epoch": 2.951904660566078, "grad_norm": 10.515625, "learning_rate": 9.03204945992041e-06, "loss": 0.849, "step": 27742 }, { "epoch": 2.9520110661842947, "grad_norm": 16.546875, "learning_rate": 9.032013928368391e-06, "loss": 0.8521, "step": 27743 }, { "epoch": 2.952117471802511, "grad_norm": 13.4609375, "learning_rate": 9.031978396816373e-06, "loss": 0.8265, "step": 27744 }, { "epoch": 2.9522238774207277, "grad_norm": 17.265625, "learning_rate": 9.031942865264356e-06, "loss": 0.9045, "step": 27745 }, { "epoch": 2.952330283038944, "grad_norm": 13.6328125, "learning_rate": 9.031907333712338e-06, "loss": 0.7868, "step": 27746 }, { "epoch": 2.952436688657161, "grad_norm": 15.203125, "learning_rate": 9.03187180216032e-06, "loss": 0.992, "step": 27747 }, { "epoch": 2.9525430942753776, "grad_norm": 11.9375, "learning_rate": 9.031836270608301e-06, "loss": 0.8903, "step": 27748 }, { "epoch": 2.952649499893594, "grad_norm": 14.0703125, "learning_rate": 9.031800739056283e-06, "loss": 0.8544, "step": 27749 }, { "epoch": 2.952755905511811, "grad_norm": 13.046875, "learning_rate": 9.031765207504264e-06, "loss": 0.9497, "step": 27750 }, { "epoch": 2.9528623111300276, "grad_norm": 17.234375, "learning_rate": 9.031729675952246e-06, "loss": 1.2821, "step": 27751 }, { "epoch": 2.952968716748244, "grad_norm": 23.625, "learning_rate": 9.031694144400228e-06, "loss": 0.8631, "step": 27752 }, { "epoch": 2.953075122366461, "grad_norm": 12.8828125, "learning_rate": 9.031658612848209e-06, "loss": 0.985, "step": 27753 }, { "epoch": 2.9531815279846776, "grad_norm": 17.796875, "learning_rate": 9.03162308129619e-06, "loss": 0.9166, "step": 27754 }, { "epoch": 2.953287933602894, "grad_norm": 11.7890625, "learning_rate": 9.031587549744174e-06, "loss": 0.9186, "step": 27755 }, { "epoch": 2.953394339221111, "grad_norm": 13.125, "learning_rate": 9.031552018192156e-06, "loss": 0.8381, "step": 27756 }, { "epoch": 2.9535007448393276, "grad_norm": 11.515625, "learning_rate": 9.031516486640137e-06, "loss": 1.0375, "step": 27757 }, { "epoch": 2.953607150457544, "grad_norm": 15.2265625, "learning_rate": 9.031480955088119e-06, "loss": 0.8294, "step": 27758 }, { "epoch": 2.953713556075761, "grad_norm": 15.3515625, "learning_rate": 9.0314454235361e-06, "loss": 0.8374, "step": 27759 }, { "epoch": 2.9538199616939775, "grad_norm": 12.203125, "learning_rate": 9.031409891984084e-06, "loss": 0.9355, "step": 27760 }, { "epoch": 2.953926367312194, "grad_norm": 14.71875, "learning_rate": 9.031374360432065e-06, "loss": 0.8115, "step": 27761 }, { "epoch": 2.954032772930411, "grad_norm": 17.421875, "learning_rate": 9.031338828880047e-06, "loss": 0.8857, "step": 27762 }, { "epoch": 2.9541391785486275, "grad_norm": 12.9140625, "learning_rate": 9.031303297328029e-06, "loss": 1.0767, "step": 27763 }, { "epoch": 2.954245584166844, "grad_norm": 13.71875, "learning_rate": 9.03126776577601e-06, "loss": 1.0089, "step": 27764 }, { "epoch": 2.9543519897850605, "grad_norm": 11.015625, "learning_rate": 9.031232234223992e-06, "loss": 0.8981, "step": 27765 }, { "epoch": 2.954458395403277, "grad_norm": 16.21875, "learning_rate": 9.031196702671973e-06, "loss": 1.0396, "step": 27766 }, { "epoch": 2.954564801021494, "grad_norm": 10.84375, "learning_rate": 9.031161171119955e-06, "loss": 0.7812, "step": 27767 }, { "epoch": 2.9546712066397105, "grad_norm": 9.625, "learning_rate": 9.031125639567937e-06, "loss": 0.8464, "step": 27768 }, { "epoch": 2.954777612257927, "grad_norm": 14.3828125, "learning_rate": 9.031090108015918e-06, "loss": 0.9178, "step": 27769 }, { "epoch": 2.954884017876144, "grad_norm": 12.703125, "learning_rate": 9.0310545764639e-06, "loss": 0.9516, "step": 27770 }, { "epoch": 2.9549904234943605, "grad_norm": 10.5546875, "learning_rate": 9.031019044911883e-06, "loss": 0.8591, "step": 27771 }, { "epoch": 2.955096829112577, "grad_norm": 12.5625, "learning_rate": 9.030983513359865e-06, "loss": 0.9095, "step": 27772 }, { "epoch": 2.955203234730794, "grad_norm": 18.765625, "learning_rate": 9.030947981807846e-06, "loss": 0.9985, "step": 27773 }, { "epoch": 2.9553096403490104, "grad_norm": 9.4765625, "learning_rate": 9.030912450255828e-06, "loss": 0.7111, "step": 27774 }, { "epoch": 2.955416045967227, "grad_norm": 10.4375, "learning_rate": 9.03087691870381e-06, "loss": 0.9211, "step": 27775 }, { "epoch": 2.955522451585444, "grad_norm": 14.2578125, "learning_rate": 9.030841387151791e-06, "loss": 0.9952, "step": 27776 }, { "epoch": 2.9556288572036604, "grad_norm": 14.1953125, "learning_rate": 9.030805855599773e-06, "loss": 0.7406, "step": 27777 }, { "epoch": 2.955735262821877, "grad_norm": 12.0859375, "learning_rate": 9.030770324047754e-06, "loss": 0.9963, "step": 27778 }, { "epoch": 2.955841668440094, "grad_norm": 14.3359375, "learning_rate": 9.030734792495736e-06, "loss": 0.7667, "step": 27779 }, { "epoch": 2.9559480740583104, "grad_norm": 9.8828125, "learning_rate": 9.030699260943718e-06, "loss": 0.8792, "step": 27780 }, { "epoch": 2.956054479676527, "grad_norm": 12.6875, "learning_rate": 9.030663729391701e-06, "loss": 0.9796, "step": 27781 }, { "epoch": 2.9561608852947434, "grad_norm": 13.71875, "learning_rate": 9.030628197839682e-06, "loss": 0.864, "step": 27782 }, { "epoch": 2.9562672909129604, "grad_norm": 12.0, "learning_rate": 9.030592666287664e-06, "loss": 0.8969, "step": 27783 }, { "epoch": 2.956373696531177, "grad_norm": 13.984375, "learning_rate": 9.030557134735646e-06, "loss": 0.9248, "step": 27784 }, { "epoch": 2.9564801021493934, "grad_norm": 14.0, "learning_rate": 9.030521603183627e-06, "loss": 0.8993, "step": 27785 }, { "epoch": 2.95658650776761, "grad_norm": 9.9296875, "learning_rate": 9.03048607163161e-06, "loss": 0.7753, "step": 27786 }, { "epoch": 2.956692913385827, "grad_norm": 15.4609375, "learning_rate": 9.030450540079592e-06, "loss": 0.8431, "step": 27787 }, { "epoch": 2.9567993190040434, "grad_norm": 15.1015625, "learning_rate": 9.030415008527574e-06, "loss": 1.0044, "step": 27788 }, { "epoch": 2.95690572462226, "grad_norm": 15.59375, "learning_rate": 9.030379476975555e-06, "loss": 0.9853, "step": 27789 }, { "epoch": 2.957012130240477, "grad_norm": 10.296875, "learning_rate": 9.030343945423537e-06, "loss": 0.7477, "step": 27790 }, { "epoch": 2.9571185358586933, "grad_norm": 13.859375, "learning_rate": 9.030308413871519e-06, "loss": 0.8977, "step": 27791 }, { "epoch": 2.95722494147691, "grad_norm": 20.171875, "learning_rate": 9.0302728823195e-06, "loss": 0.9117, "step": 27792 }, { "epoch": 2.957331347095127, "grad_norm": 16.734375, "learning_rate": 9.030237350767482e-06, "loss": 1.0237, "step": 27793 }, { "epoch": 2.9574377527133433, "grad_norm": 12.8125, "learning_rate": 9.030201819215463e-06, "loss": 0.8436, "step": 27794 }, { "epoch": 2.95754415833156, "grad_norm": 10.71875, "learning_rate": 9.030166287663445e-06, "loss": 0.8936, "step": 27795 }, { "epoch": 2.9576505639497768, "grad_norm": 13.453125, "learning_rate": 9.030130756111428e-06, "loss": 0.951, "step": 27796 }, { "epoch": 2.9577569695679933, "grad_norm": 14.34375, "learning_rate": 9.03009522455941e-06, "loss": 1.0754, "step": 27797 }, { "epoch": 2.95786337518621, "grad_norm": 13.9375, "learning_rate": 9.030059693007392e-06, "loss": 0.9893, "step": 27798 }, { "epoch": 2.9579697808044267, "grad_norm": 11.5390625, "learning_rate": 9.030024161455373e-06, "loss": 1.0386, "step": 27799 }, { "epoch": 2.9580761864226433, "grad_norm": 18.796875, "learning_rate": 9.029988629903355e-06, "loss": 0.8983, "step": 27800 }, { "epoch": 2.9581825920408598, "grad_norm": 12.1796875, "learning_rate": 9.029953098351338e-06, "loss": 0.684, "step": 27801 }, { "epoch": 2.9582889976590763, "grad_norm": 13.8046875, "learning_rate": 9.029917566799318e-06, "loss": 0.9248, "step": 27802 }, { "epoch": 2.958395403277293, "grad_norm": 14.1484375, "learning_rate": 9.0298820352473e-06, "loss": 0.9748, "step": 27803 }, { "epoch": 2.9585018088955097, "grad_norm": 14.0390625, "learning_rate": 9.029846503695281e-06, "loss": 0.7508, "step": 27804 }, { "epoch": 2.9586082145137262, "grad_norm": 13.6171875, "learning_rate": 9.029810972143263e-06, "loss": 0.9848, "step": 27805 }, { "epoch": 2.9587146201319428, "grad_norm": 11.3671875, "learning_rate": 9.029775440591244e-06, "loss": 0.9724, "step": 27806 }, { "epoch": 2.9588210257501597, "grad_norm": 15.78125, "learning_rate": 9.029739909039228e-06, "loss": 0.8678, "step": 27807 }, { "epoch": 2.958927431368376, "grad_norm": 14.3671875, "learning_rate": 9.02970437748721e-06, "loss": 0.8456, "step": 27808 }, { "epoch": 2.9590338369865927, "grad_norm": 14.5234375, "learning_rate": 9.029668845935191e-06, "loss": 0.8761, "step": 27809 }, { "epoch": 2.9591402426048097, "grad_norm": 14.703125, "learning_rate": 9.029633314383173e-06, "loss": 0.8648, "step": 27810 }, { "epoch": 2.959246648223026, "grad_norm": 13.0078125, "learning_rate": 9.029597782831154e-06, "loss": 0.9804, "step": 27811 }, { "epoch": 2.9593530538412427, "grad_norm": 12.3203125, "learning_rate": 9.029562251279137e-06, "loss": 0.9299, "step": 27812 }, { "epoch": 2.9594594594594597, "grad_norm": 12.4765625, "learning_rate": 9.029526719727119e-06, "loss": 0.9672, "step": 27813 }, { "epoch": 2.959565865077676, "grad_norm": 11.890625, "learning_rate": 9.0294911881751e-06, "loss": 0.8066, "step": 27814 }, { "epoch": 2.9596722706958927, "grad_norm": 10.7890625, "learning_rate": 9.029455656623082e-06, "loss": 0.8549, "step": 27815 }, { "epoch": 2.9597786763141096, "grad_norm": 13.5078125, "learning_rate": 9.029420125071064e-06, "loss": 0.991, "step": 27816 }, { "epoch": 2.959885081932326, "grad_norm": 11.6953125, "learning_rate": 9.029384593519046e-06, "loss": 0.78, "step": 27817 }, { "epoch": 2.9599914875505426, "grad_norm": 14.21875, "learning_rate": 9.029349061967027e-06, "loss": 0.8379, "step": 27818 }, { "epoch": 2.960097893168759, "grad_norm": 12.4921875, "learning_rate": 9.029313530415009e-06, "loss": 1.0205, "step": 27819 }, { "epoch": 2.960204298786976, "grad_norm": 12.7265625, "learning_rate": 9.02927799886299e-06, "loss": 1.0623, "step": 27820 }, { "epoch": 2.9603107044051926, "grad_norm": 9.703125, "learning_rate": 9.029242467310972e-06, "loss": 0.7681, "step": 27821 }, { "epoch": 2.960417110023409, "grad_norm": 12.875, "learning_rate": 9.029206935758955e-06, "loss": 0.956, "step": 27822 }, { "epoch": 2.9605235156416256, "grad_norm": 12.3515625, "learning_rate": 9.029171404206937e-06, "loss": 0.9712, "step": 27823 }, { "epoch": 2.9606299212598426, "grad_norm": 12.0859375, "learning_rate": 9.029135872654918e-06, "loss": 0.9599, "step": 27824 }, { "epoch": 2.960736326878059, "grad_norm": 14.625, "learning_rate": 9.0291003411029e-06, "loss": 1.1408, "step": 27825 }, { "epoch": 2.9608427324962756, "grad_norm": 9.5234375, "learning_rate": 9.029064809550882e-06, "loss": 0.7135, "step": 27826 }, { "epoch": 2.9609491381144926, "grad_norm": 10.5703125, "learning_rate": 9.029029277998865e-06, "loss": 1.017, "step": 27827 }, { "epoch": 2.961055543732709, "grad_norm": 17.5, "learning_rate": 9.028993746446847e-06, "loss": 1.2285, "step": 27828 }, { "epoch": 2.9611619493509256, "grad_norm": 12.734375, "learning_rate": 9.028958214894828e-06, "loss": 0.9541, "step": 27829 }, { "epoch": 2.9612683549691425, "grad_norm": 8.21875, "learning_rate": 9.02892268334281e-06, "loss": 0.8137, "step": 27830 }, { "epoch": 2.961374760587359, "grad_norm": 15.9375, "learning_rate": 9.028887151790791e-06, "loss": 1.0139, "step": 27831 }, { "epoch": 2.9614811662055756, "grad_norm": 17.28125, "learning_rate": 9.028851620238773e-06, "loss": 1.0049, "step": 27832 }, { "epoch": 2.9615875718237925, "grad_norm": 11.203125, "learning_rate": 9.028816088686755e-06, "loss": 0.9474, "step": 27833 }, { "epoch": 2.961693977442009, "grad_norm": 10.953125, "learning_rate": 9.028780557134736e-06, "loss": 0.8672, "step": 27834 }, { "epoch": 2.9618003830602255, "grad_norm": 13.125, "learning_rate": 9.028745025582718e-06, "loss": 0.8432, "step": 27835 }, { "epoch": 2.9619067886784425, "grad_norm": 13.9609375, "learning_rate": 9.0287094940307e-06, "loss": 0.8589, "step": 27836 }, { "epoch": 2.962013194296659, "grad_norm": 19.46875, "learning_rate": 9.028673962478683e-06, "loss": 1.0683, "step": 27837 }, { "epoch": 2.9621195999148755, "grad_norm": 11.1640625, "learning_rate": 9.028638430926664e-06, "loss": 0.8877, "step": 27838 }, { "epoch": 2.962226005533092, "grad_norm": 12.1640625, "learning_rate": 9.028602899374646e-06, "loss": 0.9978, "step": 27839 }, { "epoch": 2.9623324111513085, "grad_norm": 13.421875, "learning_rate": 9.028567367822628e-06, "loss": 0.9553, "step": 27840 }, { "epoch": 2.9624388167695255, "grad_norm": 14.125, "learning_rate": 9.028531836270609e-06, "loss": 0.8643, "step": 27841 }, { "epoch": 2.962545222387742, "grad_norm": 14.4453125, "learning_rate": 9.02849630471859e-06, "loss": 1.0356, "step": 27842 }, { "epoch": 2.9626516280059585, "grad_norm": 10.921875, "learning_rate": 9.028460773166572e-06, "loss": 0.8335, "step": 27843 }, { "epoch": 2.9627580336241754, "grad_norm": 14.4296875, "learning_rate": 9.028425241614554e-06, "loss": 0.9329, "step": 27844 }, { "epoch": 2.962864439242392, "grad_norm": 19.328125, "learning_rate": 9.028389710062536e-06, "loss": 1.2215, "step": 27845 }, { "epoch": 2.9629708448606085, "grad_norm": 19.796875, "learning_rate": 9.028354178510517e-06, "loss": 0.8591, "step": 27846 }, { "epoch": 2.9630772504788254, "grad_norm": 12.7421875, "learning_rate": 9.028318646958499e-06, "loss": 0.8432, "step": 27847 }, { "epoch": 2.963183656097042, "grad_norm": 16.1875, "learning_rate": 9.028283115406482e-06, "loss": 1.0045, "step": 27848 }, { "epoch": 2.9632900617152584, "grad_norm": 12.671875, "learning_rate": 9.028247583854464e-06, "loss": 1.0459, "step": 27849 }, { "epoch": 2.9633964673334754, "grad_norm": 13.2421875, "learning_rate": 9.028212052302445e-06, "loss": 0.8693, "step": 27850 }, { "epoch": 2.963502872951692, "grad_norm": 10.25, "learning_rate": 9.028176520750427e-06, "loss": 0.8046, "step": 27851 }, { "epoch": 2.9636092785699084, "grad_norm": 11.578125, "learning_rate": 9.028140989198409e-06, "loss": 0.9201, "step": 27852 }, { "epoch": 2.9637156841881254, "grad_norm": 15.7578125, "learning_rate": 9.028105457646392e-06, "loss": 1.3333, "step": 27853 }, { "epoch": 2.963822089806342, "grad_norm": 11.546875, "learning_rate": 9.028069926094373e-06, "loss": 1.0095, "step": 27854 }, { "epoch": 2.9639284954245584, "grad_norm": 13.671875, "learning_rate": 9.028034394542355e-06, "loss": 0.7935, "step": 27855 }, { "epoch": 2.964034901042775, "grad_norm": 11.7890625, "learning_rate": 9.027998862990337e-06, "loss": 0.9742, "step": 27856 }, { "epoch": 2.964141306660992, "grad_norm": 11.125, "learning_rate": 9.027963331438318e-06, "loss": 0.8999, "step": 27857 }, { "epoch": 2.9642477122792084, "grad_norm": 15.0703125, "learning_rate": 9.0279277998863e-06, "loss": 1.0913, "step": 27858 }, { "epoch": 2.964354117897425, "grad_norm": 14.578125, "learning_rate": 9.027892268334281e-06, "loss": 0.8975, "step": 27859 }, { "epoch": 2.9644605235156414, "grad_norm": 12.2890625, "learning_rate": 9.027856736782263e-06, "loss": 0.8896, "step": 27860 }, { "epoch": 2.9645669291338583, "grad_norm": 11.1015625, "learning_rate": 9.027821205230245e-06, "loss": 0.7792, "step": 27861 }, { "epoch": 2.964673334752075, "grad_norm": 10.1640625, "learning_rate": 9.027785673678226e-06, "loss": 0.7702, "step": 27862 }, { "epoch": 2.9647797403702913, "grad_norm": 12.609375, "learning_rate": 9.02775014212621e-06, "loss": 1.0198, "step": 27863 }, { "epoch": 2.9648861459885083, "grad_norm": 18.140625, "learning_rate": 9.027714610574191e-06, "loss": 0.8941, "step": 27864 }, { "epoch": 2.964992551606725, "grad_norm": 11.09375, "learning_rate": 9.027679079022173e-06, "loss": 0.7561, "step": 27865 }, { "epoch": 2.9650989572249413, "grad_norm": 13.171875, "learning_rate": 9.027643547470154e-06, "loss": 0.9362, "step": 27866 }, { "epoch": 2.9652053628431583, "grad_norm": 11.0234375, "learning_rate": 9.027608015918136e-06, "loss": 1.0293, "step": 27867 }, { "epoch": 2.965311768461375, "grad_norm": 17.0625, "learning_rate": 9.027572484366118e-06, "loss": 0.8458, "step": 27868 }, { "epoch": 2.9654181740795913, "grad_norm": 15.2421875, "learning_rate": 9.0275369528141e-06, "loss": 0.7974, "step": 27869 }, { "epoch": 2.9655245796978082, "grad_norm": 14.265625, "learning_rate": 9.02750142126208e-06, "loss": 1.0156, "step": 27870 }, { "epoch": 2.9656309853160248, "grad_norm": 9.890625, "learning_rate": 9.027465889710062e-06, "loss": 0.9262, "step": 27871 }, { "epoch": 2.9657373909342413, "grad_norm": 14.890625, "learning_rate": 9.027430358158044e-06, "loss": 1.0719, "step": 27872 }, { "epoch": 2.9658437965524582, "grad_norm": 15.78125, "learning_rate": 9.027394826606027e-06, "loss": 0.9638, "step": 27873 }, { "epoch": 2.9659502021706747, "grad_norm": 12.8671875, "learning_rate": 9.027359295054009e-06, "loss": 0.8235, "step": 27874 }, { "epoch": 2.9660566077888912, "grad_norm": 13.65625, "learning_rate": 9.02732376350199e-06, "loss": 1.1013, "step": 27875 }, { "epoch": 2.9661630134071078, "grad_norm": 13.1171875, "learning_rate": 9.027288231949972e-06, "loss": 0.8799, "step": 27876 }, { "epoch": 2.9662694190253243, "grad_norm": 20.390625, "learning_rate": 9.027252700397954e-06, "loss": 1.1585, "step": 27877 }, { "epoch": 2.966375824643541, "grad_norm": 9.8671875, "learning_rate": 9.027217168845937e-06, "loss": 0.7969, "step": 27878 }, { "epoch": 2.9664822302617577, "grad_norm": 16.765625, "learning_rate": 9.027181637293919e-06, "loss": 0.6788, "step": 27879 }, { "epoch": 2.9665886358799742, "grad_norm": 13.828125, "learning_rate": 9.0271461057419e-06, "loss": 0.7921, "step": 27880 }, { "epoch": 2.966695041498191, "grad_norm": 14.34375, "learning_rate": 9.027110574189882e-06, "loss": 0.8236, "step": 27881 }, { "epoch": 2.9668014471164077, "grad_norm": 13.625, "learning_rate": 9.027075042637863e-06, "loss": 0.9403, "step": 27882 }, { "epoch": 2.966907852734624, "grad_norm": 12.671875, "learning_rate": 9.027039511085845e-06, "loss": 0.8202, "step": 27883 }, { "epoch": 2.967014258352841, "grad_norm": 12.4140625, "learning_rate": 9.027003979533827e-06, "loss": 0.7282, "step": 27884 }, { "epoch": 2.9671206639710577, "grad_norm": 12.21875, "learning_rate": 9.026968447981808e-06, "loss": 1.0357, "step": 27885 }, { "epoch": 2.967227069589274, "grad_norm": 13.046875, "learning_rate": 9.02693291642979e-06, "loss": 0.8709, "step": 27886 }, { "epoch": 2.967333475207491, "grad_norm": 12.7265625, "learning_rate": 9.026897384877772e-06, "loss": 0.9329, "step": 27887 }, { "epoch": 2.9674398808257076, "grad_norm": 11.859375, "learning_rate": 9.026861853325753e-06, "loss": 0.9815, "step": 27888 }, { "epoch": 2.967546286443924, "grad_norm": 14.7578125, "learning_rate": 9.026826321773736e-06, "loss": 1.0686, "step": 27889 }, { "epoch": 2.967652692062141, "grad_norm": 14.3125, "learning_rate": 9.026790790221718e-06, "loss": 1.0664, "step": 27890 }, { "epoch": 2.9677590976803576, "grad_norm": 17.578125, "learning_rate": 9.0267552586697e-06, "loss": 0.8602, "step": 27891 }, { "epoch": 2.967865503298574, "grad_norm": 14.375, "learning_rate": 9.026719727117681e-06, "loss": 0.8591, "step": 27892 }, { "epoch": 2.967971908916791, "grad_norm": 12.3359375, "learning_rate": 9.026684195565663e-06, "loss": 1.0899, "step": 27893 }, { "epoch": 2.9680783145350076, "grad_norm": 12.875, "learning_rate": 9.026648664013644e-06, "loss": 1.066, "step": 27894 }, { "epoch": 2.968184720153224, "grad_norm": 13.015625, "learning_rate": 9.026613132461626e-06, "loss": 0.9643, "step": 27895 }, { "epoch": 2.9682911257714406, "grad_norm": 11.984375, "learning_rate": 9.026577600909608e-06, "loss": 0.9712, "step": 27896 }, { "epoch": 2.968397531389657, "grad_norm": 13.5390625, "learning_rate": 9.02654206935759e-06, "loss": 0.9377, "step": 27897 }, { "epoch": 2.968503937007874, "grad_norm": 13.03125, "learning_rate": 9.026506537805571e-06, "loss": 1.0264, "step": 27898 }, { "epoch": 2.9686103426260906, "grad_norm": 15.3125, "learning_rate": 9.026471006253554e-06, "loss": 0.9712, "step": 27899 }, { "epoch": 2.968716748244307, "grad_norm": 12.71875, "learning_rate": 9.026435474701536e-06, "loss": 0.9674, "step": 27900 }, { "epoch": 2.968823153862524, "grad_norm": 10.921875, "learning_rate": 9.026399943149517e-06, "loss": 0.9384, "step": 27901 }, { "epoch": 2.9689295594807406, "grad_norm": 16.625, "learning_rate": 9.026364411597499e-06, "loss": 0.9919, "step": 27902 }, { "epoch": 2.969035965098957, "grad_norm": 11.3671875, "learning_rate": 9.02632888004548e-06, "loss": 0.9676, "step": 27903 }, { "epoch": 2.969142370717174, "grad_norm": 14.75, "learning_rate": 9.026293348493464e-06, "loss": 1.0193, "step": 27904 }, { "epoch": 2.9692487763353905, "grad_norm": 15.4375, "learning_rate": 9.026257816941446e-06, "loss": 0.8747, "step": 27905 }, { "epoch": 2.969355181953607, "grad_norm": 15.5078125, "learning_rate": 9.026222285389427e-06, "loss": 0.9014, "step": 27906 }, { "epoch": 2.969461587571824, "grad_norm": 17.28125, "learning_rate": 9.026186753837409e-06, "loss": 0.8937, "step": 27907 }, { "epoch": 2.9695679931900405, "grad_norm": 12.71875, "learning_rate": 9.02615122228539e-06, "loss": 0.9754, "step": 27908 }, { "epoch": 2.969674398808257, "grad_norm": 12.703125, "learning_rate": 9.026115690733372e-06, "loss": 1.0249, "step": 27909 }, { "epoch": 2.969780804426474, "grad_norm": 13.5234375, "learning_rate": 9.026080159181354e-06, "loss": 0.8063, "step": 27910 }, { "epoch": 2.9698872100446905, "grad_norm": 12.1796875, "learning_rate": 9.026044627629335e-06, "loss": 1.0394, "step": 27911 }, { "epoch": 2.969993615662907, "grad_norm": 24.8125, "learning_rate": 9.026009096077317e-06, "loss": 1.0519, "step": 27912 }, { "epoch": 2.9701000212811235, "grad_norm": 12.359375, "learning_rate": 9.025973564525298e-06, "loss": 0.9855, "step": 27913 }, { "epoch": 2.9702064268993404, "grad_norm": 14.59375, "learning_rate": 9.025938032973282e-06, "loss": 1.0179, "step": 27914 }, { "epoch": 2.970312832517557, "grad_norm": 15.109375, "learning_rate": 9.025902501421263e-06, "loss": 0.8423, "step": 27915 }, { "epoch": 2.9704192381357735, "grad_norm": 10.5859375, "learning_rate": 9.025866969869245e-06, "loss": 0.7569, "step": 27916 }, { "epoch": 2.97052564375399, "grad_norm": 12.09375, "learning_rate": 9.025831438317226e-06, "loss": 0.8182, "step": 27917 }, { "epoch": 2.970632049372207, "grad_norm": 7.9609375, "learning_rate": 9.025795906765208e-06, "loss": 0.743, "step": 27918 }, { "epoch": 2.9707384549904234, "grad_norm": 12.265625, "learning_rate": 9.025760375213191e-06, "loss": 0.8127, "step": 27919 }, { "epoch": 2.97084486060864, "grad_norm": 14.7109375, "learning_rate": 9.025724843661173e-06, "loss": 1.0848, "step": 27920 }, { "epoch": 2.970951266226857, "grad_norm": 16.0, "learning_rate": 9.025689312109155e-06, "loss": 1.0602, "step": 27921 }, { "epoch": 2.9710576718450734, "grad_norm": 12.3671875, "learning_rate": 9.025653780557135e-06, "loss": 0.9347, "step": 27922 }, { "epoch": 2.97116407746329, "grad_norm": 18.078125, "learning_rate": 9.025618249005116e-06, "loss": 1.0326, "step": 27923 }, { "epoch": 2.971270483081507, "grad_norm": 13.578125, "learning_rate": 9.025582717453098e-06, "loss": 0.9527, "step": 27924 }, { "epoch": 2.9713768886997234, "grad_norm": 14.2265625, "learning_rate": 9.025547185901081e-06, "loss": 0.7961, "step": 27925 }, { "epoch": 2.97148329431794, "grad_norm": 14.390625, "learning_rate": 9.025511654349063e-06, "loss": 0.8973, "step": 27926 }, { "epoch": 2.971589699936157, "grad_norm": 15.1640625, "learning_rate": 9.025476122797044e-06, "loss": 0.7229, "step": 27927 }, { "epoch": 2.9716961055543734, "grad_norm": 12.5390625, "learning_rate": 9.025440591245026e-06, "loss": 1.0913, "step": 27928 }, { "epoch": 2.97180251117259, "grad_norm": 14.3984375, "learning_rate": 9.025405059693007e-06, "loss": 0.9256, "step": 27929 }, { "epoch": 2.971908916790807, "grad_norm": 13.4921875, "learning_rate": 9.02536952814099e-06, "loss": 0.8968, "step": 27930 }, { "epoch": 2.9720153224090233, "grad_norm": 13.3359375, "learning_rate": 9.025333996588972e-06, "loss": 1.0122, "step": 27931 }, { "epoch": 2.97212172802724, "grad_norm": 11.1484375, "learning_rate": 9.025298465036954e-06, "loss": 0.822, "step": 27932 }, { "epoch": 2.9722281336454563, "grad_norm": 12.9453125, "learning_rate": 9.025262933484936e-06, "loss": 0.9107, "step": 27933 }, { "epoch": 2.972334539263673, "grad_norm": 11.2109375, "learning_rate": 9.025227401932917e-06, "loss": 0.9189, "step": 27934 }, { "epoch": 2.97244094488189, "grad_norm": 10.953125, "learning_rate": 9.025191870380899e-06, "loss": 0.8767, "step": 27935 }, { "epoch": 2.9725473505001063, "grad_norm": 20.09375, "learning_rate": 9.02515633882888e-06, "loss": 0.7714, "step": 27936 }, { "epoch": 2.972653756118323, "grad_norm": 16.6875, "learning_rate": 9.025120807276862e-06, "loss": 1.0438, "step": 27937 }, { "epoch": 2.97276016173654, "grad_norm": 11.265625, "learning_rate": 9.025085275724844e-06, "loss": 0.8899, "step": 27938 }, { "epoch": 2.9728665673547563, "grad_norm": 16.40625, "learning_rate": 9.025049744172825e-06, "loss": 0.866, "step": 27939 }, { "epoch": 2.972972972972973, "grad_norm": 12.8046875, "learning_rate": 9.025014212620809e-06, "loss": 0.9219, "step": 27940 }, { "epoch": 2.9730793785911898, "grad_norm": 12.3125, "learning_rate": 9.02497868106879e-06, "loss": 1.0117, "step": 27941 }, { "epoch": 2.9731857842094063, "grad_norm": 13.203125, "learning_rate": 9.024943149516772e-06, "loss": 0.9483, "step": 27942 }, { "epoch": 2.9732921898276228, "grad_norm": 12.8203125, "learning_rate": 9.024907617964753e-06, "loss": 1.006, "step": 27943 }, { "epoch": 2.9733985954458397, "grad_norm": 19.53125, "learning_rate": 9.024872086412735e-06, "loss": 0.743, "step": 27944 }, { "epoch": 2.9735050010640562, "grad_norm": 12.765625, "learning_rate": 9.024836554860718e-06, "loss": 0.9541, "step": 27945 }, { "epoch": 2.9736114066822728, "grad_norm": 13.2109375, "learning_rate": 9.0248010233087e-06, "loss": 0.941, "step": 27946 }, { "epoch": 2.9737178123004897, "grad_norm": 12.6953125, "learning_rate": 9.024765491756681e-06, "loss": 0.9094, "step": 27947 }, { "epoch": 2.973824217918706, "grad_norm": 10.4140625, "learning_rate": 9.024729960204663e-06, "loss": 0.9808, "step": 27948 }, { "epoch": 2.9739306235369227, "grad_norm": 17.890625, "learning_rate": 9.024694428652645e-06, "loss": 0.8423, "step": 27949 }, { "epoch": 2.9740370291551392, "grad_norm": 15.859375, "learning_rate": 9.024658897100626e-06, "loss": 0.9045, "step": 27950 }, { "epoch": 2.974143434773356, "grad_norm": 12.625, "learning_rate": 9.024623365548608e-06, "loss": 1.1185, "step": 27951 }, { "epoch": 2.9742498403915727, "grad_norm": 15.484375, "learning_rate": 9.02458783399659e-06, "loss": 0.8589, "step": 27952 }, { "epoch": 2.974356246009789, "grad_norm": 10.9140625, "learning_rate": 9.024552302444571e-06, "loss": 0.911, "step": 27953 }, { "epoch": 2.9744626516280057, "grad_norm": 11.2734375, "learning_rate": 9.024516770892553e-06, "loss": 0.9639, "step": 27954 }, { "epoch": 2.9745690572462227, "grad_norm": 20.21875, "learning_rate": 9.024481239340536e-06, "loss": 0.8429, "step": 27955 }, { "epoch": 2.974675462864439, "grad_norm": 11.953125, "learning_rate": 9.024445707788518e-06, "loss": 0.7411, "step": 27956 }, { "epoch": 2.9747818684826557, "grad_norm": 14.4453125, "learning_rate": 9.0244101762365e-06, "loss": 1.077, "step": 27957 }, { "epoch": 2.9748882741008726, "grad_norm": 12.4140625, "learning_rate": 9.02437464468448e-06, "loss": 0.8967, "step": 27958 }, { "epoch": 2.974994679719089, "grad_norm": 11.0546875, "learning_rate": 9.024339113132462e-06, "loss": 0.9927, "step": 27959 }, { "epoch": 2.9751010853373057, "grad_norm": 13.8046875, "learning_rate": 9.024303581580444e-06, "loss": 0.8951, "step": 27960 }, { "epoch": 2.9752074909555226, "grad_norm": 10.8125, "learning_rate": 9.024268050028426e-06, "loss": 0.816, "step": 27961 }, { "epoch": 2.975313896573739, "grad_norm": 11.265625, "learning_rate": 9.024232518476407e-06, "loss": 0.9265, "step": 27962 }, { "epoch": 2.9754203021919556, "grad_norm": 16.859375, "learning_rate": 9.024196986924389e-06, "loss": 0.9701, "step": 27963 }, { "epoch": 2.9755267078101726, "grad_norm": 14.65625, "learning_rate": 9.02416145537237e-06, "loss": 0.8635, "step": 27964 }, { "epoch": 2.975633113428389, "grad_norm": 15.609375, "learning_rate": 9.024125923820352e-06, "loss": 1.0141, "step": 27965 }, { "epoch": 2.9757395190466056, "grad_norm": 9.96875, "learning_rate": 9.024090392268335e-06, "loss": 0.8233, "step": 27966 }, { "epoch": 2.9758459246648226, "grad_norm": 14.1953125, "learning_rate": 9.024054860716317e-06, "loss": 0.9099, "step": 27967 }, { "epoch": 2.975952330283039, "grad_norm": 13.140625, "learning_rate": 9.024019329164299e-06, "loss": 0.9918, "step": 27968 }, { "epoch": 2.9760587359012556, "grad_norm": 17.0, "learning_rate": 9.02398379761228e-06, "loss": 0.9801, "step": 27969 }, { "epoch": 2.976165141519472, "grad_norm": 10.515625, "learning_rate": 9.023948266060262e-06, "loss": 0.9324, "step": 27970 }, { "epoch": 2.9762715471376886, "grad_norm": 12.5390625, "learning_rate": 9.023912734508245e-06, "loss": 0.922, "step": 27971 }, { "epoch": 2.9763779527559056, "grad_norm": 15.1171875, "learning_rate": 9.023877202956227e-06, "loss": 1.2447, "step": 27972 }, { "epoch": 2.976484358374122, "grad_norm": 13.1484375, "learning_rate": 9.023841671404208e-06, "loss": 0.79, "step": 27973 }, { "epoch": 2.9765907639923386, "grad_norm": 13.203125, "learning_rate": 9.02380613985219e-06, "loss": 0.8686, "step": 27974 }, { "epoch": 2.9766971696105555, "grad_norm": 16.109375, "learning_rate": 9.023770608300172e-06, "loss": 1.0999, "step": 27975 }, { "epoch": 2.976803575228772, "grad_norm": 18.171875, "learning_rate": 9.023735076748153e-06, "loss": 1.0981, "step": 27976 }, { "epoch": 2.9769099808469885, "grad_norm": 22.0, "learning_rate": 9.023699545196135e-06, "loss": 1.0142, "step": 27977 }, { "epoch": 2.9770163864652055, "grad_norm": 20.078125, "learning_rate": 9.023664013644116e-06, "loss": 0.8537, "step": 27978 }, { "epoch": 2.977122792083422, "grad_norm": 11.515625, "learning_rate": 9.023628482092098e-06, "loss": 1.0049, "step": 27979 }, { "epoch": 2.9772291977016385, "grad_norm": 17.0625, "learning_rate": 9.02359295054008e-06, "loss": 0.6838, "step": 27980 }, { "epoch": 2.9773356033198555, "grad_norm": 13.1875, "learning_rate": 9.023557418988063e-06, "loss": 0.8426, "step": 27981 }, { "epoch": 2.977442008938072, "grad_norm": 15.75, "learning_rate": 9.023521887436044e-06, "loss": 0.7741, "step": 27982 }, { "epoch": 2.9775484145562885, "grad_norm": 13.640625, "learning_rate": 9.023486355884026e-06, "loss": 0.9494, "step": 27983 }, { "epoch": 2.9776548201745054, "grad_norm": 13.609375, "learning_rate": 9.023450824332008e-06, "loss": 0.9517, "step": 27984 }, { "epoch": 2.977761225792722, "grad_norm": 11.796875, "learning_rate": 9.02341529277999e-06, "loss": 0.8368, "step": 27985 }, { "epoch": 2.9778676314109385, "grad_norm": 12.609375, "learning_rate": 9.023379761227971e-06, "loss": 0.922, "step": 27986 }, { "epoch": 2.977974037029155, "grad_norm": 13.96875, "learning_rate": 9.023344229675953e-06, "loss": 0.9903, "step": 27987 }, { "epoch": 2.978080442647372, "grad_norm": 11.0859375, "learning_rate": 9.023308698123934e-06, "loss": 1.0015, "step": 27988 }, { "epoch": 2.9781868482655884, "grad_norm": 23.46875, "learning_rate": 9.023273166571916e-06, "loss": 0.9482, "step": 27989 }, { "epoch": 2.978293253883805, "grad_norm": 14.796875, "learning_rate": 9.023237635019897e-06, "loss": 0.8532, "step": 27990 }, { "epoch": 2.9783996595020215, "grad_norm": 15.390625, "learning_rate": 9.02320210346788e-06, "loss": 0.9962, "step": 27991 }, { "epoch": 2.9785060651202384, "grad_norm": 14.1171875, "learning_rate": 9.023166571915862e-06, "loss": 0.9288, "step": 27992 }, { "epoch": 2.978612470738455, "grad_norm": 17.71875, "learning_rate": 9.023131040363844e-06, "loss": 1.0152, "step": 27993 }, { "epoch": 2.9787188763566714, "grad_norm": 13.2578125, "learning_rate": 9.023095508811825e-06, "loss": 1.0039, "step": 27994 }, { "epoch": 2.9788252819748884, "grad_norm": 13.09375, "learning_rate": 9.023059977259807e-06, "loss": 0.9564, "step": 27995 }, { "epoch": 2.978931687593105, "grad_norm": 13.890625, "learning_rate": 9.02302444570779e-06, "loss": 0.9029, "step": 27996 }, { "epoch": 2.9790380932113214, "grad_norm": 10.921875, "learning_rate": 9.022988914155772e-06, "loss": 0.9964, "step": 27997 }, { "epoch": 2.9791444988295384, "grad_norm": 10.4921875, "learning_rate": 9.022953382603754e-06, "loss": 0.9415, "step": 27998 }, { "epoch": 2.979250904447755, "grad_norm": 14.1640625, "learning_rate": 9.022917851051735e-06, "loss": 0.9952, "step": 27999 }, { "epoch": 2.9793573100659714, "grad_norm": 11.890625, "learning_rate": 9.022882319499717e-06, "loss": 1.0308, "step": 28000 }, { "epoch": 2.9794637156841883, "grad_norm": 11.5, "learning_rate": 9.022846787947698e-06, "loss": 0.7607, "step": 28001 }, { "epoch": 2.979570121302405, "grad_norm": 15.9765625, "learning_rate": 9.02281125639568e-06, "loss": 0.9555, "step": 28002 }, { "epoch": 2.9796765269206213, "grad_norm": 14.5859375, "learning_rate": 9.022775724843662e-06, "loss": 0.9878, "step": 28003 }, { "epoch": 2.9797829325388383, "grad_norm": 17.3125, "learning_rate": 9.022740193291643e-06, "loss": 0.976, "step": 28004 }, { "epoch": 2.979889338157055, "grad_norm": 15.171875, "learning_rate": 9.022704661739625e-06, "loss": 0.8903, "step": 28005 }, { "epoch": 2.9799957437752713, "grad_norm": 12.921875, "learning_rate": 9.022669130187606e-06, "loss": 0.7832, "step": 28006 }, { "epoch": 2.980102149393488, "grad_norm": 10.9609375, "learning_rate": 9.02263359863559e-06, "loss": 0.772, "step": 28007 }, { "epoch": 2.9802085550117043, "grad_norm": 20.0, "learning_rate": 9.022598067083571e-06, "loss": 0.9699, "step": 28008 }, { "epoch": 2.9803149606299213, "grad_norm": 11.5234375, "learning_rate": 9.022562535531553e-06, "loss": 0.7123, "step": 28009 }, { "epoch": 2.980421366248138, "grad_norm": 11.2890625, "learning_rate": 9.022527003979535e-06, "loss": 0.8504, "step": 28010 }, { "epoch": 2.9805277718663543, "grad_norm": 10.0390625, "learning_rate": 9.022491472427516e-06, "loss": 0.9287, "step": 28011 }, { "epoch": 2.9806341774845713, "grad_norm": 12.5625, "learning_rate": 9.022455940875498e-06, "loss": 0.9273, "step": 28012 }, { "epoch": 2.9807405831027878, "grad_norm": 11.8515625, "learning_rate": 9.02242040932348e-06, "loss": 0.9037, "step": 28013 }, { "epoch": 2.9808469887210043, "grad_norm": 10.8828125, "learning_rate": 9.022384877771461e-06, "loss": 0.8537, "step": 28014 }, { "epoch": 2.9809533943392212, "grad_norm": 13.75, "learning_rate": 9.022349346219443e-06, "loss": 0.9107, "step": 28015 }, { "epoch": 2.9810597999574378, "grad_norm": 15.1953125, "learning_rate": 9.022313814667424e-06, "loss": 0.9871, "step": 28016 }, { "epoch": 2.9811662055756543, "grad_norm": 12.015625, "learning_rate": 9.022278283115407e-06, "loss": 0.9392, "step": 28017 }, { "epoch": 2.981272611193871, "grad_norm": 11.9921875, "learning_rate": 9.022242751563389e-06, "loss": 0.9399, "step": 28018 }, { "epoch": 2.9813790168120877, "grad_norm": 14.96875, "learning_rate": 9.02220722001137e-06, "loss": 0.8318, "step": 28019 }, { "epoch": 2.9814854224303042, "grad_norm": 10.5234375, "learning_rate": 9.022171688459352e-06, "loss": 0.9746, "step": 28020 }, { "epoch": 2.981591828048521, "grad_norm": 11.4609375, "learning_rate": 9.022136156907334e-06, "loss": 0.9047, "step": 28021 }, { "epoch": 2.9816982336667377, "grad_norm": 12.0, "learning_rate": 9.022100625355317e-06, "loss": 0.8835, "step": 28022 }, { "epoch": 2.981804639284954, "grad_norm": 13.3046875, "learning_rate": 9.022065093803299e-06, "loss": 0.8057, "step": 28023 }, { "epoch": 2.981911044903171, "grad_norm": 18.53125, "learning_rate": 9.02202956225128e-06, "loss": 1.1514, "step": 28024 }, { "epoch": 2.9820174505213877, "grad_norm": 12.921875, "learning_rate": 9.021994030699262e-06, "loss": 0.838, "step": 28025 }, { "epoch": 2.982123856139604, "grad_norm": 13.7890625, "learning_rate": 9.021958499147244e-06, "loss": 0.8189, "step": 28026 }, { "epoch": 2.9822302617578207, "grad_norm": 12.0546875, "learning_rate": 9.021922967595225e-06, "loss": 0.8463, "step": 28027 }, { "epoch": 2.982336667376037, "grad_norm": 15.28125, "learning_rate": 9.021887436043207e-06, "loss": 1.3372, "step": 28028 }, { "epoch": 2.982443072994254, "grad_norm": 13.96875, "learning_rate": 9.021851904491188e-06, "loss": 0.7578, "step": 28029 }, { "epoch": 2.9825494786124707, "grad_norm": 13.671875, "learning_rate": 9.02181637293917e-06, "loss": 0.8228, "step": 28030 }, { "epoch": 2.982655884230687, "grad_norm": 15.078125, "learning_rate": 9.021780841387152e-06, "loss": 1.0992, "step": 28031 }, { "epoch": 2.982762289848904, "grad_norm": 13.0, "learning_rate": 9.021745309835135e-06, "loss": 0.9091, "step": 28032 }, { "epoch": 2.9828686954671206, "grad_norm": 14.09375, "learning_rate": 9.021709778283117e-06, "loss": 0.9162, "step": 28033 }, { "epoch": 2.982975101085337, "grad_norm": 11.125, "learning_rate": 9.021674246731098e-06, "loss": 0.8948, "step": 28034 }, { "epoch": 2.983081506703554, "grad_norm": 12.2890625, "learning_rate": 9.02163871517908e-06, "loss": 0.975, "step": 28035 }, { "epoch": 2.9831879123217706, "grad_norm": 12.3203125, "learning_rate": 9.021603183627061e-06, "loss": 0.9439, "step": 28036 }, { "epoch": 2.983294317939987, "grad_norm": 14.609375, "learning_rate": 9.021567652075045e-06, "loss": 0.9503, "step": 28037 }, { "epoch": 2.983400723558204, "grad_norm": 10.8671875, "learning_rate": 9.021532120523026e-06, "loss": 0.8976, "step": 28038 }, { "epoch": 2.9835071291764206, "grad_norm": 11.8125, "learning_rate": 9.021496588971008e-06, "loss": 0.8874, "step": 28039 }, { "epoch": 2.983613534794637, "grad_norm": 11.015625, "learning_rate": 9.02146105741899e-06, "loss": 0.9462, "step": 28040 }, { "epoch": 2.983719940412854, "grad_norm": 12.859375, "learning_rate": 9.021425525866971e-06, "loss": 0.9112, "step": 28041 }, { "epoch": 2.9838263460310706, "grad_norm": 11.9453125, "learning_rate": 9.021389994314951e-06, "loss": 0.7394, "step": 28042 }, { "epoch": 2.983932751649287, "grad_norm": 11.0234375, "learning_rate": 9.021354462762934e-06, "loss": 0.8439, "step": 28043 }, { "epoch": 2.9840391572675036, "grad_norm": 12.28125, "learning_rate": 9.021318931210916e-06, "loss": 0.9585, "step": 28044 }, { "epoch": 2.9841455628857205, "grad_norm": 16.421875, "learning_rate": 9.021283399658898e-06, "loss": 0.9635, "step": 28045 }, { "epoch": 2.984251968503937, "grad_norm": 15.8359375, "learning_rate": 9.02124786810688e-06, "loss": 0.9884, "step": 28046 }, { "epoch": 2.9843583741221535, "grad_norm": 12.671875, "learning_rate": 9.02121233655486e-06, "loss": 0.9634, "step": 28047 }, { "epoch": 2.98446477974037, "grad_norm": 12.890625, "learning_rate": 9.021176805002844e-06, "loss": 0.8507, "step": 28048 }, { "epoch": 2.984571185358587, "grad_norm": 11.34375, "learning_rate": 9.021141273450826e-06, "loss": 0.7679, "step": 28049 }, { "epoch": 2.9846775909768035, "grad_norm": 10.78125, "learning_rate": 9.021105741898807e-06, "loss": 0.776, "step": 28050 }, { "epoch": 2.98478399659502, "grad_norm": 11.8828125, "learning_rate": 9.021070210346789e-06, "loss": 0.8655, "step": 28051 }, { "epoch": 2.984890402213237, "grad_norm": 13.234375, "learning_rate": 9.02103467879477e-06, "loss": 0.6767, "step": 28052 }, { "epoch": 2.9849968078314535, "grad_norm": 10.7578125, "learning_rate": 9.020999147242752e-06, "loss": 0.9072, "step": 28053 }, { "epoch": 2.98510321344967, "grad_norm": 7.25, "learning_rate": 9.020963615690734e-06, "loss": 0.8149, "step": 28054 }, { "epoch": 2.985209619067887, "grad_norm": 15.40625, "learning_rate": 9.020928084138715e-06, "loss": 0.9526, "step": 28055 }, { "epoch": 2.9853160246861035, "grad_norm": 9.2734375, "learning_rate": 9.020892552586697e-06, "loss": 0.777, "step": 28056 }, { "epoch": 2.98542243030432, "grad_norm": 18.296875, "learning_rate": 9.020857021034679e-06, "loss": 0.9756, "step": 28057 }, { "epoch": 2.985528835922537, "grad_norm": 17.3125, "learning_rate": 9.020821489482662e-06, "loss": 0.9894, "step": 28058 }, { "epoch": 2.9856352415407534, "grad_norm": 16.703125, "learning_rate": 9.020785957930643e-06, "loss": 0.8454, "step": 28059 }, { "epoch": 2.98574164715897, "grad_norm": 14.6015625, "learning_rate": 9.020750426378625e-06, "loss": 0.8433, "step": 28060 }, { "epoch": 2.985848052777187, "grad_norm": 12.8203125, "learning_rate": 9.020714894826607e-06, "loss": 0.8206, "step": 28061 }, { "epoch": 2.9859544583954034, "grad_norm": 15.6015625, "learning_rate": 9.020679363274588e-06, "loss": 0.9563, "step": 28062 }, { "epoch": 2.98606086401362, "grad_norm": 21.078125, "learning_rate": 9.020643831722572e-06, "loss": 1.1456, "step": 28063 }, { "epoch": 2.9861672696318364, "grad_norm": 12.9140625, "learning_rate": 9.020608300170553e-06, "loss": 0.989, "step": 28064 }, { "epoch": 2.986273675250053, "grad_norm": 13.5390625, "learning_rate": 9.020572768618535e-06, "loss": 1.0978, "step": 28065 }, { "epoch": 2.98638008086827, "grad_norm": 11.0859375, "learning_rate": 9.020537237066516e-06, "loss": 0.9242, "step": 28066 }, { "epoch": 2.9864864864864864, "grad_norm": 12.453125, "learning_rate": 9.020501705514498e-06, "loss": 0.912, "step": 28067 }, { "epoch": 2.986592892104703, "grad_norm": 15.3984375, "learning_rate": 9.02046617396248e-06, "loss": 1.0486, "step": 28068 }, { "epoch": 2.98669929772292, "grad_norm": 11.3046875, "learning_rate": 9.020430642410461e-06, "loss": 0.8957, "step": 28069 }, { "epoch": 2.9868057033411364, "grad_norm": 16.453125, "learning_rate": 9.020395110858443e-06, "loss": 0.823, "step": 28070 }, { "epoch": 2.986912108959353, "grad_norm": 15.53125, "learning_rate": 9.020359579306424e-06, "loss": 1.0721, "step": 28071 }, { "epoch": 2.98701851457757, "grad_norm": 13.203125, "learning_rate": 9.020324047754406e-06, "loss": 1.2364, "step": 28072 }, { "epoch": 2.9871249201957863, "grad_norm": 11.578125, "learning_rate": 9.02028851620239e-06, "loss": 0.8425, "step": 28073 }, { "epoch": 2.987231325814003, "grad_norm": 12.8046875, "learning_rate": 9.020252984650371e-06, "loss": 1.0068, "step": 28074 }, { "epoch": 2.98733773143222, "grad_norm": 10.5, "learning_rate": 9.020217453098353e-06, "loss": 0.9072, "step": 28075 }, { "epoch": 2.9874441370504363, "grad_norm": 11.3046875, "learning_rate": 9.020181921546334e-06, "loss": 0.9376, "step": 28076 }, { "epoch": 2.987550542668653, "grad_norm": 14.5390625, "learning_rate": 9.020146389994316e-06, "loss": 0.7204, "step": 28077 }, { "epoch": 2.98765694828687, "grad_norm": 11.921875, "learning_rate": 9.020110858442297e-06, "loss": 0.9275, "step": 28078 }, { "epoch": 2.9877633539050863, "grad_norm": 12.6796875, "learning_rate": 9.020075326890279e-06, "loss": 0.8137, "step": 28079 }, { "epoch": 2.987869759523303, "grad_norm": 15.9765625, "learning_rate": 9.02003979533826e-06, "loss": 1.0017, "step": 28080 }, { "epoch": 2.9879761651415193, "grad_norm": 19.984375, "learning_rate": 9.020004263786242e-06, "loss": 0.981, "step": 28081 }, { "epoch": 2.9880825707597363, "grad_norm": 13.171875, "learning_rate": 9.019968732234224e-06, "loss": 1.0166, "step": 28082 }, { "epoch": 2.9881889763779528, "grad_norm": 17.1875, "learning_rate": 9.019933200682205e-06, "loss": 1.0628, "step": 28083 }, { "epoch": 2.9882953819961693, "grad_norm": 16.90625, "learning_rate": 9.019897669130189e-06, "loss": 0.8414, "step": 28084 }, { "epoch": 2.988401787614386, "grad_norm": 13.4375, "learning_rate": 9.01986213757817e-06, "loss": 0.7964, "step": 28085 }, { "epoch": 2.9885081932326027, "grad_norm": 13.3671875, "learning_rate": 9.019826606026152e-06, "loss": 0.9678, "step": 28086 }, { "epoch": 2.9886145988508193, "grad_norm": 10.796875, "learning_rate": 9.019791074474134e-06, "loss": 0.88, "step": 28087 }, { "epoch": 2.9887210044690358, "grad_norm": 12.109375, "learning_rate": 9.019755542922115e-06, "loss": 0.9858, "step": 28088 }, { "epoch": 2.9888274100872527, "grad_norm": 15.6875, "learning_rate": 9.019720011370098e-06, "loss": 0.8777, "step": 28089 }, { "epoch": 2.9889338157054692, "grad_norm": 16.296875, "learning_rate": 9.01968447981808e-06, "loss": 1.1219, "step": 28090 }, { "epoch": 2.9890402213236857, "grad_norm": 11.7578125, "learning_rate": 9.019648948266062e-06, "loss": 0.9257, "step": 28091 }, { "epoch": 2.9891466269419027, "grad_norm": 15.765625, "learning_rate": 9.019613416714043e-06, "loss": 0.9482, "step": 28092 }, { "epoch": 2.989253032560119, "grad_norm": 13.28125, "learning_rate": 9.019577885162025e-06, "loss": 0.9085, "step": 28093 }, { "epoch": 2.9893594381783357, "grad_norm": 16.84375, "learning_rate": 9.019542353610006e-06, "loss": 0.9822, "step": 28094 }, { "epoch": 2.9894658437965527, "grad_norm": 12.7890625, "learning_rate": 9.019506822057988e-06, "loss": 0.867, "step": 28095 }, { "epoch": 2.989572249414769, "grad_norm": 11.71875, "learning_rate": 9.01947129050597e-06, "loss": 0.9691, "step": 28096 }, { "epoch": 2.9896786550329857, "grad_norm": 20.125, "learning_rate": 9.019435758953951e-06, "loss": 1.0882, "step": 28097 }, { "epoch": 2.9897850606512026, "grad_norm": 11.140625, "learning_rate": 9.019400227401933e-06, "loss": 1.0315, "step": 28098 }, { "epoch": 2.989891466269419, "grad_norm": 14.9140625, "learning_rate": 9.019364695849916e-06, "loss": 1.059, "step": 28099 }, { "epoch": 2.9899978718876357, "grad_norm": 11.7109375, "learning_rate": 9.019329164297898e-06, "loss": 0.996, "step": 28100 }, { "epoch": 2.990104277505852, "grad_norm": 15.09375, "learning_rate": 9.01929363274588e-06, "loss": 1.0153, "step": 28101 }, { "epoch": 2.9902106831240687, "grad_norm": 13.2421875, "learning_rate": 9.019258101193861e-06, "loss": 0.8961, "step": 28102 }, { "epoch": 2.9903170887422856, "grad_norm": 15.7265625, "learning_rate": 9.019222569641843e-06, "loss": 0.8887, "step": 28103 }, { "epoch": 2.990423494360502, "grad_norm": 18.21875, "learning_rate": 9.019187038089824e-06, "loss": 0.9593, "step": 28104 }, { "epoch": 2.9905298999787187, "grad_norm": 14.53125, "learning_rate": 9.019151506537806e-06, "loss": 0.9056, "step": 28105 }, { "epoch": 2.9906363055969356, "grad_norm": 11.2890625, "learning_rate": 9.019115974985787e-06, "loss": 0.9099, "step": 28106 }, { "epoch": 2.990742711215152, "grad_norm": 13.234375, "learning_rate": 9.019080443433769e-06, "loss": 1.0737, "step": 28107 }, { "epoch": 2.9908491168333686, "grad_norm": 13.0, "learning_rate": 9.01904491188175e-06, "loss": 0.8818, "step": 28108 }, { "epoch": 2.9909555224515856, "grad_norm": 11.921875, "learning_rate": 9.019009380329734e-06, "loss": 0.9307, "step": 28109 }, { "epoch": 2.991061928069802, "grad_norm": 12.71875, "learning_rate": 9.018973848777716e-06, "loss": 0.9455, "step": 28110 }, { "epoch": 2.9911683336880186, "grad_norm": 11.09375, "learning_rate": 9.018938317225697e-06, "loss": 0.8627, "step": 28111 }, { "epoch": 2.9912747393062356, "grad_norm": 13.8125, "learning_rate": 9.018902785673679e-06, "loss": 0.835, "step": 28112 }, { "epoch": 2.991381144924452, "grad_norm": 17.65625, "learning_rate": 9.01886725412166e-06, "loss": 0.9711, "step": 28113 }, { "epoch": 2.9914875505426686, "grad_norm": 12.015625, "learning_rate": 9.018831722569644e-06, "loss": 0.9452, "step": 28114 }, { "epoch": 2.9915939561608855, "grad_norm": 11.1875, "learning_rate": 9.018796191017625e-06, "loss": 0.9678, "step": 28115 }, { "epoch": 2.991700361779102, "grad_norm": 12.3671875, "learning_rate": 9.018760659465607e-06, "loss": 0.9723, "step": 28116 }, { "epoch": 2.9918067673973185, "grad_norm": 14.609375, "learning_rate": 9.018725127913588e-06, "loss": 0.722, "step": 28117 }, { "epoch": 2.991913173015535, "grad_norm": 13.1640625, "learning_rate": 9.01868959636157e-06, "loss": 0.9681, "step": 28118 }, { "epoch": 2.992019578633752, "grad_norm": 11.8515625, "learning_rate": 9.018654064809552e-06, "loss": 0.8212, "step": 28119 }, { "epoch": 2.9921259842519685, "grad_norm": 13.6015625, "learning_rate": 9.018618533257533e-06, "loss": 0.9024, "step": 28120 }, { "epoch": 2.992232389870185, "grad_norm": 11.796875, "learning_rate": 9.018583001705515e-06, "loss": 1.0235, "step": 28121 }, { "epoch": 2.9923387954884015, "grad_norm": 13.078125, "learning_rate": 9.018547470153497e-06, "loss": 1.0965, "step": 28122 }, { "epoch": 2.9924452011066185, "grad_norm": 13.0234375, "learning_rate": 9.018511938601478e-06, "loss": 0.8787, "step": 28123 }, { "epoch": 2.992551606724835, "grad_norm": 14.53125, "learning_rate": 9.01847640704946e-06, "loss": 1.051, "step": 28124 }, { "epoch": 2.9926580123430515, "grad_norm": 12.0703125, "learning_rate": 9.018440875497443e-06, "loss": 0.8337, "step": 28125 }, { "epoch": 2.9927644179612685, "grad_norm": 12.5625, "learning_rate": 9.018405343945425e-06, "loss": 0.8797, "step": 28126 }, { "epoch": 2.992870823579485, "grad_norm": 11.1328125, "learning_rate": 9.018369812393406e-06, "loss": 0.6331, "step": 28127 }, { "epoch": 2.9929772291977015, "grad_norm": 14.296875, "learning_rate": 9.018334280841388e-06, "loss": 0.8669, "step": 28128 }, { "epoch": 2.9930836348159184, "grad_norm": 11.8671875, "learning_rate": 9.01829874928937e-06, "loss": 0.9569, "step": 28129 }, { "epoch": 2.993190040434135, "grad_norm": 16.703125, "learning_rate": 9.018263217737353e-06, "loss": 1.1128, "step": 28130 }, { "epoch": 2.9932964460523515, "grad_norm": 15.328125, "learning_rate": 9.018227686185333e-06, "loss": 0.9719, "step": 28131 }, { "epoch": 2.9934028516705684, "grad_norm": 13.6796875, "learning_rate": 9.018192154633314e-06, "loss": 0.8734, "step": 28132 }, { "epoch": 2.993509257288785, "grad_norm": 10.3125, "learning_rate": 9.018156623081296e-06, "loss": 0.947, "step": 28133 }, { "epoch": 2.9936156629070014, "grad_norm": 9.8671875, "learning_rate": 9.018121091529277e-06, "loss": 0.8625, "step": 28134 }, { "epoch": 2.9937220685252184, "grad_norm": 12.6171875, "learning_rate": 9.01808555997726e-06, "loss": 1.0761, "step": 28135 }, { "epoch": 2.993828474143435, "grad_norm": 14.625, "learning_rate": 9.018050028425242e-06, "loss": 0.9824, "step": 28136 }, { "epoch": 2.9939348797616514, "grad_norm": 12.0, "learning_rate": 9.018014496873224e-06, "loss": 0.9389, "step": 28137 }, { "epoch": 2.994041285379868, "grad_norm": 11.3359375, "learning_rate": 9.017978965321206e-06, "loss": 0.9856, "step": 28138 }, { "epoch": 2.994147690998085, "grad_norm": 12.5078125, "learning_rate": 9.017943433769187e-06, "loss": 0.8673, "step": 28139 }, { "epoch": 2.9942540966163014, "grad_norm": 14.1953125, "learning_rate": 9.01790790221717e-06, "loss": 1.1125, "step": 28140 }, { "epoch": 2.994360502234518, "grad_norm": 12.21875, "learning_rate": 9.017872370665152e-06, "loss": 0.8082, "step": 28141 }, { "epoch": 2.9944669078527344, "grad_norm": 10.515625, "learning_rate": 9.017836839113134e-06, "loss": 0.8772, "step": 28142 }, { "epoch": 2.9945733134709513, "grad_norm": 12.4765625, "learning_rate": 9.017801307561115e-06, "loss": 1.0003, "step": 28143 }, { "epoch": 2.994679719089168, "grad_norm": 11.7578125, "learning_rate": 9.017765776009097e-06, "loss": 0.8593, "step": 28144 }, { "epoch": 2.9947861247073844, "grad_norm": 13.203125, "learning_rate": 9.017730244457079e-06, "loss": 0.95, "step": 28145 }, { "epoch": 2.9948925303256013, "grad_norm": 14.4765625, "learning_rate": 9.01769471290506e-06, "loss": 1.1299, "step": 28146 }, { "epoch": 2.994998935943818, "grad_norm": 11.59375, "learning_rate": 9.017659181353042e-06, "loss": 0.8058, "step": 28147 }, { "epoch": 2.9951053415620343, "grad_norm": 11.8046875, "learning_rate": 9.017623649801023e-06, "loss": 0.7256, "step": 28148 }, { "epoch": 2.9952117471802513, "grad_norm": 19.21875, "learning_rate": 9.017588118249005e-06, "loss": 0.9728, "step": 28149 }, { "epoch": 2.995318152798468, "grad_norm": 13.46875, "learning_rate": 9.017552586696988e-06, "loss": 0.8084, "step": 28150 }, { "epoch": 2.9954245584166843, "grad_norm": 19.453125, "learning_rate": 9.01751705514497e-06, "loss": 1.3345, "step": 28151 }, { "epoch": 2.9955309640349013, "grad_norm": 12.5390625, "learning_rate": 9.017481523592951e-06, "loss": 0.9221, "step": 28152 }, { "epoch": 2.9956373696531178, "grad_norm": 11.7265625, "learning_rate": 9.017445992040933e-06, "loss": 0.9593, "step": 28153 }, { "epoch": 2.9957437752713343, "grad_norm": 13.5234375, "learning_rate": 9.017410460488915e-06, "loss": 0.9367, "step": 28154 }, { "epoch": 2.9958501808895512, "grad_norm": 15.6640625, "learning_rate": 9.017374928936898e-06, "loss": 0.9101, "step": 28155 }, { "epoch": 2.9959565865077677, "grad_norm": 13.734375, "learning_rate": 9.01733939738488e-06, "loss": 0.8821, "step": 28156 }, { "epoch": 2.9960629921259843, "grad_norm": 12.125, "learning_rate": 9.017303865832861e-06, "loss": 0.7912, "step": 28157 }, { "epoch": 2.9961693977442008, "grad_norm": 14.28125, "learning_rate": 9.017268334280843e-06, "loss": 1.0811, "step": 28158 }, { "epoch": 2.9962758033624173, "grad_norm": 15.5390625, "learning_rate": 9.017232802728824e-06, "loss": 1.0178, "step": 28159 }, { "epoch": 2.9963822089806342, "grad_norm": 14.984375, "learning_rate": 9.017197271176806e-06, "loss": 0.8601, "step": 28160 }, { "epoch": 2.9964886145988507, "grad_norm": 12.0390625, "learning_rate": 9.017161739624788e-06, "loss": 0.9695, "step": 28161 }, { "epoch": 2.9965950202170673, "grad_norm": 11.7578125, "learning_rate": 9.01712620807277e-06, "loss": 0.9235, "step": 28162 }, { "epoch": 2.996701425835284, "grad_norm": 12.8828125, "learning_rate": 9.017090676520751e-06, "loss": 0.8406, "step": 28163 }, { "epoch": 2.9968078314535007, "grad_norm": 15.3828125, "learning_rate": 9.017055144968732e-06, "loss": 0.9451, "step": 28164 }, { "epoch": 2.9969142370717172, "grad_norm": 13.8828125, "learning_rate": 9.017019613416714e-06, "loss": 0.818, "step": 28165 }, { "epoch": 2.997020642689934, "grad_norm": 13.0859375, "learning_rate": 9.016984081864697e-06, "loss": 0.8954, "step": 28166 }, { "epoch": 2.9971270483081507, "grad_norm": 12.1875, "learning_rate": 9.016948550312679e-06, "loss": 0.9131, "step": 28167 }, { "epoch": 2.997233453926367, "grad_norm": 11.6328125, "learning_rate": 9.01691301876066e-06, "loss": 0.9332, "step": 28168 }, { "epoch": 2.997339859544584, "grad_norm": 13.3984375, "learning_rate": 9.016877487208642e-06, "loss": 0.8495, "step": 28169 }, { "epoch": 2.9974462651628007, "grad_norm": 13.3828125, "learning_rate": 9.016841955656624e-06, "loss": 0.8288, "step": 28170 }, { "epoch": 2.997552670781017, "grad_norm": 8.84375, "learning_rate": 9.016806424104605e-06, "loss": 0.7922, "step": 28171 }, { "epoch": 2.997659076399234, "grad_norm": 9.78125, "learning_rate": 9.016770892552587e-06, "loss": 0.7364, "step": 28172 }, { "epoch": 2.9977654820174506, "grad_norm": 13.8671875, "learning_rate": 9.016735361000569e-06, "loss": 0.8632, "step": 28173 }, { "epoch": 2.997871887635667, "grad_norm": 12.25, "learning_rate": 9.01669982944855e-06, "loss": 0.9991, "step": 28174 }, { "epoch": 2.9979782932538837, "grad_norm": 10.8046875, "learning_rate": 9.016664297896532e-06, "loss": 0.9304, "step": 28175 }, { "epoch": 2.9980846988721006, "grad_norm": 14.8828125, "learning_rate": 9.016628766344515e-06, "loss": 1.051, "step": 28176 }, { "epoch": 2.998191104490317, "grad_norm": 12.15625, "learning_rate": 9.016593234792497e-06, "loss": 0.8964, "step": 28177 }, { "epoch": 2.9982975101085336, "grad_norm": 8.0390625, "learning_rate": 9.016557703240478e-06, "loss": 0.7179, "step": 28178 }, { "epoch": 2.99840391572675, "grad_norm": 9.6640625, "learning_rate": 9.01652217168846e-06, "loss": 0.772, "step": 28179 }, { "epoch": 2.998510321344967, "grad_norm": 14.2109375, "learning_rate": 9.016486640136442e-06, "loss": 0.7888, "step": 28180 }, { "epoch": 2.9986167269631836, "grad_norm": 13.140625, "learning_rate": 9.016451108584425e-06, "loss": 0.8014, "step": 28181 }, { "epoch": 2.9987231325814, "grad_norm": 11.015625, "learning_rate": 9.016415577032406e-06, "loss": 0.8036, "step": 28182 }, { "epoch": 2.998829538199617, "grad_norm": 14.2890625, "learning_rate": 9.016380045480388e-06, "loss": 0.9599, "step": 28183 }, { "epoch": 2.9989359438178336, "grad_norm": 13.109375, "learning_rate": 9.01634451392837e-06, "loss": 0.9156, "step": 28184 }, { "epoch": 2.99904234943605, "grad_norm": 16.90625, "learning_rate": 9.016308982376351e-06, "loss": 0.6236, "step": 28185 }, { "epoch": 2.999148755054267, "grad_norm": 12.109375, "learning_rate": 9.016273450824333e-06, "loss": 0.9815, "step": 28186 }, { "epoch": 2.9992551606724835, "grad_norm": 14.1875, "learning_rate": 9.016237919272315e-06, "loss": 1.0867, "step": 28187 }, { "epoch": 2.9993615662907, "grad_norm": 11.1640625, "learning_rate": 9.016202387720296e-06, "loss": 0.9383, "step": 28188 }, { "epoch": 2.999467971908917, "grad_norm": 15.125, "learning_rate": 9.016166856168278e-06, "loss": 1.0827, "step": 28189 }, { "epoch": 2.9995743775271335, "grad_norm": 19.953125, "learning_rate": 9.01613132461626e-06, "loss": 0.7523, "step": 28190 }, { "epoch": 2.99968078314535, "grad_norm": 15.0078125, "learning_rate": 9.016095793064243e-06, "loss": 0.9144, "step": 28191 }, { "epoch": 2.999787188763567, "grad_norm": 27.703125, "learning_rate": 9.016060261512224e-06, "loss": 0.9437, "step": 28192 }, { "epoch": 2.9998935943817835, "grad_norm": 12.8125, "learning_rate": 9.016024729960206e-06, "loss": 0.8816, "step": 28193 }, { "epoch": 3.0, "grad_norm": 20.078125, "learning_rate": 9.015989198408187e-06, "loss": 0.9193, "step": 28194 }, { "epoch": 3.0001064056182165, "grad_norm": 9.6796875, "learning_rate": 9.015953666856169e-06, "loss": 0.6571, "step": 28195 }, { "epoch": 3.0002128112364335, "grad_norm": 11.984375, "learning_rate": 9.01591813530415e-06, "loss": 0.8065, "step": 28196 }, { "epoch": 3.00031921685465, "grad_norm": 7.95703125, "learning_rate": 9.015882603752132e-06, "loss": 0.5954, "step": 28197 }, { "epoch": 3.0004256224728665, "grad_norm": 8.2734375, "learning_rate": 9.015847072200114e-06, "loss": 0.6231, "step": 28198 }, { "epoch": 3.000532028091083, "grad_norm": 12.359375, "learning_rate": 9.015811540648095e-06, "loss": 0.7308, "step": 28199 }, { "epoch": 3.0006384337093, "grad_norm": 8.3359375, "learning_rate": 9.015776009096077e-06, "loss": 0.5809, "step": 28200 }, { "epoch": 3.0007448393275165, "grad_norm": 10.75, "learning_rate": 9.015740477544059e-06, "loss": 0.6546, "step": 28201 }, { "epoch": 3.000851244945733, "grad_norm": 12.328125, "learning_rate": 9.015704945992042e-06, "loss": 0.6501, "step": 28202 }, { "epoch": 3.00095765056395, "grad_norm": 10.2890625, "learning_rate": 9.015669414440024e-06, "loss": 0.581, "step": 28203 }, { "epoch": 3.0010640561821664, "grad_norm": 9.203125, "learning_rate": 9.015633882888005e-06, "loss": 0.558, "step": 28204 }, { "epoch": 3.001170461800383, "grad_norm": 9.25, "learning_rate": 9.015598351335987e-06, "loss": 0.5032, "step": 28205 }, { "epoch": 3.0012768674186, "grad_norm": 9.7890625, "learning_rate": 9.015562819783968e-06, "loss": 0.6124, "step": 28206 }, { "epoch": 3.0013832730368164, "grad_norm": 9.78125, "learning_rate": 9.015527288231952e-06, "loss": 0.6424, "step": 28207 }, { "epoch": 3.001489678655033, "grad_norm": 14.3203125, "learning_rate": 9.015491756679933e-06, "loss": 0.7461, "step": 28208 }, { "epoch": 3.0015960842732494, "grad_norm": 11.953125, "learning_rate": 9.015456225127915e-06, "loss": 0.5903, "step": 28209 }, { "epoch": 3.0017024898914664, "grad_norm": 11.5, "learning_rate": 9.015420693575897e-06, "loss": 0.6834, "step": 28210 }, { "epoch": 3.001808895509683, "grad_norm": 10.6328125, "learning_rate": 9.015385162023878e-06, "loss": 0.6088, "step": 28211 }, { "epoch": 3.0019153011278994, "grad_norm": 14.421875, "learning_rate": 9.01534963047186e-06, "loss": 0.7945, "step": 28212 }, { "epoch": 3.0020217067461163, "grad_norm": 11.015625, "learning_rate": 9.015314098919841e-06, "loss": 0.6642, "step": 28213 }, { "epoch": 3.002128112364333, "grad_norm": 10.2734375, "learning_rate": 9.015278567367823e-06, "loss": 0.6344, "step": 28214 }, { "epoch": 3.0022345179825494, "grad_norm": 14.875, "learning_rate": 9.015243035815805e-06, "loss": 0.6063, "step": 28215 }, { "epoch": 3.0023409236007663, "grad_norm": 14.5390625, "learning_rate": 9.015207504263786e-06, "loss": 0.6896, "step": 28216 }, { "epoch": 3.002447329218983, "grad_norm": 11.4609375, "learning_rate": 9.01517197271177e-06, "loss": 0.7077, "step": 28217 }, { "epoch": 3.0025537348371993, "grad_norm": 9.5703125, "learning_rate": 9.015136441159751e-06, "loss": 0.536, "step": 28218 }, { "epoch": 3.002660140455416, "grad_norm": 15.546875, "learning_rate": 9.015100909607733e-06, "loss": 0.7285, "step": 28219 }, { "epoch": 3.002766546073633, "grad_norm": 11.296875, "learning_rate": 9.015065378055714e-06, "loss": 0.705, "step": 28220 }, { "epoch": 3.0028729516918493, "grad_norm": 19.046875, "learning_rate": 9.015029846503696e-06, "loss": 0.6551, "step": 28221 }, { "epoch": 3.002979357310066, "grad_norm": 9.125, "learning_rate": 9.014994314951678e-06, "loss": 0.5331, "step": 28222 }, { "epoch": 3.0030857629282828, "grad_norm": 9.0703125, "learning_rate": 9.014958783399659e-06, "loss": 0.5149, "step": 28223 }, { "epoch": 3.0031921685464993, "grad_norm": 13.4375, "learning_rate": 9.01492325184764e-06, "loss": 0.6242, "step": 28224 }, { "epoch": 3.003298574164716, "grad_norm": 10.2578125, "learning_rate": 9.014887720295622e-06, "loss": 0.5258, "step": 28225 }, { "epoch": 3.0034049797829327, "grad_norm": 10.359375, "learning_rate": 9.014852188743604e-06, "loss": 0.6087, "step": 28226 }, { "epoch": 3.0035113854011493, "grad_norm": 11.0703125, "learning_rate": 9.014816657191587e-06, "loss": 0.6038, "step": 28227 }, { "epoch": 3.0036177910193658, "grad_norm": 12.7734375, "learning_rate": 9.014781125639569e-06, "loss": 0.6702, "step": 28228 }, { "epoch": 3.0037241966375823, "grad_norm": 12.3671875, "learning_rate": 9.01474559408755e-06, "loss": 0.7088, "step": 28229 }, { "epoch": 3.0038306022557992, "grad_norm": 10.7890625, "learning_rate": 9.014710062535532e-06, "loss": 0.5867, "step": 28230 }, { "epoch": 3.0039370078740157, "grad_norm": 16.390625, "learning_rate": 9.014674530983514e-06, "loss": 0.7294, "step": 28231 }, { "epoch": 3.0040434134922323, "grad_norm": 13.8359375, "learning_rate": 9.014638999431497e-06, "loss": 0.6194, "step": 28232 }, { "epoch": 3.004149819110449, "grad_norm": 11.296875, "learning_rate": 9.014603467879479e-06, "loss": 0.5499, "step": 28233 }, { "epoch": 3.0042562247286657, "grad_norm": 16.609375, "learning_rate": 9.01456793632746e-06, "loss": 0.6327, "step": 28234 }, { "epoch": 3.0043626303468822, "grad_norm": 33.21875, "learning_rate": 9.014532404775442e-06, "loss": 0.7539, "step": 28235 }, { "epoch": 3.004469035965099, "grad_norm": 10.6171875, "learning_rate": 9.014496873223423e-06, "loss": 0.4624, "step": 28236 }, { "epoch": 3.0045754415833157, "grad_norm": 14.5234375, "learning_rate": 9.014461341671405e-06, "loss": 0.7775, "step": 28237 }, { "epoch": 3.004681847201532, "grad_norm": 12.140625, "learning_rate": 9.014425810119387e-06, "loss": 0.5985, "step": 28238 }, { "epoch": 3.0047882528197487, "grad_norm": 11.71875, "learning_rate": 9.014390278567368e-06, "loss": 0.6479, "step": 28239 }, { "epoch": 3.0048946584379657, "grad_norm": 11.0, "learning_rate": 9.01435474701535e-06, "loss": 0.5939, "step": 28240 }, { "epoch": 3.005001064056182, "grad_norm": 10.984375, "learning_rate": 9.014319215463331e-06, "loss": 0.5706, "step": 28241 }, { "epoch": 3.0051074696743987, "grad_norm": 12.5625, "learning_rate": 9.014283683911313e-06, "loss": 0.6082, "step": 28242 }, { "epoch": 3.0052138752926156, "grad_norm": 11.59375, "learning_rate": 9.014248152359296e-06, "loss": 0.5959, "step": 28243 }, { "epoch": 3.005320280910832, "grad_norm": 14.59375, "learning_rate": 9.014212620807278e-06, "loss": 0.8886, "step": 28244 }, { "epoch": 3.0054266865290487, "grad_norm": 11.6796875, "learning_rate": 9.01417708925526e-06, "loss": 0.4958, "step": 28245 }, { "epoch": 3.005533092147265, "grad_norm": 11.8828125, "learning_rate": 9.014141557703241e-06, "loss": 0.6572, "step": 28246 }, { "epoch": 3.005639497765482, "grad_norm": 13.1875, "learning_rate": 9.014106026151223e-06, "loss": 0.7658, "step": 28247 }, { "epoch": 3.0057459033836986, "grad_norm": 12.75, "learning_rate": 9.014070494599206e-06, "loss": 0.6278, "step": 28248 }, { "epoch": 3.005852309001915, "grad_norm": 10.640625, "learning_rate": 9.014034963047188e-06, "loss": 0.6019, "step": 28249 }, { "epoch": 3.005958714620132, "grad_norm": 13.2265625, "learning_rate": 9.013999431495168e-06, "loss": 0.5399, "step": 28250 }, { "epoch": 3.0060651202383486, "grad_norm": 16.203125, "learning_rate": 9.01396389994315e-06, "loss": 0.9727, "step": 28251 }, { "epoch": 3.006171525856565, "grad_norm": 9.6640625, "learning_rate": 9.01392836839113e-06, "loss": 0.607, "step": 28252 }, { "epoch": 3.006277931474782, "grad_norm": 13.1171875, "learning_rate": 9.013892836839114e-06, "loss": 0.6492, "step": 28253 }, { "epoch": 3.0063843370929986, "grad_norm": 9.5390625, "learning_rate": 9.013857305287096e-06, "loss": 0.551, "step": 28254 }, { "epoch": 3.006490742711215, "grad_norm": 10.296875, "learning_rate": 9.013821773735077e-06, "loss": 0.5785, "step": 28255 }, { "epoch": 3.0065971483294316, "grad_norm": 12.671875, "learning_rate": 9.013786242183059e-06, "loss": 0.627, "step": 28256 }, { "epoch": 3.0067035539476485, "grad_norm": 13.765625, "learning_rate": 9.01375071063104e-06, "loss": 0.568, "step": 28257 }, { "epoch": 3.006809959565865, "grad_norm": 13.6015625, "learning_rate": 9.013715179079024e-06, "loss": 0.6609, "step": 28258 }, { "epoch": 3.0069163651840816, "grad_norm": 11.4765625, "learning_rate": 9.013679647527005e-06, "loss": 0.6199, "step": 28259 }, { "epoch": 3.0070227708022985, "grad_norm": 9.625, "learning_rate": 9.013644115974987e-06, "loss": 0.5427, "step": 28260 }, { "epoch": 3.007129176420515, "grad_norm": 15.3046875, "learning_rate": 9.013608584422969e-06, "loss": 0.6787, "step": 28261 }, { "epoch": 3.0072355820387315, "grad_norm": 9.96875, "learning_rate": 9.01357305287095e-06, "loss": 0.5184, "step": 28262 }, { "epoch": 3.0073419876569485, "grad_norm": 11.4453125, "learning_rate": 9.013537521318932e-06, "loss": 0.5414, "step": 28263 }, { "epoch": 3.007448393275165, "grad_norm": 9.5859375, "learning_rate": 9.013501989766913e-06, "loss": 0.49, "step": 28264 }, { "epoch": 3.0075547988933815, "grad_norm": 10.6796875, "learning_rate": 9.013466458214895e-06, "loss": 0.5128, "step": 28265 }, { "epoch": 3.007661204511598, "grad_norm": 13.0234375, "learning_rate": 9.013430926662877e-06, "loss": 0.6072, "step": 28266 }, { "epoch": 3.007767610129815, "grad_norm": 10.40625, "learning_rate": 9.013395395110858e-06, "loss": 0.5415, "step": 28267 }, { "epoch": 3.0078740157480315, "grad_norm": 11.1171875, "learning_rate": 9.013359863558842e-06, "loss": 0.5386, "step": 28268 }, { "epoch": 3.007980421366248, "grad_norm": 12.4921875, "learning_rate": 9.013324332006823e-06, "loss": 0.5456, "step": 28269 }, { "epoch": 3.008086826984465, "grad_norm": 11.3125, "learning_rate": 9.013288800454805e-06, "loss": 0.5634, "step": 28270 }, { "epoch": 3.0081932326026815, "grad_norm": 12.9375, "learning_rate": 9.013253268902786e-06, "loss": 0.5743, "step": 28271 }, { "epoch": 3.008299638220898, "grad_norm": 13.0078125, "learning_rate": 9.013217737350768e-06, "loss": 0.5167, "step": 28272 }, { "epoch": 3.008406043839115, "grad_norm": 10.625, "learning_rate": 9.01318220579875e-06, "loss": 0.6232, "step": 28273 }, { "epoch": 3.0085124494573314, "grad_norm": 8.7265625, "learning_rate": 9.013146674246733e-06, "loss": 0.4465, "step": 28274 }, { "epoch": 3.008618855075548, "grad_norm": 12.25, "learning_rate": 9.013111142694715e-06, "loss": 0.5909, "step": 28275 }, { "epoch": 3.0087252606937644, "grad_norm": 13.53125, "learning_rate": 9.013075611142696e-06, "loss": 0.6275, "step": 28276 }, { "epoch": 3.0088316663119814, "grad_norm": 17.296875, "learning_rate": 9.013040079590678e-06, "loss": 0.7628, "step": 28277 }, { "epoch": 3.008938071930198, "grad_norm": 13.625, "learning_rate": 9.01300454803866e-06, "loss": 0.6314, "step": 28278 }, { "epoch": 3.0090444775484144, "grad_norm": 16.171875, "learning_rate": 9.012969016486641e-06, "loss": 0.7737, "step": 28279 }, { "epoch": 3.0091508831666314, "grad_norm": 14.5859375, "learning_rate": 9.012933484934623e-06, "loss": 0.6898, "step": 28280 }, { "epoch": 3.009257288784848, "grad_norm": 12.4765625, "learning_rate": 9.012897953382604e-06, "loss": 0.513, "step": 28281 }, { "epoch": 3.0093636944030644, "grad_norm": 14.203125, "learning_rate": 9.012862421830586e-06, "loss": 0.5921, "step": 28282 }, { "epoch": 3.009470100021281, "grad_norm": 11.3671875, "learning_rate": 9.012826890278567e-06, "loss": 0.4945, "step": 28283 }, { "epoch": 3.009576505639498, "grad_norm": 11.140625, "learning_rate": 9.01279135872655e-06, "loss": 0.5318, "step": 28284 }, { "epoch": 3.0096829112577144, "grad_norm": 13.46875, "learning_rate": 9.012755827174532e-06, "loss": 0.7262, "step": 28285 }, { "epoch": 3.009789316875931, "grad_norm": 10.109375, "learning_rate": 9.012720295622514e-06, "loss": 0.53, "step": 28286 }, { "epoch": 3.009895722494148, "grad_norm": 16.34375, "learning_rate": 9.012684764070495e-06, "loss": 0.6751, "step": 28287 }, { "epoch": 3.0100021281123643, "grad_norm": 8.421875, "learning_rate": 9.012649232518477e-06, "loss": 0.4222, "step": 28288 }, { "epoch": 3.010108533730581, "grad_norm": 11.9765625, "learning_rate": 9.012613700966459e-06, "loss": 0.6213, "step": 28289 }, { "epoch": 3.010214939348798, "grad_norm": 12.4765625, "learning_rate": 9.01257816941444e-06, "loss": 0.6804, "step": 28290 }, { "epoch": 3.0103213449670143, "grad_norm": 12.4921875, "learning_rate": 9.012542637862422e-06, "loss": 0.7523, "step": 28291 }, { "epoch": 3.010427750585231, "grad_norm": 12.140625, "learning_rate": 9.012507106310404e-06, "loss": 0.4428, "step": 28292 }, { "epoch": 3.0105341562034473, "grad_norm": 14.4765625, "learning_rate": 9.012471574758385e-06, "loss": 0.7579, "step": 28293 }, { "epoch": 3.0106405618216643, "grad_norm": 10.640625, "learning_rate": 9.012436043206368e-06, "loss": 0.5106, "step": 28294 }, { "epoch": 3.010746967439881, "grad_norm": 14.828125, "learning_rate": 9.01240051165435e-06, "loss": 0.5853, "step": 28295 }, { "epoch": 3.0108533730580973, "grad_norm": 13.4140625, "learning_rate": 9.012364980102332e-06, "loss": 0.5882, "step": 28296 }, { "epoch": 3.0109597786763143, "grad_norm": 11.6328125, "learning_rate": 9.012329448550313e-06, "loss": 0.6605, "step": 28297 }, { "epoch": 3.0110661842945308, "grad_norm": 8.0390625, "learning_rate": 9.012293916998295e-06, "loss": 0.4417, "step": 28298 }, { "epoch": 3.0111725899127473, "grad_norm": 9.4921875, "learning_rate": 9.012258385446278e-06, "loss": 0.5543, "step": 28299 }, { "epoch": 3.0112789955309642, "grad_norm": 10.25, "learning_rate": 9.01222285389426e-06, "loss": 0.6019, "step": 28300 }, { "epoch": 3.0113854011491807, "grad_norm": 11.5546875, "learning_rate": 9.012187322342241e-06, "loss": 0.5682, "step": 28301 }, { "epoch": 3.0114918067673973, "grad_norm": 10.890625, "learning_rate": 9.012151790790223e-06, "loss": 0.6136, "step": 28302 }, { "epoch": 3.0115982123856138, "grad_norm": 11.4765625, "learning_rate": 9.012116259238205e-06, "loss": 0.4668, "step": 28303 }, { "epoch": 3.0117046180038307, "grad_norm": 11.46875, "learning_rate": 9.012080727686186e-06, "loss": 0.6548, "step": 28304 }, { "epoch": 3.0118110236220472, "grad_norm": 12.3671875, "learning_rate": 9.012045196134168e-06, "loss": 0.6088, "step": 28305 }, { "epoch": 3.0119174292402637, "grad_norm": 16.671875, "learning_rate": 9.01200966458215e-06, "loss": 0.6391, "step": 28306 }, { "epoch": 3.0120238348584807, "grad_norm": 11.3515625, "learning_rate": 9.011974133030131e-06, "loss": 0.642, "step": 28307 }, { "epoch": 3.012130240476697, "grad_norm": 16.625, "learning_rate": 9.011938601478113e-06, "loss": 0.6945, "step": 28308 }, { "epoch": 3.0122366460949137, "grad_norm": 12.1796875, "learning_rate": 9.011903069926096e-06, "loss": 0.6931, "step": 28309 }, { "epoch": 3.0123430517131307, "grad_norm": 15.421875, "learning_rate": 9.011867538374078e-06, "loss": 0.6856, "step": 28310 }, { "epoch": 3.012449457331347, "grad_norm": 14.6171875, "learning_rate": 9.011832006822059e-06, "loss": 0.6697, "step": 28311 }, { "epoch": 3.0125558629495637, "grad_norm": 13.5390625, "learning_rate": 9.01179647527004e-06, "loss": 0.6701, "step": 28312 }, { "epoch": 3.01266226856778, "grad_norm": 10.515625, "learning_rate": 9.011760943718022e-06, "loss": 0.5829, "step": 28313 }, { "epoch": 3.012768674185997, "grad_norm": 14.4140625, "learning_rate": 9.011725412166004e-06, "loss": 0.5534, "step": 28314 }, { "epoch": 3.0128750798042137, "grad_norm": 15.0625, "learning_rate": 9.011689880613986e-06, "loss": 0.6754, "step": 28315 }, { "epoch": 3.01298148542243, "grad_norm": 12.3984375, "learning_rate": 9.011654349061967e-06, "loss": 0.6506, "step": 28316 }, { "epoch": 3.013087891040647, "grad_norm": 13.6328125, "learning_rate": 9.011618817509949e-06, "loss": 0.6521, "step": 28317 }, { "epoch": 3.0131942966588636, "grad_norm": 8.5234375, "learning_rate": 9.01158328595793e-06, "loss": 0.5568, "step": 28318 }, { "epoch": 3.01330070227708, "grad_norm": 16.171875, "learning_rate": 9.011547754405912e-06, "loss": 0.8649, "step": 28319 }, { "epoch": 3.013407107895297, "grad_norm": 10.90625, "learning_rate": 9.011512222853895e-06, "loss": 0.6162, "step": 28320 }, { "epoch": 3.0135135135135136, "grad_norm": 12.6328125, "learning_rate": 9.011476691301877e-06, "loss": 0.5319, "step": 28321 }, { "epoch": 3.01361991913173, "grad_norm": 11.5859375, "learning_rate": 9.011441159749859e-06, "loss": 0.5674, "step": 28322 }, { "epoch": 3.0137263247499466, "grad_norm": 11.6640625, "learning_rate": 9.01140562819784e-06, "loss": 0.546, "step": 28323 }, { "epoch": 3.0138327303681636, "grad_norm": 13.7109375, "learning_rate": 9.011370096645822e-06, "loss": 0.7114, "step": 28324 }, { "epoch": 3.01393913598638, "grad_norm": 17.5, "learning_rate": 9.011334565093805e-06, "loss": 0.7308, "step": 28325 }, { "epoch": 3.0140455416045966, "grad_norm": 18.40625, "learning_rate": 9.011299033541787e-06, "loss": 0.6951, "step": 28326 }, { "epoch": 3.0141519472228135, "grad_norm": 14.3515625, "learning_rate": 9.011263501989768e-06, "loss": 0.5708, "step": 28327 }, { "epoch": 3.01425835284103, "grad_norm": 12.7109375, "learning_rate": 9.01122797043775e-06, "loss": 0.6134, "step": 28328 }, { "epoch": 3.0143647584592466, "grad_norm": 14.2421875, "learning_rate": 9.011192438885731e-06, "loss": 0.6788, "step": 28329 }, { "epoch": 3.014471164077463, "grad_norm": 13.9296875, "learning_rate": 9.011156907333713e-06, "loss": 0.6573, "step": 28330 }, { "epoch": 3.01457756969568, "grad_norm": 12.8984375, "learning_rate": 9.011121375781695e-06, "loss": 0.5766, "step": 28331 }, { "epoch": 3.0146839753138965, "grad_norm": 12.1796875, "learning_rate": 9.011085844229676e-06, "loss": 0.5819, "step": 28332 }, { "epoch": 3.014790380932113, "grad_norm": 24.015625, "learning_rate": 9.011050312677658e-06, "loss": 0.7204, "step": 28333 }, { "epoch": 3.01489678655033, "grad_norm": 19.5625, "learning_rate": 9.01101478112564e-06, "loss": 0.5079, "step": 28334 }, { "epoch": 3.0150031921685465, "grad_norm": 10.515625, "learning_rate": 9.010979249573623e-06, "loss": 0.6708, "step": 28335 }, { "epoch": 3.015109597786763, "grad_norm": 12.921875, "learning_rate": 9.010943718021604e-06, "loss": 0.5059, "step": 28336 }, { "epoch": 3.01521600340498, "grad_norm": 10.8203125, "learning_rate": 9.010908186469586e-06, "loss": 0.541, "step": 28337 }, { "epoch": 3.0153224090231965, "grad_norm": 16.34375, "learning_rate": 9.010872654917568e-06, "loss": 0.7839, "step": 28338 }, { "epoch": 3.015428814641413, "grad_norm": 10.2578125, "learning_rate": 9.01083712336555e-06, "loss": 0.5282, "step": 28339 }, { "epoch": 3.0155352202596295, "grad_norm": 13.296875, "learning_rate": 9.01080159181353e-06, "loss": 0.6178, "step": 28340 }, { "epoch": 3.0156416258778465, "grad_norm": 18.34375, "learning_rate": 9.010766060261512e-06, "loss": 0.8236, "step": 28341 }, { "epoch": 3.015748031496063, "grad_norm": 13.21875, "learning_rate": 9.010730528709494e-06, "loss": 0.6107, "step": 28342 }, { "epoch": 3.0158544371142795, "grad_norm": 11.7890625, "learning_rate": 9.010694997157476e-06, "loss": 0.6036, "step": 28343 }, { "epoch": 3.0159608427324964, "grad_norm": 12.5546875, "learning_rate": 9.010659465605457e-06, "loss": 0.657, "step": 28344 }, { "epoch": 3.016067248350713, "grad_norm": 14.578125, "learning_rate": 9.01062393405344e-06, "loss": 0.6494, "step": 28345 }, { "epoch": 3.0161736539689294, "grad_norm": 11.75, "learning_rate": 9.010588402501422e-06, "loss": 0.5749, "step": 28346 }, { "epoch": 3.0162800595871464, "grad_norm": 13.5703125, "learning_rate": 9.010552870949404e-06, "loss": 0.6757, "step": 28347 }, { "epoch": 3.016386465205363, "grad_norm": 11.390625, "learning_rate": 9.010517339397385e-06, "loss": 0.5004, "step": 28348 }, { "epoch": 3.0164928708235794, "grad_norm": 10.0234375, "learning_rate": 9.010481807845367e-06, "loss": 0.4584, "step": 28349 }, { "epoch": 3.016599276441796, "grad_norm": 11.0390625, "learning_rate": 9.01044627629335e-06, "loss": 0.5752, "step": 28350 }, { "epoch": 3.016705682060013, "grad_norm": 16.375, "learning_rate": 9.010410744741332e-06, "loss": 0.6341, "step": 28351 }, { "epoch": 3.0168120876782294, "grad_norm": 10.3359375, "learning_rate": 9.010375213189313e-06, "loss": 0.575, "step": 28352 }, { "epoch": 3.016918493296446, "grad_norm": 11.703125, "learning_rate": 9.010339681637295e-06, "loss": 0.5559, "step": 28353 }, { "epoch": 3.017024898914663, "grad_norm": 12.8359375, "learning_rate": 9.010304150085277e-06, "loss": 0.6511, "step": 28354 }, { "epoch": 3.0171313045328794, "grad_norm": 8.453125, "learning_rate": 9.010268618533258e-06, "loss": 0.505, "step": 28355 }, { "epoch": 3.017237710151096, "grad_norm": 11.796875, "learning_rate": 9.01023308698124e-06, "loss": 0.6177, "step": 28356 }, { "epoch": 3.017344115769313, "grad_norm": 13.59375, "learning_rate": 9.010197555429222e-06, "loss": 0.5372, "step": 28357 }, { "epoch": 3.0174505213875293, "grad_norm": 14.890625, "learning_rate": 9.010162023877203e-06, "loss": 0.654, "step": 28358 }, { "epoch": 3.017556927005746, "grad_norm": 11.671875, "learning_rate": 9.010126492325185e-06, "loss": 0.5093, "step": 28359 }, { "epoch": 3.0176633326239624, "grad_norm": 13.4765625, "learning_rate": 9.010090960773166e-06, "loss": 0.6671, "step": 28360 }, { "epoch": 3.0177697382421793, "grad_norm": 14.5859375, "learning_rate": 9.01005542922115e-06, "loss": 0.7345, "step": 28361 }, { "epoch": 3.017876143860396, "grad_norm": 13.8671875, "learning_rate": 9.010019897669131e-06, "loss": 0.5308, "step": 28362 }, { "epoch": 3.0179825494786123, "grad_norm": 9.4375, "learning_rate": 9.009984366117113e-06, "loss": 0.5209, "step": 28363 }, { "epoch": 3.0180889550968293, "grad_norm": 14.0078125, "learning_rate": 9.009948834565094e-06, "loss": 0.6589, "step": 28364 }, { "epoch": 3.018195360715046, "grad_norm": 11.0703125, "learning_rate": 9.009913303013076e-06, "loss": 0.6219, "step": 28365 }, { "epoch": 3.0183017663332623, "grad_norm": 14.234375, "learning_rate": 9.00987777146106e-06, "loss": 0.6759, "step": 28366 }, { "epoch": 3.0184081719514793, "grad_norm": 11.234375, "learning_rate": 9.009842239909041e-06, "loss": 0.5366, "step": 28367 }, { "epoch": 3.0185145775696958, "grad_norm": 8.953125, "learning_rate": 9.009806708357023e-06, "loss": 0.5586, "step": 28368 }, { "epoch": 3.0186209831879123, "grad_norm": 10.75, "learning_rate": 9.009771176805004e-06, "loss": 0.6507, "step": 28369 }, { "epoch": 3.018727388806129, "grad_norm": 9.109375, "learning_rate": 9.009735645252984e-06, "loss": 0.6885, "step": 28370 }, { "epoch": 3.0188337944243457, "grad_norm": 12.2734375, "learning_rate": 9.009700113700967e-06, "loss": 0.6378, "step": 28371 }, { "epoch": 3.0189402000425622, "grad_norm": 11.15625, "learning_rate": 9.009664582148949e-06, "loss": 0.6095, "step": 28372 }, { "epoch": 3.0190466056607788, "grad_norm": 12.75, "learning_rate": 9.00962905059693e-06, "loss": 0.6358, "step": 28373 }, { "epoch": 3.0191530112789957, "grad_norm": 13.78125, "learning_rate": 9.009593519044912e-06, "loss": 0.6503, "step": 28374 }, { "epoch": 3.0192594168972122, "grad_norm": 12.34375, "learning_rate": 9.009557987492894e-06, "loss": 0.6657, "step": 28375 }, { "epoch": 3.0193658225154287, "grad_norm": 11.1484375, "learning_rate": 9.009522455940877e-06, "loss": 0.495, "step": 28376 }, { "epoch": 3.0194722281336452, "grad_norm": 13.03125, "learning_rate": 9.009486924388859e-06, "loss": 0.6565, "step": 28377 }, { "epoch": 3.019578633751862, "grad_norm": 13.09375, "learning_rate": 9.00945139283684e-06, "loss": 0.6634, "step": 28378 }, { "epoch": 3.0196850393700787, "grad_norm": 13.6484375, "learning_rate": 9.009415861284822e-06, "loss": 0.7568, "step": 28379 }, { "epoch": 3.019791444988295, "grad_norm": 14.453125, "learning_rate": 9.009380329732804e-06, "loss": 0.7549, "step": 28380 }, { "epoch": 3.019897850606512, "grad_norm": 9.640625, "learning_rate": 9.009344798180785e-06, "loss": 0.459, "step": 28381 }, { "epoch": 3.0200042562247287, "grad_norm": 8.671875, "learning_rate": 9.009309266628767e-06, "loss": 0.5204, "step": 28382 }, { "epoch": 3.020110661842945, "grad_norm": 22.8125, "learning_rate": 9.009273735076748e-06, "loss": 0.9065, "step": 28383 }, { "epoch": 3.020217067461162, "grad_norm": 12.7109375, "learning_rate": 9.00923820352473e-06, "loss": 0.6612, "step": 28384 }, { "epoch": 3.0203234730793787, "grad_norm": 19.703125, "learning_rate": 9.009202671972712e-06, "loss": 0.8962, "step": 28385 }, { "epoch": 3.020429878697595, "grad_norm": 12.0234375, "learning_rate": 9.009167140420695e-06, "loss": 0.5446, "step": 28386 }, { "epoch": 3.0205362843158117, "grad_norm": 10.7109375, "learning_rate": 9.009131608868676e-06, "loss": 0.5158, "step": 28387 }, { "epoch": 3.0206426899340286, "grad_norm": 14.5078125, "learning_rate": 9.009096077316658e-06, "loss": 0.7224, "step": 28388 }, { "epoch": 3.020749095552245, "grad_norm": 11.625, "learning_rate": 9.00906054576464e-06, "loss": 0.4921, "step": 28389 }, { "epoch": 3.0208555011704616, "grad_norm": 10.296875, "learning_rate": 9.009025014212621e-06, "loss": 0.57, "step": 28390 }, { "epoch": 3.0209619067886786, "grad_norm": 14.328125, "learning_rate": 9.008989482660603e-06, "loss": 0.5183, "step": 28391 }, { "epoch": 3.021068312406895, "grad_norm": 10.3828125, "learning_rate": 9.008953951108586e-06, "loss": 0.4938, "step": 28392 }, { "epoch": 3.0211747180251116, "grad_norm": 11.7734375, "learning_rate": 9.008918419556568e-06, "loss": 0.6162, "step": 28393 }, { "epoch": 3.0212811236433286, "grad_norm": 13.3671875, "learning_rate": 9.00888288800455e-06, "loss": 0.6318, "step": 28394 }, { "epoch": 3.021387529261545, "grad_norm": 14.6171875, "learning_rate": 9.008847356452531e-06, "loss": 0.7339, "step": 28395 }, { "epoch": 3.0214939348797616, "grad_norm": 13.515625, "learning_rate": 9.008811824900513e-06, "loss": 0.6389, "step": 28396 }, { "epoch": 3.021600340497978, "grad_norm": 11.921875, "learning_rate": 9.008776293348494e-06, "loss": 0.6456, "step": 28397 }, { "epoch": 3.021706746116195, "grad_norm": 11.3046875, "learning_rate": 9.008740761796476e-06, "loss": 0.7126, "step": 28398 }, { "epoch": 3.0218131517344116, "grad_norm": 14.109375, "learning_rate": 9.008705230244457e-06, "loss": 0.6808, "step": 28399 }, { "epoch": 3.021919557352628, "grad_norm": 10.921875, "learning_rate": 9.008669698692439e-06, "loss": 0.6081, "step": 28400 }, { "epoch": 3.022025962970845, "grad_norm": 11.4453125, "learning_rate": 9.00863416714042e-06, "loss": 0.5557, "step": 28401 }, { "epoch": 3.0221323685890615, "grad_norm": 13.5390625, "learning_rate": 9.008598635588404e-06, "loss": 0.5526, "step": 28402 }, { "epoch": 3.022238774207278, "grad_norm": 11.984375, "learning_rate": 9.008563104036386e-06, "loss": 0.5405, "step": 28403 }, { "epoch": 3.022345179825495, "grad_norm": 15.3203125, "learning_rate": 9.008527572484367e-06, "loss": 0.51, "step": 28404 }, { "epoch": 3.0224515854437115, "grad_norm": 17.546875, "learning_rate": 9.008492040932349e-06, "loss": 0.7478, "step": 28405 }, { "epoch": 3.022557991061928, "grad_norm": 11.2109375, "learning_rate": 9.00845650938033e-06, "loss": 0.4795, "step": 28406 }, { "epoch": 3.0226643966801445, "grad_norm": 10.7734375, "learning_rate": 9.008420977828312e-06, "loss": 0.5142, "step": 28407 }, { "epoch": 3.0227708022983615, "grad_norm": 18.0, "learning_rate": 9.008385446276294e-06, "loss": 0.7873, "step": 28408 }, { "epoch": 3.022877207916578, "grad_norm": 15.1015625, "learning_rate": 9.008349914724275e-06, "loss": 0.7027, "step": 28409 }, { "epoch": 3.0229836135347945, "grad_norm": 14.7890625, "learning_rate": 9.008314383172257e-06, "loss": 0.6515, "step": 28410 }, { "epoch": 3.0230900191530115, "grad_norm": 14.6015625, "learning_rate": 9.008278851620238e-06, "loss": 0.5619, "step": 28411 }, { "epoch": 3.023196424771228, "grad_norm": 13.5390625, "learning_rate": 9.008243320068222e-06, "loss": 0.643, "step": 28412 }, { "epoch": 3.0233028303894445, "grad_norm": 11.71875, "learning_rate": 9.008207788516203e-06, "loss": 0.518, "step": 28413 }, { "epoch": 3.0234092360076614, "grad_norm": 13.25, "learning_rate": 9.008172256964185e-06, "loss": 0.6282, "step": 28414 }, { "epoch": 3.023515641625878, "grad_norm": 11.4375, "learning_rate": 9.008136725412167e-06, "loss": 0.6937, "step": 28415 }, { "epoch": 3.0236220472440944, "grad_norm": 13.296875, "learning_rate": 9.008101193860148e-06, "loss": 0.6571, "step": 28416 }, { "epoch": 3.023728452862311, "grad_norm": 11.0625, "learning_rate": 9.008065662308131e-06, "loss": 0.5266, "step": 28417 }, { "epoch": 3.023834858480528, "grad_norm": 11.8046875, "learning_rate": 9.008030130756113e-06, "loss": 0.4997, "step": 28418 }, { "epoch": 3.0239412640987444, "grad_norm": 9.7265625, "learning_rate": 9.007994599204095e-06, "loss": 0.5026, "step": 28419 }, { "epoch": 3.024047669716961, "grad_norm": 13.5859375, "learning_rate": 9.007959067652076e-06, "loss": 0.6715, "step": 28420 }, { "epoch": 3.024154075335178, "grad_norm": 17.0, "learning_rate": 9.007923536100058e-06, "loss": 0.8436, "step": 28421 }, { "epoch": 3.0242604809533944, "grad_norm": 11.4375, "learning_rate": 9.00788800454804e-06, "loss": 0.5108, "step": 28422 }, { "epoch": 3.024366886571611, "grad_norm": 12.6875, "learning_rate": 9.007852472996021e-06, "loss": 0.6279, "step": 28423 }, { "epoch": 3.0244732921898274, "grad_norm": 10.4609375, "learning_rate": 9.007816941444003e-06, "loss": 0.5678, "step": 28424 }, { "epoch": 3.0245796978080444, "grad_norm": 10.7734375, "learning_rate": 9.007781409891984e-06, "loss": 0.5148, "step": 28425 }, { "epoch": 3.024686103426261, "grad_norm": 12.7890625, "learning_rate": 9.007745878339966e-06, "loss": 0.6111, "step": 28426 }, { "epoch": 3.0247925090444774, "grad_norm": 12.7265625, "learning_rate": 9.00771034678795e-06, "loss": 0.5914, "step": 28427 }, { "epoch": 3.0248989146626943, "grad_norm": 12.1875, "learning_rate": 9.00767481523593e-06, "loss": 0.6777, "step": 28428 }, { "epoch": 3.025005320280911, "grad_norm": 14.96875, "learning_rate": 9.007639283683912e-06, "loss": 0.7394, "step": 28429 }, { "epoch": 3.0251117258991274, "grad_norm": 11.890625, "learning_rate": 9.007603752131894e-06, "loss": 0.6083, "step": 28430 }, { "epoch": 3.0252181315173443, "grad_norm": 9.5234375, "learning_rate": 9.007568220579876e-06, "loss": 0.5581, "step": 28431 }, { "epoch": 3.025324537135561, "grad_norm": 10.875, "learning_rate": 9.007532689027857e-06, "loss": 0.5168, "step": 28432 }, { "epoch": 3.0254309427537773, "grad_norm": 13.7578125, "learning_rate": 9.007497157475839e-06, "loss": 0.6682, "step": 28433 }, { "epoch": 3.025537348371994, "grad_norm": 14.046875, "learning_rate": 9.00746162592382e-06, "loss": 0.7252, "step": 28434 }, { "epoch": 3.025643753990211, "grad_norm": 16.328125, "learning_rate": 9.007426094371802e-06, "loss": 0.5597, "step": 28435 }, { "epoch": 3.0257501596084273, "grad_norm": 12.703125, "learning_rate": 9.007390562819784e-06, "loss": 0.581, "step": 28436 }, { "epoch": 3.025856565226644, "grad_norm": 13.21875, "learning_rate": 9.007355031267765e-06, "loss": 0.5807, "step": 28437 }, { "epoch": 3.0259629708448608, "grad_norm": 11.3984375, "learning_rate": 9.007319499715749e-06, "loss": 0.6544, "step": 28438 }, { "epoch": 3.0260693764630773, "grad_norm": 14.6875, "learning_rate": 9.00728396816373e-06, "loss": 0.6789, "step": 28439 }, { "epoch": 3.026175782081294, "grad_norm": 19.875, "learning_rate": 9.007248436611712e-06, "loss": 0.894, "step": 28440 }, { "epoch": 3.0262821876995107, "grad_norm": 9.65625, "learning_rate": 9.007212905059693e-06, "loss": 0.5083, "step": 28441 }, { "epoch": 3.0263885933177272, "grad_norm": 15.296875, "learning_rate": 9.007177373507675e-06, "loss": 0.7564, "step": 28442 }, { "epoch": 3.0264949989359438, "grad_norm": 14.1484375, "learning_rate": 9.007141841955658e-06, "loss": 0.5964, "step": 28443 }, { "epoch": 3.0266014045541603, "grad_norm": 14.109375, "learning_rate": 9.00710631040364e-06, "loss": 0.807, "step": 28444 }, { "epoch": 3.0267078101723772, "grad_norm": 14.2890625, "learning_rate": 9.007070778851622e-06, "loss": 0.7233, "step": 28445 }, { "epoch": 3.0268142157905937, "grad_norm": 10.1015625, "learning_rate": 9.007035247299603e-06, "loss": 0.4341, "step": 28446 }, { "epoch": 3.0269206214088102, "grad_norm": 11.6953125, "learning_rate": 9.006999715747585e-06, "loss": 0.4909, "step": 28447 }, { "epoch": 3.027027027027027, "grad_norm": 10.375, "learning_rate": 9.006964184195566e-06, "loss": 0.4743, "step": 28448 }, { "epoch": 3.0271334326452437, "grad_norm": 13.6875, "learning_rate": 9.006928652643548e-06, "loss": 0.608, "step": 28449 }, { "epoch": 3.02723983826346, "grad_norm": 10.890625, "learning_rate": 9.00689312109153e-06, "loss": 0.6865, "step": 28450 }, { "epoch": 3.027346243881677, "grad_norm": 13.9375, "learning_rate": 9.006857589539511e-06, "loss": 0.5721, "step": 28451 }, { "epoch": 3.0274526494998937, "grad_norm": 19.046875, "learning_rate": 9.006822057987493e-06, "loss": 0.8878, "step": 28452 }, { "epoch": 3.02755905511811, "grad_norm": 10.328125, "learning_rate": 9.006786526435476e-06, "loss": 0.5653, "step": 28453 }, { "epoch": 3.0276654607363267, "grad_norm": 10.5703125, "learning_rate": 9.006750994883458e-06, "loss": 0.5704, "step": 28454 }, { "epoch": 3.0277718663545437, "grad_norm": 12.046875, "learning_rate": 9.00671546333144e-06, "loss": 0.4952, "step": 28455 }, { "epoch": 3.02787827197276, "grad_norm": 15.1171875, "learning_rate": 9.006679931779421e-06, "loss": 0.6183, "step": 28456 }, { "epoch": 3.0279846775909767, "grad_norm": 9.5703125, "learning_rate": 9.006644400227403e-06, "loss": 0.4624, "step": 28457 }, { "epoch": 3.0280910832091936, "grad_norm": 13.0625, "learning_rate": 9.006608868675386e-06, "loss": 0.6297, "step": 28458 }, { "epoch": 3.02819748882741, "grad_norm": 9.6171875, "learning_rate": 9.006573337123366e-06, "loss": 0.4582, "step": 28459 }, { "epoch": 3.0283038944456266, "grad_norm": 11.3046875, "learning_rate": 9.006537805571347e-06, "loss": 0.5558, "step": 28460 }, { "epoch": 3.028410300063843, "grad_norm": 11.203125, "learning_rate": 9.006502274019329e-06, "loss": 0.654, "step": 28461 }, { "epoch": 3.02851670568206, "grad_norm": 10.8671875, "learning_rate": 9.00646674246731e-06, "loss": 0.6279, "step": 28462 }, { "epoch": 3.0286231113002766, "grad_norm": 18.265625, "learning_rate": 9.006431210915294e-06, "loss": 0.7681, "step": 28463 }, { "epoch": 3.028729516918493, "grad_norm": 11.828125, "learning_rate": 9.006395679363275e-06, "loss": 0.7028, "step": 28464 }, { "epoch": 3.02883592253671, "grad_norm": 13.28125, "learning_rate": 9.006360147811257e-06, "loss": 0.5318, "step": 28465 }, { "epoch": 3.0289423281549266, "grad_norm": 12.2265625, "learning_rate": 9.006324616259239e-06, "loss": 0.704, "step": 28466 }, { "epoch": 3.029048733773143, "grad_norm": 15.46875, "learning_rate": 9.00628908470722e-06, "loss": 0.6264, "step": 28467 }, { "epoch": 3.02915513939136, "grad_norm": 11.625, "learning_rate": 9.006253553155204e-06, "loss": 0.7313, "step": 28468 }, { "epoch": 3.0292615450095766, "grad_norm": 11.34375, "learning_rate": 9.006218021603185e-06, "loss": 0.525, "step": 28469 }, { "epoch": 3.029367950627793, "grad_norm": 13.8984375, "learning_rate": 9.006182490051167e-06, "loss": 0.6732, "step": 28470 }, { "epoch": 3.0294743562460096, "grad_norm": 16.1875, "learning_rate": 9.006146958499148e-06, "loss": 0.6718, "step": 28471 }, { "epoch": 3.0295807618642265, "grad_norm": 12.78125, "learning_rate": 9.00611142694713e-06, "loss": 0.5969, "step": 28472 }, { "epoch": 3.029687167482443, "grad_norm": 15.671875, "learning_rate": 9.006075895395112e-06, "loss": 0.6341, "step": 28473 }, { "epoch": 3.0297935731006596, "grad_norm": 12.4609375, "learning_rate": 9.006040363843093e-06, "loss": 0.6242, "step": 28474 }, { "epoch": 3.0298999787188765, "grad_norm": 12.1796875, "learning_rate": 9.006004832291075e-06, "loss": 0.6165, "step": 28475 }, { "epoch": 3.030006384337093, "grad_norm": 12.2734375, "learning_rate": 9.005969300739056e-06, "loss": 0.5637, "step": 28476 }, { "epoch": 3.0301127899553095, "grad_norm": 17.3125, "learning_rate": 9.005933769187038e-06, "loss": 0.7679, "step": 28477 }, { "epoch": 3.0302191955735265, "grad_norm": 11.234375, "learning_rate": 9.00589823763502e-06, "loss": 0.5573, "step": 28478 }, { "epoch": 3.030325601191743, "grad_norm": 10.9921875, "learning_rate": 9.005862706083003e-06, "loss": 0.6004, "step": 28479 }, { "epoch": 3.0304320068099595, "grad_norm": 12.078125, "learning_rate": 9.005827174530985e-06, "loss": 0.6368, "step": 28480 }, { "epoch": 3.030538412428176, "grad_norm": 11.1171875, "learning_rate": 9.005791642978966e-06, "loss": 0.5726, "step": 28481 }, { "epoch": 3.030644818046393, "grad_norm": 13.5859375, "learning_rate": 9.005756111426948e-06, "loss": 0.5521, "step": 28482 }, { "epoch": 3.0307512236646095, "grad_norm": 12.8125, "learning_rate": 9.00572057987493e-06, "loss": 0.7568, "step": 28483 }, { "epoch": 3.030857629282826, "grad_norm": 12.5625, "learning_rate": 9.005685048322913e-06, "loss": 0.6299, "step": 28484 }, { "epoch": 3.030964034901043, "grad_norm": 13.265625, "learning_rate": 9.005649516770894e-06, "loss": 0.628, "step": 28485 }, { "epoch": 3.0310704405192594, "grad_norm": 11.4140625, "learning_rate": 9.005613985218876e-06, "loss": 0.4792, "step": 28486 }, { "epoch": 3.031176846137476, "grad_norm": 10.3828125, "learning_rate": 9.005578453666857e-06, "loss": 0.5467, "step": 28487 }, { "epoch": 3.031283251755693, "grad_norm": 14.9140625, "learning_rate": 9.005542922114839e-06, "loss": 0.6714, "step": 28488 }, { "epoch": 3.0313896573739094, "grad_norm": 12.3203125, "learning_rate": 9.00550739056282e-06, "loss": 0.6809, "step": 28489 }, { "epoch": 3.031496062992126, "grad_norm": 11.640625, "learning_rate": 9.005471859010802e-06, "loss": 0.6216, "step": 28490 }, { "epoch": 3.0316024686103424, "grad_norm": 14.671875, "learning_rate": 9.005436327458784e-06, "loss": 0.6823, "step": 28491 }, { "epoch": 3.0317088742285594, "grad_norm": 13.515625, "learning_rate": 9.005400795906766e-06, "loss": 0.5569, "step": 28492 }, { "epoch": 3.031815279846776, "grad_norm": 15.3828125, "learning_rate": 9.005365264354747e-06, "loss": 0.7907, "step": 28493 }, { "epoch": 3.0319216854649924, "grad_norm": 11.90625, "learning_rate": 9.00532973280273e-06, "loss": 0.7045, "step": 28494 }, { "epoch": 3.0320280910832094, "grad_norm": 16.46875, "learning_rate": 9.005294201250712e-06, "loss": 0.6875, "step": 28495 }, { "epoch": 3.032134496701426, "grad_norm": 11.03125, "learning_rate": 9.005258669698694e-06, "loss": 0.5844, "step": 28496 }, { "epoch": 3.0322409023196424, "grad_norm": 13.5390625, "learning_rate": 9.005223138146675e-06, "loss": 0.5553, "step": 28497 }, { "epoch": 3.032347307937859, "grad_norm": 10.3359375, "learning_rate": 9.005187606594657e-06, "loss": 0.5653, "step": 28498 }, { "epoch": 3.032453713556076, "grad_norm": 13.3125, "learning_rate": 9.005152075042638e-06, "loss": 0.6072, "step": 28499 }, { "epoch": 3.0325601191742924, "grad_norm": 13.6953125, "learning_rate": 9.00511654349062e-06, "loss": 0.6848, "step": 28500 }, { "epoch": 3.032666524792509, "grad_norm": 12.5859375, "learning_rate": 9.005081011938602e-06, "loss": 0.6302, "step": 28501 }, { "epoch": 3.032772930410726, "grad_norm": 11.5, "learning_rate": 9.005045480386583e-06, "loss": 0.443, "step": 28502 }, { "epoch": 3.0328793360289423, "grad_norm": 11.7734375, "learning_rate": 9.005009948834565e-06, "loss": 0.615, "step": 28503 }, { "epoch": 3.032985741647159, "grad_norm": 13.21875, "learning_rate": 9.004974417282548e-06, "loss": 0.5976, "step": 28504 }, { "epoch": 3.033092147265376, "grad_norm": 11.6328125, "learning_rate": 9.00493888573053e-06, "loss": 0.6078, "step": 28505 }, { "epoch": 3.0331985528835923, "grad_norm": 12.9375, "learning_rate": 9.004903354178511e-06, "loss": 0.523, "step": 28506 }, { "epoch": 3.033304958501809, "grad_norm": 11.0078125, "learning_rate": 9.004867822626493e-06, "loss": 0.6944, "step": 28507 }, { "epoch": 3.0334113641200253, "grad_norm": 14.1171875, "learning_rate": 9.004832291074475e-06, "loss": 0.6231, "step": 28508 }, { "epoch": 3.0335177697382423, "grad_norm": 17.046875, "learning_rate": 9.004796759522456e-06, "loss": 0.8378, "step": 28509 }, { "epoch": 3.033624175356459, "grad_norm": 11.3515625, "learning_rate": 9.00476122797044e-06, "loss": 0.5826, "step": 28510 }, { "epoch": 3.0337305809746753, "grad_norm": 14.921875, "learning_rate": 9.004725696418421e-06, "loss": 0.6056, "step": 28511 }, { "epoch": 3.0338369865928922, "grad_norm": 11.796875, "learning_rate": 9.004690164866403e-06, "loss": 0.5503, "step": 28512 }, { "epoch": 3.0339433922111088, "grad_norm": 10.9296875, "learning_rate": 9.004654633314384e-06, "loss": 0.6426, "step": 28513 }, { "epoch": 3.0340497978293253, "grad_norm": 12.3125, "learning_rate": 9.004619101762366e-06, "loss": 0.5901, "step": 28514 }, { "epoch": 3.034156203447542, "grad_norm": 12.953125, "learning_rate": 9.004583570210348e-06, "loss": 0.6884, "step": 28515 }, { "epoch": 3.0342626090657587, "grad_norm": 11.5859375, "learning_rate": 9.00454803865833e-06, "loss": 0.6358, "step": 28516 }, { "epoch": 3.0343690146839752, "grad_norm": 15.484375, "learning_rate": 9.00451250710631e-06, "loss": 0.6912, "step": 28517 }, { "epoch": 3.0344754203021918, "grad_norm": 11.140625, "learning_rate": 9.004476975554292e-06, "loss": 0.5775, "step": 28518 }, { "epoch": 3.0345818259204087, "grad_norm": 15.71875, "learning_rate": 9.004441444002274e-06, "loss": 0.7913, "step": 28519 }, { "epoch": 3.034688231538625, "grad_norm": 11.2265625, "learning_rate": 9.004405912450257e-06, "loss": 0.5559, "step": 28520 }, { "epoch": 3.0347946371568417, "grad_norm": 15.046875, "learning_rate": 9.004370380898239e-06, "loss": 0.6292, "step": 28521 }, { "epoch": 3.0349010427750587, "grad_norm": 14.0234375, "learning_rate": 9.00433484934622e-06, "loss": 0.8343, "step": 28522 }, { "epoch": 3.035007448393275, "grad_norm": 11.171875, "learning_rate": 9.004299317794202e-06, "loss": 0.6209, "step": 28523 }, { "epoch": 3.0351138540114917, "grad_norm": 10.9609375, "learning_rate": 9.004263786242184e-06, "loss": 0.6792, "step": 28524 }, { "epoch": 3.0352202596297087, "grad_norm": 13.984375, "learning_rate": 9.004228254690165e-06, "loss": 0.7355, "step": 28525 }, { "epoch": 3.035326665247925, "grad_norm": 11.4921875, "learning_rate": 9.004192723138147e-06, "loss": 0.5865, "step": 28526 }, { "epoch": 3.0354330708661417, "grad_norm": 10.671875, "learning_rate": 9.004157191586129e-06, "loss": 0.5444, "step": 28527 }, { "epoch": 3.035539476484358, "grad_norm": 12.25, "learning_rate": 9.00412166003411e-06, "loss": 0.5923, "step": 28528 }, { "epoch": 3.035645882102575, "grad_norm": 12.90625, "learning_rate": 9.004086128482092e-06, "loss": 0.613, "step": 28529 }, { "epoch": 3.0357522877207916, "grad_norm": 14.9453125, "learning_rate": 9.004050596930075e-06, "loss": 0.687, "step": 28530 }, { "epoch": 3.035858693339008, "grad_norm": 10.9609375, "learning_rate": 9.004015065378057e-06, "loss": 0.623, "step": 28531 }, { "epoch": 3.035965098957225, "grad_norm": 19.4375, "learning_rate": 9.003979533826038e-06, "loss": 0.8685, "step": 28532 }, { "epoch": 3.0360715045754416, "grad_norm": 12.7578125, "learning_rate": 9.00394400227402e-06, "loss": 0.6364, "step": 28533 }, { "epoch": 3.036177910193658, "grad_norm": 12.6328125, "learning_rate": 9.003908470722001e-06, "loss": 0.7068, "step": 28534 }, { "epoch": 3.036284315811875, "grad_norm": 12.7421875, "learning_rate": 9.003872939169985e-06, "loss": 0.6122, "step": 28535 }, { "epoch": 3.0363907214300916, "grad_norm": 13.15625, "learning_rate": 9.003837407617966e-06, "loss": 0.5634, "step": 28536 }, { "epoch": 3.036497127048308, "grad_norm": 11.6953125, "learning_rate": 9.003801876065948e-06, "loss": 0.5802, "step": 28537 }, { "epoch": 3.0366035326665246, "grad_norm": 12.390625, "learning_rate": 9.00376634451393e-06, "loss": 0.622, "step": 28538 }, { "epoch": 3.0367099382847416, "grad_norm": 13.984375, "learning_rate": 9.003730812961911e-06, "loss": 0.7326, "step": 28539 }, { "epoch": 3.036816343902958, "grad_norm": 10.140625, "learning_rate": 9.003695281409893e-06, "loss": 0.5291, "step": 28540 }, { "epoch": 3.0369227495211746, "grad_norm": 13.6015625, "learning_rate": 9.003659749857874e-06, "loss": 0.6315, "step": 28541 }, { "epoch": 3.0370291551393915, "grad_norm": 10.0703125, "learning_rate": 9.003624218305856e-06, "loss": 0.6134, "step": 28542 }, { "epoch": 3.037135560757608, "grad_norm": 11.6640625, "learning_rate": 9.003588686753838e-06, "loss": 0.6603, "step": 28543 }, { "epoch": 3.0372419663758246, "grad_norm": 13.9453125, "learning_rate": 9.00355315520182e-06, "loss": 0.8046, "step": 28544 }, { "epoch": 3.0373483719940415, "grad_norm": 14.8046875, "learning_rate": 9.003517623649803e-06, "loss": 0.7741, "step": 28545 }, { "epoch": 3.037454777612258, "grad_norm": 11.0703125, "learning_rate": 9.003482092097784e-06, "loss": 0.5842, "step": 28546 }, { "epoch": 3.0375611832304745, "grad_norm": 12.953125, "learning_rate": 9.003446560545766e-06, "loss": 0.6162, "step": 28547 }, { "epoch": 3.037667588848691, "grad_norm": 10.578125, "learning_rate": 9.003411028993747e-06, "loss": 0.5156, "step": 28548 }, { "epoch": 3.037773994466908, "grad_norm": 8.765625, "learning_rate": 9.003375497441729e-06, "loss": 0.6058, "step": 28549 }, { "epoch": 3.0378804000851245, "grad_norm": 15.7734375, "learning_rate": 9.00333996588971e-06, "loss": 0.6046, "step": 28550 }, { "epoch": 3.037986805703341, "grad_norm": 15.1015625, "learning_rate": 9.003304434337692e-06, "loss": 0.689, "step": 28551 }, { "epoch": 3.038093211321558, "grad_norm": 11.65625, "learning_rate": 9.003268902785674e-06, "loss": 0.6772, "step": 28552 }, { "epoch": 3.0381996169397745, "grad_norm": 12.203125, "learning_rate": 9.003233371233655e-06, "loss": 0.618, "step": 28553 }, { "epoch": 3.038306022557991, "grad_norm": 12.7265625, "learning_rate": 9.003197839681637e-06, "loss": 0.6624, "step": 28554 }, { "epoch": 3.0384124281762075, "grad_norm": 10.21875, "learning_rate": 9.003162308129619e-06, "loss": 0.6192, "step": 28555 }, { "epoch": 3.0385188337944244, "grad_norm": 10.3984375, "learning_rate": 9.003126776577602e-06, "loss": 0.5989, "step": 28556 }, { "epoch": 3.038625239412641, "grad_norm": 9.515625, "learning_rate": 9.003091245025584e-06, "loss": 0.5556, "step": 28557 }, { "epoch": 3.0387316450308575, "grad_norm": 15.7109375, "learning_rate": 9.003055713473565e-06, "loss": 0.7646, "step": 28558 }, { "epoch": 3.0388380506490744, "grad_norm": 10.875, "learning_rate": 9.003020181921547e-06, "loss": 0.5479, "step": 28559 }, { "epoch": 3.038944456267291, "grad_norm": 9.46875, "learning_rate": 9.002984650369528e-06, "loss": 0.3866, "step": 28560 }, { "epoch": 3.0390508618855074, "grad_norm": 10.4765625, "learning_rate": 9.002949118817512e-06, "loss": 0.4626, "step": 28561 }, { "epoch": 3.0391572675037244, "grad_norm": 13.515625, "learning_rate": 9.002913587265493e-06, "loss": 0.6677, "step": 28562 }, { "epoch": 3.039263673121941, "grad_norm": 9.1015625, "learning_rate": 9.002878055713475e-06, "loss": 0.4589, "step": 28563 }, { "epoch": 3.0393700787401574, "grad_norm": 9.640625, "learning_rate": 9.002842524161456e-06, "loss": 0.5139, "step": 28564 }, { "epoch": 3.039476484358374, "grad_norm": 12.6875, "learning_rate": 9.002806992609438e-06, "loss": 0.6287, "step": 28565 }, { "epoch": 3.039582889976591, "grad_norm": 10.9296875, "learning_rate": 9.00277146105742e-06, "loss": 0.5577, "step": 28566 }, { "epoch": 3.0396892955948074, "grad_norm": 10.6484375, "learning_rate": 9.002735929505401e-06, "loss": 0.5228, "step": 28567 }, { "epoch": 3.039795701213024, "grad_norm": 10.1640625, "learning_rate": 9.002700397953383e-06, "loss": 0.5149, "step": 28568 }, { "epoch": 3.039902106831241, "grad_norm": 12.875, "learning_rate": 9.002664866401364e-06, "loss": 0.5975, "step": 28569 }, { "epoch": 3.0400085124494574, "grad_norm": 13.2109375, "learning_rate": 9.002629334849346e-06, "loss": 0.6942, "step": 28570 }, { "epoch": 3.040114918067674, "grad_norm": 16.71875, "learning_rate": 9.00259380329733e-06, "loss": 0.5592, "step": 28571 }, { "epoch": 3.040221323685891, "grad_norm": 14.4296875, "learning_rate": 9.002558271745311e-06, "loss": 0.6342, "step": 28572 }, { "epoch": 3.0403277293041073, "grad_norm": 12.1796875, "learning_rate": 9.002522740193293e-06, "loss": 0.562, "step": 28573 }, { "epoch": 3.040434134922324, "grad_norm": 15.1484375, "learning_rate": 9.002487208641274e-06, "loss": 0.7851, "step": 28574 }, { "epoch": 3.0405405405405403, "grad_norm": 12.4453125, "learning_rate": 9.002451677089256e-06, "loss": 0.5021, "step": 28575 }, { "epoch": 3.0406469461587573, "grad_norm": 12.7265625, "learning_rate": 9.002416145537239e-06, "loss": 0.7631, "step": 28576 }, { "epoch": 3.040753351776974, "grad_norm": 10.0390625, "learning_rate": 9.00238061398522e-06, "loss": 0.4508, "step": 28577 }, { "epoch": 3.0408597573951903, "grad_norm": 18.09375, "learning_rate": 9.002345082433202e-06, "loss": 0.685, "step": 28578 }, { "epoch": 3.0409661630134073, "grad_norm": 18.96875, "learning_rate": 9.002309550881182e-06, "loss": 0.7474, "step": 28579 }, { "epoch": 3.041072568631624, "grad_norm": 14.0859375, "learning_rate": 9.002274019329164e-06, "loss": 0.6484, "step": 28580 }, { "epoch": 3.0411789742498403, "grad_norm": 14.0, "learning_rate": 9.002238487777147e-06, "loss": 0.6239, "step": 28581 }, { "epoch": 3.0412853798680572, "grad_norm": 10.828125, "learning_rate": 9.002202956225129e-06, "loss": 0.5626, "step": 28582 }, { "epoch": 3.0413917854862738, "grad_norm": 16.21875, "learning_rate": 9.00216742467311e-06, "loss": 0.6706, "step": 28583 }, { "epoch": 3.0414981911044903, "grad_norm": 10.90625, "learning_rate": 9.002131893121092e-06, "loss": 0.569, "step": 28584 }, { "epoch": 3.0416045967227068, "grad_norm": 12.2734375, "learning_rate": 9.002096361569074e-06, "loss": 0.5888, "step": 28585 }, { "epoch": 3.0417110023409237, "grad_norm": 13.1171875, "learning_rate": 9.002060830017055e-06, "loss": 0.6869, "step": 28586 }, { "epoch": 3.0418174079591402, "grad_norm": 20.609375, "learning_rate": 9.002025298465038e-06, "loss": 0.8334, "step": 28587 }, { "epoch": 3.0419238135773567, "grad_norm": 10.390625, "learning_rate": 9.00198976691302e-06, "loss": 0.6219, "step": 28588 }, { "epoch": 3.0420302191955737, "grad_norm": 17.171875, "learning_rate": 9.001954235361002e-06, "loss": 0.6665, "step": 28589 }, { "epoch": 3.04213662481379, "grad_norm": 11.5078125, "learning_rate": 9.001918703808983e-06, "loss": 0.5188, "step": 28590 }, { "epoch": 3.0422430304320067, "grad_norm": 12.7890625, "learning_rate": 9.001883172256965e-06, "loss": 0.6045, "step": 28591 }, { "epoch": 3.0423494360502232, "grad_norm": 12.765625, "learning_rate": 9.001847640704947e-06, "loss": 0.5653, "step": 28592 }, { "epoch": 3.04245584166844, "grad_norm": 13.7734375, "learning_rate": 9.001812109152928e-06, "loss": 0.6889, "step": 28593 }, { "epoch": 3.0425622472866567, "grad_norm": 14.6171875, "learning_rate": 9.00177657760091e-06, "loss": 0.6444, "step": 28594 }, { "epoch": 3.042668652904873, "grad_norm": 11.078125, "learning_rate": 9.001741046048891e-06, "loss": 0.5986, "step": 28595 }, { "epoch": 3.04277505852309, "grad_norm": 14.8984375, "learning_rate": 9.001705514496873e-06, "loss": 0.5655, "step": 28596 }, { "epoch": 3.0428814641413067, "grad_norm": 9.90625, "learning_rate": 9.001669982944856e-06, "loss": 0.4839, "step": 28597 }, { "epoch": 3.042987869759523, "grad_norm": 12.484375, "learning_rate": 9.001634451392838e-06, "loss": 0.6888, "step": 28598 }, { "epoch": 3.04309427537774, "grad_norm": 10.6171875, "learning_rate": 9.00159891984082e-06, "loss": 0.7237, "step": 28599 }, { "epoch": 3.0432006809959566, "grad_norm": 12.0546875, "learning_rate": 9.001563388288801e-06, "loss": 0.6537, "step": 28600 }, { "epoch": 3.043307086614173, "grad_norm": 10.3828125, "learning_rate": 9.001527856736783e-06, "loss": 0.4223, "step": 28601 }, { "epoch": 3.0434134922323897, "grad_norm": 12.453125, "learning_rate": 9.001492325184766e-06, "loss": 0.6389, "step": 28602 }, { "epoch": 3.0435198978506066, "grad_norm": 11.9765625, "learning_rate": 9.001456793632748e-06, "loss": 0.6762, "step": 28603 }, { "epoch": 3.043626303468823, "grad_norm": 13.046875, "learning_rate": 9.00142126208073e-06, "loss": 0.561, "step": 28604 }, { "epoch": 3.0437327090870396, "grad_norm": 12.4765625, "learning_rate": 9.00138573052871e-06, "loss": 0.51, "step": 28605 }, { "epoch": 3.0438391147052566, "grad_norm": 17.015625, "learning_rate": 9.001350198976692e-06, "loss": 0.7459, "step": 28606 }, { "epoch": 3.043945520323473, "grad_norm": 10.46875, "learning_rate": 9.001314667424674e-06, "loss": 0.507, "step": 28607 }, { "epoch": 3.0440519259416896, "grad_norm": 13.546875, "learning_rate": 9.001279135872656e-06, "loss": 0.6625, "step": 28608 }, { "epoch": 3.0441583315599066, "grad_norm": 11.8359375, "learning_rate": 9.001243604320637e-06, "loss": 0.7195, "step": 28609 }, { "epoch": 3.044264737178123, "grad_norm": 11.3359375, "learning_rate": 9.001208072768619e-06, "loss": 0.6172, "step": 28610 }, { "epoch": 3.0443711427963396, "grad_norm": 13.359375, "learning_rate": 9.0011725412166e-06, "loss": 0.662, "step": 28611 }, { "epoch": 3.044477548414556, "grad_norm": 8.2734375, "learning_rate": 9.001137009664584e-06, "loss": 0.427, "step": 28612 }, { "epoch": 3.044583954032773, "grad_norm": 9.0390625, "learning_rate": 9.001101478112565e-06, "loss": 0.414, "step": 28613 }, { "epoch": 3.0446903596509896, "grad_norm": 9.8359375, "learning_rate": 9.001065946560547e-06, "loss": 0.4245, "step": 28614 }, { "epoch": 3.044796765269206, "grad_norm": 18.984375, "learning_rate": 9.001030415008529e-06, "loss": 0.7151, "step": 28615 }, { "epoch": 3.044903170887423, "grad_norm": 19.375, "learning_rate": 9.00099488345651e-06, "loss": 0.8606, "step": 28616 }, { "epoch": 3.0450095765056395, "grad_norm": 12.4921875, "learning_rate": 9.000959351904492e-06, "loss": 0.6171, "step": 28617 }, { "epoch": 3.045115982123856, "grad_norm": 9.3828125, "learning_rate": 9.000923820352473e-06, "loss": 0.4553, "step": 28618 }, { "epoch": 3.045222387742073, "grad_norm": 11.8984375, "learning_rate": 9.000888288800455e-06, "loss": 0.5803, "step": 28619 }, { "epoch": 3.0453287933602895, "grad_norm": 18.109375, "learning_rate": 9.000852757248437e-06, "loss": 0.7331, "step": 28620 }, { "epoch": 3.045435198978506, "grad_norm": 14.09375, "learning_rate": 9.000817225696418e-06, "loss": 0.7183, "step": 28621 }, { "epoch": 3.0455416045967225, "grad_norm": 19.03125, "learning_rate": 9.000781694144401e-06, "loss": 0.7468, "step": 28622 }, { "epoch": 3.0456480102149395, "grad_norm": 12.3515625, "learning_rate": 9.000746162592383e-06, "loss": 0.5192, "step": 28623 }, { "epoch": 3.045754415833156, "grad_norm": 15.421875, "learning_rate": 9.000710631040365e-06, "loss": 0.6027, "step": 28624 }, { "epoch": 3.0458608214513725, "grad_norm": 11.1015625, "learning_rate": 9.000675099488346e-06, "loss": 0.6384, "step": 28625 }, { "epoch": 3.0459672270695894, "grad_norm": 12.6328125, "learning_rate": 9.000639567936328e-06, "loss": 0.7327, "step": 28626 }, { "epoch": 3.046073632687806, "grad_norm": 12.265625, "learning_rate": 9.00060403638431e-06, "loss": 0.6294, "step": 28627 }, { "epoch": 3.0461800383060225, "grad_norm": 13.515625, "learning_rate": 9.000568504832293e-06, "loss": 0.5068, "step": 28628 }, { "epoch": 3.0462864439242394, "grad_norm": 10.84375, "learning_rate": 9.000532973280274e-06, "loss": 0.5335, "step": 28629 }, { "epoch": 3.046392849542456, "grad_norm": 11.0078125, "learning_rate": 9.000497441728256e-06, "loss": 0.6059, "step": 28630 }, { "epoch": 3.0464992551606724, "grad_norm": 14.125, "learning_rate": 9.000461910176238e-06, "loss": 0.5494, "step": 28631 }, { "epoch": 3.046605660778889, "grad_norm": 12.65625, "learning_rate": 9.00042637862422e-06, "loss": 0.5888, "step": 28632 }, { "epoch": 3.046712066397106, "grad_norm": 12.875, "learning_rate": 9.000390847072201e-06, "loss": 0.5949, "step": 28633 }, { "epoch": 3.0468184720153224, "grad_norm": 15.7578125, "learning_rate": 9.000355315520182e-06, "loss": 0.8491, "step": 28634 }, { "epoch": 3.046924877633539, "grad_norm": 16.40625, "learning_rate": 9.000319783968164e-06, "loss": 0.7842, "step": 28635 }, { "epoch": 3.047031283251756, "grad_norm": 10.7265625, "learning_rate": 9.000284252416146e-06, "loss": 0.5282, "step": 28636 }, { "epoch": 3.0471376888699724, "grad_norm": 13.4140625, "learning_rate": 9.000248720864127e-06, "loss": 0.7983, "step": 28637 }, { "epoch": 3.047244094488189, "grad_norm": 15.015625, "learning_rate": 9.00021318931211e-06, "loss": 0.5446, "step": 28638 }, { "epoch": 3.0473505001064054, "grad_norm": 17.515625, "learning_rate": 9.000177657760092e-06, "loss": 0.6225, "step": 28639 }, { "epoch": 3.0474569057246224, "grad_norm": 11.265625, "learning_rate": 9.000142126208074e-06, "loss": 0.7164, "step": 28640 }, { "epoch": 3.047563311342839, "grad_norm": 11.1484375, "learning_rate": 9.000106594656055e-06, "loss": 0.6224, "step": 28641 }, { "epoch": 3.0476697169610554, "grad_norm": 11.0078125, "learning_rate": 9.000071063104037e-06, "loss": 0.5757, "step": 28642 }, { "epoch": 3.0477761225792723, "grad_norm": 17.078125, "learning_rate": 9.000035531552019e-06, "loss": 0.7193, "step": 28643 }, { "epoch": 3.047882528197489, "grad_norm": 9.5078125, "learning_rate": 9e-06, "loss": 0.5332, "step": 28644 }, { "epoch": 3.0479889338157053, "grad_norm": 11.8046875, "learning_rate": 8.999964468447982e-06, "loss": 0.5677, "step": 28645 }, { "epoch": 3.0480953394339223, "grad_norm": 9.3671875, "learning_rate": 8.999928936895963e-06, "loss": 0.4293, "step": 28646 }, { "epoch": 3.048201745052139, "grad_norm": 10.5703125, "learning_rate": 8.999893405343945e-06, "loss": 0.5445, "step": 28647 }, { "epoch": 3.0483081506703553, "grad_norm": 14.375, "learning_rate": 8.999857873791928e-06, "loss": 0.7147, "step": 28648 }, { "epoch": 3.048414556288572, "grad_norm": 12.46875, "learning_rate": 8.99982234223991e-06, "loss": 0.6286, "step": 28649 }, { "epoch": 3.048520961906789, "grad_norm": 8.8359375, "learning_rate": 8.999786810687892e-06, "loss": 0.612, "step": 28650 }, { "epoch": 3.0486273675250053, "grad_norm": 16.765625, "learning_rate": 8.999751279135873e-06, "loss": 0.7029, "step": 28651 }, { "epoch": 3.048733773143222, "grad_norm": 7.83203125, "learning_rate": 8.999715747583855e-06, "loss": 0.4205, "step": 28652 }, { "epoch": 3.0488401787614388, "grad_norm": 12.1953125, "learning_rate": 8.999680216031838e-06, "loss": 0.7026, "step": 28653 }, { "epoch": 3.0489465843796553, "grad_norm": 12.0859375, "learning_rate": 8.99964468447982e-06, "loss": 0.7199, "step": 28654 }, { "epoch": 3.0490529899978718, "grad_norm": 15.078125, "learning_rate": 8.999609152927801e-06, "loss": 0.6072, "step": 28655 }, { "epoch": 3.0491593956160887, "grad_norm": 11.0078125, "learning_rate": 8.999573621375783e-06, "loss": 0.4741, "step": 28656 }, { "epoch": 3.0492658012343052, "grad_norm": 12.671875, "learning_rate": 8.999538089823765e-06, "loss": 0.6858, "step": 28657 }, { "epoch": 3.0493722068525217, "grad_norm": 16.78125, "learning_rate": 8.999502558271746e-06, "loss": 0.7122, "step": 28658 }, { "epoch": 3.0494786124707383, "grad_norm": 14.8125, "learning_rate": 8.999467026719728e-06, "loss": 0.685, "step": 28659 }, { "epoch": 3.049585018088955, "grad_norm": 9.0546875, "learning_rate": 8.99943149516771e-06, "loss": 0.3889, "step": 28660 }, { "epoch": 3.0496914237071717, "grad_norm": 9.6328125, "learning_rate": 8.999395963615691e-06, "loss": 0.5554, "step": 28661 }, { "epoch": 3.0497978293253882, "grad_norm": 20.140625, "learning_rate": 8.999360432063673e-06, "loss": 0.7534, "step": 28662 }, { "epoch": 3.049904234943605, "grad_norm": 12.0703125, "learning_rate": 8.999324900511656e-06, "loss": 0.6271, "step": 28663 }, { "epoch": 3.0500106405618217, "grad_norm": 15.859375, "learning_rate": 8.999289368959637e-06, "loss": 0.7011, "step": 28664 }, { "epoch": 3.050117046180038, "grad_norm": 9.578125, "learning_rate": 8.999253837407619e-06, "loss": 0.5507, "step": 28665 }, { "epoch": 3.050223451798255, "grad_norm": 14.3125, "learning_rate": 8.9992183058556e-06, "loss": 0.6234, "step": 28666 }, { "epoch": 3.0503298574164717, "grad_norm": 11.3828125, "learning_rate": 8.999182774303582e-06, "loss": 0.6002, "step": 28667 }, { "epoch": 3.050436263034688, "grad_norm": 13.9375, "learning_rate": 8.999147242751564e-06, "loss": 0.6144, "step": 28668 }, { "epoch": 3.0505426686529047, "grad_norm": 12.9609375, "learning_rate": 8.999111711199545e-06, "loss": 0.556, "step": 28669 }, { "epoch": 3.0506490742711216, "grad_norm": 11.9609375, "learning_rate": 8.999076179647527e-06, "loss": 0.6453, "step": 28670 }, { "epoch": 3.050755479889338, "grad_norm": 11.984375, "learning_rate": 8.999040648095509e-06, "loss": 0.6801, "step": 28671 }, { "epoch": 3.0508618855075547, "grad_norm": 14.7421875, "learning_rate": 8.99900511654349e-06, "loss": 0.7246, "step": 28672 }, { "epoch": 3.0509682911257716, "grad_norm": 15.34375, "learning_rate": 8.998969584991472e-06, "loss": 0.7171, "step": 28673 }, { "epoch": 3.051074696743988, "grad_norm": 31.921875, "learning_rate": 8.998934053439455e-06, "loss": 0.5905, "step": 28674 }, { "epoch": 3.0511811023622046, "grad_norm": 17.015625, "learning_rate": 8.998898521887437e-06, "loss": 0.5654, "step": 28675 }, { "epoch": 3.0512875079804216, "grad_norm": 14.21875, "learning_rate": 8.998862990335418e-06, "loss": 0.6095, "step": 28676 }, { "epoch": 3.051393913598638, "grad_norm": 12.1875, "learning_rate": 8.9988274587834e-06, "loss": 0.5474, "step": 28677 }, { "epoch": 3.0515003192168546, "grad_norm": 14.2265625, "learning_rate": 8.998791927231382e-06, "loss": 0.8121, "step": 28678 }, { "epoch": 3.051606724835071, "grad_norm": 9.8984375, "learning_rate": 8.998756395679365e-06, "loss": 0.4423, "step": 28679 }, { "epoch": 3.051713130453288, "grad_norm": 10.59375, "learning_rate": 8.998720864127347e-06, "loss": 0.472, "step": 28680 }, { "epoch": 3.0518195360715046, "grad_norm": 14.7265625, "learning_rate": 8.998685332575328e-06, "loss": 0.7053, "step": 28681 }, { "epoch": 3.051925941689721, "grad_norm": 13.1328125, "learning_rate": 8.99864980102331e-06, "loss": 0.552, "step": 28682 }, { "epoch": 3.052032347307938, "grad_norm": 15.4375, "learning_rate": 8.998614269471291e-06, "loss": 0.5457, "step": 28683 }, { "epoch": 3.0521387529261546, "grad_norm": 12.7890625, "learning_rate": 8.998578737919273e-06, "loss": 0.6518, "step": 28684 }, { "epoch": 3.052245158544371, "grad_norm": 10.9296875, "learning_rate": 8.998543206367255e-06, "loss": 0.447, "step": 28685 }, { "epoch": 3.0523515641625876, "grad_norm": 16.859375, "learning_rate": 8.998507674815236e-06, "loss": 0.7123, "step": 28686 }, { "epoch": 3.0524579697808045, "grad_norm": 9.2890625, "learning_rate": 8.998472143263218e-06, "loss": 0.5241, "step": 28687 }, { "epoch": 3.052564375399021, "grad_norm": 15.0078125, "learning_rate": 8.9984366117112e-06, "loss": 0.5569, "step": 28688 }, { "epoch": 3.0526707810172375, "grad_norm": 12.5078125, "learning_rate": 8.998401080159183e-06, "loss": 0.6769, "step": 28689 }, { "epoch": 3.0527771866354545, "grad_norm": 13.265625, "learning_rate": 8.998365548607164e-06, "loss": 0.5659, "step": 28690 }, { "epoch": 3.052883592253671, "grad_norm": 11.0, "learning_rate": 8.998330017055146e-06, "loss": 0.5287, "step": 28691 }, { "epoch": 3.0529899978718875, "grad_norm": 14.3515625, "learning_rate": 8.998294485503128e-06, "loss": 0.6558, "step": 28692 }, { "epoch": 3.0530964034901045, "grad_norm": 9.15625, "learning_rate": 8.998258953951109e-06, "loss": 0.5073, "step": 28693 }, { "epoch": 3.053202809108321, "grad_norm": 11.5703125, "learning_rate": 8.998223422399092e-06, "loss": 0.5412, "step": 28694 }, { "epoch": 3.0533092147265375, "grad_norm": 10.9453125, "learning_rate": 8.998187890847074e-06, "loss": 0.5635, "step": 28695 }, { "epoch": 3.053415620344754, "grad_norm": 11.2734375, "learning_rate": 8.998152359295056e-06, "loss": 0.5055, "step": 28696 }, { "epoch": 3.053522025962971, "grad_norm": 13.1015625, "learning_rate": 8.998116827743037e-06, "loss": 0.6665, "step": 28697 }, { "epoch": 3.0536284315811875, "grad_norm": 16.109375, "learning_rate": 8.998081296191019e-06, "loss": 0.6898, "step": 28698 }, { "epoch": 3.053734837199404, "grad_norm": 9.0, "learning_rate": 8.998045764639e-06, "loss": 0.543, "step": 28699 }, { "epoch": 3.053841242817621, "grad_norm": 19.453125, "learning_rate": 8.998010233086982e-06, "loss": 0.6459, "step": 28700 }, { "epoch": 3.0539476484358374, "grad_norm": 11.34375, "learning_rate": 8.997974701534964e-06, "loss": 0.5425, "step": 28701 }, { "epoch": 3.054054054054054, "grad_norm": 15.3125, "learning_rate": 8.997939169982945e-06, "loss": 0.8222, "step": 28702 }, { "epoch": 3.054160459672271, "grad_norm": 10.8671875, "learning_rate": 8.997903638430927e-06, "loss": 0.6326, "step": 28703 }, { "epoch": 3.0542668652904874, "grad_norm": 11.3203125, "learning_rate": 8.997868106878908e-06, "loss": 0.7107, "step": 28704 }, { "epoch": 3.054373270908704, "grad_norm": 15.4921875, "learning_rate": 8.997832575326892e-06, "loss": 0.5321, "step": 28705 }, { "epoch": 3.0544796765269204, "grad_norm": 13.046875, "learning_rate": 8.997797043774873e-06, "loss": 0.6235, "step": 28706 }, { "epoch": 3.0545860821451374, "grad_norm": 13.09375, "learning_rate": 8.997761512222855e-06, "loss": 0.574, "step": 28707 }, { "epoch": 3.054692487763354, "grad_norm": 13.8515625, "learning_rate": 8.997725980670837e-06, "loss": 0.6519, "step": 28708 }, { "epoch": 3.0547988933815704, "grad_norm": 11.71875, "learning_rate": 8.997690449118818e-06, "loss": 0.5782, "step": 28709 }, { "epoch": 3.0549052989997874, "grad_norm": 15.7578125, "learning_rate": 8.9976549175668e-06, "loss": 0.58, "step": 28710 }, { "epoch": 3.055011704618004, "grad_norm": 10.6171875, "learning_rate": 8.997619386014781e-06, "loss": 0.6455, "step": 28711 }, { "epoch": 3.0551181102362204, "grad_norm": 12.28125, "learning_rate": 8.997583854462763e-06, "loss": 0.4891, "step": 28712 }, { "epoch": 3.0552245158544373, "grad_norm": 15.125, "learning_rate": 8.997548322910745e-06, "loss": 0.5875, "step": 28713 }, { "epoch": 3.055330921472654, "grad_norm": 12.421875, "learning_rate": 8.997512791358726e-06, "loss": 0.5894, "step": 28714 }, { "epoch": 3.0554373270908703, "grad_norm": 15.484375, "learning_rate": 8.99747725980671e-06, "loss": 0.6663, "step": 28715 }, { "epoch": 3.055543732709087, "grad_norm": 11.0703125, "learning_rate": 8.997441728254691e-06, "loss": 0.6391, "step": 28716 }, { "epoch": 3.055650138327304, "grad_norm": 17.140625, "learning_rate": 8.997406196702673e-06, "loss": 0.8431, "step": 28717 }, { "epoch": 3.0557565439455203, "grad_norm": 10.7109375, "learning_rate": 8.997370665150654e-06, "loss": 0.6559, "step": 28718 }, { "epoch": 3.055862949563737, "grad_norm": 12.2578125, "learning_rate": 8.997335133598636e-06, "loss": 0.5583, "step": 28719 }, { "epoch": 3.055969355181954, "grad_norm": 18.734375, "learning_rate": 8.99729960204662e-06, "loss": 0.7476, "step": 28720 }, { "epoch": 3.0560757608001703, "grad_norm": 11.28125, "learning_rate": 8.997264070494601e-06, "loss": 0.5252, "step": 28721 }, { "epoch": 3.056182166418387, "grad_norm": 14.984375, "learning_rate": 8.997228538942582e-06, "loss": 0.6464, "step": 28722 }, { "epoch": 3.0562885720366033, "grad_norm": 11.78125, "learning_rate": 8.997193007390564e-06, "loss": 0.5867, "step": 28723 }, { "epoch": 3.0563949776548203, "grad_norm": 10.4609375, "learning_rate": 8.997157475838546e-06, "loss": 0.5838, "step": 28724 }, { "epoch": 3.0565013832730368, "grad_norm": 13.2890625, "learning_rate": 8.997121944286527e-06, "loss": 0.6437, "step": 28725 }, { "epoch": 3.0566077888912533, "grad_norm": 14.7734375, "learning_rate": 8.997086412734509e-06, "loss": 0.6864, "step": 28726 }, { "epoch": 3.0567141945094702, "grad_norm": 13.4140625, "learning_rate": 8.99705088118249e-06, "loss": 0.6925, "step": 28727 }, { "epoch": 3.0568206001276867, "grad_norm": 8.59375, "learning_rate": 8.997015349630472e-06, "loss": 0.4783, "step": 28728 }, { "epoch": 3.0569270057459033, "grad_norm": 11.71875, "learning_rate": 8.996979818078454e-06, "loss": 0.5788, "step": 28729 }, { "epoch": 3.05703341136412, "grad_norm": 11.96875, "learning_rate": 8.996944286526437e-06, "loss": 0.6376, "step": 28730 }, { "epoch": 3.0571398169823367, "grad_norm": 11.3671875, "learning_rate": 8.996908754974419e-06, "loss": 0.4443, "step": 28731 }, { "epoch": 3.0572462226005532, "grad_norm": 12.296875, "learning_rate": 8.9968732234224e-06, "loss": 0.6082, "step": 28732 }, { "epoch": 3.0573526282187697, "grad_norm": 10.6796875, "learning_rate": 8.996837691870382e-06, "loss": 0.5488, "step": 28733 }, { "epoch": 3.0574590338369867, "grad_norm": 12.015625, "learning_rate": 8.996802160318363e-06, "loss": 0.5381, "step": 28734 }, { "epoch": 3.057565439455203, "grad_norm": 11.4765625, "learning_rate": 8.996766628766345e-06, "loss": 0.5253, "step": 28735 }, { "epoch": 3.0576718450734197, "grad_norm": 11.53125, "learning_rate": 8.996731097214327e-06, "loss": 0.5937, "step": 28736 }, { "epoch": 3.0577782506916367, "grad_norm": 10.3359375, "learning_rate": 8.996695565662308e-06, "loss": 0.573, "step": 28737 }, { "epoch": 3.057884656309853, "grad_norm": 14.109375, "learning_rate": 8.99666003411029e-06, "loss": 0.7855, "step": 28738 }, { "epoch": 3.0579910619280697, "grad_norm": 14.8203125, "learning_rate": 8.996624502558271e-06, "loss": 0.6966, "step": 28739 }, { "epoch": 3.0580974675462866, "grad_norm": 13.59375, "learning_rate": 8.996588971006255e-06, "loss": 0.5807, "step": 28740 }, { "epoch": 3.058203873164503, "grad_norm": 12.203125, "learning_rate": 8.996553439454236e-06, "loss": 0.6015, "step": 28741 }, { "epoch": 3.0583102787827197, "grad_norm": 13.0390625, "learning_rate": 8.996517907902218e-06, "loss": 0.5945, "step": 28742 }, { "epoch": 3.058416684400936, "grad_norm": 11.7890625, "learning_rate": 8.9964823763502e-06, "loss": 0.5312, "step": 28743 }, { "epoch": 3.058523090019153, "grad_norm": 9.1875, "learning_rate": 8.996446844798181e-06, "loss": 0.4144, "step": 28744 }, { "epoch": 3.0586294956373696, "grad_norm": 11.875, "learning_rate": 8.996411313246163e-06, "loss": 0.5042, "step": 28745 }, { "epoch": 3.058735901255586, "grad_norm": 15.828125, "learning_rate": 8.996375781694146e-06, "loss": 0.6571, "step": 28746 }, { "epoch": 3.058842306873803, "grad_norm": 18.96875, "learning_rate": 8.996340250142128e-06, "loss": 0.691, "step": 28747 }, { "epoch": 3.0589487124920196, "grad_norm": 9.0546875, "learning_rate": 8.99630471859011e-06, "loss": 0.4518, "step": 28748 }, { "epoch": 3.059055118110236, "grad_norm": 18.578125, "learning_rate": 8.996269187038091e-06, "loss": 0.8464, "step": 28749 }, { "epoch": 3.059161523728453, "grad_norm": 13.484375, "learning_rate": 8.996233655486073e-06, "loss": 0.6234, "step": 28750 }, { "epoch": 3.0592679293466696, "grad_norm": 12.8984375, "learning_rate": 8.996198123934054e-06, "loss": 0.727, "step": 28751 }, { "epoch": 3.059374334964886, "grad_norm": 10.390625, "learning_rate": 8.996162592382036e-06, "loss": 0.5041, "step": 28752 }, { "epoch": 3.0594807405831026, "grad_norm": 9.953125, "learning_rate": 8.996127060830017e-06, "loss": 0.48, "step": 28753 }, { "epoch": 3.0595871462013196, "grad_norm": 17.1875, "learning_rate": 8.996091529277999e-06, "loss": 0.5808, "step": 28754 }, { "epoch": 3.059693551819536, "grad_norm": 9.78125, "learning_rate": 8.99605599772598e-06, "loss": 0.5643, "step": 28755 }, { "epoch": 3.0597999574377526, "grad_norm": 9.7578125, "learning_rate": 8.996020466173964e-06, "loss": 0.4547, "step": 28756 }, { "epoch": 3.0599063630559695, "grad_norm": 19.171875, "learning_rate": 8.995984934621945e-06, "loss": 0.8446, "step": 28757 }, { "epoch": 3.060012768674186, "grad_norm": 14.8828125, "learning_rate": 8.995949403069927e-06, "loss": 0.6187, "step": 28758 }, { "epoch": 3.0601191742924025, "grad_norm": 9.984375, "learning_rate": 8.995913871517909e-06, "loss": 0.5295, "step": 28759 }, { "epoch": 3.0602255799106195, "grad_norm": 11.3125, "learning_rate": 8.99587833996589e-06, "loss": 0.6515, "step": 28760 }, { "epoch": 3.060331985528836, "grad_norm": 15.734375, "learning_rate": 8.995842808413872e-06, "loss": 0.6084, "step": 28761 }, { "epoch": 3.0604383911470525, "grad_norm": 14.59375, "learning_rate": 8.995807276861854e-06, "loss": 0.6518, "step": 28762 }, { "epoch": 3.060544796765269, "grad_norm": 11.3125, "learning_rate": 8.995771745309835e-06, "loss": 0.7453, "step": 28763 }, { "epoch": 3.060651202383486, "grad_norm": 18.46875, "learning_rate": 8.995736213757817e-06, "loss": 0.8429, "step": 28764 }, { "epoch": 3.0607576080017025, "grad_norm": 15.1953125, "learning_rate": 8.995700682205798e-06, "loss": 0.6943, "step": 28765 }, { "epoch": 3.060864013619919, "grad_norm": 9.578125, "learning_rate": 8.995665150653782e-06, "loss": 0.4449, "step": 28766 }, { "epoch": 3.060970419238136, "grad_norm": 10.9140625, "learning_rate": 8.995629619101763e-06, "loss": 0.5913, "step": 28767 }, { "epoch": 3.0610768248563525, "grad_norm": 11.484375, "learning_rate": 8.995594087549745e-06, "loss": 0.4504, "step": 28768 }, { "epoch": 3.061183230474569, "grad_norm": 13.1015625, "learning_rate": 8.995558555997726e-06, "loss": 0.7448, "step": 28769 }, { "epoch": 3.0612896360927855, "grad_norm": 8.3671875, "learning_rate": 8.995523024445708e-06, "loss": 0.5514, "step": 28770 }, { "epoch": 3.0613960417110024, "grad_norm": 12.5, "learning_rate": 8.995487492893691e-06, "loss": 0.5603, "step": 28771 }, { "epoch": 3.061502447329219, "grad_norm": 13.953125, "learning_rate": 8.995451961341673e-06, "loss": 0.6666, "step": 28772 }, { "epoch": 3.0616088529474355, "grad_norm": 11.8984375, "learning_rate": 8.995416429789655e-06, "loss": 0.4997, "step": 28773 }, { "epoch": 3.0617152585656524, "grad_norm": 12.171875, "learning_rate": 8.995380898237636e-06, "loss": 0.4557, "step": 28774 }, { "epoch": 3.061821664183869, "grad_norm": 17.0625, "learning_rate": 8.995345366685618e-06, "loss": 0.6532, "step": 28775 }, { "epoch": 3.0619280698020854, "grad_norm": 8.953125, "learning_rate": 8.9953098351336e-06, "loss": 0.5998, "step": 28776 }, { "epoch": 3.0620344754203024, "grad_norm": 15.7578125, "learning_rate": 8.995274303581581e-06, "loss": 0.6221, "step": 28777 }, { "epoch": 3.062140881038519, "grad_norm": 14.6953125, "learning_rate": 8.995238772029563e-06, "loss": 0.5184, "step": 28778 }, { "epoch": 3.0622472866567354, "grad_norm": 15.46875, "learning_rate": 8.995203240477544e-06, "loss": 0.7227, "step": 28779 }, { "epoch": 3.062353692274952, "grad_norm": 13.6875, "learning_rate": 8.995167708925526e-06, "loss": 0.5416, "step": 28780 }, { "epoch": 3.062460097893169, "grad_norm": 15.109375, "learning_rate": 8.995132177373507e-06, "loss": 0.5776, "step": 28781 }, { "epoch": 3.0625665035113854, "grad_norm": 15.28125, "learning_rate": 8.99509664582149e-06, "loss": 0.6246, "step": 28782 }, { "epoch": 3.062672909129602, "grad_norm": 11.9765625, "learning_rate": 8.995061114269472e-06, "loss": 0.3847, "step": 28783 }, { "epoch": 3.062779314747819, "grad_norm": 20.90625, "learning_rate": 8.995025582717454e-06, "loss": 0.8299, "step": 28784 }, { "epoch": 3.0628857203660353, "grad_norm": 13.0, "learning_rate": 8.994990051165436e-06, "loss": 0.5303, "step": 28785 }, { "epoch": 3.062992125984252, "grad_norm": 12.0, "learning_rate": 8.994954519613417e-06, "loss": 0.9164, "step": 28786 }, { "epoch": 3.063098531602469, "grad_norm": 12.3125, "learning_rate": 8.994918988061399e-06, "loss": 0.6335, "step": 28787 }, { "epoch": 3.0632049372206853, "grad_norm": 14.234375, "learning_rate": 8.99488345650938e-06, "loss": 0.7316, "step": 28788 }, { "epoch": 3.063311342838902, "grad_norm": 12.390625, "learning_rate": 8.994847924957362e-06, "loss": 0.526, "step": 28789 }, { "epoch": 3.0634177484571183, "grad_norm": 16.203125, "learning_rate": 8.994812393405344e-06, "loss": 0.7032, "step": 28790 }, { "epoch": 3.0635241540753353, "grad_norm": 14.171875, "learning_rate": 8.994776861853325e-06, "loss": 0.6909, "step": 28791 }, { "epoch": 3.063630559693552, "grad_norm": 10.015625, "learning_rate": 8.994741330301309e-06, "loss": 0.4434, "step": 28792 }, { "epoch": 3.0637369653117683, "grad_norm": 10.0546875, "learning_rate": 8.99470579874929e-06, "loss": 0.4685, "step": 28793 }, { "epoch": 3.0638433709299853, "grad_norm": 17.765625, "learning_rate": 8.994670267197272e-06, "loss": 0.7295, "step": 28794 }, { "epoch": 3.0639497765482018, "grad_norm": 14.4765625, "learning_rate": 8.994634735645253e-06, "loss": 0.7891, "step": 28795 }, { "epoch": 3.0640561821664183, "grad_norm": 11.546875, "learning_rate": 8.994599204093235e-06, "loss": 0.4139, "step": 28796 }, { "epoch": 3.0641625877846352, "grad_norm": 15.6171875, "learning_rate": 8.994563672541218e-06, "loss": 0.9076, "step": 28797 }, { "epoch": 3.0642689934028517, "grad_norm": 10.7578125, "learning_rate": 8.9945281409892e-06, "loss": 0.5162, "step": 28798 }, { "epoch": 3.0643753990210683, "grad_norm": 11.4921875, "learning_rate": 8.994492609437181e-06, "loss": 0.5407, "step": 28799 }, { "epoch": 3.0644818046392848, "grad_norm": 12.421875, "learning_rate": 8.994457077885163e-06, "loss": 0.6486, "step": 28800 }, { "epoch": 3.0645882102575017, "grad_norm": 10.4375, "learning_rate": 8.994421546333145e-06, "loss": 0.5976, "step": 28801 }, { "epoch": 3.0646946158757182, "grad_norm": 13.171875, "learning_rate": 8.994386014781126e-06, "loss": 0.6082, "step": 28802 }, { "epoch": 3.0648010214939347, "grad_norm": 11.296875, "learning_rate": 8.994350483229108e-06, "loss": 0.4823, "step": 28803 }, { "epoch": 3.0649074271121517, "grad_norm": 12.171875, "learning_rate": 8.99431495167709e-06, "loss": 0.6613, "step": 28804 }, { "epoch": 3.065013832730368, "grad_norm": 15.3984375, "learning_rate": 8.994279420125071e-06, "loss": 0.6203, "step": 28805 }, { "epoch": 3.0651202383485847, "grad_norm": 28.890625, "learning_rate": 8.994243888573053e-06, "loss": 1.0876, "step": 28806 }, { "epoch": 3.0652266439668017, "grad_norm": 12.5546875, "learning_rate": 8.994208357021036e-06, "loss": 0.6784, "step": 28807 }, { "epoch": 3.065333049585018, "grad_norm": 14.4609375, "learning_rate": 8.994172825469018e-06, "loss": 0.6264, "step": 28808 }, { "epoch": 3.0654394552032347, "grad_norm": 16.515625, "learning_rate": 8.994137293917e-06, "loss": 0.609, "step": 28809 }, { "epoch": 3.065545860821451, "grad_norm": 11.796875, "learning_rate": 8.99410176236498e-06, "loss": 0.4869, "step": 28810 }, { "epoch": 3.065652266439668, "grad_norm": 9.8984375, "learning_rate": 8.994066230812962e-06, "loss": 0.5453, "step": 28811 }, { "epoch": 3.0657586720578847, "grad_norm": 13.78125, "learning_rate": 8.994030699260946e-06, "loss": 0.674, "step": 28812 }, { "epoch": 3.065865077676101, "grad_norm": 15.3359375, "learning_rate": 8.993995167708927e-06, "loss": 0.5353, "step": 28813 }, { "epoch": 3.065971483294318, "grad_norm": 11.671875, "learning_rate": 8.993959636156909e-06, "loss": 0.5693, "step": 28814 }, { "epoch": 3.0660778889125346, "grad_norm": 16.9375, "learning_rate": 8.99392410460489e-06, "loss": 0.7184, "step": 28815 }, { "epoch": 3.066184294530751, "grad_norm": 15.59375, "learning_rate": 8.993888573052872e-06, "loss": 0.7443, "step": 28816 }, { "epoch": 3.0662907001489677, "grad_norm": 13.90625, "learning_rate": 8.993853041500854e-06, "loss": 0.6464, "step": 28817 }, { "epoch": 3.0663971057671846, "grad_norm": 14.328125, "learning_rate": 8.993817509948835e-06, "loss": 0.6472, "step": 28818 }, { "epoch": 3.066503511385401, "grad_norm": 11.640625, "learning_rate": 8.993781978396817e-06, "loss": 0.4992, "step": 28819 }, { "epoch": 3.0666099170036176, "grad_norm": 14.3671875, "learning_rate": 8.993746446844799e-06, "loss": 0.6053, "step": 28820 }, { "epoch": 3.0667163226218346, "grad_norm": 12.5546875, "learning_rate": 8.99371091529278e-06, "loss": 0.606, "step": 28821 }, { "epoch": 3.066822728240051, "grad_norm": 13.8359375, "learning_rate": 8.993675383740762e-06, "loss": 0.6277, "step": 28822 }, { "epoch": 3.0669291338582676, "grad_norm": 15.265625, "learning_rate": 8.993639852188745e-06, "loss": 0.6888, "step": 28823 }, { "epoch": 3.0670355394764846, "grad_norm": 14.2109375, "learning_rate": 8.993604320636727e-06, "loss": 0.5618, "step": 28824 }, { "epoch": 3.067141945094701, "grad_norm": 16.796875, "learning_rate": 8.993568789084708e-06, "loss": 0.6229, "step": 28825 }, { "epoch": 3.0672483507129176, "grad_norm": 12.9765625, "learning_rate": 8.99353325753269e-06, "loss": 0.5738, "step": 28826 }, { "epoch": 3.067354756331134, "grad_norm": 14.65625, "learning_rate": 8.993497725980672e-06, "loss": 0.5325, "step": 28827 }, { "epoch": 3.067461161949351, "grad_norm": 11.546875, "learning_rate": 8.993462194428653e-06, "loss": 0.644, "step": 28828 }, { "epoch": 3.0675675675675675, "grad_norm": 11.3203125, "learning_rate": 8.993426662876635e-06, "loss": 0.5919, "step": 28829 }, { "epoch": 3.067673973185784, "grad_norm": 16.75, "learning_rate": 8.993391131324616e-06, "loss": 0.7814, "step": 28830 }, { "epoch": 3.067780378804001, "grad_norm": 16.53125, "learning_rate": 8.993355599772598e-06, "loss": 0.7724, "step": 28831 }, { "epoch": 3.0678867844222175, "grad_norm": 9.6796875, "learning_rate": 8.99332006822058e-06, "loss": 0.4355, "step": 28832 }, { "epoch": 3.067993190040434, "grad_norm": 17.859375, "learning_rate": 8.993284536668563e-06, "loss": 0.6428, "step": 28833 }, { "epoch": 3.068099595658651, "grad_norm": 10.90625, "learning_rate": 8.993249005116544e-06, "loss": 0.5406, "step": 28834 }, { "epoch": 3.0682060012768675, "grad_norm": 15.2890625, "learning_rate": 8.993213473564526e-06, "loss": 0.6387, "step": 28835 }, { "epoch": 3.068312406895084, "grad_norm": 11.5234375, "learning_rate": 8.993177942012508e-06, "loss": 0.5863, "step": 28836 }, { "epoch": 3.0684188125133005, "grad_norm": 12.8359375, "learning_rate": 8.99314241046049e-06, "loss": 0.672, "step": 28837 }, { "epoch": 3.0685252181315175, "grad_norm": 10.53125, "learning_rate": 8.993106878908473e-06, "loss": 0.5801, "step": 28838 }, { "epoch": 3.068631623749734, "grad_norm": 12.28125, "learning_rate": 8.993071347356454e-06, "loss": 0.5527, "step": 28839 }, { "epoch": 3.0687380293679505, "grad_norm": 13.703125, "learning_rate": 8.993035815804436e-06, "loss": 0.6626, "step": 28840 }, { "epoch": 3.0688444349861674, "grad_norm": 10.625, "learning_rate": 8.993000284252417e-06, "loss": 0.6048, "step": 28841 }, { "epoch": 3.068950840604384, "grad_norm": 11.1953125, "learning_rate": 8.992964752700399e-06, "loss": 0.5801, "step": 28842 }, { "epoch": 3.0690572462226005, "grad_norm": 12.6796875, "learning_rate": 8.99292922114838e-06, "loss": 0.7842, "step": 28843 }, { "epoch": 3.0691636518408174, "grad_norm": 14.28125, "learning_rate": 8.992893689596362e-06, "loss": 0.7144, "step": 28844 }, { "epoch": 3.069270057459034, "grad_norm": 13.7265625, "learning_rate": 8.992858158044344e-06, "loss": 0.7213, "step": 28845 }, { "epoch": 3.0693764630772504, "grad_norm": 13.0234375, "learning_rate": 8.992822626492325e-06, "loss": 0.5391, "step": 28846 }, { "epoch": 3.069482868695467, "grad_norm": 12.859375, "learning_rate": 8.992787094940307e-06, "loss": 0.6682, "step": 28847 }, { "epoch": 3.069589274313684, "grad_norm": 11.546875, "learning_rate": 8.99275156338829e-06, "loss": 0.5457, "step": 28848 }, { "epoch": 3.0696956799319004, "grad_norm": 11.8671875, "learning_rate": 8.992716031836272e-06, "loss": 0.5737, "step": 28849 }, { "epoch": 3.069802085550117, "grad_norm": 21.09375, "learning_rate": 8.992680500284254e-06, "loss": 0.8673, "step": 28850 }, { "epoch": 3.069908491168334, "grad_norm": 11.71875, "learning_rate": 8.992644968732235e-06, "loss": 0.6106, "step": 28851 }, { "epoch": 3.0700148967865504, "grad_norm": 14.40625, "learning_rate": 8.992609437180217e-06, "loss": 0.6859, "step": 28852 }, { "epoch": 3.070121302404767, "grad_norm": 9.90625, "learning_rate": 8.992573905628198e-06, "loss": 0.4976, "step": 28853 }, { "epoch": 3.0702277080229834, "grad_norm": 10.046875, "learning_rate": 8.99253837407618e-06, "loss": 0.5223, "step": 28854 }, { "epoch": 3.0703341136412003, "grad_norm": 12.1796875, "learning_rate": 8.992502842524162e-06, "loss": 0.6412, "step": 28855 }, { "epoch": 3.070440519259417, "grad_norm": 15.2421875, "learning_rate": 8.992467310972143e-06, "loss": 0.7265, "step": 28856 }, { "epoch": 3.0705469248776334, "grad_norm": 15.1015625, "learning_rate": 8.992431779420125e-06, "loss": 0.5751, "step": 28857 }, { "epoch": 3.0706533304958503, "grad_norm": 17.71875, "learning_rate": 8.992396247868108e-06, "loss": 0.7233, "step": 28858 }, { "epoch": 3.070759736114067, "grad_norm": 12.8984375, "learning_rate": 8.99236071631609e-06, "loss": 0.638, "step": 28859 }, { "epoch": 3.0708661417322833, "grad_norm": 15.8671875, "learning_rate": 8.992325184764071e-06, "loss": 0.6321, "step": 28860 }, { "epoch": 3.0709725473505003, "grad_norm": 15.546875, "learning_rate": 8.992289653212053e-06, "loss": 0.7741, "step": 28861 }, { "epoch": 3.071078952968717, "grad_norm": 10.828125, "learning_rate": 8.992254121660035e-06, "loss": 0.6658, "step": 28862 }, { "epoch": 3.0711853585869333, "grad_norm": 12.234375, "learning_rate": 8.992218590108016e-06, "loss": 0.5335, "step": 28863 }, { "epoch": 3.07129176420515, "grad_norm": 11.8046875, "learning_rate": 8.992183058556e-06, "loss": 0.7019, "step": 28864 }, { "epoch": 3.0713981698233668, "grad_norm": 11.734375, "learning_rate": 8.992147527003981e-06, "loss": 0.6361, "step": 28865 }, { "epoch": 3.0715045754415833, "grad_norm": 12.4921875, "learning_rate": 8.992111995451963e-06, "loss": 0.6226, "step": 28866 }, { "epoch": 3.0716109810598, "grad_norm": 9.015625, "learning_rate": 8.992076463899944e-06, "loss": 0.3781, "step": 28867 }, { "epoch": 3.0717173866780167, "grad_norm": 13.21875, "learning_rate": 8.992040932347926e-06, "loss": 0.6129, "step": 28868 }, { "epoch": 3.0718237922962333, "grad_norm": 16.765625, "learning_rate": 8.992005400795907e-06, "loss": 0.6312, "step": 28869 }, { "epoch": 3.0719301979144498, "grad_norm": 12.765625, "learning_rate": 8.991969869243889e-06, "loss": 0.6287, "step": 28870 }, { "epoch": 3.0720366035326667, "grad_norm": 10.3984375, "learning_rate": 8.99193433769187e-06, "loss": 0.6006, "step": 28871 }, { "epoch": 3.0721430091508832, "grad_norm": 19.90625, "learning_rate": 8.991898806139852e-06, "loss": 0.8527, "step": 28872 }, { "epoch": 3.0722494147690997, "grad_norm": 10.5859375, "learning_rate": 8.991863274587834e-06, "loss": 0.5748, "step": 28873 }, { "epoch": 3.0723558203873162, "grad_norm": 12.0859375, "learning_rate": 8.991827743035817e-06, "loss": 0.6129, "step": 28874 }, { "epoch": 3.072462226005533, "grad_norm": 20.453125, "learning_rate": 8.991792211483799e-06, "loss": 0.7869, "step": 28875 }, { "epoch": 3.0725686316237497, "grad_norm": 10.8984375, "learning_rate": 8.99175667993178e-06, "loss": 0.5858, "step": 28876 }, { "epoch": 3.0726750372419662, "grad_norm": 12.6015625, "learning_rate": 8.991721148379762e-06, "loss": 0.5016, "step": 28877 }, { "epoch": 3.072781442860183, "grad_norm": 12.5078125, "learning_rate": 8.991685616827744e-06, "loss": 0.5849, "step": 28878 }, { "epoch": 3.0728878484783997, "grad_norm": 12.40625, "learning_rate": 8.991650085275725e-06, "loss": 0.4953, "step": 28879 }, { "epoch": 3.072994254096616, "grad_norm": 12.078125, "learning_rate": 8.991614553723707e-06, "loss": 0.5921, "step": 28880 }, { "epoch": 3.073100659714833, "grad_norm": 9.90625, "learning_rate": 8.991579022171688e-06, "loss": 0.4683, "step": 28881 }, { "epoch": 3.0732070653330497, "grad_norm": 13.7265625, "learning_rate": 8.99154349061967e-06, "loss": 0.7823, "step": 28882 }, { "epoch": 3.073313470951266, "grad_norm": 12.78125, "learning_rate": 8.991507959067652e-06, "loss": 0.559, "step": 28883 }, { "epoch": 3.0734198765694827, "grad_norm": 11.09375, "learning_rate": 8.991472427515635e-06, "loss": 0.6061, "step": 28884 }, { "epoch": 3.0735262821876996, "grad_norm": 11.875, "learning_rate": 8.991436895963617e-06, "loss": 0.6351, "step": 28885 }, { "epoch": 3.073632687805916, "grad_norm": 10.828125, "learning_rate": 8.991401364411598e-06, "loss": 0.6471, "step": 28886 }, { "epoch": 3.0737390934241327, "grad_norm": 11.59375, "learning_rate": 8.99136583285958e-06, "loss": 0.6304, "step": 28887 }, { "epoch": 3.0738454990423496, "grad_norm": 9.3984375, "learning_rate": 8.991330301307561e-06, "loss": 0.4802, "step": 28888 }, { "epoch": 3.073951904660566, "grad_norm": 16.328125, "learning_rate": 8.991294769755545e-06, "loss": 0.7494, "step": 28889 }, { "epoch": 3.0740583102787826, "grad_norm": 8.015625, "learning_rate": 8.991259238203526e-06, "loss": 0.4971, "step": 28890 }, { "epoch": 3.074164715896999, "grad_norm": 10.9609375, "learning_rate": 8.991223706651508e-06, "loss": 0.509, "step": 28891 }, { "epoch": 3.074271121515216, "grad_norm": 10.2578125, "learning_rate": 8.99118817509949e-06, "loss": 0.6949, "step": 28892 }, { "epoch": 3.0743775271334326, "grad_norm": 15.0, "learning_rate": 8.991152643547471e-06, "loss": 0.7414, "step": 28893 }, { "epoch": 3.074483932751649, "grad_norm": 11.9453125, "learning_rate": 8.991117111995453e-06, "loss": 0.6325, "step": 28894 }, { "epoch": 3.074590338369866, "grad_norm": 15.3203125, "learning_rate": 8.991081580443434e-06, "loss": 0.7594, "step": 28895 }, { "epoch": 3.0746967439880826, "grad_norm": 16.75, "learning_rate": 8.991046048891416e-06, "loss": 0.6924, "step": 28896 }, { "epoch": 3.074803149606299, "grad_norm": 12.828125, "learning_rate": 8.991010517339398e-06, "loss": 0.593, "step": 28897 }, { "epoch": 3.074909555224516, "grad_norm": 14.2734375, "learning_rate": 8.990974985787379e-06, "loss": 0.6889, "step": 28898 }, { "epoch": 3.0750159608427325, "grad_norm": 13.5859375, "learning_rate": 8.99093945423536e-06, "loss": 0.625, "step": 28899 }, { "epoch": 3.075122366460949, "grad_norm": 14.3984375, "learning_rate": 8.990903922683344e-06, "loss": 0.6623, "step": 28900 }, { "epoch": 3.075228772079166, "grad_norm": 15.6171875, "learning_rate": 8.990868391131326e-06, "loss": 0.7798, "step": 28901 }, { "epoch": 3.0753351776973825, "grad_norm": 14.1328125, "learning_rate": 8.990832859579307e-06, "loss": 0.6287, "step": 28902 }, { "epoch": 3.075441583315599, "grad_norm": 15.1171875, "learning_rate": 8.990797328027289e-06, "loss": 0.6134, "step": 28903 }, { "epoch": 3.0755479889338155, "grad_norm": 14.5859375, "learning_rate": 8.99076179647527e-06, "loss": 0.6389, "step": 28904 }, { "epoch": 3.0756543945520325, "grad_norm": 15.28125, "learning_rate": 8.990726264923254e-06, "loss": 0.6129, "step": 28905 }, { "epoch": 3.075760800170249, "grad_norm": 14.1640625, "learning_rate": 8.990690733371235e-06, "loss": 0.7599, "step": 28906 }, { "epoch": 3.0758672057884655, "grad_norm": 16.515625, "learning_rate": 8.990655201819215e-06, "loss": 0.5884, "step": 28907 }, { "epoch": 3.0759736114066825, "grad_norm": 12.859375, "learning_rate": 8.990619670267197e-06, "loss": 0.6299, "step": 28908 }, { "epoch": 3.076080017024899, "grad_norm": 11.84375, "learning_rate": 8.990584138715179e-06, "loss": 0.577, "step": 28909 }, { "epoch": 3.0761864226431155, "grad_norm": 13.7109375, "learning_rate": 8.990548607163162e-06, "loss": 0.6376, "step": 28910 }, { "epoch": 3.076292828261332, "grad_norm": 9.6640625, "learning_rate": 8.990513075611143e-06, "loss": 0.4757, "step": 28911 }, { "epoch": 3.076399233879549, "grad_norm": 10.515625, "learning_rate": 8.990477544059125e-06, "loss": 0.6039, "step": 28912 }, { "epoch": 3.0765056394977655, "grad_norm": 10.3671875, "learning_rate": 8.990442012507107e-06, "loss": 0.5918, "step": 28913 }, { "epoch": 3.076612045115982, "grad_norm": 12.03125, "learning_rate": 8.990406480955088e-06, "loss": 0.6039, "step": 28914 }, { "epoch": 3.076718450734199, "grad_norm": 12.8984375, "learning_rate": 8.990370949403072e-06, "loss": 0.7036, "step": 28915 }, { "epoch": 3.0768248563524154, "grad_norm": 13.4609375, "learning_rate": 8.990335417851053e-06, "loss": 0.5756, "step": 28916 }, { "epoch": 3.076931261970632, "grad_norm": 12.9921875, "learning_rate": 8.990299886299035e-06, "loss": 0.6693, "step": 28917 }, { "epoch": 3.077037667588849, "grad_norm": 16.25, "learning_rate": 8.990264354747016e-06, "loss": 0.6353, "step": 28918 }, { "epoch": 3.0771440732070654, "grad_norm": 8.5234375, "learning_rate": 8.990228823194998e-06, "loss": 0.4386, "step": 28919 }, { "epoch": 3.077250478825282, "grad_norm": 14.875, "learning_rate": 8.99019329164298e-06, "loss": 0.7258, "step": 28920 }, { "epoch": 3.0773568844434984, "grad_norm": 10.09375, "learning_rate": 8.990157760090961e-06, "loss": 0.4786, "step": 28921 }, { "epoch": 3.0774632900617154, "grad_norm": 12.6484375, "learning_rate": 8.990122228538943e-06, "loss": 0.5867, "step": 28922 }, { "epoch": 3.077569695679932, "grad_norm": 12.03125, "learning_rate": 8.990086696986924e-06, "loss": 0.7307, "step": 28923 }, { "epoch": 3.0776761012981484, "grad_norm": 10.65625, "learning_rate": 8.990051165434906e-06, "loss": 0.5772, "step": 28924 }, { "epoch": 3.0777825069163653, "grad_norm": 12.0, "learning_rate": 8.99001563388289e-06, "loss": 0.6596, "step": 28925 }, { "epoch": 3.077888912534582, "grad_norm": 11.9765625, "learning_rate": 8.989980102330871e-06, "loss": 0.563, "step": 28926 }, { "epoch": 3.0779953181527984, "grad_norm": 14.78125, "learning_rate": 8.989944570778853e-06, "loss": 0.6296, "step": 28927 }, { "epoch": 3.0781017237710153, "grad_norm": 14.7265625, "learning_rate": 8.989909039226834e-06, "loss": 0.6662, "step": 28928 }, { "epoch": 3.078208129389232, "grad_norm": 15.234375, "learning_rate": 8.989873507674816e-06, "loss": 0.7442, "step": 28929 }, { "epoch": 3.0783145350074483, "grad_norm": 14.375, "learning_rate": 8.989837976122799e-06, "loss": 0.6603, "step": 28930 }, { "epoch": 3.078420940625665, "grad_norm": 12.09375, "learning_rate": 8.98980244457078e-06, "loss": 0.6099, "step": 28931 }, { "epoch": 3.078527346243882, "grad_norm": 11.15625, "learning_rate": 8.989766913018762e-06, "loss": 0.5345, "step": 28932 }, { "epoch": 3.0786337518620983, "grad_norm": 11.015625, "learning_rate": 8.989731381466744e-06, "loss": 0.5304, "step": 28933 }, { "epoch": 3.078740157480315, "grad_norm": 18.640625, "learning_rate": 8.989695849914725e-06, "loss": 0.8356, "step": 28934 }, { "epoch": 3.0788465630985318, "grad_norm": 11.828125, "learning_rate": 8.989660318362707e-06, "loss": 0.7221, "step": 28935 }, { "epoch": 3.0789529687167483, "grad_norm": 8.609375, "learning_rate": 8.989624786810689e-06, "loss": 0.5787, "step": 28936 }, { "epoch": 3.079059374334965, "grad_norm": 15.0234375, "learning_rate": 8.98958925525867e-06, "loss": 0.6684, "step": 28937 }, { "epoch": 3.0791657799531817, "grad_norm": 10.7578125, "learning_rate": 8.989553723706652e-06, "loss": 0.5864, "step": 28938 }, { "epoch": 3.0792721855713983, "grad_norm": 15.5703125, "learning_rate": 8.989518192154633e-06, "loss": 0.5463, "step": 28939 }, { "epoch": 3.0793785911896148, "grad_norm": 14.1015625, "learning_rate": 8.989482660602615e-06, "loss": 0.759, "step": 28940 }, { "epoch": 3.0794849968078313, "grad_norm": 7.8515625, "learning_rate": 8.989447129050598e-06, "loss": 0.4063, "step": 28941 }, { "epoch": 3.0795914024260482, "grad_norm": 12.5625, "learning_rate": 8.98941159749858e-06, "loss": 0.5705, "step": 28942 }, { "epoch": 3.0796978080442647, "grad_norm": 13.921875, "learning_rate": 8.989376065946562e-06, "loss": 0.6338, "step": 28943 }, { "epoch": 3.0798042136624812, "grad_norm": 12.796875, "learning_rate": 8.989340534394543e-06, "loss": 0.5823, "step": 28944 }, { "epoch": 3.079910619280698, "grad_norm": 12.375, "learning_rate": 8.989305002842525e-06, "loss": 0.6395, "step": 28945 }, { "epoch": 3.0800170248989147, "grad_norm": 14.2109375, "learning_rate": 8.989269471290506e-06, "loss": 0.679, "step": 28946 }, { "epoch": 3.0801234305171312, "grad_norm": 16.34375, "learning_rate": 8.989233939738488e-06, "loss": 0.613, "step": 28947 }, { "epoch": 3.0802298361353477, "grad_norm": 10.140625, "learning_rate": 8.98919840818647e-06, "loss": 0.4676, "step": 28948 }, { "epoch": 3.0803362417535647, "grad_norm": 11.984375, "learning_rate": 8.989162876634451e-06, "loss": 0.647, "step": 28949 }, { "epoch": 3.080442647371781, "grad_norm": 11.21875, "learning_rate": 8.989127345082433e-06, "loss": 0.5062, "step": 28950 }, { "epoch": 3.0805490529899977, "grad_norm": 16.328125, "learning_rate": 8.989091813530416e-06, "loss": 0.6674, "step": 28951 }, { "epoch": 3.0806554586082147, "grad_norm": 11.9453125, "learning_rate": 8.989056281978398e-06, "loss": 0.5596, "step": 28952 }, { "epoch": 3.080761864226431, "grad_norm": 15.984375, "learning_rate": 8.98902075042638e-06, "loss": 0.7463, "step": 28953 }, { "epoch": 3.0808682698446477, "grad_norm": 14.828125, "learning_rate": 8.988985218874361e-06, "loss": 0.6618, "step": 28954 }, { "epoch": 3.0809746754628646, "grad_norm": 12.1640625, "learning_rate": 8.988949687322343e-06, "loss": 0.6084, "step": 28955 }, { "epoch": 3.081081081081081, "grad_norm": 13.8046875, "learning_rate": 8.988914155770326e-06, "loss": 0.5321, "step": 28956 }, { "epoch": 3.0811874866992977, "grad_norm": 12.0859375, "learning_rate": 8.988878624218307e-06, "loss": 0.6958, "step": 28957 }, { "epoch": 3.081293892317514, "grad_norm": 16.734375, "learning_rate": 8.988843092666289e-06, "loss": 0.7778, "step": 28958 }, { "epoch": 3.081400297935731, "grad_norm": 13.203125, "learning_rate": 8.98880756111427e-06, "loss": 0.6339, "step": 28959 }, { "epoch": 3.0815067035539476, "grad_norm": 15.484375, "learning_rate": 8.988772029562252e-06, "loss": 0.7039, "step": 28960 }, { "epoch": 3.081613109172164, "grad_norm": 12.3359375, "learning_rate": 8.988736498010234e-06, "loss": 0.6124, "step": 28961 }, { "epoch": 3.081719514790381, "grad_norm": 9.9375, "learning_rate": 8.988700966458216e-06, "loss": 0.5396, "step": 28962 }, { "epoch": 3.0818259204085976, "grad_norm": 12.671875, "learning_rate": 8.988665434906197e-06, "loss": 0.624, "step": 28963 }, { "epoch": 3.081932326026814, "grad_norm": 10.1796875, "learning_rate": 8.988629903354179e-06, "loss": 0.4227, "step": 28964 }, { "epoch": 3.082038731645031, "grad_norm": 13.328125, "learning_rate": 8.98859437180216e-06, "loss": 0.5878, "step": 28965 }, { "epoch": 3.0821451372632476, "grad_norm": 12.5078125, "learning_rate": 8.988558840250144e-06, "loss": 0.661, "step": 28966 }, { "epoch": 3.082251542881464, "grad_norm": 13.75, "learning_rate": 8.988523308698125e-06, "loss": 0.7024, "step": 28967 }, { "epoch": 3.0823579484996806, "grad_norm": 12.2734375, "learning_rate": 8.988487777146107e-06, "loss": 0.5809, "step": 28968 }, { "epoch": 3.0824643541178975, "grad_norm": 17.3125, "learning_rate": 8.988452245594088e-06, "loss": 0.694, "step": 28969 }, { "epoch": 3.082570759736114, "grad_norm": 13.0390625, "learning_rate": 8.98841671404207e-06, "loss": 0.6985, "step": 28970 }, { "epoch": 3.0826771653543306, "grad_norm": 10.8515625, "learning_rate": 8.988381182490052e-06, "loss": 0.4952, "step": 28971 }, { "epoch": 3.0827835709725475, "grad_norm": 15.53125, "learning_rate": 8.988345650938033e-06, "loss": 0.6587, "step": 28972 }, { "epoch": 3.082889976590764, "grad_norm": 13.4140625, "learning_rate": 8.988310119386015e-06, "loss": 0.688, "step": 28973 }, { "epoch": 3.0829963822089805, "grad_norm": 10.296875, "learning_rate": 8.988274587833996e-06, "loss": 0.6901, "step": 28974 }, { "epoch": 3.0831027878271975, "grad_norm": 14.1015625, "learning_rate": 8.988239056281978e-06, "loss": 0.7241, "step": 28975 }, { "epoch": 3.083209193445414, "grad_norm": 12.015625, "learning_rate": 8.988203524729961e-06, "loss": 0.6509, "step": 28976 }, { "epoch": 3.0833155990636305, "grad_norm": 14.59375, "learning_rate": 8.988167993177943e-06, "loss": 0.6359, "step": 28977 }, { "epoch": 3.083422004681847, "grad_norm": 14.2734375, "learning_rate": 8.988132461625925e-06, "loss": 0.624, "step": 28978 }, { "epoch": 3.083528410300064, "grad_norm": 11.7109375, "learning_rate": 8.988096930073906e-06, "loss": 0.6534, "step": 28979 }, { "epoch": 3.0836348159182805, "grad_norm": 10.265625, "learning_rate": 8.988061398521888e-06, "loss": 0.549, "step": 28980 }, { "epoch": 3.083741221536497, "grad_norm": 12.3203125, "learning_rate": 8.98802586696987e-06, "loss": 0.6431, "step": 28981 }, { "epoch": 3.083847627154714, "grad_norm": 9.9375, "learning_rate": 8.987990335417853e-06, "loss": 0.4965, "step": 28982 }, { "epoch": 3.0839540327729305, "grad_norm": 12.21875, "learning_rate": 8.987954803865834e-06, "loss": 0.604, "step": 28983 }, { "epoch": 3.084060438391147, "grad_norm": 13.34375, "learning_rate": 8.987919272313816e-06, "loss": 0.551, "step": 28984 }, { "epoch": 3.0841668440093635, "grad_norm": 10.71875, "learning_rate": 8.987883740761798e-06, "loss": 0.7326, "step": 28985 }, { "epoch": 3.0842732496275804, "grad_norm": 11.5859375, "learning_rate": 8.98784820920978e-06, "loss": 0.5115, "step": 28986 }, { "epoch": 3.084379655245797, "grad_norm": 11.1484375, "learning_rate": 8.98781267765776e-06, "loss": 0.6857, "step": 28987 }, { "epoch": 3.0844860608640134, "grad_norm": 14.5546875, "learning_rate": 8.987777146105742e-06, "loss": 0.9586, "step": 28988 }, { "epoch": 3.0845924664822304, "grad_norm": 19.4375, "learning_rate": 8.987741614553724e-06, "loss": 0.6883, "step": 28989 }, { "epoch": 3.084698872100447, "grad_norm": 16.46875, "learning_rate": 8.987706083001706e-06, "loss": 0.6798, "step": 28990 }, { "epoch": 3.0848052777186634, "grad_norm": 14.1484375, "learning_rate": 8.987670551449687e-06, "loss": 0.6381, "step": 28991 }, { "epoch": 3.0849116833368804, "grad_norm": 12.8125, "learning_rate": 8.98763501989767e-06, "loss": 0.5601, "step": 28992 }, { "epoch": 3.085018088955097, "grad_norm": 14.03125, "learning_rate": 8.987599488345652e-06, "loss": 0.6173, "step": 28993 }, { "epoch": 3.0851244945733134, "grad_norm": 13.8046875, "learning_rate": 8.987563956793634e-06, "loss": 0.6203, "step": 28994 }, { "epoch": 3.08523090019153, "grad_norm": 36.84375, "learning_rate": 8.987528425241615e-06, "loss": 0.7018, "step": 28995 }, { "epoch": 3.085337305809747, "grad_norm": 10.953125, "learning_rate": 8.987492893689597e-06, "loss": 0.5309, "step": 28996 }, { "epoch": 3.0854437114279634, "grad_norm": 13.484375, "learning_rate": 8.987457362137579e-06, "loss": 0.6943, "step": 28997 }, { "epoch": 3.08555011704618, "grad_norm": 13.6328125, "learning_rate": 8.98742183058556e-06, "loss": 0.582, "step": 28998 }, { "epoch": 3.085656522664397, "grad_norm": 13.8984375, "learning_rate": 8.987386299033542e-06, "loss": 0.6056, "step": 28999 }, { "epoch": 3.0857629282826133, "grad_norm": 11.0, "learning_rate": 8.987350767481523e-06, "loss": 0.546, "step": 29000 }, { "epoch": 3.08586933390083, "grad_norm": 10.375, "learning_rate": 8.987315235929505e-06, "loss": 0.4913, "step": 29001 }, { "epoch": 3.085975739519047, "grad_norm": 12.9921875, "learning_rate": 8.987279704377488e-06, "loss": 0.6176, "step": 29002 }, { "epoch": 3.0860821451372633, "grad_norm": 11.1953125, "learning_rate": 8.98724417282547e-06, "loss": 0.6155, "step": 29003 }, { "epoch": 3.08618855075548, "grad_norm": 15.7421875, "learning_rate": 8.987208641273451e-06, "loss": 0.6817, "step": 29004 }, { "epoch": 3.0862949563736963, "grad_norm": 12.7421875, "learning_rate": 8.987173109721433e-06, "loss": 0.7084, "step": 29005 }, { "epoch": 3.0864013619919133, "grad_norm": 12.1640625, "learning_rate": 8.987137578169415e-06, "loss": 0.5491, "step": 29006 }, { "epoch": 3.08650776761013, "grad_norm": 12.8046875, "learning_rate": 8.987102046617398e-06, "loss": 0.6099, "step": 29007 }, { "epoch": 3.0866141732283463, "grad_norm": 12.59375, "learning_rate": 8.98706651506538e-06, "loss": 0.6067, "step": 29008 }, { "epoch": 3.0867205788465633, "grad_norm": 11.65625, "learning_rate": 8.987030983513361e-06, "loss": 0.5113, "step": 29009 }, { "epoch": 3.0868269844647798, "grad_norm": 14.5859375, "learning_rate": 8.986995451961343e-06, "loss": 0.6206, "step": 29010 }, { "epoch": 3.0869333900829963, "grad_norm": 10.5234375, "learning_rate": 8.986959920409324e-06, "loss": 0.4643, "step": 29011 }, { "epoch": 3.0870397957012132, "grad_norm": 11.125, "learning_rate": 8.986924388857306e-06, "loss": 0.5091, "step": 29012 }, { "epoch": 3.0871462013194297, "grad_norm": 12.2109375, "learning_rate": 8.986888857305288e-06, "loss": 0.5487, "step": 29013 }, { "epoch": 3.0872526069376462, "grad_norm": 12.421875, "learning_rate": 8.98685332575327e-06, "loss": 0.5934, "step": 29014 }, { "epoch": 3.0873590125558628, "grad_norm": 15.234375, "learning_rate": 8.98681779420125e-06, "loss": 0.8488, "step": 29015 }, { "epoch": 3.0874654181740797, "grad_norm": 10.6171875, "learning_rate": 8.986782262649232e-06, "loss": 0.5087, "step": 29016 }, { "epoch": 3.087571823792296, "grad_norm": 12.59375, "learning_rate": 8.986746731097214e-06, "loss": 0.5624, "step": 29017 }, { "epoch": 3.0876782294105127, "grad_norm": 9.96875, "learning_rate": 8.986711199545197e-06, "loss": 0.5342, "step": 29018 }, { "epoch": 3.0877846350287297, "grad_norm": 18.421875, "learning_rate": 8.986675667993179e-06, "loss": 0.7173, "step": 29019 }, { "epoch": 3.087891040646946, "grad_norm": 14.2890625, "learning_rate": 8.98664013644116e-06, "loss": 0.7273, "step": 29020 }, { "epoch": 3.0879974462651627, "grad_norm": 16.078125, "learning_rate": 8.986604604889142e-06, "loss": 0.5651, "step": 29021 }, { "epoch": 3.088103851883379, "grad_norm": 11.5078125, "learning_rate": 8.986569073337124e-06, "loss": 0.5992, "step": 29022 }, { "epoch": 3.088210257501596, "grad_norm": 9.7421875, "learning_rate": 8.986533541785107e-06, "loss": 0.4982, "step": 29023 }, { "epoch": 3.0883166631198127, "grad_norm": 16.5, "learning_rate": 8.986498010233089e-06, "loss": 0.8316, "step": 29024 }, { "epoch": 3.088423068738029, "grad_norm": 15.9140625, "learning_rate": 8.98646247868107e-06, "loss": 0.6718, "step": 29025 }, { "epoch": 3.088529474356246, "grad_norm": 14.8125, "learning_rate": 8.986426947129052e-06, "loss": 0.6762, "step": 29026 }, { "epoch": 3.0886358799744627, "grad_norm": 11.8359375, "learning_rate": 8.986391415577032e-06, "loss": 0.5813, "step": 29027 }, { "epoch": 3.088742285592679, "grad_norm": 12.375, "learning_rate": 8.986355884025015e-06, "loss": 0.5524, "step": 29028 }, { "epoch": 3.088848691210896, "grad_norm": 15.0390625, "learning_rate": 8.986320352472997e-06, "loss": 0.611, "step": 29029 }, { "epoch": 3.0889550968291126, "grad_norm": 14.625, "learning_rate": 8.986284820920978e-06, "loss": 0.7517, "step": 29030 }, { "epoch": 3.089061502447329, "grad_norm": 12.0703125, "learning_rate": 8.98624928936896e-06, "loss": 0.6557, "step": 29031 }, { "epoch": 3.089167908065546, "grad_norm": 16.65625, "learning_rate": 8.986213757816942e-06, "loss": 0.6562, "step": 29032 }, { "epoch": 3.0892743136837626, "grad_norm": 18.28125, "learning_rate": 8.986178226264925e-06, "loss": 0.6258, "step": 29033 }, { "epoch": 3.089380719301979, "grad_norm": 13.875, "learning_rate": 8.986142694712906e-06, "loss": 0.7124, "step": 29034 }, { "epoch": 3.0894871249201956, "grad_norm": 13.2109375, "learning_rate": 8.986107163160888e-06, "loss": 0.5226, "step": 29035 }, { "epoch": 3.0895935305384126, "grad_norm": 13.5625, "learning_rate": 8.98607163160887e-06, "loss": 0.598, "step": 29036 }, { "epoch": 3.089699936156629, "grad_norm": 14.0546875, "learning_rate": 8.986036100056851e-06, "loss": 0.5732, "step": 29037 }, { "epoch": 3.0898063417748456, "grad_norm": 12.125, "learning_rate": 8.986000568504833e-06, "loss": 0.5084, "step": 29038 }, { "epoch": 3.0899127473930625, "grad_norm": 13.8828125, "learning_rate": 8.985965036952814e-06, "loss": 0.6219, "step": 29039 }, { "epoch": 3.090019153011279, "grad_norm": 15.078125, "learning_rate": 8.985929505400796e-06, "loss": 0.7913, "step": 29040 }, { "epoch": 3.0901255586294956, "grad_norm": 15.03125, "learning_rate": 8.985893973848778e-06, "loss": 0.5544, "step": 29041 }, { "epoch": 3.090231964247712, "grad_norm": 11.8828125, "learning_rate": 8.98585844229676e-06, "loss": 0.5724, "step": 29042 }, { "epoch": 3.090338369865929, "grad_norm": 14.21875, "learning_rate": 8.985822910744743e-06, "loss": 0.7568, "step": 29043 }, { "epoch": 3.0904447754841455, "grad_norm": 13.40625, "learning_rate": 8.985787379192724e-06, "loss": 0.5691, "step": 29044 }, { "epoch": 3.090551181102362, "grad_norm": 12.9765625, "learning_rate": 8.985751847640706e-06, "loss": 0.6299, "step": 29045 }, { "epoch": 3.090657586720579, "grad_norm": 14.125, "learning_rate": 8.985716316088687e-06, "loss": 0.6553, "step": 29046 }, { "epoch": 3.0907639923387955, "grad_norm": 10.5625, "learning_rate": 8.985680784536669e-06, "loss": 0.5144, "step": 29047 }, { "epoch": 3.090870397957012, "grad_norm": 15.6796875, "learning_rate": 8.985645252984652e-06, "loss": 0.7088, "step": 29048 }, { "epoch": 3.090976803575229, "grad_norm": 14.359375, "learning_rate": 8.985609721432634e-06, "loss": 0.7518, "step": 29049 }, { "epoch": 3.0910832091934455, "grad_norm": 12.3828125, "learning_rate": 8.985574189880616e-06, "loss": 0.6469, "step": 29050 }, { "epoch": 3.091189614811662, "grad_norm": 12.9296875, "learning_rate": 8.985538658328597e-06, "loss": 0.573, "step": 29051 }, { "epoch": 3.0912960204298785, "grad_norm": 12.2265625, "learning_rate": 8.985503126776579e-06, "loss": 0.4229, "step": 29052 }, { "epoch": 3.0914024260480955, "grad_norm": 11.828125, "learning_rate": 8.98546759522456e-06, "loss": 0.5915, "step": 29053 }, { "epoch": 3.091508831666312, "grad_norm": 12.0625, "learning_rate": 8.985432063672542e-06, "loss": 0.6145, "step": 29054 }, { "epoch": 3.0916152372845285, "grad_norm": 15.6875, "learning_rate": 8.985396532120524e-06, "loss": 0.8436, "step": 29055 }, { "epoch": 3.0917216429027454, "grad_norm": 12.734375, "learning_rate": 8.985361000568505e-06, "loss": 0.5808, "step": 29056 }, { "epoch": 3.091828048520962, "grad_norm": 9.703125, "learning_rate": 8.985325469016487e-06, "loss": 0.5368, "step": 29057 }, { "epoch": 3.0919344541391784, "grad_norm": 11.75, "learning_rate": 8.985289937464468e-06, "loss": 0.6148, "step": 29058 }, { "epoch": 3.0920408597573954, "grad_norm": 9.9140625, "learning_rate": 8.985254405912452e-06, "loss": 0.4606, "step": 29059 }, { "epoch": 3.092147265375612, "grad_norm": 13.09375, "learning_rate": 8.985218874360433e-06, "loss": 0.5048, "step": 29060 }, { "epoch": 3.0922536709938284, "grad_norm": 12.5625, "learning_rate": 8.985183342808415e-06, "loss": 0.4795, "step": 29061 }, { "epoch": 3.092360076612045, "grad_norm": 14.0, "learning_rate": 8.985147811256397e-06, "loss": 0.6865, "step": 29062 }, { "epoch": 3.092466482230262, "grad_norm": 13.9296875, "learning_rate": 8.985112279704378e-06, "loss": 0.5313, "step": 29063 }, { "epoch": 3.0925728878484784, "grad_norm": 14.59375, "learning_rate": 8.98507674815236e-06, "loss": 0.5082, "step": 29064 }, { "epoch": 3.092679293466695, "grad_norm": 12.6328125, "learning_rate": 8.985041216600341e-06, "loss": 0.593, "step": 29065 }, { "epoch": 3.092785699084912, "grad_norm": 13.40625, "learning_rate": 8.985005685048323e-06, "loss": 0.7945, "step": 29066 }, { "epoch": 3.0928921047031284, "grad_norm": 16.65625, "learning_rate": 8.984970153496305e-06, "loss": 0.7447, "step": 29067 }, { "epoch": 3.092998510321345, "grad_norm": 14.671875, "learning_rate": 8.984934621944286e-06, "loss": 0.7956, "step": 29068 }, { "epoch": 3.093104915939562, "grad_norm": 11.7421875, "learning_rate": 8.98489909039227e-06, "loss": 0.7011, "step": 29069 }, { "epoch": 3.0932113215577783, "grad_norm": 16.171875, "learning_rate": 8.984863558840251e-06, "loss": 0.5921, "step": 29070 }, { "epoch": 3.093317727175995, "grad_norm": 9.34375, "learning_rate": 8.984828027288233e-06, "loss": 0.6803, "step": 29071 }, { "epoch": 3.0934241327942114, "grad_norm": 13.75, "learning_rate": 8.984792495736214e-06, "loss": 0.7781, "step": 29072 }, { "epoch": 3.0935305384124283, "grad_norm": 14.890625, "learning_rate": 8.984756964184196e-06, "loss": 0.611, "step": 29073 }, { "epoch": 3.093636944030645, "grad_norm": 12.015625, "learning_rate": 8.98472143263218e-06, "loss": 0.6388, "step": 29074 }, { "epoch": 3.0937433496488613, "grad_norm": 17.828125, "learning_rate": 8.98468590108016e-06, "loss": 0.6944, "step": 29075 }, { "epoch": 3.0938497552670783, "grad_norm": 9.890625, "learning_rate": 8.984650369528142e-06, "loss": 0.4816, "step": 29076 }, { "epoch": 3.093956160885295, "grad_norm": 17.03125, "learning_rate": 8.984614837976124e-06, "loss": 0.8691, "step": 29077 }, { "epoch": 3.0940625665035113, "grad_norm": 11.2734375, "learning_rate": 8.984579306424106e-06, "loss": 0.6074, "step": 29078 }, { "epoch": 3.094168972121728, "grad_norm": 15.1484375, "learning_rate": 8.984543774872087e-06, "loss": 0.6979, "step": 29079 }, { "epoch": 3.0942753777399448, "grad_norm": 12.1953125, "learning_rate": 8.984508243320069e-06, "loss": 0.5813, "step": 29080 }, { "epoch": 3.0943817833581613, "grad_norm": 16.25, "learning_rate": 8.98447271176805e-06, "loss": 0.7123, "step": 29081 }, { "epoch": 3.094488188976378, "grad_norm": 14.640625, "learning_rate": 8.984437180216032e-06, "loss": 0.6951, "step": 29082 }, { "epoch": 3.0945945945945947, "grad_norm": 12.125, "learning_rate": 8.984401648664014e-06, "loss": 0.6923, "step": 29083 }, { "epoch": 3.0947010002128112, "grad_norm": 13.0, "learning_rate": 8.984366117111997e-06, "loss": 0.7573, "step": 29084 }, { "epoch": 3.0948074058310278, "grad_norm": 13.8828125, "learning_rate": 8.984330585559979e-06, "loss": 0.6303, "step": 29085 }, { "epoch": 3.0949138114492447, "grad_norm": 10.359375, "learning_rate": 8.98429505400796e-06, "loss": 0.5997, "step": 29086 }, { "epoch": 3.095020217067461, "grad_norm": 13.0546875, "learning_rate": 8.984259522455942e-06, "loss": 0.665, "step": 29087 }, { "epoch": 3.0951266226856777, "grad_norm": 15.3984375, "learning_rate": 8.984223990903923e-06, "loss": 0.7439, "step": 29088 }, { "epoch": 3.0952330283038942, "grad_norm": 14.640625, "learning_rate": 8.984188459351905e-06, "loss": 0.5478, "step": 29089 }, { "epoch": 3.095339433922111, "grad_norm": 9.59375, "learning_rate": 8.984152927799887e-06, "loss": 0.4546, "step": 29090 }, { "epoch": 3.0954458395403277, "grad_norm": 14.328125, "learning_rate": 8.984117396247868e-06, "loss": 0.5757, "step": 29091 }, { "epoch": 3.095552245158544, "grad_norm": 12.3203125, "learning_rate": 8.98408186469585e-06, "loss": 0.6111, "step": 29092 }, { "epoch": 3.095658650776761, "grad_norm": 13.4921875, "learning_rate": 8.984046333143831e-06, "loss": 0.5626, "step": 29093 }, { "epoch": 3.0957650563949777, "grad_norm": 13.0078125, "learning_rate": 8.984010801591813e-06, "loss": 0.648, "step": 29094 }, { "epoch": 3.095871462013194, "grad_norm": 14.390625, "learning_rate": 8.983975270039796e-06, "loss": 0.7049, "step": 29095 }, { "epoch": 3.095977867631411, "grad_norm": 14.953125, "learning_rate": 8.983939738487778e-06, "loss": 0.6812, "step": 29096 }, { "epoch": 3.0960842732496276, "grad_norm": 13.671875, "learning_rate": 8.98390420693576e-06, "loss": 0.5954, "step": 29097 }, { "epoch": 3.096190678867844, "grad_norm": 12.3671875, "learning_rate": 8.983868675383741e-06, "loss": 0.5893, "step": 29098 }, { "epoch": 3.0962970844860607, "grad_norm": 10.1171875, "learning_rate": 8.983833143831723e-06, "loss": 0.5148, "step": 29099 }, { "epoch": 3.0964034901042776, "grad_norm": 21.890625, "learning_rate": 8.983797612279706e-06, "loss": 0.889, "step": 29100 }, { "epoch": 3.096509895722494, "grad_norm": 9.765625, "learning_rate": 8.983762080727688e-06, "loss": 0.5083, "step": 29101 }, { "epoch": 3.0966163013407106, "grad_norm": 11.640625, "learning_rate": 8.98372654917567e-06, "loss": 0.5617, "step": 29102 }, { "epoch": 3.0967227069589276, "grad_norm": 17.390625, "learning_rate": 8.983691017623651e-06, "loss": 0.664, "step": 29103 }, { "epoch": 3.096829112577144, "grad_norm": 16.828125, "learning_rate": 8.983655486071632e-06, "loss": 0.7286, "step": 29104 }, { "epoch": 3.0969355181953606, "grad_norm": 16.75, "learning_rate": 8.983619954519614e-06, "loss": 0.6843, "step": 29105 }, { "epoch": 3.0970419238135776, "grad_norm": 11.8046875, "learning_rate": 8.983584422967596e-06, "loss": 0.6993, "step": 29106 }, { "epoch": 3.097148329431794, "grad_norm": 11.1484375, "learning_rate": 8.983548891415577e-06, "loss": 0.6277, "step": 29107 }, { "epoch": 3.0972547350500106, "grad_norm": 14.2890625, "learning_rate": 8.983513359863559e-06, "loss": 0.7216, "step": 29108 }, { "epoch": 3.097361140668227, "grad_norm": 12.4921875, "learning_rate": 8.98347782831154e-06, "loss": 0.5472, "step": 29109 }, { "epoch": 3.097467546286444, "grad_norm": 11.4453125, "learning_rate": 8.983442296759524e-06, "loss": 0.6127, "step": 29110 }, { "epoch": 3.0975739519046606, "grad_norm": 11.109375, "learning_rate": 8.983406765207505e-06, "loss": 0.6119, "step": 29111 }, { "epoch": 3.097680357522877, "grad_norm": 15.1640625, "learning_rate": 8.983371233655487e-06, "loss": 0.7287, "step": 29112 }, { "epoch": 3.097786763141094, "grad_norm": 12.3515625, "learning_rate": 8.983335702103469e-06, "loss": 0.564, "step": 29113 }, { "epoch": 3.0978931687593105, "grad_norm": 14.0078125, "learning_rate": 8.98330017055145e-06, "loss": 0.8278, "step": 29114 }, { "epoch": 3.097999574377527, "grad_norm": 11.9609375, "learning_rate": 8.983264638999434e-06, "loss": 0.4547, "step": 29115 }, { "epoch": 3.0981059799957436, "grad_norm": 12.0234375, "learning_rate": 8.983229107447413e-06, "loss": 0.6545, "step": 29116 }, { "epoch": 3.0982123856139605, "grad_norm": 14.6484375, "learning_rate": 8.983193575895395e-06, "loss": 0.8046, "step": 29117 }, { "epoch": 3.098318791232177, "grad_norm": 12.484375, "learning_rate": 8.983158044343377e-06, "loss": 0.529, "step": 29118 }, { "epoch": 3.0984251968503935, "grad_norm": 14.1328125, "learning_rate": 8.983122512791358e-06, "loss": 0.6809, "step": 29119 }, { "epoch": 3.0985316024686105, "grad_norm": 14.0703125, "learning_rate": 8.983086981239342e-06, "loss": 0.6963, "step": 29120 }, { "epoch": 3.098638008086827, "grad_norm": 13.90625, "learning_rate": 8.983051449687323e-06, "loss": 0.6073, "step": 29121 }, { "epoch": 3.0987444137050435, "grad_norm": 11.609375, "learning_rate": 8.983015918135305e-06, "loss": 0.5875, "step": 29122 }, { "epoch": 3.0988508193232605, "grad_norm": 12.0546875, "learning_rate": 8.982980386583286e-06, "loss": 0.749, "step": 29123 }, { "epoch": 3.098957224941477, "grad_norm": 14.6484375, "learning_rate": 8.982944855031268e-06, "loss": 0.6862, "step": 29124 }, { "epoch": 3.0990636305596935, "grad_norm": 13.9609375, "learning_rate": 8.982909323479251e-06, "loss": 0.6244, "step": 29125 }, { "epoch": 3.09917003617791, "grad_norm": 13.1640625, "learning_rate": 8.982873791927233e-06, "loss": 0.5405, "step": 29126 }, { "epoch": 3.099276441796127, "grad_norm": 11.0859375, "learning_rate": 8.982838260375215e-06, "loss": 0.5864, "step": 29127 }, { "epoch": 3.0993828474143434, "grad_norm": 16.421875, "learning_rate": 8.982802728823196e-06, "loss": 0.5449, "step": 29128 }, { "epoch": 3.09948925303256, "grad_norm": 12.96875, "learning_rate": 8.982767197271178e-06, "loss": 0.6456, "step": 29129 }, { "epoch": 3.099595658650777, "grad_norm": 11.6796875, "learning_rate": 8.98273166571916e-06, "loss": 0.7185, "step": 29130 }, { "epoch": 3.0997020642689934, "grad_norm": 13.6484375, "learning_rate": 8.982696134167141e-06, "loss": 0.6091, "step": 29131 }, { "epoch": 3.09980846988721, "grad_norm": 10.8046875, "learning_rate": 8.982660602615123e-06, "loss": 0.5621, "step": 29132 }, { "epoch": 3.099914875505427, "grad_norm": 13.921875, "learning_rate": 8.982625071063104e-06, "loss": 0.6454, "step": 29133 }, { "epoch": 3.1000212811236434, "grad_norm": 15.1640625, "learning_rate": 8.982589539511086e-06, "loss": 0.7677, "step": 29134 }, { "epoch": 3.10012768674186, "grad_norm": 20.34375, "learning_rate": 8.982554007959067e-06, "loss": 0.7359, "step": 29135 }, { "epoch": 3.1002340923600764, "grad_norm": 9.9296875, "learning_rate": 8.98251847640705e-06, "loss": 0.466, "step": 29136 }, { "epoch": 3.1003404979782934, "grad_norm": 10.6875, "learning_rate": 8.982482944855032e-06, "loss": 0.4946, "step": 29137 }, { "epoch": 3.10044690359651, "grad_norm": 10.8046875, "learning_rate": 8.982447413303014e-06, "loss": 0.4977, "step": 29138 }, { "epoch": 3.1005533092147264, "grad_norm": 12.390625, "learning_rate": 8.982411881750995e-06, "loss": 0.5406, "step": 29139 }, { "epoch": 3.1006597148329433, "grad_norm": 17.296875, "learning_rate": 8.982376350198977e-06, "loss": 0.6212, "step": 29140 }, { "epoch": 3.10076612045116, "grad_norm": 12.8984375, "learning_rate": 8.98234081864696e-06, "loss": 0.4976, "step": 29141 }, { "epoch": 3.1008725260693764, "grad_norm": 11.4296875, "learning_rate": 8.982305287094942e-06, "loss": 0.5857, "step": 29142 }, { "epoch": 3.1009789316875933, "grad_norm": 13.6015625, "learning_rate": 8.982269755542924e-06, "loss": 0.545, "step": 29143 }, { "epoch": 3.10108533730581, "grad_norm": 12.71875, "learning_rate": 8.982234223990905e-06, "loss": 0.6719, "step": 29144 }, { "epoch": 3.1011917429240263, "grad_norm": 12.9453125, "learning_rate": 8.982198692438887e-06, "loss": 0.6448, "step": 29145 }, { "epoch": 3.101298148542243, "grad_norm": 10.484375, "learning_rate": 8.982163160886868e-06, "loss": 0.5029, "step": 29146 }, { "epoch": 3.10140455416046, "grad_norm": 20.4375, "learning_rate": 8.98212762933485e-06, "loss": 0.6918, "step": 29147 }, { "epoch": 3.1015109597786763, "grad_norm": 11.296875, "learning_rate": 8.982092097782832e-06, "loss": 0.5118, "step": 29148 }, { "epoch": 3.101617365396893, "grad_norm": 11.625, "learning_rate": 8.982056566230813e-06, "loss": 0.5866, "step": 29149 }, { "epoch": 3.1017237710151098, "grad_norm": 15.640625, "learning_rate": 8.982021034678795e-06, "loss": 0.6385, "step": 29150 }, { "epoch": 3.1018301766333263, "grad_norm": 13.6875, "learning_rate": 8.981985503126778e-06, "loss": 0.779, "step": 29151 }, { "epoch": 3.101936582251543, "grad_norm": 12.8046875, "learning_rate": 8.98194997157476e-06, "loss": 0.5763, "step": 29152 }, { "epoch": 3.1020429878697597, "grad_norm": 12.625, "learning_rate": 8.981914440022741e-06, "loss": 0.6533, "step": 29153 }, { "epoch": 3.1021493934879762, "grad_norm": 11.8828125, "learning_rate": 8.981878908470723e-06, "loss": 0.5054, "step": 29154 }, { "epoch": 3.1022557991061928, "grad_norm": 12.65625, "learning_rate": 8.981843376918705e-06, "loss": 0.5621, "step": 29155 }, { "epoch": 3.1023622047244093, "grad_norm": 12.5859375, "learning_rate": 8.981807845366686e-06, "loss": 0.5841, "step": 29156 }, { "epoch": 3.102468610342626, "grad_norm": 16.140625, "learning_rate": 8.981772313814668e-06, "loss": 0.6101, "step": 29157 }, { "epoch": 3.1025750159608427, "grad_norm": 9.8359375, "learning_rate": 8.98173678226265e-06, "loss": 0.4496, "step": 29158 }, { "epoch": 3.1026814215790592, "grad_norm": 11.8046875, "learning_rate": 8.981701250710631e-06, "loss": 0.5364, "step": 29159 }, { "epoch": 3.102787827197276, "grad_norm": 9.9140625, "learning_rate": 8.981665719158613e-06, "loss": 0.4667, "step": 29160 }, { "epoch": 3.1028942328154927, "grad_norm": 19.6875, "learning_rate": 8.981630187606596e-06, "loss": 0.6986, "step": 29161 }, { "epoch": 3.103000638433709, "grad_norm": 14.6171875, "learning_rate": 8.981594656054578e-06, "loss": 0.4995, "step": 29162 }, { "epoch": 3.103107044051926, "grad_norm": 12.96875, "learning_rate": 8.981559124502559e-06, "loss": 0.5723, "step": 29163 }, { "epoch": 3.1032134496701427, "grad_norm": 13.515625, "learning_rate": 8.98152359295054e-06, "loss": 0.6778, "step": 29164 }, { "epoch": 3.103319855288359, "grad_norm": 12.265625, "learning_rate": 8.981488061398522e-06, "loss": 0.5682, "step": 29165 }, { "epoch": 3.1034262609065757, "grad_norm": 16.734375, "learning_rate": 8.981452529846506e-06, "loss": 0.7483, "step": 29166 }, { "epoch": 3.1035326665247926, "grad_norm": 13.6953125, "learning_rate": 8.981416998294487e-06, "loss": 0.7436, "step": 29167 }, { "epoch": 3.103639072143009, "grad_norm": 13.1640625, "learning_rate": 8.981381466742469e-06, "loss": 0.6422, "step": 29168 }, { "epoch": 3.1037454777612257, "grad_norm": 11.03125, "learning_rate": 8.98134593519045e-06, "loss": 0.581, "step": 29169 }, { "epoch": 3.1038518833794426, "grad_norm": 14.984375, "learning_rate": 8.981310403638432e-06, "loss": 0.7608, "step": 29170 }, { "epoch": 3.103958288997659, "grad_norm": 14.5859375, "learning_rate": 8.981274872086414e-06, "loss": 0.8123, "step": 29171 }, { "epoch": 3.1040646946158756, "grad_norm": 12.171875, "learning_rate": 8.981239340534395e-06, "loss": 0.4831, "step": 29172 }, { "epoch": 3.104171100234092, "grad_norm": 10.4140625, "learning_rate": 8.981203808982377e-06, "loss": 0.6677, "step": 29173 }, { "epoch": 3.104277505852309, "grad_norm": 12.078125, "learning_rate": 8.981168277430358e-06, "loss": 0.6201, "step": 29174 }, { "epoch": 3.1043839114705256, "grad_norm": 12.0, "learning_rate": 8.98113274587834e-06, "loss": 0.6669, "step": 29175 }, { "epoch": 3.104490317088742, "grad_norm": 13.09375, "learning_rate": 8.981097214326322e-06, "loss": 0.7745, "step": 29176 }, { "epoch": 3.104596722706959, "grad_norm": 10.84375, "learning_rate": 8.981061682774305e-06, "loss": 0.5225, "step": 29177 }, { "epoch": 3.1047031283251756, "grad_norm": 11.671875, "learning_rate": 8.981026151222287e-06, "loss": 0.5847, "step": 29178 }, { "epoch": 3.104809533943392, "grad_norm": 9.796875, "learning_rate": 8.980990619670268e-06, "loss": 0.43, "step": 29179 }, { "epoch": 3.104915939561609, "grad_norm": 16.125, "learning_rate": 8.98095508811825e-06, "loss": 0.729, "step": 29180 }, { "epoch": 3.1050223451798256, "grad_norm": 19.40625, "learning_rate": 8.980919556566231e-06, "loss": 0.7725, "step": 29181 }, { "epoch": 3.105128750798042, "grad_norm": 17.734375, "learning_rate": 8.980884025014213e-06, "loss": 0.6976, "step": 29182 }, { "epoch": 3.1052351564162586, "grad_norm": 10.25, "learning_rate": 8.980848493462195e-06, "loss": 0.5756, "step": 29183 }, { "epoch": 3.1053415620344755, "grad_norm": 10.0078125, "learning_rate": 8.980812961910176e-06, "loss": 0.5906, "step": 29184 }, { "epoch": 3.105447967652692, "grad_norm": 14.3046875, "learning_rate": 8.980777430358158e-06, "loss": 0.5862, "step": 29185 }, { "epoch": 3.1055543732709086, "grad_norm": 13.984375, "learning_rate": 8.98074189880614e-06, "loss": 0.5722, "step": 29186 }, { "epoch": 3.1056607788891255, "grad_norm": 11.6953125, "learning_rate": 8.980706367254123e-06, "loss": 0.7246, "step": 29187 }, { "epoch": 3.105767184507342, "grad_norm": 13.6484375, "learning_rate": 8.980670835702104e-06, "loss": 0.6108, "step": 29188 }, { "epoch": 3.1058735901255585, "grad_norm": 11.4140625, "learning_rate": 8.980635304150086e-06, "loss": 0.6225, "step": 29189 }, { "epoch": 3.1059799957437755, "grad_norm": 13.7578125, "learning_rate": 8.980599772598068e-06, "loss": 0.6015, "step": 29190 }, { "epoch": 3.106086401361992, "grad_norm": 12.8984375, "learning_rate": 8.98056424104605e-06, "loss": 0.6737, "step": 29191 }, { "epoch": 3.1061928069802085, "grad_norm": 9.09375, "learning_rate": 8.980528709494032e-06, "loss": 0.4548, "step": 29192 }, { "epoch": 3.106299212598425, "grad_norm": 15.8984375, "learning_rate": 8.980493177942014e-06, "loss": 0.6157, "step": 29193 }, { "epoch": 3.106405618216642, "grad_norm": 11.6953125, "learning_rate": 8.980457646389996e-06, "loss": 0.5213, "step": 29194 }, { "epoch": 3.1065120238348585, "grad_norm": 12.84375, "learning_rate": 8.980422114837977e-06, "loss": 0.5331, "step": 29195 }, { "epoch": 3.106618429453075, "grad_norm": 13.515625, "learning_rate": 8.980386583285959e-06, "loss": 0.6805, "step": 29196 }, { "epoch": 3.106724835071292, "grad_norm": 48.1875, "learning_rate": 8.98035105173394e-06, "loss": 0.5159, "step": 29197 }, { "epoch": 3.1068312406895084, "grad_norm": 16.53125, "learning_rate": 8.980315520181922e-06, "loss": 0.7551, "step": 29198 }, { "epoch": 3.106937646307725, "grad_norm": 19.546875, "learning_rate": 8.980279988629904e-06, "loss": 0.8136, "step": 29199 }, { "epoch": 3.107044051925942, "grad_norm": 11.6015625, "learning_rate": 8.980244457077885e-06, "loss": 0.5509, "step": 29200 }, { "epoch": 3.1071504575441584, "grad_norm": 17.875, "learning_rate": 8.980208925525867e-06, "loss": 0.7189, "step": 29201 }, { "epoch": 3.107256863162375, "grad_norm": 12.6796875, "learning_rate": 8.98017339397385e-06, "loss": 0.5883, "step": 29202 }, { "epoch": 3.1073632687805914, "grad_norm": 13.6875, "learning_rate": 8.980137862421832e-06, "loss": 0.736, "step": 29203 }, { "epoch": 3.1074696743988084, "grad_norm": 12.5078125, "learning_rate": 8.980102330869813e-06, "loss": 0.5531, "step": 29204 }, { "epoch": 3.107576080017025, "grad_norm": 10.3046875, "learning_rate": 8.980066799317795e-06, "loss": 0.5501, "step": 29205 }, { "epoch": 3.1076824856352414, "grad_norm": 10.6640625, "learning_rate": 8.980031267765777e-06, "loss": 0.5295, "step": 29206 }, { "epoch": 3.1077888912534584, "grad_norm": 15.890625, "learning_rate": 8.979995736213758e-06, "loss": 0.8351, "step": 29207 }, { "epoch": 3.107895296871675, "grad_norm": 21.484375, "learning_rate": 8.97996020466174e-06, "loss": 0.6389, "step": 29208 }, { "epoch": 3.1080017024898914, "grad_norm": 12.65625, "learning_rate": 8.979924673109721e-06, "loss": 0.6194, "step": 29209 }, { "epoch": 3.108108108108108, "grad_norm": 11.5625, "learning_rate": 8.979889141557703e-06, "loss": 0.5609, "step": 29210 }, { "epoch": 3.108214513726325, "grad_norm": 15.203125, "learning_rate": 8.979853610005685e-06, "loss": 0.656, "step": 29211 }, { "epoch": 3.1083209193445414, "grad_norm": 9.84375, "learning_rate": 8.979818078453666e-06, "loss": 0.5878, "step": 29212 }, { "epoch": 3.108427324962758, "grad_norm": 10.2734375, "learning_rate": 8.97978254690165e-06, "loss": 0.6683, "step": 29213 }, { "epoch": 3.108533730580975, "grad_norm": 12.984375, "learning_rate": 8.979747015349631e-06, "loss": 0.6323, "step": 29214 }, { "epoch": 3.1086401361991913, "grad_norm": 14.28125, "learning_rate": 8.979711483797613e-06, "loss": 0.6823, "step": 29215 }, { "epoch": 3.108746541817408, "grad_norm": 9.3515625, "learning_rate": 8.979675952245594e-06, "loss": 0.5018, "step": 29216 }, { "epoch": 3.108852947435625, "grad_norm": 9.4375, "learning_rate": 8.979640420693576e-06, "loss": 0.4911, "step": 29217 }, { "epoch": 3.1089593530538413, "grad_norm": 14.171875, "learning_rate": 8.97960488914156e-06, "loss": 0.5796, "step": 29218 }, { "epoch": 3.109065758672058, "grad_norm": 9.796875, "learning_rate": 8.979569357589541e-06, "loss": 0.4233, "step": 29219 }, { "epoch": 3.1091721642902743, "grad_norm": 8.765625, "learning_rate": 8.979533826037523e-06, "loss": 0.4756, "step": 29220 }, { "epoch": 3.1092785699084913, "grad_norm": 12.765625, "learning_rate": 8.979498294485504e-06, "loss": 0.6398, "step": 29221 }, { "epoch": 3.109384975526708, "grad_norm": 14.9375, "learning_rate": 8.979462762933486e-06, "loss": 0.6376, "step": 29222 }, { "epoch": 3.1094913811449243, "grad_norm": 12.7421875, "learning_rate": 8.979427231381467e-06, "loss": 0.6024, "step": 29223 }, { "epoch": 3.1095977867631412, "grad_norm": 17.5625, "learning_rate": 8.979391699829449e-06, "loss": 0.7474, "step": 29224 }, { "epoch": 3.1097041923813578, "grad_norm": 10.8984375, "learning_rate": 8.97935616827743e-06, "loss": 0.5243, "step": 29225 }, { "epoch": 3.1098105979995743, "grad_norm": 11.2578125, "learning_rate": 8.979320636725412e-06, "loss": 0.583, "step": 29226 }, { "epoch": 3.109917003617791, "grad_norm": 14.03125, "learning_rate": 8.979285105173394e-06, "loss": 0.6929, "step": 29227 }, { "epoch": 3.1100234092360077, "grad_norm": 14.3203125, "learning_rate": 8.979249573621377e-06, "loss": 0.7758, "step": 29228 }, { "epoch": 3.1101298148542242, "grad_norm": 13.875, "learning_rate": 8.979214042069359e-06, "loss": 0.5448, "step": 29229 }, { "epoch": 3.1102362204724407, "grad_norm": 21.359375, "learning_rate": 8.97917851051734e-06, "loss": 0.6907, "step": 29230 }, { "epoch": 3.1103426260906577, "grad_norm": 12.953125, "learning_rate": 8.979142978965322e-06, "loss": 0.5574, "step": 29231 }, { "epoch": 3.110449031708874, "grad_norm": 11.0625, "learning_rate": 8.979107447413304e-06, "loss": 0.4295, "step": 29232 }, { "epoch": 3.1105554373270907, "grad_norm": 14.03125, "learning_rate": 8.979071915861287e-06, "loss": 0.554, "step": 29233 }, { "epoch": 3.1106618429453077, "grad_norm": 12.765625, "learning_rate": 8.979036384309268e-06, "loss": 0.6714, "step": 29234 }, { "epoch": 3.110768248563524, "grad_norm": 10.1484375, "learning_rate": 8.97900085275725e-06, "loss": 0.7009, "step": 29235 }, { "epoch": 3.1108746541817407, "grad_norm": 13.2265625, "learning_rate": 8.97896532120523e-06, "loss": 0.5902, "step": 29236 }, { "epoch": 3.1109810597999576, "grad_norm": 12.203125, "learning_rate": 8.978929789653212e-06, "loss": 0.6778, "step": 29237 }, { "epoch": 3.111087465418174, "grad_norm": 15.1796875, "learning_rate": 8.978894258101195e-06, "loss": 0.6703, "step": 29238 }, { "epoch": 3.1111938710363907, "grad_norm": 14.6171875, "learning_rate": 8.978858726549176e-06, "loss": 0.6841, "step": 29239 }, { "epoch": 3.111300276654607, "grad_norm": 10.5390625, "learning_rate": 8.978823194997158e-06, "loss": 0.4886, "step": 29240 }, { "epoch": 3.111406682272824, "grad_norm": 15.8046875, "learning_rate": 8.97878766344514e-06, "loss": 0.8635, "step": 29241 }, { "epoch": 3.1115130878910406, "grad_norm": 12.4453125, "learning_rate": 8.978752131893121e-06, "loss": 0.6372, "step": 29242 }, { "epoch": 3.111619493509257, "grad_norm": 10.8125, "learning_rate": 8.978716600341105e-06, "loss": 0.542, "step": 29243 }, { "epoch": 3.111725899127474, "grad_norm": 12.4765625, "learning_rate": 8.978681068789086e-06, "loss": 0.6318, "step": 29244 }, { "epoch": 3.1118323047456906, "grad_norm": 15.0234375, "learning_rate": 8.978645537237068e-06, "loss": 0.5611, "step": 29245 }, { "epoch": 3.111938710363907, "grad_norm": 9.5703125, "learning_rate": 8.97861000568505e-06, "loss": 0.4485, "step": 29246 }, { "epoch": 3.1120451159821236, "grad_norm": 9.53125, "learning_rate": 8.978574474133031e-06, "loss": 0.471, "step": 29247 }, { "epoch": 3.1121515216003406, "grad_norm": 10.3359375, "learning_rate": 8.978538942581013e-06, "loss": 0.5745, "step": 29248 }, { "epoch": 3.112257927218557, "grad_norm": 13.8671875, "learning_rate": 8.978503411028994e-06, "loss": 0.723, "step": 29249 }, { "epoch": 3.1123643328367736, "grad_norm": 14.46875, "learning_rate": 8.978467879476976e-06, "loss": 0.6523, "step": 29250 }, { "epoch": 3.1124707384549906, "grad_norm": 14.4453125, "learning_rate": 8.978432347924957e-06, "loss": 0.7621, "step": 29251 }, { "epoch": 3.112577144073207, "grad_norm": 11.9140625, "learning_rate": 8.978396816372939e-06, "loss": 0.5028, "step": 29252 }, { "epoch": 3.1126835496914236, "grad_norm": 13.6484375, "learning_rate": 8.97836128482092e-06, "loss": 0.6289, "step": 29253 }, { "epoch": 3.1127899553096405, "grad_norm": 13.5546875, "learning_rate": 8.978325753268904e-06, "loss": 0.7516, "step": 29254 }, { "epoch": 3.112896360927857, "grad_norm": 12.1484375, "learning_rate": 8.978290221716886e-06, "loss": 0.5293, "step": 29255 }, { "epoch": 3.1130027665460736, "grad_norm": 15.6171875, "learning_rate": 8.978254690164867e-06, "loss": 0.6813, "step": 29256 }, { "epoch": 3.11310917216429, "grad_norm": 14.890625, "learning_rate": 8.978219158612849e-06, "loss": 0.5705, "step": 29257 }, { "epoch": 3.113215577782507, "grad_norm": 11.984375, "learning_rate": 8.97818362706083e-06, "loss": 0.5685, "step": 29258 }, { "epoch": 3.1133219834007235, "grad_norm": 11.3671875, "learning_rate": 8.978148095508814e-06, "loss": 0.5275, "step": 29259 }, { "epoch": 3.11342838901894, "grad_norm": 15.1015625, "learning_rate": 8.978112563956795e-06, "loss": 0.6149, "step": 29260 }, { "epoch": 3.113534794637157, "grad_norm": 16.171875, "learning_rate": 8.978077032404777e-06, "loss": 0.838, "step": 29261 }, { "epoch": 3.1136412002553735, "grad_norm": 9.9296875, "learning_rate": 8.978041500852759e-06, "loss": 0.6202, "step": 29262 }, { "epoch": 3.11374760587359, "grad_norm": 15.5234375, "learning_rate": 8.97800596930074e-06, "loss": 0.6547, "step": 29263 }, { "epoch": 3.113854011491807, "grad_norm": 11.8671875, "learning_rate": 8.977970437748722e-06, "loss": 0.5653, "step": 29264 }, { "epoch": 3.1139604171100235, "grad_norm": 13.875, "learning_rate": 8.977934906196703e-06, "loss": 0.654, "step": 29265 }, { "epoch": 3.11406682272824, "grad_norm": 19.65625, "learning_rate": 8.977899374644685e-06, "loss": 0.7496, "step": 29266 }, { "epoch": 3.1141732283464565, "grad_norm": 13.6015625, "learning_rate": 8.977863843092667e-06, "loss": 0.6914, "step": 29267 }, { "epoch": 3.1142796339646734, "grad_norm": 11.484375, "learning_rate": 8.977828311540648e-06, "loss": 0.6538, "step": 29268 }, { "epoch": 3.11438603958289, "grad_norm": 14.6484375, "learning_rate": 8.977792779988631e-06, "loss": 0.7085, "step": 29269 }, { "epoch": 3.1144924452011065, "grad_norm": 15.828125, "learning_rate": 8.977757248436613e-06, "loss": 0.6445, "step": 29270 }, { "epoch": 3.1145988508193234, "grad_norm": 10.1953125, "learning_rate": 8.977721716884595e-06, "loss": 0.4582, "step": 29271 }, { "epoch": 3.11470525643754, "grad_norm": 10.421875, "learning_rate": 8.977686185332576e-06, "loss": 0.4735, "step": 29272 }, { "epoch": 3.1148116620557564, "grad_norm": 11.171875, "learning_rate": 8.977650653780558e-06, "loss": 0.496, "step": 29273 }, { "epoch": 3.1149180676739734, "grad_norm": 14.4921875, "learning_rate": 8.97761512222854e-06, "loss": 0.8909, "step": 29274 }, { "epoch": 3.11502447329219, "grad_norm": 9.8359375, "learning_rate": 8.977579590676521e-06, "loss": 0.5587, "step": 29275 }, { "epoch": 3.1151308789104064, "grad_norm": 16.875, "learning_rate": 8.977544059124503e-06, "loss": 0.5739, "step": 29276 }, { "epoch": 3.115237284528623, "grad_norm": 12.1328125, "learning_rate": 8.977508527572484e-06, "loss": 0.7197, "step": 29277 }, { "epoch": 3.11534369014684, "grad_norm": 15.0390625, "learning_rate": 8.977472996020466e-06, "loss": 0.6163, "step": 29278 }, { "epoch": 3.1154500957650564, "grad_norm": 11.8203125, "learning_rate": 8.97743746446845e-06, "loss": 0.5775, "step": 29279 }, { "epoch": 3.115556501383273, "grad_norm": 12.2890625, "learning_rate": 8.97740193291643e-06, "loss": 0.4891, "step": 29280 }, { "epoch": 3.11566290700149, "grad_norm": 12.96875, "learning_rate": 8.977366401364412e-06, "loss": 0.6028, "step": 29281 }, { "epoch": 3.1157693126197064, "grad_norm": 11.7109375, "learning_rate": 8.977330869812394e-06, "loss": 0.5741, "step": 29282 }, { "epoch": 3.115875718237923, "grad_norm": 12.109375, "learning_rate": 8.977295338260376e-06, "loss": 0.5995, "step": 29283 }, { "epoch": 3.11598212385614, "grad_norm": 11.9296875, "learning_rate": 8.977259806708359e-06, "loss": 0.5643, "step": 29284 }, { "epoch": 3.1160885294743563, "grad_norm": 9.7734375, "learning_rate": 8.97722427515634e-06, "loss": 0.4447, "step": 29285 }, { "epoch": 3.116194935092573, "grad_norm": 10.34375, "learning_rate": 8.977188743604322e-06, "loss": 0.4669, "step": 29286 }, { "epoch": 3.1163013407107893, "grad_norm": 12.203125, "learning_rate": 8.977153212052304e-06, "loss": 0.5709, "step": 29287 }, { "epoch": 3.1164077463290063, "grad_norm": 10.875, "learning_rate": 8.977117680500285e-06, "loss": 0.6394, "step": 29288 }, { "epoch": 3.116514151947223, "grad_norm": 9.828125, "learning_rate": 8.977082148948267e-06, "loss": 0.5165, "step": 29289 }, { "epoch": 3.1166205575654393, "grad_norm": 12.234375, "learning_rate": 8.977046617396249e-06, "loss": 0.6201, "step": 29290 }, { "epoch": 3.1167269631836563, "grad_norm": 12.6796875, "learning_rate": 8.97701108584423e-06, "loss": 0.6126, "step": 29291 }, { "epoch": 3.116833368801873, "grad_norm": 15.28125, "learning_rate": 8.976975554292212e-06, "loss": 0.5376, "step": 29292 }, { "epoch": 3.1169397744200893, "grad_norm": 12.921875, "learning_rate": 8.976940022740193e-06, "loss": 0.6099, "step": 29293 }, { "epoch": 3.1170461800383062, "grad_norm": 10.765625, "learning_rate": 8.976904491188175e-06, "loss": 0.5505, "step": 29294 }, { "epoch": 3.1171525856565228, "grad_norm": 12.3984375, "learning_rate": 8.976868959636158e-06, "loss": 0.6423, "step": 29295 }, { "epoch": 3.1172589912747393, "grad_norm": 10.8359375, "learning_rate": 8.97683342808414e-06, "loss": 0.5377, "step": 29296 }, { "epoch": 3.1173653968929558, "grad_norm": 14.4140625, "learning_rate": 8.976797896532122e-06, "loss": 0.5781, "step": 29297 }, { "epoch": 3.1174718025111727, "grad_norm": 14.59375, "learning_rate": 8.976762364980103e-06, "loss": 0.6761, "step": 29298 }, { "epoch": 3.1175782081293892, "grad_norm": 13.21875, "learning_rate": 8.976726833428085e-06, "loss": 0.5865, "step": 29299 }, { "epoch": 3.1176846137476057, "grad_norm": 12.34375, "learning_rate": 8.976691301876066e-06, "loss": 0.6711, "step": 29300 }, { "epoch": 3.1177910193658227, "grad_norm": 9.3046875, "learning_rate": 8.976655770324048e-06, "loss": 0.5032, "step": 29301 }, { "epoch": 3.117897424984039, "grad_norm": 13.3515625, "learning_rate": 8.97662023877203e-06, "loss": 0.6262, "step": 29302 }, { "epoch": 3.1180038306022557, "grad_norm": 18.09375, "learning_rate": 8.976584707220011e-06, "loss": 0.9503, "step": 29303 }, { "epoch": 3.1181102362204722, "grad_norm": 10.7734375, "learning_rate": 8.976549175667993e-06, "loss": 0.5268, "step": 29304 }, { "epoch": 3.118216641838689, "grad_norm": 23.46875, "learning_rate": 8.976513644115976e-06, "loss": 0.7304, "step": 29305 }, { "epoch": 3.1183230474569057, "grad_norm": 13.6875, "learning_rate": 8.976478112563958e-06, "loss": 0.6186, "step": 29306 }, { "epoch": 3.118429453075122, "grad_norm": 16.796875, "learning_rate": 8.97644258101194e-06, "loss": 0.6984, "step": 29307 }, { "epoch": 3.118535858693339, "grad_norm": 11.265625, "learning_rate": 8.976407049459921e-06, "loss": 0.5842, "step": 29308 }, { "epoch": 3.1186422643115557, "grad_norm": 12.7578125, "learning_rate": 8.976371517907902e-06, "loss": 0.623, "step": 29309 }, { "epoch": 3.118748669929772, "grad_norm": 8.328125, "learning_rate": 8.976335986355886e-06, "loss": 0.5238, "step": 29310 }, { "epoch": 3.118855075547989, "grad_norm": 11.015625, "learning_rate": 8.976300454803867e-06, "loss": 0.5693, "step": 29311 }, { "epoch": 3.1189614811662056, "grad_norm": 16.765625, "learning_rate": 8.976264923251849e-06, "loss": 0.867, "step": 29312 }, { "epoch": 3.119067886784422, "grad_norm": 14.6484375, "learning_rate": 8.97622939169983e-06, "loss": 0.5945, "step": 29313 }, { "epoch": 3.1191742924026387, "grad_norm": 13.296875, "learning_rate": 8.976193860147812e-06, "loss": 0.6885, "step": 29314 }, { "epoch": 3.1192806980208556, "grad_norm": 12.515625, "learning_rate": 8.976158328595794e-06, "loss": 0.6776, "step": 29315 }, { "epoch": 3.119387103639072, "grad_norm": 10.953125, "learning_rate": 8.976122797043775e-06, "loss": 0.501, "step": 29316 }, { "epoch": 3.1194935092572886, "grad_norm": 19.578125, "learning_rate": 8.976087265491757e-06, "loss": 0.7157, "step": 29317 }, { "epoch": 3.1195999148755056, "grad_norm": 14.40625, "learning_rate": 8.976051733939739e-06, "loss": 0.502, "step": 29318 }, { "epoch": 3.119706320493722, "grad_norm": 15.0625, "learning_rate": 8.97601620238772e-06, "loss": 0.563, "step": 29319 }, { "epoch": 3.1198127261119386, "grad_norm": 12.0, "learning_rate": 8.975980670835704e-06, "loss": 0.4898, "step": 29320 }, { "epoch": 3.1199191317301556, "grad_norm": 11.265625, "learning_rate": 8.975945139283685e-06, "loss": 0.6059, "step": 29321 }, { "epoch": 3.120025537348372, "grad_norm": 13.0, "learning_rate": 8.975909607731667e-06, "loss": 0.7144, "step": 29322 }, { "epoch": 3.1201319429665886, "grad_norm": 14.4609375, "learning_rate": 8.975874076179648e-06, "loss": 0.6893, "step": 29323 }, { "epoch": 3.120238348584805, "grad_norm": 14.75, "learning_rate": 8.97583854462763e-06, "loss": 0.6043, "step": 29324 }, { "epoch": 3.120344754203022, "grad_norm": 15.4375, "learning_rate": 8.975803013075612e-06, "loss": 0.6147, "step": 29325 }, { "epoch": 3.1204511598212386, "grad_norm": 21.03125, "learning_rate": 8.975767481523593e-06, "loss": 0.7946, "step": 29326 }, { "epoch": 3.120557565439455, "grad_norm": 11.046875, "learning_rate": 8.975731949971575e-06, "loss": 0.6501, "step": 29327 }, { "epoch": 3.120663971057672, "grad_norm": 15.9375, "learning_rate": 8.975696418419556e-06, "loss": 0.6919, "step": 29328 }, { "epoch": 3.1207703766758885, "grad_norm": 16.171875, "learning_rate": 8.975660886867538e-06, "loss": 0.7163, "step": 29329 }, { "epoch": 3.120876782294105, "grad_norm": 19.75, "learning_rate": 8.97562535531552e-06, "loss": 0.6441, "step": 29330 }, { "epoch": 3.120983187912322, "grad_norm": 12.7265625, "learning_rate": 8.975589823763503e-06, "loss": 0.6926, "step": 29331 }, { "epoch": 3.1210895935305385, "grad_norm": 12.96875, "learning_rate": 8.975554292211485e-06, "loss": 0.6089, "step": 29332 }, { "epoch": 3.121195999148755, "grad_norm": 12.8359375, "learning_rate": 8.975518760659466e-06, "loss": 0.5969, "step": 29333 }, { "epoch": 3.1213024047669715, "grad_norm": 12.7109375, "learning_rate": 8.975483229107448e-06, "loss": 0.6004, "step": 29334 }, { "epoch": 3.1214088103851885, "grad_norm": 14.1171875, "learning_rate": 8.97544769755543e-06, "loss": 0.7494, "step": 29335 }, { "epoch": 3.121515216003405, "grad_norm": 10.421875, "learning_rate": 8.975412166003413e-06, "loss": 0.6161, "step": 29336 }, { "epoch": 3.1216216216216215, "grad_norm": 14.2890625, "learning_rate": 8.975376634451394e-06, "loss": 0.7464, "step": 29337 }, { "epoch": 3.1217280272398384, "grad_norm": 9.7109375, "learning_rate": 8.975341102899376e-06, "loss": 0.4849, "step": 29338 }, { "epoch": 3.121834432858055, "grad_norm": 10.15625, "learning_rate": 8.975305571347357e-06, "loss": 0.7182, "step": 29339 }, { "epoch": 3.1219408384762715, "grad_norm": 13.6484375, "learning_rate": 8.975270039795339e-06, "loss": 0.6621, "step": 29340 }, { "epoch": 3.122047244094488, "grad_norm": 16.328125, "learning_rate": 8.97523450824332e-06, "loss": 0.6136, "step": 29341 }, { "epoch": 3.122153649712705, "grad_norm": 14.0390625, "learning_rate": 8.975198976691302e-06, "loss": 0.7301, "step": 29342 }, { "epoch": 3.1222600553309214, "grad_norm": 11.0703125, "learning_rate": 8.975163445139284e-06, "loss": 0.5087, "step": 29343 }, { "epoch": 3.122366460949138, "grad_norm": 11.203125, "learning_rate": 8.975127913587265e-06, "loss": 0.6064, "step": 29344 }, { "epoch": 3.122472866567355, "grad_norm": 13.75, "learning_rate": 8.975092382035247e-06, "loss": 0.7283, "step": 29345 }, { "epoch": 3.1225792721855714, "grad_norm": 11.4140625, "learning_rate": 8.97505685048323e-06, "loss": 0.5044, "step": 29346 }, { "epoch": 3.122685677803788, "grad_norm": 12.140625, "learning_rate": 8.975021318931212e-06, "loss": 0.661, "step": 29347 }, { "epoch": 3.122792083422005, "grad_norm": 11.265625, "learning_rate": 8.974985787379194e-06, "loss": 0.7469, "step": 29348 }, { "epoch": 3.1228984890402214, "grad_norm": 16.078125, "learning_rate": 8.974950255827175e-06, "loss": 0.7558, "step": 29349 }, { "epoch": 3.123004894658438, "grad_norm": 11.2265625, "learning_rate": 8.974914724275157e-06, "loss": 0.3977, "step": 29350 }, { "epoch": 3.1231113002766544, "grad_norm": 10.8046875, "learning_rate": 8.97487919272314e-06, "loss": 0.5945, "step": 29351 }, { "epoch": 3.1232177058948714, "grad_norm": 16.46875, "learning_rate": 8.974843661171122e-06, "loss": 0.6944, "step": 29352 }, { "epoch": 3.123324111513088, "grad_norm": 13.109375, "learning_rate": 8.974808129619103e-06, "loss": 0.6663, "step": 29353 }, { "epoch": 3.1234305171313044, "grad_norm": 10.1328125, "learning_rate": 8.974772598067085e-06, "loss": 0.5421, "step": 29354 }, { "epoch": 3.1235369227495213, "grad_norm": 11.6015625, "learning_rate": 8.974737066515065e-06, "loss": 0.576, "step": 29355 }, { "epoch": 3.123643328367738, "grad_norm": 13.9296875, "learning_rate": 8.974701534963048e-06, "loss": 0.5826, "step": 29356 }, { "epoch": 3.1237497339859543, "grad_norm": 12.7734375, "learning_rate": 8.97466600341103e-06, "loss": 0.6634, "step": 29357 }, { "epoch": 3.1238561396041713, "grad_norm": 13.0546875, "learning_rate": 8.974630471859011e-06, "loss": 0.6883, "step": 29358 }, { "epoch": 3.123962545222388, "grad_norm": 10.6328125, "learning_rate": 8.974594940306993e-06, "loss": 0.569, "step": 29359 }, { "epoch": 3.1240689508406043, "grad_norm": 15.109375, "learning_rate": 8.974559408754975e-06, "loss": 0.6285, "step": 29360 }, { "epoch": 3.124175356458821, "grad_norm": 16.171875, "learning_rate": 8.974523877202958e-06, "loss": 0.7044, "step": 29361 }, { "epoch": 3.124281762077038, "grad_norm": 10.3828125, "learning_rate": 8.97448834565094e-06, "loss": 0.603, "step": 29362 }, { "epoch": 3.1243881676952543, "grad_norm": 13.28125, "learning_rate": 8.974452814098921e-06, "loss": 0.5774, "step": 29363 }, { "epoch": 3.124494573313471, "grad_norm": 13.609375, "learning_rate": 8.974417282546903e-06, "loss": 0.5597, "step": 29364 }, { "epoch": 3.1246009789316878, "grad_norm": 15.2578125, "learning_rate": 8.974381750994884e-06, "loss": 0.6545, "step": 29365 }, { "epoch": 3.1247073845499043, "grad_norm": 14.109375, "learning_rate": 8.974346219442866e-06, "loss": 0.5778, "step": 29366 }, { "epoch": 3.1248137901681208, "grad_norm": 19.859375, "learning_rate": 8.974310687890848e-06, "loss": 0.7373, "step": 29367 }, { "epoch": 3.1249201957863377, "grad_norm": 11.8203125, "learning_rate": 8.974275156338829e-06, "loss": 0.516, "step": 29368 }, { "epoch": 3.1250266014045542, "grad_norm": 14.03125, "learning_rate": 8.97423962478681e-06, "loss": 0.6418, "step": 29369 }, { "epoch": 3.1251330070227707, "grad_norm": 12.8359375, "learning_rate": 8.974204093234792e-06, "loss": 0.7503, "step": 29370 }, { "epoch": 3.1252394126409873, "grad_norm": 11.3125, "learning_rate": 8.974168561682774e-06, "loss": 0.5211, "step": 29371 }, { "epoch": 3.125345818259204, "grad_norm": 12.3515625, "learning_rate": 8.974133030130757e-06, "loss": 0.5414, "step": 29372 }, { "epoch": 3.1254522238774207, "grad_norm": 11.65625, "learning_rate": 8.974097498578739e-06, "loss": 0.5156, "step": 29373 }, { "epoch": 3.1255586294956372, "grad_norm": 11.8828125, "learning_rate": 8.97406196702672e-06, "loss": 0.655, "step": 29374 }, { "epoch": 3.125665035113854, "grad_norm": 16.609375, "learning_rate": 8.974026435474702e-06, "loss": 0.6212, "step": 29375 }, { "epoch": 3.1257714407320707, "grad_norm": 14.234375, "learning_rate": 8.973990903922684e-06, "loss": 0.6881, "step": 29376 }, { "epoch": 3.125877846350287, "grad_norm": 12.3671875, "learning_rate": 8.973955372370667e-06, "loss": 0.6284, "step": 29377 }, { "epoch": 3.1259842519685037, "grad_norm": 19.6875, "learning_rate": 8.973919840818649e-06, "loss": 0.7243, "step": 29378 }, { "epoch": 3.1260906575867207, "grad_norm": 13.0703125, "learning_rate": 8.97388430926663e-06, "loss": 0.6353, "step": 29379 }, { "epoch": 3.126197063204937, "grad_norm": 12.53125, "learning_rate": 8.973848777714612e-06, "loss": 0.5622, "step": 29380 }, { "epoch": 3.1263034688231537, "grad_norm": 13.6015625, "learning_rate": 8.973813246162593e-06, "loss": 0.645, "step": 29381 }, { "epoch": 3.1264098744413706, "grad_norm": 14.703125, "learning_rate": 8.973777714610575e-06, "loss": 0.8349, "step": 29382 }, { "epoch": 3.126516280059587, "grad_norm": 11.4375, "learning_rate": 8.973742183058557e-06, "loss": 0.5863, "step": 29383 }, { "epoch": 3.1266226856778037, "grad_norm": 12.5234375, "learning_rate": 8.973706651506538e-06, "loss": 0.7179, "step": 29384 }, { "epoch": 3.1267290912960206, "grad_norm": 15.625, "learning_rate": 8.97367111995452e-06, "loss": 0.6177, "step": 29385 }, { "epoch": 3.126835496914237, "grad_norm": 11.1875, "learning_rate": 8.973635588402501e-06, "loss": 0.7967, "step": 29386 }, { "epoch": 3.1269419025324536, "grad_norm": 14.9921875, "learning_rate": 8.973600056850485e-06, "loss": 0.6526, "step": 29387 }, { "epoch": 3.1270483081506706, "grad_norm": 11.53125, "learning_rate": 8.973564525298466e-06, "loss": 0.5181, "step": 29388 }, { "epoch": 3.127154713768887, "grad_norm": 13.5078125, "learning_rate": 8.973528993746448e-06, "loss": 0.5233, "step": 29389 }, { "epoch": 3.1272611193871036, "grad_norm": 12.234375, "learning_rate": 8.97349346219443e-06, "loss": 0.6911, "step": 29390 }, { "epoch": 3.12736752500532, "grad_norm": 13.6328125, "learning_rate": 8.973457930642411e-06, "loss": 0.6208, "step": 29391 }, { "epoch": 3.127473930623537, "grad_norm": 11.9140625, "learning_rate": 8.973422399090393e-06, "loss": 0.5185, "step": 29392 }, { "epoch": 3.1275803362417536, "grad_norm": 13.015625, "learning_rate": 8.973386867538374e-06, "loss": 0.5849, "step": 29393 }, { "epoch": 3.12768674185997, "grad_norm": 11.3828125, "learning_rate": 8.973351335986356e-06, "loss": 0.5651, "step": 29394 }, { "epoch": 3.127793147478187, "grad_norm": 13.4453125, "learning_rate": 8.973315804434338e-06, "loss": 0.8509, "step": 29395 }, { "epoch": 3.1278995530964036, "grad_norm": 11.46875, "learning_rate": 8.97328027288232e-06, "loss": 0.4986, "step": 29396 }, { "epoch": 3.12800595871462, "grad_norm": 11.3203125, "learning_rate": 8.973244741330303e-06, "loss": 0.5883, "step": 29397 }, { "epoch": 3.1281123643328366, "grad_norm": 12.7265625, "learning_rate": 8.973209209778284e-06, "loss": 0.6802, "step": 29398 }, { "epoch": 3.1282187699510535, "grad_norm": 11.40625, "learning_rate": 8.973173678226266e-06, "loss": 0.6382, "step": 29399 }, { "epoch": 3.12832517556927, "grad_norm": 15.7734375, "learning_rate": 8.973138146674247e-06, "loss": 0.7676, "step": 29400 }, { "epoch": 3.1284315811874865, "grad_norm": 9.875, "learning_rate": 8.973102615122229e-06, "loss": 0.4615, "step": 29401 }, { "epoch": 3.1285379868057035, "grad_norm": 12.390625, "learning_rate": 8.973067083570212e-06, "loss": 0.3764, "step": 29402 }, { "epoch": 3.12864439242392, "grad_norm": 12.375, "learning_rate": 8.973031552018194e-06, "loss": 0.5919, "step": 29403 }, { "epoch": 3.1287507980421365, "grad_norm": 15.328125, "learning_rate": 8.972996020466175e-06, "loss": 0.739, "step": 29404 }, { "epoch": 3.1288572036603535, "grad_norm": 10.578125, "learning_rate": 8.972960488914157e-06, "loss": 0.5964, "step": 29405 }, { "epoch": 3.12896360927857, "grad_norm": 12.59375, "learning_rate": 8.972924957362139e-06, "loss": 0.5782, "step": 29406 }, { "epoch": 3.1290700148967865, "grad_norm": 8.9765625, "learning_rate": 8.97288942581012e-06, "loss": 0.5255, "step": 29407 }, { "epoch": 3.129176420515003, "grad_norm": 13.078125, "learning_rate": 8.972853894258102e-06, "loss": 0.8049, "step": 29408 }, { "epoch": 3.12928282613322, "grad_norm": 8.59375, "learning_rate": 8.972818362706083e-06, "loss": 0.4409, "step": 29409 }, { "epoch": 3.1293892317514365, "grad_norm": 12.9296875, "learning_rate": 8.972782831154065e-06, "loss": 0.561, "step": 29410 }, { "epoch": 3.129495637369653, "grad_norm": 12.125, "learning_rate": 8.972747299602047e-06, "loss": 0.5138, "step": 29411 }, { "epoch": 3.12960204298787, "grad_norm": 12.8359375, "learning_rate": 8.972711768050028e-06, "loss": 0.6253, "step": 29412 }, { "epoch": 3.1297084486060864, "grad_norm": 15.625, "learning_rate": 8.972676236498012e-06, "loss": 0.7177, "step": 29413 }, { "epoch": 3.129814854224303, "grad_norm": 12.8046875, "learning_rate": 8.972640704945993e-06, "loss": 0.5127, "step": 29414 }, { "epoch": 3.1299212598425195, "grad_norm": 9.7578125, "learning_rate": 8.972605173393975e-06, "loss": 0.4565, "step": 29415 }, { "epoch": 3.1300276654607364, "grad_norm": 11.46875, "learning_rate": 8.972569641841956e-06, "loss": 0.6321, "step": 29416 }, { "epoch": 3.130134071078953, "grad_norm": 13.9453125, "learning_rate": 8.972534110289938e-06, "loss": 0.4973, "step": 29417 }, { "epoch": 3.1302404766971694, "grad_norm": 12.625, "learning_rate": 8.97249857873792e-06, "loss": 0.5432, "step": 29418 }, { "epoch": 3.1303468823153864, "grad_norm": 12.6328125, "learning_rate": 8.972463047185901e-06, "loss": 0.8265, "step": 29419 }, { "epoch": 3.130453287933603, "grad_norm": 11.7265625, "learning_rate": 8.972427515633883e-06, "loss": 0.5175, "step": 29420 }, { "epoch": 3.1305596935518194, "grad_norm": 13.359375, "learning_rate": 8.972391984081864e-06, "loss": 0.6979, "step": 29421 }, { "epoch": 3.1306660991700364, "grad_norm": 14.3125, "learning_rate": 8.972356452529846e-06, "loss": 0.7239, "step": 29422 }, { "epoch": 3.130772504788253, "grad_norm": 11.3515625, "learning_rate": 8.97232092097783e-06, "loss": 0.5294, "step": 29423 }, { "epoch": 3.1308789104064694, "grad_norm": 10.9140625, "learning_rate": 8.972285389425811e-06, "loss": 0.5238, "step": 29424 }, { "epoch": 3.1309853160246863, "grad_norm": 12.3984375, "learning_rate": 8.972249857873793e-06, "loss": 0.7712, "step": 29425 }, { "epoch": 3.131091721642903, "grad_norm": 16.78125, "learning_rate": 8.972214326321774e-06, "loss": 0.7512, "step": 29426 }, { "epoch": 3.1311981272611193, "grad_norm": 11.78125, "learning_rate": 8.972178794769756e-06, "loss": 0.6874, "step": 29427 }, { "epoch": 3.131304532879336, "grad_norm": 13.0703125, "learning_rate": 8.972143263217739e-06, "loss": 0.6605, "step": 29428 }, { "epoch": 3.131410938497553, "grad_norm": 12.875, "learning_rate": 8.97210773166572e-06, "loss": 0.595, "step": 29429 }, { "epoch": 3.1315173441157693, "grad_norm": 17.421875, "learning_rate": 8.972072200113702e-06, "loss": 0.9491, "step": 29430 }, { "epoch": 3.131623749733986, "grad_norm": 14.03125, "learning_rate": 8.972036668561684e-06, "loss": 0.6651, "step": 29431 }, { "epoch": 3.131730155352203, "grad_norm": 12.7109375, "learning_rate": 8.972001137009666e-06, "loss": 0.5837, "step": 29432 }, { "epoch": 3.1318365609704193, "grad_norm": 10.9453125, "learning_rate": 8.971965605457647e-06, "loss": 0.5485, "step": 29433 }, { "epoch": 3.131942966588636, "grad_norm": 11.5546875, "learning_rate": 8.971930073905629e-06, "loss": 0.5527, "step": 29434 }, { "epoch": 3.1320493722068523, "grad_norm": 11.78125, "learning_rate": 8.97189454235361e-06, "loss": 0.4788, "step": 29435 }, { "epoch": 3.1321557778250693, "grad_norm": 11.8125, "learning_rate": 8.971859010801592e-06, "loss": 0.583, "step": 29436 }, { "epoch": 3.1322621834432858, "grad_norm": 19.5625, "learning_rate": 8.971823479249574e-06, "loss": 0.6937, "step": 29437 }, { "epoch": 3.1323685890615023, "grad_norm": 11.0234375, "learning_rate": 8.971787947697557e-06, "loss": 0.4923, "step": 29438 }, { "epoch": 3.1324749946797192, "grad_norm": 12.71875, "learning_rate": 8.971752416145538e-06, "loss": 0.565, "step": 29439 }, { "epoch": 3.1325814002979357, "grad_norm": 12.4765625, "learning_rate": 8.97171688459352e-06, "loss": 0.5165, "step": 29440 }, { "epoch": 3.1326878059161523, "grad_norm": 12.171875, "learning_rate": 8.971681353041502e-06, "loss": 0.5645, "step": 29441 }, { "epoch": 3.132794211534369, "grad_norm": 15.5, "learning_rate": 8.971645821489483e-06, "loss": 0.7535, "step": 29442 }, { "epoch": 3.1329006171525857, "grad_norm": 12.390625, "learning_rate": 8.971610289937467e-06, "loss": 0.8449, "step": 29443 }, { "epoch": 3.1330070227708022, "grad_norm": 11.921875, "learning_rate": 8.971574758385446e-06, "loss": 0.492, "step": 29444 }, { "epoch": 3.1331134283890187, "grad_norm": 10.2109375, "learning_rate": 8.971539226833428e-06, "loss": 0.5579, "step": 29445 }, { "epoch": 3.1332198340072357, "grad_norm": 18.09375, "learning_rate": 8.97150369528141e-06, "loss": 0.632, "step": 29446 }, { "epoch": 3.133326239625452, "grad_norm": 12.2734375, "learning_rate": 8.971468163729391e-06, "loss": 0.7651, "step": 29447 }, { "epoch": 3.1334326452436687, "grad_norm": 11.1953125, "learning_rate": 8.971432632177373e-06, "loss": 0.5233, "step": 29448 }, { "epoch": 3.1335390508618857, "grad_norm": 9.21875, "learning_rate": 8.971397100625356e-06, "loss": 0.4497, "step": 29449 }, { "epoch": 3.133645456480102, "grad_norm": 11.65625, "learning_rate": 8.971361569073338e-06, "loss": 0.6244, "step": 29450 }, { "epoch": 3.1337518620983187, "grad_norm": 12.2109375, "learning_rate": 8.97132603752132e-06, "loss": 0.5854, "step": 29451 }, { "epoch": 3.1338582677165356, "grad_norm": 13.671875, "learning_rate": 8.971290505969301e-06, "loss": 0.6589, "step": 29452 }, { "epoch": 3.133964673334752, "grad_norm": 15.875, "learning_rate": 8.971254974417283e-06, "loss": 0.6453, "step": 29453 }, { "epoch": 3.1340710789529687, "grad_norm": 11.6875, "learning_rate": 8.971219442865266e-06, "loss": 0.4769, "step": 29454 }, { "epoch": 3.134177484571185, "grad_norm": 13.8046875, "learning_rate": 8.971183911313248e-06, "loss": 0.6061, "step": 29455 }, { "epoch": 3.134283890189402, "grad_norm": 8.8984375, "learning_rate": 8.97114837976123e-06, "loss": 0.4746, "step": 29456 }, { "epoch": 3.1343902958076186, "grad_norm": 10.84375, "learning_rate": 8.97111284820921e-06, "loss": 0.4328, "step": 29457 }, { "epoch": 3.134496701425835, "grad_norm": 14.5546875, "learning_rate": 8.971077316657192e-06, "loss": 0.6897, "step": 29458 }, { "epoch": 3.134603107044052, "grad_norm": 10.6953125, "learning_rate": 8.971041785105174e-06, "loss": 0.4539, "step": 29459 }, { "epoch": 3.1347095126622686, "grad_norm": 13.7578125, "learning_rate": 8.971006253553156e-06, "loss": 0.6711, "step": 29460 }, { "epoch": 3.134815918280485, "grad_norm": 14.71875, "learning_rate": 8.970970722001137e-06, "loss": 0.6735, "step": 29461 }, { "epoch": 3.134922323898702, "grad_norm": 10.5390625, "learning_rate": 8.970935190449119e-06, "loss": 0.436, "step": 29462 }, { "epoch": 3.1350287295169186, "grad_norm": 18.59375, "learning_rate": 8.9708996588971e-06, "loss": 0.709, "step": 29463 }, { "epoch": 3.135135135135135, "grad_norm": 12.078125, "learning_rate": 8.970864127345084e-06, "loss": 0.6311, "step": 29464 }, { "epoch": 3.1352415407533516, "grad_norm": 10.75, "learning_rate": 8.970828595793065e-06, "loss": 0.5925, "step": 29465 }, { "epoch": 3.1353479463715686, "grad_norm": 13.6171875, "learning_rate": 8.970793064241047e-06, "loss": 0.5057, "step": 29466 }, { "epoch": 3.135454351989785, "grad_norm": 13.3046875, "learning_rate": 8.970757532689029e-06, "loss": 0.7037, "step": 29467 }, { "epoch": 3.1355607576080016, "grad_norm": 16.53125, "learning_rate": 8.97072200113701e-06, "loss": 0.8567, "step": 29468 }, { "epoch": 3.1356671632262185, "grad_norm": 10.59375, "learning_rate": 8.970686469584993e-06, "loss": 0.6249, "step": 29469 }, { "epoch": 3.135773568844435, "grad_norm": 17.359375, "learning_rate": 8.970650938032975e-06, "loss": 0.7791, "step": 29470 }, { "epoch": 3.1358799744626515, "grad_norm": 14.796875, "learning_rate": 8.970615406480957e-06, "loss": 0.6763, "step": 29471 }, { "epoch": 3.135986380080868, "grad_norm": 10.765625, "learning_rate": 8.970579874928938e-06, "loss": 0.4715, "step": 29472 }, { "epoch": 3.136092785699085, "grad_norm": 15.453125, "learning_rate": 8.97054434337692e-06, "loss": 0.7306, "step": 29473 }, { "epoch": 3.1361991913173015, "grad_norm": 11.578125, "learning_rate": 8.970508811824901e-06, "loss": 0.5199, "step": 29474 }, { "epoch": 3.136305596935518, "grad_norm": 13.15625, "learning_rate": 8.970473280272883e-06, "loss": 0.6122, "step": 29475 }, { "epoch": 3.136412002553735, "grad_norm": 12.765625, "learning_rate": 8.970437748720865e-06, "loss": 0.6549, "step": 29476 }, { "epoch": 3.1365184081719515, "grad_norm": 11.546875, "learning_rate": 8.970402217168846e-06, "loss": 0.6672, "step": 29477 }, { "epoch": 3.136624813790168, "grad_norm": 13.2421875, "learning_rate": 8.970366685616828e-06, "loss": 0.6209, "step": 29478 }, { "epoch": 3.136731219408385, "grad_norm": 11.0078125, "learning_rate": 8.970331154064811e-06, "loss": 0.4735, "step": 29479 }, { "epoch": 3.1368376250266015, "grad_norm": 11.9140625, "learning_rate": 8.970295622512793e-06, "loss": 0.6566, "step": 29480 }, { "epoch": 3.136944030644818, "grad_norm": 14.9453125, "learning_rate": 8.970260090960774e-06, "loss": 0.7448, "step": 29481 }, { "epoch": 3.1370504362630345, "grad_norm": 15.0859375, "learning_rate": 8.970224559408756e-06, "loss": 0.6163, "step": 29482 }, { "epoch": 3.1371568418812514, "grad_norm": 12.5234375, "learning_rate": 8.970189027856738e-06, "loss": 0.6314, "step": 29483 }, { "epoch": 3.137263247499468, "grad_norm": 13.40625, "learning_rate": 8.97015349630472e-06, "loss": 0.6039, "step": 29484 }, { "epoch": 3.1373696531176845, "grad_norm": 14.25, "learning_rate": 8.9701179647527e-06, "loss": 0.6088, "step": 29485 }, { "epoch": 3.1374760587359014, "grad_norm": 12.9296875, "learning_rate": 8.970082433200682e-06, "loss": 0.6484, "step": 29486 }, { "epoch": 3.137582464354118, "grad_norm": 12.703125, "learning_rate": 8.970046901648664e-06, "loss": 0.6429, "step": 29487 }, { "epoch": 3.1376888699723344, "grad_norm": 13.4765625, "learning_rate": 8.970011370096646e-06, "loss": 0.7552, "step": 29488 }, { "epoch": 3.1377952755905514, "grad_norm": 16.890625, "learning_rate": 8.969975838544627e-06, "loss": 0.6511, "step": 29489 }, { "epoch": 3.137901681208768, "grad_norm": 18.21875, "learning_rate": 8.96994030699261e-06, "loss": 0.5674, "step": 29490 }, { "epoch": 3.1380080868269844, "grad_norm": 13.6953125, "learning_rate": 8.969904775440592e-06, "loss": 0.6205, "step": 29491 }, { "epoch": 3.138114492445201, "grad_norm": 15.3203125, "learning_rate": 8.969869243888574e-06, "loss": 0.6202, "step": 29492 }, { "epoch": 3.138220898063418, "grad_norm": 12.4765625, "learning_rate": 8.969833712336555e-06, "loss": 0.5006, "step": 29493 }, { "epoch": 3.1383273036816344, "grad_norm": 11.2578125, "learning_rate": 8.969798180784537e-06, "loss": 0.6301, "step": 29494 }, { "epoch": 3.138433709299851, "grad_norm": 11.1640625, "learning_rate": 8.96976264923252e-06, "loss": 0.7388, "step": 29495 }, { "epoch": 3.138540114918068, "grad_norm": 13.53125, "learning_rate": 8.969727117680502e-06, "loss": 0.6774, "step": 29496 }, { "epoch": 3.1386465205362843, "grad_norm": 10.5078125, "learning_rate": 8.969691586128484e-06, "loss": 0.4899, "step": 29497 }, { "epoch": 3.138752926154501, "grad_norm": 12.734375, "learning_rate": 8.969656054576465e-06, "loss": 0.6996, "step": 29498 }, { "epoch": 3.138859331772718, "grad_norm": 15.171875, "learning_rate": 8.969620523024447e-06, "loss": 0.6539, "step": 29499 }, { "epoch": 3.1389657373909343, "grad_norm": 15.4296875, "learning_rate": 8.969584991472428e-06, "loss": 0.6536, "step": 29500 }, { "epoch": 3.139072143009151, "grad_norm": 15.6640625, "learning_rate": 8.96954945992041e-06, "loss": 0.6955, "step": 29501 }, { "epoch": 3.1391785486273673, "grad_norm": 11.375, "learning_rate": 8.969513928368392e-06, "loss": 0.6617, "step": 29502 }, { "epoch": 3.1392849542455843, "grad_norm": 10.5234375, "learning_rate": 8.969478396816373e-06, "loss": 0.7247, "step": 29503 }, { "epoch": 3.139391359863801, "grad_norm": 9.7890625, "learning_rate": 8.969442865264355e-06, "loss": 0.4375, "step": 29504 }, { "epoch": 3.1394977654820173, "grad_norm": 11.078125, "learning_rate": 8.969407333712338e-06, "loss": 0.6374, "step": 29505 }, { "epoch": 3.1396041711002343, "grad_norm": 11.9921875, "learning_rate": 8.96937180216032e-06, "loss": 0.599, "step": 29506 }, { "epoch": 3.1397105767184508, "grad_norm": 12.9453125, "learning_rate": 8.969336270608301e-06, "loss": 0.6826, "step": 29507 }, { "epoch": 3.1398169823366673, "grad_norm": 11.2265625, "learning_rate": 8.969300739056283e-06, "loss": 0.5925, "step": 29508 }, { "epoch": 3.139923387954884, "grad_norm": 14.2890625, "learning_rate": 8.969265207504264e-06, "loss": 0.6387, "step": 29509 }, { "epoch": 3.1400297935731007, "grad_norm": 9.546875, "learning_rate": 8.969229675952246e-06, "loss": 0.5796, "step": 29510 }, { "epoch": 3.1401361991913173, "grad_norm": 9.828125, "learning_rate": 8.969194144400228e-06, "loss": 0.515, "step": 29511 }, { "epoch": 3.1402426048095338, "grad_norm": 7.98046875, "learning_rate": 8.96915861284821e-06, "loss": 0.6096, "step": 29512 }, { "epoch": 3.1403490104277507, "grad_norm": 13.015625, "learning_rate": 8.969123081296191e-06, "loss": 0.5757, "step": 29513 }, { "epoch": 3.1404554160459672, "grad_norm": 15.4375, "learning_rate": 8.969087549744173e-06, "loss": 0.645, "step": 29514 }, { "epoch": 3.1405618216641837, "grad_norm": 12.15625, "learning_rate": 8.969052018192156e-06, "loss": 0.5315, "step": 29515 }, { "epoch": 3.1406682272824007, "grad_norm": 10.5703125, "learning_rate": 8.969016486640137e-06, "loss": 0.6667, "step": 29516 }, { "epoch": 3.140774632900617, "grad_norm": 19.640625, "learning_rate": 8.968980955088119e-06, "loss": 0.8635, "step": 29517 }, { "epoch": 3.1408810385188337, "grad_norm": 9.2734375, "learning_rate": 8.9689454235361e-06, "loss": 0.4857, "step": 29518 }, { "epoch": 3.1409874441370507, "grad_norm": 13.4140625, "learning_rate": 8.968909891984082e-06, "loss": 0.6148, "step": 29519 }, { "epoch": 3.141093849755267, "grad_norm": 12.765625, "learning_rate": 8.968874360432066e-06, "loss": 0.7367, "step": 29520 }, { "epoch": 3.1412002553734837, "grad_norm": 9.0234375, "learning_rate": 8.968838828880047e-06, "loss": 0.534, "step": 29521 }, { "epoch": 3.1413066609917, "grad_norm": 11.625, "learning_rate": 8.968803297328029e-06, "loss": 0.6151, "step": 29522 }, { "epoch": 3.141413066609917, "grad_norm": 14.703125, "learning_rate": 8.96876776577601e-06, "loss": 0.8074, "step": 29523 }, { "epoch": 3.1415194722281337, "grad_norm": 13.46875, "learning_rate": 8.968732234223992e-06, "loss": 0.591, "step": 29524 }, { "epoch": 3.14162587784635, "grad_norm": 14.265625, "learning_rate": 8.968696702671974e-06, "loss": 0.7126, "step": 29525 }, { "epoch": 3.141732283464567, "grad_norm": 10.0390625, "learning_rate": 8.968661171119955e-06, "loss": 0.521, "step": 29526 }, { "epoch": 3.1418386890827836, "grad_norm": 14.1328125, "learning_rate": 8.968625639567937e-06, "loss": 0.6423, "step": 29527 }, { "epoch": 3.141945094701, "grad_norm": 10.90625, "learning_rate": 8.968590108015918e-06, "loss": 0.5845, "step": 29528 }, { "epoch": 3.1420515003192167, "grad_norm": 12.2421875, "learning_rate": 8.9685545764639e-06, "loss": 0.6158, "step": 29529 }, { "epoch": 3.1421579059374336, "grad_norm": 10.9921875, "learning_rate": 8.968519044911882e-06, "loss": 0.5109, "step": 29530 }, { "epoch": 3.14226431155565, "grad_norm": 15.9921875, "learning_rate": 8.968483513359865e-06, "loss": 0.7801, "step": 29531 }, { "epoch": 3.1423707171738666, "grad_norm": 14.6796875, "learning_rate": 8.968447981807847e-06, "loss": 0.6973, "step": 29532 }, { "epoch": 3.1424771227920836, "grad_norm": 10.8046875, "learning_rate": 8.968412450255828e-06, "loss": 0.4671, "step": 29533 }, { "epoch": 3.1425835284103, "grad_norm": 13.703125, "learning_rate": 8.96837691870381e-06, "loss": 0.6369, "step": 29534 }, { "epoch": 3.1426899340285166, "grad_norm": 13.6640625, "learning_rate": 8.968341387151791e-06, "loss": 0.6327, "step": 29535 }, { "epoch": 3.1427963396467335, "grad_norm": 17.21875, "learning_rate": 8.968305855599773e-06, "loss": 0.7567, "step": 29536 }, { "epoch": 3.14290274526495, "grad_norm": 11.40625, "learning_rate": 8.968270324047755e-06, "loss": 0.5842, "step": 29537 }, { "epoch": 3.1430091508831666, "grad_norm": 11.65625, "learning_rate": 8.968234792495736e-06, "loss": 0.555, "step": 29538 }, { "epoch": 3.143115556501383, "grad_norm": 10.0703125, "learning_rate": 8.968199260943718e-06, "loss": 0.5474, "step": 29539 }, { "epoch": 3.1432219621196, "grad_norm": 9.0625, "learning_rate": 8.9681637293917e-06, "loss": 0.4451, "step": 29540 }, { "epoch": 3.1433283677378165, "grad_norm": 14.7109375, "learning_rate": 8.968128197839683e-06, "loss": 0.783, "step": 29541 }, { "epoch": 3.143434773356033, "grad_norm": 11.0, "learning_rate": 8.968092666287664e-06, "loss": 0.6362, "step": 29542 }, { "epoch": 3.14354117897425, "grad_norm": 13.9375, "learning_rate": 8.968057134735646e-06, "loss": 0.669, "step": 29543 }, { "epoch": 3.1436475845924665, "grad_norm": 12.7578125, "learning_rate": 8.968021603183627e-06, "loss": 0.6353, "step": 29544 }, { "epoch": 3.143753990210683, "grad_norm": 11.5, "learning_rate": 8.967986071631609e-06, "loss": 0.5926, "step": 29545 }, { "epoch": 3.1438603958288995, "grad_norm": 28.6875, "learning_rate": 8.967950540079592e-06, "loss": 0.7579, "step": 29546 }, { "epoch": 3.1439668014471165, "grad_norm": 10.40625, "learning_rate": 8.967915008527574e-06, "loss": 0.5264, "step": 29547 }, { "epoch": 3.144073207065333, "grad_norm": 10.2109375, "learning_rate": 8.967879476975556e-06, "loss": 0.5926, "step": 29548 }, { "epoch": 3.1441796126835495, "grad_norm": 12.25, "learning_rate": 8.967843945423537e-06, "loss": 0.7871, "step": 29549 }, { "epoch": 3.1442860183017665, "grad_norm": 12.1171875, "learning_rate": 8.967808413871519e-06, "loss": 0.5349, "step": 29550 }, { "epoch": 3.144392423919983, "grad_norm": 12.6171875, "learning_rate": 8.9677728823195e-06, "loss": 0.4884, "step": 29551 }, { "epoch": 3.1444988295381995, "grad_norm": 15.78125, "learning_rate": 8.967737350767482e-06, "loss": 0.8406, "step": 29552 }, { "epoch": 3.1446052351564164, "grad_norm": 14.578125, "learning_rate": 8.967701819215464e-06, "loss": 0.6968, "step": 29553 }, { "epoch": 3.144711640774633, "grad_norm": 13.25, "learning_rate": 8.967666287663445e-06, "loss": 0.6336, "step": 29554 }, { "epoch": 3.1448180463928495, "grad_norm": 12.71875, "learning_rate": 8.967630756111427e-06, "loss": 0.5633, "step": 29555 }, { "epoch": 3.1449244520110664, "grad_norm": 13.65625, "learning_rate": 8.96759522455941e-06, "loss": 0.7305, "step": 29556 }, { "epoch": 3.145030857629283, "grad_norm": 16.578125, "learning_rate": 8.967559693007392e-06, "loss": 0.709, "step": 29557 }, { "epoch": 3.1451372632474994, "grad_norm": 15.1796875, "learning_rate": 8.967524161455373e-06, "loss": 0.6767, "step": 29558 }, { "epoch": 3.145243668865716, "grad_norm": 6.859375, "learning_rate": 8.967488629903355e-06, "loss": 0.3025, "step": 29559 }, { "epoch": 3.145350074483933, "grad_norm": 9.7265625, "learning_rate": 8.967453098351337e-06, "loss": 0.6863, "step": 29560 }, { "epoch": 3.1454564801021494, "grad_norm": 15.7265625, "learning_rate": 8.96741756679932e-06, "loss": 0.5873, "step": 29561 }, { "epoch": 3.145562885720366, "grad_norm": 20.296875, "learning_rate": 8.967382035247301e-06, "loss": 0.7631, "step": 29562 }, { "epoch": 3.145669291338583, "grad_norm": 11.2890625, "learning_rate": 8.967346503695283e-06, "loss": 0.6042, "step": 29563 }, { "epoch": 3.1457756969567994, "grad_norm": 15.1015625, "learning_rate": 8.967310972143263e-06, "loss": 0.6638, "step": 29564 }, { "epoch": 3.145882102575016, "grad_norm": 12.109375, "learning_rate": 8.967275440591245e-06, "loss": 0.5761, "step": 29565 }, { "epoch": 3.1459885081932324, "grad_norm": 12.5234375, "learning_rate": 8.967239909039226e-06, "loss": 0.5557, "step": 29566 }, { "epoch": 3.1460949138114493, "grad_norm": 18.578125, "learning_rate": 8.96720437748721e-06, "loss": 0.7676, "step": 29567 }, { "epoch": 3.146201319429666, "grad_norm": 13.5859375, "learning_rate": 8.967168845935191e-06, "loss": 0.6071, "step": 29568 }, { "epoch": 3.1463077250478824, "grad_norm": 13.3984375, "learning_rate": 8.967133314383173e-06, "loss": 0.5866, "step": 29569 }, { "epoch": 3.1464141306660993, "grad_norm": 13.4765625, "learning_rate": 8.967097782831154e-06, "loss": 0.55, "step": 29570 }, { "epoch": 3.146520536284316, "grad_norm": 22.640625, "learning_rate": 8.967062251279136e-06, "loss": 0.7474, "step": 29571 }, { "epoch": 3.1466269419025323, "grad_norm": 12.5234375, "learning_rate": 8.96702671972712e-06, "loss": 0.623, "step": 29572 }, { "epoch": 3.1467333475207493, "grad_norm": 15.5703125, "learning_rate": 8.966991188175101e-06, "loss": 0.5381, "step": 29573 }, { "epoch": 3.146839753138966, "grad_norm": 16.4375, "learning_rate": 8.966955656623082e-06, "loss": 0.9281, "step": 29574 }, { "epoch": 3.1469461587571823, "grad_norm": 13.1484375, "learning_rate": 8.966920125071064e-06, "loss": 0.4775, "step": 29575 }, { "epoch": 3.147052564375399, "grad_norm": 16.859375, "learning_rate": 8.966884593519046e-06, "loss": 0.6649, "step": 29576 }, { "epoch": 3.1471589699936158, "grad_norm": 11.7265625, "learning_rate": 8.966849061967027e-06, "loss": 0.5485, "step": 29577 }, { "epoch": 3.1472653756118323, "grad_norm": 12.03125, "learning_rate": 8.966813530415009e-06, "loss": 0.5122, "step": 29578 }, { "epoch": 3.147371781230049, "grad_norm": 10.5703125, "learning_rate": 8.96677799886299e-06, "loss": 0.5406, "step": 29579 }, { "epoch": 3.1474781868482657, "grad_norm": 11.1484375, "learning_rate": 8.966742467310972e-06, "loss": 0.4482, "step": 29580 }, { "epoch": 3.1475845924664823, "grad_norm": 14.5546875, "learning_rate": 8.966706935758954e-06, "loss": 0.6152, "step": 29581 }, { "epoch": 3.1476909980846988, "grad_norm": 13.1015625, "learning_rate": 8.966671404206937e-06, "loss": 0.646, "step": 29582 }, { "epoch": 3.1477974037029157, "grad_norm": 10.578125, "learning_rate": 8.966635872654919e-06, "loss": 0.5627, "step": 29583 }, { "epoch": 3.1479038093211322, "grad_norm": 13.5390625, "learning_rate": 8.9666003411029e-06, "loss": 0.7574, "step": 29584 }, { "epoch": 3.1480102149393487, "grad_norm": 14.1875, "learning_rate": 8.966564809550882e-06, "loss": 0.7996, "step": 29585 }, { "epoch": 3.1481166205575652, "grad_norm": 16.296875, "learning_rate": 8.966529277998863e-06, "loss": 0.5309, "step": 29586 }, { "epoch": 3.148223026175782, "grad_norm": 15.6015625, "learning_rate": 8.966493746446847e-06, "loss": 0.6685, "step": 29587 }, { "epoch": 3.1483294317939987, "grad_norm": 10.203125, "learning_rate": 8.966458214894828e-06, "loss": 0.5472, "step": 29588 }, { "epoch": 3.148435837412215, "grad_norm": 14.6796875, "learning_rate": 8.96642268334281e-06, "loss": 0.6792, "step": 29589 }, { "epoch": 3.148542243030432, "grad_norm": 11.5625, "learning_rate": 8.966387151790792e-06, "loss": 0.6237, "step": 29590 }, { "epoch": 3.1486486486486487, "grad_norm": 12.6875, "learning_rate": 8.966351620238773e-06, "loss": 0.6512, "step": 29591 }, { "epoch": 3.148755054266865, "grad_norm": 12.1796875, "learning_rate": 8.966316088686755e-06, "loss": 0.53, "step": 29592 }, { "epoch": 3.148861459885082, "grad_norm": 15.0390625, "learning_rate": 8.966280557134736e-06, "loss": 0.7458, "step": 29593 }, { "epoch": 3.1489678655032987, "grad_norm": 8.484375, "learning_rate": 8.966245025582718e-06, "loss": 0.3904, "step": 29594 }, { "epoch": 3.149074271121515, "grad_norm": 14.59375, "learning_rate": 8.9662094940307e-06, "loss": 0.6985, "step": 29595 }, { "epoch": 3.1491806767397317, "grad_norm": 11.359375, "learning_rate": 8.966173962478681e-06, "loss": 0.5526, "step": 29596 }, { "epoch": 3.1492870823579486, "grad_norm": 19.828125, "learning_rate": 8.966138430926664e-06, "loss": 0.7806, "step": 29597 }, { "epoch": 3.149393487976165, "grad_norm": 12.859375, "learning_rate": 8.966102899374646e-06, "loss": 0.6836, "step": 29598 }, { "epoch": 3.1494998935943816, "grad_norm": 10.96875, "learning_rate": 8.966067367822628e-06, "loss": 0.5008, "step": 29599 }, { "epoch": 3.1496062992125986, "grad_norm": 11.984375, "learning_rate": 8.96603183627061e-06, "loss": 0.6483, "step": 29600 }, { "epoch": 3.149712704830815, "grad_norm": 12.8125, "learning_rate": 8.965996304718591e-06, "loss": 0.6069, "step": 29601 }, { "epoch": 3.1498191104490316, "grad_norm": 12.8671875, "learning_rate": 8.965960773166573e-06, "loss": 0.6165, "step": 29602 }, { "epoch": 3.149925516067248, "grad_norm": 13.2421875, "learning_rate": 8.965925241614554e-06, "loss": 0.6111, "step": 29603 }, { "epoch": 3.150031921685465, "grad_norm": 14.0859375, "learning_rate": 8.965889710062536e-06, "loss": 0.7377, "step": 29604 }, { "epoch": 3.1501383273036816, "grad_norm": 12.71875, "learning_rate": 8.965854178510517e-06, "loss": 0.5578, "step": 29605 }, { "epoch": 3.150244732921898, "grad_norm": 11.046875, "learning_rate": 8.965818646958499e-06, "loss": 0.6234, "step": 29606 }, { "epoch": 3.150351138540115, "grad_norm": 14.3046875, "learning_rate": 8.96578311540648e-06, "loss": 0.6061, "step": 29607 }, { "epoch": 3.1504575441583316, "grad_norm": 14.046875, "learning_rate": 8.965747583854464e-06, "loss": 0.7779, "step": 29608 }, { "epoch": 3.150563949776548, "grad_norm": 12.03125, "learning_rate": 8.965712052302445e-06, "loss": 0.5824, "step": 29609 }, { "epoch": 3.150670355394765, "grad_norm": 18.59375, "learning_rate": 8.965676520750427e-06, "loss": 0.6889, "step": 29610 }, { "epoch": 3.1507767610129815, "grad_norm": 11.8984375, "learning_rate": 8.965640989198409e-06, "loss": 0.6005, "step": 29611 }, { "epoch": 3.150883166631198, "grad_norm": 15.578125, "learning_rate": 8.96560545764639e-06, "loss": 0.6453, "step": 29612 }, { "epoch": 3.150989572249415, "grad_norm": 13.125, "learning_rate": 8.965569926094374e-06, "loss": 0.5467, "step": 29613 }, { "epoch": 3.1510959778676315, "grad_norm": 13.3671875, "learning_rate": 8.965534394542355e-06, "loss": 0.6848, "step": 29614 }, { "epoch": 3.151202383485848, "grad_norm": 13.4921875, "learning_rate": 8.965498862990337e-06, "loss": 0.7796, "step": 29615 }, { "epoch": 3.1513087891040645, "grad_norm": 11.3515625, "learning_rate": 8.965463331438318e-06, "loss": 0.583, "step": 29616 }, { "epoch": 3.1514151947222815, "grad_norm": 14.4921875, "learning_rate": 8.9654277998863e-06, "loss": 0.704, "step": 29617 }, { "epoch": 3.151521600340498, "grad_norm": 13.9453125, "learning_rate": 8.965392268334282e-06, "loss": 0.5704, "step": 29618 }, { "epoch": 3.1516280059587145, "grad_norm": 16.578125, "learning_rate": 8.965356736782263e-06, "loss": 0.6601, "step": 29619 }, { "epoch": 3.1517344115769315, "grad_norm": 12.7890625, "learning_rate": 8.965321205230245e-06, "loss": 0.5811, "step": 29620 }, { "epoch": 3.151840817195148, "grad_norm": 13.9609375, "learning_rate": 8.965285673678226e-06, "loss": 0.6532, "step": 29621 }, { "epoch": 3.1519472228133645, "grad_norm": 14.828125, "learning_rate": 8.965250142126208e-06, "loss": 0.645, "step": 29622 }, { "epoch": 3.152053628431581, "grad_norm": 9.875, "learning_rate": 8.965214610574191e-06, "loss": 0.5156, "step": 29623 }, { "epoch": 3.152160034049798, "grad_norm": 13.5390625, "learning_rate": 8.965179079022173e-06, "loss": 0.71, "step": 29624 }, { "epoch": 3.1522664396680145, "grad_norm": 9.5703125, "learning_rate": 8.965143547470155e-06, "loss": 0.4114, "step": 29625 }, { "epoch": 3.152372845286231, "grad_norm": 13.6953125, "learning_rate": 8.965108015918136e-06, "loss": 0.5808, "step": 29626 }, { "epoch": 3.152479250904448, "grad_norm": 12.46875, "learning_rate": 8.965072484366118e-06, "loss": 0.5529, "step": 29627 }, { "epoch": 3.1525856565226644, "grad_norm": 12.28125, "learning_rate": 8.9650369528141e-06, "loss": 0.6506, "step": 29628 }, { "epoch": 3.152692062140881, "grad_norm": 14.6171875, "learning_rate": 8.965001421262081e-06, "loss": 0.7123, "step": 29629 }, { "epoch": 3.152798467759098, "grad_norm": 12.359375, "learning_rate": 8.964965889710063e-06, "loss": 0.5966, "step": 29630 }, { "epoch": 3.1529048733773144, "grad_norm": 10.9296875, "learning_rate": 8.964930358158044e-06, "loss": 0.4931, "step": 29631 }, { "epoch": 3.153011278995531, "grad_norm": 13.8515625, "learning_rate": 8.964894826606026e-06, "loss": 0.7527, "step": 29632 }, { "epoch": 3.1531176846137474, "grad_norm": 12.6875, "learning_rate": 8.964859295054009e-06, "loss": 0.7733, "step": 29633 }, { "epoch": 3.1532240902319644, "grad_norm": 11.015625, "learning_rate": 8.96482376350199e-06, "loss": 0.5916, "step": 29634 }, { "epoch": 3.153330495850181, "grad_norm": 9.546875, "learning_rate": 8.964788231949972e-06, "loss": 0.58, "step": 29635 }, { "epoch": 3.1534369014683974, "grad_norm": 10.1484375, "learning_rate": 8.964752700397954e-06, "loss": 0.5067, "step": 29636 }, { "epoch": 3.1535433070866143, "grad_norm": 10.7890625, "learning_rate": 8.964717168845936e-06, "loss": 0.563, "step": 29637 }, { "epoch": 3.153649712704831, "grad_norm": 11.1328125, "learning_rate": 8.964681637293919e-06, "loss": 0.5643, "step": 29638 }, { "epoch": 3.1537561183230474, "grad_norm": 18.078125, "learning_rate": 8.9646461057419e-06, "loss": 0.7231, "step": 29639 }, { "epoch": 3.153862523941264, "grad_norm": 9.71875, "learning_rate": 8.964610574189882e-06, "loss": 0.4784, "step": 29640 }, { "epoch": 3.153968929559481, "grad_norm": 20.671875, "learning_rate": 8.964575042637864e-06, "loss": 0.6741, "step": 29641 }, { "epoch": 3.1540753351776973, "grad_norm": 12.5703125, "learning_rate": 8.964539511085845e-06, "loss": 0.6204, "step": 29642 }, { "epoch": 3.154181740795914, "grad_norm": 11.359375, "learning_rate": 8.964503979533827e-06, "loss": 0.5949, "step": 29643 }, { "epoch": 3.154288146414131, "grad_norm": 7.58203125, "learning_rate": 8.964468447981808e-06, "loss": 0.4417, "step": 29644 }, { "epoch": 3.1543945520323473, "grad_norm": 18.171875, "learning_rate": 8.96443291642979e-06, "loss": 0.7647, "step": 29645 }, { "epoch": 3.154500957650564, "grad_norm": 16.421875, "learning_rate": 8.964397384877772e-06, "loss": 0.781, "step": 29646 }, { "epoch": 3.1546073632687808, "grad_norm": 13.703125, "learning_rate": 8.964361853325753e-06, "loss": 0.7752, "step": 29647 }, { "epoch": 3.1547137688869973, "grad_norm": 11.3046875, "learning_rate": 8.964326321773735e-06, "loss": 0.5688, "step": 29648 }, { "epoch": 3.154820174505214, "grad_norm": 14.765625, "learning_rate": 8.964290790221718e-06, "loss": 0.6661, "step": 29649 }, { "epoch": 3.1549265801234307, "grad_norm": 10.21875, "learning_rate": 8.9642552586697e-06, "loss": 0.4952, "step": 29650 }, { "epoch": 3.1550329857416473, "grad_norm": 19.59375, "learning_rate": 8.964219727117681e-06, "loss": 0.7812, "step": 29651 }, { "epoch": 3.1551393913598638, "grad_norm": 11.3125, "learning_rate": 8.964184195565663e-06, "loss": 0.5413, "step": 29652 }, { "epoch": 3.1552457969780803, "grad_norm": 11.015625, "learning_rate": 8.964148664013645e-06, "loss": 0.6372, "step": 29653 }, { "epoch": 3.1553522025962972, "grad_norm": 11.4921875, "learning_rate": 8.964113132461626e-06, "loss": 0.4291, "step": 29654 }, { "epoch": 3.1554586082145137, "grad_norm": 14.703125, "learning_rate": 8.964077600909608e-06, "loss": 0.6148, "step": 29655 }, { "epoch": 3.1555650138327302, "grad_norm": 12.4375, "learning_rate": 8.96404206935759e-06, "loss": 0.5947, "step": 29656 }, { "epoch": 3.155671419450947, "grad_norm": 15.0625, "learning_rate": 8.964006537805571e-06, "loss": 0.6882, "step": 29657 }, { "epoch": 3.1557778250691637, "grad_norm": 14.1015625, "learning_rate": 8.963971006253553e-06, "loss": 0.6289, "step": 29658 }, { "epoch": 3.15588423068738, "grad_norm": 13.28125, "learning_rate": 8.963935474701536e-06, "loss": 0.6232, "step": 29659 }, { "epoch": 3.1559906363055967, "grad_norm": 11.90625, "learning_rate": 8.963899943149518e-06, "loss": 0.5394, "step": 29660 }, { "epoch": 3.1560970419238137, "grad_norm": 10.9375, "learning_rate": 8.9638644115975e-06, "loss": 0.5813, "step": 29661 }, { "epoch": 3.15620344754203, "grad_norm": 15.859375, "learning_rate": 8.96382888004548e-06, "loss": 0.7006, "step": 29662 }, { "epoch": 3.1563098531602467, "grad_norm": 13.46875, "learning_rate": 8.963793348493462e-06, "loss": 0.6861, "step": 29663 }, { "epoch": 3.1564162587784637, "grad_norm": 14.1015625, "learning_rate": 8.963757816941446e-06, "loss": 0.615, "step": 29664 }, { "epoch": 3.15652266439668, "grad_norm": 13.9765625, "learning_rate": 8.963722285389427e-06, "loss": 0.5274, "step": 29665 }, { "epoch": 3.1566290700148967, "grad_norm": 12.78125, "learning_rate": 8.963686753837409e-06, "loss": 0.6427, "step": 29666 }, { "epoch": 3.1567354756331136, "grad_norm": 13.6953125, "learning_rate": 8.96365122228539e-06, "loss": 0.548, "step": 29667 }, { "epoch": 3.15684188125133, "grad_norm": 15.890625, "learning_rate": 8.963615690733372e-06, "loss": 0.638, "step": 29668 }, { "epoch": 3.1569482868695466, "grad_norm": 13.53125, "learning_rate": 8.963580159181354e-06, "loss": 0.7069, "step": 29669 }, { "epoch": 3.157054692487763, "grad_norm": 11.3671875, "learning_rate": 8.963544627629335e-06, "loss": 0.5503, "step": 29670 }, { "epoch": 3.15716109810598, "grad_norm": 10.578125, "learning_rate": 8.963509096077317e-06, "loss": 0.6379, "step": 29671 }, { "epoch": 3.1572675037241966, "grad_norm": 15.15625, "learning_rate": 8.963473564525299e-06, "loss": 0.6704, "step": 29672 }, { "epoch": 3.157373909342413, "grad_norm": 8.6171875, "learning_rate": 8.96343803297328e-06, "loss": 0.48, "step": 29673 }, { "epoch": 3.15748031496063, "grad_norm": 15.171875, "learning_rate": 8.963402501421263e-06, "loss": 0.8128, "step": 29674 }, { "epoch": 3.1575867205788466, "grad_norm": 14.109375, "learning_rate": 8.963366969869245e-06, "loss": 0.6408, "step": 29675 }, { "epoch": 3.157693126197063, "grad_norm": 15.1796875, "learning_rate": 8.963331438317227e-06, "loss": 0.6705, "step": 29676 }, { "epoch": 3.1577995318152796, "grad_norm": 14.8984375, "learning_rate": 8.963295906765208e-06, "loss": 0.5724, "step": 29677 }, { "epoch": 3.1579059374334966, "grad_norm": 13.546875, "learning_rate": 8.96326037521319e-06, "loss": 0.6018, "step": 29678 }, { "epoch": 3.158012343051713, "grad_norm": 16.171875, "learning_rate": 8.963224843661173e-06, "loss": 0.6572, "step": 29679 }, { "epoch": 3.1581187486699296, "grad_norm": 13.9140625, "learning_rate": 8.963189312109155e-06, "loss": 0.644, "step": 29680 }, { "epoch": 3.1582251542881465, "grad_norm": 14.84375, "learning_rate": 8.963153780557136e-06, "loss": 0.6902, "step": 29681 }, { "epoch": 3.158331559906363, "grad_norm": 13.796875, "learning_rate": 8.963118249005118e-06, "loss": 0.7544, "step": 29682 }, { "epoch": 3.1584379655245796, "grad_norm": 12.7265625, "learning_rate": 8.9630827174531e-06, "loss": 0.6604, "step": 29683 }, { "epoch": 3.1585443711427965, "grad_norm": 15.28125, "learning_rate": 8.96304718590108e-06, "loss": 0.6185, "step": 29684 }, { "epoch": 3.158650776761013, "grad_norm": 13.1875, "learning_rate": 8.963011654349063e-06, "loss": 0.6591, "step": 29685 }, { "epoch": 3.1587571823792295, "grad_norm": 12.3203125, "learning_rate": 8.962976122797044e-06, "loss": 0.567, "step": 29686 }, { "epoch": 3.1588635879974465, "grad_norm": 13.0078125, "learning_rate": 8.962940591245026e-06, "loss": 0.5641, "step": 29687 }, { "epoch": 3.158969993615663, "grad_norm": 10.0703125, "learning_rate": 8.962905059693008e-06, "loss": 0.3993, "step": 29688 }, { "epoch": 3.1590763992338795, "grad_norm": 13.8046875, "learning_rate": 8.96286952814099e-06, "loss": 0.7537, "step": 29689 }, { "epoch": 3.159182804852096, "grad_norm": 13.828125, "learning_rate": 8.962833996588973e-06, "loss": 0.6554, "step": 29690 }, { "epoch": 3.159289210470313, "grad_norm": 12.4375, "learning_rate": 8.962798465036954e-06, "loss": 0.5954, "step": 29691 }, { "epoch": 3.1593956160885295, "grad_norm": 13.765625, "learning_rate": 8.962762933484936e-06, "loss": 0.5408, "step": 29692 }, { "epoch": 3.159502021706746, "grad_norm": 13.03125, "learning_rate": 8.962727401932917e-06, "loss": 0.6769, "step": 29693 }, { "epoch": 3.159608427324963, "grad_norm": 12.875, "learning_rate": 8.962691870380899e-06, "loss": 0.5118, "step": 29694 }, { "epoch": 3.1597148329431795, "grad_norm": 14.1484375, "learning_rate": 8.96265633882888e-06, "loss": 0.6241, "step": 29695 }, { "epoch": 3.159821238561396, "grad_norm": 11.9765625, "learning_rate": 8.962620807276862e-06, "loss": 0.6372, "step": 29696 }, { "epoch": 3.1599276441796125, "grad_norm": 9.9453125, "learning_rate": 8.962585275724844e-06, "loss": 0.4502, "step": 29697 }, { "epoch": 3.1600340497978294, "grad_norm": 13.8203125, "learning_rate": 8.962549744172825e-06, "loss": 0.6302, "step": 29698 }, { "epoch": 3.160140455416046, "grad_norm": 11.578125, "learning_rate": 8.962514212620807e-06, "loss": 0.7227, "step": 29699 }, { "epoch": 3.1602468610342624, "grad_norm": 14.015625, "learning_rate": 8.96247868106879e-06, "loss": 0.6381, "step": 29700 }, { "epoch": 3.1603532666524794, "grad_norm": 10.296875, "learning_rate": 8.962443149516772e-06, "loss": 0.5506, "step": 29701 }, { "epoch": 3.160459672270696, "grad_norm": 14.1953125, "learning_rate": 8.962407617964754e-06, "loss": 0.7389, "step": 29702 }, { "epoch": 3.1605660778889124, "grad_norm": 11.140625, "learning_rate": 8.962372086412735e-06, "loss": 0.6999, "step": 29703 }, { "epoch": 3.1606724835071294, "grad_norm": 12.5, "learning_rate": 8.962336554860717e-06, "loss": 0.479, "step": 29704 }, { "epoch": 3.160778889125346, "grad_norm": 11.0546875, "learning_rate": 8.9623010233087e-06, "loss": 0.4406, "step": 29705 }, { "epoch": 3.1608852947435624, "grad_norm": 14.140625, "learning_rate": 8.962265491756682e-06, "loss": 0.6802, "step": 29706 }, { "epoch": 3.160991700361779, "grad_norm": 11.3125, "learning_rate": 8.962229960204663e-06, "loss": 0.54, "step": 29707 }, { "epoch": 3.161098105979996, "grad_norm": 17.46875, "learning_rate": 8.962194428652645e-06, "loss": 0.7401, "step": 29708 }, { "epoch": 3.1612045115982124, "grad_norm": 16.84375, "learning_rate": 8.962158897100626e-06, "loss": 0.5978, "step": 29709 }, { "epoch": 3.161310917216429, "grad_norm": 11.6171875, "learning_rate": 8.962123365548608e-06, "loss": 0.6691, "step": 29710 }, { "epoch": 3.161417322834646, "grad_norm": 12.984375, "learning_rate": 8.96208783399659e-06, "loss": 0.6333, "step": 29711 }, { "epoch": 3.1615237284528623, "grad_norm": 15.8671875, "learning_rate": 8.962052302444571e-06, "loss": 0.6698, "step": 29712 }, { "epoch": 3.161630134071079, "grad_norm": 11.3984375, "learning_rate": 8.962016770892553e-06, "loss": 0.5186, "step": 29713 }, { "epoch": 3.161736539689296, "grad_norm": 13.109375, "learning_rate": 8.961981239340535e-06, "loss": 0.6342, "step": 29714 }, { "epoch": 3.1618429453075123, "grad_norm": 11.03125, "learning_rate": 8.961945707788518e-06, "loss": 0.5813, "step": 29715 }, { "epoch": 3.161949350925729, "grad_norm": 13.1171875, "learning_rate": 8.9619101762365e-06, "loss": 0.63, "step": 29716 }, { "epoch": 3.1620557565439453, "grad_norm": 15.890625, "learning_rate": 8.961874644684481e-06, "loss": 0.6881, "step": 29717 }, { "epoch": 3.1621621621621623, "grad_norm": 13.84375, "learning_rate": 8.961839113132463e-06, "loss": 0.6319, "step": 29718 }, { "epoch": 3.162268567780379, "grad_norm": 14.953125, "learning_rate": 8.961803581580444e-06, "loss": 0.5607, "step": 29719 }, { "epoch": 3.1623749733985953, "grad_norm": 13.7578125, "learning_rate": 8.961768050028426e-06, "loss": 0.5785, "step": 29720 }, { "epoch": 3.1624813790168123, "grad_norm": 16.203125, "learning_rate": 8.961732518476407e-06, "loss": 0.6447, "step": 29721 }, { "epoch": 3.1625877846350288, "grad_norm": 11.484375, "learning_rate": 8.961696986924389e-06, "loss": 0.5417, "step": 29722 }, { "epoch": 3.1626941902532453, "grad_norm": 14.2265625, "learning_rate": 8.96166145537237e-06, "loss": 0.6387, "step": 29723 }, { "epoch": 3.1628005958714622, "grad_norm": 12.15625, "learning_rate": 8.961625923820352e-06, "loss": 0.7029, "step": 29724 }, { "epoch": 3.1629070014896787, "grad_norm": 12.75, "learning_rate": 8.961590392268334e-06, "loss": 0.5517, "step": 29725 }, { "epoch": 3.1630134071078952, "grad_norm": 10.1796875, "learning_rate": 8.961554860716317e-06, "loss": 0.5263, "step": 29726 }, { "epoch": 3.1631198127261118, "grad_norm": 13.5625, "learning_rate": 8.961519329164299e-06, "loss": 0.6588, "step": 29727 }, { "epoch": 3.1632262183443287, "grad_norm": 17.8125, "learning_rate": 8.96148379761228e-06, "loss": 0.6901, "step": 29728 }, { "epoch": 3.163332623962545, "grad_norm": 16.609375, "learning_rate": 8.961448266060262e-06, "loss": 0.6882, "step": 29729 }, { "epoch": 3.1634390295807617, "grad_norm": 15.5546875, "learning_rate": 8.961412734508244e-06, "loss": 0.6162, "step": 29730 }, { "epoch": 3.1635454351989787, "grad_norm": 11.8671875, "learning_rate": 8.961377202956227e-06, "loss": 0.6143, "step": 29731 }, { "epoch": 3.163651840817195, "grad_norm": 20.078125, "learning_rate": 8.961341671404209e-06, "loss": 0.7355, "step": 29732 }, { "epoch": 3.1637582464354117, "grad_norm": 10.9921875, "learning_rate": 8.96130613985219e-06, "loss": 0.5457, "step": 29733 }, { "epoch": 3.163864652053628, "grad_norm": 10.3046875, "learning_rate": 8.961270608300172e-06, "loss": 0.4856, "step": 29734 }, { "epoch": 3.163971057671845, "grad_norm": 10.640625, "learning_rate": 8.961235076748153e-06, "loss": 0.4339, "step": 29735 }, { "epoch": 3.1640774632900617, "grad_norm": 15.453125, "learning_rate": 8.961199545196135e-06, "loss": 0.6974, "step": 29736 }, { "epoch": 3.164183868908278, "grad_norm": 14.046875, "learning_rate": 8.961164013644117e-06, "loss": 0.6692, "step": 29737 }, { "epoch": 3.164290274526495, "grad_norm": 11.140625, "learning_rate": 8.961128482092098e-06, "loss": 0.5385, "step": 29738 }, { "epoch": 3.1643966801447116, "grad_norm": 12.234375, "learning_rate": 8.96109295054008e-06, "loss": 0.6473, "step": 29739 }, { "epoch": 3.164503085762928, "grad_norm": 14.0234375, "learning_rate": 8.961057418988061e-06, "loss": 0.599, "step": 29740 }, { "epoch": 3.164609491381145, "grad_norm": 12.453125, "learning_rate": 8.961021887436045e-06, "loss": 0.5602, "step": 29741 }, { "epoch": 3.1647158969993616, "grad_norm": 19.296875, "learning_rate": 8.960986355884026e-06, "loss": 0.6059, "step": 29742 }, { "epoch": 3.164822302617578, "grad_norm": 13.703125, "learning_rate": 8.960950824332008e-06, "loss": 0.7325, "step": 29743 }, { "epoch": 3.164928708235795, "grad_norm": 12.578125, "learning_rate": 8.96091529277999e-06, "loss": 0.5522, "step": 29744 }, { "epoch": 3.1650351138540116, "grad_norm": 17.390625, "learning_rate": 8.960879761227971e-06, "loss": 0.5793, "step": 29745 }, { "epoch": 3.165141519472228, "grad_norm": 16.8125, "learning_rate": 8.960844229675953e-06, "loss": 0.6831, "step": 29746 }, { "epoch": 3.1652479250904446, "grad_norm": 12.4921875, "learning_rate": 8.960808698123934e-06, "loss": 0.7243, "step": 29747 }, { "epoch": 3.1653543307086616, "grad_norm": 20.28125, "learning_rate": 8.960773166571916e-06, "loss": 0.9128, "step": 29748 }, { "epoch": 3.165460736326878, "grad_norm": 11.96875, "learning_rate": 8.960737635019898e-06, "loss": 0.5884, "step": 29749 }, { "epoch": 3.1655671419450946, "grad_norm": 15.9296875, "learning_rate": 8.960702103467879e-06, "loss": 0.7662, "step": 29750 }, { "epoch": 3.1656735475633115, "grad_norm": 13.3515625, "learning_rate": 8.960666571915862e-06, "loss": 0.7688, "step": 29751 }, { "epoch": 3.165779953181528, "grad_norm": 16.359375, "learning_rate": 8.960631040363844e-06, "loss": 0.7462, "step": 29752 }, { "epoch": 3.1658863587997446, "grad_norm": 12.2421875, "learning_rate": 8.960595508811826e-06, "loss": 0.5384, "step": 29753 }, { "epoch": 3.165992764417961, "grad_norm": 16.40625, "learning_rate": 8.960559977259807e-06, "loss": 0.627, "step": 29754 }, { "epoch": 3.166099170036178, "grad_norm": 18.28125, "learning_rate": 8.960524445707789e-06, "loss": 0.6563, "step": 29755 }, { "epoch": 3.1662055756543945, "grad_norm": 13.03125, "learning_rate": 8.960488914155772e-06, "loss": 0.5891, "step": 29756 }, { "epoch": 3.166311981272611, "grad_norm": 16.265625, "learning_rate": 8.960453382603754e-06, "loss": 0.5523, "step": 29757 }, { "epoch": 3.166418386890828, "grad_norm": 11.6484375, "learning_rate": 8.960417851051735e-06, "loss": 0.5973, "step": 29758 }, { "epoch": 3.1665247925090445, "grad_norm": 13.859375, "learning_rate": 8.960382319499717e-06, "loss": 0.6828, "step": 29759 }, { "epoch": 3.166631198127261, "grad_norm": 11.0703125, "learning_rate": 8.960346787947699e-06, "loss": 0.466, "step": 29760 }, { "epoch": 3.166737603745478, "grad_norm": 18.25, "learning_rate": 8.96031125639568e-06, "loss": 0.7131, "step": 29761 }, { "epoch": 3.1668440093636945, "grad_norm": 13.5234375, "learning_rate": 8.960275724843662e-06, "loss": 0.6304, "step": 29762 }, { "epoch": 3.166950414981911, "grad_norm": 15.4921875, "learning_rate": 8.960240193291643e-06, "loss": 0.6379, "step": 29763 }, { "epoch": 3.1670568206001275, "grad_norm": 12.71875, "learning_rate": 8.960204661739625e-06, "loss": 0.6138, "step": 29764 }, { "epoch": 3.1671632262183445, "grad_norm": 12.875, "learning_rate": 8.960169130187607e-06, "loss": 0.6039, "step": 29765 }, { "epoch": 3.167269631836561, "grad_norm": 10.90625, "learning_rate": 8.960133598635588e-06, "loss": 0.5374, "step": 29766 }, { "epoch": 3.1673760374547775, "grad_norm": 10.3359375, "learning_rate": 8.960098067083572e-06, "loss": 0.5079, "step": 29767 }, { "epoch": 3.1674824430729944, "grad_norm": 14.171875, "learning_rate": 8.960062535531553e-06, "loss": 0.6963, "step": 29768 }, { "epoch": 3.167588848691211, "grad_norm": 13.8671875, "learning_rate": 8.960027003979535e-06, "loss": 0.748, "step": 29769 }, { "epoch": 3.1676952543094274, "grad_norm": 14.671875, "learning_rate": 8.959991472427516e-06, "loss": 0.7701, "step": 29770 }, { "epoch": 3.167801659927644, "grad_norm": 10.8125, "learning_rate": 8.959955940875498e-06, "loss": 0.5547, "step": 29771 }, { "epoch": 3.167908065545861, "grad_norm": 13.015625, "learning_rate": 8.959920409323481e-06, "loss": 0.5842, "step": 29772 }, { "epoch": 3.1680144711640774, "grad_norm": 13.78125, "learning_rate": 8.959884877771461e-06, "loss": 0.7436, "step": 29773 }, { "epoch": 3.168120876782294, "grad_norm": 14.140625, "learning_rate": 8.959849346219443e-06, "loss": 0.8729, "step": 29774 }, { "epoch": 3.168227282400511, "grad_norm": 7.60546875, "learning_rate": 8.959813814667424e-06, "loss": 0.6875, "step": 29775 }, { "epoch": 3.1683336880187274, "grad_norm": 15.90625, "learning_rate": 8.959778283115406e-06, "loss": 0.5763, "step": 29776 }, { "epoch": 3.168440093636944, "grad_norm": 10.5625, "learning_rate": 8.95974275156339e-06, "loss": 0.4779, "step": 29777 }, { "epoch": 3.168546499255161, "grad_norm": 10.78125, "learning_rate": 8.959707220011371e-06, "loss": 0.6682, "step": 29778 }, { "epoch": 3.1686529048733774, "grad_norm": 16.078125, "learning_rate": 8.959671688459352e-06, "loss": 0.8235, "step": 29779 }, { "epoch": 3.168759310491594, "grad_norm": 14.6015625, "learning_rate": 8.959636156907334e-06, "loss": 0.6826, "step": 29780 }, { "epoch": 3.168865716109811, "grad_norm": 16.75, "learning_rate": 8.959600625355316e-06, "loss": 0.7753, "step": 29781 }, { "epoch": 3.1689721217280273, "grad_norm": 21.671875, "learning_rate": 8.959565093803299e-06, "loss": 0.7958, "step": 29782 }, { "epoch": 3.169078527346244, "grad_norm": 13.5546875, "learning_rate": 8.95952956225128e-06, "loss": 0.6556, "step": 29783 }, { "epoch": 3.1691849329644604, "grad_norm": 9.8515625, "learning_rate": 8.959494030699262e-06, "loss": 0.5273, "step": 29784 }, { "epoch": 3.1692913385826773, "grad_norm": 11.25, "learning_rate": 8.959458499147244e-06, "loss": 0.5179, "step": 29785 }, { "epoch": 3.169397744200894, "grad_norm": 16.625, "learning_rate": 8.959422967595225e-06, "loss": 0.7286, "step": 29786 }, { "epoch": 3.1695041498191103, "grad_norm": 14.9921875, "learning_rate": 8.959387436043207e-06, "loss": 0.7914, "step": 29787 }, { "epoch": 3.1696105554373273, "grad_norm": 12.4765625, "learning_rate": 8.959351904491189e-06, "loss": 0.7408, "step": 29788 }, { "epoch": 3.169716961055544, "grad_norm": 13.0390625, "learning_rate": 8.95931637293917e-06, "loss": 0.6696, "step": 29789 }, { "epoch": 3.1698233666737603, "grad_norm": 14.203125, "learning_rate": 8.959280841387152e-06, "loss": 0.6403, "step": 29790 }, { "epoch": 3.169929772291977, "grad_norm": 11.859375, "learning_rate": 8.959245309835133e-06, "loss": 0.6073, "step": 29791 }, { "epoch": 3.1700361779101938, "grad_norm": 12.828125, "learning_rate": 8.959209778283117e-06, "loss": 0.8623, "step": 29792 }, { "epoch": 3.1701425835284103, "grad_norm": 12.5625, "learning_rate": 8.959174246731098e-06, "loss": 0.5835, "step": 29793 }, { "epoch": 3.170248989146627, "grad_norm": 9.1484375, "learning_rate": 8.95913871517908e-06, "loss": 0.5535, "step": 29794 }, { "epoch": 3.1703553947648437, "grad_norm": 13.5234375, "learning_rate": 8.959103183627062e-06, "loss": 0.5998, "step": 29795 }, { "epoch": 3.1704618003830602, "grad_norm": 12.2890625, "learning_rate": 8.959067652075043e-06, "loss": 0.7572, "step": 29796 }, { "epoch": 3.1705682060012768, "grad_norm": 13.1875, "learning_rate": 8.959032120523026e-06, "loss": 0.6521, "step": 29797 }, { "epoch": 3.1706746116194937, "grad_norm": 10.0390625, "learning_rate": 8.958996588971008e-06, "loss": 0.5391, "step": 29798 }, { "epoch": 3.17078101723771, "grad_norm": 13.2421875, "learning_rate": 8.95896105741899e-06, "loss": 0.7615, "step": 29799 }, { "epoch": 3.1708874228559267, "grad_norm": 13.6953125, "learning_rate": 8.958925525866971e-06, "loss": 0.6718, "step": 29800 }, { "epoch": 3.1709938284741432, "grad_norm": 15.5859375, "learning_rate": 8.958889994314953e-06, "loss": 0.6836, "step": 29801 }, { "epoch": 3.17110023409236, "grad_norm": 11.609375, "learning_rate": 8.958854462762935e-06, "loss": 0.5878, "step": 29802 }, { "epoch": 3.1712066397105767, "grad_norm": 17.125, "learning_rate": 8.958818931210916e-06, "loss": 0.7064, "step": 29803 }, { "epoch": 3.171313045328793, "grad_norm": 13.4375, "learning_rate": 8.958783399658898e-06, "loss": 0.7337, "step": 29804 }, { "epoch": 3.17141945094701, "grad_norm": 12.25, "learning_rate": 8.95874786810688e-06, "loss": 0.6073, "step": 29805 }, { "epoch": 3.1715258565652267, "grad_norm": 12.625, "learning_rate": 8.958712336554861e-06, "loss": 0.6532, "step": 29806 }, { "epoch": 3.171632262183443, "grad_norm": 12.3828125, "learning_rate": 8.958676805002843e-06, "loss": 0.6711, "step": 29807 }, { "epoch": 3.1717386678016597, "grad_norm": 11.578125, "learning_rate": 8.958641273450826e-06, "loss": 0.5181, "step": 29808 }, { "epoch": 3.1718450734198766, "grad_norm": 15.5234375, "learning_rate": 8.958605741898807e-06, "loss": 0.7794, "step": 29809 }, { "epoch": 3.171951479038093, "grad_norm": 12.546875, "learning_rate": 8.958570210346789e-06, "loss": 0.6249, "step": 29810 }, { "epoch": 3.1720578846563097, "grad_norm": 13.890625, "learning_rate": 8.95853467879477e-06, "loss": 0.6491, "step": 29811 }, { "epoch": 3.1721642902745266, "grad_norm": 11.078125, "learning_rate": 8.958499147242752e-06, "loss": 0.4735, "step": 29812 }, { "epoch": 3.172270695892743, "grad_norm": 15.25, "learning_rate": 8.958463615690734e-06, "loss": 0.634, "step": 29813 }, { "epoch": 3.1723771015109596, "grad_norm": 16.421875, "learning_rate": 8.958428084138715e-06, "loss": 0.769, "step": 29814 }, { "epoch": 3.1724835071291766, "grad_norm": 11.171875, "learning_rate": 8.958392552586697e-06, "loss": 0.6914, "step": 29815 }, { "epoch": 3.172589912747393, "grad_norm": 15.2265625, "learning_rate": 8.958357021034679e-06, "loss": 0.6646, "step": 29816 }, { "epoch": 3.1726963183656096, "grad_norm": 13.6171875, "learning_rate": 8.95832148948266e-06, "loss": 0.5567, "step": 29817 }, { "epoch": 3.1728027239838266, "grad_norm": 17.5, "learning_rate": 8.958285957930644e-06, "loss": 0.7651, "step": 29818 }, { "epoch": 3.172909129602043, "grad_norm": 10.09375, "learning_rate": 8.958250426378625e-06, "loss": 0.4662, "step": 29819 }, { "epoch": 3.1730155352202596, "grad_norm": 11.9921875, "learning_rate": 8.958214894826607e-06, "loss": 0.5216, "step": 29820 }, { "epoch": 3.173121940838476, "grad_norm": 14.0, "learning_rate": 8.958179363274588e-06, "loss": 0.5732, "step": 29821 }, { "epoch": 3.173228346456693, "grad_norm": 11.9140625, "learning_rate": 8.95814383172257e-06, "loss": 0.603, "step": 29822 }, { "epoch": 3.1733347520749096, "grad_norm": 10.109375, "learning_rate": 8.958108300170553e-06, "loss": 0.551, "step": 29823 }, { "epoch": 3.173441157693126, "grad_norm": 11.0234375, "learning_rate": 8.958072768618535e-06, "loss": 0.6155, "step": 29824 }, { "epoch": 3.173547563311343, "grad_norm": 12.734375, "learning_rate": 8.958037237066517e-06, "loss": 0.6838, "step": 29825 }, { "epoch": 3.1736539689295595, "grad_norm": 13.03125, "learning_rate": 8.958001705514498e-06, "loss": 0.6431, "step": 29826 }, { "epoch": 3.173760374547776, "grad_norm": 13.1640625, "learning_rate": 8.95796617396248e-06, "loss": 0.6238, "step": 29827 }, { "epoch": 3.1738667801659926, "grad_norm": 11.421875, "learning_rate": 8.957930642410461e-06, "loss": 0.5147, "step": 29828 }, { "epoch": 3.1739731857842095, "grad_norm": 13.328125, "learning_rate": 8.957895110858443e-06, "loss": 0.5804, "step": 29829 }, { "epoch": 3.174079591402426, "grad_norm": 11.953125, "learning_rate": 8.957859579306425e-06, "loss": 0.5129, "step": 29830 }, { "epoch": 3.1741859970206425, "grad_norm": 18.21875, "learning_rate": 8.957824047754406e-06, "loss": 0.6661, "step": 29831 }, { "epoch": 3.1742924026388595, "grad_norm": 12.484375, "learning_rate": 8.957788516202388e-06, "loss": 0.5263, "step": 29832 }, { "epoch": 3.174398808257076, "grad_norm": 14.8203125, "learning_rate": 8.957752984650371e-06, "loss": 0.6404, "step": 29833 }, { "epoch": 3.1745052138752925, "grad_norm": 10.734375, "learning_rate": 8.957717453098353e-06, "loss": 0.5194, "step": 29834 }, { "epoch": 3.1746116194935095, "grad_norm": 13.4453125, "learning_rate": 8.957681921546334e-06, "loss": 0.509, "step": 29835 }, { "epoch": 3.174718025111726, "grad_norm": 13.7734375, "learning_rate": 8.957646389994316e-06, "loss": 0.7979, "step": 29836 }, { "epoch": 3.1748244307299425, "grad_norm": 13.2734375, "learning_rate": 8.957610858442298e-06, "loss": 0.5768, "step": 29837 }, { "epoch": 3.174930836348159, "grad_norm": 11.5078125, "learning_rate": 8.957575326890279e-06, "loss": 0.5456, "step": 29838 }, { "epoch": 3.175037241966376, "grad_norm": 13.859375, "learning_rate": 8.95753979533826e-06, "loss": 0.761, "step": 29839 }, { "epoch": 3.1751436475845924, "grad_norm": 15.9765625, "learning_rate": 8.957504263786242e-06, "loss": 0.7906, "step": 29840 }, { "epoch": 3.175250053202809, "grad_norm": 11.71875, "learning_rate": 8.957468732234224e-06, "loss": 0.5859, "step": 29841 }, { "epoch": 3.175356458821026, "grad_norm": 12.2578125, "learning_rate": 8.957433200682206e-06, "loss": 0.6841, "step": 29842 }, { "epoch": 3.1754628644392424, "grad_norm": 13.453125, "learning_rate": 8.957397669130187e-06, "loss": 0.6424, "step": 29843 }, { "epoch": 3.175569270057459, "grad_norm": 13.046875, "learning_rate": 8.95736213757817e-06, "loss": 0.6367, "step": 29844 }, { "epoch": 3.175675675675676, "grad_norm": 10.6484375, "learning_rate": 8.957326606026152e-06, "loss": 0.5306, "step": 29845 }, { "epoch": 3.1757820812938924, "grad_norm": 13.328125, "learning_rate": 8.957291074474134e-06, "loss": 0.6015, "step": 29846 }, { "epoch": 3.175888486912109, "grad_norm": 12.859375, "learning_rate": 8.957255542922115e-06, "loss": 0.7874, "step": 29847 }, { "epoch": 3.1759948925303254, "grad_norm": 12.5859375, "learning_rate": 8.957220011370097e-06, "loss": 0.5163, "step": 29848 }, { "epoch": 3.1761012981485424, "grad_norm": 10.875, "learning_rate": 8.95718447981808e-06, "loss": 0.6938, "step": 29849 }, { "epoch": 3.176207703766759, "grad_norm": 12.59375, "learning_rate": 8.957148948266062e-06, "loss": 0.6513, "step": 29850 }, { "epoch": 3.1763141093849754, "grad_norm": 14.390625, "learning_rate": 8.957113416714043e-06, "loss": 0.7241, "step": 29851 }, { "epoch": 3.1764205150031923, "grad_norm": 13.96875, "learning_rate": 8.957077885162025e-06, "loss": 0.6856, "step": 29852 }, { "epoch": 3.176526920621409, "grad_norm": 13.9609375, "learning_rate": 8.957042353610007e-06, "loss": 0.5884, "step": 29853 }, { "epoch": 3.1766333262396254, "grad_norm": 15.1640625, "learning_rate": 8.957006822057988e-06, "loss": 0.651, "step": 29854 }, { "epoch": 3.1767397318578423, "grad_norm": 11.9921875, "learning_rate": 8.95697129050597e-06, "loss": 0.6829, "step": 29855 }, { "epoch": 3.176846137476059, "grad_norm": 11.4921875, "learning_rate": 8.956935758953951e-06, "loss": 0.5836, "step": 29856 }, { "epoch": 3.1769525430942753, "grad_norm": 15.53125, "learning_rate": 8.956900227401933e-06, "loss": 0.7571, "step": 29857 }, { "epoch": 3.177058948712492, "grad_norm": 10.640625, "learning_rate": 8.956864695849915e-06, "loss": 0.4377, "step": 29858 }, { "epoch": 3.177165354330709, "grad_norm": 13.9296875, "learning_rate": 8.956829164297898e-06, "loss": 0.5805, "step": 29859 }, { "epoch": 3.1772717599489253, "grad_norm": 11.4140625, "learning_rate": 8.95679363274588e-06, "loss": 0.4205, "step": 29860 }, { "epoch": 3.177378165567142, "grad_norm": 11.9609375, "learning_rate": 8.956758101193861e-06, "loss": 0.6307, "step": 29861 }, { "epoch": 3.1774845711853588, "grad_norm": 11.2265625, "learning_rate": 8.956722569641843e-06, "loss": 0.5151, "step": 29862 }, { "epoch": 3.1775909768035753, "grad_norm": 13.578125, "learning_rate": 8.956687038089824e-06, "loss": 0.5469, "step": 29863 }, { "epoch": 3.177697382421792, "grad_norm": 17.25, "learning_rate": 8.956651506537806e-06, "loss": 0.8103, "step": 29864 }, { "epoch": 3.1778037880400083, "grad_norm": 13.0078125, "learning_rate": 8.956615974985788e-06, "loss": 0.5949, "step": 29865 }, { "epoch": 3.1779101936582252, "grad_norm": 11.6640625, "learning_rate": 8.95658044343377e-06, "loss": 0.6779, "step": 29866 }, { "epoch": 3.1780165992764418, "grad_norm": 14.3828125, "learning_rate": 8.95654491188175e-06, "loss": 0.5838, "step": 29867 }, { "epoch": 3.1781230048946583, "grad_norm": 11.765625, "learning_rate": 8.956509380329732e-06, "loss": 0.5638, "step": 29868 }, { "epoch": 3.178229410512875, "grad_norm": 9.65625, "learning_rate": 8.956473848777716e-06, "loss": 0.636, "step": 29869 }, { "epoch": 3.1783358161310917, "grad_norm": 9.7421875, "learning_rate": 8.956438317225697e-06, "loss": 0.5711, "step": 29870 }, { "epoch": 3.1784422217493082, "grad_norm": 10.625, "learning_rate": 8.956402785673679e-06, "loss": 0.4916, "step": 29871 }, { "epoch": 3.178548627367525, "grad_norm": 12.4453125, "learning_rate": 8.95636725412166e-06, "loss": 0.5633, "step": 29872 }, { "epoch": 3.1786550329857417, "grad_norm": 14.15625, "learning_rate": 8.956331722569642e-06, "loss": 0.6357, "step": 29873 }, { "epoch": 3.178761438603958, "grad_norm": 15.2265625, "learning_rate": 8.956296191017625e-06, "loss": 0.7219, "step": 29874 }, { "epoch": 3.178867844222175, "grad_norm": 9.484375, "learning_rate": 8.956260659465607e-06, "loss": 0.651, "step": 29875 }, { "epoch": 3.1789742498403917, "grad_norm": 14.5546875, "learning_rate": 8.956225127913589e-06, "loss": 0.5573, "step": 29876 }, { "epoch": 3.179080655458608, "grad_norm": 13.6171875, "learning_rate": 8.95618959636157e-06, "loss": 0.5821, "step": 29877 }, { "epoch": 3.1791870610768247, "grad_norm": 15.0625, "learning_rate": 8.956154064809552e-06, "loss": 0.6945, "step": 29878 }, { "epoch": 3.1792934666950416, "grad_norm": 14.4453125, "learning_rate": 8.956118533257533e-06, "loss": 0.6771, "step": 29879 }, { "epoch": 3.179399872313258, "grad_norm": 11.5390625, "learning_rate": 8.956083001705515e-06, "loss": 0.6231, "step": 29880 }, { "epoch": 3.1795062779314747, "grad_norm": 15.53125, "learning_rate": 8.956047470153497e-06, "loss": 0.6766, "step": 29881 }, { "epoch": 3.1796126835496916, "grad_norm": 12.3359375, "learning_rate": 8.956011938601478e-06, "loss": 0.5817, "step": 29882 }, { "epoch": 3.179719089167908, "grad_norm": 10.859375, "learning_rate": 8.95597640704946e-06, "loss": 0.4883, "step": 29883 }, { "epoch": 3.1798254947861246, "grad_norm": 12.890625, "learning_rate": 8.955940875497442e-06, "loss": 0.7376, "step": 29884 }, { "epoch": 3.179931900404341, "grad_norm": 12.1796875, "learning_rate": 8.955905343945425e-06, "loss": 0.7608, "step": 29885 }, { "epoch": 3.180038306022558, "grad_norm": 12.015625, "learning_rate": 8.955869812393406e-06, "loss": 0.6205, "step": 29886 }, { "epoch": 3.1801447116407746, "grad_norm": 16.421875, "learning_rate": 8.955834280841388e-06, "loss": 0.7328, "step": 29887 }, { "epoch": 3.180251117258991, "grad_norm": 13.390625, "learning_rate": 8.95579874928937e-06, "loss": 0.5767, "step": 29888 }, { "epoch": 3.180357522877208, "grad_norm": 15.7265625, "learning_rate": 8.955763217737351e-06, "loss": 0.563, "step": 29889 }, { "epoch": 3.1804639284954246, "grad_norm": 25.828125, "learning_rate": 8.955727686185335e-06, "loss": 0.9429, "step": 29890 }, { "epoch": 3.180570334113641, "grad_norm": 9.9375, "learning_rate": 8.955692154633316e-06, "loss": 0.598, "step": 29891 }, { "epoch": 3.180676739731858, "grad_norm": 12.84375, "learning_rate": 8.955656623081296e-06, "loss": 0.6838, "step": 29892 }, { "epoch": 3.1807831453500746, "grad_norm": 11.890625, "learning_rate": 8.955621091529278e-06, "loss": 0.5522, "step": 29893 }, { "epoch": 3.180889550968291, "grad_norm": 14.484375, "learning_rate": 8.95558555997726e-06, "loss": 0.7671, "step": 29894 }, { "epoch": 3.1809959565865076, "grad_norm": 14.96875, "learning_rate": 8.955550028425243e-06, "loss": 0.6346, "step": 29895 }, { "epoch": 3.1811023622047245, "grad_norm": 11.3046875, "learning_rate": 8.955514496873224e-06, "loss": 0.5171, "step": 29896 }, { "epoch": 3.181208767822941, "grad_norm": 10.59375, "learning_rate": 8.955478965321206e-06, "loss": 0.5371, "step": 29897 }, { "epoch": 3.1813151734411576, "grad_norm": 16.390625, "learning_rate": 8.955443433769187e-06, "loss": 0.7638, "step": 29898 }, { "epoch": 3.1814215790593745, "grad_norm": 17.703125, "learning_rate": 8.955407902217169e-06, "loss": 0.7546, "step": 29899 }, { "epoch": 3.181527984677591, "grad_norm": 13.1328125, "learning_rate": 8.955372370665152e-06, "loss": 0.7588, "step": 29900 }, { "epoch": 3.1816343902958075, "grad_norm": 10.3359375, "learning_rate": 8.955336839113134e-06, "loss": 0.5742, "step": 29901 }, { "epoch": 3.181740795914024, "grad_norm": 13.203125, "learning_rate": 8.955301307561116e-06, "loss": 0.697, "step": 29902 }, { "epoch": 3.181847201532241, "grad_norm": 11.046875, "learning_rate": 8.955265776009097e-06, "loss": 0.6015, "step": 29903 }, { "epoch": 3.1819536071504575, "grad_norm": 11.6484375, "learning_rate": 8.955230244457079e-06, "loss": 0.63, "step": 29904 }, { "epoch": 3.182060012768674, "grad_norm": 9.9453125, "learning_rate": 8.95519471290506e-06, "loss": 0.506, "step": 29905 }, { "epoch": 3.182166418386891, "grad_norm": 12.5078125, "learning_rate": 8.955159181353042e-06, "loss": 0.5995, "step": 29906 }, { "epoch": 3.1822728240051075, "grad_norm": 12.171875, "learning_rate": 8.955123649801024e-06, "loss": 0.6308, "step": 29907 }, { "epoch": 3.182379229623324, "grad_norm": 11.7421875, "learning_rate": 8.955088118249005e-06, "loss": 0.6341, "step": 29908 }, { "epoch": 3.182485635241541, "grad_norm": 14.28125, "learning_rate": 8.955052586696987e-06, "loss": 0.5508, "step": 29909 }, { "epoch": 3.1825920408597574, "grad_norm": 16.8125, "learning_rate": 8.95501705514497e-06, "loss": 0.7377, "step": 29910 }, { "epoch": 3.182698446477974, "grad_norm": 11.3203125, "learning_rate": 8.954981523592952e-06, "loss": 0.5626, "step": 29911 }, { "epoch": 3.182804852096191, "grad_norm": 15.5625, "learning_rate": 8.954945992040933e-06, "loss": 0.6631, "step": 29912 }, { "epoch": 3.1829112577144074, "grad_norm": 14.8671875, "learning_rate": 8.954910460488915e-06, "loss": 0.624, "step": 29913 }, { "epoch": 3.183017663332624, "grad_norm": 11.203125, "learning_rate": 8.954874928936896e-06, "loss": 0.6748, "step": 29914 }, { "epoch": 3.1831240689508404, "grad_norm": 11.7578125, "learning_rate": 8.954839397384878e-06, "loss": 0.6393, "step": 29915 }, { "epoch": 3.1832304745690574, "grad_norm": 12.1640625, "learning_rate": 8.954803865832861e-06, "loss": 0.4332, "step": 29916 }, { "epoch": 3.183336880187274, "grad_norm": 9.2734375, "learning_rate": 8.954768334280843e-06, "loss": 0.4746, "step": 29917 }, { "epoch": 3.1834432858054904, "grad_norm": 12.8203125, "learning_rate": 8.954732802728825e-06, "loss": 0.7055, "step": 29918 }, { "epoch": 3.1835496914237074, "grad_norm": 13.8203125, "learning_rate": 8.954697271176806e-06, "loss": 0.5868, "step": 29919 }, { "epoch": 3.183656097041924, "grad_norm": 11.8671875, "learning_rate": 8.954661739624788e-06, "loss": 0.5475, "step": 29920 }, { "epoch": 3.1837625026601404, "grad_norm": 16.828125, "learning_rate": 8.95462620807277e-06, "loss": 0.7741, "step": 29921 }, { "epoch": 3.183868908278357, "grad_norm": 12.796875, "learning_rate": 8.954590676520751e-06, "loss": 0.6142, "step": 29922 }, { "epoch": 3.183975313896574, "grad_norm": 22.0, "learning_rate": 8.954555144968733e-06, "loss": 0.8265, "step": 29923 }, { "epoch": 3.1840817195147904, "grad_norm": 14.203125, "learning_rate": 8.954519613416714e-06, "loss": 0.6126, "step": 29924 }, { "epoch": 3.184188125133007, "grad_norm": 13.2109375, "learning_rate": 8.954484081864696e-06, "loss": 0.6153, "step": 29925 }, { "epoch": 3.184294530751224, "grad_norm": 11.3828125, "learning_rate": 8.95444855031268e-06, "loss": 0.5893, "step": 29926 }, { "epoch": 3.1844009363694403, "grad_norm": 12.2734375, "learning_rate": 8.95441301876066e-06, "loss": 0.5365, "step": 29927 }, { "epoch": 3.184507341987657, "grad_norm": 9.4921875, "learning_rate": 8.954377487208642e-06, "loss": 0.4924, "step": 29928 }, { "epoch": 3.184613747605874, "grad_norm": 13.4609375, "learning_rate": 8.954341955656624e-06, "loss": 0.6322, "step": 29929 }, { "epoch": 3.1847201532240903, "grad_norm": 13.546875, "learning_rate": 8.954306424104606e-06, "loss": 0.756, "step": 29930 }, { "epoch": 3.184826558842307, "grad_norm": 12.34375, "learning_rate": 8.954270892552587e-06, "loss": 0.5278, "step": 29931 }, { "epoch": 3.1849329644605233, "grad_norm": 14.9140625, "learning_rate": 8.954235361000569e-06, "loss": 0.6848, "step": 29932 }, { "epoch": 3.1850393700787403, "grad_norm": 13.6484375, "learning_rate": 8.95419982944855e-06, "loss": 0.5986, "step": 29933 }, { "epoch": 3.185145775696957, "grad_norm": 14.578125, "learning_rate": 8.954164297896532e-06, "loss": 0.6213, "step": 29934 }, { "epoch": 3.1852521813151733, "grad_norm": 12.1640625, "learning_rate": 8.954128766344514e-06, "loss": 0.501, "step": 29935 }, { "epoch": 3.1853585869333902, "grad_norm": 12.609375, "learning_rate": 8.954093234792497e-06, "loss": 0.5644, "step": 29936 }, { "epoch": 3.1854649925516068, "grad_norm": 17.296875, "learning_rate": 8.954057703240479e-06, "loss": 0.6475, "step": 29937 }, { "epoch": 3.1855713981698233, "grad_norm": 12.6953125, "learning_rate": 8.95402217168846e-06, "loss": 0.8574, "step": 29938 }, { "epoch": 3.1856778037880398, "grad_norm": 12.9609375, "learning_rate": 8.953986640136442e-06, "loss": 0.5975, "step": 29939 }, { "epoch": 3.1857842094062567, "grad_norm": 19.53125, "learning_rate": 8.953951108584423e-06, "loss": 0.6131, "step": 29940 }, { "epoch": 3.1858906150244732, "grad_norm": 12.3125, "learning_rate": 8.953915577032407e-06, "loss": 0.5479, "step": 29941 }, { "epoch": 3.1859970206426897, "grad_norm": 18.140625, "learning_rate": 8.953880045480388e-06, "loss": 0.6977, "step": 29942 }, { "epoch": 3.1861034262609067, "grad_norm": 15.6484375, "learning_rate": 8.95384451392837e-06, "loss": 0.7049, "step": 29943 }, { "epoch": 3.186209831879123, "grad_norm": 15.6875, "learning_rate": 8.953808982376351e-06, "loss": 0.7626, "step": 29944 }, { "epoch": 3.1863162374973397, "grad_norm": 12.828125, "learning_rate": 8.953773450824333e-06, "loss": 0.5372, "step": 29945 }, { "epoch": 3.1864226431155567, "grad_norm": 12.6484375, "learning_rate": 8.953737919272315e-06, "loss": 0.6442, "step": 29946 }, { "epoch": 3.186529048733773, "grad_norm": 12.5234375, "learning_rate": 8.953702387720296e-06, "loss": 0.5387, "step": 29947 }, { "epoch": 3.1866354543519897, "grad_norm": 9.6953125, "learning_rate": 8.953666856168278e-06, "loss": 0.5711, "step": 29948 }, { "epoch": 3.1867418599702066, "grad_norm": 12.9296875, "learning_rate": 8.95363132461626e-06, "loss": 0.6158, "step": 29949 }, { "epoch": 3.186848265588423, "grad_norm": 16.1875, "learning_rate": 8.953595793064241e-06, "loss": 0.708, "step": 29950 }, { "epoch": 3.1869546712066397, "grad_norm": 10.640625, "learning_rate": 8.953560261512224e-06, "loss": 0.5255, "step": 29951 }, { "epoch": 3.187061076824856, "grad_norm": 12.4375, "learning_rate": 8.953524729960206e-06, "loss": 0.5817, "step": 29952 }, { "epoch": 3.187167482443073, "grad_norm": 14.546875, "learning_rate": 8.953489198408188e-06, "loss": 0.5889, "step": 29953 }, { "epoch": 3.1872738880612896, "grad_norm": 12.96875, "learning_rate": 8.95345366685617e-06, "loss": 0.6272, "step": 29954 }, { "epoch": 3.187380293679506, "grad_norm": 12.5234375, "learning_rate": 8.95341813530415e-06, "loss": 0.7353, "step": 29955 }, { "epoch": 3.187486699297723, "grad_norm": 16.515625, "learning_rate": 8.953382603752132e-06, "loss": 0.5685, "step": 29956 }, { "epoch": 3.1875931049159396, "grad_norm": 14.9140625, "learning_rate": 8.953347072200114e-06, "loss": 0.7153, "step": 29957 }, { "epoch": 3.187699510534156, "grad_norm": 11.8046875, "learning_rate": 8.953311540648096e-06, "loss": 0.5396, "step": 29958 }, { "epoch": 3.1878059161523726, "grad_norm": 11.953125, "learning_rate": 8.953276009096077e-06, "loss": 0.5682, "step": 29959 }, { "epoch": 3.1879123217705896, "grad_norm": 11.8671875, "learning_rate": 8.953240477544059e-06, "loss": 0.593, "step": 29960 }, { "epoch": 3.188018727388806, "grad_norm": 11.9609375, "learning_rate": 8.95320494599204e-06, "loss": 0.6272, "step": 29961 }, { "epoch": 3.1881251330070226, "grad_norm": 12.9765625, "learning_rate": 8.953169414440024e-06, "loss": 0.6698, "step": 29962 }, { "epoch": 3.1882315386252396, "grad_norm": 9.9609375, "learning_rate": 8.953133882888005e-06, "loss": 0.4674, "step": 29963 }, { "epoch": 3.188337944243456, "grad_norm": 12.234375, "learning_rate": 8.953098351335987e-06, "loss": 0.5766, "step": 29964 }, { "epoch": 3.1884443498616726, "grad_norm": 12.90625, "learning_rate": 8.953062819783969e-06, "loss": 0.5429, "step": 29965 }, { "epoch": 3.1885507554798895, "grad_norm": 12.625, "learning_rate": 8.95302728823195e-06, "loss": 0.5228, "step": 29966 }, { "epoch": 3.188657161098106, "grad_norm": 15.1953125, "learning_rate": 8.952991756679934e-06, "loss": 0.8456, "step": 29967 }, { "epoch": 3.1887635667163226, "grad_norm": 17.328125, "learning_rate": 8.952956225127915e-06, "loss": 0.6908, "step": 29968 }, { "epoch": 3.188869972334539, "grad_norm": 10.3359375, "learning_rate": 8.952920693575897e-06, "loss": 0.4934, "step": 29969 }, { "epoch": 3.188976377952756, "grad_norm": 11.4375, "learning_rate": 8.952885162023878e-06, "loss": 0.5277, "step": 29970 }, { "epoch": 3.1890827835709725, "grad_norm": 11.859375, "learning_rate": 8.95284963047186e-06, "loss": 0.5892, "step": 29971 }, { "epoch": 3.189189189189189, "grad_norm": 9.7421875, "learning_rate": 8.952814098919842e-06, "loss": 0.4194, "step": 29972 }, { "epoch": 3.189295594807406, "grad_norm": 11.6640625, "learning_rate": 8.952778567367823e-06, "loss": 0.5606, "step": 29973 }, { "epoch": 3.1894020004256225, "grad_norm": 8.828125, "learning_rate": 8.952743035815805e-06, "loss": 0.5068, "step": 29974 }, { "epoch": 3.189508406043839, "grad_norm": 12.25, "learning_rate": 8.952707504263786e-06, "loss": 0.5775, "step": 29975 }, { "epoch": 3.189614811662056, "grad_norm": 11.25, "learning_rate": 8.952671972711768e-06, "loss": 0.7017, "step": 29976 }, { "epoch": 3.1897212172802725, "grad_norm": 10.875, "learning_rate": 8.952636441159751e-06, "loss": 0.5404, "step": 29977 }, { "epoch": 3.189827622898489, "grad_norm": 22.0, "learning_rate": 8.952600909607733e-06, "loss": 1.1278, "step": 29978 }, { "epoch": 3.1899340285167055, "grad_norm": 17.609375, "learning_rate": 8.952565378055714e-06, "loss": 0.6883, "step": 29979 }, { "epoch": 3.1900404341349224, "grad_norm": 13.34375, "learning_rate": 8.952529846503696e-06, "loss": 0.7157, "step": 29980 }, { "epoch": 3.190146839753139, "grad_norm": 10.859375, "learning_rate": 8.952494314951678e-06, "loss": 0.561, "step": 29981 }, { "epoch": 3.1902532453713555, "grad_norm": 12.6015625, "learning_rate": 8.95245878339966e-06, "loss": 0.7388, "step": 29982 }, { "epoch": 3.1903596509895724, "grad_norm": 13.5859375, "learning_rate": 8.952423251847641e-06, "loss": 0.5569, "step": 29983 }, { "epoch": 3.190466056607789, "grad_norm": 10.578125, "learning_rate": 8.952387720295623e-06, "loss": 0.6794, "step": 29984 }, { "epoch": 3.1905724622260054, "grad_norm": 13.2578125, "learning_rate": 8.952352188743604e-06, "loss": 0.6054, "step": 29985 }, { "epoch": 3.1906788678442224, "grad_norm": 12.3125, "learning_rate": 8.952316657191586e-06, "loss": 0.7984, "step": 29986 }, { "epoch": 3.190785273462439, "grad_norm": 12.2734375, "learning_rate": 8.952281125639569e-06, "loss": 0.7313, "step": 29987 }, { "epoch": 3.1908916790806554, "grad_norm": 11.96875, "learning_rate": 8.95224559408755e-06, "loss": 0.4752, "step": 29988 }, { "epoch": 3.190998084698872, "grad_norm": 13.953125, "learning_rate": 8.952210062535532e-06, "loss": 0.7756, "step": 29989 }, { "epoch": 3.191104490317089, "grad_norm": 12.984375, "learning_rate": 8.952174530983514e-06, "loss": 0.6667, "step": 29990 }, { "epoch": 3.1912108959353054, "grad_norm": 10.15625, "learning_rate": 8.952138999431495e-06, "loss": 0.5173, "step": 29991 }, { "epoch": 3.191317301553522, "grad_norm": 15.8671875, "learning_rate": 8.952103467879479e-06, "loss": 0.7627, "step": 29992 }, { "epoch": 3.191423707171739, "grad_norm": 12.84375, "learning_rate": 8.95206793632746e-06, "loss": 0.6292, "step": 29993 }, { "epoch": 3.1915301127899554, "grad_norm": 12.5390625, "learning_rate": 8.952032404775442e-06, "loss": 0.551, "step": 29994 }, { "epoch": 3.191636518408172, "grad_norm": 12.828125, "learning_rate": 8.951996873223424e-06, "loss": 0.5257, "step": 29995 }, { "epoch": 3.1917429240263884, "grad_norm": 11.734375, "learning_rate": 8.951961341671405e-06, "loss": 0.6208, "step": 29996 }, { "epoch": 3.1918493296446053, "grad_norm": 13.0078125, "learning_rate": 8.951925810119387e-06, "loss": 0.5416, "step": 29997 }, { "epoch": 3.191955735262822, "grad_norm": 14.265625, "learning_rate": 8.951890278567368e-06, "loss": 0.7932, "step": 29998 }, { "epoch": 3.1920621408810383, "grad_norm": 12.4375, "learning_rate": 8.95185474701535e-06, "loss": 0.643, "step": 29999 }, { "epoch": 3.1921685464992553, "grad_norm": 17.671875, "learning_rate": 8.951819215463332e-06, "loss": 0.6454, "step": 30000 }, { "epoch": 3.192274952117472, "grad_norm": 10.7109375, "learning_rate": 8.951783683911313e-06, "loss": 0.5167, "step": 30001 }, { "epoch": 3.1923813577356883, "grad_norm": 9.34375, "learning_rate": 8.951748152359295e-06, "loss": 0.4764, "step": 30002 }, { "epoch": 3.1924877633539053, "grad_norm": 15.4921875, "learning_rate": 8.951712620807278e-06, "loss": 0.594, "step": 30003 }, { "epoch": 3.192594168972122, "grad_norm": 10.65625, "learning_rate": 8.95167708925526e-06, "loss": 0.5459, "step": 30004 }, { "epoch": 3.1927005745903383, "grad_norm": 17.59375, "learning_rate": 8.951641557703241e-06, "loss": 0.6908, "step": 30005 }, { "epoch": 3.1928069802085552, "grad_norm": 12.78125, "learning_rate": 8.951606026151223e-06, "loss": 0.5963, "step": 30006 }, { "epoch": 3.1929133858267718, "grad_norm": 15.828125, "learning_rate": 8.951570494599205e-06, "loss": 0.6389, "step": 30007 }, { "epoch": 3.1930197914449883, "grad_norm": 11.6953125, "learning_rate": 8.951534963047188e-06, "loss": 0.5598, "step": 30008 }, { "epoch": 3.1931261970632048, "grad_norm": 14.546875, "learning_rate": 8.95149943149517e-06, "loss": 0.5851, "step": 30009 }, { "epoch": 3.1932326026814217, "grad_norm": 10.3046875, "learning_rate": 8.951463899943151e-06, "loss": 0.4901, "step": 30010 }, { "epoch": 3.1933390082996382, "grad_norm": 10.65625, "learning_rate": 8.951428368391133e-06, "loss": 0.5822, "step": 30011 }, { "epoch": 3.1934454139178547, "grad_norm": 12.140625, "learning_rate": 8.951392836839113e-06, "loss": 0.6342, "step": 30012 }, { "epoch": 3.1935518195360717, "grad_norm": 13.578125, "learning_rate": 8.951357305287096e-06, "loss": 0.5848, "step": 30013 }, { "epoch": 3.193658225154288, "grad_norm": 12.5625, "learning_rate": 8.951321773735077e-06, "loss": 0.5514, "step": 30014 }, { "epoch": 3.1937646307725047, "grad_norm": 8.921875, "learning_rate": 8.951286242183059e-06, "loss": 0.3638, "step": 30015 }, { "epoch": 3.1938710363907212, "grad_norm": 17.953125, "learning_rate": 8.95125071063104e-06, "loss": 0.7549, "step": 30016 }, { "epoch": 3.193977442008938, "grad_norm": 11.9296875, "learning_rate": 8.951215179079022e-06, "loss": 0.6387, "step": 30017 }, { "epoch": 3.1940838476271547, "grad_norm": 11.1015625, "learning_rate": 8.951179647527006e-06, "loss": 0.674, "step": 30018 }, { "epoch": 3.194190253245371, "grad_norm": 13.640625, "learning_rate": 8.951144115974987e-06, "loss": 0.9235, "step": 30019 }, { "epoch": 3.194296658863588, "grad_norm": 12.015625, "learning_rate": 8.951108584422969e-06, "loss": 0.5811, "step": 30020 }, { "epoch": 3.1944030644818047, "grad_norm": 13.2109375, "learning_rate": 8.95107305287095e-06, "loss": 0.6124, "step": 30021 }, { "epoch": 3.194509470100021, "grad_norm": 14.1640625, "learning_rate": 8.951037521318932e-06, "loss": 0.6778, "step": 30022 }, { "epoch": 3.194615875718238, "grad_norm": 9.4296875, "learning_rate": 8.951001989766914e-06, "loss": 0.4598, "step": 30023 }, { "epoch": 3.1947222813364546, "grad_norm": 12.0234375, "learning_rate": 8.950966458214895e-06, "loss": 0.7023, "step": 30024 }, { "epoch": 3.194828686954671, "grad_norm": 12.4921875, "learning_rate": 8.950930926662877e-06, "loss": 0.606, "step": 30025 }, { "epoch": 3.1949350925728877, "grad_norm": 11.9765625, "learning_rate": 8.950895395110858e-06, "loss": 0.6434, "step": 30026 }, { "epoch": 3.1950414981911046, "grad_norm": 15.1640625, "learning_rate": 8.95085986355884e-06, "loss": 0.6536, "step": 30027 }, { "epoch": 3.195147903809321, "grad_norm": 11.6328125, "learning_rate": 8.950824332006823e-06, "loss": 0.5961, "step": 30028 }, { "epoch": 3.1952543094275376, "grad_norm": 12.96875, "learning_rate": 8.950788800454805e-06, "loss": 0.624, "step": 30029 }, { "epoch": 3.1953607150457546, "grad_norm": 12.890625, "learning_rate": 8.950753268902787e-06, "loss": 0.645, "step": 30030 }, { "epoch": 3.195467120663971, "grad_norm": 12.203125, "learning_rate": 8.950717737350768e-06, "loss": 0.5653, "step": 30031 }, { "epoch": 3.1955735262821876, "grad_norm": 13.296875, "learning_rate": 8.95068220579875e-06, "loss": 0.588, "step": 30032 }, { "epoch": 3.195679931900404, "grad_norm": 13.28125, "learning_rate": 8.950646674246731e-06, "loss": 0.6314, "step": 30033 }, { "epoch": 3.195786337518621, "grad_norm": 14.40625, "learning_rate": 8.950611142694715e-06, "loss": 0.6805, "step": 30034 }, { "epoch": 3.1958927431368376, "grad_norm": 13.9375, "learning_rate": 8.950575611142696e-06, "loss": 0.6431, "step": 30035 }, { "epoch": 3.195999148755054, "grad_norm": 9.8203125, "learning_rate": 8.950540079590678e-06, "loss": 0.6013, "step": 30036 }, { "epoch": 3.196105554373271, "grad_norm": 14.9765625, "learning_rate": 8.95050454803866e-06, "loss": 0.6975, "step": 30037 }, { "epoch": 3.1962119599914876, "grad_norm": 17.234375, "learning_rate": 8.950469016486641e-06, "loss": 0.8136, "step": 30038 }, { "epoch": 3.196318365609704, "grad_norm": 9.578125, "learning_rate": 8.950433484934623e-06, "loss": 0.4773, "step": 30039 }, { "epoch": 3.196424771227921, "grad_norm": 9.1640625, "learning_rate": 8.950397953382604e-06, "loss": 0.5461, "step": 30040 }, { "epoch": 3.1965311768461375, "grad_norm": 11.25, "learning_rate": 8.950362421830586e-06, "loss": 0.6223, "step": 30041 }, { "epoch": 3.196637582464354, "grad_norm": 11.265625, "learning_rate": 8.950326890278568e-06, "loss": 0.5157, "step": 30042 }, { "epoch": 3.196743988082571, "grad_norm": 23.125, "learning_rate": 8.95029135872655e-06, "loss": 0.8051, "step": 30043 }, { "epoch": 3.1968503937007875, "grad_norm": 14.5703125, "learning_rate": 8.950255827174532e-06, "loss": 0.6992, "step": 30044 }, { "epoch": 3.196956799319004, "grad_norm": 9.7578125, "learning_rate": 8.950220295622514e-06, "loss": 0.4897, "step": 30045 }, { "epoch": 3.1970632049372205, "grad_norm": 13.5078125, "learning_rate": 8.950184764070496e-06, "loss": 0.5982, "step": 30046 }, { "epoch": 3.1971696105554375, "grad_norm": 15.34375, "learning_rate": 8.950149232518477e-06, "loss": 0.6743, "step": 30047 }, { "epoch": 3.197276016173654, "grad_norm": 13.625, "learning_rate": 8.950113700966459e-06, "loss": 0.6958, "step": 30048 }, { "epoch": 3.1973824217918705, "grad_norm": 10.515625, "learning_rate": 8.95007816941444e-06, "loss": 0.5547, "step": 30049 }, { "epoch": 3.1974888274100874, "grad_norm": 14.3125, "learning_rate": 8.950042637862422e-06, "loss": 0.5994, "step": 30050 }, { "epoch": 3.197595233028304, "grad_norm": 11.7421875, "learning_rate": 8.950007106310404e-06, "loss": 0.5753, "step": 30051 }, { "epoch": 3.1977016386465205, "grad_norm": 25.234375, "learning_rate": 8.949971574758385e-06, "loss": 0.7018, "step": 30052 }, { "epoch": 3.197808044264737, "grad_norm": 9.609375, "learning_rate": 8.949936043206367e-06, "loss": 0.4809, "step": 30053 }, { "epoch": 3.197914449882954, "grad_norm": 10.5390625, "learning_rate": 8.94990051165435e-06, "loss": 0.5866, "step": 30054 }, { "epoch": 3.1980208555011704, "grad_norm": 21.171875, "learning_rate": 8.949864980102332e-06, "loss": 0.8359, "step": 30055 }, { "epoch": 3.198127261119387, "grad_norm": 12.3515625, "learning_rate": 8.949829448550313e-06, "loss": 0.6227, "step": 30056 }, { "epoch": 3.198233666737604, "grad_norm": 10.2578125, "learning_rate": 8.949793916998295e-06, "loss": 0.5435, "step": 30057 }, { "epoch": 3.1983400723558204, "grad_norm": 14.5703125, "learning_rate": 8.949758385446277e-06, "loss": 0.6217, "step": 30058 }, { "epoch": 3.198446477974037, "grad_norm": 12.21875, "learning_rate": 8.94972285389426e-06, "loss": 0.6209, "step": 30059 }, { "epoch": 3.198552883592254, "grad_norm": 17.640625, "learning_rate": 8.949687322342242e-06, "loss": 0.6522, "step": 30060 }, { "epoch": 3.1986592892104704, "grad_norm": 16.75, "learning_rate": 8.949651790790223e-06, "loss": 0.7941, "step": 30061 }, { "epoch": 3.198765694828687, "grad_norm": 11.28125, "learning_rate": 8.949616259238205e-06, "loss": 0.5473, "step": 30062 }, { "epoch": 3.1988721004469034, "grad_norm": 18.671875, "learning_rate": 8.949580727686186e-06, "loss": 0.7415, "step": 30063 }, { "epoch": 3.1989785060651204, "grad_norm": 11.8359375, "learning_rate": 8.949545196134168e-06, "loss": 0.5587, "step": 30064 }, { "epoch": 3.199084911683337, "grad_norm": 10.0390625, "learning_rate": 8.94950966458215e-06, "loss": 0.75, "step": 30065 }, { "epoch": 3.1991913173015534, "grad_norm": 11.59375, "learning_rate": 8.949474133030131e-06, "loss": 0.57, "step": 30066 }, { "epoch": 3.1992977229197703, "grad_norm": 11.96875, "learning_rate": 8.949438601478113e-06, "loss": 0.5288, "step": 30067 }, { "epoch": 3.199404128537987, "grad_norm": 12.1953125, "learning_rate": 8.949403069926094e-06, "loss": 0.5809, "step": 30068 }, { "epoch": 3.1995105341562033, "grad_norm": 13.3828125, "learning_rate": 8.949367538374078e-06, "loss": 0.6738, "step": 30069 }, { "epoch": 3.19961693977442, "grad_norm": 12.6953125, "learning_rate": 8.94933200682206e-06, "loss": 0.6711, "step": 30070 }, { "epoch": 3.199723345392637, "grad_norm": 11.2265625, "learning_rate": 8.949296475270041e-06, "loss": 0.6679, "step": 30071 }, { "epoch": 3.1998297510108533, "grad_norm": 14.375, "learning_rate": 8.949260943718023e-06, "loss": 0.5717, "step": 30072 }, { "epoch": 3.19993615662907, "grad_norm": 12.421875, "learning_rate": 8.949225412166004e-06, "loss": 0.656, "step": 30073 }, { "epoch": 3.200042562247287, "grad_norm": 10.9609375, "learning_rate": 8.949189880613986e-06, "loss": 0.6454, "step": 30074 }, { "epoch": 3.2001489678655033, "grad_norm": 11.28125, "learning_rate": 8.949154349061967e-06, "loss": 0.5173, "step": 30075 }, { "epoch": 3.20025537348372, "grad_norm": 16.75, "learning_rate": 8.949118817509949e-06, "loss": 0.6717, "step": 30076 }, { "epoch": 3.2003617791019368, "grad_norm": 15.8515625, "learning_rate": 8.94908328595793e-06, "loss": 0.6424, "step": 30077 }, { "epoch": 3.2004681847201533, "grad_norm": 14.6171875, "learning_rate": 8.949047754405912e-06, "loss": 0.5431, "step": 30078 }, { "epoch": 3.2005745903383698, "grad_norm": 16.484375, "learning_rate": 8.949012222853894e-06, "loss": 0.697, "step": 30079 }, { "epoch": 3.2006809959565867, "grad_norm": 12.109375, "learning_rate": 8.948976691301877e-06, "loss": 0.6324, "step": 30080 }, { "epoch": 3.2007874015748032, "grad_norm": 13.515625, "learning_rate": 8.948941159749859e-06, "loss": 0.5203, "step": 30081 }, { "epoch": 3.2008938071930197, "grad_norm": 15.03125, "learning_rate": 8.94890562819784e-06, "loss": 0.7909, "step": 30082 }, { "epoch": 3.2010002128112363, "grad_norm": 11.8125, "learning_rate": 8.948870096645822e-06, "loss": 0.5231, "step": 30083 }, { "epoch": 3.201106618429453, "grad_norm": 12.734375, "learning_rate": 8.948834565093804e-06, "loss": 0.6478, "step": 30084 }, { "epoch": 3.2012130240476697, "grad_norm": 16.359375, "learning_rate": 8.948799033541787e-06, "loss": 0.699, "step": 30085 }, { "epoch": 3.2013194296658862, "grad_norm": 11.0859375, "learning_rate": 8.948763501989768e-06, "loss": 0.5514, "step": 30086 }, { "epoch": 3.201425835284103, "grad_norm": 10.96875, "learning_rate": 8.94872797043775e-06, "loss": 0.6764, "step": 30087 }, { "epoch": 3.2015322409023197, "grad_norm": 12.828125, "learning_rate": 8.948692438885732e-06, "loss": 0.6903, "step": 30088 }, { "epoch": 3.201638646520536, "grad_norm": 11.734375, "learning_rate": 8.948656907333713e-06, "loss": 0.6291, "step": 30089 }, { "epoch": 3.2017450521387527, "grad_norm": 11.828125, "learning_rate": 8.948621375781695e-06, "loss": 0.5235, "step": 30090 }, { "epoch": 3.2018514577569697, "grad_norm": 14.328125, "learning_rate": 8.948585844229676e-06, "loss": 0.6317, "step": 30091 }, { "epoch": 3.201957863375186, "grad_norm": 12.6328125, "learning_rate": 8.948550312677658e-06, "loss": 0.5387, "step": 30092 }, { "epoch": 3.2020642689934027, "grad_norm": 10.9765625, "learning_rate": 8.94851478112564e-06, "loss": 0.6271, "step": 30093 }, { "epoch": 3.2021706746116196, "grad_norm": 13.4296875, "learning_rate": 8.948479249573621e-06, "loss": 0.8186, "step": 30094 }, { "epoch": 3.202277080229836, "grad_norm": 17.125, "learning_rate": 8.948443718021605e-06, "loss": 0.783, "step": 30095 }, { "epoch": 3.2023834858480527, "grad_norm": 14.953125, "learning_rate": 8.948408186469586e-06, "loss": 0.6517, "step": 30096 }, { "epoch": 3.2024898914662696, "grad_norm": 14.6484375, "learning_rate": 8.948372654917568e-06, "loss": 0.7301, "step": 30097 }, { "epoch": 3.202596297084486, "grad_norm": 12.7578125, "learning_rate": 8.94833712336555e-06, "loss": 0.515, "step": 30098 }, { "epoch": 3.2027027027027026, "grad_norm": 12.8515625, "learning_rate": 8.948301591813531e-06, "loss": 0.6913, "step": 30099 }, { "epoch": 3.202809108320919, "grad_norm": 10.1484375, "learning_rate": 8.948266060261514e-06, "loss": 0.6583, "step": 30100 }, { "epoch": 3.202915513939136, "grad_norm": 11.5078125, "learning_rate": 8.948230528709494e-06, "loss": 0.5574, "step": 30101 }, { "epoch": 3.2030219195573526, "grad_norm": 15.234375, "learning_rate": 8.948194997157476e-06, "loss": 0.6644, "step": 30102 }, { "epoch": 3.203128325175569, "grad_norm": 12.5859375, "learning_rate": 8.948159465605457e-06, "loss": 0.52, "step": 30103 }, { "epoch": 3.203234730793786, "grad_norm": 14.203125, "learning_rate": 8.948123934053439e-06, "loss": 0.6135, "step": 30104 }, { "epoch": 3.2033411364120026, "grad_norm": 16.203125, "learning_rate": 8.948088402501422e-06, "loss": 0.8587, "step": 30105 }, { "epoch": 3.203447542030219, "grad_norm": 13.6484375, "learning_rate": 8.948052870949404e-06, "loss": 0.67, "step": 30106 }, { "epoch": 3.203553947648436, "grad_norm": 17.296875, "learning_rate": 8.948017339397386e-06, "loss": 0.736, "step": 30107 }, { "epoch": 3.2036603532666525, "grad_norm": 18.765625, "learning_rate": 8.947981807845367e-06, "loss": 0.7493, "step": 30108 }, { "epoch": 3.203766758884869, "grad_norm": 12.421875, "learning_rate": 8.947946276293349e-06, "loss": 0.6337, "step": 30109 }, { "epoch": 3.2038731645030856, "grad_norm": 10.8359375, "learning_rate": 8.94791074474133e-06, "loss": 0.6654, "step": 30110 }, { "epoch": 3.2039795701213025, "grad_norm": 15.0234375, "learning_rate": 8.947875213189314e-06, "loss": 0.6909, "step": 30111 }, { "epoch": 3.204085975739519, "grad_norm": 11.0625, "learning_rate": 8.947839681637295e-06, "loss": 0.4517, "step": 30112 }, { "epoch": 3.2041923813577355, "grad_norm": 11.796875, "learning_rate": 8.947804150085277e-06, "loss": 0.6271, "step": 30113 }, { "epoch": 3.2042987869759525, "grad_norm": 14.3359375, "learning_rate": 8.947768618533258e-06, "loss": 0.7468, "step": 30114 }, { "epoch": 3.204405192594169, "grad_norm": 13.3984375, "learning_rate": 8.94773308698124e-06, "loss": 0.6552, "step": 30115 }, { "epoch": 3.2045115982123855, "grad_norm": 23.4375, "learning_rate": 8.947697555429222e-06, "loss": 0.7319, "step": 30116 }, { "epoch": 3.2046180038306025, "grad_norm": 16.359375, "learning_rate": 8.947662023877203e-06, "loss": 0.7861, "step": 30117 }, { "epoch": 3.204724409448819, "grad_norm": 13.859375, "learning_rate": 8.947626492325185e-06, "loss": 0.6315, "step": 30118 }, { "epoch": 3.2048308150670355, "grad_norm": 15.1953125, "learning_rate": 8.947590960773167e-06, "loss": 0.7636, "step": 30119 }, { "epoch": 3.204937220685252, "grad_norm": 16.890625, "learning_rate": 8.947555429221148e-06, "loss": 0.565, "step": 30120 }, { "epoch": 3.205043626303469, "grad_norm": 16.078125, "learning_rate": 8.947519897669131e-06, "loss": 0.8249, "step": 30121 }, { "epoch": 3.2051500319216855, "grad_norm": 12.625, "learning_rate": 8.947484366117113e-06, "loss": 0.7159, "step": 30122 }, { "epoch": 3.205256437539902, "grad_norm": 16.09375, "learning_rate": 8.947448834565095e-06, "loss": 0.7361, "step": 30123 }, { "epoch": 3.205362843158119, "grad_norm": 12.0546875, "learning_rate": 8.947413303013076e-06, "loss": 0.6624, "step": 30124 }, { "epoch": 3.2054692487763354, "grad_norm": 16.03125, "learning_rate": 8.947377771461058e-06, "loss": 0.615, "step": 30125 }, { "epoch": 3.205575654394552, "grad_norm": 10.6875, "learning_rate": 8.947342239909041e-06, "loss": 0.5979, "step": 30126 }, { "epoch": 3.2056820600127685, "grad_norm": 13.640625, "learning_rate": 8.947306708357023e-06, "loss": 0.7862, "step": 30127 }, { "epoch": 3.2057884656309854, "grad_norm": 14.703125, "learning_rate": 8.947271176805004e-06, "loss": 0.5546, "step": 30128 }, { "epoch": 3.205894871249202, "grad_norm": 12.5, "learning_rate": 8.947235645252986e-06, "loss": 0.5785, "step": 30129 }, { "epoch": 3.2060012768674184, "grad_norm": 14.375, "learning_rate": 8.947200113700968e-06, "loss": 0.587, "step": 30130 }, { "epoch": 3.2061076824856354, "grad_norm": 17.390625, "learning_rate": 8.94716458214895e-06, "loss": 0.6299, "step": 30131 }, { "epoch": 3.206214088103852, "grad_norm": 15.2734375, "learning_rate": 8.94712905059693e-06, "loss": 0.5981, "step": 30132 }, { "epoch": 3.2063204937220684, "grad_norm": 16.625, "learning_rate": 8.947093519044912e-06, "loss": 0.7004, "step": 30133 }, { "epoch": 3.2064268993402854, "grad_norm": 16.28125, "learning_rate": 8.947057987492894e-06, "loss": 0.6069, "step": 30134 }, { "epoch": 3.206533304958502, "grad_norm": 10.1640625, "learning_rate": 8.947022455940876e-06, "loss": 0.6723, "step": 30135 }, { "epoch": 3.2066397105767184, "grad_norm": 15.75, "learning_rate": 8.946986924388859e-06, "loss": 0.7385, "step": 30136 }, { "epoch": 3.2067461161949353, "grad_norm": 11.171875, "learning_rate": 8.94695139283684e-06, "loss": 0.5859, "step": 30137 }, { "epoch": 3.206852521813152, "grad_norm": 10.5390625, "learning_rate": 8.946915861284822e-06, "loss": 0.7904, "step": 30138 }, { "epoch": 3.2069589274313683, "grad_norm": 13.03125, "learning_rate": 8.946880329732804e-06, "loss": 0.7305, "step": 30139 }, { "epoch": 3.207065333049585, "grad_norm": 13.140625, "learning_rate": 8.946844798180785e-06, "loss": 0.673, "step": 30140 }, { "epoch": 3.207171738667802, "grad_norm": 10.828125, "learning_rate": 8.946809266628767e-06, "loss": 0.5331, "step": 30141 }, { "epoch": 3.2072781442860183, "grad_norm": 16.90625, "learning_rate": 8.946773735076749e-06, "loss": 0.8539, "step": 30142 }, { "epoch": 3.207384549904235, "grad_norm": 16.671875, "learning_rate": 8.94673820352473e-06, "loss": 0.7067, "step": 30143 }, { "epoch": 3.207490955522452, "grad_norm": 11.03125, "learning_rate": 8.946702671972712e-06, "loss": 0.4813, "step": 30144 }, { "epoch": 3.2075973611406683, "grad_norm": 10.4765625, "learning_rate": 8.946667140420693e-06, "loss": 0.605, "step": 30145 }, { "epoch": 3.207703766758885, "grad_norm": 13.9296875, "learning_rate": 8.946631608868677e-06, "loss": 0.7068, "step": 30146 }, { "epoch": 3.2078101723771013, "grad_norm": 20.671875, "learning_rate": 8.946596077316658e-06, "loss": 0.7346, "step": 30147 }, { "epoch": 3.2079165779953183, "grad_norm": 10.8671875, "learning_rate": 8.94656054576464e-06, "loss": 0.5855, "step": 30148 }, { "epoch": 3.2080229836135348, "grad_norm": 11.8984375, "learning_rate": 8.946525014212621e-06, "loss": 0.6396, "step": 30149 }, { "epoch": 3.2081293892317513, "grad_norm": 12.8984375, "learning_rate": 8.946489482660603e-06, "loss": 0.6521, "step": 30150 }, { "epoch": 3.2082357948499682, "grad_norm": 15.2109375, "learning_rate": 8.946453951108585e-06, "loss": 0.64, "step": 30151 }, { "epoch": 3.2083422004681847, "grad_norm": 9.7265625, "learning_rate": 8.946418419556568e-06, "loss": 0.5333, "step": 30152 }, { "epoch": 3.2084486060864013, "grad_norm": 10.890625, "learning_rate": 8.94638288800455e-06, "loss": 0.5889, "step": 30153 }, { "epoch": 3.208555011704618, "grad_norm": 13.7578125, "learning_rate": 8.946347356452531e-06, "loss": 0.6092, "step": 30154 }, { "epoch": 3.2086614173228347, "grad_norm": 10.8515625, "learning_rate": 8.946311824900513e-06, "loss": 0.5224, "step": 30155 }, { "epoch": 3.2087678229410512, "grad_norm": 12.53125, "learning_rate": 8.946276293348494e-06, "loss": 0.6578, "step": 30156 }, { "epoch": 3.2088742285592677, "grad_norm": 40.0625, "learning_rate": 8.946240761796476e-06, "loss": 0.9641, "step": 30157 }, { "epoch": 3.2089806341774847, "grad_norm": 14.578125, "learning_rate": 8.946205230244458e-06, "loss": 0.607, "step": 30158 }, { "epoch": 3.209087039795701, "grad_norm": 11.0859375, "learning_rate": 8.94616969869244e-06, "loss": 0.5562, "step": 30159 }, { "epoch": 3.2091934454139177, "grad_norm": 12.46875, "learning_rate": 8.946134167140421e-06, "loss": 0.6901, "step": 30160 }, { "epoch": 3.2092998510321347, "grad_norm": 12.828125, "learning_rate": 8.946098635588402e-06, "loss": 0.6316, "step": 30161 }, { "epoch": 3.209406256650351, "grad_norm": 11.5859375, "learning_rate": 8.946063104036386e-06, "loss": 0.719, "step": 30162 }, { "epoch": 3.2095126622685677, "grad_norm": 14.3359375, "learning_rate": 8.946027572484367e-06, "loss": 0.5417, "step": 30163 }, { "epoch": 3.209619067886784, "grad_norm": 14.3828125, "learning_rate": 8.945992040932349e-06, "loss": 0.7404, "step": 30164 }, { "epoch": 3.209725473505001, "grad_norm": 13.5625, "learning_rate": 8.94595650938033e-06, "loss": 0.7722, "step": 30165 }, { "epoch": 3.2098318791232177, "grad_norm": 14.9140625, "learning_rate": 8.945920977828312e-06, "loss": 0.7183, "step": 30166 }, { "epoch": 3.209938284741434, "grad_norm": 14.1796875, "learning_rate": 8.945885446276294e-06, "loss": 0.5724, "step": 30167 }, { "epoch": 3.210044690359651, "grad_norm": 9.875, "learning_rate": 8.945849914724275e-06, "loss": 0.5127, "step": 30168 }, { "epoch": 3.2101510959778676, "grad_norm": 12.234375, "learning_rate": 8.945814383172257e-06, "loss": 0.6623, "step": 30169 }, { "epoch": 3.210257501596084, "grad_norm": 15.359375, "learning_rate": 8.945778851620239e-06, "loss": 0.5315, "step": 30170 }, { "epoch": 3.210363907214301, "grad_norm": 16.703125, "learning_rate": 8.94574332006822e-06, "loss": 0.7201, "step": 30171 }, { "epoch": 3.2104703128325176, "grad_norm": 12.5, "learning_rate": 8.945707788516204e-06, "loss": 0.5134, "step": 30172 }, { "epoch": 3.210576718450734, "grad_norm": 18.921875, "learning_rate": 8.945672256964185e-06, "loss": 0.7199, "step": 30173 }, { "epoch": 3.210683124068951, "grad_norm": 12.2109375, "learning_rate": 8.945636725412167e-06, "loss": 0.6978, "step": 30174 }, { "epoch": 3.2107895296871676, "grad_norm": 10.25, "learning_rate": 8.945601193860148e-06, "loss": 0.5506, "step": 30175 }, { "epoch": 3.210895935305384, "grad_norm": 15.15625, "learning_rate": 8.94556566230813e-06, "loss": 0.6075, "step": 30176 }, { "epoch": 3.2110023409236006, "grad_norm": 12.5, "learning_rate": 8.945530130756113e-06, "loss": 0.6809, "step": 30177 }, { "epoch": 3.2111087465418175, "grad_norm": 10.390625, "learning_rate": 8.945494599204095e-06, "loss": 0.543, "step": 30178 }, { "epoch": 3.211215152160034, "grad_norm": 11.7265625, "learning_rate": 8.945459067652076e-06, "loss": 0.5895, "step": 30179 }, { "epoch": 3.2113215577782506, "grad_norm": 13.5078125, "learning_rate": 8.945423536100058e-06, "loss": 0.6639, "step": 30180 }, { "epoch": 3.2114279633964675, "grad_norm": 12.7421875, "learning_rate": 8.94538800454804e-06, "loss": 0.6368, "step": 30181 }, { "epoch": 3.211534369014684, "grad_norm": 14.140625, "learning_rate": 8.945352472996021e-06, "loss": 0.5914, "step": 30182 }, { "epoch": 3.2116407746329005, "grad_norm": 12.0078125, "learning_rate": 8.945316941444003e-06, "loss": 0.5574, "step": 30183 }, { "epoch": 3.211747180251117, "grad_norm": 7.45703125, "learning_rate": 8.945281409891984e-06, "loss": 0.4096, "step": 30184 }, { "epoch": 3.211853585869334, "grad_norm": 13.4453125, "learning_rate": 8.945245878339966e-06, "loss": 0.7067, "step": 30185 }, { "epoch": 3.2119599914875505, "grad_norm": 11.875, "learning_rate": 8.945210346787948e-06, "loss": 0.6868, "step": 30186 }, { "epoch": 3.212066397105767, "grad_norm": 15.8671875, "learning_rate": 8.945174815235931e-06, "loss": 0.874, "step": 30187 }, { "epoch": 3.212172802723984, "grad_norm": 19.578125, "learning_rate": 8.945139283683913e-06, "loss": 0.8049, "step": 30188 }, { "epoch": 3.2122792083422005, "grad_norm": 17.390625, "learning_rate": 8.945103752131894e-06, "loss": 0.687, "step": 30189 }, { "epoch": 3.212385613960417, "grad_norm": 13.7890625, "learning_rate": 8.945068220579876e-06, "loss": 0.7347, "step": 30190 }, { "epoch": 3.212492019578634, "grad_norm": 12.5859375, "learning_rate": 8.945032689027857e-06, "loss": 0.4307, "step": 30191 }, { "epoch": 3.2125984251968505, "grad_norm": 12.03125, "learning_rate": 8.944997157475839e-06, "loss": 0.5126, "step": 30192 }, { "epoch": 3.212704830815067, "grad_norm": 17.71875, "learning_rate": 8.94496162592382e-06, "loss": 0.6604, "step": 30193 }, { "epoch": 3.2128112364332835, "grad_norm": 11.8359375, "learning_rate": 8.944926094371802e-06, "loss": 0.5141, "step": 30194 }, { "epoch": 3.2129176420515004, "grad_norm": 11.65625, "learning_rate": 8.944890562819784e-06, "loss": 0.5862, "step": 30195 }, { "epoch": 3.213024047669717, "grad_norm": 13.0390625, "learning_rate": 8.944855031267765e-06, "loss": 0.578, "step": 30196 }, { "epoch": 3.2131304532879335, "grad_norm": 10.4765625, "learning_rate": 8.944819499715747e-06, "loss": 0.6282, "step": 30197 }, { "epoch": 3.2132368589061504, "grad_norm": 13.640625, "learning_rate": 8.94478396816373e-06, "loss": 0.6961, "step": 30198 }, { "epoch": 3.213343264524367, "grad_norm": 13.828125, "learning_rate": 8.944748436611712e-06, "loss": 0.6269, "step": 30199 }, { "epoch": 3.2134496701425834, "grad_norm": 10.9140625, "learning_rate": 8.944712905059694e-06, "loss": 0.5082, "step": 30200 }, { "epoch": 3.2135560757608, "grad_norm": 13.6171875, "learning_rate": 8.944677373507675e-06, "loss": 0.5824, "step": 30201 }, { "epoch": 3.213662481379017, "grad_norm": 17.125, "learning_rate": 8.944641841955657e-06, "loss": 0.5472, "step": 30202 }, { "epoch": 3.2137688869972334, "grad_norm": 13.28125, "learning_rate": 8.94460631040364e-06, "loss": 0.6211, "step": 30203 }, { "epoch": 3.21387529261545, "grad_norm": 13.4140625, "learning_rate": 8.944570778851622e-06, "loss": 0.6093, "step": 30204 }, { "epoch": 3.213981698233667, "grad_norm": 10.9609375, "learning_rate": 8.944535247299603e-06, "loss": 0.6841, "step": 30205 }, { "epoch": 3.2140881038518834, "grad_norm": 13.671875, "learning_rate": 8.944499715747585e-06, "loss": 0.6775, "step": 30206 }, { "epoch": 3.2141945094701, "grad_norm": 11.0546875, "learning_rate": 8.944464184195567e-06, "loss": 0.5765, "step": 30207 }, { "epoch": 3.214300915088317, "grad_norm": 12.1640625, "learning_rate": 8.944428652643548e-06, "loss": 0.6581, "step": 30208 }, { "epoch": 3.2144073207065333, "grad_norm": 11.3125, "learning_rate": 8.94439312109153e-06, "loss": 0.566, "step": 30209 }, { "epoch": 3.21451372632475, "grad_norm": 12.484375, "learning_rate": 8.944357589539511e-06, "loss": 0.5388, "step": 30210 }, { "epoch": 3.214620131942967, "grad_norm": 12.3046875, "learning_rate": 8.944322057987493e-06, "loss": 0.4757, "step": 30211 }, { "epoch": 3.2147265375611833, "grad_norm": 13.9296875, "learning_rate": 8.944286526435475e-06, "loss": 0.6274, "step": 30212 }, { "epoch": 3.2148329431794, "grad_norm": 16.578125, "learning_rate": 8.944250994883458e-06, "loss": 0.676, "step": 30213 }, { "epoch": 3.2149393487976163, "grad_norm": 12.546875, "learning_rate": 8.94421546333144e-06, "loss": 0.5189, "step": 30214 }, { "epoch": 3.2150457544158333, "grad_norm": 13.8984375, "learning_rate": 8.944179931779421e-06, "loss": 0.7151, "step": 30215 }, { "epoch": 3.21515216003405, "grad_norm": 11.3515625, "learning_rate": 8.944144400227403e-06, "loss": 0.6329, "step": 30216 }, { "epoch": 3.2152585656522663, "grad_norm": 14.6484375, "learning_rate": 8.944108868675384e-06, "loss": 0.7803, "step": 30217 }, { "epoch": 3.2153649712704833, "grad_norm": 15.3046875, "learning_rate": 8.944073337123368e-06, "loss": 0.7739, "step": 30218 }, { "epoch": 3.2154713768886998, "grad_norm": 12.859375, "learning_rate": 8.94403780557135e-06, "loss": 0.7148, "step": 30219 }, { "epoch": 3.2155777825069163, "grad_norm": 13.609375, "learning_rate": 8.94400227401933e-06, "loss": 0.7337, "step": 30220 }, { "epoch": 3.215684188125133, "grad_norm": 11.7890625, "learning_rate": 8.94396674246731e-06, "loss": 0.6328, "step": 30221 }, { "epoch": 3.2157905937433497, "grad_norm": 10.4375, "learning_rate": 8.943931210915292e-06, "loss": 0.4788, "step": 30222 }, { "epoch": 3.2158969993615663, "grad_norm": 13.859375, "learning_rate": 8.943895679363276e-06, "loss": 0.6397, "step": 30223 }, { "epoch": 3.2160034049797828, "grad_norm": 10.8125, "learning_rate": 8.943860147811257e-06, "loss": 0.6027, "step": 30224 }, { "epoch": 3.2161098105979997, "grad_norm": 14.3828125, "learning_rate": 8.943824616259239e-06, "loss": 0.7286, "step": 30225 }, { "epoch": 3.2162162162162162, "grad_norm": 12.671875, "learning_rate": 8.94378908470722e-06, "loss": 0.6751, "step": 30226 }, { "epoch": 3.2163226218344327, "grad_norm": 10.140625, "learning_rate": 8.943753553155202e-06, "loss": 0.4359, "step": 30227 }, { "epoch": 3.2164290274526497, "grad_norm": 13.1484375, "learning_rate": 8.943718021603184e-06, "loss": 0.5823, "step": 30228 }, { "epoch": 3.216535433070866, "grad_norm": 12.3125, "learning_rate": 8.943682490051167e-06, "loss": 0.7156, "step": 30229 }, { "epoch": 3.2166418386890827, "grad_norm": 14.09375, "learning_rate": 8.943646958499149e-06, "loss": 0.5175, "step": 30230 }, { "epoch": 3.216748244307299, "grad_norm": 11.8046875, "learning_rate": 8.94361142694713e-06, "loss": 0.694, "step": 30231 }, { "epoch": 3.216854649925516, "grad_norm": 11.28125, "learning_rate": 8.943575895395112e-06, "loss": 0.6393, "step": 30232 }, { "epoch": 3.2169610555437327, "grad_norm": 17.171875, "learning_rate": 8.943540363843093e-06, "loss": 0.8156, "step": 30233 }, { "epoch": 3.217067461161949, "grad_norm": 11.7734375, "learning_rate": 8.943504832291075e-06, "loss": 0.5672, "step": 30234 }, { "epoch": 3.217173866780166, "grad_norm": 11.2578125, "learning_rate": 8.943469300739057e-06, "loss": 0.5854, "step": 30235 }, { "epoch": 3.2172802723983827, "grad_norm": 14.7265625, "learning_rate": 8.943433769187038e-06, "loss": 0.7075, "step": 30236 }, { "epoch": 3.217386678016599, "grad_norm": 16.59375, "learning_rate": 8.94339823763502e-06, "loss": 0.8037, "step": 30237 }, { "epoch": 3.217493083634816, "grad_norm": 14.8203125, "learning_rate": 8.943362706083001e-06, "loss": 0.6515, "step": 30238 }, { "epoch": 3.2175994892530326, "grad_norm": 12.5390625, "learning_rate": 8.943327174530985e-06, "loss": 0.576, "step": 30239 }, { "epoch": 3.217705894871249, "grad_norm": 11.7109375, "learning_rate": 8.943291642978966e-06, "loss": 0.5562, "step": 30240 }, { "epoch": 3.2178123004894656, "grad_norm": 10.109375, "learning_rate": 8.943256111426948e-06, "loss": 0.5379, "step": 30241 }, { "epoch": 3.2179187061076826, "grad_norm": 10.9140625, "learning_rate": 8.94322057987493e-06, "loss": 0.4691, "step": 30242 }, { "epoch": 3.218025111725899, "grad_norm": 11.6328125, "learning_rate": 8.943185048322911e-06, "loss": 0.5256, "step": 30243 }, { "epoch": 3.2181315173441156, "grad_norm": 18.78125, "learning_rate": 8.943149516770894e-06, "loss": 0.8798, "step": 30244 }, { "epoch": 3.2182379229623326, "grad_norm": 12.1953125, "learning_rate": 8.943113985218876e-06, "loss": 0.6754, "step": 30245 }, { "epoch": 3.218344328580549, "grad_norm": 12.8359375, "learning_rate": 8.943078453666858e-06, "loss": 0.7057, "step": 30246 }, { "epoch": 3.2184507341987656, "grad_norm": 12.21875, "learning_rate": 8.94304292211484e-06, "loss": 0.7149, "step": 30247 }, { "epoch": 3.2185571398169825, "grad_norm": 13.59375, "learning_rate": 8.943007390562821e-06, "loss": 0.5715, "step": 30248 }, { "epoch": 3.218663545435199, "grad_norm": 11.90625, "learning_rate": 8.942971859010802e-06, "loss": 0.5812, "step": 30249 }, { "epoch": 3.2187699510534156, "grad_norm": 12.7421875, "learning_rate": 8.942936327458784e-06, "loss": 0.7235, "step": 30250 }, { "epoch": 3.218876356671632, "grad_norm": 14.875, "learning_rate": 8.942900795906766e-06, "loss": 0.8449, "step": 30251 }, { "epoch": 3.218982762289849, "grad_norm": 14.0234375, "learning_rate": 8.942865264354747e-06, "loss": 0.6897, "step": 30252 }, { "epoch": 3.2190891679080655, "grad_norm": 13.5390625, "learning_rate": 8.942829732802729e-06, "loss": 0.5776, "step": 30253 }, { "epoch": 3.219195573526282, "grad_norm": 13.7734375, "learning_rate": 8.942794201250712e-06, "loss": 0.6489, "step": 30254 }, { "epoch": 3.219301979144499, "grad_norm": 13.03125, "learning_rate": 8.942758669698694e-06, "loss": 0.6301, "step": 30255 }, { "epoch": 3.2194083847627155, "grad_norm": 13.9296875, "learning_rate": 8.942723138146675e-06, "loss": 0.5868, "step": 30256 }, { "epoch": 3.219514790380932, "grad_norm": 14.3046875, "learning_rate": 8.942687606594657e-06, "loss": 0.7531, "step": 30257 }, { "epoch": 3.2196211959991485, "grad_norm": 12.703125, "learning_rate": 8.942652075042639e-06, "loss": 0.5597, "step": 30258 }, { "epoch": 3.2197276016173655, "grad_norm": 15.5859375, "learning_rate": 8.94261654349062e-06, "loss": 0.6525, "step": 30259 }, { "epoch": 3.219834007235582, "grad_norm": 14.1484375, "learning_rate": 8.942581011938602e-06, "loss": 0.7022, "step": 30260 }, { "epoch": 3.2199404128537985, "grad_norm": 11.765625, "learning_rate": 8.942545480386583e-06, "loss": 0.6819, "step": 30261 }, { "epoch": 3.2200468184720155, "grad_norm": 15.03125, "learning_rate": 8.942509948834565e-06, "loss": 0.6872, "step": 30262 }, { "epoch": 3.220153224090232, "grad_norm": 12.4765625, "learning_rate": 8.942474417282547e-06, "loss": 0.5957, "step": 30263 }, { "epoch": 3.2202596297084485, "grad_norm": 12.09375, "learning_rate": 8.94243888573053e-06, "loss": 0.6012, "step": 30264 }, { "epoch": 3.2203660353266654, "grad_norm": 14.875, "learning_rate": 8.942403354178512e-06, "loss": 0.7039, "step": 30265 }, { "epoch": 3.220472440944882, "grad_norm": 16.5, "learning_rate": 8.942367822626493e-06, "loss": 0.6881, "step": 30266 }, { "epoch": 3.2205788465630985, "grad_norm": 10.5859375, "learning_rate": 8.942332291074475e-06, "loss": 0.6049, "step": 30267 }, { "epoch": 3.2206852521813154, "grad_norm": 16.03125, "learning_rate": 8.942296759522456e-06, "loss": 0.6072, "step": 30268 }, { "epoch": 3.220791657799532, "grad_norm": 10.484375, "learning_rate": 8.942261227970438e-06, "loss": 0.6536, "step": 30269 }, { "epoch": 3.2208980634177484, "grad_norm": 14.65625, "learning_rate": 8.942225696418421e-06, "loss": 0.6609, "step": 30270 }, { "epoch": 3.221004469035965, "grad_norm": 12.875, "learning_rate": 8.942190164866403e-06, "loss": 0.623, "step": 30271 }, { "epoch": 3.221110874654182, "grad_norm": 16.328125, "learning_rate": 8.942154633314385e-06, "loss": 0.818, "step": 30272 }, { "epoch": 3.2212172802723984, "grad_norm": 13.4765625, "learning_rate": 8.942119101762366e-06, "loss": 0.6401, "step": 30273 }, { "epoch": 3.221323685890615, "grad_norm": 11.5234375, "learning_rate": 8.942083570210348e-06, "loss": 0.4675, "step": 30274 }, { "epoch": 3.221430091508832, "grad_norm": 10.1796875, "learning_rate": 8.94204803865833e-06, "loss": 0.5977, "step": 30275 }, { "epoch": 3.2215364971270484, "grad_norm": 12.3359375, "learning_rate": 8.942012507106311e-06, "loss": 0.5178, "step": 30276 }, { "epoch": 3.221642902745265, "grad_norm": 13.40625, "learning_rate": 8.941976975554293e-06, "loss": 0.5271, "step": 30277 }, { "epoch": 3.2217493083634814, "grad_norm": 11.4140625, "learning_rate": 8.941941444002274e-06, "loss": 0.6648, "step": 30278 }, { "epoch": 3.2218557139816983, "grad_norm": 15.8046875, "learning_rate": 8.941905912450256e-06, "loss": 0.7154, "step": 30279 }, { "epoch": 3.221962119599915, "grad_norm": 14.9765625, "learning_rate": 8.941870380898239e-06, "loss": 0.6579, "step": 30280 }, { "epoch": 3.2220685252181314, "grad_norm": 12.0859375, "learning_rate": 8.94183484934622e-06, "loss": 0.7133, "step": 30281 }, { "epoch": 3.2221749308363483, "grad_norm": 14.7890625, "learning_rate": 8.941799317794202e-06, "loss": 0.6122, "step": 30282 }, { "epoch": 3.222281336454565, "grad_norm": 16.21875, "learning_rate": 8.941763786242184e-06, "loss": 0.6998, "step": 30283 }, { "epoch": 3.2223877420727813, "grad_norm": 12.671875, "learning_rate": 8.941728254690165e-06, "loss": 0.5432, "step": 30284 }, { "epoch": 3.2224941476909983, "grad_norm": 12.40625, "learning_rate": 8.941692723138147e-06, "loss": 0.5671, "step": 30285 }, { "epoch": 3.222600553309215, "grad_norm": 13.40625, "learning_rate": 8.941657191586129e-06, "loss": 0.6349, "step": 30286 }, { "epoch": 3.2227069589274313, "grad_norm": 14.4453125, "learning_rate": 8.94162166003411e-06, "loss": 0.7422, "step": 30287 }, { "epoch": 3.222813364545648, "grad_norm": 16.3125, "learning_rate": 8.941586128482092e-06, "loss": 0.6759, "step": 30288 }, { "epoch": 3.2229197701638648, "grad_norm": 12.046875, "learning_rate": 8.941550596930074e-06, "loss": 0.5436, "step": 30289 }, { "epoch": 3.2230261757820813, "grad_norm": 13.203125, "learning_rate": 8.941515065378057e-06, "loss": 0.6587, "step": 30290 }, { "epoch": 3.223132581400298, "grad_norm": 12.4609375, "learning_rate": 8.941479533826038e-06, "loss": 0.5796, "step": 30291 }, { "epoch": 3.2232389870185147, "grad_norm": 15.65625, "learning_rate": 8.94144400227402e-06, "loss": 0.6604, "step": 30292 }, { "epoch": 3.2233453926367313, "grad_norm": 11.3046875, "learning_rate": 8.941408470722002e-06, "loss": 0.6539, "step": 30293 }, { "epoch": 3.2234517982549478, "grad_norm": 11.4453125, "learning_rate": 8.941372939169983e-06, "loss": 0.6475, "step": 30294 }, { "epoch": 3.2235582038731643, "grad_norm": 13.21875, "learning_rate": 8.941337407617967e-06, "loss": 0.5652, "step": 30295 }, { "epoch": 3.2236646094913812, "grad_norm": 16.5, "learning_rate": 8.941301876065948e-06, "loss": 0.7989, "step": 30296 }, { "epoch": 3.2237710151095977, "grad_norm": 12.1171875, "learning_rate": 8.94126634451393e-06, "loss": 0.5621, "step": 30297 }, { "epoch": 3.2238774207278142, "grad_norm": 16.390625, "learning_rate": 8.941230812961911e-06, "loss": 0.6042, "step": 30298 }, { "epoch": 3.223983826346031, "grad_norm": 12.6328125, "learning_rate": 8.941195281409893e-06, "loss": 0.5952, "step": 30299 }, { "epoch": 3.2240902319642477, "grad_norm": 11.8671875, "learning_rate": 8.941159749857875e-06, "loss": 0.515, "step": 30300 }, { "epoch": 3.224196637582464, "grad_norm": 14.0625, "learning_rate": 8.941124218305856e-06, "loss": 0.6426, "step": 30301 }, { "epoch": 3.224303043200681, "grad_norm": 13.421875, "learning_rate": 8.941088686753838e-06, "loss": 0.6016, "step": 30302 }, { "epoch": 3.2244094488188977, "grad_norm": 11.78125, "learning_rate": 8.94105315520182e-06, "loss": 0.5414, "step": 30303 }, { "epoch": 3.224515854437114, "grad_norm": 15.5234375, "learning_rate": 8.941017623649801e-06, "loss": 0.5708, "step": 30304 }, { "epoch": 3.224622260055331, "grad_norm": 11.7265625, "learning_rate": 8.940982092097784e-06, "loss": 0.6969, "step": 30305 }, { "epoch": 3.2247286656735477, "grad_norm": 12.109375, "learning_rate": 8.940946560545766e-06, "loss": 0.6144, "step": 30306 }, { "epoch": 3.224835071291764, "grad_norm": 14.4296875, "learning_rate": 8.940911028993748e-06, "loss": 0.5427, "step": 30307 }, { "epoch": 3.2249414769099807, "grad_norm": 14.1484375, "learning_rate": 8.940875497441729e-06, "loss": 0.8662, "step": 30308 }, { "epoch": 3.2250478825281976, "grad_norm": 17.578125, "learning_rate": 8.94083996588971e-06, "loss": 0.9414, "step": 30309 }, { "epoch": 3.225154288146414, "grad_norm": 11.984375, "learning_rate": 8.940804434337692e-06, "loss": 0.5192, "step": 30310 }, { "epoch": 3.2252606937646306, "grad_norm": 14.296875, "learning_rate": 8.940768902785674e-06, "loss": 0.7286, "step": 30311 }, { "epoch": 3.2253670993828476, "grad_norm": 9.0859375, "learning_rate": 8.940733371233656e-06, "loss": 0.6558, "step": 30312 }, { "epoch": 3.225473505001064, "grad_norm": 14.4921875, "learning_rate": 8.940697839681637e-06, "loss": 0.6823, "step": 30313 }, { "epoch": 3.2255799106192806, "grad_norm": 11.984375, "learning_rate": 8.940662308129619e-06, "loss": 0.6258, "step": 30314 }, { "epoch": 3.225686316237497, "grad_norm": 11.484375, "learning_rate": 8.9406267765776e-06, "loss": 0.6169, "step": 30315 }, { "epoch": 3.225792721855714, "grad_norm": 13.4140625, "learning_rate": 8.940591245025584e-06, "loss": 0.6268, "step": 30316 }, { "epoch": 3.2258991274739306, "grad_norm": 11.4375, "learning_rate": 8.940555713473565e-06, "loss": 0.7095, "step": 30317 }, { "epoch": 3.226005533092147, "grad_norm": 14.8046875, "learning_rate": 8.940520181921547e-06, "loss": 0.5942, "step": 30318 }, { "epoch": 3.226111938710364, "grad_norm": 11.15625, "learning_rate": 8.940484650369529e-06, "loss": 0.6095, "step": 30319 }, { "epoch": 3.2262183443285806, "grad_norm": 10.3203125, "learning_rate": 8.94044911881751e-06, "loss": 0.571, "step": 30320 }, { "epoch": 3.226324749946797, "grad_norm": 11.671875, "learning_rate": 8.940413587265493e-06, "loss": 0.5314, "step": 30321 }, { "epoch": 3.226431155565014, "grad_norm": 11.7890625, "learning_rate": 8.940378055713475e-06, "loss": 0.6161, "step": 30322 }, { "epoch": 3.2265375611832305, "grad_norm": 12.4453125, "learning_rate": 8.940342524161457e-06, "loss": 0.6561, "step": 30323 }, { "epoch": 3.226643966801447, "grad_norm": 13.25, "learning_rate": 8.940306992609438e-06, "loss": 0.7169, "step": 30324 }, { "epoch": 3.2267503724196636, "grad_norm": 12.1015625, "learning_rate": 8.94027146105742e-06, "loss": 0.6884, "step": 30325 }, { "epoch": 3.2268567780378805, "grad_norm": 14.703125, "learning_rate": 8.940235929505401e-06, "loss": 0.6038, "step": 30326 }, { "epoch": 3.226963183656097, "grad_norm": 14.359375, "learning_rate": 8.940200397953383e-06, "loss": 0.7436, "step": 30327 }, { "epoch": 3.2270695892743135, "grad_norm": 10.5234375, "learning_rate": 8.940164866401365e-06, "loss": 0.5958, "step": 30328 }, { "epoch": 3.2271759948925305, "grad_norm": 12.71875, "learning_rate": 8.940129334849346e-06, "loss": 0.6236, "step": 30329 }, { "epoch": 3.227282400510747, "grad_norm": 17.390625, "learning_rate": 8.940093803297328e-06, "loss": 0.7301, "step": 30330 }, { "epoch": 3.2273888061289635, "grad_norm": 11.2265625, "learning_rate": 8.940058271745311e-06, "loss": 0.5549, "step": 30331 }, { "epoch": 3.2274952117471805, "grad_norm": 13.765625, "learning_rate": 8.940022740193293e-06, "loss": 0.6861, "step": 30332 }, { "epoch": 3.227601617365397, "grad_norm": 12.1328125, "learning_rate": 8.939987208641274e-06, "loss": 0.6247, "step": 30333 }, { "epoch": 3.2277080229836135, "grad_norm": 12.0625, "learning_rate": 8.939951677089256e-06, "loss": 0.6207, "step": 30334 }, { "epoch": 3.22781442860183, "grad_norm": 12.96875, "learning_rate": 8.939916145537238e-06, "loss": 0.6066, "step": 30335 }, { "epoch": 3.227920834220047, "grad_norm": 12.390625, "learning_rate": 8.939880613985221e-06, "loss": 0.6619, "step": 30336 }, { "epoch": 3.2280272398382635, "grad_norm": 11.53125, "learning_rate": 8.939845082433203e-06, "loss": 0.6042, "step": 30337 }, { "epoch": 3.22813364545648, "grad_norm": 13.4375, "learning_rate": 8.939809550881184e-06, "loss": 0.6811, "step": 30338 }, { "epoch": 3.228240051074697, "grad_norm": 12.21875, "learning_rate": 8.939774019329166e-06, "loss": 0.6003, "step": 30339 }, { "epoch": 3.2283464566929134, "grad_norm": 11.96875, "learning_rate": 8.939738487777147e-06, "loss": 0.6477, "step": 30340 }, { "epoch": 3.22845286231113, "grad_norm": 13.90625, "learning_rate": 8.939702956225129e-06, "loss": 0.6882, "step": 30341 }, { "epoch": 3.228559267929347, "grad_norm": 13.8984375, "learning_rate": 8.93966742467311e-06, "loss": 0.7249, "step": 30342 }, { "epoch": 3.2286656735475634, "grad_norm": 30.4375, "learning_rate": 8.939631893121092e-06, "loss": 0.6846, "step": 30343 }, { "epoch": 3.22877207916578, "grad_norm": 11.734375, "learning_rate": 8.939596361569074e-06, "loss": 0.6711, "step": 30344 }, { "epoch": 3.2288784847839964, "grad_norm": 11.9765625, "learning_rate": 8.939560830017055e-06, "loss": 0.5318, "step": 30345 }, { "epoch": 3.2289848904022134, "grad_norm": 13.921875, "learning_rate": 8.939525298465037e-06, "loss": 0.7083, "step": 30346 }, { "epoch": 3.22909129602043, "grad_norm": 11.640625, "learning_rate": 8.93948976691302e-06, "loss": 0.5627, "step": 30347 }, { "epoch": 3.2291977016386464, "grad_norm": 13.0234375, "learning_rate": 8.939454235361002e-06, "loss": 0.7198, "step": 30348 }, { "epoch": 3.2293041072568633, "grad_norm": 13.328125, "learning_rate": 8.939418703808983e-06, "loss": 0.6939, "step": 30349 }, { "epoch": 3.22941051287508, "grad_norm": 17.3125, "learning_rate": 8.939383172256965e-06, "loss": 0.6708, "step": 30350 }, { "epoch": 3.2295169184932964, "grad_norm": 10.1171875, "learning_rate": 8.939347640704947e-06, "loss": 0.5061, "step": 30351 }, { "epoch": 3.229623324111513, "grad_norm": 13.3984375, "learning_rate": 8.939312109152928e-06, "loss": 0.6053, "step": 30352 }, { "epoch": 3.22972972972973, "grad_norm": 17.71875, "learning_rate": 8.93927657760091e-06, "loss": 0.719, "step": 30353 }, { "epoch": 3.2298361353479463, "grad_norm": 13.3828125, "learning_rate": 8.939241046048892e-06, "loss": 0.6391, "step": 30354 }, { "epoch": 3.229942540966163, "grad_norm": 15.09375, "learning_rate": 8.939205514496873e-06, "loss": 0.636, "step": 30355 }, { "epoch": 3.23004894658438, "grad_norm": 12.2734375, "learning_rate": 8.939169982944855e-06, "loss": 0.6141, "step": 30356 }, { "epoch": 3.2301553522025963, "grad_norm": 15.2421875, "learning_rate": 8.939134451392838e-06, "loss": 0.5771, "step": 30357 }, { "epoch": 3.230261757820813, "grad_norm": 13.8046875, "learning_rate": 8.93909891984082e-06, "loss": 0.7204, "step": 30358 }, { "epoch": 3.2303681634390298, "grad_norm": 12.4296875, "learning_rate": 8.939063388288801e-06, "loss": 0.509, "step": 30359 }, { "epoch": 3.2304745690572463, "grad_norm": 15.9375, "learning_rate": 8.939027856736783e-06, "loss": 0.6738, "step": 30360 }, { "epoch": 3.230580974675463, "grad_norm": 15.0859375, "learning_rate": 8.938992325184764e-06, "loss": 0.6727, "step": 30361 }, { "epoch": 3.2306873802936793, "grad_norm": 15.9765625, "learning_rate": 8.938956793632748e-06, "loss": 0.7097, "step": 30362 }, { "epoch": 3.2307937859118963, "grad_norm": 17.828125, "learning_rate": 8.93892126208073e-06, "loss": 0.7484, "step": 30363 }, { "epoch": 3.2309001915301128, "grad_norm": 11.9765625, "learning_rate": 8.938885730528711e-06, "loss": 0.5995, "step": 30364 }, { "epoch": 3.2310065971483293, "grad_norm": 11.1796875, "learning_rate": 8.938850198976693e-06, "loss": 0.5171, "step": 30365 }, { "epoch": 3.2311130027665462, "grad_norm": 10.8125, "learning_rate": 8.938814667424674e-06, "loss": 0.5732, "step": 30366 }, { "epoch": 3.2312194083847627, "grad_norm": 14.7578125, "learning_rate": 8.938779135872656e-06, "loss": 0.6027, "step": 30367 }, { "epoch": 3.2313258140029792, "grad_norm": 10.4375, "learning_rate": 8.938743604320637e-06, "loss": 0.4217, "step": 30368 }, { "epoch": 3.231432219621196, "grad_norm": 13.0078125, "learning_rate": 8.938708072768619e-06, "loss": 0.6029, "step": 30369 }, { "epoch": 3.2315386252394127, "grad_norm": 10.4765625, "learning_rate": 8.9386725412166e-06, "loss": 0.5217, "step": 30370 }, { "epoch": 3.231645030857629, "grad_norm": 12.796875, "learning_rate": 8.938637009664582e-06, "loss": 0.6709, "step": 30371 }, { "epoch": 3.2317514364758457, "grad_norm": 15.1328125, "learning_rate": 8.938601478112566e-06, "loss": 0.6077, "step": 30372 }, { "epoch": 3.2318578420940627, "grad_norm": 16.921875, "learning_rate": 8.938565946560547e-06, "loss": 0.7288, "step": 30373 }, { "epoch": 3.231964247712279, "grad_norm": 13.65625, "learning_rate": 8.938530415008529e-06, "loss": 0.7873, "step": 30374 }, { "epoch": 3.2320706533304957, "grad_norm": 13.484375, "learning_rate": 8.93849488345651e-06, "loss": 0.552, "step": 30375 }, { "epoch": 3.2321770589487127, "grad_norm": 11.9921875, "learning_rate": 8.938459351904492e-06, "loss": 0.7136, "step": 30376 }, { "epoch": 3.232283464566929, "grad_norm": 14.28125, "learning_rate": 8.938423820352474e-06, "loss": 0.6847, "step": 30377 }, { "epoch": 3.2323898701851457, "grad_norm": 12.375, "learning_rate": 8.938388288800455e-06, "loss": 0.7405, "step": 30378 }, { "epoch": 3.2324962758033626, "grad_norm": 14.8125, "learning_rate": 8.938352757248437e-06, "loss": 0.9088, "step": 30379 }, { "epoch": 3.232602681421579, "grad_norm": 12.078125, "learning_rate": 8.938317225696418e-06, "loss": 0.5256, "step": 30380 }, { "epoch": 3.2327090870397956, "grad_norm": 12.546875, "learning_rate": 8.9382816941444e-06, "loss": 0.6178, "step": 30381 }, { "epoch": 3.232815492658012, "grad_norm": 11.6328125, "learning_rate": 8.938246162592383e-06, "loss": 0.5882, "step": 30382 }, { "epoch": 3.232921898276229, "grad_norm": 11.8046875, "learning_rate": 8.938210631040365e-06, "loss": 0.5679, "step": 30383 }, { "epoch": 3.2330283038944456, "grad_norm": 16.453125, "learning_rate": 8.938175099488346e-06, "loss": 0.7526, "step": 30384 }, { "epoch": 3.233134709512662, "grad_norm": 11.2421875, "learning_rate": 8.938139567936328e-06, "loss": 0.5382, "step": 30385 }, { "epoch": 3.233241115130879, "grad_norm": 10.9921875, "learning_rate": 8.93810403638431e-06, "loss": 0.5546, "step": 30386 }, { "epoch": 3.2333475207490956, "grad_norm": 12.109375, "learning_rate": 8.938068504832291e-06, "loss": 0.5449, "step": 30387 }, { "epoch": 3.233453926367312, "grad_norm": 11.8046875, "learning_rate": 8.938032973280275e-06, "loss": 0.586, "step": 30388 }, { "epoch": 3.2335603319855286, "grad_norm": 10.9921875, "learning_rate": 8.937997441728256e-06, "loss": 0.607, "step": 30389 }, { "epoch": 3.2336667376037456, "grad_norm": 11.3359375, "learning_rate": 8.937961910176238e-06, "loss": 0.5633, "step": 30390 }, { "epoch": 3.233773143221962, "grad_norm": 10.9296875, "learning_rate": 8.93792637862422e-06, "loss": 0.5953, "step": 30391 }, { "epoch": 3.2338795488401786, "grad_norm": 11.7109375, "learning_rate": 8.937890847072201e-06, "loss": 0.6198, "step": 30392 }, { "epoch": 3.2339859544583955, "grad_norm": 11.8359375, "learning_rate": 8.937855315520183e-06, "loss": 0.5542, "step": 30393 }, { "epoch": 3.234092360076612, "grad_norm": 8.109375, "learning_rate": 8.937819783968164e-06, "loss": 0.5743, "step": 30394 }, { "epoch": 3.2341987656948286, "grad_norm": 15.1640625, "learning_rate": 8.937784252416146e-06, "loss": 0.6831, "step": 30395 }, { "epoch": 3.2343051713130455, "grad_norm": 11.5078125, "learning_rate": 8.937748720864127e-06, "loss": 0.6614, "step": 30396 }, { "epoch": 3.234411576931262, "grad_norm": 17.328125, "learning_rate": 8.937713189312109e-06, "loss": 0.8, "step": 30397 }, { "epoch": 3.2345179825494785, "grad_norm": 18.0625, "learning_rate": 8.937677657760092e-06, "loss": 0.7771, "step": 30398 }, { "epoch": 3.2346243881676955, "grad_norm": 12.296875, "learning_rate": 8.937642126208074e-06, "loss": 0.457, "step": 30399 }, { "epoch": 3.234730793785912, "grad_norm": 11.765625, "learning_rate": 8.937606594656056e-06, "loss": 0.4829, "step": 30400 }, { "epoch": 3.2348371994041285, "grad_norm": 15.1796875, "learning_rate": 8.937571063104037e-06, "loss": 0.635, "step": 30401 }, { "epoch": 3.234943605022345, "grad_norm": 12.3203125, "learning_rate": 8.937535531552019e-06, "loss": 0.5581, "step": 30402 }, { "epoch": 3.235050010640562, "grad_norm": 10.1328125, "learning_rate": 8.9375e-06, "loss": 0.5457, "step": 30403 }, { "epoch": 3.2351564162587785, "grad_norm": 18.484375, "learning_rate": 8.937464468447982e-06, "loss": 0.7264, "step": 30404 }, { "epoch": 3.235262821876995, "grad_norm": 11.8203125, "learning_rate": 8.937428936895964e-06, "loss": 0.5017, "step": 30405 }, { "epoch": 3.235369227495212, "grad_norm": 10.5078125, "learning_rate": 8.937393405343945e-06, "loss": 0.5293, "step": 30406 }, { "epoch": 3.2354756331134285, "grad_norm": 14.0390625, "learning_rate": 8.937357873791927e-06, "loss": 0.6892, "step": 30407 }, { "epoch": 3.235582038731645, "grad_norm": 14.984375, "learning_rate": 8.93732234223991e-06, "loss": 0.6321, "step": 30408 }, { "epoch": 3.2356884443498615, "grad_norm": 10.703125, "learning_rate": 8.937286810687892e-06, "loss": 0.692, "step": 30409 }, { "epoch": 3.2357948499680784, "grad_norm": 11.53125, "learning_rate": 8.937251279135873e-06, "loss": 0.5188, "step": 30410 }, { "epoch": 3.235901255586295, "grad_norm": 14.46875, "learning_rate": 8.937215747583855e-06, "loss": 0.6203, "step": 30411 }, { "epoch": 3.2360076612045114, "grad_norm": 11.4453125, "learning_rate": 8.937180216031837e-06, "loss": 0.5999, "step": 30412 }, { "epoch": 3.2361140668227284, "grad_norm": 13.8828125, "learning_rate": 8.93714468447982e-06, "loss": 0.7189, "step": 30413 }, { "epoch": 3.236220472440945, "grad_norm": 8.984375, "learning_rate": 8.937109152927801e-06, "loss": 0.5186, "step": 30414 }, { "epoch": 3.2363268780591614, "grad_norm": 16.84375, "learning_rate": 8.937073621375783e-06, "loss": 0.6643, "step": 30415 }, { "epoch": 3.2364332836773784, "grad_norm": 18.15625, "learning_rate": 8.937038089823765e-06, "loss": 0.7772, "step": 30416 }, { "epoch": 3.236539689295595, "grad_norm": 18.5, "learning_rate": 8.937002558271746e-06, "loss": 0.5996, "step": 30417 }, { "epoch": 3.2366460949138114, "grad_norm": 11.796875, "learning_rate": 8.936967026719728e-06, "loss": 0.638, "step": 30418 }, { "epoch": 3.236752500532028, "grad_norm": 13.28125, "learning_rate": 8.93693149516771e-06, "loss": 0.5797, "step": 30419 }, { "epoch": 3.236858906150245, "grad_norm": 13.7109375, "learning_rate": 8.936895963615691e-06, "loss": 0.6059, "step": 30420 }, { "epoch": 3.2369653117684614, "grad_norm": 12.0625, "learning_rate": 8.936860432063673e-06, "loss": 0.5148, "step": 30421 }, { "epoch": 3.237071717386678, "grad_norm": 14.8984375, "learning_rate": 8.936824900511654e-06, "loss": 0.7232, "step": 30422 }, { "epoch": 3.237178123004895, "grad_norm": 15.3671875, "learning_rate": 8.936789368959636e-06, "loss": 0.6829, "step": 30423 }, { "epoch": 3.2372845286231113, "grad_norm": 12.265625, "learning_rate": 8.93675383740762e-06, "loss": 0.4953, "step": 30424 }, { "epoch": 3.237390934241328, "grad_norm": 10.6953125, "learning_rate": 8.9367183058556e-06, "loss": 0.497, "step": 30425 }, { "epoch": 3.2374973398595444, "grad_norm": 11.734375, "learning_rate": 8.936682774303582e-06, "loss": 0.6087, "step": 30426 }, { "epoch": 3.2376037454777613, "grad_norm": 10.296875, "learning_rate": 8.936647242751564e-06, "loss": 0.5711, "step": 30427 }, { "epoch": 3.237710151095978, "grad_norm": 14.5234375, "learning_rate": 8.936611711199546e-06, "loss": 0.6754, "step": 30428 }, { "epoch": 3.2378165567141943, "grad_norm": 12.203125, "learning_rate": 8.936576179647527e-06, "loss": 0.6386, "step": 30429 }, { "epoch": 3.2379229623324113, "grad_norm": 9.59375, "learning_rate": 8.936540648095509e-06, "loss": 0.6239, "step": 30430 }, { "epoch": 3.238029367950628, "grad_norm": 11.078125, "learning_rate": 8.93650511654349e-06, "loss": 0.7041, "step": 30431 }, { "epoch": 3.2381357735688443, "grad_norm": 14.171875, "learning_rate": 8.936469584991472e-06, "loss": 0.6295, "step": 30432 }, { "epoch": 3.2382421791870613, "grad_norm": 10.3515625, "learning_rate": 8.936434053439454e-06, "loss": 0.6198, "step": 30433 }, { "epoch": 3.2383485848052778, "grad_norm": 10.6171875, "learning_rate": 8.936398521887437e-06, "loss": 0.6118, "step": 30434 }, { "epoch": 3.2384549904234943, "grad_norm": 11.0625, "learning_rate": 8.936362990335419e-06, "loss": 0.5636, "step": 30435 }, { "epoch": 3.2385613960417112, "grad_norm": 12.296875, "learning_rate": 8.9363274587834e-06, "loss": 0.6654, "step": 30436 }, { "epoch": 3.2386678016599277, "grad_norm": 13.09375, "learning_rate": 8.936291927231382e-06, "loss": 0.6759, "step": 30437 }, { "epoch": 3.2387742072781442, "grad_norm": 10.6875, "learning_rate": 8.936256395679363e-06, "loss": 0.5458, "step": 30438 }, { "epoch": 3.2388806128963608, "grad_norm": 9.25, "learning_rate": 8.936220864127347e-06, "loss": 0.4279, "step": 30439 }, { "epoch": 3.2389870185145777, "grad_norm": 13.28125, "learning_rate": 8.936185332575328e-06, "loss": 0.6203, "step": 30440 }, { "epoch": 3.239093424132794, "grad_norm": 13.078125, "learning_rate": 8.93614980102331e-06, "loss": 0.547, "step": 30441 }, { "epoch": 3.2391998297510107, "grad_norm": 18.03125, "learning_rate": 8.936114269471292e-06, "loss": 0.7413, "step": 30442 }, { "epoch": 3.2393062353692277, "grad_norm": 12.8671875, "learning_rate": 8.936078737919273e-06, "loss": 0.5355, "step": 30443 }, { "epoch": 3.239412640987444, "grad_norm": 20.640625, "learning_rate": 8.936043206367255e-06, "loss": 0.8395, "step": 30444 }, { "epoch": 3.2395190466056607, "grad_norm": 20.953125, "learning_rate": 8.936007674815236e-06, "loss": 0.8048, "step": 30445 }, { "epoch": 3.239625452223877, "grad_norm": 14.28125, "learning_rate": 8.935972143263218e-06, "loss": 0.8089, "step": 30446 }, { "epoch": 3.239731857842094, "grad_norm": 14.015625, "learning_rate": 8.9359366117112e-06, "loss": 0.6123, "step": 30447 }, { "epoch": 3.2398382634603107, "grad_norm": 12.78125, "learning_rate": 8.935901080159181e-06, "loss": 0.6651, "step": 30448 }, { "epoch": 3.239944669078527, "grad_norm": 11.9609375, "learning_rate": 8.935865548607164e-06, "loss": 0.5756, "step": 30449 }, { "epoch": 3.240051074696744, "grad_norm": 11.1875, "learning_rate": 8.935830017055146e-06, "loss": 0.6139, "step": 30450 }, { "epoch": 3.2401574803149606, "grad_norm": 13.09375, "learning_rate": 8.935794485503128e-06, "loss": 0.6499, "step": 30451 }, { "epoch": 3.240263885933177, "grad_norm": 15.53125, "learning_rate": 8.93575895395111e-06, "loss": 0.7282, "step": 30452 }, { "epoch": 3.240370291551394, "grad_norm": 19.390625, "learning_rate": 8.935723422399091e-06, "loss": 0.7642, "step": 30453 }, { "epoch": 3.2404766971696106, "grad_norm": 14.390625, "learning_rate": 8.935687890847074e-06, "loss": 0.7468, "step": 30454 }, { "epoch": 3.240583102787827, "grad_norm": 12.4453125, "learning_rate": 8.935652359295056e-06, "loss": 0.5345, "step": 30455 }, { "epoch": 3.2406895084060436, "grad_norm": 12.875, "learning_rate": 8.935616827743037e-06, "loss": 0.6143, "step": 30456 }, { "epoch": 3.2407959140242606, "grad_norm": 12.109375, "learning_rate": 8.935581296191019e-06, "loss": 0.6384, "step": 30457 }, { "epoch": 3.240902319642477, "grad_norm": 10.578125, "learning_rate": 8.935545764639e-06, "loss": 0.5375, "step": 30458 }, { "epoch": 3.2410087252606936, "grad_norm": 14.1328125, "learning_rate": 8.935510233086982e-06, "loss": 0.4709, "step": 30459 }, { "epoch": 3.2411151308789106, "grad_norm": 13.03125, "learning_rate": 8.935474701534964e-06, "loss": 0.6218, "step": 30460 }, { "epoch": 3.241221536497127, "grad_norm": 10.9375, "learning_rate": 8.935439169982945e-06, "loss": 0.5019, "step": 30461 }, { "epoch": 3.2413279421153436, "grad_norm": 14.1328125, "learning_rate": 8.935403638430927e-06, "loss": 0.6778, "step": 30462 }, { "epoch": 3.2414343477335605, "grad_norm": 12.6796875, "learning_rate": 8.935368106878909e-06, "loss": 0.6352, "step": 30463 }, { "epoch": 3.241540753351777, "grad_norm": 13.796875, "learning_rate": 8.93533257532689e-06, "loss": 0.7377, "step": 30464 }, { "epoch": 3.2416471589699936, "grad_norm": 8.96875, "learning_rate": 8.935297043774874e-06, "loss": 0.4767, "step": 30465 }, { "epoch": 3.24175356458821, "grad_norm": 12.1328125, "learning_rate": 8.935261512222855e-06, "loss": 0.8192, "step": 30466 }, { "epoch": 3.241859970206427, "grad_norm": 12.5625, "learning_rate": 8.935225980670837e-06, "loss": 0.6431, "step": 30467 }, { "epoch": 3.2419663758246435, "grad_norm": 11.1015625, "learning_rate": 8.935190449118818e-06, "loss": 0.5044, "step": 30468 }, { "epoch": 3.24207278144286, "grad_norm": 12.7734375, "learning_rate": 8.9351549175668e-06, "loss": 0.568, "step": 30469 }, { "epoch": 3.242179187061077, "grad_norm": 11.4296875, "learning_rate": 8.935119386014782e-06, "loss": 0.664, "step": 30470 }, { "epoch": 3.2422855926792935, "grad_norm": 15.0390625, "learning_rate": 8.935083854462763e-06, "loss": 0.6232, "step": 30471 }, { "epoch": 3.24239199829751, "grad_norm": 14.3359375, "learning_rate": 8.935048322910745e-06, "loss": 0.661, "step": 30472 }, { "epoch": 3.242498403915727, "grad_norm": 12.25, "learning_rate": 8.935012791358726e-06, "loss": 0.4894, "step": 30473 }, { "epoch": 3.2426048095339435, "grad_norm": 20.9375, "learning_rate": 8.934977259806708e-06, "loss": 0.8289, "step": 30474 }, { "epoch": 3.24271121515216, "grad_norm": 12.7109375, "learning_rate": 8.934941728254691e-06, "loss": 0.6955, "step": 30475 }, { "epoch": 3.2428176207703765, "grad_norm": 11.78125, "learning_rate": 8.934906196702673e-06, "loss": 0.5495, "step": 30476 }, { "epoch": 3.2429240263885935, "grad_norm": 13.4765625, "learning_rate": 8.934870665150655e-06, "loss": 0.5734, "step": 30477 }, { "epoch": 3.24303043200681, "grad_norm": 11.5546875, "learning_rate": 8.934835133598636e-06, "loss": 0.53, "step": 30478 }, { "epoch": 3.2431368376250265, "grad_norm": 12.765625, "learning_rate": 8.934799602046618e-06, "loss": 0.573, "step": 30479 }, { "epoch": 3.2432432432432434, "grad_norm": 13.0390625, "learning_rate": 8.934764070494601e-06, "loss": 0.6012, "step": 30480 }, { "epoch": 3.24334964886146, "grad_norm": 11.296875, "learning_rate": 8.934728538942583e-06, "loss": 0.5538, "step": 30481 }, { "epoch": 3.2434560544796764, "grad_norm": 18.171875, "learning_rate": 8.934693007390564e-06, "loss": 0.7022, "step": 30482 }, { "epoch": 3.243562460097893, "grad_norm": 13.4765625, "learning_rate": 8.934657475838546e-06, "loss": 0.5982, "step": 30483 }, { "epoch": 3.24366886571611, "grad_norm": 19.609375, "learning_rate": 8.934621944286527e-06, "loss": 0.7427, "step": 30484 }, { "epoch": 3.2437752713343264, "grad_norm": 14.1328125, "learning_rate": 8.934586412734509e-06, "loss": 0.7435, "step": 30485 }, { "epoch": 3.243881676952543, "grad_norm": 15.8671875, "learning_rate": 8.93455088118249e-06, "loss": 0.788, "step": 30486 }, { "epoch": 3.24398808257076, "grad_norm": 14.5703125, "learning_rate": 8.934515349630472e-06, "loss": 0.587, "step": 30487 }, { "epoch": 3.2440944881889764, "grad_norm": 18.921875, "learning_rate": 8.934479818078454e-06, "loss": 0.7083, "step": 30488 }, { "epoch": 3.244200893807193, "grad_norm": 8.6015625, "learning_rate": 8.934444286526436e-06, "loss": 0.6336, "step": 30489 }, { "epoch": 3.24430729942541, "grad_norm": 12.890625, "learning_rate": 8.934408754974419e-06, "loss": 0.5365, "step": 30490 }, { "epoch": 3.2444137050436264, "grad_norm": 13.6171875, "learning_rate": 8.9343732234224e-06, "loss": 0.688, "step": 30491 }, { "epoch": 3.244520110661843, "grad_norm": 13.296875, "learning_rate": 8.934337691870382e-06, "loss": 0.6986, "step": 30492 }, { "epoch": 3.2446265162800594, "grad_norm": 11.125, "learning_rate": 8.934302160318364e-06, "loss": 0.5147, "step": 30493 }, { "epoch": 3.2447329218982763, "grad_norm": 15.09375, "learning_rate": 8.934266628766345e-06, "loss": 0.6348, "step": 30494 }, { "epoch": 3.244839327516493, "grad_norm": 10.5703125, "learning_rate": 8.934231097214327e-06, "loss": 0.5261, "step": 30495 }, { "epoch": 3.2449457331347094, "grad_norm": 11.890625, "learning_rate": 8.934195565662308e-06, "loss": 0.6026, "step": 30496 }, { "epoch": 3.2450521387529263, "grad_norm": 13.484375, "learning_rate": 8.93416003411029e-06, "loss": 0.6795, "step": 30497 }, { "epoch": 3.245158544371143, "grad_norm": 12.59375, "learning_rate": 8.934124502558272e-06, "loss": 0.5361, "step": 30498 }, { "epoch": 3.2452649499893593, "grad_norm": 13.8125, "learning_rate": 8.934088971006253e-06, "loss": 0.6816, "step": 30499 }, { "epoch": 3.2453713556075763, "grad_norm": 16.4375, "learning_rate": 8.934053439454237e-06, "loss": 0.6586, "step": 30500 }, { "epoch": 3.245477761225793, "grad_norm": 10.09375, "learning_rate": 8.934017907902218e-06, "loss": 0.6129, "step": 30501 }, { "epoch": 3.2455841668440093, "grad_norm": 15.8515625, "learning_rate": 8.9339823763502e-06, "loss": 0.7118, "step": 30502 }, { "epoch": 3.245690572462226, "grad_norm": 10.15625, "learning_rate": 8.933946844798181e-06, "loss": 0.6193, "step": 30503 }, { "epoch": 3.2457969780804428, "grad_norm": 18.359375, "learning_rate": 8.933911313246163e-06, "loss": 0.8711, "step": 30504 }, { "epoch": 3.2459033836986593, "grad_norm": 11.6796875, "learning_rate": 8.933875781694145e-06, "loss": 0.6469, "step": 30505 }, { "epoch": 3.246009789316876, "grad_norm": 13.9375, "learning_rate": 8.933840250142128e-06, "loss": 0.6828, "step": 30506 }, { "epoch": 3.2461161949350927, "grad_norm": 13.8125, "learning_rate": 8.93380471859011e-06, "loss": 0.5226, "step": 30507 }, { "epoch": 3.2462226005533092, "grad_norm": 12.65625, "learning_rate": 8.933769187038091e-06, "loss": 0.5819, "step": 30508 }, { "epoch": 3.2463290061715258, "grad_norm": 13.1484375, "learning_rate": 8.933733655486073e-06, "loss": 0.723, "step": 30509 }, { "epoch": 3.2464354117897427, "grad_norm": 13.71875, "learning_rate": 8.933698123934054e-06, "loss": 0.7177, "step": 30510 }, { "epoch": 3.246541817407959, "grad_norm": 9.609375, "learning_rate": 8.933662592382036e-06, "loss": 0.5124, "step": 30511 }, { "epoch": 3.2466482230261757, "grad_norm": 15.8203125, "learning_rate": 8.933627060830018e-06, "loss": 0.7172, "step": 30512 }, { "epoch": 3.2467546286443922, "grad_norm": 9.953125, "learning_rate": 8.933591529278e-06, "loss": 0.5562, "step": 30513 }, { "epoch": 3.246861034262609, "grad_norm": 12.0703125, "learning_rate": 8.93355599772598e-06, "loss": 0.6074, "step": 30514 }, { "epoch": 3.2469674398808257, "grad_norm": 12.3828125, "learning_rate": 8.933520466173962e-06, "loss": 0.6514, "step": 30515 }, { "epoch": 3.247073845499042, "grad_norm": 14.8828125, "learning_rate": 8.933484934621946e-06, "loss": 0.7034, "step": 30516 }, { "epoch": 3.247180251117259, "grad_norm": 11.5234375, "learning_rate": 8.933449403069927e-06, "loss": 0.4891, "step": 30517 }, { "epoch": 3.2472866567354757, "grad_norm": 13.2578125, "learning_rate": 8.933413871517909e-06, "loss": 0.594, "step": 30518 }, { "epoch": 3.247393062353692, "grad_norm": 14.3359375, "learning_rate": 8.93337833996589e-06, "loss": 0.5434, "step": 30519 }, { "epoch": 3.2474994679719087, "grad_norm": 14.2578125, "learning_rate": 8.933342808413872e-06, "loss": 0.5766, "step": 30520 }, { "epoch": 3.2476058735901256, "grad_norm": 15.8046875, "learning_rate": 8.933307276861854e-06, "loss": 0.5651, "step": 30521 }, { "epoch": 3.247712279208342, "grad_norm": 14.4453125, "learning_rate": 8.933271745309835e-06, "loss": 0.6747, "step": 30522 }, { "epoch": 3.2478186848265587, "grad_norm": 15.109375, "learning_rate": 8.933236213757817e-06, "loss": 0.6986, "step": 30523 }, { "epoch": 3.2479250904447756, "grad_norm": 13.7890625, "learning_rate": 8.933200682205799e-06, "loss": 0.6342, "step": 30524 }, { "epoch": 3.248031496062992, "grad_norm": 15.2421875, "learning_rate": 8.93316515065378e-06, "loss": 0.8201, "step": 30525 }, { "epoch": 3.2481379016812086, "grad_norm": 12.7421875, "learning_rate": 8.933129619101763e-06, "loss": 0.731, "step": 30526 }, { "epoch": 3.2482443072994256, "grad_norm": 14.8828125, "learning_rate": 8.933094087549745e-06, "loss": 0.7212, "step": 30527 }, { "epoch": 3.248350712917642, "grad_norm": 10.4140625, "learning_rate": 8.933058555997727e-06, "loss": 0.5339, "step": 30528 }, { "epoch": 3.2484571185358586, "grad_norm": 16.1875, "learning_rate": 8.933023024445708e-06, "loss": 0.6986, "step": 30529 }, { "epoch": 3.2485635241540756, "grad_norm": 13.21875, "learning_rate": 8.93298749289369e-06, "loss": 0.7041, "step": 30530 }, { "epoch": 3.248669929772292, "grad_norm": 13.0859375, "learning_rate": 8.932951961341673e-06, "loss": 0.5693, "step": 30531 }, { "epoch": 3.2487763353905086, "grad_norm": 10.34375, "learning_rate": 8.932916429789655e-06, "loss": 0.5122, "step": 30532 }, { "epoch": 3.248882741008725, "grad_norm": 10.4140625, "learning_rate": 8.932880898237636e-06, "loss": 0.5563, "step": 30533 }, { "epoch": 3.248989146626942, "grad_norm": 15.8515625, "learning_rate": 8.932845366685618e-06, "loss": 0.6996, "step": 30534 }, { "epoch": 3.2490955522451586, "grad_norm": 14.1171875, "learning_rate": 8.9328098351336e-06, "loss": 0.7009, "step": 30535 }, { "epoch": 3.249201957863375, "grad_norm": 13.6015625, "learning_rate": 8.932774303581581e-06, "loss": 0.5873, "step": 30536 }, { "epoch": 3.249308363481592, "grad_norm": 12.4765625, "learning_rate": 8.932738772029563e-06, "loss": 0.6155, "step": 30537 }, { "epoch": 3.2494147690998085, "grad_norm": 10.40625, "learning_rate": 8.932703240477544e-06, "loss": 0.5412, "step": 30538 }, { "epoch": 3.249521174718025, "grad_norm": 17.640625, "learning_rate": 8.932667708925526e-06, "loss": 0.7988, "step": 30539 }, { "epoch": 3.2496275803362416, "grad_norm": 9.5234375, "learning_rate": 8.932632177373508e-06, "loss": 0.5753, "step": 30540 }, { "epoch": 3.2497339859544585, "grad_norm": 15.6875, "learning_rate": 8.93259664582149e-06, "loss": 0.6949, "step": 30541 }, { "epoch": 3.249840391572675, "grad_norm": 11.375, "learning_rate": 8.932561114269473e-06, "loss": 0.5687, "step": 30542 }, { "epoch": 3.2499467971908915, "grad_norm": 11.875, "learning_rate": 8.932525582717454e-06, "loss": 0.6448, "step": 30543 }, { "epoch": 3.2500532028091085, "grad_norm": 12.171875, "learning_rate": 8.932490051165436e-06, "loss": 0.7455, "step": 30544 }, { "epoch": 3.250159608427325, "grad_norm": 14.453125, "learning_rate": 8.932454519613417e-06, "loss": 0.7229, "step": 30545 }, { "epoch": 3.2502660140455415, "grad_norm": 10.3671875, "learning_rate": 8.932418988061399e-06, "loss": 0.5767, "step": 30546 }, { "epoch": 3.2503724196637584, "grad_norm": 13.5390625, "learning_rate": 8.932383456509382e-06, "loss": 0.6391, "step": 30547 }, { "epoch": 3.250478825281975, "grad_norm": 9.6953125, "learning_rate": 8.932347924957364e-06, "loss": 0.6378, "step": 30548 }, { "epoch": 3.2505852309001915, "grad_norm": 14.34375, "learning_rate": 8.932312393405344e-06, "loss": 0.7119, "step": 30549 }, { "epoch": 3.250691636518408, "grad_norm": 12.15625, "learning_rate": 8.932276861853325e-06, "loss": 0.5607, "step": 30550 }, { "epoch": 3.250798042136625, "grad_norm": 11.1953125, "learning_rate": 8.932241330301307e-06, "loss": 0.5208, "step": 30551 }, { "epoch": 3.2509044477548414, "grad_norm": 13.3515625, "learning_rate": 8.93220579874929e-06, "loss": 0.6824, "step": 30552 }, { "epoch": 3.251010853373058, "grad_norm": 15.3515625, "learning_rate": 8.932170267197272e-06, "loss": 0.5863, "step": 30553 }, { "epoch": 3.251117258991275, "grad_norm": 15.9375, "learning_rate": 8.932134735645254e-06, "loss": 0.6143, "step": 30554 }, { "epoch": 3.2512236646094914, "grad_norm": 10.4375, "learning_rate": 8.932099204093235e-06, "loss": 0.5128, "step": 30555 }, { "epoch": 3.251330070227708, "grad_norm": 13.78125, "learning_rate": 8.932063672541217e-06, "loss": 0.6785, "step": 30556 }, { "epoch": 3.2514364758459244, "grad_norm": 13.875, "learning_rate": 8.9320281409892e-06, "loss": 0.6567, "step": 30557 }, { "epoch": 3.2515428814641414, "grad_norm": 15.6328125, "learning_rate": 8.931992609437182e-06, "loss": 0.6671, "step": 30558 }, { "epoch": 3.251649287082358, "grad_norm": 11.8984375, "learning_rate": 8.931957077885163e-06, "loss": 0.5497, "step": 30559 }, { "epoch": 3.2517556927005744, "grad_norm": 13.5234375, "learning_rate": 8.931921546333145e-06, "loss": 0.6262, "step": 30560 }, { "epoch": 3.2518620983187914, "grad_norm": 18.09375, "learning_rate": 8.931886014781126e-06, "loss": 0.7168, "step": 30561 }, { "epoch": 3.251968503937008, "grad_norm": 12.3515625, "learning_rate": 8.931850483229108e-06, "loss": 0.5545, "step": 30562 }, { "epoch": 3.2520749095552244, "grad_norm": 10.328125, "learning_rate": 8.93181495167709e-06, "loss": 0.4817, "step": 30563 }, { "epoch": 3.2521813151734413, "grad_norm": 15.296875, "learning_rate": 8.931779420125071e-06, "loss": 0.6699, "step": 30564 }, { "epoch": 3.252287720791658, "grad_norm": 14.1171875, "learning_rate": 8.931743888573053e-06, "loss": 0.6704, "step": 30565 }, { "epoch": 3.2523941264098744, "grad_norm": 11.8203125, "learning_rate": 8.931708357021034e-06, "loss": 0.6105, "step": 30566 }, { "epoch": 3.2525005320280913, "grad_norm": 15.84375, "learning_rate": 8.931672825469018e-06, "loss": 0.7478, "step": 30567 }, { "epoch": 3.252606937646308, "grad_norm": 12.96875, "learning_rate": 8.931637293917e-06, "loss": 0.6375, "step": 30568 }, { "epoch": 3.2527133432645243, "grad_norm": 13.3125, "learning_rate": 8.931601762364981e-06, "loss": 0.7466, "step": 30569 }, { "epoch": 3.252819748882741, "grad_norm": 11.609375, "learning_rate": 8.931566230812963e-06, "loss": 0.5529, "step": 30570 }, { "epoch": 3.252926154500958, "grad_norm": 21.859375, "learning_rate": 8.931530699260944e-06, "loss": 0.8018, "step": 30571 }, { "epoch": 3.2530325601191743, "grad_norm": 16.75, "learning_rate": 8.931495167708928e-06, "loss": 0.7935, "step": 30572 }, { "epoch": 3.253138965737391, "grad_norm": 12.75, "learning_rate": 8.931459636156909e-06, "loss": 0.6017, "step": 30573 }, { "epoch": 3.2532453713556078, "grad_norm": 14.4765625, "learning_rate": 8.93142410460489e-06, "loss": 0.5484, "step": 30574 }, { "epoch": 3.2533517769738243, "grad_norm": 15.5390625, "learning_rate": 8.931388573052872e-06, "loss": 0.7228, "step": 30575 }, { "epoch": 3.253458182592041, "grad_norm": 10.015625, "learning_rate": 8.931353041500854e-06, "loss": 0.6442, "step": 30576 }, { "epoch": 3.2535645882102573, "grad_norm": 18.234375, "learning_rate": 8.931317509948836e-06, "loss": 1.0068, "step": 30577 }, { "epoch": 3.2536709938284742, "grad_norm": 15.1484375, "learning_rate": 8.931281978396817e-06, "loss": 0.609, "step": 30578 }, { "epoch": 3.2537773994466908, "grad_norm": 12.9453125, "learning_rate": 8.931246446844799e-06, "loss": 0.7922, "step": 30579 }, { "epoch": 3.2538838050649073, "grad_norm": 11.8671875, "learning_rate": 8.93121091529278e-06, "loss": 0.6591, "step": 30580 }, { "epoch": 3.253990210683124, "grad_norm": 11.984375, "learning_rate": 8.931175383740762e-06, "loss": 0.6331, "step": 30581 }, { "epoch": 3.2540966163013407, "grad_norm": 15.0234375, "learning_rate": 8.931139852188744e-06, "loss": 0.648, "step": 30582 }, { "epoch": 3.2542030219195572, "grad_norm": 15.1484375, "learning_rate": 8.931104320636727e-06, "loss": 0.684, "step": 30583 }, { "epoch": 3.254309427537774, "grad_norm": 7.3046875, "learning_rate": 8.931068789084708e-06, "loss": 0.4733, "step": 30584 }, { "epoch": 3.2544158331559907, "grad_norm": 16.03125, "learning_rate": 8.93103325753269e-06, "loss": 0.722, "step": 30585 }, { "epoch": 3.254522238774207, "grad_norm": 13.3671875, "learning_rate": 8.930997725980672e-06, "loss": 0.6393, "step": 30586 }, { "epoch": 3.254628644392424, "grad_norm": 12.3046875, "learning_rate": 8.930962194428653e-06, "loss": 0.645, "step": 30587 }, { "epoch": 3.2547350500106407, "grad_norm": 13.4453125, "learning_rate": 8.930926662876635e-06, "loss": 0.6104, "step": 30588 }, { "epoch": 3.254841455628857, "grad_norm": 12.609375, "learning_rate": 8.930891131324617e-06, "loss": 0.6598, "step": 30589 }, { "epoch": 3.2549478612470737, "grad_norm": 14.8203125, "learning_rate": 8.930855599772598e-06, "loss": 0.6016, "step": 30590 }, { "epoch": 3.2550542668652906, "grad_norm": 13.9296875, "learning_rate": 8.93082006822058e-06, "loss": 0.5883, "step": 30591 }, { "epoch": 3.255160672483507, "grad_norm": 10.6875, "learning_rate": 8.930784536668561e-06, "loss": 0.4829, "step": 30592 }, { "epoch": 3.2552670781017237, "grad_norm": 12.0859375, "learning_rate": 8.930749005116545e-06, "loss": 0.6163, "step": 30593 }, { "epoch": 3.25537348371994, "grad_norm": 13.625, "learning_rate": 8.930713473564526e-06, "loss": 0.674, "step": 30594 }, { "epoch": 3.255479889338157, "grad_norm": 11.21875, "learning_rate": 8.930677942012508e-06, "loss": 0.6596, "step": 30595 }, { "epoch": 3.2555862949563736, "grad_norm": 15.1015625, "learning_rate": 8.93064241046049e-06, "loss": 0.8883, "step": 30596 }, { "epoch": 3.25569270057459, "grad_norm": 10.734375, "learning_rate": 8.930606878908471e-06, "loss": 0.5137, "step": 30597 }, { "epoch": 3.255799106192807, "grad_norm": 12.75, "learning_rate": 8.930571347356454e-06, "loss": 0.5522, "step": 30598 }, { "epoch": 3.2559055118110236, "grad_norm": 16.671875, "learning_rate": 8.930535815804436e-06, "loss": 0.5513, "step": 30599 }, { "epoch": 3.25601191742924, "grad_norm": 11.375, "learning_rate": 8.930500284252418e-06, "loss": 0.4535, "step": 30600 }, { "epoch": 3.256118323047457, "grad_norm": 16.546875, "learning_rate": 8.9304647527004e-06, "loss": 0.5722, "step": 30601 }, { "epoch": 3.2562247286656736, "grad_norm": 15.4453125, "learning_rate": 8.93042922114838e-06, "loss": 0.8202, "step": 30602 }, { "epoch": 3.25633113428389, "grad_norm": 11.6953125, "learning_rate": 8.930393689596362e-06, "loss": 0.6664, "step": 30603 }, { "epoch": 3.256437539902107, "grad_norm": 12.8203125, "learning_rate": 8.930358158044344e-06, "loss": 0.7095, "step": 30604 }, { "epoch": 3.2565439455203236, "grad_norm": 13.9140625, "learning_rate": 8.930322626492326e-06, "loss": 0.6628, "step": 30605 }, { "epoch": 3.25665035113854, "grad_norm": 16.453125, "learning_rate": 8.930287094940307e-06, "loss": 0.826, "step": 30606 }, { "epoch": 3.2567567567567566, "grad_norm": 10.8984375, "learning_rate": 8.930251563388289e-06, "loss": 0.6446, "step": 30607 }, { "epoch": 3.2568631623749735, "grad_norm": 10.53125, "learning_rate": 8.930216031836272e-06, "loss": 0.7332, "step": 30608 }, { "epoch": 3.25696956799319, "grad_norm": 14.1171875, "learning_rate": 8.930180500284254e-06, "loss": 0.7805, "step": 30609 }, { "epoch": 3.2570759736114065, "grad_norm": 14.203125, "learning_rate": 8.930144968732235e-06, "loss": 0.6203, "step": 30610 }, { "epoch": 3.2571823792296235, "grad_norm": 11.7890625, "learning_rate": 8.930109437180217e-06, "loss": 0.5362, "step": 30611 }, { "epoch": 3.25728878484784, "grad_norm": 12.6015625, "learning_rate": 8.930073905628199e-06, "loss": 0.6646, "step": 30612 }, { "epoch": 3.2573951904660565, "grad_norm": 12.1328125, "learning_rate": 8.93003837407618e-06, "loss": 0.5753, "step": 30613 }, { "epoch": 3.257501596084273, "grad_norm": 12.078125, "learning_rate": 8.930002842524162e-06, "loss": 0.57, "step": 30614 }, { "epoch": 3.25760800170249, "grad_norm": 11.078125, "learning_rate": 8.929967310972143e-06, "loss": 0.5806, "step": 30615 }, { "epoch": 3.2577144073207065, "grad_norm": 13.359375, "learning_rate": 8.929931779420125e-06, "loss": 0.6447, "step": 30616 }, { "epoch": 3.257820812938923, "grad_norm": 12.8359375, "learning_rate": 8.929896247868107e-06, "loss": 0.6533, "step": 30617 }, { "epoch": 3.25792721855714, "grad_norm": 12.234375, "learning_rate": 8.929860716316088e-06, "loss": 0.7583, "step": 30618 }, { "epoch": 3.2580336241753565, "grad_norm": 11.1328125, "learning_rate": 8.929825184764071e-06, "loss": 0.5806, "step": 30619 }, { "epoch": 3.258140029793573, "grad_norm": 18.1875, "learning_rate": 8.929789653212053e-06, "loss": 0.7043, "step": 30620 }, { "epoch": 3.25824643541179, "grad_norm": 13.8984375, "learning_rate": 8.929754121660035e-06, "loss": 0.6099, "step": 30621 }, { "epoch": 3.2583528410300064, "grad_norm": 15.109375, "learning_rate": 8.929718590108016e-06, "loss": 0.7586, "step": 30622 }, { "epoch": 3.258459246648223, "grad_norm": 17.3125, "learning_rate": 8.929683058555998e-06, "loss": 0.6713, "step": 30623 }, { "epoch": 3.25856565226644, "grad_norm": 16.0, "learning_rate": 8.929647527003981e-06, "loss": 0.8115, "step": 30624 }, { "epoch": 3.2586720578846564, "grad_norm": 10.734375, "learning_rate": 8.929611995451963e-06, "loss": 0.5331, "step": 30625 }, { "epoch": 3.258778463502873, "grad_norm": 10.4921875, "learning_rate": 8.929576463899944e-06, "loss": 0.6115, "step": 30626 }, { "epoch": 3.2588848691210894, "grad_norm": 9.8828125, "learning_rate": 8.929540932347926e-06, "loss": 0.4774, "step": 30627 }, { "epoch": 3.2589912747393064, "grad_norm": 13.015625, "learning_rate": 8.929505400795908e-06, "loss": 0.6847, "step": 30628 }, { "epoch": 3.259097680357523, "grad_norm": 8.4140625, "learning_rate": 8.92946986924389e-06, "loss": 0.3813, "step": 30629 }, { "epoch": 3.2592040859757394, "grad_norm": 18.359375, "learning_rate": 8.929434337691871e-06, "loss": 0.6278, "step": 30630 }, { "epoch": 3.259310491593956, "grad_norm": 11.6484375, "learning_rate": 8.929398806139852e-06, "loss": 0.5886, "step": 30631 }, { "epoch": 3.259416897212173, "grad_norm": 12.6796875, "learning_rate": 8.929363274587834e-06, "loss": 0.6034, "step": 30632 }, { "epoch": 3.2595233028303894, "grad_norm": 14.328125, "learning_rate": 8.929327743035816e-06, "loss": 0.6282, "step": 30633 }, { "epoch": 3.259629708448606, "grad_norm": 18.0, "learning_rate": 8.929292211483799e-06, "loss": 0.706, "step": 30634 }, { "epoch": 3.259736114066823, "grad_norm": 15.1484375, "learning_rate": 8.92925667993178e-06, "loss": 0.7839, "step": 30635 }, { "epoch": 3.2598425196850394, "grad_norm": 12.953125, "learning_rate": 8.929221148379762e-06, "loss": 0.5917, "step": 30636 }, { "epoch": 3.259948925303256, "grad_norm": 12.1328125, "learning_rate": 8.929185616827744e-06, "loss": 0.5611, "step": 30637 }, { "epoch": 3.260055330921473, "grad_norm": 13.8046875, "learning_rate": 8.929150085275725e-06, "loss": 0.7996, "step": 30638 }, { "epoch": 3.2601617365396893, "grad_norm": 13.09375, "learning_rate": 8.929114553723707e-06, "loss": 0.5983, "step": 30639 }, { "epoch": 3.260268142157906, "grad_norm": 10.2109375, "learning_rate": 8.929079022171689e-06, "loss": 0.4829, "step": 30640 }, { "epoch": 3.260374547776123, "grad_norm": 17.8125, "learning_rate": 8.92904349061967e-06, "loss": 0.6568, "step": 30641 }, { "epoch": 3.2604809533943393, "grad_norm": 17.734375, "learning_rate": 8.929007959067652e-06, "loss": 0.8505, "step": 30642 }, { "epoch": 3.260587359012556, "grad_norm": 13.3046875, "learning_rate": 8.928972427515633e-06, "loss": 0.6187, "step": 30643 }, { "epoch": 3.2606937646307723, "grad_norm": 10.3203125, "learning_rate": 8.928936895963617e-06, "loss": 0.666, "step": 30644 }, { "epoch": 3.2608001702489893, "grad_norm": 12.140625, "learning_rate": 8.928901364411598e-06, "loss": 0.5649, "step": 30645 }, { "epoch": 3.260906575867206, "grad_norm": 12.9140625, "learning_rate": 8.92886583285958e-06, "loss": 0.6764, "step": 30646 }, { "epoch": 3.2610129814854223, "grad_norm": 7.92578125, "learning_rate": 8.928830301307562e-06, "loss": 0.5125, "step": 30647 }, { "epoch": 3.2611193871036392, "grad_norm": 14.5625, "learning_rate": 8.928794769755543e-06, "loss": 0.7527, "step": 30648 }, { "epoch": 3.2612257927218558, "grad_norm": 14.4609375, "learning_rate": 8.928759238203526e-06, "loss": 0.5796, "step": 30649 }, { "epoch": 3.2613321983400723, "grad_norm": 12.25, "learning_rate": 8.928723706651508e-06, "loss": 0.8325, "step": 30650 }, { "epoch": 3.2614386039582888, "grad_norm": 16.578125, "learning_rate": 8.92868817509949e-06, "loss": 0.7228, "step": 30651 }, { "epoch": 3.2615450095765057, "grad_norm": 15.1015625, "learning_rate": 8.928652643547471e-06, "loss": 0.8027, "step": 30652 }, { "epoch": 3.2616514151947222, "grad_norm": 10.90625, "learning_rate": 8.928617111995453e-06, "loss": 0.6916, "step": 30653 }, { "epoch": 3.2617578208129387, "grad_norm": 15.0703125, "learning_rate": 8.928581580443434e-06, "loss": 0.7132, "step": 30654 }, { "epoch": 3.2618642264311557, "grad_norm": 12.7421875, "learning_rate": 8.928546048891416e-06, "loss": 0.4938, "step": 30655 }, { "epoch": 3.261970632049372, "grad_norm": 14.40625, "learning_rate": 8.928510517339398e-06, "loss": 0.6016, "step": 30656 }, { "epoch": 3.2620770376675887, "grad_norm": 13.2265625, "learning_rate": 8.92847498578738e-06, "loss": 0.595, "step": 30657 }, { "epoch": 3.2621834432858057, "grad_norm": 13.609375, "learning_rate": 8.928439454235361e-06, "loss": 0.6584, "step": 30658 }, { "epoch": 3.262289848904022, "grad_norm": 12.5234375, "learning_rate": 8.928403922683343e-06, "loss": 0.5035, "step": 30659 }, { "epoch": 3.2623962545222387, "grad_norm": 13.1640625, "learning_rate": 8.928368391131326e-06, "loss": 0.6444, "step": 30660 }, { "epoch": 3.2625026601404556, "grad_norm": 13.171875, "learning_rate": 8.928332859579307e-06, "loss": 0.7094, "step": 30661 }, { "epoch": 3.262609065758672, "grad_norm": 9.0546875, "learning_rate": 8.928297328027289e-06, "loss": 0.4422, "step": 30662 }, { "epoch": 3.2627154713768887, "grad_norm": 12.0, "learning_rate": 8.92826179647527e-06, "loss": 0.4664, "step": 30663 }, { "epoch": 3.262821876995105, "grad_norm": 10.1484375, "learning_rate": 8.928226264923252e-06, "loss": 0.5592, "step": 30664 }, { "epoch": 3.262928282613322, "grad_norm": 13.9609375, "learning_rate": 8.928190733371236e-06, "loss": 0.5898, "step": 30665 }, { "epoch": 3.2630346882315386, "grad_norm": 12.4453125, "learning_rate": 8.928155201819217e-06, "loss": 0.5449, "step": 30666 }, { "epoch": 3.263141093849755, "grad_norm": 15.359375, "learning_rate": 8.928119670267199e-06, "loss": 0.7083, "step": 30667 }, { "epoch": 3.263247499467972, "grad_norm": 14.78125, "learning_rate": 8.92808413871518e-06, "loss": 0.7459, "step": 30668 }, { "epoch": 3.2633539050861886, "grad_norm": 15.28125, "learning_rate": 8.92804860716316e-06, "loss": 0.6935, "step": 30669 }, { "epoch": 3.263460310704405, "grad_norm": 11.6953125, "learning_rate": 8.928013075611144e-06, "loss": 0.6227, "step": 30670 }, { "epoch": 3.2635667163226216, "grad_norm": 10.3515625, "learning_rate": 8.927977544059125e-06, "loss": 0.5962, "step": 30671 }, { "epoch": 3.2636731219408386, "grad_norm": 11.9296875, "learning_rate": 8.927942012507107e-06, "loss": 0.6115, "step": 30672 }, { "epoch": 3.263779527559055, "grad_norm": 11.7890625, "learning_rate": 8.927906480955088e-06, "loss": 0.6132, "step": 30673 }, { "epoch": 3.2638859331772716, "grad_norm": 11.4609375, "learning_rate": 8.92787094940307e-06, "loss": 0.7231, "step": 30674 }, { "epoch": 3.2639923387954886, "grad_norm": 13.421875, "learning_rate": 8.927835417851053e-06, "loss": 0.6214, "step": 30675 }, { "epoch": 3.264098744413705, "grad_norm": 12.265625, "learning_rate": 8.927799886299035e-06, "loss": 0.5616, "step": 30676 }, { "epoch": 3.2642051500319216, "grad_norm": 13.09375, "learning_rate": 8.927764354747017e-06, "loss": 0.649, "step": 30677 }, { "epoch": 3.2643115556501385, "grad_norm": 11.9765625, "learning_rate": 8.927728823194998e-06, "loss": 0.5995, "step": 30678 }, { "epoch": 3.264417961268355, "grad_norm": 12.1484375, "learning_rate": 8.92769329164298e-06, "loss": 0.7569, "step": 30679 }, { "epoch": 3.2645243668865715, "grad_norm": 11.2265625, "learning_rate": 8.927657760090961e-06, "loss": 0.5111, "step": 30680 }, { "epoch": 3.264630772504788, "grad_norm": 12.296875, "learning_rate": 8.927622228538943e-06, "loss": 0.5866, "step": 30681 }, { "epoch": 3.264737178123005, "grad_norm": 13.8125, "learning_rate": 8.927586696986925e-06, "loss": 0.6822, "step": 30682 }, { "epoch": 3.2648435837412215, "grad_norm": 13.4921875, "learning_rate": 8.927551165434906e-06, "loss": 0.662, "step": 30683 }, { "epoch": 3.264949989359438, "grad_norm": 11.2109375, "learning_rate": 8.927515633882888e-06, "loss": 0.5211, "step": 30684 }, { "epoch": 3.265056394977655, "grad_norm": 14.7265625, "learning_rate": 8.927480102330871e-06, "loss": 0.6583, "step": 30685 }, { "epoch": 3.2651628005958715, "grad_norm": 10.640625, "learning_rate": 8.927444570778853e-06, "loss": 0.5805, "step": 30686 }, { "epoch": 3.265269206214088, "grad_norm": 13.2265625, "learning_rate": 8.927409039226834e-06, "loss": 0.5176, "step": 30687 }, { "epoch": 3.2653756118323045, "grad_norm": 15.8515625, "learning_rate": 8.927373507674816e-06, "loss": 0.6451, "step": 30688 }, { "epoch": 3.2654820174505215, "grad_norm": 14.0, "learning_rate": 8.927337976122798e-06, "loss": 0.581, "step": 30689 }, { "epoch": 3.265588423068738, "grad_norm": 12.015625, "learning_rate": 8.92730244457078e-06, "loss": 0.6942, "step": 30690 }, { "epoch": 3.2656948286869545, "grad_norm": 10.3125, "learning_rate": 8.927266913018762e-06, "loss": 0.4681, "step": 30691 }, { "epoch": 3.2658012343051714, "grad_norm": 11.0546875, "learning_rate": 8.927231381466744e-06, "loss": 0.5545, "step": 30692 }, { "epoch": 3.265907639923388, "grad_norm": 14.2109375, "learning_rate": 8.927195849914726e-06, "loss": 0.5895, "step": 30693 }, { "epoch": 3.2660140455416045, "grad_norm": 11.3515625, "learning_rate": 8.927160318362707e-06, "loss": 0.6368, "step": 30694 }, { "epoch": 3.2661204511598214, "grad_norm": 13.890625, "learning_rate": 8.927124786810689e-06, "loss": 0.6518, "step": 30695 }, { "epoch": 3.266226856778038, "grad_norm": 13.3125, "learning_rate": 8.92708925525867e-06, "loss": 0.6874, "step": 30696 }, { "epoch": 3.2663332623962544, "grad_norm": 14.640625, "learning_rate": 8.927053723706652e-06, "loss": 0.6929, "step": 30697 }, { "epoch": 3.2664396680144714, "grad_norm": 11.28125, "learning_rate": 8.927018192154634e-06, "loss": 0.5452, "step": 30698 }, { "epoch": 3.266546073632688, "grad_norm": 13.046875, "learning_rate": 8.926982660602615e-06, "loss": 0.5878, "step": 30699 }, { "epoch": 3.2666524792509044, "grad_norm": 60.96875, "learning_rate": 8.926947129050597e-06, "loss": 0.6396, "step": 30700 }, { "epoch": 3.266758884869121, "grad_norm": 9.6640625, "learning_rate": 8.92691159749858e-06, "loss": 0.6191, "step": 30701 }, { "epoch": 3.266865290487338, "grad_norm": 15.7734375, "learning_rate": 8.926876065946562e-06, "loss": 0.704, "step": 30702 }, { "epoch": 3.2669716961055544, "grad_norm": 17.765625, "learning_rate": 8.926840534394543e-06, "loss": 0.6837, "step": 30703 }, { "epoch": 3.267078101723771, "grad_norm": 13.6484375, "learning_rate": 8.926805002842525e-06, "loss": 0.6669, "step": 30704 }, { "epoch": 3.267184507341988, "grad_norm": 16.6875, "learning_rate": 8.926769471290507e-06, "loss": 0.7103, "step": 30705 }, { "epoch": 3.2672909129602044, "grad_norm": 11.3671875, "learning_rate": 8.926733939738488e-06, "loss": 0.5595, "step": 30706 }, { "epoch": 3.267397318578421, "grad_norm": 11.1953125, "learning_rate": 8.92669840818647e-06, "loss": 0.6002, "step": 30707 }, { "epoch": 3.2675037241966374, "grad_norm": 10.78125, "learning_rate": 8.926662876634451e-06, "loss": 0.4221, "step": 30708 }, { "epoch": 3.2676101298148543, "grad_norm": 15.625, "learning_rate": 8.926627345082433e-06, "loss": 0.6138, "step": 30709 }, { "epoch": 3.267716535433071, "grad_norm": 11.9140625, "learning_rate": 8.926591813530415e-06, "loss": 0.6169, "step": 30710 }, { "epoch": 3.2678229410512873, "grad_norm": 14.6015625, "learning_rate": 8.926556281978398e-06, "loss": 0.6246, "step": 30711 }, { "epoch": 3.2679293466695043, "grad_norm": 19.078125, "learning_rate": 8.92652075042638e-06, "loss": 0.857, "step": 30712 }, { "epoch": 3.268035752287721, "grad_norm": 13.1875, "learning_rate": 8.926485218874361e-06, "loss": 0.6334, "step": 30713 }, { "epoch": 3.2681421579059373, "grad_norm": 12.296875, "learning_rate": 8.926449687322343e-06, "loss": 0.5349, "step": 30714 }, { "epoch": 3.2682485635241543, "grad_norm": 14.6875, "learning_rate": 8.926414155770324e-06, "loss": 0.6457, "step": 30715 }, { "epoch": 3.268354969142371, "grad_norm": 19.328125, "learning_rate": 8.926378624218308e-06, "loss": 0.6707, "step": 30716 }, { "epoch": 3.2684613747605873, "grad_norm": 13.03125, "learning_rate": 8.92634309266629e-06, "loss": 0.6582, "step": 30717 }, { "epoch": 3.2685677803788042, "grad_norm": 10.96875, "learning_rate": 8.926307561114271e-06, "loss": 0.6159, "step": 30718 }, { "epoch": 3.2686741859970208, "grad_norm": 18.53125, "learning_rate": 8.926272029562252e-06, "loss": 0.7604, "step": 30719 }, { "epoch": 3.2687805916152373, "grad_norm": 15.359375, "learning_rate": 8.926236498010234e-06, "loss": 0.6499, "step": 30720 }, { "epoch": 3.2688869972334538, "grad_norm": 20.03125, "learning_rate": 8.926200966458216e-06, "loss": 0.7189, "step": 30721 }, { "epoch": 3.2689934028516707, "grad_norm": 11.140625, "learning_rate": 8.926165434906197e-06, "loss": 0.5374, "step": 30722 }, { "epoch": 3.2690998084698872, "grad_norm": 11.03125, "learning_rate": 8.926129903354179e-06, "loss": 0.4838, "step": 30723 }, { "epoch": 3.2692062140881037, "grad_norm": 14.28125, "learning_rate": 8.92609437180216e-06, "loss": 0.6272, "step": 30724 }, { "epoch": 3.2693126197063203, "grad_norm": 17.984375, "learning_rate": 8.926058840250142e-06, "loss": 0.7718, "step": 30725 }, { "epoch": 3.269419025324537, "grad_norm": 10.1640625, "learning_rate": 8.926023308698125e-06, "loss": 0.428, "step": 30726 }, { "epoch": 3.2695254309427537, "grad_norm": 11.4921875, "learning_rate": 8.925987777146107e-06, "loss": 0.5394, "step": 30727 }, { "epoch": 3.2696318365609702, "grad_norm": 11.125, "learning_rate": 8.925952245594089e-06, "loss": 0.604, "step": 30728 }, { "epoch": 3.269738242179187, "grad_norm": 14.609375, "learning_rate": 8.92591671404207e-06, "loss": 0.5649, "step": 30729 }, { "epoch": 3.2698446477974037, "grad_norm": 13.4921875, "learning_rate": 8.925881182490052e-06, "loss": 0.6518, "step": 30730 }, { "epoch": 3.26995105341562, "grad_norm": 13.125, "learning_rate": 8.925845650938033e-06, "loss": 0.6311, "step": 30731 }, { "epoch": 3.270057459033837, "grad_norm": 12.2890625, "learning_rate": 8.925810119386015e-06, "loss": 0.5077, "step": 30732 }, { "epoch": 3.2701638646520537, "grad_norm": 11.9609375, "learning_rate": 8.925774587833997e-06, "loss": 0.6323, "step": 30733 }, { "epoch": 3.27027027027027, "grad_norm": 14.1484375, "learning_rate": 8.925739056281978e-06, "loss": 0.7023, "step": 30734 }, { "epoch": 3.270376675888487, "grad_norm": 13.609375, "learning_rate": 8.92570352472996e-06, "loss": 0.7912, "step": 30735 }, { "epoch": 3.2704830815067036, "grad_norm": 12.25, "learning_rate": 8.925667993177941e-06, "loss": 0.5223, "step": 30736 }, { "epoch": 3.27058948712492, "grad_norm": 14.5703125, "learning_rate": 8.925632461625925e-06, "loss": 0.6598, "step": 30737 }, { "epoch": 3.2706958927431367, "grad_norm": 11.6171875, "learning_rate": 8.925596930073906e-06, "loss": 0.5605, "step": 30738 }, { "epoch": 3.2708022983613536, "grad_norm": 13.3359375, "learning_rate": 8.925561398521888e-06, "loss": 0.6231, "step": 30739 }, { "epoch": 3.27090870397957, "grad_norm": 12.8671875, "learning_rate": 8.92552586696987e-06, "loss": 0.7465, "step": 30740 }, { "epoch": 3.2710151095977866, "grad_norm": 15.46875, "learning_rate": 8.925490335417851e-06, "loss": 0.6687, "step": 30741 }, { "epoch": 3.2711215152160036, "grad_norm": 13.859375, "learning_rate": 8.925454803865835e-06, "loss": 0.7013, "step": 30742 }, { "epoch": 3.27122792083422, "grad_norm": 11.71875, "learning_rate": 8.925419272313816e-06, "loss": 0.4393, "step": 30743 }, { "epoch": 3.2713343264524366, "grad_norm": 14.046875, "learning_rate": 8.925383740761798e-06, "loss": 0.6398, "step": 30744 }, { "epoch": 3.271440732070653, "grad_norm": 11.4609375, "learning_rate": 8.92534820920978e-06, "loss": 0.5544, "step": 30745 }, { "epoch": 3.27154713768887, "grad_norm": 10.1953125, "learning_rate": 8.925312677657761e-06, "loss": 0.4636, "step": 30746 }, { "epoch": 3.2716535433070866, "grad_norm": 13.390625, "learning_rate": 8.925277146105743e-06, "loss": 0.6807, "step": 30747 }, { "epoch": 3.271759948925303, "grad_norm": 10.2578125, "learning_rate": 8.925241614553724e-06, "loss": 0.586, "step": 30748 }, { "epoch": 3.27186635454352, "grad_norm": 12.578125, "learning_rate": 8.925206083001706e-06, "loss": 0.6464, "step": 30749 }, { "epoch": 3.2719727601617365, "grad_norm": 9.2578125, "learning_rate": 8.925170551449687e-06, "loss": 0.5176, "step": 30750 }, { "epoch": 3.272079165779953, "grad_norm": 13.4921875, "learning_rate": 8.925135019897669e-06, "loss": 0.5827, "step": 30751 }, { "epoch": 3.27218557139817, "grad_norm": 11.4609375, "learning_rate": 8.925099488345652e-06, "loss": 0.5491, "step": 30752 }, { "epoch": 3.2722919770163865, "grad_norm": 10.390625, "learning_rate": 8.925063956793634e-06, "loss": 0.4932, "step": 30753 }, { "epoch": 3.272398382634603, "grad_norm": 13.2421875, "learning_rate": 8.925028425241615e-06, "loss": 0.737, "step": 30754 }, { "epoch": 3.27250478825282, "grad_norm": 10.890625, "learning_rate": 8.924992893689597e-06, "loss": 0.4731, "step": 30755 }, { "epoch": 3.2726111938710365, "grad_norm": 13.953125, "learning_rate": 8.924957362137579e-06, "loss": 0.8271, "step": 30756 }, { "epoch": 3.272717599489253, "grad_norm": 13.0078125, "learning_rate": 8.924921830585562e-06, "loss": 0.6725, "step": 30757 }, { "epoch": 3.2728240051074695, "grad_norm": 18.75, "learning_rate": 8.924886299033542e-06, "loss": 0.781, "step": 30758 }, { "epoch": 3.2729304107256865, "grad_norm": 11.890625, "learning_rate": 8.924850767481524e-06, "loss": 0.5689, "step": 30759 }, { "epoch": 3.273036816343903, "grad_norm": 12.7890625, "learning_rate": 8.924815235929505e-06, "loss": 0.6938, "step": 30760 }, { "epoch": 3.2731432219621195, "grad_norm": 11.5703125, "learning_rate": 8.924779704377487e-06, "loss": 0.6953, "step": 30761 }, { "epoch": 3.273249627580336, "grad_norm": 12.2109375, "learning_rate": 8.92474417282547e-06, "loss": 0.5306, "step": 30762 }, { "epoch": 3.273356033198553, "grad_norm": 10.328125, "learning_rate": 8.924708641273452e-06, "loss": 0.4678, "step": 30763 }, { "epoch": 3.2734624388167695, "grad_norm": 10.6953125, "learning_rate": 8.924673109721433e-06, "loss": 0.5056, "step": 30764 }, { "epoch": 3.273568844434986, "grad_norm": 12.578125, "learning_rate": 8.924637578169415e-06, "loss": 0.5784, "step": 30765 }, { "epoch": 3.273675250053203, "grad_norm": 11.546875, "learning_rate": 8.924602046617396e-06, "loss": 0.6313, "step": 30766 }, { "epoch": 3.2737816556714194, "grad_norm": 14.3046875, "learning_rate": 8.92456651506538e-06, "loss": 0.6311, "step": 30767 }, { "epoch": 3.273888061289636, "grad_norm": 12.9375, "learning_rate": 8.924530983513361e-06, "loss": 0.713, "step": 30768 }, { "epoch": 3.273994466907853, "grad_norm": 13.359375, "learning_rate": 8.924495451961343e-06, "loss": 0.6679, "step": 30769 }, { "epoch": 3.2741008725260694, "grad_norm": 13.5078125, "learning_rate": 8.924459920409325e-06, "loss": 0.6385, "step": 30770 }, { "epoch": 3.274207278144286, "grad_norm": 13.375, "learning_rate": 8.924424388857306e-06, "loss": 0.7357, "step": 30771 }, { "epoch": 3.274313683762503, "grad_norm": 12.296875, "learning_rate": 8.924388857305288e-06, "loss": 0.6959, "step": 30772 }, { "epoch": 3.2744200893807194, "grad_norm": 14.140625, "learning_rate": 8.92435332575327e-06, "loss": 0.5897, "step": 30773 }, { "epoch": 3.274526494998936, "grad_norm": 15.8203125, "learning_rate": 8.924317794201251e-06, "loss": 0.8462, "step": 30774 }, { "epoch": 3.2746329006171524, "grad_norm": 10.765625, "learning_rate": 8.924282262649233e-06, "loss": 0.5665, "step": 30775 }, { "epoch": 3.2747393062353694, "grad_norm": 12.328125, "learning_rate": 8.924246731097214e-06, "loss": 0.5855, "step": 30776 }, { "epoch": 3.274845711853586, "grad_norm": 11.484375, "learning_rate": 8.924211199545196e-06, "loss": 0.4791, "step": 30777 }, { "epoch": 3.2749521174718024, "grad_norm": 13.5625, "learning_rate": 8.924175667993179e-06, "loss": 0.6588, "step": 30778 }, { "epoch": 3.2750585230900193, "grad_norm": 13.5625, "learning_rate": 8.92414013644116e-06, "loss": 0.6682, "step": 30779 }, { "epoch": 3.275164928708236, "grad_norm": 11.0703125, "learning_rate": 8.924104604889142e-06, "loss": 0.6427, "step": 30780 }, { "epoch": 3.2752713343264523, "grad_norm": 12.265625, "learning_rate": 8.924069073337124e-06, "loss": 0.5918, "step": 30781 }, { "epoch": 3.275377739944669, "grad_norm": 12.78125, "learning_rate": 8.924033541785106e-06, "loss": 0.7326, "step": 30782 }, { "epoch": 3.275484145562886, "grad_norm": 11.4609375, "learning_rate": 8.923998010233089e-06, "loss": 0.5951, "step": 30783 }, { "epoch": 3.2755905511811023, "grad_norm": 11.75, "learning_rate": 8.92396247868107e-06, "loss": 0.4877, "step": 30784 }, { "epoch": 3.275696956799319, "grad_norm": 10.609375, "learning_rate": 8.923926947129052e-06, "loss": 0.5937, "step": 30785 }, { "epoch": 3.275803362417536, "grad_norm": 9.6015625, "learning_rate": 8.923891415577034e-06, "loss": 0.4996, "step": 30786 }, { "epoch": 3.2759097680357523, "grad_norm": 16.09375, "learning_rate": 8.923855884025015e-06, "loss": 0.7561, "step": 30787 }, { "epoch": 3.276016173653969, "grad_norm": 11.828125, "learning_rate": 8.923820352472997e-06, "loss": 0.5467, "step": 30788 }, { "epoch": 3.2761225792721858, "grad_norm": 10.328125, "learning_rate": 8.923784820920979e-06, "loss": 0.605, "step": 30789 }, { "epoch": 3.2762289848904023, "grad_norm": 14.78125, "learning_rate": 8.92374928936896e-06, "loss": 0.7117, "step": 30790 }, { "epoch": 3.2763353905086188, "grad_norm": 10.421875, "learning_rate": 8.923713757816942e-06, "loss": 0.578, "step": 30791 }, { "epoch": 3.2764417961268357, "grad_norm": 18.03125, "learning_rate": 8.923678226264923e-06, "loss": 0.6679, "step": 30792 }, { "epoch": 3.2765482017450522, "grad_norm": 9.9453125, "learning_rate": 8.923642694712907e-06, "loss": 0.5087, "step": 30793 }, { "epoch": 3.2766546073632687, "grad_norm": 9.9140625, "learning_rate": 8.923607163160888e-06, "loss": 0.6443, "step": 30794 }, { "epoch": 3.2767610129814853, "grad_norm": 10.7890625, "learning_rate": 8.92357163160887e-06, "loss": 0.6931, "step": 30795 }, { "epoch": 3.276867418599702, "grad_norm": 11.6171875, "learning_rate": 8.923536100056851e-06, "loss": 0.5385, "step": 30796 }, { "epoch": 3.2769738242179187, "grad_norm": 15.0, "learning_rate": 8.923500568504833e-06, "loss": 0.7304, "step": 30797 }, { "epoch": 3.2770802298361352, "grad_norm": 14.4921875, "learning_rate": 8.923465036952815e-06, "loss": 0.6872, "step": 30798 }, { "epoch": 3.277186635454352, "grad_norm": 11.5234375, "learning_rate": 8.923429505400796e-06, "loss": 0.509, "step": 30799 }, { "epoch": 3.2772930410725687, "grad_norm": 11.9921875, "learning_rate": 8.923393973848778e-06, "loss": 0.6717, "step": 30800 }, { "epoch": 3.277399446690785, "grad_norm": 12.3203125, "learning_rate": 8.92335844229676e-06, "loss": 0.593, "step": 30801 }, { "epoch": 3.2775058523090017, "grad_norm": 9.9296875, "learning_rate": 8.923322910744741e-06, "loss": 0.6103, "step": 30802 }, { "epoch": 3.2776122579272187, "grad_norm": 13.6484375, "learning_rate": 8.923287379192724e-06, "loss": 0.8269, "step": 30803 }, { "epoch": 3.277718663545435, "grad_norm": 15.7265625, "learning_rate": 8.923251847640706e-06, "loss": 0.5811, "step": 30804 }, { "epoch": 3.2778250691636517, "grad_norm": 11.734375, "learning_rate": 8.923216316088688e-06, "loss": 0.5822, "step": 30805 }, { "epoch": 3.2779314747818686, "grad_norm": 14.4453125, "learning_rate": 8.92318078453667e-06, "loss": 0.6726, "step": 30806 }, { "epoch": 3.278037880400085, "grad_norm": 14.3046875, "learning_rate": 8.92314525298465e-06, "loss": 0.6653, "step": 30807 }, { "epoch": 3.2781442860183017, "grad_norm": 11.9375, "learning_rate": 8.923109721432634e-06, "loss": 0.6812, "step": 30808 }, { "epoch": 3.2782506916365186, "grad_norm": 15.5078125, "learning_rate": 8.923074189880616e-06, "loss": 0.5848, "step": 30809 }, { "epoch": 3.278357097254735, "grad_norm": 11.90625, "learning_rate": 8.923038658328597e-06, "loss": 0.5367, "step": 30810 }, { "epoch": 3.2784635028729516, "grad_norm": 10.5703125, "learning_rate": 8.923003126776579e-06, "loss": 0.5348, "step": 30811 }, { "epoch": 3.2785699084911686, "grad_norm": 10.2734375, "learning_rate": 8.92296759522456e-06, "loss": 0.4739, "step": 30812 }, { "epoch": 3.278676314109385, "grad_norm": 13.8125, "learning_rate": 8.922932063672542e-06, "loss": 0.5837, "step": 30813 }, { "epoch": 3.2787827197276016, "grad_norm": 12.359375, "learning_rate": 8.922896532120524e-06, "loss": 0.6314, "step": 30814 }, { "epoch": 3.278889125345818, "grad_norm": 13.9609375, "learning_rate": 8.922861000568505e-06, "loss": 0.6561, "step": 30815 }, { "epoch": 3.278995530964035, "grad_norm": 11.171875, "learning_rate": 8.922825469016487e-06, "loss": 0.5799, "step": 30816 }, { "epoch": 3.2791019365822516, "grad_norm": 15.3046875, "learning_rate": 8.922789937464469e-06, "loss": 0.7469, "step": 30817 }, { "epoch": 3.279208342200468, "grad_norm": 11.6484375, "learning_rate": 8.92275440591245e-06, "loss": 0.6, "step": 30818 }, { "epoch": 3.2793147478186846, "grad_norm": 11.0703125, "learning_rate": 8.922718874360433e-06, "loss": 0.4912, "step": 30819 }, { "epoch": 3.2794211534369015, "grad_norm": 12.625, "learning_rate": 8.922683342808415e-06, "loss": 0.5495, "step": 30820 }, { "epoch": 3.279527559055118, "grad_norm": 13.984375, "learning_rate": 8.922647811256397e-06, "loss": 0.6897, "step": 30821 }, { "epoch": 3.2796339646733346, "grad_norm": 12.265625, "learning_rate": 8.922612279704378e-06, "loss": 0.588, "step": 30822 }, { "epoch": 3.2797403702915515, "grad_norm": 12.1875, "learning_rate": 8.92257674815236e-06, "loss": 0.5885, "step": 30823 }, { "epoch": 3.279846775909768, "grad_norm": 15.734375, "learning_rate": 8.922541216600342e-06, "loss": 0.7616, "step": 30824 }, { "epoch": 3.2799531815279845, "grad_norm": 12.078125, "learning_rate": 8.922505685048323e-06, "loss": 0.5946, "step": 30825 }, { "epoch": 3.2800595871462015, "grad_norm": 12.046875, "learning_rate": 8.922470153496305e-06, "loss": 0.6048, "step": 30826 }, { "epoch": 3.280165992764418, "grad_norm": 15.03125, "learning_rate": 8.922434621944286e-06, "loss": 0.6394, "step": 30827 }, { "epoch": 3.2802723983826345, "grad_norm": 13.984375, "learning_rate": 8.922399090392268e-06, "loss": 0.6068, "step": 30828 }, { "epoch": 3.2803788040008515, "grad_norm": 12.7734375, "learning_rate": 8.922363558840251e-06, "loss": 0.5951, "step": 30829 }, { "epoch": 3.280485209619068, "grad_norm": 10.6875, "learning_rate": 8.922328027288233e-06, "loss": 0.5029, "step": 30830 }, { "epoch": 3.2805916152372845, "grad_norm": 15.3125, "learning_rate": 8.922292495736214e-06, "loss": 0.776, "step": 30831 }, { "epoch": 3.280698020855501, "grad_norm": 13.9609375, "learning_rate": 8.922256964184196e-06, "loss": 0.7258, "step": 30832 }, { "epoch": 3.280804426473718, "grad_norm": 21.078125, "learning_rate": 8.922221432632178e-06, "loss": 0.7324, "step": 30833 }, { "epoch": 3.2809108320919345, "grad_norm": 12.484375, "learning_rate": 8.922185901080161e-06, "loss": 0.6863, "step": 30834 }, { "epoch": 3.281017237710151, "grad_norm": 13.59375, "learning_rate": 8.922150369528143e-06, "loss": 0.5911, "step": 30835 }, { "epoch": 3.281123643328368, "grad_norm": 12.4921875, "learning_rate": 8.922114837976124e-06, "loss": 0.5933, "step": 30836 }, { "epoch": 3.2812300489465844, "grad_norm": 11.6171875, "learning_rate": 8.922079306424106e-06, "loss": 0.6337, "step": 30837 }, { "epoch": 3.281336454564801, "grad_norm": 12.8359375, "learning_rate": 8.922043774872087e-06, "loss": 0.6368, "step": 30838 }, { "epoch": 3.2814428601830175, "grad_norm": 12.8359375, "learning_rate": 8.922008243320069e-06, "loss": 0.6322, "step": 30839 }, { "epoch": 3.2815492658012344, "grad_norm": 8.890625, "learning_rate": 8.92197271176805e-06, "loss": 0.4055, "step": 30840 }, { "epoch": 3.281655671419451, "grad_norm": 18.265625, "learning_rate": 8.921937180216032e-06, "loss": 0.7564, "step": 30841 }, { "epoch": 3.2817620770376674, "grad_norm": 11.984375, "learning_rate": 8.921901648664014e-06, "loss": 0.6034, "step": 30842 }, { "epoch": 3.2818684826558844, "grad_norm": 13.625, "learning_rate": 8.921866117111995e-06, "loss": 0.5221, "step": 30843 }, { "epoch": 3.281974888274101, "grad_norm": 11.5859375, "learning_rate": 8.921830585559979e-06, "loss": 0.4772, "step": 30844 }, { "epoch": 3.2820812938923174, "grad_norm": 15.875, "learning_rate": 8.92179505400796e-06, "loss": 0.6757, "step": 30845 }, { "epoch": 3.2821876995105344, "grad_norm": 13.15625, "learning_rate": 8.921759522455942e-06, "loss": 0.6479, "step": 30846 }, { "epoch": 3.282294105128751, "grad_norm": 15.5859375, "learning_rate": 8.921723990903924e-06, "loss": 0.6868, "step": 30847 }, { "epoch": 3.2824005107469674, "grad_norm": 10.421875, "learning_rate": 8.921688459351905e-06, "loss": 0.4619, "step": 30848 }, { "epoch": 3.2825069163651843, "grad_norm": 13.8125, "learning_rate": 8.921652927799887e-06, "loss": 0.6678, "step": 30849 }, { "epoch": 3.282613321983401, "grad_norm": 17.484375, "learning_rate": 8.921617396247868e-06, "loss": 0.6539, "step": 30850 }, { "epoch": 3.2827197276016173, "grad_norm": 11.1796875, "learning_rate": 8.92158186469585e-06, "loss": 0.5179, "step": 30851 }, { "epoch": 3.282826133219834, "grad_norm": 16.5625, "learning_rate": 8.921546333143832e-06, "loss": 0.6327, "step": 30852 }, { "epoch": 3.282932538838051, "grad_norm": 10.9921875, "learning_rate": 8.921510801591813e-06, "loss": 0.4571, "step": 30853 }, { "epoch": 3.2830389444562673, "grad_norm": 11.90625, "learning_rate": 8.921475270039795e-06, "loss": 0.5294, "step": 30854 }, { "epoch": 3.283145350074484, "grad_norm": 15.90625, "learning_rate": 8.921439738487778e-06, "loss": 0.6223, "step": 30855 }, { "epoch": 3.2832517556927003, "grad_norm": 21.3125, "learning_rate": 8.92140420693576e-06, "loss": 0.6956, "step": 30856 }, { "epoch": 3.2833581613109173, "grad_norm": 21.484375, "learning_rate": 8.921368675383741e-06, "loss": 0.7763, "step": 30857 }, { "epoch": 3.283464566929134, "grad_norm": 15.7578125, "learning_rate": 8.921333143831723e-06, "loss": 0.7049, "step": 30858 }, { "epoch": 3.2835709725473503, "grad_norm": 14.203125, "learning_rate": 8.921297612279705e-06, "loss": 0.7583, "step": 30859 }, { "epoch": 3.2836773781655673, "grad_norm": 14.875, "learning_rate": 8.921262080727688e-06, "loss": 0.6098, "step": 30860 }, { "epoch": 3.2837837837837838, "grad_norm": 15.90625, "learning_rate": 8.92122654917567e-06, "loss": 0.6045, "step": 30861 }, { "epoch": 3.2838901894020003, "grad_norm": 14.96875, "learning_rate": 8.921191017623651e-06, "loss": 0.6535, "step": 30862 }, { "epoch": 3.2839965950202172, "grad_norm": 11.859375, "learning_rate": 8.921155486071633e-06, "loss": 0.589, "step": 30863 }, { "epoch": 3.2841030006384337, "grad_norm": 12.84375, "learning_rate": 8.921119954519614e-06, "loss": 0.5695, "step": 30864 }, { "epoch": 3.2842094062566503, "grad_norm": 10.8828125, "learning_rate": 8.921084422967596e-06, "loss": 0.5461, "step": 30865 }, { "epoch": 3.284315811874867, "grad_norm": 15.8359375, "learning_rate": 8.921048891415577e-06, "loss": 0.7443, "step": 30866 }, { "epoch": 3.2844222174930837, "grad_norm": 17.59375, "learning_rate": 8.921013359863559e-06, "loss": 0.73, "step": 30867 }, { "epoch": 3.2845286231113002, "grad_norm": 15.5234375, "learning_rate": 8.92097782831154e-06, "loss": 0.7327, "step": 30868 }, { "epoch": 3.2846350287295167, "grad_norm": 12.28125, "learning_rate": 8.920942296759522e-06, "loss": 0.7626, "step": 30869 }, { "epoch": 3.2847414343477337, "grad_norm": 15.40625, "learning_rate": 8.920906765207506e-06, "loss": 0.6948, "step": 30870 }, { "epoch": 3.28484783996595, "grad_norm": 26.4375, "learning_rate": 8.920871233655487e-06, "loss": 0.8134, "step": 30871 }, { "epoch": 3.2849542455841667, "grad_norm": 11.1171875, "learning_rate": 8.920835702103469e-06, "loss": 0.5033, "step": 30872 }, { "epoch": 3.2850606512023837, "grad_norm": 13.265625, "learning_rate": 8.92080017055145e-06, "loss": 0.6292, "step": 30873 }, { "epoch": 3.2851670568206, "grad_norm": 13.046875, "learning_rate": 8.920764638999432e-06, "loss": 0.6693, "step": 30874 }, { "epoch": 3.2852734624388167, "grad_norm": 12.0234375, "learning_rate": 8.920729107447415e-06, "loss": 0.6341, "step": 30875 }, { "epoch": 3.285379868057033, "grad_norm": 13.734375, "learning_rate": 8.920693575895397e-06, "loss": 0.6934, "step": 30876 }, { "epoch": 3.28548627367525, "grad_norm": 12.09375, "learning_rate": 8.920658044343379e-06, "loss": 0.5573, "step": 30877 }, { "epoch": 3.2855926792934667, "grad_norm": 13.0703125, "learning_rate": 8.920622512791358e-06, "loss": 0.6811, "step": 30878 }, { "epoch": 3.285699084911683, "grad_norm": 13.5390625, "learning_rate": 8.92058698123934e-06, "loss": 0.6383, "step": 30879 }, { "epoch": 3.2858054905299, "grad_norm": 14.1328125, "learning_rate": 8.920551449687323e-06, "loss": 0.7182, "step": 30880 }, { "epoch": 3.2859118961481166, "grad_norm": 12.9296875, "learning_rate": 8.920515918135305e-06, "loss": 0.4461, "step": 30881 }, { "epoch": 3.286018301766333, "grad_norm": 18.359375, "learning_rate": 8.920480386583287e-06, "loss": 0.7746, "step": 30882 }, { "epoch": 3.28612470738455, "grad_norm": 14.3359375, "learning_rate": 8.920444855031268e-06, "loss": 0.7399, "step": 30883 }, { "epoch": 3.2862311130027666, "grad_norm": 14.8359375, "learning_rate": 8.92040932347925e-06, "loss": 0.9178, "step": 30884 }, { "epoch": 3.286337518620983, "grad_norm": 21.953125, "learning_rate": 8.920373791927233e-06, "loss": 0.7808, "step": 30885 }, { "epoch": 3.2864439242392, "grad_norm": 13.9765625, "learning_rate": 8.920338260375215e-06, "loss": 0.5001, "step": 30886 }, { "epoch": 3.2865503298574166, "grad_norm": 14.9453125, "learning_rate": 8.920302728823196e-06, "loss": 0.5469, "step": 30887 }, { "epoch": 3.286656735475633, "grad_norm": 13.015625, "learning_rate": 8.920267197271178e-06, "loss": 0.6253, "step": 30888 }, { "epoch": 3.2867631410938496, "grad_norm": 13.4921875, "learning_rate": 8.92023166571916e-06, "loss": 0.6835, "step": 30889 }, { "epoch": 3.2868695467120665, "grad_norm": 17.078125, "learning_rate": 8.920196134167141e-06, "loss": 0.6567, "step": 30890 }, { "epoch": 3.286975952330283, "grad_norm": 19.5, "learning_rate": 8.920160602615123e-06, "loss": 0.778, "step": 30891 }, { "epoch": 3.2870823579484996, "grad_norm": 15.8125, "learning_rate": 8.920125071063104e-06, "loss": 0.5331, "step": 30892 }, { "epoch": 3.287188763566716, "grad_norm": 12.4140625, "learning_rate": 8.920089539511086e-06, "loss": 0.5261, "step": 30893 }, { "epoch": 3.287295169184933, "grad_norm": 15.0078125, "learning_rate": 8.920054007959068e-06, "loss": 0.6076, "step": 30894 }, { "epoch": 3.2874015748031495, "grad_norm": 13.859375, "learning_rate": 8.920018476407049e-06, "loss": 0.6838, "step": 30895 }, { "epoch": 3.287507980421366, "grad_norm": 13.078125, "learning_rate": 8.919982944855032e-06, "loss": 0.5325, "step": 30896 }, { "epoch": 3.287614386039583, "grad_norm": 12.3515625, "learning_rate": 8.919947413303014e-06, "loss": 0.5473, "step": 30897 }, { "epoch": 3.2877207916577995, "grad_norm": 13.96875, "learning_rate": 8.919911881750996e-06, "loss": 0.7543, "step": 30898 }, { "epoch": 3.287827197276016, "grad_norm": 13.6875, "learning_rate": 8.919876350198977e-06, "loss": 0.6098, "step": 30899 }, { "epoch": 3.287933602894233, "grad_norm": 13.9375, "learning_rate": 8.919840818646959e-06, "loss": 0.7001, "step": 30900 }, { "epoch": 3.2880400085124495, "grad_norm": 13.515625, "learning_rate": 8.919805287094942e-06, "loss": 0.557, "step": 30901 }, { "epoch": 3.288146414130666, "grad_norm": 9.0078125, "learning_rate": 8.919769755542924e-06, "loss": 0.4832, "step": 30902 }, { "epoch": 3.288252819748883, "grad_norm": 8.578125, "learning_rate": 8.919734223990905e-06, "loss": 0.3971, "step": 30903 }, { "epoch": 3.2883592253670995, "grad_norm": 11.0859375, "learning_rate": 8.919698692438887e-06, "loss": 0.6945, "step": 30904 }, { "epoch": 3.288465630985316, "grad_norm": 12.5234375, "learning_rate": 8.919663160886869e-06, "loss": 0.6667, "step": 30905 }, { "epoch": 3.2885720366035325, "grad_norm": 15.390625, "learning_rate": 8.91962762933485e-06, "loss": 0.6859, "step": 30906 }, { "epoch": 3.2886784422217494, "grad_norm": 12.1796875, "learning_rate": 8.919592097782832e-06, "loss": 0.4799, "step": 30907 }, { "epoch": 3.288784847839966, "grad_norm": 9.8203125, "learning_rate": 8.919556566230813e-06, "loss": 0.5554, "step": 30908 }, { "epoch": 3.2888912534581825, "grad_norm": 13.9140625, "learning_rate": 8.919521034678795e-06, "loss": 0.7245, "step": 30909 }, { "epoch": 3.2889976590763994, "grad_norm": 10.9765625, "learning_rate": 8.919485503126777e-06, "loss": 0.5943, "step": 30910 }, { "epoch": 3.289104064694616, "grad_norm": 12.921875, "learning_rate": 8.91944997157476e-06, "loss": 0.611, "step": 30911 }, { "epoch": 3.2892104703128324, "grad_norm": 8.65625, "learning_rate": 8.919414440022742e-06, "loss": 0.4752, "step": 30912 }, { "epoch": 3.289316875931049, "grad_norm": 10.8359375, "learning_rate": 8.919378908470723e-06, "loss": 0.6843, "step": 30913 }, { "epoch": 3.289423281549266, "grad_norm": 10.5546875, "learning_rate": 8.919343376918705e-06, "loss": 0.531, "step": 30914 }, { "epoch": 3.2895296871674824, "grad_norm": 9.6640625, "learning_rate": 8.919307845366686e-06, "loss": 0.463, "step": 30915 }, { "epoch": 3.289636092785699, "grad_norm": 15.5703125, "learning_rate": 8.919272313814668e-06, "loss": 0.6268, "step": 30916 }, { "epoch": 3.289742498403916, "grad_norm": 13.2421875, "learning_rate": 8.91923678226265e-06, "loss": 0.6573, "step": 30917 }, { "epoch": 3.2898489040221324, "grad_norm": 14.28125, "learning_rate": 8.919201250710631e-06, "loss": 0.6605, "step": 30918 }, { "epoch": 3.289955309640349, "grad_norm": 14.03125, "learning_rate": 8.919165719158613e-06, "loss": 0.6176, "step": 30919 }, { "epoch": 3.290061715258566, "grad_norm": 17.21875, "learning_rate": 8.919130187606594e-06, "loss": 0.8495, "step": 30920 }, { "epoch": 3.2901681208767823, "grad_norm": 13.6015625, "learning_rate": 8.919094656054578e-06, "loss": 0.6429, "step": 30921 }, { "epoch": 3.290274526494999, "grad_norm": 22.8125, "learning_rate": 8.91905912450256e-06, "loss": 0.8411, "step": 30922 }, { "epoch": 3.290380932113216, "grad_norm": 12.828125, "learning_rate": 8.919023592950541e-06, "loss": 0.5136, "step": 30923 }, { "epoch": 3.2904873377314323, "grad_norm": 13.46875, "learning_rate": 8.918988061398523e-06, "loss": 0.6138, "step": 30924 }, { "epoch": 3.290593743349649, "grad_norm": 17.625, "learning_rate": 8.918952529846504e-06, "loss": 0.7502, "step": 30925 }, { "epoch": 3.2907001489678653, "grad_norm": 16.609375, "learning_rate": 8.918916998294487e-06, "loss": 0.7647, "step": 30926 }, { "epoch": 3.2908065545860823, "grad_norm": 12.0, "learning_rate": 8.918881466742469e-06, "loss": 0.5248, "step": 30927 }, { "epoch": 3.290912960204299, "grad_norm": 12.7421875, "learning_rate": 8.91884593519045e-06, "loss": 0.6158, "step": 30928 }, { "epoch": 3.2910193658225153, "grad_norm": 14.4765625, "learning_rate": 8.918810403638432e-06, "loss": 0.6199, "step": 30929 }, { "epoch": 3.2911257714407323, "grad_norm": 21.421875, "learning_rate": 8.918774872086414e-06, "loss": 0.834, "step": 30930 }, { "epoch": 3.2912321770589488, "grad_norm": 10.1171875, "learning_rate": 8.918739340534395e-06, "loss": 0.5208, "step": 30931 }, { "epoch": 3.2913385826771653, "grad_norm": 12.25, "learning_rate": 8.918703808982377e-06, "loss": 0.6494, "step": 30932 }, { "epoch": 3.291444988295382, "grad_norm": 12.09375, "learning_rate": 8.918668277430359e-06, "loss": 0.6299, "step": 30933 }, { "epoch": 3.2915513939135987, "grad_norm": 15.421875, "learning_rate": 8.91863274587834e-06, "loss": 0.6982, "step": 30934 }, { "epoch": 3.2916577995318153, "grad_norm": 12.3125, "learning_rate": 8.918597214326322e-06, "loss": 0.5787, "step": 30935 }, { "epoch": 3.2917642051500318, "grad_norm": 13.7265625, "learning_rate": 8.918561682774303e-06, "loss": 0.5821, "step": 30936 }, { "epoch": 3.2918706107682487, "grad_norm": 14.5390625, "learning_rate": 8.918526151222287e-06, "loss": 0.8546, "step": 30937 }, { "epoch": 3.2919770163864652, "grad_norm": 11.3203125, "learning_rate": 8.918490619670268e-06, "loss": 0.7509, "step": 30938 }, { "epoch": 3.2920834220046817, "grad_norm": 9.3125, "learning_rate": 8.91845508811825e-06, "loss": 0.4674, "step": 30939 }, { "epoch": 3.2921898276228987, "grad_norm": 15.4765625, "learning_rate": 8.918419556566232e-06, "loss": 0.6388, "step": 30940 }, { "epoch": 3.292296233241115, "grad_norm": 10.5390625, "learning_rate": 8.918384025014213e-06, "loss": 0.4397, "step": 30941 }, { "epoch": 3.2924026388593317, "grad_norm": 15.1796875, "learning_rate": 8.918348493462195e-06, "loss": 0.6312, "step": 30942 }, { "epoch": 3.2925090444775487, "grad_norm": 12.3203125, "learning_rate": 8.918312961910176e-06, "loss": 0.5895, "step": 30943 }, { "epoch": 3.292615450095765, "grad_norm": 18.140625, "learning_rate": 8.918277430358158e-06, "loss": 0.6856, "step": 30944 }, { "epoch": 3.2927218557139817, "grad_norm": 11.1484375, "learning_rate": 8.91824189880614e-06, "loss": 0.4794, "step": 30945 }, { "epoch": 3.292828261332198, "grad_norm": 13.9609375, "learning_rate": 8.918206367254121e-06, "loss": 0.6982, "step": 30946 }, { "epoch": 3.292934666950415, "grad_norm": 14.453125, "learning_rate": 8.918170835702105e-06, "loss": 0.6373, "step": 30947 }, { "epoch": 3.2930410725686317, "grad_norm": 15.5390625, "learning_rate": 8.918135304150086e-06, "loss": 0.5856, "step": 30948 }, { "epoch": 3.293147478186848, "grad_norm": 11.4921875, "learning_rate": 8.918099772598068e-06, "loss": 0.7244, "step": 30949 }, { "epoch": 3.2932538838050647, "grad_norm": 12.859375, "learning_rate": 8.91806424104605e-06, "loss": 0.7336, "step": 30950 }, { "epoch": 3.2933602894232816, "grad_norm": 11.1875, "learning_rate": 8.918028709494031e-06, "loss": 0.5693, "step": 30951 }, { "epoch": 3.293466695041498, "grad_norm": 12.171875, "learning_rate": 8.917993177942014e-06, "loss": 0.5532, "step": 30952 }, { "epoch": 3.2935731006597146, "grad_norm": 18.71875, "learning_rate": 8.917957646389996e-06, "loss": 0.6263, "step": 30953 }, { "epoch": 3.2936795062779316, "grad_norm": 13.6875, "learning_rate": 8.917922114837977e-06, "loss": 0.6469, "step": 30954 }, { "epoch": 3.293785911896148, "grad_norm": 16.75, "learning_rate": 8.917886583285959e-06, "loss": 0.6816, "step": 30955 }, { "epoch": 3.2938923175143646, "grad_norm": 10.7734375, "learning_rate": 8.91785105173394e-06, "loss": 0.5313, "step": 30956 }, { "epoch": 3.2939987231325816, "grad_norm": 19.78125, "learning_rate": 8.917815520181922e-06, "loss": 0.7217, "step": 30957 }, { "epoch": 3.294105128750798, "grad_norm": 13.921875, "learning_rate": 8.917779988629904e-06, "loss": 0.744, "step": 30958 }, { "epoch": 3.2942115343690146, "grad_norm": 11.0703125, "learning_rate": 8.917744457077886e-06, "loss": 0.6552, "step": 30959 }, { "epoch": 3.2943179399872315, "grad_norm": 20.1875, "learning_rate": 8.917708925525867e-06, "loss": 0.8057, "step": 30960 }, { "epoch": 3.294424345605448, "grad_norm": 12.8359375, "learning_rate": 8.917673393973849e-06, "loss": 0.5269, "step": 30961 }, { "epoch": 3.2945307512236646, "grad_norm": 11.890625, "learning_rate": 8.917637862421832e-06, "loss": 0.5305, "step": 30962 }, { "epoch": 3.294637156841881, "grad_norm": 9.8203125, "learning_rate": 8.917602330869814e-06, "loss": 0.4649, "step": 30963 }, { "epoch": 3.294743562460098, "grad_norm": 18.1875, "learning_rate": 8.917566799317795e-06, "loss": 0.7144, "step": 30964 }, { "epoch": 3.2948499680783145, "grad_norm": 11.203125, "learning_rate": 8.917531267765777e-06, "loss": 0.6036, "step": 30965 }, { "epoch": 3.294956373696531, "grad_norm": 13.890625, "learning_rate": 8.917495736213758e-06, "loss": 0.5548, "step": 30966 }, { "epoch": 3.295062779314748, "grad_norm": 10.4921875, "learning_rate": 8.91746020466174e-06, "loss": 0.4729, "step": 30967 }, { "epoch": 3.2951691849329645, "grad_norm": 11.3515625, "learning_rate": 8.917424673109722e-06, "loss": 0.6742, "step": 30968 }, { "epoch": 3.295275590551181, "grad_norm": 12.640625, "learning_rate": 8.917389141557703e-06, "loss": 0.7221, "step": 30969 }, { "epoch": 3.2953819961693975, "grad_norm": 12.4296875, "learning_rate": 8.917353610005685e-06, "loss": 0.6204, "step": 30970 }, { "epoch": 3.2954884017876145, "grad_norm": 12.46875, "learning_rate": 8.917318078453666e-06, "loss": 0.6679, "step": 30971 }, { "epoch": 3.295594807405831, "grad_norm": 15.3828125, "learning_rate": 8.917282546901648e-06, "loss": 0.698, "step": 30972 }, { "epoch": 3.2957012130240475, "grad_norm": 14.7890625, "learning_rate": 8.917247015349631e-06, "loss": 0.7358, "step": 30973 }, { "epoch": 3.2958076186422645, "grad_norm": 12.546875, "learning_rate": 8.917211483797613e-06, "loss": 0.6789, "step": 30974 }, { "epoch": 3.295914024260481, "grad_norm": 14.171875, "learning_rate": 8.917175952245595e-06, "loss": 0.6983, "step": 30975 }, { "epoch": 3.2960204298786975, "grad_norm": 11.125, "learning_rate": 8.917140420693576e-06, "loss": 0.6016, "step": 30976 }, { "epoch": 3.2961268354969144, "grad_norm": 13.6875, "learning_rate": 8.917104889141558e-06, "loss": 0.6029, "step": 30977 }, { "epoch": 3.296233241115131, "grad_norm": 14.265625, "learning_rate": 8.917069357589541e-06, "loss": 0.6905, "step": 30978 }, { "epoch": 3.2963396467333475, "grad_norm": 13.2421875, "learning_rate": 8.917033826037523e-06, "loss": 0.6129, "step": 30979 }, { "epoch": 3.2964460523515644, "grad_norm": 11.6796875, "learning_rate": 8.916998294485504e-06, "loss": 0.5589, "step": 30980 }, { "epoch": 3.296552457969781, "grad_norm": 12.1875, "learning_rate": 8.916962762933486e-06, "loss": 0.511, "step": 30981 }, { "epoch": 3.2966588635879974, "grad_norm": 14.1640625, "learning_rate": 8.916927231381468e-06, "loss": 0.616, "step": 30982 }, { "epoch": 3.296765269206214, "grad_norm": 13.921875, "learning_rate": 8.91689169982945e-06, "loss": 0.6643, "step": 30983 }, { "epoch": 3.296871674824431, "grad_norm": 16.109375, "learning_rate": 8.91685616827743e-06, "loss": 0.6013, "step": 30984 }, { "epoch": 3.2969780804426474, "grad_norm": 12.3125, "learning_rate": 8.916820636725412e-06, "loss": 0.6634, "step": 30985 }, { "epoch": 3.297084486060864, "grad_norm": 18.5625, "learning_rate": 8.916785105173394e-06, "loss": 0.7286, "step": 30986 }, { "epoch": 3.2971908916790804, "grad_norm": 14.4609375, "learning_rate": 8.916749573621376e-06, "loss": 0.6762, "step": 30987 }, { "epoch": 3.2972972972972974, "grad_norm": 13.46875, "learning_rate": 8.916714042069359e-06, "loss": 0.5985, "step": 30988 }, { "epoch": 3.297403702915514, "grad_norm": 12.625, "learning_rate": 8.91667851051734e-06, "loss": 0.473, "step": 30989 }, { "epoch": 3.2975101085337304, "grad_norm": 14.140625, "learning_rate": 8.916642978965322e-06, "loss": 0.6946, "step": 30990 }, { "epoch": 3.2976165141519473, "grad_norm": 12.4453125, "learning_rate": 8.916607447413304e-06, "loss": 0.6056, "step": 30991 }, { "epoch": 3.297722919770164, "grad_norm": 11.3671875, "learning_rate": 8.916571915861285e-06, "loss": 0.5993, "step": 30992 }, { "epoch": 3.2978293253883804, "grad_norm": 13.03125, "learning_rate": 8.916536384309269e-06, "loss": 0.5774, "step": 30993 }, { "epoch": 3.2979357310065973, "grad_norm": 16.515625, "learning_rate": 8.91650085275725e-06, "loss": 0.7571, "step": 30994 }, { "epoch": 3.298042136624814, "grad_norm": 13.5703125, "learning_rate": 8.916465321205232e-06, "loss": 0.5641, "step": 30995 }, { "epoch": 3.2981485422430303, "grad_norm": 10.4921875, "learning_rate": 8.916429789653213e-06, "loss": 0.5944, "step": 30996 }, { "epoch": 3.2982549478612473, "grad_norm": 16.21875, "learning_rate": 8.916394258101193e-06, "loss": 0.6092, "step": 30997 }, { "epoch": 3.298361353479464, "grad_norm": 11.328125, "learning_rate": 8.916358726549177e-06, "loss": 0.6134, "step": 30998 }, { "epoch": 3.2984677590976803, "grad_norm": 13.1328125, "learning_rate": 8.916323194997158e-06, "loss": 0.6848, "step": 30999 }, { "epoch": 3.298574164715897, "grad_norm": 14.8125, "learning_rate": 8.91628766344514e-06, "loss": 0.576, "step": 31000 }, { "epoch": 3.2986805703341138, "grad_norm": 10.9296875, "learning_rate": 8.916252131893121e-06, "loss": 0.5618, "step": 31001 }, { "epoch": 3.2987869759523303, "grad_norm": 8.5234375, "learning_rate": 8.916216600341103e-06, "loss": 0.4278, "step": 31002 }, { "epoch": 3.298893381570547, "grad_norm": 12.4375, "learning_rate": 8.916181068789086e-06, "loss": 0.5011, "step": 31003 }, { "epoch": 3.2989997871887637, "grad_norm": 12.2421875, "learning_rate": 8.916145537237068e-06, "loss": 0.4739, "step": 31004 }, { "epoch": 3.2991061928069803, "grad_norm": 12.390625, "learning_rate": 8.91611000568505e-06, "loss": 0.5785, "step": 31005 }, { "epoch": 3.2992125984251968, "grad_norm": 16.328125, "learning_rate": 8.916074474133031e-06, "loss": 0.7988, "step": 31006 }, { "epoch": 3.2993190040434133, "grad_norm": 13.484375, "learning_rate": 8.916038942581013e-06, "loss": 0.7671, "step": 31007 }, { "epoch": 3.2994254096616302, "grad_norm": 13.0703125, "learning_rate": 8.916003411028994e-06, "loss": 0.5322, "step": 31008 }, { "epoch": 3.2995318152798467, "grad_norm": 16.625, "learning_rate": 8.915967879476976e-06, "loss": 0.7706, "step": 31009 }, { "epoch": 3.2996382208980632, "grad_norm": 20.390625, "learning_rate": 8.915932347924958e-06, "loss": 0.7352, "step": 31010 }, { "epoch": 3.29974462651628, "grad_norm": 15.046875, "learning_rate": 8.91589681637294e-06, "loss": 0.6126, "step": 31011 }, { "epoch": 3.2998510321344967, "grad_norm": 19.25, "learning_rate": 8.91586128482092e-06, "loss": 0.8604, "step": 31012 }, { "epoch": 3.299957437752713, "grad_norm": 14.84375, "learning_rate": 8.915825753268902e-06, "loss": 0.6065, "step": 31013 }, { "epoch": 3.30006384337093, "grad_norm": 12.8125, "learning_rate": 8.915790221716886e-06, "loss": 0.6302, "step": 31014 }, { "epoch": 3.3001702489891467, "grad_norm": 18.6875, "learning_rate": 8.915754690164867e-06, "loss": 0.6426, "step": 31015 }, { "epoch": 3.300276654607363, "grad_norm": 11.0625, "learning_rate": 8.915719158612849e-06, "loss": 0.6515, "step": 31016 }, { "epoch": 3.30038306022558, "grad_norm": 10.3125, "learning_rate": 8.91568362706083e-06, "loss": 0.4903, "step": 31017 }, { "epoch": 3.3004894658437967, "grad_norm": 8.9453125, "learning_rate": 8.915648095508812e-06, "loss": 0.5742, "step": 31018 }, { "epoch": 3.300595871462013, "grad_norm": 14.3125, "learning_rate": 8.915612563956795e-06, "loss": 0.7549, "step": 31019 }, { "epoch": 3.3007022770802297, "grad_norm": 10.515625, "learning_rate": 8.915577032404777e-06, "loss": 0.5562, "step": 31020 }, { "epoch": 3.3008086826984466, "grad_norm": 18.296875, "learning_rate": 8.915541500852759e-06, "loss": 0.7667, "step": 31021 }, { "epoch": 3.300915088316663, "grad_norm": 12.3359375, "learning_rate": 8.91550596930074e-06, "loss": 0.707, "step": 31022 }, { "epoch": 3.3010214939348796, "grad_norm": 13.6171875, "learning_rate": 8.915470437748722e-06, "loss": 0.7587, "step": 31023 }, { "epoch": 3.301127899553096, "grad_norm": 15.3515625, "learning_rate": 8.915434906196703e-06, "loss": 0.8134, "step": 31024 }, { "epoch": 3.301234305171313, "grad_norm": 11.5390625, "learning_rate": 8.915399374644685e-06, "loss": 0.4674, "step": 31025 }, { "epoch": 3.3013407107895296, "grad_norm": 11.140625, "learning_rate": 8.915363843092667e-06, "loss": 0.6406, "step": 31026 }, { "epoch": 3.301447116407746, "grad_norm": 15.1015625, "learning_rate": 8.915328311540648e-06, "loss": 0.6456, "step": 31027 }, { "epoch": 3.301553522025963, "grad_norm": 13.9375, "learning_rate": 8.91529277998863e-06, "loss": 0.6898, "step": 31028 }, { "epoch": 3.3016599276441796, "grad_norm": 12.671875, "learning_rate": 8.915257248436613e-06, "loss": 0.6863, "step": 31029 }, { "epoch": 3.301766333262396, "grad_norm": 13.09375, "learning_rate": 8.915221716884595e-06, "loss": 0.5679, "step": 31030 }, { "epoch": 3.301872738880613, "grad_norm": 11.5078125, "learning_rate": 8.915186185332576e-06, "loss": 0.536, "step": 31031 }, { "epoch": 3.3019791444988296, "grad_norm": 13.8046875, "learning_rate": 8.915150653780558e-06, "loss": 0.6597, "step": 31032 }, { "epoch": 3.302085550117046, "grad_norm": 14.2109375, "learning_rate": 8.91511512222854e-06, "loss": 0.6327, "step": 31033 }, { "epoch": 3.302191955735263, "grad_norm": 13.046875, "learning_rate": 8.915079590676521e-06, "loss": 0.7569, "step": 31034 }, { "epoch": 3.3022983613534795, "grad_norm": 16.0625, "learning_rate": 8.915044059124503e-06, "loss": 0.6849, "step": 31035 }, { "epoch": 3.302404766971696, "grad_norm": 12.546875, "learning_rate": 8.915008527572484e-06, "loss": 0.6668, "step": 31036 }, { "epoch": 3.3025111725899126, "grad_norm": 257.5, "learning_rate": 8.914972996020466e-06, "loss": 0.5961, "step": 31037 }, { "epoch": 3.3026175782081295, "grad_norm": 14.2265625, "learning_rate": 8.914937464468448e-06, "loss": 0.6836, "step": 31038 }, { "epoch": 3.302723983826346, "grad_norm": 13.078125, "learning_rate": 8.914901932916431e-06, "loss": 0.6685, "step": 31039 }, { "epoch": 3.3028303894445625, "grad_norm": 11.6484375, "learning_rate": 8.914866401364413e-06, "loss": 0.5944, "step": 31040 }, { "epoch": 3.3029367950627795, "grad_norm": 15.546875, "learning_rate": 8.914830869812394e-06, "loss": 0.8527, "step": 31041 }, { "epoch": 3.303043200680996, "grad_norm": 12.3359375, "learning_rate": 8.914795338260376e-06, "loss": 0.6908, "step": 31042 }, { "epoch": 3.3031496062992125, "grad_norm": 15.1484375, "learning_rate": 8.914759806708357e-06, "loss": 0.733, "step": 31043 }, { "epoch": 3.303256011917429, "grad_norm": 13.90625, "learning_rate": 8.91472427515634e-06, "loss": 0.749, "step": 31044 }, { "epoch": 3.303362417535646, "grad_norm": 14.265625, "learning_rate": 8.914688743604322e-06, "loss": 0.7702, "step": 31045 }, { "epoch": 3.3034688231538625, "grad_norm": 8.1953125, "learning_rate": 8.914653212052304e-06, "loss": 0.3733, "step": 31046 }, { "epoch": 3.303575228772079, "grad_norm": 12.171875, "learning_rate": 8.914617680500286e-06, "loss": 0.738, "step": 31047 }, { "epoch": 3.303681634390296, "grad_norm": 13.59375, "learning_rate": 8.914582148948267e-06, "loss": 0.7365, "step": 31048 }, { "epoch": 3.3037880400085124, "grad_norm": 11.109375, "learning_rate": 8.914546617396249e-06, "loss": 0.5735, "step": 31049 }, { "epoch": 3.303894445626729, "grad_norm": 14.5234375, "learning_rate": 8.91451108584423e-06, "loss": 0.8061, "step": 31050 }, { "epoch": 3.304000851244946, "grad_norm": 11.6875, "learning_rate": 8.914475554292212e-06, "loss": 0.6166, "step": 31051 }, { "epoch": 3.3041072568631624, "grad_norm": 13.265625, "learning_rate": 8.914440022740194e-06, "loss": 0.7522, "step": 31052 }, { "epoch": 3.304213662481379, "grad_norm": 13.03125, "learning_rate": 8.914404491188175e-06, "loss": 0.52, "step": 31053 }, { "epoch": 3.304320068099596, "grad_norm": 10.484375, "learning_rate": 8.914368959636157e-06, "loss": 0.4609, "step": 31054 }, { "epoch": 3.3044264737178124, "grad_norm": 16.4375, "learning_rate": 8.91433342808414e-06, "loss": 0.7036, "step": 31055 }, { "epoch": 3.304532879336029, "grad_norm": 13.7890625, "learning_rate": 8.914297896532122e-06, "loss": 0.6631, "step": 31056 }, { "epoch": 3.3046392849542454, "grad_norm": 14.59375, "learning_rate": 8.914262364980103e-06, "loss": 0.6558, "step": 31057 }, { "epoch": 3.3047456905724624, "grad_norm": 11.84375, "learning_rate": 8.914226833428085e-06, "loss": 0.611, "step": 31058 }, { "epoch": 3.304852096190679, "grad_norm": 12.7734375, "learning_rate": 8.914191301876067e-06, "loss": 0.6054, "step": 31059 }, { "epoch": 3.3049585018088954, "grad_norm": 10.6796875, "learning_rate": 8.914155770324048e-06, "loss": 0.5466, "step": 31060 }, { "epoch": 3.3050649074271123, "grad_norm": 10.0234375, "learning_rate": 8.91412023877203e-06, "loss": 0.552, "step": 31061 }, { "epoch": 3.305171313045329, "grad_norm": 12.5078125, "learning_rate": 8.914084707220011e-06, "loss": 0.5656, "step": 31062 }, { "epoch": 3.3052777186635454, "grad_norm": 9.9296875, "learning_rate": 8.914049175667993e-06, "loss": 0.5274, "step": 31063 }, { "epoch": 3.305384124281762, "grad_norm": 12.2109375, "learning_rate": 8.914013644115975e-06, "loss": 0.6712, "step": 31064 }, { "epoch": 3.305490529899979, "grad_norm": 11.6640625, "learning_rate": 8.913978112563958e-06, "loss": 0.6509, "step": 31065 }, { "epoch": 3.3055969355181953, "grad_norm": 11.8671875, "learning_rate": 8.91394258101194e-06, "loss": 0.5608, "step": 31066 }, { "epoch": 3.305703341136412, "grad_norm": 18.34375, "learning_rate": 8.913907049459921e-06, "loss": 0.8449, "step": 31067 }, { "epoch": 3.305809746754629, "grad_norm": 18.15625, "learning_rate": 8.913871517907903e-06, "loss": 0.7145, "step": 31068 }, { "epoch": 3.3059161523728453, "grad_norm": 15.421875, "learning_rate": 8.913835986355884e-06, "loss": 0.5968, "step": 31069 }, { "epoch": 3.306022557991062, "grad_norm": 13.671875, "learning_rate": 8.913800454803868e-06, "loss": 0.6606, "step": 31070 }, { "epoch": 3.3061289636092788, "grad_norm": 55.71875, "learning_rate": 8.91376492325185e-06, "loss": 1.0743, "step": 31071 }, { "epoch": 3.3062353692274953, "grad_norm": 14.1953125, "learning_rate": 8.91372939169983e-06, "loss": 0.7225, "step": 31072 }, { "epoch": 3.306341774845712, "grad_norm": 12.109375, "learning_rate": 8.913693860147812e-06, "loss": 0.5071, "step": 31073 }, { "epoch": 3.3064481804639287, "grad_norm": 13.21875, "learning_rate": 8.913658328595794e-06, "loss": 0.5818, "step": 31074 }, { "epoch": 3.3065545860821453, "grad_norm": 12.7734375, "learning_rate": 8.913622797043776e-06, "loss": 0.6698, "step": 31075 }, { "epoch": 3.3066609917003618, "grad_norm": 12.53125, "learning_rate": 8.913587265491757e-06, "loss": 0.6174, "step": 31076 }, { "epoch": 3.3067673973185783, "grad_norm": 16.5, "learning_rate": 8.913551733939739e-06, "loss": 0.8589, "step": 31077 }, { "epoch": 3.3068738029367952, "grad_norm": 12.9296875, "learning_rate": 8.91351620238772e-06, "loss": 0.6985, "step": 31078 }, { "epoch": 3.3069802085550117, "grad_norm": 13.2578125, "learning_rate": 8.913480670835702e-06, "loss": 0.662, "step": 31079 }, { "epoch": 3.3070866141732282, "grad_norm": 13.8203125, "learning_rate": 8.913445139283685e-06, "loss": 0.7048, "step": 31080 }, { "epoch": 3.3071930197914448, "grad_norm": 15.2421875, "learning_rate": 8.913409607731667e-06, "loss": 0.6829, "step": 31081 }, { "epoch": 3.3072994254096617, "grad_norm": 17.140625, "learning_rate": 8.913374076179649e-06, "loss": 0.6603, "step": 31082 }, { "epoch": 3.307405831027878, "grad_norm": 12.8984375, "learning_rate": 8.91333854462763e-06, "loss": 0.5568, "step": 31083 }, { "epoch": 3.3075122366460947, "grad_norm": 16.75, "learning_rate": 8.913303013075612e-06, "loss": 0.6185, "step": 31084 }, { "epoch": 3.3076186422643117, "grad_norm": 11.4453125, "learning_rate": 8.913267481523595e-06, "loss": 0.5425, "step": 31085 }, { "epoch": 3.307725047882528, "grad_norm": 10.3984375, "learning_rate": 8.913231949971575e-06, "loss": 0.585, "step": 31086 }, { "epoch": 3.3078314535007447, "grad_norm": 12.1015625, "learning_rate": 8.913196418419557e-06, "loss": 0.5685, "step": 31087 }, { "epoch": 3.3079378591189617, "grad_norm": 12.484375, "learning_rate": 8.913160886867538e-06, "loss": 0.5968, "step": 31088 }, { "epoch": 3.308044264737178, "grad_norm": 10.8828125, "learning_rate": 8.91312535531552e-06, "loss": 0.5545, "step": 31089 }, { "epoch": 3.3081506703553947, "grad_norm": 12.8671875, "learning_rate": 8.913089823763501e-06, "loss": 0.7201, "step": 31090 }, { "epoch": 3.3082570759736116, "grad_norm": 15.9765625, "learning_rate": 8.913054292211485e-06, "loss": 0.6555, "step": 31091 }, { "epoch": 3.308363481591828, "grad_norm": 25.09375, "learning_rate": 8.913018760659466e-06, "loss": 0.84, "step": 31092 }, { "epoch": 3.3084698872100446, "grad_norm": 10.2890625, "learning_rate": 8.912983229107448e-06, "loss": 0.5852, "step": 31093 }, { "epoch": 3.308576292828261, "grad_norm": 15.328125, "learning_rate": 8.91294769755543e-06, "loss": 0.6997, "step": 31094 }, { "epoch": 3.308682698446478, "grad_norm": 13.2890625, "learning_rate": 8.912912166003411e-06, "loss": 0.61, "step": 31095 }, { "epoch": 3.3087891040646946, "grad_norm": 10.9609375, "learning_rate": 8.912876634451394e-06, "loss": 0.5414, "step": 31096 }, { "epoch": 3.308895509682911, "grad_norm": 11.3125, "learning_rate": 8.912841102899376e-06, "loss": 0.6442, "step": 31097 }, { "epoch": 3.309001915301128, "grad_norm": 11.875, "learning_rate": 8.912805571347358e-06, "loss": 0.6059, "step": 31098 }, { "epoch": 3.3091083209193446, "grad_norm": 15.046875, "learning_rate": 8.91277003979534e-06, "loss": 0.6879, "step": 31099 }, { "epoch": 3.309214726537561, "grad_norm": 10.875, "learning_rate": 8.912734508243321e-06, "loss": 0.532, "step": 31100 }, { "epoch": 3.3093211321557776, "grad_norm": 10.9453125, "learning_rate": 8.912698976691302e-06, "loss": 0.6687, "step": 31101 }, { "epoch": 3.3094275377739946, "grad_norm": 12.046875, "learning_rate": 8.912663445139284e-06, "loss": 0.645, "step": 31102 }, { "epoch": 3.309533943392211, "grad_norm": 11.453125, "learning_rate": 8.912627913587266e-06, "loss": 0.701, "step": 31103 }, { "epoch": 3.3096403490104276, "grad_norm": 12.40625, "learning_rate": 8.912592382035247e-06, "loss": 0.5873, "step": 31104 }, { "epoch": 3.3097467546286445, "grad_norm": 9.578125, "learning_rate": 8.912556850483229e-06, "loss": 0.4981, "step": 31105 }, { "epoch": 3.309853160246861, "grad_norm": 13.4140625, "learning_rate": 8.912521318931212e-06, "loss": 0.6671, "step": 31106 }, { "epoch": 3.3099595658650776, "grad_norm": 14.4375, "learning_rate": 8.912485787379194e-06, "loss": 0.9576, "step": 31107 }, { "epoch": 3.3100659714832945, "grad_norm": 10.390625, "learning_rate": 8.912450255827175e-06, "loss": 0.5113, "step": 31108 }, { "epoch": 3.310172377101511, "grad_norm": 11.3515625, "learning_rate": 8.912414724275157e-06, "loss": 0.6594, "step": 31109 }, { "epoch": 3.3102787827197275, "grad_norm": 13.828125, "learning_rate": 8.912379192723139e-06, "loss": 0.627, "step": 31110 }, { "epoch": 3.3103851883379445, "grad_norm": 15.421875, "learning_rate": 8.912343661171122e-06, "loss": 0.6213, "step": 31111 }, { "epoch": 3.310491593956161, "grad_norm": 11.625, "learning_rate": 8.912308129619104e-06, "loss": 0.5635, "step": 31112 }, { "epoch": 3.3105979995743775, "grad_norm": 11.3515625, "learning_rate": 8.912272598067085e-06, "loss": 0.5896, "step": 31113 }, { "epoch": 3.310704405192594, "grad_norm": 16.609375, "learning_rate": 8.912237066515067e-06, "loss": 0.6185, "step": 31114 }, { "epoch": 3.310810810810811, "grad_norm": 11.828125, "learning_rate": 8.912201534963048e-06, "loss": 0.7422, "step": 31115 }, { "epoch": 3.3109172164290275, "grad_norm": 15.6015625, "learning_rate": 8.91216600341103e-06, "loss": 0.6252, "step": 31116 }, { "epoch": 3.311023622047244, "grad_norm": 11.421875, "learning_rate": 8.912130471859012e-06, "loss": 0.6176, "step": 31117 }, { "epoch": 3.3111300276654605, "grad_norm": 10.6484375, "learning_rate": 8.912094940306993e-06, "loss": 0.6016, "step": 31118 }, { "epoch": 3.3112364332836774, "grad_norm": 13.0625, "learning_rate": 8.912059408754975e-06, "loss": 0.5593, "step": 31119 }, { "epoch": 3.311342838901894, "grad_norm": 13.359375, "learning_rate": 8.912023877202956e-06, "loss": 0.6456, "step": 31120 }, { "epoch": 3.3114492445201105, "grad_norm": 11.6171875, "learning_rate": 8.91198834565094e-06, "loss": 0.5105, "step": 31121 }, { "epoch": 3.3115556501383274, "grad_norm": 10.6875, "learning_rate": 8.911952814098921e-06, "loss": 0.5921, "step": 31122 }, { "epoch": 3.311662055756544, "grad_norm": 14.234375, "learning_rate": 8.911917282546903e-06, "loss": 0.5175, "step": 31123 }, { "epoch": 3.3117684613747604, "grad_norm": 12.2578125, "learning_rate": 8.911881750994884e-06, "loss": 0.8235, "step": 31124 }, { "epoch": 3.3118748669929774, "grad_norm": 11.75, "learning_rate": 8.911846219442866e-06, "loss": 0.497, "step": 31125 }, { "epoch": 3.311981272611194, "grad_norm": 15.875, "learning_rate": 8.911810687890848e-06, "loss": 0.665, "step": 31126 }, { "epoch": 3.3120876782294104, "grad_norm": 11.53125, "learning_rate": 8.91177515633883e-06, "loss": 0.6008, "step": 31127 }, { "epoch": 3.3121940838476274, "grad_norm": 12.0390625, "learning_rate": 8.911739624786811e-06, "loss": 0.7093, "step": 31128 }, { "epoch": 3.312300489465844, "grad_norm": 14.484375, "learning_rate": 8.911704093234793e-06, "loss": 0.7853, "step": 31129 }, { "epoch": 3.3124068950840604, "grad_norm": 17.234375, "learning_rate": 8.911668561682774e-06, "loss": 0.8806, "step": 31130 }, { "epoch": 3.312513300702277, "grad_norm": 18.984375, "learning_rate": 8.911633030130756e-06, "loss": 0.7579, "step": 31131 }, { "epoch": 3.312619706320494, "grad_norm": 13.953125, "learning_rate": 8.911597498578739e-06, "loss": 0.6506, "step": 31132 }, { "epoch": 3.3127261119387104, "grad_norm": 10.0859375, "learning_rate": 8.91156196702672e-06, "loss": 0.4911, "step": 31133 }, { "epoch": 3.312832517556927, "grad_norm": 15.53125, "learning_rate": 8.911526435474702e-06, "loss": 0.7343, "step": 31134 }, { "epoch": 3.312938923175144, "grad_norm": 18.859375, "learning_rate": 8.911490903922684e-06, "loss": 0.682, "step": 31135 }, { "epoch": 3.3130453287933603, "grad_norm": 11.7890625, "learning_rate": 8.911455372370665e-06, "loss": 0.6091, "step": 31136 }, { "epoch": 3.313151734411577, "grad_norm": 16.65625, "learning_rate": 8.911419840818649e-06, "loss": 0.6672, "step": 31137 }, { "epoch": 3.3132581400297934, "grad_norm": 18.28125, "learning_rate": 8.91138430926663e-06, "loss": 0.8083, "step": 31138 }, { "epoch": 3.3133645456480103, "grad_norm": 11.578125, "learning_rate": 8.911348777714612e-06, "loss": 0.5606, "step": 31139 }, { "epoch": 3.313470951266227, "grad_norm": 13.859375, "learning_rate": 8.911313246162594e-06, "loss": 0.6421, "step": 31140 }, { "epoch": 3.3135773568844433, "grad_norm": 13.4609375, "learning_rate": 8.911277714610575e-06, "loss": 0.7214, "step": 31141 }, { "epoch": 3.3136837625026603, "grad_norm": 12.7109375, "learning_rate": 8.911242183058557e-06, "loss": 0.5505, "step": 31142 }, { "epoch": 3.313790168120877, "grad_norm": 11.5859375, "learning_rate": 8.911206651506538e-06, "loss": 0.4631, "step": 31143 }, { "epoch": 3.3138965737390933, "grad_norm": 15.0078125, "learning_rate": 8.91117111995452e-06, "loss": 0.8206, "step": 31144 }, { "epoch": 3.3140029793573103, "grad_norm": 17.4375, "learning_rate": 8.911135588402502e-06, "loss": 0.6906, "step": 31145 }, { "epoch": 3.3141093849755268, "grad_norm": 11.34375, "learning_rate": 8.911100056850483e-06, "loss": 0.5307, "step": 31146 }, { "epoch": 3.3142157905937433, "grad_norm": 18.328125, "learning_rate": 8.911064525298467e-06, "loss": 0.7313, "step": 31147 }, { "epoch": 3.3143221962119602, "grad_norm": 14.0234375, "learning_rate": 8.911028993746448e-06, "loss": 0.4915, "step": 31148 }, { "epoch": 3.3144286018301767, "grad_norm": 14.609375, "learning_rate": 8.91099346219443e-06, "loss": 0.5665, "step": 31149 }, { "epoch": 3.3145350074483932, "grad_norm": 12.2578125, "learning_rate": 8.910957930642411e-06, "loss": 0.6617, "step": 31150 }, { "epoch": 3.3146414130666098, "grad_norm": 17.140625, "learning_rate": 8.910922399090393e-06, "loss": 0.8157, "step": 31151 }, { "epoch": 3.3147478186848267, "grad_norm": 14.3984375, "learning_rate": 8.910886867538375e-06, "loss": 0.5889, "step": 31152 }, { "epoch": 3.314854224303043, "grad_norm": 12.578125, "learning_rate": 8.910851335986356e-06, "loss": 0.6208, "step": 31153 }, { "epoch": 3.3149606299212597, "grad_norm": 13.203125, "learning_rate": 8.910815804434338e-06, "loss": 0.5992, "step": 31154 }, { "epoch": 3.3150670355394762, "grad_norm": 15.4921875, "learning_rate": 8.91078027288232e-06, "loss": 0.6973, "step": 31155 }, { "epoch": 3.315173441157693, "grad_norm": 11.828125, "learning_rate": 8.910744741330301e-06, "loss": 0.5776, "step": 31156 }, { "epoch": 3.3152798467759097, "grad_norm": 14.1640625, "learning_rate": 8.910709209778284e-06, "loss": 0.5582, "step": 31157 }, { "epoch": 3.315386252394126, "grad_norm": 11.46875, "learning_rate": 8.910673678226266e-06, "loss": 0.5286, "step": 31158 }, { "epoch": 3.315492658012343, "grad_norm": 12.53125, "learning_rate": 8.910638146674248e-06, "loss": 0.5758, "step": 31159 }, { "epoch": 3.3155990636305597, "grad_norm": 13.921875, "learning_rate": 8.910602615122229e-06, "loss": 0.5516, "step": 31160 }, { "epoch": 3.315705469248776, "grad_norm": 15.296875, "learning_rate": 8.91056708357021e-06, "loss": 0.6269, "step": 31161 }, { "epoch": 3.315811874866993, "grad_norm": 12.40625, "learning_rate": 8.910531552018194e-06, "loss": 0.641, "step": 31162 }, { "epoch": 3.3159182804852096, "grad_norm": 11.1953125, "learning_rate": 8.910496020466176e-06, "loss": 0.4941, "step": 31163 }, { "epoch": 3.316024686103426, "grad_norm": 11.15625, "learning_rate": 8.910460488914157e-06, "loss": 0.5489, "step": 31164 }, { "epoch": 3.316131091721643, "grad_norm": 10.6328125, "learning_rate": 8.910424957362139e-06, "loss": 0.6677, "step": 31165 }, { "epoch": 3.3162374973398596, "grad_norm": 15.15625, "learning_rate": 8.91038942581012e-06, "loss": 0.694, "step": 31166 }, { "epoch": 3.316343902958076, "grad_norm": 11.7734375, "learning_rate": 8.910353894258102e-06, "loss": 0.6276, "step": 31167 }, { "epoch": 3.3164503085762926, "grad_norm": 11.546875, "learning_rate": 8.910318362706084e-06, "loss": 0.5675, "step": 31168 }, { "epoch": 3.3165567141945096, "grad_norm": 11.109375, "learning_rate": 8.910282831154065e-06, "loss": 0.5733, "step": 31169 }, { "epoch": 3.316663119812726, "grad_norm": 13.1640625, "learning_rate": 8.910247299602047e-06, "loss": 0.6024, "step": 31170 }, { "epoch": 3.3167695254309426, "grad_norm": 14.453125, "learning_rate": 8.910211768050028e-06, "loss": 0.5467, "step": 31171 }, { "epoch": 3.3168759310491596, "grad_norm": 10.4921875, "learning_rate": 8.91017623649801e-06, "loss": 0.7549, "step": 31172 }, { "epoch": 3.316982336667376, "grad_norm": 10.34375, "learning_rate": 8.910140704945993e-06, "loss": 0.5997, "step": 31173 }, { "epoch": 3.3170887422855926, "grad_norm": 19.375, "learning_rate": 8.910105173393975e-06, "loss": 0.7102, "step": 31174 }, { "epoch": 3.317195147903809, "grad_norm": 12.171875, "learning_rate": 8.910069641841957e-06, "loss": 0.7326, "step": 31175 }, { "epoch": 3.317301553522026, "grad_norm": 14.4765625, "learning_rate": 8.910034110289938e-06, "loss": 0.6926, "step": 31176 }, { "epoch": 3.3174079591402426, "grad_norm": 10.5078125, "learning_rate": 8.90999857873792e-06, "loss": 0.5264, "step": 31177 }, { "epoch": 3.317514364758459, "grad_norm": 13.71875, "learning_rate": 8.909963047185901e-06, "loss": 0.6852, "step": 31178 }, { "epoch": 3.317620770376676, "grad_norm": 13.7421875, "learning_rate": 8.909927515633883e-06, "loss": 0.6741, "step": 31179 }, { "epoch": 3.3177271759948925, "grad_norm": 14.453125, "learning_rate": 8.909891984081865e-06, "loss": 0.6318, "step": 31180 }, { "epoch": 3.317833581613109, "grad_norm": 11.3359375, "learning_rate": 8.909856452529846e-06, "loss": 0.6467, "step": 31181 }, { "epoch": 3.317939987231326, "grad_norm": 14.84375, "learning_rate": 8.909820920977828e-06, "loss": 0.63, "step": 31182 }, { "epoch": 3.3180463928495425, "grad_norm": 13.7734375, "learning_rate": 8.909785389425811e-06, "loss": 0.6535, "step": 31183 }, { "epoch": 3.318152798467759, "grad_norm": 14.875, "learning_rate": 8.909749857873793e-06, "loss": 0.7262, "step": 31184 }, { "epoch": 3.318259204085976, "grad_norm": 10.765625, "learning_rate": 8.909714326321774e-06, "loss": 0.5573, "step": 31185 }, { "epoch": 3.3183656097041925, "grad_norm": 16.484375, "learning_rate": 8.909678794769756e-06, "loss": 0.6278, "step": 31186 }, { "epoch": 3.318472015322409, "grad_norm": 11.2421875, "learning_rate": 8.909643263217738e-06, "loss": 0.5721, "step": 31187 }, { "epoch": 3.3185784209406255, "grad_norm": 11.4453125, "learning_rate": 8.909607731665721e-06, "loss": 0.5867, "step": 31188 }, { "epoch": 3.3186848265588424, "grad_norm": 14.234375, "learning_rate": 8.909572200113702e-06, "loss": 0.7551, "step": 31189 }, { "epoch": 3.318791232177059, "grad_norm": 12.421875, "learning_rate": 8.909536668561684e-06, "loss": 0.7614, "step": 31190 }, { "epoch": 3.3188976377952755, "grad_norm": 13.1328125, "learning_rate": 8.909501137009666e-06, "loss": 0.608, "step": 31191 }, { "epoch": 3.3190040434134924, "grad_norm": 17.359375, "learning_rate": 8.909465605457647e-06, "loss": 0.7062, "step": 31192 }, { "epoch": 3.319110449031709, "grad_norm": 11.7734375, "learning_rate": 8.909430073905629e-06, "loss": 0.63, "step": 31193 }, { "epoch": 3.3192168546499254, "grad_norm": 14.3359375, "learning_rate": 8.90939454235361e-06, "loss": 0.6655, "step": 31194 }, { "epoch": 3.319323260268142, "grad_norm": 12.890625, "learning_rate": 8.909359010801592e-06, "loss": 0.6963, "step": 31195 }, { "epoch": 3.319429665886359, "grad_norm": 19.484375, "learning_rate": 8.909323479249574e-06, "loss": 0.7095, "step": 31196 }, { "epoch": 3.3195360715045754, "grad_norm": 17.21875, "learning_rate": 8.909287947697555e-06, "loss": 0.6798, "step": 31197 }, { "epoch": 3.319642477122792, "grad_norm": 13.1796875, "learning_rate": 8.909252416145539e-06, "loss": 0.7959, "step": 31198 }, { "epoch": 3.319748882741009, "grad_norm": 11.5546875, "learning_rate": 8.90921688459352e-06, "loss": 0.5761, "step": 31199 }, { "epoch": 3.3198552883592254, "grad_norm": 11.7421875, "learning_rate": 8.909181353041502e-06, "loss": 0.7113, "step": 31200 }, { "epoch": 3.319961693977442, "grad_norm": 11.0390625, "learning_rate": 8.909145821489483e-06, "loss": 0.7166, "step": 31201 }, { "epoch": 3.320068099595659, "grad_norm": 14.1171875, "learning_rate": 8.909110289937465e-06, "loss": 0.6597, "step": 31202 }, { "epoch": 3.3201745052138754, "grad_norm": 13.3125, "learning_rate": 8.909074758385448e-06, "loss": 0.672, "step": 31203 }, { "epoch": 3.320280910832092, "grad_norm": 10.5703125, "learning_rate": 8.90903922683343e-06, "loss": 0.5617, "step": 31204 }, { "epoch": 3.320387316450309, "grad_norm": 11.140625, "learning_rate": 8.909003695281412e-06, "loss": 0.6031, "step": 31205 }, { "epoch": 3.3204937220685253, "grad_norm": 10.3671875, "learning_rate": 8.908968163729391e-06, "loss": 0.6225, "step": 31206 }, { "epoch": 3.320600127686742, "grad_norm": 12.8828125, "learning_rate": 8.908932632177373e-06, "loss": 0.6313, "step": 31207 }, { "epoch": 3.3207065333049584, "grad_norm": 14.8046875, "learning_rate": 8.908897100625355e-06, "loss": 0.5549, "step": 31208 }, { "epoch": 3.3208129389231753, "grad_norm": 11.515625, "learning_rate": 8.908861569073338e-06, "loss": 0.5699, "step": 31209 }, { "epoch": 3.320919344541392, "grad_norm": 10.3515625, "learning_rate": 8.90882603752132e-06, "loss": 0.6082, "step": 31210 }, { "epoch": 3.3210257501596083, "grad_norm": 14.0546875, "learning_rate": 8.908790505969301e-06, "loss": 0.7689, "step": 31211 }, { "epoch": 3.321132155777825, "grad_norm": 14.7578125, "learning_rate": 8.908754974417283e-06, "loss": 0.619, "step": 31212 }, { "epoch": 3.321238561396042, "grad_norm": 13.3125, "learning_rate": 8.908719442865264e-06, "loss": 0.6319, "step": 31213 }, { "epoch": 3.3213449670142583, "grad_norm": 12.8046875, "learning_rate": 8.908683911313248e-06, "loss": 0.6554, "step": 31214 }, { "epoch": 3.321451372632475, "grad_norm": 9.6015625, "learning_rate": 8.90864837976123e-06, "loss": 0.5383, "step": 31215 }, { "epoch": 3.3215577782506918, "grad_norm": 11.953125, "learning_rate": 8.908612848209211e-06, "loss": 0.5366, "step": 31216 }, { "epoch": 3.3216641838689083, "grad_norm": 15.7109375, "learning_rate": 8.908577316657193e-06, "loss": 0.8615, "step": 31217 }, { "epoch": 3.321770589487125, "grad_norm": 17.265625, "learning_rate": 8.908541785105174e-06, "loss": 0.732, "step": 31218 }, { "epoch": 3.3218769951053417, "grad_norm": 16.5, "learning_rate": 8.908506253553156e-06, "loss": 0.6904, "step": 31219 }, { "epoch": 3.3219834007235582, "grad_norm": 12.0078125, "learning_rate": 8.908470722001137e-06, "loss": 0.5848, "step": 31220 }, { "epoch": 3.3220898063417748, "grad_norm": 14.265625, "learning_rate": 8.908435190449119e-06, "loss": 0.6168, "step": 31221 }, { "epoch": 3.3221962119599917, "grad_norm": 13.03125, "learning_rate": 8.9083996588971e-06, "loss": 0.5496, "step": 31222 }, { "epoch": 3.322302617578208, "grad_norm": 13.1796875, "learning_rate": 8.908364127345082e-06, "loss": 0.6969, "step": 31223 }, { "epoch": 3.3224090231964247, "grad_norm": 12.9140625, "learning_rate": 8.908328595793065e-06, "loss": 0.6995, "step": 31224 }, { "epoch": 3.3225154288146412, "grad_norm": 12.015625, "learning_rate": 8.908293064241047e-06, "loss": 0.6595, "step": 31225 }, { "epoch": 3.322621834432858, "grad_norm": 12.03125, "learning_rate": 8.908257532689029e-06, "loss": 0.6118, "step": 31226 }, { "epoch": 3.3227282400510747, "grad_norm": 10.3046875, "learning_rate": 8.90822200113701e-06, "loss": 0.4674, "step": 31227 }, { "epoch": 3.322834645669291, "grad_norm": 13.8359375, "learning_rate": 8.908186469584992e-06, "loss": 0.6026, "step": 31228 }, { "epoch": 3.322941051287508, "grad_norm": 19.1875, "learning_rate": 8.908150938032975e-06, "loss": 0.7921, "step": 31229 }, { "epoch": 3.3230474569057247, "grad_norm": 18.53125, "learning_rate": 8.908115406480957e-06, "loss": 0.6465, "step": 31230 }, { "epoch": 3.323153862523941, "grad_norm": 17.984375, "learning_rate": 8.908079874928938e-06, "loss": 0.6862, "step": 31231 }, { "epoch": 3.3232602681421577, "grad_norm": 15.6171875, "learning_rate": 8.90804434337692e-06, "loss": 0.6332, "step": 31232 }, { "epoch": 3.3233666737603746, "grad_norm": 11.9375, "learning_rate": 8.908008811824902e-06, "loss": 0.5877, "step": 31233 }, { "epoch": 3.323473079378591, "grad_norm": 14.1015625, "learning_rate": 8.907973280272883e-06, "loss": 0.6665, "step": 31234 }, { "epoch": 3.3235794849968077, "grad_norm": 13.390625, "learning_rate": 8.907937748720865e-06, "loss": 0.6199, "step": 31235 }, { "epoch": 3.3236858906150246, "grad_norm": 12.5703125, "learning_rate": 8.907902217168846e-06, "loss": 0.641, "step": 31236 }, { "epoch": 3.323792296233241, "grad_norm": 11.671875, "learning_rate": 8.907866685616828e-06, "loss": 0.6396, "step": 31237 }, { "epoch": 3.3238987018514576, "grad_norm": 13.28125, "learning_rate": 8.90783115406481e-06, "loss": 0.6478, "step": 31238 }, { "epoch": 3.3240051074696746, "grad_norm": 17.3125, "learning_rate": 8.907795622512793e-06, "loss": 0.6328, "step": 31239 }, { "epoch": 3.324111513087891, "grad_norm": 13.9609375, "learning_rate": 8.907760090960775e-06, "loss": 0.7461, "step": 31240 }, { "epoch": 3.3242179187061076, "grad_norm": 13.859375, "learning_rate": 8.907724559408756e-06, "loss": 0.6036, "step": 31241 }, { "epoch": 3.3243243243243246, "grad_norm": 16.90625, "learning_rate": 8.907689027856738e-06, "loss": 0.78, "step": 31242 }, { "epoch": 3.324430729942541, "grad_norm": 16.46875, "learning_rate": 8.90765349630472e-06, "loss": 0.7356, "step": 31243 }, { "epoch": 3.3245371355607576, "grad_norm": 9.296875, "learning_rate": 8.907617964752701e-06, "loss": 0.4761, "step": 31244 }, { "epoch": 3.324643541178974, "grad_norm": 15.5546875, "learning_rate": 8.907582433200683e-06, "loss": 0.8164, "step": 31245 }, { "epoch": 3.324749946797191, "grad_norm": 10.6875, "learning_rate": 8.907546901648664e-06, "loss": 0.5845, "step": 31246 }, { "epoch": 3.3248563524154076, "grad_norm": 19.25, "learning_rate": 8.907511370096646e-06, "loss": 0.6724, "step": 31247 }, { "epoch": 3.324962758033624, "grad_norm": 10.78125, "learning_rate": 8.907475838544627e-06, "loss": 0.6292, "step": 31248 }, { "epoch": 3.3250691636518406, "grad_norm": 11.9921875, "learning_rate": 8.907440306992609e-06, "loss": 0.5757, "step": 31249 }, { "epoch": 3.3251755692700575, "grad_norm": 11.6953125, "learning_rate": 8.907404775440592e-06, "loss": 0.799, "step": 31250 }, { "epoch": 3.325281974888274, "grad_norm": 12.4296875, "learning_rate": 8.907369243888574e-06, "loss": 0.6805, "step": 31251 }, { "epoch": 3.3253883805064905, "grad_norm": 19.03125, "learning_rate": 8.907333712336556e-06, "loss": 0.7447, "step": 31252 }, { "epoch": 3.3254947861247075, "grad_norm": 14.4765625, "learning_rate": 8.907298180784537e-06, "loss": 0.6729, "step": 31253 }, { "epoch": 3.325601191742924, "grad_norm": 12.4609375, "learning_rate": 8.907262649232519e-06, "loss": 0.6158, "step": 31254 }, { "epoch": 3.3257075973611405, "grad_norm": 18.65625, "learning_rate": 8.907227117680502e-06, "loss": 0.7508, "step": 31255 }, { "epoch": 3.3258140029793575, "grad_norm": 14.90625, "learning_rate": 8.907191586128484e-06, "loss": 0.5931, "step": 31256 }, { "epoch": 3.325920408597574, "grad_norm": 12.125, "learning_rate": 8.907156054576465e-06, "loss": 0.5944, "step": 31257 }, { "epoch": 3.3260268142157905, "grad_norm": 16.03125, "learning_rate": 8.907120523024447e-06, "loss": 0.702, "step": 31258 }, { "epoch": 3.3261332198340074, "grad_norm": 18.125, "learning_rate": 8.907084991472428e-06, "loss": 0.8043, "step": 31259 }, { "epoch": 3.326239625452224, "grad_norm": 14.3828125, "learning_rate": 8.90704945992041e-06, "loss": 0.7479, "step": 31260 }, { "epoch": 3.3263460310704405, "grad_norm": 12.5390625, "learning_rate": 8.907013928368392e-06, "loss": 0.5417, "step": 31261 }, { "epoch": 3.326452436688657, "grad_norm": 14.0078125, "learning_rate": 8.906978396816373e-06, "loss": 0.6892, "step": 31262 }, { "epoch": 3.326558842306874, "grad_norm": 11.734375, "learning_rate": 8.906942865264355e-06, "loss": 0.5015, "step": 31263 }, { "epoch": 3.3266652479250904, "grad_norm": 13.34375, "learning_rate": 8.906907333712337e-06, "loss": 0.5229, "step": 31264 }, { "epoch": 3.326771653543307, "grad_norm": 12.6171875, "learning_rate": 8.90687180216032e-06, "loss": 0.6827, "step": 31265 }, { "epoch": 3.326878059161524, "grad_norm": 14.875, "learning_rate": 8.906836270608301e-06, "loss": 0.6535, "step": 31266 }, { "epoch": 3.3269844647797404, "grad_norm": 15.3515625, "learning_rate": 8.906800739056283e-06, "loss": 0.6643, "step": 31267 }, { "epoch": 3.327090870397957, "grad_norm": 12.3046875, "learning_rate": 8.906765207504265e-06, "loss": 0.5784, "step": 31268 }, { "epoch": 3.3271972760161734, "grad_norm": 10.8125, "learning_rate": 8.906729675952246e-06, "loss": 0.5653, "step": 31269 }, { "epoch": 3.3273036816343904, "grad_norm": 13.6328125, "learning_rate": 8.906694144400228e-06, "loss": 0.6554, "step": 31270 }, { "epoch": 3.327410087252607, "grad_norm": 13.203125, "learning_rate": 8.90665861284821e-06, "loss": 0.5671, "step": 31271 }, { "epoch": 3.3275164928708234, "grad_norm": 11.21875, "learning_rate": 8.906623081296191e-06, "loss": 0.537, "step": 31272 }, { "epoch": 3.3276228984890404, "grad_norm": 11.4375, "learning_rate": 8.906587549744173e-06, "loss": 0.7296, "step": 31273 }, { "epoch": 3.327729304107257, "grad_norm": 18.03125, "learning_rate": 8.906552018192154e-06, "loss": 0.8136, "step": 31274 }, { "epoch": 3.3278357097254734, "grad_norm": 13.4453125, "learning_rate": 8.906516486640138e-06, "loss": 0.6556, "step": 31275 }, { "epoch": 3.3279421153436903, "grad_norm": 14.484375, "learning_rate": 8.90648095508812e-06, "loss": 0.6089, "step": 31276 }, { "epoch": 3.328048520961907, "grad_norm": 11.0703125, "learning_rate": 8.9064454235361e-06, "loss": 0.5751, "step": 31277 }, { "epoch": 3.3281549265801234, "grad_norm": 11.546875, "learning_rate": 8.906409891984082e-06, "loss": 0.541, "step": 31278 }, { "epoch": 3.3282613321983403, "grad_norm": 13.2734375, "learning_rate": 8.906374360432064e-06, "loss": 0.6017, "step": 31279 }, { "epoch": 3.328367737816557, "grad_norm": 11.671875, "learning_rate": 8.906338828880047e-06, "loss": 0.5632, "step": 31280 }, { "epoch": 3.3284741434347733, "grad_norm": 11.46875, "learning_rate": 8.906303297328029e-06, "loss": 0.568, "step": 31281 }, { "epoch": 3.32858054905299, "grad_norm": 14.7734375, "learning_rate": 8.90626776577601e-06, "loss": 0.663, "step": 31282 }, { "epoch": 3.328686954671207, "grad_norm": 16.21875, "learning_rate": 8.906232234223992e-06, "loss": 0.678, "step": 31283 }, { "epoch": 3.3287933602894233, "grad_norm": 11.6875, "learning_rate": 8.906196702671974e-06, "loss": 0.5783, "step": 31284 }, { "epoch": 3.32889976590764, "grad_norm": 16.796875, "learning_rate": 8.906161171119955e-06, "loss": 0.6831, "step": 31285 }, { "epoch": 3.3290061715258563, "grad_norm": 14.015625, "learning_rate": 8.906125639567937e-06, "loss": 0.655, "step": 31286 }, { "epoch": 3.3291125771440733, "grad_norm": 12.5390625, "learning_rate": 8.906090108015919e-06, "loss": 0.5549, "step": 31287 }, { "epoch": 3.32921898276229, "grad_norm": 16.921875, "learning_rate": 8.9060545764639e-06, "loss": 0.7948, "step": 31288 }, { "epoch": 3.3293253883805063, "grad_norm": 12.21875, "learning_rate": 8.906019044911882e-06, "loss": 0.6311, "step": 31289 }, { "epoch": 3.3294317939987232, "grad_norm": 14.4609375, "learning_rate": 8.905983513359863e-06, "loss": 0.7689, "step": 31290 }, { "epoch": 3.3295381996169398, "grad_norm": 15.609375, "learning_rate": 8.905947981807847e-06, "loss": 0.7801, "step": 31291 }, { "epoch": 3.3296446052351563, "grad_norm": 17.734375, "learning_rate": 8.905912450255828e-06, "loss": 0.8139, "step": 31292 }, { "epoch": 3.329751010853373, "grad_norm": 13.546875, "learning_rate": 8.90587691870381e-06, "loss": 0.6445, "step": 31293 }, { "epoch": 3.3298574164715897, "grad_norm": 14.40625, "learning_rate": 8.905841387151792e-06, "loss": 0.6274, "step": 31294 }, { "epoch": 3.3299638220898062, "grad_norm": 14.1171875, "learning_rate": 8.905805855599773e-06, "loss": 0.6021, "step": 31295 }, { "epoch": 3.330070227708023, "grad_norm": 10.28125, "learning_rate": 8.905770324047755e-06, "loss": 0.666, "step": 31296 }, { "epoch": 3.3301766333262397, "grad_norm": 18.25, "learning_rate": 8.905734792495736e-06, "loss": 0.7124, "step": 31297 }, { "epoch": 3.330283038944456, "grad_norm": 10.1171875, "learning_rate": 8.905699260943718e-06, "loss": 0.6096, "step": 31298 }, { "epoch": 3.3303894445626727, "grad_norm": 12.296875, "learning_rate": 8.9056637293917e-06, "loss": 0.7669, "step": 31299 }, { "epoch": 3.3304958501808897, "grad_norm": 16.1875, "learning_rate": 8.905628197839681e-06, "loss": 0.7949, "step": 31300 }, { "epoch": 3.330602255799106, "grad_norm": 19.140625, "learning_rate": 8.905592666287664e-06, "loss": 0.7789, "step": 31301 }, { "epoch": 3.3307086614173227, "grad_norm": 17.546875, "learning_rate": 8.905557134735646e-06, "loss": 0.7564, "step": 31302 }, { "epoch": 3.3308150670355396, "grad_norm": 16.515625, "learning_rate": 8.905521603183628e-06, "loss": 0.8743, "step": 31303 }, { "epoch": 3.330921472653756, "grad_norm": 11.890625, "learning_rate": 8.90548607163161e-06, "loss": 0.5696, "step": 31304 }, { "epoch": 3.3310278782719727, "grad_norm": 9.890625, "learning_rate": 8.905450540079591e-06, "loss": 0.4809, "step": 31305 }, { "epoch": 3.331134283890189, "grad_norm": 13.953125, "learning_rate": 8.905415008527574e-06, "loss": 0.7273, "step": 31306 }, { "epoch": 3.331240689508406, "grad_norm": 17.828125, "learning_rate": 8.905379476975556e-06, "loss": 0.637, "step": 31307 }, { "epoch": 3.3313470951266226, "grad_norm": 11.6796875, "learning_rate": 8.905343945423537e-06, "loss": 0.613, "step": 31308 }, { "epoch": 3.331453500744839, "grad_norm": 13.25, "learning_rate": 8.905308413871519e-06, "loss": 0.6565, "step": 31309 }, { "epoch": 3.331559906363056, "grad_norm": 12.1796875, "learning_rate": 8.9052728823195e-06, "loss": 0.5108, "step": 31310 }, { "epoch": 3.3316663119812726, "grad_norm": 13.0859375, "learning_rate": 8.905237350767482e-06, "loss": 0.6063, "step": 31311 }, { "epoch": 3.331772717599489, "grad_norm": 16.234375, "learning_rate": 8.905201819215464e-06, "loss": 0.6796, "step": 31312 }, { "epoch": 3.331879123217706, "grad_norm": 12.671875, "learning_rate": 8.905166287663445e-06, "loss": 0.5599, "step": 31313 }, { "epoch": 3.3319855288359226, "grad_norm": 11.7734375, "learning_rate": 8.905130756111427e-06, "loss": 0.6232, "step": 31314 }, { "epoch": 3.332091934454139, "grad_norm": 13.0390625, "learning_rate": 8.905095224559409e-06, "loss": 0.6488, "step": 31315 }, { "epoch": 3.332198340072356, "grad_norm": 10.4375, "learning_rate": 8.905059693007392e-06, "loss": 0.6558, "step": 31316 }, { "epoch": 3.3323047456905726, "grad_norm": 13.0078125, "learning_rate": 8.905024161455374e-06, "loss": 0.6588, "step": 31317 }, { "epoch": 3.332411151308789, "grad_norm": 14.3203125, "learning_rate": 8.904988629903355e-06, "loss": 0.6351, "step": 31318 }, { "epoch": 3.3325175569270056, "grad_norm": 15.390625, "learning_rate": 8.904953098351337e-06, "loss": 0.6853, "step": 31319 }, { "epoch": 3.3326239625452225, "grad_norm": 12.5546875, "learning_rate": 8.904917566799318e-06, "loss": 0.5893, "step": 31320 }, { "epoch": 3.332730368163439, "grad_norm": 11.8125, "learning_rate": 8.904882035247302e-06, "loss": 0.5441, "step": 31321 }, { "epoch": 3.3328367737816555, "grad_norm": 13.6796875, "learning_rate": 8.904846503695283e-06, "loss": 0.6793, "step": 31322 }, { "epoch": 3.3329431793998725, "grad_norm": 16.28125, "learning_rate": 8.904810972143265e-06, "loss": 0.6025, "step": 31323 }, { "epoch": 3.333049585018089, "grad_norm": 15.609375, "learning_rate": 8.904775440591246e-06, "loss": 0.636, "step": 31324 }, { "epoch": 3.3331559906363055, "grad_norm": 13.984375, "learning_rate": 8.904739909039228e-06, "loss": 0.6826, "step": 31325 }, { "epoch": 3.333262396254522, "grad_norm": 16.875, "learning_rate": 8.904704377487208e-06, "loss": 0.7705, "step": 31326 }, { "epoch": 3.333368801872739, "grad_norm": 11.0390625, "learning_rate": 8.904668845935191e-06, "loss": 0.5297, "step": 31327 }, { "epoch": 3.3334752074909555, "grad_norm": 14.21875, "learning_rate": 8.904633314383173e-06, "loss": 0.7021, "step": 31328 }, { "epoch": 3.333581613109172, "grad_norm": 14.96875, "learning_rate": 8.904597782831155e-06, "loss": 0.7395, "step": 31329 }, { "epoch": 3.333688018727389, "grad_norm": 14.359375, "learning_rate": 8.904562251279136e-06, "loss": 0.6963, "step": 31330 }, { "epoch": 3.3337944243456055, "grad_norm": 11.203125, "learning_rate": 8.904526719727118e-06, "loss": 0.6005, "step": 31331 }, { "epoch": 3.333900829963822, "grad_norm": 12.3984375, "learning_rate": 8.904491188175101e-06, "loss": 0.7438, "step": 31332 }, { "epoch": 3.334007235582039, "grad_norm": 17.078125, "learning_rate": 8.904455656623083e-06, "loss": 0.7356, "step": 31333 }, { "epoch": 3.3341136412002554, "grad_norm": 10.59375, "learning_rate": 8.904420125071064e-06, "loss": 0.6148, "step": 31334 }, { "epoch": 3.334220046818472, "grad_norm": 16.0625, "learning_rate": 8.904384593519046e-06, "loss": 0.8834, "step": 31335 }, { "epoch": 3.334326452436689, "grad_norm": 13.65625, "learning_rate": 8.904349061967027e-06, "loss": 0.6807, "step": 31336 }, { "epoch": 3.3344328580549054, "grad_norm": 15.0390625, "learning_rate": 8.904313530415009e-06, "loss": 0.7155, "step": 31337 }, { "epoch": 3.334539263673122, "grad_norm": 13.5625, "learning_rate": 8.90427799886299e-06, "loss": 0.5862, "step": 31338 }, { "epoch": 3.3346456692913384, "grad_norm": 18.453125, "learning_rate": 8.904242467310972e-06, "loss": 0.7541, "step": 31339 }, { "epoch": 3.3347520749095554, "grad_norm": 11.203125, "learning_rate": 8.904206935758954e-06, "loss": 0.5719, "step": 31340 }, { "epoch": 3.334858480527772, "grad_norm": 11.484375, "learning_rate": 8.904171404206935e-06, "loss": 0.541, "step": 31341 }, { "epoch": 3.3349648861459884, "grad_norm": 12.9921875, "learning_rate": 8.904135872654919e-06, "loss": 0.6506, "step": 31342 }, { "epoch": 3.335071291764205, "grad_norm": 11.9296875, "learning_rate": 8.9041003411029e-06, "loss": 0.5425, "step": 31343 }, { "epoch": 3.335177697382422, "grad_norm": 17.921875, "learning_rate": 8.904064809550882e-06, "loss": 0.765, "step": 31344 }, { "epoch": 3.3352841030006384, "grad_norm": 14.4765625, "learning_rate": 8.904029277998864e-06, "loss": 0.6447, "step": 31345 }, { "epoch": 3.335390508618855, "grad_norm": 9.71875, "learning_rate": 8.903993746446845e-06, "loss": 0.4923, "step": 31346 }, { "epoch": 3.335496914237072, "grad_norm": 12.4140625, "learning_rate": 8.903958214894829e-06, "loss": 0.6587, "step": 31347 }, { "epoch": 3.3356033198552884, "grad_norm": 16.984375, "learning_rate": 8.90392268334281e-06, "loss": 0.7392, "step": 31348 }, { "epoch": 3.335709725473505, "grad_norm": 13.9375, "learning_rate": 8.903887151790792e-06, "loss": 0.7102, "step": 31349 }, { "epoch": 3.335816131091722, "grad_norm": 10.0078125, "learning_rate": 8.903851620238773e-06, "loss": 0.5765, "step": 31350 }, { "epoch": 3.3359225367099383, "grad_norm": 10.046875, "learning_rate": 8.903816088686755e-06, "loss": 0.453, "step": 31351 }, { "epoch": 3.336028942328155, "grad_norm": 12.984375, "learning_rate": 8.903780557134737e-06, "loss": 0.6794, "step": 31352 }, { "epoch": 3.336135347946372, "grad_norm": 10.0234375, "learning_rate": 8.903745025582718e-06, "loss": 0.507, "step": 31353 }, { "epoch": 3.3362417535645883, "grad_norm": 15.0859375, "learning_rate": 8.9037094940307e-06, "loss": 0.641, "step": 31354 }, { "epoch": 3.336348159182805, "grad_norm": 11.125, "learning_rate": 8.903673962478681e-06, "loss": 0.5137, "step": 31355 }, { "epoch": 3.3364545648010213, "grad_norm": 10.328125, "learning_rate": 8.903638430926663e-06, "loss": 0.5827, "step": 31356 }, { "epoch": 3.3365609704192383, "grad_norm": 11.40625, "learning_rate": 8.903602899374646e-06, "loss": 0.6014, "step": 31357 }, { "epoch": 3.336667376037455, "grad_norm": 12.203125, "learning_rate": 8.903567367822628e-06, "loss": 0.6027, "step": 31358 }, { "epoch": 3.3367737816556713, "grad_norm": 15.671875, "learning_rate": 8.90353183627061e-06, "loss": 0.738, "step": 31359 }, { "epoch": 3.3368801872738882, "grad_norm": 15.9140625, "learning_rate": 8.903496304718591e-06, "loss": 0.7727, "step": 31360 }, { "epoch": 3.3369865928921048, "grad_norm": 12.4609375, "learning_rate": 8.903460773166573e-06, "loss": 0.7009, "step": 31361 }, { "epoch": 3.3370929985103213, "grad_norm": 11.5234375, "learning_rate": 8.903425241614554e-06, "loss": 0.6369, "step": 31362 }, { "epoch": 3.3371994041285378, "grad_norm": 13.671875, "learning_rate": 8.903389710062536e-06, "loss": 0.6405, "step": 31363 }, { "epoch": 3.3373058097467547, "grad_norm": 16.984375, "learning_rate": 8.903354178510518e-06, "loss": 0.7411, "step": 31364 }, { "epoch": 3.3374122153649712, "grad_norm": 13.671875, "learning_rate": 8.903318646958499e-06, "loss": 0.5734, "step": 31365 }, { "epoch": 3.3375186209831877, "grad_norm": 12.1484375, "learning_rate": 8.90328311540648e-06, "loss": 0.5212, "step": 31366 }, { "epoch": 3.3376250266014047, "grad_norm": 12.21875, "learning_rate": 8.903247583854462e-06, "loss": 0.5096, "step": 31367 }, { "epoch": 3.337731432219621, "grad_norm": 13.21875, "learning_rate": 8.903212052302446e-06, "loss": 0.5722, "step": 31368 }, { "epoch": 3.3378378378378377, "grad_norm": 16.859375, "learning_rate": 8.903176520750427e-06, "loss": 0.5858, "step": 31369 }, { "epoch": 3.3379442434560547, "grad_norm": 11.4453125, "learning_rate": 8.903140989198409e-06, "loss": 0.6548, "step": 31370 }, { "epoch": 3.338050649074271, "grad_norm": 13.8671875, "learning_rate": 8.90310545764639e-06, "loss": 0.7356, "step": 31371 }, { "epoch": 3.3381570546924877, "grad_norm": 12.375, "learning_rate": 8.903069926094372e-06, "loss": 0.5593, "step": 31372 }, { "epoch": 3.3382634603107046, "grad_norm": 15.671875, "learning_rate": 8.903034394542355e-06, "loss": 0.8373, "step": 31373 }, { "epoch": 3.338369865928921, "grad_norm": 16.78125, "learning_rate": 8.902998862990337e-06, "loss": 0.6978, "step": 31374 }, { "epoch": 3.3384762715471377, "grad_norm": 17.109375, "learning_rate": 8.902963331438319e-06, "loss": 0.7907, "step": 31375 }, { "epoch": 3.338582677165354, "grad_norm": 15.5625, "learning_rate": 8.9029277998863e-06, "loss": 0.6601, "step": 31376 }, { "epoch": 3.338689082783571, "grad_norm": 14.9140625, "learning_rate": 8.902892268334282e-06, "loss": 0.7531, "step": 31377 }, { "epoch": 3.3387954884017876, "grad_norm": 11.0078125, "learning_rate": 8.902856736782263e-06, "loss": 0.643, "step": 31378 }, { "epoch": 3.338901894020004, "grad_norm": 15.0234375, "learning_rate": 8.902821205230245e-06, "loss": 0.6412, "step": 31379 }, { "epoch": 3.3390082996382207, "grad_norm": 12.7578125, "learning_rate": 8.902785673678227e-06, "loss": 0.5794, "step": 31380 }, { "epoch": 3.3391147052564376, "grad_norm": 13.7265625, "learning_rate": 8.902750142126208e-06, "loss": 0.6291, "step": 31381 }, { "epoch": 3.339221110874654, "grad_norm": 14.3828125, "learning_rate": 8.90271461057419e-06, "loss": 0.613, "step": 31382 }, { "epoch": 3.3393275164928706, "grad_norm": 12.53125, "learning_rate": 8.902679079022173e-06, "loss": 0.6118, "step": 31383 }, { "epoch": 3.3394339221110876, "grad_norm": 16.828125, "learning_rate": 8.902643547470155e-06, "loss": 0.7953, "step": 31384 }, { "epoch": 3.339540327729304, "grad_norm": 12.0859375, "learning_rate": 8.902608015918136e-06, "loss": 0.6376, "step": 31385 }, { "epoch": 3.3396467333475206, "grad_norm": 12.234375, "learning_rate": 8.902572484366118e-06, "loss": 0.7168, "step": 31386 }, { "epoch": 3.3397531389657376, "grad_norm": 15.890625, "learning_rate": 8.9025369528141e-06, "loss": 0.7495, "step": 31387 }, { "epoch": 3.339859544583954, "grad_norm": 32.65625, "learning_rate": 8.902501421262081e-06, "loss": 0.5488, "step": 31388 }, { "epoch": 3.3399659502021706, "grad_norm": 14.9375, "learning_rate": 8.902465889710063e-06, "loss": 0.6914, "step": 31389 }, { "epoch": 3.3400723558203875, "grad_norm": 11.96875, "learning_rate": 8.902430358158044e-06, "loss": 0.7366, "step": 31390 }, { "epoch": 3.340178761438604, "grad_norm": 14.375, "learning_rate": 8.902394826606026e-06, "loss": 0.6592, "step": 31391 }, { "epoch": 3.3402851670568205, "grad_norm": 12.6484375, "learning_rate": 8.902359295054008e-06, "loss": 0.7086, "step": 31392 }, { "epoch": 3.340391572675037, "grad_norm": 12.7890625, "learning_rate": 8.902323763501991e-06, "loss": 0.7251, "step": 31393 }, { "epoch": 3.340497978293254, "grad_norm": 13.421875, "learning_rate": 8.902288231949973e-06, "loss": 0.5928, "step": 31394 }, { "epoch": 3.3406043839114705, "grad_norm": 13.7265625, "learning_rate": 8.902252700397954e-06, "loss": 0.5803, "step": 31395 }, { "epoch": 3.340710789529687, "grad_norm": 11.046875, "learning_rate": 8.902217168845936e-06, "loss": 0.5476, "step": 31396 }, { "epoch": 3.340817195147904, "grad_norm": 10.140625, "learning_rate": 8.902181637293917e-06, "loss": 0.4777, "step": 31397 }, { "epoch": 3.3409236007661205, "grad_norm": 14.8203125, "learning_rate": 8.9021461057419e-06, "loss": 0.6736, "step": 31398 }, { "epoch": 3.341030006384337, "grad_norm": 11.96875, "learning_rate": 8.902110574189882e-06, "loss": 0.6599, "step": 31399 }, { "epoch": 3.3411364120025535, "grad_norm": 13.09375, "learning_rate": 8.902075042637864e-06, "loss": 0.6589, "step": 31400 }, { "epoch": 3.3412428176207705, "grad_norm": 11.96875, "learning_rate": 8.902039511085845e-06, "loss": 0.557, "step": 31401 }, { "epoch": 3.341349223238987, "grad_norm": 10.375, "learning_rate": 8.902003979533827e-06, "loss": 0.4637, "step": 31402 }, { "epoch": 3.3414556288572035, "grad_norm": 14.4609375, "learning_rate": 8.901968447981809e-06, "loss": 0.7092, "step": 31403 }, { "epoch": 3.3415620344754204, "grad_norm": 12.4140625, "learning_rate": 8.90193291642979e-06, "loss": 0.577, "step": 31404 }, { "epoch": 3.341668440093637, "grad_norm": 14.5546875, "learning_rate": 8.901897384877772e-06, "loss": 0.5992, "step": 31405 }, { "epoch": 3.3417748457118535, "grad_norm": 10.2578125, "learning_rate": 8.901861853325753e-06, "loss": 0.5154, "step": 31406 }, { "epoch": 3.3418812513300704, "grad_norm": 11.109375, "learning_rate": 8.901826321773735e-06, "loss": 0.5215, "step": 31407 }, { "epoch": 3.341987656948287, "grad_norm": 10.296875, "learning_rate": 8.901790790221717e-06, "loss": 0.622, "step": 31408 }, { "epoch": 3.3420940625665034, "grad_norm": 13.828125, "learning_rate": 8.9017552586697e-06, "loss": 0.6624, "step": 31409 }, { "epoch": 3.3422004681847204, "grad_norm": 13.1953125, "learning_rate": 8.901719727117682e-06, "loss": 0.5963, "step": 31410 }, { "epoch": 3.342306873802937, "grad_norm": 13.2109375, "learning_rate": 8.901684195565663e-06, "loss": 0.5215, "step": 31411 }, { "epoch": 3.3424132794211534, "grad_norm": 12.84375, "learning_rate": 8.901648664013645e-06, "loss": 0.591, "step": 31412 }, { "epoch": 3.34251968503937, "grad_norm": 13.546875, "learning_rate": 8.901613132461626e-06, "loss": 0.6184, "step": 31413 }, { "epoch": 3.342626090657587, "grad_norm": 17.78125, "learning_rate": 8.90157760090961e-06, "loss": 0.8186, "step": 31414 }, { "epoch": 3.3427324962758034, "grad_norm": 14.109375, "learning_rate": 8.90154206935759e-06, "loss": 0.6857, "step": 31415 }, { "epoch": 3.34283890189402, "grad_norm": 17.40625, "learning_rate": 8.901506537805571e-06, "loss": 0.7644, "step": 31416 }, { "epoch": 3.3429453075122364, "grad_norm": 21.609375, "learning_rate": 8.901471006253553e-06, "loss": 0.6194, "step": 31417 }, { "epoch": 3.3430517131304534, "grad_norm": 18.65625, "learning_rate": 8.901435474701534e-06, "loss": 0.8514, "step": 31418 }, { "epoch": 3.34315811874867, "grad_norm": 12.7734375, "learning_rate": 8.901399943149518e-06, "loss": 0.6694, "step": 31419 }, { "epoch": 3.3432645243668864, "grad_norm": 13.2109375, "learning_rate": 8.9013644115975e-06, "loss": 0.7236, "step": 31420 }, { "epoch": 3.3433709299851033, "grad_norm": 11.03125, "learning_rate": 8.901328880045481e-06, "loss": 0.6207, "step": 31421 }, { "epoch": 3.34347733560332, "grad_norm": 15.1796875, "learning_rate": 8.901293348493463e-06, "loss": 0.6392, "step": 31422 }, { "epoch": 3.3435837412215363, "grad_norm": 11.4765625, "learning_rate": 8.901257816941444e-06, "loss": 0.5568, "step": 31423 }, { "epoch": 3.3436901468397533, "grad_norm": 10.0625, "learning_rate": 8.901222285389427e-06, "loss": 0.5025, "step": 31424 }, { "epoch": 3.34379655245797, "grad_norm": 14.0546875, "learning_rate": 8.901186753837409e-06, "loss": 0.6428, "step": 31425 }, { "epoch": 3.3439029580761863, "grad_norm": 13.640625, "learning_rate": 8.90115122228539e-06, "loss": 0.6187, "step": 31426 }, { "epoch": 3.3440093636944033, "grad_norm": 18.90625, "learning_rate": 8.901115690733372e-06, "loss": 0.9092, "step": 31427 }, { "epoch": 3.34411576931262, "grad_norm": 23.875, "learning_rate": 8.901080159181354e-06, "loss": 0.703, "step": 31428 }, { "epoch": 3.3442221749308363, "grad_norm": 15.6015625, "learning_rate": 8.901044627629336e-06, "loss": 0.6065, "step": 31429 }, { "epoch": 3.344328580549053, "grad_norm": 17.828125, "learning_rate": 8.901009096077317e-06, "loss": 0.7886, "step": 31430 }, { "epoch": 3.3444349861672698, "grad_norm": 11.0390625, "learning_rate": 8.900973564525299e-06, "loss": 0.5271, "step": 31431 }, { "epoch": 3.3445413917854863, "grad_norm": 12.8515625, "learning_rate": 8.90093803297328e-06, "loss": 0.6648, "step": 31432 }, { "epoch": 3.3446477974037028, "grad_norm": 12.1875, "learning_rate": 8.900902501421262e-06, "loss": 0.7796, "step": 31433 }, { "epoch": 3.3447542030219197, "grad_norm": 12.640625, "learning_rate": 8.900866969869245e-06, "loss": 0.7435, "step": 31434 }, { "epoch": 3.3448606086401362, "grad_norm": 11.4453125, "learning_rate": 8.900831438317227e-06, "loss": 0.582, "step": 31435 }, { "epoch": 3.3449670142583527, "grad_norm": 12.4453125, "learning_rate": 8.900795906765208e-06, "loss": 0.8324, "step": 31436 }, { "epoch": 3.3450734198765693, "grad_norm": 11.4921875, "learning_rate": 8.90076037521319e-06, "loss": 0.4859, "step": 31437 }, { "epoch": 3.345179825494786, "grad_norm": 14.25, "learning_rate": 8.900724843661172e-06, "loss": 0.6281, "step": 31438 }, { "epoch": 3.3452862311130027, "grad_norm": 13.5, "learning_rate": 8.900689312109153e-06, "loss": 0.4914, "step": 31439 }, { "epoch": 3.3453926367312192, "grad_norm": 14.9453125, "learning_rate": 8.900653780557137e-06, "loss": 0.6641, "step": 31440 }, { "epoch": 3.345499042349436, "grad_norm": 11.6171875, "learning_rate": 8.900618249005118e-06, "loss": 0.535, "step": 31441 }, { "epoch": 3.3456054479676527, "grad_norm": 12.3125, "learning_rate": 8.9005827174531e-06, "loss": 0.6092, "step": 31442 }, { "epoch": 3.345711853585869, "grad_norm": 13.4921875, "learning_rate": 8.900547185901081e-06, "loss": 0.6644, "step": 31443 }, { "epoch": 3.345818259204086, "grad_norm": 13.1640625, "learning_rate": 8.900511654349063e-06, "loss": 0.689, "step": 31444 }, { "epoch": 3.3459246648223027, "grad_norm": 13.7421875, "learning_rate": 8.900476122797045e-06, "loss": 0.6524, "step": 31445 }, { "epoch": 3.346031070440519, "grad_norm": 11.9453125, "learning_rate": 8.900440591245026e-06, "loss": 0.6715, "step": 31446 }, { "epoch": 3.346137476058736, "grad_norm": 15.8515625, "learning_rate": 8.900405059693008e-06, "loss": 0.6414, "step": 31447 }, { "epoch": 3.3462438816769526, "grad_norm": 10.25, "learning_rate": 8.90036952814099e-06, "loss": 0.6039, "step": 31448 }, { "epoch": 3.346350287295169, "grad_norm": 9.8125, "learning_rate": 8.900333996588971e-06, "loss": 0.5797, "step": 31449 }, { "epoch": 3.3464566929133857, "grad_norm": 14.4921875, "learning_rate": 8.900298465036954e-06, "loss": 0.65, "step": 31450 }, { "epoch": 3.3465630985316026, "grad_norm": 10.671875, "learning_rate": 8.900262933484936e-06, "loss": 0.4976, "step": 31451 }, { "epoch": 3.346669504149819, "grad_norm": 15.8984375, "learning_rate": 8.900227401932918e-06, "loss": 0.7003, "step": 31452 }, { "epoch": 3.3467759097680356, "grad_norm": 14.890625, "learning_rate": 8.9001918703809e-06, "loss": 0.7115, "step": 31453 }, { "epoch": 3.3468823153862526, "grad_norm": 14.328125, "learning_rate": 8.90015633882888e-06, "loss": 0.7573, "step": 31454 }, { "epoch": 3.346988721004469, "grad_norm": 11.359375, "learning_rate": 8.900120807276862e-06, "loss": 0.5626, "step": 31455 }, { "epoch": 3.3470951266226856, "grad_norm": 16.890625, "learning_rate": 8.900085275724844e-06, "loss": 0.6293, "step": 31456 }, { "epoch": 3.347201532240902, "grad_norm": 12.859375, "learning_rate": 8.900049744172826e-06, "loss": 0.6724, "step": 31457 }, { "epoch": 3.347307937859119, "grad_norm": 11.328125, "learning_rate": 8.900014212620807e-06, "loss": 0.6868, "step": 31458 }, { "epoch": 3.3474143434773356, "grad_norm": 15.109375, "learning_rate": 8.899978681068789e-06, "loss": 0.8048, "step": 31459 }, { "epoch": 3.347520749095552, "grad_norm": 14.390625, "learning_rate": 8.899943149516772e-06, "loss": 0.6591, "step": 31460 }, { "epoch": 3.347627154713769, "grad_norm": 12.5859375, "learning_rate": 8.899907617964754e-06, "loss": 0.5761, "step": 31461 }, { "epoch": 3.3477335603319855, "grad_norm": 12.28125, "learning_rate": 8.899872086412735e-06, "loss": 0.6208, "step": 31462 }, { "epoch": 3.347839965950202, "grad_norm": 12.8671875, "learning_rate": 8.899836554860717e-06, "loss": 0.6872, "step": 31463 }, { "epoch": 3.347946371568419, "grad_norm": 9.328125, "learning_rate": 8.899801023308699e-06, "loss": 0.563, "step": 31464 }, { "epoch": 3.3480527771866355, "grad_norm": 10.3671875, "learning_rate": 8.899765491756682e-06, "loss": 0.6037, "step": 31465 }, { "epoch": 3.348159182804852, "grad_norm": 16.71875, "learning_rate": 8.899729960204663e-06, "loss": 0.835, "step": 31466 }, { "epoch": 3.348265588423069, "grad_norm": 12.7578125, "learning_rate": 8.899694428652645e-06, "loss": 0.6678, "step": 31467 }, { "epoch": 3.3483719940412855, "grad_norm": 16.796875, "learning_rate": 8.899658897100627e-06, "loss": 0.7064, "step": 31468 }, { "epoch": 3.348478399659502, "grad_norm": 15.296875, "learning_rate": 8.899623365548608e-06, "loss": 0.6637, "step": 31469 }, { "epoch": 3.3485848052777185, "grad_norm": 13.8515625, "learning_rate": 8.89958783399659e-06, "loss": 0.6037, "step": 31470 }, { "epoch": 3.3486912108959355, "grad_norm": 18.65625, "learning_rate": 8.899552302444571e-06, "loss": 0.8939, "step": 31471 }, { "epoch": 3.348797616514152, "grad_norm": 13.3984375, "learning_rate": 8.899516770892553e-06, "loss": 0.7622, "step": 31472 }, { "epoch": 3.3489040221323685, "grad_norm": 12.1171875, "learning_rate": 8.899481239340535e-06, "loss": 0.6157, "step": 31473 }, { "epoch": 3.349010427750585, "grad_norm": 11.046875, "learning_rate": 8.899445707788516e-06, "loss": 0.5742, "step": 31474 }, { "epoch": 3.349116833368802, "grad_norm": 11.7265625, "learning_rate": 8.8994101762365e-06, "loss": 0.491, "step": 31475 }, { "epoch": 3.3492232389870185, "grad_norm": 12.65625, "learning_rate": 8.899374644684481e-06, "loss": 0.6027, "step": 31476 }, { "epoch": 3.349329644605235, "grad_norm": 12.6953125, "learning_rate": 8.899339113132463e-06, "loss": 0.7145, "step": 31477 }, { "epoch": 3.349436050223452, "grad_norm": 13.4140625, "learning_rate": 8.899303581580444e-06, "loss": 0.7382, "step": 31478 }, { "epoch": 3.3495424558416684, "grad_norm": 19.90625, "learning_rate": 8.899268050028426e-06, "loss": 0.8209, "step": 31479 }, { "epoch": 3.349648861459885, "grad_norm": 15.4453125, "learning_rate": 8.899232518476408e-06, "loss": 0.7813, "step": 31480 }, { "epoch": 3.349755267078102, "grad_norm": 11.2890625, "learning_rate": 8.89919698692439e-06, "loss": 0.695, "step": 31481 }, { "epoch": 3.3498616726963184, "grad_norm": 10.2265625, "learning_rate": 8.89916145537237e-06, "loss": 0.5549, "step": 31482 }, { "epoch": 3.349968078314535, "grad_norm": 11.4140625, "learning_rate": 8.899125923820352e-06, "loss": 0.6058, "step": 31483 }, { "epoch": 3.350074483932752, "grad_norm": 17.421875, "learning_rate": 8.899090392268334e-06, "loss": 0.539, "step": 31484 }, { "epoch": 3.3501808895509684, "grad_norm": 15.9375, "learning_rate": 8.899054860716316e-06, "loss": 0.7103, "step": 31485 }, { "epoch": 3.350287295169185, "grad_norm": 10.9296875, "learning_rate": 8.899019329164299e-06, "loss": 0.6436, "step": 31486 }, { "epoch": 3.3503937007874014, "grad_norm": 11.5625, "learning_rate": 8.89898379761228e-06, "loss": 0.6099, "step": 31487 }, { "epoch": 3.3505001064056184, "grad_norm": 9.6875, "learning_rate": 8.898948266060262e-06, "loss": 0.5442, "step": 31488 }, { "epoch": 3.350606512023835, "grad_norm": 16.75, "learning_rate": 8.898912734508244e-06, "loss": 0.7663, "step": 31489 }, { "epoch": 3.3507129176420514, "grad_norm": 15.859375, "learning_rate": 8.898877202956225e-06, "loss": 0.7023, "step": 31490 }, { "epoch": 3.3508193232602683, "grad_norm": 12.125, "learning_rate": 8.898841671404209e-06, "loss": 0.689, "step": 31491 }, { "epoch": 3.350925728878485, "grad_norm": 10.671875, "learning_rate": 8.89880613985219e-06, "loss": 0.4947, "step": 31492 }, { "epoch": 3.3510321344967013, "grad_norm": 11.2890625, "learning_rate": 8.898770608300172e-06, "loss": 0.5026, "step": 31493 }, { "epoch": 3.351138540114918, "grad_norm": 14.3984375, "learning_rate": 8.898735076748153e-06, "loss": 0.6882, "step": 31494 }, { "epoch": 3.351244945733135, "grad_norm": 14.2265625, "learning_rate": 8.898699545196135e-06, "loss": 0.7552, "step": 31495 }, { "epoch": 3.3513513513513513, "grad_norm": 9.46875, "learning_rate": 8.898664013644117e-06, "loss": 0.5395, "step": 31496 }, { "epoch": 3.351457756969568, "grad_norm": 11.65625, "learning_rate": 8.898628482092098e-06, "loss": 0.5482, "step": 31497 }, { "epoch": 3.351564162587785, "grad_norm": 12.21875, "learning_rate": 8.89859295054008e-06, "loss": 0.5882, "step": 31498 }, { "epoch": 3.3516705682060013, "grad_norm": 9.1484375, "learning_rate": 8.898557418988062e-06, "loss": 0.5377, "step": 31499 }, { "epoch": 3.351776973824218, "grad_norm": 11.2734375, "learning_rate": 8.898521887436043e-06, "loss": 0.5287, "step": 31500 }, { "epoch": 3.3518833794424348, "grad_norm": 12.78125, "learning_rate": 8.898486355884026e-06, "loss": 0.5896, "step": 31501 }, { "epoch": 3.3519897850606513, "grad_norm": 12.515625, "learning_rate": 8.898450824332008e-06, "loss": 0.5557, "step": 31502 }, { "epoch": 3.3520961906788678, "grad_norm": 12.9921875, "learning_rate": 8.89841529277999e-06, "loss": 0.5946, "step": 31503 }, { "epoch": 3.3522025962970847, "grad_norm": 13.0859375, "learning_rate": 8.898379761227971e-06, "loss": 0.813, "step": 31504 }, { "epoch": 3.3523090019153012, "grad_norm": 11.4609375, "learning_rate": 8.898344229675953e-06, "loss": 0.7221, "step": 31505 }, { "epoch": 3.3524154075335177, "grad_norm": 12.53125, "learning_rate": 8.898308698123934e-06, "loss": 0.6147, "step": 31506 }, { "epoch": 3.3525218131517343, "grad_norm": 11.9765625, "learning_rate": 8.898273166571916e-06, "loss": 0.59, "step": 31507 }, { "epoch": 3.352628218769951, "grad_norm": 10.6328125, "learning_rate": 8.898237635019898e-06, "loss": 0.6245, "step": 31508 }, { "epoch": 3.3527346243881677, "grad_norm": 9.9765625, "learning_rate": 8.89820210346788e-06, "loss": 0.509, "step": 31509 }, { "epoch": 3.3528410300063842, "grad_norm": 12.125, "learning_rate": 8.898166571915861e-06, "loss": 0.537, "step": 31510 }, { "epoch": 3.3529474356246007, "grad_norm": 13.96875, "learning_rate": 8.898131040363844e-06, "loss": 0.7108, "step": 31511 }, { "epoch": 3.3530538412428177, "grad_norm": 16.640625, "learning_rate": 8.898095508811826e-06, "loss": 0.8744, "step": 31512 }, { "epoch": 3.353160246861034, "grad_norm": 15.8203125, "learning_rate": 8.898059977259807e-06, "loss": 0.6613, "step": 31513 }, { "epoch": 3.3532666524792507, "grad_norm": 18.3125, "learning_rate": 8.898024445707789e-06, "loss": 0.8747, "step": 31514 }, { "epoch": 3.3533730580974677, "grad_norm": 12.734375, "learning_rate": 8.89798891415577e-06, "loss": 0.5062, "step": 31515 }, { "epoch": 3.353479463715684, "grad_norm": 17.90625, "learning_rate": 8.897953382603754e-06, "loss": 0.797, "step": 31516 }, { "epoch": 3.3535858693339007, "grad_norm": 13.015625, "learning_rate": 8.897917851051736e-06, "loss": 0.693, "step": 31517 }, { "epoch": 3.3536922749521176, "grad_norm": 11.4609375, "learning_rate": 8.897882319499717e-06, "loss": 0.5975, "step": 31518 }, { "epoch": 3.353798680570334, "grad_norm": 11.734375, "learning_rate": 8.897846787947699e-06, "loss": 0.701, "step": 31519 }, { "epoch": 3.3539050861885507, "grad_norm": 15.6484375, "learning_rate": 8.89781125639568e-06, "loss": 0.646, "step": 31520 }, { "epoch": 3.3540114918067676, "grad_norm": 15.421875, "learning_rate": 8.897775724843662e-06, "loss": 0.774, "step": 31521 }, { "epoch": 3.354117897424984, "grad_norm": 12.234375, "learning_rate": 8.897740193291644e-06, "loss": 0.5594, "step": 31522 }, { "epoch": 3.3542243030432006, "grad_norm": 13.9609375, "learning_rate": 8.897704661739625e-06, "loss": 0.6152, "step": 31523 }, { "epoch": 3.354330708661417, "grad_norm": 14.5546875, "learning_rate": 8.897669130187607e-06, "loss": 0.5459, "step": 31524 }, { "epoch": 3.354437114279634, "grad_norm": 10.9921875, "learning_rate": 8.897633598635588e-06, "loss": 0.5097, "step": 31525 }, { "epoch": 3.3545435198978506, "grad_norm": 12.2421875, "learning_rate": 8.89759806708357e-06, "loss": 0.6608, "step": 31526 }, { "epoch": 3.354649925516067, "grad_norm": 11.1328125, "learning_rate": 8.897562535531553e-06, "loss": 0.5829, "step": 31527 }, { "epoch": 3.354756331134284, "grad_norm": 14.9609375, "learning_rate": 8.897527003979535e-06, "loss": 0.5791, "step": 31528 }, { "epoch": 3.3548627367525006, "grad_norm": 12.7421875, "learning_rate": 8.897491472427517e-06, "loss": 0.7024, "step": 31529 }, { "epoch": 3.354969142370717, "grad_norm": 11.796875, "learning_rate": 8.897455940875498e-06, "loss": 0.5886, "step": 31530 }, { "epoch": 3.3550755479889336, "grad_norm": 12.765625, "learning_rate": 8.89742040932348e-06, "loss": 0.5078, "step": 31531 }, { "epoch": 3.3551819536071505, "grad_norm": 18.921875, "learning_rate": 8.897384877771463e-06, "loss": 0.836, "step": 31532 }, { "epoch": 3.355288359225367, "grad_norm": 11.609375, "learning_rate": 8.897349346219445e-06, "loss": 0.5761, "step": 31533 }, { "epoch": 3.3553947648435836, "grad_norm": 9.109375, "learning_rate": 8.897313814667425e-06, "loss": 0.4752, "step": 31534 }, { "epoch": 3.3555011704618005, "grad_norm": 14.7265625, "learning_rate": 8.897278283115406e-06, "loss": 0.7078, "step": 31535 }, { "epoch": 3.355607576080017, "grad_norm": 11.84375, "learning_rate": 8.897242751563388e-06, "loss": 0.5102, "step": 31536 }, { "epoch": 3.3557139816982335, "grad_norm": 14.2265625, "learning_rate": 8.897207220011371e-06, "loss": 0.7449, "step": 31537 }, { "epoch": 3.3558203873164505, "grad_norm": 15.21875, "learning_rate": 8.897171688459353e-06, "loss": 0.7022, "step": 31538 }, { "epoch": 3.355926792934667, "grad_norm": 11.4140625, "learning_rate": 8.897136156907334e-06, "loss": 0.61, "step": 31539 }, { "epoch": 3.3560331985528835, "grad_norm": 12.9609375, "learning_rate": 8.897100625355316e-06, "loss": 0.6851, "step": 31540 }, { "epoch": 3.3561396041711005, "grad_norm": 16.03125, "learning_rate": 8.897065093803297e-06, "loss": 0.6561, "step": 31541 }, { "epoch": 3.356246009789317, "grad_norm": 17.1875, "learning_rate": 8.89702956225128e-06, "loss": 0.6688, "step": 31542 }, { "epoch": 3.3563524154075335, "grad_norm": 16.28125, "learning_rate": 8.896994030699262e-06, "loss": 0.5827, "step": 31543 }, { "epoch": 3.35645882102575, "grad_norm": 13.40625, "learning_rate": 8.896958499147244e-06, "loss": 0.6305, "step": 31544 }, { "epoch": 3.356565226643967, "grad_norm": 11.875, "learning_rate": 8.896922967595226e-06, "loss": 0.5926, "step": 31545 }, { "epoch": 3.3566716322621835, "grad_norm": 15.0078125, "learning_rate": 8.896887436043207e-06, "loss": 0.7592, "step": 31546 }, { "epoch": 3.3567780378804, "grad_norm": 13.6484375, "learning_rate": 8.896851904491189e-06, "loss": 0.7078, "step": 31547 }, { "epoch": 3.356884443498617, "grad_norm": 13.921875, "learning_rate": 8.89681637293917e-06, "loss": 0.6222, "step": 31548 }, { "epoch": 3.3569908491168334, "grad_norm": 13.2734375, "learning_rate": 8.896780841387152e-06, "loss": 0.6908, "step": 31549 }, { "epoch": 3.35709725473505, "grad_norm": 13.0, "learning_rate": 8.896745309835134e-06, "loss": 0.5518, "step": 31550 }, { "epoch": 3.3572036603532665, "grad_norm": 14.0703125, "learning_rate": 8.896709778283115e-06, "loss": 0.5652, "step": 31551 }, { "epoch": 3.3573100659714834, "grad_norm": 15.7265625, "learning_rate": 8.896674246731099e-06, "loss": 0.6255, "step": 31552 }, { "epoch": 3.3574164715897, "grad_norm": 16.578125, "learning_rate": 8.89663871517908e-06, "loss": 0.577, "step": 31553 }, { "epoch": 3.3575228772079164, "grad_norm": 26.078125, "learning_rate": 8.896603183627062e-06, "loss": 0.549, "step": 31554 }, { "epoch": 3.3576292828261334, "grad_norm": 12.9140625, "learning_rate": 8.896567652075043e-06, "loss": 0.5555, "step": 31555 }, { "epoch": 3.35773568844435, "grad_norm": 11.5546875, "learning_rate": 8.896532120523025e-06, "loss": 0.6175, "step": 31556 }, { "epoch": 3.3578420940625664, "grad_norm": 13.9765625, "learning_rate": 8.896496588971007e-06, "loss": 0.5983, "step": 31557 }, { "epoch": 3.3579484996807833, "grad_norm": 11.140625, "learning_rate": 8.89646105741899e-06, "loss": 0.5465, "step": 31558 }, { "epoch": 3.358054905299, "grad_norm": 13.5234375, "learning_rate": 8.896425525866971e-06, "loss": 0.5622, "step": 31559 }, { "epoch": 3.3581613109172164, "grad_norm": 12.3203125, "learning_rate": 8.896389994314953e-06, "loss": 0.5722, "step": 31560 }, { "epoch": 3.358267716535433, "grad_norm": 15.828125, "learning_rate": 8.896354462762935e-06, "loss": 0.6492, "step": 31561 }, { "epoch": 3.35837412215365, "grad_norm": 9.2890625, "learning_rate": 8.896318931210916e-06, "loss": 0.4182, "step": 31562 }, { "epoch": 3.3584805277718663, "grad_norm": 12.828125, "learning_rate": 8.896283399658898e-06, "loss": 0.6375, "step": 31563 }, { "epoch": 3.358586933390083, "grad_norm": 12.984375, "learning_rate": 8.89624786810688e-06, "loss": 0.4902, "step": 31564 }, { "epoch": 3.3586933390083, "grad_norm": 12.265625, "learning_rate": 8.896212336554861e-06, "loss": 0.637, "step": 31565 }, { "epoch": 3.3587997446265163, "grad_norm": 14.1796875, "learning_rate": 8.896176805002843e-06, "loss": 0.5854, "step": 31566 }, { "epoch": 3.358906150244733, "grad_norm": 14.7734375, "learning_rate": 8.896141273450824e-06, "loss": 0.6613, "step": 31567 }, { "epoch": 3.3590125558629493, "grad_norm": 18.9375, "learning_rate": 8.896105741898808e-06, "loss": 0.7149, "step": 31568 }, { "epoch": 3.3591189614811663, "grad_norm": 10.625, "learning_rate": 8.89607021034679e-06, "loss": 0.5282, "step": 31569 }, { "epoch": 3.359225367099383, "grad_norm": 12.2421875, "learning_rate": 8.896034678794771e-06, "loss": 0.5902, "step": 31570 }, { "epoch": 3.3593317727175993, "grad_norm": 14.25, "learning_rate": 8.895999147242752e-06, "loss": 0.6864, "step": 31571 }, { "epoch": 3.3594381783358163, "grad_norm": 11.3203125, "learning_rate": 8.895963615690734e-06, "loss": 0.5514, "step": 31572 }, { "epoch": 3.3595445839540328, "grad_norm": 11.765625, "learning_rate": 8.895928084138716e-06, "loss": 0.7343, "step": 31573 }, { "epoch": 3.3596509895722493, "grad_norm": 12.6796875, "learning_rate": 8.895892552586697e-06, "loss": 0.7088, "step": 31574 }, { "epoch": 3.3597573951904662, "grad_norm": 13.6953125, "learning_rate": 8.895857021034679e-06, "loss": 0.7858, "step": 31575 }, { "epoch": 3.3598638008086827, "grad_norm": 11.7734375, "learning_rate": 8.89582148948266e-06, "loss": 0.566, "step": 31576 }, { "epoch": 3.3599702064268993, "grad_norm": 18.6875, "learning_rate": 8.895785957930642e-06, "loss": 0.7062, "step": 31577 }, { "epoch": 3.360076612045116, "grad_norm": 15.109375, "learning_rate": 8.895750426378625e-06, "loss": 0.6091, "step": 31578 }, { "epoch": 3.3601830176633327, "grad_norm": 17.125, "learning_rate": 8.895714894826607e-06, "loss": 0.7564, "step": 31579 }, { "epoch": 3.3602894232815492, "grad_norm": 11.359375, "learning_rate": 8.895679363274589e-06, "loss": 0.6127, "step": 31580 }, { "epoch": 3.3603958288997657, "grad_norm": 10.8828125, "learning_rate": 8.89564383172257e-06, "loss": 0.6275, "step": 31581 }, { "epoch": 3.3605022345179827, "grad_norm": 11.671875, "learning_rate": 8.895608300170552e-06, "loss": 0.4745, "step": 31582 }, { "epoch": 3.360608640136199, "grad_norm": 16.375, "learning_rate": 8.895572768618535e-06, "loss": 0.6512, "step": 31583 }, { "epoch": 3.3607150457544157, "grad_norm": 14.1484375, "learning_rate": 8.895537237066517e-06, "loss": 0.6278, "step": 31584 }, { "epoch": 3.3608214513726327, "grad_norm": 16.65625, "learning_rate": 8.895501705514498e-06, "loss": 0.6452, "step": 31585 }, { "epoch": 3.360927856990849, "grad_norm": 16.796875, "learning_rate": 8.89546617396248e-06, "loss": 0.6912, "step": 31586 }, { "epoch": 3.3610342626090657, "grad_norm": 9.140625, "learning_rate": 8.895430642410462e-06, "loss": 0.5306, "step": 31587 }, { "epoch": 3.361140668227282, "grad_norm": 12.75, "learning_rate": 8.895395110858443e-06, "loss": 0.5307, "step": 31588 }, { "epoch": 3.361247073845499, "grad_norm": 13.546875, "learning_rate": 8.895359579306425e-06, "loss": 0.7337, "step": 31589 }, { "epoch": 3.3613534794637157, "grad_norm": 11.9453125, "learning_rate": 8.895324047754406e-06, "loss": 0.5885, "step": 31590 }, { "epoch": 3.361459885081932, "grad_norm": 13.9140625, "learning_rate": 8.895288516202388e-06, "loss": 0.7643, "step": 31591 }, { "epoch": 3.361566290700149, "grad_norm": 14.078125, "learning_rate": 8.89525298465037e-06, "loss": 0.7229, "step": 31592 }, { "epoch": 3.3616726963183656, "grad_norm": 14.171875, "learning_rate": 8.895217453098353e-06, "loss": 0.6253, "step": 31593 }, { "epoch": 3.361779101936582, "grad_norm": 16.15625, "learning_rate": 8.895181921546334e-06, "loss": 0.7178, "step": 31594 }, { "epoch": 3.361885507554799, "grad_norm": 13.71875, "learning_rate": 8.895146389994316e-06, "loss": 0.7224, "step": 31595 }, { "epoch": 3.3619919131730156, "grad_norm": 12.8828125, "learning_rate": 8.895110858442298e-06, "loss": 0.6544, "step": 31596 }, { "epoch": 3.362098318791232, "grad_norm": 11.5234375, "learning_rate": 8.89507532689028e-06, "loss": 0.5585, "step": 31597 }, { "epoch": 3.362204724409449, "grad_norm": 13.4765625, "learning_rate": 8.895039795338261e-06, "loss": 0.6636, "step": 31598 }, { "epoch": 3.3623111300276656, "grad_norm": 18.109375, "learning_rate": 8.895004263786243e-06, "loss": 0.7664, "step": 31599 }, { "epoch": 3.362417535645882, "grad_norm": 14.0859375, "learning_rate": 8.894968732234224e-06, "loss": 0.7416, "step": 31600 }, { "epoch": 3.3625239412640986, "grad_norm": 14.6015625, "learning_rate": 8.894933200682206e-06, "loss": 0.6225, "step": 31601 }, { "epoch": 3.3626303468823155, "grad_norm": 17.171875, "learning_rate": 8.894897669130187e-06, "loss": 0.7441, "step": 31602 }, { "epoch": 3.362736752500532, "grad_norm": 12.546875, "learning_rate": 8.894862137578169e-06, "loss": 0.5764, "step": 31603 }, { "epoch": 3.3628431581187486, "grad_norm": 13.1328125, "learning_rate": 8.894826606026152e-06, "loss": 0.7191, "step": 31604 }, { "epoch": 3.362949563736965, "grad_norm": 11.5, "learning_rate": 8.894791074474134e-06, "loss": 0.5934, "step": 31605 }, { "epoch": 3.363055969355182, "grad_norm": 14.015625, "learning_rate": 8.894755542922115e-06, "loss": 0.8595, "step": 31606 }, { "epoch": 3.3631623749733985, "grad_norm": 10.6015625, "learning_rate": 8.894720011370097e-06, "loss": 0.4537, "step": 31607 }, { "epoch": 3.363268780591615, "grad_norm": 9.78125, "learning_rate": 8.894684479818079e-06, "loss": 0.5746, "step": 31608 }, { "epoch": 3.363375186209832, "grad_norm": 12.0546875, "learning_rate": 8.894648948266062e-06, "loss": 0.6617, "step": 31609 }, { "epoch": 3.3634815918280485, "grad_norm": 15.0625, "learning_rate": 8.894613416714044e-06, "loss": 0.6962, "step": 31610 }, { "epoch": 3.363587997446265, "grad_norm": 12.796875, "learning_rate": 8.894577885162025e-06, "loss": 0.6216, "step": 31611 }, { "epoch": 3.363694403064482, "grad_norm": 15.6796875, "learning_rate": 8.894542353610007e-06, "loss": 0.7332, "step": 31612 }, { "epoch": 3.3638008086826985, "grad_norm": 10.3203125, "learning_rate": 8.894506822057988e-06, "loss": 0.5091, "step": 31613 }, { "epoch": 3.363907214300915, "grad_norm": 12.1640625, "learning_rate": 8.89447129050597e-06, "loss": 0.7509, "step": 31614 }, { "epoch": 3.364013619919132, "grad_norm": 12.25, "learning_rate": 8.894435758953952e-06, "loss": 0.7201, "step": 31615 }, { "epoch": 3.3641200255373485, "grad_norm": 11.578125, "learning_rate": 8.894400227401933e-06, "loss": 0.7042, "step": 31616 }, { "epoch": 3.364226431155565, "grad_norm": 12.6328125, "learning_rate": 8.894364695849915e-06, "loss": 0.6578, "step": 31617 }, { "epoch": 3.3643328367737815, "grad_norm": 10.78125, "learning_rate": 8.894329164297896e-06, "loss": 0.5627, "step": 31618 }, { "epoch": 3.3644392423919984, "grad_norm": 12.1953125, "learning_rate": 8.89429363274588e-06, "loss": 0.6065, "step": 31619 }, { "epoch": 3.364545648010215, "grad_norm": 12.1953125, "learning_rate": 8.894258101193861e-06, "loss": 0.5477, "step": 31620 }, { "epoch": 3.3646520536284314, "grad_norm": 10.984375, "learning_rate": 8.894222569641843e-06, "loss": 0.594, "step": 31621 }, { "epoch": 3.3647584592466484, "grad_norm": 10.671875, "learning_rate": 8.894187038089825e-06, "loss": 0.5753, "step": 31622 }, { "epoch": 3.364864864864865, "grad_norm": 15.390625, "learning_rate": 8.894151506537806e-06, "loss": 0.6256, "step": 31623 }, { "epoch": 3.3649712704830814, "grad_norm": 9.2890625, "learning_rate": 8.894115974985788e-06, "loss": 0.5017, "step": 31624 }, { "epoch": 3.365077676101298, "grad_norm": 12.53125, "learning_rate": 8.89408044343377e-06, "loss": 0.5581, "step": 31625 }, { "epoch": 3.365184081719515, "grad_norm": 10.203125, "learning_rate": 8.894044911881751e-06, "loss": 0.5698, "step": 31626 }, { "epoch": 3.3652904873377314, "grad_norm": 15.7109375, "learning_rate": 8.894009380329733e-06, "loss": 0.8251, "step": 31627 }, { "epoch": 3.365396892955948, "grad_norm": 12.3359375, "learning_rate": 8.893973848777714e-06, "loss": 0.6836, "step": 31628 }, { "epoch": 3.365503298574165, "grad_norm": 12.0234375, "learning_rate": 8.893938317225698e-06, "loss": 0.6622, "step": 31629 }, { "epoch": 3.3656097041923814, "grad_norm": 10.3203125, "learning_rate": 8.893902785673679e-06, "loss": 0.523, "step": 31630 }, { "epoch": 3.365716109810598, "grad_norm": 14.296875, "learning_rate": 8.89386725412166e-06, "loss": 0.6723, "step": 31631 }, { "epoch": 3.365822515428815, "grad_norm": 13.4765625, "learning_rate": 8.893831722569642e-06, "loss": 0.6112, "step": 31632 }, { "epoch": 3.3659289210470313, "grad_norm": 14.578125, "learning_rate": 8.893796191017624e-06, "loss": 0.6794, "step": 31633 }, { "epoch": 3.366035326665248, "grad_norm": 11.4140625, "learning_rate": 8.893760659465607e-06, "loss": 0.5766, "step": 31634 }, { "epoch": 3.366141732283465, "grad_norm": 11.328125, "learning_rate": 8.893725127913589e-06, "loss": 0.6539, "step": 31635 }, { "epoch": 3.3662481379016813, "grad_norm": 15.1796875, "learning_rate": 8.89368959636157e-06, "loss": 0.7717, "step": 31636 }, { "epoch": 3.366354543519898, "grad_norm": 13.5234375, "learning_rate": 8.893654064809552e-06, "loss": 0.5525, "step": 31637 }, { "epoch": 3.3664609491381143, "grad_norm": 13.3359375, "learning_rate": 8.893618533257534e-06, "loss": 0.5619, "step": 31638 }, { "epoch": 3.3665673547563313, "grad_norm": 15.640625, "learning_rate": 8.893583001705515e-06, "loss": 0.7817, "step": 31639 }, { "epoch": 3.366673760374548, "grad_norm": 11.859375, "learning_rate": 8.893547470153497e-06, "loss": 0.5841, "step": 31640 }, { "epoch": 3.3667801659927643, "grad_norm": 14.4765625, "learning_rate": 8.893511938601478e-06, "loss": 0.6569, "step": 31641 }, { "epoch": 3.366886571610981, "grad_norm": 11.140625, "learning_rate": 8.89347640704946e-06, "loss": 0.6708, "step": 31642 }, { "epoch": 3.3669929772291978, "grad_norm": 12.8515625, "learning_rate": 8.893440875497442e-06, "loss": 0.6691, "step": 31643 }, { "epoch": 3.3670993828474143, "grad_norm": 15.859375, "learning_rate": 8.893405343945423e-06, "loss": 0.7687, "step": 31644 }, { "epoch": 3.367205788465631, "grad_norm": 12.21875, "learning_rate": 8.893369812393407e-06, "loss": 0.6412, "step": 31645 }, { "epoch": 3.3673121940838477, "grad_norm": 14.125, "learning_rate": 8.893334280841388e-06, "loss": 0.5873, "step": 31646 }, { "epoch": 3.3674185997020643, "grad_norm": 10.609375, "learning_rate": 8.89329874928937e-06, "loss": 0.6087, "step": 31647 }, { "epoch": 3.3675250053202808, "grad_norm": 18.578125, "learning_rate": 8.893263217737351e-06, "loss": 0.758, "step": 31648 }, { "epoch": 3.3676314109384977, "grad_norm": 12.203125, "learning_rate": 8.893227686185333e-06, "loss": 0.6075, "step": 31649 }, { "epoch": 3.3677378165567142, "grad_norm": 13.109375, "learning_rate": 8.893192154633316e-06, "loss": 0.5922, "step": 31650 }, { "epoch": 3.3678442221749307, "grad_norm": 13.7265625, "learning_rate": 8.893156623081298e-06, "loss": 0.6123, "step": 31651 }, { "epoch": 3.3679506277931477, "grad_norm": 12.8359375, "learning_rate": 8.89312109152928e-06, "loss": 0.6832, "step": 31652 }, { "epoch": 3.368057033411364, "grad_norm": 11.8671875, "learning_rate": 8.893085559977261e-06, "loss": 0.6524, "step": 31653 }, { "epoch": 3.3681634390295807, "grad_norm": 13.34375, "learning_rate": 8.893050028425241e-06, "loss": 0.543, "step": 31654 }, { "epoch": 3.368269844647797, "grad_norm": 12.359375, "learning_rate": 8.893014496873224e-06, "loss": 0.552, "step": 31655 }, { "epoch": 3.368376250266014, "grad_norm": 14.453125, "learning_rate": 8.892978965321206e-06, "loss": 0.6511, "step": 31656 }, { "epoch": 3.3684826558842307, "grad_norm": 9.9296875, "learning_rate": 8.892943433769188e-06, "loss": 0.4254, "step": 31657 }, { "epoch": 3.368589061502447, "grad_norm": 13.4765625, "learning_rate": 8.89290790221717e-06, "loss": 0.6968, "step": 31658 }, { "epoch": 3.368695467120664, "grad_norm": 14.203125, "learning_rate": 8.89287237066515e-06, "loss": 0.6951, "step": 31659 }, { "epoch": 3.3688018727388807, "grad_norm": 19.609375, "learning_rate": 8.892836839113134e-06, "loss": 0.9126, "step": 31660 }, { "epoch": 3.368908278357097, "grad_norm": 12.71875, "learning_rate": 8.892801307561116e-06, "loss": 0.5972, "step": 31661 }, { "epoch": 3.3690146839753137, "grad_norm": 15.5859375, "learning_rate": 8.892765776009097e-06, "loss": 0.7078, "step": 31662 }, { "epoch": 3.3691210895935306, "grad_norm": 17.515625, "learning_rate": 8.892730244457079e-06, "loss": 0.6872, "step": 31663 }, { "epoch": 3.369227495211747, "grad_norm": 13.984375, "learning_rate": 8.89269471290506e-06, "loss": 0.6201, "step": 31664 }, { "epoch": 3.3693339008299636, "grad_norm": 18.15625, "learning_rate": 8.892659181353042e-06, "loss": 0.7288, "step": 31665 }, { "epoch": 3.3694403064481806, "grad_norm": 16.171875, "learning_rate": 8.892623649801024e-06, "loss": 0.6908, "step": 31666 }, { "epoch": 3.369546712066397, "grad_norm": 16.5625, "learning_rate": 8.892588118249005e-06, "loss": 0.7355, "step": 31667 }, { "epoch": 3.3696531176846136, "grad_norm": 10.7890625, "learning_rate": 8.892552586696987e-06, "loss": 0.5759, "step": 31668 }, { "epoch": 3.3697595233028306, "grad_norm": 8.96875, "learning_rate": 8.892517055144969e-06, "loss": 0.5147, "step": 31669 }, { "epoch": 3.369865928921047, "grad_norm": 14.5859375, "learning_rate": 8.892481523592952e-06, "loss": 0.6707, "step": 31670 }, { "epoch": 3.3699723345392636, "grad_norm": 11.03125, "learning_rate": 8.892445992040933e-06, "loss": 0.4889, "step": 31671 }, { "epoch": 3.3700787401574805, "grad_norm": 16.34375, "learning_rate": 8.892410460488915e-06, "loss": 0.6833, "step": 31672 }, { "epoch": 3.370185145775697, "grad_norm": 9.6328125, "learning_rate": 8.892374928936897e-06, "loss": 0.5755, "step": 31673 }, { "epoch": 3.3702915513939136, "grad_norm": 12.3671875, "learning_rate": 8.892339397384878e-06, "loss": 0.6858, "step": 31674 }, { "epoch": 3.37039795701213, "grad_norm": 13.6875, "learning_rate": 8.89230386583286e-06, "loss": 0.7241, "step": 31675 }, { "epoch": 3.370504362630347, "grad_norm": 13.2421875, "learning_rate": 8.892268334280843e-06, "loss": 0.7315, "step": 31676 }, { "epoch": 3.3706107682485635, "grad_norm": 13.0234375, "learning_rate": 8.892232802728825e-06, "loss": 0.6217, "step": 31677 }, { "epoch": 3.37071717386678, "grad_norm": 13.0703125, "learning_rate": 8.892197271176806e-06, "loss": 0.6157, "step": 31678 }, { "epoch": 3.370823579484997, "grad_norm": 13.90625, "learning_rate": 8.892161739624788e-06, "loss": 0.6671, "step": 31679 }, { "epoch": 3.3709299851032135, "grad_norm": 12.1484375, "learning_rate": 8.89212620807277e-06, "loss": 0.6736, "step": 31680 }, { "epoch": 3.37103639072143, "grad_norm": 13.421875, "learning_rate": 8.892090676520751e-06, "loss": 0.669, "step": 31681 }, { "epoch": 3.3711427963396465, "grad_norm": 12.96875, "learning_rate": 8.892055144968733e-06, "loss": 0.7191, "step": 31682 }, { "epoch": 3.3712492019578635, "grad_norm": 10.9140625, "learning_rate": 8.892019613416714e-06, "loss": 0.5737, "step": 31683 }, { "epoch": 3.37135560757608, "grad_norm": 12.9609375, "learning_rate": 8.891984081864696e-06, "loss": 0.7326, "step": 31684 }, { "epoch": 3.3714620131942965, "grad_norm": 12.8125, "learning_rate": 8.891948550312678e-06, "loss": 0.616, "step": 31685 }, { "epoch": 3.3715684188125135, "grad_norm": 12.921875, "learning_rate": 8.891913018760661e-06, "loss": 0.7044, "step": 31686 }, { "epoch": 3.37167482443073, "grad_norm": 12.6015625, "learning_rate": 8.891877487208643e-06, "loss": 0.7965, "step": 31687 }, { "epoch": 3.3717812300489465, "grad_norm": 13.5390625, "learning_rate": 8.891841955656624e-06, "loss": 0.5841, "step": 31688 }, { "epoch": 3.3718876356671634, "grad_norm": 17.421875, "learning_rate": 8.891806424104606e-06, "loss": 0.8044, "step": 31689 }, { "epoch": 3.37199404128538, "grad_norm": 9.484375, "learning_rate": 8.891770892552587e-06, "loss": 0.4335, "step": 31690 }, { "epoch": 3.3721004469035964, "grad_norm": 11.7421875, "learning_rate": 8.891735361000569e-06, "loss": 0.5933, "step": 31691 }, { "epoch": 3.372206852521813, "grad_norm": 16.640625, "learning_rate": 8.89169982944855e-06, "loss": 0.8495, "step": 31692 }, { "epoch": 3.37231325814003, "grad_norm": 18.859375, "learning_rate": 8.891664297896532e-06, "loss": 0.7545, "step": 31693 }, { "epoch": 3.3724196637582464, "grad_norm": 9.3203125, "learning_rate": 8.891628766344514e-06, "loss": 0.4743, "step": 31694 }, { "epoch": 3.372526069376463, "grad_norm": 17.765625, "learning_rate": 8.891593234792495e-06, "loss": 0.7202, "step": 31695 }, { "epoch": 3.37263247499468, "grad_norm": 13.328125, "learning_rate": 8.891557703240479e-06, "loss": 0.6801, "step": 31696 }, { "epoch": 3.3727388806128964, "grad_norm": 12.9375, "learning_rate": 8.89152217168846e-06, "loss": 0.6181, "step": 31697 }, { "epoch": 3.372845286231113, "grad_norm": 14.765625, "learning_rate": 8.891486640136442e-06, "loss": 0.6677, "step": 31698 }, { "epoch": 3.3729516918493294, "grad_norm": 12.7421875, "learning_rate": 8.891451108584424e-06, "loss": 0.599, "step": 31699 }, { "epoch": 3.3730580974675464, "grad_norm": 10.6953125, "learning_rate": 8.891415577032405e-06, "loss": 0.524, "step": 31700 }, { "epoch": 3.373164503085763, "grad_norm": 15.5625, "learning_rate": 8.891380045480388e-06, "loss": 0.714, "step": 31701 }, { "epoch": 3.3732709087039794, "grad_norm": 12.984375, "learning_rate": 8.89134451392837e-06, "loss": 0.6711, "step": 31702 }, { "epoch": 3.3733773143221963, "grad_norm": 14.2890625, "learning_rate": 8.891308982376352e-06, "loss": 0.7859, "step": 31703 }, { "epoch": 3.373483719940413, "grad_norm": 13.0546875, "learning_rate": 8.891273450824333e-06, "loss": 0.5951, "step": 31704 }, { "epoch": 3.3735901255586294, "grad_norm": 13.328125, "learning_rate": 8.891237919272315e-06, "loss": 0.5794, "step": 31705 }, { "epoch": 3.3736965311768463, "grad_norm": 10.625, "learning_rate": 8.891202387720296e-06, "loss": 0.7192, "step": 31706 }, { "epoch": 3.373802936795063, "grad_norm": 11.0390625, "learning_rate": 8.891166856168278e-06, "loss": 0.4468, "step": 31707 }, { "epoch": 3.3739093424132793, "grad_norm": 19.90625, "learning_rate": 8.89113132461626e-06, "loss": 0.8397, "step": 31708 }, { "epoch": 3.3740157480314963, "grad_norm": 11.140625, "learning_rate": 8.891095793064241e-06, "loss": 0.5467, "step": 31709 }, { "epoch": 3.374122153649713, "grad_norm": 10.6953125, "learning_rate": 8.891060261512223e-06, "loss": 0.4785, "step": 31710 }, { "epoch": 3.3742285592679293, "grad_norm": 15.65625, "learning_rate": 8.891024729960206e-06, "loss": 0.6719, "step": 31711 }, { "epoch": 3.374334964886146, "grad_norm": 14.3125, "learning_rate": 8.890989198408188e-06, "loss": 0.6425, "step": 31712 }, { "epoch": 3.3744413705043628, "grad_norm": 16.375, "learning_rate": 8.89095366685617e-06, "loss": 0.7435, "step": 31713 }, { "epoch": 3.3745477761225793, "grad_norm": 11.9375, "learning_rate": 8.890918135304151e-06, "loss": 0.6029, "step": 31714 }, { "epoch": 3.374654181740796, "grad_norm": 15.1796875, "learning_rate": 8.890882603752133e-06, "loss": 0.6932, "step": 31715 }, { "epoch": 3.3747605873590127, "grad_norm": 11.5234375, "learning_rate": 8.890847072200114e-06, "loss": 0.5813, "step": 31716 }, { "epoch": 3.3748669929772293, "grad_norm": 13.6875, "learning_rate": 8.890811540648096e-06, "loss": 0.6323, "step": 31717 }, { "epoch": 3.3749733985954458, "grad_norm": 13.140625, "learning_rate": 8.890776009096077e-06, "loss": 0.6649, "step": 31718 }, { "epoch": 3.3750798042136623, "grad_norm": 12.3046875, "learning_rate": 8.890740477544059e-06, "loss": 0.6093, "step": 31719 }, { "epoch": 3.3751862098318792, "grad_norm": 14.7421875, "learning_rate": 8.89070494599204e-06, "loss": 0.6925, "step": 31720 }, { "epoch": 3.3752926154500957, "grad_norm": 11.1484375, "learning_rate": 8.890669414440022e-06, "loss": 0.5675, "step": 31721 }, { "epoch": 3.3753990210683122, "grad_norm": 12.234375, "learning_rate": 8.890633882888006e-06, "loss": 0.5207, "step": 31722 }, { "epoch": 3.375505426686529, "grad_norm": 15.3203125, "learning_rate": 8.890598351335987e-06, "loss": 0.645, "step": 31723 }, { "epoch": 3.3756118323047457, "grad_norm": 16.328125, "learning_rate": 8.890562819783969e-06, "loss": 0.7292, "step": 31724 }, { "epoch": 3.375718237922962, "grad_norm": 17.0625, "learning_rate": 8.89052728823195e-06, "loss": 0.6379, "step": 31725 }, { "epoch": 3.375824643541179, "grad_norm": 11.5546875, "learning_rate": 8.890491756679932e-06, "loss": 0.5877, "step": 31726 }, { "epoch": 3.3759310491593957, "grad_norm": 10.0625, "learning_rate": 8.890456225127915e-06, "loss": 0.4466, "step": 31727 }, { "epoch": 3.376037454777612, "grad_norm": 13.0390625, "learning_rate": 8.890420693575897e-06, "loss": 0.6155, "step": 31728 }, { "epoch": 3.376143860395829, "grad_norm": 24.671875, "learning_rate": 8.890385162023878e-06, "loss": 0.6833, "step": 31729 }, { "epoch": 3.3762502660140457, "grad_norm": 14.953125, "learning_rate": 8.89034963047186e-06, "loss": 0.6149, "step": 31730 }, { "epoch": 3.376356671632262, "grad_norm": 13.5, "learning_rate": 8.890314098919842e-06, "loss": 0.5963, "step": 31731 }, { "epoch": 3.3764630772504787, "grad_norm": 14.046875, "learning_rate": 8.890278567367823e-06, "loss": 0.681, "step": 31732 }, { "epoch": 3.3765694828686956, "grad_norm": 13.1875, "learning_rate": 8.890243035815805e-06, "loss": 0.5438, "step": 31733 }, { "epoch": 3.376675888486912, "grad_norm": 9.4375, "learning_rate": 8.890207504263787e-06, "loss": 0.6166, "step": 31734 }, { "epoch": 3.3767822941051286, "grad_norm": 15.359375, "learning_rate": 8.890171972711768e-06, "loss": 0.8206, "step": 31735 }, { "epoch": 3.376888699723345, "grad_norm": 10.4609375, "learning_rate": 8.89013644115975e-06, "loss": 0.532, "step": 31736 }, { "epoch": 3.376995105341562, "grad_norm": 13.53125, "learning_rate": 8.890100909607733e-06, "loss": 0.6164, "step": 31737 }, { "epoch": 3.3771015109597786, "grad_norm": 15.2578125, "learning_rate": 8.890065378055715e-06, "loss": 0.5884, "step": 31738 }, { "epoch": 3.377207916577995, "grad_norm": 12.234375, "learning_rate": 8.890029846503696e-06, "loss": 0.566, "step": 31739 }, { "epoch": 3.377314322196212, "grad_norm": 16.234375, "learning_rate": 8.889994314951678e-06, "loss": 0.7741, "step": 31740 }, { "epoch": 3.3774207278144286, "grad_norm": 13.140625, "learning_rate": 8.88995878339966e-06, "loss": 0.7124, "step": 31741 }, { "epoch": 3.377527133432645, "grad_norm": 11.2109375, "learning_rate": 8.889923251847643e-06, "loss": 0.6028, "step": 31742 }, { "epoch": 3.377633539050862, "grad_norm": 13.8828125, "learning_rate": 8.889887720295623e-06, "loss": 0.6312, "step": 31743 }, { "epoch": 3.3777399446690786, "grad_norm": 14.84375, "learning_rate": 8.889852188743604e-06, "loss": 0.737, "step": 31744 }, { "epoch": 3.377846350287295, "grad_norm": 11.9453125, "learning_rate": 8.889816657191586e-06, "loss": 0.777, "step": 31745 }, { "epoch": 3.377952755905512, "grad_norm": 13.0703125, "learning_rate": 8.889781125639568e-06, "loss": 0.5729, "step": 31746 }, { "epoch": 3.3780591615237285, "grad_norm": 11.8671875, "learning_rate": 8.88974559408755e-06, "loss": 0.6016, "step": 31747 }, { "epoch": 3.378165567141945, "grad_norm": 12.578125, "learning_rate": 8.889710062535532e-06, "loss": 0.5979, "step": 31748 }, { "epoch": 3.3782719727601616, "grad_norm": 12.0390625, "learning_rate": 8.889674530983514e-06, "loss": 0.6266, "step": 31749 }, { "epoch": 3.3783783783783785, "grad_norm": 10.734375, "learning_rate": 8.889638999431496e-06, "loss": 0.4838, "step": 31750 }, { "epoch": 3.378484783996595, "grad_norm": 8.8359375, "learning_rate": 8.889603467879477e-06, "loss": 0.572, "step": 31751 }, { "epoch": 3.3785911896148115, "grad_norm": 10.8125, "learning_rate": 8.889567936327459e-06, "loss": 0.6096, "step": 31752 }, { "epoch": 3.3786975952330285, "grad_norm": 11.9375, "learning_rate": 8.889532404775442e-06, "loss": 0.7354, "step": 31753 }, { "epoch": 3.378804000851245, "grad_norm": 11.1640625, "learning_rate": 8.889496873223424e-06, "loss": 0.6846, "step": 31754 }, { "epoch": 3.3789104064694615, "grad_norm": 13.1796875, "learning_rate": 8.889461341671405e-06, "loss": 0.6465, "step": 31755 }, { "epoch": 3.379016812087678, "grad_norm": 15.0078125, "learning_rate": 8.889425810119387e-06, "loss": 0.7089, "step": 31756 }, { "epoch": 3.379123217705895, "grad_norm": 13.7578125, "learning_rate": 8.889390278567369e-06, "loss": 0.6297, "step": 31757 }, { "epoch": 3.3792296233241115, "grad_norm": 12.828125, "learning_rate": 8.88935474701535e-06, "loss": 0.6027, "step": 31758 }, { "epoch": 3.379336028942328, "grad_norm": 14.328125, "learning_rate": 8.889319215463332e-06, "loss": 0.8826, "step": 31759 }, { "epoch": 3.379442434560545, "grad_norm": 15.421875, "learning_rate": 8.889283683911313e-06, "loss": 0.711, "step": 31760 }, { "epoch": 3.3795488401787614, "grad_norm": 13.7890625, "learning_rate": 8.889248152359295e-06, "loss": 0.6922, "step": 31761 }, { "epoch": 3.379655245796978, "grad_norm": 12.7890625, "learning_rate": 8.889212620807277e-06, "loss": 0.4794, "step": 31762 }, { "epoch": 3.379761651415195, "grad_norm": 11.578125, "learning_rate": 8.88917708925526e-06, "loss": 0.5706, "step": 31763 }, { "epoch": 3.3798680570334114, "grad_norm": 17.359375, "learning_rate": 8.889141557703242e-06, "loss": 0.7283, "step": 31764 }, { "epoch": 3.379974462651628, "grad_norm": 18.875, "learning_rate": 8.889106026151223e-06, "loss": 0.8285, "step": 31765 }, { "epoch": 3.380080868269845, "grad_norm": 12.6640625, "learning_rate": 8.889070494599205e-06, "loss": 0.6954, "step": 31766 }, { "epoch": 3.3801872738880614, "grad_norm": 12.40625, "learning_rate": 8.889034963047186e-06, "loss": 0.6588, "step": 31767 }, { "epoch": 3.380293679506278, "grad_norm": 13.1328125, "learning_rate": 8.88899943149517e-06, "loss": 0.6477, "step": 31768 }, { "epoch": 3.3804000851244944, "grad_norm": 15.5, "learning_rate": 8.888963899943151e-06, "loss": 0.6627, "step": 31769 }, { "epoch": 3.3805064907427114, "grad_norm": 11.65625, "learning_rate": 8.888928368391133e-06, "loss": 0.5633, "step": 31770 }, { "epoch": 3.380612896360928, "grad_norm": 11.546875, "learning_rate": 8.888892836839114e-06, "loss": 0.5928, "step": 31771 }, { "epoch": 3.3807193019791444, "grad_norm": 12.8046875, "learning_rate": 8.888857305287096e-06, "loss": 0.659, "step": 31772 }, { "epoch": 3.380825707597361, "grad_norm": 11.6640625, "learning_rate": 8.888821773735078e-06, "loss": 0.5575, "step": 31773 }, { "epoch": 3.380932113215578, "grad_norm": 14.3515625, "learning_rate": 8.88878624218306e-06, "loss": 0.5887, "step": 31774 }, { "epoch": 3.3810385188337944, "grad_norm": 13.875, "learning_rate": 8.888750710631041e-06, "loss": 0.6492, "step": 31775 }, { "epoch": 3.381144924452011, "grad_norm": 16.09375, "learning_rate": 8.888715179079022e-06, "loss": 0.6037, "step": 31776 }, { "epoch": 3.381251330070228, "grad_norm": 19.03125, "learning_rate": 8.888679647527004e-06, "loss": 0.7873, "step": 31777 }, { "epoch": 3.3813577356884443, "grad_norm": 12.9375, "learning_rate": 8.888644115974987e-06, "loss": 0.6057, "step": 31778 }, { "epoch": 3.381464141306661, "grad_norm": 16.09375, "learning_rate": 8.888608584422969e-06, "loss": 0.655, "step": 31779 }, { "epoch": 3.381570546924878, "grad_norm": 18.046875, "learning_rate": 8.88857305287095e-06, "loss": 0.6396, "step": 31780 }, { "epoch": 3.3816769525430943, "grad_norm": 14.6484375, "learning_rate": 8.888537521318932e-06, "loss": 0.7862, "step": 31781 }, { "epoch": 3.381783358161311, "grad_norm": 17.34375, "learning_rate": 8.888501989766914e-06, "loss": 0.6873, "step": 31782 }, { "epoch": 3.3818897637795278, "grad_norm": 12.7734375, "learning_rate": 8.888466458214895e-06, "loss": 0.6285, "step": 31783 }, { "epoch": 3.3819961693977443, "grad_norm": 18.140625, "learning_rate": 8.888430926662877e-06, "loss": 0.6991, "step": 31784 }, { "epoch": 3.382102575015961, "grad_norm": 11.9609375, "learning_rate": 8.888395395110859e-06, "loss": 0.6353, "step": 31785 }, { "epoch": 3.3822089806341773, "grad_norm": 12.9921875, "learning_rate": 8.88835986355884e-06, "loss": 0.565, "step": 31786 }, { "epoch": 3.3823153862523943, "grad_norm": 12.7578125, "learning_rate": 8.888324332006822e-06, "loss": 0.6283, "step": 31787 }, { "epoch": 3.3824217918706108, "grad_norm": 15.8125, "learning_rate": 8.888288800454805e-06, "loss": 0.632, "step": 31788 }, { "epoch": 3.3825281974888273, "grad_norm": 12.90625, "learning_rate": 8.888253268902787e-06, "loss": 0.6022, "step": 31789 }, { "epoch": 3.3826346031070442, "grad_norm": 12.9296875, "learning_rate": 8.888217737350768e-06, "loss": 0.608, "step": 31790 }, { "epoch": 3.3827410087252607, "grad_norm": 14.7734375, "learning_rate": 8.88818220579875e-06, "loss": 0.5578, "step": 31791 }, { "epoch": 3.3828474143434772, "grad_norm": 15.6328125, "learning_rate": 8.888146674246732e-06, "loss": 0.7674, "step": 31792 }, { "epoch": 3.3829538199616938, "grad_norm": 17.890625, "learning_rate": 8.888111142694713e-06, "loss": 0.8504, "step": 31793 }, { "epoch": 3.3830602255799107, "grad_norm": 12.8671875, "learning_rate": 8.888075611142696e-06, "loss": 0.6187, "step": 31794 }, { "epoch": 3.383166631198127, "grad_norm": 13.0625, "learning_rate": 8.888040079590678e-06, "loss": 0.6191, "step": 31795 }, { "epoch": 3.3832730368163437, "grad_norm": 14.3203125, "learning_rate": 8.88800454803866e-06, "loss": 0.7242, "step": 31796 }, { "epoch": 3.3833794424345607, "grad_norm": 14.546875, "learning_rate": 8.887969016486641e-06, "loss": 0.6908, "step": 31797 }, { "epoch": 3.383485848052777, "grad_norm": 13.515625, "learning_rate": 8.887933484934623e-06, "loss": 0.6651, "step": 31798 }, { "epoch": 3.3835922536709937, "grad_norm": 15.59375, "learning_rate": 8.887897953382605e-06, "loss": 0.579, "step": 31799 }, { "epoch": 3.3836986592892107, "grad_norm": 18.125, "learning_rate": 8.887862421830586e-06, "loss": 0.7884, "step": 31800 }, { "epoch": 3.383805064907427, "grad_norm": 17.3125, "learning_rate": 8.887826890278568e-06, "loss": 0.7396, "step": 31801 }, { "epoch": 3.3839114705256437, "grad_norm": 14.25, "learning_rate": 8.88779135872655e-06, "loss": 0.6172, "step": 31802 }, { "epoch": 3.3840178761438606, "grad_norm": 20.3125, "learning_rate": 8.887755827174531e-06, "loss": 0.6202, "step": 31803 }, { "epoch": 3.384124281762077, "grad_norm": 13.3671875, "learning_rate": 8.887720295622514e-06, "loss": 0.6142, "step": 31804 }, { "epoch": 3.3842306873802936, "grad_norm": 14.65625, "learning_rate": 8.887684764070496e-06, "loss": 0.7484, "step": 31805 }, { "epoch": 3.38433709299851, "grad_norm": 14.8125, "learning_rate": 8.887649232518477e-06, "loss": 0.6762, "step": 31806 }, { "epoch": 3.384443498616727, "grad_norm": 12.046875, "learning_rate": 8.887613700966459e-06, "loss": 0.6002, "step": 31807 }, { "epoch": 3.3845499042349436, "grad_norm": 20.46875, "learning_rate": 8.88757816941444e-06, "loss": 0.6743, "step": 31808 }, { "epoch": 3.38465630985316, "grad_norm": 12.234375, "learning_rate": 8.887542637862422e-06, "loss": 0.6874, "step": 31809 }, { "epoch": 3.384762715471377, "grad_norm": 12.0234375, "learning_rate": 8.887507106310404e-06, "loss": 0.5047, "step": 31810 }, { "epoch": 3.3848691210895936, "grad_norm": 15.2265625, "learning_rate": 8.887471574758385e-06, "loss": 0.5423, "step": 31811 }, { "epoch": 3.38497552670781, "grad_norm": 16.796875, "learning_rate": 8.887436043206367e-06, "loss": 0.709, "step": 31812 }, { "epoch": 3.3850819323260266, "grad_norm": 13.296875, "learning_rate": 8.887400511654349e-06, "loss": 0.5685, "step": 31813 }, { "epoch": 3.3851883379442436, "grad_norm": 10.1484375, "learning_rate": 8.887364980102332e-06, "loss": 0.5231, "step": 31814 }, { "epoch": 3.38529474356246, "grad_norm": 12.1640625, "learning_rate": 8.887329448550314e-06, "loss": 0.5711, "step": 31815 }, { "epoch": 3.3854011491806766, "grad_norm": 10.328125, "learning_rate": 8.887293916998295e-06, "loss": 0.4928, "step": 31816 }, { "epoch": 3.3855075547988935, "grad_norm": 11.1171875, "learning_rate": 8.887258385446277e-06, "loss": 0.5711, "step": 31817 }, { "epoch": 3.38561396041711, "grad_norm": 11.5859375, "learning_rate": 8.887222853894258e-06, "loss": 0.673, "step": 31818 }, { "epoch": 3.3857203660353266, "grad_norm": 15.3984375, "learning_rate": 8.887187322342242e-06, "loss": 0.6332, "step": 31819 }, { "epoch": 3.3858267716535435, "grad_norm": 15.375, "learning_rate": 8.887151790790223e-06, "loss": 0.8646, "step": 31820 }, { "epoch": 3.38593317727176, "grad_norm": 11.53125, "learning_rate": 8.887116259238205e-06, "loss": 0.6338, "step": 31821 }, { "epoch": 3.3860395828899765, "grad_norm": 12.53125, "learning_rate": 8.887080727686187e-06, "loss": 0.6771, "step": 31822 }, { "epoch": 3.386145988508193, "grad_norm": 12.09375, "learning_rate": 8.887045196134168e-06, "loss": 0.652, "step": 31823 }, { "epoch": 3.38625239412641, "grad_norm": 22.828125, "learning_rate": 8.88700966458215e-06, "loss": 0.8511, "step": 31824 }, { "epoch": 3.3863587997446265, "grad_norm": 13.203125, "learning_rate": 8.886974133030131e-06, "loss": 0.6836, "step": 31825 }, { "epoch": 3.386465205362843, "grad_norm": 13.8828125, "learning_rate": 8.886938601478113e-06, "loss": 0.7039, "step": 31826 }, { "epoch": 3.38657161098106, "grad_norm": 17.0625, "learning_rate": 8.886903069926095e-06, "loss": 0.967, "step": 31827 }, { "epoch": 3.3866780165992765, "grad_norm": 12.59375, "learning_rate": 8.886867538374076e-06, "loss": 0.6046, "step": 31828 }, { "epoch": 3.386784422217493, "grad_norm": 16.296875, "learning_rate": 8.88683200682206e-06, "loss": 0.6531, "step": 31829 }, { "epoch": 3.3868908278357095, "grad_norm": 11.0, "learning_rate": 8.886796475270041e-06, "loss": 0.6746, "step": 31830 }, { "epoch": 3.3869972334539264, "grad_norm": 13.3125, "learning_rate": 8.886760943718023e-06, "loss": 0.6555, "step": 31831 }, { "epoch": 3.387103639072143, "grad_norm": 14.109375, "learning_rate": 8.886725412166004e-06, "loss": 0.627, "step": 31832 }, { "epoch": 3.3872100446903595, "grad_norm": 13.3671875, "learning_rate": 8.886689880613986e-06, "loss": 0.711, "step": 31833 }, { "epoch": 3.3873164503085764, "grad_norm": 12.140625, "learning_rate": 8.886654349061968e-06, "loss": 0.5622, "step": 31834 }, { "epoch": 3.387422855926793, "grad_norm": 12.3046875, "learning_rate": 8.886618817509949e-06, "loss": 0.6737, "step": 31835 }, { "epoch": 3.3875292615450094, "grad_norm": 12.578125, "learning_rate": 8.88658328595793e-06, "loss": 0.6064, "step": 31836 }, { "epoch": 3.3876356671632264, "grad_norm": 15.5546875, "learning_rate": 8.886547754405912e-06, "loss": 0.7603, "step": 31837 }, { "epoch": 3.387742072781443, "grad_norm": 15.3671875, "learning_rate": 8.886512222853894e-06, "loss": 0.6836, "step": 31838 }, { "epoch": 3.3878484783996594, "grad_norm": 11.9921875, "learning_rate": 8.886476691301876e-06, "loss": 0.6724, "step": 31839 }, { "epoch": 3.3879548840178764, "grad_norm": 12.8828125, "learning_rate": 8.886441159749859e-06, "loss": 0.68, "step": 31840 }, { "epoch": 3.388061289636093, "grad_norm": 8.96875, "learning_rate": 8.88640562819784e-06, "loss": 0.4861, "step": 31841 }, { "epoch": 3.3881676952543094, "grad_norm": 16.390625, "learning_rate": 8.886370096645822e-06, "loss": 0.6654, "step": 31842 }, { "epoch": 3.388274100872526, "grad_norm": 14.8671875, "learning_rate": 8.886334565093804e-06, "loss": 0.8002, "step": 31843 }, { "epoch": 3.388380506490743, "grad_norm": 18.5625, "learning_rate": 8.886299033541785e-06, "loss": 0.9449, "step": 31844 }, { "epoch": 3.3884869121089594, "grad_norm": 21.03125, "learning_rate": 8.886263501989769e-06, "loss": 0.9605, "step": 31845 }, { "epoch": 3.388593317727176, "grad_norm": 11.953125, "learning_rate": 8.88622797043775e-06, "loss": 0.586, "step": 31846 }, { "epoch": 3.388699723345393, "grad_norm": 10.109375, "learning_rate": 8.886192438885732e-06, "loss": 0.5625, "step": 31847 }, { "epoch": 3.3888061289636093, "grad_norm": 11.1953125, "learning_rate": 8.886156907333713e-06, "loss": 0.6064, "step": 31848 }, { "epoch": 3.388912534581826, "grad_norm": 15.3359375, "learning_rate": 8.886121375781695e-06, "loss": 0.8012, "step": 31849 }, { "epoch": 3.3890189402000424, "grad_norm": 17.71875, "learning_rate": 8.886085844229677e-06, "loss": 0.8258, "step": 31850 }, { "epoch": 3.3891253458182593, "grad_norm": 10.2734375, "learning_rate": 8.886050312677658e-06, "loss": 0.5461, "step": 31851 }, { "epoch": 3.389231751436476, "grad_norm": 13.4296875, "learning_rate": 8.88601478112564e-06, "loss": 0.7683, "step": 31852 }, { "epoch": 3.3893381570546923, "grad_norm": 17.96875, "learning_rate": 8.885979249573621e-06, "loss": 0.7985, "step": 31853 }, { "epoch": 3.3894445626729093, "grad_norm": 11.640625, "learning_rate": 8.885943718021603e-06, "loss": 0.5198, "step": 31854 }, { "epoch": 3.389550968291126, "grad_norm": 13.2578125, "learning_rate": 8.885908186469586e-06, "loss": 0.7495, "step": 31855 }, { "epoch": 3.3896573739093423, "grad_norm": 11.671875, "learning_rate": 8.885872654917568e-06, "loss": 0.586, "step": 31856 }, { "epoch": 3.3897637795275593, "grad_norm": 10.6875, "learning_rate": 8.88583712336555e-06, "loss": 0.6452, "step": 31857 }, { "epoch": 3.3898701851457758, "grad_norm": 12.6484375, "learning_rate": 8.885801591813531e-06, "loss": 0.7407, "step": 31858 }, { "epoch": 3.3899765907639923, "grad_norm": 15.296875, "learning_rate": 8.885766060261513e-06, "loss": 0.6772, "step": 31859 }, { "epoch": 3.3900829963822092, "grad_norm": 13.6015625, "learning_rate": 8.885730528709496e-06, "loss": 0.576, "step": 31860 }, { "epoch": 3.3901894020004257, "grad_norm": 11.875, "learning_rate": 8.885694997157478e-06, "loss": 0.5686, "step": 31861 }, { "epoch": 3.3902958076186422, "grad_norm": 13.875, "learning_rate": 8.88565946560546e-06, "loss": 0.7255, "step": 31862 }, { "epoch": 3.3904022132368588, "grad_norm": 11.1640625, "learning_rate": 8.88562393405344e-06, "loss": 0.5624, "step": 31863 }, { "epoch": 3.3905086188550757, "grad_norm": 10.71875, "learning_rate": 8.88558840250142e-06, "loss": 0.4851, "step": 31864 }, { "epoch": 3.390615024473292, "grad_norm": 13.2734375, "learning_rate": 8.885552870949404e-06, "loss": 0.6233, "step": 31865 }, { "epoch": 3.3907214300915087, "grad_norm": 14.09375, "learning_rate": 8.885517339397386e-06, "loss": 0.6982, "step": 31866 }, { "epoch": 3.3908278357097252, "grad_norm": 13.734375, "learning_rate": 8.885481807845367e-06, "loss": 0.6398, "step": 31867 }, { "epoch": 3.390934241327942, "grad_norm": 15.671875, "learning_rate": 8.885446276293349e-06, "loss": 0.6495, "step": 31868 }, { "epoch": 3.3910406469461587, "grad_norm": 10.9296875, "learning_rate": 8.88541074474133e-06, "loss": 0.5535, "step": 31869 }, { "epoch": 3.391147052564375, "grad_norm": 14.7265625, "learning_rate": 8.885375213189312e-06, "loss": 0.7495, "step": 31870 }, { "epoch": 3.391253458182592, "grad_norm": 11.4765625, "learning_rate": 8.885339681637295e-06, "loss": 0.5823, "step": 31871 }, { "epoch": 3.3913598638008087, "grad_norm": 12.171875, "learning_rate": 8.885304150085277e-06, "loss": 0.6022, "step": 31872 }, { "epoch": 3.391466269419025, "grad_norm": 13.859375, "learning_rate": 8.885268618533259e-06, "loss": 0.6284, "step": 31873 }, { "epoch": 3.391572675037242, "grad_norm": 10.4765625, "learning_rate": 8.88523308698124e-06, "loss": 0.4605, "step": 31874 }, { "epoch": 3.3916790806554586, "grad_norm": 12.75, "learning_rate": 8.885197555429222e-06, "loss": 0.6201, "step": 31875 }, { "epoch": 3.391785486273675, "grad_norm": 11.5703125, "learning_rate": 8.885162023877203e-06, "loss": 0.5401, "step": 31876 }, { "epoch": 3.391891891891892, "grad_norm": 9.2421875, "learning_rate": 8.885126492325185e-06, "loss": 0.6819, "step": 31877 }, { "epoch": 3.3919982975101086, "grad_norm": 15.2890625, "learning_rate": 8.885090960773167e-06, "loss": 0.6514, "step": 31878 }, { "epoch": 3.392104703128325, "grad_norm": 12.3125, "learning_rate": 8.885055429221148e-06, "loss": 0.5628, "step": 31879 }, { "epoch": 3.3922111087465416, "grad_norm": 9.1484375, "learning_rate": 8.88501989766913e-06, "loss": 0.4343, "step": 31880 }, { "epoch": 3.3923175143647586, "grad_norm": 11.5390625, "learning_rate": 8.884984366117113e-06, "loss": 0.5877, "step": 31881 }, { "epoch": 3.392423919982975, "grad_norm": 13.9453125, "learning_rate": 8.884948834565095e-06, "loss": 0.7008, "step": 31882 }, { "epoch": 3.3925303256011916, "grad_norm": 12.328125, "learning_rate": 8.884913303013076e-06, "loss": 0.4965, "step": 31883 }, { "epoch": 3.3926367312194086, "grad_norm": 14.84375, "learning_rate": 8.884877771461058e-06, "loss": 0.6459, "step": 31884 }, { "epoch": 3.392743136837625, "grad_norm": 11.8359375, "learning_rate": 8.88484223990904e-06, "loss": 0.6841, "step": 31885 }, { "epoch": 3.3928495424558416, "grad_norm": 14.4765625, "learning_rate": 8.884806708357023e-06, "loss": 0.6497, "step": 31886 }, { "epoch": 3.392955948074058, "grad_norm": 14.0, "learning_rate": 8.884771176805005e-06, "loss": 0.6573, "step": 31887 }, { "epoch": 3.393062353692275, "grad_norm": 10.9140625, "learning_rate": 8.884735645252986e-06, "loss": 0.5387, "step": 31888 }, { "epoch": 3.3931687593104916, "grad_norm": 13.9921875, "learning_rate": 8.884700113700968e-06, "loss": 0.6737, "step": 31889 }, { "epoch": 3.393275164928708, "grad_norm": 14.1953125, "learning_rate": 8.88466458214895e-06, "loss": 0.5749, "step": 31890 }, { "epoch": 3.393381570546925, "grad_norm": 12.296875, "learning_rate": 8.884629050596931e-06, "loss": 0.6136, "step": 31891 }, { "epoch": 3.3934879761651415, "grad_norm": 13.859375, "learning_rate": 8.884593519044913e-06, "loss": 0.6521, "step": 31892 }, { "epoch": 3.393594381783358, "grad_norm": 17.453125, "learning_rate": 8.884557987492894e-06, "loss": 0.6793, "step": 31893 }, { "epoch": 3.393700787401575, "grad_norm": 11.0859375, "learning_rate": 8.884522455940876e-06, "loss": 0.5109, "step": 31894 }, { "epoch": 3.3938071930197915, "grad_norm": 13.875, "learning_rate": 8.884486924388857e-06, "loss": 0.5648, "step": 31895 }, { "epoch": 3.393913598638008, "grad_norm": 12.453125, "learning_rate": 8.88445139283684e-06, "loss": 0.7016, "step": 31896 }, { "epoch": 3.394020004256225, "grad_norm": 16.171875, "learning_rate": 8.884415861284822e-06, "loss": 0.6709, "step": 31897 }, { "epoch": 3.3941264098744415, "grad_norm": 15.9296875, "learning_rate": 8.884380329732804e-06, "loss": 0.7278, "step": 31898 }, { "epoch": 3.394232815492658, "grad_norm": 10.859375, "learning_rate": 8.884344798180786e-06, "loss": 0.4591, "step": 31899 }, { "epoch": 3.3943392211108745, "grad_norm": 15.25, "learning_rate": 8.884309266628767e-06, "loss": 0.6779, "step": 31900 }, { "epoch": 3.3944456267290914, "grad_norm": 11.40625, "learning_rate": 8.884273735076749e-06, "loss": 0.5234, "step": 31901 }, { "epoch": 3.394552032347308, "grad_norm": 13.9375, "learning_rate": 8.88423820352473e-06, "loss": 0.6807, "step": 31902 }, { "epoch": 3.3946584379655245, "grad_norm": 12.8515625, "learning_rate": 8.884202671972712e-06, "loss": 0.611, "step": 31903 }, { "epoch": 3.394764843583741, "grad_norm": 15.96875, "learning_rate": 8.884167140420694e-06, "loss": 0.7212, "step": 31904 }, { "epoch": 3.394871249201958, "grad_norm": 12.6953125, "learning_rate": 8.884131608868675e-06, "loss": 0.6172, "step": 31905 }, { "epoch": 3.3949776548201744, "grad_norm": 16.65625, "learning_rate": 8.884096077316658e-06, "loss": 0.6785, "step": 31906 }, { "epoch": 3.395084060438391, "grad_norm": 11.2734375, "learning_rate": 8.88406054576464e-06, "loss": 0.6871, "step": 31907 }, { "epoch": 3.395190466056608, "grad_norm": 10.546875, "learning_rate": 8.884025014212622e-06, "loss": 0.7328, "step": 31908 }, { "epoch": 3.3952968716748244, "grad_norm": 13.0078125, "learning_rate": 8.883989482660603e-06, "loss": 0.6645, "step": 31909 }, { "epoch": 3.395403277293041, "grad_norm": 15.078125, "learning_rate": 8.883953951108585e-06, "loss": 0.6511, "step": 31910 }, { "epoch": 3.395509682911258, "grad_norm": 16.09375, "learning_rate": 8.883918419556566e-06, "loss": 0.7138, "step": 31911 }, { "epoch": 3.3956160885294744, "grad_norm": 14.53125, "learning_rate": 8.88388288800455e-06, "loss": 0.5611, "step": 31912 }, { "epoch": 3.395722494147691, "grad_norm": 16.5625, "learning_rate": 8.883847356452531e-06, "loss": 0.6731, "step": 31913 }, { "epoch": 3.395828899765908, "grad_norm": 16.46875, "learning_rate": 8.883811824900513e-06, "loss": 0.7189, "step": 31914 }, { "epoch": 3.3959353053841244, "grad_norm": 18.75, "learning_rate": 8.883776293348495e-06, "loss": 0.7382, "step": 31915 }, { "epoch": 3.396041711002341, "grad_norm": 16.96875, "learning_rate": 8.883740761796476e-06, "loss": 0.7101, "step": 31916 }, { "epoch": 3.3961481166205574, "grad_norm": 11.515625, "learning_rate": 8.883705230244458e-06, "loss": 0.5809, "step": 31917 }, { "epoch": 3.3962545222387743, "grad_norm": 18.359375, "learning_rate": 8.88366969869244e-06, "loss": 0.7087, "step": 31918 }, { "epoch": 3.396360927856991, "grad_norm": 14.8125, "learning_rate": 8.883634167140421e-06, "loss": 0.7205, "step": 31919 }, { "epoch": 3.3964673334752074, "grad_norm": 16.609375, "learning_rate": 8.883598635588403e-06, "loss": 0.8441, "step": 31920 }, { "epoch": 3.3965737390934243, "grad_norm": 16.5, "learning_rate": 8.883563104036384e-06, "loss": 0.68, "step": 31921 }, { "epoch": 3.396680144711641, "grad_norm": 14.1875, "learning_rate": 8.883527572484368e-06, "loss": 0.7351, "step": 31922 }, { "epoch": 3.3967865503298573, "grad_norm": 14.3203125, "learning_rate": 8.883492040932349e-06, "loss": 0.5714, "step": 31923 }, { "epoch": 3.396892955948074, "grad_norm": 12.6640625, "learning_rate": 8.88345650938033e-06, "loss": 0.5313, "step": 31924 }, { "epoch": 3.396999361566291, "grad_norm": 9.71875, "learning_rate": 8.883420977828312e-06, "loss": 0.4167, "step": 31925 }, { "epoch": 3.3971057671845073, "grad_norm": 10.875, "learning_rate": 8.883385446276294e-06, "loss": 0.5187, "step": 31926 }, { "epoch": 3.397212172802724, "grad_norm": 10.1484375, "learning_rate": 8.883349914724276e-06, "loss": 0.4967, "step": 31927 }, { "epoch": 3.3973185784209408, "grad_norm": 10.4453125, "learning_rate": 8.883314383172257e-06, "loss": 0.407, "step": 31928 }, { "epoch": 3.3974249840391573, "grad_norm": 14.4296875, "learning_rate": 8.883278851620239e-06, "loss": 0.7361, "step": 31929 }, { "epoch": 3.397531389657374, "grad_norm": 11.8203125, "learning_rate": 8.88324332006822e-06, "loss": 0.6127, "step": 31930 }, { "epoch": 3.3976377952755907, "grad_norm": 16.796875, "learning_rate": 8.883207788516202e-06, "loss": 0.6023, "step": 31931 }, { "epoch": 3.3977442008938072, "grad_norm": 15.0859375, "learning_rate": 8.883172256964185e-06, "loss": 0.755, "step": 31932 }, { "epoch": 3.3978506065120238, "grad_norm": 15.0703125, "learning_rate": 8.883136725412167e-06, "loss": 0.7177, "step": 31933 }, { "epoch": 3.3979570121302407, "grad_norm": 13.2734375, "learning_rate": 8.883101193860149e-06, "loss": 0.7397, "step": 31934 }, { "epoch": 3.398063417748457, "grad_norm": 14.6640625, "learning_rate": 8.88306566230813e-06, "loss": 0.6234, "step": 31935 }, { "epoch": 3.3981698233666737, "grad_norm": 15.6875, "learning_rate": 8.883030130756112e-06, "loss": 0.6284, "step": 31936 }, { "epoch": 3.3982762289848902, "grad_norm": 12.296875, "learning_rate": 8.882994599204095e-06, "loss": 0.6613, "step": 31937 }, { "epoch": 3.398382634603107, "grad_norm": 13.3515625, "learning_rate": 8.882959067652077e-06, "loss": 0.6619, "step": 31938 }, { "epoch": 3.3984890402213237, "grad_norm": 12.9765625, "learning_rate": 8.882923536100058e-06, "loss": 0.6817, "step": 31939 }, { "epoch": 3.39859544583954, "grad_norm": 12.6328125, "learning_rate": 8.88288800454804e-06, "loss": 0.5724, "step": 31940 }, { "epoch": 3.398701851457757, "grad_norm": 13.9453125, "learning_rate": 8.882852472996021e-06, "loss": 0.7506, "step": 31941 }, { "epoch": 3.3988082570759737, "grad_norm": 21.0, "learning_rate": 8.882816941444003e-06, "loss": 0.7403, "step": 31942 }, { "epoch": 3.39891466269419, "grad_norm": 11.328125, "learning_rate": 8.882781409891985e-06, "loss": 0.6008, "step": 31943 }, { "epoch": 3.3990210683124067, "grad_norm": 14.09375, "learning_rate": 8.882745878339966e-06, "loss": 0.7286, "step": 31944 }, { "epoch": 3.3991274739306236, "grad_norm": 11.984375, "learning_rate": 8.882710346787948e-06, "loss": 0.6471, "step": 31945 }, { "epoch": 3.39923387954884, "grad_norm": 15.890625, "learning_rate": 8.88267481523593e-06, "loss": 0.6916, "step": 31946 }, { "epoch": 3.3993402851670567, "grad_norm": 15.1015625, "learning_rate": 8.882639283683913e-06, "loss": 0.8007, "step": 31947 }, { "epoch": 3.3994466907852736, "grad_norm": 11.2109375, "learning_rate": 8.882603752131894e-06, "loss": 0.599, "step": 31948 }, { "epoch": 3.39955309640349, "grad_norm": 11.7734375, "learning_rate": 8.882568220579876e-06, "loss": 0.6433, "step": 31949 }, { "epoch": 3.3996595020217066, "grad_norm": 16.421875, "learning_rate": 8.882532689027858e-06, "loss": 0.6912, "step": 31950 }, { "epoch": 3.3997659076399236, "grad_norm": 13.015625, "learning_rate": 8.88249715747584e-06, "loss": 0.6394, "step": 31951 }, { "epoch": 3.39987231325814, "grad_norm": 11.4453125, "learning_rate": 8.88246162592382e-06, "loss": 0.5707, "step": 31952 }, { "epoch": 3.3999787188763566, "grad_norm": 12.9375, "learning_rate": 8.882426094371802e-06, "loss": 0.5956, "step": 31953 }, { "epoch": 3.4000851244945736, "grad_norm": 12.84375, "learning_rate": 8.882390562819784e-06, "loss": 0.7503, "step": 31954 }, { "epoch": 3.40019153011279, "grad_norm": 14.0859375, "learning_rate": 8.882355031267766e-06, "loss": 0.6258, "step": 31955 }, { "epoch": 3.4002979357310066, "grad_norm": 13.671875, "learning_rate": 8.882319499715747e-06, "loss": 0.7515, "step": 31956 }, { "epoch": 3.400404341349223, "grad_norm": 12.4453125, "learning_rate": 8.882283968163729e-06, "loss": 0.673, "step": 31957 }, { "epoch": 3.40051074696744, "grad_norm": 10.8203125, "learning_rate": 8.882248436611712e-06, "loss": 0.4748, "step": 31958 }, { "epoch": 3.4006171525856566, "grad_norm": 12.0703125, "learning_rate": 8.882212905059694e-06, "loss": 0.66, "step": 31959 }, { "epoch": 3.400723558203873, "grad_norm": 17.859375, "learning_rate": 8.882177373507675e-06, "loss": 0.7601, "step": 31960 }, { "epoch": 3.4008299638220896, "grad_norm": 12.671875, "learning_rate": 8.882141841955657e-06, "loss": 0.5938, "step": 31961 }, { "epoch": 3.4009363694403065, "grad_norm": 13.34375, "learning_rate": 8.882106310403639e-06, "loss": 0.8286, "step": 31962 }, { "epoch": 3.401042775058523, "grad_norm": 11.3671875, "learning_rate": 8.882070778851622e-06, "loss": 0.5487, "step": 31963 }, { "epoch": 3.4011491806767395, "grad_norm": 16.484375, "learning_rate": 8.882035247299603e-06, "loss": 0.7836, "step": 31964 }, { "epoch": 3.4012555862949565, "grad_norm": 11.859375, "learning_rate": 8.881999715747585e-06, "loss": 0.7277, "step": 31965 }, { "epoch": 3.401361991913173, "grad_norm": 15.2890625, "learning_rate": 8.881964184195567e-06, "loss": 0.6371, "step": 31966 }, { "epoch": 3.4014683975313895, "grad_norm": 14.859375, "learning_rate": 8.881928652643548e-06, "loss": 0.7295, "step": 31967 }, { "epoch": 3.4015748031496065, "grad_norm": 14.53125, "learning_rate": 8.88189312109153e-06, "loss": 0.8478, "step": 31968 }, { "epoch": 3.401681208767823, "grad_norm": 11.0078125, "learning_rate": 8.881857589539512e-06, "loss": 0.5718, "step": 31969 }, { "epoch": 3.4017876143860395, "grad_norm": 14.640625, "learning_rate": 8.881822057987493e-06, "loss": 0.7412, "step": 31970 }, { "epoch": 3.4018940200042564, "grad_norm": 11.109375, "learning_rate": 8.881786526435475e-06, "loss": 0.5346, "step": 31971 }, { "epoch": 3.402000425622473, "grad_norm": 11.7421875, "learning_rate": 8.881750994883456e-06, "loss": 0.6499, "step": 31972 }, { "epoch": 3.4021068312406895, "grad_norm": 15.3046875, "learning_rate": 8.88171546333144e-06, "loss": 0.6845, "step": 31973 }, { "epoch": 3.402213236858906, "grad_norm": 14.2578125, "learning_rate": 8.881679931779421e-06, "loss": 0.6391, "step": 31974 }, { "epoch": 3.402319642477123, "grad_norm": 12.875, "learning_rate": 8.881644400227403e-06, "loss": 0.6672, "step": 31975 }, { "epoch": 3.4024260480953394, "grad_norm": 15.265625, "learning_rate": 8.881608868675384e-06, "loss": 0.7972, "step": 31976 }, { "epoch": 3.402532453713556, "grad_norm": 11.7109375, "learning_rate": 8.881573337123366e-06, "loss": 0.6412, "step": 31977 }, { "epoch": 3.402638859331773, "grad_norm": 11.125, "learning_rate": 8.88153780557135e-06, "loss": 0.6688, "step": 31978 }, { "epoch": 3.4027452649499894, "grad_norm": 12.5859375, "learning_rate": 8.881502274019331e-06, "loss": 0.6559, "step": 31979 }, { "epoch": 3.402851670568206, "grad_norm": 12.1484375, "learning_rate": 8.881466742467313e-06, "loss": 0.6332, "step": 31980 }, { "epoch": 3.4029580761864224, "grad_norm": 17.015625, "learning_rate": 8.881431210915294e-06, "loss": 0.6813, "step": 31981 }, { "epoch": 3.4030644818046394, "grad_norm": 12.7578125, "learning_rate": 8.881395679363276e-06, "loss": 0.6395, "step": 31982 }, { "epoch": 3.403170887422856, "grad_norm": 15.5859375, "learning_rate": 8.881360147811257e-06, "loss": 0.6924, "step": 31983 }, { "epoch": 3.4032772930410724, "grad_norm": 13.9453125, "learning_rate": 8.881324616259239e-06, "loss": 0.6322, "step": 31984 }, { "epoch": 3.4033836986592894, "grad_norm": 17.875, "learning_rate": 8.88128908470722e-06, "loss": 0.8336, "step": 31985 }, { "epoch": 3.403490104277506, "grad_norm": 14.046875, "learning_rate": 8.881253553155202e-06, "loss": 0.6847, "step": 31986 }, { "epoch": 3.4035965098957224, "grad_norm": 12.875, "learning_rate": 8.881218021603184e-06, "loss": 0.6715, "step": 31987 }, { "epoch": 3.4037029155139393, "grad_norm": 15.3828125, "learning_rate": 8.881182490051165e-06, "loss": 0.7479, "step": 31988 }, { "epoch": 3.403809321132156, "grad_norm": 11.875, "learning_rate": 8.881146958499149e-06, "loss": 0.7911, "step": 31989 }, { "epoch": 3.4039157267503724, "grad_norm": 10.421875, "learning_rate": 8.88111142694713e-06, "loss": 0.4847, "step": 31990 }, { "epoch": 3.4040221323685893, "grad_norm": 11.9453125, "learning_rate": 8.881075895395112e-06, "loss": 0.6903, "step": 31991 }, { "epoch": 3.404128537986806, "grad_norm": 13.328125, "learning_rate": 8.881040363843094e-06, "loss": 0.6629, "step": 31992 }, { "epoch": 3.4042349436050223, "grad_norm": 18.90625, "learning_rate": 8.881004832291075e-06, "loss": 0.706, "step": 31993 }, { "epoch": 3.404341349223239, "grad_norm": 12.484375, "learning_rate": 8.880969300739057e-06, "loss": 0.525, "step": 31994 }, { "epoch": 3.404447754841456, "grad_norm": 11.7890625, "learning_rate": 8.880933769187038e-06, "loss": 0.6289, "step": 31995 }, { "epoch": 3.4045541604596723, "grad_norm": 11.359375, "learning_rate": 8.88089823763502e-06, "loss": 0.5443, "step": 31996 }, { "epoch": 3.404660566077889, "grad_norm": 13.640625, "learning_rate": 8.880862706083002e-06, "loss": 0.7498, "step": 31997 }, { "epoch": 3.4047669716961053, "grad_norm": 16.828125, "learning_rate": 8.880827174530983e-06, "loss": 0.8183, "step": 31998 }, { "epoch": 3.4048733773143223, "grad_norm": 19.4375, "learning_rate": 8.880791642978967e-06, "loss": 0.701, "step": 31999 }, { "epoch": 3.404979782932539, "grad_norm": 13.1875, "learning_rate": 8.880756111426948e-06, "loss": 0.7786, "step": 32000 }, { "epoch": 3.4050861885507553, "grad_norm": 12.28125, "learning_rate": 8.88072057987493e-06, "loss": 0.6025, "step": 32001 }, { "epoch": 3.4051925941689722, "grad_norm": 12.0390625, "learning_rate": 8.880685048322911e-06, "loss": 0.5997, "step": 32002 }, { "epoch": 3.4052989997871888, "grad_norm": 15.90625, "learning_rate": 8.880649516770893e-06, "loss": 0.7785, "step": 32003 }, { "epoch": 3.4054054054054053, "grad_norm": 12.90625, "learning_rate": 8.880613985218876e-06, "loss": 0.6764, "step": 32004 }, { "epoch": 3.405511811023622, "grad_norm": 12.25, "learning_rate": 8.880578453666858e-06, "loss": 0.4924, "step": 32005 }, { "epoch": 3.4056182166418387, "grad_norm": 13.3203125, "learning_rate": 8.88054292211484e-06, "loss": 0.7134, "step": 32006 }, { "epoch": 3.4057246222600552, "grad_norm": 10.5078125, "learning_rate": 8.880507390562821e-06, "loss": 0.6019, "step": 32007 }, { "epoch": 3.405831027878272, "grad_norm": 13.640625, "learning_rate": 8.880471859010803e-06, "loss": 0.6664, "step": 32008 }, { "epoch": 3.4059374334964887, "grad_norm": 13.40625, "learning_rate": 8.880436327458784e-06, "loss": 0.7275, "step": 32009 }, { "epoch": 3.406043839114705, "grad_norm": 13.625, "learning_rate": 8.880400795906766e-06, "loss": 0.6129, "step": 32010 }, { "epoch": 3.4061502447329217, "grad_norm": 14.0546875, "learning_rate": 8.880365264354747e-06, "loss": 0.5565, "step": 32011 }, { "epoch": 3.4062566503511387, "grad_norm": 12.609375, "learning_rate": 8.880329732802729e-06, "loss": 0.6493, "step": 32012 }, { "epoch": 3.406363055969355, "grad_norm": 11.609375, "learning_rate": 8.88029420125071e-06, "loss": 0.5867, "step": 32013 }, { "epoch": 3.4064694615875717, "grad_norm": 13.734375, "learning_rate": 8.880258669698694e-06, "loss": 0.7899, "step": 32014 }, { "epoch": 3.4065758672057886, "grad_norm": 14.203125, "learning_rate": 8.880223138146676e-06, "loss": 0.6991, "step": 32015 }, { "epoch": 3.406682272824005, "grad_norm": 11.8203125, "learning_rate": 8.880187606594657e-06, "loss": 0.5377, "step": 32016 }, { "epoch": 3.4067886784422217, "grad_norm": 14.5859375, "learning_rate": 8.880152075042639e-06, "loss": 0.6947, "step": 32017 }, { "epoch": 3.406895084060438, "grad_norm": 12.6171875, "learning_rate": 8.88011654349062e-06, "loss": 0.5171, "step": 32018 }, { "epoch": 3.407001489678655, "grad_norm": 12.78125, "learning_rate": 8.880081011938602e-06, "loss": 0.6259, "step": 32019 }, { "epoch": 3.4071078952968716, "grad_norm": 11.6015625, "learning_rate": 8.880045480386584e-06, "loss": 0.5104, "step": 32020 }, { "epoch": 3.407214300915088, "grad_norm": 10.234375, "learning_rate": 8.880009948834565e-06, "loss": 0.5754, "step": 32021 }, { "epoch": 3.407320706533305, "grad_norm": 16.1875, "learning_rate": 8.879974417282547e-06, "loss": 0.7018, "step": 32022 }, { "epoch": 3.4074271121515216, "grad_norm": 13.7890625, "learning_rate": 8.879938885730528e-06, "loss": 0.6311, "step": 32023 }, { "epoch": 3.407533517769738, "grad_norm": 10.765625, "learning_rate": 8.879903354178512e-06, "loss": 0.5876, "step": 32024 }, { "epoch": 3.407639923387955, "grad_norm": 16.3125, "learning_rate": 8.879867822626493e-06, "loss": 0.6735, "step": 32025 }, { "epoch": 3.4077463290061716, "grad_norm": 17.21875, "learning_rate": 8.879832291074475e-06, "loss": 0.6746, "step": 32026 }, { "epoch": 3.407852734624388, "grad_norm": 13.296875, "learning_rate": 8.879796759522457e-06, "loss": 0.6326, "step": 32027 }, { "epoch": 3.407959140242605, "grad_norm": 17.546875, "learning_rate": 8.879761227970438e-06, "loss": 0.6503, "step": 32028 }, { "epoch": 3.4080655458608216, "grad_norm": 17.484375, "learning_rate": 8.87972569641842e-06, "loss": 0.7205, "step": 32029 }, { "epoch": 3.408171951479038, "grad_norm": 15.3828125, "learning_rate": 8.879690164866403e-06, "loss": 0.7909, "step": 32030 }, { "epoch": 3.4082783570972546, "grad_norm": 12.484375, "learning_rate": 8.879654633314385e-06, "loss": 0.5336, "step": 32031 }, { "epoch": 3.4083847627154715, "grad_norm": 10.734375, "learning_rate": 8.879619101762366e-06, "loss": 0.5469, "step": 32032 }, { "epoch": 3.408491168333688, "grad_norm": 11.40625, "learning_rate": 8.879583570210348e-06, "loss": 0.5692, "step": 32033 }, { "epoch": 3.4085975739519045, "grad_norm": 16.078125, "learning_rate": 8.87954803865833e-06, "loss": 0.8671, "step": 32034 }, { "epoch": 3.408703979570121, "grad_norm": 14.171875, "learning_rate": 8.879512507106311e-06, "loss": 0.7835, "step": 32035 }, { "epoch": 3.408810385188338, "grad_norm": 10.6875, "learning_rate": 8.879476975554293e-06, "loss": 0.6386, "step": 32036 }, { "epoch": 3.4089167908065545, "grad_norm": 21.234375, "learning_rate": 8.879441444002274e-06, "loss": 0.8649, "step": 32037 }, { "epoch": 3.409023196424771, "grad_norm": 11.921875, "learning_rate": 8.879405912450256e-06, "loss": 0.6047, "step": 32038 }, { "epoch": 3.409129602042988, "grad_norm": 13.5703125, "learning_rate": 8.879370380898238e-06, "loss": 0.6268, "step": 32039 }, { "epoch": 3.4092360076612045, "grad_norm": 10.4765625, "learning_rate": 8.879334849346221e-06, "loss": 0.5504, "step": 32040 }, { "epoch": 3.409342413279421, "grad_norm": 11.421875, "learning_rate": 8.879299317794202e-06, "loss": 0.6069, "step": 32041 }, { "epoch": 3.409448818897638, "grad_norm": 11.6875, "learning_rate": 8.879263786242184e-06, "loss": 0.6953, "step": 32042 }, { "epoch": 3.4095552245158545, "grad_norm": 11.671875, "learning_rate": 8.879228254690166e-06, "loss": 0.5431, "step": 32043 }, { "epoch": 3.409661630134071, "grad_norm": 13.6328125, "learning_rate": 8.879192723138147e-06, "loss": 0.6197, "step": 32044 }, { "epoch": 3.409768035752288, "grad_norm": 14.2265625, "learning_rate": 8.879157191586129e-06, "loss": 0.8203, "step": 32045 }, { "epoch": 3.4098744413705044, "grad_norm": 14.84375, "learning_rate": 8.87912166003411e-06, "loss": 0.644, "step": 32046 }, { "epoch": 3.409980846988721, "grad_norm": 12.140625, "learning_rate": 8.879086128482092e-06, "loss": 0.6092, "step": 32047 }, { "epoch": 3.4100872526069375, "grad_norm": 14.21875, "learning_rate": 8.879050596930074e-06, "loss": 0.665, "step": 32048 }, { "epoch": 3.4101936582251544, "grad_norm": 13.453125, "learning_rate": 8.879015065378055e-06, "loss": 0.5517, "step": 32049 }, { "epoch": 3.410300063843371, "grad_norm": 10.640625, "learning_rate": 8.878979533826039e-06, "loss": 0.6007, "step": 32050 }, { "epoch": 3.4104064694615874, "grad_norm": 16.796875, "learning_rate": 8.87894400227402e-06, "loss": 0.674, "step": 32051 }, { "epoch": 3.4105128750798044, "grad_norm": 15.40625, "learning_rate": 8.878908470722002e-06, "loss": 0.6529, "step": 32052 }, { "epoch": 3.410619280698021, "grad_norm": 14.3828125, "learning_rate": 8.878872939169983e-06, "loss": 0.7363, "step": 32053 }, { "epoch": 3.4107256863162374, "grad_norm": 11.6015625, "learning_rate": 8.878837407617965e-06, "loss": 0.5581, "step": 32054 }, { "epoch": 3.410832091934454, "grad_norm": 16.546875, "learning_rate": 8.878801876065948e-06, "loss": 0.6667, "step": 32055 }, { "epoch": 3.410938497552671, "grad_norm": 13.5390625, "learning_rate": 8.87876634451393e-06, "loss": 0.6023, "step": 32056 }, { "epoch": 3.4110449031708874, "grad_norm": 12.96875, "learning_rate": 8.878730812961912e-06, "loss": 0.5835, "step": 32057 }, { "epoch": 3.411151308789104, "grad_norm": 13.4140625, "learning_rate": 8.878695281409893e-06, "loss": 0.6788, "step": 32058 }, { "epoch": 3.411257714407321, "grad_norm": 15.3984375, "learning_rate": 8.878659749857875e-06, "loss": 0.7445, "step": 32059 }, { "epoch": 3.4113641200255373, "grad_norm": 13.71875, "learning_rate": 8.878624218305856e-06, "loss": 0.5979, "step": 32060 }, { "epoch": 3.411470525643754, "grad_norm": 17.859375, "learning_rate": 8.878588686753838e-06, "loss": 0.6789, "step": 32061 }, { "epoch": 3.411576931261971, "grad_norm": 12.1640625, "learning_rate": 8.87855315520182e-06, "loss": 0.5094, "step": 32062 }, { "epoch": 3.4116833368801873, "grad_norm": 12.5546875, "learning_rate": 8.878517623649801e-06, "loss": 0.6472, "step": 32063 }, { "epoch": 3.411789742498404, "grad_norm": 10.96875, "learning_rate": 8.878482092097783e-06, "loss": 0.6015, "step": 32064 }, { "epoch": 3.411896148116621, "grad_norm": 16.75, "learning_rate": 8.878446560545764e-06, "loss": 0.6768, "step": 32065 }, { "epoch": 3.4120025537348373, "grad_norm": 14.0390625, "learning_rate": 8.878411028993748e-06, "loss": 0.6695, "step": 32066 }, { "epoch": 3.412108959353054, "grad_norm": 11.28125, "learning_rate": 8.87837549744173e-06, "loss": 0.5381, "step": 32067 }, { "epoch": 3.4122153649712703, "grad_norm": 14.4296875, "learning_rate": 8.878339965889711e-06, "loss": 0.684, "step": 32068 }, { "epoch": 3.4123217705894873, "grad_norm": 11.0234375, "learning_rate": 8.878304434337693e-06, "loss": 0.52, "step": 32069 }, { "epoch": 3.412428176207704, "grad_norm": 11.5703125, "learning_rate": 8.878268902785674e-06, "loss": 0.634, "step": 32070 }, { "epoch": 3.4125345818259203, "grad_norm": 13.71875, "learning_rate": 8.878233371233656e-06, "loss": 0.7707, "step": 32071 }, { "epoch": 3.4126409874441372, "grad_norm": 15.046875, "learning_rate": 8.878197839681637e-06, "loss": 0.769, "step": 32072 }, { "epoch": 3.4127473930623538, "grad_norm": 16.84375, "learning_rate": 8.878162308129619e-06, "loss": 0.7634, "step": 32073 }, { "epoch": 3.4128537986805703, "grad_norm": 17.9375, "learning_rate": 8.8781267765776e-06, "loss": 0.8787, "step": 32074 }, { "epoch": 3.4129602042987868, "grad_norm": 14.8671875, "learning_rate": 8.878091245025582e-06, "loss": 0.7163, "step": 32075 }, { "epoch": 3.4130666099170037, "grad_norm": 13.65625, "learning_rate": 8.878055713473565e-06, "loss": 0.684, "step": 32076 }, { "epoch": 3.4131730155352202, "grad_norm": 12.3203125, "learning_rate": 8.878020181921547e-06, "loss": 0.6324, "step": 32077 }, { "epoch": 3.4132794211534367, "grad_norm": 16.296875, "learning_rate": 8.877984650369529e-06, "loss": 0.6256, "step": 32078 }, { "epoch": 3.4133858267716537, "grad_norm": 15.3828125, "learning_rate": 8.87794911881751e-06, "loss": 0.7538, "step": 32079 }, { "epoch": 3.41349223238987, "grad_norm": 14.9609375, "learning_rate": 8.877913587265492e-06, "loss": 0.7112, "step": 32080 }, { "epoch": 3.4135986380080867, "grad_norm": 13.5859375, "learning_rate": 8.877878055713475e-06, "loss": 0.6365, "step": 32081 }, { "epoch": 3.4137050436263037, "grad_norm": 18.203125, "learning_rate": 8.877842524161457e-06, "loss": 0.6552, "step": 32082 }, { "epoch": 3.41381144924452, "grad_norm": 11.734375, "learning_rate": 8.877806992609438e-06, "loss": 0.5834, "step": 32083 }, { "epoch": 3.4139178548627367, "grad_norm": 13.71875, "learning_rate": 8.87777146105742e-06, "loss": 0.6353, "step": 32084 }, { "epoch": 3.4140242604809536, "grad_norm": 12.1171875, "learning_rate": 8.877735929505402e-06, "loss": 0.6114, "step": 32085 }, { "epoch": 3.41413066609917, "grad_norm": 14.2578125, "learning_rate": 8.877700397953383e-06, "loss": 0.6056, "step": 32086 }, { "epoch": 3.4142370717173867, "grad_norm": 12.46875, "learning_rate": 8.877664866401365e-06, "loss": 0.7127, "step": 32087 }, { "epoch": 3.414343477335603, "grad_norm": 13.765625, "learning_rate": 8.877629334849346e-06, "loss": 0.7768, "step": 32088 }, { "epoch": 3.41444988295382, "grad_norm": 14.671875, "learning_rate": 8.877593803297328e-06, "loss": 0.675, "step": 32089 }, { "epoch": 3.4145562885720366, "grad_norm": 11.3125, "learning_rate": 8.87755827174531e-06, "loss": 0.5925, "step": 32090 }, { "epoch": 3.414662694190253, "grad_norm": 13.59375, "learning_rate": 8.877522740193293e-06, "loss": 0.6603, "step": 32091 }, { "epoch": 3.4147690998084697, "grad_norm": 11.828125, "learning_rate": 8.877487208641275e-06, "loss": 0.6273, "step": 32092 }, { "epoch": 3.4148755054266866, "grad_norm": 16.90625, "learning_rate": 8.877451677089256e-06, "loss": 0.7935, "step": 32093 }, { "epoch": 3.414981911044903, "grad_norm": 13.6953125, "learning_rate": 8.877416145537238e-06, "loss": 0.7497, "step": 32094 }, { "epoch": 3.4150883166631196, "grad_norm": 13.0390625, "learning_rate": 8.87738061398522e-06, "loss": 0.6315, "step": 32095 }, { "epoch": 3.4151947222813366, "grad_norm": 11.4375, "learning_rate": 8.877345082433203e-06, "loss": 0.6035, "step": 32096 }, { "epoch": 3.415301127899553, "grad_norm": 13.765625, "learning_rate": 8.877309550881184e-06, "loss": 0.6285, "step": 32097 }, { "epoch": 3.4154075335177696, "grad_norm": 12.1171875, "learning_rate": 8.877274019329166e-06, "loss": 0.6595, "step": 32098 }, { "epoch": 3.4155139391359866, "grad_norm": 14.34375, "learning_rate": 8.877238487777147e-06, "loss": 0.6817, "step": 32099 }, { "epoch": 3.415620344754203, "grad_norm": 10.4453125, "learning_rate": 8.877202956225129e-06, "loss": 0.6859, "step": 32100 }, { "epoch": 3.4157267503724196, "grad_norm": 9.34375, "learning_rate": 8.87716742467311e-06, "loss": 0.6523, "step": 32101 }, { "epoch": 3.4158331559906365, "grad_norm": 13.9375, "learning_rate": 8.877131893121092e-06, "loss": 0.7435, "step": 32102 }, { "epoch": 3.415939561608853, "grad_norm": 13.390625, "learning_rate": 8.877096361569074e-06, "loss": 0.6693, "step": 32103 }, { "epoch": 3.4160459672270695, "grad_norm": 12.4296875, "learning_rate": 8.877060830017056e-06, "loss": 0.6234, "step": 32104 }, { "epoch": 3.416152372845286, "grad_norm": 12.359375, "learning_rate": 8.877025298465037e-06, "loss": 0.677, "step": 32105 }, { "epoch": 3.416258778463503, "grad_norm": 24.546875, "learning_rate": 8.876989766913019e-06, "loss": 0.873, "step": 32106 }, { "epoch": 3.4163651840817195, "grad_norm": 10.1484375, "learning_rate": 8.876954235361002e-06, "loss": 0.5618, "step": 32107 }, { "epoch": 3.416471589699936, "grad_norm": 11.578125, "learning_rate": 8.876918703808984e-06, "loss": 0.5754, "step": 32108 }, { "epoch": 3.416577995318153, "grad_norm": 12.15625, "learning_rate": 8.876883172256965e-06, "loss": 0.6513, "step": 32109 }, { "epoch": 3.4166844009363695, "grad_norm": 12.1484375, "learning_rate": 8.876847640704947e-06, "loss": 0.6002, "step": 32110 }, { "epoch": 3.416790806554586, "grad_norm": 14.4140625, "learning_rate": 8.876812109152928e-06, "loss": 0.4716, "step": 32111 }, { "epoch": 3.4168972121728025, "grad_norm": 10.046875, "learning_rate": 8.87677657760091e-06, "loss": 0.4689, "step": 32112 }, { "epoch": 3.4170036177910195, "grad_norm": 17.09375, "learning_rate": 8.876741046048892e-06, "loss": 0.845, "step": 32113 }, { "epoch": 3.417110023409236, "grad_norm": 15.2265625, "learning_rate": 8.876705514496873e-06, "loss": 0.6582, "step": 32114 }, { "epoch": 3.4172164290274525, "grad_norm": 15.3984375, "learning_rate": 8.876669982944855e-06, "loss": 0.7033, "step": 32115 }, { "epoch": 3.4173228346456694, "grad_norm": 13.5, "learning_rate": 8.876634451392837e-06, "loss": 0.7792, "step": 32116 }, { "epoch": 3.417429240263886, "grad_norm": 10.0078125, "learning_rate": 8.87659891984082e-06, "loss": 0.5368, "step": 32117 }, { "epoch": 3.4175356458821025, "grad_norm": 9.9375, "learning_rate": 8.876563388288801e-06, "loss": 0.5533, "step": 32118 }, { "epoch": 3.4176420515003194, "grad_norm": 9.5078125, "learning_rate": 8.876527856736783e-06, "loss": 0.4671, "step": 32119 }, { "epoch": 3.417748457118536, "grad_norm": 14.5625, "learning_rate": 8.876492325184765e-06, "loss": 0.7142, "step": 32120 }, { "epoch": 3.4178548627367524, "grad_norm": 16.546875, "learning_rate": 8.876456793632746e-06, "loss": 0.7005, "step": 32121 }, { "epoch": 3.4179612683549694, "grad_norm": 17.9375, "learning_rate": 8.87642126208073e-06, "loss": 0.6676, "step": 32122 }, { "epoch": 3.418067673973186, "grad_norm": 13.359375, "learning_rate": 8.876385730528711e-06, "loss": 0.6482, "step": 32123 }, { "epoch": 3.4181740795914024, "grad_norm": 11.0, "learning_rate": 8.876350198976693e-06, "loss": 0.6445, "step": 32124 }, { "epoch": 3.418280485209619, "grad_norm": 12.890625, "learning_rate": 8.876314667424674e-06, "loss": 0.801, "step": 32125 }, { "epoch": 3.418386890827836, "grad_norm": 13.8984375, "learning_rate": 8.876279135872656e-06, "loss": 0.6936, "step": 32126 }, { "epoch": 3.4184932964460524, "grad_norm": 14.7265625, "learning_rate": 8.876243604320638e-06, "loss": 0.7263, "step": 32127 }, { "epoch": 3.418599702064269, "grad_norm": 12.5, "learning_rate": 8.87620807276862e-06, "loss": 0.6226, "step": 32128 }, { "epoch": 3.4187061076824854, "grad_norm": 13.0390625, "learning_rate": 8.8761725412166e-06, "loss": 0.7996, "step": 32129 }, { "epoch": 3.4188125133007023, "grad_norm": 12.328125, "learning_rate": 8.876137009664582e-06, "loss": 0.6964, "step": 32130 }, { "epoch": 3.418918918918919, "grad_norm": 17.625, "learning_rate": 8.876101478112564e-06, "loss": 0.6891, "step": 32131 }, { "epoch": 3.4190253245371354, "grad_norm": 11.6328125, "learning_rate": 8.876065946560547e-06, "loss": 0.5099, "step": 32132 }, { "epoch": 3.4191317301553523, "grad_norm": 11.1640625, "learning_rate": 8.876030415008529e-06, "loss": 0.5609, "step": 32133 }, { "epoch": 3.419238135773569, "grad_norm": 15.0859375, "learning_rate": 8.87599488345651e-06, "loss": 0.6334, "step": 32134 }, { "epoch": 3.4193445413917853, "grad_norm": 11.0234375, "learning_rate": 8.875959351904492e-06, "loss": 0.6262, "step": 32135 }, { "epoch": 3.4194509470100023, "grad_norm": 15.2109375, "learning_rate": 8.875923820352474e-06, "loss": 0.6358, "step": 32136 }, { "epoch": 3.419557352628219, "grad_norm": 11.5546875, "learning_rate": 8.875888288800455e-06, "loss": 0.6359, "step": 32137 }, { "epoch": 3.4196637582464353, "grad_norm": 12.671875, "learning_rate": 8.875852757248437e-06, "loss": 0.6049, "step": 32138 }, { "epoch": 3.4197701638646523, "grad_norm": 13.2578125, "learning_rate": 8.875817225696419e-06, "loss": 0.6275, "step": 32139 }, { "epoch": 3.4198765694828688, "grad_norm": 14.5390625, "learning_rate": 8.8757816941444e-06, "loss": 0.7621, "step": 32140 }, { "epoch": 3.4199829751010853, "grad_norm": 11.9453125, "learning_rate": 8.875746162592382e-06, "loss": 0.6297, "step": 32141 }, { "epoch": 3.420089380719302, "grad_norm": 14.3828125, "learning_rate": 8.875710631040365e-06, "loss": 0.5795, "step": 32142 }, { "epoch": 3.4201957863375188, "grad_norm": 11.1875, "learning_rate": 8.875675099488347e-06, "loss": 0.6387, "step": 32143 }, { "epoch": 3.4203021919557353, "grad_norm": 10.71875, "learning_rate": 8.875639567936328e-06, "loss": 0.4422, "step": 32144 }, { "epoch": 3.4204085975739518, "grad_norm": 16.328125, "learning_rate": 8.87560403638431e-06, "loss": 0.7925, "step": 32145 }, { "epoch": 3.4205150031921687, "grad_norm": 12.78125, "learning_rate": 8.875568504832291e-06, "loss": 0.5544, "step": 32146 }, { "epoch": 3.4206214088103852, "grad_norm": 14.5078125, "learning_rate": 8.875532973280273e-06, "loss": 0.7175, "step": 32147 }, { "epoch": 3.4207278144286017, "grad_norm": 10.859375, "learning_rate": 8.875497441728256e-06, "loss": 0.6215, "step": 32148 }, { "epoch": 3.4208342200468183, "grad_norm": 13.3359375, "learning_rate": 8.875461910176238e-06, "loss": 0.6211, "step": 32149 }, { "epoch": 3.420940625665035, "grad_norm": 12.203125, "learning_rate": 8.87542637862422e-06, "loss": 0.6276, "step": 32150 }, { "epoch": 3.4210470312832517, "grad_norm": 13.9765625, "learning_rate": 8.875390847072201e-06, "loss": 0.6862, "step": 32151 }, { "epoch": 3.4211534369014682, "grad_norm": 11.5546875, "learning_rate": 8.875355315520183e-06, "loss": 0.5429, "step": 32152 }, { "epoch": 3.421259842519685, "grad_norm": 12.8359375, "learning_rate": 8.875319783968164e-06, "loss": 0.6134, "step": 32153 }, { "epoch": 3.4213662481379017, "grad_norm": 13.5078125, "learning_rate": 8.875284252416146e-06, "loss": 0.6924, "step": 32154 }, { "epoch": 3.421472653756118, "grad_norm": 11.984375, "learning_rate": 8.875248720864128e-06, "loss": 0.7374, "step": 32155 }, { "epoch": 3.421579059374335, "grad_norm": 11.7109375, "learning_rate": 8.87521318931211e-06, "loss": 0.5492, "step": 32156 }, { "epoch": 3.4216854649925517, "grad_norm": 13.5859375, "learning_rate": 8.875177657760091e-06, "loss": 0.6681, "step": 32157 }, { "epoch": 3.421791870610768, "grad_norm": 10.2578125, "learning_rate": 8.875142126208074e-06, "loss": 0.4726, "step": 32158 }, { "epoch": 3.421898276228985, "grad_norm": 14.84375, "learning_rate": 8.875106594656056e-06, "loss": 0.731, "step": 32159 }, { "epoch": 3.4220046818472016, "grad_norm": 11.375, "learning_rate": 8.875071063104037e-06, "loss": 0.5272, "step": 32160 }, { "epoch": 3.422111087465418, "grad_norm": 11.984375, "learning_rate": 8.875035531552019e-06, "loss": 0.646, "step": 32161 }, { "epoch": 3.4222174930836347, "grad_norm": 11.6953125, "learning_rate": 8.875e-06, "loss": 0.5588, "step": 32162 }, { "epoch": 3.4223238987018516, "grad_norm": 13.0859375, "learning_rate": 8.874964468447982e-06, "loss": 0.6497, "step": 32163 }, { "epoch": 3.422430304320068, "grad_norm": 11.625, "learning_rate": 8.874928936895964e-06, "loss": 0.6751, "step": 32164 }, { "epoch": 3.4225367099382846, "grad_norm": 14.734375, "learning_rate": 8.874893405343945e-06, "loss": 0.7638, "step": 32165 }, { "epoch": 3.422643115556501, "grad_norm": 11.6328125, "learning_rate": 8.874857873791927e-06, "loss": 0.6022, "step": 32166 }, { "epoch": 3.422749521174718, "grad_norm": 13.3828125, "learning_rate": 8.874822342239909e-06, "loss": 0.5585, "step": 32167 }, { "epoch": 3.4228559267929346, "grad_norm": 15.53125, "learning_rate": 8.874786810687892e-06, "loss": 0.7634, "step": 32168 }, { "epoch": 3.422962332411151, "grad_norm": 14.0, "learning_rate": 8.874751279135874e-06, "loss": 0.7498, "step": 32169 }, { "epoch": 3.423068738029368, "grad_norm": 13.59375, "learning_rate": 8.874715747583855e-06, "loss": 0.8455, "step": 32170 }, { "epoch": 3.4231751436475846, "grad_norm": 12.5078125, "learning_rate": 8.874680216031837e-06, "loss": 0.6324, "step": 32171 }, { "epoch": 3.423281549265801, "grad_norm": 12.734375, "learning_rate": 8.874644684479818e-06, "loss": 0.5892, "step": 32172 }, { "epoch": 3.423387954884018, "grad_norm": 13.7578125, "learning_rate": 8.874609152927802e-06, "loss": 0.655, "step": 32173 }, { "epoch": 3.4234943605022345, "grad_norm": 20.421875, "learning_rate": 8.874573621375783e-06, "loss": 0.7789, "step": 32174 }, { "epoch": 3.423600766120451, "grad_norm": 14.3828125, "learning_rate": 8.874538089823765e-06, "loss": 0.7359, "step": 32175 }, { "epoch": 3.423707171738668, "grad_norm": 12.6171875, "learning_rate": 8.874502558271746e-06, "loss": 0.6109, "step": 32176 }, { "epoch": 3.4238135773568845, "grad_norm": 14.0625, "learning_rate": 8.874467026719728e-06, "loss": 0.6401, "step": 32177 }, { "epoch": 3.423919982975101, "grad_norm": 13.171875, "learning_rate": 8.87443149516771e-06, "loss": 0.6015, "step": 32178 }, { "epoch": 3.4240263885933175, "grad_norm": 11.8203125, "learning_rate": 8.874395963615691e-06, "loss": 0.6315, "step": 32179 }, { "epoch": 3.4241327942115345, "grad_norm": 12.6328125, "learning_rate": 8.874360432063673e-06, "loss": 0.5758, "step": 32180 }, { "epoch": 3.424239199829751, "grad_norm": 13.8125, "learning_rate": 8.874324900511654e-06, "loss": 0.6882, "step": 32181 }, { "epoch": 3.4243456054479675, "grad_norm": 13.0859375, "learning_rate": 8.874289368959636e-06, "loss": 0.5388, "step": 32182 }, { "epoch": 3.4244520110661845, "grad_norm": 14.0546875, "learning_rate": 8.874253837407618e-06, "loss": 0.5682, "step": 32183 }, { "epoch": 3.424558416684401, "grad_norm": 16.875, "learning_rate": 8.874218305855601e-06, "loss": 0.7381, "step": 32184 }, { "epoch": 3.4246648223026175, "grad_norm": 14.4609375, "learning_rate": 8.874182774303583e-06, "loss": 0.643, "step": 32185 }, { "epoch": 3.424771227920834, "grad_norm": 12.5703125, "learning_rate": 8.874147242751564e-06, "loss": 0.6288, "step": 32186 }, { "epoch": 3.424877633539051, "grad_norm": 10.578125, "learning_rate": 8.874111711199546e-06, "loss": 0.5863, "step": 32187 }, { "epoch": 3.4249840391572675, "grad_norm": 14.8203125, "learning_rate": 8.874076179647527e-06, "loss": 0.6929, "step": 32188 }, { "epoch": 3.425090444775484, "grad_norm": 12.3828125, "learning_rate": 8.87404064809551e-06, "loss": 0.543, "step": 32189 }, { "epoch": 3.425196850393701, "grad_norm": 18.21875, "learning_rate": 8.874005116543492e-06, "loss": 0.6918, "step": 32190 }, { "epoch": 3.4253032560119174, "grad_norm": 12.8984375, "learning_rate": 8.873969584991472e-06, "loss": 0.6164, "step": 32191 }, { "epoch": 3.425409661630134, "grad_norm": 14.1328125, "learning_rate": 8.873934053439454e-06, "loss": 0.6856, "step": 32192 }, { "epoch": 3.425516067248351, "grad_norm": 16.65625, "learning_rate": 8.873898521887435e-06, "loss": 0.8177, "step": 32193 }, { "epoch": 3.4256224728665674, "grad_norm": 13.0546875, "learning_rate": 8.873862990335419e-06, "loss": 0.6466, "step": 32194 }, { "epoch": 3.425728878484784, "grad_norm": 12.1875, "learning_rate": 8.8738274587834e-06, "loss": 0.6664, "step": 32195 }, { "epoch": 3.425835284103001, "grad_norm": 13.984375, "learning_rate": 8.873791927231382e-06, "loss": 0.7192, "step": 32196 }, { "epoch": 3.4259416897212174, "grad_norm": 16.671875, "learning_rate": 8.873756395679364e-06, "loss": 0.6835, "step": 32197 }, { "epoch": 3.426048095339434, "grad_norm": 10.5703125, "learning_rate": 8.873720864127345e-06, "loss": 0.5642, "step": 32198 }, { "epoch": 3.4261545009576504, "grad_norm": 11.5234375, "learning_rate": 8.873685332575328e-06, "loss": 0.544, "step": 32199 }, { "epoch": 3.4262609065758673, "grad_norm": 17.03125, "learning_rate": 8.87364980102331e-06, "loss": 0.6962, "step": 32200 }, { "epoch": 3.426367312194084, "grad_norm": 12.8125, "learning_rate": 8.873614269471292e-06, "loss": 0.7138, "step": 32201 }, { "epoch": 3.4264737178123004, "grad_norm": 14.3125, "learning_rate": 8.873578737919273e-06, "loss": 0.7117, "step": 32202 }, { "epoch": 3.4265801234305173, "grad_norm": 14.671875, "learning_rate": 8.873543206367255e-06, "loss": 0.7195, "step": 32203 }, { "epoch": 3.426686529048734, "grad_norm": 18.03125, "learning_rate": 8.873507674815237e-06, "loss": 0.7944, "step": 32204 }, { "epoch": 3.4267929346669503, "grad_norm": 18.59375, "learning_rate": 8.873472143263218e-06, "loss": 0.8024, "step": 32205 }, { "epoch": 3.426899340285167, "grad_norm": 13.7421875, "learning_rate": 8.8734366117112e-06, "loss": 0.6062, "step": 32206 }, { "epoch": 3.427005745903384, "grad_norm": 10.90625, "learning_rate": 8.873401080159181e-06, "loss": 0.5501, "step": 32207 }, { "epoch": 3.4271121515216003, "grad_norm": 13.671875, "learning_rate": 8.873365548607163e-06, "loss": 0.6128, "step": 32208 }, { "epoch": 3.427218557139817, "grad_norm": 11.34375, "learning_rate": 8.873330017055146e-06, "loss": 0.5832, "step": 32209 }, { "epoch": 3.4273249627580338, "grad_norm": 14.421875, "learning_rate": 8.873294485503128e-06, "loss": 0.6769, "step": 32210 }, { "epoch": 3.4274313683762503, "grad_norm": 14.625, "learning_rate": 8.87325895395111e-06, "loss": 0.6261, "step": 32211 }, { "epoch": 3.427537773994467, "grad_norm": 14.8359375, "learning_rate": 8.873223422399091e-06, "loss": 0.7122, "step": 32212 }, { "epoch": 3.4276441796126838, "grad_norm": 13.9453125, "learning_rate": 8.873187890847073e-06, "loss": 0.6685, "step": 32213 }, { "epoch": 3.4277505852309003, "grad_norm": 17.75, "learning_rate": 8.873152359295056e-06, "loss": 0.6959, "step": 32214 }, { "epoch": 3.4278569908491168, "grad_norm": 13.9140625, "learning_rate": 8.873116827743038e-06, "loss": 0.7742, "step": 32215 }, { "epoch": 3.4279633964673337, "grad_norm": 11.828125, "learning_rate": 8.87308129619102e-06, "loss": 0.5825, "step": 32216 }, { "epoch": 3.4280698020855502, "grad_norm": 11.828125, "learning_rate": 8.873045764639e-06, "loss": 0.5589, "step": 32217 }, { "epoch": 3.4281762077037667, "grad_norm": 10.71875, "learning_rate": 8.873010233086982e-06, "loss": 0.5268, "step": 32218 }, { "epoch": 3.4282826133219833, "grad_norm": 14.375, "learning_rate": 8.872974701534964e-06, "loss": 0.6739, "step": 32219 }, { "epoch": 3.4283890189402, "grad_norm": 14.28125, "learning_rate": 8.872939169982946e-06, "loss": 0.6838, "step": 32220 }, { "epoch": 3.4284954245584167, "grad_norm": 13.171875, "learning_rate": 8.872903638430927e-06, "loss": 0.6895, "step": 32221 }, { "epoch": 3.4286018301766332, "grad_norm": 10.7265625, "learning_rate": 8.872868106878909e-06, "loss": 0.5505, "step": 32222 }, { "epoch": 3.4287082357948497, "grad_norm": 10.9453125, "learning_rate": 8.87283257532689e-06, "loss": 0.4938, "step": 32223 }, { "epoch": 3.4288146414130667, "grad_norm": 12.3125, "learning_rate": 8.872797043774872e-06, "loss": 0.7439, "step": 32224 }, { "epoch": 3.428921047031283, "grad_norm": 11.1328125, "learning_rate": 8.872761512222855e-06, "loss": 0.5836, "step": 32225 }, { "epoch": 3.4290274526494997, "grad_norm": 14.875, "learning_rate": 8.872725980670837e-06, "loss": 0.5548, "step": 32226 }, { "epoch": 3.4291338582677167, "grad_norm": 9.9296875, "learning_rate": 8.872690449118819e-06, "loss": 0.4449, "step": 32227 }, { "epoch": 3.429240263885933, "grad_norm": 16.296875, "learning_rate": 8.8726549175668e-06, "loss": 0.6204, "step": 32228 }, { "epoch": 3.4293466695041497, "grad_norm": 13.265625, "learning_rate": 8.872619386014782e-06, "loss": 0.651, "step": 32229 }, { "epoch": 3.4294530751223666, "grad_norm": 9.8515625, "learning_rate": 8.872583854462763e-06, "loss": 0.4459, "step": 32230 }, { "epoch": 3.429559480740583, "grad_norm": 9.6328125, "learning_rate": 8.872548322910745e-06, "loss": 0.5371, "step": 32231 }, { "epoch": 3.4296658863587997, "grad_norm": 17.734375, "learning_rate": 8.872512791358727e-06, "loss": 0.7156, "step": 32232 }, { "epoch": 3.4297722919770166, "grad_norm": 17.0625, "learning_rate": 8.872477259806708e-06, "loss": 0.7663, "step": 32233 }, { "epoch": 3.429878697595233, "grad_norm": 10.90625, "learning_rate": 8.87244172825469e-06, "loss": 0.5347, "step": 32234 }, { "epoch": 3.4299851032134496, "grad_norm": 12.6640625, "learning_rate": 8.872406196702673e-06, "loss": 0.7007, "step": 32235 }, { "epoch": 3.430091508831666, "grad_norm": 13.4296875, "learning_rate": 8.872370665150655e-06, "loss": 0.6183, "step": 32236 }, { "epoch": 3.430197914449883, "grad_norm": 13.2890625, "learning_rate": 8.872335133598636e-06, "loss": 0.6495, "step": 32237 }, { "epoch": 3.4303043200680996, "grad_norm": 11.953125, "learning_rate": 8.872299602046618e-06, "loss": 0.5805, "step": 32238 }, { "epoch": 3.430410725686316, "grad_norm": 9.6328125, "learning_rate": 8.8722640704946e-06, "loss": 0.4207, "step": 32239 }, { "epoch": 3.430517131304533, "grad_norm": 16.296875, "learning_rate": 8.872228538942583e-06, "loss": 0.6517, "step": 32240 }, { "epoch": 3.4306235369227496, "grad_norm": 11.6484375, "learning_rate": 8.872193007390564e-06, "loss": 0.6191, "step": 32241 }, { "epoch": 3.430729942540966, "grad_norm": 11.2109375, "learning_rate": 8.872157475838546e-06, "loss": 0.5303, "step": 32242 }, { "epoch": 3.4308363481591826, "grad_norm": 10.484375, "learning_rate": 8.872121944286528e-06, "loss": 0.5493, "step": 32243 }, { "epoch": 3.4309427537773995, "grad_norm": 9.265625, "learning_rate": 8.87208641273451e-06, "loss": 0.5038, "step": 32244 }, { "epoch": 3.431049159395616, "grad_norm": 13.140625, "learning_rate": 8.872050881182491e-06, "loss": 0.7263, "step": 32245 }, { "epoch": 3.4311555650138326, "grad_norm": 13.921875, "learning_rate": 8.872015349630472e-06, "loss": 0.7472, "step": 32246 }, { "epoch": 3.4312619706320495, "grad_norm": 11.5234375, "learning_rate": 8.871979818078454e-06, "loss": 0.6119, "step": 32247 }, { "epoch": 3.431368376250266, "grad_norm": 14.1875, "learning_rate": 8.871944286526436e-06, "loss": 0.8164, "step": 32248 }, { "epoch": 3.4314747818684825, "grad_norm": 14.3203125, "learning_rate": 8.871908754974417e-06, "loss": 0.6462, "step": 32249 }, { "epoch": 3.4315811874866995, "grad_norm": 11.8125, "learning_rate": 8.8718732234224e-06, "loss": 0.6835, "step": 32250 }, { "epoch": 3.431687593104916, "grad_norm": 13.4140625, "learning_rate": 8.871837691870382e-06, "loss": 0.5867, "step": 32251 }, { "epoch": 3.4317939987231325, "grad_norm": 18.59375, "learning_rate": 8.871802160318364e-06, "loss": 0.783, "step": 32252 }, { "epoch": 3.4319004043413495, "grad_norm": 13.375, "learning_rate": 8.871766628766345e-06, "loss": 0.5691, "step": 32253 }, { "epoch": 3.432006809959566, "grad_norm": 10.9296875, "learning_rate": 8.871731097214327e-06, "loss": 0.5031, "step": 32254 }, { "epoch": 3.4321132155777825, "grad_norm": 15.5546875, "learning_rate": 8.871695565662309e-06, "loss": 0.6427, "step": 32255 }, { "epoch": 3.432219621195999, "grad_norm": 13.328125, "learning_rate": 8.87166003411029e-06, "loss": 0.6308, "step": 32256 }, { "epoch": 3.432326026814216, "grad_norm": 14.5859375, "learning_rate": 8.871624502558272e-06, "loss": 0.6755, "step": 32257 }, { "epoch": 3.4324324324324325, "grad_norm": 10.671875, "learning_rate": 8.871588971006253e-06, "loss": 0.5152, "step": 32258 }, { "epoch": 3.432538838050649, "grad_norm": 13.96875, "learning_rate": 8.871553439454235e-06, "loss": 0.6321, "step": 32259 }, { "epoch": 3.4326452436688655, "grad_norm": 16.5625, "learning_rate": 8.871517907902217e-06, "loss": 0.8002, "step": 32260 }, { "epoch": 3.4327516492870824, "grad_norm": 11.734375, "learning_rate": 8.8714823763502e-06, "loss": 0.5852, "step": 32261 }, { "epoch": 3.432858054905299, "grad_norm": 10.5234375, "learning_rate": 8.871446844798182e-06, "loss": 0.5918, "step": 32262 }, { "epoch": 3.4329644605235154, "grad_norm": 10.6875, "learning_rate": 8.871411313246163e-06, "loss": 0.6219, "step": 32263 }, { "epoch": 3.4330708661417324, "grad_norm": 12.15625, "learning_rate": 8.871375781694145e-06, "loss": 0.6045, "step": 32264 }, { "epoch": 3.433177271759949, "grad_norm": 13.5234375, "learning_rate": 8.871340250142126e-06, "loss": 0.6134, "step": 32265 }, { "epoch": 3.4332836773781654, "grad_norm": 11.3359375, "learning_rate": 8.87130471859011e-06, "loss": 0.5037, "step": 32266 }, { "epoch": 3.4333900829963824, "grad_norm": 13.46875, "learning_rate": 8.871269187038091e-06, "loss": 0.697, "step": 32267 }, { "epoch": 3.433496488614599, "grad_norm": 15.703125, "learning_rate": 8.871233655486073e-06, "loss": 0.6361, "step": 32268 }, { "epoch": 3.4336028942328154, "grad_norm": 20.9375, "learning_rate": 8.871198123934055e-06, "loss": 0.7629, "step": 32269 }, { "epoch": 3.4337092998510323, "grad_norm": 10.3359375, "learning_rate": 8.871162592382036e-06, "loss": 0.5061, "step": 32270 }, { "epoch": 3.433815705469249, "grad_norm": 11.9609375, "learning_rate": 8.871127060830018e-06, "loss": 0.5831, "step": 32271 }, { "epoch": 3.4339221110874654, "grad_norm": 13.046875, "learning_rate": 8.871091529278e-06, "loss": 0.6614, "step": 32272 }, { "epoch": 3.434028516705682, "grad_norm": 12.3828125, "learning_rate": 8.871055997725981e-06, "loss": 0.6159, "step": 32273 }, { "epoch": 3.434134922323899, "grad_norm": 15.3515625, "learning_rate": 8.871020466173963e-06, "loss": 0.6697, "step": 32274 }, { "epoch": 3.4342413279421153, "grad_norm": 13.0, "learning_rate": 8.870984934621944e-06, "loss": 0.7204, "step": 32275 }, { "epoch": 3.434347733560332, "grad_norm": 8.2265625, "learning_rate": 8.870949403069927e-06, "loss": 0.4899, "step": 32276 }, { "epoch": 3.434454139178549, "grad_norm": 10.71875, "learning_rate": 8.870913871517909e-06, "loss": 0.4951, "step": 32277 }, { "epoch": 3.4345605447967653, "grad_norm": 28.625, "learning_rate": 8.87087833996589e-06, "loss": 0.6258, "step": 32278 }, { "epoch": 3.434666950414982, "grad_norm": 12.90625, "learning_rate": 8.870842808413872e-06, "loss": 0.5872, "step": 32279 }, { "epoch": 3.4347733560331983, "grad_norm": 10.7265625, "learning_rate": 8.870807276861854e-06, "loss": 0.5061, "step": 32280 }, { "epoch": 3.4348797616514153, "grad_norm": 10.1484375, "learning_rate": 8.870771745309835e-06, "loss": 0.4712, "step": 32281 }, { "epoch": 3.434986167269632, "grad_norm": 14.90625, "learning_rate": 8.870736213757817e-06, "loss": 0.6614, "step": 32282 }, { "epoch": 3.4350925728878483, "grad_norm": 11.3125, "learning_rate": 8.870700682205799e-06, "loss": 0.6619, "step": 32283 }, { "epoch": 3.4351989785060653, "grad_norm": 14.8828125, "learning_rate": 8.87066515065378e-06, "loss": 0.7683, "step": 32284 }, { "epoch": 3.4353053841242818, "grad_norm": 12.015625, "learning_rate": 8.870629619101762e-06, "loss": 0.6926, "step": 32285 }, { "epoch": 3.4354117897424983, "grad_norm": 17.8125, "learning_rate": 8.870594087549745e-06, "loss": 0.7928, "step": 32286 }, { "epoch": 3.4355181953607152, "grad_norm": 15.9453125, "learning_rate": 8.870558555997727e-06, "loss": 0.656, "step": 32287 }, { "epoch": 3.4356246009789317, "grad_norm": 15.546875, "learning_rate": 8.870523024445708e-06, "loss": 0.815, "step": 32288 }, { "epoch": 3.4357310065971483, "grad_norm": 13.1875, "learning_rate": 8.87048749289369e-06, "loss": 0.6562, "step": 32289 }, { "epoch": 3.435837412215365, "grad_norm": 15.1875, "learning_rate": 8.870451961341672e-06, "loss": 0.8731, "step": 32290 }, { "epoch": 3.4359438178335817, "grad_norm": 10.8671875, "learning_rate": 8.870416429789655e-06, "loss": 0.5406, "step": 32291 }, { "epoch": 3.4360502234517982, "grad_norm": 12.609375, "learning_rate": 8.870380898237637e-06, "loss": 0.6506, "step": 32292 }, { "epoch": 3.4361566290700147, "grad_norm": 12.28125, "learning_rate": 8.870345366685618e-06, "loss": 0.5893, "step": 32293 }, { "epoch": 3.4362630346882317, "grad_norm": 20.421875, "learning_rate": 8.8703098351336e-06, "loss": 0.8075, "step": 32294 }, { "epoch": 3.436369440306448, "grad_norm": 10.65625, "learning_rate": 8.870274303581581e-06, "loss": 0.5326, "step": 32295 }, { "epoch": 3.4364758459246647, "grad_norm": 16.15625, "learning_rate": 8.870238772029563e-06, "loss": 0.8106, "step": 32296 }, { "epoch": 3.436582251542881, "grad_norm": 14.1015625, "learning_rate": 8.870203240477545e-06, "loss": 0.7637, "step": 32297 }, { "epoch": 3.436688657161098, "grad_norm": 12.6484375, "learning_rate": 8.870167708925526e-06, "loss": 0.6245, "step": 32298 }, { "epoch": 3.4367950627793147, "grad_norm": 13.4921875, "learning_rate": 8.870132177373508e-06, "loss": 0.6213, "step": 32299 }, { "epoch": 3.436901468397531, "grad_norm": 14.890625, "learning_rate": 8.87009664582149e-06, "loss": 0.7717, "step": 32300 }, { "epoch": 3.437007874015748, "grad_norm": 13.5078125, "learning_rate": 8.870061114269471e-06, "loss": 0.7429, "step": 32301 }, { "epoch": 3.4371142796339647, "grad_norm": 14.984375, "learning_rate": 8.870025582717454e-06, "loss": 0.6606, "step": 32302 }, { "epoch": 3.437220685252181, "grad_norm": 13.84375, "learning_rate": 8.869990051165436e-06, "loss": 0.6493, "step": 32303 }, { "epoch": 3.437327090870398, "grad_norm": 12.0546875, "learning_rate": 8.869954519613418e-06, "loss": 0.6071, "step": 32304 }, { "epoch": 3.4374334964886146, "grad_norm": 14.46875, "learning_rate": 8.869918988061399e-06, "loss": 0.5742, "step": 32305 }, { "epoch": 3.437539902106831, "grad_norm": 15.15625, "learning_rate": 8.86988345650938e-06, "loss": 0.7905, "step": 32306 }, { "epoch": 3.437646307725048, "grad_norm": 14.2421875, "learning_rate": 8.869847924957364e-06, "loss": 0.653, "step": 32307 }, { "epoch": 3.4377527133432646, "grad_norm": 14.8359375, "learning_rate": 8.869812393405346e-06, "loss": 0.6689, "step": 32308 }, { "epoch": 3.437859118961481, "grad_norm": 11.4765625, "learning_rate": 8.869776861853327e-06, "loss": 0.5853, "step": 32309 }, { "epoch": 3.4379655245796976, "grad_norm": 13.1171875, "learning_rate": 8.869741330301309e-06, "loss": 0.6248, "step": 32310 }, { "epoch": 3.4380719301979146, "grad_norm": 12.25, "learning_rate": 8.869705798749289e-06, "loss": 0.788, "step": 32311 }, { "epoch": 3.438178335816131, "grad_norm": 13.46875, "learning_rate": 8.869670267197272e-06, "loss": 0.6864, "step": 32312 }, { "epoch": 3.4382847414343476, "grad_norm": 14.0234375, "learning_rate": 8.869634735645254e-06, "loss": 0.6631, "step": 32313 }, { "epoch": 3.4383911470525645, "grad_norm": 19.265625, "learning_rate": 8.869599204093235e-06, "loss": 0.665, "step": 32314 }, { "epoch": 3.438497552670781, "grad_norm": 13.015625, "learning_rate": 8.869563672541217e-06, "loss": 0.753, "step": 32315 }, { "epoch": 3.4386039582889976, "grad_norm": 16.71875, "learning_rate": 8.869528140989198e-06, "loss": 0.7375, "step": 32316 }, { "epoch": 3.438710363907214, "grad_norm": 11.328125, "learning_rate": 8.869492609437182e-06, "loss": 0.507, "step": 32317 }, { "epoch": 3.438816769525431, "grad_norm": 16.359375, "learning_rate": 8.869457077885163e-06, "loss": 0.6276, "step": 32318 }, { "epoch": 3.4389231751436475, "grad_norm": 14.984375, "learning_rate": 8.869421546333145e-06, "loss": 0.6906, "step": 32319 }, { "epoch": 3.439029580761864, "grad_norm": 14.203125, "learning_rate": 8.869386014781127e-06, "loss": 0.7117, "step": 32320 }, { "epoch": 3.439135986380081, "grad_norm": 11.3125, "learning_rate": 8.869350483229108e-06, "loss": 0.7089, "step": 32321 }, { "epoch": 3.4392423919982975, "grad_norm": 13.7734375, "learning_rate": 8.86931495167709e-06, "loss": 0.6947, "step": 32322 }, { "epoch": 3.439348797616514, "grad_norm": 11.5, "learning_rate": 8.869279420125071e-06, "loss": 0.6485, "step": 32323 }, { "epoch": 3.439455203234731, "grad_norm": 16.1875, "learning_rate": 8.869243888573053e-06, "loss": 0.7467, "step": 32324 }, { "epoch": 3.4395616088529475, "grad_norm": 12.1796875, "learning_rate": 8.869208357021035e-06, "loss": 0.6336, "step": 32325 }, { "epoch": 3.439668014471164, "grad_norm": 13.2890625, "learning_rate": 8.869172825469016e-06, "loss": 0.5268, "step": 32326 }, { "epoch": 3.439774420089381, "grad_norm": 15.84375, "learning_rate": 8.869137293917e-06, "loss": 0.823, "step": 32327 }, { "epoch": 3.4398808257075975, "grad_norm": 12.0078125, "learning_rate": 8.869101762364981e-06, "loss": 0.5067, "step": 32328 }, { "epoch": 3.439987231325814, "grad_norm": 16.0625, "learning_rate": 8.869066230812963e-06, "loss": 0.7103, "step": 32329 }, { "epoch": 3.4400936369440305, "grad_norm": 14.1484375, "learning_rate": 8.869030699260944e-06, "loss": 0.6623, "step": 32330 }, { "epoch": 3.4402000425622474, "grad_norm": 13.4765625, "learning_rate": 8.868995167708926e-06, "loss": 0.729, "step": 32331 }, { "epoch": 3.440306448180464, "grad_norm": 12.5625, "learning_rate": 8.86895963615691e-06, "loss": 0.6148, "step": 32332 }, { "epoch": 3.4404128537986804, "grad_norm": 12.5390625, "learning_rate": 8.868924104604891e-06, "loss": 0.7484, "step": 32333 }, { "epoch": 3.4405192594168974, "grad_norm": 16.09375, "learning_rate": 8.868888573052872e-06, "loss": 0.7027, "step": 32334 }, { "epoch": 3.440625665035114, "grad_norm": 12.4375, "learning_rate": 8.868853041500854e-06, "loss": 0.7045, "step": 32335 }, { "epoch": 3.4407320706533304, "grad_norm": 15.6171875, "learning_rate": 8.868817509948836e-06, "loss": 0.7119, "step": 32336 }, { "epoch": 3.440838476271547, "grad_norm": 13.4375, "learning_rate": 8.868781978396817e-06, "loss": 0.8706, "step": 32337 }, { "epoch": 3.440944881889764, "grad_norm": 12.65625, "learning_rate": 8.868746446844799e-06, "loss": 0.5808, "step": 32338 }, { "epoch": 3.4410512875079804, "grad_norm": 12.921875, "learning_rate": 8.86871091529278e-06, "loss": 0.4999, "step": 32339 }, { "epoch": 3.441157693126197, "grad_norm": 12.6328125, "learning_rate": 8.868675383740762e-06, "loss": 0.7189, "step": 32340 }, { "epoch": 3.441264098744414, "grad_norm": 12.765625, "learning_rate": 8.868639852188744e-06, "loss": 0.588, "step": 32341 }, { "epoch": 3.4413705043626304, "grad_norm": 12.7578125, "learning_rate": 8.868604320636725e-06, "loss": 0.8481, "step": 32342 }, { "epoch": 3.441476909980847, "grad_norm": 18.703125, "learning_rate": 8.868568789084709e-06, "loss": 0.7744, "step": 32343 }, { "epoch": 3.441583315599064, "grad_norm": 16.84375, "learning_rate": 8.86853325753269e-06, "loss": 0.7462, "step": 32344 }, { "epoch": 3.4416897212172803, "grad_norm": 11.375, "learning_rate": 8.868497725980672e-06, "loss": 0.6651, "step": 32345 }, { "epoch": 3.441796126835497, "grad_norm": 15.2421875, "learning_rate": 8.868462194428653e-06, "loss": 0.7212, "step": 32346 }, { "epoch": 3.441902532453714, "grad_norm": 14.1640625, "learning_rate": 8.868426662876635e-06, "loss": 0.6617, "step": 32347 }, { "epoch": 3.4420089380719303, "grad_norm": 12.453125, "learning_rate": 8.868391131324617e-06, "loss": 0.6121, "step": 32348 }, { "epoch": 3.442115343690147, "grad_norm": 11.75, "learning_rate": 8.868355599772598e-06, "loss": 0.5112, "step": 32349 }, { "epoch": 3.4422217493083633, "grad_norm": 12.5859375, "learning_rate": 8.86832006822058e-06, "loss": 0.5365, "step": 32350 }, { "epoch": 3.4423281549265803, "grad_norm": 11.671875, "learning_rate": 8.868284536668562e-06, "loss": 0.4961, "step": 32351 }, { "epoch": 3.442434560544797, "grad_norm": 11.203125, "learning_rate": 8.868249005116543e-06, "loss": 0.5659, "step": 32352 }, { "epoch": 3.4425409661630133, "grad_norm": 12.7890625, "learning_rate": 8.868213473564526e-06, "loss": 0.6434, "step": 32353 }, { "epoch": 3.44264737178123, "grad_norm": 17.609375, "learning_rate": 8.868177942012508e-06, "loss": 0.6307, "step": 32354 }, { "epoch": 3.4427537773994468, "grad_norm": 14.265625, "learning_rate": 8.86814241046049e-06, "loss": 0.5784, "step": 32355 }, { "epoch": 3.4428601830176633, "grad_norm": 14.328125, "learning_rate": 8.868106878908471e-06, "loss": 0.7741, "step": 32356 }, { "epoch": 3.44296658863588, "grad_norm": 14.5859375, "learning_rate": 8.868071347356453e-06, "loss": 0.5795, "step": 32357 }, { "epoch": 3.4430729942540967, "grad_norm": 12.0234375, "learning_rate": 8.868035815804436e-06, "loss": 0.5427, "step": 32358 }, { "epoch": 3.4431793998723133, "grad_norm": 13.1953125, "learning_rate": 8.868000284252418e-06, "loss": 0.663, "step": 32359 }, { "epoch": 3.4432858054905298, "grad_norm": 14.453125, "learning_rate": 8.8679647527004e-06, "loss": 0.6328, "step": 32360 }, { "epoch": 3.4433922111087467, "grad_norm": 15.203125, "learning_rate": 8.867929221148381e-06, "loss": 0.6262, "step": 32361 }, { "epoch": 3.4434986167269632, "grad_norm": 11.6171875, "learning_rate": 8.867893689596363e-06, "loss": 0.5451, "step": 32362 }, { "epoch": 3.4436050223451797, "grad_norm": 19.015625, "learning_rate": 8.867858158044344e-06, "loss": 0.8611, "step": 32363 }, { "epoch": 3.4437114279633967, "grad_norm": 12.1015625, "learning_rate": 8.867822626492326e-06, "loss": 0.6285, "step": 32364 }, { "epoch": 3.443817833581613, "grad_norm": 17.6875, "learning_rate": 8.867787094940307e-06, "loss": 0.6483, "step": 32365 }, { "epoch": 3.4439242391998297, "grad_norm": 12.515625, "learning_rate": 8.867751563388289e-06, "loss": 0.5886, "step": 32366 }, { "epoch": 3.444030644818046, "grad_norm": 11.6015625, "learning_rate": 8.86771603183627e-06, "loss": 0.6455, "step": 32367 }, { "epoch": 3.444137050436263, "grad_norm": 12.2421875, "learning_rate": 8.867680500284254e-06, "loss": 0.5921, "step": 32368 }, { "epoch": 3.4442434560544797, "grad_norm": 9.3203125, "learning_rate": 8.867644968732236e-06, "loss": 0.6243, "step": 32369 }, { "epoch": 3.444349861672696, "grad_norm": 14.0078125, "learning_rate": 8.867609437180217e-06, "loss": 0.7215, "step": 32370 }, { "epoch": 3.444456267290913, "grad_norm": 12.734375, "learning_rate": 8.867573905628199e-06, "loss": 0.5973, "step": 32371 }, { "epoch": 3.4445626729091297, "grad_norm": 17.84375, "learning_rate": 8.86753837407618e-06, "loss": 0.7196, "step": 32372 }, { "epoch": 3.444669078527346, "grad_norm": 14.2734375, "learning_rate": 8.867502842524162e-06, "loss": 0.6827, "step": 32373 }, { "epoch": 3.4447754841455627, "grad_norm": 10.40625, "learning_rate": 8.867467310972144e-06, "loss": 0.638, "step": 32374 }, { "epoch": 3.4448818897637796, "grad_norm": 13.25, "learning_rate": 8.867431779420125e-06, "loss": 0.6836, "step": 32375 }, { "epoch": 3.444988295381996, "grad_norm": 12.7265625, "learning_rate": 8.867396247868107e-06, "loss": 0.66, "step": 32376 }, { "epoch": 3.4450947010002126, "grad_norm": 10.8515625, "learning_rate": 8.867360716316088e-06, "loss": 0.5423, "step": 32377 }, { "epoch": 3.4452011066184296, "grad_norm": 15.9296875, "learning_rate": 8.86732518476407e-06, "loss": 0.7211, "step": 32378 }, { "epoch": 3.445307512236646, "grad_norm": 13.46875, "learning_rate": 8.867289653212053e-06, "loss": 0.6883, "step": 32379 }, { "epoch": 3.4454139178548626, "grad_norm": 16.671875, "learning_rate": 8.867254121660035e-06, "loss": 0.7429, "step": 32380 }, { "epoch": 3.4455203234730796, "grad_norm": 13.1015625, "learning_rate": 8.867218590108016e-06, "loss": 0.6641, "step": 32381 }, { "epoch": 3.445626729091296, "grad_norm": 8.515625, "learning_rate": 8.867183058555998e-06, "loss": 0.7119, "step": 32382 }, { "epoch": 3.4457331347095126, "grad_norm": 12.2265625, "learning_rate": 8.86714752700398e-06, "loss": 0.6078, "step": 32383 }, { "epoch": 3.4458395403277295, "grad_norm": 16.0625, "learning_rate": 8.867111995451963e-06, "loss": 0.6611, "step": 32384 }, { "epoch": 3.445945945945946, "grad_norm": 14.125, "learning_rate": 8.867076463899945e-06, "loss": 0.6383, "step": 32385 }, { "epoch": 3.4460523515641626, "grad_norm": 14.203125, "learning_rate": 8.867040932347926e-06, "loss": 0.6855, "step": 32386 }, { "epoch": 3.446158757182379, "grad_norm": 13.125, "learning_rate": 8.867005400795908e-06, "loss": 0.5879, "step": 32387 }, { "epoch": 3.446265162800596, "grad_norm": 13.7109375, "learning_rate": 8.86696986924389e-06, "loss": 0.5539, "step": 32388 }, { "epoch": 3.4463715684188125, "grad_norm": 12.09375, "learning_rate": 8.866934337691871e-06, "loss": 0.5273, "step": 32389 }, { "epoch": 3.446477974037029, "grad_norm": 8.15625, "learning_rate": 8.866898806139853e-06, "loss": 0.5225, "step": 32390 }, { "epoch": 3.4465843796552456, "grad_norm": 13.546875, "learning_rate": 8.866863274587834e-06, "loss": 0.7985, "step": 32391 }, { "epoch": 3.4466907852734625, "grad_norm": 13.5234375, "learning_rate": 8.866827743035816e-06, "loss": 0.5816, "step": 32392 }, { "epoch": 3.446797190891679, "grad_norm": 10.453125, "learning_rate": 8.866792211483797e-06, "loss": 0.48, "step": 32393 }, { "epoch": 3.4469035965098955, "grad_norm": 15.3125, "learning_rate": 8.86675667993178e-06, "loss": 0.6263, "step": 32394 }, { "epoch": 3.4470100021281125, "grad_norm": 12.8359375, "learning_rate": 8.866721148379762e-06, "loss": 0.6394, "step": 32395 }, { "epoch": 3.447116407746329, "grad_norm": 11.8046875, "learning_rate": 8.866685616827744e-06, "loss": 0.6381, "step": 32396 }, { "epoch": 3.4472228133645455, "grad_norm": 13.796875, "learning_rate": 8.866650085275726e-06, "loss": 0.7904, "step": 32397 }, { "epoch": 3.4473292189827625, "grad_norm": 15.84375, "learning_rate": 8.866614553723707e-06, "loss": 0.8001, "step": 32398 }, { "epoch": 3.447435624600979, "grad_norm": 13.0390625, "learning_rate": 8.86657902217169e-06, "loss": 0.6535, "step": 32399 }, { "epoch": 3.4475420302191955, "grad_norm": 11.0859375, "learning_rate": 8.86654349061967e-06, "loss": 0.5446, "step": 32400 }, { "epoch": 3.4476484358374124, "grad_norm": 10.296875, "learning_rate": 8.866507959067652e-06, "loss": 0.5941, "step": 32401 }, { "epoch": 3.447754841455629, "grad_norm": 12.5390625, "learning_rate": 8.866472427515634e-06, "loss": 0.6149, "step": 32402 }, { "epoch": 3.4478612470738454, "grad_norm": 13.4296875, "learning_rate": 8.866436895963615e-06, "loss": 0.6742, "step": 32403 }, { "epoch": 3.447967652692062, "grad_norm": 13.28125, "learning_rate": 8.866401364411599e-06, "loss": 0.7452, "step": 32404 }, { "epoch": 3.448074058310279, "grad_norm": 10.8125, "learning_rate": 8.86636583285958e-06, "loss": 0.626, "step": 32405 }, { "epoch": 3.4481804639284954, "grad_norm": 13.9765625, "learning_rate": 8.866330301307562e-06, "loss": 0.6358, "step": 32406 }, { "epoch": 3.448286869546712, "grad_norm": 11.96875, "learning_rate": 8.866294769755543e-06, "loss": 0.6477, "step": 32407 }, { "epoch": 3.448393275164929, "grad_norm": 11.578125, "learning_rate": 8.866259238203525e-06, "loss": 0.5213, "step": 32408 }, { "epoch": 3.4484996807831454, "grad_norm": 12.0625, "learning_rate": 8.866223706651508e-06, "loss": 0.6081, "step": 32409 }, { "epoch": 3.448606086401362, "grad_norm": 14.25, "learning_rate": 8.86618817509949e-06, "loss": 0.5782, "step": 32410 }, { "epoch": 3.4487124920195784, "grad_norm": 12.2421875, "learning_rate": 8.866152643547471e-06, "loss": 0.5437, "step": 32411 }, { "epoch": 3.4488188976377954, "grad_norm": 15.3046875, "learning_rate": 8.866117111995453e-06, "loss": 0.6754, "step": 32412 }, { "epoch": 3.448925303256012, "grad_norm": 11.1640625, "learning_rate": 8.866081580443435e-06, "loss": 0.6454, "step": 32413 }, { "epoch": 3.4490317088742284, "grad_norm": 14.4140625, "learning_rate": 8.866046048891416e-06, "loss": 0.6149, "step": 32414 }, { "epoch": 3.4491381144924453, "grad_norm": 9.5234375, "learning_rate": 8.866010517339398e-06, "loss": 0.4786, "step": 32415 }, { "epoch": 3.449244520110662, "grad_norm": 9.59375, "learning_rate": 8.86597498578738e-06, "loss": 0.4804, "step": 32416 }, { "epoch": 3.4493509257288784, "grad_norm": 11.9375, "learning_rate": 8.865939454235361e-06, "loss": 0.6804, "step": 32417 }, { "epoch": 3.4494573313470953, "grad_norm": 13.5, "learning_rate": 8.865903922683343e-06, "loss": 0.7448, "step": 32418 }, { "epoch": 3.449563736965312, "grad_norm": 15.390625, "learning_rate": 8.865868391131324e-06, "loss": 0.8061, "step": 32419 }, { "epoch": 3.4496701425835283, "grad_norm": 11.4296875, "learning_rate": 8.865832859579308e-06, "loss": 0.5556, "step": 32420 }, { "epoch": 3.4497765482017453, "grad_norm": 13.4453125, "learning_rate": 8.86579732802729e-06, "loss": 0.6461, "step": 32421 }, { "epoch": 3.449882953819962, "grad_norm": 14.765625, "learning_rate": 8.86576179647527e-06, "loss": 0.6376, "step": 32422 }, { "epoch": 3.4499893594381783, "grad_norm": 13.84375, "learning_rate": 8.865726264923252e-06, "loss": 0.6755, "step": 32423 }, { "epoch": 3.450095765056395, "grad_norm": 13.15625, "learning_rate": 8.865690733371234e-06, "loss": 0.6778, "step": 32424 }, { "epoch": 3.4502021706746118, "grad_norm": 9.6953125, "learning_rate": 8.865655201819217e-06, "loss": 0.515, "step": 32425 }, { "epoch": 3.4503085762928283, "grad_norm": 12.5625, "learning_rate": 8.865619670267199e-06, "loss": 0.6433, "step": 32426 }, { "epoch": 3.450414981911045, "grad_norm": 13.03125, "learning_rate": 8.86558413871518e-06, "loss": 0.6066, "step": 32427 }, { "epoch": 3.4505213875292613, "grad_norm": 11.9765625, "learning_rate": 8.865548607163162e-06, "loss": 0.5617, "step": 32428 }, { "epoch": 3.4506277931474783, "grad_norm": 17.359375, "learning_rate": 8.865513075611144e-06, "loss": 0.7081, "step": 32429 }, { "epoch": 3.4507341987656948, "grad_norm": 14.09375, "learning_rate": 8.865477544059125e-06, "loss": 0.7023, "step": 32430 }, { "epoch": 3.4508406043839113, "grad_norm": 14.4140625, "learning_rate": 8.865442012507107e-06, "loss": 0.5439, "step": 32431 }, { "epoch": 3.4509470100021282, "grad_norm": 10.1171875, "learning_rate": 8.865406480955089e-06, "loss": 0.4616, "step": 32432 }, { "epoch": 3.4510534156203447, "grad_norm": 12.6796875, "learning_rate": 8.86537094940307e-06, "loss": 0.6487, "step": 32433 }, { "epoch": 3.4511598212385612, "grad_norm": 12.6171875, "learning_rate": 8.865335417851052e-06, "loss": 0.6679, "step": 32434 }, { "epoch": 3.451266226856778, "grad_norm": 11.4296875, "learning_rate": 8.865299886299035e-06, "loss": 0.5982, "step": 32435 }, { "epoch": 3.4513726324749947, "grad_norm": 16.265625, "learning_rate": 8.865264354747017e-06, "loss": 0.7957, "step": 32436 }, { "epoch": 3.451479038093211, "grad_norm": 16.8125, "learning_rate": 8.865228823194998e-06, "loss": 0.7513, "step": 32437 }, { "epoch": 3.451585443711428, "grad_norm": 12.2109375, "learning_rate": 8.86519329164298e-06, "loss": 0.5015, "step": 32438 }, { "epoch": 3.4516918493296447, "grad_norm": 13.0390625, "learning_rate": 8.865157760090962e-06, "loss": 0.6534, "step": 32439 }, { "epoch": 3.451798254947861, "grad_norm": 12.96875, "learning_rate": 8.865122228538943e-06, "loss": 0.7242, "step": 32440 }, { "epoch": 3.4519046605660777, "grad_norm": 10.453125, "learning_rate": 8.865086696986925e-06, "loss": 0.5787, "step": 32441 }, { "epoch": 3.4520110661842947, "grad_norm": 12.140625, "learning_rate": 8.865051165434906e-06, "loss": 0.6989, "step": 32442 }, { "epoch": 3.452117471802511, "grad_norm": 12.0078125, "learning_rate": 8.865015633882888e-06, "loss": 0.6416, "step": 32443 }, { "epoch": 3.4522238774207277, "grad_norm": 12.734375, "learning_rate": 8.86498010233087e-06, "loss": 0.6329, "step": 32444 }, { "epoch": 3.4523302830389446, "grad_norm": 10.6953125, "learning_rate": 8.864944570778853e-06, "loss": 0.4634, "step": 32445 }, { "epoch": 3.452436688657161, "grad_norm": 12.265625, "learning_rate": 8.864909039226834e-06, "loss": 0.5393, "step": 32446 }, { "epoch": 3.4525430942753776, "grad_norm": 17.234375, "learning_rate": 8.864873507674816e-06, "loss": 0.7099, "step": 32447 }, { "epoch": 3.452649499893594, "grad_norm": 14.90625, "learning_rate": 8.864837976122798e-06, "loss": 0.6999, "step": 32448 }, { "epoch": 3.452755905511811, "grad_norm": 14.0234375, "learning_rate": 8.86480244457078e-06, "loss": 0.6602, "step": 32449 }, { "epoch": 3.4528623111300276, "grad_norm": 12.1328125, "learning_rate": 8.864766913018763e-06, "loss": 0.549, "step": 32450 }, { "epoch": 3.452968716748244, "grad_norm": 13.8671875, "learning_rate": 8.864731381466744e-06, "loss": 0.758, "step": 32451 }, { "epoch": 3.453075122366461, "grad_norm": 10.4921875, "learning_rate": 8.864695849914726e-06, "loss": 0.5493, "step": 32452 }, { "epoch": 3.4531815279846776, "grad_norm": 11.1640625, "learning_rate": 8.864660318362707e-06, "loss": 0.591, "step": 32453 }, { "epoch": 3.453287933602894, "grad_norm": 10.875, "learning_rate": 8.864624786810689e-06, "loss": 0.5067, "step": 32454 }, { "epoch": 3.453394339221111, "grad_norm": 15.25, "learning_rate": 8.86458925525867e-06, "loss": 0.6218, "step": 32455 }, { "epoch": 3.4535007448393276, "grad_norm": 8.5234375, "learning_rate": 8.864553723706652e-06, "loss": 0.5098, "step": 32456 }, { "epoch": 3.453607150457544, "grad_norm": 13.9296875, "learning_rate": 8.864518192154634e-06, "loss": 0.6941, "step": 32457 }, { "epoch": 3.453713556075761, "grad_norm": 13.125, "learning_rate": 8.864482660602615e-06, "loss": 0.6243, "step": 32458 }, { "epoch": 3.4538199616939775, "grad_norm": 15.7578125, "learning_rate": 8.864447129050597e-06, "loss": 0.7611, "step": 32459 }, { "epoch": 3.453926367312194, "grad_norm": 15.515625, "learning_rate": 8.864411597498579e-06, "loss": 0.7145, "step": 32460 }, { "epoch": 3.4540327729304106, "grad_norm": 15.1875, "learning_rate": 8.864376065946562e-06, "loss": 0.6875, "step": 32461 }, { "epoch": 3.4541391785486275, "grad_norm": 13.015625, "learning_rate": 8.864340534394544e-06, "loss": 0.6558, "step": 32462 }, { "epoch": 3.454245584166844, "grad_norm": 14.1875, "learning_rate": 8.864305002842525e-06, "loss": 0.6399, "step": 32463 }, { "epoch": 3.4543519897850605, "grad_norm": 13.015625, "learning_rate": 8.864269471290507e-06, "loss": 0.7642, "step": 32464 }, { "epoch": 3.4544583954032775, "grad_norm": 17.078125, "learning_rate": 8.864233939738488e-06, "loss": 0.6843, "step": 32465 }, { "epoch": 3.454564801021494, "grad_norm": 10.140625, "learning_rate": 8.86419840818647e-06, "loss": 0.5765, "step": 32466 }, { "epoch": 3.4546712066397105, "grad_norm": 12.5078125, "learning_rate": 8.864162876634452e-06, "loss": 0.8904, "step": 32467 }, { "epoch": 3.454777612257927, "grad_norm": 14.703125, "learning_rate": 8.864127345082433e-06, "loss": 0.6283, "step": 32468 }, { "epoch": 3.454884017876144, "grad_norm": 16.046875, "learning_rate": 8.864091813530415e-06, "loss": 0.6729, "step": 32469 }, { "epoch": 3.4549904234943605, "grad_norm": 15.671875, "learning_rate": 8.864056281978396e-06, "loss": 0.7211, "step": 32470 }, { "epoch": 3.455096829112577, "grad_norm": 16.828125, "learning_rate": 8.86402075042638e-06, "loss": 0.6523, "step": 32471 }, { "epoch": 3.455203234730794, "grad_norm": 14.2265625, "learning_rate": 8.863985218874361e-06, "loss": 0.6624, "step": 32472 }, { "epoch": 3.4553096403490104, "grad_norm": 13.671875, "learning_rate": 8.863949687322343e-06, "loss": 0.7196, "step": 32473 }, { "epoch": 3.455416045967227, "grad_norm": 14.5546875, "learning_rate": 8.863914155770325e-06, "loss": 0.6593, "step": 32474 }, { "epoch": 3.455522451585444, "grad_norm": 11.9375, "learning_rate": 8.863878624218306e-06, "loss": 0.6151, "step": 32475 }, { "epoch": 3.4556288572036604, "grad_norm": 13.3515625, "learning_rate": 8.86384309266629e-06, "loss": 0.6775, "step": 32476 }, { "epoch": 3.455735262821877, "grad_norm": 8.8125, "learning_rate": 8.863807561114271e-06, "loss": 0.4709, "step": 32477 }, { "epoch": 3.455841668440094, "grad_norm": 14.953125, "learning_rate": 8.863772029562253e-06, "loss": 0.8659, "step": 32478 }, { "epoch": 3.4559480740583104, "grad_norm": 13.921875, "learning_rate": 8.863736498010234e-06, "loss": 0.643, "step": 32479 }, { "epoch": 3.456054479676527, "grad_norm": 10.8125, "learning_rate": 8.863700966458216e-06, "loss": 0.515, "step": 32480 }, { "epoch": 3.4561608852947434, "grad_norm": 11.3671875, "learning_rate": 8.863665434906197e-06, "loss": 0.4966, "step": 32481 }, { "epoch": 3.4562672909129604, "grad_norm": 14.8671875, "learning_rate": 8.863629903354179e-06, "loss": 0.6076, "step": 32482 }, { "epoch": 3.456373696531177, "grad_norm": 11.359375, "learning_rate": 8.86359437180216e-06, "loss": 0.6215, "step": 32483 }, { "epoch": 3.4564801021493934, "grad_norm": 11.90625, "learning_rate": 8.863558840250142e-06, "loss": 0.5659, "step": 32484 }, { "epoch": 3.45658650776761, "grad_norm": 19.75, "learning_rate": 8.863523308698124e-06, "loss": 0.8003, "step": 32485 }, { "epoch": 3.456692913385827, "grad_norm": 11.125, "learning_rate": 8.863487777146107e-06, "loss": 0.4888, "step": 32486 }, { "epoch": 3.4567993190040434, "grad_norm": 13.4375, "learning_rate": 8.863452245594089e-06, "loss": 0.7347, "step": 32487 }, { "epoch": 3.45690572462226, "grad_norm": 10.859375, "learning_rate": 8.86341671404207e-06, "loss": 0.6272, "step": 32488 }, { "epoch": 3.457012130240477, "grad_norm": 15.859375, "learning_rate": 8.863381182490052e-06, "loss": 0.698, "step": 32489 }, { "epoch": 3.4571185358586933, "grad_norm": 11.25, "learning_rate": 8.863345650938034e-06, "loss": 0.5354, "step": 32490 }, { "epoch": 3.45722494147691, "grad_norm": 15.21875, "learning_rate": 8.863310119386015e-06, "loss": 0.6615, "step": 32491 }, { "epoch": 3.457331347095127, "grad_norm": 16.390625, "learning_rate": 8.863274587833997e-06, "loss": 0.7635, "step": 32492 }, { "epoch": 3.4574377527133433, "grad_norm": 14.84375, "learning_rate": 8.863239056281978e-06, "loss": 0.6042, "step": 32493 }, { "epoch": 3.45754415833156, "grad_norm": 14.1328125, "learning_rate": 8.86320352472996e-06, "loss": 0.6266, "step": 32494 }, { "epoch": 3.4576505639497768, "grad_norm": 12.328125, "learning_rate": 8.863167993177942e-06, "loss": 0.5771, "step": 32495 }, { "epoch": 3.4577569695679933, "grad_norm": 11.3828125, "learning_rate": 8.863132461625923e-06, "loss": 0.5703, "step": 32496 }, { "epoch": 3.45786337518621, "grad_norm": 12.921875, "learning_rate": 8.863096930073907e-06, "loss": 0.6552, "step": 32497 }, { "epoch": 3.4579697808044263, "grad_norm": 12.7734375, "learning_rate": 8.863061398521888e-06, "loss": 0.7634, "step": 32498 }, { "epoch": 3.4580761864226433, "grad_norm": 15.9375, "learning_rate": 8.86302586696987e-06, "loss": 0.6244, "step": 32499 }, { "epoch": 3.4581825920408598, "grad_norm": 11.390625, "learning_rate": 8.862990335417851e-06, "loss": 0.6272, "step": 32500 }, { "epoch": 3.4582889976590763, "grad_norm": 9.875, "learning_rate": 8.862954803865833e-06, "loss": 0.5553, "step": 32501 }, { "epoch": 3.4583954032772932, "grad_norm": 8.9453125, "learning_rate": 8.862919272313816e-06, "loss": 0.533, "step": 32502 }, { "epoch": 3.4585018088955097, "grad_norm": 11.2578125, "learning_rate": 8.862883740761798e-06, "loss": 0.558, "step": 32503 }, { "epoch": 3.4586082145137262, "grad_norm": 15.84375, "learning_rate": 8.86284820920978e-06, "loss": 0.7801, "step": 32504 }, { "epoch": 3.4587146201319428, "grad_norm": 13.6640625, "learning_rate": 8.862812677657761e-06, "loss": 0.6106, "step": 32505 }, { "epoch": 3.4588210257501597, "grad_norm": 13.1171875, "learning_rate": 8.862777146105743e-06, "loss": 0.675, "step": 32506 }, { "epoch": 3.458927431368376, "grad_norm": 11.8671875, "learning_rate": 8.862741614553724e-06, "loss": 0.5661, "step": 32507 }, { "epoch": 3.4590338369865927, "grad_norm": 9.6484375, "learning_rate": 8.862706083001706e-06, "loss": 0.4319, "step": 32508 }, { "epoch": 3.4591402426048097, "grad_norm": 14.2578125, "learning_rate": 8.862670551449688e-06, "loss": 0.5776, "step": 32509 }, { "epoch": 3.459246648223026, "grad_norm": 10.5, "learning_rate": 8.862635019897669e-06, "loss": 0.5583, "step": 32510 }, { "epoch": 3.4593530538412427, "grad_norm": 12.7890625, "learning_rate": 8.86259948834565e-06, "loss": 0.5053, "step": 32511 }, { "epoch": 3.4594594594594597, "grad_norm": 12.015625, "learning_rate": 8.862563956793634e-06, "loss": 0.6999, "step": 32512 }, { "epoch": 3.459565865077676, "grad_norm": 13.71875, "learning_rate": 8.862528425241616e-06, "loss": 0.5738, "step": 32513 }, { "epoch": 3.4596722706958927, "grad_norm": 13.90625, "learning_rate": 8.862492893689597e-06, "loss": 0.4737, "step": 32514 }, { "epoch": 3.4597786763141096, "grad_norm": 10.765625, "learning_rate": 8.862457362137579e-06, "loss": 0.5308, "step": 32515 }, { "epoch": 3.459885081932326, "grad_norm": 10.203125, "learning_rate": 8.86242183058556e-06, "loss": 0.5514, "step": 32516 }, { "epoch": 3.4599914875505426, "grad_norm": 10.4375, "learning_rate": 8.862386299033544e-06, "loss": 0.5115, "step": 32517 }, { "epoch": 3.460097893168759, "grad_norm": 12.9453125, "learning_rate": 8.862350767481525e-06, "loss": 0.5964, "step": 32518 }, { "epoch": 3.460204298786976, "grad_norm": 14.3984375, "learning_rate": 8.862315235929507e-06, "loss": 0.8465, "step": 32519 }, { "epoch": 3.4603107044051926, "grad_norm": 12.96875, "learning_rate": 8.862279704377487e-06, "loss": 0.5906, "step": 32520 }, { "epoch": 3.460417110023409, "grad_norm": 13.625, "learning_rate": 8.862244172825469e-06, "loss": 0.6622, "step": 32521 }, { "epoch": 3.4605235156416256, "grad_norm": 11.4296875, "learning_rate": 8.862208641273452e-06, "loss": 0.5166, "step": 32522 }, { "epoch": 3.4606299212598426, "grad_norm": 14.265625, "learning_rate": 8.862173109721433e-06, "loss": 0.6979, "step": 32523 }, { "epoch": 3.460736326878059, "grad_norm": 12.578125, "learning_rate": 8.862137578169415e-06, "loss": 0.5786, "step": 32524 }, { "epoch": 3.4608427324962756, "grad_norm": 10.2578125, "learning_rate": 8.862102046617397e-06, "loss": 0.4654, "step": 32525 }, { "epoch": 3.4609491381144926, "grad_norm": 16.59375, "learning_rate": 8.862066515065378e-06, "loss": 0.7004, "step": 32526 }, { "epoch": 3.461055543732709, "grad_norm": 14.7265625, "learning_rate": 8.862030983513362e-06, "loss": 0.6556, "step": 32527 }, { "epoch": 3.4611619493509256, "grad_norm": 14.71875, "learning_rate": 8.861995451961343e-06, "loss": 0.7262, "step": 32528 }, { "epoch": 3.4612683549691425, "grad_norm": 16.59375, "learning_rate": 8.861959920409325e-06, "loss": 0.5808, "step": 32529 }, { "epoch": 3.461374760587359, "grad_norm": 13.5234375, "learning_rate": 8.861924388857306e-06, "loss": 0.6131, "step": 32530 }, { "epoch": 3.4614811662055756, "grad_norm": 15.390625, "learning_rate": 8.861888857305288e-06, "loss": 0.5991, "step": 32531 }, { "epoch": 3.4615875718237925, "grad_norm": 15.0078125, "learning_rate": 8.86185332575327e-06, "loss": 0.674, "step": 32532 }, { "epoch": 3.461693977442009, "grad_norm": 13.2890625, "learning_rate": 8.861817794201251e-06, "loss": 0.6569, "step": 32533 }, { "epoch": 3.4618003830602255, "grad_norm": 12.3046875, "learning_rate": 8.861782262649233e-06, "loss": 0.5456, "step": 32534 }, { "epoch": 3.461906788678442, "grad_norm": 15.5546875, "learning_rate": 8.861746731097214e-06, "loss": 0.8081, "step": 32535 }, { "epoch": 3.462013194296659, "grad_norm": 14.984375, "learning_rate": 8.861711199545196e-06, "loss": 0.5931, "step": 32536 }, { "epoch": 3.4621195999148755, "grad_norm": 23.9375, "learning_rate": 8.861675667993178e-06, "loss": 0.9922, "step": 32537 }, { "epoch": 3.462226005533092, "grad_norm": 13.84375, "learning_rate": 8.861640136441161e-06, "loss": 0.7253, "step": 32538 }, { "epoch": 3.462332411151309, "grad_norm": 12.859375, "learning_rate": 8.861604604889143e-06, "loss": 0.7158, "step": 32539 }, { "epoch": 3.4624388167695255, "grad_norm": 13.1015625, "learning_rate": 8.861569073337124e-06, "loss": 0.6756, "step": 32540 }, { "epoch": 3.462545222387742, "grad_norm": 11.6953125, "learning_rate": 8.861533541785106e-06, "loss": 0.6327, "step": 32541 }, { "epoch": 3.4626516280059585, "grad_norm": 13.6953125, "learning_rate": 8.861498010233087e-06, "loss": 0.656, "step": 32542 }, { "epoch": 3.4627580336241754, "grad_norm": 13.6328125, "learning_rate": 8.86146247868107e-06, "loss": 0.6176, "step": 32543 }, { "epoch": 3.462864439242392, "grad_norm": 12.859375, "learning_rate": 8.861426947129052e-06, "loss": 0.7137, "step": 32544 }, { "epoch": 3.4629708448606085, "grad_norm": 12.125, "learning_rate": 8.861391415577034e-06, "loss": 0.6489, "step": 32545 }, { "epoch": 3.4630772504788254, "grad_norm": 13.375, "learning_rate": 8.861355884025015e-06, "loss": 0.5291, "step": 32546 }, { "epoch": 3.463183656097042, "grad_norm": 14.25, "learning_rate": 8.861320352472997e-06, "loss": 0.7043, "step": 32547 }, { "epoch": 3.4632900617152584, "grad_norm": 11.3125, "learning_rate": 8.861284820920979e-06, "loss": 0.6536, "step": 32548 }, { "epoch": 3.4633964673334754, "grad_norm": 19.953125, "learning_rate": 8.86124928936896e-06, "loss": 0.7361, "step": 32549 }, { "epoch": 3.463502872951692, "grad_norm": 14.8515625, "learning_rate": 8.861213757816942e-06, "loss": 0.7554, "step": 32550 }, { "epoch": 3.4636092785699084, "grad_norm": 11.9296875, "learning_rate": 8.861178226264923e-06, "loss": 0.6486, "step": 32551 }, { "epoch": 3.4637156841881254, "grad_norm": 12.3984375, "learning_rate": 8.861142694712905e-06, "loss": 0.6157, "step": 32552 }, { "epoch": 3.463822089806342, "grad_norm": 14.625, "learning_rate": 8.861107163160888e-06, "loss": 0.7573, "step": 32553 }, { "epoch": 3.4639284954245584, "grad_norm": 15.328125, "learning_rate": 8.86107163160887e-06, "loss": 0.7405, "step": 32554 }, { "epoch": 3.464034901042775, "grad_norm": 18.109375, "learning_rate": 8.861036100056852e-06, "loss": 0.7163, "step": 32555 }, { "epoch": 3.464141306660992, "grad_norm": 11.5703125, "learning_rate": 8.861000568504833e-06, "loss": 0.6221, "step": 32556 }, { "epoch": 3.4642477122792084, "grad_norm": 12.625, "learning_rate": 8.860965036952815e-06, "loss": 0.6823, "step": 32557 }, { "epoch": 3.464354117897425, "grad_norm": 13.4453125, "learning_rate": 8.860929505400796e-06, "loss": 0.5815, "step": 32558 }, { "epoch": 3.4644605235156414, "grad_norm": 13.453125, "learning_rate": 8.860893973848778e-06, "loss": 0.5945, "step": 32559 }, { "epoch": 3.4645669291338583, "grad_norm": 17.03125, "learning_rate": 8.86085844229676e-06, "loss": 0.7278, "step": 32560 }, { "epoch": 3.464673334752075, "grad_norm": 13.5, "learning_rate": 8.860822910744741e-06, "loss": 0.6407, "step": 32561 }, { "epoch": 3.4647797403702913, "grad_norm": 11.1796875, "learning_rate": 8.860787379192723e-06, "loss": 0.5733, "step": 32562 }, { "epoch": 3.4648861459885083, "grad_norm": 13.7734375, "learning_rate": 8.860751847640706e-06, "loss": 0.5054, "step": 32563 }, { "epoch": 3.464992551606725, "grad_norm": 18.40625, "learning_rate": 8.860716316088688e-06, "loss": 0.8058, "step": 32564 }, { "epoch": 3.4650989572249413, "grad_norm": 18.015625, "learning_rate": 8.86068078453667e-06, "loss": 0.801, "step": 32565 }, { "epoch": 3.4652053628431583, "grad_norm": 10.6875, "learning_rate": 8.860645252984651e-06, "loss": 0.6157, "step": 32566 }, { "epoch": 3.465311768461375, "grad_norm": 13.515625, "learning_rate": 8.860609721432633e-06, "loss": 0.6844, "step": 32567 }, { "epoch": 3.4654181740795913, "grad_norm": 14.453125, "learning_rate": 8.860574189880616e-06, "loss": 0.7067, "step": 32568 }, { "epoch": 3.4655245796978082, "grad_norm": 12.4765625, "learning_rate": 8.860538658328597e-06, "loss": 0.7109, "step": 32569 }, { "epoch": 3.4656309853160248, "grad_norm": 12.75, "learning_rate": 8.860503126776579e-06, "loss": 0.3732, "step": 32570 }, { "epoch": 3.4657373909342413, "grad_norm": 14.7109375, "learning_rate": 8.86046759522456e-06, "loss": 0.6744, "step": 32571 }, { "epoch": 3.465843796552458, "grad_norm": 14.3125, "learning_rate": 8.860432063672542e-06, "loss": 0.8311, "step": 32572 }, { "epoch": 3.4659502021706747, "grad_norm": 15.6484375, "learning_rate": 8.860396532120524e-06, "loss": 0.6952, "step": 32573 }, { "epoch": 3.4660566077888912, "grad_norm": 14.8046875, "learning_rate": 8.860361000568506e-06, "loss": 0.7173, "step": 32574 }, { "epoch": 3.4661630134071078, "grad_norm": 10.578125, "learning_rate": 8.860325469016487e-06, "loss": 0.5224, "step": 32575 }, { "epoch": 3.4662694190253247, "grad_norm": 15.421875, "learning_rate": 8.860289937464469e-06, "loss": 0.6924, "step": 32576 }, { "epoch": 3.466375824643541, "grad_norm": 13.4375, "learning_rate": 8.86025440591245e-06, "loss": 0.6582, "step": 32577 }, { "epoch": 3.4664822302617577, "grad_norm": 15.6484375, "learning_rate": 8.860218874360432e-06, "loss": 0.7773, "step": 32578 }, { "epoch": 3.4665886358799742, "grad_norm": 14.0703125, "learning_rate": 8.860183342808415e-06, "loss": 0.6517, "step": 32579 }, { "epoch": 3.466695041498191, "grad_norm": 12.6953125, "learning_rate": 8.860147811256397e-06, "loss": 0.5944, "step": 32580 }, { "epoch": 3.4668014471164077, "grad_norm": 15.0078125, "learning_rate": 8.860112279704378e-06, "loss": 0.678, "step": 32581 }, { "epoch": 3.466907852734624, "grad_norm": 17.09375, "learning_rate": 8.86007674815236e-06, "loss": 0.7741, "step": 32582 }, { "epoch": 3.467014258352841, "grad_norm": 10.0234375, "learning_rate": 8.860041216600342e-06, "loss": 0.6133, "step": 32583 }, { "epoch": 3.4671206639710577, "grad_norm": 18.234375, "learning_rate": 8.860005685048323e-06, "loss": 0.8963, "step": 32584 }, { "epoch": 3.467227069589274, "grad_norm": 11.96875, "learning_rate": 8.859970153496305e-06, "loss": 0.6294, "step": 32585 }, { "epoch": 3.467333475207491, "grad_norm": 11.15625, "learning_rate": 8.859934621944287e-06, "loss": 0.6527, "step": 32586 }, { "epoch": 3.4674398808257076, "grad_norm": 15.859375, "learning_rate": 8.859899090392268e-06, "loss": 0.7486, "step": 32587 }, { "epoch": 3.467546286443924, "grad_norm": 10.3828125, "learning_rate": 8.85986355884025e-06, "loss": 0.5599, "step": 32588 }, { "epoch": 3.467652692062141, "grad_norm": 11.9765625, "learning_rate": 8.859828027288233e-06, "loss": 0.6167, "step": 32589 }, { "epoch": 3.4677590976803576, "grad_norm": 10.828125, "learning_rate": 8.859792495736215e-06, "loss": 0.6076, "step": 32590 }, { "epoch": 3.467865503298574, "grad_norm": 13.375, "learning_rate": 8.859756964184196e-06, "loss": 0.6583, "step": 32591 }, { "epoch": 3.4679719089167906, "grad_norm": 13.609375, "learning_rate": 8.859721432632178e-06, "loss": 0.6833, "step": 32592 }, { "epoch": 3.4680783145350076, "grad_norm": 11.6953125, "learning_rate": 8.85968590108016e-06, "loss": 0.6632, "step": 32593 }, { "epoch": 3.468184720153224, "grad_norm": 14.0703125, "learning_rate": 8.859650369528143e-06, "loss": 0.6398, "step": 32594 }, { "epoch": 3.4682911257714406, "grad_norm": 13.75, "learning_rate": 8.859614837976124e-06, "loss": 0.6199, "step": 32595 }, { "epoch": 3.4683975313896576, "grad_norm": 9.65625, "learning_rate": 8.859579306424106e-06, "loss": 0.5343, "step": 32596 }, { "epoch": 3.468503937007874, "grad_norm": 11.5078125, "learning_rate": 8.859543774872088e-06, "loss": 0.6523, "step": 32597 }, { "epoch": 3.4686103426260906, "grad_norm": 13.6484375, "learning_rate": 8.85950824332007e-06, "loss": 0.613, "step": 32598 }, { "epoch": 3.468716748244307, "grad_norm": 13.84375, "learning_rate": 8.85947271176805e-06, "loss": 0.6312, "step": 32599 }, { "epoch": 3.468823153862524, "grad_norm": 10.015625, "learning_rate": 8.859437180216032e-06, "loss": 0.5755, "step": 32600 }, { "epoch": 3.4689295594807406, "grad_norm": 14.140625, "learning_rate": 8.859401648664014e-06, "loss": 0.5817, "step": 32601 }, { "epoch": 3.469035965098957, "grad_norm": 14.34375, "learning_rate": 8.859366117111996e-06, "loss": 0.8015, "step": 32602 }, { "epoch": 3.469142370717174, "grad_norm": 12.1953125, "learning_rate": 8.859330585559977e-06, "loss": 0.5907, "step": 32603 }, { "epoch": 3.4692487763353905, "grad_norm": 16.96875, "learning_rate": 8.85929505400796e-06, "loss": 0.7752, "step": 32604 }, { "epoch": 3.469355181953607, "grad_norm": 16.09375, "learning_rate": 8.859259522455942e-06, "loss": 0.7873, "step": 32605 }, { "epoch": 3.469461587571824, "grad_norm": 15.0390625, "learning_rate": 8.859223990903924e-06, "loss": 0.6924, "step": 32606 }, { "epoch": 3.4695679931900405, "grad_norm": 9.6171875, "learning_rate": 8.859188459351905e-06, "loss": 0.5566, "step": 32607 }, { "epoch": 3.469674398808257, "grad_norm": 11.15625, "learning_rate": 8.859152927799887e-06, "loss": 0.6423, "step": 32608 }, { "epoch": 3.469780804426474, "grad_norm": 12.5859375, "learning_rate": 8.859117396247869e-06, "loss": 0.6295, "step": 32609 }, { "epoch": 3.4698872100446905, "grad_norm": 12.0546875, "learning_rate": 8.85908186469585e-06, "loss": 0.5814, "step": 32610 }, { "epoch": 3.469993615662907, "grad_norm": 11.046875, "learning_rate": 8.859046333143832e-06, "loss": 0.4983, "step": 32611 }, { "epoch": 3.4701000212811235, "grad_norm": 15.0078125, "learning_rate": 8.859010801591813e-06, "loss": 0.7206, "step": 32612 }, { "epoch": 3.4702064268993404, "grad_norm": 12.734375, "learning_rate": 8.858975270039795e-06, "loss": 0.5828, "step": 32613 }, { "epoch": 3.470312832517557, "grad_norm": 15.171875, "learning_rate": 8.858939738487777e-06, "loss": 0.7529, "step": 32614 }, { "epoch": 3.4704192381357735, "grad_norm": 18.8125, "learning_rate": 8.85890420693576e-06, "loss": 0.8375, "step": 32615 }, { "epoch": 3.47052564375399, "grad_norm": 13.9921875, "learning_rate": 8.858868675383741e-06, "loss": 0.6336, "step": 32616 }, { "epoch": 3.470632049372207, "grad_norm": 13.0859375, "learning_rate": 8.858833143831723e-06, "loss": 0.5775, "step": 32617 }, { "epoch": 3.4707384549904234, "grad_norm": 14.4921875, "learning_rate": 8.858797612279705e-06, "loss": 0.6374, "step": 32618 }, { "epoch": 3.47084486060864, "grad_norm": 13.25, "learning_rate": 8.858762080727686e-06, "loss": 0.6647, "step": 32619 }, { "epoch": 3.470951266226857, "grad_norm": 12.234375, "learning_rate": 8.85872654917567e-06, "loss": 0.6665, "step": 32620 }, { "epoch": 3.4710576718450734, "grad_norm": 14.5390625, "learning_rate": 8.858691017623651e-06, "loss": 0.7516, "step": 32621 }, { "epoch": 3.47116407746329, "grad_norm": 14.3828125, "learning_rate": 8.858655486071633e-06, "loss": 0.697, "step": 32622 }, { "epoch": 3.471270483081507, "grad_norm": 11.4609375, "learning_rate": 8.858619954519614e-06, "loss": 0.5957, "step": 32623 }, { "epoch": 3.4713768886997234, "grad_norm": 12.546875, "learning_rate": 8.858584422967596e-06, "loss": 0.5997, "step": 32624 }, { "epoch": 3.47148329431794, "grad_norm": 13.71875, "learning_rate": 8.858548891415578e-06, "loss": 0.7125, "step": 32625 }, { "epoch": 3.471589699936157, "grad_norm": 13.4375, "learning_rate": 8.85851335986356e-06, "loss": 0.6003, "step": 32626 }, { "epoch": 3.4716961055543734, "grad_norm": 11.9375, "learning_rate": 8.858477828311541e-06, "loss": 0.4959, "step": 32627 }, { "epoch": 3.47180251117259, "grad_norm": 16.953125, "learning_rate": 8.858442296759522e-06, "loss": 0.7031, "step": 32628 }, { "epoch": 3.4719089167908064, "grad_norm": 12.6796875, "learning_rate": 8.858406765207504e-06, "loss": 0.5112, "step": 32629 }, { "epoch": 3.4720153224090233, "grad_norm": 12.7578125, "learning_rate": 8.858371233655487e-06, "loss": 0.6794, "step": 32630 }, { "epoch": 3.47212172802724, "grad_norm": 15.1015625, "learning_rate": 8.858335702103469e-06, "loss": 0.6944, "step": 32631 }, { "epoch": 3.4722281336454563, "grad_norm": 12.21875, "learning_rate": 8.85830017055145e-06, "loss": 0.5796, "step": 32632 }, { "epoch": 3.4723345392636733, "grad_norm": 12.7734375, "learning_rate": 8.858264638999432e-06, "loss": 0.6402, "step": 32633 }, { "epoch": 3.47244094488189, "grad_norm": 14.9453125, "learning_rate": 8.858229107447414e-06, "loss": 0.6979, "step": 32634 }, { "epoch": 3.4725473505001063, "grad_norm": 16.328125, "learning_rate": 8.858193575895397e-06, "loss": 0.6315, "step": 32635 }, { "epoch": 3.472653756118323, "grad_norm": 19.171875, "learning_rate": 8.858158044343379e-06, "loss": 0.7855, "step": 32636 }, { "epoch": 3.47276016173654, "grad_norm": 9.4453125, "learning_rate": 8.85812251279136e-06, "loss": 0.5647, "step": 32637 }, { "epoch": 3.4728665673547563, "grad_norm": 13.0703125, "learning_rate": 8.858086981239342e-06, "loss": 0.6692, "step": 32638 }, { "epoch": 3.472972972972973, "grad_norm": 12.546875, "learning_rate": 8.858051449687322e-06, "loss": 0.6192, "step": 32639 }, { "epoch": 3.4730793785911898, "grad_norm": 13.0859375, "learning_rate": 8.858015918135305e-06, "loss": 0.5346, "step": 32640 }, { "epoch": 3.4731857842094063, "grad_norm": 16.65625, "learning_rate": 8.857980386583287e-06, "loss": 0.8164, "step": 32641 }, { "epoch": 3.4732921898276228, "grad_norm": 17.734375, "learning_rate": 8.857944855031268e-06, "loss": 0.633, "step": 32642 }, { "epoch": 3.4733985954458397, "grad_norm": 14.4765625, "learning_rate": 8.85790932347925e-06, "loss": 0.5929, "step": 32643 }, { "epoch": 3.4735050010640562, "grad_norm": 12.8515625, "learning_rate": 8.857873791927232e-06, "loss": 0.6106, "step": 32644 }, { "epoch": 3.4736114066822728, "grad_norm": 13.625, "learning_rate": 8.857838260375215e-06, "loss": 0.6567, "step": 32645 }, { "epoch": 3.4737178123004897, "grad_norm": 14.0390625, "learning_rate": 8.857802728823196e-06, "loss": 0.7004, "step": 32646 }, { "epoch": 3.473824217918706, "grad_norm": 11.6015625, "learning_rate": 8.857767197271178e-06, "loss": 0.477, "step": 32647 }, { "epoch": 3.4739306235369227, "grad_norm": 13.2578125, "learning_rate": 8.85773166571916e-06, "loss": 0.7048, "step": 32648 }, { "epoch": 3.4740370291551392, "grad_norm": 13.3515625, "learning_rate": 8.857696134167141e-06, "loss": 0.5586, "step": 32649 }, { "epoch": 3.474143434773356, "grad_norm": 12.8203125, "learning_rate": 8.857660602615123e-06, "loss": 0.6445, "step": 32650 }, { "epoch": 3.4742498403915727, "grad_norm": 10.7578125, "learning_rate": 8.857625071063104e-06, "loss": 0.5058, "step": 32651 }, { "epoch": 3.474356246009789, "grad_norm": 12.6484375, "learning_rate": 8.857589539511086e-06, "loss": 0.6233, "step": 32652 }, { "epoch": 3.4744626516280057, "grad_norm": 12.7578125, "learning_rate": 8.857554007959068e-06, "loss": 0.7619, "step": 32653 }, { "epoch": 3.4745690572462227, "grad_norm": 27.125, "learning_rate": 8.85751847640705e-06, "loss": 0.9232, "step": 32654 }, { "epoch": 3.474675462864439, "grad_norm": 16.8125, "learning_rate": 8.857482944855031e-06, "loss": 0.6935, "step": 32655 }, { "epoch": 3.4747818684826557, "grad_norm": 15.125, "learning_rate": 8.857447413303014e-06, "loss": 0.7268, "step": 32656 }, { "epoch": 3.4748882741008726, "grad_norm": 14.90625, "learning_rate": 8.857411881750996e-06, "loss": 0.6684, "step": 32657 }, { "epoch": 3.474994679719089, "grad_norm": 12.515625, "learning_rate": 8.857376350198977e-06, "loss": 0.5518, "step": 32658 }, { "epoch": 3.4751010853373057, "grad_norm": 12.6953125, "learning_rate": 8.857340818646959e-06, "loss": 0.7693, "step": 32659 }, { "epoch": 3.4752074909555226, "grad_norm": 14.765625, "learning_rate": 8.85730528709494e-06, "loss": 0.6603, "step": 32660 }, { "epoch": 3.475313896573739, "grad_norm": 10.390625, "learning_rate": 8.857269755542924e-06, "loss": 0.4949, "step": 32661 }, { "epoch": 3.4754203021919556, "grad_norm": 15.1015625, "learning_rate": 8.857234223990906e-06, "loss": 0.8553, "step": 32662 }, { "epoch": 3.4755267078101726, "grad_norm": 9.6328125, "learning_rate": 8.857198692438887e-06, "loss": 0.5598, "step": 32663 }, { "epoch": 3.475633113428389, "grad_norm": 16.109375, "learning_rate": 8.857163160886869e-06, "loss": 0.713, "step": 32664 }, { "epoch": 3.4757395190466056, "grad_norm": 13.0625, "learning_rate": 8.85712762933485e-06, "loss": 0.5811, "step": 32665 }, { "epoch": 3.475845924664822, "grad_norm": 19.96875, "learning_rate": 8.857092097782832e-06, "loss": 0.7262, "step": 32666 }, { "epoch": 3.475952330283039, "grad_norm": 10.671875, "learning_rate": 8.857056566230814e-06, "loss": 0.7112, "step": 32667 }, { "epoch": 3.4760587359012556, "grad_norm": 20.25, "learning_rate": 8.857021034678795e-06, "loss": 0.756, "step": 32668 }, { "epoch": 3.476165141519472, "grad_norm": 13.0234375, "learning_rate": 8.856985503126777e-06, "loss": 0.6595, "step": 32669 }, { "epoch": 3.476271547137689, "grad_norm": 13.8984375, "learning_rate": 8.856949971574758e-06, "loss": 0.6393, "step": 32670 }, { "epoch": 3.4763779527559056, "grad_norm": 12.2734375, "learning_rate": 8.856914440022742e-06, "loss": 0.5998, "step": 32671 }, { "epoch": 3.476484358374122, "grad_norm": 10.1796875, "learning_rate": 8.856878908470723e-06, "loss": 0.5495, "step": 32672 }, { "epoch": 3.4765907639923386, "grad_norm": 14.1875, "learning_rate": 8.856843376918705e-06, "loss": 0.7172, "step": 32673 }, { "epoch": 3.4766971696105555, "grad_norm": 16.09375, "learning_rate": 8.856807845366687e-06, "loss": 0.7285, "step": 32674 }, { "epoch": 3.476803575228772, "grad_norm": 15.921875, "learning_rate": 8.856772313814668e-06, "loss": 0.7818, "step": 32675 }, { "epoch": 3.4769099808469885, "grad_norm": 15.4375, "learning_rate": 8.85673678226265e-06, "loss": 0.6078, "step": 32676 }, { "epoch": 3.4770163864652055, "grad_norm": 11.6875, "learning_rate": 8.856701250710631e-06, "loss": 0.7057, "step": 32677 }, { "epoch": 3.477122792083422, "grad_norm": 14.9921875, "learning_rate": 8.856665719158613e-06, "loss": 0.6857, "step": 32678 }, { "epoch": 3.4772291977016385, "grad_norm": 13.375, "learning_rate": 8.856630187606595e-06, "loss": 0.6074, "step": 32679 }, { "epoch": 3.4773356033198555, "grad_norm": 16.671875, "learning_rate": 8.856594656054576e-06, "loss": 0.7979, "step": 32680 }, { "epoch": 3.477442008938072, "grad_norm": 12.078125, "learning_rate": 8.85655912450256e-06, "loss": 0.5868, "step": 32681 }, { "epoch": 3.4775484145562885, "grad_norm": 11.6015625, "learning_rate": 8.856523592950541e-06, "loss": 0.6217, "step": 32682 }, { "epoch": 3.4776548201745054, "grad_norm": 10.09375, "learning_rate": 8.856488061398523e-06, "loss": 0.5754, "step": 32683 }, { "epoch": 3.477761225792722, "grad_norm": 16.546875, "learning_rate": 8.856452529846504e-06, "loss": 0.7339, "step": 32684 }, { "epoch": 3.4778676314109385, "grad_norm": 10.34375, "learning_rate": 8.856416998294486e-06, "loss": 0.6284, "step": 32685 }, { "epoch": 3.477974037029155, "grad_norm": 13.6171875, "learning_rate": 8.85638146674247e-06, "loss": 0.7626, "step": 32686 }, { "epoch": 3.478080442647372, "grad_norm": 20.671875, "learning_rate": 8.85634593519045e-06, "loss": 0.89, "step": 32687 }, { "epoch": 3.4781868482655884, "grad_norm": 17.09375, "learning_rate": 8.856310403638432e-06, "loss": 0.7617, "step": 32688 }, { "epoch": 3.478293253883805, "grad_norm": 11.375, "learning_rate": 8.856274872086414e-06, "loss": 0.5887, "step": 32689 }, { "epoch": 3.4783996595020215, "grad_norm": 12.6328125, "learning_rate": 8.856239340534396e-06, "loss": 0.6405, "step": 32690 }, { "epoch": 3.4785060651202384, "grad_norm": 9.953125, "learning_rate": 8.856203808982377e-06, "loss": 0.6037, "step": 32691 }, { "epoch": 3.478612470738455, "grad_norm": 13.0390625, "learning_rate": 8.856168277430359e-06, "loss": 0.7045, "step": 32692 }, { "epoch": 3.4787188763566714, "grad_norm": 11.5703125, "learning_rate": 8.85613274587834e-06, "loss": 0.5283, "step": 32693 }, { "epoch": 3.4788252819748884, "grad_norm": 14.8359375, "learning_rate": 8.856097214326322e-06, "loss": 0.669, "step": 32694 }, { "epoch": 3.478931687593105, "grad_norm": 15.859375, "learning_rate": 8.856061682774304e-06, "loss": 0.763, "step": 32695 }, { "epoch": 3.4790380932113214, "grad_norm": 9.65625, "learning_rate": 8.856026151222285e-06, "loss": 0.5052, "step": 32696 }, { "epoch": 3.4791444988295384, "grad_norm": 14.625, "learning_rate": 8.855990619670269e-06, "loss": 0.7726, "step": 32697 }, { "epoch": 3.479250904447755, "grad_norm": 14.9609375, "learning_rate": 8.85595508811825e-06, "loss": 0.7302, "step": 32698 }, { "epoch": 3.4793573100659714, "grad_norm": 11.03125, "learning_rate": 8.855919556566232e-06, "loss": 0.6708, "step": 32699 }, { "epoch": 3.4794637156841883, "grad_norm": 12.0703125, "learning_rate": 8.855884025014213e-06, "loss": 0.6246, "step": 32700 }, { "epoch": 3.479570121302405, "grad_norm": 11.6484375, "learning_rate": 8.855848493462195e-06, "loss": 0.6479, "step": 32701 }, { "epoch": 3.4796765269206213, "grad_norm": 14.8359375, "learning_rate": 8.855812961910177e-06, "loss": 0.685, "step": 32702 }, { "epoch": 3.479782932538838, "grad_norm": 16.984375, "learning_rate": 8.855777430358158e-06, "loss": 0.6964, "step": 32703 }, { "epoch": 3.479889338157055, "grad_norm": 8.84375, "learning_rate": 8.85574189880614e-06, "loss": 0.5299, "step": 32704 }, { "epoch": 3.4799957437752713, "grad_norm": 10.4765625, "learning_rate": 8.855706367254121e-06, "loss": 0.5993, "step": 32705 }, { "epoch": 3.480102149393488, "grad_norm": 14.125, "learning_rate": 8.855670835702103e-06, "loss": 0.663, "step": 32706 }, { "epoch": 3.480208555011705, "grad_norm": 14.03125, "learning_rate": 8.855635304150086e-06, "loss": 0.7721, "step": 32707 }, { "epoch": 3.4803149606299213, "grad_norm": 11.8046875, "learning_rate": 8.855599772598068e-06, "loss": 0.6188, "step": 32708 }, { "epoch": 3.480421366248138, "grad_norm": 12.796875, "learning_rate": 8.85556424104605e-06, "loss": 0.5801, "step": 32709 }, { "epoch": 3.4805277718663543, "grad_norm": 12.859375, "learning_rate": 8.855528709494031e-06, "loss": 0.5476, "step": 32710 }, { "epoch": 3.4806341774845713, "grad_norm": 10.90625, "learning_rate": 8.855493177942013e-06, "loss": 0.5933, "step": 32711 }, { "epoch": 3.4807405831027878, "grad_norm": 11.8046875, "learning_rate": 8.855457646389996e-06, "loss": 0.6567, "step": 32712 }, { "epoch": 3.4808469887210043, "grad_norm": 11.296875, "learning_rate": 8.855422114837978e-06, "loss": 0.6592, "step": 32713 }, { "epoch": 3.4809533943392212, "grad_norm": 13.625, "learning_rate": 8.85538658328596e-06, "loss": 0.6267, "step": 32714 }, { "epoch": 3.4810597999574378, "grad_norm": 10.1328125, "learning_rate": 8.855351051733941e-06, "loss": 0.4487, "step": 32715 }, { "epoch": 3.4811662055756543, "grad_norm": 13.265625, "learning_rate": 8.855315520181922e-06, "loss": 0.6131, "step": 32716 }, { "epoch": 3.481272611193871, "grad_norm": 16.203125, "learning_rate": 8.855279988629904e-06, "loss": 0.6368, "step": 32717 }, { "epoch": 3.4813790168120877, "grad_norm": 19.921875, "learning_rate": 8.855244457077886e-06, "loss": 0.7847, "step": 32718 }, { "epoch": 3.4814854224303042, "grad_norm": 21.359375, "learning_rate": 8.855208925525867e-06, "loss": 0.7536, "step": 32719 }, { "epoch": 3.481591828048521, "grad_norm": 13.3046875, "learning_rate": 8.855173393973849e-06, "loss": 0.6814, "step": 32720 }, { "epoch": 3.4816982336667377, "grad_norm": 13.0, "learning_rate": 8.85513786242183e-06, "loss": 0.6187, "step": 32721 }, { "epoch": 3.481804639284954, "grad_norm": 20.828125, "learning_rate": 8.855102330869814e-06, "loss": 0.8264, "step": 32722 }, { "epoch": 3.4819110449031707, "grad_norm": 11.875, "learning_rate": 8.855066799317795e-06, "loss": 0.5223, "step": 32723 }, { "epoch": 3.4820174505213877, "grad_norm": 10.5546875, "learning_rate": 8.855031267765777e-06, "loss": 0.5931, "step": 32724 }, { "epoch": 3.482123856139604, "grad_norm": 19.53125, "learning_rate": 8.854995736213759e-06, "loss": 0.7531, "step": 32725 }, { "epoch": 3.4822302617578207, "grad_norm": 10.4921875, "learning_rate": 8.85496020466174e-06, "loss": 0.5894, "step": 32726 }, { "epoch": 3.4823366673760376, "grad_norm": 10.7578125, "learning_rate": 8.854924673109724e-06, "loss": 0.6575, "step": 32727 }, { "epoch": 3.482443072994254, "grad_norm": 16.640625, "learning_rate": 8.854889141557703e-06, "loss": 0.7987, "step": 32728 }, { "epoch": 3.4825494786124707, "grad_norm": 14.5703125, "learning_rate": 8.854853610005685e-06, "loss": 0.5846, "step": 32729 }, { "epoch": 3.482655884230687, "grad_norm": 14.3203125, "learning_rate": 8.854818078453667e-06, "loss": 0.636, "step": 32730 }, { "epoch": 3.482762289848904, "grad_norm": 14.4375, "learning_rate": 8.854782546901648e-06, "loss": 0.7859, "step": 32731 }, { "epoch": 3.4828686954671206, "grad_norm": 13.3515625, "learning_rate": 8.85474701534963e-06, "loss": 0.6495, "step": 32732 }, { "epoch": 3.482975101085337, "grad_norm": 12.2734375, "learning_rate": 8.854711483797613e-06, "loss": 0.6425, "step": 32733 }, { "epoch": 3.483081506703554, "grad_norm": 17.34375, "learning_rate": 8.854675952245595e-06, "loss": 0.8821, "step": 32734 }, { "epoch": 3.4831879123217706, "grad_norm": 13.71875, "learning_rate": 8.854640420693576e-06, "loss": 0.7115, "step": 32735 }, { "epoch": 3.483294317939987, "grad_norm": 14.8671875, "learning_rate": 8.854604889141558e-06, "loss": 0.6808, "step": 32736 }, { "epoch": 3.483400723558204, "grad_norm": 14.7890625, "learning_rate": 8.85456935758954e-06, "loss": 0.7189, "step": 32737 }, { "epoch": 3.4835071291764206, "grad_norm": 12.921875, "learning_rate": 8.854533826037523e-06, "loss": 0.5518, "step": 32738 }, { "epoch": 3.483613534794637, "grad_norm": 13.078125, "learning_rate": 8.854498294485505e-06, "loss": 0.7246, "step": 32739 }, { "epoch": 3.483719940412854, "grad_norm": 14.65625, "learning_rate": 8.854462762933486e-06, "loss": 0.6801, "step": 32740 }, { "epoch": 3.4838263460310706, "grad_norm": 12.96875, "learning_rate": 8.854427231381468e-06, "loss": 0.8461, "step": 32741 }, { "epoch": 3.483932751649287, "grad_norm": 13.1796875, "learning_rate": 8.85439169982945e-06, "loss": 0.7135, "step": 32742 }, { "epoch": 3.4840391572675036, "grad_norm": 11.2265625, "learning_rate": 8.854356168277431e-06, "loss": 0.5826, "step": 32743 }, { "epoch": 3.4841455628857205, "grad_norm": 12.171875, "learning_rate": 8.854320636725413e-06, "loss": 0.7171, "step": 32744 }, { "epoch": 3.484251968503937, "grad_norm": 16.046875, "learning_rate": 8.854285105173394e-06, "loss": 0.6645, "step": 32745 }, { "epoch": 3.4843583741221535, "grad_norm": 13.5078125, "learning_rate": 8.854249573621376e-06, "loss": 0.7018, "step": 32746 }, { "epoch": 3.48446477974037, "grad_norm": 12.8515625, "learning_rate": 8.854214042069357e-06, "loss": 0.6934, "step": 32747 }, { "epoch": 3.484571185358587, "grad_norm": 11.78125, "learning_rate": 8.85417851051734e-06, "loss": 0.6139, "step": 32748 }, { "epoch": 3.4846775909768035, "grad_norm": 17.109375, "learning_rate": 8.854142978965322e-06, "loss": 0.8628, "step": 32749 }, { "epoch": 3.48478399659502, "grad_norm": 14.8359375, "learning_rate": 8.854107447413304e-06, "loss": 0.7175, "step": 32750 }, { "epoch": 3.484890402213237, "grad_norm": 10.2421875, "learning_rate": 8.854071915861285e-06, "loss": 0.4689, "step": 32751 }, { "epoch": 3.4849968078314535, "grad_norm": 14.65625, "learning_rate": 8.854036384309267e-06, "loss": 0.6148, "step": 32752 }, { "epoch": 3.48510321344967, "grad_norm": 11.7421875, "learning_rate": 8.85400085275725e-06, "loss": 0.5934, "step": 32753 }, { "epoch": 3.485209619067887, "grad_norm": 12.8203125, "learning_rate": 8.853965321205232e-06, "loss": 0.7409, "step": 32754 }, { "epoch": 3.4853160246861035, "grad_norm": 13.0, "learning_rate": 8.853929789653214e-06, "loss": 0.7164, "step": 32755 }, { "epoch": 3.48542243030432, "grad_norm": 9.9765625, "learning_rate": 8.853894258101195e-06, "loss": 0.5308, "step": 32756 }, { "epoch": 3.485528835922537, "grad_norm": 12.21875, "learning_rate": 8.853858726549177e-06, "loss": 0.5382, "step": 32757 }, { "epoch": 3.4856352415407534, "grad_norm": 14.671875, "learning_rate": 8.853823194997158e-06, "loss": 0.6158, "step": 32758 }, { "epoch": 3.48574164715897, "grad_norm": 15.1171875, "learning_rate": 8.85378766344514e-06, "loss": 0.7012, "step": 32759 }, { "epoch": 3.4858480527771865, "grad_norm": 11.25, "learning_rate": 8.853752131893122e-06, "loss": 0.5726, "step": 32760 }, { "epoch": 3.4859544583954034, "grad_norm": 11.765625, "learning_rate": 8.853716600341103e-06, "loss": 0.6316, "step": 32761 }, { "epoch": 3.48606086401362, "grad_norm": 13.2265625, "learning_rate": 8.853681068789085e-06, "loss": 0.5912, "step": 32762 }, { "epoch": 3.4861672696318364, "grad_norm": 15.359375, "learning_rate": 8.853645537237068e-06, "loss": 0.7563, "step": 32763 }, { "epoch": 3.4862736752500534, "grad_norm": 16.890625, "learning_rate": 8.85361000568505e-06, "loss": 0.8426, "step": 32764 }, { "epoch": 3.48638008086827, "grad_norm": 12.46875, "learning_rate": 8.853574474133031e-06, "loss": 0.5857, "step": 32765 }, { "epoch": 3.4864864864864864, "grad_norm": 12.3515625, "learning_rate": 8.853538942581013e-06, "loss": 0.6694, "step": 32766 }, { "epoch": 3.486592892104703, "grad_norm": 21.140625, "learning_rate": 8.853503411028995e-06, "loss": 0.8539, "step": 32767 }, { "epoch": 3.48669929772292, "grad_norm": 11.65625, "learning_rate": 8.853467879476976e-06, "loss": 0.6513, "step": 32768 }, { "epoch": 3.4868057033411364, "grad_norm": 19.0, "learning_rate": 8.853432347924958e-06, "loss": 0.8852, "step": 32769 }, { "epoch": 3.486912108959353, "grad_norm": 13.8359375, "learning_rate": 8.85339681637294e-06, "loss": 0.7873, "step": 32770 }, { "epoch": 3.48701851457757, "grad_norm": 14.4296875, "learning_rate": 8.853361284820921e-06, "loss": 0.6953, "step": 32771 }, { "epoch": 3.4871249201957863, "grad_norm": 14.34375, "learning_rate": 8.853325753268903e-06, "loss": 0.6357, "step": 32772 }, { "epoch": 3.487231325814003, "grad_norm": 13.1484375, "learning_rate": 8.853290221716884e-06, "loss": 0.6392, "step": 32773 }, { "epoch": 3.48733773143222, "grad_norm": 10.8984375, "learning_rate": 8.853254690164868e-06, "loss": 0.6152, "step": 32774 }, { "epoch": 3.4874441370504363, "grad_norm": 15.890625, "learning_rate": 8.853219158612849e-06, "loss": 0.8884, "step": 32775 }, { "epoch": 3.487550542668653, "grad_norm": 12.1328125, "learning_rate": 8.85318362706083e-06, "loss": 0.6149, "step": 32776 }, { "epoch": 3.48765694828687, "grad_norm": 12.7890625, "learning_rate": 8.853148095508812e-06, "loss": 0.7054, "step": 32777 }, { "epoch": 3.4877633539050863, "grad_norm": 14.296875, "learning_rate": 8.853112563956794e-06, "loss": 0.6897, "step": 32778 }, { "epoch": 3.487869759523303, "grad_norm": 10.5078125, "learning_rate": 8.853077032404777e-06, "loss": 0.5372, "step": 32779 }, { "epoch": 3.4879761651415193, "grad_norm": 14.8125, "learning_rate": 8.853041500852759e-06, "loss": 0.8131, "step": 32780 }, { "epoch": 3.4880825707597363, "grad_norm": 12.4609375, "learning_rate": 8.85300596930074e-06, "loss": 0.65, "step": 32781 }, { "epoch": 3.4881889763779528, "grad_norm": 12.6875, "learning_rate": 8.852970437748722e-06, "loss": 0.6618, "step": 32782 }, { "epoch": 3.4882953819961693, "grad_norm": 14.8515625, "learning_rate": 8.852934906196704e-06, "loss": 0.69, "step": 32783 }, { "epoch": 3.488401787614386, "grad_norm": 10.125, "learning_rate": 8.852899374644685e-06, "loss": 0.5964, "step": 32784 }, { "epoch": 3.4885081932326027, "grad_norm": 9.984375, "learning_rate": 8.852863843092667e-06, "loss": 0.7649, "step": 32785 }, { "epoch": 3.4886145988508193, "grad_norm": 10.8359375, "learning_rate": 8.852828311540648e-06, "loss": 0.667, "step": 32786 }, { "epoch": 3.4887210044690358, "grad_norm": 11.3828125, "learning_rate": 8.85279277998863e-06, "loss": 0.6067, "step": 32787 }, { "epoch": 3.4888274100872527, "grad_norm": 11.5, "learning_rate": 8.852757248436612e-06, "loss": 0.6536, "step": 32788 }, { "epoch": 3.4889338157054692, "grad_norm": 11.859375, "learning_rate": 8.852721716884595e-06, "loss": 0.5509, "step": 32789 }, { "epoch": 3.4890402213236857, "grad_norm": 10.203125, "learning_rate": 8.852686185332577e-06, "loss": 0.5682, "step": 32790 }, { "epoch": 3.4891466269419027, "grad_norm": 12.1015625, "learning_rate": 8.852650653780558e-06, "loss": 0.7358, "step": 32791 }, { "epoch": 3.489253032560119, "grad_norm": 14.0703125, "learning_rate": 8.85261512222854e-06, "loss": 0.6287, "step": 32792 }, { "epoch": 3.4893594381783357, "grad_norm": 18.875, "learning_rate": 8.852579590676521e-06, "loss": 0.788, "step": 32793 }, { "epoch": 3.4894658437965527, "grad_norm": 12.796875, "learning_rate": 8.852544059124503e-06, "loss": 0.6639, "step": 32794 }, { "epoch": 3.489572249414769, "grad_norm": 11.2421875, "learning_rate": 8.852508527572485e-06, "loss": 0.6467, "step": 32795 }, { "epoch": 3.4896786550329857, "grad_norm": 15.8984375, "learning_rate": 8.852472996020466e-06, "loss": 0.8145, "step": 32796 }, { "epoch": 3.489785060651202, "grad_norm": 11.6640625, "learning_rate": 8.852437464468448e-06, "loss": 0.675, "step": 32797 }, { "epoch": 3.489891466269419, "grad_norm": 13.484375, "learning_rate": 8.85240193291643e-06, "loss": 0.5813, "step": 32798 }, { "epoch": 3.4899978718876357, "grad_norm": 18.015625, "learning_rate": 8.852366401364413e-06, "loss": 0.8171, "step": 32799 }, { "epoch": 3.490104277505852, "grad_norm": 13.140625, "learning_rate": 8.852330869812394e-06, "loss": 0.5914, "step": 32800 }, { "epoch": 3.490210683124069, "grad_norm": 14.109375, "learning_rate": 8.852295338260376e-06, "loss": 0.5667, "step": 32801 }, { "epoch": 3.4903170887422856, "grad_norm": 16.671875, "learning_rate": 8.852259806708358e-06, "loss": 0.6954, "step": 32802 }, { "epoch": 3.490423494360502, "grad_norm": 10.609375, "learning_rate": 8.85222427515634e-06, "loss": 0.605, "step": 32803 }, { "epoch": 3.4905298999787187, "grad_norm": 15.0625, "learning_rate": 8.852188743604322e-06, "loss": 0.8236, "step": 32804 }, { "epoch": 3.4906363055969356, "grad_norm": 17.765625, "learning_rate": 8.852153212052304e-06, "loss": 0.7782, "step": 32805 }, { "epoch": 3.490742711215152, "grad_norm": 15.1015625, "learning_rate": 8.852117680500286e-06, "loss": 0.7355, "step": 32806 }, { "epoch": 3.4908491168333686, "grad_norm": 11.3984375, "learning_rate": 8.852082148948267e-06, "loss": 0.5828, "step": 32807 }, { "epoch": 3.4909555224515856, "grad_norm": 12.4375, "learning_rate": 8.852046617396249e-06, "loss": 0.5933, "step": 32808 }, { "epoch": 3.491061928069802, "grad_norm": 11.9609375, "learning_rate": 8.85201108584423e-06, "loss": 0.6065, "step": 32809 }, { "epoch": 3.4911683336880186, "grad_norm": 16.40625, "learning_rate": 8.851975554292212e-06, "loss": 0.7938, "step": 32810 }, { "epoch": 3.4912747393062356, "grad_norm": 12.8125, "learning_rate": 8.851940022740194e-06, "loss": 0.6584, "step": 32811 }, { "epoch": 3.491381144924452, "grad_norm": 15.0, "learning_rate": 8.851904491188175e-06, "loss": 0.6181, "step": 32812 }, { "epoch": 3.4914875505426686, "grad_norm": 10.6953125, "learning_rate": 8.851868959636157e-06, "loss": 0.5712, "step": 32813 }, { "epoch": 3.4915939561608855, "grad_norm": 11.953125, "learning_rate": 8.851833428084139e-06, "loss": 0.6568, "step": 32814 }, { "epoch": 3.491700361779102, "grad_norm": 14.0390625, "learning_rate": 8.851797896532122e-06, "loss": 0.6197, "step": 32815 }, { "epoch": 3.4918067673973185, "grad_norm": 14.1875, "learning_rate": 8.851762364980103e-06, "loss": 0.5936, "step": 32816 }, { "epoch": 3.491913173015535, "grad_norm": 13.78125, "learning_rate": 8.851726833428085e-06, "loss": 0.7929, "step": 32817 }, { "epoch": 3.492019578633752, "grad_norm": 15.6875, "learning_rate": 8.851691301876067e-06, "loss": 0.786, "step": 32818 }, { "epoch": 3.4921259842519685, "grad_norm": 13.0703125, "learning_rate": 8.851655770324048e-06, "loss": 0.6412, "step": 32819 }, { "epoch": 3.492232389870185, "grad_norm": 9.578125, "learning_rate": 8.85162023877203e-06, "loss": 0.5656, "step": 32820 }, { "epoch": 3.492338795488402, "grad_norm": 13.0703125, "learning_rate": 8.851584707220012e-06, "loss": 0.7296, "step": 32821 }, { "epoch": 3.4924452011066185, "grad_norm": 14.7578125, "learning_rate": 8.851549175667993e-06, "loss": 0.5473, "step": 32822 }, { "epoch": 3.492551606724835, "grad_norm": 12.4296875, "learning_rate": 8.851513644115975e-06, "loss": 0.6718, "step": 32823 }, { "epoch": 3.4926580123430515, "grad_norm": 9.3671875, "learning_rate": 8.851478112563956e-06, "loss": 0.5463, "step": 32824 }, { "epoch": 3.4927644179612685, "grad_norm": 10.2890625, "learning_rate": 8.85144258101194e-06, "loss": 0.6038, "step": 32825 }, { "epoch": 3.492870823579485, "grad_norm": 18.40625, "learning_rate": 8.851407049459921e-06, "loss": 0.8214, "step": 32826 }, { "epoch": 3.4929772291977015, "grad_norm": 10.7421875, "learning_rate": 8.851371517907903e-06, "loss": 0.5073, "step": 32827 }, { "epoch": 3.4930836348159184, "grad_norm": 15.9765625, "learning_rate": 8.851335986355884e-06, "loss": 0.7485, "step": 32828 }, { "epoch": 3.493190040434135, "grad_norm": 11.71875, "learning_rate": 8.851300454803866e-06, "loss": 0.6143, "step": 32829 }, { "epoch": 3.4932964460523515, "grad_norm": 11.8046875, "learning_rate": 8.85126492325185e-06, "loss": 0.6787, "step": 32830 }, { "epoch": 3.4934028516705684, "grad_norm": 12.7578125, "learning_rate": 8.851229391699831e-06, "loss": 0.5653, "step": 32831 }, { "epoch": 3.493509257288785, "grad_norm": 10.9453125, "learning_rate": 8.851193860147813e-06, "loss": 0.5099, "step": 32832 }, { "epoch": 3.4936156629070014, "grad_norm": 14.40625, "learning_rate": 8.851158328595794e-06, "loss": 0.6089, "step": 32833 }, { "epoch": 3.493722068525218, "grad_norm": 13.5859375, "learning_rate": 8.851122797043776e-06, "loss": 0.6714, "step": 32834 }, { "epoch": 3.493828474143435, "grad_norm": 14.4453125, "learning_rate": 8.851087265491757e-06, "loss": 0.6979, "step": 32835 }, { "epoch": 3.4939348797616514, "grad_norm": 15.8828125, "learning_rate": 8.851051733939739e-06, "loss": 0.7247, "step": 32836 }, { "epoch": 3.494041285379868, "grad_norm": 13.8515625, "learning_rate": 8.85101620238772e-06, "loss": 0.519, "step": 32837 }, { "epoch": 3.494147690998085, "grad_norm": 13.046875, "learning_rate": 8.850980670835702e-06, "loss": 0.7101, "step": 32838 }, { "epoch": 3.4942540966163014, "grad_norm": 16.453125, "learning_rate": 8.850945139283684e-06, "loss": 0.7048, "step": 32839 }, { "epoch": 3.494360502234518, "grad_norm": 14.65625, "learning_rate": 8.850909607731667e-06, "loss": 0.7634, "step": 32840 }, { "epoch": 3.4944669078527344, "grad_norm": 13.90625, "learning_rate": 8.850874076179649e-06, "loss": 0.605, "step": 32841 }, { "epoch": 3.4945733134709513, "grad_norm": 19.796875, "learning_rate": 8.85083854462763e-06, "loss": 0.6998, "step": 32842 }, { "epoch": 3.494679719089168, "grad_norm": 13.59375, "learning_rate": 8.850803013075612e-06, "loss": 0.5555, "step": 32843 }, { "epoch": 3.4947861247073844, "grad_norm": 18.671875, "learning_rate": 8.850767481523594e-06, "loss": 0.7702, "step": 32844 }, { "epoch": 3.4948925303256013, "grad_norm": 18.015625, "learning_rate": 8.850731949971577e-06, "loss": 0.7226, "step": 32845 }, { "epoch": 3.494998935943818, "grad_norm": 11.46875, "learning_rate": 8.850696418419558e-06, "loss": 0.5305, "step": 32846 }, { "epoch": 3.4951053415620343, "grad_norm": 17.234375, "learning_rate": 8.85066088686754e-06, "loss": 0.7073, "step": 32847 }, { "epoch": 3.4952117471802513, "grad_norm": 15.3671875, "learning_rate": 8.85062535531552e-06, "loss": 0.6183, "step": 32848 }, { "epoch": 3.495318152798468, "grad_norm": 15.8984375, "learning_rate": 8.850589823763502e-06, "loss": 0.7423, "step": 32849 }, { "epoch": 3.4954245584166843, "grad_norm": 11.96875, "learning_rate": 8.850554292211483e-06, "loss": 0.7083, "step": 32850 }, { "epoch": 3.4955309640349013, "grad_norm": 13.1953125, "learning_rate": 8.850518760659466e-06, "loss": 0.715, "step": 32851 }, { "epoch": 3.4956373696531178, "grad_norm": 12.65625, "learning_rate": 8.850483229107448e-06, "loss": 0.5894, "step": 32852 }, { "epoch": 3.4957437752713343, "grad_norm": 10.171875, "learning_rate": 8.85044769755543e-06, "loss": 0.5703, "step": 32853 }, { "epoch": 3.495850180889551, "grad_norm": 10.9453125, "learning_rate": 8.850412166003411e-06, "loss": 0.4649, "step": 32854 }, { "epoch": 3.4959565865077677, "grad_norm": 12.8515625, "learning_rate": 8.850376634451393e-06, "loss": 0.6499, "step": 32855 }, { "epoch": 3.4960629921259843, "grad_norm": 11.9375, "learning_rate": 8.850341102899376e-06, "loss": 0.5828, "step": 32856 }, { "epoch": 3.4961693977442008, "grad_norm": 13.71875, "learning_rate": 8.850305571347358e-06, "loss": 0.7285, "step": 32857 }, { "epoch": 3.4962758033624177, "grad_norm": 13.65625, "learning_rate": 8.85027003979534e-06, "loss": 0.695, "step": 32858 }, { "epoch": 3.4963822089806342, "grad_norm": 12.625, "learning_rate": 8.850234508243321e-06, "loss": 0.6365, "step": 32859 }, { "epoch": 3.4964886145988507, "grad_norm": 10.3046875, "learning_rate": 8.850198976691303e-06, "loss": 0.4456, "step": 32860 }, { "epoch": 3.4965950202170673, "grad_norm": 13.1953125, "learning_rate": 8.850163445139284e-06, "loss": 0.626, "step": 32861 }, { "epoch": 3.496701425835284, "grad_norm": 14.4140625, "learning_rate": 8.850127913587266e-06, "loss": 0.7701, "step": 32862 }, { "epoch": 3.4968078314535007, "grad_norm": 16.671875, "learning_rate": 8.850092382035247e-06, "loss": 0.8396, "step": 32863 }, { "epoch": 3.4969142370717172, "grad_norm": 11.3671875, "learning_rate": 8.850056850483229e-06, "loss": 0.5941, "step": 32864 }, { "epoch": 3.497020642689934, "grad_norm": 12.890625, "learning_rate": 8.85002131893121e-06, "loss": 0.6976, "step": 32865 }, { "epoch": 3.4971270483081507, "grad_norm": 11.2265625, "learning_rate": 8.849985787379194e-06, "loss": 0.5437, "step": 32866 }, { "epoch": 3.497233453926367, "grad_norm": 12.5859375, "learning_rate": 8.849950255827176e-06, "loss": 0.7532, "step": 32867 }, { "epoch": 3.497339859544584, "grad_norm": 13.21875, "learning_rate": 8.849914724275157e-06, "loss": 0.6811, "step": 32868 }, { "epoch": 3.4974462651628007, "grad_norm": 17.015625, "learning_rate": 8.849879192723139e-06, "loss": 0.6686, "step": 32869 }, { "epoch": 3.497552670781017, "grad_norm": 11.9296875, "learning_rate": 8.84984366117112e-06, "loss": 0.5303, "step": 32870 }, { "epoch": 3.497659076399234, "grad_norm": 12.6640625, "learning_rate": 8.849808129619104e-06, "loss": 0.5964, "step": 32871 }, { "epoch": 3.4977654820174506, "grad_norm": 14.65625, "learning_rate": 8.849772598067085e-06, "loss": 0.6784, "step": 32872 }, { "epoch": 3.497871887635667, "grad_norm": 11.53125, "learning_rate": 8.849737066515067e-06, "loss": 0.5856, "step": 32873 }, { "epoch": 3.4979782932538837, "grad_norm": 11.578125, "learning_rate": 8.849701534963049e-06, "loss": 0.6468, "step": 32874 }, { "epoch": 3.4980846988721006, "grad_norm": 16.046875, "learning_rate": 8.84966600341103e-06, "loss": 0.7738, "step": 32875 }, { "epoch": 3.498191104490317, "grad_norm": 13.921875, "learning_rate": 8.849630471859012e-06, "loss": 0.5919, "step": 32876 }, { "epoch": 3.4982975101085336, "grad_norm": 11.25, "learning_rate": 8.849594940306993e-06, "loss": 0.6183, "step": 32877 }, { "epoch": 3.49840391572675, "grad_norm": 15.1484375, "learning_rate": 8.849559408754975e-06, "loss": 0.6359, "step": 32878 }, { "epoch": 3.498510321344967, "grad_norm": 12.015625, "learning_rate": 8.849523877202957e-06, "loss": 0.6699, "step": 32879 }, { "epoch": 3.4986167269631836, "grad_norm": 11.3359375, "learning_rate": 8.849488345650938e-06, "loss": 0.6339, "step": 32880 }, { "epoch": 3.4987231325814, "grad_norm": 12.6484375, "learning_rate": 8.849452814098921e-06, "loss": 0.7164, "step": 32881 }, { "epoch": 3.498829538199617, "grad_norm": 14.3984375, "learning_rate": 8.849417282546903e-06, "loss": 0.6935, "step": 32882 }, { "epoch": 3.4989359438178336, "grad_norm": 10.8984375, "learning_rate": 8.849381750994885e-06, "loss": 0.5183, "step": 32883 }, { "epoch": 3.49904234943605, "grad_norm": 15.25, "learning_rate": 8.849346219442866e-06, "loss": 0.7003, "step": 32884 }, { "epoch": 3.499148755054267, "grad_norm": 12.9765625, "learning_rate": 8.849310687890848e-06, "loss": 0.6488, "step": 32885 }, { "epoch": 3.4992551606724835, "grad_norm": 14.109375, "learning_rate": 8.84927515633883e-06, "loss": 0.6478, "step": 32886 }, { "epoch": 3.4993615662907, "grad_norm": 12.1484375, "learning_rate": 8.849239624786811e-06, "loss": 0.5627, "step": 32887 }, { "epoch": 3.499467971908917, "grad_norm": 11.7265625, "learning_rate": 8.849204093234793e-06, "loss": 0.6129, "step": 32888 }, { "epoch": 3.4995743775271335, "grad_norm": 14.375, "learning_rate": 8.849168561682774e-06, "loss": 0.6177, "step": 32889 }, { "epoch": 3.49968078314535, "grad_norm": 11.1875, "learning_rate": 8.849133030130756e-06, "loss": 0.5398, "step": 32890 }, { "epoch": 3.4997871887635665, "grad_norm": 14.0703125, "learning_rate": 8.849097498578738e-06, "loss": 0.635, "step": 32891 }, { "epoch": 3.4998935943817835, "grad_norm": 15.671875, "learning_rate": 8.84906196702672e-06, "loss": 0.5876, "step": 32892 }, { "epoch": 3.5, "grad_norm": 11.1015625, "learning_rate": 8.849026435474702e-06, "loss": 0.5413, "step": 32893 }, { "epoch": 3.5001064056182165, "grad_norm": 10.0, "learning_rate": 8.848990903922684e-06, "loss": 0.4394, "step": 32894 }, { "epoch": 3.500212811236433, "grad_norm": 9.5546875, "learning_rate": 8.848955372370666e-06, "loss": 0.6297, "step": 32895 }, { "epoch": 3.50031921685465, "grad_norm": 14.875, "learning_rate": 8.848919840818647e-06, "loss": 0.6069, "step": 32896 }, { "epoch": 3.5004256224728665, "grad_norm": 11.6171875, "learning_rate": 8.84888430926663e-06, "loss": 0.5684, "step": 32897 }, { "epoch": 3.500532028091083, "grad_norm": 13.8203125, "learning_rate": 8.848848777714612e-06, "loss": 0.7392, "step": 32898 }, { "epoch": 3.5006384337093, "grad_norm": 13.5390625, "learning_rate": 8.848813246162594e-06, "loss": 0.7064, "step": 32899 }, { "epoch": 3.5007448393275165, "grad_norm": 16.953125, "learning_rate": 8.848777714610575e-06, "loss": 0.5707, "step": 32900 }, { "epoch": 3.500851244945733, "grad_norm": 13.9765625, "learning_rate": 8.848742183058557e-06, "loss": 0.7389, "step": 32901 }, { "epoch": 3.50095765056395, "grad_norm": 13.4296875, "learning_rate": 8.848706651506539e-06, "loss": 0.6455, "step": 32902 }, { "epoch": 3.5010640561821664, "grad_norm": 19.9375, "learning_rate": 8.84867111995452e-06, "loss": 0.5309, "step": 32903 }, { "epoch": 3.501170461800383, "grad_norm": 15.4140625, "learning_rate": 8.848635588402502e-06, "loss": 0.7304, "step": 32904 }, { "epoch": 3.5012768674186, "grad_norm": 14.2890625, "learning_rate": 8.848600056850483e-06, "loss": 0.6421, "step": 32905 }, { "epoch": 3.5013832730368164, "grad_norm": 14.875, "learning_rate": 8.848564525298465e-06, "loss": 0.666, "step": 32906 }, { "epoch": 3.501489678655033, "grad_norm": 13.203125, "learning_rate": 8.848528993746448e-06, "loss": 0.527, "step": 32907 }, { "epoch": 3.50159608427325, "grad_norm": 15.7578125, "learning_rate": 8.84849346219443e-06, "loss": 0.6737, "step": 32908 }, { "epoch": 3.5017024898914664, "grad_norm": 9.71875, "learning_rate": 8.848457930642412e-06, "loss": 0.5758, "step": 32909 }, { "epoch": 3.501808895509683, "grad_norm": 13.3203125, "learning_rate": 8.848422399090393e-06, "loss": 0.5776, "step": 32910 }, { "epoch": 3.5019153011278994, "grad_norm": 13.0078125, "learning_rate": 8.848386867538375e-06, "loss": 0.6123, "step": 32911 }, { "epoch": 3.5020217067461163, "grad_norm": 14.1171875, "learning_rate": 8.848351335986356e-06, "loss": 0.6799, "step": 32912 }, { "epoch": 3.502128112364333, "grad_norm": 12.2734375, "learning_rate": 8.848315804434338e-06, "loss": 0.6461, "step": 32913 }, { "epoch": 3.5022345179825494, "grad_norm": 16.125, "learning_rate": 8.84828027288232e-06, "loss": 0.6931, "step": 32914 }, { "epoch": 3.502340923600766, "grad_norm": 12.7890625, "learning_rate": 8.848244741330301e-06, "loss": 0.5355, "step": 32915 }, { "epoch": 3.502447329218983, "grad_norm": 12.28125, "learning_rate": 8.848209209778283e-06, "loss": 0.6779, "step": 32916 }, { "epoch": 3.5025537348371993, "grad_norm": 10.2265625, "learning_rate": 8.848173678226266e-06, "loss": 0.6735, "step": 32917 }, { "epoch": 3.502660140455416, "grad_norm": 13.8125, "learning_rate": 8.848138146674248e-06, "loss": 0.6297, "step": 32918 }, { "epoch": 3.502766546073633, "grad_norm": 12.296875, "learning_rate": 8.84810261512223e-06, "loss": 0.7004, "step": 32919 }, { "epoch": 3.5028729516918493, "grad_norm": 11.8203125, "learning_rate": 8.848067083570211e-06, "loss": 0.6403, "step": 32920 }, { "epoch": 3.502979357310066, "grad_norm": 13.9765625, "learning_rate": 8.848031552018192e-06, "loss": 0.5784, "step": 32921 }, { "epoch": 3.5030857629282828, "grad_norm": 12.421875, "learning_rate": 8.847996020466176e-06, "loss": 0.6167, "step": 32922 }, { "epoch": 3.5031921685464993, "grad_norm": 14.2421875, "learning_rate": 8.847960488914157e-06, "loss": 0.6746, "step": 32923 }, { "epoch": 3.503298574164716, "grad_norm": 12.3515625, "learning_rate": 8.847924957362139e-06, "loss": 0.5611, "step": 32924 }, { "epoch": 3.5034049797829327, "grad_norm": 14.140625, "learning_rate": 8.84788942581012e-06, "loss": 0.6134, "step": 32925 }, { "epoch": 3.5035113854011493, "grad_norm": 11.5078125, "learning_rate": 8.847853894258102e-06, "loss": 0.6637, "step": 32926 }, { "epoch": 3.5036177910193658, "grad_norm": 21.296875, "learning_rate": 8.847818362706084e-06, "loss": 0.9054, "step": 32927 }, { "epoch": 3.5037241966375827, "grad_norm": 14.515625, "learning_rate": 8.847782831154065e-06, "loss": 0.7623, "step": 32928 }, { "epoch": 3.5038306022557992, "grad_norm": 18.953125, "learning_rate": 8.847747299602047e-06, "loss": 0.7729, "step": 32929 }, { "epoch": 3.5039370078740157, "grad_norm": 15.1328125, "learning_rate": 8.847711768050029e-06, "loss": 0.7105, "step": 32930 }, { "epoch": 3.5040434134922323, "grad_norm": 15.0, "learning_rate": 8.84767623649801e-06, "loss": 0.6557, "step": 32931 }, { "epoch": 3.5041498191104488, "grad_norm": 11.046875, "learning_rate": 8.847640704945992e-06, "loss": 0.5559, "step": 32932 }, { "epoch": 3.5042562247286657, "grad_norm": 15.765625, "learning_rate": 8.847605173393975e-06, "loss": 0.7294, "step": 32933 }, { "epoch": 3.5043626303468822, "grad_norm": 9.6171875, "learning_rate": 8.847569641841957e-06, "loss": 0.6231, "step": 32934 }, { "epoch": 3.5044690359650987, "grad_norm": 12.0703125, "learning_rate": 8.847534110289938e-06, "loss": 0.7011, "step": 32935 }, { "epoch": 3.5045754415833157, "grad_norm": 12.7109375, "learning_rate": 8.84749857873792e-06, "loss": 0.7336, "step": 32936 }, { "epoch": 3.504681847201532, "grad_norm": 12.0, "learning_rate": 8.847463047185902e-06, "loss": 0.5618, "step": 32937 }, { "epoch": 3.5047882528197487, "grad_norm": 15.71875, "learning_rate": 8.847427515633883e-06, "loss": 0.7451, "step": 32938 }, { "epoch": 3.5048946584379657, "grad_norm": 12.265625, "learning_rate": 8.847391984081865e-06, "loss": 0.6328, "step": 32939 }, { "epoch": 3.505001064056182, "grad_norm": 14.359375, "learning_rate": 8.847356452529846e-06, "loss": 0.7476, "step": 32940 }, { "epoch": 3.5051074696743987, "grad_norm": 12.1953125, "learning_rate": 8.847320920977828e-06, "loss": 0.6409, "step": 32941 }, { "epoch": 3.5052138752926156, "grad_norm": 10.1171875, "learning_rate": 8.84728538942581e-06, "loss": 0.5372, "step": 32942 }, { "epoch": 3.505320280910832, "grad_norm": 11.890625, "learning_rate": 8.847249857873793e-06, "loss": 0.6184, "step": 32943 }, { "epoch": 3.5054266865290487, "grad_norm": 14.109375, "learning_rate": 8.847214326321775e-06, "loss": 0.6658, "step": 32944 }, { "epoch": 3.5055330921472656, "grad_norm": 13.4296875, "learning_rate": 8.847178794769756e-06, "loss": 0.7362, "step": 32945 }, { "epoch": 3.505639497765482, "grad_norm": 15.25, "learning_rate": 8.847143263217738e-06, "loss": 0.6635, "step": 32946 }, { "epoch": 3.5057459033836986, "grad_norm": 10.7890625, "learning_rate": 8.84710773166572e-06, "loss": 0.4918, "step": 32947 }, { "epoch": 3.505852309001915, "grad_norm": 16.015625, "learning_rate": 8.847072200113703e-06, "loss": 0.7199, "step": 32948 }, { "epoch": 3.505958714620132, "grad_norm": 13.4609375, "learning_rate": 8.847036668561684e-06, "loss": 0.7375, "step": 32949 }, { "epoch": 3.5060651202383486, "grad_norm": 17.359375, "learning_rate": 8.847001137009666e-06, "loss": 0.6716, "step": 32950 }, { "epoch": 3.506171525856565, "grad_norm": 14.7265625, "learning_rate": 8.846965605457647e-06, "loss": 0.8115, "step": 32951 }, { "epoch": 3.5062779314747816, "grad_norm": 15.9375, "learning_rate": 8.846930073905629e-06, "loss": 0.6696, "step": 32952 }, { "epoch": 3.5063843370929986, "grad_norm": 16.21875, "learning_rate": 8.84689454235361e-06, "loss": 0.8308, "step": 32953 }, { "epoch": 3.506490742711215, "grad_norm": 13.6875, "learning_rate": 8.846859010801592e-06, "loss": 0.5178, "step": 32954 }, { "epoch": 3.5065971483294316, "grad_norm": 13.7578125, "learning_rate": 8.846823479249574e-06, "loss": 0.7406, "step": 32955 }, { "epoch": 3.5067035539476485, "grad_norm": 12.7265625, "learning_rate": 8.846787947697556e-06, "loss": 0.7151, "step": 32956 }, { "epoch": 3.506809959565865, "grad_norm": 10.4296875, "learning_rate": 8.846752416145537e-06, "loss": 0.5166, "step": 32957 }, { "epoch": 3.5069163651840816, "grad_norm": 12.6484375, "learning_rate": 8.84671688459352e-06, "loss": 0.6502, "step": 32958 }, { "epoch": 3.5070227708022985, "grad_norm": 12.4453125, "learning_rate": 8.846681353041502e-06, "loss": 0.6541, "step": 32959 }, { "epoch": 3.507129176420515, "grad_norm": 14.59375, "learning_rate": 8.846645821489484e-06, "loss": 0.6004, "step": 32960 }, { "epoch": 3.5072355820387315, "grad_norm": 16.09375, "learning_rate": 8.846610289937465e-06, "loss": 0.5608, "step": 32961 }, { "epoch": 3.5073419876569485, "grad_norm": 13.9921875, "learning_rate": 8.846574758385447e-06, "loss": 0.5992, "step": 32962 }, { "epoch": 3.507448393275165, "grad_norm": 15.4296875, "learning_rate": 8.84653922683343e-06, "loss": 0.6999, "step": 32963 }, { "epoch": 3.5075547988933815, "grad_norm": 11.9296875, "learning_rate": 8.846503695281412e-06, "loss": 0.5912, "step": 32964 }, { "epoch": 3.5076612045115985, "grad_norm": 12.6171875, "learning_rate": 8.846468163729393e-06, "loss": 0.7117, "step": 32965 }, { "epoch": 3.507767610129815, "grad_norm": 14.3203125, "learning_rate": 8.846432632177375e-06, "loss": 0.5738, "step": 32966 }, { "epoch": 3.5078740157480315, "grad_norm": 29.453125, "learning_rate": 8.846397100625357e-06, "loss": 0.7033, "step": 32967 }, { "epoch": 3.507980421366248, "grad_norm": 14.0546875, "learning_rate": 8.846361569073336e-06, "loss": 0.6326, "step": 32968 }, { "epoch": 3.508086826984465, "grad_norm": 12.6953125, "learning_rate": 8.84632603752132e-06, "loss": 0.6269, "step": 32969 }, { "epoch": 3.5081932326026815, "grad_norm": 13.0625, "learning_rate": 8.846290505969301e-06, "loss": 0.6284, "step": 32970 }, { "epoch": 3.508299638220898, "grad_norm": 13.9140625, "learning_rate": 8.846254974417283e-06, "loss": 0.5998, "step": 32971 }, { "epoch": 3.5084060438391145, "grad_norm": 11.9453125, "learning_rate": 8.846219442865265e-06, "loss": 0.7074, "step": 32972 }, { "epoch": 3.5085124494573314, "grad_norm": 9.6875, "learning_rate": 8.846183911313246e-06, "loss": 0.5618, "step": 32973 }, { "epoch": 3.508618855075548, "grad_norm": 11.265625, "learning_rate": 8.84614837976123e-06, "loss": 0.714, "step": 32974 }, { "epoch": 3.5087252606937644, "grad_norm": 10.2109375, "learning_rate": 8.846112848209211e-06, "loss": 0.7226, "step": 32975 }, { "epoch": 3.5088316663119814, "grad_norm": 11.3046875, "learning_rate": 8.846077316657193e-06, "loss": 0.6983, "step": 32976 }, { "epoch": 3.508938071930198, "grad_norm": 12.234375, "learning_rate": 8.846041785105174e-06, "loss": 0.6296, "step": 32977 }, { "epoch": 3.5090444775484144, "grad_norm": 16.0625, "learning_rate": 8.846006253553156e-06, "loss": 0.7197, "step": 32978 }, { "epoch": 3.5091508831666314, "grad_norm": 15.8828125, "learning_rate": 8.845970722001138e-06, "loss": 0.6995, "step": 32979 }, { "epoch": 3.509257288784848, "grad_norm": 14.34375, "learning_rate": 8.845935190449119e-06, "loss": 0.6984, "step": 32980 }, { "epoch": 3.5093636944030644, "grad_norm": 11.28125, "learning_rate": 8.8458996588971e-06, "loss": 0.601, "step": 32981 }, { "epoch": 3.5094701000212813, "grad_norm": 14.46875, "learning_rate": 8.845864127345082e-06, "loss": 0.6215, "step": 32982 }, { "epoch": 3.509576505639498, "grad_norm": 12.046875, "learning_rate": 8.845828595793064e-06, "loss": 0.6412, "step": 32983 }, { "epoch": 3.5096829112577144, "grad_norm": 10.3515625, "learning_rate": 8.845793064241047e-06, "loss": 0.5022, "step": 32984 }, { "epoch": 3.5097893168759313, "grad_norm": 12.3203125, "learning_rate": 8.845757532689029e-06, "loss": 0.6158, "step": 32985 }, { "epoch": 3.509895722494148, "grad_norm": 13.671875, "learning_rate": 8.84572200113701e-06, "loss": 0.6877, "step": 32986 }, { "epoch": 3.5100021281123643, "grad_norm": 11.1640625, "learning_rate": 8.845686469584992e-06, "loss": 0.6783, "step": 32987 }, { "epoch": 3.510108533730581, "grad_norm": 15.4765625, "learning_rate": 8.845650938032974e-06, "loss": 0.8072, "step": 32988 }, { "epoch": 3.5102149393487974, "grad_norm": 14.609375, "learning_rate": 8.845615406480957e-06, "loss": 0.6295, "step": 32989 }, { "epoch": 3.5103213449670143, "grad_norm": 11.0390625, "learning_rate": 8.845579874928939e-06, "loss": 0.5674, "step": 32990 }, { "epoch": 3.510427750585231, "grad_norm": 15.078125, "learning_rate": 8.84554434337692e-06, "loss": 0.7694, "step": 32991 }, { "epoch": 3.5105341562034473, "grad_norm": 11.34375, "learning_rate": 8.845508811824902e-06, "loss": 0.5554, "step": 32992 }, { "epoch": 3.5106405618216643, "grad_norm": 12.28125, "learning_rate": 8.845473280272883e-06, "loss": 0.7202, "step": 32993 }, { "epoch": 3.510746967439881, "grad_norm": 10.9140625, "learning_rate": 8.845437748720865e-06, "loss": 0.6541, "step": 32994 }, { "epoch": 3.5108533730580973, "grad_norm": 10.59375, "learning_rate": 8.845402217168847e-06, "loss": 0.6357, "step": 32995 }, { "epoch": 3.5109597786763143, "grad_norm": 11.859375, "learning_rate": 8.845366685616828e-06, "loss": 0.6627, "step": 32996 }, { "epoch": 3.5110661842945308, "grad_norm": 13.3359375, "learning_rate": 8.84533115406481e-06, "loss": 0.6129, "step": 32997 }, { "epoch": 3.5111725899127473, "grad_norm": 13.40625, "learning_rate": 8.845295622512791e-06, "loss": 0.734, "step": 32998 }, { "epoch": 3.5112789955309642, "grad_norm": 13.2734375, "learning_rate": 8.845260090960775e-06, "loss": 0.6462, "step": 32999 }, { "epoch": 3.5113854011491807, "grad_norm": 10.90625, "learning_rate": 8.845224559408756e-06, "loss": 0.656, "step": 33000 }, { "epoch": 3.5114918067673973, "grad_norm": 12.1484375, "learning_rate": 8.845189027856738e-06, "loss": 0.6204, "step": 33001 }, { "epoch": 3.511598212385614, "grad_norm": 18.75, "learning_rate": 8.84515349630472e-06, "loss": 0.9233, "step": 33002 }, { "epoch": 3.5117046180038307, "grad_norm": 10.6953125, "learning_rate": 8.845117964752701e-06, "loss": 0.5353, "step": 33003 }, { "epoch": 3.5118110236220472, "grad_norm": 18.5625, "learning_rate": 8.845082433200683e-06, "loss": 0.8984, "step": 33004 }, { "epoch": 3.5119174292402637, "grad_norm": 10.9921875, "learning_rate": 8.845046901648664e-06, "loss": 0.5063, "step": 33005 }, { "epoch": 3.5120238348584807, "grad_norm": 14.140625, "learning_rate": 8.845011370096646e-06, "loss": 0.6339, "step": 33006 }, { "epoch": 3.512130240476697, "grad_norm": 12.4296875, "learning_rate": 8.844975838544628e-06, "loss": 0.7218, "step": 33007 }, { "epoch": 3.5122366460949137, "grad_norm": 11.09375, "learning_rate": 8.84494030699261e-06, "loss": 0.5177, "step": 33008 }, { "epoch": 3.51234305171313, "grad_norm": 13.03125, "learning_rate": 8.84490477544059e-06, "loss": 0.614, "step": 33009 }, { "epoch": 3.512449457331347, "grad_norm": 11.4609375, "learning_rate": 8.844869243888574e-06, "loss": 0.696, "step": 33010 }, { "epoch": 3.5125558629495637, "grad_norm": 17.953125, "learning_rate": 8.844833712336556e-06, "loss": 0.7746, "step": 33011 }, { "epoch": 3.51266226856778, "grad_norm": 11.1875, "learning_rate": 8.844798180784537e-06, "loss": 0.6191, "step": 33012 }, { "epoch": 3.512768674185997, "grad_norm": 11.53125, "learning_rate": 8.844762649232519e-06, "loss": 0.5706, "step": 33013 }, { "epoch": 3.5128750798042137, "grad_norm": 13.3828125, "learning_rate": 8.8447271176805e-06, "loss": 0.6194, "step": 33014 }, { "epoch": 3.51298148542243, "grad_norm": 16.765625, "learning_rate": 8.844691586128484e-06, "loss": 0.6773, "step": 33015 }, { "epoch": 3.513087891040647, "grad_norm": 14.578125, "learning_rate": 8.844656054576465e-06, "loss": 0.638, "step": 33016 }, { "epoch": 3.5131942966588636, "grad_norm": 14.0703125, "learning_rate": 8.844620523024447e-06, "loss": 0.6575, "step": 33017 }, { "epoch": 3.51330070227708, "grad_norm": 15.3984375, "learning_rate": 8.844584991472429e-06, "loss": 0.7358, "step": 33018 }, { "epoch": 3.513407107895297, "grad_norm": 16.515625, "learning_rate": 8.84454945992041e-06, "loss": 0.8049, "step": 33019 }, { "epoch": 3.5135135135135136, "grad_norm": 11.828125, "learning_rate": 8.844513928368392e-06, "loss": 0.616, "step": 33020 }, { "epoch": 3.51361991913173, "grad_norm": 11.265625, "learning_rate": 8.844478396816373e-06, "loss": 0.6099, "step": 33021 }, { "epoch": 3.513726324749947, "grad_norm": 18.265625, "learning_rate": 8.844442865264355e-06, "loss": 0.7508, "step": 33022 }, { "epoch": 3.5138327303681636, "grad_norm": 15.71875, "learning_rate": 8.844407333712337e-06, "loss": 0.6709, "step": 33023 }, { "epoch": 3.51393913598638, "grad_norm": 11.34375, "learning_rate": 8.844371802160318e-06, "loss": 0.6386, "step": 33024 }, { "epoch": 3.5140455416045966, "grad_norm": 12.03125, "learning_rate": 8.844336270608302e-06, "loss": 0.5923, "step": 33025 }, { "epoch": 3.514151947222813, "grad_norm": 11.9296875, "learning_rate": 8.844300739056283e-06, "loss": 0.7846, "step": 33026 }, { "epoch": 3.51425835284103, "grad_norm": 12.7421875, "learning_rate": 8.844265207504265e-06, "loss": 0.6908, "step": 33027 }, { "epoch": 3.5143647584592466, "grad_norm": 15.40625, "learning_rate": 8.844229675952246e-06, "loss": 0.5879, "step": 33028 }, { "epoch": 3.514471164077463, "grad_norm": 12.578125, "learning_rate": 8.844194144400228e-06, "loss": 0.6152, "step": 33029 }, { "epoch": 3.51457756969568, "grad_norm": 15.1171875, "learning_rate": 8.84415861284821e-06, "loss": 0.7748, "step": 33030 }, { "epoch": 3.5146839753138965, "grad_norm": 14.3125, "learning_rate": 8.844123081296191e-06, "loss": 0.7528, "step": 33031 }, { "epoch": 3.514790380932113, "grad_norm": 12.6640625, "learning_rate": 8.844087549744173e-06, "loss": 0.6686, "step": 33032 }, { "epoch": 3.51489678655033, "grad_norm": 10.953125, "learning_rate": 8.844052018192154e-06, "loss": 0.5148, "step": 33033 }, { "epoch": 3.5150031921685465, "grad_norm": 15.953125, "learning_rate": 8.844016486640136e-06, "loss": 0.6343, "step": 33034 }, { "epoch": 3.515109597786763, "grad_norm": 10.609375, "learning_rate": 8.84398095508812e-06, "loss": 0.5801, "step": 33035 }, { "epoch": 3.51521600340498, "grad_norm": 15.484375, "learning_rate": 8.843945423536101e-06, "loss": 0.7667, "step": 33036 }, { "epoch": 3.5153224090231965, "grad_norm": 12.609375, "learning_rate": 8.843909891984083e-06, "loss": 0.6048, "step": 33037 }, { "epoch": 3.515428814641413, "grad_norm": 14.171875, "learning_rate": 8.843874360432064e-06, "loss": 0.7128, "step": 33038 }, { "epoch": 3.51553522025963, "grad_norm": 11.5546875, "learning_rate": 8.843838828880046e-06, "loss": 0.6214, "step": 33039 }, { "epoch": 3.5156416258778465, "grad_norm": 11.96875, "learning_rate": 8.843803297328029e-06, "loss": 0.676, "step": 33040 }, { "epoch": 3.515748031496063, "grad_norm": 14.4140625, "learning_rate": 8.84376776577601e-06, "loss": 0.6958, "step": 33041 }, { "epoch": 3.5158544371142795, "grad_norm": 13.15625, "learning_rate": 8.843732234223992e-06, "loss": 0.5608, "step": 33042 }, { "epoch": 3.5159608427324964, "grad_norm": 10.4921875, "learning_rate": 8.843696702671974e-06, "loss": 0.494, "step": 33043 }, { "epoch": 3.516067248350713, "grad_norm": 13.5859375, "learning_rate": 8.843661171119956e-06, "loss": 0.7708, "step": 33044 }, { "epoch": 3.5161736539689294, "grad_norm": 13.484375, "learning_rate": 8.843625639567937e-06, "loss": 0.6777, "step": 33045 }, { "epoch": 3.516280059587146, "grad_norm": 12.796875, "learning_rate": 8.843590108015919e-06, "loss": 0.6035, "step": 33046 }, { "epoch": 3.516386465205363, "grad_norm": 20.25, "learning_rate": 8.8435545764639e-06, "loss": 0.7722, "step": 33047 }, { "epoch": 3.5164928708235794, "grad_norm": 12.7421875, "learning_rate": 8.843519044911882e-06, "loss": 0.7176, "step": 33048 }, { "epoch": 3.516599276441796, "grad_norm": 14.3359375, "learning_rate": 8.843483513359864e-06, "loss": 0.5471, "step": 33049 }, { "epoch": 3.516705682060013, "grad_norm": 12.9296875, "learning_rate": 8.843447981807845e-06, "loss": 0.6516, "step": 33050 }, { "epoch": 3.5168120876782294, "grad_norm": 14.1484375, "learning_rate": 8.843412450255828e-06, "loss": 0.7707, "step": 33051 }, { "epoch": 3.516918493296446, "grad_norm": 14.3515625, "learning_rate": 8.84337691870381e-06, "loss": 0.6794, "step": 33052 }, { "epoch": 3.517024898914663, "grad_norm": 12.3984375, "learning_rate": 8.843341387151792e-06, "loss": 0.7659, "step": 33053 }, { "epoch": 3.5171313045328794, "grad_norm": 11.6640625, "learning_rate": 8.843305855599773e-06, "loss": 0.5644, "step": 33054 }, { "epoch": 3.517237710151096, "grad_norm": 17.078125, "learning_rate": 8.843270324047755e-06, "loss": 0.6674, "step": 33055 }, { "epoch": 3.517344115769313, "grad_norm": 12.65625, "learning_rate": 8.843234792495738e-06, "loss": 0.7233, "step": 33056 }, { "epoch": 3.5174505213875293, "grad_norm": 10.953125, "learning_rate": 8.843199260943718e-06, "loss": 0.5943, "step": 33057 }, { "epoch": 3.517556927005746, "grad_norm": 14.6640625, "learning_rate": 8.8431637293917e-06, "loss": 0.7955, "step": 33058 }, { "epoch": 3.517663332623963, "grad_norm": 11.8046875, "learning_rate": 8.843128197839681e-06, "loss": 0.6061, "step": 33059 }, { "epoch": 3.5177697382421793, "grad_norm": 13.3203125, "learning_rate": 8.843092666287663e-06, "loss": 0.596, "step": 33060 }, { "epoch": 3.517876143860396, "grad_norm": 13.7578125, "learning_rate": 8.843057134735646e-06, "loss": 0.7487, "step": 33061 }, { "epoch": 3.5179825494786123, "grad_norm": 11.7265625, "learning_rate": 8.843021603183628e-06, "loss": 0.4916, "step": 33062 }, { "epoch": 3.518088955096829, "grad_norm": 13.5625, "learning_rate": 8.84298607163161e-06, "loss": 0.6281, "step": 33063 }, { "epoch": 3.518195360715046, "grad_norm": 11.90625, "learning_rate": 8.842950540079591e-06, "loss": 0.5895, "step": 33064 }, { "epoch": 3.5183017663332623, "grad_norm": 12.0625, "learning_rate": 8.842915008527573e-06, "loss": 0.6199, "step": 33065 }, { "epoch": 3.518408171951479, "grad_norm": 12.1875, "learning_rate": 8.842879476975556e-06, "loss": 0.647, "step": 33066 }, { "epoch": 3.5185145775696958, "grad_norm": 9.484375, "learning_rate": 8.842843945423538e-06, "loss": 0.494, "step": 33067 }, { "epoch": 3.5186209831879123, "grad_norm": 10.1328125, "learning_rate": 8.84280841387152e-06, "loss": 0.478, "step": 33068 }, { "epoch": 3.518727388806129, "grad_norm": 16.40625, "learning_rate": 8.8427728823195e-06, "loss": 0.7553, "step": 33069 }, { "epoch": 3.5188337944243457, "grad_norm": 17.25, "learning_rate": 8.842737350767482e-06, "loss": 0.7864, "step": 33070 }, { "epoch": 3.5189402000425622, "grad_norm": 13.0546875, "learning_rate": 8.842701819215464e-06, "loss": 0.6762, "step": 33071 }, { "epoch": 3.5190466056607788, "grad_norm": 12.6875, "learning_rate": 8.842666287663446e-06, "loss": 0.6979, "step": 33072 }, { "epoch": 3.5191530112789957, "grad_norm": 14.34375, "learning_rate": 8.842630756111427e-06, "loss": 0.597, "step": 33073 }, { "epoch": 3.5192594168972122, "grad_norm": 11.4921875, "learning_rate": 8.842595224559409e-06, "loss": 0.6074, "step": 33074 }, { "epoch": 3.5193658225154287, "grad_norm": 14.75, "learning_rate": 8.84255969300739e-06, "loss": 0.7047, "step": 33075 }, { "epoch": 3.5194722281336457, "grad_norm": 11.90625, "learning_rate": 8.842524161455374e-06, "loss": 0.6088, "step": 33076 }, { "epoch": 3.519578633751862, "grad_norm": 13.859375, "learning_rate": 8.842488629903355e-06, "loss": 0.6747, "step": 33077 }, { "epoch": 3.5196850393700787, "grad_norm": 12.0390625, "learning_rate": 8.842453098351337e-06, "loss": 0.7547, "step": 33078 }, { "epoch": 3.5197914449882957, "grad_norm": 19.234375, "learning_rate": 8.842417566799319e-06, "loss": 0.6933, "step": 33079 }, { "epoch": 3.519897850606512, "grad_norm": 16.09375, "learning_rate": 8.8423820352473e-06, "loss": 0.6125, "step": 33080 }, { "epoch": 3.5200042562247287, "grad_norm": 14.953125, "learning_rate": 8.842346503695282e-06, "loss": 0.5504, "step": 33081 }, { "epoch": 3.520110661842945, "grad_norm": 13.34375, "learning_rate": 8.842310972143265e-06, "loss": 0.7277, "step": 33082 }, { "epoch": 3.5202170674611617, "grad_norm": 19.109375, "learning_rate": 8.842275440591247e-06, "loss": 0.6543, "step": 33083 }, { "epoch": 3.5203234730793787, "grad_norm": 12.4609375, "learning_rate": 8.842239909039228e-06, "loss": 0.6739, "step": 33084 }, { "epoch": 3.520429878697595, "grad_norm": 12.2109375, "learning_rate": 8.84220437748721e-06, "loss": 0.6921, "step": 33085 }, { "epoch": 3.5205362843158117, "grad_norm": 14.3515625, "learning_rate": 8.842168845935191e-06, "loss": 0.83, "step": 33086 }, { "epoch": 3.5206426899340286, "grad_norm": 10.3359375, "learning_rate": 8.842133314383173e-06, "loss": 0.5092, "step": 33087 }, { "epoch": 3.520749095552245, "grad_norm": 10.21875, "learning_rate": 8.842097782831155e-06, "loss": 0.6208, "step": 33088 }, { "epoch": 3.5208555011704616, "grad_norm": 12.5390625, "learning_rate": 8.842062251279136e-06, "loss": 0.6683, "step": 33089 }, { "epoch": 3.5209619067886786, "grad_norm": 12.4140625, "learning_rate": 8.842026719727118e-06, "loss": 0.7171, "step": 33090 }, { "epoch": 3.521068312406895, "grad_norm": 13.3046875, "learning_rate": 8.8419911881751e-06, "loss": 0.6216, "step": 33091 }, { "epoch": 3.5211747180251116, "grad_norm": 12.9765625, "learning_rate": 8.841955656623083e-06, "loss": 0.647, "step": 33092 }, { "epoch": 3.5212811236433286, "grad_norm": 11.3515625, "learning_rate": 8.841920125071064e-06, "loss": 0.6063, "step": 33093 }, { "epoch": 3.521387529261545, "grad_norm": 13.0078125, "learning_rate": 8.841884593519046e-06, "loss": 0.7537, "step": 33094 }, { "epoch": 3.5214939348797616, "grad_norm": 11.7421875, "learning_rate": 8.841849061967028e-06, "loss": 0.5637, "step": 33095 }, { "epoch": 3.5216003404979785, "grad_norm": 16.140625, "learning_rate": 8.84181353041501e-06, "loss": 0.8203, "step": 33096 }, { "epoch": 3.521706746116195, "grad_norm": 10.71875, "learning_rate": 8.84177799886299e-06, "loss": 0.6168, "step": 33097 }, { "epoch": 3.5218131517344116, "grad_norm": 14.6640625, "learning_rate": 8.841742467310972e-06, "loss": 0.651, "step": 33098 }, { "epoch": 3.521919557352628, "grad_norm": 13.5625, "learning_rate": 8.841706935758954e-06, "loss": 0.6438, "step": 33099 }, { "epoch": 3.522025962970845, "grad_norm": 13.8671875, "learning_rate": 8.841671404206936e-06, "loss": 0.6029, "step": 33100 }, { "epoch": 3.5221323685890615, "grad_norm": 11.390625, "learning_rate": 8.841635872654917e-06, "loss": 0.578, "step": 33101 }, { "epoch": 3.522238774207278, "grad_norm": 16.5625, "learning_rate": 8.8416003411029e-06, "loss": 0.6154, "step": 33102 }, { "epoch": 3.5223451798254946, "grad_norm": 12.296875, "learning_rate": 8.841564809550882e-06, "loss": 0.6044, "step": 33103 }, { "epoch": 3.5224515854437115, "grad_norm": 9.6328125, "learning_rate": 8.841529277998864e-06, "loss": 0.4823, "step": 33104 }, { "epoch": 3.522557991061928, "grad_norm": 11.4921875, "learning_rate": 8.841493746446845e-06, "loss": 0.5629, "step": 33105 }, { "epoch": 3.5226643966801445, "grad_norm": 11.125, "learning_rate": 8.841458214894827e-06, "loss": 0.7112, "step": 33106 }, { "epoch": 3.5227708022983615, "grad_norm": 12.9140625, "learning_rate": 8.84142268334281e-06, "loss": 0.7708, "step": 33107 }, { "epoch": 3.522877207916578, "grad_norm": 14.8203125, "learning_rate": 8.841387151790792e-06, "loss": 0.645, "step": 33108 }, { "epoch": 3.5229836135347945, "grad_norm": 15.375, "learning_rate": 8.841351620238774e-06, "loss": 0.7254, "step": 33109 }, { "epoch": 3.5230900191530115, "grad_norm": 14.4609375, "learning_rate": 8.841316088686755e-06, "loss": 0.798, "step": 33110 }, { "epoch": 3.523196424771228, "grad_norm": 11.234375, "learning_rate": 8.841280557134737e-06, "loss": 0.5993, "step": 33111 }, { "epoch": 3.5233028303894445, "grad_norm": 11.0859375, "learning_rate": 8.841245025582718e-06, "loss": 0.5757, "step": 33112 }, { "epoch": 3.5234092360076614, "grad_norm": 13.3359375, "learning_rate": 8.8412094940307e-06, "loss": 0.6909, "step": 33113 }, { "epoch": 3.523515641625878, "grad_norm": 14.2734375, "learning_rate": 8.841173962478682e-06, "loss": 0.6373, "step": 33114 }, { "epoch": 3.5236220472440944, "grad_norm": 12.9609375, "learning_rate": 8.841138430926663e-06, "loss": 0.5544, "step": 33115 }, { "epoch": 3.5237284528623114, "grad_norm": 10.21875, "learning_rate": 8.841102899374645e-06, "loss": 0.663, "step": 33116 }, { "epoch": 3.523834858480528, "grad_norm": 11.625, "learning_rate": 8.841067367822628e-06, "loss": 0.5031, "step": 33117 }, { "epoch": 3.5239412640987444, "grad_norm": 14.90625, "learning_rate": 8.84103183627061e-06, "loss": 0.7554, "step": 33118 }, { "epoch": 3.524047669716961, "grad_norm": 12.0859375, "learning_rate": 8.840996304718591e-06, "loss": 0.5483, "step": 33119 }, { "epoch": 3.5241540753351774, "grad_norm": 12.3046875, "learning_rate": 8.840960773166573e-06, "loss": 0.5685, "step": 33120 }, { "epoch": 3.5242604809533944, "grad_norm": 16.8125, "learning_rate": 8.840925241614554e-06, "loss": 0.596, "step": 33121 }, { "epoch": 3.524366886571611, "grad_norm": 19.609375, "learning_rate": 8.840889710062536e-06, "loss": 0.743, "step": 33122 }, { "epoch": 3.5244732921898274, "grad_norm": 11.453125, "learning_rate": 8.840854178510518e-06, "loss": 0.5872, "step": 33123 }, { "epoch": 3.5245796978080444, "grad_norm": 11.25, "learning_rate": 8.8408186469585e-06, "loss": 0.638, "step": 33124 }, { "epoch": 3.524686103426261, "grad_norm": 13.265625, "learning_rate": 8.840783115406481e-06, "loss": 0.7155, "step": 33125 }, { "epoch": 3.5247925090444774, "grad_norm": 13.2421875, "learning_rate": 8.840747583854463e-06, "loss": 0.7238, "step": 33126 }, { "epoch": 3.5248989146626943, "grad_norm": 11.28125, "learning_rate": 8.840712052302444e-06, "loss": 0.6048, "step": 33127 }, { "epoch": 3.525005320280911, "grad_norm": 12.75, "learning_rate": 8.840676520750427e-06, "loss": 0.646, "step": 33128 }, { "epoch": 3.5251117258991274, "grad_norm": 17.953125, "learning_rate": 8.840640989198409e-06, "loss": 0.775, "step": 33129 }, { "epoch": 3.5252181315173443, "grad_norm": 10.921875, "learning_rate": 8.84060545764639e-06, "loss": 0.5391, "step": 33130 }, { "epoch": 3.525324537135561, "grad_norm": 17.265625, "learning_rate": 8.840569926094372e-06, "loss": 0.6544, "step": 33131 }, { "epoch": 3.5254309427537773, "grad_norm": 16.21875, "learning_rate": 8.840534394542354e-06, "loss": 0.6381, "step": 33132 }, { "epoch": 3.5255373483719943, "grad_norm": 19.078125, "learning_rate": 8.840498862990337e-06, "loss": 0.937, "step": 33133 }, { "epoch": 3.525643753990211, "grad_norm": 16.34375, "learning_rate": 8.840463331438319e-06, "loss": 0.7541, "step": 33134 }, { "epoch": 3.5257501596084273, "grad_norm": 13.7421875, "learning_rate": 8.8404277998863e-06, "loss": 0.6036, "step": 33135 }, { "epoch": 3.525856565226644, "grad_norm": 14.9609375, "learning_rate": 8.840392268334282e-06, "loss": 0.6061, "step": 33136 }, { "epoch": 3.5259629708448608, "grad_norm": 12.984375, "learning_rate": 8.840356736782264e-06, "loss": 0.6057, "step": 33137 }, { "epoch": 3.5260693764630773, "grad_norm": 15.234375, "learning_rate": 8.840321205230245e-06, "loss": 0.5425, "step": 33138 }, { "epoch": 3.526175782081294, "grad_norm": 12.1875, "learning_rate": 8.840285673678227e-06, "loss": 0.6399, "step": 33139 }, { "epoch": 3.5262821876995103, "grad_norm": 18.296875, "learning_rate": 8.840250142126208e-06, "loss": 0.8719, "step": 33140 }, { "epoch": 3.5263885933177272, "grad_norm": 11.84375, "learning_rate": 8.84021461057419e-06, "loss": 0.5928, "step": 33141 }, { "epoch": 3.5264949989359438, "grad_norm": 12.75, "learning_rate": 8.840179079022172e-06, "loss": 0.5835, "step": 33142 }, { "epoch": 3.5266014045541603, "grad_norm": 12.7265625, "learning_rate": 8.840143547470155e-06, "loss": 0.6128, "step": 33143 }, { "epoch": 3.5267078101723772, "grad_norm": 13.3828125, "learning_rate": 8.840108015918137e-06, "loss": 0.6487, "step": 33144 }, { "epoch": 3.5268142157905937, "grad_norm": 14.0546875, "learning_rate": 8.840072484366118e-06, "loss": 0.7324, "step": 33145 }, { "epoch": 3.5269206214088102, "grad_norm": 14.671875, "learning_rate": 8.8400369528141e-06, "loss": 0.6702, "step": 33146 }, { "epoch": 3.527027027027027, "grad_norm": 14.1875, "learning_rate": 8.840001421262081e-06, "loss": 0.6866, "step": 33147 }, { "epoch": 3.5271334326452437, "grad_norm": 15.453125, "learning_rate": 8.839965889710063e-06, "loss": 0.713, "step": 33148 }, { "epoch": 3.52723983826346, "grad_norm": 14.453125, "learning_rate": 8.839930358158045e-06, "loss": 0.607, "step": 33149 }, { "epoch": 3.527346243881677, "grad_norm": 13.6328125, "learning_rate": 8.839894826606026e-06, "loss": 0.6849, "step": 33150 }, { "epoch": 3.5274526494998937, "grad_norm": 15.859375, "learning_rate": 8.839859295054008e-06, "loss": 0.7108, "step": 33151 }, { "epoch": 3.52755905511811, "grad_norm": 15.46875, "learning_rate": 8.83982376350199e-06, "loss": 0.7125, "step": 33152 }, { "epoch": 3.527665460736327, "grad_norm": 9.625, "learning_rate": 8.839788231949973e-06, "loss": 0.6096, "step": 33153 }, { "epoch": 3.5277718663545437, "grad_norm": 12.4140625, "learning_rate": 8.839752700397954e-06, "loss": 0.6116, "step": 33154 }, { "epoch": 3.52787827197276, "grad_norm": 11.15625, "learning_rate": 8.839717168845936e-06, "loss": 0.54, "step": 33155 }, { "epoch": 3.5279846775909767, "grad_norm": 12.65625, "learning_rate": 8.839681637293917e-06, "loss": 0.7154, "step": 33156 }, { "epoch": 3.528091083209193, "grad_norm": 12.4375, "learning_rate": 8.839646105741899e-06, "loss": 0.703, "step": 33157 }, { "epoch": 3.52819748882741, "grad_norm": 13.9609375, "learning_rate": 8.839610574189882e-06, "loss": 0.6269, "step": 33158 }, { "epoch": 3.5283038944456266, "grad_norm": 18.578125, "learning_rate": 8.839575042637864e-06, "loss": 0.7098, "step": 33159 }, { "epoch": 3.528410300063843, "grad_norm": 11.6953125, "learning_rate": 8.839539511085846e-06, "loss": 0.4715, "step": 33160 }, { "epoch": 3.52851670568206, "grad_norm": 8.703125, "learning_rate": 8.839503979533827e-06, "loss": 0.4025, "step": 33161 }, { "epoch": 3.5286231113002766, "grad_norm": 13.9296875, "learning_rate": 8.839468447981809e-06, "loss": 0.6205, "step": 33162 }, { "epoch": 3.528729516918493, "grad_norm": 12.109375, "learning_rate": 8.83943291642979e-06, "loss": 0.6245, "step": 33163 }, { "epoch": 3.52883592253671, "grad_norm": 12.2890625, "learning_rate": 8.839397384877772e-06, "loss": 0.5866, "step": 33164 }, { "epoch": 3.5289423281549266, "grad_norm": 13.5234375, "learning_rate": 8.839361853325754e-06, "loss": 0.7734, "step": 33165 }, { "epoch": 3.529048733773143, "grad_norm": 11.3046875, "learning_rate": 8.839326321773735e-06, "loss": 0.658, "step": 33166 }, { "epoch": 3.52915513939136, "grad_norm": 8.9921875, "learning_rate": 8.839290790221717e-06, "loss": 0.5931, "step": 33167 }, { "epoch": 3.5292615450095766, "grad_norm": 12.1875, "learning_rate": 8.839255258669698e-06, "loss": 0.5551, "step": 33168 }, { "epoch": 3.529367950627793, "grad_norm": 10.265625, "learning_rate": 8.839219727117682e-06, "loss": 0.5896, "step": 33169 }, { "epoch": 3.52947435624601, "grad_norm": 10.984375, "learning_rate": 8.839184195565663e-06, "loss": 0.63, "step": 33170 }, { "epoch": 3.5295807618642265, "grad_norm": 13.046875, "learning_rate": 8.839148664013645e-06, "loss": 0.6549, "step": 33171 }, { "epoch": 3.529687167482443, "grad_norm": 15.5703125, "learning_rate": 8.839113132461627e-06, "loss": 0.7665, "step": 33172 }, { "epoch": 3.5297935731006596, "grad_norm": 11.2421875, "learning_rate": 8.839077600909608e-06, "loss": 0.6575, "step": 33173 }, { "epoch": 3.5298999787188765, "grad_norm": 14.8828125, "learning_rate": 8.839042069357591e-06, "loss": 0.6475, "step": 33174 }, { "epoch": 3.530006384337093, "grad_norm": 16.71875, "learning_rate": 8.839006537805573e-06, "loss": 0.8241, "step": 33175 }, { "epoch": 3.5301127899553095, "grad_norm": 12.7734375, "learning_rate": 8.838971006253553e-06, "loss": 0.6773, "step": 33176 }, { "epoch": 3.530219195573526, "grad_norm": 18.703125, "learning_rate": 8.838935474701535e-06, "loss": 0.7199, "step": 33177 }, { "epoch": 3.530325601191743, "grad_norm": 12.8515625, "learning_rate": 8.838899943149516e-06, "loss": 0.6237, "step": 33178 }, { "epoch": 3.5304320068099595, "grad_norm": 13.65625, "learning_rate": 8.8388644115975e-06, "loss": 0.7435, "step": 33179 }, { "epoch": 3.530538412428176, "grad_norm": 13.1171875, "learning_rate": 8.838828880045481e-06, "loss": 0.7044, "step": 33180 }, { "epoch": 3.530644818046393, "grad_norm": 12.8125, "learning_rate": 8.838793348493463e-06, "loss": 0.6811, "step": 33181 }, { "epoch": 3.5307512236646095, "grad_norm": 10.953125, "learning_rate": 8.838757816941444e-06, "loss": 0.57, "step": 33182 }, { "epoch": 3.530857629282826, "grad_norm": 12.2265625, "learning_rate": 8.838722285389426e-06, "loss": 0.6746, "step": 33183 }, { "epoch": 3.530964034901043, "grad_norm": 15.875, "learning_rate": 8.83868675383741e-06, "loss": 0.6499, "step": 33184 }, { "epoch": 3.5310704405192594, "grad_norm": 12.328125, "learning_rate": 8.838651222285391e-06, "loss": 0.5895, "step": 33185 }, { "epoch": 3.531176846137476, "grad_norm": 11.7109375, "learning_rate": 8.838615690733372e-06, "loss": 0.6283, "step": 33186 }, { "epoch": 3.531283251755693, "grad_norm": 11.9140625, "learning_rate": 8.838580159181354e-06, "loss": 0.6194, "step": 33187 }, { "epoch": 3.5313896573739094, "grad_norm": 11.390625, "learning_rate": 8.838544627629336e-06, "loss": 0.6696, "step": 33188 }, { "epoch": 3.531496062992126, "grad_norm": 16.625, "learning_rate": 8.838509096077317e-06, "loss": 0.7298, "step": 33189 }, { "epoch": 3.531602468610343, "grad_norm": 12.734375, "learning_rate": 8.838473564525299e-06, "loss": 0.6819, "step": 33190 }, { "epoch": 3.5317088742285594, "grad_norm": 9.8671875, "learning_rate": 8.83843803297328e-06, "loss": 0.5523, "step": 33191 }, { "epoch": 3.531815279846776, "grad_norm": 13.921875, "learning_rate": 8.838402501421262e-06, "loss": 0.6165, "step": 33192 }, { "epoch": 3.5319216854649924, "grad_norm": 10.078125, "learning_rate": 8.838366969869244e-06, "loss": 0.5905, "step": 33193 }, { "epoch": 3.532028091083209, "grad_norm": 18.640625, "learning_rate": 8.838331438317227e-06, "loss": 0.9169, "step": 33194 }, { "epoch": 3.532134496701426, "grad_norm": 14.3046875, "learning_rate": 8.838295906765209e-06, "loss": 0.5588, "step": 33195 }, { "epoch": 3.5322409023196424, "grad_norm": 16.890625, "learning_rate": 8.83826037521319e-06, "loss": 0.6564, "step": 33196 }, { "epoch": 3.532347307937859, "grad_norm": 10.6328125, "learning_rate": 8.838224843661172e-06, "loss": 0.4505, "step": 33197 }, { "epoch": 3.532453713556076, "grad_norm": 13.625, "learning_rate": 8.838189312109153e-06, "loss": 0.6565, "step": 33198 }, { "epoch": 3.5325601191742924, "grad_norm": 20.4375, "learning_rate": 8.838153780557135e-06, "loss": 0.7596, "step": 33199 }, { "epoch": 3.532666524792509, "grad_norm": 13.859375, "learning_rate": 8.838118249005118e-06, "loss": 0.811, "step": 33200 }, { "epoch": 3.532772930410726, "grad_norm": 15.296875, "learning_rate": 8.8380827174531e-06, "loss": 0.8233, "step": 33201 }, { "epoch": 3.5328793360289423, "grad_norm": 14.78125, "learning_rate": 8.838047185901082e-06, "loss": 0.6619, "step": 33202 }, { "epoch": 3.532985741647159, "grad_norm": 14.609375, "learning_rate": 8.838011654349063e-06, "loss": 0.6132, "step": 33203 }, { "epoch": 3.533092147265376, "grad_norm": 14.3828125, "learning_rate": 8.837976122797045e-06, "loss": 0.5902, "step": 33204 }, { "epoch": 3.5331985528835923, "grad_norm": 13.3671875, "learning_rate": 8.837940591245026e-06, "loss": 0.611, "step": 33205 }, { "epoch": 3.533304958501809, "grad_norm": 17.109375, "learning_rate": 8.837905059693008e-06, "loss": 0.78, "step": 33206 }, { "epoch": 3.5334113641200258, "grad_norm": 14.6328125, "learning_rate": 8.83786952814099e-06, "loss": 0.7059, "step": 33207 }, { "epoch": 3.5335177697382423, "grad_norm": 11.0234375, "learning_rate": 8.837833996588971e-06, "loss": 0.6491, "step": 33208 }, { "epoch": 3.533624175356459, "grad_norm": 13.0703125, "learning_rate": 8.837798465036953e-06, "loss": 0.5884, "step": 33209 }, { "epoch": 3.5337305809746757, "grad_norm": 9.59375, "learning_rate": 8.837762933484936e-06, "loss": 0.5824, "step": 33210 }, { "epoch": 3.5338369865928922, "grad_norm": 14.4140625, "learning_rate": 8.837727401932918e-06, "loss": 0.5321, "step": 33211 }, { "epoch": 3.5339433922111088, "grad_norm": 12.296875, "learning_rate": 8.8376918703809e-06, "loss": 0.6456, "step": 33212 }, { "epoch": 3.5340497978293253, "grad_norm": 11.359375, "learning_rate": 8.837656338828881e-06, "loss": 0.669, "step": 33213 }, { "epoch": 3.5341562034475418, "grad_norm": 15.03125, "learning_rate": 8.837620807276863e-06, "loss": 0.625, "step": 33214 }, { "epoch": 3.5342626090657587, "grad_norm": 9.0859375, "learning_rate": 8.837585275724844e-06, "loss": 0.4716, "step": 33215 }, { "epoch": 3.5343690146839752, "grad_norm": 15.1171875, "learning_rate": 8.837549744172826e-06, "loss": 0.7283, "step": 33216 }, { "epoch": 3.5344754203021918, "grad_norm": 12.5859375, "learning_rate": 8.837514212620807e-06, "loss": 0.6412, "step": 33217 }, { "epoch": 3.5345818259204087, "grad_norm": 11.5625, "learning_rate": 8.837478681068789e-06, "loss": 0.5176, "step": 33218 }, { "epoch": 3.534688231538625, "grad_norm": 12.7890625, "learning_rate": 8.83744314951677e-06, "loss": 0.6383, "step": 33219 }, { "epoch": 3.5347946371568417, "grad_norm": 16.390625, "learning_rate": 8.837407617964754e-06, "loss": 0.8125, "step": 33220 }, { "epoch": 3.5349010427750587, "grad_norm": 15.265625, "learning_rate": 8.837372086412735e-06, "loss": 0.7051, "step": 33221 }, { "epoch": 3.535007448393275, "grad_norm": 15.3515625, "learning_rate": 8.837336554860717e-06, "loss": 0.8117, "step": 33222 }, { "epoch": 3.5351138540114917, "grad_norm": 13.796875, "learning_rate": 8.837301023308699e-06, "loss": 0.7105, "step": 33223 }, { "epoch": 3.5352202596297087, "grad_norm": 15.4453125, "learning_rate": 8.83726549175668e-06, "loss": 0.7553, "step": 33224 }, { "epoch": 3.535326665247925, "grad_norm": 13.3515625, "learning_rate": 8.837229960204664e-06, "loss": 0.6853, "step": 33225 }, { "epoch": 3.5354330708661417, "grad_norm": 11.6875, "learning_rate": 8.837194428652645e-06, "loss": 0.5919, "step": 33226 }, { "epoch": 3.5355394764843586, "grad_norm": 10.2578125, "learning_rate": 8.837158897100627e-06, "loss": 0.5159, "step": 33227 }, { "epoch": 3.535645882102575, "grad_norm": 12.9609375, "learning_rate": 8.837123365548608e-06, "loss": 0.6982, "step": 33228 }, { "epoch": 3.5357522877207916, "grad_norm": 12.0, "learning_rate": 8.83708783399659e-06, "loss": 0.6612, "step": 33229 }, { "epoch": 3.535858693339008, "grad_norm": 12.0859375, "learning_rate": 8.837052302444572e-06, "loss": 0.5772, "step": 33230 }, { "epoch": 3.535965098957225, "grad_norm": 16.1875, "learning_rate": 8.837016770892553e-06, "loss": 0.7875, "step": 33231 }, { "epoch": 3.5360715045754416, "grad_norm": 12.1171875, "learning_rate": 8.836981239340535e-06, "loss": 0.6395, "step": 33232 }, { "epoch": 3.536177910193658, "grad_norm": 15.96875, "learning_rate": 8.836945707788516e-06, "loss": 0.7828, "step": 33233 }, { "epoch": 3.5362843158118746, "grad_norm": 11.59375, "learning_rate": 8.836910176236498e-06, "loss": 0.6196, "step": 33234 }, { "epoch": 3.5363907214300916, "grad_norm": 12.515625, "learning_rate": 8.836874644684481e-06, "loss": 0.7213, "step": 33235 }, { "epoch": 3.536497127048308, "grad_norm": 13.453125, "learning_rate": 8.836839113132463e-06, "loss": 0.6363, "step": 33236 }, { "epoch": 3.5366035326665246, "grad_norm": 13.75, "learning_rate": 8.836803581580445e-06, "loss": 0.7461, "step": 33237 }, { "epoch": 3.5367099382847416, "grad_norm": 11.9609375, "learning_rate": 8.836768050028426e-06, "loss": 0.6362, "step": 33238 }, { "epoch": 3.536816343902958, "grad_norm": 10.4765625, "learning_rate": 8.836732518476408e-06, "loss": 0.5372, "step": 33239 }, { "epoch": 3.5369227495211746, "grad_norm": 15.3984375, "learning_rate": 8.83669698692439e-06, "loss": 0.7326, "step": 33240 }, { "epoch": 3.5370291551393915, "grad_norm": 14.140625, "learning_rate": 8.836661455372371e-06, "loss": 0.5697, "step": 33241 }, { "epoch": 3.537135560757608, "grad_norm": 12.25, "learning_rate": 8.836625923820353e-06, "loss": 0.707, "step": 33242 }, { "epoch": 3.5372419663758246, "grad_norm": 14.9921875, "learning_rate": 8.836590392268334e-06, "loss": 0.7441, "step": 33243 }, { "epoch": 3.5373483719940415, "grad_norm": 11.96875, "learning_rate": 8.836554860716316e-06, "loss": 0.5477, "step": 33244 }, { "epoch": 3.537454777612258, "grad_norm": 13.9765625, "learning_rate": 8.836519329164297e-06, "loss": 0.7504, "step": 33245 }, { "epoch": 3.5375611832304745, "grad_norm": 15.1875, "learning_rate": 8.83648379761228e-06, "loss": 0.7013, "step": 33246 }, { "epoch": 3.5376675888486915, "grad_norm": 17.671875, "learning_rate": 8.836448266060262e-06, "loss": 0.6414, "step": 33247 }, { "epoch": 3.537773994466908, "grad_norm": 12.4453125, "learning_rate": 8.836412734508244e-06, "loss": 0.6365, "step": 33248 }, { "epoch": 3.5378804000851245, "grad_norm": 10.53125, "learning_rate": 8.836377202956226e-06, "loss": 0.7355, "step": 33249 }, { "epoch": 3.537986805703341, "grad_norm": 12.8984375, "learning_rate": 8.836341671404207e-06, "loss": 0.7089, "step": 33250 }, { "epoch": 3.5380932113215575, "grad_norm": 12.5859375, "learning_rate": 8.83630613985219e-06, "loss": 0.6166, "step": 33251 }, { "epoch": 3.5381996169397745, "grad_norm": 15.5625, "learning_rate": 8.836270608300172e-06, "loss": 0.7486, "step": 33252 }, { "epoch": 3.538306022557991, "grad_norm": 13.9453125, "learning_rate": 8.836235076748154e-06, "loss": 0.7654, "step": 33253 }, { "epoch": 3.5384124281762075, "grad_norm": 16.265625, "learning_rate": 8.836199545196135e-06, "loss": 0.7524, "step": 33254 }, { "epoch": 3.5385188337944244, "grad_norm": 13.59375, "learning_rate": 8.836164013644117e-06, "loss": 0.5602, "step": 33255 }, { "epoch": 3.538625239412641, "grad_norm": 15.0625, "learning_rate": 8.836128482092098e-06, "loss": 0.6715, "step": 33256 }, { "epoch": 3.5387316450308575, "grad_norm": 13.8828125, "learning_rate": 8.83609295054008e-06, "loss": 0.6289, "step": 33257 }, { "epoch": 3.5388380506490744, "grad_norm": 16.875, "learning_rate": 8.836057418988062e-06, "loss": 0.6561, "step": 33258 }, { "epoch": 3.538944456267291, "grad_norm": 13.5234375, "learning_rate": 8.836021887436043e-06, "loss": 0.7583, "step": 33259 }, { "epoch": 3.5390508618855074, "grad_norm": 10.875, "learning_rate": 8.835986355884025e-06, "loss": 0.6072, "step": 33260 }, { "epoch": 3.5391572675037244, "grad_norm": 11.3984375, "learning_rate": 8.835950824332008e-06, "loss": 0.6289, "step": 33261 }, { "epoch": 3.539263673121941, "grad_norm": 12.515625, "learning_rate": 8.83591529277999e-06, "loss": 0.585, "step": 33262 }, { "epoch": 3.5393700787401574, "grad_norm": 11.8125, "learning_rate": 8.835879761227971e-06, "loss": 0.5509, "step": 33263 }, { "epoch": 3.5394764843583744, "grad_norm": 15.703125, "learning_rate": 8.835844229675953e-06, "loss": 0.8563, "step": 33264 }, { "epoch": 3.539582889976591, "grad_norm": 10.7890625, "learning_rate": 8.835808698123935e-06, "loss": 0.7036, "step": 33265 }, { "epoch": 3.5396892955948074, "grad_norm": 14.5, "learning_rate": 8.835773166571916e-06, "loss": 0.6135, "step": 33266 }, { "epoch": 3.539795701213024, "grad_norm": 14.0, "learning_rate": 8.835737635019898e-06, "loss": 0.6268, "step": 33267 }, { "epoch": 3.539902106831241, "grad_norm": 13.8046875, "learning_rate": 8.83570210346788e-06, "loss": 0.7155, "step": 33268 }, { "epoch": 3.5400085124494574, "grad_norm": 15.578125, "learning_rate": 8.835666571915861e-06, "loss": 0.7781, "step": 33269 }, { "epoch": 3.540114918067674, "grad_norm": 13.7890625, "learning_rate": 8.835631040363843e-06, "loss": 0.6927, "step": 33270 }, { "epoch": 3.5402213236858904, "grad_norm": 14.328125, "learning_rate": 8.835595508811826e-06, "loss": 0.5786, "step": 33271 }, { "epoch": 3.5403277293041073, "grad_norm": 15.28125, "learning_rate": 8.835559977259808e-06, "loss": 0.7364, "step": 33272 }, { "epoch": 3.540434134922324, "grad_norm": 18.796875, "learning_rate": 8.83552444570779e-06, "loss": 0.7942, "step": 33273 }, { "epoch": 3.5405405405405403, "grad_norm": 12.140625, "learning_rate": 8.83548891415577e-06, "loss": 0.7205, "step": 33274 }, { "epoch": 3.5406469461587573, "grad_norm": 16.796875, "learning_rate": 8.835453382603752e-06, "loss": 0.8095, "step": 33275 }, { "epoch": 3.540753351776974, "grad_norm": 15.078125, "learning_rate": 8.835417851051736e-06, "loss": 0.7606, "step": 33276 }, { "epoch": 3.5408597573951903, "grad_norm": 11.2109375, "learning_rate": 8.835382319499717e-06, "loss": 0.5501, "step": 33277 }, { "epoch": 3.5409661630134073, "grad_norm": 10.9296875, "learning_rate": 8.835346787947699e-06, "loss": 0.5739, "step": 33278 }, { "epoch": 3.541072568631624, "grad_norm": 14.234375, "learning_rate": 8.83531125639568e-06, "loss": 0.6946, "step": 33279 }, { "epoch": 3.5411789742498403, "grad_norm": 12.4765625, "learning_rate": 8.835275724843662e-06, "loss": 0.5629, "step": 33280 }, { "epoch": 3.5412853798680572, "grad_norm": 13.7734375, "learning_rate": 8.835240193291644e-06, "loss": 0.6069, "step": 33281 }, { "epoch": 3.5413917854862738, "grad_norm": 15.90625, "learning_rate": 8.835204661739625e-06, "loss": 0.8549, "step": 33282 }, { "epoch": 3.5414981911044903, "grad_norm": 15.7421875, "learning_rate": 8.835169130187607e-06, "loss": 0.5804, "step": 33283 }, { "epoch": 3.541604596722707, "grad_norm": 12.4609375, "learning_rate": 8.835133598635589e-06, "loss": 0.7408, "step": 33284 }, { "epoch": 3.5417110023409237, "grad_norm": 18.046875, "learning_rate": 8.83509806708357e-06, "loss": 0.7094, "step": 33285 }, { "epoch": 3.5418174079591402, "grad_norm": 13.5625, "learning_rate": 8.835062535531552e-06, "loss": 0.6642, "step": 33286 }, { "epoch": 3.5419238135773567, "grad_norm": 16.109375, "learning_rate": 8.835027003979535e-06, "loss": 0.7743, "step": 33287 }, { "epoch": 3.5420302191955733, "grad_norm": 12.5703125, "learning_rate": 8.834991472427517e-06, "loss": 0.6116, "step": 33288 }, { "epoch": 3.54213662481379, "grad_norm": 11.515625, "learning_rate": 8.834955940875498e-06, "loss": 0.6164, "step": 33289 }, { "epoch": 3.5422430304320067, "grad_norm": 10.703125, "learning_rate": 8.83492040932348e-06, "loss": 0.5966, "step": 33290 }, { "epoch": 3.5423494360502232, "grad_norm": 12.6171875, "learning_rate": 8.834884877771462e-06, "loss": 0.7042, "step": 33291 }, { "epoch": 3.54245584166844, "grad_norm": 14.453125, "learning_rate": 8.834849346219445e-06, "loss": 0.6279, "step": 33292 }, { "epoch": 3.5425622472866567, "grad_norm": 11.4375, "learning_rate": 8.834813814667426e-06, "loss": 0.6374, "step": 33293 }, { "epoch": 3.542668652904873, "grad_norm": 13.859375, "learning_rate": 8.834778283115408e-06, "loss": 0.6266, "step": 33294 }, { "epoch": 3.54277505852309, "grad_norm": 14.9765625, "learning_rate": 8.83474275156339e-06, "loss": 0.7784, "step": 33295 }, { "epoch": 3.5428814641413067, "grad_norm": 13.90625, "learning_rate": 8.83470722001137e-06, "loss": 0.705, "step": 33296 }, { "epoch": 3.542987869759523, "grad_norm": 14.359375, "learning_rate": 8.834671688459353e-06, "loss": 0.7026, "step": 33297 }, { "epoch": 3.54309427537774, "grad_norm": 16.390625, "learning_rate": 8.834636156907334e-06, "loss": 0.6721, "step": 33298 }, { "epoch": 3.5432006809959566, "grad_norm": 13.0078125, "learning_rate": 8.834600625355316e-06, "loss": 0.6065, "step": 33299 }, { "epoch": 3.543307086614173, "grad_norm": 20.875, "learning_rate": 8.834565093803298e-06, "loss": 0.8592, "step": 33300 }, { "epoch": 3.54341349223239, "grad_norm": 16.65625, "learning_rate": 8.83452956225128e-06, "loss": 0.6562, "step": 33301 }, { "epoch": 3.5435198978506066, "grad_norm": 12.7890625, "learning_rate": 8.834494030699263e-06, "loss": 0.6367, "step": 33302 }, { "epoch": 3.543626303468823, "grad_norm": 15.6171875, "learning_rate": 8.834458499147244e-06, "loss": 0.6615, "step": 33303 }, { "epoch": 3.5437327090870396, "grad_norm": 12.125, "learning_rate": 8.834422967595226e-06, "loss": 0.6272, "step": 33304 }, { "epoch": 3.5438391147052566, "grad_norm": 12.1328125, "learning_rate": 8.834387436043207e-06, "loss": 0.5438, "step": 33305 }, { "epoch": 3.543945520323473, "grad_norm": 14.78125, "learning_rate": 8.834351904491189e-06, "loss": 0.6433, "step": 33306 }, { "epoch": 3.5440519259416896, "grad_norm": 14.1484375, "learning_rate": 8.83431637293917e-06, "loss": 0.6777, "step": 33307 }, { "epoch": 3.544158331559906, "grad_norm": 18.265625, "learning_rate": 8.834280841387152e-06, "loss": 0.8899, "step": 33308 }, { "epoch": 3.544264737178123, "grad_norm": 13.140625, "learning_rate": 8.834245309835134e-06, "loss": 0.6872, "step": 33309 }, { "epoch": 3.5443711427963396, "grad_norm": 10.953125, "learning_rate": 8.834209778283115e-06, "loss": 0.4906, "step": 33310 }, { "epoch": 3.544477548414556, "grad_norm": 13.2265625, "learning_rate": 8.834174246731097e-06, "loss": 0.6378, "step": 33311 }, { "epoch": 3.544583954032773, "grad_norm": 14.203125, "learning_rate": 8.83413871517908e-06, "loss": 0.7997, "step": 33312 }, { "epoch": 3.5446903596509896, "grad_norm": 16.03125, "learning_rate": 8.834103183627062e-06, "loss": 0.6686, "step": 33313 }, { "epoch": 3.544796765269206, "grad_norm": 19.03125, "learning_rate": 8.834067652075044e-06, "loss": 0.8422, "step": 33314 }, { "epoch": 3.544903170887423, "grad_norm": 13.5625, "learning_rate": 8.834032120523025e-06, "loss": 0.6762, "step": 33315 }, { "epoch": 3.5450095765056395, "grad_norm": 14.9765625, "learning_rate": 8.833996588971007e-06, "loss": 0.6455, "step": 33316 }, { "epoch": 3.545115982123856, "grad_norm": 13.671875, "learning_rate": 8.833961057418988e-06, "loss": 0.6687, "step": 33317 }, { "epoch": 3.545222387742073, "grad_norm": 12.9765625, "learning_rate": 8.833925525866972e-06, "loss": 0.6869, "step": 33318 }, { "epoch": 3.5453287933602895, "grad_norm": 10.203125, "learning_rate": 8.833889994314953e-06, "loss": 0.5956, "step": 33319 }, { "epoch": 3.545435198978506, "grad_norm": 15.7421875, "learning_rate": 8.833854462762935e-06, "loss": 0.855, "step": 33320 }, { "epoch": 3.545541604596723, "grad_norm": 14.3671875, "learning_rate": 8.833818931210916e-06, "loss": 0.7608, "step": 33321 }, { "epoch": 3.5456480102149395, "grad_norm": 12.2421875, "learning_rate": 8.833783399658898e-06, "loss": 0.5913, "step": 33322 }, { "epoch": 3.545754415833156, "grad_norm": 12.625, "learning_rate": 8.83374786810688e-06, "loss": 0.6186, "step": 33323 }, { "epoch": 3.5458608214513725, "grad_norm": 8.953125, "learning_rate": 8.833712336554861e-06, "loss": 0.4593, "step": 33324 }, { "epoch": 3.545967227069589, "grad_norm": 12.8515625, "learning_rate": 8.833676805002843e-06, "loss": 0.6606, "step": 33325 }, { "epoch": 3.546073632687806, "grad_norm": 13.4453125, "learning_rate": 8.833641273450825e-06, "loss": 0.6776, "step": 33326 }, { "epoch": 3.5461800383060225, "grad_norm": 12.984375, "learning_rate": 8.833605741898806e-06, "loss": 0.7689, "step": 33327 }, { "epoch": 3.546286443924239, "grad_norm": 12.4296875, "learning_rate": 8.83357021034679e-06, "loss": 0.6646, "step": 33328 }, { "epoch": 3.546392849542456, "grad_norm": 13.359375, "learning_rate": 8.833534678794771e-06, "loss": 0.7288, "step": 33329 }, { "epoch": 3.5464992551606724, "grad_norm": 13.484375, "learning_rate": 8.833499147242753e-06, "loss": 0.7685, "step": 33330 }, { "epoch": 3.546605660778889, "grad_norm": 13.609375, "learning_rate": 8.833463615690734e-06, "loss": 0.6841, "step": 33331 }, { "epoch": 3.546712066397106, "grad_norm": 10.453125, "learning_rate": 8.833428084138716e-06, "loss": 0.5609, "step": 33332 }, { "epoch": 3.5468184720153224, "grad_norm": 12.234375, "learning_rate": 8.833392552586697e-06, "loss": 0.6906, "step": 33333 }, { "epoch": 3.546924877633539, "grad_norm": 13.453125, "learning_rate": 8.833357021034679e-06, "loss": 0.737, "step": 33334 }, { "epoch": 3.547031283251756, "grad_norm": 12.8359375, "learning_rate": 8.83332148948266e-06, "loss": 0.5576, "step": 33335 }, { "epoch": 3.5471376888699724, "grad_norm": 14.46875, "learning_rate": 8.833285957930642e-06, "loss": 0.5763, "step": 33336 }, { "epoch": 3.547244094488189, "grad_norm": 11.5234375, "learning_rate": 8.833250426378624e-06, "loss": 0.6301, "step": 33337 }, { "epoch": 3.547350500106406, "grad_norm": 16.90625, "learning_rate": 8.833214894826607e-06, "loss": 0.7971, "step": 33338 }, { "epoch": 3.5474569057246224, "grad_norm": 11.9453125, "learning_rate": 8.833179363274589e-06, "loss": 0.7485, "step": 33339 }, { "epoch": 3.547563311342839, "grad_norm": 16.890625, "learning_rate": 8.83314383172257e-06, "loss": 0.7701, "step": 33340 }, { "epoch": 3.547669716961056, "grad_norm": 15.421875, "learning_rate": 8.833108300170552e-06, "loss": 0.7967, "step": 33341 }, { "epoch": 3.5477761225792723, "grad_norm": 12.2734375, "learning_rate": 8.833072768618534e-06, "loss": 0.6897, "step": 33342 }, { "epoch": 3.547882528197489, "grad_norm": 11.96875, "learning_rate": 8.833037237066517e-06, "loss": 0.594, "step": 33343 }, { "epoch": 3.5479889338157053, "grad_norm": 13.203125, "learning_rate": 8.833001705514499e-06, "loss": 0.761, "step": 33344 }, { "epoch": 3.548095339433922, "grad_norm": 9.59375, "learning_rate": 8.83296617396248e-06, "loss": 0.5001, "step": 33345 }, { "epoch": 3.548201745052139, "grad_norm": 16.375, "learning_rate": 8.832930642410462e-06, "loss": 0.6814, "step": 33346 }, { "epoch": 3.5483081506703553, "grad_norm": 13.078125, "learning_rate": 8.832895110858443e-06, "loss": 0.7398, "step": 33347 }, { "epoch": 3.548414556288572, "grad_norm": 11.546875, "learning_rate": 8.832859579306425e-06, "loss": 0.6264, "step": 33348 }, { "epoch": 3.548520961906789, "grad_norm": 12.9921875, "learning_rate": 8.832824047754407e-06, "loss": 0.5436, "step": 33349 }, { "epoch": 3.5486273675250053, "grad_norm": 14.625, "learning_rate": 8.832788516202388e-06, "loss": 0.6985, "step": 33350 }, { "epoch": 3.548733773143222, "grad_norm": 11.34375, "learning_rate": 8.83275298465037e-06, "loss": 0.5068, "step": 33351 }, { "epoch": 3.5488401787614388, "grad_norm": 14.0703125, "learning_rate": 8.832717453098351e-06, "loss": 0.6679, "step": 33352 }, { "epoch": 3.5489465843796553, "grad_norm": 14.65625, "learning_rate": 8.832681921546335e-06, "loss": 0.7994, "step": 33353 }, { "epoch": 3.5490529899978718, "grad_norm": 15.5703125, "learning_rate": 8.832646389994316e-06, "loss": 0.7138, "step": 33354 }, { "epoch": 3.5491593956160887, "grad_norm": 15.890625, "learning_rate": 8.832610858442298e-06, "loss": 0.6297, "step": 33355 }, { "epoch": 3.5492658012343052, "grad_norm": 13.1953125, "learning_rate": 8.83257532689028e-06, "loss": 0.6753, "step": 33356 }, { "epoch": 3.5493722068525217, "grad_norm": 13.2890625, "learning_rate": 8.832539795338261e-06, "loss": 0.5192, "step": 33357 }, { "epoch": 3.5494786124707387, "grad_norm": 9.484375, "learning_rate": 8.832504263786243e-06, "loss": 0.5301, "step": 33358 }, { "epoch": 3.549585018088955, "grad_norm": 9.171875, "learning_rate": 8.832468732234224e-06, "loss": 0.5716, "step": 33359 }, { "epoch": 3.5496914237071717, "grad_norm": 13.59375, "learning_rate": 8.832433200682206e-06, "loss": 0.6233, "step": 33360 }, { "epoch": 3.5497978293253882, "grad_norm": 11.296875, "learning_rate": 8.832397669130188e-06, "loss": 0.6361, "step": 33361 }, { "epoch": 3.549904234943605, "grad_norm": 14.8125, "learning_rate": 8.832362137578169e-06, "loss": 0.6975, "step": 33362 }, { "epoch": 3.5500106405618217, "grad_norm": 15.484375, "learning_rate": 8.83232660602615e-06, "loss": 0.7411, "step": 33363 }, { "epoch": 3.550117046180038, "grad_norm": 13.984375, "learning_rate": 8.832291074474134e-06, "loss": 0.5802, "step": 33364 }, { "epoch": 3.5502234517982547, "grad_norm": 18.65625, "learning_rate": 8.832255542922116e-06, "loss": 0.8543, "step": 33365 }, { "epoch": 3.5503298574164717, "grad_norm": 14.7734375, "learning_rate": 8.832220011370097e-06, "loss": 0.6939, "step": 33366 }, { "epoch": 3.550436263034688, "grad_norm": 14.40625, "learning_rate": 8.832184479818079e-06, "loss": 0.6734, "step": 33367 }, { "epoch": 3.5505426686529047, "grad_norm": 12.078125, "learning_rate": 8.83214894826606e-06, "loss": 0.6399, "step": 33368 }, { "epoch": 3.5506490742711216, "grad_norm": 9.359375, "learning_rate": 8.832113416714044e-06, "loss": 0.6512, "step": 33369 }, { "epoch": 3.550755479889338, "grad_norm": 10.34375, "learning_rate": 8.832077885162025e-06, "loss": 0.5688, "step": 33370 }, { "epoch": 3.5508618855075547, "grad_norm": 9.0, "learning_rate": 8.832042353610007e-06, "loss": 0.5514, "step": 33371 }, { "epoch": 3.5509682911257716, "grad_norm": 12.078125, "learning_rate": 8.832006822057989e-06, "loss": 0.5423, "step": 33372 }, { "epoch": 3.551074696743988, "grad_norm": 12.5390625, "learning_rate": 8.83197129050597e-06, "loss": 0.6281, "step": 33373 }, { "epoch": 3.5511811023622046, "grad_norm": 10.703125, "learning_rate": 8.831935758953952e-06, "loss": 0.5762, "step": 33374 }, { "epoch": 3.5512875079804216, "grad_norm": 22.015625, "learning_rate": 8.831900227401933e-06, "loss": 0.7075, "step": 33375 }, { "epoch": 3.551393913598638, "grad_norm": 20.359375, "learning_rate": 8.831864695849915e-06, "loss": 0.8366, "step": 33376 }, { "epoch": 3.5515003192168546, "grad_norm": 12.625, "learning_rate": 8.831829164297897e-06, "loss": 0.6396, "step": 33377 }, { "epoch": 3.5516067248350716, "grad_norm": 12.6015625, "learning_rate": 8.831793632745878e-06, "loss": 0.5742, "step": 33378 }, { "epoch": 3.551713130453288, "grad_norm": 13.7578125, "learning_rate": 8.831758101193862e-06, "loss": 0.6632, "step": 33379 }, { "epoch": 3.5518195360715046, "grad_norm": 11.828125, "learning_rate": 8.831722569641843e-06, "loss": 0.5079, "step": 33380 }, { "epoch": 3.551925941689721, "grad_norm": 13.328125, "learning_rate": 8.831687038089825e-06, "loss": 0.6853, "step": 33381 }, { "epoch": 3.5520323473079376, "grad_norm": 12.796875, "learning_rate": 8.831651506537806e-06, "loss": 0.5763, "step": 33382 }, { "epoch": 3.5521387529261546, "grad_norm": 11.3515625, "learning_rate": 8.831615974985788e-06, "loss": 0.7426, "step": 33383 }, { "epoch": 3.552245158544371, "grad_norm": 10.4765625, "learning_rate": 8.831580443433771e-06, "loss": 0.658, "step": 33384 }, { "epoch": 3.5523515641625876, "grad_norm": 14.6484375, "learning_rate": 8.831544911881751e-06, "loss": 0.6855, "step": 33385 }, { "epoch": 3.5524579697808045, "grad_norm": 12.3359375, "learning_rate": 8.831509380329733e-06, "loss": 0.626, "step": 33386 }, { "epoch": 3.552564375399021, "grad_norm": 13.6796875, "learning_rate": 8.831473848777714e-06, "loss": 0.6059, "step": 33387 }, { "epoch": 3.5526707810172375, "grad_norm": 12.3671875, "learning_rate": 8.831438317225696e-06, "loss": 0.5166, "step": 33388 }, { "epoch": 3.5527771866354545, "grad_norm": 23.359375, "learning_rate": 8.83140278567368e-06, "loss": 0.917, "step": 33389 }, { "epoch": 3.552883592253671, "grad_norm": 13.328125, "learning_rate": 8.831367254121661e-06, "loss": 0.6787, "step": 33390 }, { "epoch": 3.5529899978718875, "grad_norm": 11.8203125, "learning_rate": 8.831331722569642e-06, "loss": 0.5581, "step": 33391 }, { "epoch": 3.5530964034901045, "grad_norm": 13.0703125, "learning_rate": 8.831296191017624e-06, "loss": 0.5956, "step": 33392 }, { "epoch": 3.553202809108321, "grad_norm": 17.5625, "learning_rate": 8.831260659465606e-06, "loss": 0.8054, "step": 33393 }, { "epoch": 3.5533092147265375, "grad_norm": 14.75, "learning_rate": 8.831225127913587e-06, "loss": 0.7653, "step": 33394 }, { "epoch": 3.5534156203447544, "grad_norm": 11.328125, "learning_rate": 8.83118959636157e-06, "loss": 0.6862, "step": 33395 }, { "epoch": 3.553522025962971, "grad_norm": 10.90625, "learning_rate": 8.831154064809552e-06, "loss": 0.5779, "step": 33396 }, { "epoch": 3.5536284315811875, "grad_norm": 20.515625, "learning_rate": 8.831118533257534e-06, "loss": 0.8228, "step": 33397 }, { "epoch": 3.553734837199404, "grad_norm": 10.6171875, "learning_rate": 8.831083001705515e-06, "loss": 0.5187, "step": 33398 }, { "epoch": 3.553841242817621, "grad_norm": 11.828125, "learning_rate": 8.831047470153497e-06, "loss": 0.633, "step": 33399 }, { "epoch": 3.5539476484358374, "grad_norm": 15.625, "learning_rate": 8.831011938601479e-06, "loss": 0.5938, "step": 33400 }, { "epoch": 3.554054054054054, "grad_norm": 14.09375, "learning_rate": 8.83097640704946e-06, "loss": 0.672, "step": 33401 }, { "epoch": 3.5541604596722705, "grad_norm": 12.0703125, "learning_rate": 8.830940875497442e-06, "loss": 0.6063, "step": 33402 }, { "epoch": 3.5542668652904874, "grad_norm": 10.078125, "learning_rate": 8.830905343945423e-06, "loss": 0.6247, "step": 33403 }, { "epoch": 3.554373270908704, "grad_norm": 10.7578125, "learning_rate": 8.830869812393405e-06, "loss": 0.5004, "step": 33404 }, { "epoch": 3.5544796765269204, "grad_norm": 14.0625, "learning_rate": 8.830834280841388e-06, "loss": 0.7653, "step": 33405 }, { "epoch": 3.5545860821451374, "grad_norm": 15.421875, "learning_rate": 8.83079874928937e-06, "loss": 0.7643, "step": 33406 }, { "epoch": 3.554692487763354, "grad_norm": 11.8359375, "learning_rate": 8.830763217737352e-06, "loss": 0.6098, "step": 33407 }, { "epoch": 3.5547988933815704, "grad_norm": 15.71875, "learning_rate": 8.830727686185333e-06, "loss": 0.7107, "step": 33408 }, { "epoch": 3.5549052989997874, "grad_norm": 10.953125, "learning_rate": 8.830692154633315e-06, "loss": 0.6304, "step": 33409 }, { "epoch": 3.555011704618004, "grad_norm": 14.4453125, "learning_rate": 8.830656623081298e-06, "loss": 0.7115, "step": 33410 }, { "epoch": 3.5551181102362204, "grad_norm": 10.171875, "learning_rate": 8.83062109152928e-06, "loss": 0.8012, "step": 33411 }, { "epoch": 3.5552245158544373, "grad_norm": 19.59375, "learning_rate": 8.830585559977261e-06, "loss": 0.8952, "step": 33412 }, { "epoch": 3.555330921472654, "grad_norm": 16.953125, "learning_rate": 8.830550028425243e-06, "loss": 0.6041, "step": 33413 }, { "epoch": 3.5554373270908703, "grad_norm": 14.9765625, "learning_rate": 8.830514496873225e-06, "loss": 0.7332, "step": 33414 }, { "epoch": 3.5555437327090873, "grad_norm": 17.625, "learning_rate": 8.830478965321206e-06, "loss": 0.7019, "step": 33415 }, { "epoch": 3.555650138327304, "grad_norm": 14.9296875, "learning_rate": 8.830443433769188e-06, "loss": 0.6309, "step": 33416 }, { "epoch": 3.5557565439455203, "grad_norm": 10.921875, "learning_rate": 8.83040790221717e-06, "loss": 0.5948, "step": 33417 }, { "epoch": 3.555862949563737, "grad_norm": 14.1171875, "learning_rate": 8.830372370665151e-06, "loss": 0.789, "step": 33418 }, { "epoch": 3.5559693551819533, "grad_norm": 12.328125, "learning_rate": 8.830336839113133e-06, "loss": 0.7059, "step": 33419 }, { "epoch": 3.5560757608001703, "grad_norm": 12.8515625, "learning_rate": 8.830301307561116e-06, "loss": 0.7857, "step": 33420 }, { "epoch": 3.556182166418387, "grad_norm": 13.234375, "learning_rate": 8.830265776009097e-06, "loss": 0.7146, "step": 33421 }, { "epoch": 3.5562885720366033, "grad_norm": 14.0390625, "learning_rate": 8.830230244457079e-06, "loss": 0.5074, "step": 33422 }, { "epoch": 3.5563949776548203, "grad_norm": 11.96875, "learning_rate": 8.83019471290506e-06, "loss": 0.6432, "step": 33423 }, { "epoch": 3.5565013832730368, "grad_norm": 17.1875, "learning_rate": 8.830159181353042e-06, "loss": 0.8792, "step": 33424 }, { "epoch": 3.5566077888912533, "grad_norm": 14.6640625, "learning_rate": 8.830123649801024e-06, "loss": 0.6353, "step": 33425 }, { "epoch": 3.5567141945094702, "grad_norm": 11.5078125, "learning_rate": 8.830088118249006e-06, "loss": 0.6261, "step": 33426 }, { "epoch": 3.5568206001276867, "grad_norm": 14.484375, "learning_rate": 8.830052586696987e-06, "loss": 0.6696, "step": 33427 }, { "epoch": 3.5569270057459033, "grad_norm": 16.265625, "learning_rate": 8.830017055144969e-06, "loss": 0.7441, "step": 33428 }, { "epoch": 3.55703341136412, "grad_norm": 16.328125, "learning_rate": 8.82998152359295e-06, "loss": 0.8048, "step": 33429 }, { "epoch": 3.5571398169823367, "grad_norm": 11.4921875, "learning_rate": 8.829945992040934e-06, "loss": 0.658, "step": 33430 }, { "epoch": 3.5572462226005532, "grad_norm": 11.3203125, "learning_rate": 8.829910460488915e-06, "loss": 0.6714, "step": 33431 }, { "epoch": 3.55735262821877, "grad_norm": 13.859375, "learning_rate": 8.829874928936897e-06, "loss": 0.7662, "step": 33432 }, { "epoch": 3.5574590338369867, "grad_norm": 13.28125, "learning_rate": 8.829839397384878e-06, "loss": 0.5946, "step": 33433 }, { "epoch": 3.557565439455203, "grad_norm": 13.0234375, "learning_rate": 8.82980386583286e-06, "loss": 0.7744, "step": 33434 }, { "epoch": 3.5576718450734197, "grad_norm": 11.046875, "learning_rate": 8.829768334280842e-06, "loss": 0.5973, "step": 33435 }, { "epoch": 3.5577782506916367, "grad_norm": 10.140625, "learning_rate": 8.829732802728825e-06, "loss": 0.6235, "step": 33436 }, { "epoch": 3.557884656309853, "grad_norm": 11.2890625, "learning_rate": 8.829697271176807e-06, "loss": 0.6141, "step": 33437 }, { "epoch": 3.5579910619280697, "grad_norm": 11.1171875, "learning_rate": 8.829661739624788e-06, "loss": 0.8014, "step": 33438 }, { "epoch": 3.558097467546286, "grad_norm": 20.0625, "learning_rate": 8.82962620807277e-06, "loss": 0.7178, "step": 33439 }, { "epoch": 3.558203873164503, "grad_norm": 11.5703125, "learning_rate": 8.829590676520751e-06, "loss": 0.4847, "step": 33440 }, { "epoch": 3.5583102787827197, "grad_norm": 13.484375, "learning_rate": 8.829555144968733e-06, "loss": 0.6746, "step": 33441 }, { "epoch": 3.558416684400936, "grad_norm": 12.03125, "learning_rate": 8.829519613416715e-06, "loss": 0.6817, "step": 33442 }, { "epoch": 3.558523090019153, "grad_norm": 12.640625, "learning_rate": 8.829484081864696e-06, "loss": 0.6672, "step": 33443 }, { "epoch": 3.5586294956373696, "grad_norm": 14.6171875, "learning_rate": 8.829448550312678e-06, "loss": 0.7343, "step": 33444 }, { "epoch": 3.558735901255586, "grad_norm": 14.1640625, "learning_rate": 8.82941301876066e-06, "loss": 0.6639, "step": 33445 }, { "epoch": 3.558842306873803, "grad_norm": 12.921875, "learning_rate": 8.829377487208643e-06, "loss": 0.6083, "step": 33446 }, { "epoch": 3.5589487124920196, "grad_norm": 9.546875, "learning_rate": 8.829341955656624e-06, "loss": 0.5453, "step": 33447 }, { "epoch": 3.559055118110236, "grad_norm": 12.875, "learning_rate": 8.829306424104606e-06, "loss": 0.4632, "step": 33448 }, { "epoch": 3.559161523728453, "grad_norm": 13.671875, "learning_rate": 8.829270892552588e-06, "loss": 0.5898, "step": 33449 }, { "epoch": 3.5592679293466696, "grad_norm": 9.7265625, "learning_rate": 8.829235361000569e-06, "loss": 0.6979, "step": 33450 }, { "epoch": 3.559374334964886, "grad_norm": 15.4375, "learning_rate": 8.82919982944855e-06, "loss": 0.7201, "step": 33451 }, { "epoch": 3.559480740583103, "grad_norm": 12.25, "learning_rate": 8.829164297896532e-06, "loss": 0.689, "step": 33452 }, { "epoch": 3.5595871462013196, "grad_norm": 12.4765625, "learning_rate": 8.829128766344514e-06, "loss": 0.6981, "step": 33453 }, { "epoch": 3.559693551819536, "grad_norm": 9.7421875, "learning_rate": 8.829093234792496e-06, "loss": 0.6654, "step": 33454 }, { "epoch": 3.5597999574377526, "grad_norm": 15.6328125, "learning_rate": 8.829057703240477e-06, "loss": 0.7481, "step": 33455 }, { "epoch": 3.559906363055969, "grad_norm": 12.0078125, "learning_rate": 8.82902217168846e-06, "loss": 0.5305, "step": 33456 }, { "epoch": 3.560012768674186, "grad_norm": 15.796875, "learning_rate": 8.828986640136442e-06, "loss": 0.7677, "step": 33457 }, { "epoch": 3.5601191742924025, "grad_norm": 14.25, "learning_rate": 8.828951108584424e-06, "loss": 0.8162, "step": 33458 }, { "epoch": 3.560225579910619, "grad_norm": 13.5234375, "learning_rate": 8.828915577032405e-06, "loss": 0.631, "step": 33459 }, { "epoch": 3.560331985528836, "grad_norm": 13.3359375, "learning_rate": 8.828880045480387e-06, "loss": 0.6531, "step": 33460 }, { "epoch": 3.5604383911470525, "grad_norm": 15.2265625, "learning_rate": 8.82884451392837e-06, "loss": 0.6595, "step": 33461 }, { "epoch": 3.560544796765269, "grad_norm": 15.015625, "learning_rate": 8.828808982376352e-06, "loss": 0.7134, "step": 33462 }, { "epoch": 3.560651202383486, "grad_norm": 12.8203125, "learning_rate": 8.828773450824333e-06, "loss": 0.6628, "step": 33463 }, { "epoch": 3.5607576080017025, "grad_norm": 18.59375, "learning_rate": 8.828737919272315e-06, "loss": 0.8825, "step": 33464 }, { "epoch": 3.560864013619919, "grad_norm": 17.71875, "learning_rate": 8.828702387720297e-06, "loss": 0.7107, "step": 33465 }, { "epoch": 3.560970419238136, "grad_norm": 13.0078125, "learning_rate": 8.828666856168278e-06, "loss": 0.6779, "step": 33466 }, { "epoch": 3.5610768248563525, "grad_norm": 12.546875, "learning_rate": 8.82863132461626e-06, "loss": 0.73, "step": 33467 }, { "epoch": 3.561183230474569, "grad_norm": 14.0, "learning_rate": 8.828595793064241e-06, "loss": 0.7599, "step": 33468 }, { "epoch": 3.561289636092786, "grad_norm": 10.1796875, "learning_rate": 8.828560261512223e-06, "loss": 0.6549, "step": 33469 }, { "epoch": 3.5613960417110024, "grad_norm": 15.4140625, "learning_rate": 8.828524729960205e-06, "loss": 0.7725, "step": 33470 }, { "epoch": 3.561502447329219, "grad_norm": 10.8046875, "learning_rate": 8.828489198408188e-06, "loss": 0.5737, "step": 33471 }, { "epoch": 3.561608852947436, "grad_norm": 12.2734375, "learning_rate": 8.82845366685617e-06, "loss": 0.6591, "step": 33472 }, { "epoch": 3.5617152585656524, "grad_norm": 11.2734375, "learning_rate": 8.828418135304151e-06, "loss": 0.5646, "step": 33473 }, { "epoch": 3.561821664183869, "grad_norm": 10.296875, "learning_rate": 8.828382603752133e-06, "loss": 0.5084, "step": 33474 }, { "epoch": 3.5619280698020854, "grad_norm": 12.234375, "learning_rate": 8.828347072200114e-06, "loss": 0.5759, "step": 33475 }, { "epoch": 3.562034475420302, "grad_norm": 20.96875, "learning_rate": 8.828311540648096e-06, "loss": 0.8906, "step": 33476 }, { "epoch": 3.562140881038519, "grad_norm": 21.046875, "learning_rate": 8.828276009096078e-06, "loss": 0.7197, "step": 33477 }, { "epoch": 3.5622472866567354, "grad_norm": 14.4765625, "learning_rate": 8.82824047754406e-06, "loss": 0.7196, "step": 33478 }, { "epoch": 3.562353692274952, "grad_norm": 9.9375, "learning_rate": 8.82820494599204e-06, "loss": 0.5555, "step": 33479 }, { "epoch": 3.562460097893169, "grad_norm": 11.9453125, "learning_rate": 8.828169414440022e-06, "loss": 0.5985, "step": 33480 }, { "epoch": 3.5625665035113854, "grad_norm": 14.4375, "learning_rate": 8.828133882888004e-06, "loss": 0.6087, "step": 33481 }, { "epoch": 3.562672909129602, "grad_norm": 14.734375, "learning_rate": 8.828098351335987e-06, "loss": 0.6951, "step": 33482 }, { "epoch": 3.562779314747819, "grad_norm": 11.4609375, "learning_rate": 8.828062819783969e-06, "loss": 0.7044, "step": 33483 }, { "epoch": 3.5628857203660353, "grad_norm": 14.1328125, "learning_rate": 8.82802728823195e-06, "loss": 0.6367, "step": 33484 }, { "epoch": 3.562992125984252, "grad_norm": 15.171875, "learning_rate": 8.827991756679932e-06, "loss": 0.6622, "step": 33485 }, { "epoch": 3.563098531602469, "grad_norm": 11.6875, "learning_rate": 8.827956225127914e-06, "loss": 0.4949, "step": 33486 }, { "epoch": 3.5632049372206853, "grad_norm": 16.015625, "learning_rate": 8.827920693575897e-06, "loss": 0.7917, "step": 33487 }, { "epoch": 3.563311342838902, "grad_norm": 13.921875, "learning_rate": 8.827885162023879e-06, "loss": 0.788, "step": 33488 }, { "epoch": 3.563417748457119, "grad_norm": 12.15625, "learning_rate": 8.82784963047186e-06, "loss": 0.5874, "step": 33489 }, { "epoch": 3.5635241540753353, "grad_norm": 14.1875, "learning_rate": 8.827814098919842e-06, "loss": 0.6897, "step": 33490 }, { "epoch": 3.563630559693552, "grad_norm": 13.4140625, "learning_rate": 8.827778567367823e-06, "loss": 0.5495, "step": 33491 }, { "epoch": 3.5637369653117683, "grad_norm": 11.5859375, "learning_rate": 8.827743035815805e-06, "loss": 0.5627, "step": 33492 }, { "epoch": 3.5638433709299853, "grad_norm": 13.5078125, "learning_rate": 8.827707504263787e-06, "loss": 0.7279, "step": 33493 }, { "epoch": 3.5639497765482018, "grad_norm": 11.2578125, "learning_rate": 8.827671972711768e-06, "loss": 0.6096, "step": 33494 }, { "epoch": 3.5640561821664183, "grad_norm": 12.3125, "learning_rate": 8.82763644115975e-06, "loss": 0.5191, "step": 33495 }, { "epoch": 3.564162587784635, "grad_norm": 12.2890625, "learning_rate": 8.827600909607732e-06, "loss": 0.6355, "step": 33496 }, { "epoch": 3.5642689934028517, "grad_norm": 10.5078125, "learning_rate": 8.827565378055715e-06, "loss": 0.5731, "step": 33497 }, { "epoch": 3.5643753990210683, "grad_norm": 13.7421875, "learning_rate": 8.827529846503696e-06, "loss": 0.8131, "step": 33498 }, { "epoch": 3.5644818046392848, "grad_norm": 13.8984375, "learning_rate": 8.827494314951678e-06, "loss": 0.5098, "step": 33499 }, { "epoch": 3.5645882102575017, "grad_norm": 14.609375, "learning_rate": 8.82745878339966e-06, "loss": 0.6289, "step": 33500 }, { "epoch": 3.5646946158757182, "grad_norm": 16.4375, "learning_rate": 8.827423251847641e-06, "loss": 0.6606, "step": 33501 }, { "epoch": 3.5648010214939347, "grad_norm": 10.484375, "learning_rate": 8.827387720295625e-06, "loss": 0.5559, "step": 33502 }, { "epoch": 3.5649074271121517, "grad_norm": 17.4375, "learning_rate": 8.827352188743606e-06, "loss": 0.6113, "step": 33503 }, { "epoch": 3.565013832730368, "grad_norm": 14.578125, "learning_rate": 8.827316657191588e-06, "loss": 0.5701, "step": 33504 }, { "epoch": 3.5651202383485847, "grad_norm": 11.53125, "learning_rate": 8.827281125639568e-06, "loss": 0.618, "step": 33505 }, { "epoch": 3.5652266439668017, "grad_norm": 10.2890625, "learning_rate": 8.82724559408755e-06, "loss": 0.6336, "step": 33506 }, { "epoch": 3.565333049585018, "grad_norm": 16.09375, "learning_rate": 8.827210062535533e-06, "loss": 0.6872, "step": 33507 }, { "epoch": 3.5654394552032347, "grad_norm": 12.7421875, "learning_rate": 8.827174530983514e-06, "loss": 0.5927, "step": 33508 }, { "epoch": 3.5655458608214516, "grad_norm": 15.03125, "learning_rate": 8.827138999431496e-06, "loss": 0.675, "step": 33509 }, { "epoch": 3.565652266439668, "grad_norm": 13.78125, "learning_rate": 8.827103467879477e-06, "loss": 0.6053, "step": 33510 }, { "epoch": 3.5657586720578847, "grad_norm": 16.125, "learning_rate": 8.827067936327459e-06, "loss": 0.6948, "step": 33511 }, { "epoch": 3.565865077676101, "grad_norm": 14.0703125, "learning_rate": 8.82703240477544e-06, "loss": 0.5718, "step": 33512 }, { "epoch": 3.5659714832943177, "grad_norm": 10.9453125, "learning_rate": 8.826996873223424e-06, "loss": 0.6011, "step": 33513 }, { "epoch": 3.5660778889125346, "grad_norm": 13.171875, "learning_rate": 8.826961341671406e-06, "loss": 0.6501, "step": 33514 }, { "epoch": 3.566184294530751, "grad_norm": 18.609375, "learning_rate": 8.826925810119387e-06, "loss": 0.8682, "step": 33515 }, { "epoch": 3.5662907001489677, "grad_norm": 13.359375, "learning_rate": 8.826890278567369e-06, "loss": 0.6416, "step": 33516 }, { "epoch": 3.5663971057671846, "grad_norm": 14.6875, "learning_rate": 8.82685474701535e-06, "loss": 0.7895, "step": 33517 }, { "epoch": 3.566503511385401, "grad_norm": 15.3046875, "learning_rate": 8.826819215463332e-06, "loss": 0.5768, "step": 33518 }, { "epoch": 3.5666099170036176, "grad_norm": 12.4296875, "learning_rate": 8.826783683911314e-06, "loss": 0.6115, "step": 33519 }, { "epoch": 3.5667163226218346, "grad_norm": 10.34375, "learning_rate": 8.826748152359295e-06, "loss": 0.6109, "step": 33520 }, { "epoch": 3.566822728240051, "grad_norm": 12.3203125, "learning_rate": 8.826712620807277e-06, "loss": 0.5961, "step": 33521 }, { "epoch": 3.5669291338582676, "grad_norm": 12.2265625, "learning_rate": 8.826677089255258e-06, "loss": 0.6124, "step": 33522 }, { "epoch": 3.5670355394764846, "grad_norm": 13.8359375, "learning_rate": 8.826641557703242e-06, "loss": 0.5945, "step": 33523 }, { "epoch": 3.567141945094701, "grad_norm": 12.078125, "learning_rate": 8.826606026151223e-06, "loss": 0.6623, "step": 33524 }, { "epoch": 3.5672483507129176, "grad_norm": 10.640625, "learning_rate": 8.826570494599205e-06, "loss": 0.561, "step": 33525 }, { "epoch": 3.5673547563311345, "grad_norm": 12.2734375, "learning_rate": 8.826534963047187e-06, "loss": 0.7016, "step": 33526 }, { "epoch": 3.567461161949351, "grad_norm": 12.84375, "learning_rate": 8.826499431495168e-06, "loss": 0.6679, "step": 33527 }, { "epoch": 3.5675675675675675, "grad_norm": 11.2734375, "learning_rate": 8.826463899943151e-06, "loss": 0.668, "step": 33528 }, { "epoch": 3.567673973185784, "grad_norm": 12.34375, "learning_rate": 8.826428368391133e-06, "loss": 0.5235, "step": 33529 }, { "epoch": 3.567780378804001, "grad_norm": 14.5234375, "learning_rate": 8.826392836839115e-06, "loss": 0.7247, "step": 33530 }, { "epoch": 3.5678867844222175, "grad_norm": 11.59375, "learning_rate": 8.826357305287096e-06, "loss": 0.6267, "step": 33531 }, { "epoch": 3.567993190040434, "grad_norm": 15.6875, "learning_rate": 8.826321773735078e-06, "loss": 0.6642, "step": 33532 }, { "epoch": 3.5680995956586505, "grad_norm": 14.7578125, "learning_rate": 8.82628624218306e-06, "loss": 0.7101, "step": 33533 }, { "epoch": 3.5682060012768675, "grad_norm": 13.953125, "learning_rate": 8.826250710631041e-06, "loss": 0.7522, "step": 33534 }, { "epoch": 3.568312406895084, "grad_norm": 14.6796875, "learning_rate": 8.826215179079023e-06, "loss": 0.5593, "step": 33535 }, { "epoch": 3.5684188125133005, "grad_norm": 14.4921875, "learning_rate": 8.826179647527004e-06, "loss": 0.7776, "step": 33536 }, { "epoch": 3.5685252181315175, "grad_norm": 9.453125, "learning_rate": 8.826144115974986e-06, "loss": 0.5908, "step": 33537 }, { "epoch": 3.568631623749734, "grad_norm": 12.7890625, "learning_rate": 8.82610858442297e-06, "loss": 0.6416, "step": 33538 }, { "epoch": 3.5687380293679505, "grad_norm": 8.7109375, "learning_rate": 8.82607305287095e-06, "loss": 0.4379, "step": 33539 }, { "epoch": 3.5688444349861674, "grad_norm": 12.5, "learning_rate": 8.826037521318932e-06, "loss": 0.6018, "step": 33540 }, { "epoch": 3.568950840604384, "grad_norm": 20.703125, "learning_rate": 8.826001989766914e-06, "loss": 0.7671, "step": 33541 }, { "epoch": 3.5690572462226005, "grad_norm": 12.6640625, "learning_rate": 8.825966458214896e-06, "loss": 0.5724, "step": 33542 }, { "epoch": 3.5691636518408174, "grad_norm": 12.1328125, "learning_rate": 8.825930926662877e-06, "loss": 0.5833, "step": 33543 }, { "epoch": 3.569270057459034, "grad_norm": 11.4765625, "learning_rate": 8.825895395110859e-06, "loss": 0.4943, "step": 33544 }, { "epoch": 3.5693764630772504, "grad_norm": 9.2578125, "learning_rate": 8.82585986355884e-06, "loss": 0.4943, "step": 33545 }, { "epoch": 3.5694828686954674, "grad_norm": 12.6328125, "learning_rate": 8.825824332006822e-06, "loss": 0.6218, "step": 33546 }, { "epoch": 3.569589274313684, "grad_norm": 13.375, "learning_rate": 8.825788800454804e-06, "loss": 0.7114, "step": 33547 }, { "epoch": 3.5696956799319004, "grad_norm": 17.796875, "learning_rate": 8.825753268902787e-06, "loss": 0.8493, "step": 33548 }, { "epoch": 3.569802085550117, "grad_norm": 12.25, "learning_rate": 8.825717737350769e-06, "loss": 0.6061, "step": 33549 }, { "epoch": 3.5699084911683334, "grad_norm": 10.640625, "learning_rate": 8.82568220579875e-06, "loss": 0.5474, "step": 33550 }, { "epoch": 3.5700148967865504, "grad_norm": 11.8828125, "learning_rate": 8.825646674246732e-06, "loss": 0.5753, "step": 33551 }, { "epoch": 3.570121302404767, "grad_norm": 12.015625, "learning_rate": 8.825611142694713e-06, "loss": 0.5643, "step": 33552 }, { "epoch": 3.5702277080229834, "grad_norm": 15.671875, "learning_rate": 8.825575611142695e-06, "loss": 0.7865, "step": 33553 }, { "epoch": 3.5703341136412003, "grad_norm": 11.75, "learning_rate": 8.825540079590678e-06, "loss": 0.5482, "step": 33554 }, { "epoch": 3.570440519259417, "grad_norm": 15.578125, "learning_rate": 8.82550454803866e-06, "loss": 0.7992, "step": 33555 }, { "epoch": 3.5705469248776334, "grad_norm": 15.5625, "learning_rate": 8.825469016486641e-06, "loss": 0.8166, "step": 33556 }, { "epoch": 3.5706533304958503, "grad_norm": 11.9609375, "learning_rate": 8.825433484934623e-06, "loss": 0.6164, "step": 33557 }, { "epoch": 3.570759736114067, "grad_norm": 13.078125, "learning_rate": 8.825397953382605e-06, "loss": 0.6618, "step": 33558 }, { "epoch": 3.5708661417322833, "grad_norm": 10.9375, "learning_rate": 8.825362421830586e-06, "loss": 0.5747, "step": 33559 }, { "epoch": 3.5709725473505003, "grad_norm": 14.6328125, "learning_rate": 8.825326890278568e-06, "loss": 0.6321, "step": 33560 }, { "epoch": 3.571078952968717, "grad_norm": 12.671875, "learning_rate": 8.82529135872655e-06, "loss": 0.677, "step": 33561 }, { "epoch": 3.5711853585869333, "grad_norm": 15.78125, "learning_rate": 8.825255827174531e-06, "loss": 0.7037, "step": 33562 }, { "epoch": 3.5712917642051503, "grad_norm": 11.0234375, "learning_rate": 8.825220295622513e-06, "loss": 0.6061, "step": 33563 }, { "epoch": 3.5713981698233668, "grad_norm": 13.953125, "learning_rate": 8.825184764070496e-06, "loss": 0.5001, "step": 33564 }, { "epoch": 3.5715045754415833, "grad_norm": 9.4140625, "learning_rate": 8.825149232518478e-06, "loss": 0.5248, "step": 33565 }, { "epoch": 3.5716109810598, "grad_norm": 13.8125, "learning_rate": 8.82511370096646e-06, "loss": 0.6513, "step": 33566 }, { "epoch": 3.5717173866780167, "grad_norm": 16.453125, "learning_rate": 8.82507816941444e-06, "loss": 0.6723, "step": 33567 }, { "epoch": 3.5718237922962333, "grad_norm": 13.921875, "learning_rate": 8.825042637862422e-06, "loss": 0.5969, "step": 33568 }, { "epoch": 3.5719301979144498, "grad_norm": 15.5078125, "learning_rate": 8.825007106310404e-06, "loss": 0.7501, "step": 33569 }, { "epoch": 3.5720366035326663, "grad_norm": 14.8515625, "learning_rate": 8.824971574758386e-06, "loss": 0.6224, "step": 33570 }, { "epoch": 3.5721430091508832, "grad_norm": 22.1875, "learning_rate": 8.824936043206367e-06, "loss": 0.8663, "step": 33571 }, { "epoch": 3.5722494147690997, "grad_norm": 13.671875, "learning_rate": 8.824900511654349e-06, "loss": 0.5839, "step": 33572 }, { "epoch": 3.5723558203873162, "grad_norm": 15.0703125, "learning_rate": 8.82486498010233e-06, "loss": 0.6065, "step": 33573 }, { "epoch": 3.572462226005533, "grad_norm": 10.5625, "learning_rate": 8.824829448550314e-06, "loss": 0.6402, "step": 33574 }, { "epoch": 3.5725686316237497, "grad_norm": 13.0625, "learning_rate": 8.824793916998295e-06, "loss": 0.6987, "step": 33575 }, { "epoch": 3.5726750372419662, "grad_norm": 12.4140625, "learning_rate": 8.824758385446277e-06, "loss": 0.7271, "step": 33576 }, { "epoch": 3.572781442860183, "grad_norm": 13.5234375, "learning_rate": 8.824722853894259e-06, "loss": 0.6324, "step": 33577 }, { "epoch": 3.5728878484783997, "grad_norm": 15.125, "learning_rate": 8.82468732234224e-06, "loss": 0.6629, "step": 33578 }, { "epoch": 3.572994254096616, "grad_norm": 11.2109375, "learning_rate": 8.824651790790224e-06, "loss": 0.5695, "step": 33579 }, { "epoch": 3.573100659714833, "grad_norm": 11.59375, "learning_rate": 8.824616259238205e-06, "loss": 0.5692, "step": 33580 }, { "epoch": 3.5732070653330497, "grad_norm": 10.984375, "learning_rate": 8.824580727686187e-06, "loss": 0.6374, "step": 33581 }, { "epoch": 3.573313470951266, "grad_norm": 11.015625, "learning_rate": 8.824545196134168e-06, "loss": 0.6136, "step": 33582 }, { "epoch": 3.573419876569483, "grad_norm": 9.6328125, "learning_rate": 8.82450966458215e-06, "loss": 0.5359, "step": 33583 }, { "epoch": 3.5735262821876996, "grad_norm": 10.71875, "learning_rate": 8.824474133030132e-06, "loss": 0.571, "step": 33584 }, { "epoch": 3.573632687805916, "grad_norm": 10.9765625, "learning_rate": 8.824438601478113e-06, "loss": 0.5397, "step": 33585 }, { "epoch": 3.5737390934241327, "grad_norm": 16.484375, "learning_rate": 8.824403069926095e-06, "loss": 0.8587, "step": 33586 }, { "epoch": 3.573845499042349, "grad_norm": 30.21875, "learning_rate": 8.824367538374076e-06, "loss": 0.6943, "step": 33587 }, { "epoch": 3.573951904660566, "grad_norm": 16.46875, "learning_rate": 8.824332006822058e-06, "loss": 0.7024, "step": 33588 }, { "epoch": 3.5740583102787826, "grad_norm": 17.625, "learning_rate": 8.82429647527004e-06, "loss": 0.7981, "step": 33589 }, { "epoch": 3.574164715896999, "grad_norm": 9.84375, "learning_rate": 8.824260943718023e-06, "loss": 0.5467, "step": 33590 }, { "epoch": 3.574271121515216, "grad_norm": 12.1953125, "learning_rate": 8.824225412166004e-06, "loss": 0.6612, "step": 33591 }, { "epoch": 3.5743775271334326, "grad_norm": 15.1015625, "learning_rate": 8.824189880613986e-06, "loss": 0.7234, "step": 33592 }, { "epoch": 3.574483932751649, "grad_norm": 13.5703125, "learning_rate": 8.824154349061968e-06, "loss": 0.7139, "step": 33593 }, { "epoch": 3.574590338369866, "grad_norm": 20.640625, "learning_rate": 8.82411881750995e-06, "loss": 0.7071, "step": 33594 }, { "epoch": 3.5746967439880826, "grad_norm": 13.8671875, "learning_rate": 8.824083285957931e-06, "loss": 0.6081, "step": 33595 }, { "epoch": 3.574803149606299, "grad_norm": 15.2109375, "learning_rate": 8.824047754405913e-06, "loss": 0.7617, "step": 33596 }, { "epoch": 3.574909555224516, "grad_norm": 17.1875, "learning_rate": 8.824012222853894e-06, "loss": 0.7232, "step": 33597 }, { "epoch": 3.5750159608427325, "grad_norm": 12.6640625, "learning_rate": 8.823976691301876e-06, "loss": 0.7064, "step": 33598 }, { "epoch": 3.575122366460949, "grad_norm": 10.671875, "learning_rate": 8.823941159749857e-06, "loss": 0.5304, "step": 33599 }, { "epoch": 3.575228772079166, "grad_norm": 13.09375, "learning_rate": 8.82390562819784e-06, "loss": 0.6037, "step": 33600 }, { "epoch": 3.5753351776973825, "grad_norm": 10.7109375, "learning_rate": 8.823870096645822e-06, "loss": 0.451, "step": 33601 }, { "epoch": 3.575441583315599, "grad_norm": 9.75, "learning_rate": 8.823834565093804e-06, "loss": 0.5572, "step": 33602 }, { "epoch": 3.575547988933816, "grad_norm": 11.6953125, "learning_rate": 8.823799033541785e-06, "loss": 0.5285, "step": 33603 }, { "epoch": 3.5756543945520325, "grad_norm": 13.0546875, "learning_rate": 8.823763501989767e-06, "loss": 0.6944, "step": 33604 }, { "epoch": 3.575760800170249, "grad_norm": 12.015625, "learning_rate": 8.82372797043775e-06, "loss": 0.5709, "step": 33605 }, { "epoch": 3.5758672057884655, "grad_norm": 13.640625, "learning_rate": 8.823692438885732e-06, "loss": 0.5876, "step": 33606 }, { "epoch": 3.575973611406682, "grad_norm": 16.671875, "learning_rate": 8.823656907333714e-06, "loss": 0.7681, "step": 33607 }, { "epoch": 3.576080017024899, "grad_norm": 12.3046875, "learning_rate": 8.823621375781695e-06, "loss": 0.6259, "step": 33608 }, { "epoch": 3.5761864226431155, "grad_norm": 10.09375, "learning_rate": 8.823585844229677e-06, "loss": 0.5454, "step": 33609 }, { "epoch": 3.576292828261332, "grad_norm": 14.796875, "learning_rate": 8.823550312677658e-06, "loss": 0.6567, "step": 33610 }, { "epoch": 3.576399233879549, "grad_norm": 16.671875, "learning_rate": 8.82351478112564e-06, "loss": 0.7335, "step": 33611 }, { "epoch": 3.5765056394977655, "grad_norm": 17.125, "learning_rate": 8.823479249573622e-06, "loss": 0.7696, "step": 33612 }, { "epoch": 3.576612045115982, "grad_norm": 12.0859375, "learning_rate": 8.823443718021603e-06, "loss": 0.6541, "step": 33613 }, { "epoch": 3.576718450734199, "grad_norm": 12.1015625, "learning_rate": 8.823408186469585e-06, "loss": 0.6342, "step": 33614 }, { "epoch": 3.5768248563524154, "grad_norm": 12.1171875, "learning_rate": 8.823372654917568e-06, "loss": 0.6478, "step": 33615 }, { "epoch": 3.576931261970632, "grad_norm": 13.84375, "learning_rate": 8.82333712336555e-06, "loss": 0.6058, "step": 33616 }, { "epoch": 3.577037667588849, "grad_norm": 15.890625, "learning_rate": 8.823301591813531e-06, "loss": 0.6987, "step": 33617 }, { "epoch": 3.5771440732070654, "grad_norm": 12.0859375, "learning_rate": 8.823266060261513e-06, "loss": 0.7191, "step": 33618 }, { "epoch": 3.577250478825282, "grad_norm": 11.984375, "learning_rate": 8.823230528709495e-06, "loss": 0.795, "step": 33619 }, { "epoch": 3.577356884443499, "grad_norm": 19.578125, "learning_rate": 8.823194997157478e-06, "loss": 0.7217, "step": 33620 }, { "epoch": 3.5774632900617154, "grad_norm": 45.5625, "learning_rate": 8.82315946560546e-06, "loss": 1.0268, "step": 33621 }, { "epoch": 3.577569695679932, "grad_norm": 16.609375, "learning_rate": 8.823123934053441e-06, "loss": 0.7514, "step": 33622 }, { "epoch": 3.5776761012981484, "grad_norm": 11.4609375, "learning_rate": 8.823088402501423e-06, "loss": 0.6544, "step": 33623 }, { "epoch": 3.5777825069163653, "grad_norm": 14.4765625, "learning_rate": 8.823052870949404e-06, "loss": 0.7847, "step": 33624 }, { "epoch": 3.577888912534582, "grad_norm": 11.9296875, "learning_rate": 8.823017339397386e-06, "loss": 0.6194, "step": 33625 }, { "epoch": 3.5779953181527984, "grad_norm": 17.640625, "learning_rate": 8.822981807845367e-06, "loss": 0.7287, "step": 33626 }, { "epoch": 3.578101723771015, "grad_norm": 13.4375, "learning_rate": 8.822946276293349e-06, "loss": 0.7519, "step": 33627 }, { "epoch": 3.578208129389232, "grad_norm": 11.4140625, "learning_rate": 8.82291074474133e-06, "loss": 0.532, "step": 33628 }, { "epoch": 3.5783145350074483, "grad_norm": 12.1953125, "learning_rate": 8.822875213189312e-06, "loss": 0.5994, "step": 33629 }, { "epoch": 3.578420940625665, "grad_norm": 12.6484375, "learning_rate": 8.822839681637294e-06, "loss": 0.637, "step": 33630 }, { "epoch": 3.578527346243882, "grad_norm": 11.5390625, "learning_rate": 8.822804150085277e-06, "loss": 0.5485, "step": 33631 }, { "epoch": 3.5786337518620983, "grad_norm": 12.2421875, "learning_rate": 8.822768618533259e-06, "loss": 0.7635, "step": 33632 }, { "epoch": 3.578740157480315, "grad_norm": 14.4140625, "learning_rate": 8.82273308698124e-06, "loss": 0.6682, "step": 33633 }, { "epoch": 3.5788465630985318, "grad_norm": 12.40625, "learning_rate": 8.822697555429222e-06, "loss": 0.5689, "step": 33634 }, { "epoch": 3.5789529687167483, "grad_norm": 10.859375, "learning_rate": 8.822662023877204e-06, "loss": 0.5339, "step": 33635 }, { "epoch": 3.579059374334965, "grad_norm": 13.5, "learning_rate": 8.822626492325185e-06, "loss": 0.4899, "step": 33636 }, { "epoch": 3.5791657799531817, "grad_norm": 10.78125, "learning_rate": 8.822590960773167e-06, "loss": 0.6323, "step": 33637 }, { "epoch": 3.5792721855713983, "grad_norm": 13.734375, "learning_rate": 8.822555429221148e-06, "loss": 0.5714, "step": 33638 }, { "epoch": 3.5793785911896148, "grad_norm": 14.890625, "learning_rate": 8.82251989766913e-06, "loss": 0.7023, "step": 33639 }, { "epoch": 3.5794849968078317, "grad_norm": 15.921875, "learning_rate": 8.822484366117112e-06, "loss": 0.6777, "step": 33640 }, { "epoch": 3.5795914024260482, "grad_norm": 13.4921875, "learning_rate": 8.822448834565095e-06, "loss": 0.6128, "step": 33641 }, { "epoch": 3.5796978080442647, "grad_norm": 11.4921875, "learning_rate": 8.822413303013077e-06, "loss": 0.6024, "step": 33642 }, { "epoch": 3.5798042136624812, "grad_norm": 17.15625, "learning_rate": 8.822377771461058e-06, "loss": 0.7365, "step": 33643 }, { "epoch": 3.5799106192806978, "grad_norm": 18.03125, "learning_rate": 8.82234223990904e-06, "loss": 0.8412, "step": 33644 }, { "epoch": 3.5800170248989147, "grad_norm": 14.15625, "learning_rate": 8.822306708357021e-06, "loss": 0.5944, "step": 33645 }, { "epoch": 3.5801234305171312, "grad_norm": 13.4921875, "learning_rate": 8.822271176805005e-06, "loss": 0.6422, "step": 33646 }, { "epoch": 3.5802298361353477, "grad_norm": 12.34375, "learning_rate": 8.822235645252986e-06, "loss": 0.6065, "step": 33647 }, { "epoch": 3.5803362417535647, "grad_norm": 14.5390625, "learning_rate": 8.822200113700968e-06, "loss": 0.6883, "step": 33648 }, { "epoch": 3.580442647371781, "grad_norm": 14.1640625, "learning_rate": 8.82216458214895e-06, "loss": 0.5777, "step": 33649 }, { "epoch": 3.5805490529899977, "grad_norm": 13.28125, "learning_rate": 8.822129050596931e-06, "loss": 0.5754, "step": 33650 }, { "epoch": 3.5806554586082147, "grad_norm": 21.046875, "learning_rate": 8.822093519044913e-06, "loss": 0.523, "step": 33651 }, { "epoch": 3.580761864226431, "grad_norm": 16.65625, "learning_rate": 8.822057987492894e-06, "loss": 0.9026, "step": 33652 }, { "epoch": 3.5808682698446477, "grad_norm": 12.640625, "learning_rate": 8.822022455940876e-06, "loss": 0.5791, "step": 33653 }, { "epoch": 3.5809746754628646, "grad_norm": 11.1953125, "learning_rate": 8.821986924388858e-06, "loss": 0.6225, "step": 33654 }, { "epoch": 3.581081081081081, "grad_norm": 16.78125, "learning_rate": 8.82195139283684e-06, "loss": 0.8455, "step": 33655 }, { "epoch": 3.5811874866992977, "grad_norm": 13.15625, "learning_rate": 8.821915861284822e-06, "loss": 0.715, "step": 33656 }, { "epoch": 3.5812938923175146, "grad_norm": 15.8203125, "learning_rate": 8.821880329732804e-06, "loss": 0.6798, "step": 33657 }, { "epoch": 3.581400297935731, "grad_norm": 13.625, "learning_rate": 8.821844798180786e-06, "loss": 0.7761, "step": 33658 }, { "epoch": 3.5815067035539476, "grad_norm": 13.34375, "learning_rate": 8.821809266628767e-06, "loss": 0.676, "step": 33659 }, { "epoch": 3.581613109172164, "grad_norm": 13.5859375, "learning_rate": 8.821773735076749e-06, "loss": 0.5016, "step": 33660 }, { "epoch": 3.581719514790381, "grad_norm": 11.1875, "learning_rate": 8.82173820352473e-06, "loss": 0.6289, "step": 33661 }, { "epoch": 3.5818259204085976, "grad_norm": 10.703125, "learning_rate": 8.821702671972712e-06, "loss": 0.5296, "step": 33662 }, { "epoch": 3.581932326026814, "grad_norm": 15.4375, "learning_rate": 8.821667140420694e-06, "loss": 0.7992, "step": 33663 }, { "epoch": 3.5820387316450306, "grad_norm": 10.625, "learning_rate": 8.821631608868675e-06, "loss": 0.5225, "step": 33664 }, { "epoch": 3.5821451372632476, "grad_norm": 15.21875, "learning_rate": 8.821596077316657e-06, "loss": 0.6517, "step": 33665 }, { "epoch": 3.582251542881464, "grad_norm": 11.625, "learning_rate": 8.82156054576464e-06, "loss": 0.5348, "step": 33666 }, { "epoch": 3.5823579484996806, "grad_norm": 13.6171875, "learning_rate": 8.821525014212622e-06, "loss": 0.726, "step": 33667 }, { "epoch": 3.5824643541178975, "grad_norm": 13.8125, "learning_rate": 8.821489482660603e-06, "loss": 0.685, "step": 33668 }, { "epoch": 3.582570759736114, "grad_norm": 14.234375, "learning_rate": 8.821453951108585e-06, "loss": 0.8412, "step": 33669 }, { "epoch": 3.5826771653543306, "grad_norm": 10.5859375, "learning_rate": 8.821418419556567e-06, "loss": 0.6022, "step": 33670 }, { "epoch": 3.5827835709725475, "grad_norm": 14.8984375, "learning_rate": 8.821382888004548e-06, "loss": 0.543, "step": 33671 }, { "epoch": 3.582889976590764, "grad_norm": 13.9296875, "learning_rate": 8.821347356452532e-06, "loss": 0.7395, "step": 33672 }, { "epoch": 3.5829963822089805, "grad_norm": 11.6015625, "learning_rate": 8.821311824900513e-06, "loss": 0.5652, "step": 33673 }, { "epoch": 3.5831027878271975, "grad_norm": 12.109375, "learning_rate": 8.821276293348495e-06, "loss": 0.6401, "step": 33674 }, { "epoch": 3.583209193445414, "grad_norm": 12.6796875, "learning_rate": 8.821240761796476e-06, "loss": 0.7025, "step": 33675 }, { "epoch": 3.5833155990636305, "grad_norm": 11.3203125, "learning_rate": 8.821205230244458e-06, "loss": 0.5954, "step": 33676 }, { "epoch": 3.5834220046818475, "grad_norm": 14.015625, "learning_rate": 8.82116969869244e-06, "loss": 0.7966, "step": 33677 }, { "epoch": 3.583528410300064, "grad_norm": 12.203125, "learning_rate": 8.821134167140421e-06, "loss": 0.6668, "step": 33678 }, { "epoch": 3.5836348159182805, "grad_norm": 13.0078125, "learning_rate": 8.821098635588403e-06, "loss": 0.681, "step": 33679 }, { "epoch": 3.583741221536497, "grad_norm": 14.2578125, "learning_rate": 8.821063104036384e-06, "loss": 0.719, "step": 33680 }, { "epoch": 3.5838476271547135, "grad_norm": 13.78125, "learning_rate": 8.821027572484366e-06, "loss": 0.687, "step": 33681 }, { "epoch": 3.5839540327729305, "grad_norm": 14.7421875, "learning_rate": 8.82099204093235e-06, "loss": 0.6208, "step": 33682 }, { "epoch": 3.584060438391147, "grad_norm": 15.4921875, "learning_rate": 8.820956509380331e-06, "loss": 0.7394, "step": 33683 }, { "epoch": 3.5841668440093635, "grad_norm": 12.9375, "learning_rate": 8.820920977828313e-06, "loss": 0.6205, "step": 33684 }, { "epoch": 3.5842732496275804, "grad_norm": 13.7578125, "learning_rate": 8.820885446276294e-06, "loss": 0.6573, "step": 33685 }, { "epoch": 3.584379655245797, "grad_norm": 10.7109375, "learning_rate": 8.820849914724276e-06, "loss": 0.6017, "step": 33686 }, { "epoch": 3.5844860608640134, "grad_norm": 11.265625, "learning_rate": 8.820814383172257e-06, "loss": 0.6702, "step": 33687 }, { "epoch": 3.5845924664822304, "grad_norm": 13.34375, "learning_rate": 8.820778851620239e-06, "loss": 0.6101, "step": 33688 }, { "epoch": 3.584698872100447, "grad_norm": 11.1953125, "learning_rate": 8.82074332006822e-06, "loss": 0.4672, "step": 33689 }, { "epoch": 3.5848052777186634, "grad_norm": 13.703125, "learning_rate": 8.820707788516202e-06, "loss": 0.7356, "step": 33690 }, { "epoch": 3.5849116833368804, "grad_norm": 13.03125, "learning_rate": 8.820672256964184e-06, "loss": 0.7349, "step": 33691 }, { "epoch": 3.585018088955097, "grad_norm": 9.1171875, "learning_rate": 8.820636725412167e-06, "loss": 0.553, "step": 33692 }, { "epoch": 3.5851244945733134, "grad_norm": 12.25, "learning_rate": 8.820601193860149e-06, "loss": 0.7087, "step": 33693 }, { "epoch": 3.5852309001915303, "grad_norm": 13.203125, "learning_rate": 8.82056566230813e-06, "loss": 0.7003, "step": 33694 }, { "epoch": 3.585337305809747, "grad_norm": 15.28125, "learning_rate": 8.820530130756112e-06, "loss": 0.6348, "step": 33695 }, { "epoch": 3.5854437114279634, "grad_norm": 11.9453125, "learning_rate": 8.820494599204094e-06, "loss": 0.5804, "step": 33696 }, { "epoch": 3.58555011704618, "grad_norm": 17.28125, "learning_rate": 8.820459067652077e-06, "loss": 0.6984, "step": 33697 }, { "epoch": 3.585656522664397, "grad_norm": 12.3359375, "learning_rate": 8.820423536100058e-06, "loss": 0.7125, "step": 33698 }, { "epoch": 3.5857629282826133, "grad_norm": 13.5234375, "learning_rate": 8.82038800454804e-06, "loss": 0.6825, "step": 33699 }, { "epoch": 3.58586933390083, "grad_norm": 9.859375, "learning_rate": 8.820352472996022e-06, "loss": 0.6213, "step": 33700 }, { "epoch": 3.5859757395190464, "grad_norm": 12.8984375, "learning_rate": 8.820316941444003e-06, "loss": 0.6142, "step": 33701 }, { "epoch": 3.5860821451372633, "grad_norm": 12.2109375, "learning_rate": 8.820281409891985e-06, "loss": 0.5271, "step": 33702 }, { "epoch": 3.58618855075548, "grad_norm": 17.015625, "learning_rate": 8.820245878339966e-06, "loss": 0.5615, "step": 33703 }, { "epoch": 3.5862949563736963, "grad_norm": 14.625, "learning_rate": 8.820210346787948e-06, "loss": 0.5946, "step": 33704 }, { "epoch": 3.5864013619919133, "grad_norm": 11.546875, "learning_rate": 8.82017481523593e-06, "loss": 0.4736, "step": 33705 }, { "epoch": 3.58650776761013, "grad_norm": 16.921875, "learning_rate": 8.820139283683911e-06, "loss": 0.8335, "step": 33706 }, { "epoch": 3.5866141732283463, "grad_norm": 11.140625, "learning_rate": 8.820103752131893e-06, "loss": 0.5512, "step": 33707 }, { "epoch": 3.5867205788465633, "grad_norm": 12.09375, "learning_rate": 8.820068220579876e-06, "loss": 0.6665, "step": 33708 }, { "epoch": 3.5868269844647798, "grad_norm": 14.0546875, "learning_rate": 8.820032689027858e-06, "loss": 0.6382, "step": 33709 }, { "epoch": 3.5869333900829963, "grad_norm": 13.1796875, "learning_rate": 8.81999715747584e-06, "loss": 0.5847, "step": 33710 }, { "epoch": 3.5870397957012132, "grad_norm": 11.984375, "learning_rate": 8.819961625923821e-06, "loss": 0.7905, "step": 33711 }, { "epoch": 3.5871462013194297, "grad_norm": 15.984375, "learning_rate": 8.819926094371803e-06, "loss": 0.7685, "step": 33712 }, { "epoch": 3.5872526069376462, "grad_norm": 10.7890625, "learning_rate": 8.819890562819784e-06, "loss": 0.5047, "step": 33713 }, { "epoch": 3.587359012555863, "grad_norm": 11.4296875, "learning_rate": 8.819855031267766e-06, "loss": 0.6103, "step": 33714 }, { "epoch": 3.5874654181740797, "grad_norm": 12.1484375, "learning_rate": 8.819819499715747e-06, "loss": 0.6182, "step": 33715 }, { "epoch": 3.587571823792296, "grad_norm": 11.9921875, "learning_rate": 8.819783968163729e-06, "loss": 0.5663, "step": 33716 }, { "epoch": 3.5876782294105127, "grad_norm": 13.1953125, "learning_rate": 8.81974843661171e-06, "loss": 0.5541, "step": 33717 }, { "epoch": 3.5877846350287292, "grad_norm": 14.8125, "learning_rate": 8.819712905059694e-06, "loss": 0.7044, "step": 33718 }, { "epoch": 3.587891040646946, "grad_norm": 17.59375, "learning_rate": 8.819677373507676e-06, "loss": 0.7018, "step": 33719 }, { "epoch": 3.5879974462651627, "grad_norm": 13.328125, "learning_rate": 8.819641841955657e-06, "loss": 0.6711, "step": 33720 }, { "epoch": 3.588103851883379, "grad_norm": 18.859375, "learning_rate": 8.819606310403639e-06, "loss": 0.7085, "step": 33721 }, { "epoch": 3.588210257501596, "grad_norm": 17.234375, "learning_rate": 8.81957077885162e-06, "loss": 0.6738, "step": 33722 }, { "epoch": 3.5883166631198127, "grad_norm": 12.2265625, "learning_rate": 8.819535247299604e-06, "loss": 0.6845, "step": 33723 }, { "epoch": 3.588423068738029, "grad_norm": 13.125, "learning_rate": 8.819499715747585e-06, "loss": 0.6301, "step": 33724 }, { "epoch": 3.588529474356246, "grad_norm": 13.2734375, "learning_rate": 8.819464184195567e-06, "loss": 0.6389, "step": 33725 }, { "epoch": 3.5886358799744627, "grad_norm": 19.5625, "learning_rate": 8.819428652643548e-06, "loss": 0.884, "step": 33726 }, { "epoch": 3.588742285592679, "grad_norm": 12.078125, "learning_rate": 8.81939312109153e-06, "loss": 0.6715, "step": 33727 }, { "epoch": 3.588848691210896, "grad_norm": 28.40625, "learning_rate": 8.819357589539512e-06, "loss": 0.7931, "step": 33728 }, { "epoch": 3.5889550968291126, "grad_norm": 14.0, "learning_rate": 8.819322057987493e-06, "loss": 0.6311, "step": 33729 }, { "epoch": 3.589061502447329, "grad_norm": 17.90625, "learning_rate": 8.819286526435475e-06, "loss": 0.8826, "step": 33730 }, { "epoch": 3.589167908065546, "grad_norm": 15.1171875, "learning_rate": 8.819250994883457e-06, "loss": 0.8696, "step": 33731 }, { "epoch": 3.5892743136837626, "grad_norm": 15.6015625, "learning_rate": 8.819215463331438e-06, "loss": 0.8985, "step": 33732 }, { "epoch": 3.589380719301979, "grad_norm": 15.15625, "learning_rate": 8.819179931779421e-06, "loss": 0.7907, "step": 33733 }, { "epoch": 3.589487124920196, "grad_norm": 15.3046875, "learning_rate": 8.819144400227403e-06, "loss": 0.6406, "step": 33734 }, { "epoch": 3.5895935305384126, "grad_norm": 13.2421875, "learning_rate": 8.819108868675385e-06, "loss": 0.6155, "step": 33735 }, { "epoch": 3.589699936156629, "grad_norm": 15.125, "learning_rate": 8.819073337123366e-06, "loss": 0.6951, "step": 33736 }, { "epoch": 3.5898063417748456, "grad_norm": 19.59375, "learning_rate": 8.819037805571348e-06, "loss": 0.8211, "step": 33737 }, { "epoch": 3.589912747393062, "grad_norm": 20.171875, "learning_rate": 8.819002274019331e-06, "loss": 0.8182, "step": 33738 }, { "epoch": 3.590019153011279, "grad_norm": 13.453125, "learning_rate": 8.818966742467313e-06, "loss": 0.6704, "step": 33739 }, { "epoch": 3.5901255586294956, "grad_norm": 12.875, "learning_rate": 8.818931210915294e-06, "loss": 0.7598, "step": 33740 }, { "epoch": 3.590231964247712, "grad_norm": 18.90625, "learning_rate": 8.818895679363276e-06, "loss": 0.7846, "step": 33741 }, { "epoch": 3.590338369865929, "grad_norm": 11.03125, "learning_rate": 8.818860147811258e-06, "loss": 0.4865, "step": 33742 }, { "epoch": 3.5904447754841455, "grad_norm": 14.359375, "learning_rate": 8.81882461625924e-06, "loss": 0.6545, "step": 33743 }, { "epoch": 3.590551181102362, "grad_norm": 13.1484375, "learning_rate": 8.81878908470722e-06, "loss": 0.807, "step": 33744 }, { "epoch": 3.590657586720579, "grad_norm": 18.578125, "learning_rate": 8.818753553155202e-06, "loss": 0.7917, "step": 33745 }, { "epoch": 3.5907639923387955, "grad_norm": 12.09375, "learning_rate": 8.818718021603184e-06, "loss": 0.5783, "step": 33746 }, { "epoch": 3.590870397957012, "grad_norm": 13.078125, "learning_rate": 8.818682490051166e-06, "loss": 0.6949, "step": 33747 }, { "epoch": 3.590976803575229, "grad_norm": 14.4375, "learning_rate": 8.818646958499147e-06, "loss": 0.7364, "step": 33748 }, { "epoch": 3.5910832091934455, "grad_norm": 12.7578125, "learning_rate": 8.81861142694713e-06, "loss": 0.6835, "step": 33749 }, { "epoch": 3.591189614811662, "grad_norm": 12.5703125, "learning_rate": 8.818575895395112e-06, "loss": 0.6722, "step": 33750 }, { "epoch": 3.591296020429879, "grad_norm": 13.578125, "learning_rate": 8.818540363843094e-06, "loss": 0.693, "step": 33751 }, { "epoch": 3.5914024260480955, "grad_norm": 19.0625, "learning_rate": 8.818504832291075e-06, "loss": 0.7157, "step": 33752 }, { "epoch": 3.591508831666312, "grad_norm": 18.125, "learning_rate": 8.818469300739057e-06, "loss": 0.5482, "step": 33753 }, { "epoch": 3.5916152372845285, "grad_norm": 14.9140625, "learning_rate": 8.818433769187039e-06, "loss": 0.758, "step": 33754 }, { "epoch": 3.5917216429027454, "grad_norm": 13.4375, "learning_rate": 8.81839823763502e-06, "loss": 0.6149, "step": 33755 }, { "epoch": 3.591828048520962, "grad_norm": 10.8828125, "learning_rate": 8.818362706083002e-06, "loss": 0.6849, "step": 33756 }, { "epoch": 3.5919344541391784, "grad_norm": 11.375, "learning_rate": 8.818327174530983e-06, "loss": 0.6775, "step": 33757 }, { "epoch": 3.592040859757395, "grad_norm": 13.9921875, "learning_rate": 8.818291642978965e-06, "loss": 0.7417, "step": 33758 }, { "epoch": 3.592147265375612, "grad_norm": 13.5625, "learning_rate": 8.818256111426948e-06, "loss": 0.5794, "step": 33759 }, { "epoch": 3.5922536709938284, "grad_norm": 11.421875, "learning_rate": 8.81822057987493e-06, "loss": 0.5655, "step": 33760 }, { "epoch": 3.592360076612045, "grad_norm": 13.6484375, "learning_rate": 8.818185048322911e-06, "loss": 0.5624, "step": 33761 }, { "epoch": 3.592466482230262, "grad_norm": 13.8046875, "learning_rate": 8.818149516770893e-06, "loss": 0.623, "step": 33762 }, { "epoch": 3.5925728878484784, "grad_norm": 18.921875, "learning_rate": 8.818113985218875e-06, "loss": 0.7671, "step": 33763 }, { "epoch": 3.592679293466695, "grad_norm": 13.9296875, "learning_rate": 8.818078453666858e-06, "loss": 0.5309, "step": 33764 }, { "epoch": 3.592785699084912, "grad_norm": 10.1953125, "learning_rate": 8.81804292211484e-06, "loss": 0.5823, "step": 33765 }, { "epoch": 3.5928921047031284, "grad_norm": 14.015625, "learning_rate": 8.818007390562821e-06, "loss": 0.6693, "step": 33766 }, { "epoch": 3.592998510321345, "grad_norm": 13.109375, "learning_rate": 8.817971859010803e-06, "loss": 0.6467, "step": 33767 }, { "epoch": 3.593104915939562, "grad_norm": 11.140625, "learning_rate": 8.817936327458784e-06, "loss": 0.5636, "step": 33768 }, { "epoch": 3.5932113215577783, "grad_norm": 14.8984375, "learning_rate": 8.817900795906766e-06, "loss": 0.7378, "step": 33769 }, { "epoch": 3.593317727175995, "grad_norm": 11.828125, "learning_rate": 8.817865264354748e-06, "loss": 0.6628, "step": 33770 }, { "epoch": 3.593424132794212, "grad_norm": 13.3046875, "learning_rate": 8.81782973280273e-06, "loss": 0.6558, "step": 33771 }, { "epoch": 3.5935305384124283, "grad_norm": 12.578125, "learning_rate": 8.817794201250711e-06, "loss": 0.7035, "step": 33772 }, { "epoch": 3.593636944030645, "grad_norm": 18.171875, "learning_rate": 8.817758669698692e-06, "loss": 0.8097, "step": 33773 }, { "epoch": 3.5937433496488613, "grad_norm": 13.484375, "learning_rate": 8.817723138146676e-06, "loss": 0.7031, "step": 33774 }, { "epoch": 3.593849755267078, "grad_norm": 15.1015625, "learning_rate": 8.817687606594657e-06, "loss": 0.8183, "step": 33775 }, { "epoch": 3.593956160885295, "grad_norm": 10.609375, "learning_rate": 8.817652075042639e-06, "loss": 0.6597, "step": 33776 }, { "epoch": 3.5940625665035113, "grad_norm": 14.5078125, "learning_rate": 8.81761654349062e-06, "loss": 0.7249, "step": 33777 }, { "epoch": 3.594168972121728, "grad_norm": 11.9921875, "learning_rate": 8.817581011938602e-06, "loss": 0.5374, "step": 33778 }, { "epoch": 3.5942753777399448, "grad_norm": 13.2109375, "learning_rate": 8.817545480386584e-06, "loss": 0.6028, "step": 33779 }, { "epoch": 3.5943817833581613, "grad_norm": 15.5625, "learning_rate": 8.817509948834565e-06, "loss": 0.6058, "step": 33780 }, { "epoch": 3.594488188976378, "grad_norm": 11.953125, "learning_rate": 8.817474417282547e-06, "loss": 0.5576, "step": 33781 }, { "epoch": 3.5945945945945947, "grad_norm": 17.0, "learning_rate": 8.817438885730529e-06, "loss": 0.781, "step": 33782 }, { "epoch": 3.5947010002128112, "grad_norm": 17.0625, "learning_rate": 8.81740335417851e-06, "loss": 0.7326, "step": 33783 }, { "epoch": 3.5948074058310278, "grad_norm": 11.0234375, "learning_rate": 8.817367822626494e-06, "loss": 0.6088, "step": 33784 }, { "epoch": 3.5949138114492447, "grad_norm": 10.828125, "learning_rate": 8.817332291074475e-06, "loss": 0.5642, "step": 33785 }, { "epoch": 3.595020217067461, "grad_norm": 10.75, "learning_rate": 8.817296759522457e-06, "loss": 0.5281, "step": 33786 }, { "epoch": 3.5951266226856777, "grad_norm": 13.6953125, "learning_rate": 8.817261227970438e-06, "loss": 0.6566, "step": 33787 }, { "epoch": 3.5952330283038947, "grad_norm": 15.0078125, "learning_rate": 8.81722569641842e-06, "loss": 0.7268, "step": 33788 }, { "epoch": 3.595339433922111, "grad_norm": 14.0234375, "learning_rate": 8.817190164866402e-06, "loss": 0.6381, "step": 33789 }, { "epoch": 3.5954458395403277, "grad_norm": 16.71875, "learning_rate": 8.817154633314385e-06, "loss": 0.7452, "step": 33790 }, { "epoch": 3.595552245158544, "grad_norm": 10.7578125, "learning_rate": 8.817119101762366e-06, "loss": 0.5958, "step": 33791 }, { "epoch": 3.595658650776761, "grad_norm": 14.5546875, "learning_rate": 8.817083570210348e-06, "loss": 0.7095, "step": 33792 }, { "epoch": 3.5957650563949777, "grad_norm": 10.3203125, "learning_rate": 8.81704803865833e-06, "loss": 0.5695, "step": 33793 }, { "epoch": 3.595871462013194, "grad_norm": 15.8125, "learning_rate": 8.817012507106311e-06, "loss": 0.7956, "step": 33794 }, { "epoch": 3.5959778676314107, "grad_norm": 11.3515625, "learning_rate": 8.816976975554293e-06, "loss": 0.5915, "step": 33795 }, { "epoch": 3.5960842732496276, "grad_norm": 15.640625, "learning_rate": 8.816941444002275e-06, "loss": 0.7528, "step": 33796 }, { "epoch": 3.596190678867844, "grad_norm": 12.3203125, "learning_rate": 8.816905912450256e-06, "loss": 0.7001, "step": 33797 }, { "epoch": 3.5962970844860607, "grad_norm": 11.265625, "learning_rate": 8.816870380898238e-06, "loss": 0.5934, "step": 33798 }, { "epoch": 3.5964034901042776, "grad_norm": 13.03125, "learning_rate": 8.81683484934622e-06, "loss": 0.5637, "step": 33799 }, { "epoch": 3.596509895722494, "grad_norm": 12.796875, "learning_rate": 8.816799317794203e-06, "loss": 0.6935, "step": 33800 }, { "epoch": 3.5966163013407106, "grad_norm": 17.296875, "learning_rate": 8.816763786242184e-06, "loss": 0.679, "step": 33801 }, { "epoch": 3.5967227069589276, "grad_norm": 15.8046875, "learning_rate": 8.816728254690166e-06, "loss": 0.7878, "step": 33802 }, { "epoch": 3.596829112577144, "grad_norm": 13.234375, "learning_rate": 8.816692723138147e-06, "loss": 0.6756, "step": 33803 }, { "epoch": 3.5969355181953606, "grad_norm": 14.171875, "learning_rate": 8.816657191586129e-06, "loss": 0.8036, "step": 33804 }, { "epoch": 3.5970419238135776, "grad_norm": 11.6484375, "learning_rate": 8.81662166003411e-06, "loss": 0.5461, "step": 33805 }, { "epoch": 3.597148329431794, "grad_norm": 12.5625, "learning_rate": 8.816586128482092e-06, "loss": 0.6761, "step": 33806 }, { "epoch": 3.5972547350500106, "grad_norm": 12.4375, "learning_rate": 8.816550596930074e-06, "loss": 0.6357, "step": 33807 }, { "epoch": 3.5973611406682275, "grad_norm": 13.1015625, "learning_rate": 8.816515065378055e-06, "loss": 0.5977, "step": 33808 }, { "epoch": 3.597467546286444, "grad_norm": 14.0859375, "learning_rate": 8.816479533826037e-06, "loss": 0.6126, "step": 33809 }, { "epoch": 3.5975739519046606, "grad_norm": 10.796875, "learning_rate": 8.81644400227402e-06, "loss": 0.5277, "step": 33810 }, { "epoch": 3.597680357522877, "grad_norm": 12.203125, "learning_rate": 8.816408470722002e-06, "loss": 0.6687, "step": 33811 }, { "epoch": 3.5977867631410936, "grad_norm": 24.203125, "learning_rate": 8.816372939169984e-06, "loss": 0.8107, "step": 33812 }, { "epoch": 3.5978931687593105, "grad_norm": 12.90625, "learning_rate": 8.816337407617965e-06, "loss": 0.6763, "step": 33813 }, { "epoch": 3.597999574377527, "grad_norm": 14.3046875, "learning_rate": 8.816301876065947e-06, "loss": 0.745, "step": 33814 }, { "epoch": 3.5981059799957436, "grad_norm": 12.734375, "learning_rate": 8.81626634451393e-06, "loss": 0.5255, "step": 33815 }, { "epoch": 3.5982123856139605, "grad_norm": 12.8046875, "learning_rate": 8.816230812961912e-06, "loss": 0.6817, "step": 33816 }, { "epoch": 3.598318791232177, "grad_norm": 15.3046875, "learning_rate": 8.816195281409893e-06, "loss": 0.7141, "step": 33817 }, { "epoch": 3.5984251968503935, "grad_norm": 15.828125, "learning_rate": 8.816159749857875e-06, "loss": 0.7142, "step": 33818 }, { "epoch": 3.5985316024686105, "grad_norm": 12.9296875, "learning_rate": 8.816124218305857e-06, "loss": 0.532, "step": 33819 }, { "epoch": 3.598638008086827, "grad_norm": 13.515625, "learning_rate": 8.816088686753838e-06, "loss": 0.677, "step": 33820 }, { "epoch": 3.5987444137050435, "grad_norm": 13.953125, "learning_rate": 8.81605315520182e-06, "loss": 0.5965, "step": 33821 }, { "epoch": 3.5988508193232605, "grad_norm": 12.5078125, "learning_rate": 8.816017623649801e-06, "loss": 0.7467, "step": 33822 }, { "epoch": 3.598957224941477, "grad_norm": 11.5546875, "learning_rate": 8.815982092097783e-06, "loss": 0.7169, "step": 33823 }, { "epoch": 3.5990636305596935, "grad_norm": 14.453125, "learning_rate": 8.815946560545765e-06, "loss": 0.6235, "step": 33824 }, { "epoch": 3.5991700361779104, "grad_norm": 10.46875, "learning_rate": 8.815911028993746e-06, "loss": 0.5565, "step": 33825 }, { "epoch": 3.599276441796127, "grad_norm": 13.5078125, "learning_rate": 8.81587549744173e-06, "loss": 0.5909, "step": 33826 }, { "epoch": 3.5993828474143434, "grad_norm": 11.1875, "learning_rate": 8.815839965889711e-06, "loss": 0.5644, "step": 33827 }, { "epoch": 3.59948925303256, "grad_norm": 18.359375, "learning_rate": 8.815804434337693e-06, "loss": 0.6593, "step": 33828 }, { "epoch": 3.599595658650777, "grad_norm": 12.9453125, "learning_rate": 8.815768902785674e-06, "loss": 0.7126, "step": 33829 }, { "epoch": 3.5997020642689934, "grad_norm": 13.1328125, "learning_rate": 8.815733371233656e-06, "loss": 0.6238, "step": 33830 }, { "epoch": 3.59980846988721, "grad_norm": 12.59375, "learning_rate": 8.81569783968164e-06, "loss": 0.5453, "step": 33831 }, { "epoch": 3.5999148755054264, "grad_norm": 9.96875, "learning_rate": 8.81566230812962e-06, "loss": 0.469, "step": 33832 }, { "epoch": 3.6000212811236434, "grad_norm": 11.59375, "learning_rate": 8.8156267765776e-06, "loss": 0.5855, "step": 33833 }, { "epoch": 3.60012768674186, "grad_norm": 13.046875, "learning_rate": 8.815591245025582e-06, "loss": 0.6463, "step": 33834 }, { "epoch": 3.6002340923600764, "grad_norm": 19.03125, "learning_rate": 8.815555713473564e-06, "loss": 1.0431, "step": 33835 }, { "epoch": 3.6003404979782934, "grad_norm": 10.59375, "learning_rate": 8.815520181921547e-06, "loss": 0.5329, "step": 33836 }, { "epoch": 3.60044690359651, "grad_norm": 11.9765625, "learning_rate": 8.815484650369529e-06, "loss": 0.8205, "step": 33837 }, { "epoch": 3.6005533092147264, "grad_norm": 11.359375, "learning_rate": 8.81544911881751e-06, "loss": 0.6683, "step": 33838 }, { "epoch": 3.6006597148329433, "grad_norm": 12.796875, "learning_rate": 8.815413587265492e-06, "loss": 0.6361, "step": 33839 }, { "epoch": 3.60076612045116, "grad_norm": 13.40625, "learning_rate": 8.815378055713474e-06, "loss": 0.6541, "step": 33840 }, { "epoch": 3.6008725260693764, "grad_norm": 14.8671875, "learning_rate": 8.815342524161457e-06, "loss": 0.7482, "step": 33841 }, { "epoch": 3.6009789316875933, "grad_norm": 11.984375, "learning_rate": 8.815306992609439e-06, "loss": 0.626, "step": 33842 }, { "epoch": 3.60108533730581, "grad_norm": 10.484375, "learning_rate": 8.81527146105742e-06, "loss": 0.6162, "step": 33843 }, { "epoch": 3.6011917429240263, "grad_norm": 9.8671875, "learning_rate": 8.815235929505402e-06, "loss": 0.6317, "step": 33844 }, { "epoch": 3.6012981485422433, "grad_norm": 11.0078125, "learning_rate": 8.815200397953383e-06, "loss": 0.6942, "step": 33845 }, { "epoch": 3.60140455416046, "grad_norm": 16.171875, "learning_rate": 8.815164866401365e-06, "loss": 0.7119, "step": 33846 }, { "epoch": 3.6015109597786763, "grad_norm": 10.625, "learning_rate": 8.815129334849347e-06, "loss": 0.5504, "step": 33847 }, { "epoch": 3.601617365396893, "grad_norm": 11.328125, "learning_rate": 8.815093803297328e-06, "loss": 0.5793, "step": 33848 }, { "epoch": 3.6017237710151093, "grad_norm": 11.8515625, "learning_rate": 8.81505827174531e-06, "loss": 0.6403, "step": 33849 }, { "epoch": 3.6018301766333263, "grad_norm": 13.2421875, "learning_rate": 8.815022740193291e-06, "loss": 0.6181, "step": 33850 }, { "epoch": 3.601936582251543, "grad_norm": 11.0859375, "learning_rate": 8.814987208641275e-06, "loss": 0.6032, "step": 33851 }, { "epoch": 3.6020429878697593, "grad_norm": 13.4453125, "learning_rate": 8.814951677089256e-06, "loss": 0.6822, "step": 33852 }, { "epoch": 3.6021493934879762, "grad_norm": 14.03125, "learning_rate": 8.814916145537238e-06, "loss": 0.6417, "step": 33853 }, { "epoch": 3.6022557991061928, "grad_norm": 11.25, "learning_rate": 8.81488061398522e-06, "loss": 0.53, "step": 33854 }, { "epoch": 3.6023622047244093, "grad_norm": 11.890625, "learning_rate": 8.814845082433201e-06, "loss": 0.5734, "step": 33855 }, { "epoch": 3.602468610342626, "grad_norm": 11.40625, "learning_rate": 8.814809550881184e-06, "loss": 0.5877, "step": 33856 }, { "epoch": 3.6025750159608427, "grad_norm": 11.1484375, "learning_rate": 8.814774019329166e-06, "loss": 0.5443, "step": 33857 }, { "epoch": 3.6026814215790592, "grad_norm": 15.1953125, "learning_rate": 8.814738487777148e-06, "loss": 0.7658, "step": 33858 }, { "epoch": 3.602787827197276, "grad_norm": 11.046875, "learning_rate": 8.81470295622513e-06, "loss": 0.5396, "step": 33859 }, { "epoch": 3.6028942328154927, "grad_norm": 12.2890625, "learning_rate": 8.814667424673111e-06, "loss": 0.5573, "step": 33860 }, { "epoch": 3.603000638433709, "grad_norm": 13.234375, "learning_rate": 8.814631893121092e-06, "loss": 0.5822, "step": 33861 }, { "epoch": 3.603107044051926, "grad_norm": 20.578125, "learning_rate": 8.814596361569074e-06, "loss": 1.0419, "step": 33862 }, { "epoch": 3.6032134496701427, "grad_norm": 13.046875, "learning_rate": 8.814560830017056e-06, "loss": 0.6108, "step": 33863 }, { "epoch": 3.603319855288359, "grad_norm": 12.890625, "learning_rate": 8.814525298465037e-06, "loss": 0.5944, "step": 33864 }, { "epoch": 3.603426260906576, "grad_norm": 14.6640625, "learning_rate": 8.814489766913019e-06, "loss": 0.7151, "step": 33865 }, { "epoch": 3.6035326665247926, "grad_norm": 12.28125, "learning_rate": 8.814454235361e-06, "loss": 0.7706, "step": 33866 }, { "epoch": 3.603639072143009, "grad_norm": 17.71875, "learning_rate": 8.814418703808984e-06, "loss": 0.8121, "step": 33867 }, { "epoch": 3.6037454777612257, "grad_norm": 13.546875, "learning_rate": 8.814383172256965e-06, "loss": 0.6901, "step": 33868 }, { "epoch": 3.603851883379442, "grad_norm": 13.2578125, "learning_rate": 8.814347640704947e-06, "loss": 0.6618, "step": 33869 }, { "epoch": 3.603958288997659, "grad_norm": 12.8984375, "learning_rate": 8.814312109152929e-06, "loss": 0.6047, "step": 33870 }, { "epoch": 3.6040646946158756, "grad_norm": 11.5625, "learning_rate": 8.81427657760091e-06, "loss": 0.6338, "step": 33871 }, { "epoch": 3.604171100234092, "grad_norm": 12.0, "learning_rate": 8.814241046048892e-06, "loss": 0.6368, "step": 33872 }, { "epoch": 3.604277505852309, "grad_norm": 15.3359375, "learning_rate": 8.814205514496873e-06, "loss": 0.7266, "step": 33873 }, { "epoch": 3.6043839114705256, "grad_norm": 10.546875, "learning_rate": 8.814169982944855e-06, "loss": 0.6643, "step": 33874 }, { "epoch": 3.604490317088742, "grad_norm": 15.0390625, "learning_rate": 8.814134451392837e-06, "loss": 0.6802, "step": 33875 }, { "epoch": 3.604596722706959, "grad_norm": 11.3828125, "learning_rate": 8.814098919840818e-06, "loss": 0.4974, "step": 33876 }, { "epoch": 3.6047031283251756, "grad_norm": 10.59375, "learning_rate": 8.814063388288802e-06, "loss": 0.6029, "step": 33877 }, { "epoch": 3.604809533943392, "grad_norm": 10.265625, "learning_rate": 8.814027856736783e-06, "loss": 0.7089, "step": 33878 }, { "epoch": 3.604915939561609, "grad_norm": 14.6953125, "learning_rate": 8.813992325184765e-06, "loss": 0.7285, "step": 33879 }, { "epoch": 3.6050223451798256, "grad_norm": 12.140625, "learning_rate": 8.813956793632746e-06, "loss": 0.5802, "step": 33880 }, { "epoch": 3.605128750798042, "grad_norm": 19.453125, "learning_rate": 8.813921262080728e-06, "loss": 0.7099, "step": 33881 }, { "epoch": 3.605235156416259, "grad_norm": 13.3125, "learning_rate": 8.813885730528711e-06, "loss": 0.8454, "step": 33882 }, { "epoch": 3.6053415620344755, "grad_norm": 21.671875, "learning_rate": 8.813850198976693e-06, "loss": 0.6616, "step": 33883 }, { "epoch": 3.605447967652692, "grad_norm": 13.890625, "learning_rate": 8.813814667424675e-06, "loss": 0.6481, "step": 33884 }, { "epoch": 3.6055543732709086, "grad_norm": 12.5546875, "learning_rate": 8.813779135872656e-06, "loss": 0.5965, "step": 33885 }, { "epoch": 3.6056607788891255, "grad_norm": 16.8125, "learning_rate": 8.813743604320638e-06, "loss": 0.7058, "step": 33886 }, { "epoch": 3.605767184507342, "grad_norm": 15.0, "learning_rate": 8.81370807276862e-06, "loss": 0.6509, "step": 33887 }, { "epoch": 3.6058735901255585, "grad_norm": 10.125, "learning_rate": 8.813672541216601e-06, "loss": 0.5535, "step": 33888 }, { "epoch": 3.605979995743775, "grad_norm": 12.609375, "learning_rate": 8.813637009664583e-06, "loss": 0.508, "step": 33889 }, { "epoch": 3.606086401361992, "grad_norm": 13.015625, "learning_rate": 8.813601478112564e-06, "loss": 0.6662, "step": 33890 }, { "epoch": 3.6061928069802085, "grad_norm": 18.296875, "learning_rate": 8.813565946560546e-06, "loss": 0.8073, "step": 33891 }, { "epoch": 3.606299212598425, "grad_norm": 11.3359375, "learning_rate": 8.813530415008529e-06, "loss": 0.6183, "step": 33892 }, { "epoch": 3.606405618216642, "grad_norm": 14.7890625, "learning_rate": 8.81349488345651e-06, "loss": 0.6312, "step": 33893 }, { "epoch": 3.6065120238348585, "grad_norm": 18.890625, "learning_rate": 8.813459351904492e-06, "loss": 0.787, "step": 33894 }, { "epoch": 3.606618429453075, "grad_norm": 13.8046875, "learning_rate": 8.813423820352474e-06, "loss": 0.5804, "step": 33895 }, { "epoch": 3.606724835071292, "grad_norm": 11.453125, "learning_rate": 8.813388288800456e-06, "loss": 0.5122, "step": 33896 }, { "epoch": 3.6068312406895084, "grad_norm": 11.3515625, "learning_rate": 8.813352757248437e-06, "loss": 0.6001, "step": 33897 }, { "epoch": 3.606937646307725, "grad_norm": 10.6875, "learning_rate": 8.813317225696419e-06, "loss": 0.5378, "step": 33898 }, { "epoch": 3.607044051925942, "grad_norm": 12.6875, "learning_rate": 8.8132816941444e-06, "loss": 0.736, "step": 33899 }, { "epoch": 3.6071504575441584, "grad_norm": 14.96875, "learning_rate": 8.813246162592382e-06, "loss": 0.6371, "step": 33900 }, { "epoch": 3.607256863162375, "grad_norm": 17.53125, "learning_rate": 8.813210631040364e-06, "loss": 0.7417, "step": 33901 }, { "epoch": 3.607363268780592, "grad_norm": 17.0625, "learning_rate": 8.813175099488345e-06, "loss": 0.7255, "step": 33902 }, { "epoch": 3.6074696743988084, "grad_norm": 12.0546875, "learning_rate": 8.813139567936328e-06, "loss": 0.5583, "step": 33903 }, { "epoch": 3.607576080017025, "grad_norm": 11.1328125, "learning_rate": 8.81310403638431e-06, "loss": 0.5907, "step": 33904 }, { "epoch": 3.6076824856352414, "grad_norm": 13.9609375, "learning_rate": 8.813068504832292e-06, "loss": 0.6829, "step": 33905 }, { "epoch": 3.607788891253458, "grad_norm": 15.7578125, "learning_rate": 8.813032973280273e-06, "loss": 0.5518, "step": 33906 }, { "epoch": 3.607895296871675, "grad_norm": 13.0078125, "learning_rate": 8.812997441728255e-06, "loss": 0.7578, "step": 33907 }, { "epoch": 3.6080017024898914, "grad_norm": 15.171875, "learning_rate": 8.812961910176238e-06, "loss": 0.6614, "step": 33908 }, { "epoch": 3.608108108108108, "grad_norm": 13.1171875, "learning_rate": 8.81292637862422e-06, "loss": 0.7066, "step": 33909 }, { "epoch": 3.608214513726325, "grad_norm": 19.09375, "learning_rate": 8.812890847072201e-06, "loss": 0.7794, "step": 33910 }, { "epoch": 3.6083209193445414, "grad_norm": 14.359375, "learning_rate": 8.812855315520183e-06, "loss": 0.7095, "step": 33911 }, { "epoch": 3.608427324962758, "grad_norm": 19.953125, "learning_rate": 8.812819783968165e-06, "loss": 0.8025, "step": 33912 }, { "epoch": 3.608533730580975, "grad_norm": 15.03125, "learning_rate": 8.812784252416146e-06, "loss": 0.8424, "step": 33913 }, { "epoch": 3.6086401361991913, "grad_norm": 12.75, "learning_rate": 8.812748720864128e-06, "loss": 0.725, "step": 33914 }, { "epoch": 3.608746541817408, "grad_norm": 11.0390625, "learning_rate": 8.81271318931211e-06, "loss": 0.6458, "step": 33915 }, { "epoch": 3.608852947435625, "grad_norm": 12.0625, "learning_rate": 8.812677657760091e-06, "loss": 0.6893, "step": 33916 }, { "epoch": 3.6089593530538413, "grad_norm": 10.328125, "learning_rate": 8.812642126208073e-06, "loss": 0.5401, "step": 33917 }, { "epoch": 3.609065758672058, "grad_norm": 11.765625, "learning_rate": 8.812606594656056e-06, "loss": 0.5819, "step": 33918 }, { "epoch": 3.6091721642902748, "grad_norm": 11.4765625, "learning_rate": 8.812571063104038e-06, "loss": 0.3375, "step": 33919 }, { "epoch": 3.6092785699084913, "grad_norm": 16.234375, "learning_rate": 8.812535531552019e-06, "loss": 0.8068, "step": 33920 }, { "epoch": 3.609384975526708, "grad_norm": 10.0859375, "learning_rate": 8.8125e-06, "loss": 0.6916, "step": 33921 }, { "epoch": 3.6094913811449243, "grad_norm": 11.3046875, "learning_rate": 8.812464468447982e-06, "loss": 0.6143, "step": 33922 }, { "epoch": 3.6095977867631412, "grad_norm": 13.109375, "learning_rate": 8.812428936895964e-06, "loss": 0.5932, "step": 33923 }, { "epoch": 3.6097041923813578, "grad_norm": 11.8828125, "learning_rate": 8.812393405343946e-06, "loss": 0.6555, "step": 33924 }, { "epoch": 3.6098105979995743, "grad_norm": 14.5, "learning_rate": 8.812357873791927e-06, "loss": 0.6355, "step": 33925 }, { "epoch": 3.6099170036177908, "grad_norm": 15.7421875, "learning_rate": 8.812322342239909e-06, "loss": 0.8172, "step": 33926 }, { "epoch": 3.6100234092360077, "grad_norm": 14.6875, "learning_rate": 8.81228681068789e-06, "loss": 0.7345, "step": 33927 }, { "epoch": 3.6101298148542242, "grad_norm": 12.453125, "learning_rate": 8.812251279135874e-06, "loss": 0.5851, "step": 33928 }, { "epoch": 3.6102362204724407, "grad_norm": 11.9375, "learning_rate": 8.812215747583855e-06, "loss": 0.6087, "step": 33929 }, { "epoch": 3.6103426260906577, "grad_norm": 17.09375, "learning_rate": 8.812180216031837e-06, "loss": 0.6827, "step": 33930 }, { "epoch": 3.610449031708874, "grad_norm": 15.734375, "learning_rate": 8.812144684479819e-06, "loss": 0.7237, "step": 33931 }, { "epoch": 3.6105554373270907, "grad_norm": 15.03125, "learning_rate": 8.8121091529278e-06, "loss": 0.6598, "step": 33932 }, { "epoch": 3.6106618429453077, "grad_norm": 13.640625, "learning_rate": 8.812073621375783e-06, "loss": 0.6979, "step": 33933 }, { "epoch": 3.610768248563524, "grad_norm": 14.0546875, "learning_rate": 8.812038089823765e-06, "loss": 0.5992, "step": 33934 }, { "epoch": 3.6108746541817407, "grad_norm": 13.03125, "learning_rate": 8.812002558271747e-06, "loss": 0.7209, "step": 33935 }, { "epoch": 3.6109810597999576, "grad_norm": 16.296875, "learning_rate": 8.811967026719728e-06, "loss": 0.7016, "step": 33936 }, { "epoch": 3.611087465418174, "grad_norm": 15.53125, "learning_rate": 8.81193149516771e-06, "loss": 0.8016, "step": 33937 }, { "epoch": 3.6111938710363907, "grad_norm": 10.1796875, "learning_rate": 8.811895963615691e-06, "loss": 0.5534, "step": 33938 }, { "epoch": 3.6113002766546076, "grad_norm": 11.4375, "learning_rate": 8.811860432063673e-06, "loss": 0.608, "step": 33939 }, { "epoch": 3.611406682272824, "grad_norm": 12.8828125, "learning_rate": 8.811824900511655e-06, "loss": 0.6793, "step": 33940 }, { "epoch": 3.6115130878910406, "grad_norm": 11.2421875, "learning_rate": 8.811789368959636e-06, "loss": 0.671, "step": 33941 }, { "epoch": 3.611619493509257, "grad_norm": 13.5546875, "learning_rate": 8.811753837407618e-06, "loss": 0.6635, "step": 33942 }, { "epoch": 3.6117258991274737, "grad_norm": 12.46875, "learning_rate": 8.8117183058556e-06, "loss": 0.575, "step": 33943 }, { "epoch": 3.6118323047456906, "grad_norm": 11.5859375, "learning_rate": 8.811682774303583e-06, "loss": 0.4705, "step": 33944 }, { "epoch": 3.611938710363907, "grad_norm": 12.09375, "learning_rate": 8.811647242751564e-06, "loss": 0.5594, "step": 33945 }, { "epoch": 3.6120451159821236, "grad_norm": 12.90625, "learning_rate": 8.811611711199546e-06, "loss": 0.5684, "step": 33946 }, { "epoch": 3.6121515216003406, "grad_norm": 11.796875, "learning_rate": 8.811576179647528e-06, "loss": 0.5536, "step": 33947 }, { "epoch": 3.612257927218557, "grad_norm": 12.0546875, "learning_rate": 8.81154064809551e-06, "loss": 0.7523, "step": 33948 }, { "epoch": 3.6123643328367736, "grad_norm": 19.421875, "learning_rate": 8.811505116543493e-06, "loss": 0.78, "step": 33949 }, { "epoch": 3.6124707384549906, "grad_norm": 12.109375, "learning_rate": 8.811469584991474e-06, "loss": 0.6548, "step": 33950 }, { "epoch": 3.612577144073207, "grad_norm": 17.515625, "learning_rate": 8.811434053439456e-06, "loss": 0.7998, "step": 33951 }, { "epoch": 3.6126835496914236, "grad_norm": 12.84375, "learning_rate": 8.811398521887437e-06, "loss": 0.5823, "step": 33952 }, { "epoch": 3.6127899553096405, "grad_norm": 15.671875, "learning_rate": 8.811362990335417e-06, "loss": 0.6667, "step": 33953 }, { "epoch": 3.612896360927857, "grad_norm": 15.6875, "learning_rate": 8.8113274587834e-06, "loss": 0.7046, "step": 33954 }, { "epoch": 3.6130027665460736, "grad_norm": 12.84375, "learning_rate": 8.811291927231382e-06, "loss": 0.6065, "step": 33955 }, { "epoch": 3.6131091721642905, "grad_norm": 14.671875, "learning_rate": 8.811256395679364e-06, "loss": 0.6129, "step": 33956 }, { "epoch": 3.613215577782507, "grad_norm": 15.2578125, "learning_rate": 8.811220864127345e-06, "loss": 0.563, "step": 33957 }, { "epoch": 3.6133219834007235, "grad_norm": 11.4296875, "learning_rate": 8.811185332575327e-06, "loss": 0.5482, "step": 33958 }, { "epoch": 3.61342838901894, "grad_norm": 12.2734375, "learning_rate": 8.81114980102331e-06, "loss": 0.5933, "step": 33959 }, { "epoch": 3.613534794637157, "grad_norm": 15.625, "learning_rate": 8.811114269471292e-06, "loss": 0.7157, "step": 33960 }, { "epoch": 3.6136412002553735, "grad_norm": 17.0625, "learning_rate": 8.811078737919273e-06, "loss": 0.812, "step": 33961 }, { "epoch": 3.61374760587359, "grad_norm": 11.6171875, "learning_rate": 8.811043206367255e-06, "loss": 0.5869, "step": 33962 }, { "epoch": 3.6138540114918065, "grad_norm": 15.3125, "learning_rate": 8.811007674815237e-06, "loss": 0.7212, "step": 33963 }, { "epoch": 3.6139604171100235, "grad_norm": 10.90625, "learning_rate": 8.810972143263218e-06, "loss": 0.4872, "step": 33964 }, { "epoch": 3.61406682272824, "grad_norm": 12.484375, "learning_rate": 8.8109366117112e-06, "loss": 0.6348, "step": 33965 }, { "epoch": 3.6141732283464565, "grad_norm": 22.25, "learning_rate": 8.810901080159182e-06, "loss": 0.9232, "step": 33966 }, { "epoch": 3.6142796339646734, "grad_norm": 13.8203125, "learning_rate": 8.810865548607163e-06, "loss": 0.5996, "step": 33967 }, { "epoch": 3.61438603958289, "grad_norm": 14.3203125, "learning_rate": 8.810830017055145e-06, "loss": 0.6786, "step": 33968 }, { "epoch": 3.6144924452011065, "grad_norm": 15.75, "learning_rate": 8.810794485503128e-06, "loss": 0.7143, "step": 33969 }, { "epoch": 3.6145988508193234, "grad_norm": 12.2265625, "learning_rate": 8.81075895395111e-06, "loss": 0.6133, "step": 33970 }, { "epoch": 3.61470525643754, "grad_norm": 12.4921875, "learning_rate": 8.810723422399091e-06, "loss": 0.6424, "step": 33971 }, { "epoch": 3.6148116620557564, "grad_norm": 16.125, "learning_rate": 8.810687890847073e-06, "loss": 0.7613, "step": 33972 }, { "epoch": 3.6149180676739734, "grad_norm": 11.4296875, "learning_rate": 8.810652359295054e-06, "loss": 0.4851, "step": 33973 }, { "epoch": 3.61502447329219, "grad_norm": 15.234375, "learning_rate": 8.810616827743038e-06, "loss": 0.71, "step": 33974 }, { "epoch": 3.6151308789104064, "grad_norm": 16.796875, "learning_rate": 8.81058129619102e-06, "loss": 0.7213, "step": 33975 }, { "epoch": 3.6152372845286234, "grad_norm": 15.6640625, "learning_rate": 8.810545764639001e-06, "loss": 0.6991, "step": 33976 }, { "epoch": 3.61534369014684, "grad_norm": 11.2890625, "learning_rate": 8.810510233086983e-06, "loss": 0.6843, "step": 33977 }, { "epoch": 3.6154500957650564, "grad_norm": 15.0859375, "learning_rate": 8.810474701534964e-06, "loss": 0.7033, "step": 33978 }, { "epoch": 3.615556501383273, "grad_norm": 15.3671875, "learning_rate": 8.810439169982946e-06, "loss": 0.6449, "step": 33979 }, { "epoch": 3.6156629070014894, "grad_norm": 18.546875, "learning_rate": 8.810403638430927e-06, "loss": 0.7766, "step": 33980 }, { "epoch": 3.6157693126197064, "grad_norm": 12.59375, "learning_rate": 8.810368106878909e-06, "loss": 0.6397, "step": 33981 }, { "epoch": 3.615875718237923, "grad_norm": 19.078125, "learning_rate": 8.81033257532689e-06, "loss": 0.6488, "step": 33982 }, { "epoch": 3.6159821238561394, "grad_norm": 17.078125, "learning_rate": 8.810297043774872e-06, "loss": 0.8126, "step": 33983 }, { "epoch": 3.6160885294743563, "grad_norm": 11.390625, "learning_rate": 8.810261512222854e-06, "loss": 0.5103, "step": 33984 }, { "epoch": 3.616194935092573, "grad_norm": 12.2890625, "learning_rate": 8.810225980670837e-06, "loss": 0.6827, "step": 33985 }, { "epoch": 3.6163013407107893, "grad_norm": 10.75, "learning_rate": 8.810190449118819e-06, "loss": 0.5879, "step": 33986 }, { "epoch": 3.6164077463290063, "grad_norm": 14.1171875, "learning_rate": 8.8101549175668e-06, "loss": 0.791, "step": 33987 }, { "epoch": 3.616514151947223, "grad_norm": 19.46875, "learning_rate": 8.810119386014782e-06, "loss": 0.7878, "step": 33988 }, { "epoch": 3.6166205575654393, "grad_norm": 15.390625, "learning_rate": 8.810083854462764e-06, "loss": 0.7858, "step": 33989 }, { "epoch": 3.6167269631836563, "grad_norm": 12.7109375, "learning_rate": 8.810048322910745e-06, "loss": 0.6487, "step": 33990 }, { "epoch": 3.616833368801873, "grad_norm": 14.203125, "learning_rate": 8.810012791358727e-06, "loss": 0.7161, "step": 33991 }, { "epoch": 3.6169397744200893, "grad_norm": 11.296875, "learning_rate": 8.809977259806708e-06, "loss": 0.5929, "step": 33992 }, { "epoch": 3.6170461800383062, "grad_norm": 12.6484375, "learning_rate": 8.80994172825469e-06, "loss": 0.5858, "step": 33993 }, { "epoch": 3.6171525856565228, "grad_norm": 13.2421875, "learning_rate": 8.809906196702672e-06, "loss": 0.696, "step": 33994 }, { "epoch": 3.6172589912747393, "grad_norm": 19.15625, "learning_rate": 8.809870665150655e-06, "loss": 0.8644, "step": 33995 }, { "epoch": 3.617365396892956, "grad_norm": 11.4453125, "learning_rate": 8.809835133598636e-06, "loss": 0.6727, "step": 33996 }, { "epoch": 3.6174718025111727, "grad_norm": 12.03125, "learning_rate": 8.809799602046618e-06, "loss": 0.5457, "step": 33997 }, { "epoch": 3.6175782081293892, "grad_norm": 12.90625, "learning_rate": 8.8097640704946e-06, "loss": 0.5875, "step": 33998 }, { "epoch": 3.6176846137476057, "grad_norm": 14.171875, "learning_rate": 8.809728538942581e-06, "loss": 0.6857, "step": 33999 }, { "epoch": 3.6177910193658223, "grad_norm": 15.890625, "learning_rate": 8.809693007390565e-06, "loss": 0.7316, "step": 34000 }, { "epoch": 3.617897424984039, "grad_norm": 11.3203125, "learning_rate": 8.809657475838546e-06, "loss": 0.5578, "step": 34001 }, { "epoch": 3.6180038306022557, "grad_norm": 11.21875, "learning_rate": 8.809621944286528e-06, "loss": 0.7126, "step": 34002 }, { "epoch": 3.6181102362204722, "grad_norm": 13.15625, "learning_rate": 8.80958641273451e-06, "loss": 0.6138, "step": 34003 }, { "epoch": 3.618216641838689, "grad_norm": 12.015625, "learning_rate": 8.809550881182491e-06, "loss": 0.6147, "step": 34004 }, { "epoch": 3.6183230474569057, "grad_norm": 16.25, "learning_rate": 8.809515349630473e-06, "loss": 0.7144, "step": 34005 }, { "epoch": 3.618429453075122, "grad_norm": 13.8671875, "learning_rate": 8.809479818078454e-06, "loss": 0.7555, "step": 34006 }, { "epoch": 3.618535858693339, "grad_norm": 13.3046875, "learning_rate": 8.809444286526436e-06, "loss": 0.6044, "step": 34007 }, { "epoch": 3.6186422643115557, "grad_norm": 17.8125, "learning_rate": 8.809408754974417e-06, "loss": 0.7922, "step": 34008 }, { "epoch": 3.618748669929772, "grad_norm": 14.7421875, "learning_rate": 8.809373223422399e-06, "loss": 0.6622, "step": 34009 }, { "epoch": 3.618855075547989, "grad_norm": 11.796875, "learning_rate": 8.809337691870382e-06, "loss": 0.564, "step": 34010 }, { "epoch": 3.6189614811662056, "grad_norm": 12.3984375, "learning_rate": 8.809302160318364e-06, "loss": 0.7222, "step": 34011 }, { "epoch": 3.619067886784422, "grad_norm": 12.796875, "learning_rate": 8.809266628766346e-06, "loss": 0.57, "step": 34012 }, { "epoch": 3.619174292402639, "grad_norm": 13.6875, "learning_rate": 8.809231097214327e-06, "loss": 0.645, "step": 34013 }, { "epoch": 3.6192806980208556, "grad_norm": 14.5625, "learning_rate": 8.809195565662309e-06, "loss": 0.6243, "step": 34014 }, { "epoch": 3.619387103639072, "grad_norm": 12.40625, "learning_rate": 8.80916003411029e-06, "loss": 0.5763, "step": 34015 }, { "epoch": 3.6194935092572886, "grad_norm": 10.484375, "learning_rate": 8.809124502558272e-06, "loss": 0.625, "step": 34016 }, { "epoch": 3.6195999148755056, "grad_norm": 13.4921875, "learning_rate": 8.809088971006254e-06, "loss": 0.7115, "step": 34017 }, { "epoch": 3.619706320493722, "grad_norm": 11.234375, "learning_rate": 8.809053439454235e-06, "loss": 0.5845, "step": 34018 }, { "epoch": 3.6198127261119386, "grad_norm": 10.796875, "learning_rate": 8.809017907902217e-06, "loss": 0.5592, "step": 34019 }, { "epoch": 3.619919131730155, "grad_norm": 13.4921875, "learning_rate": 8.808982376350198e-06, "loss": 0.6578, "step": 34020 }, { "epoch": 3.620025537348372, "grad_norm": 16.046875, "learning_rate": 8.808946844798182e-06, "loss": 0.6856, "step": 34021 }, { "epoch": 3.6201319429665886, "grad_norm": 13.2734375, "learning_rate": 8.808911313246163e-06, "loss": 0.4838, "step": 34022 }, { "epoch": 3.620238348584805, "grad_norm": 11.9921875, "learning_rate": 8.808875781694145e-06, "loss": 0.5255, "step": 34023 }, { "epoch": 3.620344754203022, "grad_norm": 11.859375, "learning_rate": 8.808840250142127e-06, "loss": 0.5225, "step": 34024 }, { "epoch": 3.6204511598212386, "grad_norm": 13.5859375, "learning_rate": 8.808804718590108e-06, "loss": 0.6895, "step": 34025 }, { "epoch": 3.620557565439455, "grad_norm": 18.15625, "learning_rate": 8.808769187038091e-06, "loss": 0.7707, "step": 34026 }, { "epoch": 3.620663971057672, "grad_norm": 13.484375, "learning_rate": 8.808733655486073e-06, "loss": 0.6452, "step": 34027 }, { "epoch": 3.6207703766758885, "grad_norm": 14.6328125, "learning_rate": 8.808698123934055e-06, "loss": 0.6841, "step": 34028 }, { "epoch": 3.620876782294105, "grad_norm": 12.5078125, "learning_rate": 8.808662592382036e-06, "loss": 0.6226, "step": 34029 }, { "epoch": 3.620983187912322, "grad_norm": 12.09375, "learning_rate": 8.808627060830018e-06, "loss": 0.5868, "step": 34030 }, { "epoch": 3.6210895935305385, "grad_norm": 12.1484375, "learning_rate": 8.808591529278e-06, "loss": 0.6268, "step": 34031 }, { "epoch": 3.621195999148755, "grad_norm": 10.546875, "learning_rate": 8.808555997725981e-06, "loss": 0.54, "step": 34032 }, { "epoch": 3.621302404766972, "grad_norm": 18.109375, "learning_rate": 8.808520466173963e-06, "loss": 0.8692, "step": 34033 }, { "epoch": 3.6214088103851885, "grad_norm": 14.40625, "learning_rate": 8.808484934621944e-06, "loss": 0.6791, "step": 34034 }, { "epoch": 3.621515216003405, "grad_norm": 14.0078125, "learning_rate": 8.808449403069926e-06, "loss": 0.6789, "step": 34035 }, { "epoch": 3.6216216216216215, "grad_norm": 12.6171875, "learning_rate": 8.80841387151791e-06, "loss": 0.562, "step": 34036 }, { "epoch": 3.621728027239838, "grad_norm": 11.671875, "learning_rate": 8.80837833996589e-06, "loss": 0.5143, "step": 34037 }, { "epoch": 3.621834432858055, "grad_norm": 12.640625, "learning_rate": 8.808342808413872e-06, "loss": 0.6266, "step": 34038 }, { "epoch": 3.6219408384762715, "grad_norm": 12.5234375, "learning_rate": 8.808307276861854e-06, "loss": 0.6515, "step": 34039 }, { "epoch": 3.622047244094488, "grad_norm": 10.703125, "learning_rate": 8.808271745309836e-06, "loss": 0.632, "step": 34040 }, { "epoch": 3.622153649712705, "grad_norm": 15.84375, "learning_rate": 8.808236213757819e-06, "loss": 0.7149, "step": 34041 }, { "epoch": 3.6222600553309214, "grad_norm": 14.5078125, "learning_rate": 8.808200682205799e-06, "loss": 0.5858, "step": 34042 }, { "epoch": 3.622366460949138, "grad_norm": 13.2890625, "learning_rate": 8.80816515065378e-06, "loss": 0.5911, "step": 34043 }, { "epoch": 3.622472866567355, "grad_norm": 15.2578125, "learning_rate": 8.808129619101762e-06, "loss": 0.69, "step": 34044 }, { "epoch": 3.6225792721855714, "grad_norm": 12.59375, "learning_rate": 8.808094087549744e-06, "loss": 0.6331, "step": 34045 }, { "epoch": 3.622685677803788, "grad_norm": 14.1484375, "learning_rate": 8.808058555997727e-06, "loss": 0.5988, "step": 34046 }, { "epoch": 3.622792083422005, "grad_norm": 18.078125, "learning_rate": 8.808023024445709e-06, "loss": 0.6739, "step": 34047 }, { "epoch": 3.6228984890402214, "grad_norm": 12.75, "learning_rate": 8.80798749289369e-06, "loss": 0.5582, "step": 34048 }, { "epoch": 3.623004894658438, "grad_norm": 17.046875, "learning_rate": 8.807951961341672e-06, "loss": 0.7228, "step": 34049 }, { "epoch": 3.623111300276655, "grad_norm": 14.3125, "learning_rate": 8.807916429789653e-06, "loss": 0.7331, "step": 34050 }, { "epoch": 3.6232177058948714, "grad_norm": 13.21875, "learning_rate": 8.807880898237637e-06, "loss": 0.7099, "step": 34051 }, { "epoch": 3.623324111513088, "grad_norm": 19.4375, "learning_rate": 8.807845366685618e-06, "loss": 0.8147, "step": 34052 }, { "epoch": 3.6234305171313044, "grad_norm": 11.03125, "learning_rate": 8.8078098351336e-06, "loss": 0.5605, "step": 34053 }, { "epoch": 3.6235369227495213, "grad_norm": 11.0625, "learning_rate": 8.807774303581582e-06, "loss": 0.5398, "step": 34054 }, { "epoch": 3.623643328367738, "grad_norm": 12.8671875, "learning_rate": 8.807738772029563e-06, "loss": 0.6622, "step": 34055 }, { "epoch": 3.6237497339859543, "grad_norm": 13.1640625, "learning_rate": 8.807703240477545e-06, "loss": 0.6406, "step": 34056 }, { "epoch": 3.623856139604171, "grad_norm": 13.2734375, "learning_rate": 8.807667708925526e-06, "loss": 0.6983, "step": 34057 }, { "epoch": 3.623962545222388, "grad_norm": 11.203125, "learning_rate": 8.807632177373508e-06, "loss": 0.6266, "step": 34058 }, { "epoch": 3.6240689508406043, "grad_norm": 18.953125, "learning_rate": 8.80759664582149e-06, "loss": 0.7482, "step": 34059 }, { "epoch": 3.624175356458821, "grad_norm": 18.140625, "learning_rate": 8.807561114269471e-06, "loss": 0.7792, "step": 34060 }, { "epoch": 3.624281762077038, "grad_norm": 14.8203125, "learning_rate": 8.807525582717453e-06, "loss": 0.7638, "step": 34061 }, { "epoch": 3.6243881676952543, "grad_norm": 11.8046875, "learning_rate": 8.807490051165436e-06, "loss": 0.5436, "step": 34062 }, { "epoch": 3.624494573313471, "grad_norm": 19.1875, "learning_rate": 8.807454519613418e-06, "loss": 0.6277, "step": 34063 }, { "epoch": 3.6246009789316878, "grad_norm": 15.7890625, "learning_rate": 8.8074189880614e-06, "loss": 0.7687, "step": 34064 }, { "epoch": 3.6247073845499043, "grad_norm": 12.3125, "learning_rate": 8.807383456509381e-06, "loss": 0.6908, "step": 34065 }, { "epoch": 3.6248137901681208, "grad_norm": 13.3515625, "learning_rate": 8.807347924957363e-06, "loss": 0.6388, "step": 34066 }, { "epoch": 3.6249201957863377, "grad_norm": 13.4140625, "learning_rate": 8.807312393405346e-06, "loss": 0.7272, "step": 34067 }, { "epoch": 3.6250266014045542, "grad_norm": 10.6171875, "learning_rate": 8.807276861853327e-06, "loss": 0.6284, "step": 34068 }, { "epoch": 3.6251330070227707, "grad_norm": 11.3671875, "learning_rate": 8.807241330301309e-06, "loss": 0.6686, "step": 34069 }, { "epoch": 3.6252394126409877, "grad_norm": 15.59375, "learning_rate": 8.80720579874929e-06, "loss": 0.7148, "step": 34070 }, { "epoch": 3.625345818259204, "grad_norm": 19.484375, "learning_rate": 8.807170267197272e-06, "loss": 0.7086, "step": 34071 }, { "epoch": 3.6254522238774207, "grad_norm": 14.4921875, "learning_rate": 8.807134735645254e-06, "loss": 0.6224, "step": 34072 }, { "epoch": 3.6255586294956372, "grad_norm": 12.9296875, "learning_rate": 8.807099204093235e-06, "loss": 0.6319, "step": 34073 }, { "epoch": 3.6256650351138537, "grad_norm": 11.4453125, "learning_rate": 8.807063672541217e-06, "loss": 0.4836, "step": 34074 }, { "epoch": 3.6257714407320707, "grad_norm": 11.7265625, "learning_rate": 8.807028140989199e-06, "loss": 0.6114, "step": 34075 }, { "epoch": 3.625877846350287, "grad_norm": 10.6171875, "learning_rate": 8.80699260943718e-06, "loss": 0.5435, "step": 34076 }, { "epoch": 3.6259842519685037, "grad_norm": 12.640625, "learning_rate": 8.806957077885164e-06, "loss": 0.6349, "step": 34077 }, { "epoch": 3.6260906575867207, "grad_norm": 13.4921875, "learning_rate": 8.806921546333145e-06, "loss": 0.589, "step": 34078 }, { "epoch": 3.626197063204937, "grad_norm": 15.125, "learning_rate": 8.806886014781127e-06, "loss": 0.704, "step": 34079 }, { "epoch": 3.6263034688231537, "grad_norm": 12.5703125, "learning_rate": 8.806850483229108e-06, "loss": 0.5576, "step": 34080 }, { "epoch": 3.6264098744413706, "grad_norm": 12.703125, "learning_rate": 8.80681495167709e-06, "loss": 0.5593, "step": 34081 }, { "epoch": 3.626516280059587, "grad_norm": 19.1875, "learning_rate": 8.806779420125072e-06, "loss": 0.7673, "step": 34082 }, { "epoch": 3.6266226856778037, "grad_norm": 18.8125, "learning_rate": 8.806743888573053e-06, "loss": 0.7393, "step": 34083 }, { "epoch": 3.6267290912960206, "grad_norm": 12.6640625, "learning_rate": 8.806708357021035e-06, "loss": 0.5921, "step": 34084 }, { "epoch": 3.626835496914237, "grad_norm": 12.5234375, "learning_rate": 8.806672825469016e-06, "loss": 0.7572, "step": 34085 }, { "epoch": 3.6269419025324536, "grad_norm": 12.6015625, "learning_rate": 8.806637293916998e-06, "loss": 0.5855, "step": 34086 }, { "epoch": 3.6270483081506706, "grad_norm": 12.5234375, "learning_rate": 8.806601762364981e-06, "loss": 0.6457, "step": 34087 }, { "epoch": 3.627154713768887, "grad_norm": 9.4921875, "learning_rate": 8.806566230812963e-06, "loss": 0.5506, "step": 34088 }, { "epoch": 3.6272611193871036, "grad_norm": 12.765625, "learning_rate": 8.806530699260945e-06, "loss": 0.7617, "step": 34089 }, { "epoch": 3.62736752500532, "grad_norm": 14.375, "learning_rate": 8.806495167708926e-06, "loss": 0.683, "step": 34090 }, { "epoch": 3.627473930623537, "grad_norm": 13.5546875, "learning_rate": 8.806459636156908e-06, "loss": 0.653, "step": 34091 }, { "epoch": 3.6275803362417536, "grad_norm": 10.5, "learning_rate": 8.806424104604891e-06, "loss": 0.558, "step": 34092 }, { "epoch": 3.62768674185997, "grad_norm": 11.546875, "learning_rate": 8.806388573052873e-06, "loss": 0.5213, "step": 34093 }, { "epoch": 3.6277931474781866, "grad_norm": 14.9296875, "learning_rate": 8.806353041500854e-06, "loss": 0.6526, "step": 34094 }, { "epoch": 3.6278995530964036, "grad_norm": 20.96875, "learning_rate": 8.806317509948836e-06, "loss": 0.663, "step": 34095 }, { "epoch": 3.62800595871462, "grad_norm": 11.5078125, "learning_rate": 8.806281978396817e-06, "loss": 0.5595, "step": 34096 }, { "epoch": 3.6281123643328366, "grad_norm": 15.71875, "learning_rate": 8.806246446844799e-06, "loss": 0.7996, "step": 34097 }, { "epoch": 3.6282187699510535, "grad_norm": 12.5234375, "learning_rate": 8.80621091529278e-06, "loss": 0.524, "step": 34098 }, { "epoch": 3.62832517556927, "grad_norm": 12.5625, "learning_rate": 8.806175383740762e-06, "loss": 0.6643, "step": 34099 }, { "epoch": 3.6284315811874865, "grad_norm": 15.5390625, "learning_rate": 8.806139852188744e-06, "loss": 0.8791, "step": 34100 }, { "epoch": 3.6285379868057035, "grad_norm": 15.4921875, "learning_rate": 8.806104320636726e-06, "loss": 0.6641, "step": 34101 }, { "epoch": 3.62864439242392, "grad_norm": 14.5546875, "learning_rate": 8.806068789084707e-06, "loss": 0.7642, "step": 34102 }, { "epoch": 3.6287507980421365, "grad_norm": 12.8828125, "learning_rate": 8.80603325753269e-06, "loss": 0.7436, "step": 34103 }, { "epoch": 3.6288572036603535, "grad_norm": 15.8984375, "learning_rate": 8.805997725980672e-06, "loss": 0.5667, "step": 34104 }, { "epoch": 3.62896360927857, "grad_norm": 12.625, "learning_rate": 8.805962194428654e-06, "loss": 0.5674, "step": 34105 }, { "epoch": 3.6290700148967865, "grad_norm": 14.0703125, "learning_rate": 8.805926662876635e-06, "loss": 0.7529, "step": 34106 }, { "epoch": 3.6291764205150034, "grad_norm": 18.359375, "learning_rate": 8.805891131324617e-06, "loss": 0.7714, "step": 34107 }, { "epoch": 3.62928282613322, "grad_norm": 14.0625, "learning_rate": 8.805855599772598e-06, "loss": 0.801, "step": 34108 }, { "epoch": 3.6293892317514365, "grad_norm": 11.6171875, "learning_rate": 8.80582006822058e-06, "loss": 0.5397, "step": 34109 }, { "epoch": 3.629495637369653, "grad_norm": 9.4609375, "learning_rate": 8.805784536668562e-06, "loss": 0.6582, "step": 34110 }, { "epoch": 3.62960204298787, "grad_norm": 13.0390625, "learning_rate": 8.805749005116543e-06, "loss": 0.7087, "step": 34111 }, { "epoch": 3.6297084486060864, "grad_norm": 15.703125, "learning_rate": 8.805713473564525e-06, "loss": 0.7761, "step": 34112 }, { "epoch": 3.629814854224303, "grad_norm": 12.328125, "learning_rate": 8.805677942012508e-06, "loss": 0.6851, "step": 34113 }, { "epoch": 3.6299212598425195, "grad_norm": 11.3203125, "learning_rate": 8.80564241046049e-06, "loss": 0.6475, "step": 34114 }, { "epoch": 3.6300276654607364, "grad_norm": 13.3828125, "learning_rate": 8.805606878908471e-06, "loss": 0.6595, "step": 34115 }, { "epoch": 3.630134071078953, "grad_norm": 13.3203125, "learning_rate": 8.805571347356453e-06, "loss": 0.6521, "step": 34116 }, { "epoch": 3.6302404766971694, "grad_norm": 19.75, "learning_rate": 8.805535815804435e-06, "loss": 0.7724, "step": 34117 }, { "epoch": 3.6303468823153864, "grad_norm": 11.5390625, "learning_rate": 8.805500284252418e-06, "loss": 0.5898, "step": 34118 }, { "epoch": 3.630453287933603, "grad_norm": 14.46875, "learning_rate": 8.8054647527004e-06, "loss": 0.6839, "step": 34119 }, { "epoch": 3.6305596935518194, "grad_norm": 9.859375, "learning_rate": 8.805429221148381e-06, "loss": 0.5061, "step": 34120 }, { "epoch": 3.6306660991700364, "grad_norm": 17.5, "learning_rate": 8.805393689596363e-06, "loss": 0.7973, "step": 34121 }, { "epoch": 3.630772504788253, "grad_norm": 12.1015625, "learning_rate": 8.805358158044344e-06, "loss": 0.5681, "step": 34122 }, { "epoch": 3.6308789104064694, "grad_norm": 13.1953125, "learning_rate": 8.805322626492326e-06, "loss": 0.776, "step": 34123 }, { "epoch": 3.6309853160246863, "grad_norm": 13.6328125, "learning_rate": 8.805287094940308e-06, "loss": 0.6377, "step": 34124 }, { "epoch": 3.631091721642903, "grad_norm": 15.6015625, "learning_rate": 8.80525156338829e-06, "loss": 0.6845, "step": 34125 }, { "epoch": 3.6311981272611193, "grad_norm": 12.46875, "learning_rate": 8.80521603183627e-06, "loss": 0.6773, "step": 34126 }, { "epoch": 3.6313045328793363, "grad_norm": 11.0703125, "learning_rate": 8.805180500284252e-06, "loss": 0.5236, "step": 34127 }, { "epoch": 3.631410938497553, "grad_norm": 20.671875, "learning_rate": 8.805144968732236e-06, "loss": 0.8222, "step": 34128 }, { "epoch": 3.6315173441157693, "grad_norm": 14.84375, "learning_rate": 8.805109437180217e-06, "loss": 0.6852, "step": 34129 }, { "epoch": 3.631623749733986, "grad_norm": 11.65625, "learning_rate": 8.805073905628199e-06, "loss": 0.5658, "step": 34130 }, { "epoch": 3.6317301553522023, "grad_norm": 12.078125, "learning_rate": 8.80503837407618e-06, "loss": 0.5823, "step": 34131 }, { "epoch": 3.6318365609704193, "grad_norm": 10.2421875, "learning_rate": 8.805002842524162e-06, "loss": 0.7072, "step": 34132 }, { "epoch": 3.631942966588636, "grad_norm": 12.921875, "learning_rate": 8.804967310972144e-06, "loss": 0.5549, "step": 34133 }, { "epoch": 3.6320493722068523, "grad_norm": 13.59375, "learning_rate": 8.804931779420125e-06, "loss": 0.6824, "step": 34134 }, { "epoch": 3.6321557778250693, "grad_norm": 16.984375, "learning_rate": 8.804896247868107e-06, "loss": 0.5971, "step": 34135 }, { "epoch": 3.6322621834432858, "grad_norm": 13.9453125, "learning_rate": 8.804860716316089e-06, "loss": 0.771, "step": 34136 }, { "epoch": 3.6323685890615023, "grad_norm": 15.828125, "learning_rate": 8.80482518476407e-06, "loss": 0.7152, "step": 34137 }, { "epoch": 3.6324749946797192, "grad_norm": 14.546875, "learning_rate": 8.804789653212052e-06, "loss": 0.5456, "step": 34138 }, { "epoch": 3.6325814002979357, "grad_norm": 11.0078125, "learning_rate": 8.804754121660035e-06, "loss": 0.7119, "step": 34139 }, { "epoch": 3.6326878059161523, "grad_norm": 13.7421875, "learning_rate": 8.804718590108017e-06, "loss": 0.646, "step": 34140 }, { "epoch": 3.632794211534369, "grad_norm": 11.734375, "learning_rate": 8.804683058555998e-06, "loss": 0.5567, "step": 34141 }, { "epoch": 3.6329006171525857, "grad_norm": 16.4375, "learning_rate": 8.80464752700398e-06, "loss": 0.8264, "step": 34142 }, { "epoch": 3.6330070227708022, "grad_norm": 17.421875, "learning_rate": 8.804611995451961e-06, "loss": 0.688, "step": 34143 }, { "epoch": 3.633113428389019, "grad_norm": 12.515625, "learning_rate": 8.804576463899945e-06, "loss": 0.5687, "step": 34144 }, { "epoch": 3.6332198340072357, "grad_norm": 13.3828125, "learning_rate": 8.804540932347926e-06, "loss": 0.6062, "step": 34145 }, { "epoch": 3.633326239625452, "grad_norm": 13.59375, "learning_rate": 8.804505400795908e-06, "loss": 0.6483, "step": 34146 }, { "epoch": 3.6334326452436687, "grad_norm": 15.984375, "learning_rate": 8.80446986924389e-06, "loss": 0.8176, "step": 34147 }, { "epoch": 3.6335390508618857, "grad_norm": 11.8671875, "learning_rate": 8.804434337691871e-06, "loss": 0.5025, "step": 34148 }, { "epoch": 3.633645456480102, "grad_norm": 11.1953125, "learning_rate": 8.804398806139853e-06, "loss": 0.5372, "step": 34149 }, { "epoch": 3.6337518620983187, "grad_norm": 15.734375, "learning_rate": 8.804363274587834e-06, "loss": 0.643, "step": 34150 }, { "epoch": 3.633858267716535, "grad_norm": 13.640625, "learning_rate": 8.804327743035816e-06, "loss": 0.6759, "step": 34151 }, { "epoch": 3.633964673334752, "grad_norm": 11.3515625, "learning_rate": 8.804292211483798e-06, "loss": 0.483, "step": 34152 }, { "epoch": 3.6340710789529687, "grad_norm": 13.2578125, "learning_rate": 8.80425667993178e-06, "loss": 0.6456, "step": 34153 }, { "epoch": 3.634177484571185, "grad_norm": 12.625, "learning_rate": 8.804221148379763e-06, "loss": 0.5786, "step": 34154 }, { "epoch": 3.634283890189402, "grad_norm": 12.1875, "learning_rate": 8.804185616827744e-06, "loss": 0.6036, "step": 34155 }, { "epoch": 3.6343902958076186, "grad_norm": 12.4375, "learning_rate": 8.804150085275726e-06, "loss": 0.5206, "step": 34156 }, { "epoch": 3.634496701425835, "grad_norm": 14.9296875, "learning_rate": 8.804114553723707e-06, "loss": 0.7635, "step": 34157 }, { "epoch": 3.634603107044052, "grad_norm": 12.1484375, "learning_rate": 8.804079022171689e-06, "loss": 0.5972, "step": 34158 }, { "epoch": 3.6347095126622686, "grad_norm": 17.28125, "learning_rate": 8.804043490619672e-06, "loss": 0.8583, "step": 34159 }, { "epoch": 3.634815918280485, "grad_norm": 12.015625, "learning_rate": 8.804007959067654e-06, "loss": 0.5617, "step": 34160 }, { "epoch": 3.634922323898702, "grad_norm": 14.8125, "learning_rate": 8.803972427515635e-06, "loss": 0.6408, "step": 34161 }, { "epoch": 3.6350287295169186, "grad_norm": 12.953125, "learning_rate": 8.803936895963615e-06, "loss": 0.6372, "step": 34162 }, { "epoch": 3.635135135135135, "grad_norm": 15.2890625, "learning_rate": 8.803901364411597e-06, "loss": 0.7521, "step": 34163 }, { "epoch": 3.635241540753352, "grad_norm": 15.109375, "learning_rate": 8.80386583285958e-06, "loss": 0.7618, "step": 34164 }, { "epoch": 3.6353479463715686, "grad_norm": 13.796875, "learning_rate": 8.803830301307562e-06, "loss": 0.6993, "step": 34165 }, { "epoch": 3.635454351989785, "grad_norm": 14.265625, "learning_rate": 8.803794769755544e-06, "loss": 0.639, "step": 34166 }, { "epoch": 3.6355607576080016, "grad_norm": 13.578125, "learning_rate": 8.803759238203525e-06, "loss": 0.6596, "step": 34167 }, { "epoch": 3.635667163226218, "grad_norm": 12.78125, "learning_rate": 8.803723706651507e-06, "loss": 0.5818, "step": 34168 }, { "epoch": 3.635773568844435, "grad_norm": 16.703125, "learning_rate": 8.80368817509949e-06, "loss": 0.7478, "step": 34169 }, { "epoch": 3.6358799744626515, "grad_norm": 15.5859375, "learning_rate": 8.803652643547472e-06, "loss": 0.7745, "step": 34170 }, { "epoch": 3.635986380080868, "grad_norm": 13.7421875, "learning_rate": 8.803617111995453e-06, "loss": 0.6856, "step": 34171 }, { "epoch": 3.636092785699085, "grad_norm": 13.9921875, "learning_rate": 8.803581580443435e-06, "loss": 0.7095, "step": 34172 }, { "epoch": 3.6361991913173015, "grad_norm": 15.9375, "learning_rate": 8.803546048891416e-06, "loss": 0.7809, "step": 34173 }, { "epoch": 3.636305596935518, "grad_norm": 14.6015625, "learning_rate": 8.803510517339398e-06, "loss": 0.7628, "step": 34174 }, { "epoch": 3.636412002553735, "grad_norm": 11.3046875, "learning_rate": 8.80347498578738e-06, "loss": 0.5672, "step": 34175 }, { "epoch": 3.6365184081719515, "grad_norm": 14.5703125, "learning_rate": 8.803439454235361e-06, "loss": 0.6747, "step": 34176 }, { "epoch": 3.636624813790168, "grad_norm": 14.8828125, "learning_rate": 8.803403922683343e-06, "loss": 0.8366, "step": 34177 }, { "epoch": 3.636731219408385, "grad_norm": 9.6640625, "learning_rate": 8.803368391131324e-06, "loss": 0.553, "step": 34178 }, { "epoch": 3.6368376250266015, "grad_norm": 13.0234375, "learning_rate": 8.803332859579306e-06, "loss": 0.6237, "step": 34179 }, { "epoch": 3.636944030644818, "grad_norm": 12.875, "learning_rate": 8.80329732802729e-06, "loss": 0.6735, "step": 34180 }, { "epoch": 3.637050436263035, "grad_norm": 10.9765625, "learning_rate": 8.803261796475271e-06, "loss": 0.5449, "step": 34181 }, { "epoch": 3.6371568418812514, "grad_norm": 14.7734375, "learning_rate": 8.803226264923253e-06, "loss": 0.6155, "step": 34182 }, { "epoch": 3.637263247499468, "grad_norm": 16.625, "learning_rate": 8.803190733371234e-06, "loss": 0.6675, "step": 34183 }, { "epoch": 3.6373696531176845, "grad_norm": 13.3046875, "learning_rate": 8.803155201819216e-06, "loss": 0.6235, "step": 34184 }, { "epoch": 3.6374760587359014, "grad_norm": 12.9609375, "learning_rate": 8.803119670267199e-06, "loss": 0.5463, "step": 34185 }, { "epoch": 3.637582464354118, "grad_norm": 12.4921875, "learning_rate": 8.80308413871518e-06, "loss": 0.626, "step": 34186 }, { "epoch": 3.6376888699723344, "grad_norm": 10.203125, "learning_rate": 8.803048607163162e-06, "loss": 0.5236, "step": 34187 }, { "epoch": 3.637795275590551, "grad_norm": 10.609375, "learning_rate": 8.803013075611144e-06, "loss": 0.5723, "step": 34188 }, { "epoch": 3.637901681208768, "grad_norm": 20.78125, "learning_rate": 8.802977544059126e-06, "loss": 0.9171, "step": 34189 }, { "epoch": 3.6380080868269844, "grad_norm": 11.5703125, "learning_rate": 8.802942012507107e-06, "loss": 0.5227, "step": 34190 }, { "epoch": 3.638114492445201, "grad_norm": 11.4921875, "learning_rate": 8.802906480955089e-06, "loss": 0.7121, "step": 34191 }, { "epoch": 3.638220898063418, "grad_norm": 12.0546875, "learning_rate": 8.80287094940307e-06, "loss": 0.5799, "step": 34192 }, { "epoch": 3.6383273036816344, "grad_norm": 12.6328125, "learning_rate": 8.802835417851052e-06, "loss": 0.6023, "step": 34193 }, { "epoch": 3.638433709299851, "grad_norm": 13.65625, "learning_rate": 8.802799886299034e-06, "loss": 0.7043, "step": 34194 }, { "epoch": 3.638540114918068, "grad_norm": 10.484375, "learning_rate": 8.802764354747017e-06, "loss": 0.6038, "step": 34195 }, { "epoch": 3.6386465205362843, "grad_norm": 11.3984375, "learning_rate": 8.802728823194998e-06, "loss": 0.7046, "step": 34196 }, { "epoch": 3.638752926154501, "grad_norm": 11.828125, "learning_rate": 8.80269329164298e-06, "loss": 0.5898, "step": 34197 }, { "epoch": 3.638859331772718, "grad_norm": 17.234375, "learning_rate": 8.802657760090962e-06, "loss": 0.8369, "step": 34198 }, { "epoch": 3.6389657373909343, "grad_norm": 13.0234375, "learning_rate": 8.802622228538943e-06, "loss": 0.6682, "step": 34199 }, { "epoch": 3.639072143009151, "grad_norm": 13.171875, "learning_rate": 8.802586696986925e-06, "loss": 0.7276, "step": 34200 }, { "epoch": 3.639178548627368, "grad_norm": 11.0, "learning_rate": 8.802551165434907e-06, "loss": 0.4866, "step": 34201 }, { "epoch": 3.6392849542455843, "grad_norm": 12.421875, "learning_rate": 8.802515633882888e-06, "loss": 0.6642, "step": 34202 }, { "epoch": 3.639391359863801, "grad_norm": 10.7109375, "learning_rate": 8.80248010233087e-06, "loss": 0.5752, "step": 34203 }, { "epoch": 3.6394977654820173, "grad_norm": 11.375, "learning_rate": 8.802444570778851e-06, "loss": 0.6893, "step": 34204 }, { "epoch": 3.639604171100234, "grad_norm": 15.1640625, "learning_rate": 8.802409039226835e-06, "loss": 0.7621, "step": 34205 }, { "epoch": 3.6397105767184508, "grad_norm": 14.171875, "learning_rate": 8.802373507674816e-06, "loss": 0.7293, "step": 34206 }, { "epoch": 3.6398169823366673, "grad_norm": 12.984375, "learning_rate": 8.802337976122798e-06, "loss": 0.7499, "step": 34207 }, { "epoch": 3.639923387954884, "grad_norm": 16.625, "learning_rate": 8.80230244457078e-06, "loss": 0.6417, "step": 34208 }, { "epoch": 3.6400297935731007, "grad_norm": 9.46875, "learning_rate": 8.802266913018761e-06, "loss": 0.5013, "step": 34209 }, { "epoch": 3.6401361991913173, "grad_norm": 11.0703125, "learning_rate": 8.802231381466744e-06, "loss": 0.7027, "step": 34210 }, { "epoch": 3.6402426048095338, "grad_norm": 15.4609375, "learning_rate": 8.802195849914726e-06, "loss": 0.7091, "step": 34211 }, { "epoch": 3.6403490104277507, "grad_norm": 14.8046875, "learning_rate": 8.802160318362708e-06, "loss": 0.7239, "step": 34212 }, { "epoch": 3.6404554160459672, "grad_norm": 17.046875, "learning_rate": 8.80212478681069e-06, "loss": 0.8108, "step": 34213 }, { "epoch": 3.6405618216641837, "grad_norm": 13.515625, "learning_rate": 8.80208925525867e-06, "loss": 0.6231, "step": 34214 }, { "epoch": 3.6406682272824007, "grad_norm": 12.2421875, "learning_rate": 8.802053723706652e-06, "loss": 0.699, "step": 34215 }, { "epoch": 3.640774632900617, "grad_norm": 11.890625, "learning_rate": 8.802018192154634e-06, "loss": 0.6329, "step": 34216 }, { "epoch": 3.6408810385188337, "grad_norm": 12.765625, "learning_rate": 8.801982660602616e-06, "loss": 0.6926, "step": 34217 }, { "epoch": 3.6409874441370507, "grad_norm": 20.75, "learning_rate": 8.801947129050597e-06, "loss": 0.8276, "step": 34218 }, { "epoch": 3.641093849755267, "grad_norm": 12.34375, "learning_rate": 8.801911597498579e-06, "loss": 0.682, "step": 34219 }, { "epoch": 3.6412002553734837, "grad_norm": 9.9296875, "learning_rate": 8.80187606594656e-06, "loss": 0.4916, "step": 34220 }, { "epoch": 3.6413066609917006, "grad_norm": 12.984375, "learning_rate": 8.801840534394544e-06, "loss": 0.6494, "step": 34221 }, { "epoch": 3.641413066609917, "grad_norm": 18.046875, "learning_rate": 8.801805002842525e-06, "loss": 0.7899, "step": 34222 }, { "epoch": 3.6415194722281337, "grad_norm": 13.4140625, "learning_rate": 8.801769471290507e-06, "loss": 0.6711, "step": 34223 }, { "epoch": 3.64162587784635, "grad_norm": 13.0, "learning_rate": 8.801733939738489e-06, "loss": 0.6288, "step": 34224 }, { "epoch": 3.6417322834645667, "grad_norm": 13.8671875, "learning_rate": 8.80169840818647e-06, "loss": 0.762, "step": 34225 }, { "epoch": 3.6418386890827836, "grad_norm": 12.0625, "learning_rate": 8.801662876634452e-06, "loss": 0.7115, "step": 34226 }, { "epoch": 3.641945094701, "grad_norm": 14.078125, "learning_rate": 8.801627345082433e-06, "loss": 0.7854, "step": 34227 }, { "epoch": 3.6420515003192167, "grad_norm": 13.171875, "learning_rate": 8.801591813530415e-06, "loss": 0.7696, "step": 34228 }, { "epoch": 3.6421579059374336, "grad_norm": 14.53125, "learning_rate": 8.801556281978397e-06, "loss": 0.6329, "step": 34229 }, { "epoch": 3.64226431155565, "grad_norm": 11.5859375, "learning_rate": 8.801520750426378e-06, "loss": 0.7358, "step": 34230 }, { "epoch": 3.6423707171738666, "grad_norm": 13.3046875, "learning_rate": 8.801485218874361e-06, "loss": 0.7102, "step": 34231 }, { "epoch": 3.6424771227920836, "grad_norm": 14.7578125, "learning_rate": 8.801449687322343e-06, "loss": 0.7303, "step": 34232 }, { "epoch": 3.6425835284103, "grad_norm": 11.921875, "learning_rate": 8.801414155770325e-06, "loss": 0.6379, "step": 34233 }, { "epoch": 3.6426899340285166, "grad_norm": 14.3828125, "learning_rate": 8.801378624218306e-06, "loss": 0.6824, "step": 34234 }, { "epoch": 3.6427963396467335, "grad_norm": 10.734375, "learning_rate": 8.801343092666288e-06, "loss": 0.6641, "step": 34235 }, { "epoch": 3.64290274526495, "grad_norm": 13.53125, "learning_rate": 8.801307561114271e-06, "loss": 0.8998, "step": 34236 }, { "epoch": 3.6430091508831666, "grad_norm": 11.578125, "learning_rate": 8.801272029562253e-06, "loss": 0.4824, "step": 34237 }, { "epoch": 3.6431155565013835, "grad_norm": 13.8359375, "learning_rate": 8.801236498010234e-06, "loss": 0.7379, "step": 34238 }, { "epoch": 3.6432219621196, "grad_norm": 11.0859375, "learning_rate": 8.801200966458216e-06, "loss": 0.5848, "step": 34239 }, { "epoch": 3.6433283677378165, "grad_norm": 17.859375, "learning_rate": 8.801165434906198e-06, "loss": 0.7788, "step": 34240 }, { "epoch": 3.643434773356033, "grad_norm": 10.609375, "learning_rate": 8.80112990335418e-06, "loss": 0.4978, "step": 34241 }, { "epoch": 3.64354117897425, "grad_norm": 14.2265625, "learning_rate": 8.801094371802161e-06, "loss": 0.6773, "step": 34242 }, { "epoch": 3.6436475845924665, "grad_norm": 14.1171875, "learning_rate": 8.801058840250142e-06, "loss": 0.6292, "step": 34243 }, { "epoch": 3.643753990210683, "grad_norm": 17.25, "learning_rate": 8.801023308698124e-06, "loss": 0.7268, "step": 34244 }, { "epoch": 3.6438603958288995, "grad_norm": 16.0625, "learning_rate": 8.800987777146106e-06, "loss": 0.732, "step": 34245 }, { "epoch": 3.6439668014471165, "grad_norm": 13.765625, "learning_rate": 8.800952245594089e-06, "loss": 0.6543, "step": 34246 }, { "epoch": 3.644073207065333, "grad_norm": 14.9375, "learning_rate": 8.80091671404207e-06, "loss": 0.7743, "step": 34247 }, { "epoch": 3.6441796126835495, "grad_norm": 12.4375, "learning_rate": 8.800881182490052e-06, "loss": 0.5687, "step": 34248 }, { "epoch": 3.6442860183017665, "grad_norm": 18.1875, "learning_rate": 8.800845650938034e-06, "loss": 0.7773, "step": 34249 }, { "epoch": 3.644392423919983, "grad_norm": 15.7265625, "learning_rate": 8.800810119386015e-06, "loss": 0.7213, "step": 34250 }, { "epoch": 3.6444988295381995, "grad_norm": 11.4296875, "learning_rate": 8.800774587833997e-06, "loss": 0.6002, "step": 34251 }, { "epoch": 3.6446052351564164, "grad_norm": 13.546875, "learning_rate": 8.800739056281979e-06, "loss": 0.5752, "step": 34252 }, { "epoch": 3.644711640774633, "grad_norm": 14.421875, "learning_rate": 8.80070352472996e-06, "loss": 0.7038, "step": 34253 }, { "epoch": 3.6448180463928495, "grad_norm": 10.421875, "learning_rate": 8.800667993177942e-06, "loss": 0.5473, "step": 34254 }, { "epoch": 3.6449244520110664, "grad_norm": 15.0859375, "learning_rate": 8.800632461625923e-06, "loss": 0.6108, "step": 34255 }, { "epoch": 3.645030857629283, "grad_norm": 20.640625, "learning_rate": 8.800596930073905e-06, "loss": 0.7884, "step": 34256 }, { "epoch": 3.6451372632474994, "grad_norm": 11.5390625, "learning_rate": 8.800561398521888e-06, "loss": 0.7019, "step": 34257 }, { "epoch": 3.6452436688657164, "grad_norm": 14.4140625, "learning_rate": 8.80052586696987e-06, "loss": 0.8768, "step": 34258 }, { "epoch": 3.645350074483933, "grad_norm": 10.8984375, "learning_rate": 8.800490335417852e-06, "loss": 0.6279, "step": 34259 }, { "epoch": 3.6454564801021494, "grad_norm": 14.03125, "learning_rate": 8.800454803865833e-06, "loss": 0.6741, "step": 34260 }, { "epoch": 3.645562885720366, "grad_norm": 17.421875, "learning_rate": 8.800419272313815e-06, "loss": 0.7166, "step": 34261 }, { "epoch": 3.6456692913385824, "grad_norm": 11.7421875, "learning_rate": 8.800383740761798e-06, "loss": 0.6419, "step": 34262 }, { "epoch": 3.6457756969567994, "grad_norm": 12.90625, "learning_rate": 8.80034820920978e-06, "loss": 0.6633, "step": 34263 }, { "epoch": 3.645882102575016, "grad_norm": 10.6875, "learning_rate": 8.800312677657761e-06, "loss": 0.6068, "step": 34264 }, { "epoch": 3.6459885081932324, "grad_norm": 12.8359375, "learning_rate": 8.800277146105743e-06, "loss": 0.6741, "step": 34265 }, { "epoch": 3.6460949138114493, "grad_norm": 11.5390625, "learning_rate": 8.800241614553725e-06, "loss": 0.5309, "step": 34266 }, { "epoch": 3.646201319429666, "grad_norm": 13.9296875, "learning_rate": 8.800206083001706e-06, "loss": 0.6276, "step": 34267 }, { "epoch": 3.6463077250478824, "grad_norm": 19.453125, "learning_rate": 8.800170551449688e-06, "loss": 0.697, "step": 34268 }, { "epoch": 3.6464141306660993, "grad_norm": 15.2109375, "learning_rate": 8.80013501989767e-06, "loss": 0.639, "step": 34269 }, { "epoch": 3.646520536284316, "grad_norm": 11.140625, "learning_rate": 8.800099488345651e-06, "loss": 0.6317, "step": 34270 }, { "epoch": 3.6466269419025323, "grad_norm": 12.6875, "learning_rate": 8.800063956793633e-06, "loss": 0.5025, "step": 34271 }, { "epoch": 3.6467333475207493, "grad_norm": 13.2734375, "learning_rate": 8.800028425241616e-06, "loss": 0.551, "step": 34272 }, { "epoch": 3.646839753138966, "grad_norm": 16.453125, "learning_rate": 8.799992893689597e-06, "loss": 0.8013, "step": 34273 }, { "epoch": 3.6469461587571823, "grad_norm": 11.9765625, "learning_rate": 8.799957362137579e-06, "loss": 0.6113, "step": 34274 }, { "epoch": 3.6470525643753993, "grad_norm": 15.9453125, "learning_rate": 8.79992183058556e-06, "loss": 0.8548, "step": 34275 }, { "epoch": 3.6471589699936158, "grad_norm": 10.4375, "learning_rate": 8.799886299033542e-06, "loss": 0.6278, "step": 34276 }, { "epoch": 3.6472653756118323, "grad_norm": 10.4921875, "learning_rate": 8.799850767481526e-06, "loss": 0.6176, "step": 34277 }, { "epoch": 3.647371781230049, "grad_norm": 15.6484375, "learning_rate": 8.799815235929507e-06, "loss": 0.6729, "step": 34278 }, { "epoch": 3.6474781868482657, "grad_norm": 12.9609375, "learning_rate": 8.799779704377489e-06, "loss": 0.6154, "step": 34279 }, { "epoch": 3.6475845924664823, "grad_norm": 11.2734375, "learning_rate": 8.79974417282547e-06, "loss": 0.5313, "step": 34280 }, { "epoch": 3.6476909980846988, "grad_norm": 10.578125, "learning_rate": 8.79970864127345e-06, "loss": 0.6271, "step": 34281 }, { "epoch": 3.6477974037029153, "grad_norm": 13.265625, "learning_rate": 8.799673109721434e-06, "loss": 0.5599, "step": 34282 }, { "epoch": 3.6479038093211322, "grad_norm": 10.3515625, "learning_rate": 8.799637578169415e-06, "loss": 0.5114, "step": 34283 }, { "epoch": 3.6480102149393487, "grad_norm": 14.484375, "learning_rate": 8.799602046617397e-06, "loss": 0.7663, "step": 34284 }, { "epoch": 3.6481166205575652, "grad_norm": 18.65625, "learning_rate": 8.799566515065378e-06, "loss": 0.6939, "step": 34285 }, { "epoch": 3.648223026175782, "grad_norm": 15.3828125, "learning_rate": 8.79953098351336e-06, "loss": 0.7196, "step": 34286 }, { "epoch": 3.6483294317939987, "grad_norm": 11.515625, "learning_rate": 8.799495451961343e-06, "loss": 0.6276, "step": 34287 }, { "epoch": 3.648435837412215, "grad_norm": 12.34375, "learning_rate": 8.799459920409325e-06, "loss": 0.6621, "step": 34288 }, { "epoch": 3.648542243030432, "grad_norm": 9.8203125, "learning_rate": 8.799424388857307e-06, "loss": 0.4671, "step": 34289 }, { "epoch": 3.6486486486486487, "grad_norm": 16.15625, "learning_rate": 8.799388857305288e-06, "loss": 0.7445, "step": 34290 }, { "epoch": 3.648755054266865, "grad_norm": 14.6015625, "learning_rate": 8.79935332575327e-06, "loss": 0.6558, "step": 34291 }, { "epoch": 3.648861459885082, "grad_norm": 13.625, "learning_rate": 8.799317794201251e-06, "loss": 0.6893, "step": 34292 }, { "epoch": 3.6489678655032987, "grad_norm": 16.96875, "learning_rate": 8.799282262649233e-06, "loss": 0.72, "step": 34293 }, { "epoch": 3.649074271121515, "grad_norm": 12.1796875, "learning_rate": 8.799246731097215e-06, "loss": 0.5762, "step": 34294 }, { "epoch": 3.649180676739732, "grad_norm": 10.8046875, "learning_rate": 8.799211199545196e-06, "loss": 0.6301, "step": 34295 }, { "epoch": 3.6492870823579486, "grad_norm": 12.484375, "learning_rate": 8.799175667993178e-06, "loss": 0.6291, "step": 34296 }, { "epoch": 3.649393487976165, "grad_norm": 21.78125, "learning_rate": 8.79914013644116e-06, "loss": 0.7076, "step": 34297 }, { "epoch": 3.6494998935943816, "grad_norm": 13.6015625, "learning_rate": 8.799104604889143e-06, "loss": 0.5915, "step": 34298 }, { "epoch": 3.649606299212598, "grad_norm": 13.71875, "learning_rate": 8.799069073337124e-06, "loss": 0.5928, "step": 34299 }, { "epoch": 3.649712704830815, "grad_norm": 12.5859375, "learning_rate": 8.799033541785106e-06, "loss": 0.6986, "step": 34300 }, { "epoch": 3.6498191104490316, "grad_norm": 10.703125, "learning_rate": 8.798998010233088e-06, "loss": 0.6918, "step": 34301 }, { "epoch": 3.649925516067248, "grad_norm": 13.859375, "learning_rate": 8.798962478681069e-06, "loss": 0.6681, "step": 34302 }, { "epoch": 3.650031921685465, "grad_norm": 14.2109375, "learning_rate": 8.798926947129052e-06, "loss": 0.8208, "step": 34303 }, { "epoch": 3.6501383273036816, "grad_norm": 11.4140625, "learning_rate": 8.798891415577034e-06, "loss": 0.622, "step": 34304 }, { "epoch": 3.650244732921898, "grad_norm": 11.21875, "learning_rate": 8.798855884025016e-06, "loss": 0.5913, "step": 34305 }, { "epoch": 3.650351138540115, "grad_norm": 12.6875, "learning_rate": 8.798820352472997e-06, "loss": 0.6083, "step": 34306 }, { "epoch": 3.6504575441583316, "grad_norm": 11.0625, "learning_rate": 8.798784820920979e-06, "loss": 0.5616, "step": 34307 }, { "epoch": 3.650563949776548, "grad_norm": 13.2265625, "learning_rate": 8.79874928936896e-06, "loss": 0.5707, "step": 34308 }, { "epoch": 3.650670355394765, "grad_norm": 11.5625, "learning_rate": 8.798713757816942e-06, "loss": 0.5032, "step": 34309 }, { "epoch": 3.6507767610129815, "grad_norm": 10.9765625, "learning_rate": 8.798678226264924e-06, "loss": 0.7182, "step": 34310 }, { "epoch": 3.650883166631198, "grad_norm": 8.78125, "learning_rate": 8.798642694712905e-06, "loss": 0.5456, "step": 34311 }, { "epoch": 3.650989572249415, "grad_norm": 15.453125, "learning_rate": 8.798607163160887e-06, "loss": 0.6059, "step": 34312 }, { "epoch": 3.6510959778676315, "grad_norm": 15.953125, "learning_rate": 8.79857163160887e-06, "loss": 0.7105, "step": 34313 }, { "epoch": 3.651202383485848, "grad_norm": 12.2109375, "learning_rate": 8.798536100056852e-06, "loss": 0.5837, "step": 34314 }, { "epoch": 3.6513087891040645, "grad_norm": 13.5, "learning_rate": 8.798500568504833e-06, "loss": 0.6064, "step": 34315 }, { "epoch": 3.6514151947222815, "grad_norm": 14.3671875, "learning_rate": 8.798465036952815e-06, "loss": 0.6756, "step": 34316 }, { "epoch": 3.651521600340498, "grad_norm": 13.953125, "learning_rate": 8.798429505400797e-06, "loss": 0.5487, "step": 34317 }, { "epoch": 3.6516280059587145, "grad_norm": 12.7421875, "learning_rate": 8.798393973848778e-06, "loss": 0.6707, "step": 34318 }, { "epoch": 3.651734411576931, "grad_norm": 25.375, "learning_rate": 8.79835844229676e-06, "loss": 0.8023, "step": 34319 }, { "epoch": 3.651840817195148, "grad_norm": 13.71875, "learning_rate": 8.798322910744741e-06, "loss": 0.7752, "step": 34320 }, { "epoch": 3.6519472228133645, "grad_norm": 16.578125, "learning_rate": 8.798287379192723e-06, "loss": 0.9532, "step": 34321 }, { "epoch": 3.652053628431581, "grad_norm": 11.8828125, "learning_rate": 8.798251847640705e-06, "loss": 0.7279, "step": 34322 }, { "epoch": 3.652160034049798, "grad_norm": 10.875, "learning_rate": 8.798216316088688e-06, "loss": 0.5869, "step": 34323 }, { "epoch": 3.6522664396680145, "grad_norm": 9.296875, "learning_rate": 8.79818078453667e-06, "loss": 0.4441, "step": 34324 }, { "epoch": 3.652372845286231, "grad_norm": 12.5234375, "learning_rate": 8.798145252984651e-06, "loss": 0.7683, "step": 34325 }, { "epoch": 3.652479250904448, "grad_norm": 13.859375, "learning_rate": 8.798109721432633e-06, "loss": 0.753, "step": 34326 }, { "epoch": 3.6525856565226644, "grad_norm": 16.84375, "learning_rate": 8.798074189880614e-06, "loss": 0.7962, "step": 34327 }, { "epoch": 3.652692062140881, "grad_norm": 17.734375, "learning_rate": 8.798038658328598e-06, "loss": 0.7505, "step": 34328 }, { "epoch": 3.652798467759098, "grad_norm": 13.5859375, "learning_rate": 8.79800312677658e-06, "loss": 0.7066, "step": 34329 }, { "epoch": 3.6529048733773144, "grad_norm": 13.609375, "learning_rate": 8.797967595224561e-06, "loss": 0.5936, "step": 34330 }, { "epoch": 3.653011278995531, "grad_norm": 11.5, "learning_rate": 8.797932063672542e-06, "loss": 0.5461, "step": 34331 }, { "epoch": 3.653117684613748, "grad_norm": 10.9375, "learning_rate": 8.797896532120524e-06, "loss": 0.7049, "step": 34332 }, { "epoch": 3.6532240902319644, "grad_norm": 13.484375, "learning_rate": 8.797861000568506e-06, "loss": 0.6175, "step": 34333 }, { "epoch": 3.653330495850181, "grad_norm": 12.390625, "learning_rate": 8.797825469016487e-06, "loss": 0.6746, "step": 34334 }, { "epoch": 3.6534369014683974, "grad_norm": 14.7109375, "learning_rate": 8.797789937464469e-06, "loss": 0.6776, "step": 34335 }, { "epoch": 3.653543307086614, "grad_norm": 13.0234375, "learning_rate": 8.79775440591245e-06, "loss": 0.662, "step": 34336 }, { "epoch": 3.653649712704831, "grad_norm": 13.46875, "learning_rate": 8.797718874360432e-06, "loss": 0.5688, "step": 34337 }, { "epoch": 3.6537561183230474, "grad_norm": 14.46875, "learning_rate": 8.797683342808414e-06, "loss": 0.8426, "step": 34338 }, { "epoch": 3.653862523941264, "grad_norm": 10.7890625, "learning_rate": 8.797647811256397e-06, "loss": 0.5735, "step": 34339 }, { "epoch": 3.653968929559481, "grad_norm": 12.3515625, "learning_rate": 8.797612279704379e-06, "loss": 0.6583, "step": 34340 }, { "epoch": 3.6540753351776973, "grad_norm": 11.9453125, "learning_rate": 8.79757674815236e-06, "loss": 0.6601, "step": 34341 }, { "epoch": 3.654181740795914, "grad_norm": 10.4453125, "learning_rate": 8.797541216600342e-06, "loss": 0.5335, "step": 34342 }, { "epoch": 3.654288146414131, "grad_norm": 14.078125, "learning_rate": 8.797505685048323e-06, "loss": 0.7729, "step": 34343 }, { "epoch": 3.6543945520323473, "grad_norm": 13.0234375, "learning_rate": 8.797470153496305e-06, "loss": 0.694, "step": 34344 }, { "epoch": 3.654500957650564, "grad_norm": 15.9609375, "learning_rate": 8.797434621944287e-06, "loss": 0.7222, "step": 34345 }, { "epoch": 3.6546073632687808, "grad_norm": 15.2734375, "learning_rate": 8.797399090392268e-06, "loss": 0.7, "step": 34346 }, { "epoch": 3.6547137688869973, "grad_norm": 9.8359375, "learning_rate": 8.79736355884025e-06, "loss": 0.6351, "step": 34347 }, { "epoch": 3.654820174505214, "grad_norm": 14.046875, "learning_rate": 8.797328027288231e-06, "loss": 0.6716, "step": 34348 }, { "epoch": 3.6549265801234307, "grad_norm": 15.90625, "learning_rate": 8.797292495736215e-06, "loss": 0.6835, "step": 34349 }, { "epoch": 3.6550329857416473, "grad_norm": 11.8359375, "learning_rate": 8.797256964184196e-06, "loss": 0.6197, "step": 34350 }, { "epoch": 3.6551393913598638, "grad_norm": 10.8125, "learning_rate": 8.797221432632178e-06, "loss": 0.5306, "step": 34351 }, { "epoch": 3.6552457969780807, "grad_norm": 13.3671875, "learning_rate": 8.79718590108016e-06, "loss": 0.6824, "step": 34352 }, { "epoch": 3.6553522025962972, "grad_norm": 11.2109375, "learning_rate": 8.797150369528141e-06, "loss": 0.6552, "step": 34353 }, { "epoch": 3.6554586082145137, "grad_norm": 11.015625, "learning_rate": 8.797114837976125e-06, "loss": 0.5377, "step": 34354 }, { "epoch": 3.6555650138327302, "grad_norm": 13.6953125, "learning_rate": 8.797079306424106e-06, "loss": 0.7829, "step": 34355 }, { "epoch": 3.6556714194509468, "grad_norm": 12.3828125, "learning_rate": 8.797043774872088e-06, "loss": 0.6407, "step": 34356 }, { "epoch": 3.6557778250691637, "grad_norm": 12.8359375, "learning_rate": 8.79700824332007e-06, "loss": 0.6749, "step": 34357 }, { "epoch": 3.65588423068738, "grad_norm": 14.2109375, "learning_rate": 8.796972711768051e-06, "loss": 0.6609, "step": 34358 }, { "epoch": 3.6559906363055967, "grad_norm": 11.3671875, "learning_rate": 8.796937180216033e-06, "loss": 0.6314, "step": 34359 }, { "epoch": 3.6560970419238137, "grad_norm": 12.046875, "learning_rate": 8.796901648664014e-06, "loss": 0.7111, "step": 34360 }, { "epoch": 3.65620344754203, "grad_norm": 10.03125, "learning_rate": 8.796866117111996e-06, "loss": 0.5738, "step": 34361 }, { "epoch": 3.6563098531602467, "grad_norm": 12.0390625, "learning_rate": 8.796830585559977e-06, "loss": 0.7011, "step": 34362 }, { "epoch": 3.6564162587784637, "grad_norm": 11.4609375, "learning_rate": 8.796795054007959e-06, "loss": 0.6547, "step": 34363 }, { "epoch": 3.65652266439668, "grad_norm": 17.890625, "learning_rate": 8.796759522455942e-06, "loss": 0.7856, "step": 34364 }, { "epoch": 3.6566290700148967, "grad_norm": 14.0703125, "learning_rate": 8.796723990903924e-06, "loss": 0.6581, "step": 34365 }, { "epoch": 3.6567354756331136, "grad_norm": 13.03125, "learning_rate": 8.796688459351906e-06, "loss": 0.7544, "step": 34366 }, { "epoch": 3.65684188125133, "grad_norm": 11.9296875, "learning_rate": 8.796652927799887e-06, "loss": 0.6705, "step": 34367 }, { "epoch": 3.6569482868695466, "grad_norm": 14.609375, "learning_rate": 8.796617396247869e-06, "loss": 0.6005, "step": 34368 }, { "epoch": 3.6570546924877636, "grad_norm": 12.34375, "learning_rate": 8.796581864695852e-06, "loss": 0.6654, "step": 34369 }, { "epoch": 3.65716109810598, "grad_norm": 15.046875, "learning_rate": 8.796546333143832e-06, "loss": 0.738, "step": 34370 }, { "epoch": 3.6572675037241966, "grad_norm": 13.0859375, "learning_rate": 8.796510801591814e-06, "loss": 0.6878, "step": 34371 }, { "epoch": 3.657373909342413, "grad_norm": 12.2578125, "learning_rate": 8.796475270039795e-06, "loss": 0.5274, "step": 34372 }, { "epoch": 3.65748031496063, "grad_norm": 10.5, "learning_rate": 8.796439738487777e-06, "loss": 0.6657, "step": 34373 }, { "epoch": 3.6575867205788466, "grad_norm": 15.5234375, "learning_rate": 8.796404206935758e-06, "loss": 0.7886, "step": 34374 }, { "epoch": 3.657693126197063, "grad_norm": 11.421875, "learning_rate": 8.796368675383742e-06, "loss": 0.5818, "step": 34375 }, { "epoch": 3.6577995318152796, "grad_norm": 11.671875, "learning_rate": 8.796333143831723e-06, "loss": 0.701, "step": 34376 }, { "epoch": 3.6579059374334966, "grad_norm": 11.59375, "learning_rate": 8.796297612279705e-06, "loss": 0.6477, "step": 34377 }, { "epoch": 3.658012343051713, "grad_norm": 12.9765625, "learning_rate": 8.796262080727686e-06, "loss": 0.6873, "step": 34378 }, { "epoch": 3.6581187486699296, "grad_norm": 15.125, "learning_rate": 8.796226549175668e-06, "loss": 0.7065, "step": 34379 }, { "epoch": 3.6582251542881465, "grad_norm": 14.3125, "learning_rate": 8.796191017623651e-06, "loss": 0.7403, "step": 34380 }, { "epoch": 3.658331559906363, "grad_norm": 16.1875, "learning_rate": 8.796155486071633e-06, "loss": 0.758, "step": 34381 }, { "epoch": 3.6584379655245796, "grad_norm": 12.0234375, "learning_rate": 8.796119954519615e-06, "loss": 0.7208, "step": 34382 }, { "epoch": 3.6585443711427965, "grad_norm": 14.4375, "learning_rate": 8.796084422967596e-06, "loss": 0.7167, "step": 34383 }, { "epoch": 3.658650776761013, "grad_norm": 13.34375, "learning_rate": 8.796048891415578e-06, "loss": 0.5958, "step": 34384 }, { "epoch": 3.6587571823792295, "grad_norm": 15.0, "learning_rate": 8.79601335986356e-06, "loss": 0.7406, "step": 34385 }, { "epoch": 3.6588635879974465, "grad_norm": 9.875, "learning_rate": 8.795977828311541e-06, "loss": 0.5751, "step": 34386 }, { "epoch": 3.658969993615663, "grad_norm": 13.296875, "learning_rate": 8.795942296759523e-06, "loss": 0.546, "step": 34387 }, { "epoch": 3.6590763992338795, "grad_norm": 14.671875, "learning_rate": 8.795906765207504e-06, "loss": 0.5944, "step": 34388 }, { "epoch": 3.6591828048520965, "grad_norm": 15.171875, "learning_rate": 8.795871233655486e-06, "loss": 0.6156, "step": 34389 }, { "epoch": 3.659289210470313, "grad_norm": 14.8515625, "learning_rate": 8.795835702103469e-06, "loss": 0.7511, "step": 34390 }, { "epoch": 3.6593956160885295, "grad_norm": 15.3046875, "learning_rate": 8.79580017055145e-06, "loss": 0.6285, "step": 34391 }, { "epoch": 3.659502021706746, "grad_norm": 12.9765625, "learning_rate": 8.795764638999432e-06, "loss": 0.5848, "step": 34392 }, { "epoch": 3.6596084273249625, "grad_norm": 11.6171875, "learning_rate": 8.795729107447414e-06, "loss": 0.6582, "step": 34393 }, { "epoch": 3.6597148329431795, "grad_norm": 10.2421875, "learning_rate": 8.795693575895396e-06, "loss": 0.6369, "step": 34394 }, { "epoch": 3.659821238561396, "grad_norm": 13.5703125, "learning_rate": 8.795658044343379e-06, "loss": 0.7935, "step": 34395 }, { "epoch": 3.6599276441796125, "grad_norm": 14.5859375, "learning_rate": 8.79562251279136e-06, "loss": 0.632, "step": 34396 }, { "epoch": 3.6600340497978294, "grad_norm": 12.7578125, "learning_rate": 8.795586981239342e-06, "loss": 0.713, "step": 34397 }, { "epoch": 3.660140455416046, "grad_norm": 11.9296875, "learning_rate": 8.795551449687324e-06, "loss": 0.5273, "step": 34398 }, { "epoch": 3.6602468610342624, "grad_norm": 17.140625, "learning_rate": 8.795515918135305e-06, "loss": 0.8094, "step": 34399 }, { "epoch": 3.6603532666524794, "grad_norm": 12.5703125, "learning_rate": 8.795480386583287e-06, "loss": 0.6604, "step": 34400 }, { "epoch": 3.660459672270696, "grad_norm": 16.171875, "learning_rate": 8.795444855031269e-06, "loss": 0.7917, "step": 34401 }, { "epoch": 3.6605660778889124, "grad_norm": 19.640625, "learning_rate": 8.79540932347925e-06, "loss": 0.8234, "step": 34402 }, { "epoch": 3.6606724835071294, "grad_norm": 15.4609375, "learning_rate": 8.795373791927232e-06, "loss": 0.6772, "step": 34403 }, { "epoch": 3.660778889125346, "grad_norm": 13.15625, "learning_rate": 8.795338260375213e-06, "loss": 0.6166, "step": 34404 }, { "epoch": 3.6608852947435624, "grad_norm": 14.9921875, "learning_rate": 8.795302728823197e-06, "loss": 0.6924, "step": 34405 }, { "epoch": 3.6609917003617793, "grad_norm": 13.1640625, "learning_rate": 8.795267197271178e-06, "loss": 0.6588, "step": 34406 }, { "epoch": 3.661098105979996, "grad_norm": 15.3515625, "learning_rate": 8.79523166571916e-06, "loss": 0.7354, "step": 34407 }, { "epoch": 3.6612045115982124, "grad_norm": 9.609375, "learning_rate": 8.795196134167141e-06, "loss": 0.5355, "step": 34408 }, { "epoch": 3.661310917216429, "grad_norm": 12.3359375, "learning_rate": 8.795160602615123e-06, "loss": 0.547, "step": 34409 }, { "epoch": 3.661417322834646, "grad_norm": 16.96875, "learning_rate": 8.795125071063105e-06, "loss": 0.7844, "step": 34410 }, { "epoch": 3.6615237284528623, "grad_norm": 14.359375, "learning_rate": 8.795089539511086e-06, "loss": 0.7816, "step": 34411 }, { "epoch": 3.661630134071079, "grad_norm": 14.53125, "learning_rate": 8.795054007959068e-06, "loss": 0.7689, "step": 34412 }, { "epoch": 3.6617365396892954, "grad_norm": 18.546875, "learning_rate": 8.79501847640705e-06, "loss": 0.7132, "step": 34413 }, { "epoch": 3.6618429453075123, "grad_norm": 15.0546875, "learning_rate": 8.794982944855031e-06, "loss": 0.6669, "step": 34414 }, { "epoch": 3.661949350925729, "grad_norm": 12.09375, "learning_rate": 8.794947413303013e-06, "loss": 0.6921, "step": 34415 }, { "epoch": 3.6620557565439453, "grad_norm": 11.0234375, "learning_rate": 8.794911881750996e-06, "loss": 0.5465, "step": 34416 }, { "epoch": 3.6621621621621623, "grad_norm": 11.6953125, "learning_rate": 8.794876350198978e-06, "loss": 0.6033, "step": 34417 }, { "epoch": 3.662268567780379, "grad_norm": 12.21875, "learning_rate": 8.79484081864696e-06, "loss": 0.5518, "step": 34418 }, { "epoch": 3.6623749733985953, "grad_norm": 12.9296875, "learning_rate": 8.79480528709494e-06, "loss": 0.6249, "step": 34419 }, { "epoch": 3.6624813790168123, "grad_norm": 13.3359375, "learning_rate": 8.794769755542922e-06, "loss": 0.5787, "step": 34420 }, { "epoch": 3.6625877846350288, "grad_norm": 22.140625, "learning_rate": 8.794734223990906e-06, "loss": 0.7966, "step": 34421 }, { "epoch": 3.6626941902532453, "grad_norm": 14.8515625, "learning_rate": 8.794698692438887e-06, "loss": 0.6764, "step": 34422 }, { "epoch": 3.6628005958714622, "grad_norm": 13.875, "learning_rate": 8.794663160886869e-06, "loss": 0.6887, "step": 34423 }, { "epoch": 3.6629070014896787, "grad_norm": 11.9375, "learning_rate": 8.79462762933485e-06, "loss": 0.7412, "step": 34424 }, { "epoch": 3.6630134071078952, "grad_norm": 14.1484375, "learning_rate": 8.794592097782832e-06, "loss": 0.6755, "step": 34425 }, { "epoch": 3.663119812726112, "grad_norm": 15.4609375, "learning_rate": 8.794556566230814e-06, "loss": 0.6559, "step": 34426 }, { "epoch": 3.6632262183443287, "grad_norm": 15.1015625, "learning_rate": 8.794521034678795e-06, "loss": 0.6681, "step": 34427 }, { "epoch": 3.663332623962545, "grad_norm": 13.3671875, "learning_rate": 8.794485503126777e-06, "loss": 0.6461, "step": 34428 }, { "epoch": 3.6634390295807617, "grad_norm": 17.046875, "learning_rate": 8.794449971574759e-06, "loss": 0.8024, "step": 34429 }, { "epoch": 3.6635454351989782, "grad_norm": 13.2890625, "learning_rate": 8.79441444002274e-06, "loss": 0.6666, "step": 34430 }, { "epoch": 3.663651840817195, "grad_norm": 16.0625, "learning_rate": 8.794378908470723e-06, "loss": 0.7599, "step": 34431 }, { "epoch": 3.6637582464354117, "grad_norm": 13.4921875, "learning_rate": 8.794343376918705e-06, "loss": 0.6097, "step": 34432 }, { "epoch": 3.663864652053628, "grad_norm": 11.0546875, "learning_rate": 8.794307845366687e-06, "loss": 0.5248, "step": 34433 }, { "epoch": 3.663971057671845, "grad_norm": 19.6875, "learning_rate": 8.794272313814668e-06, "loss": 0.7299, "step": 34434 }, { "epoch": 3.6640774632900617, "grad_norm": 13.8203125, "learning_rate": 8.79423678226265e-06, "loss": 0.7589, "step": 34435 }, { "epoch": 3.664183868908278, "grad_norm": 18.609375, "learning_rate": 8.794201250710632e-06, "loss": 0.8103, "step": 34436 }, { "epoch": 3.664290274526495, "grad_norm": 13.140625, "learning_rate": 8.794165719158613e-06, "loss": 0.587, "step": 34437 }, { "epoch": 3.6643966801447116, "grad_norm": 15.0, "learning_rate": 8.794130187606595e-06, "loss": 0.7939, "step": 34438 }, { "epoch": 3.664503085762928, "grad_norm": 12.71875, "learning_rate": 8.794094656054576e-06, "loss": 0.6059, "step": 34439 }, { "epoch": 3.664609491381145, "grad_norm": 27.03125, "learning_rate": 8.794059124502558e-06, "loss": 0.7763, "step": 34440 }, { "epoch": 3.6647158969993616, "grad_norm": 13.9296875, "learning_rate": 8.794023592950541e-06, "loss": 0.6343, "step": 34441 }, { "epoch": 3.664822302617578, "grad_norm": 15.2890625, "learning_rate": 8.793988061398523e-06, "loss": 0.6606, "step": 34442 }, { "epoch": 3.664928708235795, "grad_norm": 15.9453125, "learning_rate": 8.793952529846504e-06, "loss": 0.6971, "step": 34443 }, { "epoch": 3.6650351138540116, "grad_norm": 12.9296875, "learning_rate": 8.793916998294486e-06, "loss": 0.593, "step": 34444 }, { "epoch": 3.665141519472228, "grad_norm": 20.375, "learning_rate": 8.793881466742468e-06, "loss": 0.6844, "step": 34445 }, { "epoch": 3.6652479250904446, "grad_norm": 14.625, "learning_rate": 8.793845935190451e-06, "loss": 0.6652, "step": 34446 }, { "epoch": 3.6653543307086616, "grad_norm": 10.828125, "learning_rate": 8.793810403638433e-06, "loss": 0.5839, "step": 34447 }, { "epoch": 3.665460736326878, "grad_norm": 18.21875, "learning_rate": 8.793774872086414e-06, "loss": 0.7682, "step": 34448 }, { "epoch": 3.6655671419450946, "grad_norm": 14.2421875, "learning_rate": 8.793739340534396e-06, "loss": 0.6841, "step": 34449 }, { "epoch": 3.665673547563311, "grad_norm": 12.75, "learning_rate": 8.793703808982377e-06, "loss": 0.6771, "step": 34450 }, { "epoch": 3.665779953181528, "grad_norm": 12.0234375, "learning_rate": 8.793668277430359e-06, "loss": 0.5772, "step": 34451 }, { "epoch": 3.6658863587997446, "grad_norm": 9.9921875, "learning_rate": 8.79363274587834e-06, "loss": 0.6353, "step": 34452 }, { "epoch": 3.665992764417961, "grad_norm": 11.34375, "learning_rate": 8.793597214326322e-06, "loss": 0.6427, "step": 34453 }, { "epoch": 3.666099170036178, "grad_norm": 15.25, "learning_rate": 8.793561682774304e-06, "loss": 0.7375, "step": 34454 }, { "epoch": 3.6662055756543945, "grad_norm": 11.0625, "learning_rate": 8.793526151222285e-06, "loss": 0.621, "step": 34455 }, { "epoch": 3.666311981272611, "grad_norm": 16.25, "learning_rate": 8.793490619670267e-06, "loss": 0.7405, "step": 34456 }, { "epoch": 3.666418386890828, "grad_norm": 13.0859375, "learning_rate": 8.79345508811825e-06, "loss": 0.6739, "step": 34457 }, { "epoch": 3.6665247925090445, "grad_norm": 12.4609375, "learning_rate": 8.793419556566232e-06, "loss": 0.6219, "step": 34458 }, { "epoch": 3.666631198127261, "grad_norm": 14.4765625, "learning_rate": 8.793384025014214e-06, "loss": 0.5723, "step": 34459 }, { "epoch": 3.666737603745478, "grad_norm": 13.015625, "learning_rate": 8.793348493462195e-06, "loss": 0.636, "step": 34460 }, { "epoch": 3.6668440093636945, "grad_norm": 12.8203125, "learning_rate": 8.793312961910177e-06, "loss": 0.6659, "step": 34461 }, { "epoch": 3.666950414981911, "grad_norm": 13.796875, "learning_rate": 8.793277430358158e-06, "loss": 0.5903, "step": 34462 }, { "epoch": 3.667056820600128, "grad_norm": 16.046875, "learning_rate": 8.79324189880614e-06, "loss": 0.8494, "step": 34463 }, { "epoch": 3.6671632262183445, "grad_norm": 14.6015625, "learning_rate": 8.793206367254122e-06, "loss": 0.6416, "step": 34464 }, { "epoch": 3.667269631836561, "grad_norm": 11.5, "learning_rate": 8.793170835702103e-06, "loss": 0.6888, "step": 34465 }, { "epoch": 3.6673760374547775, "grad_norm": 13.15625, "learning_rate": 8.793135304150085e-06, "loss": 0.6984, "step": 34466 }, { "epoch": 3.667482443072994, "grad_norm": 12.9140625, "learning_rate": 8.793099772598068e-06, "loss": 0.5454, "step": 34467 }, { "epoch": 3.667588848691211, "grad_norm": 14.71875, "learning_rate": 8.79306424104605e-06, "loss": 0.6526, "step": 34468 }, { "epoch": 3.6676952543094274, "grad_norm": 16.34375, "learning_rate": 8.793028709494031e-06, "loss": 0.6937, "step": 34469 }, { "epoch": 3.667801659927644, "grad_norm": 12.890625, "learning_rate": 8.792993177942013e-06, "loss": 0.5405, "step": 34470 }, { "epoch": 3.667908065545861, "grad_norm": 15.125, "learning_rate": 8.792957646389995e-06, "loss": 0.8319, "step": 34471 }, { "epoch": 3.6680144711640774, "grad_norm": 13.46875, "learning_rate": 8.792922114837978e-06, "loss": 0.6283, "step": 34472 }, { "epoch": 3.668120876782294, "grad_norm": 13.15625, "learning_rate": 8.79288658328596e-06, "loss": 0.5722, "step": 34473 }, { "epoch": 3.668227282400511, "grad_norm": 10.8046875, "learning_rate": 8.792851051733941e-06, "loss": 0.6334, "step": 34474 }, { "epoch": 3.6683336880187274, "grad_norm": 11.6953125, "learning_rate": 8.792815520181923e-06, "loss": 0.7546, "step": 34475 }, { "epoch": 3.668440093636944, "grad_norm": 12.03125, "learning_rate": 8.792779988629904e-06, "loss": 0.5967, "step": 34476 }, { "epoch": 3.668546499255161, "grad_norm": 14.15625, "learning_rate": 8.792744457077886e-06, "loss": 0.7278, "step": 34477 }, { "epoch": 3.6686529048733774, "grad_norm": 13.0859375, "learning_rate": 8.792708925525867e-06, "loss": 0.7323, "step": 34478 }, { "epoch": 3.668759310491594, "grad_norm": 9.625, "learning_rate": 8.792673393973849e-06, "loss": 0.5369, "step": 34479 }, { "epoch": 3.668865716109811, "grad_norm": 10.453125, "learning_rate": 8.79263786242183e-06, "loss": 0.6094, "step": 34480 }, { "epoch": 3.6689721217280273, "grad_norm": 13.3515625, "learning_rate": 8.792602330869812e-06, "loss": 0.7385, "step": 34481 }, { "epoch": 3.669078527346244, "grad_norm": 12.265625, "learning_rate": 8.792566799317796e-06, "loss": 0.6738, "step": 34482 }, { "epoch": 3.669184932964461, "grad_norm": 10.953125, "learning_rate": 8.792531267765777e-06, "loss": 0.6386, "step": 34483 }, { "epoch": 3.6692913385826773, "grad_norm": 13.484375, "learning_rate": 8.792495736213759e-06, "loss": 0.6307, "step": 34484 }, { "epoch": 3.669397744200894, "grad_norm": 10.6875, "learning_rate": 8.79246020466174e-06, "loss": 0.6351, "step": 34485 }, { "epoch": 3.6695041498191103, "grad_norm": 14.21875, "learning_rate": 8.792424673109722e-06, "loss": 0.6515, "step": 34486 }, { "epoch": 3.669610555437327, "grad_norm": 12.2890625, "learning_rate": 8.792389141557705e-06, "loss": 0.5772, "step": 34487 }, { "epoch": 3.669716961055544, "grad_norm": 13.015625, "learning_rate": 8.792353610005687e-06, "loss": 0.659, "step": 34488 }, { "epoch": 3.6698233666737603, "grad_norm": 13.328125, "learning_rate": 8.792318078453669e-06, "loss": 0.6727, "step": 34489 }, { "epoch": 3.669929772291977, "grad_norm": 13.484375, "learning_rate": 8.792282546901648e-06, "loss": 0.5194, "step": 34490 }, { "epoch": 3.6700361779101938, "grad_norm": 15.921875, "learning_rate": 8.79224701534963e-06, "loss": 0.7565, "step": 34491 }, { "epoch": 3.6701425835284103, "grad_norm": 11.1328125, "learning_rate": 8.792211483797612e-06, "loss": 0.5019, "step": 34492 }, { "epoch": 3.670248989146627, "grad_norm": 11.0078125, "learning_rate": 8.792175952245595e-06, "loss": 0.6861, "step": 34493 }, { "epoch": 3.6703553947648437, "grad_norm": 12.140625, "learning_rate": 8.792140420693577e-06, "loss": 0.6418, "step": 34494 }, { "epoch": 3.6704618003830602, "grad_norm": 13.6640625, "learning_rate": 8.792104889141558e-06, "loss": 0.64, "step": 34495 }, { "epoch": 3.6705682060012768, "grad_norm": 12.34375, "learning_rate": 8.79206935758954e-06, "loss": 0.5742, "step": 34496 }, { "epoch": 3.6706746116194937, "grad_norm": 12.9140625, "learning_rate": 8.792033826037521e-06, "loss": 0.5962, "step": 34497 }, { "epoch": 3.67078101723771, "grad_norm": 13.6015625, "learning_rate": 8.791998294485505e-06, "loss": 0.6856, "step": 34498 }, { "epoch": 3.6708874228559267, "grad_norm": 10.75, "learning_rate": 8.791962762933486e-06, "loss": 0.4979, "step": 34499 }, { "epoch": 3.6709938284741437, "grad_norm": 13.8359375, "learning_rate": 8.791927231381468e-06, "loss": 0.732, "step": 34500 }, { "epoch": 3.67110023409236, "grad_norm": 10.5078125, "learning_rate": 8.79189169982945e-06, "loss": 0.4848, "step": 34501 }, { "epoch": 3.6712066397105767, "grad_norm": 13.0390625, "learning_rate": 8.791856168277431e-06, "loss": 0.5814, "step": 34502 }, { "epoch": 3.671313045328793, "grad_norm": 18.953125, "learning_rate": 8.791820636725413e-06, "loss": 0.5776, "step": 34503 }, { "epoch": 3.67141945094701, "grad_norm": 13.9375, "learning_rate": 8.791785105173394e-06, "loss": 0.6992, "step": 34504 }, { "epoch": 3.6715258565652267, "grad_norm": 12.78125, "learning_rate": 8.791749573621376e-06, "loss": 0.7241, "step": 34505 }, { "epoch": 3.671632262183443, "grad_norm": 11.453125, "learning_rate": 8.791714042069358e-06, "loss": 0.6239, "step": 34506 }, { "epoch": 3.6717386678016597, "grad_norm": 11.34375, "learning_rate": 8.791678510517339e-06, "loss": 0.6251, "step": 34507 }, { "epoch": 3.6718450734198766, "grad_norm": 11.1640625, "learning_rate": 8.791642978965322e-06, "loss": 0.6221, "step": 34508 }, { "epoch": 3.671951479038093, "grad_norm": 14.921875, "learning_rate": 8.791607447413304e-06, "loss": 0.8155, "step": 34509 }, { "epoch": 3.6720578846563097, "grad_norm": 14.6875, "learning_rate": 8.791571915861286e-06, "loss": 0.7806, "step": 34510 }, { "epoch": 3.6721642902745266, "grad_norm": 11.3125, "learning_rate": 8.791536384309267e-06, "loss": 0.7125, "step": 34511 }, { "epoch": 3.672270695892743, "grad_norm": 12.65625, "learning_rate": 8.791500852757249e-06, "loss": 0.5804, "step": 34512 }, { "epoch": 3.6723771015109596, "grad_norm": 16.96875, "learning_rate": 8.791465321205232e-06, "loss": 0.7775, "step": 34513 }, { "epoch": 3.6724835071291766, "grad_norm": 11.171875, "learning_rate": 8.791429789653214e-06, "loss": 0.5762, "step": 34514 }, { "epoch": 3.672589912747393, "grad_norm": 12.8671875, "learning_rate": 8.791394258101195e-06, "loss": 0.649, "step": 34515 }, { "epoch": 3.6726963183656096, "grad_norm": 13.34375, "learning_rate": 8.791358726549177e-06, "loss": 0.6299, "step": 34516 }, { "epoch": 3.6728027239838266, "grad_norm": 13.6796875, "learning_rate": 8.791323194997159e-06, "loss": 0.6906, "step": 34517 }, { "epoch": 3.672909129602043, "grad_norm": 12.84375, "learning_rate": 8.79128766344514e-06, "loss": 0.6924, "step": 34518 }, { "epoch": 3.6730155352202596, "grad_norm": 12.171875, "learning_rate": 8.791252131893122e-06, "loss": 0.6486, "step": 34519 }, { "epoch": 3.6731219408384765, "grad_norm": 15.40625, "learning_rate": 8.791216600341103e-06, "loss": 0.762, "step": 34520 }, { "epoch": 3.673228346456693, "grad_norm": 14.3046875, "learning_rate": 8.791181068789085e-06, "loss": 0.7599, "step": 34521 }, { "epoch": 3.6733347520749096, "grad_norm": 10.0859375, "learning_rate": 8.791145537237067e-06, "loss": 0.5018, "step": 34522 }, { "epoch": 3.673441157693126, "grad_norm": 10.8984375, "learning_rate": 8.79111000568505e-06, "loss": 0.5258, "step": 34523 }, { "epoch": 3.6735475633113426, "grad_norm": 13.6015625, "learning_rate": 8.791074474133032e-06, "loss": 0.8286, "step": 34524 }, { "epoch": 3.6736539689295595, "grad_norm": 13.09375, "learning_rate": 8.791038942581013e-06, "loss": 0.6339, "step": 34525 }, { "epoch": 3.673760374547776, "grad_norm": 13.9921875, "learning_rate": 8.791003411028995e-06, "loss": 0.6186, "step": 34526 }, { "epoch": 3.6738667801659926, "grad_norm": 16.40625, "learning_rate": 8.790967879476976e-06, "loss": 0.8912, "step": 34527 }, { "epoch": 3.6739731857842095, "grad_norm": 15.859375, "learning_rate": 8.790932347924958e-06, "loss": 0.6196, "step": 34528 }, { "epoch": 3.674079591402426, "grad_norm": 14.3359375, "learning_rate": 8.79089681637294e-06, "loss": 0.5859, "step": 34529 }, { "epoch": 3.6741859970206425, "grad_norm": 12.9296875, "learning_rate": 8.790861284820921e-06, "loss": 0.5362, "step": 34530 }, { "epoch": 3.6742924026388595, "grad_norm": 14.09375, "learning_rate": 8.790825753268903e-06, "loss": 0.6859, "step": 34531 }, { "epoch": 3.674398808257076, "grad_norm": 17.640625, "learning_rate": 8.790790221716884e-06, "loss": 0.7873, "step": 34532 }, { "epoch": 3.6745052138752925, "grad_norm": 9.8515625, "learning_rate": 8.790754690164866e-06, "loss": 0.5271, "step": 34533 }, { "epoch": 3.6746116194935095, "grad_norm": 14.703125, "learning_rate": 8.79071915861285e-06, "loss": 0.7785, "step": 34534 }, { "epoch": 3.674718025111726, "grad_norm": 15.515625, "learning_rate": 8.790683627060831e-06, "loss": 0.7636, "step": 34535 }, { "epoch": 3.6748244307299425, "grad_norm": 12.8359375, "learning_rate": 8.790648095508813e-06, "loss": 0.6146, "step": 34536 }, { "epoch": 3.6749308363481594, "grad_norm": 13.734375, "learning_rate": 8.790612563956794e-06, "loss": 0.6902, "step": 34537 }, { "epoch": 3.675037241966376, "grad_norm": 12.4375, "learning_rate": 8.790577032404776e-06, "loss": 0.7013, "step": 34538 }, { "epoch": 3.6751436475845924, "grad_norm": 11.3046875, "learning_rate": 8.790541500852759e-06, "loss": 0.5362, "step": 34539 }, { "epoch": 3.675250053202809, "grad_norm": 12.9921875, "learning_rate": 8.79050596930074e-06, "loss": 0.6442, "step": 34540 }, { "epoch": 3.675356458821026, "grad_norm": 12.4375, "learning_rate": 8.790470437748722e-06, "loss": 0.5895, "step": 34541 }, { "epoch": 3.6754628644392424, "grad_norm": 15.8515625, "learning_rate": 8.790434906196704e-06, "loss": 0.7342, "step": 34542 }, { "epoch": 3.675569270057459, "grad_norm": 16.046875, "learning_rate": 8.790399374644685e-06, "loss": 0.8269, "step": 34543 }, { "epoch": 3.6756756756756754, "grad_norm": 13.0078125, "learning_rate": 8.790363843092667e-06, "loss": 0.5931, "step": 34544 }, { "epoch": 3.6757820812938924, "grad_norm": 13.5, "learning_rate": 8.790328311540649e-06, "loss": 0.6162, "step": 34545 }, { "epoch": 3.675888486912109, "grad_norm": 11.4296875, "learning_rate": 8.79029277998863e-06, "loss": 0.5695, "step": 34546 }, { "epoch": 3.6759948925303254, "grad_norm": 14.6484375, "learning_rate": 8.790257248436612e-06, "loss": 0.723, "step": 34547 }, { "epoch": 3.6761012981485424, "grad_norm": 23.453125, "learning_rate": 8.790221716884593e-06, "loss": 0.8746, "step": 34548 }, { "epoch": 3.676207703766759, "grad_norm": 77.1875, "learning_rate": 8.790186185332577e-06, "loss": 0.7754, "step": 34549 }, { "epoch": 3.6763141093849754, "grad_norm": 17.546875, "learning_rate": 8.790150653780558e-06, "loss": 0.8026, "step": 34550 }, { "epoch": 3.6764205150031923, "grad_norm": 15.1484375, "learning_rate": 8.79011512222854e-06, "loss": 0.7171, "step": 34551 }, { "epoch": 3.676526920621409, "grad_norm": 12.0859375, "learning_rate": 8.790079590676522e-06, "loss": 0.669, "step": 34552 }, { "epoch": 3.6766333262396254, "grad_norm": 14.4453125, "learning_rate": 8.790044059124503e-06, "loss": 0.6619, "step": 34553 }, { "epoch": 3.6767397318578423, "grad_norm": 13.75, "learning_rate": 8.790008527572485e-06, "loss": 0.5719, "step": 34554 }, { "epoch": 3.676846137476059, "grad_norm": 15.4140625, "learning_rate": 8.789972996020466e-06, "loss": 0.8012, "step": 34555 }, { "epoch": 3.6769525430942753, "grad_norm": 18.84375, "learning_rate": 8.789937464468448e-06, "loss": 0.9684, "step": 34556 }, { "epoch": 3.6770589487124923, "grad_norm": 11.9140625, "learning_rate": 8.78990193291643e-06, "loss": 0.7056, "step": 34557 }, { "epoch": 3.677165354330709, "grad_norm": 13.90625, "learning_rate": 8.789866401364411e-06, "loss": 0.7249, "step": 34558 }, { "epoch": 3.6772717599489253, "grad_norm": 13.28125, "learning_rate": 8.789830869812395e-06, "loss": 0.6842, "step": 34559 }, { "epoch": 3.677378165567142, "grad_norm": 20.046875, "learning_rate": 8.789795338260376e-06, "loss": 0.8032, "step": 34560 }, { "epoch": 3.6774845711853583, "grad_norm": 17.953125, "learning_rate": 8.789759806708358e-06, "loss": 0.8091, "step": 34561 }, { "epoch": 3.6775909768035753, "grad_norm": 12.609375, "learning_rate": 8.78972427515634e-06, "loss": 0.6345, "step": 34562 }, { "epoch": 3.677697382421792, "grad_norm": 12.203125, "learning_rate": 8.789688743604321e-06, "loss": 0.5357, "step": 34563 }, { "epoch": 3.6778037880400083, "grad_norm": 17.171875, "learning_rate": 8.789653212052304e-06, "loss": 0.5863, "step": 34564 }, { "epoch": 3.6779101936582252, "grad_norm": 13.0234375, "learning_rate": 8.789617680500286e-06, "loss": 0.5736, "step": 34565 }, { "epoch": 3.6780165992764418, "grad_norm": 15.4140625, "learning_rate": 8.789582148948267e-06, "loss": 0.7199, "step": 34566 }, { "epoch": 3.6781230048946583, "grad_norm": 14.84375, "learning_rate": 8.789546617396249e-06, "loss": 0.6941, "step": 34567 }, { "epoch": 3.678229410512875, "grad_norm": 23.71875, "learning_rate": 8.78951108584423e-06, "loss": 0.8177, "step": 34568 }, { "epoch": 3.6783358161310917, "grad_norm": 14.28125, "learning_rate": 8.789475554292212e-06, "loss": 0.6202, "step": 34569 }, { "epoch": 3.6784422217493082, "grad_norm": 10.6171875, "learning_rate": 8.789440022740194e-06, "loss": 0.6817, "step": 34570 }, { "epoch": 3.678548627367525, "grad_norm": 13.90625, "learning_rate": 8.789404491188176e-06, "loss": 0.6827, "step": 34571 }, { "epoch": 3.6786550329857417, "grad_norm": 14.2421875, "learning_rate": 8.789368959636157e-06, "loss": 0.7808, "step": 34572 }, { "epoch": 3.678761438603958, "grad_norm": 15.75, "learning_rate": 8.789333428084139e-06, "loss": 0.8448, "step": 34573 }, { "epoch": 3.678867844222175, "grad_norm": 12.15625, "learning_rate": 8.78929789653212e-06, "loss": 0.4839, "step": 34574 }, { "epoch": 3.6789742498403917, "grad_norm": 15.015625, "learning_rate": 8.789262364980104e-06, "loss": 0.7568, "step": 34575 }, { "epoch": 3.679080655458608, "grad_norm": 13.9296875, "learning_rate": 8.789226833428085e-06, "loss": 0.6766, "step": 34576 }, { "epoch": 3.6791870610768247, "grad_norm": 11.2421875, "learning_rate": 8.789191301876067e-06, "loss": 0.7665, "step": 34577 }, { "epoch": 3.6792934666950416, "grad_norm": 16.5625, "learning_rate": 8.789155770324048e-06, "loss": 0.67, "step": 34578 }, { "epoch": 3.679399872313258, "grad_norm": 14.59375, "learning_rate": 8.78912023877203e-06, "loss": 0.6946, "step": 34579 }, { "epoch": 3.6795062779314747, "grad_norm": 13.375, "learning_rate": 8.789084707220012e-06, "loss": 0.6402, "step": 34580 }, { "epoch": 3.679612683549691, "grad_norm": 13.890625, "learning_rate": 8.789049175667993e-06, "loss": 0.7194, "step": 34581 }, { "epoch": 3.679719089167908, "grad_norm": 19.578125, "learning_rate": 8.789013644115975e-06, "loss": 0.8866, "step": 34582 }, { "epoch": 3.6798254947861246, "grad_norm": 13.90625, "learning_rate": 8.788978112563956e-06, "loss": 0.6292, "step": 34583 }, { "epoch": 3.679931900404341, "grad_norm": 11.7734375, "learning_rate": 8.788942581011938e-06, "loss": 0.5846, "step": 34584 }, { "epoch": 3.680038306022558, "grad_norm": 13.390625, "learning_rate": 8.788907049459921e-06, "loss": 0.6672, "step": 34585 }, { "epoch": 3.6801447116407746, "grad_norm": 11.5390625, "learning_rate": 8.788871517907903e-06, "loss": 0.5434, "step": 34586 }, { "epoch": 3.680251117258991, "grad_norm": 12.171875, "learning_rate": 8.788835986355885e-06, "loss": 0.8026, "step": 34587 }, { "epoch": 3.680357522877208, "grad_norm": 13.140625, "learning_rate": 8.788800454803866e-06, "loss": 0.7092, "step": 34588 }, { "epoch": 3.6804639284954246, "grad_norm": 11.546875, "learning_rate": 8.788764923251848e-06, "loss": 0.5853, "step": 34589 }, { "epoch": 3.680570334113641, "grad_norm": 15.421875, "learning_rate": 8.788729391699831e-06, "loss": 0.6764, "step": 34590 }, { "epoch": 3.680676739731858, "grad_norm": 13.4609375, "learning_rate": 8.788693860147813e-06, "loss": 0.6026, "step": 34591 }, { "epoch": 3.6807831453500746, "grad_norm": 10.4765625, "learning_rate": 8.788658328595794e-06, "loss": 0.5092, "step": 34592 }, { "epoch": 3.680889550968291, "grad_norm": 27.625, "learning_rate": 8.788622797043776e-06, "loss": 0.8016, "step": 34593 }, { "epoch": 3.680995956586508, "grad_norm": 15.03125, "learning_rate": 8.788587265491758e-06, "loss": 0.6882, "step": 34594 }, { "epoch": 3.6811023622047245, "grad_norm": 13.734375, "learning_rate": 8.78855173393974e-06, "loss": 0.7498, "step": 34595 }, { "epoch": 3.681208767822941, "grad_norm": 13.5078125, "learning_rate": 8.78851620238772e-06, "loss": 0.6268, "step": 34596 }, { "epoch": 3.6813151734411576, "grad_norm": 11.5390625, "learning_rate": 8.788480670835702e-06, "loss": 0.6336, "step": 34597 }, { "epoch": 3.681421579059374, "grad_norm": 12.5390625, "learning_rate": 8.788445139283684e-06, "loss": 0.5713, "step": 34598 }, { "epoch": 3.681527984677591, "grad_norm": 14.9296875, "learning_rate": 8.788409607731666e-06, "loss": 0.6797, "step": 34599 }, { "epoch": 3.6816343902958075, "grad_norm": 12.7265625, "learning_rate": 8.788374076179649e-06, "loss": 0.6103, "step": 34600 }, { "epoch": 3.681740795914024, "grad_norm": 13.78125, "learning_rate": 8.78833854462763e-06, "loss": 0.6361, "step": 34601 }, { "epoch": 3.681847201532241, "grad_norm": 13.2421875, "learning_rate": 8.788303013075612e-06, "loss": 0.6769, "step": 34602 }, { "epoch": 3.6819536071504575, "grad_norm": 9.4140625, "learning_rate": 8.788267481523594e-06, "loss": 0.4616, "step": 34603 }, { "epoch": 3.682060012768674, "grad_norm": 11.234375, "learning_rate": 8.788231949971575e-06, "loss": 0.5083, "step": 34604 }, { "epoch": 3.682166418386891, "grad_norm": 16.09375, "learning_rate": 8.788196418419559e-06, "loss": 0.6081, "step": 34605 }, { "epoch": 3.6822728240051075, "grad_norm": 13.484375, "learning_rate": 8.78816088686754e-06, "loss": 0.66, "step": 34606 }, { "epoch": 3.682379229623324, "grad_norm": 11.765625, "learning_rate": 8.788125355315522e-06, "loss": 0.6696, "step": 34607 }, { "epoch": 3.682485635241541, "grad_norm": 14.6875, "learning_rate": 8.788089823763503e-06, "loss": 0.6278, "step": 34608 }, { "epoch": 3.6825920408597574, "grad_norm": 9.359375, "learning_rate": 8.788054292211485e-06, "loss": 0.5307, "step": 34609 }, { "epoch": 3.682698446477974, "grad_norm": 10.8984375, "learning_rate": 8.788018760659465e-06, "loss": 0.5754, "step": 34610 }, { "epoch": 3.682804852096191, "grad_norm": 18.0, "learning_rate": 8.787983229107448e-06, "loss": 0.7746, "step": 34611 }, { "epoch": 3.6829112577144074, "grad_norm": 13.1484375, "learning_rate": 8.78794769755543e-06, "loss": 0.5393, "step": 34612 }, { "epoch": 3.683017663332624, "grad_norm": 13.890625, "learning_rate": 8.787912166003411e-06, "loss": 0.6328, "step": 34613 }, { "epoch": 3.683124068950841, "grad_norm": 12.125, "learning_rate": 8.787876634451393e-06, "loss": 0.7163, "step": 34614 }, { "epoch": 3.6832304745690574, "grad_norm": 12.78125, "learning_rate": 8.787841102899375e-06, "loss": 0.6682, "step": 34615 }, { "epoch": 3.683336880187274, "grad_norm": 13.5, "learning_rate": 8.787805571347358e-06, "loss": 0.61, "step": 34616 }, { "epoch": 3.6834432858054904, "grad_norm": 13.2265625, "learning_rate": 8.78777003979534e-06, "loss": 0.6038, "step": 34617 }, { "epoch": 3.683549691423707, "grad_norm": 16.0625, "learning_rate": 8.787734508243321e-06, "loss": 0.7542, "step": 34618 }, { "epoch": 3.683656097041924, "grad_norm": 12.546875, "learning_rate": 8.787698976691303e-06, "loss": 0.7122, "step": 34619 }, { "epoch": 3.6837625026601404, "grad_norm": 14.9140625, "learning_rate": 8.787663445139284e-06, "loss": 0.768, "step": 34620 }, { "epoch": 3.683868908278357, "grad_norm": 10.9453125, "learning_rate": 8.787627913587266e-06, "loss": 0.5667, "step": 34621 }, { "epoch": 3.683975313896574, "grad_norm": 12.4296875, "learning_rate": 8.787592382035248e-06, "loss": 0.7143, "step": 34622 }, { "epoch": 3.6840817195147904, "grad_norm": 12.65625, "learning_rate": 8.78755685048323e-06, "loss": 0.5749, "step": 34623 }, { "epoch": 3.684188125133007, "grad_norm": 13.9453125, "learning_rate": 8.78752131893121e-06, "loss": 0.736, "step": 34624 }, { "epoch": 3.684294530751224, "grad_norm": 9.40625, "learning_rate": 8.787485787379192e-06, "loss": 0.478, "step": 34625 }, { "epoch": 3.6844009363694403, "grad_norm": 10.7109375, "learning_rate": 8.787450255827176e-06, "loss": 0.5872, "step": 34626 }, { "epoch": 3.684507341987657, "grad_norm": 14.2421875, "learning_rate": 8.787414724275157e-06, "loss": 0.8215, "step": 34627 }, { "epoch": 3.684613747605874, "grad_norm": 14.3203125, "learning_rate": 8.787379192723139e-06, "loss": 0.742, "step": 34628 }, { "epoch": 3.6847201532240903, "grad_norm": 13.375, "learning_rate": 8.78734366117112e-06, "loss": 0.6787, "step": 34629 }, { "epoch": 3.684826558842307, "grad_norm": 15.9921875, "learning_rate": 8.787308129619102e-06, "loss": 0.6282, "step": 34630 }, { "epoch": 3.6849329644605238, "grad_norm": 15.1484375, "learning_rate": 8.787272598067085e-06, "loss": 0.837, "step": 34631 }, { "epoch": 3.6850393700787403, "grad_norm": 16.5, "learning_rate": 8.787237066515067e-06, "loss": 0.8436, "step": 34632 }, { "epoch": 3.685145775696957, "grad_norm": 12.1171875, "learning_rate": 8.787201534963049e-06, "loss": 0.6142, "step": 34633 }, { "epoch": 3.6852521813151733, "grad_norm": 12.6328125, "learning_rate": 8.78716600341103e-06, "loss": 0.6864, "step": 34634 }, { "epoch": 3.6853585869333902, "grad_norm": 17.796875, "learning_rate": 8.787130471859012e-06, "loss": 0.8275, "step": 34635 }, { "epoch": 3.6854649925516068, "grad_norm": 12.875, "learning_rate": 8.787094940306994e-06, "loss": 0.6295, "step": 34636 }, { "epoch": 3.6855713981698233, "grad_norm": 20.921875, "learning_rate": 8.787059408754975e-06, "loss": 0.8252, "step": 34637 }, { "epoch": 3.6856778037880398, "grad_norm": 15.5859375, "learning_rate": 8.787023877202957e-06, "loss": 0.7459, "step": 34638 }, { "epoch": 3.6857842094062567, "grad_norm": 16.703125, "learning_rate": 8.786988345650938e-06, "loss": 0.7367, "step": 34639 }, { "epoch": 3.6858906150244732, "grad_norm": 12.4140625, "learning_rate": 8.78695281409892e-06, "loss": 0.5827, "step": 34640 }, { "epoch": 3.6859970206426897, "grad_norm": 9.7890625, "learning_rate": 8.786917282546903e-06, "loss": 0.4192, "step": 34641 }, { "epoch": 3.6861034262609067, "grad_norm": 12.84375, "learning_rate": 8.786881750994885e-06, "loss": 0.7196, "step": 34642 }, { "epoch": 3.686209831879123, "grad_norm": 9.7421875, "learning_rate": 8.786846219442866e-06, "loss": 0.5635, "step": 34643 }, { "epoch": 3.6863162374973397, "grad_norm": 14.1015625, "learning_rate": 8.786810687890848e-06, "loss": 0.6811, "step": 34644 }, { "epoch": 3.6864226431155567, "grad_norm": 10.3515625, "learning_rate": 8.78677515633883e-06, "loss": 0.5678, "step": 34645 }, { "epoch": 3.686529048733773, "grad_norm": 12.4921875, "learning_rate": 8.786739624786811e-06, "loss": 0.6514, "step": 34646 }, { "epoch": 3.6866354543519897, "grad_norm": 14.5390625, "learning_rate": 8.786704093234793e-06, "loss": 0.6091, "step": 34647 }, { "epoch": 3.6867418599702066, "grad_norm": 10.984375, "learning_rate": 8.786668561682774e-06, "loss": 0.668, "step": 34648 }, { "epoch": 3.686848265588423, "grad_norm": 11.234375, "learning_rate": 8.786633030130756e-06, "loss": 0.6541, "step": 34649 }, { "epoch": 3.6869546712066397, "grad_norm": 17.53125, "learning_rate": 8.786597498578738e-06, "loss": 0.6925, "step": 34650 }, { "epoch": 3.6870610768248566, "grad_norm": 14.328125, "learning_rate": 8.78656196702672e-06, "loss": 0.6785, "step": 34651 }, { "epoch": 3.687167482443073, "grad_norm": 11.6484375, "learning_rate": 8.786526435474703e-06, "loss": 0.6452, "step": 34652 }, { "epoch": 3.6872738880612896, "grad_norm": 13.1796875, "learning_rate": 8.786490903922684e-06, "loss": 0.6086, "step": 34653 }, { "epoch": 3.687380293679506, "grad_norm": 12.2421875, "learning_rate": 8.786455372370666e-06, "loss": 0.6621, "step": 34654 }, { "epoch": 3.6874866992977227, "grad_norm": 13.8359375, "learning_rate": 8.786419840818647e-06, "loss": 0.6766, "step": 34655 }, { "epoch": 3.6875931049159396, "grad_norm": 11.484375, "learning_rate": 8.786384309266629e-06, "loss": 0.5901, "step": 34656 }, { "epoch": 3.687699510534156, "grad_norm": 10.6953125, "learning_rate": 8.786348777714612e-06, "loss": 0.6607, "step": 34657 }, { "epoch": 3.6878059161523726, "grad_norm": 15.84375, "learning_rate": 8.786313246162594e-06, "loss": 0.7126, "step": 34658 }, { "epoch": 3.6879123217705896, "grad_norm": 14.1953125, "learning_rate": 8.786277714610576e-06, "loss": 0.7812, "step": 34659 }, { "epoch": 3.688018727388806, "grad_norm": 12.8515625, "learning_rate": 8.786242183058557e-06, "loss": 0.55, "step": 34660 }, { "epoch": 3.6881251330070226, "grad_norm": 13.4140625, "learning_rate": 8.786206651506539e-06, "loss": 0.678, "step": 34661 }, { "epoch": 3.6882315386252396, "grad_norm": 11.65625, "learning_rate": 8.78617111995452e-06, "loss": 0.6285, "step": 34662 }, { "epoch": 3.688337944243456, "grad_norm": 13.109375, "learning_rate": 8.786135588402502e-06, "loss": 0.6672, "step": 34663 }, { "epoch": 3.6884443498616726, "grad_norm": 17.890625, "learning_rate": 8.786100056850484e-06, "loss": 0.8193, "step": 34664 }, { "epoch": 3.6885507554798895, "grad_norm": 12.21875, "learning_rate": 8.786064525298465e-06, "loss": 0.6824, "step": 34665 }, { "epoch": 3.688657161098106, "grad_norm": 13.125, "learning_rate": 8.786028993746447e-06, "loss": 0.5949, "step": 34666 }, { "epoch": 3.6887635667163226, "grad_norm": 16.25, "learning_rate": 8.78599346219443e-06, "loss": 0.9174, "step": 34667 }, { "epoch": 3.6888699723345395, "grad_norm": 15.546875, "learning_rate": 8.785957930642412e-06, "loss": 0.7821, "step": 34668 }, { "epoch": 3.688976377952756, "grad_norm": 14.765625, "learning_rate": 8.785922399090393e-06, "loss": 0.8608, "step": 34669 }, { "epoch": 3.6890827835709725, "grad_norm": 14.453125, "learning_rate": 8.785886867538375e-06, "loss": 0.597, "step": 34670 }, { "epoch": 3.689189189189189, "grad_norm": 10.453125, "learning_rate": 8.785851335986357e-06, "loss": 0.617, "step": 34671 }, { "epoch": 3.689295594807406, "grad_norm": 13.8125, "learning_rate": 8.785815804434338e-06, "loss": 0.5882, "step": 34672 }, { "epoch": 3.6894020004256225, "grad_norm": 12.75, "learning_rate": 8.78578027288232e-06, "loss": 0.6486, "step": 34673 }, { "epoch": 3.689508406043839, "grad_norm": 11.6875, "learning_rate": 8.785744741330301e-06, "loss": 0.6224, "step": 34674 }, { "epoch": 3.6896148116620555, "grad_norm": 12.2421875, "learning_rate": 8.785709209778283e-06, "loss": 0.681, "step": 34675 }, { "epoch": 3.6897212172802725, "grad_norm": 12.796875, "learning_rate": 8.785673678226265e-06, "loss": 0.7827, "step": 34676 }, { "epoch": 3.689827622898489, "grad_norm": 13.3828125, "learning_rate": 8.785638146674248e-06, "loss": 0.7357, "step": 34677 }, { "epoch": 3.6899340285167055, "grad_norm": 11.453125, "learning_rate": 8.78560261512223e-06, "loss": 0.6574, "step": 34678 }, { "epoch": 3.6900404341349224, "grad_norm": 11.6875, "learning_rate": 8.785567083570211e-06, "loss": 0.6461, "step": 34679 }, { "epoch": 3.690146839753139, "grad_norm": 13.3671875, "learning_rate": 8.785531552018193e-06, "loss": 0.591, "step": 34680 }, { "epoch": 3.6902532453713555, "grad_norm": 13.4765625, "learning_rate": 8.785496020466174e-06, "loss": 0.7431, "step": 34681 }, { "epoch": 3.6903596509895724, "grad_norm": 11.1875, "learning_rate": 8.785460488914158e-06, "loss": 0.6167, "step": 34682 }, { "epoch": 3.690466056607789, "grad_norm": 11.1484375, "learning_rate": 8.78542495736214e-06, "loss": 0.6012, "step": 34683 }, { "epoch": 3.6905724622260054, "grad_norm": 21.5, "learning_rate": 8.78538942581012e-06, "loss": 0.8901, "step": 34684 }, { "epoch": 3.6906788678442224, "grad_norm": 13.6171875, "learning_rate": 8.785353894258102e-06, "loss": 0.7713, "step": 34685 }, { "epoch": 3.690785273462439, "grad_norm": 14.34375, "learning_rate": 8.785318362706084e-06, "loss": 0.5875, "step": 34686 }, { "epoch": 3.6908916790806554, "grad_norm": 13.328125, "learning_rate": 8.785282831154066e-06, "loss": 0.7588, "step": 34687 }, { "epoch": 3.6909980846988724, "grad_norm": 11.7890625, "learning_rate": 8.785247299602047e-06, "loss": 0.6991, "step": 34688 }, { "epoch": 3.691104490317089, "grad_norm": 15.4921875, "learning_rate": 8.785211768050029e-06, "loss": 0.6242, "step": 34689 }, { "epoch": 3.6912108959353054, "grad_norm": 13.0859375, "learning_rate": 8.78517623649801e-06, "loss": 0.5639, "step": 34690 }, { "epoch": 3.691317301553522, "grad_norm": 14.2734375, "learning_rate": 8.785140704945992e-06, "loss": 0.6954, "step": 34691 }, { "epoch": 3.6914237071717384, "grad_norm": 15.2265625, "learning_rate": 8.785105173393974e-06, "loss": 0.7166, "step": 34692 }, { "epoch": 3.6915301127899554, "grad_norm": 14.65625, "learning_rate": 8.785069641841957e-06, "loss": 0.743, "step": 34693 }, { "epoch": 3.691636518408172, "grad_norm": 12.2109375, "learning_rate": 8.785034110289939e-06, "loss": 0.6477, "step": 34694 }, { "epoch": 3.6917429240263884, "grad_norm": 10.1875, "learning_rate": 8.78499857873792e-06, "loss": 0.5284, "step": 34695 }, { "epoch": 3.6918493296446053, "grad_norm": 16.296875, "learning_rate": 8.784963047185902e-06, "loss": 0.7959, "step": 34696 }, { "epoch": 3.691955735262822, "grad_norm": 17.09375, "learning_rate": 8.784927515633883e-06, "loss": 0.7357, "step": 34697 }, { "epoch": 3.6920621408810383, "grad_norm": 13.1875, "learning_rate": 8.784891984081867e-06, "loss": 0.6946, "step": 34698 }, { "epoch": 3.6921685464992553, "grad_norm": 12.9765625, "learning_rate": 8.784856452529847e-06, "loss": 0.599, "step": 34699 }, { "epoch": 3.692274952117472, "grad_norm": 12.4453125, "learning_rate": 8.784820920977828e-06, "loss": 0.6347, "step": 34700 }, { "epoch": 3.6923813577356883, "grad_norm": 12.0390625, "learning_rate": 8.78478538942581e-06, "loss": 0.5791, "step": 34701 }, { "epoch": 3.6924877633539053, "grad_norm": 14.1015625, "learning_rate": 8.784749857873791e-06, "loss": 0.6217, "step": 34702 }, { "epoch": 3.692594168972122, "grad_norm": 16.328125, "learning_rate": 8.784714326321775e-06, "loss": 0.7043, "step": 34703 }, { "epoch": 3.6927005745903383, "grad_norm": 12.0703125, "learning_rate": 8.784678794769756e-06, "loss": 0.5539, "step": 34704 }, { "epoch": 3.6928069802085552, "grad_norm": 17.0, "learning_rate": 8.784643263217738e-06, "loss": 0.6747, "step": 34705 }, { "epoch": 3.6929133858267718, "grad_norm": 12.5234375, "learning_rate": 8.78460773166572e-06, "loss": 0.759, "step": 34706 }, { "epoch": 3.6930197914449883, "grad_norm": 9.09375, "learning_rate": 8.784572200113701e-06, "loss": 0.4738, "step": 34707 }, { "epoch": 3.6931261970632048, "grad_norm": 14.2578125, "learning_rate": 8.784536668561684e-06, "loss": 0.6939, "step": 34708 }, { "epoch": 3.6932326026814217, "grad_norm": 16.953125, "learning_rate": 8.784501137009666e-06, "loss": 0.6433, "step": 34709 }, { "epoch": 3.6933390082996382, "grad_norm": 15.7265625, "learning_rate": 8.784465605457648e-06, "loss": 0.6935, "step": 34710 }, { "epoch": 3.6934454139178547, "grad_norm": 13.7734375, "learning_rate": 8.78443007390563e-06, "loss": 0.5752, "step": 34711 }, { "epoch": 3.6935518195360713, "grad_norm": 20.375, "learning_rate": 8.784394542353611e-06, "loss": 0.8896, "step": 34712 }, { "epoch": 3.693658225154288, "grad_norm": 12.390625, "learning_rate": 8.784359010801592e-06, "loss": 0.6185, "step": 34713 }, { "epoch": 3.6937646307725047, "grad_norm": 12.75, "learning_rate": 8.784323479249574e-06, "loss": 0.6761, "step": 34714 }, { "epoch": 3.6938710363907212, "grad_norm": 15.15625, "learning_rate": 8.784287947697556e-06, "loss": 0.6998, "step": 34715 }, { "epoch": 3.693977442008938, "grad_norm": 12.3515625, "learning_rate": 8.784252416145537e-06, "loss": 0.7144, "step": 34716 }, { "epoch": 3.6940838476271547, "grad_norm": 15.8984375, "learning_rate": 8.784216884593519e-06, "loss": 0.6698, "step": 34717 }, { "epoch": 3.694190253245371, "grad_norm": 15.859375, "learning_rate": 8.784181353041502e-06, "loss": 0.7172, "step": 34718 }, { "epoch": 3.694296658863588, "grad_norm": 9.9765625, "learning_rate": 8.784145821489484e-06, "loss": 0.5873, "step": 34719 }, { "epoch": 3.6944030644818047, "grad_norm": 11.1484375, "learning_rate": 8.784110289937465e-06, "loss": 0.6223, "step": 34720 }, { "epoch": 3.694509470100021, "grad_norm": 12.78125, "learning_rate": 8.784074758385447e-06, "loss": 0.5995, "step": 34721 }, { "epoch": 3.694615875718238, "grad_norm": 34.9375, "learning_rate": 8.784039226833429e-06, "loss": 0.8398, "step": 34722 }, { "epoch": 3.6947222813364546, "grad_norm": 15.0703125, "learning_rate": 8.78400369528141e-06, "loss": 0.6734, "step": 34723 }, { "epoch": 3.694828686954671, "grad_norm": 12.6328125, "learning_rate": 8.783968163729394e-06, "loss": 0.6017, "step": 34724 }, { "epoch": 3.694935092572888, "grad_norm": 22.1875, "learning_rate": 8.783932632177375e-06, "loss": 0.7124, "step": 34725 }, { "epoch": 3.6950414981911046, "grad_norm": 11.59375, "learning_rate": 8.783897100625357e-06, "loss": 0.7159, "step": 34726 }, { "epoch": 3.695147903809321, "grad_norm": 12.6875, "learning_rate": 8.783861569073338e-06, "loss": 0.7079, "step": 34727 }, { "epoch": 3.6952543094275376, "grad_norm": 12.9609375, "learning_rate": 8.78382603752132e-06, "loss": 0.6341, "step": 34728 }, { "epoch": 3.695360715045754, "grad_norm": 12.3203125, "learning_rate": 8.783790505969302e-06, "loss": 0.6353, "step": 34729 }, { "epoch": 3.695467120663971, "grad_norm": 14.6796875, "learning_rate": 8.783754974417283e-06, "loss": 0.6437, "step": 34730 }, { "epoch": 3.6955735262821876, "grad_norm": 12.8515625, "learning_rate": 8.783719442865265e-06, "loss": 0.6222, "step": 34731 }, { "epoch": 3.695679931900404, "grad_norm": 14.3515625, "learning_rate": 8.783683911313246e-06, "loss": 0.6721, "step": 34732 }, { "epoch": 3.695786337518621, "grad_norm": 13.296875, "learning_rate": 8.783648379761228e-06, "loss": 0.5466, "step": 34733 }, { "epoch": 3.6958927431368376, "grad_norm": 19.03125, "learning_rate": 8.783612848209211e-06, "loss": 0.7353, "step": 34734 }, { "epoch": 3.695999148755054, "grad_norm": 11.390625, "learning_rate": 8.783577316657193e-06, "loss": 0.7244, "step": 34735 }, { "epoch": 3.696105554373271, "grad_norm": 16.28125, "learning_rate": 8.783541785105175e-06, "loss": 0.6901, "step": 34736 }, { "epoch": 3.6962119599914876, "grad_norm": 16.515625, "learning_rate": 8.783506253553156e-06, "loss": 0.8148, "step": 34737 }, { "epoch": 3.696318365609704, "grad_norm": 11.90625, "learning_rate": 8.783470722001138e-06, "loss": 0.555, "step": 34738 }, { "epoch": 3.696424771227921, "grad_norm": 14.390625, "learning_rate": 8.78343519044912e-06, "loss": 0.7241, "step": 34739 }, { "epoch": 3.6965311768461375, "grad_norm": 12.140625, "learning_rate": 8.783399658897101e-06, "loss": 0.7129, "step": 34740 }, { "epoch": 3.696637582464354, "grad_norm": 13.1328125, "learning_rate": 8.783364127345083e-06, "loss": 0.6298, "step": 34741 }, { "epoch": 3.696743988082571, "grad_norm": 11.2109375, "learning_rate": 8.783328595793064e-06, "loss": 0.5363, "step": 34742 }, { "epoch": 3.6968503937007875, "grad_norm": 13.2578125, "learning_rate": 8.783293064241046e-06, "loss": 0.595, "step": 34743 }, { "epoch": 3.696956799319004, "grad_norm": 12.9609375, "learning_rate": 8.783257532689029e-06, "loss": 0.6649, "step": 34744 }, { "epoch": 3.697063204937221, "grad_norm": 12.171875, "learning_rate": 8.78322200113701e-06, "loss": 0.632, "step": 34745 }, { "epoch": 3.6971696105554375, "grad_norm": 22.46875, "learning_rate": 8.783186469584992e-06, "loss": 0.7848, "step": 34746 }, { "epoch": 3.697276016173654, "grad_norm": 11.03125, "learning_rate": 8.783150938032974e-06, "loss": 0.5288, "step": 34747 }, { "epoch": 3.6973824217918705, "grad_norm": 13.765625, "learning_rate": 8.783115406480955e-06, "loss": 0.6347, "step": 34748 }, { "epoch": 3.697488827410087, "grad_norm": 15.0546875, "learning_rate": 8.783079874928939e-06, "loss": 0.6756, "step": 34749 }, { "epoch": 3.697595233028304, "grad_norm": 10.9140625, "learning_rate": 8.78304434337692e-06, "loss": 0.6046, "step": 34750 }, { "epoch": 3.6977016386465205, "grad_norm": 11.2890625, "learning_rate": 8.783008811824902e-06, "loss": 0.6226, "step": 34751 }, { "epoch": 3.697808044264737, "grad_norm": 9.9296875, "learning_rate": 8.782973280272884e-06, "loss": 0.554, "step": 34752 }, { "epoch": 3.697914449882954, "grad_norm": 14.625, "learning_rate": 8.782937748720865e-06, "loss": 0.8061, "step": 34753 }, { "epoch": 3.6980208555011704, "grad_norm": 13.0390625, "learning_rate": 8.782902217168847e-06, "loss": 0.6552, "step": 34754 }, { "epoch": 3.698127261119387, "grad_norm": 13.15625, "learning_rate": 8.782866685616828e-06, "loss": 0.6785, "step": 34755 }, { "epoch": 3.698233666737604, "grad_norm": 14.203125, "learning_rate": 8.78283115406481e-06, "loss": 0.7324, "step": 34756 }, { "epoch": 3.6983400723558204, "grad_norm": 16.515625, "learning_rate": 8.782795622512792e-06, "loss": 0.7401, "step": 34757 }, { "epoch": 3.698446477974037, "grad_norm": 12.0234375, "learning_rate": 8.782760090960773e-06, "loss": 0.5457, "step": 34758 }, { "epoch": 3.698552883592254, "grad_norm": 18.125, "learning_rate": 8.782724559408757e-06, "loss": 0.7751, "step": 34759 }, { "epoch": 3.6986592892104704, "grad_norm": 9.859375, "learning_rate": 8.782689027856738e-06, "loss": 0.6233, "step": 34760 }, { "epoch": 3.698765694828687, "grad_norm": 13.5, "learning_rate": 8.78265349630472e-06, "loss": 0.7274, "step": 34761 }, { "epoch": 3.698872100446904, "grad_norm": 11.15625, "learning_rate": 8.782617964752701e-06, "loss": 0.5596, "step": 34762 }, { "epoch": 3.6989785060651204, "grad_norm": 10.9921875, "learning_rate": 8.782582433200683e-06, "loss": 0.594, "step": 34763 }, { "epoch": 3.699084911683337, "grad_norm": 13.6875, "learning_rate": 8.782546901648665e-06, "loss": 0.6346, "step": 34764 }, { "epoch": 3.6991913173015534, "grad_norm": 14.6953125, "learning_rate": 8.782511370096646e-06, "loss": 0.7979, "step": 34765 }, { "epoch": 3.6992977229197703, "grad_norm": 25.640625, "learning_rate": 8.782475838544628e-06, "loss": 0.7883, "step": 34766 }, { "epoch": 3.699404128537987, "grad_norm": 14.8203125, "learning_rate": 8.78244030699261e-06, "loss": 0.5996, "step": 34767 }, { "epoch": 3.6995105341562033, "grad_norm": 19.421875, "learning_rate": 8.782404775440591e-06, "loss": 0.8192, "step": 34768 }, { "epoch": 3.69961693977442, "grad_norm": 12.671875, "learning_rate": 8.782369243888573e-06, "loss": 0.6203, "step": 34769 }, { "epoch": 3.699723345392637, "grad_norm": 16.140625, "learning_rate": 8.782333712336556e-06, "loss": 0.732, "step": 34770 }, { "epoch": 3.6998297510108533, "grad_norm": 12.2734375, "learning_rate": 8.782298180784538e-06, "loss": 0.7233, "step": 34771 }, { "epoch": 3.69993615662907, "grad_norm": 11.625, "learning_rate": 8.782262649232519e-06, "loss": 0.6749, "step": 34772 }, { "epoch": 3.700042562247287, "grad_norm": 15.09375, "learning_rate": 8.7822271176805e-06, "loss": 0.6999, "step": 34773 }, { "epoch": 3.7001489678655033, "grad_norm": 12.828125, "learning_rate": 8.782191586128482e-06, "loss": 0.7792, "step": 34774 }, { "epoch": 3.70025537348372, "grad_norm": 10.4375, "learning_rate": 8.782156054576466e-06, "loss": 0.6204, "step": 34775 }, { "epoch": 3.7003617791019368, "grad_norm": 17.34375, "learning_rate": 8.782120523024447e-06, "loss": 0.6913, "step": 34776 }, { "epoch": 3.7004681847201533, "grad_norm": 12.7734375, "learning_rate": 8.782084991472429e-06, "loss": 0.5925, "step": 34777 }, { "epoch": 3.7005745903383698, "grad_norm": 14.234375, "learning_rate": 8.78204945992041e-06, "loss": 0.6208, "step": 34778 }, { "epoch": 3.7006809959565867, "grad_norm": 13.6796875, "learning_rate": 8.782013928368392e-06, "loss": 0.6966, "step": 34779 }, { "epoch": 3.7007874015748032, "grad_norm": 13.5625, "learning_rate": 8.781978396816374e-06, "loss": 0.5524, "step": 34780 }, { "epoch": 3.7008938071930197, "grad_norm": 12.28125, "learning_rate": 8.781942865264355e-06, "loss": 0.6261, "step": 34781 }, { "epoch": 3.7010002128112367, "grad_norm": 12.1640625, "learning_rate": 8.781907333712337e-06, "loss": 0.717, "step": 34782 }, { "epoch": 3.701106618429453, "grad_norm": 16.171875, "learning_rate": 8.781871802160318e-06, "loss": 0.7161, "step": 34783 }, { "epoch": 3.7012130240476697, "grad_norm": 16.625, "learning_rate": 8.7818362706083e-06, "loss": 0.7012, "step": 34784 }, { "epoch": 3.7013194296658862, "grad_norm": 10.9375, "learning_rate": 8.781800739056283e-06, "loss": 0.5069, "step": 34785 }, { "epoch": 3.7014258352841027, "grad_norm": 12.265625, "learning_rate": 8.781765207504265e-06, "loss": 0.6645, "step": 34786 }, { "epoch": 3.7015322409023197, "grad_norm": 10.2265625, "learning_rate": 8.781729675952247e-06, "loss": 0.5408, "step": 34787 }, { "epoch": 3.701638646520536, "grad_norm": 12.2109375, "learning_rate": 8.781694144400228e-06, "loss": 0.7206, "step": 34788 }, { "epoch": 3.7017450521387527, "grad_norm": 13.671875, "learning_rate": 8.78165861284821e-06, "loss": 0.5, "step": 34789 }, { "epoch": 3.7018514577569697, "grad_norm": 11.3359375, "learning_rate": 8.781623081296191e-06, "loss": 0.4925, "step": 34790 }, { "epoch": 3.701957863375186, "grad_norm": 14.015625, "learning_rate": 8.781587549744173e-06, "loss": 0.5926, "step": 34791 }, { "epoch": 3.7020642689934027, "grad_norm": 15.046875, "learning_rate": 8.781552018192155e-06, "loss": 0.664, "step": 34792 }, { "epoch": 3.7021706746116196, "grad_norm": 12.046875, "learning_rate": 8.781516486640136e-06, "loss": 0.5724, "step": 34793 }, { "epoch": 3.702277080229836, "grad_norm": 10.7578125, "learning_rate": 8.781480955088118e-06, "loss": 0.5886, "step": 34794 }, { "epoch": 3.7023834858480527, "grad_norm": 19.109375, "learning_rate": 8.781445423536101e-06, "loss": 0.8012, "step": 34795 }, { "epoch": 3.7024898914662696, "grad_norm": 14.484375, "learning_rate": 8.781409891984083e-06, "loss": 0.5718, "step": 34796 }, { "epoch": 3.702596297084486, "grad_norm": 13.296875, "learning_rate": 8.781374360432064e-06, "loss": 0.7435, "step": 34797 }, { "epoch": 3.7027027027027026, "grad_norm": 10.9765625, "learning_rate": 8.781338828880046e-06, "loss": 0.6538, "step": 34798 }, { "epoch": 3.7028091083209196, "grad_norm": 12.8046875, "learning_rate": 8.781303297328028e-06, "loss": 0.7205, "step": 34799 }, { "epoch": 3.702915513939136, "grad_norm": 12.984375, "learning_rate": 8.781267765776011e-06, "loss": 0.7429, "step": 34800 }, { "epoch": 3.7030219195573526, "grad_norm": 18.453125, "learning_rate": 8.781232234223992e-06, "loss": 0.8261, "step": 34801 }, { "epoch": 3.703128325175569, "grad_norm": 11.59375, "learning_rate": 8.781196702671974e-06, "loss": 0.5539, "step": 34802 }, { "epoch": 3.703234730793786, "grad_norm": 9.984375, "learning_rate": 8.781161171119956e-06, "loss": 0.5298, "step": 34803 }, { "epoch": 3.7033411364120026, "grad_norm": 14.2734375, "learning_rate": 8.781125639567937e-06, "loss": 0.6909, "step": 34804 }, { "epoch": 3.703447542030219, "grad_norm": 11.3125, "learning_rate": 8.781090108015919e-06, "loss": 0.7172, "step": 34805 }, { "epoch": 3.7035539476484356, "grad_norm": 11.1640625, "learning_rate": 8.7810545764639e-06, "loss": 0.656, "step": 34806 }, { "epoch": 3.7036603532666525, "grad_norm": 16.0, "learning_rate": 8.781019044911882e-06, "loss": 0.7787, "step": 34807 }, { "epoch": 3.703766758884869, "grad_norm": 15.7890625, "learning_rate": 8.780983513359864e-06, "loss": 0.7708, "step": 34808 }, { "epoch": 3.7038731645030856, "grad_norm": 12.4375, "learning_rate": 8.780947981807845e-06, "loss": 0.6413, "step": 34809 }, { "epoch": 3.7039795701213025, "grad_norm": 15.3046875, "learning_rate": 8.780912450255827e-06, "loss": 0.7409, "step": 34810 }, { "epoch": 3.704085975739519, "grad_norm": 16.15625, "learning_rate": 8.78087691870381e-06, "loss": 0.702, "step": 34811 }, { "epoch": 3.7041923813577355, "grad_norm": 14.953125, "learning_rate": 8.780841387151792e-06, "loss": 0.6971, "step": 34812 }, { "epoch": 3.7042987869759525, "grad_norm": 15.3671875, "learning_rate": 8.780805855599773e-06, "loss": 0.6578, "step": 34813 }, { "epoch": 3.704405192594169, "grad_norm": 14.6875, "learning_rate": 8.780770324047755e-06, "loss": 0.6857, "step": 34814 }, { "epoch": 3.7045115982123855, "grad_norm": 13.3046875, "learning_rate": 8.780734792495737e-06, "loss": 0.6225, "step": 34815 }, { "epoch": 3.7046180038306025, "grad_norm": 15.0390625, "learning_rate": 8.78069926094372e-06, "loss": 0.5635, "step": 34816 }, { "epoch": 3.704724409448819, "grad_norm": 13.4921875, "learning_rate": 8.780663729391702e-06, "loss": 0.6498, "step": 34817 }, { "epoch": 3.7048308150670355, "grad_norm": 17.890625, "learning_rate": 8.780628197839681e-06, "loss": 0.8014, "step": 34818 }, { "epoch": 3.7049372206852524, "grad_norm": 10.3984375, "learning_rate": 8.780592666287663e-06, "loss": 0.5977, "step": 34819 }, { "epoch": 3.705043626303469, "grad_norm": 14.5078125, "learning_rate": 8.780557134735645e-06, "loss": 0.6412, "step": 34820 }, { "epoch": 3.7051500319216855, "grad_norm": 13.0546875, "learning_rate": 8.780521603183628e-06, "loss": 0.7221, "step": 34821 }, { "epoch": 3.705256437539902, "grad_norm": 13.140625, "learning_rate": 8.78048607163161e-06, "loss": 0.6013, "step": 34822 }, { "epoch": 3.7053628431581185, "grad_norm": 14.7890625, "learning_rate": 8.780450540079591e-06, "loss": 0.6825, "step": 34823 }, { "epoch": 3.7054692487763354, "grad_norm": 15.875, "learning_rate": 8.780415008527573e-06, "loss": 0.8813, "step": 34824 }, { "epoch": 3.705575654394552, "grad_norm": 11.078125, "learning_rate": 8.780379476975554e-06, "loss": 0.7151, "step": 34825 }, { "epoch": 3.7056820600127685, "grad_norm": 9.46875, "learning_rate": 8.780343945423538e-06, "loss": 0.571, "step": 34826 }, { "epoch": 3.7057884656309854, "grad_norm": 15.5078125, "learning_rate": 8.78030841387152e-06, "loss": 0.6898, "step": 34827 }, { "epoch": 3.705894871249202, "grad_norm": 19.734375, "learning_rate": 8.780272882319501e-06, "loss": 0.681, "step": 34828 }, { "epoch": 3.7060012768674184, "grad_norm": 12.484375, "learning_rate": 8.780237350767483e-06, "loss": 0.5315, "step": 34829 }, { "epoch": 3.7061076824856354, "grad_norm": 9.15625, "learning_rate": 8.780201819215464e-06, "loss": 0.5113, "step": 34830 }, { "epoch": 3.706214088103852, "grad_norm": 13.6953125, "learning_rate": 8.780166287663446e-06, "loss": 0.6045, "step": 34831 }, { "epoch": 3.7063204937220684, "grad_norm": 10.015625, "learning_rate": 8.780130756111427e-06, "loss": 0.5848, "step": 34832 }, { "epoch": 3.7064268993402854, "grad_norm": 14.421875, "learning_rate": 8.780095224559409e-06, "loss": 0.6524, "step": 34833 }, { "epoch": 3.706533304958502, "grad_norm": 9.96875, "learning_rate": 8.78005969300739e-06, "loss": 0.561, "step": 34834 }, { "epoch": 3.7066397105767184, "grad_norm": 13.375, "learning_rate": 8.780024161455372e-06, "loss": 0.8172, "step": 34835 }, { "epoch": 3.7067461161949353, "grad_norm": 10.390625, "learning_rate": 8.779988629903355e-06, "loss": 0.5814, "step": 34836 }, { "epoch": 3.706852521813152, "grad_norm": 15.9921875, "learning_rate": 8.779953098351337e-06, "loss": 0.8298, "step": 34837 }, { "epoch": 3.7069589274313683, "grad_norm": 11.359375, "learning_rate": 8.779917566799319e-06, "loss": 0.6507, "step": 34838 }, { "epoch": 3.707065333049585, "grad_norm": 11.140625, "learning_rate": 8.7798820352473e-06, "loss": 0.5442, "step": 34839 }, { "epoch": 3.707171738667802, "grad_norm": 14.3984375, "learning_rate": 8.779846503695282e-06, "loss": 0.6457, "step": 34840 }, { "epoch": 3.7072781442860183, "grad_norm": 12.984375, "learning_rate": 8.779810972143264e-06, "loss": 0.6277, "step": 34841 }, { "epoch": 3.707384549904235, "grad_norm": 15.1875, "learning_rate": 8.779775440591247e-06, "loss": 0.6438, "step": 34842 }, { "epoch": 3.7074909555224513, "grad_norm": 10.5546875, "learning_rate": 8.779739909039228e-06, "loss": 0.5664, "step": 34843 }, { "epoch": 3.7075973611406683, "grad_norm": 13.4921875, "learning_rate": 8.77970437748721e-06, "loss": 0.6251, "step": 34844 }, { "epoch": 3.707703766758885, "grad_norm": 11.421875, "learning_rate": 8.779668845935192e-06, "loss": 0.5017, "step": 34845 }, { "epoch": 3.7078101723771013, "grad_norm": 14.21875, "learning_rate": 8.779633314383173e-06, "loss": 0.5756, "step": 34846 }, { "epoch": 3.7079165779953183, "grad_norm": 13.234375, "learning_rate": 8.779597782831155e-06, "loss": 0.5995, "step": 34847 }, { "epoch": 3.7080229836135348, "grad_norm": 17.75, "learning_rate": 8.779562251279136e-06, "loss": 0.6639, "step": 34848 }, { "epoch": 3.7081293892317513, "grad_norm": 10.9609375, "learning_rate": 8.779526719727118e-06, "loss": 0.649, "step": 34849 }, { "epoch": 3.7082357948499682, "grad_norm": 13.4921875, "learning_rate": 8.7794911881751e-06, "loss": 0.7392, "step": 34850 }, { "epoch": 3.7083422004681847, "grad_norm": 11.9765625, "learning_rate": 8.779455656623081e-06, "loss": 0.5547, "step": 34851 }, { "epoch": 3.7084486060864013, "grad_norm": 9.3515625, "learning_rate": 8.779420125071065e-06, "loss": 0.5834, "step": 34852 }, { "epoch": 3.708555011704618, "grad_norm": 12.7890625, "learning_rate": 8.779384593519046e-06, "loss": 0.5837, "step": 34853 }, { "epoch": 3.7086614173228347, "grad_norm": 12.7109375, "learning_rate": 8.779349061967028e-06, "loss": 0.6619, "step": 34854 }, { "epoch": 3.7087678229410512, "grad_norm": 15.6640625, "learning_rate": 8.77931353041501e-06, "loss": 0.7901, "step": 34855 }, { "epoch": 3.708874228559268, "grad_norm": 17.796875, "learning_rate": 8.779277998862991e-06, "loss": 0.773, "step": 34856 }, { "epoch": 3.7089806341774847, "grad_norm": 10.875, "learning_rate": 8.779242467310973e-06, "loss": 0.5276, "step": 34857 }, { "epoch": 3.709087039795701, "grad_norm": 11.5859375, "learning_rate": 8.779206935758954e-06, "loss": 0.6547, "step": 34858 }, { "epoch": 3.7091934454139177, "grad_norm": 13.0546875, "learning_rate": 8.779171404206936e-06, "loss": 0.6718, "step": 34859 }, { "epoch": 3.709299851032134, "grad_norm": 11.9453125, "learning_rate": 8.779135872654917e-06, "loss": 0.5723, "step": 34860 }, { "epoch": 3.709406256650351, "grad_norm": 11.65625, "learning_rate": 8.779100341102899e-06, "loss": 0.6468, "step": 34861 }, { "epoch": 3.7095126622685677, "grad_norm": 11.828125, "learning_rate": 8.779064809550882e-06, "loss": 0.7052, "step": 34862 }, { "epoch": 3.709619067886784, "grad_norm": 10.6875, "learning_rate": 8.779029277998864e-06, "loss": 0.5887, "step": 34863 }, { "epoch": 3.709725473505001, "grad_norm": 11.3671875, "learning_rate": 8.778993746446846e-06, "loss": 0.608, "step": 34864 }, { "epoch": 3.7098318791232177, "grad_norm": 12.25, "learning_rate": 8.778958214894827e-06, "loss": 0.5856, "step": 34865 }, { "epoch": 3.709938284741434, "grad_norm": 13.7109375, "learning_rate": 8.778922683342809e-06, "loss": 0.6344, "step": 34866 }, { "epoch": 3.710044690359651, "grad_norm": 19.15625, "learning_rate": 8.778887151790792e-06, "loss": 0.8151, "step": 34867 }, { "epoch": 3.7101510959778676, "grad_norm": 12.15625, "learning_rate": 8.778851620238774e-06, "loss": 0.7412, "step": 34868 }, { "epoch": 3.710257501596084, "grad_norm": 14.15625, "learning_rate": 8.778816088686755e-06, "loss": 0.593, "step": 34869 }, { "epoch": 3.710363907214301, "grad_norm": 12.78125, "learning_rate": 8.778780557134737e-06, "loss": 0.7036, "step": 34870 }, { "epoch": 3.7104703128325176, "grad_norm": 14.0078125, "learning_rate": 8.778745025582719e-06, "loss": 0.6842, "step": 34871 }, { "epoch": 3.710576718450734, "grad_norm": 12.25, "learning_rate": 8.7787094940307e-06, "loss": 0.659, "step": 34872 }, { "epoch": 3.710683124068951, "grad_norm": 12.9765625, "learning_rate": 8.778673962478682e-06, "loss": 0.5694, "step": 34873 }, { "epoch": 3.7107895296871676, "grad_norm": 18.515625, "learning_rate": 8.778638430926663e-06, "loss": 0.6881, "step": 34874 }, { "epoch": 3.710895935305384, "grad_norm": 13.4921875, "learning_rate": 8.778602899374645e-06, "loss": 0.7248, "step": 34875 }, { "epoch": 3.711002340923601, "grad_norm": 20.34375, "learning_rate": 8.778567367822627e-06, "loss": 0.9055, "step": 34876 }, { "epoch": 3.7111087465418175, "grad_norm": 14.671875, "learning_rate": 8.77853183627061e-06, "loss": 0.6389, "step": 34877 }, { "epoch": 3.711215152160034, "grad_norm": 13.484375, "learning_rate": 8.778496304718591e-06, "loss": 0.699, "step": 34878 }, { "epoch": 3.7113215577782506, "grad_norm": 11.546875, "learning_rate": 8.778460773166573e-06, "loss": 0.5485, "step": 34879 }, { "epoch": 3.711427963396467, "grad_norm": 11.609375, "learning_rate": 8.778425241614555e-06, "loss": 0.6241, "step": 34880 }, { "epoch": 3.711534369014684, "grad_norm": 11.40625, "learning_rate": 8.778389710062536e-06, "loss": 0.6453, "step": 34881 }, { "epoch": 3.7116407746329005, "grad_norm": 11.6796875, "learning_rate": 8.778354178510518e-06, "loss": 0.7329, "step": 34882 }, { "epoch": 3.711747180251117, "grad_norm": 12.6875, "learning_rate": 8.7783186469585e-06, "loss": 0.5507, "step": 34883 }, { "epoch": 3.711853585869334, "grad_norm": 13.6328125, "learning_rate": 8.778283115406481e-06, "loss": 0.6456, "step": 34884 }, { "epoch": 3.7119599914875505, "grad_norm": 14.5234375, "learning_rate": 8.778247583854463e-06, "loss": 0.5403, "step": 34885 }, { "epoch": 3.712066397105767, "grad_norm": 12.546875, "learning_rate": 8.778212052302444e-06, "loss": 0.624, "step": 34886 }, { "epoch": 3.712172802723984, "grad_norm": 14.3984375, "learning_rate": 8.778176520750426e-06, "loss": 0.677, "step": 34887 }, { "epoch": 3.7122792083422005, "grad_norm": 12.9921875, "learning_rate": 8.77814098919841e-06, "loss": 0.6735, "step": 34888 }, { "epoch": 3.712385613960417, "grad_norm": 13.015625, "learning_rate": 8.77810545764639e-06, "loss": 0.5247, "step": 34889 }, { "epoch": 3.712492019578634, "grad_norm": 10.5859375, "learning_rate": 8.778069926094372e-06, "loss": 0.5681, "step": 34890 }, { "epoch": 3.7125984251968505, "grad_norm": 14.2890625, "learning_rate": 8.778034394542354e-06, "loss": 0.6463, "step": 34891 }, { "epoch": 3.712704830815067, "grad_norm": 13.7578125, "learning_rate": 8.777998862990336e-06, "loss": 0.6868, "step": 34892 }, { "epoch": 3.712811236433284, "grad_norm": 12.8203125, "learning_rate": 8.777963331438319e-06, "loss": 0.6094, "step": 34893 }, { "epoch": 3.7129176420515004, "grad_norm": 10.015625, "learning_rate": 8.7779277998863e-06, "loss": 0.5239, "step": 34894 }, { "epoch": 3.713024047669717, "grad_norm": 11.59375, "learning_rate": 8.777892268334282e-06, "loss": 0.5125, "step": 34895 }, { "epoch": 3.7131304532879335, "grad_norm": 15.40625, "learning_rate": 8.777856736782264e-06, "loss": 0.6434, "step": 34896 }, { "epoch": 3.7132368589061504, "grad_norm": 12.3203125, "learning_rate": 8.777821205230245e-06, "loss": 0.5479, "step": 34897 }, { "epoch": 3.713343264524367, "grad_norm": 15.6328125, "learning_rate": 8.777785673678227e-06, "loss": 0.7373, "step": 34898 }, { "epoch": 3.7134496701425834, "grad_norm": 12.3203125, "learning_rate": 8.777750142126209e-06, "loss": 0.6476, "step": 34899 }, { "epoch": 3.7135560757608, "grad_norm": 14.828125, "learning_rate": 8.77771461057419e-06, "loss": 0.7195, "step": 34900 }, { "epoch": 3.713662481379017, "grad_norm": 14.6328125, "learning_rate": 8.777679079022172e-06, "loss": 0.6447, "step": 34901 }, { "epoch": 3.7137688869972334, "grad_norm": 12.3125, "learning_rate": 8.777643547470153e-06, "loss": 0.5764, "step": 34902 }, { "epoch": 3.71387529261545, "grad_norm": 13.2421875, "learning_rate": 8.777608015918137e-06, "loss": 0.626, "step": 34903 }, { "epoch": 3.713981698233667, "grad_norm": 12.5078125, "learning_rate": 8.777572484366118e-06, "loss": 0.5783, "step": 34904 }, { "epoch": 3.7140881038518834, "grad_norm": 15.140625, "learning_rate": 8.7775369528141e-06, "loss": 0.6991, "step": 34905 }, { "epoch": 3.7141945094701, "grad_norm": 12.5703125, "learning_rate": 8.777501421262082e-06, "loss": 0.6323, "step": 34906 }, { "epoch": 3.714300915088317, "grad_norm": 14.2109375, "learning_rate": 8.777465889710063e-06, "loss": 0.6073, "step": 34907 }, { "epoch": 3.7144073207065333, "grad_norm": 12.6640625, "learning_rate": 8.777430358158045e-06, "loss": 0.7181, "step": 34908 }, { "epoch": 3.71451372632475, "grad_norm": 14.296875, "learning_rate": 8.777394826606026e-06, "loss": 0.8114, "step": 34909 }, { "epoch": 3.714620131942967, "grad_norm": 12.046875, "learning_rate": 8.777359295054008e-06, "loss": 0.5736, "step": 34910 }, { "epoch": 3.7147265375611833, "grad_norm": 14.546875, "learning_rate": 8.77732376350199e-06, "loss": 0.745, "step": 34911 }, { "epoch": 3.7148329431794, "grad_norm": 14.921875, "learning_rate": 8.777288231949971e-06, "loss": 0.7313, "step": 34912 }, { "epoch": 3.714939348797617, "grad_norm": 12.1953125, "learning_rate": 8.777252700397954e-06, "loss": 0.6333, "step": 34913 }, { "epoch": 3.7150457544158333, "grad_norm": 13.6484375, "learning_rate": 8.777217168845936e-06, "loss": 0.6082, "step": 34914 }, { "epoch": 3.71515216003405, "grad_norm": 14.046875, "learning_rate": 8.777181637293918e-06, "loss": 0.7087, "step": 34915 }, { "epoch": 3.7152585656522663, "grad_norm": 12.0625, "learning_rate": 8.7771461057419e-06, "loss": 0.6476, "step": 34916 }, { "epoch": 3.715364971270483, "grad_norm": 14.65625, "learning_rate": 8.777110574189881e-06, "loss": 0.701, "step": 34917 }, { "epoch": 3.7154713768886998, "grad_norm": 12.6640625, "learning_rate": 8.777075042637862e-06, "loss": 0.7216, "step": 34918 }, { "epoch": 3.7155777825069163, "grad_norm": 24.65625, "learning_rate": 8.777039511085846e-06, "loss": 0.6652, "step": 34919 }, { "epoch": 3.715684188125133, "grad_norm": 15.09375, "learning_rate": 8.777003979533827e-06, "loss": 0.7316, "step": 34920 }, { "epoch": 3.7157905937433497, "grad_norm": 10.203125, "learning_rate": 8.776968447981809e-06, "loss": 0.6215, "step": 34921 }, { "epoch": 3.7158969993615663, "grad_norm": 14.140625, "learning_rate": 8.77693291642979e-06, "loss": 0.6324, "step": 34922 }, { "epoch": 3.7160034049797828, "grad_norm": 9.6171875, "learning_rate": 8.776897384877772e-06, "loss": 0.5328, "step": 34923 }, { "epoch": 3.7161098105979997, "grad_norm": 9.9375, "learning_rate": 8.776861853325754e-06, "loss": 0.5307, "step": 34924 }, { "epoch": 3.7162162162162162, "grad_norm": 12.8046875, "learning_rate": 8.776826321773735e-06, "loss": 0.6907, "step": 34925 }, { "epoch": 3.7163226218344327, "grad_norm": 10.6171875, "learning_rate": 8.776790790221717e-06, "loss": 0.7439, "step": 34926 }, { "epoch": 3.7164290274526497, "grad_norm": 15.1171875, "learning_rate": 8.776755258669699e-06, "loss": 0.7514, "step": 34927 }, { "epoch": 3.716535433070866, "grad_norm": 11.953125, "learning_rate": 8.77671972711768e-06, "loss": 0.6118, "step": 34928 }, { "epoch": 3.7166418386890827, "grad_norm": 16.0625, "learning_rate": 8.776684195565664e-06, "loss": 0.6354, "step": 34929 }, { "epoch": 3.7167482443072997, "grad_norm": 15.8203125, "learning_rate": 8.776648664013645e-06, "loss": 0.7582, "step": 34930 }, { "epoch": 3.716854649925516, "grad_norm": 9.6015625, "learning_rate": 8.776613132461627e-06, "loss": 0.4558, "step": 34931 }, { "epoch": 3.7169610555437327, "grad_norm": 14.90625, "learning_rate": 8.776577600909608e-06, "loss": 0.7127, "step": 34932 }, { "epoch": 3.717067461161949, "grad_norm": 13.8125, "learning_rate": 8.77654206935759e-06, "loss": 0.8877, "step": 34933 }, { "epoch": 3.717173866780166, "grad_norm": 12.9296875, "learning_rate": 8.776506537805573e-06, "loss": 0.7148, "step": 34934 }, { "epoch": 3.7172802723983827, "grad_norm": 14.828125, "learning_rate": 8.776471006253555e-06, "loss": 0.6865, "step": 34935 }, { "epoch": 3.717386678016599, "grad_norm": 13.7421875, "learning_rate": 8.776435474701536e-06, "loss": 0.5909, "step": 34936 }, { "epoch": 3.7174930836348157, "grad_norm": 17.046875, "learning_rate": 8.776399943149518e-06, "loss": 0.7505, "step": 34937 }, { "epoch": 3.7175994892530326, "grad_norm": 16.859375, "learning_rate": 8.776364411597498e-06, "loss": 0.7423, "step": 34938 }, { "epoch": 3.717705894871249, "grad_norm": 12.6484375, "learning_rate": 8.776328880045481e-06, "loss": 0.6122, "step": 34939 }, { "epoch": 3.7178123004894656, "grad_norm": 11.046875, "learning_rate": 8.776293348493463e-06, "loss": 0.5736, "step": 34940 }, { "epoch": 3.7179187061076826, "grad_norm": 19.59375, "learning_rate": 8.776257816941445e-06, "loss": 0.8406, "step": 34941 }, { "epoch": 3.718025111725899, "grad_norm": 14.1953125, "learning_rate": 8.776222285389426e-06, "loss": 0.6872, "step": 34942 }, { "epoch": 3.7181315173441156, "grad_norm": 15.3359375, "learning_rate": 8.776186753837408e-06, "loss": 0.8133, "step": 34943 }, { "epoch": 3.7182379229623326, "grad_norm": 8.578125, "learning_rate": 8.776151222285391e-06, "loss": 0.4092, "step": 34944 }, { "epoch": 3.718344328580549, "grad_norm": 15.328125, "learning_rate": 8.776115690733373e-06, "loss": 0.8277, "step": 34945 }, { "epoch": 3.7184507341987656, "grad_norm": 15.203125, "learning_rate": 8.776080159181354e-06, "loss": 0.7257, "step": 34946 }, { "epoch": 3.7185571398169825, "grad_norm": 12.53125, "learning_rate": 8.776044627629336e-06, "loss": 0.6866, "step": 34947 }, { "epoch": 3.718663545435199, "grad_norm": 14.796875, "learning_rate": 8.776009096077317e-06, "loss": 0.6926, "step": 34948 }, { "epoch": 3.7187699510534156, "grad_norm": 14.125, "learning_rate": 8.775973564525299e-06, "loss": 0.6155, "step": 34949 }, { "epoch": 3.7188763566716325, "grad_norm": 13.421875, "learning_rate": 8.77593803297328e-06, "loss": 0.6897, "step": 34950 }, { "epoch": 3.718982762289849, "grad_norm": 12.9375, "learning_rate": 8.775902501421262e-06, "loss": 0.7257, "step": 34951 }, { "epoch": 3.7190891679080655, "grad_norm": 15.765625, "learning_rate": 8.775866969869244e-06, "loss": 0.6058, "step": 34952 }, { "epoch": 3.719195573526282, "grad_norm": 15.7265625, "learning_rate": 8.775831438317226e-06, "loss": 0.6944, "step": 34953 }, { "epoch": 3.7193019791444986, "grad_norm": 11.6796875, "learning_rate": 8.775795906765209e-06, "loss": 0.7009, "step": 34954 }, { "epoch": 3.7194083847627155, "grad_norm": 9.7421875, "learning_rate": 8.77576037521319e-06, "loss": 0.5895, "step": 34955 }, { "epoch": 3.719514790380932, "grad_norm": 11.796875, "learning_rate": 8.775724843661172e-06, "loss": 0.512, "step": 34956 }, { "epoch": 3.7196211959991485, "grad_norm": 14.0703125, "learning_rate": 8.775689312109154e-06, "loss": 0.7625, "step": 34957 }, { "epoch": 3.7197276016173655, "grad_norm": 12.625, "learning_rate": 8.775653780557135e-06, "loss": 0.638, "step": 34958 }, { "epoch": 3.719834007235582, "grad_norm": 12.71875, "learning_rate": 8.775618249005117e-06, "loss": 0.5985, "step": 34959 }, { "epoch": 3.7199404128537985, "grad_norm": 14.6484375, "learning_rate": 8.7755827174531e-06, "loss": 0.69, "step": 34960 }, { "epoch": 3.7200468184720155, "grad_norm": 12.2734375, "learning_rate": 8.775547185901082e-06, "loss": 0.6728, "step": 34961 }, { "epoch": 3.720153224090232, "grad_norm": 15.390625, "learning_rate": 8.775511654349063e-06, "loss": 0.7875, "step": 34962 }, { "epoch": 3.7202596297084485, "grad_norm": 11.234375, "learning_rate": 8.775476122797045e-06, "loss": 0.6431, "step": 34963 }, { "epoch": 3.7203660353266654, "grad_norm": 13.1484375, "learning_rate": 8.775440591245027e-06, "loss": 0.6455, "step": 34964 }, { "epoch": 3.720472440944882, "grad_norm": 12.2421875, "learning_rate": 8.775405059693008e-06, "loss": 0.631, "step": 34965 }, { "epoch": 3.7205788465630985, "grad_norm": 11.7734375, "learning_rate": 8.77536952814099e-06, "loss": 0.5333, "step": 34966 }, { "epoch": 3.7206852521813154, "grad_norm": 15.8125, "learning_rate": 8.775333996588971e-06, "loss": 0.7618, "step": 34967 }, { "epoch": 3.720791657799532, "grad_norm": 14.1875, "learning_rate": 8.775298465036953e-06, "loss": 0.5159, "step": 34968 }, { "epoch": 3.7208980634177484, "grad_norm": 12.0859375, "learning_rate": 8.775262933484935e-06, "loss": 0.6946, "step": 34969 }, { "epoch": 3.721004469035965, "grad_norm": 14.0234375, "learning_rate": 8.775227401932918e-06, "loss": 0.5867, "step": 34970 }, { "epoch": 3.721110874654182, "grad_norm": 21.1875, "learning_rate": 8.7751918703809e-06, "loss": 0.9017, "step": 34971 }, { "epoch": 3.7212172802723984, "grad_norm": 11.96875, "learning_rate": 8.775156338828881e-06, "loss": 0.607, "step": 34972 }, { "epoch": 3.721323685890615, "grad_norm": 15.3359375, "learning_rate": 8.775120807276863e-06, "loss": 0.7389, "step": 34973 }, { "epoch": 3.7214300915088314, "grad_norm": 10.9375, "learning_rate": 8.775085275724844e-06, "loss": 0.5082, "step": 34974 }, { "epoch": 3.7215364971270484, "grad_norm": 18.734375, "learning_rate": 8.775049744172826e-06, "loss": 0.8894, "step": 34975 }, { "epoch": 3.721642902745265, "grad_norm": 12.984375, "learning_rate": 8.775014212620808e-06, "loss": 0.7483, "step": 34976 }, { "epoch": 3.7217493083634814, "grad_norm": 17.1875, "learning_rate": 8.774978681068789e-06, "loss": 0.658, "step": 34977 }, { "epoch": 3.7218557139816983, "grad_norm": 11.6171875, "learning_rate": 8.77494314951677e-06, "loss": 0.504, "step": 34978 }, { "epoch": 3.721962119599915, "grad_norm": 19.359375, "learning_rate": 8.774907617964752e-06, "loss": 0.6649, "step": 34979 }, { "epoch": 3.7220685252181314, "grad_norm": 9.2265625, "learning_rate": 8.774872086412736e-06, "loss": 0.5823, "step": 34980 }, { "epoch": 3.7221749308363483, "grad_norm": 15.28125, "learning_rate": 8.774836554860717e-06, "loss": 0.7439, "step": 34981 }, { "epoch": 3.722281336454565, "grad_norm": 14.03125, "learning_rate": 8.774801023308699e-06, "loss": 0.5964, "step": 34982 }, { "epoch": 3.7223877420727813, "grad_norm": 11.7265625, "learning_rate": 8.77476549175668e-06, "loss": 0.615, "step": 34983 }, { "epoch": 3.7224941476909983, "grad_norm": 11.0859375, "learning_rate": 8.774729960204662e-06, "loss": 0.732, "step": 34984 }, { "epoch": 3.722600553309215, "grad_norm": 14.734375, "learning_rate": 8.774694428652645e-06, "loss": 0.6893, "step": 34985 }, { "epoch": 3.7227069589274313, "grad_norm": 13.9609375, "learning_rate": 8.774658897100627e-06, "loss": 0.6561, "step": 34986 }, { "epoch": 3.7228133645456483, "grad_norm": 14.7734375, "learning_rate": 8.774623365548609e-06, "loss": 0.6675, "step": 34987 }, { "epoch": 3.7229197701638648, "grad_norm": 15.2578125, "learning_rate": 8.77458783399659e-06, "loss": 0.6964, "step": 34988 }, { "epoch": 3.7230261757820813, "grad_norm": 12.8984375, "learning_rate": 8.774552302444572e-06, "loss": 0.6119, "step": 34989 }, { "epoch": 3.723132581400298, "grad_norm": 9.4921875, "learning_rate": 8.774516770892553e-06, "loss": 0.5476, "step": 34990 }, { "epoch": 3.7232389870185143, "grad_norm": 10.9453125, "learning_rate": 8.774481239340535e-06, "loss": 0.6886, "step": 34991 }, { "epoch": 3.7233453926367313, "grad_norm": 15.4375, "learning_rate": 8.774445707788517e-06, "loss": 0.7387, "step": 34992 }, { "epoch": 3.7234517982549478, "grad_norm": 11.4921875, "learning_rate": 8.774410176236498e-06, "loss": 0.5725, "step": 34993 }, { "epoch": 3.7235582038731643, "grad_norm": 11.6640625, "learning_rate": 8.77437464468448e-06, "loss": 0.5227, "step": 34994 }, { "epoch": 3.7236646094913812, "grad_norm": 14.7421875, "learning_rate": 8.774339113132463e-06, "loss": 0.819, "step": 34995 }, { "epoch": 3.7237710151095977, "grad_norm": 14.421875, "learning_rate": 8.774303581580445e-06, "loss": 0.7205, "step": 34996 }, { "epoch": 3.7238774207278142, "grad_norm": 12.09375, "learning_rate": 8.774268050028426e-06, "loss": 0.6911, "step": 34997 }, { "epoch": 3.723983826346031, "grad_norm": 15.9140625, "learning_rate": 8.774232518476408e-06, "loss": 0.6775, "step": 34998 }, { "epoch": 3.7240902319642477, "grad_norm": 9.8203125, "learning_rate": 8.77419698692439e-06, "loss": 0.5698, "step": 34999 }, { "epoch": 3.724196637582464, "grad_norm": 14.109375, "learning_rate": 8.774161455372371e-06, "loss": 0.6763, "step": 35000 }, { "epoch": 3.724303043200681, "grad_norm": 10.6953125, "learning_rate": 8.774125923820353e-06, "loss": 0.5138, "step": 35001 }, { "epoch": 3.7244094488188977, "grad_norm": 15.265625, "learning_rate": 8.774090392268334e-06, "loss": 0.6611, "step": 35002 }, { "epoch": 3.724515854437114, "grad_norm": 14.03125, "learning_rate": 8.774054860716316e-06, "loss": 0.8087, "step": 35003 }, { "epoch": 3.724622260055331, "grad_norm": 15.65625, "learning_rate": 8.774019329164298e-06, "loss": 0.839, "step": 35004 }, { "epoch": 3.7247286656735477, "grad_norm": 14.015625, "learning_rate": 8.77398379761228e-06, "loss": 0.6251, "step": 35005 }, { "epoch": 3.724835071291764, "grad_norm": 13.6171875, "learning_rate": 8.773948266060263e-06, "loss": 0.6842, "step": 35006 }, { "epoch": 3.724941476909981, "grad_norm": 13.6796875, "learning_rate": 8.773912734508244e-06, "loss": 0.6289, "step": 35007 }, { "epoch": 3.7250478825281976, "grad_norm": 14.7421875, "learning_rate": 8.773877202956226e-06, "loss": 0.7079, "step": 35008 }, { "epoch": 3.725154288146414, "grad_norm": 12.390625, "learning_rate": 8.773841671404207e-06, "loss": 0.641, "step": 35009 }, { "epoch": 3.7252606937646306, "grad_norm": 12.5, "learning_rate": 8.773806139852189e-06, "loss": 0.6604, "step": 35010 }, { "epoch": 3.725367099382847, "grad_norm": 13.3828125, "learning_rate": 8.773770608300172e-06, "loss": 0.6777, "step": 35011 }, { "epoch": 3.725473505001064, "grad_norm": 11.484375, "learning_rate": 8.773735076748154e-06, "loss": 0.5548, "step": 35012 }, { "epoch": 3.7255799106192806, "grad_norm": 13.7109375, "learning_rate": 8.773699545196135e-06, "loss": 0.6159, "step": 35013 }, { "epoch": 3.725686316237497, "grad_norm": 13.1171875, "learning_rate": 8.773664013644117e-06, "loss": 0.5824, "step": 35014 }, { "epoch": 3.725792721855714, "grad_norm": 14.359375, "learning_rate": 8.773628482092099e-06, "loss": 0.7742, "step": 35015 }, { "epoch": 3.7258991274739306, "grad_norm": 12.890625, "learning_rate": 8.77359295054008e-06, "loss": 0.654, "step": 35016 }, { "epoch": 3.726005533092147, "grad_norm": 11.765625, "learning_rate": 8.773557418988062e-06, "loss": 0.6007, "step": 35017 }, { "epoch": 3.726111938710364, "grad_norm": 14.9140625, "learning_rate": 8.773521887436043e-06, "loss": 0.8143, "step": 35018 }, { "epoch": 3.7262183443285806, "grad_norm": 11.5546875, "learning_rate": 8.773486355884025e-06, "loss": 0.5751, "step": 35019 }, { "epoch": 3.726324749946797, "grad_norm": 14.6015625, "learning_rate": 8.773450824332007e-06, "loss": 0.7253, "step": 35020 }, { "epoch": 3.726431155565014, "grad_norm": 13.703125, "learning_rate": 8.77341529277999e-06, "loss": 0.5482, "step": 35021 }, { "epoch": 3.7265375611832305, "grad_norm": 12.9609375, "learning_rate": 8.773379761227972e-06, "loss": 0.6291, "step": 35022 }, { "epoch": 3.726643966801447, "grad_norm": 14.9453125, "learning_rate": 8.773344229675953e-06, "loss": 0.723, "step": 35023 }, { "epoch": 3.726750372419664, "grad_norm": 18.703125, "learning_rate": 8.773308698123935e-06, "loss": 0.9669, "step": 35024 }, { "epoch": 3.7268567780378805, "grad_norm": 10.3671875, "learning_rate": 8.773273166571916e-06, "loss": 0.62, "step": 35025 }, { "epoch": 3.726963183656097, "grad_norm": 15.125, "learning_rate": 8.7732376350199e-06, "loss": 0.721, "step": 35026 }, { "epoch": 3.7270695892743135, "grad_norm": 13.46875, "learning_rate": 8.77320210346788e-06, "loss": 0.6444, "step": 35027 }, { "epoch": 3.7271759948925305, "grad_norm": 12.5546875, "learning_rate": 8.773166571915861e-06, "loss": 0.6452, "step": 35028 }, { "epoch": 3.727282400510747, "grad_norm": 16.015625, "learning_rate": 8.773131040363843e-06, "loss": 0.6396, "step": 35029 }, { "epoch": 3.7273888061289635, "grad_norm": 12.2890625, "learning_rate": 8.773095508811824e-06, "loss": 0.583, "step": 35030 }, { "epoch": 3.72749521174718, "grad_norm": 14.328125, "learning_rate": 8.773059977259808e-06, "loss": 0.6364, "step": 35031 }, { "epoch": 3.727601617365397, "grad_norm": 13.0078125, "learning_rate": 8.77302444570779e-06, "loss": 0.5643, "step": 35032 }, { "epoch": 3.7277080229836135, "grad_norm": 11.65625, "learning_rate": 8.772988914155771e-06, "loss": 0.5371, "step": 35033 }, { "epoch": 3.72781442860183, "grad_norm": 15.84375, "learning_rate": 8.772953382603753e-06, "loss": 0.8023, "step": 35034 }, { "epoch": 3.727920834220047, "grad_norm": 12.859375, "learning_rate": 8.772917851051734e-06, "loss": 0.6067, "step": 35035 }, { "epoch": 3.7280272398382635, "grad_norm": 12.984375, "learning_rate": 8.772882319499716e-06, "loss": 0.7452, "step": 35036 }, { "epoch": 3.72813364545648, "grad_norm": 11.921875, "learning_rate": 8.772846787947699e-06, "loss": 0.6966, "step": 35037 }, { "epoch": 3.728240051074697, "grad_norm": 12.640625, "learning_rate": 8.77281125639568e-06, "loss": 0.7604, "step": 35038 }, { "epoch": 3.7283464566929134, "grad_norm": 14.28125, "learning_rate": 8.772775724843662e-06, "loss": 0.6334, "step": 35039 }, { "epoch": 3.72845286231113, "grad_norm": 14.7734375, "learning_rate": 8.772740193291644e-06, "loss": 0.7792, "step": 35040 }, { "epoch": 3.728559267929347, "grad_norm": 12.5546875, "learning_rate": 8.772704661739626e-06, "loss": 0.565, "step": 35041 }, { "epoch": 3.7286656735475634, "grad_norm": 12.1171875, "learning_rate": 8.772669130187607e-06, "loss": 0.6709, "step": 35042 }, { "epoch": 3.72877207916578, "grad_norm": 11.0078125, "learning_rate": 8.772633598635589e-06, "loss": 0.5266, "step": 35043 }, { "epoch": 3.728878484783997, "grad_norm": 13.71875, "learning_rate": 8.77259806708357e-06, "loss": 0.6077, "step": 35044 }, { "epoch": 3.7289848904022134, "grad_norm": 13.1015625, "learning_rate": 8.772562535531552e-06, "loss": 0.5544, "step": 35045 }, { "epoch": 3.72909129602043, "grad_norm": 14.6328125, "learning_rate": 8.772527003979534e-06, "loss": 0.6603, "step": 35046 }, { "epoch": 3.7291977016386464, "grad_norm": 7.8046875, "learning_rate": 8.772491472427517e-06, "loss": 0.5658, "step": 35047 }, { "epoch": 3.729304107256863, "grad_norm": 13.6484375, "learning_rate": 8.772455940875498e-06, "loss": 0.7068, "step": 35048 }, { "epoch": 3.72941051287508, "grad_norm": 13.328125, "learning_rate": 8.77242040932348e-06, "loss": 0.5585, "step": 35049 }, { "epoch": 3.7295169184932964, "grad_norm": 14.03125, "learning_rate": 8.772384877771462e-06, "loss": 0.7631, "step": 35050 }, { "epoch": 3.729623324111513, "grad_norm": 14.96875, "learning_rate": 8.772349346219443e-06, "loss": 0.6676, "step": 35051 }, { "epoch": 3.72972972972973, "grad_norm": 13.0859375, "learning_rate": 8.772313814667427e-06, "loss": 0.6667, "step": 35052 }, { "epoch": 3.7298361353479463, "grad_norm": 11.3515625, "learning_rate": 8.772278283115408e-06, "loss": 0.6786, "step": 35053 }, { "epoch": 3.729942540966163, "grad_norm": 11.234375, "learning_rate": 8.77224275156339e-06, "loss": 0.63, "step": 35054 }, { "epoch": 3.73004894658438, "grad_norm": 10.828125, "learning_rate": 8.772207220011371e-06, "loss": 0.5618, "step": 35055 }, { "epoch": 3.7301553522025963, "grad_norm": 13.4453125, "learning_rate": 8.772171688459353e-06, "loss": 0.587, "step": 35056 }, { "epoch": 3.730261757820813, "grad_norm": 13.9765625, "learning_rate": 8.772136156907335e-06, "loss": 0.7043, "step": 35057 }, { "epoch": 3.7303681634390298, "grad_norm": 15.078125, "learning_rate": 8.772100625355316e-06, "loss": 0.7405, "step": 35058 }, { "epoch": 3.7304745690572463, "grad_norm": 9.734375, "learning_rate": 8.772065093803298e-06, "loss": 0.4715, "step": 35059 }, { "epoch": 3.730580974675463, "grad_norm": 9.0546875, "learning_rate": 8.77202956225128e-06, "loss": 0.376, "step": 35060 }, { "epoch": 3.7306873802936797, "grad_norm": 14.1953125, "learning_rate": 8.771994030699261e-06, "loss": 0.7389, "step": 35061 }, { "epoch": 3.7307937859118963, "grad_norm": 13.5, "learning_rate": 8.771958499147244e-06, "loss": 0.6445, "step": 35062 }, { "epoch": 3.7309001915301128, "grad_norm": 14.2578125, "learning_rate": 8.771922967595226e-06, "loss": 0.591, "step": 35063 }, { "epoch": 3.7310065971483293, "grad_norm": 13.359375, "learning_rate": 8.771887436043208e-06, "loss": 0.7018, "step": 35064 }, { "epoch": 3.7311130027665462, "grad_norm": 13.8984375, "learning_rate": 8.77185190449119e-06, "loss": 0.6859, "step": 35065 }, { "epoch": 3.7312194083847627, "grad_norm": 19.203125, "learning_rate": 8.77181637293917e-06, "loss": 0.6817, "step": 35066 }, { "epoch": 3.7313258140029792, "grad_norm": 10.1328125, "learning_rate": 8.771780841387152e-06, "loss": 0.5628, "step": 35067 }, { "epoch": 3.7314322196211958, "grad_norm": 18.765625, "learning_rate": 8.771745309835134e-06, "loss": 0.7255, "step": 35068 }, { "epoch": 3.7315386252394127, "grad_norm": 13.8671875, "learning_rate": 8.771709778283116e-06, "loss": 0.7251, "step": 35069 }, { "epoch": 3.731645030857629, "grad_norm": 14.21875, "learning_rate": 8.771674246731097e-06, "loss": 0.5883, "step": 35070 }, { "epoch": 3.7317514364758457, "grad_norm": 12.8515625, "learning_rate": 8.771638715179079e-06, "loss": 0.7045, "step": 35071 }, { "epoch": 3.7318578420940627, "grad_norm": 9.6640625, "learning_rate": 8.771603183627062e-06, "loss": 0.6053, "step": 35072 }, { "epoch": 3.731964247712279, "grad_norm": 12.6171875, "learning_rate": 8.771567652075044e-06, "loss": 0.5121, "step": 35073 }, { "epoch": 3.7320706533304957, "grad_norm": 13.0546875, "learning_rate": 8.771532120523025e-06, "loss": 0.7455, "step": 35074 }, { "epoch": 3.7321770589487127, "grad_norm": 13.4453125, "learning_rate": 8.771496588971007e-06, "loss": 0.6759, "step": 35075 }, { "epoch": 3.732283464566929, "grad_norm": 15.2890625, "learning_rate": 8.771461057418989e-06, "loss": 0.7587, "step": 35076 }, { "epoch": 3.7323898701851457, "grad_norm": 13.59375, "learning_rate": 8.77142552586697e-06, "loss": 0.5532, "step": 35077 }, { "epoch": 3.7324962758033626, "grad_norm": 13.4453125, "learning_rate": 8.771389994314953e-06, "loss": 0.6625, "step": 35078 }, { "epoch": 3.732602681421579, "grad_norm": 10.734375, "learning_rate": 8.771354462762935e-06, "loss": 0.578, "step": 35079 }, { "epoch": 3.7327090870397956, "grad_norm": 17.9375, "learning_rate": 8.771318931210917e-06, "loss": 0.8331, "step": 35080 }, { "epoch": 3.7328154926580126, "grad_norm": 13.1328125, "learning_rate": 8.771283399658898e-06, "loss": 0.7476, "step": 35081 }, { "epoch": 3.732921898276229, "grad_norm": 13.0, "learning_rate": 8.77124786810688e-06, "loss": 0.6998, "step": 35082 }, { "epoch": 3.7330283038944456, "grad_norm": 13.6796875, "learning_rate": 8.771212336554861e-06, "loss": 0.6191, "step": 35083 }, { "epoch": 3.733134709512662, "grad_norm": 13.6484375, "learning_rate": 8.771176805002843e-06, "loss": 0.6983, "step": 35084 }, { "epoch": 3.7332411151308786, "grad_norm": 15.515625, "learning_rate": 8.771141273450825e-06, "loss": 0.6406, "step": 35085 }, { "epoch": 3.7333475207490956, "grad_norm": 16.203125, "learning_rate": 8.771105741898806e-06, "loss": 0.6698, "step": 35086 }, { "epoch": 3.733453926367312, "grad_norm": 12.109375, "learning_rate": 8.771070210346788e-06, "loss": 0.6984, "step": 35087 }, { "epoch": 3.7335603319855286, "grad_norm": 16.75, "learning_rate": 8.771034678794771e-06, "loss": 0.7617, "step": 35088 }, { "epoch": 3.7336667376037456, "grad_norm": 10.40625, "learning_rate": 8.770999147242753e-06, "loss": 0.5637, "step": 35089 }, { "epoch": 3.733773143221962, "grad_norm": 11.96875, "learning_rate": 8.770963615690734e-06, "loss": 0.5543, "step": 35090 }, { "epoch": 3.7338795488401786, "grad_norm": 19.78125, "learning_rate": 8.770928084138716e-06, "loss": 0.7968, "step": 35091 }, { "epoch": 3.7339859544583955, "grad_norm": 14.625, "learning_rate": 8.770892552586698e-06, "loss": 0.789, "step": 35092 }, { "epoch": 3.734092360076612, "grad_norm": 15.9921875, "learning_rate": 8.77085702103468e-06, "loss": 0.7904, "step": 35093 }, { "epoch": 3.7341987656948286, "grad_norm": 12.4296875, "learning_rate": 8.77082148948266e-06, "loss": 0.6032, "step": 35094 }, { "epoch": 3.7343051713130455, "grad_norm": 13.9296875, "learning_rate": 8.770785957930642e-06, "loss": 0.7943, "step": 35095 }, { "epoch": 3.734411576931262, "grad_norm": 9.828125, "learning_rate": 8.770750426378624e-06, "loss": 0.6473, "step": 35096 }, { "epoch": 3.7345179825494785, "grad_norm": 13.7578125, "learning_rate": 8.770714894826606e-06, "loss": 0.6246, "step": 35097 }, { "epoch": 3.7346243881676955, "grad_norm": 9.3203125, "learning_rate": 8.770679363274589e-06, "loss": 0.7427, "step": 35098 }, { "epoch": 3.734730793785912, "grad_norm": 16.5, "learning_rate": 8.77064383172257e-06, "loss": 0.7559, "step": 35099 }, { "epoch": 3.7348371994041285, "grad_norm": 12.5703125, "learning_rate": 8.770608300170552e-06, "loss": 0.6836, "step": 35100 }, { "epoch": 3.734943605022345, "grad_norm": 14.546875, "learning_rate": 8.770572768618534e-06, "loss": 0.8109, "step": 35101 }, { "epoch": 3.735050010640562, "grad_norm": 12.296875, "learning_rate": 8.770537237066515e-06, "loss": 0.6052, "step": 35102 }, { "epoch": 3.7351564162587785, "grad_norm": 13.28125, "learning_rate": 8.770501705514499e-06, "loss": 0.7217, "step": 35103 }, { "epoch": 3.735262821876995, "grad_norm": 11.1640625, "learning_rate": 8.77046617396248e-06, "loss": 0.6522, "step": 35104 }, { "epoch": 3.7353692274952115, "grad_norm": 11.8203125, "learning_rate": 8.770430642410462e-06, "loss": 0.5564, "step": 35105 }, { "epoch": 3.7354756331134285, "grad_norm": 15.828125, "learning_rate": 8.770395110858444e-06, "loss": 0.7411, "step": 35106 }, { "epoch": 3.735582038731645, "grad_norm": 12.8984375, "learning_rate": 8.770359579306425e-06, "loss": 0.4869, "step": 35107 }, { "epoch": 3.7356884443498615, "grad_norm": 12.625, "learning_rate": 8.770324047754407e-06, "loss": 0.6469, "step": 35108 }, { "epoch": 3.7357948499680784, "grad_norm": 11.515625, "learning_rate": 8.770288516202388e-06, "loss": 0.7189, "step": 35109 }, { "epoch": 3.735901255586295, "grad_norm": 11.390625, "learning_rate": 8.77025298465037e-06, "loss": 0.6536, "step": 35110 }, { "epoch": 3.7360076612045114, "grad_norm": 12.1875, "learning_rate": 8.770217453098352e-06, "loss": 0.627, "step": 35111 }, { "epoch": 3.7361140668227284, "grad_norm": 12.515625, "learning_rate": 8.770181921546333e-06, "loss": 0.589, "step": 35112 }, { "epoch": 3.736220472440945, "grad_norm": 15.6328125, "learning_rate": 8.770146389994316e-06, "loss": 0.666, "step": 35113 }, { "epoch": 3.7363268780591614, "grad_norm": 12.1796875, "learning_rate": 8.770110858442298e-06, "loss": 0.6736, "step": 35114 }, { "epoch": 3.7364332836773784, "grad_norm": 15.421875, "learning_rate": 8.77007532689028e-06, "loss": 0.6655, "step": 35115 }, { "epoch": 3.736539689295595, "grad_norm": 15.8671875, "learning_rate": 8.770039795338261e-06, "loss": 0.7556, "step": 35116 }, { "epoch": 3.7366460949138114, "grad_norm": 15.359375, "learning_rate": 8.770004263786243e-06, "loss": 0.7601, "step": 35117 }, { "epoch": 3.7367525005320283, "grad_norm": 13.296875, "learning_rate": 8.769968732234224e-06, "loss": 0.743, "step": 35118 }, { "epoch": 3.736858906150245, "grad_norm": 13.984375, "learning_rate": 8.769933200682206e-06, "loss": 0.6898, "step": 35119 }, { "epoch": 3.7369653117684614, "grad_norm": 10.8984375, "learning_rate": 8.769897669130188e-06, "loss": 0.6405, "step": 35120 }, { "epoch": 3.737071717386678, "grad_norm": 11.4453125, "learning_rate": 8.76986213757817e-06, "loss": 0.6362, "step": 35121 }, { "epoch": 3.7371781230048944, "grad_norm": 15.5625, "learning_rate": 8.769826606026151e-06, "loss": 0.8178, "step": 35122 }, { "epoch": 3.7372845286231113, "grad_norm": 11.875, "learning_rate": 8.769791074474133e-06, "loss": 0.6232, "step": 35123 }, { "epoch": 3.737390934241328, "grad_norm": 14.328125, "learning_rate": 8.769755542922116e-06, "loss": 0.6395, "step": 35124 }, { "epoch": 3.7374973398595444, "grad_norm": 12.9375, "learning_rate": 8.769720011370097e-06, "loss": 0.6744, "step": 35125 }, { "epoch": 3.7376037454777613, "grad_norm": 13.734375, "learning_rate": 8.769684479818079e-06, "loss": 0.5894, "step": 35126 }, { "epoch": 3.737710151095978, "grad_norm": 19.046875, "learning_rate": 8.76964894826606e-06, "loss": 0.8266, "step": 35127 }, { "epoch": 3.7378165567141943, "grad_norm": 17.796875, "learning_rate": 8.769613416714042e-06, "loss": 0.777, "step": 35128 }, { "epoch": 3.7379229623324113, "grad_norm": 15.09375, "learning_rate": 8.769577885162026e-06, "loss": 0.7153, "step": 35129 }, { "epoch": 3.738029367950628, "grad_norm": 16.4375, "learning_rate": 8.769542353610007e-06, "loss": 0.7042, "step": 35130 }, { "epoch": 3.7381357735688443, "grad_norm": 25.5625, "learning_rate": 8.769506822057989e-06, "loss": 0.9052, "step": 35131 }, { "epoch": 3.7382421791870613, "grad_norm": 11.7109375, "learning_rate": 8.76947129050597e-06, "loss": 0.5619, "step": 35132 }, { "epoch": 3.7383485848052778, "grad_norm": 11.9375, "learning_rate": 8.769435758953952e-06, "loss": 0.78, "step": 35133 }, { "epoch": 3.7384549904234943, "grad_norm": 11.71875, "learning_rate": 8.769400227401934e-06, "loss": 0.809, "step": 35134 }, { "epoch": 3.7385613960417112, "grad_norm": 12.8671875, "learning_rate": 8.769364695849915e-06, "loss": 0.5678, "step": 35135 }, { "epoch": 3.7386678016599277, "grad_norm": 14.1796875, "learning_rate": 8.769329164297897e-06, "loss": 0.7546, "step": 35136 }, { "epoch": 3.7387742072781442, "grad_norm": 14.4921875, "learning_rate": 8.769293632745878e-06, "loss": 0.6426, "step": 35137 }, { "epoch": 3.738880612896361, "grad_norm": 14.890625, "learning_rate": 8.76925810119386e-06, "loss": 0.7399, "step": 35138 }, { "epoch": 3.7389870185145777, "grad_norm": 12.1484375, "learning_rate": 8.769222569641843e-06, "loss": 0.6088, "step": 35139 }, { "epoch": 3.739093424132794, "grad_norm": 13.0234375, "learning_rate": 8.769187038089825e-06, "loss": 0.6582, "step": 35140 }, { "epoch": 3.7391998297510107, "grad_norm": 13.265625, "learning_rate": 8.769151506537807e-06, "loss": 0.6586, "step": 35141 }, { "epoch": 3.7393062353692272, "grad_norm": 15.7734375, "learning_rate": 8.769115974985788e-06, "loss": 0.6157, "step": 35142 }, { "epoch": 3.739412640987444, "grad_norm": 12.2109375, "learning_rate": 8.76908044343377e-06, "loss": 0.6283, "step": 35143 }, { "epoch": 3.7395190466056607, "grad_norm": 13.375, "learning_rate": 8.769044911881753e-06, "loss": 0.8238, "step": 35144 }, { "epoch": 3.739625452223877, "grad_norm": 10.59375, "learning_rate": 8.769009380329735e-06, "loss": 0.6372, "step": 35145 }, { "epoch": 3.739731857842094, "grad_norm": 12.7890625, "learning_rate": 8.768973848777716e-06, "loss": 0.7262, "step": 35146 }, { "epoch": 3.7398382634603107, "grad_norm": 11.375, "learning_rate": 8.768938317225696e-06, "loss": 0.7232, "step": 35147 }, { "epoch": 3.739944669078527, "grad_norm": 15.2109375, "learning_rate": 8.768902785673678e-06, "loss": 0.785, "step": 35148 }, { "epoch": 3.740051074696744, "grad_norm": 14.125, "learning_rate": 8.768867254121661e-06, "loss": 0.7407, "step": 35149 }, { "epoch": 3.7401574803149606, "grad_norm": 14.234375, "learning_rate": 8.768831722569643e-06, "loss": 0.6824, "step": 35150 }, { "epoch": 3.740263885933177, "grad_norm": 12.296875, "learning_rate": 8.768796191017624e-06, "loss": 0.7315, "step": 35151 }, { "epoch": 3.740370291551394, "grad_norm": 15.203125, "learning_rate": 8.768760659465606e-06, "loss": 0.8007, "step": 35152 }, { "epoch": 3.7404766971696106, "grad_norm": 10.5390625, "learning_rate": 8.768725127913587e-06, "loss": 0.6846, "step": 35153 }, { "epoch": 3.740583102787827, "grad_norm": 9.875, "learning_rate": 8.768689596361569e-06, "loss": 0.5038, "step": 35154 }, { "epoch": 3.740689508406044, "grad_norm": 11.6796875, "learning_rate": 8.768654064809552e-06, "loss": 0.6533, "step": 35155 }, { "epoch": 3.7407959140242606, "grad_norm": 13.640625, "learning_rate": 8.768618533257534e-06, "loss": 0.6533, "step": 35156 }, { "epoch": 3.740902319642477, "grad_norm": 11.65625, "learning_rate": 8.768583001705516e-06, "loss": 0.5605, "step": 35157 }, { "epoch": 3.7410087252606936, "grad_norm": 15.53125, "learning_rate": 8.768547470153497e-06, "loss": 0.6945, "step": 35158 }, { "epoch": 3.7411151308789106, "grad_norm": 13.5859375, "learning_rate": 8.768511938601479e-06, "loss": 0.7431, "step": 35159 }, { "epoch": 3.741221536497127, "grad_norm": 11.609375, "learning_rate": 8.76847640704946e-06, "loss": 0.5869, "step": 35160 }, { "epoch": 3.7413279421153436, "grad_norm": 13.234375, "learning_rate": 8.768440875497442e-06, "loss": 0.6685, "step": 35161 }, { "epoch": 3.74143434773356, "grad_norm": 10.3828125, "learning_rate": 8.768405343945424e-06, "loss": 0.6686, "step": 35162 }, { "epoch": 3.741540753351777, "grad_norm": 14.6484375, "learning_rate": 8.768369812393405e-06, "loss": 0.6201, "step": 35163 }, { "epoch": 3.7416471589699936, "grad_norm": 14.4375, "learning_rate": 8.768334280841387e-06, "loss": 0.7351, "step": 35164 }, { "epoch": 3.74175356458821, "grad_norm": 10.421875, "learning_rate": 8.76829874928937e-06, "loss": 0.5471, "step": 35165 }, { "epoch": 3.741859970206427, "grad_norm": 12.421875, "learning_rate": 8.768263217737352e-06, "loss": 0.7043, "step": 35166 }, { "epoch": 3.7419663758246435, "grad_norm": 13.6953125, "learning_rate": 8.768227686185333e-06, "loss": 0.7679, "step": 35167 }, { "epoch": 3.74207278144286, "grad_norm": 14.1953125, "learning_rate": 8.768192154633315e-06, "loss": 0.6414, "step": 35168 }, { "epoch": 3.742179187061077, "grad_norm": 12.71875, "learning_rate": 8.768156623081297e-06, "loss": 0.6275, "step": 35169 }, { "epoch": 3.7422855926792935, "grad_norm": 18.109375, "learning_rate": 8.76812109152928e-06, "loss": 0.7384, "step": 35170 }, { "epoch": 3.74239199829751, "grad_norm": 16.25, "learning_rate": 8.768085559977261e-06, "loss": 0.6218, "step": 35171 }, { "epoch": 3.742498403915727, "grad_norm": 15.6171875, "learning_rate": 8.768050028425243e-06, "loss": 0.6938, "step": 35172 }, { "epoch": 3.7426048095339435, "grad_norm": 12.953125, "learning_rate": 8.768014496873225e-06, "loss": 0.677, "step": 35173 }, { "epoch": 3.74271121515216, "grad_norm": 18.28125, "learning_rate": 8.767978965321206e-06, "loss": 0.7552, "step": 35174 }, { "epoch": 3.742817620770377, "grad_norm": 13.03125, "learning_rate": 8.767943433769188e-06, "loss": 0.5017, "step": 35175 }, { "epoch": 3.7429240263885935, "grad_norm": 12.4296875, "learning_rate": 8.76790790221717e-06, "loss": 0.6014, "step": 35176 }, { "epoch": 3.74303043200681, "grad_norm": 13.4375, "learning_rate": 8.767872370665151e-06, "loss": 0.7957, "step": 35177 }, { "epoch": 3.7431368376250265, "grad_norm": 9.3203125, "learning_rate": 8.767836839113133e-06, "loss": 0.5064, "step": 35178 }, { "epoch": 3.743243243243243, "grad_norm": 15.8828125, "learning_rate": 8.767801307561114e-06, "loss": 0.8149, "step": 35179 }, { "epoch": 3.74334964886146, "grad_norm": 13.28125, "learning_rate": 8.767765776009098e-06, "loss": 0.6931, "step": 35180 }, { "epoch": 3.7434560544796764, "grad_norm": 16.0, "learning_rate": 8.76773024445708e-06, "loss": 0.6917, "step": 35181 }, { "epoch": 3.743562460097893, "grad_norm": 11.921875, "learning_rate": 8.767694712905061e-06, "loss": 0.6968, "step": 35182 }, { "epoch": 3.74366886571611, "grad_norm": 17.40625, "learning_rate": 8.767659181353042e-06, "loss": 0.7624, "step": 35183 }, { "epoch": 3.7437752713343264, "grad_norm": 11.59375, "learning_rate": 8.767623649801024e-06, "loss": 0.6832, "step": 35184 }, { "epoch": 3.743881676952543, "grad_norm": 12.125, "learning_rate": 8.767588118249006e-06, "loss": 0.7277, "step": 35185 }, { "epoch": 3.74398808257076, "grad_norm": 11.265625, "learning_rate": 8.767552586696987e-06, "loss": 0.6569, "step": 35186 }, { "epoch": 3.7440944881889764, "grad_norm": 16.0, "learning_rate": 8.767517055144969e-06, "loss": 0.7553, "step": 35187 }, { "epoch": 3.744200893807193, "grad_norm": 14.140625, "learning_rate": 8.76748152359295e-06, "loss": 0.6594, "step": 35188 }, { "epoch": 3.74430729942541, "grad_norm": 10.4140625, "learning_rate": 8.767445992040932e-06, "loss": 0.5569, "step": 35189 }, { "epoch": 3.7444137050436264, "grad_norm": 19.671875, "learning_rate": 8.767410460488915e-06, "loss": 0.8618, "step": 35190 }, { "epoch": 3.744520110661843, "grad_norm": 12.9453125, "learning_rate": 8.767374928936897e-06, "loss": 0.7063, "step": 35191 }, { "epoch": 3.74462651628006, "grad_norm": 11.3046875, "learning_rate": 8.767339397384879e-06, "loss": 0.5111, "step": 35192 }, { "epoch": 3.7447329218982763, "grad_norm": 16.609375, "learning_rate": 8.76730386583286e-06, "loss": 0.7873, "step": 35193 }, { "epoch": 3.744839327516493, "grad_norm": 14.671875, "learning_rate": 8.767268334280842e-06, "loss": 0.6837, "step": 35194 }, { "epoch": 3.7449457331347094, "grad_norm": 17.625, "learning_rate": 8.767232802728823e-06, "loss": 0.809, "step": 35195 }, { "epoch": 3.7450521387529263, "grad_norm": 19.921875, "learning_rate": 8.767197271176807e-06, "loss": 0.8846, "step": 35196 }, { "epoch": 3.745158544371143, "grad_norm": 13.7578125, "learning_rate": 8.767161739624788e-06, "loss": 0.5988, "step": 35197 }, { "epoch": 3.7452649499893593, "grad_norm": 13.453125, "learning_rate": 8.76712620807277e-06, "loss": 0.629, "step": 35198 }, { "epoch": 3.745371355607576, "grad_norm": 14.1640625, "learning_rate": 8.767090676520752e-06, "loss": 0.6505, "step": 35199 }, { "epoch": 3.745477761225793, "grad_norm": 19.015625, "learning_rate": 8.767055144968733e-06, "loss": 0.7644, "step": 35200 }, { "epoch": 3.7455841668440093, "grad_norm": 14.15625, "learning_rate": 8.767019613416715e-06, "loss": 0.6687, "step": 35201 }, { "epoch": 3.745690572462226, "grad_norm": 13.296875, "learning_rate": 8.766984081864696e-06, "loss": 0.6868, "step": 35202 }, { "epoch": 3.7457969780804428, "grad_norm": 13.4609375, "learning_rate": 8.766948550312678e-06, "loss": 0.6597, "step": 35203 }, { "epoch": 3.7459033836986593, "grad_norm": 13.578125, "learning_rate": 8.76691301876066e-06, "loss": 0.7437, "step": 35204 }, { "epoch": 3.746009789316876, "grad_norm": 15.1796875, "learning_rate": 8.766877487208641e-06, "loss": 0.8073, "step": 35205 }, { "epoch": 3.7461161949350927, "grad_norm": 15.09375, "learning_rate": 8.766841955656625e-06, "loss": 0.7167, "step": 35206 }, { "epoch": 3.7462226005533092, "grad_norm": 12.4921875, "learning_rate": 8.766806424104606e-06, "loss": 0.5992, "step": 35207 }, { "epoch": 3.7463290061715258, "grad_norm": 19.953125, "learning_rate": 8.766770892552588e-06, "loss": 0.8021, "step": 35208 }, { "epoch": 3.7464354117897427, "grad_norm": 14.078125, "learning_rate": 8.76673536100057e-06, "loss": 0.6435, "step": 35209 }, { "epoch": 3.746541817407959, "grad_norm": 17.34375, "learning_rate": 8.766699829448551e-06, "loss": 0.7022, "step": 35210 }, { "epoch": 3.7466482230261757, "grad_norm": 12.046875, "learning_rate": 8.766664297896533e-06, "loss": 0.5683, "step": 35211 }, { "epoch": 3.7467546286443927, "grad_norm": 12.0390625, "learning_rate": 8.766628766344514e-06, "loss": 0.5616, "step": 35212 }, { "epoch": 3.746861034262609, "grad_norm": 11.0859375, "learning_rate": 8.766593234792496e-06, "loss": 0.5768, "step": 35213 }, { "epoch": 3.7469674398808257, "grad_norm": 13.21875, "learning_rate": 8.766557703240477e-06, "loss": 0.7461, "step": 35214 }, { "epoch": 3.747073845499042, "grad_norm": 14.9921875, "learning_rate": 8.766522171688459e-06, "loss": 0.6669, "step": 35215 }, { "epoch": 3.7471802511172587, "grad_norm": 12.625, "learning_rate": 8.766486640136442e-06, "loss": 0.7841, "step": 35216 }, { "epoch": 3.7472866567354757, "grad_norm": 12.21875, "learning_rate": 8.766451108584424e-06, "loss": 0.5657, "step": 35217 }, { "epoch": 3.747393062353692, "grad_norm": 19.15625, "learning_rate": 8.766415577032405e-06, "loss": 0.9277, "step": 35218 }, { "epoch": 3.7474994679719087, "grad_norm": 16.359375, "learning_rate": 8.766380045480387e-06, "loss": 0.672, "step": 35219 }, { "epoch": 3.7476058735901256, "grad_norm": 14.65625, "learning_rate": 8.766344513928369e-06, "loss": 0.7256, "step": 35220 }, { "epoch": 3.747712279208342, "grad_norm": 13.2265625, "learning_rate": 8.766308982376352e-06, "loss": 0.782, "step": 35221 }, { "epoch": 3.7478186848265587, "grad_norm": 12.703125, "learning_rate": 8.766273450824334e-06, "loss": 0.5892, "step": 35222 }, { "epoch": 3.7479250904447756, "grad_norm": 14.34375, "learning_rate": 8.766237919272315e-06, "loss": 0.6807, "step": 35223 }, { "epoch": 3.748031496062992, "grad_norm": 20.59375, "learning_rate": 8.766202387720297e-06, "loss": 0.7162, "step": 35224 }, { "epoch": 3.7481379016812086, "grad_norm": 14.546875, "learning_rate": 8.766166856168278e-06, "loss": 0.7469, "step": 35225 }, { "epoch": 3.7482443072994256, "grad_norm": 14.3828125, "learning_rate": 8.76613132461626e-06, "loss": 0.8262, "step": 35226 }, { "epoch": 3.748350712917642, "grad_norm": 11.7734375, "learning_rate": 8.766095793064242e-06, "loss": 0.5836, "step": 35227 }, { "epoch": 3.7484571185358586, "grad_norm": 13.109375, "learning_rate": 8.766060261512223e-06, "loss": 0.7742, "step": 35228 }, { "epoch": 3.7485635241540756, "grad_norm": 16.328125, "learning_rate": 8.766024729960205e-06, "loss": 0.6253, "step": 35229 }, { "epoch": 3.748669929772292, "grad_norm": 11.8125, "learning_rate": 8.765989198408186e-06, "loss": 0.5469, "step": 35230 }, { "epoch": 3.7487763353905086, "grad_norm": 14.140625, "learning_rate": 8.765953666856168e-06, "loss": 0.6015, "step": 35231 }, { "epoch": 3.748882741008725, "grad_norm": 17.96875, "learning_rate": 8.765918135304151e-06, "loss": 0.8564, "step": 35232 }, { "epoch": 3.748989146626942, "grad_norm": 12.625, "learning_rate": 8.765882603752133e-06, "loss": 0.6111, "step": 35233 }, { "epoch": 3.7490955522451586, "grad_norm": 15.6015625, "learning_rate": 8.765847072200115e-06, "loss": 0.6607, "step": 35234 }, { "epoch": 3.749201957863375, "grad_norm": 16.875, "learning_rate": 8.765811540648096e-06, "loss": 0.6917, "step": 35235 }, { "epoch": 3.7493083634815916, "grad_norm": 11.2265625, "learning_rate": 8.765776009096078e-06, "loss": 0.6027, "step": 35236 }, { "epoch": 3.7494147690998085, "grad_norm": 11.328125, "learning_rate": 8.76574047754406e-06, "loss": 0.6384, "step": 35237 }, { "epoch": 3.749521174718025, "grad_norm": 12.0625, "learning_rate": 8.765704945992041e-06, "loss": 0.5857, "step": 35238 }, { "epoch": 3.7496275803362416, "grad_norm": 9.625, "learning_rate": 8.765669414440023e-06, "loss": 0.5291, "step": 35239 }, { "epoch": 3.7497339859544585, "grad_norm": 10.4296875, "learning_rate": 8.765633882888004e-06, "loss": 0.5258, "step": 35240 }, { "epoch": 3.749840391572675, "grad_norm": 11.6875, "learning_rate": 8.765598351335986e-06, "loss": 0.6591, "step": 35241 }, { "epoch": 3.7499467971908915, "grad_norm": 14.4609375, "learning_rate": 8.765562819783969e-06, "loss": 0.7462, "step": 35242 }, { "epoch": 3.7500532028091085, "grad_norm": 13.890625, "learning_rate": 8.76552728823195e-06, "loss": 0.6288, "step": 35243 }, { "epoch": 3.750159608427325, "grad_norm": 11.6953125, "learning_rate": 8.765491756679932e-06, "loss": 0.5551, "step": 35244 }, { "epoch": 3.7502660140455415, "grad_norm": 12.0390625, "learning_rate": 8.765456225127914e-06, "loss": 0.623, "step": 35245 }, { "epoch": 3.7503724196637584, "grad_norm": 10.046875, "learning_rate": 8.765420693575896e-06, "loss": 0.4686, "step": 35246 }, { "epoch": 3.750478825281975, "grad_norm": 12.875, "learning_rate": 8.765385162023879e-06, "loss": 0.6421, "step": 35247 }, { "epoch": 3.7505852309001915, "grad_norm": 15.828125, "learning_rate": 8.76534963047186e-06, "loss": 0.6572, "step": 35248 }, { "epoch": 3.7506916365184084, "grad_norm": 15.1953125, "learning_rate": 8.765314098919842e-06, "loss": 0.7269, "step": 35249 }, { "epoch": 3.750798042136625, "grad_norm": 13.4140625, "learning_rate": 8.765278567367824e-06, "loss": 0.65, "step": 35250 }, { "epoch": 3.7509044477548414, "grad_norm": 15.8125, "learning_rate": 8.765243035815805e-06, "loss": 0.7576, "step": 35251 }, { "epoch": 3.751010853373058, "grad_norm": 15.3515625, "learning_rate": 8.765207504263787e-06, "loss": 0.6645, "step": 35252 }, { "epoch": 3.751117258991275, "grad_norm": 14.3515625, "learning_rate": 8.765171972711768e-06, "loss": 0.6507, "step": 35253 }, { "epoch": 3.7512236646094914, "grad_norm": 20.359375, "learning_rate": 8.76513644115975e-06, "loss": 1.0483, "step": 35254 }, { "epoch": 3.751330070227708, "grad_norm": 12.1953125, "learning_rate": 8.765100909607732e-06, "loss": 0.6066, "step": 35255 }, { "epoch": 3.7514364758459244, "grad_norm": 14.4921875, "learning_rate": 8.765065378055713e-06, "loss": 0.4981, "step": 35256 }, { "epoch": 3.7515428814641414, "grad_norm": 11.8515625, "learning_rate": 8.765029846503697e-06, "loss": 0.7431, "step": 35257 }, { "epoch": 3.751649287082358, "grad_norm": 12.4609375, "learning_rate": 8.764994314951678e-06, "loss": 0.635, "step": 35258 }, { "epoch": 3.7517556927005744, "grad_norm": 14.6015625, "learning_rate": 8.76495878339966e-06, "loss": 0.8719, "step": 35259 }, { "epoch": 3.7518620983187914, "grad_norm": 15.1796875, "learning_rate": 8.764923251847641e-06, "loss": 0.6639, "step": 35260 }, { "epoch": 3.751968503937008, "grad_norm": 14.1171875, "learning_rate": 8.764887720295623e-06, "loss": 0.7419, "step": 35261 }, { "epoch": 3.7520749095552244, "grad_norm": 13.125, "learning_rate": 8.764852188743606e-06, "loss": 0.5721, "step": 35262 }, { "epoch": 3.7521813151734413, "grad_norm": 13.796875, "learning_rate": 8.764816657191588e-06, "loss": 0.7704, "step": 35263 }, { "epoch": 3.752287720791658, "grad_norm": 9.828125, "learning_rate": 8.76478112563957e-06, "loss": 0.571, "step": 35264 }, { "epoch": 3.7523941264098744, "grad_norm": 13.5859375, "learning_rate": 8.764745594087551e-06, "loss": 0.5613, "step": 35265 }, { "epoch": 3.7525005320280913, "grad_norm": 10.8203125, "learning_rate": 8.764710062535533e-06, "loss": 0.5178, "step": 35266 }, { "epoch": 3.752606937646308, "grad_norm": 13.359375, "learning_rate": 8.764674530983514e-06, "loss": 0.6307, "step": 35267 }, { "epoch": 3.7527133432645243, "grad_norm": 13.359375, "learning_rate": 8.764638999431496e-06, "loss": 0.5524, "step": 35268 }, { "epoch": 3.7528197488827413, "grad_norm": 11.484375, "learning_rate": 8.764603467879478e-06, "loss": 0.5112, "step": 35269 }, { "epoch": 3.752926154500958, "grad_norm": 16.390625, "learning_rate": 8.76456793632746e-06, "loss": 0.722, "step": 35270 }, { "epoch": 3.7530325601191743, "grad_norm": 11.78125, "learning_rate": 8.76453240477544e-06, "loss": 0.6498, "step": 35271 }, { "epoch": 3.753138965737391, "grad_norm": 13.0234375, "learning_rate": 8.764496873223422e-06, "loss": 0.6464, "step": 35272 }, { "epoch": 3.7532453713556073, "grad_norm": 11.921875, "learning_rate": 8.764461341671406e-06, "loss": 0.5798, "step": 35273 }, { "epoch": 3.7533517769738243, "grad_norm": 12.90625, "learning_rate": 8.764425810119387e-06, "loss": 0.5487, "step": 35274 }, { "epoch": 3.753458182592041, "grad_norm": 11.6171875, "learning_rate": 8.764390278567369e-06, "loss": 0.6028, "step": 35275 }, { "epoch": 3.7535645882102573, "grad_norm": 16.40625, "learning_rate": 8.76435474701535e-06, "loss": 0.6752, "step": 35276 }, { "epoch": 3.7536709938284742, "grad_norm": 11.8359375, "learning_rate": 8.764319215463332e-06, "loss": 0.5898, "step": 35277 }, { "epoch": 3.7537773994466908, "grad_norm": 12.8984375, "learning_rate": 8.764283683911314e-06, "loss": 0.743, "step": 35278 }, { "epoch": 3.7538838050649073, "grad_norm": 11.8359375, "learning_rate": 8.764248152359295e-06, "loss": 0.7257, "step": 35279 }, { "epoch": 3.753990210683124, "grad_norm": 10.3671875, "learning_rate": 8.764212620807277e-06, "loss": 0.4831, "step": 35280 }, { "epoch": 3.7540966163013407, "grad_norm": 11.296875, "learning_rate": 8.764177089255259e-06, "loss": 0.5024, "step": 35281 }, { "epoch": 3.7542030219195572, "grad_norm": 11.671875, "learning_rate": 8.76414155770324e-06, "loss": 0.501, "step": 35282 }, { "epoch": 3.754309427537774, "grad_norm": 14.328125, "learning_rate": 8.764106026151223e-06, "loss": 0.6651, "step": 35283 }, { "epoch": 3.7544158331559907, "grad_norm": 11.3203125, "learning_rate": 8.764070494599205e-06, "loss": 0.5059, "step": 35284 }, { "epoch": 3.754522238774207, "grad_norm": 13.234375, "learning_rate": 8.764034963047187e-06, "loss": 0.6859, "step": 35285 }, { "epoch": 3.754628644392424, "grad_norm": 17.28125, "learning_rate": 8.763999431495168e-06, "loss": 0.7804, "step": 35286 }, { "epoch": 3.7547350500106407, "grad_norm": 14.234375, "learning_rate": 8.76396389994315e-06, "loss": 0.6086, "step": 35287 }, { "epoch": 3.754841455628857, "grad_norm": 12.4765625, "learning_rate": 8.763928368391133e-06, "loss": 0.5801, "step": 35288 }, { "epoch": 3.7549478612470737, "grad_norm": 14.6484375, "learning_rate": 8.763892836839115e-06, "loss": 0.759, "step": 35289 }, { "epoch": 3.7550542668652906, "grad_norm": 13.9921875, "learning_rate": 8.763857305287096e-06, "loss": 0.7122, "step": 35290 }, { "epoch": 3.755160672483507, "grad_norm": 14.75, "learning_rate": 8.763821773735078e-06, "loss": 0.8018, "step": 35291 }, { "epoch": 3.7552670781017237, "grad_norm": 13.578125, "learning_rate": 8.76378624218306e-06, "loss": 0.5257, "step": 35292 }, { "epoch": 3.75537348371994, "grad_norm": 15.1875, "learning_rate": 8.763750710631041e-06, "loss": 0.6855, "step": 35293 }, { "epoch": 3.755479889338157, "grad_norm": 12.6875, "learning_rate": 8.763715179079023e-06, "loss": 0.5543, "step": 35294 }, { "epoch": 3.7555862949563736, "grad_norm": 10.9765625, "learning_rate": 8.763679647527004e-06, "loss": 0.5522, "step": 35295 }, { "epoch": 3.75569270057459, "grad_norm": 10.0625, "learning_rate": 8.763644115974986e-06, "loss": 0.5344, "step": 35296 }, { "epoch": 3.755799106192807, "grad_norm": 12.015625, "learning_rate": 8.763608584422968e-06, "loss": 0.6295, "step": 35297 }, { "epoch": 3.7559055118110236, "grad_norm": 19.28125, "learning_rate": 8.763573052870951e-06, "loss": 0.7946, "step": 35298 }, { "epoch": 3.75601191742924, "grad_norm": 12.3828125, "learning_rate": 8.763537521318933e-06, "loss": 0.5841, "step": 35299 }, { "epoch": 3.756118323047457, "grad_norm": 14.7109375, "learning_rate": 8.763501989766914e-06, "loss": 0.6921, "step": 35300 }, { "epoch": 3.7562247286656736, "grad_norm": 22.4375, "learning_rate": 8.763466458214896e-06, "loss": 0.706, "step": 35301 }, { "epoch": 3.75633113428389, "grad_norm": 18.09375, "learning_rate": 8.763430926662877e-06, "loss": 0.8374, "step": 35302 }, { "epoch": 3.756437539902107, "grad_norm": 14.828125, "learning_rate": 8.763395395110859e-06, "loss": 0.6394, "step": 35303 }, { "epoch": 3.7565439455203236, "grad_norm": 15.0078125, "learning_rate": 8.76335986355884e-06, "loss": 0.6645, "step": 35304 }, { "epoch": 3.75665035113854, "grad_norm": 11.671875, "learning_rate": 8.763324332006822e-06, "loss": 0.7006, "step": 35305 }, { "epoch": 3.756756756756757, "grad_norm": 16.765625, "learning_rate": 8.763288800454804e-06, "loss": 0.9507, "step": 35306 }, { "epoch": 3.7568631623749735, "grad_norm": 13.9296875, "learning_rate": 8.763253268902785e-06, "loss": 0.7627, "step": 35307 }, { "epoch": 3.75696956799319, "grad_norm": 15.90625, "learning_rate": 8.763217737350769e-06, "loss": 0.729, "step": 35308 }, { "epoch": 3.7570759736114065, "grad_norm": 13.5625, "learning_rate": 8.76318220579875e-06, "loss": 0.6277, "step": 35309 }, { "epoch": 3.757182379229623, "grad_norm": 11.2890625, "learning_rate": 8.763146674246732e-06, "loss": 0.7718, "step": 35310 }, { "epoch": 3.75728878484784, "grad_norm": 11.984375, "learning_rate": 8.763111142694714e-06, "loss": 0.5541, "step": 35311 }, { "epoch": 3.7573951904660565, "grad_norm": 13.3046875, "learning_rate": 8.763075611142695e-06, "loss": 0.677, "step": 35312 }, { "epoch": 3.757501596084273, "grad_norm": 14.78125, "learning_rate": 8.763040079590677e-06, "loss": 0.6835, "step": 35313 }, { "epoch": 3.75760800170249, "grad_norm": 12.9765625, "learning_rate": 8.76300454803866e-06, "loss": 0.5699, "step": 35314 }, { "epoch": 3.7577144073207065, "grad_norm": 10.96875, "learning_rate": 8.762969016486642e-06, "loss": 0.6574, "step": 35315 }, { "epoch": 3.757820812938923, "grad_norm": 11.59375, "learning_rate": 8.762933484934623e-06, "loss": 0.6295, "step": 35316 }, { "epoch": 3.75792721855714, "grad_norm": 12.3359375, "learning_rate": 8.762897953382605e-06, "loss": 0.594, "step": 35317 }, { "epoch": 3.7580336241753565, "grad_norm": 11.828125, "learning_rate": 8.762862421830586e-06, "loss": 0.627, "step": 35318 }, { "epoch": 3.758140029793573, "grad_norm": 11.1328125, "learning_rate": 8.762826890278568e-06, "loss": 0.4539, "step": 35319 }, { "epoch": 3.75824643541179, "grad_norm": 13.046875, "learning_rate": 8.76279135872655e-06, "loss": 0.6046, "step": 35320 }, { "epoch": 3.7583528410300064, "grad_norm": 10.8671875, "learning_rate": 8.762755827174531e-06, "loss": 0.5674, "step": 35321 }, { "epoch": 3.758459246648223, "grad_norm": 16.15625, "learning_rate": 8.762720295622513e-06, "loss": 0.8159, "step": 35322 }, { "epoch": 3.75856565226644, "grad_norm": 10.5078125, "learning_rate": 8.762684764070495e-06, "loss": 0.556, "step": 35323 }, { "epoch": 3.7586720578846564, "grad_norm": 13.984375, "learning_rate": 8.762649232518478e-06, "loss": 0.6353, "step": 35324 }, { "epoch": 3.758778463502873, "grad_norm": 15.3984375, "learning_rate": 8.76261370096646e-06, "loss": 0.6636, "step": 35325 }, { "epoch": 3.7588848691210894, "grad_norm": 13.2265625, "learning_rate": 8.762578169414441e-06, "loss": 0.8011, "step": 35326 }, { "epoch": 3.7589912747393064, "grad_norm": 13.15625, "learning_rate": 8.762542637862423e-06, "loss": 0.6705, "step": 35327 }, { "epoch": 3.759097680357523, "grad_norm": 12.078125, "learning_rate": 8.762507106310404e-06, "loss": 0.6726, "step": 35328 }, { "epoch": 3.7592040859757394, "grad_norm": 12.90625, "learning_rate": 8.762471574758386e-06, "loss": 0.7196, "step": 35329 }, { "epoch": 3.759310491593956, "grad_norm": 11.2265625, "learning_rate": 8.762436043206367e-06, "loss": 0.6344, "step": 35330 }, { "epoch": 3.759416897212173, "grad_norm": 13.5703125, "learning_rate": 8.762400511654349e-06, "loss": 0.6453, "step": 35331 }, { "epoch": 3.7595233028303894, "grad_norm": 16.5625, "learning_rate": 8.76236498010233e-06, "loss": 0.7914, "step": 35332 }, { "epoch": 3.759629708448606, "grad_norm": 12.453125, "learning_rate": 8.762329448550312e-06, "loss": 0.7053, "step": 35333 }, { "epoch": 3.759736114066823, "grad_norm": 18.09375, "learning_rate": 8.762293916998296e-06, "loss": 0.7489, "step": 35334 }, { "epoch": 3.7598425196850394, "grad_norm": 13.3515625, "learning_rate": 8.762258385446277e-06, "loss": 0.5777, "step": 35335 }, { "epoch": 3.759948925303256, "grad_norm": 13.0234375, "learning_rate": 8.762222853894259e-06, "loss": 0.6026, "step": 35336 }, { "epoch": 3.760055330921473, "grad_norm": 19.703125, "learning_rate": 8.76218732234224e-06, "loss": 0.763, "step": 35337 }, { "epoch": 3.7601617365396893, "grad_norm": 13.671875, "learning_rate": 8.762151790790222e-06, "loss": 0.6787, "step": 35338 }, { "epoch": 3.760268142157906, "grad_norm": 11.234375, "learning_rate": 8.762116259238205e-06, "loss": 0.5962, "step": 35339 }, { "epoch": 3.760374547776123, "grad_norm": 20.921875, "learning_rate": 8.762080727686187e-06, "loss": 0.9379, "step": 35340 }, { "epoch": 3.7604809533943393, "grad_norm": 13.7109375, "learning_rate": 8.762045196134169e-06, "loss": 0.688, "step": 35341 }, { "epoch": 3.760587359012556, "grad_norm": 12.4375, "learning_rate": 8.76200966458215e-06, "loss": 0.6842, "step": 35342 }, { "epoch": 3.7606937646307728, "grad_norm": 9.0234375, "learning_rate": 8.761974133030132e-06, "loss": 0.6447, "step": 35343 }, { "epoch": 3.7608001702489893, "grad_norm": 11.9375, "learning_rate": 8.761938601478113e-06, "loss": 0.6047, "step": 35344 }, { "epoch": 3.760906575867206, "grad_norm": 17.875, "learning_rate": 8.761903069926095e-06, "loss": 0.6379, "step": 35345 }, { "epoch": 3.7610129814854223, "grad_norm": 11.09375, "learning_rate": 8.761867538374077e-06, "loss": 0.4428, "step": 35346 }, { "epoch": 3.761119387103639, "grad_norm": 12.046875, "learning_rate": 8.761832006822058e-06, "loss": 0.7627, "step": 35347 }, { "epoch": 3.7612257927218558, "grad_norm": 14.78125, "learning_rate": 8.76179647527004e-06, "loss": 0.7205, "step": 35348 }, { "epoch": 3.7613321983400723, "grad_norm": 14.9609375, "learning_rate": 8.761760943718021e-06, "loss": 0.8599, "step": 35349 }, { "epoch": 3.7614386039582888, "grad_norm": 12.8515625, "learning_rate": 8.761725412166005e-06, "loss": 0.5653, "step": 35350 }, { "epoch": 3.7615450095765057, "grad_norm": 14.1015625, "learning_rate": 8.761689880613986e-06, "loss": 0.6431, "step": 35351 }, { "epoch": 3.7616514151947222, "grad_norm": 12.1328125, "learning_rate": 8.761654349061968e-06, "loss": 0.6717, "step": 35352 }, { "epoch": 3.7617578208129387, "grad_norm": 12.1796875, "learning_rate": 8.76161881750995e-06, "loss": 0.5821, "step": 35353 }, { "epoch": 3.7618642264311557, "grad_norm": 11.3359375, "learning_rate": 8.761583285957931e-06, "loss": 0.6548, "step": 35354 }, { "epoch": 3.761970632049372, "grad_norm": 14.4921875, "learning_rate": 8.761547754405913e-06, "loss": 0.7016, "step": 35355 }, { "epoch": 3.7620770376675887, "grad_norm": 12.5234375, "learning_rate": 8.761512222853894e-06, "loss": 0.7468, "step": 35356 }, { "epoch": 3.7621834432858057, "grad_norm": 14.8125, "learning_rate": 8.761476691301876e-06, "loss": 0.6611, "step": 35357 }, { "epoch": 3.762289848904022, "grad_norm": 14.59375, "learning_rate": 8.761441159749858e-06, "loss": 0.7515, "step": 35358 }, { "epoch": 3.7623962545222387, "grad_norm": 12.1484375, "learning_rate": 8.761405628197839e-06, "loss": 0.5963, "step": 35359 }, { "epoch": 3.7625026601404556, "grad_norm": 16.015625, "learning_rate": 8.761370096645822e-06, "loss": 0.7827, "step": 35360 }, { "epoch": 3.762609065758672, "grad_norm": 9.9375, "learning_rate": 8.761334565093804e-06, "loss": 0.4572, "step": 35361 }, { "epoch": 3.7627154713768887, "grad_norm": 13.625, "learning_rate": 8.761299033541786e-06, "loss": 0.654, "step": 35362 }, { "epoch": 3.7628218769951056, "grad_norm": 10.9921875, "learning_rate": 8.761263501989767e-06, "loss": 0.7034, "step": 35363 }, { "epoch": 3.762928282613322, "grad_norm": 11.078125, "learning_rate": 8.761227970437749e-06, "loss": 0.5785, "step": 35364 }, { "epoch": 3.7630346882315386, "grad_norm": 12.7890625, "learning_rate": 8.761192438885732e-06, "loss": 0.6583, "step": 35365 }, { "epoch": 3.763141093849755, "grad_norm": 11.7890625, "learning_rate": 8.761156907333714e-06, "loss": 0.6342, "step": 35366 }, { "epoch": 3.7632474994679717, "grad_norm": 12.15625, "learning_rate": 8.761121375781695e-06, "loss": 0.7642, "step": 35367 }, { "epoch": 3.7633539050861886, "grad_norm": 12.8671875, "learning_rate": 8.761085844229677e-06, "loss": 0.7405, "step": 35368 }, { "epoch": 3.763460310704405, "grad_norm": 13.765625, "learning_rate": 8.761050312677659e-06, "loss": 0.6106, "step": 35369 }, { "epoch": 3.7635667163226216, "grad_norm": 13.328125, "learning_rate": 8.76101478112564e-06, "loss": 0.6359, "step": 35370 }, { "epoch": 3.7636731219408386, "grad_norm": 16.265625, "learning_rate": 8.760979249573622e-06, "loss": 0.6939, "step": 35371 }, { "epoch": 3.763779527559055, "grad_norm": 11.328125, "learning_rate": 8.760943718021603e-06, "loss": 0.6198, "step": 35372 }, { "epoch": 3.7638859331772716, "grad_norm": 13.109375, "learning_rate": 8.760908186469585e-06, "loss": 0.6479, "step": 35373 }, { "epoch": 3.7639923387954886, "grad_norm": 14.0390625, "learning_rate": 8.760872654917567e-06, "loss": 0.8768, "step": 35374 }, { "epoch": 3.764098744413705, "grad_norm": 11.7421875, "learning_rate": 8.76083712336555e-06, "loss": 0.6382, "step": 35375 }, { "epoch": 3.7642051500319216, "grad_norm": 11.015625, "learning_rate": 8.760801591813532e-06, "loss": 0.4944, "step": 35376 }, { "epoch": 3.7643115556501385, "grad_norm": 13.53125, "learning_rate": 8.760766060261513e-06, "loss": 0.6476, "step": 35377 }, { "epoch": 3.764417961268355, "grad_norm": 14.5078125, "learning_rate": 8.760730528709495e-06, "loss": 0.6597, "step": 35378 }, { "epoch": 3.7645243668865715, "grad_norm": 11.921875, "learning_rate": 8.760694997157476e-06, "loss": 0.6225, "step": 35379 }, { "epoch": 3.7646307725047885, "grad_norm": 13.7734375, "learning_rate": 8.76065946560546e-06, "loss": 0.7146, "step": 35380 }, { "epoch": 3.764737178123005, "grad_norm": 14.921875, "learning_rate": 8.760623934053441e-06, "loss": 0.6692, "step": 35381 }, { "epoch": 3.7648435837412215, "grad_norm": 10.875, "learning_rate": 8.760588402501423e-06, "loss": 0.552, "step": 35382 }, { "epoch": 3.764949989359438, "grad_norm": 13.796875, "learning_rate": 8.760552870949404e-06, "loss": 0.6112, "step": 35383 }, { "epoch": 3.765056394977655, "grad_norm": 14.203125, "learning_rate": 8.760517339397386e-06, "loss": 0.8128, "step": 35384 }, { "epoch": 3.7651628005958715, "grad_norm": 12.96875, "learning_rate": 8.760481807845368e-06, "loss": 0.7247, "step": 35385 }, { "epoch": 3.765269206214088, "grad_norm": 13.8828125, "learning_rate": 8.76044627629335e-06, "loss": 0.739, "step": 35386 }, { "epoch": 3.7653756118323045, "grad_norm": 13.5703125, "learning_rate": 8.760410744741331e-06, "loss": 0.6758, "step": 35387 }, { "epoch": 3.7654820174505215, "grad_norm": 12.5859375, "learning_rate": 8.760375213189312e-06, "loss": 0.6435, "step": 35388 }, { "epoch": 3.765588423068738, "grad_norm": 12.8359375, "learning_rate": 8.760339681637294e-06, "loss": 0.818, "step": 35389 }, { "epoch": 3.7656948286869545, "grad_norm": 13.7421875, "learning_rate": 8.760304150085276e-06, "loss": 0.5624, "step": 35390 }, { "epoch": 3.7658012343051714, "grad_norm": 9.8125, "learning_rate": 8.760268618533259e-06, "loss": 0.6438, "step": 35391 }, { "epoch": 3.765907639923388, "grad_norm": 16.015625, "learning_rate": 8.76023308698124e-06, "loss": 0.6327, "step": 35392 }, { "epoch": 3.7660140455416045, "grad_norm": 13.2734375, "learning_rate": 8.760197555429222e-06, "loss": 0.62, "step": 35393 }, { "epoch": 3.7661204511598214, "grad_norm": 15.171875, "learning_rate": 8.760162023877204e-06, "loss": 0.5966, "step": 35394 }, { "epoch": 3.766226856778038, "grad_norm": 14.9296875, "learning_rate": 8.760126492325185e-06, "loss": 0.6622, "step": 35395 }, { "epoch": 3.7663332623962544, "grad_norm": 10.9453125, "learning_rate": 8.760090960773167e-06, "loss": 0.6162, "step": 35396 }, { "epoch": 3.7664396680144714, "grad_norm": 16.734375, "learning_rate": 8.760055429221149e-06, "loss": 0.7474, "step": 35397 }, { "epoch": 3.766546073632688, "grad_norm": 12.2734375, "learning_rate": 8.76001989766913e-06, "loss": 0.7183, "step": 35398 }, { "epoch": 3.7666524792509044, "grad_norm": 10.8671875, "learning_rate": 8.759984366117112e-06, "loss": 0.5619, "step": 35399 }, { "epoch": 3.7667588848691214, "grad_norm": 14.3359375, "learning_rate": 8.759948834565093e-06, "loss": 0.6076, "step": 35400 }, { "epoch": 3.766865290487338, "grad_norm": 15.09375, "learning_rate": 8.759913303013077e-06, "loss": 0.746, "step": 35401 }, { "epoch": 3.7669716961055544, "grad_norm": 10.90625, "learning_rate": 8.759877771461058e-06, "loss": 0.7215, "step": 35402 }, { "epoch": 3.767078101723771, "grad_norm": 12.296875, "learning_rate": 8.75984223990904e-06, "loss": 0.6395, "step": 35403 }, { "epoch": 3.7671845073419874, "grad_norm": 12.4765625, "learning_rate": 8.759806708357022e-06, "loss": 0.6042, "step": 35404 }, { "epoch": 3.7672909129602044, "grad_norm": 12.6796875, "learning_rate": 8.759771176805003e-06, "loss": 0.7219, "step": 35405 }, { "epoch": 3.767397318578421, "grad_norm": 13.3515625, "learning_rate": 8.759735645252986e-06, "loss": 0.7594, "step": 35406 }, { "epoch": 3.7675037241966374, "grad_norm": 16.375, "learning_rate": 8.759700113700968e-06, "loss": 0.7711, "step": 35407 }, { "epoch": 3.7676101298148543, "grad_norm": 13.5078125, "learning_rate": 8.75966458214895e-06, "loss": 0.6176, "step": 35408 }, { "epoch": 3.767716535433071, "grad_norm": 22.53125, "learning_rate": 8.759629050596931e-06, "loss": 0.987, "step": 35409 }, { "epoch": 3.7678229410512873, "grad_norm": 14.078125, "learning_rate": 8.759593519044913e-06, "loss": 0.8021, "step": 35410 }, { "epoch": 3.7679293466695043, "grad_norm": 16.609375, "learning_rate": 8.759557987492895e-06, "loss": 0.7891, "step": 35411 }, { "epoch": 3.768035752287721, "grad_norm": 15.828125, "learning_rate": 8.759522455940876e-06, "loss": 0.8117, "step": 35412 }, { "epoch": 3.7681421579059373, "grad_norm": 10.4609375, "learning_rate": 8.759486924388858e-06, "loss": 0.4617, "step": 35413 }, { "epoch": 3.7682485635241543, "grad_norm": 11.7109375, "learning_rate": 8.75945139283684e-06, "loss": 0.675, "step": 35414 }, { "epoch": 3.768354969142371, "grad_norm": 12.6796875, "learning_rate": 8.759415861284821e-06, "loss": 0.5877, "step": 35415 }, { "epoch": 3.7684613747605873, "grad_norm": 13.8046875, "learning_rate": 8.759380329732804e-06, "loss": 0.8916, "step": 35416 }, { "epoch": 3.7685677803788042, "grad_norm": 16.515625, "learning_rate": 8.759344798180786e-06, "loss": 0.7781, "step": 35417 }, { "epoch": 3.7686741859970208, "grad_norm": 12.2578125, "learning_rate": 8.759309266628767e-06, "loss": 0.5762, "step": 35418 }, { "epoch": 3.7687805916152373, "grad_norm": 12.9921875, "learning_rate": 8.759273735076749e-06, "loss": 0.7291, "step": 35419 }, { "epoch": 3.7688869972334538, "grad_norm": 13.3828125, "learning_rate": 8.75923820352473e-06, "loss": 0.6492, "step": 35420 }, { "epoch": 3.7689934028516707, "grad_norm": 13.9453125, "learning_rate": 8.759202671972712e-06, "loss": 0.6743, "step": 35421 }, { "epoch": 3.7690998084698872, "grad_norm": 11.9375, "learning_rate": 8.759167140420694e-06, "loss": 0.6563, "step": 35422 }, { "epoch": 3.7692062140881037, "grad_norm": 13.84375, "learning_rate": 8.759131608868675e-06, "loss": 0.7049, "step": 35423 }, { "epoch": 3.7693126197063203, "grad_norm": 11.7421875, "learning_rate": 8.759096077316657e-06, "loss": 0.6954, "step": 35424 }, { "epoch": 3.769419025324537, "grad_norm": 14.4296875, "learning_rate": 8.759060545764639e-06, "loss": 0.6408, "step": 35425 }, { "epoch": 3.7695254309427537, "grad_norm": 11.875, "learning_rate": 8.75902501421262e-06, "loss": 0.6963, "step": 35426 }, { "epoch": 3.7696318365609702, "grad_norm": 12.2109375, "learning_rate": 8.758989482660604e-06, "loss": 0.6322, "step": 35427 }, { "epoch": 3.769738242179187, "grad_norm": 12.515625, "learning_rate": 8.758953951108585e-06, "loss": 0.6436, "step": 35428 }, { "epoch": 3.7698446477974037, "grad_norm": 11.4375, "learning_rate": 8.758918419556567e-06, "loss": 0.671, "step": 35429 }, { "epoch": 3.76995105341562, "grad_norm": 12.1640625, "learning_rate": 8.758882888004548e-06, "loss": 0.521, "step": 35430 }, { "epoch": 3.770057459033837, "grad_norm": 13.03125, "learning_rate": 8.75884735645253e-06, "loss": 0.5958, "step": 35431 }, { "epoch": 3.7701638646520537, "grad_norm": 12.9296875, "learning_rate": 8.758811824900513e-06, "loss": 0.6388, "step": 35432 }, { "epoch": 3.77027027027027, "grad_norm": 11.984375, "learning_rate": 8.758776293348495e-06, "loss": 0.5824, "step": 35433 }, { "epoch": 3.770376675888487, "grad_norm": 12.4765625, "learning_rate": 8.758740761796477e-06, "loss": 0.7062, "step": 35434 }, { "epoch": 3.7704830815067036, "grad_norm": 10.453125, "learning_rate": 8.758705230244458e-06, "loss": 0.464, "step": 35435 }, { "epoch": 3.77058948712492, "grad_norm": 11.78125, "learning_rate": 8.75866969869244e-06, "loss": 0.6057, "step": 35436 }, { "epoch": 3.770695892743137, "grad_norm": 12.6640625, "learning_rate": 8.758634167140421e-06, "loss": 0.61, "step": 35437 }, { "epoch": 3.7708022983613536, "grad_norm": 21.46875, "learning_rate": 8.758598635588403e-06, "loss": 0.7644, "step": 35438 }, { "epoch": 3.77090870397957, "grad_norm": 10.4140625, "learning_rate": 8.758563104036385e-06, "loss": 0.6637, "step": 35439 }, { "epoch": 3.7710151095977866, "grad_norm": 12.375, "learning_rate": 8.758527572484366e-06, "loss": 0.6425, "step": 35440 }, { "epoch": 3.771121515216003, "grad_norm": 14.4609375, "learning_rate": 8.758492040932348e-06, "loss": 0.7417, "step": 35441 }, { "epoch": 3.77122792083422, "grad_norm": 13.4921875, "learning_rate": 8.758456509380331e-06, "loss": 0.5706, "step": 35442 }, { "epoch": 3.7713343264524366, "grad_norm": 10.328125, "learning_rate": 8.758420977828313e-06, "loss": 0.5229, "step": 35443 }, { "epoch": 3.771440732070653, "grad_norm": 11.1640625, "learning_rate": 8.758385446276294e-06, "loss": 0.5536, "step": 35444 }, { "epoch": 3.77154713768887, "grad_norm": 14.7421875, "learning_rate": 8.758349914724276e-06, "loss": 0.6409, "step": 35445 }, { "epoch": 3.7716535433070866, "grad_norm": 12.6796875, "learning_rate": 8.758314383172258e-06, "loss": 0.7417, "step": 35446 }, { "epoch": 3.771759948925303, "grad_norm": 12.03125, "learning_rate": 8.758278851620239e-06, "loss": 0.5847, "step": 35447 }, { "epoch": 3.77186635454352, "grad_norm": 11.28125, "learning_rate": 8.75824332006822e-06, "loss": 0.6811, "step": 35448 }, { "epoch": 3.7719727601617365, "grad_norm": 12.75, "learning_rate": 8.758207788516202e-06, "loss": 0.6639, "step": 35449 }, { "epoch": 3.772079165779953, "grad_norm": 16.828125, "learning_rate": 8.758172256964184e-06, "loss": 0.6514, "step": 35450 }, { "epoch": 3.77218557139817, "grad_norm": 14.0078125, "learning_rate": 8.758136725412166e-06, "loss": 0.6462, "step": 35451 }, { "epoch": 3.7722919770163865, "grad_norm": 9.484375, "learning_rate": 8.758101193860149e-06, "loss": 0.6232, "step": 35452 }, { "epoch": 3.772398382634603, "grad_norm": 18.125, "learning_rate": 8.75806566230813e-06, "loss": 0.6707, "step": 35453 }, { "epoch": 3.77250478825282, "grad_norm": 15.953125, "learning_rate": 8.758030130756112e-06, "loss": 0.6766, "step": 35454 }, { "epoch": 3.7726111938710365, "grad_norm": 12.0078125, "learning_rate": 8.757994599204094e-06, "loss": 0.528, "step": 35455 }, { "epoch": 3.772717599489253, "grad_norm": 11.875, "learning_rate": 8.757959067652075e-06, "loss": 0.5732, "step": 35456 }, { "epoch": 3.7728240051074695, "grad_norm": 17.171875, "learning_rate": 8.757923536100059e-06, "loss": 0.7679, "step": 35457 }, { "epoch": 3.7729304107256865, "grad_norm": 13.640625, "learning_rate": 8.75788800454804e-06, "loss": 0.6878, "step": 35458 }, { "epoch": 3.773036816343903, "grad_norm": 11.671875, "learning_rate": 8.757852472996022e-06, "loss": 0.6213, "step": 35459 }, { "epoch": 3.7731432219621195, "grad_norm": 17.6875, "learning_rate": 8.757816941444003e-06, "loss": 0.7158, "step": 35460 }, { "epoch": 3.773249627580336, "grad_norm": 15.5234375, "learning_rate": 8.757781409891985e-06, "loss": 0.6817, "step": 35461 }, { "epoch": 3.773356033198553, "grad_norm": 15.3125, "learning_rate": 8.757745878339967e-06, "loss": 0.6707, "step": 35462 }, { "epoch": 3.7734624388167695, "grad_norm": 16.140625, "learning_rate": 8.757710346787948e-06, "loss": 0.7711, "step": 35463 }, { "epoch": 3.773568844434986, "grad_norm": 11.1015625, "learning_rate": 8.75767481523593e-06, "loss": 0.4636, "step": 35464 }, { "epoch": 3.773675250053203, "grad_norm": 13.171875, "learning_rate": 8.757639283683911e-06, "loss": 0.58, "step": 35465 }, { "epoch": 3.7737816556714194, "grad_norm": 13.1640625, "learning_rate": 8.757603752131893e-06, "loss": 0.5826, "step": 35466 }, { "epoch": 3.773888061289636, "grad_norm": 15.2421875, "learning_rate": 8.757568220579875e-06, "loss": 0.5798, "step": 35467 }, { "epoch": 3.773994466907853, "grad_norm": 13.703125, "learning_rate": 8.757532689027858e-06, "loss": 0.6525, "step": 35468 }, { "epoch": 3.7741008725260694, "grad_norm": 11.0625, "learning_rate": 8.75749715747584e-06, "loss": 0.592, "step": 35469 }, { "epoch": 3.774207278144286, "grad_norm": 11.5859375, "learning_rate": 8.757461625923821e-06, "loss": 0.6247, "step": 35470 }, { "epoch": 3.774313683762503, "grad_norm": 14.84375, "learning_rate": 8.757426094371803e-06, "loss": 0.8466, "step": 35471 }, { "epoch": 3.7744200893807194, "grad_norm": 14.8671875, "learning_rate": 8.757390562819784e-06, "loss": 0.6443, "step": 35472 }, { "epoch": 3.774526494998936, "grad_norm": 14.6015625, "learning_rate": 8.757355031267768e-06, "loss": 0.7041, "step": 35473 }, { "epoch": 3.774632900617153, "grad_norm": 11.609375, "learning_rate": 8.75731949971575e-06, "loss": 0.6635, "step": 35474 }, { "epoch": 3.7747393062353694, "grad_norm": 16.875, "learning_rate": 8.75728396816373e-06, "loss": 0.7098, "step": 35475 }, { "epoch": 3.774845711853586, "grad_norm": 11.7890625, "learning_rate": 8.75724843661171e-06, "loss": 0.6247, "step": 35476 }, { "epoch": 3.7749521174718024, "grad_norm": 12.8125, "learning_rate": 8.757212905059692e-06, "loss": 0.6355, "step": 35477 }, { "epoch": 3.775058523090019, "grad_norm": 10.6953125, "learning_rate": 8.757177373507676e-06, "loss": 0.5744, "step": 35478 }, { "epoch": 3.775164928708236, "grad_norm": 10.9921875, "learning_rate": 8.757141841955657e-06, "loss": 0.553, "step": 35479 }, { "epoch": 3.7752713343264523, "grad_norm": 9.8359375, "learning_rate": 8.757106310403639e-06, "loss": 0.4886, "step": 35480 }, { "epoch": 3.775377739944669, "grad_norm": 13.5859375, "learning_rate": 8.75707077885162e-06, "loss": 0.7677, "step": 35481 }, { "epoch": 3.775484145562886, "grad_norm": 13.734375, "learning_rate": 8.757035247299602e-06, "loss": 0.5852, "step": 35482 }, { "epoch": 3.7755905511811023, "grad_norm": 11.6015625, "learning_rate": 8.756999715747585e-06, "loss": 0.5254, "step": 35483 }, { "epoch": 3.775696956799319, "grad_norm": 16.484375, "learning_rate": 8.756964184195567e-06, "loss": 0.7492, "step": 35484 }, { "epoch": 3.775803362417536, "grad_norm": 11.984375, "learning_rate": 8.756928652643549e-06, "loss": 0.6779, "step": 35485 }, { "epoch": 3.7759097680357523, "grad_norm": 14.828125, "learning_rate": 8.75689312109153e-06, "loss": 0.6219, "step": 35486 }, { "epoch": 3.776016173653969, "grad_norm": 12.28125, "learning_rate": 8.756857589539512e-06, "loss": 0.5688, "step": 35487 }, { "epoch": 3.7761225792721858, "grad_norm": 15.3203125, "learning_rate": 8.756822057987493e-06, "loss": 0.6408, "step": 35488 }, { "epoch": 3.7762289848904023, "grad_norm": 12.3125, "learning_rate": 8.756786526435475e-06, "loss": 0.6239, "step": 35489 }, { "epoch": 3.7763353905086188, "grad_norm": 14.96875, "learning_rate": 8.756750994883457e-06, "loss": 0.7472, "step": 35490 }, { "epoch": 3.7764417961268357, "grad_norm": 11.5390625, "learning_rate": 8.756715463331438e-06, "loss": 0.8071, "step": 35491 }, { "epoch": 3.7765482017450522, "grad_norm": 15.140625, "learning_rate": 8.75667993177942e-06, "loss": 0.7897, "step": 35492 }, { "epoch": 3.7766546073632687, "grad_norm": 11.0859375, "learning_rate": 8.756644400227403e-06, "loss": 0.552, "step": 35493 }, { "epoch": 3.7767610129814857, "grad_norm": 15.90625, "learning_rate": 8.756608868675385e-06, "loss": 0.67, "step": 35494 }, { "epoch": 3.776867418599702, "grad_norm": 11.8203125, "learning_rate": 8.756573337123366e-06, "loss": 0.5846, "step": 35495 }, { "epoch": 3.7769738242179187, "grad_norm": 10.9609375, "learning_rate": 8.756537805571348e-06, "loss": 0.5542, "step": 35496 }, { "epoch": 3.7770802298361352, "grad_norm": 13.3125, "learning_rate": 8.75650227401933e-06, "loss": 0.6591, "step": 35497 }, { "epoch": 3.7771866354543517, "grad_norm": 13.109375, "learning_rate": 8.756466742467313e-06, "loss": 0.5547, "step": 35498 }, { "epoch": 3.7772930410725687, "grad_norm": 11.7109375, "learning_rate": 8.756431210915295e-06, "loss": 0.5885, "step": 35499 }, { "epoch": 3.777399446690785, "grad_norm": 21.375, "learning_rate": 8.756395679363276e-06, "loss": 0.7703, "step": 35500 }, { "epoch": 3.7775058523090017, "grad_norm": 17.015625, "learning_rate": 8.756360147811258e-06, "loss": 0.8195, "step": 35501 }, { "epoch": 3.7776122579272187, "grad_norm": 16.1875, "learning_rate": 8.75632461625924e-06, "loss": 0.7371, "step": 35502 }, { "epoch": 3.777718663545435, "grad_norm": 16.28125, "learning_rate": 8.756289084707221e-06, "loss": 0.7623, "step": 35503 }, { "epoch": 3.7778250691636517, "grad_norm": 12.53125, "learning_rate": 8.756253553155203e-06, "loss": 0.6984, "step": 35504 }, { "epoch": 3.7779314747818686, "grad_norm": 9.671875, "learning_rate": 8.756218021603184e-06, "loss": 0.5731, "step": 35505 }, { "epoch": 3.778037880400085, "grad_norm": 14.5234375, "learning_rate": 8.756182490051166e-06, "loss": 0.6978, "step": 35506 }, { "epoch": 3.7781442860183017, "grad_norm": 11.84375, "learning_rate": 8.756146958499147e-06, "loss": 0.6436, "step": 35507 }, { "epoch": 3.7782506916365186, "grad_norm": 13.7421875, "learning_rate": 8.756111426947129e-06, "loss": 0.6772, "step": 35508 }, { "epoch": 3.778357097254735, "grad_norm": 13.0078125, "learning_rate": 8.756075895395112e-06, "loss": 0.6675, "step": 35509 }, { "epoch": 3.7784635028729516, "grad_norm": 11.046875, "learning_rate": 8.756040363843094e-06, "loss": 0.5999, "step": 35510 }, { "epoch": 3.7785699084911686, "grad_norm": 10.4609375, "learning_rate": 8.756004832291076e-06, "loss": 0.5765, "step": 35511 }, { "epoch": 3.778676314109385, "grad_norm": 14.234375, "learning_rate": 8.755969300739057e-06, "loss": 0.6897, "step": 35512 }, { "epoch": 3.7787827197276016, "grad_norm": 14.6875, "learning_rate": 8.755933769187039e-06, "loss": 0.5974, "step": 35513 }, { "epoch": 3.778889125345818, "grad_norm": 11.9375, "learning_rate": 8.75589823763502e-06, "loss": 0.6158, "step": 35514 }, { "epoch": 3.778995530964035, "grad_norm": 16.5, "learning_rate": 8.755862706083002e-06, "loss": 0.7952, "step": 35515 }, { "epoch": 3.7791019365822516, "grad_norm": 9.2109375, "learning_rate": 8.755827174530984e-06, "loss": 0.6164, "step": 35516 }, { "epoch": 3.779208342200468, "grad_norm": 14.9765625, "learning_rate": 8.755791642978965e-06, "loss": 0.7091, "step": 35517 }, { "epoch": 3.7793147478186846, "grad_norm": 12.3203125, "learning_rate": 8.755756111426947e-06, "loss": 0.6543, "step": 35518 }, { "epoch": 3.7794211534369015, "grad_norm": 15.046875, "learning_rate": 8.75572057987493e-06, "loss": 0.6005, "step": 35519 }, { "epoch": 3.779527559055118, "grad_norm": 16.375, "learning_rate": 8.755685048322912e-06, "loss": 0.7283, "step": 35520 }, { "epoch": 3.7796339646733346, "grad_norm": 12.984375, "learning_rate": 8.755649516770893e-06, "loss": 0.6013, "step": 35521 }, { "epoch": 3.7797403702915515, "grad_norm": 10.59375, "learning_rate": 8.755613985218875e-06, "loss": 0.4386, "step": 35522 }, { "epoch": 3.779846775909768, "grad_norm": 12.125, "learning_rate": 8.755578453666856e-06, "loss": 0.5852, "step": 35523 }, { "epoch": 3.7799531815279845, "grad_norm": 22.109375, "learning_rate": 8.75554292211484e-06, "loss": 0.8792, "step": 35524 }, { "epoch": 3.7800595871462015, "grad_norm": 10.6328125, "learning_rate": 8.755507390562821e-06, "loss": 0.5983, "step": 35525 }, { "epoch": 3.780165992764418, "grad_norm": 13.5703125, "learning_rate": 8.755471859010803e-06, "loss": 0.6187, "step": 35526 }, { "epoch": 3.7802723983826345, "grad_norm": 11.328125, "learning_rate": 8.755436327458785e-06, "loss": 0.6638, "step": 35527 }, { "epoch": 3.7803788040008515, "grad_norm": 10.6875, "learning_rate": 8.755400795906766e-06, "loss": 0.5295, "step": 35528 }, { "epoch": 3.780485209619068, "grad_norm": 14.5859375, "learning_rate": 8.755365264354748e-06, "loss": 0.7106, "step": 35529 }, { "epoch": 3.7805916152372845, "grad_norm": 17.109375, "learning_rate": 8.75532973280273e-06, "loss": 0.8261, "step": 35530 }, { "epoch": 3.7806980208555014, "grad_norm": 24.25, "learning_rate": 8.755294201250711e-06, "loss": 0.9912, "step": 35531 }, { "epoch": 3.780804426473718, "grad_norm": 13.3671875, "learning_rate": 8.755258669698693e-06, "loss": 0.8609, "step": 35532 }, { "epoch": 3.7809108320919345, "grad_norm": 16.828125, "learning_rate": 8.755223138146674e-06, "loss": 0.734, "step": 35533 }, { "epoch": 3.781017237710151, "grad_norm": 12.703125, "learning_rate": 8.755187606594658e-06, "loss": 0.6688, "step": 35534 }, { "epoch": 3.7811236433283675, "grad_norm": 12.046875, "learning_rate": 8.75515207504264e-06, "loss": 0.718, "step": 35535 }, { "epoch": 3.7812300489465844, "grad_norm": 15.0234375, "learning_rate": 8.75511654349062e-06, "loss": 0.7554, "step": 35536 }, { "epoch": 3.781336454564801, "grad_norm": 12.125, "learning_rate": 8.755081011938602e-06, "loss": 0.5969, "step": 35537 }, { "epoch": 3.7814428601830175, "grad_norm": 13.4140625, "learning_rate": 8.755045480386584e-06, "loss": 0.7372, "step": 35538 }, { "epoch": 3.7815492658012344, "grad_norm": 12.1796875, "learning_rate": 8.755009948834566e-06, "loss": 0.6232, "step": 35539 }, { "epoch": 3.781655671419451, "grad_norm": 12.3203125, "learning_rate": 8.754974417282547e-06, "loss": 0.7011, "step": 35540 }, { "epoch": 3.7817620770376674, "grad_norm": 13.515625, "learning_rate": 8.754938885730529e-06, "loss": 0.7803, "step": 35541 }, { "epoch": 3.7818684826558844, "grad_norm": 9.875, "learning_rate": 8.75490335417851e-06, "loss": 0.4965, "step": 35542 }, { "epoch": 3.781974888274101, "grad_norm": 11.6953125, "learning_rate": 8.754867822626492e-06, "loss": 0.6781, "step": 35543 }, { "epoch": 3.7820812938923174, "grad_norm": 14.6328125, "learning_rate": 8.754832291074474e-06, "loss": 0.7538, "step": 35544 }, { "epoch": 3.7821876995105344, "grad_norm": 10.3125, "learning_rate": 8.754796759522457e-06, "loss": 0.5395, "step": 35545 }, { "epoch": 3.782294105128751, "grad_norm": 10.5390625, "learning_rate": 8.754761227970439e-06, "loss": 0.5259, "step": 35546 }, { "epoch": 3.7824005107469674, "grad_norm": 11.8046875, "learning_rate": 8.75472569641842e-06, "loss": 0.554, "step": 35547 }, { "epoch": 3.7825069163651843, "grad_norm": 12.015625, "learning_rate": 8.754690164866402e-06, "loss": 0.618, "step": 35548 }, { "epoch": 3.782613321983401, "grad_norm": 14.78125, "learning_rate": 8.754654633314383e-06, "loss": 0.5898, "step": 35549 }, { "epoch": 3.7827197276016173, "grad_norm": 14.6015625, "learning_rate": 8.754619101762367e-06, "loss": 0.7182, "step": 35550 }, { "epoch": 3.782826133219834, "grad_norm": 10.8828125, "learning_rate": 8.754583570210348e-06, "loss": 0.6248, "step": 35551 }, { "epoch": 3.782932538838051, "grad_norm": 17.640625, "learning_rate": 8.75454803865833e-06, "loss": 0.8093, "step": 35552 }, { "epoch": 3.7830389444562673, "grad_norm": 13.2421875, "learning_rate": 8.754512507106311e-06, "loss": 0.7005, "step": 35553 }, { "epoch": 3.783145350074484, "grad_norm": 16.578125, "learning_rate": 8.754476975554293e-06, "loss": 0.6068, "step": 35554 }, { "epoch": 3.7832517556927003, "grad_norm": 13.875, "learning_rate": 8.754441444002275e-06, "loss": 0.7307, "step": 35555 }, { "epoch": 3.7833581613109173, "grad_norm": 19.46875, "learning_rate": 8.754405912450256e-06, "loss": 0.8685, "step": 35556 }, { "epoch": 3.783464566929134, "grad_norm": 14.0703125, "learning_rate": 8.754370380898238e-06, "loss": 0.6573, "step": 35557 }, { "epoch": 3.7835709725473503, "grad_norm": 11.5390625, "learning_rate": 8.75433484934622e-06, "loss": 0.6466, "step": 35558 }, { "epoch": 3.7836773781655673, "grad_norm": 14.921875, "learning_rate": 8.754299317794201e-06, "loss": 0.7622, "step": 35559 }, { "epoch": 3.7837837837837838, "grad_norm": 13.25, "learning_rate": 8.754263786242184e-06, "loss": 0.6037, "step": 35560 }, { "epoch": 3.7838901894020003, "grad_norm": 9.828125, "learning_rate": 8.754228254690166e-06, "loss": 0.567, "step": 35561 }, { "epoch": 3.7839965950202172, "grad_norm": 14.09375, "learning_rate": 8.754192723138148e-06, "loss": 0.6665, "step": 35562 }, { "epoch": 3.7841030006384337, "grad_norm": 18.5625, "learning_rate": 8.75415719158613e-06, "loss": 0.7108, "step": 35563 }, { "epoch": 3.7842094062566503, "grad_norm": 13.5703125, "learning_rate": 8.75412166003411e-06, "loss": 0.6908, "step": 35564 }, { "epoch": 3.784315811874867, "grad_norm": 14.71875, "learning_rate": 8.754086128482092e-06, "loss": 0.64, "step": 35565 }, { "epoch": 3.7844222174930837, "grad_norm": 15.890625, "learning_rate": 8.754050596930074e-06, "loss": 0.8327, "step": 35566 }, { "epoch": 3.7845286231113002, "grad_norm": 11.046875, "learning_rate": 8.754015065378056e-06, "loss": 0.6405, "step": 35567 }, { "epoch": 3.784635028729517, "grad_norm": 18.40625, "learning_rate": 8.753979533826037e-06, "loss": 0.7344, "step": 35568 }, { "epoch": 3.7847414343477337, "grad_norm": 15.8984375, "learning_rate": 8.753944002274019e-06, "loss": 0.7236, "step": 35569 }, { "epoch": 3.78484783996595, "grad_norm": 12.7578125, "learning_rate": 8.753908470722002e-06, "loss": 0.6359, "step": 35570 }, { "epoch": 3.7849542455841667, "grad_norm": 14.4765625, "learning_rate": 8.753872939169984e-06, "loss": 0.6784, "step": 35571 }, { "epoch": 3.785060651202383, "grad_norm": 11.5703125, "learning_rate": 8.753837407617965e-06, "loss": 0.665, "step": 35572 }, { "epoch": 3.7851670568206, "grad_norm": 12.203125, "learning_rate": 8.753801876065947e-06, "loss": 0.5096, "step": 35573 }, { "epoch": 3.7852734624388167, "grad_norm": 13.125, "learning_rate": 8.753766344513929e-06, "loss": 0.637, "step": 35574 }, { "epoch": 3.785379868057033, "grad_norm": 11.5625, "learning_rate": 8.753730812961912e-06, "loss": 0.6121, "step": 35575 }, { "epoch": 3.78548627367525, "grad_norm": 13.609375, "learning_rate": 8.753695281409894e-06, "loss": 0.7003, "step": 35576 }, { "epoch": 3.7855926792934667, "grad_norm": 12.640625, "learning_rate": 8.753659749857875e-06, "loss": 0.6215, "step": 35577 }, { "epoch": 3.785699084911683, "grad_norm": 13.5859375, "learning_rate": 8.753624218305857e-06, "loss": 0.5461, "step": 35578 }, { "epoch": 3.7858054905299, "grad_norm": 14.7734375, "learning_rate": 8.753588686753838e-06, "loss": 0.6341, "step": 35579 }, { "epoch": 3.7859118961481166, "grad_norm": 10.3125, "learning_rate": 8.75355315520182e-06, "loss": 0.517, "step": 35580 }, { "epoch": 3.786018301766333, "grad_norm": 14.0546875, "learning_rate": 8.753517623649802e-06, "loss": 0.614, "step": 35581 }, { "epoch": 3.78612470738455, "grad_norm": 13.6640625, "learning_rate": 8.753482092097783e-06, "loss": 0.6451, "step": 35582 }, { "epoch": 3.7862311130027666, "grad_norm": 11.328125, "learning_rate": 8.753446560545765e-06, "loss": 0.5814, "step": 35583 }, { "epoch": 3.786337518620983, "grad_norm": 19.796875, "learning_rate": 8.753411028993746e-06, "loss": 0.7855, "step": 35584 }, { "epoch": 3.7864439242392, "grad_norm": 16.359375, "learning_rate": 8.753375497441728e-06, "loss": 0.6884, "step": 35585 }, { "epoch": 3.7865503298574166, "grad_norm": 11.015625, "learning_rate": 8.753339965889711e-06, "loss": 0.5381, "step": 35586 }, { "epoch": 3.786656735475633, "grad_norm": 13.234375, "learning_rate": 8.753304434337693e-06, "loss": 0.5673, "step": 35587 }, { "epoch": 3.7867631410938496, "grad_norm": 15.0390625, "learning_rate": 8.753268902785674e-06, "loss": 0.6945, "step": 35588 }, { "epoch": 3.7868695467120665, "grad_norm": 12.625, "learning_rate": 8.753233371233656e-06, "loss": 0.5685, "step": 35589 }, { "epoch": 3.786975952330283, "grad_norm": 12.671875, "learning_rate": 8.753197839681638e-06, "loss": 0.763, "step": 35590 }, { "epoch": 3.7870823579484996, "grad_norm": 14.8984375, "learning_rate": 8.753162308129621e-06, "loss": 0.777, "step": 35591 }, { "epoch": 3.787188763566716, "grad_norm": 11.0546875, "learning_rate": 8.753126776577603e-06, "loss": 0.5914, "step": 35592 }, { "epoch": 3.787295169184933, "grad_norm": 10.828125, "learning_rate": 8.753091245025584e-06, "loss": 0.5408, "step": 35593 }, { "epoch": 3.7874015748031495, "grad_norm": 11.3125, "learning_rate": 8.753055713473566e-06, "loss": 0.6422, "step": 35594 }, { "epoch": 3.787507980421366, "grad_norm": 11.3515625, "learning_rate": 8.753020181921546e-06, "loss": 0.633, "step": 35595 }, { "epoch": 3.787614386039583, "grad_norm": 13.5625, "learning_rate": 8.752984650369529e-06, "loss": 0.7108, "step": 35596 }, { "epoch": 3.7877207916577995, "grad_norm": 8.640625, "learning_rate": 8.75294911881751e-06, "loss": 0.4436, "step": 35597 }, { "epoch": 3.787827197276016, "grad_norm": 18.546875, "learning_rate": 8.752913587265492e-06, "loss": 0.7077, "step": 35598 }, { "epoch": 3.787933602894233, "grad_norm": 14.8515625, "learning_rate": 8.752878055713474e-06, "loss": 0.7875, "step": 35599 }, { "epoch": 3.7880400085124495, "grad_norm": 12.03125, "learning_rate": 8.752842524161455e-06, "loss": 0.6574, "step": 35600 }, { "epoch": 3.788146414130666, "grad_norm": 13.96875, "learning_rate": 8.752806992609439e-06, "loss": 0.7405, "step": 35601 }, { "epoch": 3.788252819748883, "grad_norm": 15.6484375, "learning_rate": 8.75277146105742e-06, "loss": 0.7657, "step": 35602 }, { "epoch": 3.7883592253670995, "grad_norm": 13.40625, "learning_rate": 8.752735929505402e-06, "loss": 0.7181, "step": 35603 }, { "epoch": 3.788465630985316, "grad_norm": 12.921875, "learning_rate": 8.752700397953384e-06, "loss": 0.6477, "step": 35604 }, { "epoch": 3.788572036603533, "grad_norm": 10.265625, "learning_rate": 8.752664866401365e-06, "loss": 0.4718, "step": 35605 }, { "epoch": 3.7886784422217494, "grad_norm": 13.1328125, "learning_rate": 8.752629334849347e-06, "loss": 0.77, "step": 35606 }, { "epoch": 3.788784847839966, "grad_norm": 14.640625, "learning_rate": 8.752593803297328e-06, "loss": 0.7011, "step": 35607 }, { "epoch": 3.7888912534581825, "grad_norm": 9.421875, "learning_rate": 8.75255827174531e-06, "loss": 0.5055, "step": 35608 }, { "epoch": 3.788997659076399, "grad_norm": 11.640625, "learning_rate": 8.752522740193292e-06, "loss": 0.6509, "step": 35609 }, { "epoch": 3.789104064694616, "grad_norm": 13.7109375, "learning_rate": 8.752487208641273e-06, "loss": 0.6851, "step": 35610 }, { "epoch": 3.7892104703128324, "grad_norm": 13.5234375, "learning_rate": 8.752451677089257e-06, "loss": 0.6585, "step": 35611 }, { "epoch": 3.789316875931049, "grad_norm": 14.8515625, "learning_rate": 8.752416145537238e-06, "loss": 0.6313, "step": 35612 }, { "epoch": 3.789423281549266, "grad_norm": 11.6953125, "learning_rate": 8.75238061398522e-06, "loss": 0.6429, "step": 35613 }, { "epoch": 3.7895296871674824, "grad_norm": 13.0859375, "learning_rate": 8.752345082433201e-06, "loss": 0.6459, "step": 35614 }, { "epoch": 3.789636092785699, "grad_norm": 13.4765625, "learning_rate": 8.752309550881183e-06, "loss": 0.7644, "step": 35615 }, { "epoch": 3.789742498403916, "grad_norm": 9.5859375, "learning_rate": 8.752274019329166e-06, "loss": 0.6751, "step": 35616 }, { "epoch": 3.7898489040221324, "grad_norm": 15.5703125, "learning_rate": 8.752238487777148e-06, "loss": 0.8912, "step": 35617 }, { "epoch": 3.789955309640349, "grad_norm": 12.40625, "learning_rate": 8.75220295622513e-06, "loss": 0.6714, "step": 35618 }, { "epoch": 3.790061715258566, "grad_norm": 10.84375, "learning_rate": 8.752167424673111e-06, "loss": 0.5811, "step": 35619 }, { "epoch": 3.7901681208767823, "grad_norm": 12.4921875, "learning_rate": 8.752131893121093e-06, "loss": 0.6439, "step": 35620 }, { "epoch": 3.790274526494999, "grad_norm": 14.7265625, "learning_rate": 8.752096361569074e-06, "loss": 0.6228, "step": 35621 }, { "epoch": 3.790380932113216, "grad_norm": 9.484375, "learning_rate": 8.752060830017056e-06, "loss": 0.6066, "step": 35622 }, { "epoch": 3.7904873377314323, "grad_norm": 15.2578125, "learning_rate": 8.752025298465037e-06, "loss": 0.6553, "step": 35623 }, { "epoch": 3.790593743349649, "grad_norm": 11.0390625, "learning_rate": 8.751989766913019e-06, "loss": 0.5421, "step": 35624 }, { "epoch": 3.790700148967866, "grad_norm": 15.828125, "learning_rate": 8.751954235361e-06, "loss": 0.6336, "step": 35625 }, { "epoch": 3.7908065545860823, "grad_norm": 13.640625, "learning_rate": 8.751918703808982e-06, "loss": 0.7243, "step": 35626 }, { "epoch": 3.790912960204299, "grad_norm": 13.5078125, "learning_rate": 8.751883172256966e-06, "loss": 0.6648, "step": 35627 }, { "epoch": 3.7910193658225153, "grad_norm": 16.328125, "learning_rate": 8.751847640704947e-06, "loss": 0.86, "step": 35628 }, { "epoch": 3.791125771440732, "grad_norm": 9.90625, "learning_rate": 8.751812109152929e-06, "loss": 0.5694, "step": 35629 }, { "epoch": 3.7912321770589488, "grad_norm": 15.609375, "learning_rate": 8.75177657760091e-06, "loss": 0.7933, "step": 35630 }, { "epoch": 3.7913385826771653, "grad_norm": 14.046875, "learning_rate": 8.751741046048892e-06, "loss": 0.5646, "step": 35631 }, { "epoch": 3.791444988295382, "grad_norm": 14.2890625, "learning_rate": 8.751705514496874e-06, "loss": 0.6458, "step": 35632 }, { "epoch": 3.7915513939135987, "grad_norm": 12.9375, "learning_rate": 8.751669982944855e-06, "loss": 0.6023, "step": 35633 }, { "epoch": 3.7916577995318153, "grad_norm": 16.484375, "learning_rate": 8.751634451392837e-06, "loss": 0.6893, "step": 35634 }, { "epoch": 3.7917642051500318, "grad_norm": 11.1875, "learning_rate": 8.751598919840818e-06, "loss": 0.7079, "step": 35635 }, { "epoch": 3.7918706107682487, "grad_norm": 11.96875, "learning_rate": 8.7515633882888e-06, "loss": 0.5393, "step": 35636 }, { "epoch": 3.7919770163864652, "grad_norm": 13.515625, "learning_rate": 8.751527856736783e-06, "loss": 0.7471, "step": 35637 }, { "epoch": 3.7920834220046817, "grad_norm": 11.6640625, "learning_rate": 8.751492325184765e-06, "loss": 0.6375, "step": 35638 }, { "epoch": 3.7921898276228987, "grad_norm": 9.1640625, "learning_rate": 8.751456793632747e-06, "loss": 0.4601, "step": 35639 }, { "epoch": 3.792296233241115, "grad_norm": 14.0859375, "learning_rate": 8.751421262080728e-06, "loss": 0.6577, "step": 35640 }, { "epoch": 3.7924026388593317, "grad_norm": 15.28125, "learning_rate": 8.75138573052871e-06, "loss": 0.6622, "step": 35641 }, { "epoch": 3.7925090444775487, "grad_norm": 13.25, "learning_rate": 8.751350198976693e-06, "loss": 0.6269, "step": 35642 }, { "epoch": 3.792615450095765, "grad_norm": 13.3359375, "learning_rate": 8.751314667424675e-06, "loss": 0.6801, "step": 35643 }, { "epoch": 3.7927218557139817, "grad_norm": 12.5703125, "learning_rate": 8.751279135872656e-06, "loss": 0.6462, "step": 35644 }, { "epoch": 3.792828261332198, "grad_norm": 10.6796875, "learning_rate": 8.751243604320638e-06, "loss": 0.6231, "step": 35645 }, { "epoch": 3.792934666950415, "grad_norm": 12.2109375, "learning_rate": 8.75120807276862e-06, "loss": 0.571, "step": 35646 }, { "epoch": 3.7930410725686317, "grad_norm": 13.6640625, "learning_rate": 8.751172541216601e-06, "loss": 0.7415, "step": 35647 }, { "epoch": 3.793147478186848, "grad_norm": 15.6484375, "learning_rate": 8.751137009664583e-06, "loss": 0.9037, "step": 35648 }, { "epoch": 3.7932538838050647, "grad_norm": 17.375, "learning_rate": 8.751101478112564e-06, "loss": 0.6317, "step": 35649 }, { "epoch": 3.7933602894232816, "grad_norm": 14.921875, "learning_rate": 8.751065946560546e-06, "loss": 0.8077, "step": 35650 }, { "epoch": 3.793466695041498, "grad_norm": 11.2578125, "learning_rate": 8.751030415008528e-06, "loss": 0.5904, "step": 35651 }, { "epoch": 3.7935731006597146, "grad_norm": 14.4375, "learning_rate": 8.750994883456511e-06, "loss": 0.7433, "step": 35652 }, { "epoch": 3.7936795062779316, "grad_norm": 14.484375, "learning_rate": 8.750959351904492e-06, "loss": 0.6886, "step": 35653 }, { "epoch": 3.793785911896148, "grad_norm": 12.5078125, "learning_rate": 8.750923820352474e-06, "loss": 0.6402, "step": 35654 }, { "epoch": 3.7938923175143646, "grad_norm": 13.5546875, "learning_rate": 8.750888288800456e-06, "loss": 0.5996, "step": 35655 }, { "epoch": 3.7939987231325816, "grad_norm": 10.6015625, "learning_rate": 8.750852757248437e-06, "loss": 0.5697, "step": 35656 }, { "epoch": 3.794105128750798, "grad_norm": 15.375, "learning_rate": 8.750817225696419e-06, "loss": 0.7811, "step": 35657 }, { "epoch": 3.7942115343690146, "grad_norm": 13.015625, "learning_rate": 8.7507816941444e-06, "loss": 0.6314, "step": 35658 }, { "epoch": 3.7943179399872315, "grad_norm": 13.890625, "learning_rate": 8.750746162592382e-06, "loss": 0.6715, "step": 35659 }, { "epoch": 3.794424345605448, "grad_norm": 13.28125, "learning_rate": 8.750710631040364e-06, "loss": 0.6657, "step": 35660 }, { "epoch": 3.7945307512236646, "grad_norm": 11.0, "learning_rate": 8.750675099488345e-06, "loss": 0.5177, "step": 35661 }, { "epoch": 3.7946371568418815, "grad_norm": 11.15625, "learning_rate": 8.750639567936327e-06, "loss": 0.5892, "step": 35662 }, { "epoch": 3.794743562460098, "grad_norm": 13.6171875, "learning_rate": 8.75060403638431e-06, "loss": 0.6497, "step": 35663 }, { "epoch": 3.7948499680783145, "grad_norm": 17.453125, "learning_rate": 8.750568504832292e-06, "loss": 0.7178, "step": 35664 }, { "epoch": 3.794956373696531, "grad_norm": 13.265625, "learning_rate": 8.750532973280273e-06, "loss": 0.7724, "step": 35665 }, { "epoch": 3.7950627793147476, "grad_norm": 14.40625, "learning_rate": 8.750497441728255e-06, "loss": 0.7193, "step": 35666 }, { "epoch": 3.7951691849329645, "grad_norm": 111.625, "learning_rate": 8.750461910176237e-06, "loss": 0.7021, "step": 35667 }, { "epoch": 3.795275590551181, "grad_norm": 12.8125, "learning_rate": 8.75042637862422e-06, "loss": 0.5485, "step": 35668 }, { "epoch": 3.7953819961693975, "grad_norm": 14.328125, "learning_rate": 8.750390847072202e-06, "loss": 0.6606, "step": 35669 }, { "epoch": 3.7954884017876145, "grad_norm": 13.90625, "learning_rate": 8.750355315520183e-06, "loss": 0.6505, "step": 35670 }, { "epoch": 3.795594807405831, "grad_norm": 12.5390625, "learning_rate": 8.750319783968165e-06, "loss": 0.5795, "step": 35671 }, { "epoch": 3.7957012130240475, "grad_norm": 15.359375, "learning_rate": 8.750284252416146e-06, "loss": 0.8104, "step": 35672 }, { "epoch": 3.7958076186422645, "grad_norm": 16.5, "learning_rate": 8.750248720864128e-06, "loss": 0.6861, "step": 35673 }, { "epoch": 3.795914024260481, "grad_norm": 19.46875, "learning_rate": 8.75021318931211e-06, "loss": 0.7927, "step": 35674 }, { "epoch": 3.7960204298786975, "grad_norm": 12.296875, "learning_rate": 8.750177657760091e-06, "loss": 0.6737, "step": 35675 }, { "epoch": 3.7961268354969144, "grad_norm": 10.0, "learning_rate": 8.750142126208073e-06, "loss": 0.645, "step": 35676 }, { "epoch": 3.796233241115131, "grad_norm": 15.0859375, "learning_rate": 8.750106594656054e-06, "loss": 0.7349, "step": 35677 }, { "epoch": 3.7963396467333475, "grad_norm": 14.0546875, "learning_rate": 8.750071063104038e-06, "loss": 0.7097, "step": 35678 }, { "epoch": 3.7964460523515644, "grad_norm": 12.9921875, "learning_rate": 8.75003553155202e-06, "loss": 0.6145, "step": 35679 }, { "epoch": 3.796552457969781, "grad_norm": 18.46875, "learning_rate": 8.750000000000001e-06, "loss": 0.8039, "step": 35680 }, { "epoch": 3.7966588635879974, "grad_norm": 12.75, "learning_rate": 8.749964468447983e-06, "loss": 0.6681, "step": 35681 }, { "epoch": 3.796765269206214, "grad_norm": 10.1328125, "learning_rate": 8.749928936895964e-06, "loss": 0.6372, "step": 35682 }, { "epoch": 3.796871674824431, "grad_norm": 12.421875, "learning_rate": 8.749893405343947e-06, "loss": 0.6912, "step": 35683 }, { "epoch": 3.7969780804426474, "grad_norm": 13.734375, "learning_rate": 8.749857873791927e-06, "loss": 0.7449, "step": 35684 }, { "epoch": 3.797084486060864, "grad_norm": 11.1484375, "learning_rate": 8.749822342239909e-06, "loss": 0.6691, "step": 35685 }, { "epoch": 3.7971908916790804, "grad_norm": 14.3671875, "learning_rate": 8.74978681068789e-06, "loss": 0.7468, "step": 35686 }, { "epoch": 3.7972972972972974, "grad_norm": 12.1875, "learning_rate": 8.749751279135872e-06, "loss": 0.8032, "step": 35687 }, { "epoch": 3.797403702915514, "grad_norm": 13.953125, "learning_rate": 8.749715747583855e-06, "loss": 0.6373, "step": 35688 }, { "epoch": 3.7975101085337304, "grad_norm": 14.375, "learning_rate": 8.749680216031837e-06, "loss": 0.7356, "step": 35689 }, { "epoch": 3.7976165141519473, "grad_norm": 13.0859375, "learning_rate": 8.749644684479819e-06, "loss": 0.603, "step": 35690 }, { "epoch": 3.797722919770164, "grad_norm": 14.78125, "learning_rate": 8.7496091529278e-06, "loss": 0.7086, "step": 35691 }, { "epoch": 3.7978293253883804, "grad_norm": 17.609375, "learning_rate": 8.749573621375782e-06, "loss": 0.8048, "step": 35692 }, { "epoch": 3.7979357310065973, "grad_norm": 12.9296875, "learning_rate": 8.749538089823765e-06, "loss": 0.5333, "step": 35693 }, { "epoch": 3.798042136624814, "grad_norm": 14.5703125, "learning_rate": 8.749502558271747e-06, "loss": 0.7449, "step": 35694 }, { "epoch": 3.7981485422430303, "grad_norm": 15.5, "learning_rate": 8.749467026719728e-06, "loss": 0.7223, "step": 35695 }, { "epoch": 3.7982549478612473, "grad_norm": 15.703125, "learning_rate": 8.74943149516771e-06, "loss": 0.7094, "step": 35696 }, { "epoch": 3.798361353479464, "grad_norm": 11.90625, "learning_rate": 8.749395963615692e-06, "loss": 0.7389, "step": 35697 }, { "epoch": 3.7984677590976803, "grad_norm": 13.5625, "learning_rate": 8.749360432063673e-06, "loss": 0.6957, "step": 35698 }, { "epoch": 3.7985741647158973, "grad_norm": 10.4296875, "learning_rate": 8.749324900511655e-06, "loss": 0.6395, "step": 35699 }, { "epoch": 3.7986805703341138, "grad_norm": 14.4921875, "learning_rate": 8.749289368959636e-06, "loss": 0.728, "step": 35700 }, { "epoch": 3.7987869759523303, "grad_norm": 16.125, "learning_rate": 8.749253837407618e-06, "loss": 0.711, "step": 35701 }, { "epoch": 3.798893381570547, "grad_norm": 15.0703125, "learning_rate": 8.7492183058556e-06, "loss": 0.6191, "step": 35702 }, { "epoch": 3.7989997871887633, "grad_norm": 13.296875, "learning_rate": 8.749182774303581e-06, "loss": 0.5337, "step": 35703 }, { "epoch": 3.7991061928069803, "grad_norm": 15.2890625, "learning_rate": 8.749147242751565e-06, "loss": 0.8367, "step": 35704 }, { "epoch": 3.7992125984251968, "grad_norm": 12.40625, "learning_rate": 8.749111711199546e-06, "loss": 0.6635, "step": 35705 }, { "epoch": 3.7993190040434133, "grad_norm": 12.9921875, "learning_rate": 8.749076179647528e-06, "loss": 0.6825, "step": 35706 }, { "epoch": 3.7994254096616302, "grad_norm": 10.453125, "learning_rate": 8.74904064809551e-06, "loss": 0.5871, "step": 35707 }, { "epoch": 3.7995318152798467, "grad_norm": 11.6875, "learning_rate": 8.749005116543491e-06, "loss": 0.6524, "step": 35708 }, { "epoch": 3.7996382208980632, "grad_norm": 15.0625, "learning_rate": 8.748969584991474e-06, "loss": 0.727, "step": 35709 }, { "epoch": 3.79974462651628, "grad_norm": 14.1484375, "learning_rate": 8.748934053439456e-06, "loss": 0.5652, "step": 35710 }, { "epoch": 3.7998510321344967, "grad_norm": 12.7890625, "learning_rate": 8.748898521887438e-06, "loss": 0.5965, "step": 35711 }, { "epoch": 3.799957437752713, "grad_norm": 19.09375, "learning_rate": 8.748862990335419e-06, "loss": 0.7847, "step": 35712 }, { "epoch": 3.80006384337093, "grad_norm": 13.1640625, "learning_rate": 8.7488274587834e-06, "loss": 0.6042, "step": 35713 }, { "epoch": 3.8001702489891467, "grad_norm": 11.765625, "learning_rate": 8.748791927231382e-06, "loss": 0.7023, "step": 35714 }, { "epoch": 3.800276654607363, "grad_norm": 15.75, "learning_rate": 8.748756395679364e-06, "loss": 0.777, "step": 35715 }, { "epoch": 3.80038306022558, "grad_norm": 13.375, "learning_rate": 8.748720864127346e-06, "loss": 0.7192, "step": 35716 }, { "epoch": 3.8004894658437967, "grad_norm": 14.1875, "learning_rate": 8.748685332575327e-06, "loss": 0.7439, "step": 35717 }, { "epoch": 3.800595871462013, "grad_norm": 15.6015625, "learning_rate": 8.748649801023309e-06, "loss": 0.7769, "step": 35718 }, { "epoch": 3.8007022770802297, "grad_norm": 11.796875, "learning_rate": 8.748614269471292e-06, "loss": 0.5919, "step": 35719 }, { "epoch": 3.8008086826984466, "grad_norm": 14.9765625, "learning_rate": 8.748578737919274e-06, "loss": 0.6137, "step": 35720 }, { "epoch": 3.800915088316663, "grad_norm": 13.2265625, "learning_rate": 8.748543206367255e-06, "loss": 0.7011, "step": 35721 }, { "epoch": 3.8010214939348796, "grad_norm": 14.2578125, "learning_rate": 8.748507674815237e-06, "loss": 0.6497, "step": 35722 }, { "epoch": 3.801127899553096, "grad_norm": 17.34375, "learning_rate": 8.748472143263218e-06, "loss": 0.7594, "step": 35723 }, { "epoch": 3.801234305171313, "grad_norm": 12.8515625, "learning_rate": 8.7484366117112e-06, "loss": 0.5402, "step": 35724 }, { "epoch": 3.8013407107895296, "grad_norm": 13.2578125, "learning_rate": 8.748401080159182e-06, "loss": 0.826, "step": 35725 }, { "epoch": 3.801447116407746, "grad_norm": 12.5, "learning_rate": 8.748365548607163e-06, "loss": 0.6631, "step": 35726 }, { "epoch": 3.801553522025963, "grad_norm": 12.5546875, "learning_rate": 8.748330017055145e-06, "loss": 0.7721, "step": 35727 }, { "epoch": 3.8016599276441796, "grad_norm": 15.2109375, "learning_rate": 8.748294485503127e-06, "loss": 0.7168, "step": 35728 }, { "epoch": 3.801766333262396, "grad_norm": 12.796875, "learning_rate": 8.74825895395111e-06, "loss": 0.6838, "step": 35729 }, { "epoch": 3.801872738880613, "grad_norm": 18.25, "learning_rate": 8.748223422399091e-06, "loss": 0.8904, "step": 35730 }, { "epoch": 3.8019791444988296, "grad_norm": 13.484375, "learning_rate": 8.748187890847073e-06, "loss": 0.6933, "step": 35731 }, { "epoch": 3.802085550117046, "grad_norm": 13.3125, "learning_rate": 8.748152359295055e-06, "loss": 0.6378, "step": 35732 }, { "epoch": 3.802191955735263, "grad_norm": 14.2421875, "learning_rate": 8.748116827743036e-06, "loss": 0.6533, "step": 35733 }, { "epoch": 3.8022983613534795, "grad_norm": 15.09375, "learning_rate": 8.74808129619102e-06, "loss": 0.6941, "step": 35734 }, { "epoch": 3.802404766971696, "grad_norm": 14.3984375, "learning_rate": 8.748045764639001e-06, "loss": 0.6368, "step": 35735 }, { "epoch": 3.802511172589913, "grad_norm": 18.171875, "learning_rate": 8.748010233086983e-06, "loss": 0.7594, "step": 35736 }, { "epoch": 3.8026175782081295, "grad_norm": 14.4453125, "learning_rate": 8.747974701534964e-06, "loss": 0.6886, "step": 35737 }, { "epoch": 3.802723983826346, "grad_norm": 12.828125, "learning_rate": 8.747939169982946e-06, "loss": 0.7169, "step": 35738 }, { "epoch": 3.8028303894445625, "grad_norm": 10.7734375, "learning_rate": 8.747903638430928e-06, "loss": 0.5148, "step": 35739 }, { "epoch": 3.802936795062779, "grad_norm": 13.0703125, "learning_rate": 8.74786810687891e-06, "loss": 0.6878, "step": 35740 }, { "epoch": 3.803043200680996, "grad_norm": 14.8359375, "learning_rate": 8.74783257532689e-06, "loss": 0.685, "step": 35741 }, { "epoch": 3.8031496062992125, "grad_norm": 12.6953125, "learning_rate": 8.747797043774872e-06, "loss": 0.825, "step": 35742 }, { "epoch": 3.803256011917429, "grad_norm": 14.1640625, "learning_rate": 8.747761512222854e-06, "loss": 0.6622, "step": 35743 }, { "epoch": 3.803362417535646, "grad_norm": 10.2421875, "learning_rate": 8.747725980670836e-06, "loss": 0.6307, "step": 35744 }, { "epoch": 3.8034688231538625, "grad_norm": 14.8984375, "learning_rate": 8.747690449118819e-06, "loss": 0.634, "step": 35745 }, { "epoch": 3.803575228772079, "grad_norm": 13.5625, "learning_rate": 8.7476549175668e-06, "loss": 0.6088, "step": 35746 }, { "epoch": 3.803681634390296, "grad_norm": 16.765625, "learning_rate": 8.747619386014782e-06, "loss": 0.9302, "step": 35747 }, { "epoch": 3.8037880400085124, "grad_norm": 9.578125, "learning_rate": 8.747583854462764e-06, "loss": 0.4542, "step": 35748 }, { "epoch": 3.803894445626729, "grad_norm": 16.40625, "learning_rate": 8.747548322910745e-06, "loss": 0.7751, "step": 35749 }, { "epoch": 3.804000851244946, "grad_norm": 13.078125, "learning_rate": 8.747512791358727e-06, "loss": 0.6358, "step": 35750 }, { "epoch": 3.8041072568631624, "grad_norm": 15.59375, "learning_rate": 8.747477259806709e-06, "loss": 0.676, "step": 35751 }, { "epoch": 3.804213662481379, "grad_norm": 13.0703125, "learning_rate": 8.74744172825469e-06, "loss": 0.644, "step": 35752 }, { "epoch": 3.804320068099596, "grad_norm": 16.46875, "learning_rate": 8.747406196702672e-06, "loss": 0.8964, "step": 35753 }, { "epoch": 3.8044264737178124, "grad_norm": 13.1953125, "learning_rate": 8.747370665150653e-06, "loss": 0.6308, "step": 35754 }, { "epoch": 3.804532879336029, "grad_norm": 10.3671875, "learning_rate": 8.747335133598637e-06, "loss": 0.533, "step": 35755 }, { "epoch": 3.804639284954246, "grad_norm": 14.78125, "learning_rate": 8.747299602046618e-06, "loss": 0.7998, "step": 35756 }, { "epoch": 3.8047456905724624, "grad_norm": 15.0, "learning_rate": 8.7472640704946e-06, "loss": 0.6523, "step": 35757 }, { "epoch": 3.804852096190679, "grad_norm": 17.09375, "learning_rate": 8.747228538942581e-06, "loss": 0.8087, "step": 35758 }, { "epoch": 3.8049585018088954, "grad_norm": 14.8125, "learning_rate": 8.747193007390563e-06, "loss": 0.6577, "step": 35759 }, { "epoch": 3.805064907427112, "grad_norm": 16.71875, "learning_rate": 8.747157475838546e-06, "loss": 0.7195, "step": 35760 }, { "epoch": 3.805171313045329, "grad_norm": 10.5078125, "learning_rate": 8.747121944286528e-06, "loss": 0.5048, "step": 35761 }, { "epoch": 3.8052777186635454, "grad_norm": 12.5234375, "learning_rate": 8.74708641273451e-06, "loss": 0.6414, "step": 35762 }, { "epoch": 3.805384124281762, "grad_norm": 11.828125, "learning_rate": 8.747050881182491e-06, "loss": 0.5179, "step": 35763 }, { "epoch": 3.805490529899979, "grad_norm": 17.6875, "learning_rate": 8.747015349630473e-06, "loss": 0.7919, "step": 35764 }, { "epoch": 3.8055969355181953, "grad_norm": 15.1875, "learning_rate": 8.746979818078454e-06, "loss": 0.7307, "step": 35765 }, { "epoch": 3.805703341136412, "grad_norm": 12.625, "learning_rate": 8.746944286526436e-06, "loss": 0.6331, "step": 35766 }, { "epoch": 3.805809746754629, "grad_norm": 15.2109375, "learning_rate": 8.746908754974418e-06, "loss": 0.6531, "step": 35767 }, { "epoch": 3.8059161523728453, "grad_norm": 13.0, "learning_rate": 8.7468732234224e-06, "loss": 0.6244, "step": 35768 }, { "epoch": 3.806022557991062, "grad_norm": 11.9140625, "learning_rate": 8.746837691870381e-06, "loss": 0.6494, "step": 35769 }, { "epoch": 3.8061289636092788, "grad_norm": 13.625, "learning_rate": 8.746802160318364e-06, "loss": 0.8482, "step": 35770 }, { "epoch": 3.8062353692274953, "grad_norm": 12.3046875, "learning_rate": 8.746766628766346e-06, "loss": 0.6663, "step": 35771 }, { "epoch": 3.806341774845712, "grad_norm": 11.3125, "learning_rate": 8.746731097214327e-06, "loss": 0.6217, "step": 35772 }, { "epoch": 3.8064481804639287, "grad_norm": 14.2578125, "learning_rate": 8.746695565662309e-06, "loss": 0.6687, "step": 35773 }, { "epoch": 3.8065545860821453, "grad_norm": 13.6015625, "learning_rate": 8.74666003411029e-06, "loss": 0.6455, "step": 35774 }, { "epoch": 3.8066609917003618, "grad_norm": 13.5859375, "learning_rate": 8.746624502558272e-06, "loss": 0.7555, "step": 35775 }, { "epoch": 3.8067673973185783, "grad_norm": 15.0390625, "learning_rate": 8.746588971006254e-06, "loss": 0.7869, "step": 35776 }, { "epoch": 3.8068738029367952, "grad_norm": 12.8671875, "learning_rate": 8.746553439454235e-06, "loss": 0.6905, "step": 35777 }, { "epoch": 3.8069802085550117, "grad_norm": 17.90625, "learning_rate": 8.746517907902217e-06, "loss": 0.8206, "step": 35778 }, { "epoch": 3.8070866141732282, "grad_norm": 12.9765625, "learning_rate": 8.746482376350199e-06, "loss": 0.6586, "step": 35779 }, { "epoch": 3.8071930197914448, "grad_norm": 13.1796875, "learning_rate": 8.74644684479818e-06, "loss": 0.6659, "step": 35780 }, { "epoch": 3.8072994254096617, "grad_norm": 14.234375, "learning_rate": 8.746411313246164e-06, "loss": 0.6771, "step": 35781 }, { "epoch": 3.807405831027878, "grad_norm": 10.2421875, "learning_rate": 8.746375781694145e-06, "loss": 0.6016, "step": 35782 }, { "epoch": 3.8075122366460947, "grad_norm": 12.9765625, "learning_rate": 8.746340250142127e-06, "loss": 0.6212, "step": 35783 }, { "epoch": 3.8076186422643117, "grad_norm": 19.546875, "learning_rate": 8.746304718590108e-06, "loss": 0.7709, "step": 35784 }, { "epoch": 3.807725047882528, "grad_norm": 15.671875, "learning_rate": 8.74626918703809e-06, "loss": 0.7613, "step": 35785 }, { "epoch": 3.8078314535007447, "grad_norm": 12.8125, "learning_rate": 8.746233655486073e-06, "loss": 0.6473, "step": 35786 }, { "epoch": 3.8079378591189617, "grad_norm": 11.71875, "learning_rate": 8.746198123934055e-06, "loss": 0.5405, "step": 35787 }, { "epoch": 3.808044264737178, "grad_norm": 12.6171875, "learning_rate": 8.746162592382036e-06, "loss": 0.718, "step": 35788 }, { "epoch": 3.8081506703553947, "grad_norm": 11.90625, "learning_rate": 8.746127060830018e-06, "loss": 0.6104, "step": 35789 }, { "epoch": 3.8082570759736116, "grad_norm": 15.890625, "learning_rate": 8.746091529278e-06, "loss": 0.692, "step": 35790 }, { "epoch": 3.808363481591828, "grad_norm": 16.15625, "learning_rate": 8.746055997725981e-06, "loss": 0.5765, "step": 35791 }, { "epoch": 3.8084698872100446, "grad_norm": 10.7734375, "learning_rate": 8.746020466173963e-06, "loss": 0.6089, "step": 35792 }, { "epoch": 3.8085762928282616, "grad_norm": 12.9296875, "learning_rate": 8.745984934621945e-06, "loss": 0.7318, "step": 35793 }, { "epoch": 3.808682698446478, "grad_norm": 14.6953125, "learning_rate": 8.745949403069926e-06, "loss": 0.6621, "step": 35794 }, { "epoch": 3.8087891040646946, "grad_norm": 13.0546875, "learning_rate": 8.745913871517908e-06, "loss": 0.681, "step": 35795 }, { "epoch": 3.808895509682911, "grad_norm": 9.1015625, "learning_rate": 8.745878339965891e-06, "loss": 0.4717, "step": 35796 }, { "epoch": 3.8090019153011276, "grad_norm": 11.3984375, "learning_rate": 8.745842808413873e-06, "loss": 0.6379, "step": 35797 }, { "epoch": 3.8091083209193446, "grad_norm": 12.1953125, "learning_rate": 8.745807276861854e-06, "loss": 0.6486, "step": 35798 }, { "epoch": 3.809214726537561, "grad_norm": 11.5625, "learning_rate": 8.745771745309836e-06, "loss": 0.5624, "step": 35799 }, { "epoch": 3.8093211321557776, "grad_norm": 14.625, "learning_rate": 8.745736213757817e-06, "loss": 0.7605, "step": 35800 }, { "epoch": 3.8094275377739946, "grad_norm": 13.0078125, "learning_rate": 8.7457006822058e-06, "loss": 0.5623, "step": 35801 }, { "epoch": 3.809533943392211, "grad_norm": 18.734375, "learning_rate": 8.745665150653782e-06, "loss": 0.7129, "step": 35802 }, { "epoch": 3.8096403490104276, "grad_norm": 13.625, "learning_rate": 8.745629619101764e-06, "loss": 0.7057, "step": 35803 }, { "epoch": 3.8097467546286445, "grad_norm": 12.7265625, "learning_rate": 8.745594087549744e-06, "loss": 0.6158, "step": 35804 }, { "epoch": 3.809853160246861, "grad_norm": 19.109375, "learning_rate": 8.745558555997725e-06, "loss": 0.8966, "step": 35805 }, { "epoch": 3.8099595658650776, "grad_norm": 12.96875, "learning_rate": 8.745523024445709e-06, "loss": 0.72, "step": 35806 }, { "epoch": 3.8100659714832945, "grad_norm": 15.8828125, "learning_rate": 8.74548749289369e-06, "loss": 0.8173, "step": 35807 }, { "epoch": 3.810172377101511, "grad_norm": 14.3046875, "learning_rate": 8.745451961341672e-06, "loss": 0.695, "step": 35808 }, { "epoch": 3.8102787827197275, "grad_norm": 10.6171875, "learning_rate": 8.745416429789654e-06, "loss": 0.6208, "step": 35809 }, { "epoch": 3.8103851883379445, "grad_norm": 15.1640625, "learning_rate": 8.745380898237635e-06, "loss": 0.7339, "step": 35810 }, { "epoch": 3.810491593956161, "grad_norm": 13.8203125, "learning_rate": 8.745345366685619e-06, "loss": 0.7973, "step": 35811 }, { "epoch": 3.8105979995743775, "grad_norm": 10.96875, "learning_rate": 8.7453098351336e-06, "loss": 0.5059, "step": 35812 }, { "epoch": 3.810704405192594, "grad_norm": 12.765625, "learning_rate": 8.745274303581582e-06, "loss": 0.7541, "step": 35813 }, { "epoch": 3.810810810810811, "grad_norm": 17.921875, "learning_rate": 8.745238772029563e-06, "loss": 0.7524, "step": 35814 }, { "epoch": 3.8109172164290275, "grad_norm": 18.421875, "learning_rate": 8.745203240477545e-06, "loss": 0.6902, "step": 35815 }, { "epoch": 3.811023622047244, "grad_norm": 16.53125, "learning_rate": 8.745167708925527e-06, "loss": 0.6319, "step": 35816 }, { "epoch": 3.8111300276654605, "grad_norm": 11.09375, "learning_rate": 8.745132177373508e-06, "loss": 0.6606, "step": 35817 }, { "epoch": 3.8112364332836774, "grad_norm": 15.4609375, "learning_rate": 8.74509664582149e-06, "loss": 0.778, "step": 35818 }, { "epoch": 3.811342838901894, "grad_norm": 13.765625, "learning_rate": 8.745061114269471e-06, "loss": 0.7814, "step": 35819 }, { "epoch": 3.8114492445201105, "grad_norm": 10.9609375, "learning_rate": 8.745025582717453e-06, "loss": 0.5294, "step": 35820 }, { "epoch": 3.8115556501383274, "grad_norm": 12.3828125, "learning_rate": 8.744990051165435e-06, "loss": 0.5593, "step": 35821 }, { "epoch": 3.811662055756544, "grad_norm": 22.875, "learning_rate": 8.744954519613418e-06, "loss": 0.8596, "step": 35822 }, { "epoch": 3.8117684613747604, "grad_norm": 15.28125, "learning_rate": 8.7449189880614e-06, "loss": 0.8458, "step": 35823 }, { "epoch": 3.8118748669929774, "grad_norm": 15.3359375, "learning_rate": 8.744883456509381e-06, "loss": 0.7065, "step": 35824 }, { "epoch": 3.811981272611194, "grad_norm": 20.859375, "learning_rate": 8.744847924957363e-06, "loss": 0.7594, "step": 35825 }, { "epoch": 3.8120876782294104, "grad_norm": 15.5625, "learning_rate": 8.744812393405344e-06, "loss": 0.969, "step": 35826 }, { "epoch": 3.8121940838476274, "grad_norm": 13.3828125, "learning_rate": 8.744776861853328e-06, "loss": 0.5652, "step": 35827 }, { "epoch": 3.812300489465844, "grad_norm": 14.328125, "learning_rate": 8.74474133030131e-06, "loss": 0.7191, "step": 35828 }, { "epoch": 3.8124068950840604, "grad_norm": 16.578125, "learning_rate": 8.74470579874929e-06, "loss": 0.8039, "step": 35829 }, { "epoch": 3.8125133007022773, "grad_norm": 12.84375, "learning_rate": 8.744670267197272e-06, "loss": 0.7699, "step": 35830 }, { "epoch": 3.812619706320494, "grad_norm": 13.7265625, "learning_rate": 8.744634735645254e-06, "loss": 0.9178, "step": 35831 }, { "epoch": 3.8127261119387104, "grad_norm": 11.796875, "learning_rate": 8.744599204093236e-06, "loss": 0.6752, "step": 35832 }, { "epoch": 3.812832517556927, "grad_norm": 17.515625, "learning_rate": 8.744563672541217e-06, "loss": 0.658, "step": 35833 }, { "epoch": 3.8129389231751434, "grad_norm": 11.8984375, "learning_rate": 8.744528140989199e-06, "loss": 0.6185, "step": 35834 }, { "epoch": 3.8130453287933603, "grad_norm": 13.6484375, "learning_rate": 8.74449260943718e-06, "loss": 0.6321, "step": 35835 }, { "epoch": 3.813151734411577, "grad_norm": 14.2734375, "learning_rate": 8.744457077885162e-06, "loss": 0.6366, "step": 35836 }, { "epoch": 3.8132581400297934, "grad_norm": 15.8671875, "learning_rate": 8.744421546333145e-06, "loss": 0.708, "step": 35837 }, { "epoch": 3.8133645456480103, "grad_norm": 15.53125, "learning_rate": 8.744386014781127e-06, "loss": 0.6849, "step": 35838 }, { "epoch": 3.813470951266227, "grad_norm": 12.71875, "learning_rate": 8.744350483229109e-06, "loss": 0.6166, "step": 35839 }, { "epoch": 3.8135773568844433, "grad_norm": 15.0859375, "learning_rate": 8.74431495167709e-06, "loss": 0.7898, "step": 35840 }, { "epoch": 3.8136837625026603, "grad_norm": 12.8671875, "learning_rate": 8.744279420125072e-06, "loss": 0.6759, "step": 35841 }, { "epoch": 3.813790168120877, "grad_norm": 14.78125, "learning_rate": 8.744243888573053e-06, "loss": 0.5861, "step": 35842 }, { "epoch": 3.8138965737390933, "grad_norm": 13.78125, "learning_rate": 8.744208357021035e-06, "loss": 0.6762, "step": 35843 }, { "epoch": 3.8140029793573103, "grad_norm": 16.515625, "learning_rate": 8.744172825469017e-06, "loss": 0.6953, "step": 35844 }, { "epoch": 3.8141093849755268, "grad_norm": 18.21875, "learning_rate": 8.744137293916998e-06, "loss": 0.7741, "step": 35845 }, { "epoch": 3.8142157905937433, "grad_norm": 10.6328125, "learning_rate": 8.74410176236498e-06, "loss": 0.5536, "step": 35846 }, { "epoch": 3.8143221962119602, "grad_norm": 12.8828125, "learning_rate": 8.744066230812963e-06, "loss": 0.621, "step": 35847 }, { "epoch": 3.8144286018301767, "grad_norm": 15.8515625, "learning_rate": 8.744030699260945e-06, "loss": 0.7227, "step": 35848 }, { "epoch": 3.8145350074483932, "grad_norm": 15.6796875, "learning_rate": 8.743995167708926e-06, "loss": 0.5653, "step": 35849 }, { "epoch": 3.8146414130666098, "grad_norm": 12.9453125, "learning_rate": 8.743959636156908e-06, "loss": 0.6034, "step": 35850 }, { "epoch": 3.8147478186848267, "grad_norm": 11.4765625, "learning_rate": 8.74392410460489e-06, "loss": 0.5076, "step": 35851 }, { "epoch": 3.814854224303043, "grad_norm": 13.0234375, "learning_rate": 8.743888573052873e-06, "loss": 0.5462, "step": 35852 }, { "epoch": 3.8149606299212597, "grad_norm": 15.8671875, "learning_rate": 8.743853041500854e-06, "loss": 0.7413, "step": 35853 }, { "epoch": 3.8150670355394762, "grad_norm": 22.53125, "learning_rate": 8.743817509948836e-06, "loss": 0.9248, "step": 35854 }, { "epoch": 3.815173441157693, "grad_norm": 12.4453125, "learning_rate": 8.743781978396818e-06, "loss": 0.6568, "step": 35855 }, { "epoch": 3.8152798467759097, "grad_norm": 16.453125, "learning_rate": 8.7437464468448e-06, "loss": 0.6416, "step": 35856 }, { "epoch": 3.815386252394126, "grad_norm": 11.21875, "learning_rate": 8.743710915292781e-06, "loss": 0.7494, "step": 35857 }, { "epoch": 3.815492658012343, "grad_norm": 18.96875, "learning_rate": 8.743675383740762e-06, "loss": 0.8757, "step": 35858 }, { "epoch": 3.8155990636305597, "grad_norm": 11.2421875, "learning_rate": 8.743639852188744e-06, "loss": 0.6153, "step": 35859 }, { "epoch": 3.815705469248776, "grad_norm": 12.9453125, "learning_rate": 8.743604320636726e-06, "loss": 0.6094, "step": 35860 }, { "epoch": 3.815811874866993, "grad_norm": 18.46875, "learning_rate": 8.743568789084707e-06, "loss": 0.7891, "step": 35861 }, { "epoch": 3.8159182804852096, "grad_norm": 13.234375, "learning_rate": 8.743533257532689e-06, "loss": 0.5814, "step": 35862 }, { "epoch": 3.816024686103426, "grad_norm": 14.78125, "learning_rate": 8.743497725980672e-06, "loss": 0.708, "step": 35863 }, { "epoch": 3.816131091721643, "grad_norm": 10.890625, "learning_rate": 8.743462194428654e-06, "loss": 0.4904, "step": 35864 }, { "epoch": 3.8162374973398596, "grad_norm": 12.328125, "learning_rate": 8.743426662876635e-06, "loss": 0.5518, "step": 35865 }, { "epoch": 3.816343902958076, "grad_norm": 14.71875, "learning_rate": 8.743391131324617e-06, "loss": 0.7071, "step": 35866 }, { "epoch": 3.816450308576293, "grad_norm": 11.4921875, "learning_rate": 8.743355599772599e-06, "loss": 0.6199, "step": 35867 }, { "epoch": 3.8165567141945096, "grad_norm": 11.703125, "learning_rate": 8.74332006822058e-06, "loss": 0.7052, "step": 35868 }, { "epoch": 3.816663119812726, "grad_norm": 16.109375, "learning_rate": 8.743284536668562e-06, "loss": 0.6584, "step": 35869 }, { "epoch": 3.8167695254309426, "grad_norm": 11.8984375, "learning_rate": 8.743249005116543e-06, "loss": 0.6042, "step": 35870 }, { "epoch": 3.816875931049159, "grad_norm": 15.3515625, "learning_rate": 8.743213473564525e-06, "loss": 0.6387, "step": 35871 }, { "epoch": 3.816982336667376, "grad_norm": 13.328125, "learning_rate": 8.743177942012507e-06, "loss": 0.7089, "step": 35872 }, { "epoch": 3.8170887422855926, "grad_norm": 14.6640625, "learning_rate": 8.74314241046049e-06, "loss": 0.6906, "step": 35873 }, { "epoch": 3.817195147903809, "grad_norm": 9.7109375, "learning_rate": 8.743106878908472e-06, "loss": 0.4919, "step": 35874 }, { "epoch": 3.817301553522026, "grad_norm": 13.75, "learning_rate": 8.743071347356453e-06, "loss": 0.7033, "step": 35875 }, { "epoch": 3.8174079591402426, "grad_norm": 11.046875, "learning_rate": 8.743035815804435e-06, "loss": 0.5199, "step": 35876 }, { "epoch": 3.817514364758459, "grad_norm": 11.9453125, "learning_rate": 8.743000284252416e-06, "loss": 0.6546, "step": 35877 }, { "epoch": 3.817620770376676, "grad_norm": 22.84375, "learning_rate": 8.7429647527004e-06, "loss": 0.8303, "step": 35878 }, { "epoch": 3.8177271759948925, "grad_norm": 23.09375, "learning_rate": 8.742929221148381e-06, "loss": 1.0366, "step": 35879 }, { "epoch": 3.817833581613109, "grad_norm": 11.0703125, "learning_rate": 8.742893689596363e-06, "loss": 0.5517, "step": 35880 }, { "epoch": 3.817939987231326, "grad_norm": 13.7890625, "learning_rate": 8.742858158044345e-06, "loss": 0.6687, "step": 35881 }, { "epoch": 3.8180463928495425, "grad_norm": 10.4921875, "learning_rate": 8.742822626492326e-06, "loss": 0.5269, "step": 35882 }, { "epoch": 3.818152798467759, "grad_norm": 10.453125, "learning_rate": 8.742787094940308e-06, "loss": 0.6069, "step": 35883 }, { "epoch": 3.818259204085976, "grad_norm": 12.3359375, "learning_rate": 8.74275156338829e-06, "loss": 0.6825, "step": 35884 }, { "epoch": 3.8183656097041925, "grad_norm": 15.625, "learning_rate": 8.742716031836271e-06, "loss": 0.7996, "step": 35885 }, { "epoch": 3.818472015322409, "grad_norm": 12.703125, "learning_rate": 8.742680500284253e-06, "loss": 0.6158, "step": 35886 }, { "epoch": 3.818578420940626, "grad_norm": 11.125, "learning_rate": 8.742644968732234e-06, "loss": 0.5363, "step": 35887 }, { "epoch": 3.8186848265588424, "grad_norm": 12.9375, "learning_rate": 8.742609437180217e-06, "loss": 0.6745, "step": 35888 }, { "epoch": 3.818791232177059, "grad_norm": 11.6953125, "learning_rate": 8.742573905628199e-06, "loss": 0.647, "step": 35889 }, { "epoch": 3.8188976377952755, "grad_norm": 13.90625, "learning_rate": 8.74253837407618e-06, "loss": 0.6562, "step": 35890 }, { "epoch": 3.819004043413492, "grad_norm": 21.671875, "learning_rate": 8.742502842524162e-06, "loss": 0.7523, "step": 35891 }, { "epoch": 3.819110449031709, "grad_norm": 13.078125, "learning_rate": 8.742467310972144e-06, "loss": 0.6305, "step": 35892 }, { "epoch": 3.8192168546499254, "grad_norm": 12.2578125, "learning_rate": 8.742431779420125e-06, "loss": 0.5664, "step": 35893 }, { "epoch": 3.819323260268142, "grad_norm": 13.6328125, "learning_rate": 8.742396247868107e-06, "loss": 0.684, "step": 35894 }, { "epoch": 3.819429665886359, "grad_norm": 15.71875, "learning_rate": 8.742360716316089e-06, "loss": 0.7385, "step": 35895 }, { "epoch": 3.8195360715045754, "grad_norm": 17.484375, "learning_rate": 8.74232518476407e-06, "loss": 0.7093, "step": 35896 }, { "epoch": 3.819642477122792, "grad_norm": 13.125, "learning_rate": 8.742289653212052e-06, "loss": 0.7799, "step": 35897 }, { "epoch": 3.819748882741009, "grad_norm": 14.1328125, "learning_rate": 8.742254121660034e-06, "loss": 0.666, "step": 35898 }, { "epoch": 3.8198552883592254, "grad_norm": 12.5234375, "learning_rate": 8.742218590108017e-06, "loss": 0.588, "step": 35899 }, { "epoch": 3.819961693977442, "grad_norm": 13.9921875, "learning_rate": 8.742183058555998e-06, "loss": 0.6817, "step": 35900 }, { "epoch": 3.820068099595659, "grad_norm": 12.0625, "learning_rate": 8.74214752700398e-06, "loss": 0.5415, "step": 35901 }, { "epoch": 3.8201745052138754, "grad_norm": 16.859375, "learning_rate": 8.742111995451962e-06, "loss": 0.7618, "step": 35902 }, { "epoch": 3.820280910832092, "grad_norm": 10.59375, "learning_rate": 8.742076463899943e-06, "loss": 0.7274, "step": 35903 }, { "epoch": 3.820387316450309, "grad_norm": 12.3125, "learning_rate": 8.742040932347927e-06, "loss": 0.6129, "step": 35904 }, { "epoch": 3.8204937220685253, "grad_norm": 12.28125, "learning_rate": 8.742005400795908e-06, "loss": 0.6157, "step": 35905 }, { "epoch": 3.820600127686742, "grad_norm": 15.3359375, "learning_rate": 8.74196986924389e-06, "loss": 0.7916, "step": 35906 }, { "epoch": 3.8207065333049584, "grad_norm": 12.78125, "learning_rate": 8.741934337691871e-06, "loss": 0.6768, "step": 35907 }, { "epoch": 3.8208129389231753, "grad_norm": 10.5859375, "learning_rate": 8.741898806139853e-06, "loss": 0.6314, "step": 35908 }, { "epoch": 3.820919344541392, "grad_norm": 10.2578125, "learning_rate": 8.741863274587835e-06, "loss": 0.7099, "step": 35909 }, { "epoch": 3.8210257501596083, "grad_norm": 14.203125, "learning_rate": 8.741827743035816e-06, "loss": 0.7017, "step": 35910 }, { "epoch": 3.821132155777825, "grad_norm": 13.625, "learning_rate": 8.741792211483798e-06, "loss": 0.6199, "step": 35911 }, { "epoch": 3.821238561396042, "grad_norm": 14.3359375, "learning_rate": 8.74175667993178e-06, "loss": 0.6713, "step": 35912 }, { "epoch": 3.8213449670142583, "grad_norm": 13.7265625, "learning_rate": 8.741721148379761e-06, "loss": 0.6333, "step": 35913 }, { "epoch": 3.821451372632475, "grad_norm": 14.9765625, "learning_rate": 8.741685616827744e-06, "loss": 0.8016, "step": 35914 }, { "epoch": 3.8215577782506918, "grad_norm": 13.703125, "learning_rate": 8.741650085275726e-06, "loss": 0.7259, "step": 35915 }, { "epoch": 3.8216641838689083, "grad_norm": 11.0859375, "learning_rate": 8.741614553723708e-06, "loss": 0.724, "step": 35916 }, { "epoch": 3.821770589487125, "grad_norm": 14.2109375, "learning_rate": 8.741579022171689e-06, "loss": 0.6571, "step": 35917 }, { "epoch": 3.8218769951053417, "grad_norm": 13.34375, "learning_rate": 8.74154349061967e-06, "loss": 0.6547, "step": 35918 }, { "epoch": 3.8219834007235582, "grad_norm": 15.640625, "learning_rate": 8.741507959067654e-06, "loss": 0.6806, "step": 35919 }, { "epoch": 3.8220898063417748, "grad_norm": 13.5625, "learning_rate": 8.741472427515636e-06, "loss": 0.6035, "step": 35920 }, { "epoch": 3.8221962119599917, "grad_norm": 16.703125, "learning_rate": 8.741436895963617e-06, "loss": 0.7436, "step": 35921 }, { "epoch": 3.822302617578208, "grad_norm": 16.265625, "learning_rate": 8.741401364411599e-06, "loss": 0.7434, "step": 35922 }, { "epoch": 3.8224090231964247, "grad_norm": 13.2109375, "learning_rate": 8.741365832859579e-06, "loss": 0.6865, "step": 35923 }, { "epoch": 3.8225154288146417, "grad_norm": 12.9609375, "learning_rate": 8.741330301307562e-06, "loss": 0.6992, "step": 35924 }, { "epoch": 3.822621834432858, "grad_norm": 11.53125, "learning_rate": 8.741294769755544e-06, "loss": 0.5865, "step": 35925 }, { "epoch": 3.8227282400510747, "grad_norm": 11.703125, "learning_rate": 8.741259238203525e-06, "loss": 0.6941, "step": 35926 }, { "epoch": 3.822834645669291, "grad_norm": 9.7109375, "learning_rate": 8.741223706651507e-06, "loss": 0.6356, "step": 35927 }, { "epoch": 3.8229410512875077, "grad_norm": 15.109375, "learning_rate": 8.741188175099489e-06, "loss": 0.6616, "step": 35928 }, { "epoch": 3.8230474569057247, "grad_norm": 12.171875, "learning_rate": 8.741152643547472e-06, "loss": 0.6544, "step": 35929 }, { "epoch": 3.823153862523941, "grad_norm": 14.1171875, "learning_rate": 8.741117111995453e-06, "loss": 0.593, "step": 35930 }, { "epoch": 3.8232602681421577, "grad_norm": 13.515625, "learning_rate": 8.741081580443435e-06, "loss": 0.657, "step": 35931 }, { "epoch": 3.8233666737603746, "grad_norm": 13.390625, "learning_rate": 8.741046048891417e-06, "loss": 0.6518, "step": 35932 }, { "epoch": 3.823473079378591, "grad_norm": 13.671875, "learning_rate": 8.741010517339398e-06, "loss": 0.6392, "step": 35933 }, { "epoch": 3.8235794849968077, "grad_norm": 10.4609375, "learning_rate": 8.74097498578738e-06, "loss": 0.6034, "step": 35934 }, { "epoch": 3.8236858906150246, "grad_norm": 12.8515625, "learning_rate": 8.740939454235361e-06, "loss": 0.647, "step": 35935 }, { "epoch": 3.823792296233241, "grad_norm": 14.0234375, "learning_rate": 8.740903922683343e-06, "loss": 0.6587, "step": 35936 }, { "epoch": 3.8238987018514576, "grad_norm": 13.78125, "learning_rate": 8.740868391131325e-06, "loss": 0.6206, "step": 35937 }, { "epoch": 3.8240051074696746, "grad_norm": 16.40625, "learning_rate": 8.740832859579306e-06, "loss": 0.6727, "step": 35938 }, { "epoch": 3.824111513087891, "grad_norm": 9.5, "learning_rate": 8.740797328027288e-06, "loss": 0.5781, "step": 35939 }, { "epoch": 3.8242179187061076, "grad_norm": 9.5078125, "learning_rate": 8.740761796475271e-06, "loss": 0.4113, "step": 35940 }, { "epoch": 3.8243243243243246, "grad_norm": 13.5, "learning_rate": 8.740726264923253e-06, "loss": 0.5879, "step": 35941 }, { "epoch": 3.824430729942541, "grad_norm": 14.2734375, "learning_rate": 8.740690733371234e-06, "loss": 0.8049, "step": 35942 }, { "epoch": 3.8245371355607576, "grad_norm": 17.015625, "learning_rate": 8.740655201819216e-06, "loss": 0.7613, "step": 35943 }, { "epoch": 3.824643541178974, "grad_norm": 11.4921875, "learning_rate": 8.740619670267198e-06, "loss": 0.6961, "step": 35944 }, { "epoch": 3.824749946797191, "grad_norm": 11.3515625, "learning_rate": 8.740584138715181e-06, "loss": 0.6391, "step": 35945 }, { "epoch": 3.8248563524154076, "grad_norm": 13.1171875, "learning_rate": 8.740548607163163e-06, "loss": 0.5729, "step": 35946 }, { "epoch": 3.824962758033624, "grad_norm": 14.9375, "learning_rate": 8.740513075611144e-06, "loss": 0.7477, "step": 35947 }, { "epoch": 3.8250691636518406, "grad_norm": 12.7578125, "learning_rate": 8.740477544059126e-06, "loss": 0.6491, "step": 35948 }, { "epoch": 3.8251755692700575, "grad_norm": 14.765625, "learning_rate": 8.740442012507107e-06, "loss": 0.775, "step": 35949 }, { "epoch": 3.825281974888274, "grad_norm": 13.765625, "learning_rate": 8.740406480955089e-06, "loss": 0.7179, "step": 35950 }, { "epoch": 3.8253883805064905, "grad_norm": 15.328125, "learning_rate": 8.74037094940307e-06, "loss": 0.6143, "step": 35951 }, { "epoch": 3.8254947861247075, "grad_norm": 8.6953125, "learning_rate": 8.740335417851052e-06, "loss": 0.6419, "step": 35952 }, { "epoch": 3.825601191742924, "grad_norm": 11.40625, "learning_rate": 8.740299886299034e-06, "loss": 0.7262, "step": 35953 }, { "epoch": 3.8257075973611405, "grad_norm": 12.5390625, "learning_rate": 8.740264354747015e-06, "loss": 0.727, "step": 35954 }, { "epoch": 3.8258140029793575, "grad_norm": 13.25, "learning_rate": 8.740228823194999e-06, "loss": 0.5917, "step": 35955 }, { "epoch": 3.825920408597574, "grad_norm": 14.09375, "learning_rate": 8.74019329164298e-06, "loss": 0.7566, "step": 35956 }, { "epoch": 3.8260268142157905, "grad_norm": 19.390625, "learning_rate": 8.740157760090962e-06, "loss": 0.6676, "step": 35957 }, { "epoch": 3.8261332198340074, "grad_norm": 11.9453125, "learning_rate": 8.740122228538943e-06, "loss": 0.5226, "step": 35958 }, { "epoch": 3.826239625452224, "grad_norm": 13.3671875, "learning_rate": 8.740086696986925e-06, "loss": 0.6767, "step": 35959 }, { "epoch": 3.8263460310704405, "grad_norm": 14.4140625, "learning_rate": 8.740051165434907e-06, "loss": 0.7791, "step": 35960 }, { "epoch": 3.8264524366886574, "grad_norm": 12.28125, "learning_rate": 8.740015633882888e-06, "loss": 0.7232, "step": 35961 }, { "epoch": 3.826558842306874, "grad_norm": 15.4765625, "learning_rate": 8.73998010233087e-06, "loss": 0.7602, "step": 35962 }, { "epoch": 3.8266652479250904, "grad_norm": 11.6796875, "learning_rate": 8.739944570778852e-06, "loss": 0.6294, "step": 35963 }, { "epoch": 3.826771653543307, "grad_norm": 11.0390625, "learning_rate": 8.739909039226833e-06, "loss": 0.7335, "step": 35964 }, { "epoch": 3.8268780591615235, "grad_norm": 14.515625, "learning_rate": 8.739873507674816e-06, "loss": 0.8403, "step": 35965 }, { "epoch": 3.8269844647797404, "grad_norm": 13.5625, "learning_rate": 8.739837976122798e-06, "loss": 0.6691, "step": 35966 }, { "epoch": 3.827090870397957, "grad_norm": 18.078125, "learning_rate": 8.73980244457078e-06, "loss": 0.7682, "step": 35967 }, { "epoch": 3.8271972760161734, "grad_norm": 17.265625, "learning_rate": 8.739766913018761e-06, "loss": 0.8639, "step": 35968 }, { "epoch": 3.8273036816343904, "grad_norm": 9.1328125, "learning_rate": 8.739731381466743e-06, "loss": 0.706, "step": 35969 }, { "epoch": 3.827410087252607, "grad_norm": 13.21875, "learning_rate": 8.739695849914726e-06, "loss": 0.6761, "step": 35970 }, { "epoch": 3.8275164928708234, "grad_norm": 15.015625, "learning_rate": 8.739660318362708e-06, "loss": 0.7678, "step": 35971 }, { "epoch": 3.8276228984890404, "grad_norm": 10.75, "learning_rate": 8.73962478681069e-06, "loss": 0.634, "step": 35972 }, { "epoch": 3.827729304107257, "grad_norm": 13.203125, "learning_rate": 8.739589255258671e-06, "loss": 0.6171, "step": 35973 }, { "epoch": 3.8278357097254734, "grad_norm": 13.078125, "learning_rate": 8.739553723706653e-06, "loss": 0.5849, "step": 35974 }, { "epoch": 3.8279421153436903, "grad_norm": 13.65625, "learning_rate": 8.739518192154634e-06, "loss": 0.7117, "step": 35975 }, { "epoch": 3.828048520961907, "grad_norm": 12.328125, "learning_rate": 8.739482660602616e-06, "loss": 0.774, "step": 35976 }, { "epoch": 3.8281549265801234, "grad_norm": 15.9453125, "learning_rate": 8.739447129050597e-06, "loss": 0.6354, "step": 35977 }, { "epoch": 3.8282613321983403, "grad_norm": 12.5078125, "learning_rate": 8.739411597498579e-06, "loss": 0.5655, "step": 35978 }, { "epoch": 3.828367737816557, "grad_norm": 15.75, "learning_rate": 8.73937606594656e-06, "loss": 0.7088, "step": 35979 }, { "epoch": 3.8284741434347733, "grad_norm": 7.703125, "learning_rate": 8.739340534394542e-06, "loss": 0.3158, "step": 35980 }, { "epoch": 3.82858054905299, "grad_norm": 13.5390625, "learning_rate": 8.739305002842526e-06, "loss": 0.7524, "step": 35981 }, { "epoch": 3.828686954671207, "grad_norm": 15.09375, "learning_rate": 8.739269471290507e-06, "loss": 0.7415, "step": 35982 }, { "epoch": 3.8287933602894233, "grad_norm": 14.7265625, "learning_rate": 8.739233939738489e-06, "loss": 0.7735, "step": 35983 }, { "epoch": 3.82889976590764, "grad_norm": 14.8828125, "learning_rate": 8.73919840818647e-06, "loss": 0.8592, "step": 35984 }, { "epoch": 3.8290061715258563, "grad_norm": 13.4765625, "learning_rate": 8.739162876634452e-06, "loss": 0.7531, "step": 35985 }, { "epoch": 3.8291125771440733, "grad_norm": 12.1015625, "learning_rate": 8.739127345082434e-06, "loss": 0.618, "step": 35986 }, { "epoch": 3.82921898276229, "grad_norm": 20.546875, "learning_rate": 8.739091813530415e-06, "loss": 0.8305, "step": 35987 }, { "epoch": 3.8293253883805063, "grad_norm": 14.90625, "learning_rate": 8.739056281978397e-06, "loss": 0.7311, "step": 35988 }, { "epoch": 3.8294317939987232, "grad_norm": 12.8203125, "learning_rate": 8.739020750426378e-06, "loss": 0.5772, "step": 35989 }, { "epoch": 3.8295381996169398, "grad_norm": 12.65625, "learning_rate": 8.73898521887436e-06, "loss": 0.6668, "step": 35990 }, { "epoch": 3.8296446052351563, "grad_norm": 16.40625, "learning_rate": 8.738949687322343e-06, "loss": 0.7472, "step": 35991 }, { "epoch": 3.829751010853373, "grad_norm": 12.953125, "learning_rate": 8.738914155770325e-06, "loss": 0.7019, "step": 35992 }, { "epoch": 3.8298574164715897, "grad_norm": 11.953125, "learning_rate": 8.738878624218306e-06, "loss": 0.4842, "step": 35993 }, { "epoch": 3.8299638220898062, "grad_norm": 12.0625, "learning_rate": 8.738843092666288e-06, "loss": 0.6764, "step": 35994 }, { "epoch": 3.830070227708023, "grad_norm": 14.3515625, "learning_rate": 8.73880756111427e-06, "loss": 0.5552, "step": 35995 }, { "epoch": 3.8301766333262397, "grad_norm": 15.875, "learning_rate": 8.738772029562253e-06, "loss": 0.7728, "step": 35996 }, { "epoch": 3.830283038944456, "grad_norm": 17.625, "learning_rate": 8.738736498010235e-06, "loss": 0.6936, "step": 35997 }, { "epoch": 3.830389444562673, "grad_norm": 12.859375, "learning_rate": 8.738700966458216e-06, "loss": 0.5745, "step": 35998 }, { "epoch": 3.8304958501808897, "grad_norm": 21.640625, "learning_rate": 8.738665434906198e-06, "loss": 0.8305, "step": 35999 }, { "epoch": 3.830602255799106, "grad_norm": 14.53125, "learning_rate": 8.73862990335418e-06, "loss": 0.714, "step": 36000 }, { "epoch": 3.8307086614173227, "grad_norm": 12.046875, "learning_rate": 8.738594371802161e-06, "loss": 0.6589, "step": 36001 }, { "epoch": 3.830815067035539, "grad_norm": 13.5703125, "learning_rate": 8.738558840250143e-06, "loss": 0.6407, "step": 36002 }, { "epoch": 3.830921472653756, "grad_norm": 13.7265625, "learning_rate": 8.738523308698124e-06, "loss": 0.8025, "step": 36003 }, { "epoch": 3.8310278782719727, "grad_norm": 15.078125, "learning_rate": 8.738487777146106e-06, "loss": 0.6965, "step": 36004 }, { "epoch": 3.831134283890189, "grad_norm": 14.1328125, "learning_rate": 8.738452245594087e-06, "loss": 0.8064, "step": 36005 }, { "epoch": 3.831240689508406, "grad_norm": 14.1875, "learning_rate": 8.73841671404207e-06, "loss": 0.6431, "step": 36006 }, { "epoch": 3.8313470951266226, "grad_norm": 14.1640625, "learning_rate": 8.738381182490052e-06, "loss": 0.6926, "step": 36007 }, { "epoch": 3.831453500744839, "grad_norm": 12.1875, "learning_rate": 8.738345650938034e-06, "loss": 0.8277, "step": 36008 }, { "epoch": 3.831559906363056, "grad_norm": 13.125, "learning_rate": 8.738310119386016e-06, "loss": 0.6098, "step": 36009 }, { "epoch": 3.8316663119812726, "grad_norm": 12.1796875, "learning_rate": 8.738274587833997e-06, "loss": 0.6536, "step": 36010 }, { "epoch": 3.831772717599489, "grad_norm": 16.359375, "learning_rate": 8.73823905628198e-06, "loss": 0.67, "step": 36011 }, { "epoch": 3.831879123217706, "grad_norm": 12.8359375, "learning_rate": 8.73820352472996e-06, "loss": 0.5754, "step": 36012 }, { "epoch": 3.8319855288359226, "grad_norm": 20.171875, "learning_rate": 8.738167993177942e-06, "loss": 0.7531, "step": 36013 }, { "epoch": 3.832091934454139, "grad_norm": 13.375, "learning_rate": 8.738132461625924e-06, "loss": 0.6218, "step": 36014 }, { "epoch": 3.832198340072356, "grad_norm": 14.28125, "learning_rate": 8.738096930073905e-06, "loss": 0.6795, "step": 36015 }, { "epoch": 3.8323047456905726, "grad_norm": 18.09375, "learning_rate": 8.738061398521887e-06, "loss": 0.684, "step": 36016 }, { "epoch": 3.832411151308789, "grad_norm": 10.3828125, "learning_rate": 8.73802586696987e-06, "loss": 0.5903, "step": 36017 }, { "epoch": 3.832517556927006, "grad_norm": 13.140625, "learning_rate": 8.737990335417852e-06, "loss": 0.7607, "step": 36018 }, { "epoch": 3.8326239625452225, "grad_norm": 12.0, "learning_rate": 8.737954803865833e-06, "loss": 0.5998, "step": 36019 }, { "epoch": 3.832730368163439, "grad_norm": 12.890625, "learning_rate": 8.737919272313815e-06, "loss": 0.6384, "step": 36020 }, { "epoch": 3.8328367737816555, "grad_norm": 14.4140625, "learning_rate": 8.737883740761797e-06, "loss": 0.7923, "step": 36021 }, { "epoch": 3.832943179399872, "grad_norm": 10.34375, "learning_rate": 8.73784820920978e-06, "loss": 0.6642, "step": 36022 }, { "epoch": 3.833049585018089, "grad_norm": 12.640625, "learning_rate": 8.737812677657761e-06, "loss": 0.5936, "step": 36023 }, { "epoch": 3.8331559906363055, "grad_norm": 14.921875, "learning_rate": 8.737777146105743e-06, "loss": 0.6514, "step": 36024 }, { "epoch": 3.833262396254522, "grad_norm": 14.6953125, "learning_rate": 8.737741614553725e-06, "loss": 0.7836, "step": 36025 }, { "epoch": 3.833368801872739, "grad_norm": 14.578125, "learning_rate": 8.737706083001706e-06, "loss": 0.6748, "step": 36026 }, { "epoch": 3.8334752074909555, "grad_norm": 16.4375, "learning_rate": 8.737670551449688e-06, "loss": 0.8764, "step": 36027 }, { "epoch": 3.833581613109172, "grad_norm": 10.9140625, "learning_rate": 8.73763501989767e-06, "loss": 0.7179, "step": 36028 }, { "epoch": 3.833688018727389, "grad_norm": 13.7265625, "learning_rate": 8.737599488345651e-06, "loss": 0.6942, "step": 36029 }, { "epoch": 3.8337944243456055, "grad_norm": 10.7421875, "learning_rate": 8.737563956793633e-06, "loss": 0.6304, "step": 36030 }, { "epoch": 3.833900829963822, "grad_norm": 11.0703125, "learning_rate": 8.737528425241614e-06, "loss": 0.5667, "step": 36031 }, { "epoch": 3.834007235582039, "grad_norm": 13.546875, "learning_rate": 8.737492893689598e-06, "loss": 0.6294, "step": 36032 }, { "epoch": 3.8341136412002554, "grad_norm": 12.78125, "learning_rate": 8.73745736213758e-06, "loss": 0.6886, "step": 36033 }, { "epoch": 3.834220046818472, "grad_norm": 13.5078125, "learning_rate": 8.73742183058556e-06, "loss": 0.6672, "step": 36034 }, { "epoch": 3.834326452436689, "grad_norm": 13.6171875, "learning_rate": 8.737386299033542e-06, "loss": 0.6823, "step": 36035 }, { "epoch": 3.8344328580549054, "grad_norm": 13.8359375, "learning_rate": 8.737350767481524e-06, "loss": 0.5535, "step": 36036 }, { "epoch": 3.834539263673122, "grad_norm": 11.8828125, "learning_rate": 8.737315235929507e-06, "loss": 0.5536, "step": 36037 }, { "epoch": 3.8346456692913384, "grad_norm": 12.5234375, "learning_rate": 8.737279704377489e-06, "loss": 0.5957, "step": 36038 }, { "epoch": 3.8347520749095554, "grad_norm": 12.3125, "learning_rate": 8.73724417282547e-06, "loss": 0.6457, "step": 36039 }, { "epoch": 3.834858480527772, "grad_norm": 15.09375, "learning_rate": 8.737208641273452e-06, "loss": 0.7708, "step": 36040 }, { "epoch": 3.8349648861459884, "grad_norm": 12.2890625, "learning_rate": 8.737173109721434e-06, "loss": 0.6071, "step": 36041 }, { "epoch": 3.835071291764205, "grad_norm": 13.234375, "learning_rate": 8.737137578169415e-06, "loss": 0.6953, "step": 36042 }, { "epoch": 3.835177697382422, "grad_norm": 14.3203125, "learning_rate": 8.737102046617397e-06, "loss": 0.7859, "step": 36043 }, { "epoch": 3.8352841030006384, "grad_norm": 20.015625, "learning_rate": 8.737066515065379e-06, "loss": 0.8397, "step": 36044 }, { "epoch": 3.835390508618855, "grad_norm": 14.3125, "learning_rate": 8.73703098351336e-06, "loss": 0.6903, "step": 36045 }, { "epoch": 3.835496914237072, "grad_norm": 10.3203125, "learning_rate": 8.736995451961342e-06, "loss": 0.5589, "step": 36046 }, { "epoch": 3.8356033198552884, "grad_norm": 13.3671875, "learning_rate": 8.736959920409325e-06, "loss": 0.7885, "step": 36047 }, { "epoch": 3.835709725473505, "grad_norm": 9.1953125, "learning_rate": 8.736924388857307e-06, "loss": 0.4704, "step": 36048 }, { "epoch": 3.835816131091722, "grad_norm": 17.359375, "learning_rate": 8.736888857305288e-06, "loss": 0.7842, "step": 36049 }, { "epoch": 3.8359225367099383, "grad_norm": 11.7421875, "learning_rate": 8.73685332575327e-06, "loss": 0.5954, "step": 36050 }, { "epoch": 3.836028942328155, "grad_norm": 18.96875, "learning_rate": 8.736817794201252e-06, "loss": 0.9452, "step": 36051 }, { "epoch": 3.836135347946372, "grad_norm": 14.8984375, "learning_rate": 8.736782262649233e-06, "loss": 0.7806, "step": 36052 }, { "epoch": 3.8362417535645883, "grad_norm": 14.453125, "learning_rate": 8.736746731097215e-06, "loss": 0.6026, "step": 36053 }, { "epoch": 3.836348159182805, "grad_norm": 12.6953125, "learning_rate": 8.736711199545196e-06, "loss": 0.6788, "step": 36054 }, { "epoch": 3.8364545648010218, "grad_norm": 14.0, "learning_rate": 8.736675667993178e-06, "loss": 0.7572, "step": 36055 }, { "epoch": 3.8365609704192383, "grad_norm": 13.15625, "learning_rate": 8.73664013644116e-06, "loss": 0.753, "step": 36056 }, { "epoch": 3.836667376037455, "grad_norm": 12.734375, "learning_rate": 8.736604604889141e-06, "loss": 0.674, "step": 36057 }, { "epoch": 3.8367737816556713, "grad_norm": 20.484375, "learning_rate": 8.736569073337124e-06, "loss": 0.8032, "step": 36058 }, { "epoch": 3.836880187273888, "grad_norm": 16.3125, "learning_rate": 8.736533541785106e-06, "loss": 0.7076, "step": 36059 }, { "epoch": 3.8369865928921048, "grad_norm": 14.46875, "learning_rate": 8.736498010233088e-06, "loss": 0.8163, "step": 36060 }, { "epoch": 3.8370929985103213, "grad_norm": 19.453125, "learning_rate": 8.73646247868107e-06, "loss": 0.9205, "step": 36061 }, { "epoch": 3.8371994041285378, "grad_norm": 17.671875, "learning_rate": 8.736426947129051e-06, "loss": 0.8028, "step": 36062 }, { "epoch": 3.8373058097467547, "grad_norm": 11.09375, "learning_rate": 8.736391415577034e-06, "loss": 0.6486, "step": 36063 }, { "epoch": 3.8374122153649712, "grad_norm": 14.703125, "learning_rate": 8.736355884025016e-06, "loss": 0.7329, "step": 36064 }, { "epoch": 3.8375186209831877, "grad_norm": 14.3671875, "learning_rate": 8.736320352472997e-06, "loss": 0.6553, "step": 36065 }, { "epoch": 3.8376250266014047, "grad_norm": 7.0703125, "learning_rate": 8.736284820920979e-06, "loss": 0.4576, "step": 36066 }, { "epoch": 3.837731432219621, "grad_norm": 14.0703125, "learning_rate": 8.73624928936896e-06, "loss": 0.6611, "step": 36067 }, { "epoch": 3.8378378378378377, "grad_norm": 9.8515625, "learning_rate": 8.736213757816942e-06, "loss": 0.4501, "step": 36068 }, { "epoch": 3.8379442434560547, "grad_norm": 12.4921875, "learning_rate": 8.736178226264924e-06, "loss": 0.5772, "step": 36069 }, { "epoch": 3.838050649074271, "grad_norm": 11.921875, "learning_rate": 8.736142694712905e-06, "loss": 0.6335, "step": 36070 }, { "epoch": 3.8381570546924877, "grad_norm": 10.578125, "learning_rate": 8.736107163160887e-06, "loss": 0.5057, "step": 36071 }, { "epoch": 3.8382634603107046, "grad_norm": 15.3515625, "learning_rate": 8.736071631608869e-06, "loss": 0.6502, "step": 36072 }, { "epoch": 3.838369865928921, "grad_norm": 8.125, "learning_rate": 8.736036100056852e-06, "loss": 0.5316, "step": 36073 }, { "epoch": 3.8384762715471377, "grad_norm": 12.734375, "learning_rate": 8.736000568504834e-06, "loss": 0.6869, "step": 36074 }, { "epoch": 3.838582677165354, "grad_norm": 17.09375, "learning_rate": 8.735965036952815e-06, "loss": 0.7657, "step": 36075 }, { "epoch": 3.838689082783571, "grad_norm": 9.4921875, "learning_rate": 8.735929505400797e-06, "loss": 0.4472, "step": 36076 }, { "epoch": 3.8387954884017876, "grad_norm": 12.8046875, "learning_rate": 8.735893973848778e-06, "loss": 0.7678, "step": 36077 }, { "epoch": 3.838901894020004, "grad_norm": 11.0234375, "learning_rate": 8.73585844229676e-06, "loss": 0.6965, "step": 36078 }, { "epoch": 3.8390082996382207, "grad_norm": 12.1953125, "learning_rate": 8.735822910744742e-06, "loss": 0.8009, "step": 36079 }, { "epoch": 3.8391147052564376, "grad_norm": 13.8046875, "learning_rate": 8.735787379192723e-06, "loss": 0.6786, "step": 36080 }, { "epoch": 3.839221110874654, "grad_norm": 15.421875, "learning_rate": 8.735751847640705e-06, "loss": 0.6498, "step": 36081 }, { "epoch": 3.8393275164928706, "grad_norm": 13.375, "learning_rate": 8.735716316088686e-06, "loss": 0.6778, "step": 36082 }, { "epoch": 3.8394339221110876, "grad_norm": 16.015625, "learning_rate": 8.73568078453667e-06, "loss": 0.7703, "step": 36083 }, { "epoch": 3.839540327729304, "grad_norm": 17.15625, "learning_rate": 8.735645252984651e-06, "loss": 0.8135, "step": 36084 }, { "epoch": 3.8396467333475206, "grad_norm": 13.0703125, "learning_rate": 8.735609721432633e-06, "loss": 0.7408, "step": 36085 }, { "epoch": 3.8397531389657376, "grad_norm": 15.53125, "learning_rate": 8.735574189880615e-06, "loss": 0.7524, "step": 36086 }, { "epoch": 3.839859544583954, "grad_norm": 15.3359375, "learning_rate": 8.735538658328596e-06, "loss": 0.6665, "step": 36087 }, { "epoch": 3.8399659502021706, "grad_norm": 13.71875, "learning_rate": 8.73550312677658e-06, "loss": 0.6956, "step": 36088 }, { "epoch": 3.8400723558203875, "grad_norm": 13.8125, "learning_rate": 8.735467595224561e-06, "loss": 0.7913, "step": 36089 }, { "epoch": 3.840178761438604, "grad_norm": 10.671875, "learning_rate": 8.735432063672543e-06, "loss": 0.5289, "step": 36090 }, { "epoch": 3.8402851670568205, "grad_norm": 12.6875, "learning_rate": 8.735396532120524e-06, "loss": 0.6226, "step": 36091 }, { "epoch": 3.8403915726750375, "grad_norm": 14.59375, "learning_rate": 8.735361000568506e-06, "loss": 0.8937, "step": 36092 }, { "epoch": 3.840497978293254, "grad_norm": 12.3359375, "learning_rate": 8.735325469016487e-06, "loss": 0.5932, "step": 36093 }, { "epoch": 3.8406043839114705, "grad_norm": 12.0625, "learning_rate": 8.735289937464469e-06, "loss": 0.6151, "step": 36094 }, { "epoch": 3.840710789529687, "grad_norm": 12.8515625, "learning_rate": 8.73525440591245e-06, "loss": 0.7159, "step": 36095 }, { "epoch": 3.8408171951479035, "grad_norm": 13.0703125, "learning_rate": 8.735218874360432e-06, "loss": 0.5932, "step": 36096 }, { "epoch": 3.8409236007661205, "grad_norm": 12.9765625, "learning_rate": 8.735183342808414e-06, "loss": 0.8232, "step": 36097 }, { "epoch": 3.841030006384337, "grad_norm": 16.453125, "learning_rate": 8.735147811256396e-06, "loss": 0.654, "step": 36098 }, { "epoch": 3.8411364120025535, "grad_norm": 12.40625, "learning_rate": 8.735112279704379e-06, "loss": 0.6476, "step": 36099 }, { "epoch": 3.8412428176207705, "grad_norm": 11.2265625, "learning_rate": 8.73507674815236e-06, "loss": 0.4921, "step": 36100 }, { "epoch": 3.841349223238987, "grad_norm": 9.34375, "learning_rate": 8.735041216600342e-06, "loss": 0.6355, "step": 36101 }, { "epoch": 3.8414556288572035, "grad_norm": 16.140625, "learning_rate": 8.735005685048324e-06, "loss": 0.6358, "step": 36102 }, { "epoch": 3.8415620344754204, "grad_norm": 10.921875, "learning_rate": 8.734970153496305e-06, "loss": 0.5517, "step": 36103 }, { "epoch": 3.841668440093637, "grad_norm": 17.71875, "learning_rate": 8.734934621944287e-06, "loss": 0.8611, "step": 36104 }, { "epoch": 3.8417748457118535, "grad_norm": 13.625, "learning_rate": 8.734899090392268e-06, "loss": 0.5963, "step": 36105 }, { "epoch": 3.8418812513300704, "grad_norm": 12.9921875, "learning_rate": 8.73486355884025e-06, "loss": 0.679, "step": 36106 }, { "epoch": 3.841987656948287, "grad_norm": 11.6328125, "learning_rate": 8.734828027288232e-06, "loss": 0.6482, "step": 36107 }, { "epoch": 3.8420940625665034, "grad_norm": 11.8515625, "learning_rate": 8.734792495736213e-06, "loss": 0.6775, "step": 36108 }, { "epoch": 3.8422004681847204, "grad_norm": 14.09375, "learning_rate": 8.734756964184197e-06, "loss": 0.6422, "step": 36109 }, { "epoch": 3.842306873802937, "grad_norm": 15.03125, "learning_rate": 8.734721432632178e-06, "loss": 0.6312, "step": 36110 }, { "epoch": 3.8424132794211534, "grad_norm": 12.3359375, "learning_rate": 8.73468590108016e-06, "loss": 0.6596, "step": 36111 }, { "epoch": 3.84251968503937, "grad_norm": 15.9375, "learning_rate": 8.734650369528141e-06, "loss": 0.6911, "step": 36112 }, { "epoch": 3.842626090657587, "grad_norm": 12.0, "learning_rate": 8.734614837976123e-06, "loss": 0.6496, "step": 36113 }, { "epoch": 3.8427324962758034, "grad_norm": 13.5234375, "learning_rate": 8.734579306424106e-06, "loss": 0.7168, "step": 36114 }, { "epoch": 3.84283890189402, "grad_norm": 11.3984375, "learning_rate": 8.734543774872088e-06, "loss": 0.5639, "step": 36115 }, { "epoch": 3.8429453075122364, "grad_norm": 11.9453125, "learning_rate": 8.73450824332007e-06, "loss": 0.5941, "step": 36116 }, { "epoch": 3.8430517131304534, "grad_norm": 14.6015625, "learning_rate": 8.734472711768051e-06, "loss": 0.6762, "step": 36117 }, { "epoch": 3.84315811874867, "grad_norm": 12.0859375, "learning_rate": 8.734437180216033e-06, "loss": 0.6521, "step": 36118 }, { "epoch": 3.8432645243668864, "grad_norm": 14.125, "learning_rate": 8.734401648664014e-06, "loss": 0.5815, "step": 36119 }, { "epoch": 3.8433709299851033, "grad_norm": 15.7734375, "learning_rate": 8.734366117111996e-06, "loss": 0.727, "step": 36120 }, { "epoch": 3.84347733560332, "grad_norm": 11.4609375, "learning_rate": 8.734330585559978e-06, "loss": 0.5564, "step": 36121 }, { "epoch": 3.8435837412215363, "grad_norm": 13.890625, "learning_rate": 8.73429505400796e-06, "loss": 0.6066, "step": 36122 }, { "epoch": 3.8436901468397533, "grad_norm": 12.6015625, "learning_rate": 8.73425952245594e-06, "loss": 0.5173, "step": 36123 }, { "epoch": 3.84379655245797, "grad_norm": 15.0390625, "learning_rate": 8.734223990903924e-06, "loss": 0.801, "step": 36124 }, { "epoch": 3.8439029580761863, "grad_norm": 9.7109375, "learning_rate": 8.734188459351906e-06, "loss": 0.5717, "step": 36125 }, { "epoch": 3.8440093636944033, "grad_norm": 19.5625, "learning_rate": 8.734152927799887e-06, "loss": 0.6685, "step": 36126 }, { "epoch": 3.84411576931262, "grad_norm": 14.53125, "learning_rate": 8.734117396247869e-06, "loss": 0.7321, "step": 36127 }, { "epoch": 3.8442221749308363, "grad_norm": 14.3359375, "learning_rate": 8.73408186469585e-06, "loss": 0.7944, "step": 36128 }, { "epoch": 3.8443285805490532, "grad_norm": 22.40625, "learning_rate": 8.734046333143834e-06, "loss": 0.8779, "step": 36129 }, { "epoch": 3.8444349861672698, "grad_norm": 13.328125, "learning_rate": 8.734010801591815e-06, "loss": 0.8232, "step": 36130 }, { "epoch": 3.8445413917854863, "grad_norm": 14.21875, "learning_rate": 8.733975270039797e-06, "loss": 0.6983, "step": 36131 }, { "epoch": 3.8446477974037028, "grad_norm": 12.9609375, "learning_rate": 8.733939738487777e-06, "loss": 0.6855, "step": 36132 }, { "epoch": 3.8447542030219193, "grad_norm": 12.03125, "learning_rate": 8.733904206935759e-06, "loss": 0.5817, "step": 36133 }, { "epoch": 3.8448606086401362, "grad_norm": 13.0703125, "learning_rate": 8.73386867538374e-06, "loss": 0.6319, "step": 36134 }, { "epoch": 3.8449670142583527, "grad_norm": 16.28125, "learning_rate": 8.733833143831723e-06, "loss": 0.7026, "step": 36135 }, { "epoch": 3.8450734198765693, "grad_norm": 16.703125, "learning_rate": 8.733797612279705e-06, "loss": 0.7362, "step": 36136 }, { "epoch": 3.845179825494786, "grad_norm": 16.46875, "learning_rate": 8.733762080727687e-06, "loss": 0.5872, "step": 36137 }, { "epoch": 3.8452862311130027, "grad_norm": 12.34375, "learning_rate": 8.733726549175668e-06, "loss": 0.7204, "step": 36138 }, { "epoch": 3.8453926367312192, "grad_norm": 17.984375, "learning_rate": 8.73369101762365e-06, "loss": 0.7716, "step": 36139 }, { "epoch": 3.845499042349436, "grad_norm": 14.0234375, "learning_rate": 8.733655486071633e-06, "loss": 0.6437, "step": 36140 }, { "epoch": 3.8456054479676527, "grad_norm": 13.140625, "learning_rate": 8.733619954519615e-06, "loss": 0.6147, "step": 36141 }, { "epoch": 3.845711853585869, "grad_norm": 13.1484375, "learning_rate": 8.733584422967596e-06, "loss": 0.6916, "step": 36142 }, { "epoch": 3.845818259204086, "grad_norm": 14.03125, "learning_rate": 8.733548891415578e-06, "loss": 0.724, "step": 36143 }, { "epoch": 3.8459246648223027, "grad_norm": 12.4140625, "learning_rate": 8.73351335986356e-06, "loss": 0.6351, "step": 36144 }, { "epoch": 3.846031070440519, "grad_norm": 11.8984375, "learning_rate": 8.733477828311541e-06, "loss": 0.5401, "step": 36145 }, { "epoch": 3.846137476058736, "grad_norm": 15.6953125, "learning_rate": 8.733442296759523e-06, "loss": 0.6142, "step": 36146 }, { "epoch": 3.8462438816769526, "grad_norm": 12.015625, "learning_rate": 8.733406765207504e-06, "loss": 0.6286, "step": 36147 }, { "epoch": 3.846350287295169, "grad_norm": 15.453125, "learning_rate": 8.733371233655486e-06, "loss": 0.6404, "step": 36148 }, { "epoch": 3.846456692913386, "grad_norm": 13.890625, "learning_rate": 8.733335702103468e-06, "loss": 0.5607, "step": 36149 }, { "epoch": 3.8465630985316026, "grad_norm": 14.40625, "learning_rate": 8.733300170551451e-06, "loss": 0.6156, "step": 36150 }, { "epoch": 3.846669504149819, "grad_norm": 15.2265625, "learning_rate": 8.733264638999433e-06, "loss": 0.7719, "step": 36151 }, { "epoch": 3.8467759097680356, "grad_norm": 16.703125, "learning_rate": 8.733229107447414e-06, "loss": 0.7883, "step": 36152 }, { "epoch": 3.846882315386252, "grad_norm": 13.59375, "learning_rate": 8.733193575895396e-06, "loss": 0.7143, "step": 36153 }, { "epoch": 3.846988721004469, "grad_norm": 15.2265625, "learning_rate": 8.733158044343377e-06, "loss": 0.6127, "step": 36154 }, { "epoch": 3.8470951266226856, "grad_norm": 17.109375, "learning_rate": 8.73312251279136e-06, "loss": 0.8004, "step": 36155 }, { "epoch": 3.847201532240902, "grad_norm": 9.3046875, "learning_rate": 8.733086981239342e-06, "loss": 0.4208, "step": 36156 }, { "epoch": 3.847307937859119, "grad_norm": 13.4609375, "learning_rate": 8.733051449687324e-06, "loss": 0.6553, "step": 36157 }, { "epoch": 3.8474143434773356, "grad_norm": 16.578125, "learning_rate": 8.733015918135305e-06, "loss": 0.83, "step": 36158 }, { "epoch": 3.847520749095552, "grad_norm": 13.1484375, "learning_rate": 8.732980386583287e-06, "loss": 0.5789, "step": 36159 }, { "epoch": 3.847627154713769, "grad_norm": 18.09375, "learning_rate": 8.732944855031269e-06, "loss": 0.7224, "step": 36160 }, { "epoch": 3.8477335603319855, "grad_norm": 15.15625, "learning_rate": 8.73290932347925e-06, "loss": 0.6722, "step": 36161 }, { "epoch": 3.847839965950202, "grad_norm": 12.3671875, "learning_rate": 8.732873791927232e-06, "loss": 0.6393, "step": 36162 }, { "epoch": 3.847946371568419, "grad_norm": 13.03125, "learning_rate": 8.732838260375214e-06, "loss": 0.5983, "step": 36163 }, { "epoch": 3.8480527771866355, "grad_norm": 11.421875, "learning_rate": 8.732802728823195e-06, "loss": 0.6718, "step": 36164 }, { "epoch": 3.848159182804852, "grad_norm": 12.796875, "learning_rate": 8.732767197271178e-06, "loss": 0.6026, "step": 36165 }, { "epoch": 3.848265588423069, "grad_norm": 14.4765625, "learning_rate": 8.73273166571916e-06, "loss": 0.6039, "step": 36166 }, { "epoch": 3.8483719940412855, "grad_norm": 14.5625, "learning_rate": 8.732696134167142e-06, "loss": 0.6693, "step": 36167 }, { "epoch": 3.848478399659502, "grad_norm": 13.0234375, "learning_rate": 8.732660602615123e-06, "loss": 0.6883, "step": 36168 }, { "epoch": 3.8485848052777185, "grad_norm": 18.296875, "learning_rate": 8.732625071063105e-06, "loss": 0.7838, "step": 36169 }, { "epoch": 3.8486912108959355, "grad_norm": 8.984375, "learning_rate": 8.732589539511086e-06, "loss": 0.4339, "step": 36170 }, { "epoch": 3.848797616514152, "grad_norm": 17.75, "learning_rate": 8.732554007959068e-06, "loss": 0.7198, "step": 36171 }, { "epoch": 3.8489040221323685, "grad_norm": 10.4453125, "learning_rate": 8.73251847640705e-06, "loss": 0.7168, "step": 36172 }, { "epoch": 3.849010427750585, "grad_norm": 12.34375, "learning_rate": 8.732482944855031e-06, "loss": 0.6398, "step": 36173 }, { "epoch": 3.849116833368802, "grad_norm": 16.265625, "learning_rate": 8.732447413303013e-06, "loss": 0.6344, "step": 36174 }, { "epoch": 3.8492232389870185, "grad_norm": 15.53125, "learning_rate": 8.732411881750994e-06, "loss": 0.7882, "step": 36175 }, { "epoch": 3.849329644605235, "grad_norm": 14.1953125, "learning_rate": 8.732376350198978e-06, "loss": 0.7803, "step": 36176 }, { "epoch": 3.849436050223452, "grad_norm": 21.765625, "learning_rate": 8.73234081864696e-06, "loss": 1.0515, "step": 36177 }, { "epoch": 3.8495424558416684, "grad_norm": 16.359375, "learning_rate": 8.732305287094941e-06, "loss": 0.8482, "step": 36178 }, { "epoch": 3.849648861459885, "grad_norm": 15.0078125, "learning_rate": 8.732269755542923e-06, "loss": 0.7791, "step": 36179 }, { "epoch": 3.849755267078102, "grad_norm": 13.6015625, "learning_rate": 8.732234223990904e-06, "loss": 0.7241, "step": 36180 }, { "epoch": 3.8498616726963184, "grad_norm": 14.7109375, "learning_rate": 8.732198692438888e-06, "loss": 0.6979, "step": 36181 }, { "epoch": 3.849968078314535, "grad_norm": 13.09375, "learning_rate": 8.732163160886869e-06, "loss": 0.7438, "step": 36182 }, { "epoch": 3.850074483932752, "grad_norm": 21.296875, "learning_rate": 8.73212762933485e-06, "loss": 0.6448, "step": 36183 }, { "epoch": 3.8501808895509684, "grad_norm": 14.3359375, "learning_rate": 8.732092097782832e-06, "loss": 0.7554, "step": 36184 }, { "epoch": 3.850287295169185, "grad_norm": 11.9140625, "learning_rate": 8.732056566230814e-06, "loss": 0.7049, "step": 36185 }, { "epoch": 3.850393700787402, "grad_norm": 17.515625, "learning_rate": 8.732021034678796e-06, "loss": 0.7441, "step": 36186 }, { "epoch": 3.8505001064056184, "grad_norm": 16.0, "learning_rate": 8.731985503126777e-06, "loss": 0.7509, "step": 36187 }, { "epoch": 3.850606512023835, "grad_norm": 10.640625, "learning_rate": 8.731949971574759e-06, "loss": 0.5602, "step": 36188 }, { "epoch": 3.8507129176420514, "grad_norm": 9.3359375, "learning_rate": 8.73191444002274e-06, "loss": 0.5384, "step": 36189 }, { "epoch": 3.850819323260268, "grad_norm": 11.1484375, "learning_rate": 8.731878908470722e-06, "loss": 0.649, "step": 36190 }, { "epoch": 3.850925728878485, "grad_norm": 12.4375, "learning_rate": 8.731843376918705e-06, "loss": 0.7312, "step": 36191 }, { "epoch": 3.8510321344967013, "grad_norm": 12.125, "learning_rate": 8.731807845366687e-06, "loss": 0.76, "step": 36192 }, { "epoch": 3.851138540114918, "grad_norm": 10.71875, "learning_rate": 8.731772313814668e-06, "loss": 0.5711, "step": 36193 }, { "epoch": 3.851244945733135, "grad_norm": 12.2421875, "learning_rate": 8.73173678226265e-06, "loss": 0.7108, "step": 36194 }, { "epoch": 3.8513513513513513, "grad_norm": 16.046875, "learning_rate": 8.731701250710632e-06, "loss": 0.8785, "step": 36195 }, { "epoch": 3.851457756969568, "grad_norm": 13.4765625, "learning_rate": 8.731665719158613e-06, "loss": 0.7524, "step": 36196 }, { "epoch": 3.851564162587785, "grad_norm": 11.734375, "learning_rate": 8.731630187606595e-06, "loss": 0.7429, "step": 36197 }, { "epoch": 3.8516705682060013, "grad_norm": 18.609375, "learning_rate": 8.731594656054577e-06, "loss": 0.845, "step": 36198 }, { "epoch": 3.851776973824218, "grad_norm": 11.71875, "learning_rate": 8.731559124502558e-06, "loss": 0.6645, "step": 36199 }, { "epoch": 3.8518833794424348, "grad_norm": 10.5390625, "learning_rate": 8.73152359295054e-06, "loss": 0.6677, "step": 36200 }, { "epoch": 3.8519897850606513, "grad_norm": 9.7109375, "learning_rate": 8.731488061398523e-06, "loss": 0.4898, "step": 36201 }, { "epoch": 3.8520961906788678, "grad_norm": 13.28125, "learning_rate": 8.731452529846505e-06, "loss": 0.7014, "step": 36202 }, { "epoch": 3.8522025962970847, "grad_norm": 10.8671875, "learning_rate": 8.731416998294486e-06, "loss": 0.6565, "step": 36203 }, { "epoch": 3.8523090019153012, "grad_norm": 11.25, "learning_rate": 8.731381466742468e-06, "loss": 0.6174, "step": 36204 }, { "epoch": 3.8524154075335177, "grad_norm": 11.875, "learning_rate": 8.73134593519045e-06, "loss": 0.5355, "step": 36205 }, { "epoch": 3.8525218131517343, "grad_norm": 12.9453125, "learning_rate": 8.731310403638433e-06, "loss": 0.6874, "step": 36206 }, { "epoch": 3.852628218769951, "grad_norm": 15.7890625, "learning_rate": 8.731274872086414e-06, "loss": 0.7836, "step": 36207 }, { "epoch": 3.8527346243881677, "grad_norm": 11.4453125, "learning_rate": 8.731239340534396e-06, "loss": 0.6255, "step": 36208 }, { "epoch": 3.8528410300063842, "grad_norm": 11.03125, "learning_rate": 8.731203808982378e-06, "loss": 0.5195, "step": 36209 }, { "epoch": 3.8529474356246007, "grad_norm": 10.96875, "learning_rate": 8.73116827743036e-06, "loss": 0.6367, "step": 36210 }, { "epoch": 3.8530538412428177, "grad_norm": 11.7890625, "learning_rate": 8.73113274587834e-06, "loss": 0.6824, "step": 36211 }, { "epoch": 3.853160246861034, "grad_norm": 12.296875, "learning_rate": 8.731097214326322e-06, "loss": 0.7129, "step": 36212 }, { "epoch": 3.8532666524792507, "grad_norm": 14.8046875, "learning_rate": 8.731061682774304e-06, "loss": 0.7172, "step": 36213 }, { "epoch": 3.8533730580974677, "grad_norm": 15.3125, "learning_rate": 8.731026151222286e-06, "loss": 0.7298, "step": 36214 }, { "epoch": 3.853479463715684, "grad_norm": 13.078125, "learning_rate": 8.730990619670267e-06, "loss": 0.7125, "step": 36215 }, { "epoch": 3.8535858693339007, "grad_norm": 12.890625, "learning_rate": 8.730955088118249e-06, "loss": 0.71, "step": 36216 }, { "epoch": 3.8536922749521176, "grad_norm": 18.40625, "learning_rate": 8.730919556566232e-06, "loss": 0.7885, "step": 36217 }, { "epoch": 3.853798680570334, "grad_norm": 12.21875, "learning_rate": 8.730884025014214e-06, "loss": 0.5798, "step": 36218 }, { "epoch": 3.8539050861885507, "grad_norm": 11.046875, "learning_rate": 8.730848493462195e-06, "loss": 0.5815, "step": 36219 }, { "epoch": 3.8540114918067676, "grad_norm": 12.3671875, "learning_rate": 8.730812961910177e-06, "loss": 0.6792, "step": 36220 }, { "epoch": 3.854117897424984, "grad_norm": 12.6171875, "learning_rate": 8.730777430358159e-06, "loss": 0.6233, "step": 36221 }, { "epoch": 3.8542243030432006, "grad_norm": 12.609375, "learning_rate": 8.73074189880614e-06, "loss": 0.6599, "step": 36222 }, { "epoch": 3.8543307086614176, "grad_norm": 12.4140625, "learning_rate": 8.730706367254122e-06, "loss": 0.5296, "step": 36223 }, { "epoch": 3.854437114279634, "grad_norm": 12.6328125, "learning_rate": 8.730670835702103e-06, "loss": 0.6021, "step": 36224 }, { "epoch": 3.8545435198978506, "grad_norm": 11.5859375, "learning_rate": 8.730635304150085e-06, "loss": 0.5177, "step": 36225 }, { "epoch": 3.854649925516067, "grad_norm": 12.7578125, "learning_rate": 8.730599772598067e-06, "loss": 0.5759, "step": 36226 }, { "epoch": 3.8547563311342836, "grad_norm": 14.6953125, "learning_rate": 8.73056424104605e-06, "loss": 0.6485, "step": 36227 }, { "epoch": 3.8548627367525006, "grad_norm": 17.765625, "learning_rate": 8.730528709494031e-06, "loss": 0.838, "step": 36228 }, { "epoch": 3.854969142370717, "grad_norm": 13.1015625, "learning_rate": 8.730493177942013e-06, "loss": 0.8129, "step": 36229 }, { "epoch": 3.8550755479889336, "grad_norm": 14.9765625, "learning_rate": 8.730457646389995e-06, "loss": 0.7247, "step": 36230 }, { "epoch": 3.8551819536071505, "grad_norm": 13.9609375, "learning_rate": 8.730422114837976e-06, "loss": 0.5876, "step": 36231 }, { "epoch": 3.855288359225367, "grad_norm": 11.2265625, "learning_rate": 8.73038658328596e-06, "loss": 0.7019, "step": 36232 }, { "epoch": 3.8553947648435836, "grad_norm": 12.953125, "learning_rate": 8.730351051733941e-06, "loss": 0.636, "step": 36233 }, { "epoch": 3.8555011704618005, "grad_norm": 16.046875, "learning_rate": 8.730315520181923e-06, "loss": 0.912, "step": 36234 }, { "epoch": 3.855607576080017, "grad_norm": 14.6640625, "learning_rate": 8.730279988629904e-06, "loss": 0.6177, "step": 36235 }, { "epoch": 3.8557139816982335, "grad_norm": 12.859375, "learning_rate": 8.730244457077886e-06, "loss": 0.6713, "step": 36236 }, { "epoch": 3.8558203873164505, "grad_norm": 14.5390625, "learning_rate": 8.730208925525868e-06, "loss": 0.6697, "step": 36237 }, { "epoch": 3.855926792934667, "grad_norm": 15.9296875, "learning_rate": 8.73017339397385e-06, "loss": 0.6764, "step": 36238 }, { "epoch": 3.8560331985528835, "grad_norm": 28.5, "learning_rate": 8.730137862421831e-06, "loss": 0.8677, "step": 36239 }, { "epoch": 3.8561396041711005, "grad_norm": 12.96875, "learning_rate": 8.730102330869812e-06, "loss": 0.6362, "step": 36240 }, { "epoch": 3.856246009789317, "grad_norm": 13.1484375, "learning_rate": 8.730066799317794e-06, "loss": 0.6747, "step": 36241 }, { "epoch": 3.8563524154075335, "grad_norm": 11.46875, "learning_rate": 8.730031267765777e-06, "loss": 0.579, "step": 36242 }, { "epoch": 3.85645882102575, "grad_norm": 12.4296875, "learning_rate": 8.729995736213759e-06, "loss": 0.7185, "step": 36243 }, { "epoch": 3.856565226643967, "grad_norm": 10.625, "learning_rate": 8.72996020466174e-06, "loss": 0.6284, "step": 36244 }, { "epoch": 3.8566716322621835, "grad_norm": 11.4765625, "learning_rate": 8.729924673109722e-06, "loss": 0.5105, "step": 36245 }, { "epoch": 3.8567780378804, "grad_norm": 10.0, "learning_rate": 8.729889141557704e-06, "loss": 0.5956, "step": 36246 }, { "epoch": 3.8568844434986165, "grad_norm": 15.1640625, "learning_rate": 8.729853610005685e-06, "loss": 0.6995, "step": 36247 }, { "epoch": 3.8569908491168334, "grad_norm": 13.921875, "learning_rate": 8.729818078453669e-06, "loss": 0.8539, "step": 36248 }, { "epoch": 3.85709725473505, "grad_norm": 11.0859375, "learning_rate": 8.72978254690165e-06, "loss": 0.4962, "step": 36249 }, { "epoch": 3.8572036603532665, "grad_norm": 19.296875, "learning_rate": 8.729747015349632e-06, "loss": 0.7925, "step": 36250 }, { "epoch": 3.8573100659714834, "grad_norm": 18.125, "learning_rate": 8.729711483797614e-06, "loss": 0.8403, "step": 36251 }, { "epoch": 3.8574164715897, "grad_norm": 15.2109375, "learning_rate": 8.729675952245593e-06, "loss": 0.7715, "step": 36252 }, { "epoch": 3.8575228772079164, "grad_norm": 13.8984375, "learning_rate": 8.729640420693577e-06, "loss": 0.7292, "step": 36253 }, { "epoch": 3.8576292828261334, "grad_norm": 11.140625, "learning_rate": 8.729604889141558e-06, "loss": 0.6663, "step": 36254 }, { "epoch": 3.85773568844435, "grad_norm": 15.84375, "learning_rate": 8.72956935758954e-06, "loss": 0.5306, "step": 36255 }, { "epoch": 3.8578420940625664, "grad_norm": 14.78125, "learning_rate": 8.729533826037522e-06, "loss": 0.5953, "step": 36256 }, { "epoch": 3.8579484996807833, "grad_norm": 13.421875, "learning_rate": 8.729498294485503e-06, "loss": 0.7059, "step": 36257 }, { "epoch": 3.858054905299, "grad_norm": 15.25, "learning_rate": 8.729462762933486e-06, "loss": 0.6023, "step": 36258 }, { "epoch": 3.8581613109172164, "grad_norm": 11.6015625, "learning_rate": 8.729427231381468e-06, "loss": 0.5304, "step": 36259 }, { "epoch": 3.8582677165354333, "grad_norm": 12.2578125, "learning_rate": 8.72939169982945e-06, "loss": 0.5296, "step": 36260 }, { "epoch": 3.85837412215365, "grad_norm": 11.9375, "learning_rate": 8.729356168277431e-06, "loss": 0.5668, "step": 36261 }, { "epoch": 3.8584805277718663, "grad_norm": 15.5078125, "learning_rate": 8.729320636725413e-06, "loss": 0.7583, "step": 36262 }, { "epoch": 3.858586933390083, "grad_norm": 13.9140625, "learning_rate": 8.729285105173395e-06, "loss": 0.7645, "step": 36263 }, { "epoch": 3.8586933390082994, "grad_norm": 12.5390625, "learning_rate": 8.729249573621376e-06, "loss": 0.6735, "step": 36264 }, { "epoch": 3.8587997446265163, "grad_norm": 13.6640625, "learning_rate": 8.729214042069358e-06, "loss": 0.6568, "step": 36265 }, { "epoch": 3.858906150244733, "grad_norm": 14.8359375, "learning_rate": 8.72917851051734e-06, "loss": 0.6891, "step": 36266 }, { "epoch": 3.8590125558629493, "grad_norm": 11.65625, "learning_rate": 8.729142978965321e-06, "loss": 0.6317, "step": 36267 }, { "epoch": 3.8591189614811663, "grad_norm": 16.578125, "learning_rate": 8.729107447413304e-06, "loss": 0.7511, "step": 36268 }, { "epoch": 3.859225367099383, "grad_norm": 13.4375, "learning_rate": 8.729071915861286e-06, "loss": 0.7434, "step": 36269 }, { "epoch": 3.8593317727175993, "grad_norm": 15.4765625, "learning_rate": 8.729036384309267e-06, "loss": 0.7639, "step": 36270 }, { "epoch": 3.8594381783358163, "grad_norm": 15.8203125, "learning_rate": 8.729000852757249e-06, "loss": 0.7079, "step": 36271 }, { "epoch": 3.8595445839540328, "grad_norm": 12.1171875, "learning_rate": 8.72896532120523e-06, "loss": 0.583, "step": 36272 }, { "epoch": 3.8596509895722493, "grad_norm": 14.4140625, "learning_rate": 8.728929789653214e-06, "loss": 0.7792, "step": 36273 }, { "epoch": 3.8597573951904662, "grad_norm": 16.4375, "learning_rate": 8.728894258101196e-06, "loss": 0.8113, "step": 36274 }, { "epoch": 3.8598638008086827, "grad_norm": 15.40625, "learning_rate": 8.728858726549177e-06, "loss": 0.6501, "step": 36275 }, { "epoch": 3.8599702064268993, "grad_norm": 12.5234375, "learning_rate": 8.728823194997159e-06, "loss": 0.6595, "step": 36276 }, { "epoch": 3.860076612045116, "grad_norm": 12.5, "learning_rate": 8.72878766344514e-06, "loss": 0.7365, "step": 36277 }, { "epoch": 3.8601830176633327, "grad_norm": 11.8515625, "learning_rate": 8.728752131893122e-06, "loss": 0.6843, "step": 36278 }, { "epoch": 3.8602894232815492, "grad_norm": 14.0078125, "learning_rate": 8.728716600341104e-06, "loss": 0.72, "step": 36279 }, { "epoch": 3.860395828899766, "grad_norm": 13.390625, "learning_rate": 8.728681068789085e-06, "loss": 0.6761, "step": 36280 }, { "epoch": 3.8605022345179827, "grad_norm": 11.96875, "learning_rate": 8.728645537237067e-06, "loss": 0.578, "step": 36281 }, { "epoch": 3.860608640136199, "grad_norm": 9.8359375, "learning_rate": 8.728610005685048e-06, "loss": 0.5752, "step": 36282 }, { "epoch": 3.8607150457544157, "grad_norm": 11.96875, "learning_rate": 8.728574474133032e-06, "loss": 0.5247, "step": 36283 }, { "epoch": 3.860821451372632, "grad_norm": 14.1875, "learning_rate": 8.728538942581013e-06, "loss": 0.7948, "step": 36284 }, { "epoch": 3.860927856990849, "grad_norm": 12.8359375, "learning_rate": 8.728503411028995e-06, "loss": 0.614, "step": 36285 }, { "epoch": 3.8610342626090657, "grad_norm": 12.015625, "learning_rate": 8.728467879476977e-06, "loss": 0.6209, "step": 36286 }, { "epoch": 3.861140668227282, "grad_norm": 12.1875, "learning_rate": 8.728432347924958e-06, "loss": 0.543, "step": 36287 }, { "epoch": 3.861247073845499, "grad_norm": 12.421875, "learning_rate": 8.72839681637294e-06, "loss": 0.6758, "step": 36288 }, { "epoch": 3.8613534794637157, "grad_norm": 15.453125, "learning_rate": 8.728361284820921e-06, "loss": 0.81, "step": 36289 }, { "epoch": 3.861459885081932, "grad_norm": 15.7734375, "learning_rate": 8.728325753268903e-06, "loss": 0.6367, "step": 36290 }, { "epoch": 3.861566290700149, "grad_norm": 17.1875, "learning_rate": 8.728290221716885e-06, "loss": 0.7546, "step": 36291 }, { "epoch": 3.8616726963183656, "grad_norm": 12.921875, "learning_rate": 8.728254690164866e-06, "loss": 0.5953, "step": 36292 }, { "epoch": 3.861779101936582, "grad_norm": 10.5859375, "learning_rate": 8.728219158612848e-06, "loss": 0.5468, "step": 36293 }, { "epoch": 3.861885507554799, "grad_norm": 20.171875, "learning_rate": 8.728183627060831e-06, "loss": 0.8236, "step": 36294 }, { "epoch": 3.8619919131730156, "grad_norm": 14.546875, "learning_rate": 8.728148095508813e-06, "loss": 0.7042, "step": 36295 }, { "epoch": 3.862098318791232, "grad_norm": 13.3984375, "learning_rate": 8.728112563956794e-06, "loss": 0.835, "step": 36296 }, { "epoch": 3.862204724409449, "grad_norm": 14.046875, "learning_rate": 8.728077032404776e-06, "loss": 0.8684, "step": 36297 }, { "epoch": 3.8623111300276656, "grad_norm": 13.921875, "learning_rate": 8.728041500852758e-06, "loss": 0.5722, "step": 36298 }, { "epoch": 3.862417535645882, "grad_norm": 15.578125, "learning_rate": 8.72800596930074e-06, "loss": 0.6992, "step": 36299 }, { "epoch": 3.8625239412640986, "grad_norm": 17.296875, "learning_rate": 8.727970437748722e-06, "loss": 0.7762, "step": 36300 }, { "epoch": 3.8626303468823155, "grad_norm": 11.28125, "learning_rate": 8.727934906196704e-06, "loss": 0.5464, "step": 36301 }, { "epoch": 3.862736752500532, "grad_norm": 17.3125, "learning_rate": 8.727899374644686e-06, "loss": 0.6219, "step": 36302 }, { "epoch": 3.8628431581187486, "grad_norm": 10.578125, "learning_rate": 8.727863843092667e-06, "loss": 0.4893, "step": 36303 }, { "epoch": 3.862949563736965, "grad_norm": 11.3828125, "learning_rate": 8.727828311540649e-06, "loss": 0.6527, "step": 36304 }, { "epoch": 3.863055969355182, "grad_norm": 12.9765625, "learning_rate": 8.72779277998863e-06, "loss": 0.6817, "step": 36305 }, { "epoch": 3.8631623749733985, "grad_norm": 12.4921875, "learning_rate": 8.727757248436612e-06, "loss": 0.6268, "step": 36306 }, { "epoch": 3.863268780591615, "grad_norm": 11.9296875, "learning_rate": 8.727721716884594e-06, "loss": 0.6419, "step": 36307 }, { "epoch": 3.863375186209832, "grad_norm": 14.2109375, "learning_rate": 8.727686185332575e-06, "loss": 0.7191, "step": 36308 }, { "epoch": 3.8634815918280485, "grad_norm": 13.390625, "learning_rate": 8.727650653780559e-06, "loss": 0.6361, "step": 36309 }, { "epoch": 3.863587997446265, "grad_norm": 14.75, "learning_rate": 8.72761512222854e-06, "loss": 0.7702, "step": 36310 }, { "epoch": 3.863694403064482, "grad_norm": 16.25, "learning_rate": 8.727579590676522e-06, "loss": 0.6113, "step": 36311 }, { "epoch": 3.8638008086826985, "grad_norm": 10.8125, "learning_rate": 8.727544059124503e-06, "loss": 0.6229, "step": 36312 }, { "epoch": 3.863907214300915, "grad_norm": 12.9609375, "learning_rate": 8.727508527572485e-06, "loss": 0.653, "step": 36313 }, { "epoch": 3.864013619919132, "grad_norm": 17.21875, "learning_rate": 8.727472996020467e-06, "loss": 0.6729, "step": 36314 }, { "epoch": 3.8641200255373485, "grad_norm": 13.5390625, "learning_rate": 8.727437464468448e-06, "loss": 0.6966, "step": 36315 }, { "epoch": 3.864226431155565, "grad_norm": 14.2109375, "learning_rate": 8.72740193291643e-06, "loss": 0.7615, "step": 36316 }, { "epoch": 3.864332836773782, "grad_norm": 15.1875, "learning_rate": 8.727366401364411e-06, "loss": 0.6863, "step": 36317 }, { "epoch": 3.8644392423919984, "grad_norm": 11.515625, "learning_rate": 8.727330869812393e-06, "loss": 0.5669, "step": 36318 }, { "epoch": 3.864545648010215, "grad_norm": 9.7421875, "learning_rate": 8.727295338260376e-06, "loss": 0.5806, "step": 36319 }, { "epoch": 3.8646520536284314, "grad_norm": 12.1953125, "learning_rate": 8.727259806708358e-06, "loss": 0.6772, "step": 36320 }, { "epoch": 3.864758459246648, "grad_norm": 12.8203125, "learning_rate": 8.72722427515634e-06, "loss": 0.5011, "step": 36321 }, { "epoch": 3.864864864864865, "grad_norm": 14.6796875, "learning_rate": 8.727188743604321e-06, "loss": 0.588, "step": 36322 }, { "epoch": 3.8649712704830814, "grad_norm": 20.921875, "learning_rate": 8.727153212052303e-06, "loss": 0.8254, "step": 36323 }, { "epoch": 3.865077676101298, "grad_norm": 12.09375, "learning_rate": 8.727117680500286e-06, "loss": 0.6338, "step": 36324 }, { "epoch": 3.865184081719515, "grad_norm": 11.1171875, "learning_rate": 8.727082148948268e-06, "loss": 0.5987, "step": 36325 }, { "epoch": 3.8652904873377314, "grad_norm": 17.515625, "learning_rate": 8.72704661739625e-06, "loss": 0.8804, "step": 36326 }, { "epoch": 3.865396892955948, "grad_norm": 14.71875, "learning_rate": 8.727011085844231e-06, "loss": 0.7088, "step": 36327 }, { "epoch": 3.865503298574165, "grad_norm": 11.625, "learning_rate": 8.726975554292212e-06, "loss": 0.6527, "step": 36328 }, { "epoch": 3.8656097041923814, "grad_norm": 14.8984375, "learning_rate": 8.726940022740194e-06, "loss": 0.7002, "step": 36329 }, { "epoch": 3.865716109810598, "grad_norm": 16.109375, "learning_rate": 8.726904491188176e-06, "loss": 0.6391, "step": 36330 }, { "epoch": 3.865822515428815, "grad_norm": 12.953125, "learning_rate": 8.726868959636157e-06, "loss": 0.6629, "step": 36331 }, { "epoch": 3.8659289210470313, "grad_norm": 13.90625, "learning_rate": 8.726833428084139e-06, "loss": 0.608, "step": 36332 }, { "epoch": 3.866035326665248, "grad_norm": 10.421875, "learning_rate": 8.72679789653212e-06, "loss": 0.6215, "step": 36333 }, { "epoch": 3.866141732283465, "grad_norm": 11.546875, "learning_rate": 8.726762364980102e-06, "loss": 0.7897, "step": 36334 }, { "epoch": 3.8662481379016813, "grad_norm": 11.2421875, "learning_rate": 8.726726833428085e-06, "loss": 0.6139, "step": 36335 }, { "epoch": 3.866354543519898, "grad_norm": 15.453125, "learning_rate": 8.726691301876067e-06, "loss": 0.7917, "step": 36336 }, { "epoch": 3.8664609491381143, "grad_norm": 15.015625, "learning_rate": 8.726655770324049e-06, "loss": 0.7501, "step": 36337 }, { "epoch": 3.8665673547563313, "grad_norm": 11.8671875, "learning_rate": 8.72662023877203e-06, "loss": 0.525, "step": 36338 }, { "epoch": 3.866673760374548, "grad_norm": 14.609375, "learning_rate": 8.726584707220012e-06, "loss": 0.6153, "step": 36339 }, { "epoch": 3.8667801659927643, "grad_norm": 13.1328125, "learning_rate": 8.726549175667995e-06, "loss": 0.6228, "step": 36340 }, { "epoch": 3.866886571610981, "grad_norm": 11.609375, "learning_rate": 8.726513644115975e-06, "loss": 0.6368, "step": 36341 }, { "epoch": 3.8669929772291978, "grad_norm": 12.46875, "learning_rate": 8.726478112563957e-06, "loss": 0.5767, "step": 36342 }, { "epoch": 3.8670993828474143, "grad_norm": 14.296875, "learning_rate": 8.726442581011938e-06, "loss": 0.8124, "step": 36343 }, { "epoch": 3.867205788465631, "grad_norm": 11.1484375, "learning_rate": 8.72640704945992e-06, "loss": 0.6925, "step": 36344 }, { "epoch": 3.8673121940838477, "grad_norm": 13.703125, "learning_rate": 8.726371517907903e-06, "loss": 0.6487, "step": 36345 }, { "epoch": 3.8674185997020643, "grad_norm": 11.5703125, "learning_rate": 8.726335986355885e-06, "loss": 0.6615, "step": 36346 }, { "epoch": 3.8675250053202808, "grad_norm": 11.7421875, "learning_rate": 8.726300454803866e-06, "loss": 0.6634, "step": 36347 }, { "epoch": 3.8676314109384977, "grad_norm": 15.3046875, "learning_rate": 8.726264923251848e-06, "loss": 0.6752, "step": 36348 }, { "epoch": 3.8677378165567142, "grad_norm": 12.7890625, "learning_rate": 8.72622939169983e-06, "loss": 0.6893, "step": 36349 }, { "epoch": 3.8678442221749307, "grad_norm": 10.734375, "learning_rate": 8.726193860147813e-06, "loss": 0.6195, "step": 36350 }, { "epoch": 3.8679506277931477, "grad_norm": 12.578125, "learning_rate": 8.726158328595795e-06, "loss": 0.7047, "step": 36351 }, { "epoch": 3.868057033411364, "grad_norm": 10.9296875, "learning_rate": 8.726122797043776e-06, "loss": 0.527, "step": 36352 }, { "epoch": 3.8681634390295807, "grad_norm": 9.5703125, "learning_rate": 8.726087265491758e-06, "loss": 0.6727, "step": 36353 }, { "epoch": 3.8682698446477977, "grad_norm": 13.546875, "learning_rate": 8.72605173393974e-06, "loss": 0.6398, "step": 36354 }, { "epoch": 3.868376250266014, "grad_norm": 12.5703125, "learning_rate": 8.726016202387721e-06, "loss": 0.6559, "step": 36355 }, { "epoch": 3.8684826558842307, "grad_norm": 14.1796875, "learning_rate": 8.725980670835703e-06, "loss": 0.6889, "step": 36356 }, { "epoch": 3.868589061502447, "grad_norm": 15.2578125, "learning_rate": 8.725945139283684e-06, "loss": 0.8301, "step": 36357 }, { "epoch": 3.8686954671206637, "grad_norm": 10.65625, "learning_rate": 8.725909607731666e-06, "loss": 0.5848, "step": 36358 }, { "epoch": 3.8688018727388807, "grad_norm": 13.828125, "learning_rate": 8.725874076179647e-06, "loss": 0.7528, "step": 36359 }, { "epoch": 3.868908278357097, "grad_norm": 15.8515625, "learning_rate": 8.72583854462763e-06, "loss": 0.751, "step": 36360 }, { "epoch": 3.8690146839753137, "grad_norm": 14.8984375, "learning_rate": 8.725803013075612e-06, "loss": 0.7342, "step": 36361 }, { "epoch": 3.8691210895935306, "grad_norm": 10.9296875, "learning_rate": 8.725767481523594e-06, "loss": 0.7114, "step": 36362 }, { "epoch": 3.869227495211747, "grad_norm": 17.09375, "learning_rate": 8.725731949971575e-06, "loss": 0.8711, "step": 36363 }, { "epoch": 3.8693339008299636, "grad_norm": 13.359375, "learning_rate": 8.725696418419557e-06, "loss": 0.6429, "step": 36364 }, { "epoch": 3.8694403064481806, "grad_norm": 18.765625, "learning_rate": 8.725660886867539e-06, "loss": 0.6828, "step": 36365 }, { "epoch": 3.869546712066397, "grad_norm": 12.2421875, "learning_rate": 8.725625355315522e-06, "loss": 0.6115, "step": 36366 }, { "epoch": 3.8696531176846136, "grad_norm": 13.53125, "learning_rate": 8.725589823763504e-06, "loss": 0.6888, "step": 36367 }, { "epoch": 3.8697595233028306, "grad_norm": 14.9609375, "learning_rate": 8.725554292211485e-06, "loss": 0.8824, "step": 36368 }, { "epoch": 3.869865928921047, "grad_norm": 13.328125, "learning_rate": 8.725518760659467e-06, "loss": 0.7012, "step": 36369 }, { "epoch": 3.8699723345392636, "grad_norm": 12.8828125, "learning_rate": 8.725483229107448e-06, "loss": 0.7827, "step": 36370 }, { "epoch": 3.8700787401574805, "grad_norm": 13.3671875, "learning_rate": 8.72544769755543e-06, "loss": 0.6719, "step": 36371 }, { "epoch": 3.870185145775697, "grad_norm": 16.8125, "learning_rate": 8.725412166003412e-06, "loss": 0.8709, "step": 36372 }, { "epoch": 3.8702915513939136, "grad_norm": 12.1796875, "learning_rate": 8.725376634451393e-06, "loss": 0.7998, "step": 36373 }, { "epoch": 3.87039795701213, "grad_norm": 16.140625, "learning_rate": 8.725341102899375e-06, "loss": 0.7736, "step": 36374 }, { "epoch": 3.870504362630347, "grad_norm": 12.0390625, "learning_rate": 8.725305571347356e-06, "loss": 0.6317, "step": 36375 }, { "epoch": 3.8706107682485635, "grad_norm": 13.546875, "learning_rate": 8.72527003979534e-06, "loss": 0.6793, "step": 36376 }, { "epoch": 3.87071717386678, "grad_norm": 13.1484375, "learning_rate": 8.725234508243321e-06, "loss": 0.6325, "step": 36377 }, { "epoch": 3.8708235794849966, "grad_norm": 15.390625, "learning_rate": 8.725198976691303e-06, "loss": 0.674, "step": 36378 }, { "epoch": 3.8709299851032135, "grad_norm": 14.7734375, "learning_rate": 8.725163445139285e-06, "loss": 0.7836, "step": 36379 }, { "epoch": 3.87103639072143, "grad_norm": 13.7890625, "learning_rate": 8.725127913587266e-06, "loss": 0.755, "step": 36380 }, { "epoch": 3.8711427963396465, "grad_norm": 13.4296875, "learning_rate": 8.725092382035248e-06, "loss": 0.5883, "step": 36381 }, { "epoch": 3.8712492019578635, "grad_norm": 13.90625, "learning_rate": 8.72505685048323e-06, "loss": 0.736, "step": 36382 }, { "epoch": 3.87135560757608, "grad_norm": 11.234375, "learning_rate": 8.725021318931211e-06, "loss": 0.6649, "step": 36383 }, { "epoch": 3.8714620131942965, "grad_norm": 10.0, "learning_rate": 8.724985787379193e-06, "loss": 0.6507, "step": 36384 }, { "epoch": 3.8715684188125135, "grad_norm": 10.90625, "learning_rate": 8.724950255827174e-06, "loss": 0.701, "step": 36385 }, { "epoch": 3.87167482443073, "grad_norm": 11.7890625, "learning_rate": 8.724914724275158e-06, "loss": 0.6882, "step": 36386 }, { "epoch": 3.8717812300489465, "grad_norm": 12.5625, "learning_rate": 8.724879192723139e-06, "loss": 0.6659, "step": 36387 }, { "epoch": 3.8718876356671634, "grad_norm": 15.484375, "learning_rate": 8.72484366117112e-06, "loss": 0.6873, "step": 36388 }, { "epoch": 3.87199404128538, "grad_norm": 14.171875, "learning_rate": 8.724808129619102e-06, "loss": 0.623, "step": 36389 }, { "epoch": 3.8721004469035964, "grad_norm": 13.1640625, "learning_rate": 8.724772598067084e-06, "loss": 0.7088, "step": 36390 }, { "epoch": 3.8722068525218134, "grad_norm": 13.3984375, "learning_rate": 8.724737066515067e-06, "loss": 0.7361, "step": 36391 }, { "epoch": 3.87231325814003, "grad_norm": 12.2734375, "learning_rate": 8.724701534963049e-06, "loss": 0.619, "step": 36392 }, { "epoch": 3.8724196637582464, "grad_norm": 11.90625, "learning_rate": 8.72466600341103e-06, "loss": 0.578, "step": 36393 }, { "epoch": 3.872526069376463, "grad_norm": 19.8125, "learning_rate": 8.724630471859012e-06, "loss": 0.8536, "step": 36394 }, { "epoch": 3.87263247499468, "grad_norm": 12.53125, "learning_rate": 8.724594940306994e-06, "loss": 0.7899, "step": 36395 }, { "epoch": 3.8727388806128964, "grad_norm": 14.640625, "learning_rate": 8.724559408754975e-06, "loss": 0.6909, "step": 36396 }, { "epoch": 3.872845286231113, "grad_norm": 14.0546875, "learning_rate": 8.724523877202957e-06, "loss": 0.6821, "step": 36397 }, { "epoch": 3.8729516918493294, "grad_norm": 13.3125, "learning_rate": 8.724488345650939e-06, "loss": 0.7565, "step": 36398 }, { "epoch": 3.8730580974675464, "grad_norm": 13.140625, "learning_rate": 8.72445281409892e-06, "loss": 0.544, "step": 36399 }, { "epoch": 3.873164503085763, "grad_norm": 13.703125, "learning_rate": 8.724417282546902e-06, "loss": 0.6487, "step": 36400 }, { "epoch": 3.8732709087039794, "grad_norm": 13.8125, "learning_rate": 8.724381750994885e-06, "loss": 0.7259, "step": 36401 }, { "epoch": 3.8733773143221963, "grad_norm": 11.8828125, "learning_rate": 8.724346219442867e-06, "loss": 0.6359, "step": 36402 }, { "epoch": 3.873483719940413, "grad_norm": 16.453125, "learning_rate": 8.724310687890848e-06, "loss": 0.6874, "step": 36403 }, { "epoch": 3.8735901255586294, "grad_norm": 13.75, "learning_rate": 8.72427515633883e-06, "loss": 0.7986, "step": 36404 }, { "epoch": 3.8736965311768463, "grad_norm": 14.203125, "learning_rate": 8.724239624786811e-06, "loss": 0.7017, "step": 36405 }, { "epoch": 3.873802936795063, "grad_norm": 13.859375, "learning_rate": 8.724204093234793e-06, "loss": 0.6859, "step": 36406 }, { "epoch": 3.8739093424132793, "grad_norm": 17.421875, "learning_rate": 8.724168561682775e-06, "loss": 0.7645, "step": 36407 }, { "epoch": 3.8740157480314963, "grad_norm": 15.328125, "learning_rate": 8.724133030130756e-06, "loss": 0.6978, "step": 36408 }, { "epoch": 3.874122153649713, "grad_norm": 12.1796875, "learning_rate": 8.724097498578738e-06, "loss": 0.7194, "step": 36409 }, { "epoch": 3.8742285592679293, "grad_norm": 10.4609375, "learning_rate": 8.72406196702672e-06, "loss": 0.7206, "step": 36410 }, { "epoch": 3.8743349648861463, "grad_norm": 12.890625, "learning_rate": 8.724026435474701e-06, "loss": 0.6085, "step": 36411 }, { "epoch": 3.8744413705043628, "grad_norm": 12.25, "learning_rate": 8.723990903922684e-06, "loss": 0.6883, "step": 36412 }, { "epoch": 3.8745477761225793, "grad_norm": 13.1796875, "learning_rate": 8.723955372370666e-06, "loss": 0.601, "step": 36413 }, { "epoch": 3.874654181740796, "grad_norm": 15.6171875, "learning_rate": 8.723919840818648e-06, "loss": 0.7137, "step": 36414 }, { "epoch": 3.8747605873590123, "grad_norm": 8.5625, "learning_rate": 8.72388430926663e-06, "loss": 0.4081, "step": 36415 }, { "epoch": 3.8748669929772293, "grad_norm": 11.140625, "learning_rate": 8.72384877771461e-06, "loss": 0.659, "step": 36416 }, { "epoch": 3.8749733985954458, "grad_norm": 10.1796875, "learning_rate": 8.723813246162594e-06, "loss": 0.613, "step": 36417 }, { "epoch": 3.8750798042136623, "grad_norm": 10.515625, "learning_rate": 8.723777714610576e-06, "loss": 0.6112, "step": 36418 }, { "epoch": 3.8751862098318792, "grad_norm": 11.453125, "learning_rate": 8.723742183058557e-06, "loss": 0.656, "step": 36419 }, { "epoch": 3.8752926154500957, "grad_norm": 15.21875, "learning_rate": 8.723706651506539e-06, "loss": 0.7673, "step": 36420 }, { "epoch": 3.8753990210683122, "grad_norm": 13.734375, "learning_rate": 8.72367111995452e-06, "loss": 0.6839, "step": 36421 }, { "epoch": 3.875505426686529, "grad_norm": 12.953125, "learning_rate": 8.723635588402502e-06, "loss": 0.6559, "step": 36422 }, { "epoch": 3.8756118323047457, "grad_norm": 11.3125, "learning_rate": 8.723600056850484e-06, "loss": 0.6011, "step": 36423 }, { "epoch": 3.875718237922962, "grad_norm": 9.78125, "learning_rate": 8.723564525298465e-06, "loss": 0.6035, "step": 36424 }, { "epoch": 3.875824643541179, "grad_norm": 15.4921875, "learning_rate": 8.723528993746447e-06, "loss": 0.6155, "step": 36425 }, { "epoch": 3.8759310491593957, "grad_norm": 13.890625, "learning_rate": 8.723493462194429e-06, "loss": 0.6828, "step": 36426 }, { "epoch": 3.876037454777612, "grad_norm": 13.96875, "learning_rate": 8.723457930642412e-06, "loss": 0.619, "step": 36427 }, { "epoch": 3.876143860395829, "grad_norm": 14.359375, "learning_rate": 8.723422399090393e-06, "loss": 0.7634, "step": 36428 }, { "epoch": 3.8762502660140457, "grad_norm": 9.8671875, "learning_rate": 8.723386867538375e-06, "loss": 0.5413, "step": 36429 }, { "epoch": 3.876356671632262, "grad_norm": 12.7578125, "learning_rate": 8.723351335986357e-06, "loss": 0.5063, "step": 36430 }, { "epoch": 3.8764630772504787, "grad_norm": 12.234375, "learning_rate": 8.723315804434338e-06, "loss": 0.758, "step": 36431 }, { "epoch": 3.8765694828686956, "grad_norm": 10.421875, "learning_rate": 8.72328027288232e-06, "loss": 0.6209, "step": 36432 }, { "epoch": 3.876675888486912, "grad_norm": 15.203125, "learning_rate": 8.723244741330302e-06, "loss": 0.6575, "step": 36433 }, { "epoch": 3.8767822941051286, "grad_norm": 13.3984375, "learning_rate": 8.723209209778283e-06, "loss": 0.6789, "step": 36434 }, { "epoch": 3.876888699723345, "grad_norm": 16.546875, "learning_rate": 8.723173678226265e-06, "loss": 0.7806, "step": 36435 }, { "epoch": 3.876995105341562, "grad_norm": 12.421875, "learning_rate": 8.723138146674246e-06, "loss": 0.6258, "step": 36436 }, { "epoch": 3.8771015109597786, "grad_norm": 13.3984375, "learning_rate": 8.72310261512223e-06, "loss": 0.6802, "step": 36437 }, { "epoch": 3.877207916577995, "grad_norm": 12.671875, "learning_rate": 8.723067083570211e-06, "loss": 0.663, "step": 36438 }, { "epoch": 3.877314322196212, "grad_norm": 10.390625, "learning_rate": 8.723031552018193e-06, "loss": 0.5902, "step": 36439 }, { "epoch": 3.8774207278144286, "grad_norm": 14.1484375, "learning_rate": 8.722996020466174e-06, "loss": 0.7015, "step": 36440 }, { "epoch": 3.877527133432645, "grad_norm": 15.984375, "learning_rate": 8.722960488914156e-06, "loss": 0.6667, "step": 36441 }, { "epoch": 3.877633539050862, "grad_norm": 13.7421875, "learning_rate": 8.72292495736214e-06, "loss": 0.7441, "step": 36442 }, { "epoch": 3.8777399446690786, "grad_norm": 16.40625, "learning_rate": 8.722889425810121e-06, "loss": 0.6647, "step": 36443 }, { "epoch": 3.877846350287295, "grad_norm": 10.7421875, "learning_rate": 8.722853894258103e-06, "loss": 0.5732, "step": 36444 }, { "epoch": 3.877952755905512, "grad_norm": 14.296875, "learning_rate": 8.722818362706084e-06, "loss": 0.7258, "step": 36445 }, { "epoch": 3.8780591615237285, "grad_norm": 10.1171875, "learning_rate": 8.722782831154066e-06, "loss": 0.5427, "step": 36446 }, { "epoch": 3.878165567141945, "grad_norm": 12.640625, "learning_rate": 8.722747299602047e-06, "loss": 0.6328, "step": 36447 }, { "epoch": 3.878271972760162, "grad_norm": 13.96875, "learning_rate": 8.722711768050029e-06, "loss": 0.7439, "step": 36448 }, { "epoch": 3.8783783783783785, "grad_norm": 15.796875, "learning_rate": 8.72267623649801e-06, "loss": 0.5956, "step": 36449 }, { "epoch": 3.878484783996595, "grad_norm": 14.3203125, "learning_rate": 8.722640704945992e-06, "loss": 0.6537, "step": 36450 }, { "epoch": 3.8785911896148115, "grad_norm": 13.3125, "learning_rate": 8.722605173393974e-06, "loss": 0.5935, "step": 36451 }, { "epoch": 3.878697595233028, "grad_norm": 11.6484375, "learning_rate": 8.722569641841955e-06, "loss": 0.5815, "step": 36452 }, { "epoch": 3.878804000851245, "grad_norm": 11.8515625, "learning_rate": 8.722534110289939e-06, "loss": 0.577, "step": 36453 }, { "epoch": 3.8789104064694615, "grad_norm": 12.609375, "learning_rate": 8.72249857873792e-06, "loss": 0.5716, "step": 36454 }, { "epoch": 3.879016812087678, "grad_norm": 14.8203125, "learning_rate": 8.722463047185902e-06, "loss": 0.7003, "step": 36455 }, { "epoch": 3.879123217705895, "grad_norm": 15.8203125, "learning_rate": 8.722427515633884e-06, "loss": 0.8013, "step": 36456 }, { "epoch": 3.8792296233241115, "grad_norm": 15.0859375, "learning_rate": 8.722391984081865e-06, "loss": 0.6941, "step": 36457 }, { "epoch": 3.879336028942328, "grad_norm": 19.0625, "learning_rate": 8.722356452529848e-06, "loss": 0.8133, "step": 36458 }, { "epoch": 3.879442434560545, "grad_norm": 14.46875, "learning_rate": 8.72232092097783e-06, "loss": 0.712, "step": 36459 }, { "epoch": 3.8795488401787614, "grad_norm": 21.671875, "learning_rate": 8.72228538942581e-06, "loss": 0.7396, "step": 36460 }, { "epoch": 3.879655245796978, "grad_norm": 18.71875, "learning_rate": 8.722249857873792e-06, "loss": 0.7658, "step": 36461 }, { "epoch": 3.879761651415195, "grad_norm": 11.9765625, "learning_rate": 8.722214326321773e-06, "loss": 0.674, "step": 36462 }, { "epoch": 3.8798680570334114, "grad_norm": 15.7734375, "learning_rate": 8.722178794769756e-06, "loss": 0.6911, "step": 36463 }, { "epoch": 3.879974462651628, "grad_norm": 10.953125, "learning_rate": 8.722143263217738e-06, "loss": 0.646, "step": 36464 }, { "epoch": 3.880080868269845, "grad_norm": 15.921875, "learning_rate": 8.72210773166572e-06, "loss": 0.8229, "step": 36465 }, { "epoch": 3.8801872738880614, "grad_norm": 11.171875, "learning_rate": 8.722072200113701e-06, "loss": 0.5723, "step": 36466 }, { "epoch": 3.880293679506278, "grad_norm": 10.40625, "learning_rate": 8.722036668561683e-06, "loss": 0.6607, "step": 36467 }, { "epoch": 3.8804000851244944, "grad_norm": 12.0390625, "learning_rate": 8.722001137009666e-06, "loss": 0.7295, "step": 36468 }, { "epoch": 3.8805064907427114, "grad_norm": 16.765625, "learning_rate": 8.721965605457648e-06, "loss": 0.7962, "step": 36469 }, { "epoch": 3.880612896360928, "grad_norm": 17.90625, "learning_rate": 8.72193007390563e-06, "loss": 0.7703, "step": 36470 }, { "epoch": 3.8807193019791444, "grad_norm": 12.171875, "learning_rate": 8.721894542353611e-06, "loss": 0.5355, "step": 36471 }, { "epoch": 3.880825707597361, "grad_norm": 12.8359375, "learning_rate": 8.721859010801593e-06, "loss": 0.6374, "step": 36472 }, { "epoch": 3.880932113215578, "grad_norm": 13.203125, "learning_rate": 8.721823479249574e-06, "loss": 0.7361, "step": 36473 }, { "epoch": 3.8810385188337944, "grad_norm": 9.4921875, "learning_rate": 8.721787947697556e-06, "loss": 0.6009, "step": 36474 }, { "epoch": 3.881144924452011, "grad_norm": 13.9609375, "learning_rate": 8.721752416145537e-06, "loss": 0.6624, "step": 36475 }, { "epoch": 3.881251330070228, "grad_norm": 14.4375, "learning_rate": 8.721716884593519e-06, "loss": 0.6611, "step": 36476 }, { "epoch": 3.8813577356884443, "grad_norm": 11.0078125, "learning_rate": 8.7216813530415e-06, "loss": 0.6995, "step": 36477 }, { "epoch": 3.881464141306661, "grad_norm": 14.640625, "learning_rate": 8.721645821489484e-06, "loss": 0.7029, "step": 36478 }, { "epoch": 3.881570546924878, "grad_norm": 17.453125, "learning_rate": 8.721610289937466e-06, "loss": 0.8183, "step": 36479 }, { "epoch": 3.8816769525430943, "grad_norm": 15.6640625, "learning_rate": 8.721574758385447e-06, "loss": 0.719, "step": 36480 }, { "epoch": 3.881783358161311, "grad_norm": 14.6640625, "learning_rate": 8.721539226833429e-06, "loss": 0.664, "step": 36481 }, { "epoch": 3.8818897637795278, "grad_norm": 9.75, "learning_rate": 8.72150369528141e-06, "loss": 0.477, "step": 36482 }, { "epoch": 3.8819961693977443, "grad_norm": 11.7890625, "learning_rate": 8.721468163729392e-06, "loss": 0.6394, "step": 36483 }, { "epoch": 3.882102575015961, "grad_norm": 12.1484375, "learning_rate": 8.721432632177375e-06, "loss": 0.6615, "step": 36484 }, { "epoch": 3.8822089806341777, "grad_norm": 12.9921875, "learning_rate": 8.721397100625357e-06, "loss": 0.5624, "step": 36485 }, { "epoch": 3.8823153862523943, "grad_norm": 12.953125, "learning_rate": 8.721361569073339e-06, "loss": 0.7279, "step": 36486 }, { "epoch": 3.8824217918706108, "grad_norm": 17.1875, "learning_rate": 8.72132603752132e-06, "loss": 0.7608, "step": 36487 }, { "epoch": 3.8825281974888273, "grad_norm": 15.1796875, "learning_rate": 8.721290505969302e-06, "loss": 0.7109, "step": 36488 }, { "epoch": 3.882634603107044, "grad_norm": 12.90625, "learning_rate": 8.721254974417283e-06, "loss": 0.6198, "step": 36489 }, { "epoch": 3.8827410087252607, "grad_norm": 16.8125, "learning_rate": 8.721219442865265e-06, "loss": 0.8152, "step": 36490 }, { "epoch": 3.8828474143434772, "grad_norm": 11.96875, "learning_rate": 8.721183911313247e-06, "loss": 0.5416, "step": 36491 }, { "epoch": 3.8829538199616938, "grad_norm": 16.578125, "learning_rate": 8.721148379761228e-06, "loss": 0.8037, "step": 36492 }, { "epoch": 3.8830602255799107, "grad_norm": 13.9609375, "learning_rate": 8.72111284820921e-06, "loss": 0.7098, "step": 36493 }, { "epoch": 3.883166631198127, "grad_norm": 13.109375, "learning_rate": 8.721077316657193e-06, "loss": 0.7976, "step": 36494 }, { "epoch": 3.8832730368163437, "grad_norm": 11.3515625, "learning_rate": 8.721041785105175e-06, "loss": 0.5851, "step": 36495 }, { "epoch": 3.8833794424345607, "grad_norm": 12.71875, "learning_rate": 8.721006253553156e-06, "loss": 0.6399, "step": 36496 }, { "epoch": 3.883485848052777, "grad_norm": 11.5859375, "learning_rate": 8.720970722001138e-06, "loss": 0.6018, "step": 36497 }, { "epoch": 3.8835922536709937, "grad_norm": 11.0, "learning_rate": 8.72093519044912e-06, "loss": 0.8001, "step": 36498 }, { "epoch": 3.8836986592892107, "grad_norm": 12.28125, "learning_rate": 8.720899658897101e-06, "loss": 0.5907, "step": 36499 }, { "epoch": 3.883805064907427, "grad_norm": 11.125, "learning_rate": 8.720864127345083e-06, "loss": 0.6367, "step": 36500 }, { "epoch": 3.8839114705256437, "grad_norm": 11.65625, "learning_rate": 8.720828595793064e-06, "loss": 0.6876, "step": 36501 }, { "epoch": 3.8840178761438606, "grad_norm": 10.3046875, "learning_rate": 8.720793064241046e-06, "loss": 0.5782, "step": 36502 }, { "epoch": 3.884124281762077, "grad_norm": 14.1171875, "learning_rate": 8.720757532689028e-06, "loss": 0.6595, "step": 36503 }, { "epoch": 3.8842306873802936, "grad_norm": 14.609375, "learning_rate": 8.72072200113701e-06, "loss": 0.9048, "step": 36504 }, { "epoch": 3.8843370929985106, "grad_norm": 14.5078125, "learning_rate": 8.720686469584992e-06, "loss": 0.7185, "step": 36505 }, { "epoch": 3.884443498616727, "grad_norm": 15.265625, "learning_rate": 8.720650938032974e-06, "loss": 0.8628, "step": 36506 }, { "epoch": 3.8845499042349436, "grad_norm": 17.9375, "learning_rate": 8.720615406480956e-06, "loss": 0.8543, "step": 36507 }, { "epoch": 3.88465630985316, "grad_norm": 15.671875, "learning_rate": 8.720579874928937e-06, "loss": 0.8034, "step": 36508 }, { "epoch": 3.8847627154713766, "grad_norm": 14.1640625, "learning_rate": 8.72054434337692e-06, "loss": 0.7955, "step": 36509 }, { "epoch": 3.8848691210895936, "grad_norm": 12.5, "learning_rate": 8.720508811824902e-06, "loss": 0.5925, "step": 36510 }, { "epoch": 3.88497552670781, "grad_norm": 16.0, "learning_rate": 8.720473280272884e-06, "loss": 0.7738, "step": 36511 }, { "epoch": 3.8850819323260266, "grad_norm": 16.234375, "learning_rate": 8.720437748720865e-06, "loss": 0.7709, "step": 36512 }, { "epoch": 3.8851883379442436, "grad_norm": 12.890625, "learning_rate": 8.720402217168847e-06, "loss": 0.7429, "step": 36513 }, { "epoch": 3.88529474356246, "grad_norm": 13.0703125, "learning_rate": 8.720366685616829e-06, "loss": 0.5877, "step": 36514 }, { "epoch": 3.8854011491806766, "grad_norm": 11.78125, "learning_rate": 8.72033115406481e-06, "loss": 0.6541, "step": 36515 }, { "epoch": 3.8855075547988935, "grad_norm": 12.421875, "learning_rate": 8.720295622512792e-06, "loss": 0.5899, "step": 36516 }, { "epoch": 3.88561396041711, "grad_norm": 13.0546875, "learning_rate": 8.720260090960773e-06, "loss": 0.6498, "step": 36517 }, { "epoch": 3.8857203660353266, "grad_norm": 10.375, "learning_rate": 8.720224559408755e-06, "loss": 0.5573, "step": 36518 }, { "epoch": 3.8858267716535435, "grad_norm": 12.796875, "learning_rate": 8.720189027856738e-06, "loss": 0.5898, "step": 36519 }, { "epoch": 3.88593317727176, "grad_norm": 20.09375, "learning_rate": 8.72015349630472e-06, "loss": 0.8927, "step": 36520 }, { "epoch": 3.8860395828899765, "grad_norm": 11.25, "learning_rate": 8.720117964752702e-06, "loss": 0.6123, "step": 36521 }, { "epoch": 3.8861459885081935, "grad_norm": 13.390625, "learning_rate": 8.720082433200683e-06, "loss": 0.6704, "step": 36522 }, { "epoch": 3.88625239412641, "grad_norm": 12.265625, "learning_rate": 8.720046901648665e-06, "loss": 0.6715, "step": 36523 }, { "epoch": 3.8863587997446265, "grad_norm": 9.7265625, "learning_rate": 8.720011370096646e-06, "loss": 0.5193, "step": 36524 }, { "epoch": 3.886465205362843, "grad_norm": 12.09375, "learning_rate": 8.719975838544628e-06, "loss": 0.6095, "step": 36525 }, { "epoch": 3.88657161098106, "grad_norm": 19.890625, "learning_rate": 8.71994030699261e-06, "loss": 0.8359, "step": 36526 }, { "epoch": 3.8866780165992765, "grad_norm": 13.25, "learning_rate": 8.719904775440591e-06, "loss": 0.6019, "step": 36527 }, { "epoch": 3.886784422217493, "grad_norm": 11.109375, "learning_rate": 8.719869243888573e-06, "loss": 0.4913, "step": 36528 }, { "epoch": 3.8868908278357095, "grad_norm": 14.296875, "learning_rate": 8.719833712336554e-06, "loss": 0.6944, "step": 36529 }, { "epoch": 3.8869972334539264, "grad_norm": 11.890625, "learning_rate": 8.719798180784538e-06, "loss": 0.6425, "step": 36530 }, { "epoch": 3.887103639072143, "grad_norm": 11.4609375, "learning_rate": 8.71976264923252e-06, "loss": 0.5994, "step": 36531 }, { "epoch": 3.8872100446903595, "grad_norm": 13.6640625, "learning_rate": 8.719727117680501e-06, "loss": 0.7323, "step": 36532 }, { "epoch": 3.8873164503085764, "grad_norm": 12.8359375, "learning_rate": 8.719691586128483e-06, "loss": 0.6718, "step": 36533 }, { "epoch": 3.887422855926793, "grad_norm": 11.46875, "learning_rate": 8.719656054576464e-06, "loss": 0.6582, "step": 36534 }, { "epoch": 3.8875292615450094, "grad_norm": 12.0625, "learning_rate": 8.719620523024447e-06, "loss": 0.5966, "step": 36535 }, { "epoch": 3.8876356671632264, "grad_norm": 16.28125, "learning_rate": 8.719584991472429e-06, "loss": 0.7855, "step": 36536 }, { "epoch": 3.887742072781443, "grad_norm": 10.59375, "learning_rate": 8.71954945992041e-06, "loss": 0.6175, "step": 36537 }, { "epoch": 3.8878484783996594, "grad_norm": 14.9921875, "learning_rate": 8.719513928368392e-06, "loss": 0.8591, "step": 36538 }, { "epoch": 3.8879548840178764, "grad_norm": 10.9921875, "learning_rate": 8.719478396816374e-06, "loss": 0.6372, "step": 36539 }, { "epoch": 3.888061289636093, "grad_norm": 13.015625, "learning_rate": 8.719442865264355e-06, "loss": 0.7217, "step": 36540 }, { "epoch": 3.8881676952543094, "grad_norm": 16.390625, "learning_rate": 8.719407333712337e-06, "loss": 0.7472, "step": 36541 }, { "epoch": 3.8882741008725263, "grad_norm": 15.5, "learning_rate": 8.719371802160319e-06, "loss": 0.8156, "step": 36542 }, { "epoch": 3.888380506490743, "grad_norm": 12.3203125, "learning_rate": 8.7193362706083e-06, "loss": 0.6677, "step": 36543 }, { "epoch": 3.8884869121089594, "grad_norm": 14.8515625, "learning_rate": 8.719300739056282e-06, "loss": 0.7276, "step": 36544 }, { "epoch": 3.888593317727176, "grad_norm": 13.9609375, "learning_rate": 8.719265207504265e-06, "loss": 0.6651, "step": 36545 }, { "epoch": 3.8886997233453924, "grad_norm": 12.28125, "learning_rate": 8.719229675952247e-06, "loss": 0.7147, "step": 36546 }, { "epoch": 3.8888061289636093, "grad_norm": 15.7265625, "learning_rate": 8.719194144400228e-06, "loss": 0.7371, "step": 36547 }, { "epoch": 3.888912534581826, "grad_norm": 10.8984375, "learning_rate": 8.71915861284821e-06, "loss": 0.4987, "step": 36548 }, { "epoch": 3.8890189402000424, "grad_norm": 10.5859375, "learning_rate": 8.719123081296192e-06, "loss": 0.5896, "step": 36549 }, { "epoch": 3.8891253458182593, "grad_norm": 14.4296875, "learning_rate": 8.719087549744173e-06, "loss": 0.6792, "step": 36550 }, { "epoch": 3.889231751436476, "grad_norm": 13.5078125, "learning_rate": 8.719052018192155e-06, "loss": 0.7783, "step": 36551 }, { "epoch": 3.8893381570546923, "grad_norm": 13.578125, "learning_rate": 8.719016486640136e-06, "loss": 0.6949, "step": 36552 }, { "epoch": 3.8894445626729093, "grad_norm": 12.015625, "learning_rate": 8.718980955088118e-06, "loss": 0.5536, "step": 36553 }, { "epoch": 3.889550968291126, "grad_norm": 14.3203125, "learning_rate": 8.7189454235361e-06, "loss": 0.6531, "step": 36554 }, { "epoch": 3.8896573739093423, "grad_norm": 13.140625, "learning_rate": 8.718909891984083e-06, "loss": 0.6211, "step": 36555 }, { "epoch": 3.8897637795275593, "grad_norm": 12.8515625, "learning_rate": 8.718874360432065e-06, "loss": 0.6745, "step": 36556 }, { "epoch": 3.8898701851457758, "grad_norm": 12.1484375, "learning_rate": 8.718838828880046e-06, "loss": 0.6181, "step": 36557 }, { "epoch": 3.8899765907639923, "grad_norm": 13.0078125, "learning_rate": 8.718803297328028e-06, "loss": 0.6339, "step": 36558 }, { "epoch": 3.8900829963822092, "grad_norm": 13.78125, "learning_rate": 8.71876776577601e-06, "loss": 0.8205, "step": 36559 }, { "epoch": 3.8901894020004257, "grad_norm": 13.5859375, "learning_rate": 8.718732234223991e-06, "loss": 0.6365, "step": 36560 }, { "epoch": 3.8902958076186422, "grad_norm": 10.890625, "learning_rate": 8.718696702671974e-06, "loss": 0.6792, "step": 36561 }, { "epoch": 3.8904022132368588, "grad_norm": 20.421875, "learning_rate": 8.718661171119956e-06, "loss": 0.7254, "step": 36562 }, { "epoch": 3.8905086188550757, "grad_norm": 11.4140625, "learning_rate": 8.718625639567937e-06, "loss": 0.6096, "step": 36563 }, { "epoch": 3.890615024473292, "grad_norm": 12.3515625, "learning_rate": 8.718590108015919e-06, "loss": 0.697, "step": 36564 }, { "epoch": 3.8907214300915087, "grad_norm": 15.125, "learning_rate": 8.7185545764639e-06, "loss": 0.6636, "step": 36565 }, { "epoch": 3.8908278357097252, "grad_norm": 15.1484375, "learning_rate": 8.718519044911882e-06, "loss": 0.7351, "step": 36566 }, { "epoch": 3.890934241327942, "grad_norm": 12.5703125, "learning_rate": 8.718483513359864e-06, "loss": 0.7282, "step": 36567 }, { "epoch": 3.8910406469461587, "grad_norm": 10.03125, "learning_rate": 8.718447981807846e-06, "loss": 0.477, "step": 36568 }, { "epoch": 3.891147052564375, "grad_norm": 15.5625, "learning_rate": 8.718412450255827e-06, "loss": 0.6576, "step": 36569 }, { "epoch": 3.891253458182592, "grad_norm": 16.09375, "learning_rate": 8.718376918703809e-06, "loss": 0.69, "step": 36570 }, { "epoch": 3.8913598638008087, "grad_norm": 11.328125, "learning_rate": 8.718341387151792e-06, "loss": 0.7834, "step": 36571 }, { "epoch": 3.891466269419025, "grad_norm": 11.2265625, "learning_rate": 8.718305855599774e-06, "loss": 0.6925, "step": 36572 }, { "epoch": 3.891572675037242, "grad_norm": 11.953125, "learning_rate": 8.718270324047755e-06, "loss": 0.773, "step": 36573 }, { "epoch": 3.8916790806554586, "grad_norm": 12.78125, "learning_rate": 8.718234792495737e-06, "loss": 0.6362, "step": 36574 }, { "epoch": 3.891785486273675, "grad_norm": 20.03125, "learning_rate": 8.718199260943718e-06, "loss": 0.9025, "step": 36575 }, { "epoch": 3.891891891891892, "grad_norm": 16.6875, "learning_rate": 8.718163729391702e-06, "loss": 0.8724, "step": 36576 }, { "epoch": 3.8919982975101086, "grad_norm": 11.734375, "learning_rate": 8.718128197839683e-06, "loss": 0.6413, "step": 36577 }, { "epoch": 3.892104703128325, "grad_norm": 12.6171875, "learning_rate": 8.718092666287665e-06, "loss": 0.6873, "step": 36578 }, { "epoch": 3.892211108746542, "grad_norm": 11.5078125, "learning_rate": 8.718057134735647e-06, "loss": 0.5014, "step": 36579 }, { "epoch": 3.8923175143647586, "grad_norm": 13.0390625, "learning_rate": 8.718021603183626e-06, "loss": 0.6101, "step": 36580 }, { "epoch": 3.892423919982975, "grad_norm": 14.21875, "learning_rate": 8.71798607163161e-06, "loss": 0.6668, "step": 36581 }, { "epoch": 3.8925303256011916, "grad_norm": 12.984375, "learning_rate": 8.717950540079591e-06, "loss": 0.6455, "step": 36582 }, { "epoch": 3.892636731219408, "grad_norm": 12.3984375, "learning_rate": 8.717915008527573e-06, "loss": 0.6597, "step": 36583 }, { "epoch": 3.892743136837625, "grad_norm": 12.3984375, "learning_rate": 8.717879476975555e-06, "loss": 0.6602, "step": 36584 }, { "epoch": 3.8928495424558416, "grad_norm": 12.8046875, "learning_rate": 8.717843945423536e-06, "loss": 0.6557, "step": 36585 }, { "epoch": 3.892955948074058, "grad_norm": 14.84375, "learning_rate": 8.71780841387152e-06, "loss": 0.745, "step": 36586 }, { "epoch": 3.893062353692275, "grad_norm": 11.453125, "learning_rate": 8.717772882319501e-06, "loss": 0.5326, "step": 36587 }, { "epoch": 3.8931687593104916, "grad_norm": 16.28125, "learning_rate": 8.717737350767483e-06, "loss": 0.6523, "step": 36588 }, { "epoch": 3.893275164928708, "grad_norm": 11.2734375, "learning_rate": 8.717701819215464e-06, "loss": 0.7485, "step": 36589 }, { "epoch": 3.893381570546925, "grad_norm": 10.921875, "learning_rate": 8.717666287663446e-06, "loss": 0.5201, "step": 36590 }, { "epoch": 3.8934879761651415, "grad_norm": 11.5390625, "learning_rate": 8.717630756111428e-06, "loss": 0.6061, "step": 36591 }, { "epoch": 3.893594381783358, "grad_norm": 17.0625, "learning_rate": 8.71759522455941e-06, "loss": 0.659, "step": 36592 }, { "epoch": 3.893700787401575, "grad_norm": 10.8828125, "learning_rate": 8.71755969300739e-06, "loss": 0.6461, "step": 36593 }, { "epoch": 3.8938071930197915, "grad_norm": 13.4765625, "learning_rate": 8.717524161455372e-06, "loss": 0.7022, "step": 36594 }, { "epoch": 3.893913598638008, "grad_norm": 11.6875, "learning_rate": 8.717488629903354e-06, "loss": 0.7094, "step": 36595 }, { "epoch": 3.894020004256225, "grad_norm": 13.2109375, "learning_rate": 8.717453098351337e-06, "loss": 0.6245, "step": 36596 }, { "epoch": 3.8941264098744415, "grad_norm": 10.4765625, "learning_rate": 8.717417566799319e-06, "loss": 0.6078, "step": 36597 }, { "epoch": 3.894232815492658, "grad_norm": 9.7734375, "learning_rate": 8.7173820352473e-06, "loss": 0.5801, "step": 36598 }, { "epoch": 3.8943392211108745, "grad_norm": 13.4140625, "learning_rate": 8.717346503695282e-06, "loss": 0.8861, "step": 36599 }, { "epoch": 3.8944456267290914, "grad_norm": 12.59375, "learning_rate": 8.717310972143264e-06, "loss": 0.4935, "step": 36600 }, { "epoch": 3.894552032347308, "grad_norm": 11.046875, "learning_rate": 8.717275440591245e-06, "loss": 0.5681, "step": 36601 }, { "epoch": 3.8946584379655245, "grad_norm": 12.1171875, "learning_rate": 8.717239909039229e-06, "loss": 0.5071, "step": 36602 }, { "epoch": 3.894764843583741, "grad_norm": 15.9921875, "learning_rate": 8.71720437748721e-06, "loss": 0.6843, "step": 36603 }, { "epoch": 3.894871249201958, "grad_norm": 13.8203125, "learning_rate": 8.717168845935192e-06, "loss": 0.6987, "step": 36604 }, { "epoch": 3.8949776548201744, "grad_norm": 13.1953125, "learning_rate": 8.717133314383173e-06, "loss": 0.5929, "step": 36605 }, { "epoch": 3.895084060438391, "grad_norm": 10.890625, "learning_rate": 8.717097782831155e-06, "loss": 0.5486, "step": 36606 }, { "epoch": 3.895190466056608, "grad_norm": 12.6328125, "learning_rate": 8.717062251279137e-06, "loss": 0.6428, "step": 36607 }, { "epoch": 3.8952968716748244, "grad_norm": 8.9921875, "learning_rate": 8.717026719727118e-06, "loss": 0.5006, "step": 36608 }, { "epoch": 3.895403277293041, "grad_norm": 14.2421875, "learning_rate": 8.7169911881751e-06, "loss": 0.6137, "step": 36609 }, { "epoch": 3.895509682911258, "grad_norm": 12.1953125, "learning_rate": 8.716955656623081e-06, "loss": 0.8929, "step": 36610 }, { "epoch": 3.8956160885294744, "grad_norm": 10.140625, "learning_rate": 8.716920125071063e-06, "loss": 0.6082, "step": 36611 }, { "epoch": 3.895722494147691, "grad_norm": 13.71875, "learning_rate": 8.716884593519046e-06, "loss": 0.7412, "step": 36612 }, { "epoch": 3.895828899765908, "grad_norm": 20.28125, "learning_rate": 8.716849061967028e-06, "loss": 0.6065, "step": 36613 }, { "epoch": 3.8959353053841244, "grad_norm": 12.7109375, "learning_rate": 8.71681353041501e-06, "loss": 0.5596, "step": 36614 }, { "epoch": 3.896041711002341, "grad_norm": 14.359375, "learning_rate": 8.716777998862991e-06, "loss": 0.6369, "step": 36615 }, { "epoch": 3.896148116620558, "grad_norm": 15.265625, "learning_rate": 8.716742467310973e-06, "loss": 0.6413, "step": 36616 }, { "epoch": 3.8962545222387743, "grad_norm": 12.71875, "learning_rate": 8.716706935758954e-06, "loss": 0.6244, "step": 36617 }, { "epoch": 3.896360927856991, "grad_norm": 13.2109375, "learning_rate": 8.716671404206936e-06, "loss": 0.6566, "step": 36618 }, { "epoch": 3.8964673334752074, "grad_norm": 15.0546875, "learning_rate": 8.716635872654918e-06, "loss": 0.7363, "step": 36619 }, { "epoch": 3.896573739093424, "grad_norm": 18.21875, "learning_rate": 8.7166003411029e-06, "loss": 0.8107, "step": 36620 }, { "epoch": 3.896680144711641, "grad_norm": 13.296875, "learning_rate": 8.71656480955088e-06, "loss": 0.7886, "step": 36621 }, { "epoch": 3.8967865503298573, "grad_norm": 13.984375, "learning_rate": 8.716529277998864e-06, "loss": 0.7005, "step": 36622 }, { "epoch": 3.896892955948074, "grad_norm": 18.71875, "learning_rate": 8.716493746446846e-06, "loss": 0.7949, "step": 36623 }, { "epoch": 3.896999361566291, "grad_norm": 9.984375, "learning_rate": 8.716458214894827e-06, "loss": 0.6664, "step": 36624 }, { "epoch": 3.8971057671845073, "grad_norm": 13.3671875, "learning_rate": 8.716422683342809e-06, "loss": 0.6866, "step": 36625 }, { "epoch": 3.897212172802724, "grad_norm": 15.2421875, "learning_rate": 8.71638715179079e-06, "loss": 0.6418, "step": 36626 }, { "epoch": 3.8973185784209408, "grad_norm": 13.6015625, "learning_rate": 8.716351620238774e-06, "loss": 0.6395, "step": 36627 }, { "epoch": 3.8974249840391573, "grad_norm": 18.6875, "learning_rate": 8.716316088686755e-06, "loss": 0.786, "step": 36628 }, { "epoch": 3.897531389657374, "grad_norm": 13.6640625, "learning_rate": 8.716280557134737e-06, "loss": 0.639, "step": 36629 }, { "epoch": 3.8976377952755907, "grad_norm": 13.4765625, "learning_rate": 8.716245025582719e-06, "loss": 0.7316, "step": 36630 }, { "epoch": 3.8977442008938072, "grad_norm": 15.203125, "learning_rate": 8.7162094940307e-06, "loss": 0.7219, "step": 36631 }, { "epoch": 3.8978506065120238, "grad_norm": 12.3046875, "learning_rate": 8.716173962478682e-06, "loss": 0.6311, "step": 36632 }, { "epoch": 3.8979570121302407, "grad_norm": 13.7421875, "learning_rate": 8.716138430926664e-06, "loss": 0.7067, "step": 36633 }, { "epoch": 3.898063417748457, "grad_norm": 12.0625, "learning_rate": 8.716102899374645e-06, "loss": 0.7195, "step": 36634 }, { "epoch": 3.8981698233666737, "grad_norm": 12.8515625, "learning_rate": 8.716067367822627e-06, "loss": 0.7294, "step": 36635 }, { "epoch": 3.8982762289848907, "grad_norm": 13.78125, "learning_rate": 8.716031836270608e-06, "loss": 0.795, "step": 36636 }, { "epoch": 3.898382634603107, "grad_norm": 15.484375, "learning_rate": 8.715996304718592e-06, "loss": 0.7747, "step": 36637 }, { "epoch": 3.8984890402213237, "grad_norm": 12.78125, "learning_rate": 8.715960773166573e-06, "loss": 0.6859, "step": 36638 }, { "epoch": 3.89859544583954, "grad_norm": 12.5703125, "learning_rate": 8.715925241614555e-06, "loss": 0.6186, "step": 36639 }, { "epoch": 3.8987018514577567, "grad_norm": 98.6875, "learning_rate": 8.715889710062536e-06, "loss": 0.8223, "step": 36640 }, { "epoch": 3.8988082570759737, "grad_norm": 15.1796875, "learning_rate": 8.715854178510518e-06, "loss": 0.6486, "step": 36641 }, { "epoch": 3.89891466269419, "grad_norm": 11.546875, "learning_rate": 8.7158186469585e-06, "loss": 0.6396, "step": 36642 }, { "epoch": 3.8990210683124067, "grad_norm": 15.1953125, "learning_rate": 8.715783115406481e-06, "loss": 0.6391, "step": 36643 }, { "epoch": 3.8991274739306236, "grad_norm": 10.078125, "learning_rate": 8.715747583854463e-06, "loss": 0.6061, "step": 36644 }, { "epoch": 3.89923387954884, "grad_norm": 10.734375, "learning_rate": 8.715712052302444e-06, "loss": 0.6915, "step": 36645 }, { "epoch": 3.8993402851670567, "grad_norm": 9.6875, "learning_rate": 8.715676520750426e-06, "loss": 0.5334, "step": 36646 }, { "epoch": 3.8994466907852736, "grad_norm": 17.265625, "learning_rate": 8.715640989198408e-06, "loss": 1.0047, "step": 36647 }, { "epoch": 3.89955309640349, "grad_norm": 10.53125, "learning_rate": 8.715605457646391e-06, "loss": 0.6945, "step": 36648 }, { "epoch": 3.8996595020217066, "grad_norm": 12.984375, "learning_rate": 8.715569926094373e-06, "loss": 0.794, "step": 36649 }, { "epoch": 3.8997659076399236, "grad_norm": 18.234375, "learning_rate": 8.715534394542354e-06, "loss": 0.747, "step": 36650 }, { "epoch": 3.89987231325814, "grad_norm": 11.4140625, "learning_rate": 8.715498862990336e-06, "loss": 0.6285, "step": 36651 }, { "epoch": 3.8999787188763566, "grad_norm": 12.3125, "learning_rate": 8.715463331438317e-06, "loss": 0.58, "step": 36652 }, { "epoch": 3.9000851244945736, "grad_norm": 12.2265625, "learning_rate": 8.7154277998863e-06, "loss": 0.6708, "step": 36653 }, { "epoch": 3.90019153011279, "grad_norm": 11.15625, "learning_rate": 8.715392268334282e-06, "loss": 0.6516, "step": 36654 }, { "epoch": 3.9002979357310066, "grad_norm": 13.1171875, "learning_rate": 8.715356736782264e-06, "loss": 0.7138, "step": 36655 }, { "epoch": 3.900404341349223, "grad_norm": 11.6875, "learning_rate": 8.715321205230246e-06, "loss": 0.5481, "step": 36656 }, { "epoch": 3.90051074696744, "grad_norm": 10.6640625, "learning_rate": 8.715285673678227e-06, "loss": 0.5928, "step": 36657 }, { "epoch": 3.9006171525856566, "grad_norm": 13.2421875, "learning_rate": 8.715250142126209e-06, "loss": 0.594, "step": 36658 }, { "epoch": 3.900723558203873, "grad_norm": 12.0859375, "learning_rate": 8.71521461057419e-06, "loss": 0.5739, "step": 36659 }, { "epoch": 3.9008299638220896, "grad_norm": 11.8046875, "learning_rate": 8.715179079022172e-06, "loss": 0.5977, "step": 36660 }, { "epoch": 3.9009363694403065, "grad_norm": 13.40625, "learning_rate": 8.715143547470154e-06, "loss": 0.7498, "step": 36661 }, { "epoch": 3.901042775058523, "grad_norm": 16.375, "learning_rate": 8.715108015918135e-06, "loss": 0.7367, "step": 36662 }, { "epoch": 3.9011491806767395, "grad_norm": 12.8671875, "learning_rate": 8.715072484366118e-06, "loss": 0.8311, "step": 36663 }, { "epoch": 3.9012555862949565, "grad_norm": 13.28125, "learning_rate": 8.7150369528141e-06, "loss": 0.7673, "step": 36664 }, { "epoch": 3.901361991913173, "grad_norm": 14.375, "learning_rate": 8.715001421262082e-06, "loss": 0.6749, "step": 36665 }, { "epoch": 3.9014683975313895, "grad_norm": 10.3828125, "learning_rate": 8.714965889710063e-06, "loss": 0.7389, "step": 36666 }, { "epoch": 3.9015748031496065, "grad_norm": 17.78125, "learning_rate": 8.714930358158045e-06, "loss": 0.8065, "step": 36667 }, { "epoch": 3.901681208767823, "grad_norm": 16.515625, "learning_rate": 8.714894826606028e-06, "loss": 0.7649, "step": 36668 }, { "epoch": 3.9017876143860395, "grad_norm": 12.71875, "learning_rate": 8.714859295054008e-06, "loss": 0.7617, "step": 36669 }, { "epoch": 3.9018940200042564, "grad_norm": 13.921875, "learning_rate": 8.71482376350199e-06, "loss": 0.7657, "step": 36670 }, { "epoch": 3.902000425622473, "grad_norm": 11.9375, "learning_rate": 8.714788231949971e-06, "loss": 0.6397, "step": 36671 }, { "epoch": 3.9021068312406895, "grad_norm": 12.0078125, "learning_rate": 8.714752700397953e-06, "loss": 0.6024, "step": 36672 }, { "epoch": 3.9022132368589064, "grad_norm": 11.4296875, "learning_rate": 8.714717168845936e-06, "loss": 0.5015, "step": 36673 }, { "epoch": 3.902319642477123, "grad_norm": 12.34375, "learning_rate": 8.714681637293918e-06, "loss": 0.5821, "step": 36674 }, { "epoch": 3.9024260480953394, "grad_norm": 14.296875, "learning_rate": 8.7146461057419e-06, "loss": 0.6555, "step": 36675 }, { "epoch": 3.902532453713556, "grad_norm": 12.6171875, "learning_rate": 8.714610574189881e-06, "loss": 0.6481, "step": 36676 }, { "epoch": 3.9026388593317725, "grad_norm": 13.8046875, "learning_rate": 8.714575042637863e-06, "loss": 0.7667, "step": 36677 }, { "epoch": 3.9027452649499894, "grad_norm": 15.953125, "learning_rate": 8.714539511085844e-06, "loss": 0.7475, "step": 36678 }, { "epoch": 3.902851670568206, "grad_norm": 13.2578125, "learning_rate": 8.714503979533828e-06, "loss": 0.6189, "step": 36679 }, { "epoch": 3.9029580761864224, "grad_norm": 14.40625, "learning_rate": 8.71446844798181e-06, "loss": 0.7468, "step": 36680 }, { "epoch": 3.9030644818046394, "grad_norm": 13.453125, "learning_rate": 8.71443291642979e-06, "loss": 0.6183, "step": 36681 }, { "epoch": 3.903170887422856, "grad_norm": 10.4296875, "learning_rate": 8.714397384877772e-06, "loss": 0.4626, "step": 36682 }, { "epoch": 3.9032772930410724, "grad_norm": 15.984375, "learning_rate": 8.714361853325754e-06, "loss": 0.6941, "step": 36683 }, { "epoch": 3.9033836986592894, "grad_norm": 12.359375, "learning_rate": 8.714326321773736e-06, "loss": 0.5606, "step": 36684 }, { "epoch": 3.903490104277506, "grad_norm": 15.921875, "learning_rate": 8.714290790221717e-06, "loss": 0.7207, "step": 36685 }, { "epoch": 3.9035965098957224, "grad_norm": 18.625, "learning_rate": 8.714255258669699e-06, "loss": 0.737, "step": 36686 }, { "epoch": 3.9037029155139393, "grad_norm": 14.6328125, "learning_rate": 8.71421972711768e-06, "loss": 0.7094, "step": 36687 }, { "epoch": 3.903809321132156, "grad_norm": 16.109375, "learning_rate": 8.714184195565662e-06, "loss": 0.7049, "step": 36688 }, { "epoch": 3.9039157267503724, "grad_norm": 14.4375, "learning_rate": 8.714148664013645e-06, "loss": 0.5368, "step": 36689 }, { "epoch": 3.9040221323685893, "grad_norm": 16.890625, "learning_rate": 8.714113132461627e-06, "loss": 0.7327, "step": 36690 }, { "epoch": 3.904128537986806, "grad_norm": 20.5625, "learning_rate": 8.714077600909609e-06, "loss": 0.739, "step": 36691 }, { "epoch": 3.9042349436050223, "grad_norm": 11.6015625, "learning_rate": 8.71404206935759e-06, "loss": 0.6715, "step": 36692 }, { "epoch": 3.904341349223239, "grad_norm": 11.3671875, "learning_rate": 8.714006537805572e-06, "loss": 0.7386, "step": 36693 }, { "epoch": 3.904447754841456, "grad_norm": 14.578125, "learning_rate": 8.713971006253555e-06, "loss": 0.6364, "step": 36694 }, { "epoch": 3.9045541604596723, "grad_norm": 14.328125, "learning_rate": 8.713935474701537e-06, "loss": 0.6556, "step": 36695 }, { "epoch": 3.904660566077889, "grad_norm": 11.5546875, "learning_rate": 8.713899943149518e-06, "loss": 0.7774, "step": 36696 }, { "epoch": 3.9047669716961053, "grad_norm": 11.4375, "learning_rate": 8.7138644115975e-06, "loss": 0.6679, "step": 36697 }, { "epoch": 3.9048733773143223, "grad_norm": 14.96875, "learning_rate": 8.713828880045481e-06, "loss": 0.8461, "step": 36698 }, { "epoch": 3.904979782932539, "grad_norm": 22.359375, "learning_rate": 8.713793348493463e-06, "loss": 0.7951, "step": 36699 }, { "epoch": 3.9050861885507553, "grad_norm": 11.8203125, "learning_rate": 8.713757816941445e-06, "loss": 0.6583, "step": 36700 }, { "epoch": 3.9051925941689722, "grad_norm": 13.1640625, "learning_rate": 8.713722285389426e-06, "loss": 0.6463, "step": 36701 }, { "epoch": 3.9052989997871888, "grad_norm": 16.609375, "learning_rate": 8.713686753837408e-06, "loss": 0.6815, "step": 36702 }, { "epoch": 3.9054054054054053, "grad_norm": 15.796875, "learning_rate": 8.71365122228539e-06, "loss": 0.7002, "step": 36703 }, { "epoch": 3.905511811023622, "grad_norm": 14.1875, "learning_rate": 8.713615690733373e-06, "loss": 0.7094, "step": 36704 }, { "epoch": 3.9056182166418387, "grad_norm": 27.234375, "learning_rate": 8.713580159181354e-06, "loss": 0.6042, "step": 36705 }, { "epoch": 3.9057246222600552, "grad_norm": 10.9375, "learning_rate": 8.713544627629336e-06, "loss": 0.6523, "step": 36706 }, { "epoch": 3.905831027878272, "grad_norm": 14.40625, "learning_rate": 8.713509096077318e-06, "loss": 0.6594, "step": 36707 }, { "epoch": 3.9059374334964887, "grad_norm": 14.53125, "learning_rate": 8.7134735645253e-06, "loss": 0.6413, "step": 36708 }, { "epoch": 3.906043839114705, "grad_norm": 16.46875, "learning_rate": 8.713438032973281e-06, "loss": 0.8605, "step": 36709 }, { "epoch": 3.906150244732922, "grad_norm": 18.0625, "learning_rate": 8.713402501421262e-06, "loss": 0.8877, "step": 36710 }, { "epoch": 3.9062566503511387, "grad_norm": 15.3515625, "learning_rate": 8.713366969869244e-06, "loss": 0.6861, "step": 36711 }, { "epoch": 3.906363055969355, "grad_norm": 12.015625, "learning_rate": 8.713331438317226e-06, "loss": 0.7436, "step": 36712 }, { "epoch": 3.9064694615875717, "grad_norm": 11.234375, "learning_rate": 8.713295906765207e-06, "loss": 0.6295, "step": 36713 }, { "epoch": 3.906575867205788, "grad_norm": 12.9375, "learning_rate": 8.71326037521319e-06, "loss": 0.6429, "step": 36714 }, { "epoch": 3.906682272824005, "grad_norm": 12.546875, "learning_rate": 8.713224843661172e-06, "loss": 0.6669, "step": 36715 }, { "epoch": 3.9067886784422217, "grad_norm": 15.4921875, "learning_rate": 8.713189312109154e-06, "loss": 0.7569, "step": 36716 }, { "epoch": 3.906895084060438, "grad_norm": 17.859375, "learning_rate": 8.713153780557135e-06, "loss": 0.6778, "step": 36717 }, { "epoch": 3.907001489678655, "grad_norm": 13.375, "learning_rate": 8.713118249005117e-06, "loss": 0.7021, "step": 36718 }, { "epoch": 3.9071078952968716, "grad_norm": 10.1015625, "learning_rate": 8.713082717453099e-06, "loss": 0.5321, "step": 36719 }, { "epoch": 3.907214300915088, "grad_norm": 15.25, "learning_rate": 8.713047185901082e-06, "loss": 0.765, "step": 36720 }, { "epoch": 3.907320706533305, "grad_norm": 12.8984375, "learning_rate": 8.713011654349064e-06, "loss": 0.7026, "step": 36721 }, { "epoch": 3.9074271121515216, "grad_norm": 12.25, "learning_rate": 8.712976122797045e-06, "loss": 0.5657, "step": 36722 }, { "epoch": 3.907533517769738, "grad_norm": 13.4140625, "learning_rate": 8.712940591245027e-06, "loss": 0.7727, "step": 36723 }, { "epoch": 3.907639923387955, "grad_norm": 14.7578125, "learning_rate": 8.712905059693008e-06, "loss": 0.6197, "step": 36724 }, { "epoch": 3.9077463290061716, "grad_norm": 13.59375, "learning_rate": 8.71286952814099e-06, "loss": 0.6289, "step": 36725 }, { "epoch": 3.907852734624388, "grad_norm": 14.46875, "learning_rate": 8.712833996588972e-06, "loss": 0.6955, "step": 36726 }, { "epoch": 3.907959140242605, "grad_norm": 19.671875, "learning_rate": 8.712798465036953e-06, "loss": 0.7506, "step": 36727 }, { "epoch": 3.9080655458608216, "grad_norm": 14.3046875, "learning_rate": 8.712762933484935e-06, "loss": 0.6873, "step": 36728 }, { "epoch": 3.908171951479038, "grad_norm": 10.703125, "learning_rate": 8.712727401932916e-06, "loss": 0.5767, "step": 36729 }, { "epoch": 3.9082783570972546, "grad_norm": 11.6953125, "learning_rate": 8.7126918703809e-06, "loss": 0.6863, "step": 36730 }, { "epoch": 3.9083847627154715, "grad_norm": 12.5625, "learning_rate": 8.712656338828881e-06, "loss": 0.5766, "step": 36731 }, { "epoch": 3.908491168333688, "grad_norm": 12.859375, "learning_rate": 8.712620807276863e-06, "loss": 0.5997, "step": 36732 }, { "epoch": 3.9085975739519045, "grad_norm": 14.7109375, "learning_rate": 8.712585275724844e-06, "loss": 0.7992, "step": 36733 }, { "epoch": 3.908703979570121, "grad_norm": 17.28125, "learning_rate": 8.712549744172826e-06, "loss": 0.6903, "step": 36734 }, { "epoch": 3.908810385188338, "grad_norm": 12.21875, "learning_rate": 8.712514212620808e-06, "loss": 0.6822, "step": 36735 }, { "epoch": 3.9089167908065545, "grad_norm": 11.4375, "learning_rate": 8.71247868106879e-06, "loss": 0.7005, "step": 36736 }, { "epoch": 3.909023196424771, "grad_norm": 12.6328125, "learning_rate": 8.712443149516771e-06, "loss": 0.6351, "step": 36737 }, { "epoch": 3.909129602042988, "grad_norm": 13.0546875, "learning_rate": 8.712407617964753e-06, "loss": 0.6583, "step": 36738 }, { "epoch": 3.9092360076612045, "grad_norm": 18.046875, "learning_rate": 8.712372086412734e-06, "loss": 0.9617, "step": 36739 }, { "epoch": 3.909342413279421, "grad_norm": 12.921875, "learning_rate": 8.712336554860717e-06, "loss": 0.6127, "step": 36740 }, { "epoch": 3.909448818897638, "grad_norm": 10.15625, "learning_rate": 8.712301023308699e-06, "loss": 0.6097, "step": 36741 }, { "epoch": 3.9095552245158545, "grad_norm": 14.2578125, "learning_rate": 8.71226549175668e-06, "loss": 0.684, "step": 36742 }, { "epoch": 3.909661630134071, "grad_norm": 12.625, "learning_rate": 8.712229960204662e-06, "loss": 0.7006, "step": 36743 }, { "epoch": 3.909768035752288, "grad_norm": 13.7265625, "learning_rate": 8.712194428652644e-06, "loss": 0.661, "step": 36744 }, { "epoch": 3.9098744413705044, "grad_norm": 11.8828125, "learning_rate": 8.712158897100627e-06, "loss": 0.729, "step": 36745 }, { "epoch": 3.909980846988721, "grad_norm": 11.28125, "learning_rate": 8.712123365548609e-06, "loss": 0.5782, "step": 36746 }, { "epoch": 3.910087252606938, "grad_norm": 15.9375, "learning_rate": 8.71208783399659e-06, "loss": 0.6813, "step": 36747 }, { "epoch": 3.9101936582251544, "grad_norm": 11.3359375, "learning_rate": 8.712052302444572e-06, "loss": 0.5803, "step": 36748 }, { "epoch": 3.910300063843371, "grad_norm": 12.3515625, "learning_rate": 8.712016770892554e-06, "loss": 0.5584, "step": 36749 }, { "epoch": 3.9104064694615874, "grad_norm": 16.359375, "learning_rate": 8.711981239340535e-06, "loss": 0.8306, "step": 36750 }, { "epoch": 3.910512875079804, "grad_norm": 14.515625, "learning_rate": 8.711945707788517e-06, "loss": 0.6968, "step": 36751 }, { "epoch": 3.910619280698021, "grad_norm": 18.765625, "learning_rate": 8.711910176236498e-06, "loss": 0.7351, "step": 36752 }, { "epoch": 3.9107256863162374, "grad_norm": 15.984375, "learning_rate": 8.71187464468448e-06, "loss": 0.6031, "step": 36753 }, { "epoch": 3.910832091934454, "grad_norm": 17.171875, "learning_rate": 8.711839113132462e-06, "loss": 0.753, "step": 36754 }, { "epoch": 3.910938497552671, "grad_norm": 14.125, "learning_rate": 8.711803581580443e-06, "loss": 0.6715, "step": 36755 }, { "epoch": 3.9110449031708874, "grad_norm": 16.96875, "learning_rate": 8.711768050028427e-06, "loss": 0.5769, "step": 36756 }, { "epoch": 3.911151308789104, "grad_norm": 12.0625, "learning_rate": 8.711732518476408e-06, "loss": 0.6049, "step": 36757 }, { "epoch": 3.911257714407321, "grad_norm": 12.3515625, "learning_rate": 8.71169698692439e-06, "loss": 0.7029, "step": 36758 }, { "epoch": 3.9113641200255373, "grad_norm": 13.9765625, "learning_rate": 8.711661455372371e-06, "loss": 0.6427, "step": 36759 }, { "epoch": 3.911470525643754, "grad_norm": 12.9375, "learning_rate": 8.711625923820353e-06, "loss": 0.6269, "step": 36760 }, { "epoch": 3.911576931261971, "grad_norm": 11.3359375, "learning_rate": 8.711590392268335e-06, "loss": 0.5924, "step": 36761 }, { "epoch": 3.9116833368801873, "grad_norm": 10.7265625, "learning_rate": 8.711554860716316e-06, "loss": 0.5583, "step": 36762 }, { "epoch": 3.911789742498404, "grad_norm": 13.734375, "learning_rate": 8.711519329164298e-06, "loss": 0.6952, "step": 36763 }, { "epoch": 3.911896148116621, "grad_norm": 13.734375, "learning_rate": 8.71148379761228e-06, "loss": 0.7857, "step": 36764 }, { "epoch": 3.9120025537348373, "grad_norm": 17.84375, "learning_rate": 8.711448266060261e-06, "loss": 0.8827, "step": 36765 }, { "epoch": 3.912108959353054, "grad_norm": 10.5859375, "learning_rate": 8.711412734508244e-06, "loss": 0.6314, "step": 36766 }, { "epoch": 3.9122153649712708, "grad_norm": 11.953125, "learning_rate": 8.711377202956226e-06, "loss": 0.7703, "step": 36767 }, { "epoch": 3.9123217705894873, "grad_norm": 15.15625, "learning_rate": 8.711341671404208e-06, "loss": 0.6888, "step": 36768 }, { "epoch": 3.912428176207704, "grad_norm": 12.84375, "learning_rate": 8.711306139852189e-06, "loss": 0.6365, "step": 36769 }, { "epoch": 3.9125345818259203, "grad_norm": 17.390625, "learning_rate": 8.71127060830017e-06, "loss": 0.78, "step": 36770 }, { "epoch": 3.912640987444137, "grad_norm": 17.0625, "learning_rate": 8.711235076748154e-06, "loss": 0.7057, "step": 36771 }, { "epoch": 3.9127473930623538, "grad_norm": 12.6640625, "learning_rate": 8.711199545196136e-06, "loss": 0.6173, "step": 36772 }, { "epoch": 3.9128537986805703, "grad_norm": 11.203125, "learning_rate": 8.711164013644117e-06, "loss": 0.6793, "step": 36773 }, { "epoch": 3.9129602042987868, "grad_norm": 11.953125, "learning_rate": 8.711128482092099e-06, "loss": 0.6848, "step": 36774 }, { "epoch": 3.9130666099170037, "grad_norm": 13.3125, "learning_rate": 8.71109295054008e-06, "loss": 0.495, "step": 36775 }, { "epoch": 3.9131730155352202, "grad_norm": 13.734375, "learning_rate": 8.711057418988062e-06, "loss": 0.711, "step": 36776 }, { "epoch": 3.9132794211534367, "grad_norm": 16.8125, "learning_rate": 8.711021887436044e-06, "loss": 0.6692, "step": 36777 }, { "epoch": 3.9133858267716537, "grad_norm": 14.7890625, "learning_rate": 8.710986355884025e-06, "loss": 0.7427, "step": 36778 }, { "epoch": 3.91349223238987, "grad_norm": 15.640625, "learning_rate": 8.710950824332007e-06, "loss": 0.6994, "step": 36779 }, { "epoch": 3.9135986380080867, "grad_norm": 12.484375, "learning_rate": 8.710915292779988e-06, "loss": 0.5935, "step": 36780 }, { "epoch": 3.9137050436263037, "grad_norm": 12.9921875, "learning_rate": 8.710879761227972e-06, "loss": 0.6691, "step": 36781 }, { "epoch": 3.91381144924452, "grad_norm": 16.765625, "learning_rate": 8.710844229675953e-06, "loss": 0.8759, "step": 36782 }, { "epoch": 3.9139178548627367, "grad_norm": 13.984375, "learning_rate": 8.710808698123935e-06, "loss": 0.7188, "step": 36783 }, { "epoch": 3.9140242604809536, "grad_norm": 13.7421875, "learning_rate": 8.710773166571917e-06, "loss": 0.6177, "step": 36784 }, { "epoch": 3.91413066609917, "grad_norm": 9.6796875, "learning_rate": 8.710737635019898e-06, "loss": 0.6064, "step": 36785 }, { "epoch": 3.9142370717173867, "grad_norm": 18.1875, "learning_rate": 8.710702103467882e-06, "loss": 0.8308, "step": 36786 }, { "epoch": 3.914343477335603, "grad_norm": 11.6953125, "learning_rate": 8.710666571915863e-06, "loss": 0.7036, "step": 36787 }, { "epoch": 3.91444988295382, "grad_norm": 13.4609375, "learning_rate": 8.710631040363845e-06, "loss": 0.7498, "step": 36788 }, { "epoch": 3.9145562885720366, "grad_norm": 14.3359375, "learning_rate": 8.710595508811825e-06, "loss": 0.7005, "step": 36789 }, { "epoch": 3.914662694190253, "grad_norm": 10.546875, "learning_rate": 8.710559977259806e-06, "loss": 0.6018, "step": 36790 }, { "epoch": 3.9147690998084697, "grad_norm": 14.6953125, "learning_rate": 8.71052444570779e-06, "loss": 0.6855, "step": 36791 }, { "epoch": 3.9148755054266866, "grad_norm": 12.9296875, "learning_rate": 8.710488914155771e-06, "loss": 0.562, "step": 36792 }, { "epoch": 3.914981911044903, "grad_norm": 14.6015625, "learning_rate": 8.710453382603753e-06, "loss": 0.6394, "step": 36793 }, { "epoch": 3.9150883166631196, "grad_norm": 11.6484375, "learning_rate": 8.710417851051734e-06, "loss": 0.6867, "step": 36794 }, { "epoch": 3.9151947222813366, "grad_norm": 12.8828125, "learning_rate": 8.710382319499716e-06, "loss": 0.6704, "step": 36795 }, { "epoch": 3.915301127899553, "grad_norm": 13.3125, "learning_rate": 8.710346787947698e-06, "loss": 0.6305, "step": 36796 }, { "epoch": 3.9154075335177696, "grad_norm": 11.390625, "learning_rate": 8.710311256395681e-06, "loss": 0.6038, "step": 36797 }, { "epoch": 3.9155139391359866, "grad_norm": 13.4921875, "learning_rate": 8.710275724843662e-06, "loss": 0.6761, "step": 36798 }, { "epoch": 3.915620344754203, "grad_norm": 12.28125, "learning_rate": 8.710240193291644e-06, "loss": 0.6166, "step": 36799 }, { "epoch": 3.9157267503724196, "grad_norm": 10.5, "learning_rate": 8.710204661739626e-06, "loss": 0.6047, "step": 36800 }, { "epoch": 3.9158331559906365, "grad_norm": 13.25, "learning_rate": 8.710169130187607e-06, "loss": 0.6512, "step": 36801 }, { "epoch": 3.915939561608853, "grad_norm": 19.40625, "learning_rate": 8.710133598635589e-06, "loss": 0.8196, "step": 36802 }, { "epoch": 3.9160459672270695, "grad_norm": 12.703125, "learning_rate": 8.71009806708357e-06, "loss": 0.7191, "step": 36803 }, { "epoch": 3.9161523728452865, "grad_norm": 13.4140625, "learning_rate": 8.710062535531552e-06, "loss": 0.6332, "step": 36804 }, { "epoch": 3.916258778463503, "grad_norm": 12.6953125, "learning_rate": 8.710027003979534e-06, "loss": 0.6641, "step": 36805 }, { "epoch": 3.9163651840817195, "grad_norm": 10.515625, "learning_rate": 8.709991472427515e-06, "loss": 0.6293, "step": 36806 }, { "epoch": 3.916471589699936, "grad_norm": 12.3984375, "learning_rate": 8.709955940875499e-06, "loss": 0.5941, "step": 36807 }, { "epoch": 3.9165779953181525, "grad_norm": 16.421875, "learning_rate": 8.70992040932348e-06, "loss": 0.8354, "step": 36808 }, { "epoch": 3.9166844009363695, "grad_norm": 12.6640625, "learning_rate": 8.709884877771462e-06, "loss": 0.6787, "step": 36809 }, { "epoch": 3.916790806554586, "grad_norm": 11.6796875, "learning_rate": 8.709849346219443e-06, "loss": 0.7427, "step": 36810 }, { "epoch": 3.9168972121728025, "grad_norm": 15.8984375, "learning_rate": 8.709813814667425e-06, "loss": 0.7282, "step": 36811 }, { "epoch": 3.9170036177910195, "grad_norm": 14.1953125, "learning_rate": 8.709778283115408e-06, "loss": 0.656, "step": 36812 }, { "epoch": 3.917110023409236, "grad_norm": 16.703125, "learning_rate": 8.70974275156339e-06, "loss": 0.8361, "step": 36813 }, { "epoch": 3.9172164290274525, "grad_norm": 15.546875, "learning_rate": 8.709707220011372e-06, "loss": 0.8999, "step": 36814 }, { "epoch": 3.9173228346456694, "grad_norm": 19.015625, "learning_rate": 8.709671688459353e-06, "loss": 0.8852, "step": 36815 }, { "epoch": 3.917429240263886, "grad_norm": 11.625, "learning_rate": 8.709636156907335e-06, "loss": 0.6887, "step": 36816 }, { "epoch": 3.9175356458821025, "grad_norm": 16.703125, "learning_rate": 8.709600625355316e-06, "loss": 0.7575, "step": 36817 }, { "epoch": 3.9176420515003194, "grad_norm": 13.65625, "learning_rate": 8.709565093803298e-06, "loss": 0.5941, "step": 36818 }, { "epoch": 3.917748457118536, "grad_norm": 11.2265625, "learning_rate": 8.70952956225128e-06, "loss": 0.5972, "step": 36819 }, { "epoch": 3.9178548627367524, "grad_norm": 11.15625, "learning_rate": 8.709494030699261e-06, "loss": 0.5367, "step": 36820 }, { "epoch": 3.9179612683549694, "grad_norm": 11.921875, "learning_rate": 8.709458499147243e-06, "loss": 0.6237, "step": 36821 }, { "epoch": 3.918067673973186, "grad_norm": 17.578125, "learning_rate": 8.709422967595226e-06, "loss": 0.6475, "step": 36822 }, { "epoch": 3.9181740795914024, "grad_norm": 13.8828125, "learning_rate": 8.709387436043208e-06, "loss": 0.6352, "step": 36823 }, { "epoch": 3.918280485209619, "grad_norm": 13.4296875, "learning_rate": 8.70935190449119e-06, "loss": 0.6434, "step": 36824 }, { "epoch": 3.918386890827836, "grad_norm": 15.75, "learning_rate": 8.709316372939171e-06, "loss": 0.7142, "step": 36825 }, { "epoch": 3.9184932964460524, "grad_norm": 11.8984375, "learning_rate": 8.709280841387153e-06, "loss": 0.6907, "step": 36826 }, { "epoch": 3.918599702064269, "grad_norm": 15.1328125, "learning_rate": 8.709245309835134e-06, "loss": 0.7649, "step": 36827 }, { "epoch": 3.9187061076824854, "grad_norm": 15.09375, "learning_rate": 8.709209778283116e-06, "loss": 0.7027, "step": 36828 }, { "epoch": 3.9188125133007023, "grad_norm": 15.5546875, "learning_rate": 8.709174246731097e-06, "loss": 0.7511, "step": 36829 }, { "epoch": 3.918918918918919, "grad_norm": 16.453125, "learning_rate": 8.709138715179079e-06, "loss": 0.7206, "step": 36830 }, { "epoch": 3.9190253245371354, "grad_norm": 12.5703125, "learning_rate": 8.70910318362706e-06, "loss": 0.5514, "step": 36831 }, { "epoch": 3.9191317301553523, "grad_norm": 16.609375, "learning_rate": 8.709067652075044e-06, "loss": 0.7818, "step": 36832 }, { "epoch": 3.919238135773569, "grad_norm": 15.8984375, "learning_rate": 8.709032120523025e-06, "loss": 0.7638, "step": 36833 }, { "epoch": 3.9193445413917853, "grad_norm": 15.1484375, "learning_rate": 8.708996588971007e-06, "loss": 0.6466, "step": 36834 }, { "epoch": 3.9194509470100023, "grad_norm": 14.0625, "learning_rate": 8.708961057418989e-06, "loss": 0.7686, "step": 36835 }, { "epoch": 3.919557352628219, "grad_norm": 13.8828125, "learning_rate": 8.70892552586697e-06, "loss": 0.7349, "step": 36836 }, { "epoch": 3.9196637582464353, "grad_norm": 13.859375, "learning_rate": 8.708889994314952e-06, "loss": 0.7088, "step": 36837 }, { "epoch": 3.9197701638646523, "grad_norm": 18.921875, "learning_rate": 8.708854462762935e-06, "loss": 0.7857, "step": 36838 }, { "epoch": 3.9198765694828688, "grad_norm": 12.890625, "learning_rate": 8.708818931210917e-06, "loss": 0.5602, "step": 36839 }, { "epoch": 3.9199829751010853, "grad_norm": 10.75, "learning_rate": 8.708783399658898e-06, "loss": 0.6267, "step": 36840 }, { "epoch": 3.9200893807193022, "grad_norm": 13.234375, "learning_rate": 8.70874786810688e-06, "loss": 0.7064, "step": 36841 }, { "epoch": 3.9201957863375188, "grad_norm": 11.546875, "learning_rate": 8.708712336554862e-06, "loss": 0.5982, "step": 36842 }, { "epoch": 3.9203021919557353, "grad_norm": 12.5234375, "learning_rate": 8.708676805002843e-06, "loss": 0.6161, "step": 36843 }, { "epoch": 3.9204085975739518, "grad_norm": 15.3046875, "learning_rate": 8.708641273450825e-06, "loss": 0.7464, "step": 36844 }, { "epoch": 3.9205150031921683, "grad_norm": 17.03125, "learning_rate": 8.708605741898806e-06, "loss": 0.6887, "step": 36845 }, { "epoch": 3.9206214088103852, "grad_norm": 15.1328125, "learning_rate": 8.708570210346788e-06, "loss": 0.704, "step": 36846 }, { "epoch": 3.9207278144286017, "grad_norm": 13.6484375, "learning_rate": 8.70853467879477e-06, "loss": 0.6193, "step": 36847 }, { "epoch": 3.9208342200468183, "grad_norm": 11.1328125, "learning_rate": 8.708499147242753e-06, "loss": 0.5274, "step": 36848 }, { "epoch": 3.920940625665035, "grad_norm": 18.109375, "learning_rate": 8.708463615690735e-06, "loss": 0.8265, "step": 36849 }, { "epoch": 3.9210470312832517, "grad_norm": 10.6171875, "learning_rate": 8.708428084138716e-06, "loss": 0.6255, "step": 36850 }, { "epoch": 3.9211534369014682, "grad_norm": 14.71875, "learning_rate": 8.708392552586698e-06, "loss": 0.6445, "step": 36851 }, { "epoch": 3.921259842519685, "grad_norm": 14.1171875, "learning_rate": 8.70835702103468e-06, "loss": 0.6316, "step": 36852 }, { "epoch": 3.9213662481379017, "grad_norm": 14.1015625, "learning_rate": 8.708321489482661e-06, "loss": 0.8459, "step": 36853 }, { "epoch": 3.921472653756118, "grad_norm": 12.3671875, "learning_rate": 8.708285957930643e-06, "loss": 0.7301, "step": 36854 }, { "epoch": 3.921579059374335, "grad_norm": 13.265625, "learning_rate": 8.708250426378624e-06, "loss": 0.6471, "step": 36855 }, { "epoch": 3.9216854649925517, "grad_norm": 12.3828125, "learning_rate": 8.708214894826606e-06, "loss": 0.6865, "step": 36856 }, { "epoch": 3.921791870610768, "grad_norm": 13.9140625, "learning_rate": 8.708179363274587e-06, "loss": 0.6045, "step": 36857 }, { "epoch": 3.921898276228985, "grad_norm": 14.2578125, "learning_rate": 8.70814383172257e-06, "loss": 0.6718, "step": 36858 }, { "epoch": 3.9220046818472016, "grad_norm": 16.296875, "learning_rate": 8.708108300170552e-06, "loss": 0.7846, "step": 36859 }, { "epoch": 3.922111087465418, "grad_norm": 10.453125, "learning_rate": 8.708072768618534e-06, "loss": 0.6058, "step": 36860 }, { "epoch": 3.9222174930836347, "grad_norm": 12.6953125, "learning_rate": 8.708037237066516e-06, "loss": 0.6307, "step": 36861 }, { "epoch": 3.9223238987018516, "grad_norm": 13.5, "learning_rate": 8.708001705514497e-06, "loss": 0.775, "step": 36862 }, { "epoch": 3.922430304320068, "grad_norm": 11.828125, "learning_rate": 8.70796617396248e-06, "loss": 0.6311, "step": 36863 }, { "epoch": 3.9225367099382846, "grad_norm": 13.3125, "learning_rate": 8.707930642410462e-06, "loss": 0.69, "step": 36864 }, { "epoch": 3.922643115556501, "grad_norm": 11.171875, "learning_rate": 8.707895110858444e-06, "loss": 0.5792, "step": 36865 }, { "epoch": 3.922749521174718, "grad_norm": 16.828125, "learning_rate": 8.707859579306425e-06, "loss": 0.8184, "step": 36866 }, { "epoch": 3.9228559267929346, "grad_norm": 14.46875, "learning_rate": 8.707824047754407e-06, "loss": 0.702, "step": 36867 }, { "epoch": 3.922962332411151, "grad_norm": 14.203125, "learning_rate": 8.707788516202389e-06, "loss": 0.7087, "step": 36868 }, { "epoch": 3.923068738029368, "grad_norm": 12.421875, "learning_rate": 8.70775298465037e-06, "loss": 0.805, "step": 36869 }, { "epoch": 3.9231751436475846, "grad_norm": 14.6484375, "learning_rate": 8.707717453098352e-06, "loss": 0.6756, "step": 36870 }, { "epoch": 3.923281549265801, "grad_norm": 14.0, "learning_rate": 8.707681921546333e-06, "loss": 0.5687, "step": 36871 }, { "epoch": 3.923387954884018, "grad_norm": 12.203125, "learning_rate": 8.707646389994315e-06, "loss": 0.6858, "step": 36872 }, { "epoch": 3.9234943605022345, "grad_norm": 14.609375, "learning_rate": 8.707610858442297e-06, "loss": 0.5879, "step": 36873 }, { "epoch": 3.923600766120451, "grad_norm": 11.3828125, "learning_rate": 8.70757532689028e-06, "loss": 0.6456, "step": 36874 }, { "epoch": 3.923707171738668, "grad_norm": 19.40625, "learning_rate": 8.707539795338261e-06, "loss": 0.8136, "step": 36875 }, { "epoch": 3.9238135773568845, "grad_norm": 12.0625, "learning_rate": 8.707504263786243e-06, "loss": 0.6366, "step": 36876 }, { "epoch": 3.923919982975101, "grad_norm": 10.765625, "learning_rate": 8.707468732234225e-06, "loss": 0.581, "step": 36877 }, { "epoch": 3.924026388593318, "grad_norm": 13.375, "learning_rate": 8.707433200682206e-06, "loss": 0.5739, "step": 36878 }, { "epoch": 3.9241327942115345, "grad_norm": 11.875, "learning_rate": 8.707397669130188e-06, "loss": 0.6347, "step": 36879 }, { "epoch": 3.924239199829751, "grad_norm": 12.0546875, "learning_rate": 8.70736213757817e-06, "loss": 0.6305, "step": 36880 }, { "epoch": 3.9243456054479675, "grad_norm": 17.28125, "learning_rate": 8.707326606026151e-06, "loss": 0.7079, "step": 36881 }, { "epoch": 3.924452011066184, "grad_norm": 11.71875, "learning_rate": 8.707291074474133e-06, "loss": 0.6698, "step": 36882 }, { "epoch": 3.924558416684401, "grad_norm": 14.1875, "learning_rate": 8.707255542922114e-06, "loss": 0.6942, "step": 36883 }, { "epoch": 3.9246648223026175, "grad_norm": 14.3359375, "learning_rate": 8.707220011370098e-06, "loss": 0.7103, "step": 36884 }, { "epoch": 3.924771227920834, "grad_norm": 15.5, "learning_rate": 8.70718447981808e-06, "loss": 0.6879, "step": 36885 }, { "epoch": 3.924877633539051, "grad_norm": 12.359375, "learning_rate": 8.70714894826606e-06, "loss": 0.6524, "step": 36886 }, { "epoch": 3.9249840391572675, "grad_norm": 12.3125, "learning_rate": 8.707113416714042e-06, "loss": 0.607, "step": 36887 }, { "epoch": 3.925090444775484, "grad_norm": 12.84375, "learning_rate": 8.707077885162024e-06, "loss": 0.6849, "step": 36888 }, { "epoch": 3.925196850393701, "grad_norm": 13.7109375, "learning_rate": 8.707042353610007e-06, "loss": 0.6226, "step": 36889 }, { "epoch": 3.9253032560119174, "grad_norm": 12.046875, "learning_rate": 8.707006822057989e-06, "loss": 0.7199, "step": 36890 }, { "epoch": 3.925409661630134, "grad_norm": 12.625, "learning_rate": 8.70697129050597e-06, "loss": 0.6975, "step": 36891 }, { "epoch": 3.925516067248351, "grad_norm": 17.515625, "learning_rate": 8.706935758953952e-06, "loss": 0.728, "step": 36892 }, { "epoch": 3.9256224728665674, "grad_norm": 19.4375, "learning_rate": 8.706900227401934e-06, "loss": 0.8185, "step": 36893 }, { "epoch": 3.925728878484784, "grad_norm": 17.046875, "learning_rate": 8.706864695849915e-06, "loss": 0.8098, "step": 36894 }, { "epoch": 3.925835284103001, "grad_norm": 13.6328125, "learning_rate": 8.706829164297897e-06, "loss": 0.7301, "step": 36895 }, { "epoch": 3.9259416897212174, "grad_norm": 13.5390625, "learning_rate": 8.706793632745879e-06, "loss": 0.675, "step": 36896 }, { "epoch": 3.926048095339434, "grad_norm": 13.015625, "learning_rate": 8.70675810119386e-06, "loss": 0.743, "step": 36897 }, { "epoch": 3.926154500957651, "grad_norm": 11.3984375, "learning_rate": 8.706722569641842e-06, "loss": 0.7136, "step": 36898 }, { "epoch": 3.9262609065758673, "grad_norm": 13.953125, "learning_rate": 8.706687038089825e-06, "loss": 0.7013, "step": 36899 }, { "epoch": 3.926367312194084, "grad_norm": 12.6640625, "learning_rate": 8.706651506537807e-06, "loss": 0.6632, "step": 36900 }, { "epoch": 3.9264737178123004, "grad_norm": 20.578125, "learning_rate": 8.706615974985788e-06, "loss": 0.8406, "step": 36901 }, { "epoch": 3.926580123430517, "grad_norm": 10.890625, "learning_rate": 8.70658044343377e-06, "loss": 0.5591, "step": 36902 }, { "epoch": 3.926686529048734, "grad_norm": 10.5078125, "learning_rate": 8.706544911881752e-06, "loss": 0.6245, "step": 36903 }, { "epoch": 3.9267929346669503, "grad_norm": 12.171875, "learning_rate": 8.706509380329735e-06, "loss": 0.61, "step": 36904 }, { "epoch": 3.926899340285167, "grad_norm": 14.3515625, "learning_rate": 8.706473848777716e-06, "loss": 0.6168, "step": 36905 }, { "epoch": 3.927005745903384, "grad_norm": 13.5390625, "learning_rate": 8.706438317225698e-06, "loss": 0.6235, "step": 36906 }, { "epoch": 3.9271121515216003, "grad_norm": 14.15625, "learning_rate": 8.70640278567368e-06, "loss": 0.709, "step": 36907 }, { "epoch": 3.927218557139817, "grad_norm": 14.8125, "learning_rate": 8.706367254121661e-06, "loss": 0.6646, "step": 36908 }, { "epoch": 3.9273249627580338, "grad_norm": 15.265625, "learning_rate": 8.706331722569643e-06, "loss": 0.7326, "step": 36909 }, { "epoch": 3.9274313683762503, "grad_norm": 12.328125, "learning_rate": 8.706296191017624e-06, "loss": 0.569, "step": 36910 }, { "epoch": 3.927537773994467, "grad_norm": 11.765625, "learning_rate": 8.706260659465606e-06, "loss": 0.5209, "step": 36911 }, { "epoch": 3.9276441796126838, "grad_norm": 11.8046875, "learning_rate": 8.706225127913588e-06, "loss": 0.6053, "step": 36912 }, { "epoch": 3.9277505852309003, "grad_norm": 14.21875, "learning_rate": 8.70618959636157e-06, "loss": 0.6256, "step": 36913 }, { "epoch": 3.9278569908491168, "grad_norm": 14.359375, "learning_rate": 8.706154064809551e-06, "loss": 0.645, "step": 36914 }, { "epoch": 3.9279633964673337, "grad_norm": 13.7421875, "learning_rate": 8.706118533257534e-06, "loss": 0.749, "step": 36915 }, { "epoch": 3.9280698020855502, "grad_norm": 13.765625, "learning_rate": 8.706083001705516e-06, "loss": 0.6981, "step": 36916 }, { "epoch": 3.9281762077037667, "grad_norm": 10.7578125, "learning_rate": 8.706047470153497e-06, "loss": 0.6882, "step": 36917 }, { "epoch": 3.9282826133219833, "grad_norm": 13.84375, "learning_rate": 8.706011938601479e-06, "loss": 0.6665, "step": 36918 }, { "epoch": 3.9283890189402, "grad_norm": 18.21875, "learning_rate": 8.70597640704946e-06, "loss": 0.819, "step": 36919 }, { "epoch": 3.9284954245584167, "grad_norm": 14.5078125, "learning_rate": 8.705940875497442e-06, "loss": 0.6968, "step": 36920 }, { "epoch": 3.9286018301766332, "grad_norm": 16.578125, "learning_rate": 8.705905343945424e-06, "loss": 0.7775, "step": 36921 }, { "epoch": 3.9287082357948497, "grad_norm": 9.578125, "learning_rate": 8.705869812393405e-06, "loss": 0.46, "step": 36922 }, { "epoch": 3.9288146414130667, "grad_norm": 11.3125, "learning_rate": 8.705834280841387e-06, "loss": 0.6327, "step": 36923 }, { "epoch": 3.928921047031283, "grad_norm": 15.1484375, "learning_rate": 8.705798749289369e-06, "loss": 0.7515, "step": 36924 }, { "epoch": 3.9290274526494997, "grad_norm": 12.53125, "learning_rate": 8.705763217737352e-06, "loss": 0.6348, "step": 36925 }, { "epoch": 3.9291338582677167, "grad_norm": 10.765625, "learning_rate": 8.705727686185334e-06, "loss": 0.5603, "step": 36926 }, { "epoch": 3.929240263885933, "grad_norm": 14.3359375, "learning_rate": 8.705692154633315e-06, "loss": 0.6289, "step": 36927 }, { "epoch": 3.9293466695041497, "grad_norm": 14.5859375, "learning_rate": 8.705656623081297e-06, "loss": 0.7769, "step": 36928 }, { "epoch": 3.9294530751223666, "grad_norm": 13.0078125, "learning_rate": 8.705621091529278e-06, "loss": 0.7281, "step": 36929 }, { "epoch": 3.929559480740583, "grad_norm": 17.953125, "learning_rate": 8.705585559977262e-06, "loss": 0.7564, "step": 36930 }, { "epoch": 3.9296658863587997, "grad_norm": 13.875, "learning_rate": 8.705550028425243e-06, "loss": 0.6227, "step": 36931 }, { "epoch": 3.9297722919770166, "grad_norm": 14.3671875, "learning_rate": 8.705514496873225e-06, "loss": 0.6273, "step": 36932 }, { "epoch": 3.929878697595233, "grad_norm": 11.359375, "learning_rate": 8.705478965321206e-06, "loss": 0.5173, "step": 36933 }, { "epoch": 3.9299851032134496, "grad_norm": 9.3203125, "learning_rate": 8.705443433769188e-06, "loss": 0.5913, "step": 36934 }, { "epoch": 3.9300915088316666, "grad_norm": 14.0390625, "learning_rate": 8.70540790221717e-06, "loss": 0.747, "step": 36935 }, { "epoch": 3.930197914449883, "grad_norm": 15.859375, "learning_rate": 8.705372370665151e-06, "loss": 0.6169, "step": 36936 }, { "epoch": 3.9303043200680996, "grad_norm": 17.03125, "learning_rate": 8.705336839113133e-06, "loss": 0.7238, "step": 36937 }, { "epoch": 3.930410725686316, "grad_norm": 15.78125, "learning_rate": 8.705301307561115e-06, "loss": 0.8658, "step": 36938 }, { "epoch": 3.9305171313045326, "grad_norm": 13.1484375, "learning_rate": 8.705265776009096e-06, "loss": 0.6827, "step": 36939 }, { "epoch": 3.9306235369227496, "grad_norm": 11.4609375, "learning_rate": 8.70523024445708e-06, "loss": 0.5351, "step": 36940 }, { "epoch": 3.930729942540966, "grad_norm": 13.203125, "learning_rate": 8.705194712905061e-06, "loss": 0.806, "step": 36941 }, { "epoch": 3.9308363481591826, "grad_norm": 16.234375, "learning_rate": 8.705159181353043e-06, "loss": 0.7329, "step": 36942 }, { "epoch": 3.9309427537773995, "grad_norm": 13.7578125, "learning_rate": 8.705123649801024e-06, "loss": 0.7199, "step": 36943 }, { "epoch": 3.931049159395616, "grad_norm": 14.125, "learning_rate": 8.705088118249006e-06, "loss": 0.6773, "step": 36944 }, { "epoch": 3.9311555650138326, "grad_norm": 11.8203125, "learning_rate": 8.705052586696987e-06, "loss": 0.7321, "step": 36945 }, { "epoch": 3.9312619706320495, "grad_norm": 15.796875, "learning_rate": 8.705017055144969e-06, "loss": 0.7018, "step": 36946 }, { "epoch": 3.931368376250266, "grad_norm": 14.484375, "learning_rate": 8.70498152359295e-06, "loss": 0.7034, "step": 36947 }, { "epoch": 3.9314747818684825, "grad_norm": 13.0703125, "learning_rate": 8.704945992040932e-06, "loss": 0.6822, "step": 36948 }, { "epoch": 3.9315811874866995, "grad_norm": 13.03125, "learning_rate": 8.704910460488914e-06, "loss": 0.6615, "step": 36949 }, { "epoch": 3.931687593104916, "grad_norm": 21.53125, "learning_rate": 8.704874928936897e-06, "loss": 0.8201, "step": 36950 }, { "epoch": 3.9317939987231325, "grad_norm": 13.84375, "learning_rate": 8.704839397384879e-06, "loss": 0.5902, "step": 36951 }, { "epoch": 3.9319004043413495, "grad_norm": 12.265625, "learning_rate": 8.70480386583286e-06, "loss": 0.6058, "step": 36952 }, { "epoch": 3.932006809959566, "grad_norm": 12.4296875, "learning_rate": 8.704768334280842e-06, "loss": 0.6348, "step": 36953 }, { "epoch": 3.9321132155777825, "grad_norm": 14.75, "learning_rate": 8.704732802728824e-06, "loss": 0.7078, "step": 36954 }, { "epoch": 3.932219621195999, "grad_norm": 12.7421875, "learning_rate": 8.704697271176805e-06, "loss": 0.6101, "step": 36955 }, { "epoch": 3.932326026814216, "grad_norm": 9.859375, "learning_rate": 8.704661739624789e-06, "loss": 0.6122, "step": 36956 }, { "epoch": 3.9324324324324325, "grad_norm": 13.421875, "learning_rate": 8.70462620807277e-06, "loss": 0.8354, "step": 36957 }, { "epoch": 3.932538838050649, "grad_norm": 15.46875, "learning_rate": 8.704590676520752e-06, "loss": 0.7542, "step": 36958 }, { "epoch": 3.9326452436688655, "grad_norm": 16.171875, "learning_rate": 8.704555144968733e-06, "loss": 0.8341, "step": 36959 }, { "epoch": 3.9327516492870824, "grad_norm": 11.921875, "learning_rate": 8.704519613416715e-06, "loss": 0.7027, "step": 36960 }, { "epoch": 3.932858054905299, "grad_norm": 21.21875, "learning_rate": 8.704484081864697e-06, "loss": 0.795, "step": 36961 }, { "epoch": 3.9329644605235154, "grad_norm": 11.828125, "learning_rate": 8.704448550312678e-06, "loss": 0.6044, "step": 36962 }, { "epoch": 3.9330708661417324, "grad_norm": 10.609375, "learning_rate": 8.70441301876066e-06, "loss": 0.5778, "step": 36963 }, { "epoch": 3.933177271759949, "grad_norm": 10.703125, "learning_rate": 8.704377487208641e-06, "loss": 0.6847, "step": 36964 }, { "epoch": 3.9332836773781654, "grad_norm": 14.0390625, "learning_rate": 8.704341955656623e-06, "loss": 0.5945, "step": 36965 }, { "epoch": 3.9333900829963824, "grad_norm": 15.4375, "learning_rate": 8.704306424104606e-06, "loss": 0.7496, "step": 36966 }, { "epoch": 3.933496488614599, "grad_norm": 15.0546875, "learning_rate": 8.704270892552588e-06, "loss": 0.67, "step": 36967 }, { "epoch": 3.9336028942328154, "grad_norm": 10.890625, "learning_rate": 8.70423536100057e-06, "loss": 0.7213, "step": 36968 }, { "epoch": 3.9337092998510323, "grad_norm": 12.8515625, "learning_rate": 8.704199829448551e-06, "loss": 0.7309, "step": 36969 }, { "epoch": 3.933815705469249, "grad_norm": 14.4140625, "learning_rate": 8.704164297896533e-06, "loss": 0.7316, "step": 36970 }, { "epoch": 3.9339221110874654, "grad_norm": 12.375, "learning_rate": 8.704128766344514e-06, "loss": 0.5841, "step": 36971 }, { "epoch": 3.9340285167056823, "grad_norm": 11.7109375, "learning_rate": 8.704093234792496e-06, "loss": 0.5953, "step": 36972 }, { "epoch": 3.934134922323899, "grad_norm": 11.1015625, "learning_rate": 8.704057703240478e-06, "loss": 0.5658, "step": 36973 }, { "epoch": 3.9342413279421153, "grad_norm": 13.0, "learning_rate": 8.704022171688459e-06, "loss": 0.6144, "step": 36974 }, { "epoch": 3.934347733560332, "grad_norm": 11.953125, "learning_rate": 8.70398664013644e-06, "loss": 0.6752, "step": 36975 }, { "epoch": 3.9344541391785484, "grad_norm": 13.8125, "learning_rate": 8.703951108584424e-06, "loss": 0.8002, "step": 36976 }, { "epoch": 3.9345605447967653, "grad_norm": 11.7109375, "learning_rate": 8.703915577032406e-06, "loss": 0.6379, "step": 36977 }, { "epoch": 3.934666950414982, "grad_norm": 12.4375, "learning_rate": 8.703880045480387e-06, "loss": 0.7031, "step": 36978 }, { "epoch": 3.9347733560331983, "grad_norm": 19.0625, "learning_rate": 8.703844513928369e-06, "loss": 0.67, "step": 36979 }, { "epoch": 3.9348797616514153, "grad_norm": 14.2265625, "learning_rate": 8.70380898237635e-06, "loss": 0.7225, "step": 36980 }, { "epoch": 3.934986167269632, "grad_norm": 16.640625, "learning_rate": 8.703773450824334e-06, "loss": 0.7724, "step": 36981 }, { "epoch": 3.9350925728878483, "grad_norm": 12.3203125, "learning_rate": 8.703737919272315e-06, "loss": 0.5734, "step": 36982 }, { "epoch": 3.9351989785060653, "grad_norm": 11.3125, "learning_rate": 8.703702387720297e-06, "loss": 0.5335, "step": 36983 }, { "epoch": 3.9353053841242818, "grad_norm": 15.921875, "learning_rate": 8.703666856168279e-06, "loss": 0.9589, "step": 36984 }, { "epoch": 3.9354117897424983, "grad_norm": 17.390625, "learning_rate": 8.70363132461626e-06, "loss": 0.7729, "step": 36985 }, { "epoch": 3.9355181953607152, "grad_norm": 18.328125, "learning_rate": 8.703595793064242e-06, "loss": 0.9714, "step": 36986 }, { "epoch": 3.9356246009789317, "grad_norm": 13.625, "learning_rate": 8.703560261512223e-06, "loss": 0.7237, "step": 36987 }, { "epoch": 3.9357310065971483, "grad_norm": 14.3515625, "learning_rate": 8.703524729960205e-06, "loss": 0.6221, "step": 36988 }, { "epoch": 3.935837412215365, "grad_norm": 13.1640625, "learning_rate": 8.703489198408187e-06, "loss": 0.6248, "step": 36989 }, { "epoch": 3.9359438178335817, "grad_norm": 15.3359375, "learning_rate": 8.703453666856168e-06, "loss": 0.6641, "step": 36990 }, { "epoch": 3.9360502234517982, "grad_norm": 9.6796875, "learning_rate": 8.70341813530415e-06, "loss": 0.5593, "step": 36991 }, { "epoch": 3.9361566290700147, "grad_norm": 12.4765625, "learning_rate": 8.703382603752133e-06, "loss": 0.6984, "step": 36992 }, { "epoch": 3.9362630346882317, "grad_norm": 19.046875, "learning_rate": 8.703347072200115e-06, "loss": 0.8087, "step": 36993 }, { "epoch": 3.936369440306448, "grad_norm": 12.546875, "learning_rate": 8.703311540648096e-06, "loss": 0.5704, "step": 36994 }, { "epoch": 3.9364758459246647, "grad_norm": 13.890625, "learning_rate": 8.703276009096078e-06, "loss": 0.6627, "step": 36995 }, { "epoch": 3.936582251542881, "grad_norm": 13.8515625, "learning_rate": 8.70324047754406e-06, "loss": 0.6308, "step": 36996 }, { "epoch": 3.936688657161098, "grad_norm": 15.921875, "learning_rate": 8.703204945992041e-06, "loss": 0.6173, "step": 36997 }, { "epoch": 3.9367950627793147, "grad_norm": 14.75, "learning_rate": 8.703169414440023e-06, "loss": 0.7181, "step": 36998 }, { "epoch": 3.936901468397531, "grad_norm": 12.734375, "learning_rate": 8.703133882888004e-06, "loss": 0.7167, "step": 36999 }, { "epoch": 3.937007874015748, "grad_norm": 14.484375, "learning_rate": 8.703098351335986e-06, "loss": 0.6695, "step": 37000 }, { "epoch": 3.9371142796339647, "grad_norm": 12.1171875, "learning_rate": 8.703062819783968e-06, "loss": 0.6054, "step": 37001 }, { "epoch": 3.937220685252181, "grad_norm": 12.234375, "learning_rate": 8.703027288231951e-06, "loss": 0.6309, "step": 37002 }, { "epoch": 3.937327090870398, "grad_norm": 13.09375, "learning_rate": 8.702991756679933e-06, "loss": 0.5615, "step": 37003 }, { "epoch": 3.9374334964886146, "grad_norm": 11.8828125, "learning_rate": 8.702956225127914e-06, "loss": 0.6443, "step": 37004 }, { "epoch": 3.937539902106831, "grad_norm": 14.921875, "learning_rate": 8.702920693575896e-06, "loss": 0.8003, "step": 37005 }, { "epoch": 3.937646307725048, "grad_norm": 18.34375, "learning_rate": 8.702885162023877e-06, "loss": 0.7188, "step": 37006 }, { "epoch": 3.9377527133432646, "grad_norm": 11.078125, "learning_rate": 8.70284963047186e-06, "loss": 0.7065, "step": 37007 }, { "epoch": 3.937859118961481, "grad_norm": 13.765625, "learning_rate": 8.702814098919842e-06, "loss": 0.7924, "step": 37008 }, { "epoch": 3.937965524579698, "grad_norm": 13.4140625, "learning_rate": 8.702778567367824e-06, "loss": 0.615, "step": 37009 }, { "epoch": 3.9380719301979146, "grad_norm": 11.4765625, "learning_rate": 8.702743035815805e-06, "loss": 0.6783, "step": 37010 }, { "epoch": 3.938178335816131, "grad_norm": 16.65625, "learning_rate": 8.702707504263787e-06, "loss": 0.7717, "step": 37011 }, { "epoch": 3.9382847414343476, "grad_norm": 11.59375, "learning_rate": 8.702671972711769e-06, "loss": 0.6625, "step": 37012 }, { "epoch": 3.938391147052564, "grad_norm": 16.0625, "learning_rate": 8.70263644115975e-06, "loss": 0.7795, "step": 37013 }, { "epoch": 3.938497552670781, "grad_norm": 13.9765625, "learning_rate": 8.702600909607732e-06, "loss": 0.5702, "step": 37014 }, { "epoch": 3.9386039582889976, "grad_norm": 11.3125, "learning_rate": 8.702565378055713e-06, "loss": 0.6038, "step": 37015 }, { "epoch": 3.938710363907214, "grad_norm": 11.6484375, "learning_rate": 8.702529846503695e-06, "loss": 0.621, "step": 37016 }, { "epoch": 3.938816769525431, "grad_norm": 14.4375, "learning_rate": 8.702494314951678e-06, "loss": 0.6766, "step": 37017 }, { "epoch": 3.9389231751436475, "grad_norm": 16.421875, "learning_rate": 8.70245878339966e-06, "loss": 0.8351, "step": 37018 }, { "epoch": 3.939029580761864, "grad_norm": 16.375, "learning_rate": 8.702423251847642e-06, "loss": 0.7572, "step": 37019 }, { "epoch": 3.939135986380081, "grad_norm": 15.765625, "learning_rate": 8.702387720295623e-06, "loss": 0.6951, "step": 37020 }, { "epoch": 3.9392423919982975, "grad_norm": 14.421875, "learning_rate": 8.702352188743605e-06, "loss": 0.9297, "step": 37021 }, { "epoch": 3.939348797616514, "grad_norm": 18.796875, "learning_rate": 8.702316657191588e-06, "loss": 0.7516, "step": 37022 }, { "epoch": 3.939455203234731, "grad_norm": 15.0234375, "learning_rate": 8.70228112563957e-06, "loss": 0.6804, "step": 37023 }, { "epoch": 3.9395616088529475, "grad_norm": 14.2578125, "learning_rate": 8.702245594087551e-06, "loss": 0.6774, "step": 37024 }, { "epoch": 3.939668014471164, "grad_norm": 10.453125, "learning_rate": 8.702210062535533e-06, "loss": 0.7292, "step": 37025 }, { "epoch": 3.939774420089381, "grad_norm": 12.21875, "learning_rate": 8.702174530983515e-06, "loss": 0.5817, "step": 37026 }, { "epoch": 3.9398808257075975, "grad_norm": 10.640625, "learning_rate": 8.702138999431496e-06, "loss": 0.5494, "step": 37027 }, { "epoch": 3.939987231325814, "grad_norm": 12.0546875, "learning_rate": 8.702103467879478e-06, "loss": 0.574, "step": 37028 }, { "epoch": 3.940093636944031, "grad_norm": 12.4453125, "learning_rate": 8.70206793632746e-06, "loss": 0.5932, "step": 37029 }, { "epoch": 3.9402000425622474, "grad_norm": 14.59375, "learning_rate": 8.702032404775441e-06, "loss": 0.7022, "step": 37030 }, { "epoch": 3.940306448180464, "grad_norm": 15.5546875, "learning_rate": 8.701996873223423e-06, "loss": 0.7792, "step": 37031 }, { "epoch": 3.9404128537986804, "grad_norm": 12.203125, "learning_rate": 8.701961341671404e-06, "loss": 0.6309, "step": 37032 }, { "epoch": 3.940519259416897, "grad_norm": 14.0859375, "learning_rate": 8.701925810119387e-06, "loss": 0.7111, "step": 37033 }, { "epoch": 3.940625665035114, "grad_norm": 11.6875, "learning_rate": 8.701890278567369e-06, "loss": 0.5467, "step": 37034 }, { "epoch": 3.9407320706533304, "grad_norm": 16.40625, "learning_rate": 8.70185474701535e-06, "loss": 0.7583, "step": 37035 }, { "epoch": 3.940838476271547, "grad_norm": 10.265625, "learning_rate": 8.701819215463332e-06, "loss": 0.5546, "step": 37036 }, { "epoch": 3.940944881889764, "grad_norm": 15.21875, "learning_rate": 8.701783683911314e-06, "loss": 0.692, "step": 37037 }, { "epoch": 3.9410512875079804, "grad_norm": 12.9609375, "learning_rate": 8.701748152359296e-06, "loss": 0.621, "step": 37038 }, { "epoch": 3.941157693126197, "grad_norm": 17.640625, "learning_rate": 8.701712620807277e-06, "loss": 0.6779, "step": 37039 }, { "epoch": 3.941264098744414, "grad_norm": 15.3046875, "learning_rate": 8.701677089255259e-06, "loss": 0.7461, "step": 37040 }, { "epoch": 3.9413705043626304, "grad_norm": 13.0703125, "learning_rate": 8.70164155770324e-06, "loss": 0.645, "step": 37041 }, { "epoch": 3.941476909980847, "grad_norm": 9.96875, "learning_rate": 8.701606026151222e-06, "loss": 0.6502, "step": 37042 }, { "epoch": 3.941583315599064, "grad_norm": 10.8203125, "learning_rate": 8.701570494599205e-06, "loss": 0.571, "step": 37043 }, { "epoch": 3.9416897212172803, "grad_norm": 15.203125, "learning_rate": 8.701534963047187e-06, "loss": 0.6003, "step": 37044 }, { "epoch": 3.941796126835497, "grad_norm": 13.7421875, "learning_rate": 8.701499431495168e-06, "loss": 0.8441, "step": 37045 }, { "epoch": 3.941902532453714, "grad_norm": 12.359375, "learning_rate": 8.70146389994315e-06, "loss": 0.6371, "step": 37046 }, { "epoch": 3.9420089380719303, "grad_norm": 17.84375, "learning_rate": 8.701428368391132e-06, "loss": 0.7329, "step": 37047 }, { "epoch": 3.942115343690147, "grad_norm": 12.0625, "learning_rate": 8.701392836839115e-06, "loss": 0.602, "step": 37048 }, { "epoch": 3.9422217493083633, "grad_norm": 13.3125, "learning_rate": 8.701357305287097e-06, "loss": 0.6629, "step": 37049 }, { "epoch": 3.9423281549265803, "grad_norm": 12.0546875, "learning_rate": 8.701321773735078e-06, "loss": 0.793, "step": 37050 }, { "epoch": 3.942434560544797, "grad_norm": 14.9375, "learning_rate": 8.70128624218306e-06, "loss": 0.7108, "step": 37051 }, { "epoch": 3.9425409661630133, "grad_norm": 12.6875, "learning_rate": 8.701250710631041e-06, "loss": 0.6393, "step": 37052 }, { "epoch": 3.94264737178123, "grad_norm": 13.359375, "learning_rate": 8.701215179079023e-06, "loss": 0.6572, "step": 37053 }, { "epoch": 3.9427537773994468, "grad_norm": 15.90625, "learning_rate": 8.701179647527005e-06, "loss": 0.6186, "step": 37054 }, { "epoch": 3.9428601830176633, "grad_norm": 11.1015625, "learning_rate": 8.701144115974986e-06, "loss": 0.5849, "step": 37055 }, { "epoch": 3.94296658863588, "grad_norm": 14.875, "learning_rate": 8.701108584422968e-06, "loss": 0.6483, "step": 37056 }, { "epoch": 3.9430729942540967, "grad_norm": 16.734375, "learning_rate": 8.70107305287095e-06, "loss": 0.6298, "step": 37057 }, { "epoch": 3.9431793998723133, "grad_norm": 13.7265625, "learning_rate": 8.701037521318933e-06, "loss": 0.7523, "step": 37058 }, { "epoch": 3.9432858054905298, "grad_norm": 20.328125, "learning_rate": 8.701001989766914e-06, "loss": 0.7365, "step": 37059 }, { "epoch": 3.9433922111087467, "grad_norm": 14.15625, "learning_rate": 8.700966458214896e-06, "loss": 0.7248, "step": 37060 }, { "epoch": 3.9434986167269632, "grad_norm": 13.4453125, "learning_rate": 8.700930926662878e-06, "loss": 0.7039, "step": 37061 }, { "epoch": 3.9436050223451797, "grad_norm": 21.1875, "learning_rate": 8.70089539511086e-06, "loss": 0.8164, "step": 37062 }, { "epoch": 3.9437114279633967, "grad_norm": 11.7734375, "learning_rate": 8.70085986355884e-06, "loss": 0.6502, "step": 37063 }, { "epoch": 3.943817833581613, "grad_norm": 18.59375, "learning_rate": 8.700824332006822e-06, "loss": 0.7543, "step": 37064 }, { "epoch": 3.9439242391998297, "grad_norm": 13.625, "learning_rate": 8.700788800454804e-06, "loss": 0.6999, "step": 37065 }, { "epoch": 3.9440306448180467, "grad_norm": 13.265625, "learning_rate": 8.700753268902786e-06, "loss": 0.6195, "step": 37066 }, { "epoch": 3.944137050436263, "grad_norm": 20.1875, "learning_rate": 8.700717737350767e-06, "loss": 0.857, "step": 37067 }, { "epoch": 3.9442434560544797, "grad_norm": 18.078125, "learning_rate": 8.700682205798749e-06, "loss": 0.9116, "step": 37068 }, { "epoch": 3.944349861672696, "grad_norm": 11.2890625, "learning_rate": 8.700646674246732e-06, "loss": 0.6765, "step": 37069 }, { "epoch": 3.9444562672909127, "grad_norm": 12.9453125, "learning_rate": 8.700611142694714e-06, "loss": 0.6238, "step": 37070 }, { "epoch": 3.9445626729091297, "grad_norm": 13.75, "learning_rate": 8.700575611142695e-06, "loss": 0.7144, "step": 37071 }, { "epoch": 3.944669078527346, "grad_norm": 14.015625, "learning_rate": 8.700540079590677e-06, "loss": 0.7533, "step": 37072 }, { "epoch": 3.9447754841455627, "grad_norm": 12.5, "learning_rate": 8.700504548038659e-06, "loss": 0.6339, "step": 37073 }, { "epoch": 3.9448818897637796, "grad_norm": 14.109375, "learning_rate": 8.700469016486642e-06, "loss": 0.7231, "step": 37074 }, { "epoch": 3.944988295381996, "grad_norm": 12.96875, "learning_rate": 8.700433484934623e-06, "loss": 0.6013, "step": 37075 }, { "epoch": 3.9450947010002126, "grad_norm": 12.328125, "learning_rate": 8.700397953382605e-06, "loss": 0.7013, "step": 37076 }, { "epoch": 3.9452011066184296, "grad_norm": 12.296875, "learning_rate": 8.700362421830587e-06, "loss": 0.5777, "step": 37077 }, { "epoch": 3.945307512236646, "grad_norm": 11.6953125, "learning_rate": 8.700326890278568e-06, "loss": 0.5998, "step": 37078 }, { "epoch": 3.9454139178548626, "grad_norm": 11.171875, "learning_rate": 8.70029135872655e-06, "loss": 0.5632, "step": 37079 }, { "epoch": 3.9455203234730796, "grad_norm": 10.421875, "learning_rate": 8.700255827174531e-06, "loss": 0.5337, "step": 37080 }, { "epoch": 3.945626729091296, "grad_norm": 10.640625, "learning_rate": 8.700220295622513e-06, "loss": 0.6213, "step": 37081 }, { "epoch": 3.9457331347095126, "grad_norm": 16.171875, "learning_rate": 8.700184764070495e-06, "loss": 0.7714, "step": 37082 }, { "epoch": 3.9458395403277295, "grad_norm": 13.75, "learning_rate": 8.700149232518476e-06, "loss": 0.6265, "step": 37083 }, { "epoch": 3.945945945945946, "grad_norm": 11.5078125, "learning_rate": 8.70011370096646e-06, "loss": 0.6446, "step": 37084 }, { "epoch": 3.9460523515641626, "grad_norm": 14.796875, "learning_rate": 8.700078169414441e-06, "loss": 0.8376, "step": 37085 }, { "epoch": 3.946158757182379, "grad_norm": 10.1171875, "learning_rate": 8.700042637862423e-06, "loss": 0.5395, "step": 37086 }, { "epoch": 3.946265162800596, "grad_norm": 11.625, "learning_rate": 8.700007106310404e-06, "loss": 0.5762, "step": 37087 }, { "epoch": 3.9463715684188125, "grad_norm": 15.75, "learning_rate": 8.699971574758386e-06, "loss": 0.6451, "step": 37088 }, { "epoch": 3.946477974037029, "grad_norm": 9.5859375, "learning_rate": 8.699936043206368e-06, "loss": 0.528, "step": 37089 }, { "epoch": 3.9465843796552456, "grad_norm": 10.734375, "learning_rate": 8.69990051165435e-06, "loss": 0.7327, "step": 37090 }, { "epoch": 3.9466907852734625, "grad_norm": 13.15625, "learning_rate": 8.69986498010233e-06, "loss": 0.7354, "step": 37091 }, { "epoch": 3.946797190891679, "grad_norm": 13.5546875, "learning_rate": 8.699829448550312e-06, "loss": 0.6555, "step": 37092 }, { "epoch": 3.9469035965098955, "grad_norm": 14.0, "learning_rate": 8.699793916998294e-06, "loss": 0.6117, "step": 37093 }, { "epoch": 3.9470100021281125, "grad_norm": 15.140625, "learning_rate": 8.699758385446277e-06, "loss": 0.5877, "step": 37094 }, { "epoch": 3.947116407746329, "grad_norm": 11.6171875, "learning_rate": 8.699722853894259e-06, "loss": 0.6586, "step": 37095 }, { "epoch": 3.9472228133645455, "grad_norm": 9.1953125, "learning_rate": 8.69968732234224e-06, "loss": 0.6674, "step": 37096 }, { "epoch": 3.9473292189827625, "grad_norm": 15.7734375, "learning_rate": 8.699651790790222e-06, "loss": 0.7007, "step": 37097 }, { "epoch": 3.947435624600979, "grad_norm": 14.859375, "learning_rate": 8.699616259238204e-06, "loss": 0.6696, "step": 37098 }, { "epoch": 3.9475420302191955, "grad_norm": 13.3203125, "learning_rate": 8.699580727686187e-06, "loss": 0.7194, "step": 37099 }, { "epoch": 3.9476484358374124, "grad_norm": 14.6640625, "learning_rate": 8.699545196134169e-06, "loss": 0.6174, "step": 37100 }, { "epoch": 3.947754841455629, "grad_norm": 11.1328125, "learning_rate": 8.69950966458215e-06, "loss": 0.5426, "step": 37101 }, { "epoch": 3.9478612470738454, "grad_norm": 14.328125, "learning_rate": 8.699474133030132e-06, "loss": 0.5838, "step": 37102 }, { "epoch": 3.9479676526920624, "grad_norm": 10.0, "learning_rate": 8.699438601478114e-06, "loss": 0.6509, "step": 37103 }, { "epoch": 3.948074058310279, "grad_norm": 13.71875, "learning_rate": 8.699403069926095e-06, "loss": 0.6648, "step": 37104 }, { "epoch": 3.9481804639284954, "grad_norm": 11.671875, "learning_rate": 8.699367538374077e-06, "loss": 0.5298, "step": 37105 }, { "epoch": 3.948286869546712, "grad_norm": 16.453125, "learning_rate": 8.699332006822058e-06, "loss": 0.6339, "step": 37106 }, { "epoch": 3.9483932751649284, "grad_norm": 11.9765625, "learning_rate": 8.69929647527004e-06, "loss": 0.6264, "step": 37107 }, { "epoch": 3.9484996807831454, "grad_norm": 14.125, "learning_rate": 8.699260943718022e-06, "loss": 0.7051, "step": 37108 }, { "epoch": 3.948606086401362, "grad_norm": 13.703125, "learning_rate": 8.699225412166003e-06, "loss": 0.6798, "step": 37109 }, { "epoch": 3.9487124920195784, "grad_norm": 12.6015625, "learning_rate": 8.699189880613986e-06, "loss": 0.6986, "step": 37110 }, { "epoch": 3.9488188976377954, "grad_norm": 14.0078125, "learning_rate": 8.699154349061968e-06, "loss": 0.632, "step": 37111 }, { "epoch": 3.948925303256012, "grad_norm": 14.9375, "learning_rate": 8.69911881750995e-06, "loss": 0.679, "step": 37112 }, { "epoch": 3.9490317088742284, "grad_norm": 13.4765625, "learning_rate": 8.699083285957931e-06, "loss": 0.7074, "step": 37113 }, { "epoch": 3.9491381144924453, "grad_norm": 11.6953125, "learning_rate": 8.699047754405913e-06, "loss": 0.6093, "step": 37114 }, { "epoch": 3.949244520110662, "grad_norm": 12.015625, "learning_rate": 8.699012222853896e-06, "loss": 0.6243, "step": 37115 }, { "epoch": 3.9493509257288784, "grad_norm": 16.140625, "learning_rate": 8.698976691301878e-06, "loss": 0.7165, "step": 37116 }, { "epoch": 3.9494573313470953, "grad_norm": 15.953125, "learning_rate": 8.698941159749858e-06, "loss": 0.6697, "step": 37117 }, { "epoch": 3.949563736965312, "grad_norm": 14.6796875, "learning_rate": 8.69890562819784e-06, "loss": 0.7268, "step": 37118 }, { "epoch": 3.9496701425835283, "grad_norm": 12.6484375, "learning_rate": 8.698870096645821e-06, "loss": 0.6608, "step": 37119 }, { "epoch": 3.9497765482017453, "grad_norm": 10.140625, "learning_rate": 8.698834565093804e-06, "loss": 0.5628, "step": 37120 }, { "epoch": 3.949882953819962, "grad_norm": 11.0859375, "learning_rate": 8.698799033541786e-06, "loss": 0.6297, "step": 37121 }, { "epoch": 3.9499893594381783, "grad_norm": 12.7265625, "learning_rate": 8.698763501989767e-06, "loss": 0.7179, "step": 37122 }, { "epoch": 3.950095765056395, "grad_norm": 18.3125, "learning_rate": 8.698727970437749e-06, "loss": 0.757, "step": 37123 }, { "epoch": 3.9502021706746118, "grad_norm": 12.0234375, "learning_rate": 8.69869243888573e-06, "loss": 0.6643, "step": 37124 }, { "epoch": 3.9503085762928283, "grad_norm": 10.765625, "learning_rate": 8.698656907333714e-06, "loss": 0.543, "step": 37125 }, { "epoch": 3.950414981911045, "grad_norm": 14.796875, "learning_rate": 8.698621375781696e-06, "loss": 0.8109, "step": 37126 }, { "epoch": 3.9505213875292613, "grad_norm": 13.46875, "learning_rate": 8.698585844229677e-06, "loss": 0.6358, "step": 37127 }, { "epoch": 3.9506277931474783, "grad_norm": 17.0, "learning_rate": 8.698550312677659e-06, "loss": 0.6928, "step": 37128 }, { "epoch": 3.9507341987656948, "grad_norm": 14.2109375, "learning_rate": 8.69851478112564e-06, "loss": 0.6859, "step": 37129 }, { "epoch": 3.9508406043839113, "grad_norm": 15.53125, "learning_rate": 8.698479249573622e-06, "loss": 0.7751, "step": 37130 }, { "epoch": 3.9509470100021282, "grad_norm": 15.484375, "learning_rate": 8.698443718021604e-06, "loss": 0.7442, "step": 37131 }, { "epoch": 3.9510534156203447, "grad_norm": 11.5859375, "learning_rate": 8.698408186469585e-06, "loss": 0.6364, "step": 37132 }, { "epoch": 3.9511598212385612, "grad_norm": 14.46875, "learning_rate": 8.698372654917567e-06, "loss": 0.6857, "step": 37133 }, { "epoch": 3.951266226856778, "grad_norm": 12.8984375, "learning_rate": 8.698337123365548e-06, "loss": 0.6307, "step": 37134 }, { "epoch": 3.9513726324749947, "grad_norm": 14.3671875, "learning_rate": 8.698301591813532e-06, "loss": 0.7394, "step": 37135 }, { "epoch": 3.951479038093211, "grad_norm": 11.1953125, "learning_rate": 8.698266060261513e-06, "loss": 0.634, "step": 37136 }, { "epoch": 3.951585443711428, "grad_norm": 13.1015625, "learning_rate": 8.698230528709495e-06, "loss": 0.6903, "step": 37137 }, { "epoch": 3.9516918493296447, "grad_norm": 12.96875, "learning_rate": 8.698194997157477e-06, "loss": 0.6074, "step": 37138 }, { "epoch": 3.951798254947861, "grad_norm": 10.046875, "learning_rate": 8.698159465605458e-06, "loss": 0.4883, "step": 37139 }, { "epoch": 3.951904660566078, "grad_norm": 13.7109375, "learning_rate": 8.698123934053441e-06, "loss": 0.6503, "step": 37140 }, { "epoch": 3.9520110661842947, "grad_norm": 16.375, "learning_rate": 8.698088402501423e-06, "loss": 0.6766, "step": 37141 }, { "epoch": 3.952117471802511, "grad_norm": 13.59375, "learning_rate": 8.698052870949405e-06, "loss": 0.7038, "step": 37142 }, { "epoch": 3.9522238774207277, "grad_norm": 13.3984375, "learning_rate": 8.698017339397386e-06, "loss": 0.6785, "step": 37143 }, { "epoch": 3.952330283038944, "grad_norm": 20.984375, "learning_rate": 8.697981807845368e-06, "loss": 0.7409, "step": 37144 }, { "epoch": 3.952436688657161, "grad_norm": 16.328125, "learning_rate": 8.69794627629335e-06, "loss": 0.7604, "step": 37145 }, { "epoch": 3.9525430942753776, "grad_norm": 11.859375, "learning_rate": 8.697910744741331e-06, "loss": 0.6737, "step": 37146 }, { "epoch": 3.952649499893594, "grad_norm": 13.3671875, "learning_rate": 8.697875213189313e-06, "loss": 0.6256, "step": 37147 }, { "epoch": 3.952755905511811, "grad_norm": 17.53125, "learning_rate": 8.697839681637294e-06, "loss": 0.8243, "step": 37148 }, { "epoch": 3.9528623111300276, "grad_norm": 12.34375, "learning_rate": 8.697804150085276e-06, "loss": 0.6441, "step": 37149 }, { "epoch": 3.952968716748244, "grad_norm": 13.625, "learning_rate": 8.697768618533257e-06, "loss": 0.6312, "step": 37150 }, { "epoch": 3.953075122366461, "grad_norm": 11.703125, "learning_rate": 8.69773308698124e-06, "loss": 0.6223, "step": 37151 }, { "epoch": 3.9531815279846776, "grad_norm": 13.640625, "learning_rate": 8.697697555429222e-06, "loss": 0.7142, "step": 37152 }, { "epoch": 3.953287933602894, "grad_norm": 12.7265625, "learning_rate": 8.697662023877204e-06, "loss": 0.6629, "step": 37153 }, { "epoch": 3.953394339221111, "grad_norm": 13.9921875, "learning_rate": 8.697626492325186e-06, "loss": 0.7066, "step": 37154 }, { "epoch": 3.9535007448393276, "grad_norm": 10.4609375, "learning_rate": 8.697590960773167e-06, "loss": 0.4887, "step": 37155 }, { "epoch": 3.953607150457544, "grad_norm": 9.9609375, "learning_rate": 8.697555429221149e-06, "loss": 0.7669, "step": 37156 }, { "epoch": 3.953713556075761, "grad_norm": 16.90625, "learning_rate": 8.69751989766913e-06, "loss": 0.7858, "step": 37157 }, { "epoch": 3.9538199616939775, "grad_norm": 13.4453125, "learning_rate": 8.697484366117112e-06, "loss": 0.8713, "step": 37158 }, { "epoch": 3.953926367312194, "grad_norm": 10.203125, "learning_rate": 8.697448834565094e-06, "loss": 0.5594, "step": 37159 }, { "epoch": 3.954032772930411, "grad_norm": 19.4375, "learning_rate": 8.697413303013075e-06, "loss": 0.7666, "step": 37160 }, { "epoch": 3.9541391785486275, "grad_norm": 15.6484375, "learning_rate": 8.697377771461059e-06, "loss": 0.6568, "step": 37161 }, { "epoch": 3.954245584166844, "grad_norm": 17.078125, "learning_rate": 8.69734223990904e-06, "loss": 0.732, "step": 37162 }, { "epoch": 3.9543519897850605, "grad_norm": 13.5390625, "learning_rate": 8.697306708357022e-06, "loss": 0.7143, "step": 37163 }, { "epoch": 3.954458395403277, "grad_norm": 14.6875, "learning_rate": 8.697271176805003e-06, "loss": 0.77, "step": 37164 }, { "epoch": 3.954564801021494, "grad_norm": 11.2890625, "learning_rate": 8.697235645252985e-06, "loss": 0.4959, "step": 37165 }, { "epoch": 3.9546712066397105, "grad_norm": 13.8828125, "learning_rate": 8.697200113700968e-06, "loss": 0.695, "step": 37166 }, { "epoch": 3.954777612257927, "grad_norm": 11.6484375, "learning_rate": 8.69716458214895e-06, "loss": 0.6036, "step": 37167 }, { "epoch": 3.954884017876144, "grad_norm": 12.1015625, "learning_rate": 8.697129050596931e-06, "loss": 0.6931, "step": 37168 }, { "epoch": 3.9549904234943605, "grad_norm": 13.6953125, "learning_rate": 8.697093519044913e-06, "loss": 0.7459, "step": 37169 }, { "epoch": 3.955096829112577, "grad_norm": 11.1875, "learning_rate": 8.697057987492895e-06, "loss": 0.6016, "step": 37170 }, { "epoch": 3.955203234730794, "grad_norm": 9.8828125, "learning_rate": 8.697022455940876e-06, "loss": 0.6013, "step": 37171 }, { "epoch": 3.9553096403490104, "grad_norm": 9.421875, "learning_rate": 8.696986924388858e-06, "loss": 0.6418, "step": 37172 }, { "epoch": 3.955416045967227, "grad_norm": 16.359375, "learning_rate": 8.69695139283684e-06, "loss": 0.7293, "step": 37173 }, { "epoch": 3.955522451585444, "grad_norm": 10.5625, "learning_rate": 8.696915861284821e-06, "loss": 0.6621, "step": 37174 }, { "epoch": 3.9556288572036604, "grad_norm": 9.453125, "learning_rate": 8.696880329732803e-06, "loss": 0.6108, "step": 37175 }, { "epoch": 3.955735262821877, "grad_norm": 14.859375, "learning_rate": 8.696844798180786e-06, "loss": 0.7376, "step": 37176 }, { "epoch": 3.955841668440094, "grad_norm": 17.21875, "learning_rate": 8.696809266628768e-06, "loss": 0.7245, "step": 37177 }, { "epoch": 3.9559480740583104, "grad_norm": 15.5234375, "learning_rate": 8.69677373507675e-06, "loss": 0.683, "step": 37178 }, { "epoch": 3.956054479676527, "grad_norm": 16.0, "learning_rate": 8.696738203524731e-06, "loss": 0.7082, "step": 37179 }, { "epoch": 3.9561608852947434, "grad_norm": 13.78125, "learning_rate": 8.696702671972712e-06, "loss": 0.7457, "step": 37180 }, { "epoch": 3.9562672909129604, "grad_norm": 20.640625, "learning_rate": 8.696667140420694e-06, "loss": 0.6556, "step": 37181 }, { "epoch": 3.956373696531177, "grad_norm": 17.5, "learning_rate": 8.696631608868676e-06, "loss": 0.6774, "step": 37182 }, { "epoch": 3.9564801021493934, "grad_norm": 17.109375, "learning_rate": 8.696596077316657e-06, "loss": 0.8308, "step": 37183 }, { "epoch": 3.95658650776761, "grad_norm": 13.7421875, "learning_rate": 8.696560545764639e-06, "loss": 0.7843, "step": 37184 }, { "epoch": 3.956692913385827, "grad_norm": 13.703125, "learning_rate": 8.69652501421262e-06, "loss": 0.7339, "step": 37185 }, { "epoch": 3.9567993190040434, "grad_norm": 13.5078125, "learning_rate": 8.696489482660602e-06, "loss": 0.7618, "step": 37186 }, { "epoch": 3.95690572462226, "grad_norm": 10.8125, "learning_rate": 8.696453951108585e-06, "loss": 0.591, "step": 37187 }, { "epoch": 3.957012130240477, "grad_norm": 11.421875, "learning_rate": 8.696418419556567e-06, "loss": 0.6835, "step": 37188 }, { "epoch": 3.9571185358586933, "grad_norm": 15.828125, "learning_rate": 8.696382888004549e-06, "loss": 0.8172, "step": 37189 }, { "epoch": 3.95722494147691, "grad_norm": 17.3125, "learning_rate": 8.69634735645253e-06, "loss": 0.7197, "step": 37190 }, { "epoch": 3.957331347095127, "grad_norm": 14.53125, "learning_rate": 8.696311824900512e-06, "loss": 0.772, "step": 37191 }, { "epoch": 3.9574377527133433, "grad_norm": 13.703125, "learning_rate": 8.696276293348495e-06, "loss": 0.61, "step": 37192 }, { "epoch": 3.95754415833156, "grad_norm": 14.25, "learning_rate": 8.696240761796477e-06, "loss": 0.6984, "step": 37193 }, { "epoch": 3.9576505639497768, "grad_norm": 14.1953125, "learning_rate": 8.696205230244458e-06, "loss": 0.6586, "step": 37194 }, { "epoch": 3.9577569695679933, "grad_norm": 16.390625, "learning_rate": 8.69616969869244e-06, "loss": 0.9735, "step": 37195 }, { "epoch": 3.95786337518621, "grad_norm": 10.9921875, "learning_rate": 8.696134167140422e-06, "loss": 0.535, "step": 37196 }, { "epoch": 3.9579697808044267, "grad_norm": 12.8828125, "learning_rate": 8.696098635588403e-06, "loss": 0.6229, "step": 37197 }, { "epoch": 3.9580761864226433, "grad_norm": 13.78125, "learning_rate": 8.696063104036385e-06, "loss": 0.6372, "step": 37198 }, { "epoch": 3.9581825920408598, "grad_norm": 13.9296875, "learning_rate": 8.696027572484366e-06, "loss": 0.6012, "step": 37199 }, { "epoch": 3.9582889976590763, "grad_norm": 15.40625, "learning_rate": 8.695992040932348e-06, "loss": 0.7627, "step": 37200 }, { "epoch": 3.958395403277293, "grad_norm": 12.8515625, "learning_rate": 8.69595650938033e-06, "loss": 0.677, "step": 37201 }, { "epoch": 3.9585018088955097, "grad_norm": 9.9140625, "learning_rate": 8.695920977828313e-06, "loss": 0.5483, "step": 37202 }, { "epoch": 3.9586082145137262, "grad_norm": 12.46875, "learning_rate": 8.695885446276294e-06, "loss": 0.6746, "step": 37203 }, { "epoch": 3.9587146201319428, "grad_norm": 12.6796875, "learning_rate": 8.695849914724276e-06, "loss": 0.6868, "step": 37204 }, { "epoch": 3.9588210257501597, "grad_norm": 12.4921875, "learning_rate": 8.695814383172258e-06, "loss": 0.6601, "step": 37205 }, { "epoch": 3.958927431368376, "grad_norm": 17.140625, "learning_rate": 8.69577885162024e-06, "loss": 0.8185, "step": 37206 }, { "epoch": 3.9590338369865927, "grad_norm": 15.7890625, "learning_rate": 8.695743320068221e-06, "loss": 0.8073, "step": 37207 }, { "epoch": 3.9591402426048097, "grad_norm": 10.953125, "learning_rate": 8.695707788516203e-06, "loss": 0.6235, "step": 37208 }, { "epoch": 3.959246648223026, "grad_norm": 11.5625, "learning_rate": 8.695672256964184e-06, "loss": 0.5531, "step": 37209 }, { "epoch": 3.9593530538412427, "grad_norm": 12.09375, "learning_rate": 8.695636725412166e-06, "loss": 0.7056, "step": 37210 }, { "epoch": 3.9594594594594597, "grad_norm": 13.71875, "learning_rate": 8.695601193860147e-06, "loss": 0.5763, "step": 37211 }, { "epoch": 3.959565865077676, "grad_norm": 14.640625, "learning_rate": 8.69556566230813e-06, "loss": 0.8013, "step": 37212 }, { "epoch": 3.9596722706958927, "grad_norm": 15.0546875, "learning_rate": 8.695530130756112e-06, "loss": 0.8319, "step": 37213 }, { "epoch": 3.9597786763141096, "grad_norm": 17.625, "learning_rate": 8.695494599204094e-06, "loss": 0.7946, "step": 37214 }, { "epoch": 3.959885081932326, "grad_norm": 11.5546875, "learning_rate": 8.695459067652075e-06, "loss": 0.5652, "step": 37215 }, { "epoch": 3.9599914875505426, "grad_norm": 13.3125, "learning_rate": 8.695423536100057e-06, "loss": 0.6527, "step": 37216 }, { "epoch": 3.960097893168759, "grad_norm": 12.265625, "learning_rate": 8.69538800454804e-06, "loss": 0.5433, "step": 37217 }, { "epoch": 3.960204298786976, "grad_norm": 10.7890625, "learning_rate": 8.695352472996022e-06, "loss": 0.6026, "step": 37218 }, { "epoch": 3.9603107044051926, "grad_norm": 13.84375, "learning_rate": 8.695316941444004e-06, "loss": 0.7723, "step": 37219 }, { "epoch": 3.960417110023409, "grad_norm": 14.5703125, "learning_rate": 8.695281409891985e-06, "loss": 0.6927, "step": 37220 }, { "epoch": 3.9605235156416256, "grad_norm": 17.34375, "learning_rate": 8.695245878339967e-06, "loss": 0.7083, "step": 37221 }, { "epoch": 3.9606299212598426, "grad_norm": 12.734375, "learning_rate": 8.695210346787948e-06, "loss": 0.6161, "step": 37222 }, { "epoch": 3.960736326878059, "grad_norm": 15.5625, "learning_rate": 8.69517481523593e-06, "loss": 0.6863, "step": 37223 }, { "epoch": 3.9608427324962756, "grad_norm": 12.4921875, "learning_rate": 8.695139283683912e-06, "loss": 0.736, "step": 37224 }, { "epoch": 3.9609491381144926, "grad_norm": 11.8203125, "learning_rate": 8.695103752131893e-06, "loss": 0.6475, "step": 37225 }, { "epoch": 3.961055543732709, "grad_norm": 15.9296875, "learning_rate": 8.695068220579875e-06, "loss": 0.7426, "step": 37226 }, { "epoch": 3.9611619493509256, "grad_norm": 14.125, "learning_rate": 8.695032689027856e-06, "loss": 0.737, "step": 37227 }, { "epoch": 3.9612683549691425, "grad_norm": 13.78125, "learning_rate": 8.69499715747584e-06, "loss": 0.6969, "step": 37228 }, { "epoch": 3.961374760587359, "grad_norm": 15.6953125, "learning_rate": 8.694961625923821e-06, "loss": 0.6779, "step": 37229 }, { "epoch": 3.9614811662055756, "grad_norm": 11.1328125, "learning_rate": 8.694926094371803e-06, "loss": 0.6486, "step": 37230 }, { "epoch": 3.9615875718237925, "grad_norm": 11.390625, "learning_rate": 8.694890562819785e-06, "loss": 0.7674, "step": 37231 }, { "epoch": 3.961693977442009, "grad_norm": 11.3671875, "learning_rate": 8.694855031267766e-06, "loss": 0.6114, "step": 37232 }, { "epoch": 3.9618003830602255, "grad_norm": 10.8515625, "learning_rate": 8.69481949971575e-06, "loss": 0.7269, "step": 37233 }, { "epoch": 3.9619067886784425, "grad_norm": 9.8671875, "learning_rate": 8.694783968163731e-06, "loss": 0.6181, "step": 37234 }, { "epoch": 3.962013194296659, "grad_norm": 13.8515625, "learning_rate": 8.694748436611713e-06, "loss": 0.8214, "step": 37235 }, { "epoch": 3.9621195999148755, "grad_norm": 8.3125, "learning_rate": 8.694712905059694e-06, "loss": 0.4572, "step": 37236 }, { "epoch": 3.962226005533092, "grad_norm": 21.171875, "learning_rate": 8.694677373507674e-06, "loss": 0.7425, "step": 37237 }, { "epoch": 3.9623324111513085, "grad_norm": 12.8125, "learning_rate": 8.694641841955658e-06, "loss": 0.6902, "step": 37238 }, { "epoch": 3.9624388167695255, "grad_norm": 9.4609375, "learning_rate": 8.694606310403639e-06, "loss": 0.5239, "step": 37239 }, { "epoch": 3.962545222387742, "grad_norm": 11.125, "learning_rate": 8.69457077885162e-06, "loss": 0.4649, "step": 37240 }, { "epoch": 3.9626516280059585, "grad_norm": 13.8515625, "learning_rate": 8.694535247299602e-06, "loss": 0.7369, "step": 37241 }, { "epoch": 3.9627580336241754, "grad_norm": 12.0, "learning_rate": 8.694499715747584e-06, "loss": 0.7282, "step": 37242 }, { "epoch": 3.962864439242392, "grad_norm": 13.203125, "learning_rate": 8.694464184195567e-06, "loss": 0.6812, "step": 37243 }, { "epoch": 3.9629708448606085, "grad_norm": 13.234375, "learning_rate": 8.694428652643549e-06, "loss": 0.7107, "step": 37244 }, { "epoch": 3.9630772504788254, "grad_norm": 10.96875, "learning_rate": 8.69439312109153e-06, "loss": 0.7364, "step": 37245 }, { "epoch": 3.963183656097042, "grad_norm": 11.9921875, "learning_rate": 8.694357589539512e-06, "loss": 0.654, "step": 37246 }, { "epoch": 3.9632900617152584, "grad_norm": 12.8046875, "learning_rate": 8.694322057987494e-06, "loss": 0.6244, "step": 37247 }, { "epoch": 3.9633964673334754, "grad_norm": 17.65625, "learning_rate": 8.694286526435475e-06, "loss": 0.7835, "step": 37248 }, { "epoch": 3.963502872951692, "grad_norm": 14.359375, "learning_rate": 8.694250994883457e-06, "loss": 0.7509, "step": 37249 }, { "epoch": 3.9636092785699084, "grad_norm": 12.96875, "learning_rate": 8.694215463331438e-06, "loss": 0.6487, "step": 37250 }, { "epoch": 3.9637156841881254, "grad_norm": 14.4765625, "learning_rate": 8.69417993177942e-06, "loss": 0.6627, "step": 37251 }, { "epoch": 3.963822089806342, "grad_norm": 13.4609375, "learning_rate": 8.694144400227402e-06, "loss": 0.64, "step": 37252 }, { "epoch": 3.9639284954245584, "grad_norm": 13.4609375, "learning_rate": 8.694108868675385e-06, "loss": 0.6384, "step": 37253 }, { "epoch": 3.964034901042775, "grad_norm": 12.6328125, "learning_rate": 8.694073337123367e-06, "loss": 0.6734, "step": 37254 }, { "epoch": 3.964141306660992, "grad_norm": 10.7109375, "learning_rate": 8.694037805571348e-06, "loss": 0.5728, "step": 37255 }, { "epoch": 3.9642477122792084, "grad_norm": 15.546875, "learning_rate": 8.69400227401933e-06, "loss": 0.5973, "step": 37256 }, { "epoch": 3.964354117897425, "grad_norm": 13.0078125, "learning_rate": 8.693966742467311e-06, "loss": 0.6091, "step": 37257 }, { "epoch": 3.9644605235156414, "grad_norm": 14.234375, "learning_rate": 8.693931210915295e-06, "loss": 0.7403, "step": 37258 }, { "epoch": 3.9645669291338583, "grad_norm": 12.6875, "learning_rate": 8.693895679363276e-06, "loss": 0.6445, "step": 37259 }, { "epoch": 3.964673334752075, "grad_norm": 12.5703125, "learning_rate": 8.693860147811258e-06, "loss": 0.8867, "step": 37260 }, { "epoch": 3.9647797403702913, "grad_norm": 13.828125, "learning_rate": 8.69382461625924e-06, "loss": 0.6462, "step": 37261 }, { "epoch": 3.9648861459885083, "grad_norm": 10.2734375, "learning_rate": 8.693789084707221e-06, "loss": 0.5367, "step": 37262 }, { "epoch": 3.964992551606725, "grad_norm": 13.765625, "learning_rate": 8.693753553155203e-06, "loss": 0.6621, "step": 37263 }, { "epoch": 3.9650989572249413, "grad_norm": 12.9375, "learning_rate": 8.693718021603184e-06, "loss": 0.699, "step": 37264 }, { "epoch": 3.9652053628431583, "grad_norm": 14.578125, "learning_rate": 8.693682490051166e-06, "loss": 0.6198, "step": 37265 }, { "epoch": 3.965311768461375, "grad_norm": 15.4609375, "learning_rate": 8.693646958499148e-06, "loss": 0.7483, "step": 37266 }, { "epoch": 3.9654181740795913, "grad_norm": 13.0234375, "learning_rate": 8.69361142694713e-06, "loss": 0.6143, "step": 37267 }, { "epoch": 3.9655245796978082, "grad_norm": 10.859375, "learning_rate": 8.69357589539511e-06, "loss": 0.5502, "step": 37268 }, { "epoch": 3.9656309853160248, "grad_norm": 12.8046875, "learning_rate": 8.693540363843094e-06, "loss": 0.5938, "step": 37269 }, { "epoch": 3.9657373909342413, "grad_norm": 14.3515625, "learning_rate": 8.693504832291076e-06, "loss": 0.635, "step": 37270 }, { "epoch": 3.9658437965524582, "grad_norm": 14.375, "learning_rate": 8.693469300739057e-06, "loss": 0.7255, "step": 37271 }, { "epoch": 3.9659502021706747, "grad_norm": 13.9453125, "learning_rate": 8.693433769187039e-06, "loss": 0.7487, "step": 37272 }, { "epoch": 3.9660566077888912, "grad_norm": 10.625, "learning_rate": 8.69339823763502e-06, "loss": 0.5972, "step": 37273 }, { "epoch": 3.9661630134071078, "grad_norm": 13.28125, "learning_rate": 8.693362706083002e-06, "loss": 0.6758, "step": 37274 }, { "epoch": 3.9662694190253243, "grad_norm": 14.0234375, "learning_rate": 8.693327174530984e-06, "loss": 0.6762, "step": 37275 }, { "epoch": 3.966375824643541, "grad_norm": 13.8359375, "learning_rate": 8.693291642978965e-06, "loss": 0.6741, "step": 37276 }, { "epoch": 3.9664822302617577, "grad_norm": 12.8828125, "learning_rate": 8.693256111426947e-06, "loss": 0.7477, "step": 37277 }, { "epoch": 3.9665886358799742, "grad_norm": 13.4140625, "learning_rate": 8.693220579874929e-06, "loss": 0.6061, "step": 37278 }, { "epoch": 3.966695041498191, "grad_norm": 13.765625, "learning_rate": 8.693185048322912e-06, "loss": 0.6722, "step": 37279 }, { "epoch": 3.9668014471164077, "grad_norm": 11.7265625, "learning_rate": 8.693149516770893e-06, "loss": 0.6236, "step": 37280 }, { "epoch": 3.966907852734624, "grad_norm": 12.8671875, "learning_rate": 8.693113985218875e-06, "loss": 0.6893, "step": 37281 }, { "epoch": 3.967014258352841, "grad_norm": 10.8515625, "learning_rate": 8.693078453666857e-06, "loss": 0.5882, "step": 37282 }, { "epoch": 3.9671206639710577, "grad_norm": 13.859375, "learning_rate": 8.693042922114838e-06, "loss": 0.7169, "step": 37283 }, { "epoch": 3.967227069589274, "grad_norm": 14.140625, "learning_rate": 8.693007390562822e-06, "loss": 0.7761, "step": 37284 }, { "epoch": 3.967333475207491, "grad_norm": 16.859375, "learning_rate": 8.692971859010803e-06, "loss": 0.7615, "step": 37285 }, { "epoch": 3.9674398808257076, "grad_norm": 10.84375, "learning_rate": 8.692936327458785e-06, "loss": 0.6205, "step": 37286 }, { "epoch": 3.967546286443924, "grad_norm": 15.859375, "learning_rate": 8.692900795906766e-06, "loss": 0.9206, "step": 37287 }, { "epoch": 3.967652692062141, "grad_norm": 14.796875, "learning_rate": 8.692865264354748e-06, "loss": 0.8026, "step": 37288 }, { "epoch": 3.9677590976803576, "grad_norm": 14.5859375, "learning_rate": 8.69282973280273e-06, "loss": 0.6804, "step": 37289 }, { "epoch": 3.967865503298574, "grad_norm": 13.359375, "learning_rate": 8.692794201250711e-06, "loss": 0.7859, "step": 37290 }, { "epoch": 3.967971908916791, "grad_norm": 12.140625, "learning_rate": 8.692758669698693e-06, "loss": 0.771, "step": 37291 }, { "epoch": 3.9680783145350076, "grad_norm": 12.0, "learning_rate": 8.692723138146674e-06, "loss": 0.5663, "step": 37292 }, { "epoch": 3.968184720153224, "grad_norm": 15.328125, "learning_rate": 8.692687606594656e-06, "loss": 0.7241, "step": 37293 }, { "epoch": 3.9682911257714406, "grad_norm": 13.1953125, "learning_rate": 8.69265207504264e-06, "loss": 0.6286, "step": 37294 }, { "epoch": 3.968397531389657, "grad_norm": 15.078125, "learning_rate": 8.692616543490621e-06, "loss": 0.8416, "step": 37295 }, { "epoch": 3.968503937007874, "grad_norm": 10.640625, "learning_rate": 8.692581011938603e-06, "loss": 0.5453, "step": 37296 }, { "epoch": 3.9686103426260906, "grad_norm": 14.84375, "learning_rate": 8.692545480386584e-06, "loss": 0.675, "step": 37297 }, { "epoch": 3.968716748244307, "grad_norm": 13.3671875, "learning_rate": 8.692509948834566e-06, "loss": 0.6044, "step": 37298 }, { "epoch": 3.968823153862524, "grad_norm": 12.3125, "learning_rate": 8.692474417282547e-06, "loss": 0.5693, "step": 37299 }, { "epoch": 3.9689295594807406, "grad_norm": 12.3515625, "learning_rate": 8.692438885730529e-06, "loss": 0.85, "step": 37300 }, { "epoch": 3.969035965098957, "grad_norm": 11.9375, "learning_rate": 8.69240335417851e-06, "loss": 0.5664, "step": 37301 }, { "epoch": 3.969142370717174, "grad_norm": 11.0546875, "learning_rate": 8.692367822626492e-06, "loss": 0.602, "step": 37302 }, { "epoch": 3.9692487763353905, "grad_norm": 18.59375, "learning_rate": 8.692332291074474e-06, "loss": 0.8086, "step": 37303 }, { "epoch": 3.969355181953607, "grad_norm": 15.875, "learning_rate": 8.692296759522455e-06, "loss": 0.7758, "step": 37304 }, { "epoch": 3.969461587571824, "grad_norm": 12.7421875, "learning_rate": 8.692261227970439e-06, "loss": 0.6332, "step": 37305 }, { "epoch": 3.9695679931900405, "grad_norm": 18.875, "learning_rate": 8.69222569641842e-06, "loss": 0.7446, "step": 37306 }, { "epoch": 3.969674398808257, "grad_norm": 18.90625, "learning_rate": 8.692190164866402e-06, "loss": 0.8497, "step": 37307 }, { "epoch": 3.969780804426474, "grad_norm": 13.375, "learning_rate": 8.692154633314384e-06, "loss": 0.6842, "step": 37308 }, { "epoch": 3.9698872100446905, "grad_norm": 11.015625, "learning_rate": 8.692119101762365e-06, "loss": 0.7164, "step": 37309 }, { "epoch": 3.969993615662907, "grad_norm": 12.65625, "learning_rate": 8.692083570210348e-06, "loss": 0.6144, "step": 37310 }, { "epoch": 3.9701000212811235, "grad_norm": 15.28125, "learning_rate": 8.69204803865833e-06, "loss": 0.6611, "step": 37311 }, { "epoch": 3.9702064268993404, "grad_norm": 12.9296875, "learning_rate": 8.692012507106312e-06, "loss": 0.6777, "step": 37312 }, { "epoch": 3.970312832517557, "grad_norm": 13.4140625, "learning_rate": 8.691976975554293e-06, "loss": 0.6426, "step": 37313 }, { "epoch": 3.9704192381357735, "grad_norm": 11.7109375, "learning_rate": 8.691941444002275e-06, "loss": 0.6011, "step": 37314 }, { "epoch": 3.97052564375399, "grad_norm": 16.265625, "learning_rate": 8.691905912450256e-06, "loss": 0.752, "step": 37315 }, { "epoch": 3.970632049372207, "grad_norm": 13.515625, "learning_rate": 8.691870380898238e-06, "loss": 0.5874, "step": 37316 }, { "epoch": 3.9707384549904234, "grad_norm": 14.7265625, "learning_rate": 8.69183484934622e-06, "loss": 0.7031, "step": 37317 }, { "epoch": 3.97084486060864, "grad_norm": 14.078125, "learning_rate": 8.691799317794201e-06, "loss": 0.6572, "step": 37318 }, { "epoch": 3.970951266226857, "grad_norm": 13.921875, "learning_rate": 8.691763786242183e-06, "loss": 0.7579, "step": 37319 }, { "epoch": 3.9710576718450734, "grad_norm": 14.9140625, "learning_rate": 8.691728254690166e-06, "loss": 0.7021, "step": 37320 }, { "epoch": 3.97116407746329, "grad_norm": 13.2265625, "learning_rate": 8.691692723138148e-06, "loss": 0.5966, "step": 37321 }, { "epoch": 3.971270483081507, "grad_norm": 14.546875, "learning_rate": 8.69165719158613e-06, "loss": 0.6747, "step": 37322 }, { "epoch": 3.9713768886997234, "grad_norm": 10.9609375, "learning_rate": 8.691621660034111e-06, "loss": 0.7573, "step": 37323 }, { "epoch": 3.97148329431794, "grad_norm": 12.9921875, "learning_rate": 8.691586128482093e-06, "loss": 0.6597, "step": 37324 }, { "epoch": 3.971589699936157, "grad_norm": 15.2734375, "learning_rate": 8.691550596930076e-06, "loss": 0.6995, "step": 37325 }, { "epoch": 3.9716961055543734, "grad_norm": 12.421875, "learning_rate": 8.691515065378056e-06, "loss": 0.8718, "step": 37326 }, { "epoch": 3.97180251117259, "grad_norm": 13.0546875, "learning_rate": 8.691479533826037e-06, "loss": 0.7234, "step": 37327 }, { "epoch": 3.971908916790807, "grad_norm": 12.1640625, "learning_rate": 8.691444002274019e-06, "loss": 0.5702, "step": 37328 }, { "epoch": 3.9720153224090233, "grad_norm": 12.765625, "learning_rate": 8.691408470722e-06, "loss": 0.6758, "step": 37329 }, { "epoch": 3.97212172802724, "grad_norm": 15.6953125, "learning_rate": 8.691372939169984e-06, "loss": 0.679, "step": 37330 }, { "epoch": 3.9722281336454563, "grad_norm": 13.1796875, "learning_rate": 8.691337407617966e-06, "loss": 0.6997, "step": 37331 }, { "epoch": 3.972334539263673, "grad_norm": 15.171875, "learning_rate": 8.691301876065947e-06, "loss": 0.638, "step": 37332 }, { "epoch": 3.97244094488189, "grad_norm": 13.4375, "learning_rate": 8.691266344513929e-06, "loss": 0.6216, "step": 37333 }, { "epoch": 3.9725473505001063, "grad_norm": 12.8359375, "learning_rate": 8.69123081296191e-06, "loss": 0.6428, "step": 37334 }, { "epoch": 3.972653756118323, "grad_norm": 13.125, "learning_rate": 8.691195281409894e-06, "loss": 0.6614, "step": 37335 }, { "epoch": 3.97276016173654, "grad_norm": 15.0390625, "learning_rate": 8.691159749857875e-06, "loss": 0.7465, "step": 37336 }, { "epoch": 3.9728665673547563, "grad_norm": 11.2265625, "learning_rate": 8.691124218305857e-06, "loss": 0.5673, "step": 37337 }, { "epoch": 3.972972972972973, "grad_norm": 13.25, "learning_rate": 8.691088686753839e-06, "loss": 0.5921, "step": 37338 }, { "epoch": 3.9730793785911898, "grad_norm": 15.9921875, "learning_rate": 8.69105315520182e-06, "loss": 0.7167, "step": 37339 }, { "epoch": 3.9731857842094063, "grad_norm": 14.6953125, "learning_rate": 8.691017623649802e-06, "loss": 0.7148, "step": 37340 }, { "epoch": 3.9732921898276228, "grad_norm": 11.546875, "learning_rate": 8.690982092097783e-06, "loss": 0.6082, "step": 37341 }, { "epoch": 3.9733985954458397, "grad_norm": 11.890625, "learning_rate": 8.690946560545765e-06, "loss": 0.688, "step": 37342 }, { "epoch": 3.9735050010640562, "grad_norm": 12.203125, "learning_rate": 8.690911028993747e-06, "loss": 0.7741, "step": 37343 }, { "epoch": 3.9736114066822728, "grad_norm": 12.59375, "learning_rate": 8.690875497441728e-06, "loss": 0.5891, "step": 37344 }, { "epoch": 3.9737178123004897, "grad_norm": 13.3671875, "learning_rate": 8.69083996588971e-06, "loss": 0.6257, "step": 37345 }, { "epoch": 3.973824217918706, "grad_norm": 11.7890625, "learning_rate": 8.690804434337693e-06, "loss": 0.6053, "step": 37346 }, { "epoch": 3.9739306235369227, "grad_norm": 11.2578125, "learning_rate": 8.690768902785675e-06, "loss": 0.5734, "step": 37347 }, { "epoch": 3.9740370291551392, "grad_norm": 12.7265625, "learning_rate": 8.690733371233656e-06, "loss": 0.6961, "step": 37348 }, { "epoch": 3.974143434773356, "grad_norm": 11.25, "learning_rate": 8.690697839681638e-06, "loss": 0.5813, "step": 37349 }, { "epoch": 3.9742498403915727, "grad_norm": 11.1328125, "learning_rate": 8.69066230812962e-06, "loss": 0.6534, "step": 37350 }, { "epoch": 3.974356246009789, "grad_norm": 14.5625, "learning_rate": 8.690626776577603e-06, "loss": 0.6906, "step": 37351 }, { "epoch": 3.9744626516280057, "grad_norm": 14.0078125, "learning_rate": 8.690591245025584e-06, "loss": 0.6201, "step": 37352 }, { "epoch": 3.9745690572462227, "grad_norm": 11.65625, "learning_rate": 8.690555713473566e-06, "loss": 0.6033, "step": 37353 }, { "epoch": 3.974675462864439, "grad_norm": 14.671875, "learning_rate": 8.690520181921548e-06, "loss": 0.7692, "step": 37354 }, { "epoch": 3.9747818684826557, "grad_norm": 14.203125, "learning_rate": 8.69048465036953e-06, "loss": 0.6834, "step": 37355 }, { "epoch": 3.9748882741008726, "grad_norm": 13.453125, "learning_rate": 8.69044911881751e-06, "loss": 0.7264, "step": 37356 }, { "epoch": 3.974994679719089, "grad_norm": 13.765625, "learning_rate": 8.690413587265492e-06, "loss": 0.7926, "step": 37357 }, { "epoch": 3.9751010853373057, "grad_norm": 12.5625, "learning_rate": 8.690378055713474e-06, "loss": 0.7224, "step": 37358 }, { "epoch": 3.9752074909555226, "grad_norm": 10.96875, "learning_rate": 8.690342524161456e-06, "loss": 0.5987, "step": 37359 }, { "epoch": 3.975313896573739, "grad_norm": 19.265625, "learning_rate": 8.690306992609437e-06, "loss": 0.5786, "step": 37360 }, { "epoch": 3.9754203021919556, "grad_norm": 13.3828125, "learning_rate": 8.69027146105742e-06, "loss": 0.7319, "step": 37361 }, { "epoch": 3.9755267078101726, "grad_norm": 12.4296875, "learning_rate": 8.690235929505402e-06, "loss": 0.62, "step": 37362 }, { "epoch": 3.975633113428389, "grad_norm": 13.875, "learning_rate": 8.690200397953384e-06, "loss": 0.8162, "step": 37363 }, { "epoch": 3.9757395190466056, "grad_norm": 9.765625, "learning_rate": 8.690164866401365e-06, "loss": 0.5976, "step": 37364 }, { "epoch": 3.9758459246648226, "grad_norm": 11.90625, "learning_rate": 8.690129334849347e-06, "loss": 0.5432, "step": 37365 }, { "epoch": 3.975952330283039, "grad_norm": 10.7265625, "learning_rate": 8.690093803297329e-06, "loss": 0.5287, "step": 37366 }, { "epoch": 3.9760587359012556, "grad_norm": 13.1171875, "learning_rate": 8.69005827174531e-06, "loss": 0.629, "step": 37367 }, { "epoch": 3.976165141519472, "grad_norm": 13.9765625, "learning_rate": 8.690022740193292e-06, "loss": 0.5751, "step": 37368 }, { "epoch": 3.9762715471376886, "grad_norm": 11.5390625, "learning_rate": 8.689987208641273e-06, "loss": 0.7972, "step": 37369 }, { "epoch": 3.9763779527559056, "grad_norm": 11.171875, "learning_rate": 8.689951677089255e-06, "loss": 0.6656, "step": 37370 }, { "epoch": 3.976484358374122, "grad_norm": 13.8828125, "learning_rate": 8.689916145537238e-06, "loss": 0.6846, "step": 37371 }, { "epoch": 3.9765907639923386, "grad_norm": 14.765625, "learning_rate": 8.68988061398522e-06, "loss": 0.6561, "step": 37372 }, { "epoch": 3.9766971696105555, "grad_norm": 11.890625, "learning_rate": 8.689845082433202e-06, "loss": 0.5712, "step": 37373 }, { "epoch": 3.976803575228772, "grad_norm": 12.265625, "learning_rate": 8.689809550881183e-06, "loss": 0.6678, "step": 37374 }, { "epoch": 3.9769099808469885, "grad_norm": 11.0703125, "learning_rate": 8.689774019329165e-06, "loss": 0.4923, "step": 37375 }, { "epoch": 3.9770163864652055, "grad_norm": 9.9921875, "learning_rate": 8.689738487777148e-06, "loss": 0.4927, "step": 37376 }, { "epoch": 3.977122792083422, "grad_norm": 15.9140625, "learning_rate": 8.68970295622513e-06, "loss": 0.7631, "step": 37377 }, { "epoch": 3.9772291977016385, "grad_norm": 11.546875, "learning_rate": 8.689667424673111e-06, "loss": 0.5557, "step": 37378 }, { "epoch": 3.9773356033198555, "grad_norm": 13.5703125, "learning_rate": 8.689631893121093e-06, "loss": 0.6954, "step": 37379 }, { "epoch": 3.977442008938072, "grad_norm": 13.46875, "learning_rate": 8.689596361569074e-06, "loss": 0.6197, "step": 37380 }, { "epoch": 3.9775484145562885, "grad_norm": 15.59375, "learning_rate": 8.689560830017056e-06, "loss": 0.6781, "step": 37381 }, { "epoch": 3.9776548201745054, "grad_norm": 13.6484375, "learning_rate": 8.689525298465038e-06, "loss": 0.6189, "step": 37382 }, { "epoch": 3.977761225792722, "grad_norm": 12.703125, "learning_rate": 8.68948976691302e-06, "loss": 0.6466, "step": 37383 }, { "epoch": 3.9778676314109385, "grad_norm": 13.7109375, "learning_rate": 8.689454235361001e-06, "loss": 0.838, "step": 37384 }, { "epoch": 3.977974037029155, "grad_norm": 11.6875, "learning_rate": 8.689418703808982e-06, "loss": 0.6145, "step": 37385 }, { "epoch": 3.978080442647372, "grad_norm": 10.7109375, "learning_rate": 8.689383172256964e-06, "loss": 0.665, "step": 37386 }, { "epoch": 3.9781868482655884, "grad_norm": 11.765625, "learning_rate": 8.689347640704947e-06, "loss": 0.6113, "step": 37387 }, { "epoch": 3.978293253883805, "grad_norm": 10.15625, "learning_rate": 8.689312109152929e-06, "loss": 0.6213, "step": 37388 }, { "epoch": 3.9783996595020215, "grad_norm": 14.3828125, "learning_rate": 8.68927657760091e-06, "loss": 0.8045, "step": 37389 }, { "epoch": 3.9785060651202384, "grad_norm": 10.890625, "learning_rate": 8.689241046048892e-06, "loss": 0.5816, "step": 37390 }, { "epoch": 3.978612470738455, "grad_norm": 14.9375, "learning_rate": 8.689205514496874e-06, "loss": 0.6247, "step": 37391 }, { "epoch": 3.9787188763566714, "grad_norm": 11.890625, "learning_rate": 8.689169982944855e-06, "loss": 0.6918, "step": 37392 }, { "epoch": 3.9788252819748884, "grad_norm": 12.390625, "learning_rate": 8.689134451392837e-06, "loss": 0.5441, "step": 37393 }, { "epoch": 3.978931687593105, "grad_norm": 12.96875, "learning_rate": 8.689098919840819e-06, "loss": 0.6197, "step": 37394 }, { "epoch": 3.9790380932113214, "grad_norm": 12.1796875, "learning_rate": 8.6890633882888e-06, "loss": 0.5419, "step": 37395 }, { "epoch": 3.9791444988295384, "grad_norm": 11.984375, "learning_rate": 8.689027856736782e-06, "loss": 0.5622, "step": 37396 }, { "epoch": 3.979250904447755, "grad_norm": 11.1953125, "learning_rate": 8.688992325184765e-06, "loss": 0.675, "step": 37397 }, { "epoch": 3.9793573100659714, "grad_norm": 13.796875, "learning_rate": 8.688956793632747e-06, "loss": 0.7289, "step": 37398 }, { "epoch": 3.9794637156841883, "grad_norm": 21.21875, "learning_rate": 8.688921262080728e-06, "loss": 0.7599, "step": 37399 }, { "epoch": 3.979570121302405, "grad_norm": 12.4921875, "learning_rate": 8.68888573052871e-06, "loss": 0.6178, "step": 37400 }, { "epoch": 3.9796765269206213, "grad_norm": 13.3984375, "learning_rate": 8.688850198976692e-06, "loss": 0.6755, "step": 37401 }, { "epoch": 3.9797829325388383, "grad_norm": 11.3125, "learning_rate": 8.688814667424675e-06, "loss": 0.7036, "step": 37402 }, { "epoch": 3.979889338157055, "grad_norm": 11.5703125, "learning_rate": 8.688779135872656e-06, "loss": 0.6213, "step": 37403 }, { "epoch": 3.9799957437752713, "grad_norm": 12.2265625, "learning_rate": 8.688743604320638e-06, "loss": 0.7415, "step": 37404 }, { "epoch": 3.980102149393488, "grad_norm": 13.859375, "learning_rate": 8.68870807276862e-06, "loss": 0.6264, "step": 37405 }, { "epoch": 3.9802085550117043, "grad_norm": 15.046875, "learning_rate": 8.688672541216601e-06, "loss": 0.6522, "step": 37406 }, { "epoch": 3.9803149606299213, "grad_norm": 10.4375, "learning_rate": 8.688637009664583e-06, "loss": 0.4775, "step": 37407 }, { "epoch": 3.980421366248138, "grad_norm": 12.7265625, "learning_rate": 8.688601478112565e-06, "loss": 0.6843, "step": 37408 }, { "epoch": 3.9805277718663543, "grad_norm": 13.484375, "learning_rate": 8.688565946560546e-06, "loss": 0.6028, "step": 37409 }, { "epoch": 3.9806341774845713, "grad_norm": 20.09375, "learning_rate": 8.688530415008528e-06, "loss": 0.9192, "step": 37410 }, { "epoch": 3.9807405831027878, "grad_norm": 9.6171875, "learning_rate": 8.68849488345651e-06, "loss": 0.516, "step": 37411 }, { "epoch": 3.9808469887210043, "grad_norm": 16.265625, "learning_rate": 8.688459351904493e-06, "loss": 0.6435, "step": 37412 }, { "epoch": 3.9809533943392212, "grad_norm": 9.9375, "learning_rate": 8.688423820352474e-06, "loss": 0.5788, "step": 37413 }, { "epoch": 3.9810597999574378, "grad_norm": 14.7578125, "learning_rate": 8.688388288800456e-06, "loss": 0.8943, "step": 37414 }, { "epoch": 3.9811662055756543, "grad_norm": 11.9375, "learning_rate": 8.688352757248437e-06, "loss": 0.6278, "step": 37415 }, { "epoch": 3.981272611193871, "grad_norm": 12.3671875, "learning_rate": 8.688317225696419e-06, "loss": 0.7254, "step": 37416 }, { "epoch": 3.9813790168120877, "grad_norm": 16.203125, "learning_rate": 8.6882816941444e-06, "loss": 0.7065, "step": 37417 }, { "epoch": 3.9814854224303042, "grad_norm": 15.1953125, "learning_rate": 8.688246162592382e-06, "loss": 0.7078, "step": 37418 }, { "epoch": 3.981591828048521, "grad_norm": 17.28125, "learning_rate": 8.688210631040364e-06, "loss": 0.8667, "step": 37419 }, { "epoch": 3.9816982336667377, "grad_norm": 11.1171875, "learning_rate": 8.688175099488345e-06, "loss": 0.6308, "step": 37420 }, { "epoch": 3.981804639284954, "grad_norm": 13.8671875, "learning_rate": 8.688139567936327e-06, "loss": 0.8252, "step": 37421 }, { "epoch": 3.981911044903171, "grad_norm": 12.4765625, "learning_rate": 8.688104036384309e-06, "loss": 0.6461, "step": 37422 }, { "epoch": 3.9820174505213877, "grad_norm": 11.4609375, "learning_rate": 8.688068504832292e-06, "loss": 0.503, "step": 37423 }, { "epoch": 3.982123856139604, "grad_norm": 12.390625, "learning_rate": 8.688032973280274e-06, "loss": 0.8793, "step": 37424 }, { "epoch": 3.9822302617578207, "grad_norm": 13.2890625, "learning_rate": 8.687997441728255e-06, "loss": 0.7392, "step": 37425 }, { "epoch": 3.982336667376037, "grad_norm": 14.828125, "learning_rate": 8.687961910176237e-06, "loss": 0.7592, "step": 37426 }, { "epoch": 3.982443072994254, "grad_norm": 11.8125, "learning_rate": 8.687926378624218e-06, "loss": 0.7165, "step": 37427 }, { "epoch": 3.9825494786124707, "grad_norm": 11.8203125, "learning_rate": 8.687890847072202e-06, "loss": 0.6123, "step": 37428 }, { "epoch": 3.982655884230687, "grad_norm": 12.9375, "learning_rate": 8.687855315520183e-06, "loss": 0.684, "step": 37429 }, { "epoch": 3.982762289848904, "grad_norm": 11.328125, "learning_rate": 8.687819783968165e-06, "loss": 0.5974, "step": 37430 }, { "epoch": 3.9828686954671206, "grad_norm": 14.078125, "learning_rate": 8.687784252416147e-06, "loss": 0.8565, "step": 37431 }, { "epoch": 3.982975101085337, "grad_norm": 13.46875, "learning_rate": 8.687748720864128e-06, "loss": 0.7794, "step": 37432 }, { "epoch": 3.983081506703554, "grad_norm": 11.890625, "learning_rate": 8.68771318931211e-06, "loss": 0.612, "step": 37433 }, { "epoch": 3.9831879123217706, "grad_norm": 12.640625, "learning_rate": 8.687677657760091e-06, "loss": 0.6734, "step": 37434 }, { "epoch": 3.983294317939987, "grad_norm": 22.859375, "learning_rate": 8.687642126208073e-06, "loss": 0.7209, "step": 37435 }, { "epoch": 3.983400723558204, "grad_norm": 14.40625, "learning_rate": 8.687606594656055e-06, "loss": 0.6935, "step": 37436 }, { "epoch": 3.9835071291764206, "grad_norm": 14.6171875, "learning_rate": 8.687571063104036e-06, "loss": 0.7302, "step": 37437 }, { "epoch": 3.983613534794637, "grad_norm": 14.1015625, "learning_rate": 8.68753553155202e-06, "loss": 0.6015, "step": 37438 }, { "epoch": 3.983719940412854, "grad_norm": 15.4375, "learning_rate": 8.687500000000001e-06, "loss": 0.8011, "step": 37439 }, { "epoch": 3.9838263460310706, "grad_norm": 18.421875, "learning_rate": 8.687464468447983e-06, "loss": 0.6851, "step": 37440 }, { "epoch": 3.983932751649287, "grad_norm": 9.21875, "learning_rate": 8.687428936895964e-06, "loss": 0.6753, "step": 37441 }, { "epoch": 3.9840391572675036, "grad_norm": 13.125, "learning_rate": 8.687393405343946e-06, "loss": 0.735, "step": 37442 }, { "epoch": 3.9841455628857205, "grad_norm": 12.6015625, "learning_rate": 8.68735787379193e-06, "loss": 0.6567, "step": 37443 }, { "epoch": 3.984251968503937, "grad_norm": 11.7890625, "learning_rate": 8.68732234223991e-06, "loss": 0.5391, "step": 37444 }, { "epoch": 3.9843583741221535, "grad_norm": 10.78125, "learning_rate": 8.687286810687892e-06, "loss": 0.7055, "step": 37445 }, { "epoch": 3.98446477974037, "grad_norm": 13.4765625, "learning_rate": 8.687251279135872e-06, "loss": 0.7469, "step": 37446 }, { "epoch": 3.984571185358587, "grad_norm": 17.765625, "learning_rate": 8.687215747583854e-06, "loss": 0.8773, "step": 37447 }, { "epoch": 3.9846775909768035, "grad_norm": 11.296875, "learning_rate": 8.687180216031837e-06, "loss": 0.5583, "step": 37448 }, { "epoch": 3.98478399659502, "grad_norm": 9.9921875, "learning_rate": 8.687144684479819e-06, "loss": 0.7103, "step": 37449 }, { "epoch": 3.984890402213237, "grad_norm": 11.40625, "learning_rate": 8.6871091529278e-06, "loss": 0.6525, "step": 37450 }, { "epoch": 3.9849968078314535, "grad_norm": 12.6171875, "learning_rate": 8.687073621375782e-06, "loss": 0.5895, "step": 37451 }, { "epoch": 3.98510321344967, "grad_norm": 9.8359375, "learning_rate": 8.687038089823764e-06, "loss": 0.498, "step": 37452 }, { "epoch": 3.985209619067887, "grad_norm": 18.203125, "learning_rate": 8.687002558271747e-06, "loss": 0.7139, "step": 37453 }, { "epoch": 3.9853160246861035, "grad_norm": 13.6171875, "learning_rate": 8.686967026719729e-06, "loss": 0.6512, "step": 37454 }, { "epoch": 3.98542243030432, "grad_norm": 12.375, "learning_rate": 8.68693149516771e-06, "loss": 0.696, "step": 37455 }, { "epoch": 3.985528835922537, "grad_norm": 15.1875, "learning_rate": 8.686895963615692e-06, "loss": 0.7756, "step": 37456 }, { "epoch": 3.9856352415407534, "grad_norm": 15.4296875, "learning_rate": 8.686860432063673e-06, "loss": 0.6649, "step": 37457 }, { "epoch": 3.98574164715897, "grad_norm": 10.0859375, "learning_rate": 8.686824900511655e-06, "loss": 0.5724, "step": 37458 }, { "epoch": 3.985848052777187, "grad_norm": 17.75, "learning_rate": 8.686789368959637e-06, "loss": 0.6023, "step": 37459 }, { "epoch": 3.9859544583954034, "grad_norm": 9.6953125, "learning_rate": 8.686753837407618e-06, "loss": 0.6706, "step": 37460 }, { "epoch": 3.98606086401362, "grad_norm": 14.9765625, "learning_rate": 8.6867183058556e-06, "loss": 0.6916, "step": 37461 }, { "epoch": 3.9861672696318364, "grad_norm": 14.2265625, "learning_rate": 8.686682774303581e-06, "loss": 0.6309, "step": 37462 }, { "epoch": 3.986273675250053, "grad_norm": 15.8828125, "learning_rate": 8.686647242751563e-06, "loss": 0.7211, "step": 37463 }, { "epoch": 3.98638008086827, "grad_norm": 12.9140625, "learning_rate": 8.686611711199546e-06, "loss": 0.6534, "step": 37464 }, { "epoch": 3.9864864864864864, "grad_norm": 12.71875, "learning_rate": 8.686576179647528e-06, "loss": 0.7082, "step": 37465 }, { "epoch": 3.986592892104703, "grad_norm": 13.1171875, "learning_rate": 8.68654064809551e-06, "loss": 0.6264, "step": 37466 }, { "epoch": 3.98669929772292, "grad_norm": 13.7109375, "learning_rate": 8.686505116543491e-06, "loss": 0.6976, "step": 37467 }, { "epoch": 3.9868057033411364, "grad_norm": 12.578125, "learning_rate": 8.686469584991473e-06, "loss": 0.7003, "step": 37468 }, { "epoch": 3.986912108959353, "grad_norm": 13.671875, "learning_rate": 8.686434053439456e-06, "loss": 0.6676, "step": 37469 }, { "epoch": 3.98701851457757, "grad_norm": 20.171875, "learning_rate": 8.686398521887438e-06, "loss": 0.7525, "step": 37470 }, { "epoch": 3.9871249201957863, "grad_norm": 12.328125, "learning_rate": 8.68636299033542e-06, "loss": 0.6737, "step": 37471 }, { "epoch": 3.987231325814003, "grad_norm": 12.1640625, "learning_rate": 8.686327458783401e-06, "loss": 0.5167, "step": 37472 }, { "epoch": 3.98733773143222, "grad_norm": 14.546875, "learning_rate": 8.686291927231383e-06, "loss": 0.676, "step": 37473 }, { "epoch": 3.9874441370504363, "grad_norm": 12.875, "learning_rate": 8.686256395679364e-06, "loss": 0.6814, "step": 37474 }, { "epoch": 3.987550542668653, "grad_norm": 14.4296875, "learning_rate": 8.686220864127346e-06, "loss": 0.7221, "step": 37475 }, { "epoch": 3.98765694828687, "grad_norm": 12.40625, "learning_rate": 8.686185332575327e-06, "loss": 0.7654, "step": 37476 }, { "epoch": 3.9877633539050863, "grad_norm": 13.25, "learning_rate": 8.686149801023309e-06, "loss": 0.6805, "step": 37477 }, { "epoch": 3.987869759523303, "grad_norm": 15.65625, "learning_rate": 8.68611426947129e-06, "loss": 0.6654, "step": 37478 }, { "epoch": 3.9879761651415193, "grad_norm": 18.546875, "learning_rate": 8.686078737919274e-06, "loss": 0.697, "step": 37479 }, { "epoch": 3.9880825707597363, "grad_norm": 14.65625, "learning_rate": 8.686043206367255e-06, "loss": 0.6942, "step": 37480 }, { "epoch": 3.9881889763779528, "grad_norm": 10.8984375, "learning_rate": 8.686007674815237e-06, "loss": 0.5099, "step": 37481 }, { "epoch": 3.9882953819961693, "grad_norm": 12.5234375, "learning_rate": 8.685972143263219e-06, "loss": 0.6276, "step": 37482 }, { "epoch": 3.988401787614386, "grad_norm": 11.6953125, "learning_rate": 8.6859366117112e-06, "loss": 0.5669, "step": 37483 }, { "epoch": 3.9885081932326027, "grad_norm": 12.046875, "learning_rate": 8.685901080159182e-06, "loss": 0.5886, "step": 37484 }, { "epoch": 3.9886145988508193, "grad_norm": 16.78125, "learning_rate": 8.685865548607163e-06, "loss": 0.5569, "step": 37485 }, { "epoch": 3.9887210044690358, "grad_norm": 14.390625, "learning_rate": 8.685830017055145e-06, "loss": 0.6412, "step": 37486 }, { "epoch": 3.9888274100872527, "grad_norm": 16.6875, "learning_rate": 8.685794485503127e-06, "loss": 0.6965, "step": 37487 }, { "epoch": 3.9889338157054692, "grad_norm": 18.40625, "learning_rate": 8.685758953951108e-06, "loss": 0.781, "step": 37488 }, { "epoch": 3.9890402213236857, "grad_norm": 21.21875, "learning_rate": 8.685723422399092e-06, "loss": 0.6923, "step": 37489 }, { "epoch": 3.9891466269419027, "grad_norm": 11.6875, "learning_rate": 8.685687890847073e-06, "loss": 0.5793, "step": 37490 }, { "epoch": 3.989253032560119, "grad_norm": 16.421875, "learning_rate": 8.685652359295055e-06, "loss": 0.7513, "step": 37491 }, { "epoch": 3.9893594381783357, "grad_norm": 15.515625, "learning_rate": 8.685616827743036e-06, "loss": 0.7072, "step": 37492 }, { "epoch": 3.9894658437965527, "grad_norm": 12.625, "learning_rate": 8.685581296191018e-06, "loss": 0.6798, "step": 37493 }, { "epoch": 3.989572249414769, "grad_norm": 15.8203125, "learning_rate": 8.685545764639001e-06, "loss": 0.6266, "step": 37494 }, { "epoch": 3.9896786550329857, "grad_norm": 14.0, "learning_rate": 8.685510233086983e-06, "loss": 0.6585, "step": 37495 }, { "epoch": 3.9897850606512026, "grad_norm": 15.8984375, "learning_rate": 8.685474701534965e-06, "loss": 0.7066, "step": 37496 }, { "epoch": 3.989891466269419, "grad_norm": 12.5703125, "learning_rate": 8.685439169982946e-06, "loss": 0.5924, "step": 37497 }, { "epoch": 3.9899978718876357, "grad_norm": 14.828125, "learning_rate": 8.685403638430928e-06, "loss": 0.7031, "step": 37498 }, { "epoch": 3.990104277505852, "grad_norm": 11.90625, "learning_rate": 8.68536810687891e-06, "loss": 0.5443, "step": 37499 }, { "epoch": 3.9902106831240687, "grad_norm": 15.0859375, "learning_rate": 8.685332575326891e-06, "loss": 0.7717, "step": 37500 }, { "epoch": 3.9903170887422856, "grad_norm": 12.109375, "learning_rate": 8.685297043774873e-06, "loss": 0.7917, "step": 37501 }, { "epoch": 3.990423494360502, "grad_norm": 13.6953125, "learning_rate": 8.685261512222854e-06, "loss": 0.7078, "step": 37502 }, { "epoch": 3.9905298999787187, "grad_norm": 18.375, "learning_rate": 8.685225980670836e-06, "loss": 0.9425, "step": 37503 }, { "epoch": 3.9906363055969356, "grad_norm": 16.5, "learning_rate": 8.685190449118817e-06, "loss": 0.7542, "step": 37504 }, { "epoch": 3.990742711215152, "grad_norm": 11.9921875, "learning_rate": 8.6851549175668e-06, "loss": 0.6314, "step": 37505 }, { "epoch": 3.9908491168333686, "grad_norm": 14.4609375, "learning_rate": 8.685119386014782e-06, "loss": 0.7149, "step": 37506 }, { "epoch": 3.9909555224515856, "grad_norm": 14.28125, "learning_rate": 8.685083854462764e-06, "loss": 0.6659, "step": 37507 }, { "epoch": 3.991061928069802, "grad_norm": 16.078125, "learning_rate": 8.685048322910746e-06, "loss": 0.7482, "step": 37508 }, { "epoch": 3.9911683336880186, "grad_norm": 22.953125, "learning_rate": 8.685012791358727e-06, "loss": 0.9319, "step": 37509 }, { "epoch": 3.9912747393062356, "grad_norm": 12.984375, "learning_rate": 8.684977259806709e-06, "loss": 0.6718, "step": 37510 }, { "epoch": 3.991381144924452, "grad_norm": 12.9453125, "learning_rate": 8.68494172825469e-06, "loss": 0.7081, "step": 37511 }, { "epoch": 3.9914875505426686, "grad_norm": 17.21875, "learning_rate": 8.684906196702672e-06, "loss": 0.718, "step": 37512 }, { "epoch": 3.9915939561608855, "grad_norm": 15.1171875, "learning_rate": 8.684870665150654e-06, "loss": 0.7348, "step": 37513 }, { "epoch": 3.991700361779102, "grad_norm": 12.0078125, "learning_rate": 8.684835133598635e-06, "loss": 0.6308, "step": 37514 }, { "epoch": 3.9918067673973185, "grad_norm": 11.0703125, "learning_rate": 8.684799602046618e-06, "loss": 0.658, "step": 37515 }, { "epoch": 3.991913173015535, "grad_norm": 9.9921875, "learning_rate": 8.6847640704946e-06, "loss": 0.6228, "step": 37516 }, { "epoch": 3.992019578633752, "grad_norm": 14.3125, "learning_rate": 8.684728538942582e-06, "loss": 0.8019, "step": 37517 }, { "epoch": 3.9921259842519685, "grad_norm": 13.859375, "learning_rate": 8.684693007390563e-06, "loss": 0.8384, "step": 37518 }, { "epoch": 3.992232389870185, "grad_norm": 13.65625, "learning_rate": 8.684657475838545e-06, "loss": 0.6313, "step": 37519 }, { "epoch": 3.9923387954884015, "grad_norm": 12.2890625, "learning_rate": 8.684621944286528e-06, "loss": 0.672, "step": 37520 }, { "epoch": 3.9924452011066185, "grad_norm": 15.828125, "learning_rate": 8.68458641273451e-06, "loss": 0.836, "step": 37521 }, { "epoch": 3.992551606724835, "grad_norm": 11.6953125, "learning_rate": 8.684550881182491e-06, "loss": 0.6323, "step": 37522 }, { "epoch": 3.9926580123430515, "grad_norm": 12.984375, "learning_rate": 8.684515349630473e-06, "loss": 0.6146, "step": 37523 }, { "epoch": 3.9927644179612685, "grad_norm": 11.75, "learning_rate": 8.684479818078455e-06, "loss": 0.7087, "step": 37524 }, { "epoch": 3.992870823579485, "grad_norm": 18.6875, "learning_rate": 8.684444286526436e-06, "loss": 0.7756, "step": 37525 }, { "epoch": 3.9929772291977015, "grad_norm": 13.8828125, "learning_rate": 8.684408754974418e-06, "loss": 0.6891, "step": 37526 }, { "epoch": 3.9930836348159184, "grad_norm": 11.6953125, "learning_rate": 8.6843732234224e-06, "loss": 0.5792, "step": 37527 }, { "epoch": 3.993190040434135, "grad_norm": 14.484375, "learning_rate": 8.684337691870381e-06, "loss": 0.655, "step": 37528 }, { "epoch": 3.9932964460523515, "grad_norm": 14.2734375, "learning_rate": 8.684302160318363e-06, "loss": 0.7194, "step": 37529 }, { "epoch": 3.9934028516705684, "grad_norm": 12.8359375, "learning_rate": 8.684266628766346e-06, "loss": 0.638, "step": 37530 }, { "epoch": 3.993509257288785, "grad_norm": 11.7265625, "learning_rate": 8.684231097214328e-06, "loss": 0.5619, "step": 37531 }, { "epoch": 3.9936156629070014, "grad_norm": 12.5625, "learning_rate": 8.68419556566231e-06, "loss": 0.7331, "step": 37532 }, { "epoch": 3.9937220685252184, "grad_norm": 11.3515625, "learning_rate": 8.68416003411029e-06, "loss": 0.6223, "step": 37533 }, { "epoch": 3.993828474143435, "grad_norm": 13.8359375, "learning_rate": 8.684124502558272e-06, "loss": 0.6488, "step": 37534 }, { "epoch": 3.9939348797616514, "grad_norm": 11.6796875, "learning_rate": 8.684088971006254e-06, "loss": 0.7114, "step": 37535 }, { "epoch": 3.994041285379868, "grad_norm": 10.2578125, "learning_rate": 8.684053439454236e-06, "loss": 0.628, "step": 37536 }, { "epoch": 3.994147690998085, "grad_norm": 15.9765625, "learning_rate": 8.684017907902217e-06, "loss": 0.6109, "step": 37537 }, { "epoch": 3.9942540966163014, "grad_norm": 12.7734375, "learning_rate": 8.683982376350199e-06, "loss": 0.6419, "step": 37538 }, { "epoch": 3.994360502234518, "grad_norm": 11.984375, "learning_rate": 8.68394684479818e-06, "loss": 0.6683, "step": 37539 }, { "epoch": 3.9944669078527344, "grad_norm": 13.3125, "learning_rate": 8.683911313246162e-06, "loss": 0.6428, "step": 37540 }, { "epoch": 3.9945733134709513, "grad_norm": 13.0859375, "learning_rate": 8.683875781694145e-06, "loss": 0.674, "step": 37541 }, { "epoch": 3.994679719089168, "grad_norm": 18.609375, "learning_rate": 8.683840250142127e-06, "loss": 0.9928, "step": 37542 }, { "epoch": 3.9947861247073844, "grad_norm": 13.953125, "learning_rate": 8.683804718590109e-06, "loss": 0.7873, "step": 37543 }, { "epoch": 3.9948925303256013, "grad_norm": 13.3125, "learning_rate": 8.68376918703809e-06, "loss": 0.786, "step": 37544 }, { "epoch": 3.994998935943818, "grad_norm": 20.359375, "learning_rate": 8.683733655486072e-06, "loss": 0.6519, "step": 37545 }, { "epoch": 3.9951053415620343, "grad_norm": 12.1640625, "learning_rate": 8.683698123934055e-06, "loss": 0.5863, "step": 37546 }, { "epoch": 3.9952117471802513, "grad_norm": 15.3125, "learning_rate": 8.683662592382037e-06, "loss": 0.7784, "step": 37547 }, { "epoch": 3.995318152798468, "grad_norm": 10.8359375, "learning_rate": 8.683627060830018e-06, "loss": 0.7685, "step": 37548 }, { "epoch": 3.9954245584166843, "grad_norm": 12.046875, "learning_rate": 8.683591529278e-06, "loss": 0.6404, "step": 37549 }, { "epoch": 3.9955309640349013, "grad_norm": 12.140625, "learning_rate": 8.683555997725981e-06, "loss": 0.6613, "step": 37550 }, { "epoch": 3.9956373696531178, "grad_norm": 16.828125, "learning_rate": 8.683520466173963e-06, "loss": 0.6833, "step": 37551 }, { "epoch": 3.9957437752713343, "grad_norm": 13.7265625, "learning_rate": 8.683484934621945e-06, "loss": 0.838, "step": 37552 }, { "epoch": 3.9958501808895512, "grad_norm": 12.9921875, "learning_rate": 8.683449403069926e-06, "loss": 0.6426, "step": 37553 }, { "epoch": 3.9959565865077677, "grad_norm": 15.2890625, "learning_rate": 8.683413871517908e-06, "loss": 0.6919, "step": 37554 }, { "epoch": 3.9960629921259843, "grad_norm": 12.2265625, "learning_rate": 8.68337833996589e-06, "loss": 0.5847, "step": 37555 }, { "epoch": 3.9961693977442008, "grad_norm": 13.03125, "learning_rate": 8.683342808413873e-06, "loss": 0.6211, "step": 37556 }, { "epoch": 3.9962758033624173, "grad_norm": 14.4921875, "learning_rate": 8.683307276861854e-06, "loss": 0.9216, "step": 37557 }, { "epoch": 3.9963822089806342, "grad_norm": 16.609375, "learning_rate": 8.683271745309836e-06, "loss": 0.8112, "step": 37558 }, { "epoch": 3.9964886145988507, "grad_norm": 13.6875, "learning_rate": 8.683236213757818e-06, "loss": 0.7206, "step": 37559 }, { "epoch": 3.9965950202170673, "grad_norm": 14.625, "learning_rate": 8.6832006822058e-06, "loss": 0.7338, "step": 37560 }, { "epoch": 3.996701425835284, "grad_norm": 11.28125, "learning_rate": 8.683165150653783e-06, "loss": 0.5752, "step": 37561 }, { "epoch": 3.9968078314535007, "grad_norm": 10.6640625, "learning_rate": 8.683129619101764e-06, "loss": 0.6095, "step": 37562 }, { "epoch": 3.9969142370717172, "grad_norm": 13.1796875, "learning_rate": 8.683094087549746e-06, "loss": 0.6871, "step": 37563 }, { "epoch": 3.997020642689934, "grad_norm": 13.71875, "learning_rate": 8.683058555997727e-06, "loss": 0.7018, "step": 37564 }, { "epoch": 3.9971270483081507, "grad_norm": 13.5078125, "learning_rate": 8.683023024445707e-06, "loss": 0.7075, "step": 37565 }, { "epoch": 3.997233453926367, "grad_norm": 12.546875, "learning_rate": 8.68298749289369e-06, "loss": 0.7117, "step": 37566 }, { "epoch": 3.997339859544584, "grad_norm": 12.6015625, "learning_rate": 8.682951961341672e-06, "loss": 0.9278, "step": 37567 }, { "epoch": 3.9974462651628007, "grad_norm": 12.796875, "learning_rate": 8.682916429789654e-06, "loss": 0.8345, "step": 37568 }, { "epoch": 3.997552670781017, "grad_norm": 15.515625, "learning_rate": 8.682880898237635e-06, "loss": 0.6594, "step": 37569 }, { "epoch": 3.997659076399234, "grad_norm": 14.2890625, "learning_rate": 8.682845366685617e-06, "loss": 0.7562, "step": 37570 }, { "epoch": 3.9977654820174506, "grad_norm": 13.671875, "learning_rate": 8.6828098351336e-06, "loss": 0.6729, "step": 37571 }, { "epoch": 3.997871887635667, "grad_norm": 18.015625, "learning_rate": 8.682774303581582e-06, "loss": 0.843, "step": 37572 }, { "epoch": 3.9979782932538837, "grad_norm": 10.546875, "learning_rate": 8.682738772029563e-06, "loss": 0.7271, "step": 37573 }, { "epoch": 3.9980846988721006, "grad_norm": 13.5703125, "learning_rate": 8.682703240477545e-06, "loss": 0.7781, "step": 37574 }, { "epoch": 3.998191104490317, "grad_norm": 12.59375, "learning_rate": 8.682667708925527e-06, "loss": 0.5981, "step": 37575 }, { "epoch": 3.9982975101085336, "grad_norm": 16.234375, "learning_rate": 8.682632177373508e-06, "loss": 0.5844, "step": 37576 }, { "epoch": 3.99840391572675, "grad_norm": 9.7578125, "learning_rate": 8.68259664582149e-06, "loss": 0.5029, "step": 37577 }, { "epoch": 3.998510321344967, "grad_norm": 11.328125, "learning_rate": 8.682561114269472e-06, "loss": 0.6539, "step": 37578 }, { "epoch": 3.9986167269631836, "grad_norm": 12.0390625, "learning_rate": 8.682525582717453e-06, "loss": 0.6092, "step": 37579 }, { "epoch": 3.9987231325814, "grad_norm": 16.4375, "learning_rate": 8.682490051165435e-06, "loss": 0.6398, "step": 37580 }, { "epoch": 3.998829538199617, "grad_norm": 14.7890625, "learning_rate": 8.682454519613416e-06, "loss": 0.7576, "step": 37581 }, { "epoch": 3.9989359438178336, "grad_norm": 12.453125, "learning_rate": 8.6824189880614e-06, "loss": 0.6054, "step": 37582 }, { "epoch": 3.99904234943605, "grad_norm": 11.359375, "learning_rate": 8.682383456509381e-06, "loss": 0.5264, "step": 37583 }, { "epoch": 3.999148755054267, "grad_norm": 13.046875, "learning_rate": 8.682347924957363e-06, "loss": 0.7442, "step": 37584 }, { "epoch": 3.9992551606724835, "grad_norm": 12.3359375, "learning_rate": 8.682312393405344e-06, "loss": 0.608, "step": 37585 }, { "epoch": 3.9993615662907, "grad_norm": 12.015625, "learning_rate": 8.682276861853326e-06, "loss": 0.5649, "step": 37586 }, { "epoch": 3.999467971908917, "grad_norm": 12.4296875, "learning_rate": 8.68224133030131e-06, "loss": 0.7113, "step": 37587 }, { "epoch": 3.9995743775271335, "grad_norm": 15.53125, "learning_rate": 8.682205798749291e-06, "loss": 0.6776, "step": 37588 }, { "epoch": 3.99968078314535, "grad_norm": 12.0390625, "learning_rate": 8.682170267197273e-06, "loss": 0.6617, "step": 37589 }, { "epoch": 3.999787188763567, "grad_norm": 11.2109375, "learning_rate": 8.682134735645254e-06, "loss": 0.5121, "step": 37590 }, { "epoch": 3.9998935943817835, "grad_norm": 11.609375, "learning_rate": 8.682099204093236e-06, "loss": 0.698, "step": 37591 }, { "epoch": 4.0, "grad_norm": 11.9296875, "learning_rate": 8.682063672541217e-06, "loss": 0.4425, "step": 37592 } ], "logging_steps": 1, "max_steps": 281940, "num_input_tokens_seen": 0, "num_train_epochs": 30, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 1.282991361490944e+18, "train_batch_size": 4, "trial_name": null, "trial_params": null }